A Review of H₂ Generation and H₂O Distribution in the Earth’s Interior

Abstract

Hydrogen (H) is the most abundant element in the solar system. In the Earth’s interior, it primarily exists in the form of hydrogen gas, water, atomic hydrogen, and hydroxyl groups. Hydrogen gas, as a clean energy source, is widely distributed within the Earth and is mainly generated through serpentinization, with minor contributions from water radiolysis, rock fracturing, biological activity, etc. Hydrogen sequestration occurs mainly through clay adsorption, entrapment under rock layers, dissolution in water, and fluid inclusions. Besides being present as pore water, hydrogen in the deep Earth predominantly resides in minerals as point defects related to hydrogen species (e.g., OH⁻, H⁺). During the Earth’s evolution, substantial hydrogen was stored in the deep Earth through accretion, and surface water has been transported into the Earth’s interior via subducting slabs; meanwhile, it can migrate upward through magmatic activity and mantle plumes. The inputs and outputs constitute the global hydrogen cycle. Hydrogen concentration and distribution are highly heterogeneous across the crust, mantle and core. The upper mantle is likely mostly dry, while the Earth’s core is potentially a large reservoir of hydrogen. Small amounts of hydrogen can profoundly influence the physicochemical properties of the Earth’s interior materials, as well as the dynamic processes within the Earth’s interior.

1. Introduction

Hydrogen (H), as the most abundant and lightest element in the solar system, is increasingly becoming a focus in the field of Earth science. A majority of studies mainly concentrate on its origin, forms, distribution, geochemical behavior, and impact on the physical properties of materials within the Earth’s interior. Hydrogen gas, being one form of hydrogen, is a clean new energy source that has already been widely used in industrial production and in life. According to the International Energy Agency’s Global Hydrogen Review 2023, global hydrogen consumption in 2022 reached 95 million tons [1], with main uses including (Figure 1a): (1) oil refining (44%), mainly for removing impurities (e.g., sulfur) from crude oil and upgrading heavy crude oil; (2) ammonia and methanol production (50%), primarily as feedstock for synthesizing chemical products; and (3) steelmaking (6%), where primary steel can be produced using a mixture of hydrogen and carbon monoxide as a reducing agent for direct reduced iron. Furthermore, hydrogen has broad application prospects. In the transportation sector, hydrogen fuel cells have high weight–energy density, enabling faster acceleration and lighter weight for vehicles. Hydrogen fuel cell vehicles have been proposed as potential replacements for internal combustion engine vehicles. In the construction industry, hydrogen can be used for district heating or cooling. In the power industry, power generation using hydrogen or hydrogen-based fuels (e.g., ammonia) can effectively reduce carbon emissions. Hydrogen-based fuels are also an option for large-scale and long-term energy storage to balance seasonal electricity demand or fluctuations in renewable power generation [2,3].

Currently, most hydrogen used is produced from fossil fuels, referred to as black or gray hydrogen [4]. As shown in Figure 1b, hydrogen produced from natural gas without carbon capture, utilization and storage (CCUS) accounts for about 62%, hydrogen produced from coal without CCUS accounts for about 21%, by-product hydrogen from naphtha reforming in refineries and the petrochemical industry accounts for about 16%, while low-emission hydrogen produced via water electrolysis or using fossil fuels with CCUS accounts for only about 1% [1]. At present, the energy transitions gradually from fossil fuels to non-fossil energy sources. Although hydrogen as a low-carbon energy source holds an important position among clean energies, most hydrogen currently used is a secondary energy produced from other sources, accompanied by carbon emissions during production. In contrast, vast amounts of hydrogen gas are stored in the Earth’s interior; it is a carbon-free source that is labeled white and orange hydrogen [4]. It is estimated that the cost of producing 1 kg of hydrogen could be 2 to 10 times higher than that of extracting natural hydrogen; hence, the exploitation and utilization of natural hydrogen would be a potentially important research direction in future [5].

The Earth formed from the gas and dust of the solar nebula through gravitational accretion and collisions. Therefore, it is suggested that the Earth’s interior is a vast hydrogen reservoir, especially the outer core [4,6,7]. Hydrogen transports from the deep Earth to the surface through degassing processes; therefore, it is an important source of natural hydrogen, which can possibly be directly utilized to serve social development. The hydrogen species in the Earth’s interior are primarily water, molecular hydrogen, hydroxyl (OH), and hydrides. Their presence profoundly influences the physical and chemical properties of rocks and geological processes within the Earth, from the microscopic to the global scale. The study summarizes the origin, cycle, and reservoirs of hydrogen within the Earth’s interior, which not only aids in understanding the evolutionary history of our planet but also provides crucial insights into key scientific issues such as tectonics, magmatic activity, and the origin of life.

2. Generation of Natural Hydrogen

The generation of natural hydrogen stems from both biological and abiotic processes. Biological processes are mainly anaerobic decay of organic matter, fermentation, and nitrogen-fixing bacteria; however, the hydrogen produced in these processes is rapidly consumed by other coexisting organisms [8]. Abiotic processes primarily include serpentinization, water radiolysis, rock fracturing, and magma degassing, as shown in

Table 1. Main abiotic sources of natural hydrogen (the data from [9]).

Source of Hydrogen	Principle of Hydrogen Production	Estimated Annual Fluxes (mol/a)
Hydrothermal Alteration	$3\mathrm{FeO}+{\mathrm{H}}_2\mathrm{O}={\mathrm{Fe}}_3{\mathrm{O}}_4+{\mathrm{H}}_2$ $\mathrm{FeS}+{\mathrm{H}}_2\mathrm{S}={\mathrm{FeS}}_2+{\mathrm{H}}_2$	1012
Radiolysis of Water	$\mathrm{U}, \mathrm{T}, \mathrm{K} \to \mathsf{\alpha }, \mathsf{\beta }, \mathsf{\gamma }$, $\mathsf{\alpha }, \mathsf{\beta }, \mathsf{\gamma } + {\mathrm{H}}_2\mathrm{O} \to {\mathrm{H}}_2$	4.7 × 1010
Rock Fracturing	$2(\equiv \mathrm{Si}·)+{\mathrm{H}}_2\mathrm{O} \to {\mathrm{H}}_2+2(\equiv \mathrm{SiOH})$	/
Magma Degassing	${\mathrm{H}}_2\mathrm{S}+{\mathrm{H}}_2\mathrm{O}={\mathrm{SO}}_2+3{\mathrm{H}}_2$	4.96 × 109

[9]. The hydrogen generated through these abiotic processes is considered to potentially be utilizable. The following section primarily summarizes the abiotic processes.

2.1. Serpentinization

Serpentinization is a metamorphic process where mainly ultrabasic rocks (i.e., iron-bearing mantle rocks) react with water, transforming into serpentine minerals and producing hydrogen, primarily referring to the reaction of olivine with water in which Fe²⁺ in olivine oxidizes to Fe³⁺ and hydrogen in water reduces to H₂. The reaction is as follows:

$$3{\mathrm{Fe}}_2{\mathrm{SiO}}_4 + 3{\mathrm{H}}_2\mathrm{O} = {\mathrm{Fe}}_3{\mathrm{Si}}_2{\mathrm{O}}_5{\left(\mathrm{OH}\right)}_4 + {\mathrm{H}}_{2\left(\mathrm{aq}\right)} + {\mathrm{Fe}}_3{\mathrm{O}}_4 +{\mathrm{SiO}}_{2\left(\mathrm{aq}\right)}$$

(1)

This process is mainly influenced by Fe partitioning. During the reaction, if Fe²⁺ preferentially enters products, e.g., serpentine and brucite, the generated H₂ will decrease. Conversely, when Fe²⁺ preferentially enters the product magnetite (Fe₃O₄), the amount of hydrogen increases [10].

The quantity of hydrogen during serpentinization is affected by many factors, which are summarized in Table 2. The increase in the water–rock ratio enhances the generation of hydrogen, whereas the hydrogen concentration in the solution decreases with an increasing water: rock ratio due to the amount of water increasing, and concentrations are highest when the least amount of H₂ is generated per a given amount of rock [10].

Additionally, temperature significantly affects the reaction rate of serpentinization, which is most active within the temperature range of 200 to 310 °C. At temperatures lower than this range, the reaction rate becomes very limited, and more Fe²⁺ enters brucite as temperature decreases, leading to the decrease in H₂ production. At higher temperatures, the reaction rate decreases rapidly due to thermodynamic constraints, and the extent of serpentinization also diminishes, resulting in less H₂ production [8,10].

Furthermore, SiO₂, CO₂, Al content, and pH also significantly impact H₂ production [11,12,13,14,15,16]. During serpentinization, a Mg-rich, Si-poor layer forms on the olivine surface as the solution changes from acidic to alkaline (pH > 9). Magnesium detaches from this surface layer and dissolves into the solution, then it can react with OH- in the solution to form brucite. The decrease of magnesium content in the solution allows olivine to continue dissolving, thereby increasing the serpentinization reaction rate. However, the higher the pH, the lower the H₂ production due to more Fe²⁺ entering the brucite and serpentine [15]. Moreover, the addition of CO₂ during serpentinization can lead to carbonate formation that facilitates Fe²⁺ entering carbonate products more rapidly, ultimately resulting in the reduction in H₂ production. Nevertheless, the inhibitory effect of CO₂ on H₂ generation diminishes with increasing temperature because high temperatures promote olivine dissolution in CO₂-rich environments, releasing more Fe²⁺ for reaction and simultaneously inhibiting the formation of brucite and magnesite [14,17,18]. In Al-bearing solutions, the primary form of Al is ${\mathrm{Al}\left(\mathrm{OH}\right)}_4^{-}$ which can adsorb onto the positively charged olivine surface, forming Al-Si complexes that promote olivine dissolution and the nucleation of Al-rich serpentine [11]. The serpentinization reaction rate and H₂ generation are also influenced by SiO₂. The participation of SiO₂-rich geological fluids in the hydrothermal alteration of olivine can reduce the hydrogen concentration in the fluid by 1–2 orders of magnitude [16].

Serpentinization is the primary mechanism for hydrogen generation within the Earth’s interior. By this process, hydrogen can be artificially produced by injecting water into reactive geological structures; correspondingly, the hydrogen-rich water from recovery wells surrounding the injection points can be collected [4]. The total peridotite volume in the upper 7 km of the crust is estimated at 10²⁰ kg [19]. Complete oxidation of this mass could theoretically yield up to 100 trillion tons of H₂, equivalent to a production rate of approximately 2–4 kg of H₂ per cubic meter of rock [4]. Furthermore, tectonic activity continuously replenishes peridotites at a rate of 10¹² kg per year, implying an even larger long-term hydrogen production potential [19].

Table 2. The main factors affecting the H₂ production during serpentinization.

Factors	Product H2	Source of Factors	References
T	Increases first, then decreases with T	Geothermal energy	[10]
water: rock	Increases with this ratio	Infiltration/dehydration	[10]
pH	Decreases with pH	Alteration of olivine	[15]
CO2	Suppressed at low T; the effect disappears at >270 °C	CO2-rich hydrothermal fluids	[14,17,18]
Al	Increases with Al	Coexisting minerals	[11]
SiO2	Decreases (<400–515 °C) with SiO2	Coexisting minerals	[16]

Table 2. The main factors affecting the H₂ production during serpentinization.

Factors	Product H2	Source of Factors	References
T	Increases first, then decreases with T	Geothermal energy	[10]
water: rock	Increases with this ratio	Infiltration/dehydration	[10]
pH	Decreases with pH	Alteration of olivine	[15]
CO2	Suppressed at low T; the effect disappears at >270 °C	CO2-rich hydrothermal fluids	[14,17,18]
Al	Increases with Al	Coexisting minerals	[11]
SiO2	Decreases (<400–515 °C) with SiO2	Coexisting minerals	[16]

2.2. Radiolysis of Water

The Earth’s interior contains large amounts of radioactive elements, such as U, Th, and K, which release energy during their decay. The energy is high enough to decompose water molecules into hydrogen atoms, hydroxyl radicals, hydrogen gas, hydrogen peroxide, hydrated electrons, and hydronium ions.

These products subsequently enter the solution and react with other substances to generate H₂ [20]:

$$\mathrm{H}· +\mathrm{H}· = {\mathrm{H}}_2$$

(2)

$$2{\mathrm{e}}_{\mathrm{aq}}^{-}+2{\mathrm{H}}_2\mathrm{O}=2{\mathrm{OH}}^{-}+{\mathrm{H}}_2$$

(3)

$${\mathrm{H}}^{+}+{\mathrm{e}}_{\mathrm{aq}}^{-}+{\mathrm{H}}_2\mathrm{O}={\mathrm{OH}}^{-}+{\mathrm{H}}_2$$

(4)

The amount of H₂ produced by radiolysis primarily depends on the concentration of radionuclides, the availability of water in pore spaces and fractures, and the concentration of dissolved anions and cations in the fluid, including dissolved salts [9]. Since water primarily resides in the pore and fracture spaces of rocks, rock permeability and porosity are important factors for H₂ generation via water radiolysis. It has been estimated that only 1% of the total energy from the radiogenic decay is absorbed by pore water [20]. Additionally, if salts dissolve in water, new covalent bonds form between cations and the oxygen in water, and between anions and the hydrogen in water. These new covalent bonds can weaken the energy of the O–H bonds in water and reduce the energy required to break them, facilitating the production of additional H₂ under irradiation. Therefore, the higher the salt concentration in saline solutions, the more dissolved the cations and anions; correspondingly, the H₂ production increases [21]. Additionally, H₂ yields from radiolysis can also be increased by the presence of certain minerals, e.g., zeolites; this is relevant for natural systems and in the context of decontaminating radioactive water [22].

Furthermore, some studies indicate that the decay products of radioactive elements can also react with water to produce H₂. For example, ⁴⁰Ca, produced from the decay of ⁴⁰K, can react with water to yield Ca(OH)₂ and H₂ [8]. Radiolytic H₂ production is a relatively ubiquitous process in the Earth’s crust. The Precambrian crust that constitutes about 70% of the continental crust can generate approximately 1.6–4.7 × 10¹⁰ mol of H₂ per year [23]. Radioactive elements in basaltic oceanic crust and marine sedimentary rocks can also cause water radiolysis [24,25], It is estimated that the annual H₂ production from radiolysis in the oceanic crust is approximately 2 × 10⁸ mol [26]. Radionuclide concentrations are generally higher in continental crust than in basaltic oceanic crust; therefore, H₂ production will typically be higher in continental crust compared to oceanic crust at the same porosity.

2.3. Rock Fracturing

It has been discovered that hydrogen concentrations are anomalously high in the soil within the fault zone, whereas hydrogen levels in soils away from the fault zone were similar to atmospheric concentrations [27]. By measuring CH₄, CO₂, and O₂ concentrations in the soil and comparing H₂ levels in areas with minimal biological activity, they ruled out biological processes as the cause of the H₂ anomaly, and hypothesized that fresh surfaces generated by rock fracturing react with groundwater to produce H₂ [27]. It has been experimentally evidenced that H₂ gas can be released by crushing granite and quartz, through a chemical reaction between water and such radicals as Si· and Si–O· derived from the break of Si–O bonds by crushing. The Si–O· radical reacts with O₂ to form SiO₂·, which decomposes into Si· at 180–220 °C. The Si· radical then reacts with water to produce H₂ [28]:

$$2\mathrm{Si}· + 2{\mathrm{H}}_2\mathrm{O} \to 2\mathrm{SiOH} + {\mathrm{H}}_2$$

(5)

This reaction has a temperature dependence. Below 200 °C, the increase in Si· with increasing temperature leads to the rise in H₂ production, while above 200 °C, the radicals become unstable, and Si–O· and Si· tend to react with H₂O to form SiOH without H₂ production [28].

Moreover, trace water present at structural defect sites within nominally anhydrous minerals in igneous and metamorphic rocks can undergo redox reactions to produce H₂ [29]:

$$2{\mathrm{OH}}^{-} \rightleftharpoons {\mathrm{O}}_2^{2^{-}} + {\mathrm{H}}_2$$

(6)

If the produced H₂ diffuses away from the defect site to become an interstitial molecule or completely diffuses out of the mineral structure, the reaction becomes irreversible. H₂ generated by this reaction is stored within the rock and can escape through fractures when the rock breaks [29].

This “mechanical” method of H₂ generation via rock fracturing may be widespread in fault zones. Faults are ubiquitous in geological structures such as orogenic belts, subduction zones, continental rifts, passive continental margins, spreading centers, transform faults, and fracture zones. In creeping faults, H₂ generation might be continuous, whereas it could be intermittent in locked faults and limited to slip events [9]. In addition, this H₂ generation mechanism might occur in any region where silicate mineral fragmentation happens.

H₂ produced by rock fracturing serves as an energy source for some microorganisms, which aids in understanding the evolution of ancient life. It has been indicated that after small-magnitude earthquakes H₂ concentrations in fracture zones can reach 1.1 mol/kg—a concentration sufficient to sustain a hydrogen-based subsurface lithoautotrophic microbial ecosystem. In addition, this process of earthquake-generated H₂ sustaining subsurface microbial activity might have occurred as early as 3.8 billion years ago [30]. The study from Telling et al. [31] suggested that H₂ produced from the pulverization of subglacial bedrock could have served as an energy source for subglacial life during past global glaciation periods.

2.4. Other Reactions

The oxidation of sulfides in nature can also generate H₂. The formation of pyrite (FeS₂) is accompanied by H₂ production. Amorphous FeS, hematite (Fe₂O₃), and magnetite (Fe₃O₄) can react with H₂S to produce pyrite and H₂ [32,33]. The high H₂ content observed in hydrothermal systems of the Indian Ocean Ridge can be generated by the precipitation of metal sulfides [34]:

$${\mathrm{Cu}}^{+} + {\mathrm{Fe}}^{2+} + 2{\mathrm{H}}_2\mathrm{S} = {\mathrm{CuFeS}}_2 + 0.5{\mathrm{H}}_2+ 3{\mathrm{H}}^{+}$$

(7)

Additionally, magmatic gas emissions containing H₂ can be observed during or after volcanic eruptions. This involves the conversion of H₂S and SO₂ in magma which is primarily controlled by pressure, where the yields of SO₂ and H₂ increase with the decrease in pressure [35]. H₂ in magmatic gases may also originate from reactions between CH₄ and H₂O, which continuously produce H₂ at temperatures around 1200 °C [36].

Within the Earth’s interior, the process where Fe²⁺ oxidizes to Fe³⁺ and generates H₂ occurs not only in serpentinization reactions but also in other Fe-bearing minerals. For example, Fe-bearing amphiboles in the continental crust and subduction zones can undergo oxidation–dehydrogenation during prograde metamorphism [37]:

$$2{\mathrm{Fe}}^{2+} + 2{\mathrm{OH}}^{-} = 2{\mathrm{Fe}}^{3+} + {\mathrm{O}}^{2-} + {\mathrm{H}}_2$$

(8)

3. Hydrogen Occurrence

Natural hydrogen can be adsorbed by clay minerals. It was found that clay minerals (primarily illite, chlorite, and kaolinite) in the Cigar Lake area of northern Saskatchewan, Canada, can adsorb up to 500 ppm of hydrogen. Over the 1.4-billion-year lifespan of this deposit, 4%–17% of the hydrogen generated by water radiolysis was adsorbed by clay minerals [38]. Hydrogen is adsorbed onto the surfaces of clay minerals via van der Waals forces. The adsorption capacity of clay minerals is influenced by temperature, pressure, and pore structure. It increases with rising pressure, and pore sizes smaller than 30 nm can effectively enhance the adsorption capacity of clay minerals. However, the kinetic energy of hydrogen molecules increases with temperature, leading to a reduction in the adsorption capacity of clay minerals [39,40].

Natural hydrogen can also exist as free hydrogen dissolved in water and be stored sealed by low-permeability cap rocks. A study by Maiga et al. [41] on hydrogen-bearing reservoirs in the Bourakébougou field, Mali, revealed that hydrogen primarily resides in karst pores within carbonate rocks at shallow depths. The solubility of hydrogen in water increases with pressure. At greater depths, hydrogen is mainly dissolved in water. The diabase with few fractures overlying the carbonate reservoir acts as a cap rock sealing the hydrogen. In contrast, diabase at deeper depths exhibits high fracture density, facilitating the upward migration of hydrogen [42]. Porous media primarily store hydrogen through structural trapping and cap rock seals. Cap rocks, due to their low permeability and small pore sizes, are considered to generate high capillary forces that prevent hydrogen from passing through, thereby providing a geological seal against leakage. Ideally, hydrogen remains trapped in the reservoir as long as the capillary pressure threshold of the cap rock is not exceeded [42].

Hydrogen (H₂) may dissolve in minerals. Due to their extremely small size, hydrogen molecules can occupy interstitial sites in the mineral lattice. The solubility of hydrogen in minerals increases with increasing pressure and hydrogen fugacity [43].

Hydrogen can also be stored within fluid inclusions. Aqueous fluids can form fluid inclusions within minerals at temperatures corresponding to the brittle–ductile transition (e.g., 600–800 °C for olivine). When olivine cools below approximately 400 °C, the water within these inclusions can participate in serpentinization reactions, generating hydrogen that becomes stored within the inclusions [44,45].

4. Hydrogen in the Deep Earth

Besides H₂ and free water, hydrogen, mainly as point defects such as hydroxyl OH⁻), is present within mineral structures in the deep Earth. Surface water enters the Earth’s interior primarily through subduction zones. Because the subducting lithosphere in contact with the ocean is relatively oxidized, hydrogen enters the mantle predominantly as an oxidized form: either H₂O within rock pores or OH⁻ within mineral structures [46]. With slab subducting to deep depths, hydrous minerals transform into more stable high-pressure hydrous phases. Water can even reach the core–mantle boundary through dense hydrous silicates and react with substances like Fe and Si in the outer core to form FeH_x, which subsequently enters the outer core [47,48,49]. Based on experimental and computational data as well as natural samples, the water content in the Earth’s interior was estimated, as shown in

Table 3. Summary of the estimated water content in Earth’s layers.

Layers	Water Budget (Ocean Mass)	Water Content	References
Crust	~1.23	1.3 wt% and 1.5 wt% in continental and oceanic crust, respectively	[52]
Upper mantle	0.02–0.07	50–200 ppm water	[53]
Transition zone	0.6–2.62	2000 ppm water	[52]
		1.4–1.5 wt% water based on experimental and natural samples data	[54,55]
Lower mantle	0.2–2.9	100 ppm water	[52]
		1340 ppm H2O in average	[56]
Core	37–73	0.3–0.6 wt% H based on experimentally determined metal-silicate partition coefficient of hydrogen and machine-learning methods	[57,58]

Note: 1 ocean mass is about 1.4 × 10²¹ kg. The water contents of the lower mantle and core have enormous uncertainties due to the samples from these regions for water analyses being unavailable. In the core, H present in metal was calculated as H₂O for comparison with the silicate oxide layers of the Earth.

and Figure 2. The mantle is thought to be a huge water reservoir, with estimates ranging from about a quarter to more than four ocean masses [50,51], whereas there is large uncertainty around the concentrations of H₂O in the Earth’s interior, primarily arising from the less-well-known concentrations for the lower mantle and core.

4.1. The Crust

The crust, being the Earth’s outermost layer, comprises the continental crust and the oceanic crust. In the shallow continental crust, water primarily exists as free water within rock pores or fractures. Pore water at depths less than 4 km is considered to be part of the groundwater reservoir, while pore water at depths greater than 4 km is regarded as part of the continental crust reservoir [52]. Besides pore water, the crust contains many water-bearing minerals such as mica, amphibole and clay minerals in their igneous and metamorphic units and water-rich sediments. Moreover, the crustal water reservoir also includes water dissolved within nominally anhydrous minerals (

Table 4. Water solubility of major crustal minerals.

Mineral	Water Solubility	T (K)	P (GPa)	Data Source
Quartz	3.7–32.2 ppm	1073	0.2–1	[63]
Plagioclase	36–248 ppm	973–1123	0.5	[64]
Albite	75–80 ppm	1023	0.2	[65]
Labradorite	90–285 ppm	973–1123	0.2–1	[65]
Anorthoclase	170–240 ppm	1023–1073	0.2	[65]
Orthopyroxene	34–117 ppm	1073–1173	0.2–1	[66]
Clinopyroxene	14–169 ppm	1073–1173	0.2–1	[66]
	142–1144 ppm	873–1273	1	[67]
Rutile	48–956 ppm	773–1173	0.5–2	[68]

). During the cooling of basaltic magma into igneous rocks or the metamorphic transformation of igneous rocks in the continental crust, water can be incorporated into the lattice defects of nominally anhydrous minerals such as pyroxene, feldspar, quartz and olivine [59,60,61,62]. The oceanic crust reservoir, on the other hand, includes all water between the sediment/seawater interface and atop the upper mantle [52]. Based on calculations using amphibolite and granulite in the continental crust and basalt and gabbro in the oceanic crust, the average water content was estimated to be 1.3 wt% and 1.5 wt%, respectively [52].

4.2. The Upper Mantle

The primary minerals in the upper mantle are olivine, pyroxene, and garnet (Table 5). The measurements of water partition coefficients for synthetic and natural peridotite minerals indicate that the order of water content among these mantle minerals, from highest to lowest, is: pyroxene > garnet > olivine [69]. The study on xenoliths representing upper mantle lithologies also indicated that clinopyroxene, orthopyroxene, olivine and garnet contain 342–413, 169–201, 3–54 and 0 to <3 ppm H₂O, respectively, and the average water contents of xenoliths are approximately 113 ppm [70]. As mineral proportions of peridotite change with depth in the mantle, the primary water-bearing mineral shifts from pyroxene at shallow depths to olivine at greater depths. With the depth increasing from 70 km to 300 km, the water partition coefficient between pyroxene and olivine ($D_{{\mathrm{H}}_2\mathrm{O}}^{\mathrm{pyroxene}/\mathrm{olivine}}$) changes from 10 to 1.3; correspondingly, water content in olivine varies from 22 pm to 120 ppm. Therefore, the H₂O content of olivine increases by a factor ~5 from the shallow to deep upper mantle [69], meaning that olivine is the principal host of H₂O in the deep upper mantle. Combining data on the water storage capacity of olivine with constraints on partition coefficients between olivine, pyroxene, and garnet, the water storage capacity of the upper mantle was estimated at 50–200 ppm [53].

There are several mechanisms of water incorporation into silicate crystal structure proposed by available studies. The substitution of a Si⁴⁺ by an Al³⁺ ion plus a H⁺ is one possible mechanism, whereas it cannot account for the observed rapid uptake of H ions that diffuse millimeters in a few hours at high temperature; in contrast, Al ions diffuse a few nanometers. The second is the diffusion of H ions in iron-bearing silicate minerals, which can be charge compensated by a counter flux of holes, that is, the absence of electrons in the reduction reaction Fe³⁺ → Fe²⁺ + hole. The third mechanism is the incorporation of H ions into anhydrous minerals, which can be charge compensated by a parallel flux of Si, Fe, or Mg cation vacancies or possibly by a counter flux of oxygen interstitials [71]. Under water-saturated conditions, hydrogen mainly enters the structures of upper mantle minerals as OH⁻ substituting for magnesium vacancies. Under water-undersaturated conditions, H+ and Al³⁺ co-substitute for Si⁴⁺ in the structures of olivine and pyroxene, or 2H⁺ substitutes for Mg²⁺ [72]. Aluminum content is the main factor influencing the water content in pyroxene, which decreases with increasing depth; correspondingly, the water content in pyroxene also decreases with depth [53,72,73]. Olivine also has the same feature as pyroxene. Recent studies further demonstrate that oxygen fugacity influences water solubility in upper mantle minerals by affecting the ratio of Fe³⁺ to Fe²⁺. In pyroxenes and garnet, water solubility exhibits a negative correlation with oxygen fugacity, which is primarily attributed to the reduction of Fe³⁺ to Fe²⁺ and its incorporation mechanism coupled with hydroxyl (OH⁻) groups. In contrast, water solubility in olivine shows a positive correlation with oxygen fugacity, likely due to the coupled substitution of H⁺ and Fe³⁺ for two Mg²⁺ vacancies [74].

Table 5. Water solubility of major mantle minerals.

Mineral	Water Solubility	T (K)	P (GPa)	Data Source
Olivine	40–385 ppm	1373–1623	1–7	[75]
Orthopyroxene	101–269 ppm	973–1373	1.5	[76]
	867 ppm	1373	7.5	[77]
	785–1647 ppm	1673	12–14	[78]
	75–219 ppm	1373	2–3	[66]
Clinopyroxene	714 ppm	1373	10	[77]
	84–394 ppm	1373	2–3	[66]
Pyrope	99–1024 ppm	1373–1473	5–9	[79]
	14–31 ppm	1073–1273	1.5–3	[74]
Wadsleyite	1.8–2.3 wt%	1673	16.5	[80]
	0.9–2.2 wt%	1173–1673	15	[81]
	4475–9370 ppm	1673	12–14	[78]
	0.58–2.42 wt%	1500–2100	17.5–21	[82]
Ringwoodite	1–1.25 wt%	1673	16.5	[80]
	0.25–2.06 wt%	1600–2000	23	[54]
Majorite	900–3000 ppm	1670–2270	20	[83]
Bridgmanite	1099 ppm	3690	33	[84]
	0.19–0.24 wt%	1873–1923	25.5	[85]
	100 ppm	1573	25	[86]
Aluminous bridgmanite	1100–1400 ppm	1473	25–26	[86]
Fe-, Al-bearing bridgmanite	1400–1800 ppm	1673–1873	25	[86]
	90–110 ppm	1273–1473	26	[86]
	1020 ppm	2073	24	[87]
Stishovite	1.3 wt%	623–823	10	[88]
	128–521 ppm	1573–2373	22	[89]
Aluminous stishovite	844 ppm	1473–1673	10–15	[90]
	296–3611 ppm	1600–2100	23–32	[91]
Davemaoite	0.37–0.38 wt%	1873–1893	25.5	[85]
	5100 ppm	2173	25	[86]
	0.5–1 wt%	1400–2200	19–120	[92]
Periclase	21–112 ppm	1673–2073	25	[86]
Ferropericlase	25–180 ppm	1673–2073	25	[86]
	0.19 wt%	1923	25.5	[85]
	14–79 ppm	1673–2273	25	[93]

Table 5. Water solubility of major mantle minerals.

Mineral	Water Solubility	T (K)	P (GPa)	Data Source
Olivine	40–385 ppm	1373–1623	1–7	[75]
Orthopyroxene	101–269 ppm	973–1373	1.5	[76]
	867 ppm	1373	7.5	[77]
	785–1647 ppm	1673	12–14	[78]
	75–219 ppm	1373	2–3	[66]
Clinopyroxene	714 ppm	1373	10	[77]
	84–394 ppm	1373	2–3	[66]
Pyrope	99–1024 ppm	1373–1473	5–9	[79]
	14–31 ppm	1073–1273	1.5–3	[74]
Wadsleyite	1.8–2.3 wt%	1673	16.5	[80]
	0.9–2.2 wt%	1173–1673	15	[81]
	4475–9370 ppm	1673	12–14	[78]
	0.58–2.42 wt%	1500–2100	17.5–21	[82]
Ringwoodite	1–1.25 wt%	1673	16.5	[80]
	0.25–2.06 wt%	1600–2000	23	[54]
Majorite	900–3000 ppm	1670–2270	20	[83]
Bridgmanite	1099 ppm	3690	33	[84]
	0.19–0.24 wt%	1873–1923	25.5	[85]
	100 ppm	1573	25	[86]
Aluminous bridgmanite	1100–1400 ppm	1473	25–26	[86]
Fe-, Al-bearing bridgmanite	1400–1800 ppm	1673–1873	25	[86]
	90–110 ppm	1273–1473	26	[86]
	1020 ppm	2073	24	[87]
Stishovite	1.3 wt%	623–823	10	[88]
	128–521 ppm	1573–2373	22	[89]
Aluminous stishovite	844 ppm	1473–1673	10–15	[90]
	296–3611 ppm	1600–2100	23–32	[91]
Davemaoite	0.37–0.38 wt%	1873–1893	25.5	[85]
	5100 ppm	2173	25	[86]
	0.5–1 wt%	1400–2200	19–120	[92]
Periclase	21–112 ppm	1673–2073	25	[86]
Ferropericlase	25–180 ppm	1673–2073	25	[86]
	0.19 wt%	1923	25.5	[85]
	14–79 ppm	1673–2273	25	[93]

4.3. The Transition Zone

The Earth’s transition zone, located between 410 km and 660 km depth, primarily contains wadsleyite and ringwoodite, which have enormous water storage capacities (Table 5). Olivine transforms into wadsleyite at around 410 km depth, and wadsleyite transforms into ringwoodite at around 520 km depth. The presence of water can accelerate these transformation rates [94,95,96]. Furthermore, with increasing pressure, clinopyroxene and orthopyroxene, along with pyrope, transform into majorite. In the mantle transition zone, majorite becomes the primary garnet phase [96].

The water content in wadsleyite and ringwoodite decreases with the increase in temperature. The effect of temperature on the water content of iron-free ringwoodite is more significant than on iron-bearing ringwoodite. For instance, when temperature rises from 1800 K to 2000 K, the H₂O solubility in iron-free ringwoodite decreases by 50%, while it only decreases from about 1.5 wt% to 1.15 wt% for iron-bearing ringwoodite as temperature increases from 1600 K to 2000 K [54]. The study on wadsleyite showed that the water solubility in it decreases significantly when increasing temperature from ~2.2 wt% H₂O at 900 °C down to 0.9 wt% at 1400 °C at 15 GPa, but the effect of pressure on the water content is not marked [81]. Based on the water content of synthetic minerals under conditions of the transition zone, it was inferred that ringwoodite can store 0.8–1.2 wt% water [54]. Analysis of natural diamond inclusions containing ringwoodite by Pearson et al. [55] found water contents of 1.4–1.5 wt%, leading to the inference that the transition zone contains about 1 wt% water. The water content of majorite is influenced by temperature, pressure, and Mg/Fe contents. Compared to ringwoodite and wadsleyite, the water content of majorite is extremely low, and it is generally not considered to play a dominant role in water storage in the transition zone [97,98]. Using an average transition zone water content estimate of 2000 ppm, it was inferred that the transition zone holds the equivalent of about 0.6 ocean masses of water [52]. Therefore, the transition zone is considered the most potential “water reservoir” in the Earth’s deep interior [99].

The transition zone acts as a filter for water. The downward injection of cold subducting slabs causes the surrounding mantle to rise. When wadsleyite ascends into the upper mantle and transforms back into olivine, the lower water solubility of olivine can lead to partial melting at the 410 km discontinuity, with some incompatible elements entering the melt. The residual partially dehydrated peridotite will remain saturated with respect to hydrous fluid. If that peridotite were then advected to a region of lower storage capacity, additional fluid or melt would be expelled from the solid. The melt, being denser than the upper mantle but less dense than the transition zone, accumulates around 410 km depth. When this melt encounters a subducting slab, it can be carried into the deep mantle. Most of the water diffuses back into the transition zone, while the remaining water present in hydrous minerals continues to migrate downward with the slab. At the 660 km boundary, subducting slabs tend to flatten, and minerals transform from transition zone assemblages to lower mantle minerals. Because the water solubility of lower mantle minerals is lower than that of transition zone minerals, water is released from the minerals. The rock matrix containing the exsolved water develops sufficient porosity, allowing water to percolate from the lower mantle back into the transition zone [99].

4.4. The Lower Mantle

Ringwoodite transforms into bridgmanite and ferropericlase at a depth of approximately 660 km, and majorite garnet transforms into bridgmanite between depths of about 660 km and 800 km [83]. The primary mineral in the lower mantle is bridgmanite, followed by ferropericlase and davemaoite (Table 5). In bridgmanite, hydrogen is incorporated not only by substituting for Mg²⁺ or co-substituting with Al³⁺ for Si⁴⁺ but also as OH⁻ occupying oxygen vacancies created by the substitution of trivalent cations (Fe³⁺, Al³⁺) for Si⁴⁺. Consequently, an increase in Al content enhances the water solubility of bridgmanite [100]. Under conditions at the top of the lower mantle, the water solubility of ferropericlase (e.g., 14–79 ppm at 25 GPa and 1400–2000 °C) is considerably lower than that of bridgmanite (e.g., 1020 ppm at 24 GPa and 1800 °C). However, bridgmanite can react with water to form hydrous phases such as δ–H phases, i.e., AlOOH–MgSiO₂(OH)₂ solid solution. When bridgmanite coexists with hydrous phases such as phase δ–H solid solution and stishovite, water does not preferentially partition into bridgmanite, indicating that bridgmanite is not the primary water reservoir in the uppermost lower mantle [87,93,101,102,103].

Dense hydrous magnesium silicates (DHMSs) may also serve as water reservoirs in the lower mantle. Nevertheless, most high-pressure hydrous phases are stable at temperatures below the lower mantle adiabat; for instance, the phase D (Mg_1.14Si_1.73H_2.81O₆) has a maximum thermal stability of 1673 K in the 22–32 GPa pressure range, whereas the mantle temperature at 660 km depth is about 1850 K. Therefore, these hydrous phases mainly exist in cold subducted slabs [104,105,106,107]. At around 40 GPa, Phase D transforms into phase H which is stable in the 35–60 GPa range. When it forms a solid solution with the δ-phase, i.e., the δ–H solid solution, its stability field expands significantly, allowing it to exist within the lower mantle. Water derived from the dehydration of the subduction slab reacts with Al-rich bridgmanite in the uppermost lower mantle, forming Al-poor bridgmanite and phase δ-H solid solution. The phase δ-H solid solution then transports water into the deeper mantle [57,101,106,108]. If hydrous phases or hydrated nominally anhydrous minerals from the slab accumulate at the base of the lower mantle and are heated, the released water can promote partial melting, potentially contributing to the formation of ultra-low velocity zones [109].

Recent experiments have indicated that davemaoite is also an important water-bearing mineral in the lower mantle. Chen et al. [92] determined the water contents of 0.5–1.0 wt% in davemaoite synthesized at pressures of 19–120 GPa and temperatures of 1400–2200 K using infrared spectroscopy, and observed peak splitting in the X-ray diffraction patterns of hydrous davemaoite under lower mantle conditions, attributing this to a tetragonal distortion of its crystal structure. It was further found that water incorporation in davemaoite via H⁺ substitution for Ca²⁺ or Si⁴⁺ by density functional theory calculations, and the degree of non-cubic distortion for a davemaoite crystal with 1 wt% water content was consistent with the previously observed peak splitting [110].

It was estimated that the lower mantle contains the equivalent of about 0.2 ocean masses of water, assuming an average water content of 100 ppm [52]. However, Peslier et al. [55] states that the lower mantle could contain up to about 2.9 ocean masses of water based on a lower mantle composition of 80% bridgmanite, 11% ferropericlase, 8% davemaoite, and 1% hydrous minerals.

4.5. The Earth’s Core

The density of the Earth’s core is 8%–10% lower than that of pure iron under the same temperature and pressure conditions. This density deficit is proposed to be attributed to the incorporation of light elements; for instance, adding just 1 wt% hydrogen could account for this density difference [111,112]. During the early stages of the Earth’s formation, reactions between hydrous magma and iron could have generated FeH_x. This reaction between water and iron occurs at pressures above approximately 4 GPa [113]:

$$\mathrm{Fe} + {\mathrm{H}}_2\mathrm{O} \to \mathrm{FeO} + {\mathrm{FeH}}_{\mathrm{x}}$$

(9)

The FeHx would sink and become part of the core, while FeO would be absorbed by surrounding silicates [47,113].

At pressures exceeding about 78 GPa, iron can react directly with water [47]:

$$\mathrm{Fe} + {\mathrm{H}}_2\mathrm{O} \to {\mathrm{FeO}}_2{\mathrm{H}}_{\mathrm{x}} + {\mathrm{FeH}}_{\mathrm{x}}$$

(10)

When dense hydrous minerals (e.g., phase δ–AlOOH) transport water to the core–mantle boundary and subsequently dehydrate, the released water can react with iron in the outer core. The FeH_x and FeO₂H_x (Py-phase) can accumulate at the top of the outer core, forming oxygen-rich patches (ORP), as illustrated in Figure 3 [47,114,115,116]. These oxygen-rich patches are thought to move laterally due to mantle convection, from their cooler formation regions towards hotter areas, and ultimately mix with mantle silicates to form large low-shear-velocity provinces (LLSVP). Some FeH_x may melt and incorporate into the outer core as temperature increases, while FeO₂H_x may gradually dehydrogenate with increasing temperature and heating duration. The released hydrogen might then migrate to the Earth’s surface through grain boundaries in mantle minerals or by forming hydrocarbons. Alternatively, it could be re-absorbed by iron in the core [115,116,117].

During the cooling and crystallization of the early magma ocean, the formation of the Py-phase would generate hydrogen which would permeate into the liquid outer core and simultaneously oxidize the mantle. In this oxidized environment, it becomes more favorable for the formation of the Py-phase. Concurrently, hydrogen residing within the silicates/oxides would continue to diffuse, potentially in a superionic state, ultimately contributing to a chemically highly heterogeneous core–mantle boundary region [118].

Furthermore, water released from mineral dehydration at the core–mantle boundary can also react with Fe-Si alloys present in the outer core:

$$\mathrm{Si} + 2{\mathrm{H}}_2\mathrm{O} = 4\mathrm{H} + {\mathrm{SiO}}_2$$

(11)

Since hydrogen behaves as a siderophile element under high pressure, hydrogen produced by this reaction enters the metallic core to form FeH_x, while the SiO₂ reacts with ferropericlase to form silicates that remain in the mantle [48,49].

The incorporation of hydrogen into the Earth’s core reduces the density of iron, which can account for the density deficit in the inner core. However, it simultaneously increases the shear wave velocity (V_s), which contradicts seismological observations showing a reduced V_s in the inner core. Therefore, other light elements besides hydrogen might also be present in the inner core [112,119]. Furthermore, Fe–H, Fe–C, and Fe–O alloys may transition into a superionic state under core conditions. In this state, C, H, and O ions can diffuse rapidly through the interstitial sites of the iron lattice, and exhibit fluid-like characteristics that reduce both the shear wave velocity and the density, aligning with seismic observations [120].

On the basis of different core formation models and the experimentally determined metal–silicate partition coefficient of hydrogen, the hydrogen content in the core was estimated to be 0.3–0.6 wt%, equivalent to 37–73 ocean masses of water, which can account for 30%–60% of the density deficit and sound velocity excess of the outer core [7]. However, using machine-learning methods, the inner core was proposed to contain 0.15–0.19 wt% H, equivalent in mass to about one ocean, while the outer core contains 0.31 wt% H, equivalent to about 33 ocean masses [58].

5. Hydrogen Cycle in the Earth’s Interior

5.1. Input of Surface Water into the Earth’s Interior

Terrestrial planets formed from an accretion disk resulting from the collapse of a vast, cold, slowly rotating cloud of gas and dust in the solar system. During planetary formation, water was likely incorporated into the growing planetesimals and planets through adsorption processes [6]. In the Earth’s case, water dissociated into OH was adsorbed onto minerals like olivine. Subsequently, this OH could react with other adsorbed H₂, high-energy protons from the solar wind, or other OH groups to form H₂O [121]. Wu et al. [122] proposed a nebular ingassing model for hydrogen, suggesting that the Earth’s mantle contains about two ocean masses of water, while the core holds approximately five ocean masses. After the planet’s initial formation, additional water may have been delivered via impacts from comets and asteroids [6]. Water delivered by asteroid and comet impacts likely represents only a minor fraction of the Earth’s water budget, contributing perhaps just a few percent of an ocean mass [123].

After the Earth’s formation, the primary mechanism for surface water entering the Earth’s interior is through the subduction of oceanic crust (Figure 2). At the initial stage of subduction, a significant portion of pore water within the rocks is expelled due to compaction. Most of the residual water is carried into the upper mantle with the subducting oceanic crust. Here, water reacts with peridotite to form hydrous minerals such as chlorite, amphibole, and serpentine. Due to scraping of the mantle wedge base by the subducting slab, hydrated peridotite can be carried deeper into the Earth along the subduction zone [124].

During subduction, metamorphic dehydration and partial melting of minerals facilitate the formation of more stable hydrous minerals, which transport water deeper into the Earth’s interior. At forearc depths, most hydrous minerals dehydrate as temperature and pressure increase. At the sub-arc depth, high-pressure hydrous phases such as lawsonite, zoisite, and phengite, along with hydrated nominally anhydrous minerals like clinopyroxene and garnet, become the primary carriers of water [125].

5.2. Upward Migration of Hydrogen

At shallow depths in subduction zones, hydrous minerals undergo dehydration reactions. The released water migrates upwards and reacts with the peridotite at the base of the mantle wedge, forming serpentine and chlorite. Simultaneously, the subducting slab drags solid material from the mantle wedge downwards. When these hydrous minerals from the base of the mantle wedge reach depths where they become unstable, a dehydration reaction occurs. The released water can form interconnected networks within the peridotite, allowing it to percolate into the overlying, hotter mantle wedge [126,127]. Concurrently, the addition of water lowers the solidus temperature of mantle rocks, leading to the formation of hydrous melts. Because the melt is less dense than the surrounding rock, it rises, and continuously dissolves silicate minerals, eventually forming magma that carries water back to the atmosphere and hydrosphere via arc volcanism [128]. When hydrous minerals are transported into the deep, reduced mantle by the subducting slab, the water released by the dehydration reaction equilibrates with the oxygen fugacity of the surrounding bulk mantle, potentially forming a hydrogen-rich fluid that contains very little water. Due to its low density and viscosity, this fluid tends to migrate upwards [129].

The temperature gradient between the Earth’s outer core and the mantle causes continuous heat flow from the core outwards, which can trigger partial melting of subducted oceanic crust and lower the solidus of mantle material, forming mantle plumes [130]. They can ascend through the mantle driven by thermal buoyancy, and the plume head entrains surrounding mantle material [131,132]. When the mantle plume head reaches the base of the lithosphere, it can cause extensive partial melting, ultimately leading to the formation of Large Igneous Provinces at the surface [133]. Early in the Earth’s formation, hydrogen might have been stored in the deep interior, possibly dissolved in silicate minerals. However, as mantle plumes rise, this hydrogen stored in the lower mantle is transported upwards, and the exothermic reactions of substances (hydrogen and helium) within the mantle plume provide the energy for its ascent [134]. Upon reaching the upper mantle, changes in oxygen fugacity may cause hydrogen to be oxidized to water or OH [43].

6. Summary and Prospectives

Hydrogen within the Earth’s interior plays a pivotal role in the occurrence and evolution of geological processes. Major progress has been made on hydrogen in the form of water (i.e., structurally bound OH in minerals) in the Earth’s interior for several decades, which includes experimental determinations of water solubility in mantle minerals and the water capacity of the deep Earth; investigations of water cycling processes inside the Earth; and understanding the influence of water on the geophysical properties of minerals. In recent years, natural hydrogen gas, as a naturally renewable energy source within the Earth, has gradually gained widespread attention. Serpentinization, as the dominant process of natural hydrogen generation, has been extensively and deeply studied. However, knowledge on the migration and accumulation of natural hydrogen in the lithosphere is currently very limited; correspondingly, experimental and field evidence is required to facilitate the exploration and exploitation of hydrogen in the future.

Even though a lot of advances in the generation, storage and cycling of the deep Earth’s hydrogen have been made, there remains some issues that need to be explored in detail:

(1)

The mechanisms for natural hydrogen formation, particularly for abiotic origin, as well as the impact factors affecting hydrogen generation should be further explored.

(2)

Uncertainty in origin and distribution: While hydrogen in the Earth’s interior is a carbon-free, sustainable resource, current research on natural hydrogen primarily focuses on surface occurrences. The genetic mechanisms and distribution patterns of hydrogen in the Earth’s interior are not yet fully understood and more work is required.

(3)

Limited understanding of production mechanisms: Hydrogen production by the serpentinization process has been studied in detail. By contrast, other hydrogen generation pathways have not yet obtained adequate research. To facilitate the exploitation and exploration of hydrogen energy resources within the Earth, comprehensive studies on these other production processes are also important.

(4)

Discrepancies in water content estimates: Water contents in nominally anhydrous minerals measured under experimental conditions often vary largely among different studies. Accordingly, water content estimates for each layer of the Earth’s interior determined by different methods (e.g., physical property, petrology, geochemistry) have remarkable discrepancies. For accurate determination of the water content in primary constituent minerals and estimates of the water content in the Earth’s interior, the integration of various analytical methods and disciplines is needed.

(5)

Challenges in sampling the deep Earth’s materials and accurately quantifying the water content of the Earth’s interior: Studies on natural samples from the deep Earth remain challenging owing to the present limited drilling depth, and magma trapped in crystals at the surface of the Earth may not originate from a single-source region and undergo significant chemical changes during their ascent, both of which make it difficult to accurately determine the hydration state of different layers in the Earth’s interior.