Automated Analysis of Sections of Geological Materials

Historically, the analysis of polished rock samples has been primarily carried out using cross-polarised optical microscopy – an often time consuming and laborious technique which relies on the user's skill to identify each phase. This may be done with a point counting sampling approach with the results extrapolated to the whole sample. While this approach can be very powerful when performed by an experienced mineralogist/petrologist, no direct measurement is made of the chemistry of the various phases or their morphology.

Large area mapping offers a significant improvement on this situation. Here, an area is defined on the sample which is broken up into a series of fields of view. These can be at high magnification meaning that features can be mapped at a high resolution with a montage produced showing the complete map of the area. Such maps will show the subtleties in compositional variation across grains and can be further processed to show the distribution of phases using AZtec AutoPhaseMap. The user can then further process the maps to name the identified phases. Here, we show an example of such maps in Fig. 1.

Fig. 1: Large area EDS layered image (top) and AutoPhaseMap (bottom) of a section of a geological material.

The characterisation of samples of this nature can be further automated and the speed can be significantly increased using a feature analysis approach. This approach is particularly suited to larger samples and routine analysis owing to the time savings that are possible. Fig. 2 shows the results of an automated characterisation of the same sample as shown in Fig. 1. Here, grains are identified using grey level thresholding of backscattered electron images. Morphology and chemistry is then recorded for each grain at rates of up to 30,000 grains per hour. Each spectrum is then immediately quantified and classified according to a user generated classification scheme – classifications can be based on either chemistry or morphology or both. The results enable a full calculation of the bulk mineralogy (see Fig. 3) and chemistry to be performed based upon data from every single grain – avoiding any issues related to subjectivity or bias. As many fields recorded at high magnification make up each large area, the montaged area can be interrogated at the original magnification – revealing both the large and small scale structure of the sample – see zoomed-in area in Fig. 2. The technique can equally be applied to solid samples, as shown here, or to particulated materials.

Fig. 2: Classified montaged AZtecFeature image. Pink detailing shows where complexity is present at a resolution higher than shown here. This will often delineate grain boundaries. Key shows the number of grains of each classified phase.

Fig. 3: Table reported directly from AZtecFeature showing the bulk mineralogy of the sample.

Conclusion

Automated feature analysis of large area section samples enables a quick and accurate characterisation to be made. The measurement of both chemistry and morphology on a grain by grain basis allows for the calculation of comprehensive summary statistics for the sample.