Analysis of Specificity and Limitations Applying the Receiver Operating Characteristic Curve and Laser-Induced...

Area of Science:

• Analytical Chemistry
• Geoscience
• Spectroscopy

Background:

• Laser-induced breakdown spectroscopy (LIBS) is a versatile technique for material analysis with minimal sample preparation and real-time capabilities.
• Accurate differentiation of high-iron ore from mineralogically similar, lower-iron ores is crucial for mining and geological applications.
• The effectiveness of LIBS in distinguishing iron ore from a diverse range of geologically relevant minerals requires thorough investigation.

Purpose of the Study:

• To evaluate the specificity of Laser-induced breakdown spectroscopy (LIBS) in differentiating iron ore from various mineralogically similar ores.
• To assess the impact of mineral matrix effects on the diagnostic performance of LIBS for iron ore identification.
• To compare the efficacy of LIBS with chemometric models like PCA+LDA and KNN for ore classification.

Main Methods:

• Laser-induced breakdown spectroscopy (LIBS) was employed to analyze iron ore samples against a panel of lower-iron content minerals.
• Spectral data were analyzed using receiver operating characteristic (ROC) curve analysis to determine specificity.
• Principal Component Analysis (PCA) combined with Linear Discriminant Analysis (LDA), and K-Nearest Neighbors (KNN) were applied for classification.

Main Results:

• LIBS demonstrated high specificity (>70%) in distinguishing iron ore from biotite, dolomite, chalcopyrite, rutile, olivine, and astrophyllite.
• Statistically insignificant results were observed when differentiating iron ore from limonite, chromite, and calcite, indicating spectral or compositional similarities.
• Classification models (PCA+LDA, KNN) confirmed the resilience of LIBS but highlighted the influence of mineral matrix on diagnostic performance.

Conclusions:

• LIBS shows significant potential for identifying iron ore from specific mineral types, offering a rapid and effective analytical method.
• The technique's performance is constrained by spectral similarities with certain ores like limonite, chromite, and calcite.
• Further research into advanced spectral processing and chemometric approaches is warranted to overcome matrix effects and enhance LIBS accuracy for complex ore mixtures.