Minerals in hot waterr · 2007-08-28 · volving mineral equilibria in superheated steam, Morey with respect to mineral solubilities (Morey, 1942), and Goranson (1938) with respect

American Mineralogist, Volume 71, pages 655473, 1986

Minerals in hot waterr

H.lss P. EucsrnnDepartment of Earth and Planetary Sciences, Johns Hopkins University, Baltimore, Maryland 21218

We are at an embryonic stage in our study of metal complex equilibria at high temperatures andPressures'

seward, 1981, p. 126

High-pressure/high-temperature solution chemistry . . . desemes far more experimental attention thanit has so far received.

Helgeson, 1981, p. 172

AssrRAcr

Two centuries ago James Hall initiated the study of mineral-fluid interactions at elevatedtemperatures and pressures, and interest in hydrothermal synthesis has expanded eversince. The solubility of minerals in supercritical steam is dictated largely by the propertiesof the solvent. With increasing temperature, the dielectric constant of water decreasessubstantially, enhancing the degree of association of dissolved acids, bases, and salts.Pressure has the opposite effect through electrostriction, the collapse ofthe solvent aroundthe hydrated ions. Metal complexing in supercritical chloride solutions depends on tem-perature, pressure, and chloride ion concentration. Dominant metal complexes can beidentified from conductivity, spectroscopic, or solubility measurements. In the supercriticalregion, step-wise association reactions are endothermic and driven by the large entropygain associated with the release of water from the structured environment to the bulksolvent. Because of decreased electrostatic shielding, metalJigand interaction is largelyelectrostatic and ligand-field stabilization eneryies are low.

Solubility studies on Fe, Sn, and W minerals in HCI-NaCI solutions can be used toidentifu the dominant solute species. With increasing temperature, low-charge and neutralspecies become more abundant, but higher pressures favor dissociation and high ligand-ion concentrations favor anionic species. Modeling of metal transport in hydrothermalsystems depends critically on speciation information, but reliable data in the supercriticalregion remain scarce.

INrnooucrroN

Are rocks and minerals made from fire or from water?Two hundred years ago this question was debated pas-sionately by Hutton in Scotland, the proponent ofvolcanicaction, and by Werner in Germany, who maintained thatall rocks were precipitated from aqueous solutions. Fromtoday's perception it is easy to acknowledge that bothviews contain substantial truths. More interesting and morechallenging is a search for common ground, a search thatincludes the interaction of minerals with aqueous fluidsat elevated temperatures and pressures, a topic that hasoccupied much of my professional life. For a young stu-dent exposed to the traditions of Paul Niggli and GeroldSchwarzenbach in Ziirich, this choice of topic was ob-vious. The path, however, has been neither short norstraight, and 35 years later I am just beginning to graspthe importance of the chemistry of hot aqueous solutions.If you feel that as president of MSA I should focus moreon minerals and less on their environment, permit me to

' Ad"pt d f-m the Presidential Address at the annual meetingof the Mineralogical Society of America, October 29, 1985, inOrlando, Florida.

0003{04xl86/0506{655$02.00 655

paraphrase the saying about famous men and their wom-en: "behind every important mineral there once was afluid."

In this discourse I will focus on the fluid and thoseproperties that are important for our understanding ofhydrothermal alteration, geothermal systems, ore depos-its, metamorphic reactions, igneous processes, and evenmantle degassing. After a brief historic review of hydro-thermal experiments I will discuss speciation and metaltransport in supercritical aqueous fluids, using data andinsights gained recently with my students and associates,especially Glenn Wilson (1986). In the end, I hope toconvince you that this frontier of geochemistry holds thekey to our understanding of mineral-formation processesfor a large portion ofthe earth.

150 vglns oF HyDRoTHERMAL syNTHEsrs: FnoprHar.r, ro Tutrr,n

Sir James Hall of Dunglass, Scotland, is generally ac-knowledged as the father of experimental geology. This isbased on three sets of experiments which he carried outbetween 1790 and 1825 designed to test Hutton's theories:(1) crystallization of fused dolerites and basalts as well as

656 EUGSTER: MINERALS IN HOT WATER

Fig. l. Synthesis ofcassiterite, achieved by passing SnCl, andHrO vapors tfuough a heated ceramic tube. From Daubr6e (1849).

lavas from Etna and Vesuvius, (2) recrystallization of chalkto marble in heated, sealed gun barrels and (3) compres-sion of layers of clay to imitate folding and faulting. Myaccount of Hall's work is taken from Geikie's book TheFounders of Geology (1905), in turn based on the firstGeorge Huntington Williams lectures delivered at JohnsHopkins University in 1896. For our present topic, Hall'ssecond experiment is most significant. Hutton had con-cluded that basalts, dolerites. and other members of the"whinstone" family had formed by fusion, and yet manyof them, especially their vesicles, contained calcite. Howcould calcite retain its CO, during fusion? Hutton pro-posed that pressure was the answer, and Hall, his friendand admirer, set out to test this suggestion. He enclosedchalk powder in gun barrels and exposed them to thehighest temperatures obtainable in a glass factory. CO,was not released during the experiments, and a substanceclosely resembling marble was produced. The results vin-dicated Hutton's conjecture and were published in 1805as an "Account of a Series of Experiments Showing theEtrects of Compression in Modifying the Action of Heat"(Hall, 1805).

Hall had no disciples, but his methods and approachesinspired a group of French mining engineers and geo-chemists, especially Daubr6e, S6narmont, and Friedel whocame to dominate experimental petrology until the endof the century. Daubr6e, an outstanding experimentalist,covered as wide a range of topics as Hall did, includingrock deformation and meteorites. His first paper (Dau-br6e, 1841) dealt with tin deposits he had visited in Sax-ony and Cornwall and the importance of the "mineral-izers" B, F, PO4, and Cl. His famous synthesis of cassiteriteconsisted ofpassing "SnClr" and water vapor through aheated ceramic tube at atmospheric pressure (Fig. l): "Ifone passes through a ceramic tube heated to orange-whitecolor two streams, one of tin chloride and the other ofwater, the mutual decomposition to stannic acid and hy-drochloric acid is achieved with great ease" (Daubr6e,1849, p. l3l-r32).,

Quartz, TiOr, and apatite were obtained in a similar

2 Translations by the author.

Fig.2. Autoclaves used by S6narmont (1851) and Daubr6e( I 8 5 7) in their hydrothermal experiments. From Daubr6e ( I 8 79).

manner. S6narmont (1851) followed Hall's lead and syn-thesized a large number of minerals, including quarlz,hematite, siderite, magnesite, stibnite, pyrite, sphalerite,chalcocite, and others by the "wet method" in sealed gunbarrels, in other words, hydrothermal synthesis. He con-cluded that "ore veins are nothing else but immense ca-nals, more or less obstructed, through which once passedwaters capable of incrustation. These disgorging liquidsand gases must be comparable to those ofgeysers and hotsprings" G,. 132). Of his experiments he demanded that"a chemical synthesis which reproduces a single mineralspecies contemporaneous with a particular ore vein mustbe applicable at the same time to all other minerals pres-ent" (p. 130).

Starting materials were contained in sealed, evacuatedglass ampules filled with water to 500/o of their volume,which were placed in brass gun barrels also containingwater. One end of the barrel was soldered shut with a steelplug, whereas the other was sealed with a threaded nutand art annealed copper disc. The barrels were placed atthe top of the gas furnace at the foundry of Ivry andcovered with coal dust and heated to 200-300'C. S6nar-mont (1851) warned of the possible explosions "whichare frequent and of great violence, in spite of the smallquantity of liquid" (p. 136).

Daubr6e (1857) followed similar procedures in his hy-drothermal experiments, except that he used steel bombs.Figure 2 shows two of these bombs. Feldspar + quartzwere produced by heating kaolin with water from thePlombidres hot springs for 2 days.

Daubr6e also mentions diopside, wollastonite, and pec-tolite. These and later experiments involving zeolites weresummarized in his book (Daubr6e, 1879). Friedel and

EUGSTER: MINERALS IN HOT WATER 657

Sarasin (1879) initiated their extensive hydrothermalstudies by synthesizing orthoclase from potassium silicateand AlClr. The early history of hydrothermal synthesiswas summarized by Doelter (1890), Niggli and Morey(1913), and Morey and Ingerson (1937).

Toward the end of the century two other sets of famousgeochemical experiments were carried out: those of Van'tHoffon marine evaporites and the studies by Vogt (1904)on magmatic differentiation. Although not concerned withhydrothermal conditions, Van't Hof and his colleaguescarried out the first systematic series of experiments de-signed to solve a petrologic problem (see Eugster, 1971,for a historical account). With the founding of the Geo-physical Laboratory in Washington in 1905, the center ofexperimental petrology moved to the USA, where it hasremained ever since. Between the two world wars, Moreyand Goranson made the most notable contributions in-volving mineral equilibria in superheated steam, Moreywith respect to mineral solubilities (Morey, 1942), andGoranson (1938) with respect to silicate melting at highwater pressures. After World War II, hydrothermal ex-perimentation grew rapidly to encompass all geochemicalaspects, spurred by the technical innovations of O. F.Tuttle (1948). The Tuttle press consists of a hot cone-in-cone seal and was used for the classic papers on the systemMgO-SiOr-HrO (Bowen and Tuttle, 1949) and on feldsparequilibria (Bowen and Tuttle, 1950). Because of leachingproblems, it has been replaced by the use of welded noble-metal tubes placed in cold-seal rod bombs, also a Tuttleinvention. The study by Tuttle and Bowen (1958) on thesystem granite-HrO remains a landmark in experimentalpetrology. I added procedures for controlling redox re-actions in hydrothermal systems (Eugster, 1957), Hemley(1959) focused on acid-base reactions, whereas Franck(1956) measured dissociation equilibria at high P and Tusing conductivity methods. These techniques have re-mained essential for hydrothermal studies to this day. Theexplosive growth of experimental petrology during thefifties emanated largely from the Geophysical Laboratory,Harvard University, and Pennsylvania State University.With the successful synthesis of diamond by General Elec-tric, the focus shifted to manfle mineralogy and ultra-highpressures, relegating hydrothermal research to a lesser role.Nevertheless, studies of mineral solubilities at severalhundred to several thousand bars water pressure and tem-peratures from the subcritical to the supercritical rangehave remained the continuing con@rn of a small numberof experimentalists. Examples are the classic investiga-tions of Kennedy on quartz solubility (1950) and the ex-ploratory work of Morey and Hesselgesser (1951) on thesolubilities of a variety of silicates. Geochemists, econom-ic geologists, and metamorphic petrologists interested inthe transport of mineral matter and the growth and dis-solution of minerals have relied heavily on these and sub-sequent contributions. The recent literature has been sum-marized by Holland and Malinin (1979) and Barnes (1979).Here I would like to point out that many of the funda-mental problems of mineral-fluid interaction remain un-

200 4 0 0 6 0 0 8 o o o c

Fig. 3. Density, p (in g/cm3), of water vapor as a function ofP (kbar) and I("C). Also shown are the L + V curve and criticalpoint. From the data ofBurnham et al. (1969).

solved and must be dealt with if we ever hope to under-stand geothermal systems, hydrothermal ore deposits, andmetamorphic processes. Most of the technical difficultieshave been overcome. but we remain faced with an areaof geochemistry in which correspondingly few data havebeen collected and in which predictions far outstrip facts.

Vnnv Hor wATER

Minerals dissolve in aqueous fluids through the for-mation of one or more aqueous species in the form ofsimple ions or complexes. The nature of these speciesdepends on pressure, temperature, and the compositionof the fluid. Knowledge of the nature and concentrationof the most abundant species is essential if we must knowwhether a fluid is supersaturated, saturated, or undersat-urated with respect to a particular mineral, that is, whetherthe fluid is capable of precipitating or dissolving that min-eral. Kinetic studies of mineral-fluid interaction also de-pend on speciation information. Speciation questions havebeen dealt with extensively with respect to surface waters,ground waters, and hot-spring waters. For those appli-cations a large body of information is available (see, forinstance, Stumm and Morgan, 198 I ; Truesdell and Jones,1974; Henley et al., 1984). In contrast, no more than afew dozen publications contain data on near-critical andsupercritical speciation, foremost among them the con-tributions of Franck (1981), Quist and Marshall (1968),Seward (1976, 1984), Frantz and Marshall (1984), andCrerar et al. (1978). Helgeson and coworkers (1981) havemade significant progress in predicting solute species andtheir thermodynamic properties over a wide range of Pand T, but experimental calibration is largely lacking.Melal-complex formation at elevated P arrd T was dis-cussed recently by Seward ( I 98 I ) and by Susak and Crerar(1985). Most data were obtained by spectroscopic meth-ods at the pressure of the liquid * vapor equilibrium, andhence conclusions are applicable particularly to lhe sub-critical low-pressure region. Here I would like to empha-

k b


kb

2 0 0 4 0 0 600 0c

Fig. 4. Dielectric constant, e, ofwater vapor as a function ofP(kbar) and I("C). Compiled from earlier databy Seward (1981).For more recent data, see Pitzer (1983) and McKenzie and Helge-son (1984).

size the properties of supercritical fluids, where pressureplays a more dominant role. This is evident from thedensity of water vapor as a function of P and Z (Fig. 3).Density changes with pressure can be substantial, partic-ularly at the higher temperatures.

The remarkable ability of water to disassemble a tightlybonded ionic crystal such as NaCl (melting point, 800.5eC!)is based on its polar nature-its ability to insulate positivefrom negative charges by surrounding them with waterdipoles. In the hydration shells surrounding the ions, watermolecules are more ordered and more densely packedthan in the bulk solvent. In general, anions are larger thancations and hence less hydrated because of the more dif-fuse localization oftheir charge. The solvating ability ofwater is reflected in its large dielectric constant, e, whichunder ambient conditions exceeds 80, that is, the presenceof water decreases an imposed electric field 8O-fold.

Pressure and temperature profoundly affect the solventproperties ofwater as expressed by changes in e (see Fig.4). With increasing temperature, e decreases substantiallybecause of the greater kinetic energy of the dipoles. Con-versely, e increases with increasing pressure at constanttemperature because of the decrease in volume associatedwith hydration. In the supercritical region ofinterest here,e is 25 or less. For a recent summary see Pitzer (1983).

In the hydrothermal region, the dissociation constantofwater, K*, varies overfourorders ofmagnitude, roughlyfrom l0-t3 to l0-e (see, for instance, Marshall and Franck,l98l). These changes with Pand Zare due to changes inhydrogen bonding of the water molecule, the dielectricconstant of water, and the partial molal volume of thesolvated ions. Figure 5 shows K* as a function of waterdensity, pnro, with isotherms and isobars contoured. Iso-baric temperature increases initially lead to increased dis-sociation ofthe solvent, largely owing to a decrease in thehydrogen bonding. Eventually, however, the decrease inthe dielectric constant becomes dominant, and the iso-therms pass through a maximum, leading to increased

1 . 0 . 8 . 6 . 4o ̂ zo

Fig. 5. Dissociation constant for water vapor, K*, as a func-tion of density, p rro. Isotherms ('C, solid lines), isobars ftbar,lighter dashed lines), and the L + V curve (heavy dashed line)have been contoured. Calculated from various data (see Marshalland Franck, 1981) by Eugster (1981).

association of the solvent. The temperatures at which theK* isobars of Figure 5 pass through a maximum increasewith increasing pressure, in agreement with the effect ofpressure on e illustrated in Figure 4. The behavior of theisotherms in Figure 5 are more straightforward. For everytemperature considered, K* increases monotonically withincreasing pressrue, because of electrostatic collapse of thesolvent around the ions. As expressed by Ritzert andFranck (1968, p. 805), "at all temperatures ion dissocia-tion is favored by compression. This is based on the denserpacking of the water molecules in the hydration shells ofthe ions." The effect ofa pressure increase is greatest atthe higher temperatures, particularly above the criticaltemperature, a fact illustrated by the P-?" dependence ofe illustrated in Figure 4. Above 400t, a pressure changeof as little as I kbar causes a dramatic shift in e.

Vrnv nor soLUTEs

The behavior ofthe solutes also is governed by changesin the dielectric constant and by electrostriction. For iso-baric temperature increases, e decreases, favoring associ-ation of the solutes. Conversely, an isothermal increasein pressure favors dissociation of the solutes caused bythe rise in e as a consequence of the improved hydrostaticshielding. The response of strong electrolytes is best il-lustrated by NaCl, for which data are available from Quistand Marshall (1968), illustrated in Figure 6. K*"", de-creases with temperature and between 300 and 400'C,depending upon water pressure, reaches a value of one,that is, roughly half of the Na and Cl is present as un-charged NaClo molecules:

NaClo + Na* * Cl- (1)

INa-][Cl-]AN"cr : tNacr.f

.


o k b

200 400 600 0c

Fig. 6. Dissociation constant of NaCl in water vapor as afunction of temperature (',C). Isobars have been contoured inkilobars. From the data of Quist and Marshall (1968).

Toward higher temperatures, K*"", continues to decreasedramatically, and above 500'C, NaCl is largely associated.As with the solvent, an increase in pressure favors dis-sociation owing to the tighter packing of the HrO mole-cules around Na* and Cl-, increasing the value of K*".,.

The data of Figure 6 were obtained from conductivitymeasurements in dilute solutions. At high salt concentra-tions, conductivities decrease to about one-third that indilute solutions, and there is little temperature depen-dence between 300 and 600'C (Hwang et al., 1970; Klos-termeier, 1973). Franck (1981, p. 73) concluded that "itis not yet possible to decide to what extent this diminutionofcharge carriers can be attributed to well defined asso-ciations of salt molecules, ion-pairs or triple ions."

Clearly, the molal K".", values of Quist and Marshall(1968) cannot be used for concentrated solutions unlessactivity coefficient corrections are made. In dilute solu-tions. the concentration offree chloride ions can be cal-culated as a function of P, T, and total chloride molality,

- 1

I

6 - 2

oCt

uoo/./'uo9

600-

109

l o g ( C t ) t o t

Fig. 7. Molality of free chloride ions as a function of totalchloride concentration in NaCl-HrO solutions at I -kbar pressure.Isotherms are contoured in C.

4oo soo 600 oc

Fig. 8. Dissociation constant of HCI as a function of tem-perature ("rC). Pressure is contoured in kilobars. From the dataof Frantz and Marshall (1984). For comparison, the NaCl dataofFigure 6 are shown as dashed curves.

(/n..),o,, by combining Equation I with the electrical neu-

trality condition (Eq. 2). For a solution of NaCl, we have

[Na*] + [H.]: [Cf] + [OHl, Q)

which simplifres to

[Na.] t [Cl-]

for a solution with a near-neutral pH and a total chloridemolality of at least l0-4. Results for a constant pressureof I kbar are shown in Figure 7. As the temperature in-creases, the concentration ofchloride ions and with it theionic strength, 1, drop dramatically, reflecting increasedassociation. On the other hand. the Cl- concentration in-creases with increasing total chloride. An increase in pres-sure would raise the curves of Figure 7 toward higher Cl-values. The diference between the molality of free chlo-ride ions and the total chloride molality of the solutionis largest in the supercritical region at moderate pressures.Here fluids rich in total chloride do not have many free

Y

z - 2YCD

-4

I

o

o- 3

- 4- 1- 2

400 5oo 600 oc

Fig. 9. Molality of free chloride ions for a 0.1-molar solutionof HCl-HrO as a function of temperature (oC). Pressure is con-toured in kilobars.


Vfr-

Fig. I 0. Stepwise association ofCd ions at 25oC as a functionofincreasing cadmium iodide molality. From Robinson and Stokes(re5e).

chloride ions. This distinction is crucial, because the for-mation of metal-chloride complexes depends on the ac-tivity of the chloride ions and not the total chlorinity ofthe solutions. In this region of dominant association, NaClceases to be the strong electrolyte it is at room tempera-ture. Because it is based on changes in the solvent, thisfate is shared by strong electrolytes in general: "as onemoves into a low water-density and low dielectric-con-stant domain, a lot of solutes that we call strong electro-Iytes become weak electrolytes" (Pitzer, 1981, p. 175).

HCI behaves in a similar fashion. Figure 8 is based onthe dissociation data of Frantz and Marshall (1984)

Kncr : lH.llcl-ltHcrl

A comparison with Figure 6 shows that HCI is somewhatmore associated than NaCl at the same P and 7. In thesupercritical region, K". is two to three orders of mag-nitude smaller than K*"",. In fact, the 2-kbar isobar forK"", coincides with the I -kbar K""., isobar. Figure 9 showsthe molality of chloride ions for a solution of 0. I mol(HCl + Cf) as a function of temperature. Isobars from0.5- to 4-kbar water pressure are contoured. For similarP, T, and total chloride molality, supercritical HCI solu-tions contain even fewer free chloride ions than do NaClsolutions.

Dissociation data for other supercritical electrolytes areless complete, but the general trends with P and Zare thesame, since they are imposed by changes in the solvent.Strong acids, strong bases, and strong salts become weakacids, weak bases, and weak salts. Similarly, electrolytesthat are weak at ambient temperatures become fully as-sociated and are present dominantly as uncharged mol-ecules. Consequently, ionic strengths of concentrated so-lutions are much lower than they are at room temperature.Because ofthe decreased electrostatic shielding in the low-density-low-dielectric-constant region, interactions be-tween metal ions and ligands are largely electrostatic andhard in terms of Pearson's (1963) classification (see alsoCrerar et al., 1985). Ligand-fietd stabilization eneryies(LFSEs) in this region are greatly diminished.

A g C l .

l o g C l -

Fig. 11. Schematic drawing of the stepwise complexing of Agions as a function of chloride ion activity in equilibrium withcerargyrite. After Seward (1976).

Another important efect of the low ionic strengths ofconcentrated solutions at pressures and temperatures whereassociation predominates is related to activity coefrcients.In high-ionic-strength electrolyte solutions at room tem-perature, activity coemcients have been modeled suc-cessfully by adding two virial coemcients, the Pitzer coef-ficients, to the Debye-Hiickel term (Harvie and Weare,1980). We have used this method to predict the effects ofevaporating seawater (Eugster et al., 1980) and to definethe phase relations and brine compositions in the systemNa-K-Ca-Mg-Cl-SO4-HrO (Harvie et aL., 1982). Harvieet al. (1984) have extended the approach to include car-bonate and bicarbonate ions. At high temperatures, neu-tral complexes predominate, and such complexes can beassigned unit activity coefficients (Helgeson et al., 1981).For the charged complexes, a Debye-Hiickel treatmentmay be adequate, because of the lesser importance ofshort-range order. Debye-Hiickel constants as a functionofPand l'have been presented by Helgeson and Kirkham(1976). The definition ofi, the radius ofclosest approach,remains to be clarified for each individual ion (see Helge-son et al., l98l). Nevertheless, activity-coefrcient correc-tions in concentrated solutions at high P and Tare not asdifficult as they are at room temperatwe.

Errucr oF LTcAND coNcENTRATToN

Stepwise formation of metal complexing takes place ata given P and Zin response to increasing ligand activity.This has been demonstrated at room temperature, forinstance, for cadmium iodide, where the following se-quence occurs as a function of iodide molality (see Fig.l0):

oo,

q,o

(3)

C d 2 * - C d I * - C d I r - 6 6 1 - .

EUGSTER: MINERALS IN HOT WATER 66r

1 80c

1 5 0 0 5 0

2 7 7 0

l o g C l -

Fig. 12. Abundances of individual silver-chloro complexesas a function of total chloride concentration and temperature("C). After Seward (1976).

In chloride solutions a similar situation has been re-ported by Seward (1976) for the case of silver chloridecomplexes, for temperatures from l8 to 353'C at pressuresof the L + V equilibrium. With increasing 4cr-, achievedby adding NaCl, the following steps were observed:

Ag* - AgCl - AgClt - AgCl3- - AgCl?-.

These steps are illustrated schematically in Figure I l, withthe silver concentration in solution controlled by the sol-ubility of cerargyrite, AgCl. For a given temperature, thesolubility curve can be interpreted as the sum ofseveralchloride complexes which change in their abundance withchanging NaCl concentrations. Seward's (1976) computerfit to his data is reproduced in Figure 12, with each com-plex identified by its ligand number. Figure 12 illustratesthe simplification in speciation that takes place with risingtemperature through increasing association. Below 100oC,

Table l. Enthalpies (AI1, in kJlmol) of stepwise association forsilver and lead-chloro complexes (from Seward, 1976,1984)

25 0C

1 0 0

%

o

1 0 0

%

5 0

200 0

3 o o o

1 0 0

%

5 0

0

l o g C l -

Fig. 13. Abundances ofindividual lead-chloro complexes asa function of total chloride concentration and temperature ("tC).After Seward (1984).

all five species are found, whereas above 300'C, only AgCland AgClr- are abundant. It should be remembered thatat 350"C at the L * V curve, K*""' is 0.01-that is, for atotal chloride molality of 0.1, the molality of the freechloride ion is only 0.026. At higher pressures, for thesame temperature, larger values of [Cl-] can be reachedby adding NaCl, presumably resulting in the formationof AgCll- and AgCll-, even at 350'C and higher temper-atures. Further experiments are necessary to check thisconjecture.

Seward (1984) also investigated the stepwise formationof lead chloride complexes to 300'C, but using spectro-photometric methods. At 25'C, five species of the typePbCl;-" were found with n : 0, l, 2,3, and 4, whereasat 300'C only PbCl*, PbCl! and PbCI; were abundant.The stepwise sequence in response to increasing Cl- istypical for a bivalent cation:

Pb2* * PbCl* - Pbclg - PbCl;' PbCU- + ....

A comparison of the 25 and 300'C results is shown in

Table 2. Entropies (AS, in J/K.moD) of stepwise association forsilver and lead-chloro complexes (from Seward, 1976,1984)

1 0 0

%

0

1 0 0

%

1 0 0

%

5 0

- , -Ascl tgcl,2 egct!- plcl- pbct2 Pbcl

3

- ? -AscI Lgcr2 eecti- Pucl- PbcI2 Pbcl -

2 5 - 1 2 . 5 - 1 7 . 6 - t 7 , 4 2 . 4 6 1 2 . 3 - r 1 . 5 52

58

130

35

l l 0

r90

20 -22 -57

55 38 -36

1 3 0 4 3

280 170

2 5

1 5 0

250

3 5 0

- 5 5

9 . 5

60

t 50 0

250 36 .6

3 . 8 - 9 , 6 2 9 . 9 r 5 . 5 7 . 1

4 . 8 6 9 . 1 4 9 . 5 3 I . 8

7 7 . 91 5 1

662 EUGSTER:MINERALS IN HOT WATER

oc

400

300

200

1 0 0

2 4 6

m N a G l

Fig. 14. Coordination change from sixfold to fourfold for Fe,Co, Cu, and Ni as a function of temperature and total chloridemolality for pressures ofthe L + V curve. After Susak and Crerar(1 985) .

Figure 13, confirming the general trends exhibited by sil-ver chloride complexing. Seward (1976, 1984) has ex-tracted thermodynamic parameters from the equilibriumconstants for the silver and lead complexes. He found thatwith rising temperatures, enthalpies (see Table l) and en-tropies (see Table 2) become increasingly positive. At el-evated temperatures, complexing reactions between thesemetals and chloride are all endothermic, that is, heat isrequired to replace the water molecules of the hydrationshell with chloride ions. For instance. for an octahedralcomplex we have

(Pb(H,O)Jr. + Cl- + (pbc(H,o)s)- + H'o. (4)

Because the water is released from the structured envi-ronment of the complex into the highly disordered bulksolvent, a large entropy increase results, decreasing thefree energy and stabilizing the complex. This situation isanalogous to dehydration reactions during progressivemetamorphism, which are driven largely by the entropygain of the water liberated from the crystal structure. AsTable 2 shows, for a given complex the entropy increaseswith increasing temperature because of the increasing dis-order in the bulk solvent. Furthermore, "the large positiveentropies also imply that inner sphere complex formationbecomes more important at higher temperatures" (Sew-ard, 1984, p.129).

On the other hand, as the number of chloride ligandsincreases in the complex, both Afl and A,S decrease. Thisindicates a decrease in solvation linked to the decrease ofpositive charges (Seward, 1984, p. 129).

All ofthe data collected by Seward (1976, 1984) referto the relatively low pressures of the L + V equilibrium,200 bars or less. Increasing pressure, particularly in thesupercritical region, will have a pronounced efect by sta-

bilizing the more hydrated, lower-ligand species, increas-ing positive charges and degree ofsolvation. The effect ofpressure is thus opposite to the effect ofan increase in thechloride ion activity.

Pressure has a profound influence also on the coordi-nation number of complexes, though data on this pointare scarce in the supercritical region (see Seward, 1981).Increasing temperature and chloride ion concentration fa-vor transition from octahedral to tetrahedal coordination.Figure 14 is taken from Susak and Crerar (1985) andshows data for the VI + IV transition for Fe, Co, Cu, andNi as a function of Tnand total chloride molality. Pressweis that of the L + V equilibrium. Lowering the coordi-nation number decreases the charge on the inner-spherecomplex which is favored in a low-dielectric-constant en-yironment where coulombic attraction between metal ionsand ligands is dominant. Pressure, however, stabilizes thehigher coordination state, because it increases e, favorsdissociation and solvation, and by electrostriction leadsto a volume decrease. This has been documented for Niand Co at 500qC and 6 kbar by Liidemann and Franck(1968). In the supercritical region, where water densitiesare comparatively low (see Figure 3), pressure will havea particularly pronounced effect in stabilizing 6-fold co-ordination. Although not yet reported, higher coordina-tions of8 or 12 are possible.

Spncr,c,troN AND sol,uBrl,rry DETERMINATToNS ATHIGH P AND T

In general, speciation information is derived from con-ductivity, solubility, or spectroscopic measurements. Inthis section, preliminary to discussing our data on Fe, Sn,and W and the solubilities of magnetite, cassiterite, andwolframite, the experimental procedures and data analysismethods used in those studies will be summarized briefly(see also Wilson, 1986; Eugster and Wilson, 1985; Eugsteret al., 1986).

Mineral reactions involving HCI have been measuredsuccessfully using the Ag-AgCl buffer of Frantz and Eugs-ter (1973). The assemblage Ag + AgCl in conjunctionwith an oxygen buffer controls/"", at P and T (Fig. l5c).This is necessary for reactions that consume or releasesignificant amounts of HCl. However, because it restrictschloride ion molality to a single value for a given P andf, it is not a suitable method for speciation studies thatdepend on as large a range of Cl- molality as possible. Toovercome this limitation, Frantz and Popp (1979) usedthe "unbuffered" method, later termed the "modified Ag-AgCl buffer" (Boctor et al., 1980), shown in Figure l5b.The assemblage Ag + AgCl is placed with the charge, butno oxygen buffer is used. Instead, hydrogen sensors (Chouand Eugster, 1976) arc placed inside, which monitor/"r.This, in conjunction with Ag + AgCl in the charge system,allows calculation of/ro at P and 7. Because /o, is notcontrolled, the chloride ion molality can be varied inde-pendently. Using this method, Frantz and Popp (1979),Popp and Frantz (1979, 1980), and Boctor et al. (1980)determined speciation in chloride fluids for Mg, Ca, Na,


H M

N N O+ H 2 o B

Ag-AsC l

and Fe. The data for Mg and Ca were later retracted (Frantzand Marshall, 1982).

For metals that may have a valence in the fluid differentfrom that in the solid, the "modified Ag-AgCl buffer"method cannot be used. In such cases,62 must be strictlycontrolled. Chloride concentration must be fixed by thebulk composition of the fluid, and speciation is evaluatedfrom the chemical analysis after quench by calculation ofthe relevant equilibria back to P and T (Wilson, 1986;Wilson and Eugster, I 984, I 98 5; Eugster et al., I 986). Theexperimental arrangement is depicted in Figure l5a. Totalchloride molality and, through it, chloride ion molality atP and T are varied by adding diferent amounts of HCl,NaCl, or other soluble chlorides to the starting solution.To maximize solution volume, the oxygen bufer is placedin the smaller, inside tube, commonly 3.0-mm diameterPt, whereas the charge is placed in the larger, 4.4-mm Autube. After quench in a rapid-quench vessel (Rudert etal.,1976), the tube is opened in a glove box purged withwater-saturated argon to prevent oxidation and evapo-ration. The pH is measured with a microelectrode, andaliquot solution samples are extracted for analysis. Chlo-ride concentration is measured on the chloridometer or,preferably, the ion chromatograph. Metals are determinedby AA, ion chromatograph, or other appropriate methods.For each experiment, P, T, and fo2 are known. We alsoknow the mass balances for chloride and for each of themetals, as well as the electrical neutrality condition forambient P and T. For a single metal of valence x whichis fully dissociated, we have

x[Me*+]t,zs * [H+1t'zs: [Cl-]t 'x + [OH-1t'zs. (5)

Quench is very rapid, carried out at pressure, and oxi-dation is not possible. We have not observed evidence ofmineral precipitation during or after quench. Consequent-ly, the metal valence at P and T must also be x. Assumingthe presence of a single dominant metal-chloride complexwith the ligation number n, the electrical neutrality con-dition at P and lis

[H*p. + (n - n)[MeCl;-,pr: [Cl-]P,r + [OH-r.r. (6)

The chloride mass balance is given by

Cl.,: [Cl-]a" + [HC1pr + n[MeCl;-'y'r (7)

H M + H 2 O

m i n e r a l

+Ag-A9C l

+

H2O + HCI

m i n e r a l

+

Ag-AgCl

+

and the metal mass balance by

Me,o, - [MeCl;-,p' (8)

with Cl., and Me,o, obtained by chemical analysis afterquench. The unknown quantities that must be calculatedfor each experiment, assuming a single dominant mono-nuclear metal-chloride complex, are [H*]P'r [Cl-pt[Hctop' and [MeCl;-'].'. At this point, x is known, butn is not. In addition to Equations 6, 7, and 8, the disso-ciation constant for HCI is also known for Pand T (Frantzand Marshall, 1984)

K".':Iry. (e)tHcll

No additional independent relationship is available, andthe five unknowns cannot be solved for directly. However,we know that n must be a small integer of value 0, l, 2,3, or 4, and following Wilson (1986), we solve Equations6-9 hr each experiment, assuming a value for n and com-paring the results with the predicted slope. For instance,the oxide of a bivalent metal should dissolve in a chloridesolution by a reaction such as

Meo + 2Il* + nCl- + MeCll-" + Hro (10)

with an equilibrium constant lKro of

log K,o : log [MeCl?-"1 + 2 pH - n log [Cl-] (l l)

for a standard state of pure solids and pure HrO at P andT. lf n was chosen correctly, plotting log [MeCl]-'] + 2pH vs. log [Cl-] for each experiment should produce astraight line of slope n and intercept of log K'0. Compar-ison ofthe five or so plots will quickly reveal the properchoice for n, provided the scatter in the experimentaldatais not excessive.

The one-species assumption leads to less equivocal re-sults than the "variation of slope" method traditionallyused in speciation determinations (see, for instance, Sew-ard, 1976; Crerar and Barnes, 1976; Ftantz and Popp,1979). Wilson (1986) has extended the calculations toinclude two or three dominant complexes, using least-squares procedures instead of the visual trial-and-errormethod described here. Wilson and Eugster (1985) suc-cessfully tested the method on their cassiterite-solubility

+H20

+ H203m HCI

Fig. 15. Experimental arrangements for supercritical speciation determinations from mineral solubilities. (A) Method of Wilson(1986). See also Wilson and Eugster (1984, 1985), Eugster and Wilson (1985), Eugster et al. (1986). (B) "Unbuffered" or "modifiedbuffer" method ofFrantz and Popp (1979) and Boctor et al. (1980). (C) Ag-AgCl bufer method ofFrantz and Eugster (1973). Fordetails see text.


data as well as on the talc data ofFrantz and Popp (1979).Ifpolynuclear species are present, however, solubility dataalone cannot yield unequivocal speciation data.

The experimental procedures and data analysis de-scribed here are applicable to ligands other than chloride,such as (OH-), (HS-), (HCO;). They do depend, however,on the availability of reliable data for the dissociationconstants of the relevant acids, bases, and salts. For in-stance, if Reaction l0 proceeds in the presence of NaCl,Equations 6 and 7 must be modified as follows:

[H*p' * [Nx+]r.r + (x - n)lMeCli-,|'r :

lCll'.' + [OH-1e.r (6a)

and

Cl . , : [Cl - ]e '+ [HC11e'+ [NaCl]r'r + n[MeCl;-']". (7a)

The additional information obtained from

YrFe3Oo + 2H* + nCl- + FeCl?-, + HrO + y6O, (14)

for n: 2. The y intercept is 7.4, representing log K,o, andit is consistent with the results ofChou and Eugster (1977).Consequently, magnetite dissolution at this P and ?'canbe written as

%Fe3Oo + 2H* + 2Cl- + FeClr,", + HrO + luO} (l4a)

At 650'C, both n : 2 andn : 3 produce plots compatiblewith the experiments available, making FeCl; anotherpossible solute. Further work is necessary to clarifu thispoint. Experiments performed in NaCl solutions at 650"C,2 kbar,,6, controlled by NNO are shown in Figure 17with NaCl,",, molalities ranging from 0.5 to 2.8. Total Feconcentrations in solution range from 2 x 1g-s to 4 xlO-a mol/L and are surprisingly low. Quench pH was be-tween 8 and 9, and reversals were achieved by addingFeCl, to the starting solution. Magnetite solubility in NaClsolutions at 650'C, 2kbar can be expressed by

log [Fe] : -4.18 + l.29log [NaCl]. (15)

The increase in total Fe with NaCl could be due to theformation of FeCl; or perhaps a hydroxy-chloro complex.The low Fe contents of supercritical NaCl solutions inequilibrium with magnetite demonstrate that such fluidsare not effective solvents even at high P and ?". Similarconclusions were reached by Seyfried and Bischoff(1981)at lower temperatures from experiments of basalt alter-ation with NaCl brines. This can be interpreted by con-sidering the reaction

FerOnu + 6NaCl- =. 3FeCla"or + 3NarO,., * t/2O2. (16)

For Reaction 16 to proceed to the right, a solid or meltmust be present with sufficient affinity to dissolve NarOand allow FeCl, to form. The amount of magnetite dis-solved will be proportional to the amount of NarO con-sumed by that phase.

In acid solutions, we have found that total Fe molalityin solution was, within experimental error, half the totalchloride molality, suggesting a reaction such as

FerOuu + 6HCl - 3FeClzr.or + 3HrO * t/2O2. 07)

During each experiment, essentially all of the initial HCIis consumed; that is, the amount of magnetite dissolvedis directly proportional to the amount of HCI added. This,of course, is the case only when no other source of HCIis present such as the Ag-AgCl buffer. Reaction 17 willproceed to the right until all HCI is exhausted.

There is no difficulty in dissolving magnetite in basicNaCl solutions, because of the presence of NaOH:

FerOo,", * 6NaCl"o, + 3HrO -3FeClr,- + 6NaOH,", * t/2O2. (18)

However, no relevant speciation studies have yet beenreported, and other complexes, such as Fe(OH); are sureto be present as well in such solutions.

At subcritical temperatures, FeCl, is replaced by FeCl*and Fe2* as dominant Fe species. Dissociation constants

(r2)

allows the calculations to be carried out as before. IfthepH is high enough for NaOHo to become a significantspecies, the constant

KN,o': tT=alt=o=Ir (13)INaOH]

must be included. Unfortunately, K,, is not well knownin the supercritical region.

MlcNnrrrr, cAsslrERITE, AND woLFRAMTTEsot,uBrr,rry AND Fe, Sn, W spEcrATroN

Magnetite solubility in supercritical chloride fluids wasdetermined by Chou and Eugster (1977) using the Ag-AgCl buffer (Frantz and Eugster, 1973). Fe in solution wasdetermined to be bivalent. FeCl, was thought to be theprincipal solute, but no definite proofwas provided. Boc-tor et al. (1980) identified FeCl, as the main solute inequilibrium with hematite at 400-600.C, l-2 kbar. Wanget al. (1984; see also Eugster and Wilson, 1985) havedetermined magnetite solubility in HCI and NaCl solu-tions at 600 and 650'C, 2 kbar, with /o, controlled byhematite-magnetite (HM) and nickel-nickel oxide (NNO)bufers. Fe-Cl complexes were identified using the pro-cedures of Wilson (1986), described above. Assuming asingle dominant Fe-Cl complex, abundances at p and Tmust be calculated for the following species: H*, Na*,FeClj-,, Cl-, OH-, HCl, and NaCl. Mass balances for Cl.Na, and Fe were determined after quench, dissociationconstants for HrO, HCl, and NaCl are available from theliterature, and the electrical neutrality condition at p andZprovides the seventh relation among the seven unknownspecies. Again, n is obtained by trial and error. At 600.C,2 kbar, n was found to be 2, confirming FeCl! as theprincipal Fe-Cl complex. Figure 16 shows a plot of log[FeCl]-'l + t/6log fo, * 2 pH vs. log [Ct-] according rothe reaction


Magnet i {e in Chlor ide Sol

600'C , "kbn - 2 -

-2.5 -2.O -t '5 't.o -o.5

Log Ct-

Fig. 16. Solubility of magnetite in HCI-NaCI-H'O solutionsas a function of free chloride ion molality at 600€, 2 kbar, and,fo, controiled by NNO. The solid line is drawn with a slope of2, indicating that FeCl, is the dominant solute. From Wang andEugster (in prep.).

for FeClr, according to

FeCl, + FeCl* + Cl- (19)

and

FeCl* = Fe2* + Cl-, (20)

have been reported by Crerar et al. (1978) at 300'C, butthe data are not suftciently complete to allow extrapo-lations (see also Wood and Crerar, 1985).

Cassiterite, SnOr, is about 100 times less soluble thanmagnetite. Wilson (1986) studied its dissolution and spe-ciation behavior between 400 and 700'C, at I .5-kbar pres-sure, and his results are summarized briefly here (see alsoEugster and Wilson, I 985). Dissolution can be representedby the reaction

SnOr , " , * xH*+nC l -

- Sncl;-, + {n,o * *o,; (2r)' ' ) - 4

x was found to be 2 or 4, depending on /o' and n wasfound to be I or 2 for the stannous complexes (SnCl* andSnClr) and 3 for the stannic complex (SnCli). Figure 18shows the results for400 and 500'C with6, fixed by NNO.A slope of I conforms to all experiments, indicating thereactlon

SnOr,", + 2}l* + Cl- - SnCl* + HrO + YzOr, (22)

and the intercept gives

log K,, : 29.3 - (27 500/T'), (22a)

where Z is in kelvins. Note that dissolution entails re-duction of tin valence. Figure 19 gives the results for HMat 500 and 600qC. Here SnCl r* is dominant, and we have

SnOr,., + 4H' + 3Cr - SnCli + 2HrO (23)

and

log K,r: 33.2 - (18 100/Z), (23a)

6so" C . z k4 NNO in NaCl

magnel i le

^

^ - - ' / - a

- o.zs 0 0,2^r o,5L o g ( m C t ) r o r

Fig. 17. Solubility of magnetite in NaCl-HrO solutions at650f,2 kbar, and/o, controlled by NNO, drawn as a functionof total chloride molality. From Wang and Eugster (in prep.).

60OC and dominant at 700qC with NNO. The stannousand stannic complexes are related by

SnCl* + 2Cl- + 2H' + SnCl3* + H2, Q4)

and this reaction has been plotted in Figure 20 for 500'C,1.5 kbar as a function of pH and chloride molality at Pand T, with /", controlled by quatz-fayalite-magnetite(QFM) and HM. Obviously, higher Cl- and lower pHfavor the stannic complex. Equation 24 has also beenplotted in Figure 2l as a function of fo, and Z, with thepH controlled by the reaction

3KAlSi3O8 + 2H* + KAlrAlSiO'o(OH),+ 6SiO, + 2K* (25)

- 3

L

+

. 9e - *

ttl

-{- )

6

i^D'ht+

. l T

r l

-{2

1 .5 kb

-1.2 -O.8 -O.4 0 0.4

log (Cl-)"''

Fig. 18. Solubility of cassiterite in HCI-HrO solutions, as afunction of chloride ion molality at 400 and 500f, l 5 kbar, and

fi controlled by NNO. The solid lines have a slope of I' indi-cating that SnCl* is the dominant solute. From Wilson (1986).

-10.o

- ' t2.o

-14.0

where lis in kelvins. SnCl, was found to be abundant at

666

1.5 kb MH

.1 4 - 1.O -0.6 -O.2

log(Cl-)"''

Fig. 19. Solubility of cassiterite in HCt-HrO solutions as afunction ofchloride ion molality at 500 and 600"C, I kbar, andf, controlled by MH. The electrical neutrality condition afterquench indicated that Sn4+ !\'as present in solution. The solidlines have a slope of 3, appropriate for SnClr+ as dominant solute.From Wilson (1986).

proceeding in a solution with a total chloride molality of2. The assemblage K-feldspar + muscovite is commonlyobserved in cassiterite deposits, and hence we may con-clude that for such deposits SnClf does not appear to bean important solute species. In order to evaluate Sn con-centrations in solutions responsible for Sn-W deposits,Equation 22 has been plotted in Figure 22 for QFM, NNO,and HM and the pH defined by K-feldspar + albite +muscovite + 2 m (NaCl + KCI) at temperatures between

50ooc 1.5 kb

- L5 - 1.O -0.5 0 0.5

log (Cl-)"''

Fig. 20. Oxidation of SnCl* to Snclf as a function of pH andCl- molalities at 500qC and 1.5-kbar pressure with/o, controlledby QFM and MH. From Wilson (1986).

EUGSTER: MINERALS IN HOT WATER

e

2

caEo'ot

o,o

SnGfj +HrO * SnCl* + 1 t 2O2+ 2H' + 2Cl-

Ksp-ab-ms-qtz2m Cl-., r

5$9

500 550 600

Temperature (oG)

Fig.21. Oxidation ofSnCl'to SnClf as a function of/o, andtemperature with pH defined by the assemblage K-feldspar +albite + muscovite + quartz in a 2 m total chloride solution at1.5-kbar pressure. The MH and QFM curves are shown for com-parison. From Wilson (1986).

400 and 600'C. For these conditions, Sn concentration insolution increases with f and decreases with increasing,fo' For solutions to reach I ppm Sn, for NNO and feld-spar-mica assemblages, a temperature of at last 450'Cmust be reached. This temperature is higher than the de-positional temperature generally deduced from fluid-in-clusion studies. Precipitation ofcassiterite can be initiatedby a drop in ?", oxidation, an increase in pH and probably

SnOr+2Ff '+Ql- iSnCl '+H20 + 1I202

Ksp-ab-ms-qtz

2m Cl.t

4$

400 450 500 550 600TEMP (.C)

Fig. 22. Solubility of cassiterite in HCI-HrO solutions as afunction of temperature and/o, defined by QFM, NNO, or MH.The assemblage K-feldspar + albite + muscovite + quartz de-fines pH, and 2 n KCI defines the chloride molality. From Wilson(1e86).

F t0.0

Ol! '..o,o

6.0

d 1 6o)o1 2 0

G

Io

QFM

T

f.e'


5 0 0 0 c N N O

?\stl

t^""'"$a9 '

1 . 5 K b- . i t e ( i t e

o 3 t ' . -

-1 - . s o

t o g ( C t - ) p . r

Fig. 23. Comparison of the solubility of magnetite and cas-siterite in HCI-H2O solutions, 500'C, 2 kbar (magnetite), 1.5 kbar(cassiterite) andf, controlled by NNO. The slopes of the linesare 2 and I , corresponding to FeCl, and SnCl*, respectively. Datafrom Chou and Eugster (1977) and Wilson (1986).

also a drop in pressure. Precipitation can continue onlyif the H* or HCI produced during deposition is consumed,an aspect to be discussed shortly.

As mentioned earlier, cassiterite is much less solublethan magnetite, and a quantitative comparison is givenin Figure 23. Because Fe is bivalent in solution, even at/o, values of HM (see also Boctor et al., 1980), magnetiteprecipitation also requires oxidation, but less so than cas-siterite deposition.

The solubility of wolframite, (Fe,Mn)WOo, was studiedby Wang, Eugster, and Wilson (see Eugster and Wilson,1985) bV combining solubility experiments on tungsticoxide, WOr, in HCI-HrO solutions at 600'C, 2 kbar withdata on FeCl. According to the 300t spectroscopic mea-surements of Wesoloski et al. (1984) and the data of Iva-nova and Khodakovskiy (1972), tungstic acid, HrWOo, isthe only abundant species to be expected in the super-critical region, with bitungstate, HWO;, and tungstate,WO?-, significant possibly only at high pH. Our very pre-liminary data are compatible with the reaction

W O 3 ( " ) + H 2 O + H 2 W O 4 (26)

using initial HCI molalities of 0-3. Calculated pH at Pand ?"ranges from2to 6, and log Kru is -2.7 + 0.5. Whenmore complete and reliable information for Kru becomesavailable, free eneryies for HWO. can be combined withthose for FeCl, (Chou and Eugster, 1977; Boctor et al.,1980) to calculate ferberite dissolution and precipitation

according to

FeWOn", + 2HCl + FeClz + HrwO4. Q7)

Ferberite can be precipitated through a decrease in chlo-ride or HCI molality, a decrease in pH, an increase inFeCl, or H2WO., and possibly also a decrease in P andI, but further work is necessary to clarifu transport anddeposition of wolframite.

Lrc.lNos orHER THAN CHLoRTDE

By far the largest amount of information available onsupercritical complexes deals with chloride ligands. It isessential that more and better data be collected on (OH)-HrO, carbonate-bicarbonate, and bisulfide complexes.Many other ligands may contribute to mass transport inspecific settings, such as fluoride, borate, phosphate, ar-senate, and others, but even less quantitative informationis available on their role.

For (OH) complexes, the most extensive data set refersto Si(OH)o in equilibrium with quartz. Quartz solubilitywas used recently by McKenzie and Helgeson (1984) toestimate the dielectric constant of water, and an equationvalid to 900'C and l0 kbar has been provided by Fournierand Potter (1982). According to Crerar and Anderson(1971) and Walther and Orville (1982), hydrated silicicacid is the dominant solute. and the dissolution reactioncan be written as

SiO2(,) + (n + 2)HrO + HoSiOo'nHrO. (28)

Crerar and Anderson (1971) suggest that n:2. At sub-critical temperatures and high pH, other species becomeimportant, such as HrSiO; and HrSiOi-, as well as poly-nuclear species (see, for instance, Stumm and Morgan,1981). Because of the high degree of association in su-percritical fluids, it is unlikely that the activity of (OH)-can get high enough for reactions such as

HoSiO4 + (OH)- + (HrSiOa)- + H2O (29)

to proceed sufficiently to the right to make HrSiO; animportant species, except perhaps in the high-pressure'low-temperature region. Data on Krn as a function of Pand Tare essential if we are to understand silica transportand deposition in supercritical fluids.

Judging from its role in subcritical fluids, (OH)- mustalso be an important ligand in neutral to basic supercriticalfluids, although the high end of the pH scale is not ac-cessible because of increased association. Speciation stud-ies must be carried out in alkaline solutions for all majoras well as the less abundant cations. Alteration ofbasalticand ultrabasic crust by meteoric waters is the most likelyenvironment for basic supercritical fluids to form. (OH)-complexes are particularly important for a number of ore-forming elements such as Mo, W, Nb, Ta, and As.

Our information on supercritical carbonate and bicar-bonate complexes is equally scanty. Reliable solubilitydata for calcite above 350€ are not available. We canspeculate that in situations oflow solvent density, neutralcarbonate or bicarbonate complexes, such as NaHCOS,

; 6

^ lxo l +>15

o,o 4


NarCO!, or CaCO!, will be dominant, whereas chargedbicarbonate complexes, such as CaHCOi should be re-stricted to high-density, high-pH conditions. The exis-tence of neutral bicarbonate complexes can perhaps beinferred from the presence of nahcolite, NaHCOr, crystalsin some fluid inclusions (see Roedder, 1985).

Bisulfide complexing has been reviewed by Barnes(1979).Information is extremely scarce above 350.C, in-cluding the dissociation behavior of HrS.

The question of fluoride complexing is raised often inconnection with Sn-W deposits because ofthe abundanceof topaz, lepidolite, and fluorite. Preliminary data (Wil-son, 1986) show that cassiterite is at least 100 times lesssoluble in HF solutions than it is in HCI solutions, in-dicating that fluorine is extracted very efficiently by thesolids and that fluorine complexes are not important forthe transport of Sn in hydrothermal fluids. Ian Plimer(pers. comm.) has remarked recently that'Just becausethe bridesmaid is at the wedding, this does not mean thatshe is getting married."

Rnoox, AcrD-BAsE, AND ExCHANGE REAcrroNS:How no METALS vrovn?

Supercritical fluids are involved in the formation ofhydrothermal ore deposits such as those ofthe porphyrycopper and tin-tungsten types. It is appropriate to ask howsuch fluids acquire the metals they transport and depositas ore minerals, and how these processes are affected bythe properties of the fluids. Somewhat arbitrarily we candivide the processes into redox, acid-base, and exchangereactions, even though a particular transport or depositionevent may involve all three aspects. Studies of mineralsolubilities in hydrothermal fluids have shown that forredox conditions of crust and mantle, transition and othermetals often exhibit a lower valence in the aqueous fluidsthan they do in oxide or silicate minerals. For instance,Boctor et al. (1980) found ferrous chloride, FeClr, to bethe principal solute in equilibrium with hematite, FerOr,at 400-600'C, l-2-kbar pressure. Similarly, Wilson andEugster (1984) reported stannous chloride, SnCl*, as thedominant solute in equilibrium with cassiterite, SnOr, at400-500'C, 1.5 kbar, andforcontrolled by NNO. In oxy-gen-rich solids, such metals may have a higher valencebecause metal-oxygen bonds are dominant, whereas inthe fluid, oxygen is bound largely to hydrogen. This dif-ference implies that reduction is an important aspect ofthe dissolution of minerals containing these metals, justas deposition of the ore minerals requires oxidation. No-ble metals, such as Au or Ag, on the other hand, areoxidized during dissolution. Their deposition is accom-panied by reduction. No free oxygen is present in hydro-thermal fluids, but electron transfer can be accomplishedin a variety of ways, such as by changes in the HrlHrO,HrS/H2SO4, CO2/CH' ratios in the fluid or by changes inthe oxidation states in the crystal lattices. Deposition maybe triggered by the stability or instability ofa particularmineral in response to changes in P, T, pH, or solutioncomposition. Examples of oxidation and reduction reac-

tions during ore deposition are

3FeCl, + 4}J2O - FerOr + 6HCl + Hr, (30)

SnCl* + 2H2O - SnO, + 2H* + Cl- + Hr, (31)

2 A u C l + H 2 - 2 A u + 2 H C l , (32)

and

FeCl, + 2HrSO4 + 7I{z - FeS, + 211Cl + 8HrO. (33)

Normally, the hydrogen produced (oxidation) or con-sumed (reduction) will be involved in a coupled reactionwith ore or country-rock minerals, with solids providingmost of the redox buffering. Differential loss or additionofhydrogen is possible also. Identification ofthe natureof the coupled processes requires detailed textural andcompositional studies. An example of a coupled, oxygen-conserving redox reaction involving the simultaneous de-position ofcassiterite and arsenopyrite was proposed byHeinrich and Eadington (1986):

3SnCl, + 2HrAsO, + 2FeCl, + 2HrS -3SnO, + 2FeAsS + l0HCl. (34)

During this reaction, Sn is oxidized from 2 to 4 and arsenicis reduced from 3 to 2.

The cassiterite deposits ofDachang, southeastern China,are associated with extensive sulfide mineralization, in-cluding pyrrhotite, pyrite, sphalerite, and sulfosalts. Sed-imentologic evidence points to Devonian evaporites asthe source for the sulfur (Eugster, 1983). Cassiterite andsulfide precipitation can be formulated as a combinedredox reaction, using, for the sake of simplicity, SnCl, andFeCl, as the solutes and pyrite as the sulfide.

7SnCl, + FeCl, + 2H2SO4 + 6HrO -7SnO, + FeS, + 16HCl. (35)

Acquisition of ore-forming elements by the evolvingfluids is strongly dependent upon redox processes. Else-where, I have argued (Eugster, 1985a) that large, highlycharged cations (LHCs) such as Sn, W, As, Sb, and Moare mobilized early by fluids with a low/o, and low pH,whereas the bivalent octahedral cations (BOCs) such asFe2*, M12+ and Znz+ are preferentially mobilized duringlater, more oxidizing conditions. However, the details ofthe acquisition process are more difficult to reconstructthan depositional processes.

The importance of acid-base reactions in metamorphicand ore-forming environments was pointed out by Hem-ley (1959), who studied reactions such as

KAl3Si3O'o(OH)' + 6SiO' + 2KCl +3KAlSi3O8 + 2HCl, (36)

3AlrSirOs(OH)o + 2KCl +2KAlrAlSi3O,o(OH), + 2HCl + 3HrO, (37)

and

NaAl3Si3Oro(OH), + 6SiO, + 2NaCl =3NaAlSi3OE + 2HCl. (38)


In each of these mineral-alteration reactions, Al is con-served in the solids, a metal chloride is consumed, andHCI is released. A similar effect was observed in the hy-drothermal alteration of basalts by circulating sea water.Seyfried and Bischoff(1981) found that the precipitationof a magnesian chlorite was accompanied by the releaseof Ca2* and H+ to sea water. This acidity generated by thewater-rock interaction is responsible for the solubilizationof heavy metals. In contrast, near-neutral chloride solu-tions are not effective solvents for heavy metals (Eugsterand Wilson, 1985). The key to forming efective ore fluidsappears to be the production ofacidity or alkalinity. Acid-ity can be produced in a variety ofways (see, for instance,Eugster 1985a, 1985b). In supercritical fluids the domi-nant processes will be (l) mineral precipitation, (2) min-eral alteration, (3) boiling of melts, (4) fluid immiscibility,and (5) oxidation. It is clear from Reactions 30 to 35 thatprecipitation of ore minerals and release of ligands fromthe metal complexes produce acidity. Similarly, mineralalterations such as those of Reactions 36-38 are importantsources ofacidity as is the vesiculation ofhydrous silicatemelts. The latter process is induced by crystallization ofanhydrous phases (see, for instance, Burnham, I 979). Sep-aration of a vapor phase promotes the enrichment ofacidssuch as HCl, HF, H3BO. in the fluid. Judging from thewidespread acid alteration associated with porphyry cop-per and tin-tungsten deposits, boiling of melts is a signif-icant source for acid production. HCI dissolved in themelt is strongly fractionated into the aqueous fluid. Hol-land (1972) studiedthe distribution ofZn and Mn betweenNaCl-bearing granitic melt and fluid at 850'C, 2 kbar.Upon quench, he observed low pH values, and this I haveinterpreted (Eugster, 1985a) as due to hydrolyzation ofNaCl dissolved in the melt by the reaction

2NaCl -"n, f H2O16uiay - 2HCl1s"4 + NarO,-",,,

being displaced to the right. At 700-750'C, Kilinc andBurnham (l 972) found the ratio (n",) solution /(m) meltto be between 13 and 83. Consequently, as long as NaClis dissolved in the silicate melt, HCI production uponvesiculation continues, particularly if an immiscible NaClliquid is present in the melt.

Fluid fractionation during subuitical boiling has beeninvoked as an important process responsible, for instance,for calcite precipitation in geothermal fields, with calcitesupersaturation triggered by loss ofCO, to the gas phase.Recently Gehrig (1980), Hendel and Hollister (1981), Sis-son et al. (1981), Bowers and Helgeson (1983a, 1983b),and Skippen and Trommsdorf (1986) have pointed outthat NaCl-rich HrO-CO, fluids can remain immiscible totemperatures and pressures well above the critical pointof HrO (see also Takenouchi and Kennedy, I 96 5). In otherwords, subcritical boiling can occur over the whole rangeof conditions normally associated with supercritical fl uids,provided substantial amounts of CO, and NaCl are pres-ent. The metamorphism of carbonate rocks is the mostlikely environment to provide these conditions. We mayexpect that during such boiling, volatile complexes such

as HCl, HF, and some metal chlorides are preferentiallyfractionated into the COr-rich aqueous vapor.

Oxidation can be a powerful acid-producing mecha-nism, as, for instance, in the oxidation of pyrite respon-sible for acid mine drainage. Similar processes are possibleunder supercritical conditions provided an oxygen sourceexists:

2FeS, + 4HrO + 7V2Or - FerO, + 4H2SO4.

Metal transport is possible also in alkaline supercriticalsolutions, where hydroxy complexes are the dominantsolutes. For a bivalent metal, Me2*, we can expectMe(OH)*, Me(OH)!, Me(OH), , and Me(OH)?- to be pres-ent depending on pH, P, and T. Seward (1981) has sum-marized the relevant data for Fe2*, Pb, Zn, Ni, UOr, andAl, but few of the measurements extend much beyond300"C, the limit for most spectrophotometric cells.

Exchange reactions between minerals and neutral ornear-neutral supercritical fluids are effective mechanismsfor metal acquisition and enrichment. Exchange constantsfor major elements in chloride solutions were discussedby Eugster and Ilton (1983). No data are available for theless abundant metals important to ore deposition. Onewould expect that metals such as Zrr, Cu, and Pb couldreadily be picked up by chloride-rich hydrothermal fluidspassing through wall rocks of appropriate composition,but the relevant experiments remain to be carried out.

More information is available on metal partitioningbetween silicate melts and hydrothermal fluids. Earlierwork on this subject has been summarized by Candelaand Holland (1984), who also measured the partitioningof Cu and Mo between silicate melts andaqueous chloridefluids at 750",C. Manning and Henderson (1984) studiedthe behavior of W in silicate melt-aqueous fluid systemsat 800'C, and I kbar, using various ligands in the fluid.In general, speciation has not been fully defined in suchfluids.

Sutvrvr,lnY

Supercritical aqueous solutions are very different fromelectrolyte solutions near room temperature, and thesedifferences have profound consequences on how waterdissolves minerals and transports metals and lig;ands. Nearroom temperature, water is exceptionally effective in dis-solving ionic crystals at any pH, whereas covalent struc-tures, such as silicates, dissolve only at low or high pH.Dissolution rates vary widely from mineral to mineral.At high temperature, on the other hand, water is a goodsolvent for most minerals. Dissolution is influenced byacid and ligand concentrations and is dominated by theproperties of the solvent through the formation of near-neutral complexes. Rates of dissolution of silicates appearto be independent of mineral constitution. According toWalther and Wood (1986, p. 208), "at 25"C silicate dis-solution rates are pH-dependent and vary widely fromone mineral to another. At high temperatures (300-700"C),however, the available data indicate that most mineralsdissolve in aqueous and HrO-CO, fluids with similar rate


---> 3 ct-Fig. 24. Schematic drawing of the effect ofP, I, and chloride

ion molality on the stepwise complexing of a bivalent metal, M.

constants if their rates are nonnalized to a constant num-ber of oxygen gram atoms." Above 300.C, dissolutionrates can be expressed by the relation (Wood and Walther,I 983)

-2900log k: - 6.85 mol oxygen atoms/(cm,.s).

In supercritical fluids oflow density, normally strong acids,bases, and salts become weak acids, bases, and salts,whereas otherwise weak acids, bases, and salts form neu-tral, associated complexes.

Because of the decrease in the dielectric constant, thedegree of association of solutes increases with increasingtemperature. Electrostriction causes pressure to have theopposite effect, enhancing dissociation ofboth solute andsolvent. Pressure effects are particularly important in thelow-density supercritical fluids to be expected in manymetamorphic and igneous environments. The decrease inelectric shielding in such fluids leads to dominantly elec-trostatic interaction between metal ions and ligands. Con-sequently, ligand-field effects are greatly diminished. Ionicstrengths are low even in concentrated solutions, and ac-tivity coefrcients can be approximated by a Debye-Hiickeltreatment. Minerals dissolve largely through the forma-tion of inner-sphere complexes. Such complexes are tet-rahedrally or octahedrally coordinated, depending uponpressure, although higher coordination numbers are pos-sible. With increasing ligand-ion concentrations, stepwise

complex formation takes place with ligands replacing H'Omolecules in the complex. Such association reactions di-minish the number of positive charges and with it extentof solvation. They are endothermic and are driven by theentropy gain associated with dehydration. The formationof anionic complexes is restricted by the low abundanceof ligand ions in these fluids. In Figure 24, the effects ofP, ?i and chloride-ion activity on complexing of a bivalentmetal have been summarized schematically. Rising tem-perature restricts the abundance of cations and of freechloride ions, enhancing the dominance of near-neutralcomplexes. A rise of pressure counteracts the temperatureeffect.

At high temperatures, where minerals dissolve primar-ily through formation of complexes of no or low charge,information on such complexes has been obtained mainlyfrom spectroscopic measurements in the low-pressure,subcritical L + V region. Data on supercritical complexesare much scarcer and confined largely to chloride ions asligands. Mineral-solubility determinations can providespeciation information along with solubility products. Thetrial-and-error procedure developed by Glenn Wilson andillustrated here for cassiterite and magnetite solubility iscapable ofdefining the abundances ofone or more ligands.Spectroscopic data would be invaluable in confi.rming theseconclusions. Metal transport can be accomplished also bycarbonate-bicarbonate, bisulfide, and hydroxy complexes,but no supercritical data are available.

Redox, acid-base, and exchange reactions are the prin-cipal mechanisms by which metals are acquired by hy-drothermal fluids. Redox reactions often seem to be cou-pled processes with no net oxygen transfer. Acids, whetherdissociated or associated, are effective in releasing metalsfrom minerals into the fluid. Such acids are produced bymineral precipitation, mineral alteration, boiling of melts,unmixing of fluids, and oxidation. Acid formed duringprecipitation of ore minerals is neutralized by wall-rockalteration. Metals are soluble also in alkaline fluids throughthe formation of hydroxy or carbonate complexes, butlittle is known of such fluids in the supercritical region.

Mineral dissolution and precipitation reactions occuron the molecular level, and yet their consequences are asfar-reaching as some of the more glamorous global pro-cesses. Supercritical fluids are involved in ore depositionand metamorphic events, in igneous crystallization andmantle degassing, and yet we remain largely ignorant oftheir chemical properties and their interaction with min-erals.

AcxNowr-rnGMENTS

I would like to express my gratitude to Glenn Wilson for con-tributing substantially to the ideas expressed in this address andfor permission to use unpublished material. Terry Seward freelyshared insights and unpublished data, and his studies have beenan inspiration for Glenn and me. Glenn, Terry, and DimitriSverjensky made helpful comments on the manuscript. This workhas been supported by NSF grants EAR 8206177 and EAR841 1050.

EUGSTER: MINERALS IN HOT WATER 67r

RnrnneNcnsBarnes, H.L. (1979) Solubilities of ore minerals. In H.L. Barnes,

Ed. Geochemistry of hydrothermal ore deposits (2nd edition),404-460. John Wiley, New York.

Boctor, N.2., Popp, R.K., and Frantz, J.D. (1980) Mineral so-lution equilibria IV. Solubilities and the thermodynamic prop-erties of FeC$ in the system FerOr-Hr-HrO-HCl. Geochimicaet Cosmochimica Acta, 44, I 509-1 5 18.

Bowen, N.L., and Tuttle, O.F. (19a9) The system MgO-SiOr-HrO. Geological Society of America Bulletin, 60,439-460.

- ( I 950) The system NaAlSi,O8-KAlSi3Os-H2O. Journal ofGeology, 58, 489-5 1 1.

Bowers, T.S., and Helgeson, H.C. (1983a) Calculation of thethermodynamic and geochemical consequences of nonidealmixing in the system HrO-COr-NaCl on phase relations ingeologic systems: Equation of state for HrO-COr-NaCl fluidsat high pressures and temperatures. Geochimica et Cosmo-chimica Acta, 47, 1247-127 5.

- (1983b) Calculation of the thermodynamic and geochem-ical consequences of nonideal mixing in the system HrO-COr-NaCl on phase relations in geologic systems: Metamorphicequilibria at high pressures and temperatures. American Min-eralogist, 68, 1059-1075.

Burnham, C.W. (1979) Magmas and hydrothermal fluids. In H.L.Barnes, Ed. Geochemistry of hydrothermal ore deposits (2ndedition), 7l-136. John Wiley, New York.

Burnham, C.W., Holloway, J.R., and Davis, N.F. (1969) Ther-modynamic properties of water to 1000"C and 10,000 bars.Geological Society of America Special Paper 132, l-96.

Candela, P.A., and Holland, H.D. (1984) The partitioning ofcopper and molybdenum between silicate melts and aqueousfluids. Geochimica et Cosmochimica Acta, 48, 373-380.

Chou, I-M., and Eugster, H.P. (1976) Hydrothermal acid-basebuffers: Fugacity control and dissociation constants ofHBr andHI. Contributions to Mineralogy and Petrology, 56,77-100.

-(19'77) Solubility of magnetite in supercritical chloridesolutions. American Journal of Science, 27 7, 1296-13 14.

Crerar, D.A., and Anderson, G.M. (1971) Solubility and solva-tion reactions of quartz in dilute hydrothermal solutions.Chemical Geology, 8, 107-122.

Crerar, D.A., and Barnes, H.L. (1976) Ore solution chemistry-V. Solubilities of chalcopyrite and chalcocite assemblages inhydrothermal solution at 200-350qC. Economic Geology, 71,722-794.

Crerar, D.A., Susak, N.J., Borcsik, M., and Schwartz, S. (1978)Solubility of the buffer assemblage pyrite + pyrrhotite + mag-netite in NaCI solutions from 200 to 350€. Economic Geol-ogy ,7 1 ,772-794.

Crerar, D.A., Wood, S., and Brantly, S. (1985) Chemical controlson solubility of ore-forming minerals in hydrothermal solu-tions. Canadian Mineralogist, 23, 332-352.

Daubr6e, A. (1841) M6moire sur le gisement, la constitution etI'origine des amas de minerai d'6tain. Annales des Mines, 3's6rie, t. XX, 65-112.

-(1849) Recherches sur la production artificielle de quel-ques espdces min6rales cristallines paniculiOrement de I'oxyded'6tain, de I'oxyde de titane et du quartz. Observations surI'origine des filons titaniferes des Alpes. Annales des Mines, 4"s6rie, t. XVI, 129-155.

- (1857) Observations sur le m6tamorphisme et recherchesexpErimentales sur quelques agents qui ont pu le produire.Annales des Mines, 5" s6rie, t. ){ll,289-326.

-(1879) Etudes synthetiques de g6ologie exp6rimentale.Dunod, Editeur, Paris, 828 p.

Doelter, C. (1890) Allgemeine chemische Mineralogie, Verlagvon Wilhelm Engelmann, l*ipzig,278 p.

Eugster, H.P. (1957) Heterogeneous reactions involving oxida-tion and reduction at high pressures and temperatures. Journalof Chemical Physics, 26,387.

- ( I 97 l) The beginnings ofexperimental petrology. Science,1 6 , 4 8 1 - 4 8 9 .

-(1981) Metamorphic solutions and reactions. In F.E.Wickman and D.T. Rickard, Eds. Chemistry and geochemistryof solutions at high temperatures and pressures. Physics andChemistry of the Earth, 13/14,461-507. Pergamon Press, NewYork.

- ( I 983) Cassiterite-sulfide deposition at Dachang, GuangxiProvince, S.E. China. Geological Society America Absfactswith Programs, 15, 568.

-(1985a) Granites and hydrothermal ore deposits: A geo-chemical framework. Mineralogical Magazine, 49, 7 -23.

- (1985b) Oil shales, evaporites and ore deposits. Geochim-ica et Cosmochimica Acta, 49,619-635.

Eugster, H.P., and llton, E.S. (1983) Mg-Fe fractionation in meta-morphic environments. In S. Saxena and A.B. Thompson, Eds.Advances in physical geochemistry, 4, ll4-140. Springer-Ver-lag, New York.

Eugster, H.P., and Wilson, G.A. (1985) Transport and depositionof ore-forming elements in hydrothermal systems associatedwith granites. In C. Hatls, Ed. High heat production granites,hydrothermal circulation and ore genesis, 87-98. Institute ofMining and Metallurgy, London.

Eugster, H.P., Harvie, C.E., and Weare, J.H. (1980) Mineralequilibria in the six component sea water system, Na-K-Mg-Ca-SOo-HrO, at 25C. Geochimica et Cosmochimica Actz, 44,1335-1347.

Eugster, H.P., Chou, I-M., and Wilson, G.A. (1986) Mineralsolubility and speciation in supercritical chloride fluids' In G.C'Ulmerand H.L. Bames, Eds. Hydrothermal experimental tech-niques, in press. John Wiley, New York.

Fournier, R.O., and Potter, R.W. (1982) A revised and expandedsilica (quartz) geothemometer. Geothermal Research CouncilBul let in, 11,3-9.

Franck, E.U. (1956) Hochverdichteter Wasserdampf I. Elektro-lytische Leit{iihigkeit in KCl-HrO-losungen bis 750'C. ZEit-schrift {iir Physikalische Chemie, Neue Folge, 8,92-106.

- (1981) Survey of selected non-thermodynamic propertiesand chemical phenomena of fluids and fluid mixtures. In D.Rickard and F. Wickman, Eds. Chemistry and geochemistryof solutions at high temperatures and pressures. Physics andChemistry of the Earth, l3114, 65-88. Pergamon Press, NewYork.

Frantz, J. D., and Eugster, H.P. (1973) Acid-base buffers: Use ofAg + AgCl in the experimental control of solution equilibriaat elevated pressures and temperatures. American Journal ofScience, 273,268-286.

Frantz,J. D., and Marshall, W.L. (1982) Electrical conductancesand ionization constants ofcalcium chloride and magnesiumchloride in aqueous solutions at temperatures to 600"C andpressures to 4000 bars. American Journal of Science, 282,r66Gr693.

- (1984) Electrical conductances and ionization constantsof salts, acids, and bases in supercritical aqueous fluids: I'Hydrochloric acid from 100'to 700"C and at pressures to 4000bars. American Journal of Science, 284, 651-667.

Frantz, J. D., and Popp, R.K. (1979) Mineral solution equilib-ria-I. An experimental study of complexing and thermody-namic properties of aqueous MgCl, in the system MgO-SiOr-HrO-HCl. Geochimica et Cosmochimica Acfa, 43, 1223-1239 .

Friedel, C., and Sarasin, E. (1879) Surla productiond'une matidrefeldspathique. Society of Mineralogy Bulletin, 2, 158-161.

Gehrig, M. (1980) Phasengleichgewichte and PVT'Daten ternilrerMischungen aus Wasser, Kohlendioxid und Natriumchloridbis 3 kbr und 550"C. Doctoral dissertation, published byHochschul Verlag, Freiburg, Germany.

Geikie, A. (1905) The founders of geology. Macmillan, London,486 p.

Goranson, R.W. (1938) Silicate-water systems: Phase equilibriain the NaAlSitOr-HrO and KAlSirOr-HrO systems at high tem-


peratnres and pressures. American Journal of Science. 35A.7 t-91.

Hall? J. (1805) Account of a series of experiments showing theefects of compression in modifuing the action of heat. RoyalSociety of Edinburgh Transactions, vi, 7 1.

Harvie, C.8., and Weare, J.H. (1980) The prediction of mineralsolubilities in natural waters: The Na-K-Mg-Ca-Cl-SO4-HrOsystem from zero to high concentration at 25rC,. Geochimicaet Cosmochimica Acta, 44,981-997.

Harvie, C.E., Eugster, H.P., and Weare, J.H. (1982) Mineralequilibria in the six-component seawater system, Na-K-Mg-Ca-SO4-Cl-HrO at25t,.II: Compositions of the saturated so-lutions. Geochimica et Cosmochimica Acta, 46, 1603-1618.

Harvie, C.E. Moller, N., and Weare J.H. (1984) The predictionof mineral solubilities in natural waters: The Na-K-Mg-Ca-H-CI-SO4-OH-HCO3-CO3-COr-H2O system to high ionic strengthsal 25f. Geochimica et Cosmochimica Acta, 48,723-751.

Heinrich, C.A., and Eadington, P.J. (1986) Thermodynamic pre-dictions of the hydrothermal chemistry of arsenic, and theirsignificance for the paragenetic sequence of some cassiterite-arsenopyrite-base metal sulfide deposits. Economic Geology,81, in press.

Helgeson, H.C. (1981) Prediction of the thermodynamic prop-erties ofelectrolytes at high pressures and temperatures. In D.Rickard and F. Wickman, Eds. Chemistry and geochemistryof solutions at high temperatures and pressures. Physics andChemistry ofthe Earth, l3 and 14, 133-177. Pergamon Press,New York.

Helgeson, H.C., and Kirkham, D.H. (1976) Theoretical predic-tion of the thermodynamic properties of aqueous electrolytesat high pressures and temperatures (III): Equation of state foraqueous species at infinite dilution. American Journal ofSci-ence,276,97-240.

Helgeson, H.C., and Flowers, G.C. (1981) Theoretical predictionofthe thermodynamic behavior of aqueous electrolytes at highpressures and temperatures. IV. Calculation of activity andosmotic coefrcients and apparent molal and standard and rel-ative molal properties to 600€ and 5 kb. American Journalof Science, 281, 1249-1493.

Hemley, J.J. (1959) Some mineralogical equilibria in the systemKrO-AlrO3-SiOr-HrO. American Journal of Scienc,e, 257, 241-270.

Hendel, E.M., and Hollister, L.S. (1981) An empirical solvus forCO2-HrO-2.6 wt. % salt. Geochimica et Cosmochimica Acta.45.225-228.

Henley, R.W., Truesdell, A.H., and Barton, P.B. (1984) Fluid-mineral equilibria in hydrothermal systems. Reviews in Eco-nomic Geology, l, l-267.

Holland, H.D. (1972) Granites, solutions, and base metal de-posits. Economic Geology, 67, 281-301.

Holland, H.D., and Malinin, S.D. (1979) The solubility and oc-currence of non-ore minerals. In H.L. Barnes. Ed. Geochem-istry ofhydrothermal ore deposits (2nd edition), 46 l-508. JohnWiley, New York.

Hwang, J.U., Liidemann, H.D., and Hartmann, D. (1970) Dieelektrische Leitliihiekeit konzentrierter wdssriger Alkalihalo-genid-Liisungen bei hohen Driicken und Temperaturen. Hightemperatures-high pressures, 2, 651469.

Ivanova, G., and Khodakovskiy , J. (1972) Form ofW in hydro-thermal solutions. Geokhimiya, 1426-1433.

Kennedy, G.C. (1950) A portion of the system silica-water. Eco-nomic Geology, 45, 629453.

Kilinc, I.A., and Bumham, C.W. (1972) Partitioning of chloridebetween a silicate melt and coexisting aqueous phase from 2tq8 kilobars. Economic Geology, 67,231-235.

Klostermeier, W. (1 973) Elektrische L€itfiihigkeit konzentrierterwiissriger Alkalihalogenidliisungen bei hohen Driicken und

- Temperaturen. Ph.D. thesis, Universitet Karlsruhe, Germany.lifuls6ann, H.D., and Franck, E.U. (1968) Absorptionsspeltren

bei hohen Driicken und Temperaturen. II. Kobalt (ID- undNickel flI)-halogenid in konzentrierten Alkalihalogenid-L6-

suogen. Berichte der Bunsengesellschaft fiir Physik und Chem-ile,72, 514-523.

Manning, A.C., and Henderson, P. ( 1984) The behaviour of tung-sten in granitic melt-vapour systems. Contributions to Min-eralogy and Petrology, 86, 286-293.

Marshall, W.L., and Franck, E.U. (1981) Ion product of watersubstance, G-1000"C, l-10,000 bars. New international for-mulation and its background. Journal ofPhysics and Chem-istry Reference Data, 10, 295-303.

McKenzie, W.F., and Helgeson, H.C. (1984) Estimation of thedielectric constant of HrO from experimental solubilities ofquartz, and calculation of the thermodynamic properties ofaqueous species to 900qC at 2 Kb. Geochimica et Cosmochim-ica Acta, 48, 2167-217 8.

Morey, C.W. (1942) Solubility of solids in water vapor. Pro-ceedings of the American Society for Testing Materials, 42,980-988.

Morey, G.W., and Hesselgesser, J.M. (1951) The solubility ofsome minerals in super-heated steam a1 fiigh pressures, Eco-nomic Geology, 46, 821-835.

Morey, G.W., and Ingerson, E. (1937) The pneumatolytic andhydrothermal alteration and synthesis of sitcates. EconomicGeology 32, supplement to no. 5,607-760.

Niggli, P., and Morey, G.W. (1913) Mitteilungen aus dem La-boratorium fiir Geophysik des Carnegie Institutes in Wash-ington. 17. Die hydrothermale Silikatbildung. Zeitschrift fiirAnorganische Chemie, 83, 369-416.

Pearson, R.G. (1963) Hard and soft acids and bases. Journal ofthe American Chemical Society, 85, 3533-3539.

Pitzer, K.S. (1981) Discussion of H.C. Helgeson, Prediction ofthermodynamic properties of electrolytes. In D. fuckard andF. Wickman, Eds. Chemistry and geochemistry of solutions athigh temperatures and pressures. Physics and Chemistry oftheEarfh, 13/14, 175. Pergamon Press, New York.

-(1983) Dielectric constant of water at very high temper-ature and pressure. National Academy of Sciences USA Pro-ceedings, 80, 457 5-457 6.

Popp, R.K., and Frantz, J.D. (1979)Mineral solutionequilibria-II. An experimental study of mineral solubilities and the ther-modynamic properties of aqueous CaCL in the system CaO-SiOr-HrO-HCl. Geochimica et Cosmochimica Acta, 43, 1777-1790.

-(1980) Mineral solution equilibria-Ill. The system NarO-AlrO3-SiOr-HrO-HCl. Geochimica et Cosmochimica Acta, 44,r029-t037.

Quist, A.S., and Marshall, W.L. (1968) Electrical conductancesofaqueous sodium chloride solutions from 0 to 800'C and atpressures to 4000 bars. Journal ofPhysical Chemistry,72,684-703.

Ritzert, G. von, and Franck, E.U. (1968) Elektrische Leitfiihigkeitwdssriger Ldsungen bei hohen Temperaturen und Driicken. I.KCl, BaClr, Ba(OH), und MgSOo bis 750"C und 6 kbar. Be-richte der Bunsengesellschaft fiir Physik und Chemie,72,798-808.

Robinson, R.A., and Stokes, R.H. (1959) Electrolyte solutions(2nd edition). Butterworths, London, 571 p.

Roedder, E. (1985) Fluid inclusions. Mineralogical Society ofAmerica Reviews in Mineralogy, 12,644 p.

Rudert, V., Chou, I-M., and Eugster, H.P. (1976) Temperaturegradients in rapid-quench cold-seal pressure vessels. AmericanMineralogist, 6 l, 101 2-101 5.

S6narmont, H. de. (1851) Exp6riences sur la formation des mi-n6raux par voie humide dans les gites m6tallifdres concr6-tionn6s. Annales de Chimie et de Physique, Troisidme Series,t29-175.

Seward, T.M. (1976) The stability of chloride complexes of silverin hydrothermal solutions up to 350"C. Geochimica et Cos-mochimica Acta, 40, 1329-1341.

-(1981) Metal complex formation in aqueous solutions atelevated temperatures and pressures. In D. Rickard and F.Wickman, Eds. Chemistry and geochemistry of solutions at


high temperatures and pressures. Physics and Chemistry oftheEafih, 13/14, I l3-132. Pergamon Press, New York.

-(1984) The formation of Pb (II) chtoride complexes to30OC: A spectrophotometric study. Geochimica et Cosmo-chimica Acta, 48, l2l-134.

Seyfried, W.E., and Bischoff, J.t . (1981) Experimental seawater-basalt interaction at 300qC, 500 bars, chemical exchange, sec-ondary mineral formation and implications for the transportof heavy metals. Geochimica et Cosmochimica Acta,45, 135-148.

Sisson, V.B., Crawford, M.L., and Thompson, P.H. (1981) COr-brine immiscibility at high temperatures, evidence from cal-careous metasedimentary rocks. Contributions to Mineralogyand Petrology, 7 8, 37 l-37 8.

Skippen, G., and Trommsdofi V. (1986) The influence ofNaCland KCI on phase relations in metamorphosed carbonate rocks.American Journal ofScience, 286, 8l-104.

Stumm, W., and Morgan, J.J. (1981) Aquatic chemistry (2ndedition). John Wiley, New York, 780 p.

Susak, N.J., and Crerar, D.A. (1985) Spectra and coordinationchanges of transition metals in hydrothermal solutions: Im-plications for ore genesis. Geochimica et Cosmochimica Acta,49,555-564.

Takenouchi, S., and Kennedy, G.C. (1965) The solubility of car-bon dioxide in NaCl solution at high temperatures and pres-sures. American Journal of Science, 263, 44 5-454.

Truesdell, A.H., and Jones, B.F. (197a) wnree: A computer pro-gram for calculating chemical equilibria of natural waters. U.S.Geological Survey Journal of Research, 2, 233-248.

Tuttle, O.F. (1948) A new hydrothermal quenching apparatus.American Joumal of Science 246,628435.

Tuttle, O.F., and Bowen, N.L. (1958) Origin of granite in thelight of experimental studies in the system NaAlSi3Or-KAl-Si3O8-SiO2-HrO. Geological Society of America Memoir 74,153 p .

Vogt, J.H.L. (1904) Die Silikatschmelzlbsungen mit besondererRiicksicht auf die Mineralbildung und die Schmelzpunkt-Er-

niedrigung. Skrifter Videnskabs-Selskabet l. Christiania, l-326. I :8 , l -61 , I I : I , l -136.

Walther, J.V., and Orville, P.M. (1982) Volatile production andtransport in regional metamorphism. Contributions to Min-eralogy and Petrology, 7 9, 252-257.

Walther, J.V., and Wood, B.J. ( I 98 6) Mineral-fluid reaction rates.In J.V. Walther and B.J. Wood, Eds. Fluid-rock interactionsduring metamorphism. Advances in Physical Geochemistry,5, 19+213. Springer-Verlag, New York.

Wang, S., Eugster, H.P., and Wilson, G. (1984) Solubility ofmagnetite and iron speciation in supercritical HCI and NaClsolutions. Geological Society of America Abstracts with Pro-grams, 16, 686.

Wesoloski, D., Drummond S.E., Mesmer, R.8., and Ohmoto, H.(1984) Hydrolysis equilibria oftungsten (VI) in aqueous so-dium chloride solutions to 300€. Inorganic Chemistry, 23,rt2vtt32.

Wilson, G.A. (1986) Cassiterite solubility and metal-chloridespeciation in supercritical solutions. Ph.D. thesis, Johns Hop-kins University, Baltimore, Maryland.

Wilson, G.A., and Eugster, H.P. (1984) Cassiterite solubility andmetal-chloride speciation in supercritical solutions. GeologicalSociety of America Annual Meeting Abstracts with Programs,16,696.

- ( I 98 5) Metal-chloride speciation from supercritical sol-ubility parameters. Geological Society of America Abstractswi th Programs, I7 ,75 l .

Wood, B.J., and Walther, J.V. (1983) Rates of hydrothermalreactions. Science, 222, 41341 5.

Wood, S.A., and Crerar, D.A. (1985) A numerical method forobtaining multiple linear regression parameters with physicallyrealistic signs and magnitudes: Applications to the determi-nation of equitibrium constants from solubility data. Geo-chimica et Cosmochimica Acta, 49, 165-172.

MnNuscnrrr RECETVED JnNunnv 16, 1986Mllnrscmpr AccEprED JeNu,cnv 30, 1986

Documents

Minerals in hot waterr · 2007-08-28 · volving mineral equilibria in superheated steam, Morey with respect to mineral solubilities (Morey, 1942), and Goranson (1938) with respect