Structural Elucidation and Predictive Model Generation of Olive Oil Classification using a GC/Q-TOF MS and Multivariate Analysis

Importance of the Topic

Extra virgin olive oil (EVOO) commands a premium price based on sensory quality, yet routine tasting panels can be subjective and resource intensive. Analytical methods that reliably predict sensory outcomes offer a rapid, reproducible alternative for quality assurance and fraud prevention in the expanding global olive oil market.

Objectives and Study Overview

This work aimed to develop a predictive classification model that distinguishes EVOO samples likely to pass or fail the official sensory panel test. By combining untargeted gas chromatography/quadrupole time-of-flight mass spectrometry (GC/Q-TOF MS) with multivariate data analysis, the study evaluated whether the presence and abundance of specific chemical markers could forecast sensory grades.

Methodology and Instrumentation

Sample Preparation and Data Workflow:

• Ten olive oil samples with known sensory outcomes were diluted 1:10 in cyclohexane and injected via cold splitless mode.
• GC separation employed a 30 m DB-5 MS column with a temperature gradient from 45 °C to 320 °C.
• Mass spectra were acquired in electron impact (EI) and positive chemical ionization (PCI) modes over m/z 40–800 at 5 Hz.
• Mass Profiler Professional guided filtering reduced 442 detected components to statistically relevant features via abundance filters, retention alignment, and a volcano plot (fold change ≥ 4, p < 0.05).
• Partial least squares discriminant analysis (PLS-DA) built a classification model based on five marker compounds.

Used Instrumentation

• Agilent 7890 GC coupled to 7200 Q-TOF MS
• Multi‐mode injector with 1 µL cold splitless injection
• Helium carrier gas at 1.3 mL/min
• Mass Profiler Professional software for data filtering and model building
• Molecular Structure Correlator for substructure searches using ChemSpider

Main Results and Discussion

From the untargeted profiling, 442 unique compounds were detected, of which five showed significant accumulation in oils failing the sensory test. The PLS-DA model using these markers achieved perfect classification of both training and test sets (100% accuracy). Identified marker compounds included n-hexadecanoic acid, ethyl octadecanoate, squalene, α-cubebene and one other unsaturated hydrocarbon. Complementary PCI data provided confirmatory elemental formulas, and EI spectra facilitated structure assignment via library comparisons.

Benefits and Practical Application

• Objective, chemistry-based grading reduces dependency on sensory panels.
• Rapid screening of EVOO authenticity and quality in production or import control.
• Flexibility to incorporate new markers and expand to larger sample cohorts.

Future Trends and Potential Applications

Scaling the model with broader sample sets will improve robustness and generalizability. Integration of high‐resolution substructure correlation and expanded spectral databases can refine marker identification. Automated QC workflows and inline GC/Q-TOF instruments may enable real-time quality monitoring in processing lines.

Conclusion

This study demonstrates that accurate‐mass GC/Q-TOF MS combined with multivariate analysis can predict EVOO sensory classification with high confidence. Although based on a limited sample set, the approach paves the way for a standardized, rapid chemical assay for olive oil quality control.