Students guide to SHELXTL:

We will first give an overview of the programs structure and then deal step by step with
how it is used to solve and refine a structure.
Overview of program structure
The program only requires three files of which one (name.p4p) is only used at the very
beginning. An overview of the program and its structure is given below:

Overview of SHELXTL

name.hkl, name.p4p

SHELXTL/XPREP

name.prp, name.pcf

name.ins

SHELXTL/XS

name.lst

name.res

SHELXTL/XP

edit

name.ins

SHELXTL/XL

name.lst

name.res

Thus the initial input files are the name.hkl file and the name.p4p.
The hkl file is the measured data as obtained from the X-ray diffractometer and contains
(usually) the results of several thousand measurements. Each individual measurement
would have the format:
2

3 12041.0 163.468

The first three numbers are those of the set of Miller planes from which the reflection
came. The fourth number is the measured intensity and the fifth number is the error in the
measured intensity.
The p4p file contains the cell constants (three lengths, a, b, c, and three angles, )
and other information about how the data was collected.
By using these two files several things can be done at the beginning using the program
XPREP:
1. Examine the experimentally measured data
2. Determine the space group
3. Write the initial instruction file (name.ins) for solving the structure.
Once we have the initial name.ins file we can use it to solve the structure using Direct
Methods via the program XS. [These methods are based on probabilities about the
distribution of electron density within the unit cell. Two factors on which it is based are
the fact that in a real unit cell the electron density has to be always a positive number. A
second fact is that the electron density is not evenly spread out but concentrated in certain
locations, i.e. atom positions. The net result is that in favorable circumstances (and we
will only look at favorable circumstances) we can obtain a solution which gives us atomic
positions].
The output from XS is a file called name.res (results file). All calculations give such a file
[as well as a name.lst file which details what the program actually did]. This file is
visualized with the aid of the program XPREP (which is the graphical viewer that we can
use to look at the molecules/solution).
Here we can look at the results of the calculation (either structure solution or structure
refinement) determine what atoms are present and name them. After this is done the
result is then written to the name.ins file.
This new name.ins file is then used as input to the program XL (structure refinement
program) to improve the model. The XL program produces a new name.res file. This file
is then visualized in the XP structure viewer and the model improved by including more
atoms. The improved model is then again written to the name.ins file which is then used
for further refinement using XL.
These steps are repeated until all the atoms have been found, identified, and labeled.
Detailed instructions of how to use SHELXTL
A XPREP
1. First we have to start the program. When we open we see that there is a menu bar
across the top with several options. The only ones that we will be using are Project,
XREP, XS, XL, XP, XCIF, and Edit. We always begin with Project which either starts a
new project or opens an already existing Project.
2. So we use the left most pull down menu called Project.

3. We see that we have several options so we have to make a choice:

(a) New if we are beginning a new project
(b) Open if we are coming back to a project we have already started working on
We will assume that this is a new project and go through those steps.
4. We chose Open. Now we have to browse through the directories to find the files for
our project. Each project will have its own folder. In our case the folders will be in a
folder on the Desktop.
5. We open the folder of the structure we want to work on.
6. Now in the window we see two files listed; name.hkl and name.p4p. We can click on
either to highlight it.
7. Now at the top we have to give the project a name. In principle we could call it
anything but usually we give it the same name as the file name. This project name is what
we would look for if we were coming back to re-visit the structure so giving it a name
would help us to find it better (after all if you are like me you might have been dealing
with several hundred projects).
8. After we have named the project and chosen the file the window closes and we see the
project name and project path. The project path is where all programs will go to find the
files for input into programs and where all output files will be written.
9. Now we are ready to begin. Our first task is to examine the data, determine the space
group and proposed molecular formula, and write the files which will be used to solve the
structure.
10. All mentioned in 9 is achieved through the use of the program XPREP so we open
this program from our graphic user interface.
11. When we look at this pull down menu we see we have two options. XPREP or
HUGXPREP. We could actually use either in most cases. However, we would usually
chose XPREP unless the data file was too large to be read by XPREP, in which case we
would use HUGXPREP (huge XPREP).
12. After choosing XPREP we see several things:
(i) The program has automatically read in the data from the data files, name.hkl and
name.p4p.
(ii) The program has looked at lattice type and made a suggested choice which is in
square brackets. Whenever the program makes a choice about some parameter it is
always enclosed in square brackets. This choice is based on systematically absent
reflections in the hkl class of reflections. Here the types of lattice are listed as:
P A B C I F Obv Rev All
Unless we have good reasons not to, we always accept the choice made by the program.
Since the choice made by the program is indicated in the square brackets (e.g., [P]) if we

want accept this choice all we have to do is use the Enter key. This is also true for the
other options mentioned below. If we agree with the choice made by the program all we
have to do is press the enter key.
13. After choosing the lattice type the program next wants to search for higher metric
symmetry (option [H] in the menu). This is because to find such higher metric symmetry
we need a whole data set which is usually not available until after all the data has been
collected so we can only do it at this stage.
14. This is a very important step so we should discuss it a little. Depending on crystal
system (triclinic, monoclinic, orthorhombic, etc) there are various classes of reflections
which should have the same value (minor variations are allowed, like crystal shape,
absorption problems, etc). For instance in the monoclinic system the following reflections
should be equivalent:
hkl = h-kl = -hk-l = -h-k-l
For orthorhombic
hkl = hk-l = h-kl = -hkl= h-k-l = -hk-l = -h-kl = -h-k-l
So the program averages the reflections which should be equivalent for a given system
and reports a merging R factor [R(int)]. This value should be as low as possible (ideally
zero but since these are real measurements this is never the case). In our case we will
accept any value less than 0.10. If there are several options we chose the one with the one
with the highest symmetry which has a reasonable value for R(int). Usually the program
has made this choice for us.
15. Now we are given the option of either determining the space group or inputting it [S].
We will accept this.
16. If we wanted we would choose [I] at this stage but we usually want to let the program
determine the space group so we choose [S].
17. Now the program gives us the option of changing the crystal system but usually we
go with the choice of the program.
18. Now the program gives us the option of changing the crystal lattice but usually we go
with the choice of the program.
19. The program now groups reflections into classes and looks for systematically absent
reflections. Based on these absences it suggests a possible space group. Sometimes there
is more than one choice as the systematically absent reflections only give information
about translational elements of symmetry such as glide planes and screw axes.
20. Based on the systematically absent reflections and taking into account some other
factors, such as the frequency with which the space group has been found in other known
structures, the program makes a suggestion as to what is the correct space group. Usually
this is correct and we can accept its choice.

21. After we have accepted the space group the program has highlighted option C (define
unit cell contents). This is where we are going to enter the expected formula for the
compound we are working on so we accept this.
22. In some instances the expected formula has already been entered during data
collection and this will appear at the top of the page. If this is present we will accept it.
However, in other cases the formula was not entered and we have to enter ourselves.
There are some conventions as to how we enter the formula (although the order in which
we enter the elements does not affect the correct operation of the programs). We always
start with carbon, then hydrogen, and then the other elements in order of increasing
atomic number. Single letter elements are entered as upper case (Capital letters) and
elements with symbols containing two letters have the first letter upper case and the
second letter lower case. After each element symbol we give the number of that type of
element in the expected formula. As an example:
C25 H28 N4 O4 Cl2 Cu
In our case for this compound we will use the formula:
C12 H8 N4 O2
23. After we enter the formula the program does some calculations based on the formula.
(i) Since it knows what the cell constants are and hence the volume of the unit cell, based
on the formula that was entered it will calculate the number of formula units in the unit
cell (to the nearest whole number).
(ii) It calculates the density. In the old days it was expected that there would be available
the measured density but these days very few people measure the density of their crystals.
(iii) It calculates the average atomic volume for non-hydrogen atoms. In most crystals
this number is usually very close to 18 so if we see that number it is a good sign.
(iv) It calculates the per cent (%) composition. If you have previously had you compound
analyzed these values should agree.
24. After entering the formula and checking that the number makes sense we accept
option [E] to get back to the main menu.
25. Now we see that the program is highlighting option [F] which sets up the files we
need to solve the structure. We accept this.
26. The program now creates the *.ins file needed to solve the structure and displays the
result.
27. Now we have to name the files.
(i) Usually we have the name already suggested in square brackets, e.g. [CM2]. If this is
the case we accept this.
(ii) Occasionally there is no suggestion as to name. This happens when the order of the
axes in the unit cell has to be changed to get the space group in the conventional setting.
This is quite common if the crystal system is orthorhombic.
(iii) So we either accept the name suggested by the program or enter a new name.

(iv) The program now asks if we want to overwrite the hkl file and has option [N]. We
have to say Y so that the program writes the necessary files
27. After writing the files the program has option [Q] highlighted as it thinks we are
finished. We accept this.
28. Now we need to look at the *.ins file and edit it (if necessary). To do this we use the
edit option from the menu bar and pull down to show Edit ins
29. This opens the *.ins file for editing. We see it looks like:
TITL cm09-2 in P2(1)/c
CELL 0.71073 12.1674 6.9859 12.7296 90.000 107.201 90.000
ZERR 4.00 0.0003 0.0001 0.0003 0.000 0.003 0.000
LATT 1
SYMM -X, 0.5+Y, 0.5-Z
SFAC C H N
UNIT 45 45 9
TREF
HKLF 4
END
There are several things to note here.
(i) This is a file used to solve the structure using Direct Methods.
(ii) It consists of a series of lines, each of which begins with a four letter code
(iii) The first it TITL. This just gives a title and can be anything. XPREP has given a title
which includes the structure code (file name) and space group.
(iv) The next line starts with CELL and reports first the wavelength used. If the files has
been written by XPREP it will always have the value 0.71073 which is the wavelength of
X-ray generated from a Mo anode in the X-ray tube. This is the default wavelength.
However, sometimes the data has been obtained using a Cu anode which has a different
wavelength (1.54178 ). This is the case for this structure so the line:
CELL 0.71073 12.1674 6.9859 12.7296 90.000 107.201 90.000
Should be changed to:
CELL 1.54178 12.1674 6.9859 12.7296 90.000 107.201 90.000
The rest of the line reports the cell constants.
(v) The next line starts with ZERR. This is a composite of Z (the number of formula units
in the unit cell) and ERR, the errors in the cell constants.
(vi) The next line reports the lattice type (P A B C I F etc) as a code (e.g., 1 for P, 7 for C,
etc). The value can be + or -. Here we see 1 (which is positive). This means that the unit
cell (but not necessarily the molecule) contains a center of symmetry (center of inversion)
as part of its symmetry elements.
(vii) The next line lists the unique set of symmetry elements which defines the space
group. In the case of the space group P21/c these are four such elements:
1. x, y, z
2. -x, -y, -z
3. -x, 0.5+y, 0.5-z

4. x, 0.5-y, 0.5+z
Element 1 is always present in every space group. Elements 1 and 2 indicate a center of
inversion. 3 is the result of the screw axis and glide plane. 4 is the same as 3 with the sign
of x, y, and z changed. Since LATT 1 has already indicated the space group is
centrosymmetric and all space groups contain element 1, both of these are not necessary.
Element 3 is unique and thus is specified on the SYMM card (SYMM -X, 0.5+Y, 0.5-Z).
Element 4 is just 3 with the signs changed due to the fact that for every x,y,z element
there is a -x, -y, -z element. So the program only has to list the unique element and all
others can be derived from this.
(ix) The next line indicates what elements are present (or expected as at this point we
havent solved the structure and dont know for sure what is there). Usually in this line C
is first then H followed by the rest of the atom types in order of increasing atoms number.
(x) This list the unit cell contents for each type of atom listed in the SFAC line. The
numbers come from the formula we entered in XPREP multiplied by the Z value on line
2.
(xi) The first seven line will be in every *.ins file from now on (although we might have
to edit the SFAC and UNIT lines if when we solve the structure it turns out to be different
from what we expected).
(xii) The next line just contains TREF. This is the instruction to solving program (XS)
to attempt to solve using Direct Methods. If you wanted to use an alternative method, like
the Heavy Atoms method then this would change to PATT. This could be used if the
molecule contained an atom (or atoms) which were much heavier than the rest. In most
cases you will be using TREF.
(xiii) The next line:
HKLF 4
Just tells the program what format to expect in the *.hkl file containing the measured
data.
30. There is one line we should enter after the first 7 lines. This is to report at what
temperature the data was collected. In this case it was -163 C so we add the line:
TEMP -163
after the UNIT line.
31. Now we are finished editing the *.ins file. If we made any changes we should save
and exit
32. Now we attempt to solve it using the program XS on the main menu bar. This has
no pull down menu so as soon as we use it the program commences.
33. The important points to notice from the output on the screen are:
(i) The CFOM number which the Program has used which is indicated by having *
after the value (e.g. 0.048*) .
(ii) The RE values reported
Both these are near the bottom of the output. Both the CFOM values should be as small
as possible. Values for Re less than 0.2 usually indicate that the structure has been solved.

34. When a program (like XS or XL) does any calculations it uses an *.ins file
(instruction file) and writes a *.res file (results file). In addition it also writes a *.lst file.
This file contains details about what the program actually did. It is useful to track down
problems or to understand what the program did.
35. Now we have to look at what the solution looks like. To do this we use the program
XP on the top menu bar. This reads the *.res by default.
36. When we open the program the program there are a lot of words followed by:
XP>>
This is where we enter four letter commands which tell the program what to do. If we are
unsure of what commands are possible we can type:
HELP
This will lit all possible commands. If we want information about a particular command
we type:
HELP ARAD
This would list details about how to use this command.
37. In our case usually the first command whenever we open XP is:
FMOL (or FMOL/N)
This reads in the contents of the *.res file. In this case it will read in the contents of the
*.res which contains the structure solution (if XS was successful).
What has happened when XS was used in step 32 above is the program has phased some
of the data (obtained the correct phase) using probability methods and then calculated a
map of the electron density of the unit cell. From this map of electron density it has
searched for possible locations of atoms (areas where the electron density is larger than
the surrounding areas, called peaks). It has then obtained both the peak height (electron
density) and location for each peak and written it to the *.res file as an unassigned atom
(the program does not know what kind of atom corresponds to each peak). Unassigned
peaks are given the symbol Q followed by a number corresponding to where it is in the
list of peaks. Thus the largest peak would be Q1, the next largest Q2, etc.
38. When we type FMOL it reads in the *.res file and lists the contents (in this case are
series of Q values).
39. We could look at this by using the following commands:
MPLN
PROJ (or DIAG)
(i) The MPLN instruction calculates the best mean plane through all the atoms and then
makes this plane perpendicular to the screen (usually the best view for displaying the
molecules).
(ii) PROJ displays the molecule in a way that we can manipulate it using the tools on the
right side of the screen (mouse or keyboard driven). Using this we can move the molecule
to get better views. DIAG also displays the molecule but in a fixed view but does also
allow us to view the molecule while also displaying:
XP>>

which enables us to type in commands.

40. This first view has the look of an organic molecule (we see rings) but there appear to
be Q peaks in areas where there should be no atoms. This is because the XS program has
just listed a certain number of peaks (usually more than there are atoms) and some just
correspond to noise. It would help us if we could remove those peaks which are just
noise. To do this we have to have the command line (XP>>). So if you have used DIAG
to display you can enter the necessary commands but if you used PROJ (which has the
advantage of being able to rotate the molecule) we need to exit this. This can be easily
accomplished by pressing the Esc key once of twice.
41. With the XP>> displayed we type in INFO (short for information) which lists
information about the Q peaks (or atoms if there were any). There are 12 columns of data.
What we are interested at this stage is contained in the last column under peak. This is
the peak height. On looking at the values we see that they gradually decline for the first
18 Qs (Q1 to Q18) and then there is a sudden change from 155.91 to 54.61. This is an
indication of where the real atoms end and noise begins (this drop off is usually present in
an organic compound but often not when the compound contains a heavy atom).
Removing peaks after Q18 will improve our view.
42. We can remove these peaks using the command:
KILL Q19 to Q23
43. Now we can look at the result using DIAG (pressing enter twice). This displays the
molecule while allowing us to use commands (XP>>).
44. We see that it looks like an organic molecule. We know it contains C, H, N, and O. If
we know what the structure is supposed to look like then we can make atom assignments.
We can also use peak height to help us. Obviously O contains more electrons than N, and
N more than C. So we would expect to that the O and N atoms should be the larger peaks
and thus near the top of the list.
45. We can use the PICK instruction by typing this in after XP>> and pressing enter.
46. Now we see the Q peak with the largest number (Q18) is flashing. This means we can
label it. We type in an appropriate label (C1) and press enter. Now Q17 is flashing. If we
want to number the Cs systematically then this would probably be C5 so we type this in.
Now Q16 is flashing so we type C6.
47. If we make a mistake and want to go back and change labels on any atom then this
can be done by pressing the Backspace () key. If we are not sure about a label for a
particular Q and want to go on to the next Q peak then we can press the space bar. If we
want to delete a particular Q without labeling it we press the enter key.
48. We keep going in this way until all atoms are labeled.

49. Once we label the last Q peak and press enter again we get back to the main menu.
You can also get back to main menu at any time during the labeling process by pressing
the / (forward slash) key.
50. If we are not happy with labeling we have chosen and want to rename some atoms
this can be done in two ways:
(i) PICK followed by a list of atoms to be renamed. Then as each is highlighted type in a
new label.
(ii) NAME old name new name
For each atom to be renamed.
(iii) We can always check on any atom and see what bond lengths and bond angles it
makes with other attached atoms using the command:
BANG (Bond distances and angles) and then the name of an atom, e.g.,
BANG C23 to see bond distances and angles for atoms attached to C23.
51. If we are satisfied with the result of the atom labeling then we should save this result
by using the command:
SAVE filename this filename can be any name but you should be able to remember it.
I usually use the file name that corresponds to the structure being solved. This writes a
file name.sav that only XP can read. This can be re-called at any time while in XP by
using the command:
NEXT filename
You should make a habit to write a sav file every time you use XP and before you write
out the results to the ins file.
52. Now you are ready to write the named atom list to the *.ins file for refinement
(improvement of the structure). This is done using the command:
FILE filename The filename in this case will be cm09-2 but in general will be the
name of the structure you are working on.
After carrying out this command the program asks where to get the instructions from and
has its suggestion in square brackets [cm09-2.res]. Usually this is correct so you just
press the enter key. After that you can quit the program by entering either:
QUIT
Or
EXIT
53. Now we want to look at the *.ins file again to see how it has changed. So we use the
Edit ins option and look at the file.
54. We see several things have changed:
(i) The first seven lines are unchanged (8 lines if we have a TEMP -163)
(ii) There are some new lines. The first has:
L.S. 4
L.S. means carry out least squares refinement and 4 means 4 cycles of least squares
refinement. The 4 is there for historical reasons to do with computer speed. In ancient
times it would take several hours to do 3 or 4 cycles and the result would be 90%

converged after this number. Now with modern computers 4 cycles might take less than a
second. So we want to increase this number to at least 8.
(iii) The next three instructions:
BOND
FMAP 2
PLAN 20
Should be replaced with
ACTA 135
CONF
BOND $H
55. What do these all mean?
(i) ACTA 135 ACTA comes from the crystallographic journal ACTA
CRYSTALLOGRAPHICA. This instruction is made up of two parts. ACTA means
prepare all the files needed at the end for publication (say in ACTA
CRYSTALLOGRAPHICA). It also does what the FMAP 2 and BOND instructions
would do so these become unnecessary. The 135 calculates data coverage out to a 2theta value of 135. This is needed later.
(ii) CONF CONF is short for conformation. This is an instruction to prepare a torsion
angle table for all atoms in the structure.
(iii) BOND $H This is an instruction to prepare a bond distance and bond angles table
and include values involving the Hs
56 The next line is:
FVAR
1.00000
(i) FVAR means free variable. These are values the program will refine to get better
values during the refinement process. There can be 7 free variables on a line and if more
are needed then there can be a second (or more) line starting with FVAR.
(ii) There is a value on this line 1.00000. This is the first free variable and is always
reserved for the overall scale factor (OSF). This is used to put the observed and
calculated data (Fs or structure factors) on the same scale during refinement. This is
necessary because while the calculated structure factor, Fc, is defined by what atoms are
in the molecule and thus is a fixed number, the observed structure factor, Fo, (which is
just the square root of the intensity as measured on the diffractometer i.e. Fo = I) is not a
fixed number and values can vary depending on what size crystal was used.
57. The next lines contain the atom list. There are several important things to note:
(i) atom label: the first four fields (spaces) are reserved for atom label. Thus an atom label
cannot be longer than this (C123a would not be an allowable label).
(ii) The next column contains a number which corresponds to that atom type in the SFAC
line. It is this number and not the label which determines the atom type.
(iii) The next three lines list the coordinates of each atom. These coordinates are
expressed as fractions of the corresponding cell lengths (x as a fraction of a, y as a
fraction of b, z as a fraction of c).
(iv) The next number (11.00000) is the occupancy factor for each atom. This number is
made up of two parts: 1 and 1.00000. The first 1 tells the program that the next number

(whatever it is) cannot be changed. The number 1.00000 is the actual value and means
that the atom is a full weight atom (i.e. not on a symmetry element of the space group).
(v) The last column contains the thermal parameters. This is a measure of the extent to
which each atom is vibrating as all atoms are in motion except at absolute zero
temperature. This value starts out at 0.05000 but will be refined (improved) during the
least squares refinement.
58. Now that we have looked at the *.ins file we can carry out the refinement. First close
the ins file and save if any changes have been made.
59. Now use the command on the menu line XL. When you pull down there are two
values XL and XH. The first is for normal size structures while the latter is for those
problems which are too big for XL. We choose XL.
60. The program carries out the number of cycles we requested in the L.S. instruction in e
the *.ins files used. The details of what it did are in the *.lst file produced (along with the
new *.res). In the printout you see on the screen, however, there are several interesting
features.
(i) It reports a R1 values (about 6 lines from bottom). The R value is the criterion on
which the quality of a crystal structure is based. We are trying to construct a model of the
molecule so that the calculated structure factors (Fc) agree with the observed structure
factors (Fc). So R1 reflects that difference. For perfect agreement R1 = 0 since:
R1 = [(Fo Fc)]/ Fo
However in practice any value less than 0.05 is considered a good (publishable) result.
At this stage we are not finished as we havent found all the atoms (no Hs yet) and are
not totally sure that our atom assignment type is correct for all atoms. Therefore the R
value is not yet near 0.05 but is certainly better that what you would expect if the atoms
were randomly scattered in the unit cell (R > 0.5)
(ii) Three lines from bottom it reports the highest remaining peak in the (difference)
Fourier. A difference Fourier is where the contribution of the known structure is
subtracted to find any missing atoms (or miss-assigned atoms). Missing atoms like C, N,
or O would result in peaks of height > 4 (even higher for heavier atoms). The largest
peak is only around 2 so this tells us that we have probably found all non-hydrogen
atoms.
61. If the previous step was successful, now we want to improve this result. First we look
at the menu under Edit and select copy res to ins. Then we look in the *.ins file to
check what has been done.
(i) We see that the thermal parameters (which started out all being 0.05000) all now have
their own values.
(ii) Below the END instruction we see the list of Q (unassigned) peaks (none of which are
larger than 4)
62. Now we have to improve the model. One way to do this is to allow the atoms to
vibrate anisotropically. Up to now we have only been allowing the atoms to vibrate

isotropically (equally in all directions). Consider the following diagram showing two
atoms bonded together:
X

Obviously, due to the constraints of the bond the atoms should more easily vibrate in a
direction perpendicular to the bond that in a direction parallel to the bond. Thus our
isotropic vibration is not realistic. We can improve this is we allow non-equal amplitudes
of vibration in different directions, .i.e., instead of having the vibration of the atoms
describe a sphere, have these vibrations describe an ellipse. An ellipse is characterized by
6 parameters, 3 for the amplitudes in the x, y, and z directions, and 3 to describe the
orientation of this ellipse in space.
To do this we add to the *.ins file the four letter command:
ANIS
This can be placed anywhere between the UNIT and FVAR lines.
63. Now we close and save the *.ins file and do more refinement using XL.
64. We see that the R value drops significantly and the size of the highest peak in the
Fourier also drops significantly.
65. Now we look at the structure again using XP:
XP>> FMOL
XP>> MPLN
XP>> PROJ (or DIAG)
Now we see that most of the unassigned Q peaks look like they are H atoms attach to C
atoms.
However, one disconcerting aspect is that these Q peaks have bonds drawn to more than
1 C. This is due to the covalent radius built into the C (and other) atoms and the Q peaks.
These radii are used by the program to construct bonds. When ever two atoms are closer
than the sum of their covalent radii a bond is drawn between them in the XP display.
The default covalent radius for a Q peak is the same as that for C. Since these Qs are
mainly Hs and since the covalent radius of a H atom is much less than that of a C these
Qs come within bonding distance of more than one C, hence the additional bonds.
66. How are we going to assign the Hs to the atoms. The easiest way is to let the
program do this for us:
(i) Kill at Q peaks by typing:
KILL $Q The $ is used to indicate all of the following types of atoms so $Q is all
Qs , $H is all Hs, $C is all Cs etc.
(ii) Now we can add Hs to all C atoms by typing:

HADD $C This adds H atoms to all Cs. Not only that but the program can distinguish
between sp2, sp3 or aromatic C atoms and add the appropriate number of H atoms in the
correct location.
(iii) After using this command we should look and see if everything has been done
correctly. So we use PROJ so we can rotate the molecule to look at it from all sides. If it
looks correct we can press the escape key.
(iv) Now we should save as before by using the command:
SAVE filename
(v) Now we write the result to the *.ins file by using the command:
FILE filename
67. Now we use the Edit function to look at the *.ins file.
(i) we see that Hs have been added after each C which has Hs attached
(ii) Before and after each H (or set of Hs if there is more than one H on consecutive
lines) there are AFIX followed by a number. The first AFIX tells the program what kind
of H is present. The second AFIX 0 finishes.
(iii) AFIX commands have the form:
AFIX mn
m stands for the type of H, n for how it is treated during refinement.
m = 1 This is for a single H attached to an sp3 C
m = 2 This is for a two Hs attached to an sp3 C
m = 3 This is for a three Hs attached to an sp3 C
m = 4 This is for a single Hs attached to an sp2 or aromatic C
m = 8 This is for a single H attached to and O (e.g., alcohol)
m = 9 This is for a terminal ==CH2 or ==NH2+ group.
m = 13 This is for a three Hs attached to an sp3 the CH3 group is allowed to rotate to find
the best fit to the electron density
n = 3 This means the H is riding on the C to which it is attached, i.e., when the C
moves during refinement the H also moves to preserve its geometry.
n = 7 This is used along with m = 13 to give AFIX 137. This allows the CH3 group to
rotate to find the best fit to the electron density.
We usually do not have to make any changes as the program has set this up for us.
68. Now we close the *.ins file (and save if we made any changes) and do further
refinement to improve the model.
69. Now we again use the Edit function on the menu bar to copy the *.ins file to the *.res
file. Now we are getting near the end of the refinement process and have to take into
account some new factors. The first is the weighting scheme. If we open the *.ins file we
see just above the FVAR line:
WGHT 0.1000
This is the weighting scheme used during the least squares refinement. The weighting
scheme assigns weights to each observation (data point from hkl file) based on its error
value. So far we havent had to worry about this but now we are getting to the end of the
refinement we have to use the best available weighting scheme. If we have just copied the

*res file to the *ins file this is found in the *.ins file after the END card. We copy this
whole line (usually consisting of two numbers) and past it in the *ins file replacing the
previous WGHT line. Now we close and save the *.ins file and do further refinement.
Step 69 is repeated until the suggested weighting scheme is the same as the previous one.
70. Now we have to check that the formula is correct. This is done by looking at the list
file (*.lst) in the pull-down menu in SHELXTL. The formula is incorrect is the list of
atoms in the ins file does not match the number of atoms as speificied on the UNIT card.
If the formula is incorrect (and only if it is incorrect if it correct this is not included in
the *.lst file) and the items do not match then in the list you will see something like:
Unit-cell contents from UNIT instruction and atom list resp.
C
H
N
O
SE
BR

60.00
50.00
4.00
10.00
5.00
4.00

58.00
25.00
4.00
6.00
4.00
4.00

The numbers for each atoms should match. If you have found all atoms (including Hs)
then you have to edit the *.ins file to update the UNIT line with the correct formula. At
the same time you should make sure that the ZERR line is correct for the correct number
of molecules (or formula units) in the unit cell.
71. The last thing we have to do now is have a sorted atom list contained in the *.ins file.
This is best accomplished in XP so open this program from the menu bar and type in:
>>FMOL
Now we want to sort the atom list. First see what the list of atoms looks like by typing:
>>INFO
72. Usually the best order for the atoms is to have them in the order of decreasing atom
type (as listed in the SFAC card). Thus the heaviest would be listed first and the lightest
last (H atoms will automatically be sorted when you list the atom names so we do not
have to sort them). The command to sort is:
>>SORT
This command is followed by the list of atom names to be sorted (on the same line as the
SORT).
You do not have to list all atoms in 1 line as you can use this command multiple times.
73. Once you have sorted the list you have to save it so enter:
>>SAVE filename
>>FILE filename
>>QUIT
REFINEMENT IS FINISHED!

Now what happens? We have to get the structure ready for publication. This is a three
step process:
1. Preparing tables
2. Structure validation using checkcif
3. Preparing diagrams