Hydrothermal gas equilibria: the H2O-H2-CO2-CO-CH4 system

Geochimica et Cosmochimica Acta, Vol. 62, No. 15, pp. 2673–2687, 1998 Copyright © 1998 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/98 $19.00 1 .00

Pergamon

PII S0016-7037(98)00181-1

Hydrothermal gas equilibria: The H2O-H2-CO2-CO-CH4 system GIOVANNI CHIODINI1 and LUIGI MARINI2 1

Osservatorio Vesuviano, via Manzoni 249, 80122 Napoli, Italy Dipartimento di Scienze della Terra, Universita` di Genova, Corso Europa 26, 16132 Genova, Italy

2

(Received November 26, 1997; accepted in revised form May 8, 1998)

Abstract—The difficulty in measuring reservoir gas concentrations in geothermal systems often forces the use of gas ratios in a separated vapor phase to investigate reservoir conditions. Measured CO/CO2 and H2/H2O ratios of fumarolic fluids and vapors from geothermal wells representative of twenty-two different hydrothermal systems are consistent with theoretical values obtained from either of two commonly used redox buffers, indicating that CO and H2 attain chemical equilibrium in the hydrothermal reservoir. Use of different fO2-buffers has little effect on these functions. Many measured CH4/CO2 ratios are, instead, inconsistent with theoretical values obtained with any redox buffer. Since CH4/CO2 ratios are strongly affected by redox conditions in the gas equilibration zone, this disagreement between measured and theoretical values likely indicates that either no unique fO2-buffer is active in all the hydrothermal environments or that CH4 is not in equilibrium with the other gases. The weight of CH4 on the 3log(XCO/XCO2) 1 log(XCO/XCH4) function is relatively small. Therefore this function and the log(XCO/XCO2) 2 log(XH2/XH2O) function, both of which are independent upon redox conditions, were used. These functions gave reasonable estimates of the equilibrium temperature and either the fraction of separated steam or the fraction of condensed steam in each sample. From these data, the CO/CO2, H2/H2O, and H2/CO ratios in the hypothetical single saturated vapor phase were calculated and used to investigate fO2 and fCO2 distributions in the considered twenty-two hydrothermal systems. Recalculated fCO2 values are generally consistent, within one-half log-unit, with the full equilibrium function of Giggenbach (1984, 1988) although production of thermometamorphic CO2 might locally take place. It is evident that no unique fO2-buffer is active in all the hydrothermal environments. This fact imply that CH4 could have attained chemical equilibrium with other gas species in the H2O-H2-CO2-CO-CH4 system. Copyright © 1998 Elsevier Science Ltd to the thermodynamic data by Stull et al. (1969). The results of these calculations are expressed in suitable graphics, also showing the effects of possible processes (boiling and steam condensation) experienced by deep fluids during the upflow towards the surface. The composition of vapors from twenty-two geothermal systems are then compared with theoretical data, in order to (1) establish if these graphic methods have a general validity for evaluating thermodynamic conditions at depth under the hypothesis of chemical equilibrium among gas species and (2) investigate the secondary processes affecting gas compositions.

INTRODUCTION

Gas equilibria in hydrothermal environments have been extensively investigated in the last 25 years, mainly for the exploration and exploitation of geothermal resources (e.g., Tonani, 1973; D’Amore and Nuti, 1977; Giggenbach, 1980, 1991, 1993; D’Amore and Panichi, 1980; D’Amore and Celati, 1983; Nehring and D’Amore, 1984; Arnorsson and Gunnlaugsson, 1985; Bertrami et al., 1985; Arnorsson, 1987, 1990; Chiodini and Cioni 1989; Arnorsson et al., 1990; D’Amore, 1991) and subordinately for the assessment of the risk of hydrothermal eruptions (e.g., Chiodini et al., 1991a, 1992, 1993a). The H2O-H2-CO2-CO-CH4 system played a pivotal role in most of these studies although equilibria involving S-bearing gases, such as H2S and less frequently COS (Chiodini et al., 1991b), and N-compounds, such as N2 and NH3, were also taken into consideration to derive suitable geoindicators. More recently two geothermometers, based on the H2/Ar and CO2/Ar ratios, were proposed, supposing Ar to be present in hydrothermal fluids in relative contents close to those of airsaturated groundwater (Giggenbach, 1991). Following a similar approach, a CO2/N2 geothermometer was suggested by Arnorsson (1987). This study is restricted to hydrothermal gas equilibria in the H2O-H2-CO2-CO-CH4 system. Graphical techniques for vapor discharges from fumaroles and geothermal wells are reviewed below. First, the theoretical compositions of gaseous mixtures in chemical equilibrium under fixed temperature-pressure-redox or temperature-pressure conditions are computed, referring

2. ANALYTICAL DATA

The measured data used in this study refer to hydrothermal systems, covering a wide range of volcanic and nonvolcanic environments, for which complete analytical data are available on H2O, H2, CO2, CO, and CH4 concentrations in gas effluents (Tables 1 and 2). Many data are on hydrothermal fluids discharged from natural manifestations, i.e., the fumarolic systems of Vulcano, Campi Flegrei, Vesuvio, Ischia, Pantelleria, and Lipari (Italy), Nisyros (Greece), Teide (Canarias Island, Spain), Ahuachapan (El Salvador), Montserrat (West Indies), Guagua Pichincha (Ecuador), and Tambora (Indonesia). Furthermore either vapor discharges or separated vapors from geothermal boreholes drilled in the following eleven geothermal fields are considered: Larderello, Travale, Amiata, Bagnore, Mofete (Italy), Kizildere (Turkey), Nisyros (Greece), Cagua, Alto Peak and Mahagnao (Philippines), and St. Lucia (West Indies). Gas 2673

2674

G. Chiodini and L. Marini

Hydrothermal gas equilibria

2675

(1) Chiodini et al. (1992) (2) Data from DST (Perugia, Italy) and IGGI-CNR (Pisa, Italy) (3) Chiodini et al. (1991a) (4) Tedesco (1996) (5) Barberi et al. (1987) (6) Data from IGGI-CNR (Pisa, Italy) (7) Marini et al. (1991) (8) Chiodini et al. (1994) (9) Chiodini et al. (1996) (10) Guidi et al. (1990) (11) Chiodini et al. (1993a) (12) Chiodini and Raco (1991)

concentrations in total (liquid1vapor) fluids are known for Nagqu (China), also included in the considered database. Most of these fluids, either coming from active volcanic

areas or elsewhere, have typical hydrothermal characteristics, in that, (1) H2S is the most abundant S-species, whereas SO2 is present in close to undetectable amounts (Giggenbach, 1980);

2676

G. Chiodini and L. Marini

Table 2. Composition of the vapor discharged by geothermal wells and reservoir temperature (Tres °C). Analytical data are expressed in mmol/mol. In the table are reported the following computed parameters: equilibrium temperatures (To °C), fractions of separated vapor (s), or water pressures (PH2O bar)

(1) Where not specified, the reservoir temperatures are from the sample reference (2) Data from DST (Perugia, Italy) and IGGI-CNR (Pisa, Italy) (3) D’Amore et al. (1987) (4) Guidi et al. (1990) (5) Average values of 3 samples. Data from IGGI-CNR (Pisa, Italy) (6) Battistelli et al. (1991) (7) Average values of 4 samples. Chiodini et al. (1993a) (8) Giggenbach (1993) (9) Average value of 4 samples. D’Amore et al. (1990) (10) The Amiata Geothermal field produce from two levels whose temperatures (330 and 225°C) are reported in D’Amore et al. (1987) (11) Atkinson et al. (1977) (12) Na-K temperature (Fournier, 1979). The chemistry of the water separated at atmospheric pressure is from Carella and Guglielminetti (1983)

(2) other acid gas species, e.g., HCl and HF are also virtually absent; (3) CH4 contents are always much greater than CO concentrations. Only some fumarolic discharges of Guagua Pichincha crater and Forgia Vecchia, Vulcano are characterized by detectable SO2 (Marini et al., 1991; Chiodini et al., 1995). 3. THEORETICAL COMPOSITIONS OF HYDROTHERMAL GASES

All the theoretical compositions, which are reported on the diagrams described below, are computed assuming the attainment of chemical equilibrium in the hydrothermal environ-

ments and its quenching during the upflow of hydrothermal fluids towards the surface. In general, these hypotheses are reasonable because gas species spend (1) a comparatively long time (which is probably sufficient for the attainment of chemical equilibrium) in the hydrothermal reservoirs and (2) a relatively short lapse of time in the upflow channels connecting the reservoirs to the surface outlets. The second condition is generally valid not only for geothermal wells, but also for fumaroles, as they are are often located along faults or other features of high vertical permeability. As shown below, to calculate the theoretical composition of

Hydrothermal gas equilibria

hydrothermal gases, the fugacity of H2O and, depending on the approach, the fugacities of CO2 and O2 as well, have to be known at the temperatures of interest. Depending on P, T distribution, the main component of hydrothermal fluids, H2O, can be present in different physical states, i.e., liquid, gas (called in this work superheated vapor to avoid misunderstanding) and saturated liquid 1 vapor. For coexisting vapor and liquid water, changes of log f H2O with temperature are closely approximated by the following equation (Giggenbach, 1980): log f H2O 5 5.510 2 2048/T

(1)

where T is in K. Water fugacity is not constrained by temperature alone and becomes an additional external variable both for a single liquid phase at pressures higher than saturation and for a single vapor phase at pressures lower than saturation. Carbon dioxide fugacity in full equilibrium hydrothermal systems is fixed, at any given temperature, by univariant reactions involving calcite, a Ca-Al-silicate, K-feldspar, K-mica and chalcedony (Giggenbach, 1984, 1988). These minerals constitute the thermodynamically stable alteration assemblage resulting from isochemical recrystallization of an average crustal rock. Values of f CO2 and temperature corresponding to these reactions are given with adequate precision by (T in K): log f CO2 5 0.0168 ~T 2 273.15! 2 3.78

(2)

Also Arnorsson and Gunnlaugsson (1985) recognized that CO2 fugacity is governed, above 230°C, by a buffer comprising calcite, epidote, prehnite, and quartz and proposed an empirical geothermometer that agrees closely with expression 2. As previously suggested by Grant (1982) and Giggenbach (1982), it is unlikely that these f CO2-buffers are ubiquitously efficacious in natural hydrothermal environments, probably because a continuous flux of CO2 can occur through the hydrothermal systems (Mahon et al., 1980). Although in some portions of natural hydrothermal systems f CO2-buffers may be locally active, in general f CO2 acts as an externally fixed parameter. According to Giggenbach (1987) the most suitable parameter describing redox potentials of most natural fluids is RH 5 log rH 5 log (f H2/f H2O) > log (XH2/XH2O)

(3)

He proposed that the redox potentials of hydrothermal systems are governed by the (FeO)-(FeO1.5) couple, which controls also the redox state of basaltic and andesitic magmas as earlier recognized by Fudali (1965). The RH-values referring to the (FeO)-(FeO1.5) buffer are fixed by the reaction H2O 1 2(FeO) 5 H2 1 2(FeO1.5)

(4)

the equilibrium constant of which is equal to RH and is practically temperature independent at –2.82 6 0.02. Considering formation of water from H2 and O2, reaction 4 can be rewritten as

2677

Similar equations relating f O2 values and temperatures in hydrothermal environments were proposed by many authors (e.g., Tonani, 1973; D’Amore and Nuti, 1977; D’Amore and Panichi, 1980; D’Amore and Gianelli, 1984). For the purpose of the following discussion, the empirical relationship of D’Amore and Panichi (1980) is reported here log f O2 5 8.20 2 23643/T

(7)

Oxygen fugacities of oil field waters at 400 bar were computed by Helgeson et al. (1993) through Gibbs free energy minimization calculations, assuming metastable equilibrium among calcite, albite, and relevant organic and inorganic aqueous species. The strict relationship between obtained values of log f O2 and temperature suggests that redox conditions in oil field waters are controlled by metastable equilibrium among carbonate and carboxylate species. At any temperature in the range 100 –150°C, values of log f O2 of oil field waters are approximately 2 and 4 log units lower than those given by Eqns. 6 and 7, respectively. 3.1. Single Vapor Phase It is initially assumed that the considered gas species attain chemical equilibrium in a single vapor phase. The following three reactions provide the equilibrium constraints in the H2OH2-CO2-CO-CH4 system H2O 5 H2 1 1⁄2 O2

(8)

CO2 5 CO 1 1⁄2 O2

(9)

CO2 1 2H2O 5 CH4 1 2O2

(10)

Production of CH4 could take place also through reaction of elemental C and H2. Although elemental C (graphite) may be present in some natural hydrothermal systems such as Cerro Prieto, Mexico (Nehring and D’Amore, 1984) and in Northern Latium, Italy (Chiodini, 1994), these are probably peculiar cases rather than the general situation. The most general approach, i.e., Eqn. 10 is preferred here. The equilibrium constants of reactions 8 –10 can be written as log f H2 5 log KH2 2 1⁄2log f O2 1 log f H2O

(11)

log f CO 5 log KCO 2 1⁄2log f O2 1 log f CO2

(12)

log f CH4 5 log KCH4 2 2log f O2 1 2log f H2O 1 log f CO2

(13)

The temperature dependence of the thermodynamic constants KH2, KCO, and KCH4 is given by the following equations, based on the thermodynamic data by Stull et al. (1969): log KH2 5 212707/T 1 2.548

(14)

(5)

log KCO 5 214955/T 1 5.033

(15)

and log f O2 can be linked to temperature through the following equation:

log KCH4 5 242007/T 1 0.527

(16)

4(FeO1.5) 5 O2 1 4(FeO)

log f O2 5 10.736 2 25414/T

(6)

To solve Eqns. 11–13 fugacities of H2O, CO2, and O2 have to be known at the temperatures of interest.

2678

G. Chiodini and L. Marini

Fig. 1. Fugacities of H2, CO, and CH4 at different temperatures, for a vapor phase coexisting with an infinitesimally small amount of liquid water, calculated assuming that f CO2 is constrained by the full equilibrium function of Giggenbach (1984, 1988), and f O2 is governed by either the FeO-FeO1.5 hydrothermal buffer of Giggenbach (1987, solid lines) or the f O2-buffer of D’Amore and Panichi (1980, dashed lines). Fugacities of H2O and CO2, which are simple temperature functions, are also shown.

3.1.2. Theoretical fugacities of H2, CO, and CH4 Assuming that f CO2 is constrained by Eqn. 2, f O2 is governed by a suitable buffer, e.g., Eqns. 6 or 7, and treating f H2O as either a simple temperature function, for coexisting liquid plus vapor (Eqn. 1), or an externally fixed parameter, for single phase environments, Eqns. 11–13 allow one to compute fugacities of H2, CO, and CH4 at different temperatures (Fig. 1). Use of the two O2-buffers of Giggenbach (1987) and D’Amore and Panichi (1980) leads to different theoretical fugacities of H2, CO, and CH4. Carbon monoxide shows the largest changes with temperature, and it is, therefore, the best geothermometer. As fugacities of single gas species are practically impossible to be determined for fumarolic effluents and only with limited accuracy for well discharges, Fig. 1 cannot be used to compare theoretical and analytical compositions. 3.1.3. Theoretical log-ratios H2 /H2O, CO/CO2, and CH4 /CO2 Most of the disadvantages of referring to fugacities of single gas species are overcome by working with isomolar concentration ratios, which can be analytically determined on both fumarolic fluids and well discharges. In order to meet these requirements, the equilibrium constants of reactions 8 –10 can be reformulated as log (f H2/f H2O) 5 log KH2 2 1⁄2log f O2 > log (XH2/XH2O)

(17)

log (f CO/f CO2) 5 log KCO 2 1⁄2log f O2 > log (XCO/XCO2)

(18)

Fig. 2. Plot of log(XH2/XH2O) vs. log(XCO/XCO2). The theoretical grid assumes that redox conditions in the gas equilibration zone are controlled by the FeO-FeO1.5 hydrothermal buffer of Giggenbach (1987). Compositions of both the vapor and liquid phases (thick solid lines) are shown. Compositions of the vapor phase separated in a single-step (SSVS) at different temperatures from a liquid phase initially at T o 5 150, 200, 250, 300, and 350°C are also shown (thin solid lines) as well as the compositions resulting from single-step vapor separation at two fixed T s temperatures (100 and 200°C) starting from any initial temperature (dashed lines). Circles 5 fumarolic vapors; squares 5 vapors from geothermal wells. Codes are given in Tables 1 and 2.

log (f CH4/f CO2) 5 log KCH2 2 2log f O2 1 2log f H2O > log (XCH4/XCO2)

(19)

For gas phases largely made up of water vapor, i.e., when XH2O . 0.8, the ratios of fugacity coefficients GH2/GH2O, GCO/ GCO2, and GCH4/GCO2 do not deviate significantly from 1, in the typical P,T range of hydrothermal systems, 100 –374°C and 1–220 bar (Ryzhenko and Volkov, 1971; Ryzhenko and Malinin, 1971; Naumov et al., 1974). Therefore, use of ratios of mole fractions in the vapor phase, Xi values, introduces negligible errors (Giggenbach, 1987). The theoretical values obtained from Eqns. 17–19 are reported as vapor lines and superheated vapor grids in Figs. 2 and 4 (redox potentials from Eqn. 6) and in Figs. 3 and 5 (redox potentials from Eqn. 7). It must be underscored that equilibrium values of log(XH2/XH2O) and log(XCO/XCO2) depend upon the temperature and the redox potential in the gas equilibration zone only, whereas theoretical log(XCH4/XCO2) ratios are also controlled by water fugacity. As a consequence, log(XH2/XH2O) and log(XCO/XCO2) in pure saturated vapors are equal to those in superheated vapors at any T and redox condition (vapor lines in Figs. 2 and 3), whereas log(XCH4/XH2O) in pure saturated vapors (vapor lines in Figs. 4 and 5) are different from those in superheated vapors (superheated vapor grids in Figs. 4 and 5).

Hydrothermal gas equilibria

Fig. 3. Plot of log(XH2/XH2O) vs. log(XCO/XCO2). The theoretical grid assumes that redox conditions in the gas equilibration zone are controlled by the f O2-buffer of D’Amore and Panichi (1980). Phase lines and symbols as in Fig. 2. Codes are given in Tables 1 and 2.

3.1.4. Theoretical sums of log-ratios H2 /H2O, CO/CO2 and CH4 /CO2 Geoindicators independent on the redox potential can be derived summing reactions 8 –10 in order to eliminate O2. This

2679

Fig. 5. Plot of log(XCH4/XCO2) vs. log(XCO/XCO2). Theoretical grids for coexisting liquid plus vapor phases (solid lines) and superheated vapors (dashed lines) are computed considering that redox conditions in the gas equilibration zone are governed by the f O2-buffer of D’Amore and Panichi (1980). Symbols as in Fig. 2. Codes are given in Tables 1 and 2.

exercise produces the following five reactions, each including four and excluding one of the considered five constituents: CO2 1 H2 5 CO 1 H2O

(20)

CH4 1 2H2O 5 4 H2 1 CO2

(21)

3CO2 1 CH4 5 4CO 1 2H2O

(22)

CH4 1 H2O 5 3H2 1 CO

(23)

CH4 1 CO2 5 2H2 1 2CO

(24)

Only two of these five reactions are mutually independent, as all the five constituents are involved in any two of these reactions. Let us consider, for instance, reaction 20 or water-gas shift reaction WGS and reaction 22 here indicated with the acronym CCC. Their equilibrium constant espressions are log (f CO/f CO2) 2 log (f H2/f H2O) 5 log KWGS 5 log (XCO/XCO2) 2 log (XH2/XH2O)

(25)

3log (f CO/f CO2) 1 log (f CO/f CH4) 5 log KCCC 2 2 log f H2O 5 3log (XCO/XCO2) 1 log (XCO/XCH4)

(26)

The temperature dependence of the thermodynamic constants KWGS and KCCC is given by the following equations: Fig. 4. Plot of log(XCH4/XCO2) vs. log(XCO/XCO2). Theoretical grids for coexisting liquid plus vapor phases (solid lines) and superheated vapors (dashed lines) are calculated assuming that redox potentials in the gas equilibration zone are controlled by the FeO-FeO1.5 hydrothermal buffer (Giggenbach, 1987). Symbols as in Fig. 2. Codes are given in Tables 1 and 2.

log KWGS 5 22248/T 1 2.485

(27)

log KCCC 5 217813/T 1 19.605

(28)

Equilibrium values of log(XCO/XCO2) 2 log(XH2/XH2O) depend upon the temperature in the gas equilibration zone only,

2680

G. Chiodini and L. Marini

log (XH2/XH2O)L 5 log KH2 2 1⁄2 log f O2 2 log BH2

(34)

log (XCO/XCO2)L 5 log KCO 2 1⁄2 log fO2 2 log (BCO/BCO2)

(35)

log (XCH4/XCO2)L 5 log KCH4 2 2 log f O2 1 2 log f H2O 2 log (BCH4/BCO2)

(36)

log(XCO/XCO2)L 2 log (XH2/XH2O)L 5 log KWGS 2 log (BCO/BCO2) 1 log BH2

(37)

3log (XCO/XCO2)L 1log (XCO/XCH4)L 5 log KCCC 2 2 log f H2O 2 3log (BCO/BCO2) 2 log (BCO/BCH4)

Fig. 6. Plot of 3log(XCO/XCO2) 1 log(XCO/XCH4) vs. log(XH2O/XH2) 1 log(XCO/XCO2). Phase lines and symbols as in Fig. 2. The compositions of superheated vapors equilibrated at different temperature-PH2O values (solid lines) and the compositions of equilibrated, single vapor phases affected by steam condensation at T c 5 100°C (dashed lines) are also shown. The curves labelled c 5 0.5, c 5 0.8, and c 5 0.9 refer to different fractions of condensed steam. Codes are given in Tables 1 and 2.

whereas theoretical values of 3log(XCO/XCO2) 1 log(XCO/ XCH4) are also controlled by water fugacity. Theoretical values of these sums of log-ratios are shown as vapor lines and superheated vapor grids in Fig. 6. 3.3. Single Saturated Liquid Phase Equilibrium gas contents in a single saturated liquid phase are conveniently computed using the vapor-liquid distribution coefficient, Bi, which can be defined as Bi 5 (Xi/XH2O) vapor / (Xi/XH2O) liquid

(29)

The temperature dependence of vapor-liquid distribution coefficients is described, in the 100 –340°C range, by the following equations: log BCO2 5 4.7593 2 0.01092 (T-273.15)

(30)

log BH2 5 6.2283 2 0.01403 (T-273.15)

(31)

log BCH4 5 6.0783 2 0.01383 (T-273.15)

(32)

log BCO 5 6.3173 2 0.01388 (T-273.15)

(33)

given by Giggenbach (1980) for CO2, H2, and CH4 and by Bertrami et al. (1985) for CO. Above the upper temperature limit, Bi values are obtained interpolating between the value at 340°C and 1, i.e., the theoretical value at 374°C for all gas species. Insertion of Eqn. 29 in Eqns. 17, 18, 19, 25, and 26 leads to the following relationships:

(38)

Compositions described by Eqns. 34 –38 can be reached in fumaroles and steam wells only through complete isothermal evaporation of an equilibrated liquid. This process requires a considerable supply of heat from an external source, and it is possible under peculiar situations only, usually absent in most normal hydrothermal systems. Nevertheless they represent a useful reference and would be applicable to any unflashed samples of a deep liquid phase. The theoretical values log(XH2/ XH2O)L, log(XCO/XCO2)L, log(XCH4/XCO2)L, log(XCO/XCO2)L 2 log(XH2/XH2O)L, and 3log(XCO/XCO2)L 1 log(XCO/XCH4)L are reported as liquid lines in Figs. 2 – 6. 3.4. Steam Produced Through Boiling of a Liquid Phase To compare theoretical and measured data, the effect of secondary phenomena affecting rising hydrothermal fluids have to be taken into account. As it is not possible to separate the effects of two or more secondary phenomena for a given sample, we implicitly assume that only one of these phenomena is important in each case. Among these secondary phenomena, cooling processes (boiling and steam condensation) can bring about large chemical changes, due to distribution of different gas species between separating liquid and vapor phases. In order to derive gas geothermometers for fumarolic fluids with outlet temperature close to 100°C, Arnorsson and Gunnlaugsson (1985) calculated CO2, H2S, and H2 contents (as well as related ratios and a Fischer-Tropsch quotient) in steam formed through adiabatic boiling and maximum degassing to atmospheric pressure of natural geothermal liquids. Also according to Chiodini et al. (1992), a realistic model for explaining the genesis of fumarolic fluids is through boiling of a liquid phase. To calculate the effects of boiling on the theoretical values of log(XH2/XH2O)L, log(XCO/XCO2)L, log(XCH4/XCO2)L, log(XCO/ XCO2)L 2 log(XH2/XH2O)L and 3log(XCO/XCO2)L 1 log(XCO/ XCH4)L, it is assumed attainment of chemical equilibrium in a single saturated liquid phase at initial (equilibrium) temperature T o, followed by adiabatic single-step steam separation at temperature T s. The mole fraction of the i-th species in the separated vapor phase, (Xi)Ts, is constrained by the following mass and enthalpy balances (modified from Henley et al., 1984): (Xi/XH2O)L, To 5 (Xi/XH2O)L, Ts (1 2 s) 1 (Xi/XH2O)V, Ts s HL, To 5 HL,Ts (1 2 s) 1 HV,Ts s

(39) (40)

where H is the enthalpy of the vapor (V) and of the liquid (L) at the indicated temperature (Keenan et al., 1969), and s is the

Hydrothermal gas equilibria

fraction of steam separated through boiling. Values of s are obtained solving Eqn. 40. Equation 39 can be rearranged to give log (Xi/Xj)Ts 5 log (XiXj)To 2 log (s1 (1 2 s)/Bi,Ts) 1 log (s1 (1 2 s)/Bj,Ts) 2 log Bi,To 1 log Bj,To

(41)

The following relations are obtained inserting expression 41 into Eqns. 17, 18, 19, 25, and 26: log (XH2/XH2O)Ts 5 log KH2 2 1⁄2 log f O2 2 log (s1 (1 2 s)/BH2,Ts) 2 log BH2,To

(42)

log (XCO/XCO2)Ts 5 log KCO 21⁄2 log f O2 2 log (s1 (1 2 s)/BCO,Ts) 1 log (s1 (1 2 s)/BCO2,Ts) 2 log BCO,To 1 log BCO2,To

(43)

log (XCH4/XCO2)Ts 5 log KCH4 2 2 log f O2 1 2 log f H2O

(44)

log (XCO/XCO2)Ts 2 log (XH2/XH2O)Ts 5 log KWGS 2 log (s1 (1 2 s)/BCO,Ts) 2 log BCO,To 1 log (s1 (1 2s)/BCO2,Ts) 1 log BCO2,To 1 log (s1 (1 2 s)/BH2,Ts) 1 log BH2,To

(45)

3log (XCO/XCO2)Ts 1 log (XCO/XCH4)Ts 5 log KCCC 2 2 log f H2O 2 4log (s1 (1 2 s)/BCO,Ts) 2 4log BCO,To 1 log (s1 (1 2 s)/BCH4,Ts) 1 log BCH4,To 1 3log (s1 (1 2 s)/BCO2,Ts) 1 3log BCO2,To

respectively. The compositional field for vapors generated through single-step steam separation from boiling liquids, labelled as vapor 1 liquid in Figs. 2, 3, and 6, could be alternatively interpreted as due to vapor gain processes. The approach by Giggenbach (1980) and other similar models (e.g., D’Amore and Celati, 1983; Arnorsson et al., 1990; D’Amore, 1991) are based on total fluid composition. It is also required that total fluid composition measured at the surface is fully representative of reservoir conditions (e.g., absence of gas-slip effect). These vapor gain models cannot be applied to vapors which are generated through steam separation from boiling liquids, such as those discharged by fumaroles. Vapor gain models cannot be applied also to geothermal wells of which only the composition of the separated vapor phase is known, such as all the cases considered here, with the exception of Nagqu geothermal well. 4. COMPARISON BETWEEN THEORETICAL AND ANALYTICAL COMPOSITIONS

4.1. Single log-ratios H2 /H2O, CO/CO2, and CH4 /CO2

2 log (s1 (1 2 s)/BCH4,Ts) 2 log BCH4,To 1 log (s1 (1 2 s)/BCO2,Ts) 1 log BCO2,To

2681

(46)

The theoretical log ratios and sums of log ratios values in the vapor phase separated at different temperatures from a liquid phase initially at 150, 200, 250, 300, and 350°C are shown in Figs. 2– 6 as single-step vapor separation lines (SSVS lines). In Figs. 2, 3, and 6 the theoretical values of vapor separated at 100 and 200°C starting from any initial temperature are represented as T s lines. As already pointed out by Arnorsson and Gunnlaugsson (1985), boiling phenomena occurring in the upflow zones of natural hydrothermal systems may depart from the adiabatic model. Besides they observed that steam separation can take place at different pressures, especially when the boiling water experiences a significant horizontal displacement. However occurrence of either multi-step or continuous separation (see equations in Henley et al., 1984) would quickly move gas composition out of the space delimited by the liquid and vapor lines in Figs. 2– 6. Therefore, these processes do not seem important, at least in the systems considered here. Giggenbach (1980) showed that the contents of CO2, H2S, NH3, H2, N2, and CH4 in the fluids discharged from geothermal wells of Wairakei, Broadlands, and Kawerau (New Zealand) are closely approximated by two theoretical processes, either addition to or removal from an equilibrium liquid phase of an equilibrium vapor phase, defined as vapor gain and vapor loss,

In the diagrams log(XH2/XH2O) vs. log(XCO/XCO2) analytical data are compared with the expected values of hypothetical hydrothermal fluids whose redox conditions are governed by the f O2-buffers of Giggenbach (1980), Fig. 2, and D’Amore and Panichi (1980), Fig. 3. In both diagrams observed values generally fall between the vapor line and the 100°C T s line; that is, in the expected compositional fields for vapors generated through single-step steam separation from boiling liquids of different initial temperatures. The position of the points along the single-step vapor separation (SSVS) lines is indicative of the extent of the boiling process. Points located near the vapor line have relatively high separation temperatures and low fractions of separated vapor, whereas points close to the 100°C T s line are relatable to more energetic boiling processes, i.e., higher fractions of separated vapor. Obviously equilibrium and separation temperatures and s values estimated for the natural vapors show some differences depending on the redox buffer of reference. In the diagrams log(XCH4/XCO2) vs. log(XCO/XCO2) of Figs. 4 and 5, only some points are consistent with the theoretical grids for hydrothermal fluids whose redox conditions are controlled by either the FeO-FeO1.5 hydrothermal buffer (Giggenbach, 1987) or the f O2-buffers by D’Amore and Panichi (1980). The interpretation of CH4 data is complicated by different causes. (1) A first problem is due to the overlap of the field of saturated vapors with the field of superheated vapors. In principle, some deviations below the field limited by the vapor and liquid curves could be attributed to equilibration in a superheated vapor phase, particularly considering the D’Amore-Panichi buffer (Fig. 5). However, a large number of points lying above the vapor line for this buffer are not explained. (2) A second problem might be due to kinetic reasons. The sluggish behaviour of CH4 could explain the disagreement between theoretical and analytical data. Studies of the crater fumaroles of White Island (Giggenbach, 1987) and Vulcano Island (Chiodini et al., 1993b, 1995) have shown that the two most reactive gas species in the considered

2682

G. Chiodini and L. Marini

system are H2 and CO. They react very fast, reequilibrating at outlet conditions under the redox potential fixed by the SO2-H2S magmatic gas buffer (Giggenbach, 1987), for temperatures above 300°C (H2) and 300 – 400°C (CO). Methane is undetectable in Vulcano crater fumaroles and shows a large variability at White Island, where the CH4CO2 system seems to attain equilibrium only within the very slow moving deep parts of the two-phase, hydrothermal system surrounding the magmatic system (Giggenbach, 1987). Although redox potentials prevailing in hydrothermal systems are more reducing than those of the magmatic White Island and Vulcano Island systems, it seems likely that H2 and CO are fast-reacting species and CH4 is one of the slowest species to equilibrate in the hydrothermal environments as well (Giggenbach, 1991). For example, Nehring and D’Amore (1984) recognized that H2-CO2 equilibrium temperatures of Cerro Prieto geothermal wells are systematically 30°C lower than CH4-H2-CO2 temperatures and concluded that reactions involved in the equilibration of H2-CO2 respond faster than those involving CH4. An attempt to quantify the kinetics of chemical equilibration between CO2 and CH4 was presented by Giggenbach (1997). Assuming that the conversion of CO2 to CH4 may be described by a pseudo-first order reaction, the reaction half-time would be about 12a at 300°C, 500a at 200°C and 160ka at 100°C. However, these are very preliminary estimates based on limited experimental data and further researches are needed to clarify the kinetics of conversion of CO2 to CH4 in the presence of catalysts occurring in natural environments (Giggenbach, 1997). (3) Finally, it is possible that gas compositions of the twentytwo hydrothermal systems can be controlled by variable redox conditions and that no unique fO2 buffer is ubiquitously efficacious in natural hydrothermal systems. The effect of different redox conditions is particularly strong for the log(XCH4/XCO2) ratio whose dependence on log fO2 is four times larger than for both the log(XCO/XCO2) and the log(XH2/XH2O) ratios (see Eqns. 17–19). For example the theoretical log(XCH4/XCO2) values, computed with the two redox buffers described by Eqns. 6 and 7 and reported in Figs. 4 and 5, are similar only at relatively high temperatures while they differ by several orders of magnitude at low temperatures. In Fig. 4 some measured points are consistent with the expected compositions based on the FeO-FeO1.5 hydrothermal buffer, whereas other points show lower XCH4/XCO2 ratios. Most of the fumarolic fluids with low XCH4/XCO2 ratios come from active volcanic areas (such as Guagua Pichincha, Teide, Campi Flegrei, Ischia, Vulcano Island, etc.) and their low XCH4/XCO2 ratios could be caused by comparatively more oxidizing redox conditions due to input of magmatic fluids into the roots of these hydrothermal systems. This process is actually evident at Guagua Pichincha and Forgia Vecchia, Vulcano, where fumarolic fluids exhibit significant contents of SO2. 4.2. Sums of log-ratios H2 /H2O, CO/CO2 and CH4 /CO2 Giggenbach (1987, 1991) criticized the use of equations involving sums of log-ratios to evaluate gas equilibria pointing out that “the major disadvantage of this approach is the uncer-

tainty arising from the implied assumption that all the reactants are still present in the sample in proportions representative of the deeper equilibrium environment. The degree of uncertainty increases with the number of constituents as each may be affected to different degrees by secondary processes, e.g., caused by different kinetic responses to changes in temperature, pressure, or redox potential during the rise of the fluid mixture to the surface.” This observation is true also for the H2O-H2CO2-CO-CH4 system. In principle, CH4 is not a good reaction partner of H2 or CO, as reactions involving CH4 are much slower than those involving H2 and CO. Therefore, these species should not be combined such as in reaction 22. In practice, however, CO weighs four times more than CH4 on KCCC and can thus be used with some confidence. The use of equations involving sums of log-ratios also has the advantage, with respect to the use of single log-ratios, that it is not affected by redox potentials, which can be estimated afterwards (see chapter 5). The computed redox-independent variables log(XCO/XCO2) 2 log(XH2/XH2O), corresponding to reaction 20, and 3log(XCO/ XCO2) 1 log(XCO/XCH4), corresponding to reaction 22, are compared with the correspondent analytical values in Fig. 6. As in Figs. 2–5 the results of theoretical models are reported as vapor, liquid, SSVS and Ts lines in Fig. 6. The vapor and liquid lines divide the diagram into three fields: the field of superheated vapors above the vapor line, the field of saturated vapors generated by boiling processes (labelled vapor 1 liquid), and the field of the liquids that have experienced vapor loss to the right of the liquid line. On the basis of the two independent relations plotted in Fig. 6 two parameters can be estimated for each measured point. For most points, which fall between the vapor line and the T s 5 100°C line, these two parameters are the equilibrium temperature T o and the separation temperature T s (and consequently the fraction of separated vapor s). The few points which fall between the T s 5 100°C line and the liquid line (three samples from Ischia, I, and one from Guagua Pichincha, G) can be considered representative of equilibrated liquids that have experienced extensive isothermal evaporation (s > 1). The fumaroles of Vesuvio (Ve) and Vulcano Forgia Vecchia (Vf), which fall above the critical point of pure water, can be considered representative of a pure equilibrated vapor phase. The wells of Santa Lucia (SL) and Cagua (Ph) geothermal systems, which are located above the vapor line, are possibly fed by superheated vapors. Among the fumaroles lying above the vapor line only those of Solfatara di Pozzuoli, Campi Flegrei (S) can be considered fully representative of an equilibrated vapor phase, i.e., no condensation and separation of liquid droplets took place. In fact, these discharges are characterized by outlet temperatures close to 160°C, much higher than the boiling point of water, and very high flow rates. In these cases the two estimated parameters are equilibrium temperature and water partial pressure. For Teide Volcano (Te), Pantelleria (P), and Vulcano Island-VI (V) fumaroles the possible influence of steam condensation cannot be ruled out. These fumaroles have discharge temperatures equal to or sligthly lower than the boiling point of water and low flow rates, suggesting the possible occurrence of steam condensation. This is evident for Vulcano fumarole VI, where the condensed steam collects in a mud pool at few

Hydrothermal gas equilibria

meters from the fumaroles. In these cases the estimated parameters are the equilibration temperature and the fraction of condensed steam as shown below. The importance of steam condensation, particularly on geothermometers based on the concentrations of single gas constituents, was recognized by Arnorsson and Gunnlaugsson (1985). In principle, steam condensation can take place through either (1) conductive heat loss or (2) addition of cold shallow water to the uprising hydrothermal fluids or both. Arnorsson (1987) proposed a method to evaluate separately the effects of each process of steam condensation on gas geothermometers. The equilibrium temperature and the fraction of condensed steam were estimated assuming attainment of equilibrium in a single saturated vapor phase and occurrence of steam condensation through conductive heat loss at constant temperature, T c 5 100°C. The expected molar ratios in the fluid after condensation, (Xi/Xj)c and the initial (equilibrium) values, Xi/ Xj, are related through Xi/Xj 5 (Xi/Xj)c (1 2 c 1 c/Bi,Tc)/(1 2 c 1 c/Bj,Tc)

(47)

where c is the fraction of condensed steam and Bi,Tc the vapor-liquid distribution coefficient of gas species i at condensation temperature. The following relations are obtained inserting Eqn. 47 in expressions 25 and 26: log (XCO/XCO2)c 2 log (XH2/XH2O)c 5 log KWGS 2 log (1 2 c 1 c/BCO,Tc) 1 log (1 2 c 1 c/BCO2,Tc) 1 log (1 2 c 1 c/BH2,Tc)

(48)

3log (XCO/XCO2)c 1 log (XCO/XCH4)c 5 log KCCC 2 2 log f H2O 24log (1 2c 1 c/BCO,Tc) 1 3log (1 2 c 1 c/BCO2,Tc) 1 log (1 2 c 1 c/BCH4,Tc)

(49)

Theoretical log(XCO/XCO2)c 2 log(XH2/XH2O)c and 3log(XCO/ XCO2)c 1 log(XCO/XCH4)c values define in Fig. 6 a steam condensation grid. Comparison of these theoretical values with the analytical values allows one to estimate the fractions of condensed steam and the equilibrium temperatures. 4.3. Estimations of Equilibrium Temperatures and Either Fractions of Separated Vapor s, or Fractions of Condensed Steam c, or Water Pressures All the results reported in Tables 1 and 2 (equilibrium temperatures and either fractions of separated vapor s, or fractions of condensed steam c, or water pressures), are based on Fig. 6, that is on the sums of log-ratios log(XCO/XCO2) 2 log(XH2/XH2O) and 3log(XCO/XCO2) 1 log(XCO/XCH4). However, equilibrium temperatures and other variables (s, c, PH2O) cannot be inferred from mere inspection of Fig. 6 with sufficient accuracy. For the few samples representative of a liquid phase, s has been assumed equal to 1, and the equilibrium temperature has been computed from Eqn. 38. For the samples representative of equilibrated, superheated vapor phases the equilibrium temperatures and water partial pressures have been computed solving Eqns. 25 and 26. For most samples, which are representative of vapors separated from boiling liquids, equilibrium and separation temperatures have been obtained by

2683

means of a computer code which calculates theoretical log(XCO/XCO2)Ts 2 log(XH2/XH2O)Ts and 3log(XCO/XCO2)Ts 1 log(XCO/XCH4)Ts values (Eqns. 45 and 46) at varying T o and Ts, until convergence with corresponding analytical values is attained. For the samples representative of vapor phases affected by condensation, theoretical log(XCO/XCO2)c 2 log(XH2/XH2O)c and 3log(XCO/XCO2)c 1 log(XCO/XCH4)c values (Eqns. 48 and 49) have been calculated by means of a computer code at varying T o and at fixed T c 5 100°C, until the difference between theoretical and analytical values is , 0.001 log-units. Similar computations have been done to obtain the equilibrium temperatures from the log-ratios H2 /H2O and CO/CO2, referring to the two f O2-buffers by Giggenbach (1987) and D’Amore and Panichi (1980), that is to Figs. 2 and 3, respectively. A check of the reliability of the equilibrium temperatures obtained using both the two redox-independent sums of logratios (T o in Table 2) and the log-ratios H2 /H2O and CO/CO2 is possible only for the geothermal wells, whose deep temperatures were measured by physical methods. All the equilibrium temperatures are comparable with measured temperatures within some tens of degrees Celsius (Fig. 7). Only in the case of St. Lucia and Cagua, equilibrium temperatures deviate strongly from reservoir temperatures. This fact was already recognized for Cagua and attributed to inflow to the well of vapors from shallow levels (Giggenbach, 1993). Apart from these two exceptions, the different geothermometric approaches provide comparable results, as they are mainly controlled by the CO/CO2 ratio. 5. EVALUATION OF VARIABLES CONTROLLING GAS EQUILIBRIA

Based on analytical data, equilibrium temperatures and either s values or c values (estimated by means of two redox-independent sums of log-ratios), the RH and RC values as well as the log(XH2/XCO) ratio in the hypothetical (equilibrium) single saturated vapor phase were calculated for each sample. RC values are defined, by analogy with RH values, as RC 5 log rC 5 log (f CO/f CO2) > log (XCO/XCO2)

(50)

Obviously, for samples with s 5 0 and c 5 0 (i.e., pure equilibrium vapor and superheated vapors), analytical ratios are equal to RH, RC and log(XH2/XCO) values in the single saturated vapor phase. For samples with s . 0, RH, RC and log(XH2/XCO) values in equilibrium vapors are obtained by means of Eqn. 41. For samples affected by condensation, RH, RC and log(XH2/ XCO) values in equilibrium vapors are computed using Eqn. 47. Computed RH, RC, and log(XH2/XCO) values, which have been obtained by filtering the effects of secondary processes and referring analytical data to the same reference physical state of water, can now be used to investigate redox conditions and CO2 fugacities governing gas compositions in the considered hydrothermal systems. In the RH vs. RC plot (Fig. 8), hydrothermal vapors make up an alignment clearly different from that of high temperature (up to 700°C) crater fumaroles of Vulcano Island (Chiodini et al., 1993b, 1995), which are taken as representative of typical magmatic gases. As expected, redox potentials of these magmatic gases are controlled by the H2S-SO2 buffer (Giggenbach,

2684

G. Chiodini and L. Marini

Fig. 8. Plot of log(XH2/XH2O) vs. log(XCO/XCO2). Values of fumarolic vapors (circles) and vapors from geothermal wells (squares) are referred to the hypothetical (equilibrium) single saturated vapor phase. Codes are given in Tables 1 and 2. Triangles refer to crater fumarolic fluids of Vulcano Island. The H2S-SO2 magmatic gas buffer (Giggenbach, 1987), the FeO-FeO1.5 hydrothermal buffer (Giggenbach, 1987), the calcite-anhydrite buffer (Giggenbach, 1993) and the magnetite-hematite buffer are shown for reference.

1987). Hydrothermal vapors show instead more reducing conditions, which are intermediate between the FeO-FeO1.5 hydrothermal buffer (Giggenbach, 1987) and the calcite-anhydrite buffer (Giggenbach, 1993). The latter refers to the following reaction: CaSO4 1 CO2 1 4H2 5 CaCO3 1 H2S 1 3H2O

(51)

for f CO2/f H2S values of 10 and 100. The only exception is constituted by the Forgia Vecchia fumarole of Vulcano Island, whose redox conditions are evidently controlled by the H2SSO2 magmatic buffer, indicating that these are very immature, magmatic-derived vapors. Further details on redox potentials of hydrothermal vapors are obtained through Fig. 9, where only the systems represented by numerous samples are considered. RH and RC values are compared with several redox buffers. Those by Giggenbach (1987) and D’Amore and Panichi (1980) have been already introduced. Other redox buffers shown here are the fO2-function by D’Amore and Gianelli (1984), the mH2-functions by Arnorsson and Gunnlaugsson (1985) and C-CO2 buffer for fCO2 values of 1 and 10 bar (Giggenbach, 1993). The empirical f O2-buffer by D’Amore and Gianelli (1984) Fig. 7. Comparison between different equilibrium temperatures and measured reservoir temperatures for vapors discharged from geothermal wells. (a) equilibrium temperatures from Fig. 2 (f O2-buffer of Giggenbach, 1987); (b) equilibrium temperatures from Fig. 3 (fO2buffer of D’Amore and Panichi, 1980); (c) equilibrium temperatures from Fig. 8 (two redox-independent sums of log-ratios). Codes are given in Tables 2.

log f O2 5 23.808 213708.31/T 22074882/T2

(52)

can be related, for temperatures . 150°C, to different mineral assemblages, constituted by epidotes, chlorites, amphiboles, and pyroxenes of variable compositions, as well as K-feldspar, albite, quartz, and pyrite.

Hydrothermal gas equilibria

Fig. 9. Plot of log(XH2/XH2O) vs. log(XCO/XCO2). Values of fumarolic vapors and vapors from geothermal wells are referred to the hypothetical (equilibrium) single saturated vapor phase. Different hydrothermal redox buffers are also shown (dashed lines). The heavy solid lines refer to the linear regression equations for the hydrothermal systems (or provinces) which are represented by numerous samples.

The temperature functions of H2 concentrations by Arnorsson and Gunnlaugsson (1985) are log mH2 5 23.04 210763.54/T 1 7.003 log T

(53)

log mH2 5 11.98 1 0.08489 T 1 8254.09/T 2 27.587 log T (54) Equation 53 is for all waters above 300°C and waters in the range 200 –300°C if Cl . 500 ppm, whereas Eqn. 54 is for all waters below 200°C and waters in the range 200 –300°C if Cl , 500 ppm. Pyrite 1 epidote 1 prehnite 6 pyrrhotite 6 magnetite 6 chlorite may govern these redox buffers (Arnorsson and Gunnlaugsson, 1985). Corresponding RH and RC values are readily calculated considering the temperature dependence of the vapor-liquid distribution coefficient of H2, Eqn. 31 and log KWGS, Eqn. 27. Inspection of Fig. 9 shows that samples from different hydrothermal systems (or provinces) plot along different alignments, indicating that f O2-temperature relations are governed by different redox buffers in each system (or province). Samples from the Tuscan geothermal systems (Larderello, Travale, Amiata, Bagnore), which are hosted by sedimentary and metamorphic formations, closely fit the f O2-buffer by D’Amore and Gianelli (1984). This correlation is not too surprising as this buffer was derived considering these geothermal systems plus The Geysers and Cerro Prieto. All the other samples plotted in Fig. 9 come from active volcanic areas and distribute in a comparatively wide belt below the alignment of the Tuscan geothermal fields, that is towards more oxidizing conditions, which could indicate a minor magmatic influence. As a whole, the Campanian geothermal fluids (Solfatara, Ischia, and Vesuvio) and those from Vulcano (VSF and VI) spread around the

2685

f O2-buffer of D’Amore and Panichi (1980), while even more oxidizing conditions are indicated by Guagua Pichincha fumaroles. The samples from the fumarolic fields of Montserrat, which were collected 3– 4 years before the onset of the ongoing volcanic eruption (25 September 1995, when a lava dome began growing at the western foot of Castle Peak dome; Chiodini et al., 1996 and references therein), spread, instead, from the relatively reducing conditions of the Tuscan systems to the comparatively oxidising conditions of Guagua Pichincha fumaroles. These observations confirm that no unique f O2-buffer is able to explain the distribution of temperatures and redox conditions for all the hydrothermal environments. Many geothermal systems located in volcanic areas are in fact made up of both mature environments, where alteration mineralogy is characterized by coexisting K-feldspar 1 K-mica, and immature environments, where the presence of more acidic and more oxidizing fluids determine the development of peculiar mineralogic assemblages including kaolinite, alunite, and anhydrite (Giggenbach, 1993). Besides, the variability of redox conditions in the hydrothermal environments imply that CH4 could have attained chemical equilibrium with other gas species, as already recognized by Giggenbach (1980) for the fluids discharged from the geothermal wells of Wairakei, Kawerau, and Broadlands (New Zealand). Since CH4 is the least reactive member of a family of very unreactive molecules, i.e., the hydrocarbons, attainment of chemical equilibrium in the H2O-H2-CO2-CO-CH4 system likely requires catalytic assistance, perhaps by Fe oxides or other metal oxide systems, especially for T , 300°C. Although further investigations are necessary to understand the kinetics of the conversion of CO2 to CH4 in the presence of natural catalysts, first results indicate that reaction half-time are not particularly large (see above) and should not prevent equilibration of CH4 with other gas species in the H2O-H2-CO2-COCH4 system, at least for temperature of 200 –300°C or higher (Giggenbach, 1997). Computed log(XH2/XCO) ratios in the hypothetical (equilibrium) single saturated vapor phases were introduced in the following equation (adapted from Chiodini and Cioni, 1989), log f CO2 5 22.485 12248/T 2log (XH2/XCO) 1log f H2O

(55)

together with equilibrium temperatures and water fugacities in order to obtain theoretical CO2 fugacities. Results are plotted against equilibrium temperature in Fig. 10, which shows also the full equilibrium function of Giggenbach (1984, 1988) and f CO2, T values constrained by relevant thermometamorphic reactions. Most measured points either plot along the full equilibrium line or deviate from it less than one-half log f CO2 unit. Carbon dioxide fugacities in fumarolic fluids discharged from Mt Vesuvius crater, Ischia, and Larderello might be controlled by these thermometamorphic reactions, consistent with their geological-structural setting (e.g., Barberi and Leoni, 1980; Gianelli, 1985). There are also some systems characterized by exceptionally high f CO2 and comparatively low temperatures, such as Kizildere and Nagqu. Their f CO2 values are not internally controlled by mineral-solution reactions, but they act as traps for CO2 of external provenance (Chiodini, 1994), similar

2686

G. Chiodini and L. Marini

Fig. 10. Plot of f CO2 values calculated for the hypothetical (equilibrium) single saturated vapor phase vs. equilibrium temperatures. Symbols as in Fig. 2. Codes are given in Tables 1 and 2. Closed circles refer to geothermal reservoirs of Central Italy. The full equilibrium function of Giggenbach (1984, 1988) and f CO2, T values of relevant metamorphic reactions are also shown.

to some hydrothermal systems of Central Italy, whose data are reported for comparison. 6. CONCLUSIONS

The interpretation of steam discharges from twenty-two different hydrothermal systems, covering a wide range of volcanic and nonvolcanic environments, for which analytical data are available on H2O, H2, CO2, CO, and CH4 contents, allowed us to infer thermodynamic conditions at depth assuming chemical equilibrium among gas species and considering the influence of secondary processes, such as boiling and condensation. The following three different approaches to gas equilibria in the H2O-H2-CO2-CO-CH4 system were investigated: (1) use of fugacities of single gas constituents, which has no practical relevance as they cannot be measured in fumarolic discharges; (2) utilization of CO/CO2, H2/H2O, and CH4/CO2 ratios that are governed by three independent variables, such as temperature, redox potential and either water fugacity or the fraction of separated vapor; (3) use of two sums of log-ratios, that are controlled by two independent variables, such as temperature and either water fugacity or the fraction of separated vapor, and are independent of the redox potential. Theoretical values were plotted in suitable graphics, also showing the effects of secondary processes (steam separation and steam condensation), together with measured data (fumarolic fluids and vapors from geothermal wells). Analytical data spread throughout the theoretical grids in the log(XCO/ XCO2) vs. log(XH2/XH2O) plot, which is slightly affected by the choice of redox buffer. In the log(XCH4/XCO2) vs. log (XCO/ XCO2) plot, the field of saturated vapors overlaps that of superheated vapors, thus complicating the interpretation of measured data. In addition, a large number of analytical data plot outside the theoretical grid, using either the redox buffer by Giggenbach (1987) or that of D’Amore and Panichi (1980). As the CH4/CO2 ratio is strongly affected by redox conditions in the

gas equilibration zone, this disagreement between measured and theoretical values suggests that either no unique redox buffer is active in all the hydrothermal environments or that CH4 is not in equilibrium with the other gases. The diagrams that make use of two sums of log-ratios [e.g., log(XCO/XCO2) - log(XH2/XH2O) vs. 3log(XCO/XCO2) 1 log(XCO/XCH4)] are independent of redox conditions. In addition, the weight of CH4 on these functions is relatively small and these sums of log-ratios are mainly controlled by the CO/CO2 and H2/H2O ratios. Therefore these functions can be used as geoindicators that allow the equilibrium f O2 and f CO2 values to be calculated subsequently. Based on this data interpretation, the vapors of the considered twenty-two hydrothermal systems confirm that (1) most recalculated f CO2 values are consistent, within one-half log-unit, with the full equilibrium function of Giggenbach (1984, 1988) although thermometamorphic reactions might be active in some systems in agreement with geological evidence; (2) no unique f O2-buffer explains the distribution of temperatures and redox potentials in the considered hydrothermal environments. The latter finding can explain the apparently complex behaviour of CH4, whose fugacity is strongly affected by the redox conditions prevailing in the gas equilibration zone. In the considered hydrothermal systems, CH4 could have attained chemical equilibrium with other gas species, under redox conditions controlled by different f O2-buffers, consistently with previous findings by Giggenbach (1980) and with the available data on the kinetics of the conversion of CO2 to CH4 (Giggenbach, 1997). Acknowledgements—We wish to express our appreciation to Bill Evans for his helpful comments on the first version of the manuscript. We also wish to thank Roberto Cioni, who started to use CO for geothermometric-geobarometric purposes in hydrothermal and volcanic systems and encouraged us to follow him.

REERENCES Arno´rsson S. (1987) Gas chemistry of the Krı´suvı´k geothermal field, Iceland, with special reference to evaluation of steam condensation in upflow zones. Jo¨kull 37, 32– 47. Arno´rsson S. (1990) Gas chemistry of geothermal systems. In Geochemistry of Gaseous Elements and Compounds (ed. S. S. Augustithis), pp. 187–222. Theophrastus Publ. Arno´rsson S. and Gunnlaugsson E. (1985) New gas geothermometers for geothermal exploration - Calibration and application. Geochim. Cosmochim. Acta 49, 1307–1325. Arno´rsson S., Bjo¨rnsson S., Muna Z. W., and Bwire-Ojiambo S. (1990) The use of gas chemistry to evaluate boiling processes and initial steam fractions in geothermal reservoirs with an example from the Olkaria field, Kenya. Geothermics 19, 497–514. Atkinson P., Celati R., Corsi R., and Ramey H. (1977) Thermodynamic behavior of the Bagnore geothermal field. Proc. Larderello Work. Geothermal Resour. Assess. Reservoir Eng. ENEL-ERDA, 78 –131 Barberi F. and Leoni L. (1980) Metamorphic carbonate ejecta from Vesuvius Plinian eruptions: Evidence of the occurrence of shallow magma chambers. Bull. Volcanol. 43, 107–120. Barberi F. et al. (1987) The Island of Sumbawa: A major structural discontinuity in the Indonesian arc. Boll. Soc. Geol. It. 106, 547– 620. Battistelli A., Rivera R. J., D’Amore F., Wu F., Rossi R and Luzi C. (1991) A preliminary assessment of a medium-enthalpy geothermal resource in Nagqu (Tibet), People’s Republic of China. Proc. 16th Work. Geotherm. Reserv. Eng., Standford Univ. Bertrami R., Cioni R., Corazza E., D’Amore F., and Marini L. (1985) Carbon monoxide in geothermal gases. Reservoir temperature cal-

Hydrothermal gas equilibria culation at Larderello (Italy). Geotherm. Res. Council Trans. 9, 299 –303. Carella R. and Guglielminetti M. (1983) Multiple reservoirs in the Mofete Fields, Naples, Italy. Proc. 9th Work. Geotherm. Reserv. Eng., Standford Univ. Chiodini G. (1994) Temperature, pressure and redox conditions governing the composition of the cold CO2 gases discharged in north Latium (Central Italy). Appl. Geochem. 9, 287–295. Chiodini G. and Cioni R. (1989) Gas geobarometry for hydrothermal systems and its application to some Italian geothermal areas. Appl. Geochem. 4, 465– 472. Chiodini G. and Raco B. (1991) L’esplosione freatica di Ahuashuca (El Salvador): Rapporto sulla missione effettuata dal 18 al 19 Dicembre 1990. Internal report 2/91 IGGI–CNR. Chiodini G., Cioni R., Raco B., and Taddeucci G. (1991a) Gas geobarometry applied to evaluate phreatic explosion hazard at Vulcano Island (Sicily, Italy). Acta Vulcanol. 1, 193–198. Chiodini G., Cioni R., Raco B., and Scandiffio G. (1991b) Carbonyl sulphide (COS) in geothermal fluids: An example from the Larderello field (Italy). Geothermics 20, 319 –327. Chiodini G., Cioni R., Guidi M., Marini L., Raco B., and Taddeucci G. (1992) Gas geobarometry in boiling hydrothermal systems: A possible tool to evaluate the hazard of hydrothermal explosions. Acta Vulcanol. 2, 99 –107. Chiodini G., Cioni R., Leonis C., Marini L., and Raco B. (1993a). Fluid geochemistry of Nisyros island, Dodecanese, Greece. J. Volcanol. Geotherm. Res. 56, 95–112. Chiodini G., Cioni R., and Marini L. (1993b) Reactions governing the chemistry of crater fumaroles from Vulcano Island, Italy, and implications for volcanic surveillance. Appl. Geochem. 8, 357–371. Chiodini G. et al. (1994) Chemical and isotopic variations in the fumarole fluids of Guagua Pichincha volcano during the period 1988 –1992. Acta Vulcanol. 5, 137–139. Chiodini G., Cioni R., Marini L., and Panichi C. (1995) Origin of the fumarolic fluids of Vulcano Island, Italy, and implications for volcanic surveillance. Bull. Volcanol. 57, 99 –110. Chiodini G. et al. (1996). Fluid geochemistry of Montserrat Island, West Indies. Bull. Volcanol 58, 380 –392. D’Amore F. (1991) Gas geochemistry as a link between geothermal exploration and exploitation. In Application of Geochemistry in Geothermal Reservoir Development. (ed. F. D’Amore), pp. 93–117. UNITAR D’Amore F. and Celati R. (1983) Methodology for calculating steam quality in geothermal reservoir. Geothermics 12, 129 –140. D’Amore F. and Gianelli G. (1984) Mineral assemblages and oxygen and sulfur fugacities in natural water-rock interaction processes. Geochim. Cosmochim. Acta 48, 847– 857. D’Amore F. and Nuti S. (1977) Notes on the chemistry of geothermal gases. Geothermics 6, 39 – 45. D’Amore F. and Panichi C. (1980) Evaluation of deep temperature of hydrothermal systems by a new gas-geothermometer. Geochim. Cosmochim. Acta 44, 549 –556. D’Amore F., Fancelli R., Saracco L., and Truesdell A. (1987) Gas geothermometry based on CO content: Application in Italian geothermal fields. Proc. 12th Work. Geothermal Reservoir Eng. Standford University, 48 –53. D’Amore F., Rivera J. R., Giusti D., and Rossi R. (1990) Preliminary geochemical and thermodynamic assesment of the geothermal resources, Sulphur Springs area, St Lucia, W.I. Appl. Geochem. 5, 587– 604. Fournier R. O. (1979) A revised equation for the Na/K geothermometer. Geotherm. Res. Council Trans. 3, 221–224. Fudali R. F. (1965) Oxygen fugacities of basaltic and andesitic magmas. Geochim. Cosmochim. Acta 29, 1063–1075. Gianelli G. (1985) On the origin of geothermal CO2 by metamorphic processes. Boll. Soc. Geol. It. 104, 575–584.

2687

Giggenbach W. F. (1980) Geothermal gas equilibria. Geochim. Cosmochim. Acta 44, 2021–2032. Giggenbach W. F. (1982) Geothermal mineral equilibria. Reply to a Comment by M. A. Grant. Geochim. Cosmochim. Acta 46, 2681– 2683. Giggenbach W. F. (1984) Mass transfer in hydrothermal alterations systems. Geochim. Cosmochim. Acta 48, 2693–2711. Giggenbach W. F. (1987) Redox processes governing the chemistry of fumarolic gas discharges from White Island, New Zeland. Appl. Geochem. 2, 143–161. Giggenbach W. F. (1988) Geothermal solute equilibria. Derivation of Na-K-Mg-Ca geoindicators. Geochim. Cosmochim. Acta 52, 2749 – 2765. Giggenbach W. F. (1991) Chemical techniques in geothermal exploration. In Application of Geochemistry in Geothermal Reservoir Development. (ed. F. D’Amore), pp. 119 –144. UNITAR. Giggenbach W. F. (1993) Redox control of gas compositions in Philippine volcanic-hydrothermal systems. Geothermics 22, 575–587. Giggenbach W. F. (1997) Relative importance of thermodynamic and kinetic processes in governing the chemical and isotopic composition of carbon gases in high-heatflow sedimentary basins. Geochim. Cosmochim. Acta 61, 3763–3785. Grant M. A. (1982) On the lack of a unique relation between CO2 partial pressure and temperature in geothermal systems. Comment on Geothermal mineral equilibria by W. F. Giggenbach. Geochim. Cosmochim. Acta 46, 2677–2680. Guidi M., Marini L., and Principe C. (1990) Hydrogeochemistry of Kizildere geothermal system and nearby region. Geotherm. Res. Council Trans. 14, 901–908. Helgeson H. C., Knox A. M., Owens C. E., and Shock E. L. (1993) Petroleum, oil field waters, and authigenic mineral assemblages: Are they in metastable equilibrium in hydrocarbon reservoirs? Geochim. Cosmochim. Acta 57, 3295–3339. Henley R. W., Truesdell A. H., Barton P. B.Jr, Whitney J. A. (1984) Fluid-mineral equilibria in hydrothermal systems. Rev. Econ. Geol. 1. Keenan J. H., Keyes F. G., Hill P. G., and Moore J. G. (1969) Steam Tables. Thermodynamic Properties of Water Including Vapor, Liquid, and Solid Phases (International System of Units-S.i.). Wiley. Mahon W. A. J., McDowell G. D., and Finlayson J. B. (1980) Carbon dioxide: Its role in geothermal systems. New Zealand J. Sci. 23, 133–148. Marini L., Agostini A., Cioni R., Guidi M., and Leon O. (1991). Guagua Pichincha volcano, Ecuador: Fluid geochemistry in volcanic surveillance. J. Volcanol. Geotherm. Res. 46, 21–35. Naumov G. B., Ryzhenko B. N., and Khodakovsky I. L. (1974) Handbook of Thermodynamic Data. USGS Report WRD–74 – 001. Nehring N. L. and D’Amore F. (1984) Gas chemistry and thermometry of the Cerro Prieto, Mexico, geothermal field. Geothermics 13, 75– 89. Ryzhenko B. N. and Malinin S. D. (1971) The fugacity rule for the systems CO2-H2O, CO2-CH4, CO2-N2, and CO2-H2. Geochem. Int. 562–574. Ryzhenko B. N. and Volkov V. P. (1971) Fugacity coefficients of some gases over a broad range of temperatures and pressures. Geochem. Int. 468 – 481. Stull D. R., Westrum E. F., and Sinke G. G. (1969) The Chemical Thermodynamics of Organic Compounds. Wiley. Tedesco D. (1996) Chemical and isotopic investigations of fumarolic gases from Ischia island (southern Italy): Evidences of magmatic and crustal contribution. J. Volcanol. Geotherm. Res. 74, 233–242. Tonani F. (1973) Equilibria that control the hydrogen content of geothermal gases. Unpublished Phillips Petroleum Co. report.