This is the author version of an article published as: Ong, S. and Ayoko, Godwin and Kokot, S. and Morawska, Lidia (2007) Polycyclic aromatic hydrocarbons in house dust samples: Source identification and apportionment. In Proceedings 14th International IUAPPA World Congress, Brisbane, Australia. Copyright 2007 Please consult Author
Accessed from http://eprints.qut.edu.au
Polycyclic aromatic hydrocarbons in house dust samples: Source identification and apportionment S. T. Ong a, G. A. Ayoko a*, S. Kokotb and L. Morawskaa a
International Laboratory for Air Quality and Health and b Inorganic Materials Research Program, School of Physical and Chemical Sciences, Queensland University of Technology, GPO 2434, Brisbane QLD 4001
[email protected]
Abstract House dust is a heterogeneous matrix, consisting of a variety of inorganic, organic and biological materials. Once pollutants are adsorbed onto house dust particles, they either do not degrade at all or degrade at rates that are relatively slower than their ambient counterparts. Thus house dusts are useful reservoirs for chronic exposure to indoor pollutants. In this study, house dust samples from suburban houses in Brisbane, Australia were collected in summer 2004 and winter 2005. Given the growing need to understand the potential risks of indoor pollutants and to develop appropriate control strategies, the objective of the study was to use receptor-oriented models to estimate the number of sources, their compositions and the contribution of each source to the samples. Thus the polycyclic aromatic hydrocarbon (PAH) composition data were analyzed with advanced factor analysis models. Four factors were required to reproduce the summer data well and each factor had distinctive compositions that suggested that natural gas utilities, cooking, vehicle emissions and miscellaneous combustion processes are the main sources of PAHs in the samples. The implications of the results and of the observed correlation between the building characteristics and the PAH profiles on the quality of these indoor microenvironments and the development of control strategies are discussed. Keywords: House dust, PAHs, source identification.
1. Introduction PAHs are widespread contaminants with structures, which have fused aromatic rings. These are arranged linearly, angularly or in clusters. They enter the environment from numerous sources, especially combustion processes such as wood burning, motor vehicle exhaust, cigarette smoking, cooking and agricultural waste burning. Thus studies of the profile and levels of PAHs in various environmental matrices have attracted substantial research interests in the past three decades. Such studies are driven in part by the health effects of PAHs; benz[a]anthracene, benzo[a]pyrene, benzo[b]fluoranthene, benzo[j]fluoranthene, benzo[k]fluoranthene, chrysene, dibenz[a,h]anthracene, and indeno [1,2,3c,d]pyrene, have caused tumours in laboratory animals (Clemons et al 1998; Baek, 1991) and there are well placed fears that exposure to PAHs might cause grave health effects in humans through inhalation, ingestion and prolonged dermal contact. Consequently, the US Environmental Protection Agency has designated 16 PAHs as priority pollutants, which must be monitored in the environment. House dust is a complex mixture of biologically derived material, particulate matter deposited from the indoor aerosol, and soil particles brought indoors by foot. It is a heterogeneous matrix, which
consists of debris and particles from the human and earth origins. A wide variety of organic and inorganic compounds such as toxic elements and semi-volatile organic compounds as well as biological materials may be adsorbed onto these particles (Butte et al., 2003, & 2004). Once these pollutants are absorbed onto house dust particles, they either do not degrade at all or degrade slowly (Butte et al., 2003, & 2004). Thus, house dust can act as a reservoir of pollutants, which may be inhaled through re-suspension into air, ingested accidentally by children or absorbed through the skin. For this reason, several adverse health effects, including childhood leukemia, developmental abnormalities, retardation of IQ and motor skills, and attention disorders have been associated with the exposure of infants to contaminants found in house dust (see for example Lioy et al, 2000). A number of studies have also shown that, on average, people stay in one indoor environment or another for more than 80% of the day (e.g. Ayoko et al, 2004). Therefore, the determination of sources and levels of indoor toxicants is of immense importance in the assessment of population exposure to environmental pollutants. Given the growing need to understand the real and potential effects of indoor pollutants, and to develop appropriate control strategies, an important
1
objective of the study was to use receptor models to identify and to apportion the sources of PAHs in house dust samples. The approach adopted in the study assumes that the chemical constituents of house dust derived from different sources exhibit characteristic attributes that distinguish them. Thus if such characteristics can be identified, then it should be possible to identify the sources of these chemicals in house dust. House dusts from several urban and suburban houses were therefore sampled, and their polycyclic aromatic hydrocarbons composition assessed as a function of their particle sizes. The resulting data set was subsequently subjected to (i) multi-criteria decision making methods to rank the level of contaminants in the house dust, and (ii) multivariate statistical methods to identify patterns in the data and (iii) receptor modeling to identify the possible sources of the contaminants and estimate quantitatively the contribution of each source to the PAH levels in the dust samples. The results are discussed in terms of the development of control measures for indoor pollution.
2. Experimental Methods 2.1.
Description of the Sampling Sites
Dust from carpets was collected from fifteen houses located in different urban and suburban residential buildings in Brisbane, Australia. 2.2 Questionnaire Residents were asked to fill out a questionnaire to provide information on the indoor and outdoor characteristics of their residences. The first part of the questionnaire was based on the external features of the buildings and included questions about the type of roads that are close to the buildings, type of building materials, and surrounding area. The second part consisted of questions about the interior of the buildings, use of chemical sprays, indoor activities such as cooking, heating and smoking, numbers of cars in the house, frequency of cleaning, and fraction of windows in the house that are regularly opened. 2.3 Sampling Method and Chemical Analysis Fifteen bulk dust samples were collected from various suburbs located around Brisbane, Australia. The samples analysed in this paper collected from December 2004 to February, 2005. Control samples were also taken at five selected houses during these sampling periods. Samples from these houses were collected over consecutive weeks. In addition, another five houses were sampled fortnightly.
A 1600 Watt Hoover VC9009 vacuum cleaner with powered floor head that was fitted with a motor and air outlet filters and micro-filtered clean air bags was used for the sample collection. Horizontal and vertical sweeps were alternatively employed to vacuum the carpet. After sample collection, the dust bags were sealed in air-tight plastic bags prior to transportation to the laboratory. Initially, all the bulk samples contained large pieces of debris (plastic, hair, paper, grass etc). These artifacts were removed by manual sieving before the rest of the samples were sieved by means of a mechanical shaker (Endecotts Ltd) to give sub-samples with < 45 μm, 45-89 μm and 90 μm particle sizes. 0.5 g of each of the sieved sub-samples was extracted by Soxhlet extraction using a mixture of solvents: 1: 2 hexane:dichloromethane (DCM). All solvents were of analytical grade and were supplied by either by Ajax Fine Chem, Australian Chemical Reagents or Lab-Scan Analytical Sciences. The extracts were then passed through a silica mesh no. 60 contained in columns that are topped with 1cm of anhydrous sodium sulfate and eluted with DCM in order to remove all polar contaminants. Using a rotary evaporator, the excess solvent was initially reduced to about 5ml before it was further reduced to approximately 0.5ml by a gentle flow of ultra-pure nitrogen gas. As the extract volume reduced acetonitrile was gradually added in order to exchange the sample solvent. The samples were then filtered through the syringe Teflon filters (pore size 0.45 µm) and quantitatively made up to 1 mL in AR acetonitrile. The blank samples were extracted in the same manner. The samples, which contain perdeuterated acenaphthene-D10, chrysene-D12, perylene-D12, and phenanthrene-D10 internal standards were analyzed on an Hewlett Packard 1100 Series HPLC system with reference to external standards made up of sixteen EPA priority PAH standard (QTM PAH Mix) containing napthalene (NAP), acenapthylene (ACY), acenapthene (ACE), fluorene (FLU), phenanthrene (PHE), anthracene (ANT), fluoranthene (FLT), pyrene (PYR), benz(a)anthracene (BAA), chrysene (CHR), benzo(b)fluoranthene (BBF), benzo(a)pyrene (BAP), dibenz(a, h)anthracene (DBA), benzo(g, h, i)perylene (BGP), benzo(k)fluoranthene (BKF), indeno(1, 2, 3cd)pyrene (IND). The analysis was performed using a dedicated PAH column (LichroCART 250-4), UV/VIS detector operated at 220 and 254 nm, and isocratic elution (50% water + 50% acetonitrile).
2
3. Data Analysis The application of Principal Component Analysis (PCA) to environmental data is associated with significant setbacks because the outcomes are correlated with but not proportional to source contributions (Anderson, et al, 2002). As a consequence, the results of PCA cannot be used directly for source apportionment. In addition, PCA does not provide physically explainable solutions without recourse to rotation techniques such as Varimax. Unfortunately, no wholly satisfactory rotation method has been found (Xie et al, 1999). Despite the limitations of PCA, when it is coupled with absolute principal component scores or multi-linear regression, it becomes a powerful tool for source identification and source apportionment (Guo et, al, 2004;; Anderson et al, 2002). PCA/APCS was performed as described by Guo et al, 2004 on a data matrix that consisted of 45 observations and 16 variables (the 16 priority US EPA PAHs). The main purpose of the PCA was to reduce the number of inter-correlated variables in the original data to fewer number of factors, which are orthogonal to each other. The first factor accounts for the highest variability in the data and subsequent factors account for progressively less amount of the data variance. However, as stated earlier, factors are correlated with but are not proportional to source contributions. Therefore, the results of the PCA were further subjected to a receptor modeling approach that is based on multiple linear regressions of the absolute principal component scores. Such modeling approach has previously been used for source apportionment of PAHs and volatile organic compounds in air (Larsen and Baker, 2003 and Simcik et al, 1999) but has not, as far as we aware, been previously employed for source apportionment of PAHs in dust samples. The main objective of its application to the current data set was to determine the percentage contribution of each source to the amount of PAH found in each sample. Thus, the absolute factor scores were used as independent variables and the multiple linear regressions were performed using the concentration of PAH as dependent variables in accordance with the following equation (Guo et al, 2004).
coefficient for the source p and PAH (i), APCSp is the absolute principal component score of the rotated factor p for a sample, APCSp x bpi represents the contribution of source p to Ci. The values of Ci, bio and bpi have the same dimensions as the original concentration of the PAHs.
4. Results and Discussion 4.1 Survey of the pollutants Sixteen US EPA priority PAH were identified and quantified from the samples. Some of the higher molecular weight PAHs were not detected in many of the houses. However, low molecular weight PAHs such as naphthalene, acenapthylene and acenapthene were detected in all of the house dust samples. Table 1 gives the summary statistics of the PAH concentrations found in the summer samples (N= 45). Table 1: Concentrations (μg/g) of the PAHs in the summer house dust samples (n=45). Mean
Max
Min
SD
95% Percentile
NAP
124.21
320.13
19.01
63.28
127.95
ACY
6.92
17.80
0.00
4.94
5.49
ACE
201.69
709.90
0.50
215.58
133.81
FLU
12.63
46.00
0.00
12.93
9.04
PHE
8.77
22.00
2.00
4.96
7.80
ANT
3.61
13.39
0.27
3.24
2.67
FLT
53.64
144.00
6.80
34.24
50.00
PYR
77.25
226.19
0.76
63.59
72.04
CHR
5.42
26.01
0.48
5.30
3.60
BAA
6.39
32.42
1.01
5.54
5.05
BBF
11.30
200.00
0.82
29.70
5.80
BKF
3.48
10.60
0.00
2.37
2.79
BAP
2.97
10.00
0.00
2.49
2.20
DBA
20.08
84.10
0.00
18.79
15.18
BGP
6.08
52.00
0.00
9.71
1.44
IND
3.34
15.59
0.00
4.55
1.15
p
C i = bi0 + ∑ b p i ( APCS ) p
…(1)
p =1
Where p = 1, 2, …., n, bi0 is the constant term of multiple regression for PAH (i) and represents the average contribution of the PAH from sources that were not determined by the principal component analysis, bpi is the multiple linear regression
4.2 Source apportionment The initial analysis produced four factors, which accounted for 37.7, 23.6, 10.2 and 7.3 % of the data variance respectively. By examining the factor loadings closely, it was possible to identify which of
3
the PAHs are most significantly found in the main source represented by each factor. The coefficients obtained from the linear multiple regressions were used to convert the absolute principal component scores into the mass concentrations for each source as depicted in equation 1. A comparison of the predicted mass for all sources with the observed mass is presented as Figure 1. The squared correlation coefficient (R2) of 0.9538 obtained for this linear relationship indicates that the resolved mass effectively accounted for most of the variation in the PAH mass concentrations in the house dust samples.
Factor 2: This source accounts for 11.6% of the PAHs in the dust samples and contains PAHs such as BAA, BBF, BKF, BAP, DBA and IND, many of which are found in natural gas combustion and meat cooking operation (Rogge et al, 1991 and Rogge et al, 1993). Therefore the source is associated with cooking operations. It is well-known that BAA is a tracer for natural gas combustion (Simcik et al, 1999). C1 (frequency of cooking) and W (fraction of window opened) correlate with these PAHs in the loadings plot (Figure 4) and this corroborates the assignment of this source.
Predicted VS Observed 1600 1400 1200 1000 800 600 400 200 0
P r e d ic te d P A H c o n c e n tr a tio n (u g /g )
y = 0.9875x R2 = 0.9538
0
500
1000
source profile presented above, its main constituents are NAP, FLU, PHE, ANT, FLT, PYR and CHR. PHE, FLU, ANT and CHR are tracers for diesel exhaust emission (Tavares et al, 2004). Thus, this factor was assigned diesel emission, in keeping with the diagnostic ratio discussed later below, which showed the importance of diesel emissions to the concentrations of PAHs in the samples. Furthermore, the loadings plot (Figure 4) indicates that these PAHs are displayed opposite DISC1 (distance from the nearest major road) on PC1, suggesting that as the distance decrease the PAH concentration increase.
1500
2000
Observed PAH Concentration (ug/g)
Figure 1: Observed versus predicted concentrations in the house dusts.
PAH
The regression coefficients were also used to convert the component score coefficient matrix into the source profiles. Four sources were identified and the compositional profiles of each source are displayed as Figure 2 while the source contributions are presented as Figure 3. The four sources fitted the data well and offer physically interpretable information on the PAH contents of the house dust samples. Factor 1: This factor contributes 51.3% of the PAHs found in the house dusts. Based on the
Factor 3: The percentage contribution of this source to the PAHs is 35.5%. ACE and ACY are the highest loaded PAHs on this factor but NAP, FLU, PYR, BBF and DBA are also present in the factor. Elevated levels of ACE and ACY have been associated with oil fumes (Fang et al, 2004 ; Wang and Zhu, 2003; Larsen III and Baker, 2003) and smoking (Gundel et al, 1995). Although in the loadings plot (Figure 4), ACE and ACY are loaded fairly opposite the presence of smokers in a residence (S), there is no conclusive evidence that this factor is exclusively due to smoking. Therefore, it is tentatively assigned “miscellaneous combustion”. Factor 4: Only 1.5% of the PAHs is coming from this source, which is dominated by BBF and BGP. The latter is a tracer for petrol vehicle emission (Larsen and Baker, 2003; Simcik et al, 1999). Surprisingly, this source makes the least contribution to the PAHs in the sample, possibly because higher molecular weight PAHs such as BGP predominates in smaller particles, which deposit slowly (Fang et al, 2004).
4.3 Diagnostic ratios Diagnostic ratios of PAHs have previously been used to identify the sources of PAHs in ambient air (Fang et al, 2004 and references therein). This principle was extended to the current study. Thus,
4
-0.01 0.03
0.02
0.01
0 BAP
BKF
BBF
BAA
CHR
PYR
FLT
ANT
PHE
IND
0.04
IND
0.05 DBA
Factor 4 BGP
PAH
DBA
-0.001
BGP
BAP
BKF
BBF
BAA
CHR
PYR
FLT
ANT
PHE
0
FLU
ACE
ACY
PAH S145 S245 S345 S445 S545 S645 S745 S845 S945 S1045 S1145 S1245 S1345 S1445 S1545 S189 S289 S389 S489 S589 S689 S789 S889 S989 S1089 S1189 S1289 S1389 S1489 S1589 S190 S290 S390 S490 S590 S690 S790 S890 S990 S1090 S1190 S1290 S1390 S1490 S1590
0.002 40
0 20
IND
BGP
DBA
BAP
BKF
BBF
BAA
CHR
PYR
FLT
ANT
PHE
FLU
ACE
ACY
NAP
0.008
FACTOR 2 S145 S245 S345 S445 S545 S645 S745 S845 S945 S1045 S1145 S1245 S1345 S1445 S1545 S189 S289 S389 S489 S589 S689 S789 S889 S989 S1089 S1189 S1289 S1389 S1489 S1589 S190 S290 S390 S490 S590 S690 S790 S890 S990 S1090 S1190 S1290 S1390 S1490 S1590
-0.001 PAH
0.006
0.004
-0.004
0.0015
0.001 PAH concentration (ug/g)
IND
BGP
DBA
BAP
BKF
BBF
BAA
CHR
PYR
FLT
ANT
PHE
FLU
ACE
ACY
NAP
0.001 S145 S245 S345 S445 S545 S645 S745 S845 S945 S1045 S1145 S1245 S1345 S1445 S1545 S189 S289 S389 S489 S589 S689 S789 S889 S989 S1089 S1189 S1289 S1389 S1489 S1589 S190 S290 S390 S490 S590 S690 S790 S890 S990 S1090 S1190 S1290 S1390 S1490 S1590
PAH(ug/g)
0
FLU
ACE
-0.0005 NAP
PAH(ug/g)
-0.0005
ACY
PAH(ug/g) -0.002
NAP
PAH (ug/g)
S145 S245 S345 S445 S545 S645 S745 S845 S945 S1045 S1145 S1245 S1345 S1445 S1545 S189 S289 S389 S489 S589 S689 S789 S889 S989 S1089 S1189 S1289 S1389 S1489 S1589 S190 S290 S390 S490 S590 S690 S790 S890 S990 S1090 S1190 S1290 S1390 S1490 S1590
the mean IND/(IND+BGP) ratio for all of the samples was 0.64, which according to Fang et al, 2004 (IND/IND+BGP = 0.35 - 0.7) indicated the importance of diesel combustion to the PAHs in the samples. Similarly, the mean FLU(FLU+PYR) ratio of 0.7 obtained in this work is in accord with the range of 0.6-0.7 expected for samples associated with diesel combustion (Fang et al, 2004). 1000 800
600
0.0015
Factor 1
60
FACTOR 1
400
200
-200
0
300 250 200 150 100 50 0 -50 -100 FACTOR 2
0.0005
300 250 200 150 100 50 0 -50 -100 FACTOR 3
80 FACTOR 4
-20 0
Factor 3
Figure 3: Source contribution for the four factor solution for the summer samples.
0.0005
-0.02 PAH
Figure 2: Source profile for the four factor solution for the summer samples.
Figure 4: Loadings plot for the PAH concentration of the house dusts (where S = Smoking status of occupants; C1 = frequency of cooking; C3 =frequency of cleaning; H =heating).
5
5. Conclusion In this study, receptor modeling was used to identify and apportion the sources of PAHs in house dust samples. The results indicated that four main sources that contributed PAHs to the samples were diesel emission, cooking activities, miscellaneous combustion and petrol vehicle emissions. Thus, any attempt to reduce the PAH contents of house dusts must take measures that target these sources.
6. Acknowledgements We thank McKenzie Lim for experimental assistance to STO throughout this project.
References Anderson M. J., Daly, E. P., Miller, S. L., Milford, J.B., (2002) Source apportionment of exposures to volatile organic compounds: II. Application of receptor models to TEAM study data, Atmospheric Environment, 36, 3643-3658. Ayoko GA. Morawska L., et al. 2004. ‘Application of multicriteria decision making methods to air quality in the microenvironments of residential houses in Brisbane, Australia’, Environmental Science and Technology, Vol 38, 2609-2616. Baek S. O., Field R. A., Goldstone M. E., Kirk P. W., Lester J. N., Perry R. 1991, A review of atmospheric polycyclic aromatic hydrocarbons: sources, fate and behaviour. Water Air Soil Pollution Vol 60, pp 279-300. Brans J P., Mareschall B., ‘PROMETHEE-GAIA visual interaction modeling for multi-criteria location problems. Internal Report, University of Brussels, 1989, STOOTW/244. Butte W. 2003. In: Morawska L and Salthammer T (Ed) Indoor Environment. airborne particles and settled dust, Wiley-VCH, Verlag GmbH & Co. KGaA, Weinheim, pp 407-434. Butte W. 2004. In: Pluschke, P (Ed) Indoor Air Pollution, The Handbook of Environmental Chemistry, Vol 4, Springer-Verlag, Wursburg, pp 90-116. Clemons, J. M., Allan L. M., Marvin C. H., Wu Z., McCary B. E., Byrant D. W., Zaharewski, T. R. 1998 ‘Evidence of estrogen- and TCDD-like activities in crude and fractionated extracts of PM10 air particulate materials suing in vitro gene expression assays, Environmental Science and Technology 32, 1853-1860. Fang G-C., Chang, C-N., Wu, Y-S., Fu, P.P-C., Yang, I-L., Chen, M-H., (2004) Characterisation, identification of ambient air and road dust polycyclic aromatic hydrocarbons in central Taiwan, Taichung, Science of the Total Environment, 327, 135-146.
Guo, H., Wang, T., Louie P.K.K., 2004. Source apportionment of ambient non-methane hydrocarbons in Hong Kong: Application of a principal component analysis/absolute principal component scores (PCA/APCS) receptor model. Environmental Pollution 129, 489 – 498. Larsen III, R.K. and Baker, J. E., (2003) Source apportionment of polycylic aromatic hydrocarbons in the urban atmosphere: a comparison of three methods, Environmental Science and Technology, 37, 1873-1881. Lioy, P.J., Edwards, R.D., Freeman, N., Gurunathan, S., Pellizzari, E., Adgate, J.L., Quackenboss and Sexton, K., (2000) House dusts levels of selected insecticides and a herbicide measured by the EL and LWW samplers and comparisons to hand rinses and urine metabolites, Journal of Exposure Analysis and Environmental Epidemiology, 10, 327-340. Rogge, W. F., Hidermann, L.M., Mazurek, M.A., Cass, G.R. (1991) Sources of fine organic aerosol. 1. Charboilers and meat cooking operations, Environmental Science and Technology, 25, 1112-1125. Rogge, W. F., Hidermann, L.M., Mazurek, M.A., Cass, G.R. (1993) Sources of fine organic aerosol. 5. Natural gas home applicances, Environmental Science and Technology, 27, 2736-2744. Simcik, M.F., Eisenreich, S.J., Lioy, P.J., (1999) Source apportionment and source/sink relationships of PAHs in the costal atmosphere of Chicago and Lake Michigan, Atmospheric Environment 33, 5071-5079. Singh, K.P., Malik, A., Sinha, S.,(2005) Water quality assessment and apportionment of pollution sources of Gomti river (India) using multivariate statistical techniques- a case study, Analytica Chimica Acta, 355-374. Travares Jr., M., Pinto, J.P., Souza, A.L., Scarminio, I.S., Solci, M.C. (2004) Emission of polycyclic aromatic hydrocarbons from diesel engine in a bus station, Londrina, Brazil, Atmospheric Environment, 38, 5039-5044. Xie, Y-L, Hopke, P.K., Paatero, P., Barrie, L.A., Li, S-M. (1999) Idenfication of source nature and seasonal variations of Arctic aerosol by positive matrix fatorization, Journal of the Atmospheirc Sciences, 56, 249-260. Zhu, L., Wang, J.(2003) Sources and patterns of polycyclic aromatic hydrocarbons pollution in kitchen air, China, Chemosphere, 50, 611-618. Gundel, L.A.., Lee, V.C., Mahanama, K.R.R., Stevens, R. K., Daisey, J.M., (1995) Direct determination of the phase distributions of semivolatile polycyclic aromatic hydrocarbons using annular denuders, Atmospheric Environment, 29, 1719-1733.
6
