Gas hydrates are a type of ice-like clathrate compounds that existed in nature, which have attracted considerable attention because of their unique properties [1–5] and various applications. For example, artificial gas hydrates are considered one of the most promising materials for gas storage [6–8] and separation [9–11]. The formation of gas hydrates in hydrophobic pores has been widely reported as an effective method to enhance gas hydrate formation kinetics and capacity [12–18]. It has been reported that the stoichiometry of gas to water in artificial gas hydrates formed on porous carbon is much higher than that calculated from the crystal structure [9, 19]. An acceptable hypothesis is that the enhanced gas in artificial gas hydrates is attributed to additional gas adsorption in gaps between formed gas hydrates and pore surface or semiclathrate. However, important questions about what kind of gaps and how they can be formed are still not clarified, because there is no valid scientific method to distinguish this type of gaps and semiclathrate in the pore, and quantitative analysis of the gas in gas hydrates or adsorptive phase is hardly carried out. Some techniques such as molecular simulations [11, 20], in situ Raman spectroscopy [21], solid-state nuclear magnetic resonance [10, 22], neutron diffraction [23, 24], and synchrotron X-ray powder diffraction [25] could only be employed to confirm that the formation of gas hydrates is dependent on porous materials. There is no doubt that gas hydrates could be formed on porous materials, but the details about the formation process and whether they are formed inside the pores or not are still ambiguous.

A common understanding is that porous materials could be employed to enhance the gas hydrate formation due to the nanoconfinement effect. After the clusters of gas hydrates form in the nanoconfined space, blank space appears to allow for the gas occurrence as shown in Figure 1. Due to lack of convincing characterization techniques, however, the formation of gas hydrates in the confined space has not been experimentally proven yet. In comparison with this classic mechanism, we discovered a different process (I-IV in Figure 1) for the gas hydrate formation based on the investigation of CO₂ hydrate formation on porous carbon. In this case, when CO₂ was introduced into porous carbon fully filled with water, the considerable sorption of CO₂ was observed on account of a consequence of interesting processes including (I) the replacement of water adsorbed on the carbon surface by CO₂ and (II-IV) further replacement of water in the pore by CO₂ accompanied with the growth of gas hydrate crystals out of the pore.

In process (I), although the calculated binding energy of C-CO₂ is slightly weaker than that of C-H₂O, a portion of water directly adsorbed on the carbon surface could still be replaced by CO₂ driven by the entropy. During the process (II), the water inside the pore is further replaced by CO₂ upon increasing CO₂ chemical potential. Finally, gas hydrates are formed outside the confined space, and CO₂ is condensed in the pore by route III. The gas hydrates could also be formed by following another route (IV). When the pressure reaches the value for the CO₂ hydrate formation, microhydrate crystals initially form in the nanoconfined space and then move out to grow into large crystals in order to decrease the surface free energy. Meanwhile, the CO₂ adsorption and condensation occur in the pore. Therefore, it was deduced that the gas hydrates could not be formed in the nanoconfined space. The excess gas observed in artificial gas hydrates resulted from the gas adsorption and condensation in the nanoconfined space when water is driven to form gas hydrates outside the pore. To confirm this assumption, we designed a rigorous experiment to investigate the gas hydrate formation behavior in the nanoconfined space and achieved a novel understanding about the formation behavior of gas hydrates on porous carbon materials. The pathway is totally different with previous hypothesis of the gas hydrate formation inside nanopores. Furthermore, a great enhancement of the selectivity and regeneration for the gas separation on porous materials was achieved based on the mechanism we proposed.

We selected CO₂ as a representative gas for the study. The adsorptive CO₂ could be condensed in confined space with a constant density easy for quantitative analysis. Thus, the CO₂ hydrate formation on classic ordered mesoporous carbon CMK-3 and irregular porous activated carbon (AC) was investigated by thermodynamic experiments, thermodynamic calculations, kinetic analysis, density functional theory (DFT) calculation, and in situ Raman spectroscopy. The porous properties calculated from N₂ isotherm curves at 77 K (Figures 2(a) and 2(b)) including Brumauer–Emmett–Teller (BET) surface area, DFT pore size, pore volume, and water vapor uptake (Figures S1 and S2) are summarized in Table S1. The pore volumes of CMK-3 and AC are 1.58 and 1.0 cm³/g, respectively. CMK-3 presents an ordered mesoporous structure as shown in Figure 2(c). The scanning electron microscopy (SEM) image of AC indicates a layered structure (Figure 2(d)).

When comparing CO₂ isotherms determined on CMK-3 at 273 K with different mass ratios of water to carbon (Rw), obvious differences were observed as shown in Figure 2(e). The CO₂ isotherm on dry CMK-3 presents a type IV isotherm classified by IUPAC, because of the condensation of CO₂ in the mesopore (3.8 nm). When Rw = 1 (i.e., the pore is full of water), the CO₂ uptake is obviously lower than that on the dry sample under the pressure < 15 atm, since adsorptive sites are occupied by water molecule. As compared with the CO₂ sorption curve in water (Figure S3), the slight CO₂ uptake is mainly attributed to the CO₂ adsorption on the carbon surface. Obviously, the carbon material with different water ratios presents similar CO₂ uptake before 15 atm, and the slight difference should be due to the dissolved CO₂. It was interestingly found that a small step occurred under 15.1 atm and a large step happened under pressure > 25 atm. The first step around 14-15 atm observed on all wet CMK-3 should be assigned to the gas hydrate formation. The uptake at the first step increases dependent on the Rw value obviously. When Rw > 1, water would overflow the pore volume and occupy the gap of the particles to form the gas hydrates under relatively lower pressure comparing with the formation in the confined space. Based on the equation dG=-SdT+VdP+\lambda dA, the smaller the gas hydrate crystal form, the higher the excess surface Gibbs free energy. According to the equation (\partial G/\partial p{)}_T=V, a higher formation pressure is needed when small crystal forms. Similar gas hydrate formation pressure on the porous material to that in bulky water system implies that the gas hydrates are formed out of the pore.

The second step that occurred at relatively higher pressure should be attributed to the gas hydrate formation. However, it was noted that the second step had a similar uptake increase no matter what the Rw value was. The molar ratio of water to CO₂ defined as x was listed in Table 1. It is obvious that the x value (7.67) is higher than the theoretical one (5.75) when water is completely converted into the CO₂ hydrate. The reason is that 5¹² small cages are preferably occupied by water molecule, and only 6 large cages are retained for CO₂ in a unit cell. In past, this excess uptake was attributed to (1) excess adsorption in the gap between the gas hydrate and carbon surface or (2) the formation of incomplete cages. In the present case, we propose more feasible mechanism that the excess uptake comes from the CO₂ condensation in the mesopore of CMK-3, where water is replaced by CO₂. This mechanism is also supported by other experimental results. When Rw = 2.01 (Figure 2(e)), water outside the pore is firstly converted to the CO₂ hydrate under relatively lower pressure of 14 atm at 273 K. With pressurizing CO₂, fragments of water clusters start to form in the pore in order to reduce the surface Gibbs free energy. These fragments and microcrystals tend to combine and grow up out of the pore. However, no sharp increase of uptake could be found under this condition, indicating that the migration of gas hydrate fragments cannot occur independently, unless accompanied with the CO₂ condensation in the mesopore. Cooperating with the CO₂ condensation, the water cluster migrates from inside pore to outside for further growth of the CO₂ hydrate crystals. Meanwhile, the pore gradually becomes empty and allows for the CO₂ condensation to occur. By comparing the CO₂ uptake under the two steps and the CO₂ adsorption amount on dry carbon (Table 1), it is clear that the mechanism we proposed here is reasonable. When subtracting the adsorptive amount of CO₂ from the total uptake, the gas hydrates obtained on all wet CMK-3 show reasonable x value to the theoretical one. The result could also be obtained from the desorption branch of isotherm curves. As shown in Figure S4, the desorption branch of CO₂ on CMK-3 (Rw = 0.96) shows an obvious hysteresis loop, which is much different from the desorption branch of CO₂ on dry CMK-3. To make it clearer, the estimated gas hydrate amount formed on CMK-3 (6.83 mmol/g) was deducted from the desorption branch. It was found that, when the pressure is >23.7 atm, the desorption branch meets well with that on dry CMK-3, indicating a similar desorption mechanism. When the pressure is decreased, the CO₂ uptake on wet CMK-3 decreases more quickly than that on dry sample. During this step, the gas hydrate decomposes into free water molecule to reoccupy the pore of CMK-3 and accelerate the CO₂ desorption. When calculating the CO₂ uptake by desorption on dry carbon between 12.1-23.7 atm, the released gas amount by gas hydrate decomposition is about 6.91, which is consistent with the value estimated from the adsorption branch.

When similar experiments were carried out on AC with irregular slit pore size around 2 nm, clearly different isotherm curves were obtained. The CO₂ isotherm (Figure 2(f)) on dry AC showed type I curve because of its small pore size (2 nm). Different from the isotherm on wet CMK-3, the CO₂ isotherm on wet AC showed only one inflexion. When the water loading amount is equal to the pore volume, the adsorptive amount exhibits a smooth increase around 15 atm at 273 K, which is very different from the sharp increase at 14 atm observed on wet CMK-3. It is notable that the uptake on wet AC with Rw = 1.58 is similar to that on its dry sample. It was deduced that the smooth increase is attributed to the replacement of water by CO₂. When Rw = 1.58, the uptake reaches the maximum value on dry AC, indicating that no water is converted to gas hydrates. The same result was observed on AC with Rw = 2.28. In this case, although the water loading obviously exceeded the pore volume of AC, no gas hydrate formation was observed around 14 atm, which was different from the results on CMK-3. The water out of the AC pore cannot form gas hydrates even after 48 h. The reason may be attributed to special porous structure of AC. The pore in AC has a V shape. The water replaced by CO₂ could be located on the top of the V pore, where the pressure for the gas hydrate formation is too high to reach when the CO₂ condensation takes place at the bottom of the pore.

A small hysteresis loop was observed in the desorption branch (Rw = 2.28) (Figure S5), which is very different from the CO₂ desorption branch on dry AC, indicating a different desorption mechanism. In this case, the adsorbed CO₂ molecule is easier to depart than that on dry carbon. At 1 atm, the CO₂ desorption could achieve more than 99%. When the pressure decreases, the adsorbed CO₂ dissociates from the surface and is replaced by water molecule. Under the replacement effect of water, the CO₂ dispersion occurs more easily than that on dry AC, attributing to both enthalpy and entropy. On one hand, rebinding of water to carbon surface compensates the enthalpy penalty when CO₂ disperses from the surface. One the other hand, the increase of translation entropy caused by CO₂ desorption is higher than the reduction of translation entropy caused by H₂O adsorption, driving the process proceeding forward. When Rw increases to 3.1, partial water out of the V pore could form gas hydrates at relatively lower pressure, driving water in the pore to participate in the CO₂ hydrate formation. The isotherm curve (Rw = 3.1) shows that the sharp increase which occurred at about 15 atm corresponds to the CO₂ hydrate formation. By calculating molar ratio of water to gas, the value of x = 7.65 is well consistent with the theoretical one (7.67), indicating that water on AC is completely converted to the gas hydrate under this condition. It was also noticed that, after subtracting the uptake caused by the gas hydrate formation, the isotherm curve (Rw = 3.1) is similar to others (Rw = 1.58 and 2.28), suggesting that the mechanism of CO₂ adsorption on AC with pressure > 15 atm under different Rw is similar. The desorption branch (Rw = 3.1) shows a small hysteresis loop (Figure S5). The sharp decrease of CO₂ uptake occurs at about 13.9 atm corresponding to the hydrate crystal dissociation and CO₂ desorption from the nanopore accompanied with the reoccupation of water molecule in the nanopore. Similar to the case of CMK-3, the CO₂ dispersion process is promoted by the replacement of water molecule after the gas hydrate decomposition.

In order to distinguish the mechanisms corresponding to the gas uptake obtained from thermodynamically static studies in each step, kinetic adsorptive experiments were carried out. Both of the gas adsorption and hydrate formation are exothermic processes. The gas hydrate formation process is much slower than the adsorption process, because the former includes the inducing period for the nucleation and long period for the crystal growth. Therefore, the inducing period and slow growth rate with heat release are unique features in the hydrate formation. Kinetic curves with temperature changes are shown in Figures 2(g) and 2(h). From the kinetic curve, it was found that temperature peaks which occurred on both materials are accompanied with inflection points of uptake. As compared with blank experiment (Figure S6), when loading CO₂ on dry CMK-3, only one temperature peak around 18°C was observed, attributing to the release of adsorptive heat, gas expanding heat, and heat transferred from the environment. In the case of Rw = 0.96 and 2.01, the first peak of temperature decreases to about 9°C because of high capacity of water. On wet CMK-3, the second inflection point occurs after about 20 min, and slow decline should be attributed to the phase transferring heat.

A similar trend of temperature changes was observed on AC as well. When loading CO₂ on dry AC, the temperature peak reaches about 45°C, which is much higher than that on CMK-3 due to higher adsorptive amount and stronger potential intensity of the pore in AC. When loading CO₂ on wet AC, the first temperature peak decreases to 20°C, because of high heat capacity of water. When Rw = 1.58, only one temperature peak corresponding to a sharp uptake increase without inducing time was observed. This is strong evidence for the conclusion that CO₂ hydrates cannot be formed in this case. When Rw = 3.1, the peak of temperature around 8°C occurs with relatively longer platform, indicating that the process is accompanied with a slow exothermic process assigned to the gas hydrate formation.

In order to obtain the accurate value of the heat, the isotherm curves (Figures S7-S12) at different temperatures were collected for phase changing enthalpy analysis. According to the Clausius-Clapeyron equation, adsorptive enthalpy on CMK-3 with different Rw was obtained (see Supporting Information for more details). The enthalpy of the first step assigned to the hydrate formation out of pore is 64.7 kJ/mol. The enthalpy of the second step is 31.0 kJ/mol, which is lower than the enthalpy (64.7 kJ/mol) of the CO₂ hydrate formation, but higher than CO₂ condensation enthalpy (19.8 kJ/mol) derived from the CO₂ adsorption on dry CMK-3. Since the enthalpy is not related to the pathway, the enthalpy of the entire process could be calculated from the initial state (gases CO₂ + liquid water) and the final state (condensed CO₂ + CO₂ hydrate). The total calculated enthalpy of the second step is 32.0 kJ/mol, which is amazingly consistent with the value calculated according to the Clausius-Clapeyron equation (31.0-32.6 kJ/mol). These results further support our conclusion that the condensation and hydrate formation occur simultaneously at the second step. Similarly, by calculating the enthalpy from isotherm curves on wet AC (Rw = 3.1) at different temperatures (Figure S12), the enthalpy of the phase transfer corresponding to the sharp increase of uptake is 68.0 kJ/mol. Similar enthalpy on wet AC (20.0 kJ/mol, Rw = 1.58) and on dry AC (23.0 kJ/mol) indicates the same adsorptive mechanism.

It is worth noting that the initial of uptake (8.2 mmol/g) on wet CMK-3 agrees with the adsorptive amount (9 mmol/g) estimated from surface area of CMK-3 (998 m²/g), but it is obviously lower than that on dry sample (21 mmol/g). When introducing CO₂ in the system, the first process which happened is that CO₂ replaces part of water adsorbed on the material surface. After the first process, CO₂ molecule adsorbs on the surface of the pore, and the center of the pore is still filled by water molecule. Under experimental pressures with inducing time, the water in the pore is supersaturated by hydrate nucleus. Thus, CO₂ molecule prefers to condense in the center of the pore, and the hydrate nucleus tends to grow up to large crystals. Under these conditions, water in the pore continuously transfers out for the conversion, and CO₂ is continuously condensed in free space. The kinetic curves of wet CMK-3 indicate that the surface adsorption by CO₂ occurs first, and then the gas hydrates are formed along with the CO₂ condensation. The final result is that water in the pore is replaced by condensed CO₂.

With regard to AC, it is different from CMK-3. The initial adsorptive amount on wet AC (Rw = 1.58 and 3.1) is similar to that on dry AC, indicating a complete replacement at first. On CMK-3, initially CO₂ could only replace the water adsorbed on the carbon surface, while, on AC, CO₂ could replace almost all water in the pore. This is because the pore size of these two materials is different. The pore size of AC (2 nm) is smaller than that of CMK-3 (3.8 nm), resulting in stronger potential intensity in the pore of AC and leading to the CO₂ condensation in the pore of AC. As aforementioned, no gas hydrate could be formed on wet AC (Rw = 1.58). By increasing water loading amount, it was found that gas hydrates are formed and approach the theoretic hydrate volume. It is understandable that, upon increasing Rw, water is replaced and squeezed out of wedgy pore by CO₂ molecule. Part of water out of the pore could form CO₂ hydrates under this condition. It is very interesting that, once the gas hydrates start to form out of the pore, water inside the pore is driven out of the pore to participate in the CO₂ hydrate formation. Comparing with the CO₂ adsorption on CMK-3 having a dependent replacing process with the gas hydrate formation, the replacement on AC could occur anytime, i.e., before, during, or after the gas hydrate formation.

After initial replacement process on CMK-3 and AC, whether water in middle layer of pores is replaced by CO₂ with pressurizing CO₂ is determined by the pore size. In small pores, after initial adsorption of CO₂ on the carbon surface, there is no space for water molecule, and water is squeezed out of the pore completely. When the pore size is large enough for water to occupy middle layer of the pore, the binding energy of H₂O-H₂O and H₂O-CO₂ is 25 kJ/mol and 19.8 kJ/mol, respectively, calculated by DFT. Obviously, the CO₂-CO₂ interaction (8.3 kJ/mol) is weaker than that of CO₂-H₂O or H₂O-H₂O, meaning that the CO₂ insertion is not a spontaneous process. To overcome the enthalpy barrier, the CO₂ potential in the gas phase must be enhanced by increasing the CO₂ pressure. When the replacement of water by CO₂ in middle layer of the pore is achieved, CO₂ is located in the pore and condensed. The condensation pressure in 2 nm pore is obviously lower than that in 3 nm pore, since the overlap potential is generated by carbon walls. As observed from static adsorption experiments, the pressure for the water replacement by CO₂ in 2 nm slit pore is about 16 atm. In contrast, in 3.8 nm pore of CMK-3, no replacement could be observed. The pressure inducing this process in 3.8 nm pore is much higher than that for the gas hydrate formation. Therefore, the gas hydrate formation occurs firstly, leading to the transportation of water molecule from inside pore to outside in order to participate in the formation of CO₂ hydrate crystals.

Based on static and kinetic studies, the process of introducing CO₂ in wet carbon systems includes an initial replacement of water by CO₂ on the first layer of carbon surface, the replacement of water in middle layer of the pore, and the formation of CO₂ hydrates. Then, we calculated the entropy, enthalpy, and free energy extracted from ab initio calculations of CO₂-water system confined in layered graphene model with slit size of 1, 1.5, 2, 3, and 4 nm (Figure 3(a)). The replacing process could occur at the carbon surface before the gas hydrate formation, although the carbon atom on the surface of CMK-3 has higher binding energy with H₂O than with CO₂ as shown in Figure 3(b). Therefore, the replacement process of water by CO₂ should be dominated by the entropy.

The entropy of free molecules (CO₂ and H₂O) and adsorbed molecules on the carbon surface and in CO₂ hydrates was calculated from their vibrations (Figures S13 and S14). The total entropy is a sum of translational, rotational, and internal vibrational components: S_{\mathrm{t}\mathrm{o}\mathrm{t}}=S_{\mathrm{t}\mathrm{r}\mathrm{a}\mathrm{n}\mathrm{s}}+S_{\mathrm{r}\mathrm{o}\mathrm{t}}+S_{\mathrm{v}\mathrm{i}\mathrm{b}}. Six vibration modes of water and CO₂ were obtained, corresponding to bending vibration, symmetric stretching vibration, asymmetric stretching vibration, and rotational vibration. The detailed frequency is listed in Table S2. When comparing the frequency of molecules before and after the adsorption, it was found that obvious changes occur at low frequency region corresponding to rotational vibration. The water rotational vibration is much restricted in carbon pore especially micropore, which accounts for 50% loss of entropy. In contrast, the CO₂ rotational frequency is not changed obviously, meaning that the entropy of water replacement by CO₂ on the carbon surface is favorable, which could compensate the loss of enthalpy. By taking account of the pore size effect, the relationship between pressure, pore size, and temperature when the Gibbs energy of the process is less than zero is shown in Figure 3(c). Obviously, the larger the pore size and the higher the temperature, the higher the pressure needed to make Gibbs free energy less than zero. In addition to the entropy-driven mechanism for the replacement process, in large pore, the replacement of H₂O by CO₂ in single layer region could be spontaneously driven not only by entropy but also by enthalpy. As shown in Figure 3(a), when CO₂ molecule is tiled on the carbon (001) mesh along the C-O-C orientation, the π-π stacking interaction enables one side of the π electrons to focus near the carbon surface, while the opposite side of π electrons is weakened, leading to the situation that lone pair electrons from the {\mathrm{O}}_{\mathrm{H}\mathrm{2}\mathrm{O}} atom have strong interaction with positively charged {\mathrm{C}}_{\mathrm{C}\mathrm{O}\mathrm{2}} atom. If lone pair electrons of water directly bind with the surface of carbon, {\mathrm{H}}_{\mathrm{H}\mathrm{2}\mathrm{O}} must direct to CO₂ molecule. If lone pair electrons of water direct to {\mathrm{C}}_{\mathrm{C}\mathrm{O}\mathrm{2}}, the π electrons of CO₂ would still have strong interaction with the carbon surface. When compared with the other structure of C-H₂O-CO₂, simulation results show that the C-CO₂-H₂O structure is more stable. Thus, the stable configuration makes the replacement of water by CO₂ feasible.

To further confirm the conclusion that CO₂ hydrates could be formed outside the pore under specific conditions, in situ Raman studies were performed on CMK-3 and AC. The Raman spectrum of these materials displays characteristic peaks in the spectral range of 1200 to 1800 cm⁻¹. The G feature has a characteristic peak around 1582 cm⁻¹, assigning to all carbon structures having sp² hybridization [26]. The band around 1351 cm⁻¹ (λ = 532 nm) is known as the D band (defect-induced), and it requires a structural defect to be active in honeycomb carbon lattice. On account of the curvature in contrast to the perfect honeycomb lattice of graphite, the G band splits into the G+ and G− bands centered around 1571 and 1593 cm⁻¹, respectively (Figures 3(d) and 3(e)). Then, the in situ Raman spectra were collected on CMK-3 with water ratio of 1 (Rw = 1). At the initial loading of CO₂, the CO₂ adsorption occurs quickly according to the kinetic study results, and the Raman spectrum is separated into several parts (Figure 3(d)(I)). The Raman spectrum of CO₂ in adsorptive phase shows symmetric stretching (1286 cm⁻¹) and bending vibration (1389 cm⁻¹). When CO₂ molecule is enclosed by the hydrate network, the spectrum is broadened and red shifted (Figure 3(d)(II)). When the hydration is completed, symmetric stretching and bending vibration peaks of CO₂ at 1278 and 1386 cm⁻¹ shift to 1277 and 1382 cm⁻¹, respectively, assigning to CO₂ vibration in 5¹² cages. Similar changes were observed on AC as well. Symmetric stretching and bending vibration peaks (Figure 3(e)(I)) shift when water loading ratio is 3.1, indicating the CO₂ hydrate formation. At Rw = 1.58, no obvious shifts of CO₂ vibration peaks (Figure 3(e)(II)) were observed, indicting no hydrate formation under this loading ratio, which agrees well with earlier experimental results.

Based on our experimental results on both representatives (CMK-3 and active carbon AC) and DFT studies, the mechanism of gas hydrate formation on porous materials has been well revealed. The water coming from the inner pore could occupy the gap between particles, and the formed gas hydrate crystals fill the empty space. The established conclusion not only provides a new theory to the gas hydrate formation on porous materials, but also offers a reference for further application of gas separation.

The water molecule preadsorbed in the nanopore plays the role of recognizer to different gas (Figure 4(a)). After investigating several gas adsorption/desorption isotherms (CO₂, CH₄, and C₂H₆ in Figures S15-S17) on wet carbon materials, it was found that the replacing pressure is related to the gas type. In the case of CH₄, the water in the pore could only be driven out to form gas hydrate crystals out of pore at high pressure. In comparison to CH₄ and C₂H₆ (Figure 4(b)) on AC (Rw = 1.58), C₂H₆ could replace water adsorbed in the pore at a relatively low pressure. By utilizing this difference, the enhancement of CH₄/C₂H₆ selectivity from 4 (on dry AC) to 24 (on wet AC) was achieved by calculating from the breakthrough curves (Figure 4(c)). Only C₂H₆ could replace the water adsorbed on AC at current experiment conditions on account of its high binding energy and low entropy penalty. Due to the low binding energy in the pore, CH₄ cannot simply replace water, leading to an ultralow adsorptive amount on wet AC in comparison with that on dry AC (Figure S18). It should be noted that our finding here is much different from the gas separation based on gas hydrates. The enhancement of selectivity could only be achieved in the case that gas hydrates do not form. Once gas hydrates form, water would be driven out of pore to form the gas hydrate cluster, which increases methane competitive occupation in the empty pore of AC and the cage of formed gas hydrates. The increased methane uptake would reduce the CH₄/C₂H₆ selectivity. Therefore, the enhanced gas separation in our study follows a different mechanism. More importantly, the regeneration time of C₂H₆ on wet AC reduces to 1/6 on dry AC (Figure 4(d)). Thus, by simply using water in the pore, the drawback of regenerating strong adsorptive gas on porous adsorbents that hinders their industrial applications was overcome.

When the C₂H₆ partial pressure decreases, the water molecule could reoccupy the pore to promote desorption of C₂H₆ molecule. This enhancement could also be understood from the entropy change of the whole process. When water molecule is replaced by gas molecule, the entropy penalty should be compensated by the enthalpy. The CH₄ binding energy (-27.39 kJ) is much lower than that of C₂H₆ (-39.09 kJ) in the pore, so that water could only be replaced by C₂H₆. During the regeneration process, both of entropy and enthalpy are beneficial to the C₂H₆ desorption and water reoccupation in the pore of AC.

Based on the thermodynamic and kinetic study on CO₂ adsorption combined with DFT calculations, we have proposed a new mechanism of the gas hydrate formation on porous materials, which is quite different to previous hypothesis. We have discovered that the gas hydrates could not form and stay in the nanopore, and the excess gas observed in artificial gas hydrates resulted from the gas adsorption and condensation in the nanospace when water is driven out to form gas hydrates out of the pore. The established conclusion not only provides a theory to the gas hydrate formation on porous materials, but also offers a guideline for further application of gas hydrates in the gas separation. The significant improvement of gas selectivity and regeneration could be achieved by controlling the gas hydrate formation process accurately.

According to the new understanding, the method to enhance the gas hydrate formation by using hydrophobic porous materials should be reestimated. Upon the gas hydrate formation on porous materials, the preloaded water could be replaced by adsorbates such as CO₂, methane, and ethane. Low packing density is always an issue limiting the application of porous materials in the gas storage. Our research by combining the gas storage in the pore and the gas hydrate formation in the empty space overcomes the bottleneck of low packing density and would greatly enhance the gas storage capacity of porous materials under practical conditions. Thus, the present mechanism might be applicable to guide the separation of other binary gas systems.