Correlating solvation shell dynamics and ion transport in highly ordered nanoporous polymers Authors: Ranadeb Ball, Christopher W Johnson, Lizhu Zhang, Rebecca N Hopkins, Junkyu Hwang, Karen I Winey, Amish J Patel, Chinedum O Osuji, Jessica M Anna Published: 2026 ABSTRACT AbstractA molecular-level understanding of the role of hydration in ion transport is essential for developing next generation polymeric anion exchange membranes (AEMs). However, progress towards this goal is hindered by the structural heterogeneity of conventional membranes, complex ion-water and ion-membrane interactions, and dynamics that occur over femtosecond to millisecond timescales. We use RH-dependent 2DIR, EIS, and MD simulation to investigate the role of hydration in ion transport in a highly ordered synthetic AEM with structural uniformity. 2DIR reveals sub-diffusive ion hopping (τ2), while EIS measurements yield ion conductivity (σ) and enthalpic and entropic barriers for ion transport. A nonlinear relationship between τ2 and σ indicates strong ion-ion correlations, and a deviation from Nernst-Einstein behavior. We find that hydration primarily regulates ion hopping rates through tuning ion-pair interactions, thus influencing the activation barrier for ion transport in extremely nanoconfined AEMs. MD simulations reveal that a single-contact exchange structural transport mechanism dictates ion diffusion across all hydration levels. The associative ion hopping model based on multiple cation-anion contacts provides new design principles for efficient ion transport with improved permeability.Subject terms: Infrared spectroscopy, Chemical engineering, Molecular dynamics Introduction Polymeric ion-exchange membranes (PEMs) have propelled technological advancement in many areas, including fuel cells1–6, redox flow batteries7,8, ion separation8 and water filtration8–10. Achieving selective ion transport through PEMs is key towards harnessing them for suitable applications9,10. However, this is a difficult task as ion transport is often dictated by a complex interplay between ion-water and ion-membrane interactions that impact different processes that evolve over femtosecond to millisecond timescales. Furthermore, conventional PEMs are amorphous and tend to undergo swelling with increasing water content. Swelling results in structural changes and phase separation, leading to alteration of the spatial distance between the charged headgroups inside the pores that alter ion-water and ion-membrane interactions, which ultimately impacts ion transport and selectivity11. In contrast to the conventional PEMs, highly ordered PEMs that undergo minimal change in the transport limiting dimension, offer a unique platform for investigating the role of hydration in ion transport, where increasing water content can be decoupled from structural changes9,10,12. A better understanding of the molecular mechanism of ion transport in this newer class of PEMs at varying RHs is required for further development of next-generation high-performance PEMs. In this manuscript, we investigate the microscopic ion transport model while exploring the role of hydration in ion transport in a polymerized, highly ordered anion exchange membrane (AEM). Recent molecular dynamics (MD) simulations highlighted the role of water and membrane charged headgroups in facilitating ion diffusion in AEMs, wherein the ion diffusion follows vehicular or structural transport mechanisms depending on the hydration number, λ (number of water molecules per charged headgroup)13–18. Typically, vehicular ion transport is predominant at higher hydration numbers, where the ions move along with their solvation shell, which is mainly comprised of water molecules. In contrast, hopping-mediated structural transport typically occurs at lower hydration numbers. Experimental efforts using quasi-elastic neutron scattering (QENS)6,19,20, pulsed-field gradient NMR12,21–24, electrochemical impedance spectroscopy (EIS)12,15,24,25, and, more recently, two-dimensional infrared spectroscopy (2DIR)15,26,27, have revealed different timescales of processes that influence ion diffusion in PEMs. QENS performed on commercial membranes extracted characteristic sub-nanosecond (sub-ns) timescales for ion hopping dynamics under variable RH conditions6,19, while NMR and EIS studies provided key insights into the long-range motion associated with species diffusion and polymer dynamics12,22,24,25. 2DIR studies have provided fs-ps molecular-level dynamical information, primarily focusing on the role of water dynamics in ion transport in PEMs15,26,27. Though these studies have provided important insights into the mechanism of ion transport, they were performed on amorphous membranes and thus are inevitably impacted by the structural heterogeneity and RH-dependent swelling of the membranes11. These factors invoke a myriad of ion transport pathways in amorphous membranes, each with a non-uniform energy landscape (inhomogeneously distributed free energy of activation, ΔG‡) along the transport coordinate. Such complexities originating from nanoscale heterogeneity plague our effort to meaningfully correlate ion permeability to microscopic ion solvation shell dynamics in amorphous membranes. Recently, Osuji et al. developed self-assembled nanoporous membranes with long-range hexagonal symmetry, referred to as HEX membranes10,12. The HEX membranes, compared to typical amorphous membranes, are highly ordered with uniform sub-nanometer pore size and exhibit minimal swelling in the transport-limiting dimension as a function of hydration. This behavior makes them an ideal platform for isolating the effects of hydration on local dielectric behavior (ion-ion and ion water interactions), without significant swelling or polymer chain relaxation, granting further insight into the dynamic molecular-scale processes that govern ion transport. Here, we apply 2DIR, EIS, and MD simulations to the HEX membranes under varying RHs to solely probe the effect of hydration on the ion transport (Fig. 1a). Structural characterization and experimental measurements for probing ion transport in HEX-75. a Schematic of 2DIR spectroscopy on IR-active mobile ions incorporated in a HEX-75 membrane, where hydrogen bond dynamics and ion hopping dynamics are extracted to obtain a mechanistic understanding of ion transport. b SAXS profile of the HEX-75 membrane, showing the q100,q111 and q200 reflections at 25 °C and 30% RH, with the inset schematic of a representative HI mesophase. c Sorption profile of HEX-75 membrane with SCN− as a counter ion. Error bars indicate the standard deviation of ten minutes of measurement after equilibration. d Relative pore size ratio Sx/Sx,0 as a function of λ, with an inset picture of the hydrated clear film. e Conductivity of SCN− within the HEX-75 membrane as a function of inverse temperature and relative humidity. Error bars indicate the standard deviation of four separate measurements. f Loss angle relaxation time as a function of inverse temperature and relative humidity. g Temperature-dependent Arrhenius activation energy as a function of λ. Error bars indicate the standard error of the fit when n = 5 (or 4 in the case of λ = 1.03). Building on the previous 2DIR studies on PEMs15,26,27, which have focused on probing the ultrafast dynamics of water, we take a modified approach by directly probing the ultrafast dynamics and ion conductivity of the IR active pseudohalide ion, thiocyanate (SCN−), incorporated in a HEX membrane (HEX-75, with a transport limiting dimension of ~0.7 nm). 2DIR can selectively probe the coordination shell structure and ultrafast dynamics of IR-active species28–33. Obtaining RH-dependent 2DIR of SCN− enabled us to study the ion coordination and microscopic ion hopping dynamics of the mobile ions as a function of hydration. We found the vibrational lifetime of SCN− increased monotonically with decreasing RH, indicating RH-dependent structural changes in the ion coordination shell. Further, the spectral diffusion of SCN− evolves with two distinct temporal components: a < 10 ps component assigned to ultrafast hydrogen bond fluctuations, and a > 250 ps component associated with ion hopping, as supported by MD simulations. Contrary to conventional ion hopping mechanisms in amorphous membranes, inspection of the MD trajectories revealed a single-contact exchange mechanism dominating the ion transport in HEX-75 at all hydration levels. In addition to measuring the sub-diffusive ion hopping timescales using RH-dependent 2DIR, RH and temperature-dependent EIS were performed to extract ion conductivity, and enthalpic and entropic activation barriers. An exponential relationship was observed for bulk scale ion conductivity and the ion hopping timescales extracted from 2DIR, indicating a deviation from Nernst-Einstein (NE) behavior owing to strong ion-ion correlations under the nanoconfinement. Further, a linear relationship between the ion hopping timescales and the enthalpic activation barrier for overall ion transport (Ea) was observed at all RHs. Taken all together, 2DIR, EIS, and MD simulations reveal an associative ion hopping model across all hydration levels (λ), where both water molecules and cationic headgroups facilitate the motion of the mobile ions by ion-water and ion-ion interactions. Future AEM design requires integration of these key correlations to improve AEM performance under extreme nanoconfinement. Results RH dependent X-ray scattering & sorption profiles AEM membranes were prepared through polymerization of a doubly crosslinkable quaternary ammonium (QA+) surfactant following the procedure reported previously (Supplementary Note 1)12. We performed RH-dependent small-angle X-ray scattering (SAXS), and sorption of a polymerized AEM with SCN− as a mobile co-ion. SAXS profiles revealed a lattice parameter of 3.62 nm at 30% RH and 3.69 nm at 90% RH, indicating only 1.8% change in the transport limiting dimension as a function of RH (Supplementary Fig. 4, Supplementary Note 3). Figure 1b and d show this in the context of the original SAXS profile of an SCN− co-ion membrane. Sorption analysis performed at all RHs characterized the change in hydration number λ = [H2O]/[QA+], ranging from 1 to 7 over the range of 20% to 90% RH (Fig. 1c). Systematic RH variation without appreciable changes in the transport-limited dimension of HEX-75 implies a control over the local dielectric, and associated Bjerrum length, as a function of λ. Conductivity of SCN− Temperature-dependent conductivity of HEX-75 for a range of temperatures between 25–45 °C was measured at different RH values to extract the activation barrier for ion diffusion (Ea) as a function of λ (Supplementary Table 1, Supplementary Note 4). Figure 1e plots the ionic conductivity (σ) vs. inverse of temperature at all RHs. The dielectric relaxation time of the loss angle shown in Fig. 1f portrays the relationship between the shifting dipoles under an AC field and the interactions of the mobile co-ions. As RH increases, the relaxation time, τ0, decreases significantly due to a favorable dielectric environment, primarily settling into a region of unsolvated λ (<2) and solvated λ (>2) relaxations. The activation energy for ion diffusion was determined by the Arrhenius fit of the temperature-dependent conductivity data (σ) at constant RH values (σ=σ0exp[−EA/RT]). As λ increases, the activation energy decreases in a sigmoidal fashion, as shown in Fig. 1g. Activation energies span from 63 kJ mol−1 to 45 kJ mol−1 as a function of hydration, pointing to the role of water in the ion coordination shell on the ion transport in extreme nanoconfinement. RH dependent FTIR & 2DIR Figure 2a plots the FTIR spectra in the CN stretching region of SCN− incorporated in HEX-75 equilibrated at different RHs ranging from 11% (in red) to 100% (in black). Compared to the central frequency (νmax) of free SCN− in bulk D2O (2064 cm−1), membrane incorporation results in a redshift of ~11.5 cm−1, indicating a significant change in ion coordination under extreme nanoconfinement. We attribute this to a combined effect of the dehydration of SCN− as well as the ion-pair formation with membrane QA+ headgroups under the nanoconfinement of HEX-75. Voigt lineshape fitting of the RH-dependent FTIR spectra (Supplementary Table 2, Supplementary Note 6) shows that the full width half maxima (FWHM) value decreases as RH decreases, from a value of 34.1 cm−1 at 100% RH to 27.6 cm−1 at 11% RH, while the νmax shifts by only 1 cm−1. The change in FWHM could result from changes in the lifetime and/or an increase in inhomogeneous broadening due to the change in hydration (Fig. 1b). To deconvolute contributions from different line broadening mechanisms, we collected RH-dependent 2DIR spectra of SCN− in HEX-75. Figure 2c shows the representative waiting time (t2 in Fig. 1a) dependent 2DIR spectra equilibrated at RH of 90%, 50%, 30% and 11%. See Supplementary Fig. 7, Supplementary Note 7 for additional RH-2DIR spectra. For the fundamental CN stretching peak in the FTIR spectra, there exists a peak pair associated with the CN stretch in the 2DIR spectra. The negative amplitude peaks (blue) are associated with the ν = 0 to ν = 1 transition and assigned to the ground state bleach (GSB) and stimulated emission (SE). The positive amplitude peaks (red) arise from the ν = 1 to ν = 2 transition of the CN stretching mode and are assigned to the excited state absorption (ESA). The 2DIR spectra evolve as a function of t2 time. We extract the vibrational lifetimes from the analysis of the decay in peak amplitude and determine the timescales for spectral diffusion from analysis of the 2D line shapes. Relative humidity-dependent FTIR and 2DIR spectra of thiocyanate in HEX-75. a FTIR spectra of thiocyanate incorporated in HEX-75 equilibrated in RH = 11% (red), 30%, 40%, 50%, 60%, 70%, 90%, 100% (black). The gray dashed line shows the central frequency of SCN− in bulk D2O. b Full-width half maxima (FWHM) obtained by Voigt lineshape fitting of the FTIR spectra as a function of RH (error bars represent 95% confidence bounds). c Normalized 2DIR spectra of SCN− in Hex-75 membranes at waiting times t2 = 0.5 ps, 10 ps, 25 ps and 50 ps for membranes equilibrated at 90%, 50%, 30% and 11% relative humidity (top to bottom). RH dependent vibrational dynamics Figure 3a plots the amplitude of the ESA as a function of t2. Biexponential fitting was performed to extract the decay times (Supplementary Table 3, Supplementary Note 8). The fast component primarily arises from IVR processes. The slow component of the population decay is assigned to the vibrational lifetime. The vibrational lifetime of the confined SCN− (across all RHs) is longer than that of SCN− in bulk D2O (lifetime of 18.3 ±0.19 ps). This is attributed to the SCN− - QA+ interactions, where the interactions with the QA+ groups decrease the number of water molecules present in the first coordination shell of SCN− thus limiting relaxation pathways compared to bulk D2O. Figure 3b plots the vibrational lifetimes of SCN− in HEX-75 as a function of RH to highlight the impact of hydration on the vibrational relaxation. We find that the vibrational lifetime increases with decreasing RH. This is attributed to a decrease in the number of water molecules, which serve as energy acceptors, in the first coordination shell of SCN−. Decreasing the number of acceptors results in hindered vibrational energy relaxation34. Thus, the vibrational lifetime serves as a descriptor of the ion coordination shell in extreme nanoconfinement. Vibrational dynamics on SCN− confined in HEX-75 as a function of RH. a 2DIR ESA signal amplitude (population) is plotted as a function of waiting time, t2, between pump and probe pulses as a function of increasing RH. The gray line represents decay in bulk D2O. b Vibrational lifetimes extracted from fitting the population decays are plotted as a function of RH, along with the linear fit of R2 = 0.9771 between the two parameters. c The nodal line slope (NLS) is plotted as a function of waiting time, t2, at different RH values (11% circle, 30% star, 40% cross, 50% square, 60% diamond, 70% downward pointing triangle, 90% upward pointing triangle, 100% right pointing triangle). The biexponential fits are plotted as solid lines. d Short time constants obtained from spectral diffusion analysis represent the hydrogen bond dynamics and are plotted as a function of RH. e Long time constants obtained from spectral diffusion analysis represent the ion hopping timescales and are plotted as a function of RH. All error bars represent 95% confidence bounds in Fig. 3b, d, e. The nodal line slope (NLS) approach was employed to extract the RH-dependent spectral diffusion dynamics of SCN− incorporated in HEX-75. The NLS method was chosen as we found it to be less impacted by diagonal scatter compared to other methods. Figure 3c plots the NLS as a function of t2 for all RHs studied along with their biexponential fits (see Supplementary Table 4 for extracted parameters, Supplementary Note 9). As the amplitudes indicate, the slower component has a larger contribution to the spectral diffusion across all RH values studied. Figure 3d, e plots the fast (τ1) and slow (τ2) time constants as a function of RH. Previous studies have observed hindered hydrogen bond (HB) dynamics in confinement, where a sub-ps (<1 ps) component was assigned to HB fluctuations, and a longer 9–10 ps component was assigned to HB rearrangement27. In our work, we assign the fast τ1 timescales (5.4–7.4 ps) to HB dynamics (fluctuations and rearrangements) in the ion-coordination shell. The longer time constants (τ2) are an order of magnitude larger than the τ1 values with time constants ranging from 413 ±16 ps at RH = 11% to 256 ±21 ps at RH = 100% (Supplementary Table 4, Supplementary Note 9). Additionally, Fig. 3e shows that the τ2 timescales are more sensitive to variation in RH as compared to τ1 (Fig. 3d), indicating the role of hydration in the longer-time spectral diffusion dynamics. Based on the coordination shell structure of SCN−, with contributions from the QA+ ions and water molecules (Fig. 1a), and the extracted timescales for spectral diffusion, we assign the longer timescale spectral diffusion process to ion hopping dynamics. This would be consistent with an exchange-type of spectral diffusion process35,36. As discussed in the next section, MD simulations confirm this assignment. Ion hopping dynamics from MD simulation To obtain a microscopic picture of ion transport and gain insights into the role of water, we performed all-atom molecular dynamics simulations of HEX-75 at different λ, spanning a wide range of RH. Figure 4a shows a representative simulation snapshot, where the inset highlights the SCN− coordination shell made up of water molecules and QA+ headgroups in the nanoconfinement of HEX-75. Here, we focus on characterizing the dynamics around SCN− by monitoring the coordinating QA+ groups in the first coordination shell, with the assumption that the exchange of the QA+ groups in the first coordination shell of SCN− gives rise to the longer time scale spectral diffusion process observed in the 2DIR experiments. The exchange of the coordinating QA+ not only reflects the change of the local environment around SCN−, which contributes to the spectral diffusion in the 2D IR experiment, but also potentially relates to the SCN− hopping dynamics that result in the ion diffusion probed by EIS. Studying such dynamics can help establish a connection between experimental observables and elementary molecular processes governing ion transport. Influence of hydration on the ion hopping dynamics probed by MD simulation. a Simulation snapshot of HEX-75 with SCN− as a counterion. The surfactants and water molecules are shown in licorice representation (carbon: cyan, nitrogen: blue, oxygen: red) and SCN− ions are shown in space-fill representation. b Visualization of a representative hopping event of a selected SCN− ion: before hopping, the ion is coordinated to QA23 and QA24; after hopping, the SCN-QA23 contact is replaced by an SCN-QA25 contact. c Time series of the two nearest-neighbor QA⁺ indices for the same SCN⁻ ion. Each line segment represents a change in QA⁺ coordination. The predominance of vertical and horizontal segments indicates that, in most transitions, only one of the two coordinating QA⁺ groups changes at a given time, consistent with the proposed single-contact exchange mechanism. d Cage correlation function, C(Δt), computed using the two nearest QA⁺ neighbors at different time delays, Δt. The decay of autocorrelation reflects the exchange dynamics of QA+ groups in the coordination shell of SCN−. e The intermediate time constant, τ2,sim., from the tri-exponential fit at different λ. As λ decreases, τ2,sim increases, indicating a slower exchange of QA+ neighbors. f The ion diffusion length Ldiff=6Dτ2,sim at different λ values where D is the diffusivity of SCN− calculated from the linear fit of the mean-squared-displacement (see “Methods”). The Ldiff is smaller than Lc (red dashed line), the thickness of the QA+ coordination shell of SCN−, indicating the transport of SCN− is coupled with structural reorganization of its QA+ coordination shell. To illustrate the SCN− hopping event, we follow one representative SCN− ion (Fig. 4b). On average, each SCN− ion is coordinated by two QA⁺ groups (Supplementary Fig. 10d, Supplementary Note 11). At 9.8 ns, the ion is coordinated by QA23 and QA24. After 0.2 ns (at 10.0 ns), it transiently interacts with three QA⁺ groups in an unstable configuration that then relaxes when the SCN-QA23 interaction is replaced by SCN-QA25. In Fig. 4c, the indices of two nearest QA⁺ neighbors of the selected SCN− are shown, with QA⁺ indices labeled in the order they encounter the SCN− ion. Each line segment represents a change in the identity of the coordinating QA⁺ groups. Most segments are vertical or horizontal, indicating that during most transitions, only one of the two neighboring QA⁺ groups is replaced. This suggests that SCN− hops via a mechanism where one of the existing SCN-QA interactions is broken and replaced by a new one, while the other SCN-QA interaction stays intact. We refer to this as the single-contact exchange mechanism, as only one of the QA contacting groups is exchanged at a time. Although here we illustrate this mechanism using one chosen SCN−, in Supplementary Note 18 we showed that this single-contact exchange is the dominant hopping mechanism for all the SCN− in the simulation and across different hydration levels. To characterize the dynamics of the single-contact exchange mechanism, we computed the following two-neighbor cage correlation function36, which tracks whether both nearest QA⁺ neighbors coordinating SCN− remain the same after a time lag Δt: where ⟨…⟩ denotes a time average, NSCN is the number of SCN− ions, and δ is the Kronecker delta. Here, nnk1(t) and nnk2(t) are the indices of the first and second nearest QA⁺ neighbors of the k-th SCN− ion. The product of the Kronecker delta equals 1 if both coordinating QA⁺ groups remain unchanged after time Δt and 0 otherwise. As a result, the faster the SCN− ion transports, the faster its nearest QA+ neighbor changes as a function of t, which results in a faster decay in C(Δt). Figure 4d plots the CΔt at four different hydration levels, where we find a faster decay of CΔt with increasing λ. The correlation function was fitted with a tri-exponential decay (Eq. 2, see Supplementary Fig. 11, Supplementary Table 6, Supplementary Note 12): The shortest timescale, τ1,sim, is related to the fast and local thermal fluctuations. The longest timescale, τ3,sim, is related to the dynamics of trapped SCN− in the simulation, associated with negligible change in their nearest neighbor QA+ with time (Supplementary Note 13). The intermediate timescale, τ2,sim, corresponds to the timescale at which SCN− changes one of its neighboring QA+ and thus correlates with the ion hopping timescale. τ2,sim at different λ are shown in Fig. 4e. τ2,sim ranges from 326 ps to 410 ps, agreeing with the timescale of τ2 from the 2DIR spectral diffusion analysis. This suggests that τ2 can be assigned to the single-contact exchange-based hopping dynamics of SCN− in HEX-75. In Supplementary Note 15, we also use a random walk model with a hopping time of τ2,sim. The random walk model can reproduce the trend observed in the diffusion coefficients from the simulation, providing support that τ2,sim is related to the ion hopping that leads to the ion diffusion. In addition, τ2,sim increases as λ decreases, agreeing with the trend observed from the analysis of the 2DIR spectral diffusion. In the Supplementary Note 14, we also quantify the time scale of the fluctuation of the electric field projected on the CN bond, which is also closely related to the frequency fluctuation tracked by 2DIR. We found that as λ decreases, the electric field fluctuations also slow down (see Supplementary Table 7, Supplementary Note 14), which is the same trend observed for the QA+ neighbor dynamics. From the MD simulations, we also extracted information on how hydration impacts the mechanism of ion transport. Analysis of the neighboring QA+ group dynamics (Fig. 4b, c) indicates that ion transport is mainly structural, where the ion hops from one QA+ group to another while the second neighboring QA+ does not diffuse by any appreciable amount. To further confirm the structural transport, we compare the ion diffusion length (Ldiff) to the thickness of the coordination shell of SCN− (Lc)35,37, which is the radius of the first minimum in the radial distribution (Supplementary Note 11). Ldiff is related to the ion hopping time constant according to the following equation (Eq. 3), where D is the diffusivity of SCN− calculated from the linear fit of the mean-squared-displacement (see “Methods”). When Ldiff>Lc, the transport mode is vehicular, and when Ldiff99% pure) and D2O (>99.9% pure) were purchased from Millipore Sigma. Prior to vibrational spectroscopy experiments, the membranes with all bromide counterions were exchanged with 0.1 mM sodium thiocyanate (NaSCN) solution in D2O for 1–3 min. After the incorporation of the IR-active ions into the membrane, the membranes were subjected to an equilibration chamber for 24 h at a fixed relative humidity (RH). The equilibration time was chosen based on the time evolution of the water band in the membrane, which was probed with FTIR. To perform spectroscopic measurements, the equilibrated membrane, with a thickness of 30–40 microns, was loaded into a Harrick transmission cell with 2 mm thick CaF2 windows. The Harrick cell was sealed to avoid exposure to water from the atmosphere during the spectroscopic measurements. Conductivity measurements were performed after 72 h of solvent exchange in 1 M potassium thiocyanate (Sigma Aldrich, 99% pure) solution. Linear spectroscopy A JASCO FTIR-4600 spectrometer was used for collecting steady-state FTIR spectra at a spectral resolution of 2 cm−1. Two-dimensional infrared spectroscopy (2DIR) A 4 W commercial Ti:Sapphire regenerative amplifier (Coherent Libra) with 1 kHz repetition rate was used to generate 800 nm ~100 fs laser pulses. A portion of the 800 nm output was directed into a commercial two-stage optical parametric amplifier (Light Conversion, TOPAS C) coupled with a commercial DFG (Light Conversion) to generate ~100 fs mid-IR pulses (~18 mW) with a central frequency of ~2050 cm−1 and spectral width of ~140 cm−1. The mid-IR pulse was split by a CaF2 wedge, with 4% being used for the probe (E3) and reference pulse, and 96% was directed into a 4f-pulse shaper (PhaseTech) to generate a phase-locked pump-pulse pair (E1, E2). The pulse shaper was used for both pulse compression, along with the creation of the t1 time axis, ranging from −6 to 0 ps with 30 fs step sizes. The pulse shaper output (E1, E2) was directed to a motorized delay stage (Newport ILS250CCL) for creating the t2 waiting time between the pump pair and the probe pulse (E3). 2DIR spectra were obtained in the pump-probe geometry and heterodyne detected on the top row of a 64 × 2 MCT array detector using a FPAS spectrometer (Infrared Systems Development). A portion of the probe pulse was imaged on the bottom row of the array detector and used for multichannel referencing to increase the signal-to-noise ratio43. To remove scattering and background signals, a conventional four-frame (0, 0), (0, π), (π, π), (π, 0) phase cycling scheme was used. The excitation frequency axis was generated through a fast Fourier transform during data processing. All 2DIR spectra were collected at magic angle. Electrochemical impedance spectroscopy (EIS) DC conductivity and loss angle analyses were collected through 2-probe EIS, run on a Solartron system controlled via Modulab MTS. RH (%) and temperature control were achieved with a Thermotron 3000 Humidity Chamber. All measurements were taken after at least half an hour of equilibration time, with four separate measurements of the impedance spectra. Small-angle X-ray scattering SAXS measurements were performed on a Xenocs 2.0 Dual Emission X-ray source (DEXS) facility at the University of Pennsylvania. The sample-detector distance was set at 0.363 m, using a Cu-source with a 1.54 Å wavelength and calibrated with a silver behenate standard. Measurements at different relative humidities were performed at 25 °C in a Xenocs humidity chamber, with at least half an hour of equilibration and then a ten-minute collection under 0.7 mm ×0.4 mm collimation. Sorption analysis A TA vapor sorption analyzer was used to measure the hydration number of the material. Membranes were loaded into a metalized quartz sample pan, and measurements were performed at 25 °C across 20–90% RH under N2 flow. Samples were equilibrated for at least three hours, with the mean and standard deviation over the last ten minutes of measurement expressed as the results. Molecular dynamics simulation All molecular dynamics simulations were performed using the GROMACS 2018.6 package44. The SPC/E water model was used for all simulations45. Lennard-Jones parameters and bonded interaction parameters for the surfactant molecule were obtained from the CHARMM36 force field46. Partial charges were assigned using the CGenFF online server. The Lennard-Jones parameters and partial charges for SCN− were obtained from previous work47. A cutoff distance of 1.0 nm was used for all short-range interactions, and long-range electrostatic interactions were treated by the particle mesh Ewald algorithm48. All bonds involving hydrogens were constrained using the LINCS algorithm49. All simulations were carried out in the NPT ensemble. The Bussi stochastic velocity rescaling thermostat50 was used with a time constant of 0.5 ps to maintain the temperature of the system at 300 K. The Parrinello-Rahman barostat51 with a time constant of 5 ps was used to maintain system pressure at 1 bar. To setup the starting geometry a single surfactant molecule was first equilibrated in vacuum. According to SAXS data, the surfactant in its extended conformation is too large to fit into the cylinder in the HEX phase. As a result, we applied a biasing potential, ϕRg, on the radius of gyration, Rg, to reduce the surfactant size to be consistent with the experimental results. Here we used ϕ=1kJmol−1. Note that this biasing potential is only used for preparing the initial configuration of the surfactant molecule for HEX phase construction and is not applied in the subsequent simulations. The initial structure of a cylinder in the HEX phase was obtained by stacking 30 surfactant molecules such that the hydrophobic tails are placed close to the cylinder center and quaternary ammonium head groups are placed near the cylinder surface. The cylinder is placed so that its axis lies along the z-axis. With periodic boundary conditions, this effectively makes the cylinder infinitely long, which helps stabilize the cylinder. A single cylinder with 30 SCN− counterions was equilibrated in water for 20 ns. Then an HI unit cell was prepared by stacking four cylinders (where each cylinder has 30 counterions) in a 2D plane on a hexagonal lattice so that the center-to-center distance between adjacent cylinders agreed with SAXS. Further details regarding the system preparation can be found in Supplementary Note 17. The structure was then equilibrated for 20 ns. Validation for equilibration can be found in the Supplementary Note 1612. Ion diffusion coefficients were calculated from the slope of the time-dependent mean-squared-displacement (MSD), defined as where N is the number of SCN− ions in the system and xi(t) is the position of i-th ion at time t. The first 100 ps of the MSD were discarded to account for early time ballistic motion. Supplementary information Supplementary Information Transparent Peer Review file Acknowledgments This work was supported by the Vagelos Institute of Energy Science and Technology (VIEST) at the University of Pennsylvania. R.B. and C.W.J. acknowledge support made available through the VIEST as part of a graduate fellowship. The work reported in this paper is part of C.W.J.’s PhD thesis52. Author contributions R.B., J.M.A., C.O.O., and A.J.P. conceptualized the research. C.W.J., J.H., and C.O.O. synthesized the membranes, performed structural characterization and sorption analysis as a function of RH. C.W.J., K.I.W., and C.O.O. performed the electrochemical characterization on the membranes using RH and temperature-dependent EIS and analyzed the results. R.B., R.N.H., and J.M.A. performed RH-dependent FTIR and 2DIR and analyzed the results with the in-house codes to extract population relaxation and spectral diffusion. L.Z. and A.J.P. performed MD simulations at different hydration levels and analyzed and interpreted the results. All the authors contributed to discussion of results and writing the manuscript. Peer review Peer review information Nature Communications thanks Hesam Makki and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding J.M.A acknowledges additional funding support through the National Science Foundation (2437814). C.O.O. acknowledges additional funding support through the National Science Foundation (DMR 2223705). Work was carried out in part at the Dual Source and Environmental X-ray scattering facility operated by the Laboratory for Research on the Structure of Matter at the University of Pennsylvania (NSF DMR 2309043). Data availability All data needed to support the findings of this study are provided in the manuscript and supplementary information files. Raw data can be obtained from the corresponding authors upon request. Code availability The MD simulations were performed with the GROMACS 2018.6 package. All the necessary details related to MD simulation are provided in the method section and the GitHub link (https://github.com/lizhuzh/2026NatCommPaper). All the in-house code for MD and 2DIR data analysis can be obtained from the corresponding authors upon request. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Ranadeb Ball, Christopher W. Johnson, Lizhu Zhang. Contributor Information Amish J. Patel, Email: pamish@seas.upenn.edu Chinedum O. Osuji, Email: cosuji@seas.upenn.edu Jessica M. Anna, Email: jmanna@pitt.edu Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-73642-8. References Associated Data Supplementary Materials Supplementary Information Transparent Peer Review file Data Availability Statement All data needed to support the findings of this study are provided in the manuscript and supplementary information files. Raw data can be obtained from the corresponding authors upon request. The MD simulations were performed with the GROMACS 2018.6 package. All the necessary details related to MD simulation are provided in the method section and the GitHub link (https://github.com/lizhuzh/2026NatCommPaper). All the in-house code for MD and 2DIR data analysis can be obtained from the corresponding authors upon request.