# Changeset 3078

Ignore:
Timestamp:
Sep 15, 2017 3:57:16 PM (6 years ago)
Message:

Update to 2 frame

Location:
Tutorials/MCsimanneal
Files:
57 deleted
22 edited

Unmodified
Removed
• ## Tutorials/MCsimanneal/MCSA in GSAS.htm

 r2390 vondreele Von Dreele 4 768 vondreele 7 1389 2015-03-16T21:14:00Z 2016-07-28T18:42:00Z 1 2431 13861 2017-09-15T20:53:00Z 26 2658 15155 Argonne National Laboratory 115 32 16260 126 35 17778 16.00 Print 124 Clean Clean /* Font Definitions */ @font-face {font-family:"Cambria Math"; panose-1:2 4 5 3 5 4 6 3 2 4; mso-font-charset:0; mso-generic-font-family:roman; mso-font-pitch:variable; mso-font-signature:-536870145 1107305727 0 0 415 0;} @font-face {font-family:Calibri; panose-1:2 15 5 2 2 2 4 3 2 4; mso-generic-font-family:swiss; mso-font-pitch:variable; mso-font-signature:-536870145 1073786111 1 0 415 0;} mso-font-signature:-536859905 -1073732485 9 0 511 0;} @font-face {font-family:Cambria; font-family:"Calibri",sans-serif; mso-fareast-font-family:"Times New Roman"; mso-fareast-theme-font:minor-fareast; mso-bidi-font-family:Calibri;} mso-fareast-theme-font:minor-fareast;} span.SpellE {mso-style-name:"";

In these exercises you will use GSAS-II to solve the structures of 3-aminoquinoline and α-d-lactose monohydrate from powder diffraction data via Monte Carlo/Simulated Annealing (MC/SA). The data sets were kindly provided by Peter Stephens (SUNY StonyBrook) diffraction data via Monte Carlo/Simulated Annealing (MC/SA). The data sets were kindly provided by Peter Stephens (SUNY StonyBrook) and were originally collected on NSLS beam line X3b1. The MC/SA technique is needed in these cases because as you will see the data does not extend very far style='font:7.0pt "Times New Roman"'>      Use the Import/Powder Data/from GSAS powder data file menu item to read the data file into the current GSAS-II project. Change the file directory to MCsimanneal/dataImport/Powder Data/from GSAS powder data file menu item to read the data file into the current GSAS-II project. Change the file directory to MCsimanneal/data to find the file; you may need to change the file type to All files (*.*) to find the desired file.

style='mso-bidi-font-weight:normal'>All files (*.*) to find the desired file.

Quinoli.gda data file in the first dialog and press Open. There will be a Dialog box asking Is this the file you want? Press Yes button to proceed.

mso-hansi-theme-font:minor-latin;mso-bidi-theme-font:minor-latin'>Quinoli.gda data file in the first dialog and press Open. There will be a Dialog box asking Is this the file you want? Press Yes button to proceed.

Select the BNL-I.instprm instrument parameter file in the second dialog and press Open. Youll have to change the file type to *.instprm (GSAS-II instrument parameter file). The powder pattern will immediately appear.

mso-ascii-theme-font:minor-latin;mso-hansi-theme-font:minor-latin;mso-bidi-theme-font: minor-latin'>BNL-I.instprm instrument parameter file in the second dialog and press Open. Youll have to change the file type to *.instprm (GSAS-II instrument parameter file). The powder pattern will immediately appear.

picked 21 peaks, skipping a few very weak ones. Then use Peak fitting/LSQ PeakFit to fit them. You will need to vary at least Peak Fitting/PeakFit to fit them (it will ask to save the project; I called it 2-amino). You will need to vary at least X & Orthorhombic-P). The correct unit cell should almost immediately appear with an M20 ~200 with a=7.748, b=7.651, cell should almost immediately appear with an M20 ~370 with a=7.748, b=7.650, c=12.736, Vol=755.01. NB: your solution may have a & b switched. Do Q; there is a zero point at the end that needs to be excluded. Use the right mouse button and pick a point just inside the end avoiding the zero. Make a note of the exact d-spacing (for me 1.8133) using the mouse cursor on the plot.

there are a few zero points at the end that need to be excluded. Use the right mouse button and pick a point just inside the end avoiding the zeros. Make a note of the exact d-spacing (for me 1.8126) using the mouse cursor on the plot.

Do Pawley refinement box and enter the d-spacing (1.8133) into the Pawley dmin box.

box, enter the d-spacing (1.8126) into the Pawley dmin box and set the Pawley neg. wt.  to 0.001. This will suppress possible negative Pawley reflection intensities.

When done the

id="Picture_x0020_2" o:spid="_x0000_i1051" type="#_x0000_t75" style='width:468pt; height:263.25pt;visibility:visible;mso-wrap-style:square'>

for a single line of text. Do Edit/Add powder histograms; a dialog box will appear

; a dialog box will appear

Select either choice; the desired data set will be added to this phase for analysis. The data tab now shows the new data set.

Do Set All and then OK; the desired data set will be added to this phase for analysis. The data tab now shows the new data set.

id="Picture_x0020_5" o:spid="_x0000_i1048" type="#_x0000_t75" style='width:468pt; height:345pt;visibility:visible;mso-wrap-style:square'>

This completes the Pawley refinement setup; be careful that you didnt skip a step as it might not work correctly if you did.

This completes the Pawley refinement setup; be careful that you didnt skip a step as it might not work correctly if you did.

Step 5. Pawley refinement

mso-hansi-theme-font:minor-latin'>aminoquinoline.bck0.gpx. It can be used to recover from a bad refinement. A progress dialog box will appear showing the residual as the refinement proceeds. My refinement completed with Rwp ~12.9%; a dialog box appears asking if you wish to load the result. Press OK. To see the plot select the PWDR line in the GSAS-II data tree (Ive adjusted the width and height).

showing the residual as the refinement proceeds. My refinement (after 2-3 tries) completed with Rwp ~13%; a dialog box appears asking if you wish to load the result. Press OK. To see the plot select the PWDR line in the GSAS-II data tree (Ive adjusted the width and height).

minor-latin;mso-hansi-theme-font:minor-latin'>sig to 100000; mso-ascii-theme-font:minor-latin;mso-hansi-theme-font:minor-latin'>1000; for the peak only refine

id="Picture_x0020_7" o:spid="_x0000_i1046" type="#_x0000_t75" style='width:468pt; height:207.75pt;visibility:visible;mso-wrap-style:square'>

style='font-family:"Calibri",sans-serif;mso-ascii-theme-font:minor-latin; mso-hansi-theme-font:minor-latin'>Calculate/Refine
; my Rwp fell to ~8.5%. In Rwp fell to ~10%. In Background select Calculate /Refine again. I got Rwp~7.5% at convergence and the fit looked like

c

/Refine again. I got Rwp~8.3% at convergence and the fit looked like

src="MCSA%20in%20GSAS_files/image027.jpg" v:shapes="Picture_x0020_9">

MC/SA structure solution consists of optimizing the position mso-hansi-theme-font:minor-latin'>Residue rigid bodies and then do Edit Body/Import XYZ. Change the file type to XYZ file (*.xyz) and select aminoquinoline.xyz. The Rigid bodies window will show

mso-ascii-theme-font:minor-latin;mso-hansi-theme-font:minor-latin'>Edit Body/Import XYZ. Change the file type to XYZ file (*.xyz) and select aminoquinoline.xyz. The Rigid bodies window will show

You should change the Residue name to something meaningful, I used amino.

If you select the

id="Picture_x0020_32" o:spid="_x0000_i1042" type="#_x0000_t75" style='width:468pt; height:406.5pt;visibility:visible;mso-wrap-style:square'>

The orientation is determined by rotation and translation about a coordinate system centered on C1 and defined by the vectors C1-C2 and C1-C3; you can change these via the pull downs (you cant pick the same atom for two of these so pick them appropriately; I chose C2, C3 & C6). My new model is

C1-C3. The C1-C2 vector defines the X-axis (red), the cross product of C1-C2 and C1-C3 defines the Z-axis (blue; default view direction) and the Y-axis is defined as Z-axis cross X-axis (green; down toward C5). You can change these via the pull downs (you cant pick the same atom for two of these so pick them appropriately; I chose C2, C3 & C6). At each choice the structure will be transformed accordingly. My new model is

Notice that the model includes H-atoms; they are not really needed for MC/SA and will just make computation times a bit longer than necessary. You should change the Residue name to something meaningful, I used amino.

The molecular coordinate system has been changed by you assignment of reference atoms. The C2-C3 vector is now the X-axis (red), the C2-C6 vector is ~ the Y-axis (green) and the Z-axis (blue) is ~normal to the molecular plane. The rigid body coordinates reflect this change

The Cart z values are all close to zero and C3 has only a nonzero Cart x. Notice that the model includes H-atoms; they are not really needed for MC/SA and will just make computation times a bit longer than necessary. They can be removed by selecting the Strip H-atoms box; for now leave them in.

Next go to

id="_x0000_i1039" type="#_x0000_t75" style='width:468pt;height:212.25pt; visibility:visible;mso-wrap-style:square'>

And the plot shows the arbitrarily positioned molecules.

And the plot shows the molecular XYZ-axes lined up along the crystal abc-axes; this is expected from our choice of molecular reference atoms (C2, C3 & C6).

You see the four equivalent molecules for space group P212121 whose centers fall within the unit cell; the reference one is that in the lower left corner with C1 at the origin. You can enter values in the x,y,z boxes and see how the structure changes. Similarly, you can enter an angle in Oa or a vector in the Oi,Oj,Okx,y,z boxes and see how the structure changes. Similarly, you can enter an angle in Oa or a vector in the Oi,Oj,Ok  box to see the effects of rotating the molecule. Each of these parameters has a defined range for the MC/SA optimization run we are about to do; they define the multidimensional search volume. Within that volume will be a number of correct solutions to the structure problem all related by translational and space group symmetry. One can limit this volume by choosing more limited ranges, but here well just go with the default. We want to optimize the position and orientation of this molecule; check the x, y & x, y & z boxes and select Oa, Oi,Oj,OkOi,Oj,Ok Vary?

normal'>Reflection set as it give slightly better performance in MC/SA.

gives slightly better performance in MC/SA.

Next, you need to set the minimum d-spacing (No. trials; this is the number of random tries at each temperature in the annealing schedule. I also used 4 MC/SA runs When done the General window should look like

this is the number of random tries at each temperature in the annealing schedule. I also used 4 MC/SA runs When done the General window should look like

id="Picture_x0020_36" o:spid="_x0000_i1035" type="#_x0000_t75" style='width:468pt; height:253.5pt;visibility:visible;mso-wrap-style:square'>

id="Picture_x0020_37" o:spid="_x0000_i1034" type="#_x0000_t75" style='width:468pt; height:405pt;visibility:visible;mso-wrap-style:square'>

If you are lucky (like I was!) then the result is clearly a good solution with a very low Residual (~4.8%). This problem typically gives suitable solutions with Residuals of 4-5%. If your cell indexing resulted in a & b swapped, you should view the structure along b (green). If no good solution appears (e.g. molecules clashing), then you should just rerun MC/SA perhaps using more runs or more trials. Be sure to set the Keep box for any solutions you want to retain; the others will be cleared before the next MC/SA run starts. When you think you have a good one, Select it; the parameters will be copied to the appropriate boxes in the upper part of the MC/SA window and the structure is drawn again. Next we want to refine this good solution with a very low Residual (~7.2%). This problem typically gives suitable solutions with Residuals of 7%; if you had stripped the H-atoms, the best residual is ~10%. If your cell indexing resulted in a & b swapped, you should view the structure along a (red). If no good solution appears (e.g. molecules clashing & high residuals), then you should just rerun MC/SA perhaps using more runs or more trials.

The MC/SA calculations can take advantage of multicore machines. First set the number of cores by doing File/Preferences; select Multiprocessing_cores from the pulldown and enter the number you wish to use (I suggest Maxcores-1). Then press Save current settings; your choice becomes available immediately and is then set for all your future uses of GSAS-II.  Do Compute/Multi MC/SA to use it.

Be sure to set the Keep box for any solutions you want to retain; the others will be cleared before the next MC/SA run starts. When you think you have a good one, Select it; the parameters will be copied to the appropriate boxes in the upper part of the MC/SA window and the structure is drawn again. Next we want to refine this solution so go back to the Compute/MC/SA; be sure to select the best one before starting. The residual should drop to a be sure to select the best one before starting. The residual might drop to a much lower level. This refinement can be repeated with tighter restriction on the ranges; I used 5% for a final run and got

the ranges. My result (no improvement on 7%) was

This is clearly a good result (R ~ 2.6%!) and the structure

This is clearly a good result (R ~ 7.2%!) and the structure

is identical to the published one

Has a different origin choice (a & b might also be switched) than the published one

src="MCSA%20in%20GSAS_files/image088.jpg" v:shapes="Picture_x0020_24">

as drawn by the Mercury program

id="Picture_x0020_40" o:spid="_x0000_i1030" type="#_x0000_t75" style='width:468pt; height:402.75pt;visibility:visible;mso-wrap-style:square'>

And they will be listed in the Atoms table.

id="Picture_x0020_41" o:spid="_x0000_i1029" type="#_x0000_t75" style='width:468pt; height:288.75pt;visibility:visible;mso-wrap-style:square'>

It is now ready for Rietveld refinement; this needs to use minor-latin;mso-hansi-theme-font:minor-latin'>Calculate/Refine from the main GSAS-II data tree window; my Rwp was ~9%. Next check the Refine unit cell in the Cryst. size & mustrain boxes and do another Calculate/Refine; I got an Rwp ~8.5% for this. Weve not refined atom positions or thermal parameters; to do this we need to describe the structure as a rigid body.

mso-ascii-theme-font:minor-latin;mso-hansi-theme-font:minor-latin'>. size & mustrain boxes and do another Calculate/Refine (twice) I got an Rwp ~8.2% for this. Weve not refined atom positions or thermal parameters; to do this we need to describe the structure as a rigid body.

Step 9. Rigid body model

minor-latin'>RB Models tab; it will be empty. Do Edit/Assign mso-ascii-theme-font:minor-latin;mso-hansi-theme-font:minor-latin'>Edit Body/Assign atoms to rigid body; then select

id="Picture_x0020_42" o:spid="_x0000_i1028" type="#_x0000_t75" style='width:468pt; height:195.75pt;visibility:visible;mso-wrap-style:square'>

reorientation as the three atoms are selected. Press Ready when done; the structure will be drawn with yellow bonds indicating that it is now a rigid minor-latin;mso-hansi-theme-font:minor-latin'>OK
when done; the structure will be drawn with orange bonds indicating that it is now a rigid body and not independent atoms. The

To refine the rigid body parameters check the

To refine the rigid body origin position check the Refine? box, select Calculate/Refine; after a couple of refinement runs I got an Rwp ~7.9% and the RBModels window shows the new parameters

and the profile shows the fit

~7.9% and the RB Models window shows the new parameters

and the profile shows the fit

This completes the structure analysis for 3-aminoquinoline.

• ## Tutorials/MCsimanneal/MCSA in GSAS_files/filelist.xml

 r2390
Note: See TracChangeset for help on using the changeset viewer.