Foods and Contaminants Analysis Using Multidimensional Gas Chromatography: An Interview with Philip J. Marriott and Michelle S. S. Amaral

1. How has multidimensional gas chromatography (MDGC) improved the detection and analysis of trace contaminants in complex food samples compared to traditional GC?
At the outset, it is useful to state that there are effectively two variants to what might be broadly classified as the MDGC method. So to make it clear, we will remind readers of the distinction. The first is a traditional two-column discrete heart-cut operation, where section(s) of eluted compounds from a first column are transferred to a second column. With two disparate stationary phases in the two columns, the concept is that compounds not resolved on the first phase column may be resolved on the second column. We refer to the two columns as the first dimension (¹D), and the second dimension (²D), from which the term multidimensional separation arises. Whilst there is no guarantee that the compounds will be resolved, choosing a different phase does increase the likelihood of resolution. Thus the whole basis for the MDGC method is that resolution is (expected to be) improved. Method variations to this MDGC idea include performing multiple heart-cuts; this may include performing the heart-cuts and continuing the analysis at the prevailing temperature, or cooling the oven before continuing the analysis under a separate programmed temperature operation, In all cases, using some approach such as cryo-trapping the heart-cuts to reduce dispersion of the peaks transferred from the ¹D column is advised.The second two-column operation is comprehensive two-dimensional gas chromatography (GC×GC). The essential difference is the use of a modulation mechanism that assists in the transfer process, and usually should operate at a modulation period faster than the width at baseline of the ¹D peak. The ²D column should complete each analysis within the period of the modulation process. It should therefore generate very narrow peaks at the end of ²D. And using a cryogenic or thermal modulator accompanies ‘peak compression’, which translates as both sharp, and enhanced response modulation.

Thus MDGC in general has significantly enhanced the separation and detection of trace contaminants in complex food samples by providing higher resolution and peak capacity compared to traditional gas chromatography (GC). This means that adding another separation step (“dimension”) and using cryogenics or other peak focusing methods allows us to efficiently circumvent co-elutions of target compounds. Having said this, trace analysis of contaminants is favored because (i) peaks are better resolved from otherwise overlapping matrix peaks, which for trace analysis might also be from considerably more abundant interfering peaks; (ii) phase bleed that might be present throughout the ¹D column analysis can be resolved from analyte peaks (especially when using GC×GC); (iii) mass spectrometry data are expected to be much more reliable (for example, improved comparison with database spectra) due to now having completely resolved compounds, and (iv) with cryogenic modulation, the mass abundance is larger, with better signal-to-noise ratios, leading to improved MS matching, and/or peak detection. Hence compounds can be measured at lower detection limits. This all translates as improved trace analysis.

2. What are the primary advantages of heart-cut (H/C) MDGC and GC×GC in identifying compounds for food safety and quality control?
Both techniques offer excellent separation and detailed analysis of complex mixtures, which is important for accurate food safety and quality assessments. The difference between them is how they are applied. We will highlight this here. A further difference between the heart-cut MDGC and GC×GC methods is that for most MDGC approaches you need to design a method based upon known compounds that is, it is a targeted method. Contrast this with GC×GC, which is essentially an untargeted volatile analysis method; now all volatile compounds are measured, giving a high resolution 2D separation space, such that each compound locates in its own (ideally unique) coordinate position based on first (¹t_R) and second (²t_R) dimension retentions. We might call this a 2D chemical property map. So a given compound locates in the same position across all samples provided conditions are not changed. This should make identification of compounds highly reproducible, and especially for impurities, very well defined once they are identified. The value of this property cannot be overstated.

For known compounds, MDGC has the primary benefit of increased resolution allowing for the isolation and further separation of certain regions of interest, which is crucial for identifying specific contaminants and biomarkers. This then improves analysis according to the principles previously expressed. For instance, if the focus of the analysis is to investigate the authenticity of a food product by assessing the specific enantiomeric ratios of its chemical markers (with enantioseparation used as the ²D column), MDGC can help cutting and resolving chiral compounds in regions where these target analytes are expected to elute. The same idea works for analysis of chemical contaminants.

The untargeted method of GC×GC essentially displays the full volatile profile of the sample – ideally nothing is hidden underneath large overlapping peaks. If, for instance, pesticide residues in a vegetable or fruit sample are of interest, then these can be measured at low levels, each in a unique 2D position that can be monitored with a mass spectrometer or a selective detector, and quantification can be conducted by using calibration solutions or suitable internal standards. It may be that sample analysis in this case for pesticide residues could be simply based on confirming peak positions, or noting their absence. As a further example, when investigating the adulteration of a certain food product, not only by one or a few compounds, but by mixing another complex food product (often less expensive), the analyst may need to examine the whole sample composition to confirm the fraud.

3. Can you explain the importance of sample preparation and derivatization in preparing food samples for MDGC? Derivatization may still be required if this affords best chromatographic peaks for particular analytes. For instance, this might include fatty acids, and amino acids; this enhances the volatility and thermal stability of analytes. MDGC and GC×GC do not change the basic nature of the gas chromatographic process with respect to suitability of compounds towards GC analysis. However, it can be that use of new columns reduce the need for derivatization, but we cannot always use new columns! The higher resolution of MDGC methods is still of advantage. Sample preparation is now potentially much less important, depending on the reason for the sample prep. Removing materials that could harm the GC system – injector / columns – would still be advised.But if sample prep is to simplify the sample due to interferences, then it is possible that the removal of interferences is less a concern due to the increased peak capacity available to the GC×GC method. This might be one aspect in favor of GC×GC, and others might be apparent to analysts according to their circumstances.

4. How does MDGC–MS facilitate the identification of food fraud markers, such as unapproved substitutions or adulterations?
The role of any GC method to successfully detect food fraud will depend on the nature of the substitution and adulteration required to be monitored. If the solution can be found by increased peak capacity, and the degree of separation of analytes, then this is a vote in favor of a MDGC / GC×GC approach. As exemplified in question 2, it can facilitate the detection of food fraud by providing detailed information of certain (bio)chemical markers, with MS providing the key to unlock molecular identity. This is usually achieved by using mass spectrometry data, to accurately identify compounds, determine isotopic or enantiomeric ratios, and monitor selected ions. This technique, therefore, makes it possible to identify subtle differences that indicate unapproved substitutions and adulterations. If identification relies upon the use of a specific mass spectrometry (MS) approach such as isotope ratio MS, then MDGC will likely be a valuable technology since it is important to achieve quantitative resolution of a target analyte in order to assess isotopic information. On the contrary, isotope ratio mass spectrometry (IRMS) has been little used with GC×GC. If adulteration can be tested by use of enantioselective analysis, then this is a classical application of MDGC with heart-cutting of the target compound to an enantioselective column used as the second (²D) column. A GC×GC method with chiral separation requires a ¹D enantioselective column, due to the difficulty in achieving chiral resolution on the very short ²D column which is used in GC×GC. In this case, precise enantioselective ratios are best determined provided adequate ¹D resolution is achieved, due to the modulation process in GC×GC. 

5. What role does MDGC play in untargeted profiling for comprehensive chemical composition analysis of foods?
As indicated above, the classical MDGC method is a targeted method. However we have sought to extend this to a suitable high resolution separation method by using extended sampling of the ¹D column effluent (for example 1 min), cryo-trapping the effluent, then rapid delivery to a high resolution ²D column (for example 0.1 mm ID, 5-7 m long) on which all compounds are eluted within the 1 min sampling time.Here, peaks on the ²D column average about 1 s peak width, so the effective peak capacity for each ²D analysis is about 60. This translated our MDGC method to a ‘comprehensive’ approach, since it can analyze all compounds.

By contrast, GC×GC is by definition an untargeted profiling method. All injected volatile organic compounds (VOCs) will be measured subject to them being eluted from the column within the temperature limits set, and if fully resolved will provide a detailed and complete composition of volatiles in the sample. We have previously called GC×GC the ultimate tool for volatile metabolite profiling. In a previous work of our research group, for example, the analysis of agarwood oxygenated sesquiterpenes has really benefited from this analytical approach, increasing the number of resolved peaks by more than 3-fold. Both MDGC and GC×GC were used in this study. Hop samples were also analyzed in the same study with similaroutcomes (Yan et al., 2018; https://doi.org/10.1021/acs.analchem.8b00142).

6. How do you approach data interpretation for MDGC, especially in managing complex datasets and using chemometric tools?
MDGC and GC×GC data analysis and interpretation is still a complex task that involves multiple steps, such as integration, normalization, baseline correction, peak alignment, mass spectra investigation, statistical analysis, and, depending on the aim of the study, this is followed by chemometric methods, such as principal component analysis (PCA) and hierarchical cluster analysis (HCA). Happily, chemometric tools are improving all the time, with better computing power and refined metrics. In terms of comparison of samples, for example authentic vs adulterated samples, if the GC analysis (here, GC×GC analysis) better quantifies sample compositions, and if embedded in the volatile profile is the evidence that samples differ arising from the adulteration, then it stands to reason that we now have a better differentiation of authentic vs fake samples. To some extent, the question can be to secure a valid dataset of authentic samples, such as from guaranteed manufacture, to then contrast with the non-genuine. Fortunately, many instrument companies recognize the need to incorporate chemometric platforms into their software, so they have made advances in this area, and the analyst’s task is considerably simplified.

7. What is the potential for portable MDGC systems in on-site food testing, and what challenges remain in this area?
Perhaps at the moment translating the basic GC method into fully and functional MDGC and GC×GC technologies for remote and/or portable systems is a step too far, although conceptually portable MDGC systems have the potential to revolutionize on-site food testing by providing flexibility and rapid analysis capabilities. Just establishing a portable technology for reliable GC-MS is still to emerge, although there are efforts in this area. Remote operation, where all the facilities (power, gas supply) for relatively classical instrumentation are available, functions well in dedicated environments such as some operations in petroleum refining, and the Advanced Global Atmospheric Gases Experiment (AGAGE) labs for environmental air analysis, and these areas include relatively sophisticated MDGC technologies. But GC×GC methodologies seem to be little reported. However, the use of portability of MDGC and GC×GC presents additional challenges that only dedicated research studies are addressing.

8. Which recent advancements in MDGC technology or methodology have most influenced your approach to food analysis?
Advances in GC×GC technology, such as improved modulators and modern data analysis software, perhaps integrating new artificial intelligence (AI) tools, are some of the things we particularly anticipate seeing in more studies. These technologies have the potential to improve accuracy and efficiency in food analysis. Perhaps the message of how Unilever incorporate MDGC and GC×GC into their analytical strategy (Personal Communication, Prof H. G. Janssen) provides an excellent model of using a variety of advanced instrumental approaches to food analysis integrated into the decision-making for routine quality assurance in this area. In this case, Unilever profiles all new processes by using GC×GC to establish the key information available from a particular process, and how it is best interpreted by a range of available methods. Based on this information, a decision is made as to which technique is required to assess the process. If a basic GC–MS method is adequate, then this will be used. GC×GC always is available to provide troubleshooting as and when necessary. Our lab only touches on a few aspects of food analysis, within our focus on innovation in instrumental analysis, and so we must defer to those with specific expertise for guidance.

9. What has been the response from other researchers to this review paper?
In terms of the published article, the response has recorded 4 citations in 2023, and 8 citations in 2024 thus far. One would hope that downloads might represent the interest a little better. Within our own University, this paper was noted and resulted in an invitation (Michelle Amaral) to give a seminar for the Master of Food Science and Agribusiness in May 2024. The invitation to participate in this interview we take as a further recognition and is appreciated by the authors of the review.

10. What are your next steps to continuing coverage of this type of research in food products?
We continue to address specific aspects of food analysis as the need arises, such as when students express interest in particular projects, or when we offer projects to various cohorts of students (for instance in a taught Master course which offers a research project unit), or when collaborators are interested in a project. Coffee continues to be an interest, and ranges from investigating roasting conditions, to monitoring specific profiles of S- and N-containing compounds, or profiling complete VOC to compare various coffees (Arabica, Robusta, Liberica). Yeast extract samples and integrating GC-olfactometry (GC-O) with GC×GC is a recent interest. Fatty acids analysis has been a particular interest for some time, largely for the complexity of samples, the unique ‘multidimensionality’ represented by these samples that makes them ideally suited to a MDGC / GC×GC approach, and the broad range of sample types for which fatty acids (FA) and fatty acid methyl esters (FAME) analysis is required. Essential oils and flavors is an extraordinarily productive area for GC×GC, and hopefully some recent initiatives in this area will be reported shortly. In recent work we refer to ‘shape-shifting molecules’ that accompany ‘molecular interconversions’ in GC, and encompasses an area that includes fascinating 2D patterns that are presented in the GC×GC result. The manner of how GC×GC reveals information about the kinetics and reactivity of a range of molecular systems is truly intriguing.

We envisage the development of more advanced and user-friendly systems, the integration of GC×GC with other analytical techniques, and its application to a broader range of products. Continued collaboration with other researchers and the dissemination of findings through publications and conferences will also be essential for advancing the field, which will certainly contribute to improving methods for food safety and quality control.

Reference

(1) Nolvachai, Y.; Amaral, M. S.; Marriott, P. J. Foods and Contaminants Analysis Using Multidimensional Gas Chromatography: An Update of Recent Studies, Technology, and Applications. Anal. Chem. 2023, 95 (1), 238–263. DOI: 10.1021/acs.analchem.2c04680

Michelle S.S. Amaral completed a B.Sc. in Chemistry (major in Natural Products Chemistry) at the Federal Institute of Rio de Janeiro (IFRJ, Brazil) in 2014, a M.Sc. in Chemistry at the Federal University of Rio de Janeiro (UFRJ, Brazil) in 2016, and a Ph.D. in Chemistry at Monash University (MU, Australia) in 2023. She is currently a postdoctoral research assistant at Monash University, where she has also worked as a teaching associate and exam supervisor. Her main research interests include the development of sample prep and analytical methods for gas chromatography profiling of flavors & fragrances, foods and natural products. Additionally, she is also interested in the investigation of green technologies (biocatalysis) for different applications in the aforementioned area.

Professor Marriott’s research commenced at LaTrobe University (Ph.D.) then University of Bristol (postdoc). Academic appointments include the National University of Singapore, at RMIT and Monash University in Melbourne. The invention of cryogenic modulation in GC, resulted in extensive research in GC×GC and MDGC, and mass spectrometry.A substantial number of overseas visitors – about 80 – have visited his lab for collaborative research. His research includes device development, fundamentals of GC×GC technology, and a broad applications base applying MDGC and GC×GC to a diverse array of separations: pesticides; petrochemicals; pharmaceuticals; natural products; toxins. In the food area, these technologies are applied to olfactometry, flavors, juices, coffee, beer, hop, fatty acids, polyphenols, food safety, herbs and spices and so forth. He has had visiting appointments in China, South Korea, Singapore, Brazil, and Malaysia.