A Spectral Detection Method Based on Integrated and Partition Modeling for Trace Copper in High-Concentration Zinc...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Chemometrics

Background:

• High zinc concentrations in smelting solutions mask trace copper spectral signals.
• Spectral overlap between zinc and copper complicates accurate trace copper detection.

Purpose of the Study:

• To develop a robust spectrophotometric method for trace copper detection in high-concentration zinc solutions.
• To address spectral signal masking and overlap issues in zinc smelting analysis.

Main Methods:

• Continuous wavelet transform for derivative spectra preprocessing to enhance copper peaks.
• Interval partition modeling to select optimal characteristic wavelengths for copper.
• Partial least squares integrated modeling with Adaboost algorithm for concentration detection.

Main Results:

• The proposed method effectively reduced wavelength screening complexity and ensured stable detection performance.
• Achieved a predicted root mean square error of 0.0307 for copper.
• Obtained a correlation coefficient of 0.9978 and an average relative prediction error of 3.14%.

Conclusions:

• The integrated and partition modeling spectrophotometric method accurately detects trace copper in high-concentration zinc liquid.
• This approach offers improved sensitivity and linearity for copper ion analysis.
• The method provides a stable and effective solution for challenging analytical matrices in industrial settings.