Method for identification of unknown substances

DETAILED DESCRIPTION OF THE INVENTION The invention is particularly useful in the field of petrography.
In a preferred embodiment, the invention consists of an automated point count system for identification of the content of various minerals and the porosity present in a rock sample of an unknown composition.
The method is a chemically based system that uses the electron-beam of a SEM to excite x-ray spectra from sample points on the surface of a substance analyzed.
In this embodiment, a reference set of known minerals arc analyzed to create a training database of normalized x-ray spectra values from which a fuzzy classification system is derived.
This preferred embodiment also includes one non-mineral substance in the reference set of known "minerals.
" This non-mineral substance is used to fill and identify pore space.
Then x-ray spectra from analysis points on rock samples of unknown composition are processed by the fuzzy classification system to identify each sample point as one of the reference set of known minerals or as unclassifiable.
Repetition of the process and compilation of the analysis results on numerous analysis points results in an estimate of the overall mineral composition and porosity of the sample.
FIG.
1 is a flowchart illustrating the primary steps of the inventive method for determining the composition of an unknown substance.
First, at step 100, a training database of x-ray spectra from at least one sample of each of a plurality of known substances is created.
Next, at step 200, a fuzzy classification system is derived from the training database.
Preferably, the fuzzy classification system comprises a plurality of substance membership functions wherein each substance is characterized by its content of each of a plurality of pre-selected elements, the content being defined by an element membership function.
Next, at step 300, x-ray spectra data from on or more analysis points on a sample of the unknown substance are collected.
Next, at step 400, a candidate substance is selected for each analysis point from the plurality of substance membership functions derived at step 200