Mobility and fluxes of major, minor and trace metals during basalt weathering and groundwater transport at Mt. Etna volcano (Sicily)

PII S0016-7037(00)00345-8

Mobility and fluxes of major, minor and trace metals during basalt weathering andgroundwater transport at Mt. Etna volcano (Sicily)

ALESSANDRO AIUPPA,1 PATRICK ALLARD,2 WALTER D’A LESSANDRO,3,* A GNES MICHEL,4 FRANCESCOPARELLO,1 MICHEL TREUIL,4 andMARIANO VALENZA

1

1Dipartimento di Chimica e Fisica della Terra, CFTA, via Archirafi 36, 90123 Palermo, Italy2Laboratoire des Sciences du Climat et de l’Environnement, CEA-CNRS, Orme des Merisiers, 91191 Gif/Yvette, France

3Istituto Geochimica dei Fluidi-CNR, via La Malfa 153, 90146 Palermo, Italy4Laboratoire Pierre Sue, CEA-CNRS, Saclay, 91191 Gif/Yvette, France

(Received May25, 1999;accepted in revised form February2, 2000)

Abstract—The concentrations and fluxes of major, minor and trace metals were determined in 53 samples ofgroundwaters from around Mt Etna, in order to evaluate the conditions and extent of alkali basalt weatheringby waters enriched in magma-derived CO2 and the contribution of aqueous transport to the overall metaldischarge of the volcano. We show that gaseous input of magmatic volatile metals into the Etnean aquifer issmall or negligible, being limited by cooling of the rising fluids. Basalt leaching by weakly acidic, CO2-charged water is the overwhelming source of metals and appears to be more extensive in two sectors of theS-SW (Paterno`) and E (Zafferana) volcano flanks, where out flowing groundwaters are the richest in metalsand bicarbonate of magmatic origin.

Thermodynamic modeling of the results allows to evaluate the relative mobility and chemical speciation ofvarious elements during their partitioning between solid and liquid phases through the weathering process. Thefacts that rock-forming minerals and groundmass dissolve at different rates and secondary minerals are formedare taken into account. At Mt. Etna, poorly mobile elements (Al, Th, Fe) are preferentially retained in the solidresidue of weathering, while alkalis, alkaline earth and oxo-anion-forming elements (As, Se, Sb, Mo) are moremobile and released to the aqueous system. Transition metals display an intermediate behavior and arestrongly dependent on either the redox conditions (Mn, Cr, V) or solid surface-related processes (V, Zn, Cu).

The fluxes of metals discharged by the volcanic aquifer of Etna range from 7.03 1023 t/a (Th) to 7.33104 t/a (Na). They are comparable in magnitude to the summit crater plume emissions for a series of elements(Na, K, Ca, Mg, U, V, Li) with lithophile affinity, but are minor for volatile elements. Basalt weathering atMt Etna also consumes about 2.13 105 t/a of magma-derived carbon dioxide, equivalent to ca. 7% ofcontemporaneous crater plume emissions. The considerable transport of some metals in Etna’s aquifer reflectsa particularly high chemical erosion rate, evaluated at 2.3*105 t/a, enhanced by the initial acidity of magmaticCO2-rich groundwater. Copyright © 2000 Elsevier Science Ltd

1. INTRODUCTION

Chemical weathering of crustal rocks is one of the principalprocesses controlling the geochemical cycle of elements at theEarth surface. Chemical erosion consumes atmospheric and/orendogenous carbon dioxide and extracts metals from the rocks,these latter being released to rivers and shallow groundwatersand finally discharged into the ocean (Garrels and MacKenzie,1971). Among rocks of volcanic origin, basalts are particularlysensitive to chemical erosion (Berner and Berner, 1996). There-fore, basaltic systems are useful to investigate the partitioningof chemical elements during this process.

Different approaches were developed to assess the geo-chemical mobility of elements during basalt weathering. Onefirst approach is to study the mineralogy and chemistry ofweathering profiles (Carr et al., 1980; Chesworth et al., 1981;White, 1983; Nesbitt and Wilson, 1992) and to infer a sequenceof element mobility from a group of rocks characterized by anincreasing degree of alteration. Although the results cannot beeasily generalized, a common finding is that basalt dissolutionis incongruent, since different primary minerals do not dissolve

at the same rate and different elements are partitioned betweensolution and secondary minerals in response to the aqueouschemistry. In such a framework, immobile (Al, Fe, Ti) andmobile (Na, K, SiO2) elements are commonly distinguished.

Studying the geochemistry of aqueous fluids is a secondalternative approach to gain insight into weathering processes.Generally, the relative mobility (RM) of elements is calculatedfrom their water/rock concentration ratio, normalized to so-dium:

RM 5 ~X/Na!w/~X/Na!r

wherew and r refer to the solution and the fresh host rock,respectively. This approach has been successfully applied torivers draining basaltic terrain in Iceland (Gislason et al., 1987;Gislason et al., 1996; Louvat, 1997), in the French MassifCentral (Meybeck, 1986) and in basaltic volcanic islands (Re´-union, Java and Sao Miguel; Louvat, 1997). Fluvial erosionconsumes atmospheric CO2 and produces soluble compoundswhich are later discharged into the sea, thus acting as a regu-lation system of the Earth’s environment (Gaillardet et al.,1995; Stallard, 1995; Berner and Berner, 1996). In this context,volcanic activity, which releases acidic and reduced gases tothe surface, may produce extreme conditions in which weath-

* Author to whom correspondence should be addressed ([email protected]).

Pergamon

Geochimica et Cosmochimica Acta, Vol. 64, No. 11, pp. 1827–1841, 2000Copyright © 2000 Elsevier Science LtdPrinted in the USA. All rights reserved

0016-7037/00 $20.001 .00

1827

ering processes and the resulting geochemical mobilities ofmajor and trace elements are deeply modified.

Here we report a detailed investigation of the geochemicalprocesses controlling the major, minor and trace element com-position of groundwaters hosted by Mount Etna, the largest andmost active basaltic volcano in Europe. The geochemistry ofEtnean groundwaters has been a matter of growing interest overthe past decade (Anza` et al., 1989; Allard et al., 1997; Bonfantiet al., 1996a; Bonfanti et al., 1996b; Giammanco et al., 1998;Brusca et al., in press), with the aims to understand and monitorthe relationships between groundwaters and the activity andstructure of the volcano. Based on chemical and isotopic data,it has been demonstrated that the aquifer of Etna is affected byinputs of CO2 and He-rich magmatic gas (Anza` et al., 1989;Allard et al., 1997) and is sensitive to medium-term changes inthe volcano plumbing system (Bonfanti et al., 1996a; Bonfantiet al., 1996b). Because the gas-charged groundwaters becomestrongly aggressive with respect to the host basalts, intense rockweathering occurs as the groundwaters flow through the highlypermeable volcanic strata, even though their residence timemay be short (often less than 2 y; Ferrara, 1975). The resultingchemical heterogeneities in the out flowing waters are limitedby three factors:

1. all waters are meteoric in origin (Anza` et al., 1989; Allard etal., 1997);

2. their low temperature and lack of18O-shift with respect tometeoric values exclude the influence of hydrothermal wa-ter-rock interaction (Anza` et al., 1989; Allard et al., 1997;Giammanco et al., 1998); and

3. etna lavas have a rather monotonous mineralogy and chem-istry (e.g., Tanguy et al., 1997).

Such conditions allow a quite simple modeling of the system,as the number of principal variables is restricted to the initialpH of the waters (determined by the amount of gas dissolved inthe aquifer), their redox potential (possibly influenced by thepresence of reduced species such as H2S and CH4 in the gasphase) and their residence time (or length of pathflow). Kineticeffects may additionally affect the system.

Hence, Mt. Etna offers a very good framework to study thegeochemical behavior of major, minor and trace metals duringbasalt weathering by groundwaters enriched in magmatic gas.In this work we present a detailed analysis of these elements in53 samples of Etnean groundwaters collected from all aroundthe volcano. These results are used to model the behavior of theelements during basalt weathering and, combined with thewater flow rates, to evaluate the contribution of groundwatertransport to the overall release of metals by Mount Etna.

2. HYDROGEOLOGY OF MT. ETNA

Mt. Etna, located on the eastern coast of Sicily, is a largestratovolcano (elevation, 3350 m a.s.l.; area, 1150 km2; perim-eter, 250 km) which has built upon tensional faults cutting a'20 km thick continental crust (Chester et al., 1985). Etneanvolcanism is related to the breakup of the African plate marginsince the Upper Miocene, during its collision with the Europeancontinental block (Barberi et al., 1974). Mt. Etna volcanismbegan at 0.5 Ma and consists of a lower shield unit overlain bya stratovolcano (Chester et al., 1985). The shield complex

formed over Miocene flysch sediments (rising to ca. 1300 melevation) to the NW and clayey Pleistocene formations to theSE. The present-day activity of Mt. Etna is characterized byfrequent summit and lateral eruptions (Tanguy et al., 1997) andhuge emissions of magmatic volatiles from both the summitcraters and the upper flanks, the latter as diffuse soil emanations(Allard et al., 1991; Allard et al., 1997). These emissions resultfrom open-conduit degassing of alkali basalt-hawaiite magmawhich rises from a shallow mantle diapir (D’Alessandro et al.,1997a; Hirn et al., 1997).

The average precipitation rate over Mt. Etna area is 800mm/a, 75% of which actually infiltrates into the highly perme-able volcanic strata. Evapo-transpiration accounts for 20%,whereas surface water run-off along the flanks of Mt. Etna is5% (Ogniben, 1966). The two main rivers, Alcantara andSimeto Rivers, flowing along the basis of Mt. Etna and dis-charging to the Ionian sea, are mainly fed by surface run-offfrom the Apennine-Maghrebian chain (Northern Sicily) and bygroundwaters from Mt. Etna (Aureli and Musarra, 1975).

The Etnean aquifer is hosted by the permeable basaltic lavasthat form the volcanic pile. As demonstrated by previous hy-drogeochemical studies (Anza` et al., 1989; Allard et al., 1997),all groundwaters are meteoric in origin. They seep rapidly intothe fractured volcanic strata and flow at the contact with theunderlying impermeable sedimentary basement, which forms aplane inclined towards the SE (Ogniben, 1966; Ferrara, 1975).Ogniben (1966) calculated a total input of 0.69 km3/a of waterin the Etnaen aquifer. Most cold springs discharge from thelowest part of the volcanic edifice, at the contact with sedimen-tary rocks, and many have a high flow rate (up to 103 l/s,Ogniben, 1966; Ferrara, 1975). Scarce springs outflow at ahigher elevation and with a low rate, mainly on the easternflank of the volcano, fed from small aquifers confined at theirbase by impermeable volcanoclastic deposits (Kieffer, 1970). Asignificant proportion (about 30%) of the whole water outflowfrom Etna occurs beneath seawater along the Ionian coast(Ogniben, 1966). Three principal hydrogeological basins (Fig.1), tributaries of the Alcantara (N) and Simeto (W) rivers andthe Ionian Sea (E), were distinguished by Ferrara (1975, 1990).Estimates of effective aquifer Darcy’s velocity (Ferrara, 1975,1990) and data for tritium (Allard et al., unpub. results) indicaterather short residence times of a few months to about a year formost waters, except for more saline groundwaters on the lowSW slopes, which have deeper pathflows and higher residencetime in the aquifer.

3. METHODS

Figure 1 is a schematic map of Mt. Etna volcano showing thelocation of our sampling points (circles). From a set of 150 samplescollected and analyzed for major elements and dissolved gases, weselected 53 samples on which we additionally determined the contentof minor and trace elements. The sampling points account for a totalwater out flow of 0.13 km3/a, equivalent to;20% of the overalldischarge of Etnean aquifer (0.69 km3/a; Ogniben, 1966). Most waterswere taken from drilled wells used for irrigation and drinking watersupply, but some springs are also included in our data set. Watertemperature, pH, Eh and conductivity were measured upon sampling.Values of pH, Eh and conductivity are referred to the sampling tem-perature. Samples were collected and stored in either LDPE (low-density polyethylene, major elements) or HDPE (high-density polyeth-ylene, trace elements) flacons, pre-washed in ultra-pure HNO3-aQgrade Millipore water. Major elements were analyzed by ion chroma-

1828 A. Aiuppa et al.

tography on unfiltered (Cl and SO4) and filtered (Na, K, Ca, Mg)samples. Alkalinity was determined by titration with HCl (0.1 N). Ionicbalance was computed for each sample taking into account major andminor species. All samples exhibited imbalances lower than 5%. Sam-ples to be analyzed for trace elements were filtered upon samplingthrough 0.45mm Millipore MF filters and then acidified to pH' 2 withultra-pure HNO3. Trace element analysis was carried out in Pierre SueLaboratory (CEN-Saclay) with a Fison Plasmaquad PQ21 ICP-MS.All determinations were performed with the external standard calibra-tion method, using In and Rh as internal standards. The accuracy of theresults (65%) was obtained by analyzing certified reference materials(SLR 2, SLR 3 and TM-28, NWRI Canada Center for Inland Waters).The results for both major and trace elements in Etnean groundwatersare summarized in Tables 1 and 2.

Speciation calculations for major and trace elements were performedusing PHREEQC software (Parkhurst, 1995). This program solvessimultaneously a set of functions (mole-balance equations and mass-action expressions) in order to determine the aqueous activities ofdissolved species at thermodynamic equilibrium in a given set ofconditions (T, pH,p«, total content of each metal in solution). Thedistribution of redox elements among their different valence states isassessed from the measuredp« values (p« 5 2log ae2) of watersamples. The PHREEQC database, extended with data for trace metalcomplexes from WATEQ4F and MINTEQ, was used in modeling. Themain limitation of the software is due to restricted internal consistencyin the database, as thermodynamic data were taken from variousliterature sources (Parkhurst, 1995).

4. RESULTS

Table 1 shows that the sampled waters are characterized bylow temperatures (8.4 to 22.6°C) and low conductivities (188 to1885mS/cm) and that bicarbonate is generally the main anionin solution (79.3 to 2190 mg/l). Although some compositionalheterogeneities exist, most Etnean groundwaters have a typicalbicarbonate alkaline-earth composition. Previous studies (Anza`et al., 1989; Allard et al., 1997; Giammanco et al., 1998; Brusca

et al., in press) have emphasized the key role of magma-derivedCO2 which lowers the pH (initial pH; 4) of the water ondissolution, causing leaching of the host rocks. Despite theirstrictly meteoric origin, many Etnean groundwaters displayCO2 partial pressure up to 1–3 orders of magnitude higher thanthe atmospheric value (1023.6 atm). Allard et al. (1997) andD’Alessandro et al. (1997b) have demonstrated the magmaticorigin of this CO2. During basalt weathering, CO2 is graduallyconverted into bicarbonate, which displays positive correlationswith the cation contents (Fig. 2) and a negative correlation withpH, the latter ranging between 5.94 and 7.86. This patternindicates that the amount of metals released to groundwaters isdetermined by the intensity and extent of acid attack of the hostrocks by gas-charged water. The highest concentrations ofmajor elements are observed in the most HCO3-rich ground-waters which outflow from the south-southwest (Paterno`) andeastern (Zafferana) flanks of the mountain (Anza et al., 1989;Brusca et al., in press). The waters in these two sectors alsodisplay the highest contents of magma-derived helium (Allardet al., 1997).

The Mg-rich composition of Etnean groundwaters, fallingamong the so-called “immature waters” defined by Giggenbach(1988), confirms that the weathering process acts in a very coldand shallow environment, where thermodynamic equilibriumbetween rock and solution is never reached. This facts preventsthe use of classical aqueous chemical geothermometers (i.e.,amorphous SiO2, Na/K, Na/Li), which do provide unrealisticequilibrium temperatures for all samples. The S. Venera ther-mal spring (sample 27) is one exception, in that it is heavilyaffected by thermal saline brines of marine origin rising fromthe sedimentary basement. This is reflected by the high Na, Li,B and Cl contents, by its higher emergence temperature(22.5°C) and by the rather high equilibrium temperatures givenby cation geothermometers (110–120°C).

Table 2 summarizes the results of minor and trace elementanalyses. Minor elements (Fe, Sr and Mn) display a rather wideconcentration range (from 0.1mg/l up to 10 mg/l), whereasmetal content for ultra-trace elements (Th, Pb, Sb and Cd) iscomprised between 0.01 and 1mg/l. As suggested by Aiuppa etal. (1998) and Brusca et al. (in press), the wide concentrationrange for a given element can be explained by a variable extentof water-rock interaction. As noted by Brusca et al. (in press),the highest concentrations of Li, Rb, Cs and Sr in the samplescoincide with maximum contents of Na, Mg, Ca and HCO3

(Fig. 2c), evidencing enhanced basalt leaching in response togreater dissolution of CO2. The contents of transition metal alsoincrease with increasing bicarbonate content (Fig. 2b), butdepend on the redox conditions as well. Figure 2b shows thatsome groundwaters from the Paterno` area, where reduced con-ditions prevail (210, Eh, 100), are considerably enriched inFe, Mn and Ni. The presence of groundwaters with low Ehvalues near Paterno` and Adrano villages is related to the rise ofreduced gases (mainly H2S and CH4) from deep gas reservoirshosted in the sedimentary basement (Chiodini et al., 1996;D’Alessandro et al., 1997b). Moreover, the redox conditions ofEtnean waters are dependent on hydrological factors, as re-duced waters are characterized by longer and deeper pathflowsand thus higher residence time in the aquifer. This in turnresults in oxygen depletion throughout weathering and in a

Fig. 1. Schematic map of Mt. Etna and location of water samplingsites (open circles). The gray line represents the boundary betweenvolcanic and sedimentary rocks. The main hydrogeological basinsdefined by Ferrara (1975, 1990) are delimitated and indicated as a, band c. UTM grid is also shown, vertical and horizontal lines are drawneach 10 km.

1829Mobility of metals during basalt weathering at Mt. Etna

higher degree of interaction with deep reduced fluids comingfrom the sedimentary basement.

Table 3 shows the average metal concentrations in the Et-nean aquifer, as computed from the whole data set (53 sam-ples), together with the average concentrations of trace metalsin Etnean lavas, as derived from the literature (Cristofolini andRomano, 1982; Joron and Treuil, 1984; Pennisi et al., 1988;Treuil and Joron, 1994; Michaud, 1995; and Tanguy et al.,1997) and INAA and ICP-MS analyses we performed at Lab.

Pierre Su¨e (unpublished data). Figure 3 is a scatter plot of theaverage concentration of each metal in groundwater (Cw) andin Etnean basalt (Cs). The good correlation between these twovariables suggests that the rock composition plays a major rolein controlling groundwater chemistry. Figure 3 also indicatesthat metal partitioning between rock and solution depends onthe chemical behavior of the elements. With respect to thegeneral trend, Al, Th and, to a lesser extent Fe, appear to bedepleted in groundwater, which is consistent with their strong

Table 1. Concentrations of major element in studied Etnean groundwaters.

N. Sample Type UTM Date Flow T°C pH Cond. Eh Na K Mg Ca Alk. Cl SO4 SiO2

1 Valcorrente G 0495741566 15.01.97 50 14.8 6.55 1298 653 112 27.4 131 88.8 1060 57.8 30.2 77.52 Vena S 0512141829 16.01.97 2 10.2 7.86 188 123 20.5 2.0 5.2 14.2 79.3 11.0 21.1 27.83 Piano dell’Acqua G 0508541729 16.01.97 5 8.4 6.60 218 200 15.6 5.5 13.1 17.8 122 12.1 17.8 74.64 S. Paolo W 0513241748 16.01.97 85 15.0 6.52 895 315 110 18.8 44.5 46.7 518 47.1 79.2 61.65 Guardia W 0512741691 16.01.97 50 15.2 6.04 707 143 66.7 17.6 38.6 40.9 378 28.4 54.7 77.96 Ilice W 0507741696 16.01.97 30 9.7 5.94 225 174 27.8 7.4 11.1 15.2 156 6.7 10.6 64.57 Ponteferro G 0513641720 16.01.97 50 19.8 6.72 1630 288 202 42.6 78.6 67.3 781 93.6 168 74.08 S. Giacomo G 0507841727 16.01.97 10 11.8 6.37 480 89 24.4 7.4 36.3 50.5 396 7.8 6.7 86.89 Romito S 0491341607 17.01.97 2 15.3 6.78 1240 210 142 18.4 102 50.5 956 68.4 9.1 76.7

10 Acquarossa S 0494641583 17.01.97 10 18.3 6.50 1595 190 150 13.7 130 91.2 1210 50.3 55.2 87.611 Cherubino S 0490341628 17.01.97 2 13.9 6.84 990 120 96.8 17.2 84.3 42.9 671 37.6 43.7 69.912 Acqua Grassa S 0490741588 17.01.97 5 18.0 6.10 1725 20 163 16.4 123 98.0 1210 63.5 27.4 10413 Iuncio S 0489741590 09.06.97 2 18.7 6.05 1717 210 265 21.9 141 110 1320 233 21.1 98.614 Pedara W 0505341649 11.06.97 40 14.3 7.54 267 n.m. 45.3 8.2 13.1 11.2 152 25.9 26.9 39.815 Rocca Paterno` G 0490341575 11.06.97 12 18.0 6.93 1260 n.m. 158 70.4 76.9 88.8 793 89.7 99.4 72.516 Patellina W 0492641587 11.06.97 2 17.5 6.19 1765 40 179 20.7 170 104 1770 64.9 25.0 86.917 Poggio Monaco W 0490541845 12.06.97 20 8.9 7.60 890 n.m. 224 10.6 43.6 19.6 299 283 32.6 32.618 Acqua Difesa W 0495741625 06.10.97 50 15.4 6.02 530 259 81.6 12.8 77.2 70.3 689 33.1 42.1 81.219 Currune W 0493841622 06.10.97 10 15.5 6.29 755 220 95.7 13.7 65.0 42.1 610 37.8 21.7 81.220 Casalrosato S 0511541581 08.10.97 2 17.5 7.12 621 251 67.1 25.8 29.4 54.7 91.5 88.6 120 58.021 Ciapparazzo G 0484941802 08.10.97 500 11.3 7.56 615 167 136 18.0 70.2 15.2 384 83.3 183 42.722 Russotti W 0505141931 09.10.97 10 12.8 6.68 809 234 122 14.1 105 31.9 817 51.8 31.7 53.223 S. Leonardo W 0502841589 09.10.97 100 16.4 6.35 970 212 119 19.6 108 64.7 811 118 34.1 65.924 Piano Elisi W 0500141638 09.10.97 32 19.8 6.38 860 248 96.8 29.7 80.3 36.7 555 87.9 73.0 76.325 Guzzi W 0515041674 04.11.97 60 14.9 6.15 602 201 103 21.5 54.3 46.7 378 65.9 107 64.226 Pozzillo W 0517341683 04.11.97 100 16.6 6.45 1135 110 245 28.9 88.0 48.3 543 227 230 66.327 Terme S. Venera S 0513541597 04.11.97 2 22.5 7.23 9072342 2340 55.9 218 77.6 799 3770 194 28.528 Solicchiata W 0487941702 06.11.97 10 17.2 6.90 1542 230 250 30.1 214 28.1 1180 238 128 79.429 Fontanamurata S 0489241862 06.11.97 1 14.3 7.55 331 116 37.5 10.2 20.9 47.3 220 26.9 30.7 23.630 Di Martino W 0504041921 06.11.97 10 11.5 6.60 655 160 115 18.0 96.8 22.6 738 61.0 29.3 53.531 Zummo W 0507541902 06.11.97 7 16.6 7.40 580 104 109 16.8 42.6 14.8 342 68.1 71.0 39.132 Musa W 0489341834 06.11.97 23 11.0 7.33 488 120 56.1 17.2 48.7 31.3 287 45.4 89.8 36.133 Torrerossa G 0516541829 09.12.97 700 13.4 7.63 304 126 45.1 8.6 22.2 15.0 210 28.7 31.2 36.734 Tavolone G 0511941622 09.12.97 600 16.4 7.42 581 233 93.3 12.1 54.7 16.0 403 61.7 51.8 66.535 La Gurna S 0518941801 09.12.97 50 13.4 7.25 223 167 32.6 7.0 10.3 12.2 91.5 26.6 41.3 37.536 Turchio G 0512141608 09.12.97 900 17.4 7.04 704 75 112 16.4 63.1 26.5 430 84.7 77.3 67.037 Etna acque W 0507741604 09.12.97 250 18.1 7.11 745 28 126 17.6 66.8 32.7 403 77.6 172 54.938 Rocca Campana G 0512041837 12.12.97 100 12.7 7.38 432 172 83.2 7.8 31.7 15.0 278 39.7 82.1 43.539 Gangemi W 0503741932 12.12.97 10 10.9 6.56 450 215 64.6 7.8 41.1 36.9 396 36.9 44.6 41.540 S. Caterina 2 W 0494941896 12.12.97 20 8.9 6.89 404 160 52.6 9.8 29.5 54.5 311 31.9 87.4 23.341 Favare S 0482641708 15.01.98 5 11.8 7.65 728 107 139 23.1 95.0 24.4 622 95.0 117 47.342 Picardo W 0488841676 10.02.98 35 16.1 6.75 667 40 146 23.1 122 37.5 1020 49.6 14.4 67.643 Ficominutilla W 0497641633 25.03.98 10 22.6 6.35 1520 n.m. 263 25.4 198 189 2190 57.1 26.9 n.m.44 Sacro Cuore W 0506341620 26.03.98 90 16.4 7.35 727 n.m. 92.0 16.4 36.3 17.8 345 54.6 34.6 n.m.45 SAICOP W 0505041625 26.03.98 40 18.4 7.31 475 n.m. 93.6 17.2 30.7 15.8 271 56.0 43.2 n.m.46 SAB W 0497541621 26.03.98 10 15.9 6.00 691 n.m. 119 20.3 106 74.5 1030 39.7 25.4 n.m.47 Monacella W 0511641732 27.03.98 23 15.0 6.46 485 n.m. 130 16.4 41.9 41.1 552 36.5 41.3 n.m.48 Di Mauro W 0516741736 22.04.98 10 17.6 7.53 917 102 44.1 13.3 50.9 85.4 177 55.7 249 n.m.49 S. Leonardello W 0514941726 30.06.98 20 17.7 6.52 1155 107 88.3 21.9 69.5 85.4 400 77.3 203 81.050 Petrulli W 0510641730 30.06.98 20 21.1 6.26 1885 110 213 39.5 107 71.9 830 136 226 n.m.51 Gulli W 0489141705 01.07.98 20 16.3 6.73 1710 269 197 18.0 145 27.7 1090 115 39.4 88.052 S. Domenica W 0482841711 02.07.98 56 12.7 7.65 1232 86 137 19.6 84.0 36.7 576 95.7 137 52.053 Ercolino W 0495341598 04.08.98 40 16.1 6.18 1474 36 111 20.3 98.7 76.8 924 47.9 30.7 90.0

Conductivity is expressed inmS/cm, Eh in mV and concentrations in mg/l. Alk.5 total alkalinity expressed as HCO3. n.m.5 not measured. G:drainage gallery; S: spring; W: well. UTM: location of the sampling points with UTM grid.

1830 A. Aiuppa et al.

tendency to concentrate in the weathering minerals (oxides andclays). Instead, alkalis, alkaline earth are shifted towards thelower axis, indicating their preferential solution during weath-ering. A similar behavior is also observed for a group ofelements (As, U, Sb, Se and Mo) called “oxo-hydroxo anionforming elements” (thereafter indicated as OHA elements), dueto their tendency to form water soluble anion complexes. Tran-sition metals plot in an intermediate position. Among the OHAelements, there is some uncertainty for Se, due to the lack of

data for this element in Etna basalt and the considerable un-certainty in literature data for basaltic rocks (Wedepohl, 1978).

An important observation from Figure 3 is that elementssuch as As, Zn, Cu and Sb, which usually form highly volatilecompounds during high-temperature magma degassing (e.g.,Buat-Menard and Arnold, 1978; Andres et al., 1993; Symondset al., 1992; Gauthier and Le Cloarec, 1998), are not enrichedin the aquifer with respect to typically lithophile elements (Na,Mg, Ca, Sr). We therefore infer that there is negligible input of

Table 2. Concentrations of minor and trace metals in analysed Etnean groundwaters.

N. Li Al V Cr Mn Fe Co Ni Cu Zn As Se Rb Sr Mo Cd Sb Cs Ba Pb Th U

1 27.7 5.6 57.8 6.0 0.7 281 0.38 2.3 3.3 11.7 2.9 n.a. 51.7 589 8.6 0.12 0.04 0.25 34.4 0.04 n.d. 9.62 1.4 1.3 6.9 0.7 0.4 11 0.06 n.d. 0.4 35.7 0.3 n.a. 3.8 208 2.3 n.d. 0.02 0.02 5.8 0.02 n.d. 0.53 4.9 17.1 6.0 0.7 2.8 28 0.08 0.3 0.8 27.7 0.1 n.a. 26.9 180 2.3 0.04 0.02 0.41 5.8 0.05 0.01 0.24 9.9 3.6 26.3 3.2 0.6 95 0.19 2.0 2.2 12.2 1.2 n.a. 35.9 302 14.5 0.11 0.04 0.31 19.7 0.14 n.d. 5.15 14.0 5.6 22.2 2.9 1.0 83 0.16 0.8 5.8 8.8 1.1 n.a. 43.3 404 9.8 0.02 0.04 0.30 21.5 0.14 n.d. 2.66 4.0 5.4 19.0 1.3 0.4 25 0.05 n.d. 2.3 11.0 1.1 n.a. 23.1 120 6.2 0.03 0.07 0.13 2.2 0.03 n.d. 0.27 41.0 4.2 63.4 4.6 0.8 190 0.24 1.5 2.2 14.5 3.6 n.a. 67.3 704 53.8 0.17 0.08 0.37 10.4 0.13 0.03 5.88 6.3 4.5 2.3 1.4 724 2560 4.41 2.2 0.7 3.0 n.d. n.a. 11.2 501 1.2 n.d. n.d. 0.08 6.4 n.d. n.d. 0.19 13.4 n.d. 4.5 2.7 390 3690 0.87 2.1 2.0 2.3 0.2 n.a. 24.7 434 4.0 n.d. n.d. 0.17 6.0 n.d. 0.01 0.8

10 59.1 4.5 9.4 8.3 0.8 421 0.47 16.0 8.5 8.2 1.8 n.a. 35.0 626 17.7 0.03 0.01 0.26 15.3 0.06 0.01 3.511 7.3 3.9 41.2 4.2 0.2 181 0.17 3.3 1.1 12.5 1.9 n.a. 40.9 325 11.3 n.d. 0.03 0.37 9.9 n.d. 0.02 4.812 52.5 2.7 1.6 5.8 448 5980 2.22 5.3 1.1 6.9 0.1 n.a. 25.4 802 6.3 n.d. 0.03 0.21 20.3 0.02 0.02 0.113 159 13.0 0.7 1.8 686 2050 1.99 7.9 2.0 1.9 1.8 n.a. 40.9 973 9.1 0.87 0.03 1.59 60.9 0.03 n.d. 0.514 3.5 8.6 39.7 1.2 1.2 28 0.64 2.0 2.0 6.0 4.8 n.a. 13.6 n.a. n.a. 0.95 n.d. 1.20 1.4 0.62 n.d. 1.915 45.6 1.0 26.8 2.2 74.2 2050 0.52 7.0 1.5 1.8 2.8 n.a. 50.6 n.a. 35.4 0.21 0.09 0.27 14.6 n.d. n.d. 1.916 119 1.1 1.3 2.0 1400 5630 1.23 8.5 1.2 1.5 1.3 n.a. 27.1 838 6.0 0.09 0.03 0.03 102 n.d. n.d. 0.217 0.6 6.0 30.2 1.4 75.3 93 0.94 1.8 3.8 1.4 2.4 n.a. 15.9 n.a. n.a. 0.51 n.d. 1.30 59.7 n.d. n.d. 1.118 20.2 5.3 56.7 2.4 72.4 165 2.75 4.8 6.3 13.7 2.1 3.5 46.6 563 17.2 0.63 0.09 0.57 13.5 0.03 n.d. 3.819 46.6 2.0 89.5 1.6 0.7 133 0.29 1.2 2.5 12.3 2.0 4.7 54.3 331 25.3 0.35 0.17 0.58 10.5 0.14 n.d. 3.420 7.7 n.d. 13.2 0.7 0.3 48 0.24 0.8 1.0 26.1 1.3 4.0 29.2 498 4.6 0.78 0.12 0.30 8.9 0.43 n.d. 1.821 31.0 0.2 201 1.3 3.6 105 0.19 0.4 2.3 4.2 10.5 5.2 38.3 215 35.9 0.70 0.37 0.43 10.1 0.56 n.d. 4.122 38.2 n.d. 93.6 2.9 0.3 199 0.21 0.7 2.1 50.8 2.2 3.2 54.3 250 39.0 0.41 0.20 0.65 13.3 n.d. n.d. 10.023 38.8 n.d. 60.4 4.9 0.8 284 0.34 1.1 10.8 30.0 2.6 6.2 74.9 489 21.4 0.45 0.15 0.61 15.0 0.26 n.d. 11.524 45.7 3.8 78.1 0.8 1900 179 0.67 3.4 8.6 14.7 8.3 14.1 79.8 314 n.a. 0.41 0.15 0.99 n.a. 0.32 n.d. 2.425 17.7 n.d. 22.0 1.2 0.6 116 0.21 6.4 1.5 20.5 1.4 2.5 41.4 378 12.3 0.29 0.13 0.23 13.4 0.33 n.d. 2.626 9.6 0.2 28.0 0.8 0.2 109 0.10 0.9 3.1 4.3 1.9 3.2 38.6 378 21.3 0.55 0.09 0.09 3.7 0.04 n.d. 4.727 1120 3.2 14.0 3.9 16.8 309 0.08 1.8 n.a. n.a. 21.5 66.8 88.3 2510 n.a. 0.61 0.22 0.25 14.9 0.15 n.d. 8.028 180 n.d. 28.3 6.6 27.7 150 0.24 1.2 1.8 60.4 1.9 5.4 78.4 673 15.3 0.26 0.13 0.67 29.6 n.d. n.d. 2.729 12.5 10.0 4.0 1.0 5.3 134 0.28 1.4 2.3 4.8 1.0 1.2 4.8 360 2.6 0.24 0.26 0.14 90.1 n.d. n.d. 1.230 36.1 n.d. 63.5 5.0 0.3 140 0.20 0.9 2.3 8.2 1.9 4.2 55.6 179 15.1 0.32 0.18 0.74 9.7 n.d. n.d. 7.231 8.3 0.3 62.4 0.9 0.7 73 0.16 0.2 1.8 59.4 2.3 2.5 43.6 118 44.4 0.29 0.21 0.80 5.4 0.08 n.d. 5.132 5.0 0.7 56.3 0.5 0.3 104 0.07 0.7 2.5 52.4 3.4 0.6 22.0 207 10.9 0.54 0.19 0.16 6.1 0.03 n.d. 4.933 9.9 0.7 44.6 0.8 0.8 68 0.16 0.3 0.5 4.4 2.2 1.7 21.7 60 16.7 0.30 0.19 0.25 3.9 0.02 n.d. 2.334 61.5 n.d. 8.9 1.0 96.0 70 0.63 0.9 0.2 1.2 2.1 2.7 31.2 112 12.2 0.28 0.12 0.37 4.1 0.02 n.d. 0.935 2.1 0.2 40.2 0.6 0.2 52 0.17 0.1 0.4 0.7 1.9 2.1 16.6 64 7.1 0.27 0.16 0.17 3.2 n.d. n.d. 1.036 47.6 0.2 45.5 2.1 8.8 124 0.32 1.0 1.5 1.3 3.3 2.7 42.4 171 28.2 0.30 0.17 0.41 3.9 0.05 n.d. 3.037 27.3 1.8 23.6 0.6 96.2 142 0.13 0.7 0.5 5.2 3.6 3.0 40.6 206 17.4 0.53 0.09 0.36 4.2 0.05 n.d. 4.238 9.7 0.5 27.4 0.8 0.4 67 0.16 0.3 0.2 3.1 1.7 2.9 33.0 89 21.4 0.48 0.16 0.40 2.4 0.09 n.d. 2.739 13.3 n.d. 43.8 1.3 0.4 141 0.21 0.5 3.8 87.7 1.3 1.8 26.1 161 13.6 0.75 0.20 0.34 6.4 0.08 n.d. 3.440 9.7 n.d. 22.1 1.1 2.4 330 0.27 0.7 4.3 9.4 0.9 0.9 17.8 218 6.8 0.31 0.16 0.26 4.0 0.04 n.d. 3.741 22.7 0.5 138 1.5 0.1 146 0.07 0.9 3.9 1.3 7.1 3.7 42.4 317 38.9 0.61 0.22 0.25 14.9 0.15 n.d. 8.042 37.4 1.0 5.6 6.6 2.3 427 0.13 2.3 2.9 13.2 1.4 1.7 49.4 391 n.a. 0.23 0.08 0.32 7.7 0.51 n.d. 5.543 134 n.d. 0.4 0.6 n.d. 11500 n.d. n.d. 2.3 45.7 1.8 2.5 49.4 n.a. 0.8 0.56 n.d. 0.51 54.8 0.27 n.d. 0.144 12.3 7.4 18.8 0.5 4.2 113 0.02 1.0 3.6 6.8 4.3 3.2 28.5 n.a. 36.6 10.41 n.d. 0.30 2.8 0.68 n.d. 3.445 20.7 n.d. 35.4 0.3 0.2 55 0.02 0.1 1.6 6.5 4.3 2.9 26.2 n.a. 13.2 0.67 n.d. 0.39 5.1 0.22 n.d. 4.246 19.7 1.5 12.9 0.6 16.7 390 0.29 10.8 2.5 246 1.7 2.9 38.1 n.a. 11.5 0.94 n.d. 0.54 12.0 0.46 n.d. 4.047 5.5 n.d. 15.3 0.4 0.6 157 0.02 0.3 2.6 3.7 1.0 2.4 22.5 n.a. 19.3 0.81 n.d. 0.21 13.1 0.43 n.d. 4.248 2.1 n.d. 7.3 0.4 6.9 166 0.01 2.2 1.1 15.6 2.5 5.2 22.6 n.a. 10.1 0.34 n.d. 0.02 4.3 0.37 n.d. 2.349 1.1 n.a. 15.3 0.2 0.6 215 0.05 2.5 3.4 189 0.9 3.5 24.3 723 14.6 0.22 n.d. n.d. 22.1 0.73 0.02 3.450 56.0 n.a. 64.5 0.3 0.6 218 0.08 2.7 7.4 394 2.3 8.8 61.0 888 69.0 0.45 n.d. n.d. 10.8 0.57 0.04 5.751 65.3 n.a. 2.1 0.4 76.8 590 n.d. 1.2 4.2 6.8 0.2 5.8 40.9 461 0.8 0.07 n.d. 0.34 5.5 0.02 0.03 0.052 21.1 n.a. 120 0.8 0.2 105 n.d. 1.1 5.4 12.0 6.3 4.5 33.6 409 33.5 0.27 0.02 0.14 12.6 0.14 0.03 7.153 42.2 23.8 22.2 0.3 99.0 550 n.d. n.d. 4.0 12.1 1.2 1.7 35.0 778 13.2 0.26 n.d. 0.19 26.0 0.02 0.02 4.3

Numbers as in Table 1. Concentrations inmg/l. n.d.5 below detection; n.a.5 not analyzed.

1831Mobility of metals during basalt weathering at Mt. Etna

trace metals from magmatic gas into the aquifer, and that basaltleaching is the dominant source of dissolved metals. The lackof a magmatic input of volatile metals is consistent with aprevious cooling of the magmatic gas during ascent, as indi-cated by the low temperature, low chlorine and fluorine contentand meteoric18O composition of the groundwaters. This hy-pothesis is also supported by the fact that As and Sb have aquite uniform content throughout the aquifer and display lowconcentrations also in the areas the most affected by magmaticCO2 release. We emphasize that these observations apply to aperiod (1997–early 1998) representative of typical mild volca-nic activity at Etna, consisting of both passive degassing andstrombolian eruption from the summit craters. However, we donot exclude the possibility that during major changes in themagma plumbing system, hotter, possibly metal-bearing mag-

matic vapors could reach the aquifer and affect its metal content.The relevant temperature and conductivity anomalies recorded inEtnean groundwaters before the onset of the huge 1991–1993 lavaeffusion (Bonfanti et al., 1996a,b) indicate that hotter gas andsteam may temporarily dissolve in the aquifer.

Fig. 2. Bicarbonate-metal scatter diagrams: a) HCO3-Mg; b) HCO3-Fe; c) HCO3-Rb. The fairly good correlations observed in the plotsindicate that water acidity, resulting from CO2 dissolution in theaquifer, is neutralized through rock leaching and consequent metalrelease to the solution. Water samples from Paterno` area display higherFe contents and marked Rb depletion, respectively due to particularreduced conditions and clay minerals precipitation (see text).

Table 3. Average metal concentrations in groundwater (Cw, mg/l)and Etnean lavas (Cs, mg/kg).

Groundwaters Cw mg/l Lavas Cs mg/kg

Na 105,000 27,080(15,600–2,340,000) (26,400–36,570)

Mg 57,800 30,720(5200–218,000) (23,580–39,140)

Ca 28,000 74,200(11,200–189,000) (57,360–76,780)

K 15,700 15,440(2000–42,600) (13,030–17,920)

Sr 200 1272(59.7–2507) (1031–1338)

Fe 168 77,200(11.2–11,550) (55,200–94,900)

V 54.9 305(0.4–201) (106–328)

Mn 41.9 1000(0.1–1900) (775–1700)

Rb 36.2 40(3.8–88.3) (33.4–48.5)

Li 33.0 12(0.6–1119) (11.8–12.7)

Mo 21.2 4(0.8–53.8) (2.42–4.76)

Zn 9.3 111(0.7–88) (99.3–154)

Ba 7.2 678(1.4–102) (605–823)

As 3.6 1(0.1–21.5) (0.8–1.6)

U 3.3 3(0.07–11.5) (2.1–4.1)

Se 2.9 0.05(0.6–66.8) —

Cu 1.8 90(0.2–10.8) (20–188)

Cr 1.5 32(0.3–8.3) (2–181)

Al 1.2 96,000(0.2–23.8) (87,900–99,970)

Ni 1.0 26(0.1–16) (4–54)

Cd 0.6 0.85(0.02–1) (0.11–0.9)

Cs 0.4 0.79(0.02–1.6) (0.54–0.99)

Co 0.3 30(0.05–4.4) (20–48)

Sb 0.2 0.12(0.01–0.4) (0.08–0.18)

Pb 0.1 8(0.02–0.6) (5–12.2)

Th 0.01 8.6(0.002–0.03) (6.82–16.2)

Values in parenthesis are ranges of metal contents. Data for ground-waters are weighted averages (computed basing on flow rates of springsand wells). Data for Etna basalt are from both the literature (Cristofoliniand Romano, 1982; Joron and Treuil, 1984; Pennisi et al., 1988; Treuiland Joron, 1994; Michaud, 1995; Tanguy et al., 1997) and unpublishedINAA and ICP-MS analyses that we performed at Laboratoire PierreSue (Saclay).

1832 A. Aiuppa et al.

5. DISCUSSION

5.1. Thermodynamic Evaluation of Water-RockInteractions at Etna

The fluxes of chemical elements during weathering pro-cesses closely depend on the instability of primary mineralswith respect to the solution, and on the formation of secondaryminerals. Considering the dynamic Etnean system, subject tovariable input of magmatic gas and rapid water transit, and thelow temperature at which the processes operate, one mayexpect that thermodynamic equilibrium between water androcks is seldom achieved. Thermodynamic calculations, how-ever, are useful in indicating the direction in which chemicalreactions tend to go. In our case, equilibrium calculationsrepresent a tool to evaluate the relative tendencies of primaryminerals to dissolve and of secondary minerals to form.

The saturation index (SI) of the most common lava-formingprimary minerals (olivine, clinopyroxene and plagioclase) inEtnean natural waters is shown in Figure 4 versus the pH of thesolution. SI values were computed by assuming stoichiometricdissolution of the igneous minerals at low temperature (Table4), following Gislason and Arnorsson (1993). The equilibriumconstants of dissolution reactions of pure phases were obtainedusing the SUPCRT92 software (Johnson et al., 1992). Thestandard-state Gibbs free energy of the dissolution reactions ofsolid solutions were calculated from the equation:

DG8ss~T, P! 5 XiDG8i~T, P! 1 XjDG8j~T, P!

1 nRT~Xi ln Xi 1 Xj ln Xj! (2)

whereDG8ss, DG8i and DG8j are the standard-state Gibbs freeenergy of the solid solution and of pure end-membersi and j ;Xi andXj are the relative mole fractions of pure end-membersi and j ; T and P are temperature (in degrees Kelvin) andpressure of interest (1 atm);R is the gas constant; andn is thenumber of exchange sites per unit cell.

Equilibrium constants (Kss) are then calculated from theequation:

Log Kss, ~T, P! 5 2DG8ss~T, P!/ 2.303RT (3)

The data of Tanguy et al. (1997) were used for solid solutionsof olivine phenocrysts (Fo 0.8; Fa 0.2), groundmass olivine (Fo0.5; Fa 0.5), clinopyroxene (En 0.42; Fs 0.11; Wo 0.47) andplagioclase (An 0.7; Ab 0.3).

The PHREEQC program (Parkhurst, 1995) was used tocalculate the activities of aqueous species in the selected wa-ters. These data and the equilibrium constantKssof dissolutionreactions, calculated as described above, yield SI values fromthe equation:

SI 5 Log(IAP/Kss), ~T, P! (4)

where IAP is the ionic activity product of the specific reaction.As shown in Figure 4, ferro-magnesian minerals are strongly

undersaturated in all natural waters from Mt. Etna aquifer. ApH dependence of the SI values is observed for both olivine andclinopyroxene, the highest undersaturations being associatedwith the lowest pH values. This relation confirms that thesampled waters most likely define various steps of a commonweathering process through which CO2-rich acid waters aretitrated by the dissolution of primary minerals. Figure 4 alsoshows that Etnean groundwaters are highly more undersatu-rated with respect to olivine phenocrysts than groundmassolivines, as expected from the higher Fe content of the latter.Plagioclase with (An 0.7; Ab 0.3) composition is slightlyoversaturated in Etnean waters. A pH dependence is again

Fig. 3. Average metal concentrations in the Etnean aquifer (Cw, mg/l)versus average concentrations in local basalts (Cs, mg/kg). The term“OXA” stands for oxo-hydroxo anion forming elements. Data are fromTable 3.

Fig. 4. Water saturation index (SI) of primary minerals in Etna basaltagainst the pH of the solution (open squares5 olivine phenocrysts;filled squares5 groundmass olivine; triangles5 clinopyroxene; cir-cles 5 plagioclase). SI values have been calculated by assumingstoichiometric dissolution (Table 4) of primary minerals (Gislason andArnorsson, 1993). Data for solid solution in Etna alkali basalts-hawai-ites are from Tanguy et al. (1997).

1833Mobility of metals during basalt weathering at Mt. Etna

observed, although less clearly than in the case of olivine andclinopyroxene. In particular, plagioclase oversaturation is ob-served at pH, 7, whereas in basic solutions the saturationindex SI decreases down to25. However, this pH effect mayresult from the fact that the dissolution reaction of plagioclasewas written in terms of Al31 ion and thus is dependent on H1

activity (Table 4). Al(OH)42, which is the dominant Al aqueous

species in Etnean waters (as calculated from the PHREEQCprogram), is not included in the SUPCRT92 database. If writtenin terms of Al(OH)4

2, plagioclase dissolution is virtually inde-pendent of pH, as also shown by Gislason and Arnorsson(1993) for natural waters in Iceland.

As a result of the above calculations, a sequence of “disso-lution susceptibility” of primary minerals (olivine. clinopy-roxene. plagioclase) is inferred. No data are reported foropaque mineral solid solutions (magnetite-ulvospinel, ilmenite-hematite), but strong oversaturation for magnetite and hematiteis indicated by PHREEQC calculations. Many authors (e.g.,Gislason and Arnorsson, 1993, and references quoted therein)have drawn similar conclusions. The observed solubility trendimplies that during the dissolution of primary minerals of Etnabasalt, elements such as Mg, Fe, Mn, Ni, Co and V are releasedto the solution preferentially with respect to feldspar-containedcomponents (Na, K). Calcium, a major constituent of bothplagioclase and clinopyroxene, shows intermediate behavior.

The contribution of dissolving groundmass to metal releaseis difficult to quantify but should also be considered. In theirexperimental study of basalt dissolution at 25–60°C, Gislasonand Eugster (1987) showed that basaltic glass dissolves 10–15times faster than microcrystalline basalt. As a consequence, Naand K, which are commonly enriched in basaltic glass, arelikely to be preferentially released to the solution.

Since weathering of basalt is not a congruent process, metalsreleased to the solution by the dissolution of igneous mineralsand basaltic glass may be removed by the precipitation ofsecondary minerals. Although alteration assemblages havenever been studied at Mt. Etna, data from Iceland (Kristmanns-dottir, 1978; Kristmannsdottir, 1982; Douglas, 1987) indicatethat Ca-rich zeolite, Ca-Mg-Fe smectite, kaolinite, illite, cal-cite, chalcedony and poorly crystallized iron-manganese oxideand hydroxide are minerals characteristic of the low-tempera-ture weathering of basaltic rocks. This is partially confirmed byPHREEQC calculations, showing that Etnean waters are com-monly oversaturated with respect to Ca-montmorillonite, ka-olinite, illite, K-mica, chalcedony, goethite and hematite,whereas calcite and scolecite are undersaturated. The activitydiagrams of Figure 5a,b indicate that kaolinite and (Ca-Mg-Fe)smectite are probably the stable phases in the alteration assem-

blage. Since a pH. 8 is rarely achieved at Etna, the Ca-zeolitestability field is never reached. However, given the complexityof clay minerals and the improbable attainment of chemicalequilibrium at low temperature, these speculations do not implythat other “thermodynamic unstable” phases (such as fine-grained, poorly crystalline, K-rich clays) cannot precipitatefrom the solution.

5.2. Relative Mobility of Metals in Etnean Groundwaters

The average relative mobility (RM) of each element wascomputed from Eqn. (1) and its average concentration in bothEtna groundwater and basalt (Table 3). The results are plottedin Figures 6a,b, in which elements are ranked with averageRMs increasing from left to right. Na was chosen as thereference element because of its strong chemical mobility dur-ing weathering. All analyzed elements (except Se) have RM,1, indicating either that they are fixed more efficiently than Naby the solid products of weathering, or that they are retained inthe lattice of primary minerals. Due to low Na content inrainwater (see par. 5.3), relative mobilities do not need to becorrected for the seawater component.

Deviations from the average sequence of mobility are ob-served for waters that are either strongly oxidized or stronglyreduced, conditions that affect the mobility of redox-sensitiveelements (Figs. 6a,b). A strong departure from the averagesequence is observed for sample 27 (Terme S. Venera) (Fig.6a), whose high Na content (which has to be ascribed to theinput of saline brines from the sedimentary basement ratherthan to basalt leaching) affects the computation of RM values.

The overall geochemical behavior of the analyzed elements,and their dependence on redox conditions, are discussed belowfor each category.

5.2.1. Oxo-hydroxo anion forming elements (OHA elements)

This group includes Se, Mo, As, Sb and U. The high RMvalues (from 0.5 to 1.5) displayed by these elements indicatethat they have considerable mobility in the aqueous system.This is ascribed to their strong tendency, at least in their higheroxidation states (V, VI), to form soluble oxo-hydroxo com-plexes. Like S, C and P, their high ionic potential and polariz-ing power mean that they use O22 as their exclusive ligand,leading to the formation of soluble anions.

The geochemical behavior of Se is similar to that of S, owingto the analogous electronic configuration of its outer shell(Wedepohl, 1978). As confirmed by the data of Table 5, inwhich aqueous speciation calculations are summarized, the

Table 4. Dissolution reactions of primary minerals in basalt (after Gislason and Arnorsson, 1993).

Olivine phenocrysts (Fo0.8; Fa0.2)(Mg0.8Fe0.2)2SiO4 1 4H1 N 1.6Mg21 1 0.4Fe21 1 SiO2(aq) 1 2H2OOlivine in groundmass (Fo0.5; Fa0.5)(Mg0.5Fe0.5)2SiO4 1 4H1 N Mg21 1 Fe21 1 SiO2(aq) 1 2H2OClinopyroxene (En 0.42; Fs 0.11; Wo 0.47)(Ca0.47Mg0.42Fe0.11)SiO3 1 2H1 N 0.47Ca21 1 0.42Mg21 1 0.11Fe21 1 SiO2(aq) 1 H2OPlagioclase (An 0.7; Ab 0.3)(Ca, Al)0.7(Na, Si)0.3AlSi2O8 1 7.2H1 N 0.7Ca21 1 1.7Al31 1 0.3Na1 1 2.3SiO2(aq) 1 3.6H2O

Data for solid solution composition are from Tanguy et al. (1997).

1834 A. Aiuppa et al.

soluble aqueous species of Se (selenite, HSeO32, SeO3

22) dom-inate in an oxidized environment equivalent to the shallowEtnean groundwaters. Similarly, Mo is commonly present iniron-rich minerals (oxides, olivine, pyroxenes) under reducedIII or IV valence states (Wedepohl, 1978), and is easily leachedas molybdate (MoO4

22) during the weathering of basaltic rocks.According to Brookins (1986), the transport of Sb in hydro-

thermal waters is possible in mildly oxidized conditions asSb(III) oxyions (SbO1, SbO2

2), whereas at more oxidizedconditions its mobility is limited by solid-phase solubility(Sb2O4). By contrary, PHREEQC calculations (Table 5) indi-cate that at Mt. Etna Sb occurs mostly in its (V) valence stateand is extremely mobile as soluble oxo-hydroxo anions (SbO3

2,Sb(OH)6

2). Pais and Jones (1997) also reported that Sb(OH)62 is

the common soluble Sb species in fresh waters. Solid-phaseprecipitation of Sb is probably inhibited by its low concentra-tion in the groundwaters.

The Eh-pH diagram for arsenic is shown in Figure 7a.Arsenic forms soluble compounds in both oxidized (As(V)arsenate) and reduced (As(III) arsenite) solutions. Assumingthe existence of redox equilibrium in the sampled waters(whose fact is seldom achieved in natural fluids), we cancalculate that the former prevail in most Etnean groundwaters,whereas the latter are stable only in the most reduced samples(Fig. 7a), collected from the southwestern sector of the volcanicedifice, near the village of Paterno` (Fig. 1). The strong Asdepletion found in groundwaters from the latter area is proba-bly due to their high Fe content. These iron-rich waters deposita brown precipitate upon discharge. This was studied by XRD

and chemical (INAA and ICP-MS) analysis (Mazze` andAiuppa, unpublished data) and was found to be composed of anamorphous Fe-oxy-hydroxide (Fe5 42.5%; Mn5 223 ppm;As 5 54 ppm). Despite the fact that Fe oxy-hydroxides are notstable phases in reduced waters (Fig. 7c), their formation isprobably related to departure from the original redox conditionsof the solution as water reaches shallow oxidized ground levels.Among As species, the oxidized form, arsenate, is adsorbedonto amorphous Fe-oxy-hydroxides to a greater extent thanarsenite (Pierce and Moore, 1982). We speculate that the oxi-dation of As(III) to As(V), as highly reduced Fe-rich watersflow to the surface, enhances As adsorption on to precipitatingFe-oxy-hydroxides.

Other OHA elements such as Mo, Sb and U are stronglydepleted in the reduced waters in the southwestern sector (Fig.6a) and likely share the same fate as arsenic. Uranium scav-enging by iron oxyhydroxides, for example, is well-docu-mented (Langmuir, 1978; Hsi and Langmuir, 1985). The Eh-pHdiagram for U (Fig. 7b), shows that the reduced waters areclose to the boundary between reduced (IV, V) and oxidized(VI) species. As reduced waters approach the surface, they maygain oxygen from the vadose zone. In the pH range of 5–8 theuranyl ion (UO2

21), the most common U(VI) species, isstrongly adsorbed by iron minerals such as hematite, goethiteand amorphous ferric oxyhydroxide. The rate of oxidation fromU(IV, V) to U(VI) species may be favored kinetically byadsorption on to Fe-oxyhydroxide surfaces. As described abovefor As, this results in scavenging of uranium from the ground-waters. Figure 7b suggests that the strong U mobility in these

Fig. 5a and b. Activity diagrams showing the stability field of the most common secondary minerals. Smectites and/orkaolinite are the main phases in thermodynamic equilibrium with the groundwaters. Both diagrams are drawn at the averagesilica content of Etnean groundwaters (aSiO2 5 1023 mol/l).

1835Mobility of metals during basalt weathering at Mt. Etna

shallow oxidized waters is due to the formation of solubleuranyl-carbonato complexes such as UO2(CO3)2

22 andUO2(CO3)3

42 (Table 5).Among the elements here defined as “OHA” As, Se and Mo

have a strong chalcophile affinity. Therefore, their geochemicalmobility under strongly reduced conditions is probably de-creased by the formation of insoluble sulfides (Vink, 1996).However, given the low sulfur content of Etnean waters, thisprocess may be disregarded here. Figure 7a shows that only themost reduced sample (S. Venera thermal spring) falls into thefield of solid As2S3. Despite sulfide minerals were not amongthe solid phases revealed by mineralogical analysis on theprecipitate formed by groundwaters upon discharge, we cannot

rule out that they exercise an effective sink for chalcophileelements during the weathering process at depth (Jean andBancroft, 1986; Renders and Seward, 1989).

5.2.2. Alkalis and alkaline earths

Alkali and alkaline earth elements are classified as A-typemetals (Stumm and Morgan, 1996), forming cations with inertgas-type electronic configuration in solution. Given their lowpolarizability, these ions tend to remain in solution as free ions,since they are more strongly attracted by water than by otherinorganic ligands.

Many authors have pointed out the extreme variability of theorder of mobility of alkalis and alkaline earths during low-temperature weathering of basalt. Gislason et al. (1996) showedthat the order of mobility observed in Icelandic rivers differswhen these flow through either young or old basalts, mainlybecause Ca and Sr become more mobile as the age of the rockincreases. This may be explained by enhanced plagioclasedissolution in case of old rocks. Gislason and Eugster (1987), intheir experimental study of basalt dissolution, showed that therelease of alkalis and alkaline-earths also varies between glassy(Na . K . Ca, Mg) and crystalline basalts (Na.. K . Ca.Mg). Louvat (1997) demonstrated that great differences inmetal mobility occur in rivers draining different volcanic is-lands.

Figure 6 shows the following average mobility sequence foralkali and alkaline earth elements in Etnean groundwaters: Na. Mg . K $ Rb $ Ca$ Cs$ Li . Sr.. Ba. This sequenceis constant in all the samples, thus confirming a broad litho-logical and mineralogical homogeneity of Etnean rocks. Rb andLi follow the behavior of K (rK-Rb

2 5 0.6) and Na (rNa-Li2 5

0.5), respectively. Amorphous silica (Li) and fine-grained,poorly crystalline K-Rb-rich clays probably limit the geo-chemical mobility of both elements in aqueous media (Goguel,1983). Alkaline earth elements appear to remain in solution, notremoved by secondary minerals, with calcite and zeolite beingcommonly undersaturated. A possible sink for alkaline earthelements is represented by smectite (Fig. 5).

As also suggested by Louvat (1997), barium is by far theleast mobile element in the group. Since Ba primary minerals(witherite, BaCO3; barite, BaSO4) are always undersaturated inEtnean waters, we hypothesize that Ba is retained in the weath-ered rocks by the low solubility of plagioclase in the solution(Fig. 4). The greater ionic radius of Ba compared with Caexplains its greater affinity for and thus higher retention in thesolid phase. Strontium, which has an ionic radius intermediatebetween that of Ca and Ba, actually has an intermediate mo-bility (Ca . Sr .. Ba). Finally, the well-documented Baadsorption onto solid Fe-Mn oxo-hydroxides may explain itslimited mobility during weathering (Murray, 1975; McKenzie,1989).

Our computation indicates a very high mobility of magne-sium in Etnean waters. This may be explained by the strongundersaturation of olivine and pyroxene in these waters and thelow pH and temperature during the weathering process. Thisresults in a lack of secondary minerals able to remove Mg(except for smectite).

Fig. 6. Relative mobility of metals in Etnean groundwaters. RMvalues were normalized to sodium and computed from equation (1) (seetext). Elements are arranged in four groups characterized by increasingaverage mobility (1, OXA elements; 2, alkalis and alkaline earths; 3,transition metals; 4, immobile metals). Relative metal mobilities inselected samples are compared to the average trend (}). Fig. 6a refersto samples with low Eh values: 9 (p); 12 (■); 13 (Œ); 16 (F); 51 (h).Negative spikes for Pb, V, Zn U, Sb and As are attributed to eitheradsorption onto precipitating Fe-hydroxides or/and the formation ofinsoluble species. Sample 27 (dashed line), which is characterized bythe lowest redox potential (Eh5 2342 mV), departs from the averagetrend due to its very high Na content (ascribed to the input of thermalsaline brines rising from the sedimentary basement). Fig. 6b refers tostrongly oxidized samples: 1 (‚); 2 (g); 4 (E); 5 (1) and 6 (p).Negative shifts for Mn, Co and Fe derive from lower transition metalmobility in oxidized environments.

1836 A. Aiuppa et al.

5.2.3. Transition metals

Transition metals (Zn, Mn, V, Ni, Cr, Co, Cu) occupy anintermediate position (0.01, RM , 0.1) between alkalis-alkaline earths and immobile elements (RM, 0.01; Fig. 6).Only Fe behaves similarly to immobile elements (Al and Th)which are retained in the solid residue of weathering.

The behavior of transition metals during weathering dependsfirst on the dissolution of primary ferro-magnesian minerals. Aspreviously shown (Fig. 4), olivine and pyroxene are stronglyundersaturated in Etnean waters and release trace metals to thesolution, whereas oxide minerals (magnetite-ulvo¨spinel andilmenite-hematite solid solutions) are more resistant and mayretain Fe, Cr, V and Co in the residual rock, thus limiting theirmobility. Otherwise, the formation of hydroxo complexes dur-ing hydrolysis is the most common process involving transitionmetals in the aqueous environment (Baes and Mesmer, 1976;Stumm and Morgan, 1996). Hydrolysis of cations is enhancedby a decrease in their radius and an increase in their charge.Because a correlation exists between the first hydrolysis prod-uct of metals and the solubility of their oxide and hydroxide,the mobility of transition metals is limited by the formation ofinsoluble compounds.

Additionally, transition metals can have different oxidationstates, meaning that redox transformations may play an impor-tant role in determining their geochemical mobility in solution.This is particularly the case when either a reduced or oxidizedform is insoluble whereas the other is soluble. At Mt. Etna,there is an important dependence on Eh values for most tran-sition metals, particularly Fe and Mn. The average Fe and Mnconcentrations in the southwestern sector of Etna (Paterno` and

Adrano), where extremely reduced conditions prevail (Eh,20mV), are 5 and 1 mg/l respectively. Elsewhere, groundwaterscharacterized by higher redox potential have consistently muchlower Fe and Mn concentrations (100–300 and 1–10mg/l,respectively). The Eh-pH diagram for Fe (Fig. 7c) illustratesthat the Paterno` samples are shifted towards the field of solubleFe(II) species, whereas most Etnean groundwaters fall in thestability field of the iron solid phase (goethite). Given the highFe content of reduced waters, we hypothesize that goethitesaturation is achieved only at very shallow depths, wheregroundwaters react with atmospheric oxygen.

Figure 6b shows the calculated relative mobilities for someof the more oxidized waters from the Zafferana area (eastflank), compared with the average trend. Negative spikes forMn, Co and Fe can clearly be seen, reflecting the well-docu-mented affinity and adsorption of Co on to amorphous MnO2(s)(McKenzie, 1989, and references therein).

Chromium and vanadium in Etnean groundwaters have verysimilar behavior (RMV ' RMCr). PHREEQC calculations (Ta-ble 5) indicate that both elements are mobile in oxidizedsolutions, where they form soluble complexes (chromates,HCrO4

2/CrO422; vanadates, H2VO4

2/HVO422). Under reduced

conditions, Cr(OH)3 and the hydrolysis product of the vanadylion (VOH3) are formed and both species are insoluble and/orstrongly adsorbed on iron oxo-hydroxides. Cr and V are com-monly present as V(III) and Cr(III) in Fe-rich igneous primaryminerals. Therefore, significant leaching is possible only if amildly oxidized environment exists. Although both Cr(III) toCr(VI) and V(III) to V(V) redox transformations are thermo-dynamically possible (atp« 5 2log ae2 . 10 at pH5 6;

Table 5. Results of aqueous speciation calculations carried out with PHREEQC software (Parkhurst, 1995).

Average oxidized groundwaters Average reduced waters

Al Al(OH) 42 Al(OH)2

1 (60%), Al(OH)42 (40%)

As HAsO422 (50%), H2AsO4

2 (50%) H3AsO3

Ba Ba12 (80%), BaSO4 (15%), BaHCO31 (5%) Ba12 (90%), BaHCO3

1 (10%)Ca Ca12 (90%), CaHCO3

1 (10%) Ca12 (89%), CaHCO31 (10%)

Cd Cd12 (70%), CdHCO31 (20%), CdCl (10%) Cd12 (61%), CdHCO3

1 (20%), CdCl (20%)Cr Cr(OH)3 (54%), Cr(OH)21 (46%) Cr(OH)21 (50%), Cr(OH)2

1 (40%), Cr(OH)3 (10%)Cs Cs1 Cs1

Cu Cu21 Cu21

Fe Fe12 (50%), FeHCO31 (50%) Fe12 (52%), FeHCO3

1 (47%)K K1 K1

Li Li 1 Li1

Mg Mg12 (85%), MgHCO31 (13%) Mg12 (88%), MgHCO3

1 (12%)Mn Mn12 (52%), MnHCO3

1 (48%) Mn12 (51%), MnHCO31 (49%)

Na Na1 Na1

Ni NiCO3 (90%), Ni21 (8%) NiCO3 (45%), NiHCO3 (42%), Ni21 (13%)Pb PbCO3 (62%), PbHCO3

1 (35%) PbCO3 (70%), PbHCO31 (30%)

Rb Rb1 Rb1

Sb SbO32 (95%), Sb(OH)6

2 (5%) Sb(OH)3Se HSeO3

2 (85%), SeO322 (14%) HSe2

Sr Sr12 (85%), SrHCO31 (10%) Sr12 (88%), SrHCO3

1 (12%)U UO2(CO3)2

22 U(OH)52

V H2VO42 V(OH)3 (90%), H2VO4

2 (5%)Zn Zn12 (44%), ZnHCO3

1 (39%), ZnCO3 (11%) Zn12 (50%), ZnHCO31 (50%)

The main stable aqueous species are listed. Values in parenthesis are the proportion (in %) of each species with respect to the total amount. Onlyone species is reported when it accounts for more than 95% of the total, whereas species less than 5% are neglected. Column 1 reports the averageresults for oxidized waters, while column 2 lists average data for reduced waters (samples 9, 12, 13, 16, 27, 28, 34, 51). The table shows that As,V, Sb and U aqueous species in reduced samples differ from those in average oxidized Etnean groundwaters. The carbonate-ligand is important forNi, Pb, Zn, Cd and Mn solubility, while metal-chloride complexes are negligible.

1837Mobility of metals during basalt weathering at Mt. Etna

Middelburg et al., 1988), they are characterized by slow kinet-ics (Van der Weijden and Reith, 1982; Wehrli and Stumm,1989; Van der Weijden and Van der Weijden, 1995). This fact,together with the resistant nature of Cr-V-rich opaque minerals,explains why both elements are considerably less enriched inEtnean waters than other elements forming oxo-anions (As, Sb,Se, U, Mo).

In the case of vanadium, it has been demonstrated that therate of oxygenation of the vanadyl ion (VO21) is stronglyenhanced by adsorption on to iron hydroxides (Wehrli andStumm, 1989). Scavenging of V by precipitation of iron hy-droxides explains the low V content in the reduced Fe-richwaters from Paterno` and Adrano (Figs. 6a and 7d). A similarprocess is responsible for the removal of Cu, Zn, and Pb fromsolution (Fig. 6a). Table 5 indicates that Cu is present mainly insolution as free Cu(II) ions, whereas Zn and Pb (together withNi and Cd) are strongly complexed by carbonate or bicarbon-ate, as a result of the high total carbonate species in ground-waters.

5.3. Aqueous Metal Fluxes and Rate of Chemical Erosion

Based on the average composition (Table 3) and the overalldischarge of Etnean groundwaters (0.69 km3/a; Ogniben,1966), the aqueous discharge of major, minor and trace metalsfrom Etna can be computed (Table 6). Individual metal fluxes

ranging from 7.03 1023 t/a (Th) to 7.33 104 t/a (Na) areobtained. Figure 8 compares these elemental aqueous fluxeswith published emission rates of metals from the summit cra-ters of Etna (Buat-Me´nard and Arnold, 1978; Andres et al.,1993; Gauthier and Le Cloarec, 1998). The amounts of K, Na,Mg, Ca, Sr, Li and U transported by groundwaters are compa-rable to those released to the atmosphere from the craters.Several of these elements (e.g., Mg, Ca, Sr, U) have a lithophileaffinity and are emitted principally as ash particles in thevolcanic plume. By contrast, the emission rates of volatilemetals (Pb, Cu, Cd, Zn) in the volcanic plume are up to 10,000times higher than the aqueous transport rates, further evidencethat the dominant control on the content of these elements ingroundwaters is basalt leaching.

At the same time, basalt weathering consumes carbon diox-ide. D’Alessandro et al. (1997a) estimated that at least 570 tonsof carbon dioxide are consumed daily by chemical weatheringat Mt. Etna. This minimum value corresponds to about 7% ofthe crater plume output of CO2 measured in June, 1997 (8200t/d; Allard et al., 1998), during our sampling. The magmaticderivation of this CO2 is well-constrained by carbon isotopedata (Allard et al., 1991; Allard et al., 1997; D’Alessandro etal., 1997a,b).

The large aqueous discharge of some metals at Etna resultsfrom the high meteoric water inflow and a high chemical

Fig. 7. Eh-pH diagrams for selected elements (data from Brookins, 1988): (a) Arsenic; (b) Uranium; (c) Iron; and (d)Vanadium. Superimposed on the diagrams are the Eh and pH conditions in the groundwaters under study. In constructionsthe diagrams overall equilibrium is assumed, including redox equilibia.

1838 A. Aiuppa et al.

erosion rate, the latter enhanced by the acidity of gas-chargedwaters. The chemical erosion rate can be estimated using theaverage total content of dissolved solids produced during basaltweathering (TDSbl). In order to calculate this figure, it isnecessary to subtract from the average TDS the amount ofsolutes deriving from sources other than basalt leaching. Alka-linity, that derives entirely from the dissolution of magmaticCO2, is not included in TDSbl. Chloride and SO4 may also havevarious origins, such as rainwater, magmatic gas inputs andorganic matter decay. Rainwater, in particular, is likely to be an

important source of dissolved species, and its contributionneeds to be quantified. Aiuppa et al. (submitted) have studiedthe chemical composition of rainwater samples collected on anetwork of 10 rain gages located at different altitudes along theflanks of Mt. Etna. The authors have shown that rainwaterinflows of chloride and sulfate range respectively from 0.9 to890 mg/m2 day and 0.38 to 197 mg/m2 day, depending onaltitude and distance from the sea and summit craters. Anoverall input of 8500 t/a of Cl and 4800 t/a of SO4 for the wholeEtnean area (1150 km2) has been computed using the isohyethmethod. These values correspond to about 12 and 6% of theyearly discharge of the aquifer. Therefore, we estimate Clbl andSO4,bl to be 88 and 94% of total chloride and sulphate. Theseare likely to be maximum values, as magmatic gas inputs fromdepth cannot be completely ruled out.

We consider that metals almost entirely originate from rockdissolution, since other sources such as volcanic gas input canbe discounted, as previously discussed. Metal concentrations inrainwater were measured at Nicolosi sampling station, andrevealed to be very low (Na, 3 mg/l; Mg , 0.7 mg/l).Sea-water spray is the main metal source to rainwater, assuggested by the measured Na/Cl and Mg/Cl ratios. Therefore,major cation contribution from rainwater can be calculatedfrom the Cl inflow and from the characteristic metal/chlorideratios in seawater. The computed values (Table 6) indicate thatrainwater contribution is negligible (;0.5–3%) for all majorcations.

After correcting for rainwater, the computed TDSbl results345 mg/l, which, for an aquifer discharge of 0.69 km3/a, givesa chemical erosion rate of about 2.3*105 t/a. Considering a totalsurface of 1150 km2 for the Etnean basin, we obtain a specificchemical erosion rate of about 200 t/km2/a, which can betentatively compared with chemical erosion rates on the Earth’ssurface estimated studying fluvial basins. These data are notdirectly comparable, as data presented here refer to groundwa-ter flow in a fractured aquifer instead of river run-off. Despiteuncertainties, we note that our values are considerably higherthan the average global rate for chemical erosion at the Earth’ssurface (26 t/km2/a; Berner and Berner, 1996) and fall abovethe range calculated for the world’s greatest fluvial basins(5–100 t/km2/a; Summerfield and Hulton, 1994).

6. CONCLUSIONS

The results presented have illustrated the importance ofmagmatic fluids in the aqueous cycling of metals in shallowvolcanic environments. Although a direct magmatic input ofmetals to the Etnean aquifer can be disregarded, rising mag-matic volatiles, principally CO2, do induce intense rock leach-ing. This results in metal release to the solution and enhancedmetal fluxes in the aqueous system. As such, volcanic activityproduces a “natural pollution” of the aquifer, where maximumadmissible concentrations fixed by European Union for drink-ing waters are often exceeded (Giammanco et al., 1996).

At Mt. Etna, intense basalt leaching results from the disso-lution of CO2-rich magmatic fluids into shallow groundwaterswhich flow downslope through permeable volcanic strata. Theconsequent release of major, minor and trace elements to theweathering solution depends on many factors, including:

1. the saturation state of primary minerals;

Fig. 8. Comparative emission rates of metals during groundwaterdischarge (Table 6) and summit crater volcanic activity at Etna (yearlyaverage values). The flux ranges for crater plume emissions includeboth literature data (Buat-Me`nard and Arnold, 1978; Andres et al.,1993; Gauthier and Le Cloarec, 1998) and results obtained in June 1997(mean SO2 flux of 3500 6 250 t/d; Allard et al., 1998). Lithophileelements (grey squares) and volatile metals (open squares) are distin-guished in the plot. The aqueous discharge rate of K, Na, Mg, Ca, Sr,Li, V and U is comparable to the crater emissions.

Table 6. Aqueous discharge rates of metals from Etnean aquifer.

Flux t/a Flux t/a

Na 7.3*104 (0.2*104) As 2.5Mg 4.1*104 (0.02*104) U 2.3Ca 2.0*104 (0.01*104) Se 2.0K 1.1*104 (0.01*104) Cu 1.3Sr 140 Cr 1.1Fe 118 Al 0.9V 38 Ni 0.7Mn 29 Cd 0.4Rb 25 Cs 0.3Li 23 Co 0.2Mo 15 Sb 0.11Zn 6.5 Pb 0.10Ba 5.1 Th 0.007

Elemental fluxes were computed from the weighted average ground-water composition (Table 3) and the overall groundwater dischargefrom the mountain (0.69 km3/a; Ogniben, 1966). Rainwater contribu-tion for major elements (in parenthesis) has been quantified basing onmetal/chloride ratios in rainwater samples (Aiuppa et al., submitted).

1839Mobility of metals during basalt weathering at Mt. Etna

2. the precipitation of secondary mineral phases; and3. the aqueous chemistry of each element, which may result in

the formation of either soluble or insoluble species.

The pH and Eh of water have a major effect on the direction ofthe whole process. Most elements display good correlationswith pH and alkalinity, indicating that metal partitioning intothe solution is enhanced by an acidic weathering environment.Redox processes influence the geochemical behavior of someredox-sensitive elements (Fe, Mn, V, As). The Na-normalizedrelative mobility (RM) of metals in Etna’s aquifer varies overa wide range (1025 , RM , 10), confirming that basaltweathering is not a congruent and isochemical process.

Based on the average composition and overall outflow ofEtnean groundwaters, we have estimated the aqueous dischargeof metals from Etna. The amount of K, Na, Mg, Ca, Sr, Li andU transported by groundwaters are comparable to those re-leased in the atmosphere from the summit craters, partly as ashparticles. By contrast, the emission rates of volatile metals (Pb,Cu, Cd, Zn) in the volcanic plume are up to 10,000 times higherthan those caused by aqueous transport. Basalt weathering atEtna consumes about 23 105 t/a of magma-derived CO2, about7% of the contemporaneous emission rate of magmatic CO2

from the summit craters. The chemical erosion rate responsiblefor the total dissolved solids (TDSbl) in Etnean aquifer averages2.3*105 t/a.

Acknowledgments—This research was carried out with the financialsupport of the National Group for Volcanology of Italy (G.N.V.). Theauthors wish to thank S. Bellomo and L. Brusca for help in the field andin laboratory, G. Volpicelli for major elements analysis, and J. P.Duraud and G. Meyer for the access to and assistance with ICP-MSanalyses at Lab. Pierre Su¨e (CEN-Saclay). Furthermore, we wish tothank all the owners of private wells who kindly gave us permission totake water samples and all the people in municipal administrations,public and private aqueducts for access to the wells, springs andgalleries under their responsibility. We are grateful to J. Gaillardet,S. R. Gislason and J. W. Hedenquist, whose comments significantlyimproved this paper.

REFERENCES

Aiuppa A., Allard P., D’Alessandro W., Michel A., Parello F., andTreuil M. (1998) Trace element hydrogeochemistry of Mt. Etna,Sicily: insight on water-rock interaction.Proc. of the 9th Internat.Symp. on Water Rock Interaction, Taupo, New Zealand, 30 March–3April 1998. (eds. Arehart & Hulston), pp. 195–198, Balkema, Rot-terdam.

Aiuppa A., Bonfanti P., Brusca L., D’Alessandro W., Federico C., andParello F. (submitted to Appl. Geochem.) Quantifying the environ-mental impact of volcanic emissions: Insights from the chemistry ofrainwater in the Mt. Etna area (Sicily).

Allard P., Carbonnelle J., Dajlevic D., Le Bronec J., Morel P., RobeM. C., Maurenas J. M., Faivre-Pierret R., Martin D., Sabroux J. C.,and Zettwoog P. (1991) Eruptive and diffuse emissions of CO2 fromMount Etna.Nature351,387–391.

Allard P., Jean-Baptiste P., D’Alessandro W., Parello F., Parisi B., andFlehoc C. (1997) Mantle-derived helium and carbon in groundwatersand gases of Mount Etna, Italy.Earth Planet. Sci. Letters148,501–516.

Allard P., Aiuppa A., and Loyer H. (1998) Volatile plume fluxes fromEtna and Stromboli: Plume sampling and validation. Second-termCEC Report, MVRRS contract (ENV4-CT96–0288), 61–78.

Andres R. J., Kyle P. R., and Chuan R. L. (1993) Sulphur dioxide,particle and elemental emissions from Mount Etna, Italy during July1987.Geol. Rundsch.82, 687–695.

Anza S., Dongarra` G., Giammanco S., Gottini V. Hauser S., and

Valenza M. (1989) Geochimica dei fluidi dell’Etna.Miner. Petrogr.Acta 32, 231–251.

Aureli A. and Musarra S. (1975) Idrogeologia del bacino del fiumeAlcantara (Sicilia).Proc. of the 3rd International Symposium onGroundwaters Palermo 1–5 November 1975,157–216.

Baes C. F. and Mesmer R. E. (1976)Hydrolysis of cations.Wiley-Interscience, New York.

Barberi F., Civetta L., Gasparini P., Innocenti F., Scandone R., andVillari L. (1974) Evolution of a section of the Africa-Europe plateboundary: Paleomagnetic and volcanological evidence from Sicily.Earth Planet. Sci. Lett.22, 123–132.

Berner E. K. and Berner R. A. (1996)Global environment: Water, air,and geochemical cycles.Prentice-Hall.

Bonfanti P., D’Alessandro W., Dongarra` G., Parello F., and Valenza M.(1996a) Medium term anomalies in groundwater temperature before1991–93 Mt. Etna eruption.J. Volcanol. Geotherm. Res.73, 303–308.

Bonfanti P., D’Alessandro W., Dongarra` G., Parello F., and Valenza M.(1996b) Mt. Etna eruption 1991–93: Geochemical anomalies ingroundwaters.Acta Vulcanol.8(1), 107–109.

Brookins D. G. (1986) Geochemical behavior of antimony, arsenic,cadmium and tallium: Eh-pH diagrams for 25°C, 1 bar pressure.Chem. Geol.54, 271–278.

Brookins D. G. (1988) Eh-pH diagrams for geochemistry. SpringerVerlag.

Brusca L., Aiuppa A., D’Alessandro W., Parello F., Allard P., andMichel A. (in press) Geochemical mapping of magmatic gas-water-rock interactions in the aquifer of Mount Etna volcano.J. Volcanol.Geotherm. Res.special issue “Magma degassing through volcanoes”.

Buat-Menard P. and Arnold M. (1978) The heavy metal chemistry ofatmospheric particulate matter emitted by Mount Etna volcano.Geophys. Res. Lett.5, 245–248.

Carr R. G., Rodgers K. A. and Black P. M. (1980) The chemical andmineralogical changes accompanying the laterization of basalt atKerikeri, North Auckland.J. Royal Society of New Zealand10,247–258.

Chester D. K., Duncan A. M., Guest J. E., and Kilburn C. R. J. (1985)Mount Etna: The Anatomy of a Volcano.Chapman and Hall, Lon-don, 404 pp.

Chesworth W., Dejou J., and Larroque P. (1981) The weathering ofbasalt and relative mobilities of the major elements at Belbex,France.Geochim. Cosmochim. Acta61, 1235–1243.

Chiodini G., D’Alessandro W., and Parello F. (1996) Geochemistry ofthe gases and of the waters discharged by the mud volcanoes ofPaterno`, Mt. Etna (Italy).Bull. Volcanol.58, 51–58.

Cristofolini R. and Romano R. (1982) Petrologic features of the Etneanvolcanic rocks.Mem. Soc. Geol. Ital.23, 99–115.

D’Alessandro W., Giammanco S., Parello F., and Valenza M. (1997a)CO2 output andd13C(CO2) from Mount Etna as indicators of degas-sing of shallow asthenosphere.Bull. Volcanol.58, 455–458.

D’Alessandro W., De Gregorio S., Dongarra` G., Gurrieri S., Parello F.,and Parisi B. (1997b) Chemical and isotopic characterization of thegases of Mount Etna (Italy).J. Volcanol. Geotherm. Res.78,65–76.

Douglas G. R. (1987) Manganese-rich rock coatings from Iceland.Earth Surface and Landforms12, 301–310.

Ferrara V. (1975) Idrogeologia del versante orientale dell’Etna.Proc.of the 3rd International Symposium on Groundwaters Palermo 1–5November 1975,91–134.

Ferrara V. (1990)C.N.R. Valutazione della Vulnerabilita` degli Ac-quiferi (Resp. Prof. M. Civita). Carta della vulnerabilita`all’inquinamento dell’acquifero vulcanico dell’Etna.S.E.L.C.A.Firenze.

Gaillardet J., Dupre` B., and Allegre C. J. (1995) A global geochemicalmass budget applied to the Congo Basin rivers: Erosion rates andcontinental crust composition.Geochim. Cosmochim. Acta59,3469–3485.

Garrels R. M. and Mackenzie F. T. (1971)Evolution of sedimentaryrocks.Norton, New York.

Gauthier P. J. and Le Cloarec M. F. (1998) Variability of alkali andheavy metal fluxes released by Mt. Etna volcano, Sicily, between1991 and 1995.J. Volcanol. Geotherm. Res.81, 311–326.

Giammanco S., Valenza M., Pignato S., and Giammanco G. (1996)

1840 A. Aiuppa et al.

Mg, Mn, Fe and V concentrations in the ground waters of MountEtna (Sicily, Italy).Wat. Res.30–2,378–386.

Giammanco S., Ottaviani M., Valenza M., Veschetti E., Principio E.,Giammanco G., and Pignato S. (1998) Major and trace elementsgeochemistry in the ground waters of a volcanic area: Mount Etna(Sicily, Italy). Wat. Res.32, 19–30.

Giggenbach W. F. (1988) Geothermal solute equilibria. Derivation ofNa-K-Mg-Ca geoindicators.Geochim. Cosmochim. Acta44, 2021–2032.

Gislason S. R. and Eugster H. P. (1987) Meteoric water-basalt inter-actions. II: A field study in NE Iceland.Geochim. Cosmochim. Acta51, 2841–2855.

Gislason S. R. and Arnorsson S. (1993) Dissolution of primary min-erals in natural waters: Saturation state and kinetics.Chem. Geol.105,117–135.

Gislason S. R., Arnorsson S., and Armannsson H. (1996) Chemicalweathering of basalt as deduced from the composition of precipita-tion, rivers and rocks in SW Iceland.Am. J. Sci.296,837–907.

Goguel R. L. (1983) The rare alkalis in hydrothermal alteration atWairakei and Broadlands geothermal fields.Geochim. Cosmochim.Acta 47, 429–437.

Hirn A., Nicolich R., Gallart J., Laigle M., Cernobori L., and ETNA-SEIS Scientific Group (1997) Roots of Etna volcano in faults of greatearthquakes.Earth Planet. Sci. Lett.148,171–191.

Hsi C. D. and Langmuir D. (1985) Adsorption of uranyl onto ferricoxyhydroxides: application of the surface complexation site-bindingmodel.Geochim. Cosmochim. Acta49, 1931–1941.

Kieffer G. (1970) Les de´pots detritiques et pyroclastiques du versantoriental de l’Etna.Atti. Acc. Gioenia. Sci. Nat.2, 3–32.

Kristmanndottir H. (1978) Alteration of basaltic rocks by hydrothermalactivity at 100–300°C. InInternational Clay Conference 1978(eds.,M. Mortland and V. C. Farmer), 358–367, Elsevier, Amsterdam.

Kristmanndottir H. (1982) Alteration in the IRDP drill hole comparedwith other drill holes in Iceland.J. Geophy. Res.87, 6525–6531.

Johnson J. W., Oelkers E. H., and Helgeson H. C. (1992) A softwarepackage for calculating the standards molal properties of minerals,gases, aqueous species, and reaction from 1 to 1000°C.ComputerGeosci.18, 899–947.

Jean G. E. and Bancroft M. (1986) Heavy metals adsorption by suphideminerals surfaces.Geochim. Coscochim. Acta50, 1455–1463.

Joron J. L. and Treuil M. (1984) Etude ge´ochimique et pe´trogenese deslaves de l’Etna, Sicile, Italie.Bull. Volcanol.47–4,1125–1144.

Langmuir D. (1978) Uranium solution-mineral equilibria at low tem-peratures with application to sedimentary ore deposits.Geochim.Coscochim. Acta42, 547–569.

Louvat P. (1997) Etude ge´ochimique de l’e´rosion fluviale des ıˆlesvolcaniques: l’aide des bilans d’elements majeurs et traces. The`s dedoctorat, Universite` Paris VII.

Meybeck M. (1986) Composition chimique des ruisseaux non pollue`sde France.Sci. Geol. Bull. (Strasbourg)39, 3–77.

Michaud V. (1995) Crustal xenoliths in recent hawaiites fron MounthEtna, Italy: Evidence for alkali exchanges during magma-wall rockinteraction.Chem. Geol.122,21–42.

Middleburg J. J., Van der Weijden C. H., and Woittiez J. R. W. (1988)Chemical processes affecting the mobility of major minor and traceelements during weathering of granitic rocks.Chem. Geol.68, 253–273.

Mc Kenzie R. M. (1989)Manganese oxide and hydroxides.In: Mineralin soil enviroments(eds, J. B. Dixon and S. B. Weed) Soil sciencesociety of America n.1. Madison, Wisconsin, USA.

Murray J. W. (1975) The interaction of metal ions at manganesedioxides-solution interface.Geochim. Cosmochim. Acta39, 505–519.

Nesbitt H. W. and Wilson R. E. (1992) Recent chemical weathering ofbasalt.Am. J. Sci.292,740–777.

Ogniben L. (1966) Lineamenti idrogeologici dell’Etna.Rivista Miner-aria Siciliana 100–102,151–174.

Pais I. and Jones J. B. (1997)The Handbook of trace metals.St. LuciePress. Boca Raton, Florida.

Parkhurst D. L. (1995) PHREEQC—a computer program for specia-tion, reaction path, advective transport, and inverse geochemicalcalculations. USGS Water-Resources Investigations Report95,4227.

Pierce M. L. and Moore C. B. (1982) Adsorption of arsenite andarsenate on amorphous iron hydroxide.Wat. Res.16, 1247–1253.

Pennisi M., Le Cloarec M. F., Lambert G., and Roulley J. C. (1988)Fractionation of metals in volcanic emissions.Earth Planet. Sci.Lett. 88, 284–288.

Renders P. J. and Seward T. M. (1989) The adsorption of thio gold(I)complexes by amorphous As2S3 and Sb2S3 at 25 and 90°C.Geochim.Coscochim. Acta53, 255–267.

Symonds R. B., Reed M. K., and Rose W. I. (1992) Origin, speciationand fluxes of trace-element gases at Augustine Volcano, Alaska:Insights into magma degassing and fumarolic processes.Geochim.Cosmochim. Acta56, 633–657.

Stallard R. F. (1995) Relating chemical and physical erosion. In:Chemical weathering rates of silicate minerals(eds., A. F. Whiteand S. L. Brantley),Rev. Mineral.31, Mineral Soc. Am., 543–564.

Stumm W. and Morgan J. J. (1996)Aquatic chemistry: Chemicalequilibria and rates in natural waters.Wiley-Interscience publica-tion.

Summerfield M. A. and Hulton N. J. (1994) Natural control of fluvialdenudation rates in major world drainage basins.J. Geophys. Res.99, 13871–13883.

Tanguy J. C., Condomines M., and Kieffer G. (1997) Evolution of theMount Etna magma: Constraints on the present feeding system anderuptive mechanism.J. Volcanol. Geotherm. Res.75, 221–250.

Treuil M. and Joron J. L. (1994) Etude ge´ochimique des e´lements entraces dans les laves e´mises au cour de l’e´ruption 1991–1993 del’Etna. Mise en evidence des contributions de la source, de la fusionpartielle, de la differenciation et des modalite´s de transfert desmagmas.Acta Vulcanol.4, 29–45.

Van der Weijden C. H. and Reith M. (1982) Chromium(III)-chromi-um(VI) interconversion in seawater.Mar. Chem.11, 565–572.

Van der Weijden C. H. and Van der Weijden R. D. (1995) Mobility ofmajor, minor and some redox-sensitive trace elements and rare-earthelements during weathering of four granitoids in central Portugal.Chem. Geol.125,149–167.

Wedepohl K. H. (1978)Handbook of geochemistry.Spinger, NewYork.

Wehrli B. and Stumm W. (1989) Vanadyl in natural waters: Adsorptionand hydrolysis promote oxygenation.Geochim. Cosmochim. Acta53, 69–77.

White A. F. (1983) Surface chemistry and dissolution kinetic of glassyrocks.Geochim. Cosmochim. Acta47, 805–816.

Vink B. W. (1996) Stability relations of antimony and arsenic com-pounds in the light of revised and extended Eh-pH diagrams.Chem.Geol.130,21–30.

1841Mobility of metals during basalt weathering at Mt. Etna