chem cryst
CRYSTALS manual
University of Oxford

Crystals Worked Examples

Chapter 2: Solving And Refining A Routine Structure In Crystals

2.1: Some information about the structure you are about to solve and refine

2.2: Step one: Import cell and crystal data

2.3: Step two: Import reflection data

2.4: Step three: Initial assessment of the data

2.5: Step four: Structure solution.

2.6: Step five: Check atom types

2.7: Step six: Commence refinement...

2.8: Step seven: Adding Hydrogen Atoms

2.9: Step eight: More refinement & Extinction

2.10: Step nine: Choose a nice weighting scheme

2.11: Step ten: Validation and CIF archival.

This natty material was supplied as very poor colourless crystals found congealed in the bottom of a half-abandoned flask.

A fragment of crystal (0.3 x 0.4 x 0.4 mm) was mounted in oil on a KCCD diffractometer at 190K and a data set collected in two hours.

The space group is P 21 21 21

The GUIDE will now suggest "Import reflections". Click OK.

Before importing reflections, CRYSTALS needs to know whether you plan to use F or F2 in the least squares minimisation. (You may change this option later). Click No. (for now).

The next dialog advises you about treatment of Friedel opposites. Read, and then click OK.

When asked "Do you want to merge the data", click Yes.

The filter reflection dialog then appears. This lets you omit reflections based on various thresholds. Click OK to accept an I>3 sigma(I) cutoff.

It is useful to get an idea of the quality of your data before proceeding.

The guide has moved onto the "Solve" stage, but is recommending "Initial analyses" so that you can check the data.
Click OK.
The initial analyses window appears:

Click on each tab, and convince yourself that the data looks reasonable. Some of the graphs allow you to choose cut-off limits for the data (based on I/sigma(I) or ((sin theta)/lambda)**2) if you click with the right mouse button. However, don't do this for now.
The last tab is particularly useful as it checks that the cell contents are reasonable given the cell volume, and that the number of observations is reasonable given the expected number of final parameters.

Dismiss the window by clicking Close at the bottom right.

The guide should now be recommending Run Sir92.
Click OK

There are four preset jobs for Sir92. Usually the default method will work.

SIR 92 should solve the structure in under a minute, depending on the processor speed of your machine. Click the "Quit" button (top left) twice to close down SIR.

Click Yes to import SIR's solution back into CRYSTALS.

Click Yes.
The atoms will be numbered, so that as far as possible, adjacent atoms have sequential serial numbers.

You will see that the GUIDE is now recommending "Refine posn and aniso", but first we need to sort out the structure using the graphical model interface.

Change the types of the two incorrect atoms so that the model matches the expected chemical diagram (see top of page for chemical diagram).

To rotate the structure:
Point into some empty space, hold down the left-mouse button and drag the mouse around.

To change an element:
Point at the incorrect atom with your mouse. Press the right-mouse button. Move the pointer down to "Change type", and then choose the new element type from the sub-menu:

Make sure the model matches the expected structure before continuing.

You may wish to change the model style from Ball to Ellipse so that you can see how the anisotropic temperature factors are behaving as the structure refines.
To do this click the Ellipse button on the toolbar above the model:

The guide is recommending refinement. (Refn posn and aniso)
Click OK to start.

Click OK to set up the least squares directives as specified.
Refinement will start automatically (unless you check the Advanced box, in which case you can edit the directives and choose the number of cycles).

CRYSTALS will carry out some rounds of refinement, the R-factor should drop to somewhere around 8%:

The GUIDE has decided that it is time to add hydrogen atoms.

Click OK to do this.

White atoms: Geometrically placed H
Pink atoms: Difference Fourier map H

You can see that in general the hydrogen atoms have been computed correctly (co-incident white and pink atoms), but that there is still a hydrogen missing from the Nitrogen atom. (The double C=N bond is an artefact of the number of Q peaks bonded to the neighbouring C atom - it will go away in a minute!)

There is still one H atom missing. Click Continue.

Follow the GUIDE, it will recommend more refinement, then Add Hydrogen again.

This time the missing hydrogen atom will be found in the Fourier map. It is currently labelled QH(1). Using the right-click method from step five, change the element type to Hydrogen.

Check the box that says "All H atoms have been found".
Click Continue.

Carry out some more refinement by clicking OK on the GUIDE.

This time, the refinement setup offers a choice of how to treat the H atoms:

Either leave the H's unrefined (fixed) or Ride them.

Set up and carry out the refinement by clicking OK.

Next the GUIDE recommends an extinction check:

Click OK

The extinction check graph is displayed:

It plots Fo against Fc. If extinction is a problem for the crystal, the graph will flatten out (drop under the blue Fo=Fc line) at high values of Fc.
Extinction isn't a problem here, however, two of the reflections are clearly outliers.

Remove the outliers by right-clicking on the offending points and choosing "Omit".
FYI: Likely cause - beam trap partially obscuring the image on the diffractometer - now fixed!

Click the "Do not" button to close the window and continue without an extinction correction.

Carry out one more cycle of refinement to account for the reflections just omitted.
We now have an opportunity to pick an optimal weighting scheme. The GUIDE recommends "Optimise Weights"
Click OK
This dialog lets us choose between calculated and statistical weights:

We will go with calculated for the time being. (Assumption: the model is more or less correct, therefore the residual gives a good estimate of any errors - let's fit a function to it.)

Make sure the Optimal is selected in the list, then click OK.

Read the information and pick a function to use for weights:

In this case the 3 parameter calculated scheme is fine:
Click Use 3 parameter scheme.

Click OK on the GUIDE to carry out a few last cycles of refinement.

The GUIDE now recommends "Validate" (this means that it's happy that the structure is complete).
Click OK to validate the structure.

A list of tests and any failures will appear in the text window on the left:

If the shift/esd is causing a warning then:

Change the GUIDE default option to Refine posn and aniso and click OK.

Then carry out the validation again.
This time, there should be no problems with the structure.

If all checks passed, the GUIDE will recommend Publish.
In any case, ensure that Publish is the selected option in the GUIDE and click OK.

When asked if you want 6 final cycles of least squares, click Yes.

A variety of data formats are available for publication/archiving.
Choose CIF, which contains just about everything that you need.
Click OK to write a CIF.

Open the CIF for editing if you want to. Close the Publish dialog.

From the menus choose "Results"->"Checkcif on the web" for further checks.
The cif is c:\wincrys\demo\cyclo\publish.cif.

To close CRYSTALS choose Exit Crystals from the File menu.

Alternatively, to get back to the workshop starting point, choose Demo from the Help menu.


[Getting Started | Solving And Refining A Routine Structure In Crystals | Poor Quality Data - Tetraphenylene | crysworkshop-4 ]