Exploiting Data Uncertainty for Improving the Performance of a Quantitative Analysis Model for Laser-Induced...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• Spectral noise significantly limits the accuracy and precision of laser-induced breakdown spectroscopy (LIBS) quantitative analysis.
• Conventional preprocessing methods like normalization and ensemble averaging fail to completely eliminate spectral noise.
• Data uncertainty arising from residual noise negatively impacts LIBS quantitative analysis.

Purpose of the Study:

• To propose and validate a novel method for improving LIBS quantitative analysis performance by utilizing data uncertainty.
• To enhance the robustness and accuracy of calibration models by optimizing them with preserved data uncertainty.
• To demonstrate the broad applicability of the proposed method in analyzing ash content in coal and biomass.

Main Methods:

• Developed a method that preserves data uncertainty in the calibration data matrix by using multiple spectra per sample.
• Optimized calibration models by leveraging this preserved data uncertainty to improve tolerance to spectral signal variations.
• Applied the optimized models to predict the ash content in coal and biomass samples.

Main Results:

• The optimized calibration model demonstrated superior accuracy and robustness compared to conventional methods.
• Achieved a lower root mean square error of prediction (RMSEP) for coal ash content (1.152%) versus the conventional model (1.718%).
• The optimized model also exhibited a reduced relative standard deviation (RSD) in repeated predictions and showed improved performance for biomass ash content analysis.

Conclusions:

• The proposed method effectively utilizes data uncertainty to enhance LIBS quantitative analysis.
• Optimized calibration models offer improved accuracy, robustness, and reliability in spectral analysis.
• The method shows extensive applicability for quantitative analysis of ash content in various materials like coal and biomass.