Crystallography Basel - FAQ Structure Determination

FAQ X-ray Structure Determination

Laboratory for Chemical Crystallography, University of Basel

Structure determination by single crystal X-ray diffraction experiments is a powerful tool to get three-dimensional information on chemical compounds. This text gives a overview of this method and tries to give comprehensive answers to the following questions. It is some kind of faq for this analytical method.

Many settings described here correspond to the local conditions of the X-ray lab of the University of Basel and may be different in other laboratories.

This list covers some topics of this fascinating analytical tool, but it is not complete by far. It will be extended as new questions of general interest will arise.

Markus Neuburger

Can you give a short summary of a structure determination experiment?
Why do we need X-rays?
Why do we need single crystals?
What can we "see" carrying out such a structure determination?
What equipment is needed and how does the experimental setting look like?
What is the difference between a serial diffractometer and a area detector?
How do we mount the crystal on the diffractometer?
How are images transformed to intensity data?
What can actually be measured?
If the phases cannot be measured, where do we get them from?
Why does a structure need to be refined once it is solved?
What is the final result of a structure determination?
What is "disorder" and why should we know something about?
What other special cases may turn up?
Is it always possible to solve a structure?
What software exists to work on X-ray structures?

back to top

Can you give a short summary of a structure determination experiment?

A crystal of the compound under investigation is stuck with glue on a glass fibre and mounted on the diffractometer. In our case it is equipped with a ccd area detector. After the determination of the unit cell parameters a data collection strategy is established in order to collect efficiently the required amount of data. The collected images are integrated in order to extract the reflection data. Structure solution is performed using direct methodes. After successful solution of the structure refinement is carried out until the difference between the experimental data and the structure factors derived from the molecular model is small enough to be considered satisfactory.
back to top

Why do we need X-rays?

A common and well known tool to look at small objects is the microscope. We use light which is scattered by the object we want to look at and collected again using lenses. This method gives us a magnified image of the object under examination. The limit of this kind of experiment is intrinsic to the nature of the used electromagnetic radiation. It is the wavelength of the light, we cannot look at objects smaller than the wavelenght of light which is about 10 ^-6 m. As X-rays have a wavelength of about 10 ^-10 m they are suited to look at objects in that order of magnitude.
back to top

Why do we need single crystals?

Using a normal microscope we examine one object at a time. In principle this would also be possible with X-rays. Our problem is that we cannot handle such small objects as molecules one by one. Moreover the physical events caused by the scattered X-rays on some detector would be too weak to be visible besides the noise of the electronics and other sources of errors. Crystals being ordered arrangements of molecules in all three dimensions of space provide us the possibility to overcome that problem. As all molecules of a crystal are oriented the same way the events caused by the scattering experiment get superimposed and we can measure them. One single event we can measure like this is called reflection. The typical experiment consists of 3000 to 10000 independent reflections.
back to top

What can we "see" carrying out such a structure determination?

The incident beam consists of electrons produced by the X-ray tube flying through our sample. They interact with the electrons of the atoms in our single crystal and are scattered. Therefore we "see" the electron density present around the atoms, not the atom nucleus. If we want to "see" the nucleus of the atoms, we must carry out neutron diffraction experiments which are accessible only using beamlines on synchrotrons.
back to top

What equipment is needed and how does the experimental setting look like?

The instrument used to carry out X-ray diffraction experiments is called goniometer. There are three main parts. On one end we have the equipment which produces the radiation. This may be a simple X-ray tube or similar device. Then we have the place where to put the crystal. This part is made in a way to permit to orient the crystal in any desired direction. As third part of the instrument we have the detector. We use a ccd detector permitting to collect many reflections in one go. The output of such a detector is a image showing spots where the scattered X-ray beams have hit the detector. These images are called frames.
back to top

What is the difference between a serial diffractometer and a area detector?

In the early stages of crystallography cameras equipped with ordinary films were used to take pictures of diffraction patterns produced by crystals exposed to X-rays. In later stages diffractometers were developped which were able to position the crystals in reflection position for all reflections one by one and measure the intensity for these reflections one after the other. They were called serial diffractometers and had the advantage to produce much more accurate data. The area detector of today combines the advantages of both: very accurate data, fully redundant datasets and no chemicals and films.
back to top

How do we mount the crystal on the diffractometer?

Usually the crystal is stuck with glue on a glassfibre. The choice of the glue depends on variouns factors, mainly the interaction of the crystal with the glue or the time to get dry. Sometimes crystals are put into glass capillaries. This may be needed if the substance under investigation is air sensitive. When we do low temperature work the crystal can be frozen to the glass fibre using a protective oil. Sometimes this is referred to as the "oil drop method".
back to top

How are images transformed to intensity data?

We can determine the orientation of the scattering planes of electron density. From there we can calculate the size of the main building block of the crystal. We call it unit cell. Once we know the unit cell we can calculate all reflections we should be able to observe. Special software is used to go through all the frames and to sum up all the intensity belonging to every single reflection. We use a software package called EvalCCD for this purpose.
back to top

What can actually be measured?

We can measure the intensity of the diffracted X-rays. But we do not know how the different components as parallel planes may scatter in or out of phase. This is referred to as the phase problem.
back to top

If the phases cannot be measured, where do we get them from?

The phase information tells us where the scattering planes are localized in the unit cell. Patterson found in 1934 that a modification of the equation describing the scattering process lead to the elimination of the phase term, thus to the loss of the information, where the electrons were localized. On the other hand he got distance information from which - knowing the symmetry that produced it - lead again to coordinates which could be used as a first rough structure model. This method, sometimes referred to as heavy atom method, has been almost completely forgotten as direct methodes trying to find valid estimates for the phases using statistics have become so reliable that they are widely accepted for all types of structures. In 1985, Herbert A. Hauptmann and Jerome Karle got the Nobel Prize in Chemistry fort their efforts to establish direct methods as a structure solution method, but of course the contributions by others should never be forgotten even if the space here does not permit to put all their names.
back to top

Why does a structure need to be refined once it is solved?

The first model derived from patterson vectors or from direct methods is very rough and is in most cases incomplete. We can calculate how the reflections would look like if our structural model would be present in the diffraction experiment. In the ideal case our model would produce exactly the same intensities as we have measured. In order to come as near as possible to this aim we use a iterative process to complete and optimize our model comparing the calculated structure factors (Fc) to the observed ones (Fo). In the end the difference between them should become zero. In real life this is never achieved. This difference or residual is a measure of what part of our observations cannot be explained by the model and should therefore be as low as possible, preferrably lower than 7%. It is called R-value.
back to top

What is the final result of a structure determination?

We get electron density maxima and their position in the unit cell. Period. Everything else is derived from this information. We can for instance calculate chemically sensible interatomic distances using the properties of the assigned atom types and we can draw a bond in a graphics program. But we must keep in mind that this bond does not come out of the structure determination and depends on our interpretation of the electron density map.
back to top

What is "disorder" and why should we know something about?

In theory all molecules in a crystal lattice are oriented exactly the same way. Nevertheless it is possible that slight differences are observed as there may be enough space in the lattice to permit parts of our molecule to be in different orientation at different places in the crystal. As we see the sum of all molecules together we may see electron density features not explainable with a single model. We speak of disorder in this case. Disorder can be resolved using superimposed models with partial occupancy. It is important to note that cooling down the sample may help to understand the disorder problem but will never make it disappear.
back to top

What other special cases may turn up?

When we encounter problems there may be many reasons. A possible and quite frequent reason is that we did not measure a single crystal, but a twin, thus a combination of more than one lattices. It is not always visible from the shape of the crystal if it is twinned or not. Moreover there are structures requiring more that three dimensions to describe correctly the way they are built up. They are sometimes called modulated structures.
back to top

Is it always possible to solve a structure?

The answer is simple: No. As the software available to solve structures is constantly improving the number of these structures gets less. But as we get only electron density maxima it happens also that structures are solved, but the people doing the work do not recognize the structure in the solution presented by the programs. In the meantime there is a lot of help from graphical user interfaces, but the number of structures not solved will probably never go down to zero.
back to top

What software exists to work on X-ray structures?

We use SIR92 and SIR97 by Carmelo Giacovazzo et al., University of Bari, to solve structures. We refine them using CRYSTALS, a refinement package mainly written by David Watkin and issued by the Crystallography lab of the University of Oxford.
Molecular graphics are done usually using Ortep3 forWindows, originally written by Michael N. Burnett and Carroll K. Johnson and issued by the Oak Ridge National Laboratory. Louis Farrugia's GUI to Ortep3 makes it a easy to use program even if the core code is more that 30 years old!
Another nice piece of software is Platon, a collection of crystallographic tools which also includes a graphics part. This program is mainly suited for advanced users.
Of course there exist many other software packages, as the SHELX-suite by George Sheldrick, University of Goettingen, and lots of graphics programs able to deal with coordinates from X-ray structure determinations. A list of software written in this field is available from the website of the International Union of Crystallography at www.iucr.org.

Last update 14-feb-2005

Markus Neuburger