RIETVELD REFINEMENT

Dr. Maykel Manawan

SIMULATING REAL DIFFRACTION DATA
𝑝𝑝
𝑘𝑘2
𝑝𝑝
𝑌𝑌𝑖𝑖𝑖𝑖 = 𝑌𝑌𝑖𝑖𝑖𝑖 + � � 𝐺𝐺𝑖𝑖𝑖𝑖 𝐼𝐼𝑘𝑘
𝑝𝑝 𝑘𝑘=𝑘𝑘 𝑝𝑝
1

 Yic is each individual data point i

 Yib is the intensity of the background at point i in the pattern

 k1 … k2 are the reflections contributing to data point i in the

pattern
 sometimes multiple Bragg diffraction peaks overlap, resulting in
multiple contributions to the observed intensity at a single data point
 Gik is the peak profile function
 this describes how the intensity of the diffraction peak is distributed
over a range of 2theta rather than at a single point
 IK is the intensity of the Bragg diffraction peak
 DIFFRACTION INTENSITY
𝐼𝐼𝑘𝑘 = 𝑆𝑆𝑀𝑀𝑘𝑘 𝐿𝐿𝑘𝑘 𝐹𝐹𝑘𝑘 2 𝑃𝑃𝑘𝑘 𝐴𝐴𝑘𝑘 𝐸𝐸𝑘𝑘

 S is an arbitrary scale factor

 used to adjust the relative contribution of individual phases to the overall
diffraction pattern
 M is the multiplicity of the reflection
 accounts for the fact that some observed diffraction peaks are actually the
product of multiple equivalent planes diffracting at the same 2theta;
example: (001) (100) (010) etc in cubic
 automatically calculated based on the crystal structure
 L is the Lorentz polarization factor
 F is structure factor, which is the amplitude of scattered light
due to the crystal structure
 P is the modification of intensity due to preferred orientation
 A is absorption correction
 E is extinction correction
STRUCTURE FACTOR
[ ]
m
Fhkl = ∑ N j f j exp 2πi (hx j + ky j + lz j )
j =1

 Fhkl sums the result of scattering from all of the atoms in

the unit cell to form a diffraction peak from the (hkl)
planes of atoms
 The amplitude of scattered light is determined by:
 where the atoms are on the atomic planes
expressed by the fractional coordinates xj yj zj
 what atoms are on the atomic planes
 the scattering factor fj quantifies the relative efficiency of
scattering at any angle by the group of electrons in each atom
 Nj is the fraction of every equivalent position that is occupied
by atom j
ATOMIC SCATTERING FACTOR
2
  − B sin θ 
2

f
2

=  f 0 exp   + ∆f ' 
 + (∆f ' ' ) 2

  λ 2
 

 X-rays are electromagnetic radiation that interact with an

electron, which reradiates the X-ray as a spherical wave
 The number of electrons around an atom defines how
strongly it will scatter the incident X-ray beam
 the initial strength of the atomic scattering factor is equal to the
number of electrons around the atom
 The scattering factor, f, is a product of several terms
describing the interaction of the X-ray with the electrons
around an atom
 ATOMIC SCATTERING FACTOR
 f0 at 0° is equal to the number of 40
electrons around the atom Y
Y(3+)
35
 Y and Zr are similar, but slightly Zr

different, at 0° 30 Zr(4+)
O(2-)
 Zr and Zr4+ are slightly different at 0° 25

 Y3+ and Zr4+ are identical at 0° 20

fo
 the variation with (sin θ)/λ 15
depends on size of atom
10
 smaller atoms drop off quicker
5
 at higher angles, the difference
between Y3+ and Zr4+ is more readily 0
discerned 0 0.5 1 1.5
(sin θ)/ λ
 at higher angles, the difference
between different oxidation states
(eg Zr and Zr4+) is less prominent
 TEMPERATURE FACTOR
the efficiency of scattering by an  B sin 2 θ 
 f = f 0 exp − 
atom is reduced because the atom  λ2

and its electrons are not stationary
40
 atom is vibrating about its equilibrium B=0
lattice site 35 B=1
B=10
 amount of vibration is quantified by the 30
Debye-Waller temperature factor:
25
B=8π2U2
20
 U2 is the mean-square amplitude of

f
the vibration 15

 this is for isotropic vibration: 10

sometimes B is broken down into six Bij
5
anistropic terms if the amplitude of
vibration is not the same in all 0
0 0.5 1 1.5 2
directions.
(sin θ)/ λ
DIFFRACTION PEAKS FROM LABORATORY DIFFRACTOMETERS ARE A
MIXTURE OF GAUSSIAN AND LORENTZIAN PROFILES

Counts
Counts

4000

Gaussian profile shape

2000 Lorentzian profile shape
3000

2000

1000

1000

0
0
42 44 46 48 50
45.50 46
Position [°2Theta]
Position [°2Theta]

8
 PEAK PROFILE FUNCTION (PSEUDO-VOIGT)
𝐹𝐹 Δ𝑇𝑇 = 𝜂𝜂𝜂𝜂 Δ𝑇𝑇, Γ + 1 − 𝜂𝜂 𝐺𝐺(Δ𝑇𝑇, Γ)

2 1
4𝑙𝑙𝑙𝑙𝑙 −4𝑙𝑙𝑙𝑙𝑙(Δ𝑇𝑇)2 𝐿𝐿 Δ𝑇𝑇, 𝛾𝛾𝐿𝐿 = 2
𝐺𝐺(Δ𝑇𝑇, Γ𝐺𝐺 ) = 2 𝑒𝑒𝑒𝑒𝑒𝑒 𝜋𝜋𝛾𝛾𝐿𝐿 2Δ𝑇𝑇
𝜋𝜋Γ𝐺𝐺 Γ𝐺𝐺2 1+
𝛾𝛾𝐿𝐿
180 𝑘𝑘𝜆𝜆 1 180
Γ𝐺𝐺2 = 𝑈𝑈 𝑡𝑡𝑡𝑡𝑡𝑡2 𝜃𝜃 + 𝑉𝑉 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝑊𝑊 + 𝑃𝑃/𝐶𝐶𝐶𝐶𝐶𝐶 2 𝜃𝜃 𝛾𝛾𝐿𝐿 = + 𝑆𝑆 𝑡𝑡𝑡𝑡𝑡𝑡𝜃𝜃
𝜋𝜋 𝐷𝐷 𝑐𝑐𝑐𝑐𝑐𝑐𝜃𝜃 𝜋𝜋

 Fractions of each function depend on the relative widths of each [see

mixing factor (η) in GSAS manual, η=0 is Gaussian, η=1 is Lorentzian]
 U, V & W are Cagliotti terms, derived as instrument response function for
CW neutron diffraction. Incomplete for X-rays.
 P is a crystallite broadening contribution
 K is Scherrer constant, D is size broadening and S is strain broadening
THE SYSTEMATIC VARIATION OF PEAK WIDTH WITH
2THETA

10
GAUSSIAN AND LORENTZIAN CONTRIBUTIONS
A functions to describe the Gaussian – Lorentzian contributions :
 Gaussian
 Cauchy (Lorentzian)
 Modified Lorentzian (1 &2)
 Voigt
 Pseudo Voigt
 Pseudo Voigt with FJC Asymmetry correction
 Split Pseudo Voigt
 Pearson VII
SOURCE OF PROFILE BROADENING
Peak Broadening B ( FWHM ) = Bi + Bc + Bs + BSF + ...

Instrumental • Unresolved α1 , α2 peaks

• Non-monochromaticity of the source (finite width of α peak)
Bi • Imperfect focusing

Crystallite size In the vicinity of θB the −ve of Bragg’s equation not being
satisfied
Bc
Residual Strain’ arising from dislocations, coherent
Strain
precipitates etc. leading to broadening
Bs
In the vicinity of θB the −ve of Bragg’s equation not being
Stacking fault satisfied
BSF

Other defects In principle every defect contributes to some broadening

LINE BROADENING EFFECT

B ( FWHM ) = Bi + Bc + Bs + BSF + ...

 SUBTRACTING INSTRUMENTAL BROADENING
 Instrumental broadening has to be subtracted to get the broadening effects
due to the sample
1. Internal Standard
 Mix specimen with known coarse-grained (~ 10µm) and well annealed
(strain free); does not give any broadening due to strain or crystallite size (the
only broadening is instrumental). A brittle material which can be ground into
powder form without leading to much stored strain is good.
 If the pattern of the test sample (standard) is recorded separately then the
experimental conditions should be identical (it is preferable that one or more
peaks of the standard lies close to the specimen’s peaks)

2. External Standard
 Use the same material as the standard (NIST standard: Si, CeO2, LaB6) as
the specimen to be X-rayed but with large grain size and well annealed
 PARAMETERS OF A PATTERN SIMULATION
 Global parameters affect the entire sample
 Background function
 Zero Shift
 Specimen Displacement
 POL (Lorentz polarization factor)
 Sample Geometry
 selecting the Profile Function
 Other parameters are set individually for each phase
 scale factor
 preferred orientation (March-Dollase function)
 unit cell
 atomic parameters are for each atom in the phase
 position, SOF, thermal parameter
 Peak Profile parameters
 Cagliotti peak width function parameters U,V, and W
 Profile shape parameters
 peak asymmetry profile
15
 anisotropic broadening function
 R - INDICES
1/2 1/2
2
∑ 𝑤𝑤𝑖𝑖 (𝑦𝑦𝑜𝑜,𝑖𝑖 − 𝑦𝑦𝑐𝑐,𝑖𝑖 ) 𝑁𝑁 − 𝑃𝑃
𝑅𝑅𝑤𝑤𝑤𝑤 = 𝑅𝑅𝑒𝑒𝑒𝑒𝑒𝑒 =
∑ 𝑤𝑤𝑖𝑖 (𝑦𝑦𝑜𝑜,𝑖𝑖 )2 ∑ 𝑤𝑤𝑖𝑖 (𝑦𝑦𝑜𝑜,𝑖𝑖 )2
o Rwp factor is the more valuable. Its absolute value does not depend on
the absolute value of intensities, but depends on the background. With
a high background is more easy to reach very low values. Increasing the
number of peaks (sharp peaks) is more difficult to get a good value.
 Rwp < 0.1 correspond to an acceptable refinement with a medium
complex phase
 Rwp < 0.15 a is good value for complex phase (monoclinic to
triclinic)
 Rwp < 0.08 is acceptable for a highly symmetric compound (cubic)
with few peaks
o For high background, better to look at the Rwp background subtracted.
o The Rexp is the minimum Rwp value reachable using a certain number 16

of refineable parameters.
 WSS AND GOF (OR SIGMA)
o WSS (Weighted Sum of Squares) is only used for the minimization
routines. Its absolute value depends on the intensities and number of
points.
o GoF (Goodness of Fit) is the ratio between the Rwp and Rexp and
cannot be lower then 1 (unless the weighting scheme is not correctly
valuable: for example in the case of detectors not recording exactly
the number of photons or neutrons).
o A good refinement gives GoF < 2.
o GoF is not a very good index for a noisy pattern because is quite easy
to reach a value near 1.
o For very high intensities and low noise patterns is difficult to reach a
value of 2.
o GoF is sensible to model inaccuracies.
17
LIMIT OF DETECTION

Following the rules of the statistic of Poisson distribution, the absolute and relative standard
deviation σ and σrel of measured number of count N at a 2θ position are given by:
𝜎𝜎 = 𝑁𝑁
𝑁𝑁 1
𝜎𝜎𝑟𝑟𝑟𝑟𝑟𝑟 = =
𝑁𝑁 𝑁𝑁

A usual criterion for the limit of detection (LOD) of a particular reflection, is that
𝑁𝑁𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 > 𝑁𝑁𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 + 3𝜎𝜎𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏

Example: measuring an 𝑁𝑁𝑚𝑚𝑚𝑚𝑚𝑚 of 10,000 counts, the 𝜎𝜎𝑟𝑟𝑟𝑟𝑟𝑟 is 0.01, corresponding to relative
error of 1%, the counting statistic error.

Imagine for a background of 100 counts and a small hump of 120 counts. Clearly this cannot
be classified as reflection because 3𝜎𝜎𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = 30 is obviously higher than 20. The only
solution is to increase the measurement time to improve the peak to background ratio
 WHAT IS PHYSICALLY MEANINGFUL?

Wel (1975)

19
Scherrer (1918) noted that the line profile width is inversely proportional
to crystallite size, Laue (1926) realized that using the integral breadth (IB)
gives approximate independent effect of the distribution in size and shape,
Stokes & Wilson (1944) introduce strain broadening e0 represents the
"mean"deviation from the undistorted state, Bertaut (1949) proposed an
interpretation to consider size broadening without making assumptions on
crystallite size and shape distributions. Jenkins & Snyder (1996) discussed
about Bragg extinction; crystallites larger than 2mm typically have a
sufficient number of planes to display its Darwin width. Balzar (1999)
modeling of microstructure effects; crystallite size and microstrain
comprise Lorentzian and Gaussian component convolutions varying in
2theta, Scardi & Leoni (2001, 2004), Scardi et al. (2010) the WPPM
approach for microstructure analysis. And there's a lot more.
The interpretation of size values is also difficult, as different methods may
report different quantities such as volume weighted mean, area weighted
mean and number weighted mean.
For parallel planes of atoms, with a space d between the planes and for a
given wavelength λ, Bragg’s law defines the (Bragg) angle for constructive
interference. As discussed more in detail by Jenkins & Snyder (1996), if the
incident beam angle theta is away from the Bragg angle so that the phase
shift becomes 1.1λ between adjacent planes, then the phase shift
originating of the 6th plane down the surface will be 5.5 λ resulting in
destructive interference (so-called Bragg extinction). The same is true for
the 2nd and 7th planes etc., thus no net scattering will occur. If theta is set
closer to the Bragg angle so that the phase shift becomes 1.0001 λ between
adjacent planes, then the scattering of the 1st plane will be cancelled by
scattering of
the 5.001th plane, with a phase shift of 5000.5 λ . If the crystal is too small,
then planes needed to cancel such phase shifts are not present; this is the
source of size broadening. As a result intensities at lower and higher angles
than the Bragg angle are observed. Crystallites larger than 2mm typically
have a sufficient number of planes to allow the diffraction peak to display its
inherent Darwin width.