Water contents and pressures of melts in unerupted felsic magma constrained by SEM-EDS analysis of homogenized melt inclusions in zircon

Abstract

Granitic rocks (sensu lato) represent unerupted felsic magmas crystallized in the crust. In this study, we estimate water contents of melts and crystallization pressures of zircons in granitoid magma using melt inclusions in zircon, a ubiquitous accessory mineral in granitoids. Homogenization experiments of polymineralic inclusions hosted in zircon have been conducted for a granitoid sample from the Cretaceous Gamano granodiorite in Yashiro-jima Island, southwest Japan, using a piston-cylinder high-pressure-high-temperature apparatus. SEM-EDS analysis reveals that the homogenized melt inclusions have high water contents (6.4-11.3 wt%) and high SiO₂ contents (76-78 wt% anhydrous basis) implying that they represent fractionated interstitial hydrous melts trapped in growing zircon crystals. A recently proposed machine learning-based melt-phase assemblage geobarometer yields pressures ranging from 563 to 266 MPa interpreted as crystallization pressures of the zircons. The results of this study suggest high water activity of the interstitial melts within the Gamano granodiorite magma at the time of zircon crystallization. The melt inclusions in zircons record a wide range of pressures, from intrusion of the magma into the deeper crustal levels (∼ 563-509 MPa) to final solidification at shallower levels (∼ 266 MPa).

INTRODUCTION

Water content is one of the fundamental factors controlling first-order properties of felsic melts, such as density and viscosity (see Johannes and Holtz, 1996). In the case of felsic volcanic rocks, water contents of melt are routinely determined by the analysis of matrix glass and/or glass inclusions with many methods including infrared spectroscopy (e.g., Anderson et al., 1989; Wallace and Gerlach, 1994; Yasuda et al., 2015), Raman spectroscopy (e.g., Chabiron et al., 2004), and secondary ion mass spectrometry (SIMS) analysis (e.g., Miyagi and Yurimoto, 1995). However, contrary to felsic volcanic rocks, water contents of melt that formed felsic plutonic rocks are difficult to directly determine because holocrystalline plutonic rocks lack matrix glass or glass inclusions.

Although plutonic rocks lack matrix glass or glass inclusions, crystallized melt is commonly preserved as polymineralic inclusions in zircon, a ubiquitous accessory mineral occurring in granitoids (Thomas et al., 2003). Previous studies have successfully measured the composition of melt inclusions in zircon separated from plutonic rock samples through homogenization experiments under high-pressure-high-temperature conditions (e.g., Thomas et al., 2003; Gudelius et al., 2020; Taniwaki et al., 2023). Because crystalline zircons are durable to weathering and commonly avoid alteration, zircon is an ideal host mineral for melt inclusion studies (e.g., Thomas et al., 2003) despite their tiny size making it difficult to analyze the inclusions.

Among the previous studies on melt inclusions in zircons, Gudelius et al. (2020) determined the water contents of homogenized melt inclusions by SIMS, illustrating that water contents of the melt can be directly constrained from plutonic rock samples. However, Gudelius et al. (2020) did not consider pressure, which significantly affects water solubility in silicate melts (e.g., Johannes and Holtz, 1996). Taniwaki et al. (2023) have recently proposed a new geobarometric approach based on the composition of melt inclusions in zircon, applicable to granitoid samples. These two studies collectively suggest that melt inclusions in zircon can provide information on both water contents and pressure conditions. Nevertheless, an attempt to determine water contents and pressure conditions from the same melt inclusions has not yet been reported, although both are essential in revealing magmatic process of plutonic rocks.

The objective of this study is to develop a technique to constrain both water content and pressure from melt inclusions in zircon from a granitoid sample. Recently, Geshi et al. (2017) have proposed a method for estimation of water content in volcanic glasses using a scanning electron microscopy with energy dispersive spectroscopy (SEM-EDS), which allows water content of small melt inclusions in zircon to be estimated after homogenization experiments. The advantages of this method are that (1) the SEM-EDS can analyze quite a small area (i.e., 16-36 µm², in this study), (2) sample preparation requires only single-face polish (no need for double-face polish), and (3) major element compositions are simultaneously obtained during analysis (see Geshi et al., 2017). In this study, we report new data including major element compositions and water contents of homogenized melt inclusions in zircon from a Cretaceous granitoid in Yashiro-jima Island, southwest Japan. We then interpret zircon crystallization pressures based on the melt inclusion compositions, and discuss the geological implications. The results illustrate that physicochemical conditions, such as pressures, melt compositions, and water contents, of unerupted felsic melt can be successfully constrained by the technique presented in this study.

GEOLOGICAL BACKGROUND, FIELD OBSERVATIONS AND SAMPLE DESCRIPTION

The Yashiro-jima Island is predominantly underlain by Cretaceous granitoids that intrude minor high-grade metamorphic rocks of the Ryoke belt (cf. Kodama et al., 2021). The metamorphic rocks in the northwest are within the garnet-cordierite zone, whereas those of other areas of the island belong to the sillimanite-K-feldspar zone (Ikeda, 1998) (Fig. 1). The granitoids are divided into the Gamano granodiorite in the west and the Towa granodiorite in the east (Fig. 1b, Kodama et al., 2021). The Gamano granodiorite locally exhibits a foliated structure in the vicinity of Ryoke metamorphic rocks, whereas the Towa granodiorite shows a relatively massive structure, showing a gradual transition between the two lithologies (Kodama et al., 2021). Recently, Kodama et al. (2021) have shown that the Towa granodioritic magma can be formed by fractional crystallization of the Gamano magma based on Sr-Nd isotope data and geochemical modeling, and further suggested that the Gamano granodiorite and the Towa granodiorite are derived from a single parental magma. On the other hand, Ikeda et al. (2019) have illustrated various magma processes for Gamano granodiorite genesis including differentiation, assimilation, and magma mixing/mingling based on field and petrographic observations, Sr-Nd isotope data and geochemical modeling.

Figure 1. (a) Map of the Chugoku-Shikoku region in southwest Japan showing the location of the study area. (b) Geological map of Yashiro-jima Island (modified after Kodama et al., 2021) showing sampling sites. The boundary between the garnet-cordierite metamorphic zone and sillimanite-K-feldspar zone is from Ikeda (1998). Metamorphic pressure conditions estimated by Ikeda (2004) and zircon U-Pb ages reported by Herzig et al. (1998) and Skrzypek et al. (2016) are also shown with their sample numbers.

Herzig et al. (1998) have reported zircon U-Pb ages of 96.2 ± 3 (KO-1, granite gneiss) and 95.3 ± 1 Ma (GG-1, hornblende-biotite granodiorite) from the northern area of Yashiro-jima Island (Fig. 1). Recently, Skrzypek et al. (2016) have reported a zircon U-Pb age of 93.8 ± 1 Ma from a microgranite vein (EY38M, fine-grained biotite granite) in metachert from the southwest, which likely relates to the neighboring Gamano granodiorite (Skrzypek et al., 2016) (Fig. 1). Ikeda (2004) has reported metamorphic pressure (P)-temperature (T) conditions for the Ryoke metamorphic rocks from the northwestern area (529-420 MPa and 855-849 °C, YFE3 of Ikeda, 2004, Fig. 1) and from the southwestern area (690-517 MPa and 763-688 °C, YFC7A, YFS7B, and YFD5 of Ikeda, 2004, Fig. 1). The latter represents the highest metamorphic pressure conditions of the Yanai district in the Ryoke Belt (Ikeda, 2004). Field relations (e.g., Okudaira et al., 1995) together with geochronological data (Herzig et al., 1998; Suzuki and Adachi, 1998; Skrzypek et al., 2016) have indicated that the emplacement of the Gamano granodiorite was contemporaneous with peak metamorphism (see Ikeda et al., 2019).

The sample used for the melt inclusion analysis in this study (YSR-105A) is a coarse-grained granitoid of the Gamano granodiorite collected from the northwestern area of Yashiro-jima Island, adjacent to YFE3 of Ikeda (2004), from which metamorphic P-T conditions have been previously estimated (Fig. 1b). At the outcrop, the Gamano granodiorite shows heterogenous appearance in terms of grain size and modal compositions and exhibits a concordant intrusive relation to the migmatitic foliation in the Ryoke metamorphic rocks (Fig. 2a). The sample (YSR-105A) was collected at ∼ 1 meter from the contact with Ryoke metamorphic rocks and is composed mainly of euhedral-subhedral plagioclase, subhedral-anhedral biotite, anhedral-subhedral quartz, and anhedral alkali-feldspar, and with minor amounts of accessory minerals including zircon, apatite, and opaque minerals (Fig. 2b). Microscopic petroscopy confirms the crystallization sequence of major mineral constituents, from plagioclase through biotite then quartz to finally alkali-feldspar. Zircon grains are euhedral and dominantly included in quartz, alkali-feldspar, and rim domains of biotite or occur along boundaries between main mineral constituents (Fig. 2b). Some zircon grains are also included in plagioclase.

Figure 2. (a) Field context for the sample used for homogenization experiment (YSR-105A). (b) Photomicrograph of the Gamano granodiorite sample (YSR-105A). Cross-polarized light. (c) Backscattered electron (BSE) image of a representative polymineralic quartzofeldspathic inclusion in zircon. (d) BSE image and element distribution maps for Si, Al, and K for a melt inclusion in zircon after the homogenization experiment. Afs, alkali-feldspar; Pl, plagioclase; Bt, biotite; Qz, quartz; Zrn, zircon.

ANALYTICAL TECHNIQUE

Zircon grains (mostly 250-40 µm in length, Fig. 3) were concentrated by panning and further processed with a hand magnet, and the remaining fractions were purified using heavy liquid (sodium polytungstate aqueous solution) separation. Rounded inclusions of variable size (mostly 30-3 µm) are abundant and comprise quartz, plagioclase, and alkali-feldspar with voids (Fig. 2c). These polymineralic quartzofeldspathic inclusions are interpreted as crystallized former melt inclusions trapped during crystal growth within a granitic melt (cf. Thomas et al., 2003; Taniwaki et al., 2023), although the possibility that these inclusions are mixtures of melt and trapped minerals cannot be ruled out (cf. Taniwaki et al., 2023).

Figure 3. Cathodoluminescence images of zircon grains with spot location, showing estimated H₂O (wt%) and results of MagBaTaB geobarometer (Table 1).

Homogenization experiments of the melt inclusions were performed under high pressures using a piston-cylinder type high-pressure apparatus at the Department of Earth Sciences, Graduate School of Science and Engineering, Ehime University, Japan, following the methods described in Taniwaki et al. (2023). Talc, pyrophyllite, boron nitride, and borosilicate glass were used as the pressure-transmitting media. Gudelius et al. (2020) have reported ‘volatile escape’ for homogenized melt inclusions in zircons, which occurs for inclusions connected with the crystal surface by minute fractures. In this study, the zircon samples were sealed in a Pt capsule with NaCl powder. The major advantage of using NaCl as the pressure medium in the Pt capsule is that the melt inclusions open to the surface or intersected by minute fractures of host zircon, which would cause ‘volatile escape’ from the melt inclusions during experiments (cf. Gudelius et al., 2020), can be easily recognized by distinctive analytical data (i.e., extremely higher Na₂O contents than typical for felsic melts) due to contamination of NaCl. Backscattered electron (BSE) imaging and element distribution mapping (Fig. 2d) have confirmed complete homogenization of the melt inclusions after the experiment (at 900 °C and 0.3 GPa, 4.5 h) although preliminary tests showed incomplete homogenization of melt inclusions at temperatures near zircon saturation (i.e., at 780 °C and 24 h).

Cathodoluminescence (CL) images of zircon grains after the homogenization experiments were obtained using a field-emission scanning electron microscope (FE-SEM; JEOL JSM-7000F) with a Gatan Mini-CL system at the Geodynamics Research Center, Ehime University, Japan, and a JEOL JSM-6610LV scanning electron microscope with a Gatan Mini-CL system at the Akita University, Japan. Secondary and backscattered electron imaging, element distribution mapping, and quantitative analyses were carried out using a SEM-EDS (JEOL JSM-6510LV with Oxford Instruments X-Max 20 and X-Max 50) at the Department of Earth Sciences, Graduate School of Science and Engineering, Ehime University, Japan. The analytical conditions were 15 kV accelerating voltage, 0.8 nA beam current, and 50-60 s acquisition time. We use cobalt metal for the calibration of the beam current. Obsidian was used as a standard for quantitative analysis. The software INCA was used for data analysis. Because repeated analysis of the obsidian standard with changing magnification showed significant Na-loss at higher magnification (>5000) (Supplementary Fig. S1a; Figs. S1-S4 are available online from https://doi.org/10.2465/jmps.230904), we analyzed melt inclusions with the magnification of 2000-3000 (Table 1) depending on the size of the inclusions. Negligible Na-loss during each analysis is supported by the liner relationship between the intensity of sodium Kα and analysis time (Supplementary Fig. S1b). Analysis of some small melt inclusions detected Zr (<2.1% atomic ratio), interpreted as overlapping signal from the surrounding host zircon. For such analysis, the apparent Zr data can be used to estimate the amount of Si and O (Zr:Si:O = 1:1:4) that needs to be removed from the melt inclusion analyses to account for the overlapping signal from the surrounding zircon. These extraneous Zr, Si, and O values were then subtracted from the melt inclusion analysis (Table 1). In addition, Gudelius et al. (2020) have suggested that melt inclusion compositions were not significantly biased by the melting of host zircon during their homogenization experiments (at 1100 °C and 0.2 GPa, 6 h). Therefore, the melting of host zircons into melt inclusions at lower experimental temperature (at 900 °C) is probably negligible in this study.

Table 1. Results of the SEM-EDS analysis of melt inclusions in zircon and estimated H₂O contents and pressures

Spot identifier	MI7	MI14	MI21	MI23*	MI26*	MI29	MI31-1*	MI31-2*	MI32	MI33
Domain	Core	Rim	Rim	Rim	Rim	Rim	Rim	Rim	Rim	Rim
Magnification	20000	20000	25000	20000	30000	30000	25000	30000	30000	30000
Analyzed areas (µm2)	36	36	23	36	16	16	23	16	16	16
Atomic ratio (100%)
O	68.1	67.1	66.8	65.5	66.2	66.3	65.5	66.5	67.2	68.0
Na	0.4	1.8	1.9	1.9	1.8	1.8	1.6	1.6	1.6	1.7
Al	4.9	5.2	5.3	5.2	5.3	5.0	5.3	4.6	5.1	4.9
Si	24.2	23.6	23.6	25.0	24.6	24.7	25.2	23.3	23.8	23.0
K	2.0	2.0	2.1	2.1	1.3	1.6	1.7	1.6	2.0	1.1
Ca	0.4	0.4	0.3	0.3	0.8	0.5	0.6	0.4	0.3	0.8
Fe
Zr								2.1		0.5
Original total (wt%)	101.0	100.9	98.9	92.4	89.9	97.7	17.9	19.0	97.2	100.0
Zr-corrected
O								66.4		65.8
Na								1.8		1.7
Al								5.2		4.9
Si								24.2		22.4
K								1.8		1.1
Ca								0.5		0.8
Excess O	10.7	9.8	9.3	5.3	6.8	7.2	4.8	7.9	9.8	11.5
Calculated composition (wt%) normalized to 100% with H2O
SiO2	71.9	70.0	70.0			73.1			70.4	69.0
Al2O3	12.3	12.9	13.2			12.6			12.9	12.8
FeO
CaO	1.2	1.1	0.8			1.5			0.9	2.2
Na2O	0.6	2.8	2.9			2.8			2.5	2.7
K2O	4.6	4.6	4.8			3.7			4.7	2.7
H2O	9.5	8.7	8.3			6.4			8.7	10.5
Calculated composition (wt%) normalized to 100% without H2O
SiO2	79.4	76.7	76.3			78.0			77.1	77.1
Al2O3	13.6	14.2	14.4			13.4			14.1	14.3
FeO
CaO	1.3	1.2	0.9			1.6			0.9	2.5
Na2O	0.6	3.0	3.2			3.0			2.7	3.0
K2O	5.0	5.0	5.2			4.0			5.2	3.0
A.S.I.**	1.53	1.14	1.15			1.11			1.20	1.12
Rhyolite-MELTS geothermobarometer (version 1.1.x) (Gualda and Ghiorso, 2014) quartz +1 feldspar constraint (±25 MPa)
Pressure (MPa)	No pressure calculated	251	273			158			168	306
DERP geobarometer (Wilke et al., 2017)
Pressure (MPa)	−602	292	403			148			185	274
MagMaTaB geobarometer (version 4.0) (Weber and Blundy, 2024) (with root-mean-square-error of 110-130 MPa)
Pressure (MPa)	-	373	354			395			324	563

^*Data with considerably low original totals (<94 wt%) were discarded in this study because they probably had issues with analysis or sample preparation.

^**Alumina Saturation Index: mol. Al₂O₃/(CaO + Na₂O + K₂O).

^***Average of 20 times analysis with XRF for HOb-202-1 (secondary obsidian standard).

Spot identifier	MI35	MI36	MI37-1	MI37-2	MI38	MI39	HOb-202-1 (Obsidian)
Domain	Rim	Rim	Rim	Rim	Rim	Rim
Magnification	30000	25000	30000	30000	30000	30000	30000
Analyzed areas (µm2)	16	23	16	16	16	16	16
Atomic ratio (100%)							Average (n = 6)	1σ
O	67.4	64.6	68.0	68.1	67.8	66.3	63.9	0.3
Na	1.5	5.6	1.9	1.7	1.6	1.6	2.6	0.2
Al	4.9	4.9	5.4	5.4	4.7	5.2	5.2	0.1
Si	23.9	23.6	22.8	22.6	23.0	24.5	26.0	0.1
K	2.0	0.7	1.6	1.7	1.3	2.1	2.0	0.0
Ca	0.3	0.3	0.4	0.4	0.7	0.3	0.2	0.0
Fe							0.2	0.0
Zr		0.4			1.0
Original total (wt%)	99.6	99.4	96.7	97.0	94.3	97.1	99.8
Zr-corrected
O		63.1			63.9
Na		5.6			1.6
Al		4.9			4.7
Si		23.3			22.0
K		0.7			1.3
Ca		0.3			0.7
Excess O	10.3	6.0	12.2	12.6	12.5	7.4	1.6
Calculated composition (wt%) normalized to 100% with H2O									XRF***
SiO2	70.9	70.9	67.9	67.3	69.5	72.3	76.3	0.3	75.9
Al2O3	12.2	12.6	13.6	13.7	12.5	12.9	12.8	0.1	12.5
FeO							0.6	0.1	0.6
CaO	0.9	0.7	1.1	1.2	2.1	0.9	0.3	0.3	0.5
Na2O	2.2	8.9	2.9	2.6	2.5	2.5	3.9	0.3	4.0
K2O	4.7	1.6	3.6	4.0	3.1	4.8	4.6	0.1	4.6
H2O	9.1	5.4	10.8	11.3	10.2	6.6	1.4	0.4
Calculated composition (wt%) normalized to 100% without H2O
SiO2	78.0	74.9	76.1	75.8	77.4	77.4	77.4	0.5
Al2O3	13.4	13.3	15.3	15.4	14.0	13.8	13.0	0.1
FeO							0.6	0.1
CaO	1.0	0.8	1.3	1.3	2.4	1.0	0.3	0.3
Na2O	2.4	9.4	3.3	3.0	2.8	2.7	4.0	0.2
K2O	5.2	1.7	4.1	4.5	3.4	5.2	4.6	0.1
A.S.I.**	1.16	0.72	1.27	1.26	1.10	1.18	1.08	0.09
Rhyolite-MELTS geothermobarometer (version 1.1.x) (Gualda and Ghiorso, 2014) quartz +1 feldspar constraint (±25 MPa)
Pressure (MPa)	59	No pressure calculated	310	323	275	134
DERP geobarometer (Wilke et al., 2017)
Pressure (MPa)	83	777	163	246	246	71
MagMaTaB geobarometer (version 4.0) (Weber and Blundy, 2024) (with root-mean-square-error of 110-130 MPa)
Pressure (MPa)	266	-	512	531	509	290

^*Data with considerably low original totals (<94 wt%) were discarded in this study because they probably had issues with analysis or sample preparation.

^**Alumina Saturation Index: mol. Al₂O₃/(CaO + Na₂O + K₂O).

^***Average of 20 times analysis with XRF for HOb-202-1 (secondary obsidian standard).

The estimation of water content in the homogenized melt (glass) inclusions followed the methods described by Geshi et al. (2017). The repeated analyses of secondary obsidian standards (HOb-202-1) results in 1.4 ± 0.4 wt% H₂O (n = 6, 1σ) in this study, which is consistent with the measurements by Fourier-transform infrared (FT-IR) micro-reflectance spectroscopy (1.26 ± 0.01 wt% H₂O, n = 2, 1σ) following the method described by Yasuda (2014) and an ordinary measurement of ignition loss (1.3 wt% H₂O⁺). Considering the uncertainty of the thickness evaluation for carbon coating on the sample (±5.6 nm in this study), which is given by the stoichiometric ratio from repeated analyses of standard quartz and crucial for precise quantification of oxygen (Geshi et al., 2017), uncertainty of H₂O contents in this study is within ±1.8 wt%. The analytical results of the melt inclusions together with secondary obsidian standards are listed in Table 1. We discarded data with a considerably low original total (<94 wt%, 4 of a total of 16 analysis, Table 1) because they probably had issues with analysis or sample preparation.

Whole-rock major and trace element compositions for seven samples from the Gamano granodiorite and for nine samples from the Towa granodiorite (Fig. 1) were determined by X-ray fluorescence spectrometry (XRF) (RIGAKU ZSX Primus II) at the Graduate School of Science and Engineering, Ehime University, Japan. The analytical procedure followed the methods described by Suda and Saito (2018). Whole-rock compositions are listed in Supplementary Table S1 (Supplementary Table S1 is available online from https://doi.org/10.2465/jmps.230904). Anhydrous values normalized to 100% are used in the text and Figures for all major element data.

RESULTS

Zircon grains after the homogenization experiment exhibit varieties of zoning patterns in CL images (Fig. 3), indicating that the internal structure of the grains is mostly preserved in the experimental conditions. Zircon grains have well-developed concentric oscillatory zoning. Melt inclusions are observed in both core and rim domains.

Major element compositions of 12 analyses of homogenized melt inclusions are shown in Figure 4. Among the 12 analyses, one analysis (MI36) shows a significantly higher Na₂O content (8.9 wt%) than typical for felsic melts (Fig. 4c). In this grain, the host zircon has minute fractures connecting the inclusion with the crystal surface (Supplementary Fig. S2), indicating Na contamination into the melt inclusion during the homogenization experiment through the fracture. Excluding MI36, the melt inclusions have SiO₂ contents of 76-79 wt% and alumina saturation index [A.S.I., molecular Al₂O₃/(CaO + Na₂O + K₂O)] of 1.10-1.53 (Table 1 and Fig. 4e). The SiO₂ contents of the melt inclusions are considerably higher than the whole-rock SiO₂ contents of the host sample (YSR-105A, 72.6 wt% SiO₂). The Al₂O₃ and Na₂O contents of the melt inclusions decrease with increasing SiO₂, whereas CaO and K₂O contents are constant with increasing SiO₂ (excluding MI36). TiO₂, FeO_(total), and MgO contents of the melt inclusions are below the detection limits of the SEM-EDS analysis (Table 1). The estimated water contents range from 6.4 to 11.3 wt% H₂O, which are significantly higher than the measured ignition loss (<1.1 wt%) of granitoids from Yashiro-jima Island (Fig. 4f). The melt inclusions have compositions from granodiorite to granite based on normative mineralogy (excluding MI36, Fig. 4g).

Figure 4. (a) Al₂O₃, (b) CaO, (c) Na₂O, (d) K₂O, (e) alumina saturation index (A.S.I.), and (f) H₂O versus SiO₂ of melt inclusions in zircon in comparison with the whole-rock compositions of Cretaceous granitoids from Yashiro-jima Island. Ignition loss data of the Cretaceous granitoids are also shown in (f) for comparison. (g) An-Ab-Or normative ternary diagram of the melt inclusions and whole-rock compositions. The granitoid classification is after Barker (1979).

DISCUSSION

Compositions and water contents of melt inclusions in zircons

Among the 12 analyses of melt inclusions, one analysis (MI7) shows extremely higher A.S.I. (1.53) than those of other melt inclusions (1.10-1.27, excluding MI36) (Table 1 and Fig. 4e). The measured Na₂O content (0.6 wt%) of MI7 is also significantly lower than those of other melt inclusions (2.7-3.3 wt% Na₂O, excluding MI36) (Table 1 and Fig. 4c). Because loss and/or exchange of alkalis, especially Na, is common in felsic glasses, the melt inclusions with extremely high A.S.I. and very low Na₂O (MI7) likely suffered Na loss by alteration (Taniwaki et al., 2023) although the alteration process of melt inclusions in zircon is not well understood.

Zircon grains have well-developed concentric oscillatory zoning (Fig. 3), consistent with crystallization from melt. The major element contents of the melt inclusions (76-78 wt% SiO₂, excluding MI7 and MI36) are broadly comparable to fractionated (relatively higher SiO₂) granitoids from Yashiro-jima Island (Fig. 4). These compositional characteristics suggest that fractionated melts have been trapped in zircon grains, supported by petrographic observations that zircons are dominantly included in quartz, alkali-feldspar and rim domains of biotite or occur along grain boundaries between main mineral constituents (Fig. 2b).

The estimated water contents (6.4-11.3 wt% H₂O, excluding MI7 and MI36) suggest that zircon crystals trapped hydrous fractionated melts during growth, which is consistent with the presence of voids in the melt inclusions (Fig. 2c). The estimated water contents in this study (6.4-11.3 wt%) are considerably higher than those of glass inclusions in felsic volcanic rocks reported in other studies (e.g., Aira caldera, 3.1-7.0 wt%, Table 2).

Table 2. Examples of previously reported water contents of glass inclusions in felsic volcanic rocks

Location	H2O (wt%)	Method	Reference
Bishop tuff	5.1-6.8 (Plinian deposit); 2.3-4.3 (ash-flow deposit)	Infrared spectroscopy	Anderson et al. (1989)
Pinatubo	6.07-6.58	Infrared spectroscopy	Wallace and Gerlach (1994)
Mont Dore	5.01	Raman spectroscopy	Chabiron et al. (2004)
San Pietro	2.95-3.03	Raman spectroscopy	Chabiron et al. (2004)
Guadeloupe	4.82	Raman spectroscopy	Chabiron et al. (2004)
Aira caldera	5.3-7.0 (Osumi pumice fall); 4.6-7.2 (Ito pyroclastic flow); 3.6-5.1 (Tsumaya pyroclastic flow)	SIMS	Miyagi and Yurimoto (1995)
	3.1-5.4 (Tsumaya pyroclastic flow)	Infrared spectroscopy	Yasuda et al. (2015)
	4.5-4.7 (Osumi pumice fall)	SEM-EDS	Geshi et al. (2017)

Estimation of zircon crystallization pressures

In this study, we tested three geobarometers using major element compositions of melt inclusions, including (1) rhyolite-MELTS geobarometer (Gualda et al., 2012; Gualda and Ghiorso, 2014, 2015), (2) DERP (DEtermining Rhyolite Pressures) geobarometer (Wilke et al., 2017), and (3) MagMaTaB (MAGmatic MAchine learning Thermometry And Barometry) geobarometer (Weber and Blundy, 2024). Both rhyolite-MELTS (1) and DERP (2) geobarometers are based on the pressure dependance of the cotectic curve that separates the stability fields of quartz and feldspar in haplogranitic and natural magmas (Gualda and Ghiorso, 2014; Wilke et al., 2017), whereas MagMaTaB (3) is a new machine learning-based geobarometer calibrated using a large experimental database (Weber and Blundy, 2024).

The rhyolite-MELTS geobarometer (Gualda et al., 2012; Gualda and Ghiorso, 2014, 2015) has been widely applied to estimate pressures using matrix glass and/or glass inclusion compositions of felsic volcanic rocks (e.g., Gualda and Ghiorso, 2013; Bégué et al., 2014a, 2014b; Pamukcu et al., 2015; Gualda et al., 2018, 2019; Pamukçu et al., 2020; Curry et al., 2021; Hartung et al., 2021; Pitcher et al., 2021). Recently, Taniwaki et al. (2023) successfully estimated crystallization pressures of the Miocene Miuchi granitoid pluton in southwest Japan using the compositions of homogenized melt inclusions in zircon and the application of the rhyolite-MELTS geobarometer. In this study, Rhyolite-MELTS simulations (Version 1.1.x) were used to calculate the equilibration pressures of coexisting melt, quartz and feldspar in pressure-temperature space (Supplementary Fig. S3). We used major element compositions and estimated water contents of melt inclusions in the simulations (Table 1). Models were run over a temperature range of 1000-700 °C, with steps of 1 °C, and a pressure range of 500-25 MPa in 25 MPa steps. All simulations were run with oxygen fugacity of the Ni-NiO buffer, although the effect of oxygen fugacity on pressure estimates for rhyolitic compositions is minimal (Gualda and Ghiorso, 2013). Calculations were performed with the Microsoft Excel interface (MELTS_Excel, Gualda and Ghiorso, 2015).

Of the 12 melt inclusion compositions runs with the rhyolite-MELTS geobarometer, 10 simulations yielded pressure estimates ranging from 323 to 59 MPa (with error of ±25 MPa) (Table 1, Supplementary Fig. S3). The melt inclusions with extremely high Na₂O (MI36, Na contamination during high-pressure experiments) and extremely high A.S.I. and very low Na₂O (MI7, Na loss by alteration) yielded no pressure estimates, which support previous arguments that the rhyolite-MELTS geobarometer can identify low-quality data affected by alteration, alkali-loss/exchange and analytical problems (e.g., Bégué et al., 2014b; Pamukcu et al., 2015). The results of rhyolite-MELTS simulation also suggest that the 10 melt compositions have cotectic compositions, which is consistent with the occurrence of zircons along grain boundaries between main mineral phases or as inclusions in quartz, alkali-feldspar, plagioclase, and biotite rims indicating that zircon grains crystallized from melts saturated in quartz and feldspar.

We also applied the DERP geobarometer (Wilke et al., 2017), which considers the effects of normative anorthite and water contents on the position of cotectic curve separating the stability fields of quartz and feldspar in the granitic system. Despite Taniwaki et al. (2023) could not test the DERP geobarometer because of the lack of water content estimation, our data sets including major element compositions together with water contents of melt inclusions allowed us to apply the DERP geobarometer. The results of the DERP geobarometer were 403-71 MPa (excluding MI7 and MI36, Table 1), broadly consistent with the results of the rhyolite-MELTS geobarometer (Supplementary Fig. S4a). The melt inclusions with low-quality data (MI36 and MI7) resulted in unrealistic pressure estimates (777 and −602 MPa, respectively, Table 1).

Although the pressures calculated by two independent geobarometers yielded similar results, the effect of alumina should be considered for strongly peraluminous compositions because both the rhyolite-MELTS and the DERP geobarometers are not calibrated for strongly peraluminous melts (Gualda et al., 2012; Wilke et al., 2017). Although it is well known that the quartz-feldspar cotectic curve shifts towards more quartz-rich compositions with decreasing pressures (see Johannes and Holtz, 1996), a similar relationship is also observed for strongly peraluminous melts (Johannes and Holtz, 1996). Therefore, the strongly peraluminous compositions in this study (A.S.I. = 1.10-1.27, Table 1) indicate that the estimated pressures (323-59 MPa by rhyolite-MELTS and 403-71 MPa by DERP) likely represent minimum pressures of zircon crystallization. This interpretation is consistent with the lower crystallization pressures compared to the metamorphic pressures estimated from the adjacent Ryoke metamorphic rocks (529-420 MPa, YFE3 of Ikeda, 2004, Fig. 1) and from southwestern Yashiro-jima Island (690-517 MPa, YFC7A, YFS7B, and YFD5 of Ikeda, 2004, Fig. 1) (Supplementary Fig. S4a), despite the Gamano granodiorite exhibits a concordant intrusive relation to adjacent Ryoke metamorphic rocks (Fig. 2a).

The MagMaTaB geobarometer is applicable to a wide range of melt compositions from basalt to rhyolite (Weber and Blundy, 2024). This geobarometer yielded pressures of 563-266 MPa (with root-mean-square-error of 110-130 MPa) for the compositions in this study (excluding MI7 and MI36, Table 1), assuming the coexisting mineral assemblage of quartz, plagioclase, K-feldspar, and oxide. Compared to the estimates from rhyolite-MELTS and DERP geobarometers, the pressures from MagMaTaB are more comparable to the metamorphic pressures. It is noteworthy that the MagMaTaB geobarometer is calibrated predominantly based on peraluminous experimental melts for rhyolitic compositions, comparable to the melt inclusions in this study (Supplementary Fig. S4b). On this basis, the estimates from the MagMaTaB geobarometer (563-266 MPa) are interpreted to be a more reliable indicator of the zircon crystallization pressures in this study.

Geological implications of the estimated water contents and pressures

The relationship between estimated water contents and pressures are shown in Figure 5, with a comparison of the H₂O solubility curve of haplogranitic melts (Johannes and Holtz, 1996). The average of the 10 compositions obtained in this study plot along the H₂O solubility curve (Fig. 5), suggesting that the interstitial melts trapped in zircon are close to water saturation. We interpret this as evidence for high water activity of unerupted felsic melts that formed the Gamano granodiorite.

Figure 5. Estimated water contents versus pressures for the melt inclusions. The H₂O solubility curve of haplogranitic melts (broken curve, Johannes and Holtz, 1996) is shown for comparison. Metamorphic pressure conditions estimated for Ryoke metamorphic rocks (red arrows) from Yashiro-jima Island (YFE3, YFC7A, YFC7B, and YFD5, Ikeda, 2004; Fig. 1) are also shown for comparison.

There is a wide range of estimated water contents and pressures (Fig. 5). These variations can be interpreted in two ways: (1) zircon crystallized over a range of pressures (i.e., variable depths) and trapped melts with variable water contents during the assent of the Gamano granodiorite magma, or (2) analytical uncertainty due to the small size of the melt inclusions resulted in variations of estimated water contents and pressures, such that those variations are geologically meaningless. Although the possibility of (2) cannot be ruled out, there are no significant correlations between estimated water contents/pressures and the magnification of analysis (i.e., size of analysis area, Table 1). Therefore, we interpret that the wide range of estimated pressures (563-266 MPa) represent variable depths from deeper crustal levels (i.e., initial zircon crystallization) to shallower levels (i.e., final magma solidification). It has been previously suggested that the emplacement of the Gamano granodiorite was contemporaneous with peak metamorphism (e.g., Okudaira et al., 1995; Okudaira, 1996; Suzuki and Adachi, 1998; Skrzypek et al., 2016; Ikeda et al., 2019). Concordant intrusive relations of the studied Gamano granodiorite (YSR-105A) to the adjacent Ryoke metamorphic rocks (Fig. 2a) that record metamorphic pressures of ∼ 529-420 MPa (Ikeda, 2004, Fig. 1) also support magma intrusion at deeper crustal levels.

In addition, the pressures estimated in this study are considerably higher than those from the Miocene Miuchi granitoid pluton in southwest Japan. The pressure conditions estimated for the Miuchi pluton based on melt inclusions in zircons (∼ 114-80 MPa) are consistent with field and petrographic observations suggestive of a shallow emplacement level (Taniwaki et al., 2023). The results of this study illustrate that the melt inclusions in zircon can be used for geobarometry (Taniwaki et al., 2023) working nicely with various pressures as well as for geohygrometry for the melts in unerupted felsic magmas.

SUMMARY

Water contents of melt that form felsic plutonic rocks have been difficult to directly determine because holocrystalline plutonic rocks lack matrix glass and/or glass inclusions. This paper highlights a technique using SEM-EDS and homogenized melt inclusions hosted in zircon from a granitoid sample to successfully constrain water contents of melts together with zircon crystallization pressures. The high water (6.4-11.3 wt%) and high SiO₂ (76-78 wt%) contents of the melt inclusions suggest that zircon crystals trapped fractionated hydrous melts during growth. The melt inclusion compositions yielded pressures ranging from 563 to 266 MPa (based on the MagMaTaB geobarometer) which probably represent zircon crystallization pressures from magma intrusion at deeper crustal levels (∼ 563-509 MPa) to final solidification at shallower levels (∼ 266 MPa). Because zircon is a ubiquitous accessory mineral in most felsic rock types, the approach presented here could apply to a wide range of zircon-bearing granitoids, which would provide essential data to better understand the physicochemical conditions (pressure, melt composition, and water contents) of melts in unerupted felsic magma within orogenic belts.

ACKNOWLEDGMENTS

We would like to thank two anonymous reviewers for their constructive comments, and Shumpei Yoshimura for editorial handling. We are grateful to Tomohiro Ohuchi, Hiroaki Ohfuji and Mayuko Fukuyama for their support with CL analysis. We are also grateful to Yoshimitsu Suda for providing an obsidian sample (HOb-202-1) which we used as a secondary standard for EDS analysis. We thank Yu Nishihara for his help with FT-IR micro-reflectance spectroscopy. We also thank Yuka Taniwaki for her supports for zircon melt inclusion study. We also thank Fawna J. Korhonen and Vivek P. Malviya for helping to improve our English. This study was partially supported by JSPS KAKENHI Grant Numbers JP 21J22080 to K.S. and JP 22H01323 to S.S. This work was also supported by the Joint Usage/Research Center PRIUS, Ehime University, Japan.

SUPPLEMENTARY MATERIALS

Supplementary Figures S1-S4 and Table S1 are available online from https://doi.org/10.2465/jmps.230904.

REFERENCES

•  Anderson,  A.T. Jr.,  Newman,  S.,  Williams,  S.N.,  Druitt,  T.H., et al (1989) H₂O, CO₂, Cl, and gas in Plinian and ash-flow Bishop rhyolite. Geology, 17, 221-225.
• Barker, F. (1979) Trondhjemite: Definition, environment and hypotheses of origin. In Trondhjemites, dacites and related rocks (Barker, F. Ed.). pp. 659, Developments in Petrology, 6, Elsevier, Amsterdam, 1-12.
•  Bégué,  F.,  Deering,  C.D.,  Gravley,  D.M.,  Kennedy,  B.M., et al (2014a) Extraction, storage and eruption of multiple isolated magma batches in the paired Mamaku and Ohakuri eruption, Taupo Volcanic Zone, New Zealand. Journal of Petrology, 55, 1653-1684.
•  Bégué,  F.,  Gualda,  G.A.R.,  Ghiorso,  M.S.,  Pamukcu,  A.S., et al (2014b) Phase-equilibrium geobarometers for silicic rocks based on rhyolite-MELTS. Part 2: application to Taupo Volcanic Zone rhyolites. Contributions to Mineralogy and Petrology, 168, 1082.
•  Chabiron,  A.,  Pironon,  J. and  Massare,  D. (2004) Characterization of water in synthetic rhyolitic glasses and natural melt inclusions by Raman spectroscopy. Contributions to Mineralogy and Petrology, 146, 485-492.
•  Curry,  A.,  Caricchi,  L. and  Lipman,  P.W. (2021) Magmatic evolution of zoned and unzoned ignimbrites: Evidence for a complex crustal architecture feeding four rapid-sequence, caldera-forming eruptions in the San Juan Mountains, Colorado. Journal of Petrology, 62, egab006.
•  Geshi,  N.,  Miyagi,  I. and  Saito,  G. (2017) Estimation of water concentration in volcanic glass using energy dispersal X-ray spectrometry. Bulletin of the Volcanological Society of Japan, 62, 13-22.
•  Gualda,  G.A.R.,  Ghiorso,  M.S.,  Lemons,  R.V. and  Carley,  T.L. (2012) Rhyolite-MELTS: a modified calibration of MELTS optimized for silica-rich, fluid-bearing magmatic systems. Journal of Petrology, 53, 875-890.
•  Gualda,  G.A.R. and  Ghiorso,  M.S. (2013) The Bishop Tuff giant magma body: an alternative to the Standard Model. Contributions to Mineralogy and Petrology, 166, 755-775.
•  Gualda,  G.A.R. and  Ghiorso,  M.S. (2014) Phase-equilibrium geobarometer for silicic rocks based on rhyolite-MELTS. Part 1: Principles, procedures, and evaluation of the method. Contributions to Mineralogy and Petrology, 168, 1033.
•  Gualda,  G.A.R. and  Ghiorso,  M.S. (2015) MELTS_Excel: A Microsoft Excel-based MELTS interface for research and teaching of magma properties and evolution. Geochemistry, Geophysics, Geosystems, 16, 315-324.
•  Gualda,  G.A.R.,  Gravley,  D.M.,  Connor,  M.,  Hollmann,  B., et al (2018) Climbing the crustal ladder: magma storage-depth evolution during a volcanic flare-up. Science Advances, 4, eaap7567.
•  Gualda,  G.A.R.,  Gravley,  D.M.,  Deering,  C.D. and  Ghiorso,  M.S. (2019) Magma extraction pressures and the architecture of volcanic plumbing systems. Earth and Planetary Science Letters, 522, 118-124.
•  Gudelius,  D.,  Zeh,  A.,  Almeev,  R.R.,  Wilson,  A.H., et al (2020) Zircon melt inclusions in mafic and felsic rocks of the Bushveld Complex - Constraints for zircon crystallization temperatures and partition coefficients. Geochimica et Cosmochimica Acta, 289, 158-181.
•  Hartung,  E.,  Caricchi,  L.,  Floess,  D.,  Wallis,  S. and  Harayama,  S. (2021) Establishing genetic relationships between the Takidani pluton and two large silicic eruptions in the Northern Japan Alps. Journal of Petrology, 62, egab085.
•  Herzig,  C.T.,  Kimbrough,  D.L.,  Tainosho,  Y.,  Kagami,  H., et al (1998) Late Cretaceous U/Pb zircon ages and Precambrian crustal inheritance in Ryoke granitoids, Kinki and Yanai districts, Japan. Geochemical Journal, 32, 21-31.
•  Ikeda,  T. (1998) Progressive sequence of reactions of the Ryoke metamorphism in the Yanai district, southwest Japan: the formation of cordierite. Journal of Metamorphic Geology, 16, 39-52.
•  Ikeda,  T. (2004) Pressure-temperature conditions of the Ryoke metamorphic rocks in Yanai district, SW Japan. Contributions to Mineralogy and Petrology, 146, 577-589.
•  Ikeda,  Y.,  Owada,  M.,  Nishizuka,  D. and  Kamei,  A. (2019) Magma process of Gamano granodiorite in Ryoke belt, Yanai region, Yamaguchi, Southwest Japan. The Journal of the Geological Society of Japan, 125, 167-182 (in Japanese with English abstract).
• Johannes, W. and Holtz, F. (1996) Petrogenesis and Experimental Petrology of Granitic Rocks. pp. 335, Springer, Berlin.
•  Kodama,  S.,  Owada,  M.,  Kamei,  A.,  Ikeda,  Y. and  Imuta,  J. (2021) Magmatic processes of Ryoke granitoids from Yashiro-jima Island, Yamaguchi Prefecture, SW Japan. Japanese Magazine of Mineralogical and Petrological Sciences, 49, 133-147 (in Japanese with English abstract).
•  Miyagi,  I. and  Yurimoto,  H. (1995) Water content of melt inclusions in phenocrysts using secondary ion mass spectrometer. Bulletin of the Volcanological Society of Japan, 40, 349-355.
•  Okudaira,  T. (1996) Thermal evolution of the Ryoke metamorphic belt, southwestern Japan: tectonic and numerical modeling. The Island Arc, 5, 373-385.
•  Okudaira,  T.,  Hara,  I. and  Takeshita,  T. (1995) Emplacement mechanism of the older Ryoke granites in the Yanai district, southwest Japan, with special reference to extensional deformation in the Ryoke metamorphic belt. Journal of Science of the Hiroshima University, Series C, 10, 357-366.
•  Pamukcu,  A.S.,  Gualda,  G.A.R.,  Ghiorso,  M.S.,  Miller,  C.F. and  McCracken,  R.G. (2015) Phase-equilibrium geobarometers for silicic rocks based on rhyolite-MELTS—Part 3: Application to the Peach Spring Tuff (Arizona-California-Nevada, USA). Contributions to Mineralogy and Petrology, 169, 33.
•  Pamukçu,  A.S.,  Wright,  K.A.,  Gualda,  G.A.R. and  Gravley,  D. (2020) Magma residence and eruption at the Taupo Volcanic Center (Taupo Volcanic Zone, New Zealand): insights from rhyolite-MELTS geobarometry, diffusion chronometry, and crystal textures. Contributions to Mineralogy and Petrology, 175, 48.
•  Pitcher,  B.W.,  Gualda,  G.A.R. and  Hasegawa,  T. (2021) Repetitive duality of rhyolite compositions, timescales, and storage and extraction conditions for pleistocene caldera-forming eruptions, Hokkaido, Japan. Journal of Petrology, 62, egaa106.
•  Skrzypek,  E.,  Kawakami,  T.,  Hirajima,  T.,  Sakata,  S., et al (2016) Revisiting the high temperature metamorphic field gradient of the Ryoke Belt (SW Japan): New constraints from the Iwakuni-Yanai area. Lithos, 260, 9-27.
•  Suda,  Y. and  Saito,  S. (2018) A method of X-ray fluorescence analysis of major and trace elements in silicate rocks by fused-bead method. Memoirs of the Faculty of Science. Ehime University, 21, 1-11 (in Japanese with English abstract).
•  Suzuki,  K. and  Adachi,  M. (1998) Denudation history of the high T/P Ryoke metamorphic belt, southwest Japan: constraints from CHIME monazite ages of gneisses and granitoids. Journal of Metamorphic Geology, 16, 23-37.
•  Taniwaki,  Y.,  Shimooka,  K. and  Saito,  S. (2023) Zircon melt inclusion study on the Miocene Miuchi granitoid pluton, SW Japan: A new approach to estimate crystallization pressures of granitic rocks. Lithos, 454-455, 107260.
•  Thomas,  J.B.,  Bodnar,  R.J.,  Shimizu,  N. and  Chesner,  C.A. (2003) Melt inclusions in zircon. Reviews in Mineralogy and Geochemistry, 53, 63-87.
•  Wallace,  P.J. and  Gerlach,  T.M. (1994) Magmatic vapor source for sulfur dioxide released during volcanic eruptions: Evidence from Mount Pinatubo. Science, 265, 497-499.
•  Weber,  G. and  Blundy,  J. (2024) A machine learning-based thermobarometer for magmatic liquids. Journal of Petrology, 65, egae020.
•  Wilke,  S.,  Holtz,  F.,  Neave,  D.A. and  Almeev,  R. (2017) The effect of anorthite content and water on quartz-feldspar cotectic compositions in the rhyolitic system and implications for geobarometry. Journal of Petrology, 58, 789-818.
•  Yasuda,  A. (2014) A new technique using FT-IR micro-reflectance spectroscopy for measurement of water concentrations in melt inclusions. Earth, Planets and Space, 66, 34.
•  Yasuda,  A.,  Yoshimoto,  M. and  Fujii,  T. (2015) The depth of a magma chamber associated with the Aira caldera formation. Bulletin of the Volcanological Society of Japan, 60, 381-397 (in Japanese with English abstract).