TY - JOUR
AU - Neill, Owen K
AB - Abstract Twenty-four internally heated pressure vessel experiments were conducted at 810–860°C, 1·5–4·05 kbar, and oxygen fugacities (fO2) = NNO –0·5 to NNO +2 log units (where NNO is the nickel–nickel oxide buffer), using hydrous rhyodacitic starting glasses from Mt Usu, Japan. Aqueous solutions were added to experimental charges such that the volatile phase(s) coexisting with the crystalline phases (amphibole, plagioclase, and clinopyroxene) at run conditions buffered the F, Cl, S, and CO2 concentrations in the melt. The resultant phenocryst phases and glass chemistries were analysed by electron microprobe and Fourier transform infrared spectroscopy, and final fluid Cl– contents by chloridometer. All experiments produced homogeneous glasses and large euhedral phenocrysts with minimal compositional zonation. The results of the crystal–melt partitioning data are applicable to understanding the geochemical evolution of rhyodacitic melts at fluid-saturated shallow crustal levels, the composition of fluid phases later exsolved owing to second boiling, and the effects of halogens on amphibole crystal chemistry. The residual glasses from these experiments span the dacite–rhyolite compositional join, and are hydrous (>5 wt % H2O) with F concentrations from ≤100 ppm to 0·63 wt % and Cl from ∼130 ppm to 0·72 wt %. Measured final fluid Cl contents show that Cl strongly prefers the fluid phase over the melt phase in all experiments, with DClfluid/melt ranging from 3·5 to 22·7. Amphibole compositions are calcic, Mg-rich, and typical of those found in natural calc-alkaline arc magmas. They are particularly sensitive to changes in melt halogen chemistry, with maximum amphibole F contents of 2·59 wt % (maximum DFamph/melt values of ≥15) and maximum Cl contents of 0·12 wt % (maximum DClamph/melt values of 0·40). Integration of the amphibole data with other experimental data shows that Cl incorporation is a strong function of the Mg–Cl crystallographic avoidance principle, and that addition of F to the melt strongly decreases Cl partitioning at equivalent Mg# (DClamph/melt = 0·10 ± 0·02 in experiments that were not F-doped, compared with DClamph/melt = 0·05 ± 0·01 in F-doped experiments). Plagioclase compositions are relatively restricted, with anorthite contents An59–An39, and clinopyroxene is similarly calcic, containing a significant enstatite component (Wo49–Wo30). The behaviour of Fe in the glass and crystalline phases is most significantly affected by the fO2 of the experiment. Higher total Fe contents are found in amphibole and clinopyroxene from experiments with fO2 < NNO +1, but with little effect on the absolute DFeOmineral/melt. Plagioclase, however, shows relatively decreased DFeOplag/melt reflecting its preferential incorporation of Fe3+, and less Fe3+ expected in melts crystallized at lower fO2 conditions. Comparison of the data with recently formulated amphibole geothermometers, barometers, and plagioclase hygrometers shows calculated results that are consistent with the actual experimental conditions but discrepancies arise owing to halogen-induced major element variations in amphibole, indicating the importance of routinely measuring F and Cl in both natural and experimental amphibole. INTRODUCTION The eruptive and intrusive products from magmas of bulk dacitic composition are common in global arc settings and often record evidence for multi-stage ascent and crystallization within complex plumbing systems (e.g. Cottrell et al., 1999; Costa et al., 2004; Botcharnikov et al., 2008; Kent et al., 2010). Experimental phase equilibria studies aid in deciphering the differences in petrogenetic histories of these magmas that are reflected in bulk magma chemistry, volatile contents, and the compositional evolution of the major phenocryst phases (e.g. Holtz et al., 2005; Sato et al., 2005; Muir et al., 2014; Riker et al., 2015; Andújar et al., 2016). Furthermore, such arc magmas are usually saturated in multiple phenocryst phases (such as amphibole, pyroxene, plagioclase, and Fe–Ti oxides) and may be transporting antecrysts as a result of earlier magma mixing (e.g. Halter et al., 2004; Humphreys et al., 2010; Tepley et al., 2013; Laumonier et al., 2014, and references therein). This study therefore aims to better understand and contrast the controls on the partitioning behaviour and covariations in major elements and halogens, through examination of the crystallization of a natural rhyodacite melt from Mt Usu, Japan, as a function of temperature, pressure, oxygen fugacity (fO2), and melt composition. In particular, this work examines the compositions of calcic amphibole, intermediate-An plagioclase, and clinopyroxene in equilibrium with rhyodacitic melts, containing volatile and halogen contents geologically relevant to intermediate-evolved arc magmas. Experimental conditions were chosen to closely reflect the environment of mid- to upper-crustal magma storage systems, consistent with the temperature, pressure and fO2 characteristics of active arc volcanoes and their often associated hydrothermal ore deposits (e.g. Vigneresse, 2007; Richards, 2011). The results of these experiments are applicable to constraining the geochemical evolution of shallow crustal magma storage systems typical of rhyodacitic magmas, as a function of differentiation of multiple phenocryst phases. The results also aid in the identification of antecrysts sourced from other magma batches, based on their compositions. Additionally, the study presents the first systematic data for F partitioning between amphibole and rhyodacite melt, which is an important factor when attempting to reconstruct the sequences of volatile exsolution associated with ‘second boiling’ and the F/Cl/H2O ratios of the exsolved fluids (e.g. Baker & Alletti, 2012, and references therein). Finally, the data for amphibole and plagioclase are also interpreted in conjunction with recently developed and updated geobarometers, geothermometers, and geohygrometers (e.g. Waters & Lange, 2015; Putirka, 2016) as a test of the experimental calibrations used in these models. The convention used henceforth is that for an element i, the Nernst-type partition coefficient is calculated by Dimineral/melt=Cimineral/Cimelt where Di is the partition coefficient for element i between the mineral phase and melt phase, and Cimineral and Cimelt are the concentrations of element i in those phases (in weight per cent), respectively, coexisting at equilibrium (Beattie et al., 1993). All errors reported on the partition coefficients are calculated through error propagation accounting for the standard deviation on the analytical analyses of the glasses and phenocrysts. EXPERIMENTAL METHODS Preparation of starting materials The starting glasses for these partitioning investigations were prepared from a natural nominally anhydrous rhyodacite sample (NMNH sample 108980-14) from Mt Usu, Japan (see Tomiya & Takahashi, 2005) (Table 1). The natural sample was crushed and powdered using an agate mortar and pestle and acetone. The raw Usu powder was loaded into Au80Pd20 capsules with sufficient aqueous fluid to saturate the melt composition with 6 wt % H2O, and fused in the internally heated pressure vessel (IHPV) at the American Museum of Natural History (AMNH), New York. Run conditions were 1050°C, 2·2 kbar and oxygen fugacity (fO2) ≈ NNO +2·0 (where NNO is nickel–nickel oxide buffer) for ∼100 h, and the runs were quenched isobarically (see ‘Experimental conditions’ section below). After this duration, the hydrated product glass was extracted, crushed and reloaded into another Au80Pd20 capsule with no addition of fluid, and brought back to the same run conditions for a further ∼60 h, giving a total of ∼160 h. Table 1: Major element compositions of the raw Usu dacite sample and experimental starting glasses Glass ID: . Raw Usu . 1-13-15 . 1-14-12 . 1-15-16 . 1-15-17 . 1-16-20 . wt % SiO2 70·00 67·05 66·11 66·23 65·44 68·31 TiO2 0·40 0·39 0·36 0·31 0·36 0·21 Al2O3 15·04 14·51 14·57 14·40 14·36 14·28 FeO* 3·73 3·48 2·96 2·88 3·39 2·01 MnO 0·16 0·23 0·22 0·14 0·14 0·18 MgO 0·84 0·89 0·82 0·80 0·83 0·88 CaO 3·71 3·70 3·56 3·44 3·58 3·50 Na2O 4·77 4·21 4·22 4·34 4·32 4·39 K2O 0·93 0·88 0·89 0·90 0·87 0·91 P2O5 0·15 0·15 0·15 0·14 0·15 0·11 SO3 0·00 0·00 0·00 0·02 0·02 0·05 Cl no data 0·03 0·03 0·10 0·10 0·10 F no data 0·05 0·03 0·02 0·03 0·03 H2O by diff. 0·27 4·43 6·07 6·30 6·42 5·08 Total 99·73 95·57 93·93 93·70 93·58 94·92 Molar Na/(Na + K) 0·89 0·88 0·88 0·88 0·88 0·88 Molar Al/(Ca + Na + K) 0·96 0·99 1·01 1·00 0·99 0·98 Glass ID: . Raw Usu . 1-13-15 . 1-14-12 . 1-15-16 . 1-15-17 . 1-16-20 . wt % SiO2 70·00 67·05 66·11 66·23 65·44 68·31 TiO2 0·40 0·39 0·36 0·31 0·36 0·21 Al2O3 15·04 14·51 14·57 14·40 14·36 14·28 FeO* 3·73 3·48 2·96 2·88 3·39 2·01 MnO 0·16 0·23 0·22 0·14 0·14 0·18 MgO 0·84 0·89 0·82 0·80 0·83 0·88 CaO 3·71 3·70 3·56 3·44 3·58 3·50 Na2O 4·77 4·21 4·22 4·34 4·32 4·39 K2O 0·93 0·88 0·89 0·90 0·87 0·91 P2O5 0·15 0·15 0·15 0·14 0·15 0·11 SO3 0·00 0·00 0·00 0·02 0·02 0·05 Cl no data 0·03 0·03 0·10 0·10 0·10 F no data 0·05 0·03 0·02 0·03 0·03 H2O by diff. 0·27 4·43 6·07 6·30 6·42 5·08 Total 99·73 95·57 93·93 93·70 93·58 94·92 Molar Na/(Na + K) 0·89 0·88 0·88 0·88 0·88 0·88 Molar Al/(Ca + Na + K) 0·96 0·99 1·01 1·00 0·99 0·98 Open in new tab Table 1: Major element compositions of the raw Usu dacite sample and experimental starting glasses Glass ID: . Raw Usu . 1-13-15 . 1-14-12 . 1-15-16 . 1-15-17 . 1-16-20 . wt % SiO2 70·00 67·05 66·11 66·23 65·44 68·31 TiO2 0·40 0·39 0·36 0·31 0·36 0·21 Al2O3 15·04 14·51 14·57 14·40 14·36 14·28 FeO* 3·73 3·48 2·96 2·88 3·39 2·01 MnO 0·16 0·23 0·22 0·14 0·14 0·18 MgO 0·84 0·89 0·82 0·80 0·83 0·88 CaO 3·71 3·70 3·56 3·44 3·58 3·50 Na2O 4·77 4·21 4·22 4·34 4·32 4·39 K2O 0·93 0·88 0·89 0·90 0·87 0·91 P2O5 0·15 0·15 0·15 0·14 0·15 0·11 SO3 0·00 0·00 0·00 0·02 0·02 0·05 Cl no data 0·03 0·03 0·10 0·10 0·10 F no data 0·05 0·03 0·02 0·03 0·03 H2O by diff. 0·27 4·43 6·07 6·30 6·42 5·08 Total 99·73 95·57 93·93 93·70 93·58 94·92 Molar Na/(Na + K) 0·89 0·88 0·88 0·88 0·88 0·88 Molar Al/(Ca + Na + K) 0·96 0·99 1·01 1·00 0·99 0·98 Glass ID: . Raw Usu . 1-13-15 . 1-14-12 . 1-15-16 . 1-15-17 . 1-16-20 . wt % SiO2 70·00 67·05 66·11 66·23 65·44 68·31 TiO2 0·40 0·39 0·36 0·31 0·36 0·21 Al2O3 15·04 14·51 14·57 14·40 14·36 14·28 FeO* 3·73 3·48 2·96 2·88 3·39 2·01 MnO 0·16 0·23 0·22 0·14 0·14 0·18 MgO 0·84 0·89 0·82 0·80 0·83 0·88 CaO 3·71 3·70 3·56 3·44 3·58 3·50 Na2O 4·77 4·21 4·22 4·34 4·32 4·39 K2O 0·93 0·88 0·89 0·90 0·87 0·91 P2O5 0·15 0·15 0·15 0·14 0·15 0·11 SO3 0·00 0·00 0·00 0·02 0·02 0·05 Cl no data 0·03 0·03 0·10 0·10 0·10 F no data 0·05 0·03 0·02 0·03 0·03 H2O by diff. 0·27 4·43 6·07 6·30 6·42 5·08 Total 99·73 95·57 93·93 93·70 93·58 94·92 Molar Na/(Na + K) 0·89 0·88 0·88 0·88 0·88 0·88 Molar Al/(Ca + Na + K) 0·96 0·99 1·01 1·00 0·99 0·98 Open in new tab This multiple-fusion procedure ensures that the hydrous starting glasses are as homogeneous as possible (Table 1), which was confirmed by major and minor element analysis by electron microprobe (see ‘Analytical methods’ section below). Finally, the starting glasses were crushed again and loaded into the capsules to be used in the crystallization experiments. Note that three experiments were conducted using the raw Mt. Usu powder as the starting matieral, without the pre-hydration step. Preparation of experimental capsules Pure Au, Au75Pd25, or approximately Au95Cu5 alloy capsules 1·5–3·0 cm in length with an outer diameter of 3 or 5 mm and a wall thickness of 1·1 or 1·9 mm were cleaned, bottom crimped, and welded shut using a PUK-3 tungsten-tipped gas flow arc welder. Varying masses of distilled water and aqueous alkali-chloride ± alkali-sulfate ± oxalic acid solutions ± CaF2 powder were added to the capsule, followed by 35–50 mg of hydrous starting glass powders (Table 2). The open end of the capsule was then crimped shut and welded while the bottom end was immersed in a water bath. Use of this cooling and welding technique, with the fine power controls of the welder, minimizes the potential for volatilization of the solutions during preparation. The combined mass of all of the components in the finished capsules was then recorded, and the capsule was placed in a 1 atm oven at ∼120°C for 1 h. Once removed, if the capsule showed mass losses of >1 mg it was discarded and not used for these experiments. Table 2: Experimental run conditions, fluid compositions, and run product phases Run ID . Starting powder . T (°C) . P (bar) . Duration (h) . log f O2ΔNNO . log fO2 . Fluid/melt mass ratio . Cl content initial fluid (wt %) . Cl content final fluid (wt %) . 1-13-23 raw Usu 850 2206 360 2·1 –10·7 0·06 — — 1-14-02 1-13-15 850 2190 385 2·1 –10·7 0·01 — — 1-14-04A 1-13-15 840 1500 480 –0·4 –13·4 0·04 — — 1-14-04B 1-13-15 840 1500 480 –0·4 –13·4 0·04 14·01 3·01 1-14-04C 1-13-15 840 1500 480 –0·4 –13·4 0·03 — — 1-14-05 1-13-15 840 2137 460 1·9 –11·1 0·44 9·38 7·99 1-15-12A raw Usu 835 3633 336 2·2 –10·7 0·07 — — 1-15-12B raw Usu 835 3633 336 2·2 –10·7 0·07 5·58 2·5† 1-15-15A 1-14-12 835 2947 671 –0·3 –13·5 0·10 7·38 3·31 1-15-15B 1-14-12 835 2947 671 –0·3 –13·5 0·05 — — 1-15-19B 1-15-17 835 2206 671 1·9 –11·2 0·04 14·01 4·51 1-16-02 1-15-17 845 2302 355 –0·5 –13·4 0·46 16·11 5·38 1-16-12A 1-15-17 810 3033 406 2·2 –11·4 0·07 7·20 4·52 1-16-12B 1-15-17 810 3033 406 2·2 –11·4 0·08 4·29 2·37 1-16-14A 1-15-16 850 2200 551 2·1 –10·7 0·03 14·01 no data 1-16-14B 1-15-16 850 2200 551 2·1 –10·7 0·02 — — 1-16-18A 1-15-17 820 2900 326 2·2 –11·2 0·21 7·11 7·79 1-16-18B 1-15-17 820 2900 326 2·2 –11·2 0·18 5·18 4·97 1-16-21A* 1-16-20 835 2400 480 –0·4 –13·6 0·05 14·01 1·12 1-16-21B* 1-16-20 835 2400 480 –0·4 –13·6 0·05 2·66 no data 1-16-22A* 1-16-20 840 2650 429 –0·4 –13·5 0·08 6·80 3·03 1-16-22B* 1-16-20 840 2650 429 –0·4 –13·5 0·03 14·01 1·7 1-16-23A* 1-16-20 860 4050 312 –0·5 –13·0 0·12 — — 1-16-23B* 1-16-20 860 4050 312 –0·5 –13·0 0·10 — — Run ID . Starting powder . T (°C) . P (bar) . Duration (h) . log f O2ΔNNO . log fO2 . Fluid/melt mass ratio . Cl content initial fluid (wt %) . Cl content final fluid (wt %) . 1-13-23 raw Usu 850 2206 360 2·1 –10·7 0·06 — — 1-14-02 1-13-15 850 2190 385 2·1 –10·7 0·01 — — 1-14-04A 1-13-15 840 1500 480 –0·4 –13·4 0·04 — — 1-14-04B 1-13-15 840 1500 480 –0·4 –13·4 0·04 14·01 3·01 1-14-04C 1-13-15 840 1500 480 –0·4 –13·4 0·03 — — 1-14-05 1-13-15 840 2137 460 1·9 –11·1 0·44 9·38 7·99 1-15-12A raw Usu 835 3633 336 2·2 –10·7 0·07 — — 1-15-12B raw Usu 835 3633 336 2·2 –10·7 0·07 5·58 2·5† 1-15-15A 1-14-12 835 2947 671 –0·3 –13·5 0·10 7·38 3·31 1-15-15B 1-14-12 835 2947 671 –0·3 –13·5 0·05 — — 1-15-19B 1-15-17 835 2206 671 1·9 –11·2 0·04 14·01 4·51 1-16-02 1-15-17 845 2302 355 –0·5 –13·4 0·46 16·11 5·38 1-16-12A 1-15-17 810 3033 406 2·2 –11·4 0·07 7·20 4·52 1-16-12B 1-15-17 810 3033 406 2·2 –11·4 0·08 4·29 2·37 1-16-14A 1-15-16 850 2200 551 2·1 –10·7 0·03 14·01 no data 1-16-14B 1-15-16 850 2200 551 2·1 –10·7 0·02 — — 1-16-18A 1-15-17 820 2900 326 2·2 –11·2 0·21 7·11 7·79 1-16-18B 1-15-17 820 2900 326 2·2 –11·2 0·18 5·18 4·97 1-16-21A* 1-16-20 835 2400 480 –0·4 –13·6 0·05 14·01 1·12 1-16-21B* 1-16-20 835 2400 480 –0·4 –13·6 0·05 2·66 no data 1-16-22A* 1-16-20 840 2650 429 –0·4 –13·5 0·08 6·80 3·03 1-16-22B* 1-16-20 840 2650 429 –0·4 –13·5 0·03 14·01 1·7 1-16-23A* 1-16-20 860 4050 312 –0·5 –13·0 0·12 — — 1-16-23B* 1-16-20 860 4050 312 –0·5 –13·0 0·10 — — Run ID . S content initial fluid (wt %) . Notes . Phases identified (number = vol. %) . 1-13-23 — H2O only gl (80), plag (12), oxi (4), amph (3), apa (trace) 1-14-02 — H2O only gl (86), plag (11), amph (2), oxi (2) 1-14-04A — fO2 controlled, H2O only gl (92), plag (5), cpx (2), oxi (1) 1-14-04B — fO2 controlled, Cl-bearing gl (92), plag (6), cpx (2), oxi (trace) 1-14-04C — fO2 controlled, F-bearing + H2O gl (89), plag (5), amph (4), oxi (2) 1-14-05 — Cl-bearing + H2O gl (94), amph (5), oxi (trace) 1-15-12A — H2O only gl (87), plag (5), amph (4), oxi (4) 1-15-12B — Cl-bearing + H2O gl (75), plag (14), amph (6), oxi (5) 1-15-15A — fO2 controlled, Cl-bearing gl (94), amph (3), plag (2), oxi (1) 1-15-15B — fO2 controlled, H2O only gl (89), plag (5), amph (4), oxi (2) 1-15-19B — Cl-bearing + H2O gl (92), plag (5), amph (2), cpx (1), oxi (trace) 1-16-02 0·44 fO2 controlled, Cl-bearing, S-bearing gl (94), cpx (5), oxi (trace) 1-16-12A — Cl-bearing + H2O gl (88), amph (7), plag (3), oxi (2) 1-16-12B 1·74 Cl-bearing, S-bearing gl (96), plag (3), amph (1), oxi (1) 1-16-14A — Cl-bearing, F-bearing gl (97), amph (2), oxi (1) 1-16-14B — F-bearing + H2O gl (91), plag (3), oxi (3), amph (2), cpx (1) 1-16-18A — Cl-bearing + H2O gl (88), amph (10), oxi (2) 1-16-18B — Cl-bearing + H2O gl (87), amph (10), oxi (3) 1-16-21A* — fO2 controlled, Cl-bearing + H2O gl (92), amph (7), oxi (1) 1-16-21B* — fO2 controlled, Cl-bearing + H2O gl (92), amph (5), cpx (2), oxi (1) 1-16-22A* 1·29 fO2 controlled, Cl-bearing, S-bearing gl (88), amph (8), plag (2), oxi (2) 1-16-22B* — fO2 controlled, Cl-bearing, F-bearing gl (88), amph (7), plag (3), oxi (2) 1-16-23A* 1·49 fO2 controlled, CO2-bearing, S-bearing + H2O gl (88), amph (6), cpx (3), plag (2) oxi (1) 1-16-23B* — fO2 controlled, CO2-bearing, F-bearing + H2O gl (91), amph (6), cpx (1), oxi (2) Run ID . S content initial fluid (wt %) . Notes . Phases identified (number = vol. %) . 1-13-23 — H2O only gl (80), plag (12), oxi (4), amph (3), apa (trace) 1-14-02 — H2O only gl (86), plag (11), amph (2), oxi (2) 1-14-04A — fO2 controlled, H2O only gl (92), plag (5), cpx (2), oxi (1) 1-14-04B — fO2 controlled, Cl-bearing gl (92), plag (6), cpx (2), oxi (trace) 1-14-04C — fO2 controlled, F-bearing + H2O gl (89), plag (5), amph (4), oxi (2) 1-14-05 — Cl-bearing + H2O gl (94), amph (5), oxi (trace) 1-15-12A — H2O only gl (87), plag (5), amph (4), oxi (4) 1-15-12B — Cl-bearing + H2O gl (75), plag (14), amph (6), oxi (5) 1-15-15A — fO2 controlled, Cl-bearing gl (94), amph (3), plag (2), oxi (1) 1-15-15B — fO2 controlled, H2O only gl (89), plag (5), amph (4), oxi (2) 1-15-19B — Cl-bearing + H2O gl (92), plag (5), amph (2), cpx (1), oxi (trace) 1-16-02 0·44 fO2 controlled, Cl-bearing, S-bearing gl (94), cpx (5), oxi (trace) 1-16-12A — Cl-bearing + H2O gl (88), amph (7), plag (3), oxi (2) 1-16-12B 1·74 Cl-bearing, S-bearing gl (96), plag (3), amph (1), oxi (1) 1-16-14A — Cl-bearing, F-bearing gl (97), amph (2), oxi (1) 1-16-14B — F-bearing + H2O gl (91), plag (3), oxi (3), amph (2), cpx (1) 1-16-18A — Cl-bearing + H2O gl (88), amph (10), oxi (2) 1-16-18B — Cl-bearing + H2O gl (87), amph (10), oxi (3) 1-16-21A* — fO2 controlled, Cl-bearing + H2O gl (92), amph (7), oxi (1) 1-16-21B* — fO2 controlled, Cl-bearing + H2O gl (92), amph (5), cpx (2), oxi (1) 1-16-22A* 1·29 fO2 controlled, Cl-bearing, S-bearing gl (88), amph (8), plag (2), oxi (2) 1-16-22B* — fO2 controlled, Cl-bearing, F-bearing gl (88), amph (7), plag (3), oxi (2) 1-16-23A* 1·49 fO2 controlled, CO2-bearing, S-bearing + H2O gl (88), amph (6), cpx (3), plag (2) oxi (1) 1-16-23B* — fO2 controlled, CO2-bearing, F-bearing + H2O gl (91), amph (6), cpx (1), oxi (2) * Run conducted in Au96Cu4 tubing. † Calculated through mass balance. gl,  glass; plag,  plagioclase; amph,  amphibole; cpx,  clinopyroxene; oxi,  Fe–Ti oxide; apa,  apatite. Open in new tab Table 2: Experimental run conditions, fluid compositions, and run product phases Run ID . Starting powder . T (°C) . P (bar) . Duration (h) . log f O2ΔNNO . log fO2 . Fluid/melt mass ratio . Cl content initial fluid (wt %) . Cl content final fluid (wt %) . 1-13-23 raw Usu 850 2206 360 2·1 –10·7 0·06 — — 1-14-02 1-13-15 850 2190 385 2·1 –10·7 0·01 — — 1-14-04A 1-13-15 840 1500 480 –0·4 –13·4 0·04 — — 1-14-04B 1-13-15 840 1500 480 –0·4 –13·4 0·04 14·01 3·01 1-14-04C 1-13-15 840 1500 480 –0·4 –13·4 0·03 — — 1-14-05 1-13-15 840 2137 460 1·9 –11·1 0·44 9·38 7·99 1-15-12A raw Usu 835 3633 336 2·2 –10·7 0·07 — — 1-15-12B raw Usu 835 3633 336 2·2 –10·7 0·07 5·58 2·5† 1-15-15A 1-14-12 835 2947 671 –0·3 –13·5 0·10 7·38 3·31 1-15-15B 1-14-12 835 2947 671 –0·3 –13·5 0·05 — — 1-15-19B 1-15-17 835 2206 671 1·9 –11·2 0·04 14·01 4·51 1-16-02 1-15-17 845 2302 355 –0·5 –13·4 0·46 16·11 5·38 1-16-12A 1-15-17 810 3033 406 2·2 –11·4 0·07 7·20 4·52 1-16-12B 1-15-17 810 3033 406 2·2 –11·4 0·08 4·29 2·37 1-16-14A 1-15-16 850 2200 551 2·1 –10·7 0·03 14·01 no data 1-16-14B 1-15-16 850 2200 551 2·1 –10·7 0·02 — — 1-16-18A 1-15-17 820 2900 326 2·2 –11·2 0·21 7·11 7·79 1-16-18B 1-15-17 820 2900 326 2·2 –11·2 0·18 5·18 4·97 1-16-21A* 1-16-20 835 2400 480 –0·4 –13·6 0·05 14·01 1·12 1-16-21B* 1-16-20 835 2400 480 –0·4 –13·6 0·05 2·66 no data 1-16-22A* 1-16-20 840 2650 429 –0·4 –13·5 0·08 6·80 3·03 1-16-22B* 1-16-20 840 2650 429 –0·4 –13·5 0·03 14·01 1·7 1-16-23A* 1-16-20 860 4050 312 –0·5 –13·0 0·12 — — 1-16-23B* 1-16-20 860 4050 312 –0·5 –13·0 0·10 — — Run ID . Starting powder . T (°C) . P (bar) . Duration (h) . log f O2ΔNNO . log fO2 . Fluid/melt mass ratio . Cl content initial fluid (wt %) . Cl content final fluid (wt %) . 1-13-23 raw Usu 850 2206 360 2·1 –10·7 0·06 — — 1-14-02 1-13-15 850 2190 385 2·1 –10·7 0·01 — — 1-14-04A 1-13-15 840 1500 480 –0·4 –13·4 0·04 — — 1-14-04B 1-13-15 840 1500 480 –0·4 –13·4 0·04 14·01 3·01 1-14-04C 1-13-15 840 1500 480 –0·4 –13·4 0·03 — — 1-14-05 1-13-15 840 2137 460 1·9 –11·1 0·44 9·38 7·99 1-15-12A raw Usu 835 3633 336 2·2 –10·7 0·07 — — 1-15-12B raw Usu 835 3633 336 2·2 –10·7 0·07 5·58 2·5† 1-15-15A 1-14-12 835 2947 671 –0·3 –13·5 0·10 7·38 3·31 1-15-15B 1-14-12 835 2947 671 –0·3 –13·5 0·05 — — 1-15-19B 1-15-17 835 2206 671 1·9 –11·2 0·04 14·01 4·51 1-16-02 1-15-17 845 2302 355 –0·5 –13·4 0·46 16·11 5·38 1-16-12A 1-15-17 810 3033 406 2·2 –11·4 0·07 7·20 4·52 1-16-12B 1-15-17 810 3033 406 2·2 –11·4 0·08 4·29 2·37 1-16-14A 1-15-16 850 2200 551 2·1 –10·7 0·03 14·01 no data 1-16-14B 1-15-16 850 2200 551 2·1 –10·7 0·02 — — 1-16-18A 1-15-17 820 2900 326 2·2 –11·2 0·21 7·11 7·79 1-16-18B 1-15-17 820 2900 326 2·2 –11·2 0·18 5·18 4·97 1-16-21A* 1-16-20 835 2400 480 –0·4 –13·6 0·05 14·01 1·12 1-16-21B* 1-16-20 835 2400 480 –0·4 –13·6 0·05 2·66 no data 1-16-22A* 1-16-20 840 2650 429 –0·4 –13·5 0·08 6·80 3·03 1-16-22B* 1-16-20 840 2650 429 –0·4 –13·5 0·03 14·01 1·7 1-16-23A* 1-16-20 860 4050 312 –0·5 –13·0 0·12 — — 1-16-23B* 1-16-20 860 4050 312 –0·5 –13·0 0·10 — — Run ID . S content initial fluid (wt %) . Notes . Phases identified (number = vol. %) . 1-13-23 — H2O only gl (80), plag (12), oxi (4), amph (3), apa (trace) 1-14-02 — H2O only gl (86), plag (11), amph (2), oxi (2) 1-14-04A — fO2 controlled, H2O only gl (92), plag (5), cpx (2), oxi (1) 1-14-04B — fO2 controlled, Cl-bearing gl (92), plag (6), cpx (2), oxi (trace) 1-14-04C — fO2 controlled, F-bearing + H2O gl (89), plag (5), amph (4), oxi (2) 1-14-05 — Cl-bearing + H2O gl (94), amph (5), oxi (trace) 1-15-12A — H2O only gl (87), plag (5), amph (4), oxi (4) 1-15-12B — Cl-bearing + H2O gl (75), plag (14), amph (6), oxi (5) 1-15-15A — fO2 controlled, Cl-bearing gl (94), amph (3), plag (2), oxi (1) 1-15-15B — fO2 controlled, H2O only gl (89), plag (5), amph (4), oxi (2) 1-15-19B — Cl-bearing + H2O gl (92), plag (5), amph (2), cpx (1), oxi (trace) 1-16-02 0·44 fO2 controlled, Cl-bearing, S-bearing gl (94), cpx (5), oxi (trace) 1-16-12A — Cl-bearing + H2O gl (88), amph (7), plag (3), oxi (2) 1-16-12B 1·74 Cl-bearing, S-bearing gl (96), plag (3), amph (1), oxi (1) 1-16-14A — Cl-bearing, F-bearing gl (97), amph (2), oxi (1) 1-16-14B — F-bearing + H2O gl (91), plag (3), oxi (3), amph (2), cpx (1) 1-16-18A — Cl-bearing + H2O gl (88), amph (10), oxi (2) 1-16-18B — Cl-bearing + H2O gl (87), amph (10), oxi (3) 1-16-21A* — fO2 controlled, Cl-bearing + H2O gl (92), amph (7), oxi (1) 1-16-21B* — fO2 controlled, Cl-bearing + H2O gl (92), amph (5), cpx (2), oxi (1) 1-16-22A* 1·29 fO2 controlled, Cl-bearing, S-bearing gl (88), amph (8), plag (2), oxi (2) 1-16-22B* — fO2 controlled, Cl-bearing, F-bearing gl (88), amph (7), plag (3), oxi (2) 1-16-23A* 1·49 fO2 controlled, CO2-bearing, S-bearing + H2O gl (88), amph (6), cpx (3), plag (2) oxi (1) 1-16-23B* — fO2 controlled, CO2-bearing, F-bearing + H2O gl (91), amph (6), cpx (1), oxi (2) Run ID . S content initial fluid (wt %) . Notes . Phases identified (number = vol. %) . 1-13-23 — H2O only gl (80), plag (12), oxi (4), amph (3), apa (trace) 1-14-02 — H2O only gl (86), plag (11), amph (2), oxi (2) 1-14-04A — fO2 controlled, H2O only gl (92), plag (5), cpx (2), oxi (1) 1-14-04B — fO2 controlled, Cl-bearing gl (92), plag (6), cpx (2), oxi (trace) 1-14-04C — fO2 controlled, F-bearing + H2O gl (89), plag (5), amph (4), oxi (2) 1-14-05 — Cl-bearing + H2O gl (94), amph (5), oxi (trace) 1-15-12A — H2O only gl (87), plag (5), amph (4), oxi (4) 1-15-12B — Cl-bearing + H2O gl (75), plag (14), amph (6), oxi (5) 1-15-15A — fO2 controlled, Cl-bearing gl (94), amph (3), plag (2), oxi (1) 1-15-15B — fO2 controlled, H2O only gl (89), plag (5), amph (4), oxi (2) 1-15-19B — Cl-bearing + H2O gl (92), plag (5), amph (2), cpx (1), oxi (trace) 1-16-02 0·44 fO2 controlled, Cl-bearing, S-bearing gl (94), cpx (5), oxi (trace) 1-16-12A — Cl-bearing + H2O gl (88), amph (7), plag (3), oxi (2) 1-16-12B 1·74 Cl-bearing, S-bearing gl (96), plag (3), amph (1), oxi (1) 1-16-14A — Cl-bearing, F-bearing gl (97), amph (2), oxi (1) 1-16-14B — F-bearing + H2O gl (91), plag (3), oxi (3), amph (2), cpx (1) 1-16-18A — Cl-bearing + H2O gl (88), amph (10), oxi (2) 1-16-18B — Cl-bearing + H2O gl (87), amph (10), oxi (3) 1-16-21A* — fO2 controlled, Cl-bearing + H2O gl (92), amph (7), oxi (1) 1-16-21B* — fO2 controlled, Cl-bearing + H2O gl (92), amph (5), cpx (2), oxi (1) 1-16-22A* 1·29 fO2 controlled, Cl-bearing, S-bearing gl (88), amph (8), plag (2), oxi (2) 1-16-22B* — fO2 controlled, Cl-bearing, F-bearing gl (88), amph (7), plag (3), oxi (2) 1-16-23A* 1·49 fO2 controlled, CO2-bearing, S-bearing + H2O gl (88), amph (6), cpx (3), plag (2) oxi (1) 1-16-23B* — fO2 controlled, CO2-bearing, F-bearing + H2O gl (91), amph (6), cpx (1), oxi (2) * Run conducted in Au96Cu4 tubing. † Calculated through mass balance. gl,  glass; plag,  plagioclase; amph,  amphibole; cpx,  clinopyroxene; oxi,  Fe–Ti oxide; apa,  apatite. Open in new tab Experimental conditions These crystallization experiments were performed in a steel-walled IHPV at the AMNH. The general IHPV equipment specifications and features have been described by Holloway (1971) and Holloway et al. (1992). The experiments (Table 2) were brought up to temperature and pressure within 30 min and run initially at 950°C, for at least 24 h, followed by a slow drop in temperature (10°C per 30 min) to the reported run condition. Temperature was cycled daily by ±20°C around the reported run temperature to promote the growth of large crystals at the expense of smaller ones that would be more difficult to analyse, similarly to the method discussed by Erdmann & Koepke (2016). The temperatures were measured with chromel–alumel thermocouples that are calibrated to the melting point of pure gold and positioned close to the midpoint of the capsule in the furnace. The orientation of the IHPV was adjusted from horizontal such that the temperature gradient across the length of the capsule in the hottest region of the furnace was minimized, and the gradient was consistently reduced to <10°C. Pure Ar gas pressures ranged from 1·5 to 4·05 kbar as determined with factory-calibrated, Heise Bourdon-tube-based gauges, and a recent calibration returned errors of ±2 bar for pressures >700 bar. The oxygen fugacity of the experiments ranged from NNO –0·5 (nickel–nickel oxide minus 0·5 log units) to NNO +2·0. The fO2 was controlled either by direct application of a partial pressure of pure H2 to the hottest region of the furnace, where the capsules are located, via a Shaw membrane capillary (e.g. Shaw, 1963; Scaillet et al., 1992; Schmidt et al., 1995), or by the intrinsic hydrogen fugacity (fH2) of the IHPV owing to the presence of H2 in the steel alloy of the vessel walls. The ambient fO2 buffering capacity of the IHPV was determined with an H2 sensor, and for experimental charges where water activity was ≈1, the IHPV imposes an fH2 roughly equivalent to an fO2 of NNO +2·0 at 800°C and 200 MPa. The loading pressure of H2 to the Shaw membrane was monitored with a precision of 1% relative with a factory-calibrated pressure gauge. The fO2 can then be reasonably calculated, by assuming a water activity of unity in the capsule (i.e. assuming at all times that the melt coexists with a separate ∼100% H2O volatile phase). Although volatile phases in most of the capsules were not pure H2O, the uncertainties do not significantly affect the calculated fO2 values. The accuracy on the calculated fO2 values is then taken to be ±0·4 log units. For the two experiments where CO2 was added as oxalic acid, the XH2O, and hence aH2O, was less than unity, and thus these experiments are the most reduced. In the discussion of the results, experiments conducted at ambient, unbuffered fO2 conditions are described as ‘intrinsic fO2’ and are fO2 >NNO +1·0, whereas those conducted under direct application of pure H2 are described as ‘low fO2’ and are fO2 <NNO +1·0. The durations of the experiments ranged from 312 to 671 h, with most runs conducted for at least 400 h (16 days). Prior experimental investigations have determined that Cl-bearing, aqueous fluids equilibrate with hydrous felsic melts in as little as 3–5 days (Webster, 1992; Shinohara, 1994). The experiments were concluded by rapid, isobaric quenching by cutting the power to the furnace and increasing the Ar pressure to sufficiently compensate for the pressure decrease during the thermal contraction of the cooling gas and capsule. In most cases, the temperature decreased from run conditions to below the glass transition temperature of ∼450°C in less than 8–10 s. The rapid quench method and relatively high viscosity of the melts in these experiments prevents the nucleation of nanolites and any back-reaction between the phenocrysts and melt ± fluid phase(s) at lower temperatures. Processing of run products After quenching the runs, the mass of the charge was taken and if there was any loss or gain >1 mg relative to its pre-run mass, the capsule was discarded and not used. If the mass change was minimal, the capsules to which Cl-bearing fluids had been added were cleaned with ethanol then centrifuged for ∼5 min to concentrate the volatile phase towards one end. Using a cleaned, sharpened steel-tipped probe a small puncture was then made in the capsule. The capsule was then immediately placed into a vial containing a measured mass of doubly distilled water, such that the puncture was submerged in the water. The capsules were allowed to soak for a minimum of 168 h (1 week). The soaking procedure ensures that all alkali-chloride complexes that had precipitated from the coexisting volatile phase during quench are fully redissolved in the soak solution. Following this soaking procedure, the capsules were removed from the soak solution and heated and dried in a 1 atm oven at 120°C. Non-Cl-bearing experiments were cleaned, punctured, and heated, without undergoing soaking. The two experiments with added CO2 were placed in a −20°C freezer for at least 2 h before being punctured and the masses were recorded for gravimetry constraints on the quantity of CO2-rich vapour. These capsules were then heated as normal, and the masses were recorded again. The mass difference between the post-experiment pre-punctured capsule and the post-experiment punctured ± soaked + heated capsule is taken to represent the total volatile phase budget of that capsule. Finally, the capsules were carefully opened using needle-nose pliers and a jeweller’s blade to allow extraction of the crystalline run products. The run product chips were mounted and polished in epoxy resin, and carbon-coated for electron microprobe analyses, or chips were doubly polished for Fourier transform IR spectroscopy analysis. ANALYTICAL METHODS Electron probe microanalysis (EPMA) The major, minor, and halogen element compositions of the experimental run product glasses and phenocrysts were determined by wavelength-dispersive spectrometry, using a JEOL JXA-8500F electron microprobe at the Peter Hooper GeoAnalytical Laboratory, Washington State University. An accelerating voltage of 15 kV, a beam current of 10 nA, and a beam diameter of 10 μm were used for glass analyses, whereas a 2 μm diameter beam was used for crystal analyses. Major and minor element concentrations were determined first at 10 nA, followed by determination of F, Cl, and S concentrations in the same analysis spot location using a 50 nA beam current and extended counting times to improve analytical precision. To ensure robust matrix corrections, X-ray intensity data from both passes (i.e. major and minor elements, and F, Cl, and S) were combined into a single analysis and processed together. Minor Na migration (e.g. Nielsen & Sigurdsson, 1981) observed in the hydrous glasses was corrected using the time-dependent intensity corrections of Donovan et al. (2007). For calibration, natural and synthetic standards were used: albite (Si, Na), orthoclase (Al), hornblende (Ti), basaltic glass (Fe), diopside (Ca), olivine (Mg), spessartine (Mn), MgF2 (F), KCl (K, Cl), and anhydrite (S). Given the peak shift associated with different S speciation states in silicate glasses (e.g. Wallace & Carmichael, 1994; Jugo et al., 2005), and the predominantly oxidizing conditions at which these experiments were run, anhydrite (with sulfur in the oxidized S6+ state) was selected in preference to a reduced sulfide mineral standard. All Fe concentrations are reported as total FeO and discussed as such, with Fe2+ and Fe3+ calculated based on charge balance for amphibole (using the model of Ridolfi et al., 2010) and clinopyroxene. The typical detection limits for these oxides expressed in ppm were calculated using the model of Donovan et al. (2007) and are as follows: SiO2 190, TiO2 210, Al2O3 170, FeO 600, MnO 450, MgO 110, CaO 190, Na2O 310, K2O 150, P2O5 90, SO3 10, Cl 90, and F 110. Reproducibility and monitoring of analytical drift across different analytical sessions were determined by repeat analyses of Kakanui hornblende (USNM 143965), Lake County labradorite (NMNH 115900), VG568 rhyolitic glass (NMNH 72854), and Macusani rhyolitic glass standards. Based on these check standards, the relative standard deviations of repeat analyses were consistently better than 3% for all major elements analysed (Supplementary Data Electronic Appendix 1; supplementary data are available for downloading at http://www.petrology.oxfordjournals.org). Chloridometer analysis The chloride contents of the capsule soak solutions were measured using a Labconco Digital Chloridometer. The instrument reports the measured milliequivalents of chloride per litre (mequiv Cl– l–1), which can be converted to mg l–1 or weight per cent Cl to define the chlorinity of the fluid phase coexisting in equilibrium with the rhyodacitic melt and crystals. Based on multiple analyses of a standard NaCl solution containing 3·5 wt % Cl–, the chloridometer achieves a 1σ relative precision of ∼2·5%. A 10 μl aliquot is extracted from the capsule soak solution to be analysed using the chloridometer, and the aliquot is combined with acetic acid for analysis. The dilution factor, following the soaking process, was taken into account. Blanks were run before and after each analytical session and background signal corrected. Fourier transform IR spectroscopy Analyses of 18 run product glasses to determine the concentrations of hydroxyl ion, molecular H2O, molecular CO2, and molecular CO32– in the quenched glasses were conducted at the AMNH. Absorption data were collected at 4500 cm–1 for hydroxyl, 5200 cm–1 for molecular H2O, 2350 cm–1 for CO2, and 1425 cm–1 and 1525 cm–1 for CO32–, following the method of Webster et al. (2014). The average thicknesses of doubly polished chips were measured with a Mitutoyo digitometer and three FTIR analyses were collected in different spots of each polished glass chip. The analysis locations were chosen such that vesicles and crystal phases below the polished surface in the glasses were avoided. Measurements were conducted using a Nicolet Nexus 670 FTIR spectrometer in transmittance mode, attached to a Continuum IR microscope, and N2 gas was passed through the instrument during data collection. Infrared spectral data were collected both in the mid-IR (450–4000 cm–1) and the higher-energy near-IR (up to 8500 cm–1) regions using a KBr beam splitter, an MCT/A detector, and a Globar source. Background signal was collected prior to each analysis and subtracted from the sample measurement. Glass densities were calculated after Luhr (2001) using the Gladstone–Dale rule (Gladstone & Dale, 1864; Silver et al., 1990) and EPMA data. The extinction coefficients for dacitic glasses of Yamashita et al. (1997) for H2O, Behrens et al. (2004) for CO2, and Webster et al. (2014) for CO32– were used. RESULTS Run product phase stability and crystallinity All of the 24 run products contain residual glass and Fe–Ti oxides, in varying proportions. Based on semi-quantitative energy-dispersive spectrometry analysis, these Fe–Ti oxide phases are a mixture of both magnetite and ilmenite, but no full chemical quantification was undertaken for the oxides crystallized in each run product. Where ‘Fe–Ti oxides’ as a separate phase are discussed further in the text, they should be considered as a mixture of these two minerals, and no direct fO2 constraints are implied based on their relative abundances and compositions. Plagioclase, amphibole, and clinopyroxene, along with a single apatite crystal in one run product, are the other phases that variably crystallized in these experiments (Fig. 1). ImageJ thresholding and area particle analysis shows that run product crystallinities range from ∼5 to ∼25 vol. %, and vesicle abundance averages ∼5 vol. % (Table 2). Fig. 1. Open in new tabDownload slide Reflected light (RL) and back-scattered electron (BSE) images of experimental run products, showing amphibole, clinopyroxene, plagioclase, glass, and accessory oxides and a single apatite crystal. (a) a RL image of 1-16-22A run product; (b) is the corresponding BSE image of (a) showing the unzoned crystals and euhedral crystal morphology; (c) an enlarged BSE image of the run product of experiment 1-13-23; (d) the run product of experiment 1-16-14B. (Note the similar phase assemblages and homogeneous residual glasses in all experiments.) Most amphibole and clinopyroxene crystals are euhedral and randomly oriented, and show good basal cross-sections when intersected at the polished surface (Fig. 1). Plagioclase crystals show more elongate, platy aspect ratios and are often several hundred micrometres long and ≤100 μm wide. Some larger plagioclase crystals show evidence of skeletal growth, but microanalysis reveals limited, if any, chemical zonation, and glass patches within these skeletal crystals are compositionally identical to the rest of the groundmass glass, suggesting that equilibrium is maintained. Constraining the full phase equilibria of this Mt Usu rhyodacite was beyond the scope of this investigation, and they were investigated for similar eruptive products from Mt Usu in previous studies (see Tomiya et al., 2010). Tomiya et al. (2010) did not crystallize amphibole below ∼2 kbar in their experiments on a low-K rhyolitic pumice, and clinopyroxene was not stable in any run, probably owing to lower CaO in the starting powdered sample (∼2 wt % compared with >3·5 wt % in this Mt Usu sample). The experimental conditions in this investigation were specifically selected to allow crystallization of amphibole and plagioclase (± clinopyroxene), and the experimental phase relations and stability regions broadly overlap with those from other studies of hydrous dacite–rhyolite melt compositions crystallized at these temperature, pressure, and fO2 conditions (e.g. Cottrell et al., 1999; Venezky & Rutherford, 1999; Browne et al., 2010). In our experiments it can be seen that plagioclase has a wide stability field across the conditions investigated, but was not identified in eight run products. Based on our dataset, there appears to be no systematic experimental or compositional control on the absence of plagioclase. Similarly, the majority of experimental conditions were suitable for the crystallization of amphibole, with the exception of two lower pressure runs (1·5 kbar), and a lower fO2 (NNO –0·5) run with high Cl contents in the starting fluid (>16 wt %). The addition of 0·58 wt % F to the melt of one 1·5 kbar run (1-14-04C) appears to have stabilized amphibole outside its normal stability region, as evidenced by its absence from the other two runs at these conditions. Clinopyroxene was stable in seven experiments—in three of these it was present as the only mafic phase, and in the others it was closely associated with amphibole (e.g. Fig. 1d). Glass compositions All run product glasses (Supplementary Data Electronic Appendix 2) are homogeneous based on multiple analyses (most glasses were analysed at least 10 times) across several different glass chips from the same experiment, with major elements consistently showing relative standard deviations of <5%. When plotted on an anhydrous basis on the total alkali–silica (TAS) classification, the run product glasses are predominantly rhyolitic, but span the dacite–rhyolite join (69·8–77·7 wt % SiO2 and 5·52–6·29 wt % Na2O + K2O) (Fig. 2a). The glasses show negative correlations between SiO2 vs Al2O3 and SiO2 vs CaO (Fig. 2b). MnO concentrations vary minimally, averaging 0·16 ± 0·04 wt %, but MgO ranges from 0·30 to 0·88 wt %. Neither MgO nor MnO shows a correlation with changing SiO2. Fig. 2. Open in new tabDownload slide (a) An enlarged section of the total alkali–silica diagram showing the composition of the experimental run product glasses and the raw Mt Usu dacite on an anhydrous basis; (b) major element covariance plots for the run product glasses (note the break in y-axis scale); (c) averaged glass data for each experiment showing the correlation between increasing melt Cl and melt FeO concentrations in Cl-doped experiments, but a less pronounced correlation at lower melt Cl contents. The behaviour of FeO is complex, and shows the largest variability apart from SiO2 and the halogens. FeO is greatly influenced by the fO2 conditions imposed on the experiment, and the control this has on oxide stability. For example, when the experimentally controlled fO2 values are lower than the intrinsic ∼NNO +2·0 of the IHPV, total FeO contents are higher, owing to the suppression of Fe–Ti oxide crystallization. FeO concentrations in glasses from lower fO2 experiments (where fO2 ≤ NNO +1) average 2·30 ± 0·34 wt %, whereas those crystallized at ambient fO2 conditions average 1·15 ± 0·17 wt % (Fig. 2c). The exception to this is experiment 1-14-05, which crystallized at higher fO2 conditions but shows a relatively elevated FeO of 1·92 wt % and an almost complete absence of Fe–Ti oxides. A corollary of the variable oxide stability is that the concentrations of TiO2 in the residual glasses are also affected. Lower fO2 experiments show average TiO2 concentrations of 0·28 ± 0·05 wt %, compared with 0·17 ± 0·04 wt % for higher fO2 experiments. It is also relevant to note that although Fe loss to the capsule walls was not directly constrained, this issue is likely to be negligible at these experimental conditions and in these precious metal capsule alloys (e.g. Brugier et al., 2015). The total volatile contents of the glasses are variable, depending on the concentrations in the initial materials added to the capsule and the compositions of the coexisting fluid phase(s). Chlorine ranges from natural levels (in the raw Usu rhyodacite) of ∼130 ppm to a maximum of 0·72 wt %, whereas F ranges from natural background concentrations of ∼110 ppm (close to, or below, detection limit of the EPMA) to 0·63 wt %. Sulfur contents of the run product glasses are generally low, averaging 100–300 ppm SO3. In runs where S was added to the capsule, these concentrations are slightly, but not systematically, elevated, reaching a maximum of ∼460 ppm SO3. An important observation is the strong positive correlation between increasing melt FeO and increasing melt Cl, for the experiments where Cl was added to the starting charges (Fig. 2c). In the four experiments with the highest Cl contents (ranging from 0·58 to 0·72 wt %) this factor in particular appears to have almost completely suppressed Fe–Ti oxide formation. All four experiments (1-14-04B, 1-14-05, 1-15-19B, and 1-16-02) crystallized only traces of Fe–Ti oxides compared with other experiments with lower Cl contents. Based on the difference between the measured analytical totals and 100% [i.e. the ‘by difference’ method of Devine et al. (1995)], all glasses contain >5 wt %, and average 6·97 ± 0·82 wt % H2O. The FTIR data show good agreement with the ‘by difference’ method, with most data within 1σ of the EPMA-determined concentrations (Supplementary Data Electronic Appendix 2). Carbon dioxide concentrations range from <5 to >60 ppm in the glasses that were not CO2-doped, whereas the concentration measured in one experiment with CO2 added to the starting volatile phase was 307 ± 30 ppm. The molar Na2O/(Na2O+K2O) ratios of the run product glasses are tightly clustered around 0·87 ± 0·01, whereas the aluminosities [i.e. molar Al2O3/(Na2O+K2O+CaO)] range from 0·89 to 1·08. The alkalinity of the run product glass is a strong function of the total F concentration (given F was added as CaF2), with higher F experiments trending towards peralkaline compositions (i.e. aluminosity <1) compared with peraluminous glasses (i.e. aluminosity >1) with lower F concentrations. Final fluid compositions Direct chloridometer determinations show that the final fluid compositions range from 1·12 to 7·99 wt % total Cl– (Table 2). These concentrations give a range in calculated DClfluid/melt from 3·5 to 22·7, with an average of 9·3. However, S and F concentrations of the final fluids were not directly measured. (See Discussion for further details.) Amphibole compositions A total of 179 amphibole analyses are reported from crystals across 21 amphibole-bearing run products (Supplementary Data Electronic Appendix 3). Some crystals were sufficiently large to allow multiple or transect analyses (e.g. Fig. 1a), whereas others allowed only a single analysis point. Most amphiboles are 50–100 μm in length and ∼50 μm in width when intersected at the polished surface. Amphibole formulae are calculated on the basis of 15 cations, with Fe3+ calculated stoichiometrically and assigned to the octahedral site. The end-member naming convention follows the scheme of Leake et al. (1997). The amphibole are mostly unzoned and internally homogeneous, spanning the magnesiohornblende and tschermakite subdivisions for calcic amphibole, with [Mg/(Mg + FeTOT)] (Mg#) ratios ranging from 0·55 to 0·86 and Si a.p.f.u. from 6·98 to 6·24. CaO contents vary little, and range from 10·43 to 12·76 wt %. Molar Si/AlTOT correlates positively with [Mg/(Mg + FeTOT)] and weakly negatively with A-site occupancy by (Na + K) a.p.f.u. with distinct clusters of data (Fig. 3a and b). Fig. 3. Open in new tabDownload slide Amphibole major element compositional plots showing distinct clusters of data, especially in (a) where amphiboles crystallized in lower fO2 experiments (where fO2 < NNO +1) show Mg/(Mg + FeTOT) values generally less than 0·70. In (c) the break in the y-axis scale, with the grey box highlighting an expanded portion of the data from non-F-doped experiments, should be noted. In (d) and for all following figures, the colour of the point denotes the fO2 conditions of that experiment (black vs white) whereas the shape (circle vs square vs diamond) denotes the presence or absence of F or Cl in the experiment, unless otherwise specified in the legend. (d) shows average amphibole FeO concentrations as a function of the melt FeO in that experiment, and in turn the pronounced effect of fO2 has on this variation. It should be noted also that the addition of F to the melt leads to amphibole with relatively less FeO than would be expected for those melt FeO concentrations. Colour figures available online. Halogen contents of the amphibole vary significantly, and as a function of the concentrations added to the starting experiments. Chlorine contents are higher from Cl-bearing experiments, and average ∼700 ppm, compared with ≤100 ppm (i.e. below detection limit) for most of the amphibole grown in Cl-free experiments. Fluorine concentrations average ∼0·13 wt % across both Cl-bearing and Cl-free experiments, whereas amphiboles from F-bearing experiments show much higher contents, reaching a maximum of 2·59 wt % F (Fig. 3c). There is also a positive correlation between F and total alkali contents of the amphiboles, with those from F-free experiments averaging 1·70 wt % Na2O and 0·14 wt % K2O, respectively, and those from F-bearing experiments averaging 2·13 wt % Na2O and 0·19 wt % K2O. The compositions of the experimental amphibole are also consistent with the Mg–Cl and Fe–F crystallographic avoidance principles (e.g. Morrison, 1991), but this is complicated by the oxygen fugacity of the experiments and its effect on the Fe2+/Fe3+ of the amphibole (see Discussion for further details). Sulfur concentrations were below the detection limit for all amphiboles analysed. The amphibole FeO content is strongly affected by experimental fO2 conditions, with lower fO2 amphibole (where fO2 ≤ NNO +1) averaging 12·86 ± 0·88 wt % FeO, compared with 8·77 ± 1·15 wt % FeO for amphibole crystallized in higher fO2 conditions (where fO2 > NNO +1), consistent with greater melt FeO concentrations in these experiments (Fig. 3d). The Al2O3 contents of the amphibole are varied, and range from 6·25 to 11·81 wt %, although there is no strong correlation between Al2O3 content and the experimental pressure. MnO ranges from 0·26 to 1·01 wt %, and TiO2 from 0·42 to 1·92 wt%. A small population of amphiboles, especially larger crystals grown in the CO2-bearing experiments 1-16-23A and -23B, show core–rim increases in FeO and Al2O3 and corresponding decreases in SiO2 and MgO, with sharp compositional boundaries visible in back-scattered electron (BSE) images. In one specific amphibole, the FeO concentration increases from 11·07 to 16·39 wt % and Al2O3 from 8·63 to 11·81 wt %, whereas SiO2 decreases from 48·11 to 43·17 wt % and MgO from 15·96 to 11·09 wt %. This change is consistent with a combination of the Fe2+Mg–1 and Al2Mg–1Si–1 (Tschermak’s) exchange components and is probably in response to a decrease in the fO2 of the experiment as CO2 dissolves in the melt, until the final reported fO2 is reached from initially more oxidizing conditions. For these crystals, compositions of the amphibole rims in contact with the glasses are reported (Supplementary Data Electronic Appendix 3). Plagioclase compositions A total of 119 plagioclase spot analyses are reported from crystals across 16 run products (Supplementary Data Electronic Appendix 4). Plagioclase formulae were calculated on the basis of five cations, assuming all Fe to be divalent. Similar to the amphibole, some crystals were sufficiently large for multiple analyses (e.g. Fig. 1a); however, many allow only a single spot analysis. In general, the plagioclase grains are euhedral and often larger than the amphiboles, growing up to hundreds of micrometres in length and width, and despite evidence of skeletal growth, where identifiable, less than a 5% variability in the mole per cent of the anorthite end-member component (XAn) is seen between core and rim analyses in the largest crystals. The XAn of plagioclase crystals ranges from An59 to An39, classified as labradorite or andesine (Fig. 4a). Fig. 4. Open in new tabDownload slide (a) shows the compositional range across all experimental plagioclase crystals and the generally positive correlation between MgO concentration and XAn. The limit of detection for EPMA is also shown. (b) shows the variation in plagioclase FeO and its relation to melt FeO, along with the influence of fO2 on these two values, averaged for each experiment. Although there is some scatter, MgO concentrations show a largely positive correlation with increasing anorthite component, and no correlation with experimental fO2 (Fig. 4a). TiO2 and MnO are both below the detection limit of the EPMA. Plagioclase FeO contents are variable, and range from 0·88 to 0·18 wt %, depending on the melt FeO and hence the fO2 of the experiment. K2O and MgO concentrations are generally low, averaging ∼800 ppm and ∼350 ppm, respectively. Clinopyroxene compositions A total of nine experiments crystallized clinopyroxene, with several crystals large enough for multiple spots to be analysed, giving a total of 29 analyses (Supplementary Data Electronic Appendix 5). The crystals are generally 50–200 μm in length and ∼50–75 μm in width. The clinopyroxene are mostly euhedral, and although some slight internal chemical zonation between cores and rims is visible in BSE imaging, there is a less than a 3% variability in the mole per cent of the wollastonite end-member component (XWo) in EPMA transects taken across the length of the largest crystals. There is a generally limited range in the XWo across all crystals, from Wo49 to Wo30, with the majority of clinopyroxene >Wo40, as shown on an enlarged section of the pyroxene quadrilateral with all analysis points (Fig. 5a). All clinopyroxene crystals contain a significant enstatite component, and are Mg-rich, giving Mg# values of 0·86–0·69. However, as with the amphibole and plagioclase, the FeO concentrations are a function of the melt FeO and hence fO2 of the experiment, with higher FeO (and Fe2+ a.p.f.u. by stoichiometry) in clinopyroxene from experiments where fO2 < NNO +1 (Fig. 5b). The Al2O3 concentrations range from ∼3 to >6 wt %, MnO from 0·4 to 1·5 wt %, and Na2O contents from ∼0·2 to ∼0·7 wt %. Titanium, Al, and Na a.p.f.u. correlate positively, as do Si and Mg. The K2O contents are below ∼200 ppm (EPMA detection limit) for all clinopyroxene analysed. The two samples of clinopyroxene from the lowest pressure runs (1-14-04A and 1-14-04B at 1·5 kbar) show greater than double the MnO contents, but generally half of the TiO2 and Na2O contents when compared with the runs at >2 kbar, and are more Mg-rich (Fig. 5a). Fig. 5. Open in new tabDownload slide (a) shows an enlarged section of the pyroxene compositional quadrilateral, showing the calcic experimental clinopyroxene crystals, and the effect of lower fO2 conditions (where fO2 < NNO +1) on FeTOT. The extension to more Mg-rich pyroxene compositions for those crystallized at the lowest pressures (grey shading) should also be noted. (b) shows the effect of fO2 on average melt and clinopyroxene FeO concentrations. There are two clusters of data for low fO2 experiments, with the three higher pressure runs (2·65–4·05 kbar) grouped at higher clinopyroxene FeO and lower melt FeO. Mineral–melt partition coefficients In the following sections, we present and discuss our new data for mineral–melt partitioning of major, minor, and halogen components in the three main phenocryst phases analysed. However, elements that form an integral part of the structural formulae of these phenocrysts (e.g. Mg and Ca in amphibole, or Al and Ca in plagioclase) should not be thought of as conforming to simple Nerst-like partitioning behaviour. These major components are better modelled and interpreted as undergoing coupled exchange reactions between the melt and crystallizing phases [as is often done for the fluid–melt partitioning of volatile species; e.g. Zajacz et al. (2012) and references therein], which are in turn a complex function of temperature, pressure, and fO2 conditions, and the bulk mineral and melt compositions. Therefore, these new data do not provide a simple and/or invariant partition coefficient that fully describes each mineral–melt system and that can be extrapolated beyond the conditions of these experiments. However, through comparison of these data with melt compositional parameters and through integration with literature data for similarly evolved melts and coexisting crystals, a better understanding of the largest physicochemical influences on phenocryst chemistry can be obtained. Dmineral/melt values for SiO2, MnO, MgO, CaO, and Na2O show relatively consistent partitioning behaviour with limited ranges for all three mineral phases in this rhyodacite (Table 3; Fig. 6). Conversely, TiO2, Al2O3, FeO, K2O, Cl, and F show larger variability. MgO is the most compatible element in amphibole and clinopyroxene, and one of the least compatible in plagioclase. As a first-order observation, there is no apparent systematic change in the major element partitioning behaviour for any of the three phases investigated as a function of varying temperature. Given that the temperature range of the experiments was relatively restricted, it is likely any temperature-dependent effects are small, and obscured by other parameters, such as melt structure. Similarly, although the pressure-dependent incorporation of Al into amphibole is well known (e.g. Putirka, 2016, and references therein) there is no strong correlation between amphibole DAl2O3amph/melt and experimental pressure. However, there are crystallographic constraints on the incorporation of Al and Fe as a function of fO2 conditions and Fe–F avoidance, as discussed in the following sections, which may therefore obscure this trend. Table 3: Calculated mineral–melt partition coefficients for each mineral phase present in each experiment, and associated propagated error. Mineral phase abbreviations as in Table 2 ID: . 1-13-23 . 1-14-02 . 1-14-04A . Phase: . Amph . Plag . Amph . Plag . Cpx . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·64 0·01 0·79 0·01 0·65 0·02 0·80 0·01 0·77 0·02 0·80 0·01 DTiO2min/melt 4·34 1·01 — — 8·21 1·26 — — 1·27 0·13 — — DAl2O3min/melt 0·80 0·02 2·25 0·04 0·78 0·07 2·11 0·04 0·17 0·06 2·07 0·04 DFeOmin/melt 9·04 0·47 0·53 0·11 8·51 0·62 0·56 0·05 3·78 0·19 0·25 0·03 DMnOmin/melt 5·59 0·60 — — 3·61 0·57 — — 6·03 1·54 — — DMgOmin/melt 32·67 0·78 0·08 0·03 28·80 1·20 0·07 0·02 16·98 1·42 0·07 0·01 DCaOmin/melt 5·87 0·27 4·79 0·21 4·60 0·12 3·95 0·22 6·10 0·39 3·55 0·16 DNa2Omin/melt 0·37 0·03 1·35 0·06 0·40 0·04 1·34 0·09 0·06 0·01 1·10 0·05 DK2Omin/melt 0·09 0·01 0·09 0·02 0·12 0·04 0·07 0·01 — — 0·06 0·01 DClmin/melt — — — — — — — — — — — — DFmin/melt 6·59 1·25 — — 16·63 1·05 — — — — — — ID: 1-14-04B 1-14-04C 1-14-05 1-15-12A Phase: Cpx Plag Amph Amph Amph Plag average error average error average error average error average error average error DSiO2min/melt 0·78 0·02 0·85 0·01 0·69 0·01 0·70 0·02 0·66 0·01 0·81 0·01 DTiO2min/melt 1·66 0·10 — — 3·08 0·25 4·11 0·32 4·87 1·39 — — DAl2O3min/melt 0·16 0·06 2·03 0·04 0·63 0·04 0·60 0·07 0·72 0·06 2·02 0·03 DFeOmin/melt 3·45 0·51 0·17 0·03 4·72 0·23 5·50 0·30 8·66 0·46 0·49 0·05 DMnOmin/melt 5·33 1·39 — — 2·99 1·09 3·34 0·47 3·98 0·73 — — DMgOmin/melt 19·11 1·77 0·06 0·01 26·94 0·68 23·19 0·91 28·83 0·76 0·05 0·02 DCaOmin/melt 5·67 0·81 3·27 0·12 2·96 0·08 3·47 0·05 4·24 0·08 3·58 0·22 DNa2Omin/melt 0·06 0·02 1·19 0·06 0·47 0·04 0·36 0·03 0·38 0·03 1·35 0·10 DK2Omin/melt — — 0·08 0·03 0·18 0·03 0·12 0·01 0·11 0·02 0·08 0·02 DClmin/melt — — — — — — 0·17 0·01 — — — — DFmin/melt — — — — 3·09 0·75 5·65 0·53 8·87 1·10 — — ID: . 1-13-23 . 1-14-02 . 1-14-04A . Phase: . Amph . Plag . Amph . Plag . Cpx . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·64 0·01 0·79 0·01 0·65 0·02 0·80 0·01 0·77 0·02 0·80 0·01 DTiO2min/melt 4·34 1·01 — — 8·21 1·26 — — 1·27 0·13 — — DAl2O3min/melt 0·80 0·02 2·25 0·04 0·78 0·07 2·11 0·04 0·17 0·06 2·07 0·04 DFeOmin/melt 9·04 0·47 0·53 0·11 8·51 0·62 0·56 0·05 3·78 0·19 0·25 0·03 DMnOmin/melt 5·59 0·60 — — 3·61 0·57 — — 6·03 1·54 — — DMgOmin/melt 32·67 0·78 0·08 0·03 28·80 1·20 0·07 0·02 16·98 1·42 0·07 0·01 DCaOmin/melt 5·87 0·27 4·79 0·21 4·60 0·12 3·95 0·22 6·10 0·39 3·55 0·16 DNa2Omin/melt 0·37 0·03 1·35 0·06 0·40 0·04 1·34 0·09 0·06 0·01 1·10 0·05 DK2Omin/melt 0·09 0·01 0·09 0·02 0·12 0·04 0·07 0·01 — — 0·06 0·01 DClmin/melt — — — — — — — — — — — — DFmin/melt 6·59 1·25 — — 16·63 1·05 — — — — — — ID: 1-14-04B 1-14-04C 1-14-05 1-15-12A Phase: Cpx Plag Amph Amph Amph Plag average error average error average error average error average error average error DSiO2min/melt 0·78 0·02 0·85 0·01 0·69 0·01 0·70 0·02 0·66 0·01 0·81 0·01 DTiO2min/melt 1·66 0·10 — — 3·08 0·25 4·11 0·32 4·87 1·39 — — DAl2O3min/melt 0·16 0·06 2·03 0·04 0·63 0·04 0·60 0·07 0·72 0·06 2·02 0·03 DFeOmin/melt 3·45 0·51 0·17 0·03 4·72 0·23 5·50 0·30 8·66 0·46 0·49 0·05 DMnOmin/melt 5·33 1·39 — — 2·99 1·09 3·34 0·47 3·98 0·73 — — DMgOmin/melt 19·11 1·77 0·06 0·01 26·94 0·68 23·19 0·91 28·83 0·76 0·05 0·02 DCaOmin/melt 5·67 0·81 3·27 0·12 2·96 0·08 3·47 0·05 4·24 0·08 3·58 0·22 DNa2Omin/melt 0·06 0·02 1·19 0·06 0·47 0·04 0·36 0·03 0·38 0·03 1·35 0·10 DK2Omin/melt — — 0·08 0·03 0·18 0·03 0·12 0·01 0·11 0·02 0·08 0·02 DClmin/melt — — — — — — 0·17 0·01 — — — — DFmin/melt — — — — 3·09 0·75 5·65 0·53 8·87 1·10 — — ID: . 1-15-12B . 1-15-15A . 1-15-15B . Phase: . Amph . Plag . Amph . Plag . Amph . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·66 0·01 0·82 0·00 0·70 0·01 0·88 0·00 0·65 0·01 0·82 0·01 DTiO2min/melt 5·56 0·90 — — 5·90 0·80 — — 6·76 1·16 — — DAl2O3min/melt 0·73 0·06 2·00 0·03 0·66 0·05 1·85 0·01 0·79 0·02 2·04 0·04 DFeOmin/melt 7·75 0·33 0·45 0·04 5·89 0·32 0·12 0·03 8·94 0·54 0·22 0·06 DMnOmin/melt 3·28 0·58 — — 3·21 0·26 — — 4·89 0·48 — — DMgOmin/melt 27·64 0·94 0·07 0·01 29·26 1·54 0·05 0·01 41·68 1·84 0·06 0·03 DCaOmin/melt 4·16 0·08 3·38 0·16 3·39 0·09 2·81 0·08 4·50 0·10 3·95 0·23 DNa2Omin/melt 0·37 0·03 1·33 0·05 0·39 0·02 1·22 0·05 0·40 0·06 1·43 0·09 DK2Omin/melt 0·12 0·02 0·08 0·01 0·14 0·02 0·06 0·01 0·13 0·02 0·08 0·02 DClmin/melt 0·19 0·01 — — 0·21 0·03 — — 0·29 0·05 — — DFmin/melt 10·92 1·09 — — — — — — — — — — ID: . 1-15-12B . 1-15-15A . 1-15-15B . Phase: . Amph . Plag . Amph . Plag . Amph . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·66 0·01 0·82 0·00 0·70 0·01 0·88 0·00 0·65 0·01 0·82 0·01 DTiO2min/melt 5·56 0·90 — — 5·90 0·80 — — 6·76 1·16 — — DAl2O3min/melt 0·73 0·06 2·00 0·03 0·66 0·05 1·85 0·01 0·79 0·02 2·04 0·04 DFeOmin/melt 7·75 0·33 0·45 0·04 5·89 0·32 0·12 0·03 8·94 0·54 0·22 0·06 DMnOmin/melt 3·28 0·58 — — 3·21 0·26 — — 4·89 0·48 — — DMgOmin/melt 27·64 0·94 0·07 0·01 29·26 1·54 0·05 0·01 41·68 1·84 0·06 0·03 DCaOmin/melt 4·16 0·08 3·38 0·16 3·39 0·09 2·81 0·08 4·50 0·10 3·95 0·23 DNa2Omin/melt 0·37 0·03 1·33 0·05 0·39 0·02 1·22 0·05 0·40 0·06 1·43 0·09 DK2Omin/melt 0·12 0·02 0·08 0·01 0·14 0·02 0·06 0·01 0·13 0·02 0·08 0·02 DClmin/melt 0·19 0·01 — — 0·21 0·03 — — 0·29 0·05 — — DFmin/melt 10·92 1·09 — — — — — — — — — — ID: . 1-15-19B . 1-16-02 . 1-16-12A . Phase: . Amph . Plag . Cpx . Cpx . Amph . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·68 0·02 0·83 0·01 0·73 0·02 0·77 0·01 0·70 0·02 0·85 0·02 DTiO2min/melt 7·96 2·49 — — 4·35 1·00 2·52 0·57 8·86 1·81 — — DAl2O3min/melt 0·70 0·06 1·99 0·03 0·38 0·07 0·27 0·03 0·71 0·04 1·91 0·04 DFeOmin/melt 6·57 0·47 0·47 0·07 5·73 0·44 3·75 0·21 8·69 0·50 0·45 0·05 DMnOmin/melt 3·03 0·71 — — 4·02 0·66 4·17 0·94 3·89 0·75 — — DMgOmin/melt 25·50 0·79 0·06 0·01 19·65 1·13 17·31 0·63 31·84 1·02 0·04 0·00 DCaOmin/melt 4·05 0·09 3·39 0·16 7·39 0·06 6·39 0·11 3·66 0·07 2·98 0·17 DNa2Omin/melt 0·37 0·04 1·22 0·07 0·13 0·02 0·10 0·01 0·38 0·04 1·33 0·07 DK2Omin/melt 0·14 0·02 0·07 0·01 — — — — 0·15 0·02 0·09 0·02 DClmin/melt 0·17 0·03 — — — — — — 0·13 0·03 — — DFmin/melt 12·20 0·92 — — — — — — — — — — ID: . 1-15-19B . 1-16-02 . 1-16-12A . Phase: . Amph . Plag . Cpx . Cpx . Amph . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·68 0·02 0·83 0·01 0·73 0·02 0·77 0·01 0·70 0·02 0·85 0·02 DTiO2min/melt 7·96 2·49 — — 4·35 1·00 2·52 0·57 8·86 1·81 — — DAl2O3min/melt 0·70 0·06 1·99 0·03 0·38 0·07 0·27 0·03 0·71 0·04 1·91 0·04 DFeOmin/melt 6·57 0·47 0·47 0·07 5·73 0·44 3·75 0·21 8·69 0·50 0·45 0·05 DMnOmin/melt 3·03 0·71 — — 4·02 0·66 4·17 0·94 3·89 0·75 — — DMgOmin/melt 25·50 0·79 0·06 0·01 19·65 1·13 17·31 0·63 31·84 1·02 0·04 0·00 DCaOmin/melt 4·05 0·09 3·39 0·16 7·39 0·06 6·39 0·11 3·66 0·07 2·98 0·17 DNa2Omin/melt 0·37 0·04 1·22 0·07 0·13 0·02 0·10 0·01 0·38 0·04 1·33 0·07 DK2Omin/melt 0·14 0·02 0·07 0·01 — — — — 0·15 0·02 0·09 0·02 DClmin/melt 0·17 0·03 — — — — — — 0·13 0·03 — — DFmin/melt 12·20 0·92 — — — — — — — — — — ID: . 1-16-12B . 1-16-14A . 1-16-14B . Phase: . Amph . Plag . Amph . Amph . Plag . Cpx . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·68 0·02 0·87 0·01 0·72 0·02 0·68 0·01 0·82 0·01 0·71 0·01 DTiO2min/melt 9·55 2·77 — — 2·74 0·46 3·59 0·33 — — 4·39 0·37 DAl2O3min/melt 0·72 0·02 1·84 0·02 0·66 0·06 0·71 0·02 2·02 0·05 0·49 0·07 DFeOmin/melt 9·08 0·45 0·47 0·03 4·86 0·36 6·09 0·12 0·55 0·03 7·03 0·25 DMnOmin/melt 4·24 1·53 — — 2·68 0·55 3·44 0·46 — — 4·93 0·45 DMgOmin/melt 38·91 0·39 0·05 0·02 41·48 1·44 51·21 1·10 0·08 0·02 33·52 0·41 DCaOmin/melt 4·16 0·05 3·03 0·20 2·87 0·05 3·75 0·14 3·15 0·25 6·85 0·23 DNa2Omin/melt 0·38 0·02 1·36 0·06 0·51 0·03 0·47 0·03 1·22 0·05 0·15 0·01 DK2Omin/melt 0·15 0·02 0·09 0·02 0·22 0·02 0·21 0·04 0·07 0·00 — — DClmin/melt 0·15 0·02 — — 0·05 0·01 — — — — — — DFmin/melt — — — — 3·30 0·11 3·98 1·58 — — — — ID: . 1-16-12B . 1-16-14A . 1-16-14B . Phase: . Amph . Plag . Amph . Amph . Plag . Cpx . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·68 0·02 0·87 0·01 0·72 0·02 0·68 0·01 0·82 0·01 0·71 0·01 DTiO2min/melt 9·55 2·77 — — 2·74 0·46 3·59 0·33 — — 4·39 0·37 DAl2O3min/melt 0·72 0·02 1·84 0·02 0·66 0·06 0·71 0·02 2·02 0·05 0·49 0·07 DFeOmin/melt 9·08 0·45 0·47 0·03 4·86 0·36 6·09 0·12 0·55 0·03 7·03 0·25 DMnOmin/melt 4·24 1·53 — — 2·68 0·55 3·44 0·46 — — 4·93 0·45 DMgOmin/melt 38·91 0·39 0·05 0·02 41·48 1·44 51·21 1·10 0·08 0·02 33·52 0·41 DCaOmin/melt 4·16 0·05 3·03 0·20 2·87 0·05 3·75 0·14 3·15 0·25 6·85 0·23 DNa2Omin/melt 0·38 0·02 1·36 0·06 0·51 0·03 0·47 0·03 1·22 0·05 0·15 0·01 DK2Omin/melt 0·15 0·02 0·09 0·02 0·22 0·02 0·21 0·04 0·07 0·00 — — DClmin/melt 0·15 0·02 — — 0·05 0·01 — — — — — — DFmin/melt — — — — 3·30 0·11 3·98 1·58 — — — — ID: . 1-16-18A . 1-16-18B . 1-16-21A . 1-16-21B . Phase: . Amph . Amph . Amph . Amph . Cpx . . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·71 0·01 0·72 0·01 0·69 0·01 0·71 0·01 0·77 0·00 DTiO2min/melt 4·96 1·06 4·04 0·70 4·67 0·58 4·50 0·69 3·22 0·12 DAl2O3min/melt 0·66 0·05 0·65 0·05 0·66 0·03 0·62 0·05 0·33 0·00 DFeOmin/melt 7·52 0·63 7·39 0·34 5·50 0·26 4·60 0·27 3·64 0·11 DMnOmin/melt 3·64 0·98 3·61 0·89 3·57 0·83 3·19 0·77 4·89 0·84 DMgOmin/melt 33·45 0·84 33·74 0·71 24·27 0·85 24·83 1·15 20·72 0·25 DCaOmin/melt 3·79 0·08 3·64 0·06 3·39 0·07 3·41 0·09 6·38 0·30 DNa2Omin/melt 0·35 0·04 0·33 0·03 0·38 0·04 0·40 0·04 0·11 0·01 DK2Omin/melt 0·14 0·02 0·13 0·02 0·15 0·02 0·13 0·02 — — DClmin/melt 0·10 0·02 0·10 0·03 0·18 0·03 0·17 0·02 — — DFmin/melt — — — — — — — — — — ID: . 1-16-18A . 1-16-18B . 1-16-21A . 1-16-21B . Phase: . Amph . Amph . Amph . Amph . Cpx . . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·71 0·01 0·72 0·01 0·69 0·01 0·71 0·01 0·77 0·00 DTiO2min/melt 4·96 1·06 4·04 0·70 4·67 0·58 4·50 0·69 3·22 0·12 DAl2O3min/melt 0·66 0·05 0·65 0·05 0·66 0·03 0·62 0·05 0·33 0·00 DFeOmin/melt 7·52 0·63 7·39 0·34 5·50 0·26 4·60 0·27 3·64 0·11 DMnOmin/melt 3·64 0·98 3·61 0·89 3·57 0·83 3·19 0·77 4·89 0·84 DMgOmin/melt 33·45 0·84 33·74 0·71 24·27 0·85 24·83 1·15 20·72 0·25 DCaOmin/melt 3·79 0·08 3·64 0·06 3·39 0·07 3·41 0·09 6·38 0·30 DNa2Omin/melt 0·35 0·04 0·33 0·03 0·38 0·04 0·40 0·04 0·11 0·01 DK2Omin/melt 0·14 0·02 0·13 0·02 0·15 0·02 0·13 0·02 — — DClmin/melt 0·10 0·02 0·10 0·03 0·18 0·03 0·17 0·02 — — DFmin/melt — — — — — — — — — — ID: . 1-16-22A . 1-16-22B . . Phase: . Amph . Plag . Cpx . Amph . Plag . . . average . error . average . error . . . average . error . . DSiO2min/melt 0·69 0·01 0·85 0·01 0·72 0·69 0·87 0·02 DTiO2min/melt 4·47 0·70 0·05 0·05 4·06 4·97 0·07 0·07 DAl2O3min/melt 0·64 0·04 1·88 0·03 0·38 0·65 1·87 0·03 DFeOmin/melt 6·90 0·31 0·17 0·03 5·61 5·99 0·17 0·04 DMnOmin/melt 4·63 1·31 — — 4·70 3·88 — — DMgOmin/melt 40·05 1·88 0·08 0·03 31·77 47·36 0·05 0·02 DCaOmin/melt 3·83 0·18 3·07 0·09 7·11 2·86 2·35 0·10 DNa2Omin/melt 0·37 0·04 1·29 0·05 0·13 0·47 1·41 0·07 DK2Omin/melt 0·15 0·02 0·09 0·01 — 0·17 0·10 0·02 DClmin/melt 0·18 0·02 — — — 0·10 — — DFmin/melt — — — — — 3·38 — — ID: 1-16-23A 1-16-23B Phase: Amph Plag Cpx Amph Cpx average error average error average error average error DSiO2min/melt 0·68 0·02 0·86 0·02 0·73 0·71 0·02 0·74 0·01 DTiO2min/melt 6·09 0·92 0·06 0·05 4·39 3·05 0·38 4·39 0·34 DAl2O3min/melt 0·68 0·06 1·87 0·05 0·37 0·56 0·07 0·39 0·01 DFeOmin/melt 6·25 0·74 0·21 0·04 5·17 5·85 0·38 4·92 0·33 DMnOmin/melt 3·57 0·80 — — 4·88 4·15 0·94 5·39 1·63 DMgOmin/melt 30·63 3·05 0·04 0·01 25·87 50·31 2·48 37·43 2·87 DCaOmin/melt 3·74 0·09 2·97 0·27 6·87 2·85 0·03 5·20 0·28 DNa2Omin/melt 0·41 0·03 1·35 0·10 0·15 0·47 0·04 0·15 0·01 DK2Omin/melt 0·16 0·03 0·10 0·02 — 0·24 0·04 — — DClmin/melt 0·28 0·07 — — — 0·07 0·01 — — DFmin/melt — — — — — 2·93 0·31 — — ID: . 1-16-22A . 1-16-22B . . Phase: . Amph . Plag . Cpx . Amph . Plag . . . average . error . average . error . . . average . error . . DSiO2min/melt 0·69 0·01 0·85 0·01 0·72 0·69 0·87 0·02 DTiO2min/melt 4·47 0·70 0·05 0·05 4·06 4·97 0·07 0·07 DAl2O3min/melt 0·64 0·04 1·88 0·03 0·38 0·65 1·87 0·03 DFeOmin/melt 6·90 0·31 0·17 0·03 5·61 5·99 0·17 0·04 DMnOmin/melt 4·63 1·31 — — 4·70 3·88 — — DMgOmin/melt 40·05 1·88 0·08 0·03 31·77 47·36 0·05 0·02 DCaOmin/melt 3·83 0·18 3·07 0·09 7·11 2·86 2·35 0·10 DNa2Omin/melt 0·37 0·04 1·29 0·05 0·13 0·47 1·41 0·07 DK2Omin/melt 0·15 0·02 0·09 0·01 — 0·17 0·10 0·02 DClmin/melt 0·18 0·02 — — — 0·10 — — DFmin/melt — — — — — 3·38 — — ID: 1-16-23A 1-16-23B Phase: Amph Plag Cpx Amph Cpx average error average error average error average error DSiO2min/melt 0·68 0·02 0·86 0·02 0·73 0·71 0·02 0·74 0·01 DTiO2min/melt 6·09 0·92 0·06 0·05 4·39 3·05 0·38 4·39 0·34 DAl2O3min/melt 0·68 0·06 1·87 0·05 0·37 0·56 0·07 0·39 0·01 DFeOmin/melt 6·25 0·74 0·21 0·04 5·17 5·85 0·38 4·92 0·33 DMnOmin/melt 3·57 0·80 — — 4·88 4·15 0·94 5·39 1·63 DMgOmin/melt 30·63 3·05 0·04 0·01 25·87 50·31 2·48 37·43 2·87 DCaOmin/melt 3·74 0·09 2·97 0·27 6·87 2·85 0·03 5·20 0·28 DNa2Omin/melt 0·41 0·03 1·35 0·10 0·15 0·47 0·04 0·15 0·01 DK2Omin/melt 0·16 0·03 0·10 0·02 — 0·24 0·04 — — DClmin/melt 0·28 0·07 — — — 0·07 0·01 — — DFmin/melt — — — — — 2·93 0·31 — — Open in new tab Table 3: Calculated mineral–melt partition coefficients for each mineral phase present in each experiment, and associated propagated error. Mineral phase abbreviations as in Table 2 ID: . 1-13-23 . 1-14-02 . 1-14-04A . Phase: . Amph . Plag . Amph . Plag . Cpx . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·64 0·01 0·79 0·01 0·65 0·02 0·80 0·01 0·77 0·02 0·80 0·01 DTiO2min/melt 4·34 1·01 — — 8·21 1·26 — — 1·27 0·13 — — DAl2O3min/melt 0·80 0·02 2·25 0·04 0·78 0·07 2·11 0·04 0·17 0·06 2·07 0·04 DFeOmin/melt 9·04 0·47 0·53 0·11 8·51 0·62 0·56 0·05 3·78 0·19 0·25 0·03 DMnOmin/melt 5·59 0·60 — — 3·61 0·57 — — 6·03 1·54 — — DMgOmin/melt 32·67 0·78 0·08 0·03 28·80 1·20 0·07 0·02 16·98 1·42 0·07 0·01 DCaOmin/melt 5·87 0·27 4·79 0·21 4·60 0·12 3·95 0·22 6·10 0·39 3·55 0·16 DNa2Omin/melt 0·37 0·03 1·35 0·06 0·40 0·04 1·34 0·09 0·06 0·01 1·10 0·05 DK2Omin/melt 0·09 0·01 0·09 0·02 0·12 0·04 0·07 0·01 — — 0·06 0·01 DClmin/melt — — — — — — — — — — — — DFmin/melt 6·59 1·25 — — 16·63 1·05 — — — — — — ID: 1-14-04B 1-14-04C 1-14-05 1-15-12A Phase: Cpx Plag Amph Amph Amph Plag average error average error average error average error average error average error DSiO2min/melt 0·78 0·02 0·85 0·01 0·69 0·01 0·70 0·02 0·66 0·01 0·81 0·01 DTiO2min/melt 1·66 0·10 — — 3·08 0·25 4·11 0·32 4·87 1·39 — — DAl2O3min/melt 0·16 0·06 2·03 0·04 0·63 0·04 0·60 0·07 0·72 0·06 2·02 0·03 DFeOmin/melt 3·45 0·51 0·17 0·03 4·72 0·23 5·50 0·30 8·66 0·46 0·49 0·05 DMnOmin/melt 5·33 1·39 — — 2·99 1·09 3·34 0·47 3·98 0·73 — — DMgOmin/melt 19·11 1·77 0·06 0·01 26·94 0·68 23·19 0·91 28·83 0·76 0·05 0·02 DCaOmin/melt 5·67 0·81 3·27 0·12 2·96 0·08 3·47 0·05 4·24 0·08 3·58 0·22 DNa2Omin/melt 0·06 0·02 1·19 0·06 0·47 0·04 0·36 0·03 0·38 0·03 1·35 0·10 DK2Omin/melt — — 0·08 0·03 0·18 0·03 0·12 0·01 0·11 0·02 0·08 0·02 DClmin/melt — — — — — — 0·17 0·01 — — — — DFmin/melt — — — — 3·09 0·75 5·65 0·53 8·87 1·10 — — ID: . 1-13-23 . 1-14-02 . 1-14-04A . Phase: . Amph . Plag . Amph . Plag . Cpx . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·64 0·01 0·79 0·01 0·65 0·02 0·80 0·01 0·77 0·02 0·80 0·01 DTiO2min/melt 4·34 1·01 — — 8·21 1·26 — — 1·27 0·13 — — DAl2O3min/melt 0·80 0·02 2·25 0·04 0·78 0·07 2·11 0·04 0·17 0·06 2·07 0·04 DFeOmin/melt 9·04 0·47 0·53 0·11 8·51 0·62 0·56 0·05 3·78 0·19 0·25 0·03 DMnOmin/melt 5·59 0·60 — — 3·61 0·57 — — 6·03 1·54 — — DMgOmin/melt 32·67 0·78 0·08 0·03 28·80 1·20 0·07 0·02 16·98 1·42 0·07 0·01 DCaOmin/melt 5·87 0·27 4·79 0·21 4·60 0·12 3·95 0·22 6·10 0·39 3·55 0·16 DNa2Omin/melt 0·37 0·03 1·35 0·06 0·40 0·04 1·34 0·09 0·06 0·01 1·10 0·05 DK2Omin/melt 0·09 0·01 0·09 0·02 0·12 0·04 0·07 0·01 — — 0·06 0·01 DClmin/melt — — — — — — — — — — — — DFmin/melt 6·59 1·25 — — 16·63 1·05 — — — — — — ID: 1-14-04B 1-14-04C 1-14-05 1-15-12A Phase: Cpx Plag Amph Amph Amph Plag average error average error average error average error average error average error DSiO2min/melt 0·78 0·02 0·85 0·01 0·69 0·01 0·70 0·02 0·66 0·01 0·81 0·01 DTiO2min/melt 1·66 0·10 — — 3·08 0·25 4·11 0·32 4·87 1·39 — — DAl2O3min/melt 0·16 0·06 2·03 0·04 0·63 0·04 0·60 0·07 0·72 0·06 2·02 0·03 DFeOmin/melt 3·45 0·51 0·17 0·03 4·72 0·23 5·50 0·30 8·66 0·46 0·49 0·05 DMnOmin/melt 5·33 1·39 — — 2·99 1·09 3·34 0·47 3·98 0·73 — — DMgOmin/melt 19·11 1·77 0·06 0·01 26·94 0·68 23·19 0·91 28·83 0·76 0·05 0·02 DCaOmin/melt 5·67 0·81 3·27 0·12 2·96 0·08 3·47 0·05 4·24 0·08 3·58 0·22 DNa2Omin/melt 0·06 0·02 1·19 0·06 0·47 0·04 0·36 0·03 0·38 0·03 1·35 0·10 DK2Omin/melt — — 0·08 0·03 0·18 0·03 0·12 0·01 0·11 0·02 0·08 0·02 DClmin/melt — — — — — — 0·17 0·01 — — — — DFmin/melt — — — — 3·09 0·75 5·65 0·53 8·87 1·10 — — ID: . 1-15-12B . 1-15-15A . 1-15-15B . Phase: . Amph . Plag . Amph . Plag . Amph . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·66 0·01 0·82 0·00 0·70 0·01 0·88 0·00 0·65 0·01 0·82 0·01 DTiO2min/melt 5·56 0·90 — — 5·90 0·80 — — 6·76 1·16 — — DAl2O3min/melt 0·73 0·06 2·00 0·03 0·66 0·05 1·85 0·01 0·79 0·02 2·04 0·04 DFeOmin/melt 7·75 0·33 0·45 0·04 5·89 0·32 0·12 0·03 8·94 0·54 0·22 0·06 DMnOmin/melt 3·28 0·58 — — 3·21 0·26 — — 4·89 0·48 — — DMgOmin/melt 27·64 0·94 0·07 0·01 29·26 1·54 0·05 0·01 41·68 1·84 0·06 0·03 DCaOmin/melt 4·16 0·08 3·38 0·16 3·39 0·09 2·81 0·08 4·50 0·10 3·95 0·23 DNa2Omin/melt 0·37 0·03 1·33 0·05 0·39 0·02 1·22 0·05 0·40 0·06 1·43 0·09 DK2Omin/melt 0·12 0·02 0·08 0·01 0·14 0·02 0·06 0·01 0·13 0·02 0·08 0·02 DClmin/melt 0·19 0·01 — — 0·21 0·03 — — 0·29 0·05 — — DFmin/melt 10·92 1·09 — — — — — — — — — — ID: . 1-15-12B . 1-15-15A . 1-15-15B . Phase: . Amph . Plag . Amph . Plag . Amph . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·66 0·01 0·82 0·00 0·70 0·01 0·88 0·00 0·65 0·01 0·82 0·01 DTiO2min/melt 5·56 0·90 — — 5·90 0·80 — — 6·76 1·16 — — DAl2O3min/melt 0·73 0·06 2·00 0·03 0·66 0·05 1·85 0·01 0·79 0·02 2·04 0·04 DFeOmin/melt 7·75 0·33 0·45 0·04 5·89 0·32 0·12 0·03 8·94 0·54 0·22 0·06 DMnOmin/melt 3·28 0·58 — — 3·21 0·26 — — 4·89 0·48 — — DMgOmin/melt 27·64 0·94 0·07 0·01 29·26 1·54 0·05 0·01 41·68 1·84 0·06 0·03 DCaOmin/melt 4·16 0·08 3·38 0·16 3·39 0·09 2·81 0·08 4·50 0·10 3·95 0·23 DNa2Omin/melt 0·37 0·03 1·33 0·05 0·39 0·02 1·22 0·05 0·40 0·06 1·43 0·09 DK2Omin/melt 0·12 0·02 0·08 0·01 0·14 0·02 0·06 0·01 0·13 0·02 0·08 0·02 DClmin/melt 0·19 0·01 — — 0·21 0·03 — — 0·29 0·05 — — DFmin/melt 10·92 1·09 — — — — — — — — — — ID: . 1-15-19B . 1-16-02 . 1-16-12A . Phase: . Amph . Plag . Cpx . Cpx . Amph . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·68 0·02 0·83 0·01 0·73 0·02 0·77 0·01 0·70 0·02 0·85 0·02 DTiO2min/melt 7·96 2·49 — — 4·35 1·00 2·52 0·57 8·86 1·81 — — DAl2O3min/melt 0·70 0·06 1·99 0·03 0·38 0·07 0·27 0·03 0·71 0·04 1·91 0·04 DFeOmin/melt 6·57 0·47 0·47 0·07 5·73 0·44 3·75 0·21 8·69 0·50 0·45 0·05 DMnOmin/melt 3·03 0·71 — — 4·02 0·66 4·17 0·94 3·89 0·75 — — DMgOmin/melt 25·50 0·79 0·06 0·01 19·65 1·13 17·31 0·63 31·84 1·02 0·04 0·00 DCaOmin/melt 4·05 0·09 3·39 0·16 7·39 0·06 6·39 0·11 3·66 0·07 2·98 0·17 DNa2Omin/melt 0·37 0·04 1·22 0·07 0·13 0·02 0·10 0·01 0·38 0·04 1·33 0·07 DK2Omin/melt 0·14 0·02 0·07 0·01 — — — — 0·15 0·02 0·09 0·02 DClmin/melt 0·17 0·03 — — — — — — 0·13 0·03 — — DFmin/melt 12·20 0·92 — — — — — — — — — — ID: . 1-15-19B . 1-16-02 . 1-16-12A . Phase: . Amph . Plag . Cpx . Cpx . Amph . Plag . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·68 0·02 0·83 0·01 0·73 0·02 0·77 0·01 0·70 0·02 0·85 0·02 DTiO2min/melt 7·96 2·49 — — 4·35 1·00 2·52 0·57 8·86 1·81 — — DAl2O3min/melt 0·70 0·06 1·99 0·03 0·38 0·07 0·27 0·03 0·71 0·04 1·91 0·04 DFeOmin/melt 6·57 0·47 0·47 0·07 5·73 0·44 3·75 0·21 8·69 0·50 0·45 0·05 DMnOmin/melt 3·03 0·71 — — 4·02 0·66 4·17 0·94 3·89 0·75 — — DMgOmin/melt 25·50 0·79 0·06 0·01 19·65 1·13 17·31 0·63 31·84 1·02 0·04 0·00 DCaOmin/melt 4·05 0·09 3·39 0·16 7·39 0·06 6·39 0·11 3·66 0·07 2·98 0·17 DNa2Omin/melt 0·37 0·04 1·22 0·07 0·13 0·02 0·10 0·01 0·38 0·04 1·33 0·07 DK2Omin/melt 0·14 0·02 0·07 0·01 — — — — 0·15 0·02 0·09 0·02 DClmin/melt 0·17 0·03 — — — — — — 0·13 0·03 — — DFmin/melt 12·20 0·92 — — — — — — — — — — ID: . 1-16-12B . 1-16-14A . 1-16-14B . Phase: . Amph . Plag . Amph . Amph . Plag . Cpx . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·68 0·02 0·87 0·01 0·72 0·02 0·68 0·01 0·82 0·01 0·71 0·01 DTiO2min/melt 9·55 2·77 — — 2·74 0·46 3·59 0·33 — — 4·39 0·37 DAl2O3min/melt 0·72 0·02 1·84 0·02 0·66 0·06 0·71 0·02 2·02 0·05 0·49 0·07 DFeOmin/melt 9·08 0·45 0·47 0·03 4·86 0·36 6·09 0·12 0·55 0·03 7·03 0·25 DMnOmin/melt 4·24 1·53 — — 2·68 0·55 3·44 0·46 — — 4·93 0·45 DMgOmin/melt 38·91 0·39 0·05 0·02 41·48 1·44 51·21 1·10 0·08 0·02 33·52 0·41 DCaOmin/melt 4·16 0·05 3·03 0·20 2·87 0·05 3·75 0·14 3·15 0·25 6·85 0·23 DNa2Omin/melt 0·38 0·02 1·36 0·06 0·51 0·03 0·47 0·03 1·22 0·05 0·15 0·01 DK2Omin/melt 0·15 0·02 0·09 0·02 0·22 0·02 0·21 0·04 0·07 0·00 — — DClmin/melt 0·15 0·02 — — 0·05 0·01 — — — — — — DFmin/melt — — — — 3·30 0·11 3·98 1·58 — — — — ID: . 1-16-12B . 1-16-14A . 1-16-14B . Phase: . Amph . Plag . Amph . Amph . Plag . Cpx . . average . error . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·68 0·02 0·87 0·01 0·72 0·02 0·68 0·01 0·82 0·01 0·71 0·01 DTiO2min/melt 9·55 2·77 — — 2·74 0·46 3·59 0·33 — — 4·39 0·37 DAl2O3min/melt 0·72 0·02 1·84 0·02 0·66 0·06 0·71 0·02 2·02 0·05 0·49 0·07 DFeOmin/melt 9·08 0·45 0·47 0·03 4·86 0·36 6·09 0·12 0·55 0·03 7·03 0·25 DMnOmin/melt 4·24 1·53 — — 2·68 0·55 3·44 0·46 — — 4·93 0·45 DMgOmin/melt 38·91 0·39 0·05 0·02 41·48 1·44 51·21 1·10 0·08 0·02 33·52 0·41 DCaOmin/melt 4·16 0·05 3·03 0·20 2·87 0·05 3·75 0·14 3·15 0·25 6·85 0·23 DNa2Omin/melt 0·38 0·02 1·36 0·06 0·51 0·03 0·47 0·03 1·22 0·05 0·15 0·01 DK2Omin/melt 0·15 0·02 0·09 0·02 0·22 0·02 0·21 0·04 0·07 0·00 — — DClmin/melt 0·15 0·02 — — 0·05 0·01 — — — — — — DFmin/melt — — — — 3·30 0·11 3·98 1·58 — — — — ID: . 1-16-18A . 1-16-18B . 1-16-21A . 1-16-21B . Phase: . Amph . Amph . Amph . Amph . Cpx . . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·71 0·01 0·72 0·01 0·69 0·01 0·71 0·01 0·77 0·00 DTiO2min/melt 4·96 1·06 4·04 0·70 4·67 0·58 4·50 0·69 3·22 0·12 DAl2O3min/melt 0·66 0·05 0·65 0·05 0·66 0·03 0·62 0·05 0·33 0·00 DFeOmin/melt 7·52 0·63 7·39 0·34 5·50 0·26 4·60 0·27 3·64 0·11 DMnOmin/melt 3·64 0·98 3·61 0·89 3·57 0·83 3·19 0·77 4·89 0·84 DMgOmin/melt 33·45 0·84 33·74 0·71 24·27 0·85 24·83 1·15 20·72 0·25 DCaOmin/melt 3·79 0·08 3·64 0·06 3·39 0·07 3·41 0·09 6·38 0·30 DNa2Omin/melt 0·35 0·04 0·33 0·03 0·38 0·04 0·40 0·04 0·11 0·01 DK2Omin/melt 0·14 0·02 0·13 0·02 0·15 0·02 0·13 0·02 — — DClmin/melt 0·10 0·02 0·10 0·03 0·18 0·03 0·17 0·02 — — DFmin/melt — — — — — — — — — — ID: . 1-16-18A . 1-16-18B . 1-16-21A . 1-16-21B . Phase: . Amph . Amph . Amph . Amph . Cpx . . average . error . average . error . average . error . average . error . average . error . DSiO2min/melt 0·71 0·01 0·72 0·01 0·69 0·01 0·71 0·01 0·77 0·00 DTiO2min/melt 4·96 1·06 4·04 0·70 4·67 0·58 4·50 0·69 3·22 0·12 DAl2O3min/melt 0·66 0·05 0·65 0·05 0·66 0·03 0·62 0·05 0·33 0·00 DFeOmin/melt 7·52 0·63 7·39 0·34 5·50 0·26 4·60 0·27 3·64 0·11 DMnOmin/melt 3·64 0·98 3·61 0·89 3·57 0·83 3·19 0·77 4·89 0·84 DMgOmin/melt 33·45 0·84 33·74 0·71 24·27 0·85 24·83 1·15 20·72 0·25 DCaOmin/melt 3·79 0·08 3·64 0·06 3·39 0·07 3·41 0·09 6·38 0·30 DNa2Omin/melt 0·35 0·04 0·33 0·03 0·38 0·04 0·40 0·04 0·11 0·01 DK2Omin/melt 0·14 0·02 0·13 0·02 0·15 0·02 0·13 0·02 — — DClmin/melt 0·10 0·02 0·10 0·03 0·18 0·03 0·17 0·02 — — DFmin/melt — — — — — — — — — — ID: . 1-16-22A . 1-16-22B . . Phase: . Amph . Plag . Cpx . Amph . Plag . . . average . error . average . error . . . average . error . . DSiO2min/melt 0·69 0·01 0·85 0·01 0·72 0·69 0·87 0·02 DTiO2min/melt 4·47 0·70 0·05 0·05 4·06 4·97 0·07 0·07 DAl2O3min/melt 0·64 0·04 1·88 0·03 0·38 0·65 1·87 0·03 DFeOmin/melt 6·90 0·31 0·17 0·03 5·61 5·99 0·17 0·04 DMnOmin/melt 4·63 1·31 — — 4·70 3·88 — — DMgOmin/melt 40·05 1·88 0·08 0·03 31·77 47·36 0·05 0·02 DCaOmin/melt 3·83 0·18 3·07 0·09 7·11 2·86 2·35 0·10 DNa2Omin/melt 0·37 0·04 1·29 0·05 0·13 0·47 1·41 0·07 DK2Omin/melt 0·15 0·02 0·09 0·01 — 0·17 0·10 0·02 DClmin/melt 0·18 0·02 — — — 0·10 — — DFmin/melt — — — — — 3·38 — — ID: 1-16-23A 1-16-23B Phase: Amph Plag Cpx Amph Cpx average error average error average error average error DSiO2min/melt 0·68 0·02 0·86 0·02 0·73 0·71 0·02 0·74 0·01 DTiO2min/melt 6·09 0·92 0·06 0·05 4·39 3·05 0·38 4·39 0·34 DAl2O3min/melt 0·68 0·06 1·87 0·05 0·37 0·56 0·07 0·39 0·01 DFeOmin/melt 6·25 0·74 0·21 0·04 5·17 5·85 0·38 4·92 0·33 DMnOmin/melt 3·57 0·80 — — 4·88 4·15 0·94 5·39 1·63 DMgOmin/melt 30·63 3·05 0·04 0·01 25·87 50·31 2·48 37·43 2·87 DCaOmin/melt 3·74 0·09 2·97 0·27 6·87 2·85 0·03 5·20 0·28 DNa2Omin/melt 0·41 0·03 1·35 0·10 0·15 0·47 0·04 0·15 0·01 DK2Omin/melt 0·16 0·03 0·10 0·02 — 0·24 0·04 — — DClmin/melt 0·28 0·07 — — — 0·07 0·01 — — DFmin/melt — — — — — 2·93 0·31 — — ID: . 1-16-22A . 1-16-22B . . Phase: . Amph . Plag . Cpx . Amph . Plag . . . average . error . average . error . . . average . error . . DSiO2min/melt 0·69 0·01 0·85 0·01 0·72 0·69 0·87 0·02 DTiO2min/melt 4·47 0·70 0·05 0·05 4·06 4·97 0·07 0·07 DAl2O3min/melt 0·64 0·04 1·88 0·03 0·38 0·65 1·87 0·03 DFeOmin/melt 6·90 0·31 0·17 0·03 5·61 5·99 0·17 0·04 DMnOmin/melt 4·63 1·31 — — 4·70 3·88 — — DMgOmin/melt 40·05 1·88 0·08 0·03 31·77 47·36 0·05 0·02 DCaOmin/melt 3·83 0·18 3·07 0·09 7·11 2·86 2·35 0·10 DNa2Omin/melt 0·37 0·04 1·29 0·05 0·13 0·47 1·41 0·07 DK2Omin/melt 0·15 0·02 0·09 0·01 — 0·17 0·10 0·02 DClmin/melt 0·18 0·02 — — — 0·10 — — DFmin/melt — — — — — 3·38 — — ID: 1-16-23A 1-16-23B Phase: Amph Plag Cpx Amph Cpx average error average error average error average error DSiO2min/melt 0·68 0·02 0·86 0·02 0·73 0·71 0·02 0·74 0·01 DTiO2min/melt 6·09 0·92 0·06 0·05 4·39 3·05 0·38 4·39 0·34 DAl2O3min/melt 0·68 0·06 1·87 0·05 0·37 0·56 0·07 0·39 0·01 DFeOmin/melt 6·25 0·74 0·21 0·04 5·17 5·85 0·38 4·92 0·33 DMnOmin/melt 3·57 0·80 — — 4·88 4·15 0·94 5·39 1·63 DMgOmin/melt 30·63 3·05 0·04 0·01 25·87 50·31 2·48 37·43 2·87 DCaOmin/melt 3·74 0·09 2·97 0·27 6·87 2·85 0·03 5·20 0·28 DNa2Omin/melt 0·41 0·03 1·35 0·10 0·15 0·47 0·04 0·15 0·01 DK2Omin/melt 0·16 0·03 0·10 0·02 — 0·24 0·04 — — DClmin/melt 0·28 0·07 — — — 0·07 0·01 — — DFmin/melt — — — — — 2·93 0·31 — — Open in new tab Fig. 6. Open in new tabDownload slide Major element partitioning summary plot for experimental phases, showing the range (grey bar) and average (coloured spot) across all analyses for each mineral. All data are derived from electron microprobe analyses and are based on weight per cent concentrations. Experimental fO2 and the behaviour of Fe The range in the experimental fO2 conditions in these experiments is strongly reflected in the behaviour of Fe in the glasses and phenocryst phases, and the effect is also phase dependent. Importantly for further discussion of the crystal–melt partitioning of Fe, it should be noted that the abundance and proportions of Fe–Ti oxide minerals, as a direct result of the fO2 variations documented above, will affect the melt chemistry through changes in the availability of ferric and ferrous iron to the other crystallizing phases. Therefore, a complex feedback exists between changing fO2, suppression or promotion of oxide crystallization and its effect on melt Fe contents, and thus the calculated DFeOcrystal/melt values for the other phases. Separating these independent but linked processes is beyond the scope of this investigation, but the partitioning data presented here should still be considered a useful representation of how external fO2 variations and melt compositional controls will affect the mineral–melt partitioning of Fe. In these data, amphibole DFeOamph/melt values across the experiments range from 4·04 to 10·05, but there is no systematic difference in the FeO partition coefficient between higher and lower fO2 runs (fO2 > NNO +1·0 compared with < NNO +1·0). DFeOamph/melt is 6·07 ± 1·22 at lower fO2 conditions and is 7·47 ± 1·37 at intrinsic fO2 conditions (Fig. 3d). The total calculated Fe3+ a.p.f.u. in the amphibole is also the same under both oxidizing and more reducing conditions (1·01 ± 0·1 in low fO2 experiments and 0·94 ± 0·1 in higher O2 experiments), but Mg is lower and Fe2+ is higher in amphibole crystallized in low fO2 experiments. Thus, it appears that fO2 has little impact on the actual DFeOamph/melt but clearly influences the FeTOT a.p.f.u. and Fe2+/Fe3+ ratio of the amphibole, as a result of higher melt FeO. There is also a positive correlation between DAl2O3amph/melt and DFeOamph/melt, consistent with Tschermak’s exchange. In contrast to the amphibole, plagioclase crystals from experiments with fO2 < NNO +1·0 runs show lower FeO concentrations and average DFeOplag/melt values (0·19 ± 0·04) that are less than half of those from more oxidizing runs where fO2 > NNO +1·0 (0·50 ± 0·04) (Fig. 4b). Similarly to plagioclase, the overall average DFeOcpx/melt values are lower in experiments with fO2 < NNO +1·0 compared with more oxidizing ones (>NNO +1·0), along with an apparent effect of increasing pressure, but, similarly to the amphibole, this is despite the clinopyroxene having greater FeTOT a.p.f.u. in lower fO2 experiments (Fig. 5b). The changing DFeOcpx/melt values are also associated with other major element changes, as shown in Fig. 7. There are positive correlations between increasing DAl2O3cpx/melt and DTiO2cpx/melt ⁠, DFeOcpx/melt, and DNa2Ocpx/melt ⁠, whereas DSiO2cpx/melt correlates negatively. Fig. 7. Open in new tabDownload slide A plot showing the effect of major element covariations and coupled exchange on clinopyroxene compositions. Halogen partitioning into amphibole Across all amphibole analyses, DClamph/melt ranges from 0·04 to 0·40, and averages 0·16 ± 0·07 (Fig. 6). Chlorine partitioning is most strongly related to the amphibole Mg/(Mg + FeTOT) ratio (Fig. 8a), and amphibole Si contents (Fig. 8b). When RTlnDClamph/melt is plotted versus the Mg/(Mg + FeTOT) ratio of the amphibole (Fig. 8a), where R is the gas constant and T is the experimental temperature in Kelvin, there are two distinct clusters of data, for experiments with fO2 > NNO +1·0 and those with fO2 < NNO +1·0. However, both clusters show decreasing partition coefficients for Cl as a function of increasing amphibole Mg contents, consistent with the Mg–Cl crystallographic avoidance principle. Addition of Cl to the melt has no apparent systematic effect on the absolute DClamph/melt ⁠; that is, Henrian partitioning behaviour is maintained with >0·50 wt % Cl in the melt. Importantly, in the three experiments that were doped with both halogens, it also can be seen how F incorporation affects Cl partition coefficients. There is a clear decrease in the DClamph/melt value as a function of increased F in the amphibole and melt, with these experiments showing much lower lnDClamph/melt values than would be expected from other comparable amphibole with similar Mg/(Mg + FeTOT) ratios (Fig. 8a). Fig. 8. Open in new tabDownload slide Plots of average RTlnDamphibole/melt relationships for Cl vs Mg/(Mg + FeTOT) (a) and for Cl vs SiO2 (b). Higher Fe amphiboles from lower fO2 experiments (where fO2 < NNO +1) in (a) show relatively increased Cl partition coefficients, and are consistent with the Mg–Cl relationships seen in other experimental datasets, whereas expansion of the octahedral sheet and [4]Al substitution for Si owing to Cl incorporation is responsible for the trend in (b). It should be noted also in both figures that the addition of F to the melt and amphibole leads to lower Cl partition coefficients owing to increased competition in the amphibole structure. Values of DFamph/melt can be calculated for all experiments where F concentrations in the melt were above 110 ppm, the limit of detection for EPMA measurements. Across all amphibole analyses, DFamph/melt ranges from 2·24 to 19·33, and averages 7·05 ± 4·35 (Fig. 6). There is a non-linear decrease in the DFamph/melt as the melt F increases, but F is still strongly compatible at all melt concentrations (Fig. 9a). Amphiboles crystallized from melts with F concentrations below the detection limit still contain a minimum of ∼700 ppm F, and average ∼1500 ppm (Fig. 3c), giving potential maximum DFamph/melt values of ≥20. At higher melt F contents, the partition coefficients decrease to <5, as the amphiboles incorporate up to 2·5 wt % F (Fig. 9a and b). It can also be noted that amphibole from experiments with fO2 < NNO +1·0 (and hence also those with higher Fe contents) show marginally lower DFamph/melt values, as a function of the Fe–F avoidance principle (Fig. 9a). Furthermore, in these experimental amphiboles, F incorporation is also related to changes in DTiO2amph/melt ⁠, with higher absolute F contents (i.e. experiments with the lowest DFamph/melt) also reflected in lower TiO2 contents (Fig. 9b). In summary, F is at least an order of magnitude more compatible in amphibole than Cl for all experiments and all run conditions. Fig. 9. Open in new tabDownload slide Variations in, and correlations between, the amphibole–melt F partition coefficient with changing melt F (a), and changing TiO2 partitioning (b). It should be noted that higher Fe amphiboles from lower fO2 experiments show relatively decreased F partition coefficients, and that as both melt F and amphibole F increase, calculated DFamph/melt decreases. DISCUSSION Comparison with published partitioning data Amphibole: major elements Sato et al. (2005) reported experimental partitioning data for amphibole grown in Mt Unzen dacitic melts at 800–850°C, 2–3 kbar, and fO2 ≈ NNO; that is, at very similar conditions and chemical compositions to our amphibole, although their starting glasses were up to 1 wt % richer in FeO, MgO, CaO, and K2O, and ∼1 wt % poorer in Na2O. Matjuschkin et al. (2016) reported amphibole–melt and plagioclase–melt experimental work at 850–950°C, higher pressures of 10–15 kbar, a range of fO2 from Co–CoO to magnetite–haematite (M–H), and where the experiments were buffered by mixed C–O–H–S fluids in equilibrium with trachydacite melts. Another recent study by Nandedkar et al. (2016) also provides experimental data for three amphibole–melt partitioning experiments specifically with dacitic–rhyodacitic residual liquids, which can be directly compared with our data. The experiments of Nandedkar et al. (2016) were conducted at higher pressures of 7 kbar, but at similar temperatures (830–890°C) and fO2 close to, or slightly above, NNO. The three run product glasses are more peraluminous, ranging from molar Al/(Na + K + Ca) of 1·10 to 1·20, and have lower molar Na/(Na + K) ratios ranging from 0·76 to 0·72, compared with our run product glasses (Table 4). Figure 10a and b illustrates comparisons between lnDamph/melt values for our data and those of the aforementioned investigations. In general, despite the varied temperature, pressure, fO2 conditions and fluid compositions, and differences in residual glass chemistries, there is excellent agreement between the compositions of the amphibole crystallized in all four investigations. In particular, all major-element data from Sato et al. (2005) are almost identical to our data in both the range and average for each oxide component (Table 4). It is important to note that the average DAl2O3amph/melt is greater in the Matjuschkin et al. (2016) and Nandedkar et al. (2016) studies, which is consistent with the greater experimental pressures and the pressure dependence of Al partitioning. However, in general, the partitioning relationships and their co-variations are remarkably consistent across all of the experimental data (Fig. 10a and b). The incorporation of Mg correlates positively with increasing molar melt Al/(Na + K + Ca) and polymerization (i.e. increasing ASI), and the addition of F clearly increases the lnDMgOamph/melt value (Fig. 10a). Furthermore, across a broad range in lnDamph/melt values the partitioning of FeO and TiO2 also correlates positively (Fig. 10b), with the addition of F generally decreasing the lnDFeOamph/melt values, which is consistent with the F–Ti trend described above (Fig. 9b). Table 4: Comparison of our data with published partition coefficients from other experimental investigations Data source: . This study . Sato et al. (2005) . Matjuschkin et al. (2016) . T range (°C): . 810–860 . 800–850 . 850–950 . P range (kbar): . 1·5–4·05 . 2·0–3·0 . 10·0–15·0 . fO2 range: . NNO –0·5 to NNO +2·0 . ≈NNO . Co–CoO to M–H . Glass ASI range: . 0·89–1·08 . 1·03–1·14 . 0·93–1·07 . Phase: Amph Amph Amph Max. Min. Average Max. Min. Average Max. Min. Average DSiO2min/melt 0·72 0·64 0·69 0·73 0·65 0·70 0·68 0·58 0·63 DTiO2min/melt 9·55 2·74 5·35 9·81 3·29 5·57 13·62 2·51 5·59 DAl2O3min/melt 0·80 0·56 0·68 0·74 0·57 0·65 0·89 0·69 0·78 DFeOmin/melt 9·08 4·60 6·87 10·28 4·48 5·91 9·44 2·66 5·81 DMnOmin/melt 5·59 2·68 3·73 7·00 2·63 3·81 no data DMgOmin/melt 51·21 23·19 33·93 50·55 18·44 25·08 26·17 7·35 15·69 DCaOmin/melt 5·87 2·85 3·77 6·01 2·91 3·72 4·44 2·15 3·03 DNa2Omin/melt 0·51 0·33 0·40 0·67 0·45 0·55 0·55 0·51 0·53 DK2Omin/melt 0·24 0·09 0·15 0·23 0·12 0·16 0·46 0·25 0·34 Data source: Nandedkar et al. (2016) This study Costa et al. (2004) T range (°C): 830–890 810–860 850–900 P range (kbar): 7·0 1·5–4·05 2·0–2·08 fO2 range: ≥NNO NNO –0·5 to NNO +2·0 NNO +0·3 to NNO +3·5 Glass ASI range: 1·10–1·20 0·89–1·08 0·98–1·34 Phase: Amph Plag Plag Max. Min. Average Max. Min. Average. Max. Min. Average DSiO2min/melt 0·74 0·70 0·72 0·88 0·79 0·83 0·81 0·74 0·77 DTiO2min/melt 9·31 4·76 6·67 no data no data DAl2O3min/melt 0·75 0·66 0·71 2·25 1·84 1·98 1·78 1·56 1·69 DFeOmin/melt 9·93 5·41 6·97 0·56 0·12 0·35 0·95 0·22 0·54 DMnOmin/melt 4·88 3·00 3·94 no data no data DMgOmin/melt 20·15 11·81 15·95 0·08 0·04 0·06 0·38 0·06 0·14 DCaOmin/melt 3·60 3·01 3·21 4·79 2·35 3·35 5·56 2·46 4·21 DNa2Omin/melt 0·57 0·51 0·54 1·43 1·10 1·30 1·76 0·89 1·26 DK2Omin/melt 0·19 0·15 0·17 0·10 0·06 0·08 0·18 0·07 0·12 Data source: . This study . Sato et al. (2005) . Matjuschkin et al. (2016) . T range (°C): . 810–860 . 800–850 . 850–950 . P range (kbar): . 1·5–4·05 . 2·0–3·0 . 10·0–15·0 . fO2 range: . NNO –0·5 to NNO +2·0 . ≈NNO . Co–CoO to M–H . Glass ASI range: . 0·89–1·08 . 1·03–1·14 . 0·93–1·07 . Phase: Amph Amph Amph Max. Min. Average Max. Min. Average Max. Min. Average DSiO2min/melt 0·72 0·64 0·69 0·73 0·65 0·70 0·68 0·58 0·63 DTiO2min/melt 9·55 2·74 5·35 9·81 3·29 5·57 13·62 2·51 5·59 DAl2O3min/melt 0·80 0·56 0·68 0·74 0·57 0·65 0·89 0·69 0·78 DFeOmin/melt 9·08 4·60 6·87 10·28 4·48 5·91 9·44 2·66 5·81 DMnOmin/melt 5·59 2·68 3·73 7·00 2·63 3·81 no data DMgOmin/melt 51·21 23·19 33·93 50·55 18·44 25·08 26·17 7·35 15·69 DCaOmin/melt 5·87 2·85 3·77 6·01 2·91 3·72 4·44 2·15 3·03 DNa2Omin/melt 0·51 0·33 0·40 0·67 0·45 0·55 0·55 0·51 0·53 DK2Omin/melt 0·24 0·09 0·15 0·23 0·12 0·16 0·46 0·25 0·34 Data source: Nandedkar et al. (2016) This study Costa et al. (2004) T range (°C): 830–890 810–860 850–900 P range (kbar): 7·0 1·5–4·05 2·0–2·08 fO2 range: ≥NNO NNO –0·5 to NNO +2·0 NNO +0·3 to NNO +3·5 Glass ASI range: 1·10–1·20 0·89–1·08 0·98–1·34 Phase: Amph Plag Plag Max. Min. Average Max. Min. Average. Max. Min. Average DSiO2min/melt 0·74 0·70 0·72 0·88 0·79 0·83 0·81 0·74 0·77 DTiO2min/melt 9·31 4·76 6·67 no data no data DAl2O3min/melt 0·75 0·66 0·71 2·25 1·84 1·98 1·78 1·56 1·69 DFeOmin/melt 9·93 5·41 6·97 0·56 0·12 0·35 0·95 0·22 0·54 DMnOmin/melt 4·88 3·00 3·94 no data no data DMgOmin/melt 20·15 11·81 15·95 0·08 0·04 0·06 0·38 0·06 0·14 DCaOmin/melt 3·60 3·01 3·21 4·79 2·35 3·35 5·56 2·46 4·21 DNa2Omin/melt 0·57 0·51 0·54 1·43 1·10 1·30 1·76 0·89 1·26 DK2Omin/melt 0·19 0·15 0·17 0·10 0·06 0·08 0·18 0·07 0·12 Data source: . Holtz et al. (2005) . Matjuschkin et al. (2016) . This study . T range (°C): . 775–875 . 850–950 . 835–860 . P range (kbar): . 2·0–3·0 . 10·0–15·0 . 1·5–4·05 . fO2 range: . ≈NNO . NNO to M–H . NNO –0·5 to NNO +2·0 . Glass ASI range: . 0·99–1·31 . 0·98–1·07 . 0·93–1·01 . Phase: Plag Plag Cpx Max. Min. Average Max. Min. Average Max. Min. Average DSiO2min/melt 0·87 0·80 0·83 0·91 0·87 0·90 0·78 0·71 0·75 DTiO2min/melt no data 0·14 0·06 0·10 5·87 1·27 3·53 DAl2O3min/melt 2·14 1·50 1·93 1·46 1·35 1·39 0·49 0·16 0·33 DFeOmin/melt 0·95 0·22 0·63 0·27 0·12 0·17 7·03 3·45 4·79 DMnOmin/melt no data no data 6·03 4·02 4·92 DMgOmin/melt 2·27 0·23 0·82 0·08 0·05 0·06 37·43 16·98 24·71 DCaOmin/melt 5·38 3·98 4·45 3·05 1·93 2·33 7·39 5·20 6·44 DNa2Omin/melt 2·06 0·76 1·52 1·49 1·12 1·28 0·15 0·06 0·11 DK2Omin/melt 0·22 0·11 0·17 0·19 0·12 0·15 no data Data source: Prouteau & Scaillet (2003) Andújar et al. (2016) T range (°C): 890–995 810–860 P range (kbar): 4·0–9·8 1·0–4·0 fO2 range: NNO +2·2 to NNO +3·4 NNO –0·8 to NNO +1·63 Glass ASI range: 0·93–1·22 0·81–0·98 Phase: Cpx Cpx Max. Min. Average Max. Min. Average DSiO2min/melt 0·74 0·67 0·70 0·80 0·70 0·75 DTiO2min/melt 3·96 0·55 2·10 1·58 0·65 1·02 DAl2O3min/melt 0·50 0·23 0·39 0·40 0·15 0·24 DFeOmin/melt 8·43 3·26 5·95 5·47 1·85 3·22 DMnOmin/melt 8·43 0·00 3·63 13·43 1·18 6·21 DMgOmin/melt 24·29 8·74 16·87 22·98 9·82 15·98 DCaOmin/melt 8·76 4·93 6·27 7·30 2·72 5·16 DNa2Omin/melt 0·33 0·16 0·21 0·30 0·00 0·09 DK2Omin/melt 0·09 0·01 0·03 0·07 0·00 0·03 Data source: . Holtz et al. (2005) . Matjuschkin et al. (2016) . This study . T range (°C): . 775–875 . 850–950 . 835–860 . P range (kbar): . 2·0–3·0 . 10·0–15·0 . 1·5–4·05 . fO2 range: . ≈NNO . NNO to M–H . NNO –0·5 to NNO +2·0 . Glass ASI range: . 0·99–1·31 . 0·98–1·07 . 0·93–1·01 . Phase: Plag Plag Cpx Max. Min. Average Max. Min. Average Max. Min. Average DSiO2min/melt 0·87 0·80 0·83 0·91 0·87 0·90 0·78 0·71 0·75 DTiO2min/melt no data 0·14 0·06 0·10 5·87 1·27 3·53 DAl2O3min/melt 2·14 1·50 1·93 1·46 1·35 1·39 0·49 0·16 0·33 DFeOmin/melt 0·95 0·22 0·63 0·27 0·12 0·17 7·03 3·45 4·79 DMnOmin/melt no data no data 6·03 4·02 4·92 DMgOmin/melt 2·27 0·23 0·82 0·08 0·05 0·06 37·43 16·98 24·71 DCaOmin/melt 5·38 3·98 4·45 3·05 1·93 2·33 7·39 5·20 6·44 DNa2Omin/melt 2·06 0·76 1·52 1·49 1·12 1·28 0·15 0·06 0·11 DK2Omin/melt 0·22 0·11 0·17 0·19 0·12 0·15 no data Data source: Prouteau & Scaillet (2003) Andújar et al. (2016) T range (°C): 890–995 810–860 P range (kbar): 4·0–9·8 1·0–4·0 fO2 range: NNO +2·2 to NNO +3·4 NNO –0·8 to NNO +1·63 Glass ASI range: 0·93–1·22 0·81–0·98 Phase: Cpx Cpx Max. Min. Average Max. Min. Average DSiO2min/melt 0·74 0·67 0·70 0·80 0·70 0·75 DTiO2min/melt 3·96 0·55 2·10 1·58 0·65 1·02 DAl2O3min/melt 0·50 0·23 0·39 0·40 0·15 0·24 DFeOmin/melt 8·43 3·26 5·95 5·47 1·85 3·22 DMnOmin/melt 8·43 0·00 3·63 13·43 1·18 6·21 DMgOmin/melt 24·29 8·74 16·87 22·98 9·82 15·98 DCaOmin/melt 8·76 4·93 6·27 7·30 2·72 5·16 DNa2Omin/melt 0·33 0·16 0·21 0·30 0·00 0·09 DK2Omin/melt 0·09 0·01 0·03 0·07 0·00 0·03 Open in new tab Table 4: Comparison of our data with published partition coefficients from other experimental investigations Data source: . This study . Sato et al. (2005) . Matjuschkin et al. (2016) . T range (°C): . 810–860 . 800–850 . 850–950 . P range (kbar): . 1·5–4·05 . 2·0–3·0 . 10·0–15·0 . fO2 range: . NNO –0·5 to NNO +2·0 . ≈NNO . Co–CoO to M–H . Glass ASI range: . 0·89–1·08 . 1·03–1·14 . 0·93–1·07 . Phase: Amph Amph Amph Max. Min. Average Max. Min. Average Max. Min. Average DSiO2min/melt 0·72 0·64 0·69 0·73 0·65 0·70 0·68 0·58 0·63 DTiO2min/melt 9·55 2·74 5·35 9·81 3·29 5·57 13·62 2·51 5·59 DAl2O3min/melt 0·80 0·56 0·68 0·74 0·57 0·65 0·89 0·69 0·78 DFeOmin/melt 9·08 4·60 6·87 10·28 4·48 5·91 9·44 2·66 5·81 DMnOmin/melt 5·59 2·68 3·73 7·00 2·63 3·81 no data DMgOmin/melt 51·21 23·19 33·93 50·55 18·44 25·08 26·17 7·35 15·69 DCaOmin/melt 5·87 2·85 3·77 6·01 2·91 3·72 4·44 2·15 3·03 DNa2Omin/melt 0·51 0·33 0·40 0·67 0·45 0·55 0·55 0·51 0·53 DK2Omin/melt 0·24 0·09 0·15 0·23 0·12 0·16 0·46 0·25 0·34 Data source: Nandedkar et al. (2016) This study Costa et al. (2004) T range (°C): 830–890 810–860 850–900 P range (kbar): 7·0 1·5–4·05 2·0–2·08 fO2 range: ≥NNO NNO –0·5 to NNO +2·0 NNO +0·3 to NNO +3·5 Glass ASI range: 1·10–1·20 0·89–1·08 0·98–1·34 Phase: Amph Plag Plag Max. Min. Average Max. Min. Average. Max. Min. Average DSiO2min/melt 0·74 0·70 0·72 0·88 0·79 0·83 0·81 0·74 0·77 DTiO2min/melt 9·31 4·76 6·67 no data no data DAl2O3min/melt 0·75 0·66 0·71 2·25 1·84 1·98 1·78 1·56 1·69 DFeOmin/melt 9·93 5·41 6·97 0·56 0·12 0·35 0·95 0·22 0·54 DMnOmin/melt 4·88 3·00 3·94 no data no data DMgOmin/melt 20·15 11·81 15·95 0·08 0·04 0·06 0·38 0·06 0·14 DCaOmin/melt 3·60 3·01 3·21 4·79 2·35 3·35 5·56 2·46 4·21 DNa2Omin/melt 0·57 0·51 0·54 1·43 1·10 1·30 1·76 0·89 1·26 DK2Omin/melt 0·19 0·15 0·17 0·10 0·06 0·08 0·18 0·07 0·12 Data source: . This study . Sato et al. (2005) . Matjuschkin et al. (2016) . T range (°C): . 810–860 . 800–850 . 850–950 . P range (kbar): . 1·5–4·05 . 2·0–3·0 . 10·0–15·0 . fO2 range: . NNO –0·5 to NNO +2·0 . ≈NNO . Co–CoO to M–H . Glass ASI range: . 0·89–1·08 . 1·03–1·14 . 0·93–1·07 . Phase: Amph Amph Amph Max. Min. Average Max. Min. Average Max. Min. Average DSiO2min/melt 0·72 0·64 0·69 0·73 0·65 0·70 0·68 0·58 0·63 DTiO2min/melt 9·55 2·74 5·35 9·81 3·29 5·57 13·62 2·51 5·59 DAl2O3min/melt 0·80 0·56 0·68 0·74 0·57 0·65 0·89 0·69 0·78 DFeOmin/melt 9·08 4·60 6·87 10·28 4·48 5·91 9·44 2·66 5·81 DMnOmin/melt 5·59 2·68 3·73 7·00 2·63 3·81 no data DMgOmin/melt 51·21 23·19 33·93 50·55 18·44 25·08 26·17 7·35 15·69 DCaOmin/melt 5·87 2·85 3·77 6·01 2·91 3·72 4·44 2·15 3·03 DNa2Omin/melt 0·51 0·33 0·40 0·67 0·45 0·55 0·55 0·51 0·53 DK2Omin/melt 0·24 0·09 0·15 0·23 0·12 0·16 0·46 0·25 0·34 Data source: Nandedkar et al. (2016) This study Costa et al. (2004) T range (°C): 830–890 810–860 850–900 P range (kbar): 7·0 1·5–4·05 2·0–2·08 fO2 range: ≥NNO NNO –0·5 to NNO +2·0 NNO +0·3 to NNO +3·5 Glass ASI range: 1·10–1·20 0·89–1·08 0·98–1·34 Phase: Amph Plag Plag Max. Min. Average Max. Min. Average. Max. Min. Average DSiO2min/melt 0·74 0·70 0·72 0·88 0·79 0·83 0·81 0·74 0·77 DTiO2min/melt 9·31 4·76 6·67 no data no data DAl2O3min/melt 0·75 0·66 0·71 2·25 1·84 1·98 1·78 1·56 1·69 DFeOmin/melt 9·93 5·41 6·97 0·56 0·12 0·35 0·95 0·22 0·54 DMnOmin/melt 4·88 3·00 3·94 no data no data DMgOmin/melt 20·15 11·81 15·95 0·08 0·04 0·06 0·38 0·06 0·14 DCaOmin/melt 3·60 3·01 3·21 4·79 2·35 3·35 5·56 2·46 4·21 DNa2Omin/melt 0·57 0·51 0·54 1·43 1·10 1·30 1·76 0·89 1·26 DK2Omin/melt 0·19 0·15 0·17 0·10 0·06 0·08 0·18 0·07 0·12 Data source: . Holtz et al. (2005) . Matjuschkin et al. (2016) . This study . T range (°C): . 775–875 . 850–950 . 835–860 . P range (kbar): . 2·0–3·0 . 10·0–15·0 . 1·5–4·05 . fO2 range: . ≈NNO . NNO to M–H . NNO –0·5 to NNO +2·0 . Glass ASI range: . 0·99–1·31 . 0·98–1·07 . 0·93–1·01 . Phase: Plag Plag Cpx Max. Min. Average Max. Min. Average Max. Min. Average DSiO2min/melt 0·87 0·80 0·83 0·91 0·87 0·90 0·78 0·71 0·75 DTiO2min/melt no data 0·14 0·06 0·10 5·87 1·27 3·53 DAl2O3min/melt 2·14 1·50 1·93 1·46 1·35 1·39 0·49 0·16 0·33 DFeOmin/melt 0·95 0·22 0·63 0·27 0·12 0·17 7·03 3·45 4·79 DMnOmin/melt no data no data 6·03 4·02 4·92 DMgOmin/melt 2·27 0·23 0·82 0·08 0·05 0·06 37·43 16·98 24·71 DCaOmin/melt 5·38 3·98 4·45 3·05 1·93 2·33 7·39 5·20 6·44 DNa2Omin/melt 2·06 0·76 1·52 1·49 1·12 1·28 0·15 0·06 0·11 DK2Omin/melt 0·22 0·11 0·17 0·19 0·12 0·15 no data Data source: Prouteau & Scaillet (2003) Andújar et al. (2016) T range (°C): 890–995 810–860 P range (kbar): 4·0–9·8 1·0–4·0 fO2 range: NNO +2·2 to NNO +3·4 NNO –0·8 to NNO +1·63 Glass ASI range: 0·93–1·22 0·81–0·98 Phase: Cpx Cpx Max. Min. Average Max. Min. Average DSiO2min/melt 0·74 0·67 0·70 0·80 0·70 0·75 DTiO2min/melt 3·96 0·55 2·10 1·58 0·65 1·02 DAl2O3min/melt 0·50 0·23 0·39 0·40 0·15 0·24 DFeOmin/melt 8·43 3·26 5·95 5·47 1·85 3·22 DMnOmin/melt 8·43 0·00 3·63 13·43 1·18 6·21 DMgOmin/melt 24·29 8·74 16·87 22·98 9·82 15·98 DCaOmin/melt 8·76 4·93 6·27 7·30 2·72 5·16 DNa2Omin/melt 0·33 0·16 0·21 0·30 0·00 0·09 DK2Omin/melt 0·09 0·01 0·03 0·07 0·00 0·03 Data source: . Holtz et al. (2005) . Matjuschkin et al. (2016) . This study . T range (°C): . 775–875 . 850–950 . 835–860 . P range (kbar): . 2·0–3·0 . 10·0–15·0 . 1·5–4·05 . fO2 range: . ≈NNO . NNO to M–H . NNO –0·5 to NNO +2·0 . Glass ASI range: . 0·99–1·31 . 0·98–1·07 . 0·93–1·01 . Phase: Plag Plag Cpx Max. Min. Average Max. Min. Average Max. Min. Average DSiO2min/melt 0·87 0·80 0·83 0·91 0·87 0·90 0·78 0·71 0·75 DTiO2min/melt no data 0·14 0·06 0·10 5·87 1·27 3·53 DAl2O3min/melt 2·14 1·50 1·93 1·46 1·35 1·39 0·49 0·16 0·33 DFeOmin/melt 0·95 0·22 0·63 0·27 0·12 0·17 7·03 3·45 4·79 DMnOmin/melt no data no data 6·03 4·02 4·92 DMgOmin/melt 2·27 0·23 0·82 0·08 0·05 0·06 37·43 16·98 24·71 DCaOmin/melt 5·38 3·98 4·45 3·05 1·93 2·33 7·39 5·20 6·44 DNa2Omin/melt 2·06 0·76 1·52 1·49 1·12 1·28 0·15 0·06 0·11 DK2Omin/melt 0·22 0·11 0·17 0·19 0·12 0·15 no data Data source: Prouteau & Scaillet (2003) Andújar et al. (2016) T range (°C): 890–995 810–860 P range (kbar): 4·0–9·8 1·0–4·0 fO2 range: NNO +2·2 to NNO +3·4 NNO –0·8 to NNO +1·63 Glass ASI range: 0·93–1·22 0·81–0·98 Phase: Cpx Cpx Max. Min. Average Max. Min. Average DSiO2min/melt 0·74 0·67 0·70 0·80 0·70 0·75 DTiO2min/melt 3·96 0·55 2·10 1·58 0·65 1·02 DAl2O3min/melt 0·50 0·23 0·39 0·40 0·15 0·24 DFeOmin/melt 8·43 3·26 5·95 5·47 1·85 3·22 DMnOmin/melt 8·43 0·00 3·63 13·43 1·18 6·21 DMgOmin/melt 24·29 8·74 16·87 22·98 9·82 15·98 DCaOmin/melt 8·76 4·93 6·27 7·30 2·72 5·16 DNa2Omin/melt 0·33 0·16 0·21 0·30 0·00 0·09 DK2Omin/melt 0·09 0·01 0·03 0·07 0·00 0·03 Open in new tab Fig. 10. Open in new tabDownload slide Plots of average lnDmineral/melt partitioning variations, and comparison with relevant literature values. The strong effect of F on amphibole compositions, and the good correlations between our new data and previous experimental studies on similar melt compositions, should be noted. Amphibole: halogens Whereas relatively F-rich amphibole and melts (i.e. >0·5 wt % F) are commonly reported for evolved erupted rocks, and their plutonic equivalents often associated with magmatic–hydrothermal mineralization and metasomatic alteration (e.g. Balcone-Boissard et al., 2010; Zhang et al., 2012; Chambefort et al., 2013; Iveson et al., 2016), there are few, if any, experimental partitioning data for the incorporation of F into calcic amphibole in equilibrium with dacitic melts, although data for Cl are more abundant. Some recent work by Van den Bleeken & Koga (2015) obtained a small range in DFamph/melt values of 1·18–1·85 for magnesiohastingsites of a similar major element composition to the ones in this study. Their experiments were conducted at 750–900°C, >15 kbar, and fO2 ≈ NNO, in ‘altered oceanic crust’ melts roughly equivalent to trachyandesite compositions (with ASI = 0·91) containing ∼0·5 wt % F. Those researchers conducted a multi-component regression to fit their DFamph/melt values, with some success. Application of their equation to our data, however, generates erroneously low DFamph/melt values, by almost two orders of magnitude. Van den Bleeken & Koga (2015) also reported DClamph/melt values from 0·079 to 0·625, with their values for the most oxidizing experiments (rhenium–rhenium oxide) averaging ∼0·14, which are broadly consistent with our amphibole–melt Cl partition coefficients. Furthermore, consistent with the observations in our study, Scaillet & Macdonald (2003) reported that the partitioning of F between amphibole and comenditic melt at ∼700°C, 1·5 kbar, and reducing conditions of fO2 ≈ NNO –0·1 to –3·2 decreased from ∼2·7 to ∼1 as the F content of the amphibole increased from 2 to >3 wt %. The amphiboles in that study were significantly different in composition (i.e. sodic and Fe-rich ferro-richterite end-members) but the amphibole F contents increased almost linearly over the range in melt F contents investigated. Further recent data from Bénard et al. (2017) detail F and Cl partitioning between amphibole, interstitial glasses, and homogenized melt inclusions in andesite-hosted sub-arc lithospheric mantle xenoliths. Bénard et al. (2017) showed DFamph/melt values of 3·5–3·7 (±1·5) for calcic amphibole, similar to the lower values in our study, but lower DClamph/melt values of 0·03–0·05 (±0·02) as a result of the very Mg-rich amphibole with Mg# > 0·90. In contrast to the moderately to strongly compatible behaviour of F in amphibole documented in this study and the aforementioned ones, Dalou (2011) reported DFamph/melt values of 0·36–0·63 for fluor-pargasitic hornblende at 1200°C, 12 kbar, and fO2 ≈ FMQ +1. However, the DClamph/melt values of 0·12–0·38 from Dalou (2011) are in much better accordance with the DClamph/melt values shown by the amphibole in these rhyodacite experiments. Adam et al. (1993) also crystallized amphibole with >2·5 wt % F in basaltic andesite and basanite melts at 925–1050°C, 10–20 kbar, with up to 5 wt % F added to the starting charges. They reported DFamph/melt values ranging from 0·43 to 1·0, again suggesting moderately incompatible behaviour for F in those highly depolymerized melts. The data from Cl partitioning experiments of Sato et al. (2005) and the recent DClamph/melt data from Matjuschkin et al. (2016) show an excellent correlation with our data; DClamph/melt decreases with increasing Mg/(Mg + FeTOT) values (Fig. 8a). The data from both Sato et al. (2005) and Matjuschkin et al. (2016) extend to lower amphibole Mg/(Mg + FeTOT) values than ours, and tentative extrapolation of the trend suggests that amphibole–melt Cl partition coefficients could approach unity in Fe-rich amphibole grown at very low fO2 conditions in rhyodacitic melts. This observation has important implications for the Cl contents of the magma available to partition into fluid phase(s) during subsequent fluid exsolution (e.g. Webster, 1997, 2004). The negative correlation between lnDClamph/melt and ln DSiO2amph/melt in our amphibole (Fig. 8b) and those in literature studies is consistent with previous structural analyses of amphibole (e.g. Oberti et al., 1993), which showed that the expansion of the octahedral sheet owing to Cl incorporation is accommodated by [4]Al substitution for Si. This accommodation is in addition to other mechanisms, such as Fe2+ substitution for Mg, and hence the trend for increasing Mg# versus lower Cl partitioning (Fig. 8a). In general, the greater compatibility of F over Cl is likely to largely be attributable to its smaller ionic radius (OH– = 0·132–0·137 nm; F– = 0·133 nm; Cl– = 0·181 nm), requiring greater deformation of neighbouring lattice sites to accommodate Cl relative to F. This explains why the experimental amphiboles doped with both halogens in our study show lower DClamph/melt values than would be expected at a fixed Mg#, as a result of increased F competition. The decrease in DClamph/melt with increasing melt F (and hence amphibole F) contents may also be a result of increased Cl solubility in aluminosilicate melts owing to increasing melt F, as discussed by Webster (1997). The well-defined trends observed between increasing absolute amphibole F contents and decreasing amphibole Fe and Ti contents (Figs 9a, b and 10b) illustrate the complex crystallographic substitution and partitioning relationships that exist between major and trace components in the amphibole structure. With maximum F concentrations of >2·5 wt % F in some of these experimental amphibole, F is expected to no longer behave as a simple trace component, and further work is required to accurately model its strongly compatible incorporation in calcic amphiboles crystallizing from hydrous evolved liquids, such as those found in arc settings. The other discrepancies between our new partition coefficients for both halogens and those reported from other experimental and natural samples suggest that F and Cl incorporation is likely to be a function of the prevailing fO2 conditions, its subsequent effect on Mg/(Mg + FeTOT) ratio of the amphibole, and absolute melt halogen contents. However, given the consistency in the lnDClamph/melt values across the three datasets (Fig. 8a), it is likely that temperature and pressure have a minor effect, although the Mg# of amphibole has been shown to increase with decreasing temperature at 7 kbar (e.g. Nandedkar et al., 2016). Because we added CaF2 as the F source to these experiments, and owing to the scarcity of other relevant data, the effect of melt polymerization and melt alkali–alumina ratio on F partitioning also cannot be separated from other crystallographic controls on the amphibole composition. Whereas Cl has relatively consistent partition coefficients across a large range of experimental conditions for a given amphibole composition, F is much more variable but is generally strongly compatible, despite other evidence discussed here to suggest that F shows some incompatible behaviour in certain amphibole–melt systems. Another important observation to note is the apparently stabilizing effect of F on amphibole in these experiments. The absence of amphibole in run products of two low-pressure experiments (at 1·5 kbar), combined with the presence of amphibole in another 1·5 kbar run where F was added, suggests that amphibole may not have otherwise been stable at those temperature and pressure conditions. This potential stabilization has been described elsewhere; for example, by Rutherford (2008) and Rutherford & Devine (2008) in Mt St Helens amphibole from the 2004–2006 dacite, where shallow crystallization outside the OH-bearing amphibole stability field was facilitated by high local F contents during ascent and degassing in the conduit. Outer rims of some of these amphibole contained >7000 ppm F, and the researchers speculated that slow magma ascent led to increases in F and SiO2 activity in the melt, and showed that F- and Si-rich amphibole could be stabilized at pressures as low as 50 MPa in the Mt St Helens dacitic magma at 860°C. Therefore, although no data exist for melt F contents to allow calculation of the apparent DFamph/melt values for the F-enriched rims, any variation in the partition coefficient outside the equilibrium values expected from the major element controls shown here is probably due to rapid disequilibrium crystallization during eruption. Plagioclase and clinopyroxene: major elements Plagioclase–melt partitioning data from Costa et al. (2004), Holtz et al. (2005), and Matjuschkin et al. (2016) allow for useful comparisons with our data across a similar range in conditions and melt compositions (Table 4). The data from Costa et al. (2004) and Holtz et al. (2005) have been filtered to include only glasses where measured or calculated final water concentrations are ≥6·0 wt %, consistent with the water-saturated melts in our experiments. Generally, the plagioclase grown in the trachydacite melts of Matjuschkin et al. (2016) is more sodic and exhibits lower average DAl2O3plag/melt and DCaOplag/melt ⁠, and higher DSiO2plag/melt ⁠, but there is good agreement for the other oxide components between all the datasets. Specifically, our data for DMgOplag/melt are identical to those data of Matjuschkin et al. (2016), suggesting that this difference in melt composition has a limited effect on this behaviour. The average XAn ∼51 ± 8 in the experimental plagioclase of Costa et al. (2004) is also very similar to the average of XAn ∼47 ± 4 in our experiments. It is noteworthy, therefore, that the addition of up to 1 wt % S to the starting fluids in the experimental charges has little impact on the XAn in our runs. Costa et al. (2004) showed an increase in the DFeOplag/melt value from 0·31 ± 0·06 to 0·69 ± 0·16 at increasing fO2 from ≈NNO +0·5 to ≥NNO +3, and DFeOplag/melt values from Matjuschkin et al. (2016) increase from 0·13 ± 0·02 to 0·24 ± 0·03 with increasing fO2 from ∼NNO to magnetite–haematite. These two datasets are consistent with the plagioclase data in this study (Fig. 4b). This common trend of increasing DFeOplag/melt with increasing experimental fO2 suggests that Fe3+ is the dominant species of Fe in plagioclase. Data from Lundgaard & Tegner (2004) support this hypothesis, and showed the partitioning of FeO and Fe2O3 into plagioclase across a range of magma chemistries to be independent of both fO2 and the plagioclase composition. They also showed that DFeOplag/melt is not a function of the amount of Al in plagioclase, and that Fe3+ incorporation was ∼20 times greater than that of Fe2+. Lundgaard & Tegner (2004) found that the SiO2 content of the magma exerts a first-order control on FeO and Fe2O3 partitioning between plagioclase and melt, and this trend is well developed across the range in melt compositions for the data in Fig. 10c. It can thus be summarized that the decrease in DFeOplag/melt with decreasing fO2 seen across all the experimental datasets probably reflects the change in the overall proportions of FeO and Fe2O3 in the melt, as opposed to changes in the absolute partition coefficient itself, as well as the influence of first-order controls by melt SiO2 concentrations. Two experimental datasets reported by Prouteau & Scaillet (2003) and Andújar et al. (2016) contain clinopyroxene–melt partitioning data for rhyodacitic melts from 890 to 1000°C, from 1·0 to 9·8 kbar, and fO2 from NNO –1 to +3·4 (Table 4). A strong positive correlation between DAl2O3cpx/melt and DSiO2cpx/melt can be seen in the three datasets (Fig. 10d), and is consistent with the coupled substitution of TiAl2 ⇄ MgSi2 (e.g. Fig. 7), which is generally favoured at higher experimental pressures and higher fO2. The generally more reducing conditions of the Andújar et al. (2016) experiments (fO2 < NNO +1·6) give average clinopyroxene compositions with a greater ferrosilite component and lower Mg# (XFs ≈ 22, Mg# ≈ 65), compared with the more oxidizing conditions of Prouteau & Scaillet (2003) (fO2 > NNO +2·2) and this study (XFs ≈ 13, Mg# ≈ 76; and XFs ≈ 16, Mg# ≈ 70, respectively). However, all of the three experimental sets of clinopyroxene crystals are calcic, with a large wollastonite component, and calculation of Dcpx/melt values for the oxide components shows that they broadly agree, when accounting for differing melt compositions and temperature, pressure, and fO2 conditions. Fluid–melt: halogens, sulfur, and carbon dioxide With the exception of five experiments, all capsules contained fluid:melt ratios of <10% by mass, and thus the total mass of coexisting volatile phase at the individual run conditions was probably small. Water concentrations in the quenched glasses, by difference, are broadly consistent with solubility models for crystal-free dacitic compositions (e.g. Botcharnikov et al., 2005, and references therein), showing ∼6 wt % solubility at 2·0 kbar and 1250°C. The CO2 concentration measured in the CO2-doped experiment (1-16-23A) by FTIR is largely consistent with the solubility model of Newman & Lowenstern (2002) for rhyolitic melts in equilibrium with mixed H2O–CO2 fluids. At run conditions, the model predicts a CO2 solubility of 341 ppm in the melt, with 11·2 mol % CO2 present in the volatile phase, generally in agreement with the starting fluid composition of approximately 16 mol % CO2. Using the model of Webster et al. (2015) for this Usu rhyodacite composition when hydrated with 6 wt % H2O at 850°C and at 2·5 kbar, the theoretical maximum Cl concentration in the crystal-free melt is ∼1·20 wt % when coexisting with a hydrosaline liquid ± vapour. Based on the measured Cl contents of the glasses, these melts have not reached the modelled maximum Cl concentration for the measured melt compositions. Given the relatively high concentrations of Cl in the starting fluids (up to ∼14 wt %) (Table 2), the experimental temperature–pressure conditions, and assuming that the fluid phase relations can be approximated by those of the NaCl–H2O system, it is likely that these crystallizing melts coexisted with a single-phase Cl-rich fluid at supersolvus conditions, as the critical point in the H2O–NaCl system occurs between ∼1·0 and 1·4 kbar at 600–800°C (Pitzer & Pabalan, 1986). Therefore, the calculated DClfluid/melt values (ranging from ∼4 to 22; Fig. 11) agree with previously published DClfluid/melt data for dacitic and rhyodacitic melts, in particular the results of Botcharnikov et al. (2004) and Webster et al. (2009). These studies reported DClfluid/melt ranging from ∼2 to 16 ± 6 for similar experimental conditions and bulk halogen contents. Although some scatter exists, our data suggest that in these three-phase experiments (i.e. melt + crystals + volatile phase), the partitioning of Cl into the volatile phase is partially dependent on the aluminosity of the melt (Fig. 11a). More evolved melts, with higher molar Al2O3/(CaO + Na2O + K2O) values, on average show the highest DClfluid/melt values. Superimposed on this trend is also the apparent effect of the prevailing fO2 conditions. Although we do not have quantitative constraints on the Fe2+/Fe3+ ratios of the experimental product glasses, those formed under the lowest fO2 conditions are expected to have higher Fe2+/Fe3+ ratios (e.g. Kilinc et al., 1983). Work by Bell & Webster (2015) suggested that a greater proportion of Cl could be accommodated in more reduced melts through complexation with the greater abundances of Fe2+, and hence DClfluid/melt values should decrease, as we broadly observe (Fig. 11a). Similarly, the increased melt Cl solubility with increasing melt F (Webster, 1997) is also consistent with the Cl- and F-doped experiment in our study showing the lowest calculated DClfluid/melt value (Fig. 11a), but separating this control from the effect of melt aluminosity is not possible from this dataset. The correlation between increasing fluid–melt Cl partitioning and concurrent decreases in amphibole–melt partitioning of Cl (Fig. 11b) illustrates that the aforementioned melt compositional controls on Cl solubility affect both volatile and crystal phases. The data suggest that under fluid-saturated conditions, Cl more strongly partitions into a coexisting volatile phase, rather than being incorporated into the crystallizing amphibole, when melt Cl solubility is reduced through changes in Fe2+ abundance (i.e. increasing fO2). Again, the effect of F on both DClfluid/melt and DClamph/melt values is apparent, leading to strong reductions in amphibole–melt partitioning, and modest reductions in fluid–melt partitioning, when the increased melt Cl solubility resulting from the lower fO2 conditions is also considered (Fig. 11b). Fig. 11. Open in new tabDownload slide Variations in the measured fluid–melt partitioning behaviour of Cl as a function of changing melt aluminosity (a), and the covariations between DClfluid/melt and DClamph/melt (b). Noteworthy features are the overall increase in Cl partitioning into the fluid as the melts become more alumina-rich, and the relatively decreased preference for the fluid at equivalent molar Al2O3/(CaO + Na2O + K2O) values for experiments with more Fe2+-rich melts [i.e. those crystallized at lower fO2 conditions (a)]. The combined effects of F competition on Cl incorporation into amphibole and the increased melt solubility of Cl lead to lower values for both DClfluid/melt and DClamph/melt in (b). Similar to the preference of Cl for the volatile phase relative to the rhyodacitic melts, in the four sulfur-doped experiments the behaviour of S is consistent with the recent work of Zajacz et al. (2013) and Masotta et al. (2016), who measured DSfluid/melt ≥100 between rhyodacite and aqueous fluid at oxidizing experimental compositions similar to those in this investigation. In our experiments, the majority of the total S budget partitioned strongly into the fluid phase given the low concentrations in the final glasses, and the absence of sulfide or sulfate minerals in the run products. Masotta et al. (2016) experimentally calibrated a model to predict DSfluid/melt for oxidized magmas at 2 kbar and 800–950°C, using relationships between melt structural parameters. Application of this model to our S-doped experiments yields very high predicted DSfluid/melt values of c. 50–190; this result is consistent with the measured low S concentrations in the final melts. However, the behaviour of S is complicated by its ability to be incorporated into the Au and AuCu capsule alloy (Botcharnikov et al., 2004), which may also partly explain the very low S contents (∼30 ppm SO3) in the glasses from two of the S-bearing experiments conducted in AuCu capsules. Therefore mass-balance constraints on DSfluid/melt for these experiments would not be robust, and neither would any attempt to extract information regarding changes in sulfur partitioning over the range of experimental fO2 values. Similarly, the effects of complex mixed volatile components on individual Dfluid/melt behaviours are not currently well known for these liquid compositions (e.g. Webster et al., 2014; Binder et al., 2018), and hence the effects of F, Cl, and CO2 on S partitioning are unknown. Other previous work (e.g. Webster & Holloway, 1990; Balcone-Boissard et al., 2010; Baker & Alletti, 2012; and references therein) has illustrated the tendency of F to remain in silicic melts relative to coexisting aqueous fluids. Webster & Holloway (1990) reported low DFfluid/melt values of ∼0·3 for rhyolitic melt and aqueous fluids at 2 kbar and ∼800°C where total F concentrations in the melt were less than ∼2 wt %. Thus, although F contents of the final fluids in our experiments were not directly measured and thus DFfluid/melt values were not calculated, it is also probable that the coexisting fluids contained little F relative to Cl and S at these conditions, especially owing to the crystallizing amphibole readily incorporating F, as shown here. Effect of halogens on melt structure and partitioning One potentially important melt structural parameter that must be considered when interpreting mineral–melt partition coefficients is the ratio of non-bridging oxygens to tetrahedrally coordinated cations (NBO/T ratio), which indicates the degree of melt polymerization (e.g. Mysen, 1998, 2004, and references therein). This ratio can be expressed as both an anhydrous and a hydrous parameter. In the case of these experimental glasses, given the significant concentrations of H2O dissolved in the melts, NBO/T is calculated on a hydrous basis, and assumes all hydrogen occurs as a network-modifying component, thereby depolymerizing the melt (i.e. increasing the value of NBO/T) relative to an anhydrous composition. Furthermore, Fe3+ has been shown to act as a network former whereas Fe2+ behaves as a network modifier in aluminosilicate melts (e.g. Giordano & Dingwell, 2003). However, given the relatively low total FeO concentrations, the change in the ferric:ferrous ratio as a function of fO2 has a relatively minor effect on the calculated NBO/T. Whereas H2O/molecular OH– acts to depolymerize the melts, there are more complex effects for the halogens and S. Importantly, F and Cl appear to have contrasting effects on the structure of aluminosilicate melts (Dalou & Mysen, 2015; Dalou et al., 2015) owing to the differing complexes formed by the halogens. In general, Cl predominantly dissolves via complexation with network-modifying alkaline earth metals (Ca or Mg), alkalis (Na or K), or Fe (Webster, 1997, and references therein), and at these concentration levels it is likely that the F occurs as Si–F and Al–F complexes. Thus, Cl should serve to polymerize the melt whereas F should have a depolymerizing effect. The data of Hellwig (2006) and Barber (2007) support this hypothesis, with Hellwig (2006) reporting a c. 1 log unit decrease in viscosity of dacites and rhyodacites at magmatic temperatures with addition of 0·35 wt % F (i.e. comparable with the concentrations in our glasses). Barber (2007) showed that addition of 0·03 wt % and 1·43 wt % F decreased the viscosity of rhyodacitic magmas by factors of ∼3 and ∼250, from 6·2 × 1010 Pa s in F-free melts, to 2·3 × 1010 and 2·3 × 108 Pa s, respectively. The complex behaviour of F and Cl is not accounted for in the NBO/T calculation, so in Cl-bearing experiments the hydrous NBO/T value should be taken as a maximum, whereas in F-bearing experiments it should be taken as a minimum. There are few data regarding the competing effects when both F and Cl are present in the melt, but the NBO/T for these run product glasses ranges from 0·72 ± 0·02 at the most depolymerized to 0·43 ± 0·02 at the most polymerized (Supplementary Data Electronic Appendix 2). This is a relatively restricted range, and in general this parameter appears to have little effect on the partitioning of major elements between the phenocryst and melt phases investigated here. In the case of S, Morizet et al. (2013) showed that the dissolution of S in a haplogranitic melt under oxidizing conditions did not produce appreciable changes in the melt structure, as a result of the low abundance of NBO in those melts. Thus, in our experimental glasses, although unaccounted for in the NBO/T calculation, it is likely that the effect of low S concentrations on the NBO/T values is also negligible. Whereas the NBO/T measure of melt polymerization does not seem to affect partitioning behaviour, the alumina saturation index (i.e. aluminosity, or ASI) does appear to have an influence on the phenocryst chemistry, and hence partitioning of major components (e.g. Fig. 10a). Other studies have shown strong positive correlations between increasing ASI values and increasing Dmineral/melt and Dfluid/melt coefficients for a variety of major and trace elements (e.g. Nandedkar et al., 2016). This is most often related to the tendency of melts with excess alumina, and hence those that are strongly polymerized (Webb et al. 2004; Mysen & Toplis, 2007), to reject highly charged cations, such as the rare earth and high field strength elements. The ability of a cation to bond to non-bridging oxygens to form a stable configuration results in lower Dmineral/melt values than in melts where only bridging oxygens are available. Comparison with barometers, thermometers, and hygrometers Three recent calibrations for a geobarometer and geothermometer using experimental amphibole–melt data (Putirka, 2016), and a geohygrometer using plagioclase–melt data (Waters & Lange, 2015), can be compared with the experimental data in this study. Figure 12 shows the results of the model calculations (using averages of crystal compositions from each experiment) compared with actual or measured values for pressure and temperature from Putirka (2016), and melt H2O concentrations from the plagioclase–melt model of Waters & Lange (2015) versus the melt H2O concentrations from FTIR and EPMA ‘by difference’ methods. Fig. 12. Open in new tabDownload slide Plots comparing the average experimental mineral–melt data with (a) the amphibole barometer of Putirka [2016; equation (7b)], (b) the amphibole geothermometer of Putirka [2016; equation (8)], and (c) the plagioclase hygrometer of Waters & Lange (2015). The error bars on the calculations are derived from the standard deviations of the electron microprobe analyses for the amphibole, plagioclase, and melt compositions. Errors on the absolute experimental pressure and temperature conditions are smaller than the symbols. In general, there is very good agreement between the calculated and actual experimental pressure values derived from DAlamph/melt partitioning and hydrous liquid compositional parameters (Fig. 12a), with most of the data falling close to the 1:1 line. However, amphiboles from the CO2-bearing experiments yield calculated pressures that are erroneously low. This discrepancy is probably due to the aforementioned effects of lower fO2 on FeO partitioning, and its influence on the (ferri-)tschermakite exchange in the amphibole, given the dependence of the calculated pressure on the DAlamph/melt. The pressure-independent calculation of crystallization temperatures (Fig. 12b) shows that the average amphibole compositions consistently overestimate experimental temperature by ∼35°C, with some amphibole showing more than a 60°C difference from the actual experimental temperature. However, in this case the CO2-doped experiments yield better agreement than they did in the pressure calculations. Given the partial dependence of the calculated temperature on both the Mg and Na contents of the amphibole, the lower Mg amphiboles from the lowest fO2 experiments therefore yield the lowest calculated temperatures. This observation leads to an important consideration that must be made when applying such geobarometers and geothermometers (which require accurate measurements of both melt and phenocryst composition) to natural systems—how do the Mg/Fe ratios, alkali contents, and subsequent stoichiometry of the amphibole (and melt) change as a function of increasing halogen contents, and in particular through addition of F (e.g. Fig. 10a)? The most F-rich experimental amphiboles are also the most Mg- and alkali-rich, and hence these crystals provide the highest calculated temperatures and give the largest discrepancies between actual experimental values. Therefore, although natural volcanic amphiboles are seldom as F- or Cl-rich as some of the experimental ones in this study, care must be taken when applying these barometers and thermometers (calibrated for F- and Cl-poor melts) to natural systems where independent constraints on fO2 and halogen concentrations are absent. Figure 12c shows that although there are some relatively large uncertainties on the water concentrations inferred from the EPMA ‘H2O by difference’ method (Devine et al., 1995), and generally smaller ones on the FTIR measurements, the Waters & Lange (2015) model still closely reproduces the H2O contents of the experimental glasses, based on the average An content of the plagioclase. Application of this model to the CO2-bearing experiment shows that this volatile species appears to have little effect on the plagioclase-derived melt H2O values, and similarly for plagioclase crystallizing in halogen-bearing experiments. Thus, unlike amphibole, given the insensitivity of plagioclase to variations in melt halogen contents demonstrated in our data, we suggest that this thermodynamic model can be applied to halogen-bearing hydrous melts without the need for rigorous constraints on melt F and Cl concentrations. In summary, integration of our experimental data with these recent models for constraining melt temperature, pressure, and H2O concentrations yields good agreement, and suggests that the models are useful tools for understanding natural magmatic systems. The agreement also shows that our experimental data are largely consistent with the datasets used to construct calibrations for these models. However, we suggest that although the application of such models to natural samples to reconstruct the petrogenetic histories preserved by the phenocryst chemistry still generates geologically reasonable pressure and temperature values, the errors on such calculated values could be further reduced by thorough geochemical analysis of the crystals, and subsequently a better understanding of how halogens affect the crystal–melt exchange of major elements in amphibole. Implications for rhyodacitic magma evolution Dacites and rhyolites found in arc settings often show evidence consistent with prolonged magma and crystal residence and evolution prior to explosive and/or effusive eruption (e.g. Streck et al., 2008). Significant complexity exists in the partitioning behaviour of fluid–melt–mineral systems containing multicomponent volatiles, in particular where elevated concentrations of both halogens are present. The integration of our new data with previous experimental partitioning investigations discussed here reveals that the compositions of amphibole in particular will be sensitive to punctuated inputs of new chemically variable magma batches. For example, the addition of a small amount of F to a crystallizing magma will cause a shift in the amphibole compositions towards more Mg- and alkali-rich, and Cl-poor chemistries. This change in amphibole partition coefficients and average crystal composition will then have implications for the composition of hydrothermal fluid(s) exsolved later in the history of the magma chamber. Similarly, suppression of significant amphibole fractionation should be reflected through increasing F abundances in the residual melt with continued differentiation. The consistent behaviour of Cl partitioning as a function of the Mg# number of the amphibole across a range of temperature, pressure, and melt compositions demonstrated here suggests that application to natural systems, in conjunction with previous work such as that by Sato et al. (2005) and Giesting & Filiberto (2014), can be a useful method to quantitatively reconstruct melt Cl contents. However, further work is required to constrain the systematic partitioning of F between amphibole and melt, given results demonstrating both compatibility and incompatibility in amphibole from a wide variety of melt compositions. Similarly, these new halogen partition coefficients may also be a useful mechanism for identification of antecrystic populations from earlier unpreserved magma batches, late-stage melt mixing, or recognition of the effects of fluid alteration (e.g. Iveson et al., 2016). Changes to the bulk fO2 of the magma chamber by mixing with more mafic and reduced magmas, volatile fluxing, or through auto-oxidation by preferential removal of Fe2+ in moderate-salinity Cl-bearing fluids (Bell & Simon, 2011), will also strongly affect amphibole, clinopyroxene, and plagioclase chemistry. The correlations between increasing melt FeO concentrations and increasing Cl concentrations (e.g. Fig. 1c), along with covariations in DClfluid/melt and DClamph/melt values (Fig. 11), suggest that Fe–Cl complexes are highly stable in these rhyodacitic melts. Therefore, the activities of Fe–O species in the melt are expected to be lowered, a mechanism potentially responsible for the suppression of Fe–Ti oxide crystallization seen here. High Cl and FeO contents are often documented in pantelleritic magmas (∼1 wt % Cl and >8 wt % FeO in matrix glasses; Lowenstern, 1994; Neave et al., 2012), and Bell & Webster (2015) discussed some of the implications of such Fe–Cl interactions in the melt on redox exchanges occurring in Cl-enriched arc magmas. Additionally, given the discordance seen in Fig. 1c between the experiments with lower total Cl contents and those with higher Cl contents, and the recognized high mobility of Fe in such fluids at these pressure and temperature conditions (e.g. Simon et al., 2004), it is unlikely that FeO exerts the only control on Cl solubility in these complex fluid–melt systems (e.g. Metrich & Rutherford, 1992). Our new data further confirm that external fO2 conditions, melt composition, and crystallographic controls all affect halogen incorporation into calcic amphibole. Therefore, progressive oxidation of a crystallizing magma will lead to higher amphibole Mg# and subsequent decreased Cl incorporation, and will also be accompanied by higher FeO contents in plagioclase. The overall limited variation seen in the average plagioclase anorthite contents in these experiments suggests that it is relatively insensitive to the range in temperature and pressure conditions investigated here, and thus melt composition and H2O concentration exert the strongest influence on plagioclase chemistry. Finally, although only a single small apatite crystal was documented in one non-halogen-doped experiment in this study, halogens do form a major component of apatite stoichiometry and its occurrence is widespread in arc magmas. Therefore, in higher modal abundances, it has the potential to act as a significant repository of F and Cl in natural systems, and has recently been shown to be a promising oxybarometer owing to its ability to incorporate S existing in a spectrum of oxidation states (Konecke et al., 2017). A full treatment of halogen partitioning and exchange between melt–fluid–apatite in felsic magmas is beyond the scope of our investigation, with such discussion available in studies by Webster et al. (2017) and Riker et al. (2018), and other work by Stock et al. (2016) has shown how the variability in apatite halogen and water concentrations can be used to reconstruct late-stage volatile saturation in explosive volcanic eruptions. Given the proven utility of apatite in recording the volatile systematics of igneous systems, further investigations are required to fully constrain the equilibrium halogen partitioning behaviour that would be expected in complex natural magmatic systems saturated in amphibole ± apatite ± evolved melt ± fluid, and the effect, if any, that apatite would have on F and Cl incorporation into amphibole. CONCLUSIONS Twenty-four IHPV crystallization experiments were conducted to determine the mineral–melt partitioning behaviour of major elements, minor elements, and the halogens F and Cl, between calcic amphibole, plagioclase, clinopyroxene, and hydrous rhyodacitic melts. Overall, the variability seen in partition coefficients is largely unrelated to changes in temperature and pressure (across the limited range investigated). Instead, variations in melt chemistry owing to fO2 and halogen concentrations exert the greatest impact on the composition of the phenocryst phases. In particular, plagioclase strongly prefers Fe3+ and hence shows lower FeO concentrations when crystallized under lower fO2 conditions, whereas amphibole and clinopyroxene show increased FeO and decreased MgO contents at lower fO2. Importantly, whereas Cl concentration in the melt does not affect its absolute partition coefficient into amphibole, higher melt F concentrations lead to lower DFamph/melt values, and crystallization of relatively alkali-enriched amphibole compositions, while also stabilizing amphibole at lower pressures than in F-free melts with otherwise similar compositions. There are competing effects and feedbacks between the incorporation of F and Cl into amphibole, F–Fe and Mg–Cl crystallographic avoidance, and the control of fO2 on the Mg# number of the amphibole. Given the importance of constraining amphibole and melt halogen contents shown in this study, we therefore recommend that future experimental studies aiming to investigate amphibole–melt partitioning behaviour should routinely measure F and Cl concentrations, especially where natural starting powders are used and even when starting charges are not deliberately doped with these elements. The partitioning and coupled substitution behaviours of the major components in amphibole, plagioclase, and clinopyroxene overlap with other experimental results for these phenocrysts in broadly dacitic magmas, across a range of temperature and pressure conditions. However, discrepancies are seen in particular for the halogens in amphibole, and are attributable to melt composition and phenocryst chemistry differences. Whereas calculation of fluid–melt partition coefficients for Cl suggests that it strongly partitions into a separate volatile phase, with the magnitude of this partitioning reduced when coexisting with a more Fe-rich and depolymerized melt, F appears to remains in the melt phase and is strongly compatible in amphibole. The variability in observed DFamph/melt and DClamph/melt values has potential impacts on the composition of multicomponent fluid phases exsolved later from fluid-saturated magmas. Finally, integration of our new data with recently published models for reconstructing pressures, temperatures, and H2O concentrations of crystallizing magmas give good agreement, but the complexities identified here, especially in amphibole, suggest that care must be taken when employing amphibole-based models to reconstruct the evolution of hydrous arc magmas. ACKNOWLEDGEMENTS Thanks go to Keiji Hammond at AMNH for performing the FTIR analyses, and to John Wolff for comments on earlier versions of this paper. Adam Simon, Jake Lowenstern, Wendy Bohrson, and one anonymous reviewer are also thanked for their helpful comments and suggestions that improved the clarity of the paper. The natural Mt Usu volcanic sample was kindly provided by the National Museum of Natural History, Washington, DC. FUNDING This work was funded in part by the National Science Foundation (EAR-1219480 to M. Rowe, EAR-1219484 to J. Webster). SUPPLEMENTARY DATA Supplementary data for this paper are available at Journal of Petrology online. REFERENCES Adam J. , Green T. H., Sie S. H. ( 1993 ). Proton microprobe determined partitioning of Rb, Sr, Ba, Y, Zr, Nb and Ta between experimentally produced amphiboles and silicate melts with variable F content . Chemical Geology 109 , 29 – 49 . Google Scholar Crossref Search ADS WorldCat Andújar J. , Scaillet B., Pichavant M., Druitt T. H. ( 2016 ). Generation conditions of dacite and rhyodacite via the crystallization of an andesitic magma. Implications for the plumbing system at Santorini (Greece) and the origin of tholeiitic or calc-alkaline differentiation trends in arc magmas . Journal of Petrology 57 , 1887 – 1920 . Google Scholar Crossref Search ADS WorldCat Baker D. R. , Alletti M. ( 2012 ). Fluid saturation and volatile partitioning between melts and hydrous fluids in crustal magmatic systems: the contribution of experimental measurements and solubility models . Earth-Science Reviews 114 , 298 – 324 . Google Scholar Crossref Search ADS WorldCat Balcone-Boissard H. , Villemant B., Boudon G. ( 2010 ). Behavior of halogens during the degassing of felsic magmas . Geochemistry, Geophysics, Geosystems 11 , Q09005. Google Scholar OpenURL Placeholder Text WorldCat Barber N. L. ( 2007 ). The effect of fluorine on the viscosity of dacitic melts . Geological Society of America, Abstracts with Programs 39 , 386 . Google Scholar OpenURL Placeholder Text WorldCat Beattie P. , Drake M., Jones J., Leeman W., Longhi J., McKay G., Nielsen R., Palme H., Shaw D., Takahashi E., Watson B. ( 1993 ). Terminology for trace-element partitioning . Geochimica et Cosmochimica Acta 57 , 1605 – 1606 . Google Scholar Crossref Search ADS WorldCat Behrens H. , Ohlhorst S., Holtz F., Champenois M. ( 2004 ). CO2 solubility in dacitic melts equilibrated with H2O–CO2 fluids: implications for modelling the solubility of CO2 in silicic melts . Geochimica et Cosmochimica Acta 68 , 4687 – 4703 . Google Scholar Crossref Search ADS WorldCat Bell A. S. , Simon A. ( 2011 ). Experimental evidence for the alteration of the Fe3+/ΣFe of silicate melt caused by the degassing of chlorine-bearing aqueous volatiles . Geology 39 , 499 – 502 . Google Scholar Crossref Search ADS WorldCat Bell A. S. , Webster J. D. ( 2015 ). Dissolved Cl, oxygen fugacity, and their effects on Fe behavior in a hydrous rhyodacitic melt . American Mineralogist 100 , 1595 – 1599 . Google Scholar Crossref Search ADS WorldCat Bénard A. , Koga K. T., Shimizu N., Kendrick M. A., Ionov D. A., Nebel O., Arculus R. J. ( 2017 ). Chlorine and fluorine partition coefficients and abundances in sub-arc mantle xenoliths (Kamchatka, Russia): implications for melt generation and volatile recycling processes in subduction zones . Geochimica et Cosmochimica Acta 199 , 324 – 350 . Google Scholar Crossref Search ADS WorldCat Binder B. , Wenzel T., Keppler H. ( 2018 ). The partitioning of sulfur between multicomponent aqueous fluids and felsic melts . Contributions to Mineralogy and Petrology 173 , 18 . Google Scholar Crossref Search ADS WorldCat Botcharnikov R. E. , Behrens H., Holtz F., Koepke J., Sato H. ( 2004 ). Sulfur and chlorine solubility in Mt. Unzen rhyodacitic melt at 850°C and 200 MPa . Chemical Geology 213 , 207 – 225 . Google Scholar Crossref Search ADS WorldCat Botcharnikov R. E. , Freise M., Holtz F., Behrens H. ( 2005 ). Solubility of C–O–H mixtures in natural melts: new experimental data and application range of recent models . Annals of Geophysics 48 , 633 – 646 . Google Scholar OpenURL Placeholder Text WorldCat Botcharnikov R. E. , Holtz F., Almeev R. R., Sato H., Behrens H. ( 2008 ). Storage conditions and evolution of andesitic magma prior to the 1991–95 eruption of Unzen volcano: constraints from natural samples and phase equilibria experiments . Journal of Volcanology and Geothermal Research 175 , 168 – 180 . Google Scholar Crossref Search ADS WorldCat Browne B. , Izbekov P., Eichelberger J., Churikova T. ( 2010 ). Pre-eruptive storage conditions of the Holocene dacite erupted from Kizimen Volcano, Kamchatka . International Geology Review 52 , 95 – 110 . Google Scholar Crossref Search ADS WorldCat Brugier Y.-A. , Alletti M., Pichavant M. ( 2015 ). Fe pre-enrichment: a new method to counteract iron loss in experiments on basaltic melts . American Mineralogist 100 , 2106 – 2111 . Google Scholar Crossref Search ADS WorldCat Chambefort I. , Dilles J. H., Longo A. A. ( 2013 ). Amphibole geochemistry of the Yanacocha volcanics, Peru: evidence for diverse sources of magmatic volatiles related to gold ores . Journal of Petrology 54 , 1017 – 1046 . Google Scholar Crossref Search ADS WorldCat Costa F. , Scaillet B., Pichavant M. ( 2004 ). Petrological and experimental constraints on the pre-eruption conditions of Holocene dacite from Volcán San Pedro (36°S, Chilean Andes) and the importance of sulphur in silicic subduction-related magmas . Journal of Petrology 45 , 855 – 881 . Google Scholar Crossref Search ADS WorldCat Cottrell E. , Gardner J. E., Rutherford M. J. ( 1999 ). Petrologic and experimental evidence for the movement and heating of the pre-eruptive Minoan rhyodacite (Santorini, Greece) . Contributions to Mineralogy and Petrology 135 , 315 – 331 . Google Scholar Crossref Search ADS WorldCat Dalou C. ( 2011 ). Fluorine and chlorine fractionation in the sub-arc mantle: an experimental investigation. Ph.D. thesis, Universite Blaise Pascal - Clermont-Ferrand II, pp. 355. Dalou C. , Mysen B. O. ( 2015 ). The effect of H2O on F and Cl solubility and solution mechanisms in aluminosilicate melts at high pressure and high temperature . American Mineralogist 100 , 633 – 643 . Google Scholar Crossref Search ADS WorldCat Dalou C. , Le Losq C., Mysen B. O., Cody G. D. ( 2015 ). Solubility and solution mechanisms of chlorine and fluorine in aluminosilicate melts at high pressure and high temperature . American Mineralogist 100 , 2272 – 2283 . Google Scholar Crossref Search ADS WorldCat Devine J. D. , Gardner J. E., Brack H. P., Layne G. D., Rutherford M. J. ( 1995 ). Comparison of microanalytical methods for estimating H2O contents of silicic volcanic glasses . American Mineralogist 80 , 319 – 328 . Google Scholar Crossref Search ADS WorldCat Donovan J. J. , Kremser D., Fournelle J. H. ( 2007 ). Probe for Windows User’s Guide and Reference , Enterprise edn. Probe Software . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Erdmann M. , Koepke J. ( 2016 ). Experimental temperature cycling as a powerful tool to enlarge melt pools and crystals at magma storage conditions . American Mineralogist 101 , 960 – 969 . Google Scholar Crossref Search ADS WorldCat Giesting P. A. , Filiberto J. ( 2014 ). Quantitative models linking igneous amphibole composition with magma Cl and OH content . American Mineralogist 99 , 852 – 865 . Google Scholar Crossref Search ADS WorldCat Giordano D. , Dingwell D. B. ( 2003 ). Non-Arrhenian multicomponent melt viscosity: a model . Earth and Planetary Science Letters 208 , 337 – 349 . Google Scholar Crossref Search ADS WorldCat Gladstone J. H. , Dale T. P. ( 1864 ). Researches on the refraction, dispersion, and sensitiveness of liquids . Philosophical Transactions of the Royal Society of London 153 , 317 – 343 . Google Scholar Crossref Search ADS WorldCat Halter W. E. , Heinrich C. A., Pettke T. ( 2004 ). Laser-ablation ICP-MS analysis of silicate and sulfide melt inclusions in an andesitic complex II: evidence for magma mixing and magma chamber evolution . Contributions to Mineralogy and Petrology 147 , 397 – 412 . Google Scholar Crossref Search ADS WorldCat Hellwig B. ( 2006 ). The viscosity of dacitic liquids measured at conditions relevant to explosive arc volcanism: determining the influence of temperature, silicate composition, and dissolved volatile content. MSc thesis, University of Missouri, Columbia. Holloway J. R. ( 1971 ). Internally heated pressure vessels. In: Ulmer G. C. (ed.) Research Techniques for High Temperature and Pressure . New York : Springer Verlag , pp. 217 – 258 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Holloway J. R. , Dixon J. E., Pawley A. R. ( 1992 ). An internally heated, rapid-quench, high-pressure vessel . American Mineralogist 77 , 643 – 646 . Google Scholar OpenURL Placeholder Text WorldCat Holtz F. , Sato H., Lewis J., Behrens H., Nakada S. ( 2005 ). Experimental petrology of the 1991–1995 Unzen Dacite, Japan. Part I: phase relations, phase composition and pre-eruptive conditions . Journal of Petrology 46 , 319 – 337 . Google Scholar Crossref Search ADS WorldCat Humphreys M. C. S. , Edmonds E., Christopher T., Hards V. ( 2010 ). Magma hybridisation and diffusive exchange recorded in heterogeneous glasses from Soufrière Hills Volcano, Montserrat . Geophysical Research Letters 37 , L00E06 . Google Scholar Crossref Search ADS WorldCat Iveson A. A. , Webster J. D., Rowe M. C., Neill O. K. ( 2016 ). Magmatic–hydrothermal fluids and volatile metals in the Spirit Lake pluton and Margaret Cu–Mo porphyry system, SW Washington, USA . Contributions to Mineralogy and Petrology 171 , 20. Google Scholar Crossref Search ADS WorldCat Jugo P. , Luth R., Richards J. ( 2005 ). An experimental study of the sulfur content in basaltic melts saturated with immiscible sulfide or sulfate liquids at 1300°C and 1·0 GPa . Journal of Petrology 46 , 783 – 798 . Google Scholar Crossref Search ADS WorldCat Kent A. J. R. , Darr C., Koleszar K., Salisbury M. J., Cooper K. M. ( 2010 ). Preferential eruption of andesitic magmas through recharge filtering . Nature Geoscience 3 , 631 – 636 . Google Scholar Crossref Search ADS WorldCat Kilinc A. , Carmichael I. S. E., Rivers M. L., Sack R. O. ( 1983 ). The ferric–ferrous ratio of natural silicate liquids equilibrated in air . Contributions to Mineralogy and Petrology 83 , 136 – 140 . Google Scholar Crossref Search ADS WorldCat Konecke B. A. , Fiege A., Simon A. C., Parat F., Stechern A. ( 2017 ). Co-variability of S6+, S4+, and S2– in apatite as a function of oxidation state: implications for a new oxybarometer . American Mineralogist 102 , 548 – 557 . Google Scholar Crossref Search ADS WorldCat Laumonier M. , Scaillet B., Pichavant M., Champallier R., Andújar J., Arbaret L. ( 2014 ). On the conditions of magma mixing and its bearing on andesite production in the crust . Nature Communications 5 , 5607 . Google Scholar Crossref Search ADS WorldCat Leake B. E. , Woolley A. R., Arps C. E. S., Birch W. D., Gilbert M. C., Grice J. D., Hawthorne F. C., Kato A., Kisch H. J., Krivovichev V. G., Linthout K., Laird J., Mandarino J. A., Maresch W. V., Nickel E. H., Rock N. M. S., Schumacher J. C., Smith D. C., Stephenson N. C. N., Ungaretti L., Whittaker E. J. W., Guo Y. ( 1997 ). Nomenclature of amphiboles: report of the subcommittee on amphiboles of the International Mineralogical Association, Commission on New Minerals and Mineral Names . Canadian Mineralogist 35 , 219 – 246 . Google Scholar OpenURL Placeholder Text WorldCat Lowenstern J. B. ( 1994 ). Chlorine, fluid immiscibility, and degassing in peralkaline magmas from Pantelleria, Italy . American Mineralogist 79 , 353 – 369 . Google Scholar OpenURL Placeholder Text WorldCat Luhr J. F. ( 2001 ). Glass inclusions and melt volatile contents at Paricutin Volcano, Mexico . Contributions to Mineralogy and Petrology 142 , 261 – 283 . Google Scholar Crossref Search ADS WorldCat Lundgaard K. L. , Tegner C. ( 2004 ). Partitioning of ferric and ferrous iron between plagioclase and silicate melt . Contributions to Mineralogy and Petrology 147 , 470 – 483 . Google Scholar Crossref Search ADS WorldCat Masotta M. , Keppler H., Chaudhari A. ( 2016 ). Fluid–melt partitioning of sulfur in differentiated arc magmas and the sulfur yield of explosive volcanic eruptions . Geochimica et Cosmochimica Acta 176 , 26 – 43 . Google Scholar Crossref Search ADS WorldCat Matjuschkin V. , Blundy J. D., Brooker R. A. ( 2016 ). The effect of pressure on sulphur speciation in mid- to deep-crustal arc magmas and implications for the formation of porphyry copper deposits . Contributions to Mineralogy and Petrology 171 , 66 . Google Scholar Crossref Search ADS WorldCat Metrich N. , Rutherford M. J. ( 1992 ). Experimental study of chlorine behaviour in hydrous silicic melts . Geochimica et Cosmochimica Acta 56 , 607 – 616 . Google Scholar Crossref Search ADS WorldCat Morizet Y. , Paris M., Di Carlo I., Scaillet B. ( 2013 ). Effect of sulphur on the structure of silicate melts under oxidizing conditions . Chemical Geology 358 , 131 – 147 . Google Scholar Crossref Search ADS WorldCat Morrison J. ( 1991 ). Compositional constraints on the incorporation of Cl into amphiboles . American Mineralogist 76 , 1920 – 1930 . Google Scholar OpenURL Placeholder Text WorldCat Muir D. R. , Blundy J. D., Rust A. C., Hickey J. ( 2014 ). Experimental constraints on dacite pre-eruptive magma storage conditions beneath Uturuncu Volcano . Journal of Petrology 55 , 749 – 767 . Google Scholar Crossref Search ADS WorldCat Mysen B. O. ( 1998 ). Structure and Properties of Silicate Melts . Elsevier . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Mysen B. O. ( 2004 ). Element partitioning between minerals and melt, melt composition, and melt structure . Chemical Geology 213 , 1 – 16 . Google Scholar Crossref Search ADS WorldCat Mysen B. O. , Toplis M. J. ( 2007 ). Structural behavior of Al3+ in peralkaline, metaluminous, and peraluminous silicate melts and glasses at ambient pressure . American Mineralogist 92 , 933 – 946 . Google Scholar Crossref Search ADS WorldCat Nandedkar R. H. , Hürlimann N., Ulmer P., Müntener O. ( 2016 ). Amphibole–melt trace element partitioning of fractionating calc-alkaline magmas in the lower crust: an experimental study . Contributions to Mineralogy and Petrology 171 , 71 . Google Scholar Crossref Search ADS WorldCat Neave D. A. , Fabbro G., Herd R. A., Petrone C. M., Edmonds M. ( 2012 ). Melting, differentiation and degassing at the Pantelleria Volcano, Italy . Journal of Petrology 53 , 637 – 663 . Google Scholar Crossref Search ADS WorldCat Newman S. , Lowenstern J. B. ( 2002 ). VolatileCalc: a silicate melt–H2O–CO2 solution model written in Visual Basic for Excel . Computers and Geosciences 28 , 597 – 604 . Google Scholar Crossref Search ADS WorldCat Nielsen C. H. , Sigurdsson H. ( 1981 ). Quantitative methods for electron microprobe analysis of sodium in natural and synthetic glass . American Mineralogist 66 , 547 – 552 . Google Scholar OpenURL Placeholder Text WorldCat Oberti R. , Ungaretti L., Cannillo E., Hawthorne F. C. ( 1993 ). The mechanism of Cl incorporation in amphibole . American Mineralogist 78 , 746 – 752 . Google Scholar OpenURL Placeholder Text WorldCat Pitzer K. S. , Pabalan R. T. ( 1986 ). Thermodynamics of NaCl in steam . Geochimica et Cosmochimica Acta 50 , 1445 – 1454 . Google Scholar Crossref Search ADS WorldCat Prouteau G. , Scaillet B. ( 2003 ). Experimental constraints on the origin of the 1991 Pinatubo dacite . Journal of Petrology 44 , 2203 – 2241 . Google Scholar Crossref Search ADS WorldCat Putirka K. ( 2016 ). Amphibole thermometers and barometers for igneous systems and some implications for eruption mechanisms of felsic magmas at arc volcanoes . American Mineralogist 101 , 841 – 858 . Google Scholar Crossref Search ADS WorldCat Richards J. P. ( 2011 ). Magmatic to hydrothermal metal fluxes in convergent and collided margins . Ore Geology Reviews 40 , 1 – 26 . Google Scholar Crossref Search ADS WorldCat Ridolfi F. , Renzulli A., Puerini M. ( 2010 ). Stability and chemical equilibrium of amphibole in calc-alkaline magmas: an overview, new thermobarometric formulations and application to subduction-related volcanoes . Contributions in Mineralogy and Petrology 160 , 45 – 66 . Google Scholar Crossref Search ADS WorldCat Riker J. , Humphreys M. C. S., Brooker R. A., De Hoog J. C. M. & EIMF . ( 2018 ). First measurements of OH-C exchange and temperature-dependent partitioning of OH and halogens in the system apatite–silicate melt . American Mineralogist 103 , 260 – 270 . Google Scholar Crossref Search ADS WorldCat Riker J. M. , Blundy J. D., Rust A. C., Botcharnikov R. E., Humphreys M. C. S. ( 2015 ). Experimental phase equilibria of a Mount St. Helens rhyodacite: a framework for interpreting crystallization paths in degassing silicic magmas . Contributions to Mineralogy and Petrology 170 , 6 . Google Scholar Crossref Search ADS WorldCat Rutherford M. J. ( 2008 ). Magma ascent rates. In: Putirka K. D. & Tepley F. J., III (eds) Minerals, Inclusions and Volcanic Processes . Mineralogical Society of America and Geochemical Society, Reviews in Mineralogy and Geochemistry 69 , 241 – 271 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Rutherford M. J. , Devine J. D. III ( 2008 ). Magmatic conditions and processes in the storage zone of the 2004–2006 Mount St. Helens Dacite. In: Sherrod D. R., Scott W. E. & Stauffer P. H. (eds) A Volcano Rekindled: The Renewed Eruption of Mount St. Helens, 2004–2006 . US Geological Survey, Professional Papers 1750 , 703 – 725 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sato H. , Holtz F., Behrens H., Botcharnikov R., Nakada S. ( 2005 ). Experimental petrology of the 1991–1995 Unzen dacite, Japan. Part II: Cl/OH partitioning between hornblende and melt and its implications for the origin of oscillatory zoning of hornblende phenocrysts . Journal of Petrology 46 , 339 – 354 . Google Scholar Crossref Search ADS WorldCat Scaillet B. , Macdonald R. ( 2003 ). Experimental constraints on the relationships between peralkaline rhyolites of the Kenya Rift Valley . Journal of Petrology 44 , 1867 – 1894 . Google Scholar Crossref Search ADS WorldCat Scaillet B. , Pichavant M., Roux J., Humbert G., Lefèvre A. ( 1992 ). Improvements of the Shaw membrane technique for measurement and control of fH2 at high temperatures and pressures . American Mineralogist 77 , 647 – 655 . Google Scholar OpenURL Placeholder Text WorldCat Schmidt B. C. , Scaillet B., Holtz F. ( 1995 ). Accurate control of fH2 in cold-seal pressure vessels with the Shaw membrane technique . European Journal of Mineralogy 7 , 893 – 903 . Google Scholar Crossref Search ADS WorldCat Shinohara H. ( 1994 ). Exsolution of immiscible vapour and liquid phases from a crystallizing silicate melt: implications for chlorine and metal transport . Geochimica et Cosmochimica Acta 58 , 5215 – 5221 . Google Scholar Crossref Search ADS WorldCat Silver L. A. , Ihinger P. D., Stolper E. ( 1990 ). The influence of bulk composition on the speciation of water in silicate glasses . Contributions to Mineralogy and Petrology 104 , 142 – 162 . Google Scholar Crossref Search ADS WorldCat Simon A. C. , Pettke T., Candela P. A., Piccoli P. M., Heinrich C. A. ( 2004 ). Magnetite solubility and iron transport in magmatic–hydrothermal environments . Geochimica et Cosmochimica Acta 68 , 4905 – 4914 . Google Scholar Crossref Search ADS WorldCat Stock M. J. , Humphreys M. C. S., Smith V. C., Isaia R., Pyle D. M. ( 2016 ). Late-stage volatile saturation as a potential trigger for explosive volcanic eruptions . Nature Geoscience 9 , 249 – 254 . Google Scholar Crossref Search ADS WorldCat Streck M. J. , Broderick C., Thornber C. R., Clynne M. A., Pallister J. S. ( 2008 ). Plagioclase populations and zoning in dacites of the 2004–2006 Mount St. Helens eruption: Constraints for magma origin and dynamics. In: Sherrod D. R., Scott W. E. & Stauffer P. H. (eds) A Volcano Rekindled: The Renewed Eruption of Mount St. Helens, 2004–2006 . US Geological Survey, Professional Papers 1750 , 791 – 808 . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Tepley F. J. III, De Silva S., Salas G. ( 2013 ). Magma dynamics and petrological evolution leading to the VEI 5 2000 bp eruption of El Misti volcano, southern Peru . Journal of Petrology 54 , 2033 – 2065 . Google Scholar Crossref Search ADS WorldCat Tomiya A. , Takahashi E. ( 2005 ). Evolution of the magma chamber beneath Usu volcano since 1663: a natural laboratory for observing changing phenocryst compositions and textures . Journal of Petrology 46 , 2395 – 2426 . Google Scholar Crossref Search ADS WorldCat Tomiya A. , Takahashi E., Furukawa N., Suzuki T. ( 2010 ). Depth and evolution of a silicic magma chamber: melting experiments on a low-K rhyolite from Usu volcano, Japan . Journal of Petrology 51 , 1333 – 1354 . Google Scholar Crossref Search ADS WorldCat Van den Bleeken G. , Koga K. T. ( 2015 ). Experimentally determined distribution of fluorine and chlorine upon hydrous slab melting, and implications for F–Cl cycling through subduction zones . Geochimica et Cosmochimica Acta 171 , 353 – 373 . Google Scholar Crossref Search ADS WorldCat Venezky D. Y. , Rutherford M. J. ( 1999 ). Petrology and Fe–Ti oxide reequilibration of the 1991 Mount Unzen mixed magma . Journal of Volcanology and Geothermal Research 89 , 213 – 230 . Google Scholar Crossref Search ADS WorldCat Vigneresse J. L. ( 2007 ). The role of discontinuous magma inputs in felsic magmas and ore generation . Ore Geology Reviews 30 , 181 – 216 . Google Scholar Crossref Search ADS WorldCat Wallace P. J. , Carmichael I. S. E. ( 1994 ). S-speciation in submarine basaltic glasses as determined by measurements of SKα X-ray wavelength shifts . American Mineralogist 79 , 161 – 167 . Google Scholar OpenURL Placeholder Text WorldCat Waters L. E. , Lange R. A. ( 2015 ). An updated calibration of the plagioclase–liquid hygrometer–thermometer applicable to basalts through rhyolites . American Mineralogist 100 , 2172 – 2184 . Google Scholar Crossref Search ADS WorldCat Webb S. , Müller E., Büttner H. ( 2004 ). Anomalous rheology of peraluminous melts . American Mineralogist 89 , 812 – 818 . Google Scholar Crossref Search ADS WorldCat Webster J. D. ( 1992 ). Fluid–melt interactions in Cl-rich granitic systems: effects of melt composition at 2 kbar and 800°C . Geochimica et Cosmochimica Acta 56 , 659 – 678 . Google Scholar Crossref Search ADS WorldCat Webster J. D. ( 1997 ). Chloride solubility in felsic melts and the role of chloride in magmatic degassing . Journal of Petrology 38 , 1793 – 1807 . Google Scholar Crossref Search ADS WorldCat Webster J. D. ( 2004 ). The exsolution of magmatic hydrosaline chloride liquids . Chemical Geology 210 , 33 – 48 . Google Scholar Crossref Search ADS WorldCat Webster J. D. , Holloway J. R. ( 1990 ). Partitioning of F and Cl between magmatic hydrothermal fluids and highly evolved granitic magmas. In: Stein H. J. & Hannah J. L. (eds) Ore-bearing Granite Systems; Petrogenesis and Mineralizing Processes . Geological Society of America, Special Papers 246 , 21 – 34 . Google Scholar Crossref Search ADS Google Preview WorldCat COPAC Webster J. D. , Tappen C. M., Mandeville C. W. ( 2009 ). Partitioning behavior of chlorine and fluorine in the system apatite–melt–fluid. II: felsic silicate systems at 200 MPa . Geochimica et Cosmochimica Acta 73 , 559 – 581 . Google Scholar Crossref Search ADS WorldCat Webster J. D. , Goldoff B. A., Sintoni M. F., Shimizu N., De Vivo B. ( 2014 ). C–O–H–Cl–S–F volatile solubilities, partitioning, and mixing in phonolitic–trachytic melts and aqueous–carbonic vapor ± saline liquid at 200 MPa . Journal of Petrology 55 , 2217 – 2248 . Google Scholar Crossref Search ADS WorldCat Webster J. D. , Vetere F., Botcharnikov R. E., Goldoff B., McBirney A., Doherty A. ( 2015 ). Experimental and modeled chlorine solubilities in aluminosilicate melts at 1 to 7000 bars and 700 to 1250°C: applications to magmas of Augustine Volcano, Alaska . American Mineralogist 100 , 522 – 535 . Google Scholar Crossref Search ADS WorldCat Webster J. D. , Goldoff B. A., Flesch R. N., Nadeau P. A., Silbert Z. W. ( 2017 ). Hydroxyl, Cl, and F partitioning between high-silica rhyolitic melts–apatite–fluid(s) at 50–200 MPa and 700–1000 ° C . American Mineralogist 102 , 61 – 74 . Google Scholar Crossref Search ADS WorldCat Yamashita S. , Kitamura T., Kusakabe M. ( 1997 ). Infrared spectroscopy of hydrous glasses of arc magma compositions . Geochemical Journal 31 , 169 – 174 . Google Scholar Crossref Search ADS WorldCat Zajacz Z. , Candela P. A., Piccoli P. M., Sanchez-Valle C. ( 2012 ). The partitioning of sulfur and chlorine between andesite melts and magmatic volatiles and the exchange coefficients of major cations . Geochimica et Cosmochimica Acta 89 , 81 – 101 . Google Scholar Crossref Search ADS WorldCat Zajacz Z. , Candela P. A., Piccoli P. M., Sanchez-Valle C., Wälle M. ( 2013 ). Solubility and partitioning behavior of Au, Cu, Ag and reduced S in magmas . Geochimica et Cosmochimica Acta 112 , 288 – 304 . Google Scholar Crossref Search ADS WorldCat Zhang C. , Holtz F., Ma C., Wolff P. E., Li X. ( 2012 ). Tracing the evolution and distribution of F and Cl in plutonic systems from volatile-bearing minerals: a case study from the Liujiawa pluton (Dabie orogen, China) . Contributions to Mineralogy and Petrology 164 , 859 – 879 . Google Scholar Crossref Search ADS WorldCat Author notes Present address: Earth and Environmental Sciences, University of Michigan, Ann Arbor, MI 48109, USA. © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) © The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
TI - Major Element and Halogen (F, Cl) Mineral–Melt–Fluid Partitioning in Hydrous Rhyodacitic Melts at Shallow Crustal Conditions
JF - Journal of Petrology
DO - 10.1093/petrology/egy011
DA - 2017-12-01
UR - https://www.deepdyve.com/lp/oxford-university-press/major-element-and-halogen-f-cl-mineral-melt-fluid-partitioning-in-vgn4lOzPvZ
SP - 2465
EP - 2492
VL - 58
IS - 12
DP - DeepDyve
ER -