ECN publication
JOAQUIN - Work Package 1 Action 2 and 3 - Composition and source apportionment of PM10
Staelens, J.; Mooibroek, D.; Cordell, R.; Delaunay, T.; Panteliadis, P.; Weijers, E.P.; Matheeussen, C.; Hoogerbrugge, R.; Dijkema, M.; Monks, P.; Roekens, E.
Published by: Publication date:
ECN Environment & Energy Engineering 16-3-2016
ECN report number: Document type:
ECN-E--15-079 ECN publication
Number of pages: Full text:
181 Download PDF  

This report describes a 14-month measurement campaign of the composition of particle matter (PM10)at five monitoring sites and a mobile station (trailer) in NW Europe. The study was carried out as part of the Joint Air Quality Initiative (Joaquin project, Work Package 1 Action 2 and 3). Background Particulate matter (PM) is a heterogeneous mixture of components resulting from multiple natural and anthropogenic sources. Epidemiological studies attribute the most important health impacts of air pollution to PM, although it is currently still unclear which specific particle properties (such as size and chemical composition) or sources are most relevant to health effects. Current monitoring efforts generally focus on the mass concentration of PM, in line with current air quality legislation, but this generally does not allow the assessment of different sources. To facilitate the development of healthrelevant air quality policies a better understanding of the sources and composition of PM is required. Aims The general objective of the study was to establish a link between the composition of PM10 and air pollution sources and to relate the PM10 composition to toxicological effects. Therefore, the aims were: - To determine the composition and oxidative potential of PM10 collected at five sites (four urban background sites and an industrial site) in NW Europe using a harmonized approach for aerosol sampling and laboratory analyses. - To identify and quantify sources contributing to PM10 in the NW European region using a receptor-oriented source apportionment model. - To investigate the in vitro toxicity of PM10 collected in an urban area and at a background site and to relate this to the chemical characteristics and oxidative potential of PM10. Methods From April 2013 to May 2014, aerosol samples were collected at fixed sites in Amsterdam, Antwerp, Wijk-aan-Zee, Lille and Leicester. Samples were collected daily (24 hour exposure) onto 47 mm quartz filters (Pall Tissuquartz™ filters, 2500 QAT-UP) using a sequential sampler (Leckel SEQ47/50 in Antwerp, Lille and Leicester or Dererenda PNS16 in Amsterdam and Wijk-aan-zee) with PM10 inlet, running at 2.3 m3/h for 24 h per filter. Filters were weighed before and after sampling in order to determine total PM10 collection. For pre- and post-sampling weighing filters were conditioned at 20 ± 1 °C and 50 ± 5 % relative humidity for 48 h, weighed, left for a further 24 h then re-weighed. Instrument comparability was assessed by comparisons at four sites using a mobile trailer equipped with a Leckel sampler. The trailer was also used for short-term campaigns (2-4 weeks) at a second urban background site in Amsterdam, Antwerp and Leicester. The composition of PM10 was determined for (i) filters collected every 6th day during the study period, (ii) 6-9 filters taken every 2nd day during the short campaigns at three sites and (iii) 17 filters per site taken on days with regionally enhanced PM10 levels. Filter parts were analysed for: - Metals (K, Ca, Fe, Zn, Al, Ti, V, Cr, Mn, Ni, Cu, As, Mo, Cd, Sb, Ba and Pb); - Monosaccharide anhydrides (levoglucosan, mannosan, galactosan); - Elemental and organic carbon (EC/OC, NIOSH protocol); - Water-soluble ions (Na+, K+, Mg2+, Ca2+, NH4+, NO3-, Cl-, SO42-); - Generation of free radicals (oxidative potential). The receptor-oriented model EPA-PMF 5.0.14 was used to carry out a source apportionment using the pooled data of the five sites. Results During the common sampling period at the five sites (1 June 2013 to 31 May 2014), the mean ambient PM10 concentration was highest at the site in Wijk aan Zee (annual mean of 25.0 µg/m3) and Antwerp (24.5 µg/m3), intermediate in Lille (22.4 µg/m3) and Amsterdam (20.3 µg/m3) and lowest in Leicester (16.0 µg/m3). Then number of exceedances of the EU day limit value for PM10 (50 µg/m3) ranged from 6/days per year in Lille to 20 days/year in Antwerp. The highest PM levels occurred in March and April. The PMF analysis resulted in a solution with 13 factor profiles, which could be aggregated to eight groups: secondary aerosol; furnace slacks, road wear and construction; sea spray; mineral dust; biomass burning; industrial activities; traffic emissions and brake wear; and residual oil combustion. The largest part of PM10 (40-48%) was explained by nitrate-rich (27-37%) and sulphate-rich (9-13%) secondary aerosol. The second-most important source profiles were related to sea spray and aged sea spray (11-21%). The source profile furnace slacks, road wear and construction was on the average the third most important, but there was more variation between the receptor sites. Also clear traffic and biomass burning source profiles were found. Using the calculated source profiles, dominant source contributions were evaluated for days with known PM10 composition and exceedance of the PM10 daily limit value. The highest contribution during the analysed exceedance days was found for nitrate-rich secondary aerosol, with an average of 35 ± 18 µg/m3 for all sites. The obtained profiles were used to analyse the spatial variation in the contribution per profile to PM10. The factors with the lowest spatial variation were aged sea spray and sulphate-rich secondary aerosol, reflecting the importance of long-range transport into and across the study region. The steel industry (Fe) profile had the largest COD, indicating very different source contributions between the sites. Conditional probability function plots were used to indicate likely directions of sources, while air mass back-trajectories were analysed using the HYSPLIT model. A high contribution of secondary aerosol was associated with wind coming from the northeast-east at all the sites. Levoglucosan was used as a marker of biomass burning. A distinct burning period stretching from November to March was evident at all sites. The site in Lille overall showed the highest levels of wood burning, and the contribution to PM10 in winter averaged 11.6%. The average winter PM10 contribution across all sites was 5.6%, which fell to below 1% in summer. The detrimental effects of burning on air quality are only likely to be evident in the local region unless large organised bonfire events occur when wider scale effects can be observed. The use of monosaccharide anhydride isomer ratios indicated softwoods as the primary combustion source across the sites. Overall, no chemical component of PM was predictive for all studied health endpoints. The harmful nature of levoglucosan on the viability of the bronchial epithelial cells was confirmed in the multiple model. Levoglucosan explained 29% of the variance in cytotoxicity of air samples. Conclusions Chemical characterisation of PM10 is more expensive and time-consuming than monitoring of the PM mass concentration, but gives valuable information on the breakdown of PM10 sources. A large part of PM10 (40-48%) was attributed to secondary aerosol, and in particular to the nitrate-rich aerosol source profile. The importance of nitrate-rich secondary aerosol, with ammonium nitrate as main compound, indicates that decreasing the emissions of its precursor gases (NOx and NH3) can meaningfully decrease the ambient PM10 concentrations in the study region. The source apportionment (PFM analysis) of the pooled data set of the five PM10 sampling sites in NW Europe resulted in an overview of the global sources impacting the sites. Combining the data from all sites still gives information about local sources, provided the contribution is strong enough (e.g. traffic emissions near urban sites, steel industry near an industrial site in Wijk aan Zee). Nevertheless, it is recommended to complement this pooled PMF analysis with site-specific source apportionment modelling. Whilst not being the highest contributor to PM10, wood burning already represents enough of a contribution to PM10 to cause concern in some locations. Contributions are likely to increase in coming years owing to a variety of reasons including increasing costs of conventional heating fuels and the emergence of various government schemes which encourage renewable energy usage. The results of this study gave no evidence that either the PM characteristics or the oxidative potential were a better predictor for the harmful health effect of PM compared to particle mass concentration. Nevertheless, the results increase our understanding of the composition and sources of PM, which can facilitate the development of health-relevant air quality policies.

Back to List