Metallogenesis at Santorini - a Subduction-zone Related Process. II: Geochemistry and Origin of Hydrothermal Solutions on Nea Kameni, Santorini, Greece

The most acid solutions (pH = 5.5) are anaerobic and enriched with free carbon dioxide, iron, manganese, silica, barium, copper and zinc. The deficiency of magnesium and uranium and excess of calcium and potassium in these solutions indicate major alteration at depth in the hydrothermal cell.

The results support the hypothesis that hydrothermal systems in young island arcs and subduction zones tend to be metalliferous, in contrast to hot springs located in major land areas and older island arcs with more silicic volcanism.

INTRODUCTION

Many submarine hot springs at deep-sea spreading centres supply iron, manganese, barium and copper, producing large metalliferous deposits (Boström 1980; Rona 1988). On the other hand, most hot springs in major land areas, for example Yellowstone National Park, Iceland and New Zealand, are generally poor in metals (Allen and Day 1935; Barth 1950; Waring 1965; Rowe et al. 1973). Hydrothermal systems near subduction zones such as Santorini and other volcanic centres on young calc-alkaline island arcs (Zelenov 1972) seem to develop an intermediate metallogenetic character with iron-rich solutions and exhalative-sedimentary deposits.

Ore genetic relations suggest that most exhalative-sedimentary deposits have formed at island arcs. It is, therefore, of interest to study the nature and distribution of metalliferous hot springs in such settings; these sites may also be important centres for hydrous magmas (Burnham 1979). Studies of the easily accessible subduction-zone hot-spring system at Santorini, Greece, should shed more light on this phenomenon.

Santorini is located on a major arc-shaped swell in the Aegean Sea, stretching from Methana and Aigina over Milos and Santorini to Nisyros. These islands are calc-alkaline with hot-spring deposits of varying extent (Pichler and Kussmaul 1972). These features and seismic data suggest that this arc is associated with a subduction zone (Berckhemer 1978; Meulenkamp et al. 1988). Nea and Palaea Kameni in the middle of the Santorini caldera are the centres of present fumarolic activity. They have for a long time been known for their hydrothermal activity and deposition of iron-silica-rich muds (Fouqué 1879; Brun 1911). The metalliferous process is particularly intense in two bays, one on Nea Kameni and the other on Palaea Kameni (A and P, Fig. 1) as shown by Butuzova (1969), Puchelt (1973), Puchelt et al. (1973), Smith and Cronan (1983), Boström and Widenfalk (1984), and Boström et al. (1990a; 1990b). In this paper we present chemical analyses of unfiltered surface water from bay A, and show how they produce exhalative sediments at the Kameni islands and in the caldera of Santorini.

FIELD RELATIONS

Bay A (Fig. 1a, b) cuts deeper into the island than the other bays of Kameni islands, and the innermost part is narrow and shallow. The bay is in essence the gap between two sub-parallel blocky lava flows. The hydrothermal deposits are better protected here than in other bays, but this is not the only explanation for the accumulation of hot-spring products. The springs on the south shore of the innermost basin (Fig. 1b) are more active than on other sites in the Kameni islands, which was also reported by Butuzova (1969). On the south shore of the innermost basin, two or three springs show a steady, but quiet flow and without obvious bubbling. Debouching particles indicate flow rates of 3-12 m/min. Along most of the north shore of the inner basin there are, furthermore, numerous springs which have a more intermittent flow. Bubbling occurs widely here and takes place in bursts. Spring waters at the south shore are generally clear, except for some silica-rich white flakes. Ferric hydroxide particles seem to appear at some distance (0.25-0.75 meters) from the spring orifices. It is unclear where these particles first form. They may represent material torn loose from the beach blocks or resuspended from the muddy bottom of the bay, or be due to rapid precipitation of dissolved iron. Resuspension may be the most likely explanation, since floats of iron hydroxide are most abundant at the bubbling springs of the north shore. The springs are the ultimate sources of iron and silica though. Calculations by Boström et al. (1990a) indicate that the springs are adequate to explain the exhalative-sedimentary deposition in Bay A.

Active springs produce distinct colour changes in the sea water from normal deep-blue colour in the open caldera to a pale-greenish shade; in the innermost part of some bays the water may even turn yellowish-brown. Fronts between sea water and debouching spring water are generally easy to spot. In calm weather these colour zones spread concentrically from the springs, but under windy conditions most of the outer part of Bay A may be filled with open-caldera water. Large eddies of alternating greenish spring water and open-caldera water have been observed. Because of these weather disturbances all fieldwork was done during calm weather conditions.

SAMPLING AND ANALYTICAL PROCEDURE

Water samples were collected in 1984 and 1986-88 with acid-washed polyethylene bottles immersed in the water. Temperature, dissolved oxygen, pH, alkalinity and free carbon dioxide were measured in the field. Dissolved oxygen was determined in situ with an oxygen probe connected to a YSI^® 58 instrument. The pH was measured with a portable Metrohm^® 602 pH-meter equipped with a standard glass combination electrode. The pH-meter was calibrated with standard buffer solutions in the field. Alkalinity was titrated with 1.6N H₂SO₄ on 100 ml water using HACH^® digital titrator, all samples being titrated to a pH of 4. 5. The amount of free dissolved carbon dioxide was measured by titration with 0.36 or 3.6N NaOH (depending on the amount of carbon dioxide) to a pH of 8.3. Trace elements were determined either by co-precipitation forming Mg(OH)₂ at pH 10 from one litre of unflltered water, or by direct determination in trace-element-rich waters. Samples of open-caldera sea water are generally poor in iron, aluminium, manganese, copper and zinc; for this reason some of the results for these elements are uncertain. Furthermore, determination of barium concentration in normal sea water is difficult; therefore, the barium data for open-caldera water are uncertain. However, these constituents are easily determined in the hot-spring water. The sensitivity and reproducibility of the co-precipitation method, and the required cleaning of sodium hydroxide from trace metals have been discussed by Koroleff (1984) and Boström and Boström (in press).

In the laboratory, the elemental composition of the waters was determined by atomic emission spectrometry using inductively coupled plasma as the excitation source (ICP-AES). The uranium concentration and isotope ratio were determined with alpha-ray spectrometry after chemical separation, as described by Löfvendahl (1987). The error in the uranium concentration is below ±  11% (two sigma) and in the ²³⁴U/²³⁸U activity ratio (AR) below ± 12%. The errors embrace the accumulated error in counting statistics and the error in the ²³²U spike.

RESULTS AND DISCUSSION

The analytical data (Table 1) are grouped in relation to distance from the springs (= sprd) in the inner part of Bay A. Samples of sea water in the open caldera and the strait between the islands (SWOC) are located more than 350 m from the springs. Samples from the outer zone of the bay (NKOZ) have sprd values between 300-330 m, those from the middle zone (NKMZ) 40-230 m, and those from the inner zone (NKIZ) 2-32 m. Samples obtained 0-1 m from the hydrothermal springs are referred to as NKHS.

The results show that the waters vary significantly in temperature and composition with distance from the springs (Table 1). Beyond approximately 350 m from the springs the values approach a constant sea-water value. A series of twelve fairly complete analyses of surface waters collected in 1988 (Fig. 1b) illustrates this relation, with most parameters showing a distinct correlation with distance from the springs. Oxygen saturation and pH are low in the spring waters, and outflowing surface water shows depletion in oxygen up to two hundred metres from the springs (Fig. 2b, c). Both oxygen-saturation and pH show a more or less linear increase between 0 to 300 m from the spring, levelling out seawards. This distance coincides with the calm-weather position of the front between the zone of greenish-blue spring waters and open-caldera sea water outside, suggesting that the mixing process occurs inside this boundary.

The major sea-salt components, potassium, magnesium and calcium (Fig. 2d-f) show very distinct variations, while the analytical precision for sodium is inadequate to discern any regular variation. The high calcium and potassium values in the spring waters represent 22 and 13% increases, respectively, in relation to open-caldera sea water, suggesting that much calcium and potassium has been obtained by leaching of subsurface bedrock. Magnesium shows a 5% loss and sulphur content in the spring waters is distinctly lower than in open-caldera sea water (Fig. 2f, g), indicating incorporation of these elements in rock alteration products at depth. Uranium is strongly depleted in the waters close to the springs (Table 1), while SWOC shows values close to normal sea water. The ²³⁴U/²³⁸U AR, on the other hand, shows no significant departure from the activity ratio of sea water, which is 1.14. The low uranium concentration close to the springs suggests major uptake of uranium in rocks at depth.

Silica, aluminium, iron, manganese, barium, copper and zinc (Fig. 2h-n) are enriched in the hydrothermal spring waters. The concentration of silica is highly variable though in the different springs (Fig. 2h). The excesses of iron, silica, calcium, potassium, manganese and barium in the hot-spring waters at Nea Kameni, and the deficiencies of magnesium, sulphur, oxygen and uranium are also observed in hot springs in the deep sea (Edmond et al. 1979 a, b).

However, iron-rich solutions and muds form also at island arcs, such as the Kuriles (Gorshkov 1958; Naboko 1963; Fyfe et al. 1978; Zelenov 1964), Tjiater and Banu Wuhu, Indonesia (van Bemmelen 1949; Zelenov 1964; 1972), Matupi Harbour (Ferguson and Lambert 1972), Deception Island (Elderfield 1972), and Kick 'em Jenny off Grenada (Lesser Antilles; Robson and Tomblin 1966). Similar conclusions were presented by Zelenov (1972), based on a study of hot-spring-related metalliferous processes at well-developed young island arcs in eastern and southeastern Asia. It appears that these metalliferous springs, also, are related to calc-alkaline volcanic activity in young island arcs, whereas springs in older island arcs, or in continental areas with more silicic volcanic and plutonic activity, are conspicuously poor in iron, aluminium and manganese (Allen and Day 1935; Waring 1965; Rowe et al. 1973; White et al. 1988; Boström et al. 1990a, 1990b).

Chemical compositions of hot springs at young island arcs of calc-alkaline affiliation such as the Kuriles (K 1-3), Indonesia (BW, J) and Nea Kameni (NK) are given in Table 2 and Fig. 3 a-f. The composition of hot springs in continental areas, island arcs and oceanic islands with more complex igneous development, such as Iceland (Ic 1-2), Wairakei, New Zealand (W) and Yellowstone National Park (Y 1-4) are also shown in Table 2 and Fig. 3 a-f. The data represent means of 116 analyses and show that the former group on the whole is much richer in iron, aluminium, sulphur and magnesium, and poorer in silica. It is of interest to note that even the sea water dominated hot-spring systems at Reykjanes on Iceland fall in the second group. The concentration of calcium, however, largely seems to be controlled by the degree of sea water exposure. The content of silica in the hot-spring systems, finally, seems to be correlated with both temperature of the system (Fig. 3e) and the geological classification.

The reasons for this subdivision of hot springs are poorly understood. The metal content seems to show no simple relation with temperature or rate of water-flow through the hydrothermal cell; thus the sea water dominated springs at Reykjanes, Iceland, lack equivalents to the metalliferous deposits at Nea Kameni, Banu Wuhu, Tjiater and Kuriles, in spite of higher temperatures and abundance of carbon dioxide (Kristmannsdottir 1983). The carbon dioxide content in the springs at Santorini was suggested as an explanation for their metalliferous character by Harder (1964), but this idea is difficult to defend. An overwhelming number of volcanic emanations are rich to exceptionally rich in carbon dioxide (White and Waring 1963; Chaigneau et al. 1960), whereas coexisting volcanic products and solutions are distinctly non-metalliferous. Likewise, the hot-spring systems in New Zealand with high flow rates are exceptionally poor in many trace elements. In Nea Kameni, on the other hand, temperatures and pH are not particularly spectacular, compared with many other hot-spring systems, yet the iron abundances are distinctly higher than in any of the spring systems in Yellowstone, Iceland or New Zealand.

Obviously, compositions of volcanic emanations are controlled by physico-chemical relations, but the obscure relations between metal contents and physico-chemical parameters make it hard to unveil the metal enrichment process. It is probable that the time factor and pressure effects play important roles (Stefansson 1983). This indicates that deep-seated processes at higher pressures may explain the genesis of the metalliferous solutions. Subduction zones are sites where much surface material, including sediments, may slide down to great depths, be dewatered and form hydrothermal solutions (Sillitoe 1972; Burnham 1979; Boström 1981). This could explain the plate-tectonically controlled distribution patterns suggested by the data in Table 2 and Fig. 3a-f. We, therefore, suggest the following major grouping of hot springs on the Earth:

Further studies of natural hydrothermal systems are required to solve these problems. Such studies should to a larger extent than hitherto be focused on the metalliferous system and not concentrate their main interest on the systems poor in trace metals (Weissberg et al. 1979).

CONCLUSIONS

Hot-spring systems on Nea Kameni, Santorini, show distinct metalliferous tendencies, and resemble spring systems on the Kuriles, Java and Banu Wuhu - that is, hot-spring systems associated with young, rather primitive calc-alkaline island arcs and subduction zones. Such hot-spring systems fall between the extremes shown by the highly metalliferous systems found at spreading centres on the ocean floor and hot-spring systems in Yellowstone, Iceland and New Zealand, which lack exhalative-sedimentary deposits.

------------------------------------------------

--------------------------------------------------