Relationships between Outdoor and Personal Exposure of Carbonaceous Species and Polycyclic Aromatic Hydrocarbons ( PAHs ) in Fine Particulate Matter ( PM 2 . 5 ) at Hong Kong

Personal and ambient fine particulate matter (PM2.5) samples were simultaneously collected at Hong Kong during winter in 2014. Mass concentration, organic carbon (OC), elemental carbon (EC), and polycyclic aromatic hydrocarbons (PAHs) relationships were analyzed. The correlations of personal and ambient concentrations of PM2.5, OC, and EC indicated the ambient concentrations were the factors showing influences on the personal exposures. Personal to ambient (P/A) ratios in PM2.5, OC, and EC were all > 1, suggesting influences between indoor sources and/or personal activities. Significant higher ambient ΣPAHs concentrations with P/A ratios were nevertheless < 1. The Σ15 U.S. EPA priority PAHs accounted for 50.6% and 70.8% of ΣPAHs in personal and ambient samples, respectively. The ratios of indicator compounds confirmed the origin of PAHs in personal PM2.5, which were found to be associated predominantly with traffic emissions and the influence by the indoor sources.


INTRODUCTION
Fine particulate matter with an aerodynamic diameter less than or equal to 2.5 µm (PM 2.5 ) is an environmental issue subject to major health concern.The fine particles have been observed to be associated with numerous adverse human health effects (Ito et al., 2006;Kim et al., 2015).The fine particulate matter can penetrate into the deepest section of human lungs (alveolar) and diffuse to other target extrapulmonary organs causing notable symptoms, including cardiac and respiratory morbidity and mortality (Pope III et al., 2002;Analitis et al., 2006).Typically, epidemiological studies are based on ambient air quality data collected from outdoor stationary monitoring sites.However, people spend most of their time in indoor microenvironments (> 85%) at urbanized areas (Williams et al., 2000;Jahn et al., 2013); the adverse health effects of PM 2.5 may not only be caused by ambient origin particles but also indoor pollutants (Cao et al., 2005;Baumgartner et al., 2011).In modern day individual's PM 2.5 exposures and their relationships with the corresponding ambient concentrations have been studied in many developed countries (Janssen et al., 1998;Williams et al., 2000;Noullett et al., 2006;Johannesson et al., 2007) and a few Chinese cities (Du et al., 2010;Jahn et al., 2013).
Although positive associations between PM 2.5 exposures and human health effects have generally been reported, the magnitudes of associations in different geographic locations can vary between locations (Jahn et al., 2011).Variation in chemical components of PM 2.5 , such as organic carbon (OC), elemental carbon (EC), and polycyclic aromatic hydrocarbons (PAHs) have been proposed as the links to different adverse human health outcomes (Kim et al., 2013;Baumgartner et al., 2014).OC and EC are the most important chemical components in PM 2.5 especially in highly urbanized areas (Cao et al., 2004).PAHs are products of incomplete combustion processes and ubiquitous in the atmosphere (Guo et al., 2003).PAHs comprise only in a small fraction of composition in the PM 2.5 mass; however, they are one of the most important pollutants of concerns due to their abilities to persist in environment, bioaccumulation properties, carcinogenic, and mutagenic effects (Machala et al., 2001;Boström et al., 2002).
Atmospheric fine particulate pollution in Hong Kong is mainly due to the emissions from motor vehicles, urban construction, industries, and trans-boundary pollution from the Pearl River Delta (PRD) region (Louie et al., 2005;Ho et al., 2006).Previous studies were mainly focused on characterization PM 2.5 mass in indoors and outdoors (Naumova et al., 2002;Cao et al., 2005;Lazaridis et al., 2008;Xu et al., 2015), while studies with personal exposures targeting carbonaceous species and hazardous organic compounds (e.g., PAHs) are still very limited.Reliable scientific data on personal exposure to PM 2.5 as well as their chemical characterization are essential to evaluate the potential human health effects in Hong Kong.The overall aim of the present work was to quantify personal exposure levels in different modes of conditions (outdoor and personal).This study was also targeted to enable reflection of factors specific to the individual modes.
The objectives of this study are to: (1) assess personal PM 2.5 exposures (and their hazardous chemical components) and the associations between personal exposures and ambient concentrations in Hong Kong; (2) investigate PAHs abundance and specification in ambient and personal PM 2.5 ; (3) characterize the potential sources of PAHs in personal PM 2.5 .

Ambient and Personal PM 2.5 Sampling
Ambient PM 2.5 samples were collected on the rooftop (1.5 m above the ground) of Industrial Centre building at The Hong Kong Polytechnic University (PolyU campus), Hung Hom (HH) from 15 th to 19 th of February 2014 and on the rooftop of Shaw Auditorium at Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin (ST) from 25 th to 28 th of February 2014.The Industrial Centre building at PolyU campus is a building equipped with a complete collection of engineering facilities such as additive manufacturing, digital manufacturing, electronics, intelligent automation, composites, fabrication, building services, safety, construction, design realization and aviation services all catered for research and project activities.It is assumed that the results are independent from the sampling locations.Twenty-four hour (24 h) integrated ambient PM 2.5 samples were collected by mini-volume air samplers (Airmetrics, Eugene, OR, USA) on 47 mm quartz fiber filters.The sampler was equipped with a cyclone that separated the particles with a diameter less than 2.5 µm at a flow rate of 5 L min -1 .
Personal PM 2.5 samples were collected along with the above sampling schedule.Nine non-smoker adult subjects (6 females and 3 males, aged 21-42 years at recruitment) residing in different areas of Hong Kong (adjacent to HH and ST) participated in the personal sample collection.The nine recruited participants were college students and office workers.A Leland Legacy Pump (SKC, Inc., Eighty-Four, PA, USA) was connected with a PEM (Personal Environmental Monitor) loaded with one quartz filter (37 mm, Pall Tissuquartz Filter, Pall Corporation, Ann Arbor, MI, USA), which was carried by subjects during each sampling campaign.Flow check (before and after each sampling) was performed by connecting a PEM loaded with filter to a DryCal ® air flow meter for calibration purpose.
The air purge through the filters was set at a flow rate of 10 L min -1 for the Leland/PEM samplers and collected a total air volume of 14.4 m 3 after 24 h.Two to five samples were collected from each of the subjects and a total of 35 valid personal PM 2.5 samples were obtained in this study.
A collocated sampling test was conducted to ensure the PEMs were comparable with mini-volume air samplers.Ambient PM 2.5 samples were simultaneously collected with a mini-volume air sampler and three collocated PEMs.The average personal PM 2.5 (26.0 ± 7.8 µg m -3 ) loaded with quartz filters showed comparable value with the ambient PM 2.5 samples (27.5 ± 8.8 µg m -3 ).The deviation of the PEMs loaded with quartz filters expressed in coefficient of variance (CV = standard deviation/mean (%)) ranging from 0.7% to 5.7%.

Gravimetric Analysis
All quartz fiber filters were pre-heated to 900°C for 3 h before sample collection in order to remove any organic contaminants.An average of triplicate filter weights (± 3 µg) were determined by a balance (Model MC-5; Sartorius AG, Goettingen, Germany, capacity range of 0.1 mg-5.1 g with sensitivity up to ± 1 µg) before and after equilibration with no less than 24 h prior initial filter weighing and post-sample weighing under pre-conditioned temperature (23 ± 2°C) and humidity (40 ± 5%) in a controlled weighing room.All quartz fiber filters (47 and 37 mm) were stored in a drying box (relative humidity < 40%) prior to sample collection.After sample collection, loaded filters were stored in the refrigerator (-20°C) until further chemical analysis.

Carbonaceous Species Analysis
OC and EC were analyzed (on a 0.526 cm 2 punch) by thermal analysis with optical detection following the IMPROVE protocol on a Desert Research Institute (DRI) Model 2001 Thermal/Optical Carbon Analyzer (Atmoslytic Inc., Calabasas, CA, USA) (Cao et al., 2003).The method detection limits for OC and EC were 0.8 and 0.4 µg C cm -2 , respectively, with a precision better than 10% of total carbon (TC).More information about OC/EC analysis can be found in Cao et al. (2004).

Statistical Analysis
Concentrations for PM 2.5 , OC, and EC were reported in µg m -3 .Concentrations for particle-bound PAHs were reported in ng m -3 .Two independent samples t-test was used to assess the mass difference between two variables (e.g., ambient concentrations, personal exposures).Spearman's rank correlation coefficients were used to investigate the associations between personal and ambient concentrations of PM 2.5 and each variable.The analytical results were statistically processed by the IBM SPSS Statistics 21 program (SPSS Inc., USA).All p-values were derived from 2-tailed statistic tests and a value of less than 0.05 was considered statistically significant.

Personal Exposure to PM 2.5 and Carbonaceous Aerosol
The average mass concentrations of PM 2.5 and carbonaceous aerosols (OC and EC) in personal samples and the associated ambient results are summarized in Table 1.Individual's exposure to PM 2.5 from all sampling days ranged from 16.2 to 65.1 µg m -3 with an average of 46.2 ± 18.8 µg m -3 (Table 1).The mean residential indoor and outdoor PM 2.5 concentrations in Hong Kong were in a range of 45.0-69.5 µg m -3 (Chao and Wong, 2002).Time series plots of daily personal exposure to PM 2.5 are shown in Fig. 1(a).Daily mean personal PM 2.5 exposures were in a range of 22.9-74.4µg m -3 during the sampling period.The highest individual PM 2.5 exposure (86.4 µg m -3 ) was observed in the 25 th of February.A previous study conducted at Gothenburg, Sweden showed personal exposure and urban background mean PM 2.5 concentrations were 11.0 and 10.1 µg m -3 , respectively (Molnár et al. 2014).Intraindividual coefficient of variation for each subject ranged from 5.8% to 43.8% for those sampled for no less than 3 days (Table 1).The CV of individual's exposures between days ranged from 4.4% to 71.9% with an overall mean of 29.9% (Fig. ( 1a)).As in Fig. 1(a), the lowest (14.9 µg m -3 ) and highest ambient PM 2.5 concentration (55.1 µg m -3 ) were both observed at HH.In general, average ambient PM 2.5 concentrations (33.8 ± 10.2 µg m -3 ) were considerably lower (p = 0.014) than individual's exposures during the measurement period (Fig. 2(a)).The significant lower ambient concentrations compared to personal exposures was likely the consequence of higher baseline personal exposure for the subjects, which could be due to higher indoor exposures at either the subject's apartment, other indoor microenvironments or possibly in proximity of local sources and personal activities.Personal exposure to PM 2.5 complied with the 24 h Hong Kong Air Quality Objectives of 75 µg m -3 , while 77% of personal measurements with 24 h average PM 2.5 levels exceeded World Health Organization (WHO) air quality guideline (25 µg m -3 ).The average personal PM 2.5 concentrations measured in the present study were compared with other studies (Table 2(a)).The average personal exposure to PM 2.5 measured in the developed countries show lower values than in Hong Kong, such as in American cities (ranging from 8.4 to 44.8 µg m -3 ) (Williams et al., 2000;Larson et al., 2004;Turpin et al., 2007;Brinkman et al., 2009), Canadian cities (varying from 18.0 to 22.0 µg m -3 ) (Kim et al., 2005;Noullett et al., 2006), and European cities (winter = 25.1 µg m -3 ; summer = 8.8 µg m -3 ) (Zmirou et al., 2000).In contrast, personal exposure to PM 2.5 measured in Chinese cities usually have higher values (ranging from 45.4 to 122.4 µg m -3 ) than that in Hong Kong.
Average individual OC and EC concentrations from all sampling days of personal monitoring are listed in Table 1.Time series plots of daily personal OC and EC exposures are shown in Figs.1(b) and 1(c).The individual's exposure to OC and EC in PM 2.5 varied from 3.4 to 22.7 µg m -3 and 0.7  to 5.9 µg m -3 with an average of 9.9 ± 4.5 µg m -3 and 2.2 ± 1.0 µg m -3 , respectively.The average percentage of OC and EC in personal PM 2.5 were 24.3 ± 5.6% and 5.6 ± 1.9%, which is consistent with the personal PM 2.5 exposures, as listed in Table 1.Intra-individual coefficient of variation for OC and EC for each subject ranged from 8.4% to 35.7% and from 18.2% to 61.7%, respectively.As listed in Table 2(a), personal OC ad EC concentrations measured in other developed countries usually have considerably lower values than Hong Kong, such as personal OC and EC ranging from 5.4 to 8.3 µg m -3 and 0.2 to 1.4 µg m -3 in the United State.(Wilson et al., 2000;Landis et al., 2001;Turpin et al., 2007;Brinkman et al., 2009); personal EC varied from 0.6 to 1.0 µg m -3 in Toronto and Prince George, Canada (Kim et al., 2005;Noullett et al., 2006).Individual's OC and EC exposure in urban cities of China were considerably higher than that in Hong Kong (Du et al., 2010;Zhang et al., 2015).
As reported in Figs.2(b) and 2(c), the average ambient OC and EC concentrations were 3.1 ± 2.3 and 1.6 ± 1.4 µg m -3 at ST and 3.1 ± 1.0 and 2.3 ± 0.7 µg m -3 at HH, respectively.It is noticed that OC concentrations in personal PM 2.5 (ranging from 4.7 to 14.8 µg m -3 ) observed from all subjects were significantly higher (p < 0.001) than OC (3.1 ± 1.8 µg m -3 ) in ambient samples (Figs.1(b) and 1(c)).Relatively higher personal EC concentrations were observed compared to ambient EC while no significant mass difference (p = 0.225) was reported for EC.OC/EC ratio > 2 indicates the presence of secondary organic aerosol.Average ambient OC/EC ratio was 1.7 ± 0.6 during the same period when personal samples were collected (Table 1).Average OC/EC ratios varied from 4.2 to 5.9 for personal measurements with an average of 4.9 ± 1.2, which suggested the presence of indoor sources of OC in personal samples.This is consistent with the result reported by Ho et al. (2004), which demonstrated the OC in indoor sources contributed 2-3 µg m -3 in the building near roadsides in Hong Kong.

Personal Levels of PAHs
A total of 26 particle-bound PAHs (parent-and alkyl-PAHs) were measured in this study.Daily personal exposure to ΣPAHs ranged from 0.41 to 2.83 ng m -3 with an average of 0.90 ± 0.48 ng m -3 .As shown in Table 1, individual's exposure to ΣPAHs from all sampling days varied between 0.54 and 1.87 ng m -3 , and the average concentration of ΣPAHs accounted for 0.0024 ± 0.0011% of personal PM 2.5 exposures.The results were all lower than workplace environment in natural rubber sheet factories contaminated by wood burning smoke (Choosong et al., 2010).Intraindividual coefficient of variation for each subject ranged from 10.1% to 51.9% for those sampled no less than 3 days.The CV of personal exposure to PAHs between sampling days ranged from 9.6% to 71.9% with an overall mean of 29.9%.Time series plots of daily ΣPAHs for ambient concentrations and personal exposures are shown in Fig. 1(d).Considerably higher ΣPAHs were found in ambient samples than that in personal samples on each sampling day with significant mass difference (p < 0.01).The ambient site at HH is located approximately 100 m away from the main traffic road (Guo et al., 2003).Spatial variations of ambient ΣPAHs were measured (Fig. 2(d)) with ΣPAHs concentrations observed at HH (4.00 ± 1.77 ng m -3 ) and ST (3.14 ± 0.76 ng m -3 ).It is noticed that subjects at HH exposed to higher personal ΣPAHs compared to the subjects at ST.
The personal exposure to PAHs (selected PAH compounds) measured in the present study were compared with the PAH compounds reported in other personal and/or indoor studies.Summation of the selected personal PAHs listed in Table 2(b) were found to be comparable with those reported in some of the urban cities, such as Boulder (0.69 ng m -3 ) in the U.S. (Brinkman et al., 2009) and Grenoble (1.00 ng m -3 ) in France during summertime (Zmirou et al., 2000).Generally, it was found that personal exposure to ΣPAHs in Hong Kong were considerably lower than those measured in America cities with heavy industries (e.g.refineries) nearby (varying from 2.077 to 3.081 ng m -3 ) (Turpin et al., 2007;Zhu et al., 2011) and the cities in Europe (ranging from 3.98 to 8.46 ng m -3 ) (Zmirou et al., 2000;Fromme et al., 2004).Overall, personal exposure to ΣPAHs in Hong Kong were considerably lower than those obtained in other studies in different countries.The differences in personal exposure to PAHs could be due to location of the sampling sites, different sampling times, in addition of different emission sources.Since the sampling sites in this study are located in university areas with minimal industrial actions, the lower ΣPAHs levels could be possibly attributed to these conditions.In addition, gas-particle partitioning is important in determining the transport, degradation and fate of organic contaminants in environment.The mass fraction for the particulate phase PAHs showed for molecular weights < 202 were less than 20%, whereas for molecular weights > 229 were over 90% (Huang et al., 2014).The absent of gas phase sampling could have also affected the overall outcome of the analysis.
The concentrations of each quantified PAH compound in personal and ambient samples are shown in Table 3.The summation of particle-bound PAHs groups and their percentage contribution to ΣPAHs in personal samples are illustrated in Fig. 3.Among the 26 particle-bound PAHs measured, 2-Methynaphthalene (0.04-0.52 ng m -3 ), chrysene + triphenylene (0.02-0.24 ng m -3 ), and fluoranthene (0.02-0.19 ng m -3 ) were the most abundant compounds and accounted for 25-50% of the concentration of ΣPAHs in personal samples (Table 3).The average concentrations of Σ15 U.S. EPA priority PAHs in ambient samples were about four times higher than that in personal samples (Table 3), and the Σ15 U.S. EPA priority PAHs accounted for 50.6 ± 7.8% and 72.8 ± 2.0% of ΣPAHs in personal and ambient samples, respectively.
High molecular weight PAHs (HMW-PAHs) are predominately associated with particulates while the low molecular weight PAHs (LMW-PAHs) resided in the gaseous-phase (Ho et al., 2009).Parent-PAHs are thermodynamically more stable than alkyl-PAHs.During combustion compounds formed at high temperatures, alkyl-PAHs are depleted as temperature increases; whereas alkyl-PAHs (e.g., 3-ring alkyl-PAHs) that generated at low temperatures, are abundant in petroleum (Douglas et al., 1996;Saha et al., 2009).On average, the concentrations of PAH compounds in personal samples were shown in Table 3.The average ΣHMW-PAHs/ΣLMW-PAHs ratio in personal samples was 1.5, whereas significantly higher value (9.6) was reported in ambient samples (Table 3), indicating LMW-PAHs was easily transported from outdoors and accumulated in indoors.Moreover, this suggests higher contribution of vehicle emissions (with a higher percentage contribution of ΣHMW-PAHs to ΣPAHs) in ambient samples than personal samples.

Relationship between Personal Exposure and Ambient Concentrations
The personal/ambient (P/A) ratio is an indicator of the difference between personal exposure and ambient concentrations (Noullett et al., 2010).P/A ratio greater than unity indicates personal exposure related sources (e.g., personal activities and/or indoor sources) were stronger than ambient sources, whereas weak personal exposure sources demonstrated the ratio less than unity (Wilson and Brauer, 2006).In this study, the P/A ratios for PM 2.5 and their chemical compositions (OC, EC, and particle-bound PAHs) were investigated to assess the difference between ambient concentrations and personal exposures.As shown in Fig. 4(a), daily P/A ratio of PM 2.5 ranged from 0.41 to 4.02 with an average of 1.31.The overall average P/A ratios of OC and EC were greater than unity (3.89 and 1.39, respectively).PM 2.5 and EC P/A ratios > 1 were reported in other studies (Williams et al., 2000;Noullett et al., 2006;Jahn et al., 2013).It was calculated that residents in Hong Kong (personal activities data retrieved from a 40 subjects survey) spent an average of 71.5% (± 24.5%) and 71.5% (± 23.3%) of their daily time at home for students and office workers, respectively.A considerable portion of time was spent at school (16.7% ± 17.5%) and in the office (19.2% ± 18.7%) with the rest of the daily activities divided between transportation (2.4-3.9%) and outdoors (2.0-2.9%).Similar results were reported in the previous study in Hong Kong, on average, the subjects spent more than 86% of their time indoors, 3-7% in transit and 3-7% in outdoors (Chau et al., 2002).
The patterns of ambient concentrations and personal exposures over the sampling period showed generally higher values in personal PM 2.5 , OC, and EC.In contrast, average personal exposure to ΣPAHs was significantly lower than ambient ΣPAHs on each sampling day.Spearman's correlation coefficients were applied to assess the associations between ambient and personal exposure to PM 2.5 , OC, EC, and ΣPAHs.The weak correlation (r s = 0.328, p = 0.055) between ambient and personal PM 2.5 and the intercept greater than zero from the regression analysis in Table 4 suggests that the non-ambient sources (e.g., indoor sources, personal activities) had a significant role in personal exposures.As reported in Brown et al. (2008) and Rivas et al. (2015), local traffic, indoor sources and/or personal activities can significantly affect the personal exposure to PM 2.5 and EC.Mohammadyan (2011) found that ambient PM and time spent in polluted microenvironments (e.g., buses) are the most important determinants of personal exposure to PM 2.5 .Several studies have measured the correlations between personal exposures and ambient concentrations in urban cities of the developed countries, with reported non-ambient PM 2.5 exposures ranged from 0.43 to 8.47 µg m -3 (Williams et al., 2000;Wilson and Brauer, 2006;Noullett et al., 2010).
As shown in Table 4, significant correlations (r s > 0.60, p < 0.01) between OC and EC in personal PM 2.5 and in ambient samples were observed.Moderate significant correlation (r s = 0.580, p < 0.001) between ambient and personal EC indicating that higher personal OC and EC exposures were likely due to indoor sources and/or personal  activities (e.g., transportation).Fair correlations (r s : 0.307-0.319* , p < 0.05) were observed for 3-4 ring PAHs between ambient and personal samples in this study.However, no significant correlation was found (r s = -0.108)for ΣPAHs or 5-6 ring PAHs between ambient and personal samples.Poor correlation has been observed for BghiP (6-ring PAHs) between ambient and personal samples in Zhu et al. (2011).
Different from this finding, Li et al. (2005) reported significant correlations for 5-7 ring PAHs between indoor and outdoor concentrations.

Source Identification of PAH Compounds
Among various particle-bound PAHs emission sources, the vehicle emissions have been known to be the most important contributor in urban cities of different countries (Guo et al., 2003;Li et al., 2005;Zhu et al., 2011).Guo et al. (2003) found that vehicle emissions (e.g., gasoline, diesel) were the predominant sources of airborne PAHs in Hong Kong.Gasoline vehicle exhaust contained more HMW-PAHs (e.g., benzo(a)pyrene, dibenzo(a,h)anthracene), whereas diesel truck was the major source of lighter PAHs (e.g., 4ring PAHs) (Miguel et al., 1998).It has been suggested to use PAH markers and their ratios in distinguishing emission sources (Guo et al., 2003;Zhu et al., 2011).Table 5 lists the diagnostic ratios for individual PAHs, e.g., PHE/PHE + ANT, BbF/BkF, BeP/BeP + BaP, IcdP/IcdP + BghiP, in personal and ambient samples, which were used to investigate their origin or as indicators showing aging of air samples.The values were compared with those reported in other previous studies.Khalili et al. (1995) found that the ratio of PHE/PHE + ANT was 0.50 for gasoline and 0.65 for diesel engines, respectively.The mean PHE/PHE + ANT ratio was 0.58 ± 0.11 in ambient PM 2.5 , and considerably lower values 0.42 ± 0.13 were observed in personal samples.The results indicated that the sources of PAHs dominated in ambient samples were fresh emission from gasoline and diesel engines, and aging particles could be the main sources of personal PAHs (Grimmer et al., 1983;Galarneau, 2008;Katsoyiannis et al., 2011;Tobiszewski and Namieśnik, 2012;Zhang et al., 2005).The BbF/BkF in ambient (1.12) and personal samples (1.19) were similar to those reported for automobile exhaust (1.26) (Dickhut et al., 2000).Zhu et al. (2011) reported similar median ratio of BbF/BkF (1.00) in an urban area in the U.S.
Most of the fresh exhausts have similar contents of benzo(e)pyrene (BeP) and benzo(a)pyrene (BaP) (Grimmer et al., 1983).BaP can be easily decomposed by light and oxidants, thus, BeP/BeP + BaP ratio is regarded as an index of the aging of particles.Relatively higher BeP/BeP + BaP ratio was observed in personal samples (0.78 ± 0.09) than in ambient samples (0.71 ± 0.05), indicating that emission of these two PAHs was not concurrent and BaP could be affected by vigorous activity in indoors.BeP/BeP + BaP ratios reported in this study were higher than those measured in indoors in Guangzhou (0.41-0.72) (Li et al., 2005).The ratio of IcdP/IcdP + BghiP was found to be 0.18 for gasoline emissions and 0.37 for emission from diesel engines (Grimmer et al., 1983;Guo et al., 2003).The average IcdP/IcdP + BghiP ratio in personal samples was 0.36 ± 0.06 indicating the contribution of both gasoline and diesel emissions.Significantly higher IcdP/IcdP + BghiP ratio (0.48 ± 0.06) was observed at the ambient site, which indicated that diesel emissions were the main sources.This result is similar with the finding in other studies (Li et al., 2005;Ho et al., 2009;Saha et al., 2009).

CONCLUSIONS
Personal and ambient PM 2.5 samples were simultaneously collected in Hong Kong during the winter of 2014.PM 2.5 mass, OC, EC, and ΣPAHs concentrations were determined and their correlations were investigated.Relatively higher personal exposure to PM 2.5 (42.6 ± 18.8 µg m -3 ) were reported compared to ambient concentrations (33.8 ± 10.2 µg m -3 ) and the mass difference was statistically significantly different (p < 0.05).Consistent with the PM 2.5 mass, considerably higher OC and EC concentrations were reported in personal samples (9.9 ± 4.5 and 2.2 ± 1.0 µg m -3 , respectively) compared to ambient samples (3.1 ± 1.8 and 1.9 ± 1.1 µg m -3 , respectively).Personal to ambient (P/A) ratios > 1 for PM 2.5 , OC, and EC were found in this study.Weak to moderate correlation coefficients were reported for PM 2.5 , OC, and EC, suggesting that personal exposure was less influenced by ambient sources nevertheless showed greater influences by indoor sources and/or personal activities.A different pattern was shown for ΣPAHs with significantly higher concentrations (p < 0.01) observed from ambient samples (3.46 ± 0.93 ng m -3 ) than personal samples (0.90 ± 0.48 ng m -3 ).The measured PAHs compounds showed considerably lower fractions of ΣHMW-PAHs (from benzo [b]fluoranthene to dibenzo[a,e]pyrene) in personal samples (19.8 ± 10.3%) than ambient samples (55.0 ± 8.6%), indicated that concentrations of these PAHs compounds in personal PM 2.5 were dominated by ambient sources (e.g., vehicle emissions).Higher percentages of ΣLMW-PAHs and ΣAlkyl-PAHs to ΣPAHs were observed in personal samples compared to ambient samples, indicating the importance of contribution of personal activities in indoors.
Notes: N/A denotes data not available; N.D. denotes data non-detectable; c Denotes these measurements were indoor concentrations, not personal exposure.d denotes the summation of the selected PAHs listed in Table2b; e The samples were collected on February of 2014.

Fig. 3 .
Fig. 3. Percentage contribution of different molecular weight PAHs groups to particle-bound PAHs in individuals' personal and ambient samples in Hong Kong.

Table 1 .
Ambient concentrations and individual personal exposure to PM 2.5 , organic carbon (OC), elemental carbon (EC), and Σ26PAHs.b N denotes number of sampling days.

Table 2 .
Concentrations of (a) personal exposure to PM

Table 3 .
Ambient concentration and personal exposures of PAHs in Hong Kong.

Table 4 .
Correlation analysis between ambient and personal exposure to PM 2.5 , OC, EC, and ΣPAHs during the sampling period.N denotes the number of the valid data; * Correlation is significant at the 0.05 level (2-tailed); ** Correlation is significant at the 0.01 level (2-tailed). a

Table 5 .
Diagnostic ratios of PAHs species in ambient and personal PM 2.5