Commercial tea tree oil (TTO) has to conform chemically to the international standard ISO4730: 2017, which defines acceptable ranges for the 15 main terpenoids therein. Lack of in-house capacity for the technology typically required for terpenoid quantification (GC-FID) necessitates that Australian TTO producers outsource the analyses to commercial service providers. This is associated with high cost and slow turnaround times, which have been recognized as limiting factors for TTO process optimization in the areas of quality assurance, distillation and blending. This paper describes the testing of two custom portable Raman devices with different lasers (785 and 1064 nm) for the relative quantification of the 15 main TTO components using model-based predictions. Initial testing showed that the 1064-nm device provided superior data for TTO predictions. Machine-learning algorithms were trained with spectral data from the 1064-nm device and corresponding GC-FID data from 214 TTO samples. Hold-over validation (HV) correlations between actual and predicted values using up to 53 unseen samples showed r2 at or above 0.92 for 10 and at or above 0.95 for 6 of the 15 specified terpenoids. The highly abundant key TTO quality compound terpinen-4-ol showed an HV r2 of 0.96 with a root mean square error (RMSE) of 0.37%, whereas the lowly abundant key compound 1,8-cineole showed an HV r2 of 1.00 with a RMSE of 0.13%. These results indicated a strong potential for Raman spectrometry to provide real-time product quality data at TTO distillation sites enabling direct feedback control and process optimisations.
Gloerfelt‐Tarp, F., et al. “Predicting Tea Tree Oil Distillate Composition using Portable Spectrometric Technology.” Journal of Raman Spectroscopy