Effect of Iron Mineral Transformation on Long-Term Subsurface Hydrogen Storage—Results from Geochemical Modeling

: Large-scale subsurface hydrogen storage is critical for transitioning towards renewable, economically viable, and emission-free energy technologies. Although preliminary studies on geo-chemical interactions between different minerals, aqueous ions, and other dissolved gasses with H 2 have helped partially quantify the degree of hydrogen loss in the subsurface, the long-term changes in abiotic hydrogen–brine–rock interactions are still not well understood due to variable rates of mineral dissolution/precipitation and redox transformations under different conditions of reservoirs. One of the potentially understudied aspects of these complex geochemical interactions is the role of iron on the redox interactions and subsequent impact on long-term (100 years) hydrogen cycling. The theoretical modeling conducted in this study indicates that the evolution of secondary iron-bearing minerals, such as siderite and magnetite, produced after H 2 -induced reductive dissolution of primary Fe 3+ -bearing phases can result in different degrees of hydrogen loss. Low dissolved Fe 2+ activity (<10 − 4 ) in the formation water can govern the transformation of secondary siderite to magnetite within 100 years, eventually accelerating the H 2 consumption through reductive dissolution. Quantitative modeling demonstrates that such secondary iron mineral transformations need to be studied to understand the long-term behavior of hydrogen in storage sites.


Introduction
Hydrogen is gaining popularity as a primary low-carbon energy source and a novel mode of decarbonizing fuel-intensive industries, including steel and chemical production as well as transportation, power, and heating.Common renewable energy sources, like solar and wind energy, are highly weather-dependent, expensive, and require large land surface areas for deployment [1].Another major drawback is the need for batteries to store large quantities of energy to balance the severe fluctuation in the energy supply from these sources.Hydrogen is emerging as an appealing alternative renewable energy buffer because it is consistent, reliable, and capable of meeting our growing energy demands.Hydrogen usage would also lower greenhouse gas emissions from energy-intensive industries and help reduce global warming.
Different forms of 'produced hydrogen' can be stored in a variety of ways, including cryogenic tanks, high-pressure gas cylinders, and as adsorbed components on materials with large specific surface areas like clays [2][3][4].A single 1000-megawatt (MW) hydrogenpowered plant would need more than 500 tons of daily H 2 supply [5].However, the storage and discharge capacity (MW h; hours to days) of surface hydrogen storage infrastructure, such as pipelines and tanks, is very restricted.Therefore, there is a critical need to develop massive large-scale hydrogen storage capacities to supply energy in the GW h/TW h-range [6,7].Underground hydrogen storage (UHS) in various subsurface geological formations is emerging as one of the best plausible solutions for short-and long-term large-scale hydrogen storage.
Underground salt caverns have been widely studied and proven effective for natural UHS.However, the salt caverns are restricted in spatial abundance and, therefore, have limited applicability for deploying UHS systems on a wider scale in different geographical regions [8].Hence, there is a critical need to evaluate the utility of more commonly found geological structures like sandstone aquifers or depleted oil/gas reservoirs for subsurface hydrogen storage [6,[9][10][11].These reservoirs also have several orders of magnitude higher storage capacities and sealing capabilities than salt caverns, offering a more extensive and geographically independent option [3].Therefore, there is a need for detailed geological fault/fracture characterization of the sites and the need to use artificial means like cushion gas (CO 2 or N 2 ) injection to reduce the possibility of leakage in these reservoirs [7,12,13].The integrity and stability of the storage sites could be impacted by complex geochemical processes between the minerals, formation brine, and injected gasses.
Reservoir minerals dissolve and trigger subsequent geochemical reactions that consume hydrogen [14], or directly react with the injected hydrogen, causing variable degrees of consumption [15][16][17].Dissolved ions like bicarbonate, sulfate, nitrate, iron, or some gasses in the formation water can alter the stability of hydrogen and the overall equilibrium through several complex redox and pH-dependent processes that have been investigated only in a few studies [15,16,[18][19][20].Although there are some studies on the impacts of microbes in underground hydrogen [20], this paper focuses only on understanding the abiotic reactions with iron-bearing phases.The lack of data on abiotic geochemistry in geological hydrogen storage systems raises uncertainty in the development of UHS systems.It is thus necessary to comprehend this uncertainty around hydrogen loss, and the decrease in reservoir integrity caused by abiotic geochemical reactions between the rocks, formation fluids, and hydrogen in the reservoir to enable safe geological storage [7,9,19,21].In the formation water, specific ionic species such as iron concentrations need to be critically evaluated due to their variable natural abundances, redox sensitivities, microbial affinities, and hydrogen-oxidizing potentials [16, [22][23][24][25][26].The uncertain kinetics of iron mineral dissolution as a potential controller of dissolved hydrogen concentration need to be critically evaluated.
This study uses simple kinetic batch simulations to demonstrate the abiotic redox transformations of a primary Fe 3+ oxide (hematite) and secondary Fe-bearing phases controlling the hydrogen consumption rate within a 100-year storage period.It is further proposed that low-temperature fluctuations in dissolved iron concentration (or activity) can change the overall redox equilibrium of subsurface systems and thermodynamically drive Fe mineral redox transformations.The current study highlights the need for studying the effect of abiotic mineral alterations on hydrogen storage in various subsurface settings like sedimentary basins, oil field reservoirs, and aquifers at a larger scale and temporal range.

Initial Brine Composition
Our simulations assume a hypothetical situation with 1 kg of a brine initially in contact with 1 kg of Fe 3+ -bearing mineral (hematite) and 1 bar of gaseous H 2 .Theoretical calculations are used here for long-term simulation (100 years) of a brine composition (see Table 1) that is initially in equilibrium with 1 kg of hematite and an initial dissolved H 2 activity of 10 −4 , calculated using Henry's Law (~1 bar of equilibrium pressure, after [27][28][29]).This low activity is also assumed to mimic the low solubility of hydrogen in the heterogeneous aqueous phase that is in direct contact with different minerals [30].Note that it is also emphasized here that only a fraction (~20%, after [31]) of the entire dissolved H 2 pool effectively takes part in the reactions inside pore fluids.The critical basis for most of the numerical simulations is that pO 2 and pH 2 (or respective activities) are often used interchangeably.Meanwhile, in many anoxic reservoirs, effective oxygen fugacity (or activity) may be far below the detectable limit (<<10 −15 ).Therefore, only hydrogen is assumed to be present in the aqueous medium, solely controlling the redox state instead of free oxygen.Due to its low density, gaseous hydrogen tends to accumulate at the top of the reservoir, therefore decreasing the overall opportunity to react with the reservoir minerals at a larger scale.This activity assumption also aligns with known solubility studies on hydrogen at high ionic strengths and pressures [32][33][34][35].The method for activity estimation is provided elsewhere [34] and not discussed here in detail.As a further validation of our assumptions, the initial H 2 activity used in this study (~10 −4 ) also correlates well with low-pressure estimates in [36] as well.Li et al. (2018) [37] demonstrated that, even without the presence of other major gasses, water vapor alone can effectively decrease the hydrogen pressure at elevated temperatures; therefore, the basis for simulations presented here (i.e., low hydrogen content) must be more relevant for further investigations on long term subsurface hydrogen storage [28,38,39].The pH assumed (7) in this study is not an equilibrium value, but is forced to set the initial brine composition in the Geochemist's Workbench software used here.This approach is validated on the basis that the total charge imbalance is <1% even if a pH of 7 is imposed.Moreover, the software readjusts the pH from the beginning of all the simulations which is close (~7.5) to the initial value.Bicarbonate activity is kept low (10 −3 ) to mimic conditions with low dissolved inorganic carbon.For simplicity, no initial organic carbon is considered in our calculations.Initial Fe 2+ activity varies between 10 −3 and 10 −7 to determine the overall impact of this ion on hydrogen consumption.Note that, for each case of different Fe 2+ activities, the total amount of initially dissolved iron is constant.Other major ionic activities are provided in Table 1.The effect of Si or Al is minimized, complying with the observations of very slow kinetics.The B-dot activity model used in the current version of the Geochemist's Workbench (GWB) v.17 software considers the effect of a wide range of ionic strength (0-3 molal) on the activity coefficients of significant ions [40], and is applicable to our simulations (ionic strength ~0.1 molal).Note that, to avoid site specificity, activities instead of absolute concentrations are used.

Model Setup
Our modeling basis is based upon the fact that most storage reservoirs are watersaturated, i.e., the pores are filled with aqueous media and other dissolved gasses like CO 2 and H 2 S, which are often present in a larger quantity to further dilute the gaseous H 2 .The equilibrium modeling approach computes the overall magnitude of completed mineral reactions and frequently overestimates the amount of precipitated/dissolved minerals [41,42].On the other hand, kinetic batch modeling estimates the time-dependent mineral dissolution/precipitation and considers the realistic rates of mineral reactions.The React module of Geochemist's Workbench (GWB) v.17 is used for all the kinetic simulations.To evaluate the effects of reductive dissolution of a Fe 3+ oxide (hematite) in the presence of dissolved H 2 , numerical simulations are carried out.The abiotic kinetic rate constant for its dissolution is taken from Palandri and Kharaka (2004) [43] and projected at 100 • C (rate constant k 2 ) using an Arrhenius type relationship, as follows: where E a (activation energy) = 66.2 KJ/mol and k 1 (at T 1 of 25 The kinetic mineral dissolution rate (r) is calculated as follows: where A is the specific surface area of the mineral (for hematite, 20 cm 2 /g), k 2 is the rate constant at a T 2 of 100 • C, Q is the ion activity product, and K is the equilibrium constant for the reaction, as follows: ∆G rxn = −166.0kJ.mol −1   Due to uncertainties and the lack of experimental validation of precipitation/dissolution rates of secondary iron-bearing minerals, their thermodynamic transformation reactions are assumed to be in equilibrium.Bicarbonate-methane and acetate-methane redox pairs are decoupled, as their activation requires microbial presence.Also, the high mobility of Fe in the subsurface suppresses the sulfate-or bicarbonate-reducing capability of microbes due to metabolic competition [44,45].The model considers no influence of dissolved sulfate as the absence of biology delays sulfate reduction at such low temperatures [46][47][48].Even for pyrite alteration through hydrogen, the current simulation conditions are unfavorable for the reaction [16] and, therefore, can be neglected.The suppressed phases were pure quartz, magnesite, dolomite, and pure FeO(c), which are not generally observed at lower temperatures.Additional simulations are also carried out under different conditions (P (Pressure), T (Temperature), etc.).

Results and Discussion
Due to the slow abiotic iron oxide reduction kinetics at near ambient conditions (e.g., 25 • C, low P), we will focus our discussion on conditions where high P and T can abiotically trigger mineral transformations.Many reservoirs can approach more extreme conditions and therefore are of primary interest in this study.The modeling results of a water-mineral ratio of ~1 and at 100 • C suggest that, within the simulation period of 100 years, secondary Fe minerals precipitate and re-dissolve due to progressive changes in equilibrium (Figures 1 and 2).For the case of high dissolved Fe 2+ activities of more than 10 −4 , an initial decrease in hydrogen activity by almost an order of magnitude is observed within the first couple of years (Figure 1a).Also, within this period, siderite forms as the sole secondary phase.Only after reaching maximum siderite saturation does secondary magnetite start forming, by replacing siderite as the dominant iron phase (Figure 2a).The decrease in H 2 activity is significantly lowered after this maximum siderite saturation and remains within a similar order (~10 −5 ) for the rest of the simulation period (Figure 1a).For intermediate values of Fe 2+ activities (~10 −5 ), siderite becomes completely replaced by magnetite after 60 years (dashed line in Figure 2b), and a sharp decline in H 2 activity can be observed in Figure 1b (dashed line).H 2 activity reaches 10 −6 from the initial value of 10 −4 at the end of simulation period.Notably, magnetite formation also reaches a near steady state situation after 60 years, when siderite completely disappears from the system (Figure 2b).
Finally, for very low Fe 2+ activities (<10 −6 ), calcite but not siderite becomes supersaturated, and hematite dissolution gives rise to secondary magnetite only (reaction (4)), without stabilizing any siderite (Figure 2c).This causes a relatively rapid decline in hydrogen activity by three/four orders of magnitude, as can be seen in Figure 1c.Calcite also re-dissolves slightly after initial supersaturation (Figure 2c) and buffers the pH for the rest of the simulation period.Hematite reduction results in proton consumption (reaction (3)) which probably stabilizes the calcite (reaction (5)) throughout the simulation period of 100 years.For intermediate values of Fe 2+ activities (~10 −5 ), siderite becomes completely replaced by magnetite after 60 years (dashed line in Figure 2b), and a sharp decline in H2 activity can be observed in Figure 1b (dashed line).H2 activity reaches 10 −6 from the initial value of 10 −4 at the end of simulation period.Notably, magnetite formation also reaches a near steady state situation after 60 years, when siderite completely disappears from the system (Figure 2b).
Finally, for very low Fe 2+ activities (<10 −6 ), calcite but not siderite becomes supersaturated, and hematite dissolution gives rise to secondary magnetite only (reaction (4)), without stabilizing any siderite (Figure 2c).This causes a relatively rapid decline in hydrogen activity by three/four orders of magnitude, as can be seen in Figure 1c.Calcite also redissolves slightly after initial supersaturation (Figure 2c) and buffers the pH for the rest of the simulation period.Hematite reduction results in proton consumption (reaction (3)) which probably stabilizes the calcite (reaction (5)) throughout the simulation period of 100 years.

Dissolved Fe 2+ Activity and Redox Interactions
Oxidized iron (Fe 3+ ) participates in several abiotic redox reactions involving natural electron donors such as hydrogen, acetate, sulfide, and ammonia, and subsequently

Dissolved Fe 2+ Activity and Redox Interactions
Oxidized iron (Fe 3+ ) participates in several abiotic redox reactions involving natural electron donors such as hydrogen, acetate, sulfide, and ammonia, and subsequently changes their concentrations in the aqueous systems [49].More reactive minerals such as ferrihydrite, goethite, or adsorbed Fe species on clay surfaces are more ubiquitous in nature but their rates of reductive dissolution in the presence of hydrogen are uncertain.Such uncertainty is addressed here by imposing a relatively higher dissolution rate constant upon hematite, which is within the range of studies with other reducing gasses [50,51].Microbial activities enhance Fe oxide dissolution through H 2 ; therefore, the results shown here must be the lower estimates for reduction, as only abiotic rates are considered.For this study, the water-mineral ratio is set at one to minimize thermodynamic variabilities and impose close-to-real conditions on Fe oxide dissolution.
Fe 2+ can act as a major electron donor to yield more oxidized products such as magnetite, even under anoxic conditions [22,[52][53][54].Milesi et al. (2016) [55] have suggested that ferrihydrite, siderite, and magnetite transformations are often observed in reservoirs with variable amounts of gasses.Regarding the behavior of two or more gas concentrations (or equilibrium pressures), current simulations align with previous investigations [56], which have examined the chemical behavior of a reducing gas SO 2 in the presence of CO 2 inside a reservoir.Their results implied that the primary hematite completely dissolved and gave rise to subsequent phases such as siderite and pyrite.Experimental studies [57] imply that low temperature and low pH reduction of Fe 3+ bearing phases can preferentially precipitate siderite instead of sulfides.Within 14 days, secondary siderite formed in their experiments.
Siderite dissolution and subsequent replacement by Fe 3+ bearing phases (magnetite) are observed in our simulations, which cause a relative decrease in hydrogen consumption rate.In another set of simulations where siderite precipitation was suppressed, such a delayed decrease in H 2 activity (or corresponding equilibrium pressure) was not observed.This gradual decrease in consumption can be attributed to either one (or combined) of the following two processes: (1) Simultaneous magnetite generation from siderite that causes in situ H 2 generation through the following reaction: (2) The release of Fe 2+ from siderite protonation or hydrolysis causing a suppression effect on hematite reduction through the following reactions: The accumulation of Fe 2+ in the aqueous phase thermodynamically suppresses further reductive dissolution of hematite.The release of Fe 2+ into the aqueous phase is supported by a slight increase in Fe 2+ activity after maximum siderite saturation, but this negligible increase may not fully account for the suppression of hematite reduction solely; therefore, abiotic H 2 formation is suggested as an additional mechanism for overall H 2 consumption pattern.Recently, it was proposed that low-temperature hydrous alterations of substantial amounts of Fe 2+ -bearing minerals generate sufficient quantities of hydrogen through the redox disproportionation of Fe 2+ in aqueous phases, which can also alter the redox stability of subsurface reservoirs [22,54,55,[58][59][60][61][62].The redox disproportionation of Fe 2+ in the presence of water produces Fe 3+ and H 2 .The latter exsolves from the aqueous phase once its solubility is reached.The produced Fe 3+ is expected to be removed from the solution with secondary mineral phases such as magnetite and oxyhydroxides, due to their low solubility.At more alkaline conditions, the effect of abiotic Fe 2+ disproportionation becomes apparent [53].Our study emphasizes that in situ abiotic hydrogen generation through reaction ( 6) may play a crucial role in controlling the long-term H 2 concentration in reservoirs [45,53].A summary of the major findings from the simulations is presented in Table 2.

Fe-Bearing Phase Transformations in Natural Systems
Iron-bearing clays, oxides, sulfides, and carbonates are more abundant in hydrocarbon reservoirs or aquifer sandstones [27], and eventually control the variable Fe 2+ content (activity) in the formation waters.Previous thermodynamic predictions indicated that Fe 3+ oxyhydroxide precursors must be spontaneously reduced with dissolved hydrogen due to negative ∆G rxn for oxide dissolution, especially when dissolved Fe 2+ activities are much lower than 10 −5 at neutral pH [62].This is further substantiated through our simulations.However, studies have mainly focused on low-temperature Fe oxide precipitation-dissolution behavior, due to their abundance and relative ease in synthesis [55,57,58], without considering the effects from other dissolved ions.Interactions between more ubiquitous phases such as Fe-hydroxides, surface Fe 3+ -complex, colloids, or Fe 3+ -clay and hydrogen may also show similar redox behavior in reservoirs [16,26], which raises concern regarding the applicability of the aforementioned studies.The majority of abiotic geochemical investigations have so far concentrated on simple mineral systems such as pyrite-pyrrhotite, hematite-magnetite, or ferrihydrite-Fe 2+ (aq) under the more extreme physical conditions of T ~200 • C and P > 20 MPa, due to accelerated reactions [16,17].Such redox processes are recently being appreciated to take place at temperatures below 100 • C in anoxic sediments, to consume/produce hydrogen [54,[58][59][60].Our study therefore corroborates this previously overlooked notion.
Geologically rapid iron-bearing mineral precipitation/dissolution takes place as a part of low-temperature authigenesis in hydrocarbon reservoirs [51][52][53].Such abiotic transformation rates vary from months to years under low T and near neutral pH conditions [41], imposing a further constraint on the aqueous Fe 2+ content in reservoir water.High/moderate concentrations of molecular hydrogen can abiotically promote Fe 3+ -bearing mineral dissolution within variable timescales under various thermodynamic conditions [15,17].Yekta et al. (2018) [17] mentioned that variation in the water-mineral ratio and reduction rates are primarily responsible for H 2 loss through hematite dissolution.In this study, we show that even very low quantities of H 2 (~1 bar) can be potentially preserved under low water-mineral ratios if the dissolved Fe 2+ content is high.This is also a function of kinetic rates of secondary Fe phase dissolution and/or precipitation, which is described later.A comprehensive study on the low-temperature (<100 • C) relationship between various quantities of H 2 and abundant iron-bearing phases is thus required for both long-and short-term storage considerations.

Possible Impact of Variable Ion Activities
Although changes in water-mineral ratios, pH, and temperature would also constrain such 'threshold' Fe 2+ activities in the aqueous phase, the results presented here are significant in terms of the pace of such abiotic reduction kinetics at relatively low temperatures, often encountered in reservoirs.It is acknowledged that variable bicarbonate activities may also change the overall relationship of such simple siderite-magnetite transition; especially when variable activities of dissolved Na, Ca, and Mg can buffer the effective bicarbonate activity available for initial siderite precipitation/dissolution (reactions v, vi).Current simulations have only focused on simple abiotic phase transformations that are the basis for future investigations with variations in other common cations like Na, Ca, Mg, and anions like bicarbonate and sulfate, as well as in the presence of microbial catalysis [45][46][47][48][49]. Siderite precipitation and dissolution in nature are more easily achieved than other carbonates at circumneutral pH, further strengthening the impact of dissolved iron on H 2 behavior.Furthermore, in an additional simulation run with decoupled redox between sulfate and bisulfide pairs, pyrite or amorphous FeS is found to be thermodynamically stable but has no overall impact on the distribution of iron or other major ions and minerals, agreeing with previous observations [45,46].Additionally, in conditions where the redox between sulfate and bisulfide is coupled, pyrite or amorphous FeS precipitation still does not become thermodynamically feasible without considerable biological mediation [49].This further strengthens the impact of siderite as the key controller of hydrogen quantity inside different reservoirs.Note that we are not suggesting that amorphous FeS formation will not dictate the Fe 2+ activity in the aqueous phase, but merely alters the thermodynamic stability in the simulations presented here.Systems with high sulfide activities may yield different results that need to be tested experimentally.Therefore, in many natural settings where biological activity is scarce, the abiotic transformation of iron-bearing phases must play a major role in determining the hydrogen quantity.The current study demonstrates that sedimentary reservoir oxidation due to gradual H 2 consumption can be effective within a few years if dissolved Fe 2+ activity is very low (<10 −6 ).Obtaining actual geochemical data from specific locations will therefore be beneficial for further refinement of the model.Different strategies for long-term storage cycles are needed to be evaluated for such variable H 2 consumption in the presence of different quantities of iron.

Uncertainties Related to Rates of Abiotic Reactions
The simulations presented here were carried out for a time period of 100 years to understand the long-term effects of hydrogen storage on sedimentary reservoirs.However, one key uncertainty pertaining to the models is the abiotic rate of mineral transformation.As mentioned before, siderite-magnetite dissolution or formation reactions are thought to be controlled by equilibrium with the aqueous phase due to the absence of robust kinetic data.This assumption may be valid for longer storage periods and high temperatures, such as the ones used here.Larger uncertainties related to their formation/dissolution rates will lead to deviations from model predictions.Additionally, we have not included the dissolution or precipitation rates of other potential minerals such as calcite and FeS, and consider them to form in equilibrium with the brine.This does not pose an overestimation problem as these phases are previously noted to form rapidly (within days) inside reservoirs [43,45,56].Assessing the dissolution-reprecipitation of clays and/or surface species is not within the scope of the current study.
In this study, the upper limit of temperature is assumed to be 100 • C, to show the maximum possible extent of abiotic mineral transformations.Many depleted oil and gas reservoirs reach this temperature limit and the abiotic kinetic rates of iron mineral alteration are faster [54].On the other hand, lower-temperature reservoirs might need biological mediation for these reactions.Reaction rates are often uncertain at low temperatures and, therefore, can become negligible from a practical point of view.Nonetheless, additional simulations carried out at lower temperatures (down to 50 • C) and/or higher pressures (up to 100 MPa) do not change the abiotic results significantly in the models, as the calculated kinetic rate constants do not change by orders of magnitude.More refined experimental studies should address this problem.
Lastly, the presence of microbes may greatly alter the model predictions due to increased loss of hydrogen over a longer storage time [63].To avoid complications and site specificity, only the abiotic results were targeted as the primary goal of this paper.This study focuses only on the abiotic reduction rates obtained from previous literature, and therefore may not be realistic when natural storage systems are considered.However, at the temperature assumed for the simulations, microbial activity such as methanogenesis or bacterial sulfate reduction may be diminished [20,63].Microbial growth and decay are complex processes, primarily governed by pH, T, nutrients, etc.Furthermore, the initial H 2 activity (<10 −4 ) is probably too low to sustain large-scale microbial colonies [25]; therefore, abiotic mechanisms are more relevant.Even if microbial presence consumes a significant amount of hydrogen within the first couple of months from the injection, our modeling shows that the redox-sensitive mineralogical changes inside reservoirs will still be controlled by the dissolved iron content in long-term storage.Transformations between siderite, magnetite, and other common iron-containing phases are also facilitated by microbes [46]; therefore, the results presented here may only be accelerated, but the overall prediction can remain the same.Our study is only a first-degree prediction for different sedimentary reservoirs of the world that can be targeted for long-term hydrogen storage.Experiments in the presence of microbes must be performed to validate our work.

Uncertainties Related to Reaction Dynamics
Studies that mostly considered equilibrium phase transformations [14,17,26] have overestimated the degree of chemical alteration of the geological systems due to consideration of complete reactions.Under near-surface conditions, however, reactions are rarely complete and rely upon the availability of reactant components which itself is a function of time and kinetics.This study does not consider the effect of diffusive movements of dissolved ions and gasses; therefore, it only takes into account a 0D chemical approach to understand mineral-brine-H 2 interactions.Our kinetic simulations therefore serve as the basis for future experiments that will deal with subsurface reactive-transport mechanisms triggered by the injection of hydrogen.
On the other hand, other iron-bearing minerals such as goethite and ferrihydrites may be more abundant in many sedimentary reservoirs [46], but kinetic parameters associated with their dissolution or formation are either scarce or highly variable.The current study is a modeling study that uses hematite as the primary Fe 3+ -bearing phase with established kinetic parameters.Changing the surface area or kinetic rate constants can be used as proxies for the other iron-bearing phases but needs further confirmation from controlled experiments.These may still vary by orders of magnitude as natural systems also contain surface species, organic matter, and other reactive dissolved gasses like H 2 S. Risk assessments and site selection strategies based on the models will be the focus of future studies by our research group.

Conclusions
The simulation results show, for the first time, that long-term underground hydrogen storage incorporates the risk of hydrogen loss through abiotic redox alterations of iron minerals.In particular, the following are true:

•
Dissolved Fe 2+ activity in the brine can directly impact the long-term storage of H 2 .Very low activities (and, in turn, concentrations) accelerate the abiotic hydrogen consumption within a few years.

•
Redox transformation between siderite and magnetite can play a crucial role in the H 2 consumption behavior.
Based on the model findings, it is recommended that a minute geochemical survey must be carried out before selecting sites for hydrogen storage.The presence of microbes may facilitate the entire thermodynamic relationship presented here and will be more detrimental to long-term storage considerations, but abiotic relationships must not be neglected.Higher quantities of H 2 may be required in such cases to sustain storage efficiency, which needs further experimental validation.Brines with higher ionic strengths and different ion concentrations must be studied on a field scale to understand more about the complex interactions between hydrogen and reservoir components.
Funding: This research was funded by the Department of Energy Office of Basic Energy Sciences, grant number DE-SC0023489 to S. Sharma.

Table 1 .
Major components of the brine used for modeling.

Table 2 .
Summary of the results (100-year period).