Retention Behaviour of Alkylated and Non-Alkylated Polycyclic Aromatic Hydrocarbons on Different Types of Stationary Phases in Gas Chromatography

The gas chromatographic retention behaviour of 16 polycyclic aromatic hydrocarbons (PAHs) and alkylated PAHs on a new ionic liquid stationary phase, 1,12-di(tripropylphosphonium) dodecane bis(trifluoromethanesulfonyl)imide (SLB®-ILPAH) intended for the separation of PAH mixtures, was compared with the elution pattern on more traditional stationary phases: a non-polar phenyl arylene (DB-5ms) and a semi-polar 50% phenyl dimethyl siloxane (SLB PAHms) column. All columns were tested by injections of working solutions containing 20 parental PAHs from molecular weight of 128 to 278 g/mol and 48 alkylated PAHs from molecular weight of 142 to 280 g/mol on a one dimensional gas chromatography-mass spectrometry (GC-MS) system. The SLB PAHms column allowed separation of most isomers. The SLB®-ILPAH column showed a rather different retention pattern compared to the other two columns and, therefore, provided a potential for use in comprehensive two-dimensional GC (GC×GC). The ionic liquid column and the 50% phenyl column showed good thermal stability with a low bleed profile, even lower than that of the phenyl arylene “low bleed” column.


Introduction
Ubiquitously present in the environment, polycyclic aromatic hydrocarbons (PAHs) originate from natural and anthropogenic incomplete combustion processes.They are present in air, food, water and soil.Nowadays, the PAHs originating from anthropogenic activities are unarguably predominant compared to those originating from natural sources.Humans are exposed to PAHs in almost every aspect of everyday life and, therefore, PAHs are among the most studied chemicals.During the last 50 years, the procedures for the determination of individual PAHs in complex environmental mixtures have been extensively developed and improved.In 1976, 16 specific PAHs were selected for regulation by the United States Environmental Protection Agency (U.S. EPA); the historical perspectives regarding the choice of these 16 EPA PAHs can be found in an article by Keith [1].
In 2002, the toxicities of 33 PAHs were assessed by The European Scientific Committee on Food and 15 PAHs showed clear evidence of mutagenicity/genotoxicity. Fourteen of these 15 PAHs showed clear carcinogenic effects in various types of bioassays and in experimental animals [2].Seven of these carcinogenic PAHs in the Scientific Committee on Food study are also contained in the EPA's set of 16 PAHs, while the additional seven are: benzo(j)fluoranthene, cyclopenta(cd)pyrene, dibenzo(a,e)-, dibenzo(a,h)-, dibenzo(a,i)-, dibenzo(a,l)pyrene and 5-methylchrysene.In 2006, the Joint FAO/WHO Expert Committee on Food Additives (JECFA) concluded that benzo(c)fluorene is probably also carcinogenic [3].This shows that the list of the toxic and environmentally relevant PAHs is still growing.
In non-occupational settings, food is the main source of human exposure to PAHs, followed by cigarette smoke, which in some cases may result in PAH exposure on par with the food uptake route [4,5].Other important exposure routes include traffic related air pollution and all kinds of occupational exposures.Nonetheless, the new possible exposure pathways are still being identified: e.g.synthetic turf materials used on football fields [6].
The analysis of PAHs is generally based on gas chromatography (GC) rather than on liquid chromatography (LC) because GC allows greater selectivity, resolution and sensitivity than LC [7,8].The GC systems are commonly coupled with flame ionisation detectors (FID) or mass-spectrometric detectors (MS).The GC analysis was conventionally based on non-polar stationary phases operated at relatively high temperatures [8,9].The 5% phenyl methylpolysiloxane phase (like in the DB-5 column) is still the most often applied one in PAHs analysis and it has also been recommended in a number of US-EPA methods, e.g.US EPA method 610 [10].Since the 1990s, high phenyl content stationary phases have been more frequently used, e.g.described by the producers as "50% phenyl methylpolysiloxane-like" DB-17MS [8,11], Rxi-PAH [12] or SLB PAHms [13].
Some years ago, a new group of stationary phases, based on non-bonded ionic liquids (IL) was introduced [14,15].Based on non-molecular solvents with low melting points, these stationary phases consist of organic cations plus inorganic or organic anions [16] and, therefore, the IL columns enable chromatographic separation based on a selectivity different to that provided by conventional stationary phases.Some IL columns can exhibit "dual nature" features; they allow separation of non-polar molecules as non-polar stationary phases do, while at the same time they have a high affinity for polar molecules like polyethylene glycol (wax) and cyanopropyl-siloxane stationary phases.The IL columns are more polar than the wax columns but they have higher thermal stability compared to traditional siloxane phases with a similar selectivity because they are not susceptible to back-biting reactions that result in phase degradation and column bleed [14].Siloxane-based stationary phases contain active hydroxyl groups at the terminal positions; this makes them sensitive to the oxygen catalyzed cleavage of backbone siloxane.The siloxane chain then breaks to volatile cyclic siloxanes that elute from the column as "bleed" and results in a rising baseline.
So far, the chromatographic properties of the IL columns have only been investigated in a few studies.The IL columns have been used for the separation of different classes of environmental pollutants, like polybrominated diphenyl ethers (PBDEs), polychlorinated biphenyls (PCBs) and other chlorinated compounds [17,18], alkyl phosphates, fatty acids, and petroleum distillates [19].A new IL column, SLB ® -ILPAH, intended for the separation of PAHs mixtures, recently became commercially available.This column has already been tested in terms of the retention behaviour of alkyl-substituted polycyclic aromatic sulphur heterocyclic isomers [13].
In this study, we investigated the retention behaviour of PAHs and alkylated PAHs on the SLB ® -ILPAH column and two stationary phases traditionally used for the PAH analysis: a low bleed column with a phenyl arylene polymer that is virtually equivalent to a (5%-phenyl)-methylpolysiloxane (DB-5ms) and a high phenyl content column denoted as 50% phenyl-dimethylpolysiloxane (SLB PAHms).The difference between the arylene column and a 5% phenyl-dimethylsiloxane column is that in the arylene column, the phenyl ring is built in the siloxane chain, whereas in the phenyl-dimethylsiloxane phase, the phenyl rings are positioned as substituents (side chains).Alkylated PAHs were selected because numerous isomers of these compounds are currently targeted in analyses of environmental samples.
Alkylated PAHs are recognised as environmental pollutants although they are still not regularly included in the analysis of priority PACs (e.g.16 EPA PAHs).They are ubiquitously present in the environment and are often more toxic than the parental PAHs [20,21].Alkylated PAHs have been found in the toxic fractions in several Effect Directed Analysis (EDA) studies [22][23][24][25].5-methylchrysene, 1-methylpyrene and 7,12-dimethylbenz(a)anthracene, as confirmed toxic compounds, are being included more and more in standard PAH analyses [26,27].A list of 34 PAHs (18 parental PAHs and 16 alkylated), has been recommended for toxicological screenings by the US EPA [28].In addition to the 16 traditional EPA PAHs, the list of 34 PAHs includes perylene, benzo(e)pyrene and 16 groups of C-1 to C-4 alkyl derivatives.
The determination of alkylated PAHs in complex environmental samples is problematic because of numerous coeluting isomers [19,29].It is not possible to separate all isomers of heavier PAHs in a single chromatographic run on one column but two-dimensional GC-MS analysis (GC×GC-MS) could offer a solution.GC×GC can only be fruitful if the two columns used in series are (semi-) orthogonal, or, as chemically different from each other as possible.Therefore, an assessment of new and different stationary phases with different separation mechanisms was needed.

Materials and Methods
Table 1 shows the characteristics of the three stationary phases that were tested in this study.The columns SLB PAHms and SLB ® -ILPAH (both from Supelco, Bellefonte, Pennsylvania, USA) were made available by Sigma Aldrich (Zwijndrecht, The Netherlands) and the DB-5ms was bought from Agilent, The Netherlands.The parental and alkylated-PAHs standard solutions and pure compounds (Table 2) were purchased from Sigma Aldrich (Zwijndrecht, The Netherlands).All solvents used (isooctane and toluene) were obtained in picograde quality from Merck Millipore (Amsterdam, The Netherlands).All standards were gravimetrically prepared in toluene and isooctane.The working solutions were prepared by mixing appropriate volumes from the individual stock solutions.Analyses were performed on an Agilent 6890 gas chromatograph coupled to an Agilent 5975C inert MSD with a Triple-Axis Detector.All injections were performed in the splitless mode (1 µL; splitless time 1.4 min) at 275 • C and with MS operating in total ion current mode.The oven temperature programs were set as follows: DB-5ms and SLB-PAH: isothermal at 90 • C for 10 min, then with 5 • C/min to 300 • C, SLB ® -ILPAH: isothermal at 90 • C for 6 min, then with 5 • C/min to 300 • C.
The temperature programs were optimised in order to compare the elution order and peak resolution between the columns.
The SLB ® -ILPAH is commercially available in dimensions different from the "standard" dimensions (Table 1) as discussed in the Results and discussion section.

Results and discussion
The retention times and the relative retention times to pyrene of all parental PAHs and alkylated PAHs injected on the three columns are presented in Table 2.This table also shows the coelutions of the isomers having similar mass spectra (see coloured cells).The coelutions of PAHs that are not isomers (e.g. the coelution of 7,12-dimethylbenz(a)anthracene and benzo(b)fluoranthene or 2-methylbenzo(c)phenanthrene and benz(a)anthracene on the SLB PAHms column) were not marked here because these compounds have different mass spectra and can be separated by the MS detector.However, the interferences of the fragment ions of the overlapping compounds with different base peak ions must be taken into account for accurate quantitation.
The elution order of the PAHs and the alkyl-PAHs on the phenyl arylene and the 50% phenyl-polysiloxane stationary phases is rather similar.However, the elution order on the SLB-ILPAH is different; these differences will be discussed below.The advantages and shortcomings of the three studied columns are briefly summarized in Table 3.The least polar column, phenyl arylene, shows an overlap of 19 isomers at more than 50% of the peak height and of 4 isomers at less than 50% of the peak height.Chrysene, one of the 16 EPA PAHs, coelutes with triphenylene but the rest of the 16 EPA PAHs are totally resolved.This column showed the best separation of dimethylnaphthalenes (Figure 1); 1,3-and 1,6-dimethylnaphthalenes were separated on this column only.Figure 1 shows that the dimethylnaphthalenes formed a co-eluting peaks' cluster on the ionic liquid column while on the siloxane-based columns they were much better separated.Figure 1 also shows that compared to the phenyl arylene column, the 50% phenyl-polysiloxane column shows a substantially better separation of the injected isomers.Table 2 shows that on this column only 13 isomers overlapped at more than 50% of the peak height and four isomers overlapped at less than 50% of the peak height.
Figure 2A shows that chrysene and triphenylene were partly separated on the 50% phenyl-polysiloxane column while they coeluted at the phenyl arylene column.The separation of these isomers is comparable to the separation achieved on the Rxi-PAH column (50% phenyl methylpolysiloxane-like phase) used for the PAHs analysis by Nalin et al. [12].The study of Poster et al. [8] showed that chrysene and triphenylene coelute on the comparable DB-17MS stationary phase (50% phenyl methyl-polysiloxane-like phase), are partly resolved on the non-polar DB-XLB column (proprietary phase) and totally resolved on the LC-50 column (dimethyl/50% liquid crystalline phase).Figure 2A shows that chrysene and triphenylene are totally resolved on the IL column.2.
Figure 2A shows that chrysene and triphenylene were partly separated on the 50% phenylpolysiloxane column while they coeluted at the phenyl arylene column.The separation of these isomers is comparable to the separation achieved on the Rxi-PAH column (50% phenyl methylpolysiloxane-like phase) used for the PAHs analysis by Nalin et al. [12].The study of Poster et al. [8] showed that chrysene and triphenylene coelute on the comparable DB-17MS stationary phase (50% phenyl methyl-polysiloxane-like phase), are partly resolved on the non-polar DB-XLB column (proprietary phase) and totally resolved on the LC-50 column (dimethyl/50% liquid crystalline phase).Figure 2A shows that chrysene and triphenylene are totally resolved on the IL column.
In Figure 2B we see that all isomers of methylated phenanthrenes and anthracenes were separated on the 50% phenyl-polysiloxane column, while some of these isomers coeluted on the phenyl arylene and on the SLB-ILPAH column.The IL column also demonstrated the different mechanism of retention; 4,5-methylenephenanthrene eluted before the methylated phenanthrenes and anthracenes (178-C1).Figure 3 shows that the best separation of 17 methylated benz(a)anthracenes, benzo(c)phenanthrenes and chrysenes isomers (228-C1) was achieved on the 50% phenylpolysiloxane column; only seven isomers coeluted at more than 90% of the peak height while the remaining 10 isomers were at least partly resolved (Table 2).This separation was better than the separation achieved on the phenyl arylene column, where 11 of these isomers coeluted, as well as on  2.
In Figure 2B we see that all isomers of methylated phenanthrenes and anthracenes were separated on the 50% phenyl-polysiloxane column, while some of these isomers coeluted on the phenyl arylene and on the SLB-ILPAH column.The IL column also demonstrated the different mechanism of retention; 4,5-methylenephenanthrene eluted before the methylated phenanthrenes and anthracenes (178-C1).
Figure 3 shows that the best separation of 17 methylated benz(a)anthracenes, benzo(c)phenanthrenes and chrysenes isomers (228-C1) was achieved on the 50% phenyl-polysiloxane column; only seven isomers coeluted at more than 90% of the peak height while the remaining 10 isomers were at least partly resolved (Table 2).This separation was better than the separation achieved on the phenyl arylene column, where 11 of these isomers coeluted, as well as on the SLB-ILPAH column, where 10 of these isomers coeluted.The number of the observed coelutions might be reduced by increasing the lengths of the tested columns, reducing the internal diameters and/or by improving the applied temperature programs with stable temperature periods around the elution times of isomeric clusters.1), 50% phenyl-polysiloxane (2) and SLB-ILPAH (3) stationary phases.For abbreviations see Table 2.
Figure 3 shows that the best separation of 17 methylated benz(a)anthracenes, benzo(c)phenanthrenes and chrysenes isomers (228-C1) was achieved on the 50% phenylpolysiloxane column; only seven isomers coeluted at more than 90% of the peak height while the remaining 10 isomers were at least partly resolved (Table 2).This separation was better than the separation achieved on the phenyl arylene column, where 11 of these isomers coeluted, as well as on the SLB-ILPAH column, where 10 of these isomers coeluted.The number of the observed coelutions might be reduced by increasing the lengths of the tested columns, reducing the internal diameters and/or by improving the applied temperature programs with stable temperature periods around the elution times of isomeric clusters.2.
The commercially available SLB-ILPAH column was 2/3 the length of the two other columns.The internal diameter was 3/4 of that of the other two and the film thickness 1/5 of that of the two other columns, which made the separation substantially faster.However, it is not possible to compare the dimensions of the IL column to the siloxane-based columns directly because of the different nature of an IL coating resulting in the different type of interactions between the analytes and the  The commercially available SLB-ILPAH column was 2/3 the length of the two other columns.The internal diameter was 3/4 of that of the other two and the film thickness 1/5 of that of the two other columns, which made the separation substantially faster.However, it is not possible to compare the dimensions of the IL column to the siloxane-based columns directly because of the different nature of an IL coating resulting in the different type of interactions between the analytes and the stationary phase.This IL phase shows stronger retention for heavier PAHs (Table 2); the relative retention times of the heavier PAHs on this column are higher than on the phenyl arylene and the 50% phenyl-polysiloxane columns.The SLB-ILPAH phase also showed some interesting elution shifts: 1,12-dimethylbenzo(c)phenanthrene (228-C2) eluted before benz(a)anthracene and other PAHs with MWs of 228 g/mol and 1-methylbenzo(c)phenanthrene (228-C1) eluted before 1-methylpyrene (202-C1).Also, the elution order of four PAHs from the 16 EPA PAHs-group on this IL column is different compared to the elution on the two siloxane-based columns: acenaphthylene elutes before acenaphthene and dibenz(a,h)anthracene elutes before indeno(1,2,3-cd)perylene on the SLB-ILPAH column.However, the overall separation of the isomers on the SLB-ILPAH phase is not as good as on the other two phases: 22 isomers overlap at more than 50% of the peak height.A huge advantage of this column is the total separation of chrysene from triphenylene (Figure 2A).Yet, Figure 4 shows that the highly carcinogenic benzo(a)pyrene, another PAH belonging to the group of the 16 EPA PAHs, coeluted with benzo(e)pyrene.Both isomers are separated on the phenyl-siloxane column, while benzo(a)pyrene coelutes with 8,9,11-trimethylbenz(a)anthracene.Priority toxicant 5-methylchrysene was totally separated on this column while on the other two columns it could not be totally resolved from other isomers (Figure 3).The SLB-ILPAH column also managed to separate 1-methylbenz(a)anthracene and 4-methylchrysene; these isomers (partially) coelute on the other two columns.It is plausible that increasing the length of this column to 30 m may somewhat improve the observed coelutions, but it is unlikely the pattern would improve so much that it would equal that of the other two columns.2.
Very little tailing was observed and the peak shapes obtained on all three columns were satisfactory (Figures 1-4).The variation in response obtained on the three columns was relatively small.
Figure 5 shows the column bleed of the three phases: the bleeding of the 50% phenylpolysiloxane and the SLB-ILPAH phases were comparable and several times lower than the bleeding of the phenyl arylene "low bleed" stationary phase.Overlap of 3-methylbenz(a)anthracene with 5-methylbenz(a)anthracene and 4methylbenz(a)anthracene with 6-methylbenz(a)anthracene was observed on all three columns (Figure 3).It is worth noting that these isomers could not be separated by GC×GC-MS with different column combinations either [29].The DB-5 (60 m)×LC-50 (1.2 m) column combination tested by Skoczynska et al. [29] in the analysis of the 228-C1 methylated PAHs was able to separate in the second dimension 7-methylbenz(a)anthracene from 9-methylbenz(a)anthracene isomers, two compounds that coelute on the and the 50% phenyl-polysiloxane.Significant differences in selectivity between the LC-50 and the Rxi-PAH (50% phenyl comparable to the SLB PAHms phase) were shown in the study of Nalin et al. [12].The elution pattern of methylchrysenes (228-C1) and methylbenzo(a)pyrenes (252-C1) obtained on Rxi-PAH by Nalin et al. is similar to the pattern obtained on the 50% phenyl-polysiloxane in this study (even though Nalin et al. analysed more isomers).Coupling of LC-50 in the second dimension with 50% phenyl-polysiloxane in the first dimension could, therefore, result in orthogonal separation of the coeluting isomers (e.g.7-methylbenz(a)anthracene from 9-methylbenz(a)anthracene).The SLB-ILPAH shows the strongest deviation in the retention pattern due to a different type of interactions between the analytes and the stationary phase than in the other two columns studied.Therefore, using this column together with 50% phenyl-polysiloxane may result in orthogonal separation of different PAHs isomers in one GC×GC run.Because of the "dual nature" of the IL columns, the coupling of a "standard" 50%-phenyl polysiloxane column with an IL column in a GC×GC analysis will almost certainly result in an improved separation of the PAHs isomers; a follow-up study may include the evaluation of ionic liquid stationary phases with different polarity coupled to a 50% phenyl-polysiloxane column.
Very little tailing was observed and the peak shapes obtained on all three columns were satisfactory (Figures 1-4).The variation in response obtained on the three columns was relatively small.
Figure 5 shows the column bleed of the three phases: the bleeding of the 50% phenyl-polysiloxane and the SLB-ILPAH phases were comparable and several times lower than the bleeding of the phenyl arylene "low bleed" stationary phase.

Conclusion
None of the three columns tested offers a complete separation of the injected PAH and methyl-PAH isomers.On the SLB-ILPAH column 22, isomers overlapped at more than 50% of the peak height.The phenyl arylene column showed an overlap of 19 isomers and the 50% phenylpolysiloxane phase of 13 isomers.Also, none of the columns was able to totally resolve all 16 EPA PAHs.The 50% phenyl-polysiloxane column showed the best overall resolving power and is, therefore, currently considered the best option for the PAH and methyl-PAH analysis.
However, the SLB-ILPAH column is interesting because of a strongly deviating elution pattern, which is due to the different type of interactions between the analytes and the stationary phase.That makes the ionic liquid column interesting for specific separations that cannot be obtained on one of the other two columns or possibly on other traditional phases.A huge advantage of the ionic liquid column is, for example, the total separation of chrysene from triphenylene.An additional advantage is that using this ionic liquid phase, together with e.g. the 50% phenyl-polysiloxane phase, may result in a (semi-)orthogonal separation of PAHs and methyl PAHs in one GC×GC run.

Conclusion
None of the three columns tested offers a complete separation of the injected PAH and methyl-PAH isomers.On the SLB-ILPAH column 22, isomers overlapped at more than 50% of the peak height.The phenyl arylene column showed an overlap of 19 isomers and the 50% phenyl-polysiloxane phase of 13 isomers.Also, none of the columns was able to totally resolve all 16 EPA PAHs.The 50% phenyl-polysiloxane column showed the best overall resolving power and is, therefore, currently considered the best option for the PAH and methyl-PAH analysis.
However, the SLB-ILPAH column is interesting because of a strongly deviating elution pattern, which is due to the different type of interactions between the analytes and the stationary phase.
That makes the liquid column interesting for specific separations that cannot be obtained on one of the other two columns or possibly on other traditional phases.A huge advantage of the ionic liquid column is, for example, the total separation of chrysene from triphenylene.An additional advantage is that using this ionic liquid phase, together with e.g. the 50% phenyl-polysiloxane phase, may result in a (semi-)orthogonal separation of PAHs and methyl PAHs in one GC×GC run.
The ionic liquid SLB-ILPAH column and the high phenyl content 50% phenyl-polysiloxane column both show better thermal stability with less bleeding compared to that of the phenyl arylene "low bleed" column.This low bleeding is an asset for GC×GC because often, more polar columns are used, which show higher bleeding.

Table 1 .
Stationary phases and their characteristics.