Retention behavior of isomeric polycyclic aromatic sulfur heterocycles in reversed-phase liquid chromatography

Abstract

Retention indices for 70 polycyclic aromatic sulfur heterocycles (PASHs) were determined using reversed-phase liquid chromatography (LC) on a monomeric and a polymeric C₁₈ stationary phase. Molecular shape parameters [length, breadth, thickness (T), and length-to-breadth ratio (L/B)] were calculated for all the compounds studied. Correlations between the retention on the polymeric C₁₈ phase and PASH geometry (L/B and T) were investigated for six specific PASH isomer groups with molecular mass (MM) 184 Da, 234 Da, 258 Da, 284 Da, 334 Da, and 384 Da. Similar to previous studies for polycyclic aromatic hydrocarbons (PAHs), PASH elution order on the polymeric C₁₈ phase was generally found to follow increasing L/B values. Correlation coefficients for retention vs L/B ranged from r = 0.45 (MM 184 Da) to r = 0.89 (MM 284 Da). In the case of smaller PASHs (MM ≤ 234 Da), the location of the sulfur atom in the bay-region of the structure resulted in later than expected elution of these isomers based on L/B. In the case of the larger PASHs (MM ≥ 284 Da), nonplanarity had a significant influence on earlier than predicted elution based on L/B values.

1. Introduction

Polycyclic aromatic compounds (PACs) comprise a complex class of condensed multi-ring benzenoid compounds originating from a wide variety of natural and anthropogenic sources. The parent homocyclic species, which contain only carbon and hydrogen, are the familiar polycyclic aromatic hydrocarbons (PAHs). In recent decades, PAHs are one of the most studied groups of environmental contaminants because of their high carcinogenic and mutagenic potential [1–4]. However, their sulfur analogues, polycyclic aromatic sulfur heterocycles (PASHs), have received far less attention although their presence has been confirmed in air particulate matter [5, 6], sediments [7–9], coal liquids [10], diesel [9], heavy oil [11], crude oil [9, 11–17], shale oil [10, 12, 15, 18], coal tar [12–15, 19–21], mussels [7, 8], and fish [7, 8]. In addition, their carcinogenic and mutagenic potential have been reported [22, 23]. PASHs exist in an even greater variety of structures compared to PAHs due to the presence of the sulfur atom. Therefore, the number of isomers and alkylated isomers can be extremely large, which increases the difficulty to quantify individual PASH isomers in complex natural mixtures.

Reversed-phase liquid chromatography (LC) on octadecylsilane (C₁₈) stationary phases has been shown to provide excellent separations of isomeric PAHs [24–31]. However, not all C₁₈ stationary phases provide the same selectivity for PAHs. In past decades, several studies have compared different commercial C₁₈ columns for the separation of PAHs with particular emphasis on the separation of isomeric PAHs [28, 29]. These studies demonstrated that all stationary phases were C₁₈ but some provided significantly enhanced selectivity for the separation of isomeric PAHs. One of the studies focused on the importance of the synthesis of the bonded C₁₈ phase. Typically, monomeric C₁₈ phases are prepared by reaction of monofunctional silanes [24–26] and polymeric C₁₈ phases are prepared using trifunctional silanes in the presence of water, which results in cross-linking to form silane polymers on the silica surface [24, 25].

Polymeric C₁₈ phases have been demonstrated to separate PAH isomers [24, 26, 27, 33] based on shape of the PAH solute. The relationship between the PAH shape and LC retention on polymeric C₁₈ phases was first reported by Wise et al. [24]. The shape of the PAH was defined as the length-to-breadth ratio (L/B) of the box drawn around the molecule that produces the maximum L/B value. Retention of PAH isomers was observed to increase with increasing L/B values. In later studies, the similarity of LC retention on polymeric C₁₈ phases and gas chromatography (GC) retention on liquid crystalline stationary phases was demonstrated for isomeric PAHs [31–34]. In general, the nonplanar PAH isomers, as indicated by a thickness parameter (T), elute earlier than expected based on L/B values. Farkas et al. [35] have demonstrated the usefulness of the quantitative structure-retention relationship technique (QSRR) to predict the retention indices for saturated S-heterocyclic compounds on standard nonpolar polydimethylsiloxane siloxane stationary phases.

In the current study, the LC separation of isomeric PASHs on both a monomeric and a polymeric C₁₈ stationary phase was investigated. There are few publications that have reported the reversed-phase LC separation of isomeric PASHs [36, 37]. In the current study, LC retention indices were determined for 70 PASHs on both monomeric and polymeric C₁₈ phases. Correlation of PASH geometry (L/B and T) and retention on both the monomeric and polymeric C₁₈ phases was investigated. The polymeric C₁₈ phase was the only phase to demonstrate good correlation and will be discussed in detail. This study provides the most comprehensive investigation of reversed-phase LC retention behavior of PASH to date. A similar study for alky-substituted PASHs is published elsewhere [38].

2. Material and methods

2.1. Chemicals

The following PAHs and PASHs were purchased from commercial sources with high purity (>95%): dibenzothiophene (DBT) (Acros Organics, Springfield, NJ); benzo[b]naphtho[1, 2-b]thiophene (BbN12T), benzo[b]naphtho[2, 1-b]thiophene (BbN21T), benzo[b]naphtho[2, 3-b]thiophene (BbN23T), benzo[a]anthracene (BaA), benzo[b]chrysene (BcC), and dibenzo[a,h]pyrene (DBahP) (BCR, Brussels, Belgium); naphthalene (Nap) and phenanthrene (Phe) (Fluka, Buchs, Switzerland). The remaining PASHs were synthesized in the laboratories of M.L.L. at Brigham Young University (Provo, UT). Standard Reference Material (SRM) 869b (Column Selectivity Test Mixture for Liquid Chromatography) were obtained from the Office of Standard Reference Materials at the National Institute of Standards and Technology (Gaithersburg, MD, USA). HPLC grade acetonitrile was purchased from Fisher Scientific (Pittsburgh, PA, USA).

2.2. Molecular Descriptors Calculations

Molecular shape parameter calculations have been described in detail previously [36, 39]. Briefly, ChemDraw 3D software (PerkinElmer, Waltham, MA, USA) was used to draw the molecular structures of PASHs and convert them into the mol file formats. Commercial molecular modeling programs (PC-Model and MMX, Serena Software, Bloomington, IN, USA) and algorithms were used for calculations of the molecular descriptors [length (L), breadth (B), thickness (T) and length-to-breadth ratio (L/B)].

2.3. UV-vis absorption spectrometry

LC detection of PASHs was achieved using selected wavelengths that corresponds to a compromise among the maximum absorbance wavelengths obtained with a commercial spectrophotometer (model Cary 100, Agilent) from measurements of pure standards. The light source was a 75 W pulsed xenon lamp with a 3000 nm min⁻¹ maximum scan rate. The Czerny-Turner monochromator had an accuracy (± 0.2 nm to 0.4 nm with < 0.189 nm resolution) and reproducibility (< 0.02 nm, N = 10). The monochromator 1200 lines mm⁻¹ grating (30 × 35 mm) were blazed at an angle of 8.6° at 240 nm. Detection was made with a photomultiplier tube (P928) with wavelength range from 190 to 900 nm. Measurements were made with a standard 700 µL quartz cuvette with 1 cm path length.

2.4. Liquid chromatographic retention data

LC retention index values (log I) were calculated according to equation 1 with the following index markers: (2) Nap, (3) Phe, (4) BaA, (5) BbC, and (6) DBahP [28].

$$\mathit{\text{log }}I=\frac{\text{log}{R}_x-\text{log}{R}_n}{\text{log}{R}_{n+1}-\text{log}{R}_n}$$ (1)

R is the corrected retention volume, x represents the solute, and n and n + 1 represent the lower and higher eluting PAH standards. Previous studies have investigated the retention behavior of PAHs on C₁₈ stationary phase using log I values as their basis for retention indices [34, 35]. The log I values are based on three measurements obtained from reference standards. The precision (standard deviation) of the log I values was equal to or less than ± 0.02 log I units. Baseline resolution of two components could be achieved with a difference of ~ 0.06 log I units on both the monomeric and polymeric C₁₈ phase.

2.5. Instrumentation and chromatographic conditions

LC-UV analysis was performed using a liquid chromatograph (1200 series, Agilent, Avondale, PA) coupled to a UV-vis detector (UV2000, Thermo Scientific, Waltham, MS). The LC system was equipped with a gradient pump (G1311A), a degasser (G1322A) and an auto sampler (G1329A). The instrument was computer controlled using commercial software (Chromeleon, Thermo Scientific). Separations were carried out on a monomeric (Agilent 5) and polymeric (Zorbax Eclipse PAH) C₁₈ columns purchased from Agilent (Avondale, PA) with the following characteristics: 25 cm length, 4.6 mm diameter, and 5 µm average particle diameters.

Both LC columns were characterized using the Column Selectivity Test Mixture for Liquid Chromatography (SRM 869b) using selectivity ratios (α_TBN/BaP) of tetrabenzonaphthalene (TBN) and benzo[a]pyrene (BaP). The α_TBN/BaP values were determined with a mobile phase of 85/15 (v/v) acetonitrile (ACN) – water (H₂O) at a flow-rate of 1.5 mL min⁻¹. The optimal separation and detection conditions for each isomer set are listed in Table 1. The LC retention index data (log I) were determined using these conditions.

Table 1.

LC conditions for separating isomeric groups of PASHs.

Isomer Sets	Time (min)	ACNa (%, v/v)	H2Oa (%, v/v)	Flow Rate (mL min−1)	λ (nm)
Three-ring PASHs
MM 184 Da	8.0b	85	15	1.0	254
Four-ring PASHs
MM 208 and 234 Da	30.0b	85	15	1.0	254
Five-ring PASHs
MM 258 Da	20.0b	85	15	1.0	254
MM 284 Da	0.0	85	15	1.5	254
	30.0	100	0
	60	100	0
Six-ring PASHs
MM 282 Da	30.0b	100	0	1.5	294
MM 334 Da	0.0	85	15	1.5	254
	30.0	100	0	1.5
	31.0	100	0	2.0
	60	100	0	2.0
Seven-ring PASHs
MM 306 and 332 Da	65.0b	100	0	2.0	313
MM 384 Da	70.0b	100	0	2.0	304

^aVolume fraction.

^bIsocratic conditions for the total run time.

3. Results and Discussion

In previous studies, the LC retention behavior of isomeric PAHs has been shown to be significantly different on various C₁₈ phases that can be classified as monomeric or polymeric [26–28]. Sander and Wise [28] previously proposed a scheme that uses an LC column test mixture (SRM 869) for determining α_TBN/BaP values to classify the type of C₁₈ phase. In general, the shape selective behavior of the stationary phase for separation of complex isomeric PAH mixtures increases with decreasing α_TBN/BaP. Values for α_TBN/BaP ≥ 1.7 represent monomeric C₁₈ phases, α_TBN/BaP in the range of 1.0 to 1.7 represent intermediate polymeric C₁₈ phases, and values for α_TBN/BaP ≤ 1 represent polymeric C₁₈ phases [29].

In the present study, two commercially prepared C₁₈ columns were selected and used to investigate the LC selectivity for separating isomeric PASHs. The polymeric C₁₈ phase had a α_TBN/BaP = 0.49 and the monomeric C₁₈ phase had a α_TBN/BaP value of 2.05. The LC retention behavior of isomeric PASHs on the two C₁₈ phases was investigated and the optimized conditions for the separations are listed in Table 1. The correlation between retention on the polymeric C₁₈ phase and the molecular shape parameters (L, B, T, and L/B ratio) for isomeric PASH was determined. The molecular shape parameters and retention indices (log I) for the three-, four-, five-, six- and seven-ring PASHs are summarized in Tables 2, 3, and 4. The LC retention indices reported here represent the most extensive compilation of reversed-phase LC retention data for PASHs.

Table 2.

LC retention indices and molecular shape parameters for three and four-ring PASHs isomers.

PASHs	Molecular Dimensions				Log I
	L (Å)	W (Å)	T (Å)	L/B
					Monomeric C18	Polymeric C18
Three-ring PASHs
MM 184 Da
DBT	11.62	8.02	4.06	1.45	3.28	2.84
N12T	11.28	8.07	4.16	1.40	2.84	2.90
N21T	11.28	8.07	4.06	1.40	2.78	2.74
N23T	11.67	7.52	4.07	1.55	2.91	2.89
Four-ring PASHs
MM 208 Da
P19T	11.08	9.34	4.06	1.19	3.47	3.33
MM 234 Da
A12T	13.39	8.70	4.06	1.54	3.92	3.97
A21T	13.51	8.56	4.05	1.58	3.76	3.78
A23T	14.12	7.49	4.05	1.88	3.90	4.45
BbN12T	12.72	9.31	4.39	1.37	4.00	3.75
BbN21T	13.66	8.13	4.06	1.68	4.13	4.22
BbN23T	13.90	8.17	4.06	1.70	4.05	4.05
P12T	13.49	8.11	4.15	1.66	3.90	3.99
P21T	13.35	8.12	4.05	1.65	3.70	3.82
P23T	13.43	8.75	4.05	1.54	3.70	3.67
P32T	13.74	8.75	4.06	1.57	3.66	3.64
P34T	12.27	9.23	4.06	1.32	3.73	3.50
P43T	12.27	9.23	4.06	1.33	3.86	3.75
P9,10T	11.83	10.50	4.06	1.13	3.88	3.71

Table 3.

LC retention indices and molecular shape parameters for the five-ring PASH isomers.

PASHs	Molecular Dimensions				Log I
	L (Å)	W (Å)	T (Å)	L/B
					Monomeric C18	Polymeric C18
MM 258 Da
B12P34T	11.93	10.44	4.05	1.14	4.63	4.44
B12P43T	11.92	10.46	4.05	1.14	5.02	4.09
B45P19T	13.28	9.35	4.05	1.42	4.47	4.55
B45P91T	13.31	9.37	4.10	1.42	4.47	4.55
C45T	13.81	9.28	4.05	1.49	4.76	4.74
Py12T	13.51	9.34	4.06	1.45	4.59	4.66
Py21T	13.28	9.34	4.06	1.42	4.43	4.52
Py45T	11.43	10.58	4.05	1.08	4.54	4.42
TeP112T	13.64	9.32	4.05	1.46	4.97	4.96
TriP45T	11.83	10.62	4.06	1.11	4.60	4.39
MM 284 Da
A12BT	15.85	8.81	4.06	1.80	5.37	5.08
A21BT	14.45	10.06	4.06	1.44	4.99	4.49
A23BT	14.64	9.30	4.07	1.95	5.06	4.72
B34P12T	13.52	10.57	4.84	1.28	4.89	4.25
BbP12T	15.02	9.38	4.36	1.60	5.22	5.16
BbP21T	15.92	8.17	4.07	1.95	5.36	5.43
BbP23T	15.98	8.76	4.06	1.83	4.51	4.70
BbP32T	14.99	9.46	4.06	1.59	4.91	4.52
BbP34T	14.47	9.68	4.10	1.50	5.32	4.79
BbP43T	12.40	10.30	5.35	1.20	4.91	4.06
BbP910T	13.72	10.64	4.33	1.29	5.30	4.65
DiN[12:12]T	15.11	9.34	4.06	1.62	5.40	4.91
DiN[12:21]T	15.58	8.58	4.06	1.82	5.66	5.12
DiN[12:23]T	16.14	8.23	4.06	1.96	5.47	5.43
DiN[21:12]T	12.60	10.23	5.22	1.23	4.97	4.15
DiN[21:23]T	14.53	9.746	4.20	1.49	4.99	4.53
DiN[23:23]T	16.23	8.94	4.06	1.82	5.19	4.81
TriP12T	13.73	10.69	4.09	1.29	4.76	4.39
TriP21T	13.28	10.56	4.79	1.26	4.64	4.16
TriP23T	13.70	11.28	4.07	1.22	4.51	4.21

Table 4.

LC retention indices and molecular shape parameters for the six and seven-ring PASH isomers.

PASHs	Molecular Dimensions				Log I
	L (Å)	W (Å)	T (Å)	L/B
					Monomeric C18	Polymeric C18
Six-ring PASHs
MM 282 Da
B1011C45T	13.49	9.78	4.08	1.38	5.83	5.31
B45TriP112T	11.86	10.54	4.06	1.13	5.57	4.78
Per112T	11.71	10.60	5.06	1.11	5.05	4.03
MM 334 Da
BbB56P12T	16.26	9.50	5.40	1.71	6.05	4.96
BbB56P21T	16.49	9.84	4.96	1.68	6.33	5.38
BbB56P23T	16.04	10.20	5.06	1.57	5.82	4.86
BbB56P43T	11.79	11.73	6.70	1.04	5.14	4.15
BbC12T	16.91	9.77	4.13	1.73	6.16	5.58
BbC21T	18.10	8.16	4.06	2.22	5.70	4.61
BbC23T	18.29	8.81	4.06	2.08	5.00	5.18
BbC43T	14.69	10.74	5.52	1.37	5.89	4.67
BbTriP12T	14.24	10.51	5.83	1.36	5.78	4.48
BbTriP21T	15.93	10.64	4.56	1.50	6.27	5.46
N21P910T	14.46	10.56	5.47	1.37	6.19	4.70
N[21:34]P12T	13.54	11.93	5.76	1.14	5.55	4.37
Seven-ring PASHs
MM 306 Da
B67Per112T	12.03	11.84	4.06	1.03	6.41	5.22
MM 332 Da
Diacen[12:12]T	15.29	9.82	4.06	1.56	6.25	5.83
N[218:345]Py12T	13.53	11.84	4.07	1.14	5.43	4.22
MM 384 Da
BbB56C12T	16.79	11.02	5.90	1.52	6.70	5.13
BbB56C21T	18.03	10.55	5.25	1.71	7.03	6.12
BbB56C43T	15.94	10.83	6.53	1.47	6.43	5.29
DiP[910:910]T	15.30	11.07	5.66	1.38	7.37	5.36

Table 5 summarizes the regression calculations for the correlation of L/B and LC retention on the polymeric C₁₈ phase for six groups of PASH isomers. Previous studies have used correlation coefficients (r) as a parameter for measuring the linear correlations for PAH retention and L/B [36]. The unique shape selectivity for the separation of PAHs on polymeric C₁₈ phases is demonstrated most dramatically for mixtures with a large numbers of isomers. Generally, the trend is best expressed when the correlation coefficient is close to 1. In cases were the correlation coefficient is not close to 1, it is necessary to use a proper statistical test to determine if there is a significant trend for the correlation coefficient [40]. The calculation of a t-test value is the simplest statistical method for these determinations using equation 2:

$$t_{\mathit{\text{exp}}}=\frac{|r|\sqrt{n-2}}{\sqrt{1-r^2}}$$ (2)

were n is the number of data points. The calculated t_exp value is compared to the t_crit value at the desired significance level based on the degrees of freedom (n − 2). A significant correlation does exist for the r value if the t_exp is greater than the t_crit. These concepts will be applied in the following sections for discussing the correlations between PASH retention on the polymeric C₁₈ phase and L/B values. In cases were the correlation coefficients are found to not demonstrate a significant trend, data points are removed from the correlation plots to help investigate possible explanations.

Table 5.

Correlation coefficients relating retention on the polymeric C₁₈ phase versus L/B.

PASHs	Number of PASHs	Equation	Correlation Coefficient	Degrees of Freedom	t-value, α = 0.05		Significant
					tcrit	texp
Three-ring PASHs MM 184 Da	4	y = 0.47x + 2.17	0.45	2	2.78	0.71	No
Four-ring PASHs MM 234 Da	13	y = 0.97x + 2.38	0.75	11	2.23	3.76	Yes
Five-ring PASHs MM 258 Da	10	y = 1.02x + 3.19	0.75	8	2.31	3.21	Yes
Five-ring PASHs MM 284 Da	20	y = 1.44x + 2.47	0.89	18	2.10	8.31	Yes
Six-ring PASHs MM 334 Da	12	y = 0.66x + 3.83	0.51	10	2.23	1.88	No
Seven-ring PASHs MM 384 Da	4	y = 2.54x + 1.61	0.80	2	4.30	1.89	No

3.1. Three- and four-ring PASH

The molecular structures of the three-ring PASHs (4 isomers) and four-ring PASHs (1 PASH with MM 208 Da and 13 PASH isomers with MM 234 Da) included in the present study are shown in Fig. 1. LC separations of the three-ring PASH isomers are shown in Fig. 2. Baseline resolution of all four components was not achieved on either C₁₈ phase; however, the best separation was obtained with the monomeric C₁₈ phase. Typically, polymeric C₁₈ phases provide better separation of isomeric PAHs as discussed earlier. The elution order of the three-ring PASH isomers was similar on both C₁₈ phases with the exception of DBT. On the monomeric C₁₈ phase, N21T and N23T are only partially resolved; however, on the polymeric C₁₈ phase they are baseline resolved. The elution order of the four three-ring PASH isomers on the polymeric C₁₈ phase follows the expected elution order based on L/B values with the exception of N12T, which elutes fourth on the polymeric C₁₈ phase rather than second as expected based on L/B values.

Fig. 1.

Molecular structures of the three-ring and four-ring PASHs included in the present study.

Fig. 2.

LC separations of the four three-ring PASH isomers on the monomeric and polymeric C₁₈ phases.

The correlation of retention for the four isomers on the polymeric C₁₈ phase vs. L/B (see Table 5, Fig. S1) resulted in a low correlation coefficient of r = 0.45, which provides a t_exp value of 0.71 (t_crit = 4.30, α = 0.05, n = 4) indicating that there is not a significant linear trend between PASH retention and L/B values [40]. N12T and N21T have the same L/B value but N12T has the heterocyclic sulfur atom located in the bay-region of the structure. The correlation coefficient improves to r = 0.93 when N12T is removed (see Fig. S1), however; the t_exp value (2.51) is below the t_crit value (12.71) indicating there is not a significant correlation coefficient to indicate a linear trend is represented. In this case, the number of data points is low because of the limited number of possible three ring PASH isomers (4). The influence of the sulfur in the bay-region becomes more evident with the larger PASH as discussed below.

LC separations of the four-ring PASHs on both monomeric and polymeric C₁₈ phases are shown in Fig. 3. P19T is the only PASH with MM 208 Da and elutes prior to the MM 234 Da isomers as expected. The 13 MM 234 Da isomers were not baseline resolved on either C₁₈ phase. The polymeric C₁₈ phase provided the best separation of the 13 isomers. The monomeric C₁₈ phase had significantly more co-elution between the following isomer groups: (1) P21T, P23T, P32T, and P34T and (2) P43T, P910T, and P12T.

Fig. 3.

LC separations of the 13 four-ring PASH isomers on the monomeric and polymeric C₁₈ phases.

A plot of retention on the polymeric C₁₈ phase vs. L/B for the 13 MM 234 Da isomers is shown in Fig. 4 with a correlation coefficient of r = 0.75, which provides a t_exp value of 3.76 (t_crit = 2.23, α = 0.05, n = 13) indicating that there is a significant linear trend between PASH retention and L/B values [40]. The correlation coefficient for the MM 234 Da is similar to the correlation of the four-ring cata-condensed PAHs with MM 228 Da (r = 0.80, n = 4) [36]. In general, later elution than expected based on L/B was observed for four of the five isomers (P43T, P910T, A12T, and BbN21T; numbers 6, 4, 9, and 12, respectively) with the heterocyclic sulfur atom located in a bay-region of the structure (the fifth isomer is P12T; number 10). The largest difference was with P910T (number 4), expected to elute first with an L/B of 1.13; however, it actually elutes fourth.

Fig. 4.

Plot of retention (log I) on the polymeric C₁₈ phase versus L/B value for the 13 four-ring PASHs with MM 234 Da. Data point assignment: (1) P34T, (2) P32T, (3) P23T, (4) P910T, (5) BbN12T, (6) P43T, (7) A21T, (8) P21T, (9) A12T, (10) P12T, (11) BbN23T, (12) BbN21T, and (13) A23T (See Fig. 1 for structures). PASHs with the sulfur atom located in the bay-region are identified with an asterisk (*) in the graph.

PASH isomers with relatively similar L/B and T values would be expected to have similar retention behaviors on a polymeric C₁₈ stationary phase. The following four isomers pairs meet this criteria: (1) P12T and P21T, (2) P23T and P32T, (3) P34T and P43T, and (4) A12T and A21T. In three of the isomer pairs, the isomer with the sulfur atom located in the bay-region (P12T and A12T) or fj-region (P43T) eluted several minutes later than the isomers with the sulfur not located in either region. These observations suggest that when the sulfur atom is located in the bay-region or fj-region, the sulfur atom is protected from having interactions with the stationary phase (C₁₈). In the case of P23T and P32T, the sulfur atoms are not located in the bay-region of the structure and have similar retention behaviors (log I values).

3.2. Five-ring PASH

The molecular structures of the five-ring peri-condensed PASHs with MM 258 Da (10 isomers) and five-ring cata-condensed PASHs with MM 284 Da (20 isomers) are shown in Fig. 5 and 6, respectively. The LC separations of the isomers with MM of 258 Da on the monomeric and polymeric C₁₈ phases are shown in Fig. 7. Baseline resolution of all components was not achieved on either C₁₈ phase; however, the most complete separation was achieved with the polymeric C₁₈ phase. The separation on the monomeric phase occurs over an 8 min range, whereas the separation on the polymeric phase covers a 31 min range. B45P19T and B45P91T co-elute on both C₁₈ phases as shown in Fig. 7. Similar observations were noticed in the GC separation for these two PASHs on various stationary phases, however; there was a sufficient separation difference that both PASHs provided slightly different retention index values (data not shown). These two isomers have identical L/B values and differ only in position of the sulfur atom in the heterocyclic ring. In the case of these two isomers, the location of the sulfur atom had no observed influence on the separation. However, B12P34T (position 1) and B12P43T (position 2), which only differ structurally by the location of the sulfur atom, had significantly different retention behavior on both C₁₈ phases. The separation of these two isomers is reversed on the monomeric and polymeric C₁₈ phases. B12P34T has the sulfur atom located in the bay-region and elutes 8 min after B12P43T. B45P19T and B45P91T, which co-elute, have identical L/B values and neither isomer has the sulfur atom located in the bay-region. Similar observations were observed and discussed in Section 3.1. for the MM 234 Da cata-condensed PASH isomers.

Fig. 5.

Molecular structures of the five-ring peri-condensed PASHs isomers with MM 258 Da included in the present study.

Fig. 6.

Molecular structures of the five-ring cata-condensed PASHs isomers with MM 284 Da included in the present study.

Fig. 7.

LC separations of the 10 five-ring peri-condensed PASH isomers with MM 258 Da on the monomeric and polymeric C₁₈ phases. The graph insert shows the plot of retention (log I) on the polymeric C₁₈ phase versus L/B value for the ten five-ring PASHs isomers with MM 258 Da. Data point assignment for the plot for the insert: (1) B12P43T, (2) TriP45T, (3) Py45T, (4) B12P34T, (5) Py21T, (6) B45P19T, (7) B45P91T, (8) Py12T, (9) C45T, and (10) TeP112T.

A plot of retention on the polymeric C₁₈ phase vs. L/B for the 10 MM 258 Da isomers is shown in the insert of Fig. 7 with a correlation coefficient of r = 0.75, which provides a t_exp value of 3.21 (t_crit = 2.31, α = 0.05, n = 10) indicating that there is a significant linear trend between PASH retention and L/B values [40]. From the plot it can clearly be seen that two of the isomers, B12P43T and TeP112T (numbers 1 and 10), have significantly different elution compared to the remaining eight isomers. In the case of TeP112T, the later elution may be attributed to the sulfur atom located in the bay-region of the structure as discussed in Section 3.1 and 3.2. In comparison to the five-ring peri-condensed PAHs with MM 252 Da (r = 0.36, n = 12), the correlation coefficient is significantly improved for the five-ring peri-condensed PASHs for no obvious reason [36].

Fig. 8 shows LC separations on the monomeric and polymeric C₁₈ phases of the largest isomer set in the current study with 20 isomers with a MM of 284 Da. Unfortunately, some of the reference compounds contained minor impurities (*) which are indicated in the chromatogram. The polymeric C₁₈ phase provided the most comprehensive separation over 45 min with co-elution occurring between the following isomer pairs: (1) DiN[21:12]T and TriP21T, (2) BbP32T and DiN[21:23]T, (3) BbP23T and A23BT, and (4) BbP21T and DiN[12:23]T. The elution of all 20 isomers on the monomeric phase occurs over a span of only 8 min. As a result, significantly more co-elution occurs on the monomeric phase.

Fig. 8.

LC separations of the 20 five-ring PASH isomers with MM 284 Da on the monomeric and polymeric C₁₈ phases.

The correlation of retention on the polymeric C₁₈ phase and L/B value for the 20 MM 284 Da isomers is shown in Fig. 9A with a correlation coefficient of r = 0.84, which provides a t_exp value of 8.31 (t_crit = 2.10, α = 0.05, n = 20) indicating that there is a significant linear trend between PASH retention and L/B values [40]. The correlation coefficient for the MM 284 Da is similar to the correlation of the five-ring cata-condensed PAHs with MM 278 Da (r = 0.93, n = 11) [36]. Unlike earlier results, the overall results show that the placement of the sulfur atom appears to have limited influence on the elution of MM 284 Da isomers. 14 of the 20 MM 284 Da isomers have planar molecular structures, as indicated by the similar thickness parameters (T = 4.06 Å to 4.20 Å). Early elution of nonplanar PAHs has been reported in LC separations utilizing shape-selective stationary phases [29]. BbP43T, DiN[21:12]T, TriP21T, and B34P12T are the most nonplanar isomers (T = 4.79 Å to 5.35 Å) included in this isomer set. BbP12T and BbP910T are slightly nonplanar with T values ranging from 4.33 Å to 4.36 Å. Fig. 9B shows the plot with correlation of the retention on the polymeric C₁₈ phase with the L/B value for the seven nonplanar isomers, which results in a correlation coefficient of r = 0.94. The calculated t_exp value is 6.18 (t_crit = 2.57, α = 0.05, n = 7) indicating that there is a significant linear trend between the retention of the nonplanar five-ring PASH isomers and L/B values [40]. Fig. 9C shows the plot of the correlation of retention on the polymeric C₁₈ phase and thickness (T) of these six MM 284 Da isomers, which results in a correlation coefficient of r = −0.84. The calculated t_exp value is 5.17 (t_crit = 2.23, α = 0.05, n = 13) indicating that there is a significant linear trend between the retention of nonplanar five-ring PASH isomers and T values [40]. The results confirm that the elution of planar isomers on the polymeric C₁₈ phases follow increasing L/B ratio; non-planar isomers elute earlier than expected based on L/B and have a negative correlation with degree of non-planarity (i.e., as indicated by thickness).

Fig. 9.

Plot of retention (log I) on the polymeric C₁₈ phase versus L/B value for the (A) 20 five-ring PASHs isomers with MM 284 Da. (B) Plot of retention (log I) on the polymeric C₁₈ phase versus L/B value for the six non-planar five-ring PASHs isomers with MM 284 Da. (C) Plot of retention (log I) on the polymeric C₁₈ phase versus T value for the six nonplanar five-ring PASH isomers with MM 284 Da. Data point assignment for the plot in B and C: (1) BbP43T, (2) DiN[21:12]T, (3) TriP21T, (4) TriP23T, (5) B34P12T, (6) TriP12T, (7) A21BT, (8) BbP32T, (9) DiN[21:23]T, (10) BbP910T, (11) BbP23T, (12) A23BT, (13) BbP34T, (14) DiN[23:23]T, (15) DiN[12:12]T, (16) A12BT, (17) DiN[12:21]T, (18) BbP12T, (19) BbP21T, and (20) DiN[12:23]T.

3.3. Six-ring PASH

The molecular structures of the six-ring PASHs with MM 282 Da (3 isomers) and MM 334 Da (12 isomers) are shown in Fig. 10. Fig. S2 shows the LC separations of the isomers with a MM of 282 Da on the monomeric and polymeric C₁₈ phases. Baseline resolution was achieved for the three isomers on both the monomeric and polymeric C₁₈ phases over 7.5 min and 28 min, respectively. The elution order was the same on both C₁₈ phases and followed the predicted elution based on L/B values. B45TriP112T and Per112T have similar L/B values (1.13 and 1.11, respectively) but the T values (4.06 and 5.06 Å, respectively) are drastically different. As a result, Per112T eluted significantly earlier than the other two isomers.

Fig. 10.

Molecular structures of the six-ring PASHs with MM 282 Da and MM 334 Da included in the present study.

Fig. 11 shows the LC separations of the 12 six-ring PASHs isomers of MM 334 Da on the monomeric and polymeric C₁₈ phases. Baseline resolution of all components was achieved on the polymeric C₁₈ phase over a 40 min time interval (10 to 50 min). The monomeric C₁₈ phase also provided a separation of the majority of the 12 isomers with minimal co-elution over an 8 min interval (12 to 20 min). BbC12T and N21P910T were the only isomers co-eluting on the monomeric phase. On the monomeric phase BbC23T elutes first, whereas on the polymeric phase, BbC23T elutes late in the separation (9^th of 12), as expected based on the large L/B ratio (L/B = 2.08).

Fig. 11.

LC separations of the 12 six-ring PASH isomers with MM 334 Da on the monomeric and polymeric C₁₈ phases.

Of particular interest is the fact that 9 of the 12 isomers investigated in this study are nonplanar (Table 4; T ≥ 4.20 Å, ranging from 4.56 to 6.70 Å). A plot of LC retention on the polymeric C₁₈ phase vs. L/B for the MM 334 Da isomers is shown in Fig. 12A, with a correlation coefficient of r = 0.51 and a t_exp value of 1.88 (t_crit = 2.23, α = 0.05, n = 12) indicating that there is not a significant linear trend between PASH retention and L/B values [40]. Fig. 12B shows the correlation between the retention on the polymeric C₁₈ phase vs. L/B for the nine nonplanar MM 334 Da isomers with an improved correlation coefficient of r = 0.83. The calculated t_exp value is 3.94 (t_crit = 2.36, α = 0.05, n = 9) indicating that there is a significant linear trend between the retention of nonplanar six-ring PASH isomers and L/B values [40]. Fig. 12C shows the plot of LC retention on the polymeric C₁₈ phase vs. the T values of the nine nonplanar isomers with a correlation coefficient of r = −0.92. The calculated t_exp value is 6.21 (t_crit = 2.36, α = 0.05, n = 9) indicating that there is a significant linear trend between the retention of nonplanar six-ring PASH isomers and T values [40]. Similar to the nonplanar MM 284 Da isomers, elution on the polymeric C₁₈ phase for the MM 334 Da isomers follow the trends of increasing L/B ratios and decreasing T values.

Fig. 12.

(A) Plot of retention (log I) on the polymeric C₁₈ phase versus L/B value for the 12 six-ring PASHs isomers with MM 334 Da. (B) Plot of retention (log I) on the polymeric C₁₈ phase versus L/B value for the nine nonplanar six-ring PASHs isomers with MM 334 Da. (C) Plot of retention (log I) on the polymeric C₁₈ phase versus T value for the nine nonplanar six-ring PASH isomers with MM 334 Da. Data point assignment: (1) BbB56P43T, (2) N[21:34]P12T, (3) BbTriP12T, (4) BbC21T, (5) BbC43T, (6) N21P910T, (7) BbB56P23T, (8) BbB56P12T, (9) BbC23T, (10) BbB56P21T, (11) BbTriP21T, and (12) BbC12T.

3.4. Seven-ring PASH

The molecular structures of the seven-ring PASHs with MM 306 Da, MM 332 Da, and MM 384 D are shown in Fig. 13. Even though only a limited number of seven-ring PASHs were available, the present study is the first to report the LC retention data (log I) for these isomers. Fig. 14A and Fig. 14B shows the LC separations of three PASH isomers with MM 306 and MM 332 Da on the monomeric and polymeric C₁₈ phases, respectively. Baseline resolution was achieved for the three PASHs on both C₁₈ phases; however, the separation on the polymeric phase occurred over a total of 65 min while the monomeric phase achieves the separation in 7 min. N[218:345]Py12T elutes first on both C₁₈ phases. B67Per112T elutes third on the monomeric phase but second on the polymeric phase. Fig. 14C and Fig. 14D show the LC separations of the four seven-ring PASH with MM 384 Da on the monomeric and polymeric phases, respectively. The elution order of the four isomers was different on the two columns. The most significant difference was observed for BbB56C21T, which elutes third within 10 min on the monomeric phase, but last on the polymeric phase at ~70 min. Correlation of retention on the polymeric C₁₈ phase vs. L/B (Fig. S3 and Table 5) resulted in a correlation coefficient of r = 0.80 and a t_exp value of 1.89 (t_crit = 4.30, α = 0.05, n = 4). This t_exp value indicates that there is not a significant linear trend between the retention of seven-ring PASH isomers and L/B values [40].

Fig. 13.

Molecular structures of the seven-ring PASHs with MM 306 Da, MM 332 Da and MM 384 Da included in the present study.

Fig. 14.

LC separations of seven-ring PASHs with MM 306 Da and MM 332 Da on the monomeric (A) and polymeric (B) C₁₈ phases, and LC separations of seven-ring PASHs with MM 384 Da on the monomeric (C) and polymeric (D) C₁₈ phases.

4. Conclusions

Retention indices for 70 PASHs were determined using a monomeric and polymeric C₁₈ stationary phase. The results presented here represent the most extensive characterization of the LC retention behavior of PASHs ever reported. Similar to the behavior of PAHs, isomeric PASHs are generally better resolved on the polymeric C₁₈ stationary phase than on the monomeric C₁₈ phase. Correlation between the molecular shape parameters of PASHs (L/B and T) and retention on the polymeric C₁₈ phase were similar to PAHs. Correlation coefficients for PASH retention vs. L/B ranged from r = 0.45 (MM 184 Da) to r = 0.89 (MM 284 Da). For larger PASHs (MM ≥ 284 Da), non-planarity (T) had a significant influence on the separation of PASHs. PASHs with larger T values had earlier elution then expected based on L/B values. Besides the L/B and T values, the location of the sulfur atom plays an important role in the separation mechanism of smaller PASHs (MM ≤ 258 Da). PASHs that only differ structurally by the location of the sulfur atom (similar L/B and T values) are expected to have similar retention behaviors (log I values). However, PASHs are more retentive when the sulfur atom is located in the bay-region or fj-region of the structure.