Figure 1. Contributions of sources to PM2.5 in residential outdoor and indoor, in workplace and in personal 48-h exposure samples for non-ETS exposed participants.
SYTTY
ELEMENTAL ANALYSIS AND SOURCE APPORTIONMENT OF PERSONAL PM2.5 EXPOSURES: APPLICATION OF EXPOLIS PM2.5 FILTERS AND DATA (EAS-EXPOLIS)
Project leader: Matti J. Jantunen, National Public Health Institute
(KTL), Environmental Health, P.O.Box 95, FIN-70701 Kuopio, Finland, tel
+358-17-201 340, e-mail: Matti.Jantunen@ktl.fi
| PUBLICATIONS |
| TIIVISTELMÄ SUOMEKSI |
Researcher:
Sari Alm1, Rufus Edwards2, Otto Hänninen1, Kimmo Koistinen1, Anu
Kousa3, and Tuulia Rotko4, Jutta Salo, Jouni Tuomisto1, 1) KTL-Environmental
Health, Kuopio, e-mail: Sari.Alm@ktl.fi, Otto.Hanninen@ktl.fi, Kimmo.Koistinen@ktl.fi,
Jouni.Tuomisti@ktl.fi, 2) UC Berkeley, CA, USA, Rufus_e@uclink.berkeley.edu,
3) YTV, Office for the Environment, Helsinki, Anu.Kousa@YTV.Fi, 4) HU Department
of Sociology, Helsinki, Tuulia.Rotko@helsinki.fi
Consortium: Urban Air Particles and Environmental Health
Financing SYTTY organisation: The Academy of Finland
Funding from SYTTY / Total funding of project (€): 94487
/ 396164
Person-months of work funded by SYTTY / Total person-months of work:
45 / 101,5
KEY WORDS: urban, exposure, PM2.5, elemental analysis, source
apportionment
ABSTRACT
PM2.5 samples of personal exposures as well as residential and workplace outdoor and indoor microenvironments were collected in the EXPOLIS-study in Helsinki (and six other European cities) in 1996-97. In Basel University 16 elements (by ED-XRF) and BS (reflectometry) were analysed from these samples. The resulting large database about PM2.5 concentrations and elemental constituents, variation in time and between the microenvironments, differences and similarities were applied in the present study to identify the most important sources of PM2.5 and their respective contributions to the exposure.
The five most significant sources were identified by PCA: Long range transport (24%), local traffic (23%), salt (16%, only outdoors), detergents (-, not outdoors), and soil mineral PM2.5 (12%). In parentheses is each component's power in explaining the PM2.5 variation outdoors – together 75%. Although the same sources dominated the PM2.5 variation, a different source had the highest domination in each microenvironment. The most important source for PM2.5 exposure variation was mineral dust originating from both outdoor and indoor sources.
While PCA is applied for source identification and describes each source's
contribution to total PM2.5 variation, source reconstruction requires that
the source markers are known a priori, and then quantifies the proportion
of the total PM2.5 which originates from each source in each microenvironment
and exposure. The picture is fairly clear: Three broad sources are responsible
for most ambient air PM2.5 in Helsinki; secondary (marker S), mineral (Si,
Al, CA, Ti, Fe and K), and combustion and other primary PM (COPM for short,
BS, Pb, Br, etc.). Salt (Cl and Na) was also clear, but its contribution
small. In the ETS free indoor microenvironments the proportions of the
secondary and COPM were smaller than in the ambient air and the proportion
of mineral PM was considerably larger. About half of this mineral dust
originates from indoor sources. The contributions to exposure reflect those
indoors. The total PM2.5 (mass) exposure level is similar to the respective
outdoor air concentration, but the contributions from different sources,
and in all probability the toxicity, are different.
EXTENDED ABSTRACT
1 Introduction
Epidemiological studies have provided consistent evidence regarding associations between ambient particulate matter (PM) levels and adverse health effects, such as respiratory symptoms, lung function, hospital admissions or mortality (Pope et al., 1995; Nyberg and Pershagen, 2000). The statistical links between measures of total PM mass and health outcome tend to focus attention on oversimplified concepts. PM is composed of particles with a broad range of origins chemical compositions (Chow et al., 1994; Brook et al., 1997; Eldred et al., 1997; Müller, 1999; Röösli et al., 2001).
Many epidemiological studies (Dusseldorp et al., 1995; Hoek et al., 1997; Roemer et al., 2000) have shown that the composition of the particulate matter is important for the observed health effects. Same conclusion is supported by experiments with animals and studies on occupationally exposed subjects (Spurny, 1996; Costa and Dreher, 1997; Donaldson et al., 1997; Dreher et al., 1997).
The chemical composition of aerosol particles contains information of their origin. Since almost every PM pollution source has a complex elemental signature, the knowledge of the chemical content of outdoor, indoor and personal PM2.5 provides a link to understand relationships between PM emissions, air quality and health effects.
Current regulations and measurement of PM focus on its bulk levels in ambient air. This, however, may not reflect the range of exposures of the population to particles from different sources. For example traffic has been identified as one of the PM2.5 exposure determinants by Koistinen et al (2001), but more detailed source identification could not be done without using elemental composition of the PM samples. Typically PM2.5 contains particles from combustion, atmospheric chemical, nucleation and coagulation processes. Identification of sources of exposure to these pollutants is essential for prioritization of measures to reduce exposures.
In the current work we use source apportionment techniques for identifying sources in personal exposure samples (Jantunen et al., 1998). We also examine the interaction of sources between residential indoor and outdoor microenvironments, and compare the residential results to workplace samples. Furthermore, we examine elemental residential indoor/outdoor ratios to assess seasonal and ventilation effects on the penetration of particles into homes. PM2.5 source reconstruction (Malm et al., 1994 and Brook et al., 1997) is applied to assess the contributions of sources to PM2.5 mass concentrations.
2 Methods
PM2.5 Sample collection
EXPOLIS (Helsinki) was a representative population based study of urban
adult air pollution exposures (Rotko et al., 2000). Microenvironment PM2.5
samples were collected by microenvironment monitors (MEM) inside and outside
of the residence and at the workplace of each participant. The samples
were collected during the times when he/she was expected to be in that
microenvironment during the 48-hour sampling period. In addition, each
participant carried an aluminum briefcase, containing the personal PM2.5
sampling apparatus (PEM) during the same 48-hour sampling period.
Personal exposure samples were collected separately for workday including
commuting, and for private time (Jantunen et al., 1998). Detailed descriptions
of
the PM2.5 methodology, detection limits, quality assurance and quality
control are presented in Koistinen et al. (1999) and Hänninen et al.
(2002). Only the participants, who were not exposed to ETS were considered
in the present analyses.
ELEMENTAL AND BLACK SMOKE ANALYSIS
Elemental analysis of the PM2.5 filters was performed by energy-dispersive
X-ray fluorescence spectrometry (ED-XRF) in the Institute for Mineralogy
and Petrography, University of Basel. XRF was calibrated with 7 commercially
available standards. Relative standard deviation for duplicates for elemental
analysis ranged from 0.034 (Calcium) to 0.048 (Bromine) (Oglesby et al.,
2000). A more detailed description is presented in Mathys et al. (2001).
The blackness of the PM2.5 particles on the filters (BS for black smoke)
was measured with a reflectometer according to the international standard
ISO 9835 (1993).
SOURCE IDENTIFICATION
The PM sources were identified by principal component analysis (PCA),
which uses linear combinations of element concentrations to characterize
the variation of each dimension in a multivariate space. Linear recombination
of eigenvectors of the correlation matrix of element concentrations by
applying a Varimax rotation produces the source vectors (Henry et al.,
1984). The factors represent major sources or meteorological effects to
explain common variations in the elemental concentrations. Receptor model
fundamentals have been reviewed comprehensively by Gordon (1980), Cooper
and Watson (1980) and Henry et al. (1984).
SPSS for Windows, version 9.0, was used in the PCA. Varimax rotation
was used on ln-transformed elemental concentration data. Principal components
with eigenvalues greater than 1 were entered into the source identification
analysis. Absolute coefficients below 0.40 have been omitted from tables
to facilitate identification of sources. Detailed description of the PCA
and source reconstruction methods is presented in Koistinen et al., 2002.
SOURCE RECONSTRUCTION
Specific mass contributions of sources - identified in the PCA analysis
- were calculated using source reconstruction techniques (Countess et al.,
1980, Malm et al., 1994, Brook et al., 1997). Literature data on the chemical
composition of particles from common sources were used to compute the contributions
to PM. Soil elements (Al, Si, Ca, Fe, Ti, K) were calculated to their oxides,
as described by Malm et al. (1994) and Brook et al. (1997), to estimate
the total soil PM mass according to the equation 1:
Soil PM = 2.20 Al + 2.49 Si + 1.63 Ca + 1.58 Fe + 1.94 Ti + 1.41 K. (1)
Sea salt contribution was calculated from measured Cl. Based on YTV data (22) 89% of the Cl is sea salt. Total mass of compounds (incl. Cl, Na, SO42-, Mg, Ca, K and Br) in sea salt were calculated according to Seinfeld and Pandis (1998).
Sea salt PM = 0.89 x 1.816 x Cl. (2)
Inorganic secondary PM was calculated as a sum of ammonium sulfate (from measured S) and ammonium nitrate (from average N/S ratio in YTV data, Ojanen et al., 1998).
The remaining mass was assumed to consist of primary particles emitted from combustion and unknown sources. This fraction is called COPM (combustion and other PM).
3 Results and Discussion
PCA
Several elements (Br, Fe, Mg, Na, Ni, and Pb) of the personal exposure
samples did not meet the quality assurance goals and therefore were excluded
from the analysis. Main reason for this may be the small sample volume
collected in personal sampling. Exclusion of some of the elements resulted
in a lower resolution of sources in personal samples.
The main sources and their respective contributions to exposure variations
in all microenvironments can be identified from the table 1. Source A represents
long-range transported, secondary and primary particles (see also figure
1), B other combustion sources, in Helsinki mainly traffic, C sea and de-icing
salt in outdoor air, D indoor aerosol from cleaning agents and possibly
cooking, and source E soil mineral dust from indoor and outdoor sources.
Comparison of the personal exposures versus the indoor and outdoor
microenvironment concentrations reveals interesting links. The linkage
from outdoor air to exposure is most clear for the long-range transported
secondary and primary combustion PM with no indoor or personal sources.
Source A explains a similar amount of the total elemental variation in
all microenvironments and exposure.
Outdoor-indoor-linkage is also obvious, but less pronounced for source B (local traffic and other combustion particles). Surprisingly, exposure variation due to traffic appears to be smaller than the respective variation in both indoor and outdoor concentrations.
Outdoor air salt (source C) does not appear in the personal exposures.
Exposure to the cleaning agents (source D) appears in residential indoor
samples only. The indoor air mineral dust stems from outdoor air and from
indoor dust re-suspension. Due to this the additional indoor source 'E'
is much more dominant in residential indoor spaces and personal exposures
than in outdoor air.
Sources identified in the residential outdoor and indoor microenvironments
were similar. These samples were measured during identical times, and outdoor
air particles do penetrate indoors. The sources that accounted for
the majority of variance between samples, however, occurred in a different
order in the outdoor and indoor microenvironments. Different sources dominate
the elemental variation of the PM2.5 particles in outdoors. The outdoor
particles do not penetrate equally indoors (Long et al., 2001). Particles
in automobile exhaust are in the range of 10-20 nm (gasoline) and 60-70
nm (diesel) (Ålander et al., 1989). Median particle diameter for
secondary sulfate particles from long-range transport in the Leeds particle
size survey was 1.1µm (APEG, 1999). Windblown dust and soil are abrasion
products and particle diameters are in the coarse particle fraction above
1 µm. Thus, although long-range transport dominates in the
outdoor environment followed by traffic, salt and soil, it is not surprising
that traffic particles dominate in the indoor environment as a result of
increased penetration of smaller particles into indoor environments.
Table 1. Comparison of the main PCA factors in residential outdoor (HO)
and indoor (HI), workplace (W) and 48-h personal exposures (P48).
Factor A
Factor B
Factor C
Factor D
Factor E
Low range transported Local traffic
Salt
Detergents
Soil dust
| Element | HO | HI | W | P48 | HO | HI | W | P48 | HO | HI | W | P48 | HO | HI | W | P48 | HO | HI | W | P48 |
| Al | .8 | .8 | .9 | .8 | ||||||||||||||||
| Br | .8 | .8 | .5 | |||||||||||||||||
| Ca | .7 | .8 | .4 | .8 | ||||||||||||||||
| Cl | -.4 | .9 | .6 | .6 | .8 | |||||||||||||||
| Cu | .7 | .8 | ||||||||||||||||||
| Fe | .9 | .7 | .8 | |||||||||||||||||
| K | .7 | .6 | .8 | .5 | .5 | .6 | ||||||||||||||
| Mg | .6 | .7 | ||||||||||||||||||
| Mn | .5 | .4 | ||||||||||||||||||
| Na | .8 | |||||||||||||||||||
| Ni | .6 | |||||||||||||||||||
| P | .8 | .8 | .8 | |||||||||||||||||
| Pb | .7 | .7 | -.8 | |||||||||||||||||
| S | .7 | .7 | .5 | .7 | -.8 | -.5 | -.5 | |||||||||||||
| Si | .5 | .9 | .8 | .7 | .8 | |||||||||||||||
| Zn | .9 | .8 | .9 | .8 | ||||||||||||||||
| BS | .5 | .4 | .6 | .9 | .6 | .7 | .5 | |||||||||||||
| percent of
variation explained by factor (%) |
24 | 19 | 21 | 21 | 23 | 21 | 19 | 12 | 16 | 14 | 15 | 16 | 12 | 19 | 10 | 22 |
Source reconstruction
Total PM2.5 mass concentrations in residential outdoor air, in residential
indoor air, in workplace and in personal exposure samples for non-ETS participants
were 10 µg/m3, 9.2 µg/m3, 9.1 µg/m3 and 9.2 µg/m3
respectively. Contributions of each identified source are presented in
figure 1. These mass reconstruction sources are not identical to sources
‘A’-‘E’, identified in the PCA. As an example, the PCA identified source
A (long-range transport) contains both primary combustion particles and
secondary particles.
I/O ratios for other elements Si, Al, Cl and K equal or exceed 1 during all seasons, sometimes quite considerably, with the highest ratios in fall followed by summer, pointing to substantial indoor sources. Such an indoor source may well have arisen through the re-suspension of tracked in dust and soil particles from floors, sofas and indoor surfaces. I/O ratios above 1 suggested an indoor/personal source for Cl in addition to sea salt. For indoor and personal exposure samples marine Cl was separated from indoor source. In the PCA an additional indoor/personal phosphorus/chlorine source, probably detergents, was identified, and its contribution calculated as P2O5.
The main contributors to PM2.5 in home and work microenvironments were COPM and secondary PM. The relative contributions of these source types in an urban background station (Vallila) in Helsinki (46% and 43% in Ojanen et al. (1998)) agreed with the results of the current study for residential outdoor samples (35% and 47%). Similar source contributions to fine particles in ambient air were reported by Schauer et al. (1996) in Southern California and Birmingham (Harrison et al,. 1997). ApSimon et al. (2001) modeled atmospheric transport of primary PM2.5 across Europe, obtaining 2 µg/m3 for the Helsinki area. If this is valid estimate for the current study period (1996-1997), half of the primary particles in residential outdoor air PM2.5 are from long distance sources. Results of the current work also agree with those reported by Pakkanen et al. (2001) from Helsinki, where they found 46% of the PM2.3 originating from local sources respectively. Sea salt contributions in this study also agreed well with the results reported by Ojanen et al. (1998) (2 % and 3 % respectively).
The higher soil contributions in the current work, compared to those reported by Ojanen et al. (1998), were probably due to sampling close to people’s residences compared to a central fixed site location. Source contributions in workplaces were similar to residential indoor environments as the majority of participants worked in offices and services. Kukkonen et al. (1999) has reported that ambient air quality standards have been exceeded frequently in Finland during so called “spring dust episode” due to re-suspended soil dust from the roads. Only the average contribution of COPM was higher in workplaces than in residential indoor air. This was probably due to higher traffic contribution as workplaces were measured during the daytime, while residential indoor environments were measured during the evening and night, and workplaces are generally closer to dense traffic than homes.
Figure 1. Contributions of sources to PM2.5 in residential outdoor
and indoor, in workplace and in personal 48-h exposure samples for non-ETS
exposed participants.
4 Conclusions
For adult urban population not exposed to tobacco smoke:
-Primary combustion, secondary particles and soil were dominant source
types for PM2.5 mass concentration in all microenvironments and in personal
exposure samples
-Elemental I/O ratios were dependent on source and vary between seasons
due to widely differing ventilation characteristics in the northern latitudes,
thus although overall PM2.5 I/O ratios are close to unity, the elemental
and source composition varies much more widely.
-Indoor re-suspension of soil particles was shown to be substantial.
An unidentified source associated with phosphorus and chlorine was probably
related to cleaning products and detergents.
-Contributions of sources to PM2.5 personal exposures were closely
approximated by indoor and workplace microenvironments. Ambient PM2.5 was
more dominated by long range transport and 'combustion and other primary
PM' and less by mineral dust than personal exposure.
-Population exposure assessment based on ambient fixed site monitoring
over-estimates exposures to ambient sources like traffic and long-range
transport. It does not account for the contribution of indoor sources.
Thus, the concentration – response functions with ambient fixed site data
may underestimate the true association between personal exposure to outdoor
origins and the respective health effects.
5 References
Airborne Particles Expert Group (APEG). Source apportionment of airborne particulate matter in the United Kingdom. London: Air and Environment Quality Division, Department of the Environment, Transport and the Regions 1999. Available also at http://www.environment.detr.gov.uk/airq.
Ålander T, Leskinen A, Ruotsalainen I, Raunemaa T, Rantanen L, Mikkonen S. The structure and chemical composition of combustion aerosols formed in internal combustion engines. Espoo Finland: VTT Energy, 1989. Project report, MOBILE 230T-2.
ApSimon HM, Gonzales del Campo MT, Adams HS. Modelling long-range transport of primary particulate material over Europe. Atmos Environ 2001;35:343-52.
Brook J.R., Dann T.F., Burnett R.T. The relationship among TSP, PM10, PM2.5 and inorganic constituents of atmospheric particulate matter at multiple Canadian locations. J. Air & Waste Manage. Assoc. 1997a;47:2-19.
Brook JR, Dann TF, Burnett RT. The relationship among TSP, PM10, PM2.5, and inorganic constituents of atmospheric particulate matter at multiple Canadian locations. J Air Waste Manage 1997;47:2-19.
Chow J.C., Watson J.G., Fujita E.M., Lu Z., Lawson D.R., Ashbaugh L.L. Temporal and spatial variations of PM2.5 and PM10 aerosol in the southern California air quality study. Atm. Envir. 1994;28:2061-2080.
Cooper JA, Watson JG. Receptor oriented methods of air particulate source apportionment. JAPCA J Air Waste Ma 1980;30:1116-25.
Costa D.L., Dreher K.L. Bioavailable transition metals in particulate matter mediate cardiopulmonary injury in healthy and compromised animal models. Environ Health Perspect. 1997;105 Suppl 5:1053-60.
Countess RJ, Wolff GT, Cadle SH. The denver winter aerosol: a comprehensive chemical characterization. JAPCA J Air Waste Ma 1980;30:1194-200.
Donaldson K., Brown D.M., Mitchell C. Free radical activity of PM10:
iron mediated generation of hydroxyl radicals. Environ Health Perspect.
1997;105:1285-1289.
Dreher K.L., Jaskot R.H., Lehmann J.R., Richards J.H., McGee J.K.,
Ghio A.J., Costa D.L. Soluble transition metals mediate residual oil fly
ash induced acute lung injury. J Toxicol Environ Health. 1997;50:285-305.
Dusseldorp A., Kruize H., Brunekreef B., Hofschreuder P., de_Meer G., Oudvorst A.B. Association of PM10 and airborne iron with respiratory health of adults living near a steel factory. Am J Respir Crit Care Med. 1995;152:1932-1939.
Eldred R.A., Cahill T.A., Flocchini R.G. Composition of PM2.5 and PM10 aerosols in the IMPROVE network. J. Air Waste Manage. Assoc. 1997;47:194-203.
Gordon, GE. Receptor models. Environ Sci Technol 1980;14:792-00.
Hänninen O.O., Koistinen K.J., Kousa A., Keski-Karhu J., Jantunen
M. Quantitative analysis of environmental factors in differential weighing
of blank Teflon filters. JAWMA. 2002; In print.
Harrison RM, Deacon AR, Jones MR, Appleby RS. Sources and processes
affecting concentrations of PM10 and PM2.5 particulate matter in Birmingham
(U.K.). Atmos Environ 1997;31:4103-17.
Henry CR, Lewis CW, Hopke PK, Williamson HJ. Review of receptor model fundamentals. Atmos Environ 1984;18:1507-15.
Hoek G., Schwartz J.D., Groot B., Eilers P. effects of ambient particulat matter and ozone on daily mortality in Rotterdam, the Netherlands. Arch Environ Health. 1997;52:455-463.
International Standard Organization (ISO). Ambient Air - Determination of a Black Smoke Index, International standard 9835, Geneva, Switzerland: ISO, 1993.
Jantunen M.J., Hänninen O., Katsouyanni K., Knöppel H., Kuenzli N., Lebret E., Maroni M., Saarela K., Srám R., Zmirou D. Air pollution exposure in European cities: The "Expolis" study. J. Expos. Anal. Environ. Epidem. 1998;8:495-518.
Kukkonen J, Salmi T, Saari H, Konttinen M, Karstastenpää R. Review of urban air quality in Finland. Boreal Environ Res 1999;4:55-65.
Long CM, Suh HH, Catalano PJ, Koutrakis P. Using time- and size- resolved particulate data to quantify indoor penetration and deposition behavior. Environ Sci Tech 2001;35:2089-99.
Malm WC, Sisler JF, Huffman D, Eldred RA, Cahill TA. Spatial and seasonal trends in particle concentration and optical extinction in the United States. J Geophys Res 1994;99:1347-70.
Müller K. A three year study of aerosol in northwest Saxony (Germany). Atm. Envir. 1999;33:1676-1685.
Nyberg F., Pershagen G. Epidemiologic studies on the health effects of ambient particulate air pollution. Sandinavian Journal of Work, Environment & Health. 2000;26:49-89.
Ojanen C, Pakkanen T, Aurela M, Makela T,Merilainen J, Hillamo R, Aarnio P, Koskentalo T, Hamekoski K, Rantanen L, Lappi M. The size distribution of respirable particles, their composition and sources in the metropolitan area of Helsinki, Finland. Helsinki: YTV, 1998. Pääkaupunkiseudun julkaisusarja C :7 (in Finnish).
Pakkanen TA, Kerminen V-M, Korhonen CH, Hillamo RE, Aarnio P, Koskentalo
T, Maenhaut W. Use of atmospheric elemental size distributions in estimating
aerosol sources in the Helsinki area. Atmos Environ 2001, in press.
Pope C.A., Docckery D.W., Schwartz J. Review of epidemiological evidence
of health effects of particulate air pollution. Inhalation Toxicology.
1995;7:1-18.
Roemer W., Hoek G., Brunekreef B., Clench-Aas J., Forsberg B., Pekkanen J., Schutz A. PM10 elemental composition and acute respiratory health effects in European children (PEACE project). Pollution Effects on Asthmatic Children in Europe. Eur Respir J. 2000;15:553-9.
Röösli M., G.Theis, Künzli N., Staehelin J., Mathys P., Oglesby L., Camenzind M., Braun-Fahrländer C. Temporal and spatial variation of the chemical composition of PM10 at urban and rural sites in the Basel area, Switzerland. Atmos Environ. 2001;35:3701-3713.
Schauer JJ, Rogge WF, Hildemann LM, Mazurek MA, Cass GR, Simoneit BRT. Source apportionment of airborne particulate matter using organic compounds as tracers. Atmos Environ 1996;30:3837-55.
Seinfeld JH, Pandis SN. Atmospheric chemistry and physics: from air pollution to climate change. New York (NY): John Wiley & Sons Inc, 1998:444.
Spurny K.R. Aerosol air pollution - its chemistry and size dependent
health effects. J. Aerosol Sci. 1996;27, Suppl. 1:473-474.