CO2, 13C/12C and H2O variability in natural basaltic glasses: a study comparing stepped heating and ftir spectroscopic techniques

Geochimica et Cosmochimica Acta, Vol. 63, No. 11/12, pp. 1805–1813, 1999 Copyright © 1999 Elsevier Science Ltd Printed in the USA. All rights reserved 0016-7037/99 $20.00 ⫹ .00

Pergamon

PII S0016-7037(99)00124-6

CO2, 13C/12C and H2O variability in natural basaltic glasses: A study comparing stepped heating and FTIR spectroscopic techniques COLIN G. MACPHERSON,1,2* DAVID R. HILTON,1 SALLY NEWMAN,3 and DAVID P. MATTEY2 1

Geosciences Research Division, Scripps Institution of Oceanography, UCSD, La Jolla, CA 92093-0244, USA 2 Department of Geology, Royal Holloway University of London, Egham, Surrey, TW20 0EX, UK 3 Division of Geological and Planetary Science, California Institute of Technology, Pasadena, CA 91125, USA (Received June 10, 1998; accepted in revised form March 16, 1999)

Abstract—A comparison of two independent techniques was used to assess the homogeneity of CO2 and H2O concentrations in a number of natural basaltic glasses. Variations in carbon concentration and isotopic ratio were determined by comparison of stepped heating data obtained in two different laboratories. Dissolved volatile concentrations were also obtained by stepped heating and Fourier Transform Infrared (FTIR) spectroscopy. Replicate stepped heating analyses of a mid-ocean ridge basaltic glass show that the concentration and 13C/12C of bulk magmatic and dissolved CO2 vary by less than ⫾10% and ⫾0.5‰, respectively. A similar degree of correlation is observed for replicate stepped heating analyses of Mariana Trough glasses conducted in two different laboratories. Dissolved CO2 concentrations determined by stepped heating also correlate well with concentrations measured by FTIR spectroscopy. The correspondence of results obtained in these experiments provide an upper limit to the degree of natural variation in concentrations and isotopic ratios of these volatiles in basaltic glasses and suggest that intrinsic, magmatic carbon has a relatively homogeneous distribution in these glasses. Water concentrations determined through extraction by heating and FTIR also show excellent agreement. Copyright © 1999 Elsevier Science Ltd 12 C ratios determined in a number of laboratories on different splits of individual glasses and urged “the need for a consensus” on analytical protocols to determine the extent of natural heterogeneity. Since 1986 the volatile database for natural glasses has expanded considerably but the question of individual sample homogeneity has received relatively little direct attention in the literature (however, see Mattey et al., 1989 and Pineau and Javoy, 1994). Since 1986, two analytical procedures have emerged as the most popular means of determining the CO2 and H2O characteristics of oceanic basalts. Fourier transform infra-red (FTIR) spectroscopy allows measurement of the concentration of CO2 and H2O dissolved in the glass phase (Stolper, 1982; Fine and Stolper, 1985/86). This technique has stimulated significant advances in understanding major volatile concentrations in natural basaltic melts and CO2 and H2O solubility relationships under varying pressure, temperature, hydration and oxygen fugacity conditions (Fine and Stolper, 1985/86; Dixon et al., 1988; Stolper and Holloway, 1988; Dixon et al., 1991; Pan et al., 1991; Pawley et al., 1992; Dixon et al., 1995; Jendrzejewski et al., 1997). In addition, FTIR spectroscopy is a non-destructive tool that allows the glass, and its volatiles, to be preserved for further analytical work. However, it does not detect gas either contained in vesicles or adsorbed onto surfaces (posteruptive contamination), nor does it allow isotopic analyses to be conducted on the volatile species. The second technique is stepped heating analysis that allows volatiles to be physically extracted from surfaces, vesicles and the glass phase of quenched samples. Gas yields (concentrations) can be determined manometrically with the additional advantage that isotopic analyses can be performed on each of the released components (Pineau and Javoy, 1983; Des Marais and Moore, 1984; Mattey et al., 1984; Exley at al., 1986; Sakai et al., 1984; Mattey et al., 1989; Blank et al., 1993; Macpherson

1. INTRODUCTION

Submarine volcanic glasses provide a means to directly sample volatiles that are present during the evolution and eruption of magmatic systems. A discrete gas phase may be trapped in vesicles while the confining pressure of the overlying water column results in some fraction of the volatiles remaining dissolved in the melt that is, consequently, preserved in the glass phase during quenching. Since carbon dioxide and water comprise the major part of typical magmatic gases in natural basaltic systems, analyses of these volatile phases in vesicles and associated glass matrix should provide information on the nature and character of carbon and hydrogen in the mantle source region supplying the melts (Pineau and Javoy, 1983; Des Marais and Moore, 1984; Kyser and O’Neil, 1984; Mattey et al., 1984; Exley at al., 1986; Fine and Stolper, 1985/86; Sakai et al., 1984; Dixon et al., 1988; Mattey et al., 1989; Dixon et al., 1991; Blank et al., 1993; Javoy and Pineau, 1991; Macpherson and Mattey, 1994; Pineau and Javoy, 1994; Jendrzejewski et al., 1997). A long-standing problem with this approach has been uncertainty in the extent of concentration and isotopic homogeneity for various volatile species— both within and between individual fragments or chips of submarine glass. For example, variations in volatile distribution may result from differing quench rates throughout the mass of any single erupted body. Alternatively, interaction between basaltic melts and hydrated and/or carbonated crust may introduce non mantle-derived volatiles into melts in a non-uniform manner thereby imposing a natural heterogeneity onto samples. With this in mind, Des Marais (1986) highlighted differences in CO2 concentrations and 13C/ *Author to whom correspondence should be addressed (c.macpherson @gl.rhbnc.ac.uk). 1805

1806

C. Macpherson et al.

and Mattey, 1994; Javoy and Pineau, 1994). Both carbon and hydrogen can be deposited on, or absorbed into, the outer surface of volcanic glass after eruption and most steppedheating extraction procedures regard gas evolved during low temperature heating (400°C for H2O and up to 600°C for CO2) as non-magmatic in origin (Kyser and O’Neil, 1984; Des Marais, 1986; Mattey et al., 1989). Various protocols have been employed to extract the intrinsic magmatic fraction ranging from bulk extraction to stepped heating in increments of as little as 50°C. The possibility that the range of protocols employed by different laboratories could in itself be a source of some of the “heterogeneity” observed by Des Marais (1986) has yet to be fully addressed. In this contribution we confront this possibility by comparing major volatile analyses of individual samples conducted by the stepped heating technique and by FTIR spectroscopy. First, we address the question of reproducibility (precision) in the stepped heating technique by replicate analyses of an East Pacific Rise glass. This glass has been analysed for carbon concentrations and isotopic ratios in other laboratories allowing us to also assess the accuracy of stepped heating. Subsequently, comparison of stepped heating data from two different laboratories allows us to investigate the distribution of contaminant, vesicle and dissolved volatiles in individual splits of natural glasses from the Mariana Trough. FTIR analyses of dissolved species are available for petrologically identical glasses from the Mariana Trough dredge hauls yielding further information on natural volatile homogeneity and analytical accuracy. The complete carbon isotopic dataset for the Mariana Trough suite with scientific interpretation and petrogenetic implications will be presented elsewhere (Hilton et al., in prep.). To distinguish between the different volatile species present we will use the following terms throughout this paper. Dissolved CO2 is the gas that was preserved in the glass during quenching while vesicle CO2 is gas recovered from vesicles. Note, however, that gas released from vesicles may have re-equilibrated after eruption (Mathez and Delaney, 1981; Watanabe et al., 1983; Macpherson and Mattey, 1994; Pineau and Javoy, 1994). Some studies used for comparison (e.g. Des Marais, 1986) recovered both vesicle and dissolved CO2 together, therefore, we define magmatic CO2 as the sum of the vesicle and dissolved components. Similarly for water, since the FTIR results represent dissolved hydrous species we refer to these as dissolved H2O, while the results for water extraction by melting, which include both dissolved and vesicle water, are termed magmatic H2O. 2. ANALYTICAL TECHNIQUES 2.1. Carbon Analysis by Stepped Heating As part of ongoing efforts to integrate studies of the rare gas and the major volatile systematics of oceanic basalts, we have constructed a new extraction system at the Scripps Institution of Oceanography (SIO) designed for step-wise combustion and/or pyrolysis of basaltic glass chips (Fig. 1). The all-glass system, based on that at Royal Holloway University of London (RHUL), is designed to allow simultaneous collection of CO2 and H2O extracted from volcanic glass. The SIO extraction system is constructed from Pyrex威 and fused quartz glass, and comprises two distinct parts: an “extraction” section and a “cleanup/measurement” section. Fresh chips of volcanic glass, 1–3 mm in diameter and free of large vesicles, phenocrysts and surface alteration (Mattey et al., 1989; Pineau and Javoy, 1994), were selected using a binocular microscope then washed in a 50:50 acetone : methanol

mixture and stored in a clean environment. Immediately prior to analysis glass chips weighing between 100 –140 mg were pre-cleaned in dichloromethane, weighed (to better than ⫾0.05 mg), recleaned in dichloromethane and loaded into the extraction system using the minimum exposure technique of Mattey et al. (1989). The fused quartz glass sample finger and the system were evacuated to a pressure of ⱕ10⫺6 torr for at least 1 h prior to commencing the extraction. The sample is heated using an external resistance furnace and evolved gases are collected in a liquid nitrogen-cooled U-trap close to the sample (N2 in Fig. 1). Combustion analyses are performed in a high partial pressure of oxygen gas. The O2, which is generated from a copper oxide furnace at 900°C and stored on 5A molecular sieve, promotes the oxidation of reduced gases released from the sample. After all combustion steps are completed, the O2 can be resorbed by lowering the temperature of the CuO furnace to 600°C. During pyrolysis steps exsolved gases are exposed to a second copper oxide furnace held at 600°C to promote oxidation of reduced species (e.g. CO, H2, H2S, N2O). After each 30-minute step the gases collected in the U-trap are transferred to a liquid nitrogen-cooled, all-glass, variable temperature trap in the “clean-up” part of the system. During transfer to the variable temperature trap gases collected by pyrolysis remain exposed to the 600°C copper oxide furnace. Following the transfer procedure all gases should be in either a condensed and oxidised (e.g. CO2, SO2 and H2O) or non-condensable form (e.g. rare gases and nitrogen). After discarding the non-condensable fraction, the CO2 can be isolated by heating the variable temperature trap to ⫺135°C. The CO2 yields from each step are measured using a Baratron威 capacitance manometer in a calibrated volume of the line. The CO2 is then trapped in quartz glass breakseals for later isotopic analysis (at RHUL the CO2 is passed directly into the source of a mass spectrometer). When the CO2 yield from a single step is insufficient for isotopic analysis it is combined with CO2 obtained during the subsequent step(s). The variable temperature trap is then heated to 120°C and its remaining contents are transferred to a separate water collection finger cooled by liquid nitrogen. A methanol slush is used to retain the water (from all steps) for later abundance and isotopic analysis. Procedural blanks for CO2 determined on fused quartz glass are approximately 0.1 mcm3STP for the 700°C to 900°C steps and 0.05 mcm3STP at higher temperatures, with blank CO2 having a carbon isotope ratio of c. ⫺26‰. All data discussed below have been corrected for the presence of an analytical blank. Carbon isotope ratios of CO2 extracted by stepped heating were measured using VG PRISM mass spectrometers (both at SIO and RHUL) and are reported in the usual ␦ notation relative to V-PDB. An essential part of the stepped heating procedure is combustion in oxygen at 400°C and 600°C which was conducted to remove secondary surface deposits introduced to the sample after eruption (Des Marais, 1986; Exley et al., 1987; Mattey et al., 1989). As with many previous studies, most 400°C and 600°C extractions conducted during this work produced variable CO2 yields possessing, where measured, low carbon isotope ratios (⬍ ⫺25‰), consistent with a secondary origin. However, in two cases where unusually large releases occurred at 600°C higher ␦13C values were obtained (Appendix 1), probably indicating slightly early decrepitation of vesicles (Pineau and Javoy, 1994). The abundance of the secondary carbon is dependent upon the extent of contamination experienced by each chip as a result of seafloor environment, grain size, grain geometry, laboratory storage environment and pre-cleaning protocols (Des Marais, 1986; Exley et al., 1986; Mattey et al., 1989). Therefore, carbon results for 400°C and 600°C steps are not considered further, with the two exceptions noted above. Stepped heating to higher temperatures allows sequential extraction of carbon originating in bubbles followed by CO2 dissolved in the glass (Mattey et al., 1989; Macpherson and Mattey, 1994; Pineau and Javoy, 1994; Jendrzejewski et al., 1997). In this study pyrolysis steps were conducted at 100°C intervals from 700°C to 1200°C; a procedure that has previously been demonstrated to resolve gas extracted from vesicles and dissolved CO2 (Exley et al., 1986; Macpherson and Mattey, 1994; see below). Heating beyond 1200°C, either by combustion or pyrolysis, did not produce CO2 yields significantly in excess of the procedural blank suggesting that glasses used in this study contain negligible quantities of elemental carbon (Mattey et al., 1989).

CO2,

13

C/12C and H2O variability in natural basaltic glasses

1807

Fig. 1. Schematic of the Scripps Institution of Oceanography (SIO) major volatile extraction system. The sample finger, CuO furnaces and molecular sieve make up the ‘‘extraction’’ part of the line while the variable temperature trap, manometer and collection fingers comprise the ‘‘clean-up’’ part of the system.

2.2. Reproducibility of Carbon Analysis by Stepped Heating To test the reproducibility of CO2 extraction from a natural glass by stepped heating we conducted replicate analyses of ALV981-R23, a basaltic glass from the East Pacific Rise. Assuming that magmatic carbon is released only during the 700°C step and above, our analyses suggest that this glass contains 426.9 ppm (⫾25 ppm 1␴, n ⫽ 5) of magmatic CO2 with a mean ␦13C value of ⫺6.61‰ (n ⫽ 2). This compares well with a fusion analysis of the same sample by Des Marais (1986) that, following combustions at 440°C and 635°C, recovered 436 ppm CO2 with a ␦13C value of ⫺5.6‰. By employing single-step fusion, Pineau and Javoy (1983) reported 645 ppm CO2 for this glass. However, their procedure did not include any heating steps between 120°C and 600°C so the higher yield may reflect the presence of some secondary carbon from grain surfaces. This would be consistent with the lower ␦13C value (⫺8.6‰) that was obtained. Previous studies by Exley et al. (1986), Mattey et al. (1989) and Macpherson and Mattey (1994) have demonstrated that during high resolution stepped heating (i.e. in steps of 100°C or less) two distinct carbon components are released— one at intermediate temperatures (usually below 1000°C) and the second at high temperatures (1000°C and above). These components are hosted in vesicles and dissolved in the glass, respectively. Using small heating steps allows these two components to be extracted in sequence and, hence, separated from each other. An alternative approach is to assume that there is a certain temperature on either side of which the two components are always released. Pineau and Javoy (1994) and Jendrzejewski et al. (1997) favoured the latter approach and employed a step at 900°C to collect vesicle-hosted gases followed by a fusion step to collect dissolved

gases. Since dissolved gas must diffuse out of the chips as they soften the temperature at which dissolved CO2 starts to exsolve during any one experiment will be a function of the chip size and geometry in that experiment. Furthermore, the chemistry of the glass, particularly its water content, will influence the diffusion of carbon (Dixon et al., 1997; F. Pineau, pers. comm.). Release of the dissolved CO2 is usually quite distinctive in terms of both volume of gas released and its carbon isotope ratio (Mattey et al., 1989; Macpherson and Mattey, 1994). For example, in ALV981-R23 (this work) a significant increase in the carbon release profile occurs at 1100°C in most extractions and, in these cases, this is taken as the start of the release of the dissolved CO2 (Appendix 1). However, for some glasses e.g. several Mariana Trough glasses (Appendix 1), the release of dissolved CO2 commences at 1000°C. Therefore, we prefer the high resolution stepped heating protocol because it allows us to identify uniquely the onset of release of the dissolved CO2 during each experiment. During all experiments on ALV981-R23 the major fraction of dissolved CO2 was released at 1200°C. The concentration of dissolved CO2 in ALV981-R23 was determined by summing the yield of the prominent 1000°C or 1100°C release and all subsequent steps. This gives an average dissolved CO2 concentration of 405.6 ppm (⫾8 ppm (1␴), n ⫽ 5) which is somewhat higher than an FTIR spectroscopy analysis of 347 ppm (⫾15%; Fine and Stolper, 1985/86). However, procedures employed by Fine and Stolper (1985/86) may have underestimated background subtraction (J. Dixon, pers. comm.). The reproducibility for measuring dissolved CO2 concentrations by stepped heating in this work (⫾8 ppm or 2%, 1␴) compares well with reproducibility in other laboratories employing similar techniques. For example, Pineau and Javoy (1994) repeated analysis of dissolved CO2

1808

C. Macpherson et al.

concentrations in different batches of glass from dredge 2␲D43 on the North Atlantic Ridge. They obtained dissolved CO2 concentrations of 294 ppm ⫾ 15 (1␴, n ⫽ 4), 294 ppm ⫾ 26 (1␴, n ⫽ 2) and 235 ppm ⫾ 15 (1␴, n ⫽ 2) for three different sample aliquots from dredge 2␲D43. Jendrzejewski et al. (1997) obtained 244 ppm ⫾ 9 (1␴, n ⫽ 2) for the dissolved CO2 concentration in a different EPR glass (ALV892-1a). In addition to reproducibility in CO2 contents, our results also compare favourably in reproducibility of isotopic composition. The mean weighted ␦13C value of the dissolved CO2 in our EPR glass ALV981-R23 is ⫺5.70‰ ⫾ 0.41‰ (1␴, n ⫽ 4). The precision on this measurement is similar to that determined by Mattey et al. (1989) for replicate analyses of an Indian Ocean glass and the 3 batches of glass (mentioned above) from 2␲D43 analysed by Pineau and Javoy (1994) that were reproducible to ⫾0.49‰ (1␴, n ⫽ 4), ⫾0.35‰ (1␴, n ⫽ 2), ⫾0.21‰ (1␴, n ⫽ 2), respectively. The data for the dissolved and magmatic carbon in ALV981-R23 suggest that the ␦13C value of carbon extracted from vesicles is lower than that of the dissolved CO2. This is the reverse of the relationship expected from isotopic equilibrium studies (⌬CO2-carbonate is positive at 1200°C; Mattey et al., 1990; Mattey, 1991; Scheele and Hoefs, 1992). Similar observations have been made during stepped heating analyses of several natural basaltic glasses (Mattey et al., 1984; Pineau and Javoy, 1994) and can been ascribed to post-eruptive re-equilibration of vesicle gases (Mathez and Delaney, 1981; Watanabe et al., 1983; Macpherson and Mattey, 1994; Pineau and Javoy, 1994). 2.3. Water Analysis by Stepped Heating Magmatic water contents from the Mariana glasses were determined using a two-step heating procedure at RHUL. Combustion at 400°C was employed to removed water absorbed onto surfaces (Kyser and O’Neil, 1984). The sample was then pyrolised for three hours at 1200°C. To promote oxidation of any reduced species, exsolved gases were exposed to a copper oxide furnace held at 600°C during and after the extraction (Kyser and O’Neil, 1984). All condensable products were collected at boiling nitrogen temperatures in a glass capillary. A methanol slush (⫺98°C) was used to retain water which was then sealed into the capillary. The water concentration was measured, to ⫾0.20 wt.% (2␴), by reducing the water over Zn and comparing H⫹ 2 peak-heights in a VG Prism mass spectrometer to those of known volumes of hydrogen generated from standard waters. 2.4. CO2 and H2O Concentration Analysis by FTIR

3. RESULTS

The reproducibility of carbon isotope and CO2 concentration results for ALV981-R23 (previous section), both within and between laboratories demonstrate the suitability of stepped heating analysis for determining the concentration and ␦13C value of CO2 in typical submarine glasses. Consequently, we employed the stepped heating technique to measure the concentrations of magmatic CO2 (and H2O) in a suite of glasses from the southern Mariana Trough. Carbon stable isotope ratios of the CO2 were also determined. The glasses are mainly quartz- and olivine-tholeiites that display a variety of petrogenetic affinities ranging from mid-ocean ridge basalt (MORB) to back-arc basin basalt (BABB) (Gribble et al., 1996). Based on FTIR analysis of petrologically-similar glasses from the same dredge hauls (Gribble et al., 1996), a wide range in CO2 and H2O concentrations was expected. 3.1. Magmatic Volatile Concentrations

Dissolved H2O and CO2 contents were determined by FTIR spectroscopy using the techniques described by Fine and Stolper (1985 and 1985/86) and Dixon et al. (1988), with several modifications. Doubly polished plates ⬃50 –320 ␮m thick were prepared for each sample. Spectra were taken with a Nicolet IR-Plan microscope attachment that had been modified to fit the Nicolet 60SX spectrometer. The doubleaperturing system of the microscope was set to 50 ⫻ 100 to 115 ⫻ 115 ␮m rectangles, avoiding vesicles and crystals. Each spectrum consisted of 2048 scans using a KBr beam splitter and liquid nitrogen cooled HgCdTe detector. Absorbances from 3– 4 spectra were averaged for H2O and CO2. The concentration of CO2 by FTIR is calculated using the carbonate doublet at 1515 and 1435 cm⫺1. As expected for basaltic compositions, no dissolved molecular CO2 band at 2348 cm⫺1 was observed. The absorbance for the carbonate doublet at 1515 and 1435 cm⫺1 was determined by the least squares best fit of a combination of (i) a devolatilized background, (ii) a straight line, (iii) a pure H2O(molecular) band at 1630 cm⫺1, and (iv) a pure carbonate doublet at 1435 and 1515 cm⫺1 (Newman et al., in prep.). A molar absorptivity of 375 ⫾ 20 l/mol-cm (Fine and Stolper, 1985/86), similar to that of Jendrzejewski et al. (1997; 397 ⫾ 7 l/mol-cm), and an assumed density of 2.8 g/cm3 were used to calculate the concentration from Beer’s Law: C ⫽ (A ⴱ MW)/(␳ ⴱ ␧ ⴱ d)

O-H stretching band at ⬃3550 cm⫺1 for concentrations less than ⬃1 wt.%, using a molar absorptivity of 63 ⫾ 3 l/mol-cm (P. Dobson, S. Newman, S. Epstein, and E. Stolper, unpublished data), which is similar to that of Pandya et al. (1992; 61 l/mol-cm) and Yamashita et al. (1997; 64 l/mol-cm), but lower than that of Jendrzejewski et al. (1996; 78 l/mol-cm). For concentrations greater than ⬃1 wt.%, the sum of the water contents determined by Beer’s Law (equation 1) from the near infrared bands at 5200 cm⫺1 (H2O molecules; Scholze, 1960; Bartholomew, 1980) and 4520 cm⫺1 (OH groups; Scholze, 1960; Stolper, 1982) was used, along with molar absorptivities of 0.67 ⫾ 0.03 and 0.62 ⫾ 0.07 l/mol-cm, respectively (Dixon et al., 1995). For the 11 samples used in this comparative study, the average error for H2O concentration is 5% relative. The corresponding error for CO2 concentration is 24% relative, due to the uncertainties in the background correction. These errors are 1 standard deviation of the mean of the 3– 4 measurements, taken on different spots of the same polished chip. The magnitude of the error (precision) on CO2 measurements is a result of compositional-dependent changes in baseline curvature in the part of the spectrum where the measurement is made. However, the correspondence of FTIR results with stepped heating data (see below) suggests that the FTIR data have a high degree of accuracy.

(1)

where C ⫽ concentration in weight fraction, A ⫽ absorbance, MW ⫽ molecular weight in g/␮mole, ␳ ⫽ density in g/l, ⑀⫽ molar absorptivity in l/mol-cm, and d ⫽ thickness in cm. H2O concentrations were determined from the absorbance of the

Six southern Mariana Trough glasses were analysed for CO2 by stepped heating at both SIO and RHUL (Appendix 1). Calculated CO2 concentrations and ␦13C values of dissolved and magmatic carbon are reported in Table 1. Significant differences in CO2 yield were obtained in the different laboratories for most 700°C steps as well as for two 800°C steps (Fig. 2a). This may be due to a tailing effect from the surface contamination or differences in the temperature of initial vesicle decrepitation resulting from physical properties of different grains. For the remaining 800°C and higher temperature steps there is a good correspondence between the amount of gas released during each step in the two laboratories (Fig. 2a). Some scatter around the 1:1 line probably reflects differing vesicle distributions and decrepitation temperatures, and variations in the release of dissolved CO2 resulting from diffusion out of different shaped/sized grains. The latter effect was observed during a replicate analysis of 79DS2-1 at SIO (Appendix 1). In both analyses the dissolved component started to exsolve at 1000°C and continued over the 1100°C and 1200°C steps. In the first run, these steps released 20%, 51% and 29% of the dissolved CO2, respectively. In the second extraction, the same temperature steps released 21%, 27% and 52% of the

CO2,

13

C/12C and H2O variability in natural basaltic glasses

1809

Table 1. Magmatic carbon dioxide and water concentrations in Mariana Trough basaltic glasses and Technique laboratory volatile fraction

Stepped Heating SIO

Sample

Dissolved CO2 (ppm)

Mariana Troughb 18-1 74-2 74-3 79-2 80-5d 84-2 84-3 86-4 86-6 88-1(⫺1) 88-1(⫺4) 71-1 71-3e 73-1 73-2 75-1

28.1 29.0 104.7 157.5c 134.4 231.0 231.0 96.5 146.1 204.5 148.1 136.8 142.5 117.4 117.4 7.7

EPR ALV981-R23

405.6c

13

C/12C ratio of CO2. FTIR CIT

RHUL

␦13C (‰)

⫺7.41 ⫺8.22c

Magmatic CO2 (ppm)

225.6 262.2c 327.0

␦13C (‰)

⫺9.03 ⫺10.20c

Dissolved CO2 (ppm)

142.7 149.1 151.3

␦13C (‰)

⫺8.08 ⫺8.60 ⫺7.43

Magmatic CO2 (ppm)

223.5 306.8 350.2

␦13C (‰)

⫺9.95 ⫺9.35 ⫺8.72

Magmatic Dissolved Dissolveda H2O CO2 H2O (wt.%) (ppm) (wt.%)

1.93 1.15 1.14 0.40 0.40 0.59 0.59

⫺8.94 ⫺7.26

⫺5.70c

264.3 211.7 273.6

426.9c

⫺9.77 ⫺7.61

⫺6.61c

155.7 123.1 166.5

⫺9.60 ⫺6.69 ⫺9.04

300.8 237.3 255.4

⫺9.96 ⫺7.66 ⫺9.61

1.35 1.59 1.07

30 28 95 152 104 215

2.03 2.08 1.18 0.97 2.78 0.21

126 126 199 199 111

0.58

142 13

1.46

1.30 1.34

347f

a: Gribble et al. (1996). b: The first number represents the dredge site. On-board classification of the glasses recovered from each dredge haul into petrographic types was based on features visible in hand sample such as differences in phenocryst content and assemblage. The second number gives the within haul petrographic type. See Gribble et al. (1996 and references therein) for further details of the samples and their origin. Each type was then sub-divided into splits for dissemination to the community. All analyses at Scripps Institution of Oceanography (SIO) and Royal Holloway University of London (RHUL) were performed on fragments from a single split. In most cases the FTIR analyses, at California Institute of Technology (CIT), were performed on fragments from a different split of the same petrographic type of an individual haul. In four cases where the same petrographic type was not available we compare analyses conducted on petrographically similar lavas. These four cases are illustrated in the table by italics. c: mean of replicate determinations (See Appendix 1). d: no carbon isotopic data from SIO due to isotopic fractionation of a step during mass spectrometry. e: no carbon isotopic data from SIO due to loss of a step during extraction. f: Fine and Stolper (1986).

dissolved component, respectively. While the percentage of the gas released during each step was different in the two runs, the total amount of gas released over the interval was very similar (162 ppm and 153 ppm). The consequences for the carbon isotopes are discussed below. Since the major fraction of magmatic carbon is released at 800°C and above, Fig. 2a suggests that the distribution of intrinsic CO2 in these glasses is relatively homogeneous at the scales sampled by stepped heating. Dissolved CO2 concentrations of southern Mariana Trough glasses calculated from SIO stepped heating data can be compared with FTIR data of petrologically identical glasses in 9 cases and petrologically similar splits in a further 4 cases (Table 1). For most of the southern Mariana Trough glasses, the dissolved CO2 was released during the 1100°C and 1200°C steps. Results from the two different techniques correspond well (Fig. 2b). A similar correspondence has been noted for basalt and ferrobasalt glasses from the Lau Basin (Fig. 2b). Isolated differences between results from the two techniques for the Lau Basin glasses are discussed by Macpherson and Mattey (1994). Dissolved CO2 concentrations of Mariana Trough, Lau Basin and East Pacific Rise glasses determined by stepped heating and FTIR agree to ⫾20 ppm or better at the 1␴ level. Discrepancies between magmatic CO2 concentrations

determined at SIO and RHUL are of a similar magnitude to those of dissolved CO2 (Fig. 2b). Figure 3 compares magmatic water concentrations measured by step heating at RHUL with dissolved H2O concentrations obtained by FTIR. The correlation (r2 ⫽ 0.95) is indistinguishable from a 1:1 correspondence suggesting that the two techniques show an excellent agreement in absolute results across the scales sampled (4 mm to 150 ␮m). One reviewer suggested that employing a 400°C initial step during extraction by heating may result in the loss of some magmatic water—particularly from water-rich glasses. If this were the case, the correspondence in Fig. 3 suggests that subsequent to each 400°C heating step the amount of water retained in each glass was equal to its intrinsic dissolved H2O content. Since the only other possible location for magmatic water to reside is in vesicles, the amount of water lost at 400°C during any one experiment would have to precisely equal the vesicle H2O content of that glass. This would require one of two situations to occur. The first is that vesicles decrepitate during the 400°C step. However, repeated work on carbon suggests that this event occurs at higher temperatures (Exley et al., 1986; Mattey et al., 1989; Macpherson and Mattey, 1994; Pineau and Javoy, 1994; Jendrzejewski et al., 1997; this work). Alternatively, all vesicle water would

1810

C. Macpherson et al.

Fig. 2. (a) Comparison of yields for individual steps of stepped CO2 extraction from 6 southern Mariana Trough glasses at Scripps Institution of Oceanography (SIO) and Royal Holloway University of London (RHUL). Yields are in mcm3STP/g. Numbers next to symbols indicate the temperature of that step. (b) Comparison of dissolved and magmatic CO2 concentrations in basaltic glasses measured using stepped heating (SH) at SIO and RHUL, and by Fourier transform infra-red (FTIR) spectroscopy and bulk extraction. Dissolved CO2 concentration compare SH with FTIR or SH with SH analyses for glasses from the Mariana Tough (this work; Gribble et al., 1996) and the Lau Basin (Farley and Newman, 1994; Macpherson and Mattey, 1994) and for ALV981-R23 (this work; Fine and Stolper, 1985/86). Magmatic CO2 concentrations were determined by stepped heating or bulk extraction for Mariana Trough glasses (this work) and ALV981-R23 (this work; Des Marais, 1986). See text for analytical details and uncertainty.

have to diffuse into the glass with an equal volume of (dissolved) water lost to vacuum by diffusion out of the glass. If dissolved water diffused out of the glasses under vacuum they should be H2O-undersaturated with respect to their eruption depths, but all the Mariana glasses are approximately volatilesaturated with respect to their depth of collection (Gribble et al., 1996). We prefer the simpler explanation of Fig. 3 that no magmatic water was lost during the 400°C step. The corollary is that since FTIR detects only dissolved species and bulk extraction should sample dissolved and vesicle water, the vesicles of southern Mariana Trough glasses contain a negligible fraction of their total water.

3.2.

13

C/12C of Dissolved and Magmatic CO2

Carbon isotope ratios of CO2 released during corresponding heating steps at SIO and RHUL are compared in Fig. 4a. Where a significant difference in CO2 yield occurred for the 700°C or 800°C steps, the ␦13C value of the larger release was less negative than ⫺9‰ (Appendix 1). This suggests that higher yields at these temperatures result from release of vesicle gas slightly earlier in that experiment (c.f. Pineau and Javoy, 1994). Besides sample 80DS5-1, where the carbon isotope ratio of CO2 from the 1000°C step is thought to have been fractionated during freezing into the inlet of the mass spectrometer (Appendix 1), all remaining steps show a relatively good degree of correspondence. As noted for the comparison of CO2 yields from the two stepped heating laboratories, the scatter about the 1:1 line is probably a result of differing vesicle distribution and diffusion effects related to grain geometry. The isotopic effects of diffusion cause scatter in Fig. 4a but are eliminated when ␦13C values of dissolved CO2 are calculated. This is confirmed by the correspondence between ␦13C values for the dissolved CO2 calculated from data from the two laboratories, which agree to better than ⫾0.5‰ (Fig. 4b). Similarly, the calculated isotopic ratio of dissolved CO2 in replicate analyses of 79DS2-1 at SIO differs by only 0.5‰ despite considerable differences in CO2 release profiles (see above). The isotope ratios of magmatic CO2 determinations in the two laboratories also show a high degree of correlation (Fig. 4b). 4. DISCUSSION

Fig. 3. Comparison of magmatic H2O concentrations determined by bulk extraction (this work) and dissolved H2O concentrations obtained by FTIR spectrometry (Gribble et al., 1996). See text for analytical details and uncertainty.

The results presented above suggest that individual southern Mariana Trough glasses and the single EPR glass display relatively little internal variation in the concentrations and 13 C/12C ratios of magmatic and dissolved carbon. Macpherson

CO2,

13

C/12C and H2O variability in natural basaltic glasses

1811

Fig. 4. (a) Comparison of ␦13C values determined on CO2 released by individual steps of stepped heating extractions of Mariana Trough glasses at Scripps Institution of Oceanography (SIO) and Royal Holloway University of London (RHUL). Numbers next to symbols indicate the temperature of that step. The carbon isotope ratio of the 1100°C step of 80DS5-1 measured at SIO was fractionated during analysis. (b) Comparison between ␦13C values of dissolved and magmatic CO2 in basaltic glass measured at SIO and data obtained in other laboratories. ␦13C of magmatic CO2 from ALV981-R23 from Des Marais (1986) and all other data obtained in this study.

and Mattey (1994) reached a similar conclusion for dissolved CO2 concentrations in glasses from the Lau Basin. Dissolved CO2 concentrations appear to vary by less than 10%, while carbon isotope ratios vary by less than ⫾0.5‰. The natural variation of dissolved CO2 concentrations is slightly greater than found in experimental charges equilibrated at the pressures of submarine volcanism (Jendrzejewski et al., 1997). A comparison of experimental data (Jendrzejewski et al., 1997) with observations of natural glasses (Mattey et al., 1989; Pineau et al., 1994; this work) suggests that either CO2 in natural glasses is slightly more variable in concentration than synthetic magmatic systems, or the confidence in stepped heating and FTIR techniques can be reduced by further refinements in analytical protocols. The concentration of water within individual glasses is also relatively constant and, as already noted, this implies that the water budget of the Mariana glasses erupted onto the sea floor is dominated by a dissolved component. Since magmatic carbon is present only as dissolved or vesicle CO2, the data also suggests limited variation in the concentration and carbon isotopic ratio of CO2 within vesicles of individual glasses investigated here. We note that changing the average grain size and total sample weight can serously influence measured characteristics of CO2 obtained from vesicles during stepped heating (Mattey et al., 1989). Furthermore, the speciation of carbon in vesicles may have re-equilibrated after eruption (Mathez and Delaney, 1981; Watanabe et al., 1983). Therefore, the abundance and ␦13C of CO2 extracted from vesicles may not be representative of its state during eruption (Macpherson and Mattey, 1994; Pineau and Javoy, 1994). Basaltic glasses from the Mariana Trough and Lau Basin are saturated to undersaturated with respect to the solubility of pure CO2 at their depths of eruption (Macpherson and Mattey, 1994; Gribble et al., 1996). This contrasts with the majority of spreading centre glasses which are CO2-oversaturated (Dixon et al., 1988), and suggests that these backarc basin melts were afforded a greater opportunity to degas during eruption. This may result from either slower eruption rates (Dixon et al., 1988) or the high water concentration which would force exsolution of,

and lead to apparent undersaturation in, CO2 and other volatiles (Macpherson and Mattey, 1994; Dixon et al., 1995; Macpherson et al., 1998). Higher water concentrations may also increase the nucleation rate and growth of bubbles (Dixon et al., 1997). At present, there are no recognised standards for volatile concentrations or isotope ratios in volcanic glasses. The data presented in this work suggest that the development of such standards should be possible. Moderately degassed glasses from marginal basins may prove useful in this respect since they contain easily detectable, uniform quantities of both CO2 and H2O. 5. CONCLUSIONS

Major volatile concentrations in a variety of natural basaltic glasses show relatively little variation. Dissolved CO2 and H2O concentrations, determined by stepped heating in two laboratories and by FTIR, vary by ⬍⫾10%. In addition to extracting dissolved CO2 from volcanic glasses, stepped heating in relatively small increments (100°C) can be employed to quantitatively recover contaminant and vesicle carbon for abundance and isotopic analysis. Concentrations of magmatic and, by inference, vesicle CO2 in our Mariana Trough and East Pacific Rise glasses are also relatively constant. The isotopic ratios of the dissolved carbon vary by ⬍0.5‰. Water concentrations determined for dissolved species, by FTIR, and magmatic H2O, by stepped heating, also display excellent agreement. Such glasses may prove suitable for development as standards for CO2 and H2O concentration and 13C/12C. Acknowledgments—We thank John Pace for construction of the La Jolla stepped heating system, and Professor Mark Thiemens and Joel Savarino (UCSD Chemistry) for access to, and assistance with, cross calibration of our manometric measurement system. Professor Kurt Marti and Dr. Thomas Graf (UCSD Chemistry) gave advice on the construction of a high temperature resistance furnace. Professor Martin Wahlen and Dr. Bruce Deck (SIO) are thanked for allowing us to use their quadrupole and gas source mass spectrometers. Samples were provided by Professors Peter Stoffers (Kiel), as part of an on-going study of Mariana Trough basalts, and Harmon Craig (SIO). This

1812

C. Macpherson et al.

contribution benefited from thoughtful reviews by Drs. Jackie Dixon and Franc¸oise Pineau. Work in La Jolla was supported by SIO “startup” funds to DRH. The stable isotope laboratory at RHUL is partfunded by the University of London as an Intercollegiate Analytical Facility. REFERERENCES Bartholomew, R. F., Butler, B. F., Hoover, H. L. and Wo, C. K. (1980) Infrared spectra of a water-containing glass. J. Am. Ceramic Soc. 63, 481– 485. Blank, J. G., Delaney, J. R., and Des Marais, D. J. (1993) The concentration and isotopic composition of carbon in basaltic glasses from the Juan de Fuca Ridge, Pacific Ocean. Geochim. Cosmochim. Acta 57, 875– 887. Des Marais, D. (1986) Carbon abundance measurement in oceanic basalts: The need for a consensus. Earth Planet Sci. Lett. 79, 21–26. Des Marais, D. J. and Moore, J. G. (1984) Carbon and its isotopes in mid-oceanic basaltic glasses. Earth Planet Sci. Lett. 69, 43–57. Dixon, J. E. and Stolper, E. M. (1995) An experimental study of water and carbon dioxide solubilities in mid-ocean ridge basaltic liquids. Part II: Applications to degassing. J. Petrol. 36, 33–1646. Dixon, J. E., Stolper, E., and Delaney, J. R. (1988) Infrared spectroscopic measurements of CO2 and H2O in Juan de Fuca basaltic glasses. Earth Planet Sci. Lett. 90, 87–104. Dixon, J. E., Clague, D. A., and Stolper, E. M. (1991) Degassing history of water, sulfur, and carbon in submarine lavas from Kilauea volcano, Hawaii. J. Geol. 99, 371–394. Dixon, J. E., Stolper, E. M., and Holloway, J. R. (1995) An experimental study of water and carbon dioxide solubilities in mid-ocean ridge basaltic liquids. Part I: Calibration and solubility models. J. Petrol. 36, 1607–1631. Dixon, J. E., Kingsley, R., Schilling, J. G., and Poreda, R. (1997) Water and CO2 concentrations in basaltic glasses from the Easter Microplate & Easter-Sals y Gomez Seamount Chain: Implications for water in mantle reservoirs. EOS 78, F688. Exley, R. A., Mattey, D. P., and Pillinger, C. P. (1987) Low temperature carbon components in basaltic glasses—reply to comment by H. Craig. Earth Planet Sci. Lett. 82, 387–390. Exley, R. A., Mattey, D. P., Clauge, D. A., and Pillinger, C. P. (1986) Carbon isotope systematics of a mantle hotspot: A comparison of Loihi seamount and MORB glasses. Earth Planet Sci. Lett. 78, 189 –199. Farley, K. N., and Newman, S. (1994) H2O, CO2, sulfur and ƒO2 systematics in basalts from the ancient and modern Lau backarc Basin, SW Pacific. Trans. Am. Geophys. Union (EOS) 75, 731. Fine, G. J. and Stolper, E. M. (1985) The speciation of carbon dioxide in sodium aluminosilicate glasses. Contrib. Mineral. Petrol. 91, 105–121. Fine, G. J. and Stolper, E. M. (1985/86) Dissolved carbon dioxide in basaltic glasses: Concentration and speciation. Earth Planet Sci. Lett. 76, 263–278. Gribble, R. F., Stern, R. J., Bloomer, S. H., Stuben, D., O’Hearn, T. and Newman, S. (1996) MORB mantle and subduction components interact to generate basalts in the southern Mariana Trough back-arc basin. Geochim. Cosmochim. Acta 60, 2153–2166. Hilton, D. R., Macpherson, C. G. and Hammerschmidt, K., Coupled rare gas and major volatile behaviour in the Southern Mariana Trough (in prep.). Javoy, M. and Pineau, F. (1991) The volatiles record of a “popping” rock from the Mid-Atlantic Ridge at 14°N: Chemical and isotopic composition of gas trapped in the vesicles. Earth Planet Sci. Lett. 107, 598 – 611. Jendrzejewski, N., Javoy, M. and Trull, T. (1996) Mesures quantitatives de carbone et d’eau dans les verres basaltiques naturels par spectroscopie infrarogue. Parte II: l’eau. C. R. Acad. Sci., Paris 322, 735–742.

Jendrzejewski, N., Trull, T. Pineau, F. and Javoy, M. (1997) Carbon solubility in mid-ocean ridge basaltic melt at low pressures (250 – 1950 bar). Chem. Geol. 138, 81–92. Kyser, T. K. and O’Neil, J. R. (1984) Hydrogen isotope systematics of submarine basalts. Geochim. Cosmochim. Acta. 48, 2123–2133. Macpherson, C. G. and Mattey, D. P. (1994) Carbon isotope variations of CO2 in Lau Basin basalts and ferrobasalts. Earth Planet Sci. Lett. 121, 263–276. Macpherson, C. G., Hilton, D. R., Sinton, J. M., Poreda, R. J., and Craig, H. (1998) High 3He/4He ratios in the Manus backarc basin: Implications for manle mixing and the origin of plumes in the western Pacific Ocean. Geology 26, 1007–110. Matthez, E. A. and Delaney, J. R. (1981) The nature and distribution of carbon in submarine basalts and peridotite nodules. Earth Planet Sci. Lett. 56, 217–232. Mattey, D. P. (1991) Carbon dioxide solubility and carbon isotope fractionation in basaltic melt. Geochim. Cosmochim. Acta 55, 3467– 3473. Mattey, D. P., Exley, R. A. and Pillinger, C. T. (1989) Isotopic composition of CO2 and dissolved carbon in basalt glass. Geochim. Cosmochim. Acta 53, 2377–2386. Mattey, D. P., Carr, R. C., Wright, I. P., and Pillinger, C. T. (1984) Carbon isotopes in submarine basalts. Earth Planet. Sci. Lett. 70, 196 –206. Mattey, D. P., Taylor, W. R., Green, D. H. and Pillinger, C. T. (1990) Carbon isotopic fractionation between CO2 vapour, silicate and carbonate melts: An experimental study to 30 kbar. Contrib. Mineral. Petrol. 104, 492–505. Newman, S., Stolper, E., and Stern, R., Water and carbon dioxide in the Mariana arc – back arc system. (in prep.). Pan, V., Holloway, J. R., and Hervig, R. L. (1991) The pressure and temperature dependence of carbon dioxide solubility in tholeiitic basalt melts. Geochim. Cosmochim. Acta 55, 1587–1595. Pandya, N., Muenow, D. W., and Sharma, S. K. (1992) The effect of bulk composition on the speciation of water in submarine volcanic glasses. Geochim. Cosmochim. Acta 56, 1875–1883. Pawley, A. R. Holloway, J. R., and McMillan, P. F. (1992) The effect of oxygen fugacity on the solubility of carbon-oxygen fluids in basaltic melt. Earth Planet Sci. Lett. 110, 213–225. Pineau, F. and Javoy, M. (1983) Carbon and its isotopes in mid-oceanic basaltic glasses. Earth Planet Sci. Lett. 69, 43–57. Pineau, F. and Javoy, M. (1994) Strong degassing at ridge crests: the behaviour of dissolved carbon and water in basaltic glasses at 14N, Mid-Atlantic Ridge. Earth Planet Sci. Lett. 123, 179 –198. Sakai, H. Des Marais, D. J., Ueda, A. and Moore, J. G. (1984) Concentrations and isotopic ratios of carbon, nitrogen and sulfur in ocean-floor basalts. Geochim. Cosmochim. Acta 48, 2433–2441. Scheele, N. and Hoefs, J. (1992) Carbon isotope fractionation between calcite, graphite and CO2: An experimental study. Contrib. Mineral. Petrol. 112, 34 – 45. Scholze, H. (1960) Zur Frage der Unterscheidung zwischen H2OMolekeln und OH-Gruppen in Gla¨seru und Mineralen. Naturwissenschaften 47, 226 –227. Stolper, E. (1982) Water in silicate glasses: An infrared spectroscopic study. Contrib. Mineral. Petrol. 81, 1–17. Stolper, E. and Holloway, J. R. (1988) Experimental determination of the solubility of carbon dioxide in molten basalt at low pressure. Earth Planet Sci. Lett. 87, 397– 408. Watanabe, S., Mishima, K., and Matsou, S. (1983) Isotopic ratios of carbonaceous materials incorporated in olivine crystals from the Hualalai volcano, Hawaii—An approach to mantle carbon. Geochem. J. 17, 95–104. Yamashita, S., Kitamura, T., and Kusakabe, M. (1997) Infrared spectroscopy of hydrous glasses of arc magma compositions. Geochim. J. 31, 169 –174.

104.28 102.00 102.90 101.10 99.29 103.11 106.40 99.76 101.83 99.66 103.25 105.56 101.58 110.4 101.9 101.81 109.31

105.02 105.53 102.80 102.80 112.56

98.04 91.45 102.56 94.69 97.16 102.58

SIO Mariana Trough 71DSTYPE3 74DS3-2 79DS2-1 79DS2-1 80DS5-1 88DS1-4 71DS1-1 84DS3-1 88DS1-1 74DS5-1 18DS1-3 86DS6 86DS4-2 74DS1-1 SO76GTVA1-2 74DS2-4 73GTVA1-2

East Pacific Rise ALV981-R23 ALV981-R23 ALV981-R23 ALV981-R23 ALV981-R23

RHUL Mariana Trough 71DSTYPE3 74DS3-2 79DS2-1 80DS5-1 88DS1-4 71DS1-1

2.19 2.02 0.46 11.11 2.46 6.16

9.56 15.17 9.73 6.46 5.47

8.73 12.55 12.51 4.00 31.87 0.51 13.17 4.50 15.24 26.22 9.14 14.52 17.28 14.12 18.90 14.72 7.32

400°C yield

⫺41.25 ⫺45.74 ⫺74.12 ⫺30.57 ⫺29.70 ⫺33.00

n.d. n.d. n.d. c.o. n.d.

n.d. c.o. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. c.o. n.d. n.d. c.o.

␦13C

0.91 25.04 0.18 2.90 1.79 55.06

4.96 2.64 2.17 2.20 3.12

1.49 19.58 5.87 n.r. 13.88 4.83 12.81 2.91 60.88 5.83 4.31 7.33 5.84 10.57 5.37 6.45 4.08

600°C yield

n.d. ⫺13.87 n.d. ⫺38.67 ⫺42.31 ⫺9.13

n.d. n.d. n.d. ⫺33.75 n.d.

n.d. n.d. n.d. n.d. ⫺8.64 ⫺25.11 n.d. n.d. n.d. ⫺27.60 n.d. n.d. ⫺26.43

n.d. ⫺26.46 n.d.

␦13C

n.r. CO2 yield not resolved from blank. n.d. not determined. co.o. gas combined with following step(s) for isotopic analysis. a: Isotope ratio fractionated during gas handling.

weight mg

Sample

1.68 3.82 13.79 0.47 1.58 3.17

n.r. 3.67 0.51 3.05 0.84

18.35 17.40 1.58 2.43 2.24 6.59 8.69 16.31 22.99 2.42 10.68 12.95 5.39 3.32 5.93 7.05 6.84

700°C yield

⫺22.60 ⫺18.39 ⫺8.94 n.d. ⫺25.93 ⫺15.52

n.d. n.d. c.o. c.o.

⫺8.85 ⫺8.45 n.d. n.d. n.d. n.d. n.d. c.o. c.o. c.o. c.o. c.o. c.o. c.o. c.o. c.o. c.o.

␦13C

25.45 20.75 42.16 1.68 54.42 33.94

n.r. 1.14 2.97 4.19 4.49

20.83 18.80 35.77 37.20 7.30 39.40 10.54 97.30 26.36 0.95 2.19 10.48 15.76 1.09 0.89 9.64 34.44

800°C yield

⫺9.24 ⫺11.46 ⫺9.46 ⫺18.87 ⫺9.99 ⫺7.37

n.d. n.d. c.o. c.o.

⫺10.50 ⫺11.41 ⫺12.07 lost n.d. ⫺10.87 n.d. ⫺5.78 c.o. c.o. c.o. ⫺9.57 c.o. c.o. c.o. c.o. ⫺8.49

␦13C

11.47 10.33 13.40 83.32 18.82 13.35

3.31 3.16 3.31 5.80 14.69

13.80 12.18 13.72 14.04 68.71 15.20 10.33 17.64 11.90 0.38 2.89 48.43 34.88 0.99 1.33 5.49 8.33

900°C yield

⫺11.56 ⫺13.53 ⫺11.62 ⫺9.61 ⫺10.00 ⫺9.58

n.d. n.d. n.d. ⫺17.40 ⫺14.80

⫺11.28 ⫺11.90 ⫺12.78 c.o. ⫺9.26 ⫺10.47 c.o. c.o. ⫺9.46 c.o. ⫺19.44 ⫺9.65 ⫺11.59 c.o. c.o. ⫺34.30 n.d.

␦13C

7.31 6.92 11.60 16.47 27.79 8.34

4.06 6.46 4.85 6.57 3.76

13.80 13.21 16.51 16.21 19.81 31.18 8.59 20.29 42.08 0.91 2.63 4.08 12.12 1.23 1.82 9.12 22.66

1000°C yield

⫺11.53 ⫺14.84 ⫺11.74 ⫺9.45 ⫺10.36 ⫺9.94

n.d. c.o. c.o. c.o. ⫺14.33

⫺10.13 ⫺10.35 ⫺11.75 ⫺10.44 ⫺9.94 ⫺9.25 ⫺10.64 ⫺9.82 ⫺10.24 ⫺13.79 c.o. n.d. ⫺10.94 ⫺29.76 c.o. c.o. ⫺8.43

␦13C

25.05 43.91 44.23 46.39 13.90 26.80

42.82 12.93 5.11 14.69 32.87

24.70 25.25 41.68 21.40 40.54 13.95 36.55 13.85 17.72 2.50 5.34 46.39 30.63 1.16 1.82 3.75 9.33

1100°C yield

⫺9.92 ⫺8.06 ⫺8.26 ⫺7.44 ⫺12.07 ⫺7.00

n.d. c.o. c.o. ⫺8.60 c.o.

lost ⫺7.24 ⫺8.12 c.o. ⫺10.65a ⫺11.09 ⫺7.68 ⫺9.61 ⫺9.61 c.o. c.o. ⫺10.54 ⫺9.12 c.o. c.o. c.o. ⫺10.01

␦13C

59.14 28.10 31.12 30.04 36.71 35.30

156.21 196.67 199.47 193.26 173.52

47.87 28.10 24.29 40.37 27.95 28.34 33.14 103.81 44.38 4.77 6.36 28.04 6.39 1.20 2.11 1.89 27.80

1200°C yield

⫺8.66 ⫺8.11 ⫺9.09 ⫺7.41 ⫺8.10 ⫺6.46

n.d. ⫺6.13 ⫺5.14 ⫺5.59 ⫺5.74

⫺7.86 ⫺7.56 ⫺8.39 ⫺7.70 ⫺7.48 ⫺7.55 ⫺6.80 ⫺6.64 ⫺7.60 ⫺14.48 ⫺14.47 ⫺9.38 ⫺16.99 ⫺24.39 ⫺26.59 ⫺32.98 ⫺7.70

␦13C

Table 2. Results of CO2 extraction from basaltic glass by stepped heating at Scripps Institution of Oceanography (SIO) and Royal Holloway University of London (RHUL). CO2 yields are in mcm3STP/g and 13C/12C ratios are reported in the ␦ notation relative to V-PDB.

APPENDIX 1

CO2, 13

C/12C and H2O variability in natural basaltic glasses 1813