Study of CHF3/CH2F2 Adsorption Separation in TIFSIX-2-Cu-i

Hydrofluorocarbons (HFCs) have important applications in different industries; however, they are environmentally unfriendly due to their high global warming potential (GWP). Hence, reclamation of used hydrofluorocarbons via energy-efficient adsorption-based separation will greatly contribute to reducing their impact on the environment. In particular, the separation of azeotropic refrigerants remains challenging, such as typical mixtures of CH2F2 (HFC-23) and CHF3 (HFC-32), due to a lack of adsorptive mechanisms. Metal–organic frameworks (MOFs) can provide a promising solution for the separation of CHF3–CH2F2 mixtures. In this study, the adsorption mechanism of CHF3–CH2F2 mixtures in TIFSIX-2-Cu-i was revealed at the microscopic level by combining static pure-component adsorption experiments, molecular simulations, and density-functional theory (DFT) calculations. The adsorption separation selectivity of CH2F2/CHF3 in TIFSIX-2-Cu-i is 3.17 at 3 bar under 308 K. The existence of similar TiF62− binding sites for CH2F2 or CHF3 was revealed in TIFSIX-2-Cu-i. Interactions between the fluorine atom of the framework and the hydrogen atom of the guest molecule were found to be responsible for determining the high adsorption separation selectivity of CH2F2/CHF3. This exploration is important for the design of highly selective adsorbents for the separation of azeotropic refrigerants.


Introduction
HFCs are third-generation fluorinated gases (F-gases), a class of synthetic compounds used primarily in refrigeration and air conditioning (RAC) [1][2][3].HFCs are potent greenhouse gases.So, their production and application must be phased down to meet the emission reduction target according to the Montreal Protocol [4,5].Depending on the actual production and use of refrigerants, many of the HFCs currently in use are azeotropic or near-azeotropic refrigerant mixtures [6].The current mainstream refrigerants include R-444A, R-447A, and R-448A, which are blends of HFCs (R-32, R-125, R-23, R-134a, etc.) with hydrofluoroolefins (HFOs).Difluoromethane (CH 2 F 2 , GWP = 675) and trifluoromethane (CHF 3 , GWP = 14,800) are the most common components of refrigerant mixture currently used in the refrigeration and air conditioning industry, with very high GWP [7,8].Therefore, to control HFC emissions, the first thing is to separate the components from the mixtures efficiently.However, due to the low efficiency of cryogenic distillation to separate refrigerant mixtures, the amount of refrigerant gas recovered remains low.There are very similar physical properties and molecular dynamics diameters for CHF 3 /CH 2 F 2 molecules, which makes the search for alternative technologies to energy-intensive distillation processes very challenging.In this study, CHF 3 /CH 2 F 2 was chosen as a sample to study the separation mechanism to provide a reference for the subsequent separation of HFCs.Selective adsorption technology has become an attractive solution for gas separation considering energy efficiency and environmental protection [9][10][11][12].Metal-organic framework (MOF) materials show great promise for gas storage and separation applications due to their significant advantages, such as flexible framework, tunable pore size and structure, and ultra-high specific surface area [13][14][15][16][17][18][19].
However, the practical applications of some MOFs are limited by their poor structural stability due to strong dependence on solvent molecules.The framework structure will collapse if they are exposed to air, high-strength acids, and bases for a period of time.TIFSIX-2-Cu-i is easy to regenerate and thermally stable under air atmosphere [20,21].The efficient separation of C 2 H 2 /C 2 H 4 by TIFSIX-2-M-i has been demonstrated by previous studies [20,22,23].Inspired by these findings, TIFSIX-2-Cu-i was chosen to study the separation of CHF 3 -CH 2 F 2 mixtures.To our knowledge, previous work on the mechanism of CHF 3 /CH 2 F 2 adsorption in TIFSIX-2-Cu-i is sparse, which is disadvantageous to understanding and predicting interactions between adsorbates and adsorbents.In this study, the feasibility of selective separation of CHF 3 /CH 2 F 2 by TIFSIX-2-Cu-i was evaluated for the first time.The method of adding polarization to a generic forcefield was used to obtain simulated and experimentally consistent adsorption isotherms, ensuring the accuracy of the forcefield.In this work, TIFSIX-2-Cu-i exhibits preferential adsorption of CH 2 F 2 over CHF 3 with a high CH 2 F 2 adsorption capacity (2.70 mmol/g at 298 K and 1 bar).Thermodynamic and kinetic analyses were carried out by a combination of adsorption experiments and molecular simulations.The adsorption selectivity, isosteric adsorption heat, and binding sites were investigated.In addition, these works are of great significance for exploring the adsorption separation of HFCs by fluorinated anion MOFs.

Adsorption Isotherm
To verify the accuracy of the forcefield, the present work compares the simulation adsorption isotherms for pure CHF 3 or CH 2 F 2 in TIFSIX-2-Cu-i at 288, 298, and 308 K with experimental adsorption isotherms (Figure 1).The simulation results under pressure values from 0 bar to 3 bar are in good agreement with the experimental data.Therefore, it can be inferred that the potential model and polarization forcefield parameters used are reliable for predicting the adsorption of CHF 3 and CH 2 F 2 .For all three temperatures, the trends of the CHF 3 and CH 2 F 2 isotherms are similar.These adsorption isotherms were fitted with the Langmuir isotherm model.The Langmuir equation is defined as shown in Equation (1).The parameters of the isotherms for CHF 3 and CH 2 F 2 are summarized in Table 1.q e = q m bp/(1 + bp) (1) where q e is the equilibrium adsorption capacity, q m is the maximum adsorption capacity, b is the adsorbate-adsorbent affinity coefficient, and p is the equilibrium pressure.
From Table 1, parameter b decreases with increasing temperature at the same adsorbate; parameter b is consistently larger for CH 2 F 2 than for CHF 3 at all three temperatures.The interaction of CH 2 F 2 with the framework was stronger compared to CHF 3 .Parameter q m shows a larger maximum adsorption capacity for CH 2 F 2 .sorption affinity for CH2F2 than CHF3.The adsorption capacity of both CHF3 and CH increased significantly with increasing pressure.However, the adsorption capacity CH2F2 increased faster compared to that of CHF3.These findings suggest that TIFSIX Cu-i has stronger binding ability regarding CH2F2, indicating that TIFSIX-2-Cu-i is a p tential material to separate CHF3/CH2F2 mixtures with high efficiency.

Adsorption Selectivity and Heat
In this section, thermodynamic adsorption selectivity and isosteric adsorption h were explored, which were calculated from experimental measurements.In addition, m lecular dynamics simulations were performed to explore the diffusivity of guest mo cules in TIFSIX-2-Cu-i.
Myers and Praunitz developed ideal adsorbed solution theory (IAST).Here, the m ticomponent adsorption equilibrium of CHF3 or CH2F2 was predicted using ideal a sorbed solution theory (IAST), which was calculated by the following equation [24]: where S is the selectivity of a component versus another one (e.g., CH2F2/CHF3), x is t molar fraction in the adsorbed phase, and y is the molar fraction in the gas phase.T relatively high adsorption separation selectivity is shown in Figure 2. The selectivity CH2F2/CHF3 is greater than 1.0 at all the adsorption isotherms, indicating that TIFSIX Cu-i preferentially adsorbs CH2F2.At 298 K and 308 K, selectivity increases with increasi pressure, while, at 288 K, selectivity decreases slightly with increasing pressure.At 288 the adsorption amount of CH2F2 in the high-pressure zone flattens more as pressure ris compared to CHF3.We hypothesize that most of the adsorption sites of the framewo were then occupied by CH2F2 molecules, making it difficult for the newly added CH molecules to find available adsorption sites, leading to slowdown of the adsorption rat To assess the interaction strength between the framework and gas molecules, we u lized single-component isotherms obtained at three distinct temperatures (Figure 1) to d termine the isosteric adsorption heat (Qst) of CH2F2/CHF3 on TIFSIX-2-Cu-i.The isoste adsorption heat was calculated indirectly using the Clausius−Clapeyron equation [25]: In detail, the uptake of CHF 3 and CH 2 F 2 reached 1.73 and 2.70 mmol/g at 1 bar and 298 K, respectively.At 308 K and 3 bar, the uptake of CH 2 F 2 was 3.79 mmol/g, which is almost double the CHF 3 uptake (1.99 mmol/g).At the same temperature and pressure, the adsorption capacity of CH 2 F 2 was higher than that of CHF 3 .Thus, there was a greater adsorption affinity for CH 2 F 2 than CHF 3 .The adsorption capacity of both CHF 3 and CH 2 F 2 increased significantly with increasing pressure.However, the adsorption capacity of CH 2 F 2 increased faster compared to that of CHF 3 .These findings suggest that TIFSIX-2-Cu-i has stronger binding ability regarding CH 2 F 2 , indicating that TIFSIX-2-Cu-i is a potential material to separate CHF 3 /CH 2 F 2 mixtures with high efficiency.

Adsorption Selectivity and Heat
In this section, thermodynamic adsorption selectivity and isosteric adsorption heat were explored, which were calculated from experimental measurements.In addition, molecular dynamics simulations were performed to explore the diffusivity of guest molecules in TIFSIX-2-Cu-i.
Myers and Praunitz developed ideal adsorbed solution theory (IAST).Here, the multicomponent adsorption equilibrium of CHF 3 or CH 2 F 2 was predicted using ideal adsorbed solution theory (IAST), which was calculated by the following equation [24]: where S is the selectivity of a component versus another one (e.g., CH 2 F 2 /CHF 3 ), x is the molar fraction in the adsorbed phase, and y is the molar fraction in the gas phase.The relatively high adsorption separation selectivity is shown in Figure 2. The selectivity of CH 2 F 2 /CHF 3 is greater than 1.0 at all the adsorption isotherms, indicating that TIFSIX-2-Cu-i preferentially adsorbs CH 2 F 2 .At 298 K and 308 K, selectivity increases with increasing pressure, while, at 288 K, selectivity decreases slightly with increasing pressure.At 288 K, the adsorption amount of CH 2 F 2 in the high-pressure zone flattens more as pressure rises compared to CHF 3 .We hypothesize that most of the adsorption sites of the framework were then occupied by CH 2 F 2 molecules, making it difficult for the newly added CH 2 F 2 molecules to find available adsorption sites, leading to slowdown of the adsorption rate.
where qi refers to the isosteric heat of adsorption, kJ/mol; P is the pressure, MPa; T is the temperature, K; and R is the gas constant, 8.314 J/(mol•K).As shown in Figure 3, the Qst values of CHF3 and CH2F2 were around 20 and 23 kJ/mol −1 .The heat of adsorption of CH2F2 was always higher than the heat of adsorption of CHF3 in TIFSIX-2-Cu-i.Therefore, it is directly verified that TIFSIX-2-Cu-i interacts more strongly with CH2F2 than CHF3, which leads to greater adsorption of CH2F2 than CHF3.In this part, the free diffusion behavior of CHF3 and CH2F2 in TIFSIX-2-Cu-i was explored.The corresponding mean square displacements (MSDs) obtained from the simulations are shown in Figure 4.The self-diffusion coefficients of CHF3 and CH2F2 in the TIFSIX-2-Cu-i were calculated using the Einstein relation as shown below: To assess the interaction strength between the framework and gas molecules, we utilized single-component isotherms obtained at three distinct temperatures (Figure 1) to determine the isosteric adsorption heat (Qst) of CH 2 F 2 /CHF 3 on TIFSIX-2-Cu-i.The isosteric adsorption heat was calculated indirectly using the Clausius-Clapeyron equation [25]: where q i refers to the isosteric heat of adsorption, kJ/mol; P is the pressure, MPa; T is the temperature, K; and R is the gas constant, 8.314 J/(mol•K).As shown in Figure 3, the Qst values of CHF 3 and CH 2 F 2 were around 20 and 23 kJ/mol −1 .The heat of adsorption of CH 2 F 2 was always higher than the heat of adsorption of CHF 3 in TIFSIX-2-Cu-i.Therefore, it is directly verified that TIFSIX-2-Cu-i interacts more strongly with CH 2 F 2 than CHF 3 , which leads to greater adsorption of CH 2 F 2 than CHF 3 .
Molecules 2024, 29, x FOR PEER REVIEW 4 of 14 where qi refers to the isosteric heat of adsorption, kJ/mol; P is the pressure, MPa; T is the temperature, K; and R is the gas constant, 8.314 J/(mol•K).As shown in Figure 3, the Qst values of CHF3 and CH2F2 were around 20 and 23 kJ/mol −1 .The heat of adsorption of CH2F2 was always higher than the heat of adsorption of CHF3 in TIFSIX-2-Cu-i.Therefore, it is directly verified that TIFSIX-2-Cu-i interacts more strongly with CH2F2 than CHF3, which leads to greater adsorption of CH2F2 than CHF3.In this part, the free diffusion behavior of CHF3 and CH2F2 in TIFSIX-2-Cu-i was explored.The corresponding mean square displacements (MSDs) obtained from the simulations are shown in Figure 4.The self-diffusion coefficients of CHF3 and CH2F2 in the TIFSIX-2-Cu-i were calculated using the Einstein relation as shown below: In this part, the free diffusion behavior of CHF 3 and CH 2 F 2 in TIFSIX-2-Cu-i was explored.The corresponding mean square displacements (MSDs) obtained from the simulations are shown in Figure 4.The self-diffusion coefficients of CHF 3 and CH 2 F 2 in the TIFSIX-2-Cu-i were calculated using the Einstein relation as shown below: where the average is taken over time t for the mean square displacement of the center of mass position vectors r of all the molecules N in the system; ‹›indicates the overall average.The calculated results show that the self-diffusion coefficients of CHF 3 and CH 2 F 2 are 1.15 × 10 −4 cm 2 /s and 2.18 × 10 −4 cm 2 /s, respectively.This finding showed that TIFSIX-2-Cu-i exhibited high kinetic selectivity for CH 2 F 2 over CHF 3 .TIFSIX-2-Cu-i is a doubly interpenetrated framework attributed to the much longer organic linker 4,4-dipyridylacetylene and slightly large pore sizes of about 5.2 × 5.2 Å.That makes it easier for the mixture to diffuse into the adsorption sites within the pores.The properties of this MOF can be characterized as pillared square lattice networks with a pcu topology, attributed to their pore surfaces with narrow pore sizes and highly electrostatic pore surfaces.These features combined provide exceptionally strong binding interactions with polarizable molecules, such as CHF 3 and CH 2 F 2 [26].This also enabled the two guest molecules to possess high adsorption capacity.
cules 2024, 29, x FOR PEER REVIEW 5 of where the average is taken over time t for the mean square displacement of the center mass position vectors r of all the molecules N in the system; ‹›indicates the overall avera The calculated results show that the self-diffusion coefficients of CHF3 and CH2F2 are 1.15 10 − 4 cm 2 /s and 2.18 × 10 − 4 cm 2 /s, respectively.This finding showed that TIFSIX-2-Cu-i e hibited high kinetic selectivity for CH2F2 over CHF3.TIFSIX-2-Cu-i is a doubly interpen trated framework attributed to the much longer organic linker 4,4-dipyridylacetylene a slightly large pore sizes of about 5.2 × 5.2 Å.That makes it easier for the mixture to diffu into the adsorption sites within the pores.The properties of this MOF can be characteriz as pillared square lattice networks with a pcu topology, attributed to their pore surfac with narrow pore sizes and highly electrostatic pore surfaces.These features combin provide exceptionally strong binding interactions with polarizable molecules, such CHF3 and CH2F2 [26].This also enabled the two guest molecules to possess high adsor tion capacity.

Adsorption Sites
Figure 5 shows the optimal adsorption binding sites for CHF3 and CH2F2.The sna shots obtained detailed information about the adsorption of pure CHF3 and CH2F2 TIFSIX-2-Cu-i at 298 K and 1 bar.In the doubly interpenetrated framework of TIFSIX Cu-i, the H atom of CHF3 binds with the F atom from TiF6 2− .The distance of the C-H hydrogen bond was 2.372 Å (Figure 5b).The H⋯F distance of 2.372 Å obtained by t simulation was smaller than the sum of the van der Waals radii of H and F (2.55 Å) [2 confirming the existence of electrostatic interactions for H δ+ ⋯F δ− .This was consistent w the reported binding sites of TIFSIX-2-Cu-i to short-chain alkanes (C2H2, C2H4) [20,22,2 CH2F2 has a similar binding site.The two H atoms of CH2F2 are bound at the F site virtue of a synergistic hydrogen bonding interaction.The shortest length of the C-H bond between CH2F2 and the TIF6 2− site is 2.172 Å, which is shorter than the C-H⋯F b tween CHF3 and TIF6 2− .The TIFSIX-2-Cu-i interacts more strongly with CH2F2 than CH

Adsorption Sites
Figure 5 shows the optimal adsorption binding sites for CHF 3 and CH 2 F 2 .The snapshots obtained detailed information about the adsorption of pure CHF 3 and CH 2 F 2 in TIFSIX-2-Cu-i at 298 K and 1 bar.In the doubly interpenetrated framework of TIFSIX-2-Cu-i, the H atom of CHF 3 binds with the F atom from TiF 6 2− .The distance of the C-H• • • F hydrogen bond was 2.372 Å (Figure 5b).The H• • • F distance of 2.372 Å obtained by the simulation was smaller than the sum of the van der Waals radii of H and F (2.55 Å) [27], confirming the existence of electrostatic interactions for H δ+ • • • F δ− .This was consistent with the reported binding sites of TIFSIX-2-Cu-i to short-chain alkanes (C 2 H 2 , C 2 H 4 ) [20,22,23].CH 2 F 2 has a similar binding site.The two H atoms of CH 2 F 2 are bound at the F site by virtue of a synergistic hydrogen bonding interaction.The shortest length of the C-H• • • F bond between CH 2 F 2 and the TIF 6 2− site is 2.172 Å, which is shorter than the C-H• • • F between CHF 3 and TIF 6 2− .The TIFSIX-2-Cu-i interacts more strongly with CH 2 F 2 than CHF 3 .
Radial Distribution Functions (RDFs) amount to one of the most common methods for determination of interatomic distances.The RDFs in 298 K describing the interactions between the individual atoms of the pure CHF 3 /CH 2 F 2 and the TIFSIX-2-Cu-i framework are shown in Figure 6a,b.In Figure 6a, the framework interacts preferentially with H in CHF 3 .From the investigation of the RDF of H (CHF 3 ) with each atom of the framework in Figure 6c, it was found that F (framework) interacts preferentially with H (CHF 3 ), which is the same as the snapshot conclusion of Figure 5a.Similarly, according to Figure 6b,d, it is found that F (framework) interacts preferentially with H (CH 2 F 2 ).Compared to CHF 3 , CH 2 F 2 has more H-F bonds interacting with the framework at the same time.So, we believe that the reason for the higher adsorption affinity of CH 2 F 2 than CHF 3 is due to the binding geometry of CH 2 F 2 /CHF 3 adsorbed in the supercage of TIFSIX-2-Cu-i.
Figure 5 shows the optimal adsorption binding sites for CHF3 and CH2F2.The snapshots obtained detailed information about the adsorption of pure CHF3 and CH2F2 in TIFSIX-2-Cu-i at 298 K and 1 bar.In the doubly interpenetrated framework of TIFSIX-2-Cu-i, the H atom of CHF3 binds with the F atom from TiF6 2− .The distance of the C-H⋯F hydrogen bond was 2.372 Å (Figure 5b).The H⋯F distance of 2.372 Å obtained by the simulation was smaller than the sum of the van der Waals radii of H and F (2.55 Å) [27], confirming the existence of electrostatic interactions for H δ+ ⋯F δ− .This was consistent with the reported binding sites of TIFSIX-2-Cu-i to short-chain alkanes (C2H2, C2H4) [20,22,23].CH2F2 has a similar binding site.The two H atoms of CH2F2 are bound at the F site by virtue of a synergistic hydrogen bonding interaction.The shortest length of the C-H⋯F bond between CH2F2 and the TIF6 2− site is 2.172 Å, which is shorter than the C-H⋯F between CHF3 and TIF6 2− .The TIFSIX-2-Cu-i interacts more strongly with CH2F2 than CHF3.Radial Distribution Functions (RDFs) amount to one of the most common methods for determination of interatomic distances.The RDFs in 298 K describing the interactions between the individual atoms of the pure CHF3/CH2F2 and the TIFSIX-2-Cu-i framework are shown in Figure 6a,b.In Figure 6a, the framework interacts preferentially with H in CHF3.From the investigation of the RDF of H (CHF3) with each atom of the framework in Figure 6c, it was found that F (framework) interacts preferentially with H (CHF3), which is the same as the snapshot conclusion of Figure 5a.Similarly, according to Figure 6b,d, it is found that F (framework) interacts preferentially with H (CH2F2). Compared to CHF3, CH2F2 has more H-F bonds interacting with the framework at the same time.So, we believe that the reason for the higher adsorption affinity of CH2F2 than CHF3 is due to the binding geometry of CH2F2/CHF3 adsorbed in the supercage of TIFSIX-2-Cu-i.

Redistribution of Charge Density
In order to study the charge change in TIFSIX-2-Cu-i after adsorption, DFT calculations were carried out to investigate the redistribution of charge density in this system after adsorption of CHF3/CH2F2 molecules.As shown in Figure 7, the electrons of the H

Redistribution of Charge Density
In order to study the charge change in TIFSIX-2-Cu-i after adsorption, DFT calculations were carried out to investigate the redistribution of charge density in this system after adsorption of CHF 3 /CH 2 F 2 molecules.As shown in Figure 7, the electrons of the H atoms of CHF 3 /CH 2 F 2 migrate to the F atoms of the framework due to the strong electronwithdrawing ability of the F atoms, where "-" denotes a negative charge.As shown in Figure 8, the blue area in Figure 8b is larger and darker than in Figure 8a, and the charge density transfer between CH 2 F 2 and the framework F atom is more pronounced.This may be because, compared to CHF 3 , CH 2 F 2 has more interactions between H atoms and framework F atoms, which is also the reason for the relatively high adsorption separation selectivity for CH 2 F 2 over CHF 3 .

Synthetic Procedures
Cu(BF4)2 (1 mmol), (NH4)2TiF6 (1 mmol), and 1,2-Di(pyridin-4yl) ethyne (2 mmol) were dissolved in 5 mL of water and 10 ml of methanol, and a blue slurry product was obtained after stirring at 338 K for 12 h.Then, the slurry was filtered and washed with 10 ml of methanol.The blue filter cake was heated at 393 K for 12 h under vacuum conditions to obtain TIFSIX-2-Cu-i material [26].

Synthetic Procedures
Cu(BF4)2 (1 mmol), (NH4)2TiF6 (1 mmol), and 1,2-Di(pyridin-4yl) ethyne (2 mmol) were dissolved in 5 mL of water and 10 ml of methanol, and a blue slurry product was obtained after stirring at 338 K for 12 h.Then, the slurry was filtered and washed with 10 ml of methanol.The blue filter cake was heated at 393 K for 12 h under vacuum conditions

Synthetic Procedures
Cu(BF 4 ) 2 (1 mmol), (NH 4 ) 2 TiF 6 (1 mmol), and 1,2-Di(pyridin-4yl) ethyne (2 mmol) were dissolved in 5 mL of water and 10 mL of methanol, and a blue slurry product was obtained after stirring at 338 K for 12 h.Then, the slurry was filtered and washed with 10 mL of methanol.The blue filter cake was heated at 393 K for 12 h under vacuum conditions to obtain TIFSIX-2-Cu-i material [26].

Characterization
The activated TIFSIX-2-Cu-i was subjected to X-ray diffraction characterization and scanned using a Bruker AXS D8 ADVANCE diffractometer under a CuKα radiation source operated at a voltage of 40 kV, a current of 20 mA, and a scattering angle in the range (2θ) of 5-40 degrees.The XRD pattern of TIFSIX-2-Cu-i was presented in Figure 9, which was compared to calculated patterns.ASAP 2460 (Micromeritics company, Shanghai, China) was employed in this experiment.Pore structure was characterized by the N 2 adsorption method.The experimental temperature was 77 K. Before testing, the sample was treated by 12 h vacuumization at 393 K. Specific surface area was calculated using a multipoint Brunauer-Emmett-Teller model (BET).Table 2 shows the pore parameters of TIFSIX-2-Cu-i.
Molecules 2024, 29, x FOR PEER REVIEW 8 of 14 of 5-40 degrees.The XRD pattern of TIFSIX-2-Cu-i was presented in Figure 9, which was compared to calculated patterns.ASAP 2460 (Micromeritics company, Shanghai, China) was employed in this experiment.Pore structure was characterized by the N2 adsorption method.The experimental temperature was 77 K. Before testing, the sample was treated by 12 h vacuumization at 393 K. Specific surface area was calculated using a multipoint Brunauer−Emmett−Teller model (BET).Table 2 shows the pore parameters of TIFSIX-2-Cu-i.

Single-Component Adsorption Measurements
The adsorption isotherms of CHF3 and CH2F2 were measured in the absolute pressure range of 1−3 bar in TIFSIX-2-Cu-i framework.The experimental temperatures were 288, 298, and 308 K. Excess adsorption experiments were performed using activated TIFSIX-2-Cu-i monomer.The temperature was controlled by an external circulating water bath.Before the measurements, TIFSIX-2-Cu-i was degassed at 393 K for 12 h under vacuum conditions.CHF3 and CH2F2 gas of purity 99.99% were used as adsorbates.The adsorption capacity of pure CHF3 or CH2F2 was calculated based on the pressure changes before and after adsorption.Figure 10 shows the diagram of adsorption measurements' experimental apparatus.A standard volumetric method was used to measure pure gas adsorption isotherms [28].The homemade apparatus is designed by the proposed method.

Single-Component Adsorption Measurements
The adsorption isotherms of CHF 3 and CH 2 F 2 were measured in the absolute pressure range of 1-3 bar in TIFSIX-2-Cu-i framework.The experimental temperatures were 288, 298, and 308 K. Excess adsorption experiments were performed using activated TIFSIX-2-Cu-i monomer.The temperature was controlled by an external circulating water bath.Before the measurements, TIFSIX-2-Cu-i was degassed at 393 K for 12 h under vacuum conditions.CHF 3 and CH 2 F 2 gas of purity 99.99% were used as adsorbates.The adsorption capacity of pure CHF 3 or CH 2 F 2 was calculated based on the pressure changes before and after adsorption.Figure 10 shows the diagram of adsorption measurements' experimental apparatus.A standard volumetric method was used to measure pure gas adsorption isotherms [28].The homemade apparatus is designed by the proposed method.

Models
The interpenetrated polymorph, TIFSIX-2-Cu-i, is composed of doubly interpe trated nets that are isostructural to the nets in TIFSIX-2-Cu.The independent nets are s gered with respect to one another, affording 5.2 Å pores [22,23].The crystal cells used the simulation were downloaded from the Cambridge Crystallographic Data Cen (CCDC) as structural files.Optimized structure by DFT simulation was used for furt calculations.TIFSIX-2-Cu-i is a variant of SIFSIX-2-Cu-i.Ti 4+ has a higher polarizab relative to Si 4+ [29].So, TIFSIX-2-Cu-i has a higher thermal stability (decomposition t perature of 262 °C), which may be attributed to the relatively higher polarizability of [20].TIFSIX-2-Cu-i atoms in the framework with different chemical properties are sho in Figure 11.

Models
The interpenetrated polymorph, TIFSIX-2-Cu-i, is composed of doubly interpenetrated nets that are isostructural to the nets in TIFSIX-2-Cu.The independent nets are staggered with respect to one another, affording 5.2 Å pores [22,23].The crystal cells used in the simulation were downloaded from the Cambridge Crystallographic Data Center (CCDC) as structural files.Optimized structure by DFT simulation was used for further calculations.TIFSIX-2-Cu-i is a variant of SIFSIX-2-Cu-i.Ti 4+ has a higher polarizability relative to Si 4+ [29].So, TIFSIX-2-Cu-i has a higher thermal stability (decomposition temperature of 262 • C), which may be attributed to the relatively higher polarizability of Ti 4+ [20].TIFSIX-2-Cu-i atoms in the framework with different chemical properties are shown in Figure 11.

Models
The interpenetrated polymorph, TIFSIX-2-Cu-i, is composed of doubly interpenetrated nets that are isostructural to the nets in TIFSIX-2-Cu.The independent nets are staggered with respect to one another, affording 5.2 Å pores [22,23].The crystal cells used in the simulation were downloaded from the Cambridge Crystallographic Data Center (CCDC) as structural files.Optimized structure by DFT simulation was used for further calculations.TIFSIX-2-Cu-i is a variant of SIFSIX-2-Cu-i.Ti 4+ has a higher polarizability relative to Si 4+ [29].So, TIFSIX-2-Cu-i has a higher thermal stability (decomposition temperature of 262 °C), which may be attributed to the relatively higher polarizability of Ti 4+ [20].TIFSIX-2-Cu-i atoms in the framework with different chemical properties are shown in Figure 11.

Density Functional Theory Calculations
The MOFs' structure was optimized using ab initio density functional theory (DFT) as implemented in the Vienna Ab Initio Simulation Package version 2.2 (VASP) [30], with

Density Functional Theory Calculations
The MOFs' structure was optimized using ab initio density functional theory (DFT) as implemented in the Vienna Ab Initio Simulation Package version 2.2 (VASP) [30], with the overall energy converged to within 10 −5 eV per atom.The Perdew-Burke-Ernzerhof (PBE) function of the generalized gradient approximation (GGA) [31] was used to represent the electron exchange correlation, and a cutoff energy of 500 eV was set for the plane wave.According to the Monkhorst-Pack methodology, the Brillouin zone was sampled with a series of K-point grids (2 × 2 × 4).After geometrical optimization, we obtained the electron charge density and then used Density Derived Electrostatic and Chemical (DDEC6) [32,33] to calculate the atomic charge of the net atomic charge framework for each MOF atom (Table 3).DFT simulations were used to explore the redistribution of charge density in this system after adsorption of CHF 3 or CH 2 F 2 molecules using the CP2K code [34].The DZVP-MOLOPT-SR-GTH [35] basis set and the Goedecker-Teter-Hutter [36] pseudopotential were used, and the density generalization employed was a PBE with DFT-D3 [37] dispersion corrections.

Grand Canonical Monte Carlo Simulations
All GCMC simulations were performed with the RASPA [38] code to study CHF 3 /CH 2 F 2 adsorption properties under different conditions.A grand canonical systematic (µVT) was used, where the system was under constant chemical potential, volume, and temperature.To eliminate periodic boundary conditions, we used supercell by 2 × 2 × 3 replicas of the unit cell for the calculations.The van der Waals interactions were truncated to a radius of 12 Å, and tail corrections were used to approximate the contributions beyond this truncation.In the simulations, 1 × 10 6 Monte Carlo steps were used for the equilibration and 1 × 10 7 Monte Carlo steps were used for production runs.The adsorbate molecules and the adsorbate framework were treated as rigid structures.CHF 3 and CH 2 F 2 were modeled as rigid tetrahedral molecules with five charged interaction sites.From previous simulation study, rigid body models were used to represent molecules as a collection of fixed geometric shapes that maintain a constant structure and orientation throughout the simulation.The host-guest and guest-guest interactions in the system were described by the short-range force and the electrostatic force, which are described by the Lennard-Jones and Coulomb potential functions (Equation ( 5)): where jσ ij and ε ij are the collision diameter and potential well depth, respectively, r ij is the distance between sites i and j, q i denotes the atomic charge on site i, and ε 0 is the permittivity of free space.The cross-interactions with other molecules and frameworks are obtained using the Lorentz-Berthelot mixing rule (as shown in Equations ( 6) and ( 7)): A polarizable forcefield was employed to achieve an accurate description of the adsorption behavior of CHF 3 /CH 2 F 2 in TIFSIX-2-Cu-i for molecular simulations.Backpolarization was neglected to achieve reasonable simulation times.To account for the implied polarization, we rescaled the Lennard-Jones [39] energy parameters according to the atomic polarizabilities.The Lennard-Jones energy parameters and charges of CHF 3 and CH 2 F 2 were taken from previous studies [40,41].The Lennard-Jones parameters of TIFSIX-2-Cu-i for N, C, and H were taken from the OPLS-AA forcefield, and the rest of the atoms were taken from the UFF-Dreiding hybrid forcefield [42].The equations used to adjust the parameters in this study are as follows: α i means the polarizability of atom i , α max means the max polarizability, and ε i means the initial forcefield parameter.λ and ξ are scaling factors between 0 and 1, whose values depend on the discrepancy between the experimental data and the simulation results, used to rescale the Lennard-Jones energy parameters.The detailed methodology is described in Refs.[43,44].In this study, by fitting the experimental data to the simulation results, λ was set to 0.2 and ξ to 0.9 for CH 2 F 2 , while λ was set to 0.9 and ξ to 0.01 for CHF 3 .Tables 3 and 4 summarize all the forcefield parameters, atomic polarizabilities [45,46], and atomic charges.

Molecular Dynamics Simulations Details
In this paper, RASPA code was used to perform molecular dynamics simulations of CHF 3 /CH 2 F 2 in TIFSIX-2-Cu-i.The polarized forcefield parameters from Tables 3 and 4 were employed.All MD simulations were employed for 1 ns with a time step of 0.5 fs in the NVT ensemble to explore the diffusion of the equimolar CHF 3 /CH 2 F 2 mixture in TIFSIX-2-Cu-i.The simulations were performed for 1 × 10 7 cycles, 2000 initialization cycles, and 20,000 equilibration cycles.We truncated the van der Waals interaction with a radius of 12.0 Å and used tail correction to approximate the contribution beyond this cutoff.

Conclusions
In this study, the adsorption mechanisms of CHF 3 or CH 2 F 2 in the TIFSIX-2-Cu-i framework were studied by combining single-component adsorption experiments and molecular simulations.In order to ensure consistency between the experimental data and the simulation results, a polarization forcefield was introduced.TIFSIX-2-Cu-i has excellent CH 2 F 2 adsorption capacity (3.79 mmol/g) and CH 2 F 2 /CHF 3 selectivity (3.17) at 3 bar and 308 K, making it a promising material to separate CHF 3 /CH 2 F 2 mixtures.Regarding the competitive adsorption of CHF 3 -CH 2 F 2 mixtures in TIFSIX-2-Cu-i, both the thermodynamic and kinetic selectivity of CH 2 F 2 relative to CHF 3 were observed to be relatively high.According to the combined effect of adsorption and diffusion, TIFSIX-2-Cu-i exhibits markedly preferential adsorption of CH 2 F 2 for CHF 3 .The calculated heats of

Figure 5 .
Figure 5.The typical binding sites for CH 2 F 2 (a) or CHF 3 (b) in TIFSIX-2-Cu-i.The cyan, gray, and white spheres represent fluorine, carbon, and hydrogen atoms, respectively.

Figure 6 .
Figure 6.The RDF between the framework and each atom of CHF3 (a)/CH2F2 (b); the RDF between the representative atoms on the framework and hydrogen atom of CHF3 (c)/CH2F2 (d) in 298 K.

Figure 6 .
Figure 6.The RDF between the framework and each atom of CHF 3 (a)/CH 2 F 2 (b); the RDF between the representative atoms on the framework and hydrogen atom of CHF 3 (c)/CH 2 F 2 (d) in 298 K.

Figure 8 .
Figure 8.The slices of charge density redistribution for TIFSIX-2-Cu-i and CHF3 (a)/CH2F2 (b) after molecular adsorption, which correspond to the electron transfer between hydrogen atom of CHF3/CH2F2 and the fluorine atom of TIF6 2− .

Figure 8 .
Figure 8.The slices of charge density redistribution for TIFSIX-2-Cu-i and CHF3 (a)/CH2F2 (b) after molecular adsorption, which correspond to the electron transfer between hydrogen atom of CHF3/CH2F2 and the fluorine atom of TIF6 2− .

Figure 8 .
Figure 8.The slices of charge density redistribution for TIFSIX-2-Cu-i and CHF 3 (a)/CH 2 F 2 (b) after molecular adsorption, which correspond to the electron transfer between hydrogen atom of CHF 3 /CH 2 F 2 and the fluorine atom of TIF 6 2− .

Table 1 .
Parameters of CHF 3 and CH 2 F 2 fitted by the Langmuir Adsorption Isotherm Model.

Table 3 .
Partial charges, forcefield parameters, and atomic polarizabilities corresponding to the atom types in CHF 3 and CH 2 F 2 .

Table 4 .
Forcefield parameters and the atomic polarizabilities corresponding to the atom types in the TIFSIX-2-Cu-i.