The quality of the diffraction data and, therefore, the precision of the atomic coordinates and geometric parameters is directly dependent upon the crystal quality. Sub-standard crystals from a quick recrystallisation might be good enough to determine the basic structuere, but the quality of the results could be very mediocre.
A good rule of thumb: Poor crystals ==> poor, unpublishable results and low precision.
Therefore, it is in your best interests to try and obtain crystals of the best possible quality. Growing good crystals is an art and these days it is probably the most difficult part of the entire structure determination process. Ask yourself: have you really tried very hard or conscientiously to grow good crystals; have you have applied the most appropriate technique; were the crystallisation conditions optimal? The extra effort of trying another recrystallisation is often well worthwhile.
It is very important never to hurry the process. The best crystals rarely grow overnight. It might require days or even weeks to grow a good crystal. Frustrated chemists, who have been unable to grow a suitable crystal and have left their solution forgotten at the back of their bench, only to rediscover it months later during a clean-up, often find the most beautiful crystals waiting for them.
Read these excellent documents on growing diffraction quality single crystals first.
Introduction into vapour diffusion and layering techniques
A description of crystallisation methods collated by Paul Boyle, Western University, Ontario, Canada.
By Sandy Blake, University of Nottingham. Excellent ideas accompanied by useful diagrams.
By Allen Hunter, Youngstown State University, Ohio. A discussion of similar methods to those given in Paul Boyle's treatise, but it is written in a different style and also includes some additional useful tips, particularly on how to handle those difficult and stubborn cases.
The key with each of the crystallisation methods is to allow the crystals to grow slowly. The vapour diffusion or solvent evaporation methods proceed too quickly if the flask containing the solution is left uncovered. Covering the flask with Parafilm and poking several small holes in the film with a needle is an excellent means of slowing down the vapour diffusion.
If you are crystallising by cooling a solution, the flask should be cooled slowly. If the saturation point is reached at -2°C, it is no good plunging the flask directly into the -20°C room. The cooling rate can be slowed by placing the flask in a large bath or a dewar flask, which will take longer to cool, although in some cases this still might not permit slow enough cooling. It is best to try and adjust the solution concentration so that it reaches the saturation point just a few degrees above the lowest temperature that the flask is going to reach.
Sometimes it is worthwhile to try crystallising from different solvents. Solvent polarity is just one property that can influence crystal growth. Very thin plates obtained constantly form one solvent may grow as beautiful blocks from another solvent.
Highly soluble compounds can be difficult to crystallise by evaporation, because very often nearly all of the solvent has evaporated before crystallisation begins and then the tendency is for all the material to precipitate very rapidly. In such cases it is better to try and find a solvent in which the substance is less soluble, or try the solvent diffusion process.
Crystals should be examined closely under a microscope, preferably one that has polarising capabilities.
For X-ray diffraction we must have a single crystal. The crystals should be transparent and appear to contain no flaws when viewed under the microscope. Crystals that are cloudy, have cracks, appear to have other crystals buried inside or intergrown crystals protruding from the side should be rejected. Crystals that look like a bird's feather, a fern leaf, a dandelion seed or a star are not single crystals and are totally unsuitable. Sometimes very thin plate-like crystals or needle-like fibres stack themselves into what look like single crystals and are not immediately obvious to the inexperienced eye. Crystals that appear to have many parallel lines running along the length of the crystal may be suffering from this effect and should not be used.
The crystals should have regular smooth faces which reflect light when viewed at appropriate angles. Jagged or lumpy edges are often an indication that there are several crystals joined together in a conglomerate. Crystals that look powdery on the surface are also probably unsuitable, although, if the crystal is still translucent, it may be possible to attempt to use the crystal if a better specimen is unavailable.
A good crystal rotated under a polarising microscope should become uniformly dark every 90° of rotation as the polarised light is extinguished. The presence of misaligned fragments will be obvious during this examination, since they will not darken at the same angles as the main part of the crystal and will be highlighted. Crystals in which the dark region appears to travel across the crystal, or which show streaks of colour or blotches of darkness should be treated with suspicion. Crystals that do not darken at any point during rotation under the polarising microscope should also be treated with suspicion. Usually this is a bad sign, although if the compound crystallises in a cubic space group, non-extinction of the polarised light will also be observed. However, cubic space groups occur quite rarely for complex molecules.
Ideally the best crystal is a sphere. This is to minimise absorption effects. The next best choice is a cube or a prismatic crystal with each dimension roughly equivalent.
For optimum results, it is advisable that the entire crystal be totally surrounded by the X-ray beam at all times. With the diffractometers in the Chemistry Department, this means that the maximum dimension of the crystal is about 0.25 mm. For organic compounds, it is best to try and grow the largest possible crystal within these limitations. Plates or needles of thickness down to about 0.01 mm may be satisfactory, provided the remaining dimension(s) are much bigger. Larger crystals can usually be cut, but some crystals cannot be cut without shattering. Absorption effects caused by the heavy elements also make it desirable that crystals of organometallic compounds containing these elements are a little smaller.