Pore Structure Characterization of Shale Using Gas Physisorption : Effect of Chemical Compositions

In this study, the pore structure characteristics of Canadian Horn River basin shales with various chemical compositions were evaluated using gas physisorption analyses. The samples used in this research were obtained from two different regions (shallow and deep regions) of rock cuttings during the drilling of the shale gas field located in Horn River basin. The pore size, specific surface area, total pore volume, micropore surface area, and micropore volume of the shale samples were measured using both nitrogen and CO2. The results indicated that the pore size was not a function of chemical composition, while distinct trends were observed for other macroscopic and microscopic pore-related properties. In particular, the greatest specific surface area and total pore volume were observed for silica-rich carbonate shales, while clay-rich siliceous shales exhibited the greatest micropore volume and micropore surface area. The trends clearly suggested that macroscopic and microscopic pore-related properties of the Canadian Horn River basin shales were closely related to their chemical composition. Furthermore, a stronger correlation was observed between the quartz content and the micropore-related physical properties of shales (i.e., the micropore surface area and micropore volume) in comparison to other properties.


Introduction
With an increasing natural gas consumption worldwide, and in the face of challenges involved in finding massive conventional gas resources, novel and unconventional gas resources have been actively explored.As a representative unconventional gas resource, the shale gas resources of the world are known to be abundant.More recently, the amount of gas production from shale gas has increased in Canada, the USA, and other countries, owing to the advancements in cutting-edge technologies including horizontal drilling and multi-stage hydraulic fracturing [1].Nevertheless, the porous shale medium that stores hydrocarbon has complicated and anisotropic geological structures.
Accordingly, the condition of this gas significantly varies depending on the region [2][3][4].Thus, the same evaluation criteria for estimating the gas in a given place can hardly be applied to a variety of regions in different countries, indicating that it is difficult to provide information about the gas found in each territory.The international considerations for effectively using shale gas include its successful supply and economic feasibility.Moreover, understanding of the storage and flow mechanism of shale gas is essential in optimizing the development plan and subsequent economic production of gas.

X-ray Diffraction (XRD) Analysis
For XRD analysis, the shale samples were dried at 50 • C for 5 h after being milled to a fine particle size, and random grain mounts were prepared and analyzed.XRD patterns were recorded with a Bruker D8 HRXRD X-ray diffractometer (Bruker, Karlsruhe, Germany) using Ni-filtered Cu Kα radiation at 40 kV and 40 mV, as well as a scan rate of 1 • /min with a step size of 0.01 • in 10 • -80 • 2-theta.The mineralogy was quantified by Rietveld refinement analyses using the TOPAS 3 program [18,19].

Low-Pressure N 2 and CO 2 Isotherm Analysis
Low pressure gas physisorption analyses using both nitrogen (N 2 ) and carbon dioxide (CO 2 ) gases were used to measure the pore size distribution between the pore sizes of 0.35 and 150 nm and describe the pore space of organic, microporous shales, as average pore size distributions are in the nanometer scale, into which gas can effectively penetrate.Between 1 and 2 g of cutting samples (size < 1 mm) prepared for gas physisorption analysis were first outgassed for a minimum of 4 days in a vacuum oven [7].N 2 adsorption-desorption isotherms were collected at 77 K using a Micromeritics 3Flex.The total pore volume was determined from the amount of N 2 adsorbed at a relative pressure, P/P 0 , of 0.99; where P is the gas vapor pressure in the system and P 0 is the vapor pressure at the temperature of interest.The meso-macropore size distributions (2.0-150 nm) were analyzed from adsorption data in the partial relative pressure range of 0.3-0.99 using the Barrett-Joyner-Halenda (BJH) method [20,21].The specific surface area was assessed from N 2 adsorption data in the partial relative pressure range of 0.05-0.30using the BET method [22,23] based on the following equation: where W is the weight of the nitrogen gas at relative pressure P 0 /P, W m is the weight of the monolayer adsorbent (nitrogen), and C is the BET constant which relates to the sorption energy between the adsorbent and adsorbate.Carbon dioxide (CO 2 ) adsorption data (273 K) were also collected on the same apparatus.For mircropore analysis, CO 2 adsorption isotherms were measured over a relative pressure range of 4.5 × 10 −5 −3.2 × 10 −2 at 273 K and monolayer capacities were determined using the Dubinin-Radushkevich (D-R) equation [24]: where V is the volume of the sorbed gas at equilibrium pressure (cm 3 /g, Standard Temperature and Pressure, STP), V 0 is the total micropore volume (cm 3 /g, STP), S is a constant, P is pressure and P 0 is saturation vapor pressure.The lower limit of micropore depends on the kinetic diameter of the probing gas molecule, such as 0.35 nm for CO 2 .Micropore volumes (pore size < 2.0 nm) were measured by the volume of adsorbed CO 2 (which completely fills the micropores), expressed in terms of bulk liquid at atmospheric pressure and 273 K [24,25].As discussed by Dubinin and Stoeckli [26], the theory of volume filling of micropores is a rational basis for the description of microporous structures in carbonaceous materials.The measurements for N 2 and CO 2 isotherms were repeated the times for each shale sample at the same sampling depth.

Mineralogy and Shale Lithofacies
The chemical composition of eight representative samples corresponding to each sub-region is summarized in Table  and illite contents.To increase the reliability of the results regarding the changes in the pore structure dependent on chemical composition, two shales having a similar chemical composition to each other were grouped from two different depth regions (shallow versus deep).Note that four major chemical compositions (i.e., quartz dominated shale, carbonaceous shale, clay dominated shale, and mixed shale) were eventually selected based on the information of the sCore Lithofacies Classification [27].Specifically, the Horn River basin can be divided into three groups: Muskwa, Otter Park, and Evie formation [28].Evie and Otter Park were known to be mainly composed of deep to marginal tropical shales and shallow tropical carbonates, while Muskwa consisted of deep to marginal tropical shales.Their ternary diagram based on the weight percent of minerals (Table 1) is also shown in Figure 1.Note that four major chemical compositions (i.e., quartz dominated shale, carbonaceous shale, clay dominated shale, and mixed shale) were eventually selected based on the information of the sCore Lithofacies Classification [27].Specifically, the Horn River basin can be divided into three groups: Muskwa, Otter Park, and Evie formation [28].Evie and Otter Park were known to be mainly composed of deep to marginal tropical shales and shallow tropical carbonates, while Muskwa consisted of deep to marginal tropical shales.Their ternary diagram based on the weight percent of minerals (Table 1) is also shown in Figure 1.The plotting is based on the quartz-clay-carbonate system, and the plotted data were obtained by normalizing the weight percent information summarized in Table 1.
As shown in Figure 1, two shale samples classified as the same category have a similar chemical composition but are from different depths.In the shale samples obtained from 2375 m and 2465 m underground, the quartz level was less than 40%, and carbonate was the primary component (50-80%).The sample classification and lithofacies of this study are shown in Table 2.In this study, all samples were categorized into four groups (i.e., four different chemical compositions) based on a 50 m or more difference in the sampling depth (i.e., shallow versus deep) (see Figure 1 and Table 2).1.
As shown in Figure 1, two shale samples classified as the same category have a similar chemical composition but are from different depths.In the shale samples obtained from 2375 m and 2465 m underground, the quartz level was less than 40%, and carbonate was the primary component (50-80%).The sample classification and lithofacies of this study are shown in Table 2.In this study, all samples were categorized into four groups (i.e., four different chemical compositions) based on a 50 m or more difference in the sampling depth (i.e., shallow versus deep) (see Figure 1 and Table 2).2. c Determined by averaging the BJH pore size distribution curve in Figure 3.The data in Figure 3 were determined by the BJH model from the adsorption branch of the N 2 isotherm in Figure 2. d Determined at the micropore size of 2 nm from the cumulative pore volume-micropore size curve in Figure 5.The data in Figure 5 were calculated by the D-R method from the adsorption branch of CO 2 isotherm in Figure 4.

Pore Characteristics of the Shale with Chemical Composition
Figure 2 shows the adsorption-desorption isotherm curves for every sample, which were observed as isotherm type II [6,24].In all the shale samples, a complete monolayer adsorption was observed up to the relative pressure level of approximately 0.1 before a multi-layer adsorption was initiated [24].Therefore, a multi-layer adsorption occurred on the shale surface, and the meso-macropore-level pores were found to be slit-shaped [24].The hysteresis loop of all the shale samples was confirmed as H3 (hysteresis loop type 3).Thus, the meso-macropore size distribution was determined using the desorption branch based on the BJH model [24].
Figure 3 shows the distribution of the meso-macropore pore size for all the samples.The shale samples that had four different chemical compositions exhibited slightly different curve shapes.Nevertheless, their mean pore size ranged from 18 nm to 25 nm without significant difference (see Table 2).This indicated that the chemical composition did not influence the meso-macropore size distribution, regardless of the sampling depth.However, a distinct trend in the specific surface area and the total pore volume was observed among the samples with different chemical compositions, which is valid for two different sampling regions.
Regardless of the sampling region, silica-rich carbonate shale samples consistently exhibited greater values of specific surface area than other samples (i.e., carbonate-siliceous shale, clay-rich siliceous shale and mixed shale) (Table 2).Quantitatively, for the shallow region, the BET specific surface area of sample-1 was determined to be 1.61 m 2 /g, while sample-3, sample-5, and sample-7 were 0.82, 1.00 and 1.25 m 2 /g, respectively.The same trend was also observed for the deep region.For the total pore volume in the shallow region, sample-1 exhibited 0.0123 cm 3 /g.This value is much greater than those of other samples.Similar to the specific surface area, this observation is also true for the deep region.This trend is likely due to the difference in the amount of carbonate in the samples.Previous studies reported that carbonate minerals are easily dissolved relative to other minerals, resulting in the greater development of meso-or macropores [29].Figure 4 shows the carbon dioxide adsorption isotherms of all the samples with different chemical compositions.The corresponding micropore properties (i.e., micropore volume and micropore surface area) are also presented in Figure 5 and Table 2. Unlike the observations for the specific surface area and total pore volume, the clay-rich siliceous shale samples (sample-5 and 6) exhibited the greatest values compared with the silica-rich carbonate shale (sample-1 and 2) and carbonate-siliceous shale (sample-3 and 4) samples.Note that the mixed shale samples (sample-7 and 8) were excluded from this comparison, since no clear trend was observed at either the shallow or deep regions.Specifically, for the shallow region, the micropore volume of sample-5 was found to be 0.30 mm 3 /g, which is relatively greater than those for sample-1 and sample-3 (0.23 and 0.22 mm 3 /g, respectively).
A similar trend was observed for the micropore surface area, where a greater value for sample-5 (2.17 m 2 /g) was observed than that for sample-1 (1.51 m 2 /g) and sample-3 (1.98 m 2 /g).In addition, the differences in the two micropore properties became more obvious for the deep region.Sample-6 showed a 1.6-1.8increase in the micropore volume and micropore surface area than that for sample-2 and sample-4.This suggests that the shales containing high amount of clay minerals (i.e., sample-5 and -6) show better development of micropore structures, an observation that could be explained by the unique crystalline structure of the clays (i.e., a layer structure with a stack of sheet) [30].Overall, the results from the specific surface area, the total pore volume, the micropore volume, and the micropore surface area suggest that the nanopore formation of the Canadian shale samples used in this study is likely to be affected by the chemical composition.
Figure 4 shows the carbon dioxide adsorption isotherms of all the samples with different chemical compositions.The corresponding micropore properties (i.e., micropore volume and micropore surface area) are also presented in Figure 5 and Table 2. Unlike the observations for the specific surface area and total pore volume, the clay-rich siliceous shale samples (sample-5 and 6) exhibited the greatest values compared with the silica-rich carbonate shale (sample-1 and 2) and carbonate-siliceous shale (sample-3 and 4) samples.Note that the mixed shale samples (sample-7 and 8) were excluded from this comparison, since no clear trend was observed at either the shallow or deep regions.Specifically, for the shallow region, the micropore volume of sample-5 was found to be 0.30 mm 3 /g, which is relatively greater than those for sample-1 and sample-3 (0.23 and 0.22 mm 3 /g, respectively).
A similar trend was observed for the micropore surface area, where a greater value for sample-5 (2.17 m 2 /g) was observed than that for sample-1 (1.51 m 2 /g) and sample-3 (1.98 m 2 /g).In addition, the differences in the two micropore properties became more obvious for the deep region.Sample-6 showed a 1.6-1.8increase in the micropore volume and micropore surface area than that for sample-2 and sample-4.This suggests that the shales containing high amount of clay minerals (i.e., sample-5 and -6) show better development of micropore structures, an observation that could be explained by the unique crystalline structure of the clays (i.e., a layer structure with a stack of sheet) [30].Overall, the results from the specific surface area, the total pore volume, the micropore volume, and the micropore surface area suggest that the nanopore formation of the Canadian shale samples used in this study is likely to be affected by the chemical composition.

Correlation of the Pore Structure of the Shale with Chemical Composition
Since the purpose of the present study is to examine the correlation of both the micropore surface area and the micropore volume of shales with its chemical composition, we plotted each physical property as a function of the contents of quartz, carbonate, and clay.The results are presented in Figure 6. Figure 6a shows the correlation between the micropore surface area and the chemical composition.The coefficients of determination (R-squared value) for the fitting results as a function of quartz, carbonate, and clay were determined to be 0.6359, 0.6461 and 0.1809, respectively.This indicates that the quartz content has the strongest correlation with the micropore surface area of shales.
Also note that the carbonate content exhibits a similar coefficient of determination (~0.6461) with the quartz, but, unlike quartz, it shows a negative correlation.The correlation results between the micropore volume and the chemical composition of shales are also presented in Figure 6b.The coefficient of determination of the fitting results as a function of the quartz content was determined to be 0.8020, which was greater than those of the carbonate and clay contents.This result is consistent with that for the micropore surface area.
While the quartz content has a strong relationship with the micropore volume, the clay content shows even worse correlation than the micropore surface area (the coefficient of determination = 0.0288).The coefficient of determination of carbonate was 0.5974, indicating a negative tendency similar to the micropore surface area.Accordingly, overall results suggest that the quartz content might be a good indicator for estimating the amount of shale gas adsorbed compared to the carbonate and clay contents, and that the presence of clay clearly has no correlation with the two physical properties investigated herein.

Correlation of the Pore Structure of the Shale with Chemical Composition
Since the purpose of the present study is to examine the correlation of both the micropore surface area and the micropore volume of shales with its chemical composition, we plotted each physical property as a function of the contents of quartz, carbonate, and clay.The results are presented in Figure 6. Figure 6a shows the correlation between the micropore surface area and the chemical composition.The coefficients of determination (R-squared value) for the fitting results as a function of quartz, carbonate, and clay were determined to be 0.6359, 0.6461 and 0.1809, respectively.This indicates that the quartz content has the strongest correlation with the micropore surface area of shales.
Also note that the carbonate content exhibits a similar coefficient of determination (~0.6461) with the quartz, but, unlike quartz, it shows a negative correlation.The correlation results between the micropore volume and the chemical composition of shales are also presented in Figure 6b.The coefficient of determination of the fitting results as a function of the quartz content was determined to be 0.8020, which was greater than those of the carbonate and clay contents.This result is consistent with that for the micropore surface area.
While the quartz content has a strong relationship with the micropore volume, the clay content shows even worse correlation than the micropore surface area (the coefficient of determination = 0.0288).The coefficient of determination of carbonate was 0.5974, indicating a negative tendency similar to the micropore surface area.Accordingly, overall results suggest that the quartz content might be a good indicator for estimating the amount of shale gas adsorbed compared to the carbonate and clay contents, and that the presence of clay clearly has no correlation with the two physical properties investigated herein.
Minerals 2017, 7, 66 9 of 11 Unlike our results, a previous study reported that no micropores or mesopores existed in siliceous/quartz-rich shale (high Si/Al), and that the porosity was as low as 1% [10].Conversely, clay-rich shale (low Si/Al) had a high porosity.However, it should be noted that the reported results were not obtained from the CO 2 adsorption but from Hg intrusion analysis.Additionally, the shale samples used in their study had a wide range of specific surface areas unlike those used in our study, which could lead to the difference in the correlation trend.
Since micropore volume has a positive correlation with specific surface area [8,10,31], our results have an important significance regarding the feasibility of assessing the amount of shale gas from ore-related properties of shales.Nevertheless, it may not be appropriate to generalize that the higher quartz content of shales results in the greater formation of micropores.To enhance the reliability of our outcomes, repetitive tests using a larger sample size are necessary.Additionally, diverse shale samples from various regions in Canada and other countries must be analyzed.
Minerals 2017, 7, 66 9 of 11 Unlike our results, a previous study reported that no micropores or mesopores existed in siliceous/quartz-rich shale (high Si/Al), and that the porosity was as low as 1% [10].Conversely, clay-rich shale (low Si/Al) had a high porosity.However, it should be noted that the reported results were not obtained from the CO2 adsorption but from Hg intrusion analysis.Additionally, the shale samples used in their study had a wide range of specific surface areas unlike those used in our study, which could lead to the difference in the correlation trend.
Since micropore volume has a positive correlation with specific surface area [8,10,31], our results have an important significance regarding the feasibility of assessing the amount of shale gas from ore-related properties of shales.Nevertheless, it may not be appropriate to generalize that the higher quartz content of shales results in the greater formation of micropores.To enhance the reliability of our outcomes, repetitive tests using a larger sample size are necessary.Additionally, diverse shale samples from various regions in Canada and other countries must be analyzed.

Conclusions
In the present study, pore-related physical properties such as the BET surface area, pore volume, and pore size of shales were thoroughly tested to examine the correlation with chemical composition of shales obtained from rock cuttings during drilling of the shale gas field located in Horn River basin.The key findings are presented below: 1.The N2 and CO2 gas physisorption technique was found to be suitable for analyzing the pore-related properties of shales (e.g., BET surface area, pore volume, and pore size).2. Chemical composition did not play a critical role in the BJH pore size.However, the chemical composition was critical for the specific/micropore surface area and the total/micropore volume.The silica-rich carbonate shale showed the greatest specific surface area and total pore volume, while the micropore volume and micropore surface area was found to be the greatest for the clay-rich siliceous shale samples.3.Both the micropore surface area and micropore volume were found to be a strong function of the quartz content as compared to the carbonate and clay contents.
The outcomes from the present study suggest that pore-related properties (especially micropore properties) might be a good candidate for estimating shale gas adsorption characteristics.However, acquiring more datasets for shales samples with diverse chemical compositions and from different regions, as well as correlating their microscopic properties (e.g., total/micropore volume,

Conclusions
In the present study, pore-related physical properties such as the BET surface area, pore volume, and pore size of shales were thoroughly tested to examine the correlation with chemical composition of shales obtained from rock cuttings during drilling of the shale gas field located in Horn River basin.
The key findings are presented below: 1.
The N 2 and CO 2 gas physisorption technique was found to be suitable for analyzing the pore-related properties of shales (e.g., BET surface area, pore volume, and pore size).2.
Chemical composition did not play a critical role in the BJH pore size.However, the chemical composition was critical for the specific/micropore surface area and the total/micropore volume.The silica-rich carbonate shale showed the greatest specific surface area and total pore volume, while the micropore volume and micropore surface area was found to be the greatest for the clay-rich siliceous shale samples.

3.
Both the micropore surface area and micropore volume were found to be a strong function of the quartz content as compared to the carbonate and clay contents.
The outcomes from the present study suggest that pore-related properties (especially micropore properties) might be a good candidate for estimating shale gas adsorption characteristics.However, acquiring more datasets for shales samples with diverse chemical compositions and from different regions, as well as correlating their microscopic properties (e.g., total/micropore volume, specific/micropore surface area) with their macroscopic properties (e.g., core porosity, core permeability) is guaranteed for universal application.

Figure 1 .
Figure 1.Ternary diagram of the shales.The plotting is based on the quartz-clay-carbonate system, and the plotted data were obtained by normalizing the weight percent information summarized in Table1.

Figure 1 .
Figure 1.Ternary diagram of the shales.The plotting is based on the quartz-clay-carbonate system, and the plotted data were obtained by normalizing the weight percent information summarized in Table1.

Figure 2 .
Figure 2.An average plot of nitrogen adsorption-desorption isotherms of the shale samples with different chemical compositions at two different depth regions: (a) silica-rich carbonate shale; (b) carbonate-siliceous shale; (c) clay-rich siliceous shale; and (d) mixed shale.

Figure 3 .
Figure 3.An average plot of Barrett-Joyner-Halenda (BJH) pore size distribution curves of the shale samples with different chemical compositions at two different depth regions: (a) silica-rich carbonate shale; (b) carbonate-siliceous shale; (c) clay-rich siliceous shale; and (d) mixed shale.The dV/dlog (D) represents the BJH differential pore volume of the shale samples.

Figure 2 . 11 Figure 2 .
Figure 2.An average plot of nitrogen adsorption-desorption isotherms of the shale samples with different chemical compositions at two different depth regions: (a) silica-rich carbonate shale; (b) carbonate-siliceous shale; (c) clay-rich siliceous shale; and (d) mixed shale.

Figure 3 .
Figure 3.An average plot of Barrett-Joyner-Halenda (BJH) pore size distribution curves of the shale samples with different chemical compositions at two different depth regions: (a) silica-rich carbonate shale; (b) carbonate-siliceous shale; (c) clay-rich siliceous shale; and (d) mixed shale.The dV/dlog (D) represents the BJH differential pore volume of the shale samples.

Figure 3 .
Figure 3.An average plot of Barrett-Joyner-Halenda (BJH) pore size distribution curves of the shale samples with different chemical compositions at two different depth regions: (a) silica-rich carbonate shale; (b) carbonate-siliceous shale; (c) clay-rich siliceous shale; and (d) mixed shale.The dV/dlog (D) represents the BJH differential pore volume of the shale samples.

Figure 4 .
Figure 4.An average plot of carbon dioxide adsorption isotherms of the shale samples with different chemical compositions at two different depth regions: (a) silica-rich carbonate shale; (b) carbonate-siliceous shale; (c) clay-rich siliceous shale; and (d) mixed shale.

Figure 4 .Figure 5 .
Figure 4.An average plot of carbon dioxide adsorption isotherms of the shale samples with different chemical compositions at two different depth regions: (a) silica-rich carbonate shale; (b) carbonate-siliceous shale; (c) clay-rich siliceous shale; and (d) mixed shale.

Figure 5 .
Figure 5.An average plot of cumulative micropore volume of the shale samples with different chemical compositions at two different depth regions: (a) silica-rich carbonate shale; (b) carbonate-siliceous shale; (c) clay-rich siliceous shale; and (d) mixed shale.

Figure 6 .
Figure 6.Correlation plots between two physical properties of shales ((a) micropore surface area and (b) micropore volume) and main, individual chemical components (quartz (black circle), carbonate (void square), and clay (void triangle)).Coefficient of determination (R 2 ) values for each correlation plot are also presented for comparison.

Figure 6 .
Figure 6.Correlation plots between two physical properties of shales ((a) micropore surface area and (b) micropore volume) and main, individual chemical components (quartz (black circle), carbonate (void square), and clay (void triangle)).Coefficient of determination (R 2 ) values for each correlation plot are also presented for comparison.
. All the samples include comparatively high levels of quartz, calcite, dolomite,

Table 1 .
Mineralogical composition and total organic carbon (TOC; wt %) of the shale with different sampling depths.

Table 1 .
Mineralogical composition and total organic carbon (TOC; wt %) of the shale with different sampling depths.

Table 2 .
Lithofacies and physical properties of the shales with different sampling depths.
a Calculated from the N 2 adsorption isotherms in Figure2using by BET model.b Determined at the relative pressure of 0.99 from the N 2 adsorption isotherms in Figure