Exploring the Structure–Performance Relationship of Sulfonated Polysulfone Proton Exchange Membrane by a Combined Computational and Experimental Approach

Sulfonated Polysulfone (sPSU) is emerging as a concrete alternative to Nafion ionomer for the development of proton exchange electrolytic membranes for low cost, environmentally friendly and high-performance PEM fuel cells. This ionomer has recently gained great consideration since it can effectively combine large availability on the market, excellent film-forming ability and remarkable thermo-mechanical resistance with interesting proton conductive properties. Despite the great potential, however, the morphological architecture of hydrated sPSU is still unknown. In this study, computational and experimental advanced tools are combined to preliminary describe the relationship between the microstructure of highly sulfonated sPSU (DS = 80%) and its physico-chemical, mechanical and electrochemical features. Computer simulations allowed for describing the architecture and to estimate the structural parameters of the sPSU membrane. Molecular dynamics revealed an interconnected lamellar-like structure for hydrated sPSU, with ionic clusters of about 14–18 Å in diameter corresponding to the hydrophilic sulfonic-acid-containing phase. Water dynamics were investigated by 1H Pulsed Field Gradient (PFG) NMR spectroscopy in a wide temperature range (20–120 °C) and the self-diffusion coefficients data were analyzed by a “two-sites” model. It allows to estimate the hydration number in excellent agreement with the theoretical simulation (e.g., about 8 mol H2O/mol SO3− @ 80 °C). The PEM performance was assessed in terms of dimensional, thermo-mechanical and electrochemical properties by swelling tests, DMA and EIS, respectively. The peculiar microstructure of sPSU provides a wider thermo-mechanical stability in comparison to Nafion, but lower dimensional and conductive features. Nonetheless, the single H2/O2 fuel cell assembled with sPSU exhibited better features than any earlier published hydrocarbon ionomers, thus opening interesting perspectives toward the design and preparation of high-performing sPSU-based PEMs.

illustrates the surface and cross-sectional SEM images of the sPSU membrane. The film exhibits a dense, compact and homogeneous structure. Furthermore, the sPSU membrane presents a smooth and uniform morphology without any cracks and/or defects. This clearly indicates fine quality. In Figure S2 are presented the structures extrapolated from preliminary molecular dynamics on polymer containing 9 chains (90 monomers, 72 SO3 groups) at 80 °C, 100 °C and 120 °C. This preliminary investigation was necessary since conformational information about the membrane were not available. 100 ns of molecular dynamics have been performed on these systems and, after analysis of the trajectories, the sampled conformations have been successively adopted as input in the creation of bigger models containing 270 residues (270 monomers, 216 groups).  Figure S3 shows the initial guess geometry adopted in further MD simulations. The structure was obtained packing 270 monomers and water molecules in a fictious box. Figure S3. Initial structure adopted in the molecular dynamics simulations. For clarity, solely water molecules within 5 Å from the -SO3 -are represented. Figure S4 presents the evolution of Root Mean Square Deviation, along the molecular dynamics, of C atoms composing the polymer. It can be note as, after an initial phase, the dynamics can be considered at equilibrium.
The values highlight that the rearrangement occurring during the simulation and discussed in the manuscript was obtained after relevant shift, respect to the initial positions, of each residue composing the monomer. The RDFs presented in Figure S5 emphasize the trend owned by the aromatic rings, composing the backbone of the polymer, to pair with other benzene-containing moieties. In particular, can be note as ring R1 and R4 establisher pi-pi pairing interaction with two and three aromatic rings, respectively, along the dynamics, further supporting the phase hydrophobic-hydrophilic separation characterizing this polymer. The model isolated from the MD simulation at 80 °C, and adopted for further DFT investigation, contains two sPSU monomers, with one SO3 group respectively and a cluster of 10 interconnected water molecules. The atoms labeled with "*" were kept frozen during the optimization, in order to prevent artificial movements of atoms composing the backbone of the polymer. The remaining atoms were left free to optimize. Figure S7 (a) shows the temperature evolution (from 20 up to 130 °C) of the 1 H spectra acquired on the sPSU membrane, for two representative water content, i.e. at saturation and 10 wt%. Proton spectra, referenced against pure water set at 0 ppm, were acquired with the same number of scans to compare their intensities. Clearly, both proton signals are very wide, implying a strong nanoconfinement of water molecules within the hydrophilic pores of sPSU. De facto, the corresponding linewidths ( Fig. S7 c) range between 700 and 1300 Hz depending on the initial water content, typically reflecting a solid-like configuration. Additionally, the signals are very asymmetric being the convolution of "different" water signals (free and bound to the SO3 -groups of sPSU) in fast rate of proton exchange. 1 Due to water evaporation from the membrane, the peak intensity progressively decreases during heating. In this regard, Figure S7d shows the temperature variation of the peak area, normalized to the initial water uptakes. For the 38-20 wt% of uptakes, the larger the decrement in the signal intensity can be seen above 60-80 °C, suggesting a greater contribution of free water (more mobile) to these NMR signals. Contrariwise, the signal loss is almost negligible for sPSU equilibrated at 10 wt% of water, indicating most of the water molecules are into a "bound state".  In Table S1 are reported the population (%) of clustered structure of each MD simulation. Among the thousands of structures encountered during the simulation, was possible to isolate most populated geometry representing more than the 50% of molecular dynamics (the cluster 1), adopting geometrical hierarchical clustering procedure. The results indicated the good conformational homogeneity reached by the polymer in the course of the simulation.

Model setup
The setup of sulfonated polysulfone model presented in the manuscript proceeded through preliminary investigations on structural behaviors of smaller polymer model, constituted by 10 monomers, reported in Scheme S1, as follows: -One terminal sulfonated molecule representing the head of the chain (HEA), presenting SO3 group; -Two molecules without SO3 group (NS3), retained to simulate the sulfonation level equal to 80%; -Six molecules presenting SO3 group (MOL); -One terminal sulfonated molecule representing the tail of the chain (TAI), presenting SO3 group.
This initial model contained 8 SO3 groups. The parameters of each residue have been extrapolated according to standard procedure mentioned in the main manuscript (see Molecular models section).
Scheme S1. Schematic representation of monomers adopted in the built up of the model.
The entire system was initially solvated, minimized, as depicted in Figure S8, and later has undergone to 50 ns of molecular dynamics in NPT conditions, at the fixed temperature of 20 °C, as analogously reported in the manuscript . The simulation reached the equilibrium, as highlighted by the RMSD plot reported in Figure S9. The last frame of the dynamics ( Figure S9C) was adopted as initial structure for the bigger model containing 9 chains and 72 SO3 groups, adopting the software Packmol. In the case of the 9 chains model, the system has undergone to a longer molecular dynamics (500 ns), at the selected temperatures of 80 °C, 100 °C and 120 °C. Results about RMSD trend and representation of selected frames are reported in Figure S10. The frames at 500 ns have been selected to create the model including 72 monomers, described in the main manuscript. Figure S10. On the left, Root mean square deviation as a function of simulation time for the 9 chain polymer model (top) and the initial structure created starting from the single chain model (bottom). Selected frames at 100ns, 250 ns and 500 ns are also depicted. For clarity, Sulphur and atoms belonging to the polymer are colored in yellow and cyan, respectively, while water molecules within 5 Å from the SO3 groups are represented as red surfaces.
In conclusion, the model set up can be summarized as follows: 1) Initially, MDs of single chain polymer (10 monomers, 8 SO3 groups) has been performed to investigate behavior in solution of the molecule, at 20 °C; 2) Starting from the equilibrated structure (last frame of MDs, at 50 ns, of point 1), a model containing 9 chains (90 monomers, 72 SO3 groups) has been created and three different MDs have been performed, at 80 °C, 100 °C and 120 °C; 3) The biggest model, consisting of 270 monomers and 216 SO3 groups, has been finally built up starting from last frames (500 ns) and adopted for extensive MDs study presented in the manuscript.