Estimation of metal elements content in soil using x-ray fluorescence based on multilayer perceptron

Area of Science:

• Environmental Science
• Analytical Chemistry
• Geochemistry

Background:

• X-ray fluorescence (XRF) is a key technique for rapid soil heavy metal detection.
• Improving the accuracy of XRF spectral processing is crucial for reliable soil analysis.
• Existing methods require optimization for enhanced quantitative analysis of soil contaminants.

Purpose of the Study:

• To compare different spectral preprocessing and quantitative analysis methods for XRF soil analysis.
• To evaluate the impact of denoising algorithms on the prediction accuracy of heavy metals (As, Pb, Cu, Cr, Cd).
• To identify optimal methods for accurate heavy metal quantitation in soil samples.

Main Methods:

• Analyzed 80 soil samples using X-ray fluorescence (XRF) under controlled indoor conditions.
• Applied various denoising algorithms for spectral preprocessing prior to model development.
• Developed and compared prediction models, including multilayer perceptron, for five heavy metals.

Main Results:

• Spectral preprocessing significantly improved the prediction performance of XRF models for soil heavy metals.
• The multilayer perceptron model demonstrated strong analytical capabilities for As, Pb, Cu, Cr, and Cd.
• High determination coefficients (R²) were achieved, ranging from 0.857 to 0.995, indicating excellent model accuracy.

Conclusions:

• Optimized spectral preprocessing is essential for enhancing the accuracy of XRF-based heavy metal quantitation in soil.
• The multilayer perceptron model, combined with appropriate denoising, offers a robust approach for analyzing soil contaminants.
• This study supports the potential of XRF analysis with advanced spectral processing for reliable environmental monitoring.