A Virtual Electronic Nose for the Efficient Classification and Quantification of Volatile Organic Compounds

Area of Science:

• Chemical sensing
• Machine learning applications
• Analytical chemistry

Background:

• Chemical gas sensors often lack selectivity for classifying different volatile organic compounds (VOCs).
• Accurate discrimination and quantification of VOCs remain a significant challenge in various scientific fields.

Purpose of the Study:

• To develop and demonstrate a selective gas sensing technology for discriminating, classifying, and quantifying short-chain oxygenated VOCs.
• To leverage the synergy between virtual electronic noses and machine learning for enhanced gas analysis.

Main Methods:

• Utilized virtual electronic noses combined with machine learning algorithms.
• Employed a 10-fold cross-validated quadratic support vector machine for classification.
• Applied 10-fold cross-validated partial least square regression for concentration quantification.

Main Results:

• Achieved 91% validation accuracy in classifying different VOCs and concentrations.
• Demonstrated reproducible correlation between predicted and measured values.
• Quantified VOC concentrations with coefficients of determination (R²) up to 0.99.

Conclusions:

• The proposed methodology offers an effective alternative to overcome gas sensor selectivity limitations.
• This approach enables accurate measurement of gases in the parts-per-billion range.
• The technique has broad potential applications in science and engineering requiring precise gas analysis.