A method for gaining a deeper insight into the aroma profile of olive oil

Olive oil is a valuable and ancient edible oil whose production has expanded globally, with the EU (notably Spain, Italy, Greece) as the largest producer. Its composition is predominantly glycerides (>98%), with a minor fraction containing volatile compounds and other constituents (free fatty acids, phenols, tocopherols, pigments, sterols, waxes, hydrocarbons). Volatile compounds strongly influence oil quality and consumer preference, and are affected by cultivar, ripening, environment, processing (e.g., milling and malaxation), and storage. Enzymatic pathways (e.g., lipoxygenase) generate desirable olive-like notes, while auto- and photo-oxidation yield off-odors. An accurate understanding of the native volatile profile is important for quality control and sustainability, but is hindered by the strong matrix effect of triacylglycerols in oils. The authors previously developed oiling-out assisted liquid-liquid extraction (OA-LLE), which efficiently isolates volatile compounds from oils by exploiting the oiling-out effect via two small-scale LLE steps without heating, mitigating matrix effects. OA-LLE extracted 44 and 54 aroma compounds from 5 g of coconut oil and dark chocolate, respectively. However, solvent-based extractions can co-extract non-volatiles (pigments, phenolics). Solvent-assisted flavor evaporation (SAFE) removes non-volatiles under mild conditions but suffers reduced performance in oils due to matrix effects during distillation. The authors hypothesize that combining OA-LLE with SAFE (OA-LLE + SAFE) would isolate volatiles from EVOO effectively: the dichloromethane layer from OA-LLE is subjected to SAFE to remove non-volatiles and restore distillation performance. The study aims to demonstrate OA-LLE + SAFE on EVOO, compare it with SAFE and HS-SPME, and further apply it to different EVOOs to deepen insight into olive oil aroma profiles.

Cited literature indicates that olive oil aroma depends on cultivar, ripening, environment, processing, and storage, with key contributions from LOX pathway-derived C5 and C6 aliphatic compounds for desirable notes and oxidation products for defects. Prior methods include OA-LLE (previously shown to overcome oil matrix effects in edible oils and fat-rich foods), SAFE (widely used to isolate volatiles and remove non-volatiles from oils but biased toward low-boiling volatiles due to matrix effects), and HS-SPME (solvent-free headspace method effective for low-boiling volatiles). Studies have used GC-O, GC×GC-TOFMS, and multivariate analyses to relate volatile fingerprints to sensory attributes, ripening indicators, and geographical origin. Pigments and phenolics are notable non-volatiles in olive oil and can be co-extracted by solvent methods, emphasizing the need for strategies that both overcome matrix effects and remove non-volatiles.

Experimental design: EVOO (cv. Hojiblanca, Spain) was used to compare OA-LLE, SAFE, HS-SPME, OA-LLE + SAFE (each 5.0 g EVOO; triplicate). Additional applications combined three OA-LLE extracts (total 15.0 g) prior to SAFE (OA-LLE × 3 + SAFE; single replicate) and applied to cv. Mission and cv. Lucca (Japan). Aroma extract volumes were concentrated to 200 µL for GC-MS (except HS-SPME). Model study: To enable phase separation in OA-LLE's second LLE, methanol solutions (20–100%) were extracted twice with dichloromethane; increased water in methanol reduced methanol carryover into dichloromethane, facilitating phase separation and collection of volatiles in the dichloromethane layer. OA-LLE: Followed prior protocol. EVOO was subjected to sequential extractions yielding a methanol layer (diluted to 30% with water) and dichloromethane extraction to isolate volatiles into the dichloromethane layer. Removed hexane and 30% methanol layers were odorless; the concentrated dichloromethane layer contained most volatiles but co-extracted pigments, appearing dark green. SAFE: Following Peres et al. with minor modifications, 5.0 g EVOO was mixed with 10 mL dichloromethane and distilled in a SAFE apparatus to isolate volatiles under mild conditions while removing non-volatiles. HS-SPME: Based on Vichi et al. with minor modifications, 5.0 g EVOO plus 1 µL cyclohexanol (IS) was placed in a 20 mL vial and extracted with SPME fiber; triplicates were analyzed. OA-LLE + SAFE: The dichloromethane layer from OA-LLE was charged into SAFE. The distillate was dried with 10.0 g anhydrous sodium sulfate (stored overnight at −20 °C) and concentrated to 200 µL using a Hempel column (25 × 1.0 cm, twisted glass plate) under atmospheric pressure at ~43 °C. OA-LLE × 3 + SAFE: OA-LLE performed three times on 5.0 g portions (15.0 g total); dichloromethane layers were combined, volume reduced to ~70 mL via Hempel column, then subjected to SAFE; distillate concentrated to 200 µL. GC-MS: Primary analyses used a DB-WAX Ultra Inert column; 1 µL injections, splitless; oven 30 °C for 2 min, ramp 3 °C/min to 230 °C, hold 5 min. A DB-1ms column was additionally used to resolve overlapping peaks of (E)-2-hexen-1-ol and cyclohexanol; ratios from DB-1ms were applied to deconvolute overlaps on DB-WAX. Identification used retention indices, mass spectra, and authentic standards where available. Quantification used internal standards: authentic standards or surrogate/internal standards (e.g., toluene, 2-hexanone, hexyl acetate, 2,4-decadienal, hexanoic acid, 2-hexanol, trans-2-hexenal, trans-2-heptenal, ethyl decanoate, 2-phenoxyethanol, 1-hexadecanol). Statistical analysis: PCA on peak areas (correlation matrix) to compare extraction characteristics; correlations of loadings with published boiling points (from PubChem/ChemSpider).

- OA-LLE isolated 63 aroma compounds from 5.0 g EVOO, spanning acids (10), alcohols (15), aldehydes (15), esters (10), furan (1), hydrocarbons (6), ketones (4), lactones (2). OA-LLE extracts were dark green due to co-extracted non-volatiles (pigments). Removed hexane and 30% methanol layers were odorless; OA-LLE extracts had strong olive oil-like aroma. - SAFE and HS-SPME isolated 20 and 23 compounds, respectively; extracts from SAFE were colorless and transparent. - OA-LLE + SAFE (5.0 g EVOO) produced colorless extracts and isolated 41 aroma compounds (5 acids, 16 alcohols, 5 aldehydes, 7 esters, 5 hydrocarbons, 3 ketones). C5 and C6 aliphatic markers (e.g., 3-penten-2-ol, (E)-2-hexenal) were present. - Quantitative yields (mean ± SD, n=3): OA-LLE + SAFE 43.3 ± 1.7 µg/200 µL extract vs SAFE 30.4 ± 6.6 µg/200 µL. - PCA on peak areas (OA-LLE + SAFE, SAFE, HS-SPME) explained 95.1% variance (PC1 80.5%, PC2 14.6%). OA-LLE + SAFE scored positive on PC1, while SAFE and HS-SPME scored negative; HS-SPME scored positive on PC2 and SAFE negative on PC2. Loadings correlated with boiling point (PC1 r=0.4157, p<0.05; PC2 r=−0.3494, p<0.05), indicating OA-LLE + SAFE captures both low- and relatively high-boiling compounds; HS-SPME favors lower boiling compounds. - OA-LLE × 3 + SAFE improved peak abundance versus single OA-LLE + SAFE. Numbers and totals (per 200 µL extract): Hojiblanca 59 compounds, 145.1 µg; Mission 45 compounds, 104.1 µg; Lucca 39 compounds, 231.9 µg. Compounds with RI > 2000 were dramatically increased, highlighting recovery of semi-volatiles/high-affinity species. - Specific quantified examples (Table 2, µg/200 µL): (E)-2-hexenal 12.7 (OA-LLE + SAFE), 12.5 (SAFE), — ; (Z)-3-hexenyl acetate 4.0 (OA-LLE + SAFE), 4.2 (SAFE); 1-hexanol 2.0 (OA-LLE + SAFE), 1.4 (SAFE); (E)-2-hexenoic acid 2.0 (OA-LLE + SAFE), 0.5 (SAFE); α-farnesene 3.0 (OA-LLE + SAFE), 0.8 (SAFE).

Combining OA-LLE with SAFE overcame the strong matrix effects of triacylglycerols that limit conventional methods. OA-LLE efficiently transfers most EVOO volatiles into a dichloromethane layer while excluding bulk oils; subsequent SAFE removes non-volatiles (pigments, phenolics) and restores distillation performance, yielding colorless extracts and broader volatile coverage. Compared with SAFE and HS-SPME, OA-LLE + SAFE doubled the number of identified volatiles and captured higher-boiling and semi-volatile compounds (including acids and esters with RI > 2000), while still recovering key C5–C6 LOX-derived markers. PCA confirmed method-specific selectivity: SAFE and HS-SPME favored low-boiling volatiles; HS-SPME benefited from a solvent-free chromatogram. OA-LLE × 3 + SAFE further concentrated volatiles without co-concentrating non-volatiles, enabling detection of trace/semi-volatile compounds with higher affinity to triacylglycerols. This approach is suitable for rare or precious samples due to small sample requirements and enhances the reliability of GC-MS by minimizing non-volatile contamination of inserts and columns. Broader application to other edible oils and oil-rich foods is anticipated, and integrating with advanced analytical platforms (e.g., GC×GC-TOF-MS) and multivariate sensory correlation could deepen understanding of aroma-active constituents and support objective flavor characterization and quality control in olive oil.

The study introduces and validates a two-step extraction strategy, OA-LLE followed by SAFE ("two assists"), that effectively isolates and concentrates EVOO volatiles while removing non-volatiles that hinder analysis. OA-LLE + SAFE outperformed conventional SAFE and HS-SPME by recovering approximately twice as many compounds and increasing total volatile yield, including semi-volatiles and higher-boiling species. The intensified OA-LLE × 3 + SAFE protocol enabled deeper profiling across different cultivars, revealing many compounds with RI > 2000. This method facilitates more comprehensive and cleaner aroma extracts, supporting improved analytical accuracy, quality assessment, and sustainable supply of olive oil. Future work may integrate OA-LLE + SAFE with high-resolution chromatographic techniques (e.g., GC×GC-TOF-MS) and multivariate sensory analyses (e.g., PLS) to relate volatile profiles to sensory attributes and authentication markers.

The study does not explicitly list limitations. OA-LLE × 3 + SAFE applications to each EVOO were performed once (single replicate), which may limit statistical generalizability of those specific results. Additionally, detailed OA-LLE operational parameters were largely referenced from prior work, and HS-SPME/SAFE methods included minor modifications, which may influence method-to-method comparability across studies.