Scanning LIBS mineralogical Analysis in Short, nickel ore drill core

Here we present LIBS imaging based mineralogical analysis. All the other chapters will go into the details of this process. This chapter is to try to give a quick overview of what the techniques presented on this thesis are capable of.

Drill core Sample

The sample in question is a cut drill core sample of Nickel ore from Talvivaara mine in Finland.

LIBS image of the sample

With LASOLIBS[lasolibs] we scan the rock sample. We acquire about 2 gigabytes of raw spectral data, an 8188-channel 230 x 1000 pixel LIBS image or spectral cube. Figure x shows an image slice of this data.

Preprocessing

We preprocess the data with wavelet based noise removal and moving minimum based continuum removal. This makes the data easier to deal with by increasing contrast of the interesting elemental peaks in the data.

Interesting Features Finder IFF to find the minerals

To find interesting parts of this sample we run Interesting Features Finder IFF. We find the following spectra and map what they are similar to. We keep the ones that seem worth keeping and discard the rest. Figures show the process and resulting mineral similarity for a couple of identified mineral types.

Spectral Angle Mapper SAM to make mineral map with the minerals found

We use interesting parts found in previous step and map the whole sample to these minerals using Spectral Angle Mapper SAM to obtain a complete mineralogical map of still unnamed minerals. SAM matches each measurement to the closest mineral producing the map in figure .

Fitting elemental data theoretical reference spectra with gradient descent to quantify elements. Complete geochemistry and pretty maps to give geological insight

We fit theoretical reference elemental spectra(for every realistically expected element) data to each measurement spectrum and find the correct proportions with gradient descent. We get elemental content for each measurement point and can draw elemental heatmaps:

Complete mineralogy

With some geological expertise and good references we can also name our minerals. Looking at the rock we expect at least sulphides and silicates and maybe carbonates. With good guessing and comparing the elemental content or by doing a SAM comparison to previously measured known minerals we can get a complete mineralogy for this sample.

Here we can name some of the minerals for example we have named the couple common ones here in figure.

The full process described here can be done automatically with a good reference library, but is prone to errors especially when dealing with unknown rock types or unusual mineralogy so geological insight is very much needed to interpret or check any results. These results work best as a tool for a geologist to try to understand the mineralogy.