|
1 | 1 | --- |
2 | | -title: 'Radiolysis: Monte Carlo model for estimating H2 and He production through radiolysis' |
| 2 | +title: 'LithoGas: Monte Carlo model for estimating H2 and He production through radiolysis and serpentinization' |
3 | 3 | tags: |
4 | 4 | - R |
5 | 5 | - geology |
@@ -29,13 +29,20 @@ bibliography: paper.bib |
29 | 29 | --- |
30 | 30 |
|
31 | 31 | # Summary |
32 | | -Naturally occurring hydrogen (H2) in the Earth’s subsurface represents a novel source of H2 for use in electricity generation and hard-to-abate industrial processes (e.g., ammonia, fertilizer, steel) [@IEA:2024; @Ballentine:2025; @Johnson:2025; @SherwoodLollar:2025]. Helium (He), meanwhile, is a critical non-renewable resource utilized in medical, industrial, and research fields [@Warr:2019; @Danabalan:2022; @Cheng:2023]. With a growing list of H2 and He showings in different geological settings [@SherwoodLollar:2014; @Danabalan:2022; @Miaga:2023; @Truche:2024] along with focused exploration for these gases [@Jackson:2025; @Ballentine:2025; @Hu:2025], new tools are required to prospect for and model these emerging resources. Natural H2 is produced through two dominant geologic processes [@SherwoodLollar:2014; @Ballentine:2025], radiolysis [@Lin:2005; @Warr:2023; @Higgins:2025] and serpentinization [@Coveney:1987]. Radiolytic H2 is generated when ionizing α, β, and γ particles from U, Th, K-bearing minerals breaks water into H2 and O ions. It is important to note that released alpha particles are equivalent to helium-4 atoms and the breakdown of U and Th-bearing minerals is a dominant process for the generation of helium in the subsurface. H2 formation through serpentinization requires the release of H2 due to the alteration of ultramafic rocks to serpentine in the presence of water [@SherwoodLollar:2014; @Ballentine:2025]. Current modelling of these processes is not accessible or focused on economic assessment of potential resources. To partially fill this gap in resource estimation tools, `Radiolysis` provides fast and efficient Monte Carlo modelling of H2 and He production rates, following the radiolysis quantification methods of @Warr:2023. This implementation of the previously unpublished methods provides a rapid way to digest a large number of lithogeochemical samples to understand their H2 and/or He prospectivity over geological time scales. |
33 | 32 |
|
34 | | -# Statement of need |
35 | | -New resource modelling tools are urgently required to explore for He and for natural H2, given their emerging and critical role in low-carbon energy generation and in research, medical, industrial processes [@Danabalan:2022; @Jackson:2024; @IEA:2024; @SherwoodLollar:2025]. `Radiolysis` provides this by 1) calculation of H2 and He production rates via radiolysis from rock properties; 2) backward projecting these generation rates into deep time; 3) summarization and plotting of results, including novel plotting focused on resource-estimation metrics. These functions provide the foundation for basin modelling for radiolysis-dominated H2 systems similar to exploration for hydrocarbon systems (Figure 1). |
36 | | -The Monte Carlo approach follows the equations of @Warr:2023, incorporating truncated normal distributions (controlled by min, max, mean, and standard deviation inputs) for sample geochemistry (U, Th, and K concentrations), sample physical properties (rock density and porosity), and fluid properties (fluid density) (Table 1). The approach is programmed in R to rapidly handle multiple input samples. Rock properties (porosity, density) can be user defined, however, if rock properties are unknown, then summarized rock properties can be joined to samples from The Canadian Rock Physical Property Database [@Enkin:2018], based on the known lithology. Inputs into the models are always treated as a Monte Carlo distribution, however if a single deterministic model is desired, standard deviation of any model parameter can be set to zero. |
37 | | -To assess cumulative volumes of H2, and He produced through radiolysis over geologic time periods, `Radiolysis`can project these processes backwards through geologic time. The Monte Carlo models described above are projected, taking into account the combined radiogenic decay of U, Th, and K. These production rates can then be summed to determine the cumulative production over the specified time interval (Figure 2). |
38 | | -Overall, functions in `Radiolysis` provide a straight-forward modelling of multiple samples from a single structured dataframe to data summaries that inform exploration (Figure 1). Functions include: 1) preforming the Monte Carlo modelling as an R function, ingesting data as a structured dataframe (Table 1); 2) summarizing Monte Carlo results and plotting novel source-volume-scaling plots (Figure 2a); 3) backwards projecting Monte Carlo models into geologic time by calculating cumulative volumes (Figure 2b). Source-rock-volume-scaling plots are a simple yet important development, which allows for prospecting natural H2 and He from lithogeochemical samples, or modelling H2 systems similar to what gas traditionally been done for hydrocarbon systems. Example data include dataframes structured for Monte Carlo model input (with and without known rock properties) and the summarized data from The Canadian Rock Physical Property Database [@Enkin:2018] (Figure 1). `Radiolysis` will hopefully aid in the understanding of the subsurface hydrogen and helium systems and streamline its exploration. |
| 33 | +Naturally occurring hydrogen (H2) in the Earth's subsurface represents a novel source of H2 for use in electricity generation and hard-to-abate industrial processes (e.g., ammonia, fertilizer, steel) [@IEA:2024; @Ballentine:2025; @Johnson:2025; @SherwoodLollar:2025]. Helium (He), meanwhile, is a critical non-renewable resource utilized in medical, industrial, and research fields [@Warr:2019; @Danabalan:2022; @Cheng:2023]. With a growing list of H2 and He showings in different geological settings [@SherwoodLollar:2014; @Danabalan:2022; @Miaga:2023; @Truche:2024] along with focused exploration for these gases [@Jackson:2024; @Ballentine:2025; @Hu:2025], new tools are required to prospect for and model these emerging resources. Natural H2 is produced through two dominant geologic processes [@SherwoodLollar:2014; @Ballentine:2025]: radiolysis [@Lin:2005; @Warr:2023; @Higgins:2025] and serpentinization [@Coveney:1987]. Radiolytic H2 is generated when ionizing α, β, and γ particles from U, Th, and K-bearing minerals break water into H2 and O ions. It is important to note that released alpha particles are equivalent to helium-4 atoms, and the breakdown of U and Th-bearing minerals is a dominant process for the generation of helium in the subsurface. H2 formation through serpentinization requires the release of H2 due to the oxidation state change of Fe2+ to Fe3+ in different mineral reaction in the presence of water (e.g., olivine to serpentine) [@SherwoodLollar:2014; @Ballentine:2025]. Current modelling of these processes is not accessible or focused on economic assessment of potential resources. To partially fill this gap in resource estimation tools, LithoGas provides fast and efficient Monte Carlo modelling of H2 and He production rates, following the radiolysis quantification methods of @Warr:2023 and serpentinization methods of @Ardakani:2026. This implementation provides a rapid pathway to digest a large numbers of lithogeochemical samples to understand their H2 and/or He prospectivity over geological time scales, and to scale production rates to economically relevant source rock volumes. |
| 34 | + |
| 35 | +# Statement of Need |
| 36 | + |
| 37 | +New resource modelling tools are urgently required to explore for He and natural H2, given their emerging and critical role in low-carbon energy generation and in research, medical, and industrial processes [@Danabalan:2022; @Jackson:2024; @IEA:2024; @SherwoodLollar:2025]. LithoGas addresses this need by: 1) calculating H2 and He production rates via radiolysis and serpentinization from rock geochemical and physical properties; 2) back-projecting these generation rates into deep time to estimate cumulative production; 3) summarising and plotting results, including novel source-rock-volume-scaling plots focused on resource estimation metrics. These functions provide the foundation for basin-scale modelling of radiolysis- and serpentinization-dominated H2 systems, analogous to exploration workflows for conventional hydrocarbon systems. |
| 38 | +The Monte Carlo approach (monteProd()) follows the equations of [@Warr:2023] and [@Ardakani:2026], incorporating truncated normal distributions (controlled by min/max/mean/standard deviation parameters) for sample geochemistry (Fe, U, Th, and K concentrations), physical rock properties (rock density and porosity), and fluid properties (fluid density) (Table 1). Two pathways are available for assigning rock physical properties: user-defined sample-specific distributions, or automatic lookup from The Canadian Rock Physical Property Database [@Enkin:2018] based on the known lithology. If a deterministic rather than probabilistic model is desired, the standard deviation of any model parameter can be set to zero. Serpentinization H2 production can be modelled via two methods [@Ardakani:2026] depending on data availability: an iron speciation approach using measured Fe2O3 and FeO concentrations (monteSerpFeSpecies()), or a total iron approach using bulk Fe2O3T where speciation data are unavailable (monteSerpFeTotal()). Both serpentinization methods use the change in Fe³⁺/FeT ratio between initial and current states to estimate magnetite (Fe3O4) production and the associated stoichiometric H2 yield. |
| 39 | +Monte Carlo results from multiple samples are summarised using monteSum(), which collapses the full trial distribution to minimum, mean, and maximum production rates per sample group. Source rock volume scaling plots are generated by monteH2Plot() and monteHePlot(), which scale per m³ production rates across a range of source rock volumes (0.1 to 100 km³), producing log-log plots that allow direct comparison of H2 and He prospectivity across samples and lithologies. These source rock volume scaling plots are a simple yet important development, allowing prospecting for natural H2 and He from abundant lithogeochemical samples. A secondary axis on both plots converts molar production rates (mol/year for specified source rock volume) to mass rates (kg/year for specified source rock volume) for direct economic interpretation. |
| 40 | +Lastly, production rates produced by monteProd() can be projected into deep time wit the deepTimeProd() function. For radiolysis, U, Th, and K concentrations are back calculated using radioactive decay law. For serpentinization, and average rate of serpentinization is applied. These can the then be used by the user to look at cumulative volumes produced over time. |
| 41 | + |
| 42 | +# Conclusion |
| 43 | + |
| 44 | +Overall, LithoGas provides a straightforward workflow for modelling multiple samples from a single structured input dataframe through to publication-ready summaries and plots that inform exploration. Functions include: 1) performing the Monte Carlo modelling via monteProd(), ingesting data as a structured dataframe (Table 1); 2) summarising Monte Carlo results via monteSum() and plotting novel source-volume-scaling plots via monteH2Plot() and monteHePlot() (Figure 1). Example datasets include dataframes structured for Monte Carlo model input with and without known rock properties (structuredDF) and the summarised lithology distributions from The Canadian Rock Physical Property Database (CRPPData). |
| 45 | + |
39 | 46 |
|
40 | 47 | #### Table 1: Example layout of structured dataframe required for input into Radiolysis monteProd() function. The function data from the input dataframe by column name (e.g., $uMin, $rockDenMax), as such column names must match exactly. |
41 | 48 |
|
@@ -69,6 +76,6 @@ lithLith|String|Lithologiy/category of desired rock properties from Canadian Roc |
69 | 76 | |porSD|Numeric|Standard deviation of grain porosity distribution (decimal fraction)| |
70 | 77 |
|
71 | 78 | # Figures |
72 | | - |
| 79 | + |
73 | 80 |
|
74 | | - |
| 81 | + |
0 commit comments