Laser-Induced Breakdown Spectroscopy and Principal Component Analysis for the Classification of Spectra from...

Area of Science:

• Analytical Chemistry
• Geochemistry
• Spectroscopy

Background:

• Laser-induced breakdown spectroscopy (LIBS) is a powerful technique for elemental analysis.
• Analyzing low-concentration elements in geological samples presents challenges due to sparse signal detection.
• Traditional spectral processing methods like ensemble averaging are often infeasible for such samples.

Purpose of the Study:

• To apply principal component analysis (PCA) for classifying LIBS spectra of gold ores.
• To develop a method for identifying low-concentration gold outliers in LIBS spectra.
• To overcome the limitations of ensemble averaging when gold emission lines are sparsely detected.

Main Methods:

• LIBS spectra were acquired from gold ore samples prepared as pressed pellets.
• 5000 single-shot LIBS spectra were collected per sample.
• Principal component analysis (PCA) was used to analyze spectral data, reducing variables to three principal components.
• A spectral range around the Au(I) 267.59 nm emission line was analyzed.
• Data outliers (spectra with gold emission lines) were detected using PCA statistical parameters (sample residual and Mahalanobis distance).

Main Results:

• PCA successfully discriminated between spectra with and without gold emission lines.
• The PCA approach identified 100% of data outliers compared to a standard database.
• Positive spectra, indicating gold presence, were treated as data outliers due to discrete gold distribution.
• The method proved effective even when gold concentrations were as low as 7.7 µg/g.

Conclusions:

• PCA is a viable and effective method for classifying LIBS spectra of gold ores, particularly for identifying low-concentration outliers.
• This approach overcomes the limitations of traditional methods when dealing with sparse spectral signals.
• The study demonstrates the utility of PCA in analyzing complex geological samples with challenging elemental distributions.