0% found this document useful (0 votes)

269 views

A Concise Introduction To Astrophysics

Uploaded by

Dionisis Zettas

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

269 views

A Concise Introduction To Astrophysics

Uploaded by

Dionisis Zettas

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 144

A Concise Introduction to Astrophysics

Lecture Notes for FY2450

M. Kachelrie

M. Kachelrie
Institutt for fysikk NTNU, Trondheim Norway email: [email protected]

Copyright c M. Kachelrie 2008, minor corrections 2011.

Contents
I Stellar astrophysics
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10
11 11 12 12 13 14 15 16 17 18 20 20 22 23 26 26 26 28 28 30 31 31 32 32 32 32 33 34 34 36 36 38 39

1 Continuous radiation from stars 1.1 Brightness of stars . . . . . . . . . . . . . . . . 1.2 Color of stars . . . . . . . . . . . . . . . . . . . 1.3 Black-body radiation . . . . . . . . . . . . . . . 1.3.1 Kirchho-Planck distribution . . . . . . 1.3.2 Wiens displacement law . . . . . . . . . 1.3.3 Stefan-Boltzmann law . . . . . . . . . . 1.3.4 Spectral energy density of a photon gas 1.4 Stellar distances . . . . . . . . . . . . . . . . . 1.5 Stellar luminosity and absolute magnitude scale

2 Spectral lines and their formation 2.1 Bohr-Sommerfeld model for hydrogen-like atoms . . . . . . . . . . . . . . . . 2.2 Formation of spectral lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Hertzsprung-Russel diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Telescopes and other detectors 3.1 Optical telescopes . . . . . . . . . . . . 3.1.1 Characteristics of telescopes . . . 3.1.2 Problems and limitations . . . . 3.2 Other wave-length ranges . . . . . . . . 3.3 Going beyond electromagnetic radiation 3.3.1 Neutrinos . . . . . . . . . . . . . 3.3.2 Gravitational waves . . . . . . .

. . . . . . .

4 Basic ideas of special relativity 4.1 Time dilation a Gedankenexperiment . . 4.2 Lorentz transformations and four-vectors . 4.2.1 Galilean transformations . . . . . . 4.2.2 Lorentz transformations . 4.2.3 Energy and momentum . . . . . . 4.2.4 Doppler eect . . . . . . . . . . . .

. . . . . .

5 Binary stars and stellar parameters 5.1 Keplers laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1.1 Keplers second law: The area law . . . . . . . . . . . . . . . . . . . . 5.1.2 Keplers rst law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Contents 5.1.3 Keplers third law . . . Determining stellar masses . . . 5.2.1 Mass-luminosity relation Stellar radii . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 40 41 42 43 43 43 45 46 48 48 48 49 50 51 52 54 54 55 55 55 57 58 58 58 59 59 59 59 60 61 62 63 63 63 65 65 68 70 70 71 72 73

5.2 5.3

6 Stellar atmospheres and radiation transport 6.1 The Sun as typical star . . . . . . . . . 6.2 Radiation transport . . . . . . . . . . . 6.3 Diusion and random walks . . . . . . . 6.4 Photo-, chromosphere and corona . . . .

. . . .

7 Main sequence stars and their structure 7.1 Equations of stellar structure . . . . . . . . . . . . . . 7.1.1 Mass continuity and hydrostatic equilibrium . . 7.1.2 Gas and radiation pressure . . . . . . . . . . . 7.1.3 Virial theorem . . . . . . . . . . . . . . . . . . 7.1.4 Stability of stars . . . . . . . . . . . . 7.1.5 Energy transport . . . . . . . . . . . . . . . . . 7.1.6 Thermal equilibrium and energy conservation . 7.2 Eddington luminosity and convective instability . . . . 7.3 Eddington or standard model . . . . . . . . . . . . . . 7.3.1 Heuristic derivation of L M 3 . . . . . . . . . 7.3.2 Analytical derivation of L M 3 . . . 7.3.3 Lifetime on the Main-Sequence . . . . . . . . . 7.4 Stability of stars . . . . . . . . . . . . . . . . . . . . . 7.5 Variable stars period-luminosity relation of Cepheids 7.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

8 Nuclear processes in stars 8.1 Possible energy sources of stars . . . . . . . . . . . . . . . . 8.1.1 Gravitational energy . . . . . . . . . . . . . . . . . . 8.1.2 Chemical reactions . . . . . . . . . . . . . . . . . . . 8.1.3 Nuclear fusion . . . . . . . . . . . . . . . . . . . . . 8.2 Excursion: Fundamental interactions . . . . . . . . . . . . . 8.3 Thermonuclear reactions and the Gamov peak . . . . . . . 8.4 Main nuclear burning reactions . . . . . . . . . . . . . . . . 8.4.1 Hydrogen burning: pp-chains and CNO-cycle . . . . 8.4.2 Later phases . . . . . . . . . . . . . . . . . . . . . . 8.5 Standard solar model and helioseismology . . . . . . . . . . 8.6 Solar neutrinos . . . . . . . . . . . . . . . . . . . . . . . . . 8.6.1 Solar neutrino problem and neutrino oscillations 9 End 9.1 9.2 9.3 9.4 points of stellar evolution Observations of Sirius B . . . . . . . . Pressure of a degenerate fermion gas . White dwarfs and Chandrasekhar limit Supernovae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . .

. . . .

Contents 9.5 Pulsars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 78 78 78 79 80 81 83

10 Black holes 10.1 Basic properties of gravitation . . . . . . . . . . . . . 10.2 Schwarzschild metric . . . . . . . . . . . . . . . . . . 10.2.1 Heuristic derivation . . . . . . . . . . . . . . 10.2.2 Interpretation and consequences . . . . . . . 10.3 Gravitational radiation from pulsars . . . . . . . . . 10.4 Thermodynamics and evaporation of black holes

. . . . . . . . . . . . . . .

. . . . . .

Galaxies
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

86
87 87 88 89 89 91 92 92 94 97 98 99 99 99 101 102 103 103 104 106 107 107 109 110

11 Interstellar medium and star formation 11.1 Interstellar dust . . . . . . . . . . . 11.2 Interstellar gas . . . . . . . . . . . 11.3 Star formation . . . . . . . . . . . 11.3.1 Jeans length and mass . . . 11.3.2 Protostars . . . . . . . . . . 12 Cluster of stars 12.1 Overview . . . . . . . . . . . . . 12.2 Evolution of a globular cluster . 12.3 Virial mass . . . . . . . . . . . . 12.4 Hertzsprung-Russell diagrams for

. . . . . . . . . . . . . . . clusters

. . . .

13 Galaxies 13.1 Milky Way . . . . . . . . . . . . . . . . . . . 13.1.1 Rotation curve of the Milkyway . . . . 13.1.2 Black hole at the Galactic center . . . 13.2 Normal and active galaxies . . . . . . . . . . 13.3 Normal Galaxies . . . . . . . . . . . . . . . . 13.3.1 Hubble sequence . . . . . . . . . . . . 13.3.2 Dark matter in galaxies . . . . . . . . 13.3.3 Galactic evolution . . . . . . . . . . . 13.4 Active Galaxies and non-thermal radiation . . 13.4.1 Non-thermal radiation . . . . . . . . . 13.4.2 Radio galaxies . . . . . . . . . . . . . 13.4.3 Other AGN types and unied picture

. . . . . . . . . . . .

III Cosmology
14 Overview: Universe on large scales 14.1 Problems of a static, Newtonian Universe 14.2 Einsteins cosmological principle . . . . . 14.3 Expansion of the Universe: Hubbles law . 14.3.1 Cosmic distance ladder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

113
114 114 114 115 116

Contents 15 Cosmological models for an homogeneous, isotropic universe 15.1 Friedmann-Robertson-Walker metric for an homogeneous, isotropic 15.2 Friedmann equation from Newtons and Hubbles laws . . . . . . . 15.2.1 Friedmann equation . . . . . . . . . . . . . . . . . . . . . . 15.2.2 Local energy conservation and acceleration equation . . . . 15.3 Scale-dependence of dierent energy forms . . . . . . . . . . . . . . 15.4 Cosmological models with one energy component . . . . . . . . . . 15.5 Determining and the curvature R0 from m,0 , H0 , q0 . . . . . . . 15.6 The CDM model . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Early universe 16.1 Thermal history of the Universe 16.2 Big Bang Nucleosynthesis . . . . 16.3 Structure formation . . . . . . . 16.4 Cosmic microwave background . 16.5 Ination . . . . . . . . . . . . . . Time-line of important dates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 118 121 121 123 124 125 126 128 130 130 132 134 136 137

universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

A Some formulae 139 A.1 Mathematical formulae . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 A.2 Some formulae from cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . 139 B Units and useful constants B.1 SI versus cgs units . . . . . . . . . . . . . B.2 Natural units . . . . . . . . . . . . . . . . B.3 Physical constants and measurements . . B.4 Astronomical constants and measurements B.5 Other useful quantities . . . . . . . . . . . B.6 Abbreviations: . . . . . . . . . . . . . . . B.7 Properties of main-sequence stars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 140 140 141 141 141 142 142

Astrophysicssome introductory remarks

Astronomy is with mathematics one of the oldest branches of science. It has served as basis for calendars, navigation, has been an important input for religions and was for a long time intertwined with astrology. Some of the most important steps in modern astronomy were:

Galileo performed 1609 the rst astronomical studies using a telescope. He discovered among others four Saturn moons and sun spots. Kepler (1571-1630) developed his three laws of planetary motions, based on observations of Tycho Brahe. Newton established 1687 his laws of motion and gravitation. The measurement of the distance to Venus 1761 and 1769 during its transits of the Sun with the help of the rst global measurement campaign and to the nearest stars 1838 by Bessel using trigonometric parallaxes established the rst rungs in the cosmic distance ladder. Fraunhofer discovered around 570 spectral lines in the solar light in 1814 and catalogued them. This opened together with the spectral analysis of Kirchho and Bunsen (1859) the way to study the physical properties of stars. Einsteins general theory of relativity (1916) provided the rst consistent basis to study cosmology. The Great Debate in 1920 was concerned about the question Does the Milky Way represents the whole Universe or is it just one island among many others? Opik, Shapley and Hubble showed that the latter is true. Hubble discovered 1926 that galaxies are recessing and that their velocity is increasing with distance: The universe is expanding and, extrapolating this expansion back in time led to the idea of the Big Bang. Nuclear fusion was suggested in 1920 by Eddington as source for stellar energy, the main principles were worked out after the advent of quantum mechanics by Bethe and v. Weizscker in the 1930s. a The discovery of the cosmic 2.7 Kelvin background radiation 1964 by Penzias and Wilson gave credit to the Big Bang theory.

Few other examples for the interconnection of astrophysics and physics, where astronomical observations were an important input for fundamental physics, are: Olaf C. Rmer (1644-1710) showed 1676 that the speed of light is nite by obsero vations of Jupiter moons: Light needs around 20 min to cross the Earth orbit. The 1919 solar eclipse was the rst crucial test passed by the theory of General Relativity of Einstein, while a binary system of two pulsars discovered by Hulse and Taylor in 1974 became the rst experimental evidence for the existence of gravitational waves.

Contents Observation of neutrinos from the Sun and produced by cosmic rays in the Earths atmosphere gave in the 1990s rst rm evidence that neutrinos have non-zero masses. The need for a new form of dark matter to describe correctly the formation and dynamics of galaxies requires a yet unknown extension of the current standard model of elementary particle physics. The same holds true for a new form of dark energy required for the explanation of the accelerated expansion of the universe. Astronomy as a purely observational science is unique among natural sciences; all others are based on experiments. Since the observation time is much smaller than the typical time scale for the evolution of astronomical objects, we see just a snapshot of the universe. Nevertheless it is possible to reconstruct, e.g., the evolution of stars by studying large samples. On the extragalactic scales, we can use that looking far away means looking into the past because of the nite speed of light, while on the cosmological scale we can use relics formed soon after the Big Bang as testimonies for the state of the early universe. The cosmological principle is based on the belief that the mankind and the Earth have no special role. Thus physical laws derived on Earth are valid everywhere and at all times. Astronomers and especially cosmologists are said to live these days in a golden age: There has been a tremendous increase of knowledge in the last 15 years: Telescopes and detectors on satellites explore new wavelength ranges, while new automatized ways to analyse data allow astronomers the comparative study of e.g. millions of galaxies. Astrophysics needs input of practically all sub-disciplines of physics and thus a course on astrophysics cannot be self-contained. However, the course should be accessible to students with just a general introduction to physics. Few sections of the text that are somewhat more advanced and that can be omitted are marked by stars. I will be glad to receive feedback from the readers of these notes. If you nd errors or have anysuggestions, send me an email!

Part I

Stellar astrophysics

1 Continuous radiation from stars

Practically all information we know about stars and galaxies comes from observing electromagnetic radiation, more precisely from two small windows where the Earth atmosphere is (nearly) transparent: The rst one is around the range of visible light (plus some windows in the infra-red (IR)), while the second one is the radio window in the wave-length range 1 cm < < 30 m, cf. Fig. 1.1.

1.1 Brightness of stars

Hipparcos and Ptolemy ( 150 B.C.) divided stars into six classes of brightness, called (apparent) magnitudes m: Stars with m = 6 were the faintest objects visible by eye, stars with m = 1 the brightest stars on the sky. Thus the magnitude scale m increases for fainter objects, a counter-intuitive fact that should be kept in mind. The response of the human eye is not linear but close to logarithmic to ensure a large dynamical range. With the use of photographic plates, the magnitude scale could be dened quantitatively. Pogson found 1856 that the magnitude dierence m = 5 corresponds to a ratio of energy uxes of F2 /F1 100, where the energy ux F is dened as the energy E going through the area A per time t, F = E/(At). This ratio was then used as denition, together with m = 2.12 for the polar star as reference point. Thus the ratio of received energy per time and area from two stars is connected to their magnitudes by F2 = 100(m1 m2 )/5 = 10(m1 m2 )/2.5 F1 or1 m1 m2 = 2.5 log F1 . F2 (1.1)

(1.2)

The modern denition of the magnitude scale does not agree completely with the original Greek one: For instance, the brightest star visible from the Earth, Sirius, has the magnitude m = 1.5 instead of the Ptolemic m = 1. Note also that in particular astronomers use often the equivalent name (apparent) brightness b for the energy ux received on Earth from a star or a galaxy. Ex.: The largest ground-based telescopes can detect stars or galaxies with magnitude m = 26. How much fainter are they than a m = 1 star like Antares?
F1 = 1025/2.5 = 1010 . F2 (1.3)

Thus the number of photons received per time and frequency interval from such a faint object is a factor 1010 smaller than from Antares. Since the energy ux F scales with distance r approximately
1

We denote logarithms to the base 10 with log, while we use ln for the natural logarithm.

1 Continuous radiation from stars

Figure 1.1: Left: The electromagnetic spectrum. Right: Transmission of the Earths atmosphere.
as 1/r2 , one can say equivalently that with the help of such a telescope one would see a factor 105 further than with the bare human eye.

This magnitude scale corresponds to the energy ux at Earth, i.e. the scale is not corrected for the dierent distance of stars. They are therefore called apparent magnitudes.

1.2 Color of stars

A look at Fig. 1.2 shows that the color of stars and galaxies varies considerably. A quantitative way to measure the color of stars is to use lters which sensitivity is centered at dierent wavelengths i and compare their relative brightness, m1 m2 = 2.5 log F1 . F2 (1.4)

Then m1 m2 is called the color or the color index of the object. Often used are lters for visible (V), ultraviolet (U), and blue (B) light. For instance, one denotes briey with B-V the dierence in magnitudes measured with a lter for blue and visible light, B V mB mV = 2.5 log FB . FV (1.5)

It is intuitively clear that the color of a star is connected to its temperaturethis connection will be made more precise in the next section. The color magnitudes are normalized such that their dierences m1 m2 are zero for a specic type of stars with surface temperature T 10, 000 K (cf. Sec. 2.2).

1.3 Black-body radiation

A black body is the idealization of a medium that absorbs all incident radiation. This idealization is very useful, because many objects close to thermal equilibrium, in particular stars, emit approximately blackbody radiation. In practise, black-body radiation can be approximated by radiation escaping a small hole in a cavity whose walls are in thermal equilibrium. Inside such a cavity, a gas of photons in thermal equilibrium exists.

1.3 Black-body radiation

Figure 1.2: An excerpt from the Deep Field View taken by the Hubble Space Telescope.

1.3.1 Kirchho-Planck distribution

Planck found empirically 1900 the so-called Kirchho-Planck (or simply Planck) distribution B d = 1 2h 3 d 2 h c exp( kT ) 1 (1.6)

describing the amount of energy emitted into the frequency interval [, + d] and the solid angle d per unit time and area by a body in thermal equilibrium. The function B depends only on the temperature T of the body (apart from the natural constants k, c and h) and is shown in Fig. 1.4 for dierent temperature T . The dimension of B in the cgs system of units is erg . (1.7) [B ] = Hz cm2 s sr In general, the amount of energy per frequency interval [, + d] and solid angle d crossing the perpendicular area A per time is called the (dierential) intensity I , cf. Fig. 1.3, I = dE . d d dA dt (1.8)

For the special case of the blackbody radiation, the dierential intensity is given by the Kirchho-Planck distribution, I = B . Equation (1.6) gives the spectral distribution of black-body radiation as function of the frequency . In order to obtain the corresponding formula as function of the wavelength , one has to re-express both B and d: With = c/ and thus d = c/ 2 d, it follows B d = 2hc2 1 d . 5 hc exp( kT ) 1 (1.9)

1 Continuous radiation from stars

dA cos dA cos dA

Figure 1.3: Left: A detector with surface element dA on Earth measuring radiation coming from a direction with zenith angle . Right: An imaginary detector on the surface of a star measuring radiation emitted in the direction . The Kirchho-Planck distribution contains as its two limiting cases Wiens law for highfrequencies, h kT , and the Rayleigh-Jeans law for low-frequencies h kT . In the former limit, x = h/(kT ) 1, and we can neglect the 1 in the denominator of the Planck function, 2h 3 B 2 exp(h/kT ) . (1.10) c Thus the number of photons with energy h much larger than kT is exponentially suppressed. In the opposite limit, x = h/(kT ) 1, and ex 1 = (1 + x . . .) 1 x. Hence Plancks constant h disappears from the expression for B , if the energy h of a single photon is small compared to the thermal energy kT and one obtains, B 2 2 kT . c2 (1.11)

The Rayleigh-Jeans law shows up as straight lines left from the maxima of B in Fig. 1.4.

1.3.2 Wiens displacement law

We note from Fig. 1.4 two important properties of B : Firstly, B as function of the frequency has a single maximum. Secondly, B as function of the temperature T is a monotonically increasing function for all frequencies: If T1 > T2 , then B (T1 ) > B (T2 ) for all . Both properties follow directly from taking the derivative with respect to and T . In the former c2 case, we look for the maximum of f () = 2h B as function of . Hence we have to nd the zeros of f (), h 3(ex 1) x expx = 0 with x= . (1.12) kT The equation ex (3 x) = 3 has to be solved numerically and has the solution x 2.821. Thus the intensity of thermal radiation is maximal for xmax 2.821 = hmax /(kT ) or cT 0.50K cm max or max 5.9 1010 Hz/K . T (1.13)

1.3 Black-body radiation

10-15

B in [eV cm-2 s-1 sr-1] ???

T=104K 10-20 T=103K

10-25

T=100K

10-30 1e+08

1e+09

1e+10 1e+11 in Hz

1e+12

1e+13

Figure 1.4: The Planck distribution B as function of the frequency for three dierent temperatures. Similarly one derives, using the expression for B , the value max where B peaks as max T 0.29K cm . (1.14)

This equation is called Wiens displacement law. Thus determining the color of a star tells us its temperature: A hot star is bluish and a cool star reddish. In practise, the maximum of the Planck distribution lies often outside the range of visible light and the measurement of a color index like U V is an easier method to determine the temperature T of a star. Ex.: Surface temperature of the Sun:
We can obtain a rst estimate of the surface temperature of the Sun from the know sensitivity of the human eye to light in the range 400700 nm. Assuming that the evolution worked well, i.e. that the human eye uses optimal the light from the Sun, and that the atmosphere is for all frequencies in the visible range similarly transparent, we identify the maximum in Wiens law with the center of the frequency range visible for the human eye. Thus we set max, 550 nm, and obtain T = 2.9 106 nm K/550 nm 5270 K for the surface temperature of the Sun. This simple estimate should be compared to the precise value of 5781 K for the eective temperature of the solar photosphere dened in the next subsection.

1.3.3 Stefan-Boltzmann law

Integrating Eq. (1.6) over all frequencies and possible solid angles gives the energy ux F emitted per surface area A by a black body. The angular integral consists of the solid angle d = d sin d and the factor cos taking into account that only the perpendicular area A = A cos is visible,
2 /2 /2

d cos =
0

d
0

d sin cos =
0

d sin 2 = .

(1.15)

1 Continuous radiation from stars

dl Figure 1.5: The connection between the contained energy density u and the intensity I of radiation crossing the volume dV . Next we perform the integral over d substituting x = h/(kT ),

F = where we used
x3 dx 0 ex 1

dB =
0

2 (kT )4 c2 h3

x3 dx = T 4 , ex 1

(1.16)

= 4 /15 and introduced the Stefan-Boltzmann constant, = erg 2 5 k 4 = 5.670 105 2 4 . 2 h3 15c cm K s (1.17)

If the Stefan-Boltzmann law is used to dene the temperature of a body that is only approximately in thermal equilibrium, T is called eective temperature Te .

1.3.4 Spectral energy density of a photon gas

A light ray crosses an innitesimal cylinder with volume dV = dAdl in the time dt = dl/(c cos ), cf. Fig. 1.5. Hence the spectral energy density u of photons, i.e. the number of photons per volume and energy interval, in thermal black-body radiation is u = dE dE 1 = = dV d c cos dtdAd c d B = 4 B . c (1.18)

The (total) energy density of photons u follows as u= 4 c

dB = aT 4 ,

(1.19)

where we introduced the radiation constant a. Comparing with Eqs. (1.16) and (1.17) shows that a = 4/c. Ex.: Calculate the density and the mean energy of the photons in thermal black-body radiation at
temperature T .

1.4 Stellar distances

~ star U U U U p U U U U U U U U U U U U U U U U U ~ U U~ Earth x x

Sun ~

Figure 1.6: Measurement of the trigonometric parallax. Left: The apparent movement of a nearby star during a year: Half of this angular dierence is called the parallax angle or simply the parallax p. Right: Indirect determination of the Earth-Sun distance.
The number density of black-body photons is connected to the energy density by the replacement B B /(h), n= 4 c

d
0

B 8 = 3 h c

d
0

2
h exp( kT ) 1

8 (kT /h)3 c3

dx
0

x2 2(3) = ex 1 2

kT c

where we looked up the result for the last integral in Table A.1. The mean energy of photons at temperature T follows then as E = h = 4 u = kT 2.701kT . n 30(3)

1.4 Stellar distances

Relatively nearby stars are seen at slightly dierent positions on the celestial sphere (i.e. the background of stars that are innitely far away) as the Earth moves around the Sun. Half of this angular dierence is called the parallax angle or simply the parallax p. From Fig. 1.6, one can relate the mean distance AU of the Earth to the Sun, the parallax p and the distance d to the star by AU tan p = . (1.20) d

1 Continuous radiation from stars With tan p p for p 1, one has d 1/p. Because of this simple and for observations important relation, a new length unit is introduced: The parsec is dened to be the distance from the Earth to a star that has a parallax of one arcsecond2 . Thus d[pc] = 1 p[ ] (1.21)

i.e. a star with a parallax angle of n arcseconds has a distance of 1/n parsecs. Since one arcsecond is 1/(360 60 60) = 1/206265 fraction of 2, a parsec corresponds to 206 265 AU. To express nally the unit pc in a known unit like cm, we have to determine rst the EarthSun distance, the astronomical unit AU. As rst step one measures the distance d to an inner planet at the time of its greatest elongation (i.e. the largest angular distance between the Sun and the planet). Nowadays, one uses a radar signal to Venus. Then AU = d/ cos , cf. the right panel of Fig. 1.6, and one nds 1AU = 1.496 1013 cm. As result, one determines one parsec as 1 pc = 206, 265 AU = 3.086 1018 cm = 3.26 lyr.

1.5 Stellar luminosity and absolute magnitude scale

The total luminosity L of a star is given by the product of its surface A = 4R2 and the radiation emitted per area T 4 , L = 4R2 T 4 . (1.22) Since the brightness or energy ux F was dened as F = E/(At) = L/A, we recover the inverse-square law for the energy ux at the distance r > R outside of the star, F= L . 4r2 (1.23)

The validity of the inverse-square law F(r) 1/r2 relies on the assumptions that no radiation is absorbed and that relativistic eects can be neglected. The later condition requires in particular that the relative velocity of observer and source is small compared to the velocity of light. Ex.: Luminosity and eective temperature of the Sun.
The energy ux received from the Sun at the distance of the Earth, d = 1 AU, is equal to F = 1365 W/m2 . (This energy ux is also called Solar Constant.) The solar luminosity L follows then as L = 4d2 F = 4 1033 erg/s and serves as a convenient unit in stellar astrophysics. The Stefan-Boltzmann law can then be used to dene with R 7 1010 cm the eective temperature of the Sun, Te (Sun) T 5780 K.

The brightness discussed in Sec. 1.1 is based on the observed energy uxes on Earth and does not account for the dierent distance of stars. It is therefore called apparent magnitude. To compare the intrinsic luminosities of stars, one denes the absolute magnitude M as the one a star would have at the standard distance d = 10 pc.
2

A degree, i.e. the 1/360 part of a circle is divided into 60 arcminutes and 60 60 arcseconds, 360 = 360 60 = 360 60 60 .

1.5 Stellar luminosity and absolute magnitude scale Using that F 1/d2 , the relation between m and M is from Eq. (1.2) m M = 2.5 log d 10pc
2

= 5 log

d . 10pc

(1.24)

The quantity m M is also called distance modulus. Measuring the received energy ux F from a star with known absolute magnitude establishes also the conversion between M and usual cgs or SI units, L = 3.02 1035 erg 100.4M = 3.02 1028 W 100.4M . s (1.25)

Exercises
1. Show that the mean distance between black-body photons corresponds approximately to the wave-length of photons from the maximum of the Planck distribution, n 1/3 . max 2. Derive Wiens law both for frequency and wavelength. Hint: The equations ex (5x) = 5 and ey (3y) = 3 have the solutions x = 4.965 and y = 2.821. Show that max = c/max . 3. Show that B (T1 ) > B (T2 ) for all , if T1 > T2 . 4. Derive the Earth-Sun distance D assuming that both are perfect blackbodies. Use R = 7 1010 cm, T = 5780 K and T = 288 K.

2 Spectral lines and their formation

Classical astronomy was mainly concerned with the determination of planetary orbits an understanding of the physical properties of the Sun or stars in general seemed to be impossible. The breakthrough that paved the way to stellar astrophysics was the spectral analysis of stellar light: Already Fraunhofer identied one prominent of the dark lines in the continuous spectrum of the Sun with the one emitted by salt sprinkled in a ame. The eld of spectral analysis was then established by Kirchho and Bunsen in 1859: They realized that each element produces its own pattern of spectral lines and can be thereby identied. Thus one could show that the solar light contains strong absorption lines of calcium and of previously unknown elements as helium. However, a proper understanding which factors determine the strength of various spectral lines in the spectrum of the Sun and other stars become only possible in the 1920s, after the advent of quantum mechanics. Thus we start with a brief review of the simplest atom, the hydrogen atom, in a semi-classical picture.

2.1 Bohr-Sommerfeld model for hydrogen-like atoms

Bohr postulated the existence of discrete, stationary orbits for electrons in atoms. More precisely, he required that the angular momentum J = me vr of an electron is a multiple of Plancks constant h/(2), J = me vr = n , n = 0, 1, 2, . . . (2.1)

If the electron is on a bound orbit, the electric force is balanced by the centrifugal force, me v 2 e2 = 2 r r and the quantization condition becomes me v 2 = e2 n2 2 = . 2 me rn rn (2.3) (2.2)

Solving for rn gives the allowed electron orbits in the Bohr model, rn = n2 2 , me e2 n = 1, 2, . . . (2.4)

The total energy of an electron moving in the Coulomb eld of an (innitely heavy) nucleus with electric charge e is the sum of its kinetic and potential energy, e2 e2 1 = , E = Ekin + Epot = me v 2 2 r 2r (2.5)

2.1 Bohr-Sommerfeld model for hydrogen-like atoms n= n=3 Paschen, IR n=2 Balmer series E=0 E = 1.51 eV E = 3.40 eV

Lyman series UV

n=1

E = 13.6 eV

Figure 2.1: The energy levels of the hydrogen spectrum with some lines of the rst three spectral series. where again Eq. (2.5) was used. Inserting rn gives the allowed stationary energy levels in a hydrogen atom, me e4 n = 1, 2, . . . (2.6) En = 2 2 , 2n In transition between dierent levels, only photons with discrete energies h = |En Em | (2.7)

can be absorbed or emitted. An empirical formula describing the wave-length of visible hydrogen spectral lines was found by Balmer already in 1885, 1 =R 1 1 2 4 n , (2.8)

where R is the Rydberg constant, 1/R 91.2 nm. Rewriting the general formula (2.6) for wavelengths = c/ and comparing it with the Balmer formula gives me e4 1 1 hc = |En Em | = 2 2 2 2 n m (2.9)

with n = 2 and R = me e4 /(c 3 ). Thus the Balmer series corresponds to the transitions from a level m > 2 to n = 2. Transition to the ground level n = 1 are called Lyman series and are in the UV, cf. also Fig. 2.1. Apart from transition between two discrete levels, a free electron with E > 0 can become bound or a bound electron can be emitted by an atom. These are called free-bound or boundfree transition. In this case, the energy of the emitted or absorbed photon depends on the kinetic energy of the free electron and can be arbitrary.

2 Spectral lines and their formation Ex.: Find the energy and radius of the ground state of a hydrogen atom.
Insert n = 1 in (2.6) and (2.4). Simplest in natural unit ( = c = 1) with me = 512 keV and GeV1 = 5.06 1013 cm (cf. Appendix B), and Sommerfelds ne structure constant = e2 /( c) 1/137, [in SI units = e2 /(40 c) 1/137] E1 =
2

2 512keV me4 = me c2 = 13.6eV , 2 2 2 2 1372

r1 =

me2

1 137 cm = 5.3 109 cm . me c 0.512 103 5.06 1013

Kirchhos laws Using Bohrs model we can understand the laws deduced by Kirchho empirically: A dense gas or a solid body emits continuous radiation without spectral lines. The atoms and the radiation in the body are in thermal equilibrium, and the emitted radiation follows the Planck distribution. A hot, diuse gas produces emission lines. The lines correspond to the energy dierence of atomic levels n m < n, photons are emitted.

A cool, diuse gas in front of a continuous sources produces absorption lines. Photons which energy correspond to the the energy dierence of an appropriate atomic level are absorbed.

2.2 Formation of spectral lines

A prominent line in the solar spectrum corresponds to the n = 2 n = 3 transition and is called Balmer- or H. The observation of this line means that a certain fraction of hydrogen in the solar atmosphere must be in the second, excited state. How are the populations of the dierent excited states governed? For a classical gas in thermal equilibrium, the population ni of the state i with energy Ei and degeneracy gi is given by a Boltzmann distribution, with ni = Z 1 gi exp(Ei /kT ) , (2.10)

where Z can be thought of as a normalization constant. Excitations with energy E kT are exponentially suppressed. For the population of an excited stated n > 1 relative to the ground state n1 it follows ni gi Ei E 1 = exp n1 g1 kT . (2.11)

Thus if (E2 E1 ) kT , then the state n = 2 is only weakly populated and the H line is faint. Increasing the temperature so that (E2 E1 ) kT , the relative occupation of the level n = 2 and the strength of the H line increases too. Increasing the temperature even further, electrons populate higher and higher levels, reducing the relative population of the n = 2 level and in turn the strength of the H line. Finally, hydrogen becomes ionized, when the temperature is higher than its binding energy. Thus the following picture for the strength of the H line emerges:

2.3 Hertzsprung-Russel diagram Spectral class O B A F G K M Temperature Te /K 30000 20000 10000 7000 6000 4000 3000 main spectral feature hydrogen faint, ionized helium hydrogen moderate, neutral helium moderate hydrogen strong, neutral helium very faint hydrogen moderate, single ionized heavy elements hydrogen relatively faint, single ionized heavy elements hydrogen faint, single ionized heavy elements hydrogen very faint, neutral atoms, molecules

Table 2.1: Important spectral features of main-sequence stars. For a cold star, kT E1 E2 , most hydrogen is in the ground state; the H line is very weak. Increasing the temperature, kT (E2 E1 ), the n = 2 level becomes populated; the H line becomes strong. Increasing further the temperature, kT > (E2 E1 ), higher levels become populated and n = 2 level less; the H line becomes weaker. Increasing even further the temperature, kT > E1 , Hydrogen becomes ionized; the H line disappears. Stars were classied according to the strength of the H lines as A, B, F, G, K, M, O. These classes are additionally subdivided into A0, A1, . . . A9, B0,. . . . Ordered by (decreasing) temperature, the spectral classes form the sequence O, B, A, F, G, K, M, cf. also table 2.1. (A mnemonic is oh, be a ne girl/guy, kiss me.) For the visible n = 2 transitions, it follows E2 E1 = (1/4 1) 13.6 eV or (E2 E1 )/k 1.2 105 K.

Thus in the solar atmosphere with Te 5700 K only the tiny fraction 2/8 exp(20.8) 1010 of all hydrogen atoms is in the n = 2 level. Therefore the Balmer lines in the solar spectrum are relatively faint, although the Sun consists mainly of hydrogen. On the other hand, the ionization energy of calcium is 6.11 eV, and thus a factor 2 smaller than the one hydrogen. Since the population of the dierent levels depends exponentially on the energy, already a factor 2 leads to large dierences: Practically all calcium atoms in the solar atmosphere are single ionized and in the ground state, able to absorb a photon in the visible wave-length range. More generally, the single ionized lines of relatively heavy elements as calcium are prominent in F, G and K stars, cf. Table 2.1.

2.3 Hertzsprung-Russel diagram

Between 1905 and 1913, Hertzsprung and Russel examined the distribution of stellar luminosities for dierent spectral classes. The latter quantity characterizes the surface temperature T , and since the luminosity is proportional to T 4 and the stellar surface R2 , one expects a wide distribution in this diagram. As shown in Fig. 2.2, this is not the case and stars appear only in well-dened regions of the Hertzsprung-Russel (HR) diagram. Thus arbitrary combinations of T and L, and thus of R and L are not allowed.

2 Spectral lines and their formation Most stars are found in a narrow band from the left-upper to the right-lower corner, the socalled main sequence. Stars of the main sequence of the same temperature have approximately the same luminosity and thus also radius. This correlation has to be explained by any theory of stellar astrophysics. Apart from the main sequence, there are two other star populations. Above the main sequence, there are the giant and super giant stars: They are more luminous for the same temperature as main sequence stars, hence their radius has to be larger. Much below the main sequence stars (10 magnitudes fainter) and with roughly white spectral color, there are the so called white dwarf stars. There exists various variations of the original Hertzsprung-Russel diagram. For instance, the color index can replace the temperature or the apparent magnitude can be used as measure for the luminosity, if one discusses only stars from the same cluster. The color indices in Eq. (1.4) are normalized so that their dierences m1 m2 are zero for stars of the spectral class A0. Note also the colors U, B, V,. . . denote apparent magnitudes, although astronomers use capital letters for them.

Exercises
1. Consider a gas of neutral hydrogen gas at temperature T , described by a Boltzmann distribution ni = Z 1 gi exp(Ei /kT ) with gn = 2n2 . i) At what temperature T will equal number of atoms have electrons in the ground state and in the second excited state (n = 3)? ii) What predicts the Boltzmann distribution for T ? What happens in reality?

2.3 Hertzsprung-Russel diagram

Figure 2.2: A Hertzsprung-Russel diagram. The position of a star along the main-sequence denes uniquely all important stellar parameters: Mass, radius, temperature, lifetime.

3 Telescopes and other detectors

Progress in astronomy has been driven largely by the exploration of new wavelengths ranges of the electromagnetic spectrum. The rst discovery of an extraterrestrial signal beyond the visible light was the observation of cosmic rays by Victor Hess 1912, whose origin is still unresolved. In 1935, Karl Jansky discovered an extraterrestrial radio signal and could associate it to the center and the plane of the Milky Way. After world war II, the radio band has been studied systematically. Starting from the seventies, X- and gamma-ray astronomy as well as the IR wave-lengths have been explored with satellites. Since the last decade, high-energy air Cherenkov telescopes are detecting photons with energies E > 100 GeV/c2 . This extension of the explored wave-length range made it also possible to study non-thermal radiation. At the same time there has been a continuous improvement of light receivers: The naked eye was replaced rst by photo plates and then by charge-coupled devices (CCD) similar to those used in present days digital photo cameras. Another important improvement has been the use of computers for data correction and analysis. Further progress is expected by attempts to go beyond electromagnetic radiation: There are large eorts underway to use charged particles (cosmic rays), neutrinos (from the Sun, Galactic supernovae or other sources), and gravitational waves as new means to study the Universe.

3.1 Optical telescopes

Optical telescopes can be divided into two principal classes: i) refracting telescopes using lenses and ii) reecting telescopes using mirrors (plus lenses). From the refracting telescopes two principal subclasses exist: Keplers version with two concave lenses and Galileos version with one concave and one convex lens.

3.1.1 Characteristics of telescopes

Magnication V : Many objects (binary stars, galaxies) that appear point-like to the naked eye show structure viewed by a telescope. From the left panel of Fig. 3.1, it follows that the magnication V depends on the ratio of the focal lengths fi and the apertures Di as V = 2 f1 D1 = = . 1 f2 D2 (3.1)

If an objects subtends the angle to the unaided eye, a telescope magnies this angle by the factor V . Resolution: The smallest angular distance at which two object can be separated in the image is called resolving power or resolution of the telescope. A lower limit is set by diraction, i.e. the interference of light rays from dierent positions in the aperture plane of the telescope.

3.1 Optical telescopes

Figure 3.1: Left: Schematic view of a refracting telescope (Keplerian version). Right: Diraction pattern. Consider for an one-dimensional example, when destructive interference happens: We divide the aperture into 2n segments of length d = D/(2n) and pair light rays separated by d. The phase dierence between the pairs is l sin = D/(2n) sin and thus destructive interference happens for D sin = 2n 2 or n sin n = n , D n = 1, 2, . . . (3.2)

For a circular aperture D, the rst diraction minimum is at 1 1.22/D. Two objects can be resolved, if their angular distance is larger than 1 (Rayleigh criteria). For instance, a 5 m optical telescope has a resolution of 1 1.22 550 109 m 602 180 1.22 1.34 107 0.03 , D 5m (3.3)

while the human eye resolves objects with 1 angular distance. Light gathering power The aperture or diameter D of the telescope determines the collecting area for light that is then focused on the receptor surface. Compared to the human eye with D 5mm, a 5 m telescope allows to see fainter objects with magnitude dierence m1 m2 = 2.5 log
2 D1 2 D2

= 2.5 log

25 25 106

= 15 ,

(3.4)

i.e. with a 5 m telescope stars are visible down to m = 6 + 15 = 21. The light gathering power is a measure of how strongly light is concentrated on the receptor surface and depends on the kind of object considered. The apparent brightness b of the image of a point-like object scales as b D2 : The amount of light gathered increases D2 and is still collected on a single pixel of a CCD or on one nerve of the human eye. an extended objects is constant: The increase of collected light is compensated by the larger picture of the object.

3 Telescopes and other detectors Exposure and quantum eciency With photo plates or photoelectric detectors one can collect light for hours, while the naked eye takes exposures of 1/20 seconds. Larger exposure extend thus additionally the limiting magnitude of an telescope. While the eye or a photo plate has a quantum eciency of about 1%, i.e. 1% of all photons are detected, modern detectors called charge-coupled devices (CCD) come close to a quantum eciency of about 100%. These are semiconducting detectors where a photon creates (similar as in the photoelectric eect) electrons. Spectrographs Simplest example is a prism that has typically a wave-length dependent refractive index, n 2 . Positive interference occurs in a diraction grating with grating constant 1/D when the path length dierence between neighboring rays is D(sin i + sin f ) = m, m = 1, 2, . . . (3.5)

3.1.2 Problems and limitations

Chromatic aberration Since the refractive index n of a lens (as any material) is wavelength dependent, a picture of a refracting telescopes cannot be focused on a single common point for dierent wavelengths. Together with the easier production and support of mirrors, this is the main reason for the dominance of reecting telescopes. Surface quality: The surface of mirror or lenses has to be accurate generally to /20. For an optical telescope, this corresponds to /20 25 nm, i.e. around 250 atom layers. Similarly, deviations of the various telescope parts due to thermal stress or gravity should be smaller than /20. Atmospheric seeing: For large ground-based telescopes, the resolution is limited by atmospheric seeing: Turbulence in the air leads to time-dependent refractive index along dierent lines-of-sight, n(x) + n(x, t), and causes the images of point-sources to break up into speckle patterns, which change very rapidly with time. This limitation can be overcome partially by placing telescopes at high-altitude observatories such as Mauna Kea or La Palma and using adaptive optics or image correcting with the help of computers for ground-based telescopes. Alternatively, one can go above the atmosphere, i.e. to space.

3.2 Other wave-length ranges

Radio telescopes The simplest realization would be a /2 antenna where electromagnetic waves induces resonance currents in it. To increase the sensitivity, one builds large parabolic antenna with the radio receiver in its focal point. Then the signal is electronically amplied. Since the wavelengths is in the cm to meter range, the parabolic antenna can be made as a simple metallic grid. This allows the construction of large dishes up to 100 m. On the other hand, the resolution of a radio telescope is according to Rayleighs criteria and thus very poor compared to optical telescopes. This problem can be partially overcome by combining the information from telescopes distributed in an array or even on dierent continents (Very Long Baseline interferometry).

3.2 Other wave-length ranges

Figure 3.2: The MAGIC air Cherenkov telescope: lasers are used as control for the adaptive optics.

Instruments for high-energy astronomy Extending observations to the UV and higher energies meets three main obstacles: First, the Earths atmosphere is opaque to photons with energy above 10 eV, meaning that to observe photons in the UV and beyond directly requires the placement of a detector above the Earth atmosphere. A major turning point in X-ray and gamma-ray astronomy was therefore the launch of the rst satellite-borne telescope, SAS-2, in 1972. As an additional obstacle, geometrical optic ceases to work for photons in the X-ray range (0.1 keV < E < 100 keV) and above. Finally, the ux of photons decreases with energy strongly. Small angle reection is still possible for X-rays, if the incidence angle is very small, and thus one uses a special type of grazing incidence mirror telescopes. Gamma-rays (E > MeV and higher energies) can produce e+ e pairs in the eld of a nucleus. These electrons and positrons are either detected directly or, especially if E MeV, produce secondary photons forming thereby an electromagnetic cascade. Main limitation of satellite experiments is their small collection area, which limits their use to energies < 100 GeV. On the other hand, 100 GeV is the limit where the electromagnetic cascade in the earths atmosphere from the initial photon can be detected. While the cascade dies out high in the atmosphere below 10 TeV, showers are still detectable via the Cherenkov1 emission of relativistic electrons and muons. An example for such an atmospheric Cherenkov telescope is shown in Fig. 3.2.
1

A charged particle that moves in a medium with refractive index n > 1 faster than the speed of light, v > c/n, emits electromagnetic Cherenkov radiation.

3 Telescopes and other detectors

105 LEAP 1 I(E) 1/(m2 s sr GeV) 10-5 10-10 10-15 10-20 HiRes 10-25 10-28 108 1010 1012 1014 E/eV 1016 1018 1020 Akeno PROTON

Figure 3.3: The dierential intensity I(E) of cosmic rays measured by four dierent experiments as function of kinetic energy E. The energy region inuenced by the Sun is marked yellow and an 1/E 2.7 power-law is also shown as dashed line.

3.3 Going beyond electromagnetic radiation

In principle, all stable particle types may be used to obtain information about the universe: Additionally to photons, these are the charged elementary particles (electrons e and protons p) and the stable nuclei. Moreover, the known stable neutral particles are neutrinos and the graviton. Charged particlesCosmic rays Victor Hess discovered 1912 on a balloon ight that ionizing radiation increases with altitude. He performed observations also during a solar eclipse, but found no decrease. Hence he concluded that the Sun can hardly be considered the source. Later, the study of their deections in the geomagnetic eld showed that most cosmic rays are protons. The cosmic ray spectrum, a nearly featureless power-law extending over eleven decades in energy up to a few 1020 eV, is shown in Fig. 3.3. The power-law form of the cosmic ray spectrum indicates that they are produced via non-thermal processes, in contrast to all other radiation sources like e.g. stars or (super-) novae known until the 1950s. Since our galaxy, the Milky Way, contains a magnetic eld with strength 3G in the Galactic disc of thickness d 100 pc, charged particles are deected: In a homogeneous magnetic eld, particle with charge Ze gyrate with Larmor radius RL = cp E G = pc . ZeB B Z 3 1015 eV (3.6)

Thus one cannot associate sources to the observed arrival direction of cosmic rays, except perhaps at the highest energies, E > 1019 eV or the last energy decade shown in Fig. 3.3. 30

3.3 Going beyond electromagnetic radiation

3.3.1 Neutrinos
The main advantage of neutrinos compared to photons is the extreme weakness of their interactions. For instance, a neutrino produced inside the Sun interacts only with a probability 109 on its way out. Thus neutrinos can provide information about the interior of the stars, while photons inform us only about their surface. Neutrinos are produced in weak interactions like nuclear beta decay n p + e + e that are part of the fusion processes in stars as we will discuss later in detail. Therefore they are copiously in the interior of the Sun or a supernova and are a key tool to test our theories of stellar structure and evolution. The weakness of their interactions is also their main disadvantage compared to photons: They are extremely dicult to detect. Solar neutrinos are observed since the mid-1960s, and in 2002 Raymond Davis Jr. and Masatoshi Koshiba won the Nobel Prize in Physics for the detection of cosmic neutrinos.

3.3.2 Gravitational waves

Analogously as Maxwells equations predict electromagnetic waves, general relativity predicts gravitational waves as a distortion of space-time. Passing of a gravitational wave leads (similar to a Lorentz contraction) to a time-dependent change of length scales, h(t) = L(t)/L. This change h is proportional to the amplitude of the gravitational wave, and thus the distortion is inversely proportional to the distance, h 1/d. Increasing the sensitivity of a gravitational wave experiment by a factor increases the number of potential sources therefore as 3 , in contrast to an increase as 3/2 for photon or neutrino experiments. Since gravity is very weak, the expected deviations are tiny, h < 1020 . There are several laser interferometer aiming at detecting gravitational waves.

4 Basic ideas of special relativity

4.1 Time dilation a Gedankenexperiment
Consider a light clock, i.e. a light pulse traveling between two mirrors separated by the distance L. The time for a round trip is t0 = 2L/c. Now consider the clock, i.e. the two mirrors, moving with velocity v, cf. Fig. 4.1. Then the light needs for bouncing once back and forth the time t > t0 , t2 = 4 4L2 L2 + (vt/2)2 = 2 1 + c2 c
t2 0

vt 2L
vt/ct0

= t2 + (v/c)2 t2 . 0

(4.1)

Solving for t gives t=

t0 [1 (v/c)2 ]1/2

(4.2)

i.e. the famous time dilation formula. Is this just a property of a badly constructed clock or of fundamental importance? The answer given by experiments is that time for fast moving observers indeed slows down.

4.2 Lorentz transformations and four-vectors

4.2.1 Galilean transformations
Newtons gravitational law as well as the Coulomb law are examples for an instantaneous action, x xi F(x) = K . (4.3) |x xi |3
i

The force F(x, t) depends on the distance x(t) xi (t) to all sources i (electric charges or masses) at the same time t, i.e. the force needs no time to be transmitted from xi to x. The force law is invariant under changes x = Rx + d, if |r | = |r|. The translation d could be time-dependent via d = vt. The condition that the norm of the vector r is invariant, r = r , restricts the possible transformation matrices R, rT r = rT RT Rr = rT r , (4.4)

as RT R = 1. (T denotes the transposed vector or matrix). Hence rotations R and translations d keep a force law of type (4.3) invariant. In two dimensions, the rotation matrix R can be chosen as cos sin . (4.5) R= sin cos

4.2 Lorentz transformations and four-vectors

Figure 4.1: A clock consisting of a light pulse travelling between two mirrors; left at rest and right moving with velocity v.

4.2.2 Lorentz transformations

Comparing the non-relativistic Poisson equation = 0 and Maxwells wave equation, ( 2 (1/c2 )t )A = 0 suggests that the latter are invariant under coordinate transformations that keep the space-time interval s2 x2 c2 t2 = x 2 c2 t 2 (4.6)

invariant. This means that both the time and the space dierence between two events measured in the coordinate system K and K may be dierent, but that the dierence s2 x2 c2 t2 is the same for all inertial systems. Here we call inertial system all coordinate systems that move force-free. We want now to determine the equivalent to the rotation matrices R that are called Lorentz transformations L. If we replace t by it in s2 , the dierence between two space-time events becomes the normal euclidean distance. Similarly, the identity cos2 + sin2 = 1 for an imaginary angle = i becomes cosh2 sinh2 = 1. Thus a close correspondence exists between rotations R in euclidean space that leave x2 invariant and Lorentz transformations L that leave s2 invariant. We try therefore as a guess x = ct sinh + x cosh , y = y ,

(4.7) (4.8) (4.9)

z=z ,

ct = ct cosh + x sinh .

Compared to normal coordinate rotation in euclidean space, the trigonometric functions are replaced by hyperbolic ones. Consider now in system K the origin of the system K . Then x = 0 and x = ct sinh , ct = ct cosh . Dividing the two equations gives x = tanh . ct

(4.10) (4.11) (4.12)

4 Basic ideas of special relativity Since x/t is the relative velocity of the two systems, we have identied the physical meaning of the imaginary rotation angle as tanh = Using the following identities sinh x = tanh x 1 tanh2 x 1 1 tanh x x + vt 1 (v/c)2 ct + v/cx 1 (v/c)2
2

v . c

(4.13)

1 2 1 1 2

(4.14)

cosh x =

(4.15)

together with (4.13) in (4.7) gives the standard form of the Lorentz transformations, x = ct = = (x + ct ) = (ct + x ) . (4.16) (4.17)

4.2.3 Energy and momentum

Another important example for a four-tupel consisting (from a non-relativistic point of view) of a vector and a scalar and transforming like (ct, x) under Lorentz transformations, is (E, cp). The Lorentz transformation written in this case is E = (E + cp ) x cpx = = (cp x + E ) .

(4.18) (4.19)

If we choose as system K the rest system of the particle, then p = 0 and = 1. If we would x set the rest energy E E0 equal to zero, then the energy E would be zero for arbitrary . The correct value of E0 can be obtained by considering the non-relativistic limit, = and E0 E0 + 1 2 1+ 2 (1 2 )1/2 E0 v 2 ! mv 2 . = const. + 2 2c 2 (4.20)

(4.21)

Thus we obtain the correct non-relativistic limit, if a particle at rest has the energy E0 = mc2 .

4.2.4 Doppler eect

We consider as last example how the wave-vector of a photon (/c, k) transforms under a Lorentz transformation, V c c kx = . (4.22) c [1 (V /c)2 ]1/2 34

4.2 Lorentz transformations and four-vectors Setting kx = k cos = cos , where the angle between the wave vector and the direction c of the movement of the source, we obtain the relativistic Doppler formula 1 (V /c)2 = . 1 (V /c) cos
1/2

(4.23)

For small relative velocities, v c, the quadratic term in the nominator can be neglected. Thus one reproduces the non-relativistic Doppler formula where only the radial relative velocity enters. Aberration and beaming Classical aberration was discovered 1725 by James Bradley: He discovered and later explained an apparent motion1 of celestial objects caused by the nite speed of light and the motion of the Earth in its orbit around the Sun. Assume e.g. to observe a star at the moment when it is right in the zenith. Since the Earth moves, you have to point the telescope a bit ahead and thus the position of stars changes in the course of one year. To obtain the relativistically correct form for the aberration formula, we rewrite rst the Lorentz transformations for x and t in dierential form, dx = (dx + cdt ) , dy = dy ,

(4.24)

dz = dz ,

(4.25) (4.26)

cdt = (cdt + dx ) . Then we obtain the transformation law for velocities u as u + v dx u = = dt 1 + vu /c2 u = u dx = . dt (1 + vu /c2 )

(4.27)

(4.28)

The so-called aberration formula connects the angle between the velocities u and u in the two frames, u sin u . (4.29) = tan = u (u cos + v) The most interesting case is the aberration of light. We set = /2, i.e. we ask how light emitted into one half-sphere appears in a dierent inertial system. With u = u = c it follows then c tan = (4.30) v or 1/ for v c. Thus a fast moving source emits photons mainly in a cone of opening angle 1/ around its forward direction.

This motion depends on the velocity of the observer perpendicular to the line-of-sight to the star, but is independent of the distance to the star. The deviation induced by classical aberration is much larger than the parallax we discussed already.

5 Binary stars and stellar parameters

We have discussed up to now the (surface) temperature T , the luminosity L, and the radius R as basic properties of a star. The remaining one, the stellar mass M , can be only determined by its gravitational inuence on another body. Fortunately, 50%80% of all stars are gravitationally bound in multiple star systems, most of them in binary systems. For instance, 50 binaries, 14 triple and three systems with four stars were known up to a distance of 10 pc from the Sun in the year 1980. For solar-type stars, the ratio of single:double:triple:quadruple stars is 45:46:8:1. On the other hand, most dwarf stars that are more dicult to detect are isolated stars. Classication of binary stars: Apparent or optical binaries are not really binary stars, but random coincidences along the same line-of-sight. A visual binary is a bound system that can be resolved into two stars by a telescope. William Herschel discovered 1803 the rst example when he noted that the fainter component of Castor (Castor B) had moved relative to the brighter component (Castor A). This was the rst indication that gravity acts outside the solar system. In an astrometric binary system only the brighter star is visible. Its position wobbles back and forth, because the star is moving on a periodic orbit. In a spectroscopic binary, the wavelength of spectral lines oscillates periodically due to Doppler shift. The rst case was detected 1889 by Pickering who found that the spectral lines of the brighter component of Mizar (Mizar A) were usually double and the spacing varied with a regular 104 day period. An eclipsing binary is a system that becomes periodically brighter and fainter. It is dimmed when the faint companion passes in front of the main star. Thus we are aligned to the plane of the orbit. An example visible by eye is Algol: Algol fades at intervals of 2d 20h 49m to a third of its normal brightness, then after a few hours returns to normal brightness. In a composite spectrum binary are lines of two stars with dierent spectral type mixed. Before we can use binary star systems as tools to determine stellar masses and radii, we have to review the most prominent problem of classical mechanics and astronomy, namely the gravitational two-body or Kepler problem.

5.1 Keplers laws

Kepler developed his three laws of planetary motions empirically from the observations of Tycho Brahe. The laws are 1. Planets orbit the Sun in ellipses, with the Sun in one of the two focuses.

5.1 Keplers laws 2. The line connecting the Sun and a planet sweeps out equal area in equal time. 3. The Harmonic Law states the squared orbital period P of planets measured in years equals to the third power of their major axis measured in astronomical units, (P/yr)2 = (a/AU)3 . Kepler noted already that his laws describe also the motion of the Saturn moons, if in the Harmonic Law the appropriate units are used. Newton used later the Harmonic Law to derive the 1/r2 dependence of the gravitational force. We will follow the opposite way and discuss how Keplers laws follow from Newtons law for gravitation. Ellipses: An ellipse may be dened by the condition r + r = 2a, i.e. as the set of points with a constant sum 2a of the distances r and r to the two focal points F and F . Additionally to its major axis a, an ellipse is characterized either by its minor axis b or its eccentricity e. The latter two quantities can be connected by considering the two points at the end of the minor axis b: Then r = r = a and r2 = b2 + (ae)2 or b2 = a2 (1 e2 ) .
s

(5.1)

r r
s

Any point of an ellipse can be specied by the distance r to one of its focal points and an angle that is measured counter-clockwise beginning from the major axis. From the gure, one obtains immediately (with cos(180 ) = cos()) r2 = r2 + (ae + a)2 + 2r(2ae + a) cos . Eliminating r with the help of r + r = 2a and solving for r one obtains r= a(1 e2 ) . 1 + e cos (5.3) (5.2)

As starting point, we recall how a two-body problem can be reduced to an one-body problem in the case of a central force.

5 Binary stars and stellar parameters Denoting the positions and the masses of two stars with mi and ri , i = 1, 2, respectively, the equations of motions are m11 = f (|r1 r2 |)(r1 r2 ) r (5.4) (5.5)

m22 = +f (|r1 r2 |)(r1 r2 ) . r Adding Eqs. (5.4) and (5.5), (m11 + m22 ) = (m1 + m2 )R = 0 , r r

(5.6)

i.e. we see that the center of mass (cm) moves freely. In the second step, we introduced relative coordinates r = r R and cm coordinates, R= m1 r1 + m2 r2 . m1 + m2 (5.7)

Multiplying Eq. (5.4) by m2 and (5.5) by m1 and subtracting the two equations, we obtain = f (r)r , r where we introduced the reduced mass = m1 m2 . m1 + m2 (5.9) (5.8)

Thus we can solve an one-body problem for the reduced mass moving with the distance r in the gravitational eld of the mass M = m1 + m2 .

5.1.1 Keplers second law: The area law

Consider the movement of a body under the inuence of a central force, r = f (r) . r r (5.10)

Since r r = 0, vectorial multiplication with r should simplify Eq. (5.10). Doing it, we obtain r = 0 , r that looks already similar to a conservation law. Since d r (r r) = r r +r dt
=0

(5.11)

(5.12)

we obtain indeed the conservation of angular momentum L = r r for the movement in a central potential, d d r = r (r r) = L = 0. (5.13) dt dt There are two immediate consequences: First, the movement is always in the plane perpendicular to L. Second, the area swept out by the vector r is dA = and thus also constant. 1 1 r vdt = dL 2 2 (5.14)

5.1 Keplers laws

5.1.2 Keplers rst law

We introduce the unit vector = r/r and rewrite the denition of the angular momentum L r as d r d r d = r2 r r r r . (5.15) L = r r = r (r) = r r + r r dt dt dt The rst term in the parenthesis vanishes, because of = 0. Next we take the cross r r product of the gravitational acceleration, a= with the angular momentum, aL= d r GM r2 r r 2 r dt = GM r r d r dt . (5.17) GM r r2 (5.16)

With the identity A (B C) = (A C)B (A B)C, it follows a L = GM r d r d r ( ) r r r . dt dt (5.18)

Because is a unit vector, = 1 and d( )/dt = 0, and thus r r r r r d r . dt Since L and GM are constant, we can rewrite this as a L = GM d d (v L) = (GM ) . r dt dt Integrating gives where the integration constant C is a constant vector. Taking now the dot product with r, we obtain v L = GM + C , r (5.21) (5.19)

(5.20)

r (v L) = GM r + r C . r r Applying next the identity A (B C) = (A B) C, it follows (r v) L = GM r + rC cos = GM r 1 + C cos GM .

(5.22)

(5.23)

Expressing r v as L/, dening e = C/(GM ) and solving for r, we obtain nally the equation for a conic section, i.e. Keplers rst law, r= In the case of an ellipse, r= and thus we obtain as angular momentum L = GM a(1 e2 ) . (5.26) L2 /2 . GM (1 + e cos ) a(1 e2 ) 1 + e cos (5.24)

(5.25)

5 Binary stars and stellar parameters

5.1.3 Keplers third law

We integrate Keplers second law in the form of Eq. (5.14) over one orbital period P , A = ab = Squaring and solving for P , it follows P 2 = 4 2 (ab)2 . L2 (5.28) L P. 2 (5.27)

With b2 = a2 (1 e2 ) and Eq. (5.26) for the angular momentum L, we obtain Keplers third or harmonic law, 4 2 a3 . (5.29) P2 = G(m1 + m2 )

5.2 Determining stellar masses

Visual binaries The period of a visual binary system can be determined directly, if it is observed long enough. Note that periods are typically long, from years up to 1001000 yr, because the angular separation of the two stars should be large. The plane containing the orbit may be inclined relative to the plane orthogonal to the line of sight. This angle is called inclination i and is zero when the two planes coincide and i = 90 when the orbit is seen edge-on. The inclination and the true orbit can be reconstructed from the apparent orbit, if good enough observations exist: A tilted ellipse projected on the celestial sphere remains an ellipse, but with a changed eccentricity. Therefore the stars seem to move not around the foci, and this discrepancy can be used to nd the true ellipse. In order to convert the observed angular distances into true lengths, the distance to the binary system has to be known. The total mass follows then from Keplers 3. law. Moreover, the ratio of masses can be determined from the ratio r1 /r2 (or via Doppler shift from v1 /v2 , if the system is also a spectroscopic binary). Spectroscopic binaries and Doppler shifts If two stars orbit their cms closely (AU) and fast (P hours-months) and moreover the inclination i is not close to zero, the Doppler shift of their spectral lines can be detected. The latter condition arises, because in the non-relativistic limit v c or 1 1 (v/c)2 0 1 1 = = , 1 (v/c) cos 1 (v/c) cos 1 vr /c
1/2

(5.30)

where vr is the radial velocity and was the angle between the wave vector/propagation direction and the direction of movement. Rearranging this equation and introducing = 0 it follows vr = = . (5.31) c If the orbit is circular, then the curve will be a sine curve. For an elliptical orbit, the shape of the radial velocity curve depends on the eccentricity and the inclination of the ellipse.

5.2 Determining stellar masses For simplicity, we assume orbits with small eccentricity. Then from ri = P/(2)vi , and a = r1 + r2 = P/(2)(v1 + v2 ) and thus Keplers 3. law becomes 2G(m1 + m2 ) = P (v1 + v2 )3 = P (v1r + v2r )3 . sin3 i (5.32)

with vr = v sin i. If the inclination i is not known, one can derive at least a lower limit on m1 + m2 using i = 90 . Larger samples of spectroscopic binaries allow one to improve these limits by performing a statistical analysis. One assume an isotropic distribution of the inclination angle and calculates then the expectation value of sin3 i. Since there exists an observational bias to discover values of i close to 90 , the estimation has to be corrected for such a selection eect. Often only the Doppler shift of one of the two components can be measured. Eliminating e.g. v2 using the cm condition (assuming again for simplicity circular orbits) gives v2 = Then v 1 + v2 = v 1 1 + Now Keplers law can be rewritten as P m3 sin3 i 3 2 v1r = . 2G (m1 + m2 )2 The RHS is called mass function and is again studied with statistical methods. Eclipsing binaries An eclipsing binary star is a binary system in which the orbit plane of the two stars lies nearly in the line-of-sight of the observer, sin(i) 1, so that the components undergo mutual eclipses. Thus they appear as variable stars, with periodic drops in intensity when the less luminous star obscures the other. (5.35) m1 v1 . m2 = v1 (m1 + m2 ) . m2 (5.33)

m1 m2

(5.34)

5.2.1 Mass-luminosity relation

For any binary system with derived mass also the distance is known. Thus, the luminosity can be derived from the apparent magnitude. When the observed luminosity for stars in binary systems is plotted as function of their mass, the so-called mass-luminosity relationship is obtained. Eddington derived 1924 for main-sequence stars theoretically using a rather crude model M L (5.36) = L M with 3. This means that the main-sequence in the Hertzsprung-Russel diagram can be indeed characterized by essentially a single parameter, e.g. their mass. From observations, one nds a broken power law with 1.8 for M < 0.3M , (5.37) (5.38) (5.39)

2.8 for M > 3M .

4.0 for 0.3M < M < 3M ,

5 Binary stars and stellar parameters From observations one tries to measure the number of stars as function of their mass, correcting for the bias to see more easily massive and thus brighter stars. From this, taking into account that massive stars live shorter, one can construct the initial mass function that is an important theoretical ingredient in tests of stellar evolution models.

5.3 Stellar radii

Only for the Sun, the radius can be directly determined from its angular radius and the distance. Other stars cannot be resolved as extended objects from the ground because of atmospheric seeing. If the temperature and the luminosity of a star are known, the StefanBoltzmann law can be used to dene the luminosity radius. Eclipsing binaries oer another possibility to measure stellar radii, since their light-curve depends directly on the radii of the two stars. As result, one nds that the MS stars become larger with increasing temperature, cf. Fig. 2.2.

6 Stellar atmospheres and radiation transport

6.1 The Sun as typical star
The Sun is a rather typical main-sequence star, distinguished only by its proximity to the observing astronomers on Earth. Solar radius R : From the distance D and the apparent radius of the solar disk, = 15 59.63 , one obtains R = D = 696000 km. Solar mass and density: Keplers Third Law relates the semi-major axis a of their orbits, the period P and the total mass of two gravitating bodies, a3 G = 2 (m1 + m2 ) . P2 4 (6.1)

This gives with M M , M = 1.989 1033 g. The average solar density follows as M = 4/3R3 = 1.409 g/cm3 .

Solar temperature: Measuring the Solar Constant, i.e. the energy ux of the Sun at the Earths position outside the atmosphere, one can use the Stefan-Boltzmann law to dene the eective solar temperature, T = 5762 K. Magnetic eld, sunspots: Magnetic eld strengths can be determined spectroscopically via the Zemann eect: The energy levels of atoms split with E = gmJ L , where L = eB/(2me ) is the larmor frequency, g the Lande factor and |mJ | J is the magnetic quantum number. The Sun seems to have no smooth large-scale magnetic eld; magnetic elds are concentrated in thin ux tubes. Sunspots are the ends of such ux tubes at the solar atmosphere and have eld strength up-to one T = 104 G. The abundance of sunspots varies with an average period of 11 years. With the same period, the magnetic eld of the Sun reverses sign. Rotation period: Observing sunspots, the rotation period P of the Sun can be determined. It is found that the Sun rotates dierentially, at the equator with P = 24.8 d and e.g. at latitude 70 with P 31 d. Structure: Those layers that give rise to the continuous radiation and the major part of the Fraunhofer lines are called photosphere. During total solar eclipses, one can observe the higher layer of the atmosphere. This layer is called chromosphere and is characterized by the absence of continuum and absorption lines, showing only emission lines. Afterwards the corona follows stretching outwards into the interplanetary medium.

6.2 Radiation transport

Let us idealize atoms as black discs of radius R. Then any photon hitting the disc with area R2 is absorbed. In other words, the cross-section for the absorption of a photon by a single atom is = R2 .

6 Stellar atmospheres and radiation transport Consider now a cylinder of length l and area A, lled with N atoms. Their number density n is n = N/(Al). Let us assume rst for simplicity that N A, i.e. the atoms do not overlap much and are uniformly distributed. Then the fraction of incoming radiation absorbed in the cylinder is simply N = nl (6.2) A and denes the optical depth nl. Our assumption N A corresponds to 1, an optical thin or transparent source in contrast to an optical thick source with 1. For arbitrary , we have to calculate how much radiation is absorbed passing the innitesimal distance dl, dI = Id = Indl (6.3) or Integrating gives dI = ndl . I
I I0

(6.4) I = I0 dln .

dI = I

l 0

dl n or

(6.5)

and nally I = I0 exp( ) I(l) = I0 exp (6.6)

For small , expansion of the exponential reproduces our old result: From I0 exp( ) I0 (1 + . . .), one sees that the intensity is reduced by I0 , i.e. the fraction of absorbed photons is given by . If n does not depend on the distance l, the optical depth is simply obtained by multiplying n with the path length, as in our original denition (6.2). In general, n has to be integrated along the path length l. This equation can be seen as denition and measurement description for the cross section . What is the typical size of the cross section for the absorption of photons? A reasonable 2 estimate for the excitation of an atom is r1 choosing r1 = 2 /(me ) as the rst Bohr radius, while the cross section for scattering pf photons on free electrons can be estimated 2 as r0 , where r0 = /(me c) is the classical electron radius. A theory of the radiation transport in stars needs as input scattering cross-section of photons with atoms and ions. One introduces often the opacity = /l = n. Then, Eq. (6.3) becomes dI = Idl . (6.7)

Photons can not only be absorbed, but have to be also re-emitted by atoms. Hence, we should add a source term S to the previous equation, 1 dI = I + S , dl (6.8)

where S accounts e.g. for the de-excitation of atoms. For instance, a photon can be absorbed exciting an atom to the level n, de-exciting then in several steps emitting several photons of lower energy. Ex.: A beam of light with intensity I0 enters a gas layer with constant opacity and source function
S. Derive I(l) and sketch I(l) as function of = l for S = 2I0 and S = I0 /2. We have solved already Eq. (6.8) for S = 0, where we obtained I(l) = I0 exp(l). The general solution

6.3 Diusion and random walks

for S = 0 is obtained by variation of constants, i.e. if we replace I0 by the unknown function f (l). Inserting dI df = exp(l) f exp(l) dl dl into Eq. (6.8) gives df = S exp(l) dl or f (l) = S dl exp(l) = S exp(l) + c . The integration constant c has to be chosen such that I(0) = I0 , c = I0 S. Hence the solution is I(l) = I0 exp(l) + S[1 exp(l)] . The gure shows how the intensity I(l) of the photon beam reaches asymptotically the value S of the source function. If I0 < S, more photons are replaced by the gas than absorbed and I(l) approaches S from below for . The typical distance for a change of I is .
2

1.5 I(l)/I0

0.5

0 0 0.5 1 1.5 2 tau 2.5 3 3.5 4

6.3 Diusion and random walks

What is the fate of a photon created in the center of the Sun? If we estimate its mean free-path = 1/(n) by assuming that most of hydrogen in the Sun is ionized, and thus the mean density of free electron is ne = /mH 8.4 1023 /cm3 , and use as cross section 1025 cm2 , then we obtain as estimate 10 cm. In each scattering process, the photon is approximately isotropically emitted: Thus the path of a photon from the center of the Sun to the surface is not a straight-forward travel, but resembles a random-walk. The time a photon needs to travel the distance R can be estimated in two ways: After n steps li of the same size |li | = l a photon that started at zero is at the position d = n li . The scalar product of d with itself is i=1
n n

dd=

i=1 j=1

li l j

(6.9)

and splitting the sum into the diagonal and the o-diagonal terms, we obtain
n n

d2 = nl2 + 2l2
i=1 j<i

cos ij nl2 .

(6.10)

6 Stellar atmospheres and radiation transport By assumption, the angles ij between li and lj are chosen randomly and thus the o-diagonal terms cancel against each other. A more formal argument goes as follows: The probability p(m, n) to end after n steps of unit size at the distance d = m is given in the Table as m n=0 n=1 n=2 n=3 -3 -2 -1 1/2 1/4 1/8 3/8 1/2 3/8 0 1 1 1/2 1/4 1/8 2 3

The denominator of p(m, n) is the sum of all possible paths and thus equal to 2n . The numerator is the sum of all paths ending at distance m after n steps. Each entry in the table is given by the sum of the entries to the left and right in the line above and thus equal to the binomial coecient (remember Pascals triangle). Thus p(m, n) = 2n n . m (6.11)

Since p(m, n) = p(m, n), the average distance is zero. A more useful quantity is the root mean square deviation d = k 2 1/2 , d=
k

k 2 p(k, n) 2n

1/2

= n1/2

(6.12)

If the step size is l, then we obtain again d = n1/2 l.

6.4 Photo-, chromosphere and corona

Photosphere: The photosphere gives rise to the continuous radiation and the major part of the Fraunhofer lines. Main process for continuum absorption is the bound-free transition H + H+e , for emission the reverse process. Since the electron is free, the energy of the photon is not xed. The surface of the Sun is not smooth but shows granules of 100km size: Lighter areas are surrounded by darker (i.e. cooler) ones, providing evidence for convection. Center limb darkening: An area element of a surface has the same brightness for any viewing angle (Lamberts cosine law): Although the emitted power from the area element into the direction is reduced by cos , the observed size of the area element is also reduced by that same amount. Consequently, the brightness of the solar disk should be constant, if the Sun is an ideal black-body. The observed darkening of the Sun towards the limb (cf. right panel of Fig. 6.1) can be explained by the nite extension of the photosphere and a decrease of the solar temperature towards its edge. We see by denition to = 1. At the center, a certain value r = r( = 1) is dened by this condition, while this r corresponds to 1/ cos at the limb. Thus = 1 corresponds to layers lying more outside towards the limb that are in turn cooler. Since I depends exponentially on , this explains also the sharpness of the solar atmosphere.

6.4 Photo-, chromosphere and corona

Figure 6.1: Left: Solar corona in X-rays. darkening clearly visible.

Right: Photosphere with sunspots and limb-

Chromosphere: At most wavelengths the chromosphere is optically thin, and we see through it to the photosphere. Only using special lters or during a solar eclipse, the chromosphere becomes visible. Corona: A very hot, low-density plasma that is far from thermal equilibrium. Its energy density is dominated by magnetic elds. The heating mechanism of the corona is not yet understood.

7 Main sequence stars and their structure

7.1 Equations of stellar structure
We look for spherical symmetric, static solutions of the equations of stellar structure. This requires that rotation, convection, magnetic elds B, and other eects that break rotational symmetry have only a minor inuence on the star. This assumption is in most cases very well justied.

7.1.1 Mass continuity and hydrostatic equilibrium

We denote by M (r) the mass enclosed inside a sphere with radius r and density (r),
r

M (r) = 4
0

dr r2 (r )

(7.1)

or dM (r) = 4r2 (r) . dr (7.2)

Although trivial, this equation constitutes the rst (the continuity equation) of the ve equations needed to describe the structure and evolution of stars. An important application of the continuity equation is to express physical quantities not as function of the radius r but of the enclosed mass M (r). This facilitates the computation of the stellar properties as function of time, because the mass of a star remains nearly constant during its evolution, while the stellar radius can change considerably. A radial-symmetric mass distribution M (r) produces according Gau law the same gravitational acceleration, as if it would be concentrated at the center r = 0. Therefore the gravitational acceleration produced by M (r) is g(r) = GM (r) . r2 (7.3)

If the star is in equilibrium, this acceleration is balanced by a pressure gradient from the center of the star to its surface. Since pressure is dened as force per area, P = F/A, a pressure change along the distance dr corresponds to an increment dF = dA P (P + dP )dA = dA dP = (r)dAdr
force mass

a(r)
acceleration

(7.4)

of the force F produced by the pressure gradient dP . For increasing r, the gradient dP < 0 and the resulting force dF is positive and therefore directed outward. Hydrostatic equilibrium, g(r) = a(r), requires then GM (r)(r) dP = (r)g(r) = . dr r2 (7.5)

7.1 Equations of stellar structure If the pressure gradient and gravity do not balance each other, the layer at position r is accelerated, 1 dP GM (r) + . (7.6) a(r) = 2 r (r) dr In general, we need an equation of state, P = P (, T, Yi ), that connects the pressure P with the density , the (not yet) known temperature T and the chemical composition Yi of the star. For an estimate of the central pressure Pc = P (0) of a star in hydrostatic equilibrium, we integrate (7.5) and obtain with P (R) 0,
R

Pc =
0

dP dr = G dr

dM
0

M , 4r4

(7.7)

where we used the continuity equation to substitute dr = dM/(4r2 ) by dM . If we replace furthermore r by the stellar radius R r, we obtain a lower limit for the central pressure,
M

Pc = G
0

M >G 4r4

dM
0

M M2 = . 4R4 8R4

(7.8)

Inserting values for the Sun, it follows Pc > M2 = 4 108 bar 8R4 M M
2

R R

(7.9)

The value obtained integrating the hydrostatic equation using the Solar Standard Model is Pc = 2.48 1011 bar, i.e. a factor 500 larger.

7.1.2 Gas and radiation pressure

A (relativistic or non-relativistic) particle in a box of volume L3 collides per time interval t = 2L/vx once with the yz-side of the box, if the x component of its velocity is vx . Thereby it exerts the force Fx = px /t = px vx /L. The pressure produced by N particles is then P = F/A = N px vx /(LA) = npx vx or for an isotropic distribution with v 2 = 2 2 2 2 vx + vy + vz = 3 v x 1 P = nvp . (7.10) 3 If the particles have a distribution np of momenta with

N =V
0

dp np = V
0

dv nv ,

(7.11)

then we obtain the so-called pressure integral P = 1 3

dp np vp .
0

(7.12)

Although the derivation assumed classical trajectories of the particles, the result holds for any kind of non-interacting particles, in particular also if quantum eects are important (cf. Exercise 7.1). The two most important cases in astrophysics are a classical, non-relativistic gas of atoms and a gas of photons. In the rst case, we can derive the momentum distribution np noting

7 Main sequence stars and their structure that the states describing a free particle are labelled by the continuous three-momentum p. Thus the sum over discrete quantum numbers in the Boltzmann factor is replaced by an integration over the momenta d3 p/(2)3 and the volume d3 x occupied by the system, exp(E/kT ) mv 2 d3 p exp (2)3 2kT = 1 (2)3
0

exp

mv 2 2kT

4m3 v 2 dv .

(7.13)

If we compare the RHS with Eq. (7.11) we see that we need only to normalize correctly nv . The integral can be evaluated by substituting = m/(2kT ) and noting that

dx exp(x2 )

dxx2 exp(x2 ) =

1 = 2

(7.14)

Multiplying the integrand with 4(/)3/2 , we obtain the Maxwell-Boltzmann distribution of velocities for a classical gas, nv dv = n m 2kT
2/3

exp

mv 2 2kT

4v 2 dv .

(7.15)

Because of np dp = nv dv, we can insert now nv into the pressure integral (7.12), P = 1 3

dv nv vp = n
0

2/3 0

dx x4 exp x2 = nkT .

(7.16)

The integral dx x4 exp(x2 ) has been calculated with the same trick, but now dierentiating twice the Gaussian integral with respect to . Since we use generally the mass density instead of the particle number density n, it is more convenient to introduce the gas constant R = k/mH and the mean atomic weight dened by n = /(mH ). Then the ideal gas law becomes P = nkT = RT / . (7.17) A fully ionized plasma consisting mainly of hydrogen has 1/2. The second important example is the pressure Prad of radiation, i.e. the pressure of a photon gas. With p = h/c and n d = np dp it follows Prad = 1 3

d n h .
0

(7.18)

Remembering that u d = 4/cB d, it follows Prad = aT 4 /3 , where we introduced the radiation constant a = 4/c. (7.19)

7.1.3 Virial theorem

The virial theorem is an important link between the (gravitational) potential energy and the internal (kinetic) energy of any system in equilibrium. In order to derive it for the special case of a star, we multiply both sides of the hydrostatic equilibrium Eq. (7.5) with 4r3 and integrate over r, R R GM (r) dr 4r3 dr 4r3 P = . (7.20) r2 0 0

7.1 Equations of stellar structure Next we insert dM (r) = 4r2 dr on the RHS and integrate partially the LHS with V (0) = 0 and P (R) = 0, M R GM (r) . (7.21) dM dr 4r2 P = 3 r 0 0 The RHS is the gravitational potential energy Upot of the star. We can rewrite the LHS as
M

dM
0

P = Upot .

(7.22)

2 For the special case of an ideal gas, P = nkT = 3 Ukin /V , and we obtain the virial theorem,

2Ukin = Upot .

(7.23)

Hence the average energy E = U/N of a single atom or molecule of the gas is Ekin = 1 2 Epot . This is the same result as for a free hydrogen atom, indicating that only the shape not the strength of the potential V (r) r determines the ratio of kinetic and potential energy. Ex.: Estimate the central temperature of the Sun with the virial theorem.
We approximate the Sun as a homogeneous sphere. Then the gravitational potential energy of a proton at the center of the Sun is GM mp 3.2keV/c2 . Epot = R For a thermal velocity distribution of a Maxwell-Boltzmann gas we obtain Ekin = 3 1 kT = Epot 1.6 keV/c2 . 2 2

Hence our estimate for the central temperature of the Sun is Tc 1.1 keV/c2 1.2 107 K compared to Tc = 1.57 107 K in the Solar Standard Model.

7.1.4 Stability of stars

We want to generalize the virial theorem to a gas with an equation of state such that the pressure P can be expressed as function only of the density . Such an equation of state is called polytropic, and normally written as P = K with polytropic index . To do so, we have to express the energy density as function of the pressure and the index . Combining dP/P = d/ and d/ = dV /V , we obtain V dP = P dV . Next we add pdV to both sides, d(V P ) = V dP + P dV = ( 1)P dV . or d VP 1 = P dV = dU . (7.25) (7.24)

(7.26)

7 Main sequence stars and their structure Hence the pressure and the (kinetic) energy density of a polytrope are connected by P = ( 1) U . V (7.27)

For an ideal gas, U/V = 3/2kT and P = nkT , and the adiabatic index follows as = 5/3, while for radiation, U/V = aT 4 and P = aT 4 /3, and thus = 4/3. The relation P = ( 1)U/V allows us to re-express the LHS of Eq. (7.21) as 3( 1)Ukin = Upot . A star can be only stable, if its total energy Utot = Ukin + Upot is smaller than zero, Utot = (4 3)Ukin = 3 4 Upot < 0 . 3 3 (7.29) (7.28)

For = 5/3, we obtain back our old result for an ideal gas. A star with = 4/3 has zero energy and marks the border of matter that can become gravitationally bound. Adding an arbitrary small amount of energy would disrupt such a star, while subtraction would lead to its collapse. Important examples for matter with = 4/3 are apart from radiation relativistic fermions, e.g. relativistic electrons, protons and neutrons. Stars or more generally all gravitationally bound systems, have surprising thermodynamical properties: Consider e.g. the heat capacity, CV = Ukin /V Ukin /R. Since the temperature T increases for decreasing radius R, the heat capacity of a star is negative.

7.1.5 Energy transport

There exist three dierent mechanism for the transport of energy: i) radiative energy transfer, i.e. energy transport by photons, ii) conduction, e.g. the transport of energy by valence electrons or phonons in solids, and iii) macroscopic matter ows. Conduction plays a prominent role as energy transport only for dense systems, and is therefore only relevant in the dense, nal stages of stellar evolution. Radiative energy transport For the energy ux F emitted per area and time by an layer of the star at radius r and temperature T , a transport equation similar to Eq. (6.8) for the intensity I holds, dF = nFdr = Fdr . (7.30)

Here we introduced also the opacity being the cross section per mass of a certain material. Absorption of radiation implies also a transfer of momentum to the medium. Since the momentum of photons is p = E/c and F = E/(At), a slab of matter absorbs the momentum F/c per area and time. According to Newtons second law, F = p, the absorbed momentum has to be equal to the net force applied to the layer. This force is simply the dierence in radiation pressure dPrad times the area. Thus 1 dp dF = = dPrad c dA dt (7.31)

7.1 Equations of stellar structure

Figure 7.1: In the shadowed regions convection is important for Main Sequence stars, the x axis labels the total mass of the stars, x = log(M/M ), while the y axis labels the position in the star, y = M (r)/M . and F = 1 dF c dPrad = . dr dr (7.32)

The pressure of radiation is Prad = aT 4 /3 and hence F = 4acT 3 dT c d(aT 4 ) = . 3 dr 3 dr (7.33)

The luminosity of a shell at radius r and temperature T is thus connected with a temperature gradient dT /dr as L(r) = 4r2 F(r) = 16r2 acT 3 dT . 3 dr (7.34)

Convection Convection is a cyclic mass motion carrying energy outwards, if the temperature gradient in a star becomes too large a phenomenon familiar to everybody from water close to the boiling point. In the shadowed regions of Fig. 7.1 convection is important for main-sequence stars of various masses. In the case of the Sun, convection takes places in its outer layer, M (r) > 0.98M .

7 Main sequence stars and their structure

7.1.6 Thermal equilibrium and energy conservation

Thermal equilibrium and energy conservation require that the energy density produced per time and mass by all possible processes corresponds to an increase of the luminosity L, dL = 4r2 . dr (7.35)

The energy production rate per time and mass unit, = dE/(dt), consists of three terms, = grav + nuc , (7.36)

where nuc accounts for the energy production by nuclear processes and for the energy carried away by neutrinos. Both eects will be discussed in the next chapter in more detail. The term grav is the only one that can be both positive (contraction) or negative (expansion of the star) and is therefore crucial for the stability of a star.

7.2 Eddington luminosity and convective instability

Our rst aim in this section is to derive an upper limit on the luminosity of any stable object. We divide rst Eqs. (7.32) and (7.5) and obtain dPrad L(r) = . dP 4cGM (r) (7.37)

Since the total pressure P is the sum of radiation and gas pressure, P = Prad + Pgas , and both decrease outwards, it follows that dPrad and dPgas have the same sign. Hence, dPrad < dP and L(r) < 4cGM (r) . (7.38)

This bound relies on Eq. (7.32), i.e. it assumes that the energy is transported only by radiation, and is therefore not valid in zones of the star with convection. However, for r = R, energy transport should be purely radiative if the star is stable. As noted by Eddington, Eq. (7.38) represents therefore a critical luminosity, L < LEdd = M es 4cGM = 3.2 104 L . M (7.39)

If this luminosity is exceeded, the star ejects its outer layers in a stellar wind. To obtain the numerical estimate, we inserted the opacity es for Thompson scattering on free electrons1 in the last step. Next we consider what happens if for r < R the condition (7.38) is violated. We rewrite rst the energy transport equation as equation for dPrad /dr, dPrad L(r) = . dr c 4r2 Then we solve the hydrostatic equilibrium equation, dPrad GM (r) = , dr r2
1

(7.40)

(7.41)

Since the Thompson cross section th 1/m2 , photons interact much more often with electron than with protons.

7.3 Eddington or standard model with P = Pgas + Prad for dPgas /dr, dPgas dP dPrad GM (r) L(r) GM (r) = = + = 2 2 dr dr dr r c 4r r2 1 L(r) 4cGM (r) . (7.42)

Thus for L > LEd , the expression in the parenthesis changes sign and the gas pressure increases outwards: The stratication of the gas layers becomes unstable and convection starts. However, the criterion L(r) > Ledd (r) is only a necessary condition for convection.

7.3 Eddington or standard model

An important test for any model for stellar structure is to derive the mass-luminosity relation it predicts and to compare it to observations, Eqs. (5.36).

7.3.1 Heuristic derivation of L M 3

Converting the hydrostatic equilibrium equation Eq. (7.5) into a dierence equation, dP P and dr r, gives P M /R . (7.43) For an ideal gas P T , hence P M /R T or T M/R with M/R3 . The radiation transport equation gives in the same way L R4 T 4 RT 4 . M (7.44)

Inserting T M/R shows that the luminosity is only determined by the stellar mass and the opacity, M3 L . The opacity in turn is controlled by the chemical composition of the star.

7.3.2 Analytical derivation of L M 3

Let us dene and insert it into Eq. (7.37), dPrad L = . dP 4cG M (7.46) L(r) L = M M (r) (7.45)

At the surface, = 1 by denition. In general, increases for small r, while decreases (L(r) const. and M (r) 0 for r 0). To proceed, Eddington made the assumption that the product is approximately independent from the radius r, s = const. Then we can integrate Eq. (7.46) immediately, Prad = s L P. 4cGM (7.47)

7 Main sequence stars and their structure Dening as the fraction the gas contributes to the total pressure, Prad = (1 )P and Pgas = P , we have Pgas Prad P = = . (7.48) 1 R aT 4 = T . 3(1 ) Now we can express the temperature as function of the density, T = 3R(1 ) a
1/3

Assuming an ideal gas law, Pgas = RT /, and inserting Prad = aT 4 /3, we obtain

(7.49)

1/3 .

(7.50)

The total equation of state, P = Pgas / = (R/)T , is therefore P = 3R4 (1 ) a 4 4

K 1/3

4/3 ,

(7.51)

where the factor K is a constant. Recall that an equation of state with P = K is called polytropic with index = 1 + 1/n. For such an equation of state, the continuity and the hydrostatic equation decouple and can be solved separately from the energy transport and the energy conservation equation. We multiply the hydrostatic equation by r2 / and dierentiate it then with respect to r, d dr r2 dP dr = G dM (r) = 4G(r) , dr (7.52)

where we inserted the continuity equation in the last step. This second order dierential equation was studied already a century ago. The result for the total mass M as function of the radius R is GM n1 R 3n [(n + 1)K]n . (7.53) = Mn Rn 4G For the case of interest, n = 3, the mass is independent of the radius and is determined only by K, K 3/2 M = 4M3 , (7.54) G where M3 2.02. Inserting K gives M (1 )1/2 . 2 2 (7.55)

Squaring and inserting the numerical values of the constants, we obtain Eddingtions quartic equation, M 2 4 4 1 = 0.003 . (7.56) M Its solution is shown in Fig. 7.2. What can we learn about the structure and evolution of stars from this result?

7.3 Eddington or standard model

0.8

0.6

0.4

0.2

.1e2

.1e3 x

.1e4

.1e5

Figure 7.2: The solution of Eddingtons quartic equation as function of x = 2 M/M . Remember the meaning of = Pgas /P . Thus 0 corresponds to a free gas of photons, 1 to a primordial, cold cloud of gas. Only in the small range of 2 M where has an intermediate value stars can exist. Inserting L= into Eq. (7.56), we obtain 4cGM 4 4 L = L s L M M
3

4cGM (1 ) s

(7.57)

(7.58)

i.e. again a power-law with exponent = 3 that agrees reasonably well with the observations of main-sequence stars with intermediate mass. The variation in can explain the observed scatter in the L-M plot. For stars of given composition (i.e. xed ), increases as M increases. Thus, radiation pressure is more important in case of massive stars. Nuclear reaction cause a gradual increase of and therefore an decrease of . Hence, radiation pressure becomes more important when stars become older. In their late stages, stars may eject part of their envelope, an eect called stellar winds.

7.3.3 Lifetime on the Main-Sequence

If L/L (M/M ) with 3.5, then the life-time of stars on the main-sequence relative to the one of the Sun is t M L M +1 = = . (7.59) t M L M Hence a 10M star has approximately a 300 times shorter lifetime than the Sun.

7 Main sequence stars and their structure

7.4 Stability of stars

The longevity of stars on the main-sequence, with only small changes in radius and luminosity, means that a star has a heat-thermostat built in. This self-regulation of the nuclear burning is provided by the virial theorem together with the fact that the nuclear reaction rates nuc are increasing with temperature. In order to understand this mechanism, let us consider a small perturbation of a star, compressing it. Half of the potential energy released by the contraction is radiated away, while the other half leads because of Ekin = 1/2Epot 1/R to its heating. Since nuclear reaction rates increase strongly with temperature, energy production by nuclear reaction increases and therefore also the pressure. Thus the star expands until it reaches again an equilibrium conguration.

7.5 Variable stars period-luminosity relation of Cepheids

Cepheids are named after Cephei, a star with a period of 5.4 days and m = 0.7 mag. Generally, the periods of Cepheids vary between P 1100 days. H. Leavitt studied 19081912 Cepheids in the Small and Large Magellanic Clouds, i.e. at a xed distance. She discovered a relation between the period and the luminosity of Cepheids, M = 2.81 log(P/d) 1.43 . (7.60)

Cepheids are our rst example of a standard candle. As soon as this relation is calibrated (e.g. by measuring the distance to the LMC or SMC by other means), it can be used to obtain distance to any other galaxy with Cepheides. The spectral lines of Cepheids are Doppler shifted with same period as the light-curve. This suggest radial oscillations of the star as a reason for the periodic luminosity change. Let us consider a small radial perturbation of a stellar layer close to the surface. If the radius of this layer is decreased by a small amount, its density and thus the pressure increase. The increased pressure drives the layer back, it overshots its equilibrium position, and as a result the star starts to oscillate. Next we consider the eect of a change in the opacity: The opacity of the outer layer of a normal star decreases for increasing temperatures. Thus if the radius of a layer is again decreased by a small perturbation, the density and thus the pressure and also the temperature increase. The resulting decrease of the opacity eases the radiation transport and counteracts the increase in temperature and pressure. Hence the oscillations are damped. The envelopes of Cepheid stars are characterized by a narrow range of temperatures and densities where the opacity increases with temperature. Then the temperature dependence of the opacity does not quench the oscillations, and the amplitude of the oscillations can be stable and large.

7.6 Exercises
1. Show that the relation (7.12) for the pressure of non-interacting particles holds also for a quantum gas. [Hint: Derive rst P = (U/V )S and consider then how the energy of a free particle quantized in a box changes, if you change the volume of the box.]

8 Nuclear processes in stars

8.1 Possible energy sources of stars
The origin of the radiation energy emitted by the Sun was questioned already 1846, soon after the establishment of the law of energy conservation, by J.R. Meyer. It remained mysterious for ninety years. Since the temperature on the Earth was approximately constant during the last 4 109 years, the solar luminosity should be also roughly constant. Thus we can bound the minimal energy output of the Sun as L > 6 1050 erg.

8.1.1 Gravitational energy

The mass dM (r) contained in the shell between r and r + dr feels only the gravitational attraction of the enclosed mass M (r). Finding the total gravitational energy of a spherical star amounts therefore to integrate dUpot = GM (r)dM (r)/r,
R

Upot (R) = G

M (r)dM (r) . r

(8.1)

We assume that the star is homogeneous and use M (r) = (4/3)r3 as well as dM (r) = 4r2 dr. Then Upot (R) = G (4)2 2 3
R 0

dr r4 =

3G 5

4 3

2 R 5 =

3 GM 2 . 5 R

(8.2)

M 2 /R6

According to the virial theorem, 2Ukin = Upot , the total energy U gained by the Sun contracting from innity is U = Upot /2 = 3GM 2 /(10R) while the other half is radiated away. Mayer proposed meteorites falling into the Sun, Helmholtz and Kelvin the contraction of 2 the Sun itself as energy source. In the latter case U (3GM )/(10R ) 1 1048 erg, and this energy would be consumed within the time U/L 2 1014 s = 1 107 yr compared to the age of the Earth of about 4 Gyr.

8.1.2 Chemical reactions

The typical energy scale of chemical reactions that corresponds to a fraction of the binding energy of molecules, is of the order of 1 eV. (Remember the binding energy of hydrogen is 13.6 eV.) The maximal number of molecules in the Sun is M /(2mp ) 2 1030 kg/4 1027 kg 0.5 1057 , or the total energy 0.5 1048 GeV 8 1044 erg and thus even less that in gravitational energy.

8 Nuclear processes in stars n 1.0090 p 1.0081 2p+2n 4.0342

4 He

m/u

4.0039

Table 8.1: The masses of nucleons and 4 He in atomic mass units u.

Figure 8.1: The binding energy per nucleon Eb /A as function of the nucleon number A.

8.1.3 Nuclear fusion

The total mass m(Z, N ) of a nucleus with Z protons and N neutrons is because of its binding energy Eb somewhat reduced compared to its constituent mass, Eb /c2 = ZmH + N mn m(Z, N ) . (8.3)

Measured values of the binding energy per nucleon Eb /A as function of nucleon number A are shown in Fig. 8.1. From Fig. 8.1, one recognizes that the binding energy per nucleon Eb /A has its maximum at A 56, i.e. iron 56 Fe is the most stable element. Thus there are two possible ways to release energy by nuclear reactions: Two light nuclei can be fused or a heavy one can be broken up. While Eb /A is a rather smooth function of A for A > 20, there several peaks visible for small A: 4 He, 12 C, 14 N, 16 O and 20 Ne are energetically much more favourable than their neighbouring elements. The bound states of nucleons in nuclei have a similar shell structure as electrons in atoms. Nuclei with lled shells are especially stable, as noble gases are especially stable atoms.

8.2 Excursion: Fundamental interactions interaction force coupling range particle mass decay process decay time gravitation
2 GN = ( c)/MPl graviton 0

electromagnetic e2 /r2 em = e2 /( c) 1/137 photon 0 2 1016 s

weak GF 2 GF = 2g 2 /(8MW ) W = h/(MW c) W and Z-boson 80 and 91 GeV e+ + e + 6 s 10 E 2 er/W /r2

strong s er /r2 2 s = gs /( c) 1 = h/(M c) pions 140 MeV 2 1022 s

GN m2 /r2 p

Table 8.2: Some properties of the four fundamental forces. In Tab. 8.1, the masses of 4 He and its constituent nucleons are compared. These numbers imply that the fusion of four protons to 4 He releases 26.2 MeV energy, a factor 107 more than in our estimate for chemical reactions. Converting a solar mass into helium releases M /(4mp ) 26.2 MeV 1.25 1052 erg and thus around 5% of the Sun have been already converted into helium. As shown by the failure of using fusion for the energy production on Earth and the longevity of stars fusion is a non-trivial process: All four interactions are involved in the energy release by nuclear fusion. The strong interaction leads to the binding of nucleons in nuclei, the Coulomb repulsion has to be overcome to combine them, and the weak interactions convert half of the protons into neutrons. Finally, gravitation is responsible for the connement of matter, heating it up the proto-star to the necessary start temperature and serving then as a heat regulator.

8.2 Excursion: Fundamental interactions

Bosons and forces We know four fundamental forces in nature, that are all transmitted by the exchange of bosons, i.e. particles with integer spin. Two of them are the gravitational and electromagnetic force we have already encountered. These two forces fall-o like 1/r2 and are important on macroscopic scales. By contrast, the weak and the strong force are exponentially suppressed beyond a distance which corresponds to the Compton wavelength = h/(mc) of the particles that mediate these forces at energies relevant in astrophysics. A comparison of some properties of the four fundamental forces is given in Table 8.2. The rst line shows the force law characterising the dierent interactions. The strength of an interactions depends both on how strong the charges or couplings are and how large the Compton wavelength of the exchange particles is. In a simple analogy, one can imagine the cross section for a specic reaction as a grey disc: Its area is given by 2 and its greyness depends on the magnitude of the coupling constant. Fermions and the family structure Fermions carry half-integer spin and obey the Pauli principle, a principle that is crucial for the stability of matter. Protons and neutrons dier from electrons and neutrinos in that they have also strong interactions. The discovery of the muon happened accidentally in 1936-1937. The realization that it is an exact copy of an electron having only a dierent mass came as a real surprise. Nowadays we know that there exist three dierent families or avors of fermions, but still not

8 Nuclear processes in stars why three. For instance, the electron has additionally to the muon with mass m = 207 me an even heavier copy, the tau with m = 3480 me . In the same way, three dierent neutrinos exist. They are called e , and , depending on if they are generated together with an electron, muon or tau. charged leptons neutrinos e e

8.3 Thermonuclear reactions and the Gamov peak

Coulomb barrier-classically Let us consider the forces between two protons: For r = h/(m c), the Coulomb force dominates over the strong force. The size of a nucleon, rN 1013 cm, is comparable to the Compton wave-length of the pion and we will not distinguish between them in the following. Thus nuclei should have classically the energy V Z1 Z2 e2 /rN to cross the Coulomb barrier and to reach another nucleus of size rN . For a thermal plasma of particles with reduced mass and charges Zi e, this condition reads 1 3 Z1 Z2 e2 v 2 = kT = . 2 2 r (8.4)

Specically, we obtain for protons with Z1 = Z2 = 1 and size rN 1013 cm that the temperature should be above 2e2 10 (8.5) T > 3krN 10 K .

On the other hand, we have estimated the central temperature of the Sun as Tc 107 K. Hence we should expect that only the tiny fraction of protons with v 2 > 1000 v 2 is able to cross the Coulomb barrier. Coulomb barrier-tunneling Quantum mechanically, tunneling though the Coulomb barrier is possible. The wave-function of a particle with E V < 0 is non-zero, but exponentially suppressed. In order to avoid a too strong suppression, we require that = h/p rN , Z1 Z2 e2 h2 = . 22 (8.6)

Inserting = h2 /(2Z1 Z2 e2 ) for rN in Eq. (8.5) gives

2 2 4Z1 Z2 e4 107 K . (8.7) 3h2 k Hence we expect that quantum eects lead to a not too strong suppression of fusion rates. Next we want to make this statement more precise: We can estimate the probability p(x) that a nuclei is inside the classically forbidden region as

T >

p(x)dx exp(Ax/)dx exp(Axmv/h)dx .

(8.8)

If the two particles come closer than rN , the nuclear forces dominate and they fuse. Therefore the probability for a fusion process to happen is proportional to p(rN ) with rN = 2Z1 Z2 e2 /(v 2 ) from Eq. (8.4), or p(rN )dx exp(bZ1 Z2 e2 /E 1/2 )dx . (8.9)

8.4 Main nuclear burning reactions Reaction rates We introduced the interaction depth = nl as the probability that a particle interacts once travelling the distance l through targets with density n. The rate of such interactions, i.e. the number of reactions per time follows then with l = vt as nv, if the particle moves with velocity v. Since the energies of the particles are not uniform, but distributed according a non-relativistic Maxwell-Boltzmann distribution, nv dv = m 2kT
3/2

exp

mv 2 2kT

4v 2 dv eE/kT E 1/2 dE ,

(8.10)

we should average over the distribution nv . Then the rate follows as dE (E)E eb/E
1/2

eE/kT .

(8.11)

Cross section for strong interactions are of geometrical nature, (E) 2 and with = h/p it follows (E) 1/E. Therefore it is convenient to introduce the so-called S-factor S(E) = E(E) of the reaction, dE S(E)eb/E
1/2

eE/kT .

(8.12)

If the cross-section behaves indeed as (E) 1/E, then S(E) is a slowly varying function. The reminder of the integrand is sharply peaked (Gamov peak) in the region around 10 keV: For higher energies, the integrand is suppressed because of the Boltzmann factor exp(E/kT ), while for lower energies the tunneling probability exp(b/E 1/2 ) goes to zero.

8.4 Main nuclear burning reactions

8.4.1 Hydrogen burning: pp-chains and CNO-cycle
The pp-chains are shown in detail in in the left panel of Fig. 8.2. Its main chain uses three steps: Step 1: p + p d + e+ + e Step 2: p + d 3 He + Step 2: 3 He +3 He 4 He + p + p. The CNO-cyle is shown in detail in the right panel of Fig. 8.2. The small inlet compares the temperature dependence of the pp-chain and the CNO-cycle: For solar temperatures, the contribution of the pp-chains to the solar energy production is four order of magnitudes more important than the CNO-cycle. Radius-mass relation of main-sequence stars Hydrogen is burned at nearly xed temperature T . Via the virial theorem, also the gravitational potential is nearly the same for all main-sequence stars and thus GM/R const. As a result, the radius of main-sequence stars increases approximate linearly with the stellar mass.

8.4.2 Later phases

The increasing Coulomb barrier for heavier nuclei means that the fusion of heavier nuclei requires higher and higher temperatures. Therefore the dierent fusion phases hydrogen, helium, carbon,. . . burning never coexist, but follow each-other. Since the temperature decreases outwards, the fraction of the core participating in fusion reactions becomes smaller in each new burning phase, cf. Fig. 8.3.

8 Nuclear processes in stars

p+p

H+e +

p+p+e

H+

99.75%
2

0.25% H+p
3

He + 14%

86%
3

He + He

He + 2p 99.89%
7

He + He

Be + 0.11%
7

Be + e

Li +
4

Be + p
8 *

B+
+

Li + p

2 He

Be + e +

ppI

ppII

ppIII

Figure 8.2: Left: proton-proton chains Right: CNO cycle.

Figure 8.3: Burning phases of a 15 M star.

8.5 Standard solar model and helioseismology Hydrogen burning 4p+2e 4 He+2e proceeds by pp chains and CNO cycle. no heavier elements are formed, because no stable isotopes with mass number A = 8 exist. neutrinos are produced by weak reaction in p n conversions. the typical temperature is 107 K ( 1 keV). Helium burning 4 He+4 He+4 He 8 Be+4 He 12 C triple alpha reaction builds up Be with a concentration 109 which acts a catalyzer for the production of 12 C: 12 C+4 He16 O 16 O+4 He20 Ne. the typical temperature is 108 K ( 10 keV). Carbon burning many reactions like 12 C+12 C20 Ne+4 He etc. the typical temperature is 109 K ( 100 keV).

8.5 Standard solar model and helioseismology

A standard solar model uses the i) conservation laws together with ii) an equation of state, iii) energy production by fusion processes and iv) energy transport equation to evolve the Sun in small time-step from some initial conditions to its present age of 4.5 109 yr. Agreement with present-day luminosity and radius is enforced by tuning unknown pre-solar helium abundance and a technical parameter determining convection. The structural changes in the Sun, as it evolves, are caused by the nuclear reactions occurring in the central regions of the Sun. The transmutation of four hydrogen atoms into one helium reduces the number density of particles in the central regions which decreases the pressure. The pressure decrease does not actually occur because the surrounding layers respond to the force imbalance by contracting in the central regions. Half of the gravitational energy released from the contraction goes to raising the temperature of the central regions (the other half, according to the virial theorem, is radiated away). The increased temperature, by the ideal gas law, increases the pressure of this region and restores the balance between the pressure and gravitational forces. The larger mean molecular weight increases the luminosity of the star, and the rate of nuclear reactions. Also, while the central layers contract, the outer regions expand, in a sense, compensating for the steepening temperature gradients in the central regions. Therefore, as the Sun evolves from the zero age both its luminosity and radius increase. The solar standard model can be tested by helioseismology: Inside the Sun, pressure waves with periods between 2 min and 1 h have been observed by Doppler shifts of spectral lines on the solar surface. (The amplitude is a few meters: The solar surface is rising and falling a few meters every few minutes.) The oscillation pattern depends on the sound speed in the solar interior and allows therefore to test predictions of the solar standard model in the interior.

8.6 Solar neutrinos

Solar neutrinos ux From L = 4 1033 erg/s= 2 1039 MeV/s and the energy release of 26.2 MeV per reaction, the minimal number of neutrinos produced is N = 2 1038 /s. As we

8 Nuclear processes in stars

Figure 8.4: Schematic picture of the Sun and dierent oscillation modes (left) together with the measured oscillation frequencies (right).

0.035 density differences / = ( O model ) / O 0.03 0.025 0.02 0.015 0.01 0.005 0 -0.005 -0.01 0 0.1 0.2 0.3 0.4 0.5 0.6 radius r (RO) 0.7 0.8 0.9 1 density differences / = ( O model ) / O Fine-zoned standard solar reference model same, but coarse-zoned same, but very coarse-zoned OPALeos-midT OPALeos-lowT vary LO

0.045 0.04 0.035 0.03 0.025 0.02 0.015 0.01 0.005 0 -0.005 -0.01 -0.015 0 0.1

Standard RO (R98), tO = 4.6 Gyr: reference model low RO (R78) very low RO (R508) OPALeos-midT, standard R98 OPALeos-midT, R78 OPALeos-midT, R508 very low tO = 4.5 Gyr

0.2

0.3

0.4 0.5 0.6 radius r (RO)

0.7

0.8

0.9

Figure 8.5: Relative density dierence between the solar standard model or its variations and the deduced values from helioseismology as function of the solar radius.

8.6 Solar neutrinos have seen, photon perform a random-walk. Neutrinos have much smaller interactions, = 1043 cm2 E MeV (8.13)

and thus they can escape from the Sun: The interaction depth for a neutrino in the Sun is approximately = nR = 109 . (8.14) At the distance of the Earth, this corresponds to a neutrino ux of = N 1 = 7 1010 2 . 4D2 cm s (8.15)

(Alternatively, one can derive from the known value of the Solar constant S via = 2S/(26.2 MeV).) Weak interactions in the Sun produce always electron neutrinos, i.e. p n + e+ + e , but not p n + + + or p n + + + , because the energy released in nuclear reaction and the temperature is too small too produce a or . Similarly, only e neutrinos are detected in radio-chemical reactions via inverse beta-decay, while all type of neutrinos can be detected in elastic scattering on electrons. Solar neutrino experiments Radio-chemical experiments detect neutrinos by inverse betadecay in suitable nuclei. The historically rst isotope used was chlorine, e +37 Cl 37 Ar+e , i.e. changing a neutron inside a 37 Cl nuclei into a proton, thereby converting it into a 37 Ar. The disadvantage of this reaction was its high energy threshold, E 0.814MeV, that made this experiment only sensitive to 9% of all solar neutrinos, cf. the energy spectrum of solar neutrinos in Fig. 8.6. The experiment consisted of 615 tons of C2 Cl4 solutions in a mine 1500 m underground. After exposure of around 2-3 months, a few Ar atoms were produced. They were chemically extracted and counted by their subsequent decays (halftime 35 days). Starting from rst data in 1968, a decit appeared relative to theoretical expected uxes: only 30% of predicted event number is measured. This decit was dubbed solar neutrino problem. Finding the solution to this problem required more than 30 years of intensive experimental and theoretical work. Starting from 1991, 2 Gallium experiments e +71 Ga 71 Ge + e with threshold E 233keV took data. They found 55% of the expected neutrino ux, corresponding to about 9 atoms of 71 Ge in 30 tons of solution containing 12 tons 71 Ga, after three weeks of run time. What are plausible solutions to the solar neutrino problem? The experimental results might be wrong: For instance, the extraction eciency of the few Argon or Gallium atoms might have been not 100%. The nuclear cross sections used as input for the calculation of the solar standard model were measured at higher energies and had to be extrapolation to the energy range of the Gamov peak. This extrapolation might be wrong, if an unknown resonance lies close to the Gamov peak. The solar standard model itself and in particular the central temperature Tc might have a larger error than estimated. Finally, the particle physics properties of neutrinos might be dierent than expected and e may not survive the travel to the Earth.

8 Nuclear processes in stars As explained in the next subsection, the last possibility is the correct explanation: A large fraction of electron neutrinos oscillate into muon and tau neutrinos that do not participate in inverse-beta reactions.

8.6.1 Solar neutrino problem and neutrino oscillations

We can consider the time-evolution of a neutrino analogous to the quantum mechanical problem with n states. If we denote the eigenstates of H by |n , H|n = n|n , then their timeevolution is simply |n(t ) = |n(t) eiEn (t t) . (8.16) An arbitrary (normalized) state |(t) is connected by an unitary transformation with the eigenstates |n , Un |n(t) eiEn (t t) . (8.17) |(t ) =
n

Thus the probability to measure the eigen-value n will depend on time, if the state is not an eigenstate of H. For a two-level system and the unitarity condition U U = U U 1 = 1, it follows U= cos sin sin cos . (8.18)

By denition we call the avour of a neutrino according to the produced charged leptons in weak interaction, with = e, , . These states are not necessary eigenstates of the free Hamiltonian H describing the propagation of particles. Analogous to Eq. (8.17), the time-evolution of a state describing an electron-neutrino, e (t), follows as |e (t) = Uen |n eiEn t = cos |1 eiE1 t + sin |2 eiE2 t (8.19)
n

Thus the probability to nd a neutrino in a certain avour will depend in general ( = 0) on time. The survival probability Pe e of an e is Pe e (t) = | |e (t)|e (0) |2 = 1 sin2 [(E2 E1 )t/2] sin2 (2) . (8.20)

For ultra-relativistic neutrinos, Ei = (m2 + p2 )1/2 p2 + m2 /(2p), and replacing the travel i i time t by the distance L, Pe e (L) = 1 sin2 m2 L sin2 (2) . 4E (8.21)

The two-neutrino survival probability depends on the vacuum oscillation length losc = 4 E 2.5 m m2 E MeV eV2 m2

mass squared dierence, no sensitivity to the absolute mass 68

8.6 Solar neutrinos

(10 cm -2 s -1)

8 7 6 5 4 3 2 1 0 0

SNO

SSM

5 6 6 e (10 cm -2 s -1 )

Figure 8.6: Left: The solar standard model predicts the neutrino ux and thus also the number of events that should be measured. Right: Results of the SNO experiment. the mixing angle . The measurements shown in the right panel of Fig. 8.6 demonstrated that neutrino avor conversions take place: The red band shows the ux of e , measured by inverse-beta decay reactions. This ux is just 36% of the expected ux from the solar standard model. However, the experiment was also able to measure the ux of and , cf. the green and the blue band. Summing all three favors up, the three bands cross and one obtains the value predicted by the solar standard model. Hence neutrinos oscillate, changing their avor on the way from the solar core to the Earth.

9 End points of stellar evolution

9.1 Observations of Sirius B
Sirius is with a distance of 2.6 pc the fth closest stellar system to the Sun. Analyzing the motions of Sirius 18331844, F.W. Bessel concluded that it had an unseen companion, with a orbital period P 50 yr. In 1862, A. Clark discovered this companion, Sirius B, at apastron or the time of maximal separation of the two components of the binary system. Followingup observations showed that the mass of Sirius B equals approximately the one of the Sun, M M . Sirius Bs peculiar properties were not established until the next apastron, 1915, by W.S. Adams. He noted the high temperature of Sirius B, which requires together with its small luminosity an extremely small radius and thus a large density of this star. Ex.: Find the mean density of Sirius B from its apparent magnitude, m = 8.5, and its surface temperature T = 25, 000 K. We convert rst the apparent into an absolute magnitude,
M = m 5 log(d/10pc = 11.4 , and express then the absolute magnitude as a luminosity, L = 3.02 1028 W 100.4M = 3.84 1026 W . Now we can use the Stefan-Boltzmann law and obtain R = R L L
1/2

T T

102 .

Therefore the mean density of Sirius B is a factor 106 higher than the one of the Sun. More precisely one nds = 3 106 g/cm3 .

We now apply the lower limit for the central pressure of a star in hydrostatic equilibrium, Eq. (7.9), to Sirius B, Pc > M2 = 4 108 bar 8R4 M M
2

R R

= 4 1016 bar .

(9.1)

What would be the central temperature Tc needed, if the pressure is dominated by an ideal gas? From the ideal gas law, we nd Tc = Pc 102 Tc, 109 K . nk (9.2)

For such a high central temperature, the temperature gradient dT /dr in Sirius B would be a factor 104 larger than in the Sun. This would in turn require a larger luminosity L(r) and a larger energy production rate than in a main sequence star. The solution to this puzzle is that the main source of pressure in such compact stars has a dierent origin.

9.2 Pressure of a degenerate fermion gas

Figure 9.1: Left: Hubble Space Telescope image showing Sirius A, the brightest star in our nighttime sky, along with its faint, tiny stellar companion, Sirius B. Right: Evolution of the Sun.

9.2 Pressure of a degenerate fermion gas

For a classical gas, P = nkT , and thus in the limit of zero temperature, also the pressure inside a star goes to zero. How can a star be stabilized after the fusion processes and thus energy production stopped? The Pauli principle forbids that fermions can occupy the same quantum state. In statistical mechanics, Heisenbergs uncertainty principle xp together with Paulis principle imply that each phase-space volume (1/ )dxdp can be occupied by only one fermionic state. If we use x = n1/3 and p /x n1/3 together with v = p/m valid for a nonrelativistic particles, we obtain for the pressure of a degenerate fermion gas P nvp
2 n5/3

P 5/3 .

(9.3)

For relativistic particles, we can obtain an estimate for the pressure inserting v = c, P ncp c 2 n4/3 m or P 4/3 . (9.4)

Note the following important points: 1. Both the non-relativistic and the relativistic pressure laws are polytropic equations of state, P = K . 2. A non-relativistic degenerate Fermi gas has the same adiabatic index, = 5/3, as an ideal gas, while a relativistic degenerate Fermi gas has the same adiabatic index, = 4/3, as radiation. Since = 4/3 marks the border between stable and unstable systems (cf. Eq. (7.29)), we can expect that there exists a critical mass for cold matter: Increasing the mass of a star beyond the level that its constituents become relativistic should unbind its constituents.

9 End points of stellar evolution 3. The pressure is inversely proportional to the fermion mass, P 1/m. Thus the degeneracy pressure will become important rst for electrons. Let us compute the pressure of a degenerate non-relativistic electron gas and check if it is consistent for Sirius B with the lower limit for the central pressure. First, we verify that electrons are non-relativistic: With ne /(2mp ) and 106 g/cm3 , where we assume that Np Nn , it follows ne 1 106 g/cm3 3 1029 cm3 2 1031 cm3 = (me / c)3 . 2 1.67 1024 g
5/3

(9.5)

Thus we can use the equation for the pressure of a non-relativistic degenerate electron gas, P
2 n5/3 e

(1.05 1027 erg s)2 9.11 1028 g

106 g/cm3 2 1.67 1024 g

1023 dyn/cm2 .

(9.6)

This pressure corresponds with 106 dyn/cm2 = 1bar to P = 1017 bar, and is consistent with our lower limit for central pressure of Sirius B.

9.3 White dwarfs and Chandrasekhar limit

Stars like Sirius B that are supported by the pressure of a degenerated electron gas are called white dwarf stars. They have very long cooling times because of their small surface luminosity. This type of stars is rather numerous: The mass density of man-sequence stars in the solar neighborhood is 0.04M /pc3 compared to 0.015M /pc3 in white dwarfs. The typical mass of white dwarfs lies in the range 0.4 1M , peaking at 0.6M . 2 If we combine Pc GM and P = Kn5/3 (M/R3 )5/3 = M 5/3 /R5 , we obtain R4 GM 2 KM 5/3 = R4 R5 or R= M
1012 6

1 . KM 1/3

(9.7)

Thus there exists a unique relation between mass and radius, if we neglect dierences due to small variations in their chemical composition. The radius of white dwarf stars decreases for increasing masses, suggesting also that there exists a maximal mass. In order to derive this maximal mass, we consider now the energy of a star where the pressure is dominated by a non-relativistic degenerate Fermi gas as function of its radius. The total kinetic energy is Ukin = N p2 /(2me ), where n N/R3 and p n1/3 . Thus Ukin N
2 n2/3 2

2 N 5/3 N (3+2)/3 = . 2mR2 2mR2

(9.8)

For the potential gravitational energy we use again the approximation Upot = GM 2 /R with = 1. Hence 2 N 5/3 GM 2 U (R) = Ukin + Upot . (9.9) 2mR2 R For small R, the positive term dominates and therefore a stable minimum Rmin exists for each M .

9.4 Supernovae If the Fermi gas in the star becomes however relativistic, then Ukin = N cp, or Ukin N c n1/3 and U (R) = Ukin + Upot c N 4/3 R (9.10)

c N 4/3 GM 2 . (9.11) R R Now both terms scale like 1/R. For a xed chemical composition, N/M = const., the negative term increases faster than the rst one, if M is increased. Hence, it exists a critical M so that U becomes negative and can be made arbitrary small by decreasing the radius of the star: The star collapses. This critical mass is called Chandrasekhar mass MCh and is obtained by solving Eq. (9.11) 4/3 2 for U = 0 and using M = NN mN . Then c Nmax = GNmax m2 , or N Nmax and The critical size can be determined from the condition the gas becomes relativistic, Ukin < N mc2 together with N = Nmax . Thus Nmax mc2 > c mc2 < R R> We have two options: We can 1. set m = me and obtain R > 5 108 cm. This corresponds to the radii found for white dwarf stars. 2. set m = mN and obtain R > 3 105 cm. Since already Sirius B was dicult to detect, the question arises if and how these extremely small stars can be observed. Moreover, which factors decide if a star ends as a white dwarf or as a neutron star? Note that MCh is for both cases the same, since the stellar mass is in both cases given by the sum of the nucleon masses only the main source of pressure (electron or neutrons) diers. mc c Nmax R
4/3

c Gm2 p

3/2

MPl mp

2 1057

(9.12)

MCh = Nmax mp 1.5M .

(9.13)

(9.14)

and nally

c Gm2 N c Gm2 N

1/2

(9.15)
1/2

(9.16)

9.4 Supernovae
Novae and supernovae were characterized empirically according to their luminosity and their spectral lines. Novae show a smaller luminosity increase than supernovae with peak luminosities between 10 and 106 times their average luminosity. They are recursive events with periods in the range P 1 h105 yr. By contrast, supernova are singular events. Supernovae are further divided into type II and type I, depending on the presence of Balmer lines in their spectrum, respectively.

9 End points of stellar evolution

Figure 9.2: Lightcurves of type Ia supernovae; top observed ones, bottom after rescaling. Novae and type Ia supernovae The physical interpretation of novae and supernovae does not follow precisely this division: Novae and a subset of all type I supernovae called type Ia are variations of accreting binary systems. In the case of a nova, the accreted material starts the CNO cycle in a thin layer on the stellar surface when the pressure is large enough, expelling the outermost shell of the star, but not destroying the star. In a type Ia SN, a white dwarf is driven by accretion beyond MCh and explodes. More precisely, one nds the following picture from numerical simulations: For M 1.44M , carbon in a white dwarf core starts to fuse into Fe, Co, and Ni. The burning expands outwards from the core, synthesizing lighter elements such as Mg, Si, Su. If the burning front reaches around 0.7R, the pressure of the outer layers is not sucient to conne the burning process. The white dwarf star explodes. The main part of optical luminosity comes from the decay of nickel-56, with a half-life of 6.1 days to Cobalt-56 ( = 77.1d) to iron-56. The decays produce gamma-rays, heating the ejected material. Since the white dwarf stars explode crossing the Chandrasekhar limit, M > 1.4M , the re leased total energy should vary not to much. Thus one may wonder if they are possible standard candles? This requires that the spread in luminosity, L = dE/dt, is not big. As the upper panel of Fig. 9.2, shows more luminous supernovae evolve slower. If this eect is taken into account, as done in the lower panel, the spread in luminosity is small enough to use these events as standard candles. Core collapse or type-II SN Type II or core collapse supernovae occur at the end of the fusion process in very massive stars, M > (58)M . Theses stars have an onion-like structure with a degenerate iron core. When the core is completely fused to iron, no further processes

9.5 Pulsars releasing energy are possible. Instead, photo-disintegration destroys the heavy nuclei, e.g. via + 56 Fe 4 He + 4n, and removes the thermal energy necessary to provide pressure support and the star collapses. During the collapse of the core, the density increases and the free electrons are forced together with protons to form neutrons via inverse beta decay, e + p n + e . A protoneutron star is formed. When core density reaches nuclear density, the equation of state stiens suddenly and the infalling material is reected. If the supernova is successful, a neutron star is left over. Otherwise a black hole is formed. The released energy goes mainly into neutrinos (99%), kinetic energy (1%); only 0.01% into photons.

9.5 Pulsars
Generalities of compact stars White dwarf and neutron stars have in common that their radius is strongly increased with respect to a main sequence star, RWD /R 102 and RNS /R 105 . Using two conservation laws involving the stellar radius, we can derive immediately two important properties of these compact stars: Conservation of angular momentum, L = I = const. with I = M R2 , implies (initially) fast rotation of a compact star: A star like the sun, rotating once per month, P 106 s, would rotate with a ms period when contracted down to 10 km in size.

The magnetic ux, B = BA = const., is a conserved quantity for an ideal conductor, where magnetic eld lines are frozen in the plasma. As the core collapses, the magnetic eld lines are pulled more closely together, intensifying the magnetic eld by a factor (R /RNS )2 1010 . Magnetic A stars have surface elds up to 104 G, observed eld strengths of white dwarfs and of neutron stars are of the order of B (106 108 ) G and B (1012 1012 ) G, respectively. Discovery and identication of pulsars In 1967, Hewish and collaborators detected an object emitting a radio signal with period P = 1.377s. They called the object Pulsar. Only one year later, Gold argued that pulsars are rotating neutron stars. He predicted an increase of the period as pulsars, because they should loose energy via electromagnetic radiation. The slowdown of the Crab pulsar was indeed discovered in 1969. The key observations that helped to identify pulsars with rotating neutron stars were: 1. The smallness of the observed periods, P 1ms 5s. 2. The extreme stability of the periods, P/P = 1013 . 3. The periods increase slowly. Additional information from later observations include that the source distribution is peaked in the Galactic

9 End points of stellar evolution plane, implying a Galactic origin. Moreover, several pulsars like the Crab and Vela pulsars were found in remnants of type II supernovae. Since the light travel time ct is 500 km for 1.5 ms, we can use this distance as an upper limit for the emission region (that has to be causally connected to act as good clock). This implies that a pulsar has to be a compact object and we have to choose between white dwarfs, neutron stars or black holes as possible sites for the radio emission. Possible clock mechanisms are rotation or pulsation of a single star or the orbital movement of a binary system. We now consider the dierent options: Black holes are surrounded by an event horizon. Radiation is emitted by the accretion disk surrounding a black hole. However, both accretion and radiation is very irregular process, excluding the surrounding of a black holes as source of a very regular pulsed radio signal. The shortest period for a rotating system corresponds to rotation at the break-up velocity, R 2 = GM R2 (9.17)

(v = R) or in terms of the mean density M/R3 2 = G . (9.18) Using as extreme value for the density of a white dwarf = 108 g/cm3 yields P = 2/ > 1 s. This excludes rotating white dwarfs. What about a pulsating white dwarf? The estimate 2 = G holds also for the fundamental frequency of pulsations. Higher harmonics as explanation are excluded because a superposition of frequencies would destroy the observed periodicity. Moreover, energy losses lead generally to an increase of the period in case of oscillations. In case of a binary white dwarf systems we can replace the radius R of the star by the distance d R of the system. Thus the resulting periods are certainly even larger. Finally, we consider the last option left over, neutron stars. If they are pulsating, the fundamental frequency would be with NS 106 WD typically too short, P 103 ms. A binary system of neutron stars is a copious emitter of gravitational waves as we will discuss in the next chapter. For such small distances as required by the ms periods, the life-time of these systems would be too short. In summary, the only remains option is a rotating neutron star. Rotating dipole model A simple model for a pulsar is depicted in the gure above as a neutron star with a non-aligned rotation and magnetic eld axes. Physically, on can model a pulsar as a rotating sphere endowed with a magnetic dipole moment. The energy of a rotating sphere is E = 1/2I 2 and the energy change due to a change in the rotation velocity is E = I . (9.19)

The energy loss of a rotating magnetic dipole due to the emission of electromagnetic radiation is B 2 R6 4 sin2 E= , (9.20) 6c3 where denotes the angle between rotation and dipole axis. If we dene as time-scale characterizing the slow-down of a pulsar, = I 2 6Ic3 = = 2 6 2 2 , B R sin E (9.21)

9.5 Pulsars then we can compare the estimated life-time, e.g. of the Crab nebula, 1243 yr with its true age, 19721054 = 918 yr. Also, the derived energy output E 6 1038 erg is comparable to one derived from observations. Using pulsars as tool Dispersion measure and Galactic electron density: The velocity v of electromagnetic waves in a medium is dierent from c, n = v/c = 1. The refractive index n is a function of the wave-length and thus there is a dispersion between the arrival times of a pulse at dierent wave-lengths, t = L (n(1 ) n(2 )) . c (9.22)

The deviation of the refractive index n from one is proportional to the number density of electrons, n ne . (Since ne is not constant, one should integrate along the line of sight.) Thus the time-delay t measures the integrated number density of electron along the line of sight to a pulsar. This was used originally as a check that pulsars are indeed Galactic objects. If the distance to a specic pulsar is known, one can the method as a tool to measure the electron density ne in the Galaxy. Binary systems of pulsars have been used as test for the emission of gravitational waves as predicted by general relativity, as we will discuss in the next chapter. The Faraday eect is used as measure for the Galactic magnetic eld.

10 Black holes
10.1 Basic properties of gravitation
1. In classical mechanics, the equality of gravitating mass mg = F/g and inertial mass mi = F/a is a puzzle noticed already by Newton. Knowing more forces, this puzzle becomes even stronger. Contrast the acceleration in a gravitational eld to the one in a Coulomb eld: In the latter, two independent properties, namely the charge q giving the strength of the electric force and the mass mi , the inertia of the particle, are needed. The equivalence of gravitating and inertial mass has been tested starting from Bessel, comparing e.g. the period of a pendulum of dierent materials, P = 2 mi l . mg g (10.1)

While mi = mg can be achieved for one material by a convenient choice of units, there should be deviations for test bodies with other composition. Current limits are ai /a < 1012 . 2. Newtons law postulates as the Coulomb law instantaneous an interaction. This is in contradiction to special relativity. Thus, as interactions with electromagnetic elds replace the Coulomb law, a corresponding description should be found for gravity. Moreover, the equivalence of mass and energy found in special relativity requires that, in a loose sense, energy not only mass should couple to gravity: Imagine a particleantiparticle pair falling down a gravitational potential wall, gaining energy and nally annihilating into two photons moving the gravitational potential wall outwards. If the two photons would not loose energy climbing up the gravitational potential wall, a perpetum mobile could be constructed. If all forms of energy act as sources of gravity, then the gravitational eld itself is gravitating. Thus the theory is non-linear and its mathematical structure much more complicated as Maxwells equations. 3. Gravity can be switched-o locally, just by cutting the rope of an elevator. Inside a freely falling elevator, one does not feel the eect of gravity. Motivated by 2., Einstein used 1., the principle of equivalence, and 3. to derive general relativity, a theory that describes the eect of gravity as deformation of the space-time known from special relativity.

10.2 Schwarzschild metric

Because Einsteins theory has a rather more complicated mathematical structure than Newtons, no analytical solution to the two-body problem is known. Instead, we are looking for the eect of a nite point-mass on the surrounding space-time.

10.2 Schwarzschild metric

10.2.1 Heuristic derivation

Consider a freely falling elevator in the gravitational eld of a radial-symmetric mass distribution with total mass M . Since the elevator is freely falling, no eects of gravity are felt inside and the space-time coordinates from r = should be valid inside. Let us call these coordinates K with x (parallel to movement), y , z (transversal) and t . The elevator has the velocity v at the distance r from the mass M , measured in the coordinate system K = (r, , , t) in which the mass M is at r = 0 at rest. We assume that special relativity can be used for the transformation between K at rest and K moving with v = c, as long as the gravitational eld is weak. We shall see shortly what weak means in this context. For the moment, we presume that the eect of gravity are small, if the velocity of the elevator that was at rest at r = is still small, v c. Then dt = dt dx = dy 1 2 dr (10.2) (10.3) (10.4) (10.5)

1 2 = rd

dz = r sin d .

becomes

Thus the line-element of special relativity, i.e. the innitesimal distance between two spacetime events, 2 2 ds2 = dx2 + dy + dz c2 dt2 (10.6) ds2 =

dr2 + r2 (d2 + sin2 d2 ) (1 2 )dt2 . (10.7) 1 2 Next, we want to relate the factor 1 2 to the quantities M and r. Consider the energy of the elevator with rest mass m, GmM ( 1)mc2 = 0, (10.8) r where the rst term is the kinetic energy and the second the Newtonian expression for the potential energy. According to 2), we made here the crucial assumption that gravity couples not only to the mass of the elevator but to its total energy. Dividing by mc2 gives 1 1 GM = 0. rc2 (10.9)

Remembering the denition = 1/ 1 2 and introducing = GM/c2 , we have 1 2 = 1 (10.10) r or 2 2 2 1 2 = 1 + 2 1 . (10.11) r r r In the last step, we neglected the term (/r)2 , since we attempt only an approximation for large distances, where gravity is still weak. Inserting this expression into Eq. (10.7), we obtain the metric describing the gravitational eld produced by a radial symmetric mass distribution, 2 dr2 2 2 2 2 2 2 1 . 2 + r (d + sin d ) c dt r 1 r Surprisingly, this agrees with the exact result found by Karl Schwarzschild 1916. ds2 = (10.12)

10 Black holes

10.2.2 Interpretation and consequences

Gravitational redshift As in special relativity, the line-element ds determines the time and spatial distance between two space-time events. The time measured by an observer called the proper-time d is given by d = cds. In particular, the time dierence between two events at the same point is given by setting dxi = 0. If we choose two static observers at the position r and r in the Schwarzschild metric, then we nd with dr = d = d = 0, d (r) = d (r ) g00 (r) dt g00 (r ) dt = g00 (r) . g00 (r ) (10.13)

The time-intervals d (r ) and d (r) are dierent and thus the time measured by clocks at dierent distances r from the mass M will dier too. In particular, the time measured by an observer at innity will pass faster than the time experienced in a gravitational eld, = (r) 1 2/r < (r) . (10.14)

Since frequencies are inversely proportional to time, the frequency or energy of a photon traveling from r to r will be aected by the gravitational eld as (r ) = (r) 1 2/r . 1 2/r (10.15)

Hence an observer at r will receive photons with frequency 2GM (r) (10.16) rc2 if these photons were emitted by a source at position r with frequency (r): The frequency of a photon is redshifted by a gravitational eld, in agreement with our argument against a perpetuum mobile in 2). The size of this eect is of order /c2 , where = GM/r is the Newtonian gravitational potential. We are now in position to specify more precisely what weak gravitational elds means: As long as ||/c2 1, the deviation of gtt = 1 2GM 1 2(r)/c2 from the rc2 Minkowski value gtt = 1 is small and Newtonian gravity is a sucient approximation. = 1 Schwarzschild radius What is the meaning of r = RS 2? At RS = 2GM M = 3 km c2 M (10.17)

the coordinate system (10.12) becomes ill-dened. However, this does not mean necessarily that at r = RS physical quantities like tidal forces become innite. Instead r = RS is an event horizon: We cannot obtain any information about what is going on inside RS , if the gravitating mass is concentrated inside a radius smaller than Rs 1 . An object smaller than its Schwarzschild radius is called a black hole. The black hole is fully characterized by its mass M (and possibly its angular momentum L and electric charge q). To understand this better, we consider next what happens to a photon crossing the event horizon at r = RS as seen from an observer at r = .
1

Recall that in Newtonian gravity only the enclosed mass M (r) contributes to the gravitational potential outside r for a spherically symmetric system. Thus, e.g. the Sun is not a black hole, since for all r the enclosed mass is M (r) < rc2 /2G.

10.3 Gravitational radiation from pulsars Approaching a black hole Light rays are characterized by ds2 = 0. Choosing a light ray in radial direction with d = d = 0, the metric (10.12) becomes dr = dt 1 2 r c. (10.18)

Thus light travelling towards the star, as seen from the outside, will travel slower and slower as it comes closer to the Schwarzschild radius r = 2. In fact, for an observer at innity the signal will reach r = 2 only asymptotically for t . Similarly, the communication with a freely falling space ship becomes impossible as it reaches r = Rs . A more detailed analysis shows that indeed no signal can cross the surface at r = Rs . Perihelion precession Ellipses are solutions only for a potential V (r) 1/r. General relativity generate corrections to the Newtonian 1/r potential, and as a result the perihelion of ellipses describing the motion of e.g. planets process. This eect is largest for Mercury, where /t 43 /yr. This was the main known discrepancy of the planetary motions in the solar system with Newtonian gravity at the time, when Einstein and others worked on relativistic theories of gravity. Light bending The factors (1 2/r) in the line-element ds will lead to the bending of light in gravitational elds. The measurement of the deection of light by the Sun during the solar eclipse 1919 was the rst crucial test for general relativity. Nowadays, one distinguishes three dierent cases of gravitational lensing, depending on the strength of the lensing eect: 1. Strong lensing occurs when the lens is very massive and the source is close to it: In this case light can take dierent paths to the observer and more than one image of the source will appear, either as multiple images or deformed arcs of a source. In the extreme case that a point-like source, lens and observer are aligned the image forms an Einstein ring. 2. Weak Lensing: In many cases the lens is not strong enough to form multiple images or arcs. However, the source can still be distorted and its image may be both stretched (shear) and magnied (convergence). If all sources were well known in size and shape, one could just use the shear and convergence to deduce the properties of the lens. 3. Microlensing: One observes only the usual point-like image of the source. However, the additional light bent towards the observer leads to brightening of the source. Thus microlensing is only observable as a transient phenomenon, when the lens crosses approximately the axis observer-source.

10.3 Gravitational radiation from pulsars

An accelerated system of electric charges emits dipole radiation with luminosity Lem = 2 2 |d| , 3c3 (10.19)

where the dipole moment of a system of N charges at position xi is d = N qi xi . One i=1 might guess that for the emission of gravitational radiation the replacement qi Gmi works.

10 Black holes

Figure 10.1: Left: Strong gravitational lensing of. Right: Weak gravitational lensing of Abell2218. But since mi xi = ptot = const., momentum conservation means that there exists no gravitational dipole radiation. Thus one has to go to the next term in the multipole expansion, the quadrupole term, N 1 m(k) (xi xj ij r2 ) , (10.20) Qij = 3
k=1

and nds then for the luminosity emitted into gravitational waves Lgr = G 5c5 ... |Qij |2 . (10.21)

i,j

Derive Lem d2 /c3 and Lgr GQ2 /c5 using dimensional analysis. In the case of electromagnetic dipole radiation, we use that e2 = c and the dimension of the electric charge squared is [e2 ] = erg cm. Then we compare

Ex.:

[Lem ] =

cm2 erg s3 erg ! 4 = = erg cm . s cm3 s s

In the same way, we can show that the luminosity of gravitational quadrupole radiation is proportional to ca with a = 5: The unique solution for a from [Lgr ] = is a = 5. erg cma g2 cm4 g cm2 ! cm3 a = = 3 2 s s gs s s6

The emission of gravitational radiation is negligible for all systems where Newtonian gravity is a good approximation. One of the rare examples where general relativistic eects can become important are close binary systems of compact stars. The rst such example was found 1974 by Hulse and Taylor who discovered a pulsar in a binary system via the Dopplershift of its radio pulses. The extreme precision of the periodicity of the pulsar signal makes this binary system to an ideal laboratory to test various eect of special and general relativity:

10.4 Thermodynamics and evaporation of black holes

Figure 10.2: Measured change of the period of the orbital movement for the Hulse-Taylor binary compared to the prediction of general relativity. The pulsars orbital speed changes by a factor of four during its orbit and allows us to test the usual Doppler eect. At the same time, the gravitational eld alternately strengthens at periastron and weakens at apastron, leading to variable gravitational redshift of the pulse. The system emits gravitational waves and looses thereby energy. As a result the orbit of the binary shrinks by 4mm/yr, in excellent agreement with the prediction of general relativity, cf. Fig. 10.2. The small size of the orbit leads to a precession of the Perihelion by 4.2 /yr.

10.4 Thermodynamics and evaporation of black holes

Black holes thermodynamics Classically, the mass and therefore the radius of a black hole can only increase with time. The only other quantity in physics with the same property is the entropy, dS 0. This suggests a connection between a quantity characterizing the size of the black hole and its entropy. To derive this relation, we apply the rst law of thermodynamics dU = T dS to a black hole. Its internal energy U is given by U = M c2 and thus dU = dM c2 = T dS .
!

(10.22)

A Schwarschild black hole is fully characterised by only one quantity, e.g. Rs . Hence both the temperature T and the entropy S of a black hole should be determined solely by Rs . Our experience with the thermodynamics of non-gravitating systems suggests that the entropy is an extensive quantity and thus proportional to the volume, S V . In this case, 2 2 2 dV Rs dRs Rs dM and dM Rs dV T 2 dV . On the other hand, if we assume that the entropy is proportional to the area of the event horizon, dA Rs dRs Rs dM and 1 dM Rs dA T dV . Surprisingly, several arguments show independently that the second alternative is correct; one we will sketch below.

10 Black holes Thus dU = T dS = dM c2 =

c4 dA = T d 16GRs kc3 A. 4 G

kc3 A 4 G

(10.23)

and, as rst suggested by Bekenstein, the entropy of a black hole is given by S= (10.24)

If this analogue is more than a formal coincidence, a black hole should behave similar as other thermodynamical systems we know. In particular, if its surrounded by a cooler medium, it should emit radiation and heat up the environment. Hawking could indeed show 1974 that a black hole in vacuum emits black-body radiation (Hawking radiation) with temperature kT = c3 . 8GM (10.25)

Evaporation of Black Holes The radiation of a BH is a quantum process, as indicated by the presence of Plancks constant in Eq. (10.25). We can understand the basic mechanism and the magnitude of the black hole temperature by considering a quantum uctuation near its event horizon at RS : Heisenbergs uncertainty principle allows the creation of a virtual electron-positron pair with total energy E for the time t < /E. This pair can become real, if the gravitational force acting on them during the time t can supply the energy E. More precisely, the work is done by the tidal force dF = F GM E/c2 dr = dr , r 2r3 (10.26)

because the electron-positron pair is freely falling. After the time t, the pair is separated by the distance r = ct c /E. The energy gain by the tidal force is F r (E/c2 )GM/r3 (r)2 . (10.27) If we require that F r > E and identify the energy E of the emitted pair with the temperature T of the black hole, we nd kT E < (GM/r3 )1/2 . (10.28)

The emission probability is maximal close to the event horizon, r = RS , and thus we obtain indeed kT E c3 /(GM ). This result supports the idea that the entropy of a black hole is indeed non-extensive and proportional to the surface of its event horizon. Since entropy counts the possible micro-states of a system, any promising attempt to combine gravity and quantum mechanics should provide a microscopic picture for these results.

Exercises
2 1. Assume that the surface A = 4RS of a Black Hole is emitting black-body radiation with temperature kT = c/(4RS ), where RS is its Scharzschild radius, RS = 2GM/c2 . i) Derive the luminosity L of the Black Hole. ii) Derive with E = M c2 and dE/dt = L the lifetime of a Black Hole with initial mass M. iii) Find the temperature and the luminosity of a black hole with M = M .

10.4 Thermodynamics and evaporation of black holes 2. Estimate the total entropy of the observable universe of radius 4 Gpc, comparing the entropy of the cosmic microwave radiation (T = 2.7 K and S = (4 2 /45) T 3 V ) with the entropy of all black holes assuming that each galaxy occupying the volume 10 Mpc3 contains a supermassive black hole with mass M = 106 M . 3. The Global Positioning System (GPS) consists of 24 satellites carrying with them atomic clocks at an altitude of about 20,000 km from the ground. A GPS receiver determines its current position by comparing the time signals it receives from a number of the GPS satellites (usually 4 or more) and triangulating on the known positions of each satellite. The attempted precision is below 1 m. Estimate special and general relativistic eects that have to be accounted for to achieve this precision.

Part II

Galaxies

11 Interstellar medium and star formation

11.1 Interstellar dust
Dust becomes visible by its blocking eect of star light. The combined eect of scattering and absorption of light, mainly by dust grains, is called extinction. Clearly, extinction may aects the properties like the luminosity or the distance of a star or galaxy that we deduce from its observed spectra. Surprisingly, this possibility has been ignored until the 1930s by astronomers. R.J. Trmpler found the rst clear evidence for extinction by dust examining cluster of u stars in our galaxy. He plotted the angular diameter D of these star clusters versus their distance r. For the derivation of their distance he assumed F 1/r2 . The distribution n(D/r, r) should not depend on r, but he found a systematic increase of the linear size of clusters with distance. There are only two interpretations possible: i) The Sun is at a special place in the galaxy where the size of star clusters has a minimum. ii) Some light is absorbed and thus the energy ux decreases faster than 1/r2 . If the dust between us and the source has the optical depth , then the observed intensities with (I) and without (I0 ) absorption are connected by I = I0 exp( ). We can relate the optical depth to the extinction A measured in magnitudes via A m0 m = 2.5 log(I0 /I) or A = 2.5 log(exp( )) = 2.5 log(e) 1.09 . (11.2) Thus the extinction A equals approximately the optical depth. If the spectral type and distance of a star is known, then the extinction between us and the star can be determined: The spectral type determines the absolute magnitude M , while the apparent magnitude m is observed directly. Then m = M + 5 log(r/(10pc)) + A , (11.3) (11.1)

and we can nd A for known m, M and r. On the other hand, distance measurements using spectroscopic parallaxes or standard candles like Cepheides or Supernovae of type Ia are aected by extinction. Reddening and extinction curves The scattering cross section of light on dust is wavelength dependent. In general, light with shorter wavelengths is more eciently scattered. In classical electrodynamics, one can roughly distinguish three scattering regimes, R : R2 R: R: 1 4 geometrical, Mie, Rayleigh. (11.4) (11.5) (11.6)

11 Interstellar medium and star formation

Figure 11.1: A dark cloud, Barnard 86, is silhouetted against a starry background. Measuring extinction at dierent wave-lengths allows astronomers to determine the properties of interstellar dust. One nds roughly A 1/, thus the typical size of dust grains is comparable to the wave-length of visible light. If the wave-length dependence of extinction for typical dust is known, the absolute value of the extinction in the spectrum of a certain source can be estimated by observing it at dierent wave-lengths. Polarization Light passing dust clouds becomes polarized and the degree of its polarization is found to be proportional to the extinction. Using classical electrodynamics one can show that this requires a non-spherical shape of dust grains. Since the polarization builds up scattering on dierent dust grains, the dust grains have to be moreover partially aligned. The observed polarization is most likely explained, if dust grains have a magnetic moment and become aligned by magnetic elds. Composition and sources The absorption features visible in the IR of extinction curves correspond to vibrational transitions of silicates and ice, while extinction features in the UV can be connected to the presence of carbon. This leads to the idea that dust grains are mainly dirty ice balls containing some graphite. The composition suggests that dust originates from massive stars: In their red giant phase they loose part of their envelope; silicon and carbon condenses cooling down, and a layer of hydrogen is bound around the silicon and graphite core.

11.2 Interstellar gas

Interstellar gas was rst detected by its absorption of background light. Since gas clouds are much cooler than the surface of stars, their absorption lines are narrower than those of stars, have smaller excitation energies, and can thus be distinguished from stellar absorption lines. One expects that gas clouds consist mainly of hydrogen. Since they are cool, hydrogen is in the ground state. Therefore the main absorption feature is the Lyman- line that is in the UV and thus not observable with a ground telescope. Instead, early studies concentrated on absorption lines of simple molecules like CH and CN and of atoms like Ca and So.

11.3 Star formation Nowadays, one observes gas clouds mainly with radio observation of the 21cm line of hydrogen. This wavelength corresponds to the temperature T = ch/k 0.07 K and can thus be excited even in cool clouds. The 21cm line is caused by transitions between the two hyperne levels of the hydrogen 1s ground state. The ground state of hydrogen can be occupied by an electrons with spin up or down. These two stated have the same energy, apart from a small correction due to the interaction of the electron spin with the spin of the proton. The 21cm emission line corresponds to a ip of the electron spin from parallel to anti-parallel to the spin of the proton. The 21cm emission line is an extremely useful tool to study gas clouds in our and nearby galaxies. Since their movement can be measured using the Doppler eect, they serve as tracers for the velocity distribution of matter in galaxies. Radio observations are not hampered by dust, and thus one can observe gas clouds close to the galactic center and beyond, regions that are obscured to optical observations. The Zeeman eect allows one to measure magnetic elds in the clouds and thereby to trace the galactic magnetic eld. In summary, these studies conclude that gas clouds ll 5% of the galaxy; they have an average density nH (1 10) cm3 .

11.3 Star formation

11.3.1 Jeans length and mass
A cloud of gas will collapse, if the gravitational attraction dominates over thermal pressure. A bound system has negative energy E = Epot + Ekin 0, or 3 GM 2 3M kT 5 R 2m or M 5 kT . R 2 Gm (11.7)

Here we assumed a homogeneous distribution of gas atoms of mass m, obeying the ideal gas law, in a cloud of radius R and total mass M . The smallest mass M (for xed density and 3 temperature T ) which collapses is called Jeans mass MJ = 4/3RJ with Jeans length RJ . Inserting the expression for MJ , we nd 5 kT 4 2 RJ = 3 2 Gm or RJ = 15kT 8Gm
1/2

(11.8)

The acceleration a for a homogeneous cloud is a= GM (r) 4 = Gr . 2 r 3 (11.9)

Since the density increases as the cloud collapses, we can derive an upper limit on the time required for a complete collapse, the so called free-fall time , assuming the density is constant, 1 . (11.10) (G0 )1/2
2 Since the sound speed of an ideal gas is vs = 5kT /(3m), the Jeans length can be expressed also as RJ vs / G . (11.11)

11 Interstellar medium and star formation This suggests another intuitive interpretation of the Jeans criterion: It compare the timescale of gravitational collapse (G0 )1/2 with the one pressure can react, p R/vs . For p , or R vs /(G0 )1/2 , gravitational collapse occurs. The time-scale sets the time-scale of growth of density perturbations. However, without detailed analysis we do not know the functional dependence, i.e. if e.g. (t) exp(t/ ) or (t) (t/gr ) . The same idea applies to the formation of structures on all scales in the Universe. Since gravitation can enhance only already existing density uctuation, a mechanism in the early Universe should exist that produced initial density uctuations. Density dependence If the initial density of the cloud is 0 , then = 0 or dierentiating d = 30 r r0 r
3

r0 r

(11.12)

dr =

3 dr r

d 3dr = . r

(11.13)

Thus for smaller r, the density increases faster. The free-fall time will increase towards the center, the outer material will lag behind the material closer towards the center, enhancing the dierence between core and outer layers. Rotation The acceleration in radial direction is now the dierence between gravitational and centrifugal acceleration, GM (r) r 2 . (11.14) ar = r2 Let us estimate by how much the cloud collapses until the two forces on the RHS balance each other, a = 0. 2 2 r 0 4 r 0 v0 2 GM (r) = r3 2 = r3 0 = , (11.15) r r
2 2 r 3 v0 /r0

where we used angular momentum conservation L = I = const. with I = M r2 and thus /0 = (r0 /r)2 . We now introduce the speed v0 = 0 r and solve for the fraction r/r0 the cloud collapses, 2 r 0 v0 r = . (11.16) r0 GM (r) Ex.:
Consider the collapse of a cloud with nH = 105 cm3 , T = 50K and initial rotation velocity v0 = 1km/s. The Jeans radius is 0.2pc, the Jeans mass MJ = 76M , and thus r/r0 = 0.6. Thus the cloud should expand for the chosen parameters. Angular momentum conservation is an important obstacle to the formation of dense structures, leading to fragmentation in sub-clouds and to the formation of discs perpendicular to L.

11.3 Star formation

11.3.2 Protostars
Luminosity of collapsing clouds Let us apply once again the virial theorem, Epot = 2Ekin , for the evolution of a collapsing cloud. Its total energy is E = Ekin = Epot /2 and as it collapses |Epot | increases, but only half remains as kinetic energy in the cloud. The other half has to be radiated away. The total energy E = (3/10)GM 2 /R changes in time as dE 3 GM 2 dR = dt 10 R2 dt or (11.17)

1 dE 1 dR = . (11.18) E dt R dt Thus the relative decrease of the radius is directly connected to the relative energy release of the cloud. If energy cannot be emitted eciently, the collapse is slowed down. The formation of a proto-star can be divided into the following stages: Initially, the cloud consists of molecular and atomic hydrogen plus atomic helium. Half of the energy deliberated by the collapse is transferred into the internal energy of the proto-star. In the beginning, the energy is not used to heat the gas, but for its ionization. Only after that, the proto-star heats up and its pressure increases, slowing-down the collapse. Ex.:
For a 1M proto-star with an initial radius R = 500R , nd the free-fall time (G)1/2 , the energy released and the average luminosity in the last 100 years of collapse. We use our standard approximation of a spherically symmetric and homogeneous gas cloud, E= 3 GM 2 2 1045 erg . 10 R

The average luminosity is L = E/t, or L= 2 1045 erg 7 1035 erg/s 170L . 100 107 s

Once the luminosity reached typical stellar luminosities, the cloud is called proto-star. For increasing density, the interior becomes more and more opaque to radiation, and nally only the surface can radiate. The central temperature is rising until fusion processes start. At same point, the pressure gradient is large enough to stop the collapse and a stable main sequence star is born.

12 Cluster of stars
A look at the night-sky with binoculars or a small telescope shows that stars are not uniformly distributed on the sky. They are concentrated in groups of stars like the Plejades and a thin band across the sky. The thin band with a milky appearance is the disc of our own galaxy, the Milky Way, that we will discuss in the next chapter. In this chapter, we discuss cluster of stars and in particular their evolution.

12.1 Overview
Types and properties There are two main types of star clusters, galactic or open clusters and globular clusters. Figure 12.1 shows in the left panel the open cluster M45, also called Plejades, and in the right one the globular cluster M80. 1. Galactic or open clusters have typically < 103 stars and a diameter D < 10 pc. They are close to the galactic plane and contain often dust and interstellar gas. 2. Globular clusters consist typically of 104 106 stars within a diameter of D 20100pc. The globular clusters in our galaxy form a spherical distribution with center around 8 kpc away, cf. Fig. 12.2. Assuming that the center of this distribution coincides with the center of our galaxy allowed Harlow Shapely 1918 to determine for the rst time the distance of the Sun to the galactic center. Globular clusters contain, if at all, only small amounts of dust and interstellar gas. The fraction of heavier elements of stars in a globular cluster is larger than in stars of a galactic cluster.

Dierent star populations Because of the continuous fusion in stars of hydrogen to helium and heavier elements, collectively called metals in astronomy, the initial chemical composition of stars diers. This eect can be roughly accounted for by dividing stars into two dierent populations. Population I stars are relatively young stars, which are found mainly in the galactic disc. Population II stars on the other hand are older stars, which are found e.g. in the galactic halo and globular clusters. Typical values for their initial chemical composition are X 0.70 0.75, Y 0.24 0.28, Z 0.005 0.04, Z 10
5

Pop I

(12.1)

X 0.75 0.77,

Y 0.23 0.25,

Pop II , (12.2)

where X, Y and Z denote relative mass fractions of hydrogen, helium and heavier elements, respectively. Population III stars are a hypothetical population of extremely massive and hot stars with virtually no metal content which are believed to have been formed in the early universe as the zeroth generation of stars. They have not yet been observed directly, but their existence is suggested to account for the fact that heavy elements, which could not

12.1 Overview

Figure 12.1: Left: M45 or the Pleiades stars, an open cluster of 500 mostly faint stars spread over a 2 degree (four times the diameter of the Moon) eld. Right: The globular cluster M80 containing several 100,000s of stars.

Figure 12.2: Distribution of globular clusters in the Milky Way.

12 Cluster of stars

1 observer
~

3 convergent point

Figure 12.3: The angle between v and the line-of-sight approaches zero for t and the line-of-sight converges against the convergent point. have been created in the Big Bang, are observed in the emission spectra of very early, powerful galaxies. A natural explanation for these dierences is that star formation started in the whole galaxy. The primordial material from which the rst, i.e. Pop. II, stars were formed contained only a tiny contribution of metals. Outside of the galactic disc, most gas has been used up to form this rst generation of stars. Now, the density of interstellar gas is too small in the galactic halo and globular clusters, and star formation has stopped there. By contrast, the galactic disc still contains dense regions and star formation continues there. However, the interstellar material contains the ashes of old massive stars, and therefore the metallicity of this second generation of stars (= Pop I) is larger. Proper motion and distances Some nearby stars move with a suciently large velocity such that their motion over the sky becomes visible over the years. The velocity v can be split into its radial and transverse component,
2 2 v 2 = vr + vt = (v sin )2 + (v cos )2 ,

tan =

vt . vr

(12.3)

The radial part can be measured via Doppler shift, while the transverse part is responsible for the proper motion = vt /D of the cluster at the distance D across the celestial sphere. As long as we cannot determine , we cannot determine vt and D separately from a single measurement. A solution is to observe the star (or the cluster of stars) over a longer time period. If the cluster moves away, decreases and therefore the proper motion becomes slower: the star heads towards a denite point on the celestial sphere, called convergent point. From Fig. 12.3 we see that the angle between the line of sight and the line to the convergent point is . This method is an important way to establish distances to clusters containing Cepheides.

12.2 Evolution of a globular cluster

The crucial points in our derivation of the virial theorem 2 Ukin = Upot for a star was the assumption of a gravitationally bound system in equilibrium. Thus it holds also for any other system like a cluster of stars or galaxies, if this system fullls the two conditions to be i) gravitationally bound, and ii) in equilibrium. The rst condition can be checked directly by the virial theorem, while the second one needs a more detailed discussion. In particular, we will see that a gravitationally bound system evolves and thus can be only approximately

12.2 Evolution of a globular cluster in equilibrium. In this context, we dene a system as in equilibrium or dynamically relaxed, if the interchange of energy between the members of a cluster is fast compared to the evolution of the cluster. Rms and escape velocity from the virial theorem: The total kinetic energy of cluster with mass M = N m is 1 (12.4) Ukin = M v 2 , 2 where vrms v 2 1/2 is the root mean square (rms) velocity. Applying the virial theorem and using Upot = 3GM 2 /(5D), it follows v2 = 3GM . 5D (12.5)

Ex.: Find the typical rms velocity of stars in a spherical cluster with size D = 5 pc that consists of 106 stars with average mass m = 0.5M .
v2 = 3GM = 2.5 1012 cm2 /s2 5D (12.6)

or vrms 16km/s. This value can be compared with observations. If the observed velocities are signicantly higher, the cluster cannot be gravitationally bound or its total mass has to be larger.

Crossing time: The crossing time tcr is the typical time required for a star in the cluster to travel the characteristic size D of the cluster (typically taken to be the half-mass radius). Thus, tcr D/v or tcr 3 105 yr for the values of our example. Relaxation time: The relaxation time trel is the typical time in which the stars velocity changed an amount comparable to its original velocity by gravitational encounters. Thus one might think of it as the time-scale after which the velocity distribution of stars bound in a cluster has reached an equilibrium distribution by the exchange energy and momentum with each other. But since part of the stars will have velocities v > vesc and escape, the velocity distribution of a cluster is not stationary: High-velocity stars escape, the cluster contracts and its core is heated-up. The depth for collisions of stars with each-other is = 1 1 = , nl nvt (12.7)

when n denotes the density of stars and l = vt the path traveled by a singe star. Let us call the relaxation time trel the time for which = 1. Thus trel = 1/(nv). What should we use for R (as function of v) in = R2 ? Stars are certainly gravitationally interacting with each other, if they are bound to each other. Therefore we can estimate the eective interaction range R from Ekin = Epot , mv 2 /2 = Gm2 /R or R = 2Gm/v 2 . Then trel = v3 1 = . R2 nv 4n(Gm)2 (12.8)

12 Cluster of stars Inserting 1/n = (m/M )(4/3)D3 (with M = N m as total mass of the cluster) gives trel = v3 D3 . 3G2 mM (12.9)

If the cluster is dynamically relaxed and the virial theorem applies, then v 2 = 3GM/(5D) and thus D M v4 D2 ND trel = = N tcross . (12.10) v m 3G2 M 2 v Note that trel tcross , in striking contrast to an ordinary gas. We should take into account how much momentum per collision is exchanged: In a collision at small impact parameter b the momentum transfer is larger than in one at large b. Moreover, one cannot treat relaxation just as a two-body process: Because of the innite range of the gravitational force, N -body processes with N > 2 are not negligible. Formalizing this, the relaxation time given by Eq. (12.10) becomes reduced by a logarithmic term, 12 ln(N/2), or N trel 0.1tcross . (12.11) ln(N ) Thus, trel 2 109 yr for the values of our example. Evaporation time: The evaporation time for a cluster is the time required for the cluster to dissolve through the gradual loss of stars that gain sucient velocity through encounters to escape its gravitational potential. Assuming an isolated cluster with negligible stellar evolution, the evaporation time tev can be estimated by assuming that a constant fraction of the stars in the cluster is evaporated every relaxation time. Thus, the rate of loss is dN/dt = N/trel = N/tev . The value of can be determined by noting that the escape speed vesc at a point x is related to 2 the gravitational potential Epot (x) at that point by vesc = 2Epot (x). (The total energy of a particle able to escape has to be equal or larger than zero, i.e. Ekin + Epot 0 or 2 vesc 2GM/R.) If the system is virialized (as we would expect after one relaxation time), then also v 2 = 3GM/(5R). Thus, stars with speeds above twice the RMS speed will evaporate. Assuming a Maxwellian distribution of speeds, the fraction of stars with v > 2vrms is = 7.4 103 . Therefore, the evaporation time is trel 136trel . (12.12) tev = Stellar evolution and tidal interactions with the galaxy tend to shorten the evaporation time. Using a typical trel for a globular cluster, we see that tev 1010 yr, which is comparable to the observed age of globular clusters. Emergence of order vs. the 2.nd law of thermodynamics The second law of thermodynamics, dS 0, implies generally that the order in a closed system decreases: Heat ows from a hotter to a cooler subsystem, and particles diuse from denser to under-dense regions. The standard example for this behavior are two subsystems of an ideal gas with dierent temperature and density. The entropy of each system is Si = kNi 3 ln Ti ln i + C 2 . (12.13)

12.3 Virial mass An exchange of energy U1 = U2 or particles N1 = N2 between the two systems leads to a change in the total entropy, S = S1 + S2 . With U = (3/2)N kT and thus Si = Si Ui Si Ui + Ni = +k Ui Ni Ti 3 ln Ti ln i Ni , 2 (12.14)

the total change of the entropy Stot = S1 + S2 follows as Stot = 1 1 T1 T2 U1 + k 1 3 T1 ln ln 2 T2 2 N1 . (12.15)

Therefore the second law of thermodynamics, dS 0, implies the decrease of temperature and density inhomogeneities. The crucial dierence gravity introduces is a connection between the temperature, density and size of the system. From the virial theorem we obtain N T Ukin Upot N2 M2 1/3 R V (12.16)

or V N 3 /T 3 . Thus we can eliminate and obtain S = kN 3 ln T + 2 ln N + C 2 . (12.17)

Note the changed signs of the ln T and ln N terms, indicating that the inhomogeneities in a system dominated by gravity increase.

12.3 Virial mass

For a dynamically relaxed cluster, we can use M= 5R v 2 3G (12.18)

to estimate the cluster mass M . Here, v 2 as before refers to the thermal motions of stars. Thus the center-of-mass velocity of the cluster has to be subtracted. Using Doppler 2 2 shift measurement, the radial component vr can be measured. Since v 2 = 3 vi for an 2 = 3 v 2 . Thus arbitrary cartesian component of v, we have also v r M=
2 5R vr . G

(12.19)

The main advantage of this method to determine masses is that from a representative sample of vr measurements the total cluster mass including invisible stu can be determined. Its main disadvantage is the inherent assumption of a relaxed, bound system that has to be checked, e.g. by comparing the cluster age with trel ; also there is a weak dependence of the viral mass on the exact density prole (r). This method can be applied to other bound systems as cluster of galaxies as well.

12 Cluster of stars

Figure 12.4: Hertzsprung-Russell diagram for several open clusters (left) and the globular cluster M55 (right).

12.4 Hertzsprung-Russell diagrams for clusters

Stars in a cluster have not only a common distance but were most likely also formed at the same time and from material with the same composition. Studying the Hertzsprung-Russell diagram of stars of the same cluster is therefore possible without knowing the distance to the individual stars. Moreover, a comparison of Hertzsprung-Russell diagrams of dierent clusters informs us about the evolution of stars. Main sequence tting A Hertzsprung-Russell diagram for stars of the same cluster can be made using the apparent magnitude m of the stars. In general, there is a shift up or down necessary so that the mean-sequence lies correctly in the absolute magnitude scale M . The shift gives directly the distance modulus m M , or the cluster distance d/pc = 10(mM +5)/5 . Turn-o point Hertzsprung-Russell diagrams of several clusters are shown in Fig. 12.4, left panel. For all clusters, a turn-o point is visible above which no mean-sequence stars are found. Thus the more luminous stars that are hotter and have a shorter life-time are missing in the clusters: They evolved already into white dwarfs or neutron stars. Since stars in a cluster were formed at the same time, the turn-o point xes the cluster age. Table B.7 gives the life-time of stars on the mean-sequence. The age of the oldest globular cluster, t 12 Gyr, gives a lower limit to the age of the Universe.

13 Galaxies
13.1 Milky Way
In visible light we see an accumulation of stars as a band across the sky. In the infrared or far-infrared the structure of the Milky Way consisting of a disk and a bulge starts to be revealed. The Milkyway is an example for a spiral galaxy; it can be subdivided into (cf. also Fig. 13.2): The galactic bulge with a radius of 3 kpc around the center of the galaxy. It is composed both of old (pop II) and young (pop I) stars. The galactic disk with a radius of 20 kpc and a thickness 300 pc contains also both pop I and pop II stars together with gas and dust. The disk appears not uniform, but has probably four spiral arms and a central bar. The reason why the arms of spiral galaxies are so prominent is that the brightest stars are found in the spiral arms. Spiral arms are the major regions of star formation in spiral galaxies and this is where most of the major nebulae are found. The galactic halo with a radius of 25 kpc includes globular clusters (consisting only of old stars (population II)) A dark halo with an extension of order 100 kpc contains 80%-95% of the mass of the galaxy (dark matter). The number of stars in the Milky Way is 1011 , while its total mass is estimated to be 1012 M .

13.1.1 Rotation curve of the Milkyway

In the disk of the Milky Way, stars and other matter is rotating around the center in a regular pattern, as revealed by Doppler eects. In the galactic halo and the galactic bulge, the motion is largely random. Determination of the rotation curve For simplicity, we assume a spherical mass distribution and use again the enclosed mass M (r). For circular orbits, GM (r)m/r2 = mv 2 (r)/r, or M (r) = rv 2 (r) . G (13.1)

Measurements of v(r) are mainly done using the 21cm line of gas clouds as tracer. For orbits inside the Suns orbit, the radial velocity is largest, when the distance along the line-of-sight to the Galactic center is minimal: Hence the maximal Doppler shift along one line-of-sight determines v(r) at this minimal distance rmin . If the visible stars dominate the mass distribution of the Milky Way, then we expect for small r inside the bulge, M (r) r3 and thus v(r) r. Outside the disc, for r > 20 kpc, M (r) = const. and gas clouds should follow Keplerian orbits with v(r) 1/r1/2 .

13 Galaxies

lG 180
10

kpc 0 lG 270

lG 90

-5

-10

lG 0
-15 -10 -5 0 kpc 5 10 15

Figure 13.1: Left: Schematic view of the spiral structure of the Milky Way; the position of the Sun is marked by a red dot. Right: Rotation curve of the Milky Way.

globular clusters

disc h = 300 pc Sun

8 kpc

gas/CR halo
Figure 13.2: Schematic picture of the Milky Way with a gas and dust disc of height h = 300pc.

100

13.1 Milky Way

Figure 13.3: Winding-up of the spiral arms due to dierential rotation. Figure 13.1 compares the observed rotation curve with the one obtained theoretically from the observed distribution M (r) of visible matter as stars and gas. The discrepancy indicates that a large fraction, around 90%, of the mass in the Milky Way is non-visible. What (r) corresponds to v(r) = const.? We have v 2 (r) = GM (r)/r = const. r2 or 1/r2 . Thus each radial shell of thickness dr contains the same amount of dark matter, dM (r)r2 dr const. (This makes the denition of the size or edge of the DM halo somewhat arbitrary.) We shall discuss what could be the explanation for this unseen matter later, in Sec. 13.3.2. Rotation and the spiral structure The rotation of stars around the center of the Milky Way is dierential: The observed constant rotation velocity, v(r) const., means that the rotation period increases linearly with the distance to the Galactic center, P r. Thus the spiral arms should wind up as shown schematically in Fig. 13.3. For instance, the rotation period of the Sun around the Galactic center is 2 108 yr, so the Sun completed 20 turns around the Milky Way. How does the observed spiral pattern survives? The wind-up picture assumes that the spiral arms are a denser region of stars. However, we noted already that spiral arms are instead the major regions of star formation in spiral galaxies. Star formation is triggered by gravitational perturbations, in particular during encounters with neighbor galaxies, which compress interstellar matter. The velocity with which these density perturbations travel has a priori nothing to do with the rotation velocity of individual stars.

13.1.2 Black hole at the Galactic center

Supermassive black holes (SMBH) are supposed to be in the center of each galaxy, and that includes our own galaxy, the Milky Way. One way to show the existence of a SMBH is to deduce rst the enclosed mass from rotation curves around the supposed BH. Then one has to show that no object with such a mass has a sucient long lifetime. Observationally, it is crucial to observe stars as close as possible to the BH. The large amount of dust in the Galactic bulge requires excellent IR observations. In spring 2002, a star was passing with v = 5000 km/s at 17 light hours distance or 3AU the point of closest approach to the black hole. One could determine a unique Keplerian orbit (highly elliptical (eccentricity 0.87), semimajor axis of 5.5 light days, a period of 15.2 years and an inclination of 46 degrees). Keplers 3rd law gives as enclosed mass M (r) = (3.7 1.5) 106 M , cf.

101

13 Galaxies

Figure 13.4: Enclosed mass as function of the distance to Sgr A (an object at the galactic center); the dashed-dotted line gives for comparison the enclosed mass for a DM halo with (r) r and = 1.8. Fig. 13.4. It is not possible to explain this result with a dense cluster of dark astrophysical objects because such a cluster would have the extremely short lifetime of at most a few hundred thousand years.

13.2 Normal and active galaxies

Are nebula as e.g. Andromeda galactic objects as globular clusters? Or is our galaxy just one of many in the Universe? Ernst Julius Opik estimate 1922 the distance to Andromeda as follows: The rotation velocity at the edge of Andromeda was known from Doppler measurements. Expressing then the radius R of Andromeda as R = d, where d is its distance and the angular diameter, in the virial theorem, he obtained v2 GM = . (d)2 d (13.2)

The luminosity L and the observed ux F are connected by F = L/4d2 . Opik did not know L, but he assumed that M/L is quite universal. Thus he used M/L 3 determined L as average value for the Milky Way also for Andromeda. Inserting d2 = 4F on the LHS of

102

13.3 Normal Galaxies

Figure 13.5: Hubbles classication scheme for galaxies. Eq. (13.2) gives then d v 2 L 450 kpc . 4FGM (13.3)

This distance is much larger than the size of the galactic stellar disc. Edwin Hubble measured 1924 the distance to three spiral nebula including Andromeda by observing Cepheides, conrming that these are extragalactic objects.

13.3 Normal Galaxies

The total energy emitted by a normal galaxies is the sum of the emission from each of the stars found in the galaxy. Thus the emission is roughly thermal and mainly in the infrared, visible and ultraviolet bands.

13.3.1 Hubble sequence

Spiral galaxies are two-dimensional objects, with or without bar. They make up 2/3 of all bright galaxies. They contain gas and dust; young and old stars. The stars move regularly in the disc. The surface luminosity of spiral galaxies decreases exponentially for large radii, L(r) = L0 exp(r/r0 ) with r0 5 kpc. Elliptical galaxies have a three-dimensional, elliptical appearance. Most common are dwarf elliptical galaxies, with sizes of only a few kpc. But there exist also giant elliptical galaxies with an extension up to 100 kpc. Their low gas content (< 1%) excludes the idea of an evolution from elliptical to spiral galaxies by attening. Elliptical galaxies contain no young stars. Stars are moving randomly. Their surface luminosity decreases as L(r) = L0 exp(r/r0 )1/4 with a large variation of r0 . S0 and irregular galaxies are the remaining ones.

103

13 Galaxies

Figure 13.6: Rotation curves of two galaxies, superimposed are optical images.

13.3.2 Dark matter in galaxies

Flat rotation curves as far as luminous matter extends are found in practically all galaxies. Hence, as for the Milky way, v(r) = const. corresponds to 1/r2 , compared to an exponential fall-o of luminous matter. What are potential explanations for this discrepancy? Dust and gas clouds are seen by their blocking eect on light or in radio. Their contribution to the mass density has been accounted for. Most of the mass contained in galaxies could consist of non-luminous objects, so-called MAssive Compact Halo Object (MACHOs) like brown dwarfs, Jupiter-like objects, or back holes. This possibility can be tested by microlensing, i.e. the occasional amplication of light from extragalactic stars by the gravitational lens eect. Searches for microlensing towards the Large Magellanic Cloud exclude MACHOs as main component of dark matter in the Milky Way. Gravity has to be modied. The missing matter is composed of a new stable, neutral particle: dark matter.

Modication of gravity (MOND) Newtons law of inertia F = ma could be wrong for the very small accelerations that are relevant for the dynamics of galaxies. Such an eect could be incorporated by introducing a function in Newtons law, m(a/a0 )a = F (13.4)

with (x) 1 for x and (x) x for x 0. Choosing a0 1012 cm/s2 , all Newtonian standard results and observations are reproduced. Alternatively, Newtonian gravity could be modied, (g/a0 )g = gN . (13.5) As long as gravity is the only force relevant for the dynamics, the two alternatives cannot be distinguished. MOND describes a large amount of data successful and also a relativistic generalization of MOND is possible. However, MOND ts many cosmological observations (that we shall discuss later) worse than the hypothesis that a large fraction of the mass of the universe is in form of a new dark matter particle.

104

13.3 Normal Galaxies

Figure 13.7: A potential MACHO microlensing event. Dwarf galaxies and neutrino as DM If the mass of galaxies consists indeed mainly of an elementary particle dierent to protons, then our only candidate for the dark matter particle we know are neutrinos. As neutrino oscillations show, they are massive and they are most likely stable even on time-scales much larger than the age of the universe. However, we can use an argument similar to the one we used to derive the degeneracy pressure of fermions to exclude this possibility. Since neutrinos are fermions, they cannot be packed arbitrarily dense. Together with the upper mass limit m < 3 eV 6 106 me from beta-decay experiments, the Pauli principle excludes neutrinos as main component of dark matter in galaxies. Ex.: Neutrinos as dark matter.
Consider cold neutrinos with arbitrary mass m as explanation for dark matter in spiral and dwarf galaxies. (Use M 1011 M and R = 50kpc for spiral and M 107 M and R = 0.1kpc for dwarf galaxies.) Derive a lower limit for the neutrino mass m using that neutrinos are fermions. (Hint: connect the escape velocity from the galaxy with the maximal momentum of a degenerate Fermi gas.) A neutrino bound to a galaxy has energy E = GM m/R + mv 2 /2 0, hence its momentum should be smaller than pesc = mvesc = 2GM/R. On the other hand, the number N of possible neutrino states inside the volume V = 4R3 /3 up to pesc can be estimated with Heisenbergs uncertainty principle: In one dimension xpx / > 1 and thus N (1/ 3 ) d3 xd3 p = 4 3 4 R (pesc / )3 . 3 3

Since the Pauli principle forbids that 2 fermions occupy the same state, N is equal to the maximal number of neutrinos in the galaxy. The total mass in neutrinos follows as M = Nm 4 3
2

m4 (2GM R/ 2 )3/2

105

13 Galaxies
or neglecting numerical factors For typical spiral galaxies, m > 10 eV, while for typical dwarf galaxies m > 500 eV. The current experimental upper limit is m < few eV and excludes therefore neutrinos as main component of DM. m > (G3 R3 M/ 6 )1/8 .

13.3.3 Galactic evolution

Galaxies are not isolated objects that are well-separated from each other. Instead they are bound together forming larger clusters, collide with each other and may merge. We discuss just one example for what might happen to a smaller body moving in the dark matter cloud of a larger one. For a concrete realization one can think at a dwarf galaxy like the Magellanic clouds orbiting around the Milky Way. Dynamical friction What happens when a test body of mass M moves through a background of matter? Our test body will attract matter towards his position, creating a high-density wake trailing with him. The energy transferred to the surrounding material has to be supplied by the kinetic energy of our test body. As a result, a force F know as gravitational drag or dynamical friction acts on M . We can obtain an understanding of this eect using just dimensional analysis. Since the force F can depend only on GM , v and , the unique combination with the dimension of a force is (cf. Exercise 13.1) (GM )2 F C , (13.6) v2 where C is a dimensionless constant. This constant has to be determined from numerical simulations and varies between 10100. Inserting the density distribution appropriate for a dark matter halo, = v 2 /(4r2 G), the force acting on a star cluster of mass M moving around a galaxy is F C GM 2 . 4r2 (13.7)

The angular momentum of the cluster is L = M vr. Since F acts tangentially to the obit (assumed to be circular), the torque = dL/dt = r F is simply dL = rF . dt Using v(r) = const. motivated by the observed rotation curves, we obtain Mv GM 2 dr = C . dt 4r (13.9) (13.8)

Separating the variables and integrating from the initial distance R until zero gives the lifetime of a cluster moving through background matter with density ,
0 R

dr r =

CGM 4v

dt
0

(13.10)

106

13.4 Active Galaxies and non-thermal radiation or = Ex.:

2vR2 . CGM

(13.11)

i) The age of the oldest globular cluster in the Milky Way is t = 13 Gyr. Estimate with C = 76, M = 5 106 M and v = 220 km/s the maximal distance from which globular clusters could have been spiraled into the center of the galaxy. ii) The large Magellanic Cloud (LMC), which has M = 21010 M , orbits the Milky Way at a distance of 51 kpc. Assuming that the Milky Ways dark matter halo and at rotation curve extends out to the LMC, estimate how much time it will take for the LMC to spiral into the Milky Way (C = 23). For both questions we use 1/2 C GM R= 2v

solved either for R or . Solution to ii) is R = 4 kpc and to iii) = 1.7 Gyr. Thus inside 4 kpc there should be no globular clusters, because they spiraled already into the center of the Milky Way and were dissolved, while the LMC has still some time to go. . .

13.4 Active Galaxies and non-thermal radiation

Active galaxies is a common name for all galaxies with unusual, mostly non-thermal emission not associated with stars. The modern view is that the source of the non-thermal radiation is connected to the SMBH in their center. We will therefore rst discuss the two most important mechanisms to generate non-thermal photons.

13.4.1 Non-thermal radiation

Synchrotron radiation An electron in a homogeneous magnetic eld moves on a Larmor orbit with radius rL = v mc/(eB). Because of its acceleration, the electron emits electromagnetic radiation. For typical magnetic eld strengths found in galaxies, B G, and relativistic electrons, the emitted radiation is in the radio range. We start by considering non-relativistic electrons. Then the Lorentz force is e FL = v B = ma . c (13.12)

Hence the motion parallel to B is force-free (v v = 0), while the acceleration in the plane perpendicular to B is constant, ar = (e/m)v B. Thus the electron moves along a helix. Equating the Lorentz and the centripetal force, (e/c)v B =
2 mv , r

(13.13)

and introducing the angular frequency = 2 = v /r, the cyclotron frequency 0 follows as eB or 0 /Mhz = 2.8B/G . (13.14) 0 = mc As long as electrons are non-relativistic, i.e. their total energy is dominated by their rest energy mc2 , they emit cyclotron radiation with the same frequency, independent of their velocity.

107

13 Galaxies

Figure 13.8: The emitted ux F from the BL Lacartae Markarian 421 as function of frequency , in a quiet (lower line) and a aring state (upper line). For a relativistic electron, we have to replace ma by ma and thus the cyclotron frequency eB eB becomes c = mc = E/c . However, there are two additional eects. First, the radiation is beamed and emitted in a small cone of opening angle 1/. Second, there is a time-dilation between the frame of the electron and the observer (compare with Eq. (13.16)), dt = 1 cos() 1 + ()2 /2 2 . d (13.15)

As consequence, all frequencies from 0 to 2 0 are emitted, although most energy goes into the frequency range around cr = (3/2) 2 0 . Ex.: Determine the Larmor radius rL , its orbital period P , the opening angle and the frequency cr in which the radiation is mainly emitted, for an electron with = 104 in a magnetic eld with strength typical for galaxies, B = 3G. The Larmor radius is rL = 6.0 1012 cm, the orbital period with P = 1/ = /0 = 20min, the opening angle = 0.005 and the frequency cr = 890Mhz.

Inverse Compton scattering The usual text book set-up for Compton scattering assumes an energetic photon hitting an electron at rest, transferring part of its energy to the electron. In an astrophysical environments, the opposite situation is realized more often: A fast electron hits a low-energy photon and transfers a large fraction of its energy to the photon. (Inverse Compton scattering). Thus charged electron are accelerated by electromagnetic elds, emit synchrotron radiation in the radio range as well as high-energy photons via inverse Compton

108

13.4 Active Galaxies and non-thermal radiation

Figure 13.9: Left:Schematic picture of a radio galaxy. Right: NGC4258 scattering. An example where this two dierent components of the electromagnetic radiation are visible is shown in Fig. 13.8.

13.4.2 Radio galaxies

Radio galaxies emit typically 106 times more energy in the radio range than normal galaxies. The main emission mechanism is synchrotron radiation of electrons. The radio emission comes mainly from two radio lobes separated by up to a distance of order 10 Mpc, cf. Fig. 13.9. Additionally, there is a weaker radio source in center of the galaxy. Radio galaxies appear in the visible light as giant elliptical galaxies. The lobes are caused by the interaction of a jet of high energy particles and the intergalactic medium: The medium slows down the jet, material is dragged along with the jet, while higher velocity particle are coming from behind. Thus a shocks forms and creates a Hot Spot, which is the site of most intense emission, see Fig. 13.9. Superluminal motion Consider a source moving with velocity v from A to B towards an observer. Denote the dierence in radial distance to the observer by x , the perpendicular one by x . Then a light signal from B needs x /c less time than a signal from A. Thus, the apparent time for the object to travel from A to B the distance r is tapp = r x r r r = cos = (1 cos ) , v c v c v (13.16)

where r is the distance between A and B. The apparent velocity across the sky is then vapp = x = tapp
r v (1

v sin r sin = . cos ) 1 cos

(13.17)

109

13 Galaxies For v c, we obtain the expected result vapp v sin , while for 1 the apparent velocity can exceed the speed of light. We determine the maximal value of the apparent velocity by rst deriving the maximum of vapp as function of , dvapp cos 0 (1 cos 0 ) sin2 0 = 0 d or cos 0 = . Inserting the latter condition into Eq. (13.17) gives then vapp v
max

(13.18)

1 =. (1 2 )1/2

(13.19)

Hence the apparent velocity is maximally increased by the Lorentz factor of the source towards the direction cos = . Jet beaming Another important relativistic eect is jet beaming. Consider an emission process that is isotropic in the rest frame of the emitting source. Let us use the Lorentz transformation between energies in the rest system of the source (primed) and the observer system (unprimed). In the observer system, E = (E + p cos ) (13.20)

where is the angle between the velocity of the source and the emitted photon. The maximal/minimal values of E follow directly as E = (E p ) for cos = 1, i.e. if the photons are emitted parallel and anti-parallel to the direction of the source. For a relativistic source, 1, and the observed energy of photons in the backward direction goes to zero. Moreover, a similar eect leads also to an increase of the number of photons observed in the direction parallel to the source velocity. As a result, the emitted luminosity of a relativistic source (e.g. the hot spot of a radio galaxy) moving towards us can be increased by a large factor, 103 104 , compared to isotropic emission.

13.4.3 Other AGN types and unied picture

Seyfert galaxies In 1943, Carl Seyfert noticed that certain nearby spiral galaxies have very bright, pinpoint nuclei. The spectra of these galaxies show very strong, often broad, emission lines. The brightness of the cores of Seyfert galaxies uctuates: The light from the central nucleus varies in less than a year, which implies that the emitting region must be less than one light year across. They do not have radio lobes. Most are powerful sources of infrared radiation. In addition, some emit intensely in the radio, X ray, and gamma ray regimes. Approximately 2% of all spiral galaxies are Seyfert galaxies. Seyfert galaxies are divided into two classes, based upon the widths of their spectral emission features. Seyfert 1 galaxies have hydrogen emission features with very large widths, indicating that the gas in the galaxys central regions is moving with velocities of several thousand km/sec (Seyfert 1 galaxies show velocities up to almost 0.1c). Seyfert 2 galaxies have much narrower emission features implying much lower velocities (note that the Seyfert 1 in Fig. 13.10 shows narrow features as well). BL Lacertae Main characteristics of BL Lacertae is their fast variability, with night-to-night variations of 10-30%, or a factor 100 within weeks if the source ares. Their spectra show no emission lines, and their appearance in the optical is point-like.

110

13.4 Active Galaxies and non-thermal radiation

Figure 13.10: The spectra of Seyfert 1 (left) and Seyfert 2 (right) galaxies. Quasars Quasi-stellar objects were discovered in late 1963, when a star-like object with m = 13 was identied with a strong radio source. Emission line as Balmer lines, but displaced by / = 0.15 were found. If interpreted as redshift due to expansion of Universe, quasars are at cosmological distance. For the particular case of the quasi-stellar objects found 1963, the distance is 640 Mpc. This requires enormous luminosities: The distance modulus is m M = 5 log(d/10pc) = 39 (13.21)

of M = 26. For comparison, the absolute magnitude of the Sun is 5, a dierence of 31 corresponds to a brightness ratio of 1012 . In the radio, the source emits even stronger. Energy gain from accretion onto SMBHs: The maximal energy gain is Emax = GmM/RS , where RS = 2GM/c2 , and thus Emax = mc2 /2. Modelling the accretion process gives a maximal eciency of = 10%20%. Thus the luminosity from accretion is L= c2 dm 2dt (13.22)

For dm/dt = 1M /yr, one obtains 61045 erg/s or L 1012 L . Thus the large energy output of AGN can be, at least in principle, easily explained by accretion onto an supermassive black hole. Unied picture The fast variability and the large energy output point to accretion on SMBH as main energy source of the AGN activity. We observe dierent types of AGN due to their time evolution (e.g. from Quasars to Seyfert), dierent angles of view and dierent stages (active vs. quiet due to changes in dm/dt). A cartoon sketching the dierent types of AGN is shown in Fig. 13.11.

Exercises
1. Dynamical friction. Show that the unique combination with the dimension of GM , v and with the dimension of a force is F (GM )2 /v 2 .

111

13 Galaxies

Figure 13.11: The unied picture for AGN explains the large AGN variety mainly as dierences in the viewing angle relative to the jet axis.

112

Part III

Cosmology

113

14 Overview: Universe on large scales

14.1 Problems of a static, Newtonian Universe
Newtons law of universal gravitation changed together with the Copernican principle, i.e. the assumption that all bodies in the Universe obey the same laws as measured on Earth, cosmology from a purely philosophical subject to a sub-discipline of physics. In Newtonian gravity, a static Universe has to be innite. However, a static innite Universe would be an unstable solution and over-dense regions should collapse: As we discussed for the example of star clusters, gravitationally bound systems do not have an equilibrium solution and the second law of thermodynamics, dS 0, requires their collapse. Physicists around the year 1850 were concerned about the opposite faith of the Universe: A state without temperature or density gradients called heat death. Olbers paradox addresses the question why is the night sky dark? The number of stars in shell at distance r is proportional to 4r2 dr, and thus the observed ux F 1/r2 from each shell is constant. If the Universe is innite and has no beginning, the total ux diverges. Thus the darkness of the night sky is inconsistent with an Universe extending innitely in space and time.

14.2 Einsteins cosmological principle

Einstein postulated that the Universe is homogeneous and isotropic (at each moment of its evolution). An important question is then at which scale the hierarchical structure of the universe (galaxies, cluster of galaxies, superclusters, . . . ?) stops. In the extreme case of a fractal Universe an innite sequence n of structures exists. Imagine then the case that a structure of order n contains pn stars of mass M and occupies the volume kq n . Then the density n averaged over structures of order n is n M k p q
n

with limit n = 0 for p < q and n . Hence, physical quantities O like 0 R = R (4/V ) 0 drr2 O(r) depend on the averaging scale R and the simple picture of an universe with, e.g., the average density would be not possible. Observations show that the hierarchical structure of the real universe stops at the scale of superclusters the Universe is homogeneous on scales larger than a few 100 Mpc. There is just a factor 30 between the size of the largest structures and size of the observable universe. Hubbles F 3/2 test If we correct for extinction, we can use F = L/(4r2 ) assuming an euclidean geometry of space. If furthermore all galaxies would have the same luminosity L,

114

14.3 Expansion of the Universe: Hubbles law then all galaxies brighter than > F would be closer than r. Thus the total number of galaxies brighter than a certain limiting value is N (> F) = nV = 4n 3 L 4F
3/2

F 3/2 .

(14.1)

Around 1920, it was known that star counts vary less rapidly, i.e. the edge of the Milky Way became visible. Hubble applied the same argument 1926 to galaxies and found agreement with the F 3/2 prediction. This was the rst evidence for an homogeneity of the universe on large scales.

14.3 Expansion of the Universe: Hubbles law

Hubbles law Hubble found empirically that the spectral lines of distant galaxies are redshifted, z = /0 > 1, with a rate proportional to their distance d, cz = H0 d . (14.2)

If this redshift is interpreted as Doppler eect, z = /0 = vr /c, then the recession velocity of galaxies follows as v = H0 d . (14.3) The restriction distant galaxies means more precisely that H0 d vpec few 100 km/s. In other words, the peculiar motion of galaxies caused by the gravitational attraction of nearby galaxy clusters should be small compared to the Hubble ow H0 d. Note that the interpretation of v as recession velocity is problematic. The validity of such an interpretation is certainly limited to v c. The parameter H0 is called Hubble constant and has the value H0 71+4 km/s/Mpc. We 3 will see soon that the Hubble law Eq. (14.3) is an approximation valid for z 1. In general, the Hubble constant is not constant but depends on time, H = H(t), and we will call it therefore Hubble parameter for t = t0 . Hubbles law does not imply that we live at a special place of the Universe: Imagine a balloon that is inated during the time dt so that all distances increase by the factor (1 + da). Then original distances l0 are changed to (1 + da)l0 and the recession velocity is v= da l0 = al0 . dt (14.4)

If we write l(t) = l0 a(t), then a(t) is a scale factor describing how distances or space itself is stretched and a (14.5) v = al0 = l . a Comparing Eq. (14.3) and (14.5) we can identify H as H = a/a, i.e. the Hubble parameter determines the relative change of distance per time. The Hubble constant has the dimension of an inverse time. If v would be constant, then all galaxies and space would have been expanded from one single point to its present size in the time 1 (70km/s/Mpc)1 4 1017 s 13Gyr . (14.6) H0

115

14 Overview: Universe on large scales objects nearby stars Galactic clusters MS stars in Galaxy Cepheides Type Ia supernovae spiral galaxies galaxies methods trigonometric parallaxes main sequence tting spectroscopic parallax period-luminosity relation standard candle Tully-Fisher relation Hubbles law basis of calibration radar determination of AU main sequence xed by nearby stars trig. parallaxes/moving cluster trig. parallaxes/color-magnitude diagrams color-magnitude diagrams SN nearby in galaxies nearby galaxies nearby galaxies

Table 14.1: Several distance indicators together with the underlying calibration method.
1 How precise might be t0 = H0 as our rst estimate for the age of the Universe? Since gravity should decelerate the expansion of the Universe, we might anticipate that a proper calculation should lead to a smaller age of the Universe. On the other hand, the age of the oldest stars in the Universe is estimated as 13 Gyrthus there is some tension between a decelerating universe and a large value of H0 . Note also that the actual size of the balloon at time t is not directly connected to the size of the observable part c/H0 of the balloon. The former can be much larger than c/H0 or even innite.

Hubbles law as consequence of homogeneity Consider Hubbles law as a vector equation with us at the center of the coordinate system, v = Hd . (14.7)

What sees a dierent observer at position d ? He has the velocity v = Hd relative to us. We are assuming that velocities are small and thus v v v = H(d d ) = Hd , (14.8)

where v and d denote the position relative to the new observer. A linear relation between v and d as Hubble law is the only relation compatible with homogeneity and thus the cosmological principle.

14.3.1 Cosmic distance ladder

The Table 14.1 summarizes a few methods to measure astronomical distances. The chain of overlapping methods by which astronomers establish distance scales in the universe is called cosmic distance ladder: Every extension of the distance ladder inherits all the uncertainties of the previous steps it is based on. It is therefore important to establish methods to measure cosmological distances calibrated with as few intermediate steps as possible. Hubbles law For many galaxies measurements of their redshift has been performed. Since the conversion from redshift to true distances depends on e.g. the numerical value of the Hubble parameter, astronomers often uses the redshift to indicate the extragalactic distances.

116

14.3 Expansion of the Universe: Hubbles law Tully-Fisher relation An empirical relation published 1977 between the luminosity of a spiral galaxies and the width of its spectral lines, especially the 21cm line. Since M v 2 and there is red- and blue-shifting on opposite sides, such a relation is expected. The detailed form has to derived empirically, since it depends on the distributions v(r) and L(r). One uses then the distance modulus to nd the distance from the luminosity and the apparent magnitude.

117

15 Cosmological models for an homogeneous, isotropic universe

15.1 Friedmann-Robertson-Walker metric for an homogeneous, isotropic universe
The spatial part dl of the line-element of special relativity, ds2 = c2 dt2 (dx2 + dy 2 + dz 2 ) = c2 dt2 dl2 , (15.1)

corresponds to the one of an usual euclidean three-dimensional space. Such a space is at, static, homogeneous and isotropic. An expanding universe means that at dierent times t, the distance dl between two observes at rest changes, ds2 = c2 dt2 a2 (t) dx2 + dy 2 + dz 2 = c2 dt2 a2 (t)dl2 . (15.2)

Our postulate of homogeneity requires that the dimensionless function a(t), the scale factor, depends only on t. Do other homogeneous, isotropic metrics exist? In three dimensions, the sphere S 2 is another example for an homogeneous and isotropic space. This space has constant positive curvature and its line-element is dl2 = r2 d2 + r2 sin2 d2 . Let us try to nd the equivalent expression for dl in 3+1 dimensions by considering a sphere S 3 embedded in a (ctitious!) 4-dimensional, x2 + x2 + x2 + x2 = R2 , (15.3) 1 2 3 4 euclidean space, dl2 = dx2 + dx2 + dx2 + dx2 . 1 2 2 4 (15.4) We choose x1 , x2 , x3 as the three coordinates describing the sphere and eliminate x4 using Eq. (15.3). Forming the dierential of Eq. (15.4) gives x1 dx1 + . . . + x4 dx4 = 0 and allows us thus to eliminate also dx4 , dl2 = dx2 + dx2 + dx2 + 1 2 2 (x1 dx1 + x2 dx2 + x3 dx3 )2 . R2 x2 x2 x3 1 2 3 (15.5)

This expression becomes more transparent in spherical coordinates, x1 = r cos sin , x2 = r sin sin , x3 = r cos . We calculate dxi = xi xi xi dr + d + d , r (15.6) (15.7) (15.8)

(15.9)

118

15.1 Friedmann-Robertson-Walker metric for an homogeneous, isotropic universe i.e. etc., and insert them into dl2 . After somewhat lengthy but elementary simplications we obtain dr2 dr2 dl2 = + r2 (sin2 d2 + d2 ) = + r2 d . (15.10) 1 r2 /R2 1 r2 /R2 Note that the angular part d corresponds to the usual expression for spherical coordinates, while the radial distance dr is modied by the factor (1 r2 /R(t)2 )1 . The only other form of a homogeneous, isotropic space is a hyperbolic plane with constant negative curvature. Formally, a hyperbolic plane can be identied with a sphere with imaginary radius R, i.e. it is obtained by replacement 1 r2 /R2 1 + r2 /R2 . Combining the three cases (and making the radial coordinate dimensionless, r r/R(t)), the Friedmann-Robertson-Walker (FRW) metric in its most commonly used representation follows as dr2 + r2 (sin2 d2 + d2 ) (15.11) ds2 = c2 dt2 R2 (t) 1 kr2 with k = 1 (positive/negative curvature) or k = 0 (at three-dimensional space). dx1 = cos sin dr r sin sin d + r cos cos d,

Geometry of the FRW spaces

Let us consider a sphere of xed radius at xed time, dr = dt = 0. The line-element ds simplies then to R2 (t)r2 (sin2 d + d2 ), which is the usual line-element of a sphere S 2 with radius rR(t). Thus the area of the sphere is A = 4(rR(t))2 and the circumference of a circle is L = 2rR(t), while rR(t) has the physical meaning of a length. By contrast, the radial distance between two points (r, , ) and (r + dr, , ) is dl = R(t)dr/ 1 kr2 . Thus the radius of a sphere centered at r = 0 is for k = 1 , arcsin(r) r dr r for k = 0 , = R(t) l = R(t) (15.12) 1 kr 2 0 arcsinh(r) for k = 1 . Hence for k = 0, i.e a at space, one obtains the usual result L/l = 2, while for k = 1 (spherical geometry) L/l = 2r/ arcsin(r) < 2 and for k = 1 (hyperbolic geometry) L/l = 2r/arcsinh(r) > 2. For k = 0 and k = 1, l is unbounded, while for k = +1 there exists a maximal distance lmax (t). Hence the rst two case correspond to open spaces with an innite volume, while the latter is a closed space with nite volume.

Lema tres redshift formula

A light-ray propagates with v = c or ds2 = 0. Assuming a galaxy at r = 0 and an observer at r, i.e. light rays with d = d = 0, we rewrite the FRW metric as dr dt = . (15.13) R 1 kr2 We integrate this expression between the emission and absorption times t1 and t2 of the rst light-ray, t2 r dt dr (15.14) = R 1 kr2 t1 0 119

15 Cosmological models for an homogeneous, isotropic universe

t1 + t1

! !!

! !! !!

!! !!

! !! ! !!

!! !!

! !! ! !!

! ! t2 + t2 ! ! t2

! !!

! t1 !

! !!

galaxy, r = 0

observer, r

Figure 15.1: World lines of a galaxy emitting light and an observer at comoving coordinates r = 0 and r, respectively. and between t1 + t1 and t2 + t2 for the second light-ray (see also Fig. 15.1),
t2 +t2 t1 +t1

dt = R

r 0

dr . 1 kr2 dt . R

(15.15)

The RHSs are the same and thus we can equate the LHSs,
t2 t1

dt = R

t2 +t2 t1 +t1

(15.16)

We change the integration limits, subtracting the common interval [t1 + t1 : t2 ] and obtain
t1 +t1 t1

dt = R

t2 +t2 t2

dt . R

(15.17)

Now we choose the time intervals ti as the time between two wave crests separated by the wave lengths i of an electromagnetic wave. Since these time intervals are extremely short compared to cosmological times, ti = i /c ti , we can assume R(t) as constant performing the integrals and obtain t1 t2 1 2 = or = . (15.18) R1 R2 R1 R2 The redshift z of an object is dened as the relative change in the wavelength between emission and detection, 2 1 2 z= = 1 (15.19) 1 1 or R2 2 = . (15.20) 1+z = 1 R1 This result is intuitively understandable, since the expansion of the universe stretches all lengths including the wave-length of a photon. For a massless particle like the photon, = c

120

15.2 Friedmann equation from Newtons and Hubbles laws and E = cp, and thus its frequency (energy) and its wave-length (momentum) are aected in the same way. By contrast, the energy of a non-relativistic particle with E mc2 is nearly xed. A similar calculation as for the photon shows that indeed not the energy but the momentum p = / of massive particles is red-shifted. To be able to compare Eq. (15.20) with Hubbles law, v = Hd with v = cz and H = R/R, we have to understand next how the scale factor R evolves with time.

15.2 Friedmann equation from Newtons and Hubbles laws

The Friedmann equation connects the expansion rate H of the Universe with its energy density. Since we assume the cosmological principle to be valid, i.e. we assume an homogeneous and isotropic Universe, this equation is a single ordinary dierential equation. We can derive this equation in a straightforward manner using Hubbles law together with the Newtonian expression for the energy of a gravitating system plus the insight of special relativity that gravity should couple also to energy.

15.2.1 Friedmann equation

Let us consider the (Newtonian) energy of a galaxy of mass m at the distance R in a Universe that is very close to homogeneity and isotropy, 1 GmM (R) E = Ekin + Epot = mR2 . 2 R (15.21)

According to Hubbles law we can express the recession velocity R of the galaxy as R = HR, 3 . Hence while the mass enclosed within R is given by M (R) = (4/3)R m 4 1 E = m(HR)2 Gmm R2 2 3 or 8 2E = H2 Gm . 2 mR 3 (15.22)

(15.23)

Since we assume that the universe is homogeneous, H and m cannot be functions of R. 2E Therefore also the LHS, mR2 , cannot depend on the chosen distance R to the coordinate center. However, the value of 2E/(mR2 ) is time-dependent, because the distance between us and the galaxy will change as the Universe expands. Since the mass m of our test galaxy is arbitrary, we can choose it such that |2E/(mc2 )| = 1 holds at an arbitrary moment as long as E = 0. For dierent times, the LHS scales as 1/R2 and thus we can rewrite Eq. (15.23) with k = 0, 1 as 8 kc2 = H2 Gm . (15.24) 2 R 3 Since E is constant, k is constant too. Finally, we account for the equivalence of mass and energy by including in not only the mass but also the energy density, = m c2 + rad + . . .: H
2

R R

kc2 8 G/c2 2 . 3 R

(15.25)

121

15 Cosmological models for an homogeneous, isotropic universe The result is the Friedmann equation without cosmological constant and agrees exactly with the one derived from general relativity1 . The three cases k = 1, 0, +1 in the Friedmann equation correspond to the same three cases in Eq. (15.11). Hence they distinguish the geometry of space, i.e. the geometry of 3-dimensional hypersurfaces with t = const. This geometry (k and R) can be determined by measuring two simple numbers, H0 and 0 . We set from now on c = 1. Quantities evaluated at the present time are labelled with a zero, 2 3H0 t0 , H0 , . . .. The value of H0 xes the present density 0 for k = 0 as 0 (k = 0) cr = 8G . Most cosmological quantities like cr will depend on the actual value of H0 . One hides this dependence by introducing h, H0 = 100hkm/s/Mpc . Then one can express the critical density as function of h, cr = 2.77 1011 h2 M /Mpc3 = 1.88 1029 h2 g/cm3 = 1.05 105 h2 GeV/cm3 . Thus a at universe with H0 = 100hkm/s/Mpc requires an energy density of 10 protons per cubic meter. We dene the abundance i of the dierent players in cosmology as their energy density relative to cr , i = /cr . The ones we know already are non-relativistic or cold dark matter (CDM with m ) and relativistic matter or radiation (rad ). Cosmological constant In general relativity, the absolute value of the energy matters not only energy dierences. Thus we should add a constant (the cosmological constant ) to the RHS of the Friedmann equation to account for a possible intrinsic energy and pressure of the vacuum, 2 R k 8 2 H G 2 + . (15.26) = R 3 R 3 With = 0, the simple relation = cr k = 0 still holds, if the contribution of to the energy density is included viz. 8 G = . (15.27) 3 3 Thereby one recognizes also that the cosmological constant acts as a constant energy density = 8G , or = 3H 2 . 0 For = 0, an observable additional to 0 and H0 is needed to x the geometry of the three-dimensional spatial hypersurface: We obtain this observable by a Taylor expansion of R(t), 1 R(t) = R(t0 ) + (t t0 )R(t0 ) + (t t0 )2 R(t0 ) + . . . 2 1 2 = R(t0 ) 1 + (t t0 )H0 (t t0 )2 q0 H0 + . . . , 2 (15.28) (15.29)

where q R(t0 )R(t0 )/R2 (t0 ) is called deceleration parameter: If the expansion is slowing down, R < 0 and q0 > 0.
1

In general relativity, a generalization of Gauss theorem called Birkhos theorem is valid. Thus only the enclosed mass and energy enter the Friedmann equation, and choosing R suciently small, the Newtonian approximation becomes arbitrarily precise.

122

15.2 Friedmann equation from Newtons and Hubbles laws We show now that Hubbles law is indeed an approximation for small redshift. For not too large time-dierences, we can use the expansion Eq. (15.28) and write 1z R(t) 1 = 1 + (t t0 )H0 . 1+z R0 (15.30)

Hence Hubbles law, z = (t0 t)H0 = d/cH0 , is valid as long as z H0 (t0 t) 1. Deviations from its linear form arises for z > 1 and can be used to determine q0 .

15.2.2 Local energy conservation and acceleration equation

Additionally to the Friedmann equation we need one equation that describes how the energy content of the Universe is aected by its expansion. Local energy conservation The rst law of thermodynamics becomes with dQ = 0 (no heat exchange to the outside, since no outside exists) simply dU = T dS P dV = P dV or d(R3 ) = P d(R3 ) . Dividing by dt, R + 3( + P )R = 0 , we obtain = 3( + P )H . (15.34) (15.33) (15.32) (15.31)

Thus the expansion decreases the energy density both by dilution and by the work required to expand a gas with pressure P 0. Ex.: Derive how the expansion of the Universe changes the particle number density n The Hubble ow v = Hr induces the ux vn through the surface 4r2 of a sphere with radius r, and thus N = 4r2 vn. These particles are missing inside the sphere containing the total number of particles N = V n. Hence 4 3 r n (15.35) N = 4r2 vn = 3
or n = 3vn/r = 3Hn. Thus the rst term on the RHS of Eq. (15.34) corresponds indeed due the dilution of the particle number by the expansion of the Universe. More formally, we can use that n 1/R3 and thus dn dn dR n = = 3 R . dt dR dt R

123

15 Cosmological models for an homogeneous, isotropic universe Acceleration equation An equivalent, but often more useful equation is obtained by rst multiplying the Friedmann equation with R2 , 8G 2 R2 R2 = R k + . 3 3 Taking then the time derivative, 8G R2 + (2RR) + 2RR 2RR = 3 3 inserting R = 3( + P )R into R2 and dividing by 2R gives 4G [(3 3P ) + 2] R + R R= 3 3 and nally the acceleration equation 4G R = ( + 3P ) . R 3 3 (15.39) (15.38) (15.37) (15.36)

This equation determines the (de-) acceleration of the Universe as function of its matter and energy content. Normal matter is characterized by > 0 and p 0. Thus a static solution is impossible for a universe with = 0. Such a universe is decelerating and since today R > 0, R was always negative and there was a big bang. We can understand better the physical properties of the cosmological constant by replacing by (8G) . Now we can compare the eect of normal matter and of the term on the acceleration, R 8G 4G = ( + 3P ) (15.40) R 3 3 Thus is equivalent to matter with an E.o.S. w = P/ = 1. This property can be checked using only thermodynamics: With P = (U/V )S and U = V , it follows P = . The borderline between an accelerating and decelerating universe is given by = 3P or w = 1/3. The condition < 3P violates the so-called energy condition for normal matter in equilibrium. An accelerating universe requires therefore a positive cosmological constant or a dominating form of matter that is not in equilibrium. Note that the energy contribution of relativistic matter, photons and possibly neutrinos, is much smaller than the one of non-relativistic matter (stars and cold dark matter). Thus the pressure term in the acceleration equation can be neglected at the present epoch. Measuring R/R, R/R and xes therefore the geometry of the universe.

15.3 Scale-dependence of dierent energy forms

The dependence of dierent energy forms as function of the scale factor R can derived from energy conservation, dU = P dV , if an E.o.S. P = P () = w is specied. For w = const., it follows d(R3 ) = 3P R2 dR (15.41) or eliminating P d 3 R + 3R2 = 3wR2 . dR (15.42)

124

15.4 Cosmological models with one energy component Separating the variables, 3(1 + w) we can integrate and obtain R3(1+w) 3 R R4 = const. for matter (w = 0) for radiation (w = 1/3) for (w = 1) (15.44)

d dR = , R

(15.43)

This result can be understood also from heuristic arguments: (Non-relativistic) matter means that kT m. Thus = nm nT = P and nonrelativistic matter is pressure-less, w = 0. The mass m is constant and n 1/R3 , hence is just diluted by the expansion of the universe, 1/R3 .

Cosmological constant : From 8 G = one obtains that the cosmological constant 3 3 acts as an energy density = 8G that is constant in time, independent from a possible expansion or contraction of the universe. Note that the scaling of the dierent energy forms is very dierent. It is therefore surprising that just today, the energy in matter and due to the cosmological constant is of the same order (coincidence problem).

Radiation is not only diluted but the energy of each single photon is additionally redshifted, E 1/R. Thus the energy density of radiation scales as 1/R4 . Alternatively, one can use that = aT 4 and T E 1/R.

15.4 Cosmological models with one energy component

We consider a at universe, k = 0, with one dominating energy component with E.o.S w = P/ = const.. With = 0 (R/R0 )3(1+w) , the Friedmann equation becomes 8 2 3+3w R2 = GR2 = H0 R0 R(1+3w) , 3
2 where we inserted the denition of cr = 3H0 /(8G). Separating variables we obtain

(15.45)

(3+3w)/2 0

dR R(1+3w)/2 = H0
0

dt = t0 H0

(15.46)

and hence the age of the Universe follows as 2/3 2 1/2 = t0 H 0 = 3 + 3w

for matter (w = 0) for radiation (w = 1/3) for (w = 1)

(15.47)

Models with w > 1 needed a nite time to expand from the initial singularity R(t = 0) = 0 to the current size R0 , while a Universe with only a has no beginning. In models with a hot big-bang, , T for t 0, and we should expect that classical gravity breaks down at some moment t . As long as R t with < 1, most time elapsed during the last fractions of t0 H0 . Hence our result for the age of the universe does not depend on unknown physics close to the big-bang as long as w > 1/3.

125

15 Cosmological models for an homogeneous, isotropic universe If we integrate (15.46) to the arbitrary time t, we obtain the time-dependence of the scale factor, 2/3 for matter (w = 0) t R(t) t2/(3+3w) = (15.48) t1/2 for radiation (w = 1/3) exp(t) for (w = 1) General discussion: We apply now the Friedmann and the acceleration equation to the 2 present time. Thus R0 = R0 H0 , R = q0 H0 R0 and we can neglect the pressure term in Eq. (15.39), 4G R0 2 = q0 H0 = m,0 . (15.49) R0 3 3 Thus we can determine the value of the cosmological constant from the observables m,0 , H0 and q0 via 2 = 4Gm,0 3q0 H0 . (15.50) k 8G 2 2 = 3 m,0 + 3 H0 , R0

15.5 Determining and the curvature R0 from m,0 , H0 , q0

2 Solving next the Friedmann equation (15.26) for k/R0 ,

(15.51)

we write m,0 = m cr and insert Eq. (15.50) for . Then we obtain for the curvature term k H2 = 0 (3m 2q0 2) . 2 2 R0 (15.52)

Hence the sign of 3m 2q0 2 decides about the sign of k and thus the curvature of the universe. For a universe without cosmological constant, = 0, equation (15.50) gives m = 2q0 and thus k = 1 m < 1 m = 1 m > 1 q0 < 1/2 , q0 = 1/2 , q0 > 1/2 . (15.53)

k=0

k = +1

For a at universe with = 0, m,0 = cr and k = 0, 0 = 4G

2 3H0 2 2 + H0 (q0 1) = H0 8G

3 + q0 1 2

(15.54)

and thus q0 = 1/2. In this special case, q0 < 1/2 means k = 1 and thus an innite space with negative curvature, while a nite space with positive curvature has q > 1/2. Ex.: Comparison with observations: Use the Friedmann equations applied to the present time to derive central values of and k, R0 from the observables H0 (71 4) km/s/Mpc and 0 = (0.27 0.04)cr , and q0 = 0.6. Discuss the allowed range and signicance of the values. We determine rst 2 7.1 106 cm 2 5.2 1036 s2 H0 s 3.1 1024 cm 126

15.5 Determining and the curvature R0 from m,0 , H0 , q0

2 1.8 1.6

t0H0

1.4 1.2 1 0.8 0.6 0 0.2 0.4 m 0.6 0.8 1 open m+=1

Figure 15.2: The product t0 H0 for an open universe containing only matter and for a at cosmological model with + m = 1.
The value of the cosmological constant follows as
2 2 = 4Gm,0 3q0 H0 = 3H0

3q0 2cr

1 2 2 3H0 ( 0.27 + 0.6) 0.73 3H0 2

or = 0.73. The curvature radius R follows as k 2 R0 = =

2 2 4Gm,0 H0 (q0 + 1) = 3H0

q0 + 1 2cr 3

(15.55) (15.56)

2 2 3H0 (0.135 0.02 0.4/3) = 3H0 (0.002 0.02)

thus a at universe (k = 0) is consistent with observations.

Age problem of the universe. The age of a matter-dominated universe is (expanded around 0 = 1) 1 2 1 (0 1) + . . . . t0 = (15.57) 3H0 5 Globular cluster ages require t0 13 Gyr. Using 0 = 1 leads to H0 2/3 13Gyr = 1/19.5Gyr or h 0.50. Thus a at universe with t0 = 13 Gyr without cosmological constant requires a too small value of H0 . Choosing m 0.3 increases the age by just 14%. The age t0 of a at Universe with m + = 1 is given by 3t0 H0 1 1 + = ln . 2 1 (15.58)

Requiring H0 65 km/s/Mpc and t0 13 Gyr means that the function on the RHS should be larger than 3 13Gyr 0.65/(2 9.8Gyr) 1.3 or = 0.55.

127

15 Cosmological models for an homogeneous, isotropic universe

15.6 The CDM model

We consider a at Universe containing as its only two components pressure-less matter and a cosmological constant, m + = 1. Then the curvature term in the Friedmann equation and the pressure term in the deceleration equation play no role and we can hope to solve these equations for a(t). Multiplying the deceleration equation (15.39) by two and adding it to the Friedmann equation (15.26), we eliminate m , 2 a + a a a
2

= .

(15.59)

Next we rewrite rst the LHS and then the RHS as total time derivatives: With d (aa2 ) = a3 + 2aa = aa2 a dt we obtain Integrating is now trivial, a a
2

a a

(15.60)

1 d 3 d (aa2 ) = aa2 = (a ) . dt 3 dt aa2 =

(15.61)

3 a +C. (15.62) 3 The constant C can be determined most easily by setting a(t0 ) = 1 and comparing the Friedmann equation (15.26) with (15.62) for t = t0 as C = 8Gm,0 /3. Next we introduce the new variable x = a3/2 . Then da dx da dx 2x1/3 = = , dt dt dx dt 3 and we obtain as new dierential equation x2 x2 /4 + C/3 = 0 . (15.64) Inserting the solution x(t) = A sinh( t/2) of the homogeneous equation xes A as A = 3C/. This can be expressed by the current values of i as A = m / = (1 )/ . Hence the time-dependence of the scale factor is (15.65) a(t) = A1/3 sinh2/3 ( 3t/2) . The time-scale of the expansion is set by t = 2/ 3. The present age t0 of the universe follows by setting a(t0 ) = 1 as t0 = t arctanh( ) . (15.66) (15.63)

The deceleration parameter q = /aH 2 is an important quantity for observational tests a of the CDM model. We calculate rst the Hubble parameter H(t) = and then a 2 = coth(t/t ) . a 3t (15.67)

1 q(t) = [1 3 tanh2 (t/t ) . 2

(15.68)

128

15.6 The CDM model

0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 0 0.5 1 t/t0 1.5 2 =0.9 q =0.1

Figure 15.3: The deceleration parameter q as function of t/t0 for a CDM model and various values for (0.1, 0.3, 0.5, 0.7 and 0.9 from the top to the bottom). The limiting behavior of q corresponds with q = 1/2 for t 0 and q = 1 for t as expected to the one of a at m = 1 and a = 1 universe. More interesting is the transition region and, as shown in Fig. 15.3, the transition from a decelerating to an accelerating universe happens for = 0.7 at t 0.55t0 . This can easily converted to redshift, z = a(t0 )/a(t ) 1 0.7, that is directly measured by Supernova observations.

Exercises
1. Derive the relation between temperature and time in the early (radiation dominated) universe using = gaT 4 = g 2 T 4 /30 as expression for the energy density of a gas with g relativistic degrees of freedom in the Friedmann equation. What is the temperature at t = 1 s? [Hints: The expression = g 2 T 4 /30 is valid for k = c = = 1. Then one can measure temperatures in MeV/GeV and use s1 = 6.6 1025 GeV and G1/2 = MPl = 1.2 1019 GeV.]

129

16 Early universe
16.1 Thermal history of the Universe - Time-line of important dates
Dierent energy form today. Let us summarize the relative importance of the various energy 2 forms today. The critical density cr = 3H0 /(8G) has with h = 0.7 today the numerical value cr 7.3 106 GeV/cm3 . This would corresponds to roughly 8 protons per cubic meter. However, main player today is the cosmological constant with 0.73. Next comes (pressure-less) matter with m 0.27. The energy density of cosmic microwave background (CMB) photons with temperature T = 2.7 K = 2.3 104 eV is = aT 4 = 0.4 eV/cm3 or 5 105 . Dierent energy forms as function of time The scaling of i with redshift z, 1 + z = R0 /R(t) is given by
2 H(z)2 /H0 = m,0 (1 + z)3 + rad,0 (1 + z)4

+ ,0 (tot,0 1)(1 + z)2 .

Thus the relative importance of the dierent energy forms changes: Going back in time, one enters rst the matter-dominated and then the radiation-dominated epoch. The cosmic triangle shown in Fig. 16.1 illustrates the evolution in time of the various energy components and the resulting coincidence problem: Any universe with a non-zero cosmological constant will be driven with time to a x-point with m , k 0. The only other constant state is a at universe containing only matterhowever, this solution is unstable. Hence, the question arises why we live in an epoch where all energy components have comparable size. Temperature increase as T 1/R has three main eects: Firstly, bound states like atoms and nuclei are dissolved when the temperature reaches their binding energy, kT > Eb . Sec ondly, particles with mass mX can be produced, when kT > 2mX , in reactions like XX. Thus the early Universe consists of a plasma containing more and more heavier particles that are in thermal equilibrium. Finally, reaction rates = nv increases, since n T 3 . Therefore, reactions that have became ineective today were important in the early Universe. Matter-radiation equilibrium zeq : The density of matter decreases slower than the energy density of radiation. Going backward in time, there will be therefore a time when the density of matter and radiation were equal. Before that time with redshift zeq , the universe was radiation-dominated , rad,0 (1 + zeq )4 = m,0 (1 + zeq )3 (16.1)

130

16.1 Thermal history of the Universe - Time-line of important dates

0.0

1.0
0.5
OCDM

0.5

m
1.5

1.0
SCDM

OPEN
CDM

FLAT

0.0

2.0 0.0 0.5

CLOSED
1.0

-0.5
1.5

Figure 16.1: The cosmic triangle showing the time evolution of the various energy components. or zeq = m,0 1 5400 . rad,0 (16.2)

This time is important, because i) the time-dependence of the scale factor changes from R t2/3 for a matter to R t1/2 for a radiation dominated universe, ii) the E.o.S. and thus the 2 2 speed of sound changed from w 1/3, vs = (P/)S = c2 /3 to w 0, vs = 5kT /(3m) c2 . The latter quantity determines the Jeans length and thus which structures in the Universe can collapse. Recombination zrec : Today, hydrogen and helium in the interstellar and intergalactic medium is neutral. Increasing the temperature, the fraction of ions and free electron increases, i.e. the reaction H + H + + e that is mainly controlled by the factor exp(Eb /T ) will be shifted to the right. By denition, we call recombination the time when 50% of all atoms are ionized. A naive estimate gives T Eb 13.6 eV 160.000K or zrec = 60.000. However, there are many more photons than hydrogen atoms, and therefore recombination happens latter: A more detailed calculation gives zrec 1000. Since the interaction probability of photons with neutral hydrogen is much smaller than with electrons and protons, recombination marks the time when the Universe became transparent to light. Big Bang Nucleosynthesis At Tns mn mp 1.3 MeV or t 1 s, part of protons and neutrons forms nuclei, mainly 4 He. As in the case of recombination, the large number of photons delays nucleosynthesis relative to the estimate Tns to Tns 0.1 MeV.

131

16 Early universe Quark-hadron or QCD transition Above T m 100 MeV, hadrons like protons, neutrons or pions dissolve into their fundamental constituents, quarks q. Baryogenesis The fraction baryons (=protons, neutrons, and nuclei) contribute to the critical density is certainly limited by m 0.27. All the matter observed in the Universe consists of matter (protons and electrons), and not of anti-protons (anti-protons an positrons). Thus the baryon-to-photon ratio is =
6 nb n b < 8 10 108 . n 400

(16.3)

The early q q plasma contained a tiny surplus of quarks. After all anti-matter annihilated with matter, only the small surplus of matter remained. The tiny surplus can be explained by interactions in the early Universe that were not completely symmetric with respect to an exchange of matter-antimatter.

16.2 Big Bang Nucleosynthesis

Nuclear reactions in main sequence stars are supposed to produce all the observed heavier elements up to iron. However, stellar reaction can explain at most a fraction of 5% of 4 He, while the production of the weakly bound deuterium and Lithium-7 in stars is impossible. Thus the light elements up to Li-7 are primordial. Their abundance Y is Y (D) = few 105 , Y ( 3 H) = few 105 Y ( 4 He) 0.25, Y ( 7 Li) (1 2) 107 . Observational challenge is to nd as old stars/gas clouds as possible and then to extrapolate back to the primordial values. Ex.:
Estimate of 4 He production by stars: The binding energy of 4 He is Eb = 28.3 MeV. If 25% of all nucleons were fussed into t 10 Gyr by stellar fusion, the resulting luminosity-mass ratio would be L 1 Eb erg L = . 5 2.5 Mb 4 4mp t gs M
L The observed luminosity-mass ratio is however only Mb 0.05L /M . Assuming a roughly constant luminosity of stars, they can produce only 0.05/2.5 2% of the observed 4 He. 4

He during

Big Bang Nucleosynthesis (BBN) is controlled by two parameters: The mass dierence between protons and neutrons, mn mp 1.3 MeV and the freeze-out temperature Tf of reactions converting protons into neutrons and vice versa. We will treat BBN as a two-step process: First, the weak processes freeze-out and the n/p ratio becomes xed. Still, the temperature is high compared to Eb and nuclei cannot form. Meanwhile, neutrons decay. After the temperature dropped from Tf to TNS , most remaining neutrons are built into 4 He nuclei. Gamov criterion We introduced the optical depth = nl as the probability that a photon interacts in a slab of length l lled with targets of density n. If 1, photons and targets interact eciently and are in thermal equilibrium. We can apply the same criteria to the Universe: We say a particle species A is in thermal equilibrium, as long as = nl =

132

16.2 Big Bang Nucleosynthesis nv 1. The time corresponds to the typical time-scale for the expansion of the universe, = (R/R)1 = H 1 . (Since T 1/R and thus T /T = R/R = H, the time-scale for changes of the expansion rate and of the temperature are the same.) Hence we can rewrite this condition as nv H . (16.4) A particle species goes out of equilibrium when its interaction rate becomes smaller than the expansion rate H of the universe. Decoupling of neutrinos The cross section of reactions involving neutrinos in processes like n p + e + e or e+ e is G2 E 2 . If we approximate the energy of all particle F species by their temperature T , their velocity by c and their density by n T 3 , then the interaction rate of weak processes is vn G2 T 5 . F (16.5)

The early universe is radiation-dominated, rad 1/R4 , and its curvature k/R2 is negligible. Thus the Friedmann equation simplies to H 2 = (8/3)G with = g aT 4 , where g counts the number of relativistic degrees of freedom (photons, neutrinos, electrons) at that time. Requiring (Tfr ) = H(Tfr ) gives as freeze-out temperature Tfr of weak processes Tfr 1 MeV, at time tfr 0.3 s. Thus the time-sequence is as follows At Tfr 1 MeV, the neutron-proton ratio freezes-in and can be approximated by the ratio of their equilibrium distribution in the non-relativistic limit. At Tns 0.1 MeV, practically all neutrons are bound to 4 He, with only a small admixture of other elements. Proton-neutron ratio Above Tfr , reactions like e + n p + e keep nucleons in thermal equilibrium. As we have seen, Tfr 1 MeV and thus we can treat nucleons in the nonrelativistic limit. Then their relative abundance is given by the Boltzmann factor exp (/T ) for T > Tfr . Hence for Tfr , 1 nn = exp . (16.6) np t=tfr Tf 6 As the universe cools down to Tns , neutrons decay with half-live n 886 s, and thus nn np
t=tns

As the universe cools down from Tfr to Tns , neutrons decay with half-live n 886 s.

1 tns exp 6 n

2 . 15

(16.7)

Estimate of helium abundance The synthesis if 4 He proceeds though a chain of reactions, pn d, dp 3 He, d 3 He 4 Hep. Let us assume that 4 He formation takes place instantaneously. Moreover, we assume that all neutrons are bound in 4 He. We need two neutrons to form one helium atom, n(4 He) = nn /2, and thus Y (4 He) 2nn /np 4mN nn /2 4/15 4 M (4 He) = = = = 0.235 . Mtot mN (np + nn ) 1 + nn /np 15/15 + 2/15 17 (16.8)

133

16 Early universe

Figure 16.2: Abundances of light elements as function of (left) and of the number of light neutrino species (right). Our simple estimate is not too far away from Y 0.245. The abundance of Y ( 4 He) depends mainly on exp(/Tf ): The freeze-out temperature Tfr depends in turn on the number of relativistic particles at t 1 s and was used as a method to count the number of dierent light neutrino avors, cf. right panel of Fig. 16.2. Additionally, there is a weaker dependence on the start of nucleosynthesis Tns and thus b or b . Results from detailed calculations Detailed calculations predict not only the relative amount of light elements produced, but also their absolute amount as function of e.g. the baryon-photon ratio . Requiring that the relative fraction of helium-4, deuterium and lithium-7 compared to hydrogen is consistent with observations allows one to determine or equivalently the baryon content, b h2 = 0.019 0.001. Note that b m and hence BBN shows that dark matter cannot be in form of baryons (as e.g. the MACHOs hypothesis assumes). Although the binding energy per nucleon of Carbon-12 and oxygen-16 is higher than the of 4 He, these heavier elements are not produced in BBN: At the time of 4 He production, the Coulomb barrier prevents already fusion with such large electric charges. Also, a stable intermediate element with A = 5 is missing.

16.3 Structure formation

The left panel of Figure 16.3 shows the distribution of matter obtained in a numerical simulation, while the right panel shows the observed distribution of galaxies from an astronomical survey. On both panels, galaxies are distributed in a honeycomb-like structure: There are voids visible that are essentially free of matter, while most galaxies are concentrated at the corners of a web-like structure and along the connections between the corners, so-called laments. How can we transform these visual impressions into testable mathematical expressions?

134

16.3 Structure formation

z=0
SCDM CDM

CDM

OCDM

The VIRGO Collaboration 1996

Figure 16.3: Distribution of galaxies found in numerical simulations for four dierent cosmological scenarios (left) and from observations (right). Or more precisely, how can we test which one of the four dierent numerical simulation on the left agrees best with observations? In order to make such a comparison, one employs a statistical description, i.e. one asks questions like how big is the fraction of voids with volume between V1 and V2 , or how likely is it to nd a second galaxies separated by distance r from a rst one? A brief summary of the method is as follows: Structure formation operates via gravitational instability, as we discussed for star formation, and needs as starting point a seed of primordial uctuations. One derives from observations the power spectrum P (k) = |k |2 of density uctuations today, (x) ((x) )/: k d3 x eikx (x) .

Assuming initial uctuations Pi (k), one can calculate how they are transformed by gravitational instability and diusion of dierent particle species into a power-spectrum at present time, P (k) = T (k)Pi (k) . Then one can check for which cosmological parameters (H0 , i ,. . . ) and initial spectrum Pi (k) one nds agreement between observations and predictions.
2 The Jeans radius of radiation, P = /3 equals with vs = c2 /3 approximately the Hubble radius. Since uctuations with wave-length shorter than the Jeans radius oscillate (acoustic oscillations), and only modes with longer wave lengths can grow, structure formations becomes eective only in the matter-dominated epoch.

135

16 Early universe

16.4 Cosmic microwave background

Blackbody radiation in an expanding universe
We discussed earlier that the temperate of photons decreases as T 1/R in an expanding universe. Now we want to justify that the expansion preserves the thermal spectrum in the absence of particle interactions. If reactions like f f 2 are absent or negligible, the total number N of photons is conserved and the number density n = N/V behaves as n 1/R3 . From the Kirchho-Planck distribution B d = 2h 3 1 d h 2 c e kT 1 (16.9)

describing the energy of photons emitted per time and area by a body in thermal equilibrium, we obtain the number density n of photons as n(, T )d = 4/(ch)B (, T )d. Assume that at a certain time with scale factor R the distribution of photons was initially thermal. Then the number dN of photons in the volume V and the frequency interval [ : + d] is given by h 2 d 8V 4 B d = 3 dN = V . (16.10) h ch c exp kT 1 At a dierent time with scale factor R , the frequency of a single photon scales as 1/R and thus also the frequency interval d as d 1/R . Thus the expression for dN becomes dN = 8V 8V (R /R)3 h 2 (R/R )2 d(R/R ) = 3 h(R/R ) c3 c exp exp 1 kT h 2 d
h(R/R ) kT

. 1

(16.11)

The last expression agrees with Eq. (16.10) as required by photon number conservation, dN = dN , if 1 R 1 R = or T = T . (16.12) T R T R Thus a distribution of photons remains thermal, but changes its temperature as T 1/R.

CMB and its dipole anisotropy

A background of thermal photons left over from the big-bang was rst predicted by Gamov and others. In 1964, the cosmic microwave background was detected and subsequently its isotropy and deviations from a thermal spectrum were searched for. Apart from a dipole anisotropy induced by the relative motion of the Sun with 370 km/h relative to the CMB, the temperature dierences are T /T 105 in dierent directions of the sky. In each direction, one measures a perfect blackbody spectrum.

Causality and the isotropy of the CMB

The CMB is to a high degree isotropic, i.e. two points on the sky have the same temperature. Let us estimate the maximal angular separation of a region that was in causally connected at the time of recombination. Two points with angular distance on the surface of last scattering are separated today by the linear distance l = dls , where dls = c(t0 tls ) ct0

136

16.5 Ination
(a) Curvature (b) Dark Energy

100

T (K)

tot
0.2 0.4 0.6 0.8 1.0

0.2 0.4 0.6 0.8

100

(d) Matter

T (K)

bh2
0.02 0.04 0.06

mh2
0.1 0.2 0.3 0.4 0.5

100

1000

100

1000

Figure 16.4: The inuence of several cosmological parameters on the angular power spectrum of the CMB. is the distance a photon travelled freely after its last scattering at tls . Thus the maximal angular separation of a causally connected points is with l = ctls (1 + zls ) = (1 + zls )tls 0.02 1 t0 (16.13)

The reason for this causality problem is that the universe expands slower than light travels: As the age of the universe increases, the part observable to us increases linearly, ct, while the scale factor increases only with t2/3 or (t1/2 ). Thus we see more and more regions that were never in causal contact for a radiation or matter-dominated universe. The sound horizon has approximately the same angular size, because of vs c/ 3. The exact size depends among other on the cosmological model: The sound horizon serves as a ruler at xed redshift zls to measure the geometry of space-time. Moreover, the uid of photons and nucleons performs acoustic oscillations with its fundamental frequency connected to the sound horizon plus higher harmonics. The relative size of peaks and locations gives information about cosmological parameters. Figure 16.4 shows the inuence of several cosmological parameters on the angular power spectrum as function of /. The rst panel shows that the rst peak sits indeed at 100 (or 1 ) for a at Universe, as we have found in our simple estimate (16.13). Observations by the WMAP satellite conrm with high signicance the value for b from BBN, for from type Ia supernovae, and that we live in a at Universe.

16.5 Ination
Shortcomings of the standard big-bang model Causality or horizon problem: why are even causally disconnected regions of the universe homogeneous, as we discussed for CMB?

137

16 Early universe The horizon grows like t, but the scale factor in radiation or matter dominated epoch only as t2/3 or t1/2 , respectively. Thus for any scale l contained today completely inside the horizon, there exists a time t < t0 where it crossed the horizon. A solution to the horizon problem requires that R grows faster than the horizon t. Since R t2/[3(1+w] , we need w < 1/3 or (q < 0, accelerated expansion of the universe). Flatness problem: the curvature term in the Friedmann equation is k/R2 . Thus this term decreases slower than matter ( 1/R3 ) or radiation (1/R4 ), but faster than vacuum energy. Let us rewrite the Friedmann equation as k = H2 R2 8G 0 + 1 3H 2 3H 2 (1 + z)2 , = H 2 (m+rad + rad + 1) = H 2 (tot 1) (1 + z)3 (1 + z)4 .

(16.14) The LHS scales ar the RHS for MD as and for RD as General relativity is supposed to be valid until the energy scale MPl . Most of time was RD, so we can estimate 1 + zPl = (t0 /tPl )1/2 1030 (tPl 1043 s). Thus if today |tot 1| < 1%, then the deviation had too be extremely small at tPl , |tot 1| < 102 /(1+zPl )2 1062 ! Taking the time-derivative of |tot 1| = gives |k| k = 2 2 2 H R R (16.15)

d 2|k|R d |k| = <0 |tot 1| = dt dt R2 R3

(16.16)

for R > 0. Thus tot 1 increases if the universe decelerates, i.e. R decreases (radia tion/matter dominates), and decreases if the universe accelerates , i.e. R increases (or vacuum energy dominates). The standard big-bang model contains no source for the initial uctuations required for structure formation. Solution by ination Ination is a modication of the standard big-bang model where a phase of accelerated expansion in the very early universe is introduced. For the expansion a eld called inaton with E.o.S w < 1/3 is responsible. We discuss briey how the ination solves the short-comings of standard big-bang model for the special case w = 1: Horizon problem: In contrast to the radiation or matter-dominated phase, the scale factor grows during ination faster than the horizon scale, R(t2 )/R(t1 ) = exp[(t2 t1 )H] t2 /t1 . Thus one can blow-up a small, at time t1 causally connected region, to superhorizon scales. Flatness problem: During ination R = R exp(Ht) and thus tot 1 = k = k exp(2Ht) R2 (16.17)

Thus tot 1 exponentially drives towards zero. Ination blows-up quantum uctuation to astronomical scales, generating initial ucn tuation without scale, P0 (k) = ks with ns 1, as required by observations.

138

A Some formulae
A.1 Mathematical formulae
The following integrals frequently appear in the context of calculations involving particle reactions in thermal media, where refers to the Riemann zeta function. Table A.1: Thermal integrals. MaxwellBoltzmann
0 xn dx ex

FermiDirac
0 3 2 3 7 4 120 xn dx ex +1

BoseEinstein
0 xn dx ex 1

n=2 n=3

2 6

1.8031 5.6822

23 2.40411
4 15

6.4939

A.2 Some formulae from cosmology

Redshift-time relations dz = (1 + z)H dt t= radiation-dominated universe: t= 2 1/2 (1 + z)3/2 0 3H0 1 1/2 (1 + z)2 0 2H0

matter-dominated universe:

139

B Units and useful constants

B.1 SI versus cgs units
Doing only mechanics, the dierence between the SI and the cgs system is trivial: The rst one uses kg, m and s as basic units, while the latter is based on g, cm and s. Thus derived units like energy, power, etc. dier just by powers of ten. As an example compare the energy units Joule and erg in the two systems: J = 1 Nm = 1 kgm2 /s2 = 103 g104 cm2 /s = 107 erg. Electromagnetism adds as a fundamental property of a particle its charge e. However, one can choose if one expresses the charge through the mechanical units or adds a fourth independent basic unit. This freedom can be seen e.g. in the Coulomb law between two charges, e2 (B.1) F =k 2, r where clearly only the product ke2 is measurable. In the Gaussian cgs system one sets k = 1 and uses Eq. (B.1) as denition of the electric charge: Two charges with the electrostatic charge unit esu equal to one at the distance r = 1cm repel each other with the force 1 dyn. In the SI system, one denes the current via the force between two parallel, innitely long conductors: They attract/repel each other being at the distance 1 m with the force 2 107 N, if the current 1 Ampere ows in each of them. The charge 1 C follows as Ampere times second. The value of k has to be xed experimentally and is normally expressed through 0 = 1/(4k0 ) 8.85 1012 C 2 /(N m2 ). Hence, the Coulomb law in SI units is F = The two charge units are connected by 1 C = 2.998 109 e.s.u. (B.3) 1 e2 . 40 r2 (B.2)

As a last complication, there exist two variants of the cgs system, the Gau and the LorentzHeaviside systems. They dier by the appearance of the factor 4 in Maxwells equations (Gau) or of k = 1/4 in the Coulomb law. Finally, note that it is sucient to know the numerical value of Sommerfelds nestructure constant, 1/137. In Gau cgs units, = e2 , in Lorentz-Heaviside cgs units (used in this notes) = e2 /(4) and in SI units = e2 /(40 c).

B.2 Natural units

Special relativity, the wave-particle dualism of quantum mechanics, and statistical mechanics relate many units. For instance, a particle with mass m = 1 g has the rest energy E = mc2 = (3 1010 cm/s)2 1g = 9 1020 erg. The associated temperature is E = kT , the length scales

140

B.3 Physical constants and measurements is the Compton wave length C = /mc, time C /c, etc. In any calculations, its possible to set c = = k = 1 and in the end either to the right powers of this constant by dimensional analysis, or or to use directly the values derived above.

B.3 Physical constants and measurements

Gravitational constant Plancks constant velocity of light Boltzmann constant electron mass proton mass Fine-structure constant Fermis constant Stefan-Boltzmann constant Radiation constant Rydberg constant Thomson cross-section G = 6.674 1011 m3 kg1 s2 = 6.674 108 cm3 g1 s2 = h/(2) = 1.055 1027 erg s c = 2.998 1010 cm/s k = 1.381 1023 J/K = 1.38 1016 erg/K me = 9.109 1028 g = 0.5110 MeV/c2 mp = 1.673 1024 g = 938.3 MeV/c2 = e2 /( c) 1/137.0 GF /( c)3 = 1.166 105 GeV2 = (2 5 k 4 )/(15c2 h3 ) 5.670 105 erg s1 cm2 K4 a = 4/c 7.566 1015 erg cm3 K4 R = 1.10 105 cm1 2 T = 8em /(3m2 ) = 6.652 1025 cm2 e

B.4 Astronomical constants and measurements

Astronomical Unit Parsec Tropical year Solar radius Solar mass Solar luminosity Solar apparent visual magnitude Earth equatorial radius Earth mass Age of the universe present Hubble parameter present CMB temperature present baryon density dark matter abundance AU = 1.496 1013 cm pc = 3.086 1018 cm = 3.261 ly yr = 31 556 925.2 s 107 s R = 6.960 1010 cm M = 1.998 1033 g L = 3.84 1033 erg/s m = 26.76 R = 6.378 105 cm M = 5.972 1027 g t0 = (13.7 0.2) Gyr H0 = 73 km/(s Mpc) = 100 h km/(s Mpc) T = 2.725 K nb = (2.5 0.1) 107 cm3 b = b /cr = 0.0223/h2 0.0425 DM = m b = 0.105/h2 0.20

B.5 Other useful quantities

cross section ux conversion 1 mbarn = 1027 cm2 L = 3.02 1028 W 100.4M

141

B Units and useful constants

B.6 Abbreviations:
B E F J L m M p P T U Kircho-Planck function energy of a single object energy ux E/(At), the brightness of a star observed on Earth angular momentum J = mrv luminosity, emitted energy per time mass (of a single object), apparent magnitude mass of a system of objects, absolute magnitude momentum pressure cross section temperature energy of a system of objects

B.7 Properties of main-sequence stars

The spectral class, absolute visual magnitude MV color index B-V, eective surface temperature radius T , lifetime on the main sequence, and the fraction of the spectral class out of all stars is given in the following table: SK O5 B0 B5 A0 A5 F0 F5 G0 G5 K0 K5 M0 M5 MV -6 -3.7 -0.9 0.7 2.0 2.8 3.8 4.6 5.2 6.0 7.4 8.9 12.0 B-V -0.45 -0.31 -0.17 0.0 0.16 0.30 0.45 0.57 0.70 ? 1.11 1.39 1.61 T /K 35000 21000 13500 9700 8100 7200 6500 6000 5400 4700 4000 3300 2600 lifetime/yr < 106 3 106 4 108 4 109 1 1010 6 1010 > 1011 fraction 105 103 0.01 0.02 7% 15% 75%

142

Index
CDM model, 126 aberration, chromatic, 26 aberration, relativistic, 33 Balmer formula, 19 beaming, 33, 108 Big Bang Nucleosynthesis, 130 binaries, spectroscopic, 38 binaries, visual, 38 black hole, 78 supermassive, 99, 105, 109 blackbody radiation, 10 Boltzmann distribution, 20 Chandrasekhar limit, 70 Chandrasekhar mass, 71 CNO-cycle, 61 color index, 10, 22 cosmological constant, 120 cross section, 42 dark matter, 102, 132 distance modulus, 17 dynamical friction, 104 Eddington luminosity, 52 Eddington model, 53 Ellipse, 35 force, central, 35 forces, fundamental, 59 frequency, angular, 105 frequency, cyclotron, 105 Friedmann equation, 120 galaxies active, 105 evolution, 104 radio, 107 Seyfert, 108 Gamov criterion, 130 Gamov peak, 61 gravitational radiation, 80 Hawking radiation, 82 helioseismology, 63 Hertzsprung-Russel diagram, 21, 39, 96 Hubble parameter, 113 Hubbles law, 113 ination, 135 Jeans mass, 87, 133 Keplers laws, 34 Kirchhos laws, 20 Kirchho-Planck distribution, 11 Lema tres redshift formula, 117 light gathering power, 25 MACHO, 102, 132 magnication, 24 magnitude, absolute, 16 magnitude, apparent, 9 main sequence tting, 96 main sequence stars, 22 mass function, 39 mass-luminosity relation, 39 Maxwell-Boltzmann distribution, 48 MOND, 102 neutrino dark matter, 103 oscillations, 65 opacity, 42 optical depth, 42 parallax, 15 polytrope, 49, 54

143

Index pp-chains, 61 pressure integral, 47 principle of equivalence, 76 radiation pressure, 48 random walk, 43 Rayleigh-Jeans law, 12 reduced mass, 36 relaxation time, 93 resolution, 24 Schwarzschild metric, 77 Schwarzschild radius, 78 seeing, atmospheric, 26 spectral classes, 21 stability of stars, 49 Stefan-Boltzmann law, 13 superluminal motion, 107 supernova, 71 synchrotron radiation, 105 virial mass, 95 virial theorem, 48 Wiens displacement law, 12 Wiens law, 12

144

Gouy Phase
No ratings yet
Gouy Phase
11 pages
Basic Space Plasma Physics - Revised Edition
No ratings yet
Basic Space Plasma Physics - Revised Edition
17 pages
Notes
No ratings yet
Notes
146 pages
As Trophy Sic Lesson
100% (1)
As Trophy Sic Lesson
157 pages
Quantum Field Theory
No ratings yet
Quantum Field Theory
104 pages
Intro Modern Physics
No ratings yet
Intro Modern Physics
157 pages
Hierarchy Problem in Phi4 Theory PDF
No ratings yet
Hierarchy Problem in Phi4 Theory PDF
107 pages
Landmark Experiments in Twentieth-Century Physics
From Everand
Landmark Experiments in Twentieth-Century Physics
George L. Trigg
3/5 (1)
S. K. Soni, Simmi Singh - Supersymmetry - Basics and Concepts-Narosa (2001) PDF
No ratings yet
S. K. Soni, Simmi Singh - Supersymmetry - Basics and Concepts-Narosa (2001) PDF
182 pages
How Euler Did It 59 PDEs of Fluids
No ratings yet
How Euler Did It 59 PDEs of Fluids
7 pages
Varying Constants, Gravitation and Cosmology
100% (1)
Varying Constants, Gravitation and Cosmology
155 pages
G. Barton-Elements of Green S Functions and Propagation Potentials Diffusion and Waves (Oxford Science Publications) - Oxford University Press USA (19
No ratings yet
G. Barton-Elements of Green S Functions and Propagation Potentials Diffusion and Waves (Oxford Science Publications) - Oxford University Press USA (19
479 pages
Quasar Astronomy - Weedman, D.W. - CUP - 1988 - IsBN 9780521303187
No ratings yet
Quasar Astronomy - Weedman, D.W. - CUP - 1988 - IsBN 9780521303187
231 pages
Science Illustrated Australia - March 23, 2019 PDF
No ratings yet
Science Illustrated Australia - March 23, 2019 PDF
84 pages
Solutions Exercises
No ratings yet
Solutions Exercises
109 pages
The Earth's Atmosphere - K. Saha PDF
No ratings yet
The Earth's Atmosphere - K. Saha PDF
374 pages
Red Shift Lesson
No ratings yet
Red Shift Lesson
27 pages
Intro Cosmo Ryden 1
0% (1)
Intro Cosmo Ryden 1
18 pages
Hubble Focus Galaxies 2
No ratings yet
Hubble Focus Galaxies 2
45 pages
Black Holes and
No ratings yet
Black Holes and
26 pages
Análisis Complejo (Notas)
No ratings yet
Análisis Complejo (Notas)
137 pages
Radiative
No ratings yet
Radiative
150 pages
Course of Theoretical Physics Statistical Physics Third Edition Part 1 of Volume 5 - Pergamon Press (1980)
100% (2)
Course of Theoretical Physics Statistical Physics Third Edition Part 1 of Volume 5 - Pergamon Press (1980)
562 pages
Instant Download (eBook PDF) Explanatory Supplement to the Astronomical Almanac 3rd Edition PDF All Chapters
100% (4)
Instant Download (eBook PDF) Explanatory Supplement to the Astronomical Almanac 3rd Edition PDF All Chapters
56 pages
Richard Feynman
No ratings yet
Richard Feynman
61 pages
Physics of The Solar Corona
No ratings yet
Physics of The Solar Corona
888 pages
Stat Thermal Phys Python
No ratings yet
Stat Thermal Phys Python
435 pages
Astro ph0606663 PDF
No ratings yet
Astro ph0606663 PDF
364 pages
Thompson. Elliptical Integrals.
No ratings yet
Thompson. Elliptical Integrals.
21 pages
(Eric M. Schlegel) Restless Universe Understandin PDF
No ratings yet
(Eric M. Schlegel) Restless Universe Understandin PDF
229 pages
From The Atomic Bomb To The Landau Institute - Autobiography. Top Non-Secret - I. Khalatnikov (Springer, 2012) WW
No ratings yet
From The Atomic Bomb To The Landau Institute - Autobiography. Top Non-Secret - I. Khalatnikov (Springer, 2012) WW
226 pages
XI Physics Numericals
No ratings yet
XI Physics Numericals
5 pages
The Structure of Turbulent Shea - A. A. R. Townsend
No ratings yet
The Structure of Turbulent Shea - A. A. R. Townsend
416 pages
Landau's Talk
No ratings yet
Landau's Talk
8 pages
Edwin
No ratings yet
Edwin
240 pages
Fourier Analysis On Polytopes
No ratings yet
Fourier Analysis On Polytopes
250 pages
Electromagnetic Fields and Waves - Lorrain and Corson PDF
No ratings yet
Electromagnetic Fields and Waves - Lorrain and Corson PDF
383 pages
Vacancy For Temp Demo
0% (1)
Vacancy For Temp Demo
784 pages
Pendulum With Moving Support
No ratings yet
Pendulum With Moving Support
6 pages
Handbook 19
No ratings yet
Handbook 19
101 pages
(Jeremy Bernstein) Kinetic Theory in The Expanding
No ratings yet
(Jeremy Bernstein) Kinetic Theory in The Expanding
157 pages
Lecture Notes: Condensed Matter Theory I (TKM1) : J. Schmalian
No ratings yet
Lecture Notes: Condensed Matter Theory I (TKM1) : J. Schmalian
70 pages
Untitled
No ratings yet
Untitled
288 pages
Voltage Reference
100% (1)
Voltage Reference
22 pages
Understanding Stellar Evolution - Henny J.G.L.M. Lamers
100% (2)
Understanding Stellar Evolution - Henny J.G.L.M. Lamers
209 pages
Motionmountain Volume5
No ratings yet
Motionmountain Volume5
354 pages
Cohen Tannoudji C Diu B Laloe F Quantum Mechanics 1996
No ratings yet
Cohen Tannoudji C Diu B Laloe F Quantum Mechanics 1996
887 pages
2.orbital Aspects of Satellite Communications
No ratings yet
2.orbital Aspects of Satellite Communications
65 pages
Kip Thorne Journal Gravitational Collapse Death of Star
No ratings yet
Kip Thorne Journal Gravitational Collapse Death of Star
9 pages
Libro Trench PDF
No ratings yet
Libro Trench PDF
806 pages
Guglinski Lecture 16
No ratings yet
Guglinski Lecture 16
24 pages
STELLAR STRUCTURE - O.L. Pols PDF
100% (1)
STELLAR STRUCTURE - O.L. Pols PDF
207 pages
Waves & Thermodynamics
75% (4)
Waves & Thermodynamics
71 pages
Instead of the ITER project and the TOKAMAK principle: – a new type of fusion machine
From Everand
Instead of the ITER project and the TOKAMAK principle: – a new type of fusion machine
Åke Hedberg
5/5 (1)
Light Waves
From Everand
Light Waves
Tara Haelle
No ratings yet
Beta Decay for Pedestrians
From Everand
Beta Decay for Pedestrians
Harry J. Lipkin
No ratings yet
Derivation of Newton's Law of Gravitation
From Everand
Derivation of Newton's Law of Gravitation
Balungi Francis
No ratings yet
Skript Astro
No ratings yet
Skript Astro
144 pages
Fundamentals Astrophysics Cosmology
No ratings yet
Fundamentals Astrophysics Cosmology
154 pages
Astrophysics Lecture Notes 2019 Crossfield
No ratings yet
Astrophysics Lecture Notes 2019 Crossfield
252 pages