m. hamada 1 *, m. ushioda 1 , t. fujii 2,3 and e. takahashi 1

Hydrogen concentration in plagioclase as a hygrometer of magmas: Approaches from melt inclusion analyses and hydrous melting experiments

M. Hamada1*, M. Ushioda1, T. Fujii2,3 and E. Takahashi1

1Department of Earth and Planetary Sciences, Tokyo Institute of Technology, Tokyo, Japan 2Earthquake Research Institute, University of Tokyo, Tokyo, Japan 3Crisis & Environmental Management Policy Institute, Tokyo, Japan (*E-mail: hamada@geo.titech.ac.jp)

V11C-2776

Plagioclase is one of the nominally anhydrous minerals (NAMs) which accommodates hydrogen, and plagioclase in volcanic rocks essentially contains structural OH in it. Hydrogen concentration in NAMs can be a useful proxy for dissolved H2O in silicate melts if the partitioning of hydrogen between plagioclase and melts is known. Here, we performed two parts of studies in order to determine hydrogen partitioning between plagioclase and hydrous basaltic melt: (i) analyses of low-H2O melt inclusions (H2O ≈ 0.3 wt.%) hosted by plagioclase in mid-ocean ridge basalt (MORB) and (ii) hydrogen partitioning experiments between Ca-rich plagioclase (An95) and hydrous arc basaltic melt (0.8 ~ 5.5 wt.% H2O) at 0.35 GPa. Obtained formulation is

Hydrogen in plagioclase (wt. ppm water) ≈ 94.3×√(H2O in melt, wt.%).

This empirical formulation can be used as a practical hygrometer of magmas. In this poster, we apply this formulation to understand the 1986-1987 summit eruption of Izu-Oshima volcano, a frontal-arc volcano in Izu arc, and discuss eruption process of H2O-saturated magma.

Purpose and quick summary

Part 1) Analyses of plagioclase-melt inclusion pairs from MORB

Figure is from Nakamura et al. (2007 Marine Geol.)

KH93-3 DR9

50 μm

500 μm

Whole-rock composition

SiO2 50.5 TiO2 1.52 Al2O3 15.1 FeO* 11.1 MnO 0.19 MgO 7.44 CaO 11.1 Na2O 2.80 K2O 0.10 P2O5 0.15 Total 100 (wt.%)

Studied sample: Sample# KH93-3 DR9 from Rodriguez Triple Junction in the Indian Ocean

Part 2) Hydrous melting experiments to determine

H partitioning between plagioclase and melt

1000

1050

1100

1150

1200

1250

1300

0 1 2 3 4 5 6

Experimental procedures

volcanic front

35ºN

Izu-Oshima volcano

30ºN

45ºN145ºE140ºE135ºE130ºE

Miyakejima volcano

SiO2

TiO2

Al2O3

FeO* MnO MgO CaO Na2O K2OTotal

51.2 0.9617.511.2 0.21 4.911.5 2.2 0.25100 (wt.%))

Starting material of melting experiments (MTL rock, collected from Miyakejima volcano in Izu

arc)

H2O in melt (wt.%)

Tem

pera

ture

()

℃

plagioclase-in

augite-inmagnetite-in

plagioclase-in augite-inmagnetite-in

Partitioning experiment of hydrogen between An95 plagioclase and melt

Au80Pd20 capsule

200 μm

Crushing volcanic rock

Separation of An95 plagioclase

Hydrous melting of crushed rock

10 mg of hydrous glass

1 mg of An95 plagioclase

Melting at 0.35 GPa for 24-48 hours using

internally-heated pressure vessel

Backscattered electron image of the recovered sample (1130℃, 2.6 wt.% in melt)

Obtained phase diagram of MTL rock at 0.35 GPa

An95 plagioclase

fO2~NNO+3

300

H2O in glass (wt.%)

0

50

100

150

200

250

0

0.005

0.010

0.015

0 1 2 3 4 5 6

H in

pla

gioc

lase

(w

t.ppm

wat

er)

DH

(pla

g/m

elt)

CH2O (wt. ppm) ≈94.3× CH2O (wt. %)meltplag

Hydrous melting experiments

Melt inclusions

0.01±0.005

0.008±0.002

0 1 2 3OH in glass (wt.%)

0

5

10

15

20

25

25003000350040004500500055006000

☞ Linear correlation between H concentration in plagioclase and OH in melt suggests that H is accommodated in plagioclase as OH.

Wavenumber (cm-1)

Abs

orpt

ion

per

cm

#RTJ-pl20(11 wt. ppm water, 0.3 wt.% H2O in melt inclusion)

#RTJ-pl24(38 wt. ppm water, 0.3 wt.% H2O in melt inclusion)

#MTL17(68 wt. ppm water, 0.8 wt.% H2O in melt)

#MTL05(153 wt. ppm water, 2.3 wt.% H2O in melt)

#MTL39(157 wt. ppm water, 3.5 wt.% H2O in melt)

#MTL37(210 wt. ppm water, 4.5 wt.% H2O in melt)

#MTL41(225 wt. ppm water, 5.6 wt.% H2O in melt)

Natural plagioclase in MTL lava(5 wt. ppm water)

resi

dual

res

in?

Longer hydrogen bond length d(O⋯O) Shorter hydrogen bond length d(O⋯O)

☞ Peak of infrared absorption spectra of plagioclase shifts from lower wavenumbers (peak position: 3200-3400 cm-1) under H2O-poor conditions to higher wavenumbers (peak position: 3600 cm-1) under H2O-rich conditions, meaning expansion of O-H⋯O bond length with increasing H2O. These observations suggest that hydrogen site in plagioclase slightly changes with increasing H2O, which also changes hydrogen partitioning between plagioclase and melt as shown above.

－ Representative absorption spectra of plagioclase －

☞ High hydrogen concentration in plagioclase from Izu-Oshima volcano, a frontal-arc volcano in Izu arc (≥200 wt. ppm water, Hamada et al., 2011 EPSL), suggests crystallization of plagioclase from H2O-rich (≥4 wt.%) melt.

Hamada et al. (2011, EPSL)

An80 An85 An90 An95

Anorthite content of plagioclase

☞ Hydrogen partition coefficient DH slightly decreases with increasing H2O in coexisting melt.

Internally-heated pressure vessel installed at Magma Factory, Tokyo

Tech.

40ºN

1160 ℃

1130 ℃

1050 ℃

1000 ℃

1050 ℃

☞ H concentration in plagioclase can be approximated as a square root of H2O in melt. Decreasing temperature lowers H concentration in plagioclase. However, the effect of temperature is minor.

Plagioclase (An95)Olivine (Fo78-81)ClinopyroxeneOrthopyroxeneMagnetite

16 2 0 0 0 (vol.%))

Modal composition

0

20

40

60

80

70 75 80 85 90 0

10

20

30

40

50

0 0.1 0.2 0.3 0.4 0.5

Plagioclase (An74-89)Olivine (Fo90)

51 (vol.%))

Modal composition

Cross-Nicol image of studied MORB.Quenched glass contains ~0.3 wt.% H2O.

Backscattered-electron image of plagioclase and melt inclusion.

Analytical result using FT-IR

An content of plagioclase

Hyd

roge

n co

ncen

trat

ion

in

plag

iocl

ase

(wt.

ppm

wat

er)

H2O in plagioclase-hosted melt inclusions (wt.%)

Hyd

roge

n co

ncen

trat

ion

in

plag

iocl

ase

(wt.

ppm

wat

er)

≈ 0.01

Integration of Part 1 and Part 2

Application

☞ Hydrogen concentration in plagioclase and H2O concentration in glass was measured using FT-IR.