Designing High Resolution Separations for Enhanced Characterisation of Petrochemical Samples

Advanced separation science – why petrochemical analysis?

The history of petrochemical analysis by using gas chromatography (GC) closely reflects the milestone developments in the technology that has defined improvements in GC separation, use of detection methods for improved characterisation, and exploitation of multidimensional separation approaches. The obvious reason for this is that petrochemical samples are exceedingly complex, comprising multiple chemical classes, along with heteroatomic species, requiring speciation in order to understand a range of chemical parameters. These range from processing conditions, to molecular markers, and overall characterisation. This essentially means that petroleum analysts are generally early adopters of technical advances in GC and mass spectrometry (MS).   
As an example, the analytical team at Shell (Amsterdam) [1] were quick to recognise the new capabilities offered by comprehensive two-dimensional gas chromatography (GC×GC), soon after its introduction by Liu and Phillips, [2] and worked with early modulator devices to investigate operational effects and define the applications scope of GC×GC. The US Centres for Disease Control and Prevention likewise had challenging analyses (e.g. dioxins) that demanded superior GC separations. [3]  Along with the Free University of Amsterdam, [4] one could justifiably believe that the epicentre of this new technique had a firm home in The Netherlands. The first GC×GC symposium was held at Volendam, The Netherlands, organised by key Dutch researchers and the Free University of Amsterdam academics, who were instrumental in promoting GC×GC.  
Comprehensive analysis of a sample implies analysis of all components in the sample. For a GC method, this will focus on the volatile / semi-volatile compounds. Whilst GC-FID accomplishes this task, complex (multi-component) samples will suffer indeterminate overlap that prevents adequate reporting of the components – along with limited identification. GC-MS also has the same resolution limitations, however identification can be provided in those cases where the MS allows searchable unique spectra. To some extent, deconvolution may permit overlapping components to be reported. However, the total sample composition will still largely be uncertain. The two-column separation method, GC×GC, addresses this problem by significantly increasing ‘peak capacity’ (i.e. total separable peaks) with potential to theoretically resolve up to ca. 10,000 components. Here, we expand on separation methods that describe instrumental approaches designed for high resolution separations.  

Instrumental Schematics,
Advantages and Attributes of GC×GC

An instrument schematic is shown in Fig 1, [5] with stages A-B-C-B defining a sequential arrangement of up to 4 columns, in which between each stage is a device for heart-cutting or modulation. For example, a ‘simple’ multidimensional GC separation will involve stages A-B, with a heart-cutting ‘switch’ (e.g. a microfluidic Dean’s switch. [6] In this case, the Dean’s switch cuts one or more segments from the first column (1D; A) to the second column (2D; B), and the detector (FID here) provides a monitor signal for the total first column elution. The 2D column and detector completes the analysis.
The simplest conventional GC×GC system will be described by A1-B1 with a sampling modulator device. The choice of columns largely depends on the nature of the stationary phases chosen for best separation of compounds on the 2D column, and then the column dimensions. The usual phase selection will be to combine a non-polar (NP) phase with a high polarity (P) or medium polarity (MP) phase, in either order. For instance, for petrochemicals, a 95% methyl/5% phenyl polysiloxane (NP), with a 50% methyl/50% phenyl polysiloxane (P) column is commonly used.
This normally meets the requirements for modulation ratio considerations [7] (MR = 1wb/PM); if MR = 3, the 1D peak is sampled (modulated) to produce about 3 modulated peaks; the length of the 2D column will be determined by the need to complete each 2D chromatogram within the PM time – e.g. 2-4 s. General characteristics of GC×GC are listed in Table 1, and the new nomenclature required of GC×GC has been outlined. [8]
Various advanced approaches are summarised in Fig 1. For instance, a comprehensive GC×GC×GC arrangement comprises 3 columns (A-B-C), usually of progressively shorter length, with modulation devices between 1D and 2D, and 2D and 3D respectively. [9] This necessitates data presentation in 3D space, with 3 independent axes for 1tR, 2tR and 3tR respectively. Investigations in this area are still largely exploratory, so rationalising column phase choice, and applications are awaited. The A2 and B2 options in Fig 1 describe a pressure-tuning (P/T) approach, [10] for either the 1D or 2D separation.  In this method, two columns replace either 1D or 2D, and effectively provide a variable apparent ‘polarity’ as a composite column. Thus if the two columns for the A2 coupling comprise NP and P phases, the resulting P/T 1D separation can range from a non-polar- to a polar-type separation. When coupled with the 2D column, the 2D separation space can be tuned by the effect of pressure between the two A2 1D columns.
A third option uses a ‘hybrid’ arrangement which combines both MDGC and GC×GC. [11] Briefly, this allows an operation such as selection of heart-cut region(s) from a 1D column, followed by GC×GC analysis on 2D and 3D columns. The alternative – GC×GC followed by MDGC – permits a special operation; the modulation process can use a slow PM setting, then using a flow switching device (Dean’s switch) an individual compound, or zone or class of compounds, can be cut to a 3D column.
A further comprehensive mode may be contrived, which maximises separation on the 2D column. Clearly, the short 2D column has limited ‘capacity’ so cannot separate many compounds. Using a long 2D column is incompatible with a fast PM setting. There are two possible scenarios for operation with a longer 2D column.  (1) Use a slow PM setting, and a longer 2D column, whilst ensuring that wraparound does not occur (i.e. each modulation is completed on 2D before the next sampling). Thus a PM of 1 min and 2D narrow bore column of 7 m length might be suggested. (2) Perform multiple injections, and take a narrow heart-cut for each injection. Cryofocus the heart-cut, cool the oven and elute the heart-cut as a second analysis on a long, high efficiency column.  Repeat the process but shift the heart-cut to take the next fraction, to eventually analyse the total sample. This has been recently demonstrated. [12]  Re-constructing all the 2D data presents a 2D separation space. This can be called comprehensive, since the total sample is subjected to analysis, but now with multiple injections, and much greater 2D separation.

Example applications for petrochemical
samples Classical GC×GC analysis

Fig 2A is a presentation of a crude oil sample, Late Cretaceous, classified as low-level biodegraded material. A column set of a P 1D (50% phenyl) phase and NP 2D (5% phenyl) shows that nonpolar compounds are more retained (i.e. elute at greater 2tR) than polar compounds. Choices for NP/P and P/NP phases for GC×GC have been discussed. [13] Fig 2A is a reconstruction of multiple individual modulations, one of which is shown in Fig 2B. Since this arises for a brief 6 s sampling of the 1D elution, all the compounds displayed in Fig 2B essentially overlap, before the 2D column provides additional separation. This is the rationale and justification for the GC×GC technique. Unresolved compounds can now be effectively separated on the 2D column, here in order of decreasing polarity, allowing facile quantification. Importantly, mass spectrometry will now give a relatively ‘pure’ mass spectrum of the separated compounds, as well as the total number of compounds at any given 1D position.
Each modulation may be interrogated for molecular information for the resolved compounds. The spectra of separated compounds 1. -6 are presented in Fig 2, and a few points can be made. First the individual spectra may be matched with their proposed identities using a MS database, subject to specificity. The spectra are of sufficient quality, without suffering matrix interference that arises on the 1D column. For instance, spectra for peaks 1 and 2 are very well resolved and not cross-contaminated. The suggested component identities are indicative of the various classes, but in the absence of authentic standards, absolute structures have not been confirmed.

Multidimensional GC and
new approaches to MDGC

Fig 3 shows results for a representative 15 s heart-cut of a sulfur-rich oil shale sample; both the original chromatogram in red (see column A1, Fig 1) and the 15 s heart-cut in black are illustrated in Fig 3A.[14] Only the 15 s heart-cut analytes are transferred to the 2D column (e.g. column B1 Fig 1), and all other analytes are recorded at a mid-point detector (FID Fig 1). Classically, a microfluidic Deans switch is used for the transfer process. At least 10 components are reported in the 15 s heart-cut, as shown in Fig 3B. However many more could be fitted into the 2D column ‘space’ (‘peak capacity’) between the first (most polar) component, to 10 (the least polar), maybe more than 40.  A few points are relevant here. First, the minor components shown on the 2D column are very well resolved, have well defined mass spectra, and the chemical class can be reasonably well classified. Second, the minor components are as little as 0.1% abundance of the major components here, and in all likelihood, without this additional separation, would be unlikely to be detected or identified in a 1D separation. S-, O- and aromatic compounds constitute the first 7 compounds here.
Third, the operation of the heart-cut/modulation/2D column process needs explanation. The process here functions as follows: the 15 s heart-cut is transferred to a cryogenic trap device, which focusses the heart-cut to a narrow band, so dispersion on the 1D column is essentially negated. The cryotrap rapidly (instantaneously) remobilises the heart-cut band to the 2D column, hence maximising its efficiency. The 2D column is operated according to the prevailing oven temperature program, so is ‘on-the-fly’ as it were. This suggests a range of possible method variations. For example, cooling the oven prior to releasing the heart-cut will increase the 2D peak capacity, which is the classical MDGC approach.  Using a shorter 2D column will elute the heart-cut components much faster, with some loss of peak capacity. It is possible to adjust the heart-cut time and the 2D column length so that a true ‘comprehensive’ 2D analysis is possible, taking contiguous heart-cuts e.g. every 1 min, and eluting each on the 2D column before the next is sampled into the 2D column.    

Future perspectives

The higher dimensional separations afforded by GC×GC and MDGC offer much improved separations along with additional advantages such as structured separations that group compounds into predictable retention positions; all these allow significantly better sample characterisation. These techniques are reliable, reproducible, and combined with mass spectrometry provide information-rich data on identity – as well as searchable archived data for future reference.  As protocols are established across a broader suite of applications, a reference base that will further validate these approaches for petrochemical-related samples will ensue.

Cited Literature.

[1] J. Blomberg, P. Schoenmakers, J. Beens, R. Tijssen, Comprehensive two-dimensional gas chromatography (GCxGC) and its applicability to the characterization of complex (petrochemical) mixtures, J. High Resol. Chromatogr. 20 (1997) 539-544.
[2] Z. Liu, J.B. Phillips, Comprehensive two-dimensional gas chromatography using an on-column thermal modulator interface, J. Chromatogr. Sci. 29 (1991) 227-231.
[3] D.G. Patterson, Jr, J.R. Barr, E.S. DiPietro, J. Grainger, V.E. Green, C.R. Lapeza, Jr. , V.L. Maggio, P.C. McLure, S. Sirimanne, W.E. Turner, Comprehensive two-dimensional and fast single-dimensional GC-HRMS analysis of human serum for PCDDs, PCDFs, cPCBs, congener PCBs, persistent and nonpersistent pesticides, and PAHs, Organohalogen Compd. 27 (1996) 309-314.
[4] M. Adahchour, J. Beens, R.J.J. Vreuls, U.A.T. Brinkman, Recent developments in comprehensive two-dimensional gas chromatography (GC × GC) I. Introduction and instrumental set-up, TrAC, Trends Anal. Chem. 25 (2006) 438-454.
[5] Y. Nolvachai, C. Kulsing, K.M. Sharif, Y.F. Wong, S.-T. Chin, B. Mitrevski, P.J. Marriott, Multi-column trajectory to advanced methods in comprehensive two-dimensional gas chromatography, TrAC Trends Anal. Chem. 106 (2018) 11-20.
[6] K.M. Sharif, S.-T. Chin, C. Kulsing, P.J. Marriott, The Microfluidic Deans switch: 50 years of progress, innovation and application, TrAC Trends Anal. Chem. 82 (2016) 35-54.
[7] W. Khummueng, J. Harynuk, P.J. Marriott, Modulation ratio in comprehensive two-dimensional gas chromatography, Anal. Chem. 78 (2006) 4578-4587.
[8] P.J. Marriott, S.T. Chin, B. Maikhunthod, H.G. Schmarr, S. Bieri, Multidimensional gas chromatography, TrAC Trends Anal. Chem. 34 (2012) 1-20.
[9] N.E. Watson, H.D. Bahaghighat, K. Cui, R.E. Synovec, Comprehensive three-dimensional gas chromatography with time-of-flight mass spectrometry, Anal. Chem. 89 (2017) 1793-1800.
[10] K.M. Sharif, C. Kulsing, P.J. Marriott, Pressure tuning of first dimension columns in comprehensive two-dimensional gas chromatography, Anal. Chem. 88 (2016) 9087−9094.
[11] B. Mitrevski, P.J. Marriott, Novel hybrid comprehensive two-dimensional – multi-dimensional gas chromatography for precise, high resolution characterisation of multicomponent samples, Anal. Chem. 84 (2012) 4837−4843.
[12] P.J. Marriott, S.-T. Chin, Y. Nolvachai, Techniques and application in comprehensive multidimensional gas chromatography –mass spectrometry, J. Chromatogr. A 1636 (2021) 461788.
[13] T.C. Tran, G.A. Logan, E. Grosjean, J. Harynuk, D. Ryan, P. Marriott, Comparison of column phase configurations for comprehensive two dimensional gas chromatographic analysis of crude oil and bitumen, Org. Geochem. 37 (2006) 1190-1194.
[14] M.W. Amer, B. Mitrevski, W. Roy Jackson, A.L. Chaffee, P.J. Marriott, Multidimensional and comprehensive two-dimensional gas chromatography of dichloromethane soluble products from a high sulfur Jordanian oil shale, Talanta 120 (2014) 55-63.