NEUTRON DATA BOOKLET

AFFILIATIONS
H. Brner, J. Brown, C. J. Carlile, R. Cubitt, R. Currat, A. J. Dianoux, B.
Farago, A.W. Hewat, J. Kulda, E. Lelivre-Berna, G.J. McIntyre, S. A. Mason,
R. P. May, A. Oed, J. R. Stewart, F. Tasset and J. Tribolet
- Institut Laue-Langevin, France
I. Anderson - Spallation Neutron Source, Oak Ridge, USA
D. Dubbers - University of Heidelberg, Germany
R. S. Eccleston, M. Johnson and C. C. Wilson - ISIS, Rutherford
Appleton Laboratory, Unitied Kingdom
G. Lander - Institut fr Transuranium Elements, Karlsruhe, Germany
H. Rauch - Atominstitut der sterreichischen Universitten, Austria
R. B. Von Dreele - Los Alamos National Laboratory, USA
W. Waschkowski - Technische Universitt Mnchen, Germany
Secretary: A. Mader
Institut Laue-Langevin
BP 156
38042 Grenoble Cedex 9 FRANCE

NEUTRONS

FOR SCIENCE

E-mail: [email protected]

Web: www.ill.fr

Second Edition, July 2003

Copyright 2003 Institut Laue-Langevin

Published by licence under the OCP Science imprint,
a member of the Old City Publishing Group
ISBN: 0-9704143-7-4
To order additional copies:
Old City Publishing, 628 North Second Street, Philadelphia, PA 19123 USA
Phone: + 1 215 925 4390 Fax: +1 215 925 4371 Web: oldcitypublishing.com
e-mail: [email protected]

CONTENTS
Preface for Neutron Data Booklet
1. Neutron Properties
1.1. Neutron Scattering Lengths
A. Rauch and W. Waschkowski
1.2. The Neutron and its Fundamental Interactions
D. Dubbers
2. Neutron Scattering
2.1. Small-Angle Scattering
R. P. May
2.2. Reflectometry
R. Cubitt
2.3. Time-of-Flight Neutron Diffraction
C. C. Wilson
2.4. Polarised Neutron Scattering
E. Lelivre-Berna, J. R. Stewart and F. Tasset
2.5. Magnetic Form Factors
J. Brown
2.6. Time-of-Flight Inelastic Neutron Scattering
R. S. Eccleston
2.7. Three-Axis Spectroscopy
R. Currat and J. Kulda
2.8. The Basics of Neutron Spin-Echo
B. Farago
2.9. Neutron Diffraction on Reactors
A. W. Hewat and G. J. McIntyre

3. Techniques
3.1. The Production of Neutrons
C. J. Carlile
3.2. Neutron Optics
I. Anderson
3.3. Detectors for Thermal Neutrons
A. Oed
4. Activation and Shielding
4.1. Activation Table of the Elements
M. Johnson, S. A. Mason and R. B. Von Dreele
4.2. Shielding of Radiations
H. Brner and J. Tribolet
5. Miscellaneous
5.1. Radioactivity and Radiation Protection
5.2. Physical Constants
5.3. Properties of the Elements
5.4. Periodic Table of the Elements

NEUTRON DATA BOOKLET,

SECOND EDITION
o make freely available a small pocketbook that covers the wide
field of neutron scattering, is clearly an idea which was timely. The
first edition of 5000 copies was absorbed by the growing neutron
community worldwide within a few months of being published.

The obvious need for a second edition has allowed certain corrections
to be made to the first edition inevitable given the speed with which it
was written and printed and to rectify some omissions notably a
new chapter on diffraction methods on continuous sources by Alan
Hewat and Garry McIntyre.
I want to thank Christian Vettier, ILLs Science Director, who is at the
origin of this project and the two editors Gerry Lander from Kalsruhe
and Jos Dianoux from Grenoble who have energetically pursued and
implemented the idea, and of course the authors who have once again
responded admirably to a tight deadline.
Colin Carlile
ILL Director
16th March 2003

PREFACE

FOR

NEUTRON DATA BOOKLET

elcome to the Neutron Data Booklet. The success of the X-ray

and Nuclear Physics Booklets, and the ever-increasing number
of neutron users, has led the ILL in collaboration with Old City
Publishing, to compile this "little book of facts."
We are first grateful to Christian Vettier of the ILL who persuaded us to
undertake this task and helped in many ways in getting people to
cooperate. We thank all those who contributed most sincerely; we
realise that this is not a research document, and therefore lacking in real
excitement. On the other hand, we hope they (and you the reader) will
find it above all "useful" and get to feel that having one in your pocket
is part of the dress code for a practicing neutron scatterer. Our thanks to
the secretarial help at the ILL, who are listed. They gainfully struggled
over tables to format and kept their cool.
Although this document was produced at the ILL, we recognise the
importance of spallation neutrons, and we hope you will find all you
need about neutrons in general in the pages. The editors would like to
be informed of both errors and omissions so that these may be
corrected in future editions and for when the information is loaded onto
the web. In particular, suggestions for further tables or chapters will be
warmly received, especially if they are accompanied by a "volunteer
author"!
Finally, thanks to Ian Mellanby and Guy Griffiths of Old City
Publishing for following through with this and making such a
presentable final product.

Albert-Jos Dianoux, ILL, Grenoble, France

Gerry Lander, ITU, Karlsruhe, Germany

1.1-1

NEUTRON PROPERTIES
Neutron Scattering Lengths
H. Rauch and W. Waschkowski
1. Introduction
Free neutrons are ideal tools for a lot of experiments, for instance for the
investigation of the atomic and molecular structure and of the dynamics of
condensed matter. Fission and spallation processes are used for an effective
production of free neutrons. These neutrons can be extracted from the moderator
block and are guided to the experimental devices outside the reactor shielding. For
most experiments monochromatic neutrons are needed, which have to be filtered
out from the thermal spectrum leaving the moderator. When pulsed sources or
mechanical choppers are used, the time-of-flight technique can be used to measure
the energy and the energy change of neutrons. Absorption, transmission and
scattering of neutrons are also used for non-destructive inspection of materials.
The scattering length of the neutron-nucleus system is the basic quantity
which describes the strength and character of the interaction of low-energy
neutrons with the individual nuclei and atomic structures. The values of scattering
lengths vary irregularly from one nucleus to another due to their strong dependence
on the details of the individual nuclear interaction. Therefore, low-energy neutrons
are an important tool for the investigation of the static and dynamic properties of
condensed matter since they distinguish between various elements and isotopes.
They can also be used for a detailed study of the interaction of the neutron as an
elementary particle with its surroundings. The related scattering lengths are of
fundamental interest for structure and dynamics investigations of condensed
matter, in nuclear research, for biological systems, and for other disciplines. In
most cases accurate and reliable values of scattering lengths for chemical elements
and also for separated isotopes are needed as input data for the interpretation of
experiments with neutrons.
In the past various tables of scattering lengths have been presented in the
literature, sometimes with the aim to collect all experimental results including the
references [1-6]. Thermal cross section values have been added because they are

1.1-2
also relevant in scattering experiments when extinction and background effects
have to be estimated [2,3,5,7].
2. Properties of the neutron
Neutrons have distinct particle properties, which influence the experimental scattering results. They have nearly no electrical properties: no electrical charge, no
electrical dipole momentum. Neutrons mainly obey nuclear interaction. However,
their magnetic moment couples to the local magnetic field of magnetic atoms and
ions. Neutrons also exhibit weak interaction which is responsible for neutron decay.
Table 1: Neutron Properties
Mass
Spin
magnetic moment
-decay lifetime
confinement radius
quark structure

m = 1.674928(1) 10-27 kg
s = -h/2
= -9.6491783(18)10-27 JT-1
= 885.9 0.9 s
R = 0.7 fm
udd

3. Scattering lengths
Low-energy neutrons are scattered isotropically within the center-of-mass system,
which indicates that no orbital momentum is involved in the scattering process (l =
0). This fact is equivalent to the statement that the range of interaction is much
smaller than the center-of-mass wavelength, , of the neutrons. This enables, within
the Born approximation and the ordinary scattering theory, the introduction of a
point-like interaction in the form of the Fermi pseudopotential [8]
r
r
2h 2
V (r) =
a (r)
mr
,

(1)

where mr = mnmk / (mn + mk) is the reduced mass of the neutron (mn) nucleus system.
Here, a is the free scattering length, which is related to the scattering amplitude fo of the
scattered spherical wave by fo = -a. This definition is chosen to get for an infinite repulsive potential the relation a = R, where R is the radius of this scattering potential.
Rather well known is the scattering of a plane neutron wave (wave number k)
by a single fixed nucleus (spin I = 0) which results in a spherical scattered wave:
() exp(ikr)/r. Within the Born approximation the scattering amplitude does in this
case not depend on the scattering angle and one gets
(2)
f () = a .
More rigorously, for slow neutrons, s-wave scattering dominates and leads to an swave phase shift o. The scattering amplitude fo, is a complex quantity given by

1.1-3
1

fo =

(e

-2i o

-1),

Re (f o)= a

2ik o

(3)

This relation shows that the scattering length determines the leading term of how the
phase shift depends on the momentum of the interacting particle. Thus, in the low
energy limit, the relation holds a o/k.
The total scattering cross section s is given by the imaginary part of the
scattering amplitude by the optical theorem.
4
Im(fo )
k

s =

(4)

The scattering amplitude consists of a potential and a resonance part, which can be
determined from resonance parameters.
Re(fo ) = fp + fr
1
{(1- 2 ) sin 2 o 1 cos 2 o}
2k o
1
Im(fo ) =
{1 (1- 2 ) cos 2 o 1 sin 2 o}
2k o
=

(5)

By Breit-Wigner formalism the sums can be written as

1 =

nr (E E r )

k (E E )
r

+ 4r

2 =

nr r / 2

k (E E )
r

+ 4r

(6)

where the summation must be carried out over all resonances r (all means resolved and
unresolved resonances as well as bound levels at negative energies) at the resonance
energies Er. nr stands for the neutron scattering width and r for the total (scattering
and absorption) width.
For nuclei with spin I 0, the s-wave neutron interaction leads to two
compound spin states (with I + 1/2 and I 1/2 ). Then the resonance sums must be
weighted with the spin statistical factors g.
g+ =

I +1
;
2I +1

1,2 = g +

1, 2

+ g-

g- =

1, 2

I
;
2I +1

(7)

The potential part of the scattering amplitude is the spin independent potential
scattering radius R' being of theoretical interest (Fig. 1). Systematic calculations of
the neutron-nucleus scattering parameters from experimental data have been
reported by Aleksejew et al. [9].
Neutron optical phenomena occur due to the collective interaction with many
scattering centers. Therefore, the mean phase shift <o>, or the mean interaction
potential , become relevant. In this case the momentum transfer occurs to the whole
assembly of particles and therefore, the center of mass system equals the laboratory

1.1-4

Figure 1: Measured and calculated values of the potential scattering radius [2,3].
system and the bound scattering length b, where b = a (mn + mk)/mK. Its average
value denotes the coherent scattering length bc.
b c = g + b + + g b
(8)
and its variance determines the incoherent scattering length bi
2
2
b i = g +g (b + - b - )
(9)
The related cross sections are
2
2
c = 4 b c ;
i = 4 b i
(10)
and the total scattering cross becomes
t = c + i
(11)
With the kinetic neutron energy E = h2/2mn2 the refraction index n - being the
ratio of the wave number inside (K) and outside (k) of a mean potential V of a
nuclei collective with a density N - follows from Eq. (1) and can be expressed as:
n2 1 =

V
= D 2 Nbc /
E
,

(12)

which lead to a refraction index of

n=

V
E

1 2 Nbc
2

(13)
In this simple treatment absorption and magnetic effects are neglected.
When the absorption (reaction) processes are included one obtains a complex
index of refraction in the form [10,11]
n = 1 -

2N
2

2
N
b 2c - r + i r
2
4

(14)
where r = a + i includes the absorption and incoherent scattering cross section.

1.1-5
4. Methods of measurements
Scattering lengths can be deduced from total cross section measurements, but for
this evaluation a lot of corrections are necessary. Therefore, techniques for a more
direct determination of coherent scattering lengths have been developed which are
based on neutron optical phenomena. A great variety of methods for the
determination of coherent scattering lengths exists, because all collective
interaction effects are characterized by this quantity. The standard method is based
on Bragg diffraction; whereas advanced methods use classical neutron optics,
where a more direct relation between the scattering length and the measurable
quantities exist [4,11]. That means, less corrections must be performed during the
data evaluation and the accuracy of the deduced quantity becomes much higher.
The considerable efforts to determine scattering lengths with high accuracy for
investigations in fundamental questions lead to techniques of single crystal neutron
interferometer [12] and of gravity refractometry [13]. With both techniques a
reliability of the order of b/b = 10-4 can be achieved.
Advanced methods for measurements of coherent scattering lengths are total
reflection, Christiansen filter technique, and neutron interferometry. The spin
dependent scattering lengths, which are of particular interest for theoretical
interpretations and for isotopic substitution techniques, can be obtained by polarization
techniques of neutrons and nuclei. Incoherent nuclear polarization-dependent cross
sections can be determined from a combination of free scattering cross sections with
directly measured scattering lengths as well. Free potential scattering radii, which are
of fundamental interest for the theory of optical nuclear model, can be deduced from
coherent scattering lengths substracting the resonance contribution. For more
information about these techniques we refer to literature [6,11].
5. Data of coherent scattering lengths and of thermal cross sections
This new compilation has the aim to give recommended coherent scattering lengths
for elements and if available for isotopes; this will be done including recommended
error bars according to the best possible criteria such as check of consistency of the
complete nuclear data set (absorption, energy dependence of scattering cross
section, incoherence, spin depending scattering lengths). Table 2 shows the
recommended values. For the spin dependent scattering length an alternative set of
b+ and b- values exists because in most cases (b+ - b-)2 has been measured. Column
7 of Table 2 indicates whether separate (b+ - b-) values are known (+/-) or/and
whether a strong energy dependence of the scattering lengths exists (E). The table
gives the most probable set. The explanation of the symbols is as follows:

1.1-6
Column
1:
2:
3:
4:
5:
6:
7:
8:
9:
10:
11:
*

ZSymbA
p or T1/2
I
bc
b+
bc
coh
inc
scatt
abs

nuclide: charge number Z - element symbol - mass number A

abundance (in %) or half-live T1/2
nuclear spin I
bound coherent scattering length (in fm)
spin-dependent scattering length for I+1/2 (in fm)
spin-dependent scattering length for I-1/2 (in fm)
indication whether separate (b+ - b-) values are available (+/-)
and/or a strong energy dependence exists (E)
coherent cross section in barn
incoherent cross section in barn
total cross section in barn
thermal absorption cross section in barn for 0.0253 eV
stands for estimated values

All scattering cross sections follow from known scattering lengths (Eqs. (8)(11)) but in the table separately measured values are given which results in a certain
inconsistency of the values reflecting the fact of conflicting experimental results.
A summary of all measured neutron scattering lengths until 1991 is given in
[1] and the references for the values recommended in this summary can be found in
[6]. It should be mentioned that there is still a lack of information of scattering
lengths of many isotopes and especially of many spin dependent scattering lengths.
Any reader is asked to inform the authors about new or unnoticed values. Thank
you in advance.
The authors acknowledge the valuable support of M. Hainbuchner and
E. Jericha in course of data handling.
References
[1] Koester, L., Rauch, H., Seymann, E.: Atomic Data and Nucl. Data Tab. 49
(1991) 65.
[2] Mughabghab, S.F., Divadeenam, M., Holden, N.E.: Neutron Cross Sections,
Vol. 1, Part A; Academic Press, New York, London (1981).
[3] Mughabghab, S.F.: Neutron Cross Sections, Vol. 1, Part B; Academic Press,
New York, London (1984).
[4] Sears, V.F.: Meth. Exp. Phys. 23A (1986) 521.
[5] Sears, V.F.: Neutron News 3 (1992) 26.
[6] Rauch, H., Waschkowski, W.: Landolt-Boernstein 16A, Chapt. 6, Springer
Berlin (2000).

1.1-7
[7]
[8]
[9]
[10]
[11]
[12]
[13]

CINDA 2000, "Microscopic Neutron Data". IAEA, Vienna 2000, ISSN

1011-2545.
Fermi, E., Marshall, L.: Phys. Rev. 71 (1947) 666.
Aleksejev, A., Barkanova, S., Tambergs, J., Krasta, T., Waschkowski, W.,
Knopf, K.: Z. Naturforschung 53a (1998) 855.
Goldberger, M.L., Seitz, F.: Phys. Rev. 71 (1947) 294.
Sears, V.F.: Neutron Optics, Oxford University Press (1989).
Rauch, H., Werner, S.A.: Neutron Interferometry, Clarendon Press, Oxford
(2000).
Koester, L., in: Neutron Physics (Ed. G. Hoehler); Springer Tracts in Mod.
Phys. 80, Springer Berlin (1977).

1.1-8

1.1-9

1.1-10

1.1-11

1.1-12

1.1-13

1.1-14

1.1-15

1.1-16

2.1-1

NEUTRON SCATTERING

Small-Angle Scattering
R. P. May

Introduction
Small-angle scattering (SAS) deals with the deviation of electromagnetic or particle
waves by heterogeneities in matter. Here, we shall treat the case of neutrons.
SAS allows one to study objects and structures in a size range of about 10 to
1000 at low resolution only, but on the other hand, it is not necessary to dispose of
crystals. The samples can be considered as composed of elementary volumes with a
linear dimension of a few , i.e. about the dimension of the neutron wavelengths
used. The scattering then depends on the scattering length densities of the elementary
volumes, which can readily be influenced for neutrons by modifying the isotopic
composition of the sample. This technique is known as contrast variation and has
largely determined the success of small-angle neutron scattering (SANS) in the fields
of soft condensed matter and biological structures. Other important reasons for using
neutrons rather than X-rays are the natural differences in scattering length density for
certain classes of matter, the ease with which one can use bulky sample environment
and thick samples, and of course, the sensitivity of neutrons for nuclear spins.
Small-angle scattering principles
When a plane wave in z direction ( = eikz) hits a sample, spherically symmetrical
waves, = - (b/L) eikL are emitted from the elementary scatterers in the sample, i.e. the
nuclei; L is the distance of an observation point from the sample. The complex scattering
amplitude b has the dimension of a length; it is called scattering length. The spherical
waves interfere and create a pattern on the neutron detector. The scattering geometry is
sketched in Fig. 1.

2.1-2

Figure 1. Scattering geometry:

An incoming plane wave (wave vector k0) hits two elementary volumes, 1 and
, are de2. They emit spherical waves (scattering vector k1) that, at an angle 2
phased due to the different distances of the two volumes from the wave front.
The momentum transfer Q is equal to Q = k1 - k0.

When the sample is disordered, as in the majority of the applications of SANS,

the scattering pattern is isotropic, and the scattering can be expressed as a function
of the modulus of Q: Q = |Q| = (4/) sin 2r/(L). is the wavelength of the
neutrons, L the sample-to-detector distance, and r the distance of the spot of the
scattered beam on the detector from that of the direct beam. The sample can be oriented by its preparation procedure or by external fields, in particular by a magnetic
field, e.g. for studying flux lines, or by a shear gradient.
Scattering cross sections and absolute intensity
The number of neutrons scattered by a sample into a solid-angle element of a
detector during a time t at a momentum transfer Q can be expressed as
T

Is (Q) = o t s a Ds ( ) V (Q)
(1)
Tap
d
.
o is the neutron flux (neutrons per cm2 per second) on the sample. ot is proportional to the counts in the monitor counter used for determining the duration of

2.1-3
a scattering run. Ts/Tap is the sample transmission, i.e. the ratio of the direct beam
intensities behind and in front of the sample, respectively. a and Ds are the illuminated area and the thickness of the sample, respectively, () the detector efficiency for a given wavelength, and dV/d is the differential scattering cross section
per unit volume.
dV/d can be expressed as
dV
d

(Q )

1
V

bi b e
j
i, j

iQ ri rj

(1a)

In the case of isotropic scattering, one averages all the possible configurations of the nuclei in the sample, expressed by <>:
dV
d

(Q)

1
V

< iQ ri r j >

bi b j e
i, j

(1b)

Using an attenuated beam intensity 'o with known attenuation factor fa, we
obtain from equation 1
dV
K I s (Q )
(Q ) =
d
Ds Ts I w (Q)

(2)

for data that were divided by a background-corrected water run (see below) with
K = I w (0)

Tap
fa (, 0) 'o a t

(2a)

A broad treatment of absolute scaling techniques in SANS can be found in

Wignall and Bates (1987).
The differential scattering cross section per unit mass of the sample,
dV(Q)/d, is
d m
1 d V (Q )
K
I s (Q )
=
(Q ) =
(3)
d
cs d
cs Ds Ts I w (Q)
,
with cs, the sample concentration, in gcm-3. Finally, the differential scattering
cross section per particle, dM(Q)/d, is
d M
d (Q ) M
K M
I s (Q )
=
(Q ) = m
d
d
N a cs Ds Ts N A I w (Q)

,
when the molecular mass M is known; NA is Avogadro's number.
Molecular mass
The molecular mass M can be obtained by considering

(4)

2.1-4
2

2
d M
M

(Q = 0) = b i p Vp = p vp m p = p vp

d
NA

as

M=

d m (Q )
Na
2
d
p vp

(5)

(6)

,
with vp , the (dry) partial specific volume of the particle.
Background corrections
The background in a sample is composed of a transmission-dependent part (sample
holder, e.g. a quartz cell, and matrix scattering, e.g. incoherent scattering from a
proton-containing buffer solution) and a transmission independent, but also timedependent part (electronic and neutron noise).
Of the former, the term that is proportional to the transmission (sample container scattering) is easy to treat, whereas the incoherent term (proportional to 1-T)
is more complicated. Since the empty-cell contribution is weakened by the transmission of a water sample, the empty-cell and noise contributions are corrected by
I (Q) = M I w (Q) T I e (Q) (1 T ) I Cd (Q)
w
o
w
w

Me
M Cd
Mw

(7)

Mw, Me and MCd are the neutron monitor counts of the water sample, the container
and the noise measurements (for identical geometry, wavelength etc.), and Mo is a
reference monitor for scaling. Tw is the water transmission, i.e. the ratio of the
intensity transmitted by the water sample and that of the empty cell.
In the more general case of subtracting a purely coherent container or reference
background from a sample, we get

I (Q) = M Is (Q) Ts I b (Q) 1 Ts I Cd (Q)

s
0
Tb M b
Tb M Cd

Ms

(8)

Ts and Tb are defined according to Tw above.

Debye formula
If the scattering is produced by a dilute ensemble of particles of limited size, the
scattering curve can be approximated by the first terms of a Taylor series.
According to Debye (1915), <eiQr> = (sin Qr)/Qr for randomly oriented particles.
The scattering from N particles of volume VP can be written as a sum of contributions from elementary volumes i, j with scattering densities i and j, separated by
a distance rij:

2.1-5
dV
d

(Q ) =

() ()

ri rj

sin Qrij
Qrij

dri drj

(9)

Radius of gyration
Since (sin x)/x 1 - x2/6 + x4/120 - x6/5040 +... , one can express I(Q) near Q = 0 as
(QR )2 (QR )4

G
G
I(Q) = I(Q = 0)1
+
...
3
60

(10)

where in analogy to mechanics RG is the radius of gyration of the particle or the

weight average of the radii of gyration of a mixture of particles.
Eq. 10 is approximated again in two ways, the Zimm (1948) approximation for
1/(cs I(Q))

cs
cs
(QR G )2 ...
(10a)

1 +

I( Q )

I( Q = 0 )

and the Guinier (1955) approximation that uses the similarity of the first terms in
eq. 10 with the expression e-x 1 - x + x2/2 - ... to write

(QR G )2

(10b)
.
Expression 10b was historically very useful, because it allowed one to extract RG
by plotting I(Q) vs. Q2 on logarithmic paper.
RG can be obtained without the need of absolute calibration. In practice, equations 10a and 10b do not yield the same value of RG, because the equations are not
identical, and because the fit cannot start at Q = 0 due to the beam stop.
I( Q ) I( Q = 0 ) e

Table 1. Radii of gyration (RG) of simple triaxial bodies (Mittelbach, 1964)

body
RG
sphere (radius R)
(3/5) R2
hollow sphere (radii R1 and R2)
(3/5) (R25 - R15) / (R23 - R13)
ellipsoid (semi-axes a, b, c)
(a2 + b2 + c2) / 5
parallelepiped (edge lengths A, B, C)
(A2 + B2 + C2) / 12
elliptic cylinder (semi-axes a, b; height h)
(a2 + b2) / 4 + h2/ 12 = Rc2 + h2/ 12
hollow cylinder (radii R1 and R2; height h) (R12 + R22) / 2 + h2/ 12
Rc is the cross-sectional radius of gyration
Modelling
Frequently, scattering data are analysed by comparing the experimental data
with that from models using least-squares fitting procedures, including polydisper-

2.1-6
sity treatment and instrumental smearing. The method and a catalogue of analytical and semi-analytical expressions is described in Pedersen (1997).
Some form factors for simple bodies are listed in table 2.
Table 2. Form factors for simple bodies
Sphere of radius R

sin (QR ) QR cos (QR )

P(Q) = A S2 (Q, R ) = 3

(QR)3

Two-shell sphere

P(Q) = (f1 A1 (Q, R1 ) + f2 A 2 (Q, R 2 ))

where fi and Ai are the relative scattering weights and amplitudes of the outer
and inner shells, f1 = V1 / [1 V1 + (2 1 ) V2 ], f2 = 1 f1
cylinder/disk
sin (Q d cos ) 2 J1 (Q R sin )

Pd (Q) = 0 / 2 sin
d
Q d cos
Q R sin

where 2d is the thickness, R the disk radius, J1 the first order Bessel function, and 0- is the angle between the disk normal and Q)
Pm (Q) =

Membrane, thickness 2d

2
2
( Qd) / 2
1 cos (2Qd)e

Q2

Pair-distance distribution function

Equation (9) can be rewritten to yield
dV

(Q ) =

4 0 r 2 V(r)

sin(Qr )

(11)
.
(r) is the correlation function (Debye and Bueche, 1949). Introducing p(r), the pairdistance distribution function, p(r) = r2V(r), one gets
d

dV
d

(Q) = 4 o p(r )

dr

Qr

sin(Qr )

dr.

(11a)

(Q) Qr sin(Qr ) dr

(12)

Qr

The Fourier inversion of 11a,

p(r) =

1
2 2

dV
d

can, in general, not be solved, because the scattering function is neither known at
Q 0, due to the beam stop, nor at Q , due to the limited maximum scatter-

2.1-7
ing angle and bad statistics at high Q. Glatter (1981) has introduced an indirect
Fourier transformation, a least-square method that overcomes this problem for particles of limited size rmax.
The radius of gyration can be calculated as the second moment of p(r),
r
0 p(r ) r 2 dr
r
2 0 p(r ) dr
max

R 2G =

max

(13)

Size distributions
In the case of a dilute solution of monodisperse homogeneous particles with form
factor P(Q),
d s
(Q) = n ( b i b V)2 P(Q) = n 2b V. 2 P(Q)
(14)
d

n is the number concentration of the particles (number per unit volume), bi are the
scattering lengths of the nuclei of the particle, b is the scattering-length density of
the bulk solvent, and V the volume which circumscribes the particle itself and its
associated perturbed solvent (or matrix).
For a size distribution of homogeneous particles that differ only in their dimension, not in their shape, we can write
ds

(Q) = 2b o Dn(R ) V2(R ) P(Q, R ) dR

(15)
,
where Dn(R) is the number distribution of particles, i. e. the number of particles of
size R per unit volume.
One can also define a volume distribution Dv(R) R3Dn(R), i.e. the sum of
volumes occupied by particles of dimension R per sample unit volume, and an
intensity distribution Di(R), i.e. the sum of intensity contributions per particle of
dimension R per sample unit volume.
d

Porods laws
Porod (1951) has shown that the total small-angle scattering from a sample, irrespective of the way its density is distributed, is a constant, called Porod's invariant C:

C = o I(Q) Q2dQ = 2 2V2 .

(16)

In practice it is difficult to measure the data in all the Q range necessary to evaluate Porod's invariant, but in certain cases it is possible to approximate the low-Q
(Guinier) and high-Q limits.
High-Q limit
Also to Porod, we owe an asymptotic law which is valid for high values of Q (Q >>

2.1-8
1/D), if there are sharp boundaries between the phases of the system. This is a condition that barely holds. The specific surface Os = S/V, with S, the particle (or
structure) surface, and V, the particle (or structure) volume, is given by
S
=
V

[I(Q) Q4 ]Q

(17)
.
Due to the usually very high incoherent scattering of neutrons at high Q, it is often
very difficult, if not impossible, to determine Os with sufficient precision. On the
other hand, plotting IQ4 versus Q4 ("Porod plot"), which allows one to calculate
Os from the Q4 = 0 intercept, can help us to determine a residual background,
which shows up as a constant slope for large Q.
Os =

References
Debye, P. (1915) Ann. Phys. 28, 809-823.
Guinier, A., and Fournet, G. (1955) Small Angle Scattering of X-rays, New York:
Wiley Interscience.
Mittelbach, P. (1964) Acta Phys. Austriaca 19, 53-102.
Pedersen, J.S. (1997) Adv. Coll. Interface Sci. 70, 171-210.
Wignall, G., and Bates, J. (1987) J. Appl. Crystallogr. 20, 28-40.
Zimm, B.H. (1948) J. Chem. Phys. 16, 1093.

General reading
Small-Angle X-ray Scattering, Glatter, O., and Kratky, O., eds., London: Academic
Press (1982).
Neutron, X-ray and Light Scattering: Introduction to an Investigative Tool for Colloidal and Polymeric Systems, Lindner, P., and Zemb, Th., eds., Amsterdam: North
Holland (1991).
Williams, C., May, R.P., and Guinier, A. (1994) "Small-Angle Scattering of X-rays and
Neutrons" in Characterisation of Materials, E. Lifshin ed., Vol. 2B of "Materials
Science and Technology", Weinheim: VCH Verlagsgesellschaft, pp. 611-656.

2.2-1

Reflectometry
R. Cubitt
Introduction
Neutron reflectometry is a technique for the study of planar structures with a wide
variety of materials from magnetic multi-layers to biological systems at a solidliquid interface. The method not only permits the derivation of material structures
perpendicular to the plane but also to ascertain the perfection of these layers at a
boundary which maybe modified by roughness or inter-diffusion. Many problems
such as the behaviour of lipid bi-layers in biological cells or the influence of
magnetic impurities in superconductors can be greatly simplified by synthesising
layers of the material on a substrate thus reducing the dimensionality of the system.
Reflectivity from a Thick Substrate
A beam of neutrons with its associated wave properties can behave in exactly the
same way as light when impinging on a mirror like surface. Provided any deviation
of the beam is far from satisfying the Bragg condition of the crystaline structure,
the neutron can be considered to interact with a constant potential Vo simply related
to the coherent scattering length by the relation

Vo =

2h 2
b
mn

(1)

where mn is the mass of the neutron, is the number density of atoms in the
material and b is the average coherent scattering length for these atoms. The
product b is known as the scattering length density often denoted by Nb. Most
materials have a positive b so in a positive potential a neutron has less kinetic
energy and hence a longer wavelength (opposite to light where the wavelength
shortens). We may now consider what happens to a beam approaching a surface
with a bulk potential Vo, infinitely deep.
With no structure within the surface the only potential gradient and hence
force is perpendicular to the surface. Only the normal component of the incoming
wave vector, ki is altered by the barrier potential and it is the normal component of
the kinetic energy Ei which determines whether the neutron is totally reflected
from the barrier or not.

2.2-2

Figure 1: (a) Reflection of an incident beam from an ideally flat interface. ki

and kr are the incident and scattered wave vectors, with angles i = 0 = in
the incidence plane; q is the wave vector transfer; Nb1 is the scattering length
density of the semi-infinite substrate. On the right is the scattering length
density profile as a function of depth; (b) Simulation of the specular neutron
reflectivity as a function of q from a silicon/air interface (Nb1=2.07x10-6-2).

(hk sin i )2
Ei = i

2 mn

(2)

If Ei<V o then there is total reflection and the critical value of wave vector
transfer, qc will be when Ei=Vo giving:
qc = 16Nb
as q = 2 ki sin i
(3)
Assuming the interaction is elastic then conservation of momentum means that i.e.

2.2-3
the reflection is specular provided the sample is static. Any off specular reflection
must be a result of potential gradients within the xy plane of the surface.
If Ei>Vo then the reflection is not total the neutron can either be reflected or
transmitted into the bulk of the material. The transmitted beam, kt with its normal
component of kinetic energy reduced by the potential must change direction i.e. it
is refracted. The change in the normal wave vector is:

kt2 = ki2 4Nb

(4)

allowing us to define a relative refractive index n:

n2 =

kt2 ki2// + ( ki2 4Nb )

4N
2 Nb
=
=1 2 b =1
ki2
ki2
ki

(5)

where is the neutron wavelength. For a material such as Si, Nb is 2.07x10-6-2 ,

much less than one, so for thermal neutron wavelengths we can make a good
approximation for n with the well known result:
n 1

2 Nb
2

(6)

This confirms the earlier statement about the wavelength change in the bulk
being opposite to that of light (for positive b) as we see that n is less than one. The
transmitted beam refracts towards the mirror plane and exactly at the point of total
reflection, the refracted beam travels along the surface.
All of the above discussion (apart from treating neutrons as waves) can be
derived from classical physics. In order to describe all the physical aspects of
reflectometry we must use a quantum mechanical approach. The wavefunction
describing the probability amplitude of a neutron near to the surface is

2 z
+ k2 = 0
z z

where

k2 =

2 mn
( Ei V ) k//2
h2

(7)

Solutions for this above and below the surface are:

z = eiki z + re iki z

and

Yz = teikt z

(8)

where r and t are the probability amplitudes for reflection and transmission.
Continuity of the wavefunction and its derivative give the expressions:

1+ r = t

ki (1 r ) = t kt

(9)

2.2-4
which leads directly to the classical Fresnel coefficients found in optics:
r=

ki kt
ki + kt

and

t=

2 ki
ki + kt

(10)

In reflectometry we measure the reflectivity as a function of wavevector transfer or

q. Using the expressions (3), (4) and (10) we can relate the measured reflectivity R
to q and qc. Note that what we measure is an intensity and thus is a function of the
quantum mechanical probability squared.
2
1/ 2

q (q 2 qc2 )
R = r2 =
q + q2 q2 1 / 2
(
)
c

(11)

When q>>qc this reduces to:

R

16 2 2
Nb
q4

(12)

which is the reflectivity used in the Born approximation [1]

Returning to the wavefunction within the surface (8) and using (4) we find that
when EI<V (or ki<4Nb or q<qc) we have a real solution of the form:

Yz = te

i k i2 4 N b

1/ 2

= te

1 2
qc q 2
2

1/ 2

(13)

This is a very important result as it shows that even when the potential barrier
is higher than the particle energy normal to the surface it can still penetrate to a
1 / 2
characteristic depth of ( qc2 q 2 ) . This 'evanescent' wave travels along the surface
with wave vector k// and after a very short time is ejected out of the bulk in the
specular direction. Taking the value of Nb for Si again this penetration is of the
order of 100 at q=0 rising rapidly to infinity at q=qc. No conservation laws are
broken, as the reflectivity is still unity due to the fact that this wave represents no
flux transmitted into the bulk. The result also explains why when a thin layer of
material (<100) such as Ni, which has an Nb twice as large as Si, is put onto a Si
substrate we find that qc is still defined by the Nb of Si and not of Ni. The reason
being in this case the layer is thinner than the characteristic penetration depth and
the neutrons tunnel through the barrier.
Calculating the reflectivity from such a system or one with many layers
requires a general technique such as the optical matrix method [2,3,4,5]. The
transmission and reflection from one layer to the next can be described as a matrix
multiplication product for each layer. The problem of inverting a reflectivity curve
to extract Nb as a function of depth is complex and many profiles can produce the

2.2-5

Figure 2 (a):As Figure 1a with an additional thin layer of scattering length

density Nb2. (b) Simulation of the specular neutron reflectivity as a function
of q from a 400 layer of nickel on a silicon substrate (Nb1=2.07x10-6-2,
Nb2=9.04x10-6-2).
same reflectivity curve. A useful review of the various techniques for inversion can
be found in [5].
Absorption
In reality many materials have a finite absorption cross-section. This is dealt with
by adding an imaginary part to the coherent scattering length.
btotal=bcoherent+ibabsorption

(14)

It can be seen then from (8) that the transmitted and the reflected intensity is
exponentially reduced by the presence of absorption. Even in the regime of total

2.2-6
reflection, as the evanescent wave exists in a surface region of the material,
absorption occurs reducing the total reflection amplitude to less than one. It should
not be forgotten that even materials such as Cd and Gd that are very strong
absorbers of neutrons can still have significant reflectivities.
Magnetic Materials
We have seen that a neutron in the bulk is affected by a mean potential related to the
coherent scattering length. If the material is magnetised then there is an additional
potential associated with the interaction of the magnetic dipole moment of the neutron
and variations of magnetic flux density B. This potential has the value:
Vmag = B(r )

(15)

where is the magnetic dipole moment of the neutron and B(r) is the spatially
varying magnetic field. If the incoming beam is polarized up or down with respect
to the magnetisation of the sample then the magnetic potential switches sign. The
magnetic potential can be expressed in the form of a coherent scattering length:
bm = 1.913

e2
S
me

(16)

where S is the effective spin of the magnetic atom perpendicular to the momentum
transfer of the reflection. The total coherent scattering length becomes polarization
dependent with the plus sign corresponding to a beam polarised by a polarizer with
a magnetisation in the same direction as the magnetisation of the sample:

btotal = bnuclear bm

(17)

Assuming the sample is a saturated ferro-magnet, to extract the nuclear and

magnetic parts of b two measurements must be made, one with the polarization
parallel to the magnetisation, R+ and the other with the polarization inverted, R- by
use of a flipper [6]. A very important point in polarised neutron reflectometry
(PNR) is to remember that magnetic reflection is only going to occur at a potential
boundary and from (15), that requires a step in B. If a sample has been magnetised
within the plane of a magnetic layer then Maxwells equations tell us that the step in
B is oM at the surface, where M is the magnetisation density of the layer. If the
sample is magnetised normal to the plane then B is continuous at the boundary and
there is no potential step. Magnetic reflection can only occur in the presence of inplane magnetisation components. Components of magnetisation normal to the
plane do not reflect an in-plane polarized beam but can cause spin flip i.e. a
completely polarized beam will have a certain fraction polarized in the opposite

2.2-7
direction after reflection. This can only be measured by the use of a second flipper
and polarizer or analyser. Figure 3 shows the four combinations of the initial and
final polarization states needed to measure the four reflectivities i.e. R++, R ,
R+ , R +. It should be noted that from the polarizer to the analyser there must be a
small vertical guide field that maintains the polarization on a vertical axis. In
addition to measuring these four reflectivities the four cases should also be
measured with a non-magnetic scatterer such as graphite to enable the imperfection
of the flipper and polarizer efficiencies to be taken into account. A detailed
description of neutron optics and magnetic effects can be found in ref [6,7].

Figure 3: A schematic layout of a polarised neutron reflectivity experiment with

analysis. The initially unpolarised beam is polarised vertically by the polariser
mirror. Flipper 1, when activated, inverses the polarisation allowing the
measurement of R+ and R-. The difference of these two data sets represents the
magnetic contribution to the reflectivity. If in the process of reflection the neutron
polarisation was changed (flipped) due to a magnetisation component out of the
vertical plane this can be detected by the use of an analyser mirror and second
flipper. Between the polariser and analyser a guide field (~20G) is required to
maintain the polarisation in the vertical direction.

Roughness and Interdiffusion

There is a large field of interest in what happens between materials at an intermaterial boundary. An interface may be rough with peaks and troughs over a large
range of length scales with a fractal-like structure. It may also have magnetic
roughness if the magnetisation does not sharply change at an interface. A boundary
may be smooth but with one material diffused into the other. It turns out that in both
the rough and diffuse case the specular reflectivity is reduced by a factor very much
like the Debye-Waller factor reduces scattered intensity from a crystal. Expression
(12) would be affected in the following manner:

2.2-8
16 2
2 2
R 4 Nb2 e q z
q

(16)

where is a characteristic length scale of the layer imperfection. So what happens to

the intensity lost by the exponential factor in (16)? In the case of the diffuse interface
the lost intensity must go into the transmitted beam as there are no potential gradients
in any other direction than normal to the surface. This is not the case for the rough
interface where intensity is lost by local reflections in directions away from the
specular direction or off-specular scattering. Information such as the height-height
correlation function can be deduced from this off-specular scattering [2,3].

Figure 4: Two possible interfaces which can result in identical specular

reflectivities. The data only differs in the case of the rough interface where offspecular scattering is observed.
Experimental Technique
The reflectivity can be measured as a function of q in two very different ways, each
of which has advantages and disadvantages. The first method known as
monochromatic involves making a -2 scan with a well defined wavelength and
storing the reflected intensity for each . The incoming intensity must also be
measured either at the same time with the use of a monitor and then scaled up by
the monitor efficiency or in the separate experiment. The reflectivity is simply the
ratio of these two intensities for each which is converted to q by Braggs law:
q = 4 sin( ) /

(17)

With a sample length of the order of a few cm and a starting of a fraction of a

degree, the beam must be finely collimated to ensure the sample is under
illuminated i.e. all the incoming beam strikes the reflection surface. The sample can

2.2-9
be over illuminated for very small samples but the data must be corrected for the
varying flux on the sample as is increased.

Figure 5: A typical collimation set-up within the reflection plane. When sample is
under-illuminated (minimal background), the distance between the last slit and
the sample should be as short as possible. In general the divergence out of the
plane is defined only by the source.
As the reflectivity tends to drop quite fast with increasing q (see (12) where
R1/q4) in order to gain intensity (at the price of looser resolution) the collimation slits
can be opened keeping / constant and equal to the fractional wavelength variation.
The final curve in either case must have the resultant q resolution, q, de-convolved
from the data. The q resolution is related to and by the following relation:
2

q

= +
q

(18)

This method has the advantage that the wavelength chosen can be that with the
highest flux from the neutron source and therefore for a given resolution, is the
most efficient method of using the flux available.
The alternative is to use time-of-flight (TOF). Here we keep constant and
use all the available wavelengths in the beam. The wavelength and hence q is
measured by pulsing the incoming beam and measuring the arrival time to the
detector. The resolution is the same as (18) with / replaced by T/T where T is
the pulse time width and T is the time-of -flight of the pulse. The resolution of the
time binning at the detector in principle is also a factor but in practice this is chosen
to be much smaller than T. The range of q covered for a given depends on the
useful wavelength range which is a factor of 10 for the ILL reflectometer D17.
Over this range however, the flux at the minimum and maximum wavelengths
compared to the peak flux may be more than 2 orders of magnitude smaller. The
highest q (and lowest reflectivity) is measured by the shortest wavelength. For the

2.2-10
same resolution the TOF method is less efficient than the monochromatic technique as
it will always require longer counting times to measure the same q-range to the same
statistical accuracy. The reason the ILL reflectometer has both TOF and
monochromatic options is that kinetic experiments cannot be done in the
monochromatic mode. If one had a sample which the layer structure was changing as a
function of time, only the TOF method can produce a unique R(q) for a given time
range. The -2 measurement is a sequential set of counts so each point in q is
measured at a different time. In order to solve the problem of reduced flux in TOF,
D17 has a chopper system that can continuously vary the time resolution. Provided the
reflectivity curve does not have fine structure then low a resolution measurement can
be completed in less than 1minute. The incoming beam intensity for TOF can be
measured with a monitor in which case the data has to be re-binned in time before
dividing as the monitor is not at the same distance from the pulsed source as the
detector. In addition a separate measurement of the relative efficiencies of the detector
to monitor is required as a function of wavelength. On D17 there is no monitor for this
and the incoming beam is measured directly on the detector in a separate
measurement. Care must be taken to avoid dead time losses where the local rate can be
much higher than the mean count rate observed.
About the specular reflected beam in 2 there is background that must be
subtracted before dividing through by the direct beam. This background can be
substantial for experiments involving a solid-liquid interface where the substrate can
be pure water and limits the lowest reflectivity measurable to ~10-7. With a two
dimensional multi-detector a wide range of 2 is captured which will include
background and off-specular diffraction. Instruments with a single detector require a
special measurement with the detector in an off-specular position to measure the
background. In both cases care must be taken in deciding the 2 range for background
does not include diffracted intensity from the sample itself.
PNR experiments are complicated by the fact that the polarizers and flippers in
the system are not perfect. With no sample and incoming beam directly on the analyser
we would expect no flux on the detector in the condition -+ or +- for a perfect system.
A measure of the efficiency of each flipper and the polarizers is the flipping ratio F
defined as

F1 = I+ + I +

and

F2 = I+ + I+

(19)

where F 1 and F 2 are the flipping ratios for the first and second flippers. The
intensities are measured either with the direct beam for a single detector or with a
non-magnetic scatterer for a multi-detector. The measured intensities, I correspond
to the states of the two flippers as in figure 3. Acceptable flipping ratios are of the

2.2-11
order of 40. These must be taken into account when attempting to extract the
magnetic and nuclear components of the reflectivity [6,7].
ILL instruments
At present there are three reflectometers at the ILL, two of which are externally
funded (ADAM & EVA) and therefore have limited time for ILL experiments. They
are both fixed wavelength resolution (1%) monochromatic instruments with ADAM

Figure 6: The dual mode instrument D17 at the ILL designed to take advantage
of both TOF and monochromatic methods of measuring reflectivity. (a) A side
view of the instrument in TOF mode. The monochromator is removed from the
beam and a double chopper system defines the time resolution (1-20% t/t).
Between the collimation slits is a vertically focusing guide which increases the
flux at the sample position at the price of increased vertical divergence of the
beam. The slits define the beam in the horizontal direction. (b) A vertical view of
the instrument in monochromatic mode. Here the collimation arm is rotated (~4
deg) to allow the beam reflected from the multi-layer monochromator system to
pass through to the sample. The choice of monochromators involves high and low
resolution (2-5%) plus polarising.

2.2-12
having a horizontal reflection plane and EVA vertical. EVA has been specially
designed to measure scattering from the evanescent wave mentioned above.
D17 is an ILL instrument with both monochromatic and TOF modes. The
wavelength resolution can be changed in the monochromatic mode and the time
resolution in TOF is continuously variable. All three have polarized neutrons with
polarization analysis. Unfortunately non of the three are capable of measuring
reflection from a free liquid surface which requires a special instrument which
hopefully will be built in the near future at the ILL. Further details about these
instruments can be found in the ILL web pages at http://www.ill.fr/YellowBook/.
References
[1] Born and Wolf 'Principles of Optics', Permagon Press, Oxford (1970)
[2] Phys. Rev. B 'X-ray and Neutron scattering from rough surfaces' V38 N4 P2297
[3] J. Daillant A. Gibaud (Eds.) 'X-ray and Neutron Reflectometry: Principles and
Applications', Springer (1999)
[4] V. F. Sears 'Neutron Optics', Oxford Press, Oxford (1989)
[5] X. Zhou S. Chen 'Theoretical foundation of X-ray and neutron reflectometry'
Phys. Reports V257 p223 (1995)
[6] W. Williams 'Polarized Neutrons' , Oxford Press, Oxford (1989)
[7] A. Wildes 'Polarizer-analyser correction problem in neutron polarization
analysis experiments' Rev. Sci. Inst. V70 p4241 (1999)

2.3-1

TIME-OF-FLIGHT NEUTRON
DIFFRACTION
C. C Wilson
1. Introduction
The principal characteristic of instruments for diffraction on pulsed neutron sources
is their use of time-of-flight (TOF) techniques to allow for optimal use of the white
neutron beams produced by the source. These rely on the fact that neutrons with
different energies, and hence different wavelengths, from de Broglies relationship,
travel at different velocities:
=h/mv
(1)
where m is the neutron mass and v the neutron velocity.
Thus, since in the spallation process all neutrons in a given pulse are
essentially created at the same time, t0, the higher energy, shorter wavelength
neutrons travel faster and hence arrive at sample and subsequently at detector, at an
earlier time than the lower energy, longer wavelength, slower neutrons. By
measuring the time of arrival of a neutron at the detector, and of course knowing its
flight path, we can calculate its velocity and hence its wavelength (energy). This is
the basis of TOF, which is a genuinely wavelength-sorted white beam technique of
great use in a wide range of diffraction techniques.
The wavelength and time-of-flight are related by the following expression:
= ht/mL

(2)

where t is the time-of-flight and L the total flight path.

And using Braggs law =2dsin, we get a relationship between TOF and dspacing:
t=m.L.2dsin/h
[=252.777.L.2dsin, with t in s, L in m, d in ]

(3)

We note here in passing that the commonly used unit for wavevector transfer in
neutron scattering, Q=2/d.
The measured d-spacing at a given scattering angle is therefore given by:
d=t/(252.777.L.2sin)

(4)

i.e. the measured d-spacing is directly proportional to TOF. This has important

2.3-2
consequences for the design of TOF diffraction instruments. For example, if we
consider powder diffraction, Braggs law for a monochromatic instrument is
normally expressed in the form
=2dhklsinhkl

(5)

where is the chosen (fixed) wavelength and measurement of various Bragg

reflections hkl is achieved (at d-spacing dhkl) by scanning in angle (hkl). However,
in TOF techniques we have a range of wavelengths hkl and we can rewrite the
Bragg equation as
hkl=2dhklsin

(6)

i.e. we can measure a whole range of d-spacings (Bragg reflections) at a fixed

scattering angle 2 by scanning in wavelength (TOF). This is the basic principle
of TOF diffraction as a fixed-angle, wavelength-dispersive technique (Figure 1).

Figure 1: The basic principle of TOF diffraction. The time- (wavelength-) dispersive white beam is incident on the sample, and the diffraction pattern record. The total flight path, L=L1 (source-sample dised at a fixed scattering angle 2
tance)+L2 (sample-detector distance). In this fixed geometry, the d-spacing dhkl is
directly proportional to the time-of-flight (Eq. 4) and the diffraction pattern is
recorded as shown.

2.3-3
It is also, of course, possible to measure such TOF diffraction patterns
simultaneously at many angles, in multi-detectors or in multiple detector banks.
This can either lead to higher count-rates when the patterns measured in different
detectors are summed and focused (usually in the data processing software), or
can lead to additional information being available in the diffraction pattern
measured in the different detector banks. This is a trend very much in evidence on
modern TOF diffractometers (Figure 2).

Figure 2: An example of a modern multi-detector bank TOF diffractometer, the

GEM instrument at ISIS. The detectors are arranged in a series of banks around
the sample tank, from low angles (lower left) to backscattering (upper right).
GEM is used in the study of both crystalline (powder diffraction) and disordered
materials (liquids and amorphous scattering).

2. Resolution
Among the important characteristics of a TOF diffraction instrument is that the
resolution (d/d or Q/Q) of a diffraction pattern measured at a given scattering
angle (or, for example, in a given focused detector bank)
(i)

is constant over the whole diffraction pattern (the angular term cot in
the resolution being constant) and

2.3-4
(ii) can be improved linearly with the length of the instrument (by minimising
the term L/L, where L is a constant term reflecting the finite width of the
moderator, and L is the length of the instrument).
As an example, the TOF powder diffractometer HRPD at the ISIS source has an
overall flight path of around 100m, and an effective moderator width of around 3cm
(the moderator is 5cm thick but has a poisoning layer which reduces this). Thus the
L/L term is of order 3 104. With the main detector bank in the backscattering
region (2 close to 180, close to 90), the angular term in the resolution, linear in
cot, is minimised and the overall resolution optimised. In the hi ghest resolution case,
HRPD offers d/d of around 4 104. Of particular note is that this very high d/d
resolution is maintained over the whole diffraction pattern (Figure 3).

Figure 3: Powder diffraction pattern of Al2O3 measured on the high resolution

diffractometer HRPD at ISIS. The excellent d/d resolution is essentially constant over the whole pattern.

3. Q-Range
Another important aspect of a diffraction experiment is the Q-range (d-spacing
range) accessed in a single measurement. As can simply be seen, to get to low
values of d (high values of Q), one should measure at short TOF, i.e. with short

2.3-5
wavelength, high energy neutrons. Spallation sources are rich in such neutrons and
so are well suited to short d/high Q measurements. For maximum Q-range it is also
important to measure at a range of angles (in addition to measuring over a wide
wavelength range), as lower angles give generally lower Q data, while higher
angles give access to higher Q.
4. Powder Diffraction
The principles of TOF powder diffraction have been discussed above. High
resolution diffractometers can allow the measurements of d-spacings of as low as
0.25 and refinement of structures of up to around 100 atoms.
5. Single Crystal Diffraction
Single crystal diffraction on a TOF instrument is by far most efficiently carried out
using area detectors. White beam single crystal diffraction with an area detector (and a
stationary crystal) is termed Laue diffraction, while adding wavelength discrimination

2min
Incident

2max

2min

Beam

1/min

2max

1/max

Figure 4: The geometry of TOF Laue diffraction. The two Ewald spheres are
defined by the minimum and maximum wavelengths used they have radii
min and 1/
max. The 2
min and 2
max values are defined by the angles sub1/
tended by the area detector. In this two-dimensional projection, all areas of
reciprocal space between the Ewald sphere surfaces and the dotted lines are
observed in a single measurement with a stationary crystal and detector; the
reciprocal lattice points coloured in black (along with the spaces in between).

2.3-6
using TOF techniques leads to the TOF Laue diffraction used on such instruments. The
particular feature of this technique is that it allows access to large volumes of
reciprocal space simultaneously in a single measurement (Figure 4). These reciprocal
space volumes, moreover, are completely resolved in three-dimensions.
This means that a single measurement accesses many Bragg reflections but
also offers a wide survey of reciprocal space, ideally suited to the study of other
features such as diffuse scattering which do not necessarily occur at the Bragg
reflection positions. Routine determination of organic and other structures of 100 or
more independent atoms is possible, with full anisotropic refinement of thermal
parameters, and access to d-spacings of as low as 0.25 in simpler materials.
6. Diffraction From Disordered Materials
The important aspect of studies of materials which exhibit short range order, such as
liquids, amorphous materials and highly disordered crystals, is to access a very wide
Q-range in the measured pattern, S(Q). Only by so doing, along with very accurate
determination of the instrument normalisation and background, can the pattern be
reliably inverted (by Fourier transformation) to the pair distribution function g(r).
This g(r) function gives information about the relative distributions of the
particles (e.g atoms or molecules) in the sample. While the principles of the method
used for collecting TOF diffraction data from disordered materials are very similar
to those for powder diffraction (Figure 1), in this case it is especially important to
use a wide wavelength range at multiple angles (Figure 2; the GEM instrument is
used to study both crystalline and disordered materials). However, it is also vital to
keep the detectors well calibrated (both in terms of neutron detection performance
and position) and very well shielded (to reduce background). Good secondary
collimation (shielding between the sample and detector) is therefore usually
employed, and the stationary detector arrangment is beneficial here. Q values of up
to 50 1 can be accessed.
7. Small Angle Neutron Scattering (SANS)
As its name implies, SANS is a forward scattering technique. Once again the
strength of TOF SANS is in allowing simultaneous access to a very wide Q-range.
By coupling the wavelength range allowed by the TOF technique with a fairly wide
angle subtended by the area detector usually employed, TOF SANS typically
accesses Q ranges of ~0.005 0.2 1 in a single measurement without moving the
detector. This has the benefit of allowing often complex 3D modelling to be carried
out on the scattering patterns.

2.3-7
8. Neutron Reflectivity
Once again NR on a TOF source is a fixed angle technique. It is usually carried out
on a horizontal sample as shown in Figure 5.

Figure 5: The geometry of TOF neutron reflectometry. The white beam (I) is
incident upon a horizontal sample, and the reflectivity profile (R) is recorded as
a function of TOF at a fixed angle, providing the IR(Q) used to analyse the characteristics of the sample perpendicular to the surface.

The use of a fixed horizontal sample in this way greatly simplifies the study of
liquid surfaces.
The reflectivity patterns are recorded in total reflection geometry i=r, in the
fixed detector (Figure 6), with the Q-range being provided by the wavelength
scanning of the TOF method.
Interference fringes are observed in the reflectivity profile if there is a
variation in the scattering length density perpendicular to the surface (see Chapter
on Reflectometry). The determined variation in scattering length density can reflect
changes in chemical composition, or the presence of, for example, a multilayer
structure. Detailed modelling of the scattering length density can be carried out,
most reliably if a wide range of reflectivity (and Q) have been measured
reflectivities of as low as 10 6 can be accessed. Accurate modelling of the
composition of adsorbed chemical systems is often achieved, as in SANS, by the
use of contrast variation. In this method selected components of the system are
deuterated, the large variation in scattering lengths between H and D leading to
substantial changes in the scattered intensities, thus offering extra information to
assist in reliable model building. An effective "resolution" of around an ngstrom
in layer thickness is easily obtained in NR.

S2

Monitor 1

S1

Detector

Sample

Frame Overlap
Mirrors

S3

Supermirror

S4

Monitor 2

2.3-8

Figure 6: Schematic view of the CRISP reflectometer at ISIS, showing the horizontal geometry, the arrangement of apertures, the horizontal sample and the
fixed detector.
Off-specular reflection can also be measured, usually using an area detector, and
gives information on, for example, surface roughness or on local surface
inhomogeneities.
9. Sample Environment
In common with all neutron scattering techniques, the use of sample environment
in TOF diffraction is a major strength. There are specific advantages inherent in the
TOF technique, however, related to the ability to measure the entire diffraction
pattern at a single, fixed scattering angle. For complicated, bulky sample
environment apparatus, it may therefore be possible to leave only small apertures,
for the incoming beam and at limited fixed scattering angles, yet still measure the
diffraction of interest (Figure 7).
Typical applications of this method include high pressure cells and thickwalled chemical reaction cells. The high degree of collimation means that
extraneous scattering from these bulky items is minimised, and the signal from the
(sometimes small) sample optimised.
10. Neutron Strain-Scanning
The neutron strain-scanning technique basically involves measuring diffraction

2.3-9

Figure 7: The TOF neutron diffraction technique is well suited to studying samples inside bulky sample environment apparatus, the ability to measure the
whole diffraction pattern at a single scattering angle (Eq. 6; Figure 1) being an
important factor in this. Here the sample is contained inside apparatus with only
narrow apertures for incoming and outgoing beams and at very limited scattering angles necessary.
patterns from limited gauge volumes within a bulky engineering component, and
characterises internal stresses and strains from the changes in position and shape of
the measured Bragg peaks. The benefits of TOF for neutron strain-scanning follow
from the above discussions of TOF powder diffraction and sample environment.
11. Further information
Further information on time-of-flight diffraction, instrumentation, techniques and
applications can be found on the WWW site of the ISIS Pulsed Neutron and Muon
Facility, Rutherford Appleton Laboratory, UK (http://www.isis.rl.ac.uk).

2.4-2
2 THE FUNDAMENTAL EQUATIONS
2.1 The nuclear structure factor
For a non-Bravais single crystal, i.e. a crystal with more than
r one atom per unit cell,
the nuclear structure factor at the reciprocal lattice point Q reads [1]:
r
r
N Q = j bj exp i Q rj exp Wj

( )

) (

where b j and Wj are the Fermi length and the Debye-Waller factor of the atom j
r
located at the position rj of the unit cell.
2.2 The magnetic interaction vector
For a non-Bravais
single crystal, the magnetic interaction vector at the reciprocal
r
lattice point Q is given by [1]:
2
r r r
r
r r
e

( ) 2m c

M Q =

where:

( )

Q M Q Q
r
Q 2

r rr
r r
r r
e2
= 0.2695 10-12 cm/B, M Q = f j Q m kj exp iQ rj exp Wj .
2mec 2
j

( )

( )

) ( )

r kr

f j (Q) is the form factor and m j is the Fouriemponent k of the magnetic moment of
the atom j. One must notice that because of the dipolar nature of
the magnetic
r r
interaction, the neutron is only sensitive
to the projection of M Q on a plane
r
perpendicular to the scattering vector Q .

( )

2.3 The scattering cross-section and the scattered polarisation

Using the density matrix formalism, Maleyev et al. [2,3] and Blume [4] could
r
predict the scattering cross-section t and the scattered polarisation vector P
within the Born approximation. Both fundamental equations are the sum of four
r
r
parts summarised in table 1 [8], where ki and k f are the incident and final wave
vectors of the neutron beam, and
is the Van Hove correlation function:

( A,B) =

(N

v
Q

r
, M , Qv =

1
1exp h
kbT
r
N Qv , M , Qv

(
(

))

A,B

(M

v
, Q

, NQv

2.4-3
Table 1: Scattering cross-section t
r
vector P

{ }

= N0

(2)
v0

r
. j P

{ }

= N0

(2 )3
v0

and scattered polarisation

3 THE POLARISED NEUTRON TECHNIQUES

3.1 The spin-dependent diffraction technique
The polarised neutron diffraction technique takes advantage of the incident
polarisation dependency of the cross-section t to measure precise quantitative
magnetisation distributions of single crystals. Its sensitivity to the distribution of
unpaired electrons in the unit cell is greater than for the conventional method using
unpolarised-beams (about 0.1 mB). This technique is mainly used for investigating
single crystals that are ferro- or ferri-magnetically ordered in an applied magnetic
field. It can also be applied to some antiferromagnetic materials where
r

( N * M ) 0 or

r
r
M* M 0

If the beam is perfectly polarised parallel (+) or opposite (-) to the applied field (Pi
= 1), and if the sample is ferro- or para-magnetic and has a centre of symmetry, we
get the simple expression for the flipping ratio:
r
N Q + sin ( ) M
r
+
r
R Q = t =
N Q sin ( ) M
t

( )

( )
( )

(Qr )
(Q)

where is the angle between the polarisation

vector (i.e. the applied field) rand the
r
direction of the scattering vector Q . From experimental values of R( Q ), the

2.4-5
scattering. In order to separate each of these components it is necessary to use
polarised neutron techniques. For example, in order to study spin correlations and
magnetic defect structures in systems such as paramagnets, spin glasses and
antiferromagnets, so called XYZ-polarisation analysis (XYZPA) is required [7].
In XYZPA, the longitudinal component of polarization of the scattered beam
along the arbitrarily defined x- y- and z-directions is measured. Using this
technique, and ensuring that the scattering vector is always in the x-y plane (defined
with respect to the neutron polarization vector), one can separate the nuclear,
magnetic and spin-incoherent structure factors unambiguously, on a multi-detector
spectrometer, by combining the six measured cross-sections (x-, y- and zdirections, spin-flip and non-spin-flip) given below:
xsf =

xnsf =

ysf =

ynsf =

zsf

1
2
2
cos + 1 mag + 3 si
2
1
2
2
1
+

si
sin
mag
2
3
1
2
=
mag +
si
2
3

nsf
ynsf
z

1
2
sin + 1 mag +
2
1
2
cos + 1 mag +
2
1
1
=
mag +
si + nc
2
3

1
si + nc + ii
3
1
si + nc + ii
3
+ ii

The subscripts nc, mag, si, and ii stand for nuclear coherent, magnetic, spin
incoherent and isotope incoherent contributions respectively. These equations
apply to systems with collinear magnetisation and a randomly oriented moment
direction (i.e., a paramagnetic system or an antiferromagnetic powder in zero
external field). In antiferromagnetically ordered single crystals, there will, in
general, be a strong correlation between the x and y components of the sample
v
v
magnetisation M , i.e. the components of M in the scattering plane. Therefore, in
order to separate the magnetic cross-section
r in an ordered antiferromagnetic single
v
crystal, either the angle between M and Q , or the magnetic structure factor of the
sample must be known in advance. In practice, if neither of these quantities are
known, then application of the equations will result merely in a possible change of
sign of the magnetic intensity, depending on the moment direction. In the case of
non-collinear systems, such as helical magnets, cross-terms appear in the magnetic
interaction potential and the equations do not hold [24]. The XYZPA method
cannot be applied to the study of ferromagnets since ferromagnetic domains and
demagnetisation fields will in general depolarise the neutron beam.
3.4 Zero-field spherical neutron polarimetry
The only way to measure the three components of the scattered polarisation is to
connect two independent magnetic guide-fields onto a zero-field sample chamber.
Cryopad is presently the only device
r able to achieve full vectorial control of the
neutron polarisation over a large Q range with a good precision [8]. The sample

2.4-6
chamber is maintained in a zero-field state, preventing any polarisation precession
due to a parasitic field.
In the case of magnetic structures, spherical neutron polarimetry (SNP) allows
the direct determination of the direction and phase of the magnetic interaction
r
vector, and for = 0 antiferromagnetic structures, SNP is a powerful method for
measuring precision form factors and magnetisation distributions [9].
Because transverse components contain information related to the
nuclearmagnetic interference terms, SNP is clearly the best method to study
antiferromagnetic compounds for which the magnetic and nuclear information are
located at the same place in the reciprocal space. In the inelastic case, one should be
able to observe the spinlattice correlations expected in geometrically frustrated
antiferromagnets or spin-Peierls systems like CuGe03.
Here are a few rules describing the polarisation behaviour when carrying out
SNP experiments:
The polarisation is conserved when the interaction is purely nuclear and coherent
(e.g. an isolated phonon or a nuclear Bragg peak without nuclear spin polarisation).
r
The polarisation precesses by  around M when the interaction is purely
magnetic non-chiral (e.g. a magnetic satellite of a collinear arrangement or a
magnon).
r
In the case of a rnon-collinear
magnetic structure, the intensity may dependr on Pi
r
*

M
M
Q
may create polarisation along the scattering vector
and for an helix,
.
r

When there are nuclearmagnetic interferences, the intensity depends on Pi when

r
r
r
N and M are in phase. Polarisation is created along M when N and M are in
r
r
phase (e.g. Fe2O3), and the polarisation rotates around M when N and M are in

quadrature (e.g. Cr2O3).

References
1. G.L. Squires, Thermal neutron scattering, Cambridge University Press.
2. S.V. Maleyev, Physica B 276-268 (1998) 236
3. S.V. Maleyev, V.G. Baryakhtar, R.A. Suris, Fiz. Tv. Tela 4 (1962) 3461 [Sov.
Phys. Solid state 4(12) (1963) 2533]
4. M. Blume, Phys. Rev. 130 (1963) 1670
5. G.M. Drabkin et al., Sov. Phys. JETP 20 (1965) 1548
6. R.M. Moon, T. Riste and W. Koehler, Phys. Rev. 181 (1969) 2533
7. O. Schrpf and H. Capellmann, Phys. Stat. Sol. A135 (1993) 359; O. Schrpf and

2.4-7
H Capellmann, Z. Phys. B - Condens. Matter 80 (1990) 253
8. F. Tasset, Physica B 156-157 (1989) 627; V. Nunez, P.J. Brown, J.B. Forsyth and
F. Tasset: Physica B 174 (1991) 60; P.J. Brown, J.B. Forsyth and F. Tasset: Proc. R.
Soc. Lond. A 442 (1993) 147; F. Tasset, Physica B 297 (2001) 1; P.J. Brown,
Physica B 297 (2001) 198
9. P.J. Brown, J.B. Forsyth and F. Tasset: Physica B 267-268 (1999) 215; P.J.
Brown, J.B. Forsyth, E. Lelivre-Berna and F. Tasset: J. Phys.: Condens. Matter 14
(2002) 1

2.5-1

Magnetic Form Factors

P.J. Brown

The magnetic form factor (q) is obtained from the fourier transform of the
magnetisation distribution of a single magnetic atom. Assuming that it has a unique
magnetisation direction it can be written
1)
where M gives the magnitude and direction of the moment and m(r) is a normalised
scalar function which describes how the intensity ofmagnetisationvaries over the volume of the atom. When the magnetisation arises from electrons in a single open shell
the magnetic form factor can be calculated from the radial distribution of the electrons
in that shell. The integrals from which the form factors are obtained have the form
2)
The jl are the spherical Bessel functions defined by\begin
3)
If the open shell has orbital quantum number l the form factorfor spin moment is
4)
and that for orbital moment
5)
The coefficients SLM, BLMhave to be computed from the orbitalwave-function [1].
The total spin moment Ms is given by S00 and the orbital moment ML by B00. For
small q the dipole approximation
6)
Tables 1-4 give coefficients of an analytic approximation to the j0 integrals for the
d electrons in ions of the 3d and 4d series, and the f electrons of some rare earth and
actinide ions. The approximation has the form
7)
For these expansions s = sin / in units of -1.
The integrals jl with L 0 are zero when s = 0 and have been fitted with the form
8)

2.5-2
Tables 5-13 give the coefficients obtained. For the transition metal series the fits were
made withform factor integrals calculated from Hartree-Fock wave-functions [2]. For
the rare-earth and actinide series the fits were with Dirac-Fock form factors [3, 4].
In the tables the number following the atom symbol indicates the ionisation state of
the atom. Thus the coefficients following Fe0 are for a neutral iron atom and those
following Fe2 are for Fe2+
References
[1] Marshall W and Lovesey S W (1971) Theory of thermal neutron scattering,
Chapter 6, Oxford University Press.
[2] Clementi E and Roetti C (1974) Atomic Data and Nuclear Data Tables 14, 177-478.
[3] Freeman A J and Descleaux J P (1979) J. Magn. Mag. Mater. 12, 11-21.
[4] Descleaux J P and Freeman A J (1978) J. Magn. Mag. Mater. 8, 119-129.

2.5-3

2.5-4

2.5-5

2.5-6

2.5-7

2.5-8

2.5-9

2.5-10

2.5-11

2.5-12

2.6-1

Time-of-Flight Inelastic Neutron

Scattering
R. S. Eccleston
Introduction
In the inelastic neutron scattering experiment, the quantity we measure is the
double differential cross section:
2

d
d dE

ki

b S (Q , )
2

The velocity of thermal neutrons is of the order of km s-1; consequently their energy
can be determined by measuring their time-of-flight over a distance of a few
metres. In the inelastic neutron experiment the parameters of interest are the energy
and momentum transfer, h and Q respectively.
Time-of-flight spectrometers may be divided into two classes: Direct geometry spectrometers: in which the incident energy, Ei, is defined by
a device such as a crystal or a chopper, and the final energy, Ef, is determined by
time-of-flight.
Indirect (inverted) geometry spectrometers: in which the sample is
illuminated by a white incident beam and Ef is defined by a crystal or a filter
and Ei is determined by time-of-flight.
At a pulsed source all spectrometers use the time-of-flight techniques. On steady
state sources pulsing devices such as choppers are required.

Figure 1: Distance-time plot for a direct geometry spectrometer (left) and an

indirect geometry spectrometer (right).

2.6-2
The mode of operation of each type of spectrometer can be clearly understood
if the distance-time plots are considered (figure 1).
) space
Accessible regions of (Q,
), consider the scattering
To understand the trajectories traced out through (Q,
triangles. Clearly for direct geometry spectrometers ki is fixed and kf varies as a
function of time whereas the reverse is true for indirect geometry instruments.

Figure 2: Scattering triangles for a direct geometry spectrometer (left) and an

indirect geometry spectrometer (right).
Applying the cosine rule
h 2 Q2
= Ei + E f 2 EiE f
2m

Q 2 = k 2i + k 2f 2k ik f cos

)12 cos

For direct geometry Ef can be eliminated leaving

h 2Q2
1
= 2 Ei h 2 cos [Ei (Ei h )] 2
2m

Each detector has a parabolic trajectory through (Q,) space.

To make optimum use of direct geometry spectrometers, they are equipped with
large detector arrays, giving simultaneous access to a large area of (Q,) space.
Similarly for indirect geometry Ei can be eliminated giving

[ (

h2Q 2
= 2E f + h 2 cos E f E f + h
2m

)]

The parabola are inverted. An important feature of the indirect geometry instrument
is the access to a wide range of energy transfers for energy loss.

2.6-3

) space: a direct geometry spectrometer

Figure 3: Trajectories through (Q,
(left) and an indirect geometry spectrometer (right) for detector at the given
scattering angles.
Generic instruments
The table (table 1) provides much
simplified examples of direct and
indirect geometry instruments. The
critical differences between the two are
the manner in which they cover (Q,)
space and their resolution characteristics.
On a steady state source, the moderator
as it is shown in the table is replaced by a
pulsing device such as a chopper.

Table 1: Generic instrument types.

The chopper spectrometer

The chopper is a ubiquitous feature of direct geometry spectrometers for reasons
given later in the chapter. Examples of chopper spectrometers include HET, MARI
and MAPS at ISIS and IN4, IN5 and IN6 at the ILL. In the case of some time-offlight spectrometers on steady state sources the chopper simply provides the pulsed
structure of the beam and a crystal monochromator is used to select the incident
energy, for example IN6. Other instruments such as IN5 use an array of choppers
both to provide the pulsed structure and to monochromate the beam.
On a pulsed source the design of a chopper spectrometer is in general quite
straightforward, with the key components arranged as indicated in the table. The

2.6-4
beam is monochromated using a Fermi chopper. A second chopper is usually
installed effectively to close the beamline at the time at which the proton beam hits
the spallation target, thereby preventing a large flux of epithermal neutrons entering
the spectrometer where they will thermalise and produce a background signal. In
all cases detector arrays tend to be as large as economically and physically possible
to maximise the efficiency of the spectrometer.
Resolution
For the sake of simplicity the resolution function of a chopper spectrometer is
considered. The energy resolution has two contributions: from the moderator pulse
width and from the band width of the chopper pulse. Concisely the expression can
be written:

2
2
3
3

h 2 tc
L + L3 h 2 tm
L h 2
=
1
1 + 1
+ 2
1 + 3 1

tc
Ei
L2
Ei tc
L2
Ei

Where tc is the chopper burst time width, tc is the time of flight at the chopper and
tm is the moderator pulse width. L1 is the moderator to chopper distance, L2 the
sample to detector distance and L3 the chopper to sample distance.
Single crystal experiments on a chopper spectrometer
Direct geometry spectrometers have, in the past, been thought of as being well adapted
for studying excitations in non-crystalline or polycrystalline samples, but not suitable for
single crystals. However, the development of instrumentation and experimental techniques has shown that direct geometry spectrometers can be used very effectively for
such studies allowing a broad survey of (Q,) space but also detailed, focussed studies.
From the scattering triangle we can see that an array of detectors will trace out
a sector in reciprocal space, and as we have seen earlier, each detector has a
parabolic trajectory through (Q,) space. Consequently a broad detector array
produces a surface in (Q||,Q,) space.
Consider the example of a ferromagnetic spin wave in a three-dimensional
magnetic system. As the dispersion intersects
the surface traced out by the time-of-flight
Figure 4. Use of a multidetector to
cover a region in (Q, Q) space.

2.6-5
trajectories for the detectors, scattering will be observed.
The development of large position sensitive detector (PSD) arrays provides
greater flexibility for single crystal measurements. The MAPS spectrometer at the
ISIS pulsed source has a detector area of 16m2 divided up into approximately
50000 pixels. For any given crystal orientation arbitrary cuts can be made across
the surface of the detector, providing the ability to collect data along several
crystallographic directions simultaneously. In essence, the spectrometer collects a
large (0.5Gbytes) volume of data and the choice of scans is made in software after
the experiment. For more information see reference (Perring).
Indirect geometry spectrometers
It is apparent from figure 2 that indirect geometry spectrometers offer access to a
wide range of neutron energy loss values. They also offer high resolution at the
elastic line and tend to offer broad coverage in terms of energy transfer with
reasonable resolution.
A variety of spectrometer designs offer different capabilities:
Crystal analyser spectrometer : molecular spectroscopy
Back scattering spectrometer : high resolution
Coherent excitations spectrometer : coherent excitations in single crystals.
Before discussing each of these in turn, it is useful briefly to consider the resolution
characteristics of an indirect geometry spectrometer.
The resolution of indirect geometry spectrometers
For the indirect geometry spectrometer the energy resolution contains terms
pertaining to the uncertainty in the angular spread of neutrons scattered from the
analyser, A and timing errors over the incident flight path, t. Timing errors arise
from several contributions, including moderator thickness, sample size broadening,
analyser thickness broadening, detector thickness and the finite width of data
collection time channels. If t is expressed in terms of an equivalent distance
=hit/m the energy resolution can be concisely expressed as:

2
E

h
L E
= 2 + i cot A A 1 + 2 i
Ei
L1 E f

L1 E f

1
2 2

where L 1 is the moderator to sample distance, and L 2 the sample to detector

distance. Clearly increasing A or L 1 improves resolution but physical and
instrumental limitations apply.

2.6-6
Back-scattering spectrometer
For a crystal analyser spectrometer, the resolution contains a cotA term which can
be reduced to almost zero by using a back-scattering geometry, thereby optimising
the definition of Ef. In a matched spectrometer, Ei should be determined to the same
precision, hence L 1 tends to be long. IRIS at the ISIS Facility represents an
excellent example of such an instrument. In the most frequently used mode of
operation the 002 reflection from graphite analysers defines an Ef of 1.845 meV
and provides resolution of 17eV. The analysers are cooled to reduce thermal
diffuse scattering (TDS), which broadens the elastic peak and gives rise to
additional background.
Crystal analyser spectrometers
For molecular spectroscopy, energy information is often more important than Q
information. Consequently, measuring energy transfer along a single trajectory
provides an efficient method for measuring excitations across a broad spectral
range. TOSCA at ISIS provides a good example. A graphite analyser defines an Ef
of 3meV, a cooled beryllium filter is used to remove higher order reflections. The
geometry of the secondary spectrometer is such that the sample, analyser and
detectors are parallel; this effectively time-focuses the scattered neutrons, reducing
uncertainty in the scattered neutron flight time.
Coherent excitations - PRISMA
One of the shortcomings of chopper spectrometers and the direct geometry
spectrometers described above is their inability to perform scans uniquely along

Figure 5: Schematic of the PRISMA

spectrometer.

Figure 6: Time-of-flight scan with

the PRISMA spectrometer.

2.6-7
high symmetry directions. By adopting a scattering geometry whereby the
condition sin / sinA = constant is met (figure 5), time-of-flight scan scans for all
detectors correspond to scans along a defined direction in Q space (figure 6).
The PRISMA spectrometer can operate in this mode. It is technically
challenging because the movement of analysers is limited by the risk of clashing.
Monochromating devices
Choppers
As mentioned above, choppers are key components of a direct geometry
spectrometer. For instruments using neutrons in the thermal to high energy ranges,
Fermi choppers are most appropriate; for cold neutron instruments disk choppers
offer inherent advantages.
Fermi choppers
A Fermi chopper is effectively a drum with a hole through the middle. The hole is
filled with alternating sheets of neutron absorbing material (slats) and transparent
material (slits). The slits and slats are curved, with the radius of curvature and the
slit/slat ratio optimised for specific energy ranges. On the ISIS chopper
spectrometers for example three slit packages are used to cover the incident energy
range from 15meV to 2eV. An additional slit package with very broad slits is used
to improve intensity by degrading resolution. The Fermi chopper can be rotated at
frequencies of up to 600Hz. The incident energy is defined by the phase of the
chopper relative to the incident neutron pulse. The resolution is dictated by the
slit/slat ratio of the slit package and the chopper frequency. Fermi choppers allow
continuous selection of Ei.
Disk choppers
Disk choppers are simply rotating disks with holes in them. They offer higher
transmission than Fermi choppers with the same flexibility in the selection of Ei.
Using two disks or a variable aperture chopper overcomes the need to change slit
package. They are also compact. However their frequency is limited by engineering
constraints, as is the thickness of the disks. Both these factors mean that disk
choppers are often not practical for thermal to high-energy neutron instruments.
Crystal
The use of crystal monochromators for time-of-flight spectroscopy is limited,
because they impose some geometrical constraints, and reflectivity falls as energy
rises. The geometrical constraints can be lifted by using a double bounce

2.6-8
arrangement, but at the expense of additional losses arising from the second
reflection. Focussing of the monochromator does offer the opportunity to trade
resolution for additional flux.
As discussed above, the use of crystal analysers is widespread on indirect
geometry spectrometers.
Filters
The Be filters effectively remove neutrons with an energy greater than the Bragg
cut-off of 5.2meV. Incident neutrons are scattered in the beryllium and then
absorbed in absorbing sheets within the filter. They are used to remove higher order
contamination in indirect geometry instruments. Their efficiency can be improved
by cooling, which reduces transmission of neutrons with energies above the cut-off
that arise from up-scattering by thermally excited phonons.
Nuclear resonant absorption from a foil is used for very high-energy measurements.
The eVS spectrometer at ISIS, for example, uses a filter difference technique using
uranium or gold foils to measure atomic momentum distributions in the eV range.
Reference
Neutron scattering (mostly) from low dimensional magnetic systems by T.G.
Perring, in Frontiers of Neutron Scattering, edited by A. Furrer, (World Scientific,
2000)

2.7-1

Three-axis spectroscopy
R. Currat and J. Kulda

Principle of the technique

In order to obtain information on the excitation spectrum at a given point Q in reciprocal space, a more sophisticated procedure has to be adopted than just scanning the
energy of the scattered neutrons. At each point of the scan the scattering triangle is
modified, as indicated in Fig. 1, so that ki and kf will close at the same momentum
transfer Q but the length of one of the wave vectors preferably ki is varied to
provide for the required energy transfer. As is clear from the Fig. 1, in addition to
the ki and kf modification, also all the angles of the scattering triangle will change.
Herein lies the price to be paid for the increased flexibility required for scans with
Q = const.: the measurements have to be performed in a step-by-step mode, making multichannel data acquisition quite difficult.

Figure 1: Scattering geometry for a Q = const. scan.

2.7-2
Compared to a direct-geometry TOF instrument, this disadvantage is, however, largely compensated by the high incident monochromatic flux due to the steadystate operation of the neutron source as opposed to a pulsed beam providing neutrons for only about 1/1000 of the total measurement time.
As a general rule, the TAS is the technique of choice whenever information on
excitations in single crystals is sought [1] at a well defined point or along a given
direction in the Q, space and when a detailed quantitative interpretation of the
measured spectra is intended. On the contrary, if a global overview of the excitations in a broad Q, range is the goal, the TOF may be of advantage as soon as the
number of useful detector channels can compensate the initial loss due to the time
modulation of the incident beam with a low duty-cycle.
The neutron three-axis spectrometer (TAS)

Figure 2: Schematic of a three-axis spectrometer.

2.7-3
A typical set-up of a three-axis spectrometer (TAS) is schematically displayed
in Fig. 2. The incident and scattered neutron wave vectors, ki and kf, are selected by
Bragg diffraction on the monochromator and analyser crystals, respectively.
Traditionally, large crystals of Cu, Zn or pyrolitic graphite (PG) with mosaic width
of 20 30 minutes have been used in combination with Soller collimators defining
the beam divergence of the incident and diffracted beams. On present instruments
the mosaic crystals are segmented into plates of a few centimeters size and mounted on mechanical devices (benders), which permit to control their individual orientations in order to act as curved mirrors and to focus the neutron beam horizontally and/or vertically on the sample and detector [2]. More recently, elastically bent
perfect Si and Ge crystals, offering improved focusing properties thanks to the
absence of random mosaic block misorientations, get employed for work requiring
higher resolution [3]. The monochromatic neutron flux incident on the sample is
monitored by a low-efficiency ( 10-5 - 10-4) detector (monitor 1), another similar monitor 2 is placed in the scattered beam to detect the eventual presence of
strong Bragg diffraction peaks, which might give rise to spurious signals in the
inelastic spectra. The detector is usually a single 3He proportional gas tube.
The monochromator and analyser crystals diffract together with the nominal
wavelength also its harmonics /2, /3 etc., which may be at the origin of spurious effects. Usually, the most dangerous one is /2, because it is most intense and
its energy is closest to the nominal one. Harmonic contaminations may be suppressed by using appropriate filters. For hot neutrons absorption filters are available, using sharp absorption resonances of certain nuclei in the 0.1 - 1 eV range. For
thermal neutrons pyrolithic graphite (PG) permits to suppress efficiently the second
order for two "magic" wavelengths of = 1.53 and 2.36 , corresponding to k
= 4.1 1 and 2.66 1, respectively. For cold neutrons cooled polycrystalline
beryllium permits to cut all wavelengths below 4 , ie. to transmit only neutrons
with k < 1.57 -1. Alternatively, silicon and germanium crystals, whose odd-hkl
reflections are free from 2nd order contamination, can be used for certain applications.
The TAS resolution function
The resolution volume, i.e. the region of the Q, space which corresponds to the
momentum and energy transfers of neutrons scattered and registered in a given
spectrometer configuration, can be approximated by a rather anisotropic 4D ellipsoid. Its form is determined by the convolution product of the two reciprocal space
distribution functions pi(ki) and pf(kf) which describe the transmission of the monochromator and analyser arms [4]. For a spectrometer configuration characterised by

2.7-4
nominal values Q0,0 of Q, the measured intensity is given by
I(Q0,0) = N A(kI) R(Q-Q0,0) S(Q,) dQ d,

(1)

where A(kI) and S(Q,) describe the source spectrum and the sample scattering
function, respectively. The norm of the resolution function is the product of the
norms of the two distributions:
R(Q-Q0,0) dQ d = pi(ki) pf(kf) dki d kf = VI VF .

(2)

In the gaussian approximation [4], [5] the TAS resolution function depends on the
kinematic variables ki, kf, which define the energy and momentum transfer Q,
and on the collimation angles, mosaic widths and scattering senses. For a flat mosaic monochromator crystal (i.e. in the absence of focusing) one gets:
VI = Pm ( kI ) kI3 cot M (2 )

m . 0 .1
4 m + 0 + 1
2

0 1
2

4sin 2 ( m ) 'm +0 + 1
2

(3)

where kI is the average incident neutron wave vector, kI =VI-1 ki pi(ki) dki, and
0, 0, 1 and 1 are respectively, the horizontal and vertical beam collimations
before and after the monochromator; m and m are the horizontal and vertical
mosaic widths of the monochromating crystal and Pm(kI) its the peak reflectivity.
The expression for VF is analogous.
Typical beam divergences are about 0.5 2 degrees in both horizontal and
vertical directions leading to relative momentum transfer resolution Q/Q 10-2.
The overall energy resolution lies usually in the range of 5 10%, but can be
improved in particular cases. Depending on the spectrum of the neutron source
hot, thermal or cold the TAS can be employed for studies of excitations in the
energy range from 50 eV to 200 meV.
In reality the resolution volume has a more complicated shape, influenced by
spatial variables (source size, slit widths, sample dimensions) and by the eventual
use of composite crystals made up from platelike segments to obtain approximate
horizontal and/or vertical focusing. Originally, the Gaussian approximation has
been extended to enable a more realistic description of the TAS resolution function
[6]. More recently, the RESTRAX [7] and McStas [8] computer packages became
available, which provide a highly realistic Monte-Carlo ray-tracing simulation of
the whole instruments including effects of neutron guides, slits etc. whose intrinsic
transmission profiles are not necessarily represented by gaussian distributions.

2.7-5
Preparation and optimization of a TAS experiment
The first thing to be made in preparing a TAS experiment is to define the required
neutron energy range matching the problem to be studied. Next a suitable instrument installed at a corresponding neutron source (cold, thermal or hot) is to be
selected. Among the main factors influencing such a decision we may note
the kinematic constraints, corresponding to the momentum conservation law
(closing the triangle), to be satisfied over all the required Q, range; these
constraints are particularly severe in studies of magnetic materials, where lowQ and large- often appear in combination
the resolution/intensity conditions required for a given experiment; any
improvement in resolution is paid for by a loss in countrate
the angular limitations on the chosen instrument, notably the maximum value
of the scattering angle q, which determines the maximum value of Q that may
be reached for a given set of ki, kf values
the risk of spurious signals; as a rule of thumb the final energy should be comparable to the energy transfer in the downscattering mode in order to minimise
harmonic contaminations (see below) and the same applies to the incident
energy in the upscattering mode
Table 1 indicates typical neutron energies and instrument parameters for a
quick overview. Precise values depend on the detailed characteristics of each instrument, the fluxes correspond to the instruments at the High Flux Reactor at ILL
Grenoble (France). The flux values available at smaller reactors may be 1-2 orders
of magnitude lower. Optimising the experimental conditions also involves making
choices and compromises with respect to the d-spacing of the monochromator and

Table 1: Typical neutron energies and instrument parameters for TAS spectrometers at the ILL.
TAS
Mono/analyser
Flux [107 cm-2s-1]
ki [-1]

cold

thermal

hot

PG002

PG002

Cu220

1 - 3

3 - 30

1 - 3

1 2.66

2.66 - 6

5 - 15

Qmax [-1]

12

Q [-1]

0.003

0.01

0.03

0.1 - 10

5 - 60

50 - 200

0.05 0.5

0.8 - 4

4 - 10

E [meV]

E [meV]

2.7-6
analyser crystals, collimations, the use of flat or focusing bent crystals, the scattering configurations. When carrying out this complex optimisation process one
should use one of the simulation programs mentioned above.
Data treatment.
As a rule, the intensities measured in a constant-Q scan should be corrected point
by point for the variation of the norm of the resolution function (zeroth-order resolution correction).
For a scan obtained in the constant-kf mode, VF is constant across the scan and the
measured intensities should be corrected for the variation of A(kI) VI. Since the quantity A(kI) kI VI measures the neutron flux incident on the sample, it is sufficient to use
a monitor with a 1/vn characteristic in the incident beam to normalise the counting time
per point. However the presence of higher-order harmonics (/2, /3, ) in the incident beam requires, in general, additional corrections to be applied (see below).
For a scan obtained in the constant-ki mode the measured intensities must be
corrected for the variation of Pa(kF) kF3cota. Note that Pa(kF) may vary rapidly as
a function of kF due to parasitic multiple Bragg reflections.
Spurious signals.
An element, which may perturb the measurements, is the presence of the harmonic
wavevectors 2ki, 3ki, in the primary beam or 2kf, 3kf, ... in the scattered beam due
to higher-order diffraction at the monochromator or analyser position. For instance,
the contribution of the unfiltered 2ki and 3ki harmonics to the monitor count-rate
(used to normalise the acquisition times in the constant-kf mode) can be quite significant and lead to large scan profile distorsions unless corrected for.
Harmonics also lead to spurious signals for commensurate values of ki and kf:
nki = mkf
with n,m = 1,2,3,4,..
due to elastic (coherent or incoherent) scattering on the sample (or its environment).
To minimise the probability of such processes one should limit the magnitude of the
energy transfer with respect to the incident neutron energy. A safe rule of thumb is
given by 2/3 < ki/kf < 3/2.
Spurious inelastic peaks are observed for 3-axis configurations such that 2
crystals out of 3 (sample + monochromator or sample + analyser) satisfy a Bragg
condition for a given neutron energy. These configurations are illustrated schematically in Figs. 3a and 3b.
The spurious process shown in Fig. 3a (type I) can be described as follows:
incident neutrons with the nominal wave vector ki are Bragg scattered along the
nominal scattered-beam direction (kf - direction). These neutrons do not have the

2.7-7
proper energy to be Bragg-scattered by the analyser. Nevertheless some of them
may be scattered into the detector via some other process (incoherent, diffuse or
inelastic scattering) taking place at the analyser position. The low efficiency of such
processes is offset by the high efficiency of Bragg scattering at the sample and thus,
if in the course of a constant-Q or constant- scan a configuration such as the one
shown in Fig. 3a is encountered, a significant increase in the detector counting rate
may result. Such a process can be detected with the help of a monitor placed immediately before the analyser crystal, the so-called monitor 2.

Figure 3: Configuration for spurious inelastic peaks.

Fig. 3(b) shows the reverse process (type II) where the non-Bragg scattering
occurs at the monochromator position. This latter type of event is not detected by
monitor 2.
Around any Bragg peak hkl there exist two directions in Q-space for which
spurious acoustic-like inelastic peaks will appear on the focused side of a constant-Q scan: these dangerous directions are those of kiB and kfB , where kiB and
kfB refer to the Bragg scattering configuration:
kiB kfB = hkl

When searching for long-wavelength excitations, i.e. when Q = + q with q <<

the above picture must be extended, in such a way as to include the effect of sample mosaicity.
Effects related to sample mosaicity are significant when the ratio |q|/|| becomes

2.7-8
comparable to or smaller than the sample mosaic spread, expressed in radians.
Qualitatively, the effects are twofold:
the spurious acoustic-like branches are no longer restricted to special qdirections.
the peaks occur symmetrically on both the focused and defocused sides of a
constant-Q scan.
Quantitatively, the spurion dispersion relation is found as
s (q)(THz) 2

k 2i
q
2

(4)

where the - and + signs refer to type I and type II processes, respectively.
When the spectrometer is operated in the constant kf mode, the value of ki, at which
the spurious peak should occur, is not known a priori. Since, however, eq.(1) above
is only valid to first order in |q|/|| one may replace ki by kf.
References
[1] C. Stassis, In: Neutron Scattering (K. Skold, D.L. Price, eds.), Methods of
Experimental Physics Vol. 23A, Academic Press, 1986, p.369.
W.G. Stirling, K.A. McEwen, ibid Vol. 23C, p.159.
[2] L. Pintschovius In: Proceeding of the Workshop on Focusing Bragg Optics,
Braunschweig, Germany, 1993. Nucl. Instr. And Meth. In Phys. Res. A338 (1994) 136.
W. Buehrer, ibid p. 44.
[3] J.Kulda & J.Saroun, Nucl. Inst. Meth. A379 (1996) 155-166.
[4] M.J. Cooper, B. Nathans, Acta Cryst. 23 (1967) 357-367.
[5] B. Dorner, Acta Cryst. A28 (1972) 319-327.
[6] M.Popovici, A.D. Stoica, M. Ionita, J. Appl. Cryst. 20 (1987) 90.
[7] J. Saroun, J. Kulda, Physica B 234-236 (1997) 1102.
[8] K. Lefman, K. Nielsen, A. Tennant, B. Lake, Physica B 276-8 (2000) 152-3.

2.8-1

The basics of Neutron Spin Echo

B. Farago
Introduction
Until 1974 inelastic neutron scattering consisted of producing by some means a
neutron beam of known speed and measuring the final speed of the neutrons after
the scattering event. The smaller the energy change was, the better the neutron
speed had to be defined. As the neutrons come form a reactor with an
approximately Maxwell distribution, an infinitely good energy resolution can be
achieved only at the expense of infinitely low count rate. This introduces a practical
resolution limit around 0.1eV on back-scattering instruments.
In 1972 F. Mezei discovered the method of Neutron Spin Echo. As we will see
in the following, this method decouples the energy resolution from intensity loss.
Basics
It can be shown [1] that an ensemble of polarized particle with a magnetic moment
and 1/2 spin behaves exactly like a classical magnetic moment. Entering a region
with a magnetic field perpendicular to its magnetic moment it will undergo Larmor
precession. In the case of neutrons:
= B
where B is the magnetic field, is the frequency of rotation and =2.916kHz/Oe is
the gyromagnetic ratio of the neutron.
When the magnetic field changes its direction relative to the neutron
trajectory, two limiting cases have to be distinguished:
1) Adiabatic change, when the change of the magnetic field direction (as seen by
the neutron) is slow compared to the Larmor precession. In this case the
component of the beam polarization which was parallel to the magnetic field
will be maintained and it will follow the field direction.
2) Sudden change, is just the opposite limit. In that case the polarization will not
follow the field direction. This limit is used to realize the static (Mezei)
flippers.
Now let us consider a polarized neutron beam which enters a B magnetic field
region of length l1. The total precession angle of the neutron will be:
B l

1 = 1 1
1

2.8-2
depending on the velocity (wavelength) of the neutrons. If v 1 has a finite
distribution, after a short distance (a few turns of the polarization) the beam will
appear to be completely depolarized. If now the beam will go through an other
region with opposite field B2 and length l2 the total precession angle is:
tot =

B1 l1 B2 l2

1
2

In the case of elastic scattering on the sample v1=v2= v and if B1l1 = B2l2 then tot
will be zero independently of v and we recover the original beam polarization. Let
us suppose now that a neutron is scattered with a small energy exchange between
the B1 and B2 region. In that case in leading order in :
tot =

hBl

m 3

where m is the mass of the neutron

If we put an analyzer after the second precession field and the angle between the
polarization of a neutron and the analyzer direction is , then the probability that a
neutron is transmitted is cos. We have to take the expectation value of cos over
all the scattered neutrons. At a given q the probability of the scattering with
energy exchange is by definition S(q,). Consequently the beam polarization
measured is:
cos =

cos(

hBl
) S ( q, ) d
m 3
= S ( q, t )
S ( q, ) d

So NSE directly measures the intermediate scattering function where:

hBl
t=
m 3
It is important to notice that t

3 so the resolution in t increases very rapidly with .

Implementation
In practice reversing the magnetic field is difficult as in the middle this would
create a zero field point where the beam gets easily depolarized. Instead, a
continuous horizontal field is used (conveniently produced by solenoids) and a [/2
flipper starts the precession by flipping the horizontal polarization perpendicular to
the magnetic field. The field reversal is replaced by a [ flipper which reverses the
precession plane around an axis and at the end a second [/2 flipper stops the
precession and turns the recovered polarization in the direction of the analyzer.

2.8-3

Figure 1: Sketch of a NSE spectrometer

Elements
Monochromatisation:
The biggest advantage of NSE is that it decouples monochromatisation from energy
resolution. Whenever q resolution is not so important (or it is determined by the
angular resolution rather than the wavelength distribution ), a neutron velocity selector
with high transmission (e.g. /=15% FWHM ) is the best choice. If necessary the
wavelength distribution can be reduced by installing after the analyzer a crystal
monochromator (graphite or mica) and the detector in Bragg geometry.
A further option is to use time of flight (TOF) as implemented on IN15. The
fully polychromatic beam (Maxwell distribution from the reactor) can be pulsed by
using rotating disk choppers relatively far (e.g. 20m) from the detector. By the time
the neutrons reach the detector, the pulse will be completely spread and from the
flight time we can calculate back that at any moment which wavelength at what
monochromatization is detected. Collecting data in different time channels, a whole
range of wavelengths can be used.
Polarizer:
This is a very essential part of the spectrometer, and generally supermirrors are
used. There are two possible choices, reflection or transmission mode. As the
polariser-sample distance is in the order of 2-3m the main danger in both cases is
the unwanted increase of the incoming beam divergence from the polariser. Due to
that relatively big distance here one can loose integer factors in the neutron flux on

2.8-4
the sample. Inherently the reflecting supermirrors are more susceptible to increase
the beam divergence, and also have the inconvenience that the reflection angle
might have to be adjusted to different wavelength if the useful range is large
enough. On the other hand the supermirrors of this type were easier to obtain.
Recent progress allows the production of supermirrors on Si wafers with
sufficiently good quality to make cavity like transmission devices.
As this is the most important part of the incoming beam it needs a closer
examination. If we look at the trajectory A which comes parallel with the beam
axis, we can see that it becomes very divergent after the polariser. Indeed if the

Figure 2: Cavity type transmission polariser

wavelength is sufficiently large, the first reflection on the supermirror is below the
critical angle and the polarization happens only when the neutron is bounced back
from the guide, but now with a larger incoming angle. However there is trajectory
B arriving with the same large divergence and, as shown on Fig 2, will go out of the
polariser parallel with the axis. If this very divergent neutron is missing, due to
misalignement or cuts in the guide, trajectory A will not be replaced and effectively
the center of the beam will be emptied.
 and /2 flipper:
The flipper action is based on the nonadiabatic limit. Using a flat coil (5mm thick)
perpendicular to the neutron beam with a relatively strong field inside and
arranging such that the horizontal field is small, the polarized neutrons going
through the (thin) winding do not have time to follow adiabatically the change in
the field direction. All of a sudden the beam polarization is no more parallel to the
magnetic field and it will start to precess around the field. With the appropriate
choice of the field strength and direction, the neutrons will precess exactly [ or [/2
by the time they fly through the flipper.
In the case of TOF, the flipper current has to be modulated in time to achieve the
desired action for the wavelength which is just passing through.

2.8-5
/2 Flipper
180 Precession around an axis at 45

Flipper
180 Precession around a vertical axis

Figure 3: ?/2 and ? Mezei flippers simplified scheme

Main precession coils
We have seen that the Fourier time is proportional to the field integral. If we want a good
resolution, we need high field and long integration path. Both will have a drawback.
Even the most symmetric solution for the main precession coil, a solenoid, will produce
an inhomogenous field for finite beam size. The underlying physical reason is that in the
middle of the precession coil we have a strong field and at the flippers, as was explained
above, we need a small horizontal component. Since Maxwell we know that div(B) is
zero, consequently the transition region will introduce inhomogeneities.
Making long precession coils is not necessarily optimum because that
increases the distance between the source (polariser exit) and the sample as well as
the sample-detector distance. The counting rate will decrease with R-2*R-2.
Special attention has to be payed that at least the winding is as regular as
possible to be able to calculate the necessary corrections. This is not always an
evident task as for example on IN11, the coils consist of two layers of water cooled
hollow conductor, with a total number of turns 400 and max current of 600 Amps.
Fresnel coils
How to correct the unavoidable inhomogeneities due to the finite size beam? Let us
consider the field integral difference between the trajectory on the symmetry axis
and one parallel at a distance r through the solenoid. In leading order this gives:
Bdz

2
r2
1 B( z )

dz
8 B( z ) z

2.8-6
It can be shown also that a current loop placed in a strong field region changes the
field integral by 4[I, if the trajectory passes inside the loop and by zero if it passes
outside. To correct the above calculated r2 dependence we just have to place properly
arranged current loops in the beam. The so called Fresnel coils just do that.[1]
They are usually made by printed circuit technology preparing two spirals
where the radius changes as r , placed
back to back with a thinnest possible insulator
(30 micron kapton) to minimize neutron
absorption.
Resolution
Why field inhomogeneities are bad? In fact all
we are interested in is the field integral. We
have to use finite beam size if we want to have
finite counting rates. With finite beam size we
Figure 4: Fresnel coil
have different neutron trajectories arriving on
the sample and coming to the detector. If the
field integral for different trajectories are non equal then the final precession angle
B1 l1 B2 l2 will be different for different trajectories and the detected

tot =

Figure 5: Echo
response for an
elastic scatterer
(resolution) and
sample.

polarization even for an elastic scatterer will be reduced to cos . The echo
measured on the sample will be further reduced due to the energy exchange. Using
an elastic scatterer we can measure the influence of the inhomogeneities and divide
it out. (in other words while in energy space the instrumental response has to be
deconvoluted from the measured curve to obtain the sample response, in the
Fourier transformed space the deconvolution becomes a simple division)
Nevertheless if the field integral for different trajectories is too different, then
cos quickly becomes zero and no matter how precisely we measure it, division
by zero is usually problematic!

2.8-7
Figure 6:
Resolution corrected
sample response

To put this into context lets have a look at some realistic numbers:
On IN11 we have a maximum field integral of 3*10 5 Gausscm. With 8
wavelength this gives a Fourier time of 28nsec. An exponential decrease with a
time constant of 28nsec correspond in energy space to a Lorentzian line width of 23
neV. With the incoming neutron energy (8) this is a relative energy exchange of
the order of 10 -5. The 360 degree phase difference between two trajectories
corresponds to 17 Gausscm difference in field integral. This means a requested
relative precision of better than 6*10-5 for all trajectories!
So what is the theoretical limit to the achievable resolution? There is no easy
answer. In principle with sufficiently precise in-beam correction coils (like the
Fresnels but including higher order term corrections) any resolution should be
reachable. In practice we could certainly reached already an overall correction
better than 10-4.
If one changes the collimation conditions, the neutrons will explore different
trajectories thus picking up different field inhomogeneities and the spectrometer
resolution will change. When the resolution is pushed to its limits the number of
important parameters which can give artifact increases rapidly and particular
attention has to be paid to avoid them.
Measurement sequence
Now we will detail how an experiment is really done. Once the aims of the
experiments are clear one has to decide what wavelength to use. Long wavelength
gives better resolution but lower flux.
Then it has to be decided at which precession field (Fourier time) the echo
points has to be measured. Depending on the problem studied, it can be adequate to
take equidistant points on a logarithmic scale.
The field integral on the two side of the [ flipper must be equal to a precision
of 10-5. This is done by using an additional winding on the first precession coil, the

2.8-8

Figure 7: Measurement sequence

symmetry coil. Scanning the current in this coil will go through the exact symmetry
point producing the typical echo group as shown in Fig. 7.
The periodicity of the damped oscillation is determined by the average
wavelength and the envelope is the Fourier transform of the wavelength
distribution. These depend only on the monochromator (velocity selector) so do not
carry information on the sample. (With one notable exception, when the sample is
crystalline and remonochromatize the beam. In that case the echo envelope
becomes much more extended).
With the echo we measure how much of the initial polarization we recover. For
elastic scatterer on an ideal instrument we recover it fully. During the scattering
process the initial beam polarization can change. One reason for this can be when the
sample contains hydrogen which has a high spin incoherent cross section. Spin
incoherent scattering changes the initial polarization from P to -1/3P. The scattered
intensity is thus composed of the sum of coherent and incoherent scattering decreasing
the up/down ratio. The relative weight of these can, and will change as a function of
scattering angle (or wave vector q). The measurement sequence thus starts with the
measurement of spin-up and spin down for the given scattering angle and sample
condition. This is accomplished by setting a guide field to maintain beam polarization
and counting with [ flipper on and [ flipper off. This will give us two points marked as
Up and Down on the figure 7. It would be inefficient to measure always the whole
echo group as we are interested only in its maximum amplitude. At most it is equal to
(Up-Down)/2. On an elastic scatterer we can quickly find where the center of the echo
group is as a function of the current in the symmetry coil. As we also know the
periodicity of the sinusoidal echo group it is more efficient to measure four points
placed around the center with [/2 steps. This will give us:

2.8-9
E1 = Aver+Echo*sin()
E2 = Aver-Echo*cos()
E3 = Aver-Echo*sin()
E4 = Aver+Echo*cos()
From the four measurements Aver,Echo and (the phase) can be determined, and
finally
I(q,t)/I(q,0) = 2*Echo/(Up-Down)
If we are sure that the phase is zero, then it is sufficient to measure E2 and E4.
However in general to monitor any possible phase drift, (due to current instability,
external perturbation, .... or accidental displacement of a coil.....) usually all the
four points are measured but giving more weight to E2 and E4.
Signal and Background
No general recipe can be given, besides that both should be measured and the
appropriate correction done! Nevertheless let us discuss a few typical cases and for
the sake of simplicity let us take an ideal instrument (all flippers, polariser,analyzer
having 100% efficiency, perfect field integrals).
Coherent, (spin) incoherent scattering:
As was mentioned, incoherent scattering changes the beam polarization to -1/3.
Lets consider the case when the scattered intensity is composed of I= Icoh +
Iincoh. Then we will have:
Up = Icoh+1/3Iincoh
Down = 2/3Iincoh
Echo = Icoh*fcoh(t) -1/3Iincoh*fincoh(t)
where fcoh(t),fincoh(t) denotes the time dependence (dynamics) of each
contribution and at t=0 both are equal to one.
Our usual data treatment will give
I(q,t)/I(q,0) = [Icoh*fcoh(t) -1/3Iincoh*fincoh(t)]/[Icoh-1/3Iincoh]
This means that from Up and Down we can determine the relative weight of each
contribution but the separate time dependence is impossible to extract (except if
suitable model exist and even then, it is hard to achieve the necessary statistical
accuracy) . Fortunately in many cases it is possible to minimize the incoherent
contribution by using deuterated samples. (Note the incoherent echo signal is
further reduced by the factor 1/3) In other cases it is possible to suppose that the

2.8-10
dynamics of fcoh(t) and fincoh(t) is the same or nearly identical and then:
I(q,t)/I(q,0) = [Icoh*fcoh(t) -1/3Iincoh*fincoh(t)]/[Icoh-1/3Iincoh] = f(t)
There can be also some scattering from the sample holder, solvent, etc. Some cases
can be treated in a simplified way. For example in polymer solution, scattering
from deuterated solvent gives coherent scattering, but it is too inelastic to be seen
by NSE, which means its relaxation time is out of the time window. In this case it
will influence Up-Down but gives no echo. It is sufficient to measure Up and Down
of the solvent. Similarly sample holders usually give only some elastic scattering.
Again Up,Down and a few echo points in time are sufficient to establish its
contribution.
NSE vs standard inelastic instruments
NSE not only has very high energy resolution but also has a very wide dynamical
range. With some tricks (double echo) it can be as high as tmax/tmin = 1000 [2].
With the combination of several wavelength this can be further extended by a factor
10-100. We can measure energy exchanges up to 100eV which not only overlaps
with back scattering but also with Time-Of-Flight like IN5 or IN6. When to choose
which instrument?
The fundamental difference is that NSE measures in Fourier time while TOF in
omega space. Consider a sample which has a strong (95% of the intensity) elastic
line, and well defined but weak (5%) excitations at finite energy. While this is well
separated in omega space (Fig. 8), on NSE the Fourier transform will give a small
(5%) cosine oscillation around 0.95. Statistically this is very unfavorable (Fig. 9).
Furthermore the rough wavelength distribution (around 15% FWHM) of NSE will
smear out the oscillations (t ~3). For that type of experiments TOF is better suited.
Figure 8: Spectrum
in w space measured
on a Time-Of-Flight
spectrometer.

On the other hand when all (or most of) the intensity is quasielastic, NSE can
give more precise information on the time dependence. The main reason is that the

2.8-11

Figure 9: Same
spectrum than Fig. 8
measured in the time
domain by NSE.

deconvolution of the instrumental resolution is replaced by a simple division which

leaves the data less influenced by numerical manipulation. Below is an example of
the relaxation time measured on magnetic nanoparticles imbedded in an aluminium
matrix (Fig. 10). With the combination of IN11 and IN15 (the NSE instrument
optimized for long wavelength , long Fourier times) more than four decades in time
could be covered.

Figure 10: Relaxation spectrum of diluted magnetic nanoparticles measured

by NSE.
Studying fully incoherent scatterers is also not very favorable for NSE.
Already incoherent scattering in general gives rather weak intensity, secondly the 1/3 factor in polarization decreases the statistical accuracy. These drawbacks are
somewhat lifted by now on the multidetector spin echo instruments like IN11C and
SPAN in Berlin. (see below)
Finally an interesting observation. If accidentally the incoherent contribution
is about three times higher than the coherent one, then we end up with no

2.8-12
polarization of the scattered beam and if the dynamics of the two is the same, the
experiment is unfeasible on NSE as the echo signal will be always zero.
Variants
Inelastic NSE
One could ask the question whether the method could be used in the case of well defined
excitations? The answer is yes, but with some modification. By choosing B1l1 B2l2
the same Fourier back-transformation happens but around the excitation energy 0 [1]:
0 =

m 2 B1l1 2
1 1

2 h B2 l2

cos(

hB1l1
3 ( 0 ))S( q, ) d
m1
= S( q, t )
S( q,( 0 )) d

v1 is the velocity of the incoming neutron and v2 is after the scattering

The problem here is that around =0, S(q,) usually has a strong peak (either
quasielastic scattering, diffuse elastic scattering and/or incoherent scattering or just
instrumental background) The solution is to use a triple axis spectrometer geometry
which acts as a filter, pre-cutting an interval around 0 with a coarse resolution
compared to NSE. There is a further requirement when is q dependent ( like a
phonon branch). We have to make the spin precession dependent on the incoming
and outgoing direction. Interested readers should look up the details in ref[1]. Here
we just give an easy to understand picture. Quasielastic NSE measures the Fourier
transform from to time space. This is a Fourier transform in the 1D space. Now
we switch to the (at least) 2D q- space. If is q independent (at least in the

Figure 11: Principles for measuring linewidth on a dispersion curve by NSE.

2.8-13
vicinity we are measuring, e.g. close to the Brillouin zone boundary) with
asymmetric field integral we make the Fourier transform around a finite 0 and we
obtain the line width. Now if we do the same e.g. on an acoustical branch we will
Fourier transform along a vertical line. To measure the real physical line width, we
need to do the Fourier-transformation along a line perpendicular to the dispersion
relation (Fig. 11). This refocalisation of the echo works if the spin precession
becomes dependent on the neutron direction.
Magnetic scattering
The application of NSE to magnetic scattering has special features. One is that
ferromagnetic samples can hardly be studied. The random orientation of magnetic
domains introduces Larmor precessions around random axis and unknown
precession angles normally resulting in depolarisation of the beam and the loss of
the echo signal. In some cases the application of a strong external field to align all
the domains in the same direction can help for one component of the polarization to
survive [1]
The second case is paramagnetic scattering. If the scattering vector ( q) is in
the y direction, then in the sample only the spin components which are in the xz
plane (perpendicular to q) will contribute to the magnetic scattering. As the
precession plane of the neutrons in the usual NSE geometry is the xy, the neutrons
which arrive spin parallel to q will undergo spin flip scattering, while those arriving
perpendicular to q will have 50% probability of spin flip and 50% probability of
non spin flip scattering. Accordingly the scattered beam polarization can be
decomposed as:

Px
P
1 x
1
Py + Py
2
2

0
0

We can realize that the second term is equivalent to a [ flipped beam around
the y axis. Consequently without a [ flipper the second term will give an echo with
50% amplitude of the magnetic scattering, while using a [ flipper the first term
results in a negative echo of 50% amplitude plus a full echo from the eventual
nuclear scattering as well. This means that without [ flipper only the magnetic
scattering gives an echo signal, which eliminates the need of time consuming
background measurements to separate the nuclear contribution.
Antiferromagnetic samples do not depolarize the beam, so in most cases they are
identical to paramagnetic samples. However if we measure a single-crystal it might
happen to be a single domain and depending on the spin direction one can have full
echo without [ flipper or we might need a [ flipper to see any echo.

2.8-14
Resonance or Zero Field Neutron Spin Echo (ZFNSE)
This method was introduced by Ghler & Golub [3]. We will just give a quick
explanation how to understand the basic principle. The method has clear
advantages in the case of inelastic NSE. Whether the quasielastic NSE will or will
not have better performance we will see when sufficient experience will be
accumulated with ZFNSE.
The magnetic field profile along the axis of a classical NSE schematically is
the following: Nearly zero horizontal field at the first [/2 flipper, strong (B0) field
in the first solenoid, nearly zero at the [ flipper, again strong field in the second
precession region, finally nearly zero again at the second [/2 flipper. In B0 the
neutron precesses with the Larmor frequency 0=B0. Let us choose now a rotating
frame reference which rotates with 0. In this frame the neutron seems not to
precess in the solenoids which means it sees B0=0. However now in the originally
low field region it will rotate with 0 which indicates the presence of a strong B0
field. The small field of our static flipper will appear to rotate also with 0.
Let us realize this configuration now in the lab frame. Inside the [ and [/2
flippers we must have a strong static field B0 and perpendicular to that a small field
which rotates with 0=B 0. We need a well localized strong magnetic field
perpendicular to the neutron beam, and inside a smaller coil perpendicular both to
the neutron beam and to the strong field. This small coil has to be driven by a
radiofrequency generator which will produce the necessary (resonance) rotating
field. In between these flippers a strictly zero field region has to be insured to avoid
spurious spin precession. The first prototype spectrometers all use mu-metal
shielding + earth field compensation.
Multidetector variants
At the time of writing, four instruments are developed to extend the classical
NSE to use multidetectors, and thus reducing the data acquisition time. In all cases
the problem is that useful information can be collected in all detectors only if the
echo condition can be fulfilled simultaneously for all of them. This is not a
straightforward exercise as the field integrals must be equal with a precision of
10 -5! This strict condition can be somewhat relaxed, if we can have enough
information in each individual detector to make the phase correction on a one-byone basis. This is not always possible if the scattering is weak. In that case, on an
elastic scatterer, the relative phase differences between detectors can be measured
and imposed during the data treatment.
The most straightforward implementation of such a multidetector is on the
instrument IN15 at the ILL and a somewhat later construction in KFA Jlich. Here

2.8-15
simply the solenoid diameters were increased to allow the use of a small 32cm x
32cm multidetector. The difficulty is to make a sufficiently big Fresnel coil with the
necessary precision, without absorbing a substantial fraction of the neutron beam.
On IN11 an option has been constructed, to replace the secondary solenoid with a
specially shaped sector magnet (Fig. 12). The expected field homogeneity is
inherently lower, but in many cases in the high q region the best possible resolution
is not needed anyway. The gain factor in total countrate is however 23 time higher.
This gain is sufficient to measure even the very unfavorable incoherent scatterers.

Figure 12: The IN11C

multidetector option.

In Berlin a totally new approach was initiated. If one takes a Helmholtz like
coil, but the two half generates a field opposite to each other, then in the equatorial
plane the field is guaranteed to be rotationally symmetric. The neutron beam comes
in along a radius and the detectors are placed all around. The practical problem here
is a zero field point at the sample position which would depolarize the beam. The
solution proposed is to move this zero point above the sample with the help of some
additional coils.

2.8-16
Useful numbers
Some numbers from the ILL spin-echo instruments are given in the table.
Table 1:

The miminum time can be shorter by a factor 3-5 if double echo [2] configuration is
used. The maximum reachable Fourier times can be shorter as much as a factor two
when the detector is positioned at high scattering angles.
References
[1] Mezei,F. (editor) Neutron Spin Echo Lecture Notes in Physics 128 SpringerVerlag 1979
[2] B. Farago Annual Report ILL 1988 p: 103
[3] R.Ghler, R.Golub Z.Phys, B65 (1987) 269.
Aknowledgement
The author had the chance to learn most of his knowledge from F. Mezei.

2.9-1

Neutron Diffraction on Reactors

A.W. Hewat and G.J. McIntyre

1. Introduction
The efficiency of a neutron diffractometer, whether on a reactor or a spallation
source, depends essentially on the time-averaged flux on the sample and the solid
angle of the detector. Large new detectors and focusing neutron optics are
increasingly being used to take advantage of the high flux on the sample that is
obtained with a continuous neutron source. In this article, we consider the design of
single-crystal and powder/liquids diffractometers, especially those using white beams
and large area detectors, for different types of experiment. We give the essential
formulae needed to extract structure amplitudes from the observed intensities.
2. The design of modern reactor-based diffractometers
The Shelter Island Workshop1 proposed that the intensity of a neutron diffractometer,
whether on a reactor or a pulsed neutron source, depended only on the time-averaged
flux on the sample, the sample volume, and the solid angle of the detector:
I ~ flux*sample*detector
2.1. The time-averaged flux on the sample
On a reactor, the time-averaged flux on the sample can be greatly increased by
using a wide band of wavelengths, without reducing the instrument resolution for a
particular angle the focussing angle where the diffracted Bragg angle equals the
monochromator Bragg angle (fig.1). For example, even on a high-resolution
powder diffractometer such as D2B at ILL2, the relative spread in wavelength /
can be 1% or more for a d/d resolution of 0.1%.
An extreme case of the use of a wide band of wavelengths is the quasi-Laue
diffractometer such as LADI3 and VIVALDI4 at the ILL. A very large position-sensitive
detector (PSD), here a set of neutron-sensitive image plates, collects an almost white
beam of neutrons falling on the sample, with the different wavelengths sorted out, not
by time-of-flight (TOF) or by monochromator/analyser crystals, but by the diffraction
conditions of the single-crystal sample itself (fig.3).
A continuous neutron intensity with a wide band of wavelengths, and the resulting
high time-averaged flux on the sample, is the principal advantage of reactor
diffractometers. On a TOF diffractometer, the flux on the sample is also rather constant

2.9-2

Figure 1. The focusing of a wide band of wavelengths on the D2B highresolution diffractometer. A large detector collects many simultaneous
reflections from the polycrystalline sample.
in time, since the sharp initial pulse is spread out at the sample such that the slowest
neutrons arrive just before the fastest from the next pulse. But the time-averaged flux
may be an order of magnitude lower; relatively long flight paths and a / of 0.1% are
needed to obtain a resolution d/d of 0.1%. Fortunately, the solid angle of the TOF
detector for high scattering angles (backscattering) can be large while still allowing high
resolution, since d/d ~.cot and cot is small for large . This can compensate for
the intrinsically lower flux on the sample of the TOF diffractometer, supposing that the
sample volumes are similar.
2.2 The sample volume and resolution
The sample volume limits the resolution that can be obtained with a PSD, and must therefore be considered in the design of a diffractometer. For example, the D20 instrument at
ILL uses a large PSD with an angular resolution of 0.1 at a distance of 1470 mm from the
sample. The optimal 0.1 resolution would then require a sample of less than 2.57 mm
diameter! Larger samples are normally measured on D20, with less than optimal resolution. Higher resolution could of course be obtained by increasing further the sample-detector distance, but then a very tall detector would be needed to conserve the solid angle.

2.9-3

Figure 2. The first diffraction pattern (1984) from the D2B high-resolution powder
diffractometer2. The peak intensity is rather constant over the entire d-spacing
range, unlike with X-rays where scattering falls off with angle, or with time-of-flight
techniques where intensity falls for shorter d-spacings or wavelengths.

Figure 3. The dispersion of the white band of wavelengths on the quasi-Laue

diffractometer VIVALDI. The large PSD collects many simultaneous reflections
from the single-crystal sample, where each reflection corresponds to a different
wavelength.
Alternatively, high resolution can be obtained with Sller collimators as fine
as 5 as on D2B, and then much larger samples can be used, with sample volume

2.9-4
and neutron intensity increasing with the square of the diameter. Sller collimators
also limit the detector solid angle, since even with 128x5 collimators, only 6.7% of
the 158.5 scattering angle is covered, but a sample as large as 16 mm in diameter
can be used, together with tall linear-wire detectors to increase the solid angle.
Until very large, high-resolution PSDs can be constructed, the D2B-type instrument is then more suitable for very high resolution, while the D20 instrument is more
suitable for very high intensity. There is of course a degree of overlap in application,
depending on the amount of sample actually available, and the particular problem.

Figure 4. The resolution or Full Width at Half Maximum of peaks is best at the
focussing angle (140o), which is high so as to match the FWHM required to
resolve adjacent peaks for example, a 16 cubic cell2.
The resolution is worse at angles far from the focussing angle, but by choosing a
large focussing angle the instrument resolution can be made to match the spacing
between Bragg peaks (fig. 4). There are few peaks at very low or very high angles, and
if necessary the neutron wavelength can be increased to ~6 to move reflections with
large d-spacings to higher angles for better resolution. This situation is different in the
case of the backscattering TOF diffractometer, where the resolution is equally good for
all d-spacings, but modern TOF instruments such as GEM at ISIS also use lower-angle
detectors to increase efficiency and collect data with a variety of resolutions.

2.9-5

Figure 5. The new 30 x 120 2D multi-wire PSD for D19 at ILL Grenoble. Such
large detectors match the solid angle of the best pulsed-neutron source
diffractometers, while also benefiting from the very high time-averaged neutron
flux available on the sample at a reactor source.
2.3 The detector solid angle
When only small samples are available, as in pressure cells, or when medium
resolution is sufficient, the PSD on a reactor can be as large or larger than that on a
pulsed neutron source. Figure 5 shows the new 2D PSD being constructed at ILL for
D19. This detector covers 30 vertically by 120 horizontally, covering a solid angle of
4
.sin(15o)x120/360=1.1 steradians, a record for electronic detectors.
The D19 detector is being built for measurements on single crystals and fibres,
but would also be good for fast powder or liquid diffraction, with an angular
resolution of 11 (2.5 mm wide elements at 760 mm). Using a 160o detector, a
Diffractometer for Rapid Acquisition over Ultra-Large Angles (DRACULA) might
be constructed to compete with the fastest diffractometers on the future SNS pulsed
source. (See table below.)
Time averaged sample flux
Sample volume
Detector solid angle

ILL-D20
5x107
1
0.27 sr

ILL-DRACULA
~108
1
1.45 sr

US-SNS
~2.5x107
1
3.0 sr

Table 1. A comparison of possible high-intensity powder diffractometers on

the ILL reactor and US spallation source, showing that the high time-averaged
flux at the sample, together with a D19-type PSD detector, should allow the
reactor machine DRACULA to compete with the best pulsed source.

2.9-6

Figure 6. The first neutron image-plate diffractometer LADI, developed at

EMBL-ILL Grenoble, showing the image plate (1) on a rotating drum (2)
surrounding the crystal on its supporting rod (3). The neutron beam enters
and exits through the small holes in the middle of the drum, and the remaining
components allow the laser and photo-multiplier tube (7-9) to scan the surface
of the image plate to read off the diffraction pattern.
Reactor sources look even more attractive when continuous white beams of
neutrons can be used, as with the quasi-Laue technique on image-plate detectors such as
those of LADI and VIVALDI at ILL (fig.6). Then the time-averaged flux on the sample
is very much greater than can be obtained either with monochromatic or pulsed neutron
techniques. As well, the image-plate detector covers a large solid angle (7.9 steradians);
image plates can be curved over a cylinder around the sample, while still being read out
electronically. These diffractometers are at present very new, and could be even further
improved by using super-mirror optics to increase the flux on the sample.
3 The resolution function in diffractometry
The 3-D resolution function of a diffractometer depends on many factors,
principally the incident-beam or incident-guide divergence, the monochromator
mosaic, the sample size and mosaic, and the scattering geometry, and is

2.9-7
conveniently measured or calculated as an ellipsoid in terms of differential
increments, , and , in the observational variables, , and . Following
(for the most part) the definitions of Busing & Levy5, we describe the crystal
orientation by the Euler angles , and . To allow for out-of-plane reflections we
describe the direction of a particular diffracted ray by and , the in-plane and outof-plane angles, respectively, on the surface of the Ewald sphere. 2 remains the
angle between the incident and diffracted beams, so that,
cos2 = coscos.

(1)

In single-crystal diffraction the resolution in is usually considerably better

than that in , which in turn is considerably better than that in .
The derivation of the full 3-D resolution function is straightforward but lengthy.
Since the landmark papers by Nathans and Cooper6, there have been many derivations
for various instrumental geometries. These usually derive the function in terms of
Gaussians, which facilitates the numerous convolutions particularly in three-axis
geometry with its multitude of collimators, but can disguise the effect of individual
parameters. For diffraction, Schoenborn7 derived explicitly the direction in the multidimensional function of the contribution from each of the various instrumental
parameters. The different contributions generally act along different directions,
knowledge of which may indicate how to optimise a particular measurement.
, and can be converted to reciprocal-space coordinates over the range of
interest for one reflection by a homogeneous affine transformation; for powder
diffraction (2) is then related to d*. Going to shorter wavelength () decreases
the resolution in the reciprocal directions corresponding to and . The increasing
use of PSDs with their capability to resolve single-crystal reflections in 3-D means that
we can tolerate a shorter wavelength if this is needed to give better real-space resolution
(dmin or Qmax). The principal rule, though, is to use as long a wavelength as possible to
profit from the 4 dependence of the product of the reflectivities of the monochromator
and the sample. Fig. 7 shows some resolution characteristics of the ILL single-crystal
diffractometers for chemistry and physics.
3.1 Conventional crystallography
The aim of the experiment is to measure the integrated intensities of many
reflections, even for incommensurate and magnetic structures. We are generally
interested in nearly equal resolution in all directions of reciprocal space, so the
choice of diffractometer is determined by which of 2, and is largest,
which is usually . A corollary is that instruments intended for conventional
crystallography usually avoid the use of vertically focussing monochromators.

2.9-8

Figure 7. Different ILL single-crystal diffractometers optimised for different

) versus Q (= 4sin
/
) for typical samples and
science: full-width at half-height (
choice of monochromators, and the approximate relative intensities. More intensity
is available from other optional monochromators, at the expense of resolution.
3.2 Phase transitions and critical scattering
Here the aim is often just to scan repeatedly a few reflections or a limited region of
reciprocal space while varying the temperature, magnetic field or pressure.
Concentration of the time-averaged flux into a small wavelength band clearly gives the
advantage to the monochromatic single-crystal diffractometer. The adaptability of
monochromatic diffractometers means that we can usually orient the sample to put the
reciprocal-space direction of interest most nearly along , and we can even accept a
large if the flux on the sample is increased by a vertically focussing
monochromator. One important consequence of the affine transformation between ,
and , and orthogonal reciprocal-space coordinates is that there is one orientation
(up or down) of a given reciprocal-lattice plane that gives best resolution along a given
line in that plane 8. The orientation is usually that which gives the larger | +
cos| between q steps along the line of interest.

2.9-9
3.3 Laue diffraction
The spatial resolution of the reflections is given by the projection of the crystal
shape onto the image plate, convoluted with the divergence of the incident beam,
the mosaic spread of the crystal (negligible for a crystal of good quality), and the
point spread function of the image-plate and scanner. The divergence of the
incident beam in a curved thermal guide is typically 3.5 mrad full width at half
height, which gives ~0.25 at = 1.5 , a typical wavelength of maximum flux on
a thermal guide. In terms of the angle between reciprocal-lattice directions Q1 and
Q2 the spatial resolution becomes,
Q1Q2 = w/(2R) * * * p

(2)

where w is the cross-section of the crystal perpendicular to the diffracted beam, R is

the sample-to-detector distance, is the beam divergence, is the mosaic spread,
and p is the point spread function. (The factor 1/2 is due to the relation :2
between the angle between scattering vectors and the angle between the
corresponding diffracted beams.) The spatial point-spread function of an imageplate detector is usually negligibly small (~150 m). Thus for a detector of radius
160 mm and a typical crystal dimension of 2 mm, the spot diameter is about 0.6 in
reciprocal space. The resolution is dominated by the crystal size, and for a spherical
sample is constant over the whole Laue diffraction pattern, and over all orientations
of the crystal. Allowing for twice the spot diameter between reflection centres (to
allow confident estimation of the background) we can resolve reflections up to
order 25 for a primitive cubic unit cell. A reflection of order 25 at 2 = 180 at a
wavelength of 1.5 corresponds to a unit-cell edge of 20 .
4 Structure amplitudes from integrated intensities
The reduction of the observed integrated intensities to structure amplitudes (Fhkl2) is
the same as for X-ray diffraction, except that no polarisation factor is required for
unpolarised neutron beams. For the three diffraction methods that we discuss here, the
total integrated intensity of one reflection is related to the structure amplitude by 9:
a) powder sample in a monochromatic beam
Ihkl = m Io(o) o V N2 |Fhkl|2 o3 [1/(4sin)] A* T E n s-1

(3)

b) rotating single crystal in a monochromatic beam

Ihkl = Io(o) o V N2 |Fhkl|2 o3 (1/) L A* T E n s-1

(4)

c) fixed single crystal in a polychromatic beam (Laue technique)

Ihkl = Io() V N2 |Fhkl|2 4/2sin2 A* T E n s-1

(5)

2.9-10
where m is the multiplicity factor for the powder reflection, the wavelength, V the
sample volume, N the number of unit cells per unit volume, the angular scanning
velocity, L the Lorentz factor which is related to the time-of-reflection opportunity,
A* the absorption correction, T the correction for inelastic phonon or thermaldiffuse scattering, and E the correction for extinction and multiple diffraction. To
obtain |F hkl| involves measurement of Ihkl as accurately and as efficiently as
possible, followed by application of the various correction factors. For most
powder experiments, the corrections for absorption, thermal-diffuse scattering,
extinction and multiple scattering can be ignored.
The principal disadvantage of the steady-state Laue technique is that it
integrates over all wavelengths, which gives a particularly high background if the
sample contains elements that give significant incoherent scattering. The
incoherently scattered neutron background is,
________
Iinc = Io ()inc (Vs/Vc) (1/4) n s-1 sterad-1,
(6)
where is the wavelength band-pass and inc is the total incoherent crosssection of the atoms in one unit cell, arising principally from hydrogen, which often
constitutes 50% of organic samples and inorganic hydrates. Here, Io ()inc is
averaged over the band-pass. The different units for Ihkl and Iinc emphasize the
advantage of small guide divergence to reduce the point-by-point signal-to-noise
ratio of the Laue diffraction pattern.
4.1 The generalised Lorentz factor
The expressions for Ihkl differ principally in the Lorentz factors. The Lorentz
corrections for monochromatic powder diffraction and for single-crystal Laue
diffraction are relatively simple since the sample does not rotate. Buras & Gerward 9
show how the Lorentz factors for the three techniques are related when the
monochromatic single-crystal measurement is made by an rotation in the equatorial
plane. With a large PSD, reflections may also be observed out of the equatorial plane,
for which the Lorentz factor for an rotation becomes:
L = 1/(sincos).

(7)

For some crystal structures, particularly those that are incommensurate in their
nuclear or magnetic structures, it might be preferable to perform linear scans in
reciprocal space rather than pure rotational scans. Here linear refers to the
trajectory of the detector aperture through reciprocal space; for a particular
reflection observed in such a scan the motion of the corresponding reciprocallattice point through the Ewald sphere is still along an angular trajectory. The

2.9-11
Lorentz factor is conventionally the inverse of the wavelength-independent part of
the term dz/ds that relates the scan variable s to the normal z to the Ewald sphere in
the integral of the observed counts over s:
|Fhkl|2 Ii(z) dz = Ii(s) (dz/ds) ds = i (dz/ds) si Ii(si).

(8)

Since the interval in s over which Ii is non-zero is usually small, (dz/ds) is usually
evaluated at the peak maximum and taken outside the integral. However for linear
scans, rather than evaluate dz/ds and then perform the integration over s, it is
computationally more convenient to evaluate the (ds/dz)si directly in terms of the
incremental angular shifts, i, i and i, that effect the steps s i. If the
reflection diffracts at the setting , , and 10,
(ds/dz)si = -[isincos + isinsin
+ i(cossincos cossinsin)].

(9)

Although this expression does not in general permit a separation of (ds/dz)si

into an incremental scan step and a Lorentz factor, it does correct quite generally for
the different velocities of different reciprocal-lattice points through the Ewald
sphere whatever the scan mode. |Fhkl|2 for reflections observed in linear reciprocalspace scans are thus brought to the same scale as |Fhkl|2 for reflections observed by
rotational scans.
4.2 Absorption correction
When a monochromatic beam of neutrons with incident intensity Io passes through
a homogenous plate of uniform thickness t the intensity is reduced to,
I = Io exp(-t),

(10)

which defines , the total linear absorption coefficient. To a reasonable

approximation the mass absorption coefficient /, where is the density of the
absorber, is independent of the physical state of the absorber, and / for a
compound is additive with respect to the / of its elements. This gives the
practical working expression 11:
= (n/Vc) ii

(11)

where n is the number of molecules in the unit cell, Vc is the unit-cell volume, i is the
atomic absorption coefficient of atom i, defined as (/)i (A/N) where A is the atomic
weight and N is Avogadros number, and the summation is over all atoms in one molecule.
For neutrons in the wavelength range of interest, is the sum of two terms: i)
true absorption, which is due to nuclear capture processes; and ii) apparent

2.9-12
absorption due to scattering, both coherent and incoherent. For single crystals the
absorption due to coherent scattering is usually considered separately as extinction
and multiple diffraction. Nuclear capture absorption varies as , while the
incoherent apparent absorption is independent of wavelength for most elements. A
notable exception is hydrogen, for which the dependence of inc on wavelength is
roughly linear 12,
inc = 19.2(5) + 20.6(9),

(12)

and is usually the dominant contribution to for hydrogenous materials. Sears 13 lists
the values of the true absorption (abs) and incoherent absorption (inc) cross-sections
required to calculate (= abs /1.798 + inc) for neutron experiments (reproduced
and updated in the first article in this booklet). For measurements near nuclear
resonances, it is prudent to determine the absorption coefficient experimentally.
For diffraction from a crystal bathed entirely in the beam, eq. (10) applies to
each infinitesimal volume element with t being the total path length of the incident
and diffracted beams within the crystal for that volume element. The absorption for
the whole crystal is then obtained by integration over all volume elements
I = Io (1/V)V exp(-t)dV = Io A

(13)

to give the transmission coefficient A, which is the inverse of the absorption

correction A*. Analytical integration is feasible for crystals of very regular shape
such as spheres and cylinders 11. For faceted crystals numerical integration is the
usual technique 14. These equations and methods also apply to the Laue technique,
with the wavelength different for each reflection, and to very well compacted
powder samples.
There is further absorption along the incident and diffracted beam paths
outside the crystal, the absorption in each substance, be it air, sample-capillary
walls, cryostat walls etc. also given by eq. (10). If the path length through any of
these varies significantly with crystal orientation or detector position a correction
for that absorption may need to be applied as well.
4.3 Extinction
Correction for extinction is made as in X-ray crystallography, usually following the
formulae of Becker & Coppens 15, who include the crystal form via the mean
absorption-weighted path length:
T = (1/VA)V t exp(-t)dV

(14)

which is calculated in a similar manner to the transmission factor. To check the

2.9-13

(a)

(b)

Figure 8. Reciprocal-space constructions to find the accessible regions for: (a) a

vertical cryomagnet with a single 25 blind arc; and (b) one of the ILL horizontal
cryomagnets which has restricted openings of 15, 32, 30 and 32 in the equatorial
plane. The heavy outer arcs denote the blind angles of the cryomagnets, the open
areas the accessible regions of reciprocal space, the shaded areas the inaccessible
regions, and the black dots represent a possible reciprocal lattice plane.
corrections for absorption and extinction, it is recommended to make integrated
intensity measurements of one or two reflections over a wide range of azimuthal angle
(rotation around the scattering vector). After correction for absorption and extinction,
the observed |Fhkl|2 for each reflection should be independent of the azimuthal angle.
5 Accessible regions of reciprocal space in extreme sample environments
The accessible region of reciprocal space for single-crystals in sample
environments with limited access is found by drawing the Ewald sphere when the
limit of each accessible window is in the incident beam. For the vertical and
horizontal cryomagnets often used for diffraction experiments at the ILL, the
accessible regions are as shown in Fig. 8. For a sample environment with very
restricted access the wavelength can often be varied to bring the reflection(s) of
interest into an accessible zone.
References
1. Jorgensen, J.D., Cox, D.E., Hewat, A.W., Yelon, W.B. (1985) Nuc.Inst.Meth.
B12, 525-561.
2.

Hewat, A.W. (1986) Mat.Sci.Forum 9, 69-79.

3.

Cipriani, F., Castagna, J.-C., Wilkinson, C. Oleinek, P., Lehmann, M.S (1996)

2.9-14
J. Neutron Res. 4, 79-85.
4.

Wilkinson,C., Cowan,J.A., Myles,D.A. A., Cipriani,F., McIntyre,G.J. (2002)

Neutron News 13, 37-41.

5.

Busing, W.R., Levy, H.A. (1967) Acta Cryst. 22, 457-464.

6.

Cooper, M.J., Nathans, R. (1968) Acta Cryst. A24, 619-624, 624-627.

7.

Schoenborn, B.P. (1983) Acta Cryst. A39, 315-321.

8.

McIntyre, G.J., Renault, A. (1989) Physica B 156 & 157, 880-883.

9. Buras, B. Gerward, L. (1975) Acta Cryst. A31, 372-374.

10. McIntyre, G.J., Stansfield, R.F.D. (1988) Acta Cryst. A44, 257-262.
11. International Tables for X-ray Crystallography, Volume III (1968), pp 157-200.
12. Howard, J.A.K., Johnson, O., Schultz, A.J., Stringer, A. (1987) J.Appl.Cryst.
20, 120-122.
13. Sears, V.F. (1992) Neutron News 3, 26-37.
14. Coppens, P., Leiserowitz, L., Rabinovich, D. (1965) Acta Cryst. 18, 1035-1038.
15. Becker, P.J., Coppens, P. (1974) Acta Cryst. A30, 129-147.

3.1-1

TECHNIQUES
The Production of Neutrons
C. J. Carlile

1. THE PRODUCTION OF NEUTRONS

1.1 Natural Radioactive Sources
Neutrons can be produced by fusion processes in stars, by spallation processes of
cosmic rays in the atmosphere and by the process of spontaneous fission. Other than
,n) reaction however, with which
that there is no natural source of neutrons. The (
Chadwick was first able to isolate and identify the neutron in 1932, is the nearest. In
certain light isotopes the last neutron in the nucleus is weakly bound and is released
when the compound nucleus formed following -particle bombardment decays.
Chadwick made use of the naturally occurring -emitter polonium-210 which

Figure 1: The discovery of the neutron in 1932 by Chadwick followed

experiments of particle bombardment of beryllium.

3.1-2

Figure 2: The cross-section through a portable laboratory neutron source

containing radium and beryllium. A typical source is about 1cm in diameter.

decays to lead-206 with the emission of a 5.3 MeV -particle. The bombardment of
beryllium by -particles leads to the production of neutrons by the following
exothermic reaction which is illustrated schematically in Figure 1
He4 + Be9 C12 + n + 5.7 MeV
This reaction yields a weak
source of neutrons with an energy
spectrum resembling that from a
fission source and is used nowadays in
portable neutron sources of the kind
commonly used for setting up neutron
detectors. Radium or americium is
normally the -emitter in the typical
design shown in Figure 2.
(,n) sources can also be used
for the same purpose. In this type of
source, because of the greater range
of the -ray, the two physical

Figure 3: A portable neutron source using a emitter

which by removal of the source can be switched off.

3.1-3

Figure 4: A schematic representation of the fission reaction.

components of the source can be separated making it possible to switch off the
reaction if so required by removing the radioactive source from the beryllium. The
source illustrated in Figure 3 uses antimony-124 as the -emitter in the following
endothermic reaction.
Sb124 Te124 + +
+ Be9 Be8 + n - 1.66 MeV
1.2 Fission
Uranium-235 which exists as 0.7% of naturally occurring uranium undergoes
fission with thermal neutrons with the production of, on average, 2.5 fast neutrons
and the release of ~ 180 MeV of energy per fission. In a critical assembly the
fission reaction becomes self-sustaining with 1 neutron required to trigger a further
fission, 0.5 neutrons being absorbed in other material and 1 neutron able to leave
the surface of the core and be available for use. The reaction is shown
schematically in Figure 4.
nthermal + U235 2 fission fragments + 2.5 nfast + 180 MeV
If we take the example of a 10 MW research reactor then 107 joules/sec are
released which is 3.3 x 1017 fissions/sec at 180 MeV/fission which gives 8.5 x 1017
neutrons/sec released in the whole reactor volume. The resultant neutron energy
spectrum of a reactor source is shown schematically in Figure 5.
The spectrum is composed of three distinct regions:
The first is the fast neutron region where the neutrons are produced. This is
referred to as a Lamb distribution with the peak intensity occurring between 1 and 2

3.1-4

Figure 5: The various components in a thermal reactor spectrum (intensities

not to scale).
MeV which can be described mathematically by the expression
(E)dE = f exp (-E sinh

2E ) dE

E>0.5MeV

The intermediate energy region is known as the slowing down or epithermal

region and is characterised by a 1/E intensity distribution. Thus
(E)dE =

epi
E

dE

200 meV < E < 0.5 MeV

describes the spectrum in this region where the source neutrons are losing energy in
the moderation process.
At low energies the neutron spectrum tends towards thermodynamic
equilibrium with the moderator as the neutrons, acting like a gas, both lose energy
and gain energy in collisions with nuclei in the moderator. The resultant spectrum is
described by a Maxwell-Boltzmann distribution with an effective temperature Tn ~
300 K, always somewhat higher than the physical moderator temperature since full
equilibrium is never achieved in a finite sized moderator.
E

(E)dE = th kT exp kT dE
E < 200 meV

n
n
The Maxwellian peaks in intensity at an energy of about 25 meV for a room
temperature moderator.
A stylised representation of the basic components of a research reactor, a
swimming pool reactor, is shown in Figure 6. The uranium fuel is normally contained

3.1-5

Figure 6: A schematic view of a research reactor showing the essential components.

Natural
Uranium

Gamma
Flux

Thermal
Flux
th
n/cm2/sec

Epithermal
Flux
epi
n/cm2/sec

Fast
Flux
f
n/cm2/sec

/cm2/sec

6.1013

2.1012

4.1012

2.1012

2.1014

2.1013

2.1014

1.1013

+ D2O
Enriched
Uranium
+ H2O
Table 1: Typical values of neutron and gamma fluxes for the two most
common fuel/moderator combinations in a 20 MW reactor.

3.1-6

Figure 7: (a) view of the

beam tubes and the cold
and hot moderators of
the ILL reactor. (b) The
fluxes as a function of
distance from the centre
of the core are shown in
the lower graph.

in a number of fuel rods (although there is only one at ILL) and the critical reaction is
controlled by neutron absorbing control rods (normally boron loaded). The moderator,
which can be H2O or D2O depending on the enrichment of the fuel, is often also
employed as the reactor coolant. A massive radiation shield of borated concrete and
steel surrounds the reactor which protects both experimenters and instruments.
The use of natural uranium fuel with its inherently low fissile isotope content
requires the use of heavy water D2O as the moderator since its absorption cross
section is low compared to light water H2O. When fuel enriched in the U235 isotope
is used, then the more highly absorbing H2O moderator can be used. Typical values
of the flux components for a 20 MW research reactor are shown in table I for both
combinations of fuel and moderator.

3.1-7

Figure 8: The principle behind a laboratory source of neutrons using

accelerated deuterons onto a tritium loaded target.
Clearly, whilst the source strength of both reactors illustrated in the above
table is the same (1.7 1018 n/sec), the flux is considerably higher in the enriched
uranium/H2O case. This arises simply because of the smaller core and consequent
higher power density of this combination of fuel and moderator. The use of
enriched uranium as the fuel presents its own special problems for reactor operation
however from the political point of view.
The principal object in designing a neutron beam reactor is to deliver the
maximum neutron flux at the required energy at the neutron scattering instrument.
Accordingly the noses of beam tubes are located in the region of maximum neutron
intensity which can be designed to occur just outside the reactor core.
This is achieved by under-moderating the core which delivers a high fast
neutron distribution to the surrounding reflector, where the process of moderation
raises the thermal flux to a maximum. This can also be achieved for other neutron
energies by locally inserting blocks of moderating material at higher or lower
temperatures. This is most notable in the case of liquid hydrogen or deuterium cold
sources. The distribution of beam tubes and the various components of the neutron
spectra around the high flux reactor at the Institut Laue Langevin in Grenoble is
shown in Figure 7 as an illustration. The thermal flux peaks at ~15 cm from the
edge of the reactor core.

3.1-8

Figure 9: A schematic representation of an electron linear accelerator which

produces pulsed neutron beams.

1.3 Particle Accelerator Sources

Neutrons can be generated by bombarding a target with high energy particles
produced with an accelerator. Depending on the type of accelerator the neutrons
will be produced continuously or in bursts. A number of types of reaction have been
employed, of which the following are typical:
(D,T) Fusion
Neutrons are produced in the fusion of deuterium and tritium in the following
exothermic reaction.
D2 + T3 He4 + n + 17.6 MeV
The neutron is produced with a kinetic energy of 14.1 MeV. This can be achieved
on a small scale in the laboratory with a modest 100 kV accelerator for deuterium atoms
bombarding a tritium target. Monatomic or nascent deuterium is produced by bleeding
D2 gas through a heated palladium tube. The accelerated D+ ions bombard a titanium or
zirconium target loaded with tritium in the form of the hydride TiT2 or ZrT2 as shown in
Figure 8. Before target heating becomes a problem, continuous neutron sources of
~1011 neutrons/second can be achieved relatively simply. This type of neutron source is
often used in reactor physics simulation experiments in the laboratory, for example to
measure neutron thermalisation in graphite.
Note that this reaction is the same as that which is being exploited in the
development of thermonuclear fusion reactors such as JET. In this case the energy
to initiate the binding of the deuterium and tritium atoms is derived from the
required high temperature of the plasma.

3.1-9
Bremsstrahlung from Electron Accelerators
Energetic electrons when slowed down rapidly in a heavy target emit intense radiation during the deceleration process. This is known as Bremsstrahlung or braking
radiation. The interaction of the -radiation with the target produces neutrons via the
(,n) reaction, or the (,fission) reaction when a fissile target is used.
e- Pb Pb (,n) and (,fission)
The Bremsstrahlung energy exceeds the binding energy of the last neutron in
the target. This reaction is comparable to the (,n) reaction in Sb124 referred to above. A
neutron source such as this is achieved with a moderately large purpose-built electron
linear accelerator or linac which produces electrons in bursts at frequencies of
between 25 Hz to 250 Hz at energies ~150 MeV. Examples of target materials are
tungsten, lead or depleted uranium. Such a source is illustrated in Figure 9.
A source strength of 1013 neutrons/second produced in short (i.e. < 5 s)
pulses can be readily realised. This source strength is adequate for specialised
neutron scattering measurements as for example was done at the Harwell, Toronto
and Hokkaido linacs, and also at the Frascati cyclotron source LISONE.
Spallation from Energetic Protons
The process of spallation is a common nuclear reaction occurring for high
energy particles bombarding heavy atoms. The reaction occurs above a certain
energy threshold for the incident particle, which is typically 5 - 15 MeV. The

Figure 10: A schematic representation of the spallation reaction.

3.1-10

Figure 11: The spectrum

of neutrons emitted
from heavy metal
targets by the process of
spallation indicating the
difference when fertile
material is part of the
target.

reaction, illustrated in Figure 10 is a sequential one involving the incorporation of

the incident particle (say a proton) by the target nucleus followed by an internal
nucleon cascade, an internuclear cascade where high energy neutrons are ejected,
and an evaporation process where the target nuclei de-excite by the emission of
several low energy neutrons and a variety of nucleons, photons and neutrinos.
The reaction is multifarious and in general involves several different target
nuclei. It is often likened to a cannon ball careering through a wooden-hulled ship
although it is a geological term representing the splitting of rocks. In the spallation
reaction 20 to 30 neutrons per incident particle can typically be generated. The energy
released per neutron produced is quite low being ~55 MeV again dependent upon the
incident particle energy and the target nucleus, and whether or not fission processes
play a part. The neutron source strength can be significantly increased by using a
fissile target, normally depleted uranium. The source spectrum from an 800 MeV
proton spallation reaction on both a lead target (with no fission) and a uranium-238
target (where fast fission contributes to the neutron intensity) is shown in Figure 11.
The cascade processes account for only ~3% of the source neutrons but since

3.1-11
they are produced at energies up to the energy of the incident protons these extremely
penetrating neutrons dominate the shielding requirements of the source.
In practice a spallation source is usually realised with accelerated protons.
These can be produced in a number of ways, for example:
1. Linear Accelerators such as LAMPF at Los Alamos which are high current, high
duty cycle accelerators resulting in either long pulses or at frequencies too high to
be utilised effectively in neutron scattering instruments. Accordingly particle
storage rings, which compress the long pulses, are needed.
2. Cyclotrons such as SINQ at PSI near Zurich which produce a continuous
beam of neutrons via the spallation reaction, and
3. Synchrotrons such as ISIS in the UK. Synchrotrons have operated with low
currents until relatively recent times with the implementation of rapid cycling
techniques and weak focusing magnets. Narrow neutron pulses (< 1 s) can be
produced, due to the single turn proton beam extraction method, at a modest duty
cycle (50 Hz) which is well-suited to the neutron scattering instrumentation. A
typical schematic layout of a synchrotron-based pulsed spallation neutron source
is shown in Figure 12. Multi-turn-injection into the synchrotron is achieved using
more modest linear accelerators for negative H- ions. At the point of injection the
beam passes through a thin electron stripper foil such as alumina and the resultant
protons enter the synchrotron for further acceleration. A neutron source strength of
~ 5 1016 fast neutrons/second can be produced in a pulsed manner by this method.

Figure 12: The principle of operation of a spallation neutron source using a

combination of linear accelerator and synchrotron accelerator.

3.1-12
2. MODERN NEUTRON SOURCES
2.1 The Quest for Higher Intensities
Neutron scattering techniques suffer from low data rates particularly in
comparison with x-ray diffraction or infra-red spectroscopy. Accordingly much
effort has been concentrated on improving the available intensities of neutron
sources. In Figure 13, originally due to Carpenter, the effective thermal neutron
flux of various neutron sources has been plotted chronologically. The precise
positions of various points on this graph and the use of average flux for continuous
sources and peak flux for pulsed sources has been and still is the subject of debate,
without even mentioning the multitude of questionable assumptions made in
attempting to compare such complex facilities using only one variable.
Nevertheless the visible trends are genuine and quite clearly fission reactors,
having improved rapidly in the five years following the success of the first critical
assembly at Chicago in 1942, have reached a slowly increasing quasi-asymptote
around 2 1015 n/cm2 sec. Technical advances in instrumentation, for example
neutron guides, focusing monochromators and area detectors have however
ensured a steady rise in data acquisition rates. Pulsed sources on the other hand
have not yet reached saturation of the source flux and much potential remains to be
realised with this type of neutron source before such limits are approached. The
most advanced spallation source currently under construction is the 1MW SNS

Figure 13: The effective neutron flux of various neutron sources from the
discovery of the neutron in 1932 until the present day.

3.1-13

Figure 14: A representation of the layout of the proposed European Spallation

Source with two shared pulse targets. The current design includes a long pulse
target fed directly from the Linac.
(spallation neutron source) at Oak Ridge in Tennessee which is scheduled to
generate its first neutrons in 2006. A design study has just been completed for a 10
MW European Spallation Source which promises source fluxes 30 times those of
the 160 kW ISIS source near Oxford. The layout of the ESS is shown in Figure 14.
An important limiting factor in determining the maximum neutron output of a
particular type of source is the rate of removal of the heat deposited in the target by
the nuclear reaction. Table II shows the energy released per useful neutron in the
various reactions discussed earlier. Spallation releases 3 times less energy than
fission which in turn releases 10 times less energy than photoneutron reactions
(Bremsstrahlung). Controlled thermonuclear fusion reactors offer a future promise
of neutrons for yet lower releases of energy, and consequently the potential for
higher intensities.
If additionally the neutron generating reaction is pulsed in nature, the heat
deposited in the target can be substantially reduced (perhaps by a factor of 20)
compared with a continuous source of the same equivalent source strength.

3.1-14
Process

Example

Neutron Yield

Energy Released
MeV / neutron

1. (,n) reaction

Radium-beryllium
Laboratory source

8 10-5 / alpha

6,600,000

2. (D,T) fusion

400 keV deuterons

on tritium-loaded
titanium

4 10-5/deuteron

10,000

3. Electron
Bremsstrahlung
& photofission

100 MeV electrons

on uranium

5 10-2/electron

2,000

4. Fission

U235 (n, fission) in

nuclear reactor

1 / fission

180

5. Spallation

800 MeV protons

on uranium

30 / proton

55

Laser fusion in
controlled
thermonuclear
reactor

1 / fusion

18

6. (D,T) fusion

Table 2: The neutron yield and heat released per useful neutron from the
various reactions used to generate neutrons.

Figure 15: A cut-away view of the ISIS

pulsed neutron facility.

3.1-15

Figure 16: The principle of injection of an H- beam into a closed orbit synchrotron.
Provided that the peak flux of the pulsed source can be utilised for a high
proportion of the time frame between pulses this substantial gain can be achieved
with no decrease in data rate at the neutron scattering instrument. Conversely for
the same rate of heat deposition in the target a pulsed source will, in general,
achieve a higher data rate than the continuous source.
2.2 ISIS at the Rutherford Appleton Laboratory
A cut-away view of the pulsed neutron source ISIS at the Rutherford Appleton
Laboratory in the UK is shown in Figure 15. It comprises a series of accelerators
which produce an intense 800 MeV pulsed beam of protons which is incident on a
tantalum target, producing fast neutrons by the spallation process.
We can divide the facility into nine sections for the purpose of describing it.
1.

A pre-injector high voltage generator of the Cockcroft-Walton potential

divider type. This produces pulsed voltages at 665 kV and 50 Hz, delivering a
current of 40 mA in 500 s bursts to ....

2.

... an H- ion source. The ion source produces negative hydrogen ions H- in a
caesium vapour discharge fed with monatomic hydrogen atoms generated by
passing hydrogen gas through a heated nickel tube. The H- ions sitting at the
665 kV potential of the pre-injector are driven towards the first element of ....

3.

... a linear accelerator injector of the Alvarez type consisting of a series of

potential gaps increasing in length along the linac as the H- ions gain energy.

3.1-16
The linac accelerates the H- ions to 70 MeV producing a current of 20 mA in
500 s bursts every 20 ms (i.e. at 50 Hz). The H - ion beam enters the
synchrotron via....
4.

... an injection straight where a 0.25 m thick foil of alumina Al2O3 strips the
electrons from the H - ions and the resulting protons are trapped in the
magnetic field of the synchrotron and join the acceleration process. The use of
H- ions and multi-turn injection allows the current in the synchrotron to be
increased up to the space-charge limit. Thus H- ions and protons from earlier
on in the injection process pass through the same magnetic field at the stripper
foil but with equal and opposite curvatures, as shown in Figure16.

5.

The synchrotron is 52 metres in diameter in the shape of a ten-sided polygon.

Each of the ten sides has common magnet elements for focusing the proton
beam and for bending it by 36 into the next straight section.
Each straight has an individual purpose. The first straight is used for injection
as we have seen above. Six straights are used for acceleration purposes, each
containing a radio frequency ferrite-cored accelerating cavity which feeds
energy, appropriately phased, into the proton beam. As the beam gains energy
so the frequency of the RF cavities is increased to give the proton beam a
slight increase in energy on each pass. The six RF cavities accelerate the beam
to 800 MeV at which stage the protons are circulating in two bunches on
opposite sides of the synchrotron. Each bunch is 90 nsec wide and separated
from the second bunch by 210 nsec. There are 2.5 x 1013 protons per pulse in
the accelerated beam which is equivalent to an average current of 200 A.
Two straights of the synchrotron are for vacuum pumping ports - the ring is
held at a vacuum of 5 x 10-8 mbar - and the final straight is ....

6.

... the extraction straight which contains a set of fast acting kicker magnets.
These magnets are powered at the end of the acceleration process and the two
bunches of protons in the ring are extracted simultaneously from the
synchrotron into ....

7.

... the extracted proton beamline. The proton pulse is then guided by
quadrupole magnets along an 80 metre vacuum tank to...

8.

... the target station where it strikes the tantalum target thus producing a burst
of fast neutrons by the spallation process. The target is surrounded by four
moderators two are ambient water, one is liquid methane at 110 , and one
is supercritical hydrogen vapour at 25 K - and a beryllium reflector to
concentrate the neutron flux in the region of the moderators. The moderated
neutrons pass through shutters and beam lines to reach ....

3.1-17

Figure 17: A plan view of the components of the ISIS pulsed neutron source
with the instruments arranged around the tantalum target. Upstream is the
muon target with its instrument suite.
9.

... the neutron scattering instruments. There are 18 beam holes on ISIS, nine
on either side of the target station. The present (March 2002) layout of
instruments is shown in Figure 17.

2.3 The High Flux Reactor at the Institut Laue Langevin

A plan view of the high flux reactor at the ILL in Grenoble and the associated
neutron instruments is shown in Figure 18. A short discussion of the reactor core
and the coolant system has taken place earlier in the paper.
The reactor core consists of a single fuel element made from highly enriched
(93%) uranium-235 with a mass of ~ 8.57 kg. The core has a diameter of 40 cm and
is controlled by a single, central control rod. A power of 58 MW is generated which
is removed by pumping water through the fins of the fuel element. At equilibrium
the temperature of the fuel element is 50 C. The core is surrounded by a D2O
reflector vessel of diameter 2.5 m and outside the reflector is a tank of H2O and a
concrete radiation shield. There are also three purpose-built moderators at different
temperatures to provide a wide range of neutron spectra for the various
instruments. These are a hot moderator, which is a graphite block at 2500C
providing peak flux at 200 meV, and two cold moderators, one being a spherical 25

3.1-18

Figure 18: A schematic layout of the instruments at the ILL showing the two
guide halls.
litre vessel containing liquid deuterium at 20 K with its peak intensity at 5 meV. In
addition there is an ultra cold neutron source which generates neutrons at
wavelengths of 1000 Angstroms.
The neutron beam tubes pass through the biological radiation shield of the reactor
into the region of highest thermal flux in the D2O reflector arrayed around the core of
the reactor. Some of the beam tubes are radial and thus view the core itself whilst
others are tangential to the core. The former have a more intense neutron flux than the
latter but with the disadvantage of a higher gamma flux also. There are also three
bunches of neutron conducting tubes, or guides, which view respectively the ambient
reflector and transmit a thermal neutron spectrum, and the two cold sources and
transmit a low energy or cold neutron spectrum. These guides transmit the neutrons
without a significant loss in intensity to regions distant from the reactor (40 m to 140
m) where backgrounds are low and space is available to more than treble the neutron
instrumentation which can be installed close to the reactor.
2.4 A brief Comparison of Pulsed Sources and Reactor Sources
To hazard a comparison of different neutron sources is perhaps unwise given
the large number of committed individuals and vested interests. Nevertheless my

3.1-19
Pulsed Sources
Reactors
1. More high energy neutrons
1. More low energy neutrons
2. Produce neutrons in bursts & measure 2. Easier to shield against fission neuwhen source is off. Lower backgrounds! trons. Lower backgrounds!
3. Pulsed Operation
3. Continuous Operation
If peak ~ average then data rates are equal
4. Higher peak possible
4. Higher average difficult
5. Sharp pulses give high resolution
5. Resolution can be adapted to problem
6. Pulse shape asymmetric, resolution
6. Resolution function symmetric, norfunction asymmetric
mally gaussian
7. Must use Time of Flight methods
7. Complete flexibility
constrained to source frequency
8. Horizons still to be explored.
8. Mainly tried & tested techniques
9. Seen as environmentally friendly
9. Seen as environmentally unfriendly

Table 3: Frequently stated advantages of pulsed sources and reactors. Many

of those advantages are subjective and therefore the table makes no claim to
be consistent.

personal list is given in table III. In the end it is the science which is done on the
different instruments at the various sources which is important and this is
dependent upon many other issues than a simple comparison will provide.
3. Further Reading
Neutron Physics by K H Beckurts & K Wirtz, Springer Verlag, 1964.
The Elements of Nuclear Reactor Theory S Glasstone & M C Edlund, Van
Nostrand, 1952.
Pulsed Spallation Neutron Sources for Slow Neutrons J M Carpenter, Nuclear
Instruments and Methods, (1977), 145, 91-113.
Spallation Neutron Sources for Neutron Beam Research G Manning, Contemporary
Physics, (1978), 19, 505-529.
The Neutron & the Bomb Andrew Brown 1998 Oxford University Press

3.2-1

Neutron Optics
Ian S. Anderson
1. Introduction
Owing to the low primary flux of neutrons, the beam definition devices that play
the role of defining the beam conditions (direction, divergence, energy,
polarisation, etc.) have to be highly efficient. The following sections give a (nonexhaustive) review of commonly used beam definition devices. A more detailed
review may be found in [1].
2. Collimators
A collimator is perhaps the simplest neutron optical device and is used to define the
direction and divergence of a neutron beam. The most rudimentary collimator
consists of just two slits or pinholes cut into an absorbing material and placed one
at the beginning and one at the end of a collimating distance L. The maximum
beam divergence that is transmitted with this configuration is
max = (a1 + a2)/L ,

(1)

where a1 and a2 are the widths of the slits or pinholes.

Such a device is normally used for small-angle scattering and reflectometry.
To avoid parasitic scattering by reflection from slit edges, very thin sheets of a
highly absorbing material (e.g., gd or cd foils) are used as the slit material. In cases
where a very precise edge is required, cleaved single-crystalline absorbers such as
gadolinium gallium garnet (GGG) can be employed.
As can be seen from eqn. 1, the divergence from a simple slit or pinhole
collimator depends on the aperture size. To collimate (in one dimension) a beam of
large cross section within a reasonable distance L, Soller collimators, composed of
a number of equidistant neutron-absorbing blades separated by spaces, are used.
The transmission, , of such a collimator depends on the thickness of the blades t
compared with the width of the transmission (spaces) channels s:
=

s
s+ t

Blades must be as thin and as flat as possible. If their surfaces do not reflect
neutrons, the angular dependence transmission of the function is close to the ideal
triangular form, and transmissions of 96% of the theoretical value can be obtained

3.2-2
with 10 collimation. Alternatively, thin single-crystal silicon (sapphire or quartz
are also suitable) wafers, coated with an absorbing layer (e.g., gd), can also be used
to construct microcollimators. The transmission losses through silicon can be
minimized by choosing very thin silicon wafers (200 m) coated with a few
microns of gd. The principal potential gain of this design, however, is the
possibility to build collimators with reflecting walls. If the blades of the Soller
collimator are coated with a material whose critical angle of reflection is equal to
max/2 (for one particular wavelength), then a square angular transmission function
is obtained instead of the normal triangular function, doubling the theoretical
transmission.
Soller collimators are often used in combination with single-crystal monochromators to define the wavelength resolution of an instrument. The Soller
geometry is only useful for one-dimensional collimation. For small-angle
scattering applications where two-dimensional collimation is required, a
converging pepper pot collimator can be used. [2]
Cylindrical collimators with radial blades are sometimes used to reduce
background scattering from the sample environment. This type of collimator is
particularly useful for use with position-sensitive detectors and may be oscillated
about the cylinder axis to reduce the shadowing effect of the blades. [3]
3. Crystal Monochromators
Bragg reflection from crystals is the most widely used method for selecting a welldefined wavelength band from a white neutron beam.
To obtain reasonable reflected intensities and to match typical neutron beam
divergences, crystals that reflect over an angular range of 0.2 to 0.5 degrees are
typically employed. Traditionally, mosaic crystals are preferred over perfect
crystals, although reflection from a mosaic crystal gives rise to an increase in beam
divergence with a concomitant broadening of the selected wavelength band. Thus,
collimators are often used together with mosaic monochromators to define the
initial and final divergences and therefore the wavelength spread.
Because of the beam broadening produced by mosaic crystals, elastically
deformed perfect crystals and crystals with gradients in lattice spacings may be
more suitable candidates for focusing applications since the deformation can be
modified to optimise focusing for different experimental conditions [4]. Perfect
crystals are used commonly in high-energy-resolution backscattering instruments,
interferometry, and in Bonse-Hart cameras for ultrasmall-angle scattering [5].
Mosaic crystals show a much broader diffraction profile compared with
perfect crystals with a lower peak reflectivity. An ideal mosaic crystal is assumed to

3.2-3
comprise an agglomerate of independently scattering domains or mosaic blocks
that are more or less perfect but small enough that primary extinction does not
come into play. Additionally, the intensity reflected by each block can be calculated
using the kinematic theory of Zachariasen. [6]
One assumes that the mosaic blocks are oriented almost parallel to the crystal
surface (for Bragg case) following a distribution W (B), with being the angle
formed by the incident beam and the Bragg planes and B being the Bragg angle.
The full-width-at-half-maximum of this distribution is called the mosaic spread
or mosaicity. The multiple Bragg reflections in a mosaic crystal and the concept of
secondary extinction are summarised by the Darwin equations. An exact and very
general solution of these equations has been given by Sears [7]. The physical
quantities that govern diffraction by a mosaic crystal are the absorption coefficient,
, and the scattering coefficient = Q[W ( B )] . The Q factor is given by
2
Q = 3 Fhkl
/(V 20 sin 2 B ) , where

is the wavelength, Fhkl is the structure factor, and V0

a = d /sin

is the unit cell volume. If we define

and

b = d /sin

, with d as the

crystal thickness and as the angle formed by the incident beam and the surface,
the Sears equations for the reflected and transmitted beam in symmetric Laue
(transmission) and Bragg (reflection) geometries are
1

-a
2 b
R Laue symm = 2 e (1 e )

(2)

1
= e -a (1 + e2 b )
2

(3)

T Laue symm
RBragg symm =

TBragg symm =

b
a( a + 2b) coth a( a + 2b) + ( a + b)

a( a + 2b)
a( a + 2b) cosh a( a + 2b) + ( a + b)sinh a( a + 2b)

(4)

(5)

The ideal monochromator material should have a large scattering length

density, low absorption, incoherent and inelastic crosssections, and should be
available as large single crystals with a suitable defect concentration. Relevant
parameters for some typical neutron monochromator crystals are given in Table 1.
Because of the higher reflectivities that can be obtained, neutron monochromators
are usually designed to operate in reflection geometry rather than transmission
geometry, for which the optimisation of crystal thickness is only achieved for a small
wavelength range.The maximum peak reflectivities and the optimum thicknesses for

Table 1: Some important properties of materials used for neutron monochromator crystals (in order of increasing
unit-cell volume).

3.2-4

3.2-5

Figure 1: Peak reflectivity for some typical monochromator crystals

calculated using an intrinsic mosaic of 0.1.
some selected reflections are shown in Figs. 1 and 2, respectively.
The reflection from a mosaic crystal in reciprocal space is visualised in Fig. 3.
An incident beam with small divergence is transformed into a broad exit beam. The
range of k vectors, k, selected in this process depends on the mosaic spread, , and
the incoming and outgoing beam divergences 1 and 2:
k/k = / + cot()

(6)

where is the crystal reciprocal-lattice vector ( = 2/d) and is given by:

2 2

2 2

2 2

1 2 + 1 + 2
12 + 22 + 42

(7)

The resolution can therefore be defined by collimators, and the highest resolution is
obtained in backscattering, where the wavevector spread depends only on the
intrinsic d/d of the crystal.

3.2-6

Figure 2: Wavelength dependence of the optimum thickness for various

monochromator reflections.
In some applications, the beam broadening produced by mosaic crystals can
be detrimental to the instrument performance. An interesting alternative is a
gradient crystal, that is,. single crystals with a smooth variation of the interplanar
lattice spacing along a defined crystallographic direction. As shown in Figure 3, the
diffracted phase space element has a different shape from that obtained from a
mosaic crystal. Gradients in d-spacing can be produced in various ways: thermal
gradients [8], vibrating crystals by piezoelectric excitation [9], and mixed crystals
with concentration gradients (e.g., Cu-Ge [10] and Si-Ge. [11]).
Both vertically and horizontally focusing assemblies of mosaic crystals are
employed to make better use of the neutron flux when making measurements on small
samples. Vertical focusing can lead to intensity gain factors between 2 and 5 without
affecting resolution (real-space focusing) [12]. Horizontal focusing changes the k-

3.2-7
Figure 3: Reciprocal-lattice representation of the effect of a monochromator
with reciprocal-lattice vector t on the
reciprocal-space element of a beam with
divergence a. (a) For an ideal crystal with
a lattice constant width ; (b) for a
mosaic crystal with mosaicity h, showing
that a beam with small divergence, a, is
transformed into a broad exit beam with
divergence 2h + a; (c) for a gradient
crystal with interplanar lattice spacing
changing over , showing that the divergence is not changed in this case.

Figure 4: In a curved neutron

guide, the transmission
becomes dependent: (a) the
possible types of reflection
(garland and zigzag), the
direct line-of-sight length, the
critical angle *, which is
related to the characteristic
wavelength * = *
Nb coh

(b) Transmission across the

exit of the guide for different
wavelengths, normalized to
unity at the outside edge. (c)
Total transmission of the guide
as a function of .

3.2-8
space volume that is selected by the monochromator through the variation in Bragg
angle across the monochromator surface (k-space focusing) [13]. The orientation of the
diffracted k-space volume can be modified by variation of the horizontal curvature so
that the resolution of the monochromator may be optimised with respect to a particular sample or experiment without loss of illumination. Monochromatic focusing can be
achieved. Furthermore, asymmetrically cut crystals can be used, allowing focusing
effects in real space and k-space to be decoupled. [14]
4. Mirror Reflection Devices
The refractive index, n, for neutrons of wavelength propagating in a nonmagnetic
material of atomic density N is given by the expression
= 1 i

= 1

2 N.bcoh
2

(8)

where bcoh is the mean coherent scattering length, and m is the linear absorption
coefficient. Values of the scattering-length density N.b coh for some common
materials are reported in Table 2. The refractive index for most materials is slightly
less than unity so that total external reflection can take place. Thus, neutrons can be
reflected from a smooth surface, but the critical angle of reflection, c, given by:
c =

Nbcoh

(9)

is small so that reflection can take place only at grazing incidence. The critical
angle for nickel for example is 0.1 -1.
Because of the shallowness of the critical angle, reflective optics are
traditionally bulky, and focusing devices tend to have long focal lengths. In some
cases however, depending on the beam divergence, a long mirror can be replaced
by an equivalent stack of shorter mirrors.
Neutron Guides
The principle of mirror reflection is the basis of neutron guides that are used to
transmit neutron beams to instruments situated up to 100 m away from the
source [15].
A standard neutron guide is constructed from boron glass plates assembled into a
rectangular cross section, the dimensions of which may be up to 200 mm high by 50 mm
wide. The inner, reflecting surface of the guide is coated with approximately 1200 of
nickel, 58Ni (c = 0.12 -1), or a supermirror (described subsequently). The guide is
usually evacuated to reduce losses from absorption and scattering of neutrons in air.

3.2-9

Figure 5: Illustration of
how a variation in the
bilayer period can be
used to produce a
monochromator, a
broad-band device, or a
supermirror.

Theoretically, a neutron guide that is fully illuminated by the source will

transmit a beam with a square divergence of full width 2c , in both the horizontal
and vertical directions so that the transmitted solid angle is proportional to 2. In
practice, because of imperfections in the assembly of the guide system, the
divergence profile is closer to Gaussian at the end of a long guide.
Because neutrons could undergo a large number of reflections in the guide, it is
important to achieve a high reflectivity.
The specular reflectivity is determined
by the surface roughness, and, typically,
Material
Nbcoh (10-6 -2)
values in the range 98.5 to 99% are
58Ni
13.31
achieved. Further transmission losses
Diamond
11.71
occur because of imperfections in the
Nickel
9.4
alignment of the sections that make up
Quartz
3.64
the guide.
Germanium
3.64
The great advantage of neutron
Silver
3.50
guides, in addition to the transport of neuAluminum
2.08
trons to areas of low background, is that
Silicon
2.08
they can be multiplexed, that is, one guide
Vanadium
-0.27
can serve many instruments. This is
Titanium
-1.95
achieved either by deflecting only a part
Manganese
-2.95
of the total crosssection to a given instrument or by selecting a small wavelength
Table 2: Scattering length densities
range from the guide spectrum. In the latfor some common materials
ter case, the selection device (usually a
typically used in neutron optics.

3.2-10
crystal monochromator) must have a high transmission at other wavelengths.
If the neutron guide is curved, the transmission becomes wavelength
dependent. The characteristic wavelength, *, at which the theoretical transmission
drops to 67% is related to the characteristic angle * = 2 a in the following way:

Nb coh

*,

(10)

where a is the guide width, and is the radius of curvature. For wavelengths less than ,
neutrons can be transmitted only by garland reflections along the concave wall of
the curved guide. Thus, the guide acts as a low-pass energy filter as long as its length is
longer than the direct line-of-sight length L1= 8.a. . The line-of-sight length can be
reduced by subdividing the guide into a number of narrower channels, each of which
acts as a mini-guide. The resulting device, often referred to as a neutron bender since
deviation of the beam is achieved more rapidly, is used in beam deviators
Multilayers
Schoenborn and coworkers [16] first pointed out that multilayers, comprising
alternating thin films of different scattering-length densities (N.bcoh), act like twodimensional crystals with a d-spacing given by the bilayer period. With modern
deposition techniques (usually sputtering), uniform films of thickness ranging from
~ 20 to a few 100 can be deposited over large surface areas of the order of 1 m2.
Owing to the rather large d-spacings involved, the Bragg reflection from multilayers
is generally at grazing incidence so that long devices are required to cover a typical
beam width. Alternatively, a stacked device must be used. However, with judicious
choice of the scattering-length contrast, the surface and interface roughness, and the
number of layers, reflectivities of close to 100% can be reached.
Fig. 5 illustrates how variation in the bilayer period can be used to produce a

Figure 6: Measured
reflectivity of an m = 4
supermirror. The reduction
in reflectivity at low values of
m is an artifact caused by
underillumination of the
sample during the
measurement.

3.2-11
monochromator (the minimum / that can be achieved is of the order of 1%), a
broad band device or a supermirror, so called because it is composed of a
particular sequence of bilayer thicknesses that in effect extends the region of total
mirror reflection beyond the ordinary critical angle. [17] Nowadays supermirrors
can be produced which extend the critical angle of nickel by a factor, m, between 3
and 4 with reflectivities better than 90% (Fig. 6). Such high reflectivities enable
supermirror neutron guides to be constructed with flux gains compared with nickel
guides close to the theoretical value of m2.
The choice of the layer pairs depends on the application. For non-polarising
supermirrors and broad band devices, the nickel/titanium pair is commonly used
due to the high contrast in scattering density, while for narrow band
monochromators a low contrast pair such as tungsten/silicon is more suitable.
Capillary optics
Capillary neutron optics, in which hollow glass capillaries act as waveguides are
also based on the concept of total external reflection of neutrons from a smooth
surface. The advantage of capillaries, compared with neutron guides, is that the
channel sizes are of the order of a few tens of micrometers, so that the radius of
curvature can be significantly decreased for a given characteristic wavelength (see
eqn. 10). Thus, neutrons can be efficiently deflected through large angles resulting
in a more compact optical system.
Two basic types of capillary optics exist, and the choice depends on the beam
characteristics required. Polycapillary fibers are manufactured from hollow glass
tubes several centimetres in diameter, which are heated, fused, and drawn multiple
times until bundles of thousands of micrometer-sized channels are formed having
an open area of up to 70% of the crosssection. Fiber outer diameters range from
300 to 600 mm and contain hundreds or thousands of individual channels with
inner diameters between 3 and 50 m. The channel cross section is usually
hexagonal, though square channels have been produced, and the inner channel wall
surface roughness is typically less than 10 rms, giving rise to very high
reflectivities. The principal limitations on transmission efficiency are the open area,
the acceptable divergence (note that the critical angle for glass is 1 mrad/), and
reflection losses from absorption and scattering. A typical optical device will
comprise hundreds or thousands of fibers threaded through thin screens to produce
the required shape.
The second type of capillary optic is a monolithic configuration. The
individual capillaries in monolithic optics are tapered and fused together so that no
external frame assembly is necessary. Unlike the multifiber devices, the inner

3.2-12
diameters of the channels that make up the monolithic optics vary along the length
of the component, resulting in a smaller, more compact design. Capillary optics can
be used in lenses to focus or collimate a neutron beam [18] or simply as a beam
bender.
Filters
Neutron filters are used to remove unwanted radiation from the beam while
maintaining as high a transmission as possible for neutrons of the required energy.
Two major applications can be identified: removal of fast neutrons and -rays from
the primary beam and reduction of higher-order contributions (/n) in the
secondary beam reflected from crystal monochromators. In this section, we
consider nonpolarising filters, that is, those whose transmission and removal cross
sections are independent of the neutron spin. Polarising filters are dealt with in the
section concerning polarizers.
Filter action relies on a strong variation of the neutron cross section with
energy, usually either the wavelength-dependent scattering cross section of
polycrystals or a resonant absorption cross section. Following Freund [19], the total
cross section determining the attenuation of neutrons by a crystalline solid can be
written as a sum of three terms
= abs + tds + bragg

(11)

Here, abs is the true absorption cross section, which at low energy, away from
resonances, is proportional to E-1/2. The temperature-dependent thermal diffuse
cross section, tds, describing the attenuation caused by inelastic processes can be
split into two parts depending on the neutron energy. At low energy, E kbD,
where kb is Boltzmans constant and D is the Debye temperature, single phonon
processes dominate, giving rise to a cross section sph, which is also proportional to
E-1/2. The single-phonon cross section is proportional to T7/2 at low temperatures
and to T at higher temperature T.
At higher energies E kbD, multiphonon and multiple-scattering processes
come into play, leading to a cross section mph, which increases with energy and
temperature.
The third contribution, bragg, arises from Bragg scattering in single- or
polycrystalline material. At low energies, below the Bragg cutoff ( > 2dmax),
bragg is zero. In polycrystalline materials the cross section rises steeply above the
Bragg cutoff and oscillates with increasing energy as more reflections come into
play. At higher energies, bragg decreases to zero.
In single-crystalline material above the Bragg cutoff, bragg is characterized by

3.2-13

Table 3: Characteristics of some typical elements and isotopes used as neutron filters.
a discrete spectrum of peaks whose heights and widths depend on the beam
collimation, energy resolution, and the perfection and orientation of the crystal.
Hence, a monocrystalline filter has to be tuned by careful orientation. Figures 7-11
show the total cross sections for common filter materials as a function of energy,
while Fig. 12 shows the transmission of a pyrolitic graphite filter as a function of
incident wavelength.
Cooled, polycrystalline beryllium is frequently used as a filter for neutrons
with energy less than 5 meV since there is an increase of nearly two orders of
magnitude in the attenuation cross section for higher energies. BeO, with a Bragg
cutoff at approximately 4 meV, is also commonly used.
Pyrolitic graphite, being a layered material with good crystalline properties
along the c-direction but with random orientation perpendicular to it, lies
somewhere between a polycrystal and a single crystal as far as its attenuation cross
section is concerned. Pyrolitic graphite serves as an efficient second- or third-order
filter [20] and can be tuned by slight misorientation away from the c-axis.
Resonant absorption filters show a large increase in their attenuation cross sections
at the resonant energy and are therefore used as selective filters for that energy. A list of
typical filter materials and their resonance energies is given in Table 3.

3.2-14

Figure 7: Total cross section for beryllium in an energy range where it can be
used as a filter [19].

3.2-15

Figure 8: Total cross section for sapphire in an energy range where it can be
used as a filter [19].

3.2-16

Figure 9: Total cross section for silicon in an energy range where it can be used
as a filter [19].

3.2-17

Figure 10: Total cross section for quartz in an energy range where it can be
used as a filter [19].

3.2-18

Figure 11: Total cross section for pyrolitic graphite in an energy range where
it can be used as a filter [19].

3.2-19
Refractive lenses
2
N.b of the refractive index given by
The real part of the decrement =
2
n = 1 i is small and positive for most materials. Hence, concave lenses must be
R
generally used to focus neutron beams, and the focal lengths f =
, where R is the
2
on-axis radius of curvature, are prohibitively long. Reducing R, to reduce the focal
length to practical values, would severely limit the lens aperture.
A series of N-aligned lenses with a negligible distance between each lens has a
R
focal length f =
.
coh

2 N
1

The factor reduction in focal length of a compound refractive lens allows

for a reasonable value for the radius, R, of each element. Spherical surfaces make
imperfect lenses, and only the central paraxial region of these approximate the
paraboloid of revolution surface of ideal lenses. For lenses with paraboloid
surfaces, R is the radius of curvature on axis at the apex of the paraboloid.
A suitable lens material should have a large value of and a small value of the
linear attenuation coefficient, . The figure of merit, , is listed for several

materials in Table 4.

/ [m]

Table 4: Density, linear absorption coefficient, and figure of merit of selected

materials at a wavelength of 1.8 .

3.2-20

Figure 12: Measured transmission of a typical pyrolytic graphite filter, 4 cm

thick, as a function of neutron wavelength, for neutrons traveling approximately
normal to the (0002) graphite planes. The filter orientation is tuned to achieve
minimum transmission of unwanted higher-order neutrons.
Polarizers
Methods used to polarise a neutron beam are varied, and the choice of the best
technique depends on the instrument and the experiment to be performed. The main
parameter that has to be considered when describing the effectiveness of a given
polariser is the polarising efficiency, defined as
P = (N+ - N-)/(N+ + N-) ,

(12)

where N+ and N- are the numbers of neutrons with spin parallel (+) or antiparallel
(-) to the guide field in the outgoing beam. The second important factor,
transmission of the wanted spin state, depends on various factors such as
acceptance angles, reflection, and absorption.
Single-crystal polarizers
The principle by which ferromagnetic single crystals are used to polarize and
monochromate a neutron beam simultaneously is shown in Fig. 13. A field B,
applied perpendicular to the scattering vector , saturates the atomic moments M
along the field direction B. The cross section for Bragg reflection in this geometry is
)2 + 2 FN(
) FM(
)(P
) + FM(
)2 ,
(d/d) = FN(

(13)

3.2-21

Figure 13: Geometry of a polarizing monochromator showing the lattice

planes (hkl) with |F N| = |F M|, the direction of P and , the expected spin
direction and intensity.

) and FM(
) are the nuclear and magnetic structure factors, respectively.
where FN(
The vector P describes the polarization of the incoming neutron with respect to B;
P = 1 for (+) spins, P = -1 for (-) spins, and is a unit vector in the direction of
the atomic magnetic moments.
= 1), the diffracted intensity is
Hence, for neutrons polarized parallel to B (P
proportional to [FN()+FM()]2, while, for neutrons polarized antiparallel to B
= -1), the diffracted intensity is proportional to [F N () - F M ()] 2 .The
(P
polarising efficiency of the diffracted beam is then:
P = 2 FN() FM() / [FN()2 - FM()]2,

(14)

)| = |FM(
)|.
which can be either positive or negative and has a maximum value for |FN(
Thus, a good single-crystal polariser, in addition to possessing a convenient
crystallographic structure, must be ferromagnetic at room temperature and should
contain atoms with large magnetic moments. Furthermore, large single crystals with
controllable mosaic should be available. Finally, the structure factor for the required
reflection should be high, while those for higher-order reflections should be low.
None of the three naturally occurring ferromagnetic elements (Fe, Ni, and Co)
makes efficient single-crystal polarizers. Cobolt is strongly absorbing, and the nuclear
scattering lengths of iron and nickel are too large to be balanced by their weak magnetic
moments. An exception is 57Fe, which has a rather low nuclear scattering length, and
structure-factor matching can be achieved by mixing 57Fe with Fe and 3% Si [21].

3.2-22

Table 5: Properties of polarizing crystal monochromators.

In general, to facilitate
structure factor matching, alloys
are used rather than elements. The
characteristics of some alloys used
as polarizing monochromators are
presented in Table 5. At short
wavelengths, the 200 reflection of
Co 0.92 Fe 0.08 is used to give a
positively polarized beam [F N( )
and FM() both positive], but the
absorption from cobalt is high. At
longer wavelengths, the (111)
reflection of the Heusler alloy
Figure 14: Calculated and measured
Cu2MnAl [22] is commonly used
reflectivity from the (111) reflection of a
since it has a higher reflectivity
Heusler crystal with an intrinsic mosaic
(Fig. 14) and a larger d-spacing
of 0.045.
than Co 0.92 Fe 0.08 . Since, for the
(111) reflection, F N ~ -F M , the
diffracted beam is negatively polarized. Unfortunately, the structure factor of the
(222) reflection is higher than that of the (111) reflection, leading to significant
higher-order contamination of the beam.

3.2-23
Polarizing mirrors
For a ferromagnetic material, the neutron refractive index is given by
n = 1 - 2 N(bcoh p)/2 ,

(15)

where, the magnetic scattering length, p, is defined by

p = 2(B H) m/h2N ,

(16)

where, m and are the neutron mass and magnetic moment, B is the magnetic
induction in an applied field H, and h is Plancks constant.
The - and + signs refer, respectively, to neutrons whose moments are aligned
parallel and antiparallel to B. The refractive index depends on the orientation of the
neutron spin with respect to the film magnetization, thus giving rise to two critical
angles of total reflection, - and +. Thus, reflection in an angular range between
these two critical angles gives rise to polarized beams in reflection and in
transmission. The polarization efficiency, P, is defined in terms of the reflectivity
r+ and r- of the two spin states:
P = (r+ - r-)/(r+ + r-) .

(17)

For optimum reflectivity, polarizing mirrors are usually made by depositing thin
films of ferromagnetic materials onto substrates of low surface roughness (e.g., float
glass or polished silicon). The reflection from the substrate can be reduced by
including an antireflecting layer made from, for example, gadolirium/titanium alloys.
The major limitation of such polarizers is that grazing-incidence angles must
be used and the angular range of polarization is small. This limitation can be
partially overcome by using multilayers, as described previously, in which one of
the layer materials is ferromagnetic. In this case the refractive index of the
ferromagnetic material is matched for one spin state to that of the nonmagnetic
material so that reflection does not occur. A polarizing supermirror made in this
way has an extended angular range of polarization (Figs. 15 and 16).
Note that modern deposition techniques allow the refractive index to be
adjusted readily so that matching is easily achieved. The scattering-length densities
of some commonly used layer pairs are given in Table 6.
Polarizing multilayers are also used in monochromators and broadband
devices. Depending on the application, various layer pairs can be used:
Co/Ti, Fe/Ag, Fe/Si, Fe/Ge, Fe/W, Fe 50 Co 48 V 2 /TiN, FeCoV/TiZr, and
54
63Ni
0.66 Fe0.34/V, and the range of fields used to achieve saturation varies from
about 10 to 500 Gs.
Polarizing mirrors can be used in reflection or transmission with polarization

3.2-24

Figure 15: Measured performance of an m = 3 Fe/Si polarizing supermirror

in reflection geometry. The reduction in reflectivity at low values of m is an
artefact due to underillumination of the sample during the measurement.

Figure 16: Measured performance of an m = 3 Fe/Si polarizing supermirror in

transmission geometry. The increase in transmission at low values of m is an
artifact caused by underillumination of the sample during the measurement.

3.2-25

Table 6: Scattering-length densities for some typical materials used for

polarizing multilayers. For the nonmagnetic layer, we have listed only the
simple elements that give a close match to the N(b-p) value of the
corresponding magnetic layer. In practice, excellent matching can be achieved
by using alloys (e.g., TixZry alloys allow niobium b values between -1.95 and
3.03 x 10-6 -2 to be selected) or reactive sputtering (e.g. TiNx).

efficiencies reaching 97%, although because of the low incidence angles, their use
is generally restricted to wavelengths above 2 . Various devices can be
constructed using mirror polarizers including simple reflecting mirrors, V-shaped
transmission polarizers [23], cavity polarizers [24], and benders [25].
Polarizing Filters
Polarizing filters operate by selectively removing one of the neutron spin states
from an incident beam, allowing the other spin state to be transmitted with only

3.2-26
moderate attenuation. The spin selection is obtained by preferential absorption or
scattering, so the polarizing efficiency usually increases with the thickness of the
filter, whereas the transmission decreases. A compromise must therefore be made
between polarization, P, and transmission, T. The quality factor often used is
P T [26].
The total cross sections for a generalized filter may be written as
= 0 p ,

(18)

where 0 is a spin-independent cross section and p = (+ + -)/2 is the polarization

cross section. It can be shown [27] that the ratio p/0 must be 0.65 to achieve |P|
> 0.95 and T > 0.2.
Magnetized iron was the first polarizing filter to be used. The method relies on
spin-dependent Bragg scattering from a magnetized polycrystalline block for which
p approaches 10 barns near the iron cutoff at 4 . For wavelengths in the range 3.6
to 4 , the ratio p/0 is ~ 0.59, resulting in the polarizing efficiency of 0.8 for a
transmittance of ~ 0.3. In practice, however, since iron cannot be fully saturated,
depolarization occurs, and values of P ~ 0.5 with T ~ 0.25 are more typical.
Resonance absorption polarization filters rely on the spin-dependence of the
absorption cross section of polarized nuclei at their nuclear resonance energy and can
produce efficient polarization over a wide energy range. The nuclear polarization is
normally achieved by cooling in a magnetic field, and filters based on 149Sm (Er =
0.097 eV) and 151Eu (Er = 0.32 eV and 0.46 eV) have been tested successfully [28].
The 149Sm filter has a polarizing efficiency close to 1 within a small-wavelength
range (0.85 to 1.1 ), while the transmittance is about 0.15.
Broadband polarizing filters, based on spin-dependent scattering or
absorption, provide an interesting alternative to polarizing mirrors or
monochromators because of the wider range of energy and scattering angle that can
be accepted. The most promising such filter is polarized 3He, which operates
through the huge spin-dependent neutron capture cross section that is totally
dominated by the resonance capture of neutrons with antiparallel spin. The
polarization efficiency of a 3He neutron spin filter of length, l, can be written as
Pn() = tanh[O()PHe] ,

(19)

where P He is the 3 He nuclear polarization and O() = [ 3 He] l 0 () is the

dimensionless effective absorption coefficient, also called the opacity [29]. For
gaseous 3He, the opacity can be written in more convenient units by
O = p[bar] x l[cm] x [] ,

(20)

3.2-27
where p is the 3He pressure and O =7.33 x 10-2 O.
Similarly, the residual transmission of the spin filter is given by:
Tn ( ) = exp[O( )] cosh [O( ) PHe ]

(21)

It can be seen from Fig. 17, that even at low 3 He polarization, full neutron
polarization can be achieved in the limit of large absorption at the cost of the
transmission.
3He can be polarised either by spin exchange with optically pumped Rb [30] or by
pumping of metastable 3He* atoms followed by metastable exchange collisions [31]. In
the former method, the 3He gas is polarized at the required high pressure whereas 3He*
pumping takes place at a pressure of about 1 mbar, followed by a polarization conserving compression by a factor of nearly 10,000. Although the polarization time constant
for Rb pumping is of the order of several hours, compared with fractions of a second for
3He*, pumping the latter requires several fills of the filter cell to achieve the required
pressure. Typically, nuclear polarizations PHe of 55% are achieved.
Zeeman Polarizer
The reflection width of perfect silicon crystals for thermal neutrons and the Zeeman
splitting ( = 2B) of a field of about 10 kG are comparable and therefore can be
used to polarize a neutron beam. For a monochromatic beam (energy E0 ) in a strong
magnetic field region, the result of the Zeeman splitting will be a separation into two
polarized subbeams, one polarized along B with energy E0 + B and the other
polarized antiparallel to B with energy E0 - B. The two polarized beams can be
selected by rocking a perfect crystal in the field region B [32].
Spin-orientation devices
Polarization is the state of spin orientation of an assembly of particles in a target or
beam. The beam polarization vector P is defined as the vector average of this spin
state of the assembly in the beam, often described by the density matrix
1
P). The polarization is then defined as P = Tr(
). If the polarization
= 2 (1+
vector is inclined to the field direction in a homogenous magnetic field, B, the
polarization vector will precess with the classical Larmor frequency L = ||B.
This results in a precessing vector and a precessing spin polarization. For most
experiments, it is sufficient to consider the linear polarisation vector in the direction
of an applied magnetic field. If, however, the magnetic field direction changes
along the path of the neutron, it is also possible that the direction of P will change.
If the frequency, , with which the magnetic field changes is such that
= d(B/|B|)/dt << L ,

(22)

3.2-28

Figure 17: Neutron polarization and transmission of a 3He filter with 55%
nuclear polarization. Also shown is the quality factor P T .

then the polarization vector follows the field rotation adiabatically. Alternatively,
when >> L , the magnetic field changes so rapidly that P cannot follow and the
condition is known as nonadiabatic fast passage. All spin-orientation devices are
based on these concepts.
Maintaining the direction of polarization
A polarised beam will tend to become depolarized during passage through a region
of zero field since the field direction is ill defined over the beam cross section.
Thus, to keep the polarization direction aligned along a defined quantization axis,
special precautions must be taken.
The simplest way of maintaining the polarization of neutrons is to use a guide
field to produce a well-defined field B over the whole flight path of the beam. If the
field changes direction, it has to fulfill the adiabatic condition << L, that is the
field changes must take place over a time interval that is long compared with the
Larmor period. In this case, the polarization adiabatically follows the field direction
with a maximum angle of deviation, 2 tan-1(/L) [33].

3.2-29
Rotation of the polarization direction
The polarization direction can be changed by the adiabatic change of the guide field
direction so that the direction of the polarization follows it. Such a rotation is
performed by a spin turner or spin rotator [34].
Alternatively, the direction of polarization can be rotated relative to the guide
field by using the property of precession described previously. If a polarized beam
enters a region where the field is inclined to the polarization axis, then the
polarization vector P will precess about the new field direction. The precession
angle will depend on the magnitude of the field and the time spent in the field
region. By adjustment of these two parameters together with the field direction, a
defined, though wavelength-dependent, rotation of P can be achieved. A simple
device uses the nonadiabatic fast passage through the windings of two rectangular
solenoids wound orthogonally on top of each other [35]. In this way the mechanical
rotation can be replaced by two currents the ratio of which defines the direction of
the precession field axis and the size of the fields determines the angle of the
precession. The orientation of the polarization vector can therefore be defined in
any direction.
In order to produce a continuous rotation of the polarization (i.e., a
well-defined precession) as, required in neutron spin echo applications, precession
coils are used. In the simplest case these are long solenoids for which the change of
the field integral over the cross section can be corrected by Fresnel coils [36].
Zeyen and coworkers have developed and implemented optimal field shape coils
(ofs) [37]. The field in these coils follows a cos 2 shape that results from the
optimization of the line integral homogeneity. The OFS coils can be wound over a
very small diameter, drastically reducing stray fields.
Flipping of the polarization direction
The term flipping was originally applied to the situation where the beam
polarization direction is reversed with respect to a guide field; that is. it describes a
transition of the polarization direction from parallel to antiparallel to the guide field
and vice versa. A device that produces this 180 rotation is called a -flipper. A /2
flipper, as the name suggests, produces a 90 rotation and is normally used to
initiate precession by turning the polarization at 90 to the guide field.
The most direct, wavelength-independent way of producing such a transition
is again a nonadiabatic fast passage from the region of one field direction to the
region of the other field direction. This can be realized by a current sheet like the
Dabbs foil [38], a Kjeller eight [39], or a cryoflipper [40].
Alternatively, a spin flip can be produced using a precession coil, as described

3.2-30
previously, in which the polarization direction makes a precession of just about a
direction orthogonal to the guide field direction. Normally, two orthogonally
wound coils are used, where the second, correction coil, serves to compensate the
guide field in the interior of the precession coil. Such a flipper is wavelength
dependent and can be easily tuned by varying the current in the coils.
Mechanical Choppers and Selectors
In this section we deal with devices that take advantage of the neutron flight time to
make a selection of some sort. To put things into perspective, we should mention that a
4- neutron has a reciprocal velocity of approximately 1,000 s m-1 so that accurate
neutron energy determinations can be made with flight paths of only a few meters.
Disc choppers rotating at speeds up to 20,000 rpm about an axis that is parallel
to the neutron beam are used to produce a well-defined pulse of neutrons. The discs
are made from absorbing material (at least where the beam passes) and comprise
one or more neutron-transparent apertures or slits. For polarized neutrons these
transparent slits should not be metallic, as the eddy currents in the metal moving in
even a weak guide field will strongly depolarize the beam. The pulse frequency is
determined by the number of apertures and the rotation frequency, while the duty
cycle is given by the ratio of open time to closed time in one rotation. Two such
choppers rotating in phase can be used to monochromate and pulse a beam
simultaneously [41]. In practice, more than two choppers are generally used to
avoid frame overlap of the incident and scattered beams. The time resolution of
disc choppers (and hence the energy resolution of the instrument) is determined by
the beam size, the aperture size, and the rotation speed. For a realistic beam size the
rotation speed limits the resolution.
Therefore, in modern instruments it is normal to replace a single chopper with
two counter-rotating choppers [42]. The low duty cycle of a simple disk chopper
can be improved by replacing the single slit with a series of slits either in a regular
sequence (Fourier chopper) [43] or a pseudostatistical sequence (pseudostatistical
chopper) [44] with duty cycles of 50% and 30%, respectively.
The Fermi chopper is an alternative form of neutron chopper that
simultaneously pulses and monochromates the incoming beam. It consists of a slit
package, essentially a collimator, rotating about an axis that is perpendicular to the
beam direction [45]. For optimum transmission at the required wavelength, the slits
are usually curved to provide a straight collimator in the neutron frame of
reference. The curvature also eliminates the reverse burst, that is. a pulse of
neutrons that passes when the chopper has rotated by 180.
A Fermi chopper with straight slits in combination with a monochromator

3.2-31
assembly of wide horizontal divergence can be used to time focus a polychromatic
beam, hence maintaining the energy resolution while improving the intensity [46].
Velocity selectors are used where a continuous beam is required with coarse
energy resolution. They exist in either multiple disc configurations or helical
channels rotating about an axis parallel to the beam direction [47]. Modern helical
channel selectors are made up of lightweight absorbing blades slotted into helical
grooves on the rotation axis [48]. At higher energies, where no suitable absorbing
material is available, highly scattering polymers (polymethyl metacrylate) can be
used for the blades, although in this case adequate shielding must be provided. The
neutron wavelength is determined by the rotation speed and resolutions, /;
range from 5 to practically 100% (/2 filter) can be achieved. The resolution is
fixed by the geometry of the device but can be slightly improved by tilting the
rotation axis or can be relaxed by rotating in the reverse direction for shorter
wavelengths. Nowadays transmissions of up to 94% are obtainable.
References
[1] Anderson, I. S., Brown, P.J., Carpenter, J.M., Lander, G., Pynn, R., Rowe, J.M.,
Schrpf, O., Sears, V.F., and Willis, B.T.M., Intern. Tables for Crystallography C,
2nd Edition, Kluwer Academic Publish. (1999) 426
[2] Nunes, A. C., NIM 119 (1974) 291; Glinka, C. J., Rowe, J. M., and LaRock, J.
G., J. Appl. Cryst. 19 (1986) 427
[3] Wright, A. F., Berneron, M., and Heathman, S. P., NIM 180 (1981) 655-658
[4] Maier-Leibnitz, H., Summer School on Neutron Physics, Alushta Joint Insitute
of Nuclear Physics (Russia) (1969)
[5] Bonse, U., and Hart M., Appl. Phys. Lett. 7 (1965) 238-240
[6] Zachariasen, W. H., Theory of X-ray Diffraction in Crystals, Dover, New
York (1945)
[7] Sears, V. F., Acta Cryst. A53 (1997) 35-36
[8] Alefed, B., Kerntechnik 14 (1972) 15-17
[9] Hock, R., Vogt, T., Kulda, J., Mursic, Z., Fuess, H., and Magerl A., Z. Phys B 90
(1993) 143.
[10] Freund, A. K., Guinet, P., Mareschal, J., Rustichelli, F., and Vanoni, F., J.
Cryst. Growth 13-14 (1972) 726
[11] Magerl, A., Liss, K. D., Doll, C., Madar R., and Steichele, E., NIM A 338 (1994) 83-89
[12] Riste, T., NIM 86 (1970) 1-4; Currat, R., NIM 107 (1973) 21-28

3.2-32
[13] Scherm, R., Dolling, G., Ritter, R., Schedler, E., Teuchert, W., and Wagner, V.,
NIM 143 (1977) 77-85
[14] Scherm, R. H., and Kruger, E., NIM A 338 (1994) 1
[15] Christ, J., and Spinger, T., Nukleonik 4 (1962) 23-25
[16] Schoenborn, B. P., Caspar, D. L. D., and Kammerer, O. F., J. Appl. Cryst. 7
(1974) 508-510
[17] Mezei, F., Commun. Phys. 1 (1976) 81-85; Hayter J. B. and Mook H. A., J.
Appl. Cryst. 22 (1989) 35-41
[18] Mildner, D.F.R., Chen-Mayer, H.H., and Downing, R.G., J. Phys. Soc. Jpn 65
(1996) Suppl. A, 308-311
[19] Freund, A. K., NIM 213 (1983) 495-501
[20] Shapiro, S. M., and Chesser, N. J., NIM 101 (1972) 183-186
[21] Reed, R. E., Bolling, E. D., and Harmon, H. E., Solid State Division Report.
Oak Ridge National Laboratory, USA (1973) 129
[22] Delapalme, A., Schweizer, J., Couderchon, G., and Perrier de la Bathie, R.,
NIM 95 (1971) 589-594; Freund, A., Pynn, R., Stirling, W. G., and Zeyen, C. M. E.,
Physica B 120 (1983) 86-90
[23] Majkrzak, C. F., Nunez, V., Copley, J. R. D., Ankner, J. F., and Greene, G. C.
in Neutron optical devices and applications, edited by C. F. Majkrzak and J. L.
Wood., SPIE Proc. 1738 (1992) 90-106
[24] Mezei, F. in Thin film neutron optical devices: mirrors, supermirrors,
multiplayer monochromators, polarizers and beam guides, edited by C. F.
Majkrzak., SPIE Proc. 983 (1988) 10-17
[25] Schrpf, O., Physica (Utrecht) B 156 & 157 (1989) 639-646
[26] Tassel, F. and Resouche, E., NIM A 359 (1995) 537-541
[27] Williams, W. G, Polarized Neutrons, Claredon Press, Oxford (1988)
[28] Freeman, F. F., and Williams, W. G., J. Phys. E. 11 (1978) 459-467
[29] Surkau, R., Becker, J., Ebert, M., Grossman, T., Heil, W., Hofmann, D.,
Humblot, H., Leduc, M., Otten, E. W., Rohe, D., Siemensmeyer, K., Steiner, M.,
Tasset, F., and Trautmann N., NIM A 384 (1997) 444-450
[30] Bouchiat, M., A., Carver, T., R., and Varnum, C., M., Phys. Rev. Lett. 5 (1960)
373-375; Chupp, T., E., Coulter, K., P., Hwang, S., R., Smith, T., B., and Welsh, R.,
C., J., J. Neutron Res. 5 (1996) 11-24

3.2-33
[31] Colgrove, F., D., Schearer, L., D., and Walters, K., Phys. Rev. 132 (1963)
2561-2572
[32] Forte M. and Zeyen C. M. E., NIM A 284 (1989) 147-150
[33] Schrpf, O., in Neutron spin echo, Lecture notes in physics, vol. 128, edited
by F. Mezei, Springer-Verlag (1980) 27-52
[34] Schrpf, O., and Capellman, H., Phys. Status Solidi A 135 (1993) 359-379
[35] Jones, T. J. L., and Williams, W. G., Rutherford Laboratory Report RL-77-O79/A
(1977)
[36] Mezei, F., Z. Phys. 255 (1972) 146-160
[37] Zeyen, C. M. E., and Rem, P. C., Meas. Sci. Tech. 7 (1996) 782
[38] Dabbs, J. W. T., Roberts, L. D., and Bernstein, S., Report ORNL-CF-55-5-126.
Oak Ridge National Laoratory, USA (1955)
[39] Abrahams, K., Steinsvoll, O., Bongaarts, P. J. M., and De Lange, P. W., Rev.
Sci. Instrum. 33 (1962) 524-529
[40] Forsyth, J. B., At. Energy Rev. 17 (1979) 345-412
[41] Egelstaff, P. A., Cocking, S. J., and Alexander, T. K., Inelastic Scattering of
neutrons in solids and liquids. IAEA, Vienna (1961) 165-177
[42] Hautecler, S., Legrande, E., Vansteelandt, L., dHooghe, P., Rooms, G.,
Seeger, A., Schalt, W., and Gobert, G., Proccedings of the Conference on Neutron
Scattering in the Nineties, Jlich. IAEA, (1985) 211-215; Copley J. R. D., NIM A
303 (1991) 332-341
[43] Colwell, J. F., Miller, P. H., and Whittemore, W. L., Neutron Inelastic
Scattering, Vol. II, IAEA, Vienna (1968) 429-437
[44] Hossfeld, F., Amadori, R., and Scherm, R., Instrumentation for Neutron
Inelastic Scattering, IAEA, Vienna (1970) 117
[45] Turchin, V. F., Slow neutrons, Israel Programme for Scientific Translation
(1965)
[46] Blanc, Y., Le spectrometre temps de vol IN6: characteristiques, techniques
et performances. ILL internal Report No 82BL21G, Institut Laue-Langevin,
Grenoble (1983)
[47] Dash, J. G., and Sommers, H. S., Rev. Sci. Instrum. 24 (1953) 91-96
[48] Wagner, V., Friedrich, H., Wille, P., Physica B 180&181 (1992) 938-940

3.3-1

Detectors for Thermal Neutrons

A.Oed

Introduction
The particle- or radiation-detection is based on the measurement of electric currents.
Thermal neutrons with their low velocity and without an electric charge can only
be measured subsequent to a nuclear reaction with target atoms which emit either
ionizing radiations or ionizing particles. The target isotopes commonly used for
thermal neutron detection are indicated in Tab (1). Except for Gd, all the other
reactions are fission processes where two fission particles are ejected in opposite
directions randomly oriented in space.
The Maxwell distributions of thermal neutrons are shown in Fig (1).

Figure 1: Velocity-, energy- and wave-length-distribution for neutrons at 300 o K.

Table 1: Commonly used isotopes for thermal neutron detection, reaction

products and their kinetic energies.

3.3-2
Absorption law
A neutron flux Jo [1/s] after having passed the length of x [cm] in an absorber
with the absorption length [cm], is reduced to the value
J= Jo e - (x/)
0r the relative amount of the flux which disappears in the absorber is
(Jo-J) / Jo = 1- e - (x/)
Expressed in percentage, this leads to the efficiency of the detector or absorber
Eff = (1- e -( x/))* 100
The absorption length [cm] is inversely proportional to the product of the
cross section [cm2] and the atomic density Ad [atom/cm3] of the absorber
= 1/( *Ad)
The absorption law expressed with the cross-section therefore is
J= Jo e -( *

x* Ad)

Table 2: Cross-section, absorption length and mass-absorptiondensity for

thermal neutrons. (v = 2224 m/s; = 1.78 ; E kin = 26 m eV ; T = 300 K).
Isotopes with high cross-section and therefore used in neutron detection are
marked in bold type.

3.3-3
For gases at 1 bar, the atomic density Ad is 2.7 1019 [1/cm3] and for solids and liquids
its value is about 1000 times higher. In general, the atomic density is given by
Ad = * Na / Mv [1/cm3] where
is the volumetric weight [ g/cm 3 ], N a = 6.25 10 23 the Avogadro number
[atom/ mol ] and Mv the molar weight of the absorber-material [g /mol].
Sometimes the absorption is also expressed in terms of the mass-absorptioncoefficient = * [ g / cm2]. With the surface mass density Md =x* [g/cm2] of
an absorber of x [cm] in thickness, the law reads:
J= Jo e (Md / )
The absorption length, cross section and mass absorption-coefficient for thermal
neutrons for some materials used in neutron experiments are indicated in Tab (2)
The cross-section for thermal neutrons is inversely proportional to their velocity v.
1 / v or 1 / E where E is kinetic energy

Figure 2: Neutron cross-section as function of the kinetic energy for the

isotopes commonly used in thermal neutron detection.

3.3-4
As the Broglie wavelength = h / (m n v) with h = Planckconstant and
mn= neutron mass is also inversely proportional to the velocity v, the cross-section
increases linearly with the neutron wavelength : ~ .
157Gd and 235 U have resonances at higher energies, as shown in Fig (2)
Neutron detectors
The following list is a survey of all possible neutron detectors subdivided into integrating
and counting devices. Its characteristics are indicated as advantages and disadvantages.
Without taking into account the importance of the different qualities and especially
the costs, it seems that counting detectors have more advantages than the integrating
ones. Integrating detectors are more suitable for beam monitoring, beam alignment and
neutronography whereas the counting devices are appropriate in interference pattern
recordings for structure analysis and time-depending changes inside the samples.

INTEGRATING

COUNTING

Converter foil with

Converter foil with

photogr. film;
A1,A2,A8, D3,D4,D5,D6,D7
image plate;
A1,A2,A5,A8, D3,D4,D6,D7
CCD camera;
A1,A2,A5, D3,D4,D6,D7,D8

n-Scintillator with
CCD camera;
A1,A2,A5, D3,D4,D7
Thomson tube;
A1,A2,A5,A6,A8, D3,D4,D7
Image amplifier;
A1,D2,D3,D5,D6,D7,D8

Internal converter
ionization chamber;
A1,A4,A5,A8, D2,D3
Gd loaded image plate;
A1,A2,A5,A8, D3, D4,D7

pixel semiconductor;
A1,A2,A3,A4,A5, D6,D8
pos. sensitive. gas counter;
A3,A4A5,A6,A8, D1,D2

n-Scintillator with
pos. sensitive PM;
A1,A2,A3,A4,A5,A6, A7,D8
pos. sensitive gas counter;
A3,A4,A5,A6,A7, A8, D1,D2

Internal converter
gas-counter, -array;
A3,A4,A5,A6,A7,A8, D1,D2
multi wire prop. counter.;
A3,A4,A5,A6,A7,A8, D1,D2
micro pattern gas counter;
A1,A3,A4,A5,A6,A7, A8, D2
boron diode;
A1,A2,A3,A4,A5,A6,A7, D8

3.3-5
Advantages
A1) high intensity capability
A2) high position resolution
A3) gamma discrimination
A4) on-line read-out, short time slices
A5) high dynamic
A6) high n-efficiency
A7) low noise
A8) large sensitive area

Disadvantages
D1) limited counting capacity < 1 kHz/mm2
D2) limited position resolution > 100 cr.
D3) gamma sensitivity
D4) long read-out time > 1ms /frame
D5) low dynamic < 1/100
D6) low n-efficiency < 20 %
D7) high noise > 1 count / pixel
D8) small sensitive area < 100 cm 2

Some useful constants:

k= 1.384 10-23

Joule/ K = kg(m/s)2/ K

Boltzmann constant

h= 6.6252 10-34

VAs = Joule*s= kg (m/s)2 s

Planck constant

e= 1.62 10-19

As

elementary charge

v= 2.998 108

m/s

velocity of light

me= 9.108 10-31

kg

electron mass

mp= 1.6723 10-27

kg

proton mass

mn= 1.6747 10-27

kg

neutron mass

Na= 6.25 1023

atom/Mol

Avogadro number

Gas-detectors
Ionization chamber
The simplest neutron gas detector is the ionization chamber: a condenser consisting
of two metallic plates and the gap in between filled with 3 He or 10 BF3 gas.
The two particles emitted by the nuclear reaction are slowed down in the gas.
Along their trace they produce electron-ion pairs, and with an electric field strength
of about 100 V/(cm . bar) these charges are collected on the corresponding plates
where the current can be measured. To generate one electron-ion pair in a gas, the
ionizing particle loses about 30 eV of its energy. This is why in the n (3He, p) 3H
+0.77 MeV reaction for one neutron about 22 000 electron-ion pairs are produced
and a charge of 4.1 10-15 [As] is collected. Usually ionization chambers are used as
beam monitors.
The triton range is about 1/3 of the proton range as can be seen in Fig (4). This
means that there exists a systematic deviation between the centre of gravity of the

3.3-6

Figure 3: Neutron detection efficiency as a function of the detection gap filled

with 1 bar 3 He.
charges and the location where the reaction takes place.
This deviation can be reduced be adding other gases with higher molecular
weight, preferably composed of light atoms like C, H and F in order to be less
sensitive to x- rays and gammas. For some gases, the mean energy loss of a particle
to produce one electron-ion pair, the particle range and the mean deviation from the
reaction point are indicated in Tab (3).
Proportional counter
The small amount of charges in an ionization chamber (see above) produced by a
single neutron is difficult to measure, whereas this is easy to do with a Geiger
counter {2}. A very thin wire of several microns in diameter , which serves as
anode, is located in the axe of a cylindrical tube. This condenser is filled with gas.
The electrons produced by the ionizing particles of the nuclear reaction drift
towards the positive wire where they will experience an avalanche amplification at
the very high field strength surrounding the wire. This gas amplification is
noiseless and proportional to the amount of primarily produced charges. The gas
gain reaches factors up to 105. Single neutrons are easily detectable. A typical pulse
height spectrum of a 3He +Ar +CH4 counter is shown in Fig (5).
As a matter of fact, at 770 keV only one line would be expected,
corresponding to the total reaction energy. But some of the particle traces enter into
the housing and the charges are lost. At the point of 192 keV the reaction has taken

3.3-7

Figure 4: Ionization /mm in He of the particles emitted by the reaction

n ( 3 He, p) 3 H +0.77 MeV. Ionization and Range are calculated with the
Program :TRIM86 {1}.
place directly on the inner surface of the cylinder and the proton is stopped in the
metal. Only the triton, emitted in the opposite direction, releases its complete
energy into the gas. With a higher gas pressure the amount of counts in the tail will
be smaller as the range of the particles will be shorter. Between the neutron signals
and the gamma background there is a large gap which allows a very good gamma
discrimination. Unfortunately the local counting capacity is limited to about
1 kHz / mm2. The ions generated in the gas amplification very slowly drift towards

Table 3: Mean ionization energy, particle range and deviation in some gases

3.3-8

Figure 5: Pulse height spectrum of a 3He neutron counter irradiated with a

lab. AmBe source. In order to distinguish between gamma background and
neutron signal, a boron loaded plastic absorber was installed between source
and counter in a second measurement.
the cylinder walls. At high rates their space charge reduces the applied electric field
on the wire and hence the gas gain. For a good energy resolution, which allows a
good gamma discrimination, all the primary charges of one event must be
collected. As the ionizing particles are emitted in random directions with respect to
the wire orientation, the drift time of the electrons towards the wire varies
considerably due to the inhomogeneous cylindrical field. The drift time depends on
the gas mixture and on the local electric field strength. In general, up to 10 s are
necessary for a good charge collection. The maximal count rate of the whole
counter therefore is of about 30 kHz with a pileup of 30% of the signals. The drift
velocity of electrons in a CF4 He mixture is shown in Tab (4). With He the highest
drift velocity is reached by adding CF4.
As shown recently {4}, a counter filled with N2 +CF4 is an excellent neutron
beam monitor. For a 1 cm gap filled with 1 bar N 2 the neutron efficiency is
9.05 10-5 at a wavelength of 1.8 . In the reaction n (14N, 14C) H + 627 keV, with a
cross-section of 1.8 barn , the proton is emitted with an energy of 585 keV and
delivers a charge signal which is very well separated from the gamma background.

3.3-9
Table 4: Drift velocity of electrons in
an electric field of 1.14 V/ ( cm . mb )
in a gas mixture {3}.

Multi wire proportional chamber (MWPC)

The position sensitive variant of the Geiger counter is the MWPC invented by
Charpak{5}.
It consists of a great number of thin anode wires assembled in a plane and
mounted between two plates which are composed of separated cathode strips. The
orientation of the strips is orthogonally to each other. As with Geiger counters, the
gas amplification takes place on the wires. After the separation of the electron-ion
pairs in the avalanche by the electric field, the ion cloud influences signals on the
cathode strips which allows to determine the position of the event. The achievable
position resolution perpendicular to the wires depends on the wire spacing.
With a wire spacing smaller than one millimeter the operation of this device is
extremely difficult. As the electric field between the anode frame and the cathodes is
homogeneous, the distribution of the drift time of the primary charges is more uniform
compared to a cylindrical counter and therefore the time for the charge-collection is
reduced to about 1 s. As the counting capacity only depends on this collection time,
the rate limit for a two-dimensional detector increases to about 300 kHz for the whole
sensitive surface, with a position error of 30 % at this rate. To record a higher rate, the
detector has to be segmented. A one dimensional detector with individual read-out presents such a segmentation. Each cell is able to record the rate of 300 kHz. In any case
the local counting capacity is limited to about 1 kHz/ mm2 just as for the counter tube.
Micro Pattern Gas Counter ( MPGC)
To improve the counting capacity, the cathode distance has to be reduced for a
faster evacuation of the avalanche ions; for a better position resolution, the
structure pitch of the electrodes has to be smaller. This aim has been reached with
the recently developed Micro Pattern Gas Counters. Their very small electrode
structures are manufactured by means of the photolithographic technique, a
common procedure to fabricate integrated circuits.
The first type was the Micro Strip Gas Chamber (MSGC) {6} which will be
shortly described. Its arrangement corresponds to an ionization chamber but the anode
plate is made of a glass plate on the surface of which very thin conductor-strips are

3.3-10
fixed. Small strips, with a width of several microns, are located between two larger
strips in a distance of several ten microns. The applied electric potential alternates
between each strip. The smaller strips are the anodes, the larger ones the cathodes. This
micro strip plate operates just like a proportional counter. An electron produced
anywhere in the gas volume drifts towards the positive strip where it will experience
an avalanche amplification. But now more than 90 % of the ions produced in the
avalanche are neutralised on the very close cathode strips and therefore the space
charge is significantly reduced. A local rate limit in neutron detection of 20 kHz / mm 2
was reached , 20 times higher than in a wire counter. The overall rate limit is the same
as in a MWPC, as it only depends on the charge collection time. In neutron detection
the smaller pitch of such a structure does not automatically lead to a better localisation
because the precision also depends on the intrinsic deviation between the reaction
point and the centre of gravity of the charges. The spatial resolution is only improved
if the particle range is reduced to the size of the structure pitch.
A very recent development is the Gas Electron Multiplier (GEM) {7}. Its
amplifying part consists of a Polyimide foil metallized on both sides with small
holes drilled through. A potential difference between the two surfaces generates the
necessary field strength for the avalanche amplification. This device is inexpensive
and can be produced in large sizes up to 30 cm x 30 cm. A survey of the recent
MPGC developments can be found in {8} and {9}.
The general characteristics of gaseous neutron detectors are:
1) high and noiseless internal amplification
2) very good gamma discrimination
3) large sensitive areas
4) radiation hard
5) max local rate: <= 1 kHz/ mm2 respectively < = 20 kHz/ mm2
6) total rate per cell or per sensitive surface < = 300 kHz
7) position resolution > = 1 mm
Scintillator detectors
In Fig (6) the pulse height spectra of two neutron scintillators, directly coupled to a
photomultiplier, are shown. Only the glass scintillator leads to a tolerable
separation between neutron- and gamma-signals. The photon output of the lithium
loaded ZnS- powder is higher but this is also true for its gamma sensitivity.
The characteristics of some neutron scintillators are indicated in Tab (5).
As a result of the higher density of the scintillator material, the range of the
ionizing particles is reduced to a few microns which can result in a better spatial
resolution. The light emitted from the scintillator is either directly recorded by a

3.3-11

Figure 6: Pulse height spectra of the scintillators NE 905 and NE 422

(Nuclear Enterprises) irradiated with a lab. Am-Be neutron source
(reaction: n ( 6Li , ) 3H + 4.79 MeV ) and measured with a photo multiplier
at the same amplification. In a second measurement a boron loaded plastic
absorber was installed between source and scintillator in order to distinguish
between gamma background and neutron signal.

light sensitive detector or collected by different optical means as there are lenses,
mirrors or fibres towards such a detector. This can be a photographic film, a CCD
camera, a photo multiplier (PM) or a photo-sensitive gaseous counter. But all the
integrating light-devices also record the light generated by the gammas.
On a continuous neutron beam the lithium loaded ZnS scintillator NE422 is
applicable only for measurements where its gamma sensitivity is of little
importance for the result: in beam- monitoring or neutronography. The gamma
background can only be discriminated on a pulsed neutron beam and with a
counting light detector, taking advantage of the time of flight difference between
neutrons and gammas.
Owing to the short decay time of the glass-scintillator NE 905, a local rate up to
several MHz / mm2 can be recorded with a directly connected PM. This is also valid
for a position sensitive PM. The combination of the latter with this scintillator is the
neutron detector with the highest rate capability and with a spatial resolution of about

Table 5: Characteristics of the most useful scintillators for thermal neutrons {12}.

3.3-12
200 microns. The maximum sensitive area though is still limited to 8 x 8 cm2 and the
costs for this arrangement are rather high. Unfortunately the light output of the
NE 905 scintillator is too low for single neutron detection with a CCD.
The Anger camera {10} is the combination of a scintillator and a light disperser
coupled to a bundle of photomultipliers. The localisation of the neutron event is
achieved by determining the centroid between the different PM signals. Its sensitive
area is up to several hundred cm2 and its position resolution is of several mm.
The commercially available Thomson tube looks like an inverse cathode ray
tube. The screen is replaced by a Gd Oxidescintillator, covered with a
photocathode. The light produced by the Gd-conversion electron in the scintillator
releases an electron from the photocathode into the vacuum where it is accelerated
and focused, by an electrostatic optic, onto an output screen. There it produces a
huge amount of photons. This spot can be recorded by all the light sensitive devices
mentioned above. The sensitive area has a diameter of 215 mm and a spatial
resolution of 200 to 300 microns.
For beam monitoring, a so-called Handmonitor {11} combines the lithium
loaded ZnS scintillator with a commercially available image amplifier. A neutron
beam profile can be directly seen on the screen of the device.
Photographic films have a dynamic of 1:100 at the most. A spatial resolution
of some hundred microns can be achieved depending on the optic and the thickness
of the scintillator. But as this is an integrating device, the light of the gamma
background is also recorded.
At present, R&D work is done for inorganic scintillators with a higher light
output for the future spallation neutron sources. A summary is published by Carel
W.E. van Eijk {12}
Foil detectors
Fig (7) shows the calculated pulse height spectra of the charged particles escaped
from a boron foil after the n (10B, ) 7Li reaction. The spectrum is always extended
down to the lowest channels. As the charged particles are emitted randomly in space,
their energy losses during the travelling through the material can be as high as their
complete kinetic energy, that is why only some of them escape from the foil.
The absorption length for neutrons with a wavelength of 1.8 in a metallic
10 B-foil is 19.2 microns, whereas the ranges of the particles of the reaction
n (10B,) 7Li amount to 3.9 microns respectively 1.7 microns only. This means that
the particle range is much shorter than the absorption length. Particles produced
deeper than this range cannot escape from the foil. The situation is similar for all

3.3-13

Figure 7: Calculated spectra of the charged particles escaped on the

backside from frontally irradiated 10B foils and produced in the reactions
n (10B, ) 7Li + 2.3 MeV + (0.48 MeV) and n (10B, ) 7Li + 2.79 MeV. The
efficiency for 1.8 neutrons with the indicated foil thickness would be 1.0 %,
respectively 5.5 %.
the other neutron converter foils. Therefore any combination of a particle detector
with a foil leads to an arrangement with low detection efficiency. At a foil thickness
which is about 10% smaller than the range of the particle with the higher kinetic
energy, the escape probability reaches a maximum as shown in Fig. (8). The foil
thickness for a maximum neutron detection for commonly used converters is
indicated in Tab (6).
Foil detectors are preferably used in neutronography and beam monitoring.
For the latter a very thin layer of the
absorbing material is evaporated onto
the cathode inside a gas counter or
ionization chamber. For special
applications, foil detectors are
constructed in combination with
photographic films, image plates,
CCDs and channel plates; in most cases
Gd foils are used because of their higher
neutron efficiency and an easier
handling but the disadvantage is a high
gamma sensitivity.
Figure 8: Escape probability of the
particles produced by a neutron
reaction in a foil as function of its
thickness.

3.3-14
Table 6: Foil thickness and
maximum detection efficiency
for neutrons with a wavelength
of 1.8 for frontal irradiation
and backside detection.

Other internal detectors

In a special image plate for neutron detection the storage phosphor is mixed with
Gd2O3 {11,13}. By this means the efficiency can be increased: the conversion
electron does not have to escape from a foil but is stopped in the phosphor whereby
the latter will be excited. Such image plates are available in a size of 20 x 20 cm2.
The dynamic of an image plate amounts to 1: 105. The spatial resolution of a
neutron detector is of about 100 to 200 microns. For these reasons the device finds
favour in protein crystallography with its huge amount of interference spots. But
like all the integrating detectors and particularly because of the incorporated Gd
with its higher atomic number, such a plate is rather gamma sensitive.
An excellent neutron detector which has not yet been constructed will be
proposed here: The boron-diode. Boron is, like silicon or germanium, a
semiconductor with a band gap of 1.4 eV. In the thirties, photographs used boron
photocells just as silicon cells are used today. With a surface barrier diode of boron
or a boron diode doped with silicon, a high signal of each neutron could be
measured. An energy of about 6 eV is necessary to produce one electronhole pair
in a semiconductor. 2.3 MeV are available in the n (10B, ) 7Li reaction , and
therefore a charge of 6 10 14 [As] is generated. With an integrating operation
amplifier and a capacitor of 1 pF in its loop an output signal of 60 mV could be
measured. With a boron crystal thickness of only 60 microns, the neutron (1.8 )
detection efficiency is of 97 %.
A position sensitive detector would be a CCD or a pixel device made of a
boron crystal. and a position resolution of less than 10 microns would be achieved
as the range of the particle is only 3.9 microns. Moreover the gamma sensitivity
would be rather low as boron has a low atomic number.
FAQ
find their answer in the textbook of Glenn F. Knoll {14}.

3.3-15
References
[1] The stopping and range of ions in solids: J.F. Ziegler, J.P. Biersack and
U. Littmark; Pergamon Press Inc., New York, USA (1985)
[2] H. Geiger, E. Rutherford and J.J. Thomson; Phil. Mag .XIII (1896) 392
[3] M.K. Kopp et al NIM 201 (1982) 395
[4] D. Feltin and B. Gurard, ILL NEWS 34 (2000) 8; Edit. G. Cicognani
http://www.ill.fr/News/34/034_NEWS.HTM
[5 ] G. Charpak ; NIM 62 (1968) 262
[6 ] A. Oed ; NIM A 263 (1988) 35
[7 ] F. Sauli ; NIM A 386 (1997) 531
[8 ] F. Sauli and A. Sharma ; Annual Rev. of Nucl. and Particle Sci. 49 ( 1999) 341
[9 ] A. Oed ; NIM A471 (2001) 109
[10 ] H. O. Anger ; Rev Sci. Instrum. 29 (1958) 27
[11 ] Ch. Rausch et al; SPIE Proc. 1737 (1992) 255
[12 ] Carel W.E. van Eijk ; NIM A 460 (2001) 1
[13] D.A.A. Myles et al; Physica B 241 - 243 ( 1998 ) 1122
[14] Radiation detection and measurement: Glenn F. Knoll; 3rd ed.1999;John Wiley
& Sons, Inc, New York

4.1-1

ACTIVATION

AND

SHIELDING

Activation Table of the Elements

M. Johnson, S. A. Mason and R. B. Von Dreele

The best way to determine the activation of a sample is to measure it in situ

with appropriate instruments. However in planning neutron experiments it may be
useful to know whether activation is likely to be significant. Properly estimating
activation of a sample by a neutron beam requires knowledge of the neutron spectrum, time of exposure, mass, isotopic composition, etc. The Table below allows
you to calculate roughly the activation of a sample after it has been in a neutron
beam for one day and the amount of time for it to decay to 74Bq/g (i.e. 2nCi/g) or
less, which is a typical limit for shipping a sample as nonradioactive. It also displays the anticipated exposure you may receive when removing the sample from the
instrument. The entries in this table are derived from an approximate calculation
(by M. Johnson) for decay times to 105 and to 104Bq/cm3 for 5cm3 pure solid samples of the elements exposed to a neutron beam for 1day at an intensity comparable
to that found on HIPD with LANSCE operating at 100A. These calculations were
made for pulsed sources and may somewhat overestimate activation for reactors for
some elements (no epithermal neutrons). They are augmented by calculations from
NIST of the activation from a 1 day exposure to a 107n/s-cm2 reactor thermal beam
(marked ). Storage time is the time required for a sample of the pure solid element
exposed to this standard neutron beam to decay to 74Bq/g or less. Prompt activation gives the anticipated activation for the pure solid elements 2 min after the
neutron exposure ceases. Contact dose is that expected from a 1g sample of the
pure element from the prompt activation. Elements with a dash for the entries in all
three columns do not show any activation. Those marked with a single asterisk are
radioactive before exposure to the neutron beam; apart from Tc and Pm, they are all
-particle emitters. Bismuth is a special case; it is stable before exposure to the

4.1-2
beam, but the activation product is an -emitter. A typical sample activation calculation may be found in the Web version of the table.
NB: The current Web version of this Table may be found on the ILL Web site:
(http://www.ill.fr/YellowBook/D19/help/act_table.htm).
The original Table first appeared in the LANSCE Newsletter 12/1/92. We thank
R. Pynn and V. T. Forsyth, for their cooperation. Comments to the present authors are
very welcome ([email protected], [email protected], [email protected]).
Symbol/Name

Ac
Al
Am
Sb
Ar
As
At
Ba
Bk
Be
Bi
B
Br
Cd
Ca
Cf
C
Ce
Cs
Cl
Cr
Co
Cu
Cm
Dy

actinium
aluminium
americium
antimony
argon
arsenic
astatine
barium
berkelium
beryllium
bismuth
boron
bromine
cadmium
calcium
californium
carbon
cerium
cesium
chlorine
chromium
cobalt
copper
curium
dysprosium

Mass

227
26.982
243
121.75
39.948
74.922
210
137.34
247
9.012
208.980
10.811
79.909
112.40
40.08
249
12.011
140.12
132.905
35.453
51.996
58.933
63.54
247
162.50

Storage
Time

*
21m
*
520d
19h
18d
*
<150h
*
**
18d
190d
*
<86h
54h
<2.8h
<61d
24y
7.4d
*
52h

Prompt activation Contact dose at 2.54cm

1 unit=37Bq/g
1 unit=10Gy/hr/g
(= 1nCi/g)
(=1mr/hr/g, at 1 inch)
*
1900
*
800
3500
8.4x104
*
<80
*
**
1.4x104
370
*
<40
4.6x105
<80
<40
5.2x104
1.0x104
*
5.0x105

*
2.0
*
0.7
3.0
7.3
*
<0.1
*
**
12
0.3
*
<0.1
400
<0.1
<0.1
45
8.5
*
430

4.1-3
Symbol/Name

D
Es
Er
Eu
Fm
F
Fr
Gd
Ga
Ge
Au
Hf
He
Ho
H
In
I
Ir
Fe
Kr
La
Pb
Li
Lu
Mg
Mn
Md
Hg
Mo
Nd
Ne
Np
Ni
Nb
N

Mass

deuterium
2.015
einsteinium
254
erbium
167.26
europium
151.96
fermium
253
fluorine
18.998
francium
223
gadolinium 157.25
gallium
69.72
germanium 72.59
gold
196.967
hafnium
178.49
helium
4.003
holmium
164.930
hydrogen
1.008
indium
114.82
iodine
126.904
iridium
192.2
iron
55.847
krypton
83.80
lanthanum 138.91
lead
207.19
lithium
6.939
lutetium
174.97
magnesium 24.312
manganese 54.938
mendelevium 256
mercury
200.59
molybdenum 95.94
neodymium 144.24
neon
20.183
neptunium
237
nickel
58.71
niobium
92.906
nitrogen
14.007

Storage
Time
*
78d
50y
*
*
11d
8d
<6d
29d
1.6y
20d
12d
7h
4.2y
42h
22d
1.8y
38h
*
24d
30d
15h
*
<5.5h
80m
-

Prompt activation Contact dose at 2.54cm

1 unit=37Bq/g
1 unit=10Gy/hr/g
(= 1nCi/g)
(=1mr/hr/g, at 1 inch)
*
*
600
0.5
2200
1.9
*
*
*
*
7400
6.4
3.2x104
27
1100
1.0
3000
2.5
620
0.5
4
2.8x10
24
1.1x104
9.5
5
1.2x10
100
5.0x104
43
3200
2.8
4
1.9x10
16
4
1.4x10
12
1.1x105
95
*
*
700
0.6
430
0.4
1200
1.0
*
*
<30
<0.1
2.0x104
17
-

4.1-4
Symbol/Name

Os
O
Pd
P
Pt
Pu
Po
K
Pr
Pm
Pa
Ra
Rn
Re
Rh
Rb
Ru
Sm
Sc
Se
Si
Ag
Na
Sr
S
Ta
Tc
Te
Tb
Tl
Th
Tm
Sn
Ti

Mass

osmium
190.2
oxygen
15.999
palladium
106.4
phosphorus 30.974
platinum
195.09
plutonium
242
polonium
210
potassium
39.102
praseodymium140.907
promethium 147
proctactinium 231
radium
226
radon
222
rhenium
186.2
rhodium
102.905
rubidium
85.47
ruthenium
101.07
samarium
150.35
scandium
44.956
selenium
78.96
silicon
28.086
silver
107.870
sodium
22.991
strontium
87.62
sulphur
32.064
tantalum
180.948
technetium
98
tellurium
127.60
terbium
158.924
thallium
204.37
thorium
232.038
thulium
168.934
tin
118.69
titanium
47.90

Storage
Time
41d
9d
20d
*
*
<38h
11d
*
*
*
*
53d
2h
56d
106d
35d
<1.8y
10h
7.4y
5.5d
<25h
3y
*
96h
2.1y
41m
*
3.7y
<50d
-

Prompt activation Contact dose at 2.54cm

1 unit=37Bq/g
1 unit=10Gy/hr/g
(= 1nCi/g)
(=1mr/hr/g, at 1 inch)
2300
7.1x104
230
*
*
<300
2.0x104
*
*
*
*
4.9x104
2.6x104
1800
230
6200
<90
4900
1.6x104
5700
<100
1600
*
2600
3300
460
*
7700
<40
-

2.0
60
0.2
*
*
<0.3
17
*
*
*
*
42
22
1.6
0.2
5.4
<0.1
4.2
14
5.0
<0.1
1.4
*
2.2
2.8
0.4
*
6.7
<0.1
-

4.1-5
Symbol/Name

W
U
V
Xe
Yb
Y
Zn
Zr

tungsten
uranium
vanadium
xenon
ytterbium
yttrium
zinc
zirconium

Mass

Storage
Time

183.85
238.03
50.942
131.30
173.04
88.905
65.37
91.22

15d
*
48m
7d
275d
24d
5d
79h

Prompt activation Contact dose at 2.54cm

1 unit=37Bq/g
1 unit=10Gy/hr/g
(= 1nCi/g)
(=1mr/hr/g, at 1 inch)
3.7x104
*
4.7x105
3200
780
1000
1600
<40

32
*
41
2.8
0.7
0.9
1.4
<0.1

4.2-1

SHIELDING OF RADIATIONS
H.G. Brner and J. Tribolet

Shielding of photons
The decrease in intensity of a parallel beam of photons traversing an absorber of
thickness d is given by
= 0 e d
where 0 and are the beam intensity before and after passing through the
absorber, is the attenuation coefficient for photons of energy E and for a given
material. It should be noted that the attenuation calculated by the above formula
gives only the decrease in intensity of the original beam. The total radiation
downstream of the absorber is larger, due to the presence of scattered photons, and
to the creation of secondary photons by a variety of processes, including
fluorescence and positron annihilation radiation.
The graphs in Figure 1 give values for the attenuation coefficient (expressed
in cm -1 ) for some commonly used materials like iron, lead, aluminium and
concrete. One can deduce that one needs roughly 5 cm of lead or 50 cm of concrete,
respectively, to attenuate a 10 MeV photon beam (a typical primary gamma ray
energy produced in neutron capture) by an order of magnitude.
Shielding of neutrons
Concerning the attenuation of neutrons one has to distinguish between two
categories: Thermal and subthermal neutrons on one hand and epithermal and fast
neutrons on the other one.
Thermal and subthermal neutrons can easily be stopped by capture in
materials with high neutron capture cross sections like cadmium, gadolinium,
boron and lithium. Generally a rather thin layer of absorbant material is sufficient
to stop such neutron beams. However, in the case of cadmium and gadolinium the
absorption of neutrons is accompanied by strong emission of capture gamma rays.
Therefore additional shielding is necessary to attenuate the gamma rays (see
above). This phenomenon does not exist for lithium and is much less pronounced
for boron. Consequently the latter materials are preferably used, especially on the
outside of heavy shielding in experimental areas.

4.2-2

Figure 1: Attenuation of photons.

4.2-3
To attenuate neutrons of higher energies is by far more complex because they
cannot be stopped directly via capture and have first to be slowed down. A precise
determination of a specific protection demands generally more complicated
calculations using Monte Carlo methods. For neutrons in the MeV region, three
types of materials are generally added together: dense materials (inelastic
collisions), hydrogenated materials (moderation), and absorbing materials
(capture). An example is heavy concrete in which one finds iron, hydrogen and
boron.

Figure 2: High energy part of the spectral distribution of neutrons emerging

from a standard thermal beam tube.
Specific examples:
a) Neutron guides: In the parts which are sufficiently distant from the reactor
neutrons are stopped by one to two layers (concerning ILL standards) of B4C.
b) Primary casemates: Typically 80 to 90 cm of heavy concrete are used (at ILL).
This insures simultaneous protection against neutrons and gamma rays.

4.2-4
c) Thermal neutron beam tubes: An example for the high energy part of the spectral
distribution of neutrons in a standard thermal beam tube of ILL is shown in
Figure 2. Optimized shielding against neutrons with such spectral distribution
depends on the space available:
If 10 cm: Best attenuation is obtained by using borated polyethylene
(attenuation by factor 25)
If 20 cm: Best attenuation is obtained by using 2 cm of iron, followed by 18
cm of borated polyethylene (attenuation by factor 310). However, the use of 20
cm of borated polyethylene is only about 10 percent less efficient.
If 30 cm: Best attenuation is obtained by using 12 cm of iron, followed by 18
cm of borated polyethylene ( attenuation by factor 3400).

5.1-1

MISCELLANEOUS
Radioactivity and Radiation
Protection
The International Commission on Radiation Units and Measurements (ICRU)
recommends the use of SI units. Therefore, we list SI units first, followed by cgs (or
other common) units in parentheses, where they differ.

5.1-2

5.2-1

Physical Constants

5.2-2

5.3-1

Physical Properties of the Elements

Table 5.3 lists several important properties of the elements.
Data were taken mostly from D. R. Lide, Ed., CRC Handbook of Chemistry and
Physics, 80th ed. (CRC Press, Boca Raton, Florida, 1999). Atomic weights apply to
elements as they exist naturally on earth; values in parentheses are the mass numbers for the longest-lived isotopes. Some uncertainty exists in the last digit of each
atomic weight. Specific heats are given for the elements at 25C and a pressure of
100 kPa. Densities for solids and liquids are given as specific gravities at 20C
unless otherwise indicated by a superscript temperature (inC); densities for the
gaseous elements are given in g/cm3 for the liquids at their boiling points.

5.3-2

5.3-3

5.3-4

This table is adapted with permission from the 2001 web edition of the X-Ray Data
Booklet (http://xdb.lbl.gov).

5.4

Disclaimer
Neither ILL nor any of its employees, nor any of the contributors to the
present document makes any warranty, express or implied, or assumes any legal
liability or responsibility for the accuracy, completeness, or usefulness of any
information, apparatus, product or process disclosed, or represents that its use
would not infringe privately owned rights.
The user accepts sole responsibility and any risks associated with the use and
results of material in this booklet, irrespective of the purpose to which such use or
results are applied.