Rome, 1991 ROME - ITALY . APPLICATIONS OF GEOCHEMISTRY IN GEOTHERMAL RESERVOIR t DEVELOPMENT ~ by Franco D'AMORE (Co-ordinator) ti ! a i u Nee eres i |CHEMICAL TECHNIQUES IN GEOTHERMAL EXPLORATION W.F. GIGGENBACH Chemistry Division, DSIR, Private Bag, Petone, New Zealand INTRODUCTION ‘The application of chemical techniques has become an integral part of any geothermal exploration programme. Especially during the erly, prerilling stages, geochemistry may frovige information on deep conditions and processes not obsirable by geological or Prophysical techniques. After all, the waters and gases discharged atthe surface “have been Rew" and generally carry imprints of their deeper histories with them. The constituents arcpuntered in these fluids may be subdivided into two major groups according to the type of information provided: Chemically inert, non-reactive constituents form a group which may be called tracers. Once added to the fluid phases, ideally they should remain unchange¢. providing a tag allowing their origins to be traced back to their source components. Chemically reactive species, responding to changes in their environment in a controlled and well understood manner, form a group potentially useful as geoinulcarors. Examples of the frst group are, of cours, the noble gases He ard Ar, followed by the comparatively “conservative” constituents Cl, B, Li, Rb, Cs, and N:. The species ofthe second group, such as Na, K, Mg, Ca. and SiO,, take pat in temperature-dependent interactions with the Alsilicate rock structure generally housing geothermal systems. or are, a5 ¢.g. Hs, HS, CH., and CO, involved in temperature- and pressure-dependent redox reactions among one anovher or with redox systems (Fell/Felll of the rock phase, The boundaries between the two groups are not rigid. A constituent may under one set of conditions be quite inert, e.g. Cs at temperatures >250°C, but it is incorporated into secondary zeolites at lower temperatures. Each of the constituents of geothermal fluids behaves in its own peculiar way and therefore, if used to best advantage, will provide unique information. In this introduction to the application of chemical techniques in geothermal exploration, simple procedures are described allowing an initial assessment ofthe geothermal potential of an area to be made by extracting much of the information carried by a given constituent in a comparatively straight-forward manner. The procedures frequently require an initial, “eyeball” ceevement of correlations among sets of various constituents, a task most readily carried out by tse of triangular diagrams. A special effort, therefore, is made to point out the peculiarities of this type of data presentation. The procedures are based on consituents analyzed most frequently and by readily available techniques (Giggenbach and Gogue!, 1989).| W.F. Giggenbach Sets of analytical results for quite arbitrarily selected water and ges samples, spanning a wide range of tectonic, geological and hydrological environments, are given in Tables 1 and 2. For four high temperature geothermal areas (WK, NG, ZU, and MV), pairs of samples representing well and associated spring or fumarole discharges are given. The two Bt Ruiz samples are from a highly acid voleanic spring and 2 neutral chloride spring at lower levels. The Araro samples are likely 10 represent the natural outflow of chloride water from the Los [Azufres geothermal field. The two samples from boiling springs at Manikaran and-Fang have rho obvious volcanic-magmatic associations; the Yasur water is discharged from the complex housing the active volcano, the gas sample fiom an associated “geothermal” furmarole. |The Paraso samples are from an extensive area of hot water discharges on this volcanic island; the Lake Nyos samples represent the deep, CO, charged waters there. The Waitangi Soda Spring samples are also produced from a highly CO; charged system. MO and MU are associated with ratural gas discharges along the east and west coast of New Zealand, while the White Island samples are from a spring producing highly acidic, obviously voleanie waters. ‘These data will be-used 2s examples to illustrate the various procedures. Conclusions drawn here should in no way be considered prejudicial to any ongoing or future investigations. Table 1 The chemical composition of thermal and mineralized waters from New Zealand (WK, NG, MO, MU, We), Guatemala (ZU), Costa Rica (MV), Colossus As RB), Mexico (AR), India (MA), Thailand (FN), Solomon Islands (PR), Vanuatu (YA) and Cameroon (LN) in mg/kg. ae a ee ee er fx veicate, welt un ur ie Se Sie 8 BR [Matatels‘pting Wo te 2 RE ie} He tanta tt a seo. rs 0.30 Seas |B Re senna, spring oo alm H&E Mo te ibe 2 awit, wett x vaso 210 1.90 has 0 a0 61700 py iets Sing ae Sto ar as 855 fe we “te fy aivevticss ett 205 io a te Bf °G ie ste Who hirovattey spre 73 we mos Bo ow te date as tule, aid spring wo 2 0.37 De 8 + ise 105m 1350 is tutes neutral 8s (5 ito “ta 0:36 Me a8 um feo ‘at too it avers, spring |B m3 0.8 $B a 1% wo ih honitban, seeing 9 OM OS 23a |S ‘he 1a faa, spring 2 8 Laws oF farts epring wo we acre wy GS Yo 205 a3t8 ek asus spring * fm "Boe a 8 iro io ieee Gye the B 33 loa $4075 "a is waitanak som spe. «2 as 2 0.1 vow me 8 38 iS nareec serine oF ga H So net FS ste fa navi sett 3 ee aS Gio mo ts co 3s 8 12600 ft dniee Vets spring 38 $8 Go 8) GSE Skin ato veo “a = 4a70 38700 sv senater ‘ sesso 3am 0418 «or 270 ao 5 MO = 27H 19000 120Chemical Techniques in Geothermal Exploration ‘Table The chemical composition of gases associated withthe water discharges of Table 1, on water-free basis in mmol/mol. ees wet gam oe Denn. oo $8 & eer a umathe 2 OS CONSTITUENTS PARTITIONING INTO THE LIQUID PHASE Relative Cl, SO, and HCO; contents Most geochem-al techniques may with confidence be applied only to specified types of fluids with limited ranves of compositions. For instance, most ionic solute geothermometers work" only if used wh close to neutral waters containing chloride as the major anion. Any such interpretation of geothermal water samples, therefore, is best carried out on the basis of sh inital classification, e.g. in terms of their major anions Cl, SO, and HCOs, as shown in Fig. 1. “The position of a data point in such a triangular or trilinear plot is simply obtained by first forming the sum § of the concentrations ¢, (in mg/kg) of all three constituents involved. In the present case S = ca + Csr + CHeos Q “The next step consists of the evaluation of "%-Cl* and "%-HCO,” according to "%-Cl" = 100cq/S Q and "%-HCO;" = 100cycoy/S @ 121W.F. Giggenbach Figure 1 Relative Cl, SO, and HCO, contents of thermal waters on weight (mg/kg) basis. For symbols see Table 1. Not much is gained by converting concentrations by weight (mg/kg) first to molar concentrations (mol/kg). Conversion to molar concentration units is only required in the application of more sophisticated thermodynamic techniques. Presentation of data in terms of such triangular diagrams has the advantage that none of the constituents is "prejudged” as is the case by plotting c.g. ratios where the constituent in the denominator assumes the role of a normalizing factor and can e.g. not be zero. In contrast to logarithmic plots, mixing lines in triangular diagrams remain always straight lines. The diagrams also allow the immediate “eyeball” statistical evaluation of groupings and trends. It should, however, be kept in mind that any such three-component plot tells only part of the story and that some apparent correlations may only be purely accidental. The close proximity of the data points for the White Island brine (WI), seawater (SW) and the Araro thermal spring (AR) in Fig. 1 may suggest, but in no way proves, a close genetic relationship among them. Correlations indicated by one set of components have to be checked by use of additional _independent.dat= dle telee al let| | Chemical Techniques in Geothermal Exploration tn many ways the triangular diagrams used here resemble the popular diagram proposed vy Fiper (1944) some 45 years. go for the classification of natural waters. Its major by aniage isthe use of sums of constituents dus obliterating any information carried by an igual consitvent. Also, the interpretative value ofthese diagrams i only minimal ifapplied ineothermal waters. They ae usvlly only presented a8 such and do not appear to encourage tear of deeper discussion of source components, formation processes or thermodynamic correlations. In Fig. 1 compositional ranges are indicated for several typical groups of thermal waters such a volcanic and steam-heated waters formed by the absorption of high temperature, Her-containing volcanic, or lower temperature, H,S-containing “geothermal” vapors, into Froundwaier. Most geochemical techniques are not suitable for aplication 1 these generally fe acd waters. The group best suited comprises the neutl, Tow sulfate, high chloride Teothermal” waters along the CI-HCO, axis, close to the Cl comer. Inthe case of albeit rautal, but high bicarbonate waters, considerable caution is required in the application of most *geoindicators”. Fig. 1 not only allows the weeding out of unsuitable waters, but may already provide an initial indication of mixing relationships or geographic groupings, with e.g. Cl waters forming fr eeatral core grading into HCO, waters towards the margins of a thermal area. High SO, dteam-heated waters are usvally encountered over the more elevated parts of afield. The degree of separation between data poins for high chloride and bicarbonate waters gives alreauy at idea ofthe relative degrees of interaction of the CO; charged fluids at lower temperatures, and of the HCO, contents increasing with time and istance travelled underground, a “This interpretation is supported by the shift in the position of data points representing spring discharges to higher relative HCO,-contents as compared to the well discharges at Neawha (NG), Miravalles (MV) and especially Zunil (ZU). ‘The compositional area where ionic solute geoindicators may be applied with sufficient confidence is quite small (hatched area in Fig. 1) and illustrates the need for this initial classification procedure. Relative Li, Rb and Cs contents The distribution of dzta points in Fig. 1 stresses differences among waters, largely brought about by comparative y shallow processes. An indication of the existence of a common origin, or of common, deep processes shaping the composition of surface water discharges, is provided by use of less reactive, “conservative” components, added at depth, but not affected by shallow processes, such as the rare alkalies Li, Rb and Cs. Relative Li, Rb and Cs contents of Fig. 2 again are ploted in terms of a trilinear diagram, Because of the large differences in the analytical values for the three components, normalizing or scaling factors had to be applied to achieve an even spread of data points over the plot. Omission of these scaling factors would have led to “bunching” of points in the Li comer of the plot. ' iFigure 2 Relative Li, Rb and Cs contents in the thermal water discharges on weight basis. “The numerical value given to these scaling factors depends on the choice of constellation of data points to be given special attention. Generally speaking, triangular diagrams can only cope with a spread of values for a given variable over two orders of magnitude. Any data points representing values outside this range are bunched along one of the axes with considerable loss of detail or resolution. By varying the scaling factors, different groups of data points can be shifted into the center of the diagram where resolution is highest; in other words they can be put under the magnifying glass. No matter which linear scaling factors are chosen, mixing lines always remain straight. Coordinates for Fig. 2 were calculated according to S = cy t Acq + 1 ® “GrLi" = 100JS ©) 124Chemical Techniques in Geothermal Exploration "GCs" = 1000ce/S 6 f this triangular plot, advantage is taken of its insensitivity to dilution id water or steam loss due to boiling of the rising fluids. Because of ths, dala points for deep and shallow waters, having undergone any amount of dilution oF iis oss, plt on the same postion. The preseniaton also allows the composition of average Shisil rocks from basalts to rhyolites to be shown, In the application o! py litle mineralized groun in geothermal systems with close volcanic-magmatic associations Water-rock interaction i hhat hypothetical, but well-defined endmember processes aay be described in terms of two somew! (Giggenbach, 1988): of crustal rocks in acid waters formed e.g. through absorption of magmatic vapors into deeply circulating groundwater. The waters formed in this reat dominated” alteration process contain most of the more soluble rock constituents in sroportions close to those ofthe orginal rock. The rck phase resulting from this process is Reavy depleted in the more easily leached components Initial dissolurion Final equilibrium of the fluid phase with "the rock" in ts thermodynamically stable mineral configuration formed through recrystalizaion of the origisal, thermodynamically Tnatable primary rock. The chemical composition ofthis secondary. stable rock pher> -"osrly agproaches, or isthe same (isochemical) a8 that of the original rock, This process is likely 10 arene ro completion only in stagnant systems of infinite age, The composition of the fluids in systems, however, is for a wide range of Al-silicate rocks such “rock-dominated” alter sniguely fixed and essentially only a function of temperature and salinity (chloride content, £ waters, gases and rocks in actual, dynamic hydrothermal alteration 10 correspond to the attainment of some steady state, intermediate between these two endmember processes. By comparing the compositions encountered in thermal fluid discharges with these theoretical compositions, however. its possible to assess the relative importance of the two processes in shaping the composition of a given discharge and of other processes active at depth within a geothermal system. The compositions of systems can be expected t 2 are generally far removed from the compositional area of crustal rocks, pointing to the existence of a number of secondary processes affecting ‘even these three "conservative" constituents during their rise to the surface. Ina detailed study of the behaviour of rare alkali elements during hydrothermal alteration in New Zealand's geothermal systems, Goguel (1983) found that Rb closely paralleled K in that iti already taken tp at quite high temperatures (> 300") in most K-contaning alteration products, such a illite, Geis incorporated into secondary zeolites, especially at temperatures <250°. Li is to minor degrees incorporated into secondary quartz and chlorite. Comparison of fresh and altered andesite and rhyolite in drillcores from these systems showed thatthe rare alkali’ metals have actually been added ducing rock alteration rather than leached, or in other words that they must have been dissolved from rock well below drilled depth. The data points of the waters in Fi ‘The only waters with compositions close to those of crustal rocks are the two highlyW.F. Giggenbach acidic waters from White Island (WI) and Aguas Calientes (RA), indicating rock dissolution. ‘All neutral waters have obviously lost Rb due to uptake in illite, shifting data points away from the Rb comer, LifCs ratios, especially of well discharges, are still close to those of the original rock. The increase in Li/Cs for spring discharges suggests removal of Cs through incorporation into zeolites at lower temperatures. The relative position of data points and the delineation of trends, therefore, allows e.g. the assessment of the position of thermal water discharges with respect to upflow (low Li/Cs) and outflow (high Li/Cs), and of relative degrees of water-rock interaction at high temperatures or acidities. LiCs ratios are also a function of rock type, lower in acid rocks as compared to basalts (Taylor, 1964). Relative Cl, Li and B contents “The alkali metal probably leas affected by secondary processes is Li. It may, therefore, be used as a “traces” for the initial deep rock dissolution process and as reference to evaluate the possible origin of the other two important “conservative” constituents of thermal waters, Cl and B. Fig. 3 shows their relative contents for the wide variety of thermal waters listed in Table 1. 1/100 \ on Owets O springs ESI ® '® ® a ways 19/0 NOL ® > § u 2 w0 60 --B BiM Figure 3 Relative Cl, Li and B contents in thermal waters on weight basis. For symbols see Fatiete~ 126 TFSo Chemical Techniques in Geothermal Exploration Of all these, only the very low chloride water, discharged at Fang, plots close to the composition of crustal rocks, suggesting that its compesition may be due t0 simple rock teaching. Relative Cl and B contents forall others are much higher suggesting either addition of these two elements before, during or after the rock dissolution process, or loss of Li. The tater is quite unlikely as there is no obvious major sink for this element in hydrothermal systems. Once added, Li remains largely in solution. It is, however, striking that both Cl and Bare added to the Li containing solutions in proportions close to those in crustal rocks. At high temperatures Cl occurs largely as HCI, and B as H,BO;. Both are volatile and able to be rrobilized by high temperature steam. They are, therefore, quite likely to have been introduced ‘with the magmatic vapors invoked above to lead to the formation of the deep acid brine responsible for rock dissolution. At lower temperatures the acidity of HCI increases rapidly, and it is soon converted by the rock to less volatile NaCl; B remains in its volatile form to be carried in the vapor phase even at lower temperatures. ‘The distribution of data points for most of the geothermal systems with close voleanic-magmatic associations, such as Wairakei (WK), Miravalles (MV) and Zunil (ZU) in Fig, 3 then may be explained in terms of dissolution of rock in waters formed through absorption of high temperature, high pressure magmatic vapors in deeply circulating groundwaters. ‘These vapors may contain Cl and B in proportions close to those of the rock contacted. Data points close to the Cl comer either represent waters formed through absorption of, in this case, low B/CI magmatic vapors, or waters affected by the admixture of '~«~ B/CL seawater (SW, MU, MO), Absorption of high B/CI steam, after neuualization oi =iCl, is indicated for Araro (AR) and especially Ngawha (NG). The B content of thermal fluids is likely to reflect to some degree the maturity of a geothermal system, Because of their volatility, constituents such as B, As, Sb and Hg are likely to be expelled during the early heating up stages. Fluids from “older” aydrothermal systems can be expected to be depleted in these elements. In spite of being affected to some degree by secondary chemical processes, the above five constituents Li, Rb, Cs, B and Cl provide very useful markers tor the identification of, and tracers to the origins of geothermal fluids. The remaining constituents listed in Table 1, Na, K, Mg, Ca and Si, are chemically more reactive and form the basis for a number of valuable geoindicators. ‘The Na-K-Mg-geothermometer Tonic solute geothermometers such as those based on Na-K- (Ellis and Mahon, 1967) and Na-K-Ca-contents (Fournier and Truesdell, 1973) provide powerful tools for the evaluation of deeper conditions within geothermal systems. Most of the problems in their use arise from their application to unsuitable samples. Initial weeding out on the basis of pH or their relative Cl, SO, and HCO, contents, (Fig.l), goes a long way to improve their reliability. Recently a “self-policing” technique was devised giving an automatic indication as to the suitability of a given water for the application of ionic solute geoindicators (Giggenbach, 1988). It is essentially based on the temperature dependence of the two reactions. K-feldspar + Na* = Na-feldspar + K* oOWF. Giggenbach and 2.8 K-feldspar + 1.6 water + Mg” = | 0.8 K-mica + 0.2 chlorite + 5.4 silica + 2K* (8) They both involve minerals of the full equilibrium assemblage expected to form after | isochemical recrystallization of an average crustal rock under conditions of geothermal interest. i Na, K and Mg contents of waters in equilibrium with this assemblage are accessible to rigorous evaluation, The theoretical temperature dependence of the corresponding concentration quotients then may be used to derive the two geothermometers — * s tyy = 1390K(1.75 - Lyy) - 273 © and uw = 4410/(14.0- Lyn) = 273 (10) where Ly, = log(cg/cy,) and Ly, = log (¢¢/cyq), ¢; in mg/kg. Individual application of equations (9) and (10) frequently leads to significa. div“erent apparent temperatures of equilibration. This observation is readily explained in terms of differing rates of re-adjustment of the two reactions to changes in the physical environment encountered by the rising waters. Reaction (10) was found to respond mich faster and, therefore, gives usually lower temperature estimates. The values of Lyq are also very sensitive | to the admixture of non-equilibrium acid waters, while L,, is much less affected by such shallow processes. By combining the two sub-systems, a method is obtained allowing the degree of attainment of water-rock equilibrium to be assessed and unsuitable samples to be eliminated. Again, it is based on the use of triangular diagrams. Because of the non-linear term ce, the square root is taken of the Mg contents and the coordinates for Fig. 4 are calculated j= according to \ S = cy,/1000 + cx/100 +VCyy ay | "foeNa" = cy/10S (12) I "%-Mg" = 100VCy,/S (13) Evaluation of the Na-K-Mg-geothermometer is best carried out by use of the blank diagram provided by Giggenbach (1988). Well discharges plot, as expected, close to the full equilibrium line at their deep temperatures. Their position somewhat beyond the theoretical line is due to steam loss increasing absolute solute contents of samples collected at atmospheric pressure. ‘They plot along 2 line marked *weirbox". 128Chemical Techniques in Geothermal Exploration Na/1000 {1g temperatures, For explanation of symbols see Table 1 Figure 4 Evaluation of Na-K. and Fig. 1. cers are all shifted, to varying degrees, towards the Mg corner. ‘or drops in 4,, but major decreases in tq. This trend reflects the much greater speed with which the K-Mg-geothermometer adjusts to a change in temperature as compared to the system K-Na. The shiit in K-Mg temperatures between well and spring discharges, therefore, is likely to be also a function of the time taken by the waters to rise from their deep storage horizons to the surface, The extremely large shift observed for Zunil agrees with the large increase in relative HCO; contents indicated in Fig. 1 and the large drop in relative Cs values of Fig. 2. all point to extensive interaction of the spring water with rock at comparatively low temperatures. Corresponding spring wat Their positions indicate only min ‘The high Mg content of the Zunil spring water places itin the compositional area marked siqnmature waters". For these frequently quite acid (WI, RA) or CO, rich (LN, WS) waters, application of both the K-Na- and K-Ng-geothermometer becomes doubtful, and great care Should be taken in the interpretation of the temperatures suggested. On the other hand, dataW.F. Giggenbach points for the two “formation” waters (MO, MU) probably reflect, at least, realistic Ke Mg-temperatures.. The position of MO onthe “full equilibrium’ line suggests that water-rock equilibrium is attained even for quite low temperature systems, given enough time. ‘The main advantage of Fig. 4, in addition to this preliminary evaluation of thermodynamic and Kinetic correlations, isits bility to depict the position of a large number of samples simultaneously, thus permitting the delineation of mixing trends and groupings. It 3IS0 voids the “lumping together” of variables as happens e.g. with the Na-K-Ca-geothermometer, Fig. 4 allows the effects of secondary processes on Na, K and Mg contents to be assessed individelly, Also, the positions of waters resulting from the two endmember processes, rock dissolution and equilibration, are well separated removing any doubt as to their relative importance in shaping the composition of a geothermal water. Fig. 4 may also be applied to Seawater driven hydrothermal systems. In this case, the effects of thermal processes may be acacesed in terms of the deviation of data points from the point marked SW. A presentation more auitzble to the evaluation of lower temperature waters (50° to 200°) is obtained by constructing a different diagram with cy,/400, ¢,/10 and Verge ‘A geothermomeier with rates of se-equilibration very similar to that of the K-Mg-geothermometer is based on dissolved silica contents. Extensive use is made of this technique in mixing models (Truesdell and Fournier, 1977), and a series of relationships Geseribing the temperature dependences of SiO, contents under a variety of conditions has been reported (Henley er al, 1984) At higher temperatures (> 200") silica contents 7orear t0 be governed by equilibrium with quartz, at lower temperatures with its microcrys.s.uce 7ofT, thalcedony.” A relationship allowing for ths gradual transition, and adequate in evaluating silica temperatures for spring discharges, is given by (1000/(4.55 - logesi2)) - 273 a4) 4 where csio: is in mg/kg. ‘The K-Mg-Ca-geoindicator ‘The remaining major cation in geothermal waters to be considered is Ca. Through the formation of calcite itis linked to the most important component of the vapor phase, COs. Jn addition to temperature-dependent reactions with Al-silicates, its concentrations in thermal waters can be expected to be affected by variations in the fugacity or partial pressure of CO;. The reaction governing CO; pressures in a “full equilibrium" system corresponds according Giggenbach (1988) to CarAlysilicate + K-feldspar + CO, = K-mica + calcite (15) “The temperature dependence of this univariat reaction is given with adequate precision (con in bar, t in °C) by logP con = 0.0168t - 3.78 as oOips ven be ca Chemical Techniques in Geothermal Exploration with the empirical CO; geothermometer proposed by Amérsson and Gunslavgsson (1985). The CO, pressures or contents associated with waters discharged irra ffom pools and springs, however, are virwally impossible toe determined with any ie of reliability. Rather than using the CO, contents themselves, it may be more useful to oot the effect of Peo, on other constituents ofthe deep fluid, such as Ca It agrees closely “The relationship linking CO; and Ca in a full equilibrium system is, 3K-feléspar + CO, + Cat? = K-mice + calcite + 2K* (17) ‘The equilibrium constant for this reaction can, for practical purposes, be considered to be temperature independent, its Peoe dependence corresponds to 10g (C’x!¢c) = 10g Peor + 3.0 «s) Le Eqn (18) would allow the direct evaluation of CO, pressures from measured K and Ca contents of surface water discharges. This information by itself, however, is only of use if Corresponding temperatures are also Known, as only both together allow e.g. the depth Giaritetion of the fluids to be evaluated. The geothermometer, which is likely to adjust with similar speed to changing conditions as the K-Ca-geobarometer, is the one involving K and Ms. By combining the two, 2 method is obtainet (Fig. 5) permitting the determination of Poth CO. contents and temperatures of "last equilibrium” of the waters with the rock Again, the method is best applied by use of a blank diagram provided by Giggenbach (1988). It requires analytical values of Ly. (eqn 18) to be plored versus Lan (ean 10). ‘Attainment of fluid-rock equi: brium again is indicated by the position of data points close to the full equilibrium line, Tt separates rock-domirated from fluid-dominated alteration systems. A second curve within the fluid-dominated field, sub-parallel to the full equilibrium line, marks the lower boundary, or the upper limits Of Pcoy. to which calcite can be expected to form from the increasingly acid waters. sopleths allowing ready conversion of Peo» t0 mol-ratis of CO, to HO, in the liquid (Feoxs) and vapor (Toor) Phase, are also given. They were calculated, assuming ideal gas behaviour, according to (9) Feorw = Peoo!Prno and (20) Teor = Fcory/Beor ‘The temperature dependences of the vapor pressure of liquid water and of the mass distribution coeffici 1980) ent Boos, for temperatures t from 100° to 340° are given by (Giggenbach, logPyoo = 5.51 - 2048/(t + 273) Qn logBeoo = Tcons/Fcoxs = 4.7593 - 0.01092 @W.F. Giggenbach 6 00° ty@(@C) | 290° 300° 2: 0.10 1 | i 3 100 j 4 0 ! Z i Figure _Evaluation of K-Ca-CO,-pressures and K-Mg-temperatures. Data points for the high temperature well discharges plot above the full equilibrium line and above the calcite formation field. Such behaviour implies that the CO, contents of the deep, rising waters are too low to induce rock alteration according to reaction (15). In other words, the CO, content of these waters is externally controlled, e.g. by the degree of ‘ion of the primary, CO;-contributing magmatic fluids with deeply circulating groundwater. 1 this case, the relation between Ly, and Poos provides only a qualitative indication of low CO, contents, and eqn (18) does not apply. On further rise and cooling, all the waters move into the calcite formation field, and CO; contents will be governed by reaction (15). If the fluids rise too fast, the rock system may not be able to cope with the amounts of CO, to be removed, and fluid compositions may move beyond the lower boundary of the calcite formation field. In this case, again, the correlation between Ly, and Peon breaks down, and the information obtained is only qualitative. Eqn (18) applies strictly only within the calcite formation field.ton) lon L | iol a | 7 > ¢ | | ww! i Chemical Techniques in Geothermal Exploration AAs stated above, the temperatures and CO; pressures indicated in Fig. 5 are those of "last equilibrium” of te waters with rock, They ae largely a fonction of the relative rates of ascent sae depth and the Kinetics of water-rock ineracion. Applied to a sufficiently large set of for et froma given aa, Fig. 5 allows detailed evaluation ofthe likely distribution of fluids serground and the delineation of major upflow zones. CONSTITUENTS PARTITIONING INTO THE VAPOR PHASE {As pointed out in the introduction, constituents of geothermal discharges can be grouped int either tracers or indicators. As with the liquid phase constituents, an inital evaluation of imaituents partitioning into the vapor phase is probably best carried out on the basis of a lusification scheme involving the less reactive tracer group with the hope to obtain information the possible origin of fluid phase components. on major processes governing their distribution nd on secondary processes possibly affecting their suitability for further thermodynamic valuation. ‘The most readily accessible group in thermal gas discharges consists of N;, Ar and He, A set of analyses (in mmol/mol) on water free basis of gas samples, associated with the waters of Table 1, is reported in Table 2 Relative N., He and Ar contents ‘Ar and He are noble gases, and chemically inert, Ny may take part in chemical reactions to forme.g. NH;. Generally, however, Nzis the predominant nitrogen species in thermal gases. On the basis of 2 large number of analyses of gases discharged from a wide variety of terrestrial sources, data points representing relative N;. He and Ar contents in a triangular diagram (Fis, 6) were found to occupy an area delineated by three major source components: 'A meteoric component, represented by air saturated groundwater, and contributing Ny ‘ang Ar in the molar ratio of 38 or somewhat higher, due to entrainment of air bubbles. The ratio He/Ar is <0.001. ‘A magmatic component, characterized by increased Ny/Ar ratios of around 800. but ratios of up t0 2000 have been observed. These very high N¥Ar ratios appear to be typical for “andesitic” gases associated with convergent plate boundaries. At divergent boundaries, in basaltic” gases, these ratios are usually much lower. The He content of the magmatic component is considerably higher than that of the meteoric component, (He/Ar=0.1) ‘A cmistal component made up largely of radiogenic He. Helium-+ contents can be expected to increase with increasing residence time of the gases in the crust. Free air, as indicated by the additional presence of O,, is always a contaminant, its Ny/Ar ratio is 84. “The origin of the high N; content of the magmatic component is still somewhat uncertain. Part of it may be derived through decomposition of crustal organic matter upon contact with oe EE 133WF. Giggenbach magmatic intrusion, it may directly be released from the mantle, or may have been transported with the subducted slab. The latter assumption is supported by the considerably lower relative N, content of gases associated with basaltic volcanism (Matsuo er al., 1978). A good correlation of N; with CO; and ‘He in andesitic gases suggests a very deep origin. Nofi00 1OHe - Figure 6 Relative N,, He and Ar contents in geothermal gas discharges on molar (volume) basis. For symbols see Table 2. The positions of data points in Fig. 6 cover the entire spectrum. ‘The White Island fumarole gas plots close to the magmatic endmember, as expected. The Miravalles (MV) and Zunil (ZU) well discharges are shifted towards the meteoric component indicating part ‘magmatic, part meteoric origin of the three gases. ‘The Miravalles fumarole gas is shifted towards the He comer indicating addition of an “older" component, The position of samples RA, RB and YA close to the datum for air may indicate air contamination during or after sampling. ‘The Ngawha samples occupy 2 position close to that usually observed for vapor 7 dominated systems. ‘The remainder of the samples follow a line connecting the meteoric component to the He comer. All these gases appear to contain only a minor magmatic component, even the gases | 134Chemical Techniques in Geothermal Exploration associated with convergent plate boundaries from Wairakei (WK) and Paraso (PR). The low sito for Lake Nyos is likely to reflect it basic seting, ‘The obviously non-voleanic gases EN MA, MO, MU represent essentially ar-satuated groundwater affected to varying degrees by the addition of radiogenic, crustal He. Fig. 6 allows an initial classification of gas samples in terms of their likely source components, the delineation of mixing relationships and the weeding out of samples affected by st contamination, those samples plotting in the vicinity of air being most suspect. Such fontamination may lead to partial or complete oxidation of reducing species such as H,S and, Sreourse, invalidates techniques based on species introduced with air contamination, such as Ar (sed below in the formulation of a number of geoindicators. ‘The Hy-Ar-geothermometer Most techniques 10 evaluate gas equilibration-temperatures are based on comparatively g four or more species. The major disadvantage of this aporeach is the uncertainty arising trom the implied assumption that all the reactants are still present inthe sample in proportions represenave of the deeper equilibration environment. The aeeree of uncertainty increases with the nuiber 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 (Giggenbach. 1987). complex activity quotients involvin: Of the gases listed in Table 2, hydrogen is the one reacting most quickly to changes in temperature and redox conditions (Giggenbze. 1987). As pointed out quite clearly by Taran (1986), most of the multi-component geothermometers involving CH, and NH; are actually hydrogen geothermometers. Their apparent success with well discharges (Giggenbach, 1980) is largely due to the rapid response of H, ard to the large powers with which it appears in coresponding equilibrium expressions, thus overiding variations in any ofthe other components involved. In accordance with this conclusion, Taran (1986) devised a relation based on absolute H, contents of vapor discharges. A disadvantage of this approach is the high sensitivity of all such geothermometers based on univariant reactions, such as the silica geothermometers, to secondary dilution, vapor loss and recondensaion processes and, in the case of gas samples, the need to determine representative gas/water ratios. The latter can be obtained, with limited reliability, only on well discharges, Most of these problems are overcome by formulating isomolar concentration ratios with another "well-behaved" constituent. Of the constituents listed in Table 2, CO; and H,S have been used by Amérsson and Gunnlaugsson (1985) to propose geothermometers based on the ratios Hy/CO, and H/H,S, although with limited success. Both these constituents are considerably more soluble in any coexisting liquid than H,, The proposed ratios can therefore be expected to be strongly affected by secondary boiling and recondensation processes. Also, CO; and H,S are likely to be much slower in their response to changing conditions and to be affected by a series of processes different from those controlling H contents.” He and N; show a similar solubility behaviour as H,, but have to be excluded as reliable partners because of their highly variable and ill-defined 135‘ i W.F. Giggenbach concentrations in thermal fluids. Methane is likely to be formed under equilibrium conditions but appears to become de-coupled from the H, system at t<250°. “The only remaining constituent of Table 2is Ar. It is chemically inert and is introduced almost exclusively with the meteoric component forming by far the bulk of geothermal water discharges. Measurements of “°Ar/Ar ratios show that contributions from radioactive sources are generally too low to affect "chemical" Ar contents significantly. Also, Ar contents of magmatic vapors appear to be very similar to those of air-saturated groundwater (Giggenbach, 1987). Being a minor component by mass anyway, the effect of magmatic Ar on absolute Ar contents of thermal fluids is probably minimal. Accepting the Ar of thermal fluids to be essentially derived from air-saturated groundwater, it may represent the "well-behaved" partner to be combined with H, to form a reliable and generally applicable geothermometer. Before forming such a partnership, itis essential to define first the dependence of Hy contents of thermal fluids on temperature. As pointed out on many occasions, the redox potential (e.g. Seward, 1974; D'Amore and Gianelli, 1984) and H, contents (Armérsson ané Gunnlavgsson, 1985) of thermal fluids appear to be controlled within very narrow limits, and a wide range Of potential redox buffers has been proposed. In order to bypass this somewhat futile exercise, Giggenbach (1987) suggested that redox conditions in a fully rock-equilibrated alteration system may with adequate precision be represented by one single value of -2.80 for the variable Ry = log (fo!fieo), Valid from room temperature to 1200°. It was obtained by assuming redox conditions to be controlled by the Fe({T)-Fe(II1) content of the rock phase Most geothermal systems are liquid dominated; equilibration, even among gaseous constituents, can therefore be assumed to take place largely in dissolved form. The equilibrium mol-ratio of Hy dissolved ria.1 may be oblained from the above fugacity ratio according to = fiolfacBa = tadBo = 10°/Ba (23) Where Byy is the vapor/liquid distribution coefficient of H. Its temperature dependence (Giggenbach, 1980) is given by To. logBy. = 6.23 - 0.014t (24) Values of figs may now be related tothe value of f,. Of 0.310" typical for air-saturated groundwater, ‘The temperature dependence of the Hy/AT mol-ratio in the equilibration liquid phase then corresponds to Lin = logteia!Xd = -2.8-(-6.52)- 6.23 + 0.014 = 0.014t-2.5 @5) or ty = 702.5 + Lu) (26) 136Chemical Techniques in Geothermal Exploration ecause of the very low and similar solubilities of H, and Ar, both can be assumed to virtually completely into the first few percent of vapor formed from the original eriynum iquid-phase and, therefore, to reach the surface in proportions represealaive of he sauier environment. The values Of %q and &, rewired may be in any common molar centration unit, e.g those reported in Table 2 on a wate fre basis, or simply raw data Spained directly from the gaschromatograph. ition Comparison of analytical values of Ly, with the theoretical line (Fig. 7) for the samples ised in Table 2 shows thatthe geothermometer “works” well down to quite low temperatures. The temperatures obiained for well discharges ae also close to those actually measured. In the aie of Ngawha (NG), where temperatures over much of the drilled horizons were around 230°, the tyx's of > 300° may reflect the high temperatures encountered in the deep well NG-13. The ee may, however, also have been affected by loss of Ar, through even minor vapor loss, followed by H, re-equlibration. ‘This process would shift data-points to higher values, towards the "vapor equilibration” line. similar upward shift should be observed for vapor-dominated systems where some ofthe equilibration can be assumed to have taken place in the vapor phase, In the case of boiling spring and fumarole discharges, values of Ly, appear to be almost perfectly quenched, thus providing valid information on conditions within the deep equilibration Tiquid phase. Application of eqn (26) depends vitally on the availability of reliable and uncontaminated values for Xy.. It therefore can be expected to give best results for samples which plot farthest from air in Fig. 6. Also, in the application of eqn (26) to lower temperature systems. itis ‘eseemtial that H, and He are analytically sepersied. In many such systems He contents may far uceed Hy and, if analysed together, may give rise to too high values of ty. Ar and He are readily analyzed by the techniques given in Giggenbach and Goguel (1989). The good correspondence of theoretical and observed temperatures suggests that the highly simplified framework, used to derive eqn (26), provides an adequate and possibly @ generally valid model of the actual processes occurring in hydrothermal systems. As the reference gas, Ar may be used as the reference gas, i may be used to derive a variety of other geoindicators, Within this introduction only some aspects of reactions involving the carbon gases will be discussed in some detail. Reactions involving CO,, CH, and CO ‘Assuming ronan the mol-ratio of CO; to H,0 in the deep equilibrium liquid phase, to be given by eqn (19), it may again be combined with ry, 0 give the COy/Ar geothermometer Lea = los(Xcov Xa) Tog (Peoe! PinoBeor! 0.0277t - 7.53 + 2048/(t+273) enW.F. Giggenbach ! r bax " i addition of equilibrium i vapor oF loss of argon 3 ! 2 ~ 1 i ° i | 4 i tye 70(2.5+ Lua) i 2 be Log by, Mar) ; 0 00° 200° 300° (°C) e Figuie 7 Evaluation of Hy-Ar-temperatures. Values of Ly are plotted at the measured temperatures for springs, at deep temperatures for wells. Because of the quantitatively much larger amounts of CO, involved in any re-equilibration process, as compared to H,, this geothermometer can be expected to get “out of step" much more readily. As with the Na-K-Mg-geothermometer (Fig. 4), combination of two reactions, attaining equilibrium with different velocities, may provide a technique allowing ‘an assessment to be made of the overall state of attainment of equilibrium, possibly useful in any attempt at further thermodynamic evaluation. In Fig. 8, values of Ly, (eqn (25) are plotted versus Le, (eqn (27)). Again a full equilibrium line is defined for equilibration of all gases dissolved in a single liquid phase. The horizontal line corresponds to compositions expected for equilibration in the vapor phase lines describe intermediate conditions, the addition of equilibrium vapor (Giggenbach, 1980), or the loss of Ar prior to re-equilibration. Lack of adjustment of CO, contents to lower ™. ~ — 138i Chemical Techniques in Geothermal Exploration temperature conditions, with Hy remaining in equilibrium, is indicated by a shift of datapoints downwards. HylAr-temp Le eg) «~—— ©© © 1 2 3 4 5 la 6 Figure 8 Evaluation of H,-Ar, CO,-Ar equilibration conditions. Comparison with data reported in Table 2 shows that most high-temperature spring and fumarole discharges fall close to the theoretical liquid phase, full equilibrium line. As discussed above, the position of the Ngawha samples points to admixture of equilibrium vapor or loss of Ar. Data-points well below the full equilibrium line represent obviously “immature” vapors, far from overall equilibrium. For these, tio temperatures may be read off Fig. 8: a HAT Temperature, generally quite close to actually measured discharge temperatures, and a COAr 139Be x ~ W.F. Giggenbach temperature, possibly representing conditions in some deeper environment. The lack of attainment of full equilibrium for these vapors is likely to be due to their limited contact with rock, either because of a too rapid transition from the deep high-temperature zones to much cooler shallow conditions or highly restricted porosities preventing access of the rising waters to fresh rock surfaces. ‘An ubiquitous constituent of thermal vapors is CH,. Its use in geothermometers is controversial (Arnérsson and Gunnlaugssson, 1985; Taran, 1986). It is obviously one of the slowest species (Giggendach, 1987) to equilibrate. As such, it has the potential to form the basis for a very “deep” geothermometer, remembering well conditions long forgotten by the other, “faster” constituents. In the derivation of any such geothermometer it is obvious that it should not be combined with any such fast species lest the information carried by it be swamped or mutilated by the noise generated by any ill-chosen partner. In this case, CH, is best used in combination with another slow or non-reactive species such as Ar. Such a CH,/Ar ‘geothermometer is readily devised using the procedures outlined above. Because of the unique way in which CH, is linked to the other carbon species CO, and CO, an approach involving all three is chosen here. In order to allow a species by species assessment of the effects of actual or potential secondary processes, special care is taken not to lump the three carbon species together into complex, multi-component equilibrium expressions. ‘The technique chosen is based on a diagram already used earlier to evaluate reactions controlling the composition of volcanic gases from White Island (Giggenbach, 1987) and consists of plotting Tog(xeys/Xeo:) versus log (Xco/Xco2) (Fig. 9). Two major redox buffers are considered: that involving di- and tri-valent iron of the rock, marked (FeO)/(FeO, «) and that involving the coexistence of H,S and SO; of the magmatic vapor phase. The latter is likely to be active only in high temperature volcanic systems; its effects may, however, still be felt over the deeper high temperature parts of geothermal systems or in systems where the magmatic environment reaches shallow levels. For the rock buffer two curves are shown: one representing equilibration in the vapor phase, the other in the liquid phase, Isotherms refer to attainment of equilibrium among all three species according to the reaction 3CO, + CH, = 4CO + 2H,0 (28) Where they cross a buffer line, these temperatures define equilibrium conditions also for this system. In the upper right-hand part of the diagram the isotherms refer to equilibration of the gases in the vapor phase. On crossing the rock-buffer, vapor line, the short sections connecting with the rock-buffer, liquid line describe the compositions of mixtures of liquid and vapor equilibrated gases. Over the lower parts of the diagram, the isotherms correspond to the composition of the three gases having equilibrated in a single liquid phase. Carbon monoxide is determined only rarely on geothermal vapor discharges, largely because of its very low concentrations in lower temperature (<280*) systems. The advent of specific and highly sensitive detectors for its analysis has led to the development of a number 140 i } iChemical Techniques in Geothermal Exploration -8 7 -6 -5 4 L0atrco/%eo,) { | Figure 9 Evaluation of CHy, CO-CO; equilibration conditions. i (pf techniques exploiting the special properties of reaction involving CO, especially by Tatian investigators (Bertrami er al., 1985; D'Amore et al., 1987; Chiodini and Cicni, 1989). In conjunction with CH, and CO, contents, data-points may be plotted in Fig. 9. Two examples representing Larderello, and a 150° fumarole in the Campi Flegrei (CF) are shown. As [ | 141W.F. Giggenbach expected, the Larderello vapors plot within the Tiquid-vapor coexistence field at temperatures somewhat higher than those measured at the wellhead pointing to equilibration at deeper, hotter Tevels. ‘The position ofthe point forthe Campi Flegrei below both rock-buffer lines, suggests (_) equilibration under somewhat more oxidizing conditions possibly reflecting the effects of the presence of more oxidizing magmatic fluids at quite shallow levels. \ Fig. 9 actually allows the evaluation of six different apparent temperatures of | equilibration depending on which environment or process is assumed to govern interaction ‘among the three gases. Assuming homogeneous equilibration of all three gases dissolved in a liquid phase, a temperature (1) of around 270° is indicated; if equilibrated in a vapor phase, the temperature (2) would be only about 220°. Assuming CO and CH, to have equilibrated independently, a CO-tock temperature (3) of 230° would be obtained if equilibrium was attained! in 2 liquid phase, only 145° (4) if equilibrated with rock in the vapor phase. A much higher temperature (5) would result if CO is assumed to have equilibrated deep within the system where - redox conditions are still governed by the sulfur buffer. Ina similar way, CH4/CO; ratios may be used to evaluate independent temperatures. For the Campi Flegrei (CF) only one temperature of around 420° is obtained, that for equilibrium with the rock buffer. Because of values of B, approaching unity, liquid and vapor ratios at a temperature >400° become identical. For most of the gases listed in Table 2, it is safe to assume equilibration with the rock in a single liquid phase and the rock-liquid line is likely to apply. Temperatures for most are >300°, probably reflecting the slow rates of methane equilibration thus preserving deeper temperatures. They may be obtained from the simplified relationship of the methane-carbon dioxide geothermometer The values for the two low temperature formation waters, MU and MO, are with 140° and 86° quite reasonable and suggest that methane may eventually attain rock equilibrium, given. enough time. Fig. 9 then allows a series of apparent equilibrium temperatures involving the three carbon gases to be evaluated without the need to make a prior, frequently quite arbitrary decision as to the actual circumstances of equilibrati ‘ | tc = (4625/(10.4 + log(Keps/Xco))) + 273 (29) i REFERENCES ‘Amérsson S. and Gunnlaugsson E. (1985) New gas geothermometers for geothermal exploration - Calibration and application. Geochim. Cosmochim. Acta, 49, 1307-1325. Bertrami R., Cioni R., Corazza E., D'Amore F. and Marini L. (1985) Carbon monoxide in i geothermal gases. Reservoir temperature calculations at Larderello (Italy). Geothermal & Res. Coune., Trans., 9, 299-303. & Chiodini G. and Cioni R. (1989) Gas geobarometry for hydrothermal systems and its application to various Italian geothermal areas. Appl. Geochem., 4.Chemical Techniques in Geothermal Exploration p’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. p'Amore F., Fancelli R., Saracco L. and Truesdell A.H. (1987) Gas geothermometry based on CO content -- Application in Italian geothermal fields. Proc. Stanford Reservoir Eng. Workshop, 12, 247-251. Ellis A.J. and Mahon W.A.J. (1967) Natural hydrothermal systems and experimental hot water/rock interactions (Part II). Geochim. Cosmochim. Acta, 31, 519-538. Fourier R.O. and Truesdell A.H. (1973) An empirical Na-K-Ca geothermometer for natural waters. Geochim. Cosmochim. Acta, 37, 1255-1275. Giggenbach W.F. (1980) Geothermal gas equilibria. Geochim. Cosmochim. Acta, 44, 2021-2032. Giggenbach W.F. (1987) Redox processes governing the chemistry of fumarolic gas discharges from White Island, New Zealand. 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. Giggeabach W.F. and Goguel R.L. (1989) Collection and analysis of geothermal and volcanic water and gas samples, DSIR New Zealand, Report No. CD 2387, pp. 53. Goguel R.L. (1983) The rare alkalies in hydrothermal alteration at Wairakei and Broadlands geothermal fields, New Zealand, Geochim. Cosmochim. Acta, 47, 429-437. Henley R.W., Truesdell A.H. and Barton P.B. (1984) Fluid-mineral equilibria in hydrothermal systems. Reviews in Econ. Geol., 1, 267. Matsuo S., Suzuki M. and Mizutani Y. (1978) Nitrogen to argon ratio in volcanic gases. Adv. Earth Planet., Sci., 3, \7- 25. Piper A.M. (1944) A graphic procedure in the geochemical interpretation of water-analyses. Amer. Geophys. Union, Trans., 25, 914- Seward T.M. (1974) Equilibrium and oxidation potential in geothermal waters at Broadlands, New Zealand. Amer. J. Sci, 274, 190-192. Taylor $.R. (1964) Abundance of chemical elements in the continental crust, a new table, Geochim. Cosmochim. Acta, 28, 1273-1285. Taran Y. (1986) Gas geothermometers for hydrothermal systems. Geochemistry Int., pp. 111-126.W.F. Giggenbach Sigs - ‘Truesdell A.H. and Fournier R.O. (1977) Procedure for estimating the temperature of hot water component in a mixed water using a plot of dissolved silica versus enthalpy. J. Res., US Geol. Survey 5, pp. 49-52.