Summer/winter variability of the surfactants in aerosols from Grenoble, France

Abstract

Many atmospheric aerosols seem to contain strong organic surfactants likely to enhance their cloud-forming properties. Yet, few techniques allow for the identification and characterization of these compounds. Recently, we introduced a double extraction method to isolate the surfactant fraction of atmospheric aerosol samples, and evidenced their very low surface tension (≤30 mN m⁻¹). In this work, this analytical procedure was further optimized. In addition to an optimized extraction and a reduction of the analytical time, the improved method led to a high reproducibility in the surface tension curves obtained (shapes and minimal values), illustrated by the low uncertainties on the values, ±10% or less.

The improved method was applied to PM₁₀ aerosols from the urban area of Grenoble, France collected from June 2009 to January 2010. Significant variability was observed between the samples. The minimum surface tension obtained from the summer samples was systematically lower (30 mN m⁻¹) than that of the winter samples (35–45 mN m⁻¹). Sharp transitions in the curves together with the very low surface tensions suggested that the dominating surfactants in the summer samples were biosurfactants, which would be consistent with the high biogenic activity in that season. One group of samples from the winter also displayed sharp transitions, which, together with the slightly higher surface tension, suggested the presence of weaker, possibly man-made, surfactants. A second group of curves from the winter did not display any clear transition but were similar to those of macromolecular surfactants such as polysaccharides or humic substances from wood burning. These surfactants are thus likely to originate from wood burning, the dominating source for aerosols in Grenoble in winter. These observations thus confirm the presence of surfactants from combustion processes in urban aerosols reported by other groups and illustrates the ability of our method to distinguish between different types of surfactants in atmospheric samples.

Introduction

While the thermodynamics of cloud formation is generally well understood, the properties of the aerosol particles responsible for their activation into cloud droplets, or “Cloud Condensation Nuclei (CCN) properties”, are still not completely understood, leading to large uncertainties in the climate budget (Foster et al., 2007). Over 20 years ago, correlations between the surface tension of fog water and its Dissolved Organic Matter (DOM) content provided the first evidences that organic compounds can lower the surface tension, σ, of aerosol particles (Capel et al., 1990). Similar effects were later reported for organic acids and suggested to affect cloud formation at climate scale (Facchini et al., 1999). Establishing a direct link between the presence of surfactants and the CCN properties of aerosol particles is not easy because most techniques to study atmospheric CCN can not measure the surface tension of the particles. However, such a link is suggested by atmospheric observations such as the recent measurements of CCN numbers in the Atlantic Ocean (Good et al., 2010) and Central Germany (Irwin et al., 2010), which were systematically underpredicted by models unless a surface tension of less than 50 mN m⁻¹ was assumed for the aerosol particles. Thus, knowing the surface tension of atmospheric particles or, alternatively, identifying the surfactants and determining their concentrations in aerosol particles appears to be important to predict CCN properties accurately. So far, the only way to perform such measurements is with aerosol filter samples, which are extracted, and the surface tension of the extracts measured with a tensiometer. The first studies using these methods reported surface tensions between 52 mN m⁻¹ (Mircea et al., 2005) and 60 mN m⁻¹ (Capel et al., 1990, Facchini et al., 1999, Facchini et al., 2000, Hitzenberger et al., 2002, Decesari et al., 2005, Mircea et al., 2005). The small amounts of material on the filters made these methods challenging. An additional drawback was that the extraction, usually in water, was not specific to surfactants, leading to mixtures where the contribution of the surfactants was underestimated.

Numerous works tried to identify organic compounds able to lower the surface tension of atmospheric aerosols. In particular, since the work of Facchini et al. (1999) the CCN properties of numerous organic acids were studied in laboratory, such as 3-hydroxybutanoic acid, 3-hydroxybenzoic acid, azelaic acid, cis-pinonic acid, oxalic, malonic, maleic, glutaric, citric, malic, succinic, adipic, fulvic, and phthalic acid (Li et al., 1998, Seidl, 2000, Hitzenberger et al., 2002, Tuckermann and Cammenga, 2004, Gaman et al., 2004, Kiss et al., 2005, Hyvarinen et al., 2006, Topping et al., 2007) as well as C8–C12 fatty acids (Prisle et al., 2008, Prisle et al., 2010). Other organic compounds found in aerosols were also investigated, such as levoglucosan (Rosenørn et al., 2006) and Humic-Like Substances (HULIS) (Kiss et al., 2005, Salma et al., 2006, Taraniuk et al., 2007). However, due to their modest surfactant effects or to their limited concentrations in aerosols, these compounds were all concluded to have negligible effects on the surface tension, and thus on the CCN properties, of atmospheric particles (McFiggans et al., 2006). In recent years, we identified a category of organic compounds able to lower the surface tension of solutions to about 30 mN m⁻¹, and for concentrations 5 or 6 orders of magnitude lower than organic acids: “biosurfactants” produced by microorganisms (Ekström et al., 2010). The Köhler curves for these biosurfactants, obtained from surface tension and osmometry measurements, displayed Critical Supersaturation values lower than those for ammonium sulfate and sodium chloride, indicating their better CCN efficiencies (Ekström et al., 2010).

In addition to characterizing standard biosurfactants, Ekström et al. (2010) improved the technique for isolating the surfactant fraction of atmospheric samples by introducing a double extraction procedure. Not only this new method led to complete surface tension curves for atmospheric sample extracts, evidenced by a plateau at large concentration, but it also evidenced the very low surface tension of these extracts, around 30 mN m⁻¹. This was true even for samples from regions such as the Amazonian forest, for which a surface tension of 50 mN m⁻¹ had been previously reported (Mircea et al., 2005). This confirmed that single extraction methods provided only a partial picture of the surfactants in atmospheric aerosols. The very low surface tensions obtained with the double extraction method were attributed to biosurfactants, because they are the only surfactants able to reach such low values, even for trace concentrations (Gautam and Tyagi, 2006).

This new method needed however further characterization and optimization, and application to wider range of atmospheric samples. The objective of the present work was therefore to further optimize this method and apply it to a new type of aerosols: aerosols from an urban area. In a first part, this work presents the investigation of the optimal conditions for each step of the experimental procedure, using both standard surfactants and authentic aerosol samples. In a second part, the uncertainties on the surface tension obtained with the optimized method are estimated. In a third part, the application of this method to aerosol samples from Grenoble, France, is presented.