Hydrogen Flooding of a Coal Core: Effect on Coal Swelling

Hydrogen is a clean fuel which has the potential to drastically decarbonize the energy supply chain. However, hydrogen storage is currently a key challenge; one solution to this problem is hydrogen geo‐storage, with which very large quantities of H2 can be stored economically. Possible target formations are deep coal seams, and coal permeability is a key parameter which determines how fast H2 can be injected and withdrawn again. However, it is well known that gas injection into coal can lead to coal swelling, which drastically reduces permeability. We thus injected H2 gas into a coal core and measured dynamic permeability, while imaging the core via x‐ray micro‐tomography at reservoir conditions. Importantly, no changes in coal cleat morphology or permeability were observed. We conclude that H2 geo‐storage in deep coal seams is feasible from a fundamental petro‐physical perspective; this work thus aids in the large‐scale implementation of a hydrogen economy.

• Hydrogen geo-storage in deep coal seams is feasible from a flooding and adsorption perspective • Hydrogen flooding does not cause any measurable coal swelling under deep coal seam conditions • Coal permeability is not affected by hydrogen flooding under deep coal seam conditions

Supporting Information:
Supporting Information may be found in the online version of this article.

Experimental Procedure
A deep coal seam at a depth of approximately 250 m was simulated in the laboratory. Bituminous coal (from Morgantown, West Virginia, USA; supplied by Wards Scientific US; vitrinite reflectance = 0.86) was selected and small core plugs (5 mm diameter and 10 mm length) were drilled. The coal had a porosity of 3.7% (measured via Helium porosimetry using a CoreLab UltraPoroPerm-910 instrument), and the coal was also thoroughly analyzed via XRD (performed with a RAYONS X-Rays instrument equipped with a cobalt K α radiation source at 40 kV and 40 mA), TGA (using a PerkinElmer-Thermogravimetric Analyzer-TGA 4000), ATR-FTIR (with a PerkinElmer-Spectrometer 100-FT-IR instrument), BET (performed at 77 K; to measure specific surface area, pore volume and average pore size of the coals using a Tristar II 3020 instrument) and ultimate and proximate analysis. Results are shown in Table 1 and in the Supporting Information.
The specific surface area of the coal was 0.34 m 2 /g, average pore size was 16.38 nm, pore volume was 0.0014 cm 3 /g and microporous content was 0. The coal plug was then placed into an x-ray transparent high pressure μCT cell (Iglauer & Lebedev, 2018) and vacuumed for 24 hr to remove all air from the system. The temperature was kept constant at 296 K (i.e., isothermal conditions), a pore pressure of 2.758 MPa was applied and overburden stress was raised to 6.205 MPa (i.e., 3.447 MPa effective stress was applied during the whole experiment), and the coal plug was μCT imaged at two high resolutions (1.50 and 4.00 μm, using the 3D x-ray microscope VersaXRM500 Xradia-Zeiss). Subsequently, 18,000 pore volumes H 2 gas (from BOC, HPG, 99.99 mol% purity) were injected into the plug applying a pressure drop of 0.22 MPa (2.978 MPa inlet pressure and 2.758 MPa outlet pressure). Three high precision syringe pumps (ISCO Teledyne 500D, accuracy 0.1%) were used to apply injection pressure, backpressure and overburden pressure with high accuracy. Dynamic coal permeability was measured during H 2 -flooding by measuring the H 2 flow rate through the plug at this constant pressure drop and applying Darcy's law. H 2 injection was then stopped, and the system was kept at these conditions for 24 hr. The coal core was then μCT imaged again at the same high resolutions in-situ. Note that this μCT resolution is insufficient to resolve the complete pore space in the coal matrix, which can be even of atomic volume, but cleat (=fracture) networks can be reliably imaged (e.g., compare Ramandi et al., 2016;, Y. Zhang, Xu, et al., 2016. The images were cropped, their image contrast adjusted, and outliers were removed via the application of a median filter, and the subsequent removal of outliers. The removal of outliers replaced a pixel by the median value of neighbors (radius of 2 pixels here), if the pixels was 50% brighter or darker than the neighborhood (Schneider et al., 2012). This process was performed for both bright and dark outliers three times, after which the percentage of removed outliers was below 5%. This conservative filtering was performed using Fiji ImageJ (Schindelin et al., 2012). The first and last 100 images were removed from the image stack to avoid the strongest ring artifacts. Darkest pixels were assumed to be void space as H 2 gas has a low x-ray attenuation. 3D void space (cleats) was segmented and counted in 3D using Fiji as well, taking into account integrated density, mean and standard deviation of gray values, minimum, maximum and average gray values, as well as computed centroids and centers of mass within the bounding box (Bolte & Cordelières, 2006).

Coal Cleat Network Morphology Evolution
For H 2 geo-storage project assessment it is vital to know how much H 2 can be stored in a reasonable timeframe, and how fast H 2 can be withdrawn again (Pan et al., 2021). Previously, it has been shown that substantial amounts of H 2 can be adsorbed on the coal and storage capacity is thus in principle large . However, it is currently unknown how coal swelling and the associated coal permeability is affected by H 2 -exposure; indeed, only very few data are available to assess UHS as a true economic technical option (Pan et al., 2021;F. Zhang, Zhao, et al., 2016). Potential changes in the coal cleat network characteristics (before and after H 2 flooding) were thus quantified and compared.
Importantly, H 2 exposure led to no measurable change in the coal cleat porosity, cleat network morphology or cleat size distribution (compare Figures 1  and 2 and Table 2 where the pore space morphology and the cleat network statistics before and after H 2 flooding are summarized). All phase fractions (mineral phase, maceral phase and the cleat system [void space]) remained constant during and after H 2 -flooding, and fracture nucleation or propagation  was not observed. Note that the small changes in Figures 1 and 2 and Table 2 were caused by the natural variation in coal (as the imaged volumes did not exactly overlap). We conclude that H 2 gas does not lead to swelling of the maceral phase; in contrast, CO 2 injection clearly leads to dramatic swelling of the maceral phase, even to the extent that the swelling stress is so high that it can fracture the mineral phase inside the coal (Y. Y. Zhang, Xu, et al., 2016). Compare these effects also with CH 4 and N 2 gas exposure; CH 4 exposure leads to significant maceral swelling, while N 2 exposure leads to minor maceral swelling (Zhou et al., 2013). Indeed, coal swelling follows the order CO 2 > CH 4 > N 2 > H 2 . This is related to the polarizability of these molecules (and thus the Van der Waals interaction forces between the gas molecules and the maceral phase; CO 2 polarizability is 29.1 × 10 −25 cm 3 , that of CH 4 is 25.9 × 10 −25 cm 3 , that of N 2 is 17.4 × 10 −25 cm 3 and that of H 2 is 8 × 10 −25 cm 3 ; Ahmed & Rothenberger, 2015;Rallapalli et al., 2011), which directly determines maceral-gas interaction affinity, this is also for instance expressed in a much higher adsorption capacity of coal for CO 2 when compared to H 2 Keshavarz et al., 2021). Note that CO 2 also forms hydrogen bonds with carbonyl and alcohol groups present in the coal, further increasing CO 2 -coal affinity (Fujii et al., 2002).

Dynamic Coal Permeability
Above observations are consistent with the measured dynamic coal permeability, which remained constant throughout H 2 -flooding, Figure 3. The fluctuations were within the error margin of the experiment, and as can be seen permeability did not change even after flooding the core with 18,000 pore volumes of H 2 . We conclude that injection of H 2 and subsequent withdrawal of H 2 from the storage reservoir is feasible from a fundamental fluid dynamical perspective. As high amounts of H 2 can be adsorbed, H 2 geo-storage in coal seams is a promising novel technique to store very large amounts of H 2 in a cost-effective manner.

Conclusions
H 2 geo-storage provides an alternative option for H 2 storage, at a giant scale (e.g., Tarkowski, 2019;F. Zhang, Zhao, et al., 2016). One key target formation are deep coal seams, which can adsorb large amounts of H 2 Keshavarz et al., 2021;Pan et al., 2021). However, a sufficiently large coal permeability is required for efficient H 2 injection (for storage) and H 2 withdrawal (when the fuel is needed). In this context it is well known that CO 2 injection leads to dramatic swelling of the maceral (organic) phase, for example, Zhou et al. (2013); Y. , which drastically reduces coal permeability. It is therefore of vital importance to assess how H 2 -exposure influences the cleat network in the coal and the associated coal seam permeability. As this is unknown for H 2 , we performed coal coreflooding tests and injected H 2 into a bituminous coal sample at UHS conditions (296 K, 2.758 MPa pore pressure, 3.447 MPa effective stress-which approximates the geothermal conditions prevailing at a depth of 250 m). The coal plug was imaged at high resolution with an x-ray tomograph, and coal permeability was measured in parallel. Importantly, no change in cleat morphology or permeability was observed.
We conclude that H 2 -injection into deep coal seams does not induce coal swelling or permeability reduction-UHS in coal seams is thus feasible from a fundamental physico-chemical (high H 2 adsorption) and petro-physical (high injection and withdrawal rates are possible) perspective. This work therefore provides essential insights into coal permeability behavior when exposed to H 2 gas; and thus aids in the implementation of a large-scale hydrogen economy.