Case Studies in Spectroscopy

Identification of Mineral Mixtures

Integrated Spectronics, an innovative Australian company, has developed one of the world’s most sophisticated infrared field-portable spectrometers, the Portable Infrared Mineral Analyser (PIMA). The PIMA is the size of a shoe box and works within seconds, making it easy for geologists in remote locations to measure the infrared reflectance spectra of individual rocks.

To help geologists interpret PIMA spectra, especially mixtures, CSIRO scientists have developed The Spectral Assistant (TSA), a sophisticated software tool that assists in identifying the constituent minerals of a rock sample.

TSA Version 4 is at the heart of another CSIRO package, The Spectral Geologist, which is marketed commercially and has been sold to mineral exploration companies throughout the world.

Using a database of about 500 samples representing natural variation in 42 pure minerals, TSA uses modern and fast multivariate statistical techniques to find the most likely pure minerals and the most likely mixtures of two minerals. Figure 1 illustrates its use.

TSA Version 5 is currently under development, following the completion of an AMIRA-funded project “Automated Mineralogical Logging of Drill Core, Chips and Powders”. It will be applicable to spectrometers other than the PIMA, such as the ASD, which also measure visible spectra.

Figure 1 shows the logarithms of 5 reflectance spectra. Four of these are from the database: two kaolinite (red samples) and two muscovite (blue samples). The green line shows a known mixture of kaolinite and muscovite. TSA correctly classifies this test sample as such a mixture using diagnostic features in the pure spectra, in this case those present at around 1400 and 2200 nanometres.

Automated Analysis of Bauxite Ore

Knowing the composition of bauxite ore helps to determine whether it can be processed profitably. CSIRO has worked with Alcoa of Australia Limted to help evaluate methods of constructing multivariate calibration models, relating Fourier transform infrared spectra of bauxite ore to the concentrations of its constituents. CSIRO scientists evaluated different baseline correction methods along with non-linear models. This helped show that a constituent concentration, combined with a partial least squares model, yielded near-optimal results over a broad range of concentrations. This method is now in use in an automated ore analysis system at Aloa’s Kwinana mining laboratory.

Spectroscopy in Agriculture

CSIRO Mathematics, Informatics and Statistics is collaborating with CSIRO Plant Industry to develop near infrared (NIR) monitoring protocols for a variety of food products. For whole-wheat grains, it allows protein content to be assessed on arrival at the silo. In preparation of wheat-flour dough, NIR can be used to track free and bonded water during mixing.

Experiments to assess the quality and potential of Prime Hard wheat require a cheap and efficient way of assessing dough properties. NIR spectroscopy is quicker and cheaper than direct laboratory measurements of quality indicators. CSIRO statisticians used a partial least squares method to develop predictions of dough properties from NIR spectra of grain samples.

There are many more industrial applications of spectroscopy. CSIRO can help you find the right spectroscopic solution to your analytical needs.