Secondary organic aerosol formation in the atmosphere via heterogeneous reaction of gaseous isoprene on acidic particles



[1] Water-soluble macromolecular substances with spectral properties of “humic-like substances” (HULIS) were recently found to form the major identified fraction of the organic aerosol at urban and rural sites in Europe. With primary sources identified so far (e.g., biomass combustion) it is not possible to explain the observed HULIS levels in Europe, therefore there is an ongoing search for other sources - which form HULIS in situ in the atmosphere. Here we show that secondary aerosol formation of atmospheric polymers occurs by heterogeneous reaction of isoprenoid or terpenoid emissions in the presence of a sulfuric acid aerosol catalyst. Competing oxidants such as ozone or the presence of humidity decreased the reaction yield, but the formation of humic–like substances was not disabled. Calculations indicate that the presented reaction pathway could be an additional source for HULIS in the continental aerosol.

1. Introduction

[2] Although hundreds of individual organic compounds have been identified in the organic atmospheric aerosol so far [Saxena and Hildemann, 1996], together they constitute less than 10% of the organic carbon of urban and rural aerosol [Rogge et al., 1993; Puxbaum et al., 2000]. However, the major contributors to the continental organic aerosol are recently determined “humic-like substances” (HULIS) contributing 20–50% to the water soluble organic aerosol at urban and rural sites in Europe [Havers et al., 1998; Zappoli et al., 1999; Facchini et al., 1999a; Krivacsy et al., 2001]. This means that HULIS have an impact on the hygroscopicity and the cloud-condensation nuclei formation potential of the atmospheric aerosol and are, therefore, of climatic relevance [Facchini et al., 1999b; Charlson et al., 2001]. According to the IR-spectroscopic fingerprints the origin of these atmospheric macromolecular substances has been tentatively assigned to “agricultural burning” [Mukai and Ambe, 1986; Facchini et al., 1999a]. High ambient concentrations of HULIS observed at a rural site in Hungary in samples with low “black carbon” [Zappoli et al., 1999], however, cannot be explained by biomass fires, therefore there is an ongoing search for other sources.

[3] The largest source of reactive organic species in the atmosphere, that are of terrestrial origin, are isoprene and terpene emissions [Andreae and Crutzen, 1997; Griffin et al., 1999; Simpson et al., 1999]. Whereas for the photo-oxidation of isoprene under atmospheric conditions only negligible aerosol formation was observed [Pandis et al., 1991], the atmospheric oxidation of terpenes can lead to products that have sufficiently low vapor pressure to partition between the gas and the aerosol phase [Kavouras et al., 1998]. Until now, smog chamber experiments showed that the oxidation of these volatile compounds only led to the formation of multiple oxygen-containing monomeric compounds [Hallquist et al., 1999; Seinfeld et al., 2001], and in some cases dimers [Hoffmann et al., 1998], but the formation of macromolecular compounds was not observed. However, there are other possibilities that lead to secondary aerosol formation. For example gaseous compounds can be further transformed via heterogeneous reactions between the gas phase and atmospheric particulate matter. Recently it has been shown that volatile aldehydes, which can be produced by atmospheric photochemical reactions, can contribute to secondary aerosol formation via hydration, polymerization and hemiacetal/acetal formation in the presence of an acid catalyst [Jang and Kamens, 2001]. But all of the studies dealing with the heterogeneous reaction of organic compounds with sulfuric acid were limited to carbonyls [Duncan et al., 1998; Duncan et al., 1999; Jang et al., 2002; Noziere and Riemer, 2003] - the acid catalyzed polymerization of unsaturated compounds was not considered until now. Here we show that not only carbonyls but also dienes like isoprene can contribute on secondary aerosol formation through heterogeneous reactions in the presence of an acid catalyst.

2. Methods

2.1. Polymerization Experiment

[4] Isoprene derived from a dynamic test gas generator, which was based on an open tube diffusion technique was diluted with synthetic air and lead through a quartz fiber filter impregnated with sulfuric acid. The isoprene concentration was varied between 200 and 2000 ppbv, the absolute amount of sulfuric acid ranged from 50 to 3600 ng per filter and was determined for each experiment via ion chromatography. For each experiment a sample volume of approximately 2 m3 was used, which was purged through the quartz fiber filter within 300 minutes, leading to an air flow of about 400 l h−1. Using our experimental set-up a mean reaction time for an individual isoprene molecule with sulfuric acid of ∼10 ms was achieved. To simulate atmospheric conditions some of the experiments were performed in the presence of humidity respectively oxidants such as ozone. The ozone was produced by UV irradiation of an additional synthetic air supply and introduced into the carrier gas. The reaction yield for each experiment was calculated by dividing the determined carbon mass of the formed polymer product by the carbon content of the applied isoprene mass. The organic fraction which is formed through other reaction pathways than the reaction with sulfuric acid was determined from experiments performed in the absence of the catalyst.

2.2. Characterization of the Polymerization Products

2.2.1. Total Carbon and Thermo-EGA

[5] A portion of the filter was heated with a constant rate of 20°C min−1 from room temperature to 750°C in an oxygen atmosphere. The carbon containing gases evolved from the sample were converted into CO2 over a MnO2 catalyst, and subsequently measured with a non-dispersive infrared analyzer (Maihak UNOR 6N).

2.2.2. UV/VIS-Spectra

[6] Water soluble reaction products formed in the polymerization experiment were extracted from an aliquot of the filter using ultrasonic agitation. The UV/VIS spectra of the derived sample solution was recorded with a Unicam 5625 UV/VIS spectrometer.

2.2.3. Sample Pre-Treatment for Drift Investigations

[7] After extraction of the water soluble compounds formed in the polymerization experiment from an aliquot of the filter the derived sample solution was diluted with organic free water to a final volume of 20 ml. Afterwards the pH of the sample solution was adjusted to approximately 1 by addition of 100 μl concentrated sulfuric acid. After a conditioning step the sample solution was pulled through a C-18 SPE-cartridge (IST, 221-0020-H) using a vacuum flask. The adsorbed reddish-brown colored compounds were eluted with 2 × 300 μl pure methanol from the SPE-cartridge. One drop of the derived sample solution was applied to a gold-coated object slide. After evaporation of the solvent (methanol) the residue was investigated by FTIR diffuse reflectance with a Spectratech IR-microscope attached to a Bruker VECTOR22 spectro-meter. The IR-spectra were recorded over the frequency range from 750 to 4000 cm−1 with a resolution better 1 cm−1. During the measurements the sample chamber was flushed with dried air to avoid absorption bands caused by humidity.

2.2.4. NIST-Reference Sample

[8] A 200 mg portion of reference dust sample (NIST 1648) was extracted ultrasonically with 2 ml 0.1 N NaOH for 20 minutes. The derived solution including the non soluble residue was diluted with organic free water to a total volume of 10 ml. Subsequently the sample solution was filtered through a Millipore membrane-filter to separate the non-dissolved residue. After neutralization of the filtered sample solution with a calculated amount of concentrated sulfuric acid an excess of 100 μl acid was added to achieve a similar pH compared to the synthetic samples. After filling up with ultra pure water to a final volume of 20 ml weakly polar organic sample constituents were separated from the aqueous solution using the described solid phase extraction technique. The derived reddish-brown colored solution was examined by DRIFTS.

3. Results and Discussion

[9] In a first series of experiments performed in the absence of competing reagents the development of yellowish colored products was observed, indicating the formation of larger molecules with conjugated double bonds. The formation of polymer products was evidenced by a) observation of substances of low volatility in a thermogram (Figure 1a), and b) the relatively high UV-absorbance in the visible range (Figure 1b). In the thermogram only minor peaks of monomeric compounds are seen in the temperature region from 150–220°C, while the large and extended peak in the temperature region from 250–550°C is due to substances of low volatility, such as oligomeric or polymeric reaction products from isoprene. Also the shift in the UV-spectrum to a considerable absorbance above 300 nm suggest that poly-conjugated structures are present. The most common method used for characterization of humic like substances in atmospheric aerosols is IR spectroscopy [Mukai and Ambe, 1986; Havers et al., 1998], in particular for small samples FTIR diffuse reflectance spectra (DRIFT-spectra) of the extracted reaction products were recorded [Zappoli et al., 1999; Krivacsy et al., 2001]. To this end a pre-treatment of the sample extracts by solid-phase extraction was performed to remove inorganic ions as well as highly polar organics. The DRIFT-spectra of the isolated orange colored products indicated the presence of oxygenated functional groups (Figure 2). Partially oxidized polymer formation of isoprene in the presence of oxygen has already been observed on kaolinite acid centers [Cheshchevoi et al., 1987]. The relatively well resolved absorption bands in Figure 2 could be assigned to the following structural units: carboxylic groups (COOH), hydroxyl groups (OH) and carbonyl groups (CO). Thus the obtained reaction products exhibited all the basic features of HULIS observed by different groups [Mukai and Ambe, 1986; Zappoli et al., 1999; Facchini et al., 1999a; Krivacsy et al., 2001] in extracts from atmospheric aerosols. Additionally we compare in Figure 2 the DRIFT spectra derived from extracts of a NIST standard sample (1648) of urban particulate matter and our reaction products. In this standard reference air dust 6.5% of the organic carbon could be attributed to water soluble macromolecular substances like humic acids [Havers et al., 1998]. The considerable similarity of the two IR-spectra indicates, that the macromolecular water soluble compounds formed in the laboratory experiment exhibit practically all structural properties of the HULIS isolated from the NIST standard reference material.

Figure 1.

Acid catalyzed polymerization products from Isoprene. Experiment performed with an isoprene concentration of 2 ppmv, passed through a quartz fiber filter containing 550 μg sulfuric acid. a, Thermogram of the derived products. b, UV spectrum of the aqueous extract.

Figure 2.

Comparison of FTIR spectra from isoprene polymerization products and an extract of a NIST reference dust sample.

[10] Systematic variations of the experiments indicated the following sensitivities: a decrease of the employed sulfuric acid content decreased the formation rate of polymers (Figure 3a). Likewise, a stepwise substitution of sulfuric acid by ammonium sulfate decreased the formation rate of polymeric products. Interpreting these results we concluded that free acidity is necessary for the heterogeneous polymerization of isoprene, but the “aerosol acidities” observed in background aerosols [e.g., Ferek et al., 1983; Staebler et al., 1999] are sufficient to promote the formation of polymers. In addition, it has to be regarded that atmospheric sulfate clusters as predicted from aerosol dynamic considerations [Kulmala et al., 2000] are considered also to carry free acidity. With α-pinene instead of isoprene very similar “humic-like” reaction products were obtained, that had only slight differences in the IR-spectra, in particular in the region below 1250 cm−1.

Figure 3.

Reaction yield of the acid catalyzed polymerization of isoprene for different experimental conditions. In all experiments 2 m3 synthetic air with a concentration of 750 ppbv Isoprene was passed through the quartz filter with a flow rate of 7 l min−1. a, Dependence of the reaction yield from sulfuric acid content b, Influence of ozone as competing oxidant c, Effect of the air humidity.

[11] Compounds with unsaturated functionalities (double bonds) such as isoprene react in the atmosphere with ozone, free radicals, or other oxidants and thus become converted to oxidized products. In the liquid aerosol phase the reaction with H2O2 is likely [Seinfeld and Pandis, 1998]. Thus the formation of HULIS via heterogeneous reaction with sulfuric acid could be disabled due to competing reactions of isoprene with these oxidants. To simulate an oxidative environment we performed the polymerization experiment in the presence of ozone. Compared to the results for experiments without competing oxidants we found a decreased formation yield for HULIS (Figure 3b), but still for ozone concentrations higher than the deployed isoprene a substantial formation of HULIS was found. Likewise, an increase of the humidity of the used synthetic air decreased the reaction yield for the polymeric products (Figure 3c). For conditions were the acid concentration is only about 50% (according to a relative humidity of 35%) still a significant production of HULIS was observed.

[12] Estimating that the mean concentration of humic-like compounds in the particulate phase of aerosols is around 1 μg m−3, and assuming an atmospheric life-time for aerosol particles of 5 days the production rate of humic-like material is estimated to be 10 ng m−3 h−1 or about 1 × 104 cm−3 s−1. Applying the kinetic theory of gases [e.g., Seinfeld and Pandis, 1998] and assuming an aerosol surface area of 10–100 μm2 cm−3 and a gas phase concentration of biogenic isoprenoids of 0.02–1.0 ppbv [Warneck, 1988; Kulmala and Hämeri, 2000], the molecules can be expected to collide 1 × 106 − 5 × 108 times cm−3 s−1 with aerosol particles. Thus, an overall atmospheric uptake coefficient to form humic-like compounds with the polymerization pathway from 0.01 to 2 × 10−5 would be sufficient to explain HULIS concentrations typically observed in Europe [Havers et al., 1998; Zappoli et al., 1999; Krivacsy et al., 2001]. From these considerations, it follows that the heterogeneous acid-catalyzed polymer formation from isoprene, α-pinene or other terpenoids upon reaction with acidic aerosols is likely to occur in the atmosphere. In addition to other reported mechanisms the proposed reaction pathway might be therefore an important source for HULIS in continental aerosol.

[13] It has recently been shown that the growth of nm-size atmospheric aerosols cannot be explained by the condensation of sulfuric acid [Kulmala et al., 2000]. The heterogeneous reaction mechanism proposed here could have a significant contribution to the aerosol growth, since its efficiency is improved when acidity increases as it does in atmospheric sulfate clusters [Kulmala et al., 2000]. Furthermore the uptake of biogenic gases to form polymers appears to be sufficiently effective. The presented reaction pathway will therefore help to understand the formation of natural and anthropogenic aerosols, the production of CCN, and might contribute to the “omnipresence” of mixed aerosol particles.


[14] We would like to thank T. Novakov, M. R. Hoffmann, I. Marr, and R. Hitzenberger for helpful comments, B. Mizaikoff and M. Schaufler for performing FTIR-spectra and for discussions, M. Gann for insight into polymerization reactions of dienes. The study was financed in part by the Austrian Science Foundation/FWF.