Surface Properties and Pore Structure of Anthracite, Bituminous Coal and Lignite

: Properties of coal surface and pore structure are important aspects to be investigated in coal preparation and utilization. In order to investigate the limits of different probe methods, a comprehensive approach was comparatively used to probe surface properties and pore structure of anthracite, bituminous coal and lignite. Surface morphology of the three coal samples was analyzed by scanning electron microscopy (SEM). Combining mercury intrusion porosimetry (MIP), physisorption method with carbon dioxide (CO 2 ) at 273 K and nitrogen (N 2 ) at 77 K was used to quantify a broad pore size distribution of coals, while FT-IR and water vapor sorption methods were used to study the coal surface properties. The results show that wedge-shaped pores develop with the increase of coal rank due to compression effect. The determined speciﬁc surface area (SSA) and pore volume of N 2 decrease with the increase of coal rank, while CO 2 SSA and pore volume are of a kind of U-shaped function of coal rank. MIP results indicate that that the pore size of 10–100 nm accounted for 70.7–97.5% of the total volume in the macropore range. Comparison of different methods indicates that micropores cannot be fully covered by the standard probes. CO 2 adsorption technique can only probe micropores in the range of 0.5 nm to 0.9 nm. Water vapor is not an effective probe to detect the micropores in coals, due to that the water clusters is mainly ﬁlled in mesopores and macropores. The results also show that both water vapor adsorption and FT-IR analysis can provide qualitative information of coal surface, rather than qualiﬁcation of functional groups. the pore structure and of and lignite. electron microscopy performed for pore morphology characterization. Combining MIP, physisorption method 273 and at used to quantify a broad pore size distribution of Fourier Transform infrared spectroscopy (FT-IR) and water vapor adsorption methods were used to study the coal surface properties. Finally, the measurable ranges and the limits of each method


Introduction
Coal is an extremely heterogeneous porous material containing both organic macerals and inorganic mineral matter [1]. The macromolecule structure of coal macerals offers plenty of adsorption sites for gaseous and liquid adsorbates [2][3][4]. In addition, the amorphous surface of coal enriches functional groups, possessing strong adsorption capacity for polar molecule adsorbates [5][6][7], such as water vapor. In coal preparation and utilization processes, including coalbed methane (CBM) recovery and CO 2 geosequestration, both surface properties and pore structure are important aspects to be investigated [8,9].
The pores in coal and other porous materials are commonly divided into three categories based on the pore diameter, following the International Union of Pure and Applied Chemistry (IUPAC) classification, as micropores (≤2 nm in diameter), mesopores (2-50 nm idiameter) and macropores (≥50 nm in diameter) [10]. For coal, the macromolecular structure of organic macerals forms an interconnected pore network, resulting in a broad distribution of pore size [11,12]. Coal structure varies with the degree of maturity of coal as measured by reflectance or volatile matter. Accompanying

Sample Preparation
Coal blocks were collected from three mineable coalbeds in China. Since coal rank is a key factor determining the pore structure and surface properties, different coal rank collected from three deposits were used in this study. Anthracite from No. 6 coal seam in Datong coal mine, Sichuan coalfield, medium-volatile bituminous coal from No. 2 coal seam in Zhaolou coal mine, Heze coalfield and lignite from No. 1-2 coal seam in Daliuta coal mine, Shendong coal field. The locations and stratigraphic columns of the three coal mines are shown in Figure 1. Coal blocks were firstly obtained from freshly-exposed mining faces, then sealed immediately and transported to the laboratory. In laboratory, coal lumps were ground and sieved to particles of 18-20 mesh (0.83-0.88 mm) in size for further analysis and measurement.

Sample Preparation
Coal blocks were collected from three mineable coalbeds in China. Since coal rank is a key factor determining the pore structure and surface properties, different coal rank collected from three deposits were used in this study. Anthracite from No. 6 coal seam in Datong coal mine, Sichuan coalfield, medium-volatile bituminous coal from No. 2 coal seam in Zhaolou coal mine, Heze coalfield and lignite from No. 1-2 coal seam in Daliuta coal mine, Shendong coal field. The locations and stratigraphic columns of the three coal mines are shown in Figure 1. Coal blocks were firstly obtained from freshly-exposed mining faces, then sealed immediately and transported to the laboratory. In laboratory, coal lumps were ground and sieved to particles of 18-20 mesh (0.83-0.88 mm) in size for further analysis and measurement.

Ultimate and Proximate Analyzes
Prior to surface properties and pore structure investigation, ultimate and proximate analyzes were carried out to characterize the coal samples. Flash EA2000 elemental analyzer (Thermo Fisher Scientific, Waltham, MA, USA) was used for ultimate analysis. The proximate analysis on coal samples includes fixed carbon (Cfix), volatile matters (Vdaf), ash (Aad) and moisture (M). The ultimate analysis aims for the element analysis of carbon (C), hydrogen (H), nitrogen (N) and oxygen (O).

Scanning Electron Microscopy
SEM JSM-7800F, JEOL USA, Inc. (Peabody, MA, USA) was used to visually investigate the pore morphology of the three coal samples. The coal samples prepared for SEM analysis were 1 × 1 × 1 cm in size and the magnification time was set as 10,000. After surface cleaning, the coal samples were vacuum dried at 373 K for 12 h. Before SEM scanning, the coal samples were coated with a layer of gold.

Ultimate and Proximate Analyzes
Prior to surface properties and pore structure investigation, ultimate and proximate analyzes were carried out to characterize the coal samples. Flash EA2000 elemental analyzer (Thermo Fisher Scientific, Waltham, MA, USA) was used for ultimate analysis. The proximate analysis on coal samples includes fixed carbon (C fix ), volatile matters (V daf ), ash (A ad ) and moisture (M). The ultimate analysis aims for the element analysis of carbon (C), hydrogen (H), nitrogen (N) and oxygen (O).

Scanning Electron Microscopy
SEM JSM-7800F, JEOL USA, Inc. (Peabody, MA, USA) was used to visually investigate the pore morphology of the three coal samples. The coal samples prepared for SEM analysis were 1 × 1 × 1 cm in size and the magnification time was set as 10,000. After surface cleaning, the coal samples were vacuum dried at 373 K for 12 h. Before SEM scanning, the coal samples were coated with a layer of gold.

FT-IR Analysis
FT-IR is one of the most useful methods used in instrumental analysis for research on coals and carbon materials. It allows non-destructive testing of the chemical structure of the objects being characterized [41]. In this study, surface functional groups of anthracite, bituminous coal and lignite were determined using a FT-IR Nicelet 6700 (Thermo Fisher Scientific, Waltham, MA, USA). Using He/Ne laser source, it can provide a spectral scan from 400 to 4000 cm −1 . The coal sample for FT-IR analysis (0.5-1 mg in weight) was prepared with the potassium bromide (KBr) pellet method. The coal powders and the dried KBr were ground at a mass ratio of 1:160. The obtained infrared interference spectral was transformed through Fourier calculation to identify hydroxyl, aliphatic hydrogen and carbonyl functional groups.

Low-Pressure N 2 and CO 2 Isotherm Measurements
The low-pressure isotherms of N 2 and CO 2 were measured using Accelerated Surface Area and Porosimetry System ASAP 2020M, Micrometrics Instruments Corporation USA, Inc (Norcross, GA, USA). Before each test, the coal sample (1-2 g in weight) was degassed for 12 h at 383 K to remove air, water and other volatile matters. At each pressure set-point, the sorption equilibrium was established automatically when the pressure stabilized for 30 s. The absolute pressure tolerance was set as 5 mmHg (6.66 mbar). The warm free space and cold free space were determined by helium expansion after isotherm measurement. The SSA, pore volume and pore size distribution (PSD) of the coal samples were calculated based on multiple theories [42].
The CO 2 adsorption isotherms were measured in an ice bath (273 K) under the relative pressure ranging from 0.005 to 0.032. Lower limit of the relative pressure is used for the evaluation of CO 2 adsorption as micropores are filled at lower relative pressure than mesopores [13]. The SSA and micropore volume determined by CO 2 adsorption were evaluated by Dubinin-Radushkevich (D-R) and Dubinin-Astakhov (D-A) methods. The PSD were calculated by Density Functional Theory (DFT). The N 2 adsorption-desorption isotherms were measured at 77 K under the relative pressure from 0.01 to 0.99. The SSA of the coal samples with N 2 adsorbate were evaluated by Langmuir, Barrett-Joyner-Halenda (BET) and Barrett-Joyner-Halenda (BJH) methods for comparison purpose. The pore volume and average pore diameter from N 2 isotherm were determined using BET and BJH models. The mesopore size distribution of the samples determined from N 2 adsorption were based on BJH method. In comparison, the calculation models for different adsorbates are summarized in Table 1.

Mercury Porosimetry Measurement
A Mercury Porosimetry Analyzer from Quantachrome Instruments was used to comparatively show its probe difference to different coal rank, using anthracite, bituminous coal and lignite. At the start of the measurement, the coal sample was firstly immersed in mercury container and then degassed to achieve the vacuum state. Subsequently, the pressure was gradually applied to drive the mercury into the pores. The applied mercury pressure and the intruded volume were recorded to derive the pore structure of coal.
The applied mercury intrusion pressure was started from 21 psi (0.15 MPa) to a maximum pressure of about 15,000 psi (100 MPa). The contact angle and surface tension for injection were assumed to be 130 • and 0.485 N/m to determine pore size distribution [13,18].

Water Vapor Isotherm Measurements
The gravimetric method was used to measure water vapor adsorption-desorption isotherms using the IGA gravimetric analyzer, Hiden Isochema Limited UK, Inc. (Warrington, UK), as schematically shown in Figure 2. The IGA system is an intelligent gravimetric analyzer by integrating computer-controlled precise measurement units for weight change, pressure and temperature. The pressure supply system can regulate the gas pressure from 0 to 2.0 MPa, and the weighing resolution of the microbalance is 0.2 µg ± 0.1 µg. In the isotherm measurement process, the microbalance and the pressure control units are under the thermostat control with an accuracy of ±0.1 K. degassed to achieve the vacuum state. Subsequently, the pressure was gradually applied to drive the mercury into the pores. The applied mercury pressure and the intruded volume were recorded to derive the pore structure of coal. The applied mercury intrusion pressure was started from 21 psi (0.15 MPa) to a maximum pressure of about 15000 psi (100 MPa). The contact angle and surface tension for injection were assumed to be 130° and 0.485 N/m to determine pore size distribution [13,18].

Water Vapor Isotherm Measurements
The gravimetric method was used to measure water vapor adsorption-desorption isotherms using the IGA gravimetric analyzer, Hiden Isochema Limited UK, Inc. (Warrington, UK), as schematically shown in Figure 2. The IGA system is an intelligent gravimetric analyzer by integrating computer-controlled precise measurement units for weight change, pressure and temperature. The pressure supply system can regulate the gas pressure from 0 to 2.0 MPa, and the weighing resolution of the microbalance is 0.2 μg ± 0.1 μg. In the isotherm measurement process, the microbalance and the pressure control units are under the thermostat control with an accuracy of ± 0.1 K. Water vapor isotherms on the coal samples were measured at 298 K. Prior to isotherm measurement, the coal sample was degassed at 383 K for 12 h under vacuum (10 −6 Pa) in order to remove the water content. In isotherm measurement, the adsorption/desorption equilibrium was established at a pressure point once the change in sample weight was less than 1% of equilibrium or a time-limit was reached. Then, the water vapor pressure was regulated to the next test. In preliminary tests, it was found that the weight of the coal sample was stable after 120 min adsorption and did not change further even for experiments lasting up to 20 h. Thus, the sorption equilibration time is set as 120 min, which is sufficient for gas sorption equilibrium at each pressure point. Buoyancy effect was corrected automatically in the measurement.

Ultimate and Proximate Results
The results of ultimate and proximate analyzes are presented in Table 2. In comparison, anthracite contains the highest carbon content, but the lowest volatile matters and ash content, corresponding to 73.85%, 12.81% and 13.34%, respectively, while lignite contains the lowest carbon content, but the highest volatile matters and ash content. However, the moisture content does not show such a monotonic trend. In the ultimate analysis, carbon element shows the same trend as that of carbon content in the proximate analysis. Regarding the hydrogen, nitrogen and oxygen elements, lignite is the highest, followed by bituminous coal and anthracite in sequence. Therefore, the carbon content increased with coal rank, while the volatile, ash and moisture decreased with coal rank. This Water vapor isotherms on the coal samples were measured at 298 K. Prior to isotherm measurement, the coal sample was degassed at 383 K for 12 h under vacuum (10 −6 Pa) in order to remove the water content. In isotherm measurement, the adsorption/desorption equilibrium was established at a pressure point once the change in sample weight was less than 1% of equilibrium or a time-limit was reached. Then, the water vapor pressure was regulated to the next test. In preliminary tests, it was found that the weight of the coal sample was stable after 120 min adsorption and did not change further even for experiments lasting up to 20 h. Thus, the sorption equilibration time is set as 120 min, which is sufficient for gas sorption equilibrium at each pressure point. Buoyancy effect was corrected automatically in the measurement.

Ultimate and Proximate Results
The results of ultimate and proximate analyzes are presented in Table 2. In comparison, anthracite contains the highest carbon content, but the lowest volatile matters and ash content, corresponding to 73.85%, 12.81% and 13.34%, respectively, while lignite contains the lowest carbon content, but the highest volatile matters and ash content. However, the moisture content does not show such a monotonic trend. In the ultimate analysis, carbon element shows the same trend as that of carbon content in the proximate analysis. Regarding the hydrogen, nitrogen and oxygen elements, lignite is the highest, followed by bituminous coal and anthracite in sequence. Therefore, the carbon content increased with coal rank, while the volatile, ash and moisture decreased with coal rank. This is consistent with the previous studies [16,43]. The content of hydrogen, nitrogen and oxygen elements monotonically decreased with the coal rank. This is due to the polycondensation effect and the elimination of -OH bounds during coalification process.  Figure 3 shows SEM images of the coal samples at the magnification time of 10,000. It can be seen that the pore distribution of the coal samples is inhomogeneous. Furthermore, either pore size or pore shape is different among the three coal samples. The anthracite has a major distribution of wedge-shaped pores, which is probably due to the compaction effect during coalification process. The bituminous coal is highly porous with cylindrical pores, wedge-shaped pores and bottle-shaped pores. Pores in lignite can hardly be observed at the magnification time of 10,000, which is possibly due to the less maturity of the lignite compared to anthracite and bituminous coal. is consistent with the previous studies [16,43]. The content of hydrogen, nitrogen and oxygen elements monotonically decreased with the coal rank. This is due to the polycondensation effect and the elimination of -OH bounds during coalification process.  Figure 3 shows SEM images of the coal samples at the magnification time of 10,000. It can be seen that the pore distribution of the coal samples is inhomogeneous. Furthermore, either pore size or pore shape is different among the three coal samples. The anthracite has a major distribution of wedge-shaped pores, which is probably due to the compaction effect during coalification process. The bituminous coal is highly porous with cylindrical pores, wedge-shaped pores and bottle-shaped pores. Pores in lignite can hardly be observed at the magnification time of 10,000, which is possibly due to the less maturity of the lignite compared to anthracite and bituminous coal.

FT-IR Results
The FT-IR spectra of the three coal samples are shown in Figure 4, where the difference in FT-IR spectra represents different functional group characteristics on coal surfaces. The first band at a wave number of around 3500 cm −1 is attributed to -OH groups caused by surface hydroxyl of water molecules on coal surface. The second band at the wave number of around 1622 cm −1 indicate amide carbonyl (-C=O) of the coal surface. The spectral bands from 900-700 cm −1 are attributed to the aromatic structure. The sharp absorption peak of bituminous coal at around 1434 cm −1 is attributed to C-H bond, while the peaks of anthracite and lignite at 1032 cm −1 are caused by the C-O bond of the carboxylic acid, phenol and alcohols structure. Figure 4 shows, compared to the bituminous coal and the lignite, the anthracite exhibited relatively weak FT-IR absorption peaks. This indicates that anthracite has a low concentration of functional groups on its surface.

FT-IR Results
The FT-IR spectra of the three coal samples are shown in Figure 4, where the difference in FT-IR spectra represents different functional group characteristics on coal surfaces. The first band at a wave number of around 3500 cm −1 is attributed to -OH groups caused by surface hydroxyl of water molecules on coal surface. The second band at the wave number of around 1622 cm −1 indicate amide carbonyl (-C=O) of the coal surface. The spectral bands from 900-700 cm −1 are attributed to the aromatic structure. The sharp absorption peak of bituminous coal at around 1434 cm −1 is attributed to C-H bond, while the peaks of anthracite and lignite at 1032 cm −1 are caused by the C-O bond of the carboxylic acid, phenol and alcohols structure. Figure 4 shows, compared to the bituminous coal and the lignite, the anthracite exhibited relatively weak FT-IR absorption peaks. This indicates that anthracite has a low concentration of functional groups on its surface.

Pore Structure for the Coal Samples
The N2 adsorption-desorption isotherms of the coal samples are shown in Figure 5a. The isotherms of the three coal samples uniformly show Type Ⅳ, indicating a wide distribution of pore size [10]. In the isotherms, adsorption-desorption hysteresis appeared at the high relative pressure, suggesting capillary condensation occurred within the mesopores [44]. The shape of the hysteresis loop of lignite corresponds to type C, while bituminous coal and anthracite show Type B according to de Boer classification [45]. Type C hysteresis loop illustrates slit-shaped pores in lignite, and Type B hysteresis loop reflects that a significant amount of wedge-shaped pores exist in anthracite and bituminous coal. This characteristic can be visually confirmed by SEM images as shown in Figure 3. It is noted that the hysteresis loop of lignite show a lack of closure at the low relative pressure, which is attributed to the swelling effect of coal matrix in adsorption process [46].  Figure 5b shows CO2 adsorption isotherms of the coal samples at 273 K under a low relative pressure range (≤0.032). The adsorption capacity of CO2 on the coal samples are larger than that of N2 at a similar relative pressure range. As mentioned previously, this is because CO2 molecules can enter narrow micropores when compared to N2 adsorbate [47]. From Figure 5, it can be observed that anthracite has the lowest N2 adsorption capacity, but the highest CO2 adsorption capacity,

Pore Structure for the Coal Samples
The N 2 adsorption-desorption isotherms of the coal samples are shown in Figure 5a. The isotherms of the three coal samples uniformly show Type IV, indicating a wide distribution of pore size [10]. In the isotherms, adsorption-desorption hysteresis appeared at the high relative pressure, suggesting capillary condensation occurred within the mesopores [44]. The shape of the hysteresis loop of lignite corresponds to type C, while bituminous coal and anthracite show Type B according to de Boer classification [45]. Type C hysteresis loop illustrates slit-shaped pores in lignite, and Type B hysteresis loop reflects that a significant amount of wedge-shaped pores exist in anthracite and bituminous coal. This characteristic can be visually confirmed by SEM images as shown in Figure 3. It is noted that the hysteresis loop of lignite show a lack of closure at the low relative pressure, which is attributed to the swelling effect of coal matrix in adsorption process [46].

Pore Structure for the Coal Samples
The N2 adsorption-desorption isotherms of the coal samples are shown in Figure 5a. The isotherms of the three coal samples uniformly show Type Ⅳ, indicating a wide distribution of pore size [10]. In the isotherms, adsorption-desorption hysteresis appeared at the high relative pressure, suggesting capillary condensation occurred within the mesopores [44]. The shape of the hysteresis loop of lignite corresponds to type C, while bituminous coal and anthracite show Type B according to de Boer classification [45]. Type C hysteresis loop illustrates slit-shaped pores in lignite, and Type B hysteresis loop reflects that a significant amount of wedge-shaped pores exist in anthracite and bituminous coal. This characteristic can be visually confirmed by SEM images as shown in Figure 3. It is noted that the hysteresis loop of lignite show a lack of closure at the low relative pressure, which is attributed to the swelling effect of coal matrix in adsorption process [46].  Figure 5b shows CO2 adsorption isotherms of the coal samples at 273 K under a low relative pressure range (≤0.032). The adsorption capacity of CO2 on the coal samples are larger than that of N2 at a similar relative pressure range. As mentioned previously, this is because CO2 molecules can enter narrow micropores when compared to N2 adsorbate [47]. From Figure 5, it can be observed that anthracite has the lowest N2 adsorption capacity, but the highest CO2 adsorption capacity, indicating a major distribution of micropore. The N2 adsorption capacity decreases with the increase of coal rank, while CO2 adsorption capacity shows an U-shaped function in terms of coal rank.
The estimated pore properties from N2 and CO2 isotherms are summarized in Table 1. The Langmuir SSA, BET SSA and BJH SSA determined by N2 are around as low as 1 m 2 /g. Regarding the coal rank, the SSA and pore volumes determined by N2 adsorption show a tendency to decrease with  Figure 5b shows CO 2 adsorption isotherms of the coal samples at 273 K under a low relative pressure range (≤0.032). The adsorption capacity of CO 2 on the coal samples are larger than that of N 2 at a similar relative pressure range. As mentioned previously, this is because CO 2 molecules can enter narrow micropores when compared to N 2 adsorbate [47]. From Figure 5, it can be observed that anthracite has the lowest N 2 adsorption capacity, but the highest CO 2 adsorption capacity, indicating a major distribution of micropore. The N 2 adsorption capacity decreases with the increase of coal rank, while CO 2 adsorption capacity shows an U-shaped function in terms of coal rank.
The estimated pore properties from N 2 and CO 2 isotherms are summarized in Table 1. The Langmuir SSA, BET SSA and BJH SSA determined by N 2 are around as low as 1 m 2 /g. Regarding the coal rank, the SSA and pore volumes determined by N 2 adsorption show a tendency to decrease with the increase of coal rank. The determined average pore diameter with BET and BJH methods shows that the pores probed by N 2 adsorbate fell in the mesopore range. Table 1 also shows that the CO 2 SSA, based on either D-R or D-A models, of the coal samples are larger than N 2 SSA. With CO 2 adsorbate, the determined SSA for anthracite is the largest, followed by lignite and bituminous coal. However, with N 2 adsorbate, the determined SSA decreases with the increase of coal rank. Mahajan et al. [8] argued that CO 2 is capable of measuring the total surface area of coal. Bituminous coal is found to exhibit the lowest SSA and pore volume of micropores among the three samples. This is consistent with the published results [13].
The PSD for the coal samples determined by N 2 and CO 2 adsorption are shown in Figure 6. The results indicate that the coals contain a broad PSD in the micro-and mesopore range. The calculated pore volume distribution of lignite is higher than anthracite and bituminous coal in mesopore range, with major peaks at around 9 nm and 25 nm. The bituminous coal has a major peak at around 45 nm, while anthracite has minor volume distribution in mesopore range (Figure 6a). In micropore range, anthracite has major peaks at around 0.56 nm, 0.77 nm and 0.9 nm, while bituminous coal and lignite have major peak at around 0.84 nm (Figure 6b). that the pores probed by N2 adsorbate fell in the mesopore range. Table 1 also shows that the CO2 SSA, based on either D-R or D-A models, of the coal samples are larger than N2 SSA. With CO2 adsorbate, the determined SSA for anthracite is the largest, followed by lignite and bituminous coal. However, with N2 adsorbate, the determined SSA decreases with the increase of coal rank. Mahajan et al. [8] argued that CO2 is capable of measuring the total surface area of coal. Bituminous coal is found to exhibit the lowest SSA and pore volume of micropores among the three samples. This is consistent with the published results [13].
The PSD for the coal samples determined by N2 and CO2 adsorption are shown in Figure 6. The results indicate that the coals contain a broad PSD in the micro-and mesopore range. The calculated pore volume distribution of lignite is higher than anthracite and bituminous coal in mesopore range, with major peaks at around 9 nm and 25 nm. The bituminous coal has a major peak at around 45 nm, while anthracite has minor volume distribution in mesopore range (Figure 6a). In micropore range, anthracite has major peaks at around 0.56 nm, 0.77 nm and 0.9 nm, while bituminous coal and lignite have major peak at around 0.84 nm (Figure 6b). The PSD and SSA of the coals determined by MIP method are given in Table 1 and Figure 7. It can be observed that pores in the range from 10-100 nm accounted for 70.7-97.5% of the total volume for all the coal samples. The volume of large macropore (100-1000 nm) decreases with the increase of coal rank. The SSA and total pore volume of bituminous coal are the largest, followed by lignite and anthracite in sequence. This order is different from the results of N2 and CO2, demonstrating a major distribution of macropores in bituminous coal.  The PSD and SSA of the coals determined by MIP method are given in Table 1 and Figure 7. It can be observed that pores in the range from 10-100 nm accounted for 70.7-97.5% of the total volume for all the coal samples. The volume of large macropore (100-1000 nm) decreases with the increase of coal rank. The SSA and total pore volume of bituminous coal are the largest, followed by lignite and anthracite in sequence. This order is different from the results of N 2 and CO 2 , demonstrating a major distribution of macropores in bituminous coal. The PSD and SSA of the coals determined by MIP method are given in Table 1 and Figure 7. It can be observed that pores in the range from 10-100 nm accounted for 70.7-97.5% of the total volume for all the coal samples. The volume of large macropore (100-1000 nm) decreases with the increase of coal rank. The SSA and total pore volume of bituminous coal are the largest, followed by lignite and anthracite in sequence. This order is different from the results of N2 and CO2, demonstrating a major distribution of macropores in bituminous coal.

Water Vapor Adsorption-Desorption Isotherms
Water vapor adsorption-desorption isotherms for the coal samples were measured at the relative pressure up to 0.9 as shown in Figure 8. The isotherms are type III corresponding to IUPAC classification [10]. This implies that the coal samples are of weak adsorption affinity with water. The isotherms can be divided into several sections. At low relative pressures, the amount of adsorbed water tends to be a linear increase trend with pressure, which is due to the interaction between water molecules and functional groups on coal surface. With the increase of relative pressure, the water molecules adsorb on top of the pre-adsorbed water molecules and clusters are formed. As the relative pressure increase further, the clusters merge to fill the pore space in the coals. This is shown as an rapid increase in water adsorption.
Regarding the coal rank, lignite has the largest water adsorption capacity at the whole pressure range, followed by bituminous coal, while anthracite has the smallest adsorption capacity. Combining this result with the FT-IR spectra, such adsorption capacity variation at the low pressure range is ascribed to surface properties difference. With coalification, functional group elimination and hetero-aromatic bond breaking result in a substantial change of coal surface properties, leading to density decrease of the surface functional groups, especially the oxygen-containing groups [41]. At high relative pressures, the increase rate of water adsorption on lignite and bituminous coal is larger than on anthracite. From this perspective, despite the lower kinetic diameter of water molecules, it is mainly filled in mesopores and macropores of coals.

Water Vapor Adsorption-Desorption Isotherms
Water vapor adsorption-desorption isotherms for the coal samples were measured at the relative pressure up to 0.9 as shown in Figure 8. The isotherms are type Ⅲ corresponding to IUPAC classification [10]. This implies that the coal samples are of weak adsorption affinity with water. The isotherms can be divided into several sections. At low relative pressures, the amount of adsorbed water tends to be a linear increase trend with pressure, which is due to the interaction between water molecules and functional groups on coal surface. With the increase of relative pressure, the water molecules adsorb on top of the pre-adsorbed water molecules and clusters are formed. As the relative pressure increase further, the clusters merge to fill the pore space in the coals. This is shown as an rapid increase in water adsorption.
Regarding the coal rank, lignite has the largest water adsorption capacity at the whole pressure range, followed by bituminous coal, while anthracite has the smallest adsorption capacity. Combining this result with the FT-IR spectra, such adsorption capacity variation at the low pressure range is ascribed to surface properties difference. With coalification, functional group elimination and hetero-aromatic bond breaking result in a substantial change of coal surface properties, leading to density decrease of the surface functional groups, especially the oxygen-containing groups [41]. At high relative pressures, the increase rate of water adsorption on lignite and bituminous coal is larger than on anthracite. From this perspective, despite the lower kinetic diameter of water molecules, it is mainly filled in mesopores and macropores of coals. For the water isotherms of the three coal ranks, the adsorption-desorption hysteresis can be observed, stating the adsorption of water vapor on the coals is incompletely reversible. The hysteresis is probably due to the cohesiveness characteristic of water clusters by the strong hydrogen bonding between water molecules and this interaction maintains the clusters in a relatively stable For the water isotherms of the three coal ranks, the adsorption-desorption hysteresis can be observed, stating the adsorption of water vapor on the coals is incompletely reversible. The hysteresis is probably due to the cohesiveness characteristic of water clusters by the strong hydrogen bonding between water molecules and this interaction maintains the clusters in a relatively stable state with the progress of desorption [48][49][50].

Modeling of Water Vapor Adsorption
In order to better describe water vapor adsorption on the coal samples, the description capability of D'Arcy and Watt (D-W) equation and Dent equation are checked here. The two models distinguish the contribution of primary and secondary sites to adsorption. The primary sites, represented by oxygen-containing groups, have high binding energies. Secondary sites can occur either between the first water-occupied primary sites or on previously-formed secondary sites [51][52][53]. The D-W equation was given as: where h is the relative pressure, a represents the amount of adsorbed water for a specific h, s 0 and s are the density of the primary and secondary adsorption sites, K 1 and K 2 are the attraction between adsorbate and the primary and secondary sorption site, respectively. The Dent model has a similar form with the BET theory [53], and was given as: where K 1 and K 2 are the constants related to the adsorption energies of the primary and secondary sites, respectively, m 0 represents the monolayer adsorption capacity.
Equations (1) and (2) were applied to best-fit the water vapor isotherm and the results are shown in Figure 9. The fitting parameters are calculated and summarize in Table 3. Obviously, the adsorption isotherms are well fitted by D-W and Dent plots with R-square greater than 0.98 for all cases. The s 0 and m 0 of water adsorption varied with regards to different coal samples, indicating different density of the primary sites on coal surface. Table 3 shows that the density of both primary site and secondary site of lignite is the largest, followed by bituminous coal and anthracite in sequence. Also, the parameter s of water adsorption also decreased with the coal rank, indicating loss of surface functional groups during coalification process. where h is the relative pressure, a represents the amount of adsorbed water for a specific h, s0 and s are the density of the primary and secondary adsorption sites, K1 and K2 are the attraction between adsorbate and the primary and secondary sorption site, respectively. The Dent model has a similar form with the BET theory [53], and was given as: where K1 and K2 are the constants related to the adsorption energies of the primary and secondary sites, respectively, m0 represents the monolayer adsorption capacity.
Equations (1) and (2) were applied to best-fit the water vapor isotherm and the results are shown in Figure 9. The fitting parameters are calculated and summarize in Table 3. Obviously, the adsorption isotherms are well fitted by D-W and Dent plots with R-square greater than 0.98 for all cases. The s0 and m0 of water adsorption varied with regards to different coal samples, indicating different density of the primary sites on coal surface. Table 3 shows that the density of both primary site and secondary site of lignite is the largest, followed by bituminous coal and anthracite in sequence. Also, the parameter s of water adsorption also decreased with the coal rank, indicating loss of surface functional groups during coalification process.

Comparison of Different Methods in Coal Pore Characterization
According to the analysis mentioned above, the comparison of the measurable ranges of different methods to characterizing the coal pore structure is shown in Figure 10. SEM can cover almost the whole range of pore diameter using different magnification times. However, the main problem of this technique is that it can only provide the qualitative information of the pores. Meanwhile, it is difficult to position the same adsorption sites for comparison analysis, even for advanced operators. Standard CO 2 adsorption technique can obtain a certain quantitative information of micropore in coals, as it can only probe pores in the range of 0.5 nm to 0.9 nm. For micropores falling in the range of 0.9 nm and 2 nm and smaller than 0.5 nm, CO 2 adsorption technique is unfavorable. Physisorption method with N 2 adsorbate can probe mesopores in coal, while MIP can penetrate both the mesopores and macropores. The mesopore volume determined by N 2 adsorption and MIP have similar values in this study. Regarding physisorption with water vapor, it is not an effective method to probe micropores in coals as the water clusters mainly fill mesopores and macropores, instead of micropores, even though water vapor has a lower kinetic diameter. However, in CBM exploitation and CO 2 geosequestration, proper characterization of micropore in coal is important. Therefore, development of new methods is essential to explore and characterize a wider range of micropores in coals. it is not an effective method to probe micropores in coals as the water clusters mainly fill mesopores and macropores, instead of micropores, even though water vapor has a lower kinetic diameter. However, in CBM exploitation and CO2 geosequestration, proper characterization of micropore in coal is important. Therefore, development of new methods is essential to explore and characterize a wider range of micropores in coals. This study also show that both water vapor adsorption and FT-IR analysis can provide qualitative information of coal surface properties, but ineffective in quantification of functional groups on coal surface.

Conclusions
In order to explore the surface properties and pore structure of different coal rank, and hence comparatively address the limits of different probe methods, a comprehensive approach was used to probe the surface properties and pore structure of anthracite, bituminous coal and lignite, including SEM, N2, CO2 and water vapor adsorption, MIP and FT-IR.
The results show that wedge-shaped pores develop with the increase of coal rank due to compression effect. The determined SSA and pore volume of N2 adsorbate decrease with the increase of coal rank, while CO2 SSA and pore volume show an U-shaped function of coal rank. This is corresponding to the different measurable ranges of the two probes. MIP results indicate that the pore size of 10-100 nm accounted for 70.7-97.5% of the total volume in the macropore range. The water vapor adsorption and FT-IR results show that the density of surface functional groups decreased with the coal rank, resulting in a smaller water uptake capacity of the higher rank coal. The water vapor adsorption isotherms in coals can generally be described by both D-W and Dent equations. This study also show that both water vapor adsorption and FT-IR analysis can provide qualitative information of coal surface properties, but ineffective in quantification of functional groups on coal surface.

Conclusions
In order to explore the surface properties and pore structure of different coal rank, and hence comparatively address the limits of different probe methods, a comprehensive approach was used to probe the surface properties and pore structure of anthracite, bituminous coal and lignite, including SEM, N 2 , CO 2 and water vapor adsorption, MIP and FT-IR.
The results show that wedge-shaped pores develop with the increase of coal rank due to compression effect. The determined SSA and pore volume of N 2 adsorbate decrease with the increase of coal rank, while CO 2 SSA and pore volume show an U-shaped function of coal rank. This is corresponding to the different measurable ranges of the two probes. MIP results indicate that the pore size of 10-100 nm accounted for 70.7-97.5% of the total volume in the macropore range. The water vapor adsorption and FT-IR results show that the density of surface functional groups decreased with the coal rank, resulting in a smaller water uptake capacity of the higher rank coal. The water vapor adsorption isotherms in coals can generally be described by both D-W and Dent equations.
By comparing different methods in probing pore structure and surface properties of coal, it is found that physisorption method with CO 2 adsorbate is incompetent to probe the micropores in the range of 0.9 nm and 2 nm or smaller than 0.5 nm. N 2 adsorbate can probe mesopores in coal, while MIP can penetrate both the mesopores and macropores. Also, the mesopore volume determined with two methods above is similar. Even though water vapor has a lower kinetic diameter, it is unfavorable in probing micropores in coals, as the water clusters mainly fill mesopores and macropores, instead of micropores. For the surface properties of coal, either water vapor adsorption technique or FT-IR analysis can provide qualitative information of coal surface, but incompetent to provide quantitative information for functional groups.