University of Birmingham Analysis of atmospheric concentrations of quinones and polycyclic aromatic hydrocarbons in vapour and particulate phases

(cid:1) An existing method was extended and validated to analyse larger number of quinones. (cid:1) Concentration of 17 PAH and 15 quinones have been determined. (cid:1) PAH and quinones were analysed from the same substrate air samples. (cid:1) Four quinone compounds are measured in the atmosphere for the ﬁ rst time. (cid:1) Particle e vapour relationships for quinones are assessed. (cid:1) New data for quinones in SRM are reported.


Introduction
Polycyclic aromatic hydrocarbons (PAHs) are a group of environmental pollutants containing two or more fused benzene rings. They exist ubiquitously in the environment and are formed through high-temperature incomplete combustion and pyrolysis of fossil fuels and other organic materials (Harrison et al., 1996). PAHs have been extensively studied since the mid 1970s owing to their potential mutagenic and carcinogenic properties (Giger and Blumer, 1974). Recently, increasing interest has been paid to oxygenated derivatives of PAHs, particularly quinones (Walgraeve et al., 2010), which contain two carbonyl groups in a fully conjugated dione structure. These oxygenated derivatives are reported as being more toxic than their parent PAHs, as they do not require enzymatic activation, therefore acting as direct mutagens and/or carcinogens (Pedersen et al., 2004(Pedersen et al., , 2005Walgraeve et al., 2010). Recently, the IARC has classified anthraquinone as a group 2B chemical, signifying its possible carcinogenicity to humans (Grosse et al., 2011). It has also been shown that exposure to 1,2-naphthaquinone induces mitochondrial production of H 2 O 2 , which mediates the expression of inflammatory genes (Cheng et al., 2012). There is sufficient evidence therefore, that quinones are toxic to both humans and the environment, despite the fact that the mechanisms underlying the toxicology are complex and far from fully understood.
Both gaseous and particulate matter (PM) associated quinones, like PAHs, may be released into the atmosphere through incomplete combustion processes in diesel (Layshock et al., 2010) and gasoline engines (Valavanidis et al., 2006;Jakober et al., 2007), incomplete combustion of waste (Sidhu et al., 2005) and biomass burning (Iinuma et al., 2007). Gaseous and heterogeneous atmospheric processing of PAHs can yield further quinone products via photochemistry and reactions with atmospheric oxidants (OH,NO 3 and O 3 ) or through biological transformation (Sasaki et al., 1997;Atkinson and Arey, 2007). However, quantitative data are limited on the relative importance of primary and secondary sources of quinones. Owing to their toxicity and persistence in the atmosphere, oxygenated-PAH (OPAH, hydroxylated, ketones and quinones) compounds are becoming key components in the analysis of PM (Wingfors et al., 2011).
Numerous studies have quantified PAH compounds in ambient gas and particle phase samples (Chang et al., 2006;Poster et al., 2006), while few studies have measured quinone concentrations (Walgraeve et al., 2010 and refs within). However, very few studies have characterised simultaneously both quinones and parent PAH compounds in the environment. For example, Wang et al. (2011) in a toxicity study of PM 2.5 during the Beijing Olympic Games, utilised gas chromatography-mass spectrometry (GCeMS) to detect parent PAHs, while quantifying oxygenated-PAH (3 of which were quinones) and nitrated-PAH (NPAH) compounds using electron capture negative ion chemical ionisation (NICI) (Wang et al., 2011). Albinet and co-workers measured in Marseille and in a French Alpine area 15 PAH compounds using HPLC with fluorescence/UV detection, while determining simultaneously 17 NPAH and 9 OPAH compounds -2 of them quinones e using GCeMS with NICI (Albinet et al., 2006(Albinet et al., , 2007a(Albinet et al., , 2007b(Albinet et al., , 2008a(Albinet et al., , 2008b. More recently, Wingfors et al. (2011) characterised the size distribution of aerosol and particle-bound PAH, OPAH, and n-alkanes in urban environments. They developed their method to measure 18 OPAHs (10 of which were quinones) using GCeMS, and found that only 6 of their target quinones met the detection criteria, exemplifying the difficulty in successfully detecting quinones (Wingfors et al., 2011).
In order to improve the sensitivity for the detection of quinones, some studies have exploited derivatisation techniques. Cho et al. (2004) converted four quinones to their diacetyl derivatives using zinc and acetyl acetate, prior to detection by Gas Chromatography Electron Impact Mass Spectrometry (GCeEIMS) (Cho et al., 2004). This derivatisation significantly improved the limits of detection, where their repeatability and reproducibility varied between 2 and 22%. Chung et al. (2006) expanded this methodology to measure 12 quinones, and reported that not all quinones could be converted to their diacetyl derivatives (Chung et al., 2006). Jakober et al. (2007) measured the emissions of 9 quinones from gasoline and diesel motor vehicles, using O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine derivatisation in conjunction with LCeMS (Jakober et al., 2006) and GCeMS (Jakober et al., 2007).
In this study we have applied and extended the methodology of Cho et al. (2004) to include 7 derivatised quinones in one fraction of the sample extract, while measuring 17 PAHs and 8 underivatised quinones in the second fraction. The modified methodology was validated by an analysis of urban dust (SRM 1649a and SRM 1649b) which resulted in excellent agreement with certified concentrations for PAHs and quinones (where available), and we report the first quantified values of 9 further quinones in certified urban dust samples. The methodology was also applied to provide the largest comprehensive measurement dataset for quinones and parent PAH compounds in both vapour and particulate phases from single denuder and filter collection media samples, from a trafficked site in Birmingham, UK.

List of PAHs and quinones analysed
Seventeen PAHs and fifteen quinones (see Table S1) were analysed in vapour and particulate phase airborne samples. The sample extract was divided into two fractions in order to increase the sensitivity to quinones. The first portion was analysed for 17 PAHs and 8 quinones whilst the second was derivatised to analyse 7 quinones. Analysed PAHs are listed in Table S1, and the quinones, with abbreviations in Table 1.

Atmospheric sampling
Five days of samples were collected within the campus of the University of Birmingham at a trafficked site located 2 m away from the A38 road, which has an average annual daily flow of traffic of 28,000 vehicles. Five consecutive 24-h samples were taken from 11 to 15 January 2010.
The sampling train consisted of a denuder collecting the vapour phase upstream of a MOUDI collecting the particulate phase fraction, with a polyurethane foam (PUF) located at the end of the sampling train. The denuder was designed by Delgado-Saborit et al. (submitted for publication) and consists of a set of parallel metal plates coated with XAD-4 (Gundel et al., 1995) to trap the PAH and quinone gas-phase, with an experimental efficiency ranging 85e97%. The MOUDI 110R (MSP Copley Scientific, UK) collected size fractionated particle-phase PAH and quinones samples onto PTFE (polypropylene backed) 47 mm filters (Whatman, Maidstone, UK). The aerosol was collected in the size fractions 1.8e10 mm and <1.8 mm for 24 h at 30 L min À1 . Concentrations of particulate phase compounds reported in this work refer to the total sum of compounds on both filters and the PUF, which represents condensed phase material volatilising within the MOUDI. Field blanks were collected as a quality control/quality assurance measure (See Table S2), but no further correction was applied to the analysed concentrations.
While half of the sample extract was blown down and analysed directly, a second half of the sample extract containing deuterated internal standards was derivatised following a modified version of the method described by Cho et al. (2004). Briefly, the extract was transferred to a conical glass test tube and concentrated under nitrogen to almost dryness. Acetic anhydride (200 mL) and zinc (100 mg) were added into the tube and were mixed using a vortex. The mixture was heated at 80 C with a heating block for 15 min, mixing on a vortex every 5 min. Addition of acetic anhydride and zinc was repeated (equivalent amounts) and then followed by a second 15 min heating routine with vortex mixing at 5 min intervals. The samples were then cooled to room temperature. Pentane (3 mL) and 0.5 mL of water were added. The pentane layer was transferred into a second set of silanised conical glass test tubes. The sample was further concentrated under nitrogen to 100 mL. This concentrated sample volume was transferred to a 250 mL GC insert, further evaporated under nitrogen to dryness and reconstituted in 25 mL of recovery standard (i.e. p-terphenyl-d 14 ).
Derivatised samples were stored in the GC inserter vials in a freezer at À20 C until analysis.

Analytical validation
Since the analytical methodology extends the previous Cho et al. (2004) method to include 15 quinones, and combines the extraction and analysis to analyse PAHs and quinones from the same substrate, the quality of the results of the modified methodology was validated by characterising: a) the effect of derivatisation on the additional quinones of our interest (Section 3.1.1 and Fig. S1); b) the recovery levels (Section 3.1.2 and Tables S3 and S4); c) the limits of detection (Section 3.1.2, Supplementary Information Section S1.4 and Table S2); d) the precision and accuracy (Section 3.1.2, Supplementary Information Section S1.5 and Table S3). The method was also validated characterising the concentrations of PAHs and quinones in two NIST standard reference materials as described in detail below (Section 3.1.3 and Table S5).

Derivatisation vs. non-derivatisation of quinones
Derivatisation is a procedure that has proven very useful in enhancing the sensitivity of analytical techniques to detect and quantify analytes (Rosenfeld, 2003;Cho et al., 2004;Chung et al., 2006). Cho et al. (2004) proposed a method to analyse four quinones, namely 1,2NQ, 1,4NQ, PQ and AQ, based on their diacetyl derivatives using zinc and acetyl acetate (Cho et al., 2004). Chung et al. (2006) extended this methodology to measure a larger number of quinones, but found that not all quinones investigated were fully converted to their diacetyl derivatives and did not show lower detection limits (Chung et al., 2006). A subset of the quinones were consequently measured underivatised including AceQ, 2MAQ, 2,3DMAQ, 5,12NAQ, 7,12BaAQ and AQ. BQ and 2,6DTBQ, but were found to achieve improved sensitivities when acetylated (Chung et al., 2006).
In the current work, we have evaluated the measurement of 15 quinones in ambient samples, utilising the derivatisation methodology of Cho et al. (2004), and expanding the species measured by Chung et al. (2006) to include 2MNQ; M1,4BQ; 6,12BaP and 1,6BaP. The ions corresponding to the underivatised and derivatised quinones were measured to assess the extent of derivatisation. All quinones were successfully derivatised at high concentrations (1000 pg mL À1 e14,000 pg mL À1 ), with the exception of 5,12NAQ and AceQ, in agreement with Chung et al. (2006) (see Fig. S1). At lower concentrations (0e1500 pg mL À1 , those expected in ambient air), however, M1,4BQ; 2MAQ; 2,3DMAQ; 6,12BaP and 1,6BaP did not fully derivatise, consistent with results of Chung et al. (2006). 1,2NQ, 1,4NQ, and PQ were found to fully derivatise in agreement with previous studies (Cho et al., 2004;Eiguren-Fernandez et al., 2008) and we report the derivatisation of 2MNQ for the first time. These quinones were therefore analysed in their diacetylated form. Furthermore, it was found that in contrast to Chung et al. (2006), 7,12BaAQ and AQ were successfully derivatised, which might be attributed to the fact that our method involves two additions of the acetylation reagent and the catalyst, whilst the Chung et al. (2006) methodology only repeated the addition of the catalyst, but not the acetylation agent.

Analytical performance
To assess the recovery efficiency of the three analytical methods (i.e. derivatised quinones, underivatised quinones and PAHs), 5 replicate PTFE filters, PUFs and denuders were spiked with 50 mL of standard solution containing all the natural standards and internal standards at a concentration of 200 pg mL À1 and 1000 pg mL À1 , respectively. The spiked sample substrate (i.e. filters, PUFs and denuders) were extracted and analysed as described in the Experimental Section. Results appear in Table S3 and S4.
The average recovery for the PAH spiked standards (Table S3) was 99 AE 12% for the internal standards and 95 AE 16% for the natural standards, similar to those reported for GCeMS analyses of airborne PAHs elsewhere (Escriva et al., 1994;Vasconcellos et al., 2003;Bates et al., 2008;Zhu et al., 2009;Delgado-Saborit et al., 2010). The average recovery for the derivatised quinones was 107 AE 11% and 83 AE 10% for the internal and natural standards, respectively. The lower recovery of the natural standards in comparison to the PAH method is attributed to the additional derivatisation step. The levels of recovery of the internal standards are similar to those analysed by HPLC-APCI-ITMS (Jakober et al., 2006) and GCeMS (Fernandez and Bayona, 1992;Mosi et al., 1997;Cho et al., 2004;Chung et al., 2006). The average recovery levels of the underivatised quinones were 88 AE 7% and 80 AE 6% for the internal and natural standards, respectively. Despite the lower recoveries for certain underivatised quinones (60e96%), this method demonstrated good reproducibility (standard deviation of 2e15%), confirming that the underivatised quinone methodology was deemed suitable (Harrad, 2005). The recovery values were calculated as part of the QA-QC protocols, but no further correction was applied to the results. Table S4 in the Supplementary Information presents the recovery levels calculated for denuders and PUF samples, which are consistent with recovery results calculated from filters (Table S3) as discussed above.
Limits of detection, precision and accuracy were assessed. Full details are in the Supplementary Information. To summarise, sample detection limits for the quinones varied widely, from 3 pg m À3 for 2-methyl-1,4-naphthoquinone to 168 pg m À3 for 5,12naphthacenequinone. Precisions ranged from 2.1% to 20.3%, but were mostly better than 10%, and accuracy (defined as the difference between the true quantity and the analytical results, normalised by the true quantity) ranged from 2.7% to 13.4% and was hence very good (see Table S3).

Standard reference material analysis
Four samples (1 mg) of NIST certified Urban Dust Standard Reference Material (SRM 1649a and 1649b) (SRM-1649a(SRM- , 2007SRM-1649bSRM- , 2009) were analysed to validate the accuracy and the precision of the methodology. The mass of the SRMs was accurately measured using a balance with 1 mg precision and normalised to the mg kg À1 concentrations. Experimental concentrations of PAHs were compared with the certified concentration using a one-sample t-test (SPSS 17.0). Statistically significant differences were not observed between the certified and analysed values (p > 0.05) for all PAHs in both standards, with the exception of B[k]F (p < 0.05). B[k]F was generally found to be higher in concentration in all the experimental dataset (see SI, Table S5). This compound is the second peak of two proximate chromatography peaks. The resolution factor for B[b]F and B[k]F, could indicate that the latter part of the first peak (B[b]F) may be partially integrated within the second consecutive peak (B[k]F), leading to the statistically significant difference of p < 0.05. Overall, the mean experimental concentrations obtained in this study for both SRM 1649a and 1649b were in excellent agreement with the certified concentrations and similar to those reported elsewhere (Crimmins and Baker, 2006;Delgado-Saborit et al., 2010). Table 1 compares the concentrations of quinones measured in SRM 1649a and 1649b, using the methodology described in this study, with the limited literature and certified values. Although no certified concentrations were available for the quinones studied for SRM 1649a, AQ and 7,12BaAQ concentrations were available for SRM 1649b. The measured concentrations of AQ and 7,12BaAQ found in SRM 1649b in this study are in excellent agreement with the certified concentrations. Species concentrations for the two SRMs analysed are expected to be very similar, according to the preamble to the two NIST certificates (SRM-1649a(SRM- , 2007SRM-1649bSRM- , 2009); which is demonstrated by the analyses completed for the PAH compounds (see Table S5). The quinone concentrations measured in both SRMs are also similar (see Table 1), which demonstrates the accuracy and precision of this methodology. The measurements reported here are in good agreement with Cho et al. (2004), who also analysed quinones in their derivatised form. The concentration for the most widely reported quinone, AQ, is in very good agreement with most of the literature, but in contrast to Fernandez and Bayona (1992) who measured 2 quinones utilising gel permeation chromatography coupled to normal phase liquid chromatography (Fernandez and Bayona, 1992). Those authors indicated that their lowest recoveries corresponded to the OPAH measured but did not report recovery values.
The concentration of the underivatised quinones in SRM 1649a and 1649b (shown in Table 1) are the first measurements reported, with the exception of 5,12NAQ. The concentrations of 5,12NAQ determined in SRM 1649a and 1649b are in reasonable agreement with the scarce literature. In essence, the limited information within the literature for the concentrations of quinones analysed in SRM 1649a and 1649b, means that the results reported in this study do not only demonstrate the methodology's applicability, but also enlarges the database of quantified quinone compounds for cross comparison.
In addition, the small standard deviation reported for all quinones (Table 1), and PAHs (SI , Table S5), not only shows that mass of SRM analysed is homogeneous, but also highlights that our method is suitable for the determination of a large list of PAH and quinone compounds from the same analytical matrix.

Concentrations of PAHs and quinones in ambient air
Five days of ambient air samples were collected, extracted and analysed from a trafficked site in Birmingham. Table 2 shows a breakdown of the average concentration measured for individual pollutant species. Four compounds (2-methyl-1,4-naphthoquinone, 2,6-di-tert-butyl-1,4-benzoquinone, methyl-1,4-benzoquinone and 2,3-dimethylanthraquinone) were detected which have not been reported previously in ambient air. Although the general trends of the PAH concentrations obtained are in good agreement with previous measurements at the same site, they are consistently lower. This reflects an improvement in air quality within Birmingham, as reported also by Delgado-Saborit et al. (2010).
The concentrations of quinone species measured in this study are compared with concentrations reported in the literature measured during the cold months (Table 3) in the particulate and vapour phases, when available. Comparison of concentrations with earlier work is not straightforward, as most previous studies have sampled only the particulate phase. Compounds with similar particle-associated concentrations (filter þ PUF in our study) are: 1,2NQ, 1,4NQ, AceQ, AQ, 2MAQ, 7,12BaAQ, 5,12NAQ and 1,6BaPQ. In the case of BQ, reported concentrations of the particle phase in Athens considerably exceeded that which we attribute to the particle phase, although the overall concentration (particle þ vapour) is similar. Measurements of PQ in our work exceed the other published data. The only measurement of similar magnitude to ours is from Fresno, CA (Chung et al., 2006) in winter, where generally high concentrations were experienced. The data show clearly the importance of the vapour fraction (collected by the denuder) that has not been accounted for in most previous studies.   Table S2 for details on instrument limit of detection). Fig. 1 illustrates the percentage of vapour and particulate phases (i.e. the sum of all the MOUDI stages) for each species to enable the comparison of vapour/particle phase distribution, as a function of increasing relative molecular weight (MW). The lower MW PAH were mainly found in the vapour phase, whilst heavier PAHs were present in the particle phase. Similarly, the derivatised quinones (low MW) were mainly found in the vapour phase, whilst the majority of heavier underivatised quinones were present mainly in the particulate phase. Table 2 shows the concentrations measured in the PUF located downstream of the filter, indicating that the higher concentrations measured in this substrate are related with the more volatile species (i.e. species with three or less rings). This is consistent with the work of Wilson et al. (1995), where they found that appreciable amounts of the more volatile species volatilised off from particles collected in the filter, and were captured in a backup sorbent XAD-2 trap (Wilson et al., 1995). Fig. 2 presents the gas-particle phase distribution of PAH and quinones according to the respective molecular weight (Fig. 2a) in which a sigmoidal logistic curve using 4 parameters was fitted using SigmaPlot 12.0. The sigmoidal curve fitted for PAHs and quinones was compared with that reported by Albinet et al. (2008b). The result is consistent with Albinet et al. (2008b), who reported that <10% of the 2-and 3-ring PAHs (MW < 200) and 2ring quinones (MW < 180) were mainly found in the particulate phase, whilst >70% of the 5-ring PAH (MW > 252) and 4-ring quinones (MW > 240) were present in the particulate phase. On the other hand, the partitioning of the 4-ring PAH and 3-ring quinones differs from data reported by Albinet et al. (2008b). In the present study, the base of the sigmoidal curve for PAH extends until MW < 220, which implies that <20% of the 4-ring PAHs (202 < MW < 228) were present in the particulate phase, whilst the same compounds reported by Albinet et al. (2008b) can be found in the particulate phase ranging from 20% to 90%. Likewise, the comparison of the particle phase distribution of 3-ring quinones in this study differs from that reported by Albinet et al. (2008b). The sigmoidal curve fitted to our data presents a less steep slope, which does not reach a plateau at 100%. Therefore the 3-ring quinones are distributed in both particle and vapour phases in a different proportion to that reported by Albinet et al. (2008b), whose data fits a more steep sigmoidal curve reaching a 100% plateau for quinones with MW > 240. These differences may relate to external conditions, such as different temperatures during sampling. These were À4 AE 4 C during the Albinet et al. (2008b) study, whilst temperatures ranged 1e4 C in our study. Fig. 1 shows that a larger percentage of particulate phase for PQ, AQ and 2,6DTMQ was observed in comparison to PAHs of similar molecular weights. This indicates that these quinones are less volatile than their PAH counterparts with similar molecular weights (e.g. PQ, MW ¼ 208 g mol À1 , Ps ¼ 3.11 Â 10 À4 Pa vs. Pyr, MW ¼ 202 g mol À1 , Ps ¼ 1.93 Â 10 À3 Pa). Further analysis of the physico-chemical properties of PAHs and quinones, namely molecular mass, vapour pressure as a measure of volatility; octanole water partitioning coefficient (log Kow) as a measure of polarity and Henry's Law constant as a measure of the tendency to partition to the aqueous phase was performed to disentangle which properties predict better the particle-phase partitioning (Fig. S2). This analysis showed that the trend between vapour pressure and molecular weight was similar for PAHs and quinone species. On the other hand, the correlation between log Kow and the molecular weight showed a different pattern for PAHs and quinones. The heavier the compound, the larger the log Kow; with PAHs always showing larger log Kow than quinones for similar molecular Table 3 Comparison of individual quinone concentrations (pg m À3 ) measured in this study in winter with those reported in the literature during the cold period (winter and autumn periods).

Geographical location
Site description weights. This is consistent with quinones being more polar than their parent PAHs due to the C]O bond. The values of log Kow are proportional to the values of the octanoleair partitioning coefficient (log Koa) (Lohmann and Lammel, 2004). The PAH show much higher Henry's Law constants than the quinones (Fig. S2d), due to the higher aqueous solubility of the latter. Fig. 2 shows also the distribution of particle/vapour phase for PAHs and quinones according to vapour pressure (Fig. 2b), log Kow (Fig. 2c) and Henry's Law constant (Fig. 2d). SigmaPlot 12.0 was used to fit sigmoidal logistic curves using 4 parameters for the case of vapour pressure and log Kow, whilst an exponential curve was fitted for the Henry's Law constant, for PAHs and quinones. Perhaps surprisingly, despite differing physico-chemical properties of the two groups of substances, molecular weight is a remarkably good predictor of vapour pressure (Fig. S2a)  gas-vapour partitioning for both groups of compounds ( Fig. 2a  and b).

Conclusions
This study has performed a comprehensive analysis of PAHs and quinones from the same sampling media. The analytical method uses a modification of existing analytical methodologies for the determination of PAHs and quinones in air, and includes quinones previously not reported. To ensure the appropriate performance of the analytical methodology for the extended list of quinones measured in this study, thorough quality assurance tests have been performed, including characterisation of the limits of detection, precision and accuracy. The method has been validated with the analysis of two Standard Reference Materials, for which concentrations of some quinone species are reported for the first time. The concentrations of PAHs and quinones in the gas and particulate phase in a traffic impacted scenario have been characterised and compared with concentrations reported in the literature. This work considerably extends the field measurement data on quinones by analysing for compounds not previously reported in the atmosphere, and measuring the vapour phase component. The physicochemical properties affecting the gas/particle phase distribution have been studied for PAHs and quinones. Both molecular weight and vapour pressure explain the similar trend observed in both PAHs and quinones.