Logo Conclusions

Course Material Index  Section Index  Previous Page 

Conclusions

This section on commercial applications shows how powder diffraction data is involved in patents and forensic type work. Often the data may be produced in a court of law. You may ask whether there are any tips for those who want (or need) to get involved in this type of work. The list below lists some of the points that you may want to consider before taking on this type of work:

  1. Read the literature about the substance first. Check whether reference is made to it in any databases, patents, etc. This includes single-crystal structural databases, as well as the PDF database. Be aware that databases can occasionally contain both errors and low-quality data (as mentioned in the previous case study).
  2. Check that the powder diffractometer is recently calibrated (e.g. by the use of NIST standards and associated certificates). Carefully record all runs in the diffractometer log book and archive the data properly. Good standards can be used to check both alignment, peak shape, and instrumental resolution function. Also, consider checking out the alignment at the end of the experimental period.
  3. Be sure to log, store and keep, or return any samples that are run. Courts need to establish sample provenance. Also part of the sample may be required for analysis by the opposing team.
  4. Avoid excessive sample preparation. If possible run the sample "as received" first, and then consider grinding it lightly afterwards. If possible run the sample in both reflection and transmission modes to check for preferred orientation. In some cases this can be anticipated by the use of electron microscopy or even optical microscopy. If in any doubt, repeat the diffraction scans as a test for reproducibility: this should be normal laboratory practice for all unfamiliar samples.
  5. In general, collect data using the best resolution possible, say with fine rather than course slits. There will always be some compromise between intensity and resolution due to time limitations. In order to detect very weak reflections, occasionally extra intensity with lower resolution may be the preferred option.
  6. Always index the powder pattern where possible. This can be an excellent proof that you have a pure single polymorph or phase (and so avoiding questions about "could it be a mixture?" or "are differences in peak intensities between samples due to different ratios of various components in this sample?")
  7. Even better than indexing, is either whole pattern fitting of the data (if the structure is unknown) or Rietveld refinement (if the structure is known). If the structure is unknown, consider solving it.
  8. When comparing laboratory-based data, it is often advisable to work with the data plotted in terms of scattering angle rather than d spacing. In the previous case, 2θ plots revealed much that tables of d spacings completely missed: in general, the human eye is a very good judge of peak position (and peak match) given a diffraction profile.
  9. Avoid smoothing the data. The plaintiff's expert powder diffraction witness in the previous case study was lambasted by the judge for this practice: such procedures can readily generate peaks given noisy data. So be honest with your data and don't hide low count rates.
  10. Finally, don't act as an expert in powder diffraction (or areas of the technique) if you are not! This statement may seem blindingly obvious: but do you really wish to appear a fool in a public court of law?


Course Material Index  Section Index  Previous Page 
© Copyright 1997-2006.  Birkbeck College, University of London. Author(s): Jeremy Karl Cockcroft