Volcanic eruptions inject sulfur into the atmosphere mainly in form of SO2 and H2S. The ratio of these species (H2S/SO2), which is usually used as the mirror of the oxidation state of the source magma, varies significantly according to the type of activity, tectonic setting etc. This study aims to investigate the role of the hot core of plinian and sub-plinian volcanic plumes (T > 600°C) in sulfur speciation based on the thermodynamic equilibrium in this temperature range. We consider the hot core as a box model in which 1000°C magmatic gas and 25°C ambient air are mixed and show that it functions as a hot oxidizing reactor for S species. Processes inside the hot core usually decrease the H2S content of the system but can either increase or decrease SO2 depending on initial oxidation state. Thus the SO2 injected into the atmosphere is not essentially generated directly from the magma but it can be produced in the hot core as the result of H2S oxidation. Besides, volcanic cloud compositions do not mirror the source conditions. Considering three types of tectonic settings (convergent plate, divergent plate and hot spot) we propose that H2S emission is more likely under reduced conditions in divergent plate and hot spot volcanic settings.
 In addition to high temperature S flux and speciation data, advanced measurement and observation techniques resulted in very good insight into sulfur speciation in volcanic clouds [Hunton et al., 2005; Rose et al., 2006; Clarisse et al., 2011]. Hunton et al.  studied the volcanic cloud of the Hekla 2000 eruption via in-situ aircraft observations. They reported 1 ppmv SO2 concentration in the plume but no H2S was observed. This outcome was in very good agreement with results of Rose et al.  who also detected 1–1.2 ppmv SO2 but no H2S in the 33–34 hour old volcanic cloud of the same eruption. The absence of H2S in the cooled volcanic plume (i.e., H2S/SO2 = 0) suggests that this species has been depleted via in-plume processes [Hunton et al., 2005; Rose et al., 2006]. However, the oxidation of H2S under atmospheric conditions may be of the order of 2 to 5 days [Seinfeld and Pandis, 2006; Rose et al., 2006; Aiuppa et al., 2007a]. In addition, since H2S is only slightly soluble in liquid water, its removal by cloud and raindrops at lower heights is negligible [Textor et al., 2004]. Therefore, quick depletion of H2S in a few hours old volcanic plume cannot be due to either low-temperature oxidation processes or scavenging by the liquid phase and another alternative sink for this species must exist.
 Recently, it has become clear that volcanic vents are not simply passive point sources of gas emission, but that in fact they also act as high-T reactors where mixtures of volcanic and atmospheric gases might react, generating new and previously unexpected reaction products [Gerlach, 2004; Martin et al., 2006, 2009]. Hence, these hot reactors can be considered as a main cause of changes in the sulfur speciation [Getahun et al., 1996; Africano and Bernard, 2000]. Studies over recent decades revealed that in high-temperature (T > 500°C) volcanic gases the reaction rates are sufficiently fast to ensure that the system is very close to equilibrium conditions [Symonds and Reed, 1993; Gerlach, 2004]. Considering this assumption, Martin et al.  studied the high-temperature mixture of volcanic and atmospheric gases and concluded that in lack of air (<6% air) H2S does not re-equilibrate. This is in agreement with the finding of Aiuppa et al. [2007b] who used experimental data coupled with modeling studies on the gas plume of Mount Etna. They reported that the H2S/SO2 ratio shows no systematic change with plume aging and suggested that H2S is kinetically stable during short-range transport and T < 600°C. However, in the case of higher temperatures and also more air entrainment, the sink for the H2S is still unknown.
 This study aims to investigate the effect of high-T gas-air mixing on sulfur speciation and redox state with respect to chemical, physical and thermodynamic properties of the hot core of volcanic plumes. Considering the data for hot gas (T > 600°C) [Symonds et al., 1994; Gerlach, 2004], a box model is considered to identify the processes which can change the S speciation during the first few minutes in the eruption plume. Section 2 describes the modeling concepts, procedures and scenarios. In section 3 the effects of different initial conditions and also various processes in the hot core on S speciation are discussed. In addition, the model outputs are evaluated against an observed volcanic cloud. Finally, conclusions are given in section 4.
2.1. Concepts of Mixing
 Considering the great diversity in the style of volcanic eruptions (i.e., from effusion of basaltic magma as lava flows and fire fountaining in Hawaiian eruptions to the generation of high-altitude plumes in plinian eruptions), there are different mechanisms for high-T mixing of volcanic and atmospheric gases. In general, what control the nature of mixing are temperature and eruption dynamics (i.e., turbulence and air entrainment). If the plume is able to entrain and heat enough air, it can reduce its density below that of the surrounding atmosphere and continue to rise. Otherwise, it would collapse as a low fountain and generate pyroclastic flows and surges [Sparks et al., 1997]. This study focuses on the mixing processes in the hot core of subplinian and plinian eruption plumes which usually have high temperature (T > 600°C) as well as high air entrainment due to high turbulence at the plume boundaries [Sparks et al., 1997]. For simplification, we can assume that the effect of decompression on temperature is negligible [Ogden et al., 2008]. Thus, the main cause of the cooling is ambient air entrainment.
2.2. Modeling Procedure and Scenarios
 To identify the hot core control on the H2S/SO2 ratio, a conceptual box model has been considered (Figure 1) which covers the chemical processes in the temperature range 600° < T < 1000° and thus, can be solved using thermodynamic equilibrium models. Such models have been widely used by several authors in order to simulate the mechanism of hot volcanic gas interaction with wall rock [e.g., Symonds and Reed, 1993; Africano et al., 2003] and also atmospheric gases [e.g., Gerlach, 2004; Martin et al., 2006, 2009]. Here, we use two most cited models SOLVGAS and GASWORKS [Symonds and Reed, 1993] to model the processes in the hot mixture of volcanic and atmospheric gases. SOLVGAS calculates homogeneous equilibrium (distribution of species) in a gas phase while GASWORKS computes heterogeneous equilibria among gases, solids and liquids during processes of cooling, gas-gas mixing, pressure changes and gas-rock reaction [Symonds et al., 1987; Symonds and Reed, 1993]. More details on these programs have been provided elsewhere [see Symonds and Reed, 1993] and thus, their description is not given here. A significant advantage of these models is the ease with which compositional parameters such as air to magmatic gas ratio (VA/VM) can be explored, alongside physical parameters such as temperature and pressure.
 As shown in Figure 1, we need directly sampled uncontaminated high-T volcanic gas data as model input. Such data are generally sparse [Symonds et al., 1994; Rose et al., 2006]. To optimize the model, we use the data of directly sampled high-T volcanic gases from different volcanoes reported by Gerlach  proposing three typical compositions for three different tectonic settings: convergent plate (CP), divergent plate (DP) and hot spot (HS) (Table 1). The scope of this research is identifying the sensitivity of S speciation in various volcanoes to cooling and mixing processes in the hot core of volcanic plumes. Based on the compositions given in Table 1 and in order to check the sensitivity of the model against the oxygen fugacity (fO2), we define some modeling scenarios. These scenarios have been calculated using SOLVGAS by introducing the initial composition (Table 1) as the starting point. Second, the temperature of the system has been adjusted to 1000°C for having the same initial temperature for all scenarios for comparison reasons (see Table 2, main oxidation states). Then, a range for the oxidation state of the system, which is represented by the logarithm of oxygen fugacity (log fO2), is considered by varying log fO2 by ±1 . Thus, 9 high-T volcanic gas compositions are simulated as modeling scenarios (Table 2). The pressure is held constant at 1 bar in all calculations. We note that some of the initial conditions like T and fO2 may seem unusual regarding some volcanic settings. This is done because the conceptual model of this study is not designed to reproduce a specific situation at one volcano but rather to identify the role of different initial conditions and processes in S speciation in different types of volcanoes. The impacts of these assumptions will be discussed in section 4.
Table 1. Average High-T Volcanic Gas Composition in Mole %a
Table 2. Modeling Scenarios Simulated With SOLVGAS in Mole %a
Convergent Plate (CP)
Divergent Plate (DP)
Hot Spot (HS)
Since we consider 1 mole of total gas in the system, mole fractions are similar to mole numbers. However, at the end of simulations, when ambient air has been added to the system, mole number and mole fraction are different.
 In previous studies on high-T mixtures usually one physical parameter (temperature (T) or (VA/VM)) has been considered as variable [Gerlach, 2004; Martin et al., 2006, 2009]. Here, for the first time, we consider mixing and cooling concurrently which means changing T and VA/VM simultaneously. GASWORKS has the capability of mixing two gases with different temperature and composition. At each step, we mix the hot volcanic gas (T = 1000°C) with 25°C air (78% N2, 21% O2, O.1% Ar) by increasing VA/VM which makes the system more oxidized and also cools it. Air is mixed with the magmatic gas until a temperature of 600°C is reached which corresponds approximately to VA/VM = 1.0. At T < 600°C and during the first few hours of the eruption, we assume that the changes in H2S/SO2 as well as HCl/SO2, HCl/HF, H2SO4/SO2 etc are negligible [Aiuppa et al., 2005, 2007b]. Therefore, the predicted ratios in the output of the high-T box model can be compared to the measured ratios in few hours old volcanic clouds [e.g., Hunton et al., 2005; Rose et al., 2006]. To investigate the effect of cooling and mixing on the gas composition separately, we consider two additional procedures for each scenario: cooling without air entrainment from 1000°C to 600°C as well as the mixing of magmatic gas with air at constant temperature (T = 1000°C) from VA/VM = 0 to 1.0.
 We note that gas-ash interaction in the eruption plume leads to volatile scavenging, in particular SO2, HCl and HF [Rose, 1977; Òskarsson, 1980]. Studies showed that during explosive eruptions as much as 33% of the erupted sulfur can be adsorbed on ash as acid aerosol particles and 5% of it can be trapped via gas-ash interaction [Rose, 1977; Ayris and Delmelle, 2012]. This suggests that scavenging is more important after the formation of liquid acid aerosols in volcanic plumes and is therefore thought to be negligible in a first order approximation and the temperature range of this study (600°C−1000°C) where liquid phase formation is very unlikely. However, this effect will be considered during the evaluation of Hekla 2000 eruption in section 3.4.
3.1. Cooling Without Air Entrainment
 The effect of cooling from 1000°C to 600°C on major S species in the volcanic gas (H2S, SO2, H2SO4 and SO3) is shown in Figure 2. For all volcano types it is obvious that during cooling the system favors less oxidized species (e.g., H2S) than oxidized species (e.g., SO2 and SO3). For instance, with falling temperature, H2S increases according to reaction (1) when it proceeds toward the left:
 These cooling trends are controlled by entropy effect, whereby lower temperatures favor fewer moles of gas (left side of reaction (1)). Giggenbach  proposed that the temperature dependence of the equilibrium constant (Keq) of reaction (1) can be expressed as
where f is the fugacity ratio (assumed to be equal to concentrations) and T is temperature in K. Thus, during cooling Keq decreases rapidly and results in a considerable redox ratio increase. Figure 3 displays the H2S/SO2 ratio which increases for all types of volcanoes during cooling. These changes are pronounced for the CP-M case in which the redox ratio becomes approximately 3 times larger. In other words, S redox ratio increases more drastically during cooling in the scenario with the lowest oxygen fugacity. The abundance of reduced S species at lower temperature has been reported by many researchers [e.g., Giggenbach, 1996; Aiuppa et al., 2005, 2006]. Thus in an air free magmatic gas the less the temperature, the more the H2. This effect is notable in CP volcanoes which typically have lower initial temperature (T < 1000°C). They can have higher initial H2S content and thus more potential of H2S injection into the atmosphere.
3.2. Mixing With Air at Constant Temperature
Figure 4 shows the results of the mixing of volcanic gas with air at 1000°C. Since after VA/VM = 0.2, changes in number of moles are negligible, the x axis is limited to this value. Considering the presence of molecular oxygen (O2) in the system, reaction (1), which consumes the atomic oxygen from H2O, does not control the H2S/SO2 ratio anymore. In this case, according to the reaction (3), the complete oxidation H2S to SO2 results in depletion of hydrogen sulphide in the system and also increase in SO2:
which can be further oxidizes to SO3 and H2SO4 according to reactions (4) and (5):
 These reactions become more important as soon as the H2S content of the system becomes negligible. This is a transition point in Figure 4 where H2S drastically decreases and SO3 and H2SO4 significantly increase. It occurs at VA/VM being around 0.049, 0.095 and 0.041 for CP-M, DP-M and HS-M, respectively. For comparison, the changes in S redox ratio are shown in Figure 5 where H2S/SO2 decreases with increasing VA/VM. It is obvious that H2S content becomes extremely small in the system around the above mentioned values for VA/VM. Thus, oxygen is no more consumed by the oxidation of H2S (reaction (3)) and there is more O2 available for (4) and (5) reactions. Hence, reaction (4) and accordingly reaction (5) proceed more rapidly which result in a considerable increase in SO3 and H2SO4 and also a small decrease in SO2. After this transition stage, these species become almost stable and do not change significantly. More details about the transition point are discussed in section 3.3.
 Similar to the cooling (section 3.1), changes in the H2S/SO2 ratio are more pronounced for the CP scenario which has the lowest fO2 (see Table 2). But the direction of changes is completely reversed. In the mixing procedure, the oxidation reaction (3) controls this ratio and considerably decreases it while in the cooling process, where there are no free O2 molecules, entropy controls the H2S/SO2 ratio via reaction (1) and increases it. The changes in SO2 concentration is shown in Figure 6. For example, in DP-M case SO2 reaches its maximum value at VA/VM = 0.12 which is the result of reaction (3). After this point it starts to decrease due to reaction (4) and eventually becomes constant. These trends are similar for other scenarios. For all types of volcanoes, the mixing of magmatic gas with air at 1000°C results in a chain of oxidation reactions (3) to (5) which depletes H2S, increases SO2, SO3 and H2SO4 and thus, decreases the H2S/SO2 ratio. This situation happens very fast in CP and HS volcanoes but takes longer for the DP volcanoes. In other words, more air is needed for depletion of reduced gases in DP volcanic gases.
3.3. Cooling With Air Entrainment (Cooling and Mixing Simultaneously)
 In order to simulate the mixing of hot volcanic gas with ambient air more realistically which cause the dilution, cooling and thus rising of the volcanic plume, the 1000°C magmatic gas has been mixed with 25°C air which results in simultaneous cooling and mixing. The results for major S species are shown in Figure 7. The general trends for the main scenarios are almost similar to those presented in section 3.2 where oxidation of the system at constant T causes a considerable decrease of H2S and increase of SO3 and H2SO4 via a transition point at around 950°C for CP-M and HS-M and around 900°C for DP-M. But in concurrent mixing and cooling, changes in species concentrations continue after the transition stage without reaching a constant concentration; i.e., SO3 and H2SO4 constantly increase at the expense of SO2 decrease. The entropy effect can explain these changes. By decreasing temperature, reactions (4) and (5) proceed toward the right side in order to reduce the moles of gas in the system. If the temperature remains constant (like section 3.2) the entropy effect does not play a significant role. Thus, species attain a constant concentration and would not change significantly with adding more air. By comparing different oxidation states, it is obvious that for all types of volcanoes lower initial fO2 shifts the transition stage to lower temperatures. While in all oxidized scenarios in Figure 7 the transition occurs very early at T > 950°C, in the reduced scenarios it is shifted to 800°C, 620°C and <600°C for CP-R, DP-R and HS-R, respectively. Since after the transition, SO2 keeps decreasing until 600°C, an early transition can result in a huge decrease in SO2 concentration in the final plume composition. Figure 8 shows the changes in SO2 content of the system for different volcanoes and also different oxidation states. There is no significant difference between the main and oxidized scenarios regarding the trends and values. In these cases, 65% to 75% of the initial magmatic SO2 is oxidized. But the reduced cases which have lower SO2 at 1000°C show an increase and also the highest values for this compound at 600°C. These effects can be explained by the reactions (3) to (5). The systems with lower initial oxygen fugacity, which means also higher H2S concentration or higher H2S/SO2 ratio, have more sinks for the entrained O2 and do not allow the reactions (4) and (5) to proceed very fast. While there is enough H2S in the system, its oxidation persistently consumes O2 via reaction (4) which results in production of SO2 and reduction of the H2S/SO2 ratio. An example of changes in S redox state is shown in Figure 9 for the main scenarios. It is obvious that the rate of H2S/SO2 reduction is the highest for the CP case which has the highest initial value and the lowest initial fO2 (see Table 2). This is similar to the trend shown in Figure 5 and addresses the fact that the lower the oxygen fugacity the faster the H2S/SO2 reduction. After depletion of hydrogen sulphide (when H2S/SO2 approaches 0), reaction (4) becomes a significant sink for SO2 and continuously decreases it. Therefore, reduced scenarios can potentially inject more SO2 into the atmosphere. In other words, the injected SO2 into the atmosphere is not essentially generated directly from magma but it can be produced in the hot core as the result of H2S oxidation (reaction (3)).
 As mentioned above, during mixing with ambient air (i.e., in presence of O2), the transition results in depletion of reduced species in the system. Martin et al.  proposed that the main sinks for oxygen in 1000°C magmatic gas are H2S via the reaction (3) and H2 via the reaction (6):
They showed that the location of the transition corresponds exactly to the complete removal of H2 and H2S [Martin et al., 2006]. In other words, the oxidation state of the system is controlled by the ratio of H2 and H2S content of it (i.e., two major sinks for the oxygen) to the amount of entrained oxygen from ambient air (i.e., the major source of the oxygen). Thus, this ratio can be treated as the ratio of the major sinks of the oxygen to its major source in the system and defined as
where n corresponds to mole number of each species and the number 0.21 represents the 21% of the ambient air which is assumed to be oxygen. This ratio is calculated for different volcanoes and shown in Figure 10. This figure shows unambiguously that the critical value for log(Xmix), after which the transition occurs, is between −2.50 and −3.00 for all types of volcanoes. Volcanic plume rises in the atmosphere as a result of air entrainment. Mixing of the magmatic gas with ambient air leads to addition of O2 into the system as well as cooling which, according to the entropy effect, force reactions (3) and (6) toward the right side which means consumption of H2 and H2S. Therefore, during mixing of volcanic gas with air, Xmix decreases continuously until reaching the critical value which means depletion of reduced species in the system and transition into the oxidized state.
3.4. Model Evaluation Against Observed Volcanic Clouds
 In order to evaluate the findings of this research, we compare the results to observations. Here, we consider the data of the volcanic cloud form the Hekla 2000 eruption which has been reported by Rose et al.  and Hunton et al.  (Table 3). Various studies have been conducted on this eruption [e.g., Rose et al., 2003; Moune et al., 2006, 2007, 2009; Höskuldsson et al., 2007] resulting in a very good knowledge about its petrologic and volcanologic properties as well as atmospheric transport and dispersion. This eruption is also unique regarding the direct sampling from the 33–34 hours old volcanic cloud [Rose et al., 2006]. If we assume that at temperatures less than 600°C and during the first few hours of the eruption the S speciation in the gas phase remains constant then the chemical composition in the output of the hot core (the box model in this study) could be compared to the few hours old volcanic cloud composition. It is notable that the condensation of H2SO4 can significantly affect its concentration in the gas phase even in a few hours old plume. Therefore, we omit this species in our evaluation study.
Table 3. Hekla Plume Observations on 28 February 2000a
 Remote sensing studies propose that the Hekla 2000 eruption has injected approximately 1.6 Tg volcanic gas into the atmosphere [Rose et al., 2003; Hunton et al., 2005; Rose et al., 2006]. From this value 0.2 Tg is SO2 [Rose et al., 2003] which corresponds to weight percentage of 12.5. In other words, to inject 0.2 Tg SO2 into the atmosphere, the eruption plume of Hekla should contain 12.5 wt% SO2 near the vent. The conversion of this value to mole fraction of SO2 in the plume can be done using the molar weight of volcanic gas. As Hekla is located on a mid-ocean ridge but its geochemistry also suggests a hot spot signature [Schmincke, 2004] its magmatic gas composition can be compared to both DP and HS scenarios. Petrologic studies on the Hekla magam suggest an oxygen fugacity close to FMQ (Fayalite, Magnetite, and Quartz) at 1000°C [Baldridge et al., 1973; Moune et al., 2007] which corresponds to logfO2 of −10.50 to −11.00 [Lindsley, 1991]. Although in petrology this rage is known as reduced conditions, due to our definition of the oxidation states it happens to fall into the range of the main oxidation state scenarios ( −10.89 for DP-M and −10.39 for HS-M). Therefore DP-M and HS-M scenarios are selected for the evaluation study. The gas composition in these scenarios has the molar weight of approximately 24–28 g. According to the above calculations, 12.5 wt% of this value is SO2 which corresponds to about 4.6–5.5 mole% of SO2 in the volcanic cloud. These values are satisfactorily in the range of our predictions for DP-M and HS-M with 2.7% and 6.4%, respectively (Figure 8). These scenarios also correspond to the value of 0 for H2S concentration at 600°C which is also in agreement with the observations. If we assume that the hot spot setting is dominant in Hekla, HS-M scenario has an acceptable agreement with the observations. The overestimation of this scenario can be due to neglecting either scavenging of sulfur species by volcanic ash or the effect of other processes in the warm region of the volcanic plume (T < 600°C). As it was snowing during the eruption [Höskuldsson et al., 2007], snow particles could have also washed gases out of the plume.
 It worth to mention that petrologic estimates suggest an injection of 0.6–3.8 Tg SO2 into the atmosphere during Hekla 2000 eruption [Moune et al., 2007] which when compared to the remote sensing measurement of 0.2 Tg SO2 release [Rose et al., 2003; Hunton et al., 2005; Rose et al., 2006] is surprising. This becomes less surprising when one considers the box model of this study and the effect of the hot core on S speciation. The initial SO2 content of the HS-M scenario (19.16 mole% at 1000°C) reduces by the factor of 3 once it passes through the box model (6.4% at 600°C) as the result of the hot core oxidation processes (see Figure 8). If we omit the hot core effect, then SO2 injection, based on the initial SO2 content of HS-M, is about 0.6 Tg which is close to the lower bound of the petrologic estimate. On the other hand, the above calculations show a good agreement between observed volcanic cloud and the predictions based on hot core box model. Therefore, the hot core effect on S speciation can explain the differences between petrologic estimate of S degassing and remote sensing observations of SO2 in the volcanic cloud.
4. Discussion and Conclusion
 The study of the high-T mixing of magmatic and atmospheric gases by considering a conceptual box model for the temperature range 600°C < T < 1000°C confirms the hypothesis that the hot core of volcanic plumes can functionas a hot oxidizing reactor for S species. Burgisser and Scaillet  showed that the redox state that magma records at depth does not necessarily mirror that of its escaping gases. In addition, our results show that the redox state in volcanic clouds dose not essentially reflect the magmatic gas composition at the vent. Therefore, volcanic emissions undergo significant compositional changes from the release point (i.e., magma) to the atmosphere (i.e., volcanic cloud). These changes are controlled by the effect of various processes such as cooling and mixing with the ambient air. Cooling of the magmatic gas without air entrainment favors the reduced forms of S (e.g., H2S) while mixing with air supplies molecular oxygen to the system making oxidized species dominant. During mixing with air as soon as two main reduced species reach a minimum value (H2S and H2 via (3) and (6) respectively), the transition into a completely oxidized system occurs. Aiuppa et al.  proposed the equations (8) and (9) for the thermodynamic equilibrium of reactions (3) and (6):
 They finally obtained, using relations (8) and (9), the couples of equilibrium H2/H2O and SO2/H2S gas ratios, at each temperature (at 0.1 MPa pressure) (Figure 11) [Aiuppa et al., 2011]. Schematic examples of the considered procedures in this study are shown in this figure. During cooling without adding air (blue arrow) by passing the isotherms the reduced species become favorable. The red and green arrows show the general pattern of the mixing at constant T and mixing with 25°C ambient air (i.e., simultaneous mixing and cooling), respectively. The results of this study show that when the relation between SO2/H2S and H2/H2O (x and y axis, respectively in Figure 11) reaches approximately 105, the transition to the complete oxidized system occurs. This is shown by the solid black line in Figure 11 which has the slope of 105. The fate of the oxidation state of the system and its sensitivity to initial conditions (e.g., T and fO2) can be explained using this figure. When the eruption begins the plume composition moves along the green arrow. The more the air entrainment rate, the less the slope of the line. In other words, in plinian and sub-plinian eruptions where air entrainment is the main cause of cooling and oxidation the green line tends closer to the red line and can pass the transition line quickly. On the other hand, in other eruptions where other causes may be important in cooling the green line inclines to be close to the blue line and even after reaching 600°C can remain on the left side of the transition line. Changing the initial temperature moves the starting point along the blue arrow. As shown in section 3.3, the transition happens very fast in CP volcanoes which depletes H2S and reduces SO2 concentration. But Figure 11 demonstrates that the lower initial temperature (as in CPs) makes the injection of H2S more likely. This effect can explain the detection of H2S in volcanic clouds of some CP volcanoes [Clarisse et al., 2011].
 With similar initial temperature, CP volcanoes inject S most predominantly in oxidized form and injection of H2S from such volcanoes seems to be unlikely. For DP and HS volcanoes however, even though the oxidized species are dominant too, H2S injection into the atmosphere could be expected. But it strongly depends on the oxidation state of the magma. Changing the initial fO2 means moving of the start point along 1000°C isotherm. When the end of the green arrow (at 600°C) is very far from the transition line, not only H2S is depleted but also SO2 concentration is decreased. It is obvious that the lower the initial fO2, the higher the final SO2. While the SO2 content of the system decreases in the main and oxidized scenarios, it increases in the reduced scenarios. Consequently, the reduced magmas can potentially inject more SO2 into the atmosphere even though they carry initially less SO2 in the magmatic gas. In other words, the injected SO2 into the atmosphere is not necessarily generated directly from magma but it can be produced in the hot core as the result of H2S oxidation. In such conditions, the injection of small amounts of H2S is also possible (especially in DP and HS). This may be an explanation for the known conflict between petrologic estimates and remote sensing observations regarding SO2 emissions of large explosive eruptions. Conventional petrologic methods based on glass inclusions usually underestimate the SO2 content of volcanic emissions [Gerlach et al., 1996]. However, depending on initial oxidation state, the hot core can function as a reactor for SO2 formation and convert an initially SO2-poor vapor (i.e., petrologic estimate) to a SO2-rich plume (i.e., remote sensing observation).In the Hekla 2000 eruption the conflict is the other way round; i.e petrologic methods overestimate the S injection. This can also be explained basedon the hot core effect as the hot core reduces the SO2 in this case (see section 3.4). Oxidized magmas can inject sulfur into the atmosphere mainly in form of H2SO4 and SO3 which are the products of SO2 oxidation in the hot core of the plume. In these magmas, injection of H2S into the atmosphere could be unlikely.
 Several lines of evidence confirm the fact that the hot core of volcanic plume significantly controls the S speciation in volcanic emissions and makes oxidized species like SO2 and H2SO4 more dominant in the system. Therefore the source parameters in volcanic cloud studies should be assigned considering the role of the hot core as an oxidizing reactor. This means that the implications of volcanic sulfur emissions in climate and atmospheric chemistry could be enhanced if one considers the effect of the hot core on S speciation in major explosive eruptions.
 We thank P. Delmelle and T. Mather for the in depth discussion of potential process in the hot core. Thanks also to M. H. Reed for providing us the source codes of SOLVGAS and GASWORKS as well as helpful comments. This work is supported through the Cluster of Excellence CliSAP (EXC177), University of Hamburg.