Exploring the Variation between EC and BC in a Variety of Locations

Despite intensive research over the past three decades, a generally accepted standard method to measure black carbon (BC) or elemental carbon (EC) still does not exist. Data on BC and EC concentrations are method specific and can differ widely. This work was motivated by the lack of any prior study that established the variability between these two measures of carbonaceous particulate matter. Measurements of BC and EC were performed at different locations across Asia and the South Pacific in both urban and suburban locations. Filter samples were collected during the winter of 2007 to the winter of 2010 and analyzed for both BC and EC. EC was measured using the Interagency Monitoring of Protected Visual Environments (IMPROVE_A) protocol. Black carbon was measured by the EELS reflectometer (Diffusion Systems, Ltd). Bangladesh had the highest correlation coefficient of 0.93. Bangkok, Thailand on the other hand had the lowest correlation coefficient of 0.34. A review of previously reported source apportionment of BC concentrations in these locations showed that New Zealand had the highest percentage (82%) of BC from biomass while Mongolia had the lowest percentage of 3.1%. The fraction of BC emissions from diesel vehicles was found predominant in Mumbai, India with values as high as 80%. Mongolia had the lowest emission of BC from diesel vehicle (5.4%) with coaland biomass-combustion being the dominant sources.


INTRODUCTION
Black carbon (BC) and elemental carbon (EC) in airborne particulate matter originate from the incomplete combustion of carbonaceous fuel.They are used to describe aspects of ambient particulate matter.However, they are not interchangeable because they are each defined by the way they are measured.These two particulate matter constituents are important components of the atmospheric aerosol because of their light absorbing characteristics and their possible health effects.Light absorption results in a positive (warming) forcing that exceeds that of methane (the third most important greenhouse gas) on a global scale (Jacobson, 2001;Ramanathan and Carmichael, 2008).These light absorbing particles can produce a large indirect positive forcing by enhancing cloud evaporation in the tropics (Ackerman et al., 2000).
Characterizing light absorbing aerosols is conceptually ambiguous (Watson et al., 2004(Watson et al., , 2005)).However, black carbon (BC) and elemental carbon (EC) are not measures of the same properties of particulate matter, even though, these two carbonaceous species are often well correlated (Jeong et al., 2004).Thus, careful definition of these quantities is required to allow proper discussion of their sources and how they behave in the atmosphere.Black carbon (BC) is an optical measurement that is commonly used to denote the extent of light-absorption by the sample.It is not a direct measure of the carbon concentration.In most environments, BC is produced by the incomplete combustion of fossil and bio-fuels that dominates particulate light absorption (Horvath, 1993).Elemental carbon (EC) represents thermally refractory carbon with a graphitic structure measured by oxidation and detection of the evolved gas.Emissions from diesel heavy trucks are the major source of EC in urban areas of the United States (Gray and Cass, 1998).
Optical attenuation methods such as the aethalometer (Hansen et al., 1984;Lou et al., 2005;Park et al., 2010;Lili et al., 2011) or the EELS Reflectometer (Diffusion Systems, Ltd) (Edwards et al., 1983;Biswas et al., 2003;Begum et al., 2007Begum et al., , 2011;;Cohen et al., 2011;Wimolwattanapun et al., 2011;Begum et al., 2012) have also been commonly used to determine BC.The EELS Reflectometer provides a measurement of "British Smoke" that dates back to the famous 1952 London fog episode (Edwards et al., 1983).Optical methods are easy to operate and inexpensive although they only measure light absorption and estimate BC using an assumed mass absoption coefficient.
There have been a limited number of comparisons of BC/EC relationships particularly for specific size fractions of the ambient aerosol.Early comparisons like that provided by Edwards et al. (1983) examined total suspended particulate (TSP) samples where the wide range of particle size and composition made direct mass comparisons difficult.Cyrys et al. (2003) showed good site specific correlations between BC and EC using data from Germany, the Netherlands and Sweden.Jeong et al. (2004) also found strong correlations between EC and BC in Rochester, NY and Philadelphia, PA, but again with different slopes in the two locations.The slope in Philadelphia changed during the intrusion of a large wildfire aerosol into the area.Nordmann et al. (2009) compare black and elemental carbon measurements in Europe again supporting the variability of the EC-BC relationships.
The EELS reflectometer has been widely used to provide a measure of BC typically using a single fixed mass absorption coefficient.The purpose of this work is to explore the variability in the relationship between black carbon and elemental carbon in a variety of location across Asia and the South Pacific and to provide information regarding the potential errors in the estimation of BC mass with a fixed mass absorption coefficient.

METHODS
Samples were collected at various times in different locations as documented in Table 1.The filters were sampled over the averaging times given in the table.Parallel sampling with identical Gent samplers (Hopke et al., 1997) was performed, except in Australia and Bangladesh.In Australia and Bangladesh, ASP samplers and PM 2.5 MiniVol (AirMetrics) samplers, respectively, were used.The fine filter in one sampler was a standard 0.45 μm nuclepore.A prebaked quartz fiber filter was used as the fine filter in the second sampler.The specific analytical techniques employed in this study are described below.

Thermal Carbon Analysis
Elemental carbon on the quartz filter was measured by the Interagency Monitoring of Protected Visual Environments (IMPROVE_A) protocol with a DRI 2100A system.Details of this method have been presented by Chow et al. (2007).

Black Carbon Analysis
Black carbon was measured using EELS reflectometers (Diffusion Systems, Ltd) in each country.Edwards et al. (1983) describes the basis for this measurement and Biswas et al. (2003) described its application to nuclepore filter This value was established by collecting samples of candle soot that were weighed as well as their optical density measured for white light reflectance with the EELS reflectometer (Begum et al., 2004).
The BC and EC detection limits were 0.02 μg/m 3 and 0.2 μg/m 3 , respectively, based on the approach described by the U.S. Environmental Protection Agency (40CFR136, Appendix B). Results of our replicate sample analyses showed a typical relative percentage difference of 5%.Additionally, all of the measured concentrations were large compared to the MDL values across all locations with exception of Australia and New Zealand.

Source Apportionments
To better understand the differences in mass absorption coefficients across the region, previously performed and reported source identification and apportionments of BC across all of the locations have been reviewed.Each participating group has independently analyzed PM 2.5 compositional data from samples collected at their monitoring sites and presented their results.Table 3 summarizes the source apportionments for BC and provides references to the literature in which the results have been reported.

RESULTS AND DISCUSSION
The results from the analyses of both EC and BC samples from a variety of locations are shown in Fig. 1.In most locations, the sample collection time was 24 hours with the exception of Thailand (18 hours) and Australia (72 hours, alternating with 96 hours).The total number of samples collected varied from site to site.The lowest number of samples was 10 from Dhaka, Bangladesh and the suburban site in Thailand while the highest number of sample collection was 34 from New Zealand.The lowest value for both elemental and black carbon concentration was 0.05 μg/m 3 and 0.01 μg/m 3 respectively from Australia.The highest values for EC and BC concentrations were 62.23 μg/m 3 and 53.22 μg/m 3 .The reason for this significant difference could be attributed to differences in the population densities effectiveness of air quality management systems.The sites in Australia and New Zealand are rural with low population densities whereas Dhaka is a megacity in a developing country with many substantial sources of carbonaceous materials (Begum et al., 2012).
The overall average concentrations of EC and BC were calculated as 6.21 μg/m 3 and 7.29 μg/m 3 , respectively, while their standard deviation concentrations were calculated as 2.01 μg/m 3 and 1.95 μg/m 3 .The results of the average concentrations show that BC was estimated to exceed EC by 17.39%.On the other hand, the variability of the EC results as indicated by the standard deviation of the concentrations exceeds BC by 3%.The variation in EC and BC concentration indicates the possibility of an error in the assumed mass absorption coefficient in specific locations.Minimum, maximum and average concentrations of elemental and black carbon as well as their ratios are given in the Table 2.The average BC/EC ratios were found to range from 0.47 to 3.56.Overall, BC was found to be more abundant at six sites in five different countries while EC is higher at the remaining three sites in three different countries.
To explore the variation in black and elemental carbon concentrations, regression analysis was performed for these methods.It can be seen from Fig. 1 that two of the regression coefficients were low with considerable scatter in the data.The lowest correlations were found for the two Bangkok, Thailand sites.The highest correlations were found for the samples collected in Bangladesh and in New Zealand.
Most of the slopes of the lines in Fig. 1 are reasonably close to 1 supporting the choice of the specific mass absorption coefficient as being a reasonable estimate.However, the two sites in Thailand have much lower slopes suggesting the need to reassess the absorption coefficient value.However, given the low correlation, it may mean that there is a sufficient variation in the nature of the source emissions that Table2.Minimum, maximum, average concentrations of IMPROVE EC and BC and their ratios.Cohen et al., 2011;b. Begum et al., 2011;c. Kothai et al., 2011;d. Chung et al., 2006;e. Davy et al., 2011;f. Rahman et al., 2011;g. Davy, private communication, 2011;h. Wimolwattanapun et al., 2011.there are at least two elemental carbon sources with varying light absorptivity as suggested by Jeong et al. (2004).In Malaysia, the slope is much greater than 1 but again with a very low correlation coefficient suggesting the possibility of multiple sources contributing to the BC concentrations.
Particulate matter concentrations and compositions differ from location to location (Hopke et al., 2008).The BC source apportionments at all of the locations revealed that New Zealand had highest percentage of black carbon from biomass with values as high as 82% while Mongolia had the lowest percentage of 3.1%.Emissions from diesel vehicles to black carbon concentration were found to be dominant in Mumbai, India with values up to 80% (Kothai et al., 2011).Alternatively, Mongolia has the lowest emission of BC from diesel vehicle as 5.4%.The majority of their BC emissions are from home heating and cooking through the combustion of coal and biomass (Davy et al., 2011).In general, there is a mixture of biomass burning and traffic as the primary sources of the soot.
The variations in correlation of BC and EC concentrations can be attributed to a number of factors such as differences in particle size distribution, mixing state, and chemical composition leading to variations in light absorption efficiency.This variability results from difference in source contributions such biomass combustion, vehicular emissions, etc. Jeong et al. (2004) showed that wood smoke particles had lower light absorptivity than traffic generated particles.They also found that the degree of atmospheric processing changed the absorptivity where EC collected within 2 m of a roadway was twice as light absorbing as that collected a 1 km from a large interstate highway.Thus, although the actual light absorption can be measured with very good precision and accuracy, the variability in the absorptivity per unit mass results in the variability in the EC/BC relationships shown in Fig. 1.

SUMMARY
Collection and analysis of airborne PM 2.5 samples made in this study indicate that EC and BC concentrations can vary substantially due to wide mixture of sources and degree of atmospheric processing from one location to another location across different countries .This study was performed during the winter of 2007 to the winter of 2010.Bangladesh had the highest correlation coefficient of 0.93, followed closely by New Zealand with 0.91.The two sites in Thailand and the site in Malaysia had low correlation coefficients suggesting a broader range of possible BC sources.Most sites appeared to obtain good results using a fixed mass absorption coefficient of 9.7 m 2 /g.Results obtained from the sites with the low correlation coefficients between EC and BC showed a substantial variability in the mass absorption coefficient, likely the result of multiple sources of black carbon with quite different absorptivities.

Fig. 1 .
Fig. 1.Relationships between BC and EC concentrations across Asia and Pacific regions.

Table 1 .
Logistics of the field operations for samples collection at multiple sites.

Table 3 .
Source contributions to mass concentrations of black carbon.