Development and Preliminary Analytical Evaluation of a Portable Electronic-Nose Method for Discriminating Cassia grandis-Enriched Rice Beverages Using Chemometrics and HS-SPME-GC-MS

This study presents the development and preliminary analytical evaluation of a portable electronic nose (E-nose) for differentiating rice-based beverages containing 0, 5, 10, and 20% (w/v) Cassia grandis pulp. The system was evaluated through optimization of the operating conditions, assessment of short-term repeatability and robustness, exploratory chemometric analysis, and complementary volatile characterization by headspace solid-phase microextraction coupled with gas chromatography–mass spectrometry (HS-SPME-GC-MS). The optimized conditions comprised a sample volume of 20 mL, a headspace equilibration time of 10 min, an adsorption period of 60 s, and a desorption period of 30 s. Under these conditions, the sensor responses exhibited relative standard deviations (RSDs) ranging from 1.88 to 3.67% during ten consecutive measurements. Robustness experiments involving deliberate variations in the operating parameters yielded RSDs between 0.87 and 2.48%, although reducing the desorption time to 20 s decreased the mean sensor response by 5.29%. Three independent beverage preparations were produced for each formulation, and ten E-nose measurements were obtained per formulation, resulting in 40 instrumental measurements. Principal component analysis (PCA) accounted for 96.18% of the total variance in the first two components (PC1: 87.04%; PC2: 9.14%) and revealed differences in sensor-response patterns among formulations. The control (A0) and the formulation containing 20% C. grandis pulp (A20) occupied distinct regions of the PCA space, whereas A5 and A10 exhibited greater similarity. Complementary HS-SPME-GC-MS analysis revealed formulation-dependent differences in chromatographic peak areas, particularly for terpene-related compounds. For example, the linalool peak area increased from 18.34 × 106 in A0 to 535.50 × 106 arbitrary units in A20. Nevertheless, preparation-grouped cross-validation and independent predictive validation could not be established from the available experimental records.