CCP14 Homepage - Single Crystal and Powder Diffraction - Methods, Problems, Solutions and Tutorials - Example Single Crystal Refinement in GSAS

[CCP14 Home: (Frames \| No Frames)] CCP14 Mirrors: [UK] \| [CA] \| [US] \| [AU]	What's New	Introduction	Site Map
Search the CCP14	Download Programs What do you want to do? (lists of software by crystallographic method)	Tutorials	Solutions

(This Webpage Page in No Frames Mode)

CCP14

Methods, Problems and Solutions - and Tutorials

GSAS (General Structure Analysis System) Rietveld powder diffraction and Single Crystal software

Example Single Crystal Refinement in GSAS

The CCP14 Homepage is at http://www.ccp14.ac.uk

[Back to Tutorials] [Back to Problems and Solutions] [Back to GSAS Hints/Resources]

[The reference to use for GSAS in any resulting publications is: A.C. Larson and R.B. Von Dreele, "General Structure Analysis System (GSAS)", Los Alamos National Laboratory Report LAUR 86-748 (1994).]

This example uses the raw hkl data used to determine the crystal structure of Cs₂TiSi₆O₁₅ as published in I.E. Grey, R.S. Roth, M.L. Balmer, Journal of Solid State Chemistry, 131, 38-42 (1997).

The GSAS General Structure Analysis System, while being well known for Rietveld refinement of powder diffraction data, also has a well developed single crystal refinement facility that has functionality not readily available in other refinement software. This includes the ability to refine on f' and f''.
GSAS has a large number of options. Skim through the manual and also take a browse through the menu system to get an idea of the scope of the program.
NOTE: The following is just an example of using GSAS for single crystal structure refinement. Mistakes or refinement mistakes may be present and the following guide is no substitute for thinking for yourself!!! The following is simplified for brevity and be warned you should be more gentle on refining light atoms when heavy atoms are present. Releasing light atoms thermal values too early as we too here can put you in a world of pain! Also, the GSAS fourier Map modules can be used to generating Fourier maps and looking for remaining atoms. Use of the GSAS Fourier Generation and display Module is the subject of a separate tutorial.
GSAS has a relatively gentle learning curve when doing inorganic single crystal structure refinement from scratch (compared to what can occur if you try to learn Shelx with no available Shelx gurus being nearby)
It should be noted that the GSAS manual has an excellent tutorial example of using the single crystal refinement capabilities of GSAS. The main difference here is we use Louis Farrugia's GUI WinOrtep [CCP14 Mirror] to get a graphical track on how the structure is going (GUI WinOrtep can read in GSAS EXP files). Plus in this tutorial we use a different structure to the one in the GSAS Manual. (Louis Farrugia's GUI WinStruplo can be used for Polyhedral Drawing (which can also import GSAS EXP files))

The starting information that is normally known before starting the Structure refinement:

Wavelength of radiation used.
- 0.71073 Angstrom (Mo K-alpha)
Space group
- C2/c
Unit Cell Dimensions:
- a=13.3860(0.002) b=7.4230(0.002) c=15.1340(0.001) beta=107.710(0.05)
- 13.3860 7.4230 15.1340 107.710
- 0.002 0.002 0.001 0.05
Cell contents/Composition:
- Cs₂TiSi₆O₁₅
- Z (number of formula units in the cell) =4
2-theta scan range
- 2 to 30
Colour and habit of crystals
- colorless, transparent tabular-shaped crystals.
Crystal size
- 0.24mm x 0.24mm x 0.10mm

Absorption Correction: Analytical using crystal faces (distances in mm)

FACE      2   0   0   0.12000
FACE     -2   0   0   0.12000
FACE      0   2   0   0.12000
FACE      0  -2   0   0.12000
FACE      0   0   2   0.05200
FACE      0   0  -2   0.05200

Starting Structure as provided by Dirdif, using the automatic Patterson mode, and manually put into a GSAS "input" friendly format.

i n Cs  0.11728   0.24161   0.40994 1 Cs1 i 0.01 
i n Ti  0.75000   0.25000   0.50000 1 Ti1 i 0.01
i n Si  0.49775   0.26886   0.35315 1 Si1 i 0.01 
i n Si  0.32427   0.57048   0.33326 1 Si2 i 0.01 
i n Si  0.34527   0.75968   0.35567 1 Si3 i 0.01 
i n O   0.38140   0.23510   0.33055 1 O1  i 0.01 
i n O   0.69838   0.05116   0.55721 1 O2  i 0.01 
i n O   0.60763   0.31049   0.23030 1 O3  i 0.01 
i n O   0.24664   0.53771   0.23030 1 O4  i 0.01 
i n O   0.50000   0.24996   0.25000 1 O5  i 0.01 
i n O   0.73618   0.42402   0.59014 1 O6  i 0.01 
i n O   0.41875   0.43582   0.34859 1 O7  i 0.01 
i n O   0.45180   0.07945   0.37183 1 O8  i 0.01

GSAS can read in CAD4 and Siemens P3 data and apply PSI scan corrections but the following case uses HKL data collected on an old Siemens AED 3-circle with home made data collection software.
For the raw Shelx Direction Cosine based data, the WinGX single crystal suite was used to create a Analytical Aobsorption Corrected.

To make the Shelx format HKL happier to input into GSAS, replace the normal Shelx 93/97 file ending line of 0 0 0 0.0 0.0 with END or some such equivalent piece of text. The PFE text editor [CCP14 Mirror] is an excellent ASCII file editor for Windows for doing this type of editing of large files.

Copy over the default GSAS Single Crystal X-ray parameter files of CAD4.PRM. In the following case, we use this as a dummy file to keep GSAS happy when it asks for the parameter file.

Run GSAS for PC which in the Windows case should give the following

In the menu select, Setup, Expname, browse to the directory where the HKL single crystal file is located and enter a new filename for this refinement. In this case CSTI. GSAS will prompt that this file does not exist and whether you want to create it. Confirm that you want to create this file.

In the menu select, Setup, Stxldata, which is the utility to import the Shelx HKL file (or anyother type of processed HKL file). Stxldata will not be able to find a csti.exp file so just type Yes that you want to create it.
Give the refinement a title (e.g.: Cesium Titanium Silicate refinement) and then you will be prompted with menu options. If at any point you are not sure what the characters represent, just press enter and it will present you with a brief explanation.

Before you are able to continue, you must insert a phase (basic information on the structure)
- Select I for new phase
- Enter a title (i.e., cstisilicate)
- When prompted, enter the spacegroup: C 2/c
- When prompted, enter the cell constants: 13.3860 7.4230 15.1340 107.710

Select S to enter the ESDs of the cell constants.
0.002 0.002 0.001 0.05

Type x to exit from this option.

Now you are prompted for the single crystal parameter file,
Type cad4.prm or what ever your parameter filename is titled.

When prompted for the non-GSAS reflection file, input the csti.hkl (or whatever your HKL filename is titled. (remember that as per the above instructions, the last file of the HKL file is defined with the word END.
Confirm that this filename is correct.
When prompted for how many characters from each record (or line) are to be used, use all 28 characters which includes the h, k, l, FoSQ and sFoSQ.
Confirm the first reflection when it is listed.
In this case, the csti.hkl file is readable as freeformat so type Y for yes. If in doubt you would have to refine this manually via GSAS prompted questions.
When prompted, 5 data fields are to be read.
Then enter the data fields, h k l FoSQ sFoSQ
Enter a scale factor of 1. Note that GSAS refers to data files (diffractograms or HKL files) as histograms and can refine on multiple sets of histograms/datafiles.
When prompted for the last record flag (in this case END), enter this surrounded by double quotes: "END"
When prompted, the first and last record flags in the above use of the word END is 1 and 3 so enter: 1 3
When prompted for the PHI angle, just go / to get the defaults as this is not relevant for this dataset.
When prompted if this is correct, type Y
When prompted insert the wavelength type, in this case it is Mo
If at this point, you obtain
```
run-time error M6103: MATH
- floating-point error: divide by zero
```
You most like input an incorrect option. The "last line" flag is especially vunerable to this. If this occurs to you, run sxtldata again and when prompted with:
```
Histogram  1 is single crystal constant wave length
Shall we replace it? (<Y>,N) >
```
Select Yes and have another go. It does not start from scratch but from where you were reading in the HKL file.

If you entered everything correctly, you will get the following statement of success. GSAS has imported the HKL data into its internal format. Sxtldata should then exit automatically.

Lattice constants  13.3860  7.4230 15.1340
           angles    90.00  107.71   90.00

Orientation angles     .00     .00     .00
Is this correct? (<Y>,N) >y
What radiation was used? (Cu,Mo,Other) >Mo
<sigma(I)>/<I> =   .019
SXTLDATA run ended successfully

In other situations, you may go into Expedt to setup more single crystal options such as the selection or deselection of datasets, setup of PSI scan information for SCABS. By default, GSAS will use the dataset we imported previously so we can go directly to Scmerge to merge the single crystal data.
Via the GSAS menu, select Compute, Scmerge to merge the data. R factors based on F and Fsq will be produced and these should be noted down.

The control of the refinement is performed in the Expedt module
Via the GSAS menu, select Setup, Expedt
You are normally prompted whether you wish to make a backup, not make a backup file or review a previous file. Normally Y for creating a backup is a good idea.
Enter L for Least Squares refinement setup.
Enter A for Edit Atom Parameters.
You can enter the atom positions manually, or using the above data, do an Edit Copy; then an Edit Paste into the EXPEDT window to add the atoms. (it is best to do computer edit copy, edit paste as this minimises possible errors). Doing an L for list, gives a summary of the inserted atoms for you to check.
Note that FRAC=1 is full occupancy irrespective of the atom site.
At this point, before setting any atom refinement parameters, it might be best to go into Least Squares controls, set the number of refinement cycles desired and let the Scale Factor refine to a reasonable value. Type X to exit the add atom option, L for least squares refinement controls; C for changing maximum number of cycles and set this to 9. Exit out of Expedt by selecting X until you exit.
(This can all be done in one operation if you wish: X A C 9 X X X - which is how the speed users of GSAS operate. From here on, this type of use of GSAS will be used to help with brevity - and to also show how quickly GSAS can be used once you get used to the menu system)

To run the Least Squares program via the GSAS menu, select Compute, Genles.
The output file produced by GSAS can be viewed at will via an ASCII editor such as PFE by opening the filename.lst file (in this case titled csti.lst)

Before continuing on with the refinement, the above mention GUI WinOrtep or GUi WinStruplo can be used to graphically follow the structure refinement and do quick checks on bond-lengths, bond-angles (using the Options, Write Listfile menu) and thermal values (using the Options, List RMS menu).
Just keep re-importing the file after each round of refinement.
This can also give a graphical indication if some atoms may be giving problems (false peaks in structure solution, split atoms, etc).

To set the atom positions and isotropic thermals to refine: Via the GSAS menu, select Setup, Expedt
Then Type, Y L A V 1:13 X U
(Go through the program slowly to see all the manu outputs if you are a beginner on GSAS. The above says: make a backup exp file, least squares, edit atoms, change refinement flags on atoms 1:13 setting atom positions and thermal parameters to refine)
(Like Shelx93 and above, GSAS will handle all the symmetry constraints - general positions, etc)

Then Type, L to check that all the codewords are set as desired.

Give atom editing command
 (<?>,$,C,D,E,F,I,K,L,M,S,T,U,V,X,+,-,*,/) >l
SER TYPE    X       Y       Z      FRAC   NAME    UISO    CODE  STSYM MULT FXU
  1 CS    .11728  .24161  .40994 1.00000 Cs1       .01000 I XU     1     8 000
  2 TI    .75000  .25000  .50000 1.00000 Ti1       .01000 I XU    -1     4 000
  3 SI    .49775  .26886  .35315 1.00000 Si1       .01000 I XU     1     8 000
  4 SI    .32427  .57048  .33326 1.00000 Si2       .01000 I XU     1     8 000
  5 SI    .34527  .75968  .35567 1.00000 Si3       .01000 I XU     1     8 000
  6 O     .38140  .23510  .33055 1.00000 O1        .01000 I XU     1     8 000
  7 O     .69838  .05116  .55721 1.00000 O2        .01000 I XU     1     8 000
  8 O     .60763  .31049  .23030 1.00000 O3        .01000 I XU     1     8 000
  9 O     .24664  .53771  .23030 1.00000 O4        .01000 I XU     1     8 000
 10 O     .50000  .24996  .25000 1.00000 O5        .01000 I XU   2(010)  4 000
 11 O     .73618  .42402  .59014 1.00000 O6        .01000 I XU     1     8 000
 12 O     .41875  .43582  .34859 1.00000 O7        .01000 I XU     1     8 000
 13 O     .45180  .07945  .37183 1.00000 O8        .01000 I XU     1     8 000
Phase No. 1; Phase has     13 atoms; Title: cstisilicate
Give atom editing command
 (<?>,$,C,D,E,F,I,K,L,M,S,T,U,V,X,+,-,*,/) >

Another option if you find the refinement is not stable is to set the Damping parameters to an appropriate value. As an example, to set the damping parameters for the atom positions to 7, type D 1:13 x 7. To set the damping parameters for the thermal values to 7, type D 1:13 u 7.
Using L (list) gives the following output to show that damping parameters are set.

SER TYPE    X       Y       Z      FRAC   NAME    UISO    CODE  STSYM MULT FXU
  1 CS    .11728  .24161  .40994 1.00000 Cs1       .01000 I        1     8 077
  2 TI    .75000  .25000  .50000 1.00000 Ti1       .01000 I       -1     4 077
  3 SI    .49775  .26886  .35315 1.00000 Si1       .01000 I        1     8 077
  4 SI    .32427  .57048  .33326 1.00000 Si2       .01000 I        1     8 077
  5 SI    .34527  .75968  .35567 1.00000 Si3       .01000 I        1     8 077
  6 O     .38140  .23510  .33055 1.00000 O1        .01000 I        1     8 077
  7 O     .69838  .05116  .55721 1.00000 O2        .01000 I        1     8 077
  8 O     .60763  .31049  .23030 1.00000 O3        .01000 I        1     8 077
  9 O     .24664  .53771  .23030 1.00000 O4        .01000 I        1     8 077
 10 O     .50000  .24996  .25000 1.00000 O5        .01000 I      2(010)  4 077
 11 O     .73618  .42402  .59014 1.00000 O6        .01000 I        1     8 077
 12 O     .41875  .43582  .34859 1.00000 O7        .01000 I        1     8 077
 13 O     .45180  .07945  .37183 1.00000 O8        .01000 I        1     8 077
Phase No. 1; Phase has     13 atoms; Title: cstisilicate

To exit EXPEDT type X X X.
To refine, from the select Compute, Genles.
Reload the EXP file in GUI WinOrtep to graphically check out the structure and bond-lengths/bond-angles.

To set all the Cs and Ti atoms to anisotropic (and re-refine the thermals which still have the refine flag set): Via the GSAS menu, select Setup, Expedt
Then Type, Y L A U CS A U TI A
(Go through the program slowly to see all the menu outputs if you are a beginner on GSAS. The above says: make a backup exp file, least squares, edit atoms, convert atom thermal factors on the Cs and Ti atoms)

Then type, L to check that all the codewords are set as desired. After a while, you may neglect to use this option if you are confident with the menu system.

SER TYPE    X       Y       Z      FRAC   NAME    UISO    CODE  STSYM MULT FXU
  1 CS    .11721  .24465  .40994 1.00000 Cs1              A XU     1     8 000
U11,U22,U33,U12,U13,U23 =   .02662   .02662   .02662   .00000   .00810   .00000
  2 TI    .75000  .25000  .50000 1.00000 Ti1              A XU    -1     4 000
U11,U22,U33,U12,U13,U23 =   .00748   .00748   .00748   .00000   .00227   .00000
  3 SI    .49789  .26755  .35435 1.00000 Si1       .01019 I XU     1     8 000
  4 SI    .32590  .56302  .33806 1.00000 Si2       .01488 I XU     1     8 000
  5 SI    .38018  .75933  .32898 1.00000 Si3       .08248 I XU     1     8 000
  6 O     .34209  .21094  .33470 1.00000 O1        .06893 I XU     1     8 000
  7 O     .70177  .04815  .55685 1.00000 O2        .02707 I XU     1     8 000
  8 O     .58911  .28363  .43970 1.00000 O3       -.08294 I XU     1     8 000
  9 O     .23896  .51806  .22767 1.00000 O4        .07698 I XU     1     8 000
 10 O     .50000  .28801  .25000 1.00000 O5        .03258 I XU   2(010)  4 000
 11 O     .72999  .43235  .58513 1.00000 O6        .01350 I XU     1     8 000
 12 O     .41875  .41217  .36522 1.00000 O7        .05243 I XU     1     8 000
 13 O     .44281  .07473  .35961 1.00000 O8        .07660 I XU     1     8 000
Phase No. 1; Phase has     13 atoms; Title: cstisilicate
Give atom editing command

To exit EXPEDT type X X X.
To refine, from the select Compute, Genles.
Reload the EXP file in GUI WinOrtep to graphically check out the structure and bond-lengths/bond-angles.
(If some of the atoms are going beserk, these can be reset as well as individual damping factors applied using the D (Modify atom damping factors) option applied in an analogous way to setting the above parameters). For some structures, the light atoms may have to be initially fixed and only released later after the heavy atoms have refined.

To set all the Si atoms to anisotropic (and re-refine the thermals which still have the refine flag set): Via the GSAS menu, select Setup, Expedt
Then Type, Y L A U SI A
(Go through the program slowly to see all the menu outputs if you are a beginner on GSAS. The above says: make a backup exp file, least squares, edit atoms, convert atom thermal factors on all the Silicon atoms)
Then type, L to check that all the codewords are set as desired. After a while, you may neglect to use this option if you are confident with the menu system.
To exit EXPEDT type X X X.
To refine, from the select Compute, Genles.
Reload the EXP file in GUI WinOrtep to graphically check out the structure and bond-lengths/bond-angles.
(If some of the atoms are going beserk, these can be reset as well as individual damping factors applied using the D (Modify atom damping factors) option applied in an analogous way to setting the above parameters). For some structures, the light atoms may have to be initially fixed and only released later after the heavy atoms have refined.

To set all the Oxygen atoms to anisotropic (and re-refine the thermals which still have the refine flag set): Via the GSAS menu, select Setup, Expedt
Then Type, Y L A U O A
(Go through the program slowly to see all the menu outputs if you are a beginner on GSAS. The above says: make a backup exp file, least squares, edit atoms, convert atom thermal factors on all the Oxygen atoms)
Then type, L to check that all the codewords are set as desired. After a while, you may neglect to use this option if you are confident with the menu system.
To exit EXPEDT type X X X.
To refine, from the select Compute, Genles.
Reload the EXP file in GUI WinOrtep to graphically check out the structure and bond-lengths/bond-angles. Gui WinORTEP will give a warning if any thermals have gone NPD (Non-Positive Definite) as will GSAS is you run Results, Bijcalc from the main menu.
(If some of the atoms are going beserk, these can be reset as well as individual damping factors applied using the D (Modify atom damping factors) option applied in an analogous way to setting the above parameters). For some structures, the light atoms may have to be initially fixed and only released later after the heavy atoms have refined.

At this point, you may like to do various relevant options such and refine the primary Extinction Via the GSAS menu, select Setup, Expedt
Then Type, Y L O E to get into the extinction options.
Type, T 3 N (N to not use the Lorentzian model) to set up Primary extinction.
Type, V to refine on Extinction.

Type, L to list what has been set.

Current extinction type:
Primary extinction     Value = 1.0000E-10 Refine(Y)  Damp flag = 0
Gaussian model used

To exit EXPEDT type X X X X.
To refine, from the select Compute, Genles.
Reload the EXP file in GUI WinOrtep to graphically check out the structure and bond-lengths/bond-angles. Gui WinORTEP will give a warning if any thermals have gone NPD (Non-Positive Definite) as will GSAS is you run Results, Bijcalc from the main menu.
(If some of the atoms are going beserk, these can be reset as well as individual damping factors applied using the D (Modify atom damping factors) option applied in an analogous way to setting the above parameters). For some structures, the light atoms may have to be initially fixed and only released later after the heavy atoms have refined.

For a tutorial showing how to use GSAS to generate VRML atom/Fourier Contour map output for 3D visualisation: refer to Generating a VRML based 3D Structure and Electron Density Contour Map in GSAS.

[Back to Tutorials] [Back to Problems and Solutions] [Back to GSAS Hints/Resources]

[CCP14 Home: (Frames \| No Frames)] CCP14 Mirrors: [UK] \| [CA] \| [US] \| [AU]	What's New	Introduction	Site Map
Search the CCP14	Download Programs What do you want to do? (lists of software by crystallographic method)	Tutorials	Solutions

(This Webpage Page in No Frames Mode)

If you have any queries or comments, please feel free to contact the CCP14