ARTICLE pubs.acs.org/est
Chemistry of Small Organic Molecules on Snow Grains: The Applicability of Artificial Snow for Environmental Studies Romana Kurkova,† Debajyoti Ray,† Dana Nachtigallova,‡ and Petr Klan*,†,§ †
Research Centre for Toxic Compounds in the Environment, Faculty of Science, Masaryk University, Kamenice 3, 62500 Brno, Czech Republic ‡ Institute of Organic Chemistry and Biochemistry, Flemingovo nam. 2, 16610 Prague, Czech Republic § Department of Chemistry, Faculty of Science, Masaryk University, Kamenice 5, 62500, Brno, Czech Republic
bS Supporting Information ABSTRACT: The utilization of artificial snow for environmentally relevant (photo)chemical studies was systematically investigated. Contaminated snow samples were prepared by various methods: by shock freezing of the aqueous solutions sprayed into liquid nitrogen or inside a large walk-in cold chamber at -35 °C, or by adsorption of gaseous contaminants on the surface of artificially prepared pure or natural urban snow. The specific surface area of artificial snow grains produced in liquid nitrogen was determined using valerophenone photochemistry (400440 cm2 g-1) to estimate the surface coverage by small hydrophobic organic contaminants. The dynamics of recombination/dissociation (cage effect) of benzyl radical pairs, photochemically produced from 4-methyldibenzyl ketone on the snow surface, was investigated. The initial ketone loading, c = 10-6-10-8 mol kg-1, only about 1-2 orders of magnitude higher than the contaminant concentrations commonly found in nature, was already well below monolayer coverage. We found that the efficiency of out-of-cage reactions decreased at much higher temperatures than those previously determined for frozen solutions; however, the cage effect was essentially the same no matter what technique of snow production or ketone deposition/uptake was used, including the experiments with collected natural snow. The experimental observation that the contaminant molecules are initially self-associated even at the lowest concentrations was supported by DFT calculations. We conclude that, contrary to frozen aqueous solutions, in which the impurities reside in a 3D cage (micropocket), contaminant molecules located on the artificial snow grain surface at low concentrations can be visualized in terms of a 2D cage. Artificial snow thus represents a readily available study matrix that can be used to emulate the natural chemical processes of trace contaminants occurring in natural snow.
’ INTRODUCTION It is now evident that natural and anthropogenic organic compounds are present in high latitude/altitude and related ice and snow environments in noticeable concentrations.1 The adsorption/desorption, diffusion, and conformational changes of the ice/snow organic contaminants are temperature and phase-dependent variables.2-5 The importance of hydrogenbonding, dipolar, and other noncovalent interactions in complexes of organic compounds with ice surfaces has been revealed from measurements performed using various spectroscopic techniques (e.g., 6-9). Some organic compounds can undergo primary photochemical or (either dark or photochemically initiated) secondary chemical processes on/in ice and snow.1,10 The photochemistry of these compounds depends on many factors, such as their optical and chemical properties, the presence of other reactive species, the optical and phase properties of the host matrix, r 2011 American Chemical Society
or temperature.11-23 Recent findings indicate that the absorption spectra of simple aromatic compounds, such as phenol derivatives 18 or benzene,20 exhibit bathochromic shifts to wavelengths that overlap with those of solar radiation at an air-ice interface. It has been demonstrated that the freezing of aqueous solutions of most of the organic (and inorganic) compounds causes ice and solute molecules to separate.1,3,9,24 This results in their increased local concentrations in a liquid (or quasi-liquid 25) phase covering the ice crystal surface or residing in a limited volume, referred to as micropockets or microveins,26 at the boundary of the grains of solid ice. Such a highly concentrated Received: December 7, 2010 Accepted: February 14, 2011 Revised: February 8, 2011 Published: March 02, 2011 3430
dx.doi.org/10.1021/es104095g | Environ. Sci. Technol. 2011, 45, 3430–3436
Environmental Science & Technology
ARTICLE
Scheme 1. Photochemistry of MeDBK and DBK
mixture may completely solidify below the eutectic point of the system. Contrary to this, a shock freezing preparation technique (spraying the solution into liquid nitrogen) produces artificial snow grains,27-33 which grow freely without the physical restrictions of the vessel walls. In the present study, we investigated the surface coverage of artificial snow grains by organic contaminants in relation to their surface self-association, diffusion, and bimolecular reactions. Analogous to our recent investigations of frozen aqueous solutions,34 the Norrish type I reaction of 4-methyldibenzyl ketone (MeDBK) was used as a cage effect probe.35 The selforganization of dibenzyl ketone on the ice surface was modeled using DFT calculations. The experiments were principally designed to demonstrate that artificial snow is an environmentally relevant study matrix and that the natural chemical processes of trace contaminants occurring in snow can be emulated in the laboratory.
’ METHODS Preparation of Artificial Snow Samples. Five types of
contaminated snow samples were used in this study. Pressurized aqueous solutions (organic solvents were avoided) of 4-methyldibenzyl ketone (MeDBK, c = 10-6-10-8 mol L-1), valerophenone (VP, c = 10-4-10-8 mol L-1), or a mixture of benzyl ketone (DBK, c = 10-6 mol L-1) and CuCl2 (c = 10-4-10-6 mol L-1) were either sprayed through a hollow cone brass nozzle into a large vessel containing liquid nitrogen (analogous to the Jacobi’s procedure 29 and our previous work, ref 33), or inside a large walk-in cold chamber at -35 °C. In addition, samples of artificially prepared pure snow (by using both methods), as well as natural urban snow collected outside the department during snowfall at an ambient temperature of approximately -10 °C, were placed into a glass vessel (desiccator) at -10 °C. Vapors of VP or MeDBK, carried by a cooled (-10 °C) stream of air at atmospheric pressure, were then slowly led through the sample. The size of the individual snow grains (0.05-0.3 mm) made by two methods was comparable to that of the natural snow grains (Figures S2-S4 in the Supporting Information). The equipment, procedures (such as determination of the specific surface area, photolysis, sample analyses), computational simulation methods, and some additional data are described in the Supporting Information.
’ STRATEGY Cage Effect. The triplet state of dibenzyl ketones, such as the 4-methyl derivative (MeDBK, Scheme 1), undergoes the Norrish type I reaction 36 - the rapid primary R-cleavage (k ≈109 s-1) followed by slower decarbonylation (k ≈ 107 s-1) 37 to give the corresponding benzyl radicals. The triplet radical pairs intersystem cross to singlet radical pairs prior to recombination to MeDPE or escape to give three different 1,2-diarylethane products (DPE, MeDPE, diMeDPE). The product distribution is statistical (1:2:1) in nonconstraining environments, but can be specific in a heterogeneous environment due to restricted translational diffusion of the radical intermediates.35 The photoproduct molar concentrations are then used for calculating the cage effect (CE), which is defined as the fraction of radical pairs that undergo reactions within a primary reaction cage (eq Eq 1).38
CE ¼
½MeDPE - ð½DPE þ ½diMeDPEÞ 100ð%Þ ðEq 1Þ ½DPE þ ½MeDPE þ ½diMeDPE
In this work, the cage effects on the snow grain surface provided information about the dynamics and availability of the photochemically generated benzyl radicals on the ice surface. Whereas water molecules of ice acting as the H-bond donors or acceptors are expected to restrict the motion of polar compounds, such as MeDBK containing a polar carbonyl group, only relatively weak polar π-interactions, phase, microviscosity, degree of the disorder of a layer covering the ice surface,2-5,34,39-43 and hydrophobic forces 44,45 should affect diffusion of nonpolar benzyl radicals.
’ RESULTS AND DISCUSSION Surface Coverage. The surface coverage of a solid provides information about the fraction of the adsorption sites occupied by molecules that is, the coverage of the outermost layer of the surface. The monolayer coverage can be determined from the Langmuir adsorption isotherm (e.g., refs 46,47), or calculated from the known solid surface area and the area occupied by a single molecule (e.g., ref 48). These data can further be used for calculation of the specific surface area (SSA), the measure of the area accessible to gases per unit mass. This term is an important physical property of snow.49-51 The values of SSA measured in terrestrial snow range on the order of 10 cm2 g-1 for melt-freeze 3431
dx.doi.org/10.1021/es104095g |Environ. Sci. Technol. 2011, 45, 3430–3436
Environmental Science & Technology crusts to 103 cm2 g-1 for fresh dendritic snow.49 The porous nature of snow crystals allows for adsorption of volatile and semivolatile compounds; therefore, it has been suggested that falling snowflakes are more effective at scavenging atmospheric particles than rain.52 Hagesawa and co-workers 53 and later our laboratory 27 employed valerophenone (VP) photochemistry (the Norrish type II reaction 54) for determination of the monolayer coverage of a silica-gel and alumina surface, respectively. The method is based on the comparison of the reaction rates obtained for different surface loads under the same irradiation conditions.53 When multiple layers of VP molecules are formed, it is assumed that the photoactive molecules in the outermost layer undergo the reaction but, at the same time, act as an internal optical filter for the remaining molecules (causing a decrease in the reaction efficiency). Using this method, we determined the SSA of artificial snow. Two techniques of contaminated snow production were used: (a) artificial pure snow (prepared using liquid nitrogen) was exposed to VP vapors or (b) aqueous VP solutions were sprayed into liquid nitrogen. The area of a single VP molecule was calculated from the structure optimized with the DFT method (6.2 10-19 m2); subsequently, the SSA of the artificial snow was determined to be 440 and 400 cm2 g-1 in the first and second case, respectively (Supporting Information). The first technique (vapor deposition) generated samples, in which the organic molecules must be, in principle, located on the ice grain surface. Both SSA values fall within the standard deviation of the measurements. Therefore, we conclude that when snow is produced from the VP solutions, the hydrophobic molecules are rejected to the same surface area available for vapor deposition, that is, they are largely rejected from the ice phase to the snow grain surface. However, this may not be true for small polar molecules, such as H2O2 or formaldehyde, which are assumed to be rather evenly distributed in the ice lattice of artificial snow.31 Our SSA values are similar to those found for artificial snow produced at 253 K (SSA = 580-680 cm2 g-1, adsorption of Kr) 55 or rounded grains of natural snow 49 (for example, SSA = 170550 cm2 g-1 was found for natural snow used to study phenanthrene surface adsorption 47), but more than 1 order of magnitude smaller than that of snow grains produced using a nebulizer (adsorption of nitrogen).56 These numbers represent an upper limit because the calculation assumes that all adsorbed VP molecules occupy the maximum area. Our experiments also showed that only a monomolecular layer of VP molecules should be present on the snow grain surface below cVP = ∼10-4 mol kg-1. The area occupied by VP is similar to that of many common small organic molecules within an order of magnitude. Therefore, we conclude that the surface of artificial snow grains prepared from aq solutions in which the hydrophobic contaminant concentrations are below ∼10-5 mol kg-1 should essentially be covered by a monomolecular layer (monolayer), provided that the molecules do not tend to form the second layer and leave other parts of the ice surface unoccupied.57 Only traces of such a second molecular layer were identified in case of benzaldehyde,58 which is structurally analogous to both VP and 4-methyldibenzyl ketone utilized in this work. Temperature Effect. Snow samples containing 4-methyldibenzyl ketone (MeDBK) were prepared from (a) the corresponding aqueous solutions by spraying them into liquid
ARTICLE
Figure 1. Temperature dependence of the cage effect in the photolysis of MeDBK in snow (color symbols) and frozen solution samples (0, from ref 34). Each point is the average of at least three measurements; error bars represent the standard deviation. A sigmoidal regression for the data (- - -, snow samples; --, frozen solution) is shown; the arrows depict the inflection point temperatures.
nitrogen or inside a walk-in cold chamber at -35 °C, or from (b) either artificial pure (using the same procedures as described above) or natural snow exposed to MeDBK vapors. The resulting MeDBK bulk concentrations in snow samples were kept between 10-6 and 10-8 mol kg-1 to maintain monomolecular coverage (see above). Only one set of samples had cMeDBK ≈ 10-4 mol kg-1, the highest possible concentration due to the limited MeDBK solubility in water. The samples were then irradiated at λ > 290 nm in a stainless-steel photochemical cold chamber 33 at various ambient temperatures. Photoreaction conversions were kept between 40 and 60% despite the fact that it has already been shown that conversion of MeDBK has only marginal effect on CE (refs 34,59 and this work). Figure 1 shows the temperature variations on CE calculated according to Eq 1. The statistical ratio of the photoproduct (DPE, MeDPE, diMeDPE; Scheme 1) concentrations, that is, CE = 0, was obtained for liquid solutions. CE of between 10% and 40% was found for snow samples irradiated at -10 °C; it increased with decreasing temperature and eventually leveled off to 100% below approximately -40 °C. The values fluctuated within 30% age points or less; the data dispersion was somewhat higher at higher temperatures. Only one sample with a maximum MeDBK concentration (c = 10-4 mol kg-1) gave an exceptionally low CE at -25 °C (Figure 1, dark-blue star). The curve in Figure 1 (solid line) is a sigmoidal regression for all data points (without cMeDBK = 10-4 mol kg-1) with the inflection point at approximately -16 °C. Mass balances of 60-70% were found for all photochemical experiments. MeDBK loss by evaporation from the snow samples in dark was found to be minimal (
