Smith, D. G. W., & Leibovitz, D. P. (1984) A computer-based system for identification of minerals on the basis of composition and other properties. 27th International Geological Congress. Extended Abstract, 5:169. [doi: 10.13140/RG.2.1.2471.3442]
Abstract: Developments in the techniques, theory and practice of microbeam analysis over the last quarter century have resulted in a situation in which reliable compositional data can now be obtained extremely rapidly and conveniently. However, the interpretation of these data is not always so straightforward or rapid, particularly when the microanalyst is not a fully trained mineralogist or when data are obtained from one of the less common minerals or from one of a group of compositionally similar minerals. Furthermore, certain modern procedures for the automated modal analysis of rocks and other mineral aggregates require that a very large number of identifications be performed – preferably without human intervention. The system described in this paper has been developed in response to such situations.
The system uses a purpose-developed FORTRAN IV computer program “MINIDENT” different parts of which permit the creation of analytical data files, the processing of these files to produce an “index”, and the searching of the index to find the best matches with an unknown. Although the system is b~8Bed primarily on compositional data, several other properties, such 8S reflectance, refractive indices, VHN, density, etc., have been included and can be used with (or without) compositional data to seek a match with an unknown. Each mineral entry is cross referenced to the appropriate JCPDS file number. MINIDENT produces a list of the most likely matches and prints these together with an estimated ‘reliability factor’ for each match. The program can also be used to produce an alphabetical list of all minerals with certain properties – e.g., those. containing a particular element or element combination, those which have refractive indices within a certain range, etc. It can also be used to print a summary of the information included in the data-base for a given mineral. New data can be added by means of MINIDENT and is instantly included in the data base for use in subsequent searches. Considerable emphasis has been placed on making the interactive computer software easily used and understood.
Although MINIDENT is capable of providing an identification and/or list of the most likely possibilities very rapidly and on the basis of minimal input information, it is not intended that it should supercede the JCPDS powder diffraction index. That index will normally provide the more definitive answer, although the time and effort expended in making the identification will usually be much greater. A combination of the two systems also seems possible. The normal input of d-spacings and intensities to a program searching the JCPDS files could be augmented by the output from MINIDENT, thereby greatly reducing the search time involved.
At the time of preparation of this abstract, (September 1983) the basic data for more than 2000 minerals, minerals groups or mineral series are on file and these are being added to as time and funds permit. In future, it may prove possible to add other fields to this data base – not only with 8 view to making more positive identifications but also to allowing the retrieval of a more complete summary of available information on the mineral once an identification has been made.