Oxygen Isotope Composition of Volcanic Rocks From Santorini and Christiani

The bulk of the lavas range between 7 and 9, which on a worldwide scale is somewhat higher than most andesites and dacites (6 to 8). The enrichment in ¹⁸Ο could be due to either assimilation of ¹⁸Ο-rich crustal material or to secondary low-temperature alteration processes. This phenomenon is especially pronounced in the δ ¹⁸Ο values of glassy materials like pumice. During the hydration of the glass, oxygen isotope exchange between water and silicate seems to occur. Besides these ¹⁸Ο enrichments in the whole rock samples 4 feldspar phenocrysts show "normal" values between 6.6 and 8.0.

INTRODUCTION

This study was initiated for the following purposes:

a)    to survey the oxygen isotope composition of Santorini rocks;

b)    since the Hellenic arc is built on a relatively thick continental crust, to see if there is any evidence of ¹⁸Ο-rich crustal contamination in the lavas and tuffs;

c)    to see if secondary alteration processes, i.e., hydration of the volcanic glass, have influenced the primary isotopic composition.

Oxygen isotope studies on volcanic rocks have been performed to decipher the following problems:

At the high liquid temperatures of most volcanic rocks only very small equilibrium fractionations are found between the minerals and the coexisting melt. That means, fractional crystallization alone cannot produce any appreciable δ ¹⁸Ο changes in a suite of volcanic igneous rocks (Anderson et al. 1971).

Since the δ ¹⁸O values of sedimentary rocks (δ ¹⁸O values between + 12 and + 35) are much higher than in igneous rocks derived from the mantle (δ ¹⁸Ο values between + 5.5 to + 6.5), assimilation of significant amounts of ¹⁸Ο-rich sedimentary and metamorphic rocks should be detectable in the δ ¹⁸Ο values of the contaminated igneous rocks. However, we have to keep in mind that, although the ¹⁸Ο/¹⁶Ο ratios in igneous rocks may initially reflect the varying effects of magmatic differentiation or crustal contamination, Taylor (1968), Muehlenbachs & Clayton (1972), Pineau et al. (1976) and others have shown that ¹⁸Ο variations in many volcanic rocks are due to low-temperature alteration and hydration.

Taylor (1968) and others showed that, although many rhyolites, rhyolitic obsidians and ignimbrites have very high δ ¹⁸Ο values (up to + 16 or even higher) none of these enrichments represented a primary magmatic feature. Instead, the high δ ¹⁸Ο values all resulted from an event that involved low-temperature hydrothermal exchange or hydration of the volcanic-glass by meteoric ground waters. This type of exchange enrichment in ¹⁸Ο is universally observed in the vitrophyres of welded ash flow tuffs. Taylor (1968) also showed, however, that even though the glassy groundmasses of these rocks were commonly enriched in ¹⁸Ο, the phenocrysts in such altered volcanic rocks closely preserved their high-temperature, igneous values. Therefore, in any study of such volcanic rocks, in which the major concern is the δ ¹⁸Ο value of the original magma, it is essential that the phenocryst phases be analyzed in addition to, or even in preference to the whole-rock samples. Unfortunately, only very few feldspar-phenocrysts have been analyzed.

GEOLOGICAL BACKGROUND AND SAMPLE DESCRIPTION

The structural relationship of the volcanic activity in the Aegean Sea has been described by Ninkovich & Hays (1972). The petrology of the Santorini volcano has been described in detail by Nicholls (1971). The Pliocene to recent lavas of the Santorini island group belong to four distinct series, each of high-alumina basalt-andesite-dacite type. Of the youngest historic series only the acid members have been extruded as lavas. The more basic members occur only as xenoliths. The bulk of the islands forming the caldera wall is composed of agglomerates, tuffs, ash, pumice and lapilli and a range of lavas from basalt to rhyodacite. The ratio of pyroclastics to lavas in visible deposits is around 2 : 1. Nicholls (1971) argued that each series was generated by the tapping of a differentiating high-alumina basalt in a high-level magma chamber.

Three different sample series have been analyzed:

1. lavas and pyroclastics from the Santorini volcano and Christiani   (Table 1)
2. feldspar phenocrysts from lavas and pumices (Table 3)
3. a pumice profile, which has been chemically analyzed by Puchelt & Schock (1972) (Table 2)

ANALYTICAL TECHNIQUES

Oxygen was extracted from the rocks and minerals by reacting the samples with BrF₅ at temperatures around 600^o C (for more details see Taylor & Epstein 1962). Generally, 2 or more extractions have been made on each sample.

The oxygen was quantitatively converted to CO₂ over a hot carbon rod and then measured in an ATLAS CH4 mass-spectrometer.

The oxygen isotopic data are reported as δ ¹⁸Ο values, which are defined in the usual way:

δ ¹⁸Ο = [ ( ¹⁸Ο/¹⁶Ο (sample)) / ( ¹⁸Ο/¹⁶Ο (standard)) - 1] x 1000 [‰]

All δ ¹⁸Ο values are given relative to the SMOW Standard (Standard Mean Ocean Water). The analytical error is ± 0.2 ‰.

RESULTS

In Table 1, 2 and 3 the δ ¹⁸Ο values of the analyzed samples are given.

Table 1 gives an overview of the oxygen isotope composition of various rock types from Santorini and Christiani. The δ ¹⁸Ο values range from 7.2 to 13.3, the higher δ ¹⁸Ο values above + 10 are certainly due to secondary alteration processes. Besides these exceptions, most samples fall in the δ ¹⁸Ο range between 7 and 9 ‰, which is somewhat higher than it should be if the lavas were all derived from a primary mantle source. However, a few feldspar phenocrysts separated from lavas show lower values between 6.6 and 8.0 (Table 3).

In Table 2 the isotopic composition of volcanic glasses is reported with special emphasis on the isotopic composition of pumices. Puchelt & Schock (1972) have investigated a pumice profile, sampled from the quarry south of Phira near Cape Alai. The δ ¹⁸Ο values of the volcanic glasses are normally higher than those of the volcanic rocks. There exists a relationship between the δ ¹⁸Ο value and the water content in the glassy material (see Figure 1).

DISCUSSION

As has already been mentioned in the introduction most igneous rocks derived from the lower crust or upper mantle have δ ¹⁸Ο values between + 5.5 and + 6.5 ‰. There are, however, a few exceptions. Turi & Taylor (1976) found that the leucite-bearing lavas from the Roman Province, Italy, may have high δ ¹⁸Ο values up to 11.7. They interpreted their results that the Roman magmas have strongly interacted with high ¹⁸Ο continental crust. On the other hand Muehlenbachs et al. (1974) showed that the primary unaltered volcanic rocks of Iceland are anomalously low in ¹⁸Ο content (from 1.8 to 5.7 ‰). The reason for this ¹⁸Ο depletion is not well understood. But besides these exceptions fresh basaltic rocks normally fall in the δ ¹⁸Ο  range between + 5.5 and + 6.5, while the more SiO₂-rich volcanic rocks like andesite and dacite flows have isotopic compositions in the range between 6 and 8, consistently higher because of the higher silica content.

Regarding our results on Santorini rocks, some andesites and dacites fall in this range between 6 and 8, but many analyzed rocks show somewhat higher δ ¹⁸Ο values. This may have two reasons:

1. Since Santorini is built on a relatively thick continental crust, it might indicate some contamination with ¹⁸O-rich crust. An even more sensitive indicator for such a contamination process is the ⁸⁷Sr/⁸⁶Sr ratio. Pe & Gledhill (1975) found a range of ⁸⁷Sr/⁸⁶Sr ratios for Santorini rocks between 0.7048 - 0.7060. This is somewhat higher than that reported for most other island arcs; however, the range is very similar to that for New Zealand (Ewart & Stipp 1968) and for Chile (McNutt et al. 1975). Pe & Gledhill (1975) concluded that such higher ⁸⁷Sr/⁸⁶Sr ratios are characteistic for island arcs built on continental crust. The oxygen isotope data fit with this idea; however, they could be equally interpreted in a second way.
2. As already mentioned, the ¹⁸Ο/¹⁶Ο ratio is also a sensitive indicator to weathering processes: low temperature alteration processes produce distinctly heavier δ-values. This phenomenon is especially pronounced in the pumices (see Table 2).

Taylor (1968) has also demonstrated that volcanic glasses with high H₂O contents (>2% by weight) are generally a result of secondary hydration by interaction with ground or surface waters at low temperatures. Since glass at low temperatures is metastable, it easily crystallizes (devitrifies) in the presence of water with the passage of time. The process of hydration involves exchange of oxygen in the silicate glass with large amounts of water. Water seems to be able to move in and out of the glass, presumably along cracks and imperfections present in perlitic glass. Marshall (1961) argued that water acts on glasses by leaching cations and forming hydroxyl groups with silica. Friedman (unpublished data, cited in Friedman 1977) made ~ 1 mm serial sections of a glassy selvage on submarine pillows. The analyses show decrease of water into the body of the pillow.

This indicates that water is diffusing into the glass and that its present water content is higher than it was upon eruption.

Whatever the actual mechanism is, the following figure 1 clearly demonstrates the relationship between water content in the pumices and δ ¹⁸Ο values: the higher the water content the higher the δ ¹⁸Ο values. This phenomenon has been observed several times. For instance, Garlick & Dymond (1970) showed that the δ ¹⁸Ο values of siliceous volcanic glass shards in deep-sea sediments increase with age, from + 9‰ in the Pleistocene to + 20 ‰ in the Eocene.

This type of ¹⁸Ο enrichment does not affect the phenocrysts, however, unless the primary minerals are replaced by new minerals such as kaolinite or montmorillonite (see Table 3).

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

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