Chemistry of solutions from the 13°N East Pacific Rise hydrothermal site

Chemistry of solutions from the 13°N East Pacific Rise hydrothermal site

Earth and Planetary Science Letters, 67 (1984) 297-307 297 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands [61 Chemistry ...

774KB Sizes 1 Downloads 61 Views

Earth and Planetary Science Letters, 67 (1984) 297-307

297

Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands

[61

Chemistry of solutions from the 13°N East Pacific Rise hydrothermal site G. Michard 1, F. Albar6de 2, A. Michard 2, J.-F. Minster 3, J.-L. Charlou 4 and N. Tan 5 I Laboratoire de GOochimie des Eaux, UniversitO de Paris 7, Paris (France) 2 C.R.P.G./E.N.S.G., Vandoeuvre-les-Nancy (France) ~ Institut de Physique du Globe de Paris, UniversitO Pierre et Marie Curie, Paris (France) 4 Centre Nationalpour l'Exploitation des Ocbans, Centre Ocbanologique de Bretagne, Brest (France) .s E.N.S.E.M., Nancy (France)

Received May 24, 1983 Revised version received November 29, 1983

Ten samples were recovered by the submersible " C y a n a " submersible from two groups of hydrothermal vents located 2600 m deep along the East Pacific Rise at 13°N. The ma xi mum measured temperature was 317°C and minimum pH 3.8. A systematic determination of major and trace elements has been carried out and mixing lines between a high-temperature component (HTC) and seawater are observed. The water chemistry of the HTC slightly differs for several elements at the two sites. This HTC is deprived of SO 4 and Mg and is greatly enriched in most other species. Maximum concentrations are (in units per kg): CI = 0.72 mol; Br = 1.1 mmol; Na = 0.55 mol; K = 29 mmol; Rb = 14 /~mol; Ca = 52 mmol; Sr = 170 ~mol; Mn = 750 t~mol; Fe = 1 mmol; AI = 15 ~tmol; Si = 21 mmol. For many elements, the magnitude of the anomaly relative to seawater does not compare with the results obtained from the Galapagos or East Pacific Rise 21°N. The enrichment of cations relative to seawater is likely related to the huge C1 excess through charge balance. The B r / C I ratio is close to that for seawater. However, it is not clear whether the C1 excess is due to gas release or basalt hydration (formation of amphibole chlorite or epidote). P - T dependence of SiO 2 solubility suggests that water-rock interaction last occurred at a depth in excess of 1 km below the sea floor. A mixing line of 87Sr/86Sr vs. M g / S r demonstrates that the HTCs have a nearly identical 87Sr/86Sr ratio of 0.7041 for both sites. A w a t e r / r o c k ratio of about 5 is inferred, which differs from the 1.5 value obtained at 21°N.

I. Introduction Shortly after the discovery of submarine hydrothermal activity at 21°N along the East Pacific Rise (EPR) [1-3], evidence of excess manganese and 3He in the water column at 13°N [4] led to the localization of a new hydrothermal area. The 13°N springs have been visited in January-February 1982 during the "Cyatherm" cruise with the diving saucer " C y a n a " operated by the oceanographic vessel "Suroit". At a depth of ca. 2600 m, tens of vents spout solutions at temperature in excess of Contribution I.P.G. No. 720. 0012-821X/84/$03.00

© 1984 Elsevier Science Publishers B.V.

300°C [5]. Waters were sampled from two vent groups located at 12°50'30"-103°56'10"N for the northern site (NS) and at 12o46'50"-103o56'05 '' N for the southern site (SS). The highest NS measured temperature was 317°C. At SS, the measurement failed but indication of a 284°C temperature comes from a nearby vent while a 320°C figure was obtained 30 m to the north. The aim of this paper is to present results of chemical and Sr isotopic measurements on these waters and to compare them with previous data obtained on the EPR 21°N and Galapagos vent fields. Data on REE and U [6], helium isotopes and organic chemistry will be published separately (Merlivat et al., Marty et al., in preparation).

298

2. Methods

Hydrothermal waters were collected using two different sampling devices. For hot waters ( T > 150°C), a 1.4-liter graphite syringe with an inner coating of pyrolitic carbon was designed by C N E X O from a first project of one of us. The syringe was built by the " C a r b o n e Lorraine" Company. The second device, made of A U 4 G aluminum alloy, has a content of 0.6 liter, and was used for lower-temperature springs and for dissolved organic substance studies. It was designed by two of us (F.A. and G.M.) and built by the Laboratoire d'Electricit6 et M6canique Thborique et Appliqu6e in Nancy. The graphite syringe has a significant dead volume (ca. 30 ml). Before each sampling, this volume was filled with 25-35 ml of distilled water, which produced a dilution of about 2% for all elements. This dilution affects somewhat the results, hence the analytical values have been uniformly corrected for by a same factor. This correction leaves for solution data a maximum residual error of 1% attributable to dilution effects, namely negligible relative to most analytical uncertainties. p H was measured on board as well as preliminary Si and Mn concentrations. Since the waters were rather acidic, they were stored with no preliminary acidification. Some aliquots were stored in polyethylene flasks, some others in glass bottles. Dilutions (ca. 50 times) were performed

TABLE 1 Analytical methods Element

Method

Standard

Na, K, Ca Mg Li Fe, Mn Si CI Rb, Sr

atomic absorption atomic absorption atomic absorption atomic absorption molybdate colorimetry Ag ÷ titration mass spectrometry isotope dilution flameless atomic absorption BrO 3 polarography ion chromatography

seawater distilled water seawater seawater seawater

AI Br SO 4

seawater seawater distilled water

Accuracy 2-3% 3-5% 5% 3 5% 1-2% 0.3% 1% 10% 5% 2%

upon arrival of the samples in the laboratory, within a month after sampling. The preliminary analyses showed that the samples offered a large range of dilution from a nearly pure hot end-member to pure seawater. Thus we selected analytical methods which needed small amounts of sample, allowing the measurement of a large number of elements. A summary of the analytical techniques is presented in Table 1. Most measurements were independently made in two laboratories (Paris and Brest); results largely agreed within analytical error limits. The present values correspond to Paris data. All samples contained some sulfide precipitates, which were probably not homogeneously transferred into the subsamples. As will be discussed, this could have affected the concentration of some elements. However, both the existence of well-defined mixing lines between hydrothermal water and ambient seawater and the agreement of the results obtained from different subsamples argue against any significant effect for most elements.

3. Results

The results of chemical analyses are shown in Table 2. As observed at the Galapagos spreading center and at 21°N, hot waters are deprived of SO42- and Mg [3,7]. They are enriched in S i O 2 , K , Ca, Li, Rb, Fe, Mn, A1, Sr, C1, Br and Na. In contrast with former studies, enrichment for C1 is as high as 30%. Mixing lines between the hydrothermal component and seawater (Figs. 1 and 2) are obtained by plotting all values against K concentration. The choice of K should not bias the representation and was made because K is not strongly affected by secondary precipitations, either before the exit of hydrothermal water or during mixing with ambient seawater and sampling. Positive correlations are of better quality than the negative ones. For Fe and Mn, two distinct mixing lines are obtained, one for each site. The behaviour of these two elements (especially Fe) is not a priori conservative and this difference could be related to a precipitation of secondary minerals. However, the F e / M n ratio is definitely larger (---

299 TABLE 2 Results for the samples of hydrothermal water 19G2 Na (mmol/kg) K (mmol/kg) Li (/zmol/kg) Rb (/~mol/kg) Mg (mmol/kg) Ca (mmol/kg) Sr (/Lmol/kg) Mn (~tmol/kg) Fe (~tmol/kg) AI (/.t mol/kg) SiO 2 (mmol/kg) SO2 (mmol/kg) C1 (mmol/kg) Br(mmol/kg) pH 20o 87Sr/86Sr a

24.5 551 11.1 9.9 45.4 148 576

2.8 4.0 681

22G2 26.8 602 11.9 6.6 49.0 158 663 836 5.9 18.8 3.0 712

4.33

3.83

0.70485

0.70446

24G0

24G2

26G2

467 11.0 77 1.8 48.5 12.1 88 36 0 0 0.8 27.0 546 0.92 6.90

555 29.2 663 13.6 3.5 53.0 170 765 1020 15.3 21.7 1.9 730 1.14 3.92

553 20.4 408 8.0 25.9 33.3 141 648 938 7.8 12.1 13.4 673 1.03 4.77

0.70871

0.70417

26Agl

4.1 43.2 107

5.71

0.70552

0.70721

28G0

28Ag2

32G-0

556 20.6 408 7.7 25.2 33.4 136 663 949 7.8 12.3 13.2 658 1.05 4.77

541 17.3 290 6.0 35.5 26.5 122 450 495

487 10.3 35 1.38 52.0 11.2 88 4 0 0 0.1 28.9 559 0.90 7.66

3.1 17.7 628 5.39

0.70554

0.70637

32Agl 10.3 1.73 51.4 90

6.84

0.70914

0.70891

Dives 19, 22 and 24 were made on the northern site; dives 26, 28 and 32 were made on the southern site. Samples labelled G were taken in the graphite syringes on black smokers. Samples labelled Ag were taken in aluminum syringes, in white smokers. The last digit corresponds to a particular syringe. a Range error of 2a,, is 3-7 on the last digit. Basalt values: 0.702566_+ 46 (121 ppm) for the southern site and 0.702575 + 68 (148 ppm) for the northern site. Values for graphite syringes have been corrected for 2% dilution by distilled water from the dead volume.

1.45) in the SS than in the NS (--1.30). A difference between the two sites is also revealed by Sr data. As excellent correlations are obtained between K and most elements, this difference is likely not due to the choice of the reference element. The SS trend controlled by samples 26G2 and 28G0 which were taken two days apart from the same vent. The reproducibility of the measurements is surprisingly good and suggests that seawater was added prior to sampling. Indeed, leakage of hydrothermal water was observed along this chimney (about 1 m high and 20 cm in diame-

Fig. 1. Plots of trace and major element concentrations in hydrothermal waters from 13 °N, versus K concentrations. Data from Table 2; units are per kg water. Open symbols correspond to the northern site; full symbols to the southern site. These plots correspond to mixing between "pure" hydrothermal water (maximum K concentration) and ambient seawater. The composition of the pure component can be estimated from a zero Mg concentration. For the low-temperature end member, small deviations from the ambient seawater concentrations are due to the presence of 3-4% distilled water in the syringes.

• o NCI(M) II o CI(M)

• ~ Br(rnMI e o SiO 2 (rnM)

= aAi(pM)

/ 2- -20

0.B

/

~0

oo.o-o "°~ 0.6 -,o

0.4

,0

o.~ 60

l l / I

K (mM) I

I

o

I K(mM)

I

• oCa(mM)

Mg (raM)

=.~. Rb0JM)

• D Li (pM) /

-4o

40

o

/f

(2o) 800

_2o

20

I(Io) 4.0O

// ,~)

~'o

'~ K (rnM)

4,=

~o

I0

,K (rn M) 210

30

300

S04(mM)

Sr~M)

.200

4C

°~ o

\

\

20

-I00 / \

I Mn (raM)

I ~ o = K(mM)

./? 0.6

/

d/ Io

- 1.0

o

-0.5

I K(mM) 2=o

I

IK(rnM)

/

5

0.3

I Fe(rnM)

30

/

~ /i

IO

20

/

I K(mM) 3o

Fig. 2. Same as Fig. 1. Note the differences between the high-temperature end members of the two sites.

ter); it is possible that hydrothermal water was already mixed with a m b i e n t water inside the chimney a n d that the samples represent the actual water at the m o u t h of the vent. In general, it is not k n o w n whether the difference in composition between the two sites is due to that effect, or whether it reflects differences in the pure H T C compositions. Some other elements (Si, A1, etc.) which can behave non-conservatively during mixing between h y d r o t h e r m a l a n d a m b i e n t water give fairly good mixing lines for waters sampled in the c a r b o n e device a n d stored in polyethylene flasks. Glass bottles may i n d u c e silica precipitation a n d a correlative scavenging of other elements: this is exemplified by samples 19G2 and, to a lesser extent, 22G2. The few temperature m e a s u r e m e n t s at the exit of the chimneys could u n f o r t u n a t e l y not help at the interpretation of the chemical data. Thus, the chemical composition of the H T C was deduced from an extrapolation of the mixing lines at Mg = 0. Sample 24G2 is the poorest in Mg a n d its chemical composition is fairly close to the extrapolated values for all the elements enriched in the HTC. The latter are given in Table 3 together with the values previously published for the Galapagos and E P R 2 1 ° N h y d r o t h e r m a l waters.

TABLE 3 Comparison of hot end members in hydrothermal oceanic springs

Na K Li Rb Mg Ca Sr Mn Fe SiOz C1 SO4 S

(mmol/kg) (mmol/kg) (t~mol/kg) (/~mol/kg) (mmol/kg) (mmol/kg) (mmol/kg) (~mol/kg) (t.tmol/kg) (mmol/kg) (mmol/kg) (mmol/kg) (mmol/kg)

Galapagos 21°N

13°N

259-487 18.8 680-1140 13.4-20.3 0 246-40.2 87 360-1140 + 21.9 322 595 0 +

450 24 820 26.0 0 16 90 610 1800 22 550 0 6.5

East Pacific R i s e

Reykjanes

Seawater

560 29.6 688 14.1 0 55 175 800, 1200 1050, 1850 22 740 0 -

485 44 2420 0.06 43 -

470 10.2 25 1.4 53 10.3 87 0.005 0.001 0.1 550 28 0

5.9 10.5 640 0.3

+ non-conservative. Data sources. -- Galapagos, 2]°N: Edmond et al. [3,16]; 13°N: this work; Reykjanes: Arnorsson et al. [44], Li data by C. Fouillac.

301

Aspects of the 13°N hydrothermal chemistry are discussed hereafter.

4. Silica

The calculated concentration of dissolved silica in the H T C is 22 m m o l / k g . This value is slightly higher than the most recent value of 19.4 m m o l / k g reported for 21°N [8], but similar to the Galapagos and earlier EPR 21°N estimates [3]. According to recent determinations of the activity coefficients of S i O 2 in hot saline waters [9,10], a T value of 1.05, hence an activity of dissolved silica of 23 m m o l / k g , may be assumed. At this low pH, silica ionization is negligible.

3

Recently, Walther and Helgeson [11] and Ragnarsdottir and Walther [12] retrieved experimental data on quartz solubility at high temperature and pressure: the figure of 23 m m o l / k g is greater than quartz solubility at saturation pressure whatever the temperature. In Fig. 3, silica isopleths are plotted in a P - T diagram. In this temperature range, the quartz solubility is rather sensitive to pressure, so that independent temperature or pressure estimates are needed. The minimal pressure is 600 bar; if one adopts a temperature of 350°C for the hydrothermal water, it should be increased up to ca. 1000 bar. Assuming that, within the crust, pressure is lithostatic, one can conclude that the water-rock interaction is taking place deeper than 1 km below the ocean floor. Such a depth is consistent with the estimations of the altered basaltic layer thickness and the seismic data from a nearby ridge segment [13]. This result suggests that the basalt water interactions affect a thick layer of the oceanic crust and, as discussed below, with a rather low w a t e r / r o c k ratio. This interaction must be different from the reactions between rocks and sea-water which take place on the ocean floor (palagonitization) at low temperature and with a large w a t e r / r o c k ratio.

5. Alkali elements

2.

300

350

400

T°c

Fig. 3. Silica isopleths in a P - T diagram. Data from Walther and Helgeson [11]. For an activity of 23 m m o l / k g dissolved silica, it is necessary that water-rock interaction takes place at a pressure in excess of 600 bar.

As for other hydrothermal fields (Galapagos, 21°N), the composition of the H T C shows a huge enrichment in K, Na, Rb and Li and distinctively lower K / R b and K / L i ratios compared with seawater. Uptake of a large fraction of the reacting seawater can be discarded as a factor of major chemical changes (see below). Therefore, these elements are leached from the oceanic crust during the alteration processes which result from the water-rock interaction. As pointed out by Hart and Staudigel [14], this contrasts with most observations on DSDP or dredged rocks which show a pronounced enrichment of K, Rb and Li; this is attributed to low-temperature interaction with seawater at shallow depth below the sea floor. A similar statement has been made for other hydrothermal sites and other elements like Ba or rare earth elements [6].

302 Hence, as pointed out by Staudigel and Hart [15], data from hydrothermal solutions provide information on alteration processes which differ from those recorded by the chemistry of DSDP or dredged rocks. It is likely that both kinds of studies refer to processes which do not necessarily happen concomitantly nor occur at the same stratigraphic or geotectonic site. Cold, shallow a n d / o r off-ridge alterations differ strongly from high-temperature water-rock reactions occurring at depth along the ridge axis. This raises the question of whether results from dredged rocks, DSDP samples (mostly coming from the uppermost layer 2) and ophiolites are a consistent data base for mass balance calculations relevant to the b a s a l t / s e a water exchanges on the scale of the whole oceanic crust. The proportion of oceanic crust which is affected by each of these contrasting alteration processes and their chronological relationship are far from sufficiently known at the moment.

6. Chloride

C h l o r i d e - - a n d some other elements--are definitely more concentrated (by 30%) in 13°N hydrothermal waters than in Galapagos [16] and 21°N waters [3]. The similitude of these variations with those for Na and K and the larger differences (by a factor 2) for Ca and Sr between 13°N and 21°N can be related to the influence of C1 on the equilibrium between minerals and water [17]. On the contrary, Li and Rb are identical or even less concentrated at 13°N than at 21°N; as we shall see later, this can be related to a higher w a t e r / r o c k ratio in the former case. Bromide was analyzed in order to precise the origin of chloride. As can be seen from TaMe 2, the B r / C I in hydrothermal water is not significantly different from that in seawater. The chloride increase could be related to the rock hydration upon formation of hydrous metamorphic minerals (chlorite, actinolite, epidote). A maximum amount of water involved in this reaction can be obtained by calculating the amount of water needed for the hydration of all the mag-

nesium and iron(II) present in the rock plus water system and its transformation into chlorite. For a w a t e r / r o c k ratio of 5, the increase of B r - and Ci (assumed to be unreactive) is 1%; it is 4% for a w a t e r / r o c k ratio of 1. As this range of water/rock ratio has been found reasonable for submarine hydrothermal waters ([3], see below), this process cannot explain the observed increase. If chloride were to be released from the crust, this would require a C1 concentration in rock of 6200 p p m for a w a t e r / r o c k ratio of 1, and of 3.1% for a w a t e r / r o c k ratio of 5. The mean value of CI in ridge basalts is 30-50 p p m [18,19] and can reach 400 p p m in the Azores area [18]. Recently, Byers et al. [20] measured C1 concentrations as high as 3700 p p m in andesites from the Galapagos spreading center. However, these andesites are very uncommon and differentiated rocks. Furthermore, the highest value is still two to ten times less than the values needed from the observed C1 increase in the hydrothermal water. Byers et al. [20] give no Br results but in Azores, Schilling et al. [18] found the same B r / C I ratio as in other areas along the Mid-Atlantic Ridge; this value is not significantly different from the Br/C1 ratio of seawater. A third possibility is an input of magmatic chloride upon a gas transfer process. This origin of chloride has been suggested by Sigvaldasson [21] and Henley and Ellis [22], in their discussion of the chemical composition of continental geothermal systems. In both cases, the amount of C1 present in the rocks is much too low for explaining the concentrations in hydrothermal waters. Schilling et al. [23] also found that only 1 / 3 of chloride and bromide present in seawater and evaporitic rocks have been brought from the mantle by magmatic rocks. Craig and Lupton [24] proposed a "degassing increment" to explain the 3He anomalies in seawater. A similar process would be needed for C1. Interestingly, sample 24G2 also contains particularly high 3He concentration and 3 H e / 4 H e ratio (Merlivat et al., unpublished data); this supports the idea of an exceptional process effective in this area. The Br/C1 ratio for this input would have to be similar to that for seawater.

303

7. Water/rock ratio

7.1. Heat content

Energy and matter are exchanged in huge quantities during hydrothermal processes which involve water of doubtless marine origin. But the mechanisms at play, and the nature of the original rocks and of their mineralogical changes are less obvious. From the analogy of ophiolites with the oceanic crust, the interaction would be located within a zone dominated by pillow-basalts which is ca. 1 km thick (0.3-6 km range) [25,26] or within the underlying sheeted-dyke complex which feeds the basaltic layer [27]. Typical examples like Pindos [28], Cyprus [29,30] or Oman [31,32] shed some light on these processes. The SVSr/86Sr and ~ l S o data have repeatedly demonstrated the marine origin of the water which circulates in the convection-driven systems. The resulting mineralogical association belongs to the greenschist facies (quartz-albite-epidote-chlorite-actinote). If the end product of the alteration processes may be reasonably well constrained, the nature of the original rock which reacts with seawater to produce the hydrothermal fluids is more questionable. Volcanic glasses are not expected to be found at depth and coarse-grained rocks (gabbro, peridotite) are presumably too poor in those leachable elements enriched in the H T C (alkali elements, Ba, REE). The water-rock interaction is more likely to affect the fine-grained equivalent of basaltic rocks (dolerites), provided their temperature prior to the reactions is close to the solidus or an efficient preheating of the water exists to maintain the w a t e r / r o c k ratio in a realistic range (see below). The reaction may also be localized along the thermal boundary layer of the magma chamber beneath the ridge axis. The requirement of chlorine addition to the H T C could support this latter interpretation. In view of these uncertainties, no specific mineralogical reaction will be called upon to describe the water-rock interactions. It will be parameterized by the water-rock ratio, which can be estimated by several balance calculations. The latter are not taken to be precise but provide useful information when considered simultaneously.

It is assumed that, in a magmatic environment ( T = 1150°C), rocks in contact with cold seawater are cooled to the 350°C temperature of the vents. Following Corliss et al. [33], a w a t e r / r o c k ratio can be estimated from the heat exchange equation; including a latent heat of solidification of the m a g m a and possible hydration heat does not alter significantly the results. The calculated value is of about 0.7. 7.2. "Soluble" elements

Some elements such as Li, Rb, Cs, Ba are both very soluble in water are incompatible in the basalt. It is assumed that during water-rock reaction, a large part of these elements is transferred to the hydrothermal water [16]. With these assumptions, one can write: CFR -- CAR W / R - CH w _ Cs w

where W and R are respectively the mass of water and rock in mutal reaction and CHW, Csw, CFR, CAR are respectively the concentrations of the elements in the HTC, seawater, fresh and altered rock. Assuming that CAR << CFR , an upper limit of the W / R ratio can be obtained. Using a water which previously exchanged with the rocks at low temperature instead of seawater should not greatly affect the results since it most likely lost alkalies. Using the same basalt value for Li and Rb as Edmond et al. [3,16], the W / R ratio lies between 1 and 3. The figures compare with the upper values of the Galapagos vent field and significantly exceed those calculated for 21°N [3,8]. 7.3. 87Sr/ S 6 S r ratios

The 87Sr/86Sr ratios were measured on all the collected EPR 13°N samples. This ratio cannot be simply compared to K as mixing would appear as a curved line in such a diagram. The plot of this ratio vs. M g / S r defines a mixing line between hydrothermal and seawaters (Fig. 4). The H T C has a 87Sr/86Sr value of 0.7041, i.e. higher than at

304 I

t

I

l

I

87Sr / 86Sr

I

~

SEAWATER_

..jo~

.709

.708

/ / 1

.707 /

/

21°N

.706

.705

,©/ .704

©

/

/ /

/ /

7°3 l

Mg / Sr I

0

I

I

200

I

400

I

I

600

Fig. 4. 878r/86Sr vs. Mg/Sr ratio in hydrothermal waters from 13°N, Mg/Sr unit is mole/mole. The plot shows mixing between seawater and hydrothermal water of 87Sr/86Sr ratio 0.7041. A water/rock ratio of 5.4 is implied by this value. The mixing line from the EPR 21°N hydrothermal site is shown for comparison [34].

21°N (0.7030 [34]). As the dissolution/precipitation processes should not fractionate the Sr isotopic ratios, and assuming that the mixing of rock-derived Sr and marine Sr is complete, the W/R ratio can be calculated as in Albar6de et al. [34]. Using the Sr concentration of the basalts for SrFR , a value of 5.4 is obtained at 13°N instead of 1.5 at 21°N.

7.4. Comparison As heat exchange is probably faster than chemical exchange, the thermal W/R ratio is expectedly lower than the W/R calculated from chemical and isotopic parameters. In addition to these purely kinetic differences, a significant preheating of the downfiowing seawater (heat added by conduction in the descending limb of convection cells) would also increase the effective thermal W/R ratio and help resolve this apparent contradiction. Furthermore, there is only a little latitude to bring the chemical and isotopic W/R ratio into coincidence

by either increasing K, Li, Rb or decreasing Sr contents of the fresh rock relative to the basalt, since only unusual rock types would show suitable concentration levels. It may be better envisioned that the discrepancy between the W/R values results from a second-order failure of the simple exchange models: the almost unescapable alternative is the existence of a continuum in the heat and mass exchange processes leading to preheated and K-, Li-, Rb-, etc., enriched solutions before the final outburst of the high-temperature reaction, which expells violently the modified solution towards the sea floor. Hence evidence is obtained that the W/R ratio should be small (less than 6) but that preliminary heat and mass exchanges with the conduit walls prevent from using its face value with an exaggerated confidence.

8. Sulphate depletion Even if hydrothermal waters are concentrated in sulfide ions, their total amount of sulfur is much

305 lower than that of sea water. ~345 data on solid precipitates from 21°N cluster around + 1.5%o and suggest that the sulfur present in the HTC is mostly derived from the basalt (e.g. [35,45]). These observations demand a mechanism by which this element could be quantitatively removed from sea water. It may be anhydrite or Mg3(SOa)2(OH) 2 precipitation [7]. Alternatively, the seawater sulphate is expected to be reduced by the ferrous iron present in the basalt; massive precipitation of copper and iron sulfides would produce a subsequent removal of the reduced sulfur. Anhydrite deposition would occur subsequently during the mixing of SO4-depleted, Ca-rich HTC with SO42-rich seawater [3,34]. The available experimental evidence suggests that, at temperatures in excess of 350°C, reduction of SO2 - by the basaltic ferrous iron is a major process [36]. This reduction process can be investigated in the system Ca, Fe, S, e-. Phase rule indicates that four mineral or element concentrations are necessary to define such a system and the composition of the solution. As dissolved calcium, sulfide and iron are accurately known, sulphate activity can be calculated. Such thermodynamical calculations can be thoroughly computed. However, at 350°C, the uncertainty of the thermodynamical constants is rather important. We will use here a simplified method of calculations which will enlight the induced uncertainty on the result, yet provide useful information, SOn2- activity will be calculated in each of the following hypotheses: (1) the redox processes are controlled by iron--i.e, dissolved ferrous iron and a solid containing ferric iron (hematite or magnetite); (2) the sulphate ion activity is governed by anhydrite solubility.

8.1. Redox process According to experimental studies of Franck [37] on HC1 solutions and of Crerar et al. [38] on the complexation of iron at very high temperatures, the main dissolved species in the hot end member are FeCI°2, HC1 °, H2 S°, HSO 4 and CI-. The reaction between H2S, sulphate, dissolved iron and magnetite can be summarized by:

HSO 4 + 12 FeCI~ + 12 H 2 0 = 4 Fe304 + H2S + 23 HC1 ° + C1 and: ( H S O 4 ) = (H2S)(HCl°)23(C1-) % (FeCl~)12 K The equilibrium constant K can be calculated from the studies of Sweeton and Baes [39] on the solubility of magnetite in hot waters up to 300°C and in a reducing medium and from the tabulated thermodynamic data on HSO 4 and H2S [40]. As (H2S) was not measured in the 13°N samples, the values were assumed to be similar to those at 21 °N (e.g. [7]). As the dependence of (HSO4) on (H2S) is only linear, this choice is not critical. The activity coefficients of uncharged species are taken to be 1; the activity coefficient for C1- comes from the table of activity coefficients of NaC1 up to 350°C [41]. The calculated activity of HSO 4 is 10 -17. It is very difficult to assess an uncertainty on this figure: if the log of HC1 ° or FeCI~ stability constants have an uncertainty of about 0.5 unit, the uncertainty on the result will be of about 1012!

8.2. A nhydrite solubility The dissolution reaction can be written: HSO 4- + C I - + Ca 2+= CaSO 4 + HC1 ° (HSO4-) -

(HCI°) 1 (C1-)(Ca 2+) K '

K ' is extrapolated from Yeatts and Marshall data [42] and Ca 2+ activity from isopiestic measurements in (1 - y ) NaCl, y CaC12 system up to 200°C [431. The calculated activity of H S O ; is 10 -6. The uncertainty is about one order of magnitude. Despite the large uncertainty of these calculations, mainly related to the difficulty of extrapolating the thermodynamic data up to 350°C, they can indicate that bisulphate can be controlled by a redox process. A complete reduction of the seawater sulphate b y the ferrous iron present in the basalt requires a W / R ratio of less than 6 [36]; the W / R ratio of the EPR 13°N hydrothermal solutions seems fairly close to this theoretical limit.

306

9. Conclusions The hydrothermal solutions sampled at 13°N show common features with oceanic hydrothermal waters analyzed in previous studies: --complete depletion of Mg and SO4; --large enrichment in Mn, Fe, Li, Rb, K, Ca, SiO 2.

The high silica content can be explained only by a reaction between rock and water at a rather high pressure (> 600 bar). Concentrations of soluble elements (Li, Rb) and strontium isotope ratios both indicate a water/rock ratio of 2-5, apparently larger than at 21°N. The puzzling result of this study is the high value of dissolved chlorine. Influx of a significant quantity of this element from a magmatic source is presently our preferred interpretation.

Acknowledgements This work has been made possible through the efforts of the crew, of the chief scientist R. Hekinian and of colleagues of the R / V "SuroW' and the submersible "Cyana". MM. Bougault, Echardour and Bervas from CNEXO and the company "Carbone Lorraine" have successfully mastered the difficult technology for graphite syringes. M. Murgia built the Ag-type syringes. MM. Evrard and Lavergne performed an important part of the analyses. CNEXO and the ATP-CNRS "G6ochimie-M6tallog6nie", "G6ophysique et G6ophysique des Oc6ans" and "Oc6anographie Chimique" are thanked for financial support.

References 1 J. Francheteau, H.D. Needham, P. Choukroune, T. Juteau, M. Seguret, R.D. Ballard, P.J. Fox, W. Normark, A, Carranza, D. Cordoba, J. Guerrero, C. Rangin, H. Bougault, P. Cambon and R. Hekinian, Massive deep sea sulphide ore deposits discovered on the East Pacific Rise, Nature 277, 523, 1977. 2 F.N. Spiess, K.C. McDonald, T. Atwater, R.D. Ballard, A. Carranza, D. Cordoba, C. Cox, V.M. Diaz-Garcia, J. Francheteau, J. Guerrero, J. Hawkins, R. Lhaymon, R. Hessler, T. Juteau, M. Kastner, R. Larson, B. Luyendick,

J.D. McDougall, S. Miller, W.R. Normark, J. Orcutt and C. Rangin, East Pacific Rise: hot springs and geophysical experiments, Science 207, 1421, 1980. 3 J.M. Edmond, K.L. Von Damm, R.E. McDuff and C.I. Measures, Chemistry of hot springs on the East Pacific Rise and their effluent dispersal, Nature 297, 187, 1982. 4 J. Boulegue, H. Bougauh and J.-L. Charlou. Hydrothermal activity on the East Pacific Rise between 15°N and 7°S, EOS 61,992, 1980. 5 R. Hekinian, J. Francheteau, V. Renard, R.D. Ballard, P. Choukroune, J.-L. Chemin+,e, F. Albar6de, J-.F. Minster, J-.C. Marty, J. Boulegue and J-.L. Charlou, Intense hydrothermal activity at the axis of the East Pacific Rise near 13 °N. Submersible witnesses the growth of sulfide chimney, Mar. Geophys. Res. 6, 1, 1983. 6 A. Michard, F. Albar6de, G. Michard, J-.F. Minster and J.-L. Charlou, REE and U in high temperature solutions from the EPR 13°N hydrothermal vent field, Nature 303, 795, 1983. 7 R.E. McDuff and J.-M. Edmond, On the late of sulfate during hydrothermal circulation at mid-ocean ridges, Earth Planet. Sci. Left. 57, 117, 1982. 8 K.L. Von D a m m and J.-M. Edmond, Chemistry of 21°N East Pacific Rise hydrothermal solutions, EOS 63, 1015, 1982. 9 W.L. Marshall and C.T.A. Chen, Amorphous silica solubilities, V. Prediction of solubility behaviour in aqueous mixed electrolytes to 300°C, Geochim. Cosmochirn. Acta 46, 289, 1982. 10 R.O. Fournier, A method for calculating quartz solubility in aqueous chloride solutions, Geochim. Cosmochim. Acta 47, 579, 1983. 11 J.V. Walther and H.C. Helgeson. Calculation of the thermodynamic properties of aqueous silica and the solubility of quartz and its polymorphs at high pressures and temperatures, Am. J. Sci. 277, 1315, 1977. 12 K.V. Ragnarsdottir and J.V. Walther, Pressure sensitive silica geothermometer determined from quartz solubility experiments, Geochim. Cosmochim. Acta 47, 941, 1983. 13 J.J. Ewing and G.M. Purdie, Upper crustal velocity in the ROSE area of East Pacific Rise, J. Geophys. Res. 87, 8397, 1982. 14 S.R. Hart and H. Staudigel, The control of alkalies and uranium in sea-water by ocean crust interaction, Earth Planet. Sci. Lett. 52, 202, 1982. 15 H. Staudigel and S.R. Hart, Alterations of basaltic glass: mechanisms and significance for the ocean crust sea water budget, Geochim. Cosmochim. Acta 47, 337, 1982. 16 J.M. Edmond, C. Measures, R.E. McDuff, L.H. Chan, R. Collier, B. Grant, L.I. Gordon and J.B. Corliss, Ridge crest hydrothermal activity and the balance of the major and minor elements in the ocean: the Galapagos data, Earth Planet. Sci. Lett. 46, 1, 1979. 17 G. Michard, ROle des anions mobiles dans le transport des 616ments par les solutions hydrothermales, C.R. Acad. Sci. Paris 295, 45l, 1982. 18 J.G. Schilling, M.B. Bergeron and R. Evans, Halogens in

307

19

20

21

22

23 24

25

26 27 28

29

30

31

32

33

the mantle beneath the North Atlantic, Philos. Trans. R. Soc. London, Ser. A 297, 147, 1980. E. Ito, D.M. Harris and A.T. Anderson, Alteration of oceanic crust and geologic cycling of chlorine and water, Geochim. Cosmochim. Acta 47, 1613, 1983. C.P. Byers, D.M. Muenow and M.O. Garcia, Volatiles in basalts and andesites from the Galapagos spreading center, Geochim. Cosmochim. Acta 47, 1551, 1983. G.E. Sigvaldasson, Fluid in volcanic and geothermal systems, in: Proc. Nobel Symposium, Chemistry and Geochemistry of Solutions at High Temperature and Pressure, Karlokoga, Pergamon Press, London, 1979. R.W. Henley and A.J. Ellis, Geothermal systems ancient and modern: a geochemical review, Earth Science Rev. 19, 1, 1983. J.C. Schilling, C.K. Unni and M.L. Bender, Origin of chlorine and bromine in the ocean, Nature 273, 631, 1978. H. Craig and J.E. Lupton, Helium-3 and mantle volatiles in the ocean and the oceanic crust, in: The Sea, Vol. 7, p. 391, J. Wiley, New York, N.Y., 1981. R.G. Coleman, Plate tectonic emplacement of upper mantle peridotite along continental edge, J. Geophys. Res. 76, 1212, 1971. E.M. Moores and E.D. Jackson, Ophiolites and oceanic crust, Nature 250, 136, 1974. R.G. Coleman, Ophiolites: Ancient Oceanic Lithosphere?, Springer, New York, N.Y., 1977. C.J. Allbgre, R. Montigny and Y. Bottinga. Cortege ophiolitique et cortege oc~anique. G6ochimie comparb,e et mode de gen~se, Bull. Soc. G~ol. Fr. 15, 461, 1973. Z.E. Peterman, R.G. Coleman and R.A. Hildreth, 87Sr/86Sr in mafic rocks of the Troodos massif, Cyprus, U.S. Geol. Surv. Prof. Pap. 750, 1570, 1971. T.H.E. Heaton and S.M.F. Sheppard, Hydrogen and oxygen isotopic evidences for sea water hydrothermal alteration and ore deposition, Troodos complex, Cyprus, in: Volcanic Process in Ore Genesis, Geol. Soc. London, Spec. Pap. 7, 42, 1977. M.T. McCulloch, R.T. Gregory, G.J. Wasserburg and H.P. Taylor, Jr., Sm-Nd, Rb-Sr, and 180/160 isotopic systematics in an oceanic crustal section: evidence from the Samail ophiolite, J. Geophys. Res. 86, 2721, 1981. R.T. Gregory and H.P. Taylor, Jr., An oxygen profile in a section of cretaceous oceanic crust, Samail ophiolite, Oman: evidence for 180 buffering of the oceans by deep ( > 5 km) sea water hydrothermal circulation at mid-ocean ridges, J. Geophys. Res. 86, 2737, 1981. J.B. Corliss, J.M. Edmond and J.L. Gordon, Some implications of heat/mass ratios in Galapagos Rift hydrothermal

34

35

36

37

38

39

40

41

42

43

44

45

fluid for models of seawater-rock interaction and the formation of oceanic crust, in: Edwing Symposium, Vol. 2, p. 383, M. Talwani, ed., 1979. F. Albarb.de, A. Michard, J.-F. Minster and G. Michard, 87Sr/86Sr ratios in hydrothermal waters and deposits from the East Pacific Rise at 21°N, Earth Planet. Sci. Lett. 55, 229, 1981. M. Arnold and S.M.F. Sheppard, East Pacific Rise at latitude 21°N: isotopic composition and origin of the hydrothermal sulfur, Earth Planet. Sci. Lett. 56, 148, 1981. M.J. Mottl and H.D. Holland, Chemical exchange during hydrothermal alteration of basalt by sea water, I. Experimental results for major and minor components of sea water, Geochim. Cosmochim. Acta. 42, 1103, 1978. E.U. Franck, Hochverdichteter Wasserdaupf, III. Ionendissoziation von HC1, KOH and H 20 in Oberkritischen Wasser, Z. Phys. Chem. 8, 206, 1956. D.A. Crerar, H.J. Susak, M. Borcsik and S. Schwartz, Solubility of the buffer assemblage pyrite+pyrrhotite+ magnetite in NaCI solutions from 200 to 350°C, Geochim. Cosmochim. Acta 42, 1427, 1978. F.H. Sweeton and C.F. Baes, Jr., The solubility of magnetite and hydrolysis of ferrous iron in aqueous solutions at elevated temperatures, J. Chem. Thermodyn. 2, 479, 1970. H.C. Helgeson, D. Kirkham and G.C. Flowers, Theoretical predictions of the thermodynamic properties of aqueous electrolytes, IV. Calculations of activity coefficients, osmotic coefficients and apparent molal and standard and relative partial molal properties to 600 ° and 5 kbars, Am. J. Sci. 282, 1249, 1981. L.F. Silvester and K.S. Pitzer, The thermodynamics of geothermal wells, I. Thermodynamic properties of vapor saturated NaC1 solutions, Rep. LBL-4456, 1976. L.B. Yeatts and W.L. Marshall, Apparent invariance of activity coefficients of calcium sulfate at constant ionic strength and temperature in the system CaSO4-Na2SO 4NaNO3-H20 to the critical temperature of water, J. Phys. Chem. 73, 81, 1969. H.F. Holmes, C.F. Baes, Jr. and R.E. Mesmer, Isotopic studies of aqueous solutions at elevated temperatures, III ((1-y)NaCl+yCaCl2), J. Chem. Thermodyn. 13, 101, 1981. S. Arnorsson, E. Gunnlaugsson and H. Svavarsson, The chemistry of geothermal waters in Iceland, II. Mineral equilibria and independent variables controlling water compositions, Geochim. Cosmochim. Acta 47, 547, 1983. R. Hekinian, M. Fevrier, J.-L. Bischoff, P. Picot and W.C. Shanks, Sulfide deposits from the East Pacific Rise near 21°N, Science 207, 1433, 1980.