Now that the computer-controlled diffractometer is standard laboratory equipment, the use of the digitized diffraction trace is the basis for most techniques of analysis. In addition to the geometric information that is present in the peak locations and the structure information in the intensities, there is also considerable information on the physical state of the crystal in the diffraction profile shapes. Modern studies are concentrating on this profile information, and many computer programs have been written to carry out the calculations. The diffraction traces of these well-characterized phases should be preserved in databases in a way similar to the preservation of the d-spacing-intensity data in the Powder Diffraction File.
Storing digitized diffraction information, conceptually, is a straightforward process. Initially, a set of file parameters and associated required information need to be established and programs prepared to convert experimental files into this format. Fortunately, except for setting the angular step, and intensity units, the CIF/STAR format is already an accepted structure for describing the data. There is one complication. The full diffraction trace is a convolution of the sample contribution with the effects of the source and instrument. It is really the sample component of the diffraction trace which should be preserved. Although, there is no routine answer at present to deconvoluting out this desired information, an alternative technique has been established to use the recording of pattern from a standard sample on the same instrument to establish the source/instrument contribution for later study.
A collection of patterns from a series of clay minerals has been prepared as a test case for a digitized, full-trace database. Because of the small crystallite size and 2-dimensional nature of the clay minerals, the diffraction patterns are generally broad and diffuse, so the source/instrument component is negligible. Patterns for 20 different clay minerals representing five major clay families have been recorded for both the random and oriented specimen, and the oriented specimen also has been treated with ethylene glycol or heated to elevated temperatures where appropriate for the full characterization of the mineral. The data were recorded at steps of 0.022 over the range usually employed in clay mineral analysis. In addition to the clay minerals, common accessory minerals and two instrument standards were also recorded to allow the database to be used for the characterization of natural samples and the future analysis of the profiles.
In addition to the programs for processing the database files, there is a program for identification of the components of a mixture based on the database patterns. MATCHDB compares each database pattern with an experimental pattern to determine the probability of its presence in the experimental trace and establishes a figure of merit on the probability of a fit. MATCHDB also allows further broadening effects to be modeled to improve the match and give some further information on the physical state of the crystallites. Once the identification is acceptable, GMQUANT can give the best-fit pattern weighting factors which can lead to quantification if the intensity scaling factors are known.