Deep Coal Reservoirs’ Ultimate Methane Sorption Capacity: Constraints Imposed by Multiscale Pore Networks and Fractal Geometry Authors: Yaqiong Zhang, Zhenzhi Wang, Jinping Liao, Guangbiao Tao, Junjie Xiong, Qiulei Guo Published: 2026 ABSTRACT Investigating the controlling mechanism of heterogeneous pore structure at multiple scales in deep coal reservoirs on methane adsorption capacity is a key scientific issue for the accurate evaluation of the resource capacity of deep coal gas reservoirs. Herein, ten cores (burial depth: 1888.2–2276.9 m) retrieved from the eastern flank of the Ordos Basin were characterized via LP-CO2, LT-N2, MIP, and isothermal methane sorption tests. The primary objective was to elucidate the quantitative links between pore geometry, fractal scaling, and ultimate gas storage capacity. Findings demonstrate that as burial depth rises, total pore volume tends to grow; however, micro- and mesopore volume fractions shrink while macroporosity expands, suggesting that deeper coal formations possess markedly enlarged free-gas storage capacity. The micropore fractal dimension (DV) and mesopore spatial fractal dimension (DF2) show negative correlations with the Langmuir volume (VL), whereas the macropore fractal dimension (DS) and mesopore surface fractal dimension (DF1) exhibit positive correlations with VL. These observations imply that surface roughness and spatial connectivity across variable pore-size regimes differentially modulate the methane sorption potential of coal. Specifically, an excessively high micropore fractal dimension can cause pore-throat narrowing and an increase in closed micropores, thereby reducing effective adsorption space. Conversely, an elevated macropore fractal dimension can improve pore connectivity, protect the micropore adsorption system, and enhance adsorption capacity. This research offers a scientific foundation for the precise characterization and optimized exploitation of deep CBM reservoirs. Introduction Coal, as a naturally occurring porous medium, possesses a pore network that concurrently governs two critical aspects: first, it determines the methane storage potential of coal seams; second, it serves as the primary internal factor controlling the entire extraction chainnamely desorption, diffusion, and percolationduring CBM production. , Coal exhibits a broad pore-size spectrum, conventionally classified into micro- (<2 nm), meso- (2–50 nm), and macro-pores (>50 nm) according to the IUPAC convention. These pore-scale heterogeneities impose distinctly divergent controls on the retention and transport dynamics of gas molecules within the coal matrix. , Currently, the qualitative and quantitative characterization methods for the morphological development, spatial distribution, and structural parameters of pores in coal can be broadly categorized into three major technical systems: fluid intrusion testing, ray-based nondestructive detection, and microscopic image observation. Fluid intrusion techniques operate on the fundamental principle of pressurizing specific fluid media to permeate coal pore channels; they indirectly invert pore development parameters by leveraging the response laws of fluid permeation, adsorption, and intrusion. Characterized by straightforward operation and high detection efficiency, such testing methods constitute the mainstream approach for investigating coal reservoir pores. However, these techniques exhibit clear applicability boundaries. Constrained by fluid molecular dimensions and surface interaction characteristics, different fluids can only accurately match the quantitative detection of specific pore size intervals. Moreover, they exclusively identify interconnected open pores, failing to effectively characterize isolated closed pores. Among them, high-pressure mercury intrusion porosimetry (MIP), low-pressure CO2 adsorption (LP-CO2), and low-temperature N2 adsorption (LT-N2) represent the most widely applied core detection methods within this system. , Ray-based detection characterization technologies primarily encompass advanced testing means such as nuclear magnetic resonance (NMR), industrial computed tomography (CT) scanning, small-angle X-ray scattering (SAXS), and small-angle neutron scattering (SANS). Leveraging the interactions between high-energy rays and the internal pore-fracture structures of coal, these techniques enable refined analysis of pore spatial morphology, microstructural architecture, and internal distribution patterns. They offer the distinct advantages of high detection precision and nondestructiveness. Nevertheless, the equipment procurement and operational costs for these methods are prohibitively high. Furthermore, most of these techniques are limited to two-dimensional (2D) resolution of coal’s microstructures, presenting certain shortcomings in achieving comprehensive three-dimensional (3D) spatial characterization. Microscopic image analysis employs high-precision observation instrumentssuch as transmission electron microscopy (TEM), atomic force microscopy (AFM), and scanning electron microscopy (SEM)to directly capture the micromorphology of pores and fractures on coal surfaces and cross sections. Combined with digital image processing technology, this facilitates the extraction of pore parameters and feature analysis. However, this detection approach is highly susceptible to interference from sample preparation procedures; mechanical damage incurred during specimen fabrication can destroy the pristine pore structure. Concurrently, hardware conditions such as imaging resolution and clarity directly constrain pore identification accuracy and the reliability of analytical results. In summary, all categories of pore characterization techniques possess inherent applicable scopes and methodological limitations. A single detection method struggles to completely and accurately depict the multiscale, complex coupled pore-fracture structure of coal, nor can it achieve systematic quantitative analysis across the full aperture range. , Consequently, adopting a comprehensive characterization scheme that integrates multiple methods has become a consensus in the precise characterization of coal reservoir pore structures. By consolidating the strengths of different testing techniques, the technical shortcomings of any single method can be mitigated, thereby enabling comprehensive and accurate quantitative characterization of the coal’s full-scale pore architecture. , However, coal pores and fractures exhibit significant heterogeneity and anisotropy; , thus, the macroscopic parametric statistical features (specific surface area and pore volume) derived from these methods fail to reflect complex heterogeneous characteristics such as surface roughness and spatial connectivity. The introduction of fractal theory serves as an effective tool to address this issue, offering novel insights into the influencing mechanisms of gas content in coal reservoirs. To further investigate how pore structures influence CBM occurrence and production via fractal parameters, pore characterization methods and fractal theory have been combined in numerous studies. , The influence of fractal parameters of adsorption- and seepage-pores in low-rank coals from the Junggar Basin’s southern margin on methane sorption and permeability was investigated by Fu et al. through the integration of LT-N2 and MIP data with fractal geometry. Cheng et al. employed fractal theory to examine the fractal parameters of meso- and macro-pores in high-rank coals from the southeastern Qinshui Basin. They further explored how these fractal dimensions correlate with both sorption capacity and permeability. Li et al. examined coal pore fractal parameters across varying metamorphic grades using the FHH, PCAS, and Menger sponge models. Their work further elucidated how fractal dimensions of meso/macro-pores correlate with coalification, as well as their combined influence on methane sorption capacity. Guo et al. integrated fractal theory with mercury intrusion porosimetry to calculate the macropore fractal dimensions of coal samples with different particle sizes, thereby studying pore-fracture structural characteristics and proposing a classification scheme for pore-fracture structures based on inflection points in fractal curves. The application of fractal theory has significantly advanced our understanding of the complex pore-structure heterogeneity inherent to coal reservoirs. This provides a quantitative basis for both CBM exploration and subsequent exploitation. However, current research predominantly focuses on single pore scales. There is scant literature reporting on the influence of comprehensive fractal characteristicsparticularly those of deep coal reservoirson methane adsorption capacity across multiscale pores. In summary, to clarify how multiscale pore architecture distribution relates to sorbed gas content, this study integrates LT-N2, LP-CO2, MIP, and isothermal adsorption measurements for deep coal samples from the Ordos Basin. Concurrently, based on different fractal models, quantitative analyses were conducted on the surface topography and nonuniformity of multiscale pore-fracture networks within coal-bearing strata, further clarifying how fractal pore characteristics influence gas content in deep coal formations. Experimental Methods Sample Collection Ten fresh coal cores were selected from two deep CBM wells located along the eastern flank of the Ordos Basin, with six sourced from Well P20 and four from Well DJ54 (Figure ). The eastern margin of the Ordos Basin was selected as the study area because it represents one of China‘s most important demonstration bases for CBM industrial development and has recently emerged as a critical frontier for deep CBM exploration. The two sampled wells, P20 and DJ54, are located in adjacent blocks along the eastern flank of the basin and provide a natural depth gradient (approximately 300 m interval: DJ54 at 1,888.2–1,890.6 m vs P20 at 2,195.2–2,276.9 m) within the same tectonic unit. Upon retrieval, the fresh coal samples were wrapped in preservative film and sealed with tape before being transported to the laboratory for sample logging, followed by subsequent experimental testing. To mitigate the influence of coal heterogeneity, all subsamples used in the experiments were drilled from adjacent locations of the same intact coal block. Furthermore, the coal samples were measured for their maximum vitrinite reflectance values, maceral proportions, and proximate parameters, strictly complying with ISO 7404-5:1994­(E), ISO 7404-3:2009, and ISO 17246:2024. Sample sources: (a) Location of the Ordos Basin; (b) Basic tectonic units and sampling points of the Ordos Basin. Reproduced with permission from ref . Copyright 2025, Springer Nature. Gas Adsorption and MIP Gas sorption measurements utilized powder specimens sieved to 60–80 mesh. Before testing, the samples were degassed in a vacuum environment at 150 °C for 12 h to eliminate interference from moisture, volatile impurities, and residual gases adsorbed on the sample surface, thereby ensuring the accuracy of pore structure measurements. Following degassing pretreatment, LT-N2 measurements were conducted using high-purity N2 as the adsorbate at 77 K to obtain adsorption–desorption isotherms. In strict compliance with ISO 15901-2:2006, the BET and NLDFT formalisms were applied to derive pore structural attributes, namely mesopore-size spectra, specific surface area (SSA), and pore volume. LP-CO2 tests were carried out using high-purity carbon dioxide as the adsorbate at 273 K. Based on the NLDFT model, micropore size distribution, pore volume, and SSA were obtained, and the experiments were strictly conducted in accordance with the ISO 15901-3:2007 standard. For MIP, bulk samples with a smooth surface and a volume of 1 cm3 were utilized to eliminate errors induced by the “inter-particle effect” and the “skin effect”. Testing was performed after oven-drying and degassing. In accordance with ISO 15901-1:2005, pore structural propertiesincluding SSA, pore volume, and pore-size distributionswere derived via the Washburn equation. CH4 Isothermal Adsorption For isothermal methane sorption, we selected 120 g of coal powder (60–80 mesh) and performed gravimetric tests at 60 °C. A total of nine pressure points were set, steadily increasing from 0.3 to 25 MPa, with an adsorption equilibrium time of 12 h per pressure point. We calculated each pressure-point adsorption amount from the ideal-gas equation, with all tests conducted per GB/T 19560-2008. Fractal Dimension Calculation The Vadiliev–Singh (V–S) model, derived from Dubinin–Radushkevich theory, can accurately describe the micropore filling behavior observed in low-pressure CO2 adsorption. Its fractal dimension (D) can be expressed as 1 ln⁡V=3DVln⁡S+C where V referring to cumulative pore volume, cm3/g; S to the cumulative SSA, m2/g; DV to the micropore fractal dimension; and C to a constant. Low-temperature N2 adsorption exhibits distinctly different adsorption characteristics across various relative pressure ranges. In lower relative pressure intervals, van der Waals forces dominate the adsorption process, whereas capillary condensation prevails in higher relative pressure intervals. Based on the FHH model, the calculation formula for the mesopore fractal dimension can be expressed as 2 ln⁡VV0=(DF−3)ln(lnP0P)+C where P is the current adsorption pressure, MPa; P 0 is the saturation vapor pressure of the adsorbate, MPa; V is the volume of gas adsorbed at pressure P, cm3/g; V 0 is the monolayer adsorption capacity, which can be derived from the BET specific surface area, cm3/g; DF is the fractal dimension of the mesopores; and C stands for a constant term. The fractal dimension derived from high-pressure mercury intrusion porosimetry is typically estimated using the Menger sponge model. Its formula can be written as , 3 ln(dVdP)=(DS−4)ln⁡P+C where P denotes mercury intrusion pressure, MPa; V is the pore volume measured under pressure P, cm3/g; DS refers to fractal dimension; and C denotes a constant. It should be emphasized that the three types of fractal dimensions derived from different methods serve distinct yet complementary roles in characterizing pore-structure heterogeneity. The micropore fractal dimension (DV), obtained from LP-CO2 adsorption via the V–S model, quantifies the geometric irregularity and surface heterogeneity of micropores, where the primary gas storage mechanism is micropore filling. A higher DV indicates greater microporosity heterogeneity and tortuosity but may also imply a higher proportion of constricted or closed pores. The mesopore fractal dimensions (DF1 and DF2), derived from LT-N2 adsorption via the FHH model at low and high relative pressure ranges, respectively, reflect surface roughness (DF1) and spatial structural complexity (DF2) of the mesopore system. These two parameters allow the discrimination between favorable surface adsorption sites (DF1) versus unfavorable pore-throat tortuosity and connectivity issues (DF2). The macropore fractal dimension (DS), calculated from MIP data using the Menger sponge model, characterizes the structural heterogeneity and connectivity of the macropore-fracture network, which serves as the primary seepage pathway and, crucially, as a protective framework for preserving micropore adsorption space under compaction stress. The subsequent discussion in Section will demonstrate how these fractal parameters exert differentialand even opposingcontrols on methane adsorption capacity, thereby providing mechanistic insights that conventional pore-volume and SSA analyses alone cannot offer. Results Basic Sample Information The buried depths of the collected coal samples ranged from 1888.2 to 2276.9 m. R0,max values for the coal samples fell between 2.93% and 3.40%, averaging 3.225%. The maceral composition comprised 34.1%–64.3% vitrinite, 26.6%–61.3% inertinite, and 2.2%–9.5% mineral matter, while liptinite was absent. Proximate analyses showed moisture contents of 0.62%–1.64%, ash yields of 5.75%–17.03%, volatile matter of 6.13%–10.12%, and fixed carbon of 74.39%–86.71% (Table ). 1Summary of Basic Data for the Coal Specimens from the Study Area Maceral composition (vol %)Proximate analysis (%)Sample IDDepth(m) P (MPa) T (°C) R 0,max %VitriniteInertiniteExiniteMinerals M ad A ad V ad FC ad DJ54-3-11888.215.4862.213.4062.729.5/7.80.6910.566.1382.62DJ54-3-21888.815.4962.223.3061.331.3/7.40.638.816.3384.23DJ54-3-31889.415.4962.243.3750.541.9/7.60.6214.846.5877.96DJ54-3-41890.615.5062.273.3442.449.2/8.41.6417.036.9474.39P20-82195.218.0069.882.9362.727.8/9.51.0612.567.2479.14P20-102196.318.0169.913.1441.852.5/5.70.785.756.7686.71P20-112197.118.0269.933.3062.735.1/2.21.006.766.5485.70P20-152274.318.6571.863.1564.326.6/9.10.7612.6010.1276.52P20-172275.618.6671.893.1734.161.3/4.60.707.306.2885.72P20-192276.918.6771.923.1552.540.7/6.81.068.317.3083.33aNote: R 0,max represents the maximum vitrinite reflectance; P is the reservoir pressure, estimated using a pressure gradient of 0.82 MPa/hm; T is the reservoir temperature, estimated using a temperature gradient of 2.5 °C/hm; the moisture, ash, volatile matter, and fixed carbon contents on an air-dried basis are represented by M ad, A ad, V ad, and FC ad, respectively. Summary of Basic Data for the Coal Specimens from the Study Area Note: R 0,max represents the maximum vitrinite reflectance; P is the reservoir pressure, estimated using a pressure gradient of 0.82 MPa/hm; T is the reservoir temperature, estimated using a temperature gradient of 2.5 °C/hm; the moisture, ash, volatile matter, and fixed carbon contents on an air-dried basis are represented by M ad, A ad, V ad, and FC ad, respectively. Results of Pore Structure Characterization LP-CO2 Measurement Results The CO2 adsorption isotherms for the investigated region’s coal samples all showed a steady rise with increasing relative pressure. The overall trends of the curves were similar, without distinct plateau regions. Notably, the adsorption capacities of the deeper P20 well were greater than those of the shallower DJ54 well (Figure ). Specifically, the P20-10 (2196.3 m) coal sample showed the highest adsorption capacity at 29.85 cm3/g, while the DJ54-3-4 (1890.6 m) sample exhibited the lowest at 14.16 cm3/g. LP-CO2 isotherms: (a) Well P20; (b) Well DJ54. Based on the LP-CO2 adsorption test results, the pore structure parameters (Table ) and PSD curves (Figure ) of the micropores (0.3–1.5 nm) were obtained using the NLDFT model. The evolution of pore volume and SSA was consistent across all samples: as pore size grew from 0.2 to 1.6 nm, both parameters first increased sharply, peaked at 0.4–0.7 nm, and then dropped gradually. When the pore size exceeded 0.8 nm, both parameters remained at relatively low levels with minimal fluctuation; however, the pore volume displayed a stable increase, whereas the SSA showed a steady decrease. Among the samples, the peak values for the P20 well occurred in the 0.5–0.6 nm interval, while those for the DJ54 well appeared in the 0.4–0.5 nm interval. Pore size distribution characteristics of LP-CO2: (a–b) pore volume distribution; (c–d) SSA distribution. 2Experimental Test Results LP-CO2 LT-N2 MIPIsothermal adsorptionSample IDPore volume (cm3/g)SSA (m2/g)Pore volume (cm3/g)SSA (m2/g)Pore volume (cm3/g)SSA (m2/g)VL (cm3/g)PL (MPa)P20-80.062209.410.00180.640.0293.0424.463.04P20-100.092291.810.00120.480.0262.3927.313.05P20-110.078257.520.00230.880.0295.3228.142.98P20-150.075252.790.00220.690.0414.3631.473.21P20-170.079266.140.00140.450.0555.7132.833.46P20-190.079267.180.00150.640.0395.7229.393.22DJ54-3-10.055174.460.00171.080.04310.7226.692.43DJ54-3-20.068215.930.00170.910.03610.0324.561.51DJ54-3-30.055183.280.00201.070.04425.8626.601.98DJ54-3-40.036128.100.00130.660.03821.9720.792.75 Experimental Test Results LT-N2 Measurement Results All samples showed a gradual rise in sorption uptake with increasing (P/P0), followed by a sharp rise as P/P0 approached 1, exhibiting the characteristics of a typical Type IV isotherm (Figure ). Compared with Well P20, the samples from Well DJ54 generally displayed a more pronounced hysteresis loop. Sample adsorption capacities spanned 0.70–1.38 cm3/g. The P20-11 specimen exhibited the maximum, whereas P20-19 showed the minimum. Low-temperature N2 adsorption isotherms: (a) Well P20; (b) Well DJ54. From the LT-N2 adsorption data, we derived pore volume and SSA (Table ) as well as PSD curves (Figure ) using the LNDFT and BET formalisms. For the investigated region’s coal specimens, the LNDFT pore volume ranged from 0.0012 to 0.0023 cm3/g (average 0.00171 cm3/g), and the BET SSA ranged from 0.45 to 1.08 m2/g (average 0.75 m2/g). As the pore diameter increased, the pore volume showed an increasing trend, with an inflection point uniformly at 10 nm. In contrast, the specific surface area exhibited a decreasing trend with increasing pore size; the pore size inflection point was 10 nm for the P20 well and 5 nm for the DJ54 well. Pore size distribution characteristics of LT-N2: (a–b) pore volume distribution; (c–d) SSA distribution. MIP Test Results The MIP curves of Wells P20 and DJ54 exhibited distinct differences (Figure ). For all coal samples, the pore volume gradually increased with rising pressure. For Well P20, pore volume rose rapidly at pressures below 0.1 MPa; above that value, the growth gradually leveled off. In contrast, for Well DJ54, pore volume rose gradually at pressures below 0.1 MPa, whereas it surged at elevated pressures (especially >10 MPa). Coal samples from the investigated region had pore volumes spanning 0.026–0.055 cm3/g (mean: 0.038) and SSA varying between 2.39 and 25.86 m2/g (average: 141.999). Among them, the P20-17 coal sample had the largest pore volume, and the P20-10 sample had the smallest (Table ). For Well P20, macropores larger than 40,000 nm dominated the pore volume; for Well DJ54, pore sizes smaller than 100 nm were the main contributors, and their influence gradually diminished with increasing pore size. For both wells, the SSA was mainly contributed by pores smaller than 10 nm. As the pore size increased, the contribution ratio of the SSA decreased, and the SSA of pores larger than 10 nm was almost negligible (Figure ). Characteristics of MIP curves: (a) Well P20; (b) Well DJ54. Pore diameter distribution of MIP: (a–b) pore volume distribution; (c–d) SSA distribution. Methane Isothermal Adsorption Results All coal samples exhibited a progressive rise in CH4 sorption capacity as pressure increased, but the growth rate gradually decelerated. The adsorption process falls into two phases: a sharp-rise phase and a gradual-rise phase, with the 5 MPa pressure point serving as the divider (Figure ). Among them, the P20-17 coal sample had the highest methane adsorption capacity at 29.26 cm3/g, while the DJ54-3-4 sample had the lowest at 18.75 cm3/g. Using the Langmuir equation, we derived two key sorption parametersLangmuir volume (VL) and pressure (PL)from the methane adsorption data of coal specimens (Table ). For the 10 selected coal samples, ranged from 20.79 to 32.83 cm3/g, and PL ranged from 1.51 to 3.46 MPa. Specifically, the maximum values for both VL and PL were observed in the P20-17 coal sample, while the minimum VL was found in the DJ54-3-4 sample, and the minimum PL was recorded for the DJ54-3-2 sample. Methane isothermal adsorption curves: (a) Well P20; (b) Well DJ54. Discussion Influence of Multiscale Pore Structure Characteristics on Methane Adsorption Full-Scale Pore Structure Characteristics To comprehensively characterize the multiscale pore attributes of coal reservoirsgiven the inherent scale limitations of any single detection methodthis work integrated LP-CO2, LT-N2, and MIP, following the IUPAC size-classification scheme. This combined scheme enabled the separate characterization of micro- (<2 nm), meso- (2–50 nm), and macropores (>50 nm) in turn. Integration of multiscale pore data at the 2 and 50 nm thresholds generated complete PSD profiles covering macro-, meso-, and micropores (Figure ). The P20 well showed dual peaks in its pore-volume distribution (attributed to micro- and macropores), whereas DJ54 exhibited a single micropore-dominated peak. Furthermore, the macropore peak of the P20 well was concentrated in the large-scale pore range of 3–5 μm. This indicates that compared with the DJ54 well, the P20 well was subjected to tectonic movements, leading to tensile fracturing of the coal reservoir and the generation of abundant macropores and microfractures. For SSA distribution, both wells displayed a unimodal maximum within the micropore region, centered at ∼0.5 nm; DJ54’s peak value was higher, whereas P20 exhibited greater SSA on both flanks of this peak. Full-scale pore size distribution characteristics: (a) pore volume distribution; (b) SSA distribution. Overall, the P20 well’s total pore volume ranged between 0.0872 and 0.1260 cm3/g (mean: 0.1079), with micropore, mesopore, and macropore volumes accounting for 72.08%, 1.01%, and 26.91%, respectively. DJ54’s total pore volume spanned 0.0395–0.0725 cm3/g (mean: 0.0582), with volume fractions of 91.21% (micropores), 2.40% (mesopores), and 6.39% (macropores) (Table ). For deep reservoirs, as the buried depth increased, the total pore volume increased by 185.39%, accompanied by a 19.13% decrease in the proportion of micropores, a 1.39% decrease in mesopores, and a 20.52% increase in macropores. This suggests that with increasing buried depth, the free gas storage space expands, leading to a higher proportion of free gas content. 3Multiscale Pore Parameters of the Samples Pore volume (cm3/g)SSA (m2/g)Sample IDMicMesMacTpvMicMesMacTpvP20-80.06240.00100.02380.0872209.410.60900.0227210.0417P20-100.09200.00080.02210.1149291.810.46580.0162292.2920P20-110.07750.00140.02040.0993257.520.85570.0390258.4147P20-150.07490.00120.03390.1100252.790.67140.0686253.5300P20-170.07910.00100.04590.1260266.140.44110.0567266.6378P20-190.07920.00100.02990.1101267.180.53140.0513267.7627DJ54-3-10.05460.00150.00510.0612174.461.08050.0540175.5945DJ54-3-20.06780.00140.00330.0725215.930.90650.0239216.8604DJ54-3-30.05460.00160.00350.0597183.281.07240.0289184.3813DJ54-3-40.03580.00100.00270.0395128.100.66300.0170128.7800aNote: Mic = micropores; Mes = mesopores; Mac = macropores; Tpv = total pore volume. Multiscale Pore Parameters of the Samples Note: Mic = micropores; Mes = mesopores; Mac = macropores; Tpv = total pore volume. In terms of pore volume composition, the average pore volume of the P20 well was 1.85 times that of the DJ54 well, with a higher proportion of macropores. In contrast, DJ54 had higher proportions of micropores and mesopores. Regarding total specific surface area, the P20 well ranged from 209.41 to 267.18 m2/g, averaging 258.11 m2/g, while the DJ54 well ranged from 128.10 to 215.93 m2/g, averaging 176.40 m2/g; the P20 well’s value was approximately 1.46 times that of the DJ54 well. For both wells, micropores accounted for over 99% of the SSA, leaving the contributions of meso- and macropores virtually insignificant. Based on the distinct pore-volume partitioning observed between the two wells, we propose a classification scheme for deep coal reservoir pore structures. The DJ54 well, characterized by a micropore volume fraction exceeding 90% and a macropore fraction of less than 7%, represents the “adsorption-dominated type,” wherein the pore system is overwhelmingly controlled by microporous adsorption space. In contrast, the P20 well, with a micropore fraction of approximately 70% and a substantially elevated macropore fraction of ∼27%, falls into the “adsorption–seepage composite type.” This classification reflects the progressive increase in free-gas storage capacity (macropores) relative to adsorbed-gas storage capacity (micropores) with increasing burial depth. Influence of Pores at Different Scales on Methane Adsorption For deep coal samples, the VL increased with increasing volumes of micropores (R2 = 0.689), macropores (R2 = 0.664), and total pores (R2 = 0.449), showing a positive correlation, while no significant correlation was observed with mesopore volume (Figure ). Total pore volume and VL showed a trend similar to that of micropores. This suggests that micropores exert a primary control on methane sorption capacity in deep coal formations. Relationships between pore volumes at different scales and adsorption capacity: (a) micropores; (b) mesopores; (c) macropores; (d) total pores. On the one hand, in deep coal reservoirs, micropores dominate both total pore volume (>70%) and total SSA (>99%). SSA is the key factor governing coal’s methane retention; higher SSA offers more active sites for the coal matrix, thus improving its gas adsorption capacity. , On the other hand, micropore filling dominates methane sorption in micropores. Because of their confined pore widths, the overlapping of van der Waals potentials from adjacent walls generates a strong adsorption field. − Driven by this superimposed adsorption potential, gas molecules completely occupy the micropore space via volumetric filling, making all microporous spaces effective sites for methane adsorption and storage. Therefore, the micropore volume dictates the effective site count on coal matrix surfaces; the greater the micropore volume, the larger the accessible void space and the corresponding SSA. The lack of a significant correlation between mesopore volume and VL differs from previous studies. This discrepancy is attributed to the extremely low proportion (∼2%) and limited variation of mesopore volume in the studied coal samples. Macropores themselves have a virtually negligible specific surface area, contributing very little to methane adsorption sites and thus making no direct contribution to methane adsorption. The positive correlation between macropore volume and VL is essentially controlled by the synergistic development mechanism of the pore structure and the pore-fracture connectivity protection effect. An increase in macropore volume usually indicates a more well-developed overall pore structure and stronger connectivity between pores and fractures in the coal body. This opens up or connects previously isolated and closed micropores, effectively mitigating the compressive destruction of micropores caused by compaction and ensuring the integrity of the micropore adsorption system. At the same time, macropores, mesopores, and micropores exhibit a characteristic of homologous and symbiotic development: the more developed the macropores in a coal sample, the greater its micropore volume and SSA tend to be, which in turn leads to a synchronous increase in methane adsorption capacity. Methane Adsorption in Relation to Pore Fractal Features Multiscale Pore Fractal Characteristics Using LP-CO2 data and eq , we generated a double-logarithmic plot of cumulative pore volume against SSA to derive the micropore fractal dimension (DV) (Figure ). The obtained DV values for the coal samples spanned 2.457–2.641 (mean: 2.545). The correlation coefficients varied from 0.997 to 0.999, indicating a very high fitting degree (Table ). Overall, the DV of the P20 well (mean = 2.519) was slightly lower than that of the DJ54 well (mean = 2.584). Moreover, the fluctuation amplitude of DV in the P20 well was greater than that in the DJ54 well, being 0.177 and 0.103, respectively. Fractal characteristics of micropores. 4Fractal Dimension Calculation Results LP-CO2 LT-N2 MIPSample ID D V R2 D F1 R2 D F2 R2 D S R2 P20-82.6340.9982.3130.9402.6010.9412.7920.864P20-102.5710.9972.5180.8402.5740.9392.6760.851P20-112.4750.9982.4910.9092.6430.9082.9120.822P20-152.4770.9972.5050.9492.6340.9392.8220.879P20-172.5020.9972.7060.8842.5910.9712.8610.822P20-192.4570.9972.5380.9712.6310.9502.6350.952DJ54-3-12.5770.9992.3840.9672.7310.9332.6660.990DJ54-3-22.6410.9982.3250.9652.6600.9422.4690.966DJ54-3-32.5380.9992.3080.9352.7190.9242.5120.971DJ54-3-42.5820.9992.4120.9812.7010.9622.3790.987aNote: D V = fractal dimension of micropores; D F1 = surface fractal dimension of mesopores; D F2 = spatial fractal dimension of mesopores; D S = fractal dimension of macropores. Fractal Dimension Calculation Results Note: D V = fractal dimension of micropores; D F1 = surface fractal dimension of mesopores; D F2 = spatial fractal dimension of mesopores; D S = fractal dimension of macropores. Given that van der Waals forces dominate N2 adsorption at low P/P0 and capillary condensation at high P/P0, the data were split at the 0.5 threshold. This allowed the surface fractal dimension (DF1) and its spatial counterpart (DF2) for mesopores to be derived separately (Figure ). , DF1 values spanned 2.308–2.706 (mean: 2.450) with R2 of 0.884–0.981; DF2 values spanned 2.574–2.731 (mean: 2.649) with R2 of 0.908–0.971 (Table ). Specifically, the DF1 of the P20 well was greater than that of the DJ54 well, and the fluctuation amplitude of DF1 in the P20 well was larger. In contrast, the DF2 of the P20 well was smaller than that of the DJ54 well, and the fluctuation amplitudes of DF2 for both wells were comparable. Fractal characteristics of mesopores. Considering that high-pressure mercury intrusion can cause matrix compression at high pressures, leading to measurement errors in micropore and mesopore volumes, this study exclusively utilized high-pressure mercury intrusion to characterize macropores (>50 nm) in order to ensure data accuracy and scientific rigor. Therefore, for calculating the macropore fractal dimension (DS), only the pressure and pore volume corresponding to pore sizes larger than 50 nm were selected for plotting and analysis (Figure ). DS values for the coal samples: 2.379–2.912 (mean 2.672); correlation coefficients: 0.822–0.987 (Table ). Specifically, the DS values of the P20 well were greater than those of the DJ54 well; the P20 well samples were almost entirely above the average, while all DJ54 well samples were below the average. Fractal characteristics of macropores. Relationship between Fractal Dimensions at Different Pore Scales and VL The DV ranged from 2.457 to 2.641, falling within a high-value interval and indicating an overall elevated fractal level. It exhibited a negative correlation with the VL (R2 = 0.650, Figure a). Theoretically, a higher DV value indicates stronger heterogeneity in the micropore spatial structure, more complex pore geometries, and higher tortuosity. Consequently, a larger specific surface area is provided per unit pore volume, leading to an increase in adsorption sites and, accordingly, a higher maximum methane adsorption capacity However, a negative correlation was observed between the DV and VL in this study. This can be attributed to three factors: (1) Studies indicate that the correlation of fractal dimension with adsorption capacity is not simply linear; instead, the ultimate methane adsorption capacity rises then falls with increasing fractal dimension, suggesting an excessive fractal dimension imposes a critical threshold. , (2) The primary adsorption mechanism in micropores is micropore filling. When the micropore fractal dimension is too high, the excessively rough pore morphology causes the local diameter of some micropores to decrease, approaching the kinetic diameter of methane molecules (∼0.38 nm). Such overly narrow channels hinder methane molecules from penetrating deep into the pores, resulting in these micropore spaces being ineffectively utilized and reducing the effective adsorption volume. , (3) A higher micropore fractal dimension implies the presence of more closed or semiclosed micropores in the samples. Although this portion of pore volume is included in the total pore volume, it contributes nothing to adsorbed methane. Correlation of multiscale fractal parameters with CH4 adsorption potential: (a) micropore fractal dimension; (b) mesopore surface fractal dimension; (c) mesopore spatial fractal dimension; (d) macropore fractal dimension. DF1 exhibited a positive correlation with VL (R2 = 0.790, Figure b). The underlying mechanisms can be summarized as two aspects: (1) In mesopores, methane sorption proceeds via monolayer and multilayer coverage at lower relative pressures, while capillary condensation dominates at higher relative pressures. A higher pore-surface fractal dimension (DF1) indicates rougher walls and stronger surface heterogeneity, which effectively increases the SSA and provides more active centers for initial monolayer/multilayer adsorption. (2) Crucially, such enhanced surface roughness also strengthens the van der Waals attraction for methane molecules and lowers the energy barrier for the formation of precursor capillary condensate films. This promotes the subsequent capillary condensation process, thereby significantly boosting the overall adsorption capacity within mesopores. Conversely, DF2 correlated negatively with VL (R2 = 0.195, Figure c). This adverse effect is primarily attributed to the unfavorable influence of excessive spatial complexity on pore connectivity. Mesopores serve as the essential pathways for free gas to migrate from macropores to the micropore adsorption sites. When the spatial fractal dimension (DF2) is excessively high, the pore-throat channels tend to become tortuous and constricted. Such geometric constraints not only hinder gas diffusion and disconnect the continuous migration network, , but also induce a significant “ink-bottle” effect during capillary condensation. This effect traps condensate in isolated pore bodies and blocks narrow pore throats, severely limiting the accessibility of micropores and reducing the effective mesopore filling volume. Consequently, the ultimate adsorption capacity (VL) decreases. VL rose with DS, showing a positive correlation (R2 = 0.512, Figure d). This can be attributed to two reasons: (1) A higher DS indicates better preservation of the original macropore structure, with a more well-developed and better-connected macroscopic fracture network. Coal samples with well-developed macropores have a more uniform internal stress distribution, which promotes the development of mesopores and micropores and yields larger effective micropore volumes and specific surface areas. Consequently, this manifests as a positive correlation with the Langmuir volume. (2) As macroscopic seepage channels, macropores can effectively release tectonic stress and formation pressure in deep reservoirs, thereby reducing the compressive effect of effective stress on micropores and mesopores. When the macropore fractal dimension is relatively high, the macropore-fracture network is well developed, resulting in a significant stress relaxation effect. This prevents micropores from being compacted and closed, preserving the effective adsorption space and ultimately enhancing the Langmuir volume. , Influence of Maceral Components on Methane Adsorption VL rose with vitrinite content but dropped with inertinite content (Figure ). This can primarily be attributed to three aspects: (1) Micropores in vitrinite mainly develop during coalification, whereas pores in inertinite predominantly originate from residual mesopores and macropores of the original plant cell structures. (2) For high-rank coals, the aromatic layers in vitrinite become more ordered, which gradually reduces pore size and promotes extensive micropore development. These micropores provide abundant adsorption sites. In contrast, inertinite inherently possesses extremely high thermal stability and structural stability. (3) Under the same coal rank conditions, vitrinite contains richer oxygen-containing functional groups and aliphatic hydrocarbons, and its heat of adsorption is higher than that of inertinite. Relationship between maceral components and adsorption capacity: (a) vitrinite content; (b) inertinite content. Relationship between Coal Proximate Analysis Components and Methane Adsorption Increases in moisture, ash, and volatile matter content in coal lead to a decrease in methane adsorption capacity, whereas an increase in fixed carbon content helps enhance methane adsorption (Figure ). This is because: (1) The presence of moisture results in competitive adsorption with methane for adsorption sites; simultaneously, moisture forms water films on micropore surfaces through capillary condensation, hindering methane from migrating from macropores into micropores. (2) Ash, the inorganic residue from coal combustion, has nearly negligible methane sorption capacity. Higher ash content lowers overall sorption; moreover, it densifies the coal matrix through cementation, blocking pores and fractures, thus decreasing effective porosity and SSA. (3) Volatile matter primarily comprises residual aliphatic side chains, cycloalkanes, and functional groups within the coal molecular structure. These flexible nonaromatic structures occupy interlamellar spaces that could otherwise serve as adsorption sites. Additionally, the abundant oxygenated polar groups in volatile matter increase coal hydrophilicity, promote water vapor uptake, and indirectly strengthen the competitive adsorption of water against methane. (4) Fixed carbon is the primary component of condensed aromatic rings, and a higher fixed carbon content indicates a greater degree of aromatization and polycondensation of the organic fraction in coal, resulting in enhanced micropore development. Furthermore, the surface of fixed carbon contains very few polar functional groups and exhibits strong hydrophobicity, which can reduce moisture’s competitive adsorption and boost the utilization of methane adsorption sites. , Relationships between proximate analysis results and adsorption capacity: (a) M ad; (b) A ad; (c) V ad; (d) FC ad. Measured Gas Content It is generally accepted that isothermal adsorption curves are solely employed to predict adsorbed gas content at varying pressures; however, whether such predicted values accurately reflect the actual in situ gas content requires calibration against field-measured gas content data. Given the absence of field-measured gas content data for the target coal seam in our work, we adopted the systematic well-testing results on CBM gas content reported by Wang et al. (2026) in the same region to compensate for the lack of site-specific measurements. Table details the field gas content test results under corresponding depths and coal structure conditions as documented in that reference. Using the correspondence between the tabulated data and the isothermal adsorption curves in Section , we further examined how free gas content relates to the pore architecture of coal seams in this region. 5Field-Measured Gas Content Results Sample IDDepth (m) V S (cm3/g) V a (cm3/g) V g (cm3/g) F a (%) F g (%)DJ57-8-21884.2829.2624.664.684.2815.72DJ57-8-41885.426.3221.754.5782.6417.36DJ57-8-61886.2129.9423.26.7477.4922.51DJ57-8-81887.0321.5218.323.285.1314.87P20-12192.1630.725.635.0783.4916.51P20-22192.4133.5327.286.2581.3618.64P20-32192.9732.7624.87.9675.7024.30P20-42193.4234.428.176.2381.8918.11P20-52193.9835.4428.357.0979.9920.01P20-62194.4127.3323.234.185.0015.00aWell DJ57 is adjacent to Well DJ54 investigated in this study, and its gas-bearing characteristics may provide useful reference. V S, measured gas content; V a, adsorbed gas content; V g, free gas content; F a, adsorbed gas fraction; F g, free gas fraction. Reproduced with permission from ref . Copyright 2026, Elsevier. Field-Measured Gas Content Results Well DJ57 is adjacent to Well DJ54 investigated in this study, and its gas-bearing characteristics may provide useful reference. V S, measured gas content; V a, adsorbed gas content; V g, free gas content; F a, adsorbed gas fraction; F g, free gas fraction. Reproduced with permission from ref . Copyright 2026, Elsevier. Wells DJ57 and P20 exhibit essentially comparable free-gas proportions, with DJ57 at 17.62% and P20 at 18.76%. However, the deeper Well P20 exhibits a higher total gas content; consequently, its free gas content also surpasses that of Well DJ57, being 1.92 times greater. Relationship between Isothermal Adsorption Results and Measured Gas Content A larger Langmuir volume corresponds to a greater measured gas content, exhibiting a strong positive linear correlation (R2 = 0.655); thus, the Langmuir volume can serve as a key parameter for predicting gas content (Figure b). However, in terms of specific values, except for sample P20-6, the Langmuir volumes of all coal samples are smaller than their measured gas contents, indicating that predicting the gas content of deep reservoirs based on isothermal adsorption curves under experimental conditions leads to an underestimation of gas content (Figure a). This is attributed to the following reasons: (1) Sample pretreatment for isothermal adsorption experiments causes certain damage to the adsorption potential of the coal samples. The crushing process during sample preparation destroys the primary pore network, maceral structure, and pore-fracture connectivity of the coal samples; high-temperature drying and vacuum degassing cause irreversible damage to the surface functional groups of the coal matrix, destroying some effective methane adsorption sites. (2) Adsorption isotherm tests are performed under unconfined conditions and thus cannot replicate the actual gas sorption and storage behavior of coal seams under in situ overburden stress. (3) Deep coal seams contain a high proportion of free gas; however, the Langmuir sorption model cannot account for this free-gas component, thus greatly underestimating its contribution to the total gas-in-place. Comparison of isothermal adsorption results with measured gas content: (a) difference between VL and VS; (b) correlation between VL and VS. Correlation of Pore Structure with Free Gas The controlling effects of pores at different scales on free gas content exhibit significant differences. Free gas content relates positively to micropore and macropore volumes, but shows no clear relationship with mesopore volume (Figure a). Micropores mainly accommodate adsorbed gas; as micropore volume increases, the adsorbed gas fraction rises, and the dominant control shifts from pressure to temperature with depth, facilitating the conversion of adsorbed to free gas in deep formations. Macropores, by contrast, host free gas, and their volume increase directly raises free-gas content. Relationship between pore structure and free gas: (a) pore volumes at different scales; (b) pore fractal dimensions. Pore fractal characteristics exhibit distinct correlations with free gas content (Figure b). Specifically, free-gas volume is negatively correlated with the micropore fractal parameter DV. Since micropores primarily affect free gas indirectly by influencing adsorbed gas content, this implies that an excessively high micropore fractal dimension is unfavorable for methane sorption, thus lowering the total gas content. For mesopores, free gas content is positively correlated with DF1 but negatively correlated with DF2. This contradictory relationship may explain why mesopore volume shows no significant correlation with free gas content. For macropores, free gas content is positively correlated with DS. A higher macropore fractal dimension indicates that the primary pore structure is relatively well preserved, so the free gas storage space is less disturbed by external factors, thereby facilitating more favorable gas storage. Conclusions In this study, LT-N2, LP-CO2, MIP, and isothermal adsorption tests were integrated with field-measured gas content data to investigate the influence of multiscale pore structures and fractal dimensions on CBM content in deep coal samples from the Ordos Basin. The principal findings are summarized below:1The methane adsorption capacity of deep coal reservoirs is dominated by micropores and synergistically controlled by multiscale fractal characteristics. An excessively high micropore fractal dimension induces local constriction of pore throats and increases the proportion of closed/semiclosed micropores, thereby reducing the effective adsorption space and suppressing the ultimate adsorption capacity. Conversely, a higher macropore fractal dimension improves pore connectivity, protects the micropore adsorption system, and promotes adsorption.2As burial depth increases, the pore structure of deep coal reservoirs systematically evolves from an “adsorption-dominated type” to an “adsorption–seepage composite type”. Quantitatively, the total pore volume increases with depth, accompanied by a decrease in micropore volume fraction from 91.21% to 70.08% and a marked increase in macropore volume fraction from 6.39% to 26.91%. This significant expansion of free-gas storage space drives a progressive shift in the adsorbed-to-free gas ratio, indicating that deeper coal seams possess considerably enhanced free-gas storage capacity.3Adsorption experiments under conventional laboratory conditions cannot represent the actual gas occurrence characteristics of deep reservoirs and should not be directly applied to gas content prediction, especially for deep CBM systems with a high proportion of free gas.4Free gas in deep reservoirs is primarily controlled by gas source supply from micropores and the availability of storage space in macropores, whereas mesopores exert no discernible influence on free gas content. However, an elevated micropore fractal dimension is detrimental to free gas enrichment, while a higher macropore fractal dimension facilitates free gas occurrence. 1The methane adsorption capacity of deep coal reservoirs is dominated by micropores and synergistically controlled by multiscale fractal characteristics. An excessively high micropore fractal dimension induces local constriction of pore throats and increases the proportion of closed/semiclosed micropores, thereby reducing the effective adsorption space and suppressing the ultimate adsorption capacity. Conversely, a higher macropore fractal dimension improves pore connectivity, protects the micropore adsorption system, and promotes adsorption. The methane adsorption capacity of deep coal reservoirs is dominated by micropores and synergistically controlled by multiscale fractal characteristics. An excessively high micropore fractal dimension induces local constriction of pore throats and increases the proportion of closed/semiclosed micropores, thereby reducing the effective adsorption space and suppressing the ultimate adsorption capacity. Conversely, a higher macropore fractal dimension improves pore connectivity, protects the micropore adsorption system, and promotes adsorption. 2As burial depth increases, the pore structure of deep coal reservoirs systematically evolves from an “adsorption-dominated type” to an “adsorption–seepage composite type”. Quantitatively, the total pore volume increases with depth, accompanied by a decrease in micropore volume fraction from 91.21% to 70.08% and a marked increase in macropore volume fraction from 6.39% to 26.91%. This significant expansion of free-gas storage space drives a progressive shift in the adsorbed-to-free gas ratio, indicating that deeper coal seams possess considerably enhanced free-gas storage capacity. As burial depth increases, the pore structure of deep coal reservoirs systematically evolves from an “adsorption-dominated type” to an “adsorption–seepage composite type”. Quantitatively, the total pore volume increases with depth, accompanied by a decrease in micropore volume fraction from 91.21% to 70.08% and a marked increase in macropore volume fraction from 6.39% to 26.91%. This significant expansion of free-gas storage space drives a progressive shift in the adsorbed-to-free gas ratio, indicating that deeper coal seams possess considerably enhanced free-gas storage capacity. 3Adsorption experiments under conventional laboratory conditions cannot represent the actual gas occurrence characteristics of deep reservoirs and should not be directly applied to gas content prediction, especially for deep CBM systems with a high proportion of free gas. Adsorption experiments under conventional laboratory conditions cannot represent the actual gas occurrence characteristics of deep reservoirs and should not be directly applied to gas content prediction, especially for deep CBM systems with a high proportion of free gas. 4Free gas in deep reservoirs is primarily controlled by gas source supply from micropores and the availability of storage space in macropores, whereas mesopores exert no discernible influence on free gas content. However, an elevated micropore fractal dimension is detrimental to free gas enrichment, while a higher macropore fractal dimension facilitates free gas occurrence. Free gas in deep reservoirs is primarily controlled by gas source supply from micropores and the availability of storage space in macropores, whereas mesopores exert no discernible influence on free gas content. However, an elevated micropore fractal dimension is detrimental to free gas enrichment, while a higher macropore fractal dimension facilitates free gas occurrence.