Polybaric degassing of island arc low-K tholeiitic basalt magma recorded by OH concentrations in Ca-rich plagioclase

Earth and Planetary Science Letters 308 (2011) 259–266

Contents lists available at ScienceDirect

Earth and Planetary Science Letters 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 / e p s l

Polybaric degassing of island arc low-K tholeiitic basalt magma recorded by OH concentrations in Ca-rich plagioclase Morihisa Hamada a, b,⁎, Tatsuhiko Kawamoto a, Eiichi Takahashi b, Toshitsugu Fujii c, d a

Institute for Geothermal Sciences, Kyoto University, Noguchibaru, Beppu 874-0903, Japan Department of Earth and Planetary Sciences, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551, Japan Earthquake Research Institute, University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0033, Japan d Crisis & Environment Management Policy Institute, #505, 1-22 Wakaba, Shinjuku-ku, Tokyo 160-0011, Japan b c

a r t i c l e

i n f o

Article history: Received 19 March 2011 Received in revised form 31 May 2011 Accepted 3 June 2011 Available online 24 June 2011 Editor: R.W. Carlson Keywords: water in nominally anhydrous minerals island arc basalt volcanic front melt inclusion excess degassing Izu arc

a b s t r a c t Hydrogen in nominally anhydrous minerals (NAMs) in volcanic rocks can be used as a proxy for dissolved H2O in melt prior to eruption. Plagioclase is a NAM that accommodates hydrogen in concentrations of up to hundreds of wt. ppm H2O. The species of hydrogen in volcanic plagioclase is structural OH. We report the analytical results of OH concentrations in Ca-rich plagioclase from the 1986–1987 summit eruption of IzuOshima volcano, a frontal-arc volcano in Izu arc. We demonstrate that the island arc low-K tholeiitic basalt magmas erupting from the frontal-arc volcanoes are H2O-saturated and undergo polybaric degassing during the magma ascent. The analyzed OH concentrations in plagioclase range from 20 to 300 wt. ppm H2O, and three levels of OH (20–80 wt. ppm H2O, 100–180 wt. ppm H2O, and 220–300 wt. ppm H2O) are found. These variations in OH indicate that crystallized plagioclase is equilibrated with H2O-saturated melt at three depths beneath the Izu-Oshima volcano prior to eruption: near the surface level (≈1 wt.% H2O in melt), at a 4-kmdeep magma chamber (≈3 wt.% H2O in melt), and at a 8–10-km-deep magma chamber (≈ 5 wt.% H2O in melt). It is proposed that deep-seated island arc low-K tholeiitic basalt magmas erupting from frontal-arc volcanoes are richer in H2O than previously thought, containing approximately 1 wt.% H2O based on analyses of “leaked” melt inclusions and phase equilibrium studies at “low-pressure conditions”. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Quantifying the H2O concentration in arc magmas is critical for understanding their generation, differentiation, and eruption because H2O has profound effects on melting temperatures, crystallization pathways, and explosivity. Among arc magmas, island arc low-K tholeiitic basalts from frontal-arc volcanoes are generally thought to be H2Oundersaturated and to contain approximately 1 wt.% H2O based on melt inclusion analyses (Kazahaya et al., 1994; Saito et al., 2005) and phase equilibrium studies (Grove and Baker, 1984; Kawamoto, 1996; Sakuyama, 1979). In contrast to these previous petrological studies, recent melt inclusion analyses report higher H2O concentrations in low-K tholeiitic basalts obtained from subaqueous Izu arc volcanoes (up to 5 wt.% by Straub and Layne, 2003) and Miyakejima volcano (up to 3.5 wt.% by Saito et al., 2010). At Izu-Oshima volcano, a frontal-arc volcano in Izu arc, the H2O concentration in plagioclase-hosted melt inclusions is b2 wt.% regardless of anorthite content of the host plagioclase (An87–96; Hamada

⁎ Corresponding author at: Department of Earth and Planetary Sciences, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551, Japan. Tel./fax: + 81 3 5734 3636. E-mail address: [email protected] (M. Hamada). 0012-821X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2011.06.005

and Fujii, 2007), while olivine-hosted melt inclusions contain higher H2O concentrations (up to 3.4 wt.%; Ikehata et al., 2010). There remains some uncertainty in our understanding of the H2O abundances of island arc basalts. Both Hamada and Fujii (2007) and Ikehata et al. (2010) reported that the CO2 concentration in all the melt inclusions from Izu-Oshima volcano is below the detection limit (b50 ppm), and therefore, variations in H2O concentration are not due to CO2 fluxing from the depths, a process identified at many arc basaltic volcanoes (Johnson et al., 2008; Spilliaert et al., 2006). Melting experiments of hydrous basalts have demonstrated that Ca-rich plagioclase (approximately An90) crystallizes from H2O-rich (≥3 wt.%) melt (Sisson and Grove, 1993a; Takagi et al., 2005). From these results, Hamada and Fujii (2007) assumed that the melt is originally H2Orich and that H2O vapor in trapped melt inclusions leaks through microcracks in the plagioclase phenocryst at shallow levels. This assumption can explain the low (b2 wt.%) concentrations of H2O and the undetectably low concentrations of CO2 in the plagioclase-hosted melt inclusions. However, this assumption also suggests that the H2O concentration in melt inclusions does not always represent the H2O concentration at the time the melt was trapped. An alternative method to constrain the H2O concentration in preeruptive melt is to analyze hydrogen in nominally anhydrous minerals (NAMs). The hydrogen in NAMs can be an indicator of H2O activity in melts and can act as a speedometer of the magma ascent (Demouchy

260

M. Hamada et al. / Earth and Planetary Science Letters 308 (2011) 259–266

et al., 2006; Peslier and Luhr, 2006; Peslier et al., 2008). Plagioclase is one of those NAMs, and it accommodates hydrogen up to hundreds of wt. ppm H2O. The speciation of hydrogen in plagioclase from volcanic rocks has been demonstrated to be structural OH (Johnson and Rossman, 2003, 2004). In this study, we analyzed the OH concentration in plagioclase from the Izu-Oshima summit eruptions during 1986–1987 using a Fourier transform infrared spectrometer (FTIR) to estimate the H2O concentration in the island arc low-K tholeiitic basalts. Plagioclase is easy to collect because it is the most abundant phenocryst in erupted magmas (up to 15 vol.%). The question remains of whether analyzed OH in plagioclase is intrinsic or already diffused out during eruption. Limited experiments on the diffusivity of hydrogen in feldspars have clarified that the diffusivity values range between 10 − 11 m 2/s at 900 °C and 10 − 15 m 2/s at 500 °C (K-feldspar; Kronenberg et al., 1996) or between 10 − 13 m 2/s at 1000 °C and 10 − 14 m 2/s at 800 °C (An30 plagioclase; Johnson, 2003), which allows the hydrogen in feldspars to diffuse out over periods of hours to days. To quantify this effect, we also analyzed OH in plagioclase from three distinctive types of summit eruptions of Izu-Oshima volcano during 1986–1987: the Strombolian eruption in 1986, an effusive eruption emitting lava flow in 1986, and a less energetic eruption in 1987. The variation in OH in plagioclase might be related to these eruptive styles. 2. Outline of the Izu-Oshima 1986–1987 summit eruptions The volcanic rocks of Izu-Oshima volcano are dominated by low-K tholeiitic basalts with minor andesites. During the 1986–1987 eruptions, basaltic magmas erupted from the summit vent and andesitic magmas erupted from two fissure vents in the caldera and on the outer flank. The total volume of erupted magmas was 0.053 km 3 (0.032 km 3 DRE; Endo et al., 1988). In this study, we focus on the basaltic eruptions from the summit vent. The summit eruption in 1986 was marked by two distinctive sequences: (1) Strombolian activity (November 15th, 1986, and thereafter) and (2) the emission of lava flow (November 18th, 1986 and thereafter). A continuous magma supply formed a lava lake on the summit vent, which eventually spilled as lava flows on November 18th and

thereafter. The lava flows descended down the edifice of the volcano and stopped moving within 3 days. After the 1986 eruptions, the magma head retreated into the conduit, while the level of the summit's lava lake remained steady for a year until a summit eruption on November 16th, 1987. This eruption was less energetic and was triggered by a gas explosion in the conduit, and it generated flaky bombs (Watanabe et al., 1998). The volume of erupted magma on November 1987 was 0.05% of the total erupted volume of magmas during 1986–1987 (Endo et al., 1988). Geophysical studies detected two major magma chambers beneath the Izu-Oshima volcano: a 4-km-deep magma chamber (Ida, 1995) and an 8–10-km-deep magma chamber (Mikada et al., 1997). In addition, approximately 0.011 km 3 of magma was stored in the summit's lava lake (300 m in diameter and 200 m in depth) during the 1986–1987 eruptions. The summit's lava lake and the 4km-deep magma chamber were thought to be connected through a conduit based on observations of crustal deformation in November, 1987 (Ida, 1995). 3. Studied samples We analyzed the OH concentration in plagioclase from three distinctive eruptions to relate the eruptive styles to the variations in OH in plagioclase: air-fall scoria from the Strombolian eruption in 1986, lava flow from an effusive eruption in 1986, and flaky bombs from the eruption in 1987. The compositions of the air-fall scoria from the 1986 eruption and the flaky bombs from the 1987 eruption were analyzed with an X-ray fluorescence spectrometer (RIGAKU system 3070) at the Institute for Geothermal Sciences (IGS), Kyoto University, using a calibration line optimized for a wide compositional range (Sugimoto et al., 2007). The operating conditions were 15 kV and 5.0 × 10 − 8 A. The data were corrected using a ZAF correction. Although the whole-rock major element compositions and the modal compositions of the three rocks are similar (Table 1), their groundmass textures are different. The 1986 scoria exhibits a glassy groundmass and small (b100 μm) bubbles (Fig. 1a), while the 1986 lava flow has plagioclase characterized by a b10-μm anorthite-poor rim (An70 to An75) on plagioclase phenocrysts and microcrystalline

Table 1 Whole-rock major element compositions and modal compositions of volcanic rocks from the summit eruptions of Izu-Oshima volcano during 1986–1987. The composition of lava flow (LA-III) is from Nakano and Yamamoto (1987). 1986 eruption

Major element compositions (wt.%) SiO2 TiO2 Al2O3 FeO* MnO MgO CaO Na2O K2O P2O5 Cr2O3 NiO Raw total

1987 eruption

Air-fall scoria

LA III lava*

Flaky bomb

53.6 1.20 14.7 13.1 0.22 4.70 10.2 1.84 0.38 0.10 0.01 0.01 99.1 (wt.%)

53.2 1.24 15.2 12.8 0.20 4.67 10.4 1.88 0.41 0.08 – – 99.6 (wt.%)

52.6 1.12 15.5 13.0 0.21 4.60 10.6 1.86 0.38 0.10 0.04 0.02 100.5 (wt.%)

Modal compositions (vol.%) of phenocrysts (N 0.1 mm) Plagioclase 5–10 Clinopyroxene b1 Orthopyroxene b1

10–15 b1 b1

Groundmass phases Glass Plagioclase

Plagioclase Clinopyroxene Magnetite

Glass Plagioclase Clinopyroxene Magnetite

M. Hamada et al. / Earth and Planetary Science Letters 308 (2011) 259–266

261

100µm

plag bubble

plag plag

bubble pig glass

a

mgt pig

b

mgt glass

c

Fig. 1. Backscattered electron images of volcanic rocks from the Izu-Oshima 1986–1987 summit eruptions: (a) air-fall scoria from the Strombolian eruption in 1986, (b) LA-III lava flow in 1986, and (c) flaky bomb from an eruption in 1987.

groundmass (Fig. 1b). The groundmass of the 1986 lava flow is composed of acicular plagioclase, clinopyroxene (mostly pigeonite) and magnetite. The scoria has larger bubbles than those of the lava flow, and most of them grow and coalesce to a diameter of up to 5 mm. These features are consistent with the fact that the scoria was quenched in the air, while the lava flow cooled slowly. The 1987 bomb exhibits a groundmass composed of glass, acicular plagioclase, pigeonite (up to 20 μm, which is larger than that in the 1986 lava flow) and magnetite (Fig. 1c). These features suggest that the bomb remained molten in or below the summit's lava lake for a year until the eruption on November 16th, 1987. In the three types of studied rocks, the length of plagioclase phenocrysts ranges from 200 μm to 2000 μm. We analyzed the OH concentrations in relatively larger (≥1000-μm-long) plagioclase. The major element compositions of plagioclase in the 1986 scoria and the 1987 bomb were analyzed using a JEOL JSM-5310 scanning electron microscope with an energy dispersive X-ray spectrometer (SEM-EDS) at the IGS, Kyoto University. The operating conditions were 15 kV and 5.0 × 10 − 10 A. Plagioclase in the 1986 lava flow was analyzed using a JEOL JXA-8800 electron probe micro analyzer (EPMA) at the Department of Earth and Planetary Sciences, Tokyo Institute of Technology. The operating conditions were 15 kV and 1.2 × 10 − 8 A. The values of the anorthite content of plagioclase in the 1986 scoria range from An85 to An93, those in the 1986 lava flow range from An83 to An90, and those in the 1987 bomb range from An81 to An89. Plagioclase with a lower anorthite content (bAn85) is found only in the 1986 lava flow and the 1987 bomb, and plagioclase with a higher anorthite content (NAn91) is found only in the 1986 scoria. Normal or reversal zonings within ±1 anorthite content are commonly observed in the crystal's core. 4. Analytical methods 4.1. Unpolarized mid-infrared measurements across plagioclase crystals The OH concentration in plagioclase crystals can be quantified by polarized mid-infrared spectroscopy (Johnson and Rossman, 2003, 2004), which requires analyses from three mutually perpendicular polarization directions or along each crystallographic axis. For this purpose, each plagioclase rim must be polished as a rectangular parallelepiped without a rim. A disadvantage of this technique is that only the OH concentration in the plagioclase core can be measured; the OH concentration in the removed plagioclase rim cannot be quantified. To assess the distribution of the OH concentration in the crystals, we also analyzed the rim-to-rim profiles of the unpolarized mid-infrared absorbance band area per 1 cm for three doublypolished plagioclase crystals from the 1986 scoria (samples #pl255, #pl264 and #pl265). The mid-infrared absorbance per unit thickness

was analyzed using a Jasco FTIR-6100 Fourier transform infrared (FTIR) spectrometer equipped with an infrared microscope (Jasco IRT5000) at the Department of Earth and Planetary Sciences, Tokyo Institute of Technology. This spectrometer is equipped with a heated ceramic (globar) source, Ge-coated KBr beam splitter, and N2-cooled MCT detector. The polarizer was not equipped with this FTIR. The background and sample spectra (500–7000 cm − 1) were collected over 128 scans at a resolution of 4 cm − 1. A rectangular 40 × 40 μm aperture was used. 4.2. Polarized mid-infrared measurements of the plagioclase core Relatively large (≥1 mm long), tabular crystals of plagioclase with or without minimal amounts of melt inclusions were handpicked from the crushed volcanic rocks, and the OH concentration in plagioclase was analyzed by polarized mid-infrared spectroscopy following the procedure given by Johnson and Rossman (2003, 2004). For this purpose, four mutually perpendicular planes of plagioclase were polished. The thickness of the polished plagioclase varied from 120 to 500 μm. The OH concentration in plagioclase was analyzed using a Jasco-610 FTIR equipped with a Jasco Micro-20 IR microscope at IGS, Kyoto University. This spectrometer is equipped with a halogen lamp, CaF2 beam splitter, CaF2 polarizer, and N2-cooled InSb detector. The background and sample spectra (2000–10000 cm − 1) were the accumulation of 128 scans at a resolution of 4 cm − 1. A rectangular aperture ranging from 40 × 40 μm to 100 × 100 μm was used according to the size of the polished plagioclase. Melt inclusions in the polished plagioclase were avoided in the infrared beam path under the microscope. The sum of three absorbance areas measured at random, but with mutually perpendicular polarization directions per unit thickness were converted to the OH concentration by applying the absorbance coefficient of 15.3 ± 0.7/(ppm H2O × cm 2) (Johnson and Rossman, 2003). To relate the mid-infrared spectra and the crystallographic directions, the crystallographic directions of the selected plagioclase, #pl223, were determined using electron back-scattering diffraction (EBSD) patterns produced by a JEOL JSM-7000F field emission scanning electron microscope (FE-SEM) at the Department of Earth and Planetary Science, University of Tokyo. The operating conditions were 15 kV and 8.0 × 10 − 12 A. 5. Analytical results 5.1. Unpolarized mid-infrared measurements across plagioclase crystals The unpolarized mid-infrared band area per unit thickness across the crystal was analyzed on three plagioclases from the 1986 scoria: samples #pl255 (Fig. 2a), #pl264 (Fig. 2b) and #pl265 (Fig. 2c). The

262

M. Hamada et al. / Earth and Planetary Science Letters 308 (2011) 259–266

An82 An89

An88 An83

An85

Absorbance (cm-2)

a

Distance (µm)

An88

b

Distance (µm)

c

Distance (µm)

Fig. 2. Backscattered electron images (BEIs) and line profiles of the unpolarized absorbance band area of OH in plagioclase per 1 cm thickness for (a) #pl255, (b) #pl264 and (c) #pl265. Squares in BEIs (40 × 40 μm) represent the analyzed area of OH by an FTIR.

continuous decrease from level 2, 100–180 wt. ppm H2O through level 3, 50–100 wt. ppm H2O to level 4, 20–50 wt. ppm H2O (Fig. 4a). The OH concentrations in plagioclase from the 1986 air-fall scoria span the whole range and are strictly divided into three different levels (level 1, level 2 and levels 3–4). The OH concentrations from the 1986 lava flow are the lowest (levels 3–4), and those from the 1987 flaky bomb

1.9

1.5

peak 1

1.6

peak 2

peak 3

1.7

1.4 1.3

Absorbance per mm Absorbance per mm

5.2. Polarized mid-infrared measurements of the plagioclase core Representative mid-infrared spectra of plagioclase (sample #pl223) are shown in Fig. 3. These are the integrated spectra of three peaks (≈3200 cm− 1, ≈3400 cm− 1, and ≈3600 cm− 1; Nakano et al., 2001) centered at approximately 3300 cm− 1. The absence of an absorption band of molecular H2O and OH in the silicate glass at approximately 3550 cm− 1 (Stolper, 1982) suggests that the dissolved H2O in tiny (b10 μm) melt inclusions hosted by plagioclase does not contribute to the absorption spectra of plagioclase. Therefore, the absorbance spectra shown in Fig. 3 are judged to be the OH accommodated in plagioclase. The intensity of absorption is at a maximum when the electric vector, E, is nearly parallel to the a-axis of plagioclase (Fig. 3); this result is consistent with that obtained in a previous study (Johnson and Rossman, 2003). A weak absorption band of approximately 2900 cm− 1 can be attributed to residual resin, which is sticky and cannot be completely removed using acetone. The OH concentrations in plagioclase crystals from three distinctive types of eruptions range from 20 to 300 wt. ppm H2O (Fig. 4a). The estimated errors on the H2O concentrations, ±10–15%, include uncertainties in the thickness measurement, absorbance area calculation, and absorption coefficient. The OH concentration can be categorized into four levels: level 1, 220–300 wt. ppm H2O and a

E//~a axis residual resin?

1.8

Absorbance per mm

compositions of the cores of the three plagioclase crystals are An88–89. The absorbance band areas of the three plagioclase crystals per 1 cm range from 300 through 600 cm − 2 (Fig. 2). The OH concentrations in plagioclase and the unpolarized mid-infrared band areas on plagioclase per unit thickness broadly correlate, and a value of 300– 600 cm − 2 corresponds to approximately 100 ppm H2O by weight if the spectra are on (001) (Johnson and Rossman, 2003). The absorbance per unit thickness of the plagioclase core is relatively uniform throughout the crystal within the error or is slightly higher than that of the rim (Fig. 2). Therefore, we infer that the OH concentrations in the plagioclase cores analyzed by polarized infrared spectroscopy (Johnson and Rossman, 2003, 2004) represent the OH concentration in plagioclase itself. The absorbance near the crystals' edges cannot be analyzed because of the numerous melt inclusions (Fig. 2a and b) and the divergence of the infrared beam passing through the sample (Kawamoto et al., 2003). The absorbance per unit thickness of the sodic rim (An85) is lower than that of the core (An89) (Fig. 2b), which suggests crystallization of the sodic rim from melt containing lower H2O.

a

1.2 1.6 1.5

E//~b axis

1.4 1.3

b

1.2 1.2 1.1

E//~c axis

1.0 0.9 0.8 4000

c 3500

3000

2500

Wavenumbers (cm-1) Fig. 3. Representative mid-IR spectra of plagioclase (#pl223, ≈250 wt. ppm H2O) normalized to 1 mm thickness. The polarization directions of the infrared beam are mutually perpendicular and nearly parallel to the a-, b-, and c-axes determined using EBSD patterns. Peak 1 (≈3200 cm− 1), peak 2 (≈3400 cm− 1), and peak 3 (≈3600 cm− 1) are defined by Nakano et al. (2001).

M. Hamada et al. / Earth and Planetary Science Letters 308 (2011) 259–266

263

1986 air-fall scoria 1986 lava flow 1987 flaky bomb

OH in plagioclase (wt. ppm H2O)

300 level 1

250 200 150

level 2

100 level 3 50

a

0

level 4

OH in plagioclase (wt. ppm H2O)

300 level 1 8~10-km deep magma chamber

A1

250 200

loss of hydrogen

B2

A2

level 2 ~4-km deep magma chamber

C3

B3

A3

level 3

C4

B4

A4

150 100 50 0 80

85

b

90

level 4 summit lava lake or surface

95

An content of plagioclase Fig. 4. (a) The OH concentration versus anorthite (An) content of plagioclase from the Izu-Oshima 1986–1987 summit eruptions. The OH concentration is categorized into four levels: level 1, 220–300 wt. ppm H2O; level 2, 100–180 wt. ppm H2O; level 3, 50–100 wt. ppm H2O; and level 4, 20–50 wt. ppm H2O. (b) Interpretation of the analytical results shown in (a). The thick solid curve crossing from top right to bottom left shows correlation between OH concentration in plagioclase and anorthite content of plagioclase calculated using Eq. (2) at 0.3 GPa, where the applied partition coefficient of hydrogen between plagioclase and melt is 0.004, Ca/Na = 3.0 and Al2O3/SiO2 = 0.28.

are continuous between level 3 and level 4 (Fig. 4a). No relation between different OH concentration levels and plagioclase textures is recognized. Assuming that the partition coefficient of OH between plagioclase and melt is 0.004 on a molar basis (Johnson, 2005), plagioclase crystals falling into levels 1, 2 and 4 equilibrate with hydrous melts containing ≈7 wt.%, ≈ 4 wt.% and ≈ 1 wt.% H2O, respectively. 6. Discussion 6.1. Timescale of hydrogen diffusion in plagioclase Hydrogen in plagioclase can diffuse out and re-equilibrate with H2O in a degassing melt during the magma ascent. At 1100 °C, which is a temperature calculated by applying pyroxene geothermometry to basalts from the Izu-Oshima 1986 summit eruption (Fujii et al., 1988), the estimated diffusivity of hydrogen is of the order of 10 − 12 m 2/s based on extrapolation from the diffusivity determined at lower temperatures. The diffusive loss of hydrogen from plagioclase may be modeled using a simple one-dimensional model, where a 1-mm-long plagioclase crystal containing homogeneous OH is surrounded by an infinite anhydrous melt, and hydrogen diffuses out in one-dimension as a function of time, t (Carslaw and Jaeger,

1959; Shewmon, 1983). C, the OH concentration from the crystal edge X, is expressed as C=


     4CO ∞ 1 ð2j + 1ÞπX ð2j + 1Þπ 2 sin × exp − ∑ Dt ð1Þ h h π j = 0 2j + 1

where CO is the initial OH concentration, h is the length of the crystal (h = 1 mm = 10 − 3 m), and D is the diffusion coefficient of hydrogen (D = 10 − 12 m 2/s). The calculated results show that approximately 10% of the OH is lost at the core of a 1-mm-long plagioclase after 6 h (Fig. 5). Half of the OH will be lost after 24 h, and most of the hydrogen is lost after 72 h (Fig. 5). Modeling in two- or three-dimensions should give similar results because the diffusivity of hydrogen in plagioclase was found to be isotropic based on the dehydration experiments on plagioclase (Johnson, 2003). At the onset of the 1986 summit eruption, the discharge rate of magma exceeded 2 × 10 5 m 3/h (Endo et al., 1988). The estimated radius of the conduit is N4.7 m based on the analysis of magma drainback from the summit's lava lake to the conduit in November of 1987 (Kazahaya et al., 1994). We then calculated that the magma ascent rate at the onset of the 1986 eruption must have been of the order of 10 3 m/h. It likely took only a few hours for the magma to ascend from the 4-km-deep magma chamber to the surface and several hours to

264

M. Hamada et al. / Earth and Planetary Science Letters 308 (2011) 259–266

D = 10-12 m2/s (T

Normalized OH concentration

1.0

t=1h

plagioclase from degassed melt or the re-equilibration of plagioclase with degassed melt in the conduit at shallow level.

t=6h

6.3. Possible scenario for variation of OH concentration in plagioclase The following equation gives the relationship between anorthite content and the OH concentrations in plagioclase (the solid curve in Fig. 4b). The Ca/Na molar ratio of plagioclase (An≥ 70) crystallizing from the hydrous basaltic melt is empirically formulated by Hamada and Fujii (2007) as

0.8

0.6 t = 24 h




0.4

Þ

t = 48 h

0.2

t = 72 h

0

0

0.2

0.4

ð


 Ca Ca 4100 P ðGPaÞ −800 × = × exp Na plagioclase Na melt T ðK Þ T ðK Þ ! qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi melt Al2 O3 ðwt:%Þ + 0:33 × H2 Omelt ðwt:%Þ + 2:2 × ln melt SiO2 ðwt:%Þ

0.6

0.8

1.0

Distance (mm) Fig. 5. Calculated one-dimensional diffusion profiles of hydrogen across 1-mm-long plagioclase using Eq. (1) for the first five terms (j = 0, 1, 2, 3 and 4). The values used in Eq. (1) are D = 10− 12 m2/s and h = 10− 3 m. The normalized OH concentration is given by C/CO.

ascend from the 8–10-km-deep magma chamber, if the radius of the conduit is assumed to be constant. The relatively uniform profiles of the unpolarized mid-infrared band area per unit thickness across the crystals (Fig. 2) also support the diffusive loss of hydrogen for short duration, such as 1 h (Fig. 5). Allowing that ascending magma is not completely anhydrous, the timescale of the diffusive loss of hydrogen should be longer than that simulated in Fig. 5. We expect that most of the OH in plagioclase from the air-fall scoria can be used as a proxy for dissolved H2O in deep-seated melt just prior to eruption. In the same manner, most of the OH in plagioclase from the lava flow is expected to be lost at 1 bar (anhydrous condition) after 3 days (=72 h), although the value of D will decrease by several orders of magnitude in the cooling lava flow, resulting in a longer timescale of hydrogen diffusion. Distinctive eruptive styles, i.e., explosive or effusive eruptions, should be a reasonable explanation for the systematics of the OH concentrations in plagioclase analyzed in this study. 6.2. H2O-saturated magma chambers inferred from variations of OH in plagioclase We postulate that the magmas from the Izu-Oshima 1986–1987 summit eruptions were saturated with H2O at various depths. Three different levels of OH concentration in plagioclase (levels 1, 2 and 4 in Fig. 4a) can indicate that plagioclase re-equilibrates with H2Osaturated melt in the 8–10-km-deep magma chamber (≈5 wt.% H2O in melt), in the 4-km-deep magma chamber (≈3 wt.% H2O in melt), and near the surface level (≈1 wt.% H2O in melt), respectively (Fig. 4b). Level 3, being intermediate between levels 2 and 4, represents the saturated H2O concentration in the conduit between the 4-km-deep magma chamber and the summit's lava lake. The air-fall scoria from the 1986 eruption can be interpreted as a mixture of quenched magmas at depths between the 8–10-km-deep magma chamber and the surface level during eruption, because the OH concentration of plagioclase span levels 1–4 (Fig. 4a). The OH concentration in plagioclase from the 1986 lava flow falls into levels 3 and 4 (Fig. 4a). Plagioclase in level 3 may not have completely lost hydrogen by diffusion at 1 bar. Plagioclase in levels 3 and 4 from the 1987 bomb (Fig. 4a) can be interpreted as the crystallization of

ð2Þ :

The temperature T (K) is calculated using an empirical thermometer for multiple-saturated melts (Sisson and Grove, 1993b), and it ranges from 1200 °C (0 wt.% H2O) to 1000 °C (6 wt.% H2O). We fix the pressure at 0.3 GPa to dissolve H2O up to 6 wt.% in melt. Calculating the Ca/Na ratio of plagioclase using Eq. (2) at 0.3 GPa and at a changing pressure from 0.3 to 0 GPa to simulate the magma ascent produces insignificant differences. The applied partition coefficient of hydrogen between plagioclase and melt is 0.004 (Johnson, 2005). The element ratios for the melt composition are assumed to be those of the whole-rock composition of basalts from the 1986–1987 summit eruptions (Ca/Na = 3.0, Al2O3/SiO2 = 0.28; Table 1). We define a number of regions in Fig. 4b. Regions A1, B2 and C3 are defined by matching the solid curve calculated from Eq. (2), which suggests that the plagioclase crystals in these regions preserve the OH concentrations acquired at the time of their crystallization. Plagioclase then re-equilibrates with the degassing melt at shallower depths during the magma ascent, losing OH from region A1 to regions A2, A3 and A4; from region B2 to regions B3 and B4; and from region C3 to region C4 (Fig. 4b). We now propose a scenario to explain the analyzed range of OH concentrations (Fig. 4a). In this scenario, plagioclase crystallized from H2O-saturated melts in an 8–10-km-deep magma chamber (region A1 in Fig. 4b) and in a 4-km-deep magma chamber (region B2 in Fig. 4b) (stage 1 in Fig. 6). Some of the plagioclase crystals that crystallized in the 8–10-km-deep chamber ascended and lost some portion of their hydrogen due to the re-equilibration with the surrounding degassed melt in the 4-km-deep magma chamber (regions A1 to A2 in Fig. 4b) (stage 2 in Fig. 6). An injection of hot magma triggered a Strombolian eruption (stage 3 in Fig. 6). The plagioclase in regions A3 and A4 was derived from A1 and/or A2 as a result of an incomplete diffusive loss of hydrogen near the surface level. The plagioclase in regions A1, A2, A3, A4 and B2 was brought to the surface, and was quenched without further loss of hydrogen because the timescale of the magma ascent was within several hours, which was a too short duration for the plagioclase crystal to further re-equilibrate with the degassed melt (Fig. 5). After the Strombolian eruption, the lava flows spilled from the summit's lava lake, descended down the edifice of the volcano, and stopped moving within 3 days. The slower cooling of magma in the atmosphere caused a continuous loss of hydrogen from plagioclase resulting in regions A3, A4, B4 and C4 (stage 4 in Fig. 6). During the interval between the 1986 and 1987 eruptions, the magma head retreated into the conduit while lava was left in the summit crater, which formed a gas column in the conduit beneath the summit's lava lake. In November of 1987, a gas explosion in the conduit triggered an eruption. The plagioclase from the 1987 bomb exhibited OH concentrations and anorthite contents falling within regions A3, B3, B4, C3 and C4 (Fig. 4b). Such variable OH concentrations in the plagioclase from the 1987 bomb would be produced by crystallization

M. Hamada et al. / Earth and Planetary Science Letters 308 (2011) 259–266

265

A0+A1+A2+A3+B0

A2+A3+B2+C1

A1 A 2 A3 Diffusive loss of hydrogen

4 km

A1+B0

B0

Ascent of plagioclase

plagioclase

8 km

A0

A0

10 km Injection of hot magma

stage

(1)

(2)

(3)

(4)

Fig. 6. Schematic illustrations of magma movements during the summit eruptions of Izu-Oshima volcano in 1986. (1) Plagioclase crystallized from the H2O-saturated melt at each depth. (2) Some plagioclase crystals were entrained in the ascending magma to a shallower magma chamber and lost some portions of their hydrogen due to the re-equilibration with the surrounding degassed melt. (3) Injection of hot magma triggered the Strombolian eruption on November 15th and thereafter. (4) Lava spilled from the summit's lava lake and formed lava flow on November 18th and thereafter.

and re-equilibration in the conduit between the 4-km-deep magma chamber and near the surface level. 6.4. Implications for island arc magmatism The observed variation in OH in Ca-rich plagioclase demonstrates that the island arc low-K tholeiitic basalt magma is H2O-saturated at depths where Ca-rich plagioclase crystallizes, and it degasses during the magma ascent. This assumption is contrary to consensus, which presumes that the magma is H2O-undersaturated and contains approximately 1 wt.% H2O based on melt inclusion analyses (Kazahaya et al., 1994; Saito et al., 2005) and phase equilibrium studies (Grove and Baker, 1984; Kawamoto, 1996; Sakuyama, 1979). We suggest that the H2O-saturated nature of the deep-seated magmas beneath frontalarc volcanoes has not been detected due to polybaric degassing prior to eruption. The analyzed H2O concentration in melt inclusions could represent the H2O concentration in degassed melts, and the phase equilibria of erupted volcanic rocks could represent those of degassed magmas. A polybaric degassing model proposed above will reform models on the genesis of arc magmas assuming the H2O-undersaturated magmas by Tatsumi et al. (1983). They carried out melting experiments of synthesized primary magmas under dry and hydrous conditions (H2O ≤ 3 wt.%) and constrained the multiple saturation point, where melt is saturated with olivine+ orthopyroxene+ clinopyroxene on its liquidus. They determined that the multiple saturation point of primary olivine tholeiitic basalt magma, which is equivalent to primary low-K tholeiitic basalt magma, is 11 kbar and 1320 °C under dry conditions. Based on experiments, they assumed there to be a region exceeding 1400 °C in the mantle wedge beneath Northeast Japan arc, which is higher than the calculated maximum temperature in the mantle wedge

beneath the Northeast Japan arc by thermal modeling studies (1200– 1250 °C by Iwamori, 2000; Peacock and Wang, 1999). The P–T condition of the multiple saturation point evolves toward lower temperatures and higher pressures with increasing H2O in melt (Pichavant et al., 2002; Tatsumi et al., 1983). Therefore, a plausible multiple saturation point of H2O-rich island arc low-K tholeiitic basalt magma should be at a lower temperature and higher pressure than P (11 kbar) and T (1320 °C) estimated by Tatsumi et al. (1983). If we assume a polybaric degassing model, we can explain the socalled “excess SO2 degassing” (Shinohara, 2008; Wallace, 2001), which is an observation that the mass of emitted SO2 generally exceeds the mass of SO2 dissolved in erupted magma. The excess degassing of SO2 from Izu-Oshima volcano was once explained by degassing in a magma column convecting at a shallow level (b50 MPa), assuming 1 wt.% H2O in melt, based on analyses of melt inclusions (Kazahaya et al., 1994). Instead of the model proposed by Kazahaya et al. (1994), it may be plausible to assume that SO2 exsolves into saturated H2O bubbles and degasses effectively, as experimental studies demonstrate that sulfur preferentially partitions into bubbles as SO2 in oxidizing (fO2 N Ni-NiO buffer) arc magmas (Keppler, 1999, 2010; Sharma et al., 2004; Wallace, 2001). We suggest that more H2O is brought from the mantle wedge to the surface than the previously estimated flux of H2O from arc volcanoes, e.g., 2 to 4 × 10 13 mol/yr (Fischer, 2008), based on analyses of primitive melt inclusions. 7. Conclusions Ca-rich plagioclase from the 1986–1987 summit eruptions of IzuOshima volcano contains trace amounts of OH ranging from 20 to 300 ppm H2O by weight. The OH concentrations of plagioclase are affected by eruptive styles such as Strombolian and effusive eruptions,

266

M. Hamada et al. / Earth and Planetary Science Letters 308 (2011) 259–266

because they determine the quench rate of magmas. The variation in OH in plagioclase can be reasonably explained by the hypothesis that magma is saturated with H2O at depths and undergoes degassing at shallower levels, where the plagioclase re-equilibrates with degassed, lower-H2O melt prior to eruption. The presence of such H2O-saturated island arc low-K tholeiitic basalt magma is consistent with the constraints obtained from melting experiments of hydrous basaltic magmas to crystallize Ca-rich plagioclase. The polybaric degassing model of H2O-saturated magmas also explains the so-called “excess SO2 degassing” without assuming huge volumes of unerupted magma beneath arc volcanoes. Acknowledgments We thank T. Shibata and T. Sugimoto for assistance with the analyses of the whole-rock major element composition of rocks using the XRF. We also thank H. Yoshida for operating the EBSD at the University of Tokyo, K. Komatsu and E. A. Johnson for their technical advice on infrared spectroscopy, and J. R. Holloway, A. H. Peslier and I. Kushiro for their fruitful discussions. This manuscript was improved by constructive comments by two anonymous reviewers. This study was supported by the COE program provided to M. H. and a Grant-inAid for Scientific Research given to T. K. and E. T. References Carslaw, H.S., Jaeger, J.C., 1959. Conduction of Heat in Solids. Oxford University Press, London. Demouchy, S., Jacobsen, S.D., Gaillard, F., Stern, C.R., 2006. Rapid magma ascent recorded by water diffusion profiles in mantle olivine. Geology 34, 429–432. Endo, K., Chiba, T., Taniguchi, H., Sumita, M., Tachikawa, S., Miyahara, T., Ueno, R., Miyaji, N., 1988. Tephrochronological study on the 1986–1987 eruptions of Izu-Oshima volcano, Japan (in Japanese with English abstract). Bull. Volcanol. Soc. Jpn. 33, S32–S51. Fischer, T.P., 2008. Fluxes of volatiles (H2O, CO2, N2, Cl, F) from arc volcanoes. Geochem. J. 42, 21–38. Fujii, T., Aramaki, S., Kaneko, T., Ozawa, K., Kawanabe, Y., Fukuoka, T., 1988. Petrology of the lavas and ejecta of the November, 1986 eruption of Izu-Oshima volcano (in Japanese with English abstract). Bull. Volcanol. Soc. Jpn. 33, S234–S254. Grove, T.L., Baker, M.B., 1984. Phase equilibrium controls on the tholeiitic versus calcalkaline differentiation trends. J. Geophys. Res. 89, 3253–3274. Hamada, M., Fujii, T., 2007. H2O-rich island arc low-K tholeiite magma inferred from Carich plagioclase-melt inclusion equilibria. Geochem. J. 41, 437–461. Ida, Y., 1995. Magma chamber and eruptive processes at Izu-Oshima volcano, Japan: buoyancy control of magma migration. J. Volcanol. Geotherm. Res. 66, 53–67. Ikehata, K., Yasuda, A., Notsu, K., 2010. The geochemistry of volatile species in melt inclusions and sulfide minerals from Izu-Oshima volcano, Japan. Mineral. Petrol. 99, 143–152. Iwamori, H., 2000. Deep subduction of H2O and deflection of volcanic chain towards backarc near triple junction due to lower temperature. Earth Planet. Sci. Lett. 181, 41–46. Johnson, E.A., 2003. Hydrogen in nominally anhydrous crustal minerals. Ph.D. Thesis, California Inst. Tech., Pasadena, USA. Johnson, E.A., 2005. Magmatic water contents recorded by hydroxyl concentrations in plagioclase phenocrysts from Mount St. Helens, 1980–1981. Geochim. Cosmochim. Acta 69, A743. Johnson, E.A., Rossman, G.R., 2003. The concentration and speciation of hydrogen in 1 feldspars using FTIR and H MAS NMR spectroscopy. Am. Mineral. 88, 901–911. Johnson, E.A., Rossman, G.R., 2004. A survey of hydrous species and concentrations in igneous feldspars. Am. Mineral. 89, 586–600. Johnson, E.R., Wallace, P.J., Cashman, K.V., Granados, H.D., Kent, A.J.R., 2008. Magmatic volatile contents and degassing-induced crystallization at Volcán Jorullo, Mexico: implications for melt evolution and the plumbing systems of monogenetic volcanoes. Earth Planet. Sci. Lett. 269, 478–487. Kawamoto, T., 1996. Experimental constraints on differentiation and H2O abundance of calc-alkaline magmas. Earth Planet. Sci. Lett. 144, 577–589. Kawamoto, T., Kagi, H., Yamashita, S., Handa, T., Matsukage, K., Ikemoto, Y., Moriwaki, T., Kimura, H., 2003. Mid infrared throughput with 5 μm aperture for H2O determination of an andesitic glass: comparison of synchrotron radiation source at Spring-8 with conventional light sources. Geochem. J. 37, 253–259.

Kazahaya, K., Shinohara, H., Saito, G., 1994. Excessive degassing of Izu-Oshima volcano: magma convection in a conduit. Bull. Volcanol. 56, 207–216. Keppler, H., 1999. Experimental evidence for the source of excess sulfur in explosive volcanic eruptions. Science 284, 1652–1654. Keppler, H., 2010. The distribution of sulfur between haplogranitic melts and aqueous fluids. Geochim. Cosmochim. Acta 74, 645–660. Kronenberg, A.K., Yund, R.A., Rossman, G.R., 1996. Stationary and mobile hydrogen defects in potassium feldspar. Geochim. Cosmochim. Acta 60, 4075–4094. Mikada, H., Watanabe, H., Sakashita, S., 1997. Evidence for subsurface magma bodies beneath Izu-Oshima volcano inferred from a seismic scattering analysis and possible interpretation of the magma pluming system of the 1986 eruptive activity. Phys. Earth Planet. Inter. 104, 257–269. Nakano, S., Yamamoto, T., 1987. Major element chemistry of products of the 1986 eruption of Izu-Oshima volcano (in Japanese with English abstract). Bull. Geol. Surv. Jpn. 38, 631–647. Nakano, S., Makino, K., Eriguchi, T., 2001. Microtexture and water content of alkali feldspar by Fourier-transform infrared microspectroscopy. Mineral. Mag. 65, 675–683. Peacock, S.M., Wang, K., 1999. Seismic consequences of warm versus cool subduction metamorphism: examples from southwest and northeast Japan. Science 286, 937–939. Peslier, A.H., Luhr, J.F., 2006. Hydrogen loss from olivines in mantle xenoliths from Simcoe (USA) and Mexico: mafic alkalic magma ascent rates and water budget of the sub-continental lithosphere. Earth Planet. Sci. Lett. 242, 302–319. Peslier, A.H., Woodland, A.B., Wolff, J.A., 2008. Fast kimberlite ascent rates estimated from hydrogen diffusion profiles in xenolithic mantle olivines from southern Africa. Geochim. Cosmochim. Acta 72, 2711–2722. Pichavant, M., Mysen, B.O., Macdonald, R., 2002. Source and H2O content of high-MgO magmas in island arc settings: an experimental study of a primitive calc-alkaline basalt from St. Vincent, Lesser Antilles arc. Geochim. Cosmochim. Acta 66, 2193–2209. Saito, G., Uto, K., Kazahaya, K., Shinohara, H., Kawanabe, Y., Satoh, H., 2005. Petrological characteristics and volatile content of magma from the 2000 eruption of Miyakejima Volcano, Japan. Bull. Volcanol. 67, 268–280. Saito, G., Morishita, Y., Shinohara, H., 2010. Magma plumbing system of the 2000 eruption of Miyakejima volcano, Japan, deduced from volatile and major component contents of olivine-hosted melt inclusions. J. Geophys. Res. 115 (B11202). doi:10.1029/2010JB007433. Sakuyama, M., 1979. Lateral variations of H2O contents in Quaternary magmas of northeastern Japan. Earth Planet. Sci. Lett. 43, 103–111. Sharma, K., Blake, S., Self, S., Krueger, A.J., 2004. SO2 emissions from basaltic eruptions, and the excess sulfur issue. Geophys. Res. Lett. 31 (L13612). doi:10.1029/ 2004GL019688. Shewmon, P.G., 1983. Diffusion in Solids. J. Williams Book Company, Jenks, OK, USA. Shinohara, H., 2008. Excess degassing from volcanoes and its role on eruptive and intrusive activity. Rev. Geophys. 46 (RG4055). doi:10.1029/2007RG000244. Sisson, T.W., Grove, T.L., 1993a. Experimental investigations of the role of H2O in calcalkaline differentiation and subduction zone magmatism. Contrib. Mineral. Petrol. 113, 143–166. Sisson, T.W., Grove, T.L., 1993b. Temperatures and H2O contents of low-MgO highalumina basalts. Contrib. Mineral. Petrol. 113, 167–184. Spilliaert, N., Allard, P., Métrich, N., Sobolev, A.V., 2006. Melt inclusion record of the conditions of ascent, degassing, and extrusion of volatile-rich alkali basalt during the powerful 2002 flank eruption of Mount Etna (Italy). J. Geophys. Res. 111 (B04203). doi:10.1029/2005JB003934. Stolper, E., 1982. Water in silicate glasses: an infrared spectroscopic study. Contrib. Mineral. Petrol. 81, 1–17. Straub, S.M., Layne, G.D., 2003. The systematics of chlorine, fluorine, and water in Izu arc front volcanic rocks: implications for volatile recycling in subduction zones. Geochim. Cosmochim. Acta 67, 4179–4203. Sugimoto, T., Shibata, T., Yoshikawa, M., 2007. Procedure of making a fused glass bead for whole rock major elements analyzed by X-ray fluorescence spectrometer RIGAKU SYSTEM 3070. Annual Report FY2006. Institute for Geothermal Sciences, Kyoto University. Takagi, D., Sato, H., Nakagawa, M., 2005. Experimental study of a low-alkali tholeiite at 1–5 kbar: optimal condition for the crystallization of high-An plagioclase in hydrous arc tholeiite. Contrib. Mineral. Petrol. 149, 527–540. Tatsumi, Y., Sakuyama, M., Furukawa, H., Kushiro, I., 1983. Generation of arc basalt magmas and thermal structure of the mantle wedge in subduction zones. J. Geophys. Res. 88, 5815–5825. Wallace, P.J., 2001. Volcanic SO2 emissions and the abundance and distribution of exsolved gas in magma bodies. J. Volcanol. Geotherm. Res. 108, 85–106. Watanabe, H., Okubo, S., Sakashita, S., Maekawa, T., 1998. Drain-back process of basaltic magma in the summit conduit detected by microgravity observation at Izu-Oshima volcano, Japan. Geophys. Res. Lett. 25, 2865–2868.