sign in sign up
<<
Home , Kerogen

Article

pubs.acs.org/accounts

Methane and CO2 Adsorption Capacities of Kerogen in the Eagle Ford from Molecular Simulation Published as part of the Accounts of Chemical Research special issue “Chemistry of Geologic Storage”. Peter Psarras,† Randall Holmes,‡ Vikram Vishal,§ and Jennifer Wilcox*,†

† Department of Chemical and Biological Engineering, Colorado School of Mines, Golden, Colorado 80401, United States ‡ Department of Earth System Science, Stanford University, Stanford, California 94305, United States § Department of Earth Sciences, Indian Institute of Technology Bombay, Mumbai 400076, India

*S Supporting Information

CONSPECTUS: Over the past decade, the United States has become a world leader in production, thanks in part to a large-fold increase in recovery from unconventional resources, i.e., shale rock and tight reservoirs. In an attempt to help mitigate climate change, these depleted formations are being considered for their long-term CO2 storage potential. Because of the variability in mineral and structural composition from one formation to the next (even within the same region), it is imperative to understand the adsorption behavior of CH4 and CO2 in the context of specific conditions and pore surface chemistry, i.e., relative total organic content (TOC), clay, and surface functionality. This study examines two Eagle Ford shale samples, both recovered from shale that was extracted at depths of approximately 3800 m and having low clay content (i.e., less than 5%) and similar mineral compositions but distinct TOCs (i.e., 2% and 5%, respectively). Experimentally validated models of kerogen were used to the estimate CH4 and CO2 adsorption capacities. The pore size distributions modeled were derived from low-pressure adsorption isotherm data using CO2 and N2 as probe gases for micropores and mesopores, respectively. Given the presence of in these natural systems, the role of surface chemistry on modeled kerogen pore surfaces was investigated. Several functional groups associated with surface-dissociated water were considered. Pressure conditions from 10 to 50 bar were investigated using grand canonical Monte Carlo simulations along with typical outgassing ° temperatures used in many shale characterization and adsorption studies (i.e., 60 and 250 C). Both CO2 and N2 were used as probe gases to determine the total pore volume available for gas adsorption spanning pore diameters ranging from 0.3 to 30 nm. The impacts of surface chemistry, outgassing temperature, and the inclusion of nanopores with diameters of less than 1.5 nm were determined for applications of CH4 and CO2 storage from samples of the gas-producing region of the Eagle Ford Shale. At ° 50 bar and temperatures of 60 and 250 C, CH4 adsorption increased across all surface chemistries considered by 60% and 2-fold, respectively. In the case of CO2, the surface chemistry played a role at both 10 and 50 bar. For instance, at temperatures of 60 ° and 250 C, CO2 adsorption increased across all surface chemistries by 6-fold and just over 2-fold, respectively. It was also found Downloaded via WORCESTER POLYTECHNIC INST on October 10, 2018 at 23:16:47 (UTC). that at both 10 and 50 bar, if too low an outgassing temperature is used, this may lead to a 2-fold underestimation of gas in place. See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. Finally, neglecting to include pores with diameters of less than 1.5 nm has the potential to underestimate pore volume by up to 28%. Taking into consideration these aspects of kerogen and shale characterization in general will lead to improvements in estimating the CH4 and CO2 storage potential of gas .

■ INTRODUCTION against the rock formation until fracturing occurs. The water contains micrometer-sized particles known as proppant that fill The production of natural gas from shale gas and plays “ ” comprises approximately half of the natural gas production in the expanding fracture, thereby propping open the fracture the U.S. today. According to the U.S. Energy Information network. This allows the transport of from the Administration (EIA), production from these opportunities is formation back to the wellbore and ultimately to the surface for expected to rise from approximately 14 trillion cubic feet (Tcf) recovery. According to the U.S. EIA, in 2000, 26 000 in 2015 to up to 30 Tcf in 2040, which will represent nearly hydraulically fractured wells producing 3.6 billion cubic feet 70% of total gas production in 2040. Most natural gas today is per day (Bcf/d) constituted less than 7% of U.S. production, produced through hydraulic fracturing, often in combination with horizontal drilling. This process involves forcing water- Received: January 1, 2017 based fracturing fluids under high pressure from a wellbore Published: August 1, 2017

© 2017 American Chemical Society 1818 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article while by 2015 the number of these wells had increased to 2015 at approximately 7000 million cubic feet per day (Mcf/d) 300 000 at a production rate of more than 53 Bcf/d, and steadily declined to just under 6000 Mcf/d at the end of representing nearly 70% of U.S. production. As more is learned 2016. about these complex systems, the techniques for recovering The unconventional nature of these complex systems is due unconventional natural gas are becoming more refined and to the fact that the hydrocarbons reside in a nanoporous matrix ultimately cost-effective even in a climate of low gas prices. composed of heterogeneous distributions of clay, kerogen, A question remains as to whether following the depletion of calcite, quartz, and feldspar. This is different from conventional these unconventional shale gas and tight oil reservoirs, might systems, which are composed primarily of sand- they be used as reliable storage opportunities for CO2. Because sized grains, where the grain spacing gives rise to pores with of the lack of an economic incentive for companies to reliably sizes on the order of tens of micrometers that are largely fi 6 fi store CO2 in subsurface formations, a rst step may be to connected. Simply put, it is signi cantly easier to recover mimic the CO2 utilization path that has taken place with hydrocarbons from well-connected micrometer-sized pore enhanced oil recovery (EOR) today. Investigation of the networks than from isolated nanopore networks. In addition, interaction and subsequent adsorption of CO2 within these the added complexity associated with the high clay and kerogen complex natural systems will provide greater insight into the composition can impact the rock rheological attributes and fl “ ” feasibility of using CO2 as an enhanced recovery uid in subsequent fracability of the rock, making recovery even more addition to whether the depleted reservoirs may serve as a cumbersome. Coarse-grained models have been used in the permanent sink for CO2. Gas shales are complex structural and past to assist in the prediction of recovery from conventional mineralogical systems comprising a heterogeneous mixture of sources, but it has become widely accepted that these calcite, quartz, pyrite, clay, and kerogen. Their composition techniques fail at the nanoscale since they rely on Darcy flow, impacts the “fracability” of the rock and hence their ability to which is not the transport mechanism at the nanoscale, where produce oil and gas. Lessons learned from enhanced coalbed fluid−fluid interactions compete with pore wall−fluid inter- 7 (ECBM) recovery have shown that when CO2 is used actions. This leads to the application of molecular-level models to enhance the recovery of natural gas because of its preferred or statistical mechanics to assist in elucidating the chemical − adsorption within the matrix, the CO2 coal carbon physics that determines the adsorption and transport properties interaction changes the geomechanical attributes of the coal. of hydrocarbons in gas shales. These results may be used to 1 For example, Hagin and Zoback found that CO2 injection into accurately determine the gas capacities of these reservoirs once fi samples from the Powder River Basin induced a constitutive they are depleted and to assist in de ning the interplay of CO2 change in coal properties; specifically, these samples became with subsurface fluids. more viscous, less elastic, and less brittle in response to CO2 The composition of the Eagle Ford Shale varies throughout adsorption. Studies carried out by Vishal et al.2 measuring the the formation depending on whether one is in the gas- permeability of subcritical CO2 in naturally fractured Indian producing or oil-producing regime. The oil-producing region bituminous coal over a range of down-hole stress conditions corresponds to an average clay content of 11% with a were also consistent with these findings. Depending upon the breakdown of 25% illite + mica, 43% illite/smectite, and 32% mineralogical composition of the gas shale, there is a risk that kaolinite. The gas-producing region corresponds to an average ffi the injection of CO2 may similarly lead to di culty in total clay content of 20% with a breakdown of 36% illite + mica, fracturing. However, shales that are rich in carbonate and quartz 53% illite/smectite, 7% kaolinite, and 5% chlorite and a total lead to a formation that is more brittle and potentially more organic content (TOC) between 2 and 6%.8 The two Eagle “fracable” than those that are higher in clays and kerogen, for Ford samples under investigation in the current work are instance. The brittleness index (BI), which attempts to quantify consistent with the composition from the gas-producing part of shale play brittleness as a function of mineral composition and the formation. Their mineral composition will be discussed , is typically expressed as a ratio of quartz3 (or quartz later in this Account in greater detail, but these samples contain and dolomite)4 to total soft (clay and kerogen) and hard less than 5% illite and have TOCs of 2 and 5%, respectively. (quartz, feldspar, carbonates) components. In particular, Eagle Because of the low clay content of these samples in particular, Ford shale falls into this category with average carbonate and the focus of the models in this work are on kerogen. This is a quartz contents of 60 and 18%, respectively. It is important to suitable first step since the kerogen is the hydrocarbon source understand the nature of the interaction of CO2 with the and likely where CO2 would be stored if storage in these kerogen components of shale to determine its impact on the depleted formations becomes a viable option. geomechanical properties of the rock. Molecular simulations of CO2 adsorption in the organic Named for the town of Eagle Ford, Texas, the Eagle Ford matrix using pure and functionalized surfaces as a Shale is a late-Cretaceous formation that extends from the proxy for kerogen show that the adsorption capacity is strongly Mexican border in south Texas to the East Texas Basin and is correlated with the surface heterogeneity and pore size approximately 50 miles wide and 400 miles long. With an distribution.9 Surface functionalities can interact selectively ff average thickness of 250 feet, it is underlain by the Buda with the quadrupole moment of CO2,ane ect that has been Limestone and Woodbine Formation and is succeeded by the shown to be enhanced in smaller pores (d < 2.0 nm), as Austin Chalk at a depth of approximately 4000 to 12 000 ft. overlapping wall potentials combine to further stabilize CO2 in Although historically the Eagle Ford Shale was viewed as a the pore volume. Heller and Zoback10 examined the role of for the Austin Chalk play, there has been significant adsorption in the production from gas shales sourced from the activity recently. Depending on the location, the wells drilled in Barnett, Marcellus, Montney, and Eagle Ford plays as well as the reservoir tap into the dry-gas, gas-condensate, and oil for pure carbon, used as a proxy for kerogen. The Barnett windows. Oil production peaked in 2015 at approximately 1.7 sample displayed the largest adsorption capacity as well as the million barrels per day and decreased to just over 0.9 millions highest TOC (5.3%) among all of the samples, while the Eagle barrels per day in 2016.5 Natural gas production also peaked in Ford sample was least adsorptive but not lowest in TOC (1.8%

1819 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article

Figure 1. Probe gas isotherms for two Eagle Ford samples, EF1 (squares) and EF6 (circles). The adsorption and desorption branches are represented by solid and open symbols, respectively. vs 1.2% for the Marcellus sample). This discrepancy was account for the volume consumed by the sorbed gas. attributed to the difference in clay composition (4.9% for Eagle Correction for this “double counting” can be done by Ford vs 51.4% for Marcellus) indicating that clay may play an subtracting out the free gas volume taken up by the adsorbed 15 important role in adsorption. Furthermore, the CO2 adsorption gas (termed the Gibbs correction, or excess adsorption). The capacity was 2−3 times greater than the CH adsorption excess adsorption (nσ) can be calculated from the absolute 4 ρ capacity on all of the samples. This observation is consistent loading (na), the bulk-phase gas density ( g), and the volume 16 with many reported values of the exchange ratio, i.e., the molar occupied by the bulk gas (Vg): 11 ratio of CO2 that is anticipated to displace CH4. This nσ =−nVagρg parameter largely dictates the CO2 storage potential in a material. For example, the CO2 storage potential in coal has More difficult to treat are underestimations, as these are likely been estimated to lie in the range of 260 to 1480 gigatons (Gt), caused by inadequate characterization of the porosity of the where the low and high estimates reflect conservative and material, leading to an undermeasurement of the volume due to optimistic ratios for CO2/CH4 exchange. This exchange ratio the nanopores. typically ranges from 2:1 to 3:1 but can reach as high as 5:1 for supercritical CO2 (scCO2) and varies with depth and shale ■ GAS ADSORPTION AS AN IMAGING TOOL FOR composition. DETERMINING PORE SIZE DISTRIBUTION Several studies have argued that the potential for CO2 Low-pressure gas adsorption (LPGA) has been established as storage in ECBM may become self-limiting as a result of a an effective technique to measure International Union of Pure loss of permeability caused by CO2-injection-induced coal 12,13 and Applied Chemistry (IUPAC) micropores (<2 nm), matrix swelling. However, it has been determined that the mesopores (2−50 nm), and some macropores (>50 nm).17 magnitude of swelling varies roughly linearly with the The strength of LPGA methods lies in using the resulting magnitude of absorption of CO2 within the . isotherm data to calculate pore capacities, surface areas, and Thus, as the adsorption on gas shales is approximately an order pore size distributions (PSDs) with mathematical models such of magnitude less than that observed on , the observed as nonlocal density functional theory (NLDFT) or quenched swelling strain should be at least an order of magnitude solid density functional theory (QSDFT). These models 10 smaller. account for adsorbent pore-filling mechanisms that significantly Since the exchange ratio is instrumental in predicting CO2 improve material characterization because they account for the storage potential in shales, it is important to describe a reliable interactions between adsorbent surfaces and adsorbate gases. method for estimating the gas in place (GIP). Reservoirs that Material characterization based on techniques such as the produce both gas and condensate are typically difficult to Barrett−Joyner−Halenda (BJH) method can underestimate describe, as condensate can drop out as the reservoir pressure pore characteristics by as much as 20−30% as they do not falls below the dew point. However, the shale samples account for these interactions in micropores and mesopores examined in this study fall within the “dry-gas window”, and with sizes of less than approximately 10 nm.16 A surface GIP estimates can be approached via simple gas equations of adsorbent must be selected to approximate kerogen when using state. Procedures for GIP estimates have been outlined by the DFT methods. As organic matter such as kerogen has been United States Geological Survey (USGS);14 however, discrep- established as the primary source of shale porosity18,19 by ancies in GIP estimates can be linked to several factors. studies such as that by Kuila et al.,20 who used LPGA results in Historically, overestimates have been attributed to a failure to conjunction with X-ray diffraction (XRD) and field-emission

1820 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article

○ Figure 2. Sum of CO2 pore capacity and N2 pore capacity for (left) EF1 and right (EF6). Shown are CO2 cumulative pore capacity (blue ), CO2 ● △ ▲ pore size distribution (green ), CO2 +N2 cumulative pore capacity (blue ), and pore size distribution for CO2 +N2 (green ). fi scanning electron microscopy to validate the signi cance of to pores with diameters of less than 1.5 nm, while N2 allows organic matter for pore capacities, graphitic carbon has been access to pores with diameters between 1.5 and 30 nm. Close ff established as an e ective approximation to represent the examination of the pore volume data obtained from CO2 surface adsorbent for DFT methods. Using CO2 as an adsorption indicates that EF1 with a TOC of 2% has a adsorbate at 273 K with a Quantachrome Autosorb IQ2 cumulative pore volume of 0.008 cm3/g and cumulative surface instrument results in access to pores with sizes of 0.3 to 1.5 nm, area of 21 m2/g in the range of pore sizes from 0.3 to 1.5 nm, while using N2 and Ar at 77 and 87 K, respectively, results in while EF6 with a TOC of 5% has a cumulative pore volume of access to pores ranging from at least 1.5 nm to approximately 0.006 cm3/g and cumulative surface area of 16 m2/g in the 33 nm for N2 and 27 nm for Ar on a carbon sorbent. Detailed same pore size region. The N2 adsorption experiments revealed information regarding surface area measurements on nano- a cumulative pore volume of 0.02 cm3/g and surface area of 10 porous materials can be found in the thorough review by m2/g for EF1 and a cumulative pore volume of 0.03 cm3/g and 2 Thommes and Cychosz.21 surface area of 15 m /g for EF6 in the range of pore sizes from Two Eagle Ford samples were chosen for the current 3 to 30 nm. In the case of EF1, if CO2 had not not been used to investigation and are termed EF1 and EF6 throughout this complete the PSD analysis, nearly 28% of the nanopore volume Account. The TOCs of the shale samples were measured by and just over 50% of the active surface area would have been removing all of the inorganic carbon by acid digestion. neglected. In the case of EF6, approximately 15% of the Inorganic carbonates were removed by repeatedly adding 40 nanopore volume and 70% of the active surface area would mL of HCl (1 N) to the samples until no effervescence was have been neglected. Clearly, relying solely on N2 for observed under a microscope. The samples were placed in an determination of the nanopore volume would lead to a ° significant underestimation of the total pore volume, which may oven at 60 C for approximately 4 h between successive HCl fi treatments. Once no effervescence was detected, the samples signi cantly impact the estimates for GIP in addition to the were dried at 40 °C for at least 96 h. The dried samples were CO2 storage potential. processed on a Carlo Erba NA 1500 Series 2 elemental analyzer The pore size distribution data can be used in conjunction that combusts the samples at approximately 1020 °C with pure with grand canonical Monte Carlo (GCMC) simulation results to predict adsorption isotherms. Here the individual pore . The sample compositions were determined by ff collecting X-ray diffractograms on a Rigaku model CM2029 contributions are approximated by integration of the di erential powder X-ray diffractometer. Diffractograms were collected pore volume distribution curve over discrete regions, where the over the 2θ range from 5° to 70° using a Cu Kα X-ray source. limits of integration straddle the targeted pore size. For Distinct diffraction peaks were used to match given mineral example, the contribution for a 0.7 nm pore would be phases with peak identifications from the National Institute of 0.75 nm v0.7 = ∫ Vd()d x Standards and Technology (NIST) database. The EF1 sample 0.65 nm was obtained from a core that was extracted from a depth of 3864 m. The XRD analysis showed a mineral breakdown of The adsorption isotherm is predicted as a weighted sum of approximately 12% quartz, 86% calcite, and less than 5% illite individual pore densities: and a TOC of approximately 2%. The EF6 sample was qv= ρ extracted from a depth of 3890 m and has a mineral breakdown simulated ∑ i i of approximately 24% quartz, 70% calcite, and less than 5% illite i ρ and a TOC of just over 5%. where i represents the simulated pore density and the sum is Figure 1 shows isotherms (adsorption and taken over all pore sizes i, where i spans 0.35, 0.4, 0.5, 0.6, 0.7, desorption branches) and CO2 adsorption isotherms for 0.8, 0.9, 1.0, 1.2, 1.4, 1.8, 2.2, 2.8, 3.4, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, samples EF1 and EF6, while Figure 2 shows the combined 10.0, 12.0, 14.0, 16.0, 18.0, and 20.0 nm. The PSD was PSD using both CO2 and N2 as probe gases. CO2 allows access truncated at 20.0 nm, as it has been determined that excess

1821 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article

Figure 3. Surface chemistries investigated in the current study. Presented here are top views (surface layer only) of the periodic unit cell employed in each simulation. The two-letter abbreviations used in this study follow the surface designations. adsorption for larger pore sizes is considered negligible and can Table 1. Potential Parameters and Partial Charges Used in instead be approximated by the bulk gas density under the Force-Field Calculations conditions of interest. ε σ a site /kb (K) (Å) q (e) ref ■ IMPACT OF SURFACE CHEMISTRY ON C (CO2) 27.0 2.80 +0.70 31 − ADSORPTION CAPACITY O (CO2) 79.0 3.05 0.35 31 Fundamentally, the adsorption capacity is a measure of the CH4 148 3.73 N/A 32 relative intermolecular forces (and their magnitudes) present in C (surface) 28 3.4 +1.59 (CH) 33 +0.40, +0.40 (MH) sorbate−sorbent and sorbate−sorbate interactions. These +0.36 (OH) interactions can be roughly classified into two categories: O (surface) 79 3.1 −1.12, −1.11 (CH) 33 dispersion−repulsion interactions and electrostatic interactions, −1.08 (MH) which are connected to the surface chemistry through the −0.95 (OH) following relationships. Dispersion−repulsion forces are H (surface) 30 1.31 +0.60 (CH) 33 described by the Lennard-Jones (LJ) potential: +0.05, +0.05 (MH) ⎡ ⎤ ⎛ σσ⎞12⎛ ⎞ 6 +0.54 (OH) Vr()=− 4ε⎢⎜⎟ ⎜⎟⎥ a ⎣⎝ rr⎠ ⎝ ⎠ ⎦ Charge assignments for surface atoms were calculated via Bader charge analysis. Specific charges for oxygen and are reported, where ε denotes the depth of the potential well and σ while charges reported for carbon represent the most positive values. fi The majority of surface carbon atoms were found to have negligible represents the nite distance at which the interparticle potential Bader charge. reaches zero. For interactions between unlike particles, − 22 ε Lorentz Berthelot mixing rules are employed, i.e., 12 is ε ε σ represented as the geometric mean of 1 and 2 and 12 as the quadrupole contribution, a condition that forms the basis for σ σ arithmetic mean of 1and 2. The electrostatic interaction is a selective CO2 uptake. 23 ° more complicated function of several contributions, Comparison of CH4 and CO2 Adsorption at 60 and 250 C The adsorption of CH and CO were investigated at ϕϕϕϕelec. =++P μ Q 4 2 temperatures of 60 and 250 °C, which are common outgassing where the energy contributions on the right-hand side temperatures used for shale characterization and testing. correspond to (from left to right) polarization, field-dipole, Outgassing a shale sample at 60 °C only partially removes and field-gradient quadrupole. An additional contribution from some of the smaller hydrocarbons and free water within the sorbate−sorbate interactions on the surface may be included at pores, while outgassing at an intermediate temperature of 110 high coverage. °C removes most of the free water and further outgassing at a The surface chemistries considered within this Account temperature of 250 °C is sure to remove all of the free include the pure graphitic surface, a hydrated monovacancy, a adsorbates in the pores, including water and hydrocarbons.16,21 hydroxyl group, and a carboxyl group (Figure 3). These An example of this is shown in Figures S1 and S2 in the functionalities were chosen to be representative of the actual Supporting Information. Both Ar and CO2 were used as probe chemical composition of the organic matrix of gas shale based gases to investigate the pore size distribution at 60, 110, 200, on the known presence of water vapor within the organic and 250 °C. There is a clear trend associated with increasing matrix, knowledge of the water vapor−matrix chemical pore volume and outgassing temperature. The liquid-to-vapor interaction, and support from experimental investigations in transition for water at 60 °C is approximately 20 kPa (150 − the literature.9,24 29 As per the physical descriptions previously Torr); therefore, under a vacuum of 0.01 Torr, which is a − discussed, CO2 and CH4 will interact via dispersion repulsion typical outgas setting for the Autosorb iQ2 instrument, water forces according to the parameters outlined in Table 1. will desorb from the shale. There has been concern in the shale Furthermore, electrostatic interactions will occur between the community that outgassing at 250 °Cmayresultsin fi surface and CO2 as a result of the CO2 quadrupole moment modi cation of the kerogen and possibly the clay minerals of (−13.71 × 10−40 C·m2)30 and to a lesser extent between the the shale. Within the Eagle Ford samples investigated in this surface and CH4 via the polarization contribution (the Account, illite was the primary clay mineral present. Illite is a quadrupole moment is absent in CH4). Additionally, since clay composed of two tetrahedral layers, each of which is × the polarizabilities of CO2 and CH4 are similar (ca. 2.64 bonded to opposite sides of one inner octahedral layer, which 10−24 and 2.6 × 10−24 cm3, respectively),30 differences in classifies it as a 2:1 clay.34,35 In 1966 it was concluded that only sorbate uptake can be attributed to the field-gradient adsorbed water was removed from montmorillonite at 400

1822 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article

Figure 4. (left) Artistic sketch and (right) RockEval curve for the desorption of gas species within the pore matrix as a function of outgassing temperature. Free water desorption takes place between 60 and 110 °C, and the stored hydrocarbons residing in the smaller nanopores desorb at temperatures between 110 and 250 °C.

− Figure 5. Comparison of simulated excess methane adsorption isotherms (PG, CH, OH, and MH) with experimental loadings.41 44 The simulations were conducted at 60 °C.

°C.36 Further studies have shown that no structural changes are approximation for the maximum outgassing temperature is expected in the clays at 250 °C, as structural water in 2:1 clays recommended across all shale samples, as it lies in the vicinity is not extracted until approximately 400 °C.37 More recently, of the pregeneration trough for most samples. dehydration of illite interlayers was shown to take place up to Deciding what the appropriate outgassing temperature temperatures of 280−300 °C,38 and modeling attempts have should be depends on the intention of the investigation. For corroborated experiments showing dehydration between 180 to instance, here the goal is to understand how much CH4 the gas ° 39 ° 225 C. Furthermore, an outgassing temperature of 250 Cis shales may contain in addition to determining how much CO2 lower than the 500−550 °C temperature range at which normal may be reliably stored within the pores once they are depleted. illite clay minerals dehydroxylate, which involves the removal of If one wishes to understand how CO2 interacts with water and/ 38,40 structural hydroxyls. Examination of a standard RockEval or CH4 and other smaller hydrocarbons through injection of pyrolysis curve (Figure 4 right) shows that a minimum exists enhanced gas recovery, then lower outgassing temperatures between the S1 and S2 curves. The minimum, which is labeled may be preferred since this scenario requires depiction of the here as the “pregeneration trough”, would ideally be calculated natural system, not necessarily an empty pore matrix. This independently for each shale sample. Figure 4 shows the distinction is important to recognize in determining what the desorption of gas species within the pore matrix as a function of goals of one’s study may be. outgassing temperature, with the free water desorption taking Methane adsorption isotherms simulated at 60 °C over the place between 60 and 110 °C and the stored hydrocarbons pressure range from 0 to 50 bar are compared with residing in the smaller nanopores desorbing at temperatures experimental results in Figure 5. It is important to note that between 110 and 250 °C. Hence, using 250 °Casan comparison of results from different sampleseven from

1823 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article

° Figure 6. Excess CO2 adsorption simulated at 60 C. The values are 1.9, 2.9, 1.8, and 2.2 times greater than the CH4 adsorption (Figure 5) for PG, CH, MH, and OH, respectively.

° Figure 7. Comparison of simulated CO2 and CH4 excess adsorption at 250 C. adsorption (solid circles) is shown to be 1.4, 4.3, 1.4, and 1.8 times greater than CH4 adsorption (open triangles) for PG, CH, MH, and OH, respectively. within the same regimeis difficult because of sample experimental data in Figure 5 from Rexer et al.43 on an Eagle heterogeneity and discrepancies in sample preparation, testing, Ford sample taken from the dry-gas window (8.2% quartz, and presentation. Also, the models presented in this Account 49.9% calcite, 19.5% illite, and 7.15% TOC) compare quite well neglect the clay composition for simplicity in addition to to the simulation data of the current work. Although this isolating the chemical effects of surface functionality on sample has a greater illite content than the Eagle Ford samples 42 adsorption. Recent work by Ji et al. (2012) demonstrates on which the models are based, the CH4 capacity is similar, that illite compared with other clays present in shale has the further supporting the small role that clay may play in lowest methane capacity. illite showed to have the lowest adsorption. Al Ismail41 reported results for a Permian sample of Langmuir adsorption constant and lowest BET surface area of similar composition and depth as the Eagle Ford samples the clay materials investigated, with the latter property showing examined in the current study. Furthermore, the cumulative a linear correlation to adsorption capacity. Methane adsorption pore volume of these Eagle Ford samples is smaller than that results of Ji et al.42 at 50 °C over a sample composed of 99% observed for the Permian sample (0.031 and 0.051 cm3/g, illite is included in Figure 5 for comparison. Rexer et al.44 tested respectively). an Alum Shale sample (44.4% quartz, 0.5% calcite, 5.9% illite, With the exception of the hydroxylated surface, which shows ° fi and 6.35% TOC) at 65 C, which exhibited greater CH4 an increase in CH4 adsorption over the unmodi ed surface at capacity compared with the models presented here. Additional high pressures (+16%), the functionalized surfaces generally

1824 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article

° ° Figure 8. CO2/CH4 selectivities measured at (left) 60 C and (right) 250 C. The measurement uncertainty is smaller than the height of the symbols. exhibited lower CH4 adsorption. This is intuitive from the CO2/g of modeled kerogen at 50 bar across all of the surface presentation of the sorbate−sorbent interactions discussed functionalities considered. previously together with a basic chemical understanding of polarity, as CH is a nonpolar . In particular, CH ■ IMPACT OF SURFACE CHEMISTRY ON GAS 4 4 adsorption at 10 bar is on average 0.03 mmol of CH4/g of SELECTIVITY modeled kerogen, while at 50 bar the adsorption increases to a The carbon dioxide to methane selectivity was measured as a − range of 0.05 0.08 mmol of CH4/g of modeled kerogen across function of pressure at 60 and 250 °C, and the results are all of the carbon pore surfaces considered. presented in Figure 8. Generally, the hydrated vacancy and fi Conversely, polar CO2 exhibits enhanced loading for surfaces unmodi ed surface demonstrated CO2/CH4 selectivities that modified with carboxyl (+4%) and hydroxyl (+33%) function- remained largely invariant to pressure. This behavior is typical fi alities (Figure 6). More speci cally, at 10 bar the CO2 of Langmuir adsorption observed on energetically homoge- adsorption ranges from 0.06 to 0.13 mmol of CO2/g of neous surfaces, which are capable of maintaining linear modeled kerogen and increases to a range of 0.11−0.18 mmol selectivity over a wide range. Alternatively, the hydroxylated of CO2/g of modeled kerogen at 50 bar for all of the carbon and carboxylated surfaces show a strong preference for CO2 at low and moderate pressures. This effect is minimized at higher pore surfaces considered. In addition, the CO2/CH4 exchange ratio was calculated to be 1.9 for the unmodified surface, 2.9 for pressures as pores within the microporous region become the carboxylated surface, 1.8 for the hydrated monovacancy, saturated, as it has been demonstrated that the extent of surface chemistry on enhanced CO2 uptake and selectivity converges and 2.2 for the hydroxylated surface. These results fall within fi the range of exchange ratios typically reported for these with the unmodi ed surface as the pore size enters the early mesoporous range (d > 2.0 nm).45 In spite of reduced materials.10,11 selectivity at high pressures, the carboxylated and hydroxylated Simulations were also performed at 250 °C, and the results surfaces demonstrate a strong preference for CO2 over the are shown in Figure 7. To our knowledge, no experimental data entire pressure range, suggesting that kerogen with high were available for direct comparison with the simulation results carboxyl and/or hydroxyl content should have a greater storage at low or high temperature. With the exception of the capacity for CO2. carboxylated surface, CO2 adsorption decreased to a larger extent than CH4 adsorption relative to results obtained at 60 ■ CONCLUSIONS °C; hence, the CO /CH exchange ratio is lower for these 2 4 This Account has examined two Eagle Ford shale samples, both surfaces at 250 °C (1.4, 1.4, and 1.8 for PG, MH, and OH, recovered from shale that was extracted at a depth of respectively). Conversely, CH4 adsorption decreased to a larger ° approximately 3800 m and having similar mineral compositions extent than CO2 adsorption relative to the 60 C adsorption but distinct TOCs (i.e., 2% and 5%). Given the complexity of results for the carboxylated surface, resulting in a larger CO2/ fi the nanopore structure and chemistry of gas shales, which are CH4 exchange ratio of 4.3. More speci cally, the CH4 composed primarily of interconnected clay and kerogen adsorption at 10 bar is approximately 0.005 mmol of CH4/g networks, water likely plays a role in the surface chemistry of of modeled kerogen and extends over a range of 0.01−0.02 the pores. The Eagle Ford samples considered in this work mmol of CH4/g of modeled kerogen across all of the surface were low in clay and deliberately chosen so that the kerogen functionalities considered. On the other hand, the CO2 within these systems could be explicitly modeled. On the basis adsorption ranges from 0.005 to 0.03 mmol of CO2/g of of the existence of adsorbed water and dissociated water on the modeled kerogen at 10 bar and from 0.02 to 0.05 mmol of kerogen surfaces of these natural systems, a series of surface

1825 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article functionalities were considered to determine what role, if any, times more gas stored in the pores than is realized. Similarly, surface chemistry plays in terms of CH4 and CO2 adsorption. using an outgassing temperature that is too low in estimating Adsorption was modeled up to 50 bar to be more the CO2 capacity would lead to an underestimation of what the representative of the subsurface conditions of shale. Temper- depleted shale reservoir may reliably store. ature conditions of 60 and 250 °C were considered in order to To What Extent Is Gas Adsorption Underestimated When show the dependence of the adsorption on the chosen Pore Volumes Less than 1.5 nm Are Neglected? outgassing temperature. Finally, to determine the nanopore In the cases of the EF1 and EF6 samples with 2% and 5% TOC, volume and complete PSD, both CO and N were used as 2 2 respectively, the total pore volumes (i.e., for pores less than 30 probe gases to cover the range of pore sizes from 0.3 to 30 nm. nm in diameter) are 0.028 and 0.036 cm3/g, of which 28% and Following all of these considerations, one can then determine 15%, respectively, are represented by pores that are less than the impact each has on the methane and CO storage capacity 2 1.5 nm in diameter. These are pores that go unmeasured of the Eagle Ford Shale on the basis of these particular shale without the use of CO as a probe gas. Therefore, relying solely samples. The questions that may now be answered concern the 2 on N for determining the pore volume for pores less than 30 impact of knowledge of the kerogen chemistry on gas capacity 2 nm in diameter would lead to a significant underestimation of and the extent to which gas adsorption is underestimated when the total pore volume, which may significantly impact the (1) the applied outgassing temperature is too low and (2) pore estimates of gas in place as well as CO storage potential. volumes less than 1.5 nm are neglected. 2 What Impact Does Knowledge of the Kerogen Chemistry ■ ASSOCIATED CONTENT Have on Gas Capacity? *S Supporting Information ° At temperatures of 60 and 250 C, the CH4 adsorption at 10 The Supporting Information is available free of charge on the bar was on average 0.03 and 0.005 mmol/g, respectively, for all ACS Publications website at DOI: 10.1021/acs.ac- of the surface functionalities considered. These were the only counts.7b00003. conditions under which the surface chemistry of the pores did not impact the extent of adsorption. At 50 bar, however, at Pore size distributions and corresponding cumulative ° pore volume increases as functions of outgassing temperatures of 60 and 250 C the CH4 adsorption ranged from 0.11 to 0.18 mmol/g and 0.01 to 0.02 mmol/g, temperature using and CO2 as probe gases representing increases of 60% and 100%, respectively. (PDF) In the case of CO2, which tends to be more reactive with the pore surfaces, as expected, the surface chemistry played an even ■ AUTHOR INFORMATION more dominant role. For instance, at 10 bar and temperatures Corresponding Author ° of 60 and 250 C, the CO2 adsorption ranged from 0.06 to 0.13 *E-mail: [email protected]. mmol/g and 0.005 to 0.03 mmol/g, representing 6-fold and just ORCID over 2-fold increases, respectively. These conditions have the ff Randall Holmes: 0000-0002-5047-5528 most pronounced e ect on CO2 adsorption in the simulated kerogen pores considered in this Account. At 50 bar, the Jennifer Wilcox: 0000-0001-8241-727X surface chemistry continues to be an important consideration, Notes ° i.e., at temperatures of 60 and 250 C, the CO2 adsorption fi ranges from 0.11 to 0.18 mmol/g and 0.02 to 0.05 mmol/g, The authors declare no competing nancial interest. representing 1.6-fold and 2.5-fold increases, respectively. Biographies To What Extent Is Gas Adsorption Underestimated When Peter Psarras received his B.A. in Chemistry from Miami University in the Applied Outgassing Temperature Is Too Low? 2005 and his Ph.D. in Chemistry from Cleveland State University in In the case of determining the true pore volume available to 2014. He worked as a postdoctoral associate in the Department of store either hydrocarbons or ultimately CO2, one needs to carry Energy Resources Engineering at Stanford University from 2014 to out adsorption experiments on samples with “empty” pores. 2016 and is currently a postdoctoral associate in Chemical and This includes those experiments such as low-pressure CO2 and Biological Engineering at the Colorado School of Mines. N2 adsorption experiments that provide the PSD and pore Randall Holmes received his B.S. in Civil and Environmental volume data that are used for the model development to EngineeringAtmosphere and Energy Track from Stanford Uni- simulate adsorption capacities. Using an outgassing temper- versity in 2015 and his M.S. in Earth System Science from Stanford ature that is too low will result in pores that contain residual gas University in 2017. He is currently working toward a Ph.D. in the and possibly water, which will lead to an underestimation of Emmett Interdisciplinary Program in Environment and Resources at what the pores can hold assuming that the pores are depleted of Stanford University. hydrocarbons in the case of the reliable storage of CO2. The methane adsorption simulation data from temperature Vikram Vishal received his B.S. from Presidency College in 2007, his conditions of 60 and 250 °C from this study can be used to M.S. from IIT Bombay in 2009, and his Ph.D. from IIT Bombay and estimate the residual gas that exists in the pores at the lower Monash University in 2013. He worked as a postdoctoral associate in outgassing temperature. For instance, comparing the methane the Department of Energy Resources Engineering at Stanford adsorptions at 60 and 250 °C, there remains an additional University in 2015 and is currently an Assistant Professor at IIT Bombay. 0.025 mmol of CH4/g of modeled kerogen at the outgassing temperature of 60 °C. Similarly, at 50 bar, depending upon the Jennifer Wilcox received her B.A. in Mathematics from Wellesley surface chemistry, there may remain an additional 0.04 to 0.06 College in 1998, her M.A. in Physical Chemistry from the University ° mmol of CH4/g of modeled kerogen left in the pores at 60 C. of Arizona in 2004, and her Ph.D. in Chemical Engineering from the This implies that at both pressures there may actually be 2 University of Arizona in 2004. She served as an Assistant Professor at

1826 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article

Worcester Polytechnic Institute from 2004 to 2008 and an Assistant gases, with special reference to the evaluation of surface area and pore Professor at Stanford University from 2008 to 2016. She is currently size distribution (IUPAC Technical Report). Pure Appl. Chem. 2015, an Associate Professor in Chemical and Biological Engineering at the 87, 1051−1069. Colorado School of Mines. (17) Chalmers, G. R.; Bustin, R. M.; Power, I. M. Characterization of gas shale pore systems by porosimetry, pycnometry, surface area, and field emission scanning electron microscopy/transmission electron ■ ACKNOWLEDGMENTS microscopy image analyses: Examples from the Barnett, Woodford, The authors thank Dr. Doug McCarty of Chevron and Dr. Haynesville, Marcellus, and Doig units. AAPG Bull. 2012, 96, 1099− Matthias Thommes and Dr. Katie Cychosz of Quantachrome 1119. Instruments for their helpful discussions throughout this study. (18) Ross, D. J. K.; Bustin, R. M. The importance of shale The authors additionally acknowledge Prof. Mark Zoback and composition and pore structure upon gas storage potential of shale gas − ConocoPhillips for supplying the two Eagle Ford samples reservoirs. Mar. Pet. Geol. 2009, 26, 916 927. investigated in this Account. (19) Loucks, R. G.; Reed, R. M.; Ruppel, S. C.; Jarvie, D. M. Morphology, genesis, and distribution of nanometer-scale pores in ■ REFERENCES siliceous of the Mississippian Barnett Shale. J. Sediment. Res. 2009, 79, 848−861. (1) Hagin, P. N.; Zoback, M. D. Laboratory Studies of the (20) Kuila, U.; McCarty, D. K.; Derkowski, A.; Fischer, T. B.; Topor,́ Compressibility and Permeability of Low-Rank Coal Samples from T.; Prasad, M. Nano-scale texture and porosity of organic matter and the Powder River Basin, Wyoming, USA. In Proceedings of the 44th US clay minerals in organic-rich . Fuel 2014, 135, 359−373. Rock Mechanics Symposium and 5th US−Canada Rock Mechanics (21) Thommes, M.; Cychosz, K. A. Physical adsorption character- Symposium 2010; American Rock Mechanics Association: Alexandria, ization of nanoporous materials: progress and challenges. Adsorption VA, 2010. 2014, 20, 233−250. (2) Vishal, V.; Ranjith, P. G.; Pradhan, S. P.; Singh, T. N. (22) Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids; Permeability of sub-critical carbon dioxide in naturally fractured Oxford University Press: New York, 1989. Indian bituminous coal at a range of down-hole stress conditions. Eng. − (23) Wilcox, J. Carbon Capture; Springer: New York, 2012. Geol. 2013, 167, 148 156. (24) Kostov, M. K.; Santiso, E. E.; George, A. M.; Gubbins, K. E.; (3) Jarvie, D. M.; Hill, R. J.; Ruble, T. E.; Pollastro, R. M. Nardelli, M. B. Dissociation of water on defective carbon substrates. Unconventional shale-gas systems: The Mississippian Barnett Shale of Phys. Rev. Lett. 2005, 95, 136105. north-central Texas as one model for thermogenic shale-gas (25) Bagri, A.; Grantab, R.; Medhekar, N. V.; Shenoy, V. B. Stability assessment. AAPG Bull. 2007, 91, 475−499. (4) Wang, F. P.; , J. F. W. Screening criteria for shale-gas and formation mechanisms of carbonyl-and hydroxyl-decorated holes − in graphene oxide. J. Phys. Chem. C 2010, 114, 12053−12061. systems. Gulf Coast Assoc. Geol. Soc., Trans. 2009, 59, 779 793. ’ (5) Eagle Ford Region Drilling Productivity Report; U.S. Energy (26) Kudin, K. N.; Ozbas, B.; Schniepp, H. C.; Prud Homme, R. K.; Information Administration: Washington, DC, 2016. Aksay, I. A.; Car, R. Raman spectra of oxide and − (6) Fredrich, J. T.; Menendez, B.; Wong, T. F. Imaging the pore functionalized graphene sheets. Nano Lett. 2008, 8,36 41. structure of geomaterials. Science 1995, 268, 276. (27) Liu, Y.; Wilcox, J. Molecular simulation of CO2 adsorption in (7) Travis, K. P.; Todd, B. D.; Evans, D. J. Departure from Navier- micro- and mesoporous with surface heterogeneity. Int. J. Coal Stokes hydrodynamics in confined liquids. Phys. Rev. E: Stat. Phys., Geol. 2012, 104,83−95. Plasmas, Fluids, Relat. Interdiscip. Top. 1997, 55, 4288−4295. (28) Firouzi, M.; Rupp, E. C.; Liu, C. W.; Wilcox, J. Molecular (8) Mullen, J. Lessons Learned Developing the Eagle Ford Shale. In simulation and experimental characterization of the nanoporous Proceedings of the Tight Gas Completions Conference, 2−3 November, San structures of coal and gas shale. Int. J. Coal Geol. 2014, 121, 123−128. Antonio, Texas, USA;Society of Engineers: Richardson, TX, (29) Liu, Y.; Wilcox, J. Effects of surface heterogeneity on the 2010. adsorption of CO2 in microporous carbons. Environ. Sci. Technol. 2012, (9) Liu, Y.; Wilcox, J. CO2 Adsorption on Carbon Models of Organic 46, 1940−1947. Constituents of Gas Shale and Coal. Environ. Sci. Technol. 2011, 45, (30) Ruthven, D. M. Encyclopedia of Separation Technology; Wiley: 809−814. New York, 1997. (10) Heller, R.; Zoback, M. Adsorption of methane and carbon (31) Potoff, J. J.; Siepmann, J. I. Vapor−liquid equilibria of mixtures dioxide on gas shale and pure mineral samples. J. Unconv. Oil Gas containing alkanes, carbon dioxide, and nitrogen. AIChE J. 2001, 47, − Resour. 2014, 8,14 24. 1676−1682. (11) Hendriks, C.; Graus, W.; van Bergen, F. Global Carbon Dioxide (32) Martin, M. G.; Siepmann, J. I. Transferable potentials for phase Storage Potential and Costs; Ecofys: Utrecht, The Netherlands, 2004. equilibria. 1. United-atom description of n-alkanes. J. Phys. Chem. B (12) Espinoza, D. N.; Kim, S. H.; Santamarina, J. C. CO2 geological −  − 1998, 102, 2569 2577. storage geotechnical implications. KSCE J. Civ. Eng. 2011, 15, 707 (33) Tenney, C. M.; Lastoskie, C. M. Molecular simulation of carbon 719. dioxide adsorption in chemically and structurally heterogeneous (13) Dutta, P.; Zoback, M. D. CO sequestration into the Wyodak 2 porous carbons. Environ. Prog. 2006, 25, 343−354. coal seam of Powder River BasinPreliminary reservoir character- (34) Geatches, D.; McCarty, D.; Wilcox, J. Ab initio investigations of ization and simulation. Int. J. Greenhouse Gas Control 2012, 9, 103− dioctahedral interlayer-deficient mica: Modeling particles of illite 116. − (14) Verma, M. K. Calculation of Hydrocarbon-in-Place in Gas and found within gas shale. Am. Mineral. 2014, 99, 1962 1972. Gas-Condensate ReservoirsCarbon Dioxide Sequestration; United (35)Geatches,D.L.;Wilcox,J.Abinitioinvestigationsof States Geological Survey: Reston, VA, 2012; https://pubs.usgs.gov/ dioctahedral interlayer-deficient mica: modelling 1 M polymorphs of − of/2012/1033/OF2012-1033.pdf (accessed July 13, 2017). illite found within gas shale. Eur. J. Mineral. 2014, 26, 127 144. (15) Ambrose, R. J.; Hartman, R. C.; Diaz Campos, M.; Akkutlu, I. (36) van Olphen, H. Collapse of Potassium Montmorillonite Clays −“ ” Y.; Sondergeld, C. New Pore-Scale Considerations for Shale Gas in Upon Heating Potassium Fixation .InProceedings of the Fourteenth Place Calculations. In Proceedings of the SPE Unconventional Gas National Conference on Clays and Clay Minerals; Pergamon Press: Conference, 23−25 February, Pittsburgh, Pennsylvania, USA; Society of London, 1966; pp 393−405. Petroleum Engineers: Richardson, TX, 2010. (37) Roch, G. E.; Smith, M. E.; Drachman, S. R. Solid state NMR (16) Thommes, M.; Kaneko, K.; Neimark, A. V.; Olivier, J. P.; characterization of the thermal transformation of an illite-rich clay. Rodriguez-Reinoso, F.; Rouquerol, J.; Sing, K. S. W. Physisorption of Clays Clay Miner. 1998, 46, 694−704.

1827 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828 Accounts of Chemical Research Article

(38) Drits, V. A.; McCarty, D. K. The nature of structure-bonded H2O in illite and leucophyllite from dehydration and dehydroxylation experiments. Clays Clay Miner. 2007, 55,45−58. (39) Vidal, O.; Dubacq, B. Thermodynamic modelling of clay dehydration, stability and compositional evolution with temperature, − pressure and H2O activity. Geochim. Cosmochim. Acta 2009, 73, 6544 6564. (40) Drits, V. A.; Derkowski, A.; McCarty, D. K. Kinetics of partial dehydroxylation in dioctahedral 2:1 layer clay minerals. Am. Mineral. 2012, 97, 930−950. (41) Al Ismail, M. I. Influence of nanopores on the transport of gas and gas-condensate in unconventional resources. Ph.D. Dissertation, Stanford University, Stanford, CA, 2016; https://pangea.stanford.edu/ ERE/pdf/pereports/PhD/Al-Ismail2016.pdf (accessed July 13, 2017). (42) Ji, L.; Zhang, T.; Milliken, K. L.; Qu, J.; Zhang, X. Experimental investigation of main controls to methane adsorption in clay-rich rocks. Appl. Geochem. 2012, 27, 2533−2545. (43) Rexer, T. F. T.; Mathia, E. J.; Aplin, A. C.; Thomas, K. M. High- pressure methane adsorption and characterization of pores in Posidonia shales and isolated kerogens. Energy Fuels 2014, 28, 2886−2901. (44) Rexer, T. F.; Benham, M. J.; Aplin, A. C.; Thomas, K. M. Methane adsorption on shale under simulated geological temperature and pressure conditions. Energy Fuels 2013, 27, 3099−3109. (45) Psarras, P.; He, J.; Wilcox, J. Molecular simulations of nitrogen- doped hierarchical carbon adsorbents for post-combustion CO2 capture. Phys. Chem. Chem. Phys. 2016, 18, 28747−28758.

1828 DOI: 10.1021/acs.accounts.7b00003 Acc. Chem. Res. 2017, 50, 1818−1828