Criminology Research Council grant ; (9/80)
The aim of the project was to implement a rapid and sensitive non-destructive method for the identification and comparison of small glass fragments. Two analytical techniques have been commissioned on a scanning electron microscope equipped with an energy dispersive x-ray spectrometer. These have enable accurate elemental analyses on small fragments to be obtained with little sample preparation or destruction. The initial technique was to use the conventioned method of elemental analysis in a SEM/EDS system whereby the electron beam induces x-ray fluorescence in the sample. The second technique was to generate within the SEM an x-ray beam which is then used to fluoresce the sample. The two techniques are complementary and have great significance to glass analysis. This is due to the electron beam technique. With >1000 ppm detection limits, being most sensitive for low atomic number elements typical of the major and minor elements in glasses, while the x-ray beam technique, with its >30 ppm detection limits, being most sensitive for higher atomic number elements, typical of the trace elements in glasses.
A collection has been made of 177 samples of glasses used within Australia which includes both totally produced and imported glasses. The samples are either of known origin or from forensic case studies. Quantitative elemental analyses of the glasses have been obtained using the electron induced method, together with qualitative elemental analyses using the x-ray induced method. The refractive indices of all the samples have also been determined.
In comparative analyses, quantitative electron induced major- minor elemental data have been shown effective in highlighting glasses from different classifications. This greatly enhances the predictions achieved by refractive index measurement alone. The trace obtained with the x-ray induced technique further facilitates classification. In many investigations, glass analyses are requested for fragments found at a crime scene but for which no comparison sample is available. In these cases, a glass identification or classification is required. A statistical analysis of the quantitative elemental analyses and the refractive indices of all the glasses have been undertaken using the Statistical Package for the Social , Sciences (SPSS) package, in particular using DISCRIMINANT, and then with the CLUSTAN IC package.
The use of these two packages has led to the following classifications being adopted: float, sheet, rolled (other), non-float flat (Australian, 3.6 per cent MgO), non-float flat (Australian, 3.1 per cent MgO), rolled (Japan), rolled (Belgium), lightbulb, spectacles, tableware, container, and borosilicates. Both packages have been found to be very useful in predicting the classification of a single glass fragment. The databank used as a reference contains only 134 samples yet high prediction successes were obtained. Even when the unknown fragment was of a classification not within the original selections there was a good chance that this would be highlighted by the two statistical packages. This was shown by each procedure predicting a different classification since they both operate on different selection criteria. Such problems are expected to occur until the glass data bank itself is much larger but at least the statistical procedures do tend to highlight when this occurs.
These statistical procedures have given great confidence in the approach adopted to use the major and minor elements only as a means to identify closely related samples, and classify, glasses. This is significant to other workers since it eliminates the needs for trace element analyses that may be time consuming. In the SEM/EDS system all the elements are detected simultaneously in an analysing item of 50s.