Supplementary Materials for: Geothermal energy and ore forming potential of 600 °C mid-ocean ridge hydrothermal fluids

The ∼ 4500-m-deep Iceland Deep Drilling Project (IDDP) borehole IDDP-2 in Iceland pen-etrated the root of an active seawater-recharged hydrothermal system below the Mid-Atlantic Ridge. As direct sampling of pristine free fluid was impossible, we used fluid inclusions to constrain the in situ conditions and fluid composition at the bottom of the hydrothermal con-vection cell. The fluid temperature is ∼ 600 ° C, and its pressure is near-hydrostatic ( ∼ 45 MPa). The fluid exists as two separate phases: an H 2 O-rich vapor (with an enthalpy of ∼ 59.4 kJ/mol) and an Fe-K–rich brine containing 2000 μ g/g Cu, 3.5 μ g/g Ag, 1.4 μ g/g U, and 0.14 μ g/g Au. Occasionally, the fluid inclusions coexist with rhyolite melt inclusions. These findings indicate that the borehole intersected high-energy steam, which is valuable for energy production, and discovered a potentially ore-forming brine. We suggest that similar fluids circulate deep beneath mid-ocean ridges worldwide and form volcanogenic massive sulfide Cu-Zn-Au-Ag ore deposits.


Microthermometry
Microthermometric analyses were carried out using Linkam TMS600 and Linkam TS1500 heating stages at the Lithosphere Fluid Research Lab of the Faculty of Science at the Eötvös University, Budapest (ELTE LRG). Doubly polished mineral chips were heated at a rate of 10 °C/min to approximately 20 °C below the phase transformation temperatures, at which point the heating rate was changed to 5 °C/min. The precision of temperature readings of the TMS600 and TS1500 stages is 0.2 and 2 °C, respectively. Phase transformations were determined by changing the temperature at the rate of 1 °C/min 10 °C below to 10 °C above the phase transformation temperatures and repeating this procedure a few times for each FIA. With the exception of the dissolution of S4 in brine inclusions, the Linkam TMS600 heating stage was used. The hightemperature (TS1500) stage was used to observe the dissolution of S4 and the homogenization of brine inclusions. The heating rate during these experiments was the same as in the low-T experiments.

Raman spectroscopy
Raman spectroscopy analyses were done at the Research and Instrument Core Facility Center of the Faculty of Science at the Eötvös University, Budapest (ELTE FS-RICF), both at room temperature and during the heating experiments. During room-temperature analyses a 100x objective (NA=0.9) was used, whereas during high-temperature measurements we used a 50x objective lens (NA=0.5). Analyses were carried out with a Horiba LabRam HR800 instrument, applying a 532 nm (green) laser with maximum laser energy of 130 mW (~25 mW on the sample surface), 100 m confocal pinhole, 600 to 1800 gratings. Analyses were run for 30 seconds with 2-3 repetitions. Gas and liquid components were analyzed with the maximum laser energy, whereas for mineral phases we reduced the laser power to ~2.5 mW on the sample surface with a filter, in order to avoid beam damage.
During the high-temperature analyses, spectra were collected 10 to 60 minutes after the target temperature was reached. This was especially important during the homogenization experiments with vapor-rich inclusions. For each vapor-rich inclusion, spectra were collected twice at 590 °C, first at 10 minutes after the target temperature was reached, and again after 50-60 minutes. This was done to avoid the presence of any metastable liquid film remaining upon heating. No difference was observed between the spectra collected before and after the target temperature was reached.
Raman mapping was conducted in order to identify stretching bands of a given mineral with different orientations. Maps were collected at room temperature, using a 100 μm confocal hole, 600 grooves/mm optical gratings, and 6-10 × 10 s acquisition time. The investigated spectral range was between 110 cm -1 and 1900 cm -1 (i.e. the "fingerprint" region), and between 2800 and 4000 cm-1 (i.e.: the "OH" region). The spectral resolution of the measurements was 3.0 cm -1 at 1707.9 cm -1 and 2.4 cm -1 at 3642.2 cm -1 (defined as the measured full width at half maximum values of neon atomic emission lines). The Raman maps were processed using LabSpec 5.

Electron microprobe analyses
Electron microprobe analyses of silicate melt inclusions were carried out with a JEOL JXA-8230 SuperProbe at the Institute of Earth Sciences, University of Iceland. Operating conditions were: accelerating voltage of 15 kV, beam current of 5 nA, and a beam diameter of 4 to 7 m, depending on melt inclusion size. Analyses were acquired with WDS spectrometry where Na was analyzed first to minimize the effect of Na migration caused by the electron beam. Natural and synthetic standards from Astimex Ltd and the Smithsonian Institute were used for calibration. Additionally, we used ATHO rhyolitic glass as a secondary standard.
Cathodoluminescence (CL) imaging and trace element analyses in quartz were carried out with the same instrument. For CL imaging a beam current of 50 nA was used. Trace elements were analyzed with an accelerating voltage of 15 kV, a beam current of 100 nA and a beam diameter of 10 m.

Focused Ion Beam Scanning Electron Microscopy with Energy Dispersive Spectrometryslice-and-view (FIB-SEM-EDS)
The FIB-SEM-EDS analyses were acquired with a FEI Quanta 3D Dual Beam SEM at the Research and Instrument Core Facility of the Faculty of Science, Eötvös University, Budapest (ELTE FS-RICF). The EDS analyses and SEM imaging were done using 15-20 kV acceleration voltage and 23.7-480 pA beam current, depending on magnification. The surface above a selected fluid inclusion was coated by a ~1 µm thin platinum layer in order to protect the inclusion from the high current of Ga + ions used for milling the three trenches around the inclusion. Low-current Ga + ions were used for slicing the inclusion; the distance between the slices was ~150 to 250 nm depending on the suspected size of the smallest solid phases within the inclusions previously identified by Raman measurements. High-resolution secondary electron (SE) and backscattered electron (BSE) images were taken of each newly exposed surface (slice) of the inclusion in order to calculate the exact volume proportions of the different phases.
Exposed solid phases were identified using point, or area measurements with a high-resolution EDAX SDD (Silicon Drift Detector).
Laser-Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) LA-ICP-MS analyses were acquired with an Agilent 8800 ICP-MS equipped with a GeoLas-Pro 193 nm ArF Excimer laser system at the Institute of Geological Sciences, University of Bern. GSD-1G glass was used as the standard for the calibration of trace elements and Scapolite crystal (Sca-17) for the calibration of S, Cl, Br (Seo et al., 2011). Each standard was analyzed 3 times before and after a block of ~12 fluid inclusions and host minerals to allow for true-time drift correction. Standard measurements took ~30 seconds, and the background signal was collected for >50 seconds. The spot size was 44 m for GSD1G and 60 m for Sca17. The protocol of individual fluid inclusion measurement closely followed that reported in Pettke et al. (2012). Whenever possible, a stepwise increase of the beam size prior to breaching the inclusion was employed to reduce catastrophic ablation of quartz. The final beam size varied between 18 and 44 m, depending on inclusion size, which was generally between 8 and 25 m. In the case of a few deeper inclusions, active focusing of the laser beam was also applied. Host mineral signals prior to, and after the inclusion signal, were used for host mineral correction, and where no clean host interval could be obtained from these signals, a separate host mineral measurement was done next to the inclusion using a 44 µm beam. For quantification of brine inclusions, a salinity of 5 wt% NaCl was used for internal standardisation, based on FIB-SEM results (TableS2). For vapor-rich inclusions 0.5 wt% NaCl equiv was used, which was derived from thermodynamic data (Driesner, 2007) based on the results of the heating experiments. The SILLS software was used for data reduction (Guillong et al., 2008).
2) Determination of volume proportion of phases in vapor-rich and brine inclusions Volume proportions of mineral phases in four brine inclusions were determined based on backscatter and secondary electron images collected during FIB-SEM analyses. Fluid inclusions were sliced in 150 or 200 nm steps and backscatter and secondary electron images were collected on each slice. In each inclusion, the step size was constant. Larger inclusions were sliced by using 200 nm steps. These images were processed (with various phases delineated and identified in each image) with the computer program ImageJ and volume proportions in each image were calculated. The final volume proportions in a single inclusion were derived based on 31 to 61 images, depending on the inclusion size. Along with estimated mineral densities (Table S2), these volume proportions were used for mass balance calculations to determine the chemical composition of brine inclusions. The derived NaCl contents were subsequently used as an internal standard for LA-ICP-MS analyses.
Volume proportion of liquid and vapor phases in vapor-rich inclusions were determined in a similar way based on photomicrographs of 12 vapor-rich inclusions.

1) General description of core 11:
Core 11 of the IDDP-2 borehole was collected between 4634.2 and 4642.8 m depth with a recovery of 88% (Friðleifsson et al., 2020). The main rock type is metagabbro, which generally retains its original igneous texture (Fig. S1a), but which has been pervasively hydrothermally altered (Friðleifsson et al., 2020). For example, igneous clinopyroxene has been replaced by hornblende, secondary clinopyroxene, and orthopyroxene (enstatite component between 60 and 40 mol%). Plagioclase varies in composition from andesine to anorthite and appears dusty due to abundant vapor-rich fluid inclusions. Igneous titanomagnetite was replaced by an equigranular intergrowth of magnetite and ilmenite. Hydrothermal biotite is a minor alteration phase in most samples from Core 11 (Friðleifsson et al., 2020, Zierenberg et al. in press). Clinopyroxeneorthopyroxene and hornblende-plagioclase geothermometry suggests that most of the hydrothermal alteration occurred at temperatures between ~800 and 825 °C. Magnetite-ilmenite geothermometry indicates that this mineral pair continued to re-equilibrate to temperatures near the present-day temperature of ~600 °C before being quenched by the drilling fluid during core recovery (Zierenberg et al. in press). In order to confirm that these FIAs indeed cut grain boundaries and represent the latest fluid generation, we carried out cathodoluminescence (CL) imaging. The studied secondary FIAs are generally associated with "bright" fractures, which commonly connect to bright, fine-grained quartz along grain boundaries (Fig. 2a). Rarely, "dark" fractures cut these bright fracture zones.
Although bright quartz (Fig. 2a) is more enriched in trace elements, including Ti (~200 ppm), compared to "dark" quartz (~100 ppm), this does not necessarily mean that these fractures represent higher T events. Trace element-rich quartz can also precipitate by rapid cooling (Huang and Audétat, 2012). Few fluid inclusions were analyzed in dark fractures too. No difference in chemical composition was observed compared to those in bright fractures. Therefore, we concluded that the different quartz compositions related to different luminescence are the result of how fast the fractures healed rather than changes in temperature conditions. Vapor-rich inclusions and are generally composed of a large dark vapor bubble and a thin liquid film around it (Fig. 2b, S2a, S3). In some of these inclusions, a small opaque phase can also be observed (Fig. S2a). At room temperature, the vapor phase in these inclusions is dominated by CO 2 with minor H 2 S and traces of H 2 ( Fig. S3; Table S3), whereas the liquid film is composed of low-salinity water (Fig. S4b, d). Based on the volume proportions of the liquid film and the vapor bubble in 12 inclusions at room temperature the density of the vapor-rich inclusions is estimated to be ~0.07 (±0.04) g/cm 3 , assuming that the liquid phase at room temperature is pure H 2 O. The density is likely to be higher due to the minor chloride component(s) (Table S1).
Brine inclusions are composed of four different solid phases, a vapor bubble ± a minor liquid phase (Fig. 2c, Fig. S2b, Fig. S5). The shape of the vapor bubble in these inclusions is always irregular, indicating that the liquid phase should be minor as the vapor largely fills the space remaining after the crystallization of the solid phases. Solid phase 1 (S1) is green to yellow pleochroic in plane-polarized light (Fig. S2b) and is strongly anisotropic, as seen with crossed polars (Fig. 2c). At room temperature this mineral has two characteristic Raman bands at 3451 and 200 cm -1 (Fig. S5b-d). Further smaller bands are observed in some of the spectra at 140 and 385 cm -1 . The approximate stoichiometry of this phase is (Fe,Mn) 3 (Na,K) 2 Cl 3 (OH) 5 based on FIB-SEM-EDS analyses (Table S2). Solid phase 2 (S2) is also greenish in plane-polarized light (Fig 1c) and seems to be isotropic judging from observations made with crossed polars (Fig.   S1b). This mineral has one characteristic Raman band measured at 3457 cm -1 (Fig. S3c) and its approximate stoichiometry is (Fe,Mn)(Na,K)Cl(OH) 2 (TableS2). The chemical composition and optical properties of this mineral are similar to douglasite (Palache et al., 1951). Solid phase 3 (S3) is a transparent, isotropic mineral (Fig 1c and Fig. S1b), commonly showing cubic crystal habit with no characteristic Raman bands and is a halite-sylvite solid solution dominated by the halite component (Table S1). Phase S4 (Fig 2c, Fig. S2b, Fig. S5) is a Cu-Fe sulfide (Table S1) with characteristic Raman bands at 381, 405 and 450 cm -1 (Fig. S5e), which are similar to those in molybdenite (Windom et al., 2011). Phases S1-S4 were observed by optical microscope in each brine inclusion and their phase transformations upon heating were easily recognized (see below). Based on FIB-SEM-EDS analyses, we identified further minor phases in these inclusions, such as Fe-dominated chloride-hydroxide fluid condensate (similar in composition to akageneite), Ba-rich and Ca-rich chloride phases (TableS2), and apatite. The Fe-rich chloridehydroxide fluid condensate commonly has a platy crystal habit in the opened inclusions, or coats the surfaces of other phases. When a fluid inclusion is opened during FIB-SEM analysis, a droplet of liquid escapes and after drying leaves behind material with this composition.
Silicate melt inclusions (Fig. 2d) are composed of a colorless silicate glass and a vapor bubble. The size of the vapor bubble is variable in these inclusions, and it is dominated by CO 2 and contains minor amount of H 2 S. The bubble phase of these inclusions occasionally contains the S1 phase (Fig. S2e). Although melt inclusions are rare, it occurs in the secondary FIAs with vapor-rich and brine inclusions (FigS2c and d).    The dissolution of S1, which commonly coats the surface of S2, at 179°C permitted analysis of pure S2 at high T. Table S1. Average composition of vapor-rich and brine fluid inclusions from the IDDP-2 borehole below 4500 m depth. H 2 O, CO 2 , H 2 S, and H 2 contents in vapor-rich inclusions were determined by Raman spectroscopy at 600 C in homogenized inclusions. All other data are LA-ICP-MS results. Levels of detection (LODs) in µg/g report the numbers achieved during an ideal LA-ICP-MS measurement of an equant ~25µm brine inclusion (LODs for smaller inclusions are typically higher (see Pettke et al., 2012, for discussion). H 2 was not detected in the high-temperature Raman spectra; its abundance in the bulk fluid is calculated from its concentration in the bubble phase at room T (see Table S3).  Table S2. Average composition of mineral phases and their volume proportions in brine fluid inclusions from the IDDP-2 borehole below 4500 m depth as determined by FIB-SEM-EDS.

Mass
Compositions are in atomic % (at%) and weight % (wt%); variation is 1  standard deviation (stdev). Phases S7 and S8 were too small to analyze without contamination by the surrounding minerals. Thus, we report only the least contaminated analyses. Composition of S6 was observed in three fluid inclusions and probably represents mixed analyses of S1 and S2. Ss -solid solution, b.d. -below detection. OH was calculated based on charge balance.
Major phases distinguished by optical microscopy S1,

Supplement3 -Results of fluid inclusion heating experiments
Microthermometric analyses were carried out on four samples (4635.11, A4637.80, N4635.25, G4634.48). In N4635.25, A4637.80 and G4634.48, we observed phase transformations and homogenization behavior of the brine inclusions, whereas in 4635.11 and G4634.48 phase transformations were observed in the vapor-rich inclusions. Some of these inclusions were also the subject to room-T and high-T Raman spectroscopy measurements.
Vapor-rich inclusions: The reported observations are based on measurements carried out on two fluid inclusion assemblages (FIAs) in two samples that contained five fluid inclusions (three and two inclusions, respectively). While the vapor-rich inclusions were heated, the composition of the bubble phase as well as the liquid film (Fig. 1a2, FigS2) was monitored by Raman spectroscopy. In vapor-rich inclusions, the liquid film at the inclusion wall disappears optically upon heating at 550 °C.
However, liquid H 2 O is still detectable in the Raman spectra up to at least 590 °C at the margins of each studied inclusion (Fig. S3b). At 600 °C the liquid-H 2 O signal disappeared in four out of five spectra, although in FI1 (Fig. S3b) there is a broad shoulder at the Raman shift of ~3540 cm -generally between 385 and 390 °C as determined by the Linkam TMS600 stage; lower temperatures were measured only during the high-T experiments carried out with the Linkam TS1500 stage.
The fact that the dissolution of S4 in brine inclusions and the evaporation of the liquid film in vapor-rich inclusions occur at the same temperature of 600±20 °C, we suggest that the trapping temperature of the endmember fluids to be close to this value at ~4500 m depth. Brine inclusions failed to homogenize completely to liquid at this temperature because they contain various amounts of accidently trapped vapor endmember. Thus, none of the analyzed brine inclusions are true endmember liquids.

Supplement4 -Magmatic fluid or modified seawater
We have the following arguments which suggests that the contribution of magmatic fluid to the hydrothermal fluid in this study is minimal.
1) The degassing of tholeiitic mid-oceanic ridge basalt (MORB) and its fractionation products produce magmatic fluids with low Cl/B ratios (Fig. S6 e.g., Arnórsson 1995, Aggarwal et al. 2000. This is because these melts are generally not saturated in chlorine due to their low original chlorine and H 2 O contents.    (Hart et al., 1999, Oppenheimer, 2004. 2) In tholeiitic systems the main degassing volatiles besides water are CO 2 and SO 2 , which might contribute to the hydrothermal system above a shallow magmatic body. High pressure and temperature experiments demonstrated that from a basalt at shallow depths (100 to 300 MPa pressure) sulfur is preferentially released relative to chlorine and the S/Cl ratio in the degassing fluid is high (see e.g., Lesne et al. 2011). In accordance with this, significant sulfur release was observed during the last volcanic eruption of Iceland with only a minor outgassing of Cl Reykjanes, the S/Cl ratio in the hydrothermal fluid is expected to be above seawater values (0.047). In our inclusions, the S/Cl ratios are like that in seawater (or lower), thus no increase of S was observed. This does not exclude the addition of minor magmatic fluid to the system, but significant addition is not probable.
3) The enrichment of K over Na both in the brine and vapor excludes that these fluids are direct product of magmatic degassing. Tholeiitic basalts and their fractionation products are sodic, and in many cases contain only minor K 2 O in Iceland. Experimental studies (e.g.: Clark, 1966, Ayers andEggler, 1995) suggest that Na more readily partitions to a brine phase compared to K. Therefore, a tholeiitic melt with Na/K mass ratios above 1 cannot produce a degassing brine with Na/K mass ratios below 1.

Additional References:
Aggarwal