Beyond Conventional GC–MS: Revealing Hidden Food Differences

The ever-growing demands of consumers mean that manufacturers need to turn to more advanced analytical methods to uncover fresh insights that will give them a competitive advantage. This article demonstrates two automated methods for supplementing conventional gas chromatography–mass spectrometry (GC–MS) analysis to provide additional information that is valuable for understanding real-world product performance: automated sequential tube sampling for time profiling and advanced chemometrics for rapid comparison of product lines.

With more than 75% of perceived flavor being attributed to smell, aroma plays a vital role in overall sensory experience.^1,2 Therefore, meeting consumer demands for new and improved foods and beverages has long relied on understanding the range of volatiles released from a product, typically using gas chromatography–mass spectrometry (GC–MS) in conjunction with solid-phase microextraction or headspace sampling techniques, coupled with sensory panels.

Advances in these and other methodologies now mean that it is routine to correlate analytes and their abundances with aroma characteristics,³ with the benefits being better quality control of existing products and streamlined development of new flavor sensations.

However, in the competitive food and beverage industry, manufacturers need to constantly extend their capabilities to stay ahead—a need that the development of new and improved analytical methodologies can help meet. This article presents insights into the aromas of a microwaveable curry meal and flavored sparkling waters obtained by applying two automated techniques that are readily incorporated into a GC–MS laboratory setup.

The first technique, multitube sequential air sampling, facilitates the characterization of changes in airborne volatile organic compound concentration over time without the need for continuous manual intervention or site visits. It works by sequentially drawing defined volumes of air through up to 32 sorbent-packed thermal desorption (TD) tubes, for subsequent analysis by TD–GC–MS. Sampling times and intervals can be programmed for each tube, supporting short, high‑volume sampling during initial high-concentration analyte release and less-frequent collection during aroma dissipation.

The second technique, chemometric profiling, enables more reliable, automated identification of minor differences between GC–MS profiles, for more robust characterization with less reliance on error-prone manual assessment. The software works by splitting the chromatogram into small segments, or “tiles,” and automatically recording the signal intensity for each m/z value across each tile. It then filters the data to highlight the tiles that differ between samples, so that the compounds responsible can be identified.

Experimental

Experiment 1: Time-Resolved Aroma Profiling of a Microwaveable Curry Meal

Sample Preparation: A microwaveable curry meal was acquired from a local supermarket. Outside the sampling room, the meal's film lid was pierced, and the meal was microwaved according to the cooking instructions.

Sampling: Two minutes prior to completion of the cooking time, an MTS-32 Pro multitube sampler (Markes) fitted with 32 conditioned 3.5 in. × 0.25 in. o.d. stainless-steel Tenax TA tubes (Markes) was activated in an otherwise empty, enclosed sampling room (tubes 1 and 2). At the end of the dwell (interval) stage (of tube 2), the cooked meal was brought into the sampling room. A room sample was collected with the covered meal present (tube 3). During the subsequent dwell stage, the film lid was removed, and the remaining samples (tubes 4–32) were collected, during which time the room was vacated. A sampling flow rate of 1 L/min was used, and each tube was sampled for 0.33 min, with a dwell time of 0.67 min, giving a total sampling schedule time of 32 min.

Analysis: Thermal desorption was carried out on a TD100-xr instrument (Markes) with a tube desorption temperature of 300 °C (10 min) and with vapors collected onto a “Material emissions” focusing trap (Markes) at 30 °C. Subsequent desorption was carried out at 320 °C (3 min) with a 3:1 outlet split. GC–MS was carried out using an HP-5ms column with helium as the carrier gas (1.5 mL/min), with an oven program starting at 35 °C (3 min), ramping at 20 °C/min to 300 °C (10 min). Detection was carried out in scan mode (m/z 35–350).

Experiment 2: Chemometric Comparison of Flavored Sparkling Water

Sample Preparation: Three flavors of sparkling water (cherry, lime, and blackberry) were obtained from the same manufacturer, in both syrup form for home preparation and as ready-to-drink cans.

Sampling: Direct immersion sorptive extraction involved placing sparkling water (15 mL) in a 20 mL vial and incubating it for 30 min at 40 °C and 120 rpm using a DVB/PDMS HiSorb probe (Markes). Experiments were carried out in triplicate.

Analysis: Thermal desorption was carried out on a Centri system (Markes), with vapors collected onto a “material emissions” focusing trap (Markes). GC–MS was carried out using a ZB-DHA-PONA (100% PDMS) column (50 m × 0.20 mm, 0.5 µm), with an oven program starting at 35 °C (3 min), ramping at 5 °C/min to 300 °C (6 min). Detection was carried out in scan mode (m/z 35–600).

Data Processing: Chemometric analysis was carried out on the ChromCompare+ platform (SepSolve).

Results and Discussion

Experiment 1: Time-Resolved Aroma Profiling of a Microwaveable Curry Meal

Following analysis of a blank run (using a clean sorbent tube) and two room blank samples (captured before introduction of the curry meal), the tubes used to sample volatiles in a room over a 30-min period following introduction of the precooked curry meal were analyzed.

The analytes present in the sampling room 6 min after opening provided herbal, woody, and spicy aroma notes characteristic of curry (Figure 1). Of note is the presence of cuminaldehyde, a compound that primarily originates from cumin seeds and indicates the use of cumin as an ingredient.

Figure 2 shows the concentration profiles of the top five abundant compounds collected in the 32 min of the sampling schedule and shows the small increase in analyte concentration between the samples taken at 2 min and 3 min, the interval when the film of the meal was removed. A more significant increase is apparent at 4 and 5 min, and concentrations peak at 6–7 min before dissipating, albeit to levels higher than before introducing the meal. This illustrates how this approach can provide representative tracking of food aroma evolution over time.

Experiment 2: Chemometric Comparison of Flavored Sparkling Water

In this experiment, three flavored sparkling waters—lime, blackberry, and cherry—were analyzed using direct-immersion sorptive extraction GC–MS, followed by automated chemometric analysis. For each flavor, commercially premixed canned products were compared with home-prepared sparkling waters produced using carbonation systems and flavored syrups from the same manufacturer.

Chemometrics was used to automatically align and compare the chromatograms, which easily identified significant differences among the three flavors, with over 30 differentiating compounds found, such as benzaldehyde, ethyl hexanoate, and geranial in the cherry-, blackberry-, and lime-flavored samples, respectively.

A principal components analysis showed that the flavor profiles of premixed and home-prepared sparkling waters were closely aligned across all three flavors, suggesting that the flavor formulation used in the syrup for the home-prepared drinks effectively replicates the ready-to-drink canned products.

However, further chemometric analysis revealed that only a single tile was statistically increased in premixed cans compared to home-prepared drinks across all flavors (Figure 3, red dot). The trace compound responsible for this difference was identified as di(2-ethylhexyl) sebacate, which is a known corrosion inhibitor for metal cans, meaning it may have entered the canned beverages through minor contamination during the production process.

The same plot also shows one feature that was significantly increased in all home-prepared beverages (Figure 3, blue dot). The compound responsible was not present in commercial reference libraries, so its identity could not be confirmed, but it closely resembled compounds in the 1,3-dioxolane family.

To further explore differences arising from the production method, the data were further processed to automatically subtract the chromatogram of the canned cherry water from that of the home-prepared version, and vice versa. This highlighted the key differences between the samples, enabling rapid identification of features, including trace peaks that would be difficult to distinguish by manual inspection.

The resulting subtracted chromatograms (Figure 4) clearly show several flavor-active compounds that were elevated in one of the cherry-flavored waters, suggesting potential differences in consumer perception. Notably, anethole, a sweet, licorice-like compound,⁴ was found to be higher in the home-prepared drink. In addition, minor peaks attributable to several long-chain alkanes, caprolactam, and the previously noted di(2-ethylhexyl) sebacate were elevated in the canned beverage, likely originating from packaging or processing.

Conclusions

Meaningful aroma analysis of foods and beverages is highly challenging because it depends on having access to highly sensitive chromatographic techniques, alongside robust sensory protocols to relate that data to the real-world experience of consumers.

Therefore, automated approaches that expand the versatility of (TD–)GC–MS methods and make it easier to generate insights from the data will be welcomed by food and beverage manufacturers, and two such approaches have been illustrated here. The first—automated, time-resolved air sampling coupled with thermal desorption—is a practical method for understanding the release and persistence of aroma compounds from a cooked food product, while avoiding the risk of missing fluctuations in aroma intensity that could occur when relying on single-point measurements. Chemometric analysis of GC–MS data enables confident, automated comparison of products, effectively capturing flavor differences between products and subtle variations related to production methods.

Using such automated techniques, companies seeking to understand how their products compare against others in real-world, consumer-relevant scenarios will have greater access to high-quality information. That, in turn, will enable them to make better decisions about product development while reducing reliance on time-consuming manual processing.

References

1. Zhang, T.; Spence, C. Orthonasal Olfactory Influences on Consumer Food Behaviour. Appetite 2023, 190, 107023. DOI: 10.1016/j.appet.2023.107023
2. Hansford, N. The Power of Scent in Marketing Food Products [blog post], Air Products, https://www.airproducts.co.uk/industries/food/power-of-scent (accessed May 13, 2026).
3. Louw, S. Recent Trends in the Chromatographic Analysis of Volatile Flavor and Fragrance Compounds: Annual Review 2020. Anal Sci Adv 2021, 2, 157–170. DOI: 10.1002/ansa.202000158
4. The Good Scents Company. Information System [search facility], www.thegoodscentscompany.com (accessed February 6, 2026).