by

Roger Pynn
Los Alamos
National Laboratory

LECTURE 1: Introduction & Neutron Scattering Theory

Overview
1. Introduction and theory of neutron scattering
1.
2.
3.
4.
5.
6.
7.
8.

2.
3.
4.
5.
6.

Advantages/disadvantages of neutrons
Comparison with other structural probes
Elastic scattering and definition of the structure factor, S(Q)
Coherent & incoherent scattering
Inelastic scattering
Magnetic scattering
Overview of science studied by neutron scattering
References

Neutron scattering facilities and instrumentation

Diffraction
Reflectometry
Small angle neutron scattering
Inelastic scattering

Why do Neutron Scattering?

To determine the positions and motions of atoms in condensed matter
1994 Nobel Prize to Shull and Brockhouse cited these areas
(see http://www.nobel.se/physics/educational/poster/1994/neutrons.html)

Neutron advantages:

Wavelength comparable with interatomic spacings

Kinetic energy comparable with that of atoms in a solid
Penetrating => bulk properties are measured & sample can be contained
Weak interaction with matter aids interpretation of scattering data
Isotopic sensitivity allows contrast variation
Neutron magnetic moment couples to B => neutron sees unpaired electron spins

Neutron Disadvantages
Neutron sources are weak => low signals, need for large samples etc
Some elements (e.g. Cd, B, Gd) absorb strongly
Kinematic restrictions (cant access all energy & momentum transfers)

The 1994 Nobel Prize in Physics Shull & Brockhouse

Neutrons show where the atoms are.

and what the atoms do.

The Neutron has Both Particle-Like and Wave-Like Properties

Mass: mn = 1.675 x 10-27 kg
Charge = 0; Spin =
Magnetic dipole moment: n = - 1.913 N
Nuclear magneton: N = eh/4m p = 5.051 x 10-27 J T-1
Velocity (v), kinetic energy (E), wavevector (k), wavelength (),
temperature (T).
E = mnv2/2 = kBT = (hk/2)2/2mn; k = 2 / = mnv/(h/2)

Cold
Thermal
Hot

Energy (meV)
0.1 10
5 100
100 500

(nm) = 395.6 / v (m/s)

E (meV) = 0.02072 k2 (k in nm-1)

Temp (K)
1 120
60 1000
1000 6000

Wavelength (nm)
0.4 3
0.1 0.4
0.04 0.1

Comparison of Structural Probes

Note that scattering methods
provide statistically averaged
information on structure
rather than real-space pictures
of particular instances

Macromolecules, 34, 4669 (2001)

Thermal Neutrons, 8 keV X-Rays & Low Energy

Electrons:- Absorption by Matter
Note for neutrons:
H/D difference
Cd, B, Sm
no systematic A
dependence

Interaction Mechanisms

Neutrons interact with atomic nuclei via very short range (~fm) forces.
Neutrons also interact with unpaired electrons via a magnetic dipole
interaction.

Brightness & Fluxes for Neutron &

X-Ray Sources
Brightness
(s-1 m-2 ster-1)

dE/E
(%)

Divergence
(mrad2)

15

10 x 10

10

16

0.5 x 10

5 x 10

24

0.01

0.1 x 5

5 x 10
10

Neutrons

10

Rotating
Anode

10

Bending
Magnet

10

Wiggler

10

26

0.01

0.1 x 1

Undulator
(APS)

1033

0.01

0.01 x 0.1

Flux
(s-1 m-2)
11

10

17

19

1024

Cross Sections

= number of incident neutrons per cm2 per second

= total number of neutrons scattered per second /
d number of neutrons scattered per second into d
=
d
d
d 2
number of neutrons scattered per second into d & dE
=
ddE
d dE

measured in barns:
1 barn = 10-24 cm2
Attenuation = exp(-Nt)
N = # of atoms/unit volume
t = thickness

Scattering by a Single (fixed) Nucleus

range of nuclear force (~ 1fm)
is << neutron wavelength so
scattering is point-like
energy of neutron is too small
to change energy of nucleus &
neutron cannot transfer KE to a
fixed nucleus => scattering is
elastic
we consider only scattering far
from nuclear resonances where
neutron absorption is negligible
If v is the velocity of the neutron (same before and after scattering), the number of neutrons
passing through an area dS per second after scattering is :
2

v dS scat = v dS b 2 /r 2 = v b 2 d
Since the number of incident neutrons passing through unit areas is : = v incident = v
2

d v b 2 d
=
= b2
d
d

so total = 4b 2

Adding up Neutrons Scattered by Many Nuclei

r r
r
i k 0 .R i
At a nucleus located at R i the incident wave is e

so the scattered wave is scat = e

d vdS scat

=
d
vd

r r
i k0 .Ri

-b
r r r
r ri e ik '.(r Ri )
r - Ri

r r
r r r
dS
1
i
(
k
=
bi e ik '.r r r e 0 k ') .Ri
d
r - Ri

If we measure far enough away so that r >> R i we can use d = dS/r 2 to get
r r
r r r
r r
d
i ( k 0 k ') .( Ri R j )
iQ.( Ri R j )
= bi b j e
= bi b j e
d i , j
i, j

where the wavevector transfer Q is defined by

r r r
Q = k ' k0

Coherent and Incoherent Scattering

The scattering length, bi , depends on the nuclear isotope, spin relative to the
neutron & nuclear eigenstate. For a single nucleus:

bi = b + bi where b averages to zero

bi b j = b + b (bi + b j ) + bib j
2

but b = 0 and bi b j vanishes unless i = j

b

2
i

= bi b

= b
d

= b b
2

r r r
iQ .( Ri R j )

+ ( b2 b ) N
2

i, j

Coherent Scattering
(scattering depends on the
direction of Q)

Incoherent Scattering
(scattering is uniform in all directions)

Note: N = number of atoms in scattering system

Values of coh and inc

Nuclide

coh

inc

Nuclide

coh

inc

1H

1.8

80.2

0.02

5.0

2H

5.6

2.0

Fe

11.5

0.4

5.6

0.0

Co

1.0

5.2

4.2

0.0

Cu

7.5

0.5

Al

1.5

0.0

36Ar

24.9

0.0

Difference between H and D used in experiments with soft matter (contrast variation)
Al used for windows
V used for sample containers in diffraction experiments and as calibration for energy
resolution
Fe and Co have nuclear cross sections similar to the values of their magnetic cross sections
Find scattering cross sections at the NIST web site at:
http://webster.ncnr.nist.gov/resources/n-lengths/

Coherent Elastic Scattering measures the Structure

Factor S(Q) I.e. correlations of atomic positions
r
d
2
= b N.S (Q)
d
r r
iQ.Ri

Now e
i

r
1
for an assemblyof similar atoms where S (Q) =
N

r r r
iQ.( Ri R j )

i, j

ensemble

r iQr .rr
r r
r iQr .rr
r
= dr .e (r Ri ) = dr .e N (r ) where N is the nuclear number density
i

r
1
r iQr .rr
r 2
so
S (Q) =
dr .e N (r )

N
r 1 r r iQr .(rr rr ')
r r iQr .Rr
r
r
1
r
r r
or S(Q) = dr ' dr .e
N (r ) N (r ' ) = dR dr e
N (r ) N (r R)
N
N
r
r r iQr .Rr
ie
S (Q) = 1 + dR.g (R).e
r
r r r
where g(R) = (R Ri + R0 )

r
is a function of R only.

i0

g(R) is known as the static pair correlation function. It gives the probability that there is an
atom, i, at distance R from the origin of a coordinate system at time t, given that there is also a
(different) atom at the origin of the coordinate system

S(Q) and g(r) for Simple Liquids

Note that S(Q) and g(r)/ both tend to unity at large values of their arguments
The peaks in g(r) represent atoms in coordination shells
g(r) is expected to be zero for r < particle diameter ripples are truncation
errors from Fourier transform of S(Q)

Neutrons can also gain or lose energy in the scattering process: this is
called inelastic scattering

Inelastic neutron scattering measures atomic motions

The concept of a pair correlation function can be generalized:
G(r,t) = probability of finding a nucleus at (r,t) given that there is one at r=0 at t=0
Gs(r,t) = probability of finding a nucleus at (r,t) if the same nucleus was at r=0 at t=0
Then one finds:

r
d 2
2 k'

= bcoh NS (Q, )
k
d.dE coh
r
d 2
2 k'

= binc NSi (Q , )
k
d.dE inc
where
r r
r
1
r
i ( Q .r t ) r
S (Q, ) =
G (r , t )e
dr dt and
2h

(h/2)Q & (h/2) are the momentum &

energy transferred to the neutron during the
scattering process

r r
r
1
r
i ( Q. r t ) r
S i (Q , ) =
Gs ( r , t ) e
dr dt
2h

Inelastic coherent scattering measures correlated motions of atoms

Inelastic incoherent scattering measures self-correlations e.g. diffusion

Magnetic Scattering
The magnetic moment of the neutron interacts with B fields
caused, for example, by unpaired electron spins in a material
Both spin and orbital angular momentum of electrons contribute to B
Expressions for cross sections are more complex than for nuclear scattering
Magnetic interactions are long range and non-central

Nuclear and magnetic scattering have similar magnitudes

Magnetic scattering involves a form factor FT of electron spatial distribution
Electrons are distributed in space over distances comparable to neutron wavelength
Elastic magnetic scattering of neutrons can be used to probe electron distributions

Magnetic scattering depends only on component of B perpendicular to Q

For neutrons spin polarized along a direction z (defined by applied H field):
Correlations involving Bz do not cause neutron spin flip
Correlations involving Bx or By cause neutron spin flip

Coherent & incoherent nuclear scattering affects spin polarized neutrons

Coherent nuclear scattering is non-spin-flip
Nuclear spin-incoherent nuclear scattering is 2/3 spin-flip
Isotopic incoherent scattering is non-spin-flip

Magnetic Neutron Scattering is a Powerful Tool

In early work Shull and his collaborators:
Provided the first direct evidence of antiferromagnetic ordering
Confirmed the Neel model of ferrimagnetism in magnetite (Fe3O4)
Obtained the first magnetic form factor (spatial distribution of magnetic
electrons) by measuring paramagnetic scattering in Mn compounds
Produced polarized neutrons by Bragg reflection (where nuclear and magnetic
scattering scattering cancelled for one neutronspin state)
Determined the distribution of magnetic moments in 3d alloys by measuring
diffuse magnetic scattering
Measured the magnetic critical scattering at the Curie point in Fe

More recent work using polarized neutrons has:

Discriminated between longitudinal & transverse magnetic fluctuations

Provided evidence of magnetic solitons in 1-d magnets
Quantified electron spin fluctuations in correlated-electron materials
Provided the basis for measuring slow dynamics using the neutron spin-echo
technique..etc

Ne utro ns in Co n de ns e d Matte r Re s e arc h

10 000

10 0

- Chopper
Spec
SSPSS
pa lla tio
n - Ch opp
er tro meter
ILL - w itho u t s pin -ec ho
ILL - w ith s p in -e ch o

Ene rgy Trans fe r (me V)

Elastic Scattering

1
[Larger Objects R esolved]
Micelles Polymers
Prot eins in Solution
Viruses

M etallu rgical
Syst ems
Colloids

N eutron
Induced
Excit ations

Itine rant
Hyd rog en
Magnets
Modes
Cryst al
Momentum
Fields
Dist ribut ions
Molecular
Spin
Coherent
Vibrations
Wa ves
M odes in
La tt ice
Glasses
Vibra tions
and Liquids
and
Elec tronA nharmonicit y
phonon
Molecular
Int era ctions
M otions
Accurate Liquid Structu res
C rystal and
M em branes
Amorphous
Magnetic
Precision Crystallography
Proteins
Syst ems
Struct ures
Anharmonicit y
Crit ic al
S ca ttering

0.01
Aggregate
Mot ion
Polymers
and
Biological
Syst ems

10 -4

Slow er
M otions
Res olved

Diffusive
Modes
Molecular
Reorie nta tion
Tunneling
S pectroscopy
Surface
Eff ects ?

10 -6
0 .0 01

0.01

0 .1

-1
Q ( )

10

100

Neutron scattering experiments measure the number of neutrons scattered at different values of
the wavevector and energy transfered to the neutron, denoted Q and E. The phenomena probed
depend on the values of Q and E accessed.

Next Lecture
2. Neutron Scattering Instrumentation and Facilities how is
neutron scattering measured?
1. Sources of neutrons for scattering reactors & spallation sources
1. Neutron spectra
2. Monochromatic-beam and time-of-flight methods

2. Instrument components
1.
2.
3.
4.
5.
6.

Crystal monochromators and analysers

Neutron guides
Neutron detectors
Neutron spin manipulation
Choppers
etc

3. A zoo of specialized neutron spectrometers

References
Introduction to the Theory of Thermal Neutron Scattering
by G. L. Squires
Reprint edition (February 1997)
Dover Publications
ISBN 048669447
Neutron Scattering: A Primer
by Roger Pynn
Los Alamos Science (1990)
(see www.mrl.ucsb.edu/~pynn)

by
Roger Pynn
Los Alamos
National Laboratory

LECTURE 2: Neutron Scattering Instrumentation & Facilities

Overview
1. Essential messages from Lecture #1
2. Neutron Scattering Instrumentation and Facilities how is
neutron scattering measured?
1. Sources of neutrons for scattering reactors & spallation sources
1. Neutron spectra
2. Monochromatic-beam and time-of-flight methods

2. Instrument components
3. A zoo of specialized neutron spectrometers

Recapitulation of Key Messages From Lecture #1

Neutron scattering experiments measure the number of neutrons scattered
by a sample as a function of the wavevector change (Q) and the energy
change (E) of the neutron
Expressions for the scattered neutron intensity involve the positions and
motions of atomic nuclei or unpaired electron spins in the scattering sample
The scattered neutron intensity as a function of Q and E is proportional to
the space and time Fourier Transform of the probability of finding two atoms
separated by a particular distance at a particular time
Sometimes the change in the spin state of the neutron during scattering is
also measured to give information about the locations and orientations of
unpaired electron spins in the sample

What Do We Need to Do a Basic Neutron Scattering Experiment?

A source of neutrons
A method to prescribe the wavevector of the neutrons incident on the sample
(An interesting sample)
A method to determine the wavevector of the scattered neutrons
Not needed for elastic scattering

A neutron detector
Detector
Incident neutrons of wavevector k0

Scattered neutrons of
wavevector, k1
Sample

k0
Q

k1

Remember: wavevector, k, &

wavelength, , are related by:
k = mnv/(h/2) = 2/

Neutron Scattering Requires Intense Sources of Neutrons

Neutrons for scattering experiments can be produced either by nuclear
fission in a reactor or by spallation when high-energy protons strike a heavy
metal target (W, Ta, or U).
In general, reactors produce continuous neutron beams and spallation sources produce
beams that are pulsed between 20 Hz and 60 Hz
The energy spectra of neutrons produced by reactors and spallation sources are different,
with spallation sources producing more high-energy neutrons
Neutron spectra for scattering experiments are tailored by moderators solids or liquids
maintained at a particular temperature although neutrons are not in thermal equilibrium
with moderators at a short-pulse spallation sources

Both reactors and spallation sources are expensive to build and require
sophisticated operation.
SNS at ORNL will cost about $1.5B to construct & ~$140M per year to operate

Either type of source can provide neutrons for 30-50 neutron spectrometers
Small science at large facilities

About 1.5 Useful Neutrons Are Produced by Each Fission Event in

a Nuclear Reactor Whereas About 25 Neutrons Are Produced by
spallation for Each 1-GeV Proton Incident on a Tungsten Target

Artists view of spallation

Spallation
Nuclear Fission

Neutrons From Reactors and Spallation Sources Must Be

Moderated Before Being Used for Scattering Experiments
Reactor spectra are Maxwellian
Intensity and peak-width ~ 1/(E)1/2 at
high neutron energies at spallation
sources
Cold sources are usually liquid
hydrogen (though deuterium is also
used at reactors & methane is sometimes used at spallation sources)
Hot source at ILL (only one in the
world) is graphite, radiation heated.

The ESRF* & ILL* With Grenoble & the Beldonne Mountains

*ESRF = European Synchrotron Radiation Facility; ILL = Institut Laue-Langevin

The National Institute of Standards and Technology (NIST) Reactor Is a

20 MW Research Reactor With a Peak Thermal Flux of 4x1014 N/sec. It Is
Equipped With a Unique Liquid-hydrogen Moderator That Provides Neutrons
for Seven Neutron Guides

Neutron Sources Provide Neutrons for Many

Spectrometers: Schematic Plan of the ILL Facility

A ~ 30 X 20 m 2 Hall at the ILL Houses About 30 Spectrometers.

Neutrons Are Provided Through Guide Tubes

Los Alamos Neutron Science Center

Manuel Lujan Jr.
Neutron Scattering
Center
Proton
Radiography

800-MeV Linear
Accelerator

LANSCE
Visitors Center

Proton Storage
Ring

Weapons
Neutron Research

Neutron Production at LANSCE

Linac produces 20 H- (a proton + 2 electrons) pulses per second
800 MeV, ~800 sec long pulses, average current ~100 A
Each pulse consists of repetitions of 270 nsec on, 90 nsec off
Pulses are injected into a Proton Storage Ring with a period of 360 nsec
Thin carbon foil strips electrons to convert H- to H+ (I.e. a proton)
~3 x 1013 protons/pulse ejected onto neutron production target

Components of a Spallation Source

A half-mile long proton linac

and a neutron production target

.a proton accumulation ring.

A Comparison of Neutron Flux Calculations for the ESS SPSS (50 Hz, 5 MW)
& LPSS (16 Hz, 5W) With Measured Neutron Fluxes at the ILL

16

10

15

17

average flux
poisoned m.
decoupled m.
coupled m.
and long pulse

10

16

10

Flux [n/cm /s/str/]

10

10

peak flux
poisoned m.
decoupled m.
coupled m.
long pulse

17

14

15

10

14

10

Flux [n/cm /s/str/]

10

10
10
10
10

13

12

ILL hot source

ILL thermal source
ILL cold source

11

10

13

10

12

10

ILL hot source

ILL thermal source
ILL cold source

11

10

10

10
2

Wavelength []

Wavelength []

14

Instantaneous flux [n/cm /s/str/]

2.0x10

Pulsed sources only make sense if

one can make effective use of the
flux in each pulse, rather than the
average neutron flux

=4
poisoned moderator
decoupled moderator
coupled moderator
long pulse

14

1.5x10

14

1.0x10

13

5.0x10

0.0
0

500

1000

1500

Time [ s]

2000

2500

3000

Simultaneously Using Neutrons With Many Different Wavelengths

Enhances the Efficiency of Neutron Scattering Experiments

I
ki

res = bw

kf

Neutron
Source Spectrum

bw T
ki

kf

res T

Potential Performance Gain relative to use of a Single Wavelength

is the Number of Different Wavelength Slices used

The Time-of-flight Method Uses Multiple Wavelength Slices

at a Reactor or a Pulsed Spallation Source
T

Detector

Distance

res = 3956 Tp / L
bw = 3956 T / L

T p

Moderator

Time

When the neutron wavelength is determined by time-of-flight, T/Tp different

wavelength slices can be used simultaneously.

The Actual ToF Gain From Source Pulsing

Often Does Not Scale Linearly With Peak Neutron Flux

kf

ki

low rep. rate => large dynamic range

short pulse => good resolution
neither may be necessary or useful

I
ki
kf

large dynamic range may result in intensity

changes across the spectrum
ki

kf

at a traditional short-pulse source the wavelength resolution changes with wavelength

ki

kf

A Comparison of Reactors & Spallation Sources

Short Pulse Spallation Source

Reactor

Energy deposited per useful neutron is

~20 MeV

Energy deposited per useful neutron is

~ 180 MeV

Neutron spectrum is slowing down

spectrum preserves short pulses

Neutron spectrum is Maxwellian

Constant, small / at large neutron

energy => excellent resolution
especially at large Q and E

Resolution can be more easily tailored

to experimental requirements, except
for hot neutrons where monochromator
crystals and choppers are less effective

Copious hot neutrons=> very good for

measurements at large Q and E

Large flux of cold neutrons => very

good for measuring large objects and
slow dynamics

Low background between pulses =>

good signal to noise

Pulse rate for TOF can be optimized

independently for different
spectrometers

Single pulse experiments possible

Neutron polarization easier

Why Isnt There a Universal Neutron Scattering Spectrometer?

Mononchromatic
incident beam

Detector

Elastic Scattering detect all

scattered neutrons

Inelastic Scattering detect

neutrons as a function of
scattered energy (color)

Conservation of momentum => Q = kf ki

Conservation of energy => E = ( h2 m/ 8 2 ) (kf2 - ki2)
Scattering properties of sample depend only on Q and E, not on neutron
Many types of neutron scattering spectrometer are required because
the accessible Q and E depend on neutron energy and because resolution and
detector coverage have to be tailored to the science for such a signal-limited technique.

Brightness & Fluxes for Neutron &

X-ray Sources
Brightness
(s-1 m-2 ster-1)

dE/E
(%)

Divergence
(mrad2)

15

10 x 10

10

16

0.5 x 10

5 x 10

24

0.01

0.1 x 5

5 x 10
10

Neutrons

10

Rotating
Anode

10

Bending
Magnet

10

Wiggler

10

26

0.01

0.1 x 1

Undulator
(APS)

1033

0.01

0.01 x 0.1

Flux
(s-1 m-2)
11

10

17

19

1024

Instrumental Resolution
Uncertainties in the neutron
wavelength & direction of travel
imply that Q and E can only be
defined with a certain precision
When the box-like resolution
volumes in the figure are convolved,
the overall resolution is Gaussian
(central limit theorem) and has an
elliptical shape in (Q,E) space
The total signal in a scattering
experiment is proportional to the
phase space volume within the
elliptical resolution volume the better the resolution, the lower the count
rate

Examples of Specialization of Spectrometers:

Optimizing the Signal for the Science
Small angle scattering [Q = 4 sin/; (Q/Q)2 = (/)2 + (cot )2]
Small diffraction angles to observe large objects => long (20 m) instrument
poor monochromatization (/ ~ 10%) sufficient to match obtainable angular
resolution (1 cm2 pixels on 1 m2 detector at 10 m => ~ 10-3 at ~ 10-2))

Back scattering [ = /2; = 2 d sin ; / = cot +]

very good energy resolution (~neV) => perfect crystal analyzer at ~ /2
poor Q resolution => analyzer crystal is very large (several m2)

Typical Neutron Scattering Instruments

Note: relatively massive shielding; long

flight paths for time-of-flight spectrometers;
many or multi-detectors
but small science at a large facility

Components of Neutron Scattering Instruments

Monochromators
Monochromate or analyze the energy of a neutron beam using Braggs law

Collimators
Define the direction of travel of the neutron

Guides
Allow neutrons to travel large distances without suffering intensity loss

Detectors
Neutron is absorbed by 3He and gas ionization caused by recoiling particles
is detected

Choppers
Define a short pulse or pick out a small band of neutron energies

Spin turn coils

Manipulate the neutron spin using Lamor precession

Shielding
Minimize background and radiation exposure to users

Most Neutron Detectors Use 3He

3He

+ n -> 3H + p + 0.764 MeV

Ionization caused by triton and proton

is collected on an electrode
70% of neutrons are absorbed
when the product of gas
pressure x thickness x neutron
wavelength is 16 atm. cm.
Modern detectors are often position
sensitive charge division is used
to determine where the ionization
cloud reached the cathode.

A selection of neutron detectors

thin-walled stainless steel tubes
filled with high-pressure 3He.

Essential Components of Modern Neutron

Scattering Spectrometers

Pixelated detector covering a wide range

of scattering angles (vertical & horizontal)

Horizontally & vertically focusing

monochromator (about 15 x 15 cm2)

Neutron guide
(glass coated
either with Ni
or supermirror)

Rotating Choppers Are Used to Tailor Neutron

Pulses at Reactors and Spallation Sources
T-zero choppers made of Fe-Co are used at spallation sources
to absorb the prompt high-energy pulse of neutrons
Cd is used in frame overlap choppers to absorb slower
neutrons

Fast neutrons from one pulse can catch-up with

slower neutrons from a succeeding pulse and spoil
the measurement if they are not removed.This is
called frame-overlap

Larmor Precession & Manipulation of the Neutron Spin

In a magnetic field, H, the neutron
spin precesses at a rate

L = H / 2 = 2916.4 H Hz
where is the neutrons gyromagnetic
ratio & H is in Oesteds
This effect can be used to manipulate
the neutron spin
A spin flipper which turns the spin
through 180 degrees is illustrated
The spin is usually referred to as up
when the spin (not the magnetic
moment) is parallel to a (weak ~ 10
50 Oe) magnetic guide field

Next Lecture
3. Diffraction
1. Diffraction by a lattice
2. Single-crystal diffraction and powder diffraction
3. Use of monochromatic beams and time-of-flight to measure powder
diffraction
4. Rietveld refinement of powder patterns
5. Examples of science with powder diffraction

Refinement of structures of new materials

Materials texture
Strain measurements
Structures at high pressure
Microstrain peak broadening
Pair distribution functions (PDF)

by
Roger Pynn
Los Alamos
National Laboratory

LECTURE 3: Diffraction

This Lecture
2. Diffraction
1. Diffraction by a lattice
2. Single-crystal compared with powder diffraction
3. Use of monochromatic beams and time-of-flight to measure powder
diffraction
4. Rietveld refinement of powder patterns
5. Examples of science with powder diffraction

Refinement of structures of new materials

Materials texture
Strain measurements
Pair distribution functions (PDF)

From Lecture 1:
d number of neutrons scattered through angle 2 per second into d
=
d
number of incident neutrons per square cm per second
r r
r r
d
i
(
k

k
'
)
.
(
R

R
)

i
Q
.
(
R
coh coh
coh coh
0
i
j
i R j )
=
b
b
e
=
b
b
e

i
j
i
j
d coh i , j
i, j
r
r r r
where the wavevector transfer Q is defined by Q = k 'k 0
r

k0

Q
k

For elastic scattering k0=k=k:

Q = 2k sin
Q = 4 sin /

Incident neutrons of wavevector k0

2
Sample

Detector

Scattered neutrons of
wavevector, k

Neutron Diffraction
Neutron diffraction is used to measure the differential cross section, d/d
and hence the static structure of materials
Crystalline solids
Liquids and amorphous materials
Large scale structures

Depending on the scattering angle,

structure on different length scales, d,
is measured:

2 / Q = d = / 2 sin( )
For crystalline solids & liquids, use
wide angle diffraction. For large structures,
e.g. polymers, colloids, micelles, etc.
use small-angle neutron scattering

Diffraction by a Lattice of Atoms

r
1
S (Q) =
N

r r v
iQ.( Ri R j )

r r r
r
with Ri = i + ui where i is the equilibrium position

i, j

r
of atom i and ui is any displacement (e.g. thermal) from the equilibrium position.
r r r
Ignoring thermal vibrations, S(Q) is only non - zero for Q's such that Q.(i j ) = 2M .
v
r
r
r
r r r
In a Bravais lattice, we can write i = m1i a1 + m2i a2 + m3i a3 where a1 , a2 , a3 are
the primitive translation vectors of the unit cell.
r * 2 r r
Define a1 =
a2 a3 and cyclic permutations.
V0
r* r
Then ai .a j = 2 ij .
r v
r r r
r* r* r*
If Q = Ghkl = ha1 + ka2 + la3 then Q.(i j ) = 2M .
So scattering from a (frozen) lattice only
occurs when the scattering wavevector, Q,
is equal to a reciprocal lattice vector, G hkl .

a3
a2

a1

For Periodic Arrays of Nuclei, Coherent Scattering Is Reinforced Only in

Specific Directions Corresponding to the Bragg Condition:
= 2 dhkl sin() or 2 k sin() = Ghkl

Bragg Scattering from Crystals

Working through the math (see, for example, Squires' book), we find :
r r
r 2
(2 )3
d
=N
(Q Ghkl ) Fhkl (Q)

V0 hkl
d Bragg
where the unit - cell structure factor is given by
r r
r
iQ . d Wd
Fhkl (Q ) = bd e e
d

and Wd is the Debye - Waller factor tha t accounts for thermal motions of atoms
Using either single crystals or powders, neutron diffraction can be used to
measure F2 (which is proportional to the intensity of a Bragg peak) for
various values of (hkl).
Direct Fourier inversion of diffraction data to yield crystal structures is not
possible because we only measure the magnitude of F, and not its phase =>
models must be fit to the data
Neutron powder diffraction has been particularly successful at determining
structures of new materials, e.g. high Tc materials

We would be better off

if diffraction measured
phase of scattering
rather than amplitude!
Unfortunately, nature
did not oblige us.

Picture by courtesy of D. Sivia

Reciprocal Space An Array of Points (hkl)

that is Precisely Related to the Crystal Lattice

a*
O

Ghkl

b*

Ghkl = 2/dhkl

(hkl)=(260)

a* = 2(b x c)/V0, etc.

A single crystal has to be aligned precisely to record Bragg scattering

Powder A Polycrystalline Mass

All orientations of
crystallites possible
Typical Sample: 1cc powder of
10m crystallites - 109 particles
if 1m crystallites - 1012 particles

Single crystal reciprocal lattice

- smeared into spherical shells

Powder Diffraction gives Scattering on

Debye-Scherrer Cones
(220)

Incident beam
x-rays or neutrons

(200)

Sample
Braggs Law = 2dsin
Powder pattern scan 2 or

(111)

Measuring Neutron Diffraction Patterns with a

Monochromatic Neutron Beam
Use a continuous beam
of mono-energetic
neutrons.

Since we know the neutron wavevector, k,

the scattering angle gives Ghkl directly:
Ghkl = 2 k sin

Neutron Powder Diffraction using Time-of-Flight

Lo= 9-100m
Sample

Pulsed
source

2 - fixed

L1 ~ 1-2m
= 2dsin
Detector
bank

Measure scattering as
a function of time-offlight t = const*

Time-of-Flight Powder Diffraction

Use a pulsed beam with a
broad spectrum of neutron
energies and separate
different energies (velocities)
by time of flight.

Compare X-ray & Neutron Powder Patterns

10.000 0.025 159.00 CPD RRRR PbSO4 Cu Ka X-ray data 22.9.
Scan no. = 1 Lambda1,lambda2 = 1.540 Observed Profile

Counts
.0

.5

1.0

X10E 4

1.5

X-ray Diffraction - CuKa

Phillips PW1710
Higher resolution
Intensity fall-off at small d
spacings
Better at resolving small
lattice distortions

1.0
D-spacing, A

2.0

3.0

4.0

2.5

10.0 0.05 155.9 CPD RRRR PbSO4 1.909A neutron data 8.8
Scan no. = 1 Lambda = 1.9090
Observed Profile

Counts

.5

1.0

1.5

X10E 3
2.0

Neutron Diffraction - D1a, ILL

=1.909
Lower resolution
Much higher intensity at small
d-spacings
Better atomic positions/thermal
parameters

1.0
D-spacing, A

2.0

3.0

4.0

Theres more than meets the eye in a powder pattern*

Rietveld Model Ic = Ib + Yp

Intensity

Io

Ic

Yp

Ib

TOF
*Discussion of Rietveld method adapted from viewgraphs by R. Vondreele (LANSCE)

The Rietveld Model for Refining Powder Patterns

Ic = Io{khF2hmhLhP(h) + Ib}
Io - incident intensity - variable for fixed 2
kh - scale factor for particular phase
F2h - structure factor for particular reflection
mh - reflection multiplicity
Lh - correction factors on intensity - texture, etc.
P( h) - peak shape function includes instrumental resolution,
crystallite size, microstrain, etc.

How good is this function?

Protein Rietveld refinement - Very low angle fit

1.0-4.0 peaks - strong asymmetry
perfect fit to shape

Examples of Science using Powder Diffraction

Refinement of structures of new materials

Materials texture
Strain measurements
Pair distribution functions (PDF)

High-resolution Neutron Powder Diffraction in CMR manganates*

In Sr2LaMn2O 7 synchrotron data indicated

two phases at low temperature. Simultaneous
refinement of neutron powder data at 2 s
allowed two almost-isostructural phases (one
FM the other AFM) to be refined. Only
neutrons see the magnetic reflections

High resolution powder diffraction

with Nd0.7Ca0.3MnO3 showed
splitting of (202) peak due to a
transition to a previously unknown
monoclinic phase. The data showed the existence of 2 Mn sites with
different Mn-O distances. The
different sites are likely occupied
by Mn3+ and Mn4+ respectively

* E. Suard & P. G. Radaelli (ILL)

Texture: Interesting Preferred Orientation

Random powder - all crystallite orientations
equally probable - flat pole figure
Pole figure - stereographic projection of a
crystal axis down some sample direction

Loose powder
(100) random texture

(100) wire texture

Crystallites oriented along wire axis - pole figure

peaked in center and at the rim (100s are 90
apart)
Orientation Distribution Function - probability
function for texture
Metal wire

Texture Measurement by Diffraction

(220)

Non-random crystallite
orientations in sample
Incident beam
x-rays or neutrons

(200)
Sample

Debye-Scherrer cones
uneven intensity due to texture
different pattern of unevenness for different hkls
intensity pattern changes as sample is turned

(111)

Texture Determination of Titanium Wire Plate

(Wright-Patterson AFB/LANSCE Collaboration)

Possible pseudo-single crystal turbine blade material

Ti Wire (100) texture

Ti Wire Plate - does it

still have wire texture?
ND

TD
RD

TD

Bulk measurement
Neutron time-of-flight data
Rietveld refinement of texture

ND

Spherical harmonics to Lmax = 16

Very strong wire texture in plate

reconstructed (100) pole figure

Definitions of Stress and Strain

Macroscopic strain total strain measured by an extensometer
Elastic lattice strain response of lattice planes to applied
stress, measured by diffraction
hkl =

d hkl d 0
d0

Intergranular strain deviation of elastic lattice strain from

linear behavior

I = hkl

applied

Ehkl

Residual strains internal strains present with no applied force

Thermal residual strains strains that develop on cooling from
processing temperature due to anisotropic coefficients of
thermal expansion

Neutron Diffraction Measurements of Lattice Strain*

Text box

Incident neutrons

+90 Detector
measures
to loading direction
-90 Detector
measures parallel
to loading direction

The Neutron Powder

Diffractometer at LANSCE

Neutron measurements :
Non destructive, bulk, phase sensitive

Time consuming, Limited spatial resolution

* Discussion of residual strain adapted from viewgraphs by M. Bourke (LANSCE)

Why use Metal Matrix Composites ?

Higher Pay Loads
High temperatures
High pressures
Reduced Engine Weights
Reduced Fuel Consumption
Better Engine Performance

F117 (JSF)

National AeroSpace Plane

MMC Applications
Landing Gear
Rotors
Fan Blades
Structural Rods
Impellers
Landing Gears

Engine

W-Fe :- Fabrication, Microstructure & Composition

200 m diameter continuous Tungsten fibers,
Hot pressed at 1338 K for 1 hour into Kanthal
( 73 Fe, 21 Cr, 6 Al wt%) leads to residual strains after cooling
Specimens :- 200 * 25 * 2.5 mm3

10 vol.%

20 vol.%

30 vol.%

70 vol.%

Normalized Intensity

Powder diffraction data includes Bragg peaks

from both Kanthal and Tungsten.
K

W
W

d spacing (A)

Neutron Diffraction Measures Mean Residual Phase

Strains when Results for a MMC are compared to an
Undeformed Standard
4000

Kanthal

Residual strain (e)

Residual Strain (e)

3000

10 vol.% W
20 vol.% W
30 vol.% W
70 vol.% W

2000

1000

-1000

Tungsten

-2000

10 vol.% W
20 vol.% W
30 vol.% W
70 vol.% W

-3000

-4000

20

40

60

80

100

Angle to the fiber (deg)

Symbols show data points

Curves are fits of the form <> =< 11>cos2 + < 22>sin2
Pressure Vessels and Piping, Vol. 373, Fatigue, Fracture & Residual Stresses, Book No H01154

Load Sharing in MMCs can also be

Measured by Neutron Diffraction
400

Kanthal yields and ceases to bear

further load
Tungsten fibers reinforce and
strengthen the composite

350

Kanthal
Tungsten

300
Applied stress [MPa]

Initial co-deformation results

from confinement of W fibers
by the Kanthal matrix
Change of slope (125 MPa) is
the typical load sharing
behavior of MMCs:

250
200
150
100
50
0
-1000 -500

500 1000 1500 2000 2500 3000

Elastic lattice strain []

Deformation Mechanism in U/Nb Alloys

Applied Stress (MPa)

Atypical Double Plateau Stress Strain Curve.

Diffraction Results

500
400
300
200
U-6 wt% Nb
100

Typical Metal

0
0

0.02

0.04

0.06

Macroscopic Strain

NPD Diffraction Geometry

Tensile Axis
Q||

+ 90 Detector
Q

-90 Detector

Important Programmatic Answers.

1. Material is Single Phase Monoclinic ().
2. Lack of New Peaks Rules out Stress Induced Phase Transition.
3. Changes in Peak Intensity Indicate That Twinning is the Deformation Mechanism.

Pair Distribution Functions

Modern materials are often disordered.

Standard crystallographic methods lose the aperiodic
(disorder) information.
We would like to be able to sit on an atom and look at
our neighborhood.
The PDF method allows us to do that (see next slide):
First we do a neutron or x-ray diffraction experiment
Then we correct the data for experimental effects
Then we Fourier transform the data to real-space

Obtaining the Pair Distribution Function*

Structure function

2
G(r) = Q[S (Q) 1] sin QrdQ
0

Raw data

PDF
* See http://www.pa.msu.edu/cmp/billinge-group/

Structure and PDF of a High Temperature

Superconductor
The structure of La2-xSrxCuO 4
looks like this: (copper [orange]
sits in the middle of octahedra
of oxygen ions [shown shaded
with pale blue].)

The resulting PDFs look like this.

The peak at 1.9A is the Cu-O
bond.
So what can we learn about
charge-stripes from the PDF?

Effect of Doping on the Octahedra

Doping holes (positive
charges) by adding Sr
shortens Cu-O bonds
Localized holes in stripes
implies a coexistence of
short and long Cu-O inplane bonds => increase in
Cu-O bond distribution
width with doping.
We see this in the PDF: 2
is the width of the CuO
bond distribution which
increases with doping then
decreases beyond optimal
doping

Fermiliquid
like

Polaronic

Bozin et al. Phys. Rev. Lett. Submitted;

cond-mat/9907017

by
Roger Pynn
Los Alamos
National Laboratory

LECTURE 3: Surface Reflection

Surface Reflection Is Very Different From

Most Neutron Scattering
We worked out the neutron cross section by adding scattering
from different nuclei
We ignored double scattering processes because these are usually very weak

This approximation is called the Born Approximation

Below an angle of incidence called the critical angle, neutrons
are perfectly reflected from a smooth surface
This is NOT weak scattering and the Born Approximation is not applicable to
this case

Specular reflection is used:

In neutron guides
In multilayer monochromators and polarizers
To probe surface and interface structure in layered systems

This Lecture
Reflectivity measurements

Neutron wavevector inside a medium

Reflection by a smooth surface
Reflection by a film
The kinematic approximation
Graded interface
Science examples

Polymers & vesicles on a surface

Lipids at the liquid air interface
Boron self-diffusion
Iron on MgO

Rough surfaces
Shear aligned worm-like micelles

What Is the Neutron Wavevector Inside a Medium?

Comparing our expression for S(Q) with that given by Fermi' s Golden Rule, the
r 2h 2
r
nucleus - neutron potential is given by : V (r ) =
b ( r ) for a single nucleus.
m
2h 2
1
So the average potential inside the medium is :
V =
where =
bi

m
volume i
is called the nuclear Scattering Length Density (SLD)
The neutron obeys Schrodinge r' s equation :

in vacuo (r ) = e

r r
iko .r

+ 2m ( E V ) / h 2 (r ) = 0

so k 02 = 2mE / h 2 . Simlarly

k 2 = 2m( E V ) / h 2 = k02 4

where k0 is neutron wavevector in vacuo and k is the wavevecto r in a material

Since k/k 0 = n = refractive index (by definition ), and is very small (~ 10- 6 A - 2 ) we get :
n = 1-2 / 2
Since generally n < 1, neutrons are externally reflected from most materials.

Typical Values
Let us calculate the scattering length density for quartz SiO2
Density is 2.66 gm.cm-3; Molecular weight is 60.08 gm. mole-1
Number of molecules per 3 = N = 10-24(2.66/60.08)*Navagadro
= 0.0267 molecules per 3
=b/volume = N(bSi + 2bO) = 0.0267(4.15 + 11.6) 10-5 -2 =
4.21 x10-6 -2
This means that the refractive index n = 1 2 2.13 x 10-7 for
quartz
To make a neutron bottle out of quartz we require k= 0 I.e.
k02 = 4 or =(/)1/2 .
Plugging in the numbers -- = 864 or a neutron velocity of
4.6 m/s (you could out-run it!)

Only Those Thermal or Cold Neutrons With Very Low

Velocities Perpendicular to a Surface Are Reflected
k / k0 = n
The surface cannot change the neutron ve locity parallel to the surface so :
k0 cos = k cos ' = k0 n cos ' i.e n = cos/cos '
Neutrons obey Snell's Law
Since k 2 = k 02 4
i.e.

k 2 (cos 2 '+ sin 2 ' ) = k02 (cos 2 + sin 2 ) 4

k 2 sin 2 ' = k02 sin 2 4

or k z2 = k02z 4

The critical value of k 0 z for total external reflection is k 0 z = 4

3
-1
For quartz k 0critical
=
2
.
05
x
10
A
zl

(2 / ) sin critical = k critical

0 zl
critical ( o ) 0.02 ( A) for quartz
Note : critical ( ) 0.1 ( A) for nickel
o

Reflection of Neutrons by a Smooth Surface: Fresnels Law

continuity
of & & at z = 0
a I + a R = aT
(1)
r
r
r
a I k I + a R k R = aT kT

n = 1- 2/2

components perpendicular and parallel to the surface :

aI k cos + aR k cos = aT nk cos
(2)
( aI aR )k sin = aT nk sin
(3)
(1) & (2) => Snell' s Law : cos = n cos
( aI aR )
sin sin kTz
(1) & (3) =>
=n

=
(a I + a R )
sin sin k Iz
so reflectanc e is given by r = aR / aI = (k Iz kTz ) /( k Iz + kTz )

What do the Amplitudes aR and aT Look Like?

For reflection from a flat substrate, both aR and aT are complex when k0 < 4
I.e. below the critical edge. For aI = 1, we find:
2

1.5
0.5

1
0.5
0.005

0.01

0.015

0.02

0.025

0.005

-0.5

0.01

0.015

0.02

0.025

-0.5
-1

Real (red) & imaginary (green) parts of aR

plotted against k0. The modulus of aR is
plotted in blue. The critical edge is at
k0 ~ 0.009 A-1 . Note that the reflected wave is
completely out of phase with the incident wave
at k0 = 0

-1

Real (red) and imaginary (green) parts

of aT. The modulus of a T is plotted in
blue. Note that aT tends to unity at
large values of k0 as one would expect

One can also think about Neutron Reflection from a Surface as a

1-d Problem

V(z)

V(z)= 2 (z) h 2/mn

k2=k02 - 4 (z)

substrate

Where V(z) is the potential seen by

the neutron & (z) is the scattering
length density

Film

Vacuum

Fresnels Law for a Thin Film

r=(k1z-k0z)/(k1z+k0z) is Fresnels law

Evaluate with =4.10-6 A-2 gives the
red curve with critical wavevector
given by k0z = (4)1/2

0.01

0.02

k0z
0.03

0.04

0.05

-1
-2
-3
-4

If we add a thin layer on top of the

substrate we get interference fringes &
the reflectance is given by:

r01 + r12 e i 2 k1 z t
r=
1+ r01 r12 e i 2 k1 z t
and we measure the reflectivity R = r.r*

-5

Log(r.r*)

0
1
2

Film thickness = t
substrate

If the film has a higher scattering length density than the substrate we get the
green curve (if the film scattering is weaker than the substance, the green curve is
below the red one)
The fringe spacing at large k0z is ~ /t (a 250 A film was used for the figure)

Kinematic (Born) Approximation

We defined the scattering cross section in terms of an incident plane wave & a
weakly scattered spherical wave (called the Born Approximation)
This picture is not correct for surface reflection, except at large values of Qz
For large Qz, one may use the definition of the scattering cross section to
calculate R for a flat surface (in the Born Approximation) as follows:

R=

number of neutrons reflected by a sample of size L x L y

number of neutrons incident on sample ( = L x L y sin )

1
=
=
Lx Ly sin Lx Ly sin
because k x = k0 cos

d
1
d

=
d
Lx Ly sin

d dk x dk y
d k02 sin

so dk x = k0 sin d .

From the definition of a cross section we get for a smooth substrate :

2
r v r
d
r
r
4

= 2 dr dr ' e iQ.( r r ') = 2 2 Lx Ly (Qx ) (Q y ) so R = 16 2 2 / Qz4

d
Qz

It is easy to show that this is the same as the Fresnel form at large Q z

Reflection by a Graded Interface

Repeating the bottom line of the previous viewgraph but keeping the z - dependence
16 2
of gives : R = 2
Qz

16 2
iQ z z
( z )e dz = Qz4
2

d ( z ) iQ z z
dz e dz

where the second

equality follows after intergrati ng by parts.

If we replace the prefactor by the Fresnel reflectivi ty R F , we get the right answer
for a smooth interface, as well as the correct form at large Q z
d ( z ) iQz z
R = RF
e dz
dz
This can be solved analytically for several convenient forms of d/dz such
2

as 1/cosh 2 (z). This approximate equation illustrates an important point :

reflectivi ty data cannot be inverted uniquely to obtain (z), because
we generally lack important phase informatio n. This means that models
refined to fit refelctivi ty data must have good physical justificat ion.

The Goal of Reflectivity Measurements Is to Infer a

Density Profile Perpendicular to a Flat Interface
In general the results are not unique, but independent
knowledge of the system often makes them very reliable
Frequently, layer models are used to fit the data
Advantages of neutrons include:

Contrast variation (using H and D, for example)

Low absorption probe buried interfaces, solid/liquid interfaces etc
Non-destructive
Sensitive to magnetism
Thickness length scale 10 5000

Direct Inversion of Reflectivity Data is Possible*

Use different fronting or backing materials for two
measurement of the same unknown film
E.g. D2O and H2O backings for an unknown film deposited on a quartz
substrate or Si & Al2O3 as substrates for the same unknown sample
Allows Re(R) to be obtained from two simultaneous equations for
2

R1 and R2

Re(R) can be Fourier inverted to yield a unique SLD profile

Another possibility is to use a magnetic backing and

polarized neutrons
Si or Al2O3 substrate

SiO2
Unknown film
H2O or D2O

* Majkrzak et al Biophys Journal, 79,3330 (2000)

Vesicles composed of DMPC molecules fuse creating almost a perfect

lipid bilayer when deposited on the pure, uncoated quartz block*
(blue curves)
When PEI polymer was added only after quartz was covered by the lipid
bilayer, the PEI appeared to diffuse under the bilayer (red curves)
Scattering Length Density
Profiles

10

-9

0.00

0.02

0.04

0.06

0.08

Qz [-1 ]

* Data courtesy of G. Smith (LANSCE)

0.10

0.12

O
2

Quartz

0
-2 10 -6
-4 10 -6
-6 10 -6

Hydrogenated
Tails

Quartz
Head

Lipid Bilayer on Polymer

on Quartz

Polymer

= 5.9

Head

Lipid Bilayer on Quartz

2 10 -6

Reflectivity, R*Qz

10 -8

D O

-2
Scattering Length Density [ ]

Neutron Reflectivities

= 4.3

-8 10 -6
0.0

20.0

40.0

60.0

Length, z []

80.0

100.0

Polymer-Decorated Lipids at a Liquid-Air Interface*

4.5% PEG lipid
in lipids

SLD Profiles of PEG-Lipids

9.0% PEG lipid

in lipids

8 10

-2

SLD [ ]

1.3% PEG lipid

in lipids

6 10

4 10

2 10

-6

-6

Black
Blue
Green
Red

-6

-6

- pure lipid
- 1.3% PEG
- 4.5% PEG
- 9.0% PEG

0
40

60

80

100

120

140

160

Length []

mushroom-tobrush transition

X-Ray Reflectivities

Neutron Reflectivities

-7

10

neutrons see contrast between

heads (2.6), tails (-0.4),
D2O (6.4) & PEG (0.24)

R*Q

Pure lipid

10

-8

Lipid + 9% PEG
10

-9

0.05

0.1

0.15

0.2

0.25

Q z [ -1]

*Data courtesy of G. Smith (LANSCE)

x-rays see heads (0.65), but all

else has same electron density
within 10% (-0.33)

Pure lipid
10

R/RF

10

Interface broadens as PEG

concentration increases - this is
main effect seen with x-rays

10

-1

10

-2

Lipid + 9% PEG
10

-3

0.1

0.2

0.3

Qz

0.4

0.5

0.6

Non-Fickian Boron Self-Diffusion at an Interface*

Boron Self Diffusion

10

-4

10
Unnanealed sample
2.4 Hours
5.4 Hours
9.7 Hours
22.3 Hours
35.4 Hours

0.1

10

10

10

-5

-6

-7

Reflectivity

0.001
10

-8

-5

10

10

-7

10

-9

10

-10

10

-11

0.045

-9

10

0.02

0.04

0.06

0.08

0.1

0.055

Data requires density

step at interface

-11

10

0.05

0.12

0.06

0.065

0.07

0.075

0.08

Fickian diffusion
doesnt fit the data

-1

Q ( )

Standard Fickian Model

Non-Standard Fickian Model

~650

11

8 10

-6

8 10 -6

-6

7 10

B
-2

~1350

Si

5 10

4 10

3 10

2 10

1 10

-6

-6

-6

-6

-6

-6

0
-800

*Data courtesy of G. Smith (LANSCE)

6 10

Scattering Length Density (

-2

6 10

B
Scattering Length Density (

10

-6

7 10

5 10

4 10

3 10

2 10

1 10

-600

-400

-200

Distance from

0
11

B-

200
10

400

B Interface ()

600

800

-6

-6

-6

-6

-6

-6

0
-600

-400

-200

Distance from

200
11

B-

10

400

B Interface ()

600

800

Polarized Neutron Reflectometry (PNR)

Non-Spin-Flip
++ measures b + Mz
- - measures b - Mz

Spin-Flip
+- measures Mx + i My
-+ measures Mx i My

Structure, Chemistry & Magnetism of Fe(001) on MgO(001)*

X-Ray
0.20(1)nm
0

10

= 3.9(3)

1.9(2)nm

= 4.4(8)

0.44(2)nm

-1

10
X-Ray Reflectivity

-FeOOH

Neutron

-2

10

-3

10

= 5.2(3)

10.7(2)nm

= 4.4(2)

2.1(2)nm

Fe

= 8.6(4)

X-Ray
0.07(1)nm

-4

10

interface

-5

10

= 6.9(4)

0.03(1)nm

-6

10

0.1

0.2
0.3
-1
Q [ ]

0.4

0.5

= 3.1(3)

MgO

= 6.2(5)

Co-Refinement
H
0

10

R ++(obs)
R (fit)

Polarized Neutron Reflectivity

10

++

-1

R --(obs)
R --(fit)
R SF (obs)
R SF (fit)

10 -2

10

Neutron

TEM

-3

-FeOOH (1.8
B)

10 -4

10 -5
0.00

0.02

0.04

0.06
Q [ -1]

0.08

0.10

0.12

Fe(001) (2.2
B)

*Data courtesy of M. Fitzsimmons (LANSCE)

fcc Fe (4
B)

Reflection from Rough Surfaces

k2

k1
z
z=0

kt1

diffuse scattering is caused by surface roughness or inhomogeneities in

the reflecting medium
a smooth surface reflects radiation in a single (specular) direction
a rough surface scatters in various directions
specular scattering is damped by surface roughness treat as graded
interface. For a single surface with r.m.s roughness :

R = RF e

2 k Iz k1t z 2

When Does a Rough Surface Scatter Diffusely?

Rayleigh criterion

path difference:

r = 2 h sin

phase difference:

= (4h/) sin

boundary between rough and smooth: = /2

that is

h < /(8sin)

for a smooth surface

where g = 4 h sin / = Qz h

Qx-Qz
Time-of-Flight,
Transformation

Energy-Dispersive Neutron Reflectometry

Raw data in f -TOF space for a single layer.
Note that large divergence does not imply poor Qz resolution

Raw Data from SPEAR

ki

-ki

kf

TOF - . L / 4

x

Q =kf -ki

Q x =2*(cos

f -cos i )
2
Q z = *(sin f +sin i )

where

For specular reflectivity

Qz=

4 sin( )
=2k

f =i . Then,

Vandium-Carbon Multilayer specular & diffuse scattering

in f -TOF space and transformed to Qx-Qz

The Study of Diffuse Scattering From Rough Surfaces Has

Not Made Much Headway Because Interpretation Is Difficult
The theory (Distorted Wave Born Approximation) used to describe scattering
from a rough surface, works in some cases but breaks down when k z2 / k 2 >> 1,
where is the range of correlations in the surface
In some cases (e.g. faceted surfaces) one would also expect the approximation of a
using an " average surface" wavefunct ion for perturbation theory to break down.

Rmicro rough = Rsmooth e

R facet = Rsmooth e

2 k 20 s 2

2 k 0 k1t s 2

Observation of Hexagonal Packing of Thread-like Micelles

Under Shear: Scattering From Lateral Inhomogeneities
NEUTRON
BEAM

QUARTZ or SILICON

INLET
HOLES
TEST SECTION

TEFLON LIP

OUTLET
TRENCH

RESERVOIRS

TEFLON

Specularly reflected beam

z

46

gle
Sin
artz
u
Q tal
s
Cry

Up to
Microns

Flow
direction
H2O

Thread-like micelle

Scattering pattern
implies hexagonal
symmetry

by
Roger Pynn
Los Alamos
National Laboratory

LECTURE 5: Small Angle Scattering

This Lecture
5. Small Angle Neutron Scattering (SANS)
1.
2.
3.
4.
5.
6.

What is SANS and what does it measure?

Form factors and particle correlations
Guinier approximation and Porods law
Contrast and contrast variation
Deuterium labelling
Examples of science with SANS
1.
2.
3.
4.
5.

Particle correlations in colloidal suspensions

Helium bubble size distribution in steel
Verification of Gaussian statistics for a polymer chain in a melt
Structure of 30S ribosome
The fractal structure of sedimentary rocks

Note: The NIST web site at www.ncnr.nist.gov has several good resources
for SANS calculations of scattering length densities & form factors as well
as tutorials

Small Angle Neutron Scattering (SANS) Is Used to

Measure Large Objects (~1 nm to ~1 m)
Complex fluids, alloys, precipitates,
biological assemblies, glasses,
ceramics, flux lattices, long-wavelength CDWs and SDWs, critical
scattering, porous media, fractal
structures, etc

Scattering at small angles probes

large length scales

Two Views of the Components of a Typical

Reactor-based SANS Diffractometer

The NIST 30m SANS Instrument Under Construction

Where Does SANS Fit As a Structural Probe?

SANS resolves structures
on length scales of 1 1000
nm
Neutrons can be used with
bulk samples (1-2 mm thick)
SANS is sensitive to light
elements such as H, C & N
SANS is sensitive to
isotopes such as H and D

What Is the Largest Object That Can Be Measured by SANS?

Angular divergence of the neutron beam and its lack of monochromaticity contribute to finite
transverse & longitudinal coherence lengths that limit the size of an object that can be seen
by SANS

For waves emerging from slit center,

path difference is hd/L if d<<h<<L;
for waves from one edge of the slit
path difference is (h+a)d/L

r1

d
r2

+a

The variation of path difference, which

causes decreased visibility of Youngs
fringes, is ad/L so the variation in phase
difference is = kad/L
To maintain the visibility of the Youngs
interference pattern we 2 so the
coherence length, d ~ / where is the
divergence angle a/L

-a

The coherence length is the maximum distance between points in a scatterer for which
interference effects will be observable

Instrumental Resolution for SANS

Traditionally, neutron scatterers tend to think in terms of Q and E resolution

4
Q=
sin

Q
Q2

2
cos 2 . 2
=
+
2

sin 2

For SANS, ( / ) rms ~ 5% and is small, so

Q
Q2

2
= 0.0025 +
2

For equal source - sample & sample - detector distances of L and equal
apertures at source and sample of h, rms = 5 / 12 h/L.
The smallest v alue of is determined by the direct beam size : min ~ 1.5h / L
At this value of , angular resolution dominates and
Q rms ~ ( rms / min )Q min ~ rms 4 / ~ ( 2 / )h / L
The largest observable object is ~ 2/Q rms ~ h / L .
This is equal to the transvers e coherence length for the neutron and achieves
a maximum of about 5 m at the ILL 40 m SANS instrument using 15 neutrons.
Note that at the largest va lues of , set by the detector size and distance from the
sample, wavelengt h resolution dominates.

The Scattering Cross Section for SANS

r
r
ds
1
2
recall that
= b NS (Q ) where S (Q ) =
dO
N

r iQr. rr
r 2
dr .e nnuc (r )

r
2
where b is the coherent nuclear scattering length and nnuc(r ) is the
nuclear density

Since the length scale probed at small Q is >> inter-atomic spacing we may use the
scattering length density (SLD), , introduced for surface reflection and note that
nnuc(r)b is the local SLD at position r.

A uniform scattering length density only gives forward scattering (Q=0), thus SANS
measures deviations from average scattering length density.

If is the SLD of particles dispersed in a medium of SLD 0, and np(r) is the

particle number density, we can separate the integral in the definition of S(Q) into
an integral over the positions of the particles and an integral over a single particle.
We also measure the particle SLD relative to that of the surrounding medium I.e:

SANS Measures Particle Shapes and Inter-particle Correlations

d
= b
d
=

space

space

r
r iQr .( rr rr ' )
d r d r ' n N (r ) nN ( r ' ).e
3

space

r
r
d R d R ' n P ( R) n P ( R ' ) e
3

r r r
iQ .( R R ')

3
d
x.e

( 0 )

space

r v
iQ . x

particle

r 2
r iQr .Rr
d
2
3
= ( 0 ) F (Q ) N P d R.GP ( R).e
d
space
where G P is the particle - particle correlatio n function (the probabilit y that the re
r
v 2
is a particle at R if there' s one at the origin) and F(Q) is the particle form factor :
v 2
F ( Q) =

x.e

r v
iQ . x

particle

These expressions are the same as those for nuclear scattering except for the addition
of a form factor that arises because the scattering is no longer from point-like particles

Scattering for Spherical Particles

2

r 2
r iQr .rr
The particle form factor F (Q) = dr e
is determined by the particle shape.
V

For a sphere of radius R, F(Q) only depends on the magnitude of Q :

sin QR QR cos QR 3V0
Fsphere (Q) = 3V0

j1 (QR) V0 at Q = 0

3
(QR)

QR
Thus, as Q 0, the total scattering from an assembly of uncorrelated spherical
r
v
particles[i.e. when G( r ) ( r )] is proportional to the square of the particle volume
times the number of particles.
For elliptical particles
replace R by :
R (a 2 sin 2 + b 2 cos 2 )1/ 2
where is the angle between
r
the major axis (a) and Q

1
0.8

3j1(x)/x 0.6
0.4

0.2

10

Examples of Spherically-Averaged Form Factors

Form Factor for Cylinder with Q at angle to cylinder axis

F (Q, , H , R) =

4V0 sin([QH / 2] cos ) J1 (QR sin )

QH cos (QR sin ) 2
R

Fo rm

Fa c t o r f o r a

F (Q, R, r ) = 3V0

Ve s ic le

R j1 (QR) r j1 (Qr )
Q2 (R3 r3 )
2

t = R - r

Determining Particle Size From Dilute Suspensions

Particle size is usually deduced from dilute suspensions in which inter-particle

correlations are absent
In practice, instrumental resolution (finite beam coherence) will smear out
minima in the form factor
This effect can be accounted for if the spheres are mono-disperse
For poly-disperse particles, maximum entropy techniques have been used
successfully to obtain the distribution of particles sizes

Correlations Can Be Measured in Concentrated Systems

A series of experiments in the late 1980s by Hayter et al and Chen et al

produced accurate measurements of S(Q) for colloidal and micellar systems
To a large extent these data could be fit by S(Q) calculated from the mean
spherical model using a Yukawa potential to yield surface charge and screening
length

Size Distributions Have Been Measured

for Helium Bubbles in Steel
The growth of He bubbles under neutron irradiation is a key factor limiting
the lifetime of steel for fusion reactor walls
Simulate by bombarding steel with alpha particles
TEM is difficult to use because bubble are small
SANS shows that larger bubbles grow as the steel is
annealed, as a result of coalescence of small bubbles
and incorporation of individual He atoms

SANS gives bubble volume (arbitrary units on the plots) as a function of bubble size
at different temperatures. Red shading is 80% confidence interval.

Radius of Gyration Is the Particle Size Usually

Deduced From SANS Measurements
r
If we measure r from the centroid of the particle and expand the exponential
in the definition of the form factor at small Q :
rr 3 1 rr 2 3
r iQr .rr
F (Q ) = dr e V0 + i Q.r d r (Q.r ) d r + ....
2V
V
V

2
cos
sin .d

2
Q 0
= V0 1

sin .d

where rg is the radius of gyration is

Q 2 rg2
V r d r Q 2 rg2

+ ... = V0 1
+ ... V0e 6
3

V d r 6

rg = R 2 d 3r / d 3r. It is usually obtained from a fit

V

to SANS data at low Q (in the so - called Guinier region) or by plotting ln(Intensity) v Q 2 .
The slope of the data at the lowest values of Q is rg2 /3. It is easily verified that the
expression for the form factor of a sphere is a special case of this general result.

Guinier Approximations: Analysis Road Map

Generalized Guinier approximation

1;
=0
P(Q) =
Q ; = 1,2

R 2Q 2
exp
M
0

Derivative-log analysis

d ln P(Q)
Q d P(Q)
=
= 2 R2Q 2
d ln(Q)
3
P(Q) dQ

* Viewgraph courtesy of Rex Hjelm

Guinier approximations provide a

roadmap for analysis.
Information on particle
composition, shape and size.
Generalization allows for
analysis of complex mixtures,
allowing identification of domains
where each approximation
applies.

Contrast & Contrast Matching

Water

DO2O

HO2O

* Chart courtesy of Rex Hjelm

Both tubes contain borosilicate beads +

pyrex fibers + solvent. (A) solvent
refractive index matched to pyrex;. (B)
solvent index different from both beads
and fibers scattering from fibers
dominates

Isotopic Contrast for Neutrons

Hydrogen
Isotope

Scattering Length
b (fm)

Nickel
Isotope

Scattering Lengths
b (fm)

-3.7409 (11)

58

15.0 (5)

6.674 (6)

60

2.8 (1)

4.792 (27)

61

7.60 (6)

62

-8.7 (2)

64

-0.38 (7)

H
D

Ni
Ni
Ni
Ni
Ni

Verification of of the Gaussian Coil Model for a Polymer Melt

One of the earliest important
results obtained by SANS was
the verification of that rg~N-1/2
for polymer chains in a melt
A better experiment was done
3 years later using a small
amount of H-PMMA in D-PMMA
(to avoid the large incoherent
background) covering a MW
range of 4 decades

SANS Has Been Used to Study Bio-machines

Capel and Moore (1988) used the fact that

prokaryotes can grow when H is replaced
by D to produce reconstituted ribosomes
with various pairs of proteins (but not
rRNA) deuterated
They made 105 measurements of interprotein distances involving 93 30S protein
pairs over a 12 year period. They also
measured radii of gyration
Measurement of inter-protein distances
is done by Fourier transforming the form
factor to obtain G(R)
They used these data to solve the
ribosomal structure, resolving ambiguities
by comparison with electron microscopy

Porod Scattering
2

Let us examine the behavior of F(Q) (QR) 4 at large values of Q for a spherical
particle (i.e. Q >> 1/R where R is the sphere radius)
2

2
4
2 sin QR QR. cos QR
4
2 sin QR
F(Q) (QR) = 9V
(QR) = 9V QR cos QR
3
(
QR
)

9V 2 cos2 QR as Q

= 9V 2 / 2 on average (the oscillations will be smeared out by resolution)

9V 2
2A
Thus F(Q)
= 4 where A is the area of the sphere' s surface.
4
2(QR)
Q
2

This is Porod's law and holds as Q for any particle shape provided the particle
surface is smooth.
Another way to obtain it is to expand G(r) = 1 - ar + br 2 + .. [with a = A/(2V)] at small r
and to evaluate the form factor with this (Debye) form for the correlation function.

Scattering From Fractal Systems

Fractals are systems that are self-similar under a change of scale I.e. R -> CR
For a mass fractal the number of particles within a sphere of radius R is
proportional to RD where D is the fractal dimension

Thus
4pR2 dR.G( R) = number of particles between distance R and R + dR = cR D 1dR
G( R) = (c / 4 ) R D 3
r
r i Qr .Rr
2
D 3
and S (Q) = dR.e G ( R) =
dR
.
R
.
sin
QR
.(
c
/
4

)
R
Q
c 1
const
D 2
=
dx
.
x
.
sin
x
=
2 QD
QD
const
For a surface fractal, one can prove that S (Q) 6 Ds which reduces to the Porod
Q
form for smooth surfaces of dimension 2.

Typical Intensity Plot for SANS From Disordered

Systems
Zero Q intercept - gives particle volume if
concentration is known
Guinier region (slope = -rg2/3 gives particle size)
ln(I)
Mass fractal dimension (slope = -D)

Porod region - gives surface area and

surface fractal dimension
{slope = -(6-Ds)}
ln(Q)

Sedimentary Rocks Are One of the Most

Extensive Fractal Systems*

Variation of the average number of SEM

features per unit length with feature size.
Note the breakdown of fractality (Ds=2.8
to 2.9) for lengths larger than 4 microns
*A. P. Radlinski (Austr. Geo. Survey)

SANS & USANS data from sedimentary rock

showing that the pore-rock interface is a surface
fractal (Ds=2.82) over 3 orders of magnitude in
length scale

References
Viewgraphs describing the NIST 30-m SANS instrument
http://www.ncnr.nist.gov/programs/sans/tutorials/30mSANS_desc.pdf

by
Roger Pynn
Los Alamos
National Laboratory

LECTURE 6: Inelastic Scattering

We Have Seen How Neutron Scattering

Can Determine a Variety of Structures
X-Ray
0.20(1)nm
= 3.9(3)

1.9(2)nm

-FeOOH

Neutron
= 4.4(8)

0.44(2)nm

= 5.2(3)

10.7(2)nm

Fe

= 4.4(2)

2.1(2)nm

interface

= 8.6(4)

0.07(1)nm
= 6.9(4)

0.03(1)nm

= 3.1(3)

crystals

MgO

surfaces & interfaces

= 6.2(5)

disordered/fractals

biomachines

but what happens when the atoms are moving?

Can we determine the directions and
time-dependence of atomic motions?
Can well tell whether motions are periodic?
Etc.
These are the types of questions answered
by inelastic neutron scattering

The Neutron Changes Both Energy & Momentum

When Inelastically Scattered by Moving Nuclei

The Elastic & Inelastic Scattering Cross

Sections Have an Intuitive Similarity
The intensity of elastic, coherent neutron scattering is proportional to the
spatial Fourier Transform of the Pair Correlation Function, G(r) I.e. the
probability of finding a particle at position r if there is simultaneously a
particle at r=0
The intensity of inelastic coherent neutron scattering is proportional to
the space and time Fourier Transforms of the time-dependent pair
correlation function function, G(r,t) = probability of finding a particle at
position r at time t when there is a particle at r=0 and t=0.
For inelastic incoherent scattering, the intensity is proportional to the
space and time Fourier Transforms of the self-correlation function, Gs(r,t)
I.e. the probability of finding a particle at position r at time t when the
same particle was at r=0 at t=0

The Inelastic Scattering Cross Section

r
r
d 2
d 2
2 k'
2 k'

Recall that
= bcoh k NS (Q, ) and d.dE = binc k NS i (Q, )
d

.
dE

coh

inc
r r
r r
r
r
1
r
1
r
i ( Q. r t ) r
i (Q . r t ) r
where S (Q, ) =
G (r , t )e
dr dt and S i (Q, ) =
Gs (r , t )e
dr dt

2h
2h

and the correlation functions that are intuitively similar to those for the elastic scattering case :
r
1
r
r r
r
r
1
r r
r r r
r
G( r , t ) = N (r ,0) N ( r + R, t ) dr and Gs (r , t ) = ( r R j (0)) ( r + R R j (t )) dr
N
N j
The evaluation of the correlatio n functions (in which the ' s and - functions have to be treated
as non - commuting quantum mechanical operators) is mathematic ally tedious. Details can be
found, for example, in the books by Squires or Marshal and Lovesey.

Examples of S(Q,) and Ss(Q,)

Expressions for S(Q,) and Ss(Q,) can be worked out for a
number of cases e.g:
Excitation or absorption of one quantum of lattice vibrational
energy (phonon)
Various models for atomic motions in liquids and glasses
Various models of atomic & molecular translational & rotational
diffusion
Rotational tunneling of molecules
Single particle motions at high momentum transfers
Transitions between crystal field levels
Magnons and other magnetic excitations such as spinons
Inelastic neutron scattering reveals details of the shapes of
interaction potentials in materials

A Phonon is a Quantized Lattice Vibration

Consider linear chain of particles of mass M coupled by
springs. Force on nth particle is
Fn = 0un + 1 (u n1 + u n +1 ) + 2 (u n 2 + u n +2 ) + ...
First neighbor force constant

displacements

Equation of motion is Fn = Mu&&n

4
i ( qnat )
2
2 1
u
(
t
)
=
A
e
with

sin
( qa )

Solution is: n
q
q

M
2
q = 0,

2 4
N 2
,
,......
L
L
2 L

1.4
1.2
1

0.8

0.6
0.4

Phonon Dispersion Relation:

Measurable by inelastic neutron scattering

0.2
-1

-0.5

0.5

qa/2

Inelastic Magnetic Scattering of Neutrons

In the simplest case, atomic spins in a ferromagnet precess
about the direction of mean magnetization
r r r r
H = J (l l ' ) Sl .Sl ' = H 0 + h q bq+ bq
l ,l '

exchange coupling

with
h q = 2S ( J 0 J q )

ground state energy

r iqr.lr
where J q = J (l )e
l

h q = Dq 2 is the dispersion relation for a ferromagnet

Spin wave animation courtesy of A. Zheludev (ORNL)

spin waves (magnons)

Fluctuating spin is
perpendicular to mean spin
direction => spin-flip
neutron scattering

Measured Inelastic Neutron Scattering Signals in Crystalline

Solids Show Both Collective & Local Fluctuations*

Spin waves collective

excitations

Crystal Field splittings

(HoPd2Sn) local excitations

Local spin resonances (e.g. ZnCr2 O4)

* Courtesy of Dan Neumann, NIST

Measured Inelastic Neutron Scattering Signals in Liquids

Generally Show Diffusive Behavior

Simple liquids (e.g. water)

Complex Fluids (e.g. SDS)

Quantum Fluids (e.g. He in porous silica)

Measured Inelastic Neutron Scattering in Molecular

Systems Span Large Ranges of Energy

Vibrational spectroscopy
(e.g. C60)

Polymers

Molecular reorientation
(e.g. pyrazine)

Rotational tunneling
(e.g. CH3I)

Proteins

Atomic Motions for Longitudinal & Transverse Phonons

r 2
Q=
(0.1,0,0)
a
a

Q
Transverse phonon
r
eT = (0,0.1,0) a

Q
Longitudinal phonon
r
eL = (0.1,0,0) a

r
r
r i (Qr . Rrl t )
Rl = Rl 0 + es e

Transverse Optic and Acoustic Phonons

Acoustic
r
ered = (0,0.1,0) a
r
eblue = ( 0,0.14,0) a

Optic
r
ered = (0,0.1,0)a
r
eblue = (0,0.14,0) a

r
r 0 r i (Qr .Rr t )
Rlk = Rlk + es e l

Phonons the Classical Use for Inelastic Neutron Scattering

Coherent scattering measures scattering from single phonons
d 2

d dE

rr 2
2
r r r

k'
(Q .e s )
1 1
2W

= coh
e
(ns +
) ( m s ) ( Q q G )

k MV 0
s
2
s
G
coh 1

Note the following features:

Energy & momentum delta functions => see single phonons (labeled s)
Different thermal factors for phonon creation (ns+1) & annihilation (ns)
Can see phonons in different Brillouin zones (different recip. lattice vectors, G)
Cross section depends on relative orientation of Q & atomic motions (es)
Cross section depends on phonon frequency (s) and atomic mass (M)
In general, scattering by multiple
excitations is either insignificant
or a small correction (the presence of
other phonons appears in the DebyeWaller factor, W)

The Workhorse of Inelastic Scattering Instrumentation at

Reactors Is the Three-axis Spectrometer

Q
kF

kI

scattering triangle

The Accessible Energy and Wavevector

Transfers Are Limited by Conservation Laws
Neutron cannot lose more than its initial kinetic energy &
momentum must be conserved

Triple Axis Spectrometers Have Mapped Phonons

Dispersion Relations in Many Materials
Point by point measurement in (Q,E)
space
Usually keep either kI or kF fixed
Choose Brillouin zone (I.e. G) to maximize
scattering cross section for phonons
Scan usually either at constant-Q
(Brockhouse invention) or constant-E

Phonon dispersion of 36Ar

What Use Have Phonon Measurements Been?

Quantifying interatomic potentials in metals, rare gas solids, ionic
crystals, covalently bonded materials etc
Quantifying anharmonicity (I.e. phonon-phonon interactions)
Measuring soft modes at 2nd order structural phase transitions
Electron-phonon interactions including Kohn anomalies
Roton dispersion in liquid He
Relating phonons to other properties such as superconductivity,
anomalous lattice expansion etc

Examples of Phonon Measurements

Soft mode

Phonons in 36Ar validation

of LJ potential

Roton dispersion in 4He

Kohn anomalies
in 110Cd
Phonons in 110Cd

Time-of-flight Methods Can Give Complete Dispersion Curves at a

Single Instrument Setting in Favorable Circumstances

CuGeO3 is a 1-d magnet. With the unique axis parallel to the incident
neutron beam, the complete magnon dispersion can be obtained

Much of the Scientific Impact of Neutron Scattering Has Involved

the Measurement of Inelastic Scattering
Ne utrons in Condens ed Matter Res e arch
10000

100

SPSS - Chopper
Spectrometer
Spallation
- Chopper
ILL - without s pin-echo
ILL - with s pin-e cho

Ene rgy Trans fe r (me V)

Elastic Scattering

1
[Larger Objects Resolved]
Micelles Polymers
Prot eins in Solution
Viruses

Metallurgical
Syst ems
Colloids

Neutron
Induced
Excitations

Itinerant
Hydrogen
Mag nets Mo des
Crystal
Momentum
Fields
Distribut ions
Molecular
Spin
Coherent
Vibrations
Waves
Modes in
Lattice
Glasses
Vibrations
and Liquids
and
ElectronAnharmonicity
phonon
Molecular
Int eractio ns
Motions
Accurate Liquid Structures
Crystal and
Membranes
Magnetic Amorphous
Precision Crystallography
Proteins
Systems
Structures
Anharmonicity
Crit ical
Scattering

0.01
Aggregate
Motion
Polymers
and
Biological
Systems

10-4

Slower
Motions
Resolved

Dif fusive
Modes
Molecular
Reo rientation
Tunneling
Spectroscopy
Surface
Effects?

10-6
0.001

0.01

0.1

-1
Q ( )

10

100

Energy & Wavevector Transfers accessible to Neutron Scattering

Neutron Spin Echo

By
Roger Pynn*
Los Alamos National Laboratory

*With contributions from B. Farago (ILL), S. Longeville (Saclay),

and T. Keller (Stuttgart)

What Do We Need for a Basic Neutron Scattering Experiment?

A source of neutrons
A method to prescribe the wavevector of the neutrons incident on the sample
(An interesting sample)
A method to determine the wavevector of the scattered neutrons
A neutron detector
Detector
Incident neutrons of wavevector ki
Sample

ki
Q

Scattered neutrons of
wavevector, kf

kf

h2 2
E = h =
( ki k 2f )
2m

We usually measure scattering

as a function of energy (E) and
wavevector (Q) transfer

Instrumental Resolution

Uncertainties in the neutron

wavelength & direction of travel
imply that Q and E can only be
defined with a certain precision

When the box-like resolution

volumes in the figure are convolved,
the overall resolution is Gaussian
(central limit theorem) and has an
elliptical shape in (Q,E) space

The total signal in a scattering

experiment is proportional to the
phase space volume within the
elliptical resolution volume the better the resolution, the smaller the
resolution volume and the lower the count rate

The Goal of Neutron Spin Echo is to Break the Inverse

Relationship between Intensity & Resolution

Traditional define both incident & scattered wavevectors in order to

define E and Q accurately

Traditional use collimators, monochromators, choppers etc to define

both ki and kf

NSE measure as a function of the difference between appropriate

components of ki and kf (original use: measure ki kf i.e. energy change)

NSE use the neutrons spin polarization to encode the difference

between components of ki and kf

NSE can use large beam divergence &/or poor monochromatization to

increase signal intensity, while maintaining very good resolution

The Underlying Physics of Neutron Spin Echo (NSE)

Technology is Larmor Precession of the Neutrons Spin
The time evolution of the expectation value of the spin of a
spin-1/2 particle in a magnetic field can be
B
determined classically as:
r
r r
ds
s
= s B
L = B
dt
= 2913 * 2 Gauss 1 .s 1
The total precession angle of the spin, , depends on the time
the neutron spends in the field: = L t
B(Gauss)

L (103 rad.s-1)

N (msec-1)

Turns/m for
4 neutrons

10

183

29

~29

Larmor Precession allows the Neutron Spin to be Manipulated

using or /2 Spin-Turn Coils: Both are Needed for NSE
The total precession angle of the spin, , depends on the time
the neutron spends in the B field
= L t = Bd / v

d
Neutron velocity, v
B
1
Number of turns =
.B[Gauss ].d [cm].[ Angstroms ]
135.65

How does a Neutron Spin Behave when the

Magnetic Field Changes Direction?
When H rotates with
frequency , H0->H1
->H2, and the spin
trajectory is described
by a cone rolling on the
plane in which H moves
Distinguish two cases: adiabatic and sudden
Adiabatic tan() << 1 large B or small spin and field
remain co-linear this limit used to guide a neutron spin
Sudden tan() >> 1 large spin precesses around new
field direction this limit is used to design spin-turn devices

Neutron Spin Echo (NSE) uses Larmor Precession to

Code Neutron Velocities
A neutron spin precesses at the Larmor frequency in a
L = B
magnetic field, B.
The total precession angle of the spin, , depends on the time
the neutron spends in the field
r r
d
H ,

= Lt = Bd / v
B Neutron velocity, v
1
Number of turns =
.B[Gauss ].d [cm].[ Angstroms ]
135.65

The precession angle is a measure of the neutrons speed v

The Principles of NSE are Very Simple

If a spin rotates anticlockwise & then clockwise by the same
amount it comes back to the same orientation
Need to reverse the direction of the applied field
Independent of neutron speed provided the speed is constant

The same effect can be obtained by reversing the precession

angle at the mid-point and continuing the precession in the
same sense
Use a rotation

If the neutrons velocity, v, is changed by the sample, its spin

will not come back to the same orientation
The difference will be a measure of the change in the neutrons speed or
energy

In NSE*, Neutron Spins Precess Before and After Scattering & a

Polarization Echo is Obtained if Scattering is Elastic
y
x
/2
z
/2
S
F

Initially,
neutrons
are polarized
along z

Allow spins to
precess around z:
slower neutrons
precess further over
a fixed path-length

Rotate spins into

x-y precession plane

Elastic
Scattering
Event

Rotate spins
through about
x axis

Final Polarization, P = cos(1 2 )

Rotate spins
to z and
measure
polarization

Allow spins to precess

around z: all spins are in
the same direction at the
echo point if E = 0

* F. Mezei, Z. Physik, 255 (1972) 145

For Quasi-elastic Scattering, the Echo

Polarization depends on Energy Transfer
If the neutron changes energy when it scatters, the precession
phases before & after scattering, 1 & 2, will be different:
1
m(v12 v22 ) mvv
2
1 1 Bd
Bdh Bdm 2 3
=
1 2 = Bd 2 v
3
mv
2h 2
v1 v2 v
using

h =

To lowest order, the difference between 1 & 2 depends only

on (I.e. v1 v2) & not on v1 & v2 separately
The measured polarization, <P>, is the average of cos(1 - 2)
over all transmitted neutrons I.e.
P =

r
I ( ) S (Q, ) cos(1 2 )dd
r
I ( ) S (Q, )dd

Neutron Polarization at the Echo Point is a

Measure of the Intermediate Scattering Function
P

r
I ( ) S (Q, ) cos(1 2 )dd

r
I ( ) S (Q, )dd

where the " spin echo time" = Bd

2h 2

r
r
S (Q, ) cos( )d = I(Q, )

Bd
(T.m)

(nm)

(ns)

0.4

12

0.6

40

1.0

186

I(Q,t) is called the intermediate scattering function

Time Fourier transform of S(Q,) or the Q Fourier transform of G(r,t), the two
particle correlation function

NSE probes the sample dynamics as a function of time rather than

as a function of

The spin echo time, , is the correlation time

Neutron Polarization is Measured using an

Asymmetric Scan around the Echo Point
Neutron counts
counts (( per
per 50
50 sec)
sec)
Neutron

Echo Point
2000
2000

1500
1500

1000
1000

500
500

00

00

11

22

Bd2

33
44
55
66
Current
in
coil
3
(Arbitrary Units)

77

The echo amplitude decreases

when (Bd)1 differs from (Bd)2
because the incident neutron
beam is not monochromatic.
For elastic scattering:

m
P ~ I ( ) cos {( Bd )1 ( Bd ) 2 } .d

Because the echo point is the same for all neutron wavelengths,
we can use a broad wavelength band and enhance the signal intensity

What does a NSE Spectrometer Look Like?

IN11 at ILL was the First

Field-Integral Inhomogeneities cause to vary

over the Neutron Beam: They can be Corrected

Solenoids (used as main precession fields) have fields that vary as r2

away from the axis of symmetry because of end effects (div B = 0)

B
solenoid

According to Amperes law, a current

distribution that varies as r 2 can
correct the field-integral inhomogeneities for parallel paths

Similar devices can be used

to correct the integral along
divergent paths

Fresnel correction coil for IN15

Ne utro ns in Co nde ns e d Matte r Re s e arc h

10000

100

- Chopper
SSPSS
p a lla tio
n - Ch o pSpectrometer
p er
ILL - w ith o u t s p in -ec h o
ILL - w ith s p in -e ch o

Ene rgy Trans fe r (me V)

Elastic Scattering

1
[Larger Objects Resolved]
Micelles Polym ers
Proteins in Solution
Viruses

M etallurgical
Systems
Colloids

N eutron
Induced
Excitations

Itinerant
Hyd rog en
Mag nets
Mo des
Crystal
Momentum
Fields
Distributions
Mo lecu lar
Spin
Coherent
Vib rations
Waves
M odes in
Lattice
Glasses
Vibrations
and Liquids
and
ElectronA nharmonicity
phonon
Molecular
Interactio ns
M otions
Accurate Liquid Structures
C rystal and
M em branes
Amorphous
P recision Crystallography
Magnetic
Proteins
Systems
Anharmonicity
Structures
C ritical
Scattering

0.01
A ggregate
Motion
Polymers
and
Biological
Systems

10-4

Slower
M otions
R esolved

Diffusive
Modes
Molecular
Reo rientation
Tunneling
Spectroscopy
Surface
Effects?

10 -6
0 .0 01

0 .0 1

0 .1

-1
Q ( )

10

10 0

Neutron Spin Echo has significantly extended the (Q,E) range to which
neutron scattering can be applied

Neutron Spin Echo study of Deformations

of Spherical Droplets*

* Courtesy of B. Farago

The Principle of Neutron Resonant Spin Echo

Within a coil, the neutron is subjected to a

steady, strong field, B0, and a weak rf field
B1cos(t) with a frequency = 0 = B0

B0
B1(t)

Typically, B0 ~ 100 G and B1 ~ 1 G

In a frame rotating with frequency 0, the neutron spin sees a constant field
of magnitude B1

The length of the coil region is chosen so that the neutron spin precesses
around B1 thru an angle .
exit

The neutron precession phase is:

B1exit

entry
exit
exit
entry
neutron
= RF
+ ( RF
neutron
)

entry
2 RF

entry
neutron
+ 0d / v

S rotated

B1entry

S entry

NRSE spectrometer
Neutron Spin Phases
in an NRSE Spectrometer*
A
B
C
D

B=

Sample

B=0

B0

B=0

B0
B0

B0

B=0
+ +
lAB

tA

tA

+ +
d
tB
t

+
d
tC tC

+
d
tD
t

lCD

   














 


    
    

    


    
     
    










   



   

   

Echo occurs for elastic

scattering when
lAB + d = lCD + d

 


 


    

    

 
  




* Courtesy of S. Longeville

H=0

Just as for traditional NSE, if the

scattering is elastic, all neutron spins
arrive at the analyzer with unchanged
polarization, regardless of neutron
velocity.
If the neutron velocity changes, the
neutron beam is depolarized

The Measured Polarization for NRSE is given by

an Expression Similar to that for Classical NSE

Assume that v = v + v with v small and expand to lowest order, giving:

P =

r
I ( ) S (Q, ) cos( NRSE )dd
r
I ( ) S (Q, )dd

where the " spin echo time" NRSE = 2B0 (l + d )

m2
2h 2

Note the additional factor of 2 in the echo time compared with classical NSE
(a factor of 4 is obtained with bootstrap rf coils)

The echo is obtained by varying the distance, l, between rf coils

In NRSE, we measure neutron velocity using fixed clocks (the rf coils)

whereas in NSE each neutron carries its own clock whose (Larmor) rate is
set by the local magnetic field

An NRSE Triple Axis Spectrometer at HMI:

Note the Tilted Coils

Measuring Line Shapes for Inelastic Scattering

Spin echo polarization is the FT of

scattering within the spectrometer
transmission function

An echo is obtained when

d
dk1

( Bd )1 ( Bd ) 2
N1 N 2

=
0

= 2

2
k2
k1
k2
k1

Normally the lines of constant spinecho phase have no gradient in Q,

space because the phase depends
only on k
The phase lines can be tilted by using
tilted precession magnets

By Tilting the Precession-Field Region, Spin Precession Can Be

Used to Code a Specific Component of the Neutron Wavevector
If a neutron passes through a
rectangular field region at an
angle, its total precession phase
will depend only on k.

k//

k
k

L = B
d
KBd
= L t = B
=
v sin
k
with K = 0.291

(Gauss.cm.)-1

Stop precession here

Start precession here

Phonon Focusing
For a single incident neutron wavevector, kI, neutrons are
scattered to kF by a phonon of frequency o and to kf by
neighboring phonons lying on the scattering surface.
The topology of the scattering surface is related to that of the phonon dispersion
surface and it is locally flat

Provided the edges of the NSE precession field region are

parallel to the scattering surface, all neutrons with scattering
wavevectors on the scattering surface will have equal spin-echo
phase
Q0
scattering surface
kI

kF
kf

kf

Precession field region

Tilted Fields
Phonon focusing using tilted fields is available at ILL and in
Japan (JAERI).however,
The technique is more easily implemented using the NRSE
method and is installed as an option on a 3-axis spectrometers
at HMI and at Munich
Tilted fields can also be used can also be used for elastic
scattering and may be used in future to:

Increase the length scale accessible to SANS

Separate diffuse scattering from specular scattering in reflectometry
Measure in-plane order in thin films
Improve Q resolution for diffraction

An NRSE Triple Axis Spectrometer at HMI:

Note the Tilted Coils

Nanoscience & Biology Need Structural

Probes for 1-100 nm

CdSe nanoparticles
Peptide-amphiphile nanofiber
10 nm holes in PMMA

1
Actin

Si colloidal crystal

10

Structures over many length

scales in self-assembly of
ZnS and cloned viruses

Thin copolymer films

Tilted Fields for Diffraction: SANS

d

/2

/2

Any unscattered neutron (=0) experiences the same precession angles

(1 and 2) before and after scattering, whatever its angle of incidence
Precession angles are different for scattered neutrons
KBd cos
KBd
KBd

1 =
and 2 =
cos(1 2 ) cos
2
k sin
k sin( + )
k sin
KBd cos
P = dQ.S (Q). cos 2 2 Q
k sin

Spin Echo Length,

Polarization proportional to
Fourier Transform of S(Q)
r = KBd cos /(k sin ) 2

How Large is the Spin Echo Length for SANS?

Bd/sin
(Gauss.cm)

(Angstroms)

(degrees)

r
(Angstroms)

3,000

20

1,000

5,000

20

1,500

5,000

20

3,500

5,000

10

7,500

It is relatively straightforward to probe length scales of ~ 1 micron

Spin Echo SANS using Magnetic Films:

200 nm Correlation Distance Measured

Close up of SESAME (Spin Echo Scattering

Angle Measurement) Apparatus

Conclusion:
NSE Provides a Way to Separate Resolution from
Monochromatization & Beam Divergence
The method currently provides the best energy resolution for
inelastic neutron scattering (~ neV)
Both classical NSE and NRSE achieve similar energy resolution
NRSE is more easily adapted to phonon focusing I.e. measuring the energy
line-widths of phonon excitations

The method is likely to be used in future to improve (Q)

resolution for elastic scattering

Extend size range for SANS (SESANS)

May allow 100 1000 x gain in measurement speed for some SANS exps
Separate specular and diffuse scattering in reflectometry
Measure in-plane ordering in thin films (SERGIS)