Selective Water Transport in an Alanine-Functionalized Metal − Organic Framework: A Computational Study

,


■ INTRODUCTION
−4 For instance, Bonnefoy et al. postsynthetically decorated the bridging ligands of MOFs, that is, Al-MIL-101-NH 2 , In-MIL-68-NH 2 , and Zr-UiO-66-NH 2 with various enantiopure oligopeptides using microwave irradiation.The grafted MOFs were used as efficient catalysts for the chiral aldol reaction. 5Chan et al. synthesized a membrane by incorporating L-histidine (l-His) into the ZIF-8 framework using a contra diffusion method.The homochiral l-His-ZIF-8 membrane showed good selectivity for the Renantiomer of 1-phenylethanol over the S-enantiomer, with an enantiomeric excess of up to 76%. 6 Lyu et al. embedded cytochrome c (Cyt c) proteins onto the external surfaces of ZIF-8 crystals using a coprecipitation method.The peroxidase activity of Cyt c-functionalized ZIF-8 increased by 10-fold compared to free Cyt c in solution, which was then successfully used for fast and highly sensitive detection of extremely small amounts of explosive organic peroxides in solution. 7Chen et al. heated ZrCl 4 with amino-triphenyl dicarboxylic acid to form MOF nanoparticles (NMOFs) that were further covalently linked to nucleic acids via azide groups in the NMOF to form NMOF-DNA1 complex.The anticancer drug doxorubicin was loaded into the pores of NMOFs, and the drug-loaded NMOFs were locked by the ATP-aptamer or the ATP-AS1411 hybrid aptamer, which can be unlocked when ATP forms ATP−aptamer complexes in cancer cells, thereby achieving targeted drug delivery. 8In the context of bioimaging, ZIF-8 was loaded with DNAzyme fragments with fluorescent tags generating FRET fluorescence.The tagged DNA can be released in an acidic endosome, which leads to quenching of FRET fluorescence due to miRNA activity. 9ne potential application of incorporating biomolecules into MOFs, which has not been explored yet, is the imitation of biological channels.In principle, the pores of crystalline MOFs, shaped and functionalized by biological molecules, could be used to create an environment that functions like a biological channel, as has been demonstrated in other solid-state nanopores. 10−13 The highly selective and rapid water transport observed in AQPs has been attributed to the distinct structure and chemistry of the pore channel.First, at its narrowest location, which is ∼2.8 Å in diameter, the AQP channel has a selective histidine−arginine filter, which sterically blocks the hydrated ions. 13Second, electrostatic repulsion from an asparagine−proline−alanine (NPA) motif contributes to the exclusion of positively charged ionic species. 14,15As a result, the transport of water molecules, which form a water wire in the AQP channel, is allowed but not that of protons. 12,16,17−26 However, AWC platforms that rely on soft matter may be susceptible to long-term stability issues; hence, the use of crystalline porous materials, such as MOFs, could pave the way for creating long-lasting AWCs.
Using molecular simulations, we recently studied water diffusion in Ni-CPO-27 and Ni-CPO-54 MOFs, whose onedimensional pores were functionalized by grafting imidazolecarboxaldehyde/pyrazolecarboxaldehydes, zwitterionic imidazolecarboxylic acids, pyrazolecarboxylic acids, and proline on to the coordinatively unsaturated open metal sites. 27This type of coordination bond found in the proposed AWCs (i.e., coordination of −COO− on to an open metal site) was experimentally demonstrated in Zn-CPO-27 with zwitterionic proline. 28We found that the proline-grafted Ni-CPO-54 (proline-Ni-CPO-54) had the largest self-diffusion coefficient among the structures that were investigated.This finding has motivated us to expand our work to incorporate biomolecules into MOFs for AWC design.Therefore, in this work, we propose and computationally investigate a zwitterionic alaninefunctionalized Ni-CPO-27 (referred to as alanine-Ni-CPO-27 hereafter) for its potential as an AWC and application to desalination.In fact, our preliminary molecular dynamics (MD) simulations revealed that the self-diffusion coefficient of water in alanine-Ni-CPO-27 is more than 5 times larger than that in proline-Ni-CPO-54 (see Figure S1 in the Supporting Information).Ni-CPO-27 has been shown experimentally to be stable under various conditions, viz., at room and higher temperatures (100 °C) and in acidic solution. 29The stability of coordinated zwitterionic alanine molecules in Ni-CPO-27 is determined by computing the binding energy of alanine and water to Ni-CPO-27 with density functional theory (DFT) calculations.In addition, the quantum theory of atoms in molecules (QTAIM) is utilized to analyze the hydrogen bond network in alanine-Ni-CPO-27.The osmotic permeability of water in alanine-Ni-CPO-27 and its ion rejection ability are evaluated by nonequilibrium MD simulations.Finally, the free energy profiles associated with water, Na + , and Cl − transport in alanine-Ni-CPO-27 are calculated and analyzed.
■ METHODS DFT Calculations.Periodic plane-wave DFT calculations 30 were carried out in CASTEP (version 19.1) for the geometry optimization and the calculation of molecular binding energies and partial atomic charges.First, a unit cell of Ni-CPO-27 was functionalized by grafting one zwitterionic alanine or one water molecule on each coordinatively unsaturated Ni metal atom of Ni-CPO-27 to generate two structures, one fully coordinated with alanine and the other fully coordinated with water molecules, respectively.The structures were then optimized while allowing the cell dimensions to vary independently (see Figure S2 for the optimized structures).The electronic exchange and correlation potential were modeled using the PBE-TS functional, which includes a correction for weak dispersion interactions. 31The electron−core interactions were described by ultrasoft potentials, 32 while the wave function was described with a plane wave basis set with a cut-off energy of 550 eV.The convergence criteria for energy and force were set to 1.0 −4 eV/atom and 0.05 eV/Å, respectively.To accurately describe the d orbital electrons of the Ni atom, a Hubbard Uvalue 33 of 6.4 eV was used. 34Since the Ni atom contains unpaired electrons, all calculations were spin-polarized. 35As we demonstrated in our previous work, the ground state of Ni-CPO-27 was achieved when the initial spin state of the Ni atom was set to 2. 27 The binding energies of a single zwitterionic alanine and a single water molecule in Ni-CPO-27 were calculated according to where ΔE is the binding energy, E MOF+mol is the energy of the MOF coordinated with the molecules, E mol is the energy of the isolated zwitterionic alanine or water molecules, E MOF is the energy of the bare Ni-CPO-27 (obtained from a single-point DFT calculation), and N mol = 18 is the total number of molecules coordinated with the open metal sites, that is, one molecule per metal site, in the unit cell of Ni-CPO-27.
Hydrogen bonding analysis was carried out on the optimized alanine-Ni-CPO-27 structure with QTAIM using the Critic2 software. 36QTAIM analyzes the structure and bonding in terms of the electron density ρ(r), which can be obtained either from quantum mechanical calculations or precision Xray diffraction experiments. 37According to QTAIM, points where the gradient of the electron density, ▽ρ(r), is zero are defined as critical points.In particular, first-order saddle points correspond to bond critical points (BCPs).Topological parameters at BCPs, including the Laplacian of the electron density (▽ 2 ρ(r)), potential electronic energy density (V(r)), and total electronic energy density (H(r)), are correlated with intra-and intermolecular interactions, such as polar covalent bonds and hydrogen bonds.The strength of an intra-or intermolecular hydrogen bond was estimated with the following equation 38,39 MD Simulations and Free Energy Calculations.To investigate water transport in the proposed zwitterionic alanine-functionalized Ni-CPO-27, equilibrium and nonequilibrium MD (NEMD) simulations as well as free energy The Journal of Physical Chemistry C calculations were performed.For these calculations, an alanine-Ni-CPO-27 slab was constructed via the following procedure.First, the DFT-optimized unit cell of alanine-Ni-CPO-27 was replicated by 2 × 2 × 10 to create a supercell whose dimensions are 53.25 × 53.25 × 66.7 Å in the x, y, and z directions, respectively.Each unit cell contains three channels running along the z-direction, so the supercell is composed of 12 channels in total.Then, the coordinates of alanine atoms were recorded, and the alanine molecules were removed from the supercell.The [101] surface of the Ni-CPO-27 supercell was cleaved, 40 and dangling atoms were saturated with hydrogen atoms or hydroxyl groups to create a Ni-CPO-27 slab.The removed alanine molecules were put back in accordance with their recorded coordinates.The resulting alanine-Ni-CPO-27 slab was placed in the center of the simulation box whose dimensions are 53.25 × 53.25 × 210 Å in the x, y, and z directions, respectively.Water molecules were added onto both sides of the alanine-Ni-CPO-27 slab and in its channels (Figure 1a) such that the density of water outside the slab is 1 g/cm 3 .
In all MD simulations, the Ni-CPO-27 framework atoms were kept fixed; however, coordinated zwitterionic alanine molecules were treated fully flexibly.Lennard−Jones (LJ) parameters for the framework atoms were obtained from the UFF force field. 41Partial atomic charges of the framework atoms were calculated using the REPEAT method 42 by fitting them to the periodic electrostatic potential obtained from a single-point DFT calculation.The CHARMM General force field (CGenFF) 43−45 and the TIP3P potential 46 were used to model the zwitterionic alanine and water molecules, respectively.The LJ parameters between different atom types were calculated according to the Lorentz−Berthelot mixing rules.All MD simulations were performed in the canonical ensemble (NVT) using the GROMACS software package 47 and a time step of 1 fs.The temperature was kept constant at 300 K with a Nose−Hover thermostat. 48The Coulombic interactions were computed via the Ewald sum method. 49The cut-off distance used for the LJ potential and real part of the Ewald sum was 1.2 nm.
The system illustrated in Figure 1a was first equilibrated by running a 15 ns MD simulation.Then, to compute the permeability of water as a function of external pressure, NEMD simulations were performed using the method proposed by Zhu et al. 50,51 To create 100, 200, 300, 400, 450, and 500 MPa pressures acting on the alanine-Ni-CPO-27 slab, constant forces were applied to a 3 nm water slab on the left side of the left reservoir (as shown in Figure 1a) by using the pull module in a modified version of GROMACS developed by Graẗer. 52or each pressure, a 100 ns NEMD simulation was performed and water fluxes were calculated by counting the net number of water molecules that cross the mid-point of the alanine-Ni-CPO-27 slab.The flux may be converted to osmotic permeability P f (cm 3 /s) according to the following relation where J is the water flux (number of molecules per ns), N Av is Avogadro's number, and ΔC is the osmotic gradient. 50,51The latter can be linked to the hydrostatic pressure difference ΔP (MPa) via where R (cm 3 MPa K −1 mol −1 ) is the universal gas constant and T (K) is the temperature. 50,51Substituting eq 4 into eq 3, the osmotic permeability (cm 3 mol −1 ) can be expressed as a function of the hydrostatic pressure difference and flux In practice, to compute P f for this system, we use the applied pressure on the water slab instead of ΔP; this is justified because, as we demonstrate at the end of the Results section, the osmotic permeabilities obtained by using an applied pressure or ΔP of 100 MPa are equal within statistical uncertainty.
The free energy profile, F(z), associated with water transport through the alanine-Ni-CPO-27 channels was calculated according to where ρ(z) is the water density along an alanine-Ni-CPO-27 channel as a function of the z-coordinate, ρ 0 is the bulk water density in the reservoir, and k B (kcal/K) is Boltzmann's constant.Equation 6 has been widely used to calculate the free energy profiles of water transport in aquaporins and other confined systems. 53−57 A 40 ns equilibrium MD simulation, which used the final configuration of the NEMD simulation at 100 MPa, was carried out to compute ρ(z) in 0.2 Å-wide equidistant bins along the channel direction.For each bin, the density was averaged over the 40 ns trajectory.Block averages were used to obtain the statistical uncertainties associated with the calculation of water flux and free energy profiles.The diameter of the alanine-Ni-CPO-27 channel was calculated using the PoreBlazer software. 58o characterize the ion rejection ability of alanine-Ni-CPO-27, an NEMD simulation was carried out by applying forces onto two copper walls that act as pistons on both sides of the simulation box, as illustrated in Figure 1b (referred to as the piston-NEMD simulation for the remainder of the article).Forces corresponding to 101 and 1 MPa were applied to the copper walls on the left and right sides of the simulation box, respectively, such that a 100 MPa pressure difference was created across the alanine-Ni-CPO-27 slab.The water reservoir on the left side of the slab contained 55 Na + and 55 Cl − ions, resulting in a salt concentration representative of seawater salinity, that is, 35 g/L.The copper walls prevented the ions from crossing the periodic boundary.The piston-NEMD simulation was run for 100 ns.The free energy changes corresponding to the transport of Na + and Cl − ions from the bulk water into the alanine-Ni-CPO-27 channels were obtained with a combination of umbrella sampling 59−61 and the weighted histogram analysis method (WHAM). 62,63In the umbrella sampling simulations, the system's potential is artificially biased to allow the system to sample different regions of the reaction coordinate.Data collected from each restrained state are used to calculate the potential of mean force by removing the biasing potential using WHAM.The starting configuration for the umbrella sampling calculations was the final configuration of the 100 ns piston-NEMD simulation.The sampling coordinate of the umbrella sampling calculations ranged between the surface and mid-point of the alanine-Ni-CPO-27 slab.Further details of the umbrella sampling calculations can be found in the Supporting Information.

■ RESULTS
The binding energies of zwitterionic alanine and water in Ni-CPO-27 obtained from the DFT + U calculations are −31.7 and −16.9 kcal/mol, respectively.The binding energy of The Journal of Physical Chemistry C zwitterionic alanine is almost twice that of water; thus, water molecules are not expected to replace zwitterionic alanine molecules coordinated with the open metal sites of Ni-CPO-27.The stability of the zwitterionic alanine molecules in alanine-Ni-CPO-27 is not only due to its coordination with the Ni atoms (as revealed by the binding energy calculation) but also due to hydrogen bonding.The importance of hydrogen bonding in maintaining the stability of crystalline zwitterionic alanine 64 and other AWCs has been previously reported. 65To analyze the hydrogen bonding between the coordinated zwitterionic alanine molecules in alanine-Ni-CPO-27, the electron density distribution function, ρ(r), of the DFT + U optimized structure was analyzed using the QTAIM.Figure 2 shows the hydrogen bonding between the uncoordinated oxygen atom in the carboxylate group (COO − ) and one of the hydrogen atoms in the amino group (NH 3 + ) (referred to as HB1), and the hydrogen bonding between the coordinated oxygen atom in the carboxylate group and one of the hydrogen atoms in the amino group (referred to as HB2).The corresponding topological parameters of electrons at BCPs for HB1 and HB2 are shown in Table 1.The calculated electron densities, ρ(r), for HB1 and HB2 are 0.0617 and 0.0286 a.u, respectively, which lie within the range for hydrogen bonding according to QTAIM. 66Furthermore, a positive value for the Laplacian of electron density and a negative total electronic energy density for HB1 indicate medium-strength hydrogen bonding; in contrast, a positive value for the Laplacian of electron density and a positive total electronic energy density for HB2 indicates weak hydrogen bonding. 67Using eq 2, the average bond strengths of HB1 and HB2 were calculated to be −20.5 and −7.5 kcal/mol, respectively, which are consistent with medium-strength and weak hydrogen bonds, respectively. 67In particular, the medium-strength hydrogen bonds between the COO − and NH 3 + groups help stabilize alanine-Ni-CPO-27.Figure 3a shows the single-channel pure-water flux (corresponding to J in eq 5) as a function of the applied pressure obtained from the NEMD simulations.In Figure 3b, a snapshot from the pure-water NEMD simulation shows that water forms a water wire in the alanine-Ni-CPO-27 channel, that is, single-file diffusion, which has also been reported for water diffusion in other AWCs. 18,26,54,68As shown by the line fitted through the data in Figure 3a, there is a linear relationship between the water flux and applied pressure; hence, eq 5 can be used to calculate the osmotic water permeability in alanine-Ni-CPO-27.The slope of the fitted line, that is, P f N av /RT, in Figure 3a is 5.23 ± 0.8 × 10 −4 ns −1 MPa −1 , and based on this, the osmotic permeability, P f , of water in alanine-Ni-CPO-27 was calculated to be 2.2 ± 0.3 × 10 −15 cm 3 s −1 per channel, which corresponds to 7.4 ± 1 × 10 7 molecules s −1 per channel (after dividing it by the volume of a single water molecule, which is 2.99 × 10 −23 cm 369 ).It should be noted that the TIP3P water model overestimates the experimental self-diffusion coefficient of water by about a factor of 2. 70 Figure 3c shows the single-channel osmotic water permeability of alanine-Ni-CPO-27 in comparison with other AWCs reported in the literature.Nanoporous graphene (NPG) and carbon nanotube porin (CNTP) has the highest permeabilities. 71,72The osmotic permeability of water in alanine-Ni-CPO-27 is about 50 times less than that in AQP1 (1.2 × 10 −13 cm 3 /s per channel) 73 but is close to that in the hydroxy channel (OH-channel)-based AWC (7 × 10 −15 cm 3 /s per channel) 23 and larger than that in the alkyl−ureido−ethyl− imidazole (I-quartet)-based AWC (3.3 × 10 −16 cm 3 /s per channel) and AQP0 (7.5 × 10 −16 cm 3 /s per channel).
−77 Figure 4a shows the free energy profile of water in an alanine-Ni-CPO-27 channel obtained by using eq 6 based on the water density distribution from the 40 ns equilibrium MD simulation (see Figure S3 in the Supporting Information).The free energy profile shows differences at the two water−alanine-Ni-CPO-27 interfaces because the two sides of the alanine-Ni-CPO-27 slab are slightly different due to the cleavage of the surface.On the other hand, within the channel, the free energy fluctuates about a mean value.This can be explained by the alternating hydrophobic and hydrophilic regions along the channels, as illustrated in Figure 4b.While the hydrophobicity originates from the alanine molecules, the hydrophilicity is due to the gaps between the alanine molecules.As calculated from the results in Figure 4a, the mean free energy barrier associated with water transport, ⟨ΔF water ⟩, in the alanine-Ni-CPO-27 channel is 1.44 ± 0.021 kcal/mol (c.f.k b T = 0.6 kcal/mol at 300 K).In comparison, the mean free energy barriers of water reported for AQP0 and AQP1 from MD simulations, which used the CHARMM force field for the aquaporins and the TIP3P water model, at 310 K are 1.74 and 0.42 kcal/mol, respectively. 74The differences in these calculated mean free energy barriers are consistent with the differences in the singlechannel osmotic permeabilities of alanine-Ni-CPO-27, AQP0, and AQP1 (see Figure 3c).In addition, one can show that a reasonably good correlation exists between the permeabilities and integrals of the exponential of the free energy barrier (Section S5 in the Supporting Information). 78o determine the ion rejection ability of the alanine-Ni-CPO-27 channels, a piston-NEMD simulation was carried out in which the water outside the alanine-Ni-CPO-27 slab included Na + and Cl − ions and a 100 MPa pressure difference was created across the alanine-Ni-CPO-27 slab.Figure 5a shows the number density of Na + and Cl − ions along the zcoordinate of the simulation box, that is, along the direction of the alanine-Ni-CPO-27 channels.The density profiles indicate that during the course of the simulation, neither the Na + nor the Cl − ions were able to enter the alanine-Ni-CPO-27 channels.The peaks observed near the copper piston and the alanine-Ni-CPO-27 surface are due to the interaction of the ions with the surfaces. 79,80The single-channel flux of water in

The Journal of Physical Chemistry C
alanine-Ni-CPO-27 in the presence of Na + and Cl − ions obtained from the piston-NEMD simulation is 0.049 ± 0.02 water molecules per ns.Interestingly, the piston-NEMD simulation in the absence of salt ions results in the same flux, viz., 0.049 ± 0.03 water molecules per ns.These results indicate that the presence of Na + and Cl − ions does not affect the flux of water since they do not enter the alanine-Ni-CPO-27 channels.Furthermore, water fluxes obtained from the piston-NEMD simulations with a 100 MPa pressure difference across the alanine-Ni-CPO-27 slab, with or without salt ions, are equal, within the statistical uncertainty, to that obtained from the pure-water NEMD simulation at a 100 MPa applied pressure (the first data point in Figure 3a), viz., 0.056 ± 0.023 water molecules per ns.Like water transport, permeation of ions is an activated process that depends on the free energy barriers associated with the channel.The largest free energy barriers for Na + , ΔF Na , and Cl − , ΔF Cl , as shown in Figure 5b, are 28.6 ± 1.3 and 39.6 ± 0.15 kcal/mol, respectively.In addition, the free energy profiles show fluctuations for both ions due to alternating hydrophobicity and hydrophilicity along the channel.The fact that no Na + and Cl − ions entered the slab during the piston-NEMD simulation implies that the  2), and (c) comparison of single-channel osmotic permeability, P f , of water in alanine-Ni-CPO-27 with those of other AWCs with different channel diameters, viz., hydroxy channels (OH-channel), 23 alkyl−ureido−ethyl−imidazole (I-quartet), 24 pillar [4]arene (PAH [4]), 26 aquaporin-1 (APQ1), 73 pillar [5]arene (PAH [5]), 21 porous organic cage (CC3), 25 aquaporin-0 (AQP0), 74 CNTP, 71 and nanoporous graphene (NPG).

■ CONCLUSIONS
In the present study, we consider a water-stable Ni-CPO-27 MOF, in which zwitterionic alanine is coordinated with the unsaturated Ni metal atoms as a novel design for an AWC.In comparison with the DFT + U calculated binding strength of −16.9 kcal/mol between water and Ni-CPO-27, that between zwitterionic alanine and Ni-CPO-27 is −31.7 kcal/mol, indicating that zwitterionic alanine would rarely be replaced by water molecules.In addition, the strengths of the two different hydrogen bonds formed between the carboxylate (COO − ) and amino (NH 3 + ) groups of the zwitterionic alanine molecules, HB1 and HB2, estimated by QTAIM are −20.47 and −7.5 kcal/mol, respectively, suggesting that the HB1 hydrogen bonds further enhance the water stability of this AWC.The NEMD results show that alanine-Ni-CPO-27 possesses a complete selectivity for water in the presence of Na + and Cl − ions with a water osmotic permeability of 2.2 ± 0.3 × 10 −15 cm 3 /s/channel, which is faster than AQP0 but slower than AQP1.This selectivity of the alanine-Ni-CPO-27 AWC is attributed to a low mean free energy barrier of 1.44 ± 0.021 kcal/mol for water transport and high free energy barriers of 28.6 ± 1.3 and 39.6 ± 0.15 kcal/mol for Na + and Cl − transport, respectively.By hosting an AWC in a crystalline porous material, the proposed amino acid-functionalized MOF offers a more promising design in terms of long-term stability compared to that based on the self-assembly of designed building templates in lipid bilayers or polymers.

* sı Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcc.3c02979.The Journal of Physical Chemistry C

Figure 1 .
Figure 1.Setup for the (a) pure water and (b) saline water NEMD simulations, and (c) a cross-sectional view of the alanine-Ni-CPO-27 slab, viz., in the direction of the channels.In (a,b), the light-blue regions on both sides of the alanine-Ni-CPO-27 slab represent bulk water and the applied forces are represented with green arrows.Water molecules in the channels of alanine-Ni-CPO-27 are not shown for clarity.Orange, cyan, red, blue, white, and navy-blue spheres represent copper, carbon, oxygen, nitrogen, hydrogen, and nickel atoms, respectively.

Figure 2 .
Figure 2. Atomistic representation of the DFT-optimized alanine-Ni-CPO-27 unit cell.The region circled by the black dashed line is magnified to illustrate the hydrogen bonding (HB1 and HB2) and the associated BCPs, which are represented by a purple sphere and an orange sphere, respectively.Navy-blue, red, cyan, blue, and white spheres represent nickel, oxygen, carbon, nitrogen, and hydrogen atoms, respectively.

Figure 4 .
Figure 4. (a) Free energy of water as a function of z-coordinate, that is, along the direction of the alanine-Ni-CPO-27 channel, obtained from an equilibrium MD simulation using eq 6.The surfaces of the alanine-Ni-CPO-27 membrane are located at approximately z = 7 and 14 nm.Error bars are shown in red.⟨ΔF water ⟩ is the mean free energy of water along the alanine-Ni-CPO-27 channel with respect to bulk water outside the slab.(b) Hydrophilic and hydrophobic regions in the alanine-Ni-CPO-27 channels.Color coding of the atoms is as per Figure 2.

Figure 5 .
Figure 5. (a) Number density of Na + and Cl − ions along the z-dimension obtained from the piston-NEMD simulation.The copper wall and alanine-Ni-CPO-27 slab are superimposed to indicate their locations in the simulation box.(b) Free energy profiles of Na + and Cl − ions along the alanine-Ni-CPO-27 channel obtained from umbrella sampling calculations.The free energies of the ions at the slab surface were taken as the reference and set to zero.Light-blue and gray areas correspond to alternating hydrophobic and hydrophilic regions, respectively, of alanine-Ni-CPO-27, as illustrated in Figure 4b.ΔF Na and ΔF Cl are the free energy barriers for Na + and Cl − ions, respectively.