Astrophysics and Space Science Proceedings

For further volumes:

http://www.springer.com/series/7395\
Principles and Perspectives
in Cosmochemistry

Lecture Notes of the Kodai School

on ‘Synthesis of Elements in Stars’ held
at Kodaikanal Observatory, India,
April 29-May 13, 2008

Aruna Goswami
Editor

Indian Institute of Astrophysics,

Bangalore, India

B. Eswar Reddy
Editor

Indian Institute of Astrophysics,

Bangalore, India

123
Editors
Aruna Goswami B. Eswar Reddy
Indian Institute of Astrophysics Indian Institute of Astrophysics
2nd Block, Koramangala 2nd Block, Koramangala
Bangalore 560034 Bangalore 560034
India India
[email protected] [email protected]

ISSN 1570-6591 e-ISSN 1570-6605

ISBN 978-3-642-10351-3 e-ISBN 978-3-642-10352-0
DOI 10.1007/978-3-642-10352-0
Springer Heidelberg Dordrecht London New York

Library of Congress Control Number: 2010921800

c Springer-Verlag Berlin Heidelberg 2010

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September 9,
1965, in its current version, and permission for use must always be obtained from Springer. Violations are
liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws
and regulations and therefore free for general use.

Cover design: eStudio Calamar S.L.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Preface

The origin of elements is among the fundamental aspects of our universe;

cosmochemistry tries to answer when, how and where the chemical elements
arose after hydrogen was created during primordial nucleosynthesis following
the Big Bang. However, quantitative answers to these fundamental questions
began to emerge only in the late fifties, with the pioneering works of Bur-
bidge, Burbidge, Fowler and Hoyle, and Cameron. Since then there had been
significant progress in the understanding of synthesis of elements in stars.
Cosmochemistry, however, remains a fertile area of research, as there
remain many outstanding problems. A comprehensive approach to cosmo-
chemistry requires a combination of a number of topics like primordial
nucleosynthesis, stellar nucleosynthesis, explosive nucleosynthesis and solar
abundance. The Kodai school on ‘Synthesis of elements in stars’ was orga-
nized to provide a glimpse of this exciting area of research to astrophysicists
of tomorrow, motivated young students from India and abroad. The lectures
are thus aimed at researchers who would like to venture deeper into this ex-
citing arena.
The school drew strength from considerable in-house expertise at IIA in a
number of areas critical for the school. A highlight of the school, however, was
the faculty participation by a number of leading astrophysicists from different
parts of the world.
Following a traditional and inspiring invocation from Upanishad and a
brief inaugural function, the school was opened for technical sessions. David
Lambert set the tone of the scientific sessions with the lead talk on ‘Synthesis
of elements in stars: an overview’. The basic properties of nuclei were ex-
plained by Arun Mangalam in a series of lectures. The lectures by C Sivaram
put the primary issue in cosmochemistry in perspective through a discussion
on cosmological nucleosynthesis of light elements. Aruna Goswami discussed
some current issues in the present understanding of the Galactic chemical
evolution. Gajendra Pandey explained how stellar spectra can be analyzed
using ‘Curve of growth technique’. Kameswara Rao of IIA talked about the
high resolution Echelle spectrograph at VBO, Kavalur and discussed some
results obtained from analysis of data acquired using this instrument. These
VI Preface

lectures provided the background for the series of lectures by other speakers
that followed. Apart from the regular class room lectures, students had ample
time for hands-on sessions coordinated by Goswami, Reddy and Pandey.
The book has been organized into three parts to address the major issues
in cosmochemistry. Part I of the book deals with stellar structure, nucle-
osynthesis and evolution of low and intermediate-mass stars. The lectures by
Simon Jeffery outline stellar evolution with discussion on the basic equations,
elementary solutions and numerical methods. Amanda Karakas’s lectures dis-
cuss nucleosynthesis of low and intermediate-mass stars covering nucleosyn-
thesis prior to the Asymptotic Giant Branch (AGB) phase, evolution during
the AGB, nucleosynthesis during the AGB phase, evolution after the AGB
and massive AGB stars. The slow neutron-capture process and yields from
AGB stars are also discussed in detail by Karakas. The lectures by S Giridhar
provide some necessary background on stellar classification.
Part II deals with explosive nucleosynthesis that plays a critical role in cos-
mochemistry. The lectures by Kamales Kar provide essential background ma-
terial on weak-interaction rates for stellar evolution, supernovae and r-process
nucleosynthesis. He also discusses in detail the solar neutrino problem. Mas-
sive stars, their evolution and nuclear reaction rates from the point of view
of astronomers and nuclear physicists are discussed by Alak Ray. His lec-
tures also describe the various stages of hydrostatic nuclear fuel burning with
illustrative examples of how the reactions are computed. He also discussed
core-collapse (thermonuclear vs. core-collapse) and supernovae in brief. The
lectures by Marcel Arnould address the phenomena of evolution of massive
stars and the concomitant non-explosive and explosive nucleosynthesis. He
highlights a number of important problems that are yet unresolved but cru-
cial for our understanding of Galactic chemical evolution. The p-process nu-
cleosynthesis attributed to the production of proton-rich elements, a topic of
great importance but yet less explored is also discussed in his lectures.
The third and the final part of the book addresses use of solar system abun-
dances to probe cosmochemistry quantitatively. The lectures by Bruce Fegley
address cosmochemistry of the major elements; while the lectures by Katha-
rina Lodders discuss elemental abundances in Solar, meteoritic and outside
the solar system.
Cosmochemistry is still an evolving branch of astrophysics, with many
challenges. The book is expected to serve as a contemporary reference material
for research in cosmochemistry. We would like to take this opportunity to
thank all the contributors for making this book a reality.

Bangalore, Aruna Goswami

April 2009 B. Eswar Reddy
 Acknowledgement
This school would not have been possible without the dedicated support of
many. We extend our sincere thanks to professor Siraj Hasan, Director, In-
dian Institute of Astrophysics and professor Vinod Krishan for their all round
support for the school.
We are particularly grateful to the school faculty from India and abroad
for readily accepting to participate, prepare lecture notes and spend time with
the students.
The organization of the school is a collective effort of the coordinators, the
convener, the members of the local organizing committee and many others. We
are thankful to the administrative department of IIA and the staff members of
Kodaikanal Solar observatory for their help and support in various activities
of the school.
Contents

Part I Stellar Structure, Nucleosynthesis and Evolution of Low

and Intermediate-mass Stars

Stellar Structure and Evolution: An Introduction

C. Simon Jeffery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Nucleosynthesis of Low and Intermediate-mass Stars
Amanda I. Karakas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Spectral Classification: Old and Contemporary
Sunetra Giridhar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

Part II Massive Stars, Core Collapse, Explosive Nucleosynthesis

Weak Interaction Rates for Stellar Evolution, Supernovae

and r-Process Nucleosynthesis
Kamales Kar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Massive stars as thermonuclear reactors and their explosions
following core collapse
Alak Ray . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209

The Evolution of Massive Stars and the Concomitant

Non-explosive and Explosive Nucleosynthesis
Marcel Arnould . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277

Part III Cosmochemistry and Solar System Abundances

Cosmochemistry
Bruce Fegley, Jr., Laura Schaefer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347
Solar System Abundances of the Elements
Katharina Lodders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
Cosmochemistry: A Perspective
Aruna Goswami . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 419
List of Contributors

C. Simon Jeffery Bruce Fegley

Armagh Observatory, College Planetary chemistry Laboratory,
Hill, Armagh ET61 9DG, Department of earth and
Northern Ireland planetary sciences, Washington
[email protected] University, St. Louis,
Amanda I. Karakas MO63130-4899, USA
Research School of [email protected]
Astronomy & Astrophysics
[email protected]
Sunetra Giridhar Laura Schaefer
Indian Institute of Astrophysics, Planetary chemistry Laboratory,
Bangalore 560034, India Department of earth
[email protected] and planetary sciences, Washington
Kamales Kar University, St. Louis,
Saha Institute of Nuclear Physics, MO63130-4899, USA
Bidhannagar, Kolkata 700064, India laura [email protected]
[email protected]
Alak Ray
Tata Institute of Fundamental Katharina Lodders
Research, Mumbai 400005, India Planetary chemistry laboratory,
[email protected] Department of earth and planetary
Marcel Arnould sciences and McDonnell centre
Institut d‘Astronomie et for the space sciences, Washington
d‘Astrophysique, Universite‘ University, Campus box,
Libre de Bruxelles, CP-226, 1169, One Brookings Drive, Saint
B-1050 Brussels, Belgium Louis, MO63130, USA
[email protected] [email protected]
 School Faculty

Marcel Arnould
Institut d‘Astronomie et d‘Astrophysique,
Universite‘ Libre de Bruxelles, CP-226, B-1050 Brussels, Belgium
[email protected]

Bruce Fegley
Planetary chemistry Laboratory
Department of earth and planetary sciences, Washington University
St. Louis, MO63130-4899, USA
[email protected]

Sunetra Giridhar
Indian Institute of Astrophysics, Bangalore 560034, India
[email protected]

Aruna Goswami
Indian Institute of Astrophysics, Bangalore 560034, India
[email protected]

C. Simon Jeffery
Armagh Observatory, College Hill, Armagh ET61 9DG, Northern Ireland
[email protected]

Kamales Kar
Saha Institute of Nuclear Physics, Bidhannagar, Kolkata 700064, India
[email protected]

Amanda I. Karakas
Research School of Astronomy & Astrophysics
[email protected]

David L. Lambert
McDonald Observatory, University of Texas at austin, Austin
[email protected]

Katharina Lodders
Planetary chemistry laboratory
Department of earth and planetary sciences and McDonnell centre for the
space sciences
Washington University, Campus box, 1169, One Brookings Drive, Saint Louis,
MO63130, USA
[email protected]
 School Faculty XIII

Arun Mangalam
Indian Institute of Astrophysics, Bangalore 560034, India
[email protected]

Gajendra Pandey
Indian Institute of Astrophysics, Bangalore 560034, India
[email protected]

N. Kameswara Rao
Indian Institute of Astrophysics, Bangalore 560034, India
[email protected]

Alak Ray
Tata Institute of Fundamental Research, Mumbai 400005, India
[email protected]

Eswar Reddy
Indian Institute of Astrophysics, Bangalore 560034, India
[email protected]

C. Sivaram
Indian Institute of Astrophysics, Bangalore 560034, India
[email protected]
 List of Participants

Tiago Mendes de Almeida

Cidade Universitaria, Sao Paulo-SP-Brazil

Sonam Arora
Department of Physics, Punjab University, India

Andrea Borch
Indian Institute of Astrophysics, Bangalore-560034, India

Bhavya B.
Department of Physics, CUSAT, Cochin, Kerala, India

R. S. Keerthi Chandar
Department of Physics, Bharathiar University, Tamil Nadu, India

K. Chandrashekhar
Indian Institute of Astrophysics, Bangalore-560034, India

Sukanta Deb
Department of Physics & Astrophysics, Delhi University, Delhi, India

Nandita Debnath
Department of Physics, Tezpur University, Napam, Sonitpur - 784028, Assam,
India

Thubstan Dorje
Indian Astronomical Observatory, Hanle, Leh-Ladakh, India

Krithika Dota
Department of Physics, Mumbai University, Mumbai, India

Koshy George
ISRO, Bangalore - 560017, India

Suruchi Goel
Physical Research Laboratory, Navrangpura, Ahmedabad - 380009, India

Gagan Gupta
Department of Physics, Panjab University, Chandigarh, India

Singh Abhishek Indrajit

University of Mumbai, Kalina Campus , Santa Cruz, Mumbai - 400098, India
 List of Participants XV

Vishal Joshi
Physical Research Laboratory, Navrangpura, Ahmedabad - 380009, India

Devika Kamath
Department of Physics, Christ College, Bangalore - 560029, India

Sreeja S. Kartha
Indian Institute of Astrophysics, Bangalore - 560034, India

Chrisphin Karthick
ARIES, Manora Peak, Nainital - 263129, India

Rajwinder Kaur
Punjab University, Patiala, Punjab, India

Pranav Kumar
Department of Physics & Astrophysics, University of Delhi, Delhi, India

Blesson Mathew
Indian Institute of Astrophysics, Bangalore - 560034, India

Ritesh K. Mishra
Physical Research Laboratory, Navrangpura, Ahmedabad - 380009, India

Rana Nandi
SINP, Theory Division, 1/AF, Bidhannagar, Kolkata - 700064, India

H. S. Nataraj
Indian Institute of Astrophysics, Bangalore - 560034, India

Vinicius Moris Placco

Cidade Universitaria, Sao Paulo-SP-Brazil

Ananta C. Pradhan
Indian Institute of Astrophysics, Bangalore - 560034, India

Yogesh Prasad
Dept. of Physics, H.N.B. Garhwal University
Srinagar Garhwal- 246174, Uttarakhand, India

Ashish Raj
Physical Research Laboratory, Navrangpura, Ahmedabad - 380009, India
XVI List of Participants

N. G. Rudraswami
Physical Research Laboratory, Navrangpura, Ahmedabad - 380009, India

Krishna Prasad Sayamanthula

Indian Institute of Astrophysics, Bangalore - 560034, India

Arul Selvam
Department of Physics, Madurai Kamaraj University, Tamil Nadu, India

S. Sujatha
M P Birla Inst. of Fundamental Research, 43/1 Race Course Road
Bharatiya Vidya Bhavan Campus, Bangalore - 560001, India

Ramya Sethuram
Indian Institute of Astrophysics, Bangalore - 560034, India

Bharat Kumar Yerra

Indian Institute of Astrophysics, Bangalore - 560034, India
 Part I

Stellar Structure, Nucleosynthesis and

Evolution of Low and Intermediate-mass Stars
Stellar Structure and Evolution:
An Introduction

C. Simon Jeffery

Armagh Observatory, College Hill, Armagh BT61 9DG, N. Ireland, UK

[email protected]

Summary. The synthesis of new elements takes place inside stars. How do stars
evolve and distribute this creation to the universe at large? This article starts with
the observables that the theory of stellar evolution aims to reproduce, and gives a
quick overview of what that theory predicts (Sects. 2–3). It presents the equations
governing stellar structure and evolution (Sects. 4–6) and the physics of stellar interi-
ors (Sects. 7–9). Approximate and numerical methods for their solution are outlined
(Sects. 10–11) and the general results of stellar structure and evolution are discussed
(Sects. 12–13). The structure and evolution of horizontal-branch stars, hydrogen-
deficient stars and other stellar remnants are also considered (Sects. 14–15).

Keywords: Stars: interiors – Stars: evolution – Stars: horizontal-branch –

Stars: AGB and post-AGB – HR and C-M diagrams – Equation of state –
Convection – Atomic Processes – Nucleosynthesis

1 Introduction

What are the stars? How do they shine? What are they made of? These
questions have challenged mankind ever since he began to explore the world
around him and appreciate the awesome splendour of the night sky. Just as
challenging are questions about what we ourselves are made of, and where we
come from. Only in the last hundred years have we started to find answers
that approach a coherent understanding of the universe we inhabit.
Fundamental to understanding the stars are measurements of distance and
brightness, colour and constancy. Any theory of what stars are and how they
behave must be able to explain these observations. Deeper insight is gained
from measurements of chemical composition and the relationships between
stars and the interstellar medium. The big story will show how elements are
manufactured by nuclear reactions deep inside the stars – nucleosynthesis –
and then transported to the stellar surface and into the interstellar medium.

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 1, c Springer-Verlag Berlin Heidelberg 2010
4 C. S. Jeffery

Table 1. The Sun

Mass: M = 1.98892(25) × 1030 kg
Radius: R = 6.9599(7) × 108 m
Surface gravity: g = 2.7397(5) × 102 m s−2
Effective temperature: Teff, = 5770(6) K
Luminosity: L = 3.826(8) × 1026 W
Surface hydrogen mass fraction: X = 0.71
Surface helium mass fraction: Y = 0.265
Surface metal mass fraction: Z = 0.025
Age: t ≈ 4.567 × 109 y

The object of these lectures is to explain the physics of stellar interiors,

to use this physics to make stellar models and hence to understand how stars
work and evolve. The lectures will demonstrate how models of stars are con-
structed, and explain how these models predict stars should evolve.
This article starts by introducing some of the fundamental observational
material (Sect. 2), and by providing an early preview of the stellar evolution
theory (Sect. 3). Fundamental timescales and the equations of stellar structure
and evolution are derived in Sects. 4–6. The micro-physics (equation of state,
opacity and nuclear physics) are discussed in Sects. 7–9. Some methods for
calculating approximate solutions and full numerical solutions are presented
(Sects. 10–11). Subsequent sections deal with the evolution of main-sequence
stars (Sect. 12), white dwarfs and supernovae (Sect. 13), horizontal-branch
stars (Sect. 14) and hydrogen-deficient stars (Sect. 15).
The text is based on a series of six lectures given at the 2008 Kodai School
on Synthesis of the Elements in Stars1 and on a more extended course given
in the University of St Andrews and Trinity College, Dublin over a period
of some twenty years. At Kodaikanal, the core material comprised four lec-
tures. Two more advanced lectures covered horizontal-branch stars (Sect. 14)
and hydrogen-deficient stars (Sect. 15). Development of the core material was
originally drawn from several seminal texts [1, 2, 3, 4, 5, 6, 7, 8, 9]. Section 15
and parts of Sect. 14 are based on [10, 11].

Variables

In considering stellar structure, we will meet a number of quantities. The most

important are:
• stellar mass M : often given in solar units M ,
• stellar radius R: often given in solar units R ,
• stellar luminosity L: often given in solar units L ,

1
Kodaikanal Observatory, Indian Institute of Astrophysics, 2008, April 29 - May
13.
 Stellar Structure and Evolution 5

Table 2. Physical Constants: CODATA 2006

speed of light in vacuum c = 2.997 92458 × 108 m s−1

gravitational constant G = 6.674 28(67) × 10−11 m3 kg−1 s−2
Planck constant h = 6.626 068 96(33) × 10−34 Js
electron charge e = 1.602 176 487(40) × 10−19 C
electron mass me = 9.109 382 15(45) × 10−31 kg
proton mass mp = 1.672 621 637(83) × 10−27 kg
Avogadro’s number NA = 6.022 141 79(30) × 1023 mol−1
atomic mass unit mu ≡ 10−3 kg mol−1 /NA kg
Boltzmann constant k = 1.380 6504(24) × 10−23 J K −1
Stefan-Boltzmann constant σ = 5.670 400(40) × 10−8 W m−2 K−4
Radiation constant a ≡ 4σ/c [J m−3 K−4 ]
Wien’s displacement constants b = λmax T = 2.897 7685(51) × 10−3 mK
b = νmax /T = 5.878 933(10) × 1010 Hz K−1

• surface gravity g: a force measured in m s−2 ,

• effective temperature Teff : usually given in degrees Kelvin.
• mass-fractions of hydrogen X, helium Y and other elements Z, respec-
tively,
• and age t: usually given in years, millions (t6 ) or billions of years (t9 ).
The flux emitted from the surface of a star is the product of the stellar surface
area (4πR2 ) and emissivity per unit area assuming that the surface radiates
as a black body:
L = 4πR2 σTeff 4
. (1)
where σ is the Stefan-Boltzmann constant. The surface gravity is simply given
by
g = GM/R2 , (2)
where G is the gravitational constant. The conservation equation for chemical
composition can be simply written
X + Y + Z = 1. (3)
In situations where the abundances of individual elements or nuclides are
important (e.g. nuclear reaction rates), relative abundances by mass will be
given as xi , where i is either the atomic number, or denotes the nuclide in some
other distinct way. Values for the Sun are given in Table 1; errors on the last
digits are shown in parentheses. Table 2 provides constants used throughout
the text and enables many equations to be evaluated.

2 The Hertzsprung–Russell Diagram

The most important correlations amongst stellar properties are contained in
a type of diagram developed independently by Ejnar Hertzsprung and Henry
6 C. S. Jeffery

Fig. 1. Hertzsprung-Russell diagram: a plot of luminosity (absolute magnitude)

against the colours of the brightest stars ranging from the high-temperature blue-
white stars on the left side of the diagram to the low temperature red stars on
the right side. Original image by Richard Powell licensed for derivative works and
redistribution under the Creative Commons Attribution ShareAlike 2.5 License

Russell [12, 13, 14]. The original Hertzsprung–Russell (HR) diagram showed
the distribution of spectral type and absolute magnitude (or brightness) for
stars with known distances (Fig. 1). The latter is required to convert an ap-
parent brightness (e.g. mV ) to an absolute magnitude (MV ). The diagram
demonstrates that stars do not appear with any combination of spectral type
and brightness, but fall on well-defined sequences, e.g. the main sequence, the
giant branch, the white dwarfs, and so on.
A more convenient form is the colour – magnitude diagram, in which either
an apparent or absolute magnitude is plotted against a photometric colour
index, being the ratio of brightness at one wavelength to that at another.
Such a diagram is particularly useful for comparing the properties of stars in
a cluster, which may be assumed to lie at approximately the same distance.
The additional supposition that all stars in a cluster are of the same age has
 Stellar Structure and Evolution 7

Fig. 2. Colour-magnitude diagram for the young galactic open clusters NGC 869
and NGC 884 (=h and χ Persei; based on [15])

important consequences for understanding stellar evolution – although the

supposition may not always be correct!
Wien’s displacement law states that there is a relation between the tem-
perature of a black body and the wavelength at which the maximum energy
is emitted: λmax = b/T . From this it was recognised that there should be a
connection between the colour of a star and its effective (or surface) tempera-
ture. With a theory of stellar atmospheres, the relationship between spectral
type, colour index and effective temperature became concrete. In addition, a
correction to account for light emitted at unobserved wavelengths allows the
apparent visual magnitude to be converted to a bolometric magnitude, and
hence to luminosity. The use of an effective temperature – luminosity diagram
is common in theoretical work (cf. Fig. 8). It is important to recognise and
understand the connections and differences between all three types of diagram.

2.1 Cluster Diagrams

Figure 1 contains stars of widely varying mass, age and composition, and can
only be constructed for stars whose distances can be measured directly. In the
case of Galactic and globular clusters, we assume that all stars formed at ap-
proximately the same time, from a gas-cloud of roughly uniform composition.
Because the cluster members are at the same distance, their relative bright-
nesses provide an HR diagram in which only the zero-point of the luminosity
axis is unknown.
8 C. S. Jeffery

Galactic or open clusters are normally associated with the hydrogen-rich

disk of the Galaxy, hence are young with metallicities (Z ≈ 0.01−0.03) similar
to the solar value. Their colour-magnitude diagrams (cf. Figs. 2 – 3) typically
show:
• Most stars on the main-sequence (MS).
• A turn-off (TO) point somewhere between F and O stars (depends on
cluster age, Fig. 3).
• A Hertzsprung gap between stars leaving the MS and Giants (Fig. 2).
• A relatively flat Giant Branch (GB) (Fig. 3).
Globular clusters are associated with the gas-poor halo of the Galaxy, and
are very old objects with generally low metallicities (Z ≈ 0.0001−0.01). Their
colour-magnitude diagrams (cf. Figs. 4 – 5) typically show:
• Only late-type stars on the MS.
• A turn-off-point around spectral-type G (depends on age and metallicity)
• A continuous sequence from the TO through subgiants up to a steep GB
• A horizontal branch (HB), red in some clusters, blue in others (usually
depends on metallicity)
• An asymptotic giant branch (AGB) sequence above the HB and parallel
to the GB

Fig. 3. Colour-magnitude diagram for the old galactic open cluster M67 (based on
[16])
 Stellar Structure and Evolution 9

Fig. 4. Colour-magnitude diagram for the metal-rich galactic globular cluster 47 Tuc
(Z = 0.004, [Fe/H] = −0.76; based on [17])

Fig. 5. Colour-magnitude diagram for the moderately metal-poor galactic globular

cluster ω Cen (Z = 0.0006, [Fe/H] = −1.62; based on [18])
10 C. S. Jeffery

Fig. 6. The mass-radius relation for both components in each of fifty eclipsing
binary stars. The primaries and secondaries are shown as squares and circles (based
on [19])

Fig. 7. The mass-luminosity relation for the same stars as in Fig. 6

 Stellar Structure and Evolution 11

2.2 The Temperature–Luminosity Relation

The most obvious empirical relation between stellar properties is provided by

the main-sequence in the HR diagram, where there exists a direct correlation
between stellar luminosity and effective temperature of the form

L ∝ Teff
α
(4)

where on average α ≈ 6, although the value is higher at both the upper and
lower ends of the main sequence.

2.3 The Mass–Luminosity and Mass-Radius Relations

Another important empirical relation between stellar properties is provided

by eclipsing binary stars, for which direct measurements of mass, radius and
luminosity can often be obtained for both components. Plotting log R and
log L against log M for main-sequence stars (Figs. 6 and 7), we find straight
lines which mean that L follows a relation

L ∝ M β. (5)

where on average β ≈ 3.8. Our theory of stellar structure must reproduce

these results.

3 Stellar Evolution – A Sneak Preview

We have seen that the luminosity of a star depends on its mass. Since the
luminosity L determines the rate at which a star uses up its available fuel,
and L goes as M β , β > 1, it is evident that stars of different mass effectively
age at different rates. The cluster colour-magnitude diagrams are essentially
snapshots of stellar evolution in which all the stars have the same age, but
cover a wide range of mass.
We cannot observe the passage of an individual star through the colour-
magnitude diagram, but we can attempt to model its evolution from birth to
death. A simplified picture of the evolution of two stars is shown in Fig. 8.
This diagram provides a convenient way to outline the basic features of stel-
lar evolution. These are common to a majority of stars, with the greatest
differences occurring at the extremes of mass.

Star Formation

At the current time in our Galaxy, it appears that star formation takes
place in massive interstellar clouds, with dimensions ≈ 10 pc, density ≈ 5 ×
109 atoms m−3 , and temperature ≈ 10 K. The Galaxy is pervaded by a mag-
netic field aligned approximately parallel to the galactic plane. This mag-
netic field is strongly tied to the ionized plasma in the interstellar medium.
12 C. S. Jeffery

Fig. 8. Schematic L − Teff diagram showing the evolution of 1 and 5 M mass stars.
(Based on a figure by J. Lattanzio. Evolution tracks computed by R. G. Izzard.)

Random turbulent process or other small perturbations in the field lead to

local potential wells where the ISM can condense, pulling the magnetic field
with it, and leading to a Rayleigh-Taylor (or runaway) instability. It is thought
that this provides the initial mechanism for the formation of dark clouds of
interstellar matter.
A number of processes including refrigeration (to keep the clouds cool) and
accretion disks (to dispose of angular momentum) enable these dark clouds
to collapse under their own gravity until they are sufficiently dense and hot
for nuclear reactions to begin. At this point, we consider a star to have been
born.
The Main Sequence
Stars spend over 90% of their lives on the main sequence. Hydrogen is con-
verted to helium in the stellar core (Fig. 9). Since four protons are slightly
heavier than one helium nucleus, the excess mass is converted into energy
which keeps the core hot, maintains internal pressure and allows the star to
shine. During main-sequence burning, the star increases its radius slightly as
the hydrogen content in the core drops. More detail will be given in Sect. 12.
The Giant Branch
As hydrogen in the core is depleted, nuclear reactions switch off, and the core
contracts. Hydrogen-rich material outside the core is compressed and heated
 Stellar Structure and Evolution 13

H H

H > He
H > He

He

H > He
He
He > C+O

H > He
He
He > C+O
C+O

H
He
C+O

Fig. 9. A simplified view of the internal chemical structure of a star during the
major phases of stellar evolution. The panels represent the main-sequence, through
the giant-branch, helium-burning, asymptotic-giant branch and white dwarf phases.
Filled circles and thick lines represent nuclear-burning regions. Not to scale

by the contraction, so hydrogen-burning shifts to a shell. In low-mass stars

(0.5 − 3M ), the outer layers of the star expand and the star becomes a red
giant. During this phase of evolution, the hydrogen-burning shell adds helium
to the core beneath it, which consequently contracts and heats. At the same
time, the shell becomes hotter, thinner and more luminous. The hydrogen-rich
outer layers (the envelope) expand, and the entire star evolves upwards along
the giant branch.
14 C. S. Jeffery

Low-mass Helium Burning

Once the helium core reaches a critical mass, nuclear burning of helium begins,
producing carbon and oxygen in a relatively long-lived phase of evolution.
The energy derived from helium-burning heats the helium core and forces it
to expand. Thus the hydrogen-burning shell may actually get weaker at this
point, the overall luminosity drops and the star contracts. At this stage, the
star may be either blue or red, corresponding to the horizontal-branches in
Figs. 4 and 5. We shall learn more about these stars in Sect. 14.

Intermediate-mass Helium Burning

In stars heavier than about 2.3M and with Z ≈ 0.02, helium burning reac-
tions start before the core becomes compact and before the hydrogen-burning
shell gets very thin. Core expansion following helium ignition is therefore mild
and the total luminosity does not drop by much. However the radius does be-
come smaller, producing a blue-loop in the Hertzsprung-Russell diagram for
the duration of core-helium burning.

Asymptotic Giant-Branch (AGB) Stars

With core-helium exhaustion, the focus of nuclear-burning in low- and inter-

mediate-mass stars shifts to a double-shell structure. Because the He-burning
shell is thermally unstable, a process known as thermal-pulsing is established.
The helium-burning shell burns faster, and so keeps going out. When sufficient
fresh helium has accumulated below the hydrogen-burning, the helium-shell
reignites, the intershell region is forced to expand, and hydrogen-burning is
extinguished. The interaction between these thermal pulses and convection in
different layers of the star makes this one of the most important processes in
cosmology; more detail is presented by Karakas (these proceedings).
Of consequence here is that when the hydrogen-burning shell is operating,
it is very thin and very powerful, so the star becomes very luminous and very
large. The hydrogen envelope is thus consumed from below (by the shell) and
blown away from the surface by a strong stellar wind.

Planetary-Nebula Formation and White-Dwarf Cooling

Ultimately, the expansion of the AGB star results in a dynamical instability

which causes the expulsion of the outer layers as a planetary nebula. Unable
to sustain nuclear reactions, whatever is left of the envelope starts to contract.
When it has cooled sufficiently the star becomes a white dwarf: Sect. 13. We
will discuss whether the white dwarf stage is the end of the story in Sect. 15.

Massive Stars

Following main-sequence burning, the ultimate evolution of stars more massive

than ≈ 8M is quite different. Initial evolution is similar, the star expands
 Stellar Structure and Evolution 15

rapidly and starts to cross the HR diagram towards the red giant region.
For M ≥ 20 M , core helium ignition occurs at a progressively earlier stage
(i.e. closer to the main-sequence) as the mass increases. Core-helium burning
arrests the redward evolution until core helium is exhausted, and redward evo-
lution resumes. As the inactive carbon-oxygen core contracts, it is sufficiently
massive that its temperature increases to the point where carbon-carbon re-
actions can occur. There follows a rapid sequence of nuclear-burning episodes
producing elements with increasing atomic weight. Hydrostatic equilibrium
is maintained until the core consists primarily of iron (56 Fe), at which point
further nuclear reactions require more energy than they release. The stellar
core collapses and a supernova explosion follows. This explosion is responsible
for the production of quantities of heavy nuclei quite distinct from those cre-
ated in AGB stars. These processes are discussed in greater detail by Arnould
(these proceedings).

Binary Stars

This very simple description applies to those stars which evolve as single stars
or as members of a wide binary system which do not interact. It is increasingly
clear that a large fraction of stars are born in binary or multiple systems in
which two stars exchange material at some point during their evolution. The
possibilities of what can happen thereafter are too numerous to be able to
cover here, but some of the more bizarre possibilities will be considered later.

4 Stellar Time Scales

We want now to put some physical argument behind our cartoon picture of
stellar evolution. It is evident that stars such as the Sun do not change their
properties rapidly. It is important to understand and appreciate the relative
magnitudes of the timescales on which a star can change.

4.1 Dynamical (Free-Fall) Time

Consider a star as a ball of gas which is held up by the pressure forces within
it. How long would the star survive if the pressure forces were removed? The
answer is known as the free-fall or dynamical time (tff or tdyn ). A simple es-
timate is given by the time required for a body to fall through a distance of
the order R (the stellar radius) under the influence of a (constant) gravita-
tional acceleration equivalent to the surface gravity g = GM/R2 of a star of
mass M . From Newton’s 2nd law,
1 2 1 GM 2
R= gtff = t
2 2 R2 ff
⇒ tff ≈ 2.2 × 103 (R3 /M )1/2 s
16 C. S. Jeffery

where R and M are in solar units. Writing the mean density as ρ =

(3/4π)M/R3 , we also have:

tff ≈ (2πGρ/3)−1/2 . (6)

tff is also the characteristic time for a significant departure from hydrostatic

equilibrium to alter the state of a star appreciably, the time taken for a body
orbiting at the surface of the star to make one complete revolution, and the
time for a sound wave to propagate through the star.

4.2 Thermal (Kelvin) Time

Next, consider a star as a ball of hot gas which acts as reservoir of heat
energy. How long would it take for this energy to radiate away if it were
not replenished? This is the thermal or Kelvin-Helmholtz timescale (tK or
tth ). If the total kinetic (thermal) energy of the star is Ekin , the timescale is
approximately
Ekin
tK = . (7)
L
We shall see later that Ekin is related to Egrav , the gravitational binding
energy of the star, by the Virial theorem:
1
Ekin = − Egrav .
2
Since we can write
 M
Gm GM 2
Egrav = − dm = −q (8)
0 r R

where q is a dimensionless constant of order unity2 , then

q GM 2 qM 2
tK = ≈ 3 × 107 y (9)
2 LR LR
where M , L and R are in solar units.
The thermal time is the relaxation time for departure of a star from ther-
mal equilibrium. It is also the time that would be required for a star to con-
tract from infinite dispersion to its present radius if L were to remain constant
during its entire contraction.

2
q = 3/5 for a sphere of uniform density, and becomes smaller with increasing
central condensation.
 Stellar Structure and Evolution 17

4.3 Nuclear Time

It is clear, for example from radio-carbon dating of rocks in the earth’s surface,
that the solar system must be very much older than the “Kelvin age” of
≈ 3 × 107 years. It is now taken for granted that the main source of stellar
energy comes from nuclear reactions. The fusion of four protons (hydrogen
nuclei) to an alpha-particle (helium nucleus) is associated with the release of
energy Q, where Q ≈ 26 MeV. The total available energy is thus

Enuc = qM/4mp Q,

where again q is a dimensionless factor of order unity representing the fraction

of the star available as nuclear fuel. The nuclear time is simply the time taken
to radiate this energy
Enuc
tnuc = . (10)
L
For hydrogen-burning in main-sequence stars,
M
tnuc ≈ 1 × 1011 q y. (11)
L
We adopt q = 0.1 for Table 3.

N
r
λ 1
R

Fig. 10. Diffusion processes in radiative energy transport

4.4 Diffusion Time

Energy released at the centre of a star is used to heat the stellar material
and hence oppose gravitational forces, otherwise collapse would be inevitable
and hydrostatic equilibrium could not be maintained. Energy liberated as
photons cannot therefore leave a star immediately, but must interact with the
18 C. S. Jeffery

stellar material by a series of scattering collisions, mainly with electrons. Since

scattering is an isotropic process, energy transport is most correctly described
by the diffusion equation. Section 5 will examine this further, but essentially
the photon-path can be described by a random-walk consisting of N steps,
each of length λ. Whilst the total distance travelled by√the photon is N λ, it
may be shown that the nett distance travelled is r = N λ, because of the
scatterings. Thus to escape from the star, the photon must travel a distance
R, which it will do in a time

R2
tdiff ≈ ≈ 5 × 105 R y (12)
λc
(for solar-type stars) which may be compared with the escape time for non-
interacting particles (such as neutrinos):

tesc = R/c = 2.3R s (13)

with R in solar units.

Table 3. Comparative timescales

Star M R L Teff tesc tff tdiff tkin tnuc

M R L K s−1 s−1 y−1 y−1 y−1
Sun 1 1 1 5760 2.3 2.2 103 1.7 105 3.0 107 1.0 1010
MS 5 3 450 15342 6.9 2.6 104 1.5 106 5.5 105 1.1 108
MS 10 5 6 300 22958 12 7.8 104 4.3 106 9.5 104 1.6 107
MS 0.2 0.3 0.0022 2280 0.69 1.6 102 1.5 104 1.8 109 9.1 1011
RG 1 100 1 500 3600 230 2.2 106 1.7 109 2.0 102 6.6 106
WD 0.6 0.01 0.001 10243 0.023 1.7 100 1.7 101 1.1 1012

t −1 t −1 t −1 t −1 t −1
MS 5 3.0 12 9.0 100 1.8 10−2 1.1 10−2
MS 10 5.0 35 2.5 101 3.2 10−3 1.6 10−3
MS 0.2 0.3 0.073 9.0 10−2 6.0 101 9.1 101
RG 1 100 1 000 1.0 104 6.6 10−6 6.6 10−4
WD 0.6 0.001 0.000 77 1.0 10−4 3.6 104

4.5 Comparative Timescales

Table 3 shows timescales for main-sequence stars of four different masses, a

red giant and a white dwarf, using an estimated radius and Eqs. (1), (5), (6),
(9), (11), (12), and (13). The upper block shows conventional units of time;
the lower block shows timescales relative to that of the Sun. The luminosities
for the red giant and white dwarf are also approximations.
 Stellar Structure and Evolution 19

5 Equations of Stellar Structure

The simplest picture of a star is that of an isolated body of gas sufficiently mas-
sive that the only significant forces are self-gravity and internal pressure. In the
simplest case, we can assume spherical symmetry and neglect the influence of
rotation, magnetic fields and external gravitational influences (in fact, under
these assumptions, equilibrium solutions will enforce spherical symmetry).

r=R

r=0
X,Y,X
Center Surface

0,0,Pc,Tc

m,l,P,T

M,L,0,Teff

Fig. 11. The variables of stellar structure

Consider a spherical system containing a fluid of mass M and radius R

(Fig. 11). The internal structure of this system is described by the distribution
of various quantities, including
• r: radius
• m(r): mass of fluid contained within a spherical shell of radius r
• l(r): luminosity or total energy passing through a shell of radius r
• T (r): temperature of the fluid at radius r
• P (r): pressure in the fluid at radius r
• ρ(r): fluid density at radius r
These quantities are expressed as a function of radius r, running from r = 0
at the stellar centre to r = R at the stellar surface. However, we could have
equally chosen a different independent variable, for example mass, m. In order
to keep the notation simple, stellar structure variables will usually be written
in short form, i.e. m ≡ m(r), l, P, T, ρ. The problem of stellar structure is
then to determine the run of these quantities throughout the star, for which
we require a system of equations.
20 C. S. Jeffery

ρ δr

Fig. 12. Derivation of the density equation

5.1 Mass Continuity

Consider a spherical shell of radius r, thickness δr (δr  r) and density ρ

(Fig. 12). It has a mass (volume × density)

δm = 4πr2 ρδr.

In the limit δr → 0,
dm
= 4πr2 ρ. (14)
dr
The same argument provides the mass of the volume enclosed within the
sphere of radius r  r
m = 4π r2 ρdr.
0

r
δr

g
P P + δP

Fig. 13. Gravitational and pressure forces in hydrostatic equilibrium

 Stellar Structure and Evolution 21

5.2 Hydrostatic Equilibrium

We must next consider the forces acting at any position within this sphere
(Fig. 13). Consider an element of material at radius r. The sphere of radius
r acts as a gravitational mass situated at the centre giving rise to an inward
gravitational force
Gm
g= 2 . (15)
r
If there is a pressure gradient through the sphere, there will be an additional
force. If the element has a thickness δr and cross-section δA, then a net force
arises if the pressure on the inner and outer surfaces are unequal, so that the
inward force is, since δP = (dP/dr)δr,
 
dP dP
P (r) + δr − P (r) δA = δrδA.
dr dr

But the mass of the volume element is δm = ρδrδA, then the sum of inward
forces due to gravity and pressure will be
 
1 dP d2 r
F = δm g + = −δm 2 . (16)
ρ dr dt

For the element considered in Fig. 13 to be in equilibrium, the parenthesis in

Eq. (16) must be zero, and by substitution from Eq. (15),

dP Gmρ
=− 2 . (17)
dr r
This is the equation of hydrostatic equilibrium which, amongst other things,
demonstrates that in order to oppose gravity, pressure must increase towards
the centre of the star.

5.3 Virial Theorem

A very important result can be obtained when an entire self-gravitating sys-

tem is in equilibrium, i.e. if Eq. (17) is satisfied at all r, since it becomes
possible to derive a simple relation between average internal pressure and the
gravitational potential energy of the system. This is done by multiplying both
sides of Eq. (17) by 4πr3 and integrating from r = 0 to r = R to obtain
 R  R
dP Gm
4πr3 dr = − 4πr2 ρdr.
0 dr 0 r

Integrate the lhs by parts (du = dP/dr.dr, dv = 4πr3 ), and substitute dm =

4πr2 ρdr
 R  M
 R Gm
4πr P 0 − 3
3
4πr P dr = −
2
dm.
0 0 r
22 C. S. Jeffery

Since P (R) = 0, the first term is zero. Substituting 4πr2 dr = dv

 V  M
Gm
−3 P dv = − dm.
0 0 r

The lhs becomes −3

P V where
P  is the volume-averaged pressure and V
is the volume of the system, and the rhs is simply the gravitational potential
energy of the system Egrav . Thus the average pressure needed to support a
system with gravitational energy Egrav and volume V is given by

1 Egrav

P  = − . (18)
3 V
The statement that the average pressure needed to support a self-gravitating
system is one third of the stored gravitational energy is called the Virial
Theorem. The physical meaning of pressure depends on the system itself, but
it can be applied to clusters of galaxies as well as to individual stars. Let us
consider two cases for gas in a star, where the equation of state relates the
pressure of the gas (P ) to the translational kinetic energy of the gas particles
(Ekin ).

Non-Relativistic Gas

For a gas of non-relativistic particles, since P = nkT = kT /V and Ekin =

3
2 kT , then
2
P = Ekin /V. (19)
3
Then for a self-gravitating system of volume V and gravitational energy Egrav ,
the gravitational and kinetic energies are related by

2Ekin + Egrav = 0

and the total energy of the system, which is the sum of kinetic and gravita-
tional energies, is given by

Etot = Ekin + Egrav = −Ekin ,

and
Etot = Egrav /2. (20)
These equations are of fundamental importance. Note the following. If a sys-
tem is in hydrostatic equilibrium and it is tightly bound, the gas particles
have high kinetic energy. They are hot. If the system evolves slowly, close
to hydrostatic equilibrium, then changes in kinetic and gravitational energies
are simply related to changes in the total energy. We shall see some of the
consequences later in considering the collapse of gas clouds.
 Stellar Structure and Evolution 23

Ultra-Relativistic Gas

For a gas of ultra-relativistic particles, Ekin = 3kT and hence

1
P = Ekin /V (21)
3
from which we find
Ekin + Egrav = 0. (22)
Thus hydrostatic equilibrium is possible only if the total energy is zero. As
the ultra-relativistic limit is approached, e.g. as the gas temperature increases,
the binding energy decreases and the system is easily disrupted. This type of
instability occurs in stars where the radiation pressure is very high (pressure
comes from photons which are, by definition, relativistic) e.g. supermassive
stars, or where pressure comes from degenerate electrons and the electron
temperature become very energetic, e.g. massive white dwarfs. Again, we shall
examine these later.

Jean’s criterion for gas-cloud collapse

An important corollary to the Virial theorem is connected with the condition

for gravitational collapse. Stars are believed to form when large clouds of
gas and dust contract. However, this contraction can only take place if the
internal pressure of the cloud due to the thermal motion of the cloud particles
is weaker than the gravitational forces compressing the cloud. Formally, a
cloud is bound if the total energy is less than zero. Knowing the gravitational
binding energy (Eq. (8)) and adopting q = 1, and setting the thermal energy
per particle to 32 kT , the total energy of the cloud is:

3 M GM 2
Ekin + Egrav = kT + .
2 μmH R
For the total system to be bound,

Ekin + Egrav < 0

implies
3kT
Mcloud > MJ = R, (23)
2GμmH
or
2 2GμmH M
Tcloud < TJ = , (24)
3 k R
or  3
3 3kT
ρcloud > ρJ = , (25)
4πM 2 2GμmH
where the critical values are known as the Jeans mass, temperature and
density.
24 C. S. Jeffery

Equation (24) illustrates an important consequence of cloud contraction; if

the system is to remain bound, i.e. if it is to continue to contract and become
a star, then the internal temperature of the collapsing cloud or protostar,
M
TI ∝ . (26)
R

l+δ l
ρ δr
r

δm
l

Fig. 14. Derivation of the energy equation

5.4 Energy Conservation

Since energy moves through the star, the consequences for stellar structure
imposed by the law of conservation of energy must be considered, whether that
energy comes from the release of nuclear or gravitational energy. Consider a
volume element dv = 4πr2 dr at radius r (Fig. 14).
The law of conservation of energy states that the total amount of energy
leaving a volume element dv must equal the total amount of energy entering
that element plus the energy lost or produced within that element.
Assuming spherical symmetry, it is convenient to consider a shell of mass
δm. Let l be the amount of energy entering the bottom of shell, and l + δl be
the amount of energy leaving the top of the shell. Let us denote the energy
absorbed or produced within the shell, per unit mass, by ε. Then we can
equate energy out to energy in plus energy produced or lost:

l + δl = l + εδm,

whence
dl
=ε (27)
dm
or, with Eq. (14),
 Stellar Structure and Evolution 25

dl
= 4πr2 ρε. (28)
dr
There are several sources/sinks of energy which we shall consider else-
where, including:
• nuclear, e.g. 4p+ 2e− → α+ 2e+ + 2e− + 4.768 × 10−29 kg → α+ 26.72 MeV
• neutrino, e.g. γ + e− → e− + ν + ν̄
• nonadiabatic expansion, i.e. dQ = T dS = dE + P dV = 0

5.5 Energy Transport

The fact that there is a temperature difference between the interior and surface
of the Sun implies that there must be a temperature gradient and hence a
flux of energy. Conservation of energy constrains the energy flux, but the way
energy is transported establishes the temperature gradient. There are three
major transport mechanisms:
• radiation: generally by photons, but also by neutrinos.
• convection.
• conduction: by electrons in white dwarfs.

Radiative Equilibrium

Energy transported by radiation obeys Fick’s law of diffusion:

daT 4
F = −D
dr
where F is the flux density, aT 4 is the radiation energy density3 and D is a
diffusion coefficient. D is related to the opacity κ by

D = c/κρ.

F must be multiplied by 4πr2 to obtain the total flux l, whence

4πr2 c daT 4
l=−
κρ dr
dT 3κρ l
⇒ = . (29)
dr 4acT 4πr2
3

Equation (29) represents the temperature gradient within the star when
the star is in radiative equilibrium. Combining with Eq. (17) (hydrostatic
equilibrium) and taking logs, we define a gradient
d ln T 3 κlP
= ∇rad = (30)
d ln P 16πac GmT 4

3
a = 4σ/c is the radiation constant.
26 C. S. Jeffery

P2* , ρ2* P2 , ρ2

δr

P1* , ρ1* P1, ρ1

Fig. 15. Convective energy transport

Convective Equilibrium

An element of gas is at some radius r. Consider its upward displacement by

a distance dr, allowing it to expand adiabatically until the pressure within
is equal to the pressure outside (Fig. 15). Then release the element. If it
continues to move upwards, the layer in question is convectively unstable and
convective motions will persist.
Let the pressure and density within and without the sphere be denoted by
P ∗ , ρ∗ , and P , ρ respectively. Then initially

P1∗ = P1 and ρ∗1 = ρ1 .

After the perturbation, P2∗ = P2 and

 1/γ
P2∗
ρ∗2 = ρ∗1 ,
P1∗

where P ρ−γ is a constant and γ = 5/3 for a highly-ionized gas. For radiative
equilibrium, we require ρ∗2 > ρ2 so that the net force (bouyancy+gravitation)
is downwards and the element will return to its starting position. Eliminating
asterisks and writing P1 = P2 + dP , we obtain
P dρ 1
> (31)
ρ dP γ
for radiative equilibrium. This condition is related to the temperature gradient
assuming some equation of state (e.g. P = ρkT /μm) so that Eq. (31) becomes

d ln T γ−1
≡ ∇ < ∇ad ≡ . (32)
d ln P γ
There are two main circumstances under which Eq. (32) fails:
 Stellar Structure and Evolution 27

1. In the centre of main-sequence stars, the radiation flux l/4πr2 can be-
come very large, whilst κρ remains small. Thus the temperature gradi-
ent d ln T /d ln P required for radiative equilibrium (Eq. (30)) becomes
large, and the material becomes convectively unstable. This gives rise to
nuclear-driven convective cores in massive stars, and also convective zones
in helium-burning stars.
2. In ionisation zones, the adiabatic exponent γ becomes smaller approaching
unity, and κ becomes large, so that radiative equilibrium may be violated
for small values of the temperature gradient. This gives rise to opacity-
driven convection in the envelopes of cool stars.

Mixed Equilibrium

Equations (30) and (32) give the temperature gradient ∇ in radiative and
convective equilibrium respectively. Where there are situations when it is clear
which equation to use, the situation frequently arises when material may be
naturally convective, but the convective efficiency ξ is sufficiently low that
radiation carries a substantial fraction of the flux. In these cases, ξ must be
derived from a suitable theory of convection.
With this in mind, it is useful to rewrite the equation for the temperature
gradient
d ln T
= ∇ = (1 − ξ) ∇rad + ξ∇ad (33)
d ln P
such that:
• ξ = 0 : radiative equilibrium
• 0< ξ <1: non-adiabatic convection
• ξ = 1 : fully adiabatic convection

5.6 The Equations of Stellar Structure

We have now derived the four basic (time-independent) equations of stel-

lar structure. These are mass continuity (Eq. (14)), hydrostatic equilibrium
(Eq. (17)), conservation of energy (Eq. (28)), and energy transport (Eq. (33)).
These form a set of coupled first order ordinary differential equations relating
one independent variable, e.g. r, to four dependent variables i.e., m, l, P, T ,
which uniquely describe the structure of the star, note that any variable could
be used as the independent variable. In an Eulerian frame, the spatial coordi-
nate r is the independent variable. For most problems in stellar structure and
evolution it is usually more convenient to work in a Lagrangian frame, with
mass as the independent variable. Transforming, we obtain:
dr 1
= ,
dm 4πr2 ρ
dl
= ε,
dm
28 C. S. Jeffery
dP Gm
=− ,
dm 4πr4
dT Gm T
=− ∇.
dm 4πr4 P
Boundary Conditions

To solve such a system, boundary conditions are required. In the Lagrangian

frame, the boundaries are at m = 0 (centre) and m = M (surface). In the
centre, the enclosed mass and luminosity are defined:

r(m = 0) = 0, (34)

l(m = 0) = 0, (35)
and at the surface temperature and pressure are defined:

T (m = M ) = Teff = (L/4πR2 )1/4 , (36)

Pgas (m = M ) = 0. (37)

Constitutive Relations

In order to close the system of four first order o.d.e.s and four boundary condi-
tions, additional equations are required to relate the principal state variables
to the local micro-physics. Thus ε, ρ, ∇ refer to energy generation, density
and energy transport, the last depending on ξ and κ, the convective efficiency
and opacity and may be expressed in terms of P , T and the local composition
of the stellar material X, Y, Z, or xi . These relations will be derived more
explicitly in Sects. 7–9.

Equation of state: ρ = ρ(P, T, xi ).

Nuclear energy generation: ε = ε(ρ, T, xi ).
Opacity: κ = κ(ρ, T, xi ).
Convective efficiency: ξ = ξ(ρ, T, xi ).
Energy transport: ∇ = ∇(ρ, T, ξ, κ, xi ).

Solution

A quantitative physical description of the stellar interior is provided by a

solution to these equations, i.e. by the run of P, T, m and l as a function of
r throughout the domain 0 < r < R, the radius of the star. These solutions
are characterised uniquely by M = m(R), the total mass of the star, and by
xi (r), the run of chemical composition through the star. Various approximate
and numerical schemes will be examined in Sects. 10–11.
 Stellar Structure and Evolution 29

6 Equations of Stellar Evolution

It will have been seen (Sect. 5) that the equations of stellar structure can be
constructed independent of time t. Solutions refer exclusively to stars in com-
plete equilibrium – they are solutions of stellar structure and not of evolution.
However, the fact that nuclear reactions change the chemical composition of
a region where nuclear energy is produced means that the structure of a star
must change in time. There are three ways in which t enters the equations of
stellar structure.

6.1 Thermal Expansion (Contraction)

If a region of a star contracts or expands it will release or absorb energy

according to the first law of thermodynamics, which states

dE = −T dS = −(dU + P dV ).

Hence the energy equation (Eq. (28)) needs to incorporate additional terms:

dL dS
= εnuc − T . (38)
dm dt

6.2 Nucleosynthesis

Nuclear reactions produce energy and transmute elements from one species to
another. A reaction i(j, k)l destroys species i and j and creates species k and
l at a rate rij . The rate of change of abundance of species i will then be given
(in mass fraction) by
dxi
= −(1 + δij )rij mu (39)
dt
where the bracket allows for more than one nucleus of species i to be destroyed
in a reaction, and mu is the mean atomic mass unit. Thus for the p-p chain:
dX
= −4rpp mu ,
dt
dY dX
=− .
dt dt

6.3 Mixing

Convection provides an efficient transport mechanism for both energy and

matter. Stellar evolution calculations have traditionally assumed that a fully
convective region is fully mixed.
However, the student should be aware that, when dealing with timescales
which are short compared with the convective turnover time, or with layers
30 C. S. Jeffery

that are close to convective boundaries, the assumption of complete mix-

ing may break down. There is currently much interest in three-dimensional
modelling of convection regions, in understanding how convection cells be-
have when they overshoot a convective boundary, and in the behaviour of
boundaries between materials of significantly different mean molecular weight
(thermohaline mixing) [20].
The consequences of mixing, whatever the mechanism, for stellar evolution
calculations is that nucleosynthesis may transform the mixture in part of a
convection zone. Alternatively a convective boundary may move to incorpo-
rate material of different composition. The new material must then be mixed
throughout the convection zone. After a time interval δt, the new composition
(xi ) in the convection zone (cz) will be given by
   
 dxi (m)

xi  = xi (m) + δt dm/ dm. (40)
cz dt cz

7 Equation of State
We adopt the premise that matter inside stars consists of an almost perfect
gas. The properties of a gas are often referred to as state variables. The macro-
scopic properties of a gas are described completely by three quantities, i.e.
any three of pressure, density, temperature, entropy, and number density.
An equation of state (EOS) describes relations between these properties,
e.g. pressure, density and temperature. The form of the equation of state
depends on whether the fluid particles can be treated as classical or non-
classical, relativistic or non-relativistic.

7.1 Gas Laws

In 1662 Robert Boyle4 performed a series of experiments which showed that
gas volume varies inversely with pressure:
P V = constant.
In 1787, the French physicist Jacques Charles found that oxygen, nitrogen,
hydrogen, carbon dioxide and air expand to the same extent over the same 80
degree change in temperature. In 1802, Joseph Louis Gay-Lussac published
results of similar experiments indicating a linear relationship between volume
and temperature:
V /T = constant.
At about the same time (1801), John Dalton observed that the total pres-
sure exerted by a gaseous mixture is equal to the sum of the partial pressures
of each individual component:
4
born at Lismore Castle, Munster, Ireland, as the seventh son and fourteenth child
of Richard Boyle, the “Great Earl of Cork”
 Stellar Structure and Evolution 31

Ptotal = P1 + P2 + . . . + Pn .

In 1834, Emile Clapeyron combined Boyle’s and Charles’ laws into the first
statement of the ideal gas law initially formulated as

P V = R(TC + 267)

where the temperature is expressed in degrees Celsius. With the constant

being improved to 273.15 and temperature being measured in degrees Kelvin
(after the Belfast born William Thompson),

P V = RTK ,

where R = NA k is the ideal gas constant.

7.2 Pressure

There are important differences between fluid deep within a star and a labo-
ratory gas. First, stellar material is ionized, i.e. it is a plasma, which allows
greater compression and interatomic distances as small as ≈ 10−15 m to be
achieved, compared with ≈ 10−10 m for neutral gases. Second, it is in ther-
modynamic equilibrium with radiation; the radiation intensity is governed by
Planck’s law. Third, the fluid particles may be non-classical and/or relativis-
tic. Therefore the effects of quantum mechanics and special relativity must be
considered.
Pressure due to a classical or quantum gas is physically expressed by col-
lisions between the gas particles and the gas boundary – thus it is directly
related to the kinetic energy of the particles. Consider a box of dimension
D containing N particles, each having velocity v and momentum p = mv.
Assuming that collisions with the boundaries are elastic, each collision im-
parts momentum 2pz , and occurs at a rate vz /2D. The rate of transfer of
momentum to unit area is then pz vz /D3 , and averaging over all particles,
N
P =
pz vz  = n
pz vz .
D3
If the gas is isotropic,

px vx  =
py vy  =
pz vz  =
p · v/3
n
⇒P =
p · v. (41)
3
In general, the relation between the energy, momentum and velocity of a
particle is
ε2p = p2 c2 + m2 c4 , v = pc2 /εp ,
In the non-relativistic limit: p  mc, so εp = mc2 + p2 /2m and v = p/m.
In this case, p · v = mv 2 and
32 C. S. Jeffery

n 2 1 2
P = mv 2 = n( mv 2 ) = Ukin (42)
3 3 2 3
where Ukin is the translational kinetic energy density of the particles. In the
ultra-relativistic limit: p  mc, so εp = pc and v = c. Hence p · v = pc2 and
1 1
P = n(pc) = Ukin . (43)
3 3

7.3 The Classical Ideal Gas

From kinetic theory, the pressure of an ideal gas can have the form

Pgas = nkT (44)

where n is the number density of particles, and k is Boltzmann’s constant.

With Eq. (42), we obtain for a non-relativistic ideal gas the mean energy per
particle
3
Ekin = Ukin /n = kT. (45)
2
The number density is related to the mass density ρ via the mean particle
weight (sometimes mean molecular weight) μ
ρ
n= .
mu μ

7.4 Mean Mass per Particle

The mean mass per particle is essentially a reduced mass. That is, for a
mixture of elements with atomic mass ak , and mass fraction xk , the mean
mass per atom is xk
μ−1
a = .
ak
k

By contrast, the mean mass per particle depends on the number of free elec-
trons per atom. Assuming that the material is fully ionized, each atom having
zk electrons,
xk (1 + zk )
μ−1 = . (46)
ak
k

It is often useful to know the mean mass per electron, which can be computed
similarly or from the identity

μ−1 = μ−1 −1
e + μa .

Consider some examples:

• neutral hydrogen: μ = μa = 1.
• ionized hydrogen: μ = μa /2 = 1/2.
 Stellar Structure and Evolution 33

• a mix of neutral hydrogen and helium with mass fractions X and Y:

X Y
μ−1 = μ−1
a = + .
1 4
• a fully-ionized hydrogen-helium-metal mix (X,Y,Z):

2X 3Y
z + 1Z
μ−1 = + + .
1 4
a

Assuming that for metals,

a ≈ 2
z and
z  1, and rearranging
 −1
3Y Z
μ= 2X + + .
4 2
If the gas is partially ionized, which is the case close to the stellar surface,
then it is necessary to solve the Saha equation to determine the distribution
of ions for every element.

7.5 Degenerate Electron Gas

The equation of state described in Sect. 7.3 is valid when the gas can be
described as classical, i.e. when the average separation of particles is large
compared with the de Broglie wavelength5 . As the average separation be-
comes small, the lightest particles will experience a breakdown of classical
physics. This occurs when the particle density n significantly exceeds the crit-
ical “quantum concentration”
 3/2
2πmkT
nQ =
h2

(in the non-relativistic limit), which is another way of stating the requirement
that the average occupation of any quantum state should be small. In the
regime where quantum effects dominate ne  nQ , which is equivalent to the
condition that
2/3
h2 ne
kT  ,
2πme
5
de Broglie wavelength: a particle of mass m moving with a velocity v will under
suitable conditions exhibit the characteristics of a wave with wavelength

λB = h/mv = h/p.

Compton wavelength: when the particle is moving relativistically, the de Broglie

wavelength may be written

λC = h/mc = 2.4 × 10−12 m

34 C. S. Jeffery

whence it may be seen that a quantum gas is a cold gas. The standard of
coldness is set by the density, not the temperature. A cold gas is described
as degenerate because the particles occupy the lowest possible energy states
and (for electrons), obey the Pauli-exclusion principal. The energy of the most
energetic electrons in a cold electron gas is known as the Fermi energy (εF );
the momentum of these particles is the Fermi momentum pF , which has the
unique property6
 1/3
3ne
pF = h.
8π
The equation of state for a degenerate gas is obtained by evaluating the
internal energy. Substituting εp = mc2 + p2 /2m into an integral over quantum
states gives for the non-relativistic case pF  me c
 
3p2F
U = m e c2 + .
10me

Since for an ideal gas PNR = (2/3)Ukin (Eq. (42)), and substituting for pF
and ne = ρ/mu μe ,
 5/3  2/3
ne p2F ρ h2 3
Pe = = KNR , KNR = . (47)
5me mu μe 5me 8π

Similarly, for an ultra-relativistic degenerate electron gas, εp = pc gives

3
U= pF c
4
and from PUR = (1/3)Ukin,
 4/3  1/3
ne pF c ρ hc 3
Pe = = KUR , KUR = . (48)
4 mu μe 4 8π

7.6 Photons

Thermal radiation may be characterized as a photon gas, consisting of zero-

mass bosons with zero chemical potential. Without proof, the photon number
density may be written [6, Sect. 2.3],
 3
k
n = bT 3 , b = 2.4048π = 2.03. × 107 K−3 m−3
hc

The internal energy density is

6
The Fermi momentum implies that the de Broglie wavelength of the most ener-
−1/3
getic electrons in a degenerate gas is comparable with ne , the average distance
between electrons.
 Stellar Structure and Evolution 35

8 π5 k4
Ur = aT 4 , a= = 7.565 × 10−16 J K−4 m−3 (49)
15 (hc)3
and the pressure due to this radiation is

Pr = U/3 = aT 4 /3. (50)

PR = aT 4/3 P=1/ 3 E

ρ = n/ NAμ
8
logT / K

P = nkT
6

P = KNRn 5/3 P = KUR n 4/3

4
P = 2/3E

−4 0 4 8 12
log ρ/ kg m−3

Fig. 16. A map of density-temperature space illustrating the different regimes for
the equation of state, including the classical, relativistic, and degenerate limits

7.7 Total Pressure

The total pressure is the sum of the gas and radiation pressures, and the gas
pressure can be written as the sum of ion and electron pressures:

Pt = Pg + Pr = Pi + Pe + Pr

Stability

Recall that the internal temperature of a star is proportional to its gravi-

tational potential energy, and hence TI ≈ M/R (Eq. (26)). Meanwhile the
particle density n ≈ M/R3 . Thus if the particles form a classical gas, the
ratio of gas to radiation pressure
36 C. S. Jeffery

Pr aTI4 /3 TI3 M 3 /R3

= ∝ ∝ ∝ M 2.
Pg (ne + ni )kTI ne + ni M/R3

For increasing mass, radiation pressure becomes increasingly important.

Which equation?

The exact form of the EOS to use depends on whether the gas particles are
classical or degenerate (depends on T /n), and on whether they are relativistic
or non-relativistic (related to n or T ). Figure 16 illustrates which EOS to
apply in the temperature-density space typically found in stars.

bound−bound bound−free free−free

h ν = 1/2m(v22−v12)

1/2mv 2
χion

n=3
h ν = χ3 − χ2
n=2
Excitation
energy

h ν = χ2 h ν = χion + 1/2mv 2

0 n=1

Fig. 17. The principle photon-electron interactions contributing to radiative opacity

in stellar interiors

8 Stellar Opacity
The equation for energy transfer (Eq. (33)) includes a term κ, representing
the opacity of stellar material to the flow of energy by radiation or conduc-
tion. Roughly speaking, it represents the ability of stellar material to absorb
radiation. It is equivalent to the inverse of its heat conductivity. In order to
 Stellar Structure and Evolution 37

evaluate the total radiative opacity, it is necessary to consider all of the mi-
croscopic processes whereby radiation at a specific frequency ν is absorbed,
and how these processes can be combined as a single macroscopic quantity.
There are four basic processes involved in the interaction of photons with
atoms.
1. bound-bound absorption – bb
2. bound-free absorption – bf
3. free-free absorption – ff
4. electron scattering – es
The first three represent true absorption because in each case a photon is
absorbed. The fourth only changes the direction of a photon. The contribution
from thermal conduction by electrons must also be considered.
In discussing the absorption processes, consider the energy level diagram
for a simple atom (Fig. 17).

8.1 Bound-Bound Absorption

The bound energy levels of an ion lie at discrete excitation energies χexc above
the lowest-lying level (or ground state). Bound electrons may only occupy
these levels, and when moving between them absorb (or emit) photons at a
precise frequency (spectral line) given by

E = hν12 = χ2 − χ1 .

If a photon of frequency ν12 collides with an ion where an electron occupies

level 1, that electron may be excited into level 2, with a probability B12 . This
transition probability (or oscillator strength) is effectively the bound-bound
absorption coefficient for a single line, abb (ν12 ) = B12 . When multiplied by
the occupation number of each level 1, and summed over transitions between
all levels in all ions, the total monochromatic bound-bound opacity is obtained

κbb (ν) = N1 a12 (ν). (51)
ions 1 2

Bound-bound absorption is very important in stellar atmospheres and stellar

envelopes, where many millions of lines, particularly from iron-group elements,
have been shown to have a profound influence on stellar models. The frequen-
cies which dominate the opacity must correspond roughly to the peak of the
Planck function, i.e. νmax /T = b (Wien’s law again). Most atomic transi-
tions have wavelengths longer than 100 Å, so for T > 5 × 106 K, bound-bound
absorption becomes less important.
38 C. S. Jeffery

8.2 Bound-Free Absorption

At some excitation energy χion , an electron is no longer bound to the parent

ion and the levels which it may occupy are no longer discrete but form a con-
tinuous sequence or continuum of states. A continuous spectrum of absorption
can be found only if the final state corresponds to a free electron. If the initial
state is bound, then we have bound-free absorption, or photoionization. The
kinetic energy of a free electron is simply E = 1/2mev 2 . The energy involved
in a bound-free transition from level n is

E = χion − χn + 1/2mev 2 .

If a photon has energy E = hν > χion − χn , it will be absorbed with a prob-

ability abf (n, ν) and will ionize the parent atom or ion. Since most elements
in stellar interiors are highly ionized, we may introduce an approximation for
the bound-free absorption coefficient for a hydrogenic atom in level n:

64π 4 me e10 Z 4 gbf

abf (n, ν) = √ (52)
3 3ch6 n5 ν 3
where gbf is the Gaunt factor – a quantum mechanical correction factor, Z 
is the effective charge of the nucleus and e, me , c, and h have their usual
meaning. For hydrogen, gbf ≈ 1, Z  ≈ 1. For other atoms, gbf can be quite
large and in most cases is not known. Under this approximation:

2π 2 me e4 Z 2
χion − χn = .
h2 n2
Summing over all ions and all levels, the total bound-free absorption coefficient
becomes
κbf (ν) = Nn abf (n, ν) (53)
ions n

where Nn represents the total number of ions per unit volume in a given
quantum state with quantum number n. Note that abf (n, ν) ∝ ν −3 , and that
Wien’s law demands νmax = b T . Hence Eq. (52) implies κ ∝ T −3 .

8.3 Free-Free Absorption

In the case that both initial and final states are unbound, an electron can
absorb a photon of frequency νff to move to a state with energy given by hνff =
1/2me(v22 − v12 ). The free-free absorption coefficient for one atom and one free
electron per unit volume can be written in the hydrogenic approximation as

4πe6 Z 2 gff
aff (ν) = √ (54)
3 3chm2e vν 3
 Stellar Structure and Evolution 39

where v is the initial electron velocity and gff is the Gaunt factor for free-free
absorption. The total absorption coefficient must be found for a sum similar
to Eq. (53), by averaging over electron velocities,

κff (ν) = Nions aff (ν)ne (v)dv (55)
ions v

normally assuming a Maxwellian velocity distribution ne (v), such that

v ≥
√
(πkT /me ).

8.4 Electron Scattering

Classically, scattering may be pictured as an elastic collision between two

particles. If hν  mc2 , where m is the mass of the scatterer and ν the fre-
quency of the photon, the scatterer is scarcely moved by the collision and
the frequency of the photon will not alter (This approximation only breaks
down for stars with very high central temperatures). Although a photon is
not absorbed, scattering slows down the rate of energy escape by continually
changing the photon direction.
Scattering is independent of frequency but does depend to some extent on
density and the degree of ionisation. In practice, for T > 100 000 K where
hydrogen and helium are fully ionized, the density dependence is negligible
but the composition is not, since this determines the number of electrons per
ion. The absorption coefficient per electron is
σe = 8πe4 /3c4 m2e (56)

and per unit volume

σe ne 4π e4 m2p
κes = = (1 + X) = 0.020(1 + X) m2 kg−1 . (57)
ρ 3 c4 m2e
For obvious reasons, this mechanism dominates at high temperatures, and
hence in fully-ionized stellar cores. It makes a negligible contribution to the
total absorption at low temperatures, except in the absence of other absorbers
such as neutral hydrogen.

8.5 Total Absorption Coefficient

The total monochromatic absorption coefficient is given by the sum of the

contributions from all four processes:
κ(ν) = κbb (ν) + κbf (ν) + κff (ν) + κes (58)

The above has completely ignored stimulated emission. It may be shown that
this effect can be accurately corrected by using a reduced absorption coefficient

κ∗ (ν) = κ(ν)(1 − e(−hν/kT ) )

40 C. S. Jeffery

The Rosseland Mean Opacity

Stellar opacity calculations must consider all atoms and ions and, for stel-
lar structure calculations, provide an appropriate average integrated over all
frequencies.
To compute this average, it is not sufficient simply to weight every fre-
quency equally since the contribution of any one transition must reflect the
relative radiative flux at that frequency. Stellar structure calculations most
commonly use a weighted average
 ∞
 ∞
1 wν
= dν wν dν .
κ 0 κν 0

in which the derivative of the Planck function maximises the opacity contri-
bution where the flux is strongest:
 ∞
 ∞
1 dBν dν dBν
= ∗
dν . (59)
κRoss 0 dT κ ν 0 dT

κRoss is known as the Rosseland mean opacity, and Bν (T ) represents the

energy density for black body radiation:

2hν 3 1
B(ν, T ) = . (60)
c2 ehν/kT − 1
Approximate Values

Since a vast number of data are required to calculate the contribution to

the Rosseland mean from many radiative processes, the calculation of stellar
opacities is a long and involved process. Also, since the total opacity (Rosse-
land mean) is an harmonic mean, the opacity must be recalculated for every
chemical mixture; thus
κ = κ(ρ, T, xi )
where xi represents the abundances of elements i, i = 1, .., N . For modern
stellar structure calculations, detailed tables of precalculated opacities are
usually used. In the case of high ionization, the Saha equation gives Ni,n , the
average number of electrons which are bound in the nth state as:

h3
Ni,n = ne n2 eχn /kT .
2(2πme kT )3/2

ne is the number density of free electrons and n2 is the statistical weight of

the nth state. Substituting this and Eq. (52) into Eq. (53) and rearranging [1,
p. 67] gives Kramers’ Law of Opacity
ḡ ρ
κbf = 4.34 × 1024 Z(1 + X) 3.5 m2 kg−1 (61)
t T
 Stellar Structure and Evolution 41

Fig. 18. Approximate opacities given by Eqs. (57), (61), (62), (63) and (65) for
ρ = T63 and X = 0.7, except κcond which is illustrated for a carbon-oxygen mixture.
κtot represents the appropriate sum of κH− , κbf and κes

(see Fig. 18). A similar derivation for free-free transitions gives

ρ
κff = 3.68 × 1021 g¯ff (X + Y )(1 + X) m2 kg−1 . (62)
T 3.5
At high temperatures, we have already seen that electron scattering domi-
nates:
κ = κes = 0.02(1 + X) m2 kg−1 .
At low temperatures, the high ionization approximation breaks down and the
opacity is dominated by a steeply rising contribution due to the dissociation
of H− [7, p. 164]:

κH− ≈ 2.5 × 10−32 (Z/0.02)ρ1/2 T 9 m2 kg−1 . (63)

The Sun

In order to put some numbers on these formulae, consider the solar interior. In
the centre, ρc ≈ 105 kg m−3 and κc ≈ 10−1 m2 kg−1 . Therefore κρ ≈ 104 m−1
and the photon mean-free path
λ ≈ 10−4 m. At lower temperatures, e.g.,
where ρ ≈ 103 kg m−3 , κ ≈ 10 m2 kg−1 . Thus κρ ≈ 104 m−1 also. The mean
temperature gradient in the Sun dT /dr ≈ 2 × 10−2 K m−1 , and therefore
42 C. S. Jeffery

the temperature difference between photon emission and absorption ΔT ≈

2 × 10−6 K. This demonstrates that radiation and matter in deep stellar
interiors is in true thermodynamic equilibrium.

8.6 Electron Conduction

In very dense stellar material, the mean free path of the photon becomes so
small that it is no longer the most efficient carrier of energy. Thermal conduc-
tion by electrons becomes the dominant transport mechanism, controlled by
a conductive opacity κcond in much the same way as conduction by photons
is controlled by the radiative opacity. From the diffusion approximation we
obtain
4acT 3
κcond = , (64)
3De ρ
where De is a diffusion coefficient for electrons with the general form De ≈
cV ve λ/3. It is relatively straightforward to obtain an approximate expression
for De , whence
 2
μ2 T
κcond ≈ 4 × 10−9 e Zc2 m2 kg−1 (65)
μI ρ
where Zc is the ion charge. Note how, in contrast to radiative opacities, the
conductive opacity is inversely proportional to density – heat travels more
easily in dense media. The total opacity to energy transport by the combined
effects of conduction and radiation is also an harmonic sum,
κ−1 = κ−1 −1
rad + κcond .

9 Thermonuclear Physics
After the discovery that atoms have masses which are integral multiples of the
mass of the hydrogen atom, it was realised that, if a suitable mechanism could
be found, all atoms could be created from the fusion of hydrogen. The problem
is that the electrostatic force implies a strong repulsion between atomic nuclei,
which all carry positive electric charge. Typically, the e-s potential energy for
2 protons, separated by 2 proton radii (≈ 10−15 m) is
Epot = e2 /4πε0 r ≈ 3 × 10−13 J. (66)
7
The average kinetic energy of a proton, at 10 K is
Ekin = 3/2kT ≈ 2 × 10−16 J (67)
which is insufficient to overcome the electrostatic potential barrier. So while
Eddington argued that the interiors of stars were likely sites for the synthesis of
the elements, antagonists (including James Jeans) pointed out the energetics
were against it. Eddington’s rejoinder was that “We do not argue with the
critic who urges that stars are not hot enough for this process; we tell him to
go and find a hotter place.”
 Stellar Structure and Evolution 43

Fig. 19. Binding energy Q per nucleon as a function of atomic mass

9.1 Fusion

Consider the interaction of two protons: the p-p reaction. The fundamental
forces which control the interaction can be expressed in four forms: (Table 4).

Table 4. Forces governing the interaction of two protons

force source range nuclear reactions

gravitational mass 1/r 2 no
electrostatic charge 1/r 2 yes
weak nuclear baryon-lepton 1/r : w  2
w
some
strong nuclear baryon-baryon 1/r s : s  2 yes

For the p-p reaction the electrostatic and strong forces are significant. Since
Ekin  Epot,max , classical physics states that, although the resultant 2p nu-
cleus would be energetically favourable, the two protons cannot approach one
another to within a separation r1 . However, quantum mechanics describes the
proton as a wavefunction ψ given by the solution of the Schrödinger equation
44 C. S. Jeffery

∂ 2ψ 2m
2
+ 2 (Ekin − Epot ) ψ = 0 (68)
∂r h̄
For r > r1 and r < r2 , (Ekin − Epot ) is positive and ψ is real:
r > r1 : ψ ∝ sin kr : k = 2m/h2 (Ekin − Epot )
r2 < r < r1 : ψ ∝ e−kr
r < r2 : ψ ∝ σ sin kr
where σ represents the probability of barrier penetration. There is therefore
a finite probability of the proton tunnelling through to a position r2 and
combining with the target proton. The tunnelling effect also allows alpha-
and beta-decay processes to occur, whereby a particle can escape from the
potential well in the atomic nucleus if it has sufficient kinetic energy.
Formally, the solution of Eq. (68) can be used to obtain an expression for
the penetration probability and hence a reaction cross-section σ.
In the case of the p-p reaction, if the tunnelling operation is successful, an
unstable nuclide consisting of 2 protons is created. What can happen next is
that either the inverse reaction occurs (one proton escapes from the nucleus)
or else one proton quickly releases a positron to remove excess electric charge,
and a neutrino to conserve momentum and lepton number, and becomes a
neutron thus forming a deuterium nucleus.
Reaction Notation
We introduce the notation
i1 (i2 , o3 )o4
where i are the input particles and o are the output particles and where i1
and o4 are the principal participants in the reaction. Both i2 and o3 may be
absent, or may represent one or more particles. Thus the p-p reaction may be
written
1
H(p, β + ν)2 H,
and neutron decay may be written
n(β − , ν̄).
Binding Energy
All atomic nuclei consist of Z protons and N neutrons. The total rest mass
of the energy of the individual particles is always greater than the rest mass
energy of the nucleus. The deficit represents the binding energy of the nucleus
Q(Z, N ) = (Zmp + N mn − m(Z, N ))c2 (69)
which is also the nett amount of energy released during the construction of
a given nuclide. Q is shown for stable nuclides as a function of atomic mass
(A = Z+N) in Fig. 19.
For any given reaction, we are interested in two quantities
 Stellar Structure and Evolution 45

1. the reaction rate: this is related to the penetration probability and the
energy distribution of the particles
2. the energy released: this is (approximately) the change in binding energy
per nucleon
In the latter case, we also need to know how much of the energy is available
for heating the stellar material (photons, electrons, positrons) and how much
is ‘lost’ in the form of neutrinos.

9.2 Reaction Rates

Penetration Probability
The penetration probability takes the form of a cross-section σ. In its simplest
form, the incoming wavefunction takes the form of a decaying exponential
(ψ(r) ∝ e−kr ), so that the penetration probability
σ ∝ (e−kr )2 .
Considering the reaction constant k and writing the equation in terms of the
relative energy E of the nuclei and the Gamow energy (a measure of the
magnitude of the Coulomb barrier)
EG = (παZi Zj )2 2μc2
where α = e2 /4πε0 hc ≈ 1/137 is the fine structure constant and μ is the
reduced mass, the penetration probability may be written [6, pp. 102–104]:
S(E)  
σ≈ exp −(EG /E)1/2 (70)
E
where S(E) is determined by the nuclear physics.
Conversion to a reaction rate (reactions per second per gram) requires an
average over particle velocities and number densities:
∞
σvN (v)dv

σv = 0 ∞ (71)
0
N (v)dv
rij = ni nj
σv/ρ. (72)
Velocity Distribution
Assuming a Maxwellian velocity distribution
 μ 3/2  
μv 2
N (v)dv = exp − 4πv 2 dv.
2πkT 2kT
Writing E = 12 μv 2 and substituting for N (v) into Eq. (71)
 1/2  3/2  ∞
8 1

σv = Eσe−E/kt dE.
πμ kT 0
46 C. S. Jeffery

Reaction Rate

Now substituting for σ and multiplying by number densities:

 1/2  3/2  ∞   1/2 
ni nj 8 1 E EG
rij = S(E) exp − − dE. (73)
ρ πμ kT 0 kt E

The product of the two exponentials (velocity distribution times penetra-

tion probability) has a maximum when the energy

E0 = (EG /4)1/3 (kT )2/3 . (74)

Fusion principally occurs when the relative particle energy E lies in a narrow
energy range around E0 known as the fusion window, given by:
4 1/6
Δ= EG (kT )5/6 . (75)
31/2 21/3
Simplifying Eq. (73) and extracting key terms,
 1/3
EG
rij ∝ ni nj S(E0 ) exp −3 ,
4kT

it may be seen that in a mixture with many species present, fusion will favour
reactions with small Coulomb barriers and high number densities.

Energy Yield

If Qij is the energy released per reaction ij, the total energy released is

εij (ρ, T, ni , nj ) = rij Qij . (76)

9.3 Reaction Networks

Hydrogen-burning (in particular) does not proceed via a single reaction, but
through a chain or cycle involving several different nuclear reactions. In this
case, the reaction rate of the network is governed by the rate of the slowest
reaction in the network, but the energy released per product nucleon will be
that for the entire cycle, thus:

εcycle ≈ rslowest Qcycle.

The principle reaction cycles during main-sequence burning are the proton-
proton (p-p) chains and the CNO cycles.
 Stellar Structure and Evolution 47

Fig. 20. The p-p I chain, with relative lifetimes for each of principal reactions

Table 5. The p-p chains. The bottom two lines show energy yield per 4 He nucleus
produced and the relative total yield over the main-sequence lifetime of a solar-type
star
p-p I p-p II p-p III
1
H(p,β + ν)2 H
2
H(p,γ)3 He
3
He(3 He,2p)4 He 3 He(α,γ)7 Be
7
Be(β − , ν)7 Li 7 Be(p,γ)8 B
7
Li(p,α)4 He 8 B→8 Be∗ + β + + ν
8
Be∗ → 2 4 He
26.2 MeV 25.7 MeV 19.1 MeV
85% 15% 0.02%

p-p chains

The simplest p-p chain begins with the p-p reaction already discussed. The
rest mass of the proton is 1.00728 u, where u is the atomic mass unit, 1/12th
the mass of a 12 C atom. The fusion of two protons, including a positron decay,
yields a deuterium nucleus with rest mass 2.01355 u. The energy excess is thus

Eexcess = δmc2 = (2mp − md )c2 = 1.502 × 10−13 J.

This is the slowest reaction in any of the p-p chains with a rate given by
2/3 1/3
rpp ∝ x2H (ρ/T6 ) exp(25.44 − 33.81/T6 ).
48 C. S. Jeffery

Fig. 21. Energy generation rates as a function of temperature for the p-p (X = 0.7),
CNO (X = 0.7, Z = 0.02) and 3α (Y = 0.98) processes, all for ρ = 105 kg m−3

Fusion with an additional proton creates a 3 He nucleus and a photon in a very

prompt reaction7 , and the energy excess has increased to

Eexcess = δmc2 = (3mp − m3 He )c2 = 8.697 × 10−13 J.

The slower fusion of two 3 He nuclei finally yields a 4 He nucleus, two protons
and a nett energy excess

Eexcess = δmc2 = (4mp − m4 He )c2 = 3.956 × 10−12 J = 24.7 MeV.

The total energy available to the star is less than this because neutrinos pro-
duced by the p-p reaction and in the p-p II and p-p III branches immediately
remove a fraction of the energy yield from the star.
As the hydrogen abundance drops and the 3 He and 4 He abundances in-
crease, additional reactions (Table 5) come into play and eventually the p-p
II and p-p III chains dominate over p-p I.
Two additional points should be noted about the p-p burning: (1) p-p
II results in the destruction of 7 Li and (2) the neutrinos produced in each
reaction have quite distinct energy signatures.
In a region not far below where the p-p and CNO reactions are equally
important, the p-p energy generation rate can be expressed as [21]:
7
This is the fastest hydrogen-burning reaction and, since 2 H is significantly abun-
dant in seawater and can be readily isolated, is the reaction most likely to used
for terrestrial power generation by nuclear fusion.
 Stellar Structure and Evolution 49

2.4x2H ρ
exp(−3.380/T9 ) W kg−1 .
1/3
εpp = 2/3
(77)
T9

Table 6. The CNO cycles

CN NO NOF OF
12
C (p, γ)13 N
13
N →13 C + β + + ν
13
C (p, γ)14 N
14
N (p, γ)15 O 14
N (p, γ)15 O
15
O →15 N + β + + ν 15 O →15 N + β + + ν
15
N (p, α)12 C 15
N (p, γ)16 O 15
N (p, γ)16 O
16 17 16
O (p, γ) F O (p, γ)17 F 16
O (p, γ)17 F
17
F →17 O + β + + ν 17 F →17 O + β + + ν 17 F →17 O + β + + ν
17
O (p, α)14 N 17
O (p, γ)18 F 17
O (p, γ)18 F
18
F → O + β + ν F →18 O + β + + ν
18 + 18
18
O (p, α)15 N 18
O (p, γ)19 F
19
F (p, α)16 O
(12 C destroyed) (16 O destroyed)

CNO cycles

The second major hydrogen-burning network is a catalytic process involving

carbon, nitrogen and oxygen and which becomes important at temperatures
above ≈ 1.5 × 107 K. While nearly all nuclei will react with protons at some
temperature, the results are only interesting from the point of view of long-
lived nuclear burning if the supply of nuclei matches the amount of reactant.
In the case of the CNO cycles, the capture of a proton by a 12 C nucleus to
form a 13 N nucleus is followed by further proton captures and β + decays.
The fourth proton-capture (by 15 N) may give 16 O, but at low temperatures
is more likely to give an α particle and a 12 C nucleus, replenishing the supply
of 12 C so that the cycle may continue (Table 6).
At higher temperatures, additional catalytic cycles come into operation
starting, successively, with 14 N, 15 N and 16 O. The energy produced is ≈25.02
MeV, independent of the cycle, but the rate is cycle dependent, since it is
controlled by the slowest reaction in each cycle. In the case of the CN and the
NO cycles, the slowest reaction is 14 N(p,γ)15 O. When the cycle is operating
in equilibrium, the local abundance of participating elements other than H,
He and 14 N will drop to a low value; 12 C and 16 O are effectively converted to
14
N by the CN and NO cycles respectively.
The overall reaction rate is determined by the 14 N(p,γ)15 O reaction,
1 1/3
rCN ∝ 2/3
exp(−15.23/T9 )
T9
50 C. S. Jeffery

where T9 = T /109 . If xN14 ≈ xN14 + xC12 (CN cycle) or xN14 ≈ xN14 + xC12 +
xO16 (CN + NO cycles), then the overall reaction rate in the region where
p-p and CNO reactions are equally important takes the form

3.2 × 1021 xH xN14 ρ

exp(−15.23/T9 ) W kg−1 .
1/3
εCNO ≈ 2/3
(78)
T9

Note that in addition to helium nuclei, thermonuclear fusion through ei-

ther p-p chains or CN cycles produces positrons, neutrinos, and high-energy
photons (γ rays). The positrons annihilate with electrons to give photons (and
some neutrinos). The photon energy is used to maintain the temperature gra-
dient through the star via the energy transfer equation. The neutrino energy
leaves the star immediately since the neutrino opacity is negligible.
The energy produced by the p-p chains and CN and CNO cycles is plotted
as function of temperature in Fig. 21. For more accurate work, equilibrium
abundances of the reactants should be computed and the most up-to-date
values for the reaction rates should be used [22, 21, 23, 24].

3α and α-capture reactions

When hydrogen is completely converted to helium by either of the above pro-

cesses, the next most energetically favourable reaction is the combustion of
helium:
4
He (α, γ)8 Be – 92 keV
8
Be (α, γ)12 C∗ – 288 keV
12 ∗
C →12 C + 2γ + 7.65 MeV

This reaction has a very high threshold, and the first two stages are en-
dothermic. After the initial α capture, the 8 Be nucleus has a high probability
of decaying to two α particles. Only if a high energy α particle collides with
the 8 Be nucleus within 10−16 s will it form an excited 12 C nucleus. In nearly
all cases this 12 C∗ nucleus will decay back to a 4 He nucleus and a 8 Be nu-
cleus. However if the forward reaction is sufficiently rapid, a small pool of
12 ∗
C nuclei will accumulate, some of which can decay by photon cascade (or
positron-electron pair emission) to give a stable 12 C nucleus and 7.27 MeV.
The energy released by this reaction is [23]:

x3He ρ2
ε3α = 5.1 × 104 exp(−4.4027/T9) W kg−1 . (79)
T93

Note the extremely steep temperature gradient for this reaction (Fig. 21), and
compare it with that for the CNO and p-p reactions.
When the 4 He abundance is reduced by conversion to 12 C, further α cap-
tures become more frequent:
12
C (α, γ) 16 O + 7.16 MeV,
16
O (α, γ) 20 Ne + 4.73 MeV,
 Stellar Structure and Evolution 51

20
Ne (α, γ) 24 Mg + 9.32 MeV
Meanwhile, any 14 N left over from the CNO-process will have been destroyed
by an additional set of α captures:
14
N (α, γ) 18 F + 4.42 MeV,
18
F (, β + ν) 18 O,
18
O (α, γ) 22 Ne + 9.67 MeV

Other reactions

When hydrogen and helium are completely exhausted, heavier elements will
burn at successively higher temperatures. Since the most abundant products
of helium burning are 12 C and 16 O, important reactions include:
T9 ≈0.5-1:
12
C + 12 C →23 Na + p + 2.24 MeV (56%)
12
C + 12 C →20 Ne + α + 4.62 MeV (44%)
T9 >1:
16
O + 16 O →31 P + p + 7.68 MeV (61%)
16
O + 16 O →28 Si + α + 9.59 MeV (21%)
16
O + 16 O →31 Si + n + 1.5 MeV (18%)
T9 >3:
28
Si “burning” – at these temperature, photodisintegration leads to a soup
of protons, neutrons and heavy nuclides in which an enormous number of
possible reactions are possible.

9.4 Nucleosynthesis of elements

As the hydrogen-burning reaction chains suggest, a complex web of nuclear

reactions will occur at the same time. Yields are determined by the branching
ratios and rates of individual reactions. The chemical composition of this
thermonuclear soup can be obtained, at least for steady-state burning, by
setting up a system of simultaneous equations that model the entire reaction
network and solving for the equilibrium abundances of each chemical species.
Even in steady-state nuclear burning, some species are destroyed and oth-
ers created, so the chemical composition of a star will change. If we know the
reaction rate rijk for a reaction in which species i and j are destroyed and
species k is produced, then the rates of change of species i (or j) and k are
simply related by
dni dnk dni dnj
|ijk = −(1 + δij )rijk |ijk = −( + ) (80)
dt dt dt dt
where ni,j,k are the relative abundances by number.
These time-dependent equations represent the fundamental equations of
stellar evolution. A succession of nuclear reactions (H, He, C and subsequent
burning) represents the mechanism whereby all the chemical species have
been created. Many exotic elements can only be produced in core-collapse
52 C. S. Jeffery

supernova explosions, often through the capture of fast neutrons (r-process).

Others are produced by the capture of slow neutrons in the envelopes of low-
and intermediate-mass stars (s-process). The basic framework for the theory
is elegantly illustrated in [25, Fig. I.2]. Such processes are described in more
detail elsewhere in these proceedings.

9.5 Neutrinos

Neutrinos are produced as electron (or positron) decay or capture products in

nearly all thermonuclear reaction networks. The neutrinos carry energy away
from the stellar interior because their interaction with the stellar material is
very small (but not zero). To give some idea of the scale, a typical neutrino
capture cross section is given by σν ≈ 10−48 ε2ν m2 where εν is the neutrino
energy in MeV. Then the mean free path is λ = 1/nσν ≈ 1018 ε−2 ν /ρ m,
where n is the number density of targets with a mean molecular weight near
unity. For ordinary stars λ is very large, but in the cores of supernovae, with
ρ ≈ 1017 kg m−3 and εν ≈ 20 MeV, then λ ≈ 25 m may be possible. The
consequences for energy transport and hence the physics of the supernova
core are dramatic.
Usually the neutrino losses are modest. Measurements of the solar neutrino
flux have been used to test models of stellar structure. However the neutrino
luminosity can be crucial during some stages of evolution – and can lead to a
negative flux gradient! This becomes particularly severe in the late stages of
stellar collapse when conditions arise under which large numbers of neutrinos
can be created. A very large burst of neutrinos is produced in a supernova
explosion, with flux approximately equal to the total photon flux. Such a large
burst was observed from SN 1987A, despite the latter being in another galaxy.
In addition to the nuclear decay/capture neutrinos, neutrinos can be pro-
duced in a number of other ways which can become important at some stages
in stellar evolution.

Urca process

The Urca process is essentially an electron capture followed by β-decay, with

the emission of a neutrino and anti-neutrino pair, e.g.
35
Cl(β − , ν)35 S →35 S(, β − ν̄)35 Cl

The energy carried by the ν ν̄ pair is lost to the star, but the process is unlikely
to be important except in very evolved cores with high densities and low
temperatures.

Photo-Neutrino Process

Electron scattering by photons may alternatively result in pair neutrino pro-

duction, with a nett energy loss as shown:
 Stellar Structure and Evolution 53

γ + β − → β − + γ → β − + ν̄ + ν
εν ≈ 103 T98 W kg−1
This can be produced by ordinary bremsstrahlung and is an important energy
loss mechanism in hot white dwarfs.
Pair-Neutrino Process
Similarly:
γ + γ ↔ β − + β + ↔ ν̄ + ν
T93
εν ≈ 1014.7 exp(−11.88/T9) W kg−1
ρ
Plasma-Neutrino Process
In a dense plasma electromagnetic waves can be quantized and behave like
relativistic Bose particles with finite mass. These “plasmons” can decay into
either β − + β + or ν̄ + ν pairs.

10 Approximate Solutions
Although general solutions of the stellar structure equations require numerical
techniques, approximate models provide valuable insight.

10.1 Polytropic Gas Spheres

Rearranging Eq. (17)
dP r2
m=− (81)
dr ρG
differentiating and substituting Eq. (14)
 
dm d dP r2
= 4πr ρ = −
2
dr dr dr ρG
 2 
1 d r dP
⇒ 2 = −4πGρ (82)
r dr ρ dr
Suppose the equation of state takes the polytropic form
P = Kργ (83)
where K is a constant and γ = (n + 1)/n. n is said to be the polytropic index.
Then Eq. (82) becomes Poisson’s equation
 
1 d r2 γ−1 dρ
Kρ = −4πGρ (84)
r2 dr γ dr
with boundary conditions ρ = ρc and dρ/dr = 0 at r = 0.
The resulting solutions are called polytropes, and are encountered in con-
nection with:
54 C. S. Jeffery

1. adiabatic changes (γ = cP /cV ) applicable to stars in adiabatic convective

equilibrium.
2. equation of state of degenerate matter γ = 5/3 (NR), γ = 4/3 (UR),
applicable in white dwarfs.
3. models where Pr /P =constant.

By writing
y = (ρ/ρc )−n
and  1/2
(2−γ)
4πGρc
x= r,
(n + 1)K

Eq. (84) may be rewritten in the more convenient form:

 
1 d 2 dy
x + yn = 0 (85)
x2 dx dx

which is known as the Lane-Emden equation, with boundary conditions y(0) =

1 and (dy/dx)x=0 = 0.
Polytropic models have played a very important role in the development
of stellar structure theory, but although popular before the computer age, still
involve the solution of quite messy differential equations.

10.2 Clayton Models

A simpler approach to solving the structure equations is to guess a form for the
density profile ρ(r) and to use this as a starting point for an approximate solu-
tion. The equations can be tackled sequentially, again starting with Eqs. (14)
and (17) to obtain the pressure structure. An equation of state provides the
temperature structure. Introducing an opacity κ(ρ, T ), Eq. (33) can be used
to obtain the power flow, which can be compared with the power flow ob-
tained from the nuclear power density ε(ρ, T ) and Eq. (28). For approximate
solutions, these last will bear little similarity.

Pressure

Clayton (1986) proposed a simple parameterisation of the pressure profile P .

The pressure gradient is directly constrained by Eq. (17) (hydrostatic equi-
librium):
dP Gmρ
=− 2 .
dr r
Near the centre, r is small and m ≈ 4πr3 ρc /3, hence

dP 4π
= − Gρ2c r.
dr 3
 Stellar Structure and Evolution 55

Near the surface m ≈ M , hence

dP GM ρ
=− 2 .
dr r
Thus the pressure gradient must be zero at the centre, initially varying
linearly with radius but again approaching zero as ρ/r2 decreases towards the
surface. The essence of the Clayton model is to guess a simple form for P (r)
which takes these into account:
dP 4π  
= − Gρ2c r exp −r2 /a2 (86)
dr 3
where a is a length parameter to be specified. Integrating, with P (R) = 0,
gives
2π 2 2     
P = Gρc a exp −r2 /a2 − exp −R2 /a2 (87)
3
ρc , a and R. By definition
defining a family of stellar models parameterised by √
the minimum in dP/dr occurs at a distance r = a/ 2 from the centre of the
star.

Temperature and Density

Corresponding expressions for temperature and density can then be obtained.

From hydrostatic equilibrium

Gmdm = −4πr4 dP

and integrating:  r
Gm2 dP 
= −4π r4 dr .
2 0 dr
Substituting for the pressure gradient:

m = (4/3)πa3 ρc Φ(x) (88)

where x = r/a and

 x
     
2
Φ (x) = 6 x5 exp −x2 dx = 6 − 3 x4 + 2x2 + 2 exp −x2 . (89)
0

Then the density can be found directly (mass equation)

  
1 dm x3 exp −x2
ρ= = ρc (90)
4πr2 dr Φ

and from the equation of state for an ideal classical gas (Eq. (41))
P μmu
T = (91)
ρk
56 C. S. Jeffery

where μ = 2/(1 + 3X + Y /2). At small r, these quantities will be reasonably

reliable.
The exponential term in Eq. (90) may be expanded as a power series to
obtain approximate expressions valid for small r:
 
5r2
ρ = ρc 1 − 2 + . . . ,
8a
 
3r2
T = Tc 1 − 2 + . . . .
8a

High ρc Models

The model is further simplified when ρc 

ρ. Then a  R and terms pro-
portional to exp(−a2 /R2 ) can be neglected. This is reasonable when applied
to the Sun, where a = R /5.4. With small a, Eq. (88) gives
√
4πρc a3 4πρc a3 6
M = m (R) = Φ (R/a) ≈ (92)
3 3
√
and the average density is about 6(a/R)3 ρc .
It can then be shown that the density at r = a is 0.53ρc and that a sphere
of radius a contains 0.28M . Substituting r = 0 into Eq. (87),
2π 2 2
Pc ≈ Gρc a
3
and eliminating a with Eq. (92) gives

Pc ≈ (π/36)1/3 GM 2/3 ρ4/3

c ≈ 0.44GM 2/3 ρ4/3
c (93)

which is expected to be valid for any homogeneous star where the mass is
concentrated towards the centre. Other simple models (e.g. polytropes) give
similar results.

Modelling the Sun

The simple model described above is characterised by ρc , a and R. Equa-

tion (92) provides a in terms of M and ρc . If we take M = M , R = R , and
ρc = 9 × 104 kg m−3 from a standard solar model, then a = R /5.4. Equa-
tions (87), (90) and (91) then provide the variation in pressure, temperature
and density inside the Sun (Fig. 22). The variation of these solutions with
mass can be investigated quite simply.
 Stellar Structure and Evolution 57

Fig. 22. Distribution of P , m, ρ, T , and l with r in the Sun as given by the

approximations (87), (88), (90), and (91) and by integrating Eq. (28) with Eqs. (77)
and (78)

10.3 Minimum Mass of a Star

We have found expressions for the central temperature of a star. This must
be greater than the ignition temperature for the hydrogen fusion if a star is
to result. For a perfect gas, the pressure given by Eq. (93) must be the same
as that given by the equation of state. Equating the pressures gives

kTc ≈ (π/36)1/3 Gμmu M 2/3 ρ1/3

c

1/3
Since Tc ∝ ρc , the temperature of a contracting gas cloud continues
to rise as it contracts until either nuclear reactions are initiated or electron
degeneracy prevents further contraction. The condition for stardom is that
the density at which electron degeneracy prevents further contraction corre-
sponds to a temperature greater than that required for the ignition of nuclear
reactions.
An estimate of this critical density/temperature is obtained by assuming
that the electrons are degenerate and the ions are classical and, to simplify the
algebra, that the star is entirely composed of hydrogen. The central pressure
is given by
Pc = KNR n5/3
e + ni kTc
where ne = ni = ρc /mu
58 C. S. Jeffery

Pc = KNR (ρc /mu )5/3 + (ρc /mu )kTc

This must satisfy hydrostatic equilibrium (Eq. (93)) and so, eliminating
Pc ,
kTc ≈ (π/36)1/3 Gmu M 2/3 ρ1/3
c − KNR (ρc /mu )2/3
which has the form
kTc = Aρ1/3
c − Bρ2/3
c ,

where kTc reaches a maximum of A2 /4B at a density of (A/2B)3 . Substitut-

ing, we find
(kTc )max ≈ (π/36)1/2 G2 M 4/3 m8/3
u /4KNR

and
Mmin ≈ (36/π)1/2 (4KNR /G2 m8/3
u )
3/4
(kTign )3/4
For Tign ≈ 1.5 × 106 K, Mmin ≈ 0.1 M .

Fig. 23. The ratio of mean radiation to total pressure as a function of mass

10.4 Maximum Mass of a Star

Pressure in the stellar core is the sum of gas pressure Pg = ρc kTc /mu μ and
radiation pressure Pr = aTc4 /3. Denoting Pg = βPc and Pr = (1 − β)Pc , T
can be eliminated to give
 1/3  4/3
3 (1 − β) kρc
Pc = (94)
a β4 mu μ

Equating this to the pressure from Eq. (93) to satisfy hydrostatic equilibrium
yields
 π 1/3  1/3  4/3
2/3 3 (1 − β) k
GM = (95)
36 a β4 mu μ
 Stellar Structure and Evolution 59

This equation can be used to plot (1–β) (the radiation pressure fraction) as a
function of mass (Fig. 23). When this quantity is more than ≈ 0.5, radiation
pressure leads to strong stellar winds and dramatic mass loss from the stellar
surface.

11 Methods for Numerical Solution

Recall that the problem is to find the structure of a star with given mass M
and chemical composition xi (m), i = 1, 2, . . . by solving the system of equa-
tions described in Sect. 5, together with microphysics such as that described
in Sects. 7–9.
In general, there is no analytic solution although approximate solutions
can be found. In practice, it is necessary to find a numerical solution. If all
four boundary conditions were given, say, at the centre, this would require a
straightforward numerical integration from the centre to the surface. This is
obviously not the case. There are two methods commonly used to deal with
the problem.

rof

δrf

r
rif

0
0 mf M
m

Fig. 24. Schematic of solution for one variable using the shooting method. The trial
solution is shown as a solid line, the desired solution as a dashed line

11.1 Shooting Method

The principle of this method divides the star into two parts, an inner and
outer part, for which separate solutions will be obtained by estimating a set
60 C. S. Jeffery

of additional boundary conditions. These are then adjusted until they match
one another. The procedure works as follows.
Inward solution. At m = M we have Ps = 0, Ts = Teff 8 . Estimate R and L.
Integrate inwards to the same point mf , where we obtain Pif , Tif , lif , and rif .
Outward solution. At m = 0 we have rc = 0, lc = 0. Estimate Pc and Tc .
Integrate outwards to some fitting mass mf , where we obtain Pof , Tof , lof and
rof .
For any arbitrary starting values of R, L, Pc , Tc we have, in general, δPf =
Pif − Pof = 0, δTf = Tif − Tof = 0, δlf = lif − lof = 0, and δrf = rif − rof = 0.
The goal is therefore to improve these starting values iteratively until δPf =
δTf = δlf = δrf = 0.
Repeat the inward solution with R + δR, L and again with R, L + δL, and
repeat the outward solution with Pc + δPc , Tc , and with Pc , Tc + δTc .
From the resulting changes in rif, rof , etc., form the derivatives ∂δrf /∂R,
∂δrf /∂L, ∂δrf /∂Pc , etc. Then corrections to R, L, Pc , Tc , may be found by
solving
⎛ ∂δrf ∂δrf ∂δrf ∂δrf ⎞ ⎛ ⎞ ⎛ ⎞
∂R ∂L ∂Pc ∂Tc δR δrf
⎜ ∂δlf ∂δlf ∂δlf ∂δlf ⎟ ⎜
⎜ ∂R ∂L ∂Pc ∂Tc ⎟ ⎜ δL ⎟ ⎜
⎟ = ⎜ δlf ⎟
⎟
⎜ ∂δpf ∂δpf ∂δpf ∂δpf ⎟ ⎝ ⎠ ⎝ (96)
⎝ ∂R ∂L ∂P ∂T ⎠ δPc δpf ⎠
c c
∂δtf ∂δtf ∂δtf ∂δtf δTc δtf
∂R ∂L ∂Pc ∂Tc

for δR, δL, δPc , and δTc . These corrections are applied and new estimates are
used to repeat the entire process, iterating until convergence is obtained, i.e.
by satisfying some criteria such as δPf < ε, δTf < ε, δlf < ε, and δrf < ε.
The result is a set of values for R, L, Pc , Tc and solutions for r, l, P , and
T on the interval m = [0, M ].

rj, lj, Pj,Tj rj+1, lj+1, Pj+1, Tj+1

mj ρj+1/2, εj+1/2 ,∇j+1/2 mj+1

Fig. 25. Discrete variables used in difference equations

11.2 Difference Method

An alternative approach can be obtained by writing the o.d.es for stellar

structure as difference equations. To do this, consider the star divided into J
discrete shells, numbered 0, . . . , j, . . . , J from the center to the surface. The
mass interior to shell j is denoted mj , with m0 = 0 and mJ = M (Fig. 25).
Replacing other continuous variables by the discrete variables rj , lj , Pj , Tj , we

8
In practice it is better to use a non-zero starting condition such as Ps = (−2/3)g/κ
 Stellar Structure and Evolution 61

⎛ ⎞
∂B1 ∂B2
∂r1 ∂r1
··· ··· ··· ··· ··· ··· ··· ··· ··· 0
⎜ ⎟
⎜ ⎟ ..
⎜ ∂B1 ∂B2
··· ··· ··· ··· ··· ··· ··· ··· ··· ⎟ .
⎜ ∂l1 ∂l1 ⎟
⎜ .. .. ⎟ ..
⎜ . ⎟
⎜ . 0 ⎟ .
⎜ .. ⎟
⎜ ∂G4,j−1 ∂G1,j−1 ∂G2,j−1 ∂G3,j−1 ∂G4,j−1 ⎟
⎜ 0 0 0 . ⎟
⎜ ∂Tj−1 ∂Tj ∂Tj ∂Tj ∂Tj ⎟
⎜ ..⎟
⎜ ∂G1,j ∂G2,j ∂G3,j ∂G4,j ∂G1,j ⎟
⎜ 0 0 0 .⎟
⎜ ∂rj ∂rj ∂rj ∂rj ∂rj+1
⎟
⎜ .. ⎟
..
⎜ . 0
∂G1,j ∂G2,j ∂G3,j ∂G4,j ∂G1,j ⎟
.
⎜ ∂lj ∂lj ∂lj ∂lj ∂lj+1 ⎟
A=⎜ ⎟
⎜ .. ⎟
..
⎜ . 0
∂G1,j ∂G2,j ∂G3,j ∂G4,j ∂G1,j
⎟.
⎜ ∂Pj ∂Pj ∂Pj ∂Pj ∂Pj+1 ⎟
⎜ .. ⎟..
⎜ ∂G1,j ∂G2,j ∂G3,j ∂G4,j ∂G1,j ⎟
⎜ . 0 ⎟ .
⎜ ∂Tj ∂Tj ∂Tj ∂Tj ∂Tj+1 ⎟
⎜ .. ⎟ ..
⎜ ∂G1,j+1 ⎟
⎜ . 0 0 0 0 0 ⎟ .
⎜ ∂Tj+1
⎟
⎜ .. .. ⎟ ..
⎜ . 0 . ⎟ .
⎜ ⎟
⎜ .. ⎟
⎜ ∂C4 ⎟
⎝ . ··· ∂C3
∂PJ ∂PJ ⎠
0 ··· ··· ··· ··· ··· ··· ··· ··· ··· ∂C3
∂TJ
∂C4
∂TJ

⎛ ⎞ ⎛ ⎞
δr1 −B1
⎜ δl1 ⎟ ⎜ −B2 ⎟
⎜ ⎟ ⎜ ⎟
⎜ . ⎟ ⎜ .. ⎟
⎜ .. ⎟ ⎜ ⎟
⎜ ⎟ ⎜ . ⎟
⎜ δT ⎟ ⎜ −G ⎟
⎜ j−1 ⎟ ⎜ 4,j−1 ⎟
⎜ ⎟ ⎜ ⎟
⎜ δrj ⎟ ⎜ −G1,j ⎟
⎜ ⎟ ⎜ ⎟
⎜ δlj ⎟ ⎜ −G2,j ⎟
x=⎜ ⎟ and b=⎜ ⎟
⎜ δPj ⎟ ⎜ −G3,j ⎟
⎜ ⎟ ⎜ ⎟
⎜ δTj ⎟ ⎜ −G4,j ⎟
⎜ ⎟ ⎜ ⎟
⎜ δrj+1 ⎟ ⎜ −G1,j+1 ⎟
⎜ ⎟ ⎜ ⎟
⎜ .. ⎟ ⎜ .. ⎟
⎜ . ⎟ ⎜ . ⎟
⎜ ⎟ ⎜ ⎟
⎝ δPJ ⎠ ⎝ −C3 ⎠
δTJ −C4

Fig. 26. Block structure for difference equations

arrive at a system containing 4(J + 1) variables. Since r0 , l0 , PJ and TJ are

given by the boundary conditions, we have 4J unknowns.
The ordinary differential equations can be discretized and Eq. (14) can be
written for example, as:
 
rj+1 − rj 1
= .
mj+1 − mj 4πr2 ρ j+1/2
62 C. S. Jeffery

Since there are 4 such equations for J shells, we have a system of 4J non-linear
simultaneous equations in 4J unknowns. Equation (14) can be rewritten in
the form:
1
G1,j ≡ (rj+1 − rj ) − (mj+1 − mj ) 2 = 0; quadj = 1, . . . , J − 1.
4πrj+1/2 ρj+1/2

Likewise Eqs. (17), (28) and (33) become G2,j , G3,j , G4,j = 0, boundary con-
ditions (34)–(37) become B1 = 0, B2 = 0, C3 = 0, C4 = 0. Hence we have a
system of non-linear equations of the form:

Ek (y1 , y2 , . . . , y4J ) = 0; k = 1, 4J. (97)

Supposing that there exists an approximate solution:

(0) (0) (0)
y1 , y2 , . . . , y4J ,

then the system can be linearized using, for example, a Newton-Raphson

(0)
iteration. In general, an approximate solution yi will not satisfy the system
Eq. (97). The goal is therefore to find a set of corrections δyi such that
(s+1) (s)
yi = yi + δyi ; i = 1, 4J,

by solving the block-diagonal system of equations

4J 
(s)
∂Ek
δyi = −E (s) ; k = 1, 4J,
i=1
∂yi

≡ A·x =b
where A, x and b are matrices (Fig. 26).
This system may be solved using a substitution scheme known as the
Henyey method [26], the corrections are applied to the approximate solution,
and then iterated to convergence [27].

12 Stellar Evolution
12.1 Pre-Main-Sequence Evolution

Before discussing the structure and evolution of main-sequence stars, it is

helpful to make some remarks about how stars reach the main sequence in
the first place.
It is well established observationally that stars form in massive interstellar
dust clouds. If the Jeans condition is satisfied (cf. Eq. (25)), then parts of
the cloud will start to collapse, leading to the release of gravitational energy.
Initially, the star will be very luminous but cool, and may be identified with
“Hot Cores” in giant molecular cloud complexes. Low-mass protostars are
 Stellar Structure and Evolution 63

fully convective and the family of solutions lies on the left side of the HR
diagram on a path known as the Hayashi track. The contracting protostar
evolves down this track, heating in the process.
Once the core temperature is high enough (hydrogen and helium must be
fully ionized) the opacity (κbf +κes ) fall sufficiently for a radiative core to form.
At this point the star contracts at constant L towards the main-sequence, and
the convective envelope shrinks. By this stage, the stars have mostly emerged
from their dusty cocoon and become visible, low-mass stars as T Tauri stars
and intermediate mass stars as Herbig Ae/Be stars.
As contraction continues, the core continues to heat until Tc exceeds the
critical value for the ignition of deuterium burning (the second reaction in the
p-p I chain). Providing the star is sufficiently massive, this will be followed
by the p-p reaction and/or the CNO-cycle and full hydrogen burning will be
established. At this point the star has reached the “zero-age main sequence”
(ZAMS).

12.2 The Zero-Age Main Sequence

The location of stars along the ZAMS is a function of mass and composition.
For radiation dominated stars

L ∝ M 3 /κ.

The mass-luminosity relation averaged over the whole main sequence

(Sect. 2.3) is empirically
L ∝ M 3.5 ,
but local variations reflect changes in the dominant physics.
At low mass, M/M < 0.4, the star is fully convective due to the high
opacity of the incompletely ionized gas, and

L ∝ 0.23M 2.3.

With increasing mass 0.4 < M/M < 7, the star first develops a ra-
diative core. By about 1.4 M , the (opacity-driven) convective envelope has
disappeared. Depending on metallicity, a nuclear-driven convective core will
develop. In this interval,
L ∝ M 4.75 .
For M/M > 7, the star maintains a radiative envelope and increasingly
massive convective core. For the most massive stars, the convective core can
comprise over 50% of the mass of the star. In this mass range,

L ∝ M 3.
64 C. S. Jeffery

Nuclear Energy
The principal nuclear energy generation also depends on the mass and compo-
sition. For Population I ZAMS stars (Z ≈ 0.02), energy produced by the p-p
chain and CNO cycle is approximately equal for M ≈ 1.9M . An increasing
contribution from the CNO cycle will develop during main-sequence burning
as the central hydrogen concentration drops.
For Population II stars (Z < 0.001) the p-p chain is generally dominant at
low masses (M < 2M ), though the CNO-cycle will continue to be important
at high masses.
Convection
Convection occurs in the core when the CNO cycle contributes more than
≈ 20% of the total energy output, since εCN ≈ T 17 implies that the tempera-
ture gradient in the core is very steep. The fact that stars in the 1 − 1.4 M
range can develop CNO burning during main-sequence evolution also implies
that such stars (including the Sun) may develop a convective core.
The significance of a convective core is twofold. In the first place, convec-
tion mixes proton-rich material from a region much larger than that which
is actually burning, thus increasing the total fuel available and extending the
potential main-sequence lifetime by a comparable amount.
Second, the shape of the evolution track is altered. In a star with a radia-
tive core nuclear burning shifts smoothly from the hydrogen-depleted core to
a thick hydrogen-poor shell and then to a thin hydrogen-rich shell. The tran-
sition from main-sequence to post main-sequence evolution is comparatively
smooth (Fig. 8). In a star with a convective core, nuclear burning switches off
abruptly as the entire convective core is exhausted. This precipitates a minor
collapse of the depleted core before ignition of the hydrogen shell, and pro-
duces a “hook” in the evolution track at the end of main-sequence evolution.
Convective envelopes are found when the hydrogen ionization zone is suf-
ficiently deep inside the star which occurs for Teff < 8 000 K and M < 1.4 M .
The mass in the convective envelope increases with decreasing M until the
entire star becomes convective at M ≤ 0.2 − 0.3 M .
Opacity
The composition of a main-sequence stars has a direct effect on both the
luminosity and effective temperature by changing the mean molecular weight
and the opacity. Since the effects are not linear, we state simply that, as
a rough guide, the core opacity controls the luminosity (by modifying the
temperature gradient and hence the rate at which energy leaves the star) and
the envelope opacity controls the radius (or Teff ).
We can deduce what will happen to the star during main-sequence evolu-
tion by considering the changes in chemical composition in the core. As hy-
drogen is consumed, X decreases and Y increases. This has two consequences.
κ is reduced (Eqs. (53), (57)) and μ is increased.
 Stellar Structure and Evolution 65

From hydrostatic equilibrium, increasing μ requires that ρc must increase.

From the Virial theorem it then follows that Tc must also increase, and this
leads to an increase in εc and L. So a main-sequence star will become more
luminous as it depletes its hydrogen core.

Distribution of Mass and Luminosity

The nuclear energy generation, the energy transport and the opacity also have
consequences for the distribution of mass m(r) and luminosity l(m) within the
star. The first indicates how centrally condensed the star is. One measure is
the fraction of the total mass located within 50% of the radius (or 12% of the
volume). Stars with small convective cores are more centrally condensed than
those with radiative cores so that a 2 M star has 95% of its total mass inside
this volume. For a 0.4 M star, the figure is 60%, and for a 10 M star, it is
90%.
The temperature exponent of the energy production rate determines where
the power is generated. Consequently the CNO-cycle generate 90% of the total
luminosity within the innermost 15% of the mass. With a “softer” temperature
exponent, p-p burning extends this figure up to about 30%.

Fig. 27. The evolution of a 5 M star through the L − Teff diagram (Based on [28,
Fig. 3], using a contemporary evolution track by R. G. Izzard.)
66 C. S. Jeffery

12.3 Evolution of a 5 M Star

Having examined the variation in stellar structure along the main sequence,
we turn our attention to the evolution of a star of given mass ( [29], Fig. 27).
We have chosen a mass of 5 M because it is relatively simple, avoiding the
high mass-loss associated with more massive stars and the convective envelope
associated with low-mass main-sequence stars.
During the main hydrogen-core burning phase, hydrogen is converted to
helium via the CNO-cycle. Consequently the star has a convective core, but
this decreases in mass from 20% to 8% of the total mass during main-sequence
burning. At the same time, the luminosity increases as hydrogen is depleted
(see above) all on a nuclear timescale tnuc ≈ 6 × 107 years.
When the core hydrogen abundance drops to Xc < 0.5, any further in-
creases in Tc fail to compensate for the drop in energy generation and the
whole star starts to contract. As nuclear reactions are extinguished, the con-
vective core vanishes, but only when Xc  0.01. Much of the star’s luminosity
now comes from the release of gravitational energy, with the core contracting
on a thermal timescale tK ≈ 2 × 106 years.
When Xc ≈ 0, approximately the tip of the “hook” in Fig. 27, the finite
temperature gradient through the star removes energy from the core faster
than εnuc so that the core is forced to contract and cool. Since the greatest cool-
ing is at the centre, the core becomes isothermal on a timescale t ≈ 105 years.
The mass of the, now, helium core is ≈ 7% of the total mass. Outside this
core, hydrogen-rich material continues to fall inwards, heats, and eventually
ignites nuclear reactions. Once ignited, heat from the thick hydrogen-burning
shell forces the outer layers of the star to expand.
The base of the hydrogen-burning shell now adds helium to the he-
lium core, which increases in mass. When the core mass Mc ≈ 0.1M (the
Schönberg-Chandrasekhar limit9 ) the core contracts very rapidly. This con-
traction and consequent heating feeds back directly into the hydrogen-burning
shell, so that the luminosity of the shell (Lsh ) increases rapidly. This does not
result in an immediate increase in L (the stellar luminosity), the stellar enve-
lope absorbs much of the excess energy, heats and expands, so that R increases
rapidly. Actually, this expansion absorbs so much energy that L falls until the
star starts to become convective and take over as the dominant transport pro-
cess in the envelope. As a consequence, the star crosses from the main sequence
to the giant branch very quickly (≈ 106 years), producing the Hertzsprung
gap in HR diagrams for young open clusters.
With the effective temperature now less than 5 000 K, the surface layers
become increasingly convective. Whilst at the base of the giant branch, the
9
The Schönberg-Chandrasekhar limit [30] represents a limit above which an
isothermal non-degenerate core can support the remainder of the star. Its value
is ≈ 0.1 − 0.15M and is valid for stars with total mass between 2 and 6 M . Less
massive stars develop an electron-degenerate core and more massive stars ignite
helium before this value is reached.
 Stellar Structure and Evolution 67

surface convection zone extends deep enough to reach material which has
been partially CNO-cycled. In a process known as first dredge-up, nitrogen
and helium-rich material dilutes the composition of the previously pristine
surface. Deeper in the star the hydrogen-burning shell becomes thinner, hotter
and continues to add mass to the core. As the Mc increases, so does Lsh and
R. Since at M = 0.5 M , the core is not degenerate, it contracts and heats
until, when Tc ≈ 108 K, 3α reactions are ignited.
Following core-helium ignition, the core expands (Rsh increases), the
hydrogen-burning shell cool, and both Lsh and L drop, as the star approaches
a long-lived phase of core-helium burning. With a smaller total radius, the
effective temperature also increases and the surface layers become fully ra-
diative during this phase. However, the high temperature exponent of the 3α
reaction means that a convection zone develops in the core. In contrast to the
CNO-cycle driven convection zone, the convective core mass increases during
most of core helium-burning, so that by the time Yc = 0, there is relatively
massive helium-poor layer surrounding a central carbon-oxygen core.
At this point, helium-burning shifts initially to a thick shell surrounding
the inert inner core, causing the outer (helium) core to contract and the
hydrogen-burning shell to move inwards and heat. In a re-run of the approach
to the giant branch, the increasing power of the hydrogen-burning shell forces
the star to expand, cool, and develop a deep convective envelope for the second
time, again dredging freshly-processed helium to the surface. Helium shell-
burning turns out to be unstable; as the star evolves up the asymptotic giant
branch, the dominant energy source continues to be the hydrogen shell, with
a luminosity which goes [31, 32] roughly as:

Lsh ≈ 104 Mcδ ; δ ≈ 2, Mc ≈ 0.86.

As the mass of intershell zone (i.e. the helium layer between the hydrogen
shell and the carbon-oxygen core) increases to around 0.004 M, the tem-
perature at its base increases to the point when 3α reactions are re-ignited.
There follows a complex interaction in which a 3α-driven convection zone is
driven upwards into the intershell, the intershell expands and suppresses the
hydrogen-burning shell. Briefly, the helium-burning shell adds about 0.001 M
carbon ash to the core beneath, but is itself extinguished after less than 5 y.
This allows the intershell to contract, the hydrogen-shell to heat, surface con-
vection to dredge-up new material from the top of the intershell, and the
hydrogen-shell to start building up the mass of the intershell until after an
interval of some 5000 y, this sequence of thermal pulses is repeated.
With increasing luminosity, the star which started with a mass of some
5 M will lose most of its mass via a stellar wind. Just as the surface layers
are being removed above, the hydrogen-envelope is being burnt at an ever-
increasing rate from beneath until there is insufficient mass to sustain the
hydrogen-burning shell. With a mass of some 0.9 M , electron-degeneracy in
the carbon-oxygen core (Sect. 13.1) will prevent it contracting and heating to
68 C. S. Jeffery

the point at which nuclear carbon burning becomes possible (Sect. 9.3). At
this point the star will effectively die, contracting to become a white dwarf
(Sect. 15.6).

12.4 Evolution at Other Masses

The parameter which defines the evolution of a star more than any other
is mass. Clearly (Sect. 4), the more massive a star, the more luminous and
the more quickly it evolves. There is not enough space here to discuss the
individual evolution of stars up and down the main-sequence and at different
metallicities. However there are some important classifications to make.
Very high-mass stars with initial masses greater than ≈ 20 M will ignite
helium ignition while the star is still a blue supergiant. Their extremely high
luminosities lead to a radiation-driven stellar wind and substantial mass-loss.
The result is likely a Wolf-Rayet star, followed by a core-collapse supernova.
High-mass stars have a final mass greater than the Chandrasekhar mass
(Sect. 13.1). In these stars, nuclear burning of carbon, oxygen and heavier
elements will be followed by collapse as a type II supernova (Sect. 13.2). This
is generally thought to occur for stars with initial masses greater than ≈ 8 M .
Intermediate-mass stars with initial masses roughly between 3 and 8 M
ignite helium in a non-degenerate core, thus avoiding both the helium core
flash and the core-collapse supernova fate. These stars will all evolve to the
asymptotic-giant branch phase and subsequently to become CO white dwarfs.
In low-mass stars with initial masses less than about 3 M , helium-ignition
occurs in a degenerate core – usually when the helium-core mass has reached
about 0.48 M . The star subsequently evolves as a horizontal-branch star
(Sect. 14). With a sufficiently massive hydrogen envelope (the exact value
depends strongly on metallicity), it will evolve up the asymptotic giant branch
after core helium exhaustion. Otherwise it will evolve directly to the white-
dwarf cooling track.
Very low-mass stars with initial masses below about 0.5 M may never
ignite helium. Being nearly fully convective, they have a large reservoir of
nuclear fuel compared with their luminosity, so their nuclear lifetimes are
long compared with the age of the Universe. Ultimately they will fade away
as degenerate helium white dwarfs.

13 Stellar Remnants
13.1 White Dwarfs

When a star runs out of nuclear fuel it contracts until either the temperature
becomes high enough to ignite another nuclear fuel source or until electron-
degeneracy provides sufficient pressure to restore hydrostatic equilibrium. Re-
call the hydrostatic equilibrium and mass continuity equations:
 Stellar Structure and Evolution 69

dP/dr = −ρGm/r2 , dm/dr = 4πr2 ρ. (98)

In the completely degenerate regime, ρ is independent of T , P = KNR (ρ/mu

μe )5/3 and we have two o.d.e.s with two b.c.s (cf. the Lane-Emden equation).
Hence hydrostatic equilibrium alone determines R and ρ as a function of mass.
Write average values in terms of the central pressure Pc , M and R:

dP/dr = −Pc /R,
3

ρ = M/R3 (99)
4π
and

m/r2  ≈ M/R2 .
Then, from Eq. (98),

Pc 3 M GM

dP/dr = − =−
R 4π R3 R2
3 GM 2
⇒ Pc = , (100)
4π R4
and substituting
ρ into the equation of state:
 5/3  5/3
ρ M 5/3 3 1
Pe = KNR = KNR . (101)
mu μe R5 4π mu μe

Eliminating P between Eqs. (100) and (101), and rearranging for R:

 2/3
KNR 3
R= 5/3
M −1/3 μ−5/3
e ≈ 1400(M/M)−1/3 (μe /2)−5/3 km.
Gmu 4π
(102)
For increasing mass, the white dwarf radius decreases and density increases.

Chandrasekhar Mass Limit

As the mass of a white dwarf is increased, so the density increases and

the non-relativistic approximation will break down (M > 0.2M ). Eventu-
ally the equation of state must be replaced by the ultra-relativistic version
Pe ∝ (ρ/μe )4/3 . If we consider the case where the two equations of state give
the same pressure, we can solve for the density where this transition occurs.
Equating pressures from the UR equation of state and Eq. (100):
 4/3
ρ 3 GM 2
Pe = KUR =
mu μe 4π R4

Expressing the density as above (Eq. (99)), then

70 C. S. Jeffery
 4/3  4/3
1 3 M 3 GM 2
KUR =
mu μe 4π R3 4π R4
 4/3  1/3
1 3 M 4/3 GM 2
⇒ KUR =
mu μe 4π R4 R4
from which R cancels and hence the central pressure becomes indeterminate.
Thus a white dwarf can be supported in hydrostatic equilibrium entirely by
pressure due to degenerate electrons provided those electrons are not fully
relativistic. On exceeding the relativistic limit, the radius and density become
indeterminate and the star cannot support itself. This occurs when the mass
exceeds the Chandrasekhar limit
 3/2  1/2  2  μ −2
KUR 3 1 e
M > MCh ≡ ≈ 1.4 M . (103)
G 4π mu μe 2

13.2 Type II Supernovae

When the mass of a star exceeds MCh , and following exhaustion of all avail-
able nuclear fuel, electron-degeneracy cannot provide sufficient pressure to
maintain hydrostatic equilibrium and the star will continue to contract. In
the core, some electrons will combine with heavy nuclei, since they must find
vacant energy states. However the main result is that the temperature will
rise (Virial theorem!). Several outcomes are possible.

Fe Si, Ne O, C He, H

Fig. 28. Schematic of chemical layering in a massive star immediately prior to core
collapse

Observations show that stars explode as supernovae, releasing ≈ 1043 J in

a very short time. This requires the detonation of ≈ 0.1 M of material via
 Stellar Structure and Evolution 71

proton captures onto light elements at temperatures of ≈ 109 K. This can only
occur if material rich in the necessary material collapses catastrophically.
Consider a star with fairly massive (> 3M ) CO core. Gravitational con-
traction raises T so that reactions such as

2 16 O →28 Si +4 He ⇒ 2 28 Si →56 Fe

may occur, creating a star with approximately the chemical structure illus-
trated in Fig. 28.
Conditions for explosive burning can only occur if the core collapses at a
speed  vs , the local sound speed. In a star with an iron core, vs ≈ 107 m s−1 ,
which implies a collapse timescale ≤ 3 s – essentially a dynamical timescale.
Any slower and the star can reorganize itself without an explosion, so a lot of
energy (≈ 1014 J kg−1 ) must be removed rapidly.
A number of processes, possibly, acting in series may provide this very
efficient refrigeration. The first process (Fig. 29) [33] involves the photo-
dissociation of iron:
56
Fe + γ(2.22MeV) → 12 4 He + 4n,

absorbing ≈ 2 × 1014 J kg−1 . As Tc rises and exceeds a critical value (≈

2 × 109 K), photo-dissociation occurs, absorbing the photo-dissociation energy
and temporarily halting the Tc increase until photo-dissociation is complete.
It is possible that this process alone could trigger a catastrophic collapse of
the core, followed by rapid infall, heating and ignition of proton-rich envelope
material. The resulting explosion could correspond to a type II supernova
outburst.
Additional processes occur when temperatures exceed 109 K which con-
tribute to the physics of a supernova explosion. First, when Tc ≈ 2 × 109 K,
photodissociation of 4 He occurs
4
He + γ(26MeV) → 2p + 2n,

and a similar refrigeration process is repeated (≈ 3 × 1016 J kg−1 ).

By the time, Tc ≈ 5×108 K, the neutrino luminosity is similar in magnitude
to the photon luminosity. At higher temperatures the URCA process is very
efficient:
e+ + n → p + ν̄(+10MeV),
e− + p → n + ν(+10MeV).
The neutrons and protons freed by 4 He dissociation will dramatically increase
the neutrino flux – each neutrino removing 10 MeV from the core and thus
providing a second very efficient refrigeration and contraction agent.
Subsequently, when Tc ≈ 4×1010 K and ρc ≈ 3×1014 kg m−3 , the neutrino
opacity (κν ) becomes non-zero. Neutrinos start to be reabsorbed (the reverse
URCA process) and the temperature will start to increase very sharply.
72 C. S. Jeffery

Fe O, Ne,...

photodissociation − 2.2 MeV

He + n O, Ne,...

photodissociation − 26 MeV

p+n O, Ne,...

URCA − 10 MeV

O, Ne,...

κν>0

reverse URCA

1043J

Fig. 29. Illustration of a possible sequence of processes during core collapse leading
to a type II supernova explosion. See text for more details
 Stellar Structure and Evolution 73

Finally, when Tc ≈ 1011 K and ρc ≈ 1016 kg m−3 , the contractions are

halted by the material becoming neutron-degenerate and incompressible. By
now, the outer layers of the star are effectively in free fall, and converting
gravitational potential into kinetic energy. As this infalling material collides
with the hard incompressible core, a mixture of light elements and protons is
rapidly heated to T ≈ 1011 K, resulting in a runaway thermonuclear detona-
tion.
It is evident that the omission of rotation (in particular) from this simple
picture oversimplifies the true physics of a supernova explosion, but our de-
scription demonstrates how the timescale for the sequence must be extremely
short, ≈ 0.1 s for the entire core collapse and ignition. It also demonstrates
how much power can be generated by the explosion, and that most (99%) of
it is carried by neutrinos rather than photons.

13.3 Neutron Stars

10 m

8 km
H
He
C
q n Fe O
Ne
>18
17.9

log ρ/kg m−3 17.5

14.6
7
Fig. 30. Schematic structure of a neutron star, showing core made up of quarks and
hyperons inside a degenerate neutron layer inside an iron-rich mantle. Approximate
scales and densities are also shown

The debris from a type II supernova explosion includes ejecta rich in heavy
elements which come from reaction of light elements with the mantle of the
iron core and a residual star which is made of neutron-rich material (Fig. 30).
A neutron star is believed to have a dense crust made of normal matter,
but including many heavy elements. At deeper layers, more neutron-rich (and
normally unstable) nuclei are expected and below that, layers containing free
electrons and free neutrons. In the core itself, neutrons and possibly more
74 C. S. Jeffery

exotic elementary particles including hyperons and free quarks will dominate.
The equation of state for neutron-degenerate matter in stars is not yet known,
so that the theoretical radius for a 1.5 M neutron star could be between 10
and 15 km.

14 Horizontal Branch Stars

We have demonstrated that there are two long-lived phases of nuclear-driven
stellar evolution. On the main-sequence, stars burn hydrogen in their cores.
Core-helium burning manifests in more diverse ways, depending on the mass
of the star. In the most massive stars, core helium ignition occurs while the
star is still a blue supergiant; the result is likely a Wolf-Rayet star. For stars
in the 8 – 20 M range, core helium ignition occurs as the star is crossing the
HR diagram, and will produce a yellow supergiant. After completing helium-
burning, such a star will start to burn carbon, then silicon and ultimately
become a supernova.
Stars less massive than about 8 M will avoid the supernova fate. In the
3 – 8 M range, helium ignition occurs in a non-degenerate core, so is not
explosive. Core helium burning is associated with a blueward loop through the
Cepheid instability strip, after which the star develops a double shell structure
and becomes an asymptotic giant branch star.

The helium flash

At lower masses (< 2M ), a star develops an electron-degenerate helium core

as it ascends the giant branch. This core does not increase in temperature
as the pressure increases (P ∝ ρ5/3 ) so will not ignite helium so easily. The
hydrogen-burning shell adds processed helium to the core, and itself get pro-
gressively hotter and thinner as the mass of the remaining hydrogen envelope
is reduced. In principle, the core should be isothermal with the hydrogen-
burning shell (≈ 108 K). In practice, pair-production of neutrinos in the dense
e− -degenerate core removes heat and keeps the center of the core cooler than
the outer regions.
When the hottest part of the core reaches a critical temperature (≈ 108 K),
a 3α reaction becomes increasingly likely. This occurs when the core mass
reaches about 0.48 M . Energy is liberated and provides immediate local heat-
ing. Since the local equation of state is insensitive to temperature, runaway
heating occurs and sustained helium-burning reactions begin. These stabilise
once the temperature has risen sufficiently to lift the local degeneracy and the
helium-burning flame has propagated into the centre of the star.
The details of the helium flash are still poorly understood. It is clear
that very short timescales are involved, and that nuclear-driven convection
mixes some layers of the star. Realistic 3D models of the helium-flash are
now becoming possible [34]. It is not clear how much of the star is mixed,
 Stellar Structure and Evolution 75

or whether the stellar surface reflects the violence of the helium ignition in
any way. It seems unlikely since the initially degenerate core has a very high
heat capacity; helium burning will force the core to expand, and this will
quench (but not extinguish) the hydrogen-burning shell, so the overall star
will contract.
Since the mass of the core at helium ignition depends little on prior con-
ditions, the dimensions of the star during stable core helium burning are
primarily a product of the mass (MH ) and composition (Ye , Ze ) of the hydro-
gen envelope. The envelope composition should reflect the initial composition
(Y, Z), but dredge-up of either helium or nitrogen during first giant branch
evolution could be significant.
Low-mass helium-core burning stars are identified with horizontal branch
(HB) stars in globular clusters.

14.1 Horizontal Branch Stars in Clusters and the Field

In the HR diagram for Hipparcos stars (Fig. 1), the giant branch will be seen
to contain a dense clump of stars centered at B − V = 1, Mbol = 2. Clump
stars are essentially low-mass helium-core burning stars belonging primarily to
Population I, i.e. they are metal-rich and have significant hydrogen envelopes.
In metal-poor environments, such as globular clusters and field stars at
high galactic latitudes, these stars appear on a horizontal strip extending
blueward of the giant branch. There is a clear correlation between position
on this horizontal branch and cluster metallicity. Compare, for example, the
horizontal branches in ω Cen ([Fe/H]= −1.62: Fig. 5) and 47 Tuc ([Fe/H]=
−0.76: Fig. 4).
The Second-Parameter Problem
While the location of stars on the HB is normally a function of the cluster
metallicity (Figs. 4 & 5), this is not always true as Figs. 31 & 32 show. Other
cluster pairs with similar metallicity but quite different horizontal-branch mor-
phologies include M2 and M3, both with metallicity [Fe/H]= −1.6, and Arp
2 and Rupprecht 106 with [Fe/H]= −1.7.
There must be another factor at work in these cases – sometimes known
as the second-parameter problem for globular clusters. The resolution may
have something to do with cluster age, and hence the degree of mixing that
stars currently on the HB experienced during first dredge-up, since this corre-
lates with total mass of the star leaving the main-sequence. Other suggestions
include cluster density, which may influence the average rotation rate, the
degree of mixing and the helium core mass.

14.2 Theoretical Models for Horizontal Branch Stars

Since HB stars are in a long-lived phase of evolution, the time- dependent
terms in the stellar structure equations are negligible and equilibrium solutions
76 C. S. Jeffery

Fig. 31. Colour-magnitude diagram for the metal-poor galactic globular cluster
M13 (Z = 0.0007, [Fe/H] = −1.54; based on [35])

Fig. 32. Colour-magnitude diagram for the metal-poor galactic globular clus-
ter M3, its red horizontal-branch illustrating the 2nd parameter problem (Z =
0.0007, [Fe/H] = −1.57; based on [35]
 Stellar Structure and Evolution 77

provide an excellent approximation to the overall structure. Moreover, the

chemical structure of a zero-age HB star is quite simple, having a hydrogen-
rich envelope of mass MH surrounding a helium-rich core of mass Mc . If it is
further assumed that both core and envelope are homogeneous with Yc = 1−Z
and Ye = Y, Ze = Z (Y, Z being the initial composition), then a sequence of
models may be computed for a given Mc , Y, Z and a range of MH .
In practice, it is necessary to consider how the envelope may have been
modified by convective dredge-up while the star was on the giant branch. Thus
Ye > Y though dredge-up of helium, and the relative CNO concentrations
within Z may have been altered. Both should correlate with Mi , the initial
mass of the star. Another consideration is the extent and location of core
carbon enrichment Cc during the core helium flash. An estimate of this can be
derived by computing the carbon required to provide the energy required to lift
the core electron degeneracy, and the position of the temperature maximum
immediately prior to core helium ignition. For models with very small MH
it is necessary to consider carefully the composition across the core/envelope
boundary. For large MH , this should initially correspond to the profile of the
H-burning shell in the precursor red giant.
In summary, HB models are defined by:
Mc : Helium core mass,
M : Total mass (or envelope mass MH = M − Mc ),
Ye : Envelope helium abundance,
Ze : Envelope metallicity, and
Cc : Core carbon abundance and distribution
Models of HB stars as a function of MH not surprisingly form a roughly
horizontal sequence in the L − Teff diagram. The helium core provides a lu-
minosity of some L ∝ Mcβ largely irrespective of MH 10 .
At low mass, the hydrogen envelope is essentially inert, simply providing a
high opacity envelope which makes the star larger than it would be otherwise.
Once MH exceeds some threshold, the base of the envelope becomes suffi-
ciently hot to sustain hydrogen burning by the CNO cycle, prompting both
an increase in radius and luminosity. Since increasing radius means decreasing
Teff , the models become cooler until they merge with the Giant Branch.
Models of HB stars have been computed by many researchers, including
[10]. The definitive set in common usage [36] demonstrate the locus of the HB
as a function of both Mc and Ye .

The MS – HB mass discrepancy

If the metallicity of an HB star is known, and a reasonable estimate of the he-

lium abundance can be made, then the luminosity and effective temperature
allow a realistic deduction of the total mass. In most globular clusters, this
gives answers of ≈ 0.6M . In the same clusters it is also possible to estimate
10
In fact, setting MH = 0 leads us to the helium main sequence, which is equivalent
to (and roughly parallel with) the hydrogen main sequence.
78 C. S. Jeffery

the mass of stars at the Turn-Off point, i.e., those stars which are just end-
ing their main-sequence evolution. Invariably, these turn-off stars are more
massive than the HB stars in the same cluster. Since the current HB stars
must have left the MS before the current epoch and therefore have been more
massive than the current turn-off stars, substantial mass must be lost either
while the star is on the giant-branch or at helium ignition.
The discovery of substantial stellar winds in red giants appears to account
for most of the discrepancy. In some cases, these winds appear able to remove
almost all of the envelope hydrogen, though theory suggests that the hydrogen
shell will be quenched and the star will evolve to become a helium white dwarf
if the envelope mass drops below MH ≤ 0.01M before the core ignites helium
[37, 38, 39].

14.3 Evolution of Horizontal Branch Stars

For classical HB stars containing a strong hydrogen-burning shell, initial evo-
lution is defined by a balance between He-core and H-shell. As core helium
drops and the mean molecular weight increases, the core temperature rises
to maintain hydrostatic equilibrium, so the helium core becomes more lumi-
nous and expands. The core expansion forces the hydrogen-shell outwards,
so its temperature and luminosity initially drop, and the overall luminosity
stays roughly constant. However, the total radius (and hence effective tem-
perature) drops because of the weakening of the H-burning shell, so the star
evolves horizontally to the blue in the HR diagram. Towards the end of core
helium-burning, this process reverses, the core-expansion is reduced and power
returns to the H-burning shell, causing the star to evolve back to the red and
to higher luminosity. Complete core helium exhaustion is followed by core
contraction and hydrogen-shell thinning which moves the star towards the
asymptotic giant branch. As what is now a carbon-oxygen core contracts it
becomes electron-degenerate. The increasingly powerful H-burning shell adds
helium to the layer between the CO core and the H-envelope (the intershell).
When sufficient helium has accumulated in the intershell, it ignites in an un-
stable process (i.e. nearly explosively) which leads to intershell expansion,
thermal pulses and other interesting phenomena (see Karakas: these proceed-
ings).
Semiconvection
Core-helium burning is always associated with a nuclear-driven convective
core. Because the temperature-sensitivity of the 3-α process is so steep, the
mass of the convective core increases with falling central helium abundance.
Assuming conventional convection theory the outer edge of the convective
core, with a molecular weight increasing due to its increasing carbon abun-
dance, will drive into unburnt helium with a lower mean molecular weight.
This will create a sharp discontinuity in molecular weight at the convective-
core boundary. While the upward buoyancy of a convective element in the
 Stellar Structure and Evolution 79

carbon-rich core approaches equilibrium close to the boundary, it would re-

main unstable were it to penetrate above the boundary. This has led to con-
siderable discussion about how to treat convection (or semiconvection) at the
horizontal-branch convective-core boundary. The overall result of including
some form of convective core extension is to increase the extent of the con-
vective core and hence the lifetimes of horizontal branch stars by some 20%.

14.4 Extreme Horizontal Branch Stars

Extreme horizontal-branch (EHB) stars are horizontal-branch stars with a

hydrogen envelope whose mass is insufficient to sustain hydrogen burning;i.e.
the H-burning shell is inert. They can be thought of as helium main-sequence
stars with a very thin hydrogen envelope (0.0001  MH /M  0.01).
The evolution of an EHB star is dominated by that of the core; the very
thin H-envelope plays a passive rôle. Thus the total luminosity increases dur-
ing evolution and the star follows a nearly vertical track in the HR diagram.
Following core-helium exhaustion, there being no H-burning shell, the mass
of the helium-rich intershell remains fixed. A few mild helium-shell ignitions
may occur11 , but the star is ultimately doomed not to reach the asymptotic
giant branch. Instead it will contract directly to become a hybrid He/CO
white dwarf.

Faint Blue Stars in the Galactic Disk

The observant reader will realise that the existence of an EHB star with
MH  0.01M is inconsistent with the minimum hydrogen mass anticipated
assuming conventional mass loss from a red giant (MH ≥ 0.01M). Thus the
existence of real EHB stars would appear to be impossible assuming standard
single-star evolution.
On the other hand, EHB stars have been identified with a very important
class of stars known as subdwarf B (sdB) stars. These were originally identi-
fied as faint blue stars in the galactic field. Not being associated with spiral
arms or OB associations, they had to be older than typical O and B stars.
Measurement of their surface gravities showed them to lie below the upper
main-sequence, and it was realised that they must be low-mass stars, and
identified with the extreme blue end of the horizontal branch [40]. Their dis-
tribution in the Galactic plane is consistent with a relatively old population –
they occur up to high galactic latitudes and most probably belong to the thick
disk.

11
the existence of so-called “breathing pulses” towards the end of horizontal-branch
evolution may be a numerical artefact due to incomplete treatment of core con-
vection
80 C. S. Jeffery

Extreme Horizontal Branch Stars in Clusters

Extremely-blue horizontal-branch stars were subsequently discovered in a

number of globular clusters, notably NGC 6752 [41, 42], NGC 2808 [43] and
ω Cen [44, 45]. Notably more difficult to execute than comparable studies in
the field, high-resolution spectroscopy demonstrated that cluster EHB stars
and field sdB stars are comparable. The advantage of cluster stars is that they
provide a sample at the same distance and, ostensibly, of the same age.
Both field and cluster sdB stars are predominantly hydrogen-rich. That is,
their surface helium abundances have been suppressed below typical values
for B stars by factors of 10 – 100. This can be explained by the high effective
temperature and surface gravity of these stars, which makes chemical diffu-
sion an important transport process, mediated by gravitational settling and
radiative levitation. In the case of helium, gravity normally wins and it sinks
out of sight beneath the hydrogen. However, there also exists a small fraction
(≤ 5%) of sdB (and subdwarf O) stars with very high overabundances of he-
lium, such as can only occur if nuclear-processed material is exposed on the
stellar surface. These He-rich stars may be more prevalent in some globular
clusters than in the field (ω Cen: [46]).

14.5 The Origin of EHB Stars

EHB and sdB stars pose challenges for stellar evolution theory in how to
account (a) for their very low hydrogen-envelope masses and (b) for the range
of surface helium abundances. Specifically, what mechanism removes almost
the entire hydrogen-envelope from a red-giant star? How does it still achieve
helium ignition? And how should we explain the existence of helium-rich sdB
stars?

sdB Binaries

Some answers to these questions were provided by discoveries that many sdB
stars are members of binary systems. Observationally, three (or four) distinct
groups are identified [47]:
I. sdB stars with single spectra showing no radial velocity changes (30-35%),
II. sdB stars with composite spectra (usually sdB+G) showing small or no
velocity variations. Orbital periods are thought to range from tens of days to
several years (15-20%),
III. sdB stars with single spectra showing large velocity variations with periods
of hours to days (45-50%).
The last group may be further divided into:
IIIA. those having a white dwarf or other compact companion,
IIIB. those having a low-mass main-sequence star companion.
The presence of a second star and a short-period orbit in group III require a
prior interaction between a red giant and its companion in order to explain
 Stellar Structure and Evolution 81

why the present-day orbit is smaller than the radius of the giant that birthed
the sdB star. Assuming that the existence (or otherwise) of a companion
contributes to the formation of the sdB star, it does appear that nature can
make the latter in at least four different ways.

Formation Channels

The problem of whether an EHB star can ignite helium at all has been ad-
dressed [39, 48, 49]. Models of red giants were computed which include an
enhanced mass-loss rate near to the tip of the giant branch, i.e. when the core
mass is slightly less than the critical value for 3-α ignition. In these models,
the hydrogen shell is extinguished and the star starts to evolve to the blue,
i.e. contraction towards the white dwarf sequence. Depending on conditions
when the star left the giant branch, these models show an off-center ignition
of helium either during blueward evolution or after the star reaches the white
dwarf cooling track. This off-center ignition creates a helium-burning shell
which raises the local electron degeneracy and burns its way into the core
via a series of mild helium shell flashes. Meanwhile the star evolves through
a series of loops to a position on the extreme horizontal branch. The overall
conclusion is that if a star reaches within 0.25 magnitudes of the RGB tip, he-
lium ignition will occur. The final location on the zero-age horizontal branch
only depends on the hydrogen envelope mass.
The question raised by this result is what mechanism can produce an
enhanced mass-loss rate close to the red-giant tip? Two have been proposed
that involve binary star evolution [50, 51].

Common Envelope Ejection

In the first binary-star formation channel, the sdB progenitor is a giant that
fills its Roche lobe (the gravitational equipotential surface surrounding both
stars that meets at a single point of contact between them). If it does this
near the tip of the giant branch, and if mass ratio of the progenitor mass to
that of its companion is larger than ≈ 1.2, then there will be mass transfer
on a dynamical timescale (Eq. (6)) from the more massive star. The reason is
that as mass is transferred, the mass ratio changes and (in this case) the orbit
will shrink. The ratio of the massive star radius to the Roche lobe radius and
the rate of mass transfer will both increase in a runaway process. Meanwhile
the secondary star cannot assimilate this new material quickly enough, which
will fill up and spill over the Roche surface to form a common envelope around
both stars.
With two stars orbiting one another inside this common envelope, friction
between the stars and envelope, which will not be rotating at the same rate,
will cause the two stars to spiral towards each other. Orbital energy will be
converted to heat until sufficient has been lost to eject the envelope [52]. The
result is a much closer binary (orbital period typically between 0.1 and 10 d)
82 C. S. Jeffery

consisting of the red-giant core and the secondary. If the core ignites helium
(see above) it can evolve to become an sdB/EHB star.
The secondary star may be either a main-sequence star or a white dwarf.
In the first case, the sdB star was originally the more massive star in the
binary system, since it has evolved to reach the GB first. The less massive
star must have M ≤ 0.4M in order for mass transfer to be unstable, and is
therefore likely an M-type star.
If the secondary is a white dwarf, then the sdB star must originally have
been the less massive star in the binary system. A previous phase of stable
mass transfer must have transferred the envelope of the original primary to
the secondary, allowing the core to become a helium white dwarf and revers-
ing the mass ratio in the binary system. The common envelope phase occurs
when the new primary evolves to become a red giant, and the resulting system
contains an sdB star and a white dwarf.
One challenge encountered by this scenario is that the mass predicted for
white dwarf companions in short-period sdB binaries should be < 0.5M .
Recent observations have demonstrated several sdB+WD binary systems in
which substantially higher masses are required for the unseen companion. In
some cases these masses exceed the Chandrasekhar mass limit for white dwarfs
(MCh : Sect. 13) and are a continuing challenge for stellar evolution theory [53].

Stable Roche Lobe Overflow

In the case that the mass ratio is ≤ 1.2, a similar binary to that considered
above will experience stable mass transfer [54]. With a lower initial mass ratio,
the exchange of angular momentum will lead the binary to widen and the rate
of transfer to remain low enough for matter to be assimilated comfortably by
the companion. The result will be an sdB star with a main-sequence or sub-
giant companion of increased mass and an orbital period of tens to hundreds
of days.

White Dwarf Mergers

An alternative channel for the formation of sdB stars considers the ultimate
fate of binary systems which are so evolved as to comprise two white dwarfs
in a close binary orbit. This predicates a binary system which has passed
through several stages of mass transfer (whether common-envelope or stable
Roche-lobe overflow) to leave these remnants.
In such a long-lived short-period binary, the emission of gravitational waves
can remove orbital energy from the system so that over a long time the orbit
will shrink until the stars are in contact. The merger timescale is given by

tmerge = 107 (P/h)8/3 μ−1 (M/M )−2/3 yr (104)

where P is the orbital period (in hours) and μ is the reduced mass of the
binary [55, Sect. 110],[56]. It will be seen that binaries with periods ≤ 2 h can
potentially come into contact within a realistic time frame.
 Stellar Structure and Evolution 83

Upon contact, the less massive white dwarf will have a larger radius
(Sect. 13) and will fill its Roche lobe first. Upon transferring a small amount
of mass, the white dwarf radius will increase at a rate faster than the binary
orbit will widen, and at a rate much faster than matter can be accreted by
the companion. This leads to a runaway (or dynamical) disruption of the less
massive white dwarf which will form a hot disk in Keplerian orbit around
the other more massive white dwarf. Simulations of the process show that the
disruption of the white dwarf will take only a few orbits, Porb ≈ 3 m at the
time of contact [57, 58, 59].
What happens next is a matter of some conjecture. Heating of the disk
may initiate some prompt nucleosynthesis [59, 60]. Viscosity in the dense
disk will drive angular momentum to the outer edge of the disk, and mass
will therefore fall inwards [61]. The result will be a high entropy envelope
surrounding the relatively unperturbed massive white dwarf. The envelope
will cool and material at its base will be assimilated onto the surface of the
underlying core. The next step depends on the initial white dwarfs.
When the mass of the helium-layer accreted into the low-entropy core
is sufficiently high, helium ignition will occur at the boundary between the
non-degenerate and degenerate material. The star will expand on the short
thermal timescale of the envelope (≈ 103 y) to become a yellow supergiant –
a stage which will be described in more detail in Sect. 15.7.
If both white dwarfs were helium white dwarfs [62], the helium-burning
layer will have unprocessed helium beneath. Heating from the nuclear reac-
tions will lift the degeneracy of the top layers of the helium core and allow
the helium-burning shell to burn into the center in a series of mild pulses.
At this point the star should resemble an extreme horizontal branch star.
Although the surface layers will be predominantly composed of helium from
the disrupted white dwarf, there should be sufficient hydrogen to diffuse to
the surface (helium will sink) to form the helium-poor atmosphere usually ob-
served in sdB stars. Of course, this channel will always lead to the formation
of a single EHB star.

Other formation channels

Recently, another channel has been proposed for forming single sdB stars
[63] following the merger of a red giant with a lower-mass main-sequence or
brown-dwarf companion during a common-envelope phase of evolution. This
route avoids the surface composition problem, but depends on an epoch of
enhanced mass loss following spin-up by the merger. In this respect it bears
some similarity to the single-star evolution channel, and provides a mechanism
for the enhanced mass-loss near the tip of the giant branch. However, the
expected contribution is too small to explain all observed single sdB stars.
Another merger model suggests that some post-common-envelope sdB bi-
naries could share the white dwarf merger fate [64]. The idea is that the
post-sdB star, which is a hybrid white dwarf containing a partially-burnt CO
84 C. S. Jeffery

core with an unburnt but degenerate helium layer, merges with its less-massive
helium white dwarf companion. The new star is a helium shell-burning star,
but the CO core is of sufficiently low mass that the star’s evolution and radius
is more strongly determined by the total mass of helium. Both merger routes
might equally account for the significant number of hydrogen-poor sdB and
sdO stars discussed at greater length in Sect. 15, and which are also located
around the extreme blue end of the horizontal branch [65, 66].
Work has also focused on mixing processes associated with violent burning
events, including extra mixing at convective boundaries, diffusive mixing and
mass loss. For example, it has been proposed that deep mixing driven by a He-
flash convection zone can engulf a H-rich envelope and lead to the formation
of He-rich sdO stars [67, 68].

14.6 Extreme Horizontal Branch Stars in other Galaxies

The significance of EHB stars in a cosmological context may be recognised

from resolved observations of nearby stellar systems and from the integrated
flux distributions of more distant systems [69]. Significant numbers of EHB
stars in globular clusters, including ω Cen and NGC 6752, have been noted.
Meanwhile the nearby spiral galaxy M31 shows a large EHB population very
similar to that of ω Cen.
Like globular clusters, giant elliptical galaxies are old star systems which
should principally comprise low-mass main-sequence stars, red giants and
white dwarfs, and no young blue massive stars; they are known as “old red
dead” galaxies. Therefore it was a surprise to discover several of these galaxies
show an upturn in the flux distribution at wavelengths shorter than 2000 Å
[70,71]. By creating various models for the population of these galaxies, it has
been demonstrated (as originally conjectured) that this ultraviolet excess is
due to a substantial population of EHB stars [72, 73].
Such results are significant for studies of stellar evolution. EHB stars are an
important diagnostic of age in old populations – it should be possible to use the
ultraviolet upturn as an age discriminator for distant galaxies. However, this
will depend on a clear understanding of the formation channels for EHB stars
at different epochs and metallicities in galactic evolution. Population synthesis
calculations which predict relative numbers of stars and predict emergent
spectra for galaxies of different ages and metallicities must adopt the best
possible physics to achieve this [74].

15 Late Stages of Stellar Evolution: Hydrogen-Deficient

Stars
As the section on the evolution of horizontal-branch stars has shown, the story
of stellar evolution is not a simple progression from formation to death as a
white dwarf or supernova. The late stages of evolution are frequently marked
 Stellar Structure and Evolution 85

by stars which exhibit remarkable properties – one of the most significant

being a near-complete absence of surface hydrogen [11].
The science of hydrogen-deficient stars began when Williamina Fleming
noted that “the spectrum of υ Sgr is remarkable since the hydrogen lines are
very faint and of the same intensity as the additional dark lines” [75]. Fifteen
years later observed Hγ was found to be completely absent from the spectrum
of R CrB [76].
These early observations pointed to something unusual in the spectra of
a variety of stars, but there was reluctance to accept (or even suggest) that
hydrogen might be deficient [77,78,79]. It was not until the 1930’s that quan-
titative spectral analyses forced the conclusion that in R CrB [80] and υ Sgr
[81, 82] “somehow, a very substantial amount of hydrogen had been lost”.
Forty years ago, very few hydrogen-deficient stars were known [114, 115],
and the distinctions between them were poorly understood [116]. With the
benefit of hindsight and on-line databases, it is now easier to explore their
history and identify the early landmarks (Table 7). Meanwhile, large-scale
spectroscopic surveys have produced a torrent of new stars with hydrogen-
deficient spectra (Table 8). Broadly, these hydrogen-deficient stars may be
divided into massive stars, low-mass supergiants, hot subdwarfs and white
dwarfs.

15.1 Population I and Massive Hydrogen-Deficient Stars

Wolf-Rayet Stars.

Wolf-Rayet (WR) stars were identified as peculiar from the bright bands seen
on the continuous spectra of three “small” stars in Cygnus [83] which in
retrospect were identified as emission lines due to ionised atoms including
helium. By 1894, some 55 WR stars were known, most of which had been
discovered by Fleming [84]. Remarkably, there was debate as to whether these
stars were either pre-main sequence or highly-evolved stars through the 1960’s,
while disagreement on whether they were H-deficient persisted into the 1980’s.
WR stars are found solely in spiral arms, OB associations and young clus-
ters, and hence are associated with massive star evolution. They can be clearly
divided into two sequences, one showing nitrogen-rich spectra, the other being
carbon-rich. Numerically, some 230 WR stars are known in the Galaxy, 159
having mV < 15 [118]. Hydrogen has been detected in about half of those
analyzed in detail. They are not exclusive to metal-rich environments since
about 100 have been found in the Large Magellanic Cloud, and a dozen in the
Small Magellanic Cloud.
The strong emission lines provide a clue that WR stars represent a normal
stage in the evolution of massive stars. Very high luminosities lead to high ra-
diation pressure and hence to very strong stellar winds. As the star evolves to-
wards and beyond the end of main-sequence hydrogen burning, high mass-loss
from the stellar surface exposes first nitrogen-rich material (CNO-processed
86 C. S. Jeffery

Table 7. Various classes of hydrogen-deficient star, with representative prototypes

and discovery paper

Prototype Class No. Discovery

Population I
V1679 Cyg Wolf-Rayet ≈ 230 [83]
υ Sgr H-deficient binary 5 [84]
σ Ori E Intermediate ≈ 30 [85]
h̃elium B
SN 1983N, 1984L SN Ib ≈ 80 [86]
SN 1987M SN Ic ≈ 120 [87]
Low-mass supergiants
R CrB R CrB ≈ 50 [88, 76]
HD 182040 H-deficient carbon 5 [89, 90]
HD 124448 Extreme helium B 17 [91]
MV Sgr Hot RCrB 4 [92, 93]
FG Sge Born-again 3 [94]
Hot subdwarfs
PG1544+488 sdOD / He-sdB ≈ 50 [95, 96]
BD+75◦ 325 compact He-sdO ≈ 50 [97]
BD+37◦ 1977 low-g He-sdO 5 [98]
Central stars of planetary nebulae
BD+30◦ 3639 [WC] ≈ 50 [99, 100]
A66 30 Of-WR(C) 2 [101]
PG1159–035 O(C) ≡ PG1159 ≈ 40 [102]
K 1-27 O(He) 4 [103]
White dwarfs
HZ 21 DO ≈ 50 [104]
L 930-80 DB ≈ 400 [105]
HZ 43 DC ≈ 360 [106]
DQ ≈ 120
DZ ≈ 80
AM CVn AM CVn binary 21 [107]

helium in which carbon and oxygen have been converted to nitrogen) and
then carbon-rich material (3α-processed material). These give rise in turn to
the WN and WC subclasses.

He-rich B Stars or Intermediate Helium Stars.

σ Orionis is a tightly bound group of young OB stars, with a total mass of

some 100 M or more. Berger [85] noted the spectrum of star E to show a
series of excessively strong lines of neutral helium, extending to a continuum
discontinuity on the ultraviolet side of the Balmer series limit. Subsequent
investigation has shown σ Ori E and similar stars to be chemically peculiar
 Stellar Structure and Evolution 87

Table 8. Major surveys yielding hydrogen-deficient stars

Survey Reference
HZ [106]
F Feige [108]
FB Faint Blue Stars [40]
PG Palomar-Green [95]
BPS HK Objective-Prism [109]
HS Hamburg-Schmidt [110]
HE Hamburg-ESO [111]
EC Edinburgh-Cape [112]
SDSS Sloan Digital Sky [113]

main-sequence B stars. Some 24 are catalogued [114], which may be further

divided into fast and slow rotators. Measurements of the helium/hydrogen ra-
tio show that the apparent helium abundances vary with a period of 1–10 d. In
σ Ori E, the helium anomaly is associated with a dipole magnetic field of some
104 G inclined ≈ 90◦ to the rotation axis; the magnetic caps are metal-poor.
It is generally agreed that the surface abundance distribution is governed by
diffusion, where both radiative and magnetic fields act selectively to concen-
trate particular elements in the line-forming region at specific locations on the
stellar surface e.g. [119].

H-Deficient or υ Sgr Binaries.

Despite an unusual spectrum [84], it was large radial velocity variations that
really drew attention to υ Sgr [120]. The star remained unique until [121] rec-
ognized its similarity to HD 30353 = KS Per. Two similar stars (V426 Car and
V1037 Sco) were discovered during a survey of OB+ stars [122]. BI Lyn was
recognized as the fifth “hydrogen-deficient binary” when it was accidentally
observed in a survey of subdwarf B binaries [123]. Remarkably, V426 Car had
first been noted as a faint star with a spectrum “of the fifth type” (meaning
it shows emission lines) in the former constellation of Argo in 1892 [124].
The chief characteristics of the class are strong helium lines on a metallic
spectrum, and radial velocity variations of several tens of km s−1 . The orbital
periods range from ≈ 50 − 360 d. Most of these hydrogen-deficient binaries
show emission at Hα and Hβ and also evidence of pulsations.
From high-resolution ultraviolet spectra, [125] showed that υ Sgr is a
double-lined system, and hence that the primary has a mass ≈ 3 M and
luminosity ≈ 105 L . With systemic radial velocities close to those for circu-
lar orbits around the galactic centre, and with all these stars lying less than
200 pc from the galactic plane, the hydrogen-deficient binaries are clearly pop-
ulation I stars, probably with M > MCh , and are excellent candidates for the
progenitors of SN Ib/c [126, 125].
88 C. S. Jeffery

Early models suggest at least two phases of mass transfer necessary to

explain the very low surface hydrogen abundances e.g. [127]. The current best
model for the evolution of υ Sgr [9] is that it began as a 10+3 M, 150 d binary
in which the envelope was blown to infinity, with little change of orbit, as the
more massive star approached both Roche Lobe overflow and the Cepheid
instability strip simultaneously.

H-Deficient or Type Ib and Ic Supernovae.

The Type II supernovae discussed in Sect. 13 all contain hydrogen in their

ejecta and are universally attributed to the explosions of massive stars. Type
I supernovae (SN I), by contrast, come in a greater variety. Two classes show
no evidence of hydrogen absorption lines and include those which lack the
ionized-silicon absorption line (SN Ib) and those which further lack helium
lines (SN Ic). As SN Ib age, their spectra show stronger features of helium
than SN Ia, until eventually they contain emission lines due to light elements
(e.g. oxygen, calcium and magnesium), while SN Ia spectra become dominated
by iron emission [128].
The crucial distinction between supernovae of types Ia and Ib/c is that
the latter occur near star formation sites and have never been observed in gi-
ant elliptical galaxies. Consequently they are associated with core-collapse
explosions in massive or intermediate mass stars which have already lost
their hydrogen envelopes. Possible candidates include Wolf-Rayet stars and
hydrogen-deficient binaries [126, 129].

15.2 Low-Mass Hydrogen-Deficient Supergiants

The remaining classes of hydrogen-deficient star are associated with very late
stages of evolution, either low-mass stars approaching the horizontal-branch,
or more massive stars which have already past the asymptotic-giant branch
and are evolving to become white dwarfs, or yet older stars which have been
white dwarfs and somehow found a second life. The first to consider are the
low-mass supergiants.

R CrB stars, including H-def C stars.

The R CrB stars are exceptional in that the class is defined by their dramatic
light variability rather than by a spectrum. Characterised by spectacular dim-
mings of five or more magnitudes in the course of days, followed by a slow
recovery to maximum light, the R CrB phenomenon is primarily a signature of
surface activity, rather than chemical peculiarity. Many discoveries were made
from inspection of photographic sky patrol plates e.g. [130]. Patrols searching
for microlensing events and planetary transits continue to discover new R CrB
stars, particularly in the Magellanic Clouds [131].
 Stellar Structure and Evolution 89

Fig. 33. g − Teff diagram for several classes of low-mass hydrogen-deficient stars
described in the text. Different panels identify contemporary evolutionary proposals,
not necessarily definitive or exclusive sequences. In the electronic version, classes are
separated by colour. Data have been collated from a wide variety of sources

The majority of R CrB stars would be of spectral types F and G, except

that they are missing Balmer lines and the molecular CH bands are replaced
by the Swan bands of C2 , indicating an absence of hydrogen and a super-
abundance of carbon. Approximately 40 are known in the Galaxy, with some
17 in the Large Magellanic Clouds. The overall distribution is similar to that
of the Galactic bulge or of an old thick disk. Assuming MV ≈ −5 based on
Magellanic Cloud R CrB stars, it is estimated there may be up to 400 R CrB
stars in the Galaxy.
In addition to their irregular fadings thought to be due to directed mass-
ejections, R CrB stars show low-amplitude variability with periods of 30 – 60
d and associated with pulsations. All show a substantial infrared excess at-
tributed to a warm dust shell, while a few show evidence of a more extended
nebula. It will be seen that these stars are extraordinarily rich in exotic be-
haviour [132].
While R CrB stars were discovered because of their dramatic variability,
it was fortuitous that HD 182040 was chosen for the Harvard spectral type
R0, since both the absence of hydrogen lines and CH bands was subsequently
noticed with ease [89, 90]. A further three stars with similar spectra were
90 C. S. Jeffery

subsequently confirmed to be severely hydrogen deficient [133,134]. Since these

stars are spectroscopically similar to the R CrB stars, possibly showing small-
amplitude variations [135] but missing the deep minima and infrared excesses,
the two groups have been collectively identified hydrogen-deficient carbon
stars [134]. Whether the absence of mass-ejection episodes is characteristic of
a lower luminosity, a lower carbon abundance, and/or an earlier evolutionary
age, has yet to be established.

Extreme helium stars and hot R CrB stars.

In a paper of thirteen lines, Popper [91] announced that spectra of the B2 star
HD 124448 (obtained at an elevation of 13◦ ) “show no hydrogen lines in either
emission or absorption, although helium lines are sharp and strong”. Lines of
O II and C II were also present. Popper also remarked that “faint early B-type
stars so far from the galactic plane are rare”. In the Henry Draper catalogue,
only HD 168476 (B5p) and HD 160641 (O9.5Iap) share properties with HD
124448. Subsequent surveys of luminous blue stars have identified a total of
17 bona fide extreme helium stars [115] – essentially spectral-types from late
O to A with weak or absent hydrogen lines, strong, sharp neutral helium lines,
strong C II and N II. Being low-mass supergiants, their high luminosities mean
that galactic surveys are essentially complete for these stars – those remaining
to be discovered will be highly reddened and on the far side of the Galactic
centre. Some, but not all show low-amplitude pulsations.
The kinematics of the R CrB, cool HdC and EHe stars are all similar, being
those of an old stellar population strongly concentrated towards the Galactic
Centre. They do not share the galactic rotation [136,137]. UW Cen and U Aqr
may be in halo orbits. Such a distribution is consistent with a wide range of
metallicity.
Mention should be made of four “hot R CrB” stars. MV Sgr was discovered
photometrically with cooler R CrB stars [92] and shows the typical light curve
and an infrared excess. It has a weak emission spectrum dominated by double-
peaked Fe II lines, and an absorption spectrum very similar to HD 124448 [93].
The group is quite heterogeneous [138], with MV Sgr and DY Cen possibly
representing evolved R CrB stars, and V348 Sgr and HV 2671 (in the LMC)
being something altogether different.

“Born-again” stars.

Three hydrogen-deficient stars stand out because they have been observed
to evolve from one corner of the HR diagram to the other, and back, within
a human lifetime. FG Sge came to attention [94] as a new blue star as it
brightened from being a hot white dwarf in the late 19th century to become a
cool yellow supergiant in the late 20th century [139]. In even more spectacular
fashion, V4334 Sgr (Sakurai’s object) [140, 141] brightened from a faint blue
star with mj = 21 (V > 15 in 1994) to reach V = 11 in late 1995 and to
become a yellow supergiant in 1996 [142]. Embedding itself in a thick dust
 Stellar Structure and Evolution 91

cloud it had virtually disappeared by 1999, but radio measurements suggest

that the star has now begun a blueward evolution, with Teff ≈ 12 000 K in
2006 [143]. Meanwhile the very hot central star of the planetary nebula A 58
had exhibited a nova-like outburst in 1919 [144]. Examination of historical
material and exploration of the current nebula suggest that this star followed
a rapid “red-giant excursion” similar to that of V4334 Sgr [145]. In all three
cases, the photosphere is hydrogen-deficient. V4334 Sgr appears to have lost
its hydrogen during its redward evolution [146]; when FG Sge and V605 Aql
became hydrogen-deficient is less clear [147, 148].
These stars demonstrate that somehow an extraordinary explosion can
occur inside a hot white dwarf and that this is not necessarily a rare event.
With three such “born-again” stars observed in a century, a significant fraction
of stars that become white dwarfs must go through this phase.

15.3 Hydrogen-Deficient Subdwarfs.

The discovery that certain faint blue star-like objects (quasars) might be
used to probe the farthest reaches of the universe [149] created an industry
in blue sky surveys (Table 8). Luckily for stellar science, these surveys are
strongly contaminated by all sorts of evolved stars, including white dwarfs, hot
subdwarfs and cataclysmic variables. Many of these are hydrogen-deficient,
including the PG1159 stars, and various classes of helium-rich subdwarf and
white dwarf.
The designations sdO and sdB were introduced for stars with spectra re-
sembling MK types O and B but with broadened lines of H, He I and He II
[104]. The term sdOB introduced for sdB stars which show He II 4686 [150]
has been falling out of use in recent years. The hydrogen-dominated sdB stars
have been discussed already (Sect. 14.4).
With higher quality data, greater discrimination was achieved [95] by
introducing classes sdB-O (some He I 4471) as a subclass of sdB, and dividing
sdO (with relative numbers) into sdOA (49: dominated by H absorption with
some He II), sdOB (87: dominated by He I and He II, generally some H),
sdOC (50: primarily He II, maybe some H, a trace of He I 4471) and sdOD
(30: “pure” He I, weak or absent H and He II 4686). The sdOA stars are not
important here. The sdOB and sdOC stars have been referred to elsewhere
as helium-rich sdO or He-sdO stars, while sdOD stars are also known as
helium-rich sdB or He-sdB stars. Note that two definitions for sdOB are not
equivalent, and the definition for sdOD is identical to that for the extreme
helium B stars [151].
He-sdB stars.
Little attention was paid to the stars classified sdOD in the PG survey for
several years. The prototype PG 1544+488 was discussed in a poster paper
for which only an abstract was published [96]. JL 87 was the first to be re-
ported in detail [152]. Subsequent observations demonstrate that the class is
92 C. S. Jeffery

quite heterogenous, with a wide range of hydrogen/helium ratios. The biggest

surprise was that PG 1544+488 turns out to be a double-lined spectroscopic
binary comprising two helium-rich subdwarfs [153]. A decade ago, some 50
He-sdB stars were known from the FB, PG, BPS, and HS surveys [115]. The
HE and Sloan surveys have added a further 5 definite and 11 possible He-sdB
stars, while better spectroscopy has shown several of the original 50 would be
better described as He-sdO.

Compact He-sdO stars.

The sdO classification causes some confusion because class members span a
range of 2 dex in surface gravity (and hence in luminosity-to-mass ratio). PG
classifications do not demonstrate this, but a finer scheme is more successful
[154]. Higher-gravity stars lie close to the extreme horizontal-branch / helium
main-sequence. A link with the He-sdB stars seems likely (Fig. 33).
An early description referred to these stars as “narrow-line O-type sub-
dwarfs”, including BD+75◦ 325, HD 127493, HZ 44 and BD+25◦ 4655 [97].
Like the He-sdB stars, there is considerable diversity as regards binarity and
abundance. The stars can be divided (from classification spectra) into those
which are carbon-rich, and those which are not. For the most part, all are
nitrogen rich. Several occur in composite spectrum binaries, and one [155,
HE 0301-3039] is a double He-sdO system similar to the double He-sdB; oth-
ers are apparently single. With substantial numbers of both helium-enriched
and helium-deficient compact sdO stars now identified and analysed spectro-
scopically [156], an explanation of their origin is tantalizingly close – but not
fully formed.

Low-gravity He-sdO stars.

The first indication that some He-sdO stars might have a much lower sur-
face gravity was obtained from spectroscopy of BD+37◦ 1977 [98], which was
both the hottest star in the original list observed by the TD1A ultraviolet
satellite and clearly more luminous than BD+75◦ 325 (see above). The group
remains sparse, with only five established members, including BD+37◦ 1977,
BD+37◦ 442, LSE 159, LSE 259 and LSE 263, though the boundaries with
other groups (e.g. EHe stars, compact He-sdO) are ill-defined. These stars are
important; if R CrB and Extreme Helium stars contract to the white dwarf
sequence, at some point they should look like these low-gravity He-sdO stars
(Fig. 33). Although we know five excellent examples brighter than V = 12,
fainter counterparts do not appear in surveys more recent than [157].

15.4 Central Stars of Planetary Nebulae

Planetary nebulae attract attention partly because they are photogenic, partly
because of their extreme physics and partly because they harbour a very hot
compact star at their center. Their spectra are diverse and can be clearly
 Stellar Structure and Evolution 93

divided into hydrogen-rich and hydrogen-deficient sequences, roughly in the

ratio 70:40 [158]. Both sequences can be subdivided, the latter as follows.

[WC] stars.

As in massive, Population I, WC stars, the spectra of [WC] stars are dom-

inated by broad emission from He I, II, C II - IV, N II, III and other light
ions. The group extends from early to very late types. Effective temperatures
are difficult to determine for stars with optically thick winds, but equivalent
measurements indicate surface temperatures ranging from 22 000 K (V348 Sgr:
[WC12]) to 140 000 K (Sanduleak 3: [WC3]).

Of-WR(C) stars.

The spectrum is dominated by strong C emission lines. He II 4686 is a strong

emission line, narrow in Of(C) and broad in the Of-WR(C). He II 4200, 4541
are in absorption. The two established Of-WR(C) stars are Abell 30 and Abell
78, both of which also show H-deficient material in the inner parts of their
nebulae.

O(He) stars.

Intermediate between the [WC] stars and the PG1159/O(C) stars, the O(He)
stars show He II, absorption and C IV, N V and O VI emission. Four are
known [159]. Two have planetary nebulae (K 1-27 and LoTr 4) and two, so far,
do not. The prototype K 1-27 was recognised from a survey of planetary neb-
ula central stars because of the strong ratio of the He II 4686 to [O III 4959],
which marked it out as “the most extreme example of the class of old nebu-
lae with very high excitations” [103]. Pulsations have not been detected [160].

O(C) / PG1159 stars.

According to [158], the spectra of O(C) planetary nebula central stars are
dominated by an absorption line spectrum of C. During the course of the PG
survey, a small group of eight stars showed spectra dominated by helium, but
distinguished by a broad absorption from He II 4686, C IV 4658 and possible
other C and N contributors [95]. The O(C) and PG1159 spectral classes are
essentially equivalent, and it is evident that stars of similar spectral type are
to be found both with and without planetary nebulae. Stars of both O(He)
and O(C) classes appear with and without nebulae in roughly the ratio 1:1.
The PG1159 classification has grown to include some stars showing sharp
emission lines, some having less broad absorption and one showing no he-
lium [161, H 1504+65]. The group had attracted immediate attention because
94 C. S. Jeffery

a fraction, including the prototype PG 1159-035, showed complex pulsations

[102]. It subsequently became clear that these stars were amongst the hottest
known, effective temperatures in excess of 100 000 K were reliably established.
H 1504+065 is not only very hot (≈ 200 000 K), its surface is deficient in hy-
drogen and helium, and dominated by a carbon/oxygen mixture with excep-
tionally high neon and magnesium [162, 163].
Measurements of effective temperature and surface gravity place these
stars around the knee where a contracting post-asymptotic giant-branch star
becomes fully degenerate and begins to cool down along the white dwarf cool-
ing sequence. Several hydrogen-deficient spectral classes correspond to this
evolutionary domain, including the hottest [WC] and Of-WR(C)stars, and
the DO white dwarfs (Fig. 33), demanding precision in measurement and rais-
ing important questions about stellar evolution [164].

15.5 White Dwarfs

H-deficient white dwarfs.

It is no surprise that Fleming was involved in the discovery of white dwarfs –

she had described the spectrum of 40 Eri B as being of type A, in apparent
contradiction to its faint absolute magnitude [165]. A difficulty in identifying
the first hydrogen-deficient white dwarf is linked to the development of a robust
classification scheme. The “D” designation was introduced [166], with the
suggestion to use DC for continuous spectra, followed by DA, DF, DG, . . . DM.
To this would be added a two-digit number indicating the estimated surface
temperature in thousands of degrees. Five stars were identified as “DBn”
[106], but the “Bn” appears to derive from analogy with MK types rather
than adherence to Luyten’s suggestion, since the DB0 spectrum of HZ 43
appears “nearly continuous”. The first DB white dwarfs in the modern sense
appear to be L 1573-31 and L 930-80 [105], with no hydrogen lines and He
I “outstanding in intensity, . . . probably stronger than in any known stellar
absorption spectrum”. By this time, Luyten had adopted yet another notation
to subdivide the DAs and DBs.
As the first example of a DC white dwarf, HZ 43 is remarkable in other
ways, and especially since space observatories were established to explore the
ultraviolet spectrum and beyond. Early UV observations pointed to a temper-
ature > 100 000 K [167]. Within a decade, X-ray observations had discovered
the DO/PG1159 star H 1504+65 [161] – where we have already been.
The class of very hot DO white dwarfs with He II stronger than He I was
established [104] with the “interesting star” HZ 21 [104]. In the PG survey
[95], the classification [95] is very broad and a composite of four subgroups
defined by [168], including DAO (which show hydrogen), a “cool” DOc group,
which shows He I and He II, a “hot” DOh group, which shows only He II, and
the PG1159 group described above. The overlap between DO and PG1159
can be confusing if the definitions are not used with care.
 Stellar Structure and Evolution 95

If the continuous spectra of DC white dwarfs lack both hydrogen and

helium, then it is fair to suspect that they are composed of elements with
low opacity at high temperatures. Cooler white dwarfs lacking both H and
He include the DQ white dwarfs (C I or C2 absorption) and DZ white dwarfs
(Ca II, Mg II, . . . ).
Far more complete descriptions of the classification and statistics of white
dwarfs will be found in the McCook et al. catalogue [169] and its online sup-
plements. These introduce a system of half-integer subclasses linked to the
inverse effective temperature (10 × 5040/Teff ). Whether a spectral classifi-
cation should depend on an analysis remains contentious – but it is clearly
necessary to distinguish the huge spectral range exhibited within each white
dwarf class (DQ, for example). As sky surveys go deeper, there is no shortage
of new white dwarfs to discover, with 9316 confirmed in SDSS Data Release
4 [170, 171].

AM CVn binaries.

AM CVn and its ilk stand amongst the most exotic of binary stars. The white
dwarf HZ 29 (=AM CVn) differed from normal helium-rich white (DB) dwarfs
since the helium lines appear shallow and almost double [107]. These stars re-
mained enigmatic through the 1980’s, until 4m telescopes and high-efficiency
spectrographs showed them to be interacting white dwarfs with periods in
the 5 – 65 min range, i.e. double stars separated by tens of earth radii. A
wide variety of phenomena are seen; photometric variations with periods dif-
ferent to the orbital period, spectra ranging from broad absorption line sys-
tems (AM CVn) to those dominated by He I and He II emission (ES Cet).
They include the shortest-period binary known (RX J0806.3+1527=HM Cnc:
Porb = 321 s) [172], which would fit inside Saturn. Systems appear to show
either a high or low state, as in conventional cataclysmic variables, except that
in a double-degenerate binary, the transferred material impacts directly onto
the stellar surface, with the “disk” more likely being debris filling the space
between the accretor’s surface and the Roche lobe. Over 20 systems are now
known, thanks most recently to the SDSS, with a space density upwards of
10−6 pc−3 [173]. The question arises whether these very compact binary stars
will eventually merge, and what the products might be: a gamma-ray burst
or an R CrB star perhaps? Meanwhile, it seems likely they will dominate the
gravitational wave background to be measured by LISA in the 0.3 – 1 mHz
range [174].

15.6 Post-AGB evolution

Such a diverse collection of exotic stars places severe demands on the theory
of stellar evolution. In the chapter by Karakas in this volumn, the evolution
of low-mass stars up to the asymptotic-giant branch has been discussed more
96 C. S. Jeffery

or less in detail. In the next few paragraphs, we briefly outline some of the
possibilities for “post-AGB” evolution.

“Normal” post-AGB evolution

Conventional theory posits that once the hydrogen-envelope mass has dropped
below some critical limit, the thermally-pulsing AGB star will develop a su-
perwind that expels the remaining outer layers. These layers initially form an
expanding dusty shell or envelope around the star, obscuring it from view at
visible wavelengths. The shell might be called a proto-planetary nebula and
may be visible at infrared wavelengths. Since the hydrogen-burning shell in
the AGB remnant has been extinguished, the star contracts and heats. As the
dust-shell expands, the central star will become visible, initially as an F- or
A- star. The contraction proceeds on a thermal timescale (Eq. (9)).
The more massive post-AGB stars (M ≥ 0.6M ) with higher luminos-
ity contract more quickly (L ≥ 104 L , tK ≈ 10 000 y). When their surface
temperature exceeds 30 000 K, surface hydrogen is fully ionized, the Lyman
continuum opacity drops and ultraviolet photons are able to ionize the much-
expanded shell, which now becomes visible as a planetary nebula. Schemati-
cally we might represent the evolution as follows:

AGB → protoPN → PN → WD.

Less massive post-AGB stars contract more slowly (L ≤ 104 L , tK ≈

100 000 y) so that the shell will have dispersed completely before the stel-
lar remnant is hot enough to ionize it. These central stars may appear first as
low-gravity F-stars, and later as sdO stars. Their evolution from spectral-type
F- to sdO passes through spectral-types A and B; in the absence of an infrared
dust shell, such stars are difficult to distinguish from main-sequence stars of
similar spectral-type, except that their rotation velocities will be small. In
this case:
AGB → pAGB → sdO → WD.

The late thermal pulse

In some cases the last helium shell flash will not occur while the star is on the
asymptotic-giant branch [175, 176]. If the mass of the intershell is sufficiently
high after the superwind phase and the onset of envelope contraction, residual
heating can re-ignite the helium shell while the star is contracting. Since the
hydrogen-envelope has very little mass and is relatively dense, the expansion
of the intershell will not be absorbed by the hydrogen envelope and the result
will be an immediate expansion of the star back towards the giant branch.
This will occur on the mean thermal timescale of the envelope ≈ 2 × 102
and 103 y, depending on the phase of the thermal pulse cycle at which the
star originally left the AGB and hence on precisely when shell ignition occurs.
Following ignition, the shell will drive a convection zone through the intershell
 Stellar Structure and Evolution 97

region, but it is not currently thought that there will be contact between H-
and He-rich material.
Once the star has fully expanded to become a yellow or red supergiant,
a deep surface convection zone will develop which will dredge helium- and
carbon-rich material to the surface, and will completely dilute what is left of
the hydrogen envelope. As the star contracts for a second time, its surface will
be helium- and carbon-rich. Any planetary nebula ejected at the tip of the
AGB phase may still be ionized, since the nebula recombination time is long
compared with the thermal timescale of the envelope. Additional helium-rich
material may also have been ejected, or be visible as a stellar wind.
As the energy dissipated in such a late thermal pulse is radiated away from
the star, the envelope contracts and the surface heats. The star will be very
luminous, possibly super-Eddington, driving a helium- and carbon-rich stellar
wind visible as optical emission lines, giving rise to [WC-L] and [WC-E] type
spectra. The surface heating will eventually ionize the inner portions of the
nebula expelled when the star returned to the giant phase. As the star further
contracts its spectrum will start to resemble that of the PG1159 stars before
cooling down the white dwarf sequence for a second time, possibly as a DB
white dwarf [177, 178, 164]:

AGB → PN → LTP → YSG → [WC−L] → [WC−E]

→ O(C) → PG1159 → DB/DQ.

The very late thermal pulse

The possibility that the last thermal pulse could occur even later, i.e. when
the star has reached the white dwarf cooling sequence and when the envelope
is even more compressed [176], leads to the possibility of an even more violent
helium-shell flash and expansion on a timescale of ≤ 10 y. In this case, theory
suggests that flash-driven convection can mix material all the way through the
He-rich intershell and into the H-rich surface. The surface should appear to
become hydrogen-poor and carbon-rich either immediately following the shell
flash, or very soon after the surface convection starts to develop [179, 180].
The rapid expansion and prompt change in surface composition observed in
Sakurai’s object are thought to be due to such a very late thermal pulse. The
near ballistic expansion expels a dusty envelope, so that the star itself remains
invisible for much of the time it spends as a yellow supergiant. Due to the
low envelope mass, subsequent contraction would also be more rapid, taking
≤ 100 y to return to the white dwarf sequence. The combination of extreme
flash mixing and subsequent dredge-up could result in a surface dominated by
carbon and oxygen, such as that seen in the hot DO white dwarf H1504+65
and the hot DQ white dwarfs [181]:

AGB → PN → WD → VLTP → YSG → [WC−E] → H1504 + 65 → DQ.

98 C. S. Jeffery

15.7 Double-Degenerate Mergers

It has become clear that the R CrB and EHe stars probably cannot be formed
as a result of a late thermal pulse; in particular their surface carbon abun-
dances are too low [164,60]. It has already been seen (Sect. 14.5) how a binary
system comprising two white dwarfs could decay and merge within a Hubble
time. If the resulting star re-ignites helium to produce a yellow supergiant
[182], the immediate result would be as follows.

He+He white dwarf mergers

The evolution of a star formed from the merger of two helium white dwarfs
considered a 0.4 M helium white dwarf accreting helium at approximately
half the Eddington accretion rate (10−5 M yr−1 [62]). Helium ignites at the
core-envelope boundary when 0.067 M has been accreted and stars expands
to become a yellow supergiant in ≈ 103 yr. The accretion is switched off ar-
tificially once a pre-selected final mass has been reached, whereupon the star
evolves towards the helium main sequence as previously described (Sect. 14.5).
The significance for EHes is that the surface layers of the product should
consist primarily of CNO-processed helium, with some contamination by resid-
ual hydrogen from the surface layers of the original white dwarfs. The mod-
els can reproduce the observed, mass, radius, luminosity, pulsation period
(and Ṗ ), and surface composition for the EHe star V652 Her with remarkable
success [183, 184]. The models also anticipate a population of helium- and
nitrogen-rich subdwarf O and B stars:

He+He WD → DD → EHe → He−sdB → He−sdO → DB.

CO+He white dwarf mergers

In considering the merger of a helium white dwarf with a 0.5 or 0.6 M CO

white dwarf [185], it was assumed (as for the He+He merger) that the CO
white dwarf accretes matter from the debris disk at 10−5 M yr−1 . Helium
ignition occurs once sufficient material has been accreted, forcing the star to
expand. Helium accretion is switched off at a pre-assigned final mass. In this
case, the helium-burning shell is stable, and maintains the star as a supergiant
until the helium envelope has been burnt through, leaving a CO core of 0.7 −
0.9 M as chosen. The core being more massive, the luminosity of the helium-
burning shell will be substantially greater than in the case of the He+He
merger. At this point, the remaining envelope cannot support itself and the
star contracts on a thermal timescale at constant luminosity until it becomes
a white dwarf.
This model is successful in accounting for many of the observed properties
of more luminous EHes, and also RCrB stars [186]. The absence of any EHe
stars in a binary, the observed radii, masses, and surface gravities, observed
contraction rates, surface abundances and number densities are all in satisfac-
tory agreement with evolution models for CO+He WD mergers which result
 Stellar Structure and Evolution 99

in star with a final mass of 0.7 − 0.9 M . Presumably such an EHe continues
to contract to become a white dwarf. There is no indication that a nebula ex-
ists around any EHes, except possibly MV Sgr. Stars that may represent the
descendants of EHes include the low-gravity helium-rich sdO stars and the
O(He) stars [187]. After that, these stars should become relatively high-mass
CO white dwarfs, i.e.
CO + HeWD → DD → RCB → EHe → HesdO+ → O(He) → DB.

16 Conclusion
These lecture notes have attempted to outline the fundamental theory of
stellar structure and evolution from first principles. They have tried to demon-
strate the principles by which mainstream stars work. They have also intro-
duced some of modern theory which attempts to explain how various exotic
stars were formed. To do so within six lectures and within the constraints of a
volume such as this, much useful material has had to be omitted. Many inter-
esting analytic approximations have had to be overlooked, including homology
relations and specific solutions of the Lane-Emden equation. With hindsight,
discussion of the microphysics has been less than rigorous. The omission of
stellar winds, rotation, non-standard mixing and magnetic fields is no longer
satisfactory. We have not addressed the physics of very massive or of low-mass
stars. The former have a direct impact on nucleosynthesis in the galaxy. The
latter are statistically very significant. Neither have we discussed the general
consequences of binary star evolution, supernovae explosions, or the evolu-
tion of entire stellar populations on nucleosynthesis yields. Fortunately, some
of these topics are discussed by other contributors to these proceedings. For
others, we are obliged to leave the reader to make their own exploration in
the full knowledge that this is a rewarding and fulfilling endeavour.

Acknowledgments
The author is indebted to the organisers of the Kodai School for the invita-
tion to participate, financial support and indefatigable hospitality. He is also
grateful for their patience while these extended notes were being prepared for
publication. Drs Chris Tout and Amanda Karakas made invaluable contribu-
tions to the preparation of the text. In addition to the texts cited, extensive
use was made of the SIMBAD database, operated at CDS, Strasbourg, France,
and of NASA’s Astrophysics Data System.

References
1. M. Schwarzschild, Structure and evolution of the stars, Princeton Univer-
sity Press (1958)
100 C. S. Jeffery

2. R. J. Tayler, The stars: Their structure and evolution, London: Wykeham

(1970)
3. I. S. Shklovskii, Stars, their birth, life and death, San Francisco: W. H.
Freeman, p13 (1978)
4. R. kippenhahn, A. Weigert, Stellar Structure and Evolution, Springer-
Verlag: Berlin Heidelberg New York (1990)
5. E. Böhm-Vitense, Introduction to stellar astrophysics, Vol. 3 - Stellar
structure and evolution, Cambridge University Press (1992)
6. A. C. Phillips, The physics of stars, Chichester, UK: Wiley (1994)
7. C. J. Hansen, S. D. Kawaler, Stellar interiors. Physical Principles, Struc-
ture, and Evolution, Springer-Verlag: Berlin Heidelberg New York (1994)
8. D. Prialnik, An introduction to the Theory of Stellar Structure and Evo-
lution, Cambridge University Press (2000)
9. P. Eggleton, Evolutionary Processes in Binary and Multiple Stars,
Cambridge University Press (2006)
10. C. S. Jeffery, Studies in Stellar Structure and Evolution, PhD Thesis,
University of St. Andrews (1982)
11. C. S. Jeffery, in Hydrogen-Deficient Stars, Astronomical Society of the
Pacific Conference Series Vol. 391, p3 (2008)
12. E. Hertzsprung, Publikationen des Astrophysikalischen Observatoriums
zu Potsdam, 63 (1911)
13. H. N. Russell, Popular Astronomy 22, 275 (1914)
14. H. N. Russell, Popular Astronomy 22, 331 (1914)
15. S. C. Keller, E. K. Grebel, G. J. Miller, K. M. Yoss, AJ 122, 248 (2001)
16. K. G. Stassun, M. Van den Berg, R. D. Mathieu, F. Verbunt, A&A 382,
899 (2002)
17. J. E. Hesser, W. E. Harris, D. A. Vandenberg, J. W. B. Allwrite, P. Shott,
P.B. Stetson, PASP 99, 739 (1987)
18. S. C. Rey, Y. W. Lee, C. H. Ree, J. M. Joo, Y. J. Sohn, A. R. Walker,
AJ 127, 958 (2004)
19. K. F. Khalliulin, Master’s Thesis, Sternberg Astronomical Institute
(1996)
20. P. P. Eggleton, D. S. P. Dearborn, J. C. Lattanzio, ApJ 677, 581 (2008)
21. W. A. Fowler, G. R. Caughlan, B. A. Zimmerman, ARA&A 13, 69 (1975)
22. W. A. Fowler, G. R. Caughlan, B. A. Zimmerman, ARA&A 5, 525 (1967)
23. M. J. Harris, W. A. Fowler, G. R. Caughlan, B. A. Zimmerman, ARA&A
21, 165 (1983)
24. M. Aikawa, M. Arnould, S. Goriely, A. Jorisson, K. Takahashi, A&A 441,
1195 (2005)
25. E. M. Burbidge, G. R. Burbidge, W. A. Fowler, F. Hoyle, Reviews of
Modern Physics 29, 547, (1957)
26. L. G. Henyey, L. Wilets, K.H. Böhm, R Lelevior, R. D. Levee, ApJ 129,
628 (1959)
27. R. Kippenhahn, A. Weigert, E. Hofmeister, Meth. Comp. Phys. 7, 53
(1967)
28. I. Iben, Jr., Q. J. R. Astron. Soc. 26, 1 (1985)
29. I. J. Iben, ApJ 143, 483 (1966)
30. M. Schönberg and S. Chandrasekhar, ApJ 96, 161, (1942)
31. U.Uus, Nauk. Inf. 17, 32 (1970)
 Stellar Structure and Evolution 101

32. C. S. Jeffery, MNRAS 235, 1287 (1988)

33. F. Hoyle, W. A. Fowler, ApJ 132, 565 (1960)
34. D. S. P. Dearborn, J. C. Lattanzio, P. P. Eggleton, ApJ 639, 405, (2006)
35. S. C. Rey, S. J. Yoon, Y. W. Lee, B. Chaboyer, A. Sarajedini, AJ 122,
3219 (2001)
36. B. Dorman, R. T. Rood, R. W. O’Connell, ApJ 419, 596, (1993)
37. P. Demarque, J. G. Mengel, ApJ 164, 317 (1971)
38. A. V. Sweigart, J. G. Mengel, P. Demarque, A&A 30, 13 (1974)
39. M. Castellani, V. Castellani, ApJ 407, 649, (1993)
40. J. L. Greenstein, A. I. Sargent, ApJS 28, 157 (1974)
41. V. Calo, V. Castellani, J. Danziger, R. Gilmozzi, R. D. Cannon, P. W.
Hill, A. Boksenberg, MNRAS 222, 55 (1986)
42. U Heber, A&A 155, 33 (1986)
43. C. Sosin, B. Dorman, S. G. Djorgovski, G. Piotto, R. M. Rich, I. R. King
et al., ApJ 480, L35 (1997)
44. J. H. Whitney, R. W. O’Connell, R. T. Rood, B. Dorman, W. B.
Landsman, K. P. Cheng et al., AJ 108, 1350 (1994)
45. J. H. Whitney, R. T. Rood, R. W. O’Connell, N. L. D’Cruz, B. Dorman,
W. B. Landsman et al., ApJ 495, 281 (1998)
46. S. Moehler, S. Dreizler, T. Lanz, G. Bono, A. V. Sweigart, A. Calamida
et al., in Hot Subdwarf Stars and Related Objects, Astronomical Society
of the Pacific Conference Series, vol. 392, ed. by U. Heber, C. S. Jeffery,
R. Napiwotzki, p47 (2008)
47. R. A. Saffer, E. M. Green, T. Bowers, in 12th European Workshop on
White Dwarfs, Astronomical Society of the Pacific Conference Series, vol.
226, ed. by J. L. Provencal, H. L. Shipman, J. MacDonald, S. Goodchild,
p408 (2001)
48. N. L. D’Cruz, B. Dorman, R. T. Rood, R. W. O’Connell, ApJ 466, 359
(1996)
49. T. M. Brown, A. V. Sweigart, T. Lanz, W. B. Landsman, I. Hubeny, ApJ
562, 368 (2001)
50. Z. Han, P. Podsiadlowski, P. F. L. Maxted, T. R. Marsh, N. Ivanova,
MNRAS 336, 449 (2002)
51. Z. Han, P. Podsiadlowski, P. F. L. Maxted, T. R. Marsh, MNRAS 341,
669 (2003)
52. B. Paczynski, Comments on Astrophysics 6, 95 (1976)
53. S. Geier, C. Karl, H. Edelmann, U. Heber, R. Napiwotzki, Astronomische
Nachrichten 328, 708 (2007)
54. J. G. Mengel, J. Norris, P. G. Gross, ApJ 204, 488 (1976)
55. L. Landau, E. M. Lifshitz, The classical theory of fields , Pergamon Press,
Oxford (1958)
56. T. R. Marsh, V. S. Dhillon, S. R. Duck, MNRAS 275, 828 (1995)
57. W. Benz, A. G. W. Cameron, W. H. Press, R. L. Bowers, ApJ 348, 647
(1990)
58. L. Segretain, G. Chabrier, R. Mochkovitch, ApJ 481, 355, (1997)
59. J. Guerrero, E. Garcı́a-Berro, J. Isern, A&A 413, 257 (2004)
60. G.C. Claton, T.R. Geballe, F. Herwig, C. Fryer, M. Asplund, ApJ 662,
1220 (2007)
61. D. Lynden-Bell, J. E. Pringle, MNRAS 168, 603 (1974)
102 C. S. Jeffery

62. H. Saio, C. S. Jeffery, MNRAS 313, 671 (2000)

63. M. Politano, R. E. Taam, M. van der Sluys, B. Willems, ApJ 687, L99
(2008)
64. S. Justham, P. Podsiadlowski, Z. Han, On the origin of the single and
binary helium-rich sdO stars
65. A. Ahmad, C. S. Jeffery in 14th European Workshop on White Dwarfs,
Astronomical Society of the Pacific Conference Series, vol. 334, ed. by
D. Koester, S. Mochler, p291 (2005)
66. R. Napiwotzki in Hot subdwarf star and related objects, Astronomical
Society of the Pacific Conference Series, vol 392, ed. by U. Heber, C. S.
Jeffery, R. Napiwotzki, p139 (2008)
67. A. V. Sweigart, in The Third Conference on Faint Blue Stars ed. by
A. G. D. Philip, J. Liebert, R. Saffer, D. S. Hayes, p3 (1997)
68. M. M. Miller Bertolami, L. G. Althaus, K. Unglaub, A. Weiss, A&A 491,
253 (2008)
69. T. M. Brown, New Astronomy Review 49, 474 (2005)
70. A. D. Code, G. A. Welch, ApJ 228, 95 (1979)
71. F. Bertola, M. Capaccioli, A. V. Holm, J. B. Oke, ApJ 237, L65 (1980)
72. T. M. Brown, C. W. Bowers, R. A. Kimble, A. V. Sweigart, H. C.
Fergyson, ApJ 532, 308 (2000)
73. S. Yi, P. Demarque, A. J. Oemler, ApJ 492, 480 (1998)
74. P. Podsiadlowski, Z. Han, A. E. Lynas-Gray, D. Brown, in Hot Subdwarf
Stars and Related Objects, Astronomical Society of the Pacific Conference
Series, Vol. 392, ed. by U. Heber, C. S. Jeffery, R. Napiwotzki, p15 (2008)
75. W. P. Fleming, Astron. Nachr. 126, 165 (1891)
76. H. Ludendorff, Astron. Nachr. 173, 1 (1907)
77. A. H. Joy, M. H. Humason, PASP 35, 325 (1923)
78. J. S. Plaskett, Publications of the Dominion Astrophysical Observatory
Victoria 4, 1 (1926)
79. C. H. Payne, Stellar Atmospheres; a Contribution to the Observational
Study of High Temperature in the Reversing Layers of Stars. PhD thesis,
Radcliffe College (1925)
80. L. Berman, ApJ 81, 369 (1935)
81. O. Struve, F. Sherman, ApJ 91, 428 (1940)
82. J. L. Greenstein, ApJ 91, 438 (1940)
83. C. Wolf, G. Rayet, Comptes Rendus 65, 292 (1867)
84. M. Fleming, Astron. Nachr. 128, 403 (1891)
85. J. Berger, Compt. rend. séances Acad. Sci. 242, 2300 (1956)
86. J. H. Elias, K. Matthews, G. Neugebauer, S. E. Persson, ApJ 296, 379
(1985)
87. A. V. Filippenko, A. C. Porter, W. L. W. Sargent, AJ 100, 1575 (1990)
88. E. Pigott, H. C. Englefield, Philosophical Transactions Series I 87, 133
(1797)
89. R. H. Curtiss, Publications of Michigan Observatory 2, 182 (1916)
90. W. C. Rufus, Publications of Michigan Observatory 3, 257 (1923)
91. D. M. Popper, PASP 54, 160 (1942)
92. I. E. Woods, Harvard College Observatory Bulletin 855, 22 (1928)
93. G. H. Herbig, ApJ 140, 1317 (1964)
94. C. Hoffmeister, Astron. Nachr. 274, 176 (1944)
 Stellar Structure and Evolution 103

95. R. F. Green, M. Schmidt, J. Liebert, ApJS 61, 305 (1986)

96. U. Heber, S. Dreizler, K. de Boer, S. Moehler, T. Richtler, Astron Ges.,
Abstr. Ser. 1, 16 (1988)
97. J. Greenstein, G. Münch, Carnegie Inst. Washington Year Book 53, 11
(1953)
98. S. C Wolff, C. A. Pilachowski, R. D. Wolstencroft, ApJ 194, L83 (1974)
99. C. S. Beals, Trans. IAU 6, 248 (1938)
100. L. F. Smith, L. H. Aller, ApJ 157, 1245 (1969)
101. M. Cohen, H. S. Hudson, S. L. Odell, W. A. Stein, MNRAS 181, 233
(1977)
102. J. T. McGraw, J. Liebert, S. G. Starrfield, R. Green, in IAU Colloq. 53:
White Dwarfs and variable Degenerate Stars, ed. by H.M. van Horn, V.
weidemann, p377 (1979)
103. K. G. Henize, A. P. Fairall, PASP 93, 435 (1981)
104. J. L. Greenstein, ApJ 144, 496, (1966)
105. W. J. Luyten, ApJ 116, 283, (1952)
106. M. L. Humason, F. Zwicky, ApJ 105, 85 (1947)
107. J. L. Greenstein, M. S. Matthews, ApJ 126, 14 (1957)
108. J. Feige, ApJ 128, 267, (1958)
109. T. C. Beers, S. P. Doinidis, K. E. Griffin, G. W. Preston, S. A. Shectman,
AJ 103, 267 (1992)
110. H. J. Hagen, D. Groote, D. Engels, D. Reimers, A&AS 111, 195 (1995)
111. L. Wisotzki, T. Koehler, D. Groote, D. Reimers, A&AS 115, 227 (1996)
112. R. S. Stobie, D. Kilkenny, D. O’Donoghue, A Chen, C. Koen, D. H.
Morgan et al., MNRAS 287, 848 (1997)
113. C. Stoughton, R. H. Lupton, M. Bernardi, M. R. Blanton, S. Burles, F.
J. Castander et al., AJ 123, 485 (2002)
114. J. S. Drilling, P. W. Hill in IAU Colloq. 87: Hydrogen Deficient Stars and
Related Objects, Astrophysics and Space Science Library, vol. 128, ed. by
K. Hunger, D. Schönberner, N. Kameswara Rao, p 499 (1986)
115. C. S. Jeffery, U. Heber, P. W. Hill, S. Dreizler, J. S. Drilling, W. A.
Lawson et al. in Hydrogen Deficient Stars, Astronomical Society of the
Pacific Conference Series, vol 96, ed. by C. S. Jeffery, U. Heber p471
(1996)
116. A. S. Dinger, Ap&SS 6, 118 (1970)
117. W. W. Campbell, A&A 13, 448 (1894)
118. K. A. van der Hucht, New Astronomy Review 45, 135 (2001)
119. K. Hunger, D. Groote, A&A 351, 554 (1999)
120. W. W. Campbell, ApJ 10, 241 (1899)
121. W. P. Bidelman, ApJ 111, 333 (1950)
122. J. S. Drilling, ApJ 242, L43 (1980)
123. C. S. Jeffery, R. Aznar Cuadrado, A&A 378, 936 (2001)
124. W. P. Fleming, A&A 11, 765 (1892)
125. R. E. Dudley, C. S. Jeffery, MNRAS 247, 400 (1990)
126. A. Uomoto, ApJ 310, L35 (1986)
127. D. Schönberner, J.S. Drilling, ApJ 268, 225 (1983)
128. A. V. Filippenko, ARA&A 35, 309, (1997)
129. J. C Wheeler, R. P. Harkness, Reports on Progress in Physics 53, 1467
(1990)
104 C. S. Jeffery

130. D. Hoffleit, AJ 64, 241 (1959)

131. C. Alcock, R. A. Allsman, D. R. Alves, T. S. Axelrod, A. Becker, D.P.
Bennett et al., ApJ 470, 583 (1996)
132. G. C. Clayton, PASP 108, 225 (1996)
133. W. P. Bidelman, ApJ 117, 25 (1953)
134. B. Warner, MNRAS 137, 119 (1967)
135. W. A. Lawson, P. L. Cottrell, MNRAS 285, 266 (1997)
136. C. S. Jeffery, J. S. Drilling, U. Heber, MNRAS 226, 317 (1987)
137. P. L. Cottrell, W. A. Lawson, PASA 15, 179 (1998)
138. O. De Marco, G. C. Clayton, F. Herwig, D. L. Pollacco, J. S. Clark, D.
Kilkenny, AJ 123, 3387 (2002)
139. A. M. van Genderen, A. Gautschy, A&A 294, 453 (1995)
140. S. Nakano, Y. Sakurai, M. Hazen, R. H. McNaught, S. Benetti, H. W.
Duerbeck et al., IAU Circ. No. 6322, 1 (1996)
141. H. W. Duerbeck, S. Benetti, ApJ 468, L111 (1996)
142. H. W. Duerbeck, W. Liller, C. Sterken, S. Benetti, A. M. van Genderen,
J. Arts et al., AJ 119, 2360 (2000)
143. P. A. M. van Hoof, M. Hajduk, A. A. Zijlstra, F. Herwig, A. Evans, G.C.
van de Steene et al., A&A 471, L9 (2007)
144. K. Lundmark, PASP 33, 314 (1921)
145. G. C Clayton, O. De Marco, AJ 114, 2679 (1997)
146. M. Asplund, D. L. Lambert, T. Kipper, D. Pollacco, M. D. Shetrone,
A&A 343, 507 (1999)
147. C. S. Jeffery, D. Schönberner, A&A 459, 885 (2006)
148. G. C Clayton, F. Kerber, N. Pirzkal, O. De Marco, P. A. Crowther, J.
M. Fedrow, ApJ 646, L69 (2006)
149. A. Sandage, ApJ 141, 1560 (1965)
150. B. Baschek, J. Norris, ApJ 199, 694 (1975)
151. J. S. Drilling, in Hydrogen Deficient Stars, Astronomical Society of the
Pacific Conference Series, vol. 96 ed. by C. S. Jeffery, U. Heber, p461
(1996)
152. H. Schulz, G. Wegner, U. Heber, PASP 103, 435 (1991)
153. A. Ahmad, C. S. Jeffery, A&A 413, 323 (2004)
154. J. S. Drilling, S. Moehler, C. S. Jeffery, U. Heber, R. Napiwotzki, in The
Garrison Festschrift, ed. by R. O. Gray, C. J. Corbally, A. G. D. Philip,
p27 (2003)
155. T. Lisker, U. Heber, R. Napiwotzki, N. Christlieb, D. Reimers,
D. Homeier, Ap&SS 291, 351 (2004)
156. A. Stroeer, U. Heber, T. Lisker, R. Napiwotzki, S. Dreizler, N. Christlieb,
D. Reimers, A&A 462, 269 (2007)
157. J. S. Drilling, ApJ 270, L13 (1983)
158. R. H. Mendez, in Evolution of Stars: the Photospheric Abundances Con-
nection, IAU Symposium, vol. 145 ed. by G. Michaud, A. V. Tutukov,
p375 (1991)
159. T. Rauch, S. Dreizler, B. Wolff, A&A 338, 651 (1998)
160. R. Ciardullo, H. E. Bond, AJ 111, 2332 (1996)
161. J. A. Nousek, H. L. Shipman, J. B. Holberg, J. Liebert, S. H. Pravdo,
N.E. White, P. Giommi, ApJ 309, 230 (1986)
162. K. Werner, A&A 251, 147 (1991)
 Stellar Structure and Evolution 105

163. K. Werner, T. Rauch, M. A. Barstow, J. W. Kruk, A&A 421, 1169 (2004)

164. K. Werner, F. Herwig, PASP 118, 183 (2006)
165. E. L. Schatzman, White dwarfs, Amsterdam, North-Holland Pub. Co.;
New York, Interscience Publishers (1958)
166. W. J. Luyten, ApJ 101, 131 (1945)
167. B. Margon, J. Liebert, M. Lampton, H. Spinrad, S. Bowyer, G. Gatewood,
ApJ 209, 525 (1976)
168. F. Wesemael, R. F. Green, J. Liebert, ApJS 58, 379 (1985)
169. G. P. McCook, E. M. Sion, ApJS 121, 1 (1999)
170. D. J. Eisenstein, J. Liebert, D. Koester, S. J. Kleinmann, A. Nitta, P. S.
Smith et al., AJ 132, 676 (2006)
171. D. J. Eisenstein, J. Liebert, H. C. Harris, S. J. Kleinmann, A. Nitta, N.
Silvestri et al., ApJS 167, 40 (2006)
172. G. Ramsay, P. Hakala, M. Cropper, MNRAS 332, L2 (2002)
173. G. H. A. Roelofs, P. J. Groot, G. F. Benedict, B.E. McArthur, D. Steeghs,
L. Morales-Rueda et al., ApJ 666, 1174 (2007)
174. G. Nelemans, in The Astrophysics of Cataclysmic Variables and Related
Objects, Astronomical Society of the Pacific Conference Series, vol 330,
ed. by J. M. Hameury, J. P. Lasota, p27 (2005)
175. M. Y. Fujimoto, PASJ 29, 331 (1977)
176. D. Schönberner, A&A 79, 108 (1979)
177. K. Werner, U. Heber, A&A 247, 476 (1991)
178. C.S. Jeffery, CCP7 Newsletter 21, 5 (1994)
179. F. Herwig, T. Blöcker, N. Langer, T. Driebe, A&A 349, L5 (1999)
180. F. Herwig, ApJ 554, L71 (2001)
181. P. Dufour, G. Fontaine, J. Liebert, G. D. Schmidt, N. Behara, ApJ 683,
978 (1008)
182. R. F. Webbink, ApJ 277, 355 (1984)
183. C. S. Jeffery, P. W. Hill, U. Heber, A&A 346, 491 (1999)
184. C. S. Jeffery, V. M. Woolf, D. L. Pollacco, A&A 376, 497 (2001)
185. H. Saio, C. S. Jeffery, MNRAS 333, 121 (2002)
186. G. C. Clayton, in Hydrogen-deficient Stars, Astronomical Society of the
Pacific Conference Series, vol 391, ed. by A. Werner, T. Rauch, p17
(2008)
187. C. S. Jeffery, in Hydrogen-deficient Stars, Astronomical Society of the
Pacific Conference Series, vol 391, ed. by A. Werner, T. Rauch, p53
(2008)
Nucleosynthesis of Low and Intermediate-mass
Stars

Amanda I. Karakas

Research School of Astronomy & Astrophysics, Mount Stromlo Observatory,

Weston Creek ACT 2611, Australia [email protected]

Summary. The asymptotic giant branch (AGB) is the last nuclear-burning phase
for low and intermediate-mass stars with initial masses between about 0.8 to 8
solar masses. The AGB phase of evolution is very short, comprising less than 1
per cent of the main-sequence lifetime, nevertheless, it is on the AGB that the
richest nucleosynthesis occurs for this mass range. The nucleosynthesis is driven by
thermal instabilities of the helium-burning shell, the products of which are dredged
to the stellar surface by recurrent mixing episodes. Hot bottom burning occurs in the
most massive AGB stars, and this also alters the surface composition. I review the
evolution and nucleosynthesis from the main sequence through to the tip of the AGB.
The nucleosynthesis that occurs during the AGB is explored in detail, including a
discussion of the effects of hot bottom burning. I finish with a brief review of the
slow-neutron-capture process that produces elements heavier than iron.

Keywords: stars: AGB and post-AGB stars — planetary nebulae: general —

nuclear reactions, nucleosynthesis, abundances

1 Introduction
All stars in the mass range 0.8M to ∼ 8M , including our own Sun, will
become asymptotic giant branch (AGB) stars (see Herwig [1] for a recent
review). The AGB is the last nuclear burning phase of stellar evolution for this
mass range, and is brief, lasting much less than 1% of the time spent during
core hydrogen (H) burning on the main sequence. AGB stars are evolved
objects and are found in the high-luminosity, low temperature region of the
Hertzsprung-Russell (HR) diagram. AGB stars have evolved through core H
and helium (He) burning and are now sustained against gravitational collapse
by alternate H-shell and He-shell burning. The He-burning shell is thermally
unstable and flashes or pulses every 104 years, depending on the stellar mass.
AGB stars are often observed to be chemically different from their less evolved
counterparts, and show enrichments in carbon and heavy elements synthesized

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 2, c Springer-Verlag Berlin Heidelberg 2010
108 Amanda I. Karakas

by the slow neutron-capture process (the s-process). AGB stars are observed
with spectral classes M, MS, S, SC and C, where C-type stars are carbon-
rich and have C/O ratios exceeding unity. AGB stars are observed to be
long-period Mira, semi-regular or irregular variables with periods ≥ 100 days.
Many AGB stars are also observed to be losing mass rapidly (typical mass-
loss rates Ṁ ∼ 10−5 M year−1) through slow, dense outflows (velocities ∼ 10
km/second). After the ejection of the envelope, the AGB phase is terminated
and the star quickly evolves (in ≈ 105 years) to become a post-AGB star,
before finally ending its life as a dense white dwarf. For a review of the post-
AGB and planetary nebulae phase of evolution, we refer to [2] and [3].
The range of bolometric luminosities and masses of AGB stars can be
estimated from the large sample of long-period variables observed in the
Large Magellanic Cloud (LMC). The luminosities range from Mbol ≈ −3.8
to a maximum of ≈ −7.1, that corresponds to current masses between about
≈ 1M to 8M [4]. In Figure 1 we show the colour-magnitude diagram (CMD)
for the globular cluster M3. A CMD is just a type of HR diagram in which we
use colour index rather than spectral class on the horizontal axis; and use the
apparent visual magnitude, V , for the vertical axis. While most of the stars in
Figure 1 are located on the main sequence, there are distinct bands to the top
right of the diagram and an almost horizontal band above the main sequence.
Each of these bands, including the main sequence, represents a distinct phase
in the life of low to intermediate-mass stars. In the luminous red part of the
diagram (top-right corner) there are a few AGB stars. Very few of the AGB
stars in globular clusters are more luminous than the tip of the first giant
branch which occurs at a bolometric luminosity of Mbol ≈ −3.6 [5].
AGB stars are particularly important because they are a significant site of
nucleosynthesis. The nucleosynthesis during the AGB leads to the production
of carbon, nitrogen, fluorine and heavy elements such as barium and lead.
Recurrent mixing episodes bring the freshly synthesized material from the core
to the envelope, and strong stellar winds ensure that this material is expelled
into the interstellar medium (ISM). For these reasons, AGB stars are major
factories for the production of the elements in the Universe [7, 1]. Knowledge
of AGB evolution and nucleosynthesis is vital to obtain an estimate of the
contribution of low and intermediate-mass stars to the chemical evolution of
galaxies and stellar systems such as M3. For example, a third of the carbon
in our Galaxy today is estimated to have been produced in AGB stars [8].
In these lecture notes, we are primarily interested in the changes to the sur-
face composition of low and intermediate mass stars during the AGB. Before
we get to the AGB stage, it is necessary that we first examine the evolution
and nucleosynthesis that occurs prior to the AGB and we cover this material
in Sect. 3. AGB evolution is discussed next in Sect. 4, and the nucleosynthe-
sis in Sect. 5. We briefly review the s-process in Sect. 6. We begin with some
important preliminaries.
 Nucleosynthesis of Low and Intermediate-mass Stars 109

12
M3 AGB

14
HB FGB
16

V
18
20

main sequence
22

– 0.5 0 0.5 1 1.5

B–V
Fig. 1. Colour-magnitude diagram for M3, using data from [6]. The approximate
positions of the main sequence, first giant branch (FGB), horizontal branch (HB),
and asymptotic giant branch (AGB) are labelled

2 Some preliminaries

In the following lecture notes, it is assumed that the reader has an under-
standing of the basics of stellar evolution and nucleosynthesis. We refer to
other lectures given at the Kodai School, along with the following textbooks:
Clayton [9], Rolfs & Rodney [10], and Iliadis [11] cover all aspects of nuclear
astrophysics. Stellar interiors and evolution are covered by Hansen, Kawaler,
& Trimble [12]. The chemical evolution of galaxies and stellar systems are
discussed by Pagel [13], along with an overview of stellar evolution and nu-
clear astrophysics. The topic of AGB stars is covered in detail in [14]. Lugaro
[15] reviews stardust in meteorites, with an emphasis on pre-solar grains that
originated in AGB stars.
Before we begin, it is important to make a few definitions. In the following
we will refer to low-mass stars as stars with initial masses between 0.8 to
∼ 2.25M, and intermediate-mass stars as stars with an initial masses be-
tween 2.25 and ∼ 8M . Stars in the range 0.8M to 8M evolve through
central hydrogen and helium burning and enter the AGB with an electron-
degenerate carbon oxygen (C-O) core. The cores of these stars do not reach
110 Amanda I. Karakas

the ≈ 800 million K temperatures required for carbon burning. Stars with
initial masses between 9–11M may also become AGB stars after core carbon
burning; these objects are referred to as super-AGB stars. Stars more massive
than about 11M evolve through the central carbon, neon, oxygen and silicon
burning stages and end their lives as core collapse supernovae. We do not dis-
cuss these massive stars and refer the reader to the lecture notes by Arnould,
see also Woosley, Heger, & Weaver [16] and references therein for details.
The dividing mass which separates low and intermediate mass is not ar-
bitrarily chosen but corresponds to the transition from degenerate to non-
degenerate core helium ignition (which is at about 2.25M). The lower mass
limit of 0.8M is the minimum mass required to ignite helium and evolve
through central helium burning. The upper mass limit of 8M is the maxi-
mum mass which avoids core carbon ignition. The mass ranges above are valid
for stars with a composition similar to that of our Sun, which is composed of
approximately 70% hydrogen by mass, 28% helium (or more precisely, 4 He),
and about 2% metals. The metal content is referred to as the global metal-
licity, Z, and is comprised of about half 16 O. Using the solar abundances of
Anders & Grevesse [17], we obtain a global solar metallicity of Z = 0.019.
The revised solar abundances of Asplund, Grevesse & Sauval [18] has reduced
the global metallicity to Z ≈ 0.015, primarily because the oxygen abundance
of the Sun has been reduced by about a factor of ∼ 2 [19].
In real stars, where it is impossible to measure the composition of every
element, the Fe/H ratio has been taken to represent the global metallicity of
the star due to its ease of measurement. It is usually given as a logarithmic
ratio relative to the Fe/H ratio in the Sun, that is,
[Fe/H] = log10 (Fe/H)star − log10 (Fe/H) , (1)
where [Fe/H] = 0 corresponds to a star with the same iron abundance as
in our Sun; a star with [Fe/H] = −1 has an atmospheric composition that
has 10 times less iron than in our Sun. Metal-poor stars are defined as those
with [Fe/H] < 0, and metal-rich stars with [Fe/H] > 0. In metal-poor stars,
the definitions given above are shifted to lower mass, where for example, the
minimum mass for degenerate He ignition is about ≈ 1.75M at Z = 0.0001.
Ancient very metal-poor stars with [Fe/H] ≤ −2 have been found in the halo
of our Milky Way Galaxy [20], where the most metal-poor stars discovered
have [Fe/H] < −5 [21]. Stars with [Fe/H] < −3 behave very differently to
their metal-rich counterparts, however, we will be discussing the evolution
and nucleosynthesis of solar-like stars with Z ≈ Z or [Fe/H] ≈ 0, unless
otherwise indicated.
The distribution of stellar birth masses is governed by the initial mass
function. The initial mass function of the local solar neighbourhood [22, 23]
indicates that the majority (∼ 90%) of stars have masses less than 0.8M ,
and that the remainder of the stars (about 10%) have masses between about
0.8 to 8M . Massive stars comprise much less than 1% of all stars. From
the perspective of the origin of the elements in the Universe, it is instructive
 Nucleosynthesis of Low and Intermediate-mass Stars 111

Table 1. Stellar lifetimes for stars of initial masses between 0.8 to 25M . The
lifetimes for the 25 and 15M stars are from Woosley et al. [16]; lifetimes for the
lower mass stars are from Karakas & Lattanzio [24]. Lifetimes are given in Myr,
which is units of 106 years, or in Gyr, which is units of 109 years

Initial mass (M ) Main-sequence lifetime Total stellar lifetime

25 6.7 Myr 7.5 Myr
15 11 Myr 13 Myr
5 78 Myr 102 Myr
2 0.87 Gyr 1.2 Gyr
1 9.2 Gyr 12 Gyr
0.8 20 Gyr 32 Gyr

to examine the stellar lifetimes of stars in each of these mass groupings. The
most important lifetime is the time a star spends on the main sequence, fusing
hydrogen to helium in the core. This is because it is the longest lived phase
of stellar evolution and therefore representative of the entire nuclear burning
lifetime of a star. In Table 1 we show the main sequence and total stellar
lifetime for a selection of initial stellar masses.
Stars progressively enrich the interstellar medium with the products of stel-
lar nucleosynthesis. This occurs because massive stars explode as supernovae,
releasing vast quantities of processed material, whereas lower mass stars like
our Sun slowly lose their outer envelopes by stellar winds. From Table 1 it is
clear that stars more massive than about 8M evolve and die very quickly, in
under 20 Myr, short by even geological standards. Lower mass stars, on the
other hand, evolve much more slowly, with a star like our Sun spending about
10 Gyr on the main sequence. Stars less massive than ≈ 0.8M will spend so
long on the main sequence that they will not have had a chance to evolve and
release their processed material into the interstellar medium. These very-low
mass stars have not yet contributed to the chemical enrichment of our Galaxy.
The stars that evolve through the asymptotic giant branch phase are the
0.8 to 8M objects with relatively long lives. Initial mass function consid-
erations ensure that there are very many of them in the Galaxy, and this
means that they contribute in a significant way to the origin and evolution
of elements. Their contribution will be quite different to massive stars that
explode as supernovae, owing to the different time scales of their contribu-
tion. However, as an example of their importance to the composition of the
Galaxy, it has been estimated that up to 90% of the dust in the Galaxy today
originated in the outflows from AGB stars [25]. If we want to answer the
question “Where did the elements come from?”, it is essential that AGB stars
be studied alongside their more massive cousins!
112 Amanda I. Karakas

4 3
Log 10 (Luminosity)

5M
2 1

1M
0

4.4 4.2 4 3.8 3.6

Log(Teff)

Fig. 2. Hertzsprung-Russell diagram for a 1M and a 5M star of solar composi-
tion. Evolutionary tracks have been plotted from the zero-age main sequence to the
asymptotic giant branch. The location of the main sequence is noted by the position
of the label. The x-axis is the effective temperature of the model star, shown in units
of log 10 Teff , whereas the y-axis is the luminosity, in units of log 10 (L/L )

3 Evolution and nucleosynthesis prior to the AGB

All stars begin their nuclear-burning life on the main sequence, burning hy-
drogen to helium in their cores. The majority of a star’s nuclear-burning
life is spent on the main sequence, which is why we find most of the stars
observed in M3 (Figure 1) in this phase of evolution. The stars in M3 typi-
cally have masses between about 0.18 to 0.8M [26], which means that their
main-sequence lifetimes are exceedingly long (≥10 Gyr). In the following, we
describe the evolution of stars of masses 1M and 5M of solar composition
(taken as Z = 0.02 in this case), and note that the 0.8M stars of M3 will
undergo the same evolution. One main difference is that all of the 1 and 5M
stars have long evolved away in old globular clusters such as M3. The 1M
star spans the range of evolutionary behaviours observed for low-mass stars,
which go through the core helium flash and do not experience the second
dredge-up. The 5M is representative of intermediate-mass stars which do
experience the second dredge-up.
 Nucleosynthesis of Low and Intermediate-mass Stars 113

In Figure 2 we show the theoretical evolutionary tracks of a 1 and 5M

star on a HR diagram. In this figure, and in what follows we show model
results from Karakas & Lattanzio [24] and Karakas [27]. These models were
computed using the Monash version of the Mount Stromlo Stellar Structure
code, and we refer to [24] and references therein for details. It is important
to remember that the results presented here do depend on the details of the
stellar models, with different codes sometimes giving different results. For
example, the inclusion of core overshoot during the main sequence will lower
the upper mass limit from ∼ 8M to ∼ 5 − 6M of a star that enters the
AGB with a C-O core (e.g., [28]).
From a quick inspection of Figure 2 it is clear that the 5M is much
brighter on the main sequence than the 1M star by almost three orders of
magnitude. The location of the first giant branch (also commonly referred to
as the “red giant branch”) is in the upper right of the diagram. A 1M red-
giant star will be cooler (Teff ≈ 3, 100 K) than the 5 (∼ 4, 500 K), although
it has a similar luminosity (log10 (L/L ) ≈ 3.2). The luminosity of the 5M
is much higher during the AGB phase. The other main difference is the blue-
ward extension of the core helium burning track for the 5M model, which is
not at all apparent for the lower mass star. Let’s now look at the evolution of
the 1M star in more detail.

3.1 The Evolution of a 1M Star

Figure 3 shows a schematic HR diagram for a 1M star [29, 30]. Core hy-
drogen burning occurs radiatively, with the central temperature and density
increasing with the mean molecular weight (points 1–3). The temperature
during core H burning is ≈ 15 × 106 K, which means that the main energy
generation reactions are the pp chains. At core H exhaustion, the hydrogen
profile is shown in inset (a) in Figure 3, and the star now leaves the main
sequence. Following this, the star crosses the Hertzsprung Gap (points 5–7),
and nuclear burning is established in a shell surrounding the contracting 4 He
core. Inset (b) shows the advance of the H-shell during this evolution. Simul-
taneously, the outer layers expand and as a consequence become convective,
due to an increase in the opacity. When the star reaches the Hayashi limit
(point 7), convection has extended all the way from the surface to the deep
interior, and the star is now on the first giant branch (FGB). The star is very
big (up to ∼ 200 times the radius on the main sequence) but most of the mass
in the core is within a small fraction of the total radius. A consequence of
this is that the outer layers are only tenuously held onto the star and can be
lost through an outflow of gas called a stellar wind. About ∼30% of a star’s
total mass can be lost on the FGB, depending upon the length of time a star
spends in this phase of evolution.
During the star’s ascent of the giant branch, the convective envelope moves
inward, mixing the outer layers with internal matter that has experienced par-
tial H-burning, shown in inset (c) of Figure 3. This mixing event is known as
114 Amanda I. Karakas

convective

convective
T)

envelope

envelope
g( He
lo
CO core
core

He shell H shell
inset (d)
H shell
Beyond point 15 is the
inset (f) Thermally pulsing AGB

inset (e) First Thermal 9 = Core He Flash

Pulse = 15
10
10 11 12 13 14 convection
11

Beyond point 14 the 8

Early AGB begins

H abundance
12
10 11 12 13 14
He abundance
H abundance

Core He
Log(L)

13 exhaustion
= 14
14
78
mass fraction 13 inset (c)
12
11
10 mass fraction

8 = First Dredge-Up

7
1
H abundance

6
5
2
H abundance

4 = core H
3 exhaustion
3 4
5
4 2 6
7 inset (b)

inset (a) 1 = ZAMS mass fraction

mass fraction

Log(Teff)

Fig. 3. Hertzsprung-Russell diagram for a 1M star of solar composition, from the
zero-age main sequence to the tip of the asymptotic giant branch. Figure provided
by J. C. Lattanzio
 Nucleosynthesis of Low and Intermediate-mass Stars 115

the first dredge-up (FDU) and affects the surface composition of low to inter-
mediate mass stars. Note that the region reached by convection was partially
processed by hydrogen burning during the main sequence phase, but is ex-
ternal to the location of the H-shell. The composition of this material is still
mostly hydrogen but with some added 4 He together with the products of
CN cycling. The main surface abundance changes are an increase in the 4 He
abundance by ∼ 5%, a decrease in the 12 C abundance by about 20%, and an
increase in the 14 N and 13 C abundances by 30% and a factor of ∼ 2, respec-
tively. The number ratio of 12 C/13 C drops from its initial value to lie between
18 and 30 [31, 32]. The FDU leaves behind a sharp composition discontinuity
exterior to the position of the H-burning shell. In the lowest-mass stars the
relatively long lifetime on the first giant branch allows the H-shell to reach the
composition discontinuity and erase it. This does not occur in intermediate
mass stars, which leave the first giant branch before the H-shell can erase the
discontinuity.
During the ascent of the FGB, the He core continues to contract and heat
and becomes electron degenerate. Neutrino energy losses from the centre cause
the temperature maximum to move outward, as shown in inset (d) of Figure 3
and in Figure 4. The FGB lifetime is terminated when the necessary temper-
atures for central helium ignition are reached, at about 100 million K. At this
temperature the triple alpha reactions are ignited at the point of maximum
temperature but with a degenerate equation of state. The temperature and
density are essentially decoupled and this leads to a violent helium ignition
that is referred to as the core helium flash (point 9). During a core helium
flash, the surface luminosity does not change greatly but the helium lumi-
nosity may reach up to 1010 L during the first flash. The helium-burning
luminosity oscillates a few times after the main He-flash, before settling down
to quiescent core He-burning (see Figure 4). During the flash, about 3% of
the 4 He in the core is converted into 12 C, however it seems unlikely that there
is any mixing between the core and envelope in solar metallicity stars. There
are indications that such mixing takes place in extremely low-metallicity stars
with [Fe/H] ≤ −3 (e.g., [33, 34, 35]).
The maximum initial mass for the core He-flash to occur, Mflash , is about
2.25M at Z = 0.02 using the models of Karakas, Lattanzio, & Pols [36] that
include no convective overshoot. Models with convective overshoot during
core H burning find that the upper mass limit occurs at lower mass, where
Mflash ≤ 1.6M [37]. Note that Mflash is also reduced in lower metallicity
models. For a star with an initial mass of 1M , the first giant branch is the
next longest phase of evolution after the main sequence, and the star will
typically spend ≥ 2.8 Gyr there. This is also why the FGB of M3 is the second
most populated branch in Figure 1.
Following the core helium flash, the star quickly (∼ 106 years) moves to
the Horizontal Branch (HB), where it burns 4 He in a convective core, and
hydrogen in a shell (that provides most of the luminosity). This corresponds
to points 10–13 in Figure 3. The coulomb repulsion is larger for He than for H,
116 Amanda I. Karakas

Fig. 4. (Top panel) radiated (dashed line) and He-burning (solid line) luminosity
during the core helium flash for the 1M , Z = 0.004 model. (Bottom panel) mass of
the temperature maximum as a function of time. At the flash peak, the maximum
temperature is 0.2M from the centre of the star. The evolution of a 1M , Z = 0.02
star would be very similar although somewhat less extreme owing to the higher initial
metallicity

hence more energy is required by the triple alpha process to maintain the star
in hydrostatic equilibrium. This is coupled with the fact that less energy
is released by the triple alpha process per reaction than during hydrogen
burning. For example, the reaction 8 Be + 4 He → 12 C releases 7.65 MeV of
energy per reaction, compared to ∼ 26 MeV per 4p → 4 He reaction. The
overall result is that the core helium burning lifetime is about 100 Myrs,
compared to 10 Gyrs during the main sequence.
Helium burning increases the fraction of 12 C, which in turn increases the
abundance of 16 O from the reaction 12 C(α, γ)16 O. It is during core helium
burning that the star first experiences semiconvection: the outer layers of the
convective core become stable to the Schwarzschild convection criteria but
unstable to that of Ledoux, see [38], [39] and [12] for a detailed description of
 Nucleosynthesis of Low and Intermediate-mass Stars 117

this phenomenon. The semiconvection causes the composition profile to adjust

itself to produce convective neutrality with the resulting profiles as shown in
inset (e) of Figure 3. Note that semiconvection is different to core overshoot-
ing, which extends the convective border beyond the formal boundary set by
the Schwarzschild criteria. If overshooting is included the core hydrogen and
helium burning lifetimes are longer, because more fuel is added to the core,
along with increases in the mass of the H-exhausted core at the beginning of
the AGB (e.g., [40, 41, 42]).
Following core helium exhaustion (point 14) the star ascends the giant
branch for the second time, and this is known as the asymptotic giant branch.
The properties of the C-O core left after core He-burning depend on a number
of factors including the numerical treatment of convection and the inclusion
of convective overshoot. The uncertain rate of the 12 C(α, γ)16 O reaction also
helps to determine the final C/O ratio and composition of the core [43,44]. As
the star ascends the AGB, the core becomes electron degenerate, and the star’s
energy output is provided by the He-burning shell (which lies immediately
above the C-O core), along with the H-burning shell. The structure is shown in
inset (f) in Figure 3, and schematically in Figure 10. The AGB is divided into
two regimes: the early-AGB, prior to the first instability of the He-burning
shell, and the thermally-pulsing AGB (TP-AGB) beyond this (point 15 in
Figure 3).

3.2 The Evolution of a 5M Star

The evolution of a 5M is qualitatively similar to the 1M discussed above.

The main difference during the main sequence is that the higher temperatures
in the core cause CNO cycling to be the main source of H fusion. The higher
temperature dependence of these reactions (T ∝ 1017 compared to 104 for the
pp chains) results in the formation of a convective core. The main-sequence
evolution of the 5M star corresponds to points 1–4 in Figure 5. Following
core H exhaustion, the core contracts and H burning is established in a shell as
the star crosses the Hertzsprung Gap, points 5–7 and inset (b) of Figure 5. As
the star ascends the first giant branch, the inward movement of the convective
envelope (point 8) reaches regions where partial hydrogen burning occurred
during the main sequence. Thus the products of CN cycling (14 N and 13 C)
are mixed to the surface in much the same way as for lower mass stars; see
also inset (c) of Figure 5. The depth of the first dredge-up relative to the total
mass of the star is less for the 5M than the 1M , a point we come back to
in Sect. 3.3.
For these more massive stars, the contracting He cores do not become
electron-degenerate during the ascent of the FGB and the ignition of 4 He
occurs in the centre under non-degenerate conditions. The star then settles
down to a period of central He-burning that lasts about 20 Myrs. The core
He-burning timescale is much shorter than the 90 Myrs spent on the main
sequence. Intermediate-mass stars spend longer during core He-burning than
 118

inset (d) inset (e) TP-AGB

10 10 11 12 13 14 COcore

conv-
ective
11

envelope
after He shell
before H shell
12 inset (f)
10 11 12 13 14
unchanged 15 = Second
convection

H abundance
Dredge-Up

He abundance
13 mass fraction
Core He 9 = Core He ignition 8
12

H abundance
14 13

He abundance
exhaustion = 14
Amanda I. Karakas

mass fraction
11
6 10 H abundance
5
7 8 = First Dredge-Up 7 8

Log(L)
inset (c)
3 4 = Core H exhaustion
2 mass fraction
ZAMS = 1
inset (b)
1

2
7
H abundance

H abundance
3 6
4 5
inset (a) mass fraction
4
mass fraction
Log(Teff)

Fig. 5. Hertzsprung-Russell diagram for a 5M star of solar composition, from the zero-age main sequence to the tip of the asymptotic
giant branch. Figure provided by J. C. Lattanzio
 Nucleosynthesis of Low and Intermediate-mass Stars 119

on the FGB (20 compared to 3 Myrs). This is different to lower-mass stars

that have very long FGB lifetimes. These lifetimes affect the relative ratio
of main sequence to FGB and core He-burning stars that are visible in any
cluster.
The extent of the blue loops seen in the HR diagram in Figure 2 is de-
termined by the competition between the two burning sources: central He-
burning and H-shell burning. It is during the blue-ward excursion across the
HR diagram that the star may cross the instability strip and be observed as
a Cepheid variable (points 10–14). Following core He exhaustion, the strong
expansion of the star caused by the structural re-adjustment to He-shell burn-
ing causes the H-shell to be extinguished as the star begins the ascent of the
asymptotic giant branch. With the entropy barrier of the H-shell gone, the
convective envelope moves inward and reaches regions where complete H burn-
ing had taken place (during the main sequence), resulting in an increase in
the surface composition of 4 He and 14 N. In the process the mass of the H-
exhausted core is reduced, because while 4 He is mixed outward, H is mixed
inward by the convective envelope, see inset (e) in Figure 5. This inward move-
ment of the convective envelope and reduction of the H-exhausted core mass
during the early-AGB phase is known as the second dredge-up (SDU). There
is a critical minimum mass below which the SDU does not occur (∼ 4.5M
at Z = 0.02); see Sect. 3.3 and Figure 6. Following the SDU, the H-shell is
re-ignited and the first He-shell instability soon follows. The star is now on
the TP-AGB, where the structure is qualitatively the same for all masses.

3.3 The First and Second Dredge-up

Prior to the TP-AGB, the surface composition of a star will be altered by the
first and possibly second dredge-up events. In this section we look at these
changes in more detail. In Figure 6 we show the innermost mass layer reached
by the convective envelope during the first (solid lines) and second (dashed
lines) dredge-up as a function of the stellar mass. The black lines represent
the Z = 0.02 models, the red lines the Z = 0.008 models, and the blue lines
the Z = 0.004 models. The FDU does not extend very far inward in the most
massive models, depending on Z, but extends further inward during the SDU
(Figure 6). The depth reached by the SDU is approximately the same for all
the 5 and 6M models, regardless of the initial metallicity. From Figure 6 we
predict that the largest changes to the surface composition from the FDU are
seen for stars of ≈ 2.5M, and the largest changes from the SDU for the most
massive stars that experience it (6M in the case of the models shown in the
figure).
In Figure 7 we illustrate the changes made to the surface abundance ratios
of 12 C/13 C, 14 N/15 N, 16 O/17 O, and 16 O/18 O for the Z = 0.02 models after
the FDU and SDU mixing events. All isotopic ratios are given as number
ratios. The 12 C/13 C ratio shows a noticeable decline after the FDU, from the
initial value of 90 to values between 20 and 30 [31, 32]. The decrease in the
120 Amanda I. Karakas

Fig. 6. Innermost mass layer (Mdu /M0 ) reached by the convective envelope during
the first (solid lines) and second dredge-up (dashed lines) as a function of the stellar
mass, M0

carbon isotopic ratio is caused by the convective envelope moving inward to

a position interior to the 13 C-rich region left behind after core H burning. In
Figure 8 we show the composition as a function of the stellar mass (in M )
at the end of core H-burning for the 1M and 3M , Z = 0.02 models. The
abundances are in number fraction, Y , where the mass fraction X = Y A,
and A is the  atomic mass. By definition, the mass fraction of all species sums
such that X = 1. We will use the number fraction Y throughout the text.
The position of the maximum inward penetration of the convective envelope
is also noted. For all models, hydrogen burning leaves behind a region rich in
13
C that is engulfed by the convective envelope during the FDU.
Theoretical predictions of the 12 C/13 C ratio after the FDU for intermediate-
mass stars are in good agreement with the observations, to within ∼ 25%
[45, 31, 46, 32]. The situation is the opposite for low-mass stars where the pre-
dicted trend, which has the 12 C/13 C ratio increasing with decreasing mass,
has been found to be inconsistent with the observations e.g., [47]. The same
is true for Population II giants in globular clusters, where the deviation be-
tween theory and observation is more extreme [48, 49, 50, 51]. The same is
true for the C/N ratio, which is also observed to be lower in red-giant stars
than predicted by standard stellar evolution theory [31]. The observations of
low 12 C/13 C and C/N ratios have been interpreted as evidence for extra mix-
ing taking place between the base of the convective envelope and the H-shell.
Observations indicate that the conflict between theory and observation does
not arise until after the deepest first dredge-up, hence this mixing most likely
takes place after the H-shell has erased the composition discontinuity left by
the FDU event. The physical mechanism that causes the extra mixing is still
 Nucleosynthesis of Low and Intermediate-mass Stars 121

Fig. 7. The surface abundance ratios (by number) of 12 C/13 C, 14 N/15 N, 16 O/17 O,
and 16 O/18 O as a function of the stellar mass for the Z = 0.02 models. The solid
line (with points) show the ratios after the FDU and the dashed red line (with stars)
show the ratios after the SDU. The initial ratio is indicated on each panel

uncertain, with meridional circulation caused by rotation [52] and thermo-

haline mixing suggested as possible candidates [53, 54, 55]. Most algorithms
used in deep mixing models are parameterised in some way to reproduce the
observed data [56, 57, 32, 58]. For example, Boothroyd & Sackmann [32] take
material from the base of the convective envelope into a region just hot enough
for some nuclear processing, and then transport the material back to the enve-
lope. The temperature and depth to which the material is mixed, along with
the amount of material in the circulation current, is parametrized to obtain
the observed abundances.
From Figure 7 and 8 we see that the 14 N/15 N ratio increases dramatically
in all models after the FDU and SDU events, regardless of the initial mass or
metallicity. For a 2M star, the increase is about a factor of 4, whereas this
increases to about a factor of 6 for 3–4M stars. The composition profiles
(Figure 8) demonstrate the mechanism for the large increase in the 14 N/15 N
ratio. The increase can be attributed to the production of 14 N at the expense
of 12 C but also to a reduction in the abundance of 15 N from CNO cycling.
The equilibrium value of the 14 N/15 N ratio is 2.5 × 104 [9], much higher than
the 14 N/15 N ratios seen in stellar models after either the FDU or SDU. The
maximum inward extent of the convective envelope reaches the first 14 N peak
122 Amanda I. Karakas

–3

O-16 N-14
–4

C-13
–5 C-12 O-17

–6
Log Y

–7
O-18 N-15

–8
F-19
M = 1 Z = 0.02
–9
Maximum depth reached by FDU

0.0 0.1 0.2 0.3 0.4 0.5

mass
–3

O-16 N-14 C-12

–4

–5 O-17
C-13
–6
Log Y

–7
O-18

–8 F-19
N-15

–9 M=3 Z=0.02
Maximum depth reached by FDU
0.0 0.5 1.0 1.5 2.0 2.5
mass

Fig. 8. The composition profile for the C, N, and O isotopes as a function of the
interior mass for the 1M and 3M Z = 0.02 models. The unit on the y-axis
is the logarithm of the number fraction, Y , where the mass fraction is given by
X = Y A, and A is the atomic mass. The composition profile is a snap-shot of the
interior composition of the star at an instant in time, in this case at the end of core
hydrogen exhaustion
 Nucleosynthesis of Low and Intermediate-mass Stars 123

Fig. 9. The fractional changes to the surface 23

Na abundances after the FDU (solid
lines) and SDU (dashed lines)

but only in models with the deepest FDU, or for those models that experience
the SDU, does the convective envelope reach the second 14 N plateau caused
by ON cycling. Hence, these stars are predicted to have the highest 14 N/15 N
ratios.
Figure 7 shows that the oxygen isotopes experience only small changes
from the first and second dredge up, with an increase in 17 O, and depletions
in 18 O, depending on the initial mass and metallicity. Figure 8 shows that the
convective envelope may move into the region depleted in 18 O and enriched
in 17 O, but leaves the 16 O abundance essentially unchanged. For elements
heavier than oxygen, the first and second dredge-up affects the surface abun-
dances of 7 Li, 19 F, 23 Na, and to a lesser extent the Ne, Mg and Al isotopes.
Lithium is easily destroyed in stellar atmospheres at relatively low tempera-
tures and is depleted by burning during the pre-main sequence phase. Hence
the FDU results in strong surface depletions of this light element. Fluorine is
also depleted at the surface in all but the lowest mass models, with maximum
depletions on the order of ∼ 20%. In Figure 9 we show the fractional change
to the surface 23 Na abundance as a result of the FDU and SDU. For elements
heavier than O, this element shows the largest change, with increases of up
to a factor of 2 for the most massive models after the SDU.

4 Evolution during the AGB

We now focus on the thermally-pulsing AGB phase of evolution, which alters

the surface abundances of the models in two distinct and important ways.
The first is through the operation of the third dredge-up, which can occur
periodically after each thermal pulse (TP) and is the mechanism for turning
124 Amanda I. Karakas

ec tion p
onv dge
−U
nC Dre
rive
h−d
Flas

CO degenerate
core Intershell region:
about 75% 4He Convective
about 22% 12C Envelope
about 2% 22Ne
< 1 % 16O

Helium burning shell

Hydrogen burning shell

Fig. 10. Schematic structure of an AGB star showing the degenerate C-O core
surrounded by a He-burning shell above the core, and a H-burning shell below the
deep convective envelope. The burning shells are separated by an intershell region
rich in helium (∼ 75%) and carbon (∼ 22%), with some oxygen and 22 Ne. From [36]

(single) stars into carbon-rich stars, where the C/O ratio of the envelope
exceeds unity. The second mechanism is hot bottom burning (HBB). In this
section we focus on the structure and evolution of AGB stars, with a discussion
of the nucleosynthesis in Sect. 5.
There are many reviews on AGB evolution including Iben [59], Frost &
Lattanzio [29], Wood [60], Busso, Gallino, & Wasserburg [7], and Herwig [1].
Here we briefly review the main features. Following core He-exhaustion, the
structure of the star is shown schematically in Figure 10. The H-exhausted
core of the star, composed primarily of carbon and oxygen, begins to contract.
At the same time the outer layers expand and the star ascends the giant branch
for the second time. He-burning is ignited in a thin layer around the degenerate
C-O core but instabilities quickly set in, due to the thinness of the burning
shell. At this stage the star is said to have entered the thermally-pulsing AGB
(TP-AGB) phase of evolution. This stage of evolution is characterized by
relatively long periods of quiescent H-shell burning, known as the interpulse
phase, interrupted by instabilities of the He-shell. The ashes of the H-shell
increase the thickness of the hydrogen depleted region, until eventually the
next thermonuclear runaway or thermal pulse. During a TP, the He-shell burns
fiercely, producing up to ∼ 108 L for a few hundred years. This enormous
 Nucleosynthesis of Low and Intermediate-mass Stars 125

Fig. 11. Temporal evolution of the radiated (solid line) and He-burning (dashed
line) luminosities during the TP-AGB for the 5M , Z = 0.02 model. The AGB life
cycle is shown by the repetition of the He-shell flashes (24 in total), followed in each
case by much longer periods of H-shell burning. The interpulse period of this star is
≈ 7, 000 years

quantity of energy does not reach the surface but goes into expanding the outer
layers of the star, causing the H-shell to become extinguished. In Figure 11
we show the He-shell and radiated luminosity from a 5M , Z = 0.02 model
during the TP-AGB phase. The energy from the TPs does not noticeably
affect the radiated luminosity, which stays approximately constant at about
20,000L.
The energy produced by the flash powers a convective region in the He-
shell, which has the effect of homogenizing abundances in this region. In Fig-
ure 12 we show the convective regions in the intershell during the first 5 TPs
of a 1.9M , Z = 0.008 model. The teardrop-shaped green pockets represent
flash-driven convection. Following a TP the star expands, this is shown by
the outward movement of the base of the convective envelope, at ≈ 0.56M .
When the flash dies down, the outer convective envelope moves inward in
mass to the region mixed by intershell convection. This phase is known as the
third dredge-up (TDU), and can occur after each thermal pulse. Note that
there are no TDU events visible in Figure 12, where the base of the convec-
tive envelope is seen to move inward but does not penetrate the flash-driven
convective region. The result of the third dredge-up is to mix the products
of partial He-burning (mostly 4 He and 12 C, see Figure 10) to the surface.
This is the process which turns (single) stars into carbon-rich stars. Following
dredge-up, the star contracts and the H-burning shell is re-ignited and the star
enters a new interpulse phase. The cycle of interpulse–thermal pulse–dredge-
up may occur many times on the AGB, depending on the initial mass and
composition, as well as on the mass-loss rate.
126 Amanda I. Karakas

0.60

0.58
moss (solor mosses)

0.56

0.54

0.52

0.50

0.48
1000 1500 2000 2500 3000 3500
model number
Fig. 12. Convective regions for the 1.9M , Z = 0.008 model during the first
five thermal pulses. The He-intershell extends from about 0.52M to 0.56M . The
teardrop-shaped pockets correspond to the flash-driven convective region that ex-
tends over most of the He-intershell. These have the effect of homogenizing the
abundances within the He-intershell. Convective regions are shaded in green and
radiative zones in magenta. The x-axis is nucleosynthesis time-step number, which
is a proxy for time. For this model, the duration of the convective zones are about
250 years

In the most massive AGB models, the convective envelope can dip into the
top of the H-shell, resulting in nuclear burning at the base of the convective
envelope. This phenomena is known as hot bottom burning (HBB) and can
also dramatically alter the surface composition. It has been known for some
time that intermediate mass stars over about 5M develop deep convective
envelopes with very high temperatures at the base, allowing for nuclear burn-
ing and some energy generation [61]. The observational evidence that HBB is
occurring in massive AGB stars came from the lack of luminous carbon-rich
stars in the Magellanic Clouds [4]. Many of these luminous, O-rich stars were
later found to be rich in lithium [62,63,64]. The first detailed calculations were
not made until the early 1990’s by Blöcker & Schönberner [65] and Lattanzio
[66]. Blöcker & Schönberner found that the linear core-mass-luminosity rela-
tion first proposed by Paczynski [67] does not apply to stellar models with
HBB. Boothroyd, Sackmann, & Ahern [68] found that HBB prevents the
 Nucleosynthesis of Low and Intermediate-mass Stars 127

Fig. 13. The mass of the hydrogen-exhausted core (solid line) and helium-exhausted
core (dashed line) as a function of time for the 3M , Z = 0.02 model (top) and 5M ,
Z = 0.02 model (lower)

formation of a carbon-rich atmosphere by burning 12 C into 14 N and provided

a mechanism for the lack of bright C-stars in the LMC.
In summary, the AGB evolutionary cycle can be broken down into four
distinct phases:
1. The on-phase or thermal pulse, which is when the He-shell burns brightly
producing up to 100 million L in a short time (∼ 102 years). The energy
drives a convective zone in the He-intershell.
2. The power-down phase when the He-shell dies down. The enormous
amount of energy from the TP drives an expansion of the whole star and
the H-shell is pushed to cooler regions and is extinguished.
3. The third dredge-up phase, which is when the convective envelope
moves inwards into regions previously mixed by the flash-driven convec-
tion. Carbon and other He-burning products are mixed to the stellar sur-
face.
128 Amanda I. Karakas

Interpulse period

Mass of H−exhausted core ΔMdredge

ΔMH

ΔMdredge
λ=
ΔMH

Time

Fig. 14. The definition of λ, shown schematically, where the x–axis represents time
and the y–axis represents the mass of the H-exhausted core

4. The interpulse phase, which is when the whole star contracts after the
TDU; the H-shell is re-ignited and provides most of the surface luminosity
for the next 104 years or so.
In Figure 13 we show the evolution of the H and He-exhausted cores for
the 3M and 5M , Z = 0.02 models. These models both show third dredge-
up. For the 5M model, deep third dredge-up starts after the fifth TP, this
corresponds to a decrease in the H-exhausted core mass. The 3M model, on
the other hand, does not experience much dredge up until the ∼ 13th thermal
pulse. One striking difference between the 3 and 5M models is the mass of
the He-intershell, which is much smaller by about a factor of five in the 5M
model. This mass difference has important consequences for the production of
heavy elements in low-mass AGB stars. This is because at a given dredge-up
efficiency, more matter is mixed into the envelope, at a smaller dilution, in
lower mass stars than in intermediate-mass AGB stars. We can quantify this
by defining λ, which is a parameter used to measure the depth, or efficiency of
the third dredge-up. If there is dredge-up, then a fraction of the H-exhausted
core will be mixed into the envelope
ΔMdredge
λ= . (2)
ΔMcore
where λ is the third dredge-up efficiency parameter, ΔMdredge is the mass
mixed into the envelope, and ΔMcore is the amount by which the He-intershell
increased over the previous interpulse phase; see Figure 14.
The value of λ depends on physical parameters such as the core mass and
metallicity of the star. Exactly how λ depends on these quantities is unknown.
From the above definition it should be clear that when λ = 1, the core mass
does not grow but remains constant. In Figure 15 we show the efficiency of the
TDU as a function of the core mass for two 3M , Z = 0.02 models, computed
 Nucleosynthesis of Low and Intermediate-mass Stars 129

with two different stellar evolutionary codes. The top panel shows results from
Karakas et al. [36] whereas the lower panel shows results from Straniero et al.
[69]. The behaviour of λ with core mass is quite different, with the Karakas
model [36] showing more efficient third dredge-up at the same core mass.

Fig. 15. (Top) λ as a function of the core mass for the 3M , Z = 0.02 model.
(Bottom) Same, but using data from [69]

AGB nucleosynthesis not only depends on the efficiency of the third

dredge-up but also on the minimum core mass for thermal pulses, and on
the minimum core mass for the onset of the third dredge-up. These three
quantities were parameterised by Karakas et al. [36], and were found to vary
as a function of the total mass, envelope mass, and metallicity in a complex
130 Amanda I. Karakas

Fig. 16. The maximum value of λ calculated from the Z = 0.02, 0.008 and 0.004
models. From [36]

way. In Figure 16 we show the variation of the maximum λ calculated from

grids of stellar models covering a range of mass and metallicity. The general
trend is that λ increases with increasing stellar mass, at a given Z. λ also
increases with decreasing metallicity, at a given mass (see also [70]). This
means that it should be easier to make carbon stars in lower metallicity or
higher mass models. These three quantities still suffer from large uncertainties,
mostly because of our lack of understanding about how convection operates in
stellar interiors. Different stellar codes predict different behaviour (Figure 15;
[71, 72]).
It is important to know if the stellar models are providing an accurate
description of the efficiency of mixing in AGB stars. For example, the models
shown in Figure 16 do not predict any TDU for models less than 2M at
Z = 0.02. This is in conflict with observations of carbon stars that suggest
that the minimum mass for carbon stars in the Milky Way Galaxy is ≈ 1.5M
[73]. Thus models such as those presented in Karakas et al. [36] do not predict
an accurate picture of TDU mixing in low-mass AGB models. It is possible
to constrain the efficiency of the third dredge-up using other observables?
The two closest satellite galaxies of the Milky Way are the LMC and Small
Magellanic Cloud (SMC) and they both have thousands of known carbon stars
[74,75]. We know the distances to these two galaxies reasonably well, enabling
us to construct carbon-star luminosity functions (CSLF).
The convenient fact that the stellar luminosity on the AGB is a nearly
linear function of the H-exhausted core mass has stimulated the development
of synthetic AGB evolution models, as a quick way of simulating populations
of AGB stars. The main observational constraint which models must face is
the CSLF for the Magellanic Clouds. Synthetic AGB evolution calculations
 Nucleosynthesis of Low and Intermediate-mass Stars 131

performed by Groenewegen & de Jong [76] and Marigo, Bressan, & Chiosi
[77], treat λ as a constant free parameter, calibrated by comparison with the
CSLF. These studies suggest λ ∼ 0.6 at a core mass of 0.58M is required to fit
the CSLF of the Magellanic Clouds (see Chapter 3 in [14] for further details).
Other calculations take parameterizations of λ and the minimum core mass
for TDU [78,36] and adjust as required to fit the CSLF [79]. Stancliffe, Izzard,
& Tout [80] were able to reproduce the CSLF of the LMC by computing new
AGB models using the Cambridge STARS code, which predicts deeper TDU
at smaller core masses than Straniero et al. [69] or Karakas et al. [36].

4.1 Carbon stars

Thermal pulses and dredge-up may occur many times during the TP-AGB
phase, as shown by Figures 11 and 13. Each TDU episode mixes 12 C from the
He-intershell into the envelope and this has the effect of slowly increasing the
C/O ratio of the surface. Repeated TDU episodes can explain the transition
from M-type (C/O ≈ 0.5, similar to the Sun) to carbon star:
M → MS → S → SC → C, (3)
where SC-type stars have C/O equal to unity, and C-type stars have C/O
> 1 by definition. Many carbon stars also have surface enrichments of heavy
elements synthesized by the s-process (e.g., Ba, and Tc) [81, 82, 83, 84]. In
particular, enrichments in the radioactive element technetium is an indication
that the star is a real AGB star that has experienced “recent” nucleosynthesis
and mixing [85,86]. This is because the half-life of 99 Tc (the isotope produced
by the s-process) is 210,000 years, much shorter than the main sequence life-
time of low-mass stars.
There are other classes of carbon-rich stars including the cool and warm
R-type stars, the 13 C-rich J-type stars, CH-stars, and dwarf carbon stars [73].
Barium stars also show enrichments in carbon and heavy elements, although
they have C/O < 1 in general. Possibly ∼ 20% of all very metal-poor with
[Fe/H] < −2 are also carbon rich, with [C/Fe] ∼ 2 in some cases [20]. Stars in
these other classes are not on the AGB and are not responsible for producing
their own carbon enrichments. Some of them, such as the barium and CH-
type stars, are all known binaries [87,88] and thus presumably obtained their
carbon from a former AGB companion. These stars are also devoid of Tc
enrichments [89,90]. The warm R-type stars are all single stars [91], and may
all result from some type of binary-star merger event (see Izzard, Jeffery &
Lattanzio [92] and references therein). The rare J-type stars, with very low
12
C/13 C ratios, are still a mystery [93, 94].

4.2 Luminosity variability

Luminosity variability is a common feature of AGB stars. The cause of the

variability is mainly radial and non-radial pulsations of the envelope, although
132 Amanda I. Karakas

there is some evidence for variations caused by ejections of dust shells or orbit-
ing dust clouds. While thermal pulses generally occur over a much longer time-
span than the observed luminosity variability (104 years compared to a few
hundred days), they can interact with the shorter-term pulsations. There are
three groups of long-period variables that include AGB stars: Mira-type vari-
ables (large amplitude), and the semi-regular and irregular variables (smaller
amplitude than the Miras). There are also dust-enshrouded variables that are
too faint to be observed in the optical; e.g., OH/IR stars [95]. Stellar pulsation
theory, when combined with pulsation periods and absolute luminosities, can
be used to derive stellar masses [96]. The LMC and SMC provide large sam-
ples of variable AGB stars, where the entire mass range covered by AGB stars
(from 0.8 to 8M ) is observed [4]. The radial pulsation modes of AGB stars
are essentially confined to the convective envelope, with the nuclear-burning
core acting as a point source of energy and gravity. Hence, energy transport
is dominated by convection. We refer the reader to computations by Fox &
Wood [97] and Ostlie & Cox [98] for details.

4.3 Mass loss

Mass loss is an essential component of any study involving AGB stars. This is
because it is mass loss that terminates the AGB phase by removing the con-
vective envelope, and this in turn determines the level of chemical enrichment
by setting the number of thermal pulses and mixing events. Our knowledge
of the dependence of AGB mass loss on stellar properties is rather poor, al-
though advances are being made in the theory of AGB winds [99,100,101,102].
These winds are driven by the combination of stellar pulsation and radia-
tion pressure acting on dust grains. In AGB evolutionary calculations (e.g.,
[103, 104, 105, 24]) the mass-loss laws [106, 107] are based on observational
estimates and theoretical estimates, respectively. One important feature of
these laws is that the mass-loss rate increases very rapidly with increasing
stellar luminosity. This implies that most of the mass is lost right at the end
of the AGB.

5 Nucleosynthesis during the AGB

It is on the TP-AGB that the richest nucleosynthesis occurs for low and
intermediate-mass stars, even though stars spend such a short amount of
time there compared to previous evolutionary phases. The nucleosynthesis is
driven by thermal instabilities of the helium-burning shell, reviewed in Sect. 4.
Of particular importance is the action of repeated third dredge-up events that
mix the products of He-burning to the stellar surface. Material from the He-
shell will become part of the next hydrogen shell, where they will experience
proton captures during the next interpulse period. For this reason, not only
do we need to consider nucleosynthesis in the thermal pulse itself but also
 Nucleosynthesis of Low and Intermediate-mass Stars 133

the effect of subsequent nucleosynthesis of the dredged up material in the H-

shell. Hot bottom burning occurs in the most massive AGB stars, which takes
place when the base of the envelope becomes hot enough to sustain proton-
capture nucleosynthesis. Overall, the TP-AGB gives rise to a combination of
H and He-processed material that is expelled by the star as its envelope is
lost through stellar winds.

5.1 Nucleosynthesis in the hydrogen-burning shell

(p, γ ) (p, γ ) (p, γ )

21Ne 22Na 25Mg 26Al 27Si

(β+)
(β+) (β+) (β+)

21Na 22Ne 25Al 26Mg (β+)

(p, γ )
(p, γ ) (p, γ ) (p, γ )

19 20 23 24 27 28
F Ne Na Mg Al Si
(p, γ ) (p, α ) (p, γ ) (p, α ) (p, γ )
Ne−Na Mg−Al

Fig. 17. Reactions of the Ne-Na and Mg-Al chains. Unstable isotopes are denoted
by dashed circles

Hydrogen is burnt to 4 He via the CNO cycles, but the Ne-Na and Mg-
Al chains also operate. The main result from the CNO cycle, besides the
conversion of H to He, is an increase in the abundances of 13 C and 14 N from the
destruction of other CNO species. The isotopes 12 C and 15 N are first destroyed
by the CN cycle, which comes into equilibrium quickly. Later, the oxygen
isotopes 16 O and 18 O are also destroyed to produce 14 N. The abundance of
17
O can be enhanced by the CNO cycle, depending on the uncertain rate of
the 17 O + p branching reactions. The H-shell ashes are important because
they affect the nucleosynthesis that occurs in the He-shell.
In the left part of Figure 17 we show the reactions of the Ne-Na chain
[10, 44], where unstable isotopes are denoted by dashed circles. The main
result of the Ne-Na chain is the production of 23 Na at the expense of the
neon isotopes, primarily 22 Ne, which begins to be destroyed at about 20 mil-
lion K. The production of sodium by the Ne-Na chain was examined in detail
by [108], who predicted that AGB stars could play an integral role in the
chemical evolution of sodium in the Galaxy. The rare neon isotope, 21 Ne is
134 Amanda I. Karakas

completely destroyed at temperatures over 40 million K. The dominant 20 Ne

is not significantly altered by H-shell burning, but the destruction of 23 Na at
temperatures over 80 million K can lead to a slight enhancement in the 20 Ne
abundance. The rate by which 23 Na is destroyed is important for determin-
ing sodium yields (see [109, 110]). Whether there is leakage out of the Ne-Na
chain into the Mg-Al chain depends on the relative rates of the uncertain
23
Na(p,α)20 Ne and 23 Na(p,γ)24 Mg reactions.
Magnesium and aluminium are altered in the H-burning shell via the acti-
vation of the Mg-Al chain, which begins operation at temperatures of about 30
million K [44]. This involves the radioactive nuclide 26 Al which has a ground
state 26 Alg that has to be considered a separate species from the short-lived
(τ1/2 = 6.35 s) isomeric state 26 Alm , since they are out of thermal equilibrium
at the relevant temperatures [44]. Hereafter, when we refer to 26 Al we are
referring to the ground-state, 26 Alg . In the right part of Figure 17 we show
the reactions involved in the Mg-Al chain [10, 44]. The first isotope in the
Mg-Al chain to be affected is 25 Mg, which is burnt to 26 Al. The lifetime of
β-decay relative to proton capture generally favours proton capture within the
H-burning shell. This produces the unstable 27 Si which β-decays (with a life-
time on the order of a few seconds) to 27 Al. The rate of 26 Mg + p is slow until
the temperature reaches about 60 million K, which leads to small reductions
in 26 Mg in the H-shells of AGB models. The abundance of 26 Mg is enhanced
by the β-decay of 26 Al in the H-shell ashes. Proton capture on 24 Mg requires
higher temperatures than those required for the other reactions in the Mg-Al
chain. For that reason AGB models show little change in the abundance of
this isotope due to the slow rate of proton capture at temperatures below
about 70 million K.
Even in the lowest mass AGB models the Ne-Na and Mg-Al chains are
active in the H-shell during the TP-AGB, depleting 22 Ne and 25 Mg and pro-
ducing 23 Na and 26 Al. The products of the H-shell are processed through
He-shell burning before being dredged to the surface, therefore only those
products that are not altered by α or neutron-captures will make it into the
envelope (e.g., 17 O and 23 Na, depending on the temperature in the He-shell,
whereas 14 N and 26 Al are destroyed).

5.2 Nucleosynthesis during thermal pulses

A He-shell flash occurs in the intershell region, which is composed mostly

of the ashes of H-shell burning (∼ 98% 4 He and 2% 14 N). The two main
energy-generating reactions are:
1. the triple-alpha process: effectively 3 4 He → 12 C,
2. the 12 C(α, γ)16 O reaction, which is relatively unimportant during ther-
mal pulses owing to the short He-burning timescale.
The He-burning shell in AGB stars is a rich source of nucleosynthesis.
The main result is that some of the 4 He in the shell is converted into 12 C by
 Nucleosynthesis of Low and Intermediate-mass Stars 135

Fig. 18. Logarithm of the surface C/O ratio as a function of time during the TP-
AGB for two 3M models of different metallicity. The top panel shows the C/O ratio
from the Z = 0.02 model, the lower panel shows results for the lower metallicity
Z = 0.004 model. Time has been scaled such that the start of the TP-AGB is t = 0

partial He burning, leaving the composition of the shell roughly 70–75% 4 He

(by mass) and 20–25% 12 C. There is a few percent (by mass) each of 16 O and
22
Ne, and trace amounts of other species including 17 O, 23 Na, 25 Mg, 26 Mg,
and 19 F. The exact composition of the He-intershell after a thermal pulse
depends on the mass and composition of the He-shell before the pulse, the
duration of the shell flash, as well as the peak temperature and density under
which the burning takes place. These quantities in turn depend on the mass
and metallicity of the star, and they evolve with time. This is because the core
is contracting while the star evolves along the TP-AGB, causing the He-shell
to become slightly more electron degenerate. Thus thermal pulses occur under
136 Amanda I. Karakas

somewhat more extreme conditions at the end of the TP-AGB (i.e., higher
temperatures and densities in a thinner shell) than at the beginning.
AGB stars are one of the most important sources of 12 C in the Galaxy.
An estimate of the contribution of 12 C from AGB stars suggests that they
produce roughly one third of the Galaxy’s inventory of 12 C, providing roughly
the same amount as core-collapse supernovae and Wolf-Rayet stars [8]. These
quantitative estimates are hindered by uncertainties in the depth and onset
of the third dredge-up, as discussed in Sect. 4. In Figure 18 we show the
log C/O ratio as a function of time during the TP-AGB for two 3M models,
one of solar composition and the other with Z = 0.004 ([Fe/H] ∼ −0.7). It
is evident that the lower metallicity model becomes a carbon star in fewer
thermal pulses, and that the final C/O ratio is higher. This occurs for two
reasons. First, there is less oxygen in the lower metallicity model so it is easier
to obtain the situation where C/O > 1. Second, dredge-up is more efficient.
During a thermal pulse, the 12 C(α, γ)16 O reaction does not have time
to produce much 16 O, resulting in a 16 O intershell composition that is ≈
1%. This is the case for all standard model calculations that do not assume
any convective overshoot of the flash-driven convective region [70]. Herwig
[111] does include convective overshoot at the inner border of the flash driven
convective zone and finds that some C and O from the C-O core is mixed into
the intershell. This has the effect of increasing the C and O abundances to
∼ 40 and ∼ 20%, respectively. There is a wealth of other He-burning products
including 22 Ne, 25 Mg, 26 Mg, and 27 Al [112, 113, 114], plus species produced
through the combined operation of the He and H-burning shells including 19 F
[115], and 23 Na [108].
The isotope 22 Ne is produced by the chain 14 N(α, γ)18 F, where 18 F
β-decays to 18 O allowing for the chain 18 O(α, γ)22 Ne. The composition of
22
Ne in the He-intershell is fairly high, at ≈ 2%. This is because the abun-
dant 14 N is completely converted into 22 Ne during the TP; see the top panel
of Figure 19. From this figure we note that the He-intershell is also enriched
in 23 Na. Sodium was not produced by He-burning, but was synthesized dur-
ing H-shell burning during the previous interpulse (by the CNO cycle and
Ne-Na chain, respectively). The next TDU episode will mix some fraction of
the He-intershell into the envelope. The effect of many of these efficient TDU
events will greatly affect the composition of the surface, shown in the bottom
panel of Figure 19. The 22 Ne abundance is seen to increase by almost an order
of magnitude (∼ 1 dex). If the 22 Ne abundance exceeds or is equal to the 20 Ne
abundance we should expect an enhancement in the elemental Ne composition
and this is the case for the 4M model.
If the peak temperature of the TP reaches about 300 × 106 K, the neutron-
rich Mg isotopes, 25 Mg and 26 Mg, can be synthesized by the 22 Ne(α, n)25 Mg
and 22 Ne(α, γ)26 Mg reactions. These two 22 Ne + α reactions have similar al-
though uncertain rates at He-shell burning temperatures (Figure 20; see also
[117, 118]). Owing to the relatively high temperatures required for these two
reactions, they are predicted to only occur efficiently in the most massive
 Nucleosynthesis of Low and Intermediate-mass Stars 137

Fig. 19. (Top panel) Composition profile for the 4M , Z = 0.004 model during
the 20th thermal pulse. The peak temperature in the He-shell during this pulse is
360 million K. (Bottom panel) Evolution of the 12 C, 22 Ne, 25 Mg, and 26 Mg surface
abundances of the 4M , Z = 0.004 model during the TP-AGB. The effect of HBB
can be seen on the surface abundance of 12 C
138 Amanda I. Karakas

–8

reaction rate (cm3 / mol / s)

–9
–10
–11
–12
–13 22Ne(a, g)26Mg
–14 22Ne(a, n)25Mg
–15
–16
0.2 0.25 0.3 0.35 0.4
Temperature (109 K)
Fig. 20. The NACRE rates for the 22 Ne(α, n)25 Mg and 22 Ne(α, γ)26 Mg reactions
at typical He-shell burning temperatures [116]. The reaction rate uncertainties are
not shown but for the 22 Ne(α, n)25 Mg reaction the uncertainties span more than an
order of magnitude at 0.3 GK

AGB stars (≥ 4M ). In Figure 21 we show the 22 Ne, 25 Mg, and 26 Mg intershell
abundances during five thermal pulses for the 4M , Z = 0.004 model. It is
evident that some of the 22 Ne is being converted into the neutron-rich Mg
isotopes. The initial decrease in the abundances of 22 Ne and the Mg isotopes
is caused by the flash-driven convective zone moving into a region that was
depleted in these species during the previous interpulse. Once the temperature
of the TP increases, so do the abundances. The duration of the flash-driven
convective zone for the 4M model is ≈ 26 years, and it covers a mass of
approximately 6 × 10−3 M . The third dredge-up event that follows each of
these pulses will enrich the surface, shown in Figure 19.
The He-intershell of lower mass AGB stars will only reach 300×106 K
during the last few TPs, hence the 22 Ne(α, n)25 Mg reaction is only marginally
activated near the end of the AGB. This reaction is particularly important
because it produces free neutrons that can be captured by iron-peak elements
during the s-process. It is the dominant neutron-producing reaction in the
He and C-burning shells of massive stars [119, 120, 121], and the dominant
neutron source in massive AGB stars [122].

5.3 Comparison with observations: Intershell abundances

The PG1159 class of post-AGB stars are hot, helium-rich objects that show
intershell matter of the preceding AGB phase on their surface. For this reason,
their chemical analyses allow for a direct insight into nucleosynthesis processes
during the AGB phase [123,124]. PG1159 stars are quite rare, with only about
 Nucleosynthesis of Low and Intermediate-mass Stars 139

Fig. 21. The intershell abundances 22 Ne (plus signs), 25 Mg (open circles) and 26 Mg
(closed circles) as a function of pulse number for the 4M , Z = 0.004 model. In this
diagram we show the intershell abundances during the 15th to the 20th pulse, but
only during the time when the convective shell is present; the x-axis is the (scaled)
duration of the convective pocket

two dozen known, and their atmospheres are mostly composed of the products
of partial He-burning (helium, carbon and oxygen) [125, 124]. Spectroscopic
observations of PG 1159 stars reveal oxygen mass fractions as high as 20% and
carbon as high as 40% [125], at odds with standard stellar models that pre-
dict ≈ 1% and 25% O and C, respectively. Spectroscopic observations of neon
reveal mass fractions of about 2 to 5% [125, 126], consistent with the mod-
els. Werner, Rauch, & Kruk [127] also find strong F overabundances (∼250
times solar) in a number of PG1159 stars, consistent with the F intershell
abundances of Lugaro et al. [128]. Furthermore, the high F abundances in
PG1159 stars confirm the conclusion from abundance determinations in red
giants that F is synthesized in AGB stars.
To reproduce the large observed oxygen abundances, extra-mixing pro-
cesses are required to bring 16 O from the C-O core into the intershell region.
The diffusive convective overshoot models of Herwig [111] have intershell
abundances that are consistent with the abundance patterns observed in PG
1159 central stars. The degenerate thermal pulses found by Frost, Lattanzio,
& Wood [129] may also have a similar effect. It is unclear however, if this
overshooting occurs for all AGB stars or only for AGB stars that produce
PG 1159 stars as a consequence of late and very late TPs [2]. For example,
the observed abundance analyses in intrinsic and extrinsic AGB stars, as well
as in SiC grains, suggest that such an overshoot is uncommon [71].
140 Amanda I. Karakas

5.4 Fluorine production in AGB stars

Fig. 22. The composition of the He-intershell for a 3M , Z = 0.02 model. The
upper panel shows the composition during a thermal pulse, where the flash-driven
convective pocket is shown (at its full extent it reaches from 0.664M almost all
the way to the H-shell at 0.679M ). The lower panel shows the composition of the
He-intershell after the thermal pulse has died down, before the next third dredge-up
takes place. The location of the base of the outer convective envelope is ≈ 0.681M
 Nucleosynthesis of Low and Intermediate-mass Stars 141

The fragile element fluorine can be produced inside the He-intershell of

AGB stars. Jorissen, Smith & Lambert [130] discovered that the [F/O] abun-
dance correlates with the C/O ratio in AGB stars, and that some N-type car-
bon stars show surface enrichments a factor of 30 above the solar ratio. The
increase in the C/O ratio is clearly a result of the third dredge-up and thermal
pulses, hence it was concluded that the carbon and fluorine are produced in
the same region in the star and mixed together to the surface. Jorissen et al.
[130] examined the many pathways that fluorine, or more precisely the isotope
19
F, could be produced and concluded that the most likely chain is
14
N(α, γ)18 F(β + ν)18 O(p, α)15 N(α, γ)19 F. (4)

Fluorine production takes place in the He-intershell, a region that is essentially

devoid of protons, and has a very low abundance of 13 C and 15 N. Hence other
reactions are required to produce the protons and the 15 N. These other reac-
tions include 13 C(α, n)16 O that is required to produce free neutrons; these are
subsequently captured by 14 N(n, p)14 C to produce free protons. In Figure 22
we show the composition of the He-intershell during and after a thermal pulse.
The increase in 15 N and 18 O during the pulse is evident; both of these species
are subsequently destroyed and are absent in the lower panel. The resulting
19
F abundance in the He-intershell is more than an order of magnitude greater
than in the envelope. Fluorine can also be destroyed via 19 F(α, p)22 Ne, which
is more efficient at temperatures over ≈ 300 × 106 K [128, 131]. In Figure 23
we show the surface abundance evolution of 19 F for the 3M , Z = 0.02, not-
ing the large increase over the TP-AGB through the repeated action of third
dredge-up mixing events. The fluorine abundance increases by about a factor
of four.
An example of the many reaction rates that affect the production of flu-
orine is the alternative proton production reaction 18 F(α, p)21 Ne [132]. In-
cluding the 18 F(α, p)21 Ne reaction reduces the abundance of 18 O because it
competes with 18 O production via the 18 F(β + ν)18 O decay (the half life of 18 F
is 109 minutes). However, the extra amount of protons from (α, p) enhances
the 18 O(p, α)15 N reaction rate, even though 18 O production has been deprived
from the decay. This is shown in Figure 24 for the 3M , Z = 0.008 model,
where we also show an example of the effect of reaction rate uncertainties on
the nucleosynthesis predictions. We refer the reader to [115], [133], [128], and
[132] for more details on the complex production of fluorine in AGB stars.
The cosmic origin of fluorine is not yet completely understood, where
Type II SN explosions [134] and stellar winds from Wolf-Rayet stars [135]
play a significant role in producing this fragile element alongside AGB stars
[136]. Certainly AGB stars and their progeny (e.g., post-AGB stars and plan-
etary nebulae) are the only confirmed site of fluorine production thus far
[130, 127, 137, 138], with no clear indication for enhanced F abundances re-
sulting from the ν-process in a region shaped by past SNe [139]. The recent
observations of a greatly enhanced fluorine abundance ([F/Fe] = 2.90) in a
142 Amanda I. Karakas

Fig. 23. The abundance evolution of 19 F at the surface for the 3M , Z = 0.02
model during the TP-AGB. The 19 F abundance increases by about a factor of 4

carbon-enhanced metal-poor halo star [140] represents further strong moti-

vation to better understand the details of fluorine production in AGB stars
[141].

5.5 Extra-mixing process on the AGB

There is evidence that some type of extra mixing process operates between
the H-shell and the base of the envelope in low-mass AGB stars. This mixing
is sometimes referred to as cool bottom processing. Matter from the base of
the envelope is slowly mixed down to a hot region near the H-shell, with the
result that the surface abundances of the star are slowly altered. The main
surface abundance changes are a decrease in the 12 C/13 C ratio and an increase
in the nitrogen abundance. This is qualitatively similar to the extra mixing
required on the first giant branch. The idea that extra-mixing processes are
operating in low-mass AGB stars is supported by the lower than predicted
12
C/13 C ratios and enrichments in Li and N in some carbon-rich AGB stars
[142]. The C, O, and Al isotopic ratios found in meteoritic silicon carbide
and oxide grains from AGB stars also point toward some sort of extra-mixing
process operating in the parent stars [46, 7, 143, 144, 145]. In these cases, the
mixing should not be so efficient to prevent the formation of a carbon-rich
atmosphere. We refer to Busso et al. [7] and Nollett, Busso, & Wasserburg
[143] for more details.
 Nucleosynthesis of Low and Intermediate-mass Stars 143

Fig. 24. Comparison of fluorine abundances observed by [130] and model predictions
for the 3M , Z = 0.008 model. The predictions are normalized such that the initial
19
F abundance corresponds to the average F abundance observed in K and M stars.
Each symbol on the prediction lines represents a TDU episode. Solid lines represent
calculations performed using no 18 F(α, p)21 Ne reaction, which are equivalent to using
the current lower limit, recommended value and Brussels library rate. Dotted lines
are calculations performed using the current upper limit of the rate. Figure provided
by Maria Lugaro

5.6 Hot bottom burning

In intermediate-mass AGB stars over about 4M (depending on Z), the base
of the convective envelope can dip into the top of the H-burning shell, causing
proton-capture nucleosynthesis to occur there (Figure 25). This phenomenon
is known as hot bottom burning, and can change the surface composition be-
cause the entire envelope is exposed to the hot burning region a few thousand
times per interpulse period. For a review of HBB nucleosynthesis see Lattanzio
et al. [146]. If the base of the envelope is sufficiently hot, the Ne-Na and Mg-
Al chains (Figure 17) may operate alongside the CNO cycle; 7 Li production
is also possible via the Cameron-Fowler mechanism [63, 146]. HBB converts
the 12 C dredged into the envelope to 14 N, which can prevent the C/O ratio
exceeding unity, while keeping the 12 C/13 C ratio near the equilibrium value
∼ 4. Frost et al. [147] noted that intermediate-mass AGB stars may become
144 Amanda I. Karakas

Fig. 25. The temperature at the base of the envelope during the TP-AGB phase
for a 6M , Z = 0.02 model

luminous, optically obscured carbon stars near the end of the TP-AGB, when
mass loss has removed much of the envelope, extinguishing HBB but allowing
dredge-up to continue. HBB stars may be important for the production of
many elements including nitrogen, lithium, sodium, as well as the magnesium
and aluminium isotopes.

5.7 The production of lithium by HBB

The observations of Wood, Bessell, & Fox [4] were the first to suggest that
the oxygen-rich luminous AGB stars in the Magellanic Clouds are undergoing
CNO cycling at the base of the convective envelope, converting the dredged up
carbon to nitrogen. The discovery that these stars are also rich in Li [62,63,64]
gave further credibility to the idea that HBB was actually occurring in massive
AGB stars. The production of 7 Li is thought to occur via the Cameron-Fowler
mechanism [148]: Some 3 He, created earlier in the evolution (during central H-
burning), captures an α-particle to create 7 Be. The 7 Be can either 1) capture
a proton to complete the PP III chain, or 2) capture an electron to produce
7
Li. Whether the 7 Be follows path 1) or path 2) depends critically on the
temperature of the region. Owing to efficient mixing in the convective envelope
(where the convective turnover time is ≈ 1 year), some of the 7 Be is mixed into
a cooler region which prevents proton capture. The 7 Be will undergo electron-
capture instead, producing 7 Li. The 7 Li is also subject to proton capture
and is eventually mixed into the hot temperature region and subsequently
destroyed. Once the envelope is depleted in 3 He, 7 Li production stops. The
Li-rich regime lasts for ∼100,000 years for the 6M , Z = 0.02 model shown
in Figure 26. Time-dependent mixing is required to produce 7 Li in a HBB
 Nucleosynthesis of Low and Intermediate-mass Stars 145

Fig. 26. The surface abundance of 7 Li during the TP-AGB phase for a 6M ,
Z = 0.02 model

calculation because the nuclear timescale for the reactions involved in the
Cameron-Fowler mechanism are similar to the convective turnover timescale
(see Fig. 2 in [149]).
It is still an open question whether or not AGB stars contribute to the
production of 7 Li in the Galaxy [150, 151]. There are many uncertainties in-
volved in the production of 7 Li in AGB models, including the mass-loss rates
used and the treatment of convective mixing [152]. Mass-loss rates for AGB
stars, such as the formula given by Vassiliadis & Wood [106] and Blöcker
[107], have a superwind phase which occurs during the final few thermal
pulses. The superwind phase results in a period of rapid mass loss, and most
of the convective envelope is lost during this time. Thus the composition of
the envelope at the start of the superwind phase critically determines the
contribution AGB stars make to the enrichment of the interstellar medium.
In Figure 26 most of the 7 Li has been destroyed by the time the superwind
phase starts.

5.8 HBB and the C, N, and O isotopes

Hot bottom burning first alters the envelope abundance of the CNO isotopes
by the CN cycle and later, by the ON cycles. There is also a fourth cycle
involving the destruction of 19 F to produce 16 O. The CN cycle burns 12 C
first into 13 C and later into 14 N, which reduces the 12 C/13 C ratio from the
pre-AGB value of ≈ 20 to close to 4 − 5, see Figure 27. The 14 N abundance
146 Amanda I. Karakas

increases from CN cycling and the rare nitrogen isotope, 15 N, is destroyed

by proton capture, resulting in large increases in the nitrogen isotope ratio.
HBB quickly destroys the fragile 19 F in the envelope by proton captures via
19
F(p, α)16 O. The nitrogen enrichment expected from HBB has been observa-
tionally confirmed by McSaveney et al. [153] for stars in the LMC and SMC.
Intermediate-mass AGB stars with HBB are considered to be one of the
major producers of 14 N in the Galaxy [154]. The 14 N produced by HBB has
a primary and secondary component, depending on whether the carbon and
oxygen used in the CNO reactions is produced by the He-burning shell (indi-
cating primary production), or was initially present when the star first formed
(indicating secondary production). Primary 14 N is produced when the TDU
mixes primary 12 C from the He-shell into the envelope of an AGB star with
HBB. Standard models of Type II supernovae cannot account for the primary
nitrogen component in the Galaxy [155, 156, 157], which make low-metallicity
intermediate mass AGB stars promising candidates. Low-metallicity, rapidly
rotating massive stars may also supply a significant fraction of the nitrogen
in the early Universe [158, 159]. If the principal source of primary nitrogen
is intermediate-mass stars, then there should be a lag between the release
of nitrogen into the interstellar medium and the release of oxygen, which is
produced by Type II supernovae and released very soon after a period of star
formation [160].
There are two burning cycles that involve the oxygen isotopes and these
follow the same pattern as described for the H-shell. Both operate simulta-
neously but what effect each has on the envelope abundance depends on the
branching ratio between the two 17 O + p reaction rates. Overall, there is a
net decrease in the 16 O and 18 O abundances, along with an increase in the
17
O abundance in the envelope. The decrease in the surface 16 O abundance
is only significant in the most massive AGB models, depending on the con-
vective model employed in the calculation [152, 104]. Hence the elemental
oxygen abundances observed in AGB stars should be indicative of their initial
abundances, except in the most massive objects.
When the mass of the envelope has been reduced below ≈ 1.5M , HBB
ceases but the continuation of dredge-up can turn the star into an obscured
carbon star, depending on the initial mass and metallicity [147, 118]. Other
helium burning products will also be mixed into the envelope, and in Figure 27
we see that while the 19 F abundance initially decreases, it increases again after
the cessation of HBB. The final surface abundance of fluorine is similar to the
initial. Hence, the composition of stars that are experiencing HBB depends
crucially on the evolutionary state of the star, as well as on the mass. If near
the beginning of the TP-AGB, the star is predicted to have a low 12 C/13 C
ratio (≈ 5) and low 19 F abundance, and it would be enriched in nitrogen and
lithium. Conversely, near the end of the TP-AGB that same star would be
Li-poor and nitrogen rich, and it would have a 12 C/13 C ratio near ∼ 10.
 Nucleosynthesis of Low and Intermediate-mass Stars 147

Fig. 27. The surface abundances evolution of C, N, and F during the TP-AGB
phase for the 6M , Z = 0.02 model. The final 12 C/13 C and C/O ratios are ∼ 10
∼ 0.4, respectively
148 Amanda I. Karakas

5.9 HBB and the Ne, Mg, and Al isotopes

When the temperature is high enough for the Ne-Na and Mg-Al chains to
operate1 , both 23 Na and 26 Al are produced at the expense of 22 Ne and 25 Mg.
HBB is responsible for the largest increase in the surface abundance of 23 Na,
25
Mg, and 26 Al, and is the only nucleosynthesis site in intermediate-mass
stars capable of altering the 24 Mg abundance. The 26 Al production can be
substantial with 26 Al/27 Al ratios of the order of a few ×10−1 predicted in the
most massive models.
The Ne-Na and Mg-Al chains follow the same sequence as seen in the
H-shell, except that temperatures of at least ∼90 million K are required before
24
Mg is substantially depleted. In Figure 28 we show the time variation of
various surface abundances for the 6M , Z = 0.02 and Z = 0.004 models.
Significant enrichments of 22 Ne, 25 Mg, and 26 Mg are seen in both models,
although the behaviour exhibited by the lower metallicity Z = 0.004 model
is more extreme. This is because the temperature reached at the base of the
envelope is 98 × 106 K (compared to 82 million for the Z = 0.02 model),
and the lower metallicity model experiences 100 thermal pulses with efficient
mixing. The final 26 Al/27 Al ratios are 0.015 and 0.6, respectively.
The model calculations shown in Figure 28 are subject to severe reaction
rate uncertainties (along with uncertainties that affect the structure of the
model, such as convection and mass loss). In Figure 29 we show results of
calculations using the NACRE [116] rates for the Ne-Na and Mg-Al chains
compared to the rates from [109] used in Figure 28. The main difference
is that the models using the NACRE rates produce significant amounts of
sodium at the expense of 22 Ne. Izzard et al. [110] examined the effect of
reaction rate uncertainties on the operation of the Ne-Na and Mg-Al chains
during HBB in intermediate-mass AGB stars of various compositions. The
main conclusion is that more accurate determinations of the following reaction
rates: 22 Ne(p, γ)23 Na, 23 Na(p, γ)24 Mg, 25 Mg(p, γ)26 Al, and 26 Mg(p, γ)27 Si, are
required to obtain reliable estimates of the Ne, Na, Mg, and Al yields from
AGB stars.

5.10 Yields from AGB stars

To study the chemical evolution of galaxies and stellar systems, it is essential
to have an estimate of what elements are produced (or destroyed) by stars of
different mass ranges. Stellar yields provide this information, by quantifying
the amount (in mass) of species i that is expelled into the interstellar medium
over the course of a star’s life. The definition of the yield that we use is given
by the following expression
 τ
dM
Mi = [X(i) − X0 (i)] dt, (5)
0 dt
1
higher for HBB than in the H-shell, because the density is lower at the base of
the convective envelope
 Nucleosynthesis of Low and Intermediate-mass Stars 149

Fig. 28. The surface abundances evolution of 22 Ne, 23 Na, and the neutron-rich Mg
isotopes during the TP-AGB for the 6M , Z = 0.02 model (top panel), and for the
6M , Z = 0.004 model (bottom panel)
150 Amanda I. Karakas

Fig. 29. Same as Figure 28, but using the NACRE rates for the Ne-Na and Mg-Al
chains
 Nucleosynthesis of Low and Intermediate-mass Stars 151

where Mi is the yield of species i (in solar masses), dM/dt is the current
mass-loss rate, X(i) and X0 (i) refer to the current and initial mass fraction
of species i, and τ is the total lifetime of the stellar model. The yield can
be negative, in the case where the element is destroyed, and positive if it is
produced.
Equation (5) can be re-written as
 τ  τ
dM dM
Mi = X(i) dt − X0 (i) dt. (6)
0 dt 0 dt

The first integral gives the total amount of species i expelled into the inter-
stellar medium:  τ
dM
Mitot = X(i) dt, (7)
0 dt
where Mitot is in solar masses, and is always positive.
Yields from AGB stars are provided by a number of authors. Many AGB
yields available in the literature are from synthetic AGB computations that
use fitting formula to estimate the evolution during the TP-AGB. Synthetic
AGB models have successfully been used to model AGB populations [76], and
compute stellar yields [161, 162, 163, 79]. The yields published by Forestini &
Charbonnel [112] employed a combination of full, or detailed, AGB models and
synthetic. The AGB yields of [164], [103], [105], and [24] were computed from
detailed AGB model computations. An example of the yields from Karakas &
Lattanzio [24] are shown in Figure 30. The yields have been weighted by the
initial mass function of Kroupa, Tout, & Gilmore [165], and we show results
from the Z = 0.02 (solar), 0.008 (LMC) and 0.004 (SMC) metallicity models.
For comparison we also show the yields from a number of different synthetic
AGB calculations, and from [164]; see the figure caption for details.
Many of the parameterizations used in synthetic evolution studies are de-
rived from detailed stellar models, such as the growth of the H-exhausted core
with time, and as such are only accurate over the range in mass and metallic-
ity of the stellar models they are based upon. An example is provided by van
den Hoek & Groenewegen [162] who computed AGB yields for initial masses
between 0.9 and 8M whereas the interpulse-period-core mass relation [70]
they use was only derived for stars with initial masses between 1 and 3M .
What affect this has on the yields is unclear since this relation will affect the
number of TPs during the TP-AGB phase and hence the level of chemical en-
richment. Recent improvements in computer power mean that grids of detailed
AGB models can be produced in a reasonable time [114, 118, 103]; however,
producing yields from more than ∼ 20 AGB stars for any given metallicity
range is still challenging. For this reason synthetic models are still preferred
for some applications e.g., [110].
152 Amanda I. Karakas

Fig. 30. Weighted yield of 12 C as a function of the initial mass for the Z = 0.02
(top), the Z = 0.008 (middle) and the Z = 0.004 models (bottom). We show re-
sults from [24](black solid points), [162] (open magenta squares), [112] (solid green
squares), [163] (open red circles), [164] (solid aqua triangles), and [79] (blue crosses)
 Nucleosynthesis of Low and Intermediate-mass Stars 153

6 The slow neutron-capture process

For heavy elements with atomic masses greater than ≈ 56 their proton num-
bers are so large that electrostatic repulsion inhibits charged particle reactions
(e.g., proton and α captures) except under very specific conditions. Most
heavy nuclei are instead formed by neutron addition onto abundant Fe-peak
elements. For this reason, neutron-capture processes are secondary. That is,
they require that some Fe-peak nuclei (e.g., 56 Fe) is already present in the
star. The solar abundance distribution is characterized by peaks that can be
explained by
1. the slow-neutron-capture process, the s-process,
2. the rapid-neutron-capture process, the r-process,
3. proton-rich nuclei are much less abundant in the solar system than
nuclei synthesized by the s- and r-processes.
The seminal paper by Burbidge, Burbidge, Fowler, & Hoyle [166] (hereafter
B2 FH) laid down the foundations for these processes, and Wallerstein et al.
[167] provides an updated review on B2 FH. An historical overview and details
of how the s-process operates in AGB stars has been reviewed by Meyer [168],
Busso et al. [7], and Herwig [1]. The textbooks by Clayton [9] and Lugaro
[15] provide excellent discussions of the s-process at a level aimed at graduate
students.
We will concern ourselves with the s-process, which occurs under condi-
tions of relatively low neutron densities (Nn ∼ 107 neutrons/cm3 ). In this
case the timescale for neutron-capture is much slower, in general, than the
β-decay rate of unstable isotopes. Hence, the s-process will produce isotopes
along the valley of β-stability. Examples of s-process elements include Sr,
Zr, Nb, Ba, and La, where the dominant isotopes are mainly produced by
the s-process. There are also some isotopes that are only produced by the
s-process and examples include 86 Sr, 96 Mo, 104 Pd, and 116 Sn. During the r-
process, neutron densities as high as Nn ∼ 1025 neutrons/cm3 ensures that
the timescale for neutron-capture is much faster than the β-decay rates. The
r-process will produce isotopes right up to the neutron drip line, which de-
cay to stable, neutron-rich isotopes once the neutron flux is gone. Given the
extreme conditions required for the r-process, it has been hypothesized to
occur during supernovae explosions [169, 170] but other sites have also been
proposed including colliding neutron stars [171], and black hole/neutron star
mergers [172]. We refer to the reviews by Meyer [168] and Arnould, Goriely, &
Takahashi [173] for further details.
The solar abundance distribution shows peaks that are constrained by nu-
clei with a magic number of neutrons (n = 50, 80, 82, 126). Note that for lighter
elements there are also peaks at n = 2, 8, 20, and 28. A nucleus composed of
a magic number of protons and neutrons is very stable and considered to be
“doubly magic”. Examples include 16 O (with 8 protons and neutrons), and
208
Pb, with 82 protons and 126 neutrons. The stability of nuclei with a magic
154 Amanda I. Karakas

number of neutrons can be explained when nuclei are described as systems

ruled by quantum mechanics, as done for atomic systems. In practice, nu-
clei with a magic number of neutrons are very stable against neutron-capture
and have low neutron-capture cross sections. Hence these nuclei act as bottle-
necks and are consequently seen as s-process peaks. Note that 56 Ni, made in
abundance by supernovae, is doubly magic, with 28 protons and 28 neutrons.
The neutron exposure, τ , is defined by

τ = Nn vT dt, (8)

where Nn is the neutron density and vT the thermal velocity of the neutrons.
The neutron exposure is a time-averaged neutron flux with units of mbarn−1
(where 1 barn is equal to 10−24 cm2 ), see [7] for further details. There are three
distinct components of the s-process that require distinct neutron exposures:
1. The weak component: This produces most of the s-isotopes with atomic
mass ≤ 90, from Fe to Sr, and can be described by a mean neutron
exposure of τ ≈ 0.06 mbarn−1 .
2. The main component: This is responsible for the s-process isotopes from
90 ≤ A ≤ 204, from Sr to Pb, where the mean τ ≈ 0.3 mbarn−1 .
3. The strong component: Required to reproduce the solar system Pb abun-
dance, requiring τ ≈ 7.0 mbarn−1 .
The weak component is thought to originate in the He and C-burning shells
of massive stars [121], the main component in low-mass AGB stars, and the
strong component, that produces Pb, has been hypothesized to occur in very
low metallicity, low-mass AGB stars [168, 7].

6.1 Neutron sources operating in AGB stars

Neutron-capture processes require a source of free neutrons, given that neu-

trons are unstable and decay in ≈ 10 minutes. There are two important neu-
tron sources available during He-shell burning in AGB stars:
14
1. N(α, γ)18 F(β + ν) 18 O(α, γ)22 Ne(α, n)25 Mg.
12
2. C(p, γ)13 N(β + ν)13 C(α, n)16 O.
The 22 Ne(α, n)25 Mg was first identified as a neutron source for AGB stars by
[174]. Note that B2 FH suggested that the 21 Ne(α, n)24 Mg reaction might be
an important neutron source, but the scarcity of 21 Ne in the He-intershell
meant that the reaction is not efficient at providing neutrons. The inter-
shell region is rich in 14 N from CNO cycling, and during a thermal pulse,
14
N can suffer successive α captures to produce 22 Ne. If the temperature ex-
ceeds about 300×106 K, 22 Ne can capture an α particle to produce 25 Mg
and 26 Mg, in almost equal proportions (see Figure 20). Neutrons are released
by the 22 Ne(α, n)25 Mg reaction. Given the high temperatures required for
the 22 Ne(α, n)25 Mg to operate efficiently, it is theoretically predicted to be
 Nucleosynthesis of Low and Intermediate-mass Stars 155

efficient in the most massive AGB stars, with initial masses ≥ 4M . The
temperatures required for the 22 Ne(α, n)25 Mg may also be reached during the
last few thermal pulses of lower-mass AGB stars, where it has a marginal in-
fluence on the final s-process abundance distribution prior to the subsequent
third dredge-up [175].
The 13 C(α, n)16 O reaction requires the operation of both proton and
α-capture reactions to occur in the He-shell, a region normally devoid of
protons. There is some 13 C left over in the He-intershell after the end of
CNO cycling, but the study by Gallino et al. [176] has shown that this is
not enough to account for the s-process enrichments of AGB stars. Hence,
some mixing of protons from the convective envelope into the top layers of
the He-intershell is required. Straniero et al. [177] discovered that the 13 C
burns under radiative conditions because the 13 C(α, n)16 O reaction occurs
at relatively low temperatures (T ≥ 90 × 106 K). The neutrons are released
in the 13 C pocket, and the s-processing occurs between pulses in the same
layers where the 13 C was produced. When the next convective thermal pulse
occurs, it ingests this s-element rich layer, mixing it over most of the in-
tershell. Observational [81, 178, 84] and theoretical evidence [176, 179, 175]
suggests that the 13 C(α, n)16 O reaction is the dominant neutron source in
low-mass AGB stars, with initial masses ≤ 3M . The timescales for neutron
production during the interpulse are much longer (≥ 103 years) than during
the convective pulse (∼ 10 years), resulting in much lower neutron densities
(∼ 108 neutrons/cm3 ) than the 22 Ne source (∼ 1013 neutrons/cm3 ). Together,
the timescale for neutron production and the neutron source determine the
resulting s-process element distribution.

13
6.2 Partial mixing and the formation of C pockets

For the 13 C(α, n)16 O reaction to occur efficiently, some partial mixing is re-
quired at the border between the H-rich envelope and the C-rich intershell.
Protons are mixed down into the top 1/10th of the 4 He and 12 C-rich region,
where the CN cycle is able to partially operate to produce a 13 C pocket (along
with a 14 N pocket). In Figure 31 we show an example of calculations performed
with the inclusion of a partially-mixed zone, where the resulting 13 C pocket
covers ≈ 10% of the He-intershell. Note the large increase in Ge isotopes in the
model with a partially mixed zone, and the lack of any s-process enrichments
in the model with no pocket. It is important that there not be too many pro-
tons, because in this case the CN cycle goes to completion and 14 N is mostly
synthesized. Both 14 N and 26 Al are neutron poisons, which means that they
are efficient neutron absorbers and will change the resulting abundance dis-
tribution. The details of how the 13 C pocket forms and its extent in mass in
the He-intershell are still unknown, although various mechanisms have been
proposed including convective overshoot, rotation, and gravity waves (see [1]
for a discussion of the relative merits of each mechanism). Models that include
artificial 13 C pockets produce s-process abundance distributions that fit the
156 Amanda I. Karakas

Fig. 31. Composition profile showing the intershell abundances just before the last
computed thermal pulse. The shaded region is the inner edge of the convective
envelope. We show abundances from the 3M , Z = 0.012 model with a partially
mixed zone (a), and without (b). Using models from [180]
 Nucleosynthesis of Low and Intermediate-mass Stars 157

observational data reasonably well, with a small spread in the pocket sizes
required [181, 182].

6.3 The s-process in massive AGB stars

Fig. 32. Section of the chart of the nuclides from Kr to Zr. Stable isotopes are
depicted with dark backgrounds and white labels. Unstable isotopes are depicted
on a lighter background with black labels. The light arrow shows the neutron flow
for low-neutron densities, whereas the dark, dashed arrows show the flow for high
neutron densities (Nn ≥ 108 n/cm−3 ). The black solid arrows shows the direction of
the flow for all densities. Isotope with magic neutron number = 50 are surrounded
by the open rectangle. Figure provided by Mark van Raai

Luminous, O-rich AGB stars in the LMC and SMC exhibit strong molec-
ular bands of ZrO, indicating that the atmospheres of these stars are enriched
in the s-process element Zr [4]. Recent observations by Garcı́a-Hernández
et al. [122] showed that luminous OH/IR stars in the Galaxy have enhanced
rubidium abundances. This evidence suggest that massive AGB stars do pro-
duce s-process elements and that the neutron source operating in these stars
is the 22 Ne(α, n)25 Mg reaction. The contribution of these stars to the total
Galactic inventory of s-process elements is still expected to be small [183].
Travaglio et al. [183] estimate that intermediate-mass stars contribute ≈ 8, 6,
6, 1, and 5% toward the solar-system composition of Sr, Y, Zr, Nb, and Mo,
respectively.
158 Amanda I. Karakas

Rubidium production is an indicator of high neutron densities and the

operation of the 22 Ne(α, n)25 Mg reaction. This is because at high neutron
densities two branching points open that allow Rb to be synthesized (Fig-
ure 32). At Nn = 5 × 108 n/cm3 , ∼ 86% of the neutron flux goes through
85
Kr allowing for 85 Kr(n, γ)86 Kr(n, γ)87 Kr that decays to 87 Rb. Note that
87
Rb has a magic number of neutrons and is fairly stable against neutron-
capture. Also, high neutron densities allow neutrons to bypass the branching
point at 86 Rb (in low-mass stars this isotope decays to 86 Sr), allowing for the
chain 86 Rb(n, γ)87 Rb. In low-mass stars, the magic 88 Sr is produced instead.
For this reason, the elemental ratios of Rb/Sr and Rb/Zr are indicators of the
neutron densities, and was used as evidence that the 13 C(α, n)16 O reaction is
the major neutron source in low-mass AGB stars [178].

7 Concluding remarks
The AGB phase is the last nuclear burning phase for stars with initial masses
between about 0.8M to 8M . The AGB phase of evolution is very short,
comprising less than 1 per cent of the main-sequence lifetime, nevertheless, it
is on the AGB that the richest nucleosynthesis occurs for this mass range. The
nucleosynthesis is driven by thermal instabilities of the helium-burning shell,
the products of which are dredged to the stellar surface by recurrent mixing
episodes. Hot bottom burning occurs in the most massive AGB stars, and
this also alters the surface composition. AGB stars are important factories
for producing many elements including carbon, nitrogen, fluorine, and heavy
elements synthesized by the s-process. It is now estimated that perhaps up to
half of all elements heavier than iron are made by the s-process in low-mass
AGB stars [7]. While much is known about the inner workings of AGB stars,
there are many unknowns that render predictions uncertain. In particular, our
lack of knowledge about convective mixing processes in stars is perhaps the
greatest uncertainty, since it overshadows so much of AGB evolution. Other
modelling uncertainties such as mass-loss rates, opacities, and reaction rate
uncertainties are also important.
In these lecture notes, we have reviewed the structure and evolution of
low and intermediate-mass stars prior to the AGB phase, along with the sur-
face abundance changes predicted by models for the first and second dredge-
up mixing events. We reviewed the evolution and nucleosynthesis during the
AGB, including a brief discussion of the slow neutron-capture process. There
are many difficult problems left unsolved including the formation of 13 C pock-
ets, and the extent of extra-mixing processes during the first and asymptotic
giant branch phases. The process driving mass loss in AGB stars also poses a
severe challenge for the future. It will be important to find techniques to accu-
rately measure many of the reaction rates of interest for AGB nucleosynthesis.
It is also essential that we understand nucleosynthesis in low-metallicity AGB
stars if we are to unravel the puzzle surrounding the abundances of many
 Nucleosynthesis of Low and Intermediate-mass Stars 159

carbon-rich metal-poor stars in the Galaxy. The mass-transfer processes that

occur in binary star systems are poorly understood but these need to be un-
tangled. Finally, AGB stars are a fascinating, challenging phase of evolution
to study with much progress yet to be made.

Acknowledgments
AIK warmly thanks the organisers of the Kodai School on Synthesis of the
Elements in Stars for the hospitality and the opportunity to attend and give
lectures at the school. AIK also thanks John Lattanzio, Maria Lugaro, Simon
Campbell, Peter Wood, Pilar Gil-Pons, Lionel Siess, and Mark van Raai for
discussions and/or material that helped in the preparation of the lecture notes
given at the School. She thanks Robin Humble, Ross Church, and Maria Lu-
garo for help with proof-reading her lecture notes.

References
1. F. Herwig, ARA&A 43, 435 (2005)
2. T. Blöcker, Ap&SS 275, 1 (2001)
3. H. van Winckel, ARA&A 41 391 (2003)
4. P. R. Wood, M. S. Bessell, M. W. Fox, ApJ 272, 99 (1983)
5. D. A. VandenBerg, F. J. Swenson, F. J. Rogers, C. A. Iglesias, D. R.
Alexander, ApJ 532, 430 (2000)
6. R. Buonanno, C. E. Corsi, A. Buzzoni, C. Cacciari, F. R. Ferraro, F. Fusi
Pecci, A&A 290, 69 (1994)
7. M. Busso, R. Gallino, G. J. Wasserburg, ARA&A 37, 239 (1999)
8. L. M. Dray, C. A. Tout, A. I. Karakas, J. C. Lattanzio, MNRAS 338,
973 (2003)
9. D. D. Clayton in Principles of stellar evolution and nucleosynthesis, Uni-
versity of Chicago Press (1983)
10. C. E. Rolfs, W. S. Rodney in Cauldrons in the cosmos: Nuclear astro-
physics, University of Chicago Press, (1988)
11. C. Iliadis in Nuclear Physics of Stars, Wiley-VCH Verlag (2007)
12. C. J. Hansen, S. D. Kawaler, V. Trimble in Stellar interiors : physical
principles, structure, and evolution, Springer-Verlag, (2004).
13. B. E. J. Pagel in Nucleosynthesis and Chemical Evolution of Galaxies,
Cambridge University Press (1997)
14. H. J. Habing, H. Olofsson in Asymptotic Giant Branch Stars, Springer
(2004)
15. M. Lugaro in Stardust from meteorites. An introduction to presolar grains,
World Scientific (2005)
16. S. E. Woosley, A. Heger, T. A. Weaver, Rev. Mod. Phys. 74, 1015 (2002)
17. E. Anders, N. Grevesse, Geochim. Cosmochim. Acta 53, 197 (1989)
18. M. Asplund, N. Grevesse, A. J. Sauval, ASPS 336, 25 (2005)
160 Amanda I. Karakas

19. C. Allende Prieto, D. L. Lambert, M. Asplund, ApJ 556, L63 (2001)

20. T. C. Beers, N. Christlieb, ARA&A 43, 531 (2005)
21. A. Frebel, W. Aoki, N. Christlieb, H. Ando, M. Asplund, P. S. Barklem
et al., Nature 434, 871 (2005)
22. E. E. Salpeter, ApJ, 121, (1955)
23. P. Kroupa, Science 295, 82 (2002)
24. A. I. Karakas, J. C. Lattanzio, PASA 24, 103 (2007)
25. G. C. Sloan, K. E. Kraemer, P. R. Wood, A. A. Zijlstra, J. Bernard-Salas,
D. Devost et al., ApJ 686, 1056 (2008)
26. G. Marconi, R. Buonanno, E. Carretta, F. R. Ferraro, P. Montegriffo,
F. Fusi Pecci et al., MNRAS 293, 479 (1998)
27. A. I. Karakas in Asymptotic Giant Branch Stars: their influence on bi-
nary systems and the interstellar medium, PhD thesis, Monash University
(2003)
28. F. Fagotto, A. Bressan, G. Bertelli, C. Chiosi, A&AS 104, 365 (1994)
29. C. Frost, J. Lattanzio, ArXiv Astrophysics e-prints, astro-ph/9601017
(1996)
30. J. C. Lattanzio A. I. Boothroyd, American Institute of Physics Conference
Series 402, ed. by T. J. Bernatowicz, E. Zinner, 85 (1997)
31. C. Charbonnel, A&A 282, 811 (1994)
32. A. I. Boothroyd, I.-J. Sackmann, ApJ 510, 232 (1999)
33. H. Schlattl, S. Cassisi, M. Salaris, A. Weiss, ApJ 559, 1082 (2001)
34. I. Picardi, A. Chieffi, M. Limongi, O. Pisanti, G. Miele, G. Mangano,
G. Imbriani, ApJ 609, 1035 (2004)
35. T. Suda, M. Y. Fujimoto, N. Itoh, ApJ 667, 1206 (2007)
36. A. I. Karakas, J. C. Lattanzio, O. R. Pols, PASA 19, 515 (2002)
37. G. Bertelli, A. Bressan, C. Chiosi, K. Angerer, A&AS 66, 191 (1986)
38. V. Castellani, P. Giannone, A. Renzini, A&AS 10, 355 (1971)
39. V. Trimble, PASP 104, 1 (1992)
40. C.-W. Chin, R. B. Stothers, ApJ 77, 299 (1991)
41. V. Castellani, S. Degl’Innocenti, M. Marconi, P. G. Prada Moroni,
P. Sestito, A&A 404, 645 (2003)
42. P. Ventura, M. Castellani, C. W. Straka, A&A 440, 623 (2005)
43. G. Imbriani, M. Limongi, L. Gialanella, F. Terrasi, O. Straniero,
A. Chieffi, ApJ 558, 903 (2001)
44. M. Arnould, S. Goriely, A. Jorissen, A&A 347, 572 (1999)
45. M. F. El Eid, A&A 285, 915 (1994)
46. G. J. Wasserburg, A. I. Boothroyd, I.-J. Sackmann, ApJ 447, L37 (1995)
47. K. K. Gilroy, ApJ 347, 835 (1989)
48. M. D. Shetrone, AJ 112, 2639 (1996)
49. C. A. Pilachowski, C. Sneden, R. P. Kraft, G. E. Langer, AJ 112, 545
(1996)
50. R. Gratton, C. Sneden, E. Carretta, ARA&A 42, 385 (2004)
51. J. G. Cohen, M. M. Briley, P. B. Stetson, AJ 130, 1177 (2005)
52. A. V. Sweigart, J. G. Mengel, ApJ 229, 624 (1979)
53. P. P. Eggleton, D. S. P. Dearborn, J. C. Lattanzio, Science 314, 1580
(2006)
54. P. P. Eggleton, D. S. P. Dearborn, J. C. Lattanzio, ApJ 677, 581 (2008)
55. C. Charbonnel, J.-P. Zahn, A&A 467, L15 (2007)
 Nucleosynthesis of Low and Intermediate-mass Stars 161

56. G. H. Smith, C. A. Tout, MNRAS 256, 449 (1992)

57. C. Charbonnel, ApJ 453, L41 (1995)
58. P. A. Denissenkov, C. A. Tout, MNRAS, 316, 395 (2000)
59. I. Iben, Jr. In IAU Symp. 145: Evolution of Stars: the Photospheric Abun-
dance Connection ed. by G. Michaud, A. V. Tutukov, 257 (1991)
60. P. R. Wood. In Planetary Nebulae, IAU Symposium vol. 180, ed. by
H. J. Habing, H. J. G. L. M. Lamers, 297 (1997)
61. J. M. Scalo, K. H. Despain, R. K. Ulrich, ApJ 196, 805 (1975)
62. V. V. Smith, D. L. Lambert, ApJ 345, L75 (1989)
63. V. V. Smith, D. L. Lambert, ApJ 361, L69 (1990)
64. B. Plez, V. V. Smith, D. L. Lambert, ApJ, 418, 812 (1993)
65. T. Bloecker, D. Schoenberner, A&A 244, L43 (1991)
66. J. C. Lattanzio, PASA 10, 120 (1992)
67. B. Paczyński, Acta Astronomica 20, 47 (1970)
68. A. I. Boothroyd, I.-J. Sackmann, S. C. Ahern, ApJ 416, 762 (1993)
69. O. Straniero, A. Chieffi, M. Limongi, M. Busso, R. Gallino, C. Arlandini,
ApJ 478, 332 (1997)
70. A. I. Boothroyd, I.-J. Sackmann, ApJ 328, 653 (1988)
71. M. Lugaro, F. Herwig, J. C. Lattanzio, R. Gallino, O. Straniero, ApJ
586, 1305 (2003)
72. R. J. Stancliffe, C. A. Tout, O. R. Pols, MNRAS 352, 984 (2004)
73. G. Wallerstein, G. R. Knapp, ARA&A 36, 369 (1998)
74. J. A. Frogel, J. Mould, V. M. Blanco, ApJ 352, 96 (1990)
75. M. A. T. Groenewegen, A&A 425, 595 (2004)
76. M. A. T. Groenewegen, T. de Jong, A&A 267, 410 (1993)
77. P. Marigo, A. Bressan, C. Chiosi, A&A 313, 545 (1996)
78. N. Mowlavi, A&A 344, 617 (1999)
79. R. G. Izzard, C. A. Tout, A. I. Karakas, O. R. Pols, MNRAS 350, 407
(2004)
80. R. J. Stancliffe, R. G. Izzard, C. A. Tout, MNRAS 356, L1 (2005)
81. V. V. Smith, D. L. Lambert, ApJ 311, 843 (1986)
82. V. V. Smith, D. L. Lambert, A. McWilliam, ApJ 320, 862 (1987)
83. V. V. Smith, D. L. Lambert, ApJS 72, 387 (1990)
84. C. Abia, I. Domı́nguez, R. Gallino, M. Busso, S. Masera, O. Straniero
et al., ApJ, 579, 817 (2002)
85. I. R. Little-Marenin, S. J. Little, AJ 84, 1374 (1979)
86. S. Uttenthaler, J. Hron, T. Lebzelter, M. Busso, M. Schultheis, H. U.
Käufl, A&A 463, 251 (2007)
87. R. D. McClure, A. W. Woodsworth, ApJ 352, 709 (1990)
88. R. D. McClure, PASP 109, 536 (1997)
89. V. V. Smith, D. L. Lambert, ApJ 333, 219 (1988)
90. J. A. Brown, V. V. Smith, D. L. Lambert, E. J. Dutchover, K. H. Hinkle,
H. R. Johnson, AJ 99, 1930 (1990)
91. R. D. McClure, PASP 109, 256 (1997)
92. R. G. Izzard, C. S. Jeffery, J. Lattanzio, A&A 470, 661 (2007)
93. T. Lloyd Evans, MNRAS 220, 723 (1986)
94. C. Abia, J. Isern, ApJ 536, 438 (2000)
95. J. Herman, H. J. Habing, A&AS 59, 523 (1985)
96. T. Lebzelter, P. R. Wood, A&A 475, 643 (2007)
162 Amanda I. Karakas

97. M. W. Fox, P. R. Wood, ApJ 259, 198 (1982)

98. D. A. Ostlie, A. N. Cox, ApJ 311, 864 (1986)
99. G. H. Bowen, ApJ 329, 299 (1988)
100. J. M. Winters, T. Le Bertre, K. S. Jeong, L.-Å. Nyman, N. Epchtein,
A&A, 409, 715 (2003)
101. P. Woitke, A&A, 452, 537 (2006)
102. B. Freytag, S. Höfner, A&A 483, 571 (2008)
103. F. Herwig, ApJS 155, 651 (2004)
104. P. Ventura, F. D’Antona, A&A 439, 1075 (2005)
105. R. J. Stancliffe, C. S. Jeffery, MNRAS 375, 1280 (2007)
106. E. Vassiliadis, P. R. Wood, ApJ 413, 641 (1993)
107. T. Blöcker, A&A 297, 727 (1995)
108. N. Mowlavi, A&A 350, 73 (1999)
109. C. Iliadis, J. M. D’Auria, S. Starrfield, W. J. Thompson, M. Wiescher,
ApJS 134, 151 (2001)
110. R. G. Izzard, M. Lugaro, A. I. Karakas, C. Iliadis, M. van Raai, A&A
466, 641 (2007)
111. F. Herwig, A&A 360, 952 (2000)
112. M. Forestini, C. Charbonnel, A&AS 123, 241 (1997)
113. A. I. Karakas J. C. Lattanzio, PASA 20, 393 (2003)
114. A. I. Karakas, J. C. Lattanzio, PASA 20, 279 (2003)
115. M. Forestini, S. Goriely, A. Jorissen, M. Arnould, A&A 261, 157 (1992)
116. C. Angulo, M. Arnould, M. Rayet, P. Descouvemont, D. Baye,
C. Leclercq-Willain et al., Nucl. Phys. A 656, 3 (1999)
117. F. Kaeppeler, M. Wiescher, U. Giesen, J. Goerres, I. Baraffe, M. El Eid
et al., A. Chieffi, ApJ 437, 396 (1994)
118. A. I. Karakas, M. Lugaro, M. Wiescher, J. Goerres, C. Ugalde, ApJ 643,
471 (2006)
119. L.-S. The, M. F. El Eid, B. S. Meyer, ApJ 533, 998 (2000)
120. L.-S. The, M. F. El Eid, B. S. Meyer, ApJ 655, 1058 (2007)
121. M. Heil, F. Käppeler, E. Uberseder, R. Gallino, M. Pignatari, Phys. Rev.
C 77(1), 015808 (2008)
122. D. A. Garcı́a-Hernández, P. Garcı́a-Lario, B. Plez, F. D’Antona, A. Man-
chado, J. M. Trigo-Rodrı́guez, Science 314, 1751 (2006)
123. R. Napiwotzki, Rev. Mod. Astron. 11, 3 (1998)
124. K. Werner, Ap&SS 275, 27 (2001)
125. K. Werner, T. Rauch, A&A 284, L5 (1994)
126. K. Werner, B. Wolff, A&A 347, L9 (1999)
127. K. Werner, T. Rauch, J. W. Kruk, A&A 433, 641 (2005)
128. M. Lugaro, C. Ugalde, A. I. Karakas, J. Görres, M. Wiescher, J. C.
Lattanzio, R. C. Cannon, ApJ 615, 934 (2004)
129. C. A. Frost, J. C. Lattanzio, P. R. Wood, ApJ 500, 355 (1998)
130. A. Jorissen, V. V. Smith, D. L. Lambert, A&A, 261, 164 (1992)
131. C. Ugalde, R. E. Azuma, A. Couture, J. Görres, H. Y. Lee, E. Stech et al.,
Phys. Rev. C 77(3), 035801 (2008)
132. A. I. Karakas, H. Y. Lee, M. Lugaro, J. Görres, M. Wiescher, ApJ 676,
1254 (2008)
133. N. Mowlavi, A. Jorissen, M. Arnould, A&A 311, 803 (1996)
134. S. E. Woosley, T. A. Weaver, ApJS 101, 181 (1995)
 Nucleosynthesis of Low and Intermediate-mass Stars 163

135. G. Meynet, M. Arnould, A&A 355, 176 (2000)

136. A. Renda, Y. Fenner, B. K. Gibson, A. I. Karakas, J. C. Lattanzio,
S. Campbell, A. Chieffi et al., MNRAS 354, 575 (2004)
137. Y. Zhang, X.-W. Liu, ApJ 631, L61 (2005)
138. G. Pandey, ApJ 648, L143 (2006)
139. S. R. Federman, Y. Sheffer, D. L. Lambert, and V. V. Smith, ApJ 619,
884 (2005)
140. S. C. Schuler, K. Cunha, V. V. Smith, T. Sivarani, T. C. Beers, Y. S. Lee,
ApJ 667, L81 (2007)
141. M. Lugaro, S. E. de Mink, R. G. Izzard, S. W. Campbell, A. I. Karakas,
S. Cristallo et al., A&A 484, L27 (2008)
142. C. Abia, J. Isern, MNRAS 289, L11 (1997)
143. K. M. Nollett, M. Busso, G. J. Wasserburg, ApJ 582, 1036 (2003)
144. E. Zinner, L. R. Nittler, R. Gallino, A. I. Karakas, M. Lugaro,
O. Straniero, J. C. Lattanzio, ApJ 650, 350 (2006)
145. E. Zinner, PASA 25, 7 (2008)
146. J. Lattanzio, C. Frost, R. Cannon, P. R. Wood, Mem. Soc. Astron. Ital-
iana 67, 729 (1996)
147. C. A. Frost, R. C. Cannon, J. C. Lattanzio, P. R. Wood, M. Forestini,
A&A 332, L17 (1998)
148. A. G. W. Cameron, W. A. Fowler, ApJ 164, 111 (1971)
149. A. I. Boothroyd, I.-J. Sackmann, ApJ 393, L21 (1992)
150. D. Romano, F. Matteucci, P. Ventura, F. D’Antona, A&A 374, 646
(2001)
151. C. Travaglio, S. Randich, D. Galli, J. Lattanzio, L. M. Elliott,
M. Forestini, F. Ferrini, ApJ 559, 909 (2001)
152. P. Ventura, F. D’Antona, A&A 431, 279 (2005)
153. J. A. McSaveney, P. R. Wood, M. Scholz, J. C. Lattanzio, K. H. Hinkle,
MNRAS 378, 1089 (2007)
154. R. B. C. Henry, M. G. Edmunds, J. Köppen, ApJ 541, 660 (2000)
155. F. X. Timmes, S. E. Woosley, T. A. Weaver, ApJS 98, 617 (1995)
156. M. Samland, ApJ 496, 155 (1998)
157. A. Goswami & N. Prantzos, A&A 359, 191 (2000)
158. C. Chiappini, R. Hirschi, G. Meynet, S. Ekström, A. Maeder,
F. Matteucci, A&A 449, L27 (2006)
159. R. Hirschi, A&A 461, 571 (2007)
160. M. Pettini, S. L. Ellison, J. Bergeron, P. Petitjean, A&A 391, 21 (2002)
161. A. Renzini, M. Voli, A&A 94, 175 (1981)
162. L. B. van den Hoek, M. A. T. Groenewegen, A&AS 123, 305 (1997)
163. P. Marigo, A&A 370, 194 (2001)
164. P. Ventura, F. D’Antona, I. Mazzitelli, A&A 393, 215 (2002)
165. P. Kroupa, C. A. Tout, G. Gilmore, MNRAS 262, 545 (1993)
166. E. M. Burbidge, G. R. Burbidge, W. A. Fowler, F. Hoyle, Rev. Mod.
Phys. 29, 547 (1957)
167. G. Wallerstein, I. J. Iben, P. Parker, A. M. Boesgaard, G. M. Hale, A. E.
Champagne et al., Rev. Mod. Phy. 69, 995 (1997)
168. B. S. Meyer, ARA&A 32, 153 (1994)
169. C. L. Fryer, F. Herwig, A. Hungerford, F. X. Timmes, ApJ 646, L131
(2006)
164 Amanda I. Karakas

170. S. Wanajo, ApJ 650, L79 (2006)

171. D. Argast, M. Samland, F.-K. Thielemann, Y.-Z. Qian, A&A 416, 997
(2004)
172. R. Surman, G. C. McLaughlin, M. Ruffert, H.-T. Janka, W. R. Hix, ApJ
679, L117 (2008)
173. M. Arnould, S. Goriely, K. Takahashi, Phys. Rep. 450, 97 (2007)
174. A. G. W. Cameron, AJ 65, 485 (1960)
175. M. Busso, R. Gallino, D. L. Lambert, C. Travaglio, V. V. Smith, ApJ
557, 802 (2001)
176. R. Gallino, C. Arlandini, M. Busso, M. Lugaro, C. Travaglio, O. Straniero,
A. Chieffi, M. Limongi, ApJ 497, 388 (1998)
177. O. Straniero, R. Gallino, M. Busso, A. Chiefei, C. M. Raiteri, M. Limongi,
M. Salaris, ApJ 440, L85, (1995)
178. C. Abia, M. Busso, R. Gallino, I. Domı́nguez, O. Straniero, J. Isern, ApJ
559, 1117 (2001)
179. S. Goriely, N. Mowlavi, A&A 362, 599 (2000)
180. A. I. Karakas, M. Lugaro, and R. Gallino, ApJ 656, L73 (2007)
181. S. Goriely, L. Siesu, A&A 421, L25 (2004)
182. A. Bonačić Marinović, M. Lugaro, M. Reyniers, H. van Winckel, A&A
472, L1, (2007)
183. C. Travaglio, R. Gallino, E. Arnone, J. Cowan, F. Jordan, C. Sneden,
ApJ 601, 864 (2004)
Spectral Classification: Old and Contemporary

Sunetra Giridhar

Indian Institute of Astrophysics, Bangalore 560034, India

Summary. Beginning with a historical account of the spectral classification, its re-
finement through additional criteria is presented. The line strengths and ratios used
in two dimensional classifications of each spectral class are described. A parallel
classification scheme for metal-poor stars and the standards used for classification
are presented. The extension of spectral classification beyond M to L and T and
spectroscopic classification criteria relevant to these classes are described. Contem-
porary methods of classifications based upon different automated approaches are
introduced.

Keywords: Spectral classification, luminosity classes, metal-poor stars,

M-L-T spectral classes

1 Historical Account of Spectral Classification

In 1866, Fr Angelo Secchi a Jesuit astronomer working in Italy observed pris-

matic spectra of about 4000 stars visually and divided stars in four broad
spectral classes using common absorption features of hydrogen. During 1886-
97, Henry Draper Memorial Survey at Harvard carried out a systematic pho-
tographic spectroscopy of stars brighter than 9th magnitude covering entire
sky using telescopes at Harvard and Arequipa, Peru under the leadership of
E.C. Pickering. The Henry Draper Catalog was published in 9 volumes of the
Annals of Harvard College Observatory between 1918 and 1924. It contains
rough positions, magnitudes, spectral classifications for 225,300 stars. Earlier
work by W. Fleming essentially subdivided the previously used Secchi classes
(I to IV) into more specific classes, by giving letters from A to N. The strength
of hydrogen lines being the main classifier, the spectral type A was assigned
to stars with strongest hydrogen lines followed by B, C with weaker hydro-
gen lines. This system was found to be unsatisfactory since the line strengths
of other lines varied irregularly and so did the B−V color. This system was
improvised by A. Maury, A.J. Cannon and E. Pickering who re-arranged the

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 3, c Springer-Verlag Berlin Heidelberg 2010
166 Sunetra Giridhar

spectral sequence taking into consideration the changes in other lines and this
new spectral sequence was also a sequence according to the color of the stars.
But well-known stars had been already assigned the older spectral classes for
long hence it was not possible to change them. We, therefore, have a spectral
sequence essentially temperature dependent but goes like OBAFGKM. At the
cool end the classification becomes more complex with parallel branches of
R, N, S stars. While the M stars have TiO bands, S stars display ZrO bands,
while R and N are carbon stars showing strong bands of various molecules
with carbon. These have more recently been merged into a unified carbon
classifier C scheme of C1, C2 etc with the old N0 starting at roughly C6. An-
other subset of cool carbon stars are the J-type stars, which are characterized
by the strong 13 CN molecules in addition to those of 12 CN.
Each of the above mentioned spectral classes OBAFGKM have been sub-
divided into ten subclasses e.g. A0, A1 ... A9.

1.1 Luminosity Effects in Stellar Spectra

E. Hertzsprung suggested in 1905 that spectral line widths were related to the
luminosity of the stars. He pointed out that at a given apparent magnitude,
the low proper motion stars would be at larger distance from us than the high
proper motion stars of the same apparent magnitude and hence of higher
intrinsic luminosity. These low proper motion stars were found to exhibit
narrower spectral lines, so, Hertzsprung concluded that these narrow line stars
have larger intrinsic luminosity than the broad line stars.
In 1943, W. Morgan, P.C. Keenan and E. Kellman introduced luminosity
as second classification parameter. Morgan noticed the near constancy of the
gravity along the main sequence in HR diagram and luminosity class param-
eter was an attempt to identify stars of different gravities and hence radii at
nearly constant temperature.
The above mentioned system also known as Yerkes Spectral Classification.
Within the system, six luminosity classes are defined on the basis of standard
stars over the observed luminosity range.
The Six classes are:
Ia: most luminous supergiants
Ib: less luminous supergiants
II: luminous giants
III: normal giants
IV: subgiants
V: main sequence stars
The main sequence class (dwarfs) are the stars at the main sequence,
sustaining themselves through the conversion of hydrogen to helium by nuclear
fusion in the stellar core, giant is a post main sequence star which is no longer
burning hydrogen at the core but has H-burning shell outside the core. Giants
as well as supergiants have comparable mass to dwarfs but have expanded
to a much larger radii resulting in a decrease in their surface gravities. The
 Spectral Classification 167

spectral lines appear broad in the dense atmospheres of dwarfs primarily due
to pressure broadening and Stark broadening while the same line would appear
narrow in the low gravity atmospheres of supergiants.
The luminosity effects are not restricted to the narrowing of strong lines.
The line strengths and ratios of line strengths of neutral and ionized species
also show remarkable variations over spectral classes and luminosity types and
have been used for defining the subclasses and luminosity types. In addition,
there are well-known luminosity indicators such as the emission components in
the lines of Ca II H and K in late type stars which are related to the luminosity
(absolute magnitude) of the stars and the calibration of this relationship has
been carried out by Wilson and Bappu [1]. The strength of near IR OI triplet
at 7771-75 Å has been used by Osmer, Arellano and collaborators [2, 3, 4]
for absolute magnitude calibration of A−G stars. In the next section, we
will describe the line strengths and their ratios which are used to define the
spectral classes and luminosity types.

Fig. 1. The luminosity effect in late O stars from [5] with author’s permission
168 Sunetra Giridhar

2 Classification Criteria for various spectral types

Beginning from O stars which are hottest, we will briefly describe the spec-
tral characteristic for each spectral class and also list the spectral type and
luminosity class indicators. Most of the information for this chapter is taken
from Jaschek and Jaschek [6]. The figures used in this section illustrating
the different spectral types and also the luminosity effects at various spec-
tral types are taken from ‘A Digital Spectral Classification Atlas v1.02’ by
R.O. Gray with his permission. More figures can be found on the website
http://nedwww.ipac.caltech.edu/level5/Gray/frames.html.

O-type Stars

These are hot stars with temperature range of 28,000 K to 50,000 K. These
stars exhibit strong lines of neutral and ionized helium. The strength of He
II lines decrease and He I as well as H I lines increase in strength towards
later O-types. Over the spectral range O4 to B0 the line strength of He II
line at 4541 Å decreases from 800 mÅ to 200 mÅ, while He I line at 4471
Å increases from 100 mÅ to 1000 mÅ and that of H I line at 4341 Å in-
creases from 1.5 Å to 2.5 Å . The O-type spectra also exhibit the features
of doubly and triply ionized carbon, nitrogen, oxygen and silicon. The line
C III λ4649
ratios such as He II λ4686 are used for luminosity classification. Similarly, the
Si IV λ4089
ratio He I λ4143 serves as good luminosity indicator in late O type stars. Wolf-
Royet stars are a special family of O-type stars that are characterized by broad
emission lines of ionized helium, carbon (WC sequence) or nitrogen (WN se-
quence). The WC stars exhibit emission lines of He II such as He II λ4686,
ionised carbon such as C II λ4267, C III λλ3609, 4187, 4325, 4650 etc. C IV
λλ4441, 4658, 4758 etc. and lines of O II, O III, O IV and O V. These stars
are subdivided into WC classes from WC2−WC10 based upon the line ratios
of CO III λ5696 C III λ5696
V λ5592 , C IV λ5805 etc.
The WN stars exhibit emission lines of He II, N III λλ4097, 4640, 5314,
N IV λλ3483, 4057 and N V λλ4605, 4622. These stars are also subdivided
into WN subclasses using the line ratios such as N III λ4640
He II λ4686 .

B-type Stars

The B type spectra contains lines of He I, H I, C II, C III, N II, N III, O II,
Si II, Si III, Si IV, Mg II. The lines of higher ionized states of C, N, O are
present in early B stars. The maximum strength of He I line reaches near B2.
Many B stars are fast rotators and emission lines are present in some of them.
The line ratios Si III SiII λ4128−30
Si IV , He I λ4121 and
Si II λ4128−30
He Iλ4144 are used for spectral
class determination. The luminosity criteria used include the ratios of features
at λ4119λ4144
(Si IV +He II) λ4481 (Mg II)
(He I) , λ4471 (He I) which increase with luminosity. Profiles
of Balmer lines become narrower with luminosity. These are blue white stars
with temperature range of 10,000 K to 28,000 K.
 Spectral Classification 169

Fig. 2. The luminosity effect in B stars is illustrated. The figure is taken from [5]
with author’s permission

Fig. 3. The luminosity effect in A5 to G0 stars are illustrated. The figure is taken
from [5] with author’s permission
170 Sunetra Giridhar

A-type Stars

These are white stars with temperature range of 7,500 K to 10,000 K. The
A type stars exhibit strong hydrogen lines of Balmer series. The hydrogen
lines are strongest at spectral type A2. The Hγ has a strength of 13.6 Å at
A0, 17 Å at A2 and decreases to 13 Å at A7 and to 8Å at F0. The similar
pattern is followed by other lines of Balmer series. The metallic lines gradually
increase in strength from A0 to A9. The helium lines are absent. To assign
Ca Iλ4227 Mg IIλ4481 Mg IIλ4481
subclasses the line ratios MgIIλ4481 , FFeeIIIλ4045
λ4173 , F e Iλ4485 , F e IIλ4416 are
useful. However other line ratios are also used. The luminosity criteria used are
blend ratios such as F eMg
Iλλ4383−85 F e IIλ4417
IIλ4481 , Mg IIλ4481 which become stronger towards
higher luminosity. The hydrogen lines also become narrower towards higher
luminosity. The additional luminosity indicators are Sr IIλ4215 F e IIλ4351
Ca Iλ4226 , Mg IIλ4481
which increase at higher luminosity. The near infrared O I triplet at λλ 7771-75
is a very good indicator of luminosity for A-F stars.
Additional sub-classification of A type stars such as Am and Ap is done
based on their chemical peculiarities, magnetic fields, rotation and the pres-
ence of emission lines in their spectra.

Fig. 4. The luminosity effect in F stars is illustrated. The figure is taken from [5]
with author’s permission
 Spectral Classification 171

F-type Stars

These are white-yellow stars with temperature range of 6,000 K to 7,500 K.

The F type spectra have large number of metallic lines, Ca II H and K lines
are very strong which become stronger than hydrogen lines of Balmer series.
The hydrogen lines although present, are on decline from F0−F9. The G band
of CH molecule makes its appearance around F3. The Ca II at λ3933 increases
from 6.5 Å at F0 to 17.0 Å at G0, Ca I at λ4226 increases from 0.25 Å at F0
to 1.1 Å at G0 while Hγ decreases from 8 Å at F0 to 3 Å at G0. Due to large
number of lines present, at the resolution used for classification, most feature
are blended. For the spectral class the criteria used are F e IHλ4045
δ
, Ca IHλ4226
γ
,
Mn I λλ4030−34
Si II λλ4128−32 etc.
T i II λ4444
The Ca II lines show positive luminosity effect, ratios Mg II λ4481 ,
Sr II λ4077 Sr II λ4077
F e I λ4045 , Hδ are used for luminosity classification.

G-type Stars

These are yellow stars with temperature range of 4,900 K to 6,000 K. In G type
stars the hydrogen lines are further weakened and become comparable to the
strength of metal lines. Metal lines are stronger and more numerous towards
later G, molecular bands of CH and CN become visible. The spectral types are
obtained by taking the ratio of metal lines with those of hydrogen lines e.g.
using the ratios such as F e IHλ4384
γ
, F e IHλ4143
δ
, F e IHλ4045
δ
, Ca IHλ4226
δ
etc. For
spectral types later than G5 the Ca IHλ4226δ
can be used. The line ratios such
Cr Iλ4254 Cr I λ4742
as F e Iλ4250 or F e I λ4271 are recommended for stars showing compositional
anomalies such as weak metal line stars or weak G band stars.
Luminosity effects at low dispersion can be seen through CN bands. The
ratio of the Sr I+Fe I blend at Caλ4216
I λ4226 is known to be luminosity sensitive.
Ratios of Y II+Fe I at F eλ4376
Iλ4383 , Sr II λ4077
Hδ is also known to be luminosity
sensitive but will not be suitable for stars with anomalous s−process abun-
dances. Mg I triplet at λλ5167−83 are luminosity sensitive for spectral type
range G8−K5.

K-type Stars

These are orange stars with temperature range of 3,500 K to 4,900 K. In these
stars the hydrogen lines are very weak but strong numerous metal lines are
seen. The Ca II lines are very strong and CH molecular band like G band
becomes very strong. In late K stars TiO and VO bands are also seen.
The line ratios for spectral type used are Cr I λ4254 Cr I λ4254 Cr I λ4254
F e I λ4250 , F e I λ4260 , F e I λ4271 .
T i I λ3999 F e I λ4144 Ca I λ4226
Additional qualifiers are F e I λ4005 , Hδ , F e I λ4250 . The TiO band
becomes visible at K7 and MgH at K5.
The CN band increases with luminosity, so does F H δ 4101 Sr II λ4077
e I λ4071 , F e I λ4063 ,
T i II λ4400,08
F e I λ4405 also increase with luminosity.
172 Sunetra Giridhar

Fig. 5. The luminosity effect in G stars is illustrated. The figure is taken from [5]
with author’s permission

Carbon Stars

A carbon star is a late type giant with strong bands of carbon compound but
no metallic oxide band. In their spectra, very intense bands of C2 , CN and CH
are present but no bands of TiO, VO are seen. The carbon stars as a group
have been studied by Secchi [7]. Although in older classification they were
given two different types R and N, the R type stars were similar to late G and
early K but exhibited very strong Swan C2 bands around λ4700 and λ4395
which are as strong as G-band of CH. In N stars the Swan band is so strong
that the spectrum appears chopped up into sections of different intensities.
Although based upon CN band (at λ4216 and λ3833) strengths a sequence R0,
R1 to R10=N0 and subsequently upto N7 can be defined, the same pattern is
not shown by Swan C2 bands, which become strongest at R5 and weakest at
N0 and further strengthening towards later N types. A more detailed study
by Shane [8] revealed that temperature variations over the subclasses are not
large and branching of these stars R and N stars is caused by abundance
difference in C and O. If oxygen is more abundant than carbon then the
spectrum is dominated by oxides like TiO (M stars). Keenan and Morgan [9]
used the term ‘carbon stars instead of R and N wherein the overabundance of
carbon varied from star to star. They established the sequence of carbon stars
by means of temperature index of FCr Iλ4254
e I λ4250 and the strength of resonance
lines of Na I at λλ5890−96 which are good temperature indicators for the late
 Spectral Classification 173

spectral type stars. The temperature sequence C0 to C7 covers full range of

R0 to N7 which in temperature is very similar to the sequence of G4−M4.
J stars are another group of carbon stars characterized by unusually strong
isotopic bands of carbon which implies very low 12 C/13 C ratio.

Fig. 6. The luminosity effect in M2 stars is illustrated. The figure is taken from
[5] with author’s permission

M-type Stars

These are red stars with temperature range of 2,000 K to 3,500 K. The spectra
of M stars are dominated by strong bands of TiO, VO, LaO. The bands used
for classification upto M2 are those of TiO at λλ4584, 4761, 4954, 5448. After
M3 these bands saturate, those at λλ5759, 5810 saturate after M5. The VO
bands at λλ5737, 7373, 7865, 7896 become conspicuous after M7.
The luminosity effect in early M type stars can be seen through decrease
of Ca I λ4226 at higher luminosity; a similar negative luminosity depen-
dence is exhibited by Cr I feature at λλ4254−74−90. Ratios Sr II λ4077
F e I λ4263 and
(Y II+F e I)λ4376
F e I λ4386 increase with luminosity. The K I λ7699, Na I λ8183 and
λ8185 decrease in intensity from dwarfs to giants. The Ca II triplet at λλ8498,
8542 and 8662 is a very important luminosity indicator in these stars. It is
very weak in dwarfs but becomes very strong in giants and supergiants.
174 Sunetra Giridhar

S-type Stars

These stars are similar in temperature to K5 to M stars but exhibit strong

bands of ZrO with band heads at λλ4640, 5551 and 6474.

3 New Spectral types L and T

These are the coolest stars with temperatures less than 2500 K. Since they are
very cool, they emit mostly in infrared wavelengths. The original classification
scheme of L dwarfs based on red to far-red spectral region has been described
in [10, 11]. The L stars have a temperature range of 2500 K to 1300 K and
exhibit some overlapping features with M stars such as TiO, VO although
these features do not remain as strong. These features are very weak in L3
and almost disappear at L6 leaving only metal hydride bands of FeH, CaH,
CrH. These bands have maximum strength at mid L but become weak in late
L types where H2 O presents a strong feature at 9300 Å. The atomic features
such as resonance lines of Rb I at 7800 Å, 7948 Å and Cs I at 8521 Å,
8945 Å can be seen in early L and grow in strength throughout the sequence.
The resonance line of Li I at 6708 Å is present in L dwarfs. The K I resonance
doublet shows very remarkable change. Being somewhat narrower and weak at
M9, it broadens and becomes strong through L0 onwards such that at L5 two
separate cores are indistinguishable and doublet looking like a broad trough
∼ 600 Å wide; at L8 it is ∼ 1000 Å wide. The Na I resonance doublet also
follow the same pattern by becoming broad and shallow at L5 .
These spectral variations like disappearance of TiO, VO and other molecu-
lar features at later L subclasses are believed to be caused by formation of var-
ious types of condensates at the temperatures lower than 2600 K. Lodders [12]
have made extensive chemical equilibrium calculations and the atmospheric
composition changes as the material is removed from gas to solid phase.

3.1 The T dwarfs

As the brown dwarfs cool to temperatures below 1400 K, a drastic change

occurs between 1400 K to 1200 K when CH4 bands become strong while CO
band weaken. The CH4 bands are extremely broad in near infrared and they
even modify the near infrared broad band colors. Strong methane absorption
at 1.6 and 2.2 μ reduces more than half flux from H and K pass-band giving
them unrealistic colors.
A set of flux ratios measuring the strengths of methane and H2 O have
been attempted to classify the T dwarfs by Burgasser and co-workers [14]
and others. Considerable progress will be made through deeper IR surveys.
 Spectral Classification 175

Fig. 7. The spectral variations seen from late M to L stars are illustrated. The
figure is taken from the website www.stsci.edu/inr/ldwarfs originally published in
[10]

4 Modification of MK system

The MK system has gone through extensive research and refinement. The
important work in refining and extending the MK system in dimensions be-
yond the traditional two dimensional spectral type luminosity class grid has
been carried out by Keenan and co-workers and the use of third ‘metallicity
parameter’ is recommended. Grey [16] has carried out the extension of MK
system to metal-weak F and G stars. A grid of metal-poor standard stars has
been prepared with the objective of providing a mean of classifying metal-
weak stars using a standard of similar metallicity rather than the old method
of comparing the metallic line strengths of the program stars with that of a
considerable earlier solar composition standard.
The earlier attempt of deriving spectral types for metal-poor stars re-
lied upon hydrogen line strengths or G-band strength due to their strong
176 Sunetra Giridhar

Fig. 8. The drastic variations in KI resonance doublet is illustrated above. The

figure was originally published in [13]

temperature dependence. However these classifiers are not adequate; the hy-
drogen line strengths do depend upon other stellar parameters and are not
suitable at the later spectral types. Corbally [17] has pointed out that the
ratio of Cr I resonance triplet at 4254, 4274, 4290 Å to the neighboring lines
of Fe I at 4251, 4272, 4362 Å arising from excited level is nearly independent
of metallicity and hence can serve as temperature indicator. For luminosity
classification the ratios of features like T iFII, F e IIλλ4172−9
e I λ4203, λ4271 are useful for F
Sr II λ4077 F e Iλ4383
stars while for G stars F e I λ4063, λ4046 and Y II λ4375 features may be useful
since these lines (blends) are very strong and remain discernible even at low
metallicity. A list of standard stars for metallicity classification can be found
in [16].
 Spectral Classification 177

Fig. 9. The strengthening of CH4 bands in T stars is illustrated above. The figure
was originally published in [15]

5 Contemporary methods of spectral classification

Stellar classification has been an important tool in stellar and Galactic astron-
omy since it provides empirical measure of the fundamental stellar parameters
such as temperature, luminosity and metallicity.
The massive surveys both ground based as well as from space missions pro-
vide large number of stellar spectra covering distant components of Galaxy.
To understand the complex evolutionary history of our Galaxy, rapid and ac-
curate methods of stellar classification are necessary. A short review of the
automated procedures are presented here. The most commonly used auto-
mated spectral classification methods are based on (a) Minimum Distance
Method (MDM) (b) Gaussian Probability Method (GPM) (c) Principal Com-
ponent Analysis (PCA) and (d) Artificial Neural Network (ANN). We chose
to describe only two of them to introduce the automated approach of classi-
fication.
178 Sunetra Giridhar

Fig. 10. The spectra of a few metal-poor stars are shown above

In MDM, the classification is done by minimizing the distance metric be-

tween the object to be classified and each member of a set of templates. The
object is assigned the class of the template, which gives the smallest distance.
In this approach the number of templates used to define subclasses limit the
accuracy of classification. Interpolation can be made to make intra-class as-
signment. Katz [18] used this method with χ2 minimizing on high resolution
Elodie spectra using a large number of reference stars of known Tef f , log g
and [M/H] to derive atmospheric parameters of target stars. These authors
achieved accuracy of 86 K in Tef f , 0.28 in log g and 0.35 in [M/H].
The Neural Network methods have become very popular due to their
speed and objectivity as explained in the papers [19, 20, 21, 22] and more re-
cently in [23]. These are computational methods which can provide non-linear
parametrized mapping between an input vector (a spectrum for example) and
one or more outputs like SpT, LC or Tef f , log g and [M/H]. For the network
to give required input-output mapping, it must be trained with the help of
representative data patterns. These could be stellar spectra (or a set of line
strengths measured from a spectrum) for which classification or stellar param-
eters are well determined. The training procedure is a numerical least square
error minimization method. The training proceeds by optimizing the network
parameters (weights) to give minimum classification error. Once the network
is trained the weights are fixed, the network can be used to produce output
SpT, LC or Tef f , log g and [M/H] for an unclassified spectrum.
 Spectral Classification 179

The ANN has been used in very large number of stellar applications. Vieira
and Ponz [24] have used ANN on low-resolution IUE spectra and have de-
termined SpT with an accuracy of 1.1 subclass. Bailer-Jones and Irwin [19]
used ANN to classify spectra from Michigan Spectral Survey with an accuracy
of 1.09 SpT. Prieto and co-workers [25] used ANN in their search of metal-
poor stars. Snider and co-workers [26] used ANN for the three dimensional
classification of metal-poor stars.
We have made a modest effort to use ANN for parametrization of a sample
of stars in the temperature range 4500 to 8000 K. We have used a medium
resolution Cassegrain spectrograph with the 2.3 m Vainu Bappu Telescope at
VBO, Kavalur, India to get spectra at resolution (R) of 2000. Using 90 spectra
for stars of known parameters, and ANN of 680:11:3 architecture we could
attain an accuracy of ± 200 K in temperature and ± 0.3 dex in metallicity.
It is very important to envisage an approach that would give quick, reliable
spectral classifications (or stellar parameters) for stars falling in all regions
of HR diagram. The pipeline procedures are being developed for the future
ambitious missions such as GAIA and PAN-STARS.

References
1. O. C. Wilson, M. K. V. Bappu, ApJ 125, 661 (1957)
2. P. S. Osmer, ApJS 24, 247 (1972)
3. A. Arellano Ferro, S. Giridhar, A. Goswami, MNRAS 250, 1, (1991)
4. A. Arellano Ferro, S. Giridhar, E.Rojo Arellano, RMxAA 39, 3, (2003)
5. R. O. Gray, A Digital Spectral Classifcation Atlas V1.02, (2003)
6. C. Jaschek, M. Jaschek, The classification of stars, Cambridge University
Press, Cambridge (1987)
7. A. Secchi, Mem. Soc. Ital. Sciencze (2) 3, 73 (1868)
8. C. D. Shane, Lick Bull. 13, 123 (1928)
9. C. P. Keenan, W. W. Morgan, ApJ 94, 501 (1941)
10. J. D. Kirkpatrick, I. N. Reid, J. Liebert et al., ApJ 519, 802 (1999)
11. J. D. Kirkpatrick, I. N. Reid, J. Liebert et al., AJ 120, 447 (2000)
12. K. Lodders, ApJ 519, 793 (1999)
13. I. N. Reid, J. D. Kirkpatrick et al., AJ 119, 369 (1999)
14. A. J. Burgasser et al. ApJ 524, 421 (2002)
15. B. R. Oppenheimer, S. R. Kulkarni et al., ApJ 502, 932 (1998)
16. R. O. Gray, AJ 98, 1049 (1989)
17. C. J. Corbally, AJ 94, 161 (1987)
18. D. Katz, C. Soubiran et al., A&A 338, 151 (1998)
19. C. A. L. Bailer-Jones, M. Irwin, T. von Hippel, MNRAS 298, 361, (1998)
20. C. A. L. Bailer-Jones, An Introduction to Artificial Neural Networks in
Automated Data Analysis in Astronomy ed. by R. Gupta, H. P. Singh
et al., (New Delhi, London : Narosa Pub. House), p83 (2002)
21. T. von Hippel, L. J. Storrie-Lombardi et al., MNRAS 269, 97 (1994)
180 Sunetra Giridhar

22. H. P. Singh, C. A. L. Bailer-Jones, R. Gupta, Principal Component Anal-

ysis and its Application to Stellar Spectra in Automated Data Analysis
in Astronomy ed. by R. Gupta, H. P.Singh et al., (New Delhi, London :
Narosa Pub. House), p69 (2002)
23. S. Giridhar, S. Muneer & A. Goswami, MmSAI 77, 1130, (2006)
24. E. F. Vieira, J. D. Ponz, A&AS 111, 393 (1995)
25. C. Allende Prieto, R. Rebolo, et al., AJ 120, 1516 (2000)
26. S. Snider, C. Allende Prieto, T. von Hippel et al., ApJ 562, 528 (2001)
 Part II

Massive Stars, Core Collapse,

Explosive Nucleosynthesis
Weak Interaction Rates for Stellar Evolution,
Supernovae and r-Process Nucleosynthesis

Kamales Kar

Saha Institute of Nuclear Physics, Bidhannagar, Kolkata - 700064, INDIA

Summary. The role of weak interaction processes in the late stage of evolution of
massive stars and during supernova explosions is described. Models for calculating
the β-decay, electron capture and neutrino capture rates on nuclei are reviewed.
The neutrinos produced in the sun in the weak interaction reaction processes when
detected on earth gave rise to the solar neutrino deficit problem. The resolution
of this problem is also mentioned. Roughly half the elements heavier than iron are
produced through r-process nucleosynthesis. The methods of calculating the β-decay
rates of very neutron-rich nuclei taking part in the r-process are also presented.

Keywords: Weak interaction rate, β-decay, Electron capture, Stellar collapse,

Supernovae, r-process nucleosynthesis

1 Introduction

Different weak interaction processes like β-decay and electron capture are very
crucial at the late stage of stellar burning and gravitational collapse for the
type II supernovae. The energy of the shock wave that eventually causes the
explosion and the outward propagation of the shock depends on the rates of
these processes. This article describes this aspect and also presents the nu-
clear structure models used to calculate the rates of these processes. The fu-
sion reactions in the sun also involve some of the weak reactions that produce
neutrinos. The detection of the solar neutrinos led to the solar neutrino puz-
zle. That is also covered in this article along with the β-decay rates needed
for r-process nucleosynthesis. In Sect. 2 we mention some basic facts of nu-
clear physics that will be repeatedly used. Section 3 gives an overview of core
collapse supernovae. Section 4 points out the stages of the supernova where
electron capture, β-decay and neutrino capture become important. Section 5
describes the different structure models used to calculate the rates and Sect. 6
gives the results of incorporating some improved electron and β-decay rates

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 4, c Springer-Verlag Berlin Heidelberg 2010
184 Kamales Kar

for the supernova simulations. Section 7 is concerned with the weak interac-
tion reactions that produce neutrinos in the pp-chain of stellar burning and
how one explains the observed deficit of neutrinos in terrestrial detectors. Fi-
nally in Sect. 8 we discuss the r-process nucleosynthesis through which highly
neutron-rich heavy nuclei are produced. The possible sites are discussed and
the models for calculating the β-decay rates are presented.

2 Some Nuclear Physics Basics

A nucleus X with atomic mass A and atomic number Z will be denoted as
A
Z X. Sometimes we also write it as (N, Z) where the number of neutrons in the
nucleus is given by N = A − Z. The binding energy per nucleon for the nuclei
have a range from a fraction of MeV, for the nuclei very close to drip line
to 8-9 Mev, for the iron type of nuclei. The nuclear energy eigenstates have
angular momentum, isospin and parity as good quantum numbers. The total
angular momentum is the sum of the angular momenta of all the nucleons. The
ground state angular momenta of even(Z)-even(N) nuclei are always zero. The
ground state angular momentum for the odd A nuclei or the odd-odd nuclei
are predicted by simple rules based on the shell structure of the nucleons.

Isospin

The neutron and the proton are considered to form the isospin doublet with
T = 1/2 and with MT 1/2 and -1/2 respectively. Denoting |n >= |1/2, 1/2 >
and |p >= |1/2, −1/2 > one gets t+ |p >= |n > and t− |n >= |p > where
t+ = tx + ity and t− = tx − ity . A nucleus (N,Z) has isospin |T, MT > with
the eigenvalue MT = (N − Z)/2.

2.1 Shell Model

The nucleons inside a nucleus, due to the dominant mean field, have a shell
structure with the quantum numbers (nlj) (standing for the radial, orbital
angular momentum and total angular momentum quantum numbers respec-
tively) defining each orbit, similar to atomic spectroscopy. In the limit of
non-interacting fermions the nucleons occupy the orbits of lowest possible en-
ergies filling up the shells (1s1/2 ), (1p3/2 , 1p1/2 ), (1d5/2 , 2s1/2 , 1d3/2 ), ...and
the ground state energy is the sum of the energy of the nucleons. Each shell
differs in energy from the lower by 1h̄ω. The 1-body Hamiltonian is given by

H0 = Σλ λ a†λ aλ (1)

where the summation is over the orbits λ. For example 16 O has the first two
shells completely filled and is one of the magic nuclei. 20 Ne has 16 O as a closed
core and has 4 nucleons outside (MT = 0 for the ground state) occupying the
 Weak Interaction Rates 185

d5/2 orbit. The nuclear ground state wavefunction is the antisymmetrised

product of the single particle states.
But the nuclei indeed have residual interaction among the nucleons but
that does not destroy the shell structure. The additional Hamiltonian is given
by 
H = Σ < kl|H|ij > a†l a†k ai aj (2)
giving rise to non-diagonal matrix elements in the two-particle matrix rep-
resentation. Now the numerical procedure of calculating the energy eigen-
values and wavefunctions for a nucleus given a model space with specified
active single particle orbits and two body nuclear matrix elements is known
as the ‘shell model’ in nuclear structure [1]. The model also can calculate
the transition matrix elements of different excitation operators. As nuclear
states have (J, T ) and parity as conserved quantities the shell model consists
of the following steps: a) construct all the m-particle basis states with fixed
(J, T ) for a nucleus with m valence nucleons by distributing them over the
active orbits, b) calculate the matrix elements of the total Hamiltonian in
this m-particle basis and c) finally diagonalising the matrix to obtain the
eigenvalues and eigenfunctions. Each state in the original space is called a
configuration and this shell model diagonalisation gives rise to configuration
mixing. These spaces often have large dimensions. For example, for 28 Si for
states with (J, T ) as (2, 0) and (2, 1), the matrix dimensions are 3276 and
5768 respectively, staying within the sd-shell. For highly excited states in
shell model studies one may need to include single particle orbits from the
higher shell in the process enlarging the space and its dimension.

2.2 β-decay

The β − and the β + are given as :

(N, Z) → (N − 1, Z + 1) + e− + ν̄e (3)

(N, Z) → (N + 1, Z − 1) + e+ + νe (4)
with the release of energy Qβ − = [M (N, Z) − M (N − 1, Z + 1)]c and
2

Qβ + = [M (N, Z) − M (N + 1, Z − 1) − 2me ]c2 respectively. The process of

electron capture [EC] has the reaction

(N, Z) + e− → (N + 1, Z − 1) + νe (5)

Here the release of energy is QEC = [M (N, Z] − M [N + 1, Z − 1)]c2 + Ee ,

where Ee is the energy of the electron absorbed. Finally the neutrino capture
reaction is
(N, Z) + νe → (N − 1, Z + 1) + e− (6)
with the corresponding energy release of
Qν = [M (N, Z) − M (N − 1, Z + 1)]c2 + Eν . The half-life t1/2 for a β-decay
from the ground state |i > is given by
186 Kamales Kar

t1/2 = ln 2/(Σf wf i ) (7)

where the transition probability wf i is summed over all final states to which
the transitions are possible energetically and given by the expression

wf i = (mc2 /h̄) (Γ 2 / π 2 ) |Mf i |2 f (Z, ) (8)

with 0 = E0 /(me c2 ),  = E/(me c2 ) and Γ is related to the weak inter-

action constant. The phase factor f (Z, ) has the expression [2]
 0
f (Z, ) = dF (Z, )(0 − )2 (2 − 1)1/2 (9)
0

where F (Z, ) is the Coulomb distortion factor. The squared nuclear matrix
is
|Mf i |2 = |CF |2 |MF (f i)|2 + |CGT |2 |MGT (f i)|2 (10)
This has contribution from the allowed transitions by Fermi and Gamow-Teller
(GT) operators. They have the form

MF (f i)|2 = | < f |Σk t± (k)|i > |2 /(2Ji + 1)f orβ ± decay (11)

|MGT (f i)|2 = | < f |Σk σ(k)t± (k)|i > |2 /(2Ji + 1)f orβ ± decay (12)
The transition probability for the process of electron capture in stellar tem-
peratures T is given by

λEC = (ln2/K)Σi ((2Ji + 1)exp(−Ei /kT )/G(Z, A, T )) × Σj Bij (GT )Φij (13)

where Φij is the phase space factor for EC and K is a constant.


The electron capture transitions change the initial isospin T to T = T+1.

3 Overview of Core Collapse Supernovae

Supernova (SN) explosions are one of the most violent explosions in the uni-
verse where energies of a few times 1053 ergs are released in a few tens of
seconds. Though long ago Baade and Zwicky associated this with the death
of large stars but the exact mechanism of the explosions was one of the out-
standing problems of the last century which is still not understood completely.
Here we shall be concerned with the explosions that start with the collapse
of the core. Though from the outer regions of the star different forms of elec-
tromagnetic radiation give information about the physical processes in the
envelope but nothing except the neutrinos can come from the much denser
core region. The production of neutrinos in SN was spectacularly confirmed
by the 11 and 8 neutrinos detected in two different terrestrial neutrino de-
tectors during the explosion of Supernova 1987A. The theory of core collapse
 Weak Interaction Rates 187

Fig. 1. The stellar evolution leading to collapse for stars of different masses (taken
from [6])

supernova involves elements of astrophysics, particle physics, nuclear physics

and statistical mechanics [3, 4, 5].
The stellar burning stages leading to the late silicon burning in the pre-
supernova stage is depicted in Figure 1. Massive stars (with mass > 8M )
collapse once the nuclear burning in the centre of the core stops and the core
mass exceeds the Chandrasekhar mass. A combination of electron capture and
photodissociation of the iron type nuclei initiates the process. The fraction of
neutrons are small at this stage and the fraction of protons much smaller. The
radius of the core is a few thousand kilometres and the temperature less than
1 MeV. The density of matter at the start of the collapse is around 109 g/cm3 .
At this stage electron capture on the nuclei and on the very small fraction of
free protons reduce the electron density and hence the pressure. The electron
fraction is expressed by Ye , the ratio of the number of electrons to the number
of nucleons.
When the density reaches the value around 1011 g/cm3 , the neutrino mean
free path becomes close to the core radius and so the neutrinos that were
escaping the star earlier get trapped inside. So beyond this density the lepton
fraction Yl becomes constant. The entropy change during this period is not
enough to break up the nuclei and hence the nuclei survive through this stage.
Beyond the nuclear matter density the nucleons touch each other and pressure
waves start travelling from the centre, the region of highest density. Matter
becomes stiffer, the pressure waves stop going outward at the ‘sonic point’ and
188 Kamales Kar

eventually a shock wave develops there. The timescale of this whole process
is tens of milliseconds. Eventually the shock gathers enough energy to start
moving outward. In the early eighties it was thought that this leads to the
explosion. But the simulations done showed otherwise. The shock in almost
all simulations lost energy fast by dissociating the nuclei in its path and soon
became a static shock after moving a few hundred kilometres. The typical
dissociation reaction on a iron nucleus is:
56
F e → 13α + 4n (14)

α → 2n + 2p (15)
Cooling through neutrino emission also worked against the shock motion. For
every 0.1 M , dissociation energy ∼ 1.7 × 1051 ergs is lost and that clearly
indicates why the shock stops moving out. Except for a few special cases
known as ‘prompt’ explosions for smaller star masses with lower entropies and
softer equations of state, all simulations gave no explosion. Now one believes
that the energising of the stalled shock takes place through a process known
as ‘delayed neutrino heating’ where the neutrinos and the weak reactions
involved are crucial.
From the central region of the core a huge number of neutrinos of all
three flavors radiate out. They originate from the high temperature region
deep inside where ν ν̄ pairs are produced thermally. These neutrinos through
charge current and neutral current interactions heat up the matter at the
stalled region over a timescale of a few seconds. Though a very tiny fraction
of the neutrinos interacts but their number is still large enough to revitalise
the shock over a long timescale. Unfortunately as yet very few simulations
lead to the shock reaching the edge of the core. In other words, even today
with a lot of improvements in the simulations taking into account large-scale
convection, relativistic treatment of the matter etc. very few simulations give
rise to an explosion. One of the few cases where explosion is seen is given in
Figure 2 [7].
One should mention here that some stars with mass in the range of
9-10M end up with a O−N e core instead of the normal iron core. These stars
even in 1 dimensional simulations, are seen to have the stalled shock moving
out in 150 ms after bounce and mass shells start ‘being ablated’ from the
protoneutron star about 50 ms later giving rise to a successful explosion [8].
The research work today concentrates on better treatment of convection
by combing 2D/3D hydrodynamics for the matter coupled with the neutrino
radiative transport. Improved pre-supernova and collapse conditions and more
detailed treatment of the neutrino-matter interaction are the other areas of
activity. One also envisages that an instability of accretion shock to non-
radial deformation (particularly l = 1, 2) known as Standing Accretion Shock
Instability (SASI) can lead to a breakthrough in the understanding of the
energy transport [9].
 Weak Interaction Rates 189

Fig. 2. The results of a hydrodynamic simulation showing an explosion (taken

from [7])

4 Weak Interaction Processes in Supernova Evolution

We discuss the different stages of stellar evolution and core collapse supernovae
where the β-decay and its inverse processes play important roles:

4.1 At the Pre-SN Stage

As already described electron capture on nuclei start at the late stages of

stellar burning. It is also realised that the β-decay of the unstable nuclei
increase Ye and compete with the electron capture. We have seen in equations
(5) and (3) that in the case of EC only GT transitions participate (T → T + 1
is not possible by the Fermi operator) where for β-decay, Fermi transitions
are, in principle, allowed. Figure 3 gives a sketch of the contributions from
Fermi and GT for β-decay and EC. The timescale of silicon burning is a few
days to a month. During the late phase of Si burning successive addition of
alpha particles to silicon and other nuclei around that mass take place at
temperatures T = 4 × 109 (i.e. 350 keV) with the end product being the
‘iron group’ nuclei (like Mn, Fe, Co, Ni,...). Near the end of this burning
stage Nuclear Statistical Equilibrium (NSE) is reached and the system is then
characterised by ρ, T and Ye .

4.2 At the Collapse Stage

As the collapse proceeds the electron captures along with the β-decays of the
neutron-rich nuclei control the change in Ye . But as the density increases so
does the electron chemical potential and when it exceeds the Q-value of the
190 Kamales Kar

Electron capture

Beta decay 20
15

15

E (MeV)
10
E (MeV)

10
5
GT−
5
Finite T F
0
(Z, A) GT+
0
(Z+1, A)
Fig. 3. The Fermi and GT resonances observed through β-decay and electron cap-
ture between nuclei (N,Z) and (N-1,Z+1) (taken from [6])

β-decay, the decay essentially stops. Also the equation of state of the matter
makes the average nucleus heavier i.e. the mass number slowly increases from
the initial value of 56. When the neutron number crosses the value of 40,
the allowed electron capture starts getting blocked. This nuclear structure
effect, first pointed out by Fuller [10], reduces the rate of decrease of Ye
substantially. We can understand this effect with the example of the nucleus
74
Ge given in Figure 4. 74 Ge has 32 protons and 42 neutrons. In the extreme
single particle model i.e. with 1-body H only, the neutrons fill the f p-shell
with the 4 orbits 1f7/2 , 2p3/2 , 2p1/2 and 1f5/2 completely. Two extra neutrons
occupy the 1g11/2 orbits. The protons on the other hand outside the closed Ca
core occupy the 1f7/2 and 2p3/2 orbits. On these valence protons no allowed
electron capture takes place as the corresponding 4 neutron orbits are full. Of
course the contribution from the closed Ca core is zero too. Hence one sees
that for these nuclei with N > Z once N crosses 40 allowed electron captures
stop. The rate of ‘forbidden’ electron capture is 2 to 3 orders lower. This effect
is known as the ‘blocking’ of the EC rate.

4.3 Mechanisms of Unblocking

But actually complete blocking of the rates does not start suddenly. Below
we describe the mechanisms for unblocking which operate in reality. Firstly,
there is always an ensemble of nuclei and not just one. So while 74 Ge may
 Weak Interaction Rates 191

protons neutrons
x x 1g9/2

–
2p1/2

–
1f5/2
FULL

xx xx –
2p3/2

–
1f7/2

Fig. 4. The occupation of the different orbits for the nucleus 74

Ge

have its EC rate blocked lighter nuclei in the ensemble with N < 40 with
smaller abundances will have rates with allowed transitions. Secondly the
extreme single particle model does not take into account the residual inter-
action and the configuration mixing. For the 74 Ge case, a small contribution
in the n-sector of (2p1/2 )1 (2g9/2 )3 to the dominant (2p1/2 )2 (1g9/2 )2 will make
a 2p3/2 → 2p1/2 transition possible. Thirdly one can also thermally excite
a neutron from the 2p1/2 to the 1g9/2 orbit causing the allowed transition.
Finally a small deformation of the nucleus from spherical shape will lead to
the Nilsson model being operative instead of the spherical shell model. There
each orbit no longer has the (2j + 1) degeneracy but each |m| component has
different energy depending on the deformation where the degeneracy for each
one is 2 coming from the states with quantum numbers ‘m’ and ‘-m’ having
the same energy. This also leads to non-zero allowed transition rates.

4.4 At the late time neutrino heating stage

The heating of the matter through neutrinos take place through the charged
current electron neutrino capture reactions given by Equation (6). As matter
behind the shock is heated by neutrinos coming from deep inside the core, its
temperature rises and that leads to radiation of neutrinos. Bethe and Wilson
[11] first pointed out that the electron neutrino absorption on free nucleons
and neutrino-electron elastic scattering (both charged and neutral current
contributions) is able to make the shock move out again. A self-consistent
192 Kamales Kar

treatment of the heating and the dissociation of the nuclei was done [12].
However Haxton [13] stressed the importance of neutral current contribution
in elastic scattering off nuclei given by

(N, Z) + νx = (N, Z) + νx (16)

where ‘x’ stands for all three flavors, particularly the part that arises due to
the excited states of the nucleus. The net rate of gain of energy by a gram of
matter is given by
Ė = Ė1 − Ė2 + Ėscatt (17)
where Ė1 is the heating rate and Ė2 is the rate of energy loss due to emission
of neutrinos. Ėscatt is the heating through the scattering processes mentioned
above. The nuclei involved are the ‘iron group’ nuclei near the shock front,
in particular. The amount of heating through these reactions are carefully
calculated. Also processes that can help in getting extra heating are also
investigated. One such idea was the flavor mixing through matter enhanced
oscillation increasing the heating. But one finds that with the realistic masses
the mixing cannot take place within a few hundred kilometres as needed in
the core collapse scenario.

5 Nuclear Models for Calculation of the Weak

Interaction Rates
5.1 Systematics with simple shell structure

The first realistic estimation of the electron capture and β-decay rates in-
cluding the contribution from the giant resonance was by Fuller, Fowler and
Newman (FFN) [14]. While for the sd-shell nuclei shell model results were
used, for the f p-shell nuclei low-lying experimental values of logf t were supple-
mented by contributions from a collective strength whose energy position and
magnitude were estimated using systematics and single particle shell structure
for nuclei with A < 60.
Working on improved FFN ideas Aufderheide et al. [15] calculated the
EC and β-decay rates for nuclei beyond A = 60 and for different density-
temperature grid points for the presupernova evolution and collapse. They also
tabulated the important nuclei that need to be considered in the ensemble.

5.2 Statistical models for β strength

One of the earliest models that treat the β-decay strength in a statistical
way is the ‘Gross Theory’. Here assuming enough final states, the strength
function is given as an integral of a one-particle strength function multiplied
by a weight function to take into account the Pauli blocking and the density
of single particle states [16, 17]. The strength function took into account the
 Weak Interaction Rates 193

allowed and forbidden transitions and also the pairing and sum rules in a
simple approach. It was almost independent of the nucleus considered except
for considering its Q-value. Later the theory was improved [18, 19] by taking
into account the shell effects and modifying the strength function. This is
called the ‘Semi-gross Theory’ or the second generation gross theory.
The other approach involved constructing the giant resonance using ele-
ments of spectral distribution theory [20,21,22] and experimental information
on the resonance. Here one first recognises the sum rule quantities correspond-
ing to the transition strength due to the excitation operator O. The sum rule
strength SO is given by

SO = Σf | < f |O|i > |2 = Σf < i|O† |f >< f |O|i >=< i|O† O|i > (18)

This is the total strength of the giant resonance plus the strength of the dis-
crete states in the low-lying excitation region. For the Fermi and GT β-decay
transitions these sum rules are exactly given by the Ikeda sum rule expres-
sions [23]:
Sβ − (GT ) − Sβ + (GT ) = 3(N − Z) (19)
Sβ − (F ) − Sβ + (F ) = (N − Z) (20)
However these sum rules give the difference of the total electron β-decay
and the total positron β-decay strengths. For finding the β strength sums
individually, Kar, Ray and Sarkar [24] expressed the strength sum in terms
of the average neutron and average proton occupancies. These occupancies
averaged over the configurations were evaluated using the spectral distribution
theory. One finds that the assumption that the GT strength distribution in
final energy is close to a Gaussian is not a bad approximation to start with
and fixes the centroid by the best fit to observations in (p, n) reactions [25]

GT = IAS + 26A−1/3 − 18.5(N − Z)/A (21)

where GT is the GT centroid energy and IAS is the energy of the Isobaric
Analog State (IAS). Deviation from the Gaussian can also be handled here by
using a skewness. The width of the GT resonance is left as a free parameter
and fixed by the best fit to observed half-lives of nuclei in this region. This
form of the β strength function is then used to calculate the rates for the
presupernova β-decay. There are also some special excited states of the mother
nucleus which have large overlaps with states in the ground state region of the
daughter and their contributions are also taken into account. These so-called
‘back resonances’, first pointed out by Fuller, can become important for the
higher temperatures.
The statistical method using the spectral distribution theory including
orbits higher than the f p-shell was recently used to evaluate the β-decay
rates for nuclei involved in the collapse stage but with A > 65. The half-lives
using this method are compared to the experimental values in Table 1 [26].
The rates are given in Table 2.
194 Kamales Kar

Table 1. Half-lives of nuclei with A > 65 using statistical model [26]

expt. calc
Sr No. Nucleus Z Qval T1/2 T1/2
(MeV)
66
1 Co 27 10.0 0.23 0.32
67
2 Ni 28 3.385 21.0 26.91
68
3 Co 27 9.30 0.18 0.20
68
4 Ni 28 2.06 19.0 18.50
68
5 Cu 29 4.46 31.1 31.23
69
6 Co 27 9.30 0.27 0.36
70
7 Ni 28 5.36 11.4 12.04
71
8 Ni 28 6.90 1.86 1.62
72
9 Cu 29 8.22 6.60 6.13
74
10 Cu 29 9.99 1.59 0.40

Table 2. Rates for β − decay for typical supernova densities and temperatures [26]
Nucleus ρ(gms/cc) Temperature in o K
Ye 2× 109 3 × 109 4 × 109 5 × 109
Rates (s−1 )
69 Co 109 0.47 7.25 × 10−3 1.18 × 10−1 2.96 × 10−1 3.63 × 10−1
108 0.47 1.64 × 10−2 2.62 × 10−1 6.37 × 10−1 7.53 × 10−1
107 0.47 1.94 × 10−2 3.05 × 10−1 7.32 × 10−1 8.55 × 10−1
109 0.45 7.47 × 10−3 1.22 × 10−1 3.05 × 10−1 3.73 × 10−1
108 0.45 1.66 × 10−2 2.64 × 10−1 6.41 × 10−1 7.57 × 10−1
107 0.45 1.94 × 10−2 3.06 × 10−1 7.33 × 10−1 8.56 × 10−1
68 Ni 109 0.47 8.38 × 10−6 7.81 × 10−4 6.00 × 10−3 1.59 × 10−2
108 0.47 2.18 × 10−3 4.38 × 10−2 1.34 × 10−1 1.93 × 10−1
107 0.47 4.55 × 10−3 8.08 × 10−2 2.20 × 10−1 2.90 × 10−1
109 0.45 1.08 × 10−5 9.18 × 10−4 6.76 × 10−3 1.75 × 10−2
108 0.45 2.25 × 10−3 4.50 × 10−2 1.36 × 10−1 1.97 × 10−1
107 0.45 4.57 × 10−3 8.11 × 10−2 2.20 × 10−1 2.91 × 10−1

5.3 Microscopic Models

Among the microscopic models, the most accurate one for the presupernova
and supernova calculations is the shell model. Over the years very large space
shell model calculations for nuclei in the f p-shell were found to very successful
and weak interaction rates have been calculated extensively using the shell
model results [27, 6].
Shell model calculations in the f p-shell using the modified Kuo-Brown
(KB3) interaction with the valence orbits f7/2 , p3/2 , p1/2 and f5/2 have been
carried out using the Strasbourg-Madrid [28] code. They reproduce the ex-
perimental spectra well. Information on the GT matrix elements come from
the (p, n) and (n, p) reactions for the β − and β + decays. Calculations for GT
 Weak Interaction Rates 195

transition matrix elements using wavefunctions with the KB3 interaction us-
ing the same code again showed good agreement. This is seen in Figure 5.
The figure compares the GT strength for β + transition as a function of en-

1.0 0.5
54 51
Fe FFN V
0.8 Exp 0.4
SM
0.6 0.3

0.4 0.2

0.2 0.1

0.0 0.0
56 55
Fe Mn
0.6
GT Strength

0.4
0.4
0.2
0.2

0.0 0.0
58Ni 59Co

0.6
1.0
0.4
0.5
0.2

0.0 0.0
–2 0 2 4 6 8 10 12 –2 0 2 4 6 8 10 12
E (MeV) E (MeV)

Fig. 5. Comparison of GT matrix elements calculated by shell model and observed

in (n, p) experiments (taken from [6])

ergy for the shell model with the same obtained through (n,p) reactions for
the 6 nuclei 54 Fe, 56 Fe, 58 Ni, 51 V, 55 Mn and 59 Co. The figure also shows the
centroid of the GT distribution used in FFN. Similar comparisons were done
for β − transitions as well. The success of the shell model results motivated
the authors in calculating β − , β + and EC rates for a large of nuclei in the
mass range of 40 < A < 65 for a wide range of density and temperature. For
the nuclei in the range 65 < A < 80 Pruet and Fuller [29] did the calculations
extending the FFN model alongwith taking into account the GT resonance.

5.4 Calculations with Improved Rates

The detailed shell model results in agreement with experiments led to better
EC and β-decay rates. The EC rates were often reduced compared to the FFN
196 Kamales Kar

earlier values. For example the EC rate on 60 Co was two orders of magnitude
lower for typical presupernova conditions! Presupernova evolution of stars in
the mass range of 11M to 40M were carried out using the full FFN and
the rates based on the shell model results (LMP) by redoing the Woosley and
Weaver calculations [31] only changing the weak rates from FFN to LMP.
These results are displayed in Figures 6 to 9 [30, 6].
Figure 6 plots the change of Ye for 15M and 25M stars as a function
of time till core bounce. The different nuclei that are most important at a
particular time are also given both for EC and β-decay.

0.50

55
Fe EC
0.48 56Mn β−
57
Fe
56
Mn
Ye

0.46
57
Fe
58Mn 57
Fe
0.44 64Co

15 M
0.42
106 105 104 103 102 101 100
0.50
54
Fe
54
Mn
0.48
56
Fe
Ye

56
Mn
0.46 57
Fe 1
56
Mn H
58
Mn
25 M
0.44
105 104 103 102 101 100
Time till core collapse (s)

Fig. 6. The change of Ye with time using the improved weak interaction rates (taken
from [6])

Figure 7 displays the comparison of the central value of Ye at the onset of

collapse with the two rates FFN and LMP for the range of star masses from
11M to 40M . This clearly shows that for LMP the central Ye is higher for
all stars considered. The next two Figures 8 and 9 shows the iron core size and
the central entropy for the stars again for the two different weak rates. Now
let us remind us that the expression for the Chandrasekhar mass generalised
for finite temperature is [30]
 Weak Interaction Rates 197

MCh = 5.83Ye2 [1 + (Se )2 /(πYe )2 ] (22)

where Se is the electron entropy at the non-zero temperature.

0.450

0.445

0.440

0.435
Ye

0.430

0.425

WW
0.420 LMP

0.020

0.015
ΔYe

0.010

0.005
10 15 20 25 30 35 40
Star Mass (M)

Fig. 7. Comparison of the central value of Ye using the FFN and LMP rates; the
lower panel shows the change in the values. The most important nuclei at each stage
is also given (taken from [6])

In Figure 9 it is seen that for star masses upto 20M the central entropy
is lower for the LMP rates. So in this mass range the increase in the Chan-
drasekhar mass due to higher Ye gets neutralised by the lower Se and no strong
effect is seen upto 20M . On the other hand the mass of the homologous core
which depends on Ye2 at the time of neutrino trapping is expected to increase
for these lower mass stars with LMP rates. The shock energy also increases
with Ye and with less mass to travel through, the new rates are supposed to
help the shock propagation for this mass range. But this effect may not be
enough to result in ‘prompt’ explosions. But more studies on numerical stellar
evolution both at the preSN and the SN stages with realistic weak interaction
rates are needed.
198 Kamales Kar

2.0

1.8
MFe (M)

1.6

1.4 WW
LMP

0.1
ΔMFe (M)

0.0

– 0.1

– 0.2
10 15 20 25 30 35 40
Star Mass (M)

Fig. 8. Comparison of the iron core sizes using the FFN and LMP rates; the lower
panel shows the change in the values (taken from [6])

6 Weak Interaction Processes During pp-Chain

and Solar Neutrino Problem
In the pp-chain i.e. the first stage of nuclear burning there are a number of
reactions that produce neutrinos. We list them below:

Neutrinos produced in the fusion reactions in the sun (pp chain):

1. p + p →2 H + e+ + νe (pp) [Eν < 0.42 MeV]

2. p + e− + p →2 H + νe (pep) [Eν = 1.442 MeV]

3. 2 H + p →3 H + γ

4. 3 He +3 He → α + 2p

5. 3 He +4 He →7 Be + γ
 Weak Interaction Rates 199

1.2

1.1

central entropy / baryon (kB)

1.0

0.9

0.8

0.7

WW
0.6
LMP

0.1

0.0
ΔS (kB)

–0.1

–0.2
10 15 20 25 30 35 40
Star Mass (M)

Fig. 9. The central entropy for 11-40M star mass using the LMP rates compared
to the same using the old FFN rates; the lower panel shows the change in the values
(taken from [6])

6. 3 He + p → α + e+ + νe (hep) [Eν < 18.77 MeV]

7. 7 Be + e− →7 Li + νe [Eν = 0.861 MeV(90%) or 0.331 MeV(10%)]

8. 7 Be + p →8 B + γ

9. 8 B →8 Be∗ + e+ + νe [Eν = 15 MeV]

10. 8 Be∗ → 2α

The reactions all put together give rise to the following fusion reaction:

4p → α + 2e+ + 2νe + 25M eV (23)

The spectra of the neutrinos produced in these reactions are given in Figure 10.
200 Kamales Kar

Fig. 10. The spectrum of the neutrinos produced by different reactions in the pp-
chain in the sun (taken from [33])

The energy threshold of the detectors with chlorine (Homestake), water

(Kamiokande/SuperK) and gallium (GALLEX/SAGE) are 0.814 MeV, 7.5
MeV and 0.233 MeV respectively. So the neutrinos produced by the first re-
action, known as the ‘pp’ reaction, with a maximum energy of 0.42 MeV
were not seen with chlorine or water and first detected by the gallium detec-
tors. Now for all these three reactions the number of neutrinos detected were
substantially lower, by a factor of 2-3 than the predictions of theory based
on the Standard Solar Model (SSM) [32]. This is seen clearly in Figure 11.
On the other hand for the Sudbury Neutrino Observatory (SNO) though the
charged current reaction (CC) (d + νe → p + p + e− ) with a threshold of 5
MeV sees only 35% of the predicted events, but the neutral current events
(NC) (d + νx → p + n + νx ) with a threshold of 2.2 MeV is able to register
all the neutrinos. This proved once for all that the solar neutrinos in their
transit get converted from the electron flavor to other flavors but their to-
tal flux remains unchanged. As beyond 5 MeV the neutrinos come almost
fully from the 8 B decay one gets from the SNO NC results, the neutrino flux
ΦN C = [5.09 ± 0.44(stat)−0.43
+0.46
(sys)] × 106 cm−2 s−1 .

6.1 Neutrino Oscillation

The mechanism of conversion of neutrinos from one flavor to another is seen

to be the neutrino flavor oscillation. We briefly describe below how in a two
 Weak Interaction Rates 201

SAGE GALLEX
+ SNO SNO
Superk GNO
Kamiokande

Fig. 11. The total rates as predicted by theory for the different detectors compared
with the observed rates (taken from [33])

flavor scenario in vacuum the neutrino oscillation occurs. Let the two flavor
eigenstates be |νe > and |νμ > whereas the mass eigenstates are |ν1 > and
|ν2 >. If the oscillation angle is θ then we have
    
νe cosθ sinθ ν1
= (24)
νμ −sinθ cosθ ν2

As the mass eigenstates evolve in time as e−iEt one gets

|νe (t) >= cosθ e−iE1 t |ν1 > +sinθ e−iE2 t |ν2 > (Ei2 = m2i + p2 ) (25)

leading to
Pνμ νe = sin2 (2θ) sin2 (Δm2 L/(4E)) (26)
This shows that the probability of a electron type neutrino getting converted
into a mu type neutrino has an oscillatory dependence on (L/E) where L is the
length travelled and E is the energy. The two mass eigenstates have the same
momentum here so that the energies are given by Ei2 = m2i + p2 . The mass
square difference is Δm2 = m22 − m21 . With three neutrino flavors instead of
two, one has three oscillation angles and two mass squared differences but the
probabilities can be similarly be expressed in terms of the oscillation angles
and the mass squared differences.
When there is matter in the travel path of the neutrinos the νe s interact
with the electrons in the matter and give rise to an extra potential (propor-
tional to ne ) in the mass matrix. This results in matter-enhanced neutrino
202 Kamales Kar

oscillation (MSW effect). The observed depletion of the solar electron type
neutrino by different detectors has been explained using certain mass dif-
ferences and oscillation angles [34]. The allowed region in the mass squared
difference and oscillation angle with a global fit to the solar and reactor experi-
ments all put together is shown in Figure 12. This shows how the uncertainties
got reduced with more precise experimental input in 2004.

20
All free + Lum. constraint All free + Lum. constraint
18
Δm212 (10–5 eV2)

16
14
12
10
8
6
4 Before Nu2004 After Nu2004

0.2 0.4 0.6 0.2 0.4 0.6 0.8

tan2θ12 tan2θ 12
Fig. 12. More accurate determination of the oscillation parameters with time (taken
from [33])

7 β-decay Rates for r-Process Nucleosynthesis

One of the most basic questions in nuclear astrophysics is: How do the nuclei
heavier than iron get produced? This question was first answered by Burbidge,
Burbidge, Fowler and Hoyle in 1957 [35]. They proposed that these elements
are produced through the slow (s) and rapid (r) neutron-capture processes.
The words rapid/slow refer to the rate of neutron capture compared to the
rate of β-decay in the astrophysical conditions. Figure 13 shows the nuclei
involved in the r- and s-processes. The s-process path stays close to the valley
of stability whereas the r-process path moves staying close to the drip line.
The figure also shows the nuclei involved in the rp-process; these are proton
rich nuclei where capture of protons are involved and that the rate is compared
to the β + rates.
 Weak Interaction Rates 203

Fig. 13. The s- and r-process path and the nuclear chart

7.1 s-process and r-process

For the s-process nuclei the neutrons get captured by one stable nucleus going
to another stable nucleus and so on. Once an unstable isotope is reached,
there is a β-decay. So the yield Ys (A) of these nuclei can be expressed as
dYs (A)/dt = φn [σn,γ (A − 1)Ys (A − 1) − σn,γ (A)Ys (A)] (27)
At equilibrium then one gets Ys (A)/Ys (A − 1) = σn,γ (A − 1)/σn,γ (A). As the
n-capture rates are very small for magic neutron number nuclei like 88 Sr
(N =50), 138 Ba (N =82), 208 Pb (N =82) that explains the observed large abun-
dances for shell closure nuclei.
For the rapid process, if one takes into account the different competing
processes the general equation for the evolution of the yield is given by [36]

dY (Z, A)/dt = nn < vσn,γ (Z, A − 1) > Y (Z, A − 1) + λγ,n (Z, A + 1)Y (Z, A + 1)
+λβ0 (Z − 1, A)Y (Z − 1, A) + λβ1 (Z − 1, A + 1)Y (Z − 1, A + 1)
+λβ2 (Z − 1, A + 2)Y (Z − 1, A + 2) + λβ3 (Z − 1, A + 3)Y (Z − 1, A + 3)
−nn < vσn,γ (Z, A) > Y (Z, A) − λγ,n (Z, A)Y (Z, A)
−[λβ0 (Z, A) + λβ1 (Z, A) + λβ2 (Z, A) + λβ3 (Z, A)]Y (Z, A) (28)

where nn is the neutron number density, nn < vσn,γ (Z, A) > is the ther-
mally averaged neutron capture rate and λβn (Z, A) is the rate for β-decay
204 Kamales Kar

followed by the emission of ‘n’ neutrons. Here the charged particle reactions
are considered unimportant. When the n−capture and photodisintegration
occur much faster than the β-decay, one gets for equilibrium

Y (Z, A + 1)/Y (Z, A) = nn < vσn,γ (Z, A) > /λγ,n (Z, A + 1) (29)

Then using the expressions for the averaged (n, γ) and (γ, n) processes one
sees that the most abundant isotopes in different isotopic chains have approxi-
mately the same separation energy of about 2-3 MeV for r-process conditions.
Later network calculations get the equilibrium at neutron density greater than
1020 cm−3 .
It was first seen that (n, γ) ↔ (γ, n) equilibrium is obtained at T> 2 × 109
K and Nn > 1020 cm−3 [37]. Later network calculations get the equilibrium
at neutron density of 1020 cm−3 for temperatures 2 × 109 K and at a density
of 1028 cm−3 for half the temperature. One also notes that at temperatures
greater than 6 × 109 K Nuclear Statistical Equilibrium (NSE) is achieved
where the forward strong and electromagnetic reactions are balanced by their
reverse reactions.
In this description Fission Cycling has not been taken into account. But if
the r-process involves nuclei with Z > 80, then fission should be included. In
that case the heaviest nucleus fissions and a cyclic flow occurs between this
nucleus and its fission fragments in the presence of large neutron abundance.
The solar r-process abundances are normally obtained by subtracting the
calculated s-process abundances from the observed solar abundances. A typi-
cal example is given in Figure 14.

7.2 Possible r-process site

Curiously enough no r-process site is clearly identified yet. But, of course,

there are good candidates. A neutron rich environment is the primary necessity
for the r-process. The delayed neutrino driven core collapse is considered to
be one of more promising ones. The neutrino driven wind starting from the
protoneutron star surface is hot (temperatures more than 1010 K) and has low
density and high entropy. An entropy reaching a value of 400 kB per baryon
in about ten seconds showed the onset of r-process [38] in one dimensional
simulation. But another calculation got entropy much lower and not showing
the development of the r-process [40]. A better understanding of the neutrino
driven wind is required to establish that as a r-process site. Several analytical
models to treat this wind exist in literature. Among the spherically symmetric
steady-state models both Newtonian adiabatic and general relativistic versions
exist. There are also numerical models available [39]. On the other hand there
are claims that the heavy r-nuclei with mass number A > 130 cannot be
produced in the neutrino-driven winds [41].
The possibility of r-process in accretion induced delayed core collapse SN
has also been studied [42]. The other possibility lies in the neutron star merger
 Weak Interaction Rates 205

a 0.5

–0.5
log (Abundance)

–1

–1.5

–2

–2.5
90 110 130 150 170 190 210
Mass Number (A )

Fig. 14. The r-processes abundances by subtracting the s-process abundances from
the observed values (taken from [36])

models for the r-process. During the merger of two neutron stars or a neutron
star and a black hole decompression of the cold neutron star matter takes
place which can be considered as a site for high density r-process. But this
has the drawback that the rates of the mergers are much less compared to
core collapse SN rates and it has difficulty in reproducing the abundances of
nuclei in the mass region 90 - 110.

7.3 Models for calculation of β-decay rates for r-process nuclei

For the r-process the models for calculating β-decay rates can again be di-
vided into microscopic and statistical categories. Among the microscopic ones
shell model is of limited use as this involves very neutron-rich nuclei all over
the periodic table. Beyond the f p-shell nuclei shell model has been applied
to nuclei with either a few valence particles or with more particles but with
not too many valence orbits. The microscopic theory that has been widely
used is the ‘Random Phase Approximation’ (RPA) and its different improved
version. We refer here to the review by Arnould, Goriely and Takahashi [39]
for a detailed description and references. The effective nucleon-nucleon inter-
action is often taken to be of the spin-isospin type (στ .στ ) and one builds
206 Kamales Kar

the particle-hole excitations of the charge exchange mode. The Quasi-particle

RPA to take into account the pairing along with the models for masses like
the extension of the droplet model, FRDM [43] has also been used extensively.
Among the statistical models one uses the ‘Gross Theory’ or the ‘Semi-
gross Theory’ described before. Here we shall briefly mention the beginning
of another statistical approach following the spectral distribution methods
applied earlier for the presupernova problem. In this treatment [44, 45], the
GT sum rule strength has contributions from the low-lying states and the giant
resonance. Nuclei involved are divided into two categories- one set for which
some experimentally measured values of log ft are available in the ground state
region and the other set, normally the nuclei with larger neutron fractions,
where no experimental values of log ft exists. The sharp Fermi Resonance
above the IAS has very little contribution but is taken into account. The GT
resonance is again taken as a Gaussian with the strength centroid given by
the Bertsch and Esbensen expression [46]

EGT − = EIAS + ΔEs.o. + 2[[kστ SGT − ]/3 − (N − Z)kτ ] (30)

Here the ΔEs.o. is the contribution coming from the spin-orbit interaction and
one takes kτ = 28.5/A MeV and kστ = 23/A MeV [29]. The method has been
used to calculate the β-decay rates of nuclei in the range 115 < A < 140. The
width of the GT giant resonance is set to give a best fit to the observed half-
lives. The two sets of nuclei with known values of log ft and no known ones
are treated separately and the best fit value for the width for the second case
naturally comes out larger. The calculated and the experimental half-lives are
given for all the nuclei in Figure 14. The figure also give a straight line fit of
half-life as a function of Q−value given by

λβ − = ln2/t1/2 = 10−4 × (Qβ − /M eV )5 (31)

where t1/2 is the half-life and the Q value is expressed in MeV. This line
actually represents the case where the GT matrix element has a fixed value
independent of energy [39]. The rates are then calculated for typical r-process
densities and temperatures and the method is used to predict half-lives of
nuclei very close to the drip line where no experimental information available.

8 Concluding Remarks
This article discusses the importance of the weak interaction rates in differ-
ent astrophysical processes. Though a lot of effort has gone in to calculate
these rates accurately, more work is still needed, particularly for the highly
neutron-rich nuclei. Radioactive ion beam experiments are also expected to
give valuable information for these nuclei. So one hopes that these will help
in a more accurate description of these processes in the near future.
The author thanks Palash B. Pal for help in preparing the manuscript.
 Weak Interaction Rates 207
1000

100

10
log(t(1/2) / sec)

0.1

0.01

0.001
10
log(Q / MeV)

Fig. 15. The half-lives of nuclei with 115 < A < 140 calculated by the statistical
method [45] compared with a form with only the Q5 dependence

References
1. A. deShalit, H. Feshbach, Theoretical Nuclear Physics Vol I, John Wiley and
Sons (New York) (1974)
2. C. S. Wu, S. Mozkowski, Beta Decay, Inter Science Publishers, New York (1966)
3. H. A. Bethe, G. E. Brown, Sc. American 25, 60 (1985)
4. H. A. Bethe, Rev. Mod. Phys. 60, 801 (1990)
5. H-Th. Janka, K. Langanke, A. Marek, G. Martinez-Pinedo, B. Muller, Phys.
Rep. 442, 38 (2007)
6. K. Langanke, G. Martinez-Pinedo, Rev. Mod. Phys. 75, 819 (2003)
7. G. G. Raffelt, M. Th. Keil, R. Buras, H.-Th. Janka, M. Rampp, astro-
ph/0303226; Proc. 4th Workshop on Neutrino Oscillations and their Origin
NOON (2003)
8. F. S. Kitaura, H.-Th. Janka, W. Hillebrandt, A&A 450, 345 (2006)
9. J. M. Blondin, A. Mezzacappa, C. Demarino, ApJ 584, 971 (2004)
10. G. M. Fuller, ApJ 252, 741 (1982)
11. H. A. Bethe, J. R. Wilson, ApJ 295, 14 (1985)
12. A. Ray, K. Kar, ApJ 319, 143 (1987)
13. W. C. Haxton, Phys. Rev. Lett. 60, 1999 (1988)
208 Kamales Kar

14. G. M. Fuller, W. A. Fowler, M. J. Newman, ApJS 42, 447 (1982); 48, 279
(1975); ApJ 252, 715 (1982); 293 1 (1985)
15. M. B. Aufderheide, G. E. Brown, T. T. S. Kuo, D. B. Stout, P. Vogel, ApJ 362,
241 (1990)
16. K. Takahashi, M. Yamada, Prog. Theor. Phys. 41, 1470 (1969)
17. K. Takahashi, M. Yamada, T. Kondoh, At. Data Nucl. Tables 12, 101 (1973)
18. T. Tachibana, M. Yamada, Y. Yoshida, Prog. Theor. Phys. 84, 641 (1990)
19. H. Nakata, T. Tachibana, M. Yamada, Nucl. Phys. A 625, 521 (1997)
20. J. B. French, V. K. B. Kota, Ann. Rev. Nuc. Part. Sci. 32, 35 (1982)
21. V. K. B. Kota, K. Kar, Prmana- J. Phys. 32, 647 (1989)
22. K. Kar, Nucl. Phys. A 368, 285 (1981)
23. K. Ikeda, S. Fujii, J.I. Fujita, Phys. Lett. 3, 271 (1963)
24. K. Kar, A. Ray, S. Sarkar, ApJ 434, 662 (1994)
25. K. Nakayama, A. Pio Galeao, F. Krmpotic, Phys. Lett. B 114, 217 (1982)
26. D. Majumdar, K. Kar, Pramana- Jour. Phys. 68, 423 (2007)
27. K. Langanke, G. Martinez-Pinedo, At. Nucl. Data Tables 79, 1 (2001)
28. E. Caurrier, G. Martinez-Pinedo, F. Nowacki, A. Poves, J. Retamosa, A. P.
Zuker, Phys. Rev. C 59, 2033 (1999)
29. J. Pruet, G. M. Fuller, ApJS 149, 189 (2003)
30. A. Heger, K. Langanke, G. Martinez-Pinedo, S. E. Woosley, ApJ 560, 307 (2001)
31. S. E. Woosley, T. A. Weaver, ApJS 101, 181 (1995)
32. J. N. Bahcall, A. Serenelli, S. Basu, ApJ 621, l85 (2005)
33. Homepage of John Bahcall: http://www.sns.ias.edu/ jnb/
34. A. Bandyopadhyay, S. Choubey, S. Goswami, K. Kar, Phys. Lett. B 519, 83
(2001)
35. E. M. Burbidge, G. M. Burbidge, W. A. Fowler, F. Hoyle, Rev. Mod. Phys. 29,
547 (1957)
36. Y.Z. Qian, Prog. Part. Nucl. Phys. 50, 153 (2003); astro-ph/0301422
37. A. G. Cameron, J. J. Cowan and J. W. Truran, Ap&SS 91, 235 (1983)
38. S. E. Woosley, J. R. Wilson, G. J. Mathews et al., ApJ 433, 229 (1994)
39. M. Arnould, S. Goriely, K. Takahashi, Phys. Rep. 450, 97 (2007)
40. K. Takahashi, J. Witti, H.-Th. Janka, A&A 286, 857 (1994)
41. Y. Z. Qian, G. J. Wasserburg, Phys. Rep. 442, 237 (2007)
42. L. Dessart, A. Burrows, C. D. Ott et al., ApJ 644, 1063 (2006)
43. P. Moller, J. Nix, W. D. Myers, W. J. Swiatecki, At. Data Nucl. Data Tables
59, 185 (1995)
44. K. Kar, S. Chakravarti, V. R. Manfredi, Pramana -J. Phys. 48, 123 (2007)
45. K. Kar, S. Chakravarti (unpublished)
46. G. F. Bertsch, H. Esbensen, Phys. Prog. Phys. 50, 607 (1987)
Massive stars as thermonuclear reactors
and their explosions following core collapse

Alak Ray

Tata Institute of Fundamental Research, Mumbai 400 005, India [email protected]

Summary. Nuclear reactions transform atomic nuclei inside stars. This is the pro-
cess of stellar nucleosynthesis. The basic concepts of determining nuclear reaction
rates inside stars are reviewed. How stars manage to burn their fuel so slowly most
of the time are also considered. Stellar thermonuclear reactions involving protons
in hydrostatic burning are discussed first. Then I discuss triple alpha reactions in
the helium burning stage. Carbon and oxygen survive in red giant stars because
of the nuclear structure of oxygen and neon. Further nuclear burning of carbon,
neon, oxygen and silicon in quiescent conditions are discussed next. In the subse-
quent core-collapse phase, neutronization due to electron capture from the top of
the Fermi sea in a degenerate core takes place. The expected signal of neutrinos
from a nearby supernova is calculated. The supernova often explodes inside a dense
circumstellar medium, which is established due to the progenitor star losing its out-
ermost envelope in a stellar wind or mass transfer in a binary system. The nature
of the circumstellar medium and the ejecta of the supernova and their dynamics are
revealed by observations in the optical, IR, radio, and X-ray bands, and I discuss
some of these observations and their interpretations.

Keywords: nuclear reactions, nucleosynthesis, abundances; stars: interiors;

supernovae: general; neutrinos; circumstellar matter; X-rays: stars

1 Introduction
The sun is not commonly considered a star and few would think of stars as
nuclear reactors. Yet, that is the way it is, and even our own world is made
out of the “fall-out” from stars that blew up and spewed radioactive debris
into the nascent solar system.
Nuclear Astrophysics is the field concerning “the synthesis and Evolution
of atomic nuclei, by thermonuclear reactions, from the Big Bang to the present.
What is the origin of the matter of which we are made?”[1]. Our high entropy
universe resulting from the Big Bang, contains many more photons per particle

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 5, c Springer-Verlag Berlin Heidelberg 2010
210 A. Ray

of matter with mass, e.g. protons and neutrons. Because of the high entropy
as the universe expanded, there was time to manufacture elements only upto
helium and the major products of cosmic nucleosynthesis remained the light
elements hydrogen and helium1 . Stars formed from this primordial matter.
They used these elements, hydrogen and helium as fuel to generate energy like
a giant nuclear reactor2 . In the process, the stars could shine and manufacture
higher atomic number elements like carbon, oxygen, calcium and iron which
we and our world are made of. The heavy elements are either dredged up from
the core of the star to the surface of the star from which they are dispersed by
stellar wind or directly ejected into the interstellar medium when a (massive)
star explodes. The stardust is the source of heavy elements for new generation
of stars and sun-like systems.
Our sun is not a massive star. It burns hydrogen in a set of nuclear re-
actions called the pp-chain, whereas the more massive stars presently burn
their hydrogen by the so-called CNO-cycle3 . Nevertheless, to put nuclear re-
actions in stars in perspective, we shall start with a discussion of how these
reactions proceed within the sun. There is a correspondence between the evo-
lutionary state of a star, its external appearance4 and internal core conditions
and the nuclear fuel it burns, – a sort of a mapping between the astronomers
Hertzsprung-Russel diagram and the nuclear physicist’s chart of the nuclide
[4], until nuclear burning takes place on too rapid a time scale.
The problem of evolution of stars to their explosion and subsequent in-
teraction with the circumstellar medium has many time scales (ranging from
tens of milliseconds to tens of thousands of years) and macroscopic length
scales (from dimensions effectively that of a white dwarf to that of a super-
nova remnant, i.e from few thousand kilometers to many tens of light years).
The physics of supernova explosions is complex and diverse and in many parts,
the explosion mechanism is still an unsolved problem. Even the constraining
parameters and ingredients which makes the SN explode are still controversial
(see e.g. the discussion in [96]). It is possible that the identification of the key

1
Note however suggestions [5,6] that early generation of stars called Pop III objects
can also contribute to the abundance of 4 He seen in the universe today and the
entire helium may not be a product of big bang nucleosynthesis alone.
2
Our sun is slowly burning hydrogen into helium and is presently not exactly
the same when it just started burning hydrogen. It will appear different once it
exhausts all hydrogen it can burn in its core. In other words, nuclear reactions in
stellar interiors determine the life-cycle of stars, apart from providing them with
internal power for heat and light and manufacturing all the heavier elements that
the early universe could not.
3
Note however that CN cycle may have driven an early stage convection in the
young Sun.
4
Astronomers classify stars according to their colors and (absorption) line spectra
and luminosities. Meghnad Saha showed [2, 3] the link between the classification
scheme and temperature (and thermal ionization) of stellar atmosphere.
 Thermonuclear fusion and collapse and explosion in massive stars 211

aspects in the explosion may require seminal observations about the conditions
in the supernova core other than the indirect evidence such as explosion asym-
metries, pulsar kicks or nucleosynthetic yields. Such observations may involve
the future detection of strong neutrino signals and gravitational waves from
a galactic supernova in future. Detectable neutrino signals from a supernova
was seen in the case of SN 1987A, but since that target was in a neighboring
satellite galaxy (the Large Magellanic Cloud), the number of neutrinos were
too small to distinguish the fine points of theoretical issues.
Since nuclear astrophysics is not usually taught at the master’s level nu-
clear physics specialization in our universities, these lecture notes are meant
to be an introduction to the subject and a pointer to literature and Inter-
net resources5. The emphasis in the first part of these lecture notes is on the
nuclear reactions in the stars and how these are calculated, rather than how
stars evolve. The latter usually forms a core area of stellar astrophysics.
This article is organized essentially in the same sequence that a massive
star burns successively higher atomic number elements in its core, until it col-
lapses and explodes in a supernova. The introductory part discusses how the
rates of thermonuclear reactions in (massive) stars are calculated, what the
different classes of reactions are and how the stars (usually) manage to burn
their fuels so slowly6 . The middle part describes the nuclear physics during
the collapse phase of the massive star. The last part describes a few typi-
cal examples of what can be learned by optical, IR and X-ray studies about
nucleosynthesis and dynamics of explosion in supernovae and supernova rem-
nants such as Cassiopeia A, SN 1987A etc. Only core-collapse supernovae
are discussed in these lectures, those that arise from massive stars (e.g. stars
more massive than 8M with typical solar metallicity at the time they start
5
See for example, [7] for a course of nuclear astrophysics, and the International
Conference Proceedings under the title: “Nuclei in the Cosmos” for periodic re-
search conferences in the field. Valuable nuclear astrophysics datasets in machine
readable formats can be found at sites: [8], [9]. A new and updated version of
the nuclear reactions rate library REACLIB for astrophysics is now being main-
tained as a public, web-based version at [10]. A complementary effort to develop
software tools to streamline the process of getting the latest and best informa-
tion into this new library is available at [11] (see [12]). Much of the material
discussed in the first part of these notes can be found in textbooks in the sub-
ject, see e.g. [13], [14], [4], [15], [1] etc. There is also a recent book on the
subject by Richard Boyd [16] that among other topics describes terrestrial and
space born instruments operating in service to nuclear astrophysics. A Workshop
on Solar fusion cross sections for the pp chain and CNO cycle held in 2009 by
the Institute of Nuclear Theory is expected to result in a Reviews of Modern
Physics article. Supernovae of various types are the sites where nuclear reactions
in stars or explosions are of prime importance. For Internet resources to two
recent Schools on these topics, see http://icts.tifr.res.in/sites/Sgrb/Programme
and http://www.tifr.res.in/∼sn2004.
6
These issues were discussed in an earlier SERC School [17].
212 A. Ray

burning hydrogen in their cores, i.e. at the “Zero Age Main Sequence”, before
any mass was lost from their surface). We shall not discuss the type Ia SNe7
in these lectures. Abundance of elements in our galaxy Milky Way give im-
portant information about how stars affect the element and isotopic evolution
in various parts of the galaxy. For a study of the evolution of elements from
C to Zn in the galactic halo and disk, with yields of massive stars and type
Ia SNe, see [20] and elsewhere in these Proceedings.

2 Stars and their thermonuclear reactions

While referring to Sir Ernest Rutherford “breaking down the atoms of oxy-
gen and nitrogen”, Arthur Eddington remarked: “what is possible in the
Cavendish Laboratory may not be too difficult in the sun” [21]. Indeed this
is the case, but for the fact that a star does this by fusion reactions, rather
than by transfer reactions, – in the process giving out heat and light and
manufacturing fresh elements. Of all the light elements, hydrogen is the most
important one in this regard, because: a) it has a large universal abundance,
b) considerable energy evolution is possible with it because of the large bind-
ing energies of nuclei that can be generated from its burning and c) its small
charge and mass allows it to penetrate easily through the potential barriers
of other nuclei. A star burns its fuel in thermonuclear reactions in the core
where the confinement of the fuel is achieved in the star’s own gravitational
field. These reactions remain “controlled”, or self-regulated8 , as long as the
stellar material remains non-degenerate.
7
The supernovae are classified by astronomers on the basis of their optical spectra
at the time of their maximum light output. Those that do not show the presence
of hydrogen in their spectra are classified as type I SNe. A subclass of them, type
Ia’s are believed to arise from thermonuclear explosions in the electron degenerate
cores of stars less massive than 8M and are very useful to map the geometry of
our universe, because they serve as calibratable “standard” candles. These “ther-
monuclear supernovae” are usually more luminous in the optical bands than the
core-collapse varieties, but while the former class put out several MeVs of energy
per nucleon, the core-collapse SNe or ccSNe, emit several hundreds of MeVs per
nucleon from their central engines (mostly in down-scattered neutrinos). Apart
from the missing hydrogen lines, the type Ia SNe show an absorption “trough”
in their spectra around maximum light at 6150 Å, due to blue shifted Si II lines
[18]. Of the other type I SNe which show no hydrogen and no Si II trough, some
show helium in their early spectra (type Ib) and others have little or no helium
and have a composition enhanced in oxygen (type Ic) [19]. These, (Ib and Ic)
together with the type IIs constitute the core collapse SNe.
8
There are however examples to the contrary when thermonuclear reactions take
place in an explosive manner, e.g. when a whole white dwarf (resulting from an
evolved intermediate mass star) undergoes merger with another and explodes, as
nuclear fuel (carbon) is ignited under degenerate conditions, such as in a type Ia
supernova; explosive thermonuclear reactions also take place in post-bounce core
 Thermonuclear fusion and collapse and explosion in massive stars 213

The recognition of the quantum mechanical tunneling effect prompted

Atkinson and Houtermans [22] to work out the qualitative treatment of en-
ergy production in stars. They suggested that the nucleus serves as both a
cooking pot and a trap. Binding energy difference of four protons and two
electrons (the nuclear fuel) and their ash, the helium nucleus, some 26.7 MeV
is converted to heat and light that we receive from the sun9 . The photons in
the interior are scattered many a times, for tens of millions of years, before
they reach the surface of the sun. Some of the reactions produce neutrinos,
which because of their small cross-section for interaction, are not stopped or
scattered by overlying matter, – but stream out straight from the core. Neu-
trinos are thus the best probes of the stellar core [23, 106], while the photons
bear information from their surface of last scattering – the photosphere.

2.1 Why do the stars burn slowly: a look at Gamow peaks

The sun has been burning for at least 4.6 billion years10. How does it manage
to burn so slowly11 ? Under the ambient conditions in the core, the relevant

collapse supernovae, when the hydrodynamic shock ploughs through unburnt Si-
or O-layers in the mantle.
9
Lord Kelvin surmised in the nineteenth century that the solar luminosity is
supplied by the gravitational contraction of the sun. Given the solar luminosity,
this immediately defined a solar lifetime (the so-called Kelvin-Helmholtz time):
τKH = GM 2
/R L ∼ few × 107 yr. This turned out to be much shorter than
the estimated age of the earth at that time known from fossil records, and led
to a famous debate between Lord Kelvin and the fossil geologist Cuvier. In fact,
as noted above modern estimates of earth’s age are much longer and therefore
the need to maintain sunshine for such a long time requires that the amount
of energy radiated by the sun during its lifetime is much larger than its grav-
itational energy or its internal (thermal) energy: L × tlife  GM 2
/R . This
puzzle was resolved only later with the realization that the star can tap its much
larger nuclear energy reservoir in its core through thermonuclear reactions. The
luminosity of the sun however is determined by an interplay of atomic and grav-
itational physics that controls the opacity, chemical composition, the balance of
pressure forces against gravity, etc. Nuclear physics determines how fast nuclear
reactions go under feedback control determined by the ambient conditions.
10
Lord Rutherford [24] determined the age of a sample of pitchblende, to be 700
million years, by measuring the amount of uranium and radium and helium re-
tained in the rock and by calculating the annual output of alpha particles. The
oldest rock found is from Southwest Greenland: ≈ 3.8 Gyr old [15]. Radioactive
dating of meteorites point to their formation and the solidification of the earth
about 4.55 ± 0.07 years ago [25]. Since the sun and the solar system formed
only slightly before, their age at isolation and condensation from the interstellar
medium is taken to be 4.6 Gyr [26].
11
The Nobel prize citation of Hans Bethe (1967) who solved this problem, noted
that this “concerns an old riddle. How has it been possible for the sun to emit
light and heat without exhausting its source not only during the thousands of
214 A. Ray

thermonuclear reaction cross sections are very small12 . For reactions involving
charged particles, nuclear physicists often encounter cross-sections near the
Coulomb barrier of the order of millibarns. One can obtain a characteristic
luminosity LC based on this cross section and the nuclear energy released per
reaction [15]:

LC ∼ N ΔE/τC (1)
−2
where  ≈ 10 is the fraction of total number of solar nuclei N ∼ 1057
that take part in nuclear fusion reactions generating typically ΔE ∼ 25 MeV
in hydrogen to helium conversion. Here, the τC is the characteristic time scale
for reactions, which becomes minuscule for the cross-sections at the Coulomb
barrier and the ambient density and relative speed of the reactants etc:

1 10−8 s
τC ∼ = (2)
nσv [n/(1026 cm−3 )][σ/1 mbarn][v/109 cm s−1 ]
This would imply a characteristic luminosity of Lc ≈ 1020 L , even for a
small fraction of the solar material taking part in the reactions (i.e.  ∼ 10−2 ).
If this was really the appropriate cross-section for the reaction, the sun would
have been gone very quickly indeed. Instead the cross-sections are much less
than that at the Coulomb barrier penetration energy (say at proton energies
of 1 MeV), to allow for a long lifetime of the sun (in addition, weak-interaction
process gives a smaller cross-section for some reactions than electromagnetic
process, – see Sect. 3.1).
Stellar nuclear reactions can be either: a) charged particle reactions (both
target and projectile are nuclei) or b) neutral particle (neutron) induced re-
actions. Both sets of reactions can go through either a resonant state of an
intermediate nucleus or can be a non-resonant reaction. In the former reac-
tion, the intermediate state could be a narrow unstable state, which decays
into other particles or nuclei. In general, a given reaction can involve both

centuries the human race has existed but also during the enormously long time
when living beings needing the sun for their nourishment have developed and
flourished on our earth thanks to this source? The solution of this problem seemed
even more hopeless when better knowledge of the age of the earth was gained.
None of the energy sources known of old could come under consideration. A very
important part of his work resulted in eliminating a great number of thinkable
nuclear processes under the conditions at the center of the sun, after which only
two possible processes remained..... (Bethe) attempted a thorough analysis of
these and other thinkable processes necessary to make it reasonably certain that
these processes, and only these, are responsible for the energy generation in the
sun and similar stars.”
12
This makes the experimental verification of the reaction cross-sections a very
challenging task, requiring in some cases, extremely high purity targets and pro-
jectiles so that relevant small event rates are not swamped by other reaction
channels and products (see Rolfs and Rodney, Chapter 5 [13]).
 Thermonuclear fusion and collapse and explosion in massive stars 215

types of reaction channels. In charged particle induced reactions, the cross-

section for both reaction mechanisms drops rapidly with decreasing energy,
due to the effect of the Coulomb barrier (and thus it becomes more difficult
to measure stellar reaction cross-sections accurately). In contrast, the neu-
tron induced reaction cross-section is very large and increases with decreasing
energy (here, resonances may√ be superposed on a smooth non-resonant yield
which follows the 1/v ∼ 1/ E dependence). These reaction rates and cross-
sections can be then directly measured at stellar energies that are relevant (if
such nuclei are long lived or can be generated). The schematic dependence
of the cross-sections are shown in Fig. 1. We shall not discuss the neutron
capture elements in these notes.

NEUTRONS
RESONANCE
CROSS SECTION σ (E)

1
LAW
V
NARROW
(log scale)

RESONANCE

BROAD RESONANCE

NON–RESONANT CHARGED
PROCESS PARTICLES

wg a Γp wg a Γg
0 EC
INTERACTION ENERGY E
Fig. 1. Dependence of total cross-sections on the interaction energy for neutrons
(top panel) and charged particles (bottom panel). Note the presence of resonances
(narrow or broad) superimposed on a slowly varying non-resonant cross-section (af-
ter [13])

2.2 Gamow peak and the astrophysical S-factor

The sun and other “main-sequence” stars (burning hydrogen in their core qui-
escently) evolve very slowly by adjusting their central temperature such that
216 A. Ray

the average thermal energy of a nucleus is small compared to the Coulomb

repulsion an ion-ion pair encounters. This is how stars can live long for astro-
nomically long times. A central temperature T ≥ 107 K (or T7 ≥ 1, hereafter
a subscript x to a quantity, indicates that quantity in units of 10x) is required
for sufficient kinetic energy of the reactants to overcome the Coulomb barrier
and for thermonuclear reactions involving hydrogen to proceed at an effective
rate, even though fusion reactions have positive Q values i.e. net energy is lib-
erated out of the reactions. The classical turning point radius for a projectile
of charge Z2 and kinetic energy Ep (in a Coulomb potential VC = Z1 Z2 e2 /r,
and effective height of the Coulomb barrier EC = Z1 Z2 e2 /Rn = 550 keV for a
p + p reaction), is: rcl = Z1 Z2 e2 /Ep . Thus, classically a p + p reaction would
proceed only when the kinetic energy exceeds 550 keV. Since the number of
particles traveling at a given speed is given by the Maxwell Boltzmann (MB)
distribution φ(E), only the tail of the MB distribution above 550 keV is effec-
tive when the typical thermal energy is 0.86 keV ( T9 = 0.01). The ratio of the
tails of the MB distributions: φ(550 keV)/φ(0.86 keV) is quite minuscule, and
thus classically at typical stellar temperatures this reaction will be virtually
absent.
Although classically a particle with projectile energy Ep cannot penetrate
beyond the classical turning point, quantum mechanically, one has a finite
value of the squared wave function at the nuclear radius Rn : |ψ(Rn )|2 . The
probability that the incoming particle penetrates the barrier is:

|ψ(Rn )|2
P = (3)
|ψ(Rc )|2
where ψ(r) are the wave-functions at corresponding points. Bethe [27]
solved the Schrodinger equation for the Coulomb potential and obtained the
transmission probability:-
 
 tan−1 (Rc /Rn − 1)1/2 Rn 
P = exp − 2KRc − (4)
(Rc /Rn − 1)1/2 Rc
with K = [2μ/h̄2 (Ec − E)]1/2 . This probability reduces to a much simpler
relation at the low energy limit: E  Ec , which is equivalent to the clas-
sical turning point Rc being much larger than the nuclear radius Rn . The
probability is:

μ
P = exp(−2πη) = exp[−2πZ1 Z2 e2 /(h̄v)] = exp[−31.3Z1Z2 ( )1/2 ] (5)
E
where in the second equality, μ is the reduced mass in Atomic Mass Units
and E is the center of mass energy in keV. The exponential quantity involving
the square brackets in the second expression is called the “Gamow factor”.
The reaction cross-section between particles of charge Z1 and Z2 has this
exponential dependence due to the Gamow factor. In addition, because the
 Thermonuclear fusion and collapse and explosion in massive stars 217

cross-sections are essentially “areas”: proportional to π(λ/2πh̄)2 ∝ 1/E, it

is customary to write the cross-section, with these two energy dependences
filtered out:

exp(−2πη)
σ(E) = S(E) (6)
E
where the factor S(E) is called the astrophysical S-factor. The S-factor
may contain degeneracy factors due to spin, e.g. [(2J + 1)/(2J1 + 1)(2J2 + 1)]
as reaction cross-sections are summed over final states and averaged over ini-
tial states. Because the rapidly varying parts of the cross-section (with energy)
are thus filtered out, the S-factor is a slowly varying function of center of mass
energy, at least for the non-resonant reactions. It is thus much safer to ex-
trapolate S(E) to the energies relevant for astrophysical environments from
the laboratory data, which is usually generated at higher energies (due to
difficulties of measuring small cross-sections), than directly extrapolating the
σ(E), which contains the Gamow transmission factor (see Fig. 2). Addition-
ally, in order to relate σ(E) and S(E), quantities measured in the laboratory
to these relevant quantities in the solar interior, a correction factor f0 due to
the effects of electron screening needs to be taken into account [28].
In the stellar core with a temperature T, reacting particles have many
different velocities (energies) according to a Maxwell Boltzmann distribution:-
 3/2  
μ μv2
φ(v) = 4πv 2 exp − ∝ E1/2 exp[−E/kT] (7)
2πkT 2kT
Nuclear cross-section or the reaction rates which also depend upon the
relative velocity (or equivalently the center of mass energy) therefore need to
be averaged over the thermal velocity (energy) distribution. Therefore, the
thermally averaged reaction rate per particle pair is:

 ∞  ∞
 8 1/2 1
< σv >= φ(v)σ(v)vdv = σ(E)E exp(−E/kT)dE
0 πμ (kT )3/2 0
(8)
The thermally averaged reaction rate per pair is, utilizing the astrophysical
S-factor and the energy dependence of the Gamow-factor:
 ∞
 8 1/2 1  E b 
< σv >= 3/2
S(E)exp − − √ dE (9)
πμ (kT ) 0 kT E
with b2 = EG = 2μ(πe2 Z1 Z2 /h̄)2 = 0.978μZ12Z22 MeV, EG being called
the Gamow energy. Note that in the expression√for the reaction  rate above,
at low energies, the exponential term exp(−b/ E) = exp(− (EG /E)) be-
comes very small whereas at high energies the Maxwell Boltzmann factor
E1/2 exp(−E/kT) vanishes. Hence there would be a peak (at energy, say, E0 )
of the integrand for the thermally averaged reaction rate per pair (see Fig. 3).
The exponential part of the energy integrand can be approximated as:
218 A. Ray

Cross-Section σ (E)
(log scale)

Coulomb
barrier

Measurements

Extropolation
(linear scale)
S(E) factor

Coulomb
Lowest EL of
barrier
direct measurement

Energy E
Fig. 2. Cross-section and astrophysical S-factor for charged particle reactions as a
function of beam energy. The effective range of energy in stellar interiors is usually
far less than the Coulomb barrier energy EC or the lower limit EL of laboratory
measurements. The y-scale is logarithmic for cross-section but linear for S-factor;
thus the cross section drops sharply in regions of astrophysical interest, while the
change is much less severe for the S-factor. The extrapolation of laboratory data to
lower energies relevant for astrophysical situations is more reliable for S-factor

 
 E   E − E 0 2
exp − − bE−1/2 ∼ C exp − (10)
kT Δ/2
where
−1/2
C = exp(−E0 /kT − bE0 ) = exp(−3E0 /kT) = exp(−τ )
2 1
E0 = (bkT /2) 3 = 1.22keV(Z21 Z22 μT26 ) 3

1 1
Δ = 4(E0 kT /3) 2 = 0.75keV(Z21 Z22 AT56 ) 6

Since most stellar reactions happen in a fairly narrow band of energies,

S(E) will have a nearly constant value over this band averaging to S0 . With
this, the reaction rate per pair of particles, turns out to be:
 Thermonuclear fusion and collapse and explosion in massive stars 219

Maxwell Gamow
Boltzmann Peak
Tunneling
α exp – E through
KT coulomb
barrier

ΔEO α exp EG
Probability

kT
EO
Energy E

Fig. 3. The Gamow peak is a convolution of the energy distribution of the Maxwell
Boltzmann probability and the quantum mechanical Coulomb barrier transmission
probability. The peak in the shaded region near energy E0 is the Gamow peak that
gives the highest probability for charged particle reactions to take place. Usually the
Gamow peak is at a much higher energy than kT , and in the figure the ordinate scale
(for the Gamow peak) is magnified with respect to those of the M-B and barrier
penetration factors. See also Table 1

 ∞
 2  32 S0 E−E0 2 S0
< σv >= e−τ −4( Δ )
dE = 4.51014 τ 2 e−τ cm3 s−1
kT πμ 0 AZ1 Z2
(11)
Here,
1
τ = 3E0 /kT = 42.5(Z12 Z22 μ/T6 ) 3 (12)
The maximum value of the integrand in the above equation is:

Imax = exp(−τ )

The values of E0 , Imax , Δ, etc., apart from the Coulomb barrier for several
reactions are tabulated in Table 1 for T6 = 15.
As the nuclear charge increases, the Coulomb barrier increases, and
the Gamow peak E0 also shifts towards higher energies. Note how rapidly the
maximum of the integrand Imax decreases with the nuclear charge and the
Coulomb barriers. The effective width Δ is a geometric mean of E0 and kT,
and Δ/2 is much less rapidly varying between reactions (for kT  E0 ). The
rapid variation of Imax indicates that of several nuclei present in the stel-
lar core, those nuclear pairs will have the largest reaction rates, which have
220 A. Ray

the smallest Coulomb barrier. The relevant nuclei will be consumed most
rapidly at that stage. (Note however that for the p+p reaction, apart from
the Coulomb barrier, the strength of the weak force, which transforms a pro-
ton to a neutron also comes into play).
When nuclei of the smallest Coulomb barrier are consumed, there is a
temporary dip in the nuclear generation rate, and the star contracts gravi-
tationally until the temperature rises to a point where nuclei with the next
lowest Coulomb barrier will start burning. At that stage, further contraction is
halted. The star goes through well defined stages of different nuclear burning
phases in its core dictated by the height of the Coulomb barriers of the fuels.
Note also from the Table 1, how far E0 , the effective mean energy of reaction
is below the Coulomb barrier at the relevant temperature. The stellar burning
is so slow because the reactions are taking place at such a far sub-Coulomb
region, and this is why the stars can last so long.
Table 1. Parameters of the thermally averaged reaction rates at T6 = 15
Reaction Coulomb Gamow Imax Δ (Δ)Imax
Barrier Peak (E0 ) (e−3E0 /kT ) (keV)
(MeV) (keV)
p+p 0.55 5.9 1.1 × 10−6 6.4 7 × 10−6
p+N 2.27 26.5 1.8 × 10−27 13.6 2.5 × 10−26
α + C12 3.43 56 3 × 10−57 19.4 5.9 × 10−56
O16 + O16 14.07 237 6.2 × 10−239 40.4 2.5 × 10−237

The above discussion assumes that a bare nuclear Coulomb potential is

seen by the charged projectile. For nuclear reactions measured in the labora-
tory, the target nuclei are in the form of atoms with electron cloud surrounding
the nucleus and giving rise to a screened potential – the total potential then
goes to zero outside the atomic radius. The effect of the screening is to reduce
the effective height of the Coulomb barrier. Atoms in the stellar interiors are
in most cases in highly stripped state, and nuclei are immersed in a sea of
free electrons which tend to cluster near the nucleus. When the stellar den-
2 1/2
sity increases,
 the2 so called Debye-Huckel radius RD = (kT /4πe ρNA ξ) ,
(here: ξ = i (Zi + Zi )Xi /Ai ) which is a measure of this cluster “radius”,
decreases, and the effect of shielding upon the reaction cross-section becomes
more important. This shielding effect enhances thermonuclear reactions in-
−3/2
side the star. The enhancement factor f0 = exp(0.188Z1Z2 ξρ1/2 T6 , varies
between 1 and 2 for typical densities and compositions [28] but can be large
at high densities.
 Thermonuclear fusion and collapse and explosion in massive stars 221

3 Hydrogen burning: the pp chain

The quantitative aspects of the problem of solar energy production with de-
tails of known nuclear physics of converting hydrogen into helium was first
worked out by von Weizsäcker (1937-38) [32], [33] and Bethe & Critchfield
(1938-1939) [34], which made it clear that two different sets of reactions: the
p-p chains and the CN cycle can do this conversion. This happens in the core
of the star initially (at the “main sequence” stage), and then later in the life
of a star in a shell of burning hydrogen around an inert core of He.
In the first generation of stars in the galaxy only the p-p cycle may have
operated. In second generation, heavier elements like C, N from the ashes
of burning in previous stars are available and they too can act as catalysts
to have thermonuclear fusion of hydrogen to helium. Since in the very first
generation, the heavier elements like C, N were practically absent, all stars
including the massive stars, burnt hydrogen through the p-p cycle. [A recent
discovery ([35]) of a low-mass star with an iron abundance as low as 1/200,000
of the solar value (compare the previous record of lowest iron abundance less
than 1/10,000 that of the sun), suggests that such first generation stars are
still around].
c
The sun with a central temperature of 15.7 million degrees, (T6 = 15.7)
burns by p-p chains. Slightly more massive star (with central temperature
T6 ≥ 20) burns H by the CNO cycle also. Davis et al.s’ solar neutrino exper-
iment [23], which in 1968 had only an upper limit of the neutrino flux, itself
put a limit of less than 9% of the sun’s energy is produced by the carbon-
nitrogen cycle (the more recent upper limit [36] is 7.3%, from an analysis
of several solar neutrino experiments, including the Kamland measurements).
Note however that for the standard solar model, the actual contribution of
CNO cycle to solar luminosity is ∼ 1.5% [15]). In CNO cycle, nuclei such as
C, N, O serve as “catalysts” do in a chemical reaction. The pp-chain and the
CNO cycle reaction sequences are illustrated in Figs. 4 and 10.
The pp-chain begins with the reaction p + p → d + e+ + νe . Bethe and
Critchfield [34] showed that weak nuclear reaction is capable of converting
a proton into a neutron during the brief encounter of a scattering event.
(This reaction overcomes the impasse posed by the instability of 2 He in the
p + p →2 He reaction via the strong or electromagnetic interactions, and as
the next nucleus 3 Li reachable via these interactions is also unstable as a
final product). Since a hydrogen atom is less massive than a neutron, such a
conversion would be endothermic (requiring energy), except for the fact that
a neutron in a deuterium nucleus 2 D can form a bound state with the proton
with a binding energy of 2.224 MeV – thus making the reaction exothermic
with an available kinetic energy of 0.42 MeV. The released positron soon pair
annihilates into photons making the total energy released to be 1.442 MeV.
Because of the low Coulomb barrier, in the p+p reaction (Ec = 0.55 MeV),
a star like the sun would have consumed all its hydrogen quickly (note the
relatively large value of (Δ)Imax in Table 1), were it not slowed down by the
222 A. Ray

4
THE REACTIONS OF THE P–P CHAIN, 4P He + 2e+ + 2ne + Qeff
(BP2000 model)
p(p, e+, n) d En = 0.26 MeV
d (p, g ) 3He Fn = 6×1010cm–2s–1

85% 15%

3
He (3He, 2P)4He 3
He (α, g )7Be
15% 0.02%

En = 0.81 MeV
7Be(e–, n) 7Li 7Be(p, g ) 8B
En = 7.3 MeV
Fn = 0.47×1010cm–2s–1 7
Li(p, α) 4He 8
B(e , ν) 8Be*
+
Fn = 5.8×10+8cm–2s–1
8
Be*(α) 4He

CHAIN I CHAIN II CHAIN III

Qeff = 26.20 MeV Qeff = 25.66 MeV Qeff = 19.17 MeV
(2.0% loss) (4.0% loss) (28.3% loss)

Fig. 4. The p-p chain starts with the formation of deuterium and 3 He. Thereafter,
3
He is consumed in the sun 85% of the time through ppI chain, whereas pp II and
pp III chains together account for 15% of the time in the Bahcall Pinsonneault 2000
solar model. The pp III chain occurs only 0.02% of the time, but the 8 B β + -decay
provides the higher energy neutrinos (average Ēν = 7.3 MeV). The net result of
the chains is the conversion of four protons to a helium, with the effective Q-values
(reduced from 26.73 MeV) as shown, due to loss of energy in escaping neutrinos. See
[38, 37] for updated branching ratios and neutrino fluxes for BPS2008(AGS) model

weakness of the weak interactions. The calculation of probability of deuteron

formation consists of two separate considerations: 1) the penetration of a mu-
tual potential barrier in a collision of two protons in a thermal bath and
2) the probability of the β-decay and positron and neutrino emission. Bethe
and Critchfield used the original Fermi theory (point interaction) for the sec-
ond part, which is adequate for the low energy process.

3.1 Cross-section for deuterium formation

The total Hamiltonian H for the p-p interaction can be written as a sum of
nuclear term Hn and a weak-interaction term Hw . As the weak interaction
term is small compared to the nuclear term, first order perturbation theory
can be applied and Fermi’s “Golden rule”, gives the differential cross-section
as:
 Thermonuclear fusion and collapse and explosion in massive stars 223

2πρ(E)
dσ = | < f |Hw |i > |2 (13)
h̄vi
here ρ(E) = dN/dE, is the density of final states in the interval dE and
vi is the relative velocity of the incoming particles. For a given volume V, the
number of states dn between p and p+dp is:-

4πp2e dpe 4πp2ν dpν

dN = dne dnν = (V 3
)(V ) (14)
h h3
By neglecting the recoil energy of deuterium (since this is much heavier
than the outgoing positron in the final state) and neglecting the mass of the
electron neutrino, we have:

E = Ee + Eν = Ee + cpν (15)
and dE = dEν = cpν , for a given Ee and,

16π 2 V 2 2
ρ(E) = dN (E)/dE = dne (dnν /dE) = p (E − Ee )2 dpe = ρ(Ee )dpe
c3 h 6 e
(16)
The matrix element that appears in the differential cross section, may be
written in terms of the initial state wave function Ψi of the two protons in the
entrance channel and the final state wave function Ψf as:

Hif = [Ψd Ψe Ψν ]∗ Hβ Ψi dτ (17)

If the energy of the electron is large compared to Z × Rydberg (Rydberg

R∞√= 2π 2 me4 /ch3 ), then a plane wave approximation is a good one: Ψe =
1/( V )exp(ike .r) where the wave-function is normalized over volume V. (For
lower energies, typically 200 keV or less, the electron wave-function could be
strongly affected by nuclear charge (see [39])). Apart from this, the final state
wave function: [Ψd Ψe Ψν ] has a deuteron part Ψd whose radial part rapidly
vanishes outside the nuclear domain (R0 ), so that the integration need not
extend much beyond r  R0 (for example, the deuteron radius Rd = 1.7
fm). Note that because of the Q-value of 0.42 MeV for the reaction, the
kinetic energy of the electron (Ke ≤ 0.42 MeV) and the average energy of
the neutrinos (Ēν = 0.26 MeV) are low enough so that for both electrons and
neutrino wave-functions, the product kR0 ≤ 2.2 × 10−3 and the exponential
can be approximated by the first term of the Taylor expansion:
√ √
Ψe = 1/( V )[1 + i(ke .r)] ∼ 1/( V) (18)
and √
Ψν ∼ 1/( V )
Then the expectation value of the Hamiltonian, for a coupling constant g is:
224 A. Ray

 
g
Hif = [Ψd Ψe Ψν ]∗ Hβ Ψi dτ = [Ψd ]∗ Ψi dτ (19)
V
The dτ integration can be broken into space and spin parts Mspace and
Mspin :

2π 16π 2 2 2
dσ = 2
g Mspin Mspace p2e (E − Ee )2 dpe (20)
h̄vi c3 h6
The total cross-section upto an electron energy of E is proportional to:

 E  W
(me c2 )5
p2e (E − Ee )2 dpe = (We2 − 1)1/2 (W − We )2 We dWe (21)
0 c3 1

where W = (E + me c2 )/me c2 . The integral over W can be shown as:

W4 3W 2 2 W
f (W ) = (W 2 − 1)1/2 [ − − ]+ ln[W + (W2 − 1)1/2 ] (22)
30 20 15 4
so that:
m5e c4
σ= f (W )g 2 Mspace
2 2
Mspin (23)
2π 3 h̄7
At large energies, the factor f (W ) behaves as:
1 5
f (W ) ∝ W 5 ∝
E (24)
30
The final state nucleus (deuterium in its ground state) in the reaction:
p + p → d + e+ + νe , has Jfπ = 1+ , with a predominant relative orbital angular
momentum lf = 0 and Sf = 1 (triplet S-state). For a maximally probable
super-allowed transition, there is no change in the orbital angular momentum
between the initial and final states of the nuclei. Hence for such transitions,
the initial state two protons in the p + p reaction must have li = 0. Since the
protons are identical particles, Pauli principle requires Si = 0, so that the total
wave-function will be antisymmetric in space and spin coordinates. Thus, we
have a process: |Si = 0, li = 0 >→ |Sf = 1, lf = 0 >. This is a pure Gamow-
Teller13 transition with coupling constant g = CA (the axial vector coupling
component can be obtained, from the pure GT decay 6 He(0+ ) →6 Li(1+ )).
13
In the beta-decay allowed approximation, we neglect the variation of the lep-
ton wave-functions over the nuclear volume and the nuclear momentum (this is
equivalent to neglecting all total lepton orbital angular momenta L > 0). The
total angular momentum carried off by the leptons is their total spin: i.e. S = 1
or 0, since each lepton has s=1
2
. When the lepton spins in the final state are anti-
parallel, se +sν = stot = 0 the process is the Fermi transition with Vector coupling
constant g = CV (e.g. a pure Fermi decay: 14 O(Jiπ = 0+ ) →14 N (Jfπ = 0+ )).
When the final state lepton spins are parallel, se + sν = stot = 1, the process is
 Thermonuclear fusion and collapse and explosion in massive stars 225

The spin matrix element in the above expression for energy integrated
cross-section σ, is obtained from summing over the final states and averaging
over the initial states and dividing by 2 to take into account that we have two
identical particles in the initial state. Thus,
2 2
1 m5 c 4 Mspace Mspin
λ= = 3 7 f (W )g 2 (25)
τ 2π h̄ vi 2
2 (2J+1)
where, Mspin = = 3. And the space matrix element is:
(2J1 +1)(2J2 +1)
 ∞
Mspace = χf (r)χi (r)r2 dr (26)
0
3/2
in units of cm . The above integral contains the radial parts of the nu-
clear wave-functions χ(r), and involves Coulomb wave-functions for barrier
penetration at (low) stellar energies. The integral has been evaluated by nu-
merical methods ( [42]), and Fig. 5 shows schematically how the Mspace is
evaluated for the overlap of the deuterium ground state wave-function with
the initial pair of protons state. (See also [43], [44] for details of calculations of
the overlap integral and writing the astrophysical S-factor in terms of the beta
decay rate of the neutron [43] which takes into account of radiative correc-
tions to the axial-vector part of the neutron decay through an effective matrix
Gamow-Teller with g = CA . For Fermi coupling, there is no change in the (total)
angular momentum between the initial and final states of the nuclei (ΔJ = 0).
For the Gamow-Teller coupling, the selection rules are: ΔJ = 0 or ±1 (but the
possibility ΔJ = 0 is excluded between two states of zero angular momentum).
The size of the matrix element for a transition depends on the overlap of the
wave-functions in the initial and final states. In the case of “mirror pair” of nuclei
(the nucleus AZ = (2Z + 1)Z is the mirror of the nucleus (2Z + 1)Z+1 ), the wave-
functions are very much alike as shown through simple heuristic arguments [40].
For these nuclei, f t-values range from ∼ 1000 − 5000 and are called super-allowed
transitions. For super-allowed transitions, which have maximum decay probabil-
ities, there are no changes in the orbital angular momentum between the initial
and final states of the nuclei. In the p + p → D + e+ + νe reaction, the initial
proton state is antisymmetric to an interchange of space and spin coordinates
and the final deuteron is symmetric in this respect. (In fact when the two protons
are in the S state (which is most favorable for their close approach), their spins
will be anti-parallel (a singlet state) whereas the ground state of the deuteron is
a triplet S state). If this were the complete description of the exchange symmetry
properties of the Gamow-Teller transition (permitting a change of spin direction
of the proton as it transforms to a neutron, changing the total spin by one unit)
advocated here this would actually be forbidden. However in the use of configu-
ration space in beta-decay process one must include isotopic spin as well. The 1 S
state of the two protons is symmetric to exchange of this coordinate, whereas the
deuteron (consisting of both a proton and a neutron) function is antisymmetric
in this coordinate. In the complete coordinate system the transition is from an
initial antisymmetric state to another antisymmetric final state accompanied by
a positron emission ( [41]).
226 A. Ray

element, the assumption being that these are the same as that for the proton
beta decay in the pp reaction above). In the overlap integral one needs only
the S-wave part for the wave-function of the deuteron ψd , as the D-wave part
makes no contribution to the matrix element [42], although its contribution
to the normalization has to be accounted for. The wave-function of the initial
two-proton system ψp is normalized to a plane wave of unit amplitude, and
again only the S-wave part is needed. The asymptotic form of ψp (well outside
the range of nuclear forces) is given in terms of regular and irregular Coulomb
functions and has to be defined through quantities related to the S-wave phase
shifts in p-p scattering data). The result is a minuscule total cross-section of
σ = 10−47 cm2 at a laboratory beam energy of Ep = 1 MeV, which cannot be
measured experimentally even with milliampere beam currents.
The reaction p + p → d + e+ + νe is a non-resonant reaction and at all
energies the rate varies smoothly with energy (and with stellar temperatures),
with S(0) = 3.8 × 10−22 keV barn and dS(0)/dE = 4.2 × 10−24 barn. As for
example, the central temperature of the sun T6 = 15, this gives: < σv >pp =
1.2 × 10−43 cm3 s−1 . For density in the center of the sun ρ = 100 gm cm−3
and equal mixture of hydrogen and helium (XH = XHe = 0.5), the mean life
of a hydrogen nucleus against conversion to deuterium is τH (H) = 1/NH <
σv >pp ∼ 1010 yr. This is comparable to the age of the old stars. The reaction
is so slow primarily because of weak interactions and to a lesser extent due
to the smallness of the Coulomb barrier penetration factor (which contributes
a factor ∼ 10−2 in the rate), and is the primary reason why stars consume
their nuclear fuel of hydrogen so slowly. For a calculation of the weak capture
of protons on protons using calculated wavefunctions obtained from modern,
realistic high precision interactions, see [29].

3.2 Deuterium burning

Once deuterium is produced in the weak interaction mediated p + p reaction,
the main way this is burnt in the sun turns out to be:

d + p → 3 He + γ (27)
3
This is a non-resonant direct capture reaction to the He ground state with
a Q-value of 5.497 MeV and S(0) = 2.5 × 10−3 keV barn. The angle averaged
cross-sections measured as a function of proton + deuterium center of mass
energy, where the capture transitions were observed in gamma-ray detectors
at several angles to the incident proton beam direction, are well explained by
the direct capture model (see Fig. 6 after [13]). The LUNA collaboration [31]
has measured the cross section down to a 2.5 keV c.m. energy, well below the
solar Gamow peak using a segmented Bismuth germanate (BGO) gamma-ray
detector and found the S(E) factor to be in fair agreement with extrapolation
of data at higher energies.
The reactions comprising the rest of the (three) pp-chains start out with
the predominant product of deuterium burning: 3 He (manufactured from d+p
 Thermonuclear fusion and collapse and explosion in massive stars 227

Coulomb
V(r)
Barrier
Projectile
0
–2.22 MeV Radius r
Ground
Nuclear
state of
Potential
Deuterum
–Vo
Ro = Nuclear
Radius

Deuterium Ground State

χ f (r)

0
Ro Radius r

P + P Scattering
χ i(r)

0
Ro Radius r

Product ~ Mspace
χ i(r)χ f(r)r2

0
Ro RADIUS r

Fig. 5. Schematic representation (after [13]) of the numerical calculation of the

spatial part of the matrix element Mspace in the p + p → d + e+ + νe reaction. The
top part shows the potential well of depth V0 and nuclear radius R0 of deuterium
with binding energy of −2.22 MeV. The next part shows the radius dependence
of the deuterium radial wave function χd (r). The wave-function extends far outside
the nuclear radius with appreciable amplitude due to the loose binding of deuterium
ground state. The p-p wave-function χpp (r) which comprise the li = 0 initial state
has small amplitude inside the final nuclear radius. The radial part of the integrand
entering into the calculation of Mspace is a convolution of both χd and χpp in the
second and third panels and is given with the hatched shading in the bottom panel.
It has the major contribution far outside the nuclear radius

reaction) as the starting point. The only other reactions with a S(0) greater
than the above are: d(d, p)t , d(d, n)3 He , d(3 He, p)4 He, and d(3 He, γ)5 Li.
However, because of the overwhelmingly large number of protons in the
stellar thermonuclear reactors, the process involving protons on deuterium
dominates. The rate of this reaction is so fast compared to its precursor:
228 A. Ray

Proton – projectile
γ – Bremsstrahlung

(Plane wave)
Valence – proton
2D 2D in available orbit
(Standing wave)
Target
(Spectator) FINAL NUCLEUS 3He
EP
Q = 5.9494MeV Ex Jπ
d+p
Direct
capture
(DC 0) +
o 1/2

Fig. 6. The Direct Capture reaction d(p, γ)3 He to form 3 He in its ground state. The
proton projectile (shown as a plane wave) radiates away a bremsstrahlung photon
to be captured in a “valence” orbital around the 2 D

p + p → d + e+ νe , that the overall rate of the pp-chain is not determined by

this reaction.
One can show that the abundance ratio of deuterium to hydrogen in a
quasi-equilibrium has an extremely small value, signifying that deuterium
is destroyed in thermonuclear burning. The time dependence of deuterium
abundance D is:

dD H2
= rpp − rpd = < σv >pp −HD < σv >pd (28)
dt 2
The self regulating system eventually reaches a state of quasi-equilibrium
and has:

(D/H) =< σv >pp /(2 < σv >pd ) = 5.6 × 10−18 (29)

−18
at T6 = 5 and 1.7 × 10 at T6 = 40. For the solar system however, this
ratio is 1.5 × 10−4 and the observed (D/H)obs ratio in the cosmos is ∼ 10−5 .
The higher cosmic ratio is due to primordial nucleosynthesis in the early phase
of the universe before the stars formed. (The primordial deuterium abundance
is a key quantity used to determine the baryon density in the universe). Stars
only destroy the deuterium in their core due to the above reaction.
 Thermonuclear fusion and collapse and explosion in massive stars 229

3.3 3 He burning

The pp-chain-I is completed (see Fig. 4) through the burning of 3 He via the
reaction:
3
He + 3 He → p + p + 4 He (30)
with an S-factor: S(0) = 5500 keV barn and Q-value = 12.86 MeV. In
addition, the reaction:
3
He + D → 4 He + p (31)
has an S-factor: S(0) = 6240 keV barn, but since the deuterium concentra-
tion is very small as argued above, the first reaction dominates the destruction
of 3 He even though both reactions have comparable S(0) factors. The cross
section for the former reaction near the Gamow energy for the sun, has been
measured in [30].
3
He can also be consumed by reactions with 4 He (the latter is pre-existing
from the gas cloud from which the star formed and is synthesized in the early
universe and in Pop III objects). These reactions proceed through Direct Cap-
tures and lead to the pp II and pp III parts of the chain (happening 15% of
the time). Note that the reaction 3 He(α, γ)7 Be together with the subsequent
reaction 7 Be(p, γ)8 B control the production of high energy neutrinos in the
sun and are particularly important for the 37 Cl solar neutrino detector con-
structed by Ray Davis and collaborators.

3.4 Reactions involving 7 Be

As shown in Fig. 4, about 15% of the time, 3 He is burned with 4 He radiatively

to 7 Be. Subsequent reactions involving 7 Be as a first step in alternate ways
complete the fusion process: 4H → 4 He in the pp II and pp III chains.

Electron capture process

The first step of the pp II chain is the electron capture reaction on 7 Be :

7
Be + e− → 7 Li + νe (see Fig 7). This decay goes both to the ground state
−
of 7 Li as well as to its first excited state at EX = 0.478 keV, Jπ = 12 ) –
the percentage of decays to the excited state being 10.4 % in the laboratory.
The energy released in the reaction with a Q-value of 0.862 keV is carried
away by escaping mono-energetic neutrinos with energies: Eν = 862 and 384
keV. The measured laboratory mean life of the decay is τ = 76.9d. The cap-
ture rate in the laboratory can be obtained from Fermi’s Golden Rule and
utilizing the fact that the wave-functions of both the initial nucleus and the
final one vanish rapidly outside the nuclear domain and the electron wave-
function in that domain can be approximated as its value at r = 0 and
the neutrino wave-function by a plane wave normalizes to volume V, so that
230 A. Ray
√ √
Hif = Ψe (0)g/ V Ψ7∗Li Ψ7 Be dτ = Ψe (0)gMn / V , where Mn represents the
nuclear matrix element and the resultant capture rate is:

λEC = 1/τEC = (g 2 Mn2 /πc3 h̄4 )Eν2 |Ψe (0)|2 (32)

In the laboratory capture process, any of the various electron shells contribute
to the capture rate; however the K-shell gives the dominant contribution. At
temperatures inside the sun, e.g. T6 = 15, nuclei such as 7 Be are largely
ionized. The nuclei however are immersed in a sea of free electrons resulting
from the ionization process and therefore electron capture from continuum
states is possible (see e.g., [45], [46]). Since all factors in the capture of
continuum electrons in the sun are approximately the same as those in the
case of atomic electron capture, except for the respective electron densities,
the 7 Be lifetime in a star, τs is related to the terrestrial lifetime τt by:

τf r 2|Ψt (0)|2
∼ (33)
τt |Ψf r (0)|2
where |Ψf r (0)|2 is the density of the free electrons ne = ρ/mH at the nucleus,
ρ being the stellar density. The factor of 2 in the denominator takes care
of the two spin states of calculation of the λt whereas the corresponding
λf r is calculated by averaging over these two orientations. Taking account
of distortions of the electron wave-functions due to the thermally averaged
Coulomb interaction with nuclei of charge Z and contribution due to hydrogen
(of mass fraction XH ) and heavier nuclei, one gets the continuum capture rate
as:

2|Ψt (0)|2 τt
τf r = (34)
(ρ/MH )[(1 + XH )/2]2πZα(me c2 /3kT )1/2
with |Ψe (0)|2 ∼ (Z/a0 )3 /π. Bahcall et al. [44] obtained for the 7 Be nucleus a
lifetime:
1/2
T6
τf r (7 Be) = 4.72 × 108 s
ρ(1 + XH )
The temperature dependence comes from the nuclear Coulomb field cor-
rections to the electron wave-function which are thermally averaged. For
solar condition the above rate [46] gives a continuum capture lifetime of
τf r (7 Be) = 140d as compared to the terrestrial mean life of τt = 76.9d.
Actually, under stellar conditions, there is a partial contribution from some
7
Be atoms which are only partially ionized, leaving electrons in the inner
K-shell. So the contributions of such partially ionized atoms have to be taken
into account. Under solar conditions the K-shell electrons from partially ion-
ized atoms give another 21% increase in the total decay rate. Including this,
gives the solar lifetime of a 7 Be nucleus as: τ (7 Be) = 120d. In addition, the
solar fusion reactions have to be corrected for plasma electrostatic screening
enhancement effects. For a recent discussion of the issues see [47, 48].
 Thermonuclear fusion and collapse and explosion in massive stars 231

Ex(keV) Jπ Ex(keV) Jπ
–
0 3/2
–
7
Be
(τav = 77d) 10.4%
–
478 1/2
Q = 862 keV ELECTRON–
CAPTURE 89.6%
g
–
0 3/2
7Li

Fig. 7. Electron capture on 7 Be nucleus. The capture proceeds 10.4% of the time
to the first excited state of 7 Li at 478 keV, followed by a decay to the ground state
by the emission of a photon. The average energy of the escaping neutrinos (which
are from the pp II chain) is 814 keV

Capture reaction leading to 8 B

Apart from the electron capture reaction, the 7 Be that is produced is partly
consumed by proton capture via: 7 Be(p, α)8 B reaction. Under solar condi-
tions, this reaction happens only 0.02% of the time. The proton capture on
7
Be proceeds at energies away from the 640 keV resonance via the direct cap-
ture process. Since the product 7 Li nucleus emits an intense γ-ray flux of 478
keV, this prevents the direct measurement of the direct capture to ground state
γ-ray yield. The process is studied indirectly by either the delayed positron or
the breakup of the product 8 B nucleus into two alpha particles. This reaction
has a weighted average S(0) = 0.0238 keVbarn [49]. The 7 Be(p, α)8 B reac-
tion cross section measurement has been attempted both by direct capture
reactions as well as by the Coulomb dissociation of 8 B. For a comparison of
the S17 (0) factors determined by the two methods and a critical review of the
differences of direct and indirect methods, see [50].
The product 8 B is a radioactive nucleus with a lifetime τ = 1.1 s:
8
B → 8 Be + e+ + νe (35)
8 π +
The positron decay of B(J = 2 ) goes mainly to the Γ = 1.6 MeV broad
excited state in 8 Be at excitation energy Ex = 2.94 MeV (J π = 2+ ) due to
the selection rules (see Fig. 8). This excited state has very short lifetime and
quickly decays into two α-particles. This completes the pp III part of the pp-
chain. The average energy of the neutrinos from 8 B reactions is: Ēν (8 B) = 7.3
MeV. These neutrinos, having relatively high energy, play an important role
in several solar neutrino experiments. The neutrino spectrum is not the same
232 A. Ray

as that obtained from allowed β-decay approximation and is affected by the

broad α-unstable 8 Be final state. Winter et al. [51] measured the total energy
of the α-particles emitted following the β-decay of 8 B and determined the
neutrino energy spectrum corrected for recoil order effects in the 8 Be final
state and constructed from the decay strength function.

Ex Jπ

0 2+T = 1
–
8
B
(τav = 1.1s) π
>99% Ex J
(MeV)
Q = 16.957MeV
e+ – DECAY
2.9 2+
<1%

0 0+ – 0.092MeV
8
Be 4He + 4He

8B 8Be* + e++ ν
4He + 4He
Fig. 8. The decay scheme of 8 B with positron emission, which goes to the first ex-
cited state of 8 Be at EX = 2.9 MeV with a width of Γ = 1.6 MeV. The 8 Be nucleus
itself fissions into two alpha particles. The neutrinos accompanying the positron
decay of 8 B are the higher energy solar neutrinos with Ēν = 7.3 MeV

4 The CNO cycle and hot CNO

The sun gets most of its energy generation through the pp-chain reactions (see
Fig. 9). However, as the central temperature (in stars more massive than the
sun) gets higher, the CNO cycle (see below for reaction sequence) comes to
dominate over the pp-chain at T6 near 20 (this changeover assumes the solar
CNO abundance, the transition temperature depends upon CNO abundance
in the star; in fact, if one is able to isolate the neutrino flux from the Sun’s
weak CN cycle, as in say future extended Borexino or SNO+ experiments
 Thermonuclear fusion and collapse and explosion in massive stars 233

with the capability to detect low energy neutrinos, this could in turn directly
constrain the metallicity of the Sun’s core14 [38]. The early generation of stars
(usually referred to as the Population II (Pop II) stars, although there is an
even earlier generation of Pop III metal poor massive stars15 ) generated energy
primarily through the pp-chain. The Pop II stars are still shining in globular
clusters, and being of mass lower than that of the sun, are very old. Most other
stars that we see today are later generation stars formed from the debris of
heavier stars that contained heavy elements apart from (the most abundant)
hydrogen. Thus in the second and third generation stars (which are slightly
heavier than the sun) where higher central temperatures are possible because
of higher gravity, hydrogen burning can take place through faster chain of
reactions involving heavy elements C, N, and O which have some reasonable
abundance (exceeding 1%)) compared to other heavy elements like Li, Be, B
which are extremely low in abundance. The favored reactions involve heavier
elements (than those of the pp-chain) which have the smallest Coulomb bar-
riers but with reasonably high abundance. Even though the Coulomb barriers
of Li, Be, B are smaller than those of C, N, O (when protons are the lighter
reactants (projectiles)), they lose out due to their lower abundance.
In 1937-1938, Bethe and von Weizsäcker independently suggested the CN
part of the cycle, which goes as:
12
C(p, γ)13 N (e+ νe )13 C(p, γ)14 N (p, γ)15 O(e+ ν)15 N (p, α)12 C (36)
This has the net result, as before: 4p → He + 2e + 2νe with a Q = 26.73.
4 +

In these reactions, the 12 C and 14 N act merely as catalysts as their nuclei

are “returned” at the end of the cycle. Therefore the 12 C nuclei act as seeds
that can be used over and over again, even though the abundance of the
seed material is minuscule compared to the hydrogen. But note that there is
a loss of the catalytic material from the CN cycle that takes place through
the 15 N (p, γ)16 O reactions. However, the catalytic material is subsequently
returned to the CN cycle by the reaction: 16 O(p, γ)17 F (e+ νe )17 O(p, α)14 N .
In the CN cycle (see Fig 10), the two neutrinos involved in the beta decays
(of 13 N (t1/2 = 9.97min) and 15 O (t1/2 = 122.24s)) are of relatively low en-
ergy and most of the total energy Q = 26.73 MeV from the conversion of four
protons into helium is deposited in the stellar thermonuclear reactor. The rate
of the energy production is governed by the slowest thermonuclear reaction in
the cycle. Here nitrogen isotopes have the highest Coulomb barriers in charged
particle reactions, because of their Z = 7. Among them 14 N (p, γ)15 O is the

14
A measurement of CN-cycle neutrino flux (with an expected total flux of about 5×
108 cm−2 s−1 ) would test an assumption of the Standard Solar Model that during
the early pre-main-sequence Hayashi phase the Sun became homogeneous due to
convective mixing and that subsequent evolution has not appreciably altered the
distribution of metals [38].
15
Though subsequently when Carbon has been synthesized by triple-α process,
these Pop III stars turn on their CN cycle.
234 A. Ray

ENERGY PRODUCTION e (MeV / g.s)

1025 E
L
CYC
1020 CNO

1015
P–P CHAIN
1010

105

100

10–5

100 101 T 102 103

STELLAR TEMPERATURE T6

Fig. 9. Comparison of the temperature dependence of the p-p chain and the CNO
cycle energy production. The points marked for the solar central temperature T =
T6 = 15.7 are shown on both graphs. The CNO cycle generation dominates over
the pp-chain at temperatures higher than T6 = 20, so that for sun like stars, the
pp-chain dominates. For more massive stars, the CNO cycle dominates as long as
one of the catalysts: C, N, or O have initial mass concentration at least 1%. Note
the logarithmic scales of the graph and how both rates drop sharply with decreasing
temperature, with that of CNO cycle even more drastic due to higher Coulomb
barriers

slowest because this reaction having a final state photon is governed by electro-
magnetic forces while that involving the other nitrogen isotope: 15 N (p, α)12 C
is governed by strong forces and is therefore faster.
From the CN cycle, there is actually a branching off from 15 N by the reac-
tion 15 N (p, γ)16 O mentioned above. This involves isotopes of oxygen, and is
called the ON cycle; finally the nitrogen is returned to the CN cycle through
14
N . Together the CN and the ON cycles, constitutes the CNO bi-cycle. The
two cycles differ considerably in their relative cycle-rates: the ON cycle oper-
ates only once for every 1000 cycles of the main CN cycle. This can be gauged
from the S(0) factors of the two sets of reactions branching off from 15 N : for
the 15 N (p, α)12 C reaction S(0) = 65 MeV b, whereas for 15 N (p, γ)16 O, it is
64 keV b, i.e. a factor of 1000 smaller.
 Thermonuclear fusion and collapse and explosion in massive stars 235

CNO – bicycle
(p,g ) (p,a )
13 14 17
C N O
(p,g ) (e+,n )
(e+,ν)
CN – Cycle
13 15 17
Slowest fusion N O F
(p, g ) reaction (e.m.)
13 14
C N +
(e ,n ) (p,g )
(p,g )
m (p,g )
T1/2 = 9.97 (e+,ν)
12 15 18
13 15 C N O
N O (p,a ) (p,g )
(p,g ) (e+,n ) 1000 : 1 cycles
(p,g )
12 15
C N
(p, a) 13
C 14
N 17
O 18
F
strong forces
(e+,n )
13 15 17
N O F
(p,g ) (p,g )
12 13 16 18 19 20
C N O O F Ne
(p,g )
(p,a)

Fig. 10. The various CNO cycles. The left part is the CN cycle where only C and
N serve as catalysts for the conversion of four protons into 4 He. Here the slowest
fusion reaction is (p,γ) reaction on 14 N whereas the slower β-decay has a half-life of
9.97m. In the CNO bi-cycle (right part), there is leakage from the CN cycle to the
ON cycle through the branching at 15 N . The flow is returned to the CN cycle (which
cycles 1000 times for each ON cycle) through 17 O(p, α)14 N . The right bottom part
represents additional cycles linking into the CNO cycle through the 17 O(p, γ)18 F
reaction [13]

4.1 Hot CNO and rp-process

The above discussion of CNO cycle is relevant for typical temperatures T6 ≤

80. These are found in quiescently hydrogen burning stars with solar compo-
sition which are only slightly more massive than the sun. There are situations
where the hydrogen burning takes place at temperatures (T ∼ 108 − 109 K)
which are in excess of those found in the interiors of the ordinary “main
sequence” stars. Examples of these are: hydrogen burning at the accreting
surface of a neutron star or in the explosive burning on the surface of a white
dwarf, i.e. novae, or the outer layers of a supernova shock heated material in
the stellar mantle. These hot CNO cycles operate under such conditions on
a rapid enough time scale (few seconds) so that even “normally” β-unstable
nuclei like 13 N will live long enough to be burned by thermonuclear charged
particle reactions, before they are able to β-decay [52], [53]. So, unlike the
normal CNO the amount of hydrogen to helium conversion in hot CNO is
236 A. Ray

limited by the β-decay lifetimes of the proton-rich nuclei like: 14 O and 15 O

rather than the proton capture rate of 14 N . Wallace and Woosley [54] showed
that for temperatures, T ≥ 5 × 108 K, nucleosynthesised material can leak out
of the cycles. This leads to a diversion from lighter to heavier nuclei and is
known as the rapid proton capture or rp-process. The flow between the hot
CNO cycle and the rp capture process in X-ray bursts from the atmosphere
of a neutron star is regulated by the 15 O(α, γ)19 N e reaction. Another alpha
capture reaction 18 N e(α, γ)21 N a, continuously processes the available 4 He
nuclei flowing towards heavier elements. For a recent discussion on the hot
CNO and the rp process on accreting neutron stars, see [55].
The nucleosynthesis path of rp-process of rapid proton addition is anal-
ogous to the r-process of neutron addition (for neutron capture processes in
the early galaxy see [56]). The hot hydrogen bath converts CNO nuclei into
isotopes near the region of proton unbound nuclei (the proton drip line).
For each neutron number, a maximum mass number A is reached where the
proton capture must wait until β + -decay takes place before the buildup of
heavier nuclei (for an increased neutron number) can take place. Unlike the
r-process the rate of the rp-process is increasingly hindered due to the in-
creasing Coulomb barrier of heavier and higher-Z nuclei to proton projectiles.
Thus the rp-process does not extend all the way to the proton drip line but
runs close to the beta-stability valley and runs through where the β + -decay
rate compares favorably with the proton captures.

5 Helium burning and the triple-α reaction

After hydrogen burning in the core of the star has exhausted its fuel, the
helium core contracts slowly. Its density and temperature goes up as gravita-
tional energy released is converted to internal kinetic energy. The contraction
also heats hydrogen at the edge of the helium core, igniting the hydrogen to
burn in a shell. At a still later stage in the star’s evolution, the core has con-
tracted enough to reach central temperature density conditions: T6 = 100−200
and ρc = 102 − 105 gm cm−3 when the stellar core settles down to burn 4 He
in a stable manner. The product of helium burning is 12 C. Since in nature,
the A = 5 and A = 8 nuclei are not stable, the question arises as to how
helium burning bridges this gap. A direct interaction of three α particles to
produce a 12 C nucleus would seem at first sight, to be too improbable (as
was mentioned, for example, in Bethe’s 1939 paper [57], which was primarily
on the CN cycle). However, Öpik [58] and Salpeter [43], [59] independently
proposed a two step process where in the first step, two α particles interact to
produce a 8 Be nucleus in its ground state (which is unstable to α-breakup),
followed by the unstable nucleus interacting with another α-particle process
to produce a 12 C nucleus.
Thus the triple alpha reaction begins with the formation of 8 Be that has
a lifetime of only 1 × 10−16 s (this is found from the width Γ = 6.8 eV of the
 Thermonuclear fusion and collapse and explosion in massive stars 237

ground state and is the cause of the A = 8 mass gap). This is however long
compared to the transit time 1 × 10−19 s of two α-particles to scatter past
each other non-resonantly with kinetic energies comparable to the Q-value of
the reaction namely, Q = −92.1 keV. So it is possible to have an equilibrium
build-up of a small quantity of 8 Be in equilibrium with its decay or reaction
products: α + α → 8 Be. The equilibrium concentration of the 8 Be nucleus
can be calculated through the Saha equation

N1 N2  2π 3/2 3 (2J + 1) ER
N12 = h̄ exp(− ) (37)
2 μkT (2J1 + 1)(2J2 + 1) kT
at the relevant temperature T6 = 11 and ρ = 105 gm cm−3 to be:

N (8 Be)
= 5.2 × 10−10 (38)
N (4 He)
Salpeter suggested that this small quantity of 8 Be serves as the seed for
the second stage of the triple α-capture into the 12 C nucleus. It was however
shown by Hoyle [61] that the amount of 12 C produced for the conditions inside
a star at the tip of the red-giant branch is insufficient to explain the observed
abundance, unless the reaction proceeds through a resonance process [60].
The presence of such a resonance greatly speeds up the rate of the triple-α
process which then proceeds through an s-wave (l = 0) resonance in 12 C near
the threshold of 8 Be + α reaction. Since 8 Be and 4 He both have J π = 0+ ,
an s-wave resonance would imply that the resonant state in question has
to be 0+ in the 12 C nucleus. Hoyle suggested the excitation energy to be:
EX ∼ 7.68 MeV in the 12 C nucleus and this state was experimentally found
by W.A. Fowler’s group ( [62]) with spin-parity: J π = 0+ . This state has a
total width ( [13]) Γ = 8.9 ± 1.08 eV, most of which lies in Γα , due to the
major propensity of the 12 C nucleus to break-up through α-decay. (The decay
of the excited state of 12 C by γ-rays cannot go directly to the ground state,
since the resonance state as well as the ground state of the 12 C nucleus have
both J π = 0+ and 0+ → 0+ decays are forbidden. This partial width due to
gamma-decay is several thousand times smaller than that due to α-decay).
So, Γ = Γα + Γrad ∼ Γα and Γrad = Γγ + Γe+ e− = 3.67 ± 0.50 meV. Again
the radiative width Γrad is dominated by the width due to photon width
deexcitation: Γγ = 3.58 ± 0.46 meV. (Note the scales of millielectron Volts).
The reaction scheme for the first and the second parts of the triple-alpha
reaction is given in Fig. 11. The locations of the Gamow energy regions near
the above resonance state (for several stellar temperatures) are shown only
schematically.
The reaction rate for the 12 C formation can be calculated by using the
properties of the resonant state and the thermally averaged cross-section:

r3α = N8 Be Nα < σv >8 Be+α (39)

238 A. Ray

FIRST STEP: α+α 8

Be
Ex Jπ

ER = |Q| 0 0+ Γ = 6.8eV
Q = –92keV 8
Be
α+α

Gamow Peak ΔE
SECOND STEP : 8Be(α, g )12C
Ex Jπ E0 T6 = 1000
(keV)
7654 0+ T6 = 300
ER(3α) = 379keV ER = 287keV
T6 = 100 E0, ΔE
g e+ –e –
Q = 7275keV Q = 7367keV
3α Be + α
8 4439 2+

g
0 0+
12C

Fig. 11. Triple alpha process of 12 C synthesis. In the first step a small amount of
8
Be nuclei builds up in equilibrium with its decay products (forward and backward
reactions involve alpha particles). The second step involves a capture of another
alpha particle by the unstable 8 Be nucleus which proceeds via an s-wave resonance
state in the product nucleus 12 C located close to the Gamow energy window for
temperatures indicated schematically by the three-way arrows on the right

Here N8 Be and Nα are the number densities of interacting 8 Be and 4 He

nuclei and the angular brackets denote thermal averaging over a Maxwell
Boltzmann distribution ψ(E). This averaging leads to:
 ∞
r3α = N8 Be Nα ψ(E)v(E)σ(E)dE (40)
0
with
2 E dE
ψ(E) = √ exp(−E/kT)
π kT (kTE)1/2
and
 λ 2 2J + 1 Γ1 Γ2
σ(E) = π
2π (2J1 + 1)(2J2 + 1) (E − ER )2 + (Γ/2)2
is the Breit-Wigner resonant reaction cross section with the resonant energy
centroid at E = ER . The total width Γ is a sum of all decay channel widths
such as Γ1 = Γα and Γ2 = Γγ . If the width Γ is only a few eVs then the
 Thermonuclear fusion and collapse and explosion in massive stars 239

functions ψ(E) and v(E) can be pulled ∞ out of the integral. Then, the reaction
rate will contain an integral like: 0 σBW (E)dE = 2π(λ/2πh̄)2 ωΓ1 Γ2 /Γ ,
where ω = (2J + 1)/[(2J1 + 1)(2J2 + 1)] and the functions pulled out of the
integral need to be evaluated at E = ER . Since most of the time the excited
state of the 12 C ∗ breaks-up into α-particles, we have Γ1 = Γα dominating
over Γγ and (Γ1 Γ2 /Γ ) ∼ Γ2 . This limit usually holds for resonances of energy
sufficiently high so that the incident particle width (Γ1 ) dominates the natural
width of the state (Γ2 ). In that case, we can use the number density of the
8
Be nuclei in equilibrium with the α-particle nuclei bath as described by Saha
equilibrium condition:

h3
N (8 Be) = Nα2 ωf exp(−Er /kT) (41)
(2πμkT )3/2
where f is the screening factor. It is possible to get the overall triple-
alpha reaction rate by calculating the equilibrium concentration of the excited
(resonant) state of 12 C reached by the 8 Be + α →12 C ∗ reaction and then
multiplying that concentration by the gamma-decay rate Γγ /h̄ which leads
to the final product of 12 C. So, the reaction rate for the final step of the
triple-alpha reaction turns out to be:
3/2

2π 
r3α = N8 Be Nα h̄2 ωf Γ2 exp(−Er /kT) (42)
μkT
where μ is the reduced mass of the reactants 8 Be and α particle. This
further reduces by the above argument to:
 3
Nα3 3/2 2πh̄2 Γα Γγ
r3α→12 C = 3 f exp(−Q/kT) (43)
2 Mα kT Γ h̄
The Q-value of the reaction is the sum of ER (8 Be + α) = 287 keV and
ER (α + α) = |Q| = 92 keV and turns out to be: Q3α = (M12 C ∗ − 3Mα )c2 =
379.38 ± 0.20keV [63]. Numerically, the energy generation rate for the triple-
alpha reaction is:

r3α Q3α ρ2 X 3
3α = = 3.9 × 1011 3 α f exp(−42.94/T8) erg gm−1 s−1 (44)
ρ T8
The triple alpha reaction has a very strong temperature dependence: near
a value of temperature T0 , one can show that the energy generation rate is:
T n
(T ) = (T0 )( ) (45)
T0
where, n = 42.9/T8 − 3. Thus at a sufficiently high temperature and den-
sity, the helium gas is very highly explosive, so that a small temperature rise
gives rise to greatly accelerated reaction rate and energy liberation. When
helium thermonuclear burning is ignited in the stellar core under degenerate
conditions, an unstable and sometimes an explosive condition develops.
240 A. Ray
12 12
6 Survival of C in red giant stars and C(α, γ)16 O
reaction
The product of the triple-alpha reactions 12 C, is burned into 16 O by α-capture
reactions:
12
C + α →16 O + γ (46)
If this reaction proceeds too efficiently in helium burning Red giant stars,
then all the carbon will be burned up to oxygen. Carbon is however the most
abundant element in the universe after hydrogen, helium and oxygen, and the
cosmic C/O ratio is about 0.6. In fact, the O and C burning reactions and the
conversion of He into C and O take place in similar stellar core temperature
and density conditions. Major ashes of He burning in Red Giant stars are
C and O. Red Giants are the source of the galactic supply of 12 C and 16 O.
Fortuitous circumstances of the energy level structures of these alpha-particle
nuclei are in fact important for the observed abundance of oxygen and carbon.
For example, if as in the case of the 3α reaction, there was a resonance in
the 12 C(α, γ)16 O reaction near the Gamow window for He burning conditions
(i.e. T9 ∼ 0.1 − 0.2), then the conversion of 12 C → 16 O would have proceeded
at a very rapid rate. However, the energy level diagram of 16 O shows that for
temperatures upto about T9 ∼ 2, there is no level available in 16 O to foster
a resonant reaction behavior (Fig. 12). But since this nucleus is found in na-
ture, its production must go through either: 1) a non-resonant direct capture
reaction or 2) non-resonant captures into the tails of nearby resonances (i.e.
sub-threshold reactions). In Fig. 12, also shown on the left of the 16 O energy
levels, is the threshold for the 12 C +4 He reaction, drawn at the appropriate
level with respect to the ground state of the 16 O nucleus. The Gamow energy
regions drawn on the extreme right for temperatures T9 = 0.1 and above, in-
dicates that for the expected central temperatures, the effective stellar (center
of mass) energy region is near E0 = 0.3 MeV. This energy region is reached
by the low energy tail of a broad resonance centered at ECM = 2.42 MeV
above the threshold (the J π = 1− state at 9.58 MeV above the ground state
of 16 O) with a (relatively large) resonance width of 400 keV. On the other
hand, there are two sub-threshold resonances in 16 O (at EX = 7.12 MeV and
EX = 6.92 MeV), i.e. −45 keV and −245 keV below the α-particle thresh-
old that have J π = 1− and J π = 2+ , that contribute to stellar burning rate
by their high energy tails. However, electric dipole (E1) γ-decay of the 7.12
MeV state is inhibited by isospin selection rules. Had this not been the case,
the 12 C(α, γ)16 O reaction would have proceeded fast and 12 C would have
been consumed during helium burning itself. The two sub-threshold states
at −45 keV and −245 keV give contributions to the astrophysical S-factor of:
S1− (E0 ) = 0.1 MeV barn and S2+ (E0 ) = 0.2 MeV barn respectively at the rel-
evant stellar energy E0 = 0.3 MeV. The state at ECM = 2.42 MeV (J π = 1−
state at 9.58 MeV) gives a contribution: S1− (E0 ) = 1.5 × 10−3MeV barn. The
 Thermonuclear fusion and collapse and explosion in massive stars 241

total S-factor at E0 = 0.3 MeV is therefore close to 0.3 MeV barn. These then
provide low enough S or cross-section to not burn away the 12 C entirely to
16
O, so that C/O ∼ 0.1 at the least.

Ex(keV) Jπ
Γc.m.(keV)
Ec.m(keV) 10957 0 –

10367 4+
3195 25
9847 2+ 1– 400 keV
2685 9580 – 0.62
1
2418 400 30
2–
8872
10
E0 0.6
7162
7117 1– 0.3 E0 = 0.3 MeV
12C + 4He
6917 2+ (Stellar Gamow Peak)
–45 6130 3–
–245 – 45keV(1–)
6049 0+
– 245keV
2+

0 0+
16
O

Fig. 12. Energy levels of 16 O nucleus near and above the alpha-particle threshold of
capture on 12 C. Shown on the right are effective stellar energy regions corresponding
to the temperatures given near the three-way arrows. The reaction rate is influenced
mainly by the high energy tails of two sub-threshold resonances in 16 O at ER =
−45 keV and ER = −245 keV, plus the low energy tail of another high-lying broad
resonance at 9580 keV

16
Additionally, O nuclei are not burnt away by further α-capture in the
reaction:
16
O +4 He → 20
Ne + γ (47)
20
A look at the level schemes of N e (see Fig. 13) shows the existence of
a EX = 4.97 MeV state (J π = 2− ) in the Gamow window. However, this
state cannot form in the resonance reaction due to considerations of parity
conservation (unnatural parity of the resonant state)16 . The lower 4.25 MeV
state (J π = 4+ ) in 20 N e also cannot act as a sub-threshold resonance as it

16
Whether or not a resonant state can be formed or accessed via a given reac-
tion channel depends upon the angular momentum and parity conservation laws.
The spins of the particles in the entrance channel, j1 , j2 and relative angular
242 A. Ray

lies too far below threshold and is formed in the g-wave state. Therefore only
direct capture reactions seem to be operative, which for (α, γ) reactions lead
to cross-sections in the range of nanobarns or below. Thus the destruction of
16
O via: 16 O(α, γ)20 N e reaction proceeds at a very slow rate during the stage
of helium burning in Red Giant stars, for which the major ashes are carbon
and oxygen and these elements have their galactic origin in the Red Giants.
To summarize, the synthesis of two important elements for the evolution
of life as we know on the earth have depended upon fortuitous circumstances
of nuclear properties and selection rules for nuclear reactions. These are: 1)
the mass of the unstable lowest (ground) state of 8 Be being close to the
combined mass of two α-particles; 2) there is a resonance in 12 C at 7.65 MeV
which enhances the alpha addition reaction (the second step); and 3) parity
conservation has protected 16 O from being destroyed in the 16 O(α, γ)20 N e
reactions by making the 4.97 MeV excited state in 20 N e of unnatural parity.
The experimental determination of the reaction rate 12 C(α, γ)16 O has been
an important goal in nuclear astrophysics for several decades. Its cross sec-
tion at the position of the Gamow window for a typical stellar temperature of
2.5 × 108 K is comparable to that of weak interaction cross-sections. At those
energies, this reaction is practically a non-resonant reaction and its cross-
section is determined by the tails of interfering resonance and sub-threshold
states [64]. The low cross section and the complexity of low energy contribu-
tions to the reaction rate makes a reliable prediction difficult [65, 66].

7 Advanced stages of thermonuclear burning

As the helium burning progresses, the stellar core is increasingly made up of
C and O. At the end of helium burning, all hydrogen and helium is converted
into a mixture17 of C and O, and since H, He are most abundant elements
in the original gas from which the star formed, the amount of C and O are
far more in the core than the traces of heavy elements in the gas cloud.
Between these two products, the Coulomb barrier for further thermonuclear
reaction involving the products is lower for C nuclei. At first the C+O rich

momentum l adds upto the angular momentum of the resonant state J = j1 +j2 +l.
Therefore, for spin-less particles like the closed shell nuclei 4 He,16 O (j1 = 0, j2 =
0), we have J = l. In the entrance channel of the reacting particles, the parity
would be: (−1)l π(j1 )π(j2 ) = (−1)l=0 (1)(1). If the parity of the resonance state
were the same as that of the entrance channel, then the resultant state would
have been a “natural parity” state. However, since the 4.97 MeV state in 20 N e
has an assignment: J π = 2− , this is an “unnatural parity” state.
17
Note however the caveat: if the amount of 12 C is little (either due to a long stellar
lifetime of He burning or due to a larger rate of the 12 C + α →16 O + γ reaction
whose estimate outlined in the earlier section is somewhat uncertain), then the
star may directly go from He-burning stage to the O-burning or Ne-burning stage
skipping C-burning altogether ( [67]).
 Thermonuclear fusion and collapse and explosion in massive stars 243
+
294 2
9.61 3+

a 7.65 0+
(–0.08) a 0 (7.37) 0+ E0
4 8
He Be
8
Be–MEQUILIBRIUMM
(Sufficient 8Be 4.44 2+
due to small
mass differences) 9.58 1–
12 8.87 2–
“CREATION” OF C
(Due to thermal resonance) + a
0(7.18) 0 7.12 1– E0
12 6.92 2+
C 6.13 3–
6.05 0+

‘”SURVIVAL “ OF C12
(Due to lack of
7.00 4–
thermal resonance
but 16O produced via 6.72 0+
5.78 1–
subthreshold resonances) a 5.62 3–
0(4.73) 0– E0
4.97 2–
4+
18
O 4.25

BLOCKING 1.63 2+
OF NUCLEOSYNTHESIS
0 0+
20Ne

Fig. 13. Energy levels of nuclei participating in thermonuclear reactions during the
helium burning stage in red giant stars (after [13]). Survival of both 12 C and 16 O
in red giants, from which terrestrial abundances result, depends upon the fortuitous
circumstances of nuclear level structures and other properties of these nuclei

core is surrounded by He burning shells and a helium rich layer, which in turn
may be surrounded by hydrogen burning shell and the unignited hydrogen
rich envelope. When the helium burning ceases to provide sufficient power,
the star begins to contract again under its own gravity and as implied by
the Virial theorem the temperature of the helium exhausted core rises. The
contraction continues until either the next nuclear fuel begins to burn at rapid
enough rate or until electron degeneracy pressure halts the infall.

7.1 Carbon burning

Stars somewhat more massive than about 3 M contract until the temperature
is large enough for carbon to interact with itself (stars less massive on the main
244 A. Ray

sequence may settle as degenerate helium white dwarfs). For stars which are
more massive than M ≥ 8−10 M (mass on the main sequence, - not the mass
of the C+O core), the contracting C+O core remains non-degenerate until C
starts burning at T ∼ 5 × 108 K and ρ = 3 × 106 gcm−3 . Thereafter sufficient
power is generated and the contraction stops and quiescent (hydrostatic, not
explosive) C-burning proceeds (see Fig. 14).

0.6 M Inward white dwarf C-flash

6 0.4 M
He-
Degenerate flash
region
5
EF>>kT

4 He
0.7 M
log ρc

He
kT
≥
EF 4M
3 He
=O 15.6 M
2 EF < kT
qu in
ce
se -ma
en
e

1
Pr

Non-degenerate region
IM
O
7.0 7.5 8.0 8.5 9.0
log Tc

Fig. 14. Tracks in the core temperature, density plane of stars of various masses
(at the start of hydrogen burning i.e. main sequence masses). Note that a star of
mass M ∼ 15M ignites all its fuels in non-degenerate conditions, whereas a star of
mass M ∼ 4M ignites carbon under strongly degenerate conditions. (After [68])

The combined mass of two reacting 12 C nuclei falls at an excitation energy

of 14 MeV in the compound nucleus of 24 M g. At this energy there are many
compound nuclear states, and the most effective range of stellar energies (the
Gamow window) at the relevant temperature is about 1 MeV; hence a number
of resonant states can contribute to the decay of the compound nucleus, and
 Thermonuclear fusion and collapse and explosion in massive stars 245

even the large angular momentum resonances may be important because the
penetration factors in massive nuclei are not affected by centrifugal barriers.
The carbon on carbon burning can proceed through multiple, energetically
allowed reaction channels, listed below:
12
C + 12 C → 20
N e + 4 He (Q = 4.62 MeV) (48)

→ 23
N a + p (Q = 2.24 MeV) (49)

→ 23
M g + n (Q = −2.62 MeV) (50)
At the temperatures where carbon burning starts, the neutron liberating
reactions requires too much particle kinetic energy to be effective. In addition,
based on laboratory measurements at higher energies compared to the stellar
energies, the electromagnetic decay channel (24 M g + γ) and the three parti-
cle channel (16 O + 2α) have lower probability compared to the two particle
channels: 23 N a + p and 20 N e + α. The latter two channels have nearly equal
probabilities (see [14]; at the lowest center of mass energies for which cross-
sections are measured in the laboratory for the proton and α channels, (i.e.
about 2.45 MeV [69]), the branching ratios were bp ∼ 0.6 and bα ∼ 0.4), and
therefore the direct products of carbon burning are likely to be 23 N a, 20 N e,
protons and alpha particles. The rate for this reaction per pair of 12 C nuclei
is ( [70]):

36.55(1 + 0.1T9 )1/3 2

logλ12,12 = logf12,12 + 4.3 − 1/3
− logT9 (51)
T9 3

the factor f12,12 is a screening factor. Now, at the temperatures of 12 C

burning, the liberated protons and alpha particles can be quickly consumed
through the reaction chain: 12 C(p, γ)13 N (e+ νe )13 C(α, n)16 O. Thus, the net
effect is that the free proton is converted into a free neutron (which may
be further captured) and the α-particle is consumed with 12 C into 16 O. The
α-particles are also captured by other alpha-particle nuclei, resulting in, at
the end of carbon burning in nuclei like: 16 O, 20 N e, 24 M g and 28 Si. These
secondary reactions augment the energy released by the initial carbon reaction
and Reeves (1959) estimated that each pair of 12 C nuclei release about 13
MeV of energy. Towards the end of carbon burning phase there are also other
reactions such as: 12 C +16 O and 12 C +20 N e which take place. But these are
less rapid and are not expected to play major roles compared to the 12 C +12 C
reactions, due to their increased Coulomb barriers. A recent discussion of the
heavy ion reactions involving C and O is contained in [1] Sect. 3.6.
During the carbon-burning and subsequent stages, the dominant energy
loss from the star is due to neutrinos streaming out directly from the stellar
thermonuclear furnace, rather than by photons from the surface. The neutrino
luminosity is a sensitive function of core temperature and quickly outshines the
246 A. Ray

surface photon luminosity of the star at carbon burning stage. The (thermal)
evolutionary time scale of the star, due to the neutrino emission becomes very
short and the core evolves rapidly, – so rapidly (compared to the “cooling”
time scale Kelvin-Helmholtz time: τKH ∼ GM 2 /RLph ) that the conditions in
the core are “uncommunicated” to the surface, since this happens by photon
diffusion. The surface conditions (e.g. the temperature) then does not change
much as the core goes beyond the carbon burning stage, and it may not be
possible just by looking at a star’s surface conditions whether the core is
close to a supernova stage or has many thousands of years of hydrostatic
thermonuclear burning to go.

7.2 Neon burning

The result of carbon burning is mainly neon, sodium and magnesium, but
aluminum and silicon are also produced in small quantities by the capture of
α, p and n released during carbon burning. When carbon fuel is exhausted,
again the core contracts and its temperature Tc goes up. At approximately
T9 ∼ 1, energetic photons from the high energy tail of the Planck distribution
function can begin to disintegrate the 20 N e ash (see Fig. 13) so that one has
the reaction: 20 N e + γ →16 O +4 He.
Nucleons in a nucleus are bound with typical binding energy of several to
8 MeV. An energetic γ-ray photon is required to photo-eject a nucleon. Two
nucleon ejection requires more energy. Alpha particles are however released at
approximately the same energy as a nucleon due to the low separation energy
of an alpha particle in the nucleus. For example, the alpha separation energy
in 20 N e is 4.73 MeV. Thus, the major photo-nuclear reactions are: (γ, n), (γ, p)
and (γ, α) processes. For a photo-disintegration reaction to proceed through
an excited state EX in the mother, the decay rate is:-
 
 EX  2JR + 1 Γγ Γα
λ(γ, α) = exp − × (52)
kT 2J0 + 1 Γ h̄

In the above equation, the first factor in square brackets on the RHS is the
probability of finding the nucleus in the excited state EX and spin JR (with
J0 being the ground state spin), while the second factor Γα /h̄ is the decay rate
of the excited state with an alpha particle emission. Now since EX = ER + Q,
we have:

exp(−Q/kT) Γα Γγ
λ(γ, α) = (2JR + 1) exp(−ER /kT) (53)
h̄(2J0 + 1) Γ
At T9 ≥ 1, the photo-disintegration is dominated by the 5.63 MeV level
in 20 N e (see Fig. 13). At approximately T9 ∼ 1.5, the photo-dissociation rate
becomes greater than the rate for alpha capture on 16 O to produce 20 N e (i.e.
the reverse reaction), thus leading effectively to the net dissociation of 20 N e.
The released 4 He reacts with the unspent 20 N e and leads to: 4 He + 20 N e →
 Thermonuclear fusion and collapse and explosion in massive stars 247

24
M g + γ. Thus the net result of the photo-dissociation of two 20 N e nuclei
is: 2 × 20 N e → 16 O + 24 M g with a net Q-value of 4.58 MeV. The brief neon
burning phase concludes at T9 close to ∼ 1.5.

7.3 Oxygen burning

At the end of the neon burning the core is left with a mixture of alpha particle
nuclei: 16 O and 24 M g. After this another core contraction phase ensues and
the core heats up, until at T9 ∼ 2, 16 O begins to react with itself:
16
O + 16 O → 28 Si + 4 He (54)

→ 32 S + γ (55)
The first reaction takes place approximately 45% of the time with a Q-
value of 9.593 MeV. In addition to Si and S, the oxygen burning phase also
produces, Ar, Ca and trace amounts of Cl, K, etc. upto Sc. Then at T9 ∼ 3,
the produced 28 Si begins to burn in what is known as the Si burning phase.

7.4 Silicon burning

As we have seen, most of the stages of stellar burning involve thermonuclear

fusion of nuclei to produce higher Z and A nuclei. The first exception to
this is neon burning where the photon field is sufficiently energetic to photo-
dissociate neon, before the temperature rises sufficiently to allow fusion reac-
tions among oxygen nuclei to overcome their Coulomb repulsion. Processing
in the neon burning phase takes place with the addition of helium nuclei to
the undissociated neon rather than overcoming the Coulomb barrier of two
neon nuclei. This trend continues in the silicon burning phase. In general, a
photo-disintegration channel becomes important when the temperature rises
to the point that the Q-value, i.e. the energy difference between the fuel and
the products is smaller than approximately 30kB T ( [71]).
With typical Q-values for reactions among stable nuclei above silicon be-
ing 8-12 MeV, photo-disintegration of the nuclear products of neon and oxy-
gen burning begins to play an important role once the temperature exceeds:
T9 ≥ 3. Then nuclei with smaller binding energies are destroyed by photo-
dissociation in favor of their more tightly bound neighbors, and many nu-
clear reactions involving α-particles, protons and neutrons interacting with all
the nuclei in the mass range A = 28 − 65 take place. In contrast to the previ-
ous burning stages where only a few nuclei underwent thermonuclear reactions
upon themselves, here the nuclear reactions are primarily of a rearrangement
type, in which a particle is photo-ejected from one nucleus and captured by
another and a given fuel nucleus is linked to a product nucleus by a multi-
tude of reaction chains and cycles and it is necessary to keep track of many
more nuclei (and many reaction processes involving these) than for previous
248 A. Ray

burning stages. More and more stable nuclei form in a nuclear reaction net-
work as the rearrangement proceeds. Since there exists a maximum in the
binding energy per nucleon at the 56 F e nucleus, the rearrangements lead to
nuclei in the vicinity of this nucleus (i.e. iron-group nuclei).
In the mass range A = 28−65, the levels in the compound nuclei that form
during silicon burning are so dense that they overlap. Moreover, at the high
temperatures that are involved (T9 = 3−5), the net reaction flux may be small
compared to the large forward and backward reactions involving a particular
nucleus and a quasi-equilibrium may ensue between groups of nuclei which
are connected between separate groups by a few, slow, rate-limiting reactions
(“bottlenecks”). However, as the available nuclear fuel(s) are consumed and
thermal energy is removed due to escaping neutrinos, various nuclear reac-
tions may no longer occur substantially rapidly (“freeze-out”). Thielemann
and Arnett [72] found that for cores of massive stars in hydrostatic cases, the
bottlenecks between quasi-equilibrium (QSE) groups coincided with Z=21 nu-
clei whereas for lower mass stars, lower temperatures and Ye and higher den-
sity this bridge involved neutron rich isotopes of Ca capturing protons. Hix
and Thielemann [71] discussed and contrasted these results with those of
earlier workers; in general the reaction flow across the boundary of the QSE
groups are influenced by the neutronization of the material, i.e. the overall Ye .
It is in this context that weak interaction processes such as electron capture
and beta decay of nuclei are important, by influencing the Ye and thereby the
reaction flow. These ultimately affect both the stellar core density and en-
tropy structures, and it is important to track and include the changing Ye of
the core material not only in the silicon burning phase, but even from earlier
oxygen burning phases. The calculation of stellar weak processes on nuclei has
spawned extensive literature (see [73], [74], [75] etc., and for a review [76]).
Iron, nickel and neighboring nuclei which are the products of the hydro-
static Si burning in the core are mostly trapped in the collapsing core that
ends up as the compact remnant and little of this may reach the interstellar
medium. However, the blast wave shock launched after the core bounce im-
pacts through the onion-like outer shells that the star’s earlier evolution has
left behind. The nearest part of the Si shell that is still unburned is heated to
sufficient temperatures to form iron peak nuclei. Regions which are not as close
can undergo incomplete Si burning and be left with substantial amounts of Si,
S, Ar, Ca and Ti. Three separate outcomes may result depending upon the ini-
tial density and the peak temperature: 1) incomplete Si burning; 2) a “normal
freezeout” and 3) an “α- rich freezeout”. In the first case, with initial tem-
peratures of about 5 × 109 K and peak density of ρ = 109 g cm−3 , significant
amounts of Si and other intermediate-mass elements remain after the charged
particle reactions freezeout in the expanding ejecta. In the normal freezeout,
an initial condition of 7 × 109 K and peak density of ρ = 109 g cm−3 leads to
complete Si burning. The “α-rich freezeout” however takes place at lower peak
densities ρ = 107 g cm−3 though similar peak temperatures: 7 × 109 K. This
ends up with an abundance of 4 He nuclei which produces a significant flow
 Thermonuclear fusion and collapse and explosion in massive stars 249

through the triple alpha reaction into intermediate mass nuclei, in addition
to the iron group nuclei products. The feedback between the rate of nuclear
recombination and that of temperature evolution in the expansion critically
controls the “α-richness” of matter and production of important radioactive
nuclei e.g. 44 T i and is one of the challenges of computational simulation of
silicon burning. For a recent discussion of the computational aspects of the
nuclear evolution during silicon burning, see [77].
In summary, a few key points concerning the thermonuclear burning of
28
Si are as follows:-
• Direct thermonuclear fusion of two 28 Si nuclei does not take place be-
cause their Coulomb barrier is too high. Instead thermonuclear fusion takes
place by successive additions of α-particles, neutrons and protons.
• Although this is actually a large network of nuclear reactions it is
called “silicon burning” because 28 Si offers the largest resistance to photo-
dissociation because of its highest binding energy among intermediate mass
nuclei.
• The source of the α-particles which are captured by 28 Si and higher
nuclei is 28 Si itself. Silicon, sulphur etc. partially melt-down into α-particles,
neutrons and protons by photo-dissociation. These then participate in reaction
networks involving quasi-equilibrium clusters linked by “bottleneck” links.
• Although beta decay and electron captures on stellar core nuclei do not
produce energy in major ways they nevertheless play a crucial role in shifting
the pathways of nuclear and thermodynamic evolution in the core conditions.
These ultimately determine the mass of the core and its entropy structure
which finally collapses in a supernova explosion.

8 Core collapse SNe: electron capture and neutrinos

At the end of nuclear burning in the core of a massive star, inert Iron group
nuclei are left in the innermost region, – inert because nuclear burning cannot
extract any further energy from the most tightly bound nuclei. At the last
stages of nuclear burning, neutrino cooling which is far more efficient than
the radiation of photons from the surface, would leave the core compact and
degenerate [78] (where electron Fermi energy is much larger than the thermal
energy kT ) due to a number of processes such as e+ /e− pair annihilation
process, photoneutrino process and plasmon decay into neutrinos. These neu-
trinos escape from the star freely at this stage since their interaction cross
section is so small, much smaller than that of the photons. In the picture
of the late stages of a spherical star that B2 FH [79] proposed one had an
onion-skin model: nested shells of progressively heavier elements, with the
densest iron core at the center. Successively high-Z elements are ignited in
increasingly central parts of the core in decreasingly lower specific entropy
environments, and are contained within entropy barriers left behind by previ-
ous generations of nuclear burning. The degenerate core becomes unstable to
250 A. Ray

gravitational collapse at roughly the same Chandrasekhar mass (modulo rela-

tively small variations in the electron fraction or entropy profiles in the core)
practically independent of the total mass of the star, which for example could
range from ∼ 8 M to ∼ 60 M [67]. This “core convergence” establishes
the important connection between the Chandrasekhar mass and the masses
of neutron stars that are formed from core-collapse supernovae (though the
details of the boundaries of the main sequence masses that lead to neutron
stars, as well as the composition of the cores, e.g. O-Mg-Ne cores vs iron
cores etc., have evolved as further research was undertaken). The supernova
explosion itself is triggered by the collapse of the core [80], followed by the
sudden release of gravitational energy. When the iron core grows in size, its
temperature increases, and when T > 7 × 109 K, iron nuclei photo-dissociate
in endothermic reactions (consuming energy, even though the free energy F =
U -TS decreases due to increased entropy) into alphas and neutrons. This
leads to the effective adiabatic index decrease below 4/3 and the iron core
becomes unstable to collapse. Collapse initiation through photo-dissociation
happens in relatively massive cores of more massive stars whereas in less mas-
sive stars the collapse may initiate due to reduction of electrons (which provide
the bulk of the supporting pressure due to their high Fermi momenta) as they
are captured by nuclei in “inverse beta decay” in regions where the electron
Fermi energy exceed the capture thresholds.
The core collapses on a dynamical scale until the infall is suddenly halted
when the central density overshoots that of the nuclear matter, at 4 − 5 ×
1014 gm cm−3 . In between, the iron core collapsed onto itself nearly freely at
about a quarter of the speed of light. Initially it had the size of the Earth,
but towards the end it is a hot, dense, neutron rich sphere about 30 km in
radius (see [81] for a review). A static accretion shock wave forms initially at
the edge of the quasi-free falling core, which soon starts propagating outward
through the outer core and mantle as more kinetic energy is brought into it
by infalling matter. The shock wave soon stalls (at least for most of the range
of main sequence masses beyond about 10 M ) which may however be helped
after about 500 milliseconds (“after a pause that refreshes” – according to
Hans Bethe) [82] by neutrinos which are freely streaming out from the inner
core, but manage to deposit enough energy (albeit a small fraction of their
total energy) “reviving” the shock. During this post-bounce phase the proto-
neutron star radiates away most of its energy (3 × 1053 erg) in the form of
neutrinos and antineutrinos of various flavors and this accounts for the larger
intrinsic energy per unit mass of the central engines of core collapse SNe.

8.1 Electron capture on nuclei and protons: a core thermometer

During the gravitational collapse, the entropy of the core stays low, which
permit the nuclei of various elements present in the core to (largely) survive
thermal disintegration and coexist with a small fraction of “dripped” nucle-
ons ( [83] hereafter BBAL). Around the density 5 × 1011 gm cm−3 , neutrinos
 Thermonuclear fusion and collapse and explosion in massive stars 251

produced through electron capture on these nuclei and free protons are
trapped in the core. Much of the information pertaining to the conditions
in which the neutrinos are originally produced, such as the nuclear and ther-
modynamic properties of the core of the supernova are altered because these
neutrinos undergo inelastic scattering with the overlying stellar matter in the
post neutrino trapping phase. Neutrinos which are emitted through electron
captures on the nuclei present in the pre-supernova and collapsing core before
it reaches neutrino trapping density ( 1012 gm cm−3 ) [84], [85], however,
stream freely through the stellar matter without any further interactions.
These pre-trapping neutrinos carry with them information on both physical
conditions within the core, as well as it’s nuclear configuration e.g. the ratio of
the number density of free protons to that of heavy nuclei. The last quantity
can depend on the nuclear equation of state relevant to collapse. Since neu-
trinos act as probes of the dynamic, thermodynamic and nuclear properties
of the pre-supernova and collapsing core, their detection and measurement of
energy spectrum can have significant implications. The time evolution of the
detected spectrum could also reveal the dynamical time scale – a clue to the
average density and mass of the stellar core which may have implications for
neutron star vs black hole formation18 . The reduction of lepton fraction dur-
ing stellar collapse has implications for shock formation stages and the overall
dynamics - even in the delayed explosion stage, since it determines, through
the original energy of the bounce shock, and the entropy profile in the outer
core, the position of the stalled shock.
The loss of neutrinos at low and intermediate densities is important in
determining the saturation ratio of the lepton to baryon ratio at the time
of core bounce (the leptons determine the pressure in the core whereas the
baryons, mainly nucleons, determine the mass of the homologous collapsing
core). Brown et al. [87] argued that the hydrodynamic collapse is nearly ho-
mologous, i.e. the density structure of the collapsing core remains self-similar
throughout the collapse until the time of bounce. This greatly simplifies the
study of various processes during the collapse and quantities in the core can
be calculated (largely semi-analytically) through the evolution of a “mean”
“one-zone” symbolizing the core properties (see e.g. [88], [89]).
The core of a massive star collapses under its own gravity when the pres-
sure support from degenerate electrons is reduced through the capture of
electrons in the stellar material. The electron capture on neutron rich heavy
nuclei initially proceeds primarily through the allowed type (Δl = 0) Gamow
Teller transitions. As core density exceeds  1011 gm/cm3 , the nuclei be-
come more and more massive and too neutron rich to allow e− -capture to
take place through allowed Gamow-Teller transitions from the ground state.
This is because the allowed states for p to n transition within the nucleus
are already filled by the neutrons (neutrons shell blocked) and the transi-
tion strength for typical captures like 56 Fe → 56 Mn used earlier (as in [90])

18
The collapse of a star could in principle also lead to a “naked singularity” [86]
252 A. Ray

is no longer representative of typical nuclear e− -capture rates. It was shown

[91],[92] that the dominant unique first forbidden transition strength was actu-
ally negligible compared to the thermally unblocked strength under the typical
core collapse conditions. Therefore, after neutron-shell blocking sets in, (when
(A,Z) > 74 Ge) the sum rule for the Gamow Teller transition operator |MGT |2
decreases from a typical value of 2.5 [90], [88] to about 0.1. The e− -capture
rate on a single nucleus X(A, Z) in the initial state i to the final state j is:

fi (T, μe , Qij )
λij = ln 2 (56)
f tij
where f tij is related to |MGT |2 by f tij = 3.596 − log |MGT |2 for allowed
Gamow-Teller type transitions (for free protons, log ftf.p. = 3.035). The fac-
tor fi (T, μe , Qij ) is the phase space factor for allowed transition, which is a
function of the ambient temperature T , the Fermi-energy of the electron μe
and the Q-value for the reaction. The neutrino energy is Eνe = Ee − Qij .
The change in entropy during collapse controls the fraction of the dripped
protons with respect to that of the heavy nuclei and this influences the overall
neutrino spectrum received on earth as the spectrum of neutrinos generated
by electron capture on protons are different from captures on heavy nuclei.
The received neutrino spectrum depends not only upon the initial conditions
from which the collapse started, but also on the details of the electron capture
properties of the stellar matter. Properties of nuclei at finite temperatures and
density during this phase of the collapse, where shell and pairing corrections
are relevant were computed in [94] and utilized to evolve self-consistently
with the electron capture physics and the consequent changes in nuclear and
thermodynamic variables.
The rate of generation of neutrinos per nucleon within energy band Eν
to Eν + dEν after accounting for the relative abundance of free protons and
nuclei, is:

dYν (Eν ) = dλf p (Eν )Xp + dλH (Eν )(1 − Xn − Xp )/A (57)
here A represents the Atomic weight of the ensemble of nuclei present in the
core, taken to be represented by a single “mean” nucleus as in [90]. The
differential neutrino production rates for free protons and heavy nuclei are:


log 2 < G > Eν2 (Eν + Qf p,H ) (Eν + Qf p,H )2 − (me c2 )2 )
dλf p,H = dEν
(f t)f p,H (me c2 )5 (1 + exp(Eν + Qf p,H − μe ))
(58)
where the Coulomb correction factor < G > has been taken as ≈ 2 for heavy
nuclei and 1 for free protons. The Q-value is given as: Q = (μ̂ + 1.297 + EGT )
assuming that the strength is concentrated in a single state and here μ̂ (= μn −
μp ) is the difference in the neutron and proton chemical potentials when free
nucleons coexist with a distribution of neutron rich nuclei in nuclear statistical
 Thermonuclear fusion and collapse and explosion in massive stars 253

equilibrium, and EGT is the energy of the Gamow-Teller Resonance centroid.

The centroids in fp-shell nuclei, found from experimental data from (n,p)
reactions have been used for characterizing GT transitions in these nuclei [93]
and are close to the value (3 MeV) used here. The GT centroid in the e-capture
direction is itself a function of the ambient temperatures as a quasi-particle
random phase approximation (QRPA) calculation shows [95]. At the relevant
high densities, when neutron rich nuclei with A > 65, are abundant and the
electron chemical potential is noticeably larger than typical nuclear Q-values,
electron capture rates are mainly sensitive to the centroid and total strength
of the GT+ distributions – these are reasonably well described within the
random phase approximation [96]. We note that the temperature dependence
of the nuclear symmetry energy can also affect the neutronization of the stellar
core prior to neutrino trapping during gravitational collapse [97], [98] since
ambient temperatures can reach upto several MeV. Not only the reaction
Q-values but also the equation of state of bulk dense matter, the free nucleon
abundances, the degree of dissociation into alpha-particles and the nuclear
internal excitations are modified by changes in the symmetry energy.
The Fermi-energy μe = 11.1(ρ10 Ye )1/3 MeV. The difference in chemical
potentials μ̂, and the relative fraction of free protons are obtained from a low
density analytic equation of state similar to that in [90] with modifications
noted in [88]. Shown in Fig. 15 is a typical “snapshot” spectrum of neutrinos
from a 15 M star’s core collapse within a narrow range of stellar core density
around 1011 gm cm−3 . Note the two separate peaks from captures on free
protons and heavy nuclei which have non-thermal spectra.

8.2 Number of neutrinos emitted and predictions of detections

The number of neutrinos and their energy distributions that could have been
detected by large underground neutrino experiments like the Super-Kamioka
experiment and the Sudbury Neutrino Observatory (SNO)19 has been cal-
culated [99]. The spectrally integrated fluence of νe at a distance of 1 kpc,
as Ye changes from 0.42 to 0.39 in a 1.4 M stellar core (of a 15 M star)
is: Fν = 4.2 × 1011 cm−2 . The energy of the infall neutrino burst up to this
stage is: Eνe = 7.2 × 1050 erg. The flux, direction and the spectra of the neu-
trinos could have been measured by the charge current dissociation of the
deuterium nucleus (νe (d, pp)e− ) in the (“classic”) SNO [100] with a fiducial
mass of 1 Kt of high purity D2 O. It would have been also possible to detect

19
The original SNO experiment with heavy water has finished taking data. Thus
the heavy water based predictions if a nearby supernova took place are only
indicative. However there are plans of an extended SNO (called SNO+) which
will use 1 kTon of liquid scintillator which will greatly reduce the lower energy
threshold of neutrinos and will remain sensitive to neutrinos from a nearby SN.
Moreover, after an accident the photomultiplier tubes in Super-Kamioka, the low
energy thresholds originally designed for are no longer operative.
254 A. Ray

7.0 Combined Spectrum

νe fluence (10–8 MeV–1 / baryon)

Heavy nuclei peak
6.0 Free proton peak

5.0

4.0

3.0

2.0

1.0

0.0
0 5 10 15 20 25
Neutrino Energy Eνe (MeV)

Fig. 15. “Snapshot” neutrino fluence (MeV−1 baryon−1 ) in a density interval

Δρ10 = 0.0002 around ρ10 = 9.8668 gm cm−3 for a 15 M star (D = 1 kpc) [99]

νe and obtain spectral information by means of the neutrino-electron scat-

tering reaction (νe + e− → νe + e− ) in SNO as well as in the light water
detector Super-Kamioka. Apart from the reactions mentioned above which
can perform spectroscopy (i.e. measure the incoming neutrino energy), the
neutral current dissociation of d by the reaction νe (d, pn)νe could have ob-
tained the total neutrino flux (of all flavors). The number of νe events which
could have been detected in the SNO detector through neutrino-induced c.c.
reaction on the target d nuclei is given by: nνe = Fν σcc (νe )Nd where Nd
(= 6.02 × 1031 ) is the total number of target nuclei present in the 1Kt de-
tector. The charge current and neutral current cross-sections (σcc (νe ) and
σnc (νx ) respectively), have been computed for the ν-d process by Bahcall et
al. [101] and accurate fits to these cross-sections between 5 to 40 MeV are
given as [102]: σi = αi (ν − th,i )2.3 where i = cc and nc, αcc = 1.7 × 10−44
cm2 , αnc = 0.85 × 10−44 cm2 , th,cc = 2.2 MeV and th,nc = 1.44 MeV.
For the H2 O based Cerenkov detector (Super-Kamioka) the (νe , e− ) scat-
tering events would be the main source of νe spectral information since the
corresponding energy thresholds for charge current and neutral current inter-
actions for ordinary water are much higher. (During the collapse stage, the
neutrino flux is almost entirely in neutrinos of the electron type; anti-neutrinos
of various kinds, as well as neutrinos of the mu or tau type are generated in
copious numbers only in the hot post core bounce phase). The relevant (νe , e− )
scattering cross-section is [103]: σe = (1/2)(4G2 m2e h̄2 /πc2 )(7/12)(ν /me c2 ).
 Thermonuclear fusion and collapse and explosion in massive stars 255

The number of detections by Super-Kamioka (mass 32 kt) and SNO for a

supernova explosion 1 kpc away, for several possible scenarios of stellar core
collapse are reported in Table 2 [99]. The 15 M star collapse is initiated
from thermodynamic conditions as in [88] (Yei = 0.420, Si /kB = 1.00, Ti =
0.7178, ρ10 = 0.37), while the 25 M star’s single zone initial conditions are
similarly derived from the data reported in [104] and an expression for the
core averaged entropy (Yei = 0.423, Si /kB = 1.14, Ti = 0.6756, ρ10 = 0.15).

Table 2. Pre-trapping neutrino detections predicted in SNO (heavy water) and Su-
per Kamioka [99] with hardness ratios up to ρ10 = 24.16 for indicated heavy nuclear
e-capture matrix elements for 15 M Fuller (1982) and 25 M WWF presupernova
stars. Note the caveat in footnote 20
Star Mass |MGT |2 tcollapse Pre-trapping Variables No. Detected Hardness Ratio†
(ms) Yef Sf /kB SNO S-K SNO S-K
15 M 1.2/0.1 120 0.3909 1.0021 82 394 0.2786 0.8540
2.5/0.1 120 0.3896 1.0085 66 344 0.2876 0.9537
25 M 1.2/0.1 190 0.3828 1.1080 120 566 0.2878 0.8319
2.5/0.1 190 0.3813 1.1204 99 499 0.2916 0.9190

†
The hardness ratio denotes the number of neutrino events in the 5 MeV ≤ Eνe ≤ 12
MeV and 12 MeV ≤ Eνe ≤ 25 MeV bands.

Neutrino spectroscopy of the final state of a star would be possible provided

that the event occurs at a relatively close distance. Although, a priori, these
may be rare events, there have been a number of historical Supernovae, as
well as detected radio pulsars within a distance of about 2 kpc. There are a
number of star forming regions nearby (such as the Orion complex - about
440 pc away), which are sites of potential supernova progenitor. A detection
by underground neutrino experiments would constrain features of theoretical
calculations of both collapse and explosion era of type II Supernovae as well as
shed light on the characteristics of the stellar core. The 19 neutrinos detected
from SN1987A were most likely to have been emitted during the post-bounce
phase as their total fluence during the proto neutron star cooling phase (at
 1058 ) is much larger than that during the collapse phase ( 1056 ).

9 Detected neutrinos from SN 1987A and future

neutrino watch

Gravitational collapse of the core of the massive star under its own grav-
ity leads to a supernova explosion. These are extremely energetic explosions
where the observable energy in the kinetic energy of the exploded debris and
electromagnetic radiation add up to several times 1051 erg. The actual energy
256 A. Ray

scale is typically 3 × 1053 erg or higher, but most of this is radiated away
neutrinos. Although the full understanding of the process of explosion in a
gravitational collapse leading to a supernova has not been achieved despite
several decades of theoretical and computational work, a watershed in the field
was achieved observationally when a supernova exploded close by in a satellite
galaxy of our own, namely SN1987A in the Large Magellanic Cloud (LMC).
A few neutrinos were detected from this supernova [105], [106], which were
the first detections of the astrophysical neutrinos from outside of our solar
system. By using the energetics of the neutrinos, their approximate spectral
distribution, the distance to the LMC it was possible to show that the overall
energy of the explosion was indeed ET ∼ 2 − 3 × 1053 erg.
In addition, the duration of the neutrino burst was of the order of a few
seconds as opposed to a few milliseconds, signifying that the observed neu-
trinos had to diffuse out of the dense and opaque stellar matter as predicted
theoretically, instead of directly streaming out of the core. The spectral char-
acteristics also indicated that the object that is radiating the neutrinos was
indeed very compact, roughly of the same dimensions as that of a protoneu-
tron star. Thus SN1987A provided the observational confirmation of the broad
aspects of the theoretical investigation of stellar collapse and explosion. For a
review of the understanding of the astrophysics of SN1987A, see [107].
Physicists are now gearing up to detect not only another supernova in our
own galaxy, but by hoping to build very large neutrino detectors, they aim
to detect supernova neutrinos from the local group of galaxies ( [108], [109]).
As neutrinos from the supernova travel directly out from the core, they arrive
a few hours ahead of the light flash from the exploding star, since light can
only be radiated from the surface of the star, and it takes the supernova
shock launched at the deep core several hours to reach the surface. In the
case of SN1987A, this time delay allowed the estimation of the size of the star
that exploded. Thus some advance warning of the supernova in the optical
band can be gotten from a “neutrino watch”. Physicists have now connected a
worldwide array of neutrino detectors through the Internet (SN Early Warning
System or SNEWS20 ) which will notify astronomers to turn their optical, UV
and other telescopes to the right direction when they find a burst of neutrinos
characteristic of a supernova explosion.

10 What X-ray spectroscopy reveals about

nucleosynthesis in SNe and SNRs
X-rays from a supernova explosion arise from the interaction of the supersonic
ejecta with the circumstellar medium (CSM). The CSM typically consists
of a slow-moving wind. When the ejecta collides with the CSM, it creates
two shocks: a high-temperature, low-density, forward-shock ploughing through
20
See the site: http://hep.bu.edu/∼snnet/
 Thermonuclear fusion and collapse and explosion in massive stars 257

the CSM (known as blast-wave shock) and a low-temperature, high-density,

reverse-shock moving into the expanding ejecta. Initially the X-rays come from
the forward-shocked shell dominated by continuum radiation, but after a few
days X-rays arise also from the reverse-shock, which can have substantial line
emission, thus providing nucleosynthetic fingerprints of the ejecta. The tem-
poral evolution of the X-ray luminosity of a supernova can yield information
on the density distribution in the outer parts of the exploding star (ρ ∝ r−n ,
here n can be in the range 7 − 12, typically n ∼ 10 for a Blue Supergiant
(BSG) and n ∼ 12 for a Red Supergiant (RSG)- see [110]). These studies are
therefore of interest to stellar structure and evolution.
To date thirty-six supernovae have been detected in the X-ray bands21 .
Among these the most extensively studied is SN 1987A because it was so
bright. Another extensively studied object, of the supernova remnant (SNR)
kind is Cassiopeia A. A SNR is just an older supernova after it has picked
up enough material from the surrounding medium, which slows it down. The
debris of the explosion, or the ejecta, radiates when it is reheated after an
initial cooling, having been hit by a inward propagating reverse shock.

10.1 Supernova Remnant Cassiopeia A

Cassiopeia A (Cas A for short and also known as 3C461 or G111.7-2.1) is

the second youngest SNR in our Milky Way galaxy. It was believed to be the
product of a SN explosion in ∼ 1672 [112] only about 3.4 kpc away [113]. The
British astronomer John Flamsteed may have recorded it as a sixth magni-
tude star in 1680, but it may have faded rapidly after explosion which could
have acted against its widespread reportage. It is a shell type SNR which was
rediscovered by radio astronomers Ryle and Smith [114] and is the brightest
extra-solar radio object. Cas A is an oxygen-rich SNR with heavy element
distribution typical of a core-collapse SN. Its closeness, young age and high
brightness across the whole electromagnetic spectrum have underscored its
importance for studying supernovae22 . A recent optical spectrum of the orig-
inal supernova near maximum brightness, obtained from scattered light echo
more than three centuries after the direct light of the explosion was received on
Earth, showed it to be a type IIb SN [116]23 , somewhat like the well studied

21
See S. Immler’s X-ray supernova page at
http://lheawww.gsfc.nasa.gov/users/immler/supernovae list.html and [111].
22
Here we discuss thermal emission from Cas A, which directly connects to nu-
cleosynthetic products. Many shell-type SNRs, including Cas A, also show non-
thermal synchrotron X-rays - see [115].
23
Krause et al. note that even with the overall lack of hydrogen emission in most
knots and the nitrogen enrichment in the remnant which is widely interpreted
as signatures of the collapse of a Wolf-Rayet star, i.e. a type Ib SN [117], Cas
A cannot be classified as arising out of a type Ib SN, since Cas A light echo
spectrum does not match well the spectrum of the prototype SN 2005bf.
258 A. Ray

prototype SN 1993J24 [123]. The estimated mass loss rate of 2×10−5 M yr−1
and a wind velocity of 10 km s−1 consistent with the hydrodynamical state
of the Cas A remnant [124] is also similar to what has been interpreted from
radio and X-ray observations of SN 1993J [125], [126].
SNR Cassiopeia A has a bright clumpy emission ring with a diameter of
about 3 . This is associated with the SNR ejecta, whereas at an outer diameter
of about 5 , there is fainter emission in a filamentary ring which is due to the
forward (“blast-wave”) shock in the circumstellar medium. The SNR has a
nearly circular appearance, though the ejecta shows a bipolar structure, and
jets (mainly in the NE, i.e. upper left of the image and a fainter extension in
the SW direction, called the “counterjet”) that were formed by the explosion
(detected both in the optical [127] and in the Si line X-ray emission [129],
[128]). The remnant has been extensively observed in the optical and more
recently with a million second exposure of the Chandra X-ray Observatory
[128] (see also XMM-Newton [130]) and includes high resolution grating X-
ray spectroscopy with Chandra [131] as also the Spitzer Space Telescope using
the Infrared Array Camera [132] and the IR Spectrograph [133].
Many of Cas A’s optically bright knots have been identified as shocked
ejecta which is still visible due to its young age. The optical radiation emitting
regions have been classified into two groups: the so called fast moving knots
(or ’FMK’s with speeds 4000 km s−1 < v < 15000 km s−1 ) and the slow
moving quasi-stationery flocculi (or ’QSF’s with speeds v < 300 km s−1 ). The
FMKs believed to comprise of SN ejecta, have H-deficient emission (lacks in
Hα) which is dominated by forbidden lines of O and S emission [134] while
the QSF emission is rich in N, and is believed to originate from the gas lost
from the star before its explosion (circumstellar envelope) which is now hit
by the blast wave shock [131].
Cas A shows the nucleosynthesis products of both hydrostatic and explo-
sive nucleosynthesis. C and O are produced in He burning, Ne and Mg first
appears through C burning, and O and Al are added with Ne burning [135].
Spectral lines of heavier elements: Si, S, Ar, Ca etc. are seen. These are pro-
duced in O burning, with alpha reactions on Mg also contributing, – they are
from zones where explosive O-burning and incomplete explosive Si-burning
occurs. Fe group elements are produced in the “Si burning” chain of reactions
and result from complete and incomplete Si-burning. Much of the layered
24
The collapse of a red supergiant [118] in a binary system [119] with the larger
star with a mass on the main sequence of ∼ 10 to 20 led to SN 1993J and could
explain its light curve and other characteristics [120], [121]. Photometry and
spectroscopy at the position of the fading SN with Hubble Space Telescope and
Keck Telescope, a decade after its explosion showed the signature of a massive
star, – the companion to the progenitor that exploded. While the binary system
initially consisted of e.g. a 15M star and 14M star in an orbit of 5.8 yr, at the
time of the explosion, the primary had a mass of 5.4M (with a helium exhausted
core of 5.1M ) and a secondary which gained mass in transfer and which we still
see today, ended up with 22M [122].
 Thermonuclear fusion and collapse and explosion in massive stars 259

nucleosynthetic structure has been preserved during the explosion process in

Cas A, with layers of N-, S-, O-rich ejecta seen beyond the outer shock [134].
However, in the optical bands, “mixed emission knots” showing N and S lines
suggest that clumps of high speed S ejecta have penetrated through outer
N-rich layers [134]. X-ray line data show that for example, in the southeast
region of the SNR, Fe is farther out and is moving faster than the Si/O region.
Hughes et al. [136] consider the high surface brightness knots enriched in Si
and S are consistent with nucleosynthetic products of explosive O burning,
whereas elsewhere, the more diffuse, lower surface brightness features with
enhanced Fe could be the result of explosive Si burning.

1 ENERGY (keV) 5

105
COUNTS PER CHANNEL

Fe L Si
S

Ar
104 Ca

Fe K
1000

100

500 1000
PULSE HEIGHT
Fig. 16. Spectrum of the entire supernova remnant Cassiopeia A taken by Chandra
(AXAF CCD Imaging Spectrometer (ACIS)) on January 30-31, 2000 against pulse
height (lower x-axis) in the spectral channel and photon energy (upper x-axis) using
a gain value of 4.8 eV per channel (Fig. reproduced by permission of the AAS and
courtesy of U. Hwang [129])

The “non-dispersed”25 X-ray CCD spectrum for the entire Cas A remnant
is shown in Fig. 16. A 50 ks exposure in January 2000 collected approximately
16 million photons during the observation of the SNR. It shows prominent
lines of Si, S, Ar, and Ca, in their Helium-like ionic state and undergoing
(n = 2 → n = 1) transitions as also the L and K transitions of Fe.
Maps were constructed of the entire remnant in the spectral line energies
of several specific elements. The Si and S X-ray line maps and the optical
25
This means the X-ray beam did not pass through any dispersive element like a
grating, but the spectrum is obtained by measuring the energy of the incident
X-ray by the intrinsic spectral sensitivity of the CCD.
260 A. Ray

Fig. 17. Top: Three-color image of Cas A with red = Si Heα (1.78–2.0 keV), blue =
Fe K (6.52–6.95 keV), and green = 4.2–6.4 keV continuum. The remnant is roughly
5’ across. Bottom left: Overexposed broadband image showing faint features. The
spectral regions are indicated (top left of this box: northeast jet; bottom left of
this box: Fe-rich region; lines at bottom right point to two southwest jet filaments).
Bottom right: On the same scale, the ratio image of the Si Heα (1.78–2.0 keV) and
1.3–1.6 keV (Mg Heα, Fe L), without subtraction of the continuum contribution.
The image highlights the jet and counterjet traced by Si emission, although features
at the lowest intensity levels are uncertain. (Figure reproduced by permission of the
AAS and courtesy of U. Hwang [128])

maps resemble each other and this demonstrated that the X-ray and optically
emitting ejecta are largely spatially coexistent. In fact the X-ray Si and S line
emission, expected to include a significant contribution from the ejecta, shows
asymmetric Doppler shifts corresponding to bulk velocities of ∼ 2000 km s−1
which are comparable to those of the optical ejecta knots [129]. Si, S, Ar, and
Ca have similar general morphologies, but examination of their images in their
line energies also show significant variations. At the positions of the brightest
knots in the Si image at the inner boundary of the shell to the northeast and
southeast the corresponding Ar and Ca X-ray images are much weaker in the
remnant. Hughes et al. [136] using the Chandra first light observation of Cas
A (∼ 5000s on 1999 August 20) showed that the Fe-rich ejecta lie outside the
S-rich material, and claimed that this is due to extensive, energetic bulk
 Thermonuclear fusion and collapse and explosion in massive stars 261

motions which caused a spatial inversion of a part of the supernova core. How-
ever, on the basis of Spitzer Space Telescope imaging and spectral data Rud-
nick and collaborators [137], [133] claim that there are two roughly spherical
shocks, (the blast wave and the reverse shocks) but “the wide spatial variations
in composition seen in infrared, optical and X-ray studies are not due to local
differences in ionization age or temperature, but instead reflect very specific
asymmetries in the geometry of the underlying explosion. These asymmetries
are ones in which the nucleosynthetic layers have remained mostly intact along
each radial direction, but the velocity profiles along different radial directions
vary over a range of approximately five” (see also [132]). In some directions,
only the upper C-burning layers have been probed by the reverse shock, while
in other directions, deeper O- and Si-burning layers have reached the reverse
shock.

150

100

50
Δδ (arcsec)

–50

–100

–150
150 100 50 0 –50 –100 –150
Δα (arcsec)

Fig. 18. Composite Spitzer image of Cas A in [Ar III] (blue), [S IV] (green), and [Ne
II] (red), showing two pronounced neon-rich crescent-shaped regions to the N and
S. The size of the image is 320” with north to the top and east to the left. Shown
are the kinematic center of the remnant at the magenta circle and X-ray localization
of the remnant’s compact object (red square, 7” to the south). Green lines indicate
the 1σ range of the “kick vector” direction of the compact object from the ejecta’s
expansion center: 169o ± 8.4o [112]. The two regions of enhanced neon abundance lie
very close to this projected direction. Several much smaller neon-enhanced regions
lie in the West along the X-ray jet direction (Fig reproduced by permission of AAS
and courtesy of J. Smith [133])
262 A. Ray

The gas infrared lines of Ar, Ne, O, Si, S and Fe are seen at different
locations, along with higher ionization states of the same elements visible in
the optical and X-ray lines. For example, strong [Ar II] regions in the IR match
very closely the places where helium- and hydrogen-like ionization states of
Si and S are seen in the X-ray and [S III] seen in the optical [134]. These
are the same regions with multiple temperatures and ionization states but
show products of oxygen burning instead of carbon burning. Other regions,
for example the crescent-shaped green feature slightly east of south in the
Spitzer IRAC line and continuum images depicting gas in Fig 2 of [132] show
strong [Ne II] emission in the IR (see also the red regions in the image Fig 18
reproduced from [133]), increasing [O III] emission in the optical and a gap in
the silicon dominated X-ray emission, are interpreted as the locations where
carbon-burning layers are presently encountering the reverse shock. Different
layers containing various types of nucleosynthetic products seems to show the
presence of characteristic types of dust, e.g. the deep layers contain silicates
while the upper layers contain dust dominated by Al2 O3 and carbon grains.
They also find evidence for circumstellar dust heated by the blast wave shock
[137].
Smith et al. [133] identify IR line emission from ejecta materials in the
interior, prior to their encounter with the reverse shock, as well as from the
post-shock bright ring. There is a dramatic increase in the electron density (by
a factor ≥ 100 to ∼ 104 cm−3 ) as well as a concomitant change in the ionization
state of the ejecta as it encounters the reverse shock. There is a clear layering of
ionization state, from low ionization species in the interior, e.g. [Si II] (8.2 eV),
higher energies on the IR bright ring [S IV] (34.8 eV) where optical emission
is also seen, and very high energies traced by X-ray line emission from H-
like and He-like K-alpha resonance lines of Si XIII (0.5 keV) and Si XIV (2.4
keV), extending beyond the IR-bright rim. In addition, they find two compact,
crescent shaped clumps with high neon abundance (mentioned above) which
are arranged symmetrically around the central neutron star, and the crescent
regions are closely aligned with the kick direction of the neutron star from the
remnant’s expansion center. These regions contain a huge amount of ionized
neon (dominated by Ne+ and Ne++ , and excluding neutral neon), almost
1.8 × 10−4 M , flowing outwards 20 degrees from the plane of the sky at
roughly −5500 km s−1 in the south and +4200 km s−1 in the north, while
the entire SNR may contain an ionized neon mass ∼ 8.6 × 10−4 M . Smith
et al. comment that the apparent macroscopic elemental mixing mentioned in
[136], may actually arise from different compositional layers of ejecta passing
through the reverse shock at present along different directions.

X-ray grating spectra of Cassiopeia A and SN 1987A

In the optical bands, thousands of individual knots have been observed to yield
kinematic information, whereas the X-rays probe more dynamically important
information since a much larger fraction of the ejecta mass is probed by X-rays
 Thermonuclear fusion and collapse and explosion in massive stars 263

(∼ 4M as compared to < 0.1M in the optical). The Chandra High Energy
Grating Spectroscopy (HETGS) [131] provided for the first time the high
spatial and spectral resolution X-ray map of Cas A, comparable to what is
achievable in the optical. Because of Cas A’s bright emission lines and narrow
bright filaments and small bright clumps that stand out against the diffuse,
continuum emission, this grating X-ray spectroscopy was possible and yielded
rich information about the kinematic and plasma properties of the emitting
knots. The HETGS has two grating arms with different dispersion directions,
with the medium energy range (MEG) covering the range 0.4 − 5.0 keV and
spectral FWHM of 0.023 Å, while the high energy range (HEG) covers 0.9 −
10.0 keV and spectral FWHM 0.012 Å. The grating spectra were obtained
by Lazendic et al. on some 17 different positions on the Cas A image, mainly
near the reverse shock regions which are X-ray bright.
As mentioned already, Cas A image in the HETGS both in the non-
dispersed (zeroth-order) as well as dispersed images in different energy bands
contain H- and He-like ionization states of O+Ne+Fe (0.65-1.2 keV), Mg
(1.25-1.55 keV), Si (1.72-2.25 keV), S (2.28-2.93 keV), Ar (2.96-3.20 keV), Ca
(3.75-4.0 keV). In addition images in the Li- and Be-like states of Fe K lines
(6.3-6.85 keV) are also obtained. A key advantage of high resolution grating
spectroscopy over the low resolution “non-dispersive” CCD spectroscopy is
that the former can be used to resolve individual lines and thereby interpret
conditions in the radiating plasma in various parts of the SNR. The He-like
ions of Si and S and other elements present in Cas A are the dominant ion
species for each element over a wide range of temperatures and they emit
strong K-shell lines of these ionic stages. For typical plasma densities in the
SNR, the He-like triplet of Si and S lines shows strong forbidden (f) and
resonance lines (r) and a comparatively weaker intercombination line (i) (see
Sect. 4.5 of [138] for an illustration of these transitions in O-ions). The G-ratio
= (f +i)/r is, for example, a good diagnostic of the electron temperature [139].
Thus the measured line ratios and abundance ratios of H- and He-like ions of
the same element (say Si) were used [131] to measure the electron temperature
kTe and the ionization time scale τ = ne t (ne being the electron density and t
being for example the time since the region was hit by the shock). The results
of [131] show that for most of the selected regions, resolved spectroscopy of Si
He-like triplet lines and Si H-like lines yields temperatures around ∼ 1 keV,
consistent with reverse shocked ejecta. However, two regions, (designated R8
and R10) had significantly higher temperatures ∼ 4 keV, and could be part
of the circumstellar material. The dominant element in Cas A seems to be
Oxygen, and the electron density in the X-ray line emitting regions seem to
vary between 20 to 200 cm−3 , which is more typical of the unshocked interior
of the IR line-emitting plasma, than that at the shock front as shown in [133]
(see above). Lazendic et al. had derived unambiguous Doppler shifts for their
selected 17 regions. While the SE region of the SNR show mostly blue shifted
velocities reaching upto −2500 km s−1 (compare the IR line velocities in the
264 A. Ray

south −5500 km s−1 ), the NW side of the SNR had extreme red shifted values
up to +4000 km s−1 (compare IR line velocity in the north: +4200 km s−1 ).
Deep high resolution X-ray spectroscopy of SN 1987A was undertaken
with Chandra HETG, 20 years after its explosion [140]. The impact of the SN
ejecta with the circumstellar medium dominates the observed luminosity, the
X-ray luminosity having brightened by a factor of 25 from that in 1999 and is
presently increasing at the rate of 40 % per year [141]. An expanding elliptical
ring is seen in the X-ray image whose brightness distribution seems to correlate
with the rapidly brightening optical hot spots on the inner circumstellar ring
observed with Hubble Space Telescope. The HETG spectrum for a total life
time of 355 ksec shows H-like and He-like lines of Si, Mg, and Ne, and O VIII
lines and bright Fe XVII lines. Fig. 19 taken from Dewey et al. [140] shows
the resolved Si and Mg triplets of (r, i, and f lines); the data have similarity
with model G-ratios (f + i)/r mentioned earlier. Since the dispersed X-ray
spectrum is a convolution of the spatial structure of the X-ray image and the
motion of the X-ray emitting gas, Chandra’s sub-arc second resolution was
important to resolve the circumstellar ring of dimensions 1.2” × 1.6”.

Mg XI

Mg XI

Mg XI
Si XIII

Si XIII
XIII

XIII
Si XIII
30

30
Si

Si
Counts / bin

Counts / bin
20

20
10

10
0

6.6 6.7 6.8 9.1 9.2 9.3

Wavelength [Angstrom] Wavelength [Angstrom]

Fig. 19. Si (left panel) and Mg (right panel) triplets (resonance, intercombination
and forbidden lines) resolved by Chandra High Energy transmission Grating Spec-
troscopy. The data are shown by the solid black histogram and an arbitrarily scaled
point-source version of the two-shock model (no spatial-velocity effects) is shown in
red to suggest the relative similarity of the data and model G-ratios [(f+i)/r]. (Fig
reproduced by permission of the AAS and courtesy of D. Dewey [140])

The global fit to the HETGS data with a two shock model yielded element
abundances and absorption column densities which are consistent within the
same 90% confidence limits that were derived from LETGS by [142]. The lower
temperature shock seems to give the same kTlow for all the LETG data sets (in
2004 and 2007) and the HETG observation, whereas the kThigh values seem
to show a general evolution towards lower values among the grating datasets
and similar trend seen in the ACIS monitoring of SN 1987A.
 Thermonuclear fusion and collapse and explosion in massive stars 265

The HETG data implies a relatively low bulk radial velocity of the shocked
gas in the ring compared to model expectation from a plane parallel strong
shock entering stationary gas and the temperature range inferred from the
spectral modeling of the emitting gas. Since the X-ray image is correlated
with the optical hot spots, it appears that the blast wave ahead of the SN
ejecta is overtaking dense clumps of circumstellar gas in the hot spots. The
blast wave may be encountering the dense clumps either at normal incidence
or at oblique incidence. In the former case, the reflected shock would give rise
to gas that has been shocked twice and having nearly stationary bilk velocity
but further elevated temperature. With more and more X-ray emission coming
from gas behind reflected shocks, there would be an increase of the fraction of
higher temperature X-ray emission, consistent with what is seen. If the blast
wave encounters dense clumps at an oblique incidence, the shocked gas will
have significant velocity component parallel to the shock surface, and a signif-
icantly fast transmitted component emitting X-rays can result. The Doppler
broadening seen in the line profiles of the Chandra data seems to suggest that
both transmitted and reflected shocks encountering the circumstellar ring at
normal and oblique angles may be at play [140].

10.2 Live radioactive decays in Cas A, SN 1987A

Gamma-ray, X-ray, optical and infrared line spectroscopy of SNe and SNRs
have been used to observe nucleosynthesis and the abundance distribution of
elements freshly synthesized (both radioactive isotopes and stable decay prod-
ucts) and to extract dynamical information about the explosion. In particular
the radioactive isotopes provide unique tracers of nucleosynthetic processes
(what, where and how much) and its related dynamics. The best examples
are the observations of gamma-ray lines in supernovae, but X-ray line spec-
troscopy of decay products of radioactive nuclei have also been attempted,
and specific elements in numerous SNRs and a few SNe identified.
After a few days, the main energy input to the SN ejecta comes from
radioactive decay. For SN 1987A, at first, the important isotopes are 56 N i
(τ1/2 = 6.1d) followed by 56 Co (τ1/2 = 78.8d). Beyond ∼ 1100 days, 57 Co is
more important and at ≥ 2000d, the dominant role is played by 44 T i. Using
bolometric and broad-band UBVRI light curves [143] estimated the masses
of the three most important radio-isotopes in SN 1987A to be 56 N i(0.069 ±
0.003M), 57 N i(0.003M) and 44 T i(1 ± (0.5 − 2) × 10−4 M ).
Among the radioactive isotopes accessible to gamma-ray astronomy, 44 T i
is a key isotope for the investigation of the inner regions of core collapse
SNe and their young SNRs. It has a half-life of 58.9 ± 0.3 yr [144] and a
decay scheme: 44 T i →44 Sc →44 Ca. The discovery [145] of the 1157 keV
gamma-ray line emission26 from Cas A with the Compton Gamma Ray Ob-
servatory was the first direct proof of synthesis of this short lived, freshly
26
J π = 2+ → 0+ transition in 44
Ca reached from 44
Sc following electron capture.
266 A. Ray

made radio-isotope in SNe. Renaud et al. [146] using the INTEGRAL space-
craft recently reported the detection of both 67.9 and 78.4 keV gamma-ray
lines of 44 Sc in Cas A. There was a clear separation of the two lines due to
an improved detection of the hard X-ray continuum up to 100 keV. The line
flux of (2.5 ± 0.3) × 10−5 cm−2 s−1 leads to a tightly constrained 44 T i mass of
+0.6
(1.6−0.3 ) × 10−4 M ). This is actually high compared to the predictions of the
standard models [135], [147], or their improved versions [148], [149]. Since
the production of 44 T i is sensitive to the explosion energy and asymmetries
and Cas A is known to be asymmetric and energetic (2 × 1051 erg instead
of the standard 1 × 1051 erg), this could be a factor in its apparent overpro-
duction compared to models. At the same time, the recent revision of the
40
Ca(α, γ)44 T i reaction rate [150] has led to an increase of 44 T i production
by a factor of ∼ 2.
Live radioactive isotopes freshly synthesized in the explosion were detected
from SN 1987A by directly detecting 1238 keV and 847 keV gamma-ray lines
[151], which provided detailed diagnostics of the nuclear burning conditions
and the explosion dynamics. Another radio-nuclide 57 Co, which is a decay
product of 57 N i made in the supernova explosion, was also detected directly
by measuring gamma-ray lines [152]. The calculated gamma-ray light curves
for 847 keV and 1238 keV and other lines [153] could be made consistent with
the Solar Maximum Mission (SMM) measurement only if the 56 Co was mixed
up to a mass coordinate of Mr ∼ 13.5M. The calculated fluxes were still
slightly smaller than the observed fluxes at t ∼ 200d, but this discrepancy
could be removed by taking account of the effect of clumpiness in the ejecta
which is more significant in the earlier phases. Also, the observed flux ratio
between the two gamma-ray lines at 847 keV and 1238 keV is close to unity
at early stages because of the smaller cross section for the 1238 keV lines
than for 847 keV. As the column depth of the overlying matter decreased
due to dilution with time in an expanding envelope, the observed flux ratio
approached the expected value of 0.68 based on branching ratios [153].
X-ray fluorescence spectrometry technique has been used by archaeologists
to determine the major and trace elements in ceramics, glass etc. Here, a sam-
ple is irradiated with X-rays and the wavelengths of the released (fluorescent)
X-rays are measured. Since different elements have characteristic wavelengths,
they can be identified and their concentrations estimated from the intensity of
the released X-rays. Trace elements analysis may, for example, help in iden-
tifying the (geographical) source of a material. A similar method has been
proposed [154] and attempted for SN 1987A [155]. It utilizes electron capture
decays of freshly synthesized radio-nuclei. Following the electron capture, the
K-shell vacancies are in most cases filled in by downward transitions from
other bound shells. As the fluorescence X-ray yields are well known, mea-
suring the X-ray line fluxes can then estimate the number of freshly synthe-
sized nuclei. For example, attempts were made to detect the 5.9 keV Kα line
from the stable nucleus 55 M n in the Chandra X-ray spectrum of SN 1987A,
which is due to the decay of radioactive 55 F e. Both 55 F e(τ1/2 = 2.7 yr) and
 Thermonuclear fusion and collapse and explosion in massive stars 267

55
Co(τ1/2 = 18 hr) are produced in explosive, incomplete Si burning as well
as in normal freezeout of nuclear statistical equilibrium, in the inner ejecta of
core collapse supernovae. However, no evidence of the 5.9 keV line emission
from Mn could be found in 400 ks of Chandra ACIS data and the upper limit
to the mean flux was < 3 × 10−7 cm−2 s−1 . Rauscher et al. [148] calculated the
ejected mass of A = 55 radioactive nuclei to be 7.7 × 10−4 M for 20M mod-
els of which most was 55 Co. If only about half this mass of 55 F e were ejected,
the reduced flux would be consistent with the observed upper limit. On the
other hand, even if the total mass inside were as much as 1 × 10−3 M , but
the 55 F e abundance was zero outside the radial velocity shells at 1500 kms−1 ,
the line flux would be still consistent with data, as at late times the emerging
flux depends sensitively on the presence of 55 F e in the outer zones.

10.3 Other X-ray supernovae

Apart from SN 1987A, which occurred close by and was therefore easy to
detect, there are supernovae with relatively high intrinsic X-ray luminosity.
SN 1993J and SN 1995N are two of them and they are at the high end of
the X-ray luminosity [156], [157], which makes it suitable for study in the
X-ray wave bands even at late stages [159]. Nymark et al. [158] have shown
that SNe with strong X-ray emission are likely to have radiative shocks and
that in these shocks a large range of temperatures contribute to the spectrum.
The cooling of the shock also affects the hydrodynamics of the flow structure
and the volume of the emitting gas and thus the total luminosity from the
interaction region. A shock is radiative depending on whether the cooling
time scale tcool of the shock is short compared to its expansion time t. Since
the cooling function (from electron bremsstrahlung or free-free scattering and
X-ray line emission from bound-bound transitions of ionized atomic species)
is a function of the shock temperature this condition translates to a condition
on the shock temperature for radiative vs. adiabatic shock. The boundary of
the two regimes depends upon the shock velocity Vs and the ejecta density
profile (given by the index η in ρ ∝ v −η ) and the separation in turn depends
upon the chemical composition of the ejecta (see Fig. 20 reproduced from
[160]). For a given shock velocity (and a composition and mass loss parameter
of the progenitor stellar wind and the ejecta velocity scale), the shock is likely
to be radiative for steeper density gradients.
As already mentioned, the spectroscopic type of SN 1993J from a type II-
like SN at early times due to detected Balmer lines soon weakened and it came
to resemble a type Ib SN, leading to a reclassification as a type IIb SN [161].
This transition is best explained if the progenitor star prior to its explosion,
had lost its hydrogen envelope due to interaction with a binary companion
[120], [121], [162], [163], [122]. If the H-envelope was indeed already thin,
then the reverse shock would transit the H-rich region quickly, and soon the
emission may be dominated by material dredged up from the nuclear processed
parts of the star in the interior. In the 4H47 model of Shigeyama et al. [164], by
268 A. Ray

Fig. 20. The border between radiative and adiabatic regions of the shocks in
supernovae as a function of shock velocity and ejecta density profile index η (ρ ∝
v −η ). The boundaries depend upon the composition of the ejecta through which the
shock may be propagating as also the mass loss parameter of the progenitor stellar
wind and the ejecta velocity scale (Fig. courtesy T. Nymark [160])

2600 days after explosion, the hydrogen-rich envelope had ∼ 0.47M. Inside
this envelope is the outer part of the He-zone which is enriched in N (about
∼ 0.3M ), while its inner part is C-rich (has ∼ 1.5M ). If the circumstellar
medium or the ejecta is clumpy, they may be hit by the reverse shock obliquely
and the shock in the clump will be slower and the shock temperature lower.
The presence of both hard X-rays and optical emission from the shocked CSM
shows the importance of a clumpy CSM, for example in SN 1987A where a
range of shock speeds are necessary to explain the observed spectrum [142].
SN 1993J is not as bright as SN 1987A or Cas A due to its larger distance,
and both Chandra and XMM-Newton data is comparatively of a moderate
signal to noise, at the low (CCD) resolution. The XMM spectrum of SN 1993J
is dominated by the Fe L emission at 0.7 - 1.0 keV and is blended with strong
Ne IX-X emission. Above 1.0 keV, emission lines which may be present are
from Mg XI-XII, Si XIII-XIV, as well as S XVI. Nymark et al. find that
the spectrum of SN 1993J is fit best by a combination of an adiabatic shock
(kTrev = 2.1 keV) propagating through a zone with CNO burning products,
with He dominant and N and Ne being the most abundant metals, and a
second radiative shock (kTrev = 1.0 keV). The latter could have been caused
by instabilities at the reverse shock front or by a clumpy CSM. The XMM
spectrum of SN 1993J and the model fits used [160] are shown in Fig 21.
SN 1995N is a type IIn supernova, a type that shows unusual optical char-
acteristics and spans a very broad range of photometric properties such as
 Thermonuclear fusion and collapse and explosion in massive stars 269

Fig. 21. XMM spectrum of SN 1993J (shown by data-points with crosses from
April 2001). The data is fitted with an adiabatic shock at kTrev = 2.1 keV (long
dashed in blue) and a radiative shock at kTrev = 1.0 keV (short dashed in red) for
a He/N dominated composition from model s15s7b [163]. (Fig. courtesy T. Nymark
[160])

decline rates at late times [165]. It is likely that these differences are related
to their progenitor’s structure, mass, composition as well as the composition
and the density profile of the CSM [166]. These supernovae show the pres-
ence of strong, narrow Balmer line emission on top of the broader emission
lines in their early spectra. The narrow emission lines may originate in the
dense and ionized circumstellar (CS) gas [167, 168]. The presence of strong
Hα emission line, the high bolometric luminosity and the broad Hα emission
base powered by the interaction of the supernova shock with the CSM, all
point towards a very dense circumstellar environment [169]. When the stellar
core collapses and explodes, the supernova lights up the slow-moving gas into
narrow emission lines leading to the type IIn supernova classification (n for
narrow emission line). This interaction of the supernova shock with the dense
CS gas is indicated by strong radio and X-ray emission detected from several
type IIn supernovae and in particular SN 1995N.
Chandra ACIS observations of SN 1995N show a best-fit line energy at
1.02 keV which is ascribed to Ne X [170]. There is also the possibility of
another line with best-fit line energy at 0.85 keV with identification as Ne IX.
The ionization potentials of all ionized Neon species upto Ne VIII are less
than 240 eV, whereas those of Ne IX and Ne X are above 1195 eV. Hence at
temperatures found by the Chandra observations, the predominant species of
Neon are expected to be Ne X and Ne XI. The mass of Neon was estimated
270 A. Ray

to be about 5 × 10−3 M − 1 × 10−2 M . This is consistent with Neon being

in the Helium layer for a 15M star where it was co-synthesized with C, N,
and He. For other stellar masses and other composition zones, the required
Neon mass is much larger than observed. Therefore, Neon in the Helium core
of the 15M star is the probable site where it was co-synthesized with C, N.

Conclusions
Much insight about how stars burn nuclear fuel, evolve and ultimately explode
has been obtained by calculations and computer simulations. Theoretical de-
velopments have been augmented by crucial astronomical observations over
decades in all bands of electromagnetic radiation and through other signal
channels like neutrinos. Most important have been the wider availability of
powerful telescopes both from the ground and space platforms and state of
the art computers. Crucial inputs to the field of nuclear astrophysics are also
coming from laboratory experiments involving radioactive ion beams (RIB)
and intense beams of energetic nucleons and nuclei. Short lived nuclei can
only be studied close to their sites of formation in the laboratory before they
decay. Such facilities will further advance the future of nuclear astrophysics.

Acknowledgments
I thank Aruna Goswami and the organizers of the School for the invitation
to Kodaikanal Observatory. Discussions with Katherina Lodders and Bruce
Fegley at the School and Claus Rolfs and Richard McCray in other meetings
are thankfully acknowledged. I thank Poonam Chandra and Firoza Sutaria
for comments on this manuscript and for long term collaboration. Nuclear
astrophysics research at Tata Institute is a part of the Plan Project: 11P-409.

References
1. W. D. Arnett, Supernovae and Nucleosynthesis, Princeton University
Press (1996)
2. M. N. Saha: Phil. Mag. 40, 472 (1920)
3. M. N. Saha: Proc. Roy. Soc. Ser. A 99, 135 (1921)
4. H. Reeves, Stellar Evolution and Nucleosynthesis, Gordon and Breach,
New York, (1968)
5. B. J. Carr, J. Bond, W. D. Arnett, ApJ 277, 445 (1984)
6. B. J. Carr, ARAA 419, 904 (1994)
7. W. Haxton, Nuclear Astrophysics Course (1999)
http://ewiserver.npl.washington.edu/phys554/phys554.html
8. Oak Ridge National Lab, Nuclear Data for Nuclear Astrophysics
http://www.phy.ornl.gov/astrophysics/data
 Thermonuclear fusion and collapse and explosion in massive stars 271

9. Lawrence Berkeley National Lab, Nuclear Astrophysics Reaction Rates

http://ie.lbl.gov/astro/astrorate.html
10. Joint Institute for Nuclear Astrophysics, REACLIB library
http://www.nscl.msu.edu/nero/db
11. Computational Infrastructure for Nuclear Astrophysics
http://www.nucastrodata.org
12. M. S. Smith et al., in CP1016, Origin of Matter and Evolution of Galaxies,
ed. by T. Suda et al., (AIP) p.466 (2008)
13. C. E. Rolfs, W. S. Rodney, Cauldrons in the Cosmos, University of
Chicago Press (1988)
14. D. D. Clayton, Principles of Stellar Evolution and Nucleosynthesis, Uni-
versity of Chicago Press (1984)
15. J. N. Bahcall: Neutrino Astrophysics, Chapter 3, Cambridge University
Press (1989); see also: http://www.sns.ias.edu/∼jnb/.
16. R. N. Boyd An Introduction to Nuclear Astrophysics, University of
Chicago Press (2008)
17. A. Ray, in 5th SERC School on Radioactive Ion Beams and Physics of Nu-
clei Away from the Line of Stability, Chandigarh, ed. by I. M. Govil, R. K.
Puri (Elite publishers, New Delhi), p. 99 (2003); arXiv:astro-ph/0405568.
18. A. V. Filippenko, in Supernovae and Stellar Evolution, proceedings of the
school and workshop in Goa, ed. by A. Ray, T. Velusamy, World Scientific,
Singapore, p 34 (1991)
19. R. P. Harkness, J. C., Wheeler, in Supernovae, ed. by A. Petschek,
Springer, Berlin, p 1 (1990)
20. A. Goswami, N. Prantzos, A&A 359, 191 (2000)
21. A. S. Eddington, Observatory 43, 341 (1920)
22. R. d’E. Atkinson, F. G. Houtermans, Z. Phys. 54, 656 (1929)
23. R. Davis, D. S. Harmer, K. C. Hoffman, Phys. Rev. Lett. 20, 1205 (1968)
24. E. Rutherford: Nature 123, 313 (1929)
25. T. Kirsten, in The origin of the solar system, ed. by S. F. Dermott, New
York: Wiley, p 267 (1978)
26. W. A. Fowler, in Proc. Welch Found. Conf. on Chemical Research, ed.
by W.D. Milligan, Houston Univ. Press, p 61 (1977)
27. H. A. Bethe, Rev. Mod. Phys. 9, 69 (1937)
28. E. E. Salpeter, Australian J. Phys. 7, 373 (1954)
29. R. Schiavilla, Phys. Rev. C 58, 1263 (1998)
30. T. Itahashi et al., Nucl. Phys. A 718, 466c (2003)
31. LUNA Collaboration, C. Casella, et al., Nucl. Phys. A 706, 203 (2002)
32. C. F. von Weizsäcker, Phys. Z. 38, 176 (1937)
33. C. F. von Weizsäcker, Phys. Z. 39, 633 (1938)
34. H. A. Bethe, C. L. Critchfield, Phys. Rev. 54, 248 (1938)
35. N. Chriestlieb et al., Nature 419, 904 (2002)
36. J. N. Bahcall, M. C. Gonzalez-Garcia, C. Pefia-Garay, Phys. Rev. Lett.
90, 131301 (2003)
37. C. Pena-Garay, A. Serenelli, arXiv: 0811.2424 (2008)
38. W. Haxton, A. Serenelli, ApJ 687, 678 (2008)
39. G. K. Schenter, P. Vogel, Nucl. Sci. Engg. 83, 393 (1983)
40. E. Fermi, Nuclear Physics, Course Notes, University of Chicago Press, p
83 (1951)
272 A. Ray

41. H. A. Bethe, C. L. Critchfield, Phys. Rev. 54, 862 (1938)

42. E. Frieman, L. Motz, Phys. Rev 89, 648 (1951)
43. E. E. Salpeter, Phys. Rev. 88, 547 (1952); also ApJ 115, 326 (1952)
44. J. N. Bahcall, R. M. May, ApJ 155, 501 (1969)
45. H. A. Bethe, R. F. Bacher, Rev. Mod. Phys. 8, 82 (1936)
46. J. N. Bahcall, C. P. Moeller, ApJ 155, 511 (1969)
47. J. N. Bahcall, L. S. Brow, A. Gruzinov, R. F. Sawyer, A&A 383, 291
(2002)
48. P. Quarati, A. M. Scarfone, J. Phys. G. 36, 025203 (2009)
49. B. W. Filippone, et al., Phys. Rev. C 28, 2222 (1983)
50. M. Gai, Phys. Rev. C 74, 025810 (2006)
51. W. T. Winter et al., Phys. Rev. C 73, 025503 (2006)
52. L. van Wormer et al., ApJ 432, 326 (1994)
53. A. E. Champagne, M. Wiescher, Ann Rev. Nucl. Part. Sci. 42, 39 (1992)
54. R. K. Wallace, S. E. Woosley, ApJS 45, 389 (1981)
55. M. Wiescher et al., Prog in part & Nucl Phys. 59, 51 (2007)
56. C. Sneden, J. J. Cowan, R. Gallino, ARAA 48, 241 (2008)
57. H. A. Bethe, Phys. Rev. 55, 103 and 434 (1939)
58. E. J. Öpik: Proc. Roy. Irish Acad. A 54, 49 (1951)
59. E. E. Salpeter, Phys. Rev. 107, 516 (1957)
60. F. Hoyle, ApJS 1, 121 (1954)
61. F. Hoyle, D. N. F. Dunbar, W. A. Wenzel, W. Whaling, Phys. Rev., 92,
1095 (1953)
62. C. W. Cook, W. A. Fowler, C. C. Lauritsen, T. Lauritsen, Phys. Rev.
107, 508 (1957)
63. J. A. Nolen, S. M. Austin, Phys. Rev. C 13, 1773 (1976)
64. C. Angulo, Exp Tools for Nuclear Astrophysics, Lecture Notes in Physics
764, 253 (2009)
65. L. Buchmann, C. A. Barnes: Nucl. Phys. A 777, 254 (2006)
66. A. Lefebvre-Schuhl, in Frontiers in Nuclear Structure, Astrophysics and
Reactions: FINUSTAR 2, ed. by P. Demetriou et al., (AIP), p 150 (2008)
67. S. E. Woosley, T. A. Weaver, ARAA 24, 205 (1986)
68. C. Hayashi, R. Hoshi, D. Sugimoto, Prog. Theor. Phys. Suppl. 22, (1951)
69. M. Mazarkis, W. Stephens, ApJ 171, L97 (1972)
70. H. Reeves, E. E. Salpeter, Phys. Rev. 116, 1505 (1959)
71. W. R. Hix, F. K. Thielemann, ApJ 460, 869 (1996)
72. F. K. Thielemann, W. D. Arnett, ApJ 295, 264 (1985)
73. G. M. Fuller, W. A. Fowler, M. J. Newman, ApJ 252, 715 (1982)
74. K. Kar, A. Ray, S. Sarkar, ApJ 434, 662 (1994)
75. J. -U. Nabi, M. Sajjad, Phys. Rev. C 77, 055802 (2008)
76. K. Langanke, G. Martinez-Pinedo, nucl-th/0203071, Rev. Mod. Phys. 75,
819 (2003)
77. W. R. Hix et al., ApJ 667, 476 (2007)
78. H. Y. Chiu, Ann of Phys 15, 1 (1961)
79. E. M. Burbidge, G. R. Burbidge, W. A. Fowler, F. Hoyle, Rev. Mod.
Phys. 29, 547 (1957)
80. G. Gamow, M. Schoenberg, Phys. Rev. 59, 539 (1941)
81. S. E. Woosley, H. -T. Janka, Nature Physics 1, 147 (2005)
82. H. A. Bethe, J. Wilson, ApJ 295, 14 (1985)
 Thermonuclear fusion and collapse and explosion in massive stars 273

83. H. A. Bethe, G. E. Brown, J. Applegate, J. Lattimer, Nucl. Phys. A 324,

487 (1979)
84. W. D. Arnett, ApJ 218, 815 (1977)
85. S. W. Bruenn, ApJS 58, 771 (1985)
86. P. S. Joshi, Scientific American 300, 36 (2009)
87. G. E. Brown, H. A. Bethe, G. Baym, Nucl. Phys. A 375, 481 (1982)
(BBB)
88. G. M. Fuller, ApJ 252, 741 (1982)
89. A. Ray, S. M. Chitre, K. Kar, ApJ 285, 766 (1984)
90. H. A. Bethe, G. E. Brown, J. Applegate, J. M. Lattimer, Nucl. Phys. A
324, 487 (1979)
91. K. Kar, A. Ray, Phys Lett A 96, 322 (1983)
92. A. Zaringhalam, Nucl. Phys. A 404, 599 (1983)
93. F. K. Sutaria, A. Ray: Phy. Rev. C 52, 3460 (1995)
94. F. K. Sutaria, A. Ray, J. A. Sheikh, P. Ring, A&A 349, 135 (1999)
95. O. Civitarese, A. Ray, Physica Scripta 59, 352 (1999)
96. H.-T. Janka, K. Langanke, A. Marek, G. Martnez-Pinedo, B. Mller, Phys.
Rept. 442, 38 (2007)
97. A. F. Fantina, P. Donati, P. M. Pizzochero, Phys. Lett. B 676, 140 (2009)
98. D. J. Dean, K. Langanke, J. M. Sampaio, Phys Rev. C 66, 45802 (2002)
99. F. K. Sutaria, A. Ray, Phys. Rev. Lett. 79, 1599 (1997)
100. Sudbury Neutrino Observatory Proposal, SNO-87-12, October (1987)
101. J. N. Bahcall, K. Kubodera, S. Nozawa, Phys. Rev. D 38 1030 (1988)
102. A. S. Burrows, in Supernovae, ed. by A. G. Petschek, Springer-Verlag
(1990)
103. L. M. Segal, Nucl. Phys. B 70, 61 (1974)
104. T. Weaver, S. E. Woosley, G. M. Fuller, Nuclear Astrophysics: Essays in
honor of J. Wilson, ed. by J. Centrella et al., 374 (1985)
105. R. M. Bionta et al., Phys. Rev. Lett., 58, 1494 (1987) (IMB collaboration)
106. K. Hirata, et al., Phys. Rev. Lett. 58, 1490 (1987) (K II collaboration)
107. W. D. Arnett, J. N. Bahcall, R. P. Kirshner, S. E. Woosley, ARAA 27,
629 (1989)
108. D. Casper, UNO A Next Generation Detector for Nucleon Decay and
Neutrino Physics
meco.ps.uci.edu/lepton workshop/talks/casper/uno.pdf (2000).
109. K. J. Chang, hep-ex/0005046 (2000)
110. R. A. Chevalier, C. Fransson, in Supernovae and Gamma-Ray Bursters,
ed. by Kurt Weiler, Lecture Notes in Physics, 598, 171 (2003)
111. S. Immler, W. H. G. Lewin, in Supernovae and Gamma-Ray Bursters,
ed. by Kurt Weiler., Lecture Notes in Physics, vol. 598, p. 91 (2003)
112. R. A. Fesen et al., ApJ 645, 283 (2006)
113. J. E. Reed, J. J. Hester, A. C. Fabian, P. F. Winkler, ApJ 440, 706 (1995)
114. M. Ryle, F. G. Smith, Nature 162, 462 (1948)
115. S. P. Reynolds, ARAA 48, 89 (2008)
116. O. Krause et al., Science 320, 1195 (2008)
117. S. E. Woosley, N. Langer, T. A. Weaver, ApJ 411, 823 (1993)
118. G. Aldering, R. M. Humphreys, M. Richmond, AJ 107, 662 (1994)
119. N. Rathnasree, A. Ray, JAA 13, 3 (1992)
120. A. Ray, K. P. Singh, F. K. Sutaria, JAA 14, 53 (1993)
274 A. Ray

121. K. Nomoto et al., Nature 364, 507 (1993)

122. J. R. Maund, S. J. Smartt, R. P. Kudritzki, P. Podsiadlowski, G. F.
Gilmore, Nature 427, 129 (2004)
123. T. Matheson et al., AJ 120, 1487 (2000)
124. R. A. Chevalier, J. Oishi, ApJ 593, L23 (2003)
125. C. Fransson, P. Lundqvist, R. A. Chevalier, ApJ 461, 993 (1996)
126. P. Chandra, A. Ray, S. Bhatnagar, ApJ 612, 974 (2004)
127. R. A. Fesen, K. S. Gunderson, ApJ 470, 967 (1996)
128. U. Hwang et al., ApJ 615, L117 (2004)
129. U. Hwang, S. S. Holt, R. Petre, ApJ 537, L119 (2000)
130. R. Willingale et al., A&A 381, 1039 (2002)
131. J. S. Lazendic, D. Dewey, N. S. Schulz, C. R. Canizares, ApJ 651, 250
(2006)
132. J. A. Ennis et al., ApJ 652, 376 (2006)
133. J. D. T. Smith et al., ApJ 693, 713 (2009)
134. R. A. Fesen et al., AJ 122, 2644 (2001)
135. S. E. Woosley, T. A. Weaver, ApJS 101, 181 (1995)
136. J. P. Hughes et al., ApJ 528, L113 (2000)
137. L. Rudnick, in The 10th International Symposium on Origin of Matter
and Evolution of Galaxies, Sapporo, Japan, ed. by Takuma Suda et al.,
AIP Conference Proceedings, Volume 1016, p. 353 (2008)
138. D. A. Liedahl, in X-ray Spectroscopy in Astrophysics, X EADN School
Amsterdam, ed. by J. van Paradijs, J. A. M. Bleeker, Springer, Berlin,
189 (1999)
139. D. Porquet et al., A&A 376, 1113 (2001)
140. D. Dewey, S. A. Zhekov, R. McCray, R. Canizares, ApJ 676, L131 (2008)
141. S. Park et al., in Supernova 1987A: 20 Years After, ed. by S. Immler, K.
W. Weiler, R. McCray (New York: AIP) 43 (2007)
142. S. A. Zhekov et al., ApJ 645, 293 (2006)
143. C. Fransson, C. Kozma, New Astron Rev. 46, 487 (2002)
144. I. Ahmad et al., Phys. Rev. C 74, 065803 (2006)
145. A. F. Iyudin et al., A&A 284, L1 (1994)
146. M. Renaud et al., ApJ 647, L41 (2006)
147. F. K. Thielemann, K. Nomoto, M. Hashimoto, ApJ 460, 408 (1996)
148. T. Rauscher et al., ApJ 576, 323 (2002)
149. M. Limongi, A. Chieffi, ApJ 592, 404 (2003)
150. H. Nassar et al., Phys. Rev. Lett. 96, 041102 (2006)
151. S. M. Matz et al., Nature 331, 416 (1988)
152. J. D. Kurfess et al., ApJ 399, L137 (1992)
153. K. Nomoto et al., in Supernovae and Stellar Evolution, proceedings of the
school and workshop in Goa, ed. by A. Ray, T. Velusamy, World Scientific,
Singapore p. 116 (1991)
154. M. D. Leising, New Astr. Reviews 46, 529 (2002)
155. M. D. Leising, ApJ 651, 1019 (2006)
156. H. U. Zimmermann, B. Aschenbach, A&A 406, 969 (2003)
157. D. W. Fox et al., MNRAS 319, 1154 (2000)
158. T. Nymark, C. Fransson, C. Kozma, A&A 449, 171 (2006)
159. P. Chandra et al., ApJ 699, 388 (2009)
160. T. Nymark, P. Chandra, C. Fransson, A&A 494, 179 (2009)
 Thermonuclear fusion and collapse and explosion in massive stars 275

161. A. V. Filippenko, T. Matheson, L. C. Ho, ApJ 415, L103 (1993)

162. Ph. Podsiadlowski et al., Nature 364, 509 (1993)
163. S. E. Woosley et al., ApJ 429, 300 (1994)
164. T. Shigeyama et al., ApJ 420, 341 (1994)
165. A. V. Filippenko, ARAA 35, 309 (1997)
166. W. Li, A. V. Filippenko, S. D. Van Dyk, J. Hu, PASP 114, 403 (2002)
167. R. B. C. Henry, D. Branch, PASP 99, 112 (1987)
168. A. V. Filippenko, in Supernova 1987A and other supernova, ESO Con-
ference and Workshop, ed. by J. Danziger and Kurt Kjaer, p 343 (1991)
169. N. N. Chugai, I. J. Danziger, Astron. Lett. 29, 649 (2003)
170. P. Chandra et al., ApJ 629, 933 (2005)
The Evolution of Massive Stars
and the Concomitant Non-explosive
and Explosive Nucleosynthesis

Marcel Arnould

Institut d’Astronomie et d’Astrophysique, Université Libre de Bruxelles, CP-226,

B-1050 Brussels, Belgium

Summary. These lectures are concerned with some aspects of the evolution of mas-
sive stars and of the concomitant nucleosynthesis. They complement other lectures
in this volume. Special emphasis is put on the production of the nuclides heavier
than iron by the r- and p-processes.

Keywords: Stars: massive – Stars: evolution – Stars: nucleosynthesis – Solar

System: composition – r-process – p-process

1 Introduction

Massive stars classically refer to those stars that are able to experience all
possible nuclear burning stages from hydrogen to the so-called silicon burn-
ing. This definition limits the relevant mass range to M > ∼ 10M , with some
uncertainty on this boundary value.
These stars have been of central concern in a myriad of observational and
theoretical works. No wonder! They indeed play a key role in many chapters of
astrophysics. In particular, they influence the physical and chemical states of
their circumstellar environments or of the interstellar medium through their
intense radiation and mass losses during their non-explosive phases of evolu-
tion, and even more so, as a result of their final supernova explosions. They
may act as triggers of star formation, are essential agents of the evolution of
the nuclidic content of the galaxies, accelerate particles to cosmic ray energies,
and leave neutron stars or black holes at the end of their evolution. They are
also the progenitors of certain γ-ray bursts.
Chapter 5 in this volume discusses some aspects of the physics and as-
trophysics of massive stars. We concentrate here on the production of the
nuclides heavier than iron by the r- and p-processes (the third mode of pro-
duction of the heavy nuclides, that is the s-process, is discussed by Karakas in
this volume). As an introduction to these topics, we briefly sketch in Sect. 2

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 6, c Springer-Verlag Berlin Heidelberg 2010
278 Marcel Arnould

the evolution of massive stars before dwelling into some details of their non-
explosive nucleosynthesis (Sect. 3). The explosion of massive stars that leads
to the supernova phenomenon is the subject of Sect. 4, and the concomitant
synthesis of the nuclides up to the iron peak is summarised in Sect. 5. The
following sections are devoted to the r- and p-processes. The separation be-
tween the s-, r- and p-process contributions to the bulk Solar System compo-
sition is discussed in Sect. 6.2, with special emphasis on the largely neglected
uncertainties that affect this splitting procedure. Deviations to the bulk com-
position in the form of isotopic anomalies are briefly mentioned in Sect. 6.3,
while Sect. 6.4 deals with the evolution of the r-nuclide content of the Galaxy.
The topics of the actinide abundances in the Solar System, in the local inter-
stellar medium or in stars, and of the r-nuclide content of the galactic cosmic
rays are covered in Sects. 6.6 and 6.7. Site-free parametric approaches of the
r-process are reviewed in Sect. 7. Section 7.1 is concerned with the so-called
canonical and multi-event models, and dynamical high-temperature r-process
approaches are presented in Sect. 7.2. Section 7.3 deals instead with a high-
density scenario. Attempts to relate the r-process to specific astrophysical
situations follow. Sections 8 and 9 review the possibility of development of
the r-process in massive star explosions and as a result of the coalescence of
two neutron stars. The review of the r-process is concluded in Sect. 10 with
some comments on the difficulties to model the evolution of the r-nuclide
content of the Galaxy and on nucleo-cosmochronology. The review of the p-
process starts with some generalities (Sect. 11). Sections 11.1 and 11.2 discuss
the possibility of production of the p-nuclides in Type II and in Type Ia su-
pernovae, while the development of the p-process in sub-Chandrasekhar white
dwarf explosions is examined in Sect. 11.3. Section 11.4 is devoted to a brief
account of the isotopic anomalies attributed to the p-process, and to attempts
to build a chronometry based on this process. Some general conclusions are
drawn in Sect. 12.

2 Some generalities about the evolution of massive stars

The structure and evolution of the stars of various masses and initial compo-
sitions are discussed in some detail in chapters 3 and 4 in this volume. We
limit ourselves here to some general considerations regarding the evolution of
massive stars (M > ∼ 10M ) only, with special emphasis on the accompanying
nucleosynthesis. In short, and as pictured very schematically in Fig. 1, the evo-
lution of the central regions of a massive star is made of successive ‘controlled’
thermonuclear burning stages and of phases of gravitational contraction. The
latter phases are responsible for a temperature increase, while the former ones
produce nuclear energy through charged-particle induced reactions. Of course,
composition changes also result from these very same reactions, as well as, at
some stages at least, from neutron-induced reactions, which in contrast do not
play any significant role in the stellar energy budget. The nuclear reactions
 Massive Stars Evolution and Nucleosynthesis 279

involved in the different non-explosive nuclear burning stages are briefly dis-
cussed in Sect. 3. Let us simply say here that such a sequence develops in
time with nuclear fuels of increasing charge number Z and at temperatures
increasing from several tens of 106 K to about 4 × 109 K. Concomitantly
the duration of the successive nuclear-burning phases decreases in a dramatic
way. This situation results from the combination of (i) a decreasing energy
production when going from H burning to the later burning stages, and (ii) an
increasing neutrino production, and the consequent energy losses, with tem-
peratures exceeding about 5 × 108 K. Figure 1 also depicts schematically that
a nuclear burning phase, once completed in the central regions, migrates into
a thin peripheral shell. As a consequence the deep stellar regions look like an
onion with various ‘skins’ of different compositions.
It is quite important to stress that the true stellar structure is certainly
much more complicated than sketched in Fig. 1, even when effects like devi-
ations from spherical symmetry (induced by rotation or certain mechanisms
of transport of matter) are neglected. This spherically symmetric picture of a
star may break down, especially during the advanced stages of the evolution
of massive stars, and would lead to a dramatic growing of the complication
of the stellar structure and evolution (e.g. [2, 3], and references therein). This
increased complexity is demonstrated by multi-dimensional simulations of the
structure of massive stars. The consideration of rotation of course brings ad-
ditional difficulties. Steady mass loss from a star may also affect its evolution
in various ways. Finally, binarity may lead to specific evolutionary patterns
resulting for the largest part from episodic mass transfers from one component
to the other.
After having experienced all the burning phases terminating with Si burn-
ing (Fig. 1), massive stars are seen to develop an Fe core that is lacking
further nuclear fuels, as any transformation of the strongly-bound Fe nuclei
is endothermic. In fact this core becomes dynamically unstable and implodes
as a result of free-electron captures and Fe photodisintegration, the former
transformation playing an especially important role at the low end of the rel-
evant stellar-mass range. Through a very complex chain of physical events
the implosion can, in certain cases at least, turn into an explosion referred to
as a supernova (SN). It is accompanied with the ejection into the interstel-
lar medium (ISM) of a substantial amount of material with kinetic energies
typically of the order of 1051 ergs. Outliers in the form of more energetic
(hypernovae) or less energetic (faint) SNe have been identified, however. The
innermost parts of the star are predicted to escape explosion. They remain
bound in a ‘residue’ which may be a neutron star (observable as a pulsar if it
is magnetized and rapidly rotating), or even a black hole. Some more details
on the modelling of SN explosions can be found in Sect. 4. A more thorough
review of the SN phenomenon can be found in [4].
Let us just note here that, observationally, a supernova is classified as
being of Type II (SN II) if it shows H-lines in its spectrum. It is likely that
most, if not all, of the exploding massive stars still have some H-envelope
280 Marcel Arnould

H burning
8
He
Mass inside star / Mo.

He burning
6
CO
C burning
H burning

SN remnant
O Ne Mg
4 He burning

C burning
Ne burning
O Mg Si
O burning

Ne burn.
Si S Si burning
2

O burn.

Si burn.
“Fe”
SN residue
0
Evolution
Supernova
explosion
Tc(K) 4x107 2x108 8x108 1.6x109 2.1x109 4x109
rc(g / cm )
3 4 760 105 3.9x106 3.6x106 3x107
Time– 7 x106 8 x105 500 0.9 0.4 1
scale y y y y y d
Fig. 1. Schematic representation of the evolution of the internal structure of a
spherically-symmetric massive (M ≈ 25M ) star. The shaded zones correspond to
nuclear burning stages. A given burning phase starts in the central regions and then
migrates into thin peripheral burning shells. Typical central temperatures Tc , central
densities ρc and durations of the core burning phases are indicated at the bottom
of the figure (adapted from [1]). In between the central nuclear burning phases
are episodes of gravitational contraction (downward arrows). The chemical symbols
represent the most abundant nuclear species left after each nuclear-burning mode
(“Fe” symbolises the iron-peak nuclei with mass numbers A in the 50 < ∼A< ∼ 60
range). If the star eventually explodes as a supernova, the most central parts may
leave a ‘residue,’ while the rest of the stellar material is ejected into the ISM, where
it is observed as a supernova ‘remnant’

left, and thus exhibit such a feature. In contrast, Type I SNe (SNI) lack H in
their ejecta. Specific spectral features have led to the identification of different
SNI subclasses, among which are the SNIa and SNIb/c supernovae. The latter
concern the fate of massive stars suffering pre-explosion mass losses that are
strong enough for ejecting their H-rich envelopes. A situation of this type is
typically encountered in stars that are massive enough to evolve through the
Wolf-Rayet phase. In contrast, SNIa are classically viewed as the explosion of
a white dwarf following its accretion of matter from a companion star in a
binary system.
 Massive Stars Evolution and Nucleosynthesis 281

3 Non-explosive stellar evolution and concomitant

nucleosynthesis
In the following we present a very brief account of the nucleosynthesis devel-
oping during the various non-explosive stages in the evolution of massive stars
depicted in Fig. 1.

3.1 Hydrogen burning

In massive stars, hydrogen burns essentially through the cold CNO cycles rep-
resented in Fig. 2 which develop at temperatures in excess of some 107 K.1
They produce fresh 4He, while the initial C, N and O are mostly transformed
into 14N. Some H is also consumed by the NeNa and MgAl chains (Fig 3),
which play only a minor role in the stellar energy budget, but are of signifi-
cance in the production of the Na to Al isotopes. Most important, the MgAl
chain might synthesize 26Al, which is a very interesting radio-active nuclide for
γ-ray astronomy and cosmochemistry. The p-p chains of H burning play only
a very minor energetic or nucleosynthesis role in massive stars, in contrast to
the situation encountered in low-mass stars (see e.g. [5]).

Fig. 2. Reactions involved in the cold CNO mode of hydrogen burning

1
A burning mode is referred to as ‘cold’ if an unstable nucleus, once produced,
has time to β-decay before suffering a particle capture. The reverse situation
characterises ‘hot’ modes instead.
282 Marcel Arnould

(p,γ) (p,γ) (p,γ)

22 25 26 27
21
Ne Na Mg Al Si

(β+) (β+) (β+) (β+)

21
Na NeNa 22
Ne
25
Al MgAl 26
Mg (β+)

(p,γ) (p,γ) (p,γ) (p,γ)

21 23Na 24 27 28
Ne Mg Al Si
(p,α) (p,γ) (p,α) (p,γ)
Fig. 3. Reactions involved in the cold NeNa and MgAl modes of hydrogen burning

Much experimental and theoretical effort has been devoted to the reactions
involved in the H-burning modes, as summarized in NACRE [6], which also
provides typical uncertainties still affecting the relevant reactions (see also
[7]). The yields from the cold CNO, NeNa and MgAl burning modes based
on the NACRE adopted rates and their uncertainties have been analyzed by
[8] based on the NACRE data and on schematic stellar conditions. These
predictions, as well as others based on specific stellar models (see e.g. [9, 10]
for stars with masses M < 10M) demonstrate that better determinations
of certain reaction rates would be desirable in order to set up meaningful
comparisons between certain abundance predictions and observations.

3.2 Helium burning and the s-process

The reactions involved in the He-burning stage have been discussed in many
places (e.g. [5], NACRE [6]). The main ones are displayed in Fig 4. They
develop at temperatures in excess of 108 K, and mainly transform 4He into
12
C and 16O, with some limited contribution to the abundance of some heav-
ier α-particle nuclei (esp. 20Ne,....), at least in massive enough stars. Of very
special and dramatic importance for the theories of stellar evolution and of
nucleosynthesis is the famed 12C (α , γ) 16O reaction, which has been the sub-
ject of a flurry of experimental investigations, as well as of theoretical efforts
(for a review, see [11]; see also [12] for a recent re-analysis of the case). In
spite of that, uncertainties remain, and preclude certain nuclear astrophysics
predictions to be made at a satisfactory level (see e.g. [13]).
 Massive Stars Evolution and Nucleosynthesis 283

STABLE NUCLEUS 25 Mg

UNSTABLE NUCLEUS 22 Ne

18 F

16 18 O

14 N

12 13 C

(1)
8 Be
(2)

(3)
4 He

Fig. 4. Main reactions involved in the burning of helium. Arrows of types (1), (2)
and (3) refer to (γ, α), (α, γ) and (α.n) reactions

Other α-particle induced reactions are of special importance, and have

been the subject of many dedicated experimental and theoretical works. This
is particularly the case for 13C (α , n) 16O and 22Ne (α , n) 25Mg . These reac-
tions are considered as the main sources of neutrons for the s-process nucle-
osynthesis, the latter one being of special significance in massive stars.
The proper treatment of the s-process also requires the knowledge of a
host of neutron-capture cross sections at typical energies from about 10 to
100 keV on targets in the whole 12 ≤ A ≤ 210 mass range. Much dedicated
experimental work has led to a substantial improvement in our knowledge of
relevant (n,γ), as well as (n,p) and (n,α) cross sections. Some data are still
lacking, especially for neutron-captures by unstable targets close to the valley
of nuclear stability that can be involved in the s-process. In these cases, the
astrophysical rates of the competing β-decays have also to be known.

3.3 Carbon burning

Carbon burning develops at temperatures in excess of about 5 × 108 K, and

produces mainly nuclides in the 20 <∼A< ∼ 27 mass range through reactions
the most important of which are displayed in Fig. 5. The C-burning phase
raises the very interesting question of the fusion of light heavy ions below
the Coulomb barrier, and in particular of the origin of the very pronounced
structures observed in the 12C + 12C fusion cross section at low energies. The
284 Marcel Arnould

devoted experimental (e.g. [14], and references therein) and theoretical works
do not provide fully satisfactory solutions.

a
Panel
12C(12C,α)20Ne
12C(12C,
Ρ)23Ne

12C( 13 + 13 16 12C(n,g )13C(a ,n)16O

P,g ) N(b ) C(a , n) O
25Mg(n,g )26Mg( ,g )27Al
P
16
O(a ,g )20Ne
20Ne(n,g )21Ne

b Panel
16
O(a ,g )20Ne
12
C(12C,α)20Ne
(P,α)20Ne
12
C(12C,P)23Na (P,g )24Mg

(α,P)26Mg
22
Ne(α,n)25Mg(P,g )26Al(b + )26Mg(P,g )27Al
22
Ne(P,g )23Na
20
Ne(n,g )21Ne(α,n)24Mg
20
Ne(α,g )24Mg
12
C(n,g )13C(α,n)16O
Fig. 5. Most important reactions involved in the carbon burning core of massive
stars. Panels (a) and (b) concern the first and later stages of the burning in a
Population I 25 M star (from [15])
 Massive Stars Evolution and Nucleosynthesis 285

3.4 Neon, oxygen, and silicon burning

As shown in Fig. 6, the Ne-burning phase is initiated at temperatures some-

what in excess of 109 K by 20Ne (γ , α) 16O , the first major energetically sig-
nificant photodisintegration reaction experienced by a star in the course of
its evolution. A complementary 20Ne destruction channel is 20Ne (α , γ) 24Mg .
Both α-capture reactions have been studied experimentally and theoretically
[6]. The main ashes of Ne burning are 16O. A variety of other nuclides up to
28
Si are also synthesised.

20Ne(g, a)16O

20Ne(a, g) (a, g )
24Mg 28Si(a, g )32S
21Ne(a, n) (n, g )25Mg(a, n)

(α, n)29Si
(α, P )26
Mg (α, p)30Si
23
Na (P, g )27Al
(P, α) Ne20
(n, g )28Al(b –)28Si

22
Ne(α, n )25Mg

20Ne(n, g )21Ne

28
Si(n, g )29Si(n, g )30Si
25
Mg(n, g )26Mg
Fig. 6. Most important reactions involved in the central Ne burning of massive
stars (here a Population I 25 M star (from [15])

The main reactions involved in the central burning of oxygen are dis-
played in Fig. 7. They develop at temperatures in excess of about 2 × 109 K
286 Marcel Arnould

and produce mainly 28Si, as well as a variety of nuclides in the approximate

30 <∼ A  40 mass range. It is important to emphasize that free electron cap-
tures also start playing the important role of changing
 the neutron excess
during the course of the burning. It is defined as i (Ni − Zi )Yi , where Ni (Zi )
is the number of (bound plus free) neutrons (protons), and Yi is the mole
fraction of nuclide i (Yi = Xi /Ai with Xi the mass fraction of i and Ai its
atomic weight). This change has an important impact on the composition of
the central region of a massive star at its presupernova stage (see below). Also
note that, on the nuclear side, the 16O+16O fusion reaction that governs the
O-burning phase does not exhibit intricacies comparable to those encountered
with 12C+12C.
Silicon burning comprises a very complex pattern of nuclear reactions that
has been analysed in particular detail by [16]. It develops at temperatures in
excess of about 4 × 109 K, and is mainly made of a nuclear flow starting at
28
Si and leading to the production of lighter species through a series of photo-
disintegrations. The nucleons and α-particles released by these reactions can
be captured by the remaining 28Si, which generates a nuclear flow moving
upward. The reactions involved in the downward and upward moving flows
progressively lead to a situation where limited ensembles of nuclei reach a
situation of ‘Nuclear Statistical Equilibrium (NSE)’. These ensembles are re-
ferred to as ‘Quasi-Statistical Equilibrium (QSE)’ clusters. In this situation,
the relative abundances of nuclei belonging to the same QSE cluster are ob-
tained from the equations of equilibrium statistical mechanics (the so-called
nuclear Saha equations; see below). In such a regime, it is not required to
solve a full nuclear reaction network. In contrast, the relative abundances of
species belonging to different QSE clusters can be obtained only from the
consideration of a reaction network. Figure 8 displays a snapshot of the Si
burning pattern of nuclear flows at T = 3.5 × 109 K in a Population I 25 M
star. Two QSE clusters are seen to have developed, corresponding to two en-
sembles of nuclei connected by a large number of reactions (which translates
in the figure by grey patches). These QSE clusters are interconnected by a
smaller number of reactions that are too slow for allowing these clusters to
be in equilibrium at the selected time. Note that the boundaries of the QSE
clusters and abundances are time-dependent, as a result of changes in tem-
perature and in neutron excess, the latter being due to the operation of free
electron captures. The general trend is that the two QSE clusters displayed
in Fig. 8 are more fragmented at earlier epochs. At later times, a QSE cluster
involving lighter nuclei develops, and all the clusters finally merge to lead to a
Nuclear Statistical Equilibrium (NSE) state.2 This regime has been described
in detail by e.g. [17], and is not reviewed here. Let us simply say that the

2
Note that weak interaction processes remain out of equilibrium as long as neu-
trinos are not equilibrated with matter and radiation. This is the case as long as
the density remains lower than about 1011 g/cm3 . At higher densities, a state of
so-called complete equilibrium is obtained.
 Massive Stars Evolution and Nucleosynthesis 287

(p, g )39K(p, g )40Ca

(g , p)35Cl(g , p)34S(a, g )38Ar

(a, g )36Ar (a, p )39K↑ (a, g )42Ca

(n, g )37Ar(n, a)34S↑

(p, a)28Si(a, g )32S (a, p )35Cl↑

(16O, p)31P (p, g )32S↑ (n, g )33S(n, a)30Si(a, g )34S↑

(g, p)30Si↑

(g , n)30Si↑

16
O (16O, a)28Si↑

(g, p)30P( g , p )20Si

(16O, n)31S
(b +)31P↑ (a, n)32S↑

Fig. 7. The main reactions involved in the O burning in the core of massive stars
(here a Population I 25 M star). The reverse reactions of the underlined ones may
be activated at some point during the burning (from [15])
288 Marcel Arnould

30

20
z

10

0
0 5 10 15 20 25 30 35 40
N
Fig. 8. Illustration of the development of two QSE clusters during Si burning at
3.5 × 109 K in a Population I 25 M star (from [15])

abundances are obtained simply by solving a set of nuclear Saha equations

under the constraint of mass conservation and electrical neutrality. The re-
sulting abundances do not depend on reaction rates, but only on temperature,
nuclear binding energies, nuclear partition functions and neutron excess. As
in the case of the QSE regime, the NSE abundances are subjected to time
variations as a result of temperature and neutron excess changes. In any case,
the NSE conditions favour the production of the most stable, that is the iron
peak, nuclides. The iron core represented in Fig. 1 results.
In the framework of stellar evolution models, Si burning has often been
approximated in different ways relying on the QSE cluster concept. These
approximations are not fully appropriate, particularly in view of the time
dependence of the QSE cluster boundaries. This has been stressed by [15] who
adopt instead for the QSE and NSE regimes a detailed Si-burning network
coupled to the stellar evolution equations.
 Massive Stars Evolution and Nucleosynthesis 289

4 The explosive fate of massive stars

As said in Sect. 2, the iron core left over following Si burning suffers a dynam-
ical instability as a result of endothermic electron captures and Fe photodid-
integration. To a first approximation, this gravitational instability sets in near
the classical Chandrasekhar mass limit for cold white dwarfs, MCh = 5.83 Ye2 ,
Ye being the electron mole fraction. In the real situation of a hot stellar core,
collapse may start at masses that differ somewhat from this value, depending
on the details of the core equation of state. The reader is referred to [18]
(especially Chaps. 12 and 13) for a detailed discussion of the implosion mech-
anism and for its theoretical outcome and observable consequences. Here, we
just briefly summarise the situation.
The gravitational collapse of the iron core does not stop before the central
densities exceed the nuclear matter density ρ0 ≈ 2.5 × 1014 g cm−3 by about
a factor of two. At this point, the innermost (M < ∼ 0.5 M ) material forms
an incompressible, hot and still lepton-rich ‘proto-neutron’ star (PNS) whose
collapse is stopped abruptly, and which eventually bounces. A shock wave
powered by the gravitational binding energy released in the collapse propa-
gates supersonically into the infalling outer layers. For many years, it has been
hoped that this shock could be sufficiently strong for ejecting explosively most
of the material outside the core, producing a so-called ‘prompt core collapse
supernova’ (PCCSN) with a typical kinetic energy of 1–2 × 1051 ergs, as ob-
served. The problem is that the shock is formed roughly half-way inside the
iron core, and looses a substantial fraction of its energy in the endothermic
photodisintegrations of the iron-group nuclei located in the outermost portion
of the core. The shock energy loss is aggravated further by the escape of the
neutrinos produced by electron captures on the abundant free protons in the
shock-heated material. Detailed one-dimensional hydrodynamic simulations
conclude that the initially outgoing shock wave transforms within a few mil-
liseconds after bounce into an accretion shock. The matter behind the shock
continues to accrete on the PNS. No recent simulation is able to predict a
successful PCCSN for a Fe-core progenitor star (M > ∼ 10 M ). This failure
is illustrated in Fig. 9 for a 15 M star.
Even so, some hope to get a CCSN of a non-prompt type has been ex-
pressed if there is a way to ‘rejuvenate’ the shock efficiently enough to obtain
an explosive ejection of the material outside the PNS. This rejuvenation re-
mains a matter of intensive research. Neutrinos might well play a pivotal
role in this matter. They are produced in profusion from the internal energy
reservoir of the PNS that cools and deleptonises hundreds of milliseconds af-
ter bounce, and their total energy might amount to several 1053 ergs, that is
about 100 times the typical explosion energy of a SN II. The deposition of
a few percent of this energy would thus be sufficient to unbind the stellar
mantle and envelope, and provoke a ‘delayed’ CCSN (DCCSN) (these quali-
tative statements assume that a black hole is not formed instead of a PNS;
see below). Many attempts to evaluate the precise level of neutrino energy
290 Marcel Arnould

104

103
radius [km]

102

101

100
0.0 0.2 0.4 0.6 0.8 1.0
time [s]
Fig. 9. Radial trajectories of several mass elements of the core of a 15 M star
versus time after bounce. The trajectories are plotted for each 0.02 M up to 1 M ,
and for each 0.01 M outside this mass. The thick dashed line indicates the location
of the shock wave. The prompt shock stalls within 100 ms after reaching 150 km,
and recedes down to below 100 km. No sign of a revival of the shock that possibly
leads to a successful D(elayed-)CCSN is seen either, even after 300 ms. Instead,
a stationary accretion shock forms at several tens of km. A PNS is seen to form,
reaching 1.6 M around 1 s after bounce (from [19])

deposition have been conducted over the last decades, based on more or less
controversial simplifications of the treatment of the neutrino transport (e.g.
[20] for a recent re-analysis of the problem, which is made even more com-
plex by the due consideration of neutrino flavor mixing). In fact, theoretical
investigations and numerical simulations performed with increasing sophisti-
cation over the past two decades have not been able to come up with a clearly
successful CCSN for a Fe-core progenitor (M > ∼ 10 M ). This conclusion is
apparently robust to changes in the highly complex physical ingredients (like
 Massive Stars Evolution and Nucleosynthesis 291

the neutrino interactions, or the equation of state), and in the numerical tech-
niques (e.g. [20]). In fact, the neutrino-energy deposition should have to be
significantly enhanced over the current model values in order to trigger an
explosion. An illustration of a failed DCCSN is shown in Fig. 9.

Fig. 10. Simulation of an electron-capture supernova following the collapse of an

O-Ne core. The time evolution of the radius of various mass shells is displayed with
the inner boundaries of the O+Ne, C+O and He shells marked by thick lines. The
inner core of about 0.8 M is mainly made of Ne at the onset of collapse ( [21], and
references therein). The explosion is driven by the baryonic wind caused by neutrino
heating around the PNS. The thick solid, dashed, and dash-dotted lines mark the
neutrino spheres of νe , ν̄e , and heavy-lepton neutrinos, respectively. The thin dashed
line indicates the gain radius which separates the layers cooled from those heated by
the neutrino flow. The thick line starting at t = 0 is the outward moving supernova
shock (from [22])

This adverse circumstance may not mark the end of any hope to get a
DCCSN, however. In the case of the single stars considered here, one might
just have to limit the considerations to the stars in the approximate 9 to 10
M range that possibly develop O-Ne cores instead of iron cores at the termi-
nation of their hydrostatic evolution. Efficient endothermic electron captures
could trigger the collapse of that core, which could eventually transform into
a so-called electron-capture supernova that may be of the SN Ia or SN II
292 Marcel Arnould

type, depending upon the extent of the pre-explosion wind mass losses.3 In
these situations, illustrated in Fig. 10, the neutrino heating is indeed efficient
enough for rejuvenating the shock wave about 150 ms after bounce, and mass
shells start being ablated from the PNS surface about 50 ms later, leading
to a so-called ‘neutrino-driven wind’.4 No information is provided by the cur-
rent simulations on the conditions at times much later than a second after
bounce. Note that the predicted successful delayed electron-capture super-
nova is characterised by a low final explosion energy (of the order of 0.1 × 1051
ergs, which is roughly ten times lower than typical SN values), and by just a
small amount of ejected material (only about 0.015 M ). These features might
suggest a possible connection with some sub-luminous (faint) SN II events.
Let us stress, however, that the structure of the progenitors of the electron-
capture supernovae remains especially uncertain (e.g. [23]), which endangers
any conclusion one may draw on these SN types.
A major effort has been put recently in the development of simulations
of explosions that go beyond the one-dimensional approximation. This is
motivated not only by the difficulty of obtaining successful CCSNe in one-
dimensional simulations, but also by the mounting observational evidence that
SN explosions deviate from spherical symmetry, not to mention the possible
connection between the so-called soft long-duration gamma-ray bursts, and
grossly asymmetric explosions accompanied with narrow jets of relativistic
particles, referred to as JetSNe. The multi-dimensional extension of the sim-
ulations opens the potentiality to treat in a proper way different effects that
may turn out to be essential in the CCSN or JetSNe process. As briefly re-
viewed by e.g. [26], they include fluid instabilities, or rotation and magnetic
fields on top of the neutrino transport already built into the one-dimensional
models. Acoustic power may be another potential trigger of CCSNe [27] (see
also [24] for a brief review of multidimensional simulations).
In summary, there are obviously many crucial questions that remain to
be answered before one can hope putting together a clear and coherent pic-
ture of the CCSN fate of massive stars. The structure of the pre-supernova
stars remains uncertain in many important aspects which may have a signif-
icant impact on the properties, and even the very existence, of the explosive
fate of the massive stars. This concerns in particular the mass loss rates,

3
The range of initial masses of single stars which could experience an electron-
capture instability is still quite uncertain, and depends in particular on a subtle
competition between the growth of the stellar cores resulting from thermal pulses
developing during the Asymptotic Giant Branch evolution and their erosion re-
sulting from steady mass losses. Other stars in the approximate 8 to 12 M mass
range might end up as O-Ne white dwarfs or experience of Fe core collapse instead
of experiencing an electron-capture supernova resulting from the collapse of the
O-Ne core ([23] for a review). Binary systems might offer additional opportunities
of obtaining electron-capture supernovae (see [24] or [25])
4
Unless otherwise stated, neutrino-driven winds refer to transonic as well as sub-
sonic winds, the latter being referred to as a breeze regime (e.g. [24])
 Massive Stars Evolution and Nucleosynthesis 293

angular-momentum distributions, couplings to magnetic fields, chemical mix-

ing, not to mention multi-dimensional effects. The simulations of CCSNe
and of JetSNe face crucial problems of micro- and macro-physics nature.
Aborted model explosions are currently commonplace. PCCSNe appear to
be excluded, and so are 1D DCCSNe. Multi-dimensional simulations leave
some hope through the interplay between fluid instabilities, acoustic waves,
rotation, magnetic fields and neutrinos. Mild or weak explosions of stars de-
veloping O-Ne cores or of accreting and rotating O-Ne white dwarfs have
been obtained thus far, sometimes at the expense of an artificial enhancement
of the neutrino luminosity. Detailed three-dimensional simulations are most
needed in order to clarify the role of various mechanisms listed above, and
their precise couplings.

5 Nucleosynthesis associated with CCSN events

0.5
Log10 Production Factor

– 0.5

–1
Mg

Mn

Mo
Ga
Ge
Ne
Na

Ca

Co
Cu

Rb

Nb
Se
Sc

Fe

Zn

As
Cr
Ar

Br
Kr

Sr
Cl

Ni

Zr
Al
Si

Ti
O
C
N

P
S

Y
F

Fig. 11. Production factors for the elements between C and Mo following the SN
explosion of stars with metallicity Z = 0.02 and with different masses (13 M :
triangles; 15 M : squares; 20 M : open circles; 25 M : filled circles; 30 M : pen-
tagons; 35 M : asterisks in open circles). In all cases, the production factors have
been normalised to an oxygen production factor of unity. The lines refer to produc-
tion factors obtained by integrating over a Salpeter IMF (dn/dM ∝ M −2.35 ). The
solid line refers to an assumed relation between the mass of the SN residue, which
is equivalent to the 56Ni yields, and the progenitor mass (0.15, 0.10, 0.08, 0.07, 0.05
and 0.05 M for the 13, 15, 20, 25, 30 and 35 M stars). The dotted line is obtained
by assuming that the 56Ni yield of each SN is 0.05 M (from [28])

Detailed nucleosynthesis predictions are in general available for 1D explo-

sion models only, under the assumption that a shock wave propagates outward
through most of the supernova structure, compresses the various traversed
294 Marcel Arnould

layers, heats them up before pushing them successfully outward until their
ejection into the ISM. This expansion is of course accompanied by the cooling
of the material. This heating and cooling process of the layers hit by the su-
pernova shock wave allows some nuclear transformations to take place during
a quite brief time, modifying more or less significantly the pre-explosion com-
position of the concerned layers. The study of the composition of the ejected
material that makes up the supernova remnant is one of the main chapters of
the theory of ‘explosive nucleosynthesis.’
These lectures are mainly concerned with the explosive production of the
r- and p-nuclides heavier than iron. The explosive yields of the lighter nuclides
have been the subject of many calculations, and are not reviewed in any detail
here. Let us just emphasize that the absence of fully self-consistent CCSN
explosions imposes a parametric approach of the associated nucleosynthesis.
It involves in particular the choice of a suitable final kinetic energy of the
ejecta compatible with the observations (of the order of 1051 ergs), or the
selection of the ejected amount of 56Ni that can influence the SN light curve.
This amount relates to the mass of the residue left from the explosion, as 56Ni
is the result of the explosive burning of Si in the deepest stellar layers. The
mass of the residue is indeed a free parameter in absence of self-consistent
SN simulations. The propagation of the outward moving shock wave and the
properties of the shocked layers are computed in various ways (e.g. [28]).
Roughly speaking, the most abundant species are the result of the pre-
explosion nucleosynthesis, while less abundant ones can be more or less
substantially produced explosively. Figure 11 provides an example of the com-
position of the ejecta from various stars with metallicity Z = 0.02. The result
of an integration of the yields from individual stars over an Initial Mass Func-
tion (IMF) is also displayed. It appears that most of the resulting yields are
compatible (within a factor of about 2) with the oxygen one. Deviations may
be due to the neglect of the contribution from stars with masses lower than
13 M stars, from SNIa explosions, or from stars with metallicities different
from Z = 0.02. It has also to be stressed that many uncertainties of different
natures remain. At the level of individual stars, they concern in particular
the pre-SN and SN evolutionary stages (including multi-dimensional effects),
or the rate of certain key reactions, like 12C (α , γ) 16O . Many more large un-
certainties remain at the level of the evolution of the nuclidic content of the
galaxies predicted by so-called ‘chemical evolution models’.

6 The synthesis of the nuclides heavier than iron:

generalities
Since the early days of the development of the theory of nucleosynthesis (e.g.
[29]), it has proved operationally rewarding to introduce three categories of
heavy nuclides referred to as s-, p-, and r-nuclides. This splitting is not a
 Massive Stars Evolution and Nucleosynthesis 295

mere game. It corresponds instead to the ‘topology’ of the chart of the nu-
clides, which exhibits three categories of stable heavy nuclides: those located
at the bottom of the valley of nuclear stability, called the s-nuclides, and those
situated on the neutron-deficient or neutron-rich side of the valley, named the
p- or r-nuclides, respectively. Three different mechanisms are called for to
account for the production of these three types of stable nuclides. They are
naturally referred to as the s-, r-, and p-processes.
Over the years, the development of models for the production of these three
types of nuclides has largely relied on the bulk Solar System abundances, and
on the decomposition of these abundances into the contributions from the
s-, r- and p-processes. These data have been complemented with a myriad of
spectroscopic observations from which heavy element abundances have been
derived in a large sample of stars in different galactic locations and with
different metallicities.

6.1 The bulk Solar System composition

Since the fifties, much effort has been devoted to the derivation of a meaningful
set of elemental abundances representative of the composition of the bulk
material from which the Solar System (referred to in the following as SoS)
formed some 4.6 Gyr ago.
Early in the development of the theory of nucleosynthesis, it has been
recognised that this bulk material is made of a well-mixed blend of many
nucleosynthesis contributions to the SoS composition over the approximate
10 Gyr that have elapsed between the formations of the Galaxy and of the
SoS. The latest detailed analysis of the SoS is due to [30]. As in previous
compilations, the selected abundances are largely based on the analysis of
a special class of rare meteorites, the CI1 carbonaceous chondrites, which
are considered as the least-altered samples of available primitive SoS matter.
Solar abundances derived from spectroscopic data now come in quite good
agreement with the CI1 data for a large variety of elements. Some noticeable
exceptions exist, however (e.g. [31]).
The bulk SoS isotopic composition shows a high level of homogeneity. This
is why it is mostly based on the terrestrial data. For H and the noble gases, as
well as for Sr, Nd, Hf, Os, and Pb, some adjustments are required [30]. There
are exceptions, however, to this high bulk isotopic homogeneity. One is due to
the decay of relatively short-lived radionuclides that existed in the early SoS,
and decayed in early formed solids in the solar nebula. Also interplanetary dust
particles contain isotopic signatures apparently caused by chemical processes.
Additional isotopic ‘anomalies’ are observed in some meteoritic inclusions or
grains (see Sect. 6.3).
As it is well known, the SoS nuclidic abundance distribution exhibits a high
‘iron peak’ centered around 56Fe followed by a broad peak in the A ≈ 80 − 90
mass region, whereas double peaks show up at A = 130 ∼ 138 and 195 ∼ 208.
These peaks are superimposed on a curve decreasing rapidly with increasing
296 Marcel Arnould

mass number. It has been realized very early that these peaks demonstrate the
existence of a tight correlation between SoS abundances and nuclear neutron
shell closures.

102
101 s
r
Abundances [Si = 106 ]

s
100 r
10– 1
10– 2
10– 3
p
10– 4
10– 5
10– 6
80 100 120 140 160 180 200
A
Fig. 12. Decomposition of the solar abundances of heavy nuclides into s-process
(solid line), r-process (black dots) and p-process (open squares) contributions. The
uncertainties on the abundances of some p-nuclides due to a possible s-process con-
tamination are represented by vertical bars (from [32]). See Figs. 13 - 15 for the
uncertainties in the s- and r-nuclide data

6.2 The s-, r- and p-nuclides in the Solar System

As mentioned above, it is very useful to split the abundance distribution of the

nuclides heavier than iron into three separate distributions giving the image
of the SoS content of the p-, s- and r-nuclides. A rough representation of this
splitting is displayed in Fig. 12. In its details, the procedure of decomposition is
not as obvious as it might be thought from the very definition of the different
types of nuclides, and is to some extent dependent on the models for the
synthesis of the heavy nuclides. These models predict in particular that the
stable nuclides located on the neutron-rich (deficient) side of the valley of
nuclear stability are produced, to a first good approximation, only by the
r-(p-)process. These stable nuclides are naturally called ‘r-only’ and ‘p-only’
nuclides, and their abundances are deduced directly from the SoS abundances.
The situation is more intricate for the nuclides situated at the bottom of the
valley of nuclear stability. Some of them are produced solely by the s-process,
 Massive Stars Evolution and Nucleosynthesis 297

the typical flow of which is located very close to the valley (see the discussion
on s-process in this volume). They are referred to as ‘s-only’ nuclides, and
are encountered only when a stable r- or p-isobar exists, which ‘shields’ the
s-isobar from the r- and p-processes. As a result, only even-Z heavy elements
possess an s-only isotope. In general, a phenomenological model of the s-
process is used to fit at best the abundances of all the s-only nuclides. Such a
model is described in e.g. [24]. Once the parameters of this model have been
selected in such a way, it is used to predict the s-process contributions to the
other s-nuclides. The subtraction of these s-process contributions from the
observed SoS abundances leaves for each isotope a residual abundance that
represents the contribution to it of the r-process (if neutron-rich) or p-process
(if neutron-deficient). These nuclides of mixed origins are called ‘sr’ or ‘sp’
nuclides.
Figure 12 shows that about half of the heavy nuclei in the SoS material
come from the s-process, and the other half from the r-process, whereas the
p-process is responsible for the production of about 0.01 to 0.001 of the abun-
dances of the s- and r-isobars, except in the Mo-Ru region. It also appears
that some elements have their abundances dominated by an s- or r-nuclide.
They are naturally referred to as s- or r-elements. Clearly, p-elements do not
exist. If these global abundance patterns remain valid in other locations than
the SoS, stellar spectroscopy can provide information on the s- or r- (but
not on p-) abundances outside the SoS. Even if the dominance of the s- or
r-processes on a given element remains true in all astrophysical locations, a
wealth of observations demonstrate departures from the SoS s- and r-element
mixtures. As mentioned above, such departures exist in the SoS itself in the
form of isotopic anomalies, or in stars with different ages, galactic locations,
or evolutionary stages. The SoS abundances and their s-, r- and p-process
contributions do not have any ‘universal’ character.
From the above short description of the splitting strategy between s-, r-
and p-nuclides, it is easily understood that uncertainties affect the relative s-
and r-(p-)process contributions to the SoS abundances of the sr(p)-nuclides.
Even so, they are quite systematically put under the rug. This question clearly
deserves a careful study, especially in view of the sometimes very detailed
and far-reaching considerations that have the s-r SoS splitting as an essential
starting point.
The splitting of the SoS s-, r- and p-nuclide abundances has been reviewed
in some detail by [32, 24]. In view of its importance, we repeat here most
aspects of the procedure.
As recalled above, the SoS r-nuclide abundance distribution is obtained by
subtracting from the observed SoS abundances those predicted to originate
from the s-process. These predictions are classically based on a parametric
model, referred to as the canonical exponential model initially developed by
[33], and which has received some refinements over the years (e.g. [34]). This
model assumes that stellar material composed only of iron nuclei is subjected
to neutron densities and temperatures that remain constant over the whole
298 Marcel Arnould

period of the neutron irradiation. In addition, the SoS s-abundance pattern is

viewed as originating from a superposition of two exponential distributions of
t
the time-integrated neutron exposure, τn = 0 Nn vT dt (where Nn is the neu-
tron number density, and vT is the most probable relative neutron-nucleus
velocity at temperature T ). These distributions are traditionally held respon-
sible for the so-called weak (70 <∼A< ∼ 90) and main (A > ∼ 90) components of
the s-process. A third exponential distribution is sometimes added in order to
account for the 204 < A ≤ 209 s-nuclides. Through an adequate fitting of the
parameters of the three τ -distributions, the superposition of the two or three
resulting abundance components reproduces quite successfully the abundance
distribution of the s-only nuclides in the SoS, from which it is concluded that
the s-contribution to the sr-nuclides can be predicted reliably. It has to be
stressed that this result is rooted only in the nuclear properties of the species
involved in the s-process, and does not rely at all on specific astrophysics
scenarios. Many s-process calculations have been performed in the framework
of models for stars of various masses and initial compositions (e.g. [24], and
other contributions to this volume). Some model calculations along the line
have been used to obtain the contributions of the s- and r-processes to the
SoS abundances [35, 36]. This procedure is currently not advisable. Large un-
certainties remain in the s-abundances predicted from individual model stars.
In addition, the SoS s-nuclide abundances result from a long evolution of the
galactic composition that cannot be mimicked reliably enough.
Despite the success of the canonical model in fitting the solar s-nuclide
distribution, some of its basic assumptions deserve questioning. This concerns
in particular a presumed exponential form for the distribution of the neutron
exposures τ , which has been introduced by [33] in view of their mathemat-
ical ease in abundance calculations. In addition, the canonical model makes
it difficult in the s-nuclide abundance predictions to evaluate uncertainties of
nuclear or observational nature. As a result, the concomitant uncertainties in
the solar r-abundances are traditionally not evaluated. The shortcomings of
the canonical model are cured to a large extent by the so-called multi-event
s-process model (MES) [37]. In view of the importance to evaluate the uncer-
tainties affecting the solar distribution of the abundances of the r-nuclides,
we review the MES in some detail. A similar multi-event model has also been
developed for the r-process (MER), and is presented in [38].
The MES relies on a superposition of a given number of canonical events,
each of them being defined by a neutron irradiation on the 56Fe seed nuclei
during a time tirr at a constant temperature T and a constant neutron density
Nn . In contrast to the canonical model, no hypothesis is made concerning any
particular distribution of the neutron exposures. Only a set of canonical events
that are considered as astrophysically plausible is selected a priori. We adopt
here about 500 s-process canonical events covering ranges of astrophysical
conditions that are identified as relevant by the canonical model, that is,
1.5 × 108 ≤ T ≤ 4 × 108 K, 7.5 ≤ log Nn [cm−3 ] ≤ 10, and 40 chosen tirr -values,
 Massive Stars Evolution and Nucleosynthesis 299

corresponding to evenly distributed values of ncap in the 5 ≤ ncap ≤ 150 range,

where

ncap = A NZ,A (t = tirr ) − A NZ,A (t = 0) (1)
Z,A Z,A

is the number of neutrons captured per seed nucleus (56Fe) on the timescale
tirr , the summation extending over all the nuclides involved in the s-process.
For each of the selected canonical events, the abundances NZ,A are obtained
by solving a reaction network including 640 nuclear species between Cr and
Po. Based on these calculated abundances, an iterative inversion procedure
described in [39] allows us to identify a combination of events from the con-
sidered set that provides the best fit to the solar abundances of a selected
ensemble of nuclides. This set includes 35 nuclides comprising the s-only nu-
clides, complemented with 86 Kr and 96 Zr (largely produced by the s-process
in the canonical model), 152 Gd and 164 Er (unable in the p-process and able in
the s-process to be produced in solar abundances [32]), and 208 Pb (possibly
produced by the strong s-process component in the canonical model).
On grounds of the solar abundances of [40], it has been demonstrated in
[37] that the derived MES distribution of neutron irradiation agrees qualita-
tively with the exponential distributions assumed in the canonical model, even
though some deviations are noticed with respect to the canonical weak and
strong components.5 The MES provides an excellent fit to the abundances
of the 35 nuclides included in the considered set of species, and in fact per-
forms to a quite-similar overall quality as that of the exponential canonical
model predictions of [40]. Even a better fit than in the canonical framework is
obtained for the s-only nuclides (see [37] for details). The MES model is there-
fore expected to provide a decomposition of the solar abundances into their
s- and r-components that is likely to be more reliable than the one derived
from the canonical approach for the absence of the fundamental assumption
of exponential distributions of neutron exposures.
Compared with the canonical approach, the MES model has the major
advantage of allowing a systematic study of the various uncertainties affect-
ing the abundances derived from the parametric s-process model, and conse-
quently the residual r-nuclide abundances. The uncertainties in these residuals
have been evaluated in detail by [37] from due consideration of the uncertain-
ties in (i) the observed SoS abundances as given by [40] (see footnote1 ), (ii)
the experimental and theoretical radiative neutron-capture rates involved in
the s-process network, and in (iii) the relevant β-decay and electron-capture
rates. Total uncertainties resulting from a combination of (i) to (iii) have fi-
nally been evaluated. The results of such a study for the elements with Z ≥ 30
are displayed in Figs. 13 and 14. The corresponding SoS isotopic r-residuals

5
A MES calculation with the revised solar abundances [30, 31] has not been done,
but is expected not to give significantly different results from those reported here.
300 Marcel Arnould

1
Sr Pb
Zr Ba
Ga Ce
0.8 Rb T1
Ge Nb Sn
W
Ns / Nsolar

0.6 Y La Bi
Hf Hg
Kr Cd
Nd
0.4 Yb
Pr
In Sb
Mo Tm
0.2 Cs Re
Ag Lu
Sm Os
Pt
Eu
0
30 40 50 60 70 80
Z
Fig. 13. MES predictions of the s-process contribution to the SoS abundances Nsolar
[40] of the elements with Z ≥ 30. Uncertainties are represented by vertical bars (from
the calculations of [37])

and their uncertainties are shown in Fig. 15. They are tabulated in [24]. Dif-
ferent situations can be identified concerning the uncertainties affecting the r-
residuals. Many sr-nuclides are predicted to have a small s-process component
only. The r-process contribution to these species, referred to as r-dominant,
is clearly quite insensitive to the s-process uncertainties. The situation is just
the opposite in the case of s-dominant nuclides.
Some r-process residuals are seen to suffer from remarkably large uncer-
tainties, which quite clearly cannot be ignored when discussing the r-process
and the virtues of one or another model for this process. This concerns in
particular the elements Rb, Sr, Y, Zr, Ba, La, Ce and Pb. Some of them,
and in particular Ba or La, are often used as tracers of the levels of s- or
r-processing during the galactic history (see Sect. 6.4). Lead has also a spe-
cial status in the studies of the s-process (e.g. [89] for references), as well as
of the r-process (see Sect. 6.6). It could well be of pure s-nature if a strong
s-process component can indeed develop in some stars, but a pure r-process
origin cannot be excluded. These uncertainties largely blur any picture one
 Massive Stars Evolution and Nucleosynthesis 301

4
Ge
Se Solar abundances
Kr
Solar r- abundances
log NZ / NH + 12 3 Sr
Zr
Xe
Te Ba Pb
Mo Pd Sn
2 As
Br Ru Cd Ce Pt
Ga Nd Ir
Rb Dy Yb Os Hg
I Gd Er
1 Ag Cs
Hf
W
Y Rh Sm Au
Sb
Nb In
Pr Eu Ho
0 La Tb Re T1
Tm
Lu
Ta
–1
30 40 50 60 70 80
Z
Fig. 14. SoS r-residuals and their uncertainties for the Z ≥ 30 elements based on
the s-abundances of Fig. 13. The abundances NZ and NH refer to element Z and
to H. The ordinate used here is about 1.55 dex larger than the [Si=106 ] unit used
elsewhere

might try to draw from spectroscopic observations and from simplistic theo-
retical considerations.

6.3 Isotopic anomalies in the solar composition

The bulk SoS composition has been of focal interest since the very beginning
of the development of the theory of nucleosynthesis. Further astrophysical
interest and excitement have developed with the discovery of the fact that
a minute fraction of the SoS material has an isotopic composition deviating
from that of the bulk. Such ‘isotopic anomalies’ are observed in quite a large
suite of elements ranging from C to Nd (including the rare gases), and are now
known to be carried by high-temperature inclusions of primitive meteorites,
as well as by various types of meteoritic grains. The inclusions are formed
from SoS material out of equilibrium with the rest of the solar nebula. The
grains are considered to be of circumstellar origin, and to have survived the
process of incorporation into the SoS.
Isotopic anomalies contradict the canonical model of an homogeneous
and gaseous protosolar nebula, and provide new clues to many astrophysical
302 Marcel Arnould

102
r-only
r-abundances [Si = 106] 1 r-dominant
10
s-dominant
100

10–1

10–2

10–3
60 80 100 120 140 160 180 200 220
A
Fig. 15. SoS isotopic r-residuals corresponding to the elemental abundances of
Fig. 14. Different symbols identify different relative levels of r-process contribution.
The s-dominant nuclides are defined here as those predicted by MES to have more
than 50% of their abundances produced by the s-process. The s-process contribution
varies between 10 and 50% in the case of the r-dominant species, and does not exceed
10% for the r-only nuclides

problems, like the physics and chemistry of interstellar dust grains, the forma-
tion and growth of grains in the vicinity of objects with active nucleosynthesis,
the circumstances under which stars (and in particular SoS-type structures)
can form, as well as the early history of the Sun (in the so-called ‘T-Tauri’
phase) and of the SoS solid bodies. Last but not least, they raise the question
of their nucleosynthesis origin and offer the exciting perspective of comple-
menting the spectroscopic data for chemically peculiar stars in the confronta-
tion between abundance observations and nucleosynthesis models for a very
limited number of stellar sources, even possibly a single one. This situation is
in marked contrast with the one encountered when trying to understand the
bulk SoS composition, which results from the mixture of a large variety of nu-
cleosynthesis events, and consequently requires the modelling of the chemical
evolution of the Galaxy.
Among the identified anomalies, several concern the p-, s- and r-nuclides.
Those attributed to the r- and p-processes are discussed in some detail by
[32, 24], and are not reviewed here. As a very brief summary, let us just state
 Massive Stars Evolution and Nucleosynthesis 303

that various blends of p, s-, and r-nuclides that differ more or less markedly
from the bulk SoS mixture depicted in Sect. 6.2 are identified in a variety of
meteorites at various scales, including bulk samples, refractory inclusions or
grains interpreted from their many highly anomalous isotopic signatures as
grains of circumstellar origins. This is generally interpreted in terms of the
decoupling between the three mechanisms producing these nuclides, and of a
non-uniform mixing of their products. One of the surprises of main relevance
to these lectures is that those grains that are generally interpreted in terms
of supernova condensates do not carry the unambiguous signature of the r-
process that would be expected if indeed supernovae are the privileged r-
process providers (see Sect. 8).

6.4 Evolution of the r-nuclide content of the Galaxy

In practice, the question of the evolution of the galactic content of the nu-
clides heavier than iron concerns the s- and r-nuclides only. It is traditionally
assumed indeed that the p-nuclides are just as rare in all galactic locations as
in the SoS. In such conditions, the p-nuclide abundances outside the SoS are
out of reach of spectroscopic studies. On the other hand, the s-process and
its contribution to the galactic abundances at different epochs are discussed
elsewhere in this volume, so that we focus here only on the evolution of the
r-nuclide galactic content.
A substantial observational work has been conducted in recent years on
this subject. This effort largely relies on the abundance evolution of Eu, clas-
sified as an r-process element on grounds of SoS abundance analyses. The
main conclusions derived from this observational work may be summarized as
follows (see [24] for details and references).

(i) The Eu data summarised in Fig. 16 are classically used to support the
idea that the r-process has contributed very early to the heavy element con-
tent of the Galaxy. However, the observed Eu abundance scatter introduces
some confusion when one tries to establish a clear trend of the Eu enrichment
with metallicity. It is also difficult to identify the value of [Fe/H] at which
the signature of the s-process becomes identifiable. This conclusion relies in
particular on La abundances derived from observation (see Fig. 17), La being
classically considered as an s-element in the SoS (even if a non-negligible r-
process contribution cannot be excluded). A most useful information on the
relative evolution of the s- and r-process efficiencies in the Galaxy would be
provided by the knowledge of the isotopic composition of the neutron-capture
elements (see Sect. 6.5).
(ii) Much excitement has been raised by the observation that the patterns
of abundances of heavy neutron-capture elements between Ba and Pb in r-
process-rich metal-poor stars are remarkably similar to the SoS one. This
claimed ‘convergence’ or ‘universality’ illustrated in Fig. 18 has to be taken
304 Marcel Arnould

Fig. 16. Ratio [Eu/Fe] versus [Fe/H] for a large stellar sample. The data are from
various sources, the details of which can be found in [42]

with some care, however, as it largely relies on the assumption that the decom-
position between s- and r-process contributions in metal-poor stars is identical
to the SoS one, which has yet to be demonstrated. An interpretation of this
universality is proposed in Sect. 7.1.
(iii) No universality appears to hold for Z < ∼ 58 and for Z > ∼ 76, where this
concerns in particular the Pb-peak elements and the actinides. This situation
has far-reaching consequences, particularly regarding the possibility of build-
ing galactic chronologies on the actinide content of very metal-poor stars (see
Sect. 10). No universality is observed either for the elements lighter than Ba
(like Sr, Y or Zr). This is illustrated in Fig. 19, which shows in particular
that Sr has a quite different behaviour than the heavier Eu, with no clear
identifiable trend of [Sr/Eu], in contrast to [La/Eu].
(iv) The different behaviours of the abundance patterns of the elements below
and above Ba have laid the ground for speculations on possible different sites
of the r-process. It is discussed by [24] that the interest of these speculations
is rather limited at the present stage of desperate search for a single suitable
site for the r-process (see Sects. 8 and 9).
 Massive Stars Evolution and Nucleosynthesis 305

Fig. 17. Values of log10 (La/Eu) = log10 (NLa /NH ) − log10 (NEu /NH ) versus [Fe/H]
from various studies (different symbols). A typical error is shown as a dagger. The
three labelled points correspond to metal-poor La-rich stars. The solid line is the
total SoS La/Eu ratio, and the broken lines indicate the SoS s- and r-process abun-
dance breakdowns (from [36], where the references to the observations are provided)

6.5 Can the available isotopic data tell something about the
prevalence of the s- or of the r-process at early galactic times?

The considerations above rely on elemental abundance data. A most useful

information regarding the relative evolution of the s- and r-process efficien-
cies in the Galaxy would be provided by the knowledge of the isotopic
composition of the neutron-capture elements. Such data are unfortunately
very scarce, and largely concern Ba only. The hyperfine splitting of its
spectral lines has been used to evaluate the fractional abundance fodd =
[N (135 Ba) + N (137 Ba)]/N (Ba) of the odd Ba isotopes. This ratio has indeed
been considered as a measure of the relative contributions of the r- and s-
processes to Ba. The SoS fodd is about 0.18. The r-process fraction to Ba
is given by r/(r + s) = [fodd − fodds r
]/[fodd − fodd
s s
]. Here, the fractions fodd
r
and fodd are of the odd-mass Ba isotopes in the cases of pure s- and pure
r-processes, respectively, which for the time being cannot be evaluated in any
other way than by analysing the SoS Ba isotopic compositions. Table 1 of [24]
∼ 0.23 with the ‘standard’ value being fodd ≈ 0.10. On the
s s
gives 0.06 <∼ fodd <
306 Marcel Arnould

0.50

– 0.50

–1.50
Relative log ε

–2.50

– 3.50

– 4.50
HD 11 5444
–5.50 CS 22892–052
SS r–Process Abundances
BD + 173248
– 6.50
50 60 70 80 90
Atomic Number

Fig. 18. Heavy-element abundance patterns for three heavy-element-rich metal-

poor stars, the [Eu/Fe] ratios of which are included in Fig. 16. The solid lines
represent a scaled SoS r-nuclide distribution. Inverted triangles indicate upper limits
(from [43])

r r
other hand, any value of fodd (0 <∼ fodd < ∼ 1) is permissible in consideration of
the largely uncorrelated uncertainties, whilst its standard value is 0.66. On top
s r
of the uncertainties in fodd and fodd , the derivation of fodd from observation
is also hampered with substantial difficulties, as illustrated by the diverging
conclusions concerning HD 140283 by [46] who concludes that Ba in this star
is of typical SoS s-r mix, and by [47] who instead claims that it is of pure
r-process origin. A solar mixture of Ba isotopes is also preferred by [48] for
a sample of metal-poor main sequence or close to main sequence cool stars.
Note that all these statements are based on roughly the same fodd values. The
Ba isotopic composition has been analysed recently in 25 cool thick and thin
disc dwarf stars with −1.35 ≤ [Fe/H] ≤ 0.25 [45]. The derived fodd values
are displayed in Fig. 20. They are seen to be smaller in the thin disc than
in the thick disc, whose age is estimated to be comparable to the one of the
s r
halo. Note that the fodd and fodd values (0.10 and 0.46) indicated in Fig. 20
are those derived from the predictions of certain AGB models, and thus may
not be reliable to be applied in the problem at hand because of the large in-
trinsic uncertainties of the models themselves. In particular, to combine [45]
 Massive Stars Evolution and Nucleosynthesis 307

1.0
0.5
0.0
[Sr / Eu]

–0.5
–1.0
–1.5
–2.0
–4.0 –3.5 –3.0 –2.5 –2.0 –1.5 –1.0
[Fe/ H]
Fig. 19. Values of [Sr/Eu] versus [Fe/H] for an ensemble of stars with different
metallicities. The average relative error is shown at the bottom left. The horizontal
line represents an estimate of the pure r-process SoS value (from [44])

the metallicity-dependent AGB s-process models along the chemical evolution

history of the Galaxy with the SoS isotopic composition is quite a dubious
practice because it is then effectively dictating the possible metallicity de-
s
pendence of the r-process. Nonetheless, with these or our ‘standard’ fodd and
r
fodd values taken for granted as a measure, the observed fodd trend seemingly
suggests that the r-process contribution to Ba decreased during the galactic
evolution. However, before deriving firm conclusions, the uncertainties in the
fodd values surely do not have to be swept under the rug.

6.6 Actinides in the Solar System, in the Local Interstellar

Medium, and in stars

Actinides have a very special status in the theory of nucleosynthesis, as they

are the only ones of clear and unique r-process origin. In addition, their ra-
dioactivity, and in some cases their lifetimes commensurable with the pre-
sumed age of the Galaxy, makes them potentially suited for chronological
considerations.
For long, the abundances of the actinides have been known only in the SoS
essentially through meteorite analyses [30]. The astrophysical importance of
Th and U has been enhanced further with the first observation of Th in
308 Marcel Arnould

fraction of the odd isotopes of Ba

50 r-process

40

30

20 solar

s-process
10

–1.5 –1.0 – 0.5 0.0

[Ba/H]

Fig. 20. Values of fodd (in %) versus [Ba/H] for a sample of thick disc (black
circles) and thin disc (open circles) stars with −1.35 ≤ [Fe/H] ≤ 0.25. The displayed
s r
s-process (fodd ) and r-process (fodd ) values (horizontal lines) are derived from an
AGB star model (from [45])

some stars with different metallicities (Fig. 21). The Th data have been used
by several authors in combination with the corresponding Eu abundances to
evaluate the ages of the concerned stars. In this respect, it would clearly be
desirable to rely instead on the Th/U abundance ratio in stars, as it is done in
the SoS case. Unfortunately, detection of U is difficult because of the weakness
of its spectral lines, combined with its present low abundance in the studied
stars. In spite of these difficulties, U has been successfully measured in the
[Fe/H] = -2.9 giant CS 31082-001 ( [50], and references therein). Only upper
limits have been obtained by [49] for the 7 stars in which they have derived
Th abundances. The reliability of the age determination of specific stars based
on the use of the Th/Eu and Th/U derived from observations is discussed in
Sect. 10.
It has to be noted that 232Th, 235U and 238U all decay to Pb. The Pb
abundance has been measured in CS 31082-001 [50]. From this observation,
it is concluded [50] that more than 50% of the total Pb in this star are the
actinide progeny. This does not provide any strong constraint on the fraction of
the Pb in CS 31082-001 that is a direct (instead of an actinides decay) product
of the r-process. Lead in very metal-poor stars can indeed originate from the
s-process as well [51]. It has also to be remarked that the SoS r-process Pb
is highly uncertain, the fractional contribution of this process derived from
 Massive Stars Evolution and Nucleosynthesis 309

– 0.2
log ε (Th / Eu)

– 0.4

– 0.6

– 0.8

–1
– 3.2 –3 –2.8 –2.6 –2.4 –2.2
[Fe / H]
Fig. 21. Values of Th/Eu (in the  scale defined in Fig. 17) versus [Fe/H] obtained
from different observations (from [49]). The black circles are from [49], whereas the
open circles are earlier observations with sometimes different [Fe/H] assignments.
The solid horizontal line corresponds to the SoS value

Table 1 of [24] lying between 1 and 80%! In such conditions, Pb data in low-
metallicity stars or in the SoS can hardly provide useful information on the
r-process.
Finally, let us recall the attempts to measure the 244Pu content in the
local ISM, which may have some interesting astrophysical implications. At
present, this can be done through the analysis of dust grains of identified
interstellar origin recovered in deep-sea sediments (e.g. [52]). In a near future,
the determination of elemental and isotopic composition of the ISM grains
will be a major goal of research with their recovery to Earth by the Stardust
mission [53].

6.7 The r-nuclide content of Galactic Cosmic Rays

The measured abundances in the Galactic Cosmic Rays (GCRs) of all the ele-
ments in the approximate 30 < ∼Z < ∼ 60 range of atomic numbers are roughly
consistent with a solar composition at the source of the GCRs (the very na-
ture of which remains quite mysterious) once corrections for atomic selection
effects have been duly taken into account. In particular, no clear trend is iden-
tified in this range of atomic numbers for a specific enhancement or deficiency
of either r- or s-nuclides. The elements with Z > ∼ 40 might be overabundant
relative to Fe. This conclusion is very sensitive, however, to the modelling of
the GCR propagation conditions (see e.g. [54]).
310 Marcel Arnould

The situation appears to be different in the Z > ∼ 60 range, where obser-

vations suggest that the abundances in the ‘Pt group’ (Pt, Ir and Os) are in
excess to those around Pb (the ‘Pb group’) relative to the SoS composition
(e.g. [55]), this conclusion being relatively insensitive to the propagation con-
ditions. Various interpretations of this relative excess in the Pt group have
been proposed. One of them [54] relates to the high volatility of Pb and of
other Pb-group elements, which contrasts with the refractory character of
the Pt-group elements. In these views, the observations could well be consis-
tent with a mixture of the SoS type. Another interpretation (e.g. [55]) calls
for an r-process enrichment of the material to be accelerated to GCR ener-
gies. A note of caution is in order at this point. As in the analysis of stellar
spectroscopic data, the discussion concerning the Pt group to the Pb group
abundance ratio relies heavily on the SoS splitting between s- and r-nuclides.
As already stressed before, these two contributions to Pb are very uncertain
(these uncertainties are much lower in the Pt group case, the r-process con-
tribution to which varies from about 85 to 100%; see Table 1 of [24]). Recall
that GCRs are made of much younger material than the SoS (i.e., 20 ∼ 30
Myr old). There is no proof at this time that the r-process(es) that has(have)
contributed to this recent sample of galactic material is(are) similar to the
one(s) contained in the SoS material.
The identification of actinides in the GCRs has been made possible quite
recently by the use of the Trek detector [55]. An accurate measurement of
their abundances relative to each other and to the Pt group is within the
reach of the planned Extremely Heavy Cosmic Ray Composition Observer
(ECCO) [56]. Such data would in particular help discriminating between var-
ious GCR sources that have been proposed, including fresh supernova ejecta,
supper-bubble material, or old galactic material (see [24] for some details and
references).
As in the stellar case, information on the isotopic composition of the GCRs
would be of prime interest in helping to evaluate the fractional contribution
of freshly synthesised r-process material to this quite recent sample of galactic
material. In fact, GCR composition measurements with isotopic resolution up
to at least Z ≈ 40 are within the reach of present detector technology, as
exemplified by the R-process Isotope Observer RIO project currently under
study [57].

7 The astrophysics of the r-process: parametrized

site-free scenarios
7.1 Canonical and ‘multi-event r-process (MER)’
high-temperature models
It has already been proposed by [29] that the r-process results from the avail-
ability of neutron concentrations that are so high that neutron captures (espe-
cially of the radiative type) are faster than β-decays, at least for a substantial
 Massive Stars Evolution and Nucleosynthesis 311

number of neutron-rich nuclides that are not located too far from the valley
of nuclear stability. This is in marked contrast to the situation envisioned for
the s-process. Such conditions clearly provide a natural way to transform any
pre-existing material into very neutron-rich species, viewed as the progenitors
of the r-nuclides. A classical additional hypothesis has been that the otherwise
unspecified stellar location is hot enough to allow (γ,n) photodisintegrations
to counteract to a more or less large extent the action of the inverse radia-
tive neutron captures. Finally, it is supposed that a decrease of temperature
that allows the ‘freezing-out’ of the photodisintegrations occurs concomitantly
with a decrease of the neutron density to values that are low enough to freeze
the neutron captures.6

(n, f)

abundance peak
in r-nuclei r-process
distribution path
accumulatinon
point

S
Sr
S
Sr r
Z
r
e
lin (n, γ )
ity
bil magic (γ , n)
sta
neutron
Seed pre-freezing β–
number
post-freezing β–

Fig. 22. Schematic representation of the r-process in the (N, Z)-plane

The aforementioned requirements on neutron concentration and tempera-

ture suffice to fix qualitatively several of the main features of the nuclear flow
associated with the r-process and to identify the involved nuclear physics.
Figure 22 depicts the situation very schematically. In the course of the trans-
formation of a given seed into more neutron-rich isotopes by a series of (n,γ)
reactions, (γ,n) photodisintegrations have a rate increasing with the neutron
excess or, equivalently, with the associated decrease of the neutron separation
energy Sn . For low enough Sn , the (γ,n) reactions counteract efficiently the
radiative neutron captures. At this point, the nuclear flow may proceed to
6
Let us recall that a transformation is said to be ‘frozen’ if its typical mean lifetime
gets longer than a typical evolutionary timescale of the considered astrophysical
site.
312 Marcel Arnould

higher Z elements through β-decays. In this picture, the flow takes a special
character at neutron closed shells. The especially low Sn values just past a
magic neutron number indeed hinder the flow to proceed to more neutron-rich
species, so that β-decays drive the material closer to the valley of stability fol-
lowing a path with increasing Z at practically constant N . The β-decays of
the corresponding relatively less exotic nuclei become less probable and tend
to slow down the nuclear flow. As a consequence, some material accumulates
at nuclei with a magic neutron number. However, when the path gets close
enough to the stability line, Sn is finally large enough to allow (n,γ) reac-
tions to become more rapid than the β-decays, and to proceed without being
largely counteracted by the (γ,n) photodisintegrations. The flow then resumes
normally until a new neutron magic number is reached. In this picture, the
accumulation of matter at neutron closed shell nuclei due to the relatively
slow β-decay bottlenecks provides a natural explanation of the SoS r-process
peaks (Fig. 15), as suggested in Fig. 22.
If the nuclear flow towards increasing Z values reaches the actinide or
transactinide region, it is stopped by neutron-induced or β-delayed fissions
which lead to a recycling of a portion of the material to lower Z values.
At freezing of the neutron captures or inverse photodisintegrations, mainly
β-decays, but also spontaneous or β-delayed fissions and single or multiple
β-delayed neutron emissions, drive the neutron-rich matter towards the valley
of stability. These post-freezing transformations are shown schematically in
Fig. 22.
The hypothesized high neutron fluence and temperature has been the
framework adopted by the vast majority of studies of the r-process. In many
cases, the consequences of such an assumption have been scrutinized only from
a purely nuclear physics point of view, just considering that one astrophysical
site or the other, and in particular the inner regions of massive star supernova
explosions, could be the required neutron provider. The early works of [29]
and of [58] have proposed the simplest and most widely used form of the
r-process scenario, referred to as the canonical r-process model (see e.g. [24]
for details). It assumes constant thermodynamic conditions (temperatures,
densities, neutron concentrations) during a given irradiation time. Over this
period, pre-existing iron-peak material is driven by neutron captures into a
location of the neutron-rich region determined by the neutron fluence and
by the highly temperature-sensitive reverse photodisintegrations. When the
irradiation stops, all neutron captures and photodisintegrations are abruptly
frozen (this implies that the capture of neutrons produced in β-delayed pro-
cesses is neglected). Even if this canonical model does not make reference to
any specific astrophysics scenario, but builds on nuclear properties only, it
has helped greatly paving the way to more sophisticated approaches of the
r-process.
A parametric approach of the r-process referred to as the ‘multi-event
r-process (MER)’ has been developed recently (see [24] for details). It
drops some of the basic assumptions of the canonical model, but keeps the
 Massive Stars Evolution and Nucleosynthesis 313

Fig. 23. Comparison between the SoS r-abundances (Fig. 15; the uncertainties are
not shown) and a MER fit obtained assuming a constant temperature T9 = 1.2 (in
billion K) and a superposition of events characterized by free neutron concentrations
Nn and numbers of neutrons captured by iron seed ncap (see Eq. (1)) located in the
indicated ranges. Within these prescribed limits, the characteristics of the events
involved and their relative contributions to the displayed yields are derived from an
iterative inversion procedure (see [24] for details and references)

simplification of constant thermodynamic conditions and neutron concentra-

tions during the irradiation time, after which all nuclear reactions are frozen.
It is an especially efficient tool to carry out systematic studies of the impact
of uncertainties of nuclear physics nature on yield predictions, which, in this
approach, necessitate the solution of huge nuclear reaction networks involving
thousand of nuclear species, and the availability of an even larger body of input
nuclear data (reaction and β-decay rates, fission probabilities,...). Figure 23
provides an example of fit to the SoS r-abundance distribution obtained with
the use of MER.
Let us also note that MER has been used to interpret the high similarity
of the SoS r-abundances between Ba and Os and those observed in r-process-
rich metal-poor stars. This situation is generally interpreted as the signature
of a ‘universality’ of the r-process (see Sect. 6.4). As reviewed by [24], the
main conclusions drawn from the MER results are that (1) the pattern of
abundances in the Ba to Os range is mainly governed by nuclear physics
properties (and in particular by the fact that even Z elements have more sta-
ble isotopes that can be fed by the r-process). If this is indeed true, a possible
314 Marcel Arnould

universality in this Z range does not tell much about specific astrophysical
conditions, and (2) the convergence of abundances in the above mentioned
range does not provide any demonstration of any sort of a more global uni-
versality involving lighter and heavier elements. With time, these reservations
have received mounting support from observation, as noted in Sect. 6.4.

7.2 Dynamical high-temperature r-process approaches (DYR)

In associating the r-process with supernova explosions, several attempts to

go beyond the canonical and MER models have been made by taking into
account some evolution of the characteristics of the sites of the r-process
during its development. These models are coined ‘dynamical’ (DYR) in the
following in order to remind of the time variations of the thermodynamic state
of the r-process environment (see [24] for references). These models do not
rely on any specific explosion scenario. They just assume that a material that
is initially hot enough for allowing a nuclear statistical equilibrium (NSE) to
be achieved expands and cools in a prescribed way on some selected timescale.
This evolution is fully parameterized.
With the requirement of charge and mass conservation, and if the relevant
nuclear binding energies are known, the initial NSE composition is determined
from the application of the nuclear Saha equation (e.g. Sect. 7-2 of [17] for a
general presentation) for an initial temperature and density (or, equivalently,
entropy), and electron fraction (net electron number per baryon) Ye . These
three quantities are of course free parameters in a site-free r-process approach.
The evolution of the abundances during expansion and cooling of the material
from the NSE state is derived by solving an appropriate nuclear reaction
network. The freeze-out of the charged-particle induced reactions might be
followed by an r-process.
With temperature, density and Ye as free parameters, many choices of ini-
tial NSE compositions may clearly be made, involving a dominance of light
or heavy nuclides, as illustrated in Fig. 24. However, in view of its relevance
to the supernova models, an initial NSE at temperatures of the order of 1010
K is generally considered. It favours the recombination of essentially all the
available protons into α-particles (the region noted NSE [n,α] in Fig. 24). The
evolution of this initial composition to the stage of charged-particle induced
reaction freeze-out has been analyzed in detail by [60], and we just summarize
here some of its most important features that are of relevance to a possible
subsequent r-process:
(1) at some point in the course of the expansion and cooling of the initially α-
rich material, full NSE breaks down as a result of the slowness of a fraction of
the charged-particle reactions relative to the expansion timescale. The forma-
tion of quasi-equilibrium (QSE) clusters results. In this state, the intra-QSE
composition still follows the NSE Saha equation, but the relative inter-cluster
abundances do not, and depend on the kinetics of the nuclear flows into and
out of the QSE clusters. To be more specific, the QSE phase is dominated in
 Massive Stars Evolution and Nucleosynthesis 315
0.5
NSE [n, α]
no r – processing
0.4

Electron Fraction Ye
NSE [n, heavies]
0.3

0.2
Solar – like r – Process

0.1
strong r – processing

0
0 1 2 3
log10 [ Entropy s ]

Fig. 24. The likelihood of a DYR r-process for given combinations of the electron
fraction Ye and the entropy per baryon s. A SoS-like r-process is expected for a
suitable superposition of conditions between the black lines. The results inferred
from an initial NSE phase at low s are smoothly connected to those of various nuclear
network calculations for high s values. In the latter cases, the assumed expansion
timescales imply that the freeze-out of the charged-particle induced reactions is
reached after dynamical timescales τdyn in excess of about 50 - 100 ms. The two
dotted lines represent the contours of successful r-processing for τdyn = 50 ms (left
line) and 100 ms (right line) (see [59] for details)

its early stages by a light cluster made of neutrons, α-particles and traces of
protons, and by a cluster made of 12C and heavier species. The population
of the latter is determined mainly by the α + α + n reaction, followed by
9
Be(α, n)12 C(n, γ)13 C(α, n)16 O, as first noticed by [61].
(2) As the temperature decreases further, the QSE clusters fragment more
and more into smaller clusters until total breakdown of the QSE approxima-
tion, at which point the abundances of all nuclides have to be calculated from
a full nuclear reaction network. In the relevant α-particle-rich environment,
the reaction flows are dominated by (α, γ) and (α, n) reactions with the addi-
tion of radiative neutron captures. Nuclei as heavy as Fe or even beyond may
result. For a low enough temperature, all charged-particle-induced reactions
freeze-out, only neutron captures being still possible. This freeze-out is made
even more efficient if the temperature decrease is accompanied with a drop
of the density ρ, which is especially efficient in bringing the operation of the
ρ2 -dependent α + α + n reaction to an end.
316 Marcel Arnould

In the following, the neutron-rich α-rich process summarized above will

be referred to as the α-process for simplicity, and for keeping the terminology
introduced by [62]) (the α-process of [29] refers to a different nuclear process).
The composition of the material at the time of freeze-out depends on the
initial Ye , on the entropy s (see [60] for a detailed discussion), as well as on
the dynamical timescale τdyn . The heavy nuclei synthesized at that moment
may have on average neutron numbers close to the N = 50 closed shell, and
an average mass number around A = 100. These nuclei can be envisioned
to be the seeds for a subsequent r-process, in replacement of the iron peak
assumed in the canonical and MER models. For a robust r-process to develop,
favourable conditions have to be fulfilled at the time of the α-process freeze-
out. In particular, the ratio at that time of the neutron concentration to the
abundance of heavy neutron-rich seeds has to be high enough for allowing even
the heaviest r-nuclides to be produced. As an example, A = 200 nuclei can be
produced if an average of 100 neutrons are available per A = 90 nucleus that
could emerge from the α-process, as it is the case if Ye is close to 0.5 [62], (ii)
at lower entropies if Ye is low enough, and/or (iii) if the temperature decrease
is fast enough for avoiding a too prolific production of heavy seeds. Figure 24
sketches in a semi-quantitative way the conclusions of the discussion above
concerning the likelihood of development of a successful r-process in terms of
entropy and Ye .

7.3 A high-density r-process scenario (HIDER)

Early in the development of the theory of nucleosynthesis, an alternative to

the high-T r-process canonical model (Sects. 7.1 and 7.2) has been proposed
[63]. It relies on the fact that very high densities (say ρ > 1010 gcm−3 ) can lead
material deep inside the neutron-rich side of the valley of nuclear stability as
a result of the operation of endothermic free-electron captures, this so-called
‘neutronisation’ of the material being possible even at the T = 0 limit. The
astrophysical plausibility of this scenario in accounting for the production of
the r-nuclides has long been questioned, and has remained largely unexplored
until the study of the composition of the outer and inner crusts of neutron
stars and of the decompression of cold neutronised matter resulting from tidal
effects of a black hole on a neutron-star companion ( [24] for references). The
decompression of cold neutron star matter has recently been studied further
(Sect. 9).
In view of the renewed interest for a high-density r-process, a simple steady
flow model, referred to in the following as HIDER, may be developed. Irre-
spective of the specific details of a given astrophysical scenario, it allows to
follow in a very simple and approximate way the evolution of the composition
of an initial cold (say T = 0) highly neutronized matter under the combined
effect of β-decays and of the captures of free neutrons that are an important
initial component of the considered material. These are the only two types of
transformations that have to be considered if fissions are disregarded, and if
 Massive Stars Evolution and Nucleosynthesis 317

any heating of the material resulting from the β-decay energy deposition is
neglected, so that photodisintegrations can be ignored.

Fig. 25. Abundance distributions predicted by the steady flow HIDER for Nn =
1020 and 1021 cm−3 . The details of the adopted nuclear physics can be found in [24].
The SoS abundances are shown for illustrative purposes

The predictions of HIDER under the additional assumption of a steady

flow (dN (A)/dt = 0, N (A) being the total abundance of all the isobars with
mass number A; see [24] for details) are illustrated in Fig. 25. This model does
roughly as well in reproducing the three SoS abundance peaks as a steady state
high-T canonical model for comparable neutron densities. In other words, a
high-T environment is not a necessary condition to account either for the
location, or for the width of the observed SoS r-abundance peaks.

8 The neutrino-driven DCCSNe: a high-temperature

site for the r-process?

The neutrino-driven winds that may accompany successful DCCSNe are cer-
tainly interesting from a purely hydrodynamical point of view, even if they
may turn out not to be the key triggering agents of DCCSNe. In addition,
318 Marcel Arnould

their nucleosynthesis has been scrutinized in detail, especially following the

excitement raised by the hope that they could provide a natural site for an
α-process and for a subsequent dynamical r-process (Sect. 7.2, and e.g. [64]
for early calculations). This hope has gained support from a one-dimensional
DCCSN simulation of an iron-core progenitor predicting that entropies as
high as about 400 could be attained in the wind more than 10 seconds after
bounce (e.g. [64]). Such a high entropy allows the development of a robust r-
process for a large variety of values of the neutron excess or Ye and dynamical
timescale τdyn (see Sect. 7.2 and Fig. 24). However, another one-dimensional
iron-core DCCSN model has predicted about five times lower entropies, so
that the development of an extended r-process is severely endangered [65].
The subsequent studies have confirmed that this r-process scenario could only
be recovered at the expense of some twists that are difficult to justify in
general (e.g. [66]).
In such an unsatisfactory state of affairs, the best one can do is to try
to understand better the physics of neutrino-driven winds through the devel-
opment of (semi-)analytical models some aspects of which may be inspired
by (failed) explosion simulations, and to try to delineate on such grounds
favourable conditions for the development of the r-process. These analytical
models confirm that the wind nucleosynthesis depends on Ye , entropy s, and
τdyn , as in the α-process discussed in Sect. 7.2. The wind mass-loss rate Ṁ
is influential as well. Ultimately, the quantities acting upon the synthesis in
the neutrino-driven DCCSN model depend crucially on the details of the in-
teraction of neutrinos with the innermost supernova layers, as well as on the
mechanisms that might aid to get a successful DCCSN, and whose relative
importance remains to be quantified in detail.
Several wind models of analytical nature exist. They differ in their level
of physical sophistication and in their way to parametrize the wind char-
acteristics. In all cases, the wind is assumed to be spherically symmetric,
which appears to be a reasonable first approximation even in two-dimensional
simulations, at least late enough after core bounce. In addition, the wind
is generally treated as a stationary flow, meaning no explicit time depen-
dence of any physical quantity at a given radial position. Newtonian and
post-Newtonian descriptions of a spherically symmetric stationary neutrino-
driven (supersonic) wind or (subsonic) breeze emerging from the surface of a
PNS have been developed. The reader is referred to [24] for the presentation
of a Newtonian, adiabatic and steady-state model for the wind and breeze
regimes, and for a general-relativistic steady-state wind solution.
A comment is in order here concerning the relevance of the wind or breeze
regimes. Is one of the two favoured by the DCCSN physics? This question is
far from being just academic, as it is likely that its answer may have some
impact on the predicted development of the r-process. It is quite intricate
as well. One difficulty arises as the neutrino-driven material is likely not to
flow unperturbed to infinity in a variety of DCCSN situations. The wind may
in particular interact with matter and radiation in that portion of the star
 Massive Stars Evolution and Nucleosynthesis 319

through which the SN shock has already passed. This interaction is likely to
depend, among other things, on the pre-SN structure. As an example, it is
clearly more limited as the mass of the outer layers decreases when going from
massive SN II progenitors to SN Ib/c events whose progenitors (Wolf-Rayet
stars) have lost their extended H-rich envelope prior to the explosion. The
interaction of the material ablated from the PNS and the outer SN layers has
several important consequences. It may give rise to a reverse shock responsible
for the fallback of a more or less large amount of material onto the PNS,
and whose properties (location and strength) alter more or less deeply the
characteristics of the neutrino-ejected material. For large enough energies of
the reverse shock, the wind may indeed transform into a breeze.

Fig. 26. Evolution of density ρ and of temperature T9 (in 109 K) calculated in the
Newtonian approximation for a breeze starting about 14 km away from a 1.5M PNS
with a temperature T9 = 9. The displayed solution depends on other parameters
(including entropy and mass loss rate) that are discussed in [24]

Figure 26 displays the evolution of temperature and density for a partic-

ular breeze solution discussed in [24]. Some snapshots of the corresponding
progressive build-up of heavy nuclei by the α-process, followed by the produc-
tion of r-nuclides are shown in Fig. 27. The r-nuclide abundance distributions
calculated for a breeze solution that is slightly different from the one dis-
played in Fig. 26 (even if the PNS mass and entropy are the same; see [24] for
details) and three different mass loss rates (leading to three different cooling
timescales) are shown in Fig. 28. The influence on the r-abundances of different
320 Marcel Arnould
100

80

60
X > 10–6
Z
40 10–6 > X > 10–8
10–8 > X > 10–10
t = 0.014 s t = 0.028 s
20 T9 = 3.0; ρ = 4 104 g / cm3
T9 = 5.4; ρ = 2.3 105 g / cm3

80

60
Z
40

t = 0.015 s t = 0.40 s
20 T9 = 0.63; ρ = 1.7 102 g / cm3
T9 = 0.77; ρ = 3 102 g / cm3

0
0 50 100 150 0 50 100 150 200
N N

Fig. 27. Snapshots of the nuclear flows calculated for the breeze solution of Fig. 26.
The first three panels (t ≤ 0.15 s) describe the progressive build-up of heavy nuclei
by the α-process. These act as the seeds for the r-process that develops after the
α-process freeze-out, as shown in the last panel

Fig. 28. Distribution of the r-nuclide abundances derived for a breeze solution
obtained with a PNS mass of 1.5M , an initial electron fraction Ye = 0.48, and
different values of the mass loss rate (in units of 10−5 M
/s). The upper curve
corresponds to the SoS r-nuclide abundances normalized to i Xi = 1 (see [24] for
details, particularly on the selected breeze solution)
 Massive Stars Evolution and Nucleosynthesis 321

100 1015
10–1 Ye = 0.03
1014
Ye = 0.04
10– 2
1013
Ye = 0.10

ρ [g / cm3]
10– 3
nb [fm–3]

Ye = 0.30 1012
10– 4
Core 1011
–5
10
Inner crust 1010
10– 6
109
Outer crust
10–7
9.8 10 10.2 10.4 10.6 10.8
r[km]
Fig. 29. Density profile (expressed in baryon number density nb and matter density
ρ in a typical 1.4 M NS showing the structure of the crust. The values of the electron
fraction Ye are given at ρ = 4 1011 , 2 × 1012 , 1013 and 1014 g/cm3

values of other breeze parameters is discussed by [24], as is the influence of

the neutrino luminosity.
A discussion of the prospects of development of an r-process in a variety of
multi-dimensional simulations including rotation, magnetic fields, or accretion
in binary systems can also be found in [24]. In short, it is fair to say that no
firm conclusion can be drawn on the possibility of development of the r-process
in CCSNe.

9 Compact objects: a site for the high-density

r-process scenario?
As reminded in Sect. 7.3, the decompression of the crust of cold neutron stars
(NSs) made of a lattice of very neutron-rich nuclei immersed in a gas of
neutrons and degenerate electrons has long been envisioned as a possible site
for the development of a high-density r-process (HIDER). This decompression
could result from the coalescence of two NSs or of a NS and a black hole (BH)
in a binary system. It could also result from the ejection of material from
magnetars.
The modelling of the coalescence of two NSs or of a NS and a BH has
recently attracted a flurry of interest (see the review by [24]). These events are
indeed considered as being among the strongest known sources of gravitational
wave radiation, this emission being in fact responsible for the coalescence after
322 Marcel Arnould

Fig. 30. Final composition of the ejected NS inner crust material with four differ-
ent initial densities lower than 1012 g/cm3 (black squares). The SoS r-abundance
distribution is also shown by open dots (from [24])

typical times of the order of tens of millions to billions of years. They are also
viewed as the likely progenitors of the class of short hard gamma-ray bursts.
Magnetars may result from certain JetSNe. Their properties are reviewed
in several places (see [24] for references). They are observed as Soft Gamma
Repeaters emitting sporadically bright bursts of energy over a period that
is estimated to be of the order of 10000 years. Their most remarkable char-
acteristic is their magnetic fields whose values are typically of the order of
1014−15 G, that is 100 to 1000 times larger than classical pulsar values.
The matter ejected from magnetars or from a NS coalescence can be highly
neutron-rich, the distribution of Ye in this material being in fact reminiscent
of the neutron crust values (Fig. 29). Figure 30 shows the final composition
of the material ejected from the NS inner crust at densities ρ < 1012 g/cm3 .
Although the composition is seen to be density-dependent, it is characterized
by peaks with location and width similar to the SoS ones. For inner crust layers
at higher densities (ρ > 1012 g/cm−3 ), large neutron-to-seed ratios bring the
nuclear flow into the superheavy region, from which fission recycles material
to lighter nuclides. Figure 31 shows the abundance distribution resulting from
the decompression of a clump of material with initial density ρ = 1014 g/cm3 .
 Massive Stars Evolution and Nucleosynthesis 323

Fig. 31. Final r-abundance distribution (black squares) for a clump of material
with initial density ρ = 1014 g/cm3 expanding on a timescale τexp = 6.5 ms. The SoS
r-abundance distribution is also shown

10 Some brief comments on the modelling of the

evolution of the r-nuclide content of the Galaxy
and on nucleo-cosmochronology

A quite natural astrophysicists’ dream is to understand the wealth of data

on the evolution of the r-nuclide content of the Galaxy that are accumulat-
ing from very many spectroscopic observations (Sect. 6.4). These observations
clearly demonstrate a huge complexity that will probably keep rising as new
observations become available. The best one can thus imagine at this stage is
to explain broad trends which may be identified through the analysis of the
r-nuclide abundance information. In this exercise, one has always to keep in
mind that, at best, more or less reliable elemental abundances are derived
from spectral analyses that often rely on approximate classically used tech-
niques. This necessitates to disentangle the s- and r-process contributions to
a given elemental abundance. It is generally done by assuming that these two
nucleosynthesis contributions are at a largely metallicity-independent relative
level, and thus do not differ widely from the SoS case. This assumption cannot
yet be ascertained in any quantitative way, and is in fact not expected to hold
for the s-process.
Various galactic chemical evolution toy models have been constructed,
which often focus on the evolution of the abundances of two representative
elements, Ba (a s-process element in the SoS) and Eu (a r-process element
in the SoS). They adopt different schematic descriptions of the galactic halo
324 Marcel Arnould

and disk, and different prescriptions for the physical input quantities to these
models. In particular, fully ad-hoc assumptions are made or free parameters
are chosen concerning the r-process yields from stars of different masses and
metallicities.
Some predictions for [Eu/Fe] have been made recently by [67]. They are
based on a homogeneous one zone model in which it is assumed that stars in
prescribed mass ranges produce an artificially selected amount of r-nuclides
through the neutrino wind or prompt explosion mechanisms (Sects. 4 and 8).
As expected, the predicted [Eu/Fe] ratio is very sensitive to the selected stellar
mass ranges. This result might be optimistically considered as providing a way
to constrain the site(s) of the r-process from observation. Reality is most likely
less rosy, as very many uncertainties and severe approximations drastically
blur the picture.
The assumption of the homogeneity of the interstellar medium at all times
is dropped by [68]. The resulting inhomogeneous model might increase the
plausibility of the predictions especially at early times in the galactic his-
tory. With the granularity of the nucleosynthesis events duly considered, one
might hope to better account for the large observed scatter of the r-nuclide
abundances at very low metallicities. In addition, the model of [68] takes into
account the r-process contribution from NS mergers (Sect. 9) on top of the one
from supernovae in selected mass ranges. The many other simplifications gen-
erally made in other chemical evolution models are also adopted by [68]. This
concerns in particular the r-process yields from supernovae, as well as from
NS mergers, that are just taken to be SoS-like. From their predicted [Eu/Fe]
ratio, [68] conclude that the scenario assuming the predominance of SN II
events in the 20 to 50 M range allows the best fit to the observations. This
result is obtained for total masses of r-nuclides per supernova varying from
about 10−4 M down to about 10−7 M when going from 20 to 50 M stars.
Again, this conclusion has to be taken with great care in view of the many
uncertainties and approximations involved in the chemical evolution model.
Within the same model, it is also claimed that NS mergers are ruled out as
the major source of r-nuclides in the Galaxy. This conclusion relies on a very
approximate and highly uncertain time-dependent frequency of the events. In
order to cope at best with observational constraints, coalescence timescales
and amount of r-nuclides ejected per merger are adopted by [68] to vary from
about 0.1 to 10−4 M , depending upon other parameters of the NS merging
model.
All in all, we consider that the galactic chemical evolution models devised
up to now are by far too schematic and uncertain to provide a reliable tool to
account for the observed evolution of the r-nuclide content of the Galaxy, or
for constraining the possible sites of the r-process. A priority would clearly be
at least to identify with a reasonable confidence a single site for the r-process
before dwelling on the grand project of constructing models for the evolution
of the r-nuclide content of a whole galaxy, and of the SoS in particular.
 Massive Stars Evolution and Nucleosynthesis 325

As far as nucleo-cosmochronology is concerned, the actinides produced by

the r-process enter in particular attempts to estimate the age of the Galaxy
through their present SoS content, or through their abundances evaluated at
the surface of very metal-poor stars. These attempts face some severe prob-
lems related to the nuclear physics and astrophysics uncertainties that affect
the predictions of the actinide production. The situation is worsened further
by the especially large uncertainties in the contribution of the r-process to the
solar-system Pb and Bi content (see [24] for a brief review and references).
Concerning the 232Th - 238U and 235U - 238U pairs classically used to date
the Galaxy from their present meteoritic abundances, the opinion has been
expressed more than 20 years ago by [69] that they have just limited chrono-
metric virtues. This is in marked contrast to a widely-spread and repeated
claim following [70], and based on simple analytic models for the evolution of
Th and U in the Galaxy. In addition, the chronometric predictions based on
the observations of Th and U in very metal-poor stars have to be considered
with great care (see [24] for a brief review and references). In order for them
to be reliable, it is not only required that the production of the actinides by
the r-process is well known, but also, and very decisively, that the produc-
tion of Th with respect to U and to the Pt peak is universal. Observation
demonstrates now that this is not the case.
The bottom line of this brief review of the r-process is that unanswered
questions are by far more numerous than solved problems when one is deal-
ing with this nucleosynthesis mechanism. They concern especially the astro-
physics of the process, as no single site has been identified decisively yet. It
also raises many nuclear physics questions. In such conditions, the modelling
of the evolution of the r-nuclide content of the Galaxy and actinide-based
chronometric evaluations cannot be based on solid grounds yet. This is in fact
a very pleasing situation, as hope for many exciting discoveries is still ahead
of us!

11 The p-process: Some generalities

The nuclear physics and astrophysics aspects of the p-process have been re-
viewed in substantial detail by [32]. We limit ourselves here to a summary of
this review.
Quite clearly, the neutron-deficiency of the p-nuclides forbids their produc-
tion in neutron capture chains of the s- or r-types. In contrast, and as already
proposed very early in the development of the theory of nucleosynthesis [29],
they could well be synthesized from the destruction of pre-existing s- or r-
nuclides by different combinations of (p,γ) captures, (γ,n), (γ,p) or (γ, α)
reactions. Some β-decays, electron captures or (n,γ) reactions can possibly
complete the nuclear flow. These reactions may lead directly to the produc-
tion of a p-nuclide. In most cases, however, they are synthesized through
an unstable progenitor that transforms into the stable p-isobar through
326 Marcel Arnould

(a chain of) β-decay(s). Figure 32 displays in a very schematic way some

possible nuclear routes through which seed s-(r-) nuclides can be transformed
into p-nuclides.

Fig. 32. Schematic representation of some simple possible nuclear routes through
which seed s- or r-nuclides (black dots) can be transformed into a p-nuclide (black
square). Unstable nuclei are represented by open dots. Routes (1) and (2) are made
of a succession of (p,γ) and (γ,n) reactions leading directly to the p-nuclide. Slightly
more complicated chains involve (p,γ) reactions followed by β-decays (route (3)), or
a combination of (γ,n) and (γ,p) or (γ,α) and β-decays (route (4)). More complicated
flow patterns involving combinations of the represented ones can also be envisioned.
The p-nuclide destruction channels are not represented

The relative importance of the various ways to make p-nuclides from more
neutron-rich seeds depends critically on temperature. This relates largely to
the properties of the photodisintegration rates which are known to be strongly
increasing functions of temperature. In fact, values typically in excess of
about 1.5 × 109 K are required for the photodisintegrations to take place
on timescales commensurable with stellar evolution ones. It is also manda-
tory for these reactions to freeze-out before a too strong photoerosion of the
heavy nuclides, which would leave iron peak nuclei as the main end products.
In practice, this forces temperatures not to exceed about 3.5 × 109 K during
a short enough time only. These three constraints (abundant enough seed nu-
clei, high enough temperatures, short enough timescales for the hot phases)
 Massive Stars Evolution and Nucleosynthesis 327

are obviously complemented by the necessity of considering proton-rich layers

if (p,γ) captures are indeed envisioned.7
The details of the nuclear physics for the p-process have been discussed
by [32], and are not reviewed here. Static properties have to be known for
a large variety of neutron-deficient nuclei from about iron to lead located
between the valley of nuclear stability and the vicinity of the proton-drip line,
As made clear in Fig. 32, nucleon and α-particle capture rates by these nuclei
are needed as well. Free electron captures may also be requested.
In spite of the general consensus that the p-nuclides largely originate from
photo-eroded s- or r-nuclides, the identification of the stellar sites in which
the p-process has a chance to develop has changed most dramatically since
[29], where it has been considered that the constraints on the operation of
the p-process mentioned above could be best satisfied in the H-rich envelopes
of SN II explosions. This scenario has been put to its limits by [74]. At the
end of the sixties already, it was realized that the required high temperatures
are very unlikely to be reached in SN II envelopes. This is why it has been
suggested by [75, 76] to locate the p-process in the deep O-Ne-rich layers of
massive stars either in their pre-SN or SN phases. Subsequently, other sites
have been considered as well, like SN Ia or exploding sub-Chandrasekhar mass
white dwarfs following the accretion of He-rich material. Other possibilities
are reviewed in detail by [32].

11.1 The p-process in SN IIe

Following the suggestion by [75,76], most of the investigations of the p-process

have been concerned with the explosion of the O-Ne-rich layers associated with
SN II events. The first calculations of this type [77] made use of parametrized
conditions for the thermodynamic description of the relevant supernova lay-
ers, of solar s-nuclide seeds, and of a simplified reaction network. The most
detailed studies performed so far [78, 79] are conducted in the framework of
one-dimensional explosions of non-rotating solar-metallicity stars in the 13 to
15 ≤ M ≤ 25M mass range.
These computations demonstrate that the p-process can develop in the
O-Ne layers of the considered massive stars explosively heated to peak tem-
peratures Tm comprised approximately between 1.7 and 3.3 × 109 K. These
zones, referred to in the following as the p-process layers (PPLs), are located
far enough from the base of the supposedly ejected material for their physi-
cal properties not to be affected drastically by the very complicated physics
of the deeper SN II layers possibly of importance for the r-process (Sect. 8).
7
Note that the above considerations concern only thermonuclear modes of p-
nuclide synthesis. Some non-thermonuclear scenarios have in fact been proposed,
like production by spallation reactions in the interstellar medium [71, 72], or
photonuclear reactions triggered by non-thermalized photons [73]. These models
suffer from either too low efficiencies, or from constraints that limit their astro-
physical plausibility. They are not discussed further here.
328 Marcel Arnould

In fact, what is basically required is just a successful explosion with a typical

SN kinetic energy [32]. Another essential requirement is to have enough s- (or
possibly r-) process seeds for the production of the p-nuclides. The s-process
accompanying core He burning in massive stars can do the job. It strongly
enhances the 60 <∼A< ∼ 90 s-nuclei, while the abundances of the heavier nuclei
are only very weakly increased.

Fig. 33. Location in the (N, Z)-plane of the stable isotopes of the elements between
Fe and Bi. The p-isotopes are represented by black squares, while both the s-, r-,
sr- or sp-isotopes are identified with open squares (see Figs. 13 - 15 for details). The
p-nuclides are the progeny of unstable neutron-deficient isobars located on the down-
streaming p-process flow (thick black line; for more details on the p-process flow,
see [32]). Some possible r-process flows derived from a high-temperature parametric
model (Sect. 7) are also shown, as well as the up-streaming s-process flow (thin
black line) confined at the bottom of the valley of nuclear stability. The proton and
neutron drip lines correspond to the locations of zero proton and neutron separation
energies

Figure 33 provides a schematic view of a typical SNII p-process flow. Its

distinguishing feature is that it evolves from heavy to light nuclides as a
result of the dominant action of photodisintegrations. Figure 34 displays p-
nuclide yields in the form of normalized overproduction factors computed for
a variety of SNII explosions of solar-metallicity stars (results for each star
 Massive Stars Evolution and Nucleosynthesis 329

100

10 Pt Hg
Se Ba Dy W
Hf
< F > /F0

Kr Sn Sm Os
Xe
1 Ta
Sr Cd Yb
Pd Ce
Te Er
0.1 Ru In
Mo La Gd
Sn

0.01
80 100 120 140 160 180 200
Mass Number
Fig. 34. Yields of the p-nuclides from SNIIe. They are expressed in terms of nor-
malized overproduction factors
Fi (M )/F0 (M ). The mean overproduction factor

Fi (M ) for a star with mass M is defined as the total mass of the p-nuclide i
produced in its PPLs divided by the corresponding mass if the PPLs had a solar
composition. The normalizing factor F0 (M ) is the mean overproduction factor aver-
aged over the 35 p-nuclides listed in [32]. With these definitions, all the normalized
overproductions would be equal to unity if the derived abundance patterns were
solar. Ranges of variations of the normalized p-nuclide overproductions represented
by vertical bars cover the results obtained for individual SNII explosions of solar
metallicity stars with masses in the 13 to 25 M range. Open squares indicate the
values of the normalized overproductions obtained by integration over the IMF pro-
posed by [80]. Solid lines join different p-isotopes of the same element. The displayed
180
Ta data correspond to the sum of the 180Tag and 180Tam abundances. Roughly
speaking, 180Tam (the only Ta form to be present in the SoS) represents about half
of the total 180Ta SNII yields (see [32])

are not shown. Instead, only the ranges covered by the individual cases are
represented). It is seen that, roughly speaking, the overproduction factors
depend only relatively weakly on the star masses, this being less true, however,
for the lightest p-nuclides 74Se, 78Kr and 84Sr. As a zeroth-order approximation
to the build-up of a full galactic chemical evolution model, the p-process yields
have been averaged over an IMF, leading to the IMF-averaged normalized
overproduction factors also displayed in Fig. 34. Of course, the build-up of a
model for the temporal evolution of the galactic content of the p-nuclides not
only requires the knowledge of the stellar mass dependence of their yields,
but also their variations with the metallicity. This question has been first
330 Marcel Arnould

addressed qualitatively by [78] on grounds of calculations performed for the

supernova SN 1987A. More details about these results can be found in [32].
From Fig. 34, it is concluded that about 60% of the produced p-nuclides
fit the SoS composition within a factor of 3. Some discrepancies are also
apparent. In particular, the Mo and Ru p-isotopes are severely underproduced.
A possible remedy based on uncertainties in the seed s-process abundances
due to an uncertain rate for the neutron production reaction 22Ne (α , n) 25Mg
is discussed by [32]. More specifically, an increase of this rate leads to an
enhanced production of the Mo and Ru p-isotopes. The nuclides 113 In, 115 Sn,
138
La and 152Gd are also underproduced. No other clear nucleosynthetic source
has been identified yet for 113In. In contrast, some 115Sn and a substantial
amount of the solar 152Gd may owe their origin to the s-process. This process
also contributes to 164Er. From the level of p-process production of this nuclide
shown in Fig. 34, it appears that 164Er may well have a combined s- and p-
origin without having to fear any insuperable overproduction with respect
to the neighbouring p-nuclides. The rare odd-odd nuclide 180Tam emerges
naturally from all the considered SNII models in quantities consistent with its
SoS abundance. This interesting case is discussed further by [32]. The other
rare odd-odd p-nuclide 138La is found to be of unlikely thermonuclear origin
in SNII explosions, even if reasonable uncertainties on the production and
destruction rates are taken into account. As reviewed by [32], 138La could
result instead from the captures by the initially relatively abundant 138Ba of
electron-neutrinos streaming out of the nascent PNS in a SNII explosion.
The impact on the p-nuclide yields displayed in Fig. 34 of a change of the
SNII explosion energy, of metallicity, and of reaction rate uncertainties is also
discussed by [32].
In order for SNII to be potentially important contributors to the galactic
p-nuclides, it remains to make sure that they do not produce too much of
other species, and in particular oxygen, the abundance of which is classically
attributed to SNII explosions. This question has been examined in some detail
by [78], who conclude that the p-nuclides are globally underproduced in a solar
metallicity 25M star by a factor of about 4 ± 2 relative to oxygen when all
the abundances are normalized to the bulk SoS values. The problem of the
oxygen overproduction may be eased in lower mass SNII explosions, while it
is worsened with decreasing metallicity. It may thus be that the p-process
enrichment of the Galaxy has been slower than the oxygen enrichment. There
is at present no observational test of this prediction. The O overproduction
may also be cured by a higher than nominal production of s-nuclides, which
might also help solving the Mo-Ru underproduction problem, as discussed
above. At this point, it must also be kept in mind that uncertainties in stellar
convection introduce a large uncertainty (a factor of 2–3) in the oxygen yield
of massive stars [81]. Likewise, the p-nuclide production might be affected
as well. It has also to be stressed that the influence of multi-dimensional
phenomena, and possibly of rotation, on the relative production of oxygen
and the p-nuclides remains largely unknown.
 Massive Stars Evolution and Nucleosynthesis 331

To conclude this section, let us note the following: (1) it has been pro-
posed that some p-nuclides could be produced under certain circumstances in
Ye > 0.5 neutrino-driven winds suggested by some SNII simulations [82], or
through antineutrino absorption by intermediate-mass nuclei formed at some
distance from the PNS in a SNII explosion [83]. It remains to be convincingly
demonstrated that these scenarios can in particular help solving the Mo-Ru p-
isotope puzzle, as sometimes claimed; (2) massive stars exploding as SNIb/c
following a Wolf-Rayet phase might also provide a suitable site for the p-
process in their Ne-O-rich layers. No detailed study of this scenario has been
conducted to-date, and (3) massive enough stars might explode as so-called
‘pair-creation supernovae’. The possibility of development of the p-process in
such explosions is reviewed by [32].

11.2 The p-process in SNIa

The canonical model of SNIa explosions calls for a thermonuclear runaway

of carbon in a C-O white dwarf (WD) that has reached a mass close to the
Chandrasekhar limit (M ≈ 1.4 M ) following the accretion of matter from
a companion in a binary system. The nuclear burning in the course of the
explosion is confined to very thin layers. These propagate either diffusively
as subsonic deflagrations (‘flames’) or by shock compression as supersonic
detonations, both modes being hydrodynamically unstable to spatial pertur-
bations. A so-called ‘delayed detonation’ regime has also been proposed. It
corresponds to the formation of a detonation wave when the outward moving
deflagration flame is quenched by the expansion of the crossed layers. Some
variants to the exploding C-O WD scenario have been proposed. They include
the thermonuclear explosion of oxygen-neon WDs or the merging of two WDs.
Large uncertainties of various sorts still affect the modelling of SNIa explo-
sions (e.g. [84]). They influence more or less drastically the nucleosynthesis
predictions, including those concerning the p-process.
Figure 35 presents the total p-nuclide yields from the SNIa deflagration
model considered by [32], and with two different s-seed distributions, the
proper one being largely unknown at this time. The conclusions one can derive
from the inspection of this figure are quite similar to those resulting from the
consideration of massive star explosions (Fig. 34), or from a delayed detonation
scenario (see [32]). This concerns in particular the general pattern of the p-
nuclide distribution, its high sensitivity to the assumed seed distribution, and
the set of underproduced species (esp. 92,94Mo, 96,98Ru, 113In, 115Sn, and 138La.
Note that 180Ta is also underproduced in the deflagration model adopted to
construct Fig. 35, in contrast to the SNII predictions shown in Fig. 34). It has
of course to be kept in mind that the simulation of deflagrations is extremely
intricate, and model uncertainties certainly bring their share to the reliability
of the yield predictions.
332 Marcel Arnould

Xe Ba
104 Sn
Te
Pd Yb Hf Pt
Ce
Se Dy Er
Sm
Sr Hg
103
W
<F>

Os
102
Kr
Ta
Sn
Cd
Ru Er
101 Gd
La

Mo In

100
80 100 120 140 160 180 200
Mass Number
Fig. 35. Abundances of the p-nuclides in a SNIa deflagration model. They are
calculated with solar s-seeds (black squares), or with a seed distribution which is
representative of the s-process in solar-metallicity AGB stars (open squares) (see
[32] for details)

11.3 The p-process in sub-Chandrasekhar white dwarf explosions

A scenario referred to as a ‘sub-Chandrasekhar-mass’ supernova envisions a

C-O WD capped with a helium layer accreted by a companion, and which
explodes as the result of a hydrodynamical burning before having reached the
Chandrasekhar limit. This type of explosions may exhibit properties which
do not match easily the observed properties of typical SNIa events. It can-
not be excluded, however, that they are responsible for some special types of
events, depending in particular on the He accretion rate and on the CO-sub-
Chandrasekhar WD (SCWD) initial mass (e.g. [85]). Unidimensional simu-
lations of He cataclysmics characterized by suitably selected values of these
quantities reach the conclusion that the accreted He-rich layer can detonate.
Most commonly, this explosion is predicted to be accompanied with the C-
detonation of the CO-SCWD. In some specific cases, however, this explosive
burning might not develop, so that a remnant would be left following the He
detonation. Multidimensional calculations cast doubt on the nature, and even
occurrence, of the C-detonation in CO-SCWD (e.g. [86]).
The possibility of development of the p-process following the detonation
of the He-rich material assumed to be accreted onto a CO-SCWD has been
explored by [88] in the framework of a 1D model of the He detonation of a
non-rotating 0.8M WD. This work has been extended to a 3D simulation
by [41]. It gives results that largely confirm the yield predictions of the 1D
calculation, to which we limit the discussion in what follows.
 Massive Stars Evolution and Nucleosynthesis 333

Fig. 36. Snapshots in the nuclidic chart of flow patterns in a 1D model of a deto-
nating He layer accreted onto a 0.8M WD. The selected times and corresponding
temperatures or densities are given in different panels. The stable nuclides are indi-
cated with open squares. The magic neutron and proton numbers are identified by
vertical and horizontal double lines. The drip lines predicted by a microscopic mass
model are also shown. The abundances are coded following the grey scales shown in
each panel. At early times (bottom left panel), an r-process type of flow appears on
the neutron-rich side of the valley of nuclear stability. At somewhat later times (top
left panel), the material is pushed back to the neutron-deficient side rather close to
the valley of β-stability. As time passes (two right panels), a pn-process [87] develops

A thermonuclear runaway develops near the base of the He envelope. The

associated outward-moving detonation wave heats the matter to temperatures
around 3×109 K, and leads to the ejection of about 0.2 M into the interstellar
medium. The associated nucleosynthesis is especially complex, and is reviewed
in detail by [32]. The situation may be summarized by saying that a quite
high neutron density builds up at the beginning of the detonation, and leads
to a weak r-process. Subsequently, the nuclear flow is pushed back to the
valley of nuclear stability, and eventually into the neutron-deficient region,
by photodisintegrations that become more and more efficient as temperature
rises. A process coined ‘pn-process’ by [87] then develops. The associated
flow is displayed in Fig. 36. Its main path lies much further away from the
proton-drip line than in the classical rp-process, which eases somewhat the
nuclear physics problems. This results from the lower proton concentration
334 Marcel Arnould

than classically considered in the rp-process, and from the non-zero neutron
concentrations encountered in the He detonation. Further details on the pn-
process can be found in [32].

105 Kr Pt
Sr Yb Hf W Hg
104 Ce Er
Se Mo Ru Xe
Dy
Cd
103
Sn
<F>

102
Os
Ba Sm
101 Pd Te
La Er
0 In Gd Ta
10
Sn
10–1
80 100 120 140 160 180 200
Mass Number
Fig. 37. Overproduction of the p-nuclides resulting from the He-detonation consid-
ered in Fig. 36. The full and open symbols correspond to SoS s-seed abundances,
and to SoS seeds increased by a factor 100, respectively (from [32])

Figure 37 shows that almost all the p-nuclei are overproduced in solar pro-
portions within a factor of 3 as a combined result of the p- and pn-processes.
This includes the puzzling Mo and Ru p-isotopes, as well as 138La, which ap-
pears to be overproduced at the same level as the neighbouring p-nuclides.
This contrasts with the situation encountered in SNII (Sect. 11.1) or SNIa
(Sect. 11.2) explosions.
An embarrassment with the SCWD scenario may come from the fact that
the Ca-to-Fe nuclei are predicted to be overabundant with respect to the p-
nuclei, but 78Kr, by a factor of about 100. Taken at face value, this situation
implies that the considered He detonation cannot be an efficient source of
the bulk SoS p-nuclides (see [32] for details). In order to cure this problem,
one may envision enhancing the initial abundance of the s-seeds, which has
already been seen in Sect. 11.1 to be an essential factor determining the level
of production of the p-nuclides. Figure 37 shows that an increase by a factor
100 of the s-nuclide abundances over their SoS values makes the overpro-
duction of a substantial variety of p-nuclides comparable to the one of 78Kr
and of the Ca-to-Fe nuclei. The factor 100 enhancement would have to be
increased somewhat if the material processed in the core of the CO-SCWD by
a C-detonation were ejected along with the envelope. At this point, one essen-
tial question concerns the plausibility of the required s-nuclide enhancement.
 Massive Stars Evolution and Nucleosynthesis 335

There is no definite answer to this key question. This enrichment of the ac-
creting WD might result from its past AGB history if indeed some of its outer
s-process enriched layers could be mixed convectively with part at least of the
accreted He-rich layers before the detonation. Alternatively, the He-rich layer
accreted by the WD could be (or become) enriched in s-process elements. Such
speculations need to be confirmed by detailed simulations.
An analysis of the impact of nuclear uncertainties on the p-abundances
concludes that the results displayed in Fig. 37 are not drastically affected by
other choices of nuclear reaction sets (see [32] for details).

11.4 Some comments on the p-process isotopic anomalies

and chronometry

There is now observational evidence for the existence of isotopic anomalies

involving the p-isotopes of Kr, Sr, Mo, Xe, Ba and Sm in various meteoritic
materials [32]. These anomalies manifest themselves as excesses or deficits
of the abundances of the p-nuclides with respect to the more neutron-rich
isotopes, when comparison is made with the bulk SoS mix. In addition, two
isotopic anomalies are attributable to the now extinct neutron-deficient ra-
dionuclides 92Nbg and 146Sm which have decayed in the meteoritic material
where excesses of 92Zr and 142Nd are observed.
Some of these anomalies, in particular in Xe, are puzzling, and cannot be
explained by any p-process scenario [32]. As far as 92Nbg is concerned, there is
some observational evidence for its presence in the early Solar System. With
its half-live of 3.6 × 107 yr, some hope has been expressed that it could help
building a chronometry for the p-process. However, the controversial early
SoS 92Nbg abundance combined with the substantial uncertainties of nuclear
physics and astrophysics natures which remain in the evaluated 92Nb/ 92Mo
production ratios make by far premature any attempt to develop a meaningful
92
Nb-based p-process chronometry [32]. The radionuclide 146Sm (α-decay half-
live t1/2 = 1.03 × 108 yr) has also been proposed as a p-process chronometer.
Again, and particularly in view of uncertainties in its production by the p-
process, it is without doubt highly risky, if not impossible, to construct a
meaningful p-process chronometry based on 146Sm [32].
By way of conclusion of this brief review of the p-process, one may say
that substantial progress has been made in the understanding of this nucle-
osynthetic mechanism since [29]. A variety of astrophysical sites have been
identified as possible active p-nuclide contributors. The relative success of the
p-process calculations relying on detailed stellar models does not have to hide
some puzzling problems raised by specific cases, the most publicized ones con-
cerning 92,94Mo, 96,98Ru and 138La. Some remedies have been proposed, but
remain to be scrutinized in more detail. Some questions of nuclear physics
or astrophysics nature the study of which would help putting the p-process
modelling on safer grounds are identified by [32].
336 Marcel Arnould

12 Summary and prospects

In view of their importance in many astrophysical questions, massive stars

have been the subject of much work devoted to their evolution and nucleosyn-
thesis. In spite of this, much obviously remains to be done, particularly on
various astrophysics aspects of both the pre-supernova and supernova stages
of their evolution. Among the questions that still call for further progress, let
us mention the proper description of the transport of matter or of rotation,
and of the structural effects of magnetic fields or of various instabilities that
can develop prior to an eventual explosion. Uncertainties in the rates of some
nuclear reaction and of weak interaction processes certainly do not have to be
forgotten. Multi-dimensional simulations, especially at the late pre-supernova
phases, are clearly in need.
The supernova explosions are still poorly understood, and it is still a long
way to go to their successful simulations based on less uncertain pre-supernova
models. Here again, the multi-dimensional treatment of a variety of physical
effects, including rotation, magnetic fields, instabilities of different origins, and
the transport of neutrinos, appears to be required.
As a direct result of the remaining structural and evolutionary uncertain-
ties, the nucleosynthetic yields from massive stars have to be taken with some
care, even if general trends seem to be understood. Of course, nuclear physics
uncertainties also bring their share of difficulties to these predictions.
The r-process remains one of the most vexing nucleosynthetic puzzle. We
briefly summarise here a selection of some r-process problems (the reader is
referred to [24] for more details).
(1) The r-nuclide content of the solar system. Since [29], the bulk SoS
content of r-nuclides has been a key source of information. Uncertainties un-
fortunately still affect a variety of SoS r-nuclide abundances, which weakens to
some extent their constraining virtues on r-process models. These uncertain-
ties stem from various origins, including meteoritic data or, more importantly,
models for the SoS s-process content, from which r-abundances are tradition-
ally derived. They need to be kept in mind in particular when interpreting
spectroscopic data [see (2) below]. It is difficult to see how this situation can
be improved in a decisive way. Isotopic anomalies involving some r-nuclides
have been identified in meteorites. They do not bring much useful information
on the identification of the site and characteristics of the r-process.
(2) The evolution of the r-nuclide content of the Galaxy. A substantial
observational work has been conducted in recent years in order to identify the
level of r-process contamination of the Galaxy with time. This effort largely
relies on the abundance evolution of Eu, classified as an r-process element on
grounds of SoS abundance analyses. The main conclusions derived from this
observational work may be summarised as follows:
(i) the Eu data are classically used to support the idea that the r-process has
contributed very early to the heavy-element content of the Galaxy. However,
the scatter of the observed Eu abundances introduces some confusion when
 Massive Stars Evolution and Nucleosynthesis 337

one tries to establish a clear trend of the Eu enrichment with metallicity.

It is also difficult to identify the value of [Fe/H] at which the signature of
the s-process becomes identifiable. This conclusion relies in particular on La
abundances derived from observation, La being classically considered as an
s-element in the SoS (even if a non-negligible r-process contribution cannot
be excluded). A most useful information on the relative evolution of the s-
and r-process efficiencies in the Galaxy would be provided by the knowledge
of the isotopic composition of the neutron-capture elements. Such data are
unfortunately very scarce and mainly concern the Ba isotopic composition in
a limited sample of stars. They are still under some debate, but suggest that
the relative r-process contribution to Ba has decreased during the galactic
evolution.
(ii) Much excitement has been raised by the observation that the patterns
of abundances of heavy neutron-capture elements between Ba and Pb in r-
process-rich metal-poor stars are remarkably similar to the SoS one. This
claimed ‘convergence’ or ‘universality’ has to be taken with some care, how-
ever, as it largely relies on the assumption that the decomposition between s-
and r-process contributions in metal-poor stars is identical to that in the SoS,
a point that has yet to be demonstrated. A case is made that the convergence
does not tell much about the astrophysics of the r-process, and is largely a
signature of nuclear properties in the 58 < ∼Z < ∼ 76 range. This claim is in
marked contrast with a statement that is often found in the literature. On
the other hand, no ‘universality’ appears to hold for Z > ∼ 76. This concerns
in particular the Pb-peak elements and the actinides. This situation has far-
reaching consequences. In particular, it invalidates the many attempts that
have been made to build detailed galactic chronologies based on the actinides
content of very metal-poor stars. At best, some lower limits to the age can
be derived. No universality exists either for Z < ∼ 58 elements. In this mass
range, some universality appears to be limited at best to stars with a large
Eu overabundance.
(iii) The different behaviours of the abundance patterns of the elements below
and above Ba have been the ground for the hypothesised existence of two so-
called r-process ‘components’. The question of the number of such components
brings us more than forty years back, when it was discussed in the framework
of the newly-constructed canonical r-process model. At a time when one is des-
perately trying to identify a suitable astrophysical site for the development of
an (even limited) r-process [see (4) below], numbering the ‘components’ may
not deserve a very high priority.
(3) The nuclear physics for the r-process. This has been reviewed in some
detail by [24], and is not dealt with in these lectures. Let us just say that much
effort has been put recently in the development of microscopic nuclear models
aiming at reliable predictions of nuclear ground-state properties of thousands
of nuclides located between the valley of nuclear stability and the neutron-
drip line that may be involved in the r-process. Besides some remarkable
achievements in the field, clearly much remains to be done.
338 Marcel Arnould

(4) The astrophysics aspects of the r-process. This is clearly and by far the
most unsatisfactorily understood facet of the r-process modelling, and the one
that calls most desperately for progress. After some fifty years of research on
this subject, the identification of a fully convincing r-process astrophysical site
remains an elusive dream. The attempts conducted thus far may be briefly
summarised as follows:
(i) some simplified site-free r-process models have been devised. They have
the virtue of shedding some light on the conditions that are required in order
for an r-process to develop. Broadly speaking, they can be divided into high-
temperature scenarios and high-density scenarios. The former ones rely on the
assumption that high neutron concentrations and high temperatures are both
mandatory. They include, in order of increasing complexity (or decreasing
number of simplifying assumptions), (a) the canonical waiting-point approxi-
mation model, (b) the multi-event model, and (c) dynamical approaches. The
high-density models rely on the possibility of extensive neutronisation as a re-
sult of endothermic free-electron captures at high enough densities, followed
by the decompression of this highly-neutronised material. A simple steady-
flow model is constructed, and shows that the location and width of the SoS
r-process peaks can be roughly accounted for.
(ii) Inspired by simulations that attempt to account for successful supernova
explosions, (semi-)analytical models have been constructed in order to esti-
mate the properties of the material ablated from a proto-neutron-star follow-
ing the deposition of energy and momentum by the neutrinos streaming out
of it. These models help identifying the physical quantities that have a signif-
icant impact on the nucleosynthesis in the ablated material, and the suitable
ranges of their values for the development of a successful r-process. It appears
that entropy (depending on temperature and density), expansion timescales
(depending on the energy of the ablated material and on its mass-loss rate),
and electron fraction are decisive quantities for the r-process nucleosynthesis,
as illustrated by some network integrations. Recall also that the r-process ef-
ficiency tends to be reduced for increasing neutrino luminosities (which also
affect the electron fraction).
(iii) The conclusions derived from the analytical models referred to above
largely ruin the hope of having a successful r-process developing in the ab-
lated material from a proto-neutron star, at least if one relies on realistic
simulations of prompt or delayed iron core-collapse supernovae. The identi-
fied necessary conditions for the r-process are indeed not met in these models.
True, one may get some hope that this conclusion will be invalidated when
successful explosion simulations will at last be available!
(iv) Some hope might also come from a high-density r-process that could
develop following the decompression of crust material ejected as a result of
neutron-star coalescence. This is made plausible by some yield calculations
based on numerical neutron-star merger simulations. Additional possibilities
could be offered by the outflow of material from the discs forming around the
coalesced neutron-stars. Considering the present status of the simulations of
 Massive Stars Evolution and Nucleosynthesis 339

core-collapse supernovae and of neutron-star coalescence, one may conclude

that the latter scenario offers better potentialities for the r-process. Of course,
uncertainties of different natures affect this scenario. They concern some mod-
elling details and the calculated r-process yields. The limited efficiency of the
scenario at early times in the galactic history based on the assumed low co-
alescence frequency in the young Galaxy is often considered as incompatible
with the observed r-nuclide abundance at very low metallicity. This may well
be the case. One has to acknowledge, however, that the frequency of the coa-
lescence or magnetar events along the galactic history, as well as the amount
of ejected matter per event, remain very uncertain.
(v) Attempts to interpret the observed trends in the galactic r-nuclide abun-
dances have been conducted on grounds of models for the chemical evolution
of the Galaxy. This is a task that is seemingly impossible to complete in
the present state of affairs, given the absence of a reliable identification of
an r-process site, and the schematic nature of the available galactic chemical
evolution models, not to mention various intricacies of observational nature.
(vi) The actinides produced by the r-process enter in particular attempts
to estimate the age of the Galaxy through their present SoS content, or
through their abundances evaluated at the surface of very metal-poor stars.
These attempts face some severe problems related to the nuclear and astro-
physics uncertainties that affect the predictions of the actinides production.
The situation is worsened further by the especially large uncertainties in the
contribution of the r-process to the SoS Pb and Bi content. Concerning the
232
Th - 238U and 235U - 238U pairs classically used to date the Galaxy from
their present meteoritic abundances, it may be worth reiterating an opinion
first expressed more than two decades ago that they have just limited chrono-
metric virtues. The chronometric predictions based on the observations of Th
and U in very metal-poor stars have to be considered with great care as well.
In order for them to be reliable, it is not only required that the production
of the actinides by the r-process is well known, but, and very decisively, that
the production of Th with respect to U and to the Pt peak is ‘universal.’
Observation demonstrates now that this is not the case.
As far as the p-process is concerned, the calculations conducted in ‘realis-
tic’ models for the various proposed sites, and in particular for the O-Ne-rich
layers of SNeII, indicate, roughly speaking, that they may all provide yields
enriched with a suite of p-nuclides at a level that is compatible (within a
factor of 3 or so) with the solar values.
This relative success does not have to hide some puzzling problems raised
by specific cases, the most publicized ones concerning 92,94Mo, 96,98Ru and
138
La. It has been proposed that the underproduction of the p-isotopes of Mo
and Ru in SNII explosions is in fact due to some misrepresentation of the
production in the He-burning core of massive stars of the s-nuclide seeds for
the p-process. The modeling of the s-process in these conditions indeed faces
some nuclear physics and astrophysics uncertainties which are not always ap-
preciated at their true value. An increase of the 22Ne (α , n) 25Mg neutron
340 Marcel Arnould

production rate could help solving the underproduction of the Mo and Ru

p-isotopes, but the validation of this possibility necessitates further nuclear
physics laboratory efforts. Helium-detonating sub-Chandrasehkar-mass CO
white dwarfs could be suitable 92,94Mo and 96,98Ru producers at least if the
seed s-nuclides for the p-process are enhanced with respect to typical solar
abundances in the material experiencing the detonation. As far as 138La is
concerned, SNII explosions could well be suitable sites for the production
of this rare odd-odd nuclide mainly through νe -captures by 138Ba. Exploding
sub-Chandrasekhar-mass CO white dwarfs might also be significant 138La pro-
ducers. The other rare odd-odd nuclide 180Tam has been shown years ago to be
a natural product of the p-process in SNII explosions, a situation that is too
often incorrectly appreciated in the literature. Charged current νe -captures
on pre-existing 180Hf could efficiently complement the thermonuclear 180Tam
production. Some calculations have also suggested that it could be made by
the s-process in AGB stars. This possibility remains controversial at this time.
Another quite puzzling case which does not attract much attention is 113In. It
does not appear to be efficiently produced in the p-process scenarios reviewed
here, and no other source has been identified to-date.
A detailed interpretation of the meteoritic isotopic anomalies involving p-
nuclides still eludes us to a large extent. This concerns in particular the Xe-HL
case. It is also concluded that no reliable p-process chronometry can be built
yet on the p-radionuclides 92Nbg or 146Sm in view of the uncertainties still
affecting their production in each of the considered p-process sites and of the
possible variations of their yields from one site to the other.
At this point, it is of interest to remind of some items the study of which
would help putting the p-process modeling on safer grounds. On the astro-
physics side, let us cite (1) a more reliable estimate of the s-nuclide seeds for
the p-process in massive stars (including e.g. the effect of rotation) and in the
material accreted by white dwarfs with (sub-) Chandrasekhar mass prior to
their explosion. This question relates directly to a good knowledge of the rate
of the key neutron production reaction 22Ne (α , n) 25Mg .
(2) Some progress in the description of convection in advanced stages of the
evolution of massive stars, and in particular in their O-Ne layers. The possi-
bility of survival at the explosion stage of at least a fraction of pre-explosively
produced p-nuclides remains an uncertain issue.
(3) The modeling of the p-process in massive stars with a quite broad range
of metallicities. Yields from such stars are needed to predict the evolution of
the p-nuclide content of galaxies.
(4) The exploration of the impact on the p-process (and on many other ques-
tions) of the multi-dimensional modeling of the pre-SNII and core-collapse
SNII or SNIb/c phases.
(5) The calculation of p-process yields from a variety of Chandrasekhar mass
Type I supernova models of the deflagration or delayed detonation types,
as well as of sub-Chandrasekhar He-detonation models. In the latter case, a
 Massive Stars Evolution and Nucleosynthesis 341

study of the p-nuclide production in the framework of a 3D simulation has

been conducted recently.
On the more nuclear physics side, and apart from the already mentioned
22
Ne (α , n) 25Mg reaction, it is of substantial importance to predict reliably
the rates of thousands of nucleon or α-particle radiative captures and of the
inverse transformations. For a long time to come, almost all these data will
have to be provided by theory, but experiments have to help constraining
and improving the models as much as possible. Efforts have to be started in
order to measure directly the cross sections of photoreactions near threshold.
In addition, more experiments should have to be conducted at sub-Coulomb
energies on radiative proton and α-particle captures.
I hope that this review of the p-process will convince astrophysicsts, nu-
clear physicists and cosmochemists hunting for isotopic anomalies that the
nuclear astrophysics ‘p-nuts’ are quite appealing after all, and can be enjoyed
within a broad interdisciplinary research gathering.

Acknowledgements It is a pleasure to thank very warmly the organisers of

the School for having made my stay in Kodaikanal most enjoyable. I also
express my appreciation to all the participants to the School for their essential
contribution to the very nice atmosphere they have succeeded to create inside,
as well as outside, the lecture room.

References
1. S. E. Woosley et al., Rev. Mod. Phys. 2002, 1015 (2002)
2. C. A. Meakin, D. Arnett, ApJ 667, 448 (2007)
3. D. Arnett et al., preprint [arXiv:0809.1625] (2008)
4. H.-Th. Janka et al., Phys. Rep. 442, 38 (2007)
5. C. E. Rolfs, W. S. Rodney, Cauldrons in the Cosmos (Chicago: Univ.
Chicago Press) (1988)
6. C. Angulo et al., (the NACRE Collaboration), Nucl. Phys. A656, 3
(1999)
7. C. Iliadis et al., ApJS 142, 105 (2002)
8. M. Arnould et al., A&A 347, 572 (1999)
9. J. A. Stoesz, F. Herwig, MNRAS 340, 763 (2003)
10. R. G. Izzard et al., A&A 466, 641 (2007)
11. L. R. Buchmann, C.A. Barnes, Nucl. Phy A777, 254 (2006
12. M. Katsuma, Phys. Rev. C 78, 034606 (2008)
13. G. Imbriani et al., ApJ 558, 903 (2001)
14. T. Spilane et al., in Frontiers in Nuclear Structure, Astrophysics, and
Reactions: FINUSTAR 2, AIP Conf. Proc., Vol. 1012, Eds. P. Demetriou,
R. Julin, & S.V. Harrisopulos (New York: American Institue of Physics),
p. 184 (2008)
15. A. Chieffi et al., ApJ 502, 737 (1998)
16. S. E. Woosley et al., ApJ 175, 731 (1972)
342 Marcel Arnould

17. D.D. Clayton, Principles of Stellar Evolution and Nucleosynthesis

(Chicago: Univ. Chicago Press) (1968)
18. W. D. Arnett, Supernovae and Nucleosynthesis (Princeton: Princeton
Univ. Press) (1996)
19. K. Sumiyoshi et al., ApJ 629, 922 (2005)
20. M. Liebendörfer, et al., ApJ 620, 840 (2005)
21. K. Nomoto, ApJ 322, 206 (1987)
22. F. S. Kitaura et al., A&A 450, 345 (2006)
23. L. Siess, in Stars and Nuclei: A Tribute to Manuel Forestini, ed. by
T. Montmerle & C. Kahane, EAS Pub. Series, (Les Ulis: EDP Sciences)
19, p. 103 (2006)
24. M. Arnould et al., Phys. Rep. 450, 97 (2006)
25. H.-Th.Janka et al., A&A 485,199 (2008)
26. A. Mezzacappa, in Supernovae as Cosmological Lighthouses, ed. by
M. Turetto et al., ASP Conf. Series 342, 175 (2005)
27. A. Burrows et al., ApJ 640, 878 (2006)
28. M. Limongi, A. Chieffi, ApJ 592, 404 (2003)
29. E. M. Burbidge et al., Rev. Mod. Phys. 29, 547 (1957)
30. K. Lodders, ApJ 591, 1220 (2003)
31. M. Asplund et al., ASPC 336, 25 (2005)
32. M. Arnould, S. Goriely, Phys. Rep. 384, 1 (2003)
33. D. D. Clayton et al., Ann. Phys. 12, 331 (1961)
34. F. Käppeler et al., Rep. Prog. Phys. 52, 945 (1989)
35. C. Arlandini et al., ApJ 525,886 (1999)
36. J. Simmerer et al., ApJ 617, 1091 (2004)
37. S. Goriely, A&A 342, 881 (1999)
38. S. Goriely, M. Arnould, A&A 312, 327 (1996)
39. S. Goriely, A&A 327, 845 (1997)
40. H. Palme, H. Beer, in Landolt Börnstein, New Series, Group VI, Astron-
omy and Astrophysics, Subvolume 3a (Berlin: Springer), p. 196 (1993)
41. S. Goriely et al., A&A 444, L1 (2005)
42. G. Cescutti et al., A&A 448, 557 (2006)
43. J. W. Truran et al., PASP 114, 1293 (2002)
44. P. S. Barklem, A&A 439, 129 (2005)
45. L. Mashonkina, G. Zhao, A&A 456, 313 (2006)
46. P. Magain, A&A 297, 686 (1995)
47. D. L. Lambert, C. Allende Prieto, MNRAS 335, 325 (2002)
48. L. Mashonkina et al., A&A 343, 519 (1999)
49. S. Honda et al., ApJ 607, 474 (2004)
50. B. Plez et al., A&A 428, L9 (2004)
51. S. Van Eck et al., A&A 404, 291 (2003)
52. M. Paul et al., ApJ 558, L133 (2001)
53. D. E. Brownlee et al., ASPC 104, 223(1996)
54. J.-P. Meyeret al., ApJ 487, 182 (1997)
55. A. J. Westphal et al., Nat 396, 50 (1998)
56. A. J. Westphal et al., Adv. Space Res., 27, 797 ( 2001)
57. B. A. Weaver, A.J. Westphal, Adv. Space Res. 35, 167 (2005)
58. P. A. Seeger.et al., ApJS 11, 121 (1965)
59. R. D. Hoffman et al., ApJ 482, 951 (1997)
 Massive Stars Evolution and Nucleosynthesis 343

60. B. S. Meyer et al., ApJ 498, 808 (1998)

61. M. D. Delano, A.G.W. Cameron, Astophys. Space Sci. 10, 203 (1971)
62. S. E. Woosley, R.D. Hoffman, ApJ 395, 202 (1992)
63. S. Tsuruta, A. G. W. Cameron, Can. J. Phys., 43, 2056 (1965)
64. S. E. Woosley et al., ApJ 433, 229 (1994)
65. K. Takahashi et al., A&A 286, 857 (1994)
66. Y.-Z. Qian, S.E. Woosley, ApJ 471, 331 (1996)
67. S. Wanajo, Y. Ishimaru, Nucl.Phy. A777, 676c (2006)
68. D. Argast et al., A&A 416, 997 (2004)
69. K. Yokoi et al., A&A 117, 65 (1983)
70. W. A. Fowler, F. Hoyle, Ann. Phys. 10, 280 (1960)
71. J. Audouze, A&A 8, 436 (1970)
72. K. L. Hainebach et al., ApJ 205, 920 (1976)
73. T. G. Harrison, ApJS 36, 199 (1978)
74. J. Audouze, J. W. Truran, ApJ 202, 204 (1975)
75. M. Arnould,PhD thesis, Université Libre de Bruxelles (unpublished)
(1971)
76. M. Arnould, A&A 46, 117 (1976)
77. S. E. Woosley, W. M. Howard, ApJS 36, 285 (1978)
78. M. Rayet et al., A&A 298, 517 (1995)
79. T. Rauscher et al., ApJ 576, 323 (2002)
80. P. Kroup et al., MNRAS 262, 545 (1993)
81. T. Weaver, S.E. Woosley, Phys. Rep. 227, 65 (1993)
82. G.C. Jordan IV, B.S. Meyer, ApJ 617, L131 (2004)
83. C. Fröhlich et al., PRL 96, 142502 (2006)
84. Röpke, et al., in Proc. 13th EPS Conf. ‘Beyond Einstein - Physics of the
21st Century, Eds. A.M. Cruise & L. Ouwehand, ESA-SP 637, p. 161
(2006)
85. I. Jr. Iben, I. Jr.Tutukov, ApJ 370, 615 (1991)
86. W. Hillebrandt, J.C. Niemeyer, ARAA 38, 191 (2000)
87. S. Goriely et al., A&A 383, L27 (2002)
88. S. Goriely et al., A&A 383, L27 (2002)
89. S. Goriely, L. Siess, in From Lithium to Uranium: Elemental tracers of
early cosmic evolution ed. by V. Hill et al., Proc. of IAU symposium Nr
228, (Cambridge: Cambridge University Press), p. 451 (2005)
 Part III

Cosmochemistry and Solar System Abundances

Cosmochemistry

Bruce Fegley, Jr.1 and Laura Schaefer2

1
Planetary Chemistry Laboratory, Department of Earth and Planetary Sciences,
Washington University, St. Louis, MO 63130-4899 [email protected]
2
[email protected]

Keywords: cosmochemistry, condensation calculations, elements, solar

nebula

1 Introduction
Cosmochemistry is the chemistry of the cosmos. This is a broad topic that
ranges from the nucleosynthesis of elements in stars to their chemistry on the
Earth today. In this chapter, we describe chemical equilibrium (or condensa-
tion) calculations of the cosmochemical behavior of the elements.

2 Computational Methods

Chemical equilibrium calculations predict the distribution of each element be-

tween its gaseous, solid, and liquid compounds as a function of temperature,
pressure, and bulk elemental composition. These calculations are often called
condensation calculations because they show the stable phases that condense
out of a cooling gas with solar system elemental abundances. However, chem-
ical equilibrium calculations are path independent because the Gibbs energy
is a state function, i.e., its differential dG is an exact (or perfect) differential.
Thus, the results of chemical equilibrium calculations apply equally well to
heating or cooling of a solar composition system.
The inputs to the chemical equilibrium calculations are the temperature,
pressure, bulk elemental composition, and thermodynamic data for all com-
pounds included in the calculations. The temperatures and pressures used in
the calculations depend on the system being studied, e.g., a protoplanetary
accretion disk, the photospheric region of a cool star, the ejecta from a nova
or supernova, a planetary atmosphere, and so on. The bulk elemental compo-
sition is the set of elemental abundances that are appropriate for the system

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 7, c Springer-Verlag Berlin Heidelberg 2010
348 Fegley, Schaefer

being studied. For example, solar elemental abundances are used for a proto-
planetary accretion disk around a solar metallicity star. On the other hand,
chemical equilibrium calculations for the atmospheres of cool carbon stars use
bulk compositions that are richer in C, N, and s-process elements than solar
abundances [1].
The thermodynamic data for each compound are either the standard Gibbs
energy ΔGo or equilibrium constant Keq for forming the compound from its
constituent elements in their reference states. The reference states are gener-
ally the stable form of an element at the ambient temperature. For example,
the reaction

2H2 (gas) + O2 (gas) = 2H2 O(gas) (1)

forms water vapour from its constituent elements. A second example is the
reaction

4Al(metal) + 3O2 (gas) = 2Al2 O3 (corundum) (2)

This forms Al2 O3 , which is the mineral corundum. The reference state of Al
is metal below its melting point, liquid Al from the melting point (933.6 K) to
the boiling point (2,798 K), and Al gas at higher temperatures. The coexisting
phases at the transition points (i.e., solid and liquid at the melting point or
liquid and gas at the boiling point) are in equilibrium with one another, and
either can be used as the reference state for aluminum.
The standard Gibbs energy for the formation of water vapour, corundum,
or any other compound, from its constituent elements is calculated from tab-
ular data in literature compilations using polynomial equations. The equilib-
rium constant is related to the ΔGo of a reaction by the equation

ΔGo = −RT lnKeq (3)

The thermodynamic data come from a number of literature compilations and
from primary articles in the scientific literature. Fegley and Lodders [2] list
the data sources for a large number of compounds. Lodders [3, 4] published
some updates to this list. The other updates that have been made will be
described in future publications.
The chemical equilibrium calculations are done by sophisticated computer
codes, such as the CONDOR code [2]. This code simultaneously considers
the dual constraints of mass balance and chemical equilibrium. The opera-
tion of the CONDOR code and the general principles of chemical equilibrium
calculations are best illustrated using a simplified version of iron chemistry
in solar composition material. We define the total elemental abundance of
iron as A(Fe). This is the atomic abundance of Fe relative to 106 Si atoms
and is 838,000 Fe atoms [5]. The mole fraction (X) of total iron (ΣF e) in all
Fe-bearing compounds is
 Cosmochemistry 349

A(F e)
XΣF e = (4)
A(H + H2 + He)
The term A(H + H2 + He) is the sum of the H and He abundances with the
temperature-dependent H and H2 equilibrium taken into account. Molecular
H2 , H, and He are the three major gases in solar composition material over
most temperatures and pressures where chemical compounds exist. In the
actual computation, other gases such as H+ , CO, H2 O, N2 , Ne, other ions,
etc., are also included in the denominator; however, these are neglected in the
description here for clarity. Multiplying XΣF e by the total pressure Ptot gives
the partial pressure sum for iron

PΣF e = XΣF e PT = PF e + PF eS + PF eH + PF eO + .. (5)

Equation (5) is a mass balance equation that counts the total number of
iron atoms in all Fe-bearing gases and equates this sum to the total amount
of iron. The partial pressures of Fe-bearing gases with more than one iron
atom are multiplied by the number of iron atoms in the gas to insure that all
atoms are counted. In this example the Fe-bearing gases included in the mass
balance equation are the most abundant ones at low pressures characteristic
of cool stellar atmospheres, protoplanetary disks, and ejecta from nova and
supernova. All of these Fe-bearing gases contain one Fe atom and their partial
pressures are multiplied by unity. If any gases containing two Fe atoms, such
as Fe2 Cl4 , were present, their partial pressures would be multiplied by two.
We must now relate the partial pressures of the different Fe-bearing gases in
the mass balance equation to their chemical equilibrium abundances.
This is done by rewriting equation (5) in terms of the thermodynamic ac-
tivity of Fe (aF e ), the equilibrium constants (Ki ) for forming the Fe-bearing
gases from the constituent elements in their reference states, and the thermo-
dynamic activities and fugacities of all other elements combined with iron in
the gases.

1/2 1/2 1/2

PΣF e = XΣF e PT = aF e [KF e + KF eS f S2 + KF eH fH2 + KF eO fO2 ] (6)

The actual mass balance sum for iron in the CONDOR code includes about
21 Fe-bearing gases. The most important ones in a low-pressure, solar com-
position gas are Fe, FeS, FeH, and FeO. Analogous forms of equation (6) are
written for each element in the code. The ai and fi terms in equation (6) are
the elemental activities and fugacities of the respective elements, e.g., fH2 is
the fugacity of hydrogen.
Equation (6) combines the mass balance and chemical equilibrium con-
straints for iron. It also shows that the chemistry of iron is coupled to that
of other elements because the fugacities of sulfur, hydrogen, and oxygen are
included in equation (6). In general, the chemistry of all the elements is cou-
pled, and the mass balance equations form a set of coupled, nonlinear equa-
tions that are solved iteratively. An initial guess is assumed for the activity
350 Fegley, Schaefer

or fugacity of each element. These guesses can be optimized if the major gas
for each element is known, but this is not necessary for the code to operate
properly. The CONDOR code solves the set of mass balance equations and
returns the thermodynamic activity or fugacity for each element, the abun-
dances of all gases (molecules, radicals, atoms, and ions) included in the code,
and information on the quality of the calculated results for each element. The
convergence criterion requires that the calculated and input abundances for
each element agree to within 1 part in 1015 . However, for all practical purposes
a satisfactory solution is reached when the calculated and input abundances
for each element agree to within 1 part in 10,000 (0.01%). At present the
CONDOR code considers over 3,600 solid, liquid, and gaseous compounds for
all naturally occurring elements in the periodic table.
The CONDOR code also takes possible condensates (both solid and liquid)
into account. For example, Fe metal (or liquid Fe depending on the temper-
ature) forms if the thermodynamic activity of Fe (aF e ) is equal to or greater
than unity. The code computes the temperature at which aF e reaches unity
(the Fe metal condensation temperature), resets aF e to unity at all tempera-
tures below this point, and adds a new term for the abundance of Fe metal
to the mass balance equation.
The stabilities of condensates containing two or more elements are com-
puted considering compound formation from the elements in their respective
reference states.
For example, the reaction

F e(metal) + S(ref state) = F eS(troilite) (7)

is used for troilite, which is the most common Fe sulfide in meteorites. The
reference state of sulfur is solid orthorhombic sulfur, solid monoclinic sulfur,
liquid sulfur, or S2 gas depending on the temperature. As temperature de-
creases, reaction (7) shifts to the right and the thermodynamic activity of
troilite increases. Troilite condenses when its thermodynamic activity reaches
unity and this is calculated from

aF eS = aF e as KF eS (8)
where KF eS is the temperature-dependent equilibrium constant for troilite
formation from its constituent elements, and aF e and aS are taken from the
gas-phase equilibrium calculations described above. In general, a pure phase
such as Fe-metal, corundum (Al2 O3 ), or troilite (FeS) starts to condense from
the gas at the temperature where its thermodynamic activity reaches unity.
The thermodynamic activity is less than unity at higher temperatures where
the condensate phase is unstable, and is fixed at unity at all temperatures
where the condensate is stable.
Once troilite is stable, the fraction of Fe (αF e ) condensed in troilite is
calculated and the gas-phase abundance of total iron (PF e ) is reduced by
 Cosmochemistry 351

multiplying by (1 - αF e ). Analogous corrections are made for all elements dis-

tributed between the gas and condensates. The gas-phase and gas-solid chemi-
cal equilibria are coupled and solved simultaneously using iterative techniques.
Palme and Fegley [6] illustrate this procedure for condensation of the minerals
enstatite MgSiO3 and forsterite Mg2 SiO4 , which involve coupled equilibria for
the elements Mg, O, and Si.
The total abundance of each condensate is limited by the abundance of the
least abundant element in the condensate. For example, the mineral schreiber-
site Fe3 P forms by reaction of P-bearing gases with Fe metal at about 1300
K and 10−4 bar total pressure. Phosphorus has an atomic abundance of 8,373
atoms, which is about 1% of the atomic abundance of iron. There are 3 Fe
atoms in each molecule of schreibersite. Thus Fe3 P formation consumes only
3% of the total iron abundance, while removing all phosphorus from the gas
phase. Likewise, troilite formation removes all sulfur from the gas because
its abundance is only 53% of that of iron, while unreacted Fe metal remains
present at lower temperatures until it is consumed by formation of Fe-bearing
oxides and silicates.
Several points are worth emphasizing. The first point is mass balance. The
total amount of each element is conserved in the chemical equilibrium calcu-
lations. Thus the abundances of all gases and all condensed phases (solids
and/or liquids) sum to the total elemental abundance - no less and no more.
The second point is that chemical equilibrium is completely independent of
the size, shape, and state of aggregation of condensed phases - a point demon-
strated by Willard Gibbs over 130 years ago. Finally, the third point is that
chemical equilibrium is path independent. Thus, the results of chemical equi-
librium calculations are independent of any particular reaction. A particular
chemical reaction does not need to be specified because all possible reactions
give the same result at chemical equilibrium. This is completely different than
chemical kinetic models where the results of the model are critically dependent
on the reactions that are included. However, a chemical equilibrium calcula-
tion does not depend on kinetics, is independent of kinetics, and does not
need a particular list of reactions. This point may seem obvious, but is often
misunderstood.

3 Cosmochemical Behaviour of the Elements

This section summarizes the results of chemical equilibrium calculations for
naturally occurring elements. We review the cosmochemical behavior of the
elements in order of increasing volatility, i.e., decreasing condensation tem-
perature using the cosmochemical classification scheme.

3.1 Refractory Elements

Refractory elements are the first elements to condense (or the last ones
to evaporate) from solar composition material. The refractory elements are
352 Fegley, Schaefer

subdivided into refractory siderophiles (elements found in metal) and refrac-

tory lithophiles (elements found as oxides and/or silicates). The refractory
siderophiles are the noble or platinum group metals (Ru, Os, Rh, Ir, and Pt,
but not Pd), molybdenum Mo, tungsten W, and rhenium Re. All of the re-
fractory siderophiles are trace elements with low abundances. In the 1970s
micron-size metal nuggets composed of the refractory siderophile elements
were discovered inside Ca, Al-rich inclusions (CAIs) in the Allende meteorite.
This is a stony meteorite, which is a type CV3 carbonaceous chondrite. The
compositions of the refractory metal nuggets in Allende and other carbona-
ceous chondrites are reproduced by chemical equilibrium calculations [7, 8].
The actinides (U, Th, Pu), alkaline earths (Be, Mg, Ca, Sr, Ba), lan-
thanides (elements La - Lu), Al, and the elements in groups 3b (Sc, Y), 4b
(Ti, Zr, Hf), and 5b (V, Nb, Ta) of the periodic table are refractory lithophile
elements. The refractory lithophiles are 5% of the total mass of the rock in so-
lar composition material. Aluminum Al, calcium Ca, and titanium Ti are the
three most abundant refractory lithophiles, and they form minerals that are
the host phases for most of the less abundant refractory lithophile elements
such as the actinides, lanthanides, and transition elements in group 5b of the
periodic table. Some of the less abundant refractory lithophiles - the group
4b elements Zr, Hf, and the group 3b elements Y and Sc - condense as oxides
before any Ca, Al, Ti-bearing minerals form [9]. But the rest condense into
the more abundant host phases.
Figure 1 shows the percentage distribution of aluminum between its major
gases and the condensed phases that are stable at chemical equilibrium in a
solar composition gas as a function of temperature at a total pressure of 10−4
bar. This pressure is representative of that in the inner regions of protoplan-
etary accretion disks (such as the solar nebula) and photospheric regions of
cool stars.
At the high temperatures that are shown on the right hand side of Figure
1, all Al is in the gas phase, mainly as monatomic Al and aluminum hydroxide
AlOH gases. For example at 2000 K, about 92% of all aluminum is in Al gas,
about 7.5% is in AlOH gas, and the remaining 0.5% is in other Al-bearing
gases. Corundum Al2 O3 condenses at 1640 K and is the first Al-bearing min-
eral (see the small wedge at the top right of Figure 1). However, Figure 1
shows that corundum is only stable for a few degrees (the wedge disappears).
Grossite CaAl4 O7 and hibonite CaAl12 O19 form by reaction of corundum with
the surrounding gas. These reactions destroy all the corundum and produce
grossite and hibonite. At 1600 K, most of the Al is in these minerals. The
remainder is mainly in gaseous AlOH and Al. In fact, Figure 2 shows that
other Al-bearing gases are still present but that these are less abundant than
Al or AlOH. In turn, other reactions form the mineral gehlenite CaAl2 SiO7
from grossite and hibonite with decreasing temperature. The rest of Figure
1 shows the appearance, disappearance, and abundance of other Al-bearing
condensates formed at chemical equilibrium at lower temperatures.
 Cosmochemistry 353

100
MgAI2O4
KAISi3O8 AIOH (g)

AI2O3
KMg3AISi6O10(OH)2 other
gases
80 NagAI6Si6O24Cl2

Ca2AI2SiO7
60
% of total Al

NaAISi3O8
CaAI2Si2O8

40

Na2Mg3Al2Si8O22(OH)2
20 AI (g)
AI2O3

grossite
+
hibonite

0
500 1000 1500 2000
Temperature (K)
Fig. 1. Percentage distribution of aluminum between different phases in a solar com-
position gas as a function of temperature for a total pressure of 10−4 bar. Gaseous
species are indicated with (g), all other species are condensed phases

Figure 2 shows the mole fractions of the major Al-bearing gases that are
stable at chemical equilibrium in a solar composition gas as a function of
temperature at a total pressure of 10−4 bar. The temperature range in Figure 2
is only from 1200 - 2000 K because the mole fractions of all the Al-bearing
gases decrease strongly with decreasing temperature below the condensation
temperature of corundum. Figures 1 and 2 are complementary to one another.
Figure 1 gives the percentage distribution relative to total aluminum while
Figure 2 gives the mole fractions, defined as the partial pressure of a gas
relative to the total pressure. For example, the mole fraction of AlOH gas is
PAlOH
XAlOH = (9)
PT
354 Fegley, Schaefer

Fig. 2. Gas phase chemistry of aluminum in a solar composition gas as a function

of temperature at a total pressure of 10−4 bar. The abundances of the gases are
shown as mole fractions

This definition of mole fraction is identical to saying that the mole frac-
tion is equal to the number of moles of a gas relative to the moles of all gases,
or equivalently that the mole fraction of a gas is equal to its volume frac-
tion. Figure 2, and similar figures for other elements, show gases that are not
abundant enough to appear on a percentage distribution graph such as Figure
1. This information is important for astronomical observations of the photo-
spheric regions of cool stars, of protoplanetary accretion disks, of circumstellar
envelopes, and of the atmospheres of gas giant planets.
Table 1 lists the names and formulas of the minerals shown in Figure 1
and subsequent figures. For example, Table 1 shows that the mineral grossite
has the formula CaAl4 O7 and that MgAl2 O4 is the mineral spinel. Table 2
 Cosmochemistry 355

Table 1. Minerals included in Figures

Mineral name Formula

Alabandite MnS
Albite NaAlSi3 O8
Anorthite CaAl2 Si2 O8
Bornite Cu5 FeS4
Chlorapatite Ca5 (PO4 )3 Cl
Chromite FeCr2 O4
(Cobalt Sulfide)a Co9 S8
Corundum Al2 O3
Diopside CaMgSi2 O6
Enstatite MgSiO3
Eskolaite Cr2 O3
Fayalite Fe2 SiO4
Fluorapatite Ca5 (PO4 )3 F
Forsterite Mg2 SiO4
Gehlenite Ca2 Al2 SiO7
Glaucophane Na2 Mg3 Al2 Si8 O22 (OH)2
Grossite CaAl4 O7
Halite NaCl
Heazlewoodite Ni3 S2
Hibonite CaAl12 O19
Hydroxyapatite Ca5 (PO4 )3 OH
Ilmenite FeTiO3
Iron metal Fe
(Iron Selenide)a FeSe0.961
Karelianite V 2 O3
Orthoclase KAlSi3 O8
Perovskite CaTiO3
Phlogopite KMg3 AlSi3 O10 (OH)2
Picrochromite MgCr2 O4
Rutile TiO2
Schreibersite Fe3 P
Sodalite 3NaAlSiO4 ·NaCl
(Sodiam Bromide)a NaBr
Spinel MgAl2 O4
Talc Mg3 Si4 O10 (OH)2
Tephroite Mn2 SiO4
(Titanium Oxide)a Ti4 O7
Tremolite Ca2 Mg5 Si8 O22 (OH)2
Troilite FeS
Zincochromite ZnCr2 O4
a
No known mineral name
356 Fegley, Schaefer

Table 2. Abundances and Condensation Temperatures of the Elements in the Solar

Nebula
Element Symbol abundance Tc (K) Initial 50% Major gases
Condensate Tc (K)
{Dissolving
Species}

1 Hydrogen H 2.431×1010 182 H2 O ice - H2

2 Helium He 2.343×109 <3 He ice - He
3 Oxygen O 1.413×107 182 H2 O icea 180 CO,H2 O
4 Carbon C 7.079×106 78 CH4 · 7H2 O 40 CO, CH4
5 Neon Ne 2.148×106 9.3 Ne ice 9.1 Ne
6 Nitrogen N 1.950×106 131 NH3 ·H2 O 123 N2 , NH3
7 Magnesium Mg 1.020×106 1390 Spinel - Mg
1354 Forsteriteb 1336
8 Silicon Si ≡1.00×106 1529 Gehlenite - SiO, SiS
1290 Enstatite 1265
9 Iron Fe 8.380×105 1358 Fe metalc 1325 Fe
10 Sulfur S 4.449×105 704 FeS 655 H2 S, HS
11 Argon Ar 9.032×104 48 Ar.6H2 O 47 Ar
12 Aluminum Al 8.410×104 1640 Al2 O3 1653 Al, AlOH, Al2 O, AlS,
1620 CaAl12 O19 1615 AlH, AlO
13 Calcium Ca 6.287×104 1620 CaAl12 O19 - Ca
1529 Gehlenite 1500
14 Sodium Na 5.751×104 970 {NaAlSi3 O8 } 960 Na,NaCl
15 Nickel Ni 4.780×104 1280 {Ni} 1260 Ni
16 Chromium Cr 1.286×104 870 Cr2 O3 Cr
17 Manganese Mn 9168 1050 MnS - Mn
18 Phosphorus P 8373 1285 Fe3 P 1265 PO,P,PN,PS
19 Chlorine Cl 5237 650 Ca5 (PO4 )3 Cl - HCl,NaCl,KCl
- Na4 [Al3 Si3 O12 ]Cl 400
20 Potassium K 3692 950 {KAlSi3 O8 } 915 K,KCl,KOH
21 Titanium Ti 2422 1593 CaTiO3 1575 TiO,TiO2
22 Cobalt Co 2323 1190 {Co} 1170 Co
23 Zinc Zn 1226 660 ZnCr2 O4 650 Zn
24 Fluorine F 841.1 710 Ca5 [PO4 ]3 F 710 HF
25 Copper Cu 527.0 960 {Cu} 940 Cu
26 Vanadium V 288.4 - {VO,V2 O3 } 1250 VO2 , VO
27 Selenium Se 65.79 - ZnSe 520 H2 Se,GeSe
28 Krpton Kr 55.15 53 Kr·6H2 O 52 Kr
29 Bromine Br 11.32 370 NaBr 360 Br,HBr
a
22.75% of oxygen is condensed into rock before water ice condensation. b Major
condensed reservoir of element. c Condensation temperature of pure iron metal.
 Cosmochemistry 357

lists the elements discussed in this chapter in decreasing order of abundance.

Table 2 also gives the major gases and initial condensates of these elements.
For example, Table 2 shows that Al is the twelfth most abundant element in
solar composition material with an abundance of 8.41×104 Al atoms (per 106
Si atoms), its initial condensates are corundum Al2 O3 at 1640 K and hibonite
CaAl12 O19 at 1620 K, and its major gases are Al and AlOH.

Fig. 3. Percentage distribution of calcium between different phases in a solar com-

position gas as a function of temperature for a total pressure of 10−4 bar. Gaseous
species are indicated with (g), all other species are condensed phases. Hydroxyap-
atite (Ca5 (PO4 )3 OH) is not shown because it occurs only at 300 K (and lower) with
an abundance of 15.5% of total Ca

Figures 3 and 4 for Ca and Ti, respectively, give similar information for the
major gases and condensates of these two refractory lithophiles. For example,
358 Fegley, Schaefer

Fig. 4. Percentage distribution of titanium between different phases in a solar com-

position gas as a function of temperature for a total pressure of 10−4 bar. Gaseous
species are indicated with (g), all other species are condensed phases

monatomic Ca gas is the major Ca-bearing gas and hibonite CaAl12 O19 is
the first Ca-bearing condensate. Likewise, titanium monoxide TiO, TiO2 , and
monatomic Ti are the major Ti-bearing gases and perovskite CaTiO3 is the
first Ti-bearing condensate.
The results of the chemical equilibrium calculations for the refractory
lithophiles are also confirmed by the mineralogy and chemistry of the Ca, Al-
rich inclusions in Allende and other meteorites. The major minerals in CAIs
are the same ones predicted by the calculations, namely melilite (a solid solu-
tion of gehlenite CaAl2 SiO7 and åkermanite Ca2 MgSi2 O7 ), spinel, corundum,
grossite, hibonite, and perovskite. Chemical analyses of CAIs show that the
refractory lithophiles are enriched by an average of 20 times solar elemental
 Cosmochemistry 359

abundances, as expected for complete condensation of elements that constitute

5% of rocky material. The less abundant refractory lithophiles, such as the lan-
thanides (or rare earth elements REE) are found dissolved in perovskite and
hibonite, as predicted by chemical equilibrium calculations. Kornacki and Feg-
ley [9, 10] describe chemical equilibrium calculations for refractory lithophiles
and their abundances in CAIs.

Fig. 5. Percentage distribution of iron between different phases in a solar com-

position gas as a function of temperature for a total pressure of 10−4 bar. Gaseous
species are indicated with (g), all other species are condensed phases
360 Fegley, Schaefer

3.2 Major elements

Iron Fe, magnesium Mg, and silicon Si are the next elements to condense and
they constitute the major elements because they make up most of the rock
in solar composition material. Figure 5 shows the percentage distribution of
iron between its major gases and condensates. Iron metal condenses at 1357
K at 10−4 bar total pressure and is 50% condensed by 1325 K. The iron metal
is not pure Fe, but is an Fe-rich metal alloy containing smaller amounts of
nickel Ni, cobalt Co, and other siderophile elements. Iron metal is the only
Fe-bearing condensate until schreibersite Fe3 P forms (1285 K, 10−4 bar). As
mentioned earlier, phosphorus is about 1% as abundant as Fe, so formation of
Fe3 P consumes only 3% of the total iron metal. The formation of troilite FeS
at 704 K consumes about 55% of the remaining Fe metal. Hydrogen sulfide
is the major S-bearing gas containing about 100% of all sulfur and troilite
formation occurs via the net thermochemical reaction

H2 S(gas) + F e(metal) = F eS(troilite) + H2 (gas) (10)

This reaction has the same number of gas molecules on either side of the
reaction and the FeS formation temperature of 704 K is thus independent of
total pressure.
Chromite FeCr2 O4 formation at 660 K consumes a tiny fraction of Fe
metal, but the remainder of the metal remains present until it is oxidized
to Fe2+ -bearing silicates such as fayalite Fe2 SiO4 , shown in Figure 5. This
dissolves in solid solution with forsterite Mg2 SiO4 forming olivine.
Figures 6 and 7 show the equilibrium chemistry of magnesium and silicon.
Small amounts of magnesium and silicon condense at high temperatures as
spinel MgAl2 O4 (1390 K, shown in Figure 1), gehlenite Ca2 Al2 SiO7 (1529 K),
and åkermanite Ca2 MgSi2 O7 (not shown). However the amounts of Mg and
Si in these high temperature condensates are limited by Ca and Al, which
are much less abundant than magnesium or silicon. Most Mg and Si condense
as forsterite Mg2 SiO4 (1354 K) and enstatite MgSiO3 (1290 K). Forsterite
condenses from the gas phase while enstatite forms by reaction of nebular gas
with preexisting forsterite grains.
Figure 8 shows that at pressures greater than about 10−4 bar, Fe metal
condenses at higher temperatures than forsterite Mg2 SiO4 . Conversely, at
pressures less than about 10−4 bar, forsterite condenses at higher tempera-
tures than Fe metal. The separation between the metal and forsterite con-
densation temperatures increases with increasing pressure from the crossover
point. The planet Mercury is richer in Fe and poorer in silicates than ex-
pected from solar composition. This may indicate that the materials accreted
by Mercury formed at pressures higher than the 10−4 bar crossover point for
the iron and forsterite condensation curves.
 Cosmochemistry 361

Fig. 6. Percentage distribution of magnesium between different phases in a so-

lar composition gas as a function of temperature for a total pressure of 10−4 bar.
Gaseous species are indicated with (g), all other species are condensed phases. Spinel
(MgAl2 O4 ) and talc (Mg3 Si4 O10 (OH)2 ) are not shown due to their low abundance
(1.4% and 0.9% total Mg at 300 K, respectively)

3.3 Moderately Volatile Elements

The moderately volatile elements are the next elements to condense from
cooling solar composition material. The moderately volatile elements are a
diverse set of elements including the alkali elements (Li, Na, K, Rb, Cs);
transition metals (Cr, Co, Mn, Ni, Pd, Cu, Ag, Au, Zn); group 3 elements (B
and Ga); group 4 elements (Ge and Sn); group 5 elements (P, As, Sb); group 6
elements (S, Se, Te); and two of the halogens (F and Cl). Moderately volatile
elements have condensation temperatures between those of the major elements
(Fe, Mg, Si) and troilite FeS. The moderately volatile elements with higher
362 Fegley, Schaefer

Fig. 7. Percentage distribution of silicon between different phases in a solar com-

position gas as a function of temperature for a total pressure of 10−4 bar. The
abundances of the phases are shown as a cumulative percentage of the element.
Gaseous species are indicated with (g), all other species are condensed phases. Not
shown due to low abundance: KMg3 AlSi3 O10 (OH)2 (1.1%), Mn2 SiO4 (0.46%), and
Na8 Al6 Si6 O24 Cl2 (1.6%)

abundances condense as pure phases. For example, Figures 5 and 9 show that
sulfur condenses as troilite FeS at 704 K. As mentioned above, H2 S is the
dominant S-bearing gas at the troilite condensation temperature. Figure 9
shows that S, SH, and SiS are important and that PS and S2 are minor S-
bearing gases at higher temperatures. However, these gases are highly reactive
and cannot survive at lower temperatures. For example, the condensation of
forsterite Mg2 SiO4 and enstatite MgSiO3 removes all silicon, including SiS,
from the gas. Condensation of schreibersite Fe3 P removes PS and all other
 Cosmochemistry 363

Fig. 8. Condensation temperatures of Fe metal, Mg2 SiO4 forsterite, and MgSiO3

enstatite in a solar composition gas as a function of total pressure

P-bearing gases. Finally, monatomic S, SH, and S2 convert into H2 S with

decreasing temperature.
Figures 5 and 10 show that phosphorus condenses as schreibersite Fe3 P
at 1285 K. Schreibersite reacts with nebular gas at lower temperatures and
forms fluorapatite Ca5 (PO4 )3 F at 710 K and whitlockite Ca3 (PO4 )2 at 680
K. The whitlockite reacts with HCl in the nebular gas to form chlorapatite
Ca5 (PO4 )3 Cl at 470 K. Chondrites contain all these phosphate minerals.
Figures 11 and 12 show the equilibrium chemistry of fluorine and chlorine,
which illustrate an interesting mass balance example. Fluorine is the least
abundant element in fluorapatite and its abundance limits that of fluorap-
atite. The 3:1 ratio of P to F atoms in fluorapatite consumes about 30% of
total phosphorus, leaving 70% of total phosphorus for chlorapatite formation.
364 Fegley, Schaefer

Fig. 9. Percentage distribution of sulfur between different phases in a solar com-

position gas as a function of temperature for a total pressure of 10−4 bar. Gaseous
species are indicated with (g), all other species are condensed phases

However, this is not enough to condense all chlorine, which requires a 3:1 ratio
of P to Cl atoms to condense all chlorine in chlorapatite. In fact, the total
phosphorus abundance is too small to condense all Cl in chlorapatite. The
residual chlorine, which is about 60% of total Cl, condenses as halite NaCl
(420 K). However, this forms sodalite Na4 (AlSiO4 )3 Cl (405 K, and again at
315 K), which decomposes back to halite plus nepheline (365 K), over the
405 - 315 K range. Both minerals are found in chondrites. Halite is probably
present in many chondrites that contain water-soluble chlorine, even though
it is not reported as one of the observed minerals.
Sodium and potassium condense initially as the feldspar minerals albite
NaAlSi3 O8 and orthoclase KAlSi3 O8 . Some of the albite reacts to form halite
 Cosmochemistry 365

Fig. 10. Percentage distribution of phosphorus between different phases in a so-

lar composition gas as a function of temperature for a total pressure of 10−4 bar.
Gaseous species are indicated with (g), all other species are condensed phases. Hy-
droxyapatite (Ca5 (PO4 )3 OH) has an abundance of 70% of total P, but occurs only
at 300 K (and lower), and is therefore not shown

at 420 K. Formation of glaucophane, a hydrous mineral called an amphibole,

consumes the remaining albite at 360 K. Likewise, formation of phlogopite,
a hydrous mineral called a mica, consumes all orthoclase at 470 K. Neither
glaucophane nor phlogopite occurs in meteorites. Thus, although they become
thermodynamically stable at low temperatures, they may not form because
of slow chemical reaction rates.
Most of the moderately volatile elements have low abundances and con-
dense in solid solutions in more abundant minerals. For example, selenium Se
may dissolve in FeS as FeSe. Copper, Ag, and Au may dissolve in Fe alloy.
366 Fegley, Schaefer

Fig. 11. Percentage distribution of fluorine between different phases in a solar com-
position gas as a function of temperature for a total pressure of 10−4 bar. Gaseous
species are indicated with (g), all other species are condensed phases

Lithium, Mn, and Zn condense into forsterite and enstatite, e.g. as LiSiO3 ,
Mn2 SiO4 , MnSiO3 , Zn2 SiO4 , and ZnSiO3 . Boron and gallium substitute for
Al in feldspar and Rb and Cs substitute for K in orthoclase. Some elements
such as gallium are found in metal and silicates in meteorites, and gallium
may dissolve in both phases during condensation.

3.4 Highly Volatile Elements

The highly volatile elements (Bi, Br, Cd, Hg, I, In, Pb, Tl) condense at temper-
atures between troilite and water ice. In general, the equilibrium chemistry of
these elements is poorly known because thermodynamic data are unavailable
 Cosmochemistry 367

Fig. 12. Percentage distribution of chlorine between different phases in a solar com-
position gas as a function of temperature for a total pressure of 10−4 bar. Gaseous
species are indicated with (g), all other species are condensed phases

for some pure phases and solid solutions. Most of the highly volatile elements
condense in solid solution in more abundant minerals. For example, Lodders
[5] calculated that mercury condenses as HgS, HgSe, and HgTe dissolved in
solid solution in troilite FeS. Fifty percent of all Hg is condensed at 252 K.
Her result predicts that the CI carbonaceous chondrites should contain the
solar abundance of mercury. At present, the Hg solar abundance is estimated
from r-process and s-process systematics of the elements. Bismuth Bi, and
lead Pb condense in solid solution in iron metal alloy, while cadmium, In, and
Tl condense in solid solution in troilite FeS. Bromine and iodine condense
368 Fegley, Schaefer

in solid solution in the apatite minerals. Lodders [5] gives a good discussion
of condensation chemistry of the moderately and highly volatile elements in
solar composition material.

3.5 Atmophile Elements

The atmophile elements (H, C, N, noble gases) condense at and below the con-
densation temperature of water (liquid or ice). Liquid water or water ice con-
denses at the temperature where the partial pressure of water vapour equals
the vapour pressure over liquid water or water ice. The condensation curve
for water (liquid or ice) is
10, 000
= 38.84 − 3.83log10PT (11)
Tc (H2 O)

Thus, water ice starts to condense at 185 K at 10−4 bar total pressure in
solar composition material. Fifty percent of all water condenses by 180 K.
Liquid water condenses instead of water ice at total pressures of 3.8 bar and
above. Condensation of liquid water has important consequences such as the
formation of aqueous NH3 solutions. Aqueous NH3 solutions are stable down
to 173 K at one bar total pressure, which is the eutectic point, i.e., the lowest
melting point, in the NH3 - H2 O phase diagram [11].
The partial pressure of water vapour in solar composition material depends
on several factors. Formation of anhydrous minerals consumes about 23%
of total oxygen. The formation of hydrous minerals, such as glaucophane,
tremolite, serpentine, and talc consumes more water vapour. For example,
serpentine and brucite Mg(OH)2 form by the hydration of forsterite Mg2 SiO4
grains by water vapour

2M g2 SiO4 (f orsterite) + 3H2 O(g) =

M g3 Si2 O5 (OH)4 (serpentine) + M g(OH)2 (brucite) (12)

An analogous reaction is formation of talc and brucite by hydration of en-

statite grains

4M gSiO3 (enstatite) + 2H2 O(g) =

M g3 Si4 O10 (OH)2 (talc) + M g(OH)2 (brucite) (13)

These hydration reactions provide hydrous minerals at low temperatures in

solar composition material (∼ 250 K for serpentine and ∼ 300 K for talc).
By mass, serpentine contains 13.0% water, talc 4.8% water, and brucite 3.1%
water. Theoretical models of gas - grain kinetics predict that hydration of
forsterite and enstatite to serpentine and talc are too slow to proceed over
the lifetime of the solar nebula [12]. These predictions are consistent with
 Cosmochemistry 369

Fig. 13. Percentage distribution of oxygen between different phases in a solar com-
position gas as a function of temperature for a total pressure of 10−4 bar. Gaseous
species are indicated with (g), all other species are condensed phases. The minor
solids include: Al2 O3 (0.28%), Ca2 Mg5 Si8 O22 (OH)2 (4.15%), Ca5 (PO4 )3 (F,Cl,OH)
(0.25%), Ca3 (PO4 )2 (0.17%), CaTiO3 (0.05%), Cr2 O3 (0.14%), FeCr2 O4
(0.15%), FeTiO3 (0.05%), KAlSi3 O8 (0.21%), KMg3 AlSi3 O10 (OH)2 (0.31%),
MgAl2 O4 (0.40%), MgCr2 O4 (0.18%), Mg3 Si4 O10 (OH)2 (0.27%), Mn2 SiO4 (0.13%),
Na8 Al6 Si6 O24 Cl2 (0.44%), TiO2 (0.03%), Ti4 O7 (0.03%), VO (0.002%), V2 O3
(0.003%), ZnCr2 O4 (0.03%)
370 Fegley, Schaefer

petrological studies of meteorites containing hydrous silicates, which conclude

that hydrous minerals formed on the meteorite parent bodies.
The partitioning of carbon between CO and CH4 also affects the water
vapour partial pressure. As discussed below, CO is the major C-bearing gas
at high temperatures and low pressures while CH4 is the major C-bearing gas
at low temperatures and high pressures. Because the bulk C/O ratio is 0.5 in
solar composition material, the water vapour abundance increases by about a
factor of two when CO converts to methane. All these factors are summarized
in Figure 13, which shows the percentage distribution of oxygen between the
major O-bearing gases and minerals. At complete chemical equilibrium, the
anhydrous and hydrous minerals consume about 26% of all oxygen and about
74% is left to condense as water or water ice.
The condensation chemistry of carbon and nitrogen depends on the major
C- and N-bearing gases at low temperatures in solar composition material.
At 10−4 bar total pressure, CH4 and NH3 are the predominant C- and N-
bearing gases at temperatures below the water ice (or liquid water) conden-
sation curve. Ammonia condenses as ammonia monohydrate NH3 ·H2 O at 131
K, 10−4 bar total pressure via the net reaction

N H3 (g) + H2 O(ice) = N H3 · H2 O(ice) (14)

Ammonia monohydrate (also called ammonium hydroxide NH4 OH) is a dis-
tinct compound and is not a clathrate hydrate. Ammonia ice and/or liquid
NH3 could form in the absence of liquid water. This happens in Jupiter’s
atmosphere where NH3 ice clouds form because gravitational sedimentation
keeps liquid water droplets in a cloud layer hundreds of kilometers below the
NH3 ice clouds. Thus, no water is present in the upper atmosphere where the
NH3 ice clouds form. Liquid NH3 does not condense in the solar nebula or
on the gas giant planets in our solar system (Jupiter, Saturn, Uranus, and
Neptune) because the total pressure and/or NH3 gas abundance is too low.
The NH3 gas pressure would have to be at least as high as the triple point
pressure (0.06 bar at 195.5 K) for liquid NH3 condensation. Assuming that
all nitrogen is present as NH3 , the total pressure at low temperatures would
have to be about 444 bar for liquid NH3 to form (i.e., to reach 0.06 bar NH3
pressure at the triple point) in solar composition material. Such high pressures
are unreasonable for the solar nebula but they may occur in extrasolar gas
giant planets or in the interiors of icy planets and satellites.
Methane condenses at lower temperatures as methane clathrate hydrate
(CH4 ·7H2 O) at 78 K and CH4 ice. A clathrate hydrate is a cage compound
in which a gas molecule is trapped inside the water ice crystal lattice. One
gas atom is trapped for every seven water molecules. Clathrate hydrates form
via reactions exemplified by
CH4 (gas) + 7H2 O(ice) = CH4 · 7H2 O(clathrate hydrate) (15)
However, there is not enough water ice to condense all CH4 as a clathrate
hydrate because the CH4 to H2 O ratio is 1:7 in the clathrate hydrate while the
 Cosmochemistry 371

solar C/O ratio is 0.5. Thus, the residual CH4 condenses as CH4 ice at lower
temperatures. Hydrogen and helium do not condense because temperatures
never get low enough for solid hydrogen or liquid helium to form. Neon ice
forms at about 9 K, but it is doubtful if temperatures ever get this low in
any astronomical environment where condensation occurs. The other noble
gases (Ar, Kr, Xe) either condense as ices (solid Ar, Kr, Xe) or form clathrate
hydrates (Ar 48 K, Kr 53 K, Xe 69 K). The condensation temperatures given
for the noble gas clathrate hydrates assume that sufficient water ice is available
for reaction and that all of each noble gas forms a clathrate hydrate. If no
water ice is available, the noble gases condense as ices at lower temperatures
than their clathrates form.
The condensation chemistry of carbon and nitrogen depends on whether
or not CH4 and NH3 are the predominant C- and N-bearing gases. The two
major C-bearing gases in solar composition material over a wide P, T range
are CO and CH4 . These two gases are converted into one another by the net
reaction
CO + 3H2 = CH4 + H2 O (16)
11, 069.94
logK16 = − 1.17969logT − 8.96596 (17)
T
Equation (17) gives the equilibrium constant for reaction (16) from 298 - 2500
K. Chemical equilibrium calculations using this data and the solar elemental
abundances show that CO is stable at high temperatures and low pressures
while CH4 is stable at low temperatures and high pressures in solar compo-
sition material (e.g., see [13]). For example, at 10−4 bar total pressure, CO
is the major C-bearing gas at temperatures greater than 625 K, CH4 is the
major C-bearing gas at temperatures less than 625 K, and the two gases
have equal abundances at 625 K. The equal abundance point shifts to higher
temperatures with higher pressures, as shown in Figure 14.
However, the kinetics of the gas phase conversion of CO to CH4 become
so slow that it may not happen unless grain catalyzed reactions occur. If all
carbon remained as CO at the low temperatures in solar composition mate-
rial, then the water ice abundance was decreased below the amount which
could condense if CO were converted to CH4 . On the other hand, if CO were
efficiently converted to CH4 and/or other hydrocarbons, then a sizable frac-
tion of the total O was released from CO and was available for formation
of water ice. The water ice/rock mass ratios in “icy” bodies formed in the
solar nebula, where CO was the dominant carbon gas, are predicted to be
lower than the water ice/rock ratios in “icy” bodies formed in the subnebulae
around Jupiter and Saturn, where CH4 was the dominant carbon gas. (Plan-
etary scientists think that the Galilean satellites (Io, Europa, Ganymede, and
Callisto) of Jupiter and Titan and other regular satellites of Saturn formed
in miniature versions of the solar nebula known as subnebulae. These existed
around Jupiter and Saturn during their formation and were higher density
regions with different chemistry than the surrounding solar nebula.) To first
372 Fegley, Schaefer

Fig. 14. Percentage distribution of carbon between CO and CH4 in a solar compo-
sition gas as a function of temperature and pressure. The line shows where CO and
CH4 have equal abundances. Above the line, CO is more abundant, whereas below
the line, CH4 is more abundant

approximation, this distinction is observed and supports the chemical mod-

eling. The two major N-bearing gases in solar composition material are N2
and NH3 (see Figure 15). They are converted into one another by the net
thermochemical reaction

N2 + 3H2 = 2N H3 (18)
6051.59
logK = ( ) − 1.21176logT − 7.89739 (19)
T
Equation (19) gives the equilibrium constant from 298 - 2500 K for reaction
(18). Chemical equilibrium calculations using this data and the solar elemental
 Cosmochemistry 373

abundances show that N2 is the dominant N-bearing gas at high temperatures

and low pressures and that NH3 is the major N-bearing gas at low tempera-
tures and high pressures. Figure 15 shows the N2 /NH3 equal abundance line
over a wide P, T range. At higher temperatures, N2 is the only nitrogen-
bearing gas of any importance. Along the 10−4 bar isobar, NH3 remains the
second most abundant N-bearing gas until about 1670 K where monatomic
N becomes the second most abundant gas. However, even at 2000 K, 10−4
bar the N2 /N molecular ratio is 100,000 and all other N-bearing gases are less
abundant.

Fig. 15. Percentage distribution of nitrogen between N2 and NH3 in a solar compo-
sition gas as a function of temperature and pressure. The line shows where N2 and
NH3 have equal abundances. Above the line, N2 is more abundant, whereas below
the line, NH3 is more abundant

The N2 /NH3 equal abundance line is slightly different from the CO/CH4
equal abundance line because equal abundances of N2 and NH3 do not
374 Fegley, Schaefer

correspond to 50% of total nitrogen in each gas. Figure 15 shows that at

10−4 bar total pressure, N2 and NH3 have equal abundances at 345 K, but
50% of total nitrogen is in each gas at 320 K. At lower temperatures, NH3
first condenses as ammonium carbonate NH4 HCO3 , or ammonium carbamate
(NH4 COONH2 ). The exact amount of NH3 in these compounds depends upon
the amount of CO2 and is hard to quantify because the amount of CO2 de-
pends on the rate of the CO to CO2 conversion in different astronomical
environments such as the solar nebula or another protoplanetary accretion
disk. The remaining NH3 condenses as ammonia monohydrate (NH3 · H2 O)
at 131 K, 10−4 bar.
However, the gas phase conversion of N2 to NH3 is a slow reaction and may
not occur over long times, such as the lifetime of the solar nebula. Industrial
production of NH3 from N2 (the Bosch - Haber process) uses Fe-based cat-
alysts to speed up the reaction. Iron-rich metal grains and magnetite Fe3 O4
grains are common in chondritic meteorites and it is likely that such grains
catalyzed the N2 to NH3 conversion in the solar nebula and other solar com-
position systems, e.g. see [14]. If N2 remains the major N-bearing gas due to
kinetic limitations, N-bearing ices do not form until very low temperatures
where either N2 clathrate hydrate or N2 ice condense. Temperatures of 20 -
40 K are required for this to happen.
The most important of the atmophile elements is hydrogen, which is the
most abundant element in solar composition material. Hydrogen’s dominance
controls the chemistry of solar composition material. With the exception of
helium, which is non-reactive, hydrogen is about 1,000 times as abundant
as all other elements combined. Thus, hydrogen-bearing gases (hydrides) are
major or important gases at chemical equilibrium for many elements. A few
examples are H2 O, CH4 (at low temperatures), NH3 (at low temperatures),
H2 S, HF, HCl, and HBr.
Most hydrogen remains in elemental form because no other element
amounts to more than about 0.1% of the hydrogen elemental abundance.
Figures 16 and 17 illustrate this point. They show the mole fractions of the
major H-bearing gases. By definition, the mole fractions of all gases that are
present add up to unity. Figure 17 shows that the major H-bearing gases are
always elemental gases. With decreasing temperature the major H-bearing gas
changes from ionized hydrogen H+ (at temperatures higher than those shown
on the graph), to monatomic hydrogen H, and finally to molecular hydrogen
H2 . The hydrides and all other H-bearing gases are much less abundant than
the different allotropes of hydrogen.
Figure 16 shows the percentage distribution (on a logarithmic scale) of hy-
drogen between the major H-bearing gases over the same temperature range.
Figure 16 shows that H2 is 50% dissociated to monatomic H at 2,230 K (i.e.,
the temperature where each gas has a mole fraction of 0.50). Monatomic H is
50% ionized to H+ at 8,700 K, above the highest temperature on the graphs.
Thermal ionization of monatomic H occurs via the reaction

H(gas) = H + (gas) + e− (gas) (20)

 Cosmochemistry 375

Fig. 16. Percentage distribution of hydrogen between different phases in a so-

lar composition gas as a function of temperature for a total pressure of 10−4
bar. Gaseous species are indicated with (g). Solid phases (Na2 Mg3 Al2 Si8 O22 (OH)2 ,
Ca5 (PO4 )3 OH, Ca2 Mg5 Si8 O22 (OH)2 , KMg3 AlSi3 O10 (OH)2 , Mg3 Si4 O10 (OH)2 ) are
not shown, due to their low abundances (1 < ×10−4 %)

The electron pressure in this reaction is that produced by the ionization of

all elements in a solar composition gas - not only the electron pressure due
to hydrogen ionization. As the total pressure decreases, thermal dissociation
and ionization of hydrogen become more important at lower temperatures. For
example, hydrogen is 50% dissociated at 1,880 K and 50% ionized at 7,100 K
at 10−6 bar total pressure.
376 Fegley, Schaefer

Fig. 17. Gas phase chemistry of hydrogen in a solar composition gas as a function
of temperature at a total pressure of 10−4 bar. The abundances of the gases are
shown in mole fractions

4 Summary
In this chapter, we reviewed the methods and results of chemical equilibrium
calculations applied to solar composition material. These types of calcula-
tions are applicable to chemistry in a variety of astronomical environments
including the atmospheres and circumstellar envelopes of cool stars, the so-
lar nebula and protoplanetary accretion disks around other stars, planetary
atmospheres, and the atmospheres of brown dwarfs. The results of chemical
equilibrium calculations have guided studies of elemental abundances in mete-
orites and presolar grains and as a result have helped to refine nucleosynthetic
models of element formation in stars.
 Cosmochemistry 377

Acknowledgment This work was supported by funding from the NASA

Astrobiology and Planetary Atmospheres Program.

References
1. K. Lodders, B. Fegley, Jr.: Condensation Chemistry of Carbon Stars. In:
Astrophysical Implications of the Laboratory Study of Presolar Materials,
AIP Conference Proceedings v. 402, ed by T. J. Bernatowicz, E. Zinner
(American Institute of Physics, Woodbury, NY 1997) pp 391-423
2. B. Fegley, Jr., K. Lodders: Icarus 110, 117 (1994)
3. K. Lodders: J. Phys. Chem. Ref. Data 28, 1705 (1999)
4. K. Lodders: J. Phys. Chem. Ref. Data 33, 357 (2004)
5. K. Lodders: Astrophys. J. 591, 1220 (2003)
6. H. Palme, B. Fegley, Jr.: Earth Planet. Sci. Lett. 101, 180 (1990)
7. H. Palme, F. Wlotzka: Earth Planet. Sci. Lett. 33, 45 (1976)
8. B. Fegley, Jr., H. Palme: Earth Planet. Sci. Lett. 72, 311 (1985)
9. A. S. Kornacki, B. Fegley, Jr.: Earth Planet. Sci. Lett. 79, 217 (1986)
10. A. S. Kornacki, B. Fegley, Jr.: Proc. 14th Lunar Planet. Sci. Conf. J.
Geophys. Res. 89, B588 (1984)
11. J. S. Lewis: Earth Planet. Sci. Lett. 15, 286 (1972)
12. B. Fegley, Jr.: Trends of Volatile Elements in the Solar System. In: Work-
shop on the Origins of Solar Systems ed by J. A. Nuth, P. Sylvester (LPI,
Houston 1988) Tech. Rep. No. 88-04, p. 51.
13. K. Lodders, B. Fegley, Jr.: Icarus 155, 393 (2002)
14. B. Fegley, Jr., R. G. Prinn: Solar Nebula Chemistry: Implications for
Volatiles in the Solar System. In: The Formation and Evolution of Plan-
etary Systems, ed by H. Weaver, L. Danly (Cambridge University Press,
Cambridge, UK 1989) pp 171-211
Solar System Abundances of the Elements

Katharina Lodders

Planetary Chemistry Laboratory, Dept. of Earth & Planetary Sciences

and McDonnell Center for the Space Sciences, Washington University, Campus
Box 1169, One Brookings Drive, Saint Louis, MO 63130, USA

Summary. Representative abundances of the chemical elements for use as a solar

abundance standard in astronomical and planetary studies are summarized. Updated
abundance tables for solar system abundances based on meteorites and photospheric
measurements are presented.

Keywords: solar system abundances, solar photosphere, meteorites, CI chon-

drites

1 Motivations to Study Solar System Elemental

Abundances

The investigation of which chemical elements exist in nature and in what

quantities has a long history. The determination of elemental abundances
in various celestial objects is still a very active field in astronomy, planetary
science, and meteoritics. There are multiple motivations for studying the solar
system abundances of the chemical elements. One reason to study this overall
composition of the solar system is to understand how the diversity of planetary
compositions, including that of our home planet, can be explained, since all
planets in the solar system share a common origin from the material of the
protosolar disk (the solar nebula).
The composition of the sun determines how the sun works and evolves over
time, as composition influences the interior structure of the sun. Although the
Sun is mainly composed of H and He, other heavy elements such as C, N, O,
Ne, Fe, etc., are important opacity sources that influence the energy transport
out of the sun through radiation and convection. The sun is a typical main
sequence dwarf star and its composition is a useful baseline for comparison
to abundances in other dwarf stars and to changes that appear in advanced
stages of stellar evolution. For example, relative to the suns composition,
red giant stars show observable abundance variations that are the result of

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 8, c Springer-Verlag Berlin Heidelberg 2010
380 Katharina Lodders

nucleosynthesis operating in giant stars, and these products have been dredged
up from stellar interiors to the stellar exteriors.
The solar system abundances are a useful local galactic abundance stan-
dard because many nearby dwarf stars are similar in composition; however, in
detail there are some stochastic abundance variations (e.g [1,2,3,5]). The term
cosmic abundances should be avoided because abundances generally decrease
with galactocentric distance. There are also abundance differences between
our galaxy and galaxies at high red-shift; hence there is no generic cosmic
composition that applies to all cosmic systems.
Finally, solar abundances are a critical test of nucleosynthesis models and
models of Galactic chemical evolution [6, 7, 8]. Ideally, such models should
quantitatively explain the elemental and nuclide distributions of solar system
matter.
The sun has the most mass (>99%) of the solar system objects and there-
fore it is the prime target for studying solar system abundances. Most elements
can be measured in the sun’s photosphere, but data from the solar chromo-
sphere and corona, solar energetic particles, solar wind, and solar cosmic rays
(from solar flares), help to evaluate abundances of elements that have weak
absorption lines (because these elements are low in abundance or only have
blended absorption lines in the photospheric spectrum).
Below we will see that meteorites, smaller rocks from asteroidal objects
delivered to Earth, provide important information for solar system abundances
of non-volatile elements. Other sources to refine solar system abundances are
analysis of other solar system objects such as the gas-giant planets, comets and
the interplanetary dust particles from comets. Outside the solar system, the
compositions of hot B stars, planetary nebulae, Galactic cosmic rays (GCR),
the nearby interstellar medium (ISM) and H II regions have been employed
to amend the solar system abundances of some elements.
Solar or solar system abundance data derived from meteorites and the
solar photosphere are reviewed periodically. References [9,10,11,12,13,14,15,
16, 17] give some compilations that summarize information on photosphere
and meteoritic abundances used as solar system abundance standards since
1989.

2 Meteorites as Abundance Standards for Non-Volatile

Solar System Matter
About a century ago, it became more evident that meteorites may contain
chemical and mineralogical information about the earliest solid objects that
existed in the solar system, and that they may carry resemblance to the mate-
rials that accreted to planets like the Earth. Several different meteorite groups
are recognized. With respect to abundances, the chondritic meteorites are the
most important ones. Meteorites that contain small silicate spheres are called
chondrites (after Greek chondros for sphere). In addition to silicate minerals,
 Solar System Abundances 381

most chondrites contain FeNi metal and iron sulfide (the mineral troilite, FeS,
is most common), and a host of minor minerals. The major chondrite groups
are ordinary, enstatite and carbonaceous chondrites; each group has further
divisions based on the different proportions of their major minerals. The most
common meteorites are, as the name implies, the ordinary chondrites. The en-
statite chondrites, named after their most common silicate mineral enstatite,
contain exotic sulfides and metal phases that are only stable under highly
reducing conditions. The carbonaceous chondrites contain up to several per-
cent of carbon, depending on sub-type. The carbonaceous meteorites of Ivuna
type (abbreviated as CI chondrites for Carbonaceous and Ivuna) are the most
important ones for solar system abundance determinations.
The CI chondrites experienced severe aqueous alteration on their parent
asteroid, and if they ever contained chondrules, metal, and sulfides as the
other carbonaceous chondrites do, these phases were largely erased. The CI
chondrites mainly consist of fine grained, hydrous silicates, magnetite (F e3 O4 )
probably produced by oxidation of FeNi metal, Fe-Ni bearing sulfides other
than troilite, and various salts. Despite the absence of chondrules, the CI
chondrites are still chondrites because their overall elemental composition for
many elements is much closer to chondritic meteorites than to iron meteorites
or so-called achondrites, that mainly consist of silicates and experienced severe
melting.
Up to the 1950s, the CI chondrites were not given too much attention
for their potential role as abundance standards. The classical abundance pa-
pers by [18,19,20] used elemental analyses of silicates, metal, and sulfides from
different chondrites and assumed some representative proportion of metal, sil-
icate, and sulfides to come up with a meteoritic abundance standard. By the
1970s, the picture emerged that the CI chondrites as well as another carbona-
ceous chondrite group named CM after the Mighei meteorite may be suitable
groups to represent the solar system abundances of elements that provide the
cations in rock-forming minerals. However, the relative contents of volatile
elements is lower in CM chondrites than in CI chondrites, and a correlation
of CM- to CI chondrite abundance ratios with condensation temperatures
implies that volatility-related fractions occurred in CM chondrites.
Figure 1 shows the concentration ratios for CM chondrites to CI chon-
drites as a function of condensation temperature. The symbols indicate the
mineral phase hosting the elements. There is a smooth decrease in the concen-
tration ratio with condensation temperatures which is independent of mineral
host phase. This correlation indicates that the elemental abundances in CM
chondrites are volatility controlled. The higher concentration ratio for refrac-
tory elements plots above unity and reflects that CM chondrites accumulated
a higher proportion of refractory elements. The ratio below unity is for the
volatile elements incomplete condensation or accumulation. This limits the
use of CM chondrites as abundance standards for elements with condensation
temperatures less than ∼ 1500 K. (see chapter by B. Fegley and L. Schaefer
in this volume for condensation chemistry of the elements).
382 Katharina Lodders

mass concentration ratio CM chon./Cl chon.

w Ce
1.5 Sc U
Be CM chondrites
Zr PNb
Hf M Ca v Eu
Os Al S
Gd Ta
Re Ba
Th Ru Yb Mg Si
YTm Lu Ir
Ti Pt FePdCr As
Rh P Au
1.0 Co Ni Li Cu

Mn Sb
Ag Ge Br
Na F
K Ga Rb
Bi
Se
Cl B Cd TI
Cs Te Zn In
0.5 Pb I
Sn S

0.0
1800 1600 1400 1200 1000 800 600 400
50% condensation temperature, K, at 10–4 bar

Fig. 1. The decreasing element concentration ratio of CM- over CI chondrites indi-
cates volatility related fractionations in CM meteorites and makes them of limited
use as an abundance standard. The different symbol shapes indicate the principal
mineral host phase for the elements (circle: lithophile elements in silicate and ox-
ides; box: siderophile elements in metal alloy; chalcophile: sulfides; triangle: halogen).
Data sources for CM chondrites: [21] plus updates; CI chondrites: [17]

The EH enstatite chondrite group also has higher relative abundances of

volatile elements, however, in detail chemical fractionations of volatile and
non-volatile elements are apparent. In a comparison of the abundances of all
chondrite groups to those in the solar photosphere, only the CI chondrites
are found to have the closest match. Finally, the CI chondrites give the best
agreement (of all chondrite groups) between observed and theoretically pre-
dicted nuclide abundances as a function of mass number (see e.g., [22] for
arguments in favor of CI chondrites as standard).
 Solar System Abundances 383

Part of the reason why it took so long to recognize the significance of

CI chondrites is simply that CI chondrites are very rare. Out of the ∼ 1000
recorded observed meteorite falls from which material is preserved, only 5
CI chondrites are known. Among the 40,000 or so meteorites collected in
Antarctica, only a few are CI chondrites. The meteorites are very fragile and
decompose easily, for example, if placed in water, CI chondrites immediately
begin to disintegrate. Hence, CI chondrites that are found a long time after
their fall are not useful for abundance studies, as chemical information is easily
altered or lost.
Table 1 lists the 5 observed CI chondrite falls, which are named after the
nearest town to their fall location. Sufficient mass for study is available only
for three of them, notably the Orgueil meteorite, which is probably one of the
most and best analyzed rocks on this planet.

Table 1. Observed CI chondrite meteorite falls

Meteorite Date of Fall Country Preserved Mass
Alais 15 March 1806 France 6 kg
Orgueil 14 May 1868 France 14 kg
Tonk 22 January 1911 India 10 g
Ivuna 16 December 1938 Tanzania 0.7 kg
Revelstoke 31 March 1965 Canada ≤1g

The other reason why it took so long to realize that CI chondrites are a
chemically special meteorite group is that early chemical analysis methods
were not able to analyze trace element abundances in small samples with suf-
ficient precision. The advances came through the application of instrumental
neutron activation analyses and mass-spectroscopic measurements, which re-
vealed the chemical differences in minor element abundances. Compared to
all other chondrites, the CI chondrites have the highest relative contents of
volatile elements, such as C, O, alkalis, chalcogenides, and halogens. In con-
trast, the abundances of refractory elements such as Al, Ca, Si, Mg, Fe, alkaline
earths, and rare earth elements are less variable among chondrites, but im-
portant abundance differences for these elements exist between the chondrite
groups and provide lively discussions among meteoriticists.
At the same time that elemental abundance analyses improved for rock
samples, spectroscopic abundance determinations for the solar photosphere
advanced. A comparison of elemental abundances in the solar photosphere to
CI chondrites then showed the best agreement for most elements; the excep-
tions being H, C, N, O, and the noble gases. These elements form extremely
volatile compounds that may have never accreted to the CI chondrite parent
asteroid or were easily lost from CI chondrite material while in space or in
the terrestrial environment.
384 Katharina Lodders

2.1 Composition of CI chondrites

The reference composition of CI chondrites in Table 2 is based on a new review

by [17]; henceforth LPG09. The previous evaluations of the composition of CI
chondrites were done by [13]; henceforth L03 and [14]; henceforth PJ03. These
two evaluations used slightly different approaches to derive recommended CI
chondrite compositions. PJ03 emphasized the use of the Orgueil meteorite as
the CI standard rock, because it is the most massive of the 5 CI chondrite
falls and therefore the most analyzed one (see Table 1). L03 used data from all
CI chondrites and computed weighted average compositions. However, since
most analytical data are for the Orgueil meteorite, these weighted averages are
also dominated by this chondrite. A comparison of data from Orgueil and the
other four CI chondrites in [13] shows that compositional differences among
CI chondrites are relatively small.
In our new CI chondrite compilation [17], we included many new data on
trace elements that have become available in recent years, primarily because
of improvements in instrumentation. In particular, application of Inductively
Coupled Plasma Mass Spectrometry (ICP-MS) led to many new high quality
analyses. These data and resulting updates of CI chondrite compositions are
discussed in [17], however, an extensive discussion of the database containing
200+ references on CI chondrite analyses is deferred to the future [23].
In addition to computing average concentrations from several reliable anal-
yses for a given element, we employed the element ratio method to find stan-
dard values for element concentrations in CI chondrites. This method relies
on the fact that concentration ratios of elements with similar cosmo- and geo-
chemical properties are usually more constant than the absolute measured
element concentration. The reason for using this method is that analytical
data for CI meteorites can show variable absolute concentrations for some el-
ements, which is caused by the variable water contents and/or massive alter-
ation of fine grained matrix material. In addition, small scale heterogeneities
naturally arise because these meteorites are breccias (see [24, 25]). If there
are compositional heterogeneities in the meteorite, the measured concentra-
tion of elements residing in common mineral phases may vary from sample to
sample, however, the concentration ratios of these elements relative to each
other will remain constant. Therefore, it is sometimes practical to take the
concentration ratio of a pair of geochemically similar elements from several
samples and to use the absolute abundance of the accurately determined el-
ement of that pair to calculate the concentration of the other element. Since
there are only three CI chondrites for which larger numbers of analysis exist,
it is also useful to compare the element concentration ratios in CI chondrites
to that of other carbonaceous chondrites. The inclusion of data, e.g., from CM
chondrites, which are relatively closely related to the CI chondrites, increases
the statistics for abundance determinations by the ratio method. However, in
using other data than for CI chondrites, the explicit assumption is made that
chemical fractionations of the elements are absent between CI chondrites and
 Solar System Abundances 385

the other chondrite groups invoked. This assumption is usually justified as

long as CM chondrites and refractory elements are involved. Figure 1 shows
that the concentration ratio of refractory elements in CI and CM chondrites
plots at a comparable constant level.
The element ratio method and the direct average method usually give
consistent results within 3% for most elements, which is easily within the esti-
mated uncertainty from the statistics or individual analyses. However, includ-
ing recent new analyses, the data spread has become wider for the elements
Y, Zr, Hf, Nb and Ta. This spread seems to indicate real heterogeneities for
these elements in CI chondrite samples, and the ratio method may lead to
more reliable results (see LPG09 for more details).
The selected concentrations for the Orgueil meteorite, considered as most
representative for the CI chondrites, are listed in Table 2; details about the
data sources are in LPG09. Concentrations are given as parts per million
(ppm) by mass (10,000 ppm = 1 mass%). Corresponding atomic abundances
normalized to 106 Si atoms (the ‘cosmochemical abundance scale’) are listed
as well.

3 Photospheric abundances
Russell [26] reported the earliest comprehensive analysis of the solar photo-
sphere for 56 elements. Since then, numerous element abundance data have
been derived by spectroscopy of the solar photosphere.
Converting the absorption lines into abundances requires knowledge of line
positions of neutral and ionized atoms, as well as their transition probabilities
and lifetimes of the excited atomic states. In addition, a model of the solar
atmosphere is needed. In the past years, atomic properties have seen many
experimental updates, especially for the rare earth elements (see below). Older
solar atmospheric models used local thermodynamic equilibrium (LTE) to
describe the population of the quantum states of neutral and ionized atoms
and molecules according to the Boltzmann and Saha equations. However, the
ionization and excitation temperatures describing the state of the gas in a
photospheric layer may not be identical as required for LTE. Models that
include the deviations from LTE (=non-LTE) are used more frequently, and
deviations from LTE are modeled by including treatments for radiative and
collision processes (see, e.g., [27, 28]).
Solar atmospheric models have evolved from one dimension to more com-
plicated 2D and 3D models designed to take into account effects of convection
and granulation on radiative transfer in the solar atmosphere. Recent appli-
cations of 3D models instead of older 1D models leads to lower abundances
of several elements, notably oxygen, from previously determined values. Sig-
nificant reductions of photospheric abundances in other elements (e.g., Na,
Al, Si) were also found (see, e.g., [15]). However, different 3D model assump-
tions lead to different results, see for example the discussion by [29] for silicon.
386 Katharina Lodders

Table 2. CI chondrite composition

Z ppm σ Si = 106 σ
1 H 19700 2000 5.13E+06 5.1E+05
2 He 0.00917 0.601
3 Li 1.47 0.19 55.6 7.2
4 Be 0.0210 0.0015 0.612 0.043
5 B 0.775 0.078 18.8 1.9
6 C 34800 3500 7.60E+05 7.6E+04
7 N 2950 440 55300 8300
8 O 459000 46000 7.63E+06 7.6E+05
9 F 58.2 8.7 804 121
10 Ne 1.80E-04 0.00235
11 Na 4990 250 5.70E+04 2.8E+03
12 Mg 95800 2900 1.03E+06 3E+04
13 Al 8500 260 8.27E+04 2.5E+03
14 Si 107000 3000 1.00E+06 3E+04
15 P 967 97 8190 820
16 S 53500 2700 4.38E+05 2.2E+04
17 Cl 698 105 5170 780
18 Ar 0.00133 0.00962
19 K 544 27 3650 180
20 Ca 9220 461 60400 3000
21 Sc 5.9 0.3 34.4 1.7
22 Ti 451 36 2470 200
23 V 54.3 2.7 280 14
24 Cr 2650 80 13400 400
25 Mn 1930 58 9220 280
26 Fe 185000 6000 8.70E+05 2.6E+04
27 Co 506 15 2250 70
28 Ni 10800 300 4.83E+04 1.4E+03
29 Cu 131 13 541 54
30 Zn 323 32 1300 130
31 Ga 9.71 0.49 36.6 1.8
32 Ge 32.6 3.26 118 12
33 As 1.74 0.16 6.10 0.55
34 Se 20.3 1.42 67.5 4.7
35 Br 3.26 0.49 10.7 1.6
36 Kr 5.22E-05 1.64E-04
37 Rb 2.31 0.16 7.10 0.50
38 Sr 7.81 0.55 23.4 1.6
39 Y 1.53 0.2 4.52 0.45
40 Zr 3.62 0.4 10.4 1.0
 Solar System Abundances 387

Table 2. CI chondrite composition – Continued

Z ppm σ Si = 106 σ
41 Nb 0.279 0.028 0.788 0.079
42 Mo 0.973 0.097 2.66 0.27
44 Ru 0.686 0.041 1.78 0.11
45 Rh 0.139 0.014 0.355 0.035
46 Pd 0.558 0.028 1.38 0.07
47 Ag 0.201 0.010 0.489 0.024
48 Cd 0.674 0.047 1.57 0.11
49 In 0.0778 0.0054 0.178 0.012
50 Sn 1.63 0.24 3.60 0.54
51 Sb 0.145 0.021 0.313 0.047
52 Te 2.28 0.16 4.69 0.33
53 I 0.53 0.11 1.10 0.22
54 Xe 1.74E-04 3.48E-04
55 Cs 0.188 0.009 0.371 0.019
56 Ba 2.41 0.14 4.61 0.28
57 La 0.242 0.012 0.457 0.023
58 Ce 0.622 0.031 1.17 0.06
59 Pr 0.0946 0.0066 0.176 0.012
60 Nd 0.471 0.024 0.857 0 .043
62 Sm 0.152 0.008 0.265 0.013
63 Eu 0.0578 0.0029 0.100 0.005
64 Gd 0.205 0.010 0.342 0.017
65 Tb 0.0384 0.0027 0.063 0.004
66 Dy 0.255 0.013 0.412 0.021
67 Ho 0.0572 0.0040 0.091 0.006
68 Er 0.163 0.008 0.256 0.013
69 Tm 0.0261 0.0018 0.041 0.003
70 Yb 0.169 0.008 0.256 0.013
71 Lu 0.0253 0.0013 0.038 0.002
72 Hf 0.106 0.005 0.156 0.008
73 Ta 0.0145 0.0015 0.0210 0.0021
74 W 0.0960 0.0096 0.137 0.014
75 Re 0.0393 0.0039 0.0554 0.0055
76 Os 0.493 0.039 0.680 0.054
77 Ir 0.469 0.023 0.640 0.032
78 Pt 0.947 0.076 1.27 0.10
79 Au 0.146 0.015 0.195 0.019
80 Hg 0.350 0.070 0.458 0 .092
81 Tl 0.142 0.011 0.182 0.015
82 Pb 2.63 0.18 3.33 0.23
83 Bi 0.110 0.010 0.138 0.012
90 Th 0.0310 0.0025 0.0351 0.0028
92 U 8.10E-03 6.5E-04 8.93E-03 7.1E-04
388 Katharina Lodders

Hence abundances derived with 3D models still have to be regarded with some
caution until model assumptions and details are sorted out.
Lower abundances for some elements derived from these models also cause
problems for standard solar models that describe the evolution of the sun to its
current radius and luminosity (see [30]). Another problem is that some of the
3D abundances compare worse to meteoritic data than before. In the following
preference is given to elemental abundances derived with more conservative
solar atmospheric models; however, for some elements (e.g., P, S, Eu) the
results from 1D/2D and some 3D models produce consistent results.
The solar photospheric abundances from various literature sources are
listed in Table 3. Several new measurements have become available since the
compilation by L03 that lists references to data not described below.
Elemental abundances are normalized to 1012 atoms of hydrogen. The
ratio of the number of atoms of an element N(X), relative to the number of
hydrogen atoms, N(H) is given on a logarithmic scale, and frequently used
notations are:
A(X) = log N(X)/N(H) + 12 = logX
Uncertainties for the logarithmic scale are given in logarithmic units dex
equivalent to an uncertainty factor on a linear scale. The uncertainty relation
is σ(in%) = (10σ(in dex) − 1) × 100.
As already mentioned, most elements are determined by absorption spec-
troscopy. Exceptions are rare gases and some elements that have no accessible
or only heavily blended lines for use in quantitative spectroscopy of the pho-
tosphere (e.g., As, Se, Br, Te, I, Cs). Sunspot spectra are used for e.g., F, Cl,
In, and Tl, however, the abundance uncertainty for these elements is rather
large. For the noble gases, theoretical considerations or data from other ob-
jects must be used to obtain representative solar values (see below and [13]).
The following describes updates for several elements.

Helium: The He discovery was made from spectral lines in the coronal
spectrum during a solar eclipse in 1868; however, despite being first discov-
ered in the sun, the He abundance cannot be determined spectroscopically in
the solar photosphere. The He abundance is determined from results of he-
lioseismic models, as described below in the section for the present-day mass
fractions of H, He, and heavy elements.
Lithium: The value of A(Li) = 1.1 ± 0.1 from [31] used in [13] is kept.
The analysis from [32] of A(Li)=1.0 ± 0.1 was amended with 3D models by
[15] to give 1.05 ± 0.10. Considering the already large uncertainty and the
uncertainties in the 3D models, the previous selection has been kept.
Beryllium: The photospheric and meteoritic Be abundance determina-
tions are associated with difficulties [13]. By 2003, there were two conflicting
photospheric determinations (A(Be)= 1.15 ± 0.10 and 1.40 ± 0.09). A sub-
sequent re-analysis of the photospheric Be abundance by [33] yields A(Be) =
1.38 ± 0.09, in support of the higher photospheric value. The meteoritic value
of A(Be) = 1.41 ± 0.08 in L03 was based on chemical element systematics in
 Solar System Abundances 389

CM and CV chondrites because there was only one previous determination

of Be in CI chondrites. This situation has improved since [34] report 2 Be
analyses for the Orgueil CI chondrite. The average of the 3 Be determinations
is 0.021 ± 0.002 ppm (by mass), corresponding to a meteoritic A(Be)= 1.32
± 0.03. Within the larger uncertainties, the photospheric and meteoritic Be
abundances are in agreement, suggesting that no Be destruction has occurred
over the Sun’s lifetime. However, taken at face value, a lower Be abundance
in CI chondrites than in the photosphere is not easily accounted for by any
physical or chemical fractionation process.
Carbon: The C abundance of A(C) = 8.39 (± 0.04) from [35] was selected
in [13]. This value was confirmed by [36], and derived from the analysis of CO
by [37]. However, this C abundance is based on 3D models from one group
and an independent confirmation of this value by another model is desirable.
Nitrogen: The N abundance is among the more uncertain elemental abun-
dances. Asplund et al. [15] (henceforth also A05) derived N abundances from
neutral N (NI) and the NH molecule. An LTE analysis of the NI lines gives
A(N) = 7.88 ± 0.08, a NLTE analysis gives 7.85 ± 0.08. The N abundance
from the NH molecule and 3D models gives a lower abundance of A(N) =
7.73 ± 0.05. A05 recommend A(N) = 7.78 ± 0.06 in their table. However,
this seems quite low in comparison to previous estimates of e.g., 7.83 ± 0.11
in L03 that was based on [27]. A recent detailed study by [38] found 7.86 ±
0.12, which is recommended here.
Oxygen: The recommended O abundance of 8.73 ± 0.07 is an average
from the recent determinations by [39, 40, 41].
Caffau et al. [39] recommend A(O)= 8.76 ± 0.07 from a detailed analyses
of several oxygen lines and 3D atmospheric models. They include different
model NLTE corrections, as well as treatments of collisions with H atoms on
NLTE level populations of O. Within the (unfortunately still larger) uncer-
tainties, their O abundance is closer to most of the recent low O abundance
determinations than to the older values advocated by [9, 12]. Melendez et al.
[41] derived the O abundances from several O lines and compared various
model approaches, which lead to an average O abundance of A(O) = 8.71 ±
0.02, where the uncertainty does not really cover the real uncertainty of the
true O abundance. Using different 3D models,[40] report A(O) = 8.72 ± 0.06
from modeling of several O I lines.
A review on the problem of the solar O and other light element abundances
is given by [30]. The photospheric O abundance determination remains enig-
matic, as the value obtained by [35,33] appears to be erratically low, whereas
the older abundances quoted in compilations such as [9] (AG89), [12] (GS98),
as well as the recent study by [42] probably give values that are much too
high.
The downward adjustment in solar O by [35] was substantial compared
to the values given in the compilations by [12] (A(O)= 8.83) and [9] (A(O)
= 8.93). The decrease was mainly due to the realization that the O line com-
monly used in the abundance analysis is blended with a Ni line. The value
390 Katharina Lodders

by Caffau et al. is 0.07 dex higher (factor 1.17) than the abundances derived
by [35]; and it is also higher than the low values determined by Asplund
and coworkers. [39] give these reasons why the O abundances was previously
underestimated. In the studies by Allende Pietro et al. and Asplund et al.,
lower equivalent widths were used, and effects of collisions of H atoms in the
calculations of the statistical equilibrium of O were not considered.
Ayres et al. ( [42] ) derived the O abundance from weak CO absorptions
but did not find support for a lower O abundance. They recommend A(O) =
8.85, much closer to the value in [9] (A(O) = 8.93). However, to obtain the
absolute O abundance using the CO molecule, Ayres et al. assumed that C/O
= 0.5. Oxygen is about twice as abundant as carbon, and carbon is largely
tied into the CO molecule. An analysis of the CO abundance thus provides
only the lower limit to the total C abundance (as C is present also in other
gases such as C, CH etc), and the O abundance can only be derived if the total
C/O ratio and the C abundance is known. [42] find A(C) =8.54, and with
an assumed C/O of 0.5, their corresponding O abundance is A(O) = 8.85.
However, using the same CO lines [37] find a lower O abundance with their
3D models than reported by [42]. Given the problem that the O abundances
from CO requires various assumptions about the distribution of C and the
C/O ratio, these abundance determinations appear even more uncertain.
Adopting A(C) = 8.39 and A(O) = 8.73 ± 0.07 gives a C/O ratio of C/O =
0.457, less than the C/O ratio of 0.50 in more recent studies and compilations,
but still somewhat higher than the value of C/O = 0.427 from AG89.
The O abundance has been steadily revised downward from [26] value over
the years, likewise, the abundance values of C and N from more recent analysis
tend to be smaller. Figure 2 shows historical abundance trends for the more
abundant elements C, N, O, Si, and Fe from various sources starting with [26]
and including data sources quoted in the compilations mentioned above. The
variations on the graph appear small, however, one should not forget that the
data are plotted on a log scale and a difference of 0.5 dex corresponds to about
a factor of 3 change in abundance. The Fe abundance was a big problem until
the early 1970s when improved transition probabilities and lifetimes confirmed
that the photospheric Fe abundance had been underestimated by about a
factor of 3-10, which did not match the meteoritic value. This issue has now
been put to rest and the photospheric and meteoritic value are now in perfect
agreement. However, the issue of the C, N, and O abundances cannot be aided
with meteoritic abundances, and future analyses have to resolve this.
Neon Results by [43] from Ne I and Ne II lines in nearby, early type B
stars yield log A(Ne) = 7.97 ± 0.07, which can be taken as characteristic of the
present day ISM. If neon contributions from more massive AGB stars to the
ISM over the past 4.6 Gyr are negligible, this value may be taken as represen-
tative for the Sun. Landi et al. ([44]) obtained A(Ne) = 8.11 ± 0.1 from solar
flare measurements in the ultra-violet. This Ne abundance is derived indepen-
dently of the O abundance, unlike other Ne abundance determinations that
rely on Ne/O ratios and an adopted O abundance (see e.g., [13] and below).
 Solar System Abundances 391

The average from these studies gives A(Ne) = 8.05 ± 0.06, but considering
the larger uncertainties in the determinations, an overall uncertainty of 0.10
dex (25%) is easily warranted for the recommended value here.

9.5
9.0 carbon

8.5
8.0
7.5
8.5

8.0
A(X); normalized to A(H)=12

nitrogen
7.5
9.5

9.0

8.5
oxygen
8.0
8.0

7.0
iron
6.0
8.0

7.5
silicon
7.0
1930 1940 1950 1960 1970 1980 1990 2000 2010
Fig. 2. Photospheric abundance determinations over time

This Ne value compares relatively well to the Ne abundance adopted in

older compilations by [8] (8.09) or [11](8.08). However, the Ne values in more
recent compilations (A(Ne) = 7.87 ± 0.1 in L03, and A(Ne) = 7.84 ± 0.06
in A05) are lower by 26% (0.10 dex). The Ne abundance in previous com-
pilations was mainly based on the characteristic Ne/O ratio of 0.15 for the
local ISM and solar energetic particles. The Ne abundance was then calcu-
lated from the adopted O abundances. Since the adopted O abundances have
changed to lower values in these compilations, the Ne abundances dropped as
well. Using Ne/O = 0.15 and our selected O abundance from above, the Ne
392 Katharina Lodders

abundance from the ratio method gives A(Ne) = 7.91, about 0.14 dex lower
than the preferred Ne value which is independent of an adopted O abundance.
More recently, [45] recommends Ne/O = 0.17 ± 0.05 for the photosphere from
extreme UV measurements of supergranule cell center regions. This ratio and
the selected O abundance (A(O)=8.73) yield A(Ne) = 7.96 (± 0.15, uncer-
tainty only from ratio). This value is lower than the recommended value, but
agrees within uncertainties permitted by the given Ne/O ratio.
In principle, the Ne abundance can also be derived from solar wind data.
In 2006, Bochsler [46] reported A(Ne) = 8.08 ± 0.12. Bochsler [47] further
analyzed the solar wind data for fractionations, and finds A(Ne) =7.96 ± 0.13,
and also A(O) = 8.87 ± 0.11, with relatively large uncertainties that do not
help to resolve the issue of the uncertain O and Ne abundances.
Neon is the third most abundant heavy element after O and C, and the
heavy elements are important opacity sources that influence radiative transfer
in the sun. Good agreement of standard solar models and helioseismological
observations existed until about the year 2000. Then more sophisticated pho-
tospheric modeling began to yield lower C, N, O, Ne and other heavier element
abundances (see Figure 2, and compilations by L03, A05). A decrease in heavy
element abundances led to solar model results that no longer stood the test
from helioseismology. A detailed review of this problem is given by [30] . The
recommended N, O, and Ne abundances here are larger than previously rec-
ommended in L03 and A05, and it will be interesting to see if these abundances
can bring solar models again in closer agreement with helioseismological con-
straints.
Sodium: the previously selected Na value of 6.3 ± 0.03 in L03 is in agree-
ment with A(Na) = 6.27 measured by [3]; [15] found A(Na) = 6.17 ± 0.04
from six Na lines and their 3D model atmospheres, which is ∼ 25% lower
than the meteoritic value as well as previously determined photospheric Na
abundances. The reason for this difference is not yet clear.
Aluminum: A 3D analysis by A05 gives A(Al) = 6.37 ± 0.06, which is
lower than the previous value of 6.47 ± 0.07 from 1D models selected in L03.
The older value is kept as it is much better in agreement with meteoritic data.
The Al/Si from 3D models in A05 is 1.2 times that of CI chondrites, which is
a large difference that still needs to be understood.
Silicon: The A(Si) = 7.52 ± 0.06 selected here is from [29], which is not
that different from 7.51 ± 0.04 reported by [48] and 7.54 ± 0.05 by [27].
Phosphorus: The value of A(P) = 5.49 ± 0.04 in L03 is changed to the
recent result of A(P) = 5.46 ± 0.04 by [49]. The new value is based on 3D
atmospheric models. According to [49] the value for P with 1D models is not
significantly different. A lower value, A(P) = 5.36 ± 0.04 from a different
3D analysis was given in the compilation by A05. Here the well documented
analysis by [49] is taken for the photospheric abundance, which agrees well
with the meteoritic value of 5.43 ± 0.04.
 Solar System Abundances 393

Sulfur: New results with 3D models by [50,51] lead to A(S) = 7.14 ± 0.01.
They found that 3D models have no big effect on S abundances compared to
1D models.
Argon: The value of A(Ar) = 6.50 is derived from various independent
sources since the Ar abundance cannot be determined spectroscopically in the
photosphere (see [52]).
Potassium: [53] confirmed the K abundance of A(K) = 5.12 ± 0.03 used
in L03. The value A(K) = 5.08 ± 0.07 proposed by A05 appears too low, a
similar situation as for Na above.
Calcium: [3] find A(Ca) = 6.33 ± 0.07 in their LTE analysis of the solar
spectrum (the same value as in [54]), which is lower than the previously
selected value in L03. However, the uncertainty for the value from [3] is high.
The result from 3D models by A05 is about 5% lower. A new Ca analysis for
the photosphere using different model atmospheres is needed.
Scandium: The Sc abundance remains uncertain, [55] recommends a
range of 3.07 < A(Sc) < 3.13. To cover the range of values reported in sev-
eral recent papers A(Sc) = 3.10 with an appropriate uncertainty of 0.1 dex is
recommended.
Titanium: The value from the LTE analysis by [3] is adopted.
Chromium: [56] found A(Cr) = 5.64 ± 0.1, which is identical with the
photospheric value listed in L03 but with much smaller uncertainty.
Manganese: Two recent studies give A(Mn) = 5.37 ± 0.05 [57] and 5.36
± 0.10 [58] and [3] found the same value as [57]. These values are only slightly
lower than the value 5.39 ± 0.03 given in L03. The analyses of the photospheric
Mn abundance seem to confirm that the photospheric Mn abundance is lower
than the meteoritic value of 5.50 ± 0.01. Assuming that indeed both the
photospheric and meteoritic data are reliable, the cause of this abundance
difference must be found.
Nickel: The A(Ni) = 6.23 ± 0.04 from the LTE analysis by [3] is similar
to the previous value and has a lower uncertainty.
Zirconium: [59] found A(Zr) = 2.58 ± 0.02 from a 3D analysis.
Palladium: A value of A(Pd) = 1.66 ± 0.04 was reported by [60]
Indium: Previous determinations of the photospheric In abundance lead
to a value that is substantially higher (∼ 400%) than the CI chondritic value.
[61] suggested that this large difference could be the result of the relatively
high volatility of In. Incomplete condensation of any element into the ma-
terials assembled to the CI chondrite parent body would lead to a relative
depletion of a volatile element in CI chondrites when compared to the Sun.
At 10−4 bar total pressure, half of all In is condensed at 536 K, compara-
ble to the 50% condensation temperatures of other volatile elements like Tl
(532K), S (664K) and Pb (730K; see L03). However, the abundances of these
and other volatile elements are in closer agreement for the photosphere and
CI chondrites, so a fractionation due to volatility cannot explain the huge
photospheric In abundance.
394 Katharina Lodders

A high In abundance can also be ruled out considering the abundance

distribution of the elements and nuclides as a function of atomic mass. A
high In abundance would introduce spikes in the otherwise rather smooth
abundances curve (see below). A recent investigation by [62] shows that the
often used In line at 451.13 nm in the solar sun spot spectrum is blended by
some line of a currently not identified element, which causes an apparently
higher In abundance. Vitas et al. [62] obtained A(In) = 1.50 with no potential
blends considered, and this value is adopted here as an upper limit for the
photospheric In abundance. Their study of other In lines further indicates
that the solar In abundance is unlikely to be higher than the meteoritic value
of A(In) = 0.78, for which [62] also find support from nuclide distribution
systematics from nucleosynthesis.
Rare earth elements: In the past years, many improvements have been
made in abundance analyses of the REE through measurements of atomic
lifetimes and transition probabilities, notably by the Wisconsin group. A re-
cent paper by [63] summarizes the efforts and gives abundances for the REE.
Now the REE abundances are among the best-known abundances for the sun.
The following lists several papers on REE that appeared since 2003; the new
values for Ce, Dy, Tm, Yb, and Lu are from [63].
Praseodymium: The value of A(Pr) = 0.76 ± 0.02 from [63] is preferred
to the previously selected value of A(Pr) = 0.71 ± 0.08 in L03. The value based
on [64] of A(Pr) = 0.58 ± 0.10 adopted by [16] is erroneously low, which is
easily seen from a comparison to the well-established meteoritic value of 0.78
± 0.03.
Neodymium: [65]; A(Nd) = 1.45 ± 0.05
Samarium: [66]; A(Sm) = 1.00 ± 0.03
Europium: [67] employ a 3D hydrodynamic model atmosphere and find
A(Eu) = 0.52 ± 0.02 [67]. They find that 3D effects are negligible for the Eu
determination in the Sun. This value is identical to the value reported by [68].
Gadolinium: [69]; A(Gd) = 1.11 ± 0.03.
Holmium: [70]; A(Ho) = 0.51 ± 0.1.
Erbium: [71]; A(Er) = 0.96 ± 0.03
Hafnium: The photospheric abundance of A(Hf) = 0.88 ± 0.08 selected
in L03 is confirmed with a reanalysis by [72] who determined improved tran-
sition probabilities. Another recent re-determination of the photospheric Hf
abundance gives A(Hf) = 0.87 ± 0.04 [73].
Osmium: [74] found A(Os) = 1.25 ± 0.11. This is significantly lower
than the value of 1.45 ± 0.10 used in previous compilations. Both Os values
seem to be problematic when compared to the meteoritic value of 1.37 ± 0.03.
Assuming that the meteoritic value is reliable, the older photospheric value is
17% too low and the new one 30% too high. Grevesse et al. [16] selected the
newer, smaller value but a conservative approach is to adopt the value with
the smaller difference to the meteoritic value. The older value of A(Os) =
1.45 also appears more reasonable considering abundance systematics in the
Pt-element region. Nucleosynthesis models predict that Os should be more
 Solar System Abundances 395

abundant than Ir, as seen in CI chondrites. Overall, a new analysis of the

photospheric Os abundance is needed.
Platinum: [75] find that the photospheric Pt abundance is not very reli-
able. The value selected in L03 is kept here, but is assigned a 0.3 dex uncer-
tainty (factor of 2) to emphasize its low reliability.
Thallium: The Tl value of 0.95 ± 0.2 is from the linear average of the
end-member composition of 0.72 ≤ A(Tl) ≤ 1.1 selected in L03 that was found
for sunspot spectra. The uncertainty quoted here is to indicate the derived
range. There are no new measurements.
Thorium: The Th abundance is difficult to determine because the only
accessible Th line in the photospheric spectrum is heavily blended with Ni I
and Ni II. Caffau et al. [73] report a nominal Th abundance A(Th) = 0.08 ±
0.03, which should not be over-interpreted because of the line blends.

4 Recommended Present-Day Solar Abundances

4.1 Cosmochemical and Astronomical Abundance Scale

Conversion

In order to compare the atomic silicon-normalized CI chondrite abundances

in Table 2 (N(Si) = 106 atoms; cosmochemical abundance scale) with the
photospheric abundances on the hydrogen-normalized scale (A(H) = 12; as-
tronomical abundance scale) in Table 3, the data must be converted to a com-
mon scale. One cannot easily convert the meteoritic data to the H-normalized
astronomical abundance scale because H is depleted in meteorites. However,
a comparison can be done for the non-volatile rock-forming elements. The
difference of the logarithmic Si-normalized abundances of CI chondrites to
the abundances on the astronomical scale is more or less constant for many
elements. This shows that the relative abundances in the photosphere and CI
chondrites are similar.
The link for both abundance scales is an average conversion constant that
is calculated by subtracting the logarithm of the Si-normalized meteoritic
abundances (Table 2) from the logarithmic H-normalized photospheric abun-
dances (Table 3) for all elements heavier than neon that have uncertainties
< 0.1 dex, i.e., below ∼25%, in their photospheric abundance determinations.
There are 39 elements that qualify and Figure 3 shows a comparison of the
photospheric and CI chondritic abundances for these elements on a linear scale
(note that the conversion constant for the log scales is equivalent to a scale
factor on linear abundance scales). The scale conversion constant is 1.533 ±
0.042; and the cosmochemical and astronomical scales are coupled as:
A(X) = log N(X) + 1.533
Previously, the conversion constant was somewhat larger. For example,
AG89 used 1.554 ± 0.020 from only 12 elements, which resulted in a smaller
nominal uncertainty of the conversion constant. L03 found a value of 1.539
396 Katharina Lodders

± 0.046 for 35 elements for which photospheric abundances were determined

with less than 25% uncertainty, and used a constant of 1.540, which is exactly
the log of the ratio of Si in the astronomic to the meteoritic scale. The slightly
lower conversion factor found here is the result of the systematic decrease of
the reported photospheric abundance values.

–20% –10% +10% +20%

Pb
80
Ir

Lu Hf
Tm Er
Dy
Eu Gd
Sm
60 Nd
Pr Ce
Atomic Number

La Ba

Pd
Ru
Mo Nb
40 Y Zr
Sr
Nph / NCI
Ge
Cu Ni Co
Mn Fe
Cr V
Ti Sc
20 Ca
K
S
Si P
Mg NaAl

0.8 1.0 1.2 1.4

Atomic Abundance Ratio
Fig. 3. The photospheric/CI chondritic abundance ratios for 39 elements that are
well determined for the solar photosphere. The grey-shaded region shows agreement
within 10%

It is notable that the conversion constant can be derived from a large

range of elements with different properties (e.g., atomic number, mass, first
ionization potential, condensation temperature). The premise in linking the
meteoritic and solar data is that there are no chemical and physical fraction-
ations of the elements (except for the obvious loss of highly volatile elements
 Solar System Abundances 397

from meteorites). The small spread in the conversion factor indicates that
there is basic agreement of solar and meteoritic abundances. There is no ap-
parent dependence of the conversion factor on atomic number, mass or any
other elemental property. In addition, the solar/meteoritic abundance ratios
are independent of the geochemical character of an element, whether it is
lithophile, siderophile or chalcophile, which indicates that any chemical and
physical fractionation of silicates, metal, and sulfides did not affect CI chon-
drite abundances. A reasonable estimate for the uncertainty of the relative
scale of solar and meteoritic abundances is about 10%.

4.2 Comparison of Photospheric and Meteoritic Abundances

The photospheric and CI chondrite abundances on the astronomical abun-

dance scale are given in Table 3. Agreement within 10% for meteoritic and
photospheric data exists for 40 elements (the 39 shown in Figure 3 plus the
light element Be). This increase in elements that show good agreement is
mainly due to the recent improvements in photospheric measurements.
The largest differences are for the highly volatile elements that form low-
temperature ices and/or exist in gaseous form in the terrestrial atmosphere.
The largest depletion is for the noble gases. The depletion sequence for N, C,
H reflects the general lack of solid nitrogen compounds in meteorites and the
predominance of oxides and silicates.
Only Li is clearly consumed in the interior of Sun by nuclear reactions,
which explains the ∼150 times smaller photospheric abundance. Nominally,
B is depleted in the Sun by about 20% but within the stated uncertainties it
is apparently not affected. Beryllium is another fragile element like Li and B,
and may be subject to destruction in the Sun. However, the comparison of
abundances for the photosphere and CI chondrites indicate that there was no
Be loss in the Sun; indeed; the nominal Be abundance for the Sun is higher
than in CI chondrites.
The difference between photospheric and CI chondrite abundances exceeds
10% for 21 other elements (see LPG09 for a detailed comparison). However,
in most cases the combined uncertainties of the photospheric and meteoritic
determinations are larger than the difference in abundance, and solar and
meteoritic abundances agree within error limits. Elements with abundance
differences larger than the combined error bars are W, Rb, Ga, Hf, and Mn.
The abundances of Ga, Rb, and W need to be redetermined in the photo-
sphere to resolve the differences. There are new photospheric analyses for Hf
and Mn suggesting that the differences in photospheric and meteoritic values
could be real since there are no plausible reasons to doubt the results. Line
blending may not be the culprit as this usually leads to over-estimated abun-
dances for the photosphere (e.g., like for Indium as noted above); however,
the photospheric Mn value is ∼1.4x lower than in CI chondrites. Manganese
can be accurately measured in meteorites. Its concentration in Orgueil is sim-
ilar to that in two other CI meteorites, Alais and Ivuna (L03), and it fits
398 Katharina Lodders

with the abundances of other elements of similar volatility. It seems that an

unidentified problem in the photospheric abundance analysis may cause the
discrepancy of the meteoritic and solar Mn abundances.
The problem is reversed for Hf and the meteoritic abundance of Hf is
less than that of the photosphere. This could indicate a problem with the
photospheric abundance determination and suggest that line blending is more
severe than already corrected for in current models. However, two recent Hf
analyses using different models essentially obtain the same abundance, and if
there is a problem with the analysis, it remains elusive. The Hf concentration
in CI chondrites has been accurately determined, because Hf is important for
Lu-Hf and W-Hf dating. The very constant Lu/Hf ratio in meteorites closely
ties Hf to other refractory elements, which do not show large differences in
abundance to the sun as does Hf. This issue awaits resolution.
Overall, the agreement between photospheric and meteoritic abundances
has improved further with new photospheric and meteoritic data.

4.3 Combined Solar Abundances from CI Chondrites

and Photospheric Data

The CI chondritic and photospheric abundances can be combined to select

a set of recommended present-day solar system abundances. Here the same
procedure as in L03 is used to construct such an abundance set. The recom-
mended data are from photospheric values for ultra-highly volatile elements
like H, C, N, and O and from various sources and theory for the noble gases
(see above for He, Ne, Ar, and L03 for Kr and Xe). The CI chondrite data
are the obvious choice for elements that are only determined in CI chondrites
but also for elements that have photospheric abundance determinations with
high uncertainties. Several elements are equally well determined in CI chon-
drites and in the photosphere, and an average of their Si-normalized abun-
dances is used. The recommended present day abundances are converted to
the astronomical scale using the same conversion constant (1.533) between
the astronomical and cosmochemical abundance scales as described before.
Figure 4 shows the recommended abundances of the elements as a func-
tion of atomic number. The large abundances of H and He are not shown to
avoid scale compression in the diagram. Overall, abundances decrease rela-
tively smoothly with increasing atomic number. The regular pattern that odd
numbered elements are less abundant than their even-numbered neighbors
holds from the lightest to the heaviest elements; originally Harkins estab-
lished this observation for the lighter elements up to the Fe region. Notable
exceptions are the low abundances of Li, Be, and B, that consist of fragile
nuclei that are easily destroyed in stellar interiors. The elemental abundance
distribution is not controlled by the chemical properties of the elements but
instead by nuclear properties.
 Solar System Abundances 399

Table 3. Cl chondrite, solar (mainly photosphere), and recommended present day

solar abundances
Cl chondrites Sun recommended
Z A(X) A(X) Note A(X)
1 H 8.24 0.04 12.00 s 12.00
2 He 1.31 10.925 0.02 s,t 10.925 0.02
3 Li 3.28 0.05 1.10 0.10 m 3.28 0.05
4 Be 1.32 0.03 1.38 0.09 m 1.32 0.03
5 B 2.81 0.04 2.70 0.17 m 2.81 0.04
6 C 7.41 0.04 8.39 0.04 s 8.39 0.04
7 N 6.28 0.06 7.86 0.12 s 7.86 0.12
8 O 8.42 0.04 8.73 0.07 s 8.73 0.07
9 F 4.44 0.06 4.56 0.30 m 4.44 0.06
10 Ne -1.10 8.05 0.10 s,t 8.05 0.10
11 Na 6.29 0.02 6.30 0.03 a 6.29 0.04
12 Mg 7.55 0.01 7.54 0.06 a 7.54 0.06
13 Al 6.45 0.01 6.47 0.07 a 6.46 0.07
14 Si 7.53 0.01 7.52 0.06 m 7.53 0.06
15 P 5.45 0.04 5.46 0.04 a 5.45 0.05
16 S 7.17 0.02 7.14 0.01 a 7.16 0.02
17 Cl 5.25 0.06 5.50 0.30 m 5.25 0.06
18 Ar -0.48 6.50 0.10 s,t 6.50 0.10
19 K 5.10 0.02 5.12 0.03 a 5.11 0.04
20 Ca 6.31 0.02 6.33 0.07 m 6.31 0.02
21 Sc 3.07 0.02 3.10 0.10 m 3.07 0.02
22 Ti 4.93 0.03 4.90 0.06 m 4.93 0.03
23 V 3.98 0.02 4.00 0.02 a 3.99 0.03
24 Cr 5.66 0.01 5.64 0.01 a 5.65 0.02
25 Mn 5.50 0.01 5.37 0.05 m 5.50 0.01
26 Fe 7.47 0.01 7.45 0.08 a 7.46 0.08
27 Co 4.89 0.01 4.92 0.08 a 4.90 0.08
28 Ni 6.22 0.01 6.23 0.04 a 6.22 0.04
29 Cu 4.27 0.04 4.21 0.04 m 4.27 0.04
30 Zn 4.65 0.04 4.62 0.15 m 4.65 0.04
31 Ga 3.10 0.02 2.88 0.10 m 3.10 0.02
32 Ge 3.60 0.04 3.58 0.05 a 3.59 0.06
33 As 2.32 0.04 m 2.32 0.04
34 Se 3.36 0.03 m 3.36 0.03
35 Br 2.56 0.06 m 2.56 0.06
36 Kr -2.25 3.28 0.08 t 3.28 0.08
37 Rb 2.38 0.03 2.60 0.10 m 2.38 0.03
38 Sr 2.90 0.03 2.92 0.05 m 2.90 0.03
39 Y 2.19 0.04 2.21 0.02 a 2.20 0.04
40 Zr 2.55 0.04 2.58 0.02 a 2.57 0.04
400 Katharina Lodders

Table 3. – Continued
Cl chondrites Sun recommended
Z A(X) A(X) Note A(X)
41 Nb 1.43 0.04 1.42 0.06 a 1.42 0.07
42 Mo 1.96 0.04 1.92 0.05 a 1.94 0.06
44 Ru 1.78 0.03 1.84 0.07 m 1.78 0.03
45 Rh 1.08 0.04 1.12 0.12 a 1.10 0.13
46 Pd 1.67 0 .02 1.66 0.04 a 1.67 0.04
47 Ag 1.22 0.02 0.94 0.30 m 1.22 0.02
48 Cd 1.73 0.03 1.77 0.11 m 1.73 0.03
49 In 0.78 0.03 1.50 UL m 0.78 0.03
50 Sn 2.09 0.06 2.00 0.30 m 2.09 0.06
51 Sb 1.03 0.06 1.00 0.30 m 1.03 0.06
52 Te 2.20 0.03 m 2.20 0.03
53 I 1.57 0.08 m 1.57 0.08
54 Xe -1.93 2.27 0.08 t 2.27 0.08
55 Cs 1.10 0.02 m 1.10 0.02
56 Ba 2.20 0.03 2.17 0.07 a 2.18 0.07
57 La 1.19 0.02 1.14 0.03 m 1.19 0.02
58 Ce 1.60 0.02 1.61 0.06 a 1.60 0.06
59 Pr 0.78 0.03 0.76 0.04 a 0 .77 0.05
60 Nd 1.47 0.02 1.45 0.05 m 1.47 0.02
62 Sm 0.96 0.02 1.00 0.05 m 0.96 0.02
63 Eu 0.53 0.02 0.52 0.04 a 0.53 0.04
64 Gd 1.07 0.02 1.11 0.05 a 1.09 0.06
65 Tb 0.34 0.03 0.28 0.10 m 0.34 0.03
66 Dy 1.15 0.02 1.13 0.06 a 1.14 0.06
67 Ho 0.49 0.03 0.51 0.10 m 0.49 0.03
68 Er 0.94 0.02 0.96 0.06 a 0.95 0.06
69 Tm 0.14 0.03 0.14 0.04 m 0.14 0.03
70 Yb 0.94 0.02 0.86 0.10 m 0.94 0.02
71 Lu 0.11 0 .02 0.12 0.08 m 0.11 0.02
72 Hf 0.73 0.02 0.88 0.08 m 0.73 0.02
73 Ta -0.14 0.04 m -0.14 0.04
74 W 0 .67 0.04 1.11 0.15 m 0.67 0.04
75 Re 0.28 0.04 m 0.28 0.04
76 Os 1.37 0.03 1.45 0.11 m 1.37 0.03
77 Ir 1.34 0.02 1.38 0.05 a 1.36 0.06
78 Pt 1.64 0.03 1.74 0.30 m 1.64 0.03
79 Au 0.82 0.04 1.01 0.18 m 0.82 0.04
80 Hg 1.19 0.08 m 1 .19 0.08
81 Tl 0.79 0.03 0.95 0.20 m 0.79 0.03
82 Pb 2.06 0.03 2.00 0.06 m 2.06 0.03
83 Bi 0.67 0.04 m 0.67 0.04
90 Th 0.08 0.03 0.08 UL m 0.08 0.03
92 U -0.52 0.03 -0.47 UL m -0.52 0.03
Abundances on the astronomical scale with log N(H)=12.
Note: a = average of meteoritic and solar value; m = meteoritic value;
t = theoretical and/or indirectly determined.
UL: upper limit
 Solar System Abundances 401

108
O
107 C
Ne
Mg Si Fe
Present-day Solar System Composition
6
10 N S (H, He not shown)
Atomic Abundance, Si = 106

Ar
105 Ca Ni
Al
Na Cr
104
P Ti Mn
Cl Zn
K
103 Co
F
Cu Ge
102 Li V Se Kr
Sr
B Sc Ga Zr
101 SnTe
Xe
Ba Pb
Br Mo
As Rb RuPdCd Pt
0 Y Ce
10 Nd Os
Be Gd Dy Yb Hg
Nb I Sm Er Hf
Ag La W Ir
Rh Sb Cs
10–1 In Pr Au Tl Bi Th
Eu Ho
Tb Re
–2 Tm Lu U
10 Ta

10 20 30 40 50 60 70 80 90
Atomic Number, Z

Fig. 4. Abundances of the elements as a function of atomic number

4.4 Mass Fractions X, Y, and Z in Present-Day Solar Material

Many applications in planetary sciences and astronomy use mass fractions of

the elements rather than atomic abundances that we have dealt with so far.
Mass fractions are also involved when the He abundance is to be derived. Al-
though a value for He is listed in the tables above, the He abundance cannot be
derived from the meteoritic nor the photospheric analyses. Thus, at this point
one only has the atomic abundances of all elements relative to H or Si, except
for He. Using the atomic weights of the elements, the relative atomic elemental
abundances are converted to mass concentration ratios. Then the ratio of the
mass sum of all heavy elements relative to the mass of H is obtained, which
is needed to derive the He mass fraction from different constraints, and from
that we finally obtain the atomic He abundance.
402 Katharina Lodders

The mass fraction of H is usually abbreviated as X, that of He as Y, and

the sum of the mass fractions of all other heavy elements as Z. The overall sum
of these mass fractions is X+Y+Z=1. Absolute mass fractions of X, Y, and Z
can be derived if the ratio of Z/X is known from atomic abundance analysis
(the Z/X ratio can always be computed without knowing the He abundance),
and if either the mass fraction of H or He is known independently.
The mass fraction of He can be inferred from inversion of helioseismic data
by matching the sound speeds of H and He dominated mixtures under phys-
ical conditions appropriate for the solar convection zone. The mass fraction
Z, which combines all other heavy elements is also important as it governs
opacities and thus the density structure of the Sun’s outer convection zone.
The depth of the solar convection zone derived from helioseismic data poses
constraints on the permissible fraction of heavy elements (a detailed review
on helioseismology and the He abundance problem is given by [30]). The he-
lioseismic inversion models require Z/X ratios and heavy element abundances
(for opacities) as inputs. Therefore the He mass fraction and the He abun-
dance from such models is not independent of X/Z. Ideally, one would use the
abundance data and X/Z from the new compilation here to find the corre-
sponding He abundance from helioseismic models and fits to solar data. One
should not necessarily adopt a He abundance that is based on models that are
calibrated to different Z/X than found for the new compilation of elemental
abundances.
One can also obtain the absolute fractions of X, Y, and Z if the Z/X and the
X (hydrogen mass fraction) are known. [76,30] have shown that the estimated
mass fraction of H from helioseismic models is relatively independent of Z/X
ratios in the range of 0.0171 < Z/X < 0.0245. Their models calibrated to
Z/X=0.0171 and 0.0218 yield an average X=0.7389 ± 0.0034 [76]. If the H
mass fraction is indeed independent of the Z/X ratio, and if compositional
variations within Z (mainly governed by the mass fractions of O, C, Ne, see
below) also do not alter this conclusion much, we can use this X to estimate
the He mass fraction.
With Z/X = 0.0191 found from the abundances in Table 3, and assuming
X = 0.739, one obtains Z = 0.0141, and from this, Y = 1−X−Z = 0.2469.
This mass fraction of He corresponds to an atomic He abundance of A(He) =
10.925.
The mass fraction of heavy elements (Z=0.014) is intermediate to those
in the compilations by GS98 (Z=0.017) and G07 (Z=0.012); see Table 4 for
a comparison for present-day solar values. The He mass fraction Y is smaller
than that in previous compilations, but the (rounded) He abundance is the
same for the three compilations in Table 4. The hydrogen mass fraction of
X=0.739 [76] adopted here is essentially the same as in [15, 16], and the
smaller value for [12] seems to be due to the different model assumptions for
deriving the He abundance there.
 Solar System Abundances 403

Table 4. Protosolar mass fractions and He abundance

Present-Day Z/X X Y Z A(He)
this work 0.0191 0.7390 0.2469 0.0141 10.925
A05, G07 0.0165 0.7383 0.2495 0.0122 10.93
GS98 0.0231 0.7347 0.2483 0.0169 10.93
Note: AG89 is not included because they do not give the present-day
He abundance or Z/X. For protosolar values see below.

Table 5. Concentration of present-day solar composition (mass %)

this work A05,G07 GS98
H (=X) 73.90 73.92 73.47
He (=Y) 24.69 24.86 24.83
O 0.63 0.54 0.79
C 0.22 0.22 0.29
Ne 0.17 0.10 0.18
Fe 0.12 0.12 0.13
N 0.07 0.06 0.08
Si 0.07 0.07 0.07
Mg 0.06 0.06 0.07
S 0.03 0.03 0.05
all other elements 0.04 0.02 0.04
total heavy elements (=Z) 1.41 1.22 1.69
Note: Elements in order of decreasing concentration by mass.

Overall, the mass fractions derived here are closer to the ∼10 year old GS98
compilation than to the most recent ones by L03, A05 or G07. Ideally, the He
abundance proposed here needs to be evaluated with results from helioseismic
models calibrated to the abundances of the elements (other than He) and the
X/Z ratio found here.
The fraction for Z obtained here is higher than in the compilations by
e.g., L03, A05, G07. The mass increase in Z should help to resurrect the
standard solar models, which agreed with helioseismic constraints when the
GS98 abundances were used, but crumbled under the too low Z values that
were suggested in more recent compilations (see review by [30]). Table 5
compares the mass fractions of the most abundant elements in present-day
solar material (note that the order by mass is different from that by atomic
abundances). About half of the mass fraction of Z is from O, followed by C, Ne,
and Fe. The higher Z here comes mainly from increased O and Ne abundances,
which may help to eradicate the problem with the incompatibility of standard
solar models and recommended present-day solar abundances.
404 Katharina Lodders

5 Solar System Abundances 4.56 Gyr Ago

The data discussed above are for present-day abundances in the photosphere
and meteorites. However, two processes affected the solar abundances over
time. The first is element settling from the solar photosphere into the Sun’s
interior; the second is decay of radioactive isotopes that contribute to the
overall atomic abundance of an element. The first, discussed in the following,
is more important for the sun and large-scale modeling; the changes in iso-
topic compositions and their effects on abundances are comparably minor but
important for radiometric dating. The isotopic effects are considered in the
solar system abundance table in this section, but are not described at length
here.
Settling or diffusion of heavy elements from the photosphere to the inte-
rior boundary layer of the convection zone and beyond lowered the elemental
abundances (relative to H) from protosolar values 4.56 Gyr ago (see [30]). Over
the Sun’s lifetime, diffusion decreased abundances of elements heavier than He
by ∼13% from original protosolar values, whereas that of He dropped a little
more by about ∼15%; modeling these depletions also depends on opacities,
hence abundances. With these estimates, the proto-solar abundances (sub-
script 0) are calculated from the present-day data for the astronomical scales
as
A(He)0 = A(He) + 0.061,

and for all elements heavier than He it is

A(X)0 = A(X) + 0.053 = log N(X) + 1.586

The atomic abundances of the elements on both abundance scales are given
in Table 6. Note that on the cosmochemical abundance scale (N(Si)=106 ),
the relative abundances of the heavy elements do not change from the data
in Table 3 because the scale is normalized to Si, one of the heavy elements.
Only H and He change from present-day values: the relative H abundance is
less, and the He abundance is slightly higher (because of the higher diffusive
loss).
For completeness, the protosolar mass fractions X, Y, and Z are summa-
rized in Table 7. However, as noted before for the present day solar mass
fractions, it is up to the solar models and helioseismology to derive the best-
fitting current and proto solar He mass fractions from the given abundances
of the other elements.

6 Abundance of the Nuclides

The abundance of an element is determined by the number and abundances
of its stable isotopes, which in turn depends on the stability of the nuclei dur-
ing thermonuclear reactions in stellar interiors. Already in the 1910s, Harkins
 Solar System Abundances 405

Table 6. Solar system abundances 4.56 Gyr ago

Z N(Si)=106 σ log N(H)=12 σ
1 H 2.59E+10 12.00
2 He 2.51E+09 1.2E+08 10.986 0.02
3 Li 55.6 7.2 3.33 0.05
4 Be 0.612 0.043 1.37 0.03
5 B 18.8 1.9 2.86 0.04
6 C 7.19E+06 6.9E+05 8.44 0.04
7 N 2.12E+06 6.8E+05 7.91 0.12
8 O 1.57E+07 2.8E+06 8.78 0.07
9 F 804 121 4.49 0.06
10 Ne 3.29E+06 8.5E+05 8.10 0.10
11 Na 57700 5100 6.35 0.04
12 Mg 1.03E+06 1.5E+05 7.60 0.06
13 Al 84600 15300 6.51 0.07
14 Si 1.00E+06 2E+04 7.59 0.08
15 P 8300 1100 5.51 0.05
16 S 4.21E+05 2.4E+04 7.21 0.02
17 Cl 5170 780 5.30 0.06
18 Ar 92700 24000 6.55 0.10
19 K 3760 330 5.16 0.04
20 Ca 60400 3000 6.37 0.02
21 Sc 34.4 1.7 3.12 0.02
22 Ti 2470 200 4.98 0.03
23 V 286 20 4.04 0.03
24 Cr 13100 500 5.70 0.02
25 Mn 9220 280 5.55 0.01
26 Fe 8.48E+05 1.69E+05 7.51 0.08
27 Co 2350 500 4.96 0.08
28 Ni 49000 5000 6.28 0.04
29 Cu 541 54 4.32 0.04
30 Zn 1300 130 4.70 0.04
31 Ga 36.6 1.8 3.15 0.02
32 Ge 115 18 3.65 0.06
33 As 6.10 0.55 2.37 0.04
34 Se 67.5 4.7 3.42 0.03
35 Br 10.7 1.6 2.62 0.06
36 Kr 55.8 11.3 3.33 0.08
37 Rb 7.23 0.51 2.45 0.03
38 Sr 23.3 1.6 2.95 0.03
39 Y 4.63 0.50 2.25 0.04
40 Zr 10.8 1.2 2.62 0.04
406 Katharina Lodders

Table 6. – Continued
Z N(Si)=106 σ log N(H)=12 σ
41 Nb 0.780 0.139 1.48 0.07
42 Mo 2.55 0.40 1.99 0.06
44 Ru 1.78 0.11 1.84 0.03
45 Rh 0.370 0.128 1.15 0.13
46 Pd 1.36 0.15 1.72 0.04
47 Ag 0.489 0.024 1.28 0.02
48 Cd 1.57 0.11 1.78 0.03
49 In 0.178 0.012 0.84 0.03
50 Sn 3.60 0.54 2.14 0.06
51 Sb 0.313 0.047 1.08 0.06
52 Te 4.69 0.33 2.26 0.03
53 I 1.10 0.22 1.63 0.08
54 Xe 5.46 1.10 2.32 0.08
55 Cs 0.371 0.019 1.16 0.02
56 Ba 4.47 0.81 2.24 0.07
57 La 0.457 0.023 1.25 0.02
58 Ce 1.18 0.19 1.66 0.06
59 Pr 0.172 0.020 0.82 0.05
60 Nd 0.856 0.043 1.52 0.02
62 Sm 0.267 0.013 1.01 0.02
63 Eu 0.10 0.01 0.58 0.04
64 Gd 0.360 0.049 1.14 0.06
65 Tb 0.06 0.00 0.39 0.03
66 Dy 0.404 0.062 1.19 0.06
67 Ho 0.09 0.01 0.55 0.03
68 Er 0.262 0.042 1.00 0.06
69 Tm 0.04 0.00 0.19 0.03
70 Yb 0.256 0.013 0.99 0.02
71 Lu 0.0380 0.0019 0.17 0.02
72 Hf 0.156 0.008 0.78 0.02
73 Ta 0.0210 0.0021 -0.09 0.04
74 W 0.137 0.014 0.72 0.04
75 Re 0.0581 0.0058 0.35 0.04
76 Os 0.678 0.054 1.42 0.03
77 Ir 0.672 0.092 1.41 0.06
78 Pt 1.27 0.10 1.69 0.03
79 Au 0.195 0.019 0.88 0.04
80 Hg 0.458 0.092 1.25 0.08
81 Tl 0.182 0.015 0.85 0.03
82 Pb 3.31 0.23 2.11 0.03
83 Bi 0.138 0.012 0.73 0.04
90 Th 0.0440 0.0035 0.23 0.03
92 U 0.0238 0.0019 -0.04 0.03
 Solar System Abundances 407

Table 7. Protosolar mass fractions and He abundance

Z0 /X0 X0 Y0 Z0 A(He)0
this work(a) 0.0215 0.7112 0.2735 0.0153 10.986
A05, G07(b) 0.0185 0.7133 0.2735 0.0132 10.985
GS98(c) 0.0231 0.7086 0.2750 0.0163 10.99
AG89(d) 0.0267 0.7068 0.2743 0.0189 10.99
(a)
present-day to protosolar values are converted using (A(He)0 = A(He) + 0.061,
all other elements (except H) from A(X)0 = A(X) + 0.053

(b)
G07 suggest A(He)0 = A(He)+0.057, all other elements (except H) A(X)0 =
A(X) + 0.05.

(c)
GS98: Changes in Z due to diffusion were not assumed (Z/X = Z0 /X0 );
a ∼ 10% loss of He from the photosphere was considered.

(d)
Changes in Z due to diffusion were not assumed (Z/X = Z0 /X0 )

and Oddo made the observation that elements with even atomic numbers are
more abundant than their odd-numbered neighbors, which finds its explana-
tion in the nuclear properties of the elements (see also Figure 4). An element
is defined by its atomic number (Z), which is the number of positively charged
nucleons (=protons) in its atoms. Atoms belonging to the same element may
have different atomic masses due to a different number of neutral nucleons
(=neutrons, neutron number N). In 1913, Soddy coined the term isotope for
atoms with the same proton number but different neutron numbers after the
Greek isos-topos meaning at the same place in the periodic table of the ele-
ments. The term isotope is in specific reference to a given element; whereas in
a discussion of properties of atomic nuclei of different elements (with different
Z and N) the generic term nuclide is usually more appropriate. However, often
the terms nuclide and isotope are used as inter-exchangeable.
The sum of the number of protons (Z) and neutrons (N) is referred to as
mass number A = Z+N. The mass number A is usually used when the nuclide
abundance distributions are discussed, which is analogous to using the proton
or atomic number Z for discussing elemental distributions.
There are 280 naturally occurring nuclides that make up the 83 stable and
long-lived elements. These are all the elements up to Bi with Z = 83, except
for unstable Tc (Z = 43) and Pm (Z = 61) that only have short-lived isotopes,
but the long-lived Th and U bring the total back to 83. Here long-lived or
short-lived is with respect to the half-life of an isotope against radioactive
decay and the age of the solar system. Long-lived means then an element is
still present in measurable quantities since the solar system formed 4.6 Gyr
ago, and radioactive isotopes with half-lives above ∼0.6 Gyr usually qualify
408 Katharina Lodders

for this. Of the 280 nuclides, 266 are stable, and 14 have large half-lives such
as 40 K, 232 Th, 235 U, 238 U, that find practical use in radiometric age dating of
terrestrial and extraterrestrial samples, nuclear power reactors, and weaponry.
Considering only the atomic number, one finds that of the 83 elements,
43 have even Z, and 40 odd Z (note that Tc and Pm with only short-lived
isotopes have odd atomic numbers, but Th and U have even ones), which
reflects the higher stability of an atomic nucleus with even number of protons.
This extends further to nuclei that also have an even number of neutrons. The
proton and neutron numbers in the nuclei of the 266 stable nuclides lead to
the following groupings:
Z even, N even: 159 nuclides
Z even, N odd: 53 nuclides
Z odd, N even: 50 nuclides
Z odd, N odd: 4 nuclides (2 H, 6 Li, 10 B, 14 N)
Since an element’s abundance is the sum of the abundances of the element’s
isotopes, a lower number of odd-Z numbered nuclides (50+4) than even-Z
(159+53) means that there is a lower abundance of odd-Z elements. This is a
simple explanation for the odd-even abundance distribution noted by [77].
The mass numbers of the stable and long-lived nuclides range from A=1
(1 H) to 209 (209 Bi) except for gaps at A=5 and 8. After 209 Bi we only have
longer-lived nuclides of the actinides Th and U with the mass numbers 232,
235, and 238. Several nuclides have the same mass number but are isotopes of
different elements, simply because A is given by Z+N. In comparisons of the
nuclide distributions as function of mass number, the nuclides with the same
A (isobaric nuclides, or isobars) are often summed up.
Table 8 summarizes the nuclide abundances and Figure 5 shows the abun-
dance distribution of the nuclides as a function of mass number at the time of
solar system formation 4.56 Ga ago. Figure 5 shows that nuclides with even
A (shown as closed symbols) have usually higher abundances than the odd
numbered nuclides (open symbols). Further, the odd numbered nuclides plot
parallel to the even numbered A in a somewhat smoother distribution curve.
This behavior of nuclide distribution with mass number compares well to the
behavior of elemental distribution with atomic number (Figure 4).
Abundances peak at mass numbers for closed proton and neutron shells.
These nuclear shells are analogous to the closed electron shells that character-
ize atomic properties. The magic numbers for nuclear stability are 2, 8, 20, 28,
50, 82, and 126; and nuclides with Z and/or N equal to these magic numbers
are the ones that show large abundances in the diagram of abundance versus
mass number (A=Z+N). This is particularly notable for the light doubly-
magic nuclei with equal magic Z and N, e.g., 4 He (Z=N=2), 16 O (Z=N=8),
and 40 Ca (Z=N=20). Beyond the region of nuclides with mass numbers of
56 (the Fe-peak region), abundances decline more or less smoothly and spike
at certain mass number regions. The nuclides beyond the Fe peak are prod-
ucts from neutron-capture processes. The peaks in the distribution correspond
to regions where either nuclides are preferentially made by the slow-neutron
 Solar System Abundances 409

capture (s-) process operating in red giant stars (e.g, Y and Ba regions) or by
the rapid-neutron capture (r-) process probably operating in supernovae (e.g.,
Pt region); see, e.g.,[78,79,80] for reviews on stellar nucleosynthesis. Here the
slow and rapid are in reference to beta-decay timescales of the intermediate,
unstable nuclei produced during the neutron-capture processes. The nuclide
yields from these processes depend on the neutron energies and flux, but also
on the abundance and stability of the target nuclei against neutron-capture
which in turn depends on Z and N. Hence the abundance distribution becomes
controlled by the more stable magic nuclides that serve as bottlenecks for the
overall yields in the neutron capture processes.

1011
1
H
1010 101 89
Y odd mass numbers

109
137 195
100 Ba Pt
108 12
16
O 139
La
C
20
107 Ne
abundance, Si = 106 atoms

56
Fe 10–1
117, 119
106 Sn
40
Ca 169
105 10
–2 Tm
80 100 120 140 160 180 200
104
103
102
138
Ba 208
101 195
Pb
Pt
100 89
164
Dy
9
B Y
10–1
117, 119
10–2 Sn 169
Tm
10–3
0 20 40 60 80 100 120 140 160 180 200 220 240
mass number, A
Fig. 5. Solar system abundances of nuclides 4.56 Gyr ago (see LPG09)
410 Katharina Lodders

Table 8. Solar system nuclide abundances 4.56 Gyr ago

Z A atom% N Z A atom% N
1 H 1 99.9981 2.59E+10 15 P 31 100 8300
1 H 2 0.00194 5.03E+05
100 2.59E+10 16 S 32 95.018 400258
2 He 3 0.0166 1.03E+06 16 S 33 0.75 3160
2 He 4 99.9834 2.51E+09 16 S 34 4.215 17800
100 2.51E+09 16 S 36 0.017 72
3 Li 6 7.589 4.2 100 421245
3 Li 7 92.411 51.4 17 Cl 35 75.771 3920
100 55.6 17 Cl 37 24.229 1250
4 Be 9 100 0.612 100 5170
18 Ar 36 84.595 78400
5 B 10 19.820 3.7 18 Ar 38 15.381 14300
5 B 11 80.180 15.1 18 Ar 40 0.024 22
100 18.8 100 92700
6 C 12 98.889 7.11E+06 19 K 39 93.132 3500
6 C 13 1.111 7.99E+04 19 K* 40 0.147 6
100 7.19E+06 19 K 41 6.721 253
7 N 14 99.634 2.12E+06 100 3760
7 N 15 0.366 7.78E+03 20 Ca 40 96.941 58500
100 2.12E+06 20 Ca 42 0.647 391
8 O 16 99.763 1.57E+07 20 Ca 43 0.135 82
8 O 17 0.037 5.90E+03 20 Ca 44 2.086 1260
8 O 18 0.200 3.15E+04 20 Ca 46 0.004 2
100 1.57E+07 20 Ca 48 0.187 113
9 F 19 100 804 100 60400
21 Sc 45 100 34.4
10 Ne 20 92.9431 3.06E+06
10 Ne 21 0.2228 7.33E+03 22 Ti 46 8.249 204
10 Ne 22 6.8341 2.25E+05 22 Ti 47 7.437 184
100 3.29E+06 22 Ti 48 73.72 1820
11 Na 23 100 57700 22 Ti 49 5.409 134
22 Ti 50 5.185 128
12 Mg 24 78.992 8.10E+05 100 2470
12 Mg 25 10.003 1.03E+05 23 V 50 0.2497 0.7
12 Mg 26 11.005 1.13E+05 23 V 51 99.7503 285.7
100 1.03E+06 100 286.4
13 Al 27 100 8.46E+04 24 Cr 50 4.3452 569
14 Si 28 92.230 9.22E+05 24 Cr 52 83.7895 11000
14 Si 29 4.683 4.68E+04 24 Cr 53 9.5006 1240
14 Si 30 3.087 3.09E+04 24 Cr 54 2.3647 309
100 1.00E+06 100 13100
 Solar System Abundances 411

Table 8. – Continued
Z A atom% N Z A atom% N
25 Mn 55 100 9220 35 Br 79 50.686 5.43
35 Br 81 49.314 5.28
26 Fe 54 5.845 49600 100 10.7
26 Fe 56 91.754 7.78E+05 36 Kr 78 0.362 0.20
26 Fe 57 2.1191 18000 36 Kr 80 2.326 1.30
26 Fe 58 0.2819 2390 36 Kr 82 11.655 6.51
100 8.48E+05 36 Kr 83 11.546 6.45
27 Co 59 100 2350 36 Kr 84 56.903 31.78
36 Kr 86 17.208 9.61
28 Ni 58 68.0769 33400 100 55.8
28 Ni 60 26.2231 12900 37 Rb 85 70.844 5.121
28 Ni 61 1.1399 559 37 Rb* 87 29.156 2.108
28 Ni 62 3.6345 1780 100 7.23
28 Ni 64 0.9256 454 38 Sr 84 0.5580 0.13
100 49000 38 Sr 86 9.8678 2.30
29 Cu 63 69.174 374 38 Sr 87 6.8961 1.60
29 Cu 65 30.826 167 38 Sr 88 82.6781 19.2
100 541 100 23.3
30 Zn 64 48.63 630 39 Y 89 100 4.63
30 Zn 66 27.9 362
30 Zn 67 4.1 53 40 Zr 90 51.452 5.546
30 Zn 68 18.75 243 40 Zr 91 11.223 1.210
30 Zn 70 0.62 8 40 Zr 92 17.146 1.848
100 1300 40 Zr 94 17.38 1.873
31 Ga 69 60.108 22.0 40 Zr 96 2.799 0.302
31 Ga 71 39.892 14.6 100 10.78
100 36.6 41 Nb 93 100 0.780
32 Ge 70 21.234 24.3
32 Ge 72 27.662 31.7 42 Mo 92 14.525 0.370
32 Ge 73 7.717 8.8 42 Mo 94 9.151 0.233
32 Ge 74 35.943 41.2 42 Mo 95 15.838 0.404
32 Ge 76 7.444 8.5 42 Mo 96 16.672 0.425
100 115 42 Mo 97 9.599 0.245
33 As 75 100 6.10 42 Mo 98 24.391 0.622
42 Mo 100 9.824 0.250
34 Se 74 0.89 0.60 100 2.55
34 Se 76 9.37 6.32 44 Ru 96 5.542 0.099
34 Se 77 7.64 5.15 44 Ru 98 1.869 0.033
34 Se 78 23.77 16.04 44 Ru 99 12.758 0.227
34 Se 80 49.61 33.48 44 Ru 100 12.599 0.224
34 Se 82 8.73 5.89 44 Ru 101 17.060 0.304
412 Katharina Lodders

Table 8. – Continued
Z A atom% N Z A atom% N
44 Ru 102 31.552 0.562 52 Te 120 0.096 0.005
44 Ru 104 18.621 0.332 52 Te 122 2.603 0.122
100 1.78 52 Te 123 0.908 0.043
45 Rh 103 100 0.370 52 Te 124 4.816 0.226
52 Te 125 7.139 0.335
46 Pd 102 1.02 0.0139 52 Te 126 18.952 0.889
46 Pd 104 11.14 0.1513 52 Te 128 31.687 1.486
46 Pd 105 22.33 0.3032 52 Te 130 33.799 1.585
46 Pd 106 27.33 0.371 100 4.69
53 I 127 100 1.10
46 Pd 108 26.46 0.359
46 Pd 110 11.72 0.159 54 Xe 124 0.129 0.007
100 1.36 54 Xe 126 0.112 0.006
47 Ag 107 51.839 0.254 54 Xe 128 2.23 0.122
47 Ag 109 48.161 0.236 54 Xe 129 27.46 1.499
100 0.489 54 Xe 130 4.38 0.239
48 Cd 106 1.25 0.020 54 Xe 131 21.80 1.190
48 Cd 108 0.89 0.014 54 Xe 132 26.36 1.438
48 Cd 110 12.49 0.197 54 Xe 134 9.66 0.527
48 Cd 111 12.8 0.201 54 Xe 136 7.87 0.429
48 Cd 112 24.13 0.380 100 5.46
48 Cd 113 12.22 0.192 55 Cs 133 100 0.371
48 Cd 114 28.73 0.452
48 Cd 116 7.49 0.118 56 Ba 130 0.106 0.005
100 1.57 56 Ba 132 0.101 0.005
49 In 113 4.288 0.008 56 Ba 134 2.417 0.108
49 In 115 95.712 0.170 56 Ba 135 6.592 0.295
100 0.178 56 Ba 136 7.853 0.351
50 Sn 112 0.971 0.035 56 Ba 137 11.232 0.502
50 Sn 114 0.659 0.024 56 Ba 138 71.699 3.205
50 Sn 115 0.339 0.012 100 4.471
50 Sn 116 14.536 0.524 57 La* 138 0.091 0.000
50 Sn 117 7.676 0.277 57 La 139 99.909 0.457
50 Sn 118 24.223 0.873 100 0.457
50 Sn 119 8.585 0.309 58 Ce 136 0.186 0.002
50 Sn 120 32.593 1.175 58 Ce 138 0.250 0.003
50 Sn 122 4.629 0.167 58 Ce 140 88.450 1.043
50 Sn 124 5.789 0.209 58 Ce 142 11.114 0.131
100 3.60 100 1.180
51 Sb 121 57.213 0.179 59 Pr 141 100 0.172
51 Sb 123 42.787 0.134
100 0.313
 Solar System Abundances 413

Table 8. – Continued
Z A atom% N Z A atom% N
60 Nd 142 27.044 0.231 68 Er 166 33.503 0.088
60 Nd 143 12.023 0.103 68 Er 167 22.869 0.060
60 Nd 144 23.729 0.203 68 Er 168 26.978 0.071
60 Nd 145 8.763 0.075 68 Er 170 14.910 0.039
60 Nd 146 17.130 0.147 100 0.262
60 Nd 148 5.716 0.049 69 Tm 169 100 0.0406
60 Nd 150 5.596 0.048
100 0.856 70 Yb 168 0.12 0.0003
62 Sm 144 3.073 0.008 70 Yb 170 2.98 0.0076
62 Sm* 147 14.993 0.041 70 Yb 171 14.09 0.0361
62 Sm* 148 11.241 0.030 70 Yb 172 21.69 0.0556
62 Sm 149 13.819 0.037 70 Yb 173 16.10 0.0413
62 Sm 150 7.380 0.020 70 Yb 174 32.03 0.0821
62 Sm 152 26.742 0.071 70 Yb 176 13.00 0.0333
62 Sm 154 22.752 0.060 100 0.256
100 0.267 71 Lu 175 97.1795 0.0370
63 Eu 151 47.81 0.0471 71 Lu* 176 2.8205 0.0011
63 Eu 153 52.19 0.0514 100 0.0380
100 0.0984 72 Hf 174 0.162 0.0003
64 Gd 152 0.203 0.0007 72 Hf 176 5.206 0.0081
64 Gd 154 2.181 0.0078 72 Hf 177 18.606 0.0290
64 Gd 155 14.800 0.0533 72 Hf 178 27.297 0.0425
64 Gd 156 20.466 0.0736 72 Hf 179 13.629 0.0212
64 Gd 157 15.652 0.0563 72 Hf 180 35.100 0.0547
64 Gd 158 24.835 0.0894 100 0.156
64 Gd 160 21.864 0.0787 73 Ta 180 0.0123 2.6E-06
100 0.360 73 Ta 181 99.9877 0.0210
65 Tb 159 100 0.0634 100 0.0210
74 W 180 0.120 0.0002
66 Dy 156 0.056 0.0002 74 W 182 26.499 0.0363
66 Dy 158 0.095 0.0004 74 W 183 14.314 0.0196
66 Dy 160 2.329 0.0094 74 W 184 30.642 0.0420
66 Dy 161 18.889 0.0762 74 W 186 28.426 0.0390
66 Dy 162 25.475 0.1028 100 0.137
66 Dy 163 24.896 0.1005 75 Re 185 35.662 0.0207
66 Dy 164 28.260 0.1141 75 Re* 187 64.338 0.0374
100 0.404 100 0.0581
67 Ho 165 100 0.0910 76 Os 184 0.020 0.0001
76 Os 186 1.598 0.0108
68 Er 162 0.139 0.0004 76 Os 187 1.271 0.0086
68 Er 164 1.601 0.0042 76 Os 188 13.337 0.0904
414 Katharina Lodders

Table 8. – Continued
Z A atom% N Z A atom% N
76 Os 189 16.261 0.110 80 Hg 200 23.10 0.106
76 Os 190 26.444 0.179 80 Hg 201 13.18 0.060
76 Os 192 41.070 0.278 80 Hg 202 29.86 0.137
100 0.678 80 Hg 204 6.87 0.031
77 Ir 191 37.272 0.250 100 0.458
77 Ir 193 62.728 0.421 81 Tl 203 29.524 0.054
100 0.672 81 Tl 205 70.476 0.129
78 Pt* 190 0.014 0.0002 100 0.182
78 Pt 192 0.783 0.010 82 Pb 204 1.997 0.066
78 Pt 194 32.967 0.420 82 Pb 206 18.582 0.614
78 Pt 195 33.832 0.431 82 Pb 207 20.563 0.680
78 Pt 196 25.242 0.322 82 Pb 208 58.858 1.946
78 Pt 198 7.163 0.091 100 3.306
100 1.27 83 Bi 209 100 0.1382
79 Au 197 100 0.195 90 Th* 232 100 0.0440
80 Hg 196 0.15 0.001 92 U* 234 0.002 4.9E-07
80 Hg 198 9.97 0.046 92 U* 235 24.286 0.0058
80 Hg 199 16.87 0.077 92 U* 238 75.712 0.0180
100 0.0238

Only a few nuclides beyond the Fe-group are exclusively produced by the s-
or r- process; most nuclides have varying abundance contributions from both
processes. If the contribution from each process for each isotope is known,
the overall contribution of the r- and s- process to the elemental abundance
can be estimated. The review on heavy element synthesis by [80] includes a
recent table on the r- and s- contributions to each element. A small number
of proton-rich nuclides cannot be produced by the r- and s- process and are
produced instead by the p-process, which probably involves neutrino induced
disintegration of heavier nuclides. Like the r- process, the p- process is not
yet completely understood. However, except for Mo, where p-process isotopes
contribute about 25% to the elemental abundance, the contribution from p-
process nuclides to overall elemental abundances is usually quite small.
Table 8 lists the percent contribution of the isotope(s) for each element,
and the atomic abundance relative to 106 silicon atoms at the time of solar
system formation.
The abundances of radioactive isotopes (indicated by a star next to the
element symbol) are adjusted accordingly. Table 8 is an update to the Ta-
ble in L03, and includes several revisions of isotopic compositions, e.g., for
Mo [81],Dy [82, 83],Er [84],Yb [85],and Lu [86].

Acknowledgements I thank Bruce Fegley and Herbert Palme for discussions

and comments.
 Solar System Abundances 415

References
1. B. Edvardsson, J. Andersen, B. Gustafsson, D. L. Lambert, P. E. Nissen,
J. Tomkin, A&A 275, 101 (1993)
2. B. Nordstrom, M. Mayor, J. Andersen, J. Holmberg, F. Pont, B. R.
Jorgensen, E. H. Olsen, S. Udry, N. Mowlavi, A&A 419, 989 (2004)
3. B. E. Reddy, J. Tomkin, D. L. Lambert, C. Allende Prieto, MNRAS 340,
304 (2003)
4. B. R. Olsen, E. H. Udry, S. Mowlavi, A&A 419, 989 (2003)
5. B. E. Reddy, D. L. Lambert, C. Allende Prieto, MNRAS 367, 1329 (2006)
6. F. X. Timmes, S. E. Woosley, T. A. Weaver, ApJS 98, 617 (1995)
7. M. Samland, ApJ 496, 155 (1998)
8. A. Goswami, N. Prantzos, A&A 359 191 (2000)
9. A. Anders, N. Grevesse, Geochim. Cosmochim. Acta 53, 197 (1989)
10. H. Palme, H. Beer in Landolt-Bernstein Group VI, A&A ed. by H. H.
Voigt, Springer, Berlin, 3a, 196 (1993)
11. N. Grevesse, A. Noels, A. J. Sauval, ASPC 99, 117(1996)
12. N. Grevesse, A. J. Sauval, Space Sci. Rev. 85, 161 (1998); GS98
13. K. Lodders, ApJ 691, 1220 (2003); L03
14. H. Palme, A. Jones, in Meteorites, Comets, and Planets ed. by A. M.
Davis; Vol. 1 Treatise on Geochemistry eds. H.D. Holland & K. K.
Turekian Elsevier-Pergamon, Oxford, 41 (2003); PJ03
15. M. Asplund, N. Grevesse, A. J. Sauval, ASPC 336, 25 (2005); A05
16. N. Grevesse, M. Asplund, A. J. Sauval, Space Sci. Rev. 130, 105 (2007);
G07
17. K. Lodders, H. Palme, H. P. Gail, in Landolt-Böernstein New Series,
Group IV/4B, chapter 4.4, ed. by Trümper, J. E. (ed.), Berlin, Heidelberg,
New York: Springer-Verlag, p. 560 (2009); LPG09
18. V. M. Goldschmidt, Geochemische Verteilungsgesetzte der Elemente.
IX. Die Mengenverhltnisse der Elemente und der Atom-Arten, Skrifter
Norske Videnskaps Akademi I. Mathematisk-Naturvidenskapelig klasse,
No 4, Oslo 1937/1938, pp. 148. see also V.M. Goldschmidt, J. Amer.
Chem. Soc., 1937, 655 (1937)
19. H. Brown, Rev. Mod. Phys 21, 625 (1949)
20. H. E. Suess, H. C. Urey, Rev. Mod. Phys 28, 53 (1956)
21. K. Lodders, B. Fegley, The Planetary Scientist’s Companion, Oxford
Univ. Press (1998)
22. E. Anders, Geochim. Cosmochim. Acta 35, 516 (1971); AG89
23. K. Lodders, H. Palme, in preparation (2009)
24. A. Greshake, W. Kloeck, P. Arndt, M., Maetz, G. J. Flynn, S. Bajt, S.,
A. Bischoff, Meteoritics Planet. Sci. 33, 267 (1998)
25. A. Morlok, A. Bischoff, T. Stephan, C. Floss, E. Zinner, E. K. Jessberger,
Geochim. Cosmochim. Acta 70, 5371 (2006)
26. H. N. Russell, ApJ 70, 11 (1929)
27. H. Holweger, in Solar and galactic composition ed. by R.F. Wimmer-
Schweinsgruber, AIP Conf. Proc. 598, 23 (2001)
28. M. Steffen, & H. Holweger, A&A 387, 258 (2002)
29. J. R. Shi, T. Gehren, K. Butler, L. I. Mashonkina, G. Zhao, A&A 486,
303 (2008)
416 Katharina Lodders

30. S. Basu, H. M. Anita, Phys. Rep. 457, 217 (2008)

31. M. Carlsson, R. J. Rutten, J. H. M. J. Brules, N. G. Shchukina, A&A
288, 860 (1994)
32. E.A. Müller, E. Peytremann, R. de la Reza, Solar Physics 41, 53 (1975)
33. M. Asplund, A&A 417, 769 (2004)
34. A. Makishima, E. Nakamura, Geostandard and Geoanalytical Research
30, 245 (2006)
35. C. Allende Pietro, D. L. Lambert, M. Asplund, ApJ 573, L137 (2002)
36. M. Asplund, N. Grevesse, A. J. Sauval, C. Allende Prieto, R. Blomme,
A&A 431, 693 (2005)
37. P. C. Scott, M. Asplund, N. Grevesse, A. J. Sauval, A&A 456, 675 (2006)
38. E. Caffau, E. Maiorca, P. Bonifacio, R. Faraggiana et al., A&A 498, 877
(2009)
39. E. A. Caffau, H. G. Ludwig, M. Steffen, T. R. Ayres, P. Bonifacio, R.
Cyarel, B.Freytag, B. Plez, A&A 488, 1031 (2008)
40. H. G. Ludwig, M. Steffen in Precision Spectroscopy in Astrophysics,
Proc. of the ESO/Lisbon/Aveiro conference in Aveiro, Portugal, 11-15
Sep 2006, ed. by N.C. Santos, L. Pasquini, A.C.M. Correia, and M. Ro-
manielleo, Springer-Verlag Berlin Heidelberg, Germany, 133 (2008)
41. J. Melendez, M. Asplund, A&A 490, 817 (2008)
42. T. R. Ayres, C. Plymate, C. U. Keller, ApJS 165, 618 (2006)
43. T. Morel, K. Butler, A&A 487, 307 (2008)
44. E. Landi, U.Feldman, & G. A. Doschek, ApJ 659, 743 (2007)
45. P. R. Young, A&A 444, L45 (2005)
46. P. Bochsler, F. Auchere, R. M. Skoug, in Proc. SOHO 17, ed. by H La-
coste, Ouwehand, L., ESA SP-616, 28 (2006)
47. P. Bochsler, A&A 471, 315 (2007)
48. M. Asplund, A&A 359, 755 (2000)
49. E. Caffau, M. Steffen, L. Sbordone, H. G. Ludwig, P.Bonifacio, A&A 473,
L9 (2007)
50. E. Caffau, H. G. Ludwig A&A 467, L11 (2007)
51. E. Caffau, R. Faraggiana, P. Bonifacio, H. G. Ludwig, M. Steffen, A&A
699, 699 (2007)
52. K. Lodders, ApJ 674, 607 (2008)
53. H. W. Zhang, K. Butler, T. Gehren, J. R. Shi, G. Zhao, A&A 457, 645
(2006)
54. D. L. Lambert, B. Warner, MNRAS 138, 181 (1968)
55. H. W. Zhang, T. Gehren, G. Zhao, A&A 48, 489 (2008)
56. J. S. Sobeck, J. R. Lawler, C. Sneden, ApJ 667, 1267 (2007)
57. M. Bergemann, T. Gehren, A&A 473, 291 (2007)
58. R. Blackwell-Whitehead, M. Bergemann, A&A 472, L43 (2007)
59. G. Ljung, H. Nilsson, M. Asplund, S. Johansson, A&A 456, 1181 (2006)
60. H. L. Xu, Z. W. Sun, Z. W. Dai, Z. K. Jiang, P. Palmeri, P. Quinet, E.
Biemont, A&A 452, 357 (2006)
61. G. Gonzales, MNRAS 371, 781 (2006)
62. N. Vitas, I. Vince, M. Lugaro, O. Andriyenko, M. Gosic, R. J. Rutten,
MNRAS 384, 370 (2008)
63. C. Sneden, J. E. Lawler, J. J. Cowan, I. I. Ivans, E. A. Den Hartog, ApJS
182, 80 (2009)
 Solar System Abundances 417

64. S. Ivarsson, G. M., Wahlgren, H. G. Ludwig, Bull. A&AS 35, 1451 (2003)
65. E. A. Den Hartog, J. E. Lawler, C. Sneden, J. J. Cowan, ApJS 148, 543
(2003)
66. J. E. Lawler, E. A., Den Hartog, C. Sneden, J. J. Cowan, ApJS 162, 227
(2006)
67. A. Mucciarelli, E. Caffau, B. Freytag, H. G. Ludwig, P. Bonifacio, A&A
484, 841 (2008)
68. J. E. Lawler, M. E. Wickliffe, E. A. Den Hartog, C. Sneden, ApJ 563,
1675 (2001)
69. E. A. Den Hartoog, J. E. Lawler, C. Sneden, J. J. Cowan, ApJS 167, 292
(2006)
70. J. E. Lawler, C. Sneden, J. J. Cowan, ApJ 604, 850 (2004)
71. J. E. Lawler, C. Sneden, J. J. Cowan, J. F. Wyart, I. I. Ivans, J. S. Sobeck,
M. H.Stockett, E. A. Den Hartog, ApJS 178, 71 (2008)
72. J. E. Lawler, E. A. D. Hartog, Z. E. Labby, C. Sneden, J. J. Cowan, I. I.
Ivans, ApJS 169, 120 (2007)
73. E. Caffau, E. Maiorca, P. Bonifacio, R. Faraggiana, et al., A&A 483, 591
(2008)
74. P. Quinet, P. Palmari, E. Biemont, A. Jorrison, et al., A&A 448, 1207
(2006)
75. E. A. Den Hartog, M. T. Herd, J. E. Lawler, C. Sneden, J. J. Cowan, T.
C. Beers, ApJ 619, 639 (2005)
76. S. Basu, H. M. Anita, 2004, ApJ 606, L85 (2004)
77. W.D. Harkins, J. Am. Chem. Soc. 39, 856 (1917)
78. G. Wallerstein, I. Iben, P. Parker, A. M. Boesgaard, G. M. Hale, A. E.
Champagne, et al., Rev. Mod. Phys.69, 995 (1997)
79. S. E. Woosley, A. Heger, T. A. Weaver, Rev. Mod. Phys. 74, 1015 (2002)
80. C. Sneden, J. J. Cowan, R, Gallino, ARAA 46, 145 (2008)
81. M. E. Wieser, J. R. DeLaeter, Phys. Rev. C75, 055802 (2007)
82. I. Segal, L. Halicz, I. T. Platzner, Int. J. Mass Spec. 216, 177 (2002)
83. T. L. Chang, W. J. Li, M. T. Zhao, J. Wang, Q. Y. Qian, Int. J. Mass
Spec. 207, 13 (2001)
84. T. L. Chang, M. T. Zhao, W. J. Li, J. Wang, Q. Y. Qian, Int. J. Mass
Spec. 177, 131 (1998)
85. J. R. De Laeter, N. Bukilic, Int. J. Mass. Spec. 252, 222 (2006)
86. J. R. Laeter, N. Bukilic, Phys. Rev. C73, 045806 (2006)
Cosmochemistry: A Perspective

Aruna Goswami

Indian Institute of Astrophysics, Koramangala, Bangalore 560034

[email protected]

Cosmochemistry can be said to aim at constructing a quantitative and contin-

uous history of chemical evolution of our universe starting from the appear-
ance of hydrogenic matter immediately after the Big Bang. While the main
events of this history are embodied in stellar nucleosynthesis, the scope of the
complete history is much larger. Many important aspects of cosmochemistry
still remain unclear, although improved observations and steadily increasing
computing power hold exciting promises. The Kodai school on ‘Synthesis of
Elements in Stars’ provided glimpses of status and challenges of diverse issues
in this emerging discipline; the following paragraphs provide a synthesis of
the various contributions to arrive at a perspective.
The non-linear and multi-scale physics that govern the life of a star are not
amenable to linear or simplified calculations. However, much insight about
stellar nuclear reactions, stellar evolution as well as stellar explosions have
come from computer simulations. As also noted by Ray (this volume) astro-
nomical observations across the entire spectrum of electromagnetic radiation
and other indicators like neutrinos have significantly aided theoretical de-
velopments. A major contribution has come from the wider availability of
powerful telescopes from different platforms. Similarly, high speed computers
had given important feedback to theoretical developments in many areas. An-
other important and encouraging trend in cosmochemistry is the increasing
feedback from laboratories. Crucial inputs to the field of nuclear astrophysics
have come from laboratory experiments involving radioactive ion beams and
intense high-energy particle beams. Since short lived nuclei as found in SN
ejecta can only be studied close to their sites of formation/in the laboratory
before they decay, such studies can further contribute to the advancement of
nuclear astrophysics. With growing and affordable and increasing computing
power, therefore, computational astrophysics in general, and computational
cosmochemistry in particular, is an area that holds many promises.
Many concepts and techniques of cosmochemistry have a generic nature.
The methods and results of chemical equilibrium calculations applied to
solar composition are applicable to chemistry in a variety of astronomical

A. Goswami and B.E. Reddy (eds.), Principles and Perspectives in Cosmochemistry,

Astrophysics and Space Science Proceedings,
DOI 10.1007/978-3-642-10352-0 9, c Springer-Verlag Berlin Heidelberg 2010
420 Aruna Goswami

environments including the atmospheres and circumstellar envelopes of cool

stars, the solar nebula and protoplanetary accretion disks around other stars,
planetary atmospheres, and the atmospheres of brown dwarfs. Similarly, the
results of chemical equilibrium calculations have guided studies of elemental
abundances in meteorites and presolar grains and as a result have helped to
refine and constrain models of nucleosynthesis in stars.
Recent estimates indicate that as much as half of all elements heavier than
iron are made by the s-process in low-mass AGB stars. However, while our
understanding of the inner workings of AGB stars has significantly improved
in the recent past, there are many unknowns that render predictions uncer-
tain. As emphasized by Karakas (this volumn) our lack of knowledge about
convective mixing processes in stars is one of the greatest sources of uncer-
tainty. Other modeling uncertainties such as mass-loss rates, opacities, and
reaction rates are also important and need careful investigation.
Many challenges also remain in understanding the structure and evolu-
tion of low and intermediate-mass stars prior to the AGB phase, such as the
mechanisms of formation of 13 C pockets and the extent of extra-mixing pro-
cesses during the first dredge-up and asymptotic giant branch phases. The
processes driving mass loss in AGB stars remain a major challenge, an impor-
tant requirement is to develop techniques to accurately measure many of the
reaction rates of interest for AGB nucleosynthesis. Better understanding of
nucleosynthesis in low-metallicity AGB stars is another essential requirement
to solve the puzzle surrounding the abundances of many carbon-rich metal-
poor stars in the Galaxy. The mass-transfer processes that occur in binary star
systems also remain poorly understood; AGB stars thus remain a challenging
and critical gap in our understanding of stellar evolution.
Supernova events as well as pre-supernova stages of stars play important
roles in cosmochemistry. However, a number of aspects of astrophysics of
both pre-supernova and supernova stages of evolution of massive stars remain
unclear, as emphasized by Marcel Arnould (this volume). A major challenge
comes from the uncertainties in the rates of some nuclear reactions and of weak
interaction processes. Multi-dimensional simulations, especially of the late pre-
supernova phases, are expected to be particularly helpful to understand these
processes. Similarly, considerable efforts are still required for successful simu-
lations of supernova explosions based on less uncertain pre-supernova models.
Here again, the multi-dimensional treatment of a variety of physical effects,
including rotation, magnetic fields, instabilities of different origins, and the
transport of neutrinos needs careful inclusion.
Among the most vexing nucleosynthetic puzzles is the r-process. Some
success has been achieved with regard to p-process elements for which cal-
culations conducted with ‘realistic’ models for the O-Ne-rich layers of SNeII
broadly indicate that they may all provide yields enriched with a suite of
p-nuclides at a level that is compatible (within a factor of 3 or so) with the
solar values. As highlighted by Marcel Arnould this relative success, however,
does not remove a host of puzzling problems. It is still not clear, for example,
 Cosmochemistry: A Perspective 421

if the underproduction of the p-isotopes of Mo and Ru in SNII explosions is

in fact due to some misrepresentation of the production in the He-burning
core of massive stars of the s-nuclide seeds for the p-process. The modelling of
the s-process in these conditions indeed faces uncertainties from both nuclear
physics and astrophysics which are highly non-trivial.
The arena of cosmochemistry is emerging as a broad interdisciplinary area
of research. It is one area where astrophysical processes at different scales, both
spatial and temporal, combine to provide an integrative view of the chemical
history of the Galaxy/galaxies, their structure and origin. The primary goal
of this collection of lectures is to put together important components of cos-
mochemistry which we hope will serve as a contemporary and useful reference
for researchers in this exciting and challenging area.