STUDY FOR PARTICULATE SAMPLING, SIZING AND ANALYSIS FOR COMPOSITION by A M King 1, A M Jones 1, S R Dorling 2, J R Merefield 3, I M Stone 3, K Hall 4, G V Garner 4, P A Hall 4 and B Stokes 5 1 IMC Technical Services Ltd, 2 University of East Anglia, 3 University of Exeter, 4 Hall Analytical Laboratories Ltd and 5 CRE Group Ltd SUMMARY The aim of the project was to identify the origin of air pollution particulate matter, by simultaneously examining the size distribution within, and composition of, PM 10 and PM 2.5 particulate matter from a number of rural, semi-rural and urban UK locations. Two previous studies (ETSU Report Nos N/01/0032/REP, N/01/0033/REP), demonstrated that ambient background levels of PM 10 particulate matter form the majority component of urban levels. Exceedences of the 50µg/m 3 United Kingdom National Air Quality Standard (AQS), across the UK, were shown to be linked predominantly to national and trans-boundary phenomena (King and Dorling, 1997, 1998). These studies had been restricted to the assessment of PM 10 particulate matter, and little chemical characterisation had been undertaken. The Airborne Particles Expert Group (APEG) have come to similar conclusions on the importance of transboundary particulate matter, and have proposed changes to the National Air Quality Strategy to accommodate these findings. DETR have highlighted the implications of moving to a standard based on a finer size dust fraction, and the on-going debate over the suitability of TEOM monitors for statutory sampling. The report is comprised of four parts. Firstly, it reviews the findings of three studies, previously carried out in three distinct areas of the UK: North-East, South-East and South Wales. Concurrent archive PM 10 samples, collected in the seven months October 1996-April 1997, were available for comparative analyses. Secondly, new samples of PM 10 and PM 2.5 material, and information on meteorological conditions, were collected over the six-month period 24 March-29 September 1998, at two rural locations, in east and south-east England, chosen to target the influence of Continental air-masses, in relation to UK emissions and Prevailing winds. Concurrent PM 10 samples were also selected from South Wales, Cornwall, and from three Automatic Urban Network (AUN) sites. Samples were collected using TEOM monitors. Thirdly, the wide range of predominantly rural samples was then analysed for chemical composition, using a range of alternative and independent organic and inorganic analytical techniques, against a backdrop of meteorological analyses. Finally, results of the separate analyses were compared and analysed. i
Sampling Site Period Mean PM 10 25 Mar 28 Sep mg/m 3 (TEOM) Proportion of Stoke Ferry Valid days (190 max) N % Rural sites Wales 4 1 17.7 103 157 Wales 5 1 14.1 82 118 Wales 6 1 14.5 85 138 St Margaret s 22.9 134 187 (21.9 ) 3 ( 128) 3 Annual Mean PM 10 1998 mg/m 3 (TEOM) (180) 3 Stoke Ferry 17.1 100 184 Rochester 17.3 101 173 17.0 Mean Rural 17.3 101 AUN Urban Background Norwich 19.1 112 185 19.6 Sutton Roadside 2 20.1 117 183 20.8 Southampton 19.7 114 184 21.0 Liverpool 20.9 122 182 22.2 Leeds 20.2 117 181 21.9 Wolverhampton 18.0 105 184 18.4 Manchester 20.2 118 185 21.2 London Bexley 18.9 110 180 19.0 Mean Urban 19.6 114 20.5 PM 10 means calculated as average of all valid daily means (>75% data capture) within period. 1. South Wales sites not fully rural, and started April 98 onwards - see site details in Table 1 of main report. 2. Sutton Roadside is classified as an urban kerbside, rather than urban background site. 3 Recalculated values, rejecting six outliers identified (Figure 40 of main report) in St Margaret s data. Table 1 Urban contribution to PM 10 at eight UK cities in 1998, in comparison with six rural sites. Table 1, (Table 23 of the main report) lists summary PM 10 statistics for rural and AUN urban background sites, for the 1998 monitoring period. PM 2.5 concentrations averaged 14.6µg/m 3 at St Margaret s, and 11.6µg/m 3 at Stoke Ferry. Particulate levels at St Margaret s were thus approximately 30-35% higher, on average, than those at Stoke Ferry, in both size fractions. PM 2.5 concentrations were, on average, approximately 65% of simultaneous PM 10 levels, although the PM 2.5 /PM 10 ratio was found to vary widely from day to day (in the range 40%- 100%) and values of the ratio were closely correlated between St Margaret s, Stoke Ferry and Rochester AUN data. ii
Soluble Matter, Major ions as proportion of ACCU sample masses 100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 All High All low SF St M All 2.5 All10 0.0 Cl NO3 SO4 Na K Mg Ca Total ions, % Figure 1 Major ions, St Margaret s / Stoke Ferry, proportion of ACCU sample mass - High/ Low concentrations, by site, or according to size fraction. Proportion of Insoluble particles, by number, all ACCU samples 100.00 All High 90.00 All low 80.00 SF St M 70.00 All 2.5 All10 60.00 50.00 40.00 30.00 20.00 10.00 0.00 Soot No. ** Soil Si/quartz Flyash C/biol Others **Soot is assessed independently of other insoluble matter, on a relative scale by area. Figure 2 Proportion of Insoluble matter, by number, St Margaret s / Stoke Ferry ACCU samples - High/ Low concentrations, by site, or according to size fraction. Figures 1 and 2 (Figures 54 and 64 of main report) summarise the inorganic composition of the 1998 rural particulate samples, in terms of the proportions of major soluble ions, and classification of insoluble matter. The dominant ions are sulphate, nitrate and sodium, the dominant insoluble class was Carbon/biological material. Relative soot levels are included in the second figure, but were assessed independently. The figures illustrate there was little iii
difference, in average terms, between the composition of particulate matter, by overall concentration, rural location, or size fraction. However, significant trends were found within the analyses, which are detailed in the full report. Based on the analyses of targetted samples, collected at two rural locations over a six-month period, plus further analyses of data collected across the UK over an earlier twelve-month period, ambient particulate matter was confirmed to be a complex, ever-changing cocktail of a variety of components. Application of a number of complementary techniques, to the analysis of samples of PM 2.5 and PM 10 particulate matter, mainly collected in rural locations, provided a number of key conclusions. Continental air-masses, already associated (ETSU, 1996) with the majority of breaches of the UK PM 10 Standard for Air Quality, were, relatively, found to contain proportionately more volatile material than normal levels, as did the PM 2.5 fraction over PM 10. There is general acceptance that the trans-boundary component of PM 10 is significant, and is problematic in management terms. Consequently, it is suggested that care should be taken in the drafting of any gravimetric standard for PM 2.5, until the importance of regional scale transport of particulate material is better understood. Largely, the proportionate mix of components was independent of absolute concentration, size fraction and wind sector. However, since the concentration of individual components varied independently it was concluded that caution should be observed in the use of a single component, such as sulphate, either as an indicator of total levels, or of the contribution of another component. Soluble salts accounted for approximately 50% of the airborne mass, at all locations. Urban PM 10 samples were characterised by relatively higher soot and soil content than corresponding rural samples. Aerosol sulphate was the dominant ion present, and concentrations were highest in Continental air-masses. Aerosol nitrate concentrations showed a clear bias towards Continental origins, but also indicated a significant contribution from the UK. Aerosol chloride values demonstrated the maritime influence at all sites, both coastal and inland. Primary particles in the form of flyash showed a regional compositional variation. Carbon/biological data highlighted the different outcomes resulting from different particle size cut-offs and TEOM and ACCU systems. Only a minority part of the total organic matter was clearly quantified, as Polycyclic Aromatic Hydrocarbons (PAHs), by the use of HPLC-MS on rural samples. PAH masses iv
were generally close to the detectable limit (5ng), and in total comprised less than 0.02% of the mass of sample assessed (>1mg). ETE identified a wider range of organic materials than did the HPLC technique. Localised events associated with the release of organic material (harvesting crops) were easily recognised, and generated data with significant compositional differences. Accurate, and easily achievable, correction could be made to TEOM data, by changing default factors in the instrument s software. The corrections (a gain factor of 1.5, and an off-set of minus 3) would not only redress the current over-sampling found at average concentrations, but also the more significant under-sampling confirmed at higher levels, close to the Air Quality Standard. As a result of the above conclusions, recommendations are made in regard to the direction of future research. In particular, a better understanding of the source apportionment of UK urban particulate matter is required and should be accompanied (preferably preceded) by rural studies, which should include sites on the near continent. The benefits of a multidisciplinary approach to source apportionment work is stressed. Improvements in confidence of the relative source strengths of particulate matter depend upon the development of a fully quantitative basis for both organic and insoluble inorganic chemical analysis techniques. Inter-comparisons of rural and urban PM 10 measurements do not support the present weighting given to the significance of local traffic as a source of PM 10. Improved verification of models (currently based on this incorrect assumption) is required, using empirical data. The requirements/preferences of an individual analysis method (filter size, sample collection conditions, pre and post handling/storage of filters) should be addressed in the design of any future study. Further work is required in assessing to what extent adsorption of particulate material onto a filter occurs before, during or after the sampling process. Although sulphate dominates in terms of its contribution to the total soluble inorganic component of PM 10 and PM 2.5, and was found to correlate closely with PM 2.5 levels, the relative proportion does vary. Caution should be observed in utilising sulphate as a surrogate for total secondary particulate, especially in relation to the prediction of future trends. Nevertheless, such an approach was confirmed as being useful, for prediction purposes, until a more precise model is developed. v