Article pubs.acs.org/JPCA
Surface Organic Monolayers Control the Hygroscopic Growth of Submicrometer Particles at High Relative Humidity Christopher R. Ruehl and Kevin R. Wilson* Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States S Supporting Information *
ABSTRACT: Although many organic molecules commonly found in the atmosphere are known to be surface-active in macroscopic aqueous solutions, the impact of surface partitioning of organic molecules to a microscopic aqueous droplet interface remains unclear. Here we measure the droplet size formed, at a relative humidity (∼99.9%) just below saturation, on submicrometer particles containing an ammonium sulfate core and an organic layer of a model compound of varying thickness. The 12 model organic compounds are a series of dicarboxylic acids (C3 to C10), cis-pinonic, oleic, lauric, and myristic acids, which represent a broad range in solubility from miscible (malonic acid) to insoluble. The variation in droplet size with increasing organic aerosol fraction cannot be explained by assuming the organic material is dissolved in the bulk droplet. Instead, the wet droplet diameters exhibit a complex and nonlinear dependence on organic aerosol volume fraction, leading to hygroscopic growth that is in some cases smaller and in others larger than that predicted by bulk solubility alone. For palmitic and stearic acid, small droplets at or below the detection limit of the instrument are observed, indicating significant kinetic limitations for water uptake, which are consistent with mass accommodation coefficients on the order of 10−4. A model based on the two-dimensional van der Waals equation of state is used to explain the complex droplet growth with organic aerosol fraction and dry diameter. The model suggests that mono- and dicarboxylic acids with limited water solubility partition to the droplet surface and reduce surface tension only after a two-dimensional condensed monolayer is formed. Two relatively soluble compounds, malonic and glutaric acid, also appear to form surface phases, which increase hygroscopicity. There is a clear alternation in the threshold for droplet growth observed for odd and even carbon number diacids, which is explained in the model by differences in the excluded molecular areas of even (∼40 Å2/molecule) and odd (∼20 Å2/molecule) diacids. These differences are consistent with the odd diacids arranged at the droplet interface in “end-to-end” configurations with only one acid group in contact with the aqueous phase, which is in contrast to even carbon numbered diacids forming “folded” films with both acid groups in contact with the bulk phase. Organic matter produced by the ozonolysis of α-pinene forms surface films that exhibit similar behavior and become thinner with oxidation, allowing for greater water uptake. These results reveal a new and complex relationship between the composition of an organic aerosol and its hygroscopicity, suggesting that organic surface films might strongly influence cloud droplet formation as well as the multiphase chemistry of organic aerosols.
■
INTRODUCTION The aqueous phase chemistry of atmospheric particulate organic matter (organic aerosol, or OA) has received a great deal of recent attention. This has been driven by the desire to understand, among other things, the formation of OA from gasphase precursors,1 the chemical transformations that OA undergoes after formation,2 and the cloud condensation nuclei (CCN) activity of ambient particles.3 Liquid water typically remains in atmospheric particles even at low relative humidity (RH);4 therefore, the chemistry of microscopic aqueous droplets not only is relevant in cloudy, foggy, and humid environments, but also may be nearly ubiquitous. Despite these efforts, the extent to which OA is dissolved in the bulk phase of these droplets, is absorbed to their surfaces, or exists in separate three-dimensional phases is generally not known. Typically, observations of macroscopic solutions are used to estimate this phase partitioning,5−7 but it is not clear that these are relevant © 2014 American Chemical Society
to microscopic droplets that have surface-to-volume ratios 1000 times higher and diffusional time scales one million times faster than even the smallest macroscopic solutions. The surface− bulk partitioning of OA in microscopic aqueous droplets strongly influences their ability to form cloud droplets and likely determines the reactivity of OA in the atmosphere. Therefore, observations of such droplets, although generally more difficult to make than those of macroscopic solutions, are necessary to further constrain the climate effects and atmospheric aging of OA. From Köhler theory,8 CCN activity depends on two factors: the Raoult effect, which lowers water activity by the dissolution of solute into the droplet bulk, and the Kelvin effect, which Received: March 21, 2014 Revised: May 15, 2014 Published: May 16, 2014 3952
dx.doi.org/10.1021/jp502844g | J. Phys. Chem. A 2014, 118, 3952−3966
The Journal of Physical Chemistry A
Article
Henry’s law coefficients. Although the importance of surface chemistry has been estimated,39 the uncertainty associated with these calculations is rather large compared to that of bulk (gas or aqueous) phase processes. A better understanding of the surface composition of aqueous droplets will therefore represent a significant advance in the understanding of atmospheric organic chemistry, including formation of aqueous SOA. Finally, because wet deposition is the most efficient removal mechanism for submicrometer particles,40 the surface of microscopic aqueous droplets may play an important role in the sinks as well as the sources of atmospheric OA. The model compounds used in this study are straight-chain (normal) carboxylic and dicarboxylic acids as well as cis-pinonic and oleic acid. These compounds are selected to represent a broad range of solubility, from fully miscible (malonic) to essentially insoluble (e.g, oleic, myristic). In addition, these species have molecular weights that vary by a factor of 2.7. The dicarboxylic acids have carbon numbers from three (malonic acid) to ten (sebacic acid) and are selected because of a wellknown alternation in physical properties between even and odd carbon numbers. These properties include melting point41 and vapor pressure42 as well as surfactant properties such as the Krafft temperature43 and other micelle properties.44 Although the debate about the precise origin of this even−odd alteration has been ongoing for nearly a century,45,46 it is generally believed to originate from differences in the crystal structures of even and odd-numbered compounds; even-numbered compounds exhibit larger intermolecular attractive forces and therefore higher heats of dissolution. Odd-numbered diacids have macroscopic aqueous solubility higher than those with even numbers of carbon atoms because of the increased repulsion between oxygen atoms in the head groups when oddnumbered chains are packed together.47 It is not known whether a similar alternation of surface film properties or solubility between even and odd diacids would be observed in microscopic aqueous droplets. Here droplet diameters which form on submicrometer aerosols near water vapor saturation (at RH from 99.91 to 99.96%) are measured using a technique detailed in a previous publication.48 Many previous studies have compared aerosol hygroscopicity at lower RH (typically ≤95%) and CCN activity (i.e., hygroscopicity at RH > 100%), but measurements at intermediate RH, such as those reported here, are relatively sparse. Measurements in this RH range can help resolve discrepancies between low-RH hygroscopicity and CCN activity because only as the RH approaches 100% does the Kelvin effect become comparable to the Raoult effect in controlling hygroscopicity.48,49 The aim of this study is to ascertain if the droplet diameters measured at high RH are accurately predicted based solely on bulk solubility of the model compounds and to determine if surface activity is required to explain the observed water uptake. Additionally, we explore whether the well-known alternation of macroscopic physical properties between even- and oddnumbered diacids occurs in microscopic droplets. Finally, the results of α-pinene SOA particles that are more highly oxidized than those reported by Ruehl et al.11 are presented, linking the observations of simple model aerosols with the water uptake of more complex multicomponent particles produced by gasphase oxidation reactions.
increases the relative humidity in equilibrium with a droplet because of its curved surface and is proportional to surface tension. (Here “bulk” is used to refer to the portion of microscopic droplets not adjacent to their surfaces, whereas “macroscopic” is used to refer to solutions with volumes of ∼1 mm3 or larger). The surface activity of atmospheric organic material has long been recognized; both theoretical9,10 and observational11−13 evidence indicates that organic matter is indeed found at the surface of atmospheric aqueous droplets. Calculations based on observed surface tension depression in macroscopic solutions indicate that surface activity enhances the CCN activity of ambient aerosols.5−7 However, the high surface-to-volume ratios of atmospherically relevant microscopic droplets has led to the conclusion that the reduction of surface activity cannot significantly increase the hygroscopicity of submicrometer aerosols because the depletion of organic material from the bulk that accompanies an increase in surface concentration will both raise water activity and limit the reduction in surface tension.14 However, there is evidence that surface tension reduction can enhance the hygroscopicity of submicrometer aerosols,11,15,16 although it remains challenging to distinguish between the Kelvin and Raoult effects based on CCN data alone.17 Parameterizations of only the Raoult portion of the Köhler equation can generally reproduce the general trends in ambient aerosol hygroscopicity and composition.18−20 Typically, hygroscopicity increases with oxidation state, although other factors introduce significant spread into the relationship between a single hygroscopicity parameter and a compositional variable such as oxygen-to-carbon (O:C) ratio.21 The hygroscopicity of some environmental chamber secondary organic aerosol (SOA) increases with the oxidation level (typically expressed as an O:C ratio),22,23 similar to ambient OA,18,19 in contrast to other studies that have found that hygroscopicity is independent of oxidation level.24−26 If the surface partitioning in microscopic droplets evolves with oxidation level, a trade-off between the Kelvin and Raoult effects could help explain the lack of apparent increase in hygroscopicity. Even more simple CCN measurements of several model compounds of limited solubility show that they activate as if they were fully soluble, despite calculations that indicate that at the point of activation their bulk phase concentrations would be greater than their observed macroscopic solubility limits. This has been observed for several organic compounds, including phthalic acid,27,28 cholesterol,29 and pimelic acid.30,31 The formation of a supersaturated metastable droplet phase has been invoked to explain this observation.27,28 CCN activity consistent with full solubility has also been linked to the wettability of model organic compounds.29 If these compounds contribute to CCN activity both by lowering both water activity and surface tension, this could help explain the apparent enhanced solubility. In addition to hygroscopicity, the surface arrangement of organic molecules on microscopic droplets may strongly influence the atmospheric chemistry of OA. The aqueous interface is a unique environment in which condensed-phase species are more available for reaction with gas-phase molecules in the atmosphere. Interfacial properties such as pH can differ from their bulk aqueous values,32−34 thus potentially promoting cross-reactions such as oligomerization.35 Most efforts to model the influence of the aqueous phase on particulate organic concentrations explicitly account for only the bulk reservoir3,36−38 by utilizing properties such as osmolality and 3953
dx.doi.org/10.1021/jp502844g | J. Phys. Chem. A 2014, 118, 3952−3966
The Journal of Physical Chemistry A
■
Article
EXPERIMENTAL DESIGN Submicrometer particles composed of known amounts of ammonium sulfate and model organic compounds are generated, followed by measurements of the droplet diameters (Dwet) that form on them after exposure to RH between 99.91 and 99.96%. Mixed inorganic−organic dry particles are formed first by generating ammonium sulfate particles, followed by condensation of organic material onto these seeds in an oven (for model compounds) or a flow tube reactor (for SOA). Seed particles are produced by atomization of a 0.1% aqueous ammonium sulfate (AS) solution using an aerosol generator (Brechtel Manufacturing, Inc., model 9200). The AS aerosol is then dried to RH < 10% and size-selected using a differential mobility analyzer (DMA, TSI 3081). The seed diameter (Dseed) that is selected is typically 200 nm but ranged from 175 to 250 nm. A shape-correction factor of 1.04 is applied50 only to these seed particles. Thus, for example, dry AS particles that were size-selected at 200 nm are assumed to have a volumeequivalent diameter of 194.8 nm. Temperatures in the oven ranged from 30 °C for thin coatings (∼10 nm) of the most volatile compounds (malonic acid) to 140 °C for thick coatings (∼100 nm) of the least volatile (stearic acid), and were more finely adjusted (∼10 °C) to achieve the desired Dcoated. After the model organic compounds are condensed onto these AS seed particles and allowed to cool, they are size-selected with a second DMA to generate coated (dry) diameters (Dcoated) ranging from 175 to 420 nm. No additional shape-correction factor is applied, as internally mixed particles composed of ammonium sulfate and both succinic lauric acid are nearly spherical.50 To study secondary organic aerosol generated from the oxidation gaseous α-pinene, seed particles are sent to a flow tube reactor that has been described in detail in a previous publication.51 This reactor tube is 1.30 m long and has an inner diameter of 2.5 cm. The flow rate is controlled and set to 1 lpm, resulting in a residence time of ∼37 s. Ozone is added such that its concentration in the tube is either 1.5 or 2.5 ppm. No UV lights were used to generate OH radicals, although no radical scavenger was added, suggesting that radical chemistry, in addition to dark ozonolysis, also contributed to the oxidation of the α-pinene. After exiting the flow tube, the flow was passed through a charcoal denuder to remove gas-phase species. As with model organic compounds, the coated particles were then size-selected and no shape-correction factor was applied. These SOA results were compared to previous experiments conducted under similar conditions but using a different reactor and an ozone concentration of 0.35 ppm.11 The wet droplet diameter (Dwet) measurements have been described in detail in a previous publication,48 so only an overview will be given here. After the organic material is coated onto ammonium sulfate seeds, the size-selected particles are sent to a custom-built continuous-flow streamwise thermal gradient chamber.52 This instrument is normally configured with a positive temperature gradient for producing RH just above saturation (approximately 100−101%). For these experiments, a negative (or zero) temperature gradient is used to generate a RH between 99.91 and 99.96% along the chamber centerline.48 The total flow through the instrument is 0.9 lpm, with a sheath-to-aerosol flow ratio of 8:1, resulting in a residence time of approximately 10 s. The custom-built droplet detection chamber53 uses a phase Doppler interferometer (PDI,
Artium Technologies, Inc.) to measure Dwet and droplet velocity.54 There are a number of advantages to using a PDI to measure Dwet for these experiments. The first of these is that the view volume of this probe is determined by the intersection of two laser beams, which are focused through windows onto the chamber centerline less than 3 cm below the end of the wetted and temperature-controlled walls. This allows Dwet to be measured while the droplets are still subject to the chamber centerline RH, before the flow has been disturbed in any way. The second advantage is that Dwet is determined by the phase shift of the signal measured by pairs of photodetectors. In other instruments, Dwet is determined by signal intensity and therefore varies with laser output, photodetector sensitivity, condensation on optics, etc. However, the phase shift depends, in addition to Dwet, only on the geometry of the probe, which does not vary. Therefore, the PDI can make measurements of Dwet that are more precise than those that would be possible using signal intensity alone. A third advantage of PDI is that the velocity of the droplets is measured in addition to Dwet. At a RH near saturation, Dwet is very sensitive to T, so that very slight changes in room temperature, flow rate, pressure, etc. can cause the flow to deviate from a parabolic velocity distribution. As such, deviations in droplet velocity are a helpful diagnostic that indicate when the chamber flow has been heated or cooled slightly or when the laminar flow necessary for the generation of precise RH is disturbed. The droplet sizes are measured after 10 s of exposure to RH ∼ 99.9%. Given the small diffusional time scales in microscopic droplets, it is likely that the droplets have attained their equilibrium sizes over this time scale. Most previous measurements of dynamic surface tension in aqueous solutions have found that this effect is important on shorter (< ∼1 s) time scales.55 For aqueous phase diffusion to kinetically limit droplet growth, diffusion coefficients on the order of (1 μm)2/10 s = 10−9 cm2 s−1 would be required, which is ∼1000 times smaller than that expected for nonelectrolyte organic molecules in water at room temperature.56 Consequently, the measurements reported here likely reflect equilibrium surface tension. Furthermore, because the aqueous surface is neither created nor destroyed when these passive measurements are made, surface tension inferred from the equilibrium size of microscopic aqueous droplets could be relatively free of dynamic effects and may represent a unique probe of equilibrium surface tension.
■
RESULTS Figure 1 shows droplet diameters formed at RH = 99.9%, formed on particles composed of a 200 nm ammonium sulfate (AS) seed coated with one of eight linear dicarboxylic acids. The diameters of the coated dry particles ranged from 200 nm (
