Abstract
One of the strengths of laser-induced breakdown spectroscopy (LIBS) is that a large amount of data can be measured relatively easily in a short time, which makes LIBS interesting in many areas, from geomaterial analysis with portable handheld instruments to applications for the exploration of planetary surfaces. Statistical methods, therefore, play an important role in analyzing the data to detect not only individual compositions but also trends and correlations. In this study, we apply two approaches to explore the LIBS data of geomaterials measured with a handheld device at different locations on the Aeolian island of Vulcano, Italy. First, we use the established method, principal component analysis (PCA), and second we adopt the principle of the interesting features finder (IFF), which was recently proposed for the analysis of LIBS imaging data. With this method it is possible to identify spectra that contain emission lines of minor and trace elements that often remain undetected with variance-based methods, such as PCA. We could not detect any spectra with IFF that were not detected with PCA when applying both methods to our LIBS field data. The reason for this may be the nature of our field data, which are subject to more experimental changes than data measured in laboratory settings, such as LIBS imaging data, for which the IFF was introduced first. In conclusion, however, we found that the two approaches complement each other well, making the exploration of the data more intuitive, straightforward, and efficient.