Environ. Sci. Technol. 2002, 36, 754-761
Charring Characteristics of Atmospheric Organic Particulate Matter in Thermal Analysis JIAN ZHEN YU,* JINHUI XU, AND HONG YANG Department of Chemistry, Hong Kong University of Science & Technology, Clear Water Bay, Kowloon, Hong Kong
The charring of organic materials during carbon analysis by thermal methods makes it difficult to differentiate elemental carbon (EC) from organic carbon (OC). Failure to correct for charring results in the overestimation of EC and the underestimation of OC. The charring characteristics and thermal behaviors of aerosol OC are studied by subjecting hexane and water extracts of ambient aerosols to various analysis conditions. The complete evolution of watersoluble organic carbon (WSOC) aerosol materials is found to require a temperature as high as 850 °C and the presence of oxygen. EC would be oxidized under these thermal conditions as well. As a result, thermal methods relying only on temperature for the differentiation of EC and OC would give unreliable OC and EC concentrations. Our investigation also reveals that WSOC accounts for a large fraction (13-66%) of charring, while hexane extractable organic compounds produce little charring. The extent of charring from WSOC, defined as the ratio between pyrolytically generated EC to the total WSOC, is found to increase with the WSOC loading in each analysis when the loadings are below a certain value. This ratio remains constant when the loadings are above this value. This may account for the high variability in the extent of charring among aerosol samples from different locations as well as among samples from a single location collected at different times. Charring is reduced if the residence time at each temperature step in a helium atmosphere is sufficiently long to allow for maximum C evolution at each step. Charring is also influenced by the presence of inorganic constituents such as ammonium bisulfate. For the few tested organic materials, it is observed that ammonium bisulfate enhances the charring of starch and cellulose but reduces the charring of levoglucosan.
Introduction Carbonaceous materials make up a large fraction of ambient aerosols. Three forms of carbon species, namely, organic carbon (OC), elemental carbon (EC), and inorganic carbon, are present in ambient aerosols. The different chemical natures of each of these forms dictate that they each play different roles in various atmospheric processes. As a result, there is great interest in quantifying the various forms of carbon in aerosol particles. Thermal analysis is the most widely used method to determine the concentrations and speciation of carbon * Corresponding author phone: 852-2358-7389; fax: 852-23581594; e-mail: [email protected]
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 4, 2002
(1-3). Of the three C forms, inorganic C can be separated from the other two forms by acidification of a filter segment before carbon analysis (4-6). Determination of OC and EC by thermal methods involves programmed thermal vaporization and oxidation steps. Typically, the filter is first heated to a temperature ranging from 500 to 950 °C in an helium atmosphere or to about 350-400 °C in air (or oxygen) to remove the OC from the filter. Oxygen is then introduced to the helium carrier gas at the elevated temperature of 500 to 950 °C. When O2 is used as the primary carrier gas, the temperature is raised to 700-900 °C to oxidize the residual carbon. Differentiation between OC and EC relies on the assumption that these components can be distinguished by their volatilization and combustion properties. An interlaboratory comparison study shows that different thermal methods agree within 5-15% on the total C (TC) concentrations; however, large disagreement is seen in the breakdown between EC and OC (7). One of the major contributing factors to the disagreement has been recognized as charring formation from the organic constituents (2, 4). The OC fraction consists of a wide variety of molecular forms, ranging from simple alkanes to insoluble polymers such as cellulose (8-10). Some organic species in ambient aerosols produce pyrolytically generated EC (PEC) (i.e., char) during thermal analysis (2, 11, 12). The fraction of OC that undergoes charring can be significant. Huntzicker et al. (2) report that the average ratios of PEC to OC range from 0.19 to 0.42 for samples collected from nine cities in the United States. Efforts to differentiate PEC from original EC are hampered by a lack of a clear definition of atmospheric EC (2, 13-15). In addition, a standard for EC, as it appears in the atmosphere, does not exist, which makes it impossible to characterize the thermal behaviors of EC under different thermal analysis conditions. It is clear that without correction for charring, the EC concentrations would be overestimated while the OC values would be underestimated. Correction for charring is often carried out by measuring the amount of EC oxidation that is necessary to return the filter reflectance or transmittance to the initial value before charring occurs (2, 3, 5, 16). Uncertainty in transmittance or reflectance measurement leads to uncertainty in the OC and EC concentrations. This source of uncertainty could be large for EC determination in aerosols containing little EC and a large amount of pyrolyzable OC or for OC determination in aerosols containing a lot of EC and little OC (17). Consequently, minimizing charring has the benefit of reducing uncertainties in EC and OC determination. Charring of organic aerosol materials has long been recognized, but little has been done to understand what molecular species are easy to char and what analysis conditions minimize charring (2, 4). Charring depends on many factors, including the amount of OC, the temperatures used in analysis, the residence time at each temperature step, the presence of certain inorganic constituents, and the carrier gas used (helium versus air, for example). Our study attempts to determine what fraction of organic constituents that tends to char, the temperature at which the charred OC evolves, and the factors that influence the extent of the charring. For the study of charring characteristics of aerosol OC, it is important to have the test OC materials free of EC so that PEC can be unambiguously determined. There are no reference standards for aerosol OC. Individual organic compounds that make up the aerosol OC number in the hundreds, if not the thousands. They are rich in molecular variety and are highly variable in their concentrations. Detailed chemical analysis so far has identified less 10.1021/es015540q CCC: $22.00
2002 American Chemical Society Published on Web 01/08/2002
TABLE 1. Experimental Parameters of the Thermal Methods Used in This Study analysis method
He-1 He-2 He-3 He-4 He/O2-1b He/O2-2 He/O2-3 He/O2-4 He/O2-5 He/O2-6 He/O2-7
250 °C, 60 s 500 °C, 60 s 650 °C, 60 s 850 °C, 90 s 650 °C, 30 s 750 °C, 30 s 850 °C, 60 s 940 °C, 120 s
310 °C, 60 s 450 °C, 60 s 615 °C, 60 s 870 °C, 110 s 540 °C, 30 s 600 °C, 30 s 660 °C, 30 s 720 °C, 30 s 780 °C, 30 s 840 °C, 30 s 890 °C, 150 s
250 °C, 150 s 500 °C, 150 s 650 °C, 150 s 850 °C, 160 s 650 °C, 150 s 750 °C, 150 s 850 °C, 150 s 890 °C, 150 s
250 °C, 150 s 500 °C, 150 s 650 °C, 150 s 850 °C, 160 s
250 °C, 150 s 430 °C, 3000 s 485 °C, 150 s 650 °C, 150 s 850 °C, 160 s
The temperature program for the NIOSH method is reported in Birch (1998). b A mix of 1% oxygen in UHP helium.
FIGURE 1. Typical thermogram of an aerosol water extract when using the UST-2 thermal method. than 55% of the OC mass (10, 18). The lack of knowledge about OC composition prevents reconstruction of aerosol OC reference standards from known individual compounds. However, subfractions of OC aerosol material can be obtained by extracting particulate matter collected on filters with water or organic solvents. For the examination of aerosol charring characteristics, the solvent extracts provide realistic mixtures of a wide range of organic components and yet are free of EC.
Experimental Section Aerosol samples were collected over 12-h periods at a flow rate of 1.13 m3/min using a high-volume sampler with an impactor inlet of a 2.5 µm cut size (GT22001; Andersen Instruments, Smyrna, GA). The collection substrates were 8 × 10 in. quartz fiber filters (Pall Gelman, Ann Arbor, MI). Prior to sampling, they were baked at 550 °C overnight to remove any absorbed organic materials. The filters were kept in prebaked aluminum foil pouches during transportation and storage at -4 °C. Aerosol samples from two locations, one in Hong Kong and one in Nanjing, China, were used in this study. The sampling site in Hong Kong was on the rooftop of a building on the campus of the Hong Kong University of Science and Technology (HKUST), located on the eastern coast of Hong Kong and away from the industrial and commercial areas in the city. Depending on air flow patterns, both marine and continental air may influence the site. Sampling at this site was carried out from February 7 to 13, 2001. The Nanjing samples were collected on the campus of
Nanjing University, which is located in a residential and commercial neighborhood, from January 31 to February 5, 2001. Water extracts of the aerosol samples were obtained by removing 10-20 1 × 1.45 cm sized filter strips with a punch from the 8 × 10 in. filters and mixing them with two portions of 10 mL water in a sonication bath for 15 min. Filter debris and suspending insoluble particles were removed from the water extracts using a syringe filter (MFS-25, PTFE membrane, 0.2 µm pore size; Advantec MFS Inc., Pleasanton, CA). The water extracts were evaporated to dryness using a rotary evaporator and then reconstituted with 500 µL of water. Hexane extracts were obtained in the same manner. Various amounts of the water or hexane extracts were then spiked to prebaked filter strips. The filter strips were air-dried and submitted for carbon analysis. The water extracts contained water-soluble organic carbon (WSOC) materials along with water-soluble inorganic species such as NH4+, SO42-, and NO3-. The hexane extracts contained nonpolar organic compounds. The carbon in all of the samples was analyzed using a thermal/optical carbon analyzer (Sunset Laboratory Inc., Forest Grove, OR). The carbon analyzer uses a temperature and atmosphere-controlled oven and a transmittancemonitoring laser at a wavelength of 680 nm to produce an operational EC/OC speciation (16). The carbonaceous materials on the filter evolve into the carrier gas stream as a result of volatilization, decomposition, and combustion. The evolved carbon-containing gases are oxidized to CO2 over an manganese dioxide catalyst maintained at 870 °C, VOL. 36, NO. 4, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
FIGURE 2. Percentage of charring versus the WSOC loading for (a) aerosol samples collected from the HKUST campus and (b) one aerosol sample collected from Nanjing, China. The extent of charring, calculated as the ratio of PEC to WSOC, depends on the WSOC loading on each filter strip for analysis. catalytically reduced to methane, and then detected by a flame ionization detector (FID). Various temperature program and carrier gas combinations were used to examine the factors that control the extent of charring. Table 1 lists the analysis conditions for the five analysis methods used in this study.
Results and Discussion 1. Charring of Water-Soluble Organic Compounds and Hexane-Soluble Organic Compounds. The formation of charring from water-soluble organic compounds is evident from the thermograms of aerosol water extracts. Figure 1 shows a typical thermogram of an aerosol water extract using the UST-2 thermal method. In the UST-2 method, the first stage of analysis uses a helium atmosphere and four temperature steps (310, 450, 615, and 870 °C). The second stage is completed in a 1% O2/99% He atmosphere with seven temperature steps ranging from 540 to 890 °C. Four C peaks evolve in the first stage of analysis in He, and one broad C 756
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 4, 2002
peak evolves after the introduction of O2. The transmittance starts to drop as soon as the first C peak evolves, indicating the pyrolytic conversion of WSOC to EC or to light-absorbing intermediate OC products. Aerosol EC would evolve in the second stage of analysis, as observed in the thermograms of filters without any pretreatment. The formation of charring is indicated by the transmittance decrease and the fact that some OC evolves under the thermal analysis conditions during which aerosol EC would evolve. The amount of charring (i.e., PEC) is defined in this paper as the amount of OC that evolves in the second stage when O2 is present in the carrier gas. The percentage of charring, calculated as the ratio of PEC to the total OC, represents the extent of charring. The analysis conditions of the first stage in the UST-2 method are similar to those in the NIOSH standard method to measure EC in diesel particulates (19). The second stage of the analysis in the UST-2 method uses more temperature steps and slightly lower starting and ending temperatures than the NIOSH method. The combustion of EC at more
FIGURE 3. Thermograms of an aerosol water extract in oxidative atmospheres using (a) the UST-4 method and (b) the UST-5 method. The complete evolution of PEC generated from WSOC in aerosols may require the same high temperature and oxidative atmosphere as original EC. temperature steps results in a slower increase in the transmittance corresponding to the oxidation of EC, thereby reducing uncertainty in the EC and OC concentrations when using transmittance for the OC/EC breakdown. There has been disagreement on the chemical nature of the fraction of carbon that evolves at 850 °C in the He stage in the NIOSH method. The NIOSH method defines this fraction of carbon as part of OC (19). However, Chow et al. (15) argue that this fraction of carbon is EC. Their supporting evidence is the increasing light transmission and reflectance during this temperature step. Chow et al. suggest that the increasing light transmission and reflectance is due to the oxidation of EC by oxygen supplied by mineral oxides in the particle mixture on the filter. Our observation of the thermogram features of aerosol water extracts is consistent with theirs. As shown in Figure 1, part of the water-soluble OC evolves at 850 °C in He and in the subsequent oxidative stage, and the transmittance increases during these steps. However, the use of aerosol water extracts allows us to offer an alternative explanation. We suggest that the increasing transmittance is a result of the evolution of the light-absorbing
intermediate OC products instead of the oxidation of EC. We have two lines of evidence to support our hypothesis. First, mineral oxides are excluded from the aerosol water extracts, because they are insoluble in water. As a result, the oxidation of EC by mineral oxides is not operative in our system. Second, our experimental results do not support the hypothesis of the oxidation of PEC by trace amounts of oxygen impurities in the helium carrier gas. A same set of four aerosol water extracts were analyzed with and without an oxygen trap placed between the UHP helium cylinder and the instrument’s helium inlet. An increase in transmittance was observed at 850 °C in the helium stage, regardless of the presence or absence of the oxygen trap. In addition, the percentage of charring was similar either with or without the oxygen trap. If the O2 impurities were responsible for the transmittance increase, in the presence of the oxygen trap, we would not have observed the transmittance increase and we would have seen a lower percentage of charring. The WSOC loading (µgC/cm2) on each filter strip is observed to influence the percentage of charring of the aerosol VOL. 36, NO. 4, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
FIGURE 4. Thermograms of a same aerosol water extract using (a) the NIOSH and (b) the UST-3 thermal methods. Prolonging the duration at each temperature step reduces charring.
TABLE 2. Comparison of Charring Formation with the NIOSH and UST-3 Method sample sample 1 sample 2 sample 3 average
UST-3 NIOSH UST-3 NIOSH UST-3 NIOSH
9.32 9.62 14.19 12.93 10.59 11.89
1.36 2.49 2.44 3.79 1.32 2.66
15% 26% 17% 29% 12% 22%
0.56 0.59 0.56 0.57
water extracts. Figure 2 plots the percentage of charring versus the WSOC loading in µgC/cm2. Figure 2a is generated from 11 PM2.5 samples collected from the HKUST campus, with the low WSOC loading (