Magnesium silicate hydrate regulates the dissolution of natural ultramafic rocks and associated hydrogen generation at low temperatures

Abstract

Hydrous alteration of ultramafic rocks produces unique reducing environments accompanied by hydrogen (H₂) generation. To understand the early stages of the reaction, batch experiments were conducted at 90°C for 2 weeks using a NaNO₃ solution, natural dunite and harzburgite samples with variable degrees of serpentinization. Our results indicate that the fresh ultramafic rocks generate more H₂ than serpentinized rocks, showing that the dissolution of major minerals in fresh ultramafic rocks (i.e., pyroxene and olivine) is the dominant factor in H₂ generation. Fresh harzburgite yielded higher amounts of H₂ of up to 322.1 μmol/kg than the other ultramafic rocks. Fresher samples had higher dissolved Si and Ca concentrations in the solutions with higher H₂ generation than the serpentinized samples, which can be explained by the dissolution of pyroxene, because the main host mineral for Ca is clinopyroxene. Magnesium-bearing silicates were observed in the experiments using a fresh harzburgite, probably due to the more effective Si supply from pyroxene dissolution than olivine. Another series of experiments using a fresh harzburgite with different chemical reagents showed that the addition of Si enhanced H₂ generation, suggesting that H₂ generation was regulated by the precipitation of magnesium silicate hydrate. Thermodynamic calculations indicated that the solution chemistry during the hydration of fresh ultramafic rocks was regulated by magnesium silicate hydrate, whereas the solution chemistry during the hydration of serpentinized rocks was buffered by brucite. Our study suggests that fresh harzburgite is more favorable for H₂ generation than dunite at 90°C because of its higher pyroxene content.

Introduction

Hydrous alteration of ultramafic rocks (i.e., serpentinization) and associated hydrogen (H₂) generation have been widely recognized in Earth’s surface environments (e.g., spring waters emerging from the on-land Oman ophiolites: Neal and Stanger, 1983; Lost City: Kelley et al., 2001; Tekirova ophiolites at Çirali, Turkey: Etiope et al., 2011; Hakuba Happo hot spring at Nagano, Japan: Suda et al., 2014). Previous studies have suggested that a large proportion of global H₂ production results from serpentinization in marine systems and the continental lithosphere (e.g., Cannat et al., 2010; Lollar et al., 2014; Merdith et al., 2020; Zgonnik, 2020), much of which occurs in low-temperature (<50°C) environments. Due to ongoing efforts to counter global warming, such natural H₂ has received significant attention because it is a zero emission fuel.

These H₂-rich environments may also be modern analogs of the early Earth’s surface and possible sites of prebiotic molecular synthesis and ecosystems containing early life (Martin and Russel, 2007; Lang et al., 2010; Ménez et al., 2018; Preiner et al., 2018, 2020), as well as extraterrestrial habitats (e.g., Mars, Europa, and Enceladus; Sleep et al., 2004; Schulte et al., 2006; Glein et al., 2015; Greenberger et al., 2015; McCollom et al., 2022). However, the geochemical processes that control the hydrous alteration of natural ultramafic rocks and associated H₂ generation have yet to be clarified.

Serpentinization of ultramafic rocks occurs by the hydration of primary silicate minerals (e.g., olivine and pyroxene) during water–rock interactions, which is conceptually described by the following reactions:

  

$${3\mathrm{Mg}}_{1.8}{\mathrm{Fe}}_{0.2}{\mathrm{SiO}}_4\left(\text{olivine}{\mathrm{Fo}}_{90}\right)+4.1{\mathrm{H}}_2\mathrm{O}=1.5{\mathrm{Mg}}_3{\mathrm{Si}}_2{\mathrm{O}}_5(\mathrm{OH}{)}_4(\text{serpentine})+0.9\mathrm{Mg}(\mathrm{OH}{)}_2(\text{brucite})+0.2{\mathrm{Fe}}_3{\mathrm{O}}_4(\text{magnetite})+0.2{\mathrm{H}}_2,$$(1)

(Seyfried et al., 2007)

and

  

$$3{\mathrm{Mg}}_{1.8}{\mathrm{Fe}}_{0.2}{\mathrm{Si}}_2{\mathrm{O}}_6(\text{pyroxene})+4.1{\mathrm{H}}_2\mathrm{O}=1.5{\mathrm{Mg}}_3{\mathrm{Si}}_2{\mathrm{O}}_5(\mathrm{OH}{)}_4+0.9\mathrm{Mg}(\mathrm{OH}{)}_2+0.2{\mathrm{Fe}}_3{\mathrm{O}}_4+0.2{\mathrm{H}}_2+3{\mathrm{SiO}}_{2(\mathrm{aq})}.$$(2)

The formula (2) can be derived by replacing the reactant in formula (1) with a common orthopyroxene. In reactions (1) and (2), H₂ is generated by the reduction of water coupled with the oxidation of Fe²⁺ dissolved from olivine or pyroxene, which results in magnetite formation, although oxidized Fe³⁺ can also be hosted by other minerals (e.g., serpentine and brucite; Beard and Frost, 2016; McCollom et al., 2022). Various experimental studies have demonstrated that serpentinization and associated abiotic H₂ and hydrocarbon formation occurs at 150–400°C (McCollom and Seewald, 2001; Fu et al., 2007; McCollom, 2016). Serpentinization at lower temperatures has also been experimentally investigated using olivine and ultramafic rocks (Stevens and McKinley, 2000; Neubeck et al., 2011, 2014; Mayhew et al., 2013; Okland et al., 2014; McCollom and Donaldson, 2016). Mayhew et al. (2013) reported a rapid increase in H₂ concentration prior to 400 hours at 100°C, however, the H₂ generation mechanism in such early stages of the hydrous alteration has not been sufficiently discussed.

The mineralogical composition of starting materials is an important factor in serpentinization processes at higher temperatures (e.g., Frost and Beard, 2007; Seyfried et al., 2007; Syverson et al., 2017). Low-temperature serpentinization and associated H₂ production in nature are also affected by the mineralogy of ultramafic rocks because the dissolution of the constituent minerals controls the resultant solution chemistry (e.g., the pH and activity of aqueous silica species). Natural ultramafic rocks have often already experienced varying degrees of serpentinization due to regional metamorphism, which determines the mineral assemblage and modal proportions. However, no studies have investigated the effects of the degree of serpentinization in ultramafic rocks on hydrous alteration and H₂ generation at low temperatures. Therefore, we prepared samples of several natural ultramafic rocks with variable mineral assemblages and degrees of serpentinization, and conducted batch-type hydrous alteration experiments in neutral pH solutions at 90°C. The petrographic and geochemical features of the starting ultramafic rocks, solution chemistry, and precipitated secondary minerals during the experiments were investigated to understand the geochemical factors that control the early stages of the hydrous alteration of ultramafic rocks at low temperatures (<100°C).

The H₂ generation mechanism is difficult to be verified due to the limited H₂ yield under the slow reaction rates at a low temperature. We addressed this problem by conducting experiments using an initial material with a finer particle size than in previous studies (Stevens and McKinley, 2000; Neubeck et al., 2011, 2014; Mayhew et al., 2013; Okland et al., 2014; McCollom and Donaldson, 2016) in a simple HNO₃ solution. This experimental design might not be appropriate for estimating the H₂ generation rate in nature, but it can be used to understand the dominant dissolution/precipitation processes responsible for H₂ generation in the early stages of the system.

Materials and Methods

Sample collection and characterization

The dunite and harzburgite samples (Table 1) were collected from Konde Hill of the Soroako Ni mine, Sulawesi Island, Indonesia (samples KND-D and KND-H); the Horoman peridotite in Hokkaido, Japan (samples HRM-D and HRM-H); Wadi Hilti, northern Oman ophiolite, Oman (samples WHL-D and WHL-H); and the abandoned Wakamatsu Cr mine in Tottori Prefecture, Japan (samples WKM-D and WKM-H). The collected samples have different degrees of serpentinization.

Table 1.

Sample list of the studied ultramafic rocks

Sample name	Location	Rock type	Sample description	References
KND-D	Konde Hill of Soroako mine, Sulawesi Island, Indonesia	Dunite	Fresh	Kadarusman et al. (2004), Ito et al. (2021)
KND-H		Harzburgite	Fresh
HRM-D	Horoman peridotite, Japan	Dunite	Fresh	Takazawa et al. (1992, 2000)
HRM-H		Harzburgite	Fresh
WHL-D	Wadi Hilti, northern Oman ophiolite, Oman	Dunite	Highly serpentinized	Nicolas (1989), Boudier and Juteau (2000)
WHL-H		Harzburgite	Highly serpentinized
WKM-D	Wakamatsu mine, Japan	Dunite	Highly serpentinized	Arai (1975, 1980), Ishiwatari (1991)
WKM-H		Harzburgite	Highly serpentinized

Less-weathered and unserpentinized bedrocks of Ni laterite ores were collected from Konde Hill (Kadarusman et al., 2004; Ito et al., 2021). The Horoman peridotite samples are weakly serpentinized and were collected from the Hidaka Metamorphic Belt. The Horoman peridotite is an ultramafic massif with alpine-type peridotite intrusions that are layered (Takazawa et al., 1992, 2000). Highly serpentinized samples were collected from Wadi Hilti at the boundary between the crustal and mantle sequences on the western side of the Oman ophiolite (Nicolas, 1989; Boudier and Juteau, 2000) and from the Wakamatsu Cr mine in the Tari–Misaka ultramafic complex. The latter consists mainly of harzburgite and dunite that were subjected to contact metamorphism associated with granitic intrusions during the Late Cretaceous (Arai, 1975, 1980; Ishiwatari, 1991).

These rock samples were cut into chips and the weathered parts removed with a diamond saw. The remaining material was pulverized with an agate mortar and multi-bead shocker (PV1001; Yasui Kikai Company, Osaka, Japan). Grains with a diameter of <53 μm were sieved from the powdered samples for the whole-rock analyses. The whole-rock chemical compositions were measured by wavelength-dispersive X-ray fluorescence spectrometry (WD-XRF; MagiX PRO; Malvern PANalytical, Spectris PLC, Egham, United Kingdom) with a Rh X-ray tube on glass beads (Chikanda et al., 2019). Powdered samples (~1 g) were pre-heated in an electric furnace (EP-K-1200; Isuzu, Niigata, Japan) at 1000°C for 12 h to remove volatile components. Subsequently, 0.4 g of the pre-heated sample was placed in a Pt crucible with 4 g of Li₂B₄O₇, heated again at 1000°C for 8 min using a bead sampler (Bead and Fuse Sampler, TK-4100; AmenaTech, Tokyo, Japan) and quenched to make fused glass disks for the WD-XRF analyses.

The main minerals in the samples were identified by powder X-ray diffraction (XRD; RINT 2000; Rigaku, Tokyo, Japan) analysis with a scan speed of 2.0°/min from 5° to 70° (2θ) with a step size of 0.02° using Cu Kα radiation. The voltage was 40 kV and current was 30 mA. The XRD spectra were analyzed using the Match!3 database (Crystal Impact, Bonn, Germany).

Thin-sections were prepared from the rock chips to identify the minerals present, and were observed using an optical microscope (BX60; Olympus, Tokyo, Japan). The chemical compositions of the minerals in the thin-sections were identified by scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS; SUPRERS CAN SSX-550 and SEDX-500; Shimadzu, Kyoto, Japan).

The degree of serpentinization was quantified from the loss-on-ignition (LOI) values after heating to 700–800°C, whereby the structural water is released from serpentine. The LOI values were measured with the modified method of Tomita et al. (1992). Approximately 1 g of sample powder was pre-heated at 110°C for 2 h in an electric furnace (EP-K-1200; Isuzu, Niigata, Japan) to remove surface water and then weighed. The samples were heated again at 850°C for 2 h to remove structural water. The LOI value is the weight percent (wt.%) difference in sample weight between that pre-heated at 110°C and that heated at 850°C:

  

$$\mathrm{LOI}(\mathrm{wt.\%})=\frac{100({\mathrm{W}}_{110°\mathrm{C}}-{\mathrm{W}}_{850°\mathrm{C}})}{{\mathrm{W}}_{110°\mathrm{C}}},$$(3)

where W_T(°C) is the weight of the sample heated at a particular temperature in °C. Completely serpentinized rocks contain a maximum of ~13 wt.% structural water. The effect of other volatile constituents (e.g., CO₂ in carbonate minerals and H₂O in other phyllosilicate minerals) (Evans, 2008) was verified from petrographic observations.

Batch experiments

Prior to the experiments, 30 mL glass vials were soaked in acid (HNO₃ solution) for >6 h and then washed three times with deionized water. The dried glass vials were then wrapped in Al foil and sterilized in an electric furnace (EP-K-1200; Isuzu, Niigata, Japan) at 200°C for 12 h. Butyl rubber stoppers for the vials were boiled three times in a 0.1 M NaOH solution for 5 min and then washed three times with deionized water.

Samples of ultramafic rock powder (4 g) with a grain diameter of <53 μm, which were prepared using the same procedure as for the whole-rock analyses, were placed in an O₂-free glove box with a H₂-free N₂ atmosphere (pO₂ < 100 ppm) immediately after powdering to avoid oxidation. A NaNO₃ solution was used for the batch experiments to understand the hydrous alteration mechanisms of ultramafic rocks in a simple system where no complexes are expected to form, as predicted from thermodynamic calculations and in previous studies (e.g., the Geochemist’s Workbench with the thermo.com.V8.R6+ database; Smirnov et al., 2008; Blanc et al., 2012). A 10 mM NaNO₃ solution was prepared in the glove box using NaNO₃ (Kanto Chemical Company Incorporated, Tokyo, Japan) and O₂-free deionized water in which the gas phase was replaced by bubbling N₂ until the dissolved O₂ (DO) concentration reached <0.01 ppm at ambient temperature. The DO concentration was measured with a DO meter (LAQUA 9520-10D; Horiba, Kyoto, Japan).

The sample powders were mixed in glass vials with 20 mL of 10 mM NaNO₃ solution at a liquid/solid ratio of 5. The glass vials were sealed with butyl rubber stoppers and Al caps using a vial crimper and then gently shaken for ca. 10 s. The vials were placed upside down in an oven to minimize possible leaks of generated H₂ at 90°C for 2 weeks. The batch experiments were undertaken three times to verify the data reproducibility. A blank test was also conducted to check for any H₂ generation from the experimental equipment (e.g., the butyl rubber stoppers) as suggested by McCollom and Donaldson (2016). Batch experiments using sample HRM-H were also conducted with the addition of the following chemical reagents to investigate the effects of dissolved chemical species (Si, Ca²⁺, and Mg²⁺) on H₂ generation: fumed silica (AEROSIL^®200; Nippon Aerosil Company Limited, Tokyo, Japan), tetraethyl orthosilicate (TEOS; Tokyo Chemical Industry Company Limited, Tokyo, Japan), Ca(OH)₂ (Fujifilm Wako Pure Chemical Company, Osaka, Japan), and MgO (Kanto Chemical Company Incorporated, Tokyo, Japan). Fumed silica, Ca(OH)₂, and MgO (0.2 g) were added to the solution to adjust the initial concentrations to 5 wt.%. TEOS was added to the solutions to adjust the initial silica concentrations to 5, 10, and 50 ppm.

H₂ gas, solution, and solid analyses

Gas samples (0.5 mL) were collected from the headspace in the glass vials by inserting a syringe through the rubber stopper at room temperature after the experiments. The collected gas was introduced into a gas chromatograph equipped with a reduced gas detector (GC-RGD; Model 310C; SRI Instruments, California, USA), to measure the H₂ concentration using N₂ as a carrier gas. The detection limit for the H₂ concentration measurements was ~10 ppm. The GC-RGD was pre-heated at 80°C for 1 day prior to the measurements to stabilize the instrument. The temperatures of the column, cell, and reactor of the GC-RGD were set to 80, 170, and 290°C, respectively, with a 2.5 min retention time. The background of the obtained spectra was less than ±0.5000 mV. The H₂ concentration was calculated using a calibration curve obtained from gas measurements of 5, 10, 30, 50, 70, 100, 500, and 1000 μL of a N₂ and H₂ (H₂ = 0.930%) gas mixture. The produced H₂ is expressed as μmol of H₂ per kg of rock sample. The experiments were performed three times for each sample. The dissolved H₂ is negligible as compared with that in the gas phase (~3% of the H₂ concentration in the gas phase).

After collection of the headspace gases, 1 mL of the supernatant was filtered with a 0.45 μm polytetrafluoroethylene (PTFE) membrane filter (Merck KGaA, Darmstadt, Germany) and diluted 10 times with 0.1 mL of 1 vol.% HNO₃ (60%; Kanto Chemical Company Incorporated, Tokyo, Japan) and 8.9 mL of ultrapure water (18 MΩ∙cm) in an O₂-free glove box with a H₂-free N₂ atmosphere (pO₂ < 100 ppm). These samples were analyzed by inductively coupled plasma–atomic emission spectroscopy (ICP-AES; ICPE-9000; Shimadzu, Kyoto, Japan) to determine the solution Si, Ca, Mg, and Fe concentrations. The Si, Ca, Mg, and Fe concentrations were calibrated using 0.01, 0.05, 0.1, 0.5, 1, 5, and 10 ppm multi-element standard solutions (Merck KGaA, Darmstadt, Germany) and a Si standard solution (Fujifilm Wako Pure Chemical Company, Osaka, Japan). Residual supernatants were subjected to pH measurement at room temperature with a pH meter (LAQUA F-72; Horiba, Kyoto, Japan).

An additional scaled-up batch experiment was conducted to recover sufficient precipitates for chemical analysis and observations. These precipitates were formed during the experiments with sample HRM-H for 2 weeks, which used a larger grain size for starting sample (53–106 μm) and double the volumes of reactants, but at the same liquid/solid ratios as in the other experiments. A large difference in the particle size between the starting sample and precipitates allows them to be easily separated by centrifuging at 550 g for 3 min. The supernatants were then centrifuged at 1650 g for 35 min and freeze-dried. The recovered fine particles were placed on carbon tape, carbon-coated, and observed by field emission–scanning electron microscopy (FE-SEM; JSM-6500F; JEOL, Tokyo, Japan). The samples (5 mg) were also suspended in 1 mL of ethanol by ultrasonication for 5 min. The suspensions were then dropped onto Cu microgrids (Okenshoji Company Limited, Tokyo, Japan) and observed by transmission electron microscopy–energy dispersive X-ray spectroscopy (TEM-EDS; JEM-2010; JEOL) at 200 V with Charge Coupled Device cameras (DV300W; Gatan Incorporated, Ametek, PA and MultiScan Camera 794; Gatan Incorporated) to identify the precipitates.

Results

Characterization of the ultramafic rocks used as starting materials

Whole-rock chemical compositions

The whole-rock chemical compositions of the ultramafic rocks used in the experiments were determined by XRF spectrometry (Table 2). The dunite samples (KND-D, HRM-D, WHL-D, and WKM-D) consist of 45.5–50.3 wt.% MgO, 38.9–41.1 wt.% SiO₂, 7.84–11.2 wt.% Fe₂O₃, 0.09–1.09 wt.% Al₂O₃, 0.02–0.17 wt.% CaO, 0.11–0.16 wt.% MnO, <0.02 wt.% TiO₂, <0.01 wt.% K₂O, 0.01 wt.% P₂O₅, 0.24–0.28 wt.% Ni, and 0.26–1.25 wt.% Cr, with trace amounts of Cu (<517 ppm), Zn (79.3–157 ppm), Sc (<4.37 ppm), and V (16.5–65.0 ppm). The Na₂O contents of the dunite samples are below the detection limit. The harzburgite samples (KND-H, HRM-H, WHL-H, and WKM-H) consist of 44.0–48.5 wt.% MgO, 41.8–44.1 wt.% SiO₂, 8.18–10.1 wt.% Fe₂O₃, 0.43–0.78 wt.% Al₂O₃, 0.08–0.90 wt.% CaO, 0.12–0.15 wt.% MnO, <0.06 wt.% TiO₂, <0.02 wt.% Na₂O, <0.01 wt.% K₂O, 0.01 wt.% P₂O₅, 0.22–0.28 wt.% Ni, and 0.19–0.31 wt.% Cr, with trace amounts of Cu (<147 ppm), Zn (43.5–65.6 ppm), Sc (1.87–11.4 ppm), and V (20.2–40.7 ppm) (Table 2). The harzburgite samples tend to have higher SiO₂ and CaO contents than the dunite samples due to the presence of orthopyroxene. The LOI values are excluded from the chemical compositions in Table 2 because these are described separately in Subsubsection “Degree of serpentinization”.

Table 2.

Whole-rock chemical compositions of the ultramafic rock samples, as obtained by WD-XRF analysis

Sample name	KND-D	HRM-D	WHL-D	WKM-D	KND-H	HRM-H	WHL-H	WKM-H
wt.%
SiO2	40.5	40.1	38.9	41.1	43.5	44.1	41.8	43.2
TiO2	<d.l.	<d.l.	0.02	<d.l.	0.03	0.06	0.01	<d.l.
Al2O3	0.09	0.71	1.09	0.40	0.78	0.59	0.43	0.57
Fe2O3 (total)	7.84	11.2	9.95	8.32	8.84	8.18	10.1	9.59
MnO	0.11	0.16	0.15	0.15	0.13	0.12	0.14	0.15
MgO	45.5	47.8	47.7	50.3	44.0	45.9	48.5	46.3
CaO	0.11	0.17	0.10	0.02	0.90	0.51	0.43	0.08
Na2O	<d.l.	<d.l.	<d.l.	<d.l.	0.02	0.01	<d.l.	0.02
K2O	0.01	0.01	<d.l.	0.01	0.01	<d.l.	<d.l.	0.01
P2O5	0.01	0.01	0.01	0.01	0.01	0.01	0.01	0.01
Ni	0.28	0.26	0.28	0.24	0.23	0.22	0.28	0.27
Cr	0.26	0.34	1.25	0.26	0.31	0.22	0.31	0.19
ppm
Cu	518	29.3	3.75	<d.l.	147	<d.l.	7.09	38.7
Zn	157	79.3	82.7	115	62.7	65.6	56.2	43.5
Sc	1.13	2.59	4.37	<d.l.	11.4	1.87	5.92	3.41
V	20.9	29.0	65.0	16.5	40.7	20.2	31.4	33.9
Total	94.7	101	99.5	101	98.8	99.9	102	100

<d.l. = below detection limit.

Degree of serpentinization

Samples of the Konde Hill and Horoman peridotites (KND-D, HRM-D, KND-H, and HRM-H) have low LOI values of 0.71–2.33 wt.%, while samples from the Wadi Hilti, Oman ophiolite, and Wakamatsu mine (WHL-D, WKM-D, WHL-H, and WKM-H) have higher LOI values of 6.50–15.2 wt.% (Table 3; Fig. 1). The effect of carbonate minerals on the LOI values can be ignored, because carbonates were not detected in petrographic observations or from the XRD patterns (Subsubsection “Mineralogy”). Considering that a completely serpentinized rock contains 13 wt.% water, the samples from the Wakamatsu abandoned mine (WKM-D and WKM-H) are likely to be completely serpentinized. The LOI value of sample WKM-D exceeded the water content of a completely serpentinized rock, suggesting a contribution from other primary hydroxides such as brucite. Hereafter, we use the LOI value as a quantitative indicator of the degree of serpentinization of the starting materials. Based on the LOI value, the degree of serpentinization increases broadly in the order of samples from Konde Hill, Horoman, Wadi Hilti, and Wakamatsu.

Table 3.

Loss-on-ignition (LOI) values calculated from the weight loss after heating at 110°C and 850°C

Sample name	Sample weight after heating at 110°C (g)	Sample weight after heating at 850°C (g)	LOI (wt.%)
KND-D	1.00	0.99	1.10
HRM-D	1.01	0.99	2.33
WHL-D	0.99	0.88	11.7
WKM-D	0.99	0.84	15.2
KND-H	1.00	0.99	0.71
HRM-H	1.00	0.99	0.77
WHL-H	1.00	0.93	6.50
WKM-H	1.00	0.86	13.8

Fig. 1.

LOI values of samples KND-D, HRM-D, WHL-D, WKM-D, KND-H, HRM-H, WHL-H, and WKM-H. Red and blue bars represent dunite and harzburgite, respectively.

Mineralogy

Petrological observations of the thin-sections using an optical microscope and SEM-EDS showed that the dunite samples KND-D and HRM-D consist mainly of large grains of olivine (Fig. 2a–b), with small amounts of pentlandite, chromite, and magnetite, and rare serpentine (Table 4), which indicates the samples are relatively fresh. Olivine grains in sample HRM-D are smaller than those in sample KND-D and are partially serpentinized with mesh textures. Sample WHL-D consists mainly of olivine remaining after serpentinization (Fig. 2c), serpentine with mesh textures, and small amounts of orthopyroxene and pentlandite (Table 4). Sample WKM-D exhibits strong serpentinization (Fig. 2d) and comprises mainly serpentine with small amounts of pentlandite, magnetite, chromite, and awaruite (Fe–Ni alloy), along with olivine remaining after serpentinization and scattered pores (Table 4).

Fig. 2.

Cross-polarized light photomicrographs of the ultramafic rocks. (a) Olivine in sample KND-D. (b) Olivine in sample HRM-D. (c) Olivine and veins of serpentine in sample WHL-D. (d) Olivine and serpentine in sample WKM-D. (e) Olivine and orthopyroxene in sample KND-H. (f) Olivine and orthopyroxene in sample HRM-H. (g) Olivine, orthopyroxene, and serpentine in sample WHL-H. (h) Olivine and serpentine in sample WKM-H. (i) Pentlandite in sample HRM-D. (j) Magnetite, pentlandite, and awaruite in sample WKM-D. Ol = olivine, Srp = serpentine, Opx = orthopyroxene, Pn = pentlandite, Mag = magnetite, and Awr = awaruite.

Table 4.

Mineral assemblages of the ultramafic rock samples in the batch experiments (the number of x symbols corresponds to the mineral abundance)

Sample name	Olivine	Orthopyroxene	Clinopyroxene	Pentlandite	Chromite	Magnetite	Serpentine	Talc	Bastite	Awaruite	Minor minerals
KND-D	xxx			x	x	x					serpentine
HRM-D	xx			x	x	x	x
WHL-D	x	x		x			xxx
WKM-D				x	x	x	xxx			x	olivine
KND-H	xxx	xx	x	x	x	x
HRM-H	xxx	xx	x	x	x	x
WHL-H	x	x	x	x	x	x	xxx	x	x		awaruite
WKM-H				x	x	x	xxx	xx	x	x

x = low proportion, xx = middle proportion, xxx = high proportioin.

The harzburgite samples KND-H and HRM-H consist mainly of olivine and orthopyroxene (Fig. 2e–f) with minor amounts of clinopyroxene, pentlandite, chromite, and magnetite (Table 4). These samples do not contain serpentine, which indicates that they are relatively fresh. Sample WHL-H consists mainly of serpentine, along with olivine, orthopyroxene (Fig. 2g), clinopyroxene, pentlandite, chromite, magnetite, talc, and bastite, with small amounts of awaruite (Table 4). WKM-H consists mainly of serpentine (Fig. 2h) along with pentlandite, magnetite, chromite, talc, awaruite, and bastite (Table 4). These serpentinized harzburgite samples contain talc, probably due to high Si activity during pyroxene decomposition. Pentlandite was observed in all samples (Fig. 2i), as well as magnetite and chromite. Awaruite, an Fe–Ni alloy that is generally recognized as being a secondary product of serpentinization, was locally observed in sample WKM-D (Fig. 2j).

In the XRD spectra, sample KND-D only has peaks attributable to olivine (Fig. 3a). Sample HRM-D has major peaks attributable to olivine with minor peaks of serpentine and brucite (Fig. 3b). Sample WHL-D has a serpentine-dominant assemblage with minor olivine and brucite (Fig. 3c). Similarly, sample WKM-D also consists mainly of serpentine, with minor chlorite, brucite, and spinel (magnetite or chromite; Fig. 3d). For the harzburgite samples, XRD spectra of samples KND-H and HRM-H indicate they consist mainly of olivine with minor amounts of orthopyroxene (Fig. 3e–f). Sample WHL-H is dominated by serpentine with minor amounts of olivine, brucite, and orthopyroxene (Fig. 3g). Sample WKM-H is dominated by serpentine and talc with minor amounts of brucite, chlorite, and spinel (Fig. 3h). Peaks attributable to orthopyroxene were not observed in the XRD spectra of samples WHL-D and WKM-H, although orthopyroxene was observed petrographically, which is probably due to its low abundance in these samples. For the same reason, peaks attributable to clinopyroxene were not observed in the XRD spectra of samples KND-H, HRM-H, and WHL-H, although these samples contain a small amount of clinopyroxene (Table 4). Peaks attributable to olivine were not detected in either of the Wakamatsu samples (WKM-D and WKM-H; Fig. 3d and h), which is indicative of their intense serpentinization. These samples contain unidentifiable fine anhedral minerals.

Fig. 3.

XRD patterns of samples (a) KND-D, (b) HRM-D, (c) WHL-D, (d) WKM-D, (e) KND-H, (f) HRM-H, (g) WHL-H, and (h) WKM-H. Ol = olivine, Srp = serpentine, Brc = brucite, Chl = chlorite, Opx = orthopyroxene, and Tlc = talc.

Experimental results

Solution + ultramafic rocks

The fresh Konde Hill and Horoman peridotite samples produced more H₂ (KND-D = 99.3–120 μmol/kg, HRM-D = 81.9–103 μmol/kg, KND-H = 304–338 μmol/kg, and HRM-H = 140.2–153.8 μmol/kg) than the highly serpentinized samples from Wadi Hilti and Wakamatsu (WHL-D = 27.2–49.0 μmol/kg, WKM-D = 19.2–106 μmol/kg, WHL-H = 39.6–59.3 μmol/kg, and WKM-H = 26.3–181 μmol/kg) (Table 5; Fig. 4). The average amounts of generated H₂ in each experiment tended to decrease with increasing LOI values of the starting ultramafic rocks (i.e., degree of serpentinization; Fig. 4). Figure 4 also shows that the harzburgite samples tended to produce more H₂ than the dunite samples. In particular, sample KND-H yielded the highest amounts of generated H₂ (337.9 μmol/kg) of all the samples (Table 5; Fig. 4). Sample WKM-H had considerable variability in the amount of generated H₂, with a large standard deviation over the three experiments (Table 5; Fig. 4). Of note, H₂ was not detected in any of the blank tests (Table 5).

Table 5.

Amount of generated H₂, pH, and dissolved Si, Ca, and Mg concentrations in the recovered solutions after batch experiments for 2 weeks

Sample name	LOI	H2 (μmol/kg)	pH	Si concentration (ppm)	Ca concentration (ppm)	Mg concentration (ppm)
KND-D	1.10	99.3	8.9	4.0	5.6	44
		103	8.9	4.5	5.3	43
		120	8.1	4.0	5.4	38
HRM-D	2.33	99.8	8.7	4.7	5.7	67
		103	8.7	4.9	5.9	61
		81.9	9.0	6.2	3.1	60
WHL-D	11.7	49.0	10	0.3	3.5	7.2
		31.2	10	2.9	3.6	5.8
		27.2	10	2.0	3.5	7.1
WKM-D	15.2	19.2	11	2.0	2.7	58
		27.2	10	1.0	5.0	35
		106	10	0.9	2.9	15
KND-H	0.71	324	8.8	5.4	7.0	42
		304	8.8	5.3	7.0	39
		338	8.7	8.1	8.7	41
HRM-H	0.77	154	8.6	10	7.3	28
		145	8.4	7.0	8.5	25
		140	8.7	9.7	6.6	29
WHL-H	6.50	43.0	9.9	0.5	3.2	3.4
		39.6	10	0.8	4.1	13
		59.3	10	2.4	4.9	8.2
WKM-H	13.8	181	8.4	12	26	2.3
		26.3	10	0.3	3.1	7.2
		57.6	10	1.1	3.0	15
Blank test (NaNO3 solution only)		<detection limit	8.8	0.2	0.063	0.052

Fig. 4.

Relationship between the LOI values and amount of H₂ generated in the experiments.

The recovered solutions from the experiments with fresh samples (i.e., low LOI values; samples KND-D, HRM-D, KND-H, and HRM-H) had a pH of 8.1–9.0, while the pH of the solutions from the experiments with highly serpentinized samples (i.e., high LOI values; samples WHL-D, WKM-D, WHL-H, and WKM-H) was 8.4–10.5 (Table 5; Fig. 5a). The pH and dissolved Si concentrations in the recovered solutions are bimodal, except for sample WKM-H (Fig. 5b). The dissolved Si concentrations correlate positively with Ca concentrations (Fig. 5c), whereas the Mg concentrations have no significant correlations with pH or the Si and Ca concentrations. The Fe concentrations in the recovered solutions were below the ICP-AES detection limit (<0.2 ppm). The pH and dissolved Si, Ca, and Mg concentrations of sample WKM-H had a low reproducibility.

Fig. 5.

Solution chemistry of the recovered solutions from the batch experiments. (a) LOI versus pH, (b) pH versus Si concentration, and (c) Si versus Ca concentrations.

Solution + ultramafic rocks + chemical reagents

In the batch experiments of sample HRM-H with the addition of fumed silica, 211.0 μmol/kg of H₂ was generated at a pH of 8.21 (Table 6; Fig. 6a), which is higher than that without reagents (144.9 μmol/kg). The amounts of H₂ generated also increased from 134.9 to 231.6 μmol/kg with an increase in the TEOS addition from 5 to 50 ppm and the pH of the recovered solutions increased from 7.70 to 8.36 (Table 6; Fig. 6a). In the TEOS addition experiments, the dissolved Si and Ca concentrations in the recovered solutions were correlated positively (Fig. 6b). However, the amounts of generated H₂ decreased with the addition of Ca(OH)₂ and MgO to 16.5 and 26.5 μmol/kg, respectively (Table 6; Fig. 6b). H₂ was not detected in any of the blank tests.

Table 6.

Amount of generated H₂, pH, and dissolved Si, Ca, and Mg concentrations in the recovered solutions with the addition of fumed silica, tetraethyl orthosilicate (TEOS), Ca(OH)₂, and MgO

Added reagent	H2 (μmol/kg)	pH	Si concentration (ppm)	Ca concentration (ppm)	Mg concentration (ppm)
No reagent	144.9	8.40	7.03	8.53	24.6
+fumed silica	211.0	8.21	76.3	15.6	55.0
+TEOS 5 ppm	134.9	7.70	6.98	9.44	27.2
+TEOS 10 ppm	193.5	7.90	7.52	9.95	29.5
+TEOS 50 ppm	231.6	8.36	20.0	18.8	36.7
+Ca(OH)2	16.5	12.70	1.21	850	0.04
+MgO	26.5	10.00	1.18	5.76	44.3

Fig. 6.

(a) Amount of generated H₂ in the recovered solutions from the batch experiments using HRM-H without reagent and with the addition of fumed silica, tetraethyl orthosilicate (TEOS; 5, 10, and 50 ppm), Ca(OH)₂, and MgO. (b) Si concentration versus Ca concentration in the batch experiments using HRM-H with/without reagents.

Characterization of the precipitates

FE-SEM observations showed that the precipitates which formed during the batch experiments using sample HRM-H (no addition of chemical reagents) consisted of >10 μm-sized aggregates of needle-like and fluffy crystals (Fig. 7a). TEM observation showed tubular networks in the precipitates (Fig. 7b). The XRD pattern of the aggregates exhibited halos attributable to a 3.0 Å lattice plane spacing, which is indicative of an amorphous structure (Fig. 7c). TEM-EDS analysis showed the aggregates contained O (25.3–44.6 wt.%), Mg (21.6–37.5 wt.%), Si (2.3–32.9 wt.%), Ca (0.5–23.7 wt.%), and Na (2.0–18.0 wt.%), with small amounts of Fe (0.2–2.9 wt.%), Al (0.5–2.5 wt.%), Cr (0.1–1.1 wt.%), K (<1.7 wt.%), Cl (<1.4 wt.%), and S (<1 wt.%), which were calculated from the detected characteristic Kα X-rays (Table 7).

Fig. 7.

(a) FE-SEM image of precipitates from the batch experiment using sample HRM-H. Aggregates (>10 μm in diameter) of needle-like and fluffy crystals were observed. (b) TEM image of the precipitates with a tubular network. (c) Diffraction pattern with a halo attributed to a lattice spacing of 3.0 Å.

Table 7.

Chemical compositions of the precipitates, as determined by TEM-EDS analysis

Measurement No.	1	2	3	4	5
wt.%
O	25.8	34.8	44.6	25.3	31.9
Na	2.0	2.8	6.8	18.0	14.5
Mg	37.5	26.3	21.6	32.0	22.2
Al	2.5	1.4	0.5	0.5	0.6
Si	28.4	32.9	2.3	13.5	16.1
S	—	—	—	1.0	0.9
Cl	—	—	—	1.4	0.7
K	—	—	—	1.7	1.5
Ca	0.5	0.7	23.7	6.2	9.4
Cr	0.5	0.3	0.2	0.1	1.1
Fe	2.9	0.8	0.2	0.2	1.2
Total	100.0	100.0	100.0	100.0	100.0

Discussion

Relationship between H₂ generation and initial degrees of serpentinization

The experimental results show that the amount of generated H₂ decreases with an increasing degree of serpentinization of the starting ultramafic rock (Fig. 4). In nature, ultramafic rocks exposed at the surface have often already experienced serpentinization due to water–rock interactions during obduction of the oceanic lithosphere (Smith et al., 2006; Escartín et al., 2008). Serpentine fills the matrix of ultramafic rocks and covers the surfaces of olivine and pyroxene grains, which can prevent further serpentinization and subsequent H₂ generation (Klein et al., 2015). Although serpentine can contain small amounts of Fe²⁺ in its crystal structure, it is unlikely to be the main source of Fe²⁺ to the solution, because the dissolution rate is orders of magnitude lower than those of olivine or pyroxene at the experimental temperatures (e.g., Daval et al., 2013). This interpretation is consistent with our results; i.e., highly serpentinized ultramafic rocks generated smaller amounts of H₂ than the fresh ultramafic rocks. This further suggests that the dissolution of olivine and pyroxene is an important controlling factor on H₂ generation in the early stages of hydration of ultramafic rocks at low temperatures, which is consistent with previous serpentinization experiments (e.g., McCollom and Bach, 2009; Mayhew et al., 2013; McCollom et al., 2020). The samples that produced larger amounts of H₂ contained trace amounts of chromite and magnetite (samples KND-H and HRM-H), suggesting that the H₂ generation may have been promoted by the catalytic effect of spinel surfaces, as mentioned by Mayhew et al. (2013). H₂ generation by radical reactions (Kita et al., 1982) is four to five orders of magnitude smaller than in our experiments, and thus was negligible in our study.

The lower amounts of H₂ generated from the serpentinized samples also suggest that spinel, which are typically present in serpentinized samples, is not a dominant factor responsible for H₂ generation. Mayhew et al. (2013) suggested that spinel surface-promoted H₂ generation occurred during serpentinization at 100°C. Similarly, Geymond et al. (2023) reported that magnetite alteration can produce H₂ at low temperatures. In our study, samples WKM-D and WKM-H have high LOI values (Table 3) and do not contain detectable amounts of olivine or pyroxene (Fig. 3d and h), indicating the near-complete serpentinization of the rocks before the experiments. Instead, these samples contain spinel (Figs. 2j and 3d, h; Table 4). H₂ generation by the spinel-bearing reaction could have occurred in experiments on these serpentinized rock samples. However, the generated amounts of H₂ in these rocks was rather low (Fig. 4). Sample WKM-D had results with a poor reproducibility, but generated a high amount of H₂ (181 μmol/kg) in one batch experiment. This sample contains a trace amount of awaruite, which is a Fe–Ni alloy (Fig. 2j) that can act as a reductant and/or catalyst for H₂ generation (Sleep et al., 2004). Previous studies have suggested that Fe²⁺-containing brucite could be a Fe²⁺ source for H₂ generation at low temperatures (Beard and Frost, 2016; Miller et al., 2017). These minerals may have contributed to the moderate amounts of H₂ generation in the experiments with samples WHL-D and WHL-H (Fig. 4).

Harzburgite vs. dunite

The fresh (i.e., low LOI values) harzburgite samples produced higher amounts of H₂ (up to 322.1 μmol/kg) than the other samples used in the experiments (Fig. 4). Fresh samples (KND-D, KND-H, HRM-D, and HRM-H) tended to yield solutions with lower pH (8.1–9.0; Fig. 5a), as well as higher Si and Ca concentrations than the serpentinized samples (Fig. 5b–c). The higher Si and Ca concentrations of the resultant solutions suggest a greater contribution from clinopyroxene dissolution in the fresh samples because Ca is mainly hosted by clinopyroxene (Table 4). The results obtained using harzburgite samples tended to have higher Si and Ca concentrations in the resultant solutions than dunite samples from the same locality (i.e., KND and HRM) (Fig. 5c), which supports this interpretation. Previous studies have shown that the dissolution rate of pyroxene is about two orders of magnitude slower than that of olivine under our experimental conditions at lower pH values. However, the difference in dissolution rates between pyroxene and olivine decreases as the pH increases (Knauss et al., 1993; Oelkers and Schott, 2001; Golubev et al., 2005; Hänchen et al., 2006; Rimstidt et al., 2012) (Supplementary Fig. 1a). The trend becomes more pronounced as the temperature decreases (Supplementary Fig. 1b). Considering the errors of each experiment in terms of olivine/pyroxene dissolution rates (e.g., ~0.5 log units for the forsterite dissolution rate; Rimstidt et al., 2012), at higher pH, the dissolution of pyroxene can contribute to H₂ generation as well as olivine dissolution. The dissolution rates of olivine and pyroxene also depend on the surface area of the particles. The initial particle size of the powdered samples in the experiment was adjusted by sieving, however, differences in the pattern of the fragmentation between olivine and pyroxene may have led to variations in surface area.

Precipitates were recovered from one batch experiment using fresh harzburgite (sample HRM-H) with a larger grain size and by doubling the volumes of the reactants. TEM-EDS analysis of the precipitates showed that they consist mainly of Mg, Si, and O (Table 7). The morphology and electron diffraction patterns from the TEM observations (Fig. 7a–c) resemble those of magnesium silicate hydrates (M-S-H) reported by Lothenbach et al. (2015), although the diffraction pattern with a halo attributed to a 3.0 Å lattice plane spacing is not identical to that of pure M_xS_yH(MgO)_x(SiO₂)_y(H₂O)_z that has halos due to lattice plane spacings of 1.54, 2.57, 3.34, and 4.51 Å (Nied et al., 2016; Nishiki et al., 2023). A previous study also reported precipitation of M-S-H in diopside dissolution experiments in an acidic solution (Monasterio-Guillot et al., 2021), which was interpreted to form at a high pH (>8) on the surface of diopside grains.

The high Si concentrations in the solution were favorable for the precipitation of M-S-H (Lothenbach et al., 2015). Pyroxene can more-effectively supply Si to the solution than olivine based on the stoichiometry: Si/(Mg + Fe) = 0.5 and 1.0 for olivine and pyroxene, respectively. This may have contributed to the precipitation of M-S-H in our experiments. M-S-H formation under Si-rich conditions is described by the following reaction (modified from Nied et al., 2016):

  

$$3{\mathrm{Mg}}^{2+}+4{\mathrm{SiO}}_2+8{\mathrm{H}}_2\mathrm{O}\to (\mathrm{MgO}{)}_3({\mathrm{SiO}}_2{)}_4({\mathrm{H}}_2\mathrm{O}{)}_5+6{\mathrm{H}}^{+}.$$(4)

The precipitation of M-S-H releases protons and buffers the solution pH at 8.5–10.5 with variable Mg/Si ratios (Nied et al., 2016), which is consistent with our experimental results (Fig. 5a).

The addition of Si as either fumed silica or TEOS enhanced H₂ generation in the experiments using fresh harzburgite (sample HRM-H) (Table 6; Fig. 6a), whereas the addition of Ca(OH)₂ and MgO reduced the amount of generated H₂. The amount of generated H₂ increased with the amount of TEOS (Fig. 6a), indicating that addition of Si promoted H₂ generation. The positive correlation between Si and Ca concentrations with increasing pH indicates the dissolution of clinopyroxene (Table 6; Fig. 6b). Considering that high-Si conditions are favorable for M-S-H precipitation (Lothenbach et al., 2015), H₂ generation is likely linked to precipitation of M-S-H. The dissolution rates of olivine and pyroxene decrease with increasing pH (Supplementary Fig. 1). The pH buffering effect by M-S-H precipitation may prevent a decrease in the dissolution rate of olivine and pyroxene.

The Fe²⁺/Fe³⁺ ratio of the M-S-H-like precipitates could not be determined by electron energy loss spectroscopy due to the low concentrations in our experiments. The formation of low-crystallinity Mg-bearing silicate at ambient temperatures in natural alkaline environments has been reported in recent studies (Nishiki et al., 2020; Shimbashi et al., 2022), which are metastable relative to crystalline Mg-bearing silicates (e.g., serpentine and talc). A recent study also indicated Fe³⁺ can be incorporated in M-S-H (Marsiske et al., 2021), which might have occurred in our experiments.

Controls on solution chemistry by M-S-H and brucite

The thermodynamic data for M-S-H (M₃S₂H₅ and M₃S₄H₅) from Nied et al. (2016) are incorporated into the Thermoddem database (Blanc et al., 2012). A stability diagram of the M-S-H system constructed using the database shows that the chemical compositions of the recovered solutions plot on the phase boundary of H₄SiO_4(aq) and M₃S₄H₅ when samples KND-D, HRM-D, KND-H, and HRM-H were used as the starting materials (Fig. 8a). Although FE-SEM and TEM observations of the precipitates were only conducted in one batch experiment (HRM-H), the chemical compositions of the solutions plotted on the diagram suggest that the solution chemistry was buffered by M-S-H when fresh ultramafic rock samples were used.

Fig. 8.

(a) Log activity of the Mg²⁺/H₂–H₄SiO_4(aq) stability fields of possible minerals in the MgO–SiO₂–H₂O system at 25°C, with the chemical compositions of the recovered solutions from the batch experiments using samples KND-D, HRM-D, KND-H, and HRM-H. (b) pH–log activity diagram of the Mg²⁺ stability fields at 25°C, with the chemical compositions of the recovered solutions from the batch experiments using samples WHL-D, WKM-D, WHL-H, and WKM-H. The diagrams were calculated using the Act2 module of the Geochemist’s Workbench 10.0 (Aqueous Solutions LLC, Champaign, IL) with the Thermoddem database (Blanc et al., 2012) modified by adding thermodynamic data for M₃S₄H₅ and M₃S₂H₅ from Nied et al. (2016). Note that the thermodynamic calculations were performed at 25°C to account for the analysis temperature of the recovered samples after the experiments.

The chemical compositions of solutions from the WHL-D, WKM-D, WHL-H, and WKM-H samples plot on the phase boundary of Mg²⁺ and brucite, which suggests the solution chemistry was controlled by brucite (Fig. 8b). The starting ultramafic rocks contain brucite (Table 4; Fig. 3), which can regulate the solution chemistry at a higher pH in the experiments with serpentinized rock samples. Undetectable brucite may have also precipitated, although brucite precipitation was not observed in the recovered materials after the batch experiments. Previous studies have suggested that precipitation of brucite suppresses H₂ generation during low-temperature serpentinization (e.g., McCollom and Bach, 2009), which is consistent with our results that indicate a lower amount of H₂ is generated when serpentinized ultramafic rocks are used as the starting material.

The amounts of H₂ generated in our experiments (Table 5; Fig. 4) were higher than those in previous studies (e.g., McCollom and Donaldson, 2016). In general, there is an initial rapid reaction followed by much slower reaction rates in serpentinization experiments (e.g., Mayhew et al., 2013) due to the presence of reactive surfaces on the constituent minerals caused by sample pulverization. Compared with the reaction rates for serpentinization obtained by Leong et al. (2023), our reaction rates for H₂ generation were two to three orders of magnitude higher than those expected from the grain size. The reaction rates in our experiments are not realistic for estimates of long-term H₂ generation from ultramafic rocks in nature, given the experimental design.

Dissolution of ultramafic rocks at low temperatures

Based on the above discussion, we now consider the dominant dissolution processes of ultramafic rocks during the early stages of hydration at low temperatures (Fig. 9). When fresh ultramafic rocks are used as starting materials, olivine and pyroxene are dissolved in an O₂-free solution, which causes a pH increase. With an effective supply of Si from pyroxene dissolution, Si activity reaches saturation in terms of M-S-H, which results in M-S-H precipitation with H₂ generation caused by the oxidation of dissolved Fe²⁺. The further increase of the pH is then inhibited by buffering of the M-S-H precipitation, which leads to the dissolution of olivine and pyroxene (Fig. 9a). In this cycle, the pH is kept between 8.1 and 9.0 by the M-S-H buffer.

Fig. 9.

Mechanisms of dissolution of (a) fresh ultramafic rocks and (b) serpentinized ultramafic rocks at low temperatures.

However, the serpentinized ultramafic rocks contain smaller amounts of pyroxene and olivine than the fresh ultramafic rocks. The Si activity in the solutions does not increase sufficiently due to the dissolution of pyroxene and olivine to precipitate M-S-H. Instead, the solution reaches saturation for brucite, which buffers the pH at ~10.5 (Pokrovsky and Schott, 2004; Zhang et al., 2011) and does not promote further dissolution of serpentinized ultramafic rock (Fig. 9b).

These processes are distinct from serpentinization at higher temperatures where pyroxene decreases the amount of generated H₂ with the release of Si (e.g., Huang et al., 2021). Previous studies of high-temperature serpentinization (e.g., Frost and Beard, 2007) suggested that Si-rich solutions inhibit H₂ generation due to favorable conditions for the precipitation of crystalline silicate minerals containing Mg²⁺ and Fe²⁺, such as serpentine and talc, which have low rates of dissolution/precipitation and thus may not be able to buffer the pH. However, at the lower temperatures in our experiments, M-S-H, which is a metastable phase of serpentine or talc, controlled the dissolution/precipitation rates and enabled buffering of the solution pH.

Conclusions

Hydrous alteration experiments of natural ultramafic rocks with variable degrees of serpentinization were conducted to understand the H₂ generation mechanism during the early stages of the reaction at low temperatures (90°C, 2 weeks). The degree of serpentinization was estimated from the LOI values, which is consistent with the petrographic observations and XRD analyses. The amounts of generated H₂ increased with decreasing LOI values of the starting ultramafic rocks, indicative of dissolution of olivine and pyroxene, which were the key contributors to H₂ generation. Fresh harzburgite produced more H₂ than the other samples. The solution chemistry in the resultant solutions showed that the pyroxene dissolution contributed to the H₂ generation at higher pH and could supply Fe²⁺ for H₂ generation and Si to the solutions. Precipitation of M-S-H-like materials was recognized in the experiments using fresh harzburgite. Given the increase in H₂ generation with Si addition in the experiments, M-S-H precipitation in Si-rich solutions likely controlled the solution chemistry and H₂ generation from fresh harzburgite.

Thermodynamic calculations indicate that dissolution of ultramafic rocks at low temperatures is controlled by M-S-H or brucite. Harzburgite was more soluble and favorable for H₂ generation than dunite in our experimental system, which was likely due to the contribution of dissolved Si from pyroxene and subsequent M-S-H precipitation.

Acknowledgments

The authors thank K. Nakamura and H. Nomura of the Thin-Section Laboratory, Prototype Machining Solution Division, Global Facility Center, Hokkaido University, for preparation of the thin-sections, and the technical staff of the Nano-Micro Material Analysis Laboratory, Joint-Use Facilities, Faculty of Engineering, Hokkaido University, for support with the FE-SEM and TEM-EDS observations. This research was supported by the Nanotechnology Platform program of the Ministry of Education, Culture, Sports, Science, and Technology (MEXT), Japan, and by KAKENHI funds (Grants 17K19081 and 22K18325) from the Japan Society for the Promotion of Science (JSPS) to T. O.

References

•  Arai,  S. (1975) Contact metamorphosed dunite-harzburgite complex in the Chugoku district, western Japan. Contrib. Mineral. Petrol. 52, 1–16. https://doi.org/10.1007/BF00377998
•  Arai,  S. (1980) Dunite–harzburgite–chromitite complexes as refractory residue in the Sangun–Yamaguchi zone, western Japan. J. Petrol. 21, 141–165. https://doi.org/10.1093/petrology/21.1.141
•  Beard,  J. S. and  Frost,  B. R. (2016) The stoichiometric effects of ferric iron substitutions in serpentine from microprobe data. Int. Geol. Rev. 59, 541–547. https://doi.org/10.1080/00206814.2016.1197803
•  Blanc,  P.,  Lassin,  A.,  Piantone,  P.,  Azaroual,  M.,  Jacquemet,  N.,  Fabbri,  A. and  Gaucher,  E. C. (2012) Thermoddem: A geochemical database focused on low temperature water/rock interactions and waste materials. Appl. Geochem. 27, 2107–2116. https://doi.org/10.1016/j.apgeochem.2012.06.002
•  Boudier,  F. and  Juteau,  T. (2000) The ophiolite of Oman and United Arab Emirates. Mar. Geophys. Res. 21, 145–407. https://doi.org/10.1023/A:1026754209262
•  Cannat,  M.,  Fontaine,  F. and  Escartin,  J. (2010) Serpentinization and associated hydrogen and methane fluxes at slow spreading ridges. Diversity of Hydrothermal Systems on Slow Spreading Ocean Ridges (Rona, P. A., Devey, C. W., Dyment, J. and Murton, B. J., eds.), 241–264, American Geophysical Union. https://doi.org/10.1029/2008GM000760
•  Chikanda,  F.,  Otake,  T.,  Ohtomo,  Y.,  Ito,  A.,  Yokoyama,  T. D. and  Sato,  T. (2019) Magmatic-hydrothermal processes associated with rare earth element enrichment in the Kangankunde carbonatite complex, Malawi. Minerals 9, 442. https://doi.org/10.3390/min9070442
•  Daval,  D.,  Hellmann,  R.,  Martinez,  I.,  Gangloff,  S. and  Guyot,  F. (2013) Lizardite serpentine dissolution kinetics as a function of pH and temperature, including effects of elevated pCO₂. Chem. Geol. 351, 245–256. https://doi.org/10.1016/j.chemgeo.2013.05.020
•  Escartín,  J.,  Smith,  D. K.,  Cann,  J.,  Schouten,  H.,  Langmuir,  C. H. and  Escrig,  S. (2008) Central role of detachment faults in accretion of slow-spreading oceanic lithosphere. Nature 455, 790–794. https://doi.org/10.1038/nature07333
•  Etiope,  G.,  Schoell,  M. and  Hosgörmez,  H. (2011) Abiotic methane flux from the Chimaera seep and Tekirova ophiolites (Turkey): Understanding gas exhalation from low temperature serpentinization and implications for Mars. Earth Planet. Sci. Lett. 310, 96–104. https://doi.org/10.1016/j.epsl.2011.08.001
•  Evans,  B. W. (2008) Control of the products of serpentinization by the Fe²⁺Mg_–1 exchange potential of olivine and orthopyroxene. J. Petrol. 49, 1873–1887. https://doi.org/10.1093/petrology/egn050
•  Frost,  R. B. and  Beard,  J. S. (2007) On silica activity and serpentinization. J. Petrol. 48, 1351–1368. https://doi.org/10.1093/petrology/egm021
•  Fu,  Q.,  Lollar,  B. S.,  Horita,  J.,  Lacrampe-Couloume,  G. and  Seyfried,  W. E. (2007) Abiotic formation of hydrocarbons under hydrothermal conditions: Constraints from chemical and isotope data. Geochim. Cosmochim. Acta 71, 1982–1998. https://doi.org/10.1016/j.gca.2007.01.022
•  Geymond,  U.,  Briolet,  T.,  Combaudon,  V.,  Sissmann,  O.,  Martinez,  I.,  Duttine,  M. and  Moretti,  I. (2023) Reassessing the role of magnetite during natural hydrogen generation. Front. Earth Sci. 11, 1169356. https://doi.org/10.3389/feart.2023.1169356
•  Glein,  C. R.,  Baross,  J. A. and  Waite,  J. H. (2015) The pH of Enceladus’ ocean. Geochim. Cosmochim. Acta 162, 202–219. https://doi.org/10.1016/j.gca.2015.04.017
•  Golubev,  S. V.,  Pokrovsky,  O. S. and  Schott,  J. (2005) Experimental determination of the effect of dissolved CO₂ on the dissolution kinetics of Mg and Ca silicates at 25°C. Chem. Geol. 217, 227–238. https://doi.org/10.1016/j.chemgeo.2004.12.011
•  Greenberger,  R. N.,  Mustard,  J. F.,  Cloutis,  E. A.,  Pratt,  L. M.,  Sauer,  P. E.,  Mann,  P.,  Turner,  K.,  Dyar,  M. D. and  Bish,  D. L. (2015) Serpentinization, iron oxidation, and aqueous conditions in an ophiolite: Implications for hydrogen production and habitability on Mars. Earth Planet. Sci. Lett. 416, 21–34. https://doi.org/10.1016/j.epsl.2015.02.002
•  Hänchen,  M.,  Prigiobbe,  V.,  Storti,  G.,  Seward,  T. M. and  Mazzotti,  M. (2006) Dissolution kinetics of forsteritic olivine at 90–150°C including effects of the presence of CO₂. Geochim. Cosmochim. Acta 70, 4403–4416. https://doi.org/10.1016/j.gca.2006.06.1560
•  Huang,  R.,  Ding,  X.,  Sun,  W. and  Shang,  X. (2021) Contrasted effect of spinel and pyroxene on molecular hydrogen (H₂) production during serpentinization of olivine. Minerals 11, 794. https://doi.org/10.3390/min11080794
•  Ishiwatari,  A. (1991) Time-space distribution and petrologic diversity of Japanese ophiolites. Ophiolite Genesis and Evolution of the Oceanic Lithosphere (Peters, T., Nicolas, A., Coleman, R. G., eds.), 723–743, Springer Dordrecht.
•  Ito,  A.,  Otake,  T.,  Maulana,  A.,  Sanematsu,  K.,  Sufriadin, and  Sato,  T. (2021) Geochemical constraints on the mobilization of Ni and critical metals in laterite deposits, Sulawesi, Indonesia: A mass-balance approach. Resour. Geol. 71, 255–282. https://doi.org/10.1111/rge.12266
•  Kadarusman,  A.,  Miyashita,  S.,  Maruyama,  S.,  Parkinson,  C. D. and  Ishikawa,  A. (2004) Petrology, geochemistry and paleogeographic reconstruction of the East Sulawesi Ophiolite, Indonesia. Tectonophysics 392, 55–83. https://doi.org/10.1016/j.tecto.2004.04.008
•  Kelley,  D. S.,  Karson,  J. A.,  Blackman,  D. K.,  Früh-Green,  G. L.,  Butterfield,  D. A.,  Lilley,  M. D.,  Olson,  E. J.,  Schrenk,  M. O.,  Roe,  K. K.,  Lebon,  G. T.,  Rivizzigno,  P. and the AT3-60 Shipboard Party (2001) An off-axis hydrothermal vent field near the Mid-Atlantic Ridge at 30°N. Nature 412, 145–149. https://doi.org/10.1038/35084000
•  Kita,  I.,  Matsuo,  S. and  Wakita,  H. (1982) H₂ generation by reaction between H₂O and crushed rock: An experimental study on H₂ degassing from the active fault zone. J. Geophys. Res. 87, 10789–10795. https://doi.org/10.1029/JB087iB13p10789
•  Klein,  F.,  Grozeva,  N. G.,  Seewald,  J. S.,  McCollom,  T. M.,  Humphris,  S. E.,  Moskowitz,  B.,  Berquó,  T. S. and  Kahl,  W. A. (2015) Fluids in the crust. Experimental constraints on fluid-rock reactions during incipient serpentinization of harzburgite. Am. Mineral. 100, 991–1002. https://doi.org/10.2138/am-2015-5112
•  Knauss,  K. G.,  Nguyen,  S. N. and  Weed,  H. C. (1993) Diopside dissolution kinetics as a function of pH, CO₂, temperature, and time. Geochim. Cosmochim. Acta 57, 285–294. https://doi.org/10.1016/0016-7037(93)90431-U
•  Lang,  S. Q.,  Butterfield,  D. A.,  Schulte,  M.,  Kelley,  D. S. and  Lilley,  M. D. (2010) Elevated concentrations of formate, acetate and dissolved organic carbon found at the Lost City hydrothermal field. Geochim. Cosmochim. Acta 74, 941–952. https://doi.org/10.1016/j.gca.2009.10.045
•  Leong,  J. A.,  Nielsen,  M.,  McQueen,  N.,  Karolytė,  R.,  Hillegonds,  D. J.,  Ballentine,  C.,  Darrah,  T.,  McGillis,  W. and  Kelemen,  P. (2023) H₂ and CH₄ outgassing rates in the Samail ophiolite, Oman: Implications for low-temperature, continental serpentinization rates. Geochim. Cosmochim. Acta 347, 1–15. https://doi.org/10.1016/j.gca.2023.02.008
•  Lollar,  B. S.,  Onstott,  T. C.,  Lacrampe-Couloume,  G. and  Ballentine,  C. J. (2014) The contribution of the Precambrian continental lithosphere to global H₂ production. Nature 516, 379–382. https://doi.org/10.1038/nature14017
•  Lothenbach,  B.,  Nied,  D.,  L’Hôpital,  E.,  Achiedo,  G. and  Dauzères,  A. (2015) Magnesium and calcium silicate hydrates. Cem. Concr. Res. 77, 60–68. https://doi.org/10.1016/j.cemconres.2015.06.007
•  Marsiske,  M. R.,  Debus,  C.,  Di Lorenzo,  F.,  Bernard,  E.,  Churakov,  S. V. and  Ruiz-Agudo,  C. (2021) Immobilization of (aqueous) cations in low pH M-S-H cement. Appl. Sci. 11, 2968. https://doi.org/10.3390/app11072968
•  Martin,  W. and  Russell,  M. J. (2007) On the origin of biochemistry at an alkaline hydrothermal vent. Philos. Trans. R. Soc. B, Biol. Sci. 362, 1887–1926. https://doi.org/10.1098/rstb.2006.1881
•  Mayhew,  L. E.,  Ellison,  E. T.,  McCollom,  T. M.,  Trainor,  T. P. and  Templeton,  A. S. (2013) Hydrogen generation from low-temperature water–rock reactions. Nature Geosci. 6, 478–484. https://doi.org/10.1038/ngeo1825
•  McCollom,  T. M. (2016) Abiotic methane formation during experimental serpentinization of olivine. Proc. Natl. Acad. Sci. U.S.A 113, 13965–13970. https://doi.org/10.1073/pnas.1611843113
•  McCollom,  T. M. and  Bach,  W. (2009) Thermodynamic constraints on hydrogen generation during serpentinization of ultramafic rocks. Geochim. Cosmochim. Acta 73, 856–875. https://doi.org/10.1016/j.gca.2008.10.032
•  McCollom,  T. M. and  Donaldson,  C. (2016) Generation of hydrogen and methane during experimental low-temperature reaction of ultramafic rocks with water. Astrobiology 16, 389–406. https://doi.org/10.1089/ast.2015.1382
•  McCollom,  T. M. and  Seewald,  J. S. (2001) A reassessment of the potential for reduction of dissolved CO₂ to hydrocarbons during serpentinization of olivine. Geochim. Cosmochim. Acta 65, 3769–3778. https://doi.org/10.1016/S0016-7037(01)00655-X
•  McCollom,  T. M.,  Klein,  F.,  Moskowitz,  B.,  Berquó,  T. S.,  Bach,  W. and  Templeton,  A. S. (2020) Hydrogen generation and iron partitioning during experimental serpentinization of an olivine-pyroxene mixture. Geochim. Cosmochim. Acta 282, 55–75. https://doi.org/10.1016/j.gca.2020.05.016
•  McCollom,  T. M.,  Klein,  F. and  Ramba,  M. (2022) Hydrogen generation from serpentinization of iron-rich olivine on Mars, icy moons, and other planetary bodies. Icarus 372, 114754. https://doi.org/10.1016/j.icarus.2021.114754
•  Ménez,  B.,  Pisapia,  C.,  Andreani,  M.,  Jamme,  F.,  Vanbellingen,  Q. P.,  Brunelle,  A.,  Richard,  L.,  Dumas,  P. and  Réfrégiers,  M. (2018) Abiotic synthesis of amino acids in the recesses of the oceanic lithosphere. Nature 564, 59–63. https://doi.org/10.1038/s41586-018-0684-z
•  Merdith,  A. S.,  del Real,  P. G.,  Daniel,  I.,  Andreani,  M.,  Wright,  N. M. and  Coltice,  N. (2020) Pulsated global hydrogen and methane flux at mid-ocean ridges driven by Pangea breakup. Geochem. Geophys. Geosys. 21, e2019GC008869. https://doi.org/10.1029/2019GC008869
•  Miller,  H. M.,  Mayhew,  L. E.,  Ellison,  E. T.,  Kelemen,  P.,  Kubo,  M. and  Templeton,  A. S. (2017) Low temperature hydrogen production during experimental hydration of partially-serpentinized dunite. Geochim. Cosmochim. Acta 209, 161–183. https://doi.org/10.1016/j.gca.2017.04.022
•  Monasterio-Guillot,  L.,  Rodriguez-Navarro,  C. and  Ruiz-Agudo,  E. (2021) Kinetics and mechanisms of acid-pH weathering of pyroxenes. Geochem. Geophys. Geosyst. 22, e2021GC009711. https://doi.org/10.1029/2021GC009711
•  Neal,  C. and  Stanger,  G. (1983) Hydrogen generation from mantle source rocks in Oman. Earth Planet. Sci. Lett. 66, 315–321. https://doi.org/10.1016/0012-821X(83)90144-9
•  Neubeck,  A.,  Duc,  N. T.,  Bastviken,  D.,  Crill,  P. and  Holm,  N. G. (2011) Formation of H₂ and CH₄ by weathering of olivine at temperatures between 30 and 70°C. Geochem. Trans. 12, 6. https://doi.org/10.1186/1467-4866-12-6
•  Neubeck,  A.,  Duc,  N. T.,  Hellevang,  H.,  Oze,  C.,  Bastviken,  D.,  Bacsik,  Z. and  Holm,  N. G. (2014) Olivine alteration and H₂ production in carbonate-rich, low temperature aqueous environments. Planet. Space Sci. 96, 51–61. https://doi.org/10.1016/j.pss.2014.02.014
•  Nicolas,  A. (1989) Structure of Ophiolites and Dynamics of the Oceanic Lithosphere. Springer, 384 pp.
•  Nied,  D.,  Enemark-Rasmussen,  K.,  L’Hopital,  E.,  Skibsted,  J. and  Lothenbach,  B. (2016) Properties of magnesium silicate hydrates (M-S-H). Cem. Concr. Res. 79, 323–332. https://doi.org/10.1016/j.cemconres.2015.10.003
•  Nishiki,  Y.,  Sato,  T.,  Katoh,  T.,  Otake,  T. and  Kikuchi,  R. (2020) Precipitation of magnesium silicate hydrates in natural alkaline surface environments. Clay Sci. 24, 1–13. https://doi.org/10.11362/jcssjclayscience.24.1_1
•  Nishiki,  Y.,  Cama,  J.,  Otake,  T.,  Kikuchi,  R.,  Shimbashi,  M. and  Sato,  T. (2023) Formation of magnesium silicate hydrate (M-S-H) at pH 10 and 50°C in open-flow systems. Appl. Geochem. 148, 105544. https://doi.org/10.1016/j.apgeochem.2022.105544
•  Oelkers,  E. H. and  Schott,  J. (2001) An experimental study of enstatite dissolution rates as a function of pH, temperature, and aqueous Mg and Si concentration, and the mechanism of pyroxene/pyroxenoid dissolution. Geochim. Cosmochim. Acta 65, 1219–1231. https://doi.org/10.1016/S0016-7037(00)00564-0
•  Okland,  I.,  Huang,  S.,  Thorseth,  I. H. and  Pedersen,  R. B. (2014) Formation of H₂, CH₄ and N-species during low-temperature experimental alteration of ultramafic rocks. Chem. Geol. 387, 22–34. https://doi.org/10.1016/j.chemgeo.2014.08.003
•  Pokrovsky,  O. S. and  Schott,  J. (2004) Experimental study of brucite dissolution and precipitation in aqueous solutions: Surface speciation and chemical affinity control. Geochim. Cosmochim. Acta 68, 31–45. https://doi.org/10.1016/S0016-7037(03)00238-2
•  Preiner,  M.,  Xavier,  J. C.,  Sousa,  F. L.,  Zimorski,  V.,  Neubeck,  A.,  Lang,  S. Q.,  Greenwell,  H. C.,  Kleinermanns,  K.,  Tüysüz,  H.,  McCollom,  T. M.,  Holm,  N. G. and  Martin,  W. F. (2018) Serpentinization: Connecting geochemistry, ancient metabolism and industrial hydrogenation. Life 8, 41. https://doi.org/10.3390/life8040041
•  Preiner,  M.,  Igarashi,  K.,  Muchowska,  K. B.,  Yu,  M.,  Varma,  S. J.,  Kleinermanns,  K.,  Nobu,  M. K.,  Kamagata,  Y.,  Tüysüz,  H.,  Moran,  J. and  Martin,  W. F. (2020) A hydrogen-dependent geochemical analogue of primordial carbon and energy metabolism. Nat. Ecol. Evol. 4, 534–542. https://doi.org/10.1038/s41559-020-1125-6
•  Rimstidt,  J. D.,  Brantley,  S. L. and  Olsen,  A. A. (2012) Systematic review of forsterite dissolution rate data. Geochim. Cosmochim. Acta 99, 159–178. https://doi.org/10.1016/j.gca.2012.09.019
•  Schulte,  M.,  Blake,  D.,  Hoehler,  T. and  McCollom,  T. (2006) Serpentinization and its implications for life on the early Earth and Mars. Astrobiology 6, 364–376. https://doi.org/10.1089/ast.2006.6.364
•  Seyfried Jr.,  W. E.,  Foustoukos,  D. I. and  Fu,  Q. (2007) Redox evolution and mass transfer during serpentinization: An experimental and theoretical study at 200°C, 500 bar with implications for ultramafic-hosted hydrothermal systems at mid-ocean ridges. Geochim. Cosmochim. Acta 71, 3872–3886. https://doi.org/10.1016/j.gca.2007.05.015
•  Shimbashi,  M.,  Yokoyama,  S.,  Kikuchi,  R.,  Otake,  T. and  Sato,  T. (2022) Characteristics and formation pathways of iron- and magnesium-silicate-hydrates and smectites under natural alkaline conditions. Clays Clay Miner. 70, 492–513. https://doi.org/10.1007/s42860-022-00197-2
•  Sleep,  N. H.,  Meibom,  A.,  Fridriksson,  T.,  Coleman,  R. G. and  Bird,  D. K. (2004) H₂-rich fluids from serpentinization: Geochemical and biotic implications. Proc. Natl. Acad. Sci. U.S.A. 101, 12818–12823. https://doi.org/10.1073/pnas.0405289101
•  Smirnov,  A.,  Hausner,  D.,  Laffers,  R.,  Strongin,  D. R. and  Schoonen,  M. A. (2008) Abiotic ammonium formation in the presence of Ni-Fe metals and alloys and its implications for the Hadean nitrogen cycle. Geochem. Trans. 9, 5. https://doi.org/10.1186/1467-4866-9-5
•  Smith,  D. K.,  Cann,  J. R. and  Escartín,  J. (2006) Widespread active detachment faulting and core complex formation near 13°N on the Mid-Atlantic Ridge. Nature 442, 440–443. https://doi.org/10.1038/nature04950
•  Stevens,  T. O. and  McKinley,  J. P. (2000) Abiotic controls on H₂ production from basalt–water reactions and implications for aquifer biogeochemistry. Environ. Sci. Technol. 34, 826–831. https://doi.org/10.1021/es990583g
•  Suda,  K.,  Ueno,  Y.,  Yoshizaki,  M.,  Nakamura,  H.,  Kurokawa,  K.,  Nishiyama,  E.,  Yoshino,  K.,  Hongoh,  Y.,  Kawachi,  K.,  Omori,  S.,  Yamada,  K.,  Yoshida,  N. and  Maruyama,  S. (2014) Origin of methane in serpentinite-hosted hydrothermal systems: The CH₄–H₂–H₂O hydrogen isotope systematics of the Hakuba Happo hot spring. Earth Planet. Sci. Lett. 386, 112–125. http://dx.doi.org/10.1016/j.epsl.2013.11.001
•  Syverson,  D. D.,  Tutolo,  B. M.,  Borrok,  D. M. and  Seyfried Jr.,  W. E. (2017) Serpentinization of olivine at 300°C and 500 bars: An experimental study examining the role of silica on the reaction path and oxidation state of iron. Chem. Geol. 475, 122–134. https://doi.org/10.1016/j.chemgeo.2017.11.006
•  Takazawa,  E.,  Frey,  F.,  Shimizu,  N.,  Obata,  M.,  Bodinier,  J. L. (1992) Geochemical evidence for melt migration and reaction in the upper mantle. Nature 359, 55–58. https://doi.org/10.1038/359055a0
•  Takazawa,  E.,  Frey,  F. A.,  Shimizu,  N. and  Obata,  M. (2000) Whole rock compositional variations in an upper mantle peridotite (Horoman, Hokkaido, Japan): Are they consistent with a partial melting process? Geochim. Cosmochim. Acta 64, 695–716. https://doi.org/10.1016/S0016-7037(99)00346-4
•  Tomita,  T.,  Tanabe,  K. and  Yamamoto,  K. (1992) Determination of clay minerals by ignition loss methods with a muffle furnace. Proceedings Jpn. Soc. Civ. Eng. 445/III-18, 55–63. (Japanese with English abstract).
•  Zgonnik,  V. (2020) The occurrence and geoscience of natural hydrogen: A comprehensive review. Earth Sci. Rev. 203, 103140. https://doi.org/10.1016/j.earscirev.2020.103140
•  Zhang,  T.,  Cheeseman,  C. R. and  Vandeperre,  L. J. (2011) Development of low pH cement systems forming magnesium silicate hydrate (M-S-H). Cem. Concr. Res. 41, 439–442. https://doi.org/10.1016/j.cemconres.2011.01.016