Secondary Organic Aerosol Production from Terpene Ozonolysis. 2

Effect of Pellet Boiler Exhaust on Secondary Organic Aerosol Formation from α- .... Comparison of FTIR and Particle Mass Spectrometry for the Measure...
2 downloads 0 Views 233KB Size
Environ. Sci. Technol. 2005, 39, 7046-7054

Secondary Organic Aerosol Production from Terpene Ozonolysis. 2. Effect of NOx Concentration ALBERT A. PRESTO,‡ KARA E. HUFF HARTZ,‡ AND N E I L M . D O N A H U E * ,†,‡ Department of Chemistry and Department of Chemical Engineering, Doherty Hall, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213

We report secondary organic aerosol (SOA) yields from the ozonolysis of R-pinene in the presence of NO and NO2. Experimental conditions are characterized by the [VOC]0/ [NOx]0 ratio (ppbC/ppb), which varies from ∼1 to ∼300. SOA yield is constant for [VOC]0/[NOx]0 > ∼15 and decreases dramatically (by more than a factor of 4) as [VOC]0/[NOx]0 decreases. Aerosol production is completely suppressed in the presence of NO for [VOC]0/[NOx]0 e 4.5. Fourier transform IR analysis of filter samples reveals that nitratecontaining species contribute significantly to the total aerosol mass at low [VOC]0/[NOx]0. Yield reduction is a result of the formation of a more volatile product distribution as [VOC]0/[NOx]0 decreases; we propose that the change in the product distribution is driven by changes in the gasphase chemistry as NOx concentration increases. We also present two-product model parameters to describe aerosol production from the R-pinene/O3/NOx system under both high- and low-NOx conditions.

1. Introduction The production of secondary organic aerosol (SOA) from the oxidation of monoterpene (C10H16) species is well documented (1-10). Previous studies of aerosol production from monoterpenes have focused on photooxidation (1-4), nitrate reaction (4, 6), and ozonolysis (3, 4, 7, 8, 10). Experiments are typically conducted in environmental (or “smog”) chambers (1-15), and aerosol yields up to ∼25% have been measured for a number of different monoterpenes. Organic oxidation mechanisms are sensitive to nitrogen oxides (NOx), primarily because under high-NOx conditions NO and NO2 react with organo-peroxy radicals (RO2) that would otherwise react with other peroxy radicals (RO2 and HO2). This introduces a critical branch point in the oxidation mechanism (Figure 1), which we expect to influence SOA yields. Changes in branching will alter the reaction products, for instance, introducing organic nitrates (RONO2) at high NOx, but they will also alter the distribution of common products, for instance, changing ratios of various aldehydes, ketones, and organic acids. These products will have different vapor pressures (for example, acids are less volatile than aldehydes) (16), and thus the branching will influence SOA yields. Branching, in turn, is known to vary significantly with VOC/NOx in the urban atmosphere, ranging along a chemical * Corresponding author phone: (412)268-4415; e-mail: nmd@ andrew.cmu.edu. † Department of Chemistry. ‡ Department of Chemical Engineering. 7046

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 18, 2005

FIGURE 1. Organic mechanism branching and the VOC/NOx ratio. Organo-peroxy radicals (RO2) can react with NO2, NO, RO2, and hydroperoxy radicals (HO2). This branching strongly depends on VOC/NOx, and typical ambient conditions explore its full range. This influences the composition of product compounds, for instance, introducing organonitrates (RONO2) at high NOx, but it also influences the distribution of common products, such as carbonyls (RdO). Because secondary organic aerosol yields depend on the distribution of product vapor pressures, we expect SOA yields to be sensitive to [VOC]0/[NOx]0. coordinate from NOx-dominated to HO2-dominated conditions (17). Indeed, this branching is largely responsible for the transition from VOC-limited to NOx-limited conditions in ozone production; the “spine” on the classic empirical kinetic modeling approach (EKMA) ozone isopleth diagram (18) at 8:1 VOC/NOx corresponds to roughly 50:50 branching of RO2 between NO and HO2. Photooxidation experiments have revealed a dependence of SOA yields on [VOC]0/[NOx]0 (1), but the source of this dependence is difficult to ascertain. This is because the ratio of OH to ozone varies with [VOC]0/[NOx]0 (both in chambers and in the atmosphere). Therefore, it is difficult to separate yield changes due to variations in the rate of OH-initiated to ozone-initiated terpene oxidation from yield changes due to altered product distributions within the OH and ozonolysis mechanisms. The solution is to separate the reaction mechanisms as much as possible by conducting experiments under “pure OH” or “pure ozone” conditions. Our focus here is on pure ozone conditions, for several reasons: First, ozonolysis is an important atmospheric sink for monoterpenes (19) and is a potent source of SOA (4). Second, ozonolysis experiments are typically conducted under darkened, NOx-free conditions, largely because ozone and NOx react rapidly, making ozonolysis experiments in the presence of NOx difficult. Third, ozone concentrations are typically highest in areas that have correspondingly high concentrations of NOx (18). Fourth, ozone-alkene reactions will generate organic radicals, especially after short-lived products such as the Criegee intermediate decompose to produce OH radicals. There is thus every reason to expect these organic radicals to behave comparably to other organic radicals, with branching behavior similar to that depicted in Figure 1. Consequently, we expect SOA yields under pure 10.1021/es050400s CCC: $30.25

 2005 American Chemical Society Published on Web 08/16/2005

TABLE 1. Experiments Conducted for This Studya date 3/30/04 4/22/04 4/27/04 5/20/04 5/26/04c 6/2/04 6/4/04 6/11/04 8/3/04 8/4/04 8/10/04 8/20/04 8/25/04 8/26/04 8/31/04 9/15/04 9/21/04ad 9/21/04b 9/27/04 10/4/04a 10/4/04b 10/6/04 10/14/04a 10/14/04b 10/20/04 10/27/04 11/3/04 12/16/04 1/7/05 2/8/05

VOC0 (ppb) 210 130 235 218 205 91.2 50.4 54.4 23.8 17.9 180 20.6 16.1 20.4 205 15 40 150 156 50 200 150 50 167 10.8 152 15.0 166 11.8 12.3

O3 (ppb) 570 500 510 510 450 500 550 490 190 180 470 280 160 230 520 605 650 560 530 520 450 420 490 420 185 445 180 460 200 150

NOx (ppb) 30 2000 525 614 320 25 15 396 188 22 4.6 11 11 45 6.5 9.0 30 15 6.0 10 373 145 16 13 20 9.0 14 7.0 42 45

COC0 (µg/m-3) 346 2.07 212 271 N/A 117 39.3 3.06 0.96 7.9 286 5.63 7.30 0.00 260 7.08 32.3 218 164 46.3 167 205 45.8 226 0.96 143 3.64 300 0.00 0.00

[VOC]0/[NOx]0 (ppbC/ppb) 70 0.65 4.5 3.6 6.3 36 34 1.4 1.3 8.1 391 19 16 4.5 315 17 13 100 260 50 5.4 10 31 130 5.5 170 11 240 2.8 2.7

yield 0.29 0.0028 0.16 0.22 N/A 0.23 0.14 0.01 0.01 0.08 0.28 0.05 0.08 0.00 0.23 0.08 0.14 0.26 0.19 0.16 0.15 0.24 0.16 0.24 0.02 0.17 0.043 0.32 0.00 0.00

notes NO2 NO2 NO2 NO2 NO2 NO2 NO2 NO2 NO2 NO2

b

UV NOx,UV UV NO2 SOA seed UV, SOA seed NO2, SOA seed NO2 SOA seed NOx, UV UV, SOA seed NOx, UV NOx, UV NOx, UV, (NH4)2SO4 seed

a COC and yield values are calculated directly from the raw data; yields presented for UV-illuminated experiments are thus subject to reduction 0 via UV exposure. b Experiments are classified as either NOx or NO2. Experiments with neither label utilized ambient NOx and are applicable to both data sets. c Experimental temperature was ramped from 22 to 40 °C. SOA yield was not measured. dDates with experiments labeled as a and b are SOA seed experiments. Experiment a is the initial R-pinene injection that is used to seed the chamber. Experiment b is the second R-pinene injection, which is performed in the presence of the SOA seed particles generated by experiment a.

ozone conditions to depend on the [VOC]0/[NOx]0 ratio. It is vital that we understand this dependence, because it could strongly influence the response of SOA production to pollution control strategies, potentially changing the efficacy of VOC or NOx controls when considering multiple objectives such as ozone and SOA reductions. We have shown in a companion study (20) that SOA yields from the R-pinene + ozone reaction are reduced in the presence of UV light. Here, we present an investigation of SOA production from the ozonolysis of R-pinene in both high- and low-NOx conditions. We have conducted experiments under varying conditions of initial terpene concentration and initial [VOC]0/[NOx]0 ratio. Our results show that SOA formation is indeed dependent upon NOx concentration and that the changes in SOA yield are a result of changes in the product distribution. We also present two-product model parameters to characterize aerosol yield from the ozonolysis reaction.

2. Experimental Section 2.1. Chamber Experiments. Many of the experimental details are described elsewhere (20); here we will present details specific to this study. SOA yield experiments are conducted in a 10-m3 Teflon chamber (Welch Fluorocarbon) (21), suspended inside a temperature-controlled room (15-40 °C). Ozone and NOx concentrations are measured by gas-phase analyzers for each species (Dasibi 1008-PC and API 200A). Particle concentrations (for ∼10-800-nm particles) are monitored by a scanning mobility particle sizer (SMPS, TSI 3936). For most experiments, a gas chomatograph with a flame ionization detector (GC-FID, Perkin-Elmer AutoSystem XL; J&W Scientific DB-624 capillary column, 30 m × 0.530 mm) coupled to a preconcentrator (Entech 7100A) is available for measuring gas-phase concentrations of organic species.

The chamber is also equipped with three banks of UV lights (General Electric model 10526 black lights). Experiments involve either R-pinene + O3 + NO2 or R-pinene + O3 + NOx, where NOx is a mixture of NO and NO2. We classify experiments by the initial [VOC]0/[NO2]0 or [VOC]0/[NOx]0 ratio (ppbC/ppb), following Pandis et al. (1). Ozone is generated using a corona-discharge ozone generator (Azco HTU500AC); ozone concentrations are typically in the range of 200-500 ppb. NOx is introduced to the chamber directly from standard mixed cylinders of NO (1 ppm in N2) or NO2 (1% in N2). NOx concentrations range from ∼10 to >500 ppb, and the [VOC]0/[NOx]0 ratio (ppbC/ppb) varies from 300. The background NOx concentration in the smog chamber is ∼10 ppb; therefore no additional NOx is added to the chamber for high [VOC]0/[NOx]0 (low NOx) experiments. Such experiments are labeled neither as “NOx” or “NO2” in Table 1; results from these experiments are applicable for both the NOx and the NO2 portions of this study. Experiments using NO2 are conducted in a darkened chamber. In these experiments, both O3 and NO2 are introduced to the chamber prior to R-pinene injection, which is used to initiate the reaction. A substantial complication is the generation of NO3, which can also react rapidly with R-pinene (19).

O3 + NO2 f NO3 + O2

(R1)

NO3 + NO2 f N2O5

(R2)

NO3 + R-pinene f products

(R3)

O3 + R-pinene f products

(R4)

Because our objective is to isolate the ozonolysis reaction, VOL. 39, NO. 18, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

7047

this is a concern. The R-pinene-NO3 reaction is a poor source of SOA (4, 6), so high fractional removal of R-pinene by NO3 would suppress the measured SOA yield. This is a particular concern for experiments with [VOC]0/[NO2]0 ratios between 3 and 10. When the [VOC]0/[NO2]0 ratio is below 3, NO2 efficiently scavenges NO3 (22); at [VOC]0/[NO2]0 greater than 10, NO3 concentrations are too low to remove an appreciable amount of R-pinene. Because of the importance of this middle region with [VOC]0/[NO2]0 ratios between 3 and 10, we used GC-FID data to obtain a rough measure of the R-pinene loss rate under varying conditions of [VOC]0/[NO2]0. Experiments including NO are conducted in the presence of UV light because NO rapidly reacts with O3 to form NO2. The purpose of the UV light is to regenerate NO via NO2 photolysis.

O3 + NO f NO2 + O2 hν

(R5)

NO2 98 NO + O

(R6)

O 2 + O f O3

(R7)

Our initial experiments under UV illumination, discussed in a companion paper, revealed that the UV light systematically reduces SOA yields from R-pinene ozonolysis by an approximately constant value of 0.03 (20). However, the use of UV light is inevitable if we wish to study SOA production from R-pinene ozonolysis in the presence of NO. The UV lights allow us to maintain a NO/NO2 ratio above 0.5, while in a darkened chamber this ratio would be nearly zero after only a few seconds. The presence of UV light is arguably also more relevant to the atmosphere. Because the UV light will also slowly dissociate O3, we inject R-pinene into the chamber prior to illumination and use O3 as the initiating reagent. As we reported previously (20), the SOA yield is independent of the initiating reactant under otherwise identical experimental conditions. 2.2. Filter Samples. During most experiments, SOA is collected onto quartz fiber filters (Pall Life Sciences) for analysis by gas chromatography mass spectrometry (GCMS). Details of the filter sampling, extraction, and analysis are given elsewhere (20). During some experiments, SOA is collected onto filters for Fourier transform IR (FTIR) analysis rather than GC-MS analysis. FTIR analysis has been used to analyze filter samples of both ambient (23-25) and laboratory-generated (26) aerosol. The FTIR analysis complements GC-MS or other techniques because it is nondestructive; filters can be analyzed directly and do not require complicated extraction procedures. Extraction and derivatization processes, such as those used for the GC-MS filters collected in this study, have the potential of altering the composition of the species collected on the filter, and derivatization of filter extracts requires the identification of derivatized species rather than the actual aerosol components. Additionally, FTIR allows for the determination of the total concentration of particular functional groups in a sample. The samples are collected by drawing chamber air at 13 L min-1 across a 37-mm Teflon filter (Pall Life Sciences). Filters are scanned in a single-pass cell coupled to a 0.5cm-1 FTIR (Oriel MIR 8000) with a mercury-cadmium telluride (MCT) detector. The filters are scanned both before and after sampling, and analysis is conducted in a manner similar to the method developed by Donahue et al. (27). In this method, the prescan is used to establish a baseline absorption for the filter. Dividing the postscan by the baseline absorption allows for the removal of constant features; this method is effective at removing prominent features due to water vapor and CO2, but as noted by McClenny et al. (23), it is sometimes difficult to completely background-correct 7048

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 18, 2005

FIGURE 2. Aerosol yield for the r-pinene/O3/NO2 system for initial terpene mixing ratios of 15-30 ppb and 150-200 ppb. All experiments were conducted in a darkened chamber with the exception of the highest [VOC]0/[NO2]0 ratio for each series; in these experiments, the chamber was irradiated with UV light. Yields for the UV-irradiated experiments have been adjusted by +0.03. The highest [VOC]0/ [NO2]0 ratios tested in the 150-200 ppb experiments (∼300) are not shown. Error bars show the precision (1σ) for the calculated yields. for the large Teflon absorption features located near 1200 cm-1. We therefore focus on spectral features located far (100 wavenumbers or more) from this portion of the spectrum. The filters are also marked prior to sampling and are placed in the FTIR cell in the same orientation for both pre- and postscans. FTIR filter samples require relatively high mass loading (∼100 µg). Thus, FTIR filter samples are only collected for experiments that use high initial concentrations of R-pinene and consequently generate large quantities of SOA. The products of R-pinene ozonolysis consist primarily of oxygenated organic species, particularly aldehydes, acids, keto acids, and hydroxy acids (9, 20, 28). All of these species contain a carbonyl group, and the most prominent feature in their FTIR spectra is the CdO stretch at approximately 1725 cm-1. The large number of SOA species (approximately 10) displaying this feature makes identification of each species difficult. Thus, we will use the carbonyl peak as a proxy for the total amount of oxygenated organic species. We also expect to identify organic nitrates in samples collected at low [VOC]0/[NOx]0 ratios. Organic nitrates have been identified previously in both ambient (24) and laboratory (26) data and have features at approximately 1630, 1280, and 860 cm-1.

3. Results and Discussion Table 1 lists the aerosol yields as calculated from the raw data. Yields are calculated using the method described in a companion study (20) and assume an aerosol density of 1.0 g cm-3. All subsequent figures, however, will show adjusted yields from UV-illuminated experiments to account for UVinduced yield reduction. We adjust the yields for UVilluminated experiments (with nonzero yields) by adding 0.03, consistent with our earlier finding (20). Our experimental results and the accompanying paramaterizations may therefore overestimate the real (atmospheric) SOA yield from R-pinene ozonolysis during daylight; conversely, a crude model for the effect of sunlight would be a uniform reduction in yields of 0.03 (potentially to zero yield). 3.1. NO2 Experiments. SOA yields for the R-pinene/O3/ NO2 system were measured for initial R-pinene mixing ratios of 15-200 ppb. Results are shown in Figure 2. SOA yield changes quite dramatically as the system moves from highNOx to low-NOx conditions. The yield increases as NO2 concentration decreases and reaches an asymptote near

[VOC]0/[NO2]0 ) 15; a maximum yield of ∼0.30 is observed in the 150-200 ppb R-pinene experiments for [VOC]0/[NO2]0 ratios ranging from 62 to nearly 400. (Yields for the highest [VOC]0/[NO2]0 ratio experiments are not shown in Figure 2.) A maximum yield of 0.08 is observed for the 15-30 ppb R-pinene experiments at [VOC]0/[NO2]0 > 15. SOA yield decreases rapidly below [VOC]0/[NO2]0 of 15, reaching a minimum of 0.01 for the low-VOC experiments at [VOC]0/ [NO2]0 of 1.3 and 10, the km/kc ratio is 1.0 ( 0.30. This suggests that for low concentrations of NO2 R-pinene loss is (nearly) completely a result of ozonolysis. For [VOC]0/[NO2]0 between 3 and 10, the km/kc ratio is 1.25 ( 0.25. The increase in the km/kc ratio is consistent with a small R-pinene loss (