A Practical Guide to Solving Single Crystal Structures. Manuel A. Fernandes. 15 Mar 2006

A Practical Guide to Solving Single Crystal Structures Manuel A. Fernandes 15 Mar 2006 School of Chemistry University of the Witwatersrand Johannesburg, South Africa, 2006

Solving Crystal Structures 1 Frequently asked questions What are the aims of this manual? The aims of this document is to enable students and researchers to fully solve their crystal structures such that they are able to submit their structures to Acta Crystallographica C or E (from now on referred to as an ACTA paper; see www.iucr.org) or insert it into their thesis documents without the help of a crystallographer. What is covered by this manual? The following areas are covered: [1] The use of WinGX, SHELX-97, ORTEP-3 and PLATON to solve crystal structures. [2] The use of CIFTAB (part of SHELX-97 and incorporated into WinGX) to finish of crystal structures in association with PLATON. Included will be information on what is needed to finish a crystal structure for insertion into a paper or thesis. [3] The use of PLATON, ORTEP-3 and PovRay to draw pictures. [4] How to use all the above software to prepare your structure for an ACTA paper. [5] Some general guidelines on how to discuss your structure in a synthesis journal, a more crystallographic journal and the ACTA journals. As part of this section some discussion of crystal packing, classical hydrogen bonding and weak interactions will be covered.

Solving Crystal Structures 2 Required software In order to solve and publish crystal structures the following software will need to be downloaded and installed: WinGX. The main software package required to solve and refine structures. It includes SHELX, Sir-97 and PLATON. There are other packages that do similar things to WinGX but in my opinion WinGX is one of the easiest to use and to teach and hence is presented here. Download from http://www.chem.gla.ac.uk/~louis/wingx. PLATON. This software package allows one to pick up errors in a crystal structure solution, do many crystallographic manipulations and draw molecular and packaging diagrams of the structure. Download from http://www.chem.gla.ac.uk/~louis/software/platon. ORTEP-3. This software program allows to draw ORTEP, packing and povray diagrams. Download from http://www.chem.gla.ac.uk/~louis/software/ortep3. Mercury. This software program allows one to find and draw interesting interactions in crystal structures. It is also possible to export some of these pictures but they are often not of good quality. Download from http://www.ccdc.cam.ac.uk/mercury/. Encifer. An editor which is designed for editing CIF files. Though not a bad program I tend to prefer simpler editors such as wordpad or SCINTILLA (available at http://www.scintilla.org). Download ENCIFER from http://www.ccdc.cam.ac.uk/free_services/encifer. GhostScript and GhostView. These programs allow one to convert the graphical output files from PLATON and ORTEP-3 to other formats or to paste them directly into MS-word or MS-power point. Download from http://www.cs.wisc.edu/~ghost/. Povray. This is a scene rendering program that allows some very fancy graphics to be rendered. It is not a crystallography program. All it does is convert a set of instructions (contained in a POV file) into a picture. Some people like it though I almost never use it. Download it from http://www.povray.org. It is assumed that MS-office is installed on your computer. If you cannot get hold of MS-office then you can use OpenOffice (available from

Solving Crystal Structures 3 www.openoffice.org) instead which is a very good and free alternative allowing read and write access to all of the MS-office formats. It is also assumed that you have access to the internet and know how to use one of the following web browsers: Internet Explorer, Mozilla or Firefox (the last two are available from www.mozilla.org). In addition you will need to know how to extract data from ZIP files. You will need Winzip (www.winzip.com) or unzip (www.unzip.org) to do this though many other programs that are able to do this are available.

Solving Crystal Structures 4 Chapter 2: Solving your first crystal structure In this chapter you will be solving a simple crystal structure. In addition, you will shown how to prepare it for publication and how it was eventually published. Before getting started you will obviously have to install the recommended software packages. 2.1 Getting started Before getting started you will need to download the data from http://www.gh.wits.ac.za/manuel/solving-xtals/struct1/md1.zip. Extract the data to a directory on your computer. In my case I installed the data to D:\GH8\solving-xtal\data\struct1\ but it really does not matter. Now you will need to start and point your WinGX program to your data directory. Do this by doing the following: [1] Either double click the WinGX icon on your desktop or click the Wingx32 link on your windows start menu. It usually looks something like the Fig 2.1.. Fig 2.1 [2] Once you have done this you will see the WinGX toolbar which looks like Fig. 2.2 Fig 2.2 If you look carefully you will see that the program is currently pointing to a project called 5m_lc4_a which is in the d:\strcts\wits\carlton\5m_lc4 directory. To work on our data you will need to select a new project by clicking on File > CHANGE PROJECT > Select New Project. You will now see a menu that looks like Fig. 2.3.

Solving Crystal Structures 5 Fig 2.3 Click on Browse and work your way to the directory containing your data and double click on the INS file. In this case it's called md1_s.ins. The above window should now look like Fig 2.4. Fig 2.4

Solving Crystal Structures 6 [3] Now click OK and you should have the Model Summary information box poping up (Fig 2.5). Fig 2.5 [4] This provides you with some information about you structure. In this case the crystallographer has determined that the space group is probably P-1 and that the unit cell parameters are 6.85 Å etc. Click OK and have you will see that the project name and directory have changed those you selected (Fig 2.6). Fig 2.6 2.2 Solving the crystal structure In this document only two structure solution programs will be shown. The first involves using SHELXS-97 while an alternative is to use SIR-92 (or SIR-97 or SIR- 2002 if you have them). The SIR programs are very easy to use and often are able to solve a structure to almost completion making the structure refinement a little easier (saves you about 5-10 mins in time when you have a small but busy structure). SIR is often very useful when you have a data set that stubbornly refuses to be solved using SHELXS.

Solving Crystal Structures 7 In this manual only the use of SHELXL will be shown. Using SIR is just as easy and the results are often a little better. It is left to the reader to try to solve and refine his or her data set using SIR. To solve your structure using SHELXL clink on Solve > Shelxs-97. A window looking like Fig 2.7 should pop up. Fig 2.7 As you can see the DIRECT panel is selected which means that the structure is going to be solved by direct methods (a brief explanation for this will be given in one of the appendices). Now click OK and a new window will appear (Fig 2.8).

Solving Crystal Structures 8 Fig 2.8 Very briefly you can see that the program reports that you have an R(int) of 1.4 % and a R(sigma) of 2.5 %. These are both excellent values (see appendix AX for details). The contents of the end of this window are given in Fig 2.9. Fig 2.9

Solving Crystal Structures 9 Here you can see that the RE (a kind of R-factor) is about 19% for 12 atoms, i.e. the structure has about 12 atoms. In general you can consider a structure solution with RE less than 30% as solved though the higher the number the more difficult the refinement phase. An RE of about 19% tells us that this structure should be reasonably straight forward to refine from now on. 2.3 Refining the crystal structure At this stage you can close the SHELXS-97 output window and click on the SHELX- GRAPH icon on the WinGX toolbar (Fig 2.9). Fig 2.9 A new window looking like Fig 2.10 will appear. Fig 2.10 Moving the molecule around (simply click and hold the left mouse button on the cyan area and move the molecule around) and activating Label atoms check box gives you a picture that looks like Fig 2.11.

Solving Crystal Structures 10 Fig 2.11 As you can see the SHELXS program has already marked the probable position of the suphur atom (SIR could probably do better in this case). If you look carefully you will see that there is a five membered ring (made up of the white spots) connected through a bond to a six membered ring. As I'm sure you have noticed I have not shown you what the structure should look like. All you are given is the space group and unit cell parameters (provided by the crystallographer and mentioned earlier) as well as the basic elemental composition of the molecule which is C H N S. This is all you need to solve the structure and SHELXS has already told you where the sulphur probably is. Now activate the Label Q-peaks check box. The SHELX-GRAPH screen should now look like Fig 2.12.

Solving Crystal Structures 11 Fig 2.12 You will see that the white peaks are numerically ordered from Q1 to Q20. The important thing to realize is that this is really an ordered list that comes from the res file outputted by SHELXS. If you open the md1_s.res (open with wordpad, notepad or scintilla) you will see lines looking like the following: Q1 1 0.5960 0.7081 0.6909 11.000000 0.05 343.61 Q2 1 0.6409 0.7441 0.4527 11.000000 0.05 316.70 Q3 1 0.5069 0.7420 0.5575 11.000000 0.05 298.34 Q4 1 0.2958 0.7671 0.5428 11.000000 0.05 266.91 Q5 1 0.5085 0.7446 0.8198 11.000000 0.05 263.80 Q6 1 0.8358 0.6442 0.6896 11.000000 0.05 221.84 Q7 1 0.2200 0.7996 0.4150 11.000000 0.05 208.50 Q8 1 0.5659 0.7825 0.3266 11.000000 0.05 205.85 Q9 1 0.6738 0.6844 0.9227 11.000000 0.05 203.74 Q10 1 0.3600 0.8077 0.3014 11.000000 0.05 195.51 Q11 1 0.8881 0.5630 0.8391 11.000000 0.05 174.45 Q12 1 0.2693 0.8959 0.6229 11.000000 0.05 93.24 Q13 1 0.5305 0.7950 1.0084 11.000000 0.05 89.43 Q14 1 0.6683 0.6953 0.5759 11.000000 0.05 85.54 Q15 1 0.3025 0.8361 0.9855 11.000000 0.05 79.45 Q16 1 0.4681 0.6770 0.7555 11.000000 0.05 74.12 Q17 1 0.0000 1.0000 0.5000 10.500000 0.05 72.59 Q19 1 0.5537 0.6623 0.2472 11.000000 0.05 66.55 Q20 1 1.1178 0.3572 0.8630 11.000000 0.05 63.60 These represent residual peaks (the intensity of which is given in the last column) in the electron density map and give an indication of where the atoms of the molecule

Solving Crystal Structures 12 are. The list is ordered from the most intense peak (Q1) to the least intense peak (Q20). You can actually select how many residual peaks SHELXL will give you but more on that latter. If you look at the residual peak list you will see that peaks Q1 to Q11 are significantly more intense than the remainder and there is an obvious jump in the peak height on going from Q11 and Q12. This feature is often very useful when solving crystal structures and you should remember this effect. In this case it means that the "heavy atoms", i.e. the positions of C and N atoms, is separated from the H atoms at this point. Looking at the last SHELX-GRAPH screen (Fig 2.12) you will see that details of the Q1 peak are available to see on the side menu (look at the CURSOR atom area). This information is accessible by just holding the mouse cursor over an atom or peak you are interested in. In this case I had the mouse cursor over the Q1 peak. If you look at the Q1 in the list I gave above and compare it with this you will see that they are identical, i.e. SHELX-GRAPH is telling you in a graphical manner the information that available from the md1_s.res file directly. As I mentioned peaks Q1 to Q11 seem to be the main peaks of this structure. Together with the S atom you have the 12 atoms that were mentioned by SHELXS in its final output. Fig 2.12 shows that peaks Q1 to Q11 make up the 5 membered and 6 membered rings which are from a geometric point of view chemically sensible. At this point it is a good idea to delete the picks that do not make chemical sense. To do this simply clicking on the peaks you don't want (Fig 2.13).

Solving Crystal Structures 13 Fig 2.13 You will notice that these correspond to peaks Q12 to Q20, i.e. the weak peaks I showed in the Q list earlier. Now click on Delete > Selected atoms. This will give you a diagram that looks like Fig 2.14. Fig 2.14 You know can see what the molecule will probably look like. At this stage you have a choice or relabelling all the Q peaks to something that is chemically sensible such a C atoms. In this particular structure we know that the back bone is made up mostly of C atoms. So we can either change the labels by hand one by one or we can cheat initially. Lets rather take the second approach and ask SHELX-GRAPH to convert all the Q atoms to C atoms. To do this click on Model > Change Q >> C. You should now have something that looks like Fig 2.14.

Solving Crystal Structures 14 Fig 2.14 Though the labels are all wrong the residual peaks have all been labelled as carbons and it is now time to pass our model as we have it now through SHELXL and see how the structure refines. There are two ways to do this. The easy way is to run SHELXL through SHELX-GRAPH itself though when things go wrong people get very confused and don't know what to do. The other way is more difficult but is easier to recover from if things go wrong. For novices I prefer to teach the slightly harder way as it's hard to make mistakes this way. This is the method I will demonstrate here. First of all save the changes you have made by clicking on Save INS file. This creates a new ins file named md1_s.ins in the case of this structure. Now close SHELX- GRAPH either by clicking on the "x" on the top right corner or by clicking on File > Exit in SHELX-GRAPH. The WinGX toolbar should still be available. We now are going to pass the md1_s.ins file which contains our current structural model through SHELXL and ask it to both refine our current atom positions (the peaks we chose) and give a new list of residual peaks (or from SHELX's point of view Q peaks). To do this click on Refine > SHELXL-97. A new window will appear with information similar to that shown in Fig 2.15.

Solving Crystal Structures 15 Fig 2.15 As you can see the final Rfactor (R1) was 0.17 or 17%. The highest residual peak was 2.26 which means that the highest residual peak has an estimated height of 2.26 electrons. In this case this value does not really matter as its lies right next to the S atom. Lets now look at SHELX-GRAPH again. Click the SHELX-GRAPH icon on the WinGX toolbar. You should now have something that looks like Fig 2.16.

Solving Crystal Structures 16 Fig 2.16 As you can see almost all the residual peaks surround the S and C atoms positions. In this particular structure the reason for this is the fact that we still have isotropic atoms. What does this mean? It means that you are assuming that the atoms vibrate symmetrically around a central position leading a molecule with atoms looking those in the ORTEP 1 diagram shown in Fig 2.17. Fig 2.17 1 The meaning of an ORTEP will become clear later.

Solving Crystal Structures 17 However, the reality is that in nature very few things behave isotropically. More often than not atoms are arranged in a less random manner and atoms vibrate more in some directions than in others. The residual peaks shown in Fig 2.16 reflect this as they tell you that the spherical atom model (or isotropic) in insufficient to describe the electron density surrounding each atom. In SHELX-GRAPH take a close look at the S atom. You will see that it is surrounded by the two biggest residual peaks (Q1 and Q2). If you open the md1_s.res file using a text editor you will see that the line for the S atom looks like the following: S1 4 0.238186 0.853746 0.874298 11.00000 0.06045 In this case the last number represents the magnitude of the displacement (or size of the ellipsoid) around S1. When you have an isotropic atom, as in this case, all you need is one number as this defines the radius of the sphere around the atom (see Fig 2.17). Lets now change the description we are using for our atoms from isotropic to anisotropic as this will better describe the electron density around the molecules atoms. There are two ways to do this. If you have a complicated structure then it is often a good idea to make a few atoms anisotropic at a time. In this particular case the molecule is quite simple so its not a bad idea to make all the atoms anisotropic at the same time. To do this first uncheck the Display Q-peaks check box which will hide the Q peaks. You now have two choices. Either you select all the atoms in the molecule one by one by clicking on them or you can click on Select > All atoms. Now right click in the cyan area of SHELX-GRAPH and a new window will appear Fig 2.18.

Solving Crystal Structures 18 Fig 2.18 Now click the Set Uij's anisotropic check box followed by OK. This will set the thermal ellipsoids of all the atoms you selected to anisotropic. The meaning of this will become clear in a moment. Lets now save your changes and click on Save INS file and close SHELX-GRAPH as before. This will have saved your changes to a new version of md1_s.ins. Rerun SHELXL-97 by clicking on Refine > SHELXL-97 which will create a new window with refinement information as before. At the end of the run you will notice that making the atoms anisotropic has lowered the Rfactor (R1) to 0.088 or about 9%. An ORTEP diagram after applying anisotropic ADP's is shown in Fig 2.19. If you open the md1_s.res file now using a text editor you will see that the line for the S atom now looks like the following: S1 4 0.238089 0.853981 0.874423 11.00000 0.04365 0.08868 = 0.04904-0.00742 0.01031-0.02388 As you can see the 0.06045 value of before has been replaced by six new numbers (0.04365 0.08868 0.04904-0.00742 0.01031-0.02388). The first three numbers represent the displacement of the ellipsoid along the three orthogonal

Solving Crystal Structures 19 directions defining its shape. The last three define the orientation of the ellipsoid relative to the unit cell axes (see Fig 2.19). Fig 2.19 As you can see the ellipsoids around each atom are no longer perfectly spherical and in fact many of them look slightly elongated. Notice that the ellipsoid belonging to C001 is visually smaller than those of the rest of the molecule. There is a good reason for this. Remember that we are expecting that the structure has one or more N atoms. The way to look at ellipsoids is like this: N atoms have one more electron than C atoms. That means that assigning a C atom where a N atom should be will give you a small ellipsoid because the ellipsoid has to extend a smaller distance to account for the electron density of a C atom. On the other hand if you place a N atom where a C atom is actually present then the ellipsoid will appear much larger than those of the rest of the molecule because the ellipsoid has to be drawn out much further to account for the electron density associated with an N atom. The best way to see these effects is to wrongly assign some atoms when we have completed the refinement of this structure, i.e. change the N atom to a C and see what happens to the Rfactor and ellipsoid once you refined the structure. Also, see what happens when you change a C to an N atom. Finally change the S atom to a C atom.

Solving Crystal Structures 20 So now we strongly suspect that C001 is actually a N atom what do we do to make sure? That will become clear in a few minutes. Lets first see what residual peaks came out of the SHELXL refinement. Start by opening SHELX-GRAPH again. You should now have something that looks like Fig 2.20. Fig 2.20 Notice that the H atoms are now visible, i.e. SHELXL has found them in the difference map. If you open md1_s.res with a text editor you will see the following Q peaks: Q1 1 0.8680 0.7424 0.6674 11.00000 0.05 0.64 Q2 1 0.9055 0.5200 0.6184 11.00000 0.05 0.60 Q3 1 0.2029 0.7619 0.6217 11.00000 0.05 0.59 Q4 1 0.0788 0.8186 0.3972 11.00000 0.05 0.54 Q5 1 1.0127 0.5830 0.8634 11.00000 0.05 0.50 Q6 1 0.3036 0.8384 0.2034 11.00000 0.05 0.49 Q7 1 0.9122 0.3998 0.8498 11.00000 0.05 0.49 Q8 1 0.6682 0.8210 0.9547 11.00000 0.05 0.46 Q9 1 0.6219 0.6403 1.0049 11.00000 0.05 0.46 Q10 1 0.6585 0.7943 0.2482 11.00000 0.05 0.45 Q11 1 0.2590 0.7847 0.5174 11.00000 0.05 0.33 Q12 1 0.3693 0.7750 0.8419 11.00000 0.05 0.31 Q13 1 0.4482 0.8423 0.3135 11.00000 0.05 0.29 Q14 1 0.2911 0.7434 0.8246 11.00000 0.05 0.25 Q15 1 0.3670 0.7720 0.5587 11.00000 0.05 0.17 Q16 1 0.2845 1.0359 0.1552 11.00000 0.05 0.17 Q17 1 0.8384 0.8485 0.4381 11.00000 0.05 0.10 Q18 1 0.2529 1.0240 0.9426 11.00000 0.05 0.09 Q19 1 0.8678 0.7299 0.4022 11.00000 0.05 0.09 Q20 1 0.8531 0.8046 0.4194 11.00000 0.05 0.08

Solving Crystal Structures 21 As you can see there seems to be a jump in the height of the residual peak between Q10 and Q11. Residual peaks Q1 to Q10 seem to there all be H atoms while the rest do not seem to be important. Also notice the height of the residual peaks Q1 to Q10 are all less than 1 eå 3, i.e. less than 1 electron. This tells you that they are probably hydrogen atoms. If the peak was higher e.g. 1.5 or 2.5 eå 3 then you should consider wheter anything heavier is present instead of a H atom. Use bond length and angle measurements to make sure. What I mean by this will become clear in a few minutes. If you look at the H's around the C atoms using SHXG you will notice that atoms C006, C009 and C011 all have two potential H's attached to them, i.e. these are all probably CH 2 's. See if they are of sufficient height (account for significant electron density) by checking the Q atoms against the list above or by holding the cursor over them and reading the height of the peak one by one as before. As an alternative to opening the md1_s.res file with a text editor you can also look at the Q atom list using SHXG. Do this by clicking on Model > List Q-peaks which will bring a window up looking like Fig 2.21. Fig 2.21 Using this menu select those that are probably H atoms, i.e. Q1 to Q10 and click OK. Notice that the molecule in SHXG is much clearer now (Fig 2.22).

Solving Crystal Structures 22 Fig 2.22 Lets now insert the H atoms that we know into the structure or model. There are two ways of doing this. One way is to change the Q labels to H labels directly. Never do this unless you have a good reason 2. Instead lets tell SHELXL to geometrically calculate the positions for these H atoms. Lets do the CH 2 's first but before doing this lets check that the C-C distances associated with these atoms are typical of single bonds - about 1.45 to 1.55 Å. To check a distance in SHXG click on C009 followed by C011. At the bottom of the SHXG window you will see the distance 1.533 Å (Fig 2.23). Doing the same for C005-C009, C011-C006 and C006-C001 will confirm that these are all single bonds. Lets now add the H atoms. Click on C009, C006 and C011 then click on Model > Add Hydrogen > Methylene Group. Ignore the warning that WinGX (in this case as we know supposedly know what we are doing) and click OK. A new window will appear (Fig 2.24). 2 H atoms are weakly diffracting and as a consequence it is very difficult to determine their true position from an electron density map unless you know what you are doing. Exceptions to this occur for structures that have been done at low temperature (-100 C or lower) and generally only when you are looking at H atoms involved in H bonding. If you do this make sure that the Uij associated with H atom is reasonable - no details here because if you know what you are doing then you know what reasonable is. In addition, all H atoms of a certain type in a structure, e.g. O-H should have O-H distances within 0.05 Å of each other. Always use calculated H positions when you have a heavy atom, e.g. second or third transition metal or an iodine atom etc., in the structure.

Solving Crystal Structures 23 Fig 2.23 Fig 2.24 The default settings will do. Once you are quite familiar with SHELXL you will know if need to change these but usually the default values are correct. Click OK. The "HFIXED" atoms will now appear green. Lets now add the H atoms to the aromatic C atoms but first lets check to make sure that the distances correspond to aromatic bonds. A typical distance for an aromatic C- C bond (or one involved in conjugation) is 1.35 to 1.45 Å. Now measure C003-C004, C004-C007, C007-C010, C010-C008 and C008-C002. Notice that they are typical of conjugated bonds. Now lets add H's to these atoms. Click on C004, C007, C010 and C008 followed by Model > Add Hydrogen > Aromatic C-H. You will notice that the HFIX code is 43. This is correct so just accept the defaults again.

Solving Crystal Structures 24 Now that we have made these changes let run them through SHELXL. Click on Save INS file and close SHXG. Open the new md1_s.ins file with a text editor. You will notice that SHXG has added the following lines to the ins file: HFIX 23-1.20 C006 C009 C011 HFIX 43-1.20 C004 C007 C008 C010 This is the actual instruction that SHELXL will be given to do what you requested above. However, the addition of H atoms to a structure is temperature dependent. IN other words the C-H distance that SHELXL uses is dependent on the temperature at which the collection is done. To find out what this temperature was open the md1_s.pcf file with a text editor. The following line tells you that the data collection was done at 293 K: _diffrn_ambient_temperature 293(2) Lets now tell SHELXL what the data collection temperature was. To do this first convert the above temperature to Celsius which in this case is 20 C. Now type the following line as a new line into the ins file anywhere between the UNIT and FVAR lines: TEMP 20 Now save you changes and close the ins file. Now run this file through SHELXL - click Refine > SHELXL-97. Notice that R1 has now dropped to about 7.3 %. Close the SHELXL window and open SHXG again. Opening the res file using ORTEP-3 yields the ORTEP shown in Fig 2.25.

Solving Crystal Structures 25 Fig 2.25 Notice that ellipsoid or ADP (atomic displacement parameter) around C001 is still significantly smaller than most of the other C atoms. If you look at your SHXG window you will notice that C002 is also suspicious as no residual peak corresponding to a H atom is present. Is this perhaps also a N atom. Well lets place the N atoms one by one. Click on C001. Now right click on it. A new window will appear (Fig 2.26). Fig 2.26

Solving Crystal Structures 26 Change C001 to N1 and click OK followed by Save INS file. If we are correct R1 should drop when we run SHELXL. Run SHELXL. Note that R1 is now 6.2 % so our guess has been correct. The ORTEP diagram at this stage is shown in Fig 2.27. Fig 2.27 Notice that the size of the ADP around N1 is more like those around it. Lets see if C002 is also an N atom. If you check SHXG you will see that there is still no residual peak next to C002. Measure the C002-C003 and C001-C008 distances. For convenience you can switch off the Q-peaks (uncheck Display Q-peaks). You will see that the distances are about the distance of a C=C bond (about 1.30-134 Å). However, a C=C=C system will be linear and not bent like C008-C002-C003. It is therefore very probable that you have a C-N=C system here. Lets therefore relabel C002 to N2 and run SHELXL as we did for N1. R1 will now be about 4.4 % so this too has been a good decision. The ORTEP diagram for this molecule is shown in Fig 2.28.

Solving Crystal Structures 27 Fig 2.28 As you can see everything in Fig 2.28 looks chemically sensible so the structure is just about refined but we still have to make the structure suitable for publication. To do this we need to apply a sensible labeling scheme. Do this by following steps: 1. Delete all the H atoms but remember what types were where - select all of them and then click Delete > Selected Atoms, or click Delete > All Type... > H atoms. 2. Relabel all the C atoms by selecting C005 C009 C011 C006 C003 C004 C007 C010 C008 in that order. Right click in the cyan area. A new window will appear (Fig 2.29). In the New names text area type the replacement labels C1 C2 C3 C4 C5 C6 C7 C8 C9 and click OK (in you structure you can choose you own labeling system but make sure that it looks systematic). The new labels will now appear in SHXG. If you make a mistake in the labeling sequence just left click then right click on those atoms one by one and change their labels. 3. Now add the H atoms again, i.e. CH 2 H's on C2, C3 and C4 and aromatic H's on C6, C7, C8 and C9. Do this the way you were shown earlier. 4. Now close SHXG and refine the structure as you have seen before. Don't forget to save the new ins file before exiting SHXG.

Solving Crystal Structures 28 Fig 2.29 After refining the structure you will get a R1 of about 4.4 %, i.e. no real change. The ORTEP diagram at this stage is shown in Fig 2.30. Fig 2.30

Solving Crystal Structures 29 2.4 Preparing the structure for publication Very nice (or pretty) picture but will the structure be publishable as is? In other words are there changes that we might need to make to the structure refinement or data to make it publishable. To test this we need to run a program which is designed to check for problems with structure solutions. This program is called PLATON. However, before you use PLATON you need to create a file summarizing your current results. The format which is currently the standard is the CIF format (Crystallographic Information Format). Though you have been refining your structure you will see that no file exists with a cif extension in your current working directory. To get SHELXL to create it you will need to do the following. Open SHXG and save the current res file as a ins file (just click Save INS file). Now open md1_s.ins in a text editor and type in the following lines: OMIT -2.00 180.00 ACTA CONF HTAB BOND $H If the ins file contains an OMIT line then just replace it with the one above. Make sure that only one OMIT line exists. OMIT as defined above tells SHELXL to omit any reflections with a negative signal to noise ratio of less than minus two (yes you can get these from the data reduction programs) from our next refinement, i.e. any peak with a I/σI ratio greater than -2 will be included in our next refinement. By the way OMIT line in the beginning of a structure refinement process (e.g. at the beginning of the refinement of this structure) is usually defined as: OMIT 4 180 In other words we were telling SHELXL until now to ignore reflections that did not have a I/σI ratio higher than 4. One of the reasons this is done is that it speeds up the structure solution process as SHELXL ends up working on much less than the full data contained in the hkl file. The second number in the OMIT line simply tells SHELXL that we want all the data from a diffraction angle point of view - 180 (180 ) implying that the X-rays are being diffracted back to the X-ray source. In reality this does not happen and a more

Solving Crystal Structures 30 realistic value is anywhere from 50 (the minimum you should use as this is an ACTA or structure refinement requirement) to about 70 depending on the quality of your data. At WITS we tend to collect data so that it is 100% complete to 0.75 Å resolution which is more than good enough for standard data collections. In our case the last number is therefore anywhere between 50 (the minimum acceptable) and 56-57 (the maximum complete data range of our data collection). The next line (ACTA) tells SHELXL to generate a cif file. SHELXL will not generate a cif if you have an OMIT line other than the above, i.e., OMIT must be written as OMIT -2 or -4 or -1 even 0 might work but it cannot be greater than 0. CONF tells SHELXL that you want bond lengths and angles. HTAB tells SHELXL that you want a list of hydrogen bonds if they exist. These are listed in the lst file (as in shelxl.lst in the case of WinGX). BOND $H tells SHELXL that you want a list of bond lengths including those to H atoms (hence the $H). For more information on these commands take a look at the SHELXL manual. Once you have added all this information into the ins file save it and refine the structure again. You will see that R1 is still about 4.4 %. In addition you will notice that there are two new file in your work directory: md1_s.cif and md1_s.fcf. For now ignore the fcf file except that you should know that it contains you hkl data in CIF format and is required by the PLATON program to fully determine the quality of your data and structure refinement. Let's now do the actual validation of our current structure. We are going to do this the "hard" way by running a PLATON externally of WinGX 3. The reason that we are doing it this way is to make you realize that PLATON is actually a separate program which WinGX can run for you. To run platon double click on your PWT icon (PWT is a toolbar for PLATON; Fig 2.31) 3 To run the PLATON validation program from within WinGX click on Publish > Validate CIF > Platon Validate in the WinGX toolbar. However, you will need to install PLATON as suggested by the WinGX program just read the error message properly and follow its instructions.

Solving Crystal Structures 31 Fig 2.31 At this point a new window should appear (Fig 2.32). Fig 2.32 You now need to tell PWT which data it should work on. Click on the open folder icon (Fig 2.33) or File > Select Data File and select the newly created CIF file (md1_s.cif in this case). Fig 2.33 Now click on the tick icon (Fig 2.34) or click on Publish > CIF VALIDATE which will run the actual structure validation program. Fig 2.34 A window containing the following output (for brevity I have omitted some of the information below this) is obtained after running the PLATON validation command: #==============================================================================# # PLATON/CHECK-(220505) versus check.def version of 220305 for entry: md1_s # Data From: md1_s.cif - Data Type: CIF Bond Precision C-C = 0.0031 A # Refl Data: md1_s.fcf - Data Type: SHELXL # # Cell 6.8504(8) 7.3247(9) 9.5740(11) 87.965(2) 85.299(2) 64.705(2) # WaveLength 0.71073 Volume Reported 432.87(9) Calculated 432.87(9) # SpaceGroup from Symmetry P -1 Hall: -P 1 # Reported?? # MoietyFormula C9 H10 N2 S

Solving Crystal Structures 32 # Reported? # SumFormula C9 H10 N2 S # Reported C9 H10 N2 S # Mr = 178.26[Calc], 178.25[Rep] # Dx,gcm-3 = 1.368[Calc], 1.368[Rep] # Z = 2[Calc], 2[Rep] # Mu (mm-1) = 0.315[Calc], 0.315[Rep] # F000 = 188.0[Calc], 188.0[Rep] or F000' = 188.30[Calc] # Calculated T limits: Tmin=0.000 Tmin'=0.000 Tmax=0.000 # Reported Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2092, Th(max)= 28.32 # Obs in FCF Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2092, Th(max)= 28.32 # Calculated Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2149, Ratio = 0.97 # rho(min) = -0.27, rho(max) = 0.29 e/ang^3 # R= 0.0443( 1565), wr2= 0.1311( 2092), S = 1.008, Npar= 109 #=============================================================================== >>> The Following ALERTS were generated <<< ------------------------------------------- Format: alert-number_alert_alert-type_alert-level text 052_ALERT_1_A (Proper) Absorption Correction Method Missing..? 053_ALERT_1_A Minimum Crystal Dimension Missing (or Error)...? 054_ALERT_1_A Medium Crystal Dimension Missing (or Error)...? 055_ALERT_1_A Maximum Crystal Dimension Missing (or Error)...? 093_ALERT_1_A No su's on H-atoms, but refinement reported as. mixed 122_ALERT_1_A No _symmetry_space_group_name_h-m Given...? #=============================================================================== 029_ALERT_3_C _diffrn_measured_fraction_theta_full Low... 0.97 048_ALERT_1_C MoietyFormula Not Given...? 120_ALERT_1_C Reported SPGR? Inconsistent with Explicit P-1 125_ALERT_4_C No _symmetry_space_group_name_hall Given...? 199_ALERT_1_C Check the Reported _cell_measurement_temperature 293 K 200_ALERT_1_C Check the Reported _diffrn_ambient_temperature. 293 K 795_ALERT_4_C C-Atom in CIF Coordinate List out of Sequence.. C1 911_ALERT_3_C # Missing FCF Refl Between THmin & STh/l= 0.600 10 912_ALERT_3_C # Missing FCF Reflections Above STH/L=0.6... 48 #=============================================================================== ALERT_Level and ALERT_Type Summary ================================== 6 ALERT_Level_A = In General: Serious Problem 9 ALERT_Level_C = Check & Explain 10 ALERT_Type_1 CIF Construction/Syntax Error, Inconsistent or Missing Data. 3 ALERT_Type_3 Indicator that the Structure Quality may be Low. 2 ALERT_Type_4 Improvement, Methodology, Query or Suggestion. #=============================================================================== Explanation and Advice for the Reported ALERTS ================================================================================

Solving Crystal Structures 33 As you can see by looking at the first line I ran a 22-05-05 version of PLATON with a rule set (structure quality reference file check.def) dating back to 22-03-05. This is important as the quality indicators for crystallographic analyses become stricter with time so a newer version of either PLATON or check.def will probably yield more alerts than the ones listed above. Looking at the validation output again you will see that all sorts of structure information has been listed including R1 and GOOF. In some places you will see the words Reported (or Rep) and Calculated (or Calc). Reported indicates information that is obtained directly from your interpretation of the structure, i.e., from the CIF file. Calculated indicates the equivalent information determined from your atom coordinates, unit cell and space group information. With some exceptions (usual when a structure is disordered) the Reported and Calculated indicators should have identical information. If not you will have to fix these discrepancies between your reported information and the calculated information. The exact nature of the errors in md1 as we current have it is shown over the next few lines. There are 6 level A and 9 level B alerts as defined in this version of check.def. The format of each alert is as follows: 052_ALERT_1_A (Proper) Absorption Correction Method Missing..? The reference number (052) leads you to a paragraph of information further down the window (omitted in the above output but shown below) which gives you an indication of how to solve this particular problem. ================================================================================ ALERT_052 Type_1 CIF Construction/Syntax Error, Inconsistent or Missing Data. ================================================================================ The treatment/method of absorption(correction) should be given explicitly. This simply tells you that it very important to report ( give explicitly ) the type of absorption correction carried out for this structure. I will describe the exact nature of each of the alerts and how to solve them a little later. The next part of the alert ( ALERT_1_A ) simply tells you that this is an A alert or as Platon has it In General: Serious Problem. In this particular case it is not a serious problem but rather

Solving Crystal Structures 34 a serious omission of experimental information. The final part of the alert gives you a very short description of the problem: (Proper) Absorption Correction Method Missing. Nothing new here. Before going ahead and fixing all these alerts lets see if we can easily fix some of them. You will find the following three alerts in every structure you solve and refine: 053_ALERT_1_A Minimum Crystal Dimension Missing (or Error)...? 054_ALERT_1_A Medium Crystal Dimension Missing (or Error)...? 055_ALERT_1_A Maximum Crystal Dimension Missing (or Error)...? Obviously our cif file is missing its crystal size information. Looking at the extra information associated with 053 yields the following: ================================================================================ ALERT_053 Type_1 CIF Construction/Syntax Error, Inconsistent or Missing Data. ================================================================================ The smallest crystal dimension should be supplied in the CIF. The expected value should be a real number (i.e. not 0.35mm) If you open the md1_s.cif file in an editor and look for the following lines you will understand the problem, i.e., instead of crystal size information you have? symbols. This means that cif file is incomplete and hence triggers an error. _exptl_crystal_size_max? _exptl_crystal_size_mid? _exptl_crystal_size_min? Lets solve this easy problem. Close any validation related windows. Now open SHXG and save the current res file as an ins file (by clicking the Save INS file button). Now open md1_s.ins and md1_s.pcf (a file containing some experimental details information that should have been given to you by your crystallographer). In the pcf file you will find the following three lines as well as other experimental data including collection temperature and crystal description: _exptl_crystal_size_max 0.38 _exptl_crystal_size_mid 0.18 _exptl_crystal_size_min 0.09

Solving Crystal Structures 35 Given this information there are three ways to get rid of the crystal size alert that we are currently have. You could replace the equivalent lines in md1_s.cif with these, you could use a program called ciftab (in WinGX you would click on Publish > CIF TABLES) or you can use a SIZE card in the ins file. We will see how ciftab works in a next few pages. In the meantime let s use what I consider to be the easiest method. Insert the following line into your ins file near the top of the file but below the SFAC line but above the atom lines: SIZE 0.38 0.18 0.09 Now close the ins file (and the pcf file if you want) as well as SHXG and refine the structure again. In terms of R1 and other statistics nothing much will change however a brand new cif file will be created. Now run the new cif file through the Platon validation command as explained earlier. The output from the program should look like the following: #==============================================================================# # PLATON/CHECK-(220505) versus check.def version of 220305 for entry: md1_s # Data From: md1_s.cif - Data Type: CIF Bond Precision C-C = 0.0031 A # Refl Data: md1_s.fcf - Data Type: SHELXL # # Cell 6.8504(8) 7.3247(9) 9.5740(11) 87.965(2) 85.299(2) 64.705(2) # WaveLength 0.71073 Volume Reported 432.87(9) Calculated 432.87(9) # SpaceGroup from Symmetry P -1 Hall: -P 1 # Reported?? # MoietyFormula C9 H10 N2 S # Reported? # SumFormula C9 H10 N2 S # Reported C9 H10 N2 S # Mr = 178.26[Calc], 178.25[Rep] # Dx,gcm-3 = 1.368[Calc], 1.368[Rep] # Z = 2[Calc], 2[Rep] # Mu (mm-1) = 0.315[Calc], 0.315[Rep] # F000 = 188.0[Calc], 188.0[Rep] or F000' = 188.30[Calc] # Reported T limits: Tmin=0.890 Tmax=0.972 '?' # Calculated T limits: Tmin=0.934 Tmin'=0.887 Tmax=0.972 # Reported Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2092, Th(max)= 28.32 # Obs in FCF Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2092, Th(max)= 28.32 # Calculated Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2149, Ratio = 0.97 # rho(min) = -0.27, rho(max) = 0.28 e/ang^3 # R= 0.0444( 1565), wr2= 0.1329( 2092), S = 1.076, Npar= 109

Solving Crystal Structures 36 #=============================================================================== >>> The Following ALERTS were generated <<< ------------------------------------------- Format: alert-number_alert_alert-type_alert-level text 052_ALERT_1_A (Proper) Absorption Correction Method Missing..? 093_ALERT_1_A No su's on H-atoms, but refinement reported as. mixed 122_ALERT_1_A No _symmetry_space_group_name_h-m Given...? #=============================================================================== 029_ALERT_3_C _diffrn_measured_fraction_theta_full Low... 0.97 048_ALERT_1_C MoietyFormula Not Given...? 120_ALERT_1_C Reported SPGR? Inconsistent with Explicit P-1 125_ALERT_4_C No _symmetry_space_group_name_hall Given...? 199_ALERT_1_C Check the Reported _cell_measurement_temperature 293 K 200_ALERT_1_C Check the Reported _diffrn_ambient_temperature. 293 K 795_ALERT_4_C C-Atom in CIF Coordinate List out of Sequence.. C1 911_ALERT_3_C # Missing FCF Refl Between THmin & STh/l= 0.600 10 912_ALERT_3_C # Missing FCF Reflections Above STH/L=0.6... 48 #=============================================================================== As you can see the alerts related to the crystal dimensions are now gone. There are however two alerts that require the use of shelxl to sort them out - alerts 029 and 795. Alert 029 simply tells us that the data that was collected is 97% complete up to an angle of 28.32º (see the calculated and reported HMAX lines in the report above). Ideally the data should be 100% complete up to the chosen collection angle though this is often not possible, especially if the crystal was a bad diffractor (due to insufficient size or crystal quality or due to extensive molecular disorder within the crystal) or if the material crystallizes in a triclinic space group (really only a problem on 3 circle diffractometers like we have especially if the crystal was mounted on a unit cell axis). The minimum cut off angle is 25º below which you will have to explain why the crystal was unable to diffract normally. Alert 795 tells us that we have not sorted our atoms, i.e., we do not have a logical sequence of atoms in our cif file (and ins and res files). Lets now get rid of these two problems. Open SHXG and click on Model > Sort Atom List. This simply organizes your atom list so that the atoms are arranged in sequence, i.e. C1 C2 C3 C4 etc. You might want to look at your ins file before and after doing this just to see the change. Now click on the Save Ins Button and close SHXG. In some structures you might want to sort the atoms before adding your hydrogen atoms as these sometimes confuse the sorting algorithm especially if the

Solving Crystal Structures 37 structure contains disorder. Open the ins file in a text editor and alter the OMIT line to the following: OMIT -2 56 The 56 simply tells SHELXL that we want to limit the range of the data we use for our refinement to 28º θ (or 56º 2θ). If you were paying attention while in SHXG you would have noticed that it was quite difficult to see your structure through all the Q peaks (you might want to open SHXG to check this now but do not save anything or you will change the ins file you are busy editing). Since the structure is refined we really do not need so many extra peaks. There change the PLAN line to the following: PLAN 5 This instructs SHELXL to only give us the 5 highest residual peaks from the difference Fourier map. Now save the ins file and run SHELXL. Now run the cif file through Platon validation algorithm as before. The output from the program should look like the following: #==============================================================================# # PLATON/CHECK-(220505) versus check.def version of 220305 for entry: md1_s # Data From: md1_s.cif - Data Type: CIF Bond Precision C-C = 0.0031 A # Refl Data: md1_s.fcf - Data Type: SHELXL # # Cell 6.8504(8) 7.3247(9) 9.5740(11) 87.965(2) 85.299(2) 64.705(2) # WaveLength 0.71073 Volume Reported 432.87(9) Calculated 432.87(9) # SpaceGroup from Symmetry P -1 Hall: -P 1 # Reported?? # MoietyFormula C9 H10 N2 S # Reported? # SumFormula C9 H10 N2 S # Reported C9 H10 N2 S # Mr = 178.26[Calc], 178.25[Rep] # Dx,gcm-3 = 1.368[Calc], 1.368[Rep] # Z = 2[Calc], 2[Rep] # Mu (mm-1) = 0.315[Calc], 0.315[Rep] # F000 = 188.0[Calc], 188.0[Rep] or F000' = 188.30[Calc] # Reported T limits: Tmin=0.890 Tmax=0.972 '?' # Calculated T limits: Tmin=0.934 Tmin'=0.887 Tmax=0.972 # Reported Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2047, Th(max)= 28.00 # Obs in FCF Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2047, Th(max)= 28.00 # Calculated Hmax= 9, Kmax= 9, Lmax= 12, Nref= 2091, Ratio = 0.98 # rho(min) = -0.27, rho(max) = 0.28 e/ang^3 # R= 0.0442( 1547), wr2= 0.1322( 2047), S = 1.078, Npar= 109

Solving Crystal Structures 38 #=============================================================================== >>> The Following ALERTS were generated <<< ------------------------------------------- Format: alert-number_alert_alert-type_alert-level text 052_ALERT_1_A (Proper) Absorption Correction Method Missing..? 093_ALERT_1_A No su's on H-atoms, but refinement reported as. mixed 122_ALERT_1_A No _symmetry_space_group_name_h-m Given...? #=============================================================================== 029_ALERT_3_C _diffrn_measured_fraction_theta_full Low... 0.98 048_ALERT_1_C MoietyFormula Not Given...? 120_ALERT_1_C Reported SPGR? Inconsistent with Explicit P-1 125_ALERT_4_C No _symmetry_space_group_name_hall Given...? 199_ALERT_1_C Check the Reported _cell_measurement_temperature 293 K 200_ALERT_1_C Check the Reported _diffrn_ambient_temperature. 293 K 911_ALERT_3_C # Missing FCF Refl Between THmin & STh/l= 0.600 10 912_ALERT_3_C # Missing FCF Reflections Above STH/L=0.6... 34 #=============================================================================== As you can see alert 795 has now gone but alert 029 is still present. This shows that we have to further limit the range of the data we use for the refinement. To do this simply open SHXG (notice that there are now only 5 Q peaks) and save the res file as an ins file (click on the Save Ins Button). Now close SHXG and change the OMIT line in the ins file to the following using a text editor as before: OMIT -2 54 We are now limiting the range of the data used in the refinement to 27º θ. Now save the file and refine the structure as before. Running Platon Validation should give you the following output: #==============================================================================# # PLATON/CHECK-(220505) versus check.def version of 220305 for entry: md1_s # Data From: md1_s.cif - Data Type: CIF Bond Precision C-C = 0.0033 A # Refl Data: md1_s.fcf - Data Type: SHELXL # # Cell 6.8504(8) 7.3247(9) 9.5740(11) 87.965(2) 85.299(2) 64.705(2) # WaveLength 0.71073 Volume Reported 432.87(9) Calculated 432.87(9) # SpaceGroup from Symmetry P -1 Hall: -P 1 # Reported?? # MoietyFormula C9 H10 N2 S # Reported? # SumFormula C9 H10 N2 S # Reported C9 H10 N2 S # Mr = 178.26[Calc], 178.25[Rep] # Dx,gcm-3 = 1.368[Calc], 1.368[Rep]