Lithium isotope fractionation during magma degassing: Constraints from silicic differentiates and natural gas condensates from Piton de la Fournaise volcano (Réunion Island)

Author's personal copy Chemical Geology 284 (2011) 26–34

Contents lists available at ScienceDirect

Chemical Geology j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c h e m g e o

Lithium isotope fractionation during magma degassing: Constraints from silicic differentiates and natural gas condensates from Piton de la Fournaise volcano (Réunion Island) I. Vlastélic a,b,c,⁎, T. Staudacher d, P. Bachèlery e, P. Télouk f, D. Neuville g, M. Benbakkar a,b,c a

Clermont Université, Université Blaise Pascal, Laboratoire Magmas et Volcans, BP 10448, F-63000 Clermont-Ferrand, France CNRS, UMR 6524, LMV, F-63038 Clermont-Ferrand, France c IRD, R 163, LMV, F-63038 Clermont-Ferrand, France d Observatoire Volcanologique du Piton de la Fournaise, Institut de Physique du Globe de Paris, CNRS UMR 7154, 14 RN3, le 27èmekm, 97418, La Plaine des Cafres, La Réunion, France e Laboratoire GéoSciences Réunion, Université de La Réunion, Institut de Physique du Globe de Paris, CNRS UMR 7154, 15 Avenue René Cassin, 97715 Saint-Denis cedex 09, La Réunion, France` f Laboratoire des Sciences de la Terre, Ecole Normale Supérieure de Lyon, CNRS UMR 5570 46 Allée d'Italie, 69364 Lyon cedex 07, France g Laboratoire de Géochimie et Cosmochimie, Université Paris Diderot, Institut de Physique du Globe de Paris, CNRS UMR 7154, 1 rue Jussieu, 75238 Paris cedex 05, France b

a r t i c l e

i n f o

Article history: Received 30 August 2010 Received in revised form 31 January 2011 Accepted 2 February 2011 Available online 5 March 2011 Edited by: R.L. Rudnick Keywords: Lithium isotopes Isotopic fractionation Magma degassing Piton de la Fournaise volcano, Réunion Island

a b s t r a c t Recent volcanic products from the Piton de la Fournaise Volcano, Reunion, show pronounced depletion or enrichment in lithium and significant isotopic fractionation related to degassing. (1) trachytic pumices from the April 2007 eruption show extreme Li depletion (90%) and isotopic fractionation (δ7Li of − 21‰). The depletion of water and volatiles (Cl, F, B, Cs) in these samples suggests that Li loss occurred in response to degassing, which most likely occurred as the small, isolated volume of magma underwent extensive differentiation near the surface. Because the pre-degassing composition is relatively well known, the composition of the degassed pumice constrains the partition coefficient to 60 b DV–M b 135 and the isotopic fractionation factor, αV–M, to 1.010 at magmatic temperatures. Unlike DV–M, αV–M does not depend on whether crystallization and degassing occurred successively or concomitantly. (2) basaltic samples from the interior wall of the long-lived 1998 Piton Kapor were extensively altered by acidic gas. They also show extreme Li depletion, but barely significant isotopic fractionation (δ7Li = + 4.5‰), suggesting that hightemperature leaching of Li by volcanic gas does not significantly fractionate Li isotopes. (3) high-temperature (400–325 °C) gas condensates formed during degassing of the thick lava flow of April 2007 display high Li contents (50–100 ppm), which are consistent with Li being as volatile as Zn and Sn. Their isotopically light Li signature (average of − 1.7‰) is consistent with their derivation from isotopically heavy vapor (+ 13.5‰) if the factor of isotopic fractionation between condensate and vapor is less than 0.985. A degassingcrystallization model accounts for the evolution of trace species, which, like lithium, are volatile but also moderately incompatible. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Alkali metals are only moderately volatile, in the sense that only a small proportion (b0.1%) of the initial budget of mantle derived melts is ultimately lost by magmas (Rubin, 1997). On the other hand, alkali are largely mobilized during exsolution of H2O-rich vapor phase in magmatic systems (Sakuyama and Kushiro, 1979). Selective vapor transport of potassium has been suggested during the late stage of crystallization of oceanic magmas (Sinton and Byerly, 1980), and from bottom to top of lava lakes (Richter and Moore, 1966) although the ⁎ Corresponding author at: Laboratoire Magmas et Volcans, Observatoire de Physique du Globe de Clermont-Ferrand, UMR 6524, 5 Rue Kessler, 63038 ClermontFerrand, France. Tel.: +33 4 73 34 67 10; fax: +33 4 73 34 67 44. E-mail address: [email protected] (I. Vlastélic). 0009-2541/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.chemgeo.2011.02.002

Kilauea data on which this idea is based were subsequently questioned (Helz et al., 1994). There has been recently a growing interest in lithium, the lightest alkali metal, but above all one of the fastest diffusing elements in silicates (Jambon and Semet, 1978; Giletti and Shanahan, 1997; Richter et al., 2003). Indeed, lithium seems to be particularly sensitive to vapor transfer, either in shallow magma conduit shortly before eruption (Berlo et al., 2004; Kent et al., 2007) or during post-eruptive degassing of thick lava flows (Kuritani and Nakamura, 2006). While the use of lithium abundance as tracer of degassing processes is still in its infancy, new perspectives arise from lithium stable isotopes (6Li and 7Li) geochemistry. The main reason is that our knowledge of the laws governing lithium isotopes fractionation in nature has improved significantly in recent years. First, it is now well established that heavy lithium partitions preferentially into aqueous

Author's personal copy I. Vlastélic et al. / Chemical Geology 284 (2011) 26–34

fluids over silicate rocks, and that the magnitude of this phenomenon decreases with increasing temperature (e.g., Wunder et al., 2007; Millot et al., 2010). Second, as 6Li diffuses faster than 7Li, large isotopic fractionation arises during diffusion (e.g., Richter et al., 2003). These two types of isotopic fractionation, which, for convenience, will be simply referred to as “chemical” and “kinetic”, respectively, may occur during magma degassing (Beck et al., 2004; Rowe et al., 2008; Schiavi et al., 2010), although little is yet known on this topic. Taking advantage of the recent intense volcanic activity of Piton de la Fournaise (Réunion Island), a Li isotopic study of late-stage magmatic processes was undertaken. This paper focuses on extensively devolatilized samples (silicic differentiates) and gas condensates (natural sublimates), which provide new constraints on the behavior of Li isotopes during magma degassing. 2. Geological setting and samples Piton de la Fournaise (Réunion Island, Indian Ocean) in an intraplate shield volcano that has erupted on average once a year since 1930, and seemingly more frequently since 1998 (see a review of the most recent activity in Peltier et al. (2009)). The volcano produces dominantly transitional basalts with little compositional variability (these basalts are referred to as Steady State Basalts (SSB) following Albarède et al., 1997) and, every 15–30 years and more frequently since 2001, olivine-rich basalts (the most magnesian being commonly referred to as “oceanites”). Geophysical data support the existence of a shallow magma reservoir near sea level (2.5 km depth) and a deeper reservoir at the crust-mantle interface (7.5 km depth)

27

(Peltier et al., 2008). The volume of the storage system, in the range of 0.1–0.35 km3, and magma residence time of 15–30 years were estimated based on geochemical data (Albarède, 1993; Sigmarsson et al., 2005; Vlastélic et al., 2009a). Among the recent products of Piton de la Fournaise, this study focused on samples that underwent pronounced depletion or enrichment in lithium in relation with degassing processes (Fig. 1). They include trachytic pumices recovered near the main vent of April 2007 eruption. The pumices are almost completely glassy, highly vesicular (80–90%) and often coated by basaltic glass. These silicic differentiates, the only known at Piton de la Fournaise, are thought to originate from small batches of magmas trapped at shallow depth within the volcanic edifice (Bachèlery, in prep.). Comparing the Pb isotopic signature of the pumices to the detailed Pb isotope temporal record (Vlastélic et al., 2009a) suggests derivation from liquids erupted between 1977 and 1986. Theses pockets of differentiated liquids may have been disrupted and entrained during the paroxysmal phase of the eruption, on April 5th (see review of the eruption by Staudacher et al. (2009)). To confirm this hypothesis we also analyzed the olivine-bearing glass coating the pumices. The second type of sample is gas-altered rocks from the interior wall of the 1998 Kapor crater. These rocks, which have been extensively leached by acidic gas during the unusually long 1998 eruption (nearly 6 months), have been selected to evaluate how Li isotopes behave during gas-rock interaction. The last type of sample is gas condensates that formed during degassing of the April 2007 lava flow. Post-eruptive degassing processes, which include formation of segregation veins and gas-filter pressing (Martin and Sigmarsson, 2007), have been shown to

Fig. 1. Map of Piton de la Fournaise volcano showing the eruptions of March 1998 (Piton Kapor) and April 2007 (Piton Tremblet). Locations and photos of the studied samples are indicated by numbers: (1) gas altered basalt from the interior wall of the 1998 Kapor crater. (2) trachytic pumice and its basaltic coating (recovered in May 2007 near Piton Tremblet) were probably emitted during the paroxysmal phase of April 2007 eruption (Bachèlery et al., in prep.) (3) gas condensates from a fumarole that was active between 2007 and 2009 in the region where April 2007 lava flow is the thickest (ca. 50 m).

Author's personal copy 28

I. Vlastélic et al. / Chemical Geology 284 (2011) 26–34

efficiently mobilize lithium (Kuritani and Nakamura, 2006). Since Piton de la Fournaise volcano lacks long-lived fumaroles, the voluminous (130 Mm3) and slowly cooling lava flow of April 2007 (Staudacher et al., 2009) offers a rare possibility to sample magmatic gas, in particular in its thickest part where an active fumarole was discovered. Natural gas condensates were sampled four times at this site between August 2008 and November 2009 as the vent temperature decreased from ~ 400 to ~ 300 °C. In addition, a large number of steady-state basalts (n = 31) and olivine-rich basalts (n = 30) erupted between 1927 and 2007 were also analyzed for comparison. 3. Analytical methods The bulk major element chemistry was determined by ICP-OES (rocks) and ion chromatography (condensates). Scanning electron microscopy (SEM) was used to investigate the compositional heterogeneity of the altered basalt and to determine the mineralogy of gas condensates. Water content was determined in the trachytic pumice by Raman spectroscopy (IPGP, Paris) using a recently improved calibration curve (Le Losq et al., in press). The Raman spectra were recorded using a T64000 Jobin-Yvon spectrometer, with an excitation line at 514.532 nm. Acquisition times and laser power were set to 900s and 20 mW, respectively. Data treatment and accuracy of the method are detailed in Le Losq et al. (in press). Wholerock analysis of Cl, F and B concentrations has been assigned to the Service d'Analyse des Roches et Minéraux (CRPG-Nancy). Chlorine content was determined by a spectrophotometric method based on the formation of ferrithiocyanate, while F content was determined by potentiometry using an ion-selective electrode. The precision of F and Cl determination is 10–20% (2σ) for the concentration range (100– 500 ppm) relevant to our study (Vernet et al., 1987). Following sodium carbonate fusion, B was purified on ion-exchange resin and its concentration was determined by the colorimetric method using Carmin with a precision of ca. 25%. Other trace elements and Li isotopic compositions were analyzed on same dissolutions. Between 50 and 100 mg of sample were dissolved in HF-HNO3 (rocks) or diluted HNO3 (gas condensates). Once digested, solutions were evaporated to near dryness and re-dissolved in 7M HNO3. The dissolved samples were then split into two aliquots. The first was intended for determination of trace element abundance by quadrupole ICPMS (Agilent 7500, Laboratoire Magmas et Volcans). To this end, the aliquots were evaporated to near dryness and subsequently diluted in HNO3 0.4 M to reach a total dilution factor ranging from 5000 (rocks) to 20,000 (gas condensates). The reaction cell (He mode) was used to reduce interferences on masses ranging from 45 (Sc) to 75 (As). The signal was calibrated externally with a reference basaltic standard (BHVO-2) dissolved as samples (rocks) or with a synthetic standard (gas condensates). One of these two standards and pure HNO3 0.4 M were measured every 4 samples. The external reproducibility of the method, as estimated by running repeatedly different standards (BCR-2, BIR, BEN) is b5% for most lithophile elements and b15% for most chalcophile and siderophile elements. The second aliquot of dissolved sample was used for measurement of lithium isotopic composition. Following evaporation and conversion of ions to chloride form (with 1 ml of HCl 6 N), samples were successfully fully dissolved in 1 ml of HCl 0.5 N, and Li was purified on AG50W-X8 cation exchange resin following the two-step method described in Vlastélic et al. (2009b). The lithium chemistry blank measured by quadrupole ICP-MS is less than 0.1 ng. Given the amount of lithium required for isotopic analysis (N50 ng) and the loading limit of the first-step column (the equivalent of 50 mg of rock), the Li depleted pumices were processed through two columns and Li fractions were merged before further purification on the second-step column. Lithium isotopic compositions were measured by MC-ICPMS (Nu 500) at the Ecole Normale Supérieure de Lyon. Samples were introduced through a

desolvator (Nu DSN) at a rate of 100 μl/min, yielding a total Li beam of 4 to 6 V for 70 ng/g Li solutions. Standard operating conditions were used (RF power of 1350 W, Ar cool gas flow of 13 L/min, Ar auxiliary gas flow of 1 L/min, Ar sample gas flow of 0.8 L/min, and acceleration voltage of 4000 volts). Lithium isotopes were measured in static mode in L5 and H6 faraday cups. Measurements (60 cycles of 10s) were performed following an uptake-stabilization time of 90s. The washout procedure (300s with 0.65 N HNO3 and 300s with 0.05 N HNO3) reduced Li signal by a factor of 104. Mass fractionation was monitored externally with IRMM-016 standard using a sample-standard bracketing technique. Depending on the amount of lithium extracted, multiple measurements of each sample were performed. Repeated analysis of the USGS BHVO-2 standard (batch 759) over 4 years yielded δ7Li= +4.2‰ ±0.5 (2σ, 17 dissolutions). Repeated measurement (n = 12) of a L-SVEC solution (not processed through separation columns) during a single session yielded δ7Li= −0.20‰ ±0.04 (S.E.), raising the possibility that L-SVEC is isotopically slightly lighter than IRMM-016, as previously proposed (Millot et al., 2004). 4. Results Lithium isotopic compositions and trace element concentrations of rocks and gas condensates are reported in Table 1, together with the average composition of steady-state basalts and oceanites, the most common types of basalts erupted at Piton de la Fournaise (a complete lithium data set is provided in Supplementary Table 1). Major element chemistry of the studied samples is reported in Supplementary Table 2. 4.1. Silicic differentiates The pumice samples 0704-PS and 0704-PN have total alkali and SiO2 contents (ca. 11 wt.% and 62 wt.%, respectively) that plot within the field of trachytes (Bachelery, in prep.). Loss-on-ignition at 1000 °C is barely significant, indicating that water content in these differentiated samples is probably less than 0.5 wt.% (i.e., the maximum weight gain due to oxidation of iron). Consistently, precise determination of water content by Raman spectroscopy yielded 0.2 ± 0.1 wt.%. By comparison to steady-state basalts, the 2007 trachytic pumices are enriched in most incompatible trace elements (Fig. 2a,b). The degree of enrichment is the highest for Nb and Ta (ca. 7) and decreases progressively from light REE (ca. 5) to heavy REE (ca. 2). The increase of Zr and Hf concentration with digestion strength (Table 1) suggests the occurrence of zircons. There are also a number of negative anomalies, most of them (Eu, Sr, Ba, and Pb) being consistent with lowpressure fractionation of plagioclase feldspar. Lithium, B and Cs, together with Cl and F (not reported on the plots), all being incompatible in plagioclase, show the most pronounced anomalies, however. The lithium depletion is accompanied by strong isotopic fractionation, the two pumice samples analyzed showing extremely light lithium compositions (−21 b δ7Li b −17‰) when compared to historical lavas of Piton de la Fournaise (+2.7 b δ7Li b +4.9‰), and terrestrial rocks in general. The olivine-bearing glass (0704-BAS) coating the pumices has trace element pattern (Fig. 2c) and Li isotopic signature (δ7Li of + 4.2‰) indistinguishable from typical oceanites. 4.2. Gas-altered basalt Whole rock analysis (ICP-OES) of the altered basalt 9809–301 revealed that it is essentially made of SiO2 (58 wt.%), CaO (11 wt.%) and a volatile component lost during sample fusing. Compared to fresh basalts, this specimen shows a marked enrichment in silica and depletion in Al2O3, Fe2O3, and MgO, which have concentrations that do not exceed 2 wt.%. Inspection of the sample by SEM revealed the occurrence of a silica-rich phase (SiO2 up to 87% vol.%) and anhydrite

Author's personal copy I. Vlastélic et al. / Chemical Geology 284 (2011) 26–34

29

Table 1 Trace element concentrations and Li isotopic compositions of rocks and gas condensates. Main rock types SSBa

Devolatilized trachytic pumice

Oceanitesb 0704-PS

Long. (°S) Lat. (°E)

dupc

0704-PN

Gas altered basalt

Gas condensates

0704-BAS 9809-301-a 9809-301-b 9809-301-c

55°46.47′

55°46.32′

55°42.84′

21°16.84′

21°16.93′

21°13.65′

07.05-a

07.05-b 08.10.281 09.06.261 09.11.201

55°47.57′ 21°17.25'

Mean compositionsd

~ 400°

384°

345°

325°

Material

powder powder

powder

powder powder

chip

chip

chip

chip

chip

chip

chip

chip

chip

Li Be B Sc Ti V Cr Co Ni Cu Zn Ga As Rb Sr Y Zr Nb Cd In Sn Sb Cs Ba La Ce Pr Nd Sm Eu Tb Gd Dy Ho Er Tm Yb Lu Hf Ta W Tl Pb Bi Th U Cl F δ7Lie 2σ n

5.85 1.04 3.7 32.6 16,130 300 225 44.9 92.1 104 110 22.29 0.82 17.5 353 28.8 193 22.2 0.065 0.089 1.83 0.065 0.261 136 19.6 44.5 5.94 26.0 6.28 2.11 0.977 6.55 5.70 1.07 2.79 0.367 2.24 0.308 4.72 1.39 0.295 0.046 1.70 0.020 2.26 0.560 215 464 + 3.6 ± 0.8 31f

0.99 3.87 2.0 5.61 1276 10.8 20.1 4.14 29.2 8.47 32.2 31.4 1.49 52.6 318 51.2 456 150 0.116 0.101 3.04 0.026 0.042 223 89.9 176 20.1 69.8 12.6 1.41 1.69 10.6 9.92 1.89 5.20 0.733 4.57 0.63 10.6 8.25 0.384 0.102 2.65 0.040 12.7 2.32 80 100 − 21.0 −

1.05 4.58 n.d. 7.47 1318 12.6 151 5.03 38.7 10.2 36.2 32.9 1.85 55.4 331 61.6 736 170 0.174 0.107 3.22 0.029 0.040 229 102 182 20.7 71.8 13.0 1.46 1.80 11.2 10.5 2.01 5.63 0.826 5.35 0.73 15.2 8.49 0.375 0.104 2.68 0.043 13.6 2.65 n.d. n.d. − 19.7 ± 0.4 2g

4.01 0.69 n.d. 20.39 9040 172 443 98.4 887 59.6 104 13.1 0.432 9.92 197 16.3 116 14.1 0.041 0.056 1.04 0.040 0.143 77.2 11.7 26.6 3.52 15.3 3.68 1.20 0.576 3.87 3.39 0.635 1.65 0.218 1.35 0.186 2.83 0.882 0.168 0.026 0.900 0.023 1.41 0.336 n.d. n.d. + 4.2 ± 0.0 2g

0.356 0.029 n.d. 5.76 11,566 14.2 11.0 1.80 3.76 5.74 12.7 2.84 1.37 2.92 196 10.8 107 25.3 0.114 0.007 0.903 0.039 0.016 150 9.29 21.3 2.83 12.4 2.96 0.989 0.465 3.03 2.62 0.468 1.20 0.157 0.862 0.123 3.32 1.75 0.463 0.036 0.603 0.074 1.07 0.029 n.d. n.d. + 4.4

0.599 0.092 n.d. 10.0 9391 26.2 15.6 3.32 6.44 11.8 16.7 3.30 1.19 3.74 403 24.9 101 22.3 0.120 0.012 0.490 0.024 0.033 134 17.2 41.2 5.40 24.2 6.31 2.08 1.03 6.49 5.80 1.05 2.77 0.349 2.02 0.267 2.84 1.50 0.360 0.030 0.741 0.081 1.76 0.058 n.d. n.d. + 4.1 –

0.450 0.049 n.d. 9.95 9163 17.8 15.7 2.24 4.15 7.33 16.2 2.46 1.21 3.71 214 12.8 102 22.7 0.088 0.006 0.498 0.018 0.030 119 10.3 26.1 3.57 15.8 3.73 1.23 0.516 3.61 2.87 0.533 1.42 0.192 1.19 0.167 3.05 1.54 0.289 0.018 0.502 0.077 1.07 0.043 n.d. n.d. + 5.1 –

101 2.98 n.d. 2.80 818 469 84.8 19.85 98.95 7821 638 1.90 36.51 361 11.4 0.205 25.1 0.947 503 22.7 18.3 1.22 11.2 7.91 0.084 0.170 0.022 0.111 0.029 0.009 0.006 0.031 0.040 0.008 0.020 0.005 0.030 0.005 0.585 b 0.005 90.0 378 437 39.1 0.022 0.407 n.d. n.d. − 1.3 ± 0.6 10g

71.7 2.44 n.d. 3.27 941 564 140 21.75 109 6693 523 2.24 26.71 289 15.2 0.314 29.8 1.10 443 22.9 22.4 1.52 10.2 8.73 0.117 0.284 0.042 0.173 0.068 0.020 0.007 0.050 0.050 0.011 0.037 0.005 0.039 0.007 0.760 b 0.005 120 396 460 33.2 0.029 0.365 n.d. n.d. − 1.3 ± 0.2 2g

52.7 1.91 n.d. 3.16 762 339 101 17.04 85.49 5098 431 1.71 29.59 280 9.59 0.286 23.4 0.814 357 20.1 17.5 1.27 6.63 6.45 0.124 0.253 0.033 0.146 0.039 0.017 0.009 0.055 0.047 0.012 0.037 0.009 0.037 0.007 0.575 b0.005 49.8 409 454 25.2 0.022 0.249 n.d. n.d. − 2.8 ± 0.2 3g

92.5 2.88 n.d. 2.20 1139 1048 125 19.03 93.26 8446 600 2.66 17.47 430 24.8 0.588 26.9 1.68 461 26.0 38.8 3.34 18.3 8.22 0.437 0.947 0.127 0.573 0.163 0.048 0.024 0.136 0.126 0.025 0.071 0.010 0.062 0.009 0.680 b 0.005 234 624 836 60.5 0.033 0.601 n.d. n.d. − 1.4 ± 0.4 3g

70.4 2.25 n.d. 1.67 637 395 76.0 21.57 108 5598 523 1.47 35.11 348 5.67 0.176 20.3 0.877 359 19.8 27.5 1.59 8.14 4.82 0.069 0.151 0.022 0.096 0.024 0.008 0.004 0.025 0.029 0.007 0.028 0.003 0.023 0.004 0.465 b 0.005 64.1 443 341 31.9 0.011 0.314 n.d. n.d. − 1.8 ± 0.4 4g

3.70 0.501 2.0 20.7 n.d. 169 n.d. 125 1126 56.6 111 7.63 n.d. 8.48 175 15.2 113 10.8 n.d. n.d. n.d. n.d. 0.123 64.6 9.10 20.8 2.79 12.3 3.02 1.01 0.487 3.01 2.70 0.522 1.40 0.184 1.09 0.154 2.58 0.759 n.d. 0.023 0.802 n.d. 1.04 0.257 118 226 + 3.9 ± 0.8 21f

T(°C) ~ 400°

1.41 4.95 2.0 4.28 1076 3.18 9.80 2.78 16.7 14.6 9.27 35.1 1.63 58.0 264 46.9 267 154 0.065 0.143 4.93 0.026 0.017 229 95.6 188 21.7 75.9 13.7 1.49 1.68 11.0 9.31 1.72 4.73 0.657 4.14 0.58 7.57 8.07 0.292 0.088 1.94 0.033 12.5 0.988 117 60 − 17.0 –

All concentration in g/g (ppm).n.d.: not determined. a SSB: Steady State Basalts. Mean composition inferred from 1931 to 2007 lavas with MgO b 9%wt. b Océanites: Lavas rich in cumulative olivine. Mean composition inferred from 1931 to 2007 lavas with MgO N 20%wt. c PARR bomb dissolution. d A complete lithium data set is provided in Supplementary Table 1. Source of trace element data: Vlastélic et al. (2005, 2007) and unpublished data. e δ7Li = ((7Li / 6Li)sample / (7Li/6Li)IRMM − 1) ⁎ 1000. f number of sample analyzed. g Repeated analysis of the same dissolution.

(CaSO4). Silicification in fumarolic environment was shown to be the result of extensive leaching of major cations by acidic gas condensates (Africano and Bernard, 2000). Replicate analysis of this sample points to heterogeneous distribution of trace elements, the two extreme patterns being reported in Fig. 3. The alteration resulted in no loss in

Ba, Ta and Nb, small to significant (up to 50% in sample 9809-301a) loss of rare earth elements, Th, Pb, Zr, Hf and Y, and major to extreme (N90%) loss of Cs, Rb, U, Be and Li. In contrast with the pumice, lithium loss is accompanied here by barely significant isotopic fractionation (+4.1 b δ7Li b +5.1‰).

Author's personal copy 30

I. Vlastélic et al. / Chemical Geology 284 (2011) 26–34

a

10

Cs Ba U

B

Pb

Sr

Eu

a

Li

10

9809-301a Concentration / SSB

Concentration / SSB

0704-PS

1

0.1

b

b

10

Concentration / SSB

Concentration / SSB

1

0.1

1

0.01 10

1

0.1

0.01

0.01

Concentration / Oceanite

0.

9809-301b

0704-PN

c

1

10

0704-BAS

Cs Ba U Ta Ce Pr Nd Sm Hf Gd Dy Y Tm Yb Rb Th Nb La Be Pb Sr Zr Eu Tb Ho Er Li Lu

Fig. 3. Trace element patterns of the gas-altered basalts from the 1998 Piton Kapor. Concentrations are normalized to the average concentrations of steady-state basalts given in Table 1. Vertical lines indicate elements that underwent the largest depletion. Amongst the three samples analyzed (see Table 1), the two extreme patterns are shown here.

1

5. Discussion 5.1. Origin of lithium depletion and isotopic fractionation in silicic differentiates 0.1 Cs Ba U Nb Ta Ce Pr Na Sr Zr Eu Tb Ho Er Li Lu Rb Th K B La Be Pb Nd Sm Hf Gd Dy Y Tm Yb

Fig. 2. Trace element patterns of the April 2007 trachytic pumice (a,b) and its basaltic coating (c). Concentrations are normalized to the average concentrations of steadystate basalts (a,b) and oceanites (c), both being given in Table 1. Vertical lines are for elements showing negative anomalies.

4.3. Gas condensate

The trachytic pumices display very low lithium contents and extremely light Li isotope compositions that have not yet been reported in fresh volcanic rocks. Such a combination rules out diffusive influx of lithium, which best accounts for very low δ7Li signatures in Lirich rocks only (Rudnick and Ionov, 2007; Marschall et al., 2007). Hydrothermal alteration, which preferentially removes 7Li from solids, could yield such light signatures, but only in low-temperature environments (Wunder et al., 2007; Millot et al., 2010), which is inconsistent with the magmatic origin of pumices. Independently, the

Analysis of the gas condensates by SEM identified K–Na sulfates with K/Na suggesting the main occurrence of aphthitalite ((K,Na)3Na (SO4)2). The distribution of trace elements in gas condensates is shown on Fig. 4 where the elements are sorted according to their enrichment relative to lavas. Enrichment factor (EF) is defined as: ð1Þ

where X is the element of interest and XR the refractory element of reference. Beryllium was chosen here because of its low volatility and low abundance in lavas (Moune et al., 2006). The gas condensates are depleted (EF b 1) in rare earth elements and high field strength elements and enriched (EF N 1) in alkali and chalcophile elements, which pattern is consistent with element volatility at mafic volcanoes (Toutain et al., 1990; Rubin, 1997). Lithium (5 b EF b 6), like Zn, Sn, Rb, Sb, and As, displays a moderately volatile behavior during degassing of the April 2007 lava flow. The condensates have light and rather uniform Li isotopic signatures (− 2.8 b δ7Li b − 1.3‰), the lowest δ7Li (−2.8‰) being measured in the sample (08.10.281) with the lowest Li content (52.7 ppm, against N70 ppm in other samples).

Tl Cd

Gas condensates

Bi

1 000

W In

Enrichment Factor

EFX = ðX =XR Þgas condensate = ðX =XR Þlava

10 000

100 As

Li

Rb

10 Zn V

1 Co

0.1 0.01 0.001

U

Cu Pb

Cs

Sb Sn

Be Ni

Ga Zr Cr Nb Ti Hf Lu Ho Yb Sc Pr La Eu Tb Th Ba Sr Tm Dy Y Er Ce Nd Gd Sm

Fig. 4. Enrichment factors of trace elements in 2008–2009 gas condensates. Enrichment factors are calculated according to Eq. (1), using beryllium as reference. Normalization is made to the average composition of steady-state basalts (see Table 1). Elements are sorted according to their mean enrichment factor. For each element, the range of variation of the enrichment factor within the five samples analyzed is shown.

Author's personal copy I. Vlastélic et al. / Chemical Geology 284 (2011) 26–34

mantle-like oxygen isotope signature of pumices (Bachèlery et al., in prep.) does not support input of meteoric water. Fluxing by vapors released by melt degassing deeper in the plumbing system (Blundy et al., 2010) is another possibility, which could also explain the less dramatic loss of uranium, which is fluid-mobile when oxidized but non-volatile. Although this possibility cannot be definitely ruled out, our attempt to test it suggests that high-temperature leaching of lithium by magmatic gas does not significantly fractionate 7Li/6Li. On the other hand, the concomitant loss of B, Cs, Cl and F suggests that Li was lost during devolatilization of the magma, as it underwent extensive crystallization at shallow depth. Indeed, the water content of pumice samples (0.2 wt.%) is unexpectingly low given the large mass fraction of anhydrous crystals removed. Both major and trace elements are consistent with a crystallizing mineral assemblage made dominantly of clinopyroxene, plagioclase, olivine, Fe–Ti oxides, apatite, with no evidence for hydrous minerals such as amphibole or phlogopite. Elements (e.g., Th, Nb) that are highly incompatible in these minerals are strongly enriched (a factor of 6.2 on average) in the trachytic samples with respect to steady-state basalts. Assuming these elements do not enter the crystallizing phases (bulk DS–M ~0), the fraction of melt having crystallized during the basalt–trachyte differentiation is estimated at ca. 84%. From the water content of primitive melt (0.7–1.0%wt% according to Bureau et al. (1999)), and assuming that H2O is as incompatible as Ce (Michael, 1995), a water content of 2.9–4.1 wt.% would be expected in the trachytic melt if it evolved in a closed system. The residual water content of pumices (0.2 ± 0.1 wt.%) suggests that between 2.6 and 4.0 wt.% of water left the system and entrained a major fraction of the initial budget of lithium.

where ΔH20 is the mass fraction of vapor lost. Integrating Eq. (5) yields the expression of lithium concentration in the residual melt (CM) as a function of the initial melt content (CM 0 ): C

M

M

= C0 :F

ð2Þ

and, for lithium, CS :dmS + CV :dmV = −dðCM :mM Þ

ð3Þ

where C is the concentration of lithium. Following Villemant and Boudon (1999), we assume that the masses of crystallizing melt and exsolving vapor are proportional: dmS = k:dmV

ð4Þ

and, from Eq. (2), dmV = −dmM / (1+ k) and dmS = −dmM / (1+ 1 / k). Substituting in Eq. (3) yields: dCM dmM DV−M + kDS−M = ðβ−1Þ with β = CM mM k+1

ð5Þ

where DV–M and DS–M are the vapor–melt and crystal–melt partition coefficients of lithium, respectively. We then set dmM / mM = dF/ F, F being the mass fraction of residual melt expressed as: F = 1−ð1 + kÞΔH2 0

ð6Þ

ð7Þ

!M kðDS−M −DS−M Þ + DV−M −DV−M  7 7 6 6 Li 1 + k :F 6 Li 0

!M Li = 6 Li 7

7

ð8Þ

where (7Li / 7Li)0 is the initial isotopic ratio, and subscripts 7 and 6 denote for 7Li and 6Li species. If lithium isotopes partition equally into the crystallizing phases (DS–M = DS–M ): 7 6 !M Li = 6 Li 7

7

Li 6 Li

!M DV−M :F

 ðαV−M −1Þ

6 : 1 + k

ð9Þ

0

with   V V−M Li = 6 Li ðM6 = M7 Þ C7V = C6V D  M  = 7V−M =  M = M 6 7 ðM6 = M7 Þ C7 = C6 D6 Li = Li 

V−M

α

7

and 0

V−M

D6

 M 1 7 6 1 + ð M = M Þ Li = Li 7 6 V−M B C =D :@  M A 1 + αV−M ðM7 = M6 Þ 7 Li = 6 Li

where αV–M is the coefficient of isotopic fractionation between vapor and melt and M the molar mass. When there is only degassing (k = 0) and DV–MNN1, it is convenient to use the simplified Rayleigh equation describing the isotopic evolution as a function of C/C0 and α:

7

dmS + dmV = −dmM

ðβ−1Þ

Note that if DS–M ~ 0, β = DV–M / (1 + k) as inferred for chlorine (Villemant et al., 2008). Considering lithium isotopes as individual species, Eq. (7) may also be used to compute the evolution of lithium isotopic ratio of the residual melt:

5.2. Modeling lithium loss and isotopic fractionation during coupled degassing and crystallization To explain the composition of the dehydrated silicic differentiates, we resorted to the open system degassing-crystallization model previously used for chlorine (Villemant and Boudon, 1999; Villemant et al., 2008). However, the model has to be modified to take into account that, unlike chlorine, lithium significantly partitions into crystallizing phases. The mass balance equations in the system made of melt (M), crystals (S) and vapor (V) are:

31

Li 6 Li

!M

!M V−M V−M Li D −1Þðα −1Þ :F ð = 6 Li 0 7

=

!M !ðαV−M −1Þ Li CM : ð10Þ 6 Li 0 C0M 7

We stress that Eq. (10) only applies when elements strongly partition into the vapor phase, and is thus not valid when elements enter the crystalline assemblage significantly. The evolution of Li content and Li isotopic composition of the residual melt was modeled considering that crystallization and degassing occurred either successively (case A) or simultaneously (case B). In both cases, the starting composition is assumed to be that of steady-state basalts (Li = 5.85 ppm, δ7Li = + 3.6‰, see Table 1). It is also assumed that Li isotopes do not fractionate significantly during crystallization (Tomascak et al., 1999; Schuessler et al., 2009). In case A, Li is assumed to behave like heavy rare earth elements during low-pressure crystallization (Ryan and Langmuir, 1987). Its concentration in the undegassed trachytic melt is thus calculated by considering smooth SSBnormalized trace element pattern (Li = LiSSB.[Tm / TmSSB + Yb/ YbSSB] / 2 = 11.8 ppm), which, in turn, implies a solid-melt partition coefficient of ca. 0.6. As discussed above, the differentiated melt must have lost between 2.6 and 4.0 wt.% of water during degassing. In case B, the initial water content is assumed to be that of primitive melts (~1 wt.%). Thus, k on the order of 100 is inferred from the fraction of vapor lost (~0.8 wt.%) and the fraction of melt having crystallized (~84 wt.%). Lithium loss during degassing was modeled using Eq. (7) (Fig. 5), setting k = 0 (no crystallization) in case A. The low lithium content of the pumice requires 60 b DV–M b 95 in case A and β ~ 1.9 in case B (DV–M = 135 for DS–M = 0.6

Author's personal copy 32

I. Vlastélic et al. / Chemical Geology 284 (2011) 26–34

a

14

degassing

crystallization Ds-m=0.62

12

Li (ppm)

10 8 6

Dv-m=60

4

Dv-m=95

2 0

Li in pumice

0

20

40 crystal

F

b

60

800

2 vapor

F

(%)

3

4

(%)

crystallization + degassing

8

β=0.8

5.3. Lithium isotopic composition of volcanic vapor

β=1.0

6 Li (ppm)

1

entrainment into the vapor phase. In this scenario, our calculated DV–M values are overestimated. Fractionation of Li isotopes during degassing was modeled using Eq. (10) in case A (DV–M ≫ 1) and Eq. (9) in case B assuming (D6)V–M ~ DV–M (since αV–M ~ 1) (Fig. 6). The end-member pumice composition (Li = 1 ppm and δ7Li= −21‰) is consistent with αV–M of 1.0100 (case A) and 1.0099 (case B). The similarity of αV–M in both cases is at first glance surprising given the difference in the starting composition. It is consistent with αV–M being dependent only on the amount of lithium lost, which is the same in cases A and B. These values of αV–M are unexpectingly large (isotopic fractionation of ~10‰) for a high-temperature process. Large fractionations (1.005 b αV–M b 1.015) during magma degassing were also proposed to explain Li isotope systematics in 2003 ashes from Stromboli, suggesting that degassing fractionates lithium isotopes more efficiently than fluid-rock interaction (Schiavi et al., 2010).

The isotopic composition of Piton de la Fournaise volcanic gas has not been measured, but can reasonably be inferred from αV–M calculated for the dehydrated trachytic melt. The instantaneous composition of the vapor is given by:

4

β= 2

β=1.9

Dv-m + k.Ds-m 1+k

!V Li V−M =α 6 Li

!M Li 6 Li

7

Li in pumice

0 0

20

40

60

80

7

ð11Þ

Fcrystal (%) 10

Average SSB

5

crystallization

crystallization + degassing (α=1.0099)

-5

A degassing (α=1.0100)

-10

Pumice (basaltic glass coating) Pumice (trachytic glass)

-15

Pumice (bulk analysis) Common lavas (SSB + oceanites)

-20

α=(7Li/6Li)vapor/(7Li/6Li)melt

-25 0

2

4

6

8

10

12

14

15

B

10

δ7Li vapor)

and k = 100). Note that if β = 1 (DV–M = k(1 − DS–M) + 1), the competing effects of crystallization and degassing cancel out and the Li content of the melt remains constant. Taking 0.6 as an upper bound for DS–M, it can be estimated that lithium content in melt decreases only if DV–M N ~ 40. The values of DV–M required to explain Li depletion in pumice are larger than those estimated for the partitioning of lithium between fluids and melts, although truly relevant data lack in the literature. For granitic–rhyolitic compositions, DV–M was shown to increase from 0 to 13 with decreasing pressure from 400 MPa to 50 MPa (London et al., 1988; Webster et al., 1989), while 1 b DV–M b 30 was inferred by mass balance following degassing experiments (Koga et al., 2008). The elevated DV–M value inferred in this study could be a consequence of the basalt–trachyte differentiation process, which generally involves high concentrations of water (Ringwood, 1959). However, this does not explain the scarcity of the Li signature of the 2007 trachyte. The most comparable case, even if Li loss (~ 20%) is less extreme, is that of post-eruptive internal differentiation of the thick lava flow of Rishiri Volcano (Japan) (Kuritani and Nakamura, 2006). Based on these observations, it is suggested that DV–M could be very elevated during low-pressure, vapor-saturated crystallization, in particular if there is a long interaction (possibly 20 years for the 2007 trachyte) between exsolved vapor and melt. Another possibility is that fluxing by vapors released by deeper melts accompanied degassing (Blundy et al., 2010) and enhanced lithium

Undegassed trachytic melt

B

0

δ7Li (melt)

Fig. 5. Modeling Li content in residual melt during crystallization and degassing. Two cases are considered: (A) crystallization occurs first and degassing occurs subsequently, and (B) crystallization and degassing occur concomitantly. In both cases, the average Li content of steady-state basalts (5.85 ppm) is used for the starting composition. The target composition is that of the trachytic pumice (1 ppm Li). In case A, the Li content of melt (11.8 ppm) after differentiation and before degassing, is estimated assuming that Li displays no anomaly along trace element patterns when it is positioned between Tm and Yb. The fraction of melt having crystallized (Fcrystal) is inferred from the enrichment of the most incompatible elements, while the fraction of vapor lost (Fvapor) is constrained from the initial (assumed) and final (measured) water content. DS–M and DV–M are the crystal-melt and vapor-melt partition coefficients, respectively. Following Villemant and Boudon (1999), k is the mass ratio of crystal removed to vapor exsolved. See text for details.

A

5 0 -5

vapor accumulated

-10

vapor instantaneous

-15 0

2

4

6

8

10

12

14

Li in melt (ppm) Fig. 6. Lithium isotopic composition of the melt (upper panel) and vapor (lower panel) plotted against Li content in the residual melt. A and B refer to the two crystallizationdegassing scenarios described in Fig. 5. Upper panel: all points are data points, with the exception of the undegassed trachtytic composition which is calculated as explained in text. The straight dashed line, on which plot the pumice samples (0704-PS and 0704-PN), the coating glass (0704-BAS) as well as a bulk analysis not reported in Table 1 (Li = 3.36 ppm; δ7Li= −1.8‰ ± 0.3), suggests that the rock is made of two homogeneous components (basaltic and trachytic). Curved lines show the evolution of δ7Li in the residual melt during degassing (Eqs. (10) and (9) for cases A and B). Lower panel: vapor composition is calculated for cases A and B using Eq. (12) and Eq. (13), respectively).

Author's personal copy I. Vlastélic et al. / Chemical Geology 284 (2011) 26–34

The average isotopic ratio of the accumulated (ACC) vapor is obtained by integrating Eq. (11). Knowing that the average value of a function f(x) that is continuous over the interval a ≤ x ≤ b is 1 / (b–a).∫ f(x)dx, we infer:   V−M !M M M α 1− C =C0 Li   : 6 1− C M = C0M Li 0

!V Li = 6 Li ACC

7 6

Li Li

!M : 0

α

V−M

 1−F ðε +

1Þ

ðε + 1Þð1−F Þ



15

ð12Þ

with ε =

DV−M 6

  αV−M −1 1+k

ð13Þ

for cases A and B, respectively. During degassing of a finite batch of magma, the lithium isotopic composition of the accumulated vapor varies widely, from ~13.5‰ in the first vapors down to about ~5‰ in the case of extreme degassing, as for the April 2007 pumice (Fig. 6). The 7Li / 6Li ratio of the gas phase will subsequently increase during partial condensation, as suggested by the isotopically light Li composition of the April 2007 fumarolic condensates. While the coefficient of isotopic fractionation during condensation (αCond-V) is constrained to be less than unity, it cannot be determined precisely because both the isotopic composition of the condensing gas and the fraction of gas condensed are poorly known. To explain the composition of the April 2007 gas condensates (average of − 1.7‰), we calculated that αCond-V ≤ 0.99 is required if the condensing gas had δ7Li N 8‰, or if more than 50% of the vapor condensed. If the vapor had δ7Li = 13.5‰, then αCond-V ≤ 0.985. The trachytic pumice recorded an extreme process that is not representative of the degassing pattern of Piton de la Fournaise magmas, however. Steady state basalts display rather uniform Li content and isotopic composition (Supplementary Table 1), suggesting that lithium loss in normal degassing mode is of only a few percents and, consequently, that those volcanic emanations have very heavy Li isotopic signatures. The slightly heavier Li signature of oceanites (average of +3.9‰, Table 1) is due to the incorporation of cumulative olivine crystals having distinctly heavy composition (δ7Li = +5.2 ±1.4‰(2σ, N =10), Supplementary Table 1), which most likely reflect mineral-melt isotopic fractionation (Seitz et al., 2004) rather than a less degassed signature. Our attempt to find a distinctive lithium signature in naturally quenched samples (Pele's hairs) failed (Li =5.6 ±0.4 ppm, δ7Li = +3.4 ±0.3‰ (2σ, N =4), Supplementary Table 1). However, amongst lavas free of cumulative olivine, the highest δ7Li (+4.0 to 4.5‰) coincide with the eruption of more primitive, less degassed magmas (as during the 1998 Hudson eruption or during the paroxysmal phase of April 2007 eruption) suggesting that degassing may influence whole-rock Li signature to a measurable degree. We calculated that decreasing δ7Li from +4.5 to +3.6‰ (average of SSB) by degassing isotopically heavy lithium (αV–M of 1.010) is consistent with 9% loss of lithium with δ7Li of +14.1‰. 6. Concluding remarks The preliminary results obtained show that 7Li partitions preferentially into the vapor phase during low-pressure, vapor-saturated crystallization. This strongly suggests that Li isotopic fractionation is controlled by chemical rather than by kinetic effects. The composition of the degassed trachytic pumice places strong constraints on the magnitude of the vapor-melt isotopic fractionation (αV–M of 1.010) because the pre-degassing composition is relatively well known. Since the fraction of Li volatilized in normal degassing mode is very small, a few percent at the most, a very heavy Li isotopic signature (δ7LiN + 13‰) is expected in volcanic emanations. In addition, the Li composition of the vapor may subsequently increase as isotopically light lithium condensates. These findings and expectations are summarized in Fig. 7.

Condensation Cooled gas ?

20

7

and !V 7 Li = 6 Li ACC

Degassing 25

Exsolved fluids

10

δ7Li (‰)

7

33

5

High-temperature gas

Common lavas

0

Gas condensates 300-400°C

-5 -10 -15 -20

Vapor-differentiated lava

Measured Inferred

-25 Fig. 7. Summary of lithium isotope fractionation during magma degassing and gas condensation. Measured compositions are shown by squares while inferred compositions are shown by circles.

This behavior of lithium isotopes is consistent with that inferred for boron isotopes, but in magmatic systems only (You, 1994; Gillis et al., 2003; Kuritani and Nakamura, 2006). It is possible that Li, as with B (Rose-Koga et al., 2006), obeys somewhat different laws during evaporation and condensation processes in the ocean–atmosphere system. By comparison with heavier isotopic systems, Li isotopes seem to behave rather like Zn isotopes during degassing (preferential evaporation of heavy isotopes (Toutain et al., 2008)), but rather follow Tl isotopes during condensation (preferential removal of light isotopes (Baker et al., 2009)). However, because these isotopic systems were studied independently in different volcanic settings, it is unclear whether the contrasting isotope behaviors are fundamental or just reflect distinct degassing contexts. A more comprehensive view of trace isotopes behavior during degassing will probably require multi-isotopic approaches. Supplementary materials related to this article can be found online at doi:10.1016/j.chemgeo.2011.02.002. Acknowledgements The authors are grateful to A.J.R. Kent and an anonymous reviewer for their constructive comments, and to R. Rudnick for editorial handling. E. Rose-Koga, G. Falco and K. Koga are thanked for discussions. Thanks also to N. Bolfan-Casanova, C. Bosq, K. David, J.-L. Devidal, J.-L. Piro (LMV, Clermont-Ferrand), C. Douchet (LST, Lyon) and C. Le Losq (IPGP, Paris) for technical assistance in the lab. Thanks to SARM (CRPG, Nancy) and L. Bouvier (LaMP, Clermont-Ferrand) for providing major element analyses. This study benefited financial support from the Institut National des Sciences de l'Univers (ST, A02009 and AO2010). References Africano, F., Bernard, A., 2000. Acidic alteration in the fumarolic environment of Usu volcano, Hokkaido, Japan. Journal of Volcanology and Geothermal Research 97, 475–495. Albarède, F., 1993. Residence time analysis of geochemical fluctuations in volcanic series. Geochimica et Cosmochimica Acta 57, 615–621. Albarède, F., Luais, B., Fitton, G., Semet, M., Kaminski, E., Upton, B.G.J., Bachèlery, P., Cheminée, J.L., 1997. The geochemical regimes of Piton de la Fournaise volcano (Réunion) during the last 530 000 years. Journal of Petrology 38, 171–201. Baker, R.G.A., Rehkämper, M., Hinkley, T.K., Nielsen, S.G., Toutain, J.-P., 2009. Investigation of thallium fluxes from subaerial volcanism — implication for the present and past mass balance of thallium in the oceans. Geochimica et Cosmochimica Acta 73, 6340–6359.

Author's personal copy 34

I. Vlastélic et al. / Chemical Geology 284 (2011) 26–34

Beck, P., Barrat, J.A., Chaussidon, M., Gillet, P., Bohn, M., 2004. Li isotopic variations in single pyroxenes from the Northwest Africa 480 shergottite (NWA 480): a record of degassing of Martian magmas ? Geochimica et Cosmochimica Acta 68, 2925–2933. Berlo, K., Blundy, J., Turner, S., Cashman, K., Hawkesworth, C., Black, S., 2004. Geochemical precursors to volcanic activity at Mount St. Helens, USA. Science 306, 1167–1169. Blundy, J., Cashman, K.V., Rust, A., Witham, F., 2010. A case for CO2-rich arc magmas. Earth and Planetary Sciences Letters 290, 289–301. Bureau, H., Métrich, N., Semet, M., Staudacher, T., 1999. Fluid-magma decoupling in a hot-spot volcano. Geophysical Research Letters 23, 3501–3504. Giletti, B.J., Shanahan, T.M., 1997. Alkali diffusion in plagioclase feldspar. Chemical Geology 139, 3–20. Gillis, K.M., Coogan, L.A., Chaussidon, M., 2003. Volatile element (B, Cl, F) behaviour in the roof of an axial magma chamber from the East Pacific Rise. Earth and Planetary Sciences Letters 213, 447–462. Helz, R.T., Kirschenbaum, H.K., Marinenko, J.W., Qian, R., 1994. Whole-rock analyses of core samples from the 1967, 1975, 1979 and 1981 drillings of Kilauea Iki lava lake, Hawaii. U.S. Geological Survey Open-File Report 94–684. . Jambon, A., Semet, M.P., 1978. Lithium diffusion in silicate glasses of albite, orthoclase, and obsidian composition: an ion-microprobe determination. Earth and Planetary Sciences Letters 37, 445–450. Kent, A.J.R., Blundy, J., Cashman, K.V., Cooper, K.M., Donnelly, C., Pallister, J.S., Reagan, M., Rowe, M.C., Thornber, C.R., 2007. Vapor transfer prior to the October 2004 eruption of Mount St. Helens, Washington. Geology 35, 231–234. Koga, K.T., Cluzel, N., Rose-Koga, E., Laporte, D., Shimizu, N., 2008. Experimental demonstration of lithium-boron depletion during magma degassing. EOS. Transactions of the American Geophysical Union 89 (53) Fall Meeting Supplement, Abstract V31B-2143. Kuritani, T., Nakamura, E., 2006. Elemental fractionation in lavas during post-eruptive degassing: evidence from trachytic lavas, Rishiri Volcano, Japan. Journal of Volcanology and Geothermal Research 149, 124–138. Le Losq, C., Neuville, D.R., Morretti, R., Roux, J., Behrens, H. in press Determination of water content in silicate glasses using Raman spectrometry: implications for materials and volcanic activities. American Mineralogist. London, D., Hervig, R.L., Morgan, G.B., 1988. Melt-vapor solubilities and elemental partitioning in peraluminous granite-pegmatite systems: experimental results with Macusani glass at 200 MPa. Contribution to Mineralogy and Petrology 99, 360–373. Marschall, H.R., Pogge von Strandmann, P.A.E., Seitz, H.-M., Elliott, T., Niu, Y., 2007. The lithium isotopic composition of orogenic eclogites and deep subducted slabs. Earth and Planetary Sciences Letters 262, 563–580. Martin, E., Sigmarsson, O., 2007. Low-pressure differentiation of tholeiitic lavas as recorded in segregation veins from Reykjanes (Iceland), Lanzarote (Canary Islands) and Masaya (Nicaragua). Contributions to Mineralogy and Petrology 154, 559–573. Michael, P., 1995. Regionally distinctive source of depleted MORB: evidence from trace elements and H2O. Earth and Planetary Sciences Letters 131, 301–320. Millot, R., Guerrot, C., Vigier, N., 2004. Accurate and high-precision measurement of lithium isotopes in two reference materials by MC-ICP-MS. Geostandards and Geoanalytical Research 28, 153–159. Millot, R., Scaillet, B., Sanjuan, B., 2010. Lithium isotopes in island arc geothermal systems: Guadeloupe, Martinique (French West Indies) and experimental approach. Geochimica et Cosmochimica Acta 74, 1852–1871. Moune, S., Gauthier, P.-J., Gislason, S.R., Sigmarsson, O., 2006. Trace element degassing and enrichment in the eruptive plume of the 2000 eruption of Hekla volcano, Iceland. Geochimica et Cosmochimica Acta 70, 461–479. Peltier, A., Famin, V., Bachèlery, P., Cayol, V., Fukushima, Y., Staudacher, T., 2008. Cyclic magma storages and transfers at Piton de la Fournaise volcano (La Réunion hotspot) inferred from deformation and geochemical data. Earth and Planetary Sciences Letters 270, 180–188. Peltier, A., Bachèlery, P., Staudacher, T., 2009. Magma transport and storage at Piton de la Fournaise (La Réunion) between 1972 and 2007: a review of geophysical and geochemical data. Journal of Volcanology and Geothermal Research 184, 93–108. Richter, D.H., Moore, J.G., 1966. Petrology of the Kilauea Iki lava lake, Hawaii. U.S. Geological Survey Professional Paper 537B. . Richter, F.M., Davis, A.M., DePaolo, D.J., Watson, E.B., 2003. Isotope fractionation by chemical diffusion between molten basalt and rhyolite. Geochimica et Cosmochimica Acta 67, 3905–3923. Ringwood, A.E., 1959. Genesis of the basalt–trachyte association. Beiträge zur Mineralogie und Petrographie 6, 346–351. Rose-Koga, E.F., Sheppard, S.M.F., Chaussidon, M., Carignan, J., 2006. Boron isotopic composition of atmospheric precipitations and liquid-vapour fractionations. Geochimica et Cosmochimica Acta 70, 1603–1615. Rowe, M.C., Kent, A.J.R., Thornber, C.R., 2008. Using amphibole phenocrysts to track vapor transfer during magma crystallization and transport: an example from

Mount St. Helens, Washington. Journal of Volcanology and Geothermal Research 178, 593–607. Rubin, K., 1997. Degassing of metals and metalloids from erupting seamount and midocean ridge volcanoes: observations and predictions. Geochimica et Cosmochimica Acta 61, 3525–3542. Rudnick, R.L., Ionov, D.A., 2007. Lithium elemental and isotopic disequilibrium in minerals from peridotite xenoliths from far-east Russia: product of recent melt/ fluid-rock reaction. Earth and Planetary Sciences Letters 256, 278–293. Ryan, J.G., Langmuir, C.H., 1987. The systematics of lithium abundances in young volcanic rocks. Geochimica et Cosmochimica Acta 51, 1727–1741. Schiavi, F., Kobayashi, K., Moriguti, T., Nakamura, E., Pompilio, M., Tiepolo, M., Vannuci, R., 2010. Degassing, crystallization and eruption dynamics at Stromboli: trace element and lithium isotopic evidence from 2003 ashes. Contributions to Mineralogy and Petrology 159. doi:10.1007/s00410-009-441-2. Sakuyama, M., Kushiro, I., 1979. Vesiculation of hydrous andesitic melt and transport of alkalies by separated vapor phase. Contributions to Mineralogy and Petrology 71, 61–66. Schuessler, J.A., Schoenberg, R., Sigmarsson, O., 2009. Iron and lithium isotope systematics of the Hekla volcano, Iceland — evidence for Fe isotope fractionation during magma differentiation. Chemical Geology 258, 78–91. Seitz, H.-M., Brey, G.P., Lahaye, Y., Durali, S., Weyer, S., 2004. Lithium isotopic signatures of peridotite xenoliths and isotopic fractionation at high temperature between olivine and pyroxenes. Chemical Geology 212, 163–177. Sigmarsson, O., Condomines, M., Bachèlery, P., 2005. Magma residence time beneath the Piton de la Fournaise volcano, Réunion Island, from U-series disequilibria. Earth and Planetary Science Letters 234, 223–234. Sinton, J.M., Byerly, G.R., 1980. Silicic differentiates of abyssal oceanic magmas: evidence for late-magmatic vapor transport of Potassium. Earth and Planetary Sciences Letters 47, 423–430. Staudacher, T., Ferrazzini, V., Peltier, A., Kowalski, P., Boissier, P., Catherine, P., Lauret, F., Massin, F., 2009. The April 2007 eruption and the Dolomieu crater collapse, two major events at Piton de la Fournaise (La Réunion Island, Indian Ocean). Journal of Volcanology and Geothermal Research 184, 126–137. Tomascak, P.B., Tera, F., Helz, R.T., Walker, R.J., 1999. The absence of lithium isotope fractionation during basalt differentiation: new measurements by multicollector ICP-MS. Geochimica et Cosmochimica Acta 63, 907–910. Toutain, J.-P., Aloupogiannis, P., Delorme, H., Person, A., Blanc, P., Robaye, 1990. Vapor deposition of trace elements from degassed basaltic lava, Piton de la Fournaise volcano, Reunion Island. Journal of Volcanology and Geothermal Research 40, 257–268. Toutain, J.-P., Sonke, J., Munoz, M., Nonell, A., Polvé, M., Viers, J., Freydier, R., Sortino, F., Joron, J.-L., Sumarti, S., 2008. Evidence for Zn isotopic fractionation at Merapi volcano. Chemical Geology 253, 74–82. Vernet, M., Marin, L., Boulmier, S., Lhomme, J., Demange, J.C., 1987. Dosage du Fluor et du Chlore dans les matériaux géologiques y compris les échantillons hyperalumineux. Analysis 15, 490–498. Villemant, B., Boudon, G., 1999. H20 and halogen (F, Cl, Br) behaviour during shallow magma degassing processes. Earth and Planetary Science Letters 168, 271–286. Villemant, B., Mouatt, J., Michel, A., 2008. Andesitic magma degassing investigated through H2O vapour-melt partitioning of halogens at Soufrière Hills Volcano, Montserrat (Lesser Antilles). Earth and Planetary Science Letters 269, 212–229. Vlastélic, I., Staudacher, T., Semet, M., 2005. Rapid change of lava composition from 1998 to 2002 at Piton de la Fournaise (Réunion) inferred from Pb isotopes and trace elements: evidence for variable crustal contamination. Journal of Petrology 46, 79–107. Vlastélic, I., Peltier, A., Staudacher, T., 2007. Short-term (1998–2006) fluctuations of Pb isotopes at Piton de la Fournaise volcano (Réunion Island): Origins and constraints on the size and shape of the magma reservoir. Chemical Geology 244, 202–220. Vlastélic, I., Deniel, C., Bosq, C., Télouk, P., Boivin, P., Bachèlery, P., Famin, V., Staudacher, T., 2009a. Pb isotope geochemistry of Piton de la Fournaise historical lavas. Journal of Volcanology and Geothermal Research 184, 63–78. Vlastélic, I., Koga, K., Chauvel, C., Jacques, G., Télouk, P., 2009b. Survival of lithium isotopic heterogeneities in the mantle supported by HIMU-lavas from Rurutu Island, Austral Chain. Earth and Planetary Science Letters 286, 456–466. Webster, J.D., Holloway, J.R., Hervig, R.L., 1989. Partitioning of lithophile trace elements between H2O and H2O + CO2 fluids and topaz rhyolite melt. Economic Geology 84, 116–134. Wunder, B., Meixner, A., Romer, R.L., Feenstra, A., Schettler, G., Heinrich, W., 2007. Lithium isotope fractionation between Li-bearing staurolite, Li-mica and aqueous fluids: an experimental study. Chemical Geology 238, 277–290. You, C.-F., 1994. Comment on “Boron content and isotopic composition of oceanic basalts: geochemical and cosmochemical implications” by M. Chaussidon and A. Jambon. Earth and Planetary Science Letters 128, 727–730.