Component spectra extraction and quantitative analysis for preservative mixtures by combining terahertz spectroscopy...

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Machine Learning Applications

Background:

• Preservatives are essential in food products, often used in synergistic combinations to maximize antimicrobial efficacy.
• Accurate identification and quantification of preservative components are critical for quality control and ensuring consumer safety.
• Terahertz (THz) time-domain spectroscopy offers a non-destructive method for analyzing chemical compositions.

Purpose of the Study:

• To develop and validate a method for identifying and quantifying common preservatives (sorbic acid, potassium sorbate, sodium benzoate) in binary mixtures.
• To address challenges in analyzing multicomponent preservative systems using advanced analytical techniques.
• To leverage machine learning for spectral analysis and accurate component quantification.

Main Methods:

• Utilized terahertz (THz) time-domain spectroscopy to measure binary mixtures of sorbic acid, potassium sorbate, and sodium benzoate at various mass ratios.
• Applied singular value decomposition (SVD) to determine the number of components in the mixtures.
• Employed non-negative matrix factorization (NMF) and self-modeling mixture analysis (SMMA) for component spectra extraction.
• Developed a support vector machine for regression (SVR) model for quantitative analysis and content determination.

Main Results:

• SVD successfully identified the number of components in the mixed preservative systems.
• NMF and SMMA effectively extracted component spectra, which closely matched the spectra of pure reagents.
• The SVR model accurately predicted the content of individual components in validation mixtures, achieving a decision coefficient (R²) of 0.989.
• The combined THz spectroscopy and machine learning approach demonstrated high accuracy in analyzing preservative mixtures.

Conclusions:

• The developed approach integrating THz time-domain spectroscopy and machine learning provides a powerful strategy for analyzing complex preservative mixtures.
• This method offers a reliable and accurate means for quality monitoring and ensuring the safety of products containing preservative combinations.
• The study highlights the potential of THz spectroscopy as a fingerprint-based technique for practical applications in food safety and quality control.