Vol48/04/2005def 739 ANNALS OF GEOPHYSICS, VOL. 48, N. 4/5, August/October 2005 Key words magmatic gas – magma degassing – hy- drothermal system – crater lake – meteoric water – scrubbing – reaction path modeling 1. Introduction Gases are released from magmas both during eruptions (syn-eruptive or active degassing) and in quiescent periods that either may preceed an eruption (pre-eruptive degassing) or may not (passive degassing). Studies of pre-eruptive de- gassing are of uppermost interest since we hope that we are able to interpret the changes in the composition and/or fluxes of volatiles as signals of the evolution of the magmatic system towards an eruptive episode. However, this is not a sim- ple exercise due to the common occurrence of interfering processes, mainly magmatic gas scrubbing by waters and interaction of these aqueous solutions with rocks. The effects of wa- ter-rock interaction are enhanced by the high re- activity of most magmatic gases and are similar to what takes place in a Giggenbach’s flask dur- ing fumarolic gas sampling. In particular, CO2(g), H2S(g), HCl(g), and HF(g) are readily neutralised (1.1) (1.2) (1.3) (1.4) whereas the behavior of SO2 is unique, in that it may be disproportionate to SO42− and HS−, as indicated by the reaction (1.5) SO OH SO HS H O4 7 3 3( ) ( ) ( ) ( )laq aq aq2 4 2 2( )g + = + + - - - HF OH F H O( ) ( ) ( )laq aq 2( )g + = + - - HCl OH Cl H O( ) ( ) ( )laq aq 2( )g + = + - - H S OH HS H O( ) ( ) ( )laq aq2 2( )g + = + - - CO OH HCO( ) ( )aq aq2 3( )g + = - - Geochemical modeling of magmatic gas scrubbing Luigi Marini and Barbara Gambardella Laboratorio di Geochimica, Dipartimento per lo Studio del Territorio e delle sue Risorse (DipTeRis), Università degli Studi di Genova, Italy Abstract The EQ3/6 software package, version 7.2 was successfully used to model scrubbing of magmatic gas by pure water at 0.1 MPa, in the liquid and liquid-plus-gas regions. Some post-calculations were necessary to account for gas separation effects. In these post-calculations, redox potential was considered to be fixed by precipitation of crystalline α-sulfur, a ubiquitous and precocious process. As geochemical modeling is constrained by conser- vation of enthalpy upon water-gas mixing, the enthalpies of the gas species of interest were reviewed, adopting as reference state the liquid phase at the triple point. Our results confirm that significant emissions of highly acidic gas species (SO2(g), HCl(g), and HF(g)) are prevented by scrubbing, until dry conditions are established, at least locally. Nevertheless important outgassing of HCl(g) can take place from acid, HCl-rich brines. Moreover, these findings support the rule of thumb which is generally used to distinguish SO2-, HCl-, and HF-bearing mag- matic gases from SO2-, HCl-, and HF-free hydrothermal gases. Mailing address: Prof. Luigi Marini, Laboratorio di Geochimica, Dipartimento per lo Studio del Territorio e delle sue Risorse (DipTeRis), Università degli Studi di Ge- nova, Corso Europa 26, 16132 Genova, Italy; e-mail: lma- rini@dipteris.unige.it 740 Luigi Marini and Barbara Gambardella but it can also produce many other aqueous S- bearing species and elemental sulfur (e.g., Giggenbach, 1987, 1997). In natural settings, scrubbing of magmatic gases is carried out by 1) the liquids hosted in the hydrothermal systems frequently interposed be- tween the magma batch stationing at depth and the surface (e.g., Reed, 1997; Symonds et al., 2001; Marini et al., 2003a); 2) the lacustrine waters stored in crater lakes (e.g., Rowe et al., 1992; Delmelle and Bernard, 2000; Varekamp et al., 2001; Marini et al., 2003b), and 3) the shallow meteoric waters soaking the volcanic edifice (e.g., Symonds et al., 2001). Interaction of magmatic gases with seawater can also oc- cur, as possibly in the recent crisis of Panarea, Aeolian Islands, Italy (Chiodini et al., 2002). All these processes can be satisfactorily mod- eled by means of reaction path modeling, a pow- erful geochemical tool originally proposed by Helgeson (1968), who worked out the thermody- namic relationships needed to evaluate the irre- versible gas-water-rock exchanges. The first computer code for reaction path modeling, PATHI, was developed by Helgeson and cowork- ers, who presented several applications, including hydrothermal rock alteration and ore deposition, in a series of ground breaking scientific papers (e.g., Helgeson et al., 1969). The theoretical evolution of water composi- tion and the changes in secondary mineral as- semblages during progressive neutralization of an initially acidic aqueous solution originated by absorption of magmatic gases in meteoric water was modeled by Reed (1997), who recog- nised the appearance/disappearance of a series of intermediate pH buffers until the final, ther- modynamically stable mineral paragenesis at- tains equilibrium with the aqueous solution. Al- though this theoretical model is in line with our present understanding of most hydrothermal systems with close magmatic association (Gig- genbach, 1997), it cannot be applied through- out. For instance, this model fails to explain the origin of the deep, acidic Na-Cl-SO4 waters of the Miravalles geothermal system (Costa Rica), which are probably generated through interac- tion of deep, neutral Na-Cl waters with rocks depleted in alkali and alkali-earth metals and enriched in SiO2, Al2O3, and sulfate minerals, representing old solfataras buried by recent vol- canics, as suggested by Bob Fournier in an in- formal communication about 10 years ago. Re- action path modeling supports the Fournier’s hypothesis (Marini et al., 2003a). Gas-water-rock interactions in crater lake waters were modeled by Marini et al. (2003b) to investigate the reasons for the bimodal pH distribution in crater lake waters, which exhib- it an acidic mode at pH 0.5-1.5 and a near neu- tral mode at pH 6-6.5 (Varekamp et al., 2000). The neutralization of acid SO4-Cl waters with andesite at both low temperature and high tem- perature, followed by cooling below 100°C, rarely produce aqueous solutions with pH in the 3.5-5 range, in line with available analyti- cal data. Reaction path modeling has recently been used to model magmatic gas scrubbing taking place through mixing of magmatic gas with air- saturated water, reaction of a magmatic gas/air- saturated water mixture with rock, and other processes in an excellent paper by Symonds et al. (2001). It turns out that scrubbing by water pre- vents significant SO2(g) and most HCl(g) emis- sions until dry pathways are established, apart from moderate HCl(g) degassing from acidic hy- drothermal waters (pH < 0.5). Degassing of SO2(g) from long-resident boiling hydrothermal systems is also prevented by magmatic gas scrubbing. This paper investigates the irreversible gas- water mass exchanges taking place during addi- tion of magmatic gas to pure water at 0.1 MPa, chiefly by means of the EQ3/6 software pack- age, version 7.2 (Wolery, 1992; Wolery and Daveler, 1992). The purpose of this exercise is to test the EQ3/6 software package on a rather complex problem, which is similar to case 1 of Symonds et al. (2001). They modeled the addi- tion of magmatic gas to air-saturated water in- stead of pure water as we intend to do. Hence, our results are not fully comparable with those of Symonds et al. (2001), as we neglect dis- solved atmospheric gases (i.e., O2(aq) and N2(aq)) and, consequently, the derived N species (main- ly NH4+(aq)) in our computations. It must be stressed that the relatively high concentrations of NH4+(aq) predicted by Symonds et al. (2001) are based on the assumption of instantaneous 741 Geochemical modeling of magmatic gas scrubbing equilibration, which is rarely attained between N species in abiotic systems. On the other hand, the absence of dissolved O2 in our model deter- mines more reducing conditions, with respect to the model of Symonds et al. (2001), in the initial steps of gas-water interactions (ξ <0.4 mol) and, consequently, some differences in the concentrations of S species. 2. Setting up the model The model consists in the progressive addi- tion of hot magmatic gas into 1000 g of pure water initially at 25°C. Pressure is maintained at 0.1 MPa throughout the process. The magmatic gas is represented by sample 79-2 from Merapi Volcano, Indonesia (Le Guern et al., 1982), with a discharge temperature of 810°C and an equi- librium temperature of 915°C (Symonds et al., 1994). This is the same magmatic gas consid- ered by Symonds et al. (2001) and its chemical composition is given in table I. The overall process under investigation de- termines first the gradual heating of the aqueous solution from 25° to 100°C, second its gradual isothermal evaporation at 100°C, and third gas- gas mixing at increasing temperature. The first and second stages of the process, taking place in the pure liquid and liquid-plus-gas regions, can be modeled by means of the EQ3/6 software package, provided that some suitable pre-com- putations and post-calculations are carried out (see below). Upon total evaporation of the aque- ous solution, the subsequent mixing of the gas phase thus produced with magmatic gas cannot be modeled by means of EQ3/6. However, sim- ple mass-balance and equilibrium computations are sufficient to describe this gas-gas mixing process if the number of species considered is reasonably low. Reaction path modeling in the pure liquid and liquid-plus-gas regions was performed in reac- tion path mode, through a purely stoichiometric approach, assuming immediate attainment of chemical equilibrium. The thermodynamic data- base COM was used. This is the most complete database available in the EQ3/6 software package and includes the thermodynamic properties of several solids, aqueous species, and gases, most- ly derived from SUPCRT92 (Johnson et al., 1992). In EQ3/6 runs, activity coefficients of ionic solutes were computed by means of the B-dot equation of Helgeson (1969). Table I. Chemical composition (molar fractions) of the magmatic gas and of the gas produced through total va- porization of the liquid phase. Sample 79-2 from Merapi Volcano, Indonesia (Le Guern et al., 1982) is consid- ered to be representative of the magmatic gas by Symonds et al. (2001) and in this study as well. Gas species Magmatic Gas produced through total gas vaporization of the liquid H2O 0.8887 0.8532 CO2 0.0707 0.09026 H2 0.0154 1.49E-08 SO2 0.0115 4.60E-06 H2S 0.0112 0.00325 HCl 0.0059 0.0528 CO 0.0016 4.90E-13 S2 0.0008 8.81E-10 HF 0.0004 0.00055 Discharge temperature 810°C - Equilibrium temperature 915°C 100°C Molecular weight 20.59 21.39 742 Luigi Marini and Barbara Gambardella 2.1. The enthalpy of liquid water and gas species Modeling is also based on the assumption of conservation of enthalpy upon water-gas mix- ing. The enthalpies tabulated in SUPCRT92 are apparent standard molal enthalpies of forma- tion from the elements, as defined by Benson (1968), and cannot be used for these computa- tions. In steam tables a different reference state is chosen, either liquid water at 0°C (Denbigh, 1971) or liquid water at the triple point, i.e., 273.16 K, 0.006113 MPa. The latter reference state, with SH2O(l), triple=0, GH2O(l), triple=0, was adopt- ed by the 5th International Conference on the Properties of Steam (Helgeson and Kirkham, 1974). For our purpose, the same convention must be adopted not only for H2O, but also for all the gas species of interest. Therefore, for all gas species, we obtained the enthalpy value at the triple point consistent with this convention (see table II for details). The enthalpy of the i- th gas species at any temperature T and any pressure P (0.1 MPa in our case) can be com- puted by means of the relation (e.g., Denbigh, 1971) Table II. Triple point temperature, enthalpy at the triple point, and coefficients of the empirical power series in the temperature, which were used in this work to fit the molar heat capacity of relevant gas species and liq- uid water (CP data from Chase, 1998). Species Ttriple Htriple a b c d e f g K Jmol−1 Jmol−1K−1 Jmol−1K−2 Jmol−1K−3 Jmol−1K−4 Jmol−1K−5 Jmol−1K−6 Jmol−1K−7 H2O(l) 273.16 (a) 0.00 (a) 170.293 −9.821E-01 3.833E-03 −6.808E-06 4.755E-09 H2O(g) 273.16 (a) 45063 (a) 33.791 −7.490E-03 2.777E-05 −1.582E-08 3.049E-12 H2(g) 13.95 (b) 831.8 (b) 14.668 1.127E-01 −3.400E-04 5.106E-07 −4.052E-10 1.639E-13 −2.665E-17 CH4(g) 90.69 (c) 7555 (b) 35.277 −4.261E-02 1.946E-04 −1.535E-07 3.854E-11 CO(g) 68.13 (c) 5684 (b) 29.803 −8.247E-03 2.382E-05 −1.547E-08 3.243E-12 CO2(g) 216.58 (c) 19346 (b) 23.068 5.714E-02 −3.260E-05 6.791E-09 S2(g) 387.15 (d) 34100 (f) 26.747 2.493E-02 −2.029E-05 6.008E-09 H2S(g) 187.6 (c) 16500 (f) 33.765 −1.031E-02 4.991E-05 −3.573E-08 8.182E-12 SO2(g) 197.64 (b) 17400 (f) 28.044 4.950E-02 −2.869E-05 5.705E-09 HCl(g) 159.0 (c) 14000 (f) 28.953 2.914E-03 −1.563E-05 3.215E-08 −2.162E-11 4.849E-15 HF(g) 151 (e) 13300 (f) 29.059 1.046E-03 −4.412E-06 6.534E-09 −2.073E-12 N2(g) 63.14 (b) 2011 (b) 28.757 6.667E-03 −4.163E-05 1.066E-07 −1.103E-10 5.186E-14 −9.259E-18 NH3(g) 195.4 (c) 25055 (b) 33.142 −9.534E-03 8.160E-05 −6.472E-08 1.627E-11 (a) After Keenan et al. (1978): the reported Htriple value of 0.01 J/g for liquid water was set to 0 and the Htriple val- ue of steam (2501.3 J/g) was changed accordingly. (b) NIST Chemistry WebBook (http://webbook.nist.gov/). (c) Lide (2002). (d) Castellan (1971). (e) Extrapolated value based on Ttriple of HCl, HBr, and HI (from reference (c)) and atomic number of halogens. (f) Value computed by means of the equation ∆Hvap = 21 × 4.184 × Tvap, which is based on the Trouton’s rule; this can be applied to non-associated liquids at Tvap > 150 K (e.g., Castellan, 1971). 743 Geochemical modeling of magmatic gas scrubbing (2.1) where CP is the molar heat capacity, V is the molar volume, and α is the coefficient of ther- mal expansion of the i-th gas species. As usual- ly done in the literature (e.g., Denbigh, 1971, and references therein), the heat capacity of gas species was expressed as empirical power se- ries in the temperature (table II) (2.2) fitting the CP data reported by Chase (1998). Consequently, the first integral of eq. (2.1) was computed as follows: (2.3) ( ) ( ) ( ) ( ) C dT a T T b T T c T T d T T 2 3 4 P T T triple triple triple triple triple f = - + - + + - + - +4 4 2 2 3 3 # C a bT cT dTP 3 f= + + + +2 T- a( )H H C dT V dP1T P T P T P triple triple triple = + +# # whereas the second integral of eq. (2.1) was neglected assuming ideal gas behavior. In fact, for ideal gases V = RT/P, and (2.4) It follows that: 1−αT=0 and . This assumption is reasonable in the case under consideration, due to the comparatively small pressure increments from Ptriple to 0.1 MPa pres- sure. Computed enthalpies for H2O(l), H2O(g), H2(g), CO2(g), CO(g), CH4(g), N2(g), and NH3(g) are in excellent agreement with those reported by the NIST Chemistry WebBook (http://web- book.nist.gov/). Enthalpies of H2O(l), H2O(g), H2(g), CO2(g), CO(g), CH4(g), N2(g), NH3(g), S2(g), SO2(g), H2S(g), HCl(g), and HF(g) at 0.1 MPa pres- sure and variable temperature are shown in fig. 1. This plot shows that, at magmatic temperatures (let us say > 800 K): a) the enthalpy of any gas is significantly higher than that of liquid water; and b) H2O(g) has the highest enthalpy and is general- ly followed by CH4(g), NH3(g), CO2(g), SO2(g), S2(g), and H2S(g), whereas HCl(g), HF(g), CO(g), N2(g), and H2(g) have much lower enthalpy values. From a) it follows that the temperature of liquid water is expected to increase upon input of magmatic gases of any composition. However, considering that the main constituent of magmatic gases is by far H2O(g), followed by CO2(g) and SO2(g) (e.g., Symonds et al., 1994), the enthalpy of magmatic gases is not expected to diverge substantially from that of pure steam. 2.2. Isobaric heating of the aqueous solution from 25° to 100°C Conservation of enthalpy upon water-gas mixing is conveniently expressed by the fol- lowing equation: (2.5)H n H XH MW 1000 , , , j j m Tm w Tw w g j Tg= + / ( )V T dP1 0 P triple P - =a# . V T V VP R T 1 1 P2 2 / = =a b l Fig. 1. Enthalpies of H2O(l), H2O(g), H2(g), CO2(g), CO(g), CH4(g), N2(g), NH3(g), S2(g), SO2(g), H2S(g), HCl(g), and HF(g) at 0.1 MPa pressure and variable temperature. 744 Luigi Marini and Barbara Gambardella where indices w, g, and m indicate the initial pure water, the magmatic gas, and the aqueous solution formed upon addition of ng moles of magmatic gas to 1000 g of pure water, MWw is the molecu- lar weight of water, and Xj is the molar fraction of the j-th gas species in the magmatic gas. Both the enthalpies of solute species and the enthalpy of mixing are neglected. The temperature of the aqueous solution was derived from the enthalpy computed by eq. (2.5), based on the first relation- ship in table II solved with respect to T, i.e. (2.6) Then, temperature was related to the reaction progress variable ξ, i.e., the moles of magmatic gas added to 1000 g of pure water, by means of a second-degree polynomial equation. This is (2.7) for the considered example and was used to constrain reaction path modeling by EQ6 in the temperature range 25-99°C. It must be noted that, in the system under consideration, the aqueous solution attains sat- uration with respect to crystalline α-sulfur up- on addition of 0.022 mol of magmatic gas. This fact has to be taken into account in post-calcu- lations (see below). For ξ < 0.4 mol, the total fluid pressure, ob- tained by summing the partial pressures of gas species (mainly H2O, CO2, and H2S) given by EQ6, is lower than the external pressure of 0.1 MPa and results of EQ6 are fully representa- tive. For ξ ≥ 0.4 mol, total fluid pressure ex- ceeds 0.1 MPa, indicating that the aqueous so- lution experiences gas exsolution (degassing) until total fluid pressure attains 0.1 MPa. Sin- gle-step degassing was modeled starting from the compositions of the aqueous phase comput- ed by EQ6, referring to the following mass bal- ance: (2.8) and assuming equilibrium gas-liquid partition- ing of gas species, i.e. f-( ) n n n n f n n 1 j o j g j lH O H O H O2 2 2 = +c c cm m m =( ) . . .CT 25 0 18 696 0 28003 2+ -p pc .. H0033401-. H13 282+.0 07-= 2( )CT , ,m Tm m Tmc (2.9) where the subscripts o, g, and l indicate the ini- tial undegassed liquid, the separated gas and the degassed liquid phase, respectively, and f is the fraction of separated gas. Only CO2(g), H2S(g), and H2O(g) are present in significant contents in the separated gas phase. Water partial pressure was assumed to be fixed by temperature. The partition coefficients Bj of CO2 and H2S were computed from gas solubility data (Henry’s constants) reported in the COM thermodynam- ic database of EQ3/6, KH, j, and PH2O, by means of the equation (2.10) To compute degassing effects, f was tentatively varied until total fluid pressure equalled 0.1 MPa. At this point it was necessary to recom- pute the composition of the degassed liquid phase, a fact that is complicated by the change in redox conditions caused by H2S loss. How- ever, concurrent precipitation of crystalline α- sulfur (see above) constrains fO2, as expressed by the reaction (2.11) Oxygen fugacity was, therefore, computed based on the thermodynamic constant of reac- tion (2.11) and the activity of H2S in the de- gassed liquid phase, which is assumed to be equal to its molality. This is a reasonable as- sumption for neutral aqueous species. The com- position of the degassed liquid was then recom- puted by means of EQ3NR, introducing tem- perature, f O 2, f CO 2, f H 2S, and total Cl and F mo- lalities, as input data, and forcing pH by the electroneutrality condition. The differences be- tween the pH values thus computed and those of the initial undegassed liquid are in all cases except one less than 0.0004 pH units, probably due to rounding off effects. This virtual pH con- stancy upon gas loss is not surprising as the aqueous phase is a relatively dilute solution of strong acids (see below). O .S H 1/2O H S( ) ( ) ( ) ( )c gl2 2 2 aq+ = + .B P K ,H j j H O2 = B n n n n H O H O j j l j g 2 2 = c c m m 745 Geochemical modeling of magmatic gas scrubbing 2.3. Isobaric-isothermal evaporation at 100°C During this second stage of the overall process, temperature is kept constant at 100°C by H2O(l)-H2O(g) coexistence. Initially, progres- sive addition of magmatic gas to water was mod- eled by means of EQ6 neglecting the effects of gas separation. Knowing the enthalpy of the aqueous solution (computed by eq. (2.5)), the fraction of separated steam, y = nH2O(g)/(nH2O(g)+ + nH2O(l)), was then calculated assuming isoen- thalpic boiling, i.e. (2.12) and equilibrium gas-liquid partitioning of CO2 (eq. (2.9)). Subscripts o, l, and g refer to the un- boiled aqueous solution, the separated (boiled) liquid phase and the separated gas phase. Since H2O and CO2 represent more than 98% (on mo- lar basis) of the gas species present in the sepa- rated gas phase, other gas species were neglect- ed in eq. (2.12). Obtained y values were used to distribute the moles of each chemical species computed by EQ6 between the separated liquid and gas phases by means of simple mass bal- ances (equivalent to eq. (2.8) with y instead of f), assuming equilibrium partitioning for gas species (eq. (2.9)). Again, partition coefficients were computed from the gas solubility data re- ported in the COM thermodynamic database of EQ3/6 by means of eq. (2.10) and fO2 was cal- culated referring to reaction (2.11). Then the composition of the boiled liquid was recomput- ed by means of EQ3NR as for the first stage of the process. In this case, the differences be- tween computed pH values and those of the un- boiled aqueous solution range from 0.005 to 0.09 pH units, since the change in pH upon boiling cannot be properly accounted for by a simple mass balance like eq. (2.8). 2.4. Isobaric gas-gas mixing at 100-915°C For this stage of the process, first we com- puted the specific enthalpy of the gas mixture Hgm, referring to water only, i.e., by the equa- tion H n H n H n$ $ $+ +=Ho H O H O H O H O CO CO( ) ( ) ( ) ( ) ( ) ( )l l g g g g2 2 2 2 2 2 (2.13) in which subscript TS is the saturation tempera- ture (100°C) at the considered pressure (0.1 MPa), ng,ev indicates the moles of magmatic gas consumed to evaporate the aqueous solution completely, and ng are the moles of magmatic gas added to the system after complete vapor- ization of the liquid. The first term of the nu- merator refers to the enthalpy of the steam pro- duced through total vaporization of the aqueous solution and the first term of the denominator identifies the corresponding moles of steam. The second term of the numerator represents the enthalpy of the magmatic steam added to the system after complete vaporization of the liquid and the second term of the denominator identifies the corresponding moles of magmat- ic steam. The temperature of the gas mixture was derived from the enthalpy computed by eq. (2.13), based on the HH2O-T relationship (table II) solved with respect to T, i.e. (2.14) The molar fraction of magmatic steam in the gas mixture, xs, was readily computed as follows: (2.15) and used to calculate the concentrations of ma- jor gas species (H2O(g), CO2(g), SO2(g), H2S(g), HCl(g), and HF(g)) in the gas mixture by means of simple mass balances, based on the compo- sitions of the magmatic gas (table I) and of the gas produced through total vaporization of the aqueous solution. The concentrations of minor gas species (H2(g), CO(g), S2(g)) in the gas mix- ture were then calculated referring to the fol- lowing reactions: x MW n X n X n X 1000 , ev H O H O H O s w g g g 2 2 2 $ $ $ = + +b l . H0 1088 $ +-. H38 84 $+.1529 4-=( ) . . CT H6 975 10 m m m 2 5$ $+ - 3 c H H n X n X n X H n X MW MW 1000 1000 , , , , ev ev H O H O H O gm T w g w g g T g H O H O H O ( ) ( ) 2 2 g S g g 2 2 2 2 $ $ $ $ $ $ = + + + + + b b l l 746 Fig. 2. Changes in temperature during the progres- sive addition of hot (915°C) magmatic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simulation. (3.1) and those constrained by the H2S-SO2 magmat- ic redox buffer (Giggenbach, 1987) (3.2) The latter buffer is closely approached upon dry-up of the system, when the SO2(g) concen- tration in the gas phase equals that of H2S(g) (see below). The evolution of the RH parameter against ξ and temperature is shown in figs. 3b and 4b, re- spectively. This parameter is defined as RH= (Giggenbach, 1987) and, therefore, it is directly comparable with the analytical data of fumaroles. 3.3. Sulfur precipitation As anticipated, the aqueous solution becomes saturated with respect to crystalline α-sulfur up- on addition of 0.022 mol of magmatic gas. In the log (moles of S) - logξ plot (fig. 5), the moles of S precipitated experience a very sharp initial in- /log f /Xf log X,= H HH O H O2 2 2 2 8 . .log logf T f2 657 3 2 O H O2 2= - + 2511 .log f T 10 736 25414 O2 = - Luigi Marini and Barbara Gambardella (2.16) (2.17) (2.18) whose thermodynamic constants were obtained by means of SUPCRT92 (Johnson et al., 1992). 3. Addition of magmatic gas to pure water at 0.1 MPa: modeling results Apart from minor differences arising from the choice of modeling magmatic gas scrubbing by pure water rather than by air-saturated water, our results are very similar to those of Symonds et al. (2001). 3.1. Temperature The plot of temperature versus the reaction progress variable ξ (fig. 2) shows a compara- tively smooth increase from the initial tempera- ture of liquid water, 25°C, to the saturation tem- perature at 0.1 MPa pressure (100°C). Howev- er, the transition from the single liquid phase re- gion to the biphase liquid-plus-gas region oc- curs at ξ of 0.4 mol (see above), corresponding to a temperature of 32.4°C only. The tempera- ture of 100°C is attained through addition of 4.29 mol of magmatic gas and it is kept con- stant up to ξ of 79.3 moles by coexistence of H2O(l) and H2O(g). Upon dry-up of the system, temperature immediately experiences a very sharp increase. Then it flattens for temperatures approaching that of the magmatic gas, 915°C. 3.2. Redox potential The diagram of log f O2 versus ξ (fig. 3a) is similar in shape to fig. 2, due to the marked temperature dependence of log f O2. As shown by fig. 4a, log f O2 values are initially intermedi- ate between those fixed by the FeO-FeO1.5 hy- drothermal redox buffer (Giggenbach, 1987, temperature in Kelvin) 1 HH S S 2( ) ( ) ( )g g g2 2 2 = + H OCO H CO( ) ( ) ( ) ( )g g g g2 2 2+ = + H S H HO SO 322 ( ) ( ) ( ) ( )g g g g2 2 2= ++ 747 Geochemical modeling of magmatic gas scrubbing crease, but for ξ > 0.14 they follow a linear trend which is parallel to that of the moles of S added to the system. This means that for ξ > 0.14 mol the ratio between the amount of S removed from the system and the amount of S added to the sys- tem is constant. Since this ratio is 0.81, it can be concluded that most of the S added to the system is precipitated as crystalline α-sulfur, as already observed by Symonds et al. (2001). 3.4. Chemistry of the liquid phase The pH of the aqueous solution decreases almost linearly with logξ, up to ξ of ∼ 40 mol (fig. 6), whereas the decrease in pH becomes dramatic above this threshold, when the liquid phase is essentially a very concentrated aque- ous solution of hydrogen chloride. In particular, the «last drop» of liquid (for ξ of 79.3 moles Fig. 3a,b. Variations in a) oxygen fugacity and b) during the progressive ad- dition of hot (915°C) magmatic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simulation. Fig. 4a,b. Plot of a) log fO2 and b) RH versus temperature showing the expected changes in these variables dur- ing the progressive addition of hot (915°C) magmatic gas into 1000 g of pure water, initially at 25°C (line la- belled simulation). Pressure was kept at 0.1 MPa throughout the simulation. Also shown are the FeO-FeO1.5 hy- drothermal redox buffer and the H2S-SO2, magmatic redox buffer of Giggenbach (1987). The latter was com- puted for the same fH2O of the computer simulation. / /log logR f f X XH H H O H H O2 2 2 2,= 3 4 a a b b 748 Fig. 5. Moles of sulfur both added to the system and precipitated as crystalline α-sulfur during the progressive addition of hot (915°C) magmatic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simulation. Fig. 6. Changes in the pH of the aqueous solution during the progressive addition of hot (915°C) mag- matic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simula- tion. Fig. 7. Changes in dissolved carbonate species dur- ing the progressive addition of hot (915°C) magmat- ic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simula- tion. Fig. 8. Changes in dissolved sulfur species during the progressive addition of hot (915°C) magmatic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simu- lation. Luigi Marini and Barbara Gambardella 749 Geochemical modeling of magmatic gas scrubbing Fig. 9. Changes in dissolved chloride and fluoride species during the progressive addition of hot (915°C) magmatic gas into 1000 g of pure water ini- tially at 25°C. Pressure was kept at 0.1 MPa through- out the simulation. Fig. 10. Weight percentage of gas in the system during the progressive addition of hot (915°C) magmatic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simulation. and y of 0.993) is dominated by HCl(aq), with a molality of 20.2, followed by H+ and Cl−, which are present with equal molalities of 7.96. Since activity coefficients of ionic solutes were com- puted by means of the B-dot equation of Helge- son (1969), which is expected to be accurate only in dilute aqueous solutions (i.e., for ionic strength lower than 1 molal; Wolery, 1992), these values are highly uncertain. On the one hand, the concentrations of dis- solved CO2 and H2S (which enter preferentially the gas phase, with Bi at 100°C of 1405 and 29.8, respectively) increase with ξ in the single liquid phase but experience a marked decrease with ξ upon transition from the single liquid region to the biphase liquid-plus-gas region (figs. 7 and 8). The corresponding dissociation products, i.e., HCO3− and HS− ions, show variations similar to those of CO2 and H2S. On the other hand, the concentrations of dissolved HCl and HF (which preferentially remain into the liquid phase, with Bi at 100°C of 0.00121 and 0.00221, respective- ly) increase continuously with ξ, both in the liq- uid region and in the liquid-plus-gas region (fig. 750 Fig. 11. Variations in the molar fractions of H2O(g), CO2(g), H2S(g), SO2(g), CO(g), and S2(g), during the progres- sive addition of hot (915°C) magmatic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simulation. Fig. 12. Variations in the molar fractions of H2(g), HCl(g), and HF(g) during the progressive addition of hot (915°C) magmatic gas into 1000 g of pure water initially at 25°C. Pressure was kept at 0.1 MPa throughout the simulation. Luigi Marini and Barbara Gambardella 751 Geochemical modeling of magmatic gas scrubbing 9). Changes in the concentration of dissolved HCl and HF with ξ are controlled by the disso- ciation reactions (3.3) and (3.4) whose pK values at 100°C are −0.62 and 3.85, respectively. These pK values correspond to the iso-activity pH’s, where the activity of HX(aq) equals that of X−. Since pH decreases regularly with ξ, the iso-activity condition is attained much earlier for HF than for HCl. The distribu- tion of HF and F− is further complicated by HF dimerization (3.5) and acid dissociation of the dimer (3.6) The evolution of dissolved S species (fig. 8) is rather complicated due to the occurrence of several association/dissociation and redox reac- tions (Symonds et al., 2001). Dissolved hydro- gen sulfide is the dominant aqueous S species for ξ ≤ 3 moles, whereas HSO4− ion prevails above this threshold, indicating an increase in fO2. Concurrently, the concentration of aqueous SO2 increases, although it remains a subordi- nate S species throughout the process. Similar to what was observed for pH and Cl species, al- so some S species (e.g., HSO4− and SO4 2−) expe- rience dramatic changes approaching dry-up of the system. 3.5. Chemistry of the gas phase In the biphase liquid-plus-gas region, the weight percentage of gases in the system con- sidered increases almost linearly with ξ in the log-log plot of fig. 10, attaining 100% (dry-up of the system) for ξ of 79.3 moles. The gas phase is initially dominated by CO2(g), but H2O(g) becomes the prevailing gas species for ξ ≥ 4 .H F HF H2 2 2= + - + HF H F2 2 2= H FHF(aq) = + + - H ClHCl(aq) = ++ - moles (fig. 11). Hydrogen sulfide is the third major gas species through most of the biphase liquid-plus-gas region. Approaching dry-up of the system, HF(g) first and HCl(g) and SO2(g) af- terwards appear in the gas phase and their con- centrations increase progressively with ξ. In particular, HCl(g) experiences a large increase soon before dry-up and becomes the third ma- jor gas species, due to the high concentration of HCl(aq) in the «last drop» of liquid. In the single gas region, H2O(g) and CO2(g) are the two main constituents, initially followed by HCl(g), whose molar fraction decreases with ξ. On the contrary, the concentrations of SO2(g), H2(g), S2(g), and CO(g) increase by several orders of magnitude, and H2(g) becomes more important than HCl(g) for ξ > 1600 mol. Gas phases rela- tively rich in the magmatic component are char- acterised by similar concentrations of SO2(g), H2S(g), H2(g), and HCl(g), whereas CO(g), S2(g), and HF(g) are present in lower amounts. As already recalled, the similarity in H2S(g) and SO2(g) molar fractions explains why redox conditions are consistent with the H2S-SO2 magmatic redox buffer of Giggenbach (1987). 4. Conclusions The EQ3/6 software package, version 7.2, can be effectively used to compute the irre- versible gas-water mass exchanges taking place during addition of magmatic gas to pure water at 0.1 MPa, in the liquid and liquid-plus-gas re- gions, similar to what was done by Symonds et al. (2001) by means of the software code CHILLER. However some post-calculations are needed to take in due account the effects of gas separation. In this specific case, these post-calcu- lations were simplified by concurrent precipita- tion of crystalline α-sulfur, which fixes the redox potential. Post-calculations might be not as sim- ple as in this case if precipitation of elemental sulfur does not take place. Luckily, this process is ubiquitous and precocious, considering that it begins upon addition of 0.45 g of magmatic gas only, at least in the case considered. Since reaction path modeling is largely based on the assumption of conservation of en- thalpy upon water-gas mixing, we reviewed the 752 Luigi Marini and Barbara Gambardella enthalpies of the gas species of interest, adopt- ing as reference state the liquid phase at the triple point. Used data are presented (table II) for comparison with other possible test cases. Our results of geochemical modeling are very similar to those obtained by Symonds et al. (2001), apart from minor differences due to our choice of modeling magmatic gas scrub- bing by pure water rather than by air-saturated water, as done by Symonds et al. (2001). As already shown by Symonds et al. (2001), results of reaction path modeling carried out in this investigation indicate that scrubbing by wa- ter prevents significant emissions of strongly acidic gas species (SO2(g), HCl(g), and HF(g)) un- til dry conditions are established, at least locally. However substantial HCl(g) degassing can occur from acidic brines rich in HCl(aq). These results are also consistent with the rule of thumb which is generally used to distinguish SO2-, HCl-, and HF-bearing magmatic gases from SO2-, HCl-, and HF-free hydrothermal gases (Giggenbach, 1980, 1987; Chiodini et al., 1993). Finally, we would like to emphasize once more the power of reaction path modeling, which is also highly recommended for the cor- rect interpretation of isotope data, especially multivalent elements, e.g., sulfur (e.g., Gam- bardella, 2001; Marini et al., 2002). Acknowledgements We thank Roberto Moretti, Paolo Papale, and Guido Ventura for the excellent organisation of the ESF LESC Exploratory Workshop «Gases in magmatic evolution: from depth to atmosphere, from micro to macro-scale, from calculation to observation», held in Rome on 11-13 May 2003. We express our appreciation to Giovanni Chiodini, Johan Varekamp, and Ünsal Gemici for their constructive comments on the first ver- sion of the manuscript. Our contribution was inspired by the out- standing 2001 paper by Robert Symonds, Ter- rence Gerlach, and Mark Reed, who really did a great job. We do not intend to rival with them but to pay a tribute to them, stressing once more the importance of geochemical modeling to un- derstand natural processes. REFERENCES BENSON, S.W. (1968): Thermochemical Kinetics (Wiley, New York), pp. 223. CASTELLAN, G.W. (1971): Physical Chemistry (Addison Wesley, Reading, Massachusetts), 2nd edition, pp. 866. CHASE, M.W. Jr. (1998): NIST-JANAF Thermochemical ta- bles, J. Phys. Chem. Ref. Data, Monogr. No. 9, 4th edi- tion, pp. 1951. CHIODINI, G., R. CIONI and L. MARINI (1993): Reactions governing the chemistry of crater fumaroles from Vul- cano Island, Italy, and implications for volcanic sur- veillance, Appl. Geochem., 8, 357-371. CHIODINI, G., R. AVINO, S. CALIRO, D. GRANIERI, C. MINOP- OLI, R. MORETTI, L. PERROTTA and M. RUSSO (2002): Relazione sulle emissioni gassose antistanti l’Isola di Panarea (novembre-dicembre 2002), OV-INGV Rep., pp. 13. DELMELLE, P. and A. BERNARD (2000): Volcanic lakes, in Encyclopedia of Volcanoes, edited by H. SIGURDSSON, B.F. HOUGHTON, S.R. MCNUTT, H. RYMER and J. STIX (Academic Press, San Diego), 877-895. DENBIGH, K. (1971): The Principles of Chemical Equilibri- um (Cambridge University Press), 3rd edition, pp. 494. GAMBARDELLA, B. (2001): Concentrazione e rapporto iso- topico dello zolfo dei fluidi fumarolici della Solfatara di Pozzuoli e di altre manifestazioni campane, OV-IN- GV Rep. (unpublished). GIGGENBACH, W.F. (1980): Geothermal gas equilibria, Geochim. Cosmochim. Acta, 44, 2021-2032. GIGGENBACH, W.F. (1987): Redox processes governing the chemistry of fumarolic gas discharges from White Is- land, New Zeland, Appl. Geochem., 2, 143-161. GIGGENBACH, W.F. (1997): The origin and evolution of flu- ids in magmatic-hydrothermal systems, in Geochem- istry of Hydrothermal Ore Deposits, edited by H.L. Barnes (John Wiley & Sons, New York), 3rd edition, 737-796. HELGESON, H.C. (1968): Evaluation of irreversible reactions in geochemical processes involving minerals and aque- ous solutions, I. Thermodynamic relations, Geochim. Cosmochim. Acta, 32, 853-877. HELGESON, H.C. (1969): Thermodynamics of hydrothermal systems at elevated temperatures and pressures, Am. J. Sci., 267, 729-804. HELGESON, H.C. and D.H. KIRKHAM (1974): Theoretical prediction of the thermodynamic behavior of aqueous electrolytes at high pressures and temperatures, I. Sum- mary of the thermodynamic/electrostatic properties of the solvent, Am. J. Sci., 274, 1089-1198. HELGESON, H.C., R.M. GARRELS and F.T. MACKENZIE (1969): Evaluation of irreversible reactions in geo- chemical processes involving minerals aqueous solu- tions, II. Applications, Geochim. Cosmochim. Acta, 33, 455-481. JOHNSON, J.W., E.H. OELKERS and H.C. HELGESON (1992): SUPCRT92: a software package for calculating the stan- dard molal thermodynamic properties of minerals, gas- es, aqueous species, and reactions from 1 to 5000 bar and 0 to 1000°C, Comput. Geosci., 18, 899-947. KEENAN, J.H., F.G. KEYES, P.G. HILL and J.G. MOORE (1978): Steam Tables. Thermodynamic Properties of Water Including Vapor, Liquid, and Solid Phases (In- 753 Geochemical modeling of magmatic gas scrubbing ternational System of Units – SI), (Wiley, New York), pp. 162. LE GUERN, F., T.M. GERLACH and A. NOHL (1982): Field gas chromatograph analyses of gases from a glowing dome at Merapi volcano, Java, Indonesia, 1977, 1978, 1979, J. Volcanol. Geotherm. Res., 47, 223-245. LIDE, D.R. (2002): CRC Handbook of Chemistry and Physics (CRC Press, Boca Raton), pp. 2664. MARINI, L., B. GAMBARDELLA, C. PRINCIPE, A. ARIAS, T. BROMBACH and J.C. HUNZIKER (2002): Characteriza- tion of magmatic sulfur in the Aegean Island Arc by means of the d34S values of fumarolic gases, elemental sulfur, and hydrothermal gypsum of Nisyros and Milos islands, Earth Planet. Sci. Lett., 200, 15-31. MARINI, L., A. YOCK FUNG and E. SANCHEZ (2003a): Use of reaction path modeling to identify the processes gover- ning the generation of neutral Na-Cl and acidic Na-Cl- SO4 deep geothermal liquids at Miravalles geothermal system, Costa Rica, J. Volcanol. Geotherm. Res., 128, 363-387. MARINI, L., M. VETUSCHI ZUCCOLINI and G. SALDI (2003b): The bimodal pH distribution in volcanic lake waters, J. Volcanol. Geotherm. Res., 121, 83-98. REED, M.H. (1997): Hydrothermal alteration and its relation- ship to ore fluid composition, in Geochemistry of Hy- drothermal Ore Deposits, edited by H.L. BARNES (John Wiley & Sons, New York), 3rd edition, 303-366. ROWE, G.L. JR., S. OHSAWA, B. TAKANO, S.L. BRANTLEY, J.F. FERNANDEZ and J. BARQUERO (1992): Using crater lake chemistry to predict volcanic activity at Poás vol- cano, Costa Rica, Bull. Volcanol., 54, 494-503. SYMONDS, R.B., W.I. ROSE, G.J.S. BLUTH and T.M. GERLACH (1994): Volcanic-gas studies: methods, results, and ap- plications, in Volatiles in Magmas, edited by M.R. CAR- ROLL and J.R. HOLLOWAY, Rev. Mineral., 30, 1-66. SYMONDS, R.B., T.M. GERLACH and M.H. REED (2001): Magmatic gas scrubbing: implications for volcano monitoring, J. Volcanol. Geotherm. Res., 108, 303-341. VAREKAMP, J.C., G.B. PASTERNACK and G.L. JR. ROWE (2000): Volcanic lake systematics, II. Chemical con- straints, in Crater Lakes, edited by J.C. VAREKAMP and G.L. ROWE JR., J. Volcanol. Geotherm. Res., 97, 161- 179. VAREKAMP, J.C., A.P. OUIMETTE, S.W. HERMAN, A. BERMÚDEZ and D. DELPINO (2001): Hydrothermal element fluxes from Copahue, Argentina: a «beehive» volcano in tur- moil, Geology, 29, 1059-1062. WOLERY, T.J. (1992): EQ3NR, a computer program for geo- chemical aqueous speciation-solubility calculations: the- oretical manual, user’s guide and related documentation (version 7.0), Rep. UCRL-MA-110662 PT III (Lawrence Livermore National Laboratory, Livermore). WOLERY, T.J. and S.A. DAVELER (1992): EQ6, a computer program for reaction path modeling of aqueous geo- chemical systems: theoretical manual, user’s guide, and related documentation (version 7.0), Rep. UCRL- MA-110662 PT IV (Lawrence Livermore National Lab- oratory, Livermore).