This work builds upon recent developments in the field of second dimensional (2D) retention indices (RI) for use in comprehensive two-dimensional gas chromatography (GC×GC), expanding application to the most commonly used “normal” orthogonality column configuration, where 2D RI are rarely employed.
Journal of Chromatography A 1683 (2022) 463548 Contents lists available at ScienceDirect Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma Second dimension retention indices in “normal” orthogonality comprehensive two-dimensional gas chromatography using single standard injection generated isovolatility curves Jason Devers∗ , David I Pattison, Jan H Christensen Analytical Chemistry Group, Department of Plant and Environmental Science, Faculty of Science, University of Copenhagen, 1871 Frederiksberg C, Denmark a r t i c l e i n f o Article history: Received April 2022 Revised 29 September 2022 Accepted 29 September 2022 Available online October 2022 Keywords: Comprehensive two-dimensional gas chromatography Retention index Non-targeted analysis Wastewater Identification a b s t r a c t This work builds upon recent developments in the field of second dimensional (2D) retention indices (RI) for use in comprehensive two-dimensional gas chromatography (GC×GC), expanding application to the most commonly used “normal” orthogonality column configuration, where 2D RI are rarely employed Initially, one dimensional retention indices for 80 wastewater pollutants were determined by GC-MS on a mid-polar ZB-50 column In order to determine the 2D RIs for peaks detected in wastewater extracts separated by GC×GC -MS, a single injection of a ten-compound standard mix allowed the construction of model-generated isovolatility curves These curves were used for the determination of 2D RIs of compounds initially identified on the basis of the mass spectral match factor and 1D RIs Good agreements (average deviation of 1.7%) were observed between the calculated 2D RIs and the measured reference RIs for these compounds These results show that this approach provides an additional level of confidence for the identification of compounds detected in GC×GC-MS and demonstrates the potential of this approach for improved compound identification in non-targeted analysis © 2022 The Author(s) Published by Elsevier B.V This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) Introduction GC×GC has proven to be a powerful tool in the non-targeted analysis of complex samples The multiplicative peak capacity and separation via two independent separation mechanisms in GC×GC allows for the deconvolution, separation and therefore identification of peaks which may otherwise coelute using conventional one-dimensional (1D) GC [1] Retention indices (RI) are commonly used in 1D GC to provide an additional level of confidence in the identification of unknowns in addition to mass spectral information, [2,3] or as a primary identification tool in the absence of mass spectral information [4] The two-column setup of GC×GC offers the opportunity of using RIs in both the 1st and 2nd dimensions, allowing for the prediction of a compound’s elution position within the 2D space This additional identification criterion can increase confidence in detection by reducing the incidence of falsepositives in the non-targeted analysis of complex samples, such Abbreviations: GC×GC, Comprehensive two-dimensional gas chromatography; QTOF, Quadropole time-of-flight; RI, Retention Index; SPE, Solid phase extraction ∗ Corresponding author at: Analytical Chemistry, Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, Frederiksberg C, Copenhagen 1871, Denmark E-mail address: jpd@plen.ku.dk (J Devers) as wastewater [2,5] However, while in 1D, RI is calculated based on the relationship between an analyte’s retention time and the retention times of reference compounds (often n-alkanes), [6] using the same approach for 2D RI is not applicable This is a result of the discrete nature of each 2D modulation, in which each second dimension separation occurs over only a few seconds as defined by the modulation time, and thus analytes are eluted under pseudo isothermal conditions [7] Recent solutions have included the creation of 2D RI surfaces via the repeated injection of standards using several oven temperature ramp rates [8], and more commonly, via the creation of isovolatility curves These curves are created by plotting the 2D retention time of a series of standards, when their 2D separation occurs at different 1D retention times/pseudo isothermal separations [9] Based on these generated curves, it is then possible to determine the 2D RI of an analyte based on its 2D retention time relative to the most closely eluting isovolatility curves Generation of these isovolatility curves typically involves the continuous, [10] or repeated injection of a solution of n-alkanes, [11] fatty acid methyl esters (FAMES), or a combination thereof [12] Recent studies have demonstrated that the creation of these curves can be completed using a single injection of a standard mix whose retention times span the 2D separational space, which eliminates the need to regularly run complex continuous or multiple https://doi.org/10.1016/j.chroma.2022.463548 0021-9673/© 2022 The Author(s) Published by Elsevier B.V This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/) J Devers, D.I Pattison and J.H Christensen Journal of Chromatography A 1683 (2022) 463548 injections [13] Thus far, most of the research published on calculation of 2D RIs has been carried out using a “reverse” orthogonality column setup, where a more polar column is used in the 1st dimension and a non-polar column in the 2nd dimension [14] In this case, the 2D RIs are based on separation on the non-polar column, which gives two major advantages Firstly, there is a large database of reference RI values measured on non-polar columns for a wide variety of compounds, approximately 10 0,0 0 values in the NIST database, [15] and secondly, there is a lower observed retention index deviation and superior chromatographic reproducibility for non-polar columns compared to more polar columns [16] However, in practice, a recent literature review shows that the vast majority of GC×GC applications employ a “normal” orthogonality setup, [17] i.e using a non-polar column in the 1st dimension and a more polar column in the 2nd dimension, thus there is a need to increase the database of RIs on (mid-)polar columns, to provide a method for more confident compound identification in GC×GC-MS using a “normal” column orientation This study utilises a recently established method by Jiang et al 2019 that generates isovolatility curves using a single injection of a standard mix [13] In this case, the approach is applied to a “normal” orthogonality column setup, which is a more popular GC×GC configuration for non-targeted analysis of wastewater extracts The applicability and performance in this scenario have been assessed by comparing reference RIs measured for standards by conventional GC-MS on a mid-polar column, with the 2D RIs determined by the generated isovolatility curves, to provide confident identification of compounds in wastewater extracts ms Ionization was carried out using an electron ionisation source, operated at 230°C with an emission current of 35 μA and electron energy of 70 eV The data for calculating (see Section 2.4) the reference RIs for the mid-polar column were acquired by conventional 1D GC-MS, using an Agilent 6890N GC coupled to an Agilent 5975B single quadropole MS Separation was carried out using a ZB-50 column (60m, 0.18 mm i.d., 0.18 μm) Samples and standards of 1.0 μL were injected splitless at an inlet temperature of 280°C The primary oven temperature followed the program: initially 60°C for minute, ramped to 320°C at a rate of 3°C/min, with a final hold of five minutes, giving a total run time of 93 minutes Helium was used as the carrier gas with a constant flow rate of 1.0 mL/min 2.3 Standard and Sample Preparation Wastewater extracts were prepared using a multi-layer solid phase extraction (ml-SPE) technique previously described by Mechelke et al 2019 and adapted by Tisler et al.2021 [18,19] In summary, wastewater samples were adjusted to pH 6.5 and filtered using a Büchner vacuum filtration apparatus employing Whatman 1820-090 glass microfiber filters (Whatman, UK) Automated SPE was then executed using a Promochrom SPE-03 system (Promochrom, Richmond, Canada) and ml-SPE cartridges comprised of SupelcleanTM ENVI-Carb SPE cartridges (Sigma-Aldrich), Oasis HLB, WAX and MAX sorbent SPE cartridges (Waters), resulting in 500 times enriched samples in methanol The ten-compound mixture used for generating isovolatilty curves was prepared as described in table S-1 within the supporting information The compounds were chosen based on preliminary non-targeted analysis of wastewater extracts, with compounds spanning as much of the two-dimensional space as possible, which allowed the isovolatility curves to be defined by data points across the whole 2D separational space Experimental Section 2.1 Materials and Chemicals An influent wastewater sample was sourced from a local wastewater treatment plant (BIOFOS A/S, Avedøre, Denmark) LCMS grade methanol was sourced from Honeywell (Honeywell: 34966-2.5 L, Germany) Standard mixtures used for the generation of isovolatility curves in GC×GC-MS and determination of reference RI by GC-MS were prepared in methanol, with the supplier and solution information detailed in the supporting information for the relevant solutions (Tables S-1 and S-2) 2.4 Data Processing GC×GC-MS instrument data acquisition was performed via Mass Hunter (Agilent, version: B.07.02.1938), and the interpretation of the resultant GC×GC-MS chromatograms using GC Image (GC Image LLC, version: 2.9R1.1) In-house reference RIs were generated using Automatic Mass Spectral Deconvolution and Identification System (National Institute of Standards and Technology, Build: 149.31), automatically calculated based on the retention time of the compound of interest, relative to the retention times of separately injected n-alkane standards, using the linear retention indices relationship established by Dool and Kratz [6] In addition, reference RIs were also sourced using the NIST_RI function contained within Webchem, an R (R Foundation, version: 3.6.2) package used to retrieve chemical information from online databases [20] 2.2 Instrumentation GC×GC-MS analysis was conducted using an Agilent 7890B GC system coupled to an Agilent 7200 Accurate Mass QTOF mass spectrometer (Agilent Technologies, Palo Alto, CA, USA) The instrument was adapted for two-dimensional separation using a secondary column oven paired with a Zoex ZX2 cryogen free looptype modulator (Zoex Corporation, Houston, TX, USA) Separation was carried out using a “normal” orthogonality column set, comprising of a non-polar ZB-5 column in the first dimension (60 m, 0.25 mm i.d., 25μm), and a mid-polar ZB-50 column (1.5 m, 0.18 mm i.d., 0.18 μm) in the second dimension (Phenomenex, Torrance, CA, USA) SilTite μ-union ferrules (SGE Analytical Science, Wetherill Park, Australia) were used to connect the columns Samples and standards of 1.0 μL were injected using a 10:1 split ratio at an inlet temperature of 280°C The primary oven temperature followed the program: initially 60°C for minute, ramped to 315°C at a rate of 3°C/min, with a final hold of ten minutes, giving a total run time of 96 minutes Secondary oven and hot jet temperature were operated at a constant temperature offset of + 60°C from the primary oven, with the column temperature plateauing at 315°C and the hot jet at 360°C Helium was used as the carrier gas with a constant flow rate of 2.0 mL/min The modulation period was set to 80 0 ms, with a hot jet on-time of 850 2.5 Generation of isovolatility curves using single injection of standard mix Log(log(2 tR )) = c I 100 + d ln T + e I 100 +f (1) Equation (1) was developed by Jiang et al 2019, [13] and was used here to establish the relationship between the observed 2D retention time (2 tR ), reference retention index (2 I) and corresponding 2D elution temperature (2 T, determined by the 2D column oven temperature corresponding to the peak’s 1D elution time) of the reference compounds Using the relevant variables detailed above for each standard, yielded from a single injection of the standard J Devers, D.I Pattison and J.H Christensen Journal of Chromatography A 1683 (2022) 463548 Fig (a) GC×GC-MS chromatogram of a single injection of the RI standard mix, with each of the ten standards labelled; (b) n-alkane isovolatility curves generated via equation (1) using the tR , I and T of the standards in the chromatogram mix, the equation was solved to determine the coefficients c, d, e and f, using the COIN-OR CBC linear solver contained within the OpenSolver plugin of Microsoft Excel The solved coefficients were then substituted into equation (1), in conjunction with the known variables tR , T to calculate the experimental RI (2 I) for each analyte of interest expected due to the lack of equivalence between the wax-type columns and a phenyl siloxane ZB-50 column This observation highlighted the need to determine more relevant RIs for both our standards and compounds of interest on the column type used in our studies Thus, standards were prepared for the in-house determination of reference RIs for 80 compounds that were of interest in wastewater samples using the mid-polar ZB-50 column via 1D-GC-MS The resulting RIs for these 80 standards are given in the supporting information (Table S-2) The use of these accurate reference RIs for our ten-compound standard mix allowed the constants in equation (1) to be solved (c=-0.12, d=-0.19, e= 0.78 and f=1.36),and subsequently construction of hypothetical nalkane isovolatility curves for our GC×GC conditions, as illustrated in Fig (b) Results and Discussion In order to test the creation of standard mix generated isovolatility curves as described by Jiang et al [13] but using a “normal” orthogonality column setup, the ten-compound standard mix was analysed using GC×GC-MS, as shown in Fig (a) The standard mix was run before and after samples to ensure that no significant drift in tR occurred within the sequence, and a drift of 100 The compound identification for each peak in the blob J Devers, D.I Pattison and J.H Christensen Journal of Chromatography A 1683 (2022) 463548 Fig GC×GC-MS chromatogram of influent wastewater ml-SPE extract, with identified compounds (n=20) labelled and highlighted in light blue The absolute (2D RI) values range from - 81, with 14 out of 20 having (2D RI)