Injection artifacts in odorant analysis by gas chromatography

The study aims to evaluate various gas chromatography (GC) injection techniques to identify and minimize artifact formation during odorant analysis, a critical factor for ensuring accurate results in food and consumer product assessments. While pre-GC workup artifacts are well studied, this research focuses on those arising specifically during injection.

By testing 14 compounds across 10 GC injection methods, the study found that on-column injection—whether in the oven or via a programmable temperature vaporizing (PTV) injector—produces virtually no artifacts. In contrast, fixed high-temperature splitless injections, particularly in HS–SPME setups, led to significant artifact formation. The findings support using on-column injection for representative odorant profiling, especially in GC–olfactometry, and suggest cedryl acetate as a useful test compound for assessing artifact potential.

The original article

Injection artifacts in odorant analysis by gas chromatography

Julian Reinhardt, Martin Steinhaus 

Journal of Chromatography A, Volume 1741, 25 January 2025, 465624

https://doi.org/10.1016/j.chroma.2024.465624

licensed under CC-BY 4.0

Selected sections from the article follow. Formats and hyperlinks were adapted from the original.

In the field of odorant research, studies on artifact formation from volatile compounds during hot injection are scarce, regardless of whether hot injectors were applied to liquid samples or for thermodesorption in combination with headspace SPME or headspace SBSE.

Block and colleagues [[27], [28], [29], [30]] reported the formation of artifacts from thiosulfinates and dialkyldisulfides during GC–MS analysis and identified this as a particular problem when high injector temperatures were applied to organic solvent-based volatile isolates obtained from Allium species. Several studies demonstrated that headspace SPME–GC–MS, in combination with hot injection, led to the formation of disulfides from thiols and to the formation of sulfoxides from dialkyl sulfides [[31], [32], [33], [34]]. In particular, the formation of dimethyl disulfide from methanethiol was shown [31,33,34], both of which are potent odorants [6,7]. The same is true for dimethyl sulfide, which is an important odorant, for example, in cooked vegetables [6,7,35] and reacted to dimethyl sulfoxide in the hot injector [31,32]. A catalytic effect of metals included in the Carboxen adsorbent on the fiber was discussed; it remained, however, unclear what was serving as an electron acceptor in these oxidation reactions. Fischer et al. analyzed a solvent-based volatile isolate obtained from Pontianak orange peel oil by GC and observed the formation of styrene by elimination of hydrogen sulfide from sulfurous, burned smelling 1-phenylethane-1-thiol, a signature odorant of the peel, during hot injection [36]. Thus, in this case, hot injection not only led to a decrease in the concentration of a key odorant but, at the same time, also to the formation of a highly odor-active artifact, namely rotten egg-like smelling hydrogen sulfide. Pons and coworkers investigated the enantiomeric ratio of fenugreek-like smelling sotolon in dry white wine by GC–MS and found that the temperature of the GC injector port is crucial. High temperatures led to the racemization of sotolon. For example, at an injector temperature of 230 °C, the enantiomeric excess (ee) value decreased from 99 to 65 % [37,38]. One of us (MS) reported the formation of numerous artifacts from linalyl acetate during hot GC injection of organic solutions, among which were myrcene, (Z)-β-ocimene, (E)-β-ocimene, neryl acetate, and geranyl acetate [1].

The objective of the present study was to compare the extent of artifact formation from a selection of test compounds between a representative variety of ten different GC injection approaches. The injection techniques tested included split and splitless injections of liquid samples with fixed temperature and with the application of an injector temperature program, as well as classical headspace–SPME and headspace–SPME-Arrow sampling. Injection of a liquid sample with a cold on-column injector served as a low artifact reference approach.

2. Materials and methods

2.4. GC–FID

A Trace 1310 gas chromatograph (Thermo Fisher Scientific) was equipped with a TriPlus RSH autosampler, a flame ionization detector (FID), and one of the following three injectors: a cold on-column injector, a split splitless (S/SL) injector, or a programmable temperature vaporizing (PTV) injector. The splitless hold time in splitless injections was 2 min. The split ratio applied to split and splitless injections was 1:20. The injection volume in combination with liquid sampling was 1 μL. HS–SPME and HS–SPME-Arrow injections included sampling from 20 mL headspace vials with 5 mL sample at 40 °C during 2 min with preconditioned (250 °C, 3 min) polydimethylsiloxane/divinylbenzene (PDMS/DVB; 65 µm) fibers, followed by 2 min desorption in the hot injector (250 °C) and bake-out (250 °C, 40 min). The precolumn was an uncoated but deactivated fused silica capillary, 5 m × 0.53 mm i.d. for on-column injections, 5 m × 0.32 mm i.d. for all other injection types (Chromatographie Handel Müller; Fridolfing, Germany). The main column was a DB-FFAP, 30 m × 0.32 mm i.d., 0.25 μm film thickness (Agilent Technologies; Waldbronn, Germany). The carrier gas was helium at a constant pressure of 70 kPa. The oven temperature was 40 °C for 2 min, then increased by 6 °C/min to 230 °C, and the final temperature was held for 5 min. FID signal data were recorded by using the Chromeleon software, version 7.2.8 (Thermo Fisher Scientific), and copied to Excel (Microsoft; Redmond, WA, USA) for further evaluation. Solvent peaks were not included in the calculations of the area count percentages.

2.5. GC–Orbitrap MS

A Trace 1310 gas chromatograph (Thermo Fisher Scientific) was equipped with a TriPlus RSH autosampler and a PTV injector with a glass liner, 120 mm × 2 mm i.d., and tapered bottom. The precolumn was an uncoated but deactivated fused silica capillary, 5 m × 0.32 mm i.d. (Chromatographie Handel Müller). The main column was a DB-FFAP, 30 m × 0.32 mm i.d., 0.25 μm film thickness (Agilent). The carrier gas was helium at a constant flow of 2 mL/min. The injection volume was 1 μL. The oven temperature was 40 °C for 2 min, then increased by 6 °C/min to 230 °C. The final temperature was held for 5 min. The column end was connected to a Q Exactive GC Orbitrap mass spectrometer (Thermo Fisher Scientific). PCI spectra were generated in nominal or high-resolution mode with isobutane as the reagent gas and a scan range of m/z 84−350. EI spectra were generated in nominal or high-resolution mode at 70 eV and a scan range of m/z 35−300. Data were evaluated using the Xcalibur software (Thermo Fisher Scientific).

2.6. Headspace solid-phase microextraction (SPME)–GC–Paul trap MS

A gas chromatograph 7890B (Agilent) was equipped with a GC Sampler 80 and a Multimode injector with a straight glass liner, 78.5 mm × 0.75 mm i.d., operated in splitless mode (Agilent). The precolumn was an uncoated but deactivated fused silica capillary, 5 m × 0.32 mm i.d. (Chromatographie Handel Müller). The main column was a DB-FFAP, 30 m × 0.25 mm i.d., 0.25 μm film thickness (Agilent). The carrier gas was helium at a constant flow of 1 mL/min. Sampling and temperature programming were done as detailed in Section 2.4. The column end was connected to a 240 Ion Trap GC–MS mass spectrometer (Agilent). PCI spectra were generated in nominal resolution with methanol as the reagent gas. EI spectra were generated in nominal resolution at 70 eV The scan range was m/z 60−250. Data were evaluated using the MS Workstation software (Agilent).

3. Results and discussion

3.1. Selection of test compounds and qualitative insights into artifact formation

To identify volatiles suitable for the evaluation of different GC injection approaches with regard to the extent of artifact formation, a variety of pre-selected compounds were subjected to a comparative GC–FID analysis using a diluted sample of the test compound in DCM and, on the one hand, injection with a cold on-column injector as an artifact-avoiding reference approach or, on the other hand, isothermal hot injection at 250 °C in the splitless mode as an injection approach with high thermal impact [39]. Pre-selection of compounds considered the following criteria: i) occurrence in natural materials, particularly food, ii) previous reports on artifact formation during hot injection or structural similarities to compounds mentioned in previous reports, iii) odor activity of the test compound and/or the reported artifacts, and iv) availability in high purity. The final selection additionally considered the extent of artifact formation observed in the comparative GC–FID analysis, the success in identifying/confirming major artifacts in our lab, and the odor relevance of the artifacts found. The outcome of the test compound selection process was the set of 14 volatiles depicted in Fig. 1.

Journal of Chromatography A, Volume 1741, 25 January 2025, 465624: Fig. 1. Test compounds selected for assessing artifact formation during GC injection.

3.2. Quantitative insights into artifact formation: comparison of ten different injection approaches

Using the 14 test compounds introduced in Section 3.1, ten different GC injection approaches (Table 2) were evaluated with regard to the extent of artifact formation. All other analytical parameters were kept as constant as possible, e.g., all analyses were conducted with the same GC instrument, the injection volume of the liquid samples was always 1 µL, and the oven temperature program was identical.

Journal of Chromatography A, Volume 1741, 25 January 2025, 465624: Table 2. GC injection approaches evaluated with regard to artifact formation.

Results are depicted in Fig. 8, which shows the area percentage of the target compound in the chromatogram for each combination of test substance and injection approach, thus indicating the percentage of artifacts formed as the difference from a value of 100. Injection of a DCM solution of the test compound with a cold on-column injector served again as the low artifact reference approach (Table 2, approach 1). The reference approach was applied in duplicate, once at the beginning of the experimental campaign (1a) and once at the end (1b) to be able to detect a potential deterioration of the test compound in the solution and not misinterpret it as formation of injection artifacts. However, no substantial difference was found between 1a and 1b. As detailed in Fig. 8, only for 4MSP, with 99 %, the area percentage obtained from injection approach 1b was minimally lower than 100 %. Thus, the DCM solutions of the test compounds were considered stable over the period of the experiments, and deviations from 100 % observed with the other injection approaches were interpreted as artifact formation.

Journal of Chromatography A, Volume 1741, 25 January 2025, 465624 - Fig. 8. Degree of artifact formation after application of ten different injection approaches (cf. Table 2) to the 14 test compounds. Numerical values represent area percentages of the target compound; thus, the difference from a value of 100 indicates the area percentage of artifacts. The low artifact reference approach using cold on-column injection was applied in duplicate at the beginning (1a) and at the end (1b) of the experimental campaign.

The idea behind injection approach 2 was to simulate a classic cold on-column injector with a PTV injector run in cold on-column mode. Whereas in a classic cold on-column injector, the sample is released from the tip of the syringe needle into a section of the uncoated precolumn positioned inside the GC oven (in-oven cold on-column), cold on-column injection with a PTV injector involves the release of the sample inside a section of the precolumn within the injector (in-injector cold on-column). Thus, the volatiles need to be transferred from the column section inside the injector to the column section inside the oven. In injection approach 2, this was achieved with a shallow temperature program applied to the injector that paralleled that of the oven, thus simulating the situation in classic cold on-column injection. Consequently, this approach did neither result in a substantial formation of artifacts from the 14 compounds tested (injection approach 2 in Fig. 8).

Injection approach 3 met classic PTV injection into a glass liner in combination with splitless mode. As suggested by the manufacturer for avoiding band broadening, a fast heating rate of 300 °C/min was applied. Despite a potentially higher heat impact on the test compounds, this did not result in artifact formation (injection approach 3 in Fig. 8). It notably changed when the injector temperature was fixed at 250 °C (injection approach 4 in Fig. 8). Substantial artifact formation was found from 1-phenylethane-1-thiol (6 %), cedrol (8 %), linalyl acetate (10 %), and particularly cedryl acetate (22 %). Using the same setting but with a metal injector liner instead of the standard glass liner increased the degree of artifact formation (injection approach 5 in Fig. 8). Despite deactivation, the metal surface might have acted as a catalyst and thus fostered artifact formation as previously also suggested by Edge [74]. Switching from splitless mode (injection approach 4 in Fig. 8) to split mode (1:20) almost entirely avoided the formation of artifacts, as reflected by recoveries of the target compounds ≥97 % (injection approach 6 in Fig. 8). This suggested the time the target compounds dwelled inside the hot injector as a major driver for artifact formation [39].

Injection approaches 7 and 8 paralleled approaches 4 and 6, but instead of a modern PTV injector, a classic S/SL injector in combination with a liner packed with 10 mm silylated quartz wool was employed. This resulted in clearly higher degrees of artifact formation, particularly in the splitless mode (injection approach 7 in Fig. 8) and for some compounds as well in the split mode (injection approach 8 in Fig. 8). In the splitless mode, the percentage of artifacts increased, e.g., from cedryl acetate (32 vs. 22 %), cedrol (19 vs. 8 %), and 1-phenylethane-1-thiol (11 vs. 6 %), but most spectacularly from linalyl acetate (64 vs. 10 %) and neryl acetate (53 vs. 3 %). In the split mode, the use of the S/SL injector particularly led to the formation of artifacts from cedryl acetate (16 vs. 3 %), cedrol (6 vs. 0 %), and linalyl acetate (4 vs. 0 %). This outcome suggested that besides injection type and temperature, either the injector geometry had an additional impact, as the PTV injector was longer but substantially smaller in diameter, or, more likely, the glass wool filling included in the S/SL liner. The latter might have substantially increased the dwell time of the analytes in the hot injector.

Injection approaches 9 and 10 differed from the previous ones in that the samples were aqueous instead of DCM solutions, and sampling was accomplished from the headspace by SPME using a classic SPME fiber or a SPME-Arrow with a higher capacity, both coated with PDMS/DVB [75,76]. Desorption was accomplished in an S/SL injector using the splitless mode, as suggested in combination with SPME [77]. As easily visible from the color coding in Fig. 8, these two injection approaches (9 and 10 in Fig. 8) turned out to be overall most critical in terms of artifact formation. Only linalool (4) and 2-phenylethane-1-thiol (10) were virtually unaffected. Nine of the remaining 12 test compounds showed the highest degree of artifact formation with SPME sampling; however, artifact formation from linalyl acetate (1), neryl acetate (2), and cedrol (8) was higher during splitless injection with the S/SL injector (injection approach 7 in Fig. 8). The lowest overall recovery of a target compound was obtained after SPME-Arrow injection of cedryl acetate, namely 10 %, meaning that 90 % of the injected compound reacted to artifacts. A direct comparison of the two SPME approaches for most compounds showed a higher degree of artifact formation with the SPME-Arrow than with the classic SPME fiber. This could also be due to the overall dwell time in the hot injector, as it can be assumed that the compounds require more time to diffuse to the surface of the thicker PDMS layer of the Arrow device. Variation of the test compound concentrations in the test samples, however, did not show a substantial impact on the degree of artifact formation (data not shown).

4. Conclusions

On-column injection was confirmed as the gold standard for the artifact-avoiding application of volatiles into a gas chromatographic system. However, not only the use of a classic cold on-column injector with release of the compounds into the precolumn inside the oven resulted in virtually zero artifact formation, but also cold on-column injection inside a PTV injector in combination with a temperature program. Nevertheless, injection approaches associated with releasing the analytes into a liner, do not necessarily produce artifacts. Neither classic PTV injection in the splitless mode nor the combination of a PTV injector with fixed temperature and split injection produced artifacts. By contrast, substantial artifact formation was found with all injection approaches that combined a fixed injector temperature with splitless injection, whether of liquid samples or in combination with SPME. These approaches thus need to be carefully avoided whenever even tiny amounts of artifacts can substantially falsify the results, as is the case, e.g., during screening for odor-active compounds by GC–O [1]. However, other sources of artifacts during sample preparation always need to be additionally considered [20]. Furthermore, artifact formation is not only an issue in the field of odorant analysis, but a general problem in GC analysis, particularly of complex mixtures from natural sources [39].

In targeted approaches, at least when critical injection approaches are used in combination with critical target analytes such as compounds whose structures suggest susceptibility to unimolecular thermal reactions (e.g., eliminations), test injections are highly recommended to exclude artifact formation. Individual solutions of the appropriately diluted target compounds should be used for these injections. Parallel analysis using an on-column injection technique should be performed to exclude misinterpreting impurities as artifacts.