We have investigated OH uptake by a MgCl2-CaCl2 mixture at 0-29% RH, using a differential bead-filled flow tube coupled to a high-pressure chemical ionization mass spectrometer. The results show that the effect of RH on OH uptake by the mixture reproduces quantitatively that of sea salt. This observation is supported by results of our thermodynamic modeling which predicts that the deliquescence relative humidity of MgCl2·6H2O(cr) decreases in the presence of CaCl2·6H2O(cr), leading to higher surface acidification, and therefore, to greater enhancement in OH uptake under humid conditions. This conclusion is further strengthened by our earlier experimental observations which indicate that the enhancement of the OH uptake coefficient under humid conditions for sea salt is a factor of two larger than that for binary MgCl2-H2O solutions in equilibrium with MgCl2·6H2O(cr). Based on these results we conclude that surface composition determines, to a large extent, the heterogeneous reactivity of inorganic salt mixtures.
 To our knowledge, the heterogeneous OH + sea salt reaction has not been studied yet and no reaction mechanism has been proposed in the literature. However, many researchers have used NaCl, the main sea-salt component, as a model compound for determining the heterogeneous chemistry of sea salt. The use of NaCl as a model for sea salt was initially based on experimental observations of particle mass growth, which starts at similar RH for both sea salt and NaCl [Tang et al., 1997]. However, our thermodynamic model predicted a mixed deliquescence relative humidity (MDRH) of ∼32% for sea salt at 25°C, which differs significantly from that of 74.8% for NaCl (C. I. Christov, The deliquescence of sea salt at room temperature, manuscript in preparation, 2008). This predicted MDRH value for sea salt is also consistent with our kinetic observations of its heterogeneous reactivity [Park et al., 2008]. The RH dependence of OH uptake on sea salt was found to be similar to that of MgCl2, a lower deliquescing component, rather than that of NaCl [Park et al., 2008]. MgCl2 is a salt with a weak base cation, which dissociates in the presence of surface-adsorbed water. The Mg2+ ion readily reacts with OH−, producing Mg(OH)+, the formation of which was found to be thermodynamically favorable in aqueous solutions [Park et al., 2008; Wagman et al., 1982; Chase et al., 1985]. This hydrolysis mechanism, which results in the consumption of surface OH−, is responsible for lowering the surface pH (acidification) of MgCl2 exposed to water vapor, and was supported by the results of our earlier study which showed an increase in the OH uptake coefficient, γOH, for wet MgCl2 [Park et al., 2008].
 Although both MgCl2 and sea salt showed similar wetting behaviors, we found that relative humidity quantitatively enhanced γOH twice as much for sea salt as for MgCl2 [Park et al., 2008].
 The objective of the present kinetic study was to compare OH uptake by sea salt to that by a mixture of MgCl2 and CaCl2 under humid conditions. We also performed thermodynamic modeling to explore whether this mixture could be responsible for the γOH enhancement observed in our sea-salt uptake experiments. For this purpose, we developed an approach combining experimentation with modeling, which in addition to a comprehensive kinetics study of heterogeneous chemistry, takes into account chemical thermodynamics of equilibria and transformations occurring on the salt surface.
2.1. Chemical Kinetics Studies
 Experimental studies were conducted under flow conditions at 100 Torr and room temperature, using an experimental setup, described elswhere [Park et al., 2008]. The setup consisted of a differential bead-filled flow tube, consisting of identical reference and reaction tubes, coupled to a high-pressure chemical ionization mass spectrometer. We used the reaction H+O2+M to produce OH at 100 Torr, which was detected with SF6−. Relative humidity was controlled by mixing a water-vapor-saturated flow of N2 with dry N2 in varying proportions. Flow rates were monitored with calibrated electronic mass flow meters (Tylan), while pressure in the differential flow tube was measured with an absolute pressure gauge (MKS 1000, Baratron).
 Glass beads in the reference tube were coated with halocarbon wax, whereas those in the reaction tube were coated with the inorganic materials of interest. The bead coatings were made using techniques described previously [Park et al., 2008]. The materials used were MgCl2 anhydride (Sigma-Aldrich, 99.5%), CaCl2 dihydrate (Mallinckrodt, 99%), and sea salt (Sigma-Aldrich). During preparation, the salt coatings were exposed to ambient air. Such exposure led to prompt formation of thermodynamically stable MgCl2·6H2O(cr) and CaCl2·6H2O(cr) saturated solutions at room temperature, which is in agreement with experimental solubility data for binary and ternary solutions and with thermodynamic model predictions.
2.2. Thermodynamic Model Parameterization
 We modeled the solid-liquid equilibria in a MgCl2-CaCl2-H2O system to determine key thermodynamic parameters — such as water activity, the mean activity coefficients of MgCl2 and CaCl2 in binary and mixed solutions, DRH and MDRH, and the surface pH — which are responsible for establishing the chemical equilibrium and which, therefore, control the heterogeneous reactivity of the salts and their mixture. In our thermodynamic model, we used the Pitzer approach [Pitzer, 1973, 1991]. This allowed us to calculate, with an accuracy ranging from 2% to 6%, the activity coefficients in unsaturated solutions of electrolytes [Christov, 2005]. The fundamental Pitzer equations and this modeling approach have been discussed in detail elsewhere [Christov, 2005; Christov and Moller, 2004].
 In our model parameterization, to describe the binary MgCl2-H2O and CaCl2-H2O solution behavior, the Pitzer parameters of pure electrolytes, such as β(0), β(1), and Cϕ, had to include ion interactions. These binary solution parameters of Mg-Cl interactions were taken from the literature [Balarew et al., 1993]. The Ca-Cl pure electrolyte parameters at 25°C are calculated from 0°C-to-250°C temperature dependence equations [Christov and Moller, 2004]. Both sets of parameters are valid up to high molality, including saturation; their applicability has been previously demonstrated by solubility calculations for various binary, ternary, and multicomponent systems [Christov, 2005; Christov and Moller, 2004; Balarew et al., 1993].
 To describe solid-liquid equilibria, the solid phase should be incorporated in the solution model. The chemical potential of solids is determined by a thermodynamic equilibrium function, such as the solubility product, or lnKosp. We determined solubility products of MgCl2·6H2O(cr) and CaCl2·6H2O(cr), which crystallize from the corresponding saturated MgCl2-H2O and CaCl2-H2O solutions, using the recommended values [Rard and Miller, 1981; Zdanovskii et al., 2003] for the molality of saturation, msat, and the binary solution parameters accepted in this study.
 Our model for a MgCl2-CaCl2-H2O system at 25°C was parameterized using the experimental mixed equilibrium solubility data [Zdanovskii et al., 2003]. In this ternary system, new binary and ternary ion interactions between Mg2+, Ca2+, and Cl− were considered as well. The binary mixing Pitzer parameter θ(Mg, Ca) of 0.007 [Harvie et al., 1984] was accepted in this work to describe interactions between two ions of the same sign. To describe interactions between three ions of different signs, the ternary mixing Pitzer parameter ψ(Mg, Ca, Cl) was evaluated using the equilibrium solubility data [Zdanovskii et al., 2003] and found to be −0.017. The equilibrium solubility data [Zdanovskii et al., 2003] were also used to determine the solubility product of 2MgCl2·CaCl2·12H2O, precipitating from the saturated ternary solutions. According to the ternary system solubility data [Zdanovskii et al., 2003], the phase diagram at 25°C consists of fields of equilibrium crystallization of MgCl2·6H2O(cr), CaCl2·6H2O(cr), and 2MgCl2·CaCl2·12H2O(cr), i.e., the precipitation of the highly soluble metastable CaCl2·4H2O(cr) is not reported, and therefore, is not considered here.
 To evaluate the pH-dependence of the MgCl2-CaCl2-H2O solutions in equilibrium with bishofite, we have added the Ca acid-base parameters [Christov and Moller, 2004] — such as the binary H-Cl and Ca-OH and the mixing Ca-H, Ca-H-Cl, OH-Cl, and Ca-OH-Cl parameters — as well as the Mg acid-base parameters [Harvie et al., 1984] — such as the binary Mg-OH and the mixing Mg-H, Mg-H-Cl, and Mg-OH-Cl parameters.
3. Results and Discussion
3.1. Chemical Kinetics Studies
 Determination of the OH uptake coefficient, γOH, was based on measurements of the ratio between the OH signal with the radical-containing flux entering the reaction flow tube and the signal with the radicals flowing through the reference flow tube. Since the signal is proportional to the amount of OH remaining after the heterogeneous reaction with the bead-surface coating, the ratio is determined by OH uptake. The uptake coefficient, γOH, was extracted from the signal ratio, using a methodology developed specifically for this purpose, and described in detail elsewhere [Park et al., 2008].
Figure 1 shows the results of the γOH measurements as a function of RH for sea salt, MgCl2, and a mixture of MgCl2 and CaCl2 using the typical molal ratio for solid sea-salt aerosols: m(MgCl2)/m(CaCl2) = 7/2 [Park et al., 2008]. Shown in the figure is an enhancement in the OH uptake coefficient that is determined as a ratio between the values of gamma measured under wet conditions to that of (4.5 ± 1) · 10−3 determined under dry conditions. As Figure 1 shows, the MgCl2-CaCl2 mixture displayed the enhanced OH uptake coefficients that quantitatively reproduced those of sea salt from 0 to 29% RH, whereas pure MgCl2 displayed a less enhanced OH uptake coefficient. Mixtures of MgCl2 with other sea salt components, such as NaCl, KCl, and Na2SO4, did not change heterogeneous reactivity within experimental error.
 Although, typical experiments were 20–30 min, some were continued for 3–4 hours, showing no time dependence for OH uptake.
3.2. Thermodynamic Modeling
 The lack of a time-dependent OH uptake in our experiments was taken as evidence that supersaturation was not important under our conditions. This allowed us to exclude thermodynamically-metastable solid phases from our consideration.
 For the deliquescence behavior of the salt hydrates in the equilibrium binary systems, we calculated the mean activity coefficients, γ±, of MgCl2 and CaCl2 in the corresponding binary solutions at 25°C as a function of their molality, up to saturation, as shown in Figure 2a. Figure 2a shows that the activity coefficient increases exponentially as the solution molality increases, suggesting that the hexahydrate MgCl2·6H2O(cr) and CaCl2·6H2O(cr) salts will efficiently crystallize in the supersaturated solutions, which are thermodynamically unstable. For comparison, Figure 2a also shows the literature data [Rard and Miller, 1981; Robinson and Stokes, 1968; El Guendouzi et al., 2001], which are consistent with the calculated values. Additionally, the Figure 2a shows that γ±CaCl2 in a binary CaCl2-H2O solution, saturated with CaCl2·6H2O(cr), is much higher than γ±MgCl2 in a binary MgCl2-H2O solution, in equilibrium with MgCl2·6H2O(cr) (solid triangles). Because (γ±CaCl2)binary > (γ±MgCl2)binary at saturation, it suggests that CaCl2·6H2O(cr) crystallizes at a lower RH than MgCl2·6H2O(cr) since the equilibrium crystallization constant is inversely proportional to the cube of the mean activity coefficient [Atkins, 1998]. This suggestion is also supported by our RH modeling that predicts DRH = 22% for CaCl2·6H2O(cr) and DRH = 34.4% for MgCl2·6H2O(cr). The predicted DRH value for MgCl2·6H2O(cr) is further strengthened by results of our previous kinetic observations of its heterogeneous reactivity [Park et al., 2008].
 Mixed solutions with a MgCl2-to-CaCl2 molal ratio of 7:2, which is representative of sea-salt aerosol and was therefore used in our uptake experiments, are predicted to be in equilibrium with MgCl2·6H2O(cr). For this reason, we calculated γ±MgCl2 in the bishofite-saturated mixed MgCl2-CaCl2-H2O solutions as a function of added CaCl2 (Figure 2b). Figure 2b shows that γ±MgCl2 increases significantly, from 26.6 to 65.4, as the CaCl2 molality increases from 0.0 m to 3.0 m, suggesting that the efficiency of crystallization of MgCl2·6H2O(cr) is considerably higher in the presence of CaCl2. This implies that adding CaCl2 to MgCl2·6H2O(cr) should lead to its crystallization in the ternary system at lower RH than in the binary solution because (γ±MgCl2)ternary > (γ±MgCl2)binary.
 To verify this suggestion for the ternary MgCl2-CaCl2-H2O system, we calculated the MRDH dependence of MgCl2·6H2O(cr) on the CaCl2 molality, as shown in Figure 3. Results of our modeling show that the MRDH decreased from 34.4% in MgCl2-H2O to 19.1% in the ternary MgCl2·6H2O(cr)+2MgCl2·CaCl2·12H2O(cr) system invariant point. For uptake measurements with mMgCl2/mCaCl2 = 7:2, the MRDH was calculated to be 31.8% — i.e., the deliquescence of MgCl2·6H2O(cr) occurred at a 2.6% lower RH in the mixed, saturated ternary MgCl2-CaCl2-H2O solutions than in the saturated binary MgCl2-H2O solution.
 We also calculated the pH-dependence of MgCl2-CaCl2-H2O solutions saturated with MgCl2·6H2O(cr) that showed a decrease in pH from 4.9 to 3.9 as the CaCl2 molality increased from 0.0 m to 3.0 m. This result suggests that surface acidification of mixed MgCl2-CaCl2-H2O solutions in equilibrium with bishofite is greater than that of saturated binary solutions. This model prediction is in line with the results of our kinetics studies, which showed greater enhancement under humid conditions in the OH uptake coefficient for the mixture of MgCl2 and CaCl2, as Figure 1 shows.
4. Atmospheric Implications
 To date, only one experimental study [Park et al., 2008] has been performed at relative humidities below the deliquescence point of sea salt to investigate the impact of OH uptake on the heterogeneous chemistry of sea-salt aerosols. Results of the current study provide additional insight into sea-salt reactivity at low relative humidity, which is determined, to a large extent, by its lower deliquescing components, namely MgCl2·6H2O(cr) and CaCl2·6H2O(cr). Additionally, we found that the thermodynamics of sea salt under ambient conditions dictates the equilibrium surface composition, which in turn determines the net reactivity of sea-salt aerosols. This is important for atmospheric models, which consider the presence of these aerosols.
 We also found that, under our experimental conditions, supersaturation was of minor importance; however, if atmospheric sea-salt aerosols exist as supersaturated solutions, supersaturation and its effects on heterogeneous reactivity would be important in the atmosphere.
 We performed an experimental study of OH uptake on sea salt, MgCl2, and a mixture of MgCl2 with CaCl2 under humid conditions. To our knowledge, this is the first kinetic-thermodynamic study of the chemistry of the salt mixture. The results indicate that OH uptake by the salt mixture is enhanced in the presence of water vapor and is qualitatively the same as OH uptake by sea salt under humid conditions. The quantitatively similar wetting behaviors of sea salt and the salt mixture can be thermodynamically explained by assuming that these mixtures form similar surficial equilibrium interfaces. Based on our results, we conclude that surface composition makes a major contribution to the heterogeneous reactivity of inorganic salt mixtures.
 These results, along with our previous ones [Park et al., 2008], provide a basis for elucidating the contribution of the reactions under study to the heterogeneous chemistry of sea salt. Further research, including product studies and determination of the reaction mechanism, is clearly needed to establish the role of these reactions in the marine troposphere, and will be the subject of future work.
 This study was supported partially by NSF and DOE grants.