In a previous blog I explored the promise and pitfalls of natural or “geologic” hydrogen — hydrogen that forms spontaneously in the subsurface through a variety of chemical reactions. The conclusion was sobering: whilst the subsurface is evidently generating more hydrogen than what scientists once assumed, finding accumulations of natural hydrogen gas that can produce commercially at scale may be difficult and such accumulations may in fact be very rare.
A growing number of scientists, engineers and entrepreneurs are therefore asking a logical follow-up question: if nature can generate hydrogen underground, why not engineer the process ourselves? Rather than searching for rare natural accumulations, why not inject water into suitable rocks, trigger the reaction deliberately, and produce the hydrogen we create? This concept — known as “stimulated hydrogen” or “gold hydrogen” — is attracting serious research attention by scientists and technology startups including some early field trials.
It is a genuinely clever idea. But does it work in practice? Here is what the science currently tells us.
Alternative routes to stimulated hydrogen
Researchers have proposed various approaches to generating hydrogen underground.
Some technologies have been around for decades — at least conceptually and at lab/small pilot stage — in the context of enhanced oil recovery (EOR) methods. One such approach involves induced microbial degradation of oil in depleted reservoirs. Certain microbes, when injected into a depleted oil field, can break down residual oil and generate hydrogen as a metabolic by-product. A small-scale field trial of this approach has reportedly been conducted in the San Joaquin Basin in California, though details remain sparse.
Another EOR-related method of stimulating subsurface hydrogen is in-situ combustion or pyrolysis of natural gas in depleted oil or gas reservoirs, using heat and steam to convert residual methane into hydrogen underground. This is essentially subsurface steam methane reforming — a well-known industrial process, but one that has never been demonstrated in an actual subsurface reservoir, let alone at scale.
A recent review paper by Alkan et al (2024) describes the various methods of hydrogen production from depleted oil and gas field as technologically challenging and extremely costly (in the range of >5 to 40 USD per kg of hydrogen).
A totally different approach to stimulating subsurface hydrogen production is enhanced serpentinization of ultramafic rocks. Ultramafic rocks — iron- and magnesium-rich basement rocks found at depth across many parts of the world — react with water in a process called serpentinization, releasing hydrogen as a by-product. By injecting water into these formations, the idea is to accelerate a reaction that nature already performs, with the target to achieve rates of in-situ gas generation that are commercially useful.
The case for stimulating ultramafic rocks
With the recent spur in interest for subsurface hydrogen including its stimulation, enhanced serpentinization has attracted by far the most scientific attention, and for good reason. Unlike microbial degradation or in-situ combustion, methods that subject a valuable resource (namely oil and gas) to risky and difficult to control reactions and processes, serpentinization only interacts with the rock. Ultramafic rocks are not rare or highly valuable: while less common than ordinary crustal rocks, they occur in significant quantities worldwide, particularly in ancient continental cores and along the remnants of old ocean floors known as ophiolite complexes. The chemistry is well understood. The reaction is exothermic — it releases heat — which in theory helps sustain itself once started. And unlike hunting for a rare natural accumulation, you are engineering the process at a location of your choosing.
Laboratory and desktop studies have identified the key conditions for maximising hydrogen generation: a high magnesium-oxide content in the rock, a very dense network of micro-fractures to maximise the contact area between rock and water, large volumes of injected water and — critically — sufficiently high temperature, ideally between 200 to 300°C. At lower temperature, reaction rates are orders of magnitude less than the optimum.
Here lies the first practical challenge. For a typical continental geothermal gradient, this temperature range translates to depths of 5,700 to 9,000 metres. Even in areas with higher-than-average heat flow, you are still looking at targets of at least 3,000 metres depth. These are not shallow, cheap wells.
Generating hydrogen is the easy part
Most published research on stimulated hydrogen focuses on the generation side: what rock types work best, what temperatures are needed, what fracture networks are optimal. This is important science. But it largely sidesteps what may be the harder problem: once you have generated hydrogen in a deeply buried rock, how do you actually get it to the surface?
To understand why this is challenging, consider what is happening underground. Water is being injected into fractures in an otherwise tight rock. The rock around those fractures begins to react, generating small quantities of hydrogen gas. But here is the catch: to keep the reaction going, water saturation in the fractures must remain high. The rock essentially needs to stay wet. This means that the hydrogen gas, once generated, finds itself as a small minority phase surrounded by large quantities of water.
With increasing depth and pressure, hydrogen gas is compressed to a fraction of its surface volume. Combined with the relatively modest quantities generated relative to the large volumes of injected water, this means that gas saturation in the stimulated fractures remains very low. And in reservoir physics, low gas saturation means low relative permeability to gas: the rock will preferentially allow water to flow rather than gas. Any attempt to draw fluids to surface will therefore predominantly produce water, with only a small fraction of hydrogen gas.
This notion is not just theoretical. Work by two independent research groups — Sekar and Okoroafor (2025) and Egert et al. (2025) — illustrates these challenges with numerical modelling results. They find that even in optimistic model scenarios using horizontal wells and huge volumes of water injected with speculative chemical catalysts that accelerate the reaction tenfold, hydrogen yields over a 30-day production period nevertheless remain modest: between 2,800 and 4,600 cubic metres [1]. Meaningful production in these simulations only occurs under conditions of extraordinarily high fracture permeability and very wide damage zones around the well — conditions that may be extremely difficult to achieve in practice even in ideal subsurface settings.
What the first field pilot tells us
The Rock-Hydrogen project in Oman, led by Professor Alexis Templeton of the University of Colorado, is the world’s first dedicated field pilot of stimulated hydrogen from ultramafic rocks. In early 2024, the team drilled to 1,050 metres, cased off the upper 900 metres, and left 150 metres of relatively fresh peridotite — an ultramafic rock — exposed in the wellbore. They then pumped around 4,800 cubic metres of water into the well and shut it in for a year to allow serpentinization to proceed.
When flow testing began after twelve months of shut-in, the well reportedly produced roughly equal volumes of gas and water, with the gas being predominantly hydrogen. As a proof-of-concept this is encouraging: despite shallow depth and low temperature, water injection apparently initiated enough serpentinization to result in producible hydrogen.
However, indicative assessment of the scarce quantitative results released from this pilot paints a more sobering picture, especially on the recovery side [2]. Based on typical, literature-quoted hydrogen yield and serpentinization-related water consumption, some 1,100 cubic metres of the injected water may have been consumed by serpentinization resulting in an approximate 4,800 kg of hydrogen generated in the subsurface. Optimistically assuming that all the unused water was flowed back to surface during testing together with hydrogen gas at an approximate 1:1 gas-to-water ratio, hydrogen recovery may amount to some 330 kg— a recovery factor of around 7%. The remaining 93% of stimulated hydrogen either stayed underground — dissolved in water or as a residual gas unable to flow to surface — or escaped from the stimulated interval to shallower layers or to surface.
It is worth noting that this Oman well stimulated ultramafic rocks at a temperature of around 55-60°C only, much lower than the optimum for serpentinization. A deeper, hotter well might generate hydrogen at a much faster rate. But on the other hand, a deeper well may face even more severe producibility challenges due to higher pressure (compressing the generated gas and hence, reducing gas saturation) and tighter rock.
Other complications lurk below the surface
High water production for a modest hydrogen yield is not the only problem. Serpentinization involves mineralogy changes that cause the rock bulk volume to expand by 20 to 40%. This volumetric expansion tends to reduce permeability — possibly by orders of magnitude — with the risk of choking off both the water supply needed to sustain the reaction and the pathways available to draw the generate hydrogen towards and into a well.
At surface, the challenges continue. Produced hydrogen will inevitably be saturated with water vapour because its generation involves water. Producer wellhead pressure and flowrate are likely to change rapidly since the well is drawing fluids from a confined set of fractures in an otherwise tight rock. Separating water from gas under these conditions — using methods like temperature swing adsorption or membrane dryers — will be technically challenging and so will any required gas purification (separating hydrogen from other produced gases such as CO2 and methane).
For the microbial stimulation approach, similar producibility concerns apply. In a depleted reservoir where most of the oil has been swept by water, microbial stimulation may introduce small amounts of hydrogen gas but mobility of this gas will be low compared to the high mobility of water. Drawing these small, diffuse amounts of hydrogen to a producer well will therefore be very difficult. The alternative would be to wait for the hydrogen gas to slowly migrate upward through the reservoir and eventually form a secondary gas cap, which could subsequently be drilled and produced. However, gas migration could take a long time, secondary gas caps could be difficult to locate and loss of hydrogen through diffusion or microbial consumption (producing methane or H2S, a highly poisonous gas) are additional risks.
The final take
Stimulated hydrogen is a legitimate and intellectually compelling idea. The underlying chemistry is real, the geological settings exist in many parts of the world, and early laboratory and field results confirm that hydrogen can be generated underground through engineered processes. As a long-term research direction, it deserves continued scientific attention.
But the gap between generating hydrogen in a rock and producing it commercially to surface is large — and, based on current evidence, significantly underestimated in much of the popular and even scientific coverage of this topic. Reservoir physics, fluid mechanics and basic well engineering all point to the same conclusion: getting the gas out is harder than stimulating the subsurface to make it.
The Oman pilot is an important first step. A 330kg hydrogen yield from one month of flow testing, however, is orders of magnitude below a commercial case. Rather than more inflated resource estimates or unrealistic predictions for the time-to-commercial deployment, stimulated hydrogen needs focused engineering work on the producibility problem: novel well architectures, production strategies that can overcome water loading and surface handling, and a honest assessment of what production rates and recovery factors are realistically achievable.
Until that work is done, stimulated hydrogen should be regarded as a promising but highly immature technology — one whose resource potential in the ground substantially exceeds its current recovery potential.
Footnotes
- For context, the naturally-producing Bougou-1 well in Mali — the world’s only documented natural hydrogen producer — produces around 1,500 m³ of hydrogen per day. The optimistic simulated scenarios thus yield only 2–3 days’ worth of Mali-equivalent production over an entire 30-day production cycle.
- These figures are indicative estimates based on publicly available information about the Rock-Hydrogen pilot (GeoExpro, January 2026) and standard serpentinization stoichiometry. They have not been independently verified by the project team and should be treated as order-of-magnitude assessments only.
