Lecture notes

Rietveld Method
Isostructural Compounds

Nd(Gd)

Nd2CuO4 Gd2CuO4
a=0.39419nm a=3.8938nm
c=1.21627nm c=11.8810nm
ZNd=0.353 ZGd=0.349
XRD patterns of Solid solution NaSr4-xBaxB3O9
(0≤x≤4) :First cubic borate only with BO3

Cubic borates are estimated below 1%.

L. Wu, X.L. Chen, et. al. 2004

Phase transition of
BaTiO3 from
tetragonal to cubic
at about 132˚C
 YBa3B3O9: Phase transition and structure
determination

S. G.: P63cm (No. 185) S. G.: R-3 (No. 148)

a=9.4235(4)Å, c=17.602(1) Å a=13.0441(1)Å, c=9.5291(1) Å
1100C 1140C
X.Z. Li, X.L. Chen, et. al. 2004
LiAlB2O5: Search for new SHG materials

[B3O7]5- & [AlB2O7]5-

 Structural Data for LiAlB2O5

atoms site x y z B(Å2)

Li (1) 4e 0.0000 0.378 (1) 0.2500 2.6(3)

Li (2) 4e 0.0000 0.159 (2) -0.2500 3.2(3)

Al 8f 0.1948(2) 0.1523(2) 0.1510(2) 1.18(4)

B (1) 8f -0.0662(7) 0.3266(7) -0.0308(8) 1.5(1)

B (2) 8f 0.2341(8) -0.0050(6) 0.4071(8) 1.4(1)

O(1) 8f 0.0582(4) 0.2754(3) 0.1104(4) 1.63(8)

O(2) 8f -0.1232(3) 0.2944(3) -0.1897(3) 0.98(8)

O(3) 8f -0.1386(4) 0.4271(3) 0.0238(4) 1.76(8)

O (4) 8f 0.1581(4) 0.0184(3) 0.2428(4) 1.64(8)

O (5) 8f 0.1846 (3) 0.1071(3) -0.0311(4) 1.19(8)

Final Refinement of New compound of LiAlB2O5

50000
Iobs
40000 LiAlB2O5 Ical
peak position
Iobs-Ical
Inten sity (counts)

30000

20000

10000

-10000
20 40 60 80 100 120
2  (degree)
Structure vs Temperature:KCaCO3F

0.8120(3) 0.1880(3)
0.5 by x-ray data
LiSr4B3O9 : A comparison between structure determination
from single-crystal and powder X-ray diffraction
Single-crystal: Rint=0.0745
R1(all data)=0.0695
wR2(all data)=0.1887
with weighting scheme:
W=1/[σ2(Fo2)+(0.0000P)2+359.71P]
where P=(Fo2+2Fc2)/3
SDPD: RB=0.07
Rp=0.0609
Rwp=0.0811
Rexp=0.0314
a= 14.9470 Å S.G: Ia-3d
(ZB/ZSr)2=(3/38)2 0.6%
What is a Rietveld Refinement?
-a standard treatment of powder diffraction data to
make the final structural model achieve the
accepted criterion;

-a best known method that fully makes use of the

step-mode scanned data to dig out a lot of
structural and other information;

- a procedure for structural solution in nature.

 What can we get to perform a
Rietveld refinement?

Lattice Parameters Quantitative phase Analysis

Atomic Positions Grain size
Atomic Occupancy Incommensurate Structure
Debye Temperatures Structure factors
Crystallinity Phase transitions
Magnetic structures
……
 History Review
• Rietveld originally introduced the Profile Refinement method (Using step-
scanned data rather than integrated Powder peak intensity) (1966,1967)
• Rietveld developed first computer Program for the analysis of neutron data for
Fixed-wavelength diffractometers (1969)
• Malmos & Thomas first applied the Rietveld refinement method (RR) for
analysis of x-ray powder data collected on a Ginier Hagg focusing Camera
(1977)
• Khattack & Cox first applied the RR to x-ray powder data collected on a
diffractometer (1977)
• Conference on Diffraction Profile Anlysis Sponsored by IUCr in Poland,
suggested the term “Rietveld Method”(1978)
• Wiles and Yang developed a general computer program (D.B.W) for both x-
ray & neutron diffraction data (fixed wavelength)(1981)
• Von Dreele, Jorgensen and Windsor extended to the program to the neutron
diffraction data (1982)
• Fitch et al, 193 refined parameters,UO2 DAs.4D2O (1982)
Aminoff Prize, Stockholm,1995
H.M. Rietveld Acta crystallogr., 22, 151 (1967).

H.M.Riveted, J. Appl. Crystallogr., 2, 65 (1969).

Structural model
shift Raw data

Rietveld Refinement

Refined model
 How RM works?
The RM refines a structure by minimizing a
quantity through the Newton-Raphson
algorithm

where, yi is the observed intensity at a certain 2,

yc,i is the calculated intensity at the same angle,
wi is a weight, we usually take wi=1/yi
i=1,2,…n
=( 1 2 …p), the parameters to be refined.
Given a solution =opt(1, 2… p) that approximately satisfy
the above equation. To find a better solution, we begin an
iterative process by expanding into a Taylor series.

 2

2 p
 
2 2
   i
  1
   opt i 1   i
A1=b

1= 0+ 1

1 is a shift.
S is the scale factor of the phase 
Lh contains the Lorentz, polarisation and multiplicity factors.
Fh is the structure factor
Ah is the absorption correction
Ph is the preferred orientation function
Ω is the reflection profile function that models both
instrumental and sample effects
 n
Fhkl   f i e 2 ( hxi  kyi lzi )
i

where, fi atomic scattering factor for ith atom

xi, yi and zi the fractional coordinates for ith atom

f i  f i 0e  M
sin  2 sin  2
M  8 u ( 2
)  B(
2
)
 

The mean square displacement of the atom in a direction normal

to the reflecting planes
  i
n 1
    i  RELAX  CC
i
n

Where RELAX is relaxtion factors that are used

to control the shifts to avoid divergence;
and CC is a multiplier.
 What we need to perform a RR?

A set of step-mode scanned data,

usually 2=10-120˚ or more, step  2=0.02˚
collecting time is instrument dependent from 1-20s
for laboratory diffractometer;

An initial structural model having roughly accurate lattice

constants, correct space group and approximate atomic
positions
How we obtain an initial structural model?

- solid solutions usually adopt same structure types of their

parent compounds;
NaSr4-xBaxB3O9 (0≤x≤4)

- Compounds with same chemical formula

YBa2Cu3O7 and NdBa2Cu3O7
but always alert that exceptions are not uncommon
La2CuO4 and Nd2CuO4

- Try and error

- Ab inito structure determination

 Is the compound known?
Crystallographic Structure Databases

•ICSD (Minerals and Inorganics) •ICDD diffraction data

– http://www.fiz-karlsruhe.de/ – http:http://www.icdd.com/
– Minerals and Inorganic – Inorganic & Organic
– Over 60000 entries – Over 140000 entries

•Cambridge Structure Data Bank ） •NIST Crystal Data

– http://www.ccdc.cam.ac.uk – http://www.nist.gov/srd/nist3.htm
– Organics & Organometallics – Inorganic & Organic
– Over 250000 entries •Over 230000 entries
A new
structural
database(2003):
aimed at freely
retrieving data

18000
Patterns
already!
 Parameters in PCR file
The parameters in PCR file can be divided into three
categories

-relating only to samples, refinable

such as atomic positions, temperature factors
-relating both to samples and instruments
such as scale factors, FWHM (Full width at half maximum)
-user-specified parameters
such as BKPOS, Nba
 Codewords(I)
codewords are used to control parameters
when to be refined, when to be fixed and
when to be constrained and etc. A
codeword is formed as
C=S(10P+CC)
Where S stands for the sign mark, P is an
ordinal number set by users from 1 to p, the
maximum number of parameters
 Codeword(II)
• For example, an atom Ca position is (0,0,z) with z to be
refined from its initial value 0.1. The codeword in your
PCR file looks like the following
……
Ca1 Ca+2 0.0 0.0 0.1 …
0.0 0.0 120.5 …
……
Here, S=1, P=12, and CC=0.5. That means that z of Ca1 is
the 12nd parameter to be refined in the iterative process,
and x and y of Ca1 occupy special positions not needed to
be refined.
 Codeword(III)
• Another example: the lattice constants of a
tetragonal compound are to be refined. The
codeword in your PCR file looks as follows
3.891 3.891 11.732
51.0 51.0 61.0
In this case, constraint is put on a and b by
using the same codeword since a=b always
holds in tetragonal compounds
 Codeword(IV)
One more example: the occupancies of two kinds of atoms
at one site are to be refined. Solid solutions are the most
common among this kind of refinements. The codewords
in your PCR file are set as
Y Y+3 …… 0.8
…… 10.3
Yb Yb+3…… 0.2
…… -10.3
Only in this way are the occupancies guaranteed to satisfy
Occ(Y)+Occ(Yb)=1
 Codeword(V)
- each parameter usually controlled by one
codeword. Be alert that one codeword should be
given to two or more parameters that are irrelevant;

- there is no limit to choose ordinal numbers.

But we usually set the first ordinal numbers to
global parameters such as zero point, background
parameters and the etc.
 Modeling backgrounds
• The background intensity bi at the ith step
may be obtained by any of the following
three method.
- a specified background function, usually a
polynomial;
- linear interpolation between user-selected points
in the pattern
- A user-supplied function
 Control flags
The choice of background type is indicated by a
control flag
• Comment line(4) Job Npr Nph Nba Nex…
0 5 1 0 2
Nba:
=0 Refine background with a polynomial
=1 Read background from file COFHIL.bac
=2,3,…,N linear interpolation between N given
points
…
 11
2
bi   Bm (  1) m
m BKPOS

Where Bm are parameter to be refined

BKPOS is a user-specified parameter, origin of
polynomial function, non-refinable.

If 2=BKPOS, we see bi=B0

Users can look into their data files to set the values
of BKPOS
 Profile functions (I)

Gaussian (G) C00.5

exp( C 0 ( 2 i  2 k ) 2
/ H 2
K)
Npr=0 H K 0.5

C10.5 1

Lorentzian (L) H K (2 i  2 k ) 2
[1  C1 ]
Npr=1 HK2

Parameter to be refined: Hk, Full Width at Half

Maximum (FWHM)
 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6

5 5
Gaussian
Lorentzian
4 4
intensity

3 3

2 2

1 1

0 0

-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6

i-k

Hk=0.2
 Profile functions (II)
2C 205 1
Mod.I Lorentzian 
H K (2 i  2 k ) 2 2
Npr=2 [1  C 2 2
]
HK

C305 1
Mod.I Lorentzian 
2H K (2 i  2 k ) 2 3 / 2
Npr=3 [1  C3 2
]
HK

Parameter to be refined: Hk,

 Profile functions (III)

Psudo-Voigt
L  (1   )G
Npr=5

   0  X *  2 

Parameters to be refined: Hk, η0, X

η0 = shape
 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6
4.5 4.5
0.7G+0.3L
4.0 0.5G+0.5L
4.0
3.5 0.3G+0.7L 3.5
3.0 3.0
intensithy

2.5 2.5
2.0 2.0
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
-0.5 -0.5
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6

i-k

Pseudo-Voigt functions
Hk=0.2
 Profile functions (III)
m
C4  (2 i  2 k ) 
2

1  4 * (2  1)
1/ m
Pearson VII 2 
HK  HK 

X Y
m  m0  100  1000
2 (2 ) 2

Parameters to be refined: Hk, m0, X,Y

 Profile functions (IV)


(Mod-TCHZ pV) L( x)  G ( x)   L( x  u )G (u )du



L(x) and G(x) have different FWHMs HL and HG

HL HL 2 HL 3
  1.36603  0.47719( )  0.1116 ( )
H H H


H  H G5  AH G4 H L  BH G3 H L2  CH G2 H L3  DH G H L4  H L5  0.2

Parameters to be refined: HG and HL

Full width at half Maximum
(FWHM)
Ig
H G  U tan   V tan   W 
2

cos 2

For Npr=0…6, Hk=HG

[Y  F ( S z )]
H L  X tan  
cos 

For Npr=7, HL is required

apart from HG
Typical variations of FWHM vs 2
 Summary for the parameters to be
refined with different profiles

Npr=0, Gaussian: U, V, W, Ig 3

Npr=5, pv: U,V,W,Ig, η0(Shape), X 5

NPr=6, Pearson VII: U,V,W,Ig, η0(Shape), X,Y 6

NPr=7, TCHZpv: U,V,W,Ig, X,Y,Sz 6

 Preferred orientations (I)
Nor=0,
Rietveld-Toraya
Model PH  G2  (1  G2 ) exp(G1 ) 2
H

G1 and G2 are refinable parameters

H is the acute angle between d*H and the
normal to the crystallites (platy habit)

Note: preferred orientation vector Pr1,Pr2 and Pr3

is needed to specify a priori by users
 Preferred orientations (II)
Nor=1, modified
March’s Model
3

 sin 2  H  2
PH  G2  (1  G2 ) (G1 cos  H ) 
2

 G 1 

G1 and G2 are refinable parameters

G1<1, platy habit,

G1=1, no preferred orientation
G1>1 Needle-like habit
 Systematic line-shift
Bragg-Brentano Geometry
• Specimen displacement
 2s
 2  cos  SYCOS
R
• Specimen Transparency
1
 2  sin 2 SYSIN
2R
: the linear absorption coefficient of the sample
WDT·FWHM WDT>5,
preferably 10
Monochromator polarization correct

1  cos 2 2 cos 2 2
LP 
sin 2  cos 

 Incident angle to a monochromator

CTHM=cos22=0.8009 for a graphite

monochromator, CuK
Asymmetry correction for profiles
P1 Fa ( z )  P2 Fb ( z ) P3 Fa ( z )  P4 Fb ( z )
As ( z )  1  
tanh  H tanh 2 H

2 i  2 H  Sshf
z
FWHM

P1, P2, P3 , and P4 are parameters to be refined

AsymLim: peaks below this 2 angle limit

are corrected for asymmetry
 Occupancy
m
Occ  chemOcc 
M
m is the site multiplicity, M is the multiplicity
of the general site for a given space group.

For example, KCaFCO3, P-6m2(187)

K+1 occupies 1(a) site; O-2 3(k) site; general site
12(o), Occ(K+)=1/12=0.08333, Occ(O-2)=0.25, both
atoms’ chemical occupancy=1.0
 Agreement Factors (I)

n 1

  n 
2
y i  y ci
2

  wi yi  yci 
R p  100 i
n Rwp  100  i n 
 
y i   wi yi 
i  i 

Profile Factor Weighted Profile Factor

1
  2

 n p  Expected
Rexp  100  n  Weighted
 w y2 
 
i
i i
 Profile Factor
 Agreement Factors (II)

Rwp
 'I obs , h ' I calc ,h
S RB  100 h

Rexp  'I
h
obs , h
'

Goodness of fit Bragg Factor

indicator

 'F obs , h ' Fcalc ,h

RF  100 h

 'F
h
obs , h
'

Crystallographic RF factor
Variations of agreement factors and
esd.

Hill & Madsen, Powder Diffraction(1987)

An estimation of S for an ideal refinement

 i  yi 1
take wi 
yi

 
( yi  yci )  yi  ( yi  yi )  yi
2 2

1 1
 N
2 
2 N
 2

  wi ( y i  y ci )    i i 
w y 1

   N 2
S i
  i
 
 N P   N P  N  P
   

S≈1 since N>>P

 Gaussian Lorentzian

R.J. Hill and H.D. Flack, J.Appl. Cryst. 20 (1987) 356-361

Durbin-Watson statistic parameters
N

 i i ci i 1 i 1 ci1
w ( y  y )  w ( y  y )  2

d 2
N

 i i ci
[
i
w ( y  y )]2

 N  1 3.0901 
QD  2  
 N  P N  2 

d < QD : positive serial correlation

QD< d <4-QD: no serial correlation
d > 4-QD: negative serial correlation
 Variations of d, eds, Rwp and RB vs cycles

R.J. Hill and H.D. Flack, J.Appl. Cryst. 20 (1987) 356-361

 Comments on agreement factors
• RF and RB are more indicative of structural model fits
• Rp and Rexp are more indicative of overall profile fits
• RF ,RB , Rp and Rexp are not good indices for the
refinements of different patterns
• S should be as close as 1
• d is a more sensitive index over RF ,RB , Rp and Rexp
• At least Rp, Rwp and Rexp should be given when
submitting a paper to a journal
 Quantitative Phase Analysis
S j Z j M jV j / t j
Wj  N

S Z M V /t
i
i i i i i

where, Wj is the weight fraction for the jth phase;

Sj is scale factor for the jth the phase;
Zj is the number formula units per cell for the jth phase;
Mj is the mass of the formula unit;
Vj is the unit cell volume;
tj Brindley coefficient that comes into effect when the
linear absorption coefficients of phases in powder
differ a lot to each other.
KCaFCO3 CaCO3
Multiphase Rietveld Analysis
Result in *.out file

No absorption correction is applied

To obtain a satisfactory quantitative phase analysis
based on the Rietveld method, we should be
cautioned:
- Sample should be carefully prepared: powder is
homogeneous in compositions and have a
sufficient number of grains with random
orientations;
- Structures of phases are well known;
- Absorption correction is applied whenever the
phases differ a lot in their linear absorption
coefficients. The Brindley coefficients can be
consulted in the Fullprof Manual.
Calculation of bond length, angle and bond
valence sum
ESDs
Bond valence sum is a good indicator of
the structural validity. For details, see
I.D.Brown, Acta Crystallogr. B48,
141(1992)
 EPS and Relax factors
Forced termination when shifts<EPS*esd
Recommended EPS=0.1

 in 1   in   i  RELAX  CC

R_at: Atomic parameters, including atomic coordinates and etc.

R_an: anisotropic parameters
R_pr: profile parameters, lattice parameters, preferred
orientations and the etc;
R_gl: global parameters, such as zero shift, backgrounds and the
etc.
2 Excluded 2 regions
Suggested turn-on sequences for the
parameters
1. Scale factors
2. Zero shift
3. Background
4. FWHM
5. Shape1, X, Y,…
6. Lattice parameters (if accurate to some extent otherwise
do it before refining FWHM)
7. Atomic coordinates
8. Temperature factors, occupancies
9. Preferred orientation, GauSiz, LorSiz…
Some factors affecting refinement results
- Low instrument’s resolution. RS=0.1-0.2mm
- Too low counts. Strongest counts >10000
- Too less sample. Sample should fully cover the
sample holder window;
- Overflow in low angle region;
- Too less angle region 2>120˚;
- Too large EPS that leads to false minimum;
- Improper profile function;
- Too less WDT values .
 Error messages (I)
-‘Hole in Matrix’ : the number of parameters to be refined
NPR larger than the number of codewords
For example, you set NPR= 12, while one codeword 80.5 is
missing or 101.0 is mistyped as 11.0
- ‘Negative FWHM’: HG2 <0, meaningless! Increasing the
negative U,V, or W while set smaller Relax values in the
ensuing the refinements

Ig
H G  U tan   V tan   W 
2 2

cos 2 
 Error messages (II)
- ‘No scattering factor’ : atom identifier ‘TYP’ is not
recognizable by Fullprof.
For example, Ca+2 is accepted while Ca2+ is not
accepted.
-‘Too many reflections’: For a given point, there are too
many reflections contributing to the intensity that are
beyond the software’s capacity. Usually this results from
the false FWHM
- ‘Invalid integer’ or ‘Invalid real’: Examine the format
of parameters
Note: The software does not always give the correct
lines where errors occur. Look into the nearby lines!
MAC MXP18A-HF
chenhong FormCon2.0
理学 DMAX 2000
huming
 S j Z j M jV j / t j
Wj  N

S
j
j Z j M jV j / t j

Refine
scale
factor S
Rwp=962 138, S=0.5E-2 0.64E-3
Refine zero point along with S
Rwp=62.8 ZP≈0.05
Refine background along with S and ZP
Rwp=32.9%
Refine lattice parameters along with others
Rwp=28.9%
Refine peak profile along with other parameters
Rwp=18.9%
Refine asymmetry
Rwp=16.8%
Refine atomic coordinates: first two atoms Pb and S
the number of parameters to be refined:22
 Rwp=14.0%

Further refine atomic

coordinates of 3 O atoms
Rwp=13.1%
Refine temperature factors along other
parameters
Rwp=12.5%
Rp=8.92%, Rwp=10.8%, Rexp=6.57%
Calculate the bond lengths and bond valences
And the results are stored in *.dis
 PbSO4

a b
b
c
a
 PbSO4

Set
Fou=4
in your
Pcr file

Fourier synthesis
 Compare with the Rietveld Refinement
Round Robin

Rp=7.3%-16.6% 5.82% 8.91%

Rwp=8.2-20.0% 7.83% 10.8%

Rexp=1.5%-7.0% 4.83% 6.71%
GodF=1.3-7.4 1.6 1.6

Total 23 respondents Background

excluded
 Compare with the Rietveld Refinement
Round Robin

Range mean single crystal this work

a(Å) 8.4764-8.4859 8.4804(4) 8.482(2) 8.4818(1)
b(Å) 5.3962-5.4024 5.3989(3) 5.398(2) 5.3997(1)
c(Å) 6.9568-6.9650 6.9605(4) 6.959(2) 6.9614(1)

R.J.Hill, J. Appl.Cryst. 25, 589(1992)

 Compare with the Rietveld Refinement
Round Robin

Range mean single crystal this work

Pb x 0.1875-0.1883 0.18783(4) 0.1879(1) 0.18785(7)
z 0.1669-0.1683 0.16752(9) 0.1667(1) 0.16742(10)
S x 0.0621-0.0673 0.0642(2) 0.0633(6) 0.0638(4)
z 0.6799-0.6860 0.6838(4) 0.6842(7) 0.6834(6)
O1x 0.902-0.924 0.9083(13) 0.908(2) 0.9069(11)
z 0.585-0.601 0.5945(7) 0.596(3) 0.5929(14)
 Compare with the Rietveld Refinement
Round Robin

Range mean single crystal this work

O2 x 0.177-0.200 0.1850(11) 0.194(2) 0.1894(11)
z 0.523-0.548 0.5398(13) 0.543(2) 0.5423(14)
O3 x 0.071-0.080 0.0778(5) 0.082(1) 0.0789(6)
y 0.018-0.041 0.026(13) 0.026(2) 0.0214(9)
z 0.806-0.819 0.8139(7) 0.809(2) 0.8130(9)
 Compare with the Rietveld Refinement
Round Robin

Range mean single crystal this work

B Pb(Å2) 0.90-2.39 1.42(11) 1.48 1.59(2)
S (Å2 ) 0.29-1.37 0.98(8) 0.74 1.21(8)
O1(Å2) 0.50-4.2 1.24(10) 1.87 1.32(21)
O2(Å2) 0.1-5.8 1.31(13) 1.76 2.02(21)
O3(Å2) 0.8-4.6 1.27(11) 1.34 1.05(13)
Output file: Data.dis

Bond length, Angel, Bond Valence

( Pb )-( O1 ): 2.595(8) Pb : 2.273(9)
( Pb )-( O2 ): 3.022(4)
S:
( Pb )-( O3 ): 2.909(6) 5.714(63)
O1: 1.994(35)
(S )-(O1 ) : 1.4721( 84)
O2: 2.143(45)
(S )-(O2 ) : 1.4486(102)
(S )-(O3 ) : 1.5347( 59) O3: 1.908(21)
( S )-( O3 )-( S ): 56.10(13)
X-ray neutron
X-ray neutrons
Lattice parameters

X-ray

neutron
Agreement indices Temperature factors

FWHM
 Please cite:
Dicvol04
A. Boultif, and D. Loüer
“powder pattern indexing with the successive
dichotomy method”
J. Appl. Cryst. 37, 724-731 (2004)

Fullprof
J.Rodriguez-Carvajal
Program Fullprof.2k, version 3.30, Laboratoire
Leon Brillouin, France, June 2005.