A natural hydrogen seep in Western Australia: Observed characteristics and controls

. Natural hydrogen (H 2 ) is a promising resource for the energy industry ’ s transition to zero-carbon fuels. However, its extent and feasibility for exploitation remain unclear. A key step towards discovering subsurface dihydrogen accumulations is detecting H 2 seeps. This study presents the ﬁ rst autonomous, multi-gas monitoring of a natural hydrogen seep in Australia, where dihydrogen, carbon dioxide, and hydrogen sulphide were measured together. The research revealed signi ﬁ cant H 2 seepage on the Yilgarn Craton in Western Australia, with seasonal ﬂ uctuations: high emissions after dry summers and reduced emissions following rainfall due to increased groundwater levels. Groundwater appears to act as a seasonal inhibitor to H 2 seepage through the near subsurface potentially leading to false negatives in soil gas surveys post-rainfall and in low-lying areas. This work provides fundamental data for natural hydrogen exploration and therefore aids in the implementation of a large-scale hydrogen economy.


Introduction
The reduction of carbon dioxide emissions by 2050 requires the rapid development of alternative clean energy industries [1].One such alternative is using hydrogen (H 2 ) produced from renewable resources, producing pure water as the only waste product [2][3][4][5][6].
Manufactured hydrogen is generated by a number of mechanisms such as gasification of coal (black hydrogen), steam reforming of methane (grey hydrogen), pyrolysis of methane (turquoise hydrogen) and electrolysis of water (green, yellow, or pink depending on the power source) [7,8].In such cases, the hydrogen acts as an energy carrier for the original energy source.Natural hydrogen is directly produced from a set of geological mechanisms including radiolysis, organic activities, tectonics, and mineral alterations, and is thus a primary energy source.Natural hydrogen has an estimated full lifecycle greenhouse gas intensity of one-third that of green hydrogen [9] making it advantageous for the environment, but also more energy efficient than other forms of hydrogen.
With recent natural H 2 discoveries in Mali [10], France [11], and South Australia [12], exploration has gained strong interest internationally [10,[13][14][15].However, natural H 2 exploration is in its infancy and little data is available.Past petroleum wells rarely recorded dihydrogen gas concentrations, making this new field of research challenging.
Detecting dihydrogen surface micro seeps may be an indicator of deeper accumulations, however, yet there are no standard procedures to sample H 2 at the surface.It is therefore critical to standardize exploration techniques for soil gas sampling in the field [16].Point sampling for natural H 2 in soils has been documented by several authors [17][18][19], but continuous and long-term studies of natural H 2 seeps are limited to two sites in the Sao Francisco Basin in Brazil [20][21][22].
Variable H 2 gas emissions were measured from the soil in and around sub-circular depressions in Brazil.These sub-circular depressions may appear as salt lakes and are thought to relate to the occurrence of H 2 in the subsurface [14,21,23,24].
No studies to date have continuously monitored natural H 2 seepage in Australia or have combined long-term H 2 with other trace gases to better constrain the potential source.Additionally, prior to this work, long-term autonomous monitoring has been focussed in and around areas with sub-circular depressions; thus, significantly more research is required to enable efficient natural H 2 exploration and production.
We here present the first long-term monitoring of a natural dihydrogen seep in Australia.In a novel approach, carbon dioxide (CO 2 ) and hydrogen sulphide (H 2 S) were monitored together with H 2 , to determine whether associated trace gases could assist in identifying the geologic source.The Yilgarn Craton site represents the first natural H 2 seep identified on the flood plain of an ephemeral stream, as opposed to in and around sub-circular depressions.
We identify the scale and variability of the natural H 2 seepage and discuss potential endogenic, environmental, and climatic controls at the site named FF4 and the implications for exploration.This work therefore advances the natural H 2 energy industry and aids in the development of a low-carbon energy future.

Study location and geological context 2.1 Yilgarn Craton
The study area is located on private property on the Yilgarn Craton, in the southwest of the state of Western Australia (Fig. 1a) where bedrock comprises extensive Archean granitoid and greenstone terrains [25][26][27].The greenstones comprise metamorphosed basalts, gabbros, chlorite schists, and serpentinites and are interbedded with metamorphosed sediments and banded iron formations [28,29].They form arcuate belts of metamorphosed sedimentary and mafic volcanic rocks that lie between the granitoids at the surface and are interpreted to extend to depths of up to 5 km below the study location [30,31].Granitoids include granites, tonalites, granodiorites, and Proterozoic sedimentary rocks that overlie parts of the Craton [32,33].The Yilgarn Craton comprises an imbricated middle crust, which was subjected to subsequent crustal spreading, reworking, and injection of late plutons via rejuvenated listric faults (Fig. 1d).These reworked faults, imaged on deep seismic data, sole out near the Moho boundary and offer potential deep-seated pathways for fluid migration through the crust [30].
The geomorphology of the study area comprises an ephemeral stream in the centre of an elongated depression, with a wide rim of low-lying salt-affected flat areas for 500 m to 2 km on either side (Fig. 1c).The stream is slow-flowing and meanders seasonally across the salt pan, causing yearly variations in standing water and soil saturation.The groundwater at the FF4 site was acidic (pH 5.75) and highly saline (77,000 ppm), consistent with previous groundwater studies in the western Wongan Hills locality [34].
Point soil gas monitoring was performed at 15 locations in the area in January 2022, six of the locations were drilled within a 1 km radius including the FF4 site (Fig. 1d).The initial FF4 H 2 concentration measured during point sampling was 800 ppm [35] and the decision was made to conduct more detailed autonomous monitoring at that site in order to prove that H 2 was emitting continuously, and not simply the result of drill bit cataclasis [36], and to better understand the variations in natural H 2 emissions.

Boreholes
Three boreholes were drilled for the autonomous monitoring to a depth of between 80 cm and 100 cm initially with a 40 mm diameter auger drill bit used to minimize artifact H 2 generation via drill bit metamorphism [36].Capped PVC tube liners were placed into each borehole comprising a 20 mm diameter PVC pipe.The lower 20 cm of the liner was perforated with 5 mm holes (1 per 2.5 cm 2 spacing).During soil gas measurements, 3 mm PVC tubing was inserted through the tube liner into the perforated interval to draw gas from the base of the hole (80-100 cm below the surface).

Point sampling
Gas was detected during the initial point sampling survey by a handheld iBrid MX6.The iBrid MX6 gas detector was fitted with H 2 , CO 2 , methane (CH 4 ) and H 2 S sensors.Measurement ranges and precisions of the sensors are shown in Table 1.The iBrid MX6 is fitted with an internal pump that pumps gases across the sensors at a rate of 300 ml per minute and was calibrated prior to each survey period.

Autonomous sampling
Two stationary autonomous multi-gas detectors (Axiom Sensing; WHALI & EVA, origin: Australia) were used in this study.
The pumped autonomous logging unit (Axiom Sensing; WHALI, origin: Australia) was equipped with H 2 , CO 2, and H 2 S sensors, temperature, and pressure sensors, and a peristaltic pump capable of pumping gas vapors at 300 ml per minute, similar to the iBrid MX6.The range of the H 2 detector was 0-2000 ppm, with a precision of 10 ppm (Table 1).Automatic measurements were made for 3 min either every 15 or 30 min.
The autonomous unit was deployed at the FF4 site for 5 h on 20th February 2022, 7 days from the 23rd to the 30th of April 2022, and 7 days from the 9th to the 16th of February 2023.
The pumped autonomous WHALI unit shut down after five hours on the first day of operation (20th February 2022) due to groundwater intrusion.A water management system was subsequently added and used during April 2022 and February 2023 to evacuate and dispose of groundwater drawn through the system while pumping soil vapors to the sensors (Fig. 2).
In April 2022 the FF4H site was selected for the autonomous sampling, located 8 m up-dip from the original FF4 site, which was inundated with water.An initial H 2 concentration of 47 ppm was measured at FF4H with the handheld MX6 gas meter.
The site was inaccessible from July to November, and upon returning in November 2022, measured trace gas concentrations in the air space above the standing water in the boreholes were consistent with air, with no H 2 detected [35].
In February 2023 the original FF4 borehole was chosen for the autonomous sampling.
A second autonomous, passive measurement system, i.e. non-pumped (Axiom Sensing; EVA, origin: Australia,) was also deployed at the FF4 site during February 2023 approximately 80 cm away from the WHALI unit to compare the results of the two systems.The passive system was not available prior to that time.
As both autonomous units sit above ground, and electrochemical sensors are sensitive to temperature variations, all H 2 concentrations (cH 2 ) and hydrogen sulphide concentrations (cH 2 S) results displayed herein represent temperature-corrected concentrations according to the manufacturer's specifications [40].The following polynomial equation has been applied: x = Raw concentration y = Temperature corrected concentration z = Measured temperature at sensor Figure 3 shows the results of test measurements in air (H 2 < 1), which were recorded for two minutes every 15 min over a four-day period, where the temperature at the sensor ranged from 8.5 °C to 41 °C.The raw H 2 concentrations represent the maximum recording during each two-minute sampling period.Equation ( 1) was applied to generate the temperature-corrected H 2 concentration.This shows the extent of the sensor reaction due to temperature variation at a peak of 41 °C was 12 ppm, which corrected to 2.5 ppm.As the accuracy of the sensor in the autonomous units is 10 ppm, <10 ppm is insignificant.Therefore, the cH 2 measured can be effectively corrected with the temperature correction equation.The CO 2 sensors are infra-red sensors with internal thermometers and self-calibration, and thus CO 2 concentrations do not require environmental correction.

H 2 concentration
Normal cH 2 in air is <1 ppm, therefore, cH 2 > 1 ppm indicates H 2 seepage.However, as the autonomous monitoring unit had an accuracy of 10 ppm, for this study we consider >10 ppm as significant.Here cH 2 >> 10 ppm was measured  The amplitude of the H 2 concentrations varied strongly between monitoring periods, despite similar temperature and pressure conditions in the summers (February) of 2022 and 2023 (Table 2).In February 2022 a maximum cH 2 of 493 ppm was measured (average 326 ppm), while the highest cH 2 for April 2022 was 330 ppm (average 62 ppm) and the peak cH 2 was 52 ppm (average 26 ppm) for February 2023 (WHALI).The peak cH 2 measured by the EVA system in February 2023 was 180 ppm, but the average (4 ppm) was insignificant.
A diurnal bell-shaped trend was established during monitoring in April 2022 and February 2023, with the highest cH 2 measured in the middle part of the day and the lowest cH 2 in the evenings.

Carbon dioxide concentrations
The carbon dioxide concentrations (cCO 2 ) measured across the monitoring periods ranged from 442 to 15,558 ppm (Table 2), indicating valid soil gas measurements.In February 2022, very high and stable cCO 2 was measured across the limited monitoring period.For the first week of April 2022 monitoring period, the cCO 2 concentrations were high, irregular, and gradually increasing.A more regular oscillating pattern was observed in the second half cCO 2 peaked on the EVA unit in February 2023 at 7570 ppm an hour after installation along with a cH 2 of 16 ppm.Other than the two cH 2 peaks on the 13th of February 2023, concentrations did not reach those levels again.This may have been due to soil gas readings reaching the sensors before the borehole became inundated with groundwater.

Hydrogen sulphide concentrations
Hydrogen Sulphide concentrations (cH 2 S) were insignificant, measuring below the 0.6 ppm precision of the H 2 S sensors (Table 1, Table 2).This result varies from the headspace gas analysis conducted at the site between November 2022 and February 2023, where cH 2 S up to 31 ppm (average 3.4 ppm) was detected with vigorous agitation of the groundwater samples [35].Hydrogen sulphide is slightly soluble in water, but solubility decreases with increasing salinity [42].This suggests that gaseous H 2 S may be held within the groundwater, with degassing requiring vigorous agitation, rather than the gentle water movement created through the WHALI water management system.

Methane concentrations
Methane was measured during initial point sampling at the site in early February 2022 but was not measured by the autonomous units during continuous monitoring.Five measurements were taken using the handheld iBrid MX6 at the FF4 site on the 7th and 8th of February 2022 and measured methane concentrations from 0 to 4500 ppm.Headspace gas analysis of groundwater samples at this site between November 2022 and February 2023 measured methane concentrations from 0 to 6500 ppm [35].

WHALI pumped vs EVA passive autonomous unit
The average cH 2 for the EVA system was approximately one-quarter to one-fifth of the cH 2 measured by the WHALI system at the same time.Because the WHALI system is pumped, it draws a sample of 600-900 ml of soil gases over the sensors, which is derived from the area around the borehole.
Alternatively, the EVA system relies on whatever volume of soil gases migrate up the borehole to the sensor.If there is standing water in the borehole, soil gases must break through the interfacial tension of groundwater to rise to the sensors at the surface.cCO 2 measured by the pumped WHALI unit was approximately ten times that measured the passive EVA unit.The exception to this was one hour after both units were emplaced on the 9th of February 2023, when measurements were similar.This may have been due to there being no water saturation in the borehole for the initial part of that day.Again, the main difference is likely to be the pumping.As CO 2 is significantly heavier than H 2 , a larger concentration differential is required for the gas to migrate up the borehole, compared to lighter H 2 .

Gas samples
Gas Samples were collected from the FF4 site during August 2023 and April 2024, in order to confirm the H 2 gas emitting from the boreholes, as detected by the handheld gas sensors and autonomous sensors.The samples confirmed cH 2 in soil gas from the FF4 site.The maximum cH 2 measured from gas samples was 300 ppm (August 2023) and 530 ppm (April 2024) (Fig. 5).

Soil gas transport mechanisms
Diffusion and advective pressure pumping are the two fundamental processes governing the gas exchange between soils and the atmosphere [43][44][45][46].Diffusion occurs as gases naturally move from areas of high concentration to low concentration, driven by concentration gradients [47,48].This process is slow but operates over short distances, allowing gases to permeate through the soil matrix.The diffusion of gases through soil is influenced by physical parameters such as the density of the gases; atmospheric pressure; total porosity; soil moisture; permeability and fractures facilitating preferential flow [46,47,49].
Advective pressure pumping involves the movement of gases due to pressure differentials within the soil.This can be caused by wind, temperature and atmospheric pressure gradients, leading to the displacement of gases through soil pores [43,45,50,51].

Temperature
Diurnal H 2 emissions occurred at FF4, where cH 2 fluctuated in a bell-shaped pattern, peaking in the middle of the day, and falling to low levels during the cooler evenings (Fig. 4).While cH 2 and temperature are correlative on a daily scale, Figure 6a shows that there was no overall significant relationship between cH 2 and temperature measured at the sensor.
Soil temperatures influence diffusive gas movements through soil and fluctuate both with depth and seasonally due to factors such as solar radiation, air temperature, soil moisture content, and the thermal properties of the soil itself [52,53].The influence of air temperature and solar radiation on soil temperature is highest close to the surface, and diminishes with increasing depth [53].Cheng et al. [52] demonstrated modest seasonal variation in soil temperature at 80 cm in the low-latitude plateau of China (15-23 °C range), but minimal diurnal variation.Zeng et al. [54] found, that in arid climates, diurnal temperature variations can penetrate soils down to 1 m depth, but the hourly temperature gradient variation was low (0-0.5 °C).
The soil gas in this study was drawn from 80 cm to 1 m below the surface, where soil temperatures would be expected to vary seasonally, but not significantly on an hourly basis.Therefore, we can assume that in the summer and autumn months, the soil temperature is higher, enabling increased diffusive flow of all gases.However, soil temperature cannot account for the diurnal flux of cH 2 observed in all cases or the diurnal pattern observed in CO 2 following the precipitation event on the 26th and 27th of April 2022.
The average ambient temperatures, average cH 2 and average cCO 2 were all highest during the February 2022 monitoring.February 2023 had the next highest temperatures and average cCO 2 , but the lowest average cH 2 .This suggests a difference in the main transport mechanisms of CO 2 and H 2 through soils.The average CO 2 concentrations appear to follow the seasonal variations in soil temperature, suggesting the primary mechanism of CO 2 transport may be diffusion.
While the diurnal H 2 variations appear to trend with temperature on a daily time frame, this relationship appears to be correlative rather than causal (Fig. 6a).
The diurnal patterns observed at FF4 are consistent with those documented at the Sao Francisco site in Brazil [22].In that case, the authors suggested that while there was a correlation between temperature and cH 2 , it was coincidental, as the temperature did not vary significantly below 0.5 m depth.Rather, they postulated a The Author(s): Science and Technology for Energy Transition 79, 48 (2024) relationship between atmospheric pressure and nearsurface gas transport with a lag between the decreasing atmospheric pressure and subsequent increase in cH 2 [20,22].

Pressure
Overall, cH 2 and cCO 2 varied substantially across pressure conditions (Fig. 6c & d), suggesting that absolute atmospheric pressure is not a predictor of the amplitude of the cH 2 or cCO 2 .This suggests that an alternative control, such as advection from deep in the sub-surface, or bacterial activity controls the amount of cH 2 in the soil.However short-term relationships were seen and are discussed further here.
On the 13th of February 2023, a large pressure drop was observed on both the WHALI (Fig. 4c) and EVA (Fig. 4d) units.Following the pressure drop, two large H 2 peaks were measured on the EVA system (178 ppm and 180 ppm) and corresponded to the highest cH 2 measurement on the WHALI system (52 ppm) for February 2023.The difference in reaction by the two systems is deduced to be the difference in sampling methods.The passive EVA system does not draw water but relies on passive flux.We propose the large pulse was the result of the H 2 gas pressure building up before overcoming the interfacial tension and forcing through the groundwater to the sensor [55].In contrast, the pumped WHALI system consistently detected the elevated cH 2 over several hours, dispersed within the groundwater, by agitating the sample through the water management system (Fig. 7).cH 2 exhibited diurnal flux throughout all periods of monitoring at the FF4 site.cCO 2 followed a subdued diurnal flux pattern after the rainfall event of the 26th to 27th of April, and throughout the February 2023 monitoring period (on both the pumped WHALI unit and the passive EVA unit).The diurnal pattern is consistent with measured daily drops in pressure (Fig. 4).A similar diurnal flux pattern of cH 2 has been documented in Brazil [20,22,56,57], and diurnal CO 2 flux is well known, although for the latter samples depths are usually shallower and therefore more affected by ambient temperatures [58][59][60][61][62][63].
In both the April 2022 and February 2023 campaigns, small dips in pressure coincided with the peak diurnal H 2 concentration (Fig. 8).
The process of pressure pumping (changing barometric pressure causing upward and downward mass flow of soil gases) as a mechanism for soil gas transport is well known [22, 43-47, 50, 61, 63-65].Gases of different densities react differently to pressure pumping [46].High density gases such as CO 2 are less sensitive to pressure-induced advection, whereas low-density gases such as Helium, are highly reactive to pressure-induced advection [43,44,46,64].
We propose that the strong diurnal flux observed in the cH 2 trend is the result of the low density di-hydrogen molecule reacting to barometric pressure pumping similar to the behaviour of Helium in soil gas [43,64].Figure 8 shows temperature and pressure plots for the April 2022 monitoring period, with both plots coloured according to cH 2 .As pressure falls in response to ambient temperature increase, the H 2 concentrations increase.The asymmetry of the H 2 con- centration with temperature confirms the flow of gas through the subsurface is controlled by pressure, but the daily barometric pressure change is related to temperature, hence the appearance of cH 2 to trend with ambient temperature.
Thus, while neither pressure or temperature control the absolute H 2 soil gas concentrations in the soil, atmospheric pressure appears to enable the movement of H 2 to the surface on a diurnal basis through pressure pumping.

Groundwater levels and soil moisture
Groundwater volumes extracted by the WHALI water management system were measured in April 2022 (20.3 L)  Water levels in boreholes varied substantially between campaigns due to the higher levels of rainfall in the preceding months (Fig. 9).Rainfall was measured by the Bureau of Meteorology at the Wongan Hills Weather Station #08137, located 13 km northeast of FF4 [34].High rainfall in the second half of 2022 led to high groundwater levels [34], and saturated salt lakes in the region throughout the summer of 2023 (Fig. 10).There was a strong negative correlation between the six months rainfall before each monitoring period (BOM, 2023), and cH 2 (Fig. 9).Thus, when no water was initially observed in the FF4 borehole (summer season 2022), peak cH 2 was 493 ppm; however, as the groundwater at the FF4 site rose following significant rainfall the H 2 concentrations reduced.
In Sao Francisco, Brazil, the centres of circular depressions, which were described as flood zones, measured lower H 2 emissions to surface than the surrounding dry areas [21].
The reason for the apparent inhibition of cH 2 flow in moist soils may be due to either higher bacterial consumption in groundwater, or the physical inhibition of H 2 diffusion.
The interfacial tension between H 2 and H 2 O is high at ambient conditions, and diffusivity at surface conditions of dihydrogen in water is close to zero [55].Therefore, porous water-wet rocks, and even water itself, can form an impermeable seal for H 2 gas [10,55,66].
Experimental studies by [67] showed significantly lower H 2 diffiusion rates through water and brine saturated  sedimentary rocks as opposed to dry samples, and higher diffusion rates through sandstones vs claytones.A similar relationship can be expected in soils, where diffusion rates will be lower in brine saturated samples, and indeed is supported by the results of this study.[68] Davies et al. [35] demonstrated that high concentrations (up to 1684 ppm) of natural H 2 was trapped in groundwater and degassed into the headspace of a measurement vessel when agitated vigorously.In this study, the WHALI water management system (Fig. 2), provided significantly less agitation and cH 2 concentrations measured were lower.The high levels of H 2 liberated during headspace gas analysis by [35] suggest that physical inhibition, rather than bacterial consumption, is responsible for the reduced H 2 soil gas emissions measured at the FF4 site during periods of high groundwater.
We propose that during periods of low groundwater levels, H 2 can migrate through the subsurface via feeder systems such as fault zones into the dry soil where it flows to the surface.Conversely, when groundwater levels rise, H 2 diffusion into surface soils may be impeded by water at the terminus of the conduit fault, trapping H 2 in groundwater and preventing its fast movement to the surface.Soil moisture is known to impact the movement gases through soils as it reduces the porosity [46,47,[69][70][71] and CO 2 emissions are lowest when groundwater levels are closest to the surface [72].
While soil moisture was not monitored in this study, we use rainfall as a proxy.During the April 2022 monitoring period, CO 2 emissions were high and generally increasing for the first three days.cCO 2 dropped dramatically when the rainfall occurred on the 26th of April, and thereafter adopted a subtle diurnal pattern similar to cH 2 .We surmise that rainfall increased soil moisture, which decreased the soil temperature and free air porosity, leading to a drop in diffusive CO 2 flux.The diurnal pattern which follows suggests that pressure pumping may be the dominant mechanism of transport for heavier gases only in moist soils [46,63].
This study demonstrated that groundwater impeded the natural H 2 flow to the surface.Such inhibition may lead to false negative readings in soil gas surveys after high rainfall and in low-lying areas.Alternative exploration methods should be considered in such wet areas.

Natural hydrogen source
Understanding the source rocks for natural H 2 can de-risk exploration, by focussing efforts on areas with the most conducive geology.This study aimed to advance understanding through continuous monitoring of multiple gases at the surface.Gas analysis techniques such as isotopic analysis may help to characterize the natural H 2 source rocks and are planned for future studies.While isotopic analysis was outside of the scope of the existing study, we may still postulate the most likely sources with the continuous monitoring data available.Five natural H 2 generation mechanisms are possible at the FF4 site, namely radiolysis, water interactions with iron-bearing minerals, mantle flux, gas release from fluid inclusions, and/or bacterial activities [17,[73][74][75].

Bacterial source
Bacterial community abundance has been shown to be at its highest in the 5-10 cm zone below surface and reduce at depth [76][77][78].The soil gas measured during this study was taken from 80 cm to 1 m below surface, at which depth microbial abundance is low [76,78].
CO 2 and H 2 S are classical by-products of bacterial fermentation [72,[79][80][81][82][83].If the H 2 was produced by bacteria, elevated concentrations of cH 2 S and cCO 2 would be expected, and correlate with cH 2 .Such patterns were not observed at FF4 where cH 2 and cCO 2 did not correlate (Table 2), and cH 2 S was <0.4 ppm during autonomous monitoring.
While further bacterial soil studies are required to verify the microbial influence on the H 2 emissions, the depth of sampling combined with the observed trace gas mixtures suggest that bacterial fermentation is not the most likely source of H 2 at the FF4 site.

Iron-rich minerals and water interactions
The interaction between iron-rich minerals and water, potentially generates H 2 in the ultra-mafic greenstone belt adjacent to FF4, which projects several kilometres into the subsurface [30,31].Geochronological data suggest deformation of the Yilgarn Craton greenstone belts around 2.8-2.6 billion years ago [84][85][86].However, it remains uncertain whether mafic minerals of this age actively produce H 2 or have fully reacted.Similarly, the permeability of these formations and their ability to allow water penetration crucial for H 2 generation, as well as the release of H 2 to adjacent formations, are unknown.
Serpentinization and low-temperature iron oxidation typically produce highly alkaline fluids with a pH of 9-12 [89][90][91], and consistent generation of H 2 gas occurs at pH 8.5 and above [92].As the groundwater at FF4 is acidic (pH = 5.75), the H 2 generating system would need to be separated from the FF4 groundwater via a perched water table, which seems unlikely as it would impede H 2 migration into the groundwater system.For these reasons, serpentinization is not the preferred mechanism for the H 2 produced at the FF4 site.

Mantle flux
The volatile elements of the earth's mantle are predominantly H 2 , carbon, nitrogen, noble gases and sulphur [93].Those elements are vented as volcanic gases; water vapor, CO 2 , SO 2 , CO, H 2 S and H 2 at the surface [94][95][96].The acidic groundwater at FF4 is consistent with a deep mantle-derived fluid source [96], into which abundant CO 2 is dissolved [97].
Based on these observations, it appears that mantle flux is not the most likely source for the FF4 H 2 seep; however, a full-scale geo-chemical analysis should be conducted before dismissing this potential source mechanism including isotopic analysis of CO 2 and noble gases including helium.Such analysis was outside the scope of the current study but is planned for future work.

Fluid inclusion release
H 2 gas is known to have been trapped in Precambrian granites worldwide, and the Archean granitoids of South and Western Australia have some of the highest H 2 concentrations measured [75,100].
H 2 may be released from fluid inclusions by heating to !600 °C or by mechanical crushing [100,101].On the Yilgarn Craton, extensional faulting extends 30-35 km into the lower crust [30,102].These faults may provide conduits for migration of gases deep in the crust.Magneto-telluric surveying of the Yilgarn Craton shows zones of high conductivity stretching from the Moho to the near surface, supporting the theory of fluid migration through faults [103,104].
Heat flow modelling predicts a geothermal gradient of ~18 ± 2 °C km À1 over the Yilgarn Craton [105], such that granitoids in the lower crust at 30 km depth exist at temperatures of ~565 ± 60 °C.At such temperatures, H 2 may exsolve from fluid inclusions in the lower crust, contributing a volumetrically significant amount of H 2 into the subsurface.
The juxtaposition of listric faults extending into the lower crust and the endogenic conditions necessary for the release of H 2 from fluid inclusions into these migration conduits provides the ideal geological conditions for the H 2 source at the FF4 site.
Granites and granitoids have the highest radiogenic heat flow of all rocks, due to their relative abundance of radioac-The Author(s): Science and Technology for Energy Transition 79, 48 (2024) tive isotopes ( 40 K, 232 Th, 235 U, and 238 U) [111].Global studies show that Australia has some of the most radiogenic granites in the world, averaging at 3.53 lW/m 3 , and that the Archean granites of Western Australia are anomalously high, producing up to 9.73 lW/m 3 [111].The basement rocks of the Wongan Hills area are well documented through mineral drilling and surface mapping, and comprise granites, granodiorites, tonalites and monzonites [30,32,86,112].
Water radiolysis also produces hydronium ions resulting in the formation of temporary acidic pH spikes in the irradiated groundwater [113].Flowing groundwaters may retain an acidic signature, consistent with the low pH of the groundwaters at the FF4 site.
The Archean basement lithologies combined with the acidity of the groundwater support radiolysis as a potential source for the natural H 2 observed at FF4.

Conclusions and implications
Natural H 2 has the potential to revolutionize the energy industry by enabling zero-carbon emissions from a natural fuel source.However, exploration of natural H 2 accumulations remains limited.Thus, we present the first long-term, multi-gas monitoring of a natural H 2 seep in Australia.
H 2 emissions occurred in significant concentrations, varied in amplitude across seasons and occurred as diurnal emissions with irregular high-concentration pulses.We interpret that H 2 micro-seepage was controlled by pressure pumping and soil moisture.The large pulses are interpreted to be the result of the H 2 gas pressure overcoming the interfacial tension of groundwater resulting in a sudden release.
H 2 emissions were inhibited by long-term high groundwater levels, which indicates that groundwater at ambient temperatures and pressures can form a barrier to the upward flow of H 2 gas through the soil profile and inhibit the soil gas H 2 readings at the surface.
This work has significant implications for natural H 2 exploration using soil gas monitoring techniques.False negatives are possible and likely in areas subject to inundation.When conducting soil gas sampling it is preferable to operate during the dry season and groundwater levels need to be considered, particularly in low-lying areas.Novel exploration methods will be required in wet environments.
Our confirmation of natural H 2 seepage on the Yilgarn Craton indicates active H 2 generation in the subsurface.The fundamental information on natural H 2 seepage provided by this study de-risks H 2 exploration in Western Australia, thus aiding in the implementation of a hydrogen economy.

Figure 1 .
Figure 1.Shows (a): Yilgarn Craton and FF4 Sample location (30°55 0 39 00 S, 116°35 0 35 00 E).(b) FF4 site interpreted bedrock geology (c) site morphology of ephemeral stream and salt-affected flood plains.Google Earth Image modified after Maxar Technologies, and (d) Results of a regional point soil gas sampling program that identified elevated H 2 concentrations at the FF4 site.

Figure 3 .
Figure 3. Sensor reaction to temperature (red) while measuring air, showing raw H 2 concentration measurement (grey) and temperature-corrected H 2 concentration measurement (blue).The dotted line shows the lower accuracy of the sensor of 10 ppm.

Figure 4 .
Figure 4. Gas concentrations of H 2 , hydrogen sulphide, and carbon dioxide plus atmospheric pressure and temperature measured at the sensor from the FF4 site (a) WHALI -20th of February 2022, installed immediately after drilling of the FF4 hole (b) WHALI -23rd of April to the 30th of April 2022 (c) WHALI -9th to the 16th of February 2023 and (d) EVA -9th to the 16th of February 2023.

4. 6
Groundwater volumesWe measured groundwater volumes extracted from the WHALI unit over the April 2022 and February 2023 autonomous monitoring periods.The total groundwater extracted during the seven-day monitoring periods in April 2022 and February 2023, was 20.3 L and 30.1 L, respectively (Table2).While a greater total volume of water was extracted in February 2023, the largest volume of water extracted during a single monitoring cycle occurred in April 2022.Thus, groundwater levels fluctuated during April 2022, whereas by February 2023 the groundwater saturation was consistently high.

Figure 6 .
Figure 6.All data from autonomous monitoring periods April 2022 and February 2023 (a) cH 2 vs temperature (b) cCO 2 vs temperature (c) cH 2 vs pressure and (d) cCO 2 vs pressure.

Figure 7 .
Figure 7. cH 2 for the WHALI and EVA autonomous logging units for the February 2023 monitoring period, with atmospheric pressure.Dotted line shows the lower accuracy limit of the H 2 sensors.

Figure 8 .
Figure 8. Temperature (upper squares) and Pressure (lower circles) measured at the site FF4 from the 24th to 29th April 2022.Both graphs are coloured by cH 2 (ppm) where blue is high and red is low.

Figure 9 .
Figure 9. (a) Rainfall preceding the monitoring periods at FF4, compared to the highest cH 2 from the WHALI autonomous monitoring unit during the monitoring periods Feb 2022, April 2022 and February 2023, and cH 2 measured via gas chromatography from gas samples taken at the FF4 site.(b) Rainfall preceding the monitoring periods at FF4, compared to the highest cH 2 from the WHALI autonomous monitoring unit and measured from gas samples.Pearsons coefficient is higher for the cH 2 vs previous 6 months rainfall compared to cH 2 vs previous 3 months rainfall.

Figure 10 .
Figure 10.A nearby ephemeral salt lake shows very little standing water was present during (a) January 2022 compared with (b) February 2023.

Table 1 .
Measurement ranges and precision ranges of gas detectors used in this study.

Table 2 .
Peak gas concentrations and atmospheric conditions measured during the three monitoring periods in the WHALI autonomous unit.