Indoor / Outdoor Relationships for Organic and Elemental Carbon in PM 2 . 5 at Residential Homes in Guangzhou , China

Nine residential areas were selected in this study (three homes in urban areas, three homes near roadsides, and three homes in industrial zones) to evaluate the indoor and outdoor relationship and carbonaceous species characteristics of PM2.5 in Guangzhou, China, during summer and winter 2004. Daily (24 h) average PM2.5 samples were collected on pre-fired quartz-fiber filters with low-volume samplers and analyzed by the thermal optical reflectance (TOR) method following the Interagency Monitoring of PROtected Visual Environments (IMPROVE) protocol. The average indoor and outdoor concentrations of PM2.5 were 88.8 μg/m and 99.1 μg/m, respectively. The average indoor OC and EC concentrations were 21.7 μg/m, and 7.6 μg/m, respectively, accounting for an average of 25.5% and 8.9% indoor PM2.5 mass, respectively. The average indoor and outdoor OC/EC ratios were 3.4 and 3.0, respectively. The average I/O ratios of PM2.5, OC and EC were 0.91, 1.02 and 0.96, respectively. Poor indoor-outdoor correlations were observed for OC in the summer (R = 0.18) and winter (R = 0.33), while strong correlations (R > 0.8) were observed for EC during summer and winter. OC and EC were moderately correlated (R = 0.4) during summer, while OC and EC correlated well during winter, with a correlation coefficient of 0.64 indoors and 0.75 outdoors. Similar distributions of eight carbon fractions in indoor and outdoor TC pointed to the contributions of motor vehicle exhaust and coal-combustion sources. A simple estimation indicates that about ninety percent of carbonaceous particles in indoor air result from penetration of outdoor pollutants, and indoor sources contribute only ten percent of the indoor carbonaceous particles.


INTRODUCTION
Indoor air pollution in China is of increasing concern as public health problem is accompanied by a rapid economic growth (Ho et al., 2004;Cao et al., 2005;Zhu et al., 2010a;Cao et al., 2011).The newly construction areas have reached one billion square meters each year.Indoor air pollution may cause premature death of more than 110 thousands people each year (http://house.focus.cn/newshtml/30317.html).The effects of indoor air pollution on economic development are estimated to be equivalent of 10.7 billions USD loss in China for 2005.The first legislation (GB50325-2001) was issued to control indoor air pollution in civil buildings on in assessing human exposure and health risks.The objectives of this case study are to provide the distributions and relationships of PM 2.5 , OC, and EC in selected indoor and outdoor microenvironments in Guangzhou (urban, roadside, and industrial areas) and to evaluate the contributions of indoor/outdoor sources to indoor carbonaceous aerosol.Information obtained in this study will provide background understanding of indoor air pollutions at residential homes in Guangzhou and help form indoor/outdoor emission control strategies for PM reduction in Guangzhou as well as in Southern China.

Site Selection
Three site-types-urban, roadside, and industrial-were established for this study to represent typical outdoor environments in Guangzhou, consisting of nine paired (indoor and outdoor) sampling sites.Three urban sites (U1, U2, & U3) were selected at Haizhu district, a rapidly developing metropolitan area without obvious impacts of heavy vehicle traffic and industrial activities.Roadside sites were selected at Xinggangxi road (R1), Tianshou road (R2), and Mingyueer road (R3), which have some of the highest traffic flows in Guangzhou.Industrial sites (I1, I2, & I3) were selected within Huangpu district, which a major industrial district of Guangzhou where power plants, chemical and metallurgy factories, are located.A questionnaire was designed for gathering the information of occupancies, home and occupants' habit.Detailed characteristics of the nine pairs of sampling sites are shown in Table 1.All residential homes were under natural ventilation and air change rates weren't measured during the sampling periods.

Sampling Methods
A monitoring program for indoor and outdoor concentrations of PM 2.5 , OC, and EC, which started from 2 nd July to 13 th August 2004 (summer period) and from 29 th November 2004 to 6 th January 2005 (winter period), was performed in Guangzhou city.Paired mini-vol portable samplers with PM 2.5 impactors (Airmetrics, Eugene, OR, USA) were used to measure 24 h average mass concentrations.The flow rate of the samplers is 5 L/min.Mini-vol samplers were simultaneously put in the living room and in the balcony, or platform, or flat roof, which represent outdoor environment.The indoor sampling heights were in the range of 1-1.5 m above ground in order to simulate the breathing zone and to avoid potential interferences from excessive resuspension of particles.All PM 2.5 samples were collected on 47 mm quartz microfiber filters (QM/A) (Whatman, Maidstone, Kent, England).The filters were preheated before sampling at 900°C for 3 hours to remove carbonaceous contaminants.After collection, loaded filters were stored in a refrigerator at about 4°C before chemical analysis to prevent the evaporation of volatile components.Each filter was equilibrated for 24 h in a room with a controlled temperature (25°C) and relative humidity (40%) before and after sampling and weighed twice on an electronic microbalance with 1 μg sensitivity (Sartorius, MC5, Goettingen, Germany).The precision of mass measurement before and after sampling based on replicate weighting is 15 μg and 20 μg per filter, respectively; filters were reweighted if the difference between the replicate weighting was out of this range.Seventy-two pairs of filters were collected for carbonaceous aerosol analysis.

OC and EC Analysis
The samples were analyzed for OC and EC using a Desert Research Institute (DRI) Model 2001 Thermal/ Optical Carbon Analyzer (Atmoslytic Inc., Calabasas, CA, USA).A 0.5 cm 2 punch from the filter was analyzed for eight carbon fractions following the IMPROVE (Interagency Monitoring of Protected Visual Environments) thermal/optical reflectance (TOR) protocol (Chow et al., 1993;Fung et al., 2002;Chow et al., 2004).This produced four OC fractions (OC1, OC2, OC3, and OC4 at 120°C, 250°C, 450°C, and 550°C, respectively, in a helium atmosphere); a pyrolyzed carbon fraction (OP, determined when reflected laser light attained its original intensity after oxygen was added to the combustion atmosphere); and three EC fractions (EC1, EC2, and EC3 at 550°C, 700°C, and 800°C, respectively, in a 2% oxygen/98% helium atmosphere).IMPROVE OC is operationally defined as OC1 + OC2 + OC3 + OC4 + OP and EC is defined as EC1 + EC2 + EC3 -OP.Six blank filters were also analyzed for quality control and the sample results were corrected by the average of the blank concentrations, which were 0.96 and 0.23 μg/m 3 for OC and EC, respectively.Quality Assurance/Quality Control (QA/QC) procedures were described in Cao et al. (2003).

PM 2.5 Mass Distribution among Urban, Roadside, and Industrial Areas
The overall indoor and outdoor averages of PM 2.5 were 67.7 μg/m 3 and 74.5 μg/m 3 during summer, and the corresponding values were 109.9μg/m 3 and 123.7 μg/m 3 during winter (Table 2).Almost all the PM 2.5 concentrations at nine residences show this seasonal pattern with a summer minimum and a winter maximum, which is consistent with ambient observations in Guangzhou city (Cao et al., 2004).Most of indoor PM 2.5 concentrations at the roadside and industrial residences exceeded the new 24-h ambient (outdoor) PM 2.5 standards (75 μg/m 3 , Chinese National Ambient Air Quality Standard will be implemented in 2016), reflecting the serious PM pollution in the Guangzhou residences.
The indoor/outdoor averages of PM 2.5 in urban, roadside, and industrial microenvironments were 56.2/51.2μg/m 3 , 73.5/79.4μg/m 3 , and 73.4/92.9μg/m 3 , respectively, during summer, and they were 83.0/119.9μg/m 3 , 135.6/141.0μg/m 3 , and 111.0/110.2μg/m 3 , respectively, during winter.Larger variations of PM 2.5 concentrations were found in roadside and industrial areas than in urban areas.The ranges of indoor concentrations were found to be larger than those of corresponding outdoor concentrations at roadside and industrial homes, which may be due to the influences of indoor activities/sources, in addition to outdoor penetration, on indoor PM 2.5 concentration.Heavy traffic during the sampling period was the main cause of high PM 2.5 concentrations in roadside homes while intermittent industrial exhaust emissions contributed to elevated PM 2.5 in industrial homes.
The average indoor/outdoor (I/O) ratios are summarized in Table 2 to evaluate the difference between indoor concentrations and corresponding outdoor levels (Long et al., 2000;Li and Lin, 2003;Lai et al., 2010;Zhu et al., 2010).The I/O ratios of PM 2.5 concentrations ranged from 0.49 to 1.19 during summer and from 0.55 to 1.41 during winter, averaging approximately 0.94 and 0.88 for the two seasons.There was no significant seasonal difference observed in the I/O ratio, which was in accord with findings in Taiwan (Li and Lin, 2003).Lower I/O ratios than 1.0 for both seasons demonstrate that they may be affected by outdoor pollutant sources.

Variations of OC and EC under Indoor and Outdoor Microenvironments
The overall average indoor/outdoor OC and EC concentrations were 17.3/19.9μg/m 3 and 6.5/6.6 μg/m 3 , respectively, during summer (Table 3).They increased by different degrees during winter but showed similar seasonal patterns as PM 2.5 mass.The average indoor OC concentrations in urban, roadside, and industrial environments were 1.05 to 1.12 times as high as those of outdoor OC concentrations for both summer and winter (Table 3), implying impacts from indoor OC sources, such as cooking, cleaning solvents, waxes, etc. Gas cooking is the most popular indoor combustion in Guangzhou domestic environments.Less variations were found in EC in urban, roadside, and industrial environments.The overall averages of indoor EC were comparable to or lower than those of outdoor (Table 3), which suggests the lack of major EC sources indoors and that the majority of indoor EC can be attributed to outdoor sources.Consistent with PM 2.5 mass, indoor and outdoor average OC and EC concentrations in roadside and industrial areas were higher than at urban sites.This confirms the influence from motor vehicles and industrial emissions in Guangzhou (Cao et al., 2003(Cao et al., , 2004)).

Relationship between Indoor and outdoor EC and OC Concentrations
Correlations between the indoor and outdoor measurements imply the degree to which outdoor PM 2.5 contributes to indoors.Summer outdoor data of the I2 residence are not considered in this analysis due to poor ventilation.The results are illustrated in Fig. 1.The poor indoor-outdoor correlations of OC (R 2 = 0.18) and EC (R 2 = 0.33) in the summer indicate the presence of multiple carbon sources.However, strong indoor-outdoor correlations (R 2 > 0.8) were observed for OC and EC during winter, reflecting similar source contributions to indoor and outdoor carbonaceous particles.When outdoor OC and EC concentrations are used as independent variables and indoor OC and EC concentrations as dependent variables for regression, the resulting intercepts are 10.6 (OC) and 2.5 (EC) μg/m 3 during summer and 3.3 (OC) and -0.2 (EC) μg/m 3 during winter, as shown in Fig. 2. Each intercept roughly reflects OC (or EC) concentrations that originate exclusively from indoor emission sources because intercepts are the concentration values when outdoor OC (or EC) contributions are zero.A larger intercept indicates that a large portion of the indoor concentrations is derived from indoor sources.The percentage of the OC (or EC) intercept in the average indoor OC (or EC) concentration indicates the contribution of indoor OC (or EC) sources to observed indoor OC (or EC) concentrations.Winter results appear to be reasonable, as the percentage for OC is 12% (= 3.27/26.2),i.e., about 12% of indoor OC results from the contribution of indoor OC sources.The intercept of EC is close to zero when considering sampling and measurement uncertainties.This implies that all the indoor EC concentrations result from penetration of outdoor EC.The result is consistent with those reported by Jones et al. (2000), Na and Cocker III (2005), Cao et al. (2005), and Diapouli et al. (2011), who found that EC mostly originate outdoors.However, summer results show otherwise, which may be due to the influences of secondary OC and various meteorological factors like thundershower during summer.
The average indoor/outdoor ratios are summarized in Table 5 to evaluate the difference between indoor concentrations and the corresponding outdoor levels (Long et al., 2000;Li and Lin, 2003).The I/O ratios of OC concentrations in PM 2.5 ranged from 0.99 to 1.12 in urban, 0.88 to 1.28 in roadside, and 0.38 to 1.25 in industrial environments, with an average of 1.06, 1.05, and 0.89, respectively (Table 5).The large variations among I/O ratios of OC suggest that there was no single dominant emission source for OC.The I/O ratios of EC concentrations in urban, roadside, and rural areas ranged from 0.77 to 1.04, 0.77 to 1.11, and 0.87 to 1.11, with an average ratio of 0.91, 0.97 and 1.0, respectively (Table 5).The I/O ratios below unity imply that no significant emission sources of EC exist indoors.
During summer, OC and EC are moderately correlated with a correlation coefficient of 0.42 indoors and 0.40 outdoors, implying the complex sources of OC and EC.However, OC and EC are correlated well with a correlation

The Distributions of Eight Carbon Fractions Indoors and Outdoors
Carbon abundances in each of these fractions differ by carbon source and they have been used as source signature in source apportionment studies (Chow et al., 2003;Cao et al., 2005;Zhu et al., 2010).From example, OC1 is enriched in the emissions from biomass burning, OC2 is enriched in the emissions from coal burning, and EC1 is enriched in the emissions from motor vehicles (Cao et al., 2005).Grabowsky et al. (2011) unraveled the organic composition of four organic fractions (OC1, OC2, OC3, OC4) on a molecular level and they found no aromatic compounds in OC1 and a large variety of aromatic species in OC2 and OC3.The average percentages of 8 carbon fractions indoors and outdoors were shown in Fig. 3.The average abundances of OC1, OC2, OC3, OC4, OP, EC1, EC2, and EC3 in indoor TC during summer were 7.4%, 21.0%, 17.0%, 9.2%, 18.4%, 18.7%, 8.1%, and 0.1%, respectively.The outdoor TC during summer has a similar profile to the indoor TC, supporting the same contributing sources indoors and outdoors.The averages of OC1, OC2, OC3, OC4, OP, EC1, EC2, and EC3 in indoor TC during winter were 3.0%, 18.9%, 15.6%, 9.9%, 28.5%, 12.5%, 11.9%, and 0.1%, , respectively.The outdoor TC also has a similar profile.However, the carbon profiles were different between summer and winter.The indoor and outdoor TC display two peaks during summer, i.e., one in OC2 and another in EC1, but the TC is characterized by three peaks (OC2, OP, and EC2) during winter (Fig. 4).This demonstrates that the emission sources of OC and EC were different between the two seasons, i.e., coal burning and motor vehicle exhausts dominantly contributed to carbonaceous particles with minor impact from biomass burning during summer, but motor vehicle exhausts have an increasing impact on carbonaceous particles during winter when compared to the source signature data in Cao et al. (2005).High percentages of EC2 in TC during winter are attributed to the impact of onroad motor vehicles because EC2 is the most abundant carbon fraction in the exhaust of motor vehicles (Watson et al., 1994).The pyrolyzed carbon (OP) abundance in the TC accounts for 28% both indoors and outdoors during winter, implying the presence of substantial water-soluble OC (Yang and Yu, 2002).Comparing with the indoor TC in Hong Kong residences (Cao et al., 2005), the eight carbon fractions during summer has similar distributions with those at six residential sites, which suggests that the motor vehicle emissions have heavy impact on residential environment.

Relative Contributions of Indoor/Outdoor OC and EC Sources to Indoor Carbonaceous Particles
A simple model has been used to differentiate the relative contributions of indoor emission and outdoor-to-indoor penetration of carbonaceous particles (Cao et al., 2005).
where TC in is indoor TC concentration, OC in and EC in are the indoor OC and EC concentration, OC out and EC out are the outdoor OC and EC concentration, OC in-real and EC in-real are the real indoor OC and EC emissions, and OC out-pen and EC out-penn are outdoor-to-indoor penetration of OC and EC, respectively.The indoor and outdoor source contributions to indoor TC concentrations are shown in Fig. 4. Outdoor OC accounted for the highest fractions of indoor TC.The contribution of outdoor OC is largest in three types of homes, accounting for more than 60% of indoor TC.Outdoor EC is the secondary contributor (more than 20%) to indoor TC.Small contributions (9.1%) came from indoor OC sources and almost no contributions (1.4%) from indoor EC sources.In comparison, average outdoor OC, indoor OC, outdoor EC, and indoor EC account for 34.4%, 50.6%, 2.1%, and 12.9% of TC, respectively, in Hong Kong residences (Cao et al., 2005).On average, outdoor sources account for 89.5% of indoor TC in Guangzhou and 63.5% of indoor TC in Hong Kong.This implies that the carbonaceous pollutants in residences in Guangzhou are dominated by outdoor sources, primarily motor vehicle and industrial emissions.This study provides a preliminary estimate for indoor and outdoor source contributions to indoor TC.

CONCLUSIONS
The levels of and relationships between OC and EC concentrations inside nine residences relative to outdoor concentrations have been evaluated in Guangzhou, China.
The average indoor OC and EC concentrations were 21.7 and 7.6 μg/m 3 , respectively, and the corresponding outdoor values were 21.9 and 7.9 μg/m 3 , respectively.Carbonaceous aerosol averaged at more than one-third to half of the PM 2.5 mass both indoors and outdoors.Correlation analysis of indoor-outdoor OC and EC concentrations demonstrated that indoor OC was influenced by indoor emission sources while indoor EC concentrations were predominantly due to outdoor penetration.Similar distributions of carbon profile in TC, in terms of the relative abundance of eight carbon fractions, suggest the contributions of similar sources indoors and outdoors.A simple model indicates that the carbonaceous particles found indoors consist of about ninety percent of outdoor sources and ten percentages of indoor sources.Consequently, it is critical to control outdoor emissions to improve the indoor air quality in residences within Guangzhou.

Fig. 1 .
Fig. 1.Relationship between indoor and outdoor concentrations of OC and EC.

Fig. 2 .
Fig. 2. Relationship between OC and EC concentrations indoors and outdoors.

Fig. 4 .
Fig. 4. Relative contributions of indoor/outdoor OC and EC sources to indoor carbonaceous particles.

Table 1 .
Characteristic of indoor/outdoor sampling locations in Guangzhou city.

Table 2 .
The average 24 h indoor and outdoor PM 2.5 concentrations and I/O ratios during summer and winter at roadside, urban and industrial residences.
a Number of samples is 4 at each home; b Values represent average ± standard deviation.

Table 3 .
Average of the concentrations of OC and EC at nine residences in Guangzhou, China (unit: μg/m 3 ).
a Indoor; b Outdoor

Table 5 .
I/O ratios of OC and EC during summer and winter.
Average percentage of total carbon contributed by eight carbon fractions in PM 2.5 indoors and outdoors for nine residential homes.