High-temperature mixtures of magmatic and atmospheric gases

3.1. Concepts

The molecular composition of a chemical system will vary so as to reduce, and under true equilibrium conditions minimize, the Gibbs energy of the system within the constraints imposed by a fixed (bulk) atomic composition. At the Gibbs energy minimum, the system is said to be at equilibrium. Using thermodynamic modeling software (such as HSC Chemistry), a series of mass balance and mass-action (i.e., equilibrium) relations can be solved using a Gibbs energy minimization algorithm. This process allows the equilibrium composition of a mixture to be calculated over a range of temperatures and starting compositions. More detailed discussions of gas-phase thermodynamic models are presented elsewhere [Symonds and Reed, 1993; Gerlach, 2004].

The use of equilibrium models for volcanic gases at T < 800°C may not be valid if the composition is controlled by kinetic, rather than thermodynamic factors [Symonds and Reed, 1993]; however, this is highly dependent on the cooling history of the gas, and equilibration will continue below 800°C if the lifetime of the high-T mixture is sufficiently long. Hot magmatic gases (T > 800°C) can be assumed to attain equilibrium rapidly, and so we consider kinetic effects at this temperature to be negligible, as others have previously [Gerlach, 2004].

3.2. Models for Magmatic and Atmospheric Gas Mixtures

In previous work [Gerlach, 2004], the commercial thermochemical modeling software HSC Chemistry was used to investigate the equilibrium composition of a mixture as a function of temperature, with the mixture defined by an input composition of several major C-O-S-H-Cl-F-Br-N-Ar species. Gerlach [2004] showed that mixing (i.e., high-T oxidation of magmatic gases by atmospheric gases) increases the levels of trace oxidants such as BrO and ClO by several orders of magnitude relative to a pure magmatic gas, and that higher magmatic gas temperatures further promote the formation of these trace oxidants. This earlier model spanned 275 species, although for many of these, the equilibrium concentration is below that of the calculation threshold (10⁻³⁶ mols). HSC Chemistry can also be used to investigate the equilibrium composition of a mixture as an explicit function of the amount of mixing (V_A/V_M). In line with recent measurements of iodine speciation in volcanic plumes [Aiuppa et al., 2005], it is desirable to extend the model to include iodine chemistry.

A model for C-O-S-H-F-Cl-Br-I-N-Ar speciation was created (termed the full HSC model), spanning 475 species taken from the HSC thermochemical database (which lists over 17,000 species). The large number of species involved in the full HSC model leads to unnecessary complexity and increased calculation times. To optimize the model, the model was run using averaged magmatic (Table 1) and atmospheric gas compositions (Table 2), at 1000°C over the range 0 < V_A/V_M < 0.1 by adding successive increments of atmospheric gases (i.e., N₂, O₂ and Ar) to the magmatic gas mixture. Species that were predicted to have a mixing ratio below a threshold (10⁻¹⁸) over the entire range 0 < V_A/V_M < 0.1 were removed from the model. This procedure reduced the species set to 110 species (see Appendix A). Although there are inherent disadvantages with using an averaged magmatic composition, particularly as the mixing ratios of HBr and HI are estimated and uncertain, this averaged composition is sufficient to reveal general characteristics of high-T mixtures of magmatic and atmospheric gases.

The full and reduced HSC models were compared over 0 < V_A/V_M < 0.1 at 1000°C to verify that the inclusion, or omission of species below the rejection threshold had a negligible effect on the concentration of any other species above that threshold. Minor species (i.e., <10⁻¹⁸) are decoupled from each other, and their equilibrium concentration only depends on the concentrations of the major species. If the atomic composition of iodine is set to zero, the reduced HSC and Gerlach models generate identical results.

Several trace species that have been detected in volcanic environments, such as halogenated organic compounds (e.g., [CCl₃F] = 3.7 ppbv in dry fumarolic gas at Vulcano, Sicily [Schwandner et al., 2004]) were removed. Our investigations with HSC Chemistry show that high-temperature thermodynamic processes do not control the formation of halogenated organic compounds (as these species are unstable in the high-T mixture). Therefore it would be inappropriate to attempt to apply thermodynamic models to the genesis of these species, and low-temperature kinetic models are more appropriate.

A comparison of the full and reduced HSC models over the range 0 < V_A/V_M < 0.1 at different temperatures (700°C < T < 1300°C) reveals that while several omitted species are expected to exceed mixing ratios of 10⁻¹⁸, these mixing ratios never exceed 10⁻¹⁶. The reduced HSC model can therefore be used with confidence.

The reduced HSC model presented here has been optimized to give the broadest coverage of the important C-O-S-H-F-Cl-Br-I-N-Ar species that are likely to form in high-T mixtures of magmatic and atmospheric gases, and offers improved calculation times and usability than either the full HSC or Gerlach species sets.

3.3. Comparisons With Pure Magmatic Gas Models

Several different software packages are available to model volcanic gas equilibria, ranging from very general thermodynamic models to specialist volcanic gas models. A significant advantage of using a general thermodynamic package over a more specialist package is the ease with which compositional parameters such as V_A/V_M can be explored, alongside physical parameters such as temperature and pressure.

The reduced HSC model was compared with two specialist models: Gasmix [Kress et al., 2004] and Solvgas [Symonds and Reed, 1993]. Gasmix is a simple model for C-O-S-H-Cl-F gas speciation although recent developments in remote sensing and direct sampling [Bobrowski et al., 2003] suggest that C-O-S-H-Cl-F speciation is inadequate to completely describe the chemistry of volcanic gases. Solvgas can be extended to consider many more species of known or suspected importance in volcanic plumes (including several reactive N and Br species) and has been successfully applied to a range of important problems [Symonds et al., 1992; Symonds and Reed, 1993].

For a pure magmatic gas cooled from 1300°C to 700°C (with the atomic compositions of bromine and iodine set to zero), the three models gave excellent agreement for the major species and reveal the expected temperature stabilities well established in past work [Gerlach and Nordlie, 1975]. Although some differences are evident when other species are included in the three models (usually at mixing ratios below 10⁻⁶), the inclusion of minor species does not significantly perturb any other species (either major or minor) within the model. Figure 1 shows the results from the reduced HSC model for the major C-O-S-H species (i.e., H₂O, CO₂, SO₂, H₂S, CO) and Figure 2 shows the results for the major halogen species (i.e., HF, HCl, HBr, HI). While HF, HCl and HBr are mostly invariant with temperature, HI in magmatic gas is increasingly unstable at high temperatures [Aiuppa et al., 2005] due to the formation of atomic iodine.

Neither Solvgas nor Gasmix allow for V_A/V_M (or some equivalent parameter) to be varied over a range although it is possible manually to set concentrations for O₂ and N₂ for each data point within Solvgas. HSC Chemistry is therefore a more convenient choice for investigating the composition of high-T magmatic and atmospheric gas mixtures.

4.1. Nitrogen Chemistry

NO_x species are formed naturally (i.e., biotic or lightning fixation) and anthropogenically (i.e., combustion of fuel-N species) and play an important role in the chemistry of the troposphere [Wayne, 2000]. Mixing ratios for NO_x vary hugely and are greatly dependent on the proximity of a surface source; remote from these sources, background levels of both NO and NO₂ are usually on the order of 10 ppt [Atkinson, 2000]. Tropospheric ozone levels are strongly dependent on the concentration of NO_x species; high levels of NO_x species are responsible for “photochemical smogs” resulting from the oxidation of organic compounds leading to tropospheric ozone formation [Atkinson, 2000]. Fixed nitrogen (i.e., nitrogen species other than N₂) is an important source of nitrogen for living systems as atmospheric N₂ is essentially inert at ambient temperatures.

Elevated levels of NO_x have been detected in volcanic plumes, and it has been suggested that the source of this NO_x is thermal equilibration of atmospheric N₂ and O₂ at the high temperatures associated with the vent, or the magma itself [Huebert et al., 1999; Mather et al., 2004a, 2004b]. Subsequent quenching of the mixture with cold atmospheric air results in a plume composition with elevated NO_x levels relative to the background. Simple modeling [Mather et al., 2004a, 2004b] has shown this mechanism to be plausible.

Volcanic NO_x is formed in the presence of hot magmatic gases, so we first determine the thermodynamic stability of NO_x species within a high-temperature mixture of atmospheric and magmatic gases. Using the reduced HSC model, the equilibrium composition of mixtures in the range 0 < V_A/V_M < 0.1 were investigated at a temperature of 1000°C. In order to distinguish the effects of reaction, from dilution by the magmatic gas, the calculations were repeated using argon (to serve as an inert diluent) to generate high-temperature mixtures of atmospheric gas and argon at equivalent dilutions (solid lines). Results are shown in Figure 3.

At low V_A/V_M, the stability of both NO and NO₂ are greatly reduced in the high-T mixture of magmatic and atmospheric gases relative to a high-T mixture of inert (i.e., argon) and atmospheric gases. Fixation of N₂ to NO_x is hindered at low V_A/V_M because oxygen is scavenged by the oxidation of H₂S and H₂. At high V_A/V_M, the results of the two models are very similar, and the magmatic gases no longer inhibit the formation of NO_x. Gerlach and Nordlie [1975] termed the step-like feature in Figure 3 a compositional discontinuity; it corresponds to the complete oxidation of reduced species (H₂S and H₂) in the mixture and gives rise to a sharp increase in the stability of oxidized species such as NO. Other oxidized trace species (i.e., HNO_x, x = 1, 2, 3 or XNO, X = Cl, Br, I) will not form in the absence of H- or halogen-bearing magmatic gases.

To model the expected mixing ratios of NO_x produced by open vent degassing, further dilution and cooling of the mixture in the range V_A/V_M > 0.1 was modeled by nonreactive dilution (i.e., quenching) of the mixture formed at V_A/V_M = 0.1. Where hot magmatic gases mix with hot atmospheric air, nitrogen fixation (N₂ → NO_x) may continue beyond V_A/V_M = 0.1, and we repeat equilibrium calculations in the range 0 < V_A/V_M < 1 and model for nonreactive dilution at V_A/V_M > 1. Results are presented in Figure 4 over the range −2 < log V_A/V_M < 5.

The high levels of HNO₃ (200 ppb) detected on the crater rim of Masaya by Mather et al. [2004a] implies HNO₃ concentrations at the vent of 200–2000 ppm (corrected for 1000–10000 times dilution), far higher than the mixing ratio for HNO₃ predicted using this gas-phase model (Figure 3). The reduced HSC model also predicts that the mixing ratio of NO₂ would be small compared to the mixing ratio of NO in the high-T mixture, whereas measurements at Masaya found comparable levels of NO and NO₂ (0.1 ppb at crater rim; implying mixing ratios of 0.1–1 ppm at vent). The observed mixing ratio for NO was also less than that predicted by the model for high-T mixtures of magmatic and atmospheric gases presented here. The decrease of the ratio NO/NO₂ between that predicted in the high-T mixture (NO/NO₂ > 100) to that measured at the crater rim (NO/NO₂ ∼ 1) suggests a progressive oxidation of NO to higher oxides of nitrogen (NO₂ and HNO₃) in the cooling plume.

However, the modeling results for total reactive N (mainly NO + NO₂) are consistent with the vent concentrations of total reactive N of 200 ppm (mainly HNO₃) estimated from the measurements. Therefore high-T equilibration of the magmatic gas with atmospheric air may initially generate elevated levels of NO, which is further oxidized to NO₂ and converted to HNO₃ by low-temperature chemical processes within the near-source volcanic plume.

Oxidation of NO with HO₂ (and peroxy radicals, RO₂) is a major pathway to NO₂ formation in the troposphere [Wayne, 2000]. Gerlach [2004] reported elevated levels of H and OH in high-T magmatic gases, and suggested that reaction between H and O₂(R1a) would generate elevated levels of HO₂. Figure 5 shows that below the compositional discontinuity, the modeled mixing ratios of H and OH are comparable, as the major source of OH is thermal dissociation of water (i.e., H₂O + Δ → H + OH). Above the compositional discontinuity (where NO is stable), the mixing ratio of H is much smaller than OH as oxidative dissociation of water is now the major pathway to OH (i.e., H₂O + O → 2OH), and any H that is generated in the high-T mixture is oxidized to OH, and not HO₂. Since H and NO are stable in very different regimes, it seems unlikely that there would be sufficient HO₂ generated by (R1a) to react with NO through (R1b).

The mixing ratio of OH increases above the compositional discontinuity, and schemes can also be written for OH-initiated oxidation of NO. Although OH-initiated oxidation of NO is a minor pathway to NO₂ in the background troposphere [Wayne, 2000], the elevated levels of OH predicted for high-T magmatic and atmospheric gas mixtures would promote OH (R2) chemistry relative to HO₂(R1) chemistry.

Thermal dissociation (R3) suppresses [NO₂] and in a high-T mixture, NO₂/NO is small. Upon cooling, thermal dissociation becomes less important, and NO₂ gains stability relative to NO. The observed ratio of NO₂/NO may therefore reflect a shift (i.e., reequilibration) toward higher NO₂/NO at lower (i.e., ambient) temperatures.

Nitric acid may then be generated by reaction (R4).

If the conversion of NO to HNO₃ is quantitative on the timescale of ∼10 s, -–(R4) could generate levels of HNO₃ at the crater rim comparable to those measured by Mather et al. [2004a].

A further possibility is a heterogeneous reaction catalyzed on the surface of volcanic aerosols. (R6) generates surface-bound nitric and nitrous acids, which may be leached by water vapor to form nitric and nitrous acid droplets [Thompson and Margey, 2003]:

The photolysis rate of HNO₃ is several orders of magnitude slower than the photolysis rates of either HONO or NO₂ [Madronich and Flocke, 1998]. A nighttime peak in HNO₃ would therefore imply the involvement of either HONO or NO₂, due to the presumed absence of any such diurnal signal in NO formed from high-T chemistry. This approach may prove generally useful in determining how low-T chemistry modifies the plume composition from that predicted by high-T equilibrium modeling, and is readily testable through field experiments.

4.2. Halogen Chemistry

Monatomic halogen species (i.e., X = Br, Cl, I) are of great atmospheric importance because of their potential to participate in catalytic cycles for ozone destruction (via XO) [Wayne, 2000]. The major background sources of monatomic halogens include photochemical degradation of anthropogenic (i.e., CFCs) and biogenic halides (i.e., CH₃Br, CH₃Cl, CH₃CCl₃, CH₃I, CH₂I₂) and oxidation of sea-salt halides. However, these atomic halogens are not measurable by current remote spectroscopic techniques.

Halogen oxides (XO and OXO, X = Cl, Br, I) on the other hand are detectable, and can be used as an indicator of halogen-catalyzed ozone destruction. Field campaigns have shown a clear anti-correlation between BrO and O₃ in both the polar [Tuckermann et al., 1997] and midlatitude troposphere [Matveev et al., 2001]. Background levels of XO species are typically on the order of a few ppt; however, in the vicinity of salt lakes the mixing ratios of XO may be considerably higher, and mixing ratios of up to several hundred ppt have been measured for BrO and ClO [Hebestreit et al., 1999; Stutz et al., 2002]. The major source of iodine oxides into the troposphere is the photolysis of biogenic organic iodides (i.e., CH₃I, CH₂I₂) from marine organisms (such as algae) released into the marine boundary layer where mixing ratios of up to 6 ppt (for both IO and OIO) have been measured using DOAS methods [Alicke et al., 1999].

The halogen chemistry of volcanic plumes has previously been restricted to measurements of HCl and HF alone. As shown in Figure 2, the concentrations of HCl and HF are invariant with temperature in a pure magmatic gas. This invariance can be attributed to the stability of HCl and HF, and the relative instability of other trace Cl or F bearing species in magmatic gases. However, recent developments in remote sensing technology have led to the first measurements of BrO in volcanic plumes [Bobrowski et al., 2003], and prompted similar efforts to detect other oxidized trace halogen species such as ClO [Lee et al., 2005]. While data on volcanic bromine are currently scarce and volcanic iodine even more so, thermodynamic modeling of mixtures derived from an averaged magmatic gas is sufficient to reveal several important chemical features of volcanic halogen chemistry [Aiuppa et al., 2005].

Using the reduced HSC model, the equilibrium composition of a mixture at 1000°C (with an initial composition as given in Table 1) was calculated as a function of V_A/V_M over the range 0 < V_A/V_M < 0.1. Figures 67–8 show the results of the calculations for chlorine, bromine and iodine respectively (only HX, X and XO are displayed for clarity). Figure 9 shows the results for the diatomic halogen species (i.e., X₂ and XY). The compositional discontinuity can be seen clearly in all four figures and occurs at V_A/V_M = 0.06, whereupon elevated levels of reactive halogens (i.e., X, XO, X₂ and XY) are generated. Approximate enhancement factors (i.e., the increase in mixing ratio of each species on crossing the compositional discontinuity) are given for each species in Table 3.

The mixing ratios for reactive bromine and chlorine species are comparable to those predicted by Gerlach [2004] for an 85% magmatic: 15% atmospheric gas mixture (i.e., V_A/V_M = 0.18) but are generated at much lower values of V_A/V_M. This result may be important for open vent degassing, as the compositional discontinuity (and hence trace oxidized species) can be reached at relatively low V_A/V_M, where the mixture may still be at sufficiently high temperatures to attain full equilibrium.

The thermodynamic behavior of iodine within the mixture is markedly different from that of the other halogens. The ratio X/HX in a pure magmatic gas is significantly greater for iodine (∼1) than for either bromine (∼10⁻³) or chlorine (∼10⁻⁶). We attribute this to weakening of the H-X bond from Cl to I, and suggest that while high-T oxidative dissociation (i.e., HX + O → X + OH) may be important for generating reactive bromine and chlorine species [Gerlach, 2004], this mechanism is much less important for the production of reactive iodine. High-T equilibration of a pure magmatic gas is sufficient in itself to generate comparable levels of HI and I by thermal dissociation (i.e., HX + Δ → H + X). For the magmatic gas composition used here, the mixing ratios of Cl, Br and I are comparable at V_A/V_M = 0 and 1000°C, although the uncertainties in the mixing ratios of HBr and HI are large. At higher temperatures (i.e., 1200°C), the mixing ratios of Br and Cl at V_A/V_M = 0 increase (above those of I) as thermal dissociation becomes more important, although oxidative dissociation remains the dominant process for Br and Cl generation.

Gerlach [2004] showed that formation of primary BrO by high-T oxidation of HBr by atmospheric gases cannot account for the 1 ppbv of downwind BrO measured at Soufrière Hills Volcano [Bobrowski et al., 2003]. Gerlach [2004] suggested that low-temperature oxidation of Br radicals (generated by high-T oxidative dissociation of HBr) is the source of BrO, and used thermodynamic modeling with HSC Chemistry to show that a sufficient amount of Br is generated in the high-T mixture for this process to be plausible. The major oxidants in a cooling plume are likely to be volcanogenic OH, along with much smaller amounts of volcanogenic HO₂ and atmospheric O₃, which oxidize Br to BrO through -–(R10).

An analogous case can be made to suggest that low-T oxidation of Cl (generated by high-T oxidative dissociation of HCl) could be the source of measured 25 ppbv levels of downwind ClO measured at Sakurajima Volcano [Lee et al., 2005].

If we assume that once formed, the reactive halogens persist and are neither gained from, nor lost to form, HX species, then we can estimate the mixing ratios of the reactive halogens at various plume dilutions by treating the plume as a quenched, diluting mixture. This treatment allows for the possibility of processes that inter-convert reactive halogens, and conserve the amount of reactive halogens in the mixture. To investigate the mixing ratios of reactive halogens produced from open vent degassing, dilution and cooling of the mixture at V_A/V_M > 0.1 was modeled by nonreactive quenching of the mixture formed at V_A/V_M = 0.1. Where hot magmatic gases mix with hot atmospheric air (e.g., above a lava lake, or within a lava dome structure), oxidative dissociation (HX + O → X + OH) may continue beyond V_A/V_M = 0.1, and we repeat equilibrium calculations in the range 0 < V_A/V_M < 1 and model nonreactive dilution at V_A/V_M > 1. Results are presented in Figure 10 over the range −2 < log V_A/V_M < 5.

Figure 10 reveals that the mixing ratios of Cl and Br are expected to be comparable, representing a balance between initial concentrations of HX and the strength of the HX bond. Simultaneous determinations of HCl, HBr, ClO and BrO are essential to determine if the ratio ClO/BrO lies close to (predicted) Cl/Br in the high-T mixture, suggesting an initial high-T oxidative dissociation step. For the magmatic gas composition used here (with HI = 0.01*HBr, based on measurements from Mt. Etna [Aiuppa et al., 2005]), the mixing ratios of I are 1—2 orders of magnitude less than either Cl or Br at high V_A/V_M. Further measurements of HI (or IO) are essential to assess the importance of reactive iodine chemistry within volcanic plumes.

Reactive halogen species (i.e., X) may also participate in auto-catalytic reactions on the surface of acid aerosols whereby halogen atoms are initially oxidized to HOX. HOX may then react with HX on the surface of acid aerosols (denoted {HX}) to liberate gaseous X₂, which photolyzes readily during the day to give two halogen atoms. These halogen atoms may reenter the cycle or become oxidized to form XO through -–(R10):

(R11) acts to increase total reactive halogen due to the involvement of HX. The importance of this heterogeneous reaction will depend on the relative rates of reaction of HOX with OH (R8) and of HOX with {HX} (R11). An analogous scheme was invoked for the release of Br (and Cl) into the arctic Marine Boundary Layer during the springtime (i.e., when the photolytic step is activated) from the reaction of HOX with HX on the surface of sea-salt aerosols [Vogt et al., 1996].

The contribution of the heterogeneous pathway for each halogen could be tested by simultaneously measuring HX/SO₂ (i.e., using filter-packs and ion chromatography to determine SO₂, gaseous HX and particle HX) and XO/SO₂ at various points along the length of the plume. BrO/SO₂ has been found to be negligible at the vent (implying a nonthermal genesis), but increases to a maximum of ∼10⁻⁷ a few km downwind, and this maximum is sustained in the plume of Mt. Etna (implying that losses of BrO are small) [Bobrowski et al., 2005]. For the homogenous mechanism, the ratio HBr/SO₂ will remain constant at all points along the plume, as the mixing ratio of both species only changes with dilution, and will show little correlation with BrO/SO₂. The heterogeneous mechanism would result in a strong anti-correlation between HBr and BrO. A diurnal signal in BrO would also indicate the involvement of a photolytic step (R12) implying the action of heterogeneous chemistry, although the current method of detection (DOAS) cannot easily be applied at night.

4.3. Sulfur Chemistry

The reduced HSC model was used to calculate the compositions of a mixture at 1000°C in the range 0 < V_A/V_M < 0.1, and the results for certain sulfur species are shown in Figure 11.

Sulfur species less oxidized than SO₂ (such as H₂S and S₂) become very unstable toward the compositional discontinuity. The only S species to persist across the compositional discontinuity is SO₂ itself, and there is a maximum in the mixing ratio of SO₂ at the compositional discontinuity due to complete oxidation of reduced S species to SO₂, without the further oxidation to SO₃ and H₂SO₄ that occurs at higher V_A/V_M.

Within the model, condensation of primary H₂SO₄ (i.e., that which forms from high-T equilibrium) into droplets cannot account for the observed high levels of near-source sulfate, where (H₂SO₄ + SO₄²⁻)/SO₂ = 0.01 [Allen et al., 2002; Mather et al., 2003]. Unlike oxidized halogen species, H₂SO₄ can be detected at the vent, and it may be the case that reactive halogen chemistry occurs largely on the surface of sulfuric acid droplets. Therefore understanding sulfur chemistry may be vital to understanding other processes in the volcanic plume.

Recent isotopic studies (T. A. Mather et al., The oxygen and sulfur isotopic composition of near-source volcanic sulfate aerosol and implications for its origin, submitted to Journal of Geophysical Research, 2005) show that the near-source sulfate bears a δ¹⁸O characteristic of the magma, and does not support sulfate formation by oxidation of magmatic SO₂ to H₂SO₄ with atmospheric oxygen above the compositional discontinuity. This observation suggests that the high levels of SO₃ that are predicted to form in the high-T mixture do not react with H₂O to generate H₂SO₄. Attempts to measure SO₃ near volcanic vents, and isotopic studies of SO₃, may offer further insights into the role of atmospheric oxygen in the high-T oxidation of sulfur compounds.

Emission of H₂SO₄ from the magma at the detected levels is not predicted within the thermodynamic model presented here (or models of a similar nature) as H₂SO₄ is unstable at typical magmatic temperatures. It therefore seems unlikely that reactions such as (R13) and (R14) could be responsible for sulfate generation [Varekamp et al., 1986; Mather et al., 2003], as H₂SO₄ seems to be a very minor species in the high-T mixture (at low V_A/V_M), and H₂SO₄ generated at lower temperature (or higher V_A/V_M) would be contaminated with atmospheric oxygen.

Mather et al. [2003] highlight this class of reactions as implausible, since unless the reactions went to completion and the acid product was perfectly revolatilized, SO₄²⁻ and Cl⁻ (or F⁻) would closely associated in a single aerosol phase. Instead, in the aerosols from Masaya volcano, Cl⁻ (and F⁻) are found in the coarser (>1 μm) aerosol mode, while Na⁺ and K⁺ appear to be associated with SO₄²⁻ in the finer (<1 μm) aerosol mode.

A further possibility is the direct emission of other sulfate species such as Na₂SO₄ or K₂SO₄. Sulfates emitted from the magma may take the form of simple molecular species in the gas phase, or condensed, crystalline sulfates. Transport of these emitted sulfates to lower temperatures, followed by reaction with strong acids such as HCl (or HF) could generate high levels of sulfate and sulfuric acid, as in (R15). This reaction is chemically reasonable because HCl is a stronger acid than H₂SO₄, and so will protonate SO₄²⁻ ions to give H₂SO₄; H₂SO₄ is a weaker acid and so cannot protonate Cl⁻ ions to give HCl.

However, the absence of any association between Na⁺ and Cl⁻ in the aerosols [Mather et al., 2003] suggests (R15) is a minor process. Preliminary modeling using HSC Chemistry (by introducing Na₂SO₄, Na₂SO₃, NaCl, NaBr, NaF and Na₂O into the reduced HSC model) reveals that molecular and condensed metal sulfates are both liable to thermal decomposition in the high-T mixture (T = 1000°C, 0 < V_A/V_M < 0.1), although condensed metal sulfates appear stable above the compositional discontinuity at temperatures as low as 800°C. Metal sulfites (such as Na₂SO₃) are thermodynamically unstable with respect to the metal sulfates over 0 < V_A/V_M < 0.1 at all temperatures which can reasonably be used with the model (T > 800°C).

Within the constraints imposed by the associations in the aerosol phase, isotopic studies and modeling work, it appears clear that simple reactions such as -–(R15) cannot account for the formation of mixtures of near-source sulfates and sulfuric acid. While no simple mechanism can currently be envisaged, it is reasonable to suggest that condensed metal sulfates are more likely to survive decomposition in the high-T mixture than simple molecular species such as H₂SO₄ because condensed metal sulfates are able to distribute the excess thermal energy into lattice vibrations (eventually leading to phase changes), rather than bond vibrations which lead to decomposition of the sulphate unit. Condensed metal sulfates may be the primary mode of sulfate transport from the magma, while sulfuric acid droplets may form as a result of some secondary process. Métrich et al. [2002] suggest that significant amounts of sulfite may be present in melts, so direct emission of sulfite cannot be excluded; however, oxidation from sulfite to the detected sulfate (S^IV → S^VI) would presumably result in the incorporation of atmospheric oxygen.

4.4. Thermodynamic X_i/SO₂ Ratios

Thermodynamic X_i/SO₂ ratios vary with the composition and temperature of the magmatic gas, along with the V_A/V_M of the mixture. Thermodynamic X_i/SO₂ ratios can be calculated for each species. In Table 4 we show model ratios for the most abundant oxidized trace species of each element considered in this study (i.e., OH, SO₃, NO, Cl, Br and I) for a mixture at 1000°C and V_A/V_M = 0.1. These thermodynamic X_i/SO₂ ratios may provide the necessary inputs for initializing atmospheric reaction models (i.e., kinetic models) to further investigate the chemical evolution of volcanic plumes.