Flavor Profiling with the ecTOF – Volatilomic Comparison of Vegan and Non-Vegan Cheeses

cheese flavor profiling ecTOF

Marleen Vetter, Steffen Bräkling and Sonja Klee
TOFWERK, Switzerland

An increased awareness of health, environmental, and animal welfare issues, and a growing acceptance of vegan products, has led to a rise in non-dairy product sales [1]. Consequently, the market for meat and dairy-free alternatives is growing rapidly. Two primary commercial strategies exist for meat and dairy-free alternatives: production of a new product with a unique flavor, texture, and taste profile or the imitation of the already established non-vegan product. The latter is especially popular since customers are already familiar with a product and may want to switch to a vegan alternative.  Because of this, characterization of flavor and aroma profiles is of particular interest during the development of these types of products.

Flavor and Aroma Profiling

A large part of flavor and aroma perception arises from volatile and semi-volatile compounds released from the food product [2]. Gas chromatography coupled to mass spectrometry (GC-MS) is particularly useful at investigating these types of compounds. Typically, 70 eV electron ionization (EI) is used as the ionization method of choice in GC-MS. Unfortunately, this ionization method often results in the loss of molecular ion information, which is essential for compound identification as probabilities are reduced for identification, especially when appropriate reference standards for compounds are not available and library match information is insufficient [3].

Softer ionization techniques such as chemical ionization (CI) can overcome this issue and provide information on the empirical sum formulas of compounds when using accurate mass, high-resolution mass spectrometers. Furthermore, under controlled conditions CI can yield additional information about physio-chemical properties of the compounds of interest, such as their proton affinity. However, when trying to differentiate isomers or other similarly eluting compounds with the same sum formula, CI information alone falls short. Hence, TOFWERK developed the ecTOF, combining both EI and CI ionization in one instrument [3] to analyze molecular and fragment ions in parallel (Figure 1). With the ecTOF, data alignment is inherently provided, and tentative identification of unknown compounds, especially when combined with GC analysis, is rendered simpler and faster.

Figure 1. Principle of operation of the GC-ecTOF.

In this application note, we demonstrate comparing flavor profiles of both vegan cheese (New Roots AG, CH) and its non-vegan flavor equivalent with the ecTOF.

Compound of Interest Identification with the ecTOF

In a headspace vial, 10 g of a vegan cheese sample from New Roots AG, as well as its non-vegan equivalent (Appenzeller Classic, Appenzeller Käse GmbH, Appenzell Switzerland) were incubated for 10 minutes. After incubation, each cheese was sampled in triplicate at 60 °C for 30 mins using a Restek Polyacrylate SPME fiber (BGB Analytik, Switzerland). Full details on the method are found in Table 1.

SPME (Polyacrylate, Restek) Method10 g sample into 20 mL screw top vial, incubated for 15 mins at 60 °C, 2 cm SPME fiber exposed to headspace for 30 mins at 60 °C
Desorption and Injection  SPME held in injector port (235 °C) for 3 mins, splitless injection into Agilent GC 7890A with SPME injection sleeve (0.75 mm ID)
Carrier Gas Flow1.0 mL/min sccm He
Purge Flow10.0 mL/min
ColumnRestek Stabilwax-MS (30 m x 0.25 mm x 0.25 µm)
Septum Purge3.0 mL/min
Temperature Program35 °C for 4 mins, 10 °C/min 250 °C held for 5 mins
Flow Split1:1 CI/EI
Heated Transfer Line Temperature250 °C
Ionization SourcesStarBeam 70 eV EI source HRP CI source reagent: H3O+ [4]
Source TemperaturesEI 280 °C CI 300 °C
Mass Range1 – 450 m/Q (Th)
Table 1: Method Parameters

Initially, only the main flavor compounds of interest were investigated (Figure 2). Between the two types of cheese a clear difference between common flavor compounds such as butanoic acid and 2/3- methyl butanoic acid can be identified. Whereas the former compound has a more pungent odor and acrid flavor, the latter is known for its pleasantly sweet, fruity, and cheesy flavor profile. These are classic compounds of interest for cheese flavor profile determinations [5].

Figure 2. Comparison of key flavor compounds for the employed fermentation process. Ideally, both profiles should be very similar. However, a clear difference in peak proportionality can be seen, especially between butanoic acid and 2/3-methyl butanoic acid. a) shows a comparison of the combined EICs of m/Q 61, 89 and 103 and b) shows the proportional distribution of the peak ratios to one another. 

In comparison with a conventional target flavor compound method, analysis with the ecTOF provides the possibility of investigating non-target compounds recorded in the same analysis run. Figure 3 shows the comparison of chromatograms and different mass spectra between the vegan cheese sample and the non-vegan equivalent cheese. Significant differences between the vegan cheese and the non-vegan alternative can be identified using a volcano plot (Figure 4). A peak of interest for both the vegan and the non-vegan equivalent was identified at around 17.5 minutes (Figure 3b) and the EI and CI mass spectra for both runs are shown in Figure 3c.

 Figure 3. Comparison of vegan and non-vegan equivalent cheese flavor chromatograms. a) Comparison of the total ion chromatograms (TIC) for both EI and CI data of the two different samples.  b) Zoom-in of the chromatographic peak of interest at around 17.5 mins. c) EI and CI mass spectra detected for the peak of interest for both samples are shown.
Figure 4. Volcano plot of the vegan cheese and non-vegan cheese ecTOF data.

For the non-vegan cheese equivalent, both EI and CI information are in good agreement, pointing to 2,4,6-trimethyl-benzaldehyde (sum formula C10H12O) with a mass accuracy of 4.0 ppm, an isotopic pattern error of 5.2 % (Figure 5) and NIST library match factor 899, reverse match factor 928 and probability of 51.6 %.

Figure 5. a) CI extracted ion chromatograms of the non-vegan cheese peak at 17.5 minutes as well as b) the isotope ratio distribution and mass accuracy for m/Q 149.0988.

In the vegan cheese however, the EI mass spectrum clearly suggests that 2,3,4-trimetyl-benzaldehyde is not present, even though a chromatographic peak appears at the same retention time. The first suggested NIST hit for the peak at around 17.5 minutes (match factor 857, reverse match factor 878 and probability of 27.2 %) was benzyl alcohol (C7H8O). Not only the low probability in the NIST library search, but also a contradiction of the accurate mass of the possible molecular ion found in the CI mass spectrum (m/Q 149.0968) alludes to an inconsistency of the recorded data with the NIST library match. For many compounds, the retention index (RI) or equivalent measure for polar columns is unfortunately not reported, which was the case in this investigation. Hence, the RI information was of limited added value. Such as used here, retention index information unfortunately was of limited use in this investigation. While a more detailed examination of the peak specific mass traces in the EI chromatogram does not indicate a distinct coelution of substances, a coelution of two peaks is clearly visible the CI chromatogram (Figure 6a).

While the extracted ion chromatogram (EIC) of the CI chromatogram of m/Q 91 shows the presence of the tropylium ion ([C7H7]+) for both compounds, the EIC of m/Q 109.0652 (predicted sum formula [C7H8O]+ with a mass accuracy of 3.8 ppm) and m/Q of 149.0968 (predicted sum formula [C10H12O]+ with a mass accuracy of 5.0 ppm) markedly show the presence of two different compounds. Using this and the NIST library search information, the presence of benzyl alcohol and allyl benzyl ether (NIST hit 7, match factor 778, reverse match factor 791, probability 1.28 %) can be predicted. Since these two substances produce almost identical fragments in the 70 eV EI spectra, a separation of the two peaks using EI only is not possible (Figure 6b). Due to shifted fragment ion ratios between the two compounds that mix within the EI mass spectrum, a NIST library search alone is not reliable.

Figure 6. a) CI and b) EI extracted ion chromatograms of the vegan cheese peak at around 17.5 minutes as well as the isotope ratio distribution and mass accuracy for c) m/Q 109.0652 and d) 149.0968.


From these results, the benefits of using CI in addition to standard EI within the same GC-MS analysis to increase confidence are clearly highlighted, especially in complex samples. Whereas the ecTOF can easily identify commonly investigated flavor compounds of interest, it is also able to identify and distinguish other volatile and semi-volatile compounds which may play a crucial role in flavor development or in later processing steps. With this, the strength of the ecTOF in identifying flavor compounds in cheese samples is demonstrated.


[1] Fortune Business Insights. The global vegan food market is projected to grow from $26.16 billion in 2021 to $61.35 billion in 2028 at a CAGR of 12.95% in forecast period, 2021-2028. Vegan Food Market Report, 2022, March Issue: 1-160.

[2] McSweeney, P.L.H. and Sousa, M.J.: Biochemical pathways for the production of flavour compounds in cheeses during ripening: A review. Lait, 2000, 80 (3): 293-324.

[3] Bräkling, S. et al: Parallel Operation of Electron Ionization and Chemical Ionization for GC-MS Using a Single TOF Mass Analyzer; Anal Chem., 2022, 94 (15): 6057-6064.

[4] Bräkling, S. et al: Hydrogen Plasma-Based Medium Pressure Chemical Ionization Source for GC-TOFMS; J. Am. Soc. Mass Spectrom., 2021, 33 (3): 499–509.

[5] Schlegel, K.; Ortner, E.; Buettner, A.; Schweiggert-Weisz, U. Contribution of S. Xylosus and L. Sakei ssp. carnosus Fermentation to the Aroma of Lupin Protein Isolates. Foods, 2021,10:1257.