Analysis of Polycyclic Aromatic Hydrocarbons in Ambient Aerosols by Using One-Dimensional and Comprehensive Two-Dimensional Gas Chromatography Combined with Mass Spectrometric Method: A Comparative Study

Advanced separation technology paired with mass spectrometry is an ideal method for the analysis of atmospheric samples having complex chemical compositions. Due to the huge variety of both natural and anthropogenic sources of organic compounds, simultaneous quantification and identification of organic compounds in aerosol samples represents a demanding analytical challenge. In this regard, comprehensive two-dimensional gas chromatography with time-of-flight mass spectrometry (GC×GC-TOFMS) has become an effective analytical method. However, verification and validation approaches to quantify these analytes have not been critically evaluated. We compared the performance of gas chromatography with quadrupole mass spectrometry (GC-qMS) and GC×GC-TOFMS for quantitative analysis of eighteen target polycyclic aromatic hydrocarbons (PAHs). The quantitative obtained results such as limits of detection (LODs), limits of quantification (LOQs), and recoveries of target PAHs were approximately equivalent based on both analytical methods. Furthermore, a larger number of analytes were consistently identified from the aerosol samples by GC×GC-TOFMS compared to GC-qMS. Our findings suggest that GC×GC-TOFMS would be widely applicable to the atmospheric and related sciences with simultaneous target and nontarget analysis in a single run.


Introduction
Human health research associated with polycyclic aromatic hydrocarbons (PAHs) has raised concerns because certain PAHs are classified as probable human carcinogens [1][2][3][4] and have shown tumorigenic activity and endocrine disrupting activity in mammals [5]. e US EPA has included 16 of them in the list of priority pollutants and has established a maximum contaminant level of 0.2 μg/L for benzo [a]pyrene in drinking water [6]. In the European Union (EU), eight PAHs have been identified as priority hazardous substances in the field of water policy [7]. e EPA priority 16 PAHs and two additional PAHs are now being monitored by European agencies, and they have sought to quantify the individual concentrations of benzo [e] pyrene and perylene in environmental samples [6]. PAHs are found in ambient air in the gas phase and as sorbents to aerosols [8]. us, air monitoring of PAHs to quantify inhalation exposure and to identify other organic compounds is important for insight into photochemical reactions. e quantification and identification of organic compounds in air samples is an important feature of atmospheric chemistry and represents some demanding analytical challenges [9].
For these reasons, a key issue in current analytical methods is the ability to measure a large number of compounds with quantitative analysis for target analytes. Comprehensive two-dimensional gas chromatography (GC×GC) coupled with mass spectrometry (MS) can screen for nontarget compounds with fast identification of the compounds in an entire sample [10]. erefore, previous studies applied GC×GC-MS for the identification of numerous compounds present in air samples [11][12][13]. However, there are limitations on the validation of simultaneous quantification and identification of analytes in air samples. Correspondingly, a validation of simultaneous identification and quantification of PAHs and other compounds in air samples by GC×GC-MS is required. A TOF mass spectrometer was used to acquire sufficient data from a comprehensive two-dimensional chromatographic technique that generated multiple narrow peaks from the short secondary column [14,15]. Generally, GC coupled with quadrupole MS (GC-qMS) in the selected ion monitoring (SIM) mode has been used for quantitative analysis of PAHs in air samples because of its selective detection for specific target compounds [16,17]. However, a GC×GC-TOFMS validated method suitable for the quantification of target PAHs in an aerosol sample compared with GC-qMS in the SIM mode has not yet been reported. e aim of this study was to evaluate the effectiveness of GC×GC-TOFMS in the quantitative analysis of target PAHs as well as the fast identification of multiple compounds for aerosol samples. e validity of the quantitative results obtained by both GC×GC-TOFMS and GC-qMS in the SIM mode was demonstrated by several method performance parameters such as linearity, accuracy, and repeatability.

Air Sampling.
e total suspended particle (TSP) samples were collected at Asan Engineering Building, Ewha Womans University, Seoul, South Korea (37.56°N, 126.94°E, 20 m above ground level), with a PUF sampler (Tisch, TE-1000) on a quartz fiber filter (Quartz fiber filter, QFF, Ø10.16 cm, Whatman, UK). e sampling site is located in the mixed resident area, commercial area, forest area, and nearby roadside. A total of 67 filter samples were obtained during summer (August 12-30, 2013) and winter (January 27-February 16, 2014) and day (9 a.m.∼6 p.m.) and night (8 p.m.∼6 a.m.). Prior to sampling, the quartz fiber filters were baked for 8 h in an electric oven at 550°C to remove possible organic contaminants. e sampled filters were wrapped in aluminum foils and stored in a freezer at −20°C until analysis.

Chemicals.
All organic solvents were of GC grade and purchased from Burdick and Jackson (Phillipsburg, NJ, USA). Standard solutions of target PAHs (Table 1 for their full chemical names and information) except Per and BeP for quantitative analysis were purchased as a mixture at a concentration of 2000 µg/mL in dichloromethane from Supelco (Bellefonte, PA, USA). Per and BeP standards (>99%) were purchased from Aldrich (St. Louis, MI, USA), and a standard mixture of eighteen PAHs was prepared at a concentration of 1000 µg/mL. Deuterium-labeled internal standards of seven PAHs were purchased from Aldrich (St. Louis, MI, USA) and Chiron (Trondheim, Norway) and used for the spiking test as listed in Table 1. Working standard solutions (0.01∼10 µg/mL) were prepared and then stored at −20°C prior to use.

Preparation of Samples.
Air sampling filters were extracted with a mixture of dichloromethane and methanol (3 : 1, v/v) two times using an accelerated solvent extractor (ASE) (Dionex ASE-200) at 40°C and 1700 psi for 5 min. Prior to the extraction, seven deuterated internal standards (Nap-d8, Ace-d10, Phen-d10, Fla-d10, Chr-d12, Per-d12, and BghiPer-d12) were spiked in the filters to compensate for matrix effects during the extraction procedure. Extracts were blown down to 1 mL using a nitrogen evaporator  Table 2.

GC-qMS and GC×GC-TOFMS for Characterization of Aerosol Samples.
In most studies, separation and quantification of PAHs in aerosol samples have been analyzed using a conventional GC-qMS [18]. Flame ionization detection (FID) has also been widely used for quantification as it features a higher response to PAHs which contain only carbon and hydrogen, while oxygenates and other species that contain heteroatoms tend to have a lower response factor [19]. However, this nonspecific detector may not distinguish inferences, which include a large fraction of aliphatic and aromatic compounds in aerosol samples from alkylated PAH homologues. e coupling of GC with MS is increasingly becoming the analytical tool of choice in this regard because of its superior selectivity and sensitivity. Among the most common analyzers including TOF [20], ion trap, and qMS [21,22], qMS is the most widely adopted technique for routine analysis of PAHs [23]. GC-qMS data acquisition takes advantages of both a full mass scan range (scan mode) and specific ion masses for target analytes (SIM mode). e sensitivity in the SIM mode is higher than that in the scan mode of GC-qMS due to the increased dwell time on each monitored ion for trace analysis in some matrices such as in atmospheric aerosols [24,25]. GC-TOFMS has a much faster spectral acquisition rate than GC-qMS does, which is up to 500 full mass scans per second [26]. Consequently, this system is able to widen the application of GC×GC techniques providing very narrow chromatographic peaks, typically 50∼600 ms at the baseline with sufficient density of data points per chromatographic peak [27]. Environmental samples are generally complex, often with more than hundreds of compounds containing structural isomers and homologues spread over a wide range of concentration and volatility. Accordingly, multidimensional separation is an advanced technique offering the possibility of greatly enhanced selectivity using different separation mechanisms for the analysis of complex environmental samples [28][29][30]. In this study, a set of columns DB-5×DB-17 ms was applied to increase the resolution and peak capacity. e fast scanning Pegasus 4D TOFMS system was combined to allow efficient processing of data acquisition, handling, peak detection, and deconvolution. In the one-dimensional column, a 30 m-long DB-5 ms (5% diphenyl/95% dimethyl polysiloxane) stationary phase was used to separate analytes based on volatility and combined with a 1.17 m-long DB-17 ms column (50% diphenyl/50% dimethyl polysiloxane) allowing relative polarity-based separation. Figure 1 shows GC×GC-TOFMS chromatograms of aerosol samples collected at day and night during winter in Seoul, South Korea. To compare the identification ability of GC×GC-TOFMS with GC-qMS, analysis with GC-qMS in the scan mode was performed. A comparison of the one-dimensional chromatograms of the same samples obtained by GC-qMS is shown in Figure 2. 2D chromatograms enable the visual classification of chemically related compounds into groups. It was rare to see that the early-eluting  analytes have an extreme volatility in the chromatogram, as shown in Figure 2. Because of the large losses of these analytes during sample extraction and concentration, particle-associated semivolatile analytes were mainly detected and classified according to their aromatic and aliphatic hydrocarbon groups. Meanwhile, analytes from the GC-qMS chromatogram were separated based on their vapor pressures or boiling points. e GC×GC technique is rather well suited for group separations, and classifying compounds into chemical-related groups could be useful for source identification of atmospheric aerosols by means of the large amount of chemical data handling. e combined use with TOFMS provides rapid and reliable identification of analytes using their deconvoluted pure mass spectra. e major limitation of qMS is its limited scan rate; therefore, quantification and identification is seriously compromised because of the mass spectral skew due to the variations in ion abundances at different regions of a chromatographic peak [31,32]. e numbers of identified chromatographic peaks analyzed by GC-qMS using the same signal threshold setting from the aerosol samples collected at day and night were 35 and 64, respectively. In the case of results obtained by GC×GC-TOFMS, 251 and 297 peaks from the day-and night-time aerosol samples were, respectively, assigned by individual spectral deconvolution. As a result, phthalic anhydride and 1,2-naphthalic anhydride as the markers of secondary formation for gas-phase PAH reactions were identified in the aerosol sample, as shown in Figure 3. Since the products formed through photochemical reactions are often more toxic than their parent PAHs in atmosphere [17], significant efforts have been expended to identify the photochemical products with PAHs in the fields of atmospheric or environmental sciences. In the case of results obtained using GC-qMS, phthalic anhydride and 1,2-naphthalic anhydride were not detected in the same sample. Limitations of one-dimensional separation have been reported for these photochemical products and complex mixtures of the aerosol sample because of their diverse polarities in a single run [33,34]. Contrastively, two anhydrides associated with secondary organic aerosol formation were clearly separated and detected by GC×GC-TOFMS. erefore, it showed advantages for nontarget screening to identify molecular markers or chemical patterns more representative of the aerosol state observed in ambient air.

Validation of GC-qMS and GC×GC-TOFMS for Quantification of PAHs.
GC-qMS and GC×GC-TOFMS were tested individually in order to evaluate their analytical performances. e calibration linearity (regression coefficient, R 2 ) and relative response factor (RRF) are presented in Table 3. e RRF is the ratio between a signal produced by an individual native analyte and the corresponding isotopically labeled analogue of the analyte (as an internal standard). For calculating RRF, 2 ng of each target PAH and each corresponding deuterated internal standard was spiked, and the relative sensitivity in both the methods was compared. Despite the high-speed scanning performance of GC×GC-TOFMS, the RRFs obtained by this method were approximately equivalent to those obtained by GC-qMS. RRF expresses the sensitivity of a detector for a given substance relative to a standard substance [35,36].
us, it indicated that the sensitivity of GC×GC-TOFMS relative to target PAHs is comparable in quantitative analysis. Calibration curves were generated using the peak area for the 18 PAHs at seven concentrations ranging from 0.01 to 10 μg/mL. e linearity was assessed by calculating the regression equation and the correlation coefficient by the least squares method, as shown in Table 3. e R 2 values were greater than 0.999 for GC-qMS and 0.99 for GC×GC-TOFMS. Although data processing for quantification by GC×GC-TOFMS was derived from the combined peak areas Journal of Analytical Methods in Chemistry 5 for the slices of modulated peaks in contrast to production of the single measured peak by GC-qMS, the results meet the criteria for acceptable linearity within this calibration range.
Naturally, the development of quantitative GC×GC studies based on the quantitative results associated with sophisticated implementation for modulated peaks has been delayed where A x is the peak area of a quantifying ion for a given analyte being measured; A is is the peak area of a quantifying ion for its corresponding internal standard; C x is the concentration of a given analyte; and C is is the concentration of its corresponding internal standard. compared with qualitative reports. Recently, the approach to quantifying multiple analytes at once with comprehensive two-dimensional GC has been extensively studied in accordance with the improvement of data processing for the integration of modulated peaks [37,38]. In this study, the modulated peaks of each PAH was automatically combined and integrated by the ChromaTOF software based on a similarity of spectra within an allowable time difference between the second dimension peaks in the neighboring slices of the chromatogram. Recovery test was performed by spiking known amounts of the 18 PAH compounds in a prebaked clean filter at a final concentration of 2 μg/mL and analyses of each through all the experiment procedures were compared using the two different methods. Six duplicate tests were performed, and the results of the recovery are shown in Table 4. e average recoveries were in the range of 90.3 to 158% with relative standard deviations (RSDs) ranging from 3.9 to 28% for GC-qMS, while the recoveries were from 86.3 to 135% for GC×GC-TOFMS, with RSDs ranging from 5.7 to 45%. Most of the targeted PAH compounds were afforded acceptable recoveries, excluding F and Nap by using the two analytical methods due to the high volatility of these compounds. Compared with the reproducibility as expressed in %RSDs, the values obtained by GC-qMS were slightly lower than those obtained by GC×GC-TOFMS; however, the %RSD values of the targeted PAHs excluding F and Nap were acceptable (<20% RSD). ese observations may vary for the versatile GC×GC technique, since the reproducibility of the modulation phase is dependent on the type of modulator, the stability of the stationary phases, and the chemistry of the analyte, regarding interaction with the stationary phase as presented in several prior studies [39,40]. e LOD and LOQ were determined based on the standard deviation (SD) of the intersection of the analytical curve (s) and the slope of the curve (S) as LOD � 3.3 × (s/S) and LOQ � 10 × (s/S). e LOD and LOQ for each PAH compound obtained from both the methods are shown in Table 4. e LOD and LOQ values of the 18 PAH compounds obtained by GC-qMS were similar to the results of previous studies [10,41,42]. us, the suitability of GC×GC-TOFMS for quantification of PAHs was proven by comparing the results with those obtained using GC-qMS.

Conclusion
A fast scanning GC×GC-TOFMS was compared to a GC-qMS for the determination of PAHs in aerosol samples. For separation, identification, and characterization, GC×GC-TOFMS was advantageous over GC-qMS owing to the increased peak capacity, and its results showed enhanced detectability and structured chromatograms for nontarget analysis. e qualitative mass separation by TOFMS combined with an automated peak-finding capability provided the resolution of complex mixed mass spectra, resulting from overlapping chromatographic peaks and spectral deconvolution of individual mass spectra for unknown analytes. Furthermore, the obtained quantitative results such as LODs, LOQs, and recoveries of the 18 target PAHs were approximately equivalent for both the analytical methods. us, GC×GC-TOFMS had advantages for the simultaneous quantification and qualification of PAHs and other organic compounds in a single run. Because of its high degree of separation and capability of spectral deconvolution of overlapping peaks in highly complex samples, comprehensive GC×GC-TOFMS may become a useful platform in many other fields of research.

Conflicts of Interest
e authors declare that they have no conflicts of interest. Journal of Analytical Methods in Chemistry 7