Estimating the secondary organic aerosol contribution to PM2.5 using the EC tracer method.

Juan C. Cabada and Spyros N. Pandis, Department of Chemical Engineering, Carnegie Mellon University, 5000 Forbes Ave., Pittsburgh, PA 15213

Ramachandran Subramanian and Allen L. Robinson, Department of Mechanical Engineering, Carnegie Mellon University, 5000 Forbes Ave., Pittsburgh, PA 15213

Andrea Polidori and Barbara Turpin, Department of Environmental Sciences, Rutgers University, 14 College Farm Road, New Brunswick, NJ 08901

Abstract

The EC tracer method is applied to a series of measurements by different carbonaceous aerosol samplers in the Pittsburgh Air Quality Study (PAQS) in order to estimate the concentration of secondary organic aerosol. High-resolution measurements (2-6 hrs) and daily averaged concentrations were collected during the summer 2001 intensive (July 1 to August 4, 2001) and are used for the analysis. The various samplers used during PAQS show differences in the measured concentrations of OC and EC due to the different sampling artifacts and sampling periods.

A systematic approach for the separation of periods where SOA contributes significantly to the ambient OC levels from the periods where organic and elemental carbon concentrations are dominated by primary emissions is proposed. Ozone is used as indicator of photochemical activity to identify periods of probable secondary organic aerosol production in the area. Gaseous tracers of combustion sources (CO, NO, and NOx) are used to identify periods where most of the OC is primary. Periods dominated by primary emissions are used to establish the relationship between primary OC and EC, a tracer for primary combustion-generated carbon for the different sets of measurements for July 2001. Around 35% of the organic carbon concentration in Western Pennsylvania during July of 2001 is estimated to be secondary in origin.

1. Introduction

Carbonaceous aerosol is an important constituent of the PM2.5 (particulate matter with aerodynamic diameters less than 2.5 microns) mass in most of the U.S. Between 10 to 65% of the fine particulate mass has been identified as carbonaceous material for various regions of the country (Gray et al., 1986; Turpin et al., 1991; Seinfeld and Pandis, 1998; Tolocka et al., 2001; Lim and Turpin, 2002). Aerosol carbon is commonly classified as organic (OC) and elemental carbon (EC). OC can be directly emitted to the atmosphere in the particulate form (primary) or can be produced by gas to particle conversion processes (secondary). EC is emitted from combustion sources. Since primary OC and EC are mostly emitted from the same sources, EC can be used as a tracer for primary combustion-generated OC (Gray, 1986; Turpin and Huntzicker, 1995; Strader et al., 1999). The formation of secondary organic aerosol (SOA) increases the ambient concentration of OC and the ambient OC/EC ratio. OC to EC ratios exceeding the expected primary emission ratio are an indication of SOA formation. For Southern and Central California between 30 to 80% of the total OC has been identified as secondary in summer (Gray et al., 1986; Pandis et al., 1992; Hildemann et al., 1993; Turpin and Huntzicker, 1995; Schauer et al., 1996).

The relationship between primary OC and EC depends also on the sampling and analyses techniques used to determine the ambient OC and EC concentrations. Sample collection (i.e. use or not of denuders, filter face velocities, etc.) and different analysis techniques (i.e. Thermal Optical Transmittance vs. Thermal Optical Reflectance) affect the reported concentrations for OC and EC (Countess, 1990; Birch, 1998; Chow et al., 2001; Schmid et al., 2001). Sampling carbonaceous particulate matter from the atmosphere is challenging because of interferences from gaseous material that is adsorbed on the filters or evaporation of the collected organic material during sampling (Turpin et al., 1994; Fitz, 1990; Hering et al., 1990). Different sampling arrangements (e.g., using backup quartz filters, placing denuders upstream of the filter to remove organic gases, etc.) have been proposed in order to reduce and/or measure and correct for the positive and negative artifacts that affect the measured carbonaceous concentrations (Turpin et al., 2000).

Primary ratios of OC to EC vary from source to source and show temporal and diurnal patterns (Gray, 1986; Cabada et al. 2002) but since EC is only emitted by combustion sources, gaseous tracers of combustion (CO, NO, NOx) can be used to determine periods dominated by primary aerosol emissions. Ozone is an indicator of photochemical activity and it also can be used as a tracer for periods where secondary organic aerosol production is expected. In this case, increases in the OC to EC ratio correlated to ozone episodes are indicative of SOA production.

In this work a relationship between primary OC and EC is established for each of the different types of measurements and artifact estimation approaches. An algorithm is proposed for the determination of the primary OC/EC ratio and secondary organic aerosol concentrations are estimated. SOA results based on high-resolution and the daily-averaged samples are compared. The effect of sampling frequency on the estimates of the primary ratios is also discussed.

2. Experimental Methods and Equipment

The Pittsburgh Air Quality Study (PAQS) main site was located in Schenley park on the top of a hill just outside of Carnegie Mellon University campus, around three miles to the east of downtown Pittsburgh. The Pittsburgh supersite operated three different samplers for collecting carbonaceous aerosol (one undenuded and two denuded samplers). The undenuded sampler and a denuded in-situ analyzer collected samples every 2-6 hr, while the denuder-based sampler collected daily samples, during the 2001 summer intensive (July 1 to August 4, 2001).

Quartz fiber filters (47 mm Pallflex, QAOT), Teflon filters (2 mm pore, Whatman 7592-104) and carbon-impregnated filters (Schleicher and Schuell, GF-3649) were used to sample carbonaceous material in three different samplers. Quartz fiber filters were baked at 550°C for more than 12 hours and stored in previously cleaned glass jars until sampling and analysis. Carbon-impregnated filters were baked at 370˚C for more than 3 hours in a nitrogen atmosphere.

Undenuded sampler

PM2.5 carbonaceous aerosol samples were collected on quartz fiber filters using filter packs in a non-denuded line. This sampler consisted of two parallel lines, the first line holding a quartz fiber filter followed by a backup quartz filter and the second line having of a Teflon filter followed by a backup quartz fiber filter (Figure 1). The two backup quartz fiber filters are used to estimate the positive and negative artifact (Turpin, 2000). Five samples a day, with sampling times between 4 and 6 hrs, were collected during the summer intensive. Samples were collected during 0-6, 6-10, 10-14, 14-18 and 18-24 hrs (all in EST). The filter configuration allows two different estimates of the adsorption artifacts on the front quartz fiber filter. The first correction is done subtracting the OC collected in the backup quartz fiber filter behind the front quartz (QB,F) from the OC collected by the front quartz (QF). This approach assumes that the front quartz filter (QF) collects 100% of the carbonaceous particulate matter (no evaporation) and that both the front and the backup filter adsorb organic gases and reach equilibrium with them during the sampling period. The second correction approach subtracts the OC collected in the backup quartz filter behind the Teflon filter (QB,T) from the OC in the front quartz filter (QF). This approach assumes that the Teflon filter collects all the particles from the sampled flow with 100% efficiency and the backup quartz from this line adsorbs the same quantity of gases as the front quartz fiber filter. The EC concentration reported by the front quartz filter (QF) is used for all datasets from the undenuded sampler.

Denuded sampler

Filter packs holding a quartz filter in front of a carbon-impregnated filter (CIF) were used to collect carbonaceous material from a denuded sampling line (Figure 1). A carbon annular denuder (Novacarb monolith synthetic carbon, Mast Carbon Ltd. Guilford, UK) was used to remove organic gases and minimize the positive artifact in the quartz filter. The CIF organic carbon concentration was intended to correct for evaporation of semi-volatile material from the quartz filter (negative artifact). Sampling frequency for this unit was 24 hrs, from midnight to midnight (EST).

Semi-continuous denuded in-situ analyzer

An in-situ semi continuous carbon analyzer (Sunset Labs, Carbon Aerosol Analysis Field Instrument), similar in design to that described by Turpin et al. (1990), was used to collect and analyze carbonaceous aerosol with sampling periods of 2 to 4 hours (100 to 220 minutes sampling time plus 20 minutes for analysis). Instrument performance and PAQS protocols are described in detail by Lim et al. (2002) and Polidori et al. (2002). A parallel plate diffusion denuder (CIF; Schleicher Schuell, Keene, NH) was placed upstream of a quartz filter, which is mounted inside the analyzer (Figure 1). Cycles of sampling and analysis were alternated in order to determine the ambient concentrations of OC and EC (Lim et al., 2002).

Quartz filters from the filter pack-based samplers were analyzed using a Thermal/Optical transmittance carbon analyzer (Sunset Laboratory Inc., OC-EC Aerosol Carbon Analyzer Model-3) using the temperature steps of the NIOSH protocol (Birch, 1996; NIOSH, 1999) for the determination of OC and EC. Table 1 shows the experimental parameters for the analysis of the quartz and carbon impregnated filter during PAQS. The time length of the different temperatures steps in the method was modified to get a better split between OC and EC (Yu et al., 2002). Carbon impregnated filters were analyzed using a temperature ramp up to 340°C, during 25 minutes under a helium atmosphere. All concentrations of OC and EC were corrected for field blanks. Concentrations reported from the in-situ carbon analyzer were corrected for dynamic blanks generated by sampling with a Teflon filter upstream of the denuder (Polidori et al., 2002).

3. Carbonaceous aerosol measurements

Differences exist among the measured concentrations of organic and elemental carbon collected by the different samplers. For example, adsorption of organic gases on the front quartz filter of the undenuded lines (positive artifact) is evident as the OC measured by this line is higher than that of the denuded samplers. The magnitudes of the positive and negative artifacts depend not only on the sampling method and the atmospheric composition but also on the length of the sampling period. A detailed discussion of the artifacts using these datasets is presented by Subramanian et al. (2002).

Figure 2 shows time resolved concentrations for the different samplers during a 6-day period. Overall samplers indicate similar patterns of OC and EC concentrations, but the OC and EC concentrations from the undenuded line are almost always higher than that of the other two samplers.

The summer intensive averaged concentrations of OC and EC for all types of samplers and all artifact correction approaches at the Pittsburgh supersite is shown in figure 3. Subtracting the measured OC concentration on the backup quartz filters (QB,F, and QB,T) from the front quartz (QF) reduces the OC concentrations by 20% on average for the summer intensive. Subtracting the OC concentration of the backup quartz filter behind the Teflon filter in the parallel line of the undenuded sampler results in an average correction of around 50% on average for the summer intensive. The reported OC concentrations of the two denuded samplers agree within 10%, and give particulate OC concentrations between the QF,B and QB,T corrected undenuded sampler concentrations. However the reported average EC concentrations differ by 50%. EC concentrations reported by the undenuded and denuded in-situ analyzer agree within 10%. A detailed discussion of the potential reasons for these differences is provided by Subramanian et al. (2002).

In this work we examine the effect of the difference in sampler configuration and sampling periods of OC and EC measurements on the SOA estimates applying the EC tracer method.

4. The EC tracer method

The ratio of the ambient concentrations of particulate OC to EC includes information about the extent of secondary OC formation. Ambient OC/EC ratios greater than those characteristic of the primary emissions for a specific area are an indication of secondary aerosol formation. The EC tracer method takes advantage of the fact that primary OC and EC are mostly emitted by the same combustion sources. Primary ratios of OC to EC can be determined form a subset of ambient measurements if a large data set is available and conditions to produce SOA are unlikely (Turpin and Huntzicker, 1995; Strader et al., 1999) or by developing an emissions inventory of the principal sources for an area of interest (Gray, 1986; Cabada et al., 2002).

Assuming that OC primary can be defined by,

(1)

the contribution of secondary OC can be estimated as

(2)

where [OC]p is the primary organic aerosol concentration, [OC/EC]p is the ratio of OC to EC for the primary sources affecting the site of interest and b is the non-combustion contribution to the primary OC (Turpin and Huntzicker, 1995; Strader et al., 1999), [EC] is the measured EC concentration, [OC]S is the secondary organic aerosol contribution to the total OC and [OC] is the measured OC concentration. All of these parameters are time dependent because of the temporal variations in anthropogenic emissions and in meteorology. The application of this method requires measurements of [OC], [EC] and the determination of the [OC/EC]p ratio and the non-combustion primary OC contribution (b) for the area and period of interest (Turpin and Huntzicker, 1995).

4.1 Calculation of the primary OC/EC ratio and intercept

Diurnal variations of the ambient OC to EC ratio were observed for all the high-resolution measurements taken during the summer intensive of 2001 at PAQS. Photochemical activity, meteorology and primary emissions all contribute to these variations. Ozone concentration can be used as an indicator of photochemical activity. Carbon monoxide (CO) and nitrogen oxides (NO and NOx) can be used as tracers of combustion-related primary emissions. The primary ratio and intercept are determined from a dataset by identifying the periods where the ambient concentrations are dominated by primary emissions.

The first step in the determination of the primary OC/EC ratio is the subtraction from the original dataset of the points where rain and the corresponding storms cause significant changes to the OC/EC ratio (Figure 4a). These changes have a variety of causes (removal of aged particles and increased importance of the locally produced ones, preferential removal of secondary OC, etc.). These periods are excluded from the analysis to avoid unnecessary complications.