Structural Transitions During Formation and Rehydration of Proton Conducting Polymeric Membranes

* Knowledge of the transitions occurring during the formation of ion-conducting polymer ﬁlms and membranes is crucial to optimize material performances. The use of non-destructive scattering techniques that oﬀer high spatio-temporal resolution is essential to investigating such structural transitions, especially when combined with complementary techniques probing at diﬀerent time and spatial scales. Here, a simultaneous multi-technique study is performed on the membrane formation mechanism and the subsequent hydration of two ion-conducting polymers, the well-known commercial Naﬁon and a synthesized sulfonated poly(phenylene sulﬁde sulfone) (sPSS). The X-ray data distinguish the multi-stage processes occurring during drying. A sol-gel-membrane transition sequence is observed for both polymers. However, while Naﬁon membrane evolves from a micellar solution through the formation of a phase-separated gel, forming an oriented supported membrane, sPSS membrane evolves from a solution of dispersed polyelectrolyte chains via formation of an inhomogeneous gel, showing assembly and ionic phase separation only at the end of the drying process. Impedance spectroscopy data conﬁrm the occurrence of the sol-gel transitions, while gel-membrane transitions are detected by optical reﬂectance data. The simultaneous multi-technique approach presented here can connect the nanoscale to the macroscopic behavior, unraveling information essential to optimize membrane formation of diﬀerent ion-conducting polymers.


Introduction
[3][4] In fuel cells, ion-conducting DOI: 10.1002/marc.202000717polymers are used both in thin films, as a binder for the cathode and anode materials [5,6] and as thick membranes to allow the transport of protons during redox reactions. [7]Membrane preparation involves the use of solutions or dispersions of the polymeric materials usually processed by bar coating followed by thermal annealing.Understanding the main processes occurring at the nanoscale during membrane formation is important to optimize membrane preparation and performance. [8,9]ecently, thin membranes (< 100 µm) have received increasing interest owing to their overall low resistance to ion transport. [10]esides this, thin membranes must comply with the requirements of a low fuel crossover and chemical resistance to the harsh environment of a fuel cell. [7,10,11]This requires the use of highly stable materials.Perfluorosulfonic acid ionomers (PFSI) are the benchmark polymer electrolytes for various water and energy applications. [10]The key feature that makes PFSI rather unique is the combination of a highly hydrophobic and stable poly(tetrafluoroethylene) (PTFE) backbone with flexible hydrophilic perfluorinated sulfonic side chains.[14] Neutron and X-ray scattering techniques have been extensively used to investigate the nanostructures of ionomers.A wide range of morphological descriptions have been proposed, including a network of interconnected spherical ionic domains, [15] ribbon-like and sandwich-like structures, [16] bicontinuous interconnected network of hydrophilic domains, [17] parallel cylindrical water nanochannels, [13] and locally flat layered structures of water domains. [14]The percolated locally flat layered structure has been confirmed recently by cryo-transmission electron microscopy (TEM) [18] and further supported by molecular dynamics (MD) simulations. [19]The interconnected structure under hydrated conditions and the presence of "bulk" water channels are responsible for the unique transport properties of PFSI. [11,17,20]In contrast to the excellent transport properties of PFSI, their low performance at low relative humidity (RH), the safety and environmental issues related to their production, and the use of these materials [21] paved the way for research on alternative ionomers and electrolytes.[24][25][26] Among others, sulfonated poly(phenylene sulfide sulfone) (sPSS) or sulfonated polyarylenethioethersulfone [27][28][29] represent a particularly interesting class of amorphous [30] fluorine-free proton-conducting polymer.sPSS was first introduced in 2000 [27] and shows proton conductivity higher than Nafion in a broad range of RH. [29,31] It shows potential application in proton exchange membrane fuel cells (PEMFC). [32]][35] The nanomorphology of sPSS membranes at different degrees of sulfonation and levels of hydration have been studied in bulk using small-angle neutron, [34,35] small and wide-angle X-ray scattering (SAXS and WAXS), [30] atomic force microscopy, [30] and high-resolution transmission electron microscopy. [34]The morphological features of the membranes differ depending on the sulfonation degree, but a common clustering effect of the ion domains is observed. [34,35]The more rigid aromatic structure of sPSS and the presence of the sulfonic groups directly connected to the polymer backbone reduce the segregation and the order of the ionic domains. [34]Compared to PFSI membranes, less ordered clustered structures were observed with a strong dependence on hydration level. [30,34]Also the transport properties, similar to PFSI, are affected by both RH and temperature. [30,33,34]Understanding the membrane formation of such a promising ion-conducting polymer and revealing the structural transitions occurring during membrane formation thus represent a crucial step to understand the effect of solvent, additives, ions, and processing parameters on the final film structure and morphology.Contrary to Nafion, sPSS has a much higher ion exchange capacity (IEC) and is actually a polyelectrolyte rather than a classical ionomer. [36]Thus, differences in the film formation process reported for Nafion are expected.
Here, we have investigated the membrane formation mechanism and the structural changes occurring at the nanoscale due to subsequent hydration for Nafion and sPSS using a multitechnique approach.We use the membrane formation term here to specify our focus on micrometer thick films and clearly differentiate it from already existing studies focusing on thin-film production.Grazing incidence small and wide-angle X-ray scattering (GISAXS/GIWAXS) was used to probe the nanostructural evolution occurring during solvent evaporation.At the same time, information about the resistivity and dynamics of the system is obtained using electrochemical impedance spectroscopy (EIS).Moreover, the evolution of the solvent/solid fractions is determined from in situ thickness measurements using thin-film light reflectance (LR) coupled to transfer matrix method modeling.The structural and impedance data obtained here clearly demonstrate the existence of a sol-gel transition in both cases.However, the underlying structures involved in the sol-gel transition are quite different for the two systems, reflecting their different chemical structures and polymeric aggregation behavior in solution.

Result and Discussion
The structural reorganization taking place during deposition of µm-thick membranes of two polymers cast from solutions was first studied using a combined setup allowing simultaneous measurement of GISAXS/GIWAXS, EIS, and LR signals (Scheme 1).Further details about the experimental setup are explained in the Supporting Information.

Nafion
The time evolution of the integrated GISAXS scattered intensity (or scattering power), of the system resistance, and of the film thickness are summarized in Figure 1a.The time t = 0 s is defined here as the time when the experimental measurements are started.We note that there is a 60s delay between the wet layer spreading and the beginning of the measurements.Qualitative inspection of Figure 1a reveals a first increase of the integrated GISAXS intensity along the horizontal q y and the vertical q z direction with time, reaching its maximum value at t ≈ 60 s.After this moment, a sudden decrease in the GISAXS scattered intensity occurs in the period t = 60−140 s.For t > 140 s, the integrated GISAXS intensity decreases with time at a slower rate, eventually reaching a plateau until a "dry" homogeneous membrane is formed.
Selected horizontal I(q y ) and vertical I(q z ) GISAXS/GIWAXS intensity cuts are reported in Figure 1b,c, respectively.It is clear that the scattering power changes are also paired to changes in the shape of the X-ray intensity profiles along both the q y and q z direction.
To unveil the Nafion membrane formation mechanism and quantitatively extract information on the evolution of the structural parameters during the drying process, we have fitted the experimental curves using a combined model described by Equation S1, Supporting Information, along q y and Equation S2, Supporting Information, along q z (see Supporting Information for the detailed information).The model is composed of three components: namely I polymer , I ionomer , and I upturn .I polymer describes the scattering from polymeric entities of size R polymer such as micelles in the solution and low crystalline polymeric aggregates in the membrane form. [37,38]I ionomer describes the scattering from the phase separated nanostructure of the Nafion.As recently suggested by Dudenas and Kusoglu, [39] a Teubner-Strey model originally developed for low molecular weight surfactant systems was used here, which provides the characteristic spacing (d ionomer ) and the correlation length ( ionomer ) of the ionic domains. [40]I upturn accounts for the low angle scattering generated by the presence of large-scale inhomogeneities in the membrane. [41]The scaling factors of each of the three components (I polymer (0), I ionomer (0), and I upturn (0)) are related to the product of the scatterer volume fractions and the contrast term (squared of the difference between the scattered electron density and the surrounding media).Their evolution can be followed during processing to track the temporal changes occurring in each of the structural components.A detailed description of the equations used to fit the data, and a description of all key parameters are provided in the Supporting Information.
The evolution of the structural parameters extracted by the best fit procedure is summarized in Figure 2, together with some representative GISAXS patterns.Three main stages of drying for Nafion can be observed and are discussed in detail below.
Scheme 1. Scheme of the experimental configuration used to simultaneously measure the time evolution of the film reflectance spectrum, the electrochemical impedance spectra, and the GISAXS/GIWAXS data during drying of bar coated polymer solutions.

Stage I, Micellar Aggregation (t = 0-60 s)
In the starting solution, we observe that Nafion chains aggregate to form rod-like micelles, in agreement with previous literature. [42,43]A cylindrical core-shell form factor was found to successfully reproduce the scattering of these Nafion aggregates, especially the intensity oscillation located at ≈6 nm −1 .The core is composed of aggregated PTFE backbones, while the solvated pendant chains form the shell.At the beginning of the experiment, fitting results provide a core radius of ≈1 nm with a shell thickness of ≈1.5 nm, in close agreement with published results. [44]As the initial solid content is about 10 wt%, the micelles show some non-negligible spatial correlation and a scattering bump associated with the presence of a structure factor is observed together with the micellar form factor (see Figure S2a, S2d, Supporting Information).With time, the micellar suspension rapidly evolves in the early stages of drying and the polymeric structures further concentrate as solvent molecules evaporate, which results in an increase of the term I polymer (0).The average inter-micellar spacing d polymer as calculated from the best fit evolves from ≈7.2 to ≈6 nm.The average cross-sectional radius R polymer = R core + R shell of the Nafion cylindrical aggregates decreases from ≈2.5 to ≈1.5 nm, as a result of the shrinking of the shell thickness during drying (i.e., contraction of the side pendant chains during the micellar packing phase), while the core remains almost constant at a value of about 0.9-1 nm.The system is still in the sol phase.In this time interval, no interference pattern can be detected in the reflectance spectrum of the system, indicating that a well-defined flat interface is not yet formed (see Figure S3, Supporting Information).

Stage II, Micellar Coalescence and Sol-Gel Transition (t = 60-140 s)
For t > 60 s the integrated GISAXS scattering intensity starts to decrease (Figure 1a) and the scattering signal shifts slightly towards higher q-values, indicating that the micelles get closer to each other.For t > 100 s, d polymer decreases rapidly with time, reaching a value of 4 nm at t = 135 s (Figure 2b).Concomitantly, the shape of the GISAXS profile significantly changes (Figure 1b,c).At t = 140 s, a better-defined scattering peak located at q ≈ 2 nm −1 is observed.This signal corresponds to the so-called ionomer peak, typically observed for perfluorosulfonated ionomers like Nafion. [10]This change in the scattering profiles is an indication that a transition occurs upon drying.The micelles start to coalesce and form a swollen network of Nafion chains where the ionic groups are located at the solvent/polymer interface.The characteristic length scale for the ionic domains at t = 140 s is d ionomer = 2∕q max = 3.7 nm in the in-plane q y direction and 3.3 nm in the vertical q z direction.Large-scale inhomogeneities dominate the signal at the low angles, and a clear intensity upturn appears in the I(q y ) profiles for q y < 0.5 nm −1 (see Figure 1b). [12]The sol-gel transition observed by GISAXS is well detected by EIS that measures a drastic increase in the sample resistivity R from ≈7 to ≈70 kOhm in this time range.The structural evolution reported here, from the micellar suspension to a phase-separated swollen gel, is in line with what was recently reported during the formation of a ≈100 nm Nafion thin film. [39]hus, the tenfold increase in the system resistivity R observed here can be associated with the formation of the gel phase, in which the mobility of the ionic species is drastically reduced with respect to the sol phase.

Stage III, Formation of a Swollen Membrane and Subsequent Drying (t > 140 s)
At t ≈ 140 s, a clear Fabry-Perot interference pattern is detected in the reflectance spectra of the drying system (see Figure S3, Supporting Information).This observation suggests the formation of flat and parallel interfaces between the polymer-solvent system, the substrate on the bottom, and the air on the top, indicating the formation of a swollen membrane.Figure 1a (V) shows the temporal evolution of the film thickness extracted from the interference pattern (see Supporting Information for calculations).As expected, the film thickness decreases with time upon drying.At about 265 s, a transition in the water desorption regime is detected in the optical data (See dashed line Figure S5, Supporting Information).At this time, the solvent content is ≈50% in volume with respect to the initial value (see Figure 3a).A second dramatic increase in the sample resistance is measured by EIS for t > 180 s, with the last resistivity value measured of R ≈16 M Ω at t = 280 s.At longer times, the film thickness decreases monotonically with time.In agreement with the literature, this behavior is characteristic of a two-stage mechanism where the solvent desorption occurs initially through Fickian diffusion (linear part of the curve up to 190 s in Figure 3a) followed by a non-Fickian behavior where intermolecular forces play a major role in the desorption kinetics. [45,46]Increasing the temperature above ambient value (increase to 50 °C at t = 2000 s in Figure 2) induces further film shrinkage and polymer chain aggregation.In this stage of drying, the evolution of the GISAXS signal closely follows the evolution of the film thickness.The solvent is released mostly from the ionic domains causing a decrease in the intensity of the ionomer signal I ionomer (0) and of the d ionomer value (Figure 2).In this stage, chain aggregation also occurs and the GIWAXS signal (q z > 8 nm −1 ) increases with time (see Figure S6a, Supporting Information).
The formation of a well-defined swollen polymer-air interface is also confirmed by the appearance of a peak at low q z -values, called the Yoneda peak (Figure 1c, Figure S2e, S2f, Supporting Information).The intensity of the Yoneda peak increases with time and shits to high q z -values, especially when the sample is further heated at T = 50 °C.As the Yoneda peak position along q z is associated with the critical angle of the material  c (that is related to the system density  as  c ∝ √ ), the shift towards high q-values (i.e., high scattering angles) implies that the density of the system increases upon solvent evaporation.This is expected as the bulk density of dry Nafion is about 2-2.2 g cm −3 .Measurement of the sample thickness with time allows us to explore the dynamics of the evaporation process during Nafion film formation.As described in the Supporting Information and in literature, [45,47] it is possible to correlate the normalized spectral position of any interference fringe resulting from the film with its thickness and to the solvent content with respect to the initial and final values by means of simple equations.Figure 3 shows the comparison of the normalized solvent content together with the normalized GISAXS intensities for the ionomer and polymer components as obtained from the best fit of the horizontal I norm (q y ) and vertical I norm (q z ) GISAXS intensities.Comparison between the evolution of the normalized solvent fraction (extracted from the thickness data) and the normalized signal intensity along q y and q z for the ionomer and polymer components (extracted from GISAXS modeling).b) Evolution of the ratio between the normalized GISAXS intensities and the normalized solvent fraction (I norm (q)/SC norm ).
Figure 3a shows how the GISAXS signals decrease with similar velocity with respect to the solvent content, with the exception of the I norm polymer (q y ) component, which shows a faster decay.In order to highlight differences in the dynamics, the data can be plotted as the ratio between the normalized GISAXS signals and the normalized solvent content, I norm (q)/SC norm (Figure 3b).For 140 s < t < 600 s, the ratio I norm (q)/SC norm approaches 1.This means that the macroscopic change in thickness is coincident with the nanostructural changes experienced by the swollen membrane.Regarding the I norm polymer (q y )∕SC norm ratio, its value is < 1, as a result of the faster evolution of this signal with respect to the solvent volume content.At t ≈600 s, a transition is visible in the optical data associated with a self-stress relaxation process [47] (see Figure 3a).Such relaxation does not seem to be associated with any particular change in the GISAXS signals.However, for t > 600 s, a positive deviation in the polymer signals with respect to the thickness is observed in Figure 3b, while the signals associated with the ionomer domains still follow the thickness evolution.Interestingly, 600 s is about the time when most of the GIWAXS signal evolution occurs (see Figure S6, Supporting Information).Thus, the positive deviations in the I norm polymer (q z )∕SC norm could be associated with the development of a significant degree of Nafion chain aggregation, significantly restricting the mobility of the polymer network.The earlier drying kinetic arrest of the polymer matrix along the in-plane direction has a direct impact on the membrane structure.At t = 1000 s, the d ionomer is 3.3 nm along the in-plane y-direction and 2.65 nm along the vertical out-of-plane z-direction.On the contrary, the length-scale of the ionic domains  ionomer is almost the same, 1.4 nm along the in-plane y-direction and 1.5 nm along the vertical out-of-plane z-direction.This means that the polymer layer separating two adjacent ionic domains is thicker along the in-plane direction.Obviously, the faster evolution of the polymer component along the in-plane direction does not allow the polymeric chain to pack efficiently in-plane, while more compact and better packing is achieved along the direction perpendicular to the membrane surface.It is remarkable that a very similar anisotropy in the spacing of the ionomer domains reported here for a ≈6 µm has been observed for ≈100 nm thin Nafion films. [39]he values extracted from the best fit for the spacing d ionomer and the characteristic length scale  ionomer of the ionomer domains can be used to calculate the so-called strength factor f a (Equation S7, Supporting Information).The value of f a provides information about the sample microstructure. [48,49]For lamellar systems, it is usually found that f a = −1 while for disordered systems f a = 1.Bicontinuous phases usually have −0.5 < f a < 0. As shown in Figure S7, Supporting Information, f a is about −0.9 ÷ −0.8 at ambient temperature, suggesting a good degree of ordering and a layered structure for the swollen Nafion film, in agreement with other reports. [14]This layered structure shows the tendency to orient with the layers parallel to the substrate, as deduced from the GISAXS patterns where the intensity of the ionomer signal is regrouped towards the vertical direction (see Figure 2).Nevertheless, this orientation is weak as the membrane thickness used here is quite large.When the temperature is increased above ambient temperature, f a increases, suggesting a progressive disordering of the system.
Besides the ionomer peak, contributions to the GISAXS profile from the polymeric aggregates and the large-scale inhomogeneities are also present (Figure S2b, S2c, Supporting Information).As a mild drying temperature well below the Nafion glass transition (≈100 °C) is used here, the degree of crystallinity reached by the membrane is very limited.Only a main broad GIWAXS reflection is observed located at q z ≈ 11.8-11.7 nm −1 , corresponding to the WAXS scattering peak from amorphous Nafion. [38]Thus, rather than the usually observed matrix peak, [50,51] only a weak, diffuse signal due to the aggregated Nafion chains is present in the GISAXS profiles in the range 0.5 nm −1 < q y < 1 nm −1 (Figure 1b).A clear matrix peak appears on the same membrane once annealed at 150 °C as discussed below.
After the film formation process, we have studied the hydration of the deposited film under a controlled humidity ) Variation of the ionomer peak position (q) and the related ionic domain size (d) along q y and q z directions as a function of the RH.c) GISAXS intensity cuts as a function of the RH along horizontal (q y ) direction and d) vertical (q z ) direction.Inset in the d) quadrant shows the presence of a crystalline peak (highlighted by *) in the GIWAXS intensity cut as commented in the main text.
environment.Note that before the hydration experiment, the sample was dried 12 h at 150 °C under vacuum, to further dry the film and also mimic what is usually done in the recast Nafion membrane processing.Figure 4 reports the results of the RH-controlled experiment.With respect to the Nafion film as obtained from in situ casting experiments at T = 50 °C, the GISAXS profiles now show a clear matrix knee centered at q y ≈ 0.65 nm −1 .At the same time, the GIWAXS signal along q z shows a peak at q z ≈ 12.1 nm −1 , indicating the development of crystalline domains in the deposited membrane (see inset in Figure 4d).During exposure to increasing RH, a film thickness increase is recorded as a result of the water uptake (see Figure 4a).The thickness increase is paired to a decrease in the sample resistivity (i.e., increase in ionic conductivity), in agreement with what is expected for ionomers. [11,20]At the same time, the ionomer peak shifts towards low q-values, that is, the spacing between ionic domains increases (Figure 4b).However, the evolution of d ionomer with RH is different than the thickness evolution, meaning that the behaviour observed during the in situ membrane formation is not observed during hydration of the recast annealed membranes.This can be the result of the development of crystalline domains that modify the membrane swelling behavior.Interestingly the thermal treatment has also removed the difference in the in-plane and vertical spacing of the ionomer domains.The absorption of water also causes a shift in the Yoneda peak towards lower q z -values (see arrow in Figure 4d).A shift in the peak to lower q-values (i.e., lower scattering angles) is observed as a result of a decrease in the density of the material due to water absorption (dry Nafion density is higher than water and about 2-2.2 g cm −3 ).
The results summarized in Figures 1 and 4 clearly suggest that the multi-technique experimental approach presented here is a powerful tool to explore the structure-property relationship in Nafion simultaneously at the macroscopic and nanoscopic scale.
In the next paragraphs, we will show that this experimental approach is not limited to good film forming and high scattering systems such as Nafion, but also to other alternative interesting proton-conducting systems such as sulfonated aromatic polyelectrolytes, thus extending the validity of our experimental approach.

. Sulfonated Poly(Phenylene Sulfide Sulfone)
Here we focus on the membrane formation process of a specific sulfonated poly(phenylene sulfide sulfone) polyelectrolyte (named here as sPSS) with IEC of 3.20 meq g −1 .The system is cast in the proton form from a 10 wt% solution in Nmethylpyrrolidone (NMP).A wet layer of 30 µm has been applied using blade coating similar to what was done for the Nafion solution above.
The summary of the in situ results from the simultaneous GISAXS/GIWAXS/LR/EIS measurements is presented in Fig- ure 5a.Similar to what was observed for Nafion, the evolution of the horizontal and vertical integrated scattered X-ray intensity as a function of time shows multi-stage progress, although the scattering intensity changes in a different way with respect to Nafion (panels I and II from top of Figure 2a).A first dramatic decrease in the scattering intensity is observed along the vertical q z direction during the first 1000 s of drying.On the contrary, the scattering intensity along q y shows a step-wise increase in the same time window.For longer times, a second drop in the scattering intensity is observed for a very long drying time and only after the temperature is increased to accelerate solvent evaporation.The changes in the horizontal and vertical intensity cuts at selected representative times are reported in Figure 5b,c.The difference between the shape of the scattering curves of the sPSS and Nafion is obvious.
In order to quantitatively extract the structural changes occurring during the membrane formation process of the sPSS polymer, we have fitted the experimental intensity cuts along q y using Equation S1, Supporting Information, and along q z using Equation S2, Supporting Information (see Supporting Information for detailed information on the modeling used).Similar to what was seen for Nafion, the scattering intensity along the q y direction can be well described using a three-component model.Two of them, I ionomer (q) and I upturn (q) have the same nature as those seen for Nafion.However, for what concerns the scattering of the polymeric entities, I polymer (q), a different equation must be used for sPSS.NMP is a good solvent for sPSS and the polymer is present in solution as a polymer coil, contrary to the cylindrical core-shell aggregates formed by Nafion.
The solution SAXS profile for a 5 wt% sPSS/NMP solution is reported in Figure S8, Supporting Information.The SAXS profile in the log-log plot shows a transition between a q −1 and a q −2 power-law behavior, typical for semi-rigid polyelectrolyte coils. [52] region with the q −1 slope is also observed for the coated wet layer (see Figure 5b).Thus, we have used a modified Ornstein-Zernike [52,53] equation for I polymer (q) that is able to describe scattering from semi-flexible chains of cross-sectional radius r c and Figure 6.Summary of the time evolution of the structural parameters extracted from the best fit of the a) horizontal I(q y ) and b) vertical I(q z ) GISAXS intensity cuts during film formation of an 8 µm sPSS film.Note that one point out of six has been plotted in most of the curves for clarity.Representative corrected GISAXS patterns at selected times of the membrane formation process are also reported in the top line.The intensity scale is from low intensity (blue) to high intensity (yellow) in the range of 200 to 1000 counts.
length L (Equation S14, Supporting Information).Examples of the obtained best fit curves together with the single contributions of the GISAXS model for three representative instants of the drying process are reported in Figure S9, Supporting Information.The evolution of the structural parameters extracted by fitting the GISAXS data acquired during drying is summarized in Figure 6a,b together with some representative GISAXS patterns.Four distinct stages of drying can be observed.

Stage I, Concentrated Solution Regime (t = 0-190s)
Right after coating of the wet layer, only a weak diffuse scattering is recorded at small angles (see Figure 5b,c and Figure S9a, Supporting Information).The initial sPSS concentration is ≈10 wt%.Considering that for polyelectrolytes such as sPSS, the crossover between dilute to semi-dilute solution occurs at concentrations significantly lower than the ones for neutral polymers, the initial sPSS/NMP solution can be considered as semi-dilute. [54]The sPSS chains are thus well above their overlap concentration c*.The scattering profile along q y for the sPSS chains in NMP 5 s after the deposition shows the signature of a q −1 slope in the log-log plot for 1 < q < 5 nm −1 , indicative of a semi-rigid polymer chain behavior as discussed above. [51]itting of the data using Equation S14, Supporting Information, provides a rigid chain segment length value of L ≈1.2 nm and a chain cross-sectional radius r c of ≈0.4 nm for t = 5 s (see Figure S5a, S5d, Supporting Information).Thus, the chains are not so much stretched at the beginning of the drying process.As commonly observed for semi-dilute solutions, long-range electrostatic interactions and/or large-scale density fluctuations are present, and the intensity at low angles rises steeply following a power law ∝q −m , with 2 < m < 4. [55] The crossover point q m is often used to estimate the chain persistence length or at least its lower bound limit. [52,56]For sPSS in NMP at 10 wt% we observe  = 1∕q m = 3.5 nm.In concentrated solutions (and thus also in the gel phase as discussed below), the chains overlap to form a network characterized by a random mesh, and  is often associated with the distance between chains (or mesh size). [52]As observed in Figure 6a,  decreases upon drying due to the increase in polymer concentration, as commonly observed for semi-dilute and concentrated polyelectrolyte solutions. [55]2.2.Stage II, Sol-Gel Transition (t = 190-600 s) For t > 240 s the scattering intensity at low angles along q y (<0.6 nm −1 ) described by the I upturn (q) component rapidly increases (Figure 6a).Power law fitting in the low and high scattering angle regions of the GISAXS profile for t = 600 s reveals exponents of ≈ −3 and ≈ −1, respectively (see Figure S10, Supporting Information).This kind of scattering profile is often observed for polyelectrolyte gels, where clustering of the polymeric chains at large scales has occurred. [55]Thus, the drastic increase in I upturn (0) in the time range of 190 to 600 s is associated with a sol-gel transition in sPSS.In the same temporal range, the sample resistance measured by EIS exhibits a drastic increase at t ≈490 s.Interestingly, this is also the time when the middle point of the step in the I upturn (0) versus time is recorded and where the I polymer (0) versus time curve has a minimum value (Figure 6a).Thus, similar to what was observed for Nafion, EIS captures the gel point of the system, where a drastic decrease in the ionic conductivity is recorded due to a reduction in the dynamics of the system when forming a gel phase.At the end of the gelation process, the crossover point analysis provides a value of  ≈ 2.4 nm.The length L seems to increase significantly upon crossing the sol-gel transition and at t ≈500 s a value of L ≈ 3 nm is found.This value, together with the extended q −1 regime observed at high angles, suggests chain stretching and local alignment.The sol-gel transition is also detected along the q z direction where a step-like change is observed for both the I polymer (0) component and for the GIWAXS peak I WAXS located at 12.5 nm −1 (see Figure S11a, Supporting Information).The latter term decreases since the beginning of the experiment and during the whole sol-gel transition, indicating that the GIWAXS signal contains mostly contributions from the NMP solvent.

Stage III, Formation of a Swollen Membrane and Subsequent Mild Drying (t = 600-4690 s)
After the sol-gel transition has occurred, the sPSS scattering profiles do not change significantly in shape, but only in intensity.Nevertheless, a weak Yoneda peak appears at t ≈500 s and rapidly increases in intensity with time (see Figure S11b, Supporting Information).Drying of the membrane also causes a shift in the Yoneda peak position.The appearance of a measurable Yoneda peak means that a defined swollen membrane-air interface is formed.Indeed, even if with some delay, clear Fabry-Perot interference fringes are detected in the LR profiles at t ≈670 s, confirming the formation of a well-defined flat polymer-air interface after the sol-gel transition.The film thickness determined at this moment is 8.7 µm and obviously decreases with time.Contrary to Nafion, we could not find a clear correlation between the temporal evolution of the X-ray data and the thickness for sPSS.This is most probably the result of the observed inhomogeneous structure of the system presenting clustering at large scales with the formation of regions with different densities of polymer chains and solvents.With time, the scattering intensity associated with the polymer network slightly increases, reaching its maximum at about 2200 s.The values of the correlation length L and the persistence length  practically coincide in the gel phase, as expected.For t > 2500 s, the scattered intensity gradually decreases, due to solvent molecules leaving the swollen polymer network and the value of L and also decrease, reaching values of  ≈2 nm and L ≈ 1.4 nm, respectively.

Stage IV, Final Membrane Drying and Ionic Nanostructuring (t > 4690 s)
At t ≈ 4700 s, when the temperature is increased to 70 °C in order to further remove residual NMP solvent, the scattering intensity further decreases.Interestingly, for t > 5200 s, the scattering profiles show a further evolution both along q z and q y and a clear, although weak, correlation peak centered at q y = 2,0 nm −1 is visible (see Figure 5b and Figure S9c, Supporting Information).This peak is associated with the formation of the ionic network and its position describes the average distance between adjacent ionic domains in sPSS ( d ionomer = 2∕ q max y = 3 nm).A similar signal is also visible along q z , but less clear and slightly shifted towards higher q-values (see Figure 5c).The strength factor f a initially has values close to −0.9 ÷ −0.8, but rapidly increases to f a > 0 indicating a large disorder in the dry state (see Figure S12, Supporting Information).
Similar to what was done for Nafion, we have explored the effect of hydration on the casted membranes by exposing the sample to increasing RH atmosphere after having fully dried the casted film (see Figure 7).sPSS macroscopic membrane swelling sets in when RH > 10% while no change is measured at the nanoscale by GISAXS (Figure 7a,b).
The sample resistivity remains quite high and not measurable with the used electrode assembly until RH = 75%.When going from RH = 75 to 90%, the resistivity drops more than 1 order of magnitude.At the same time, a significant increment in the ionomer peak intensity is recorded along both q y and q z .The dspacing for the ionic domains at RH = 90% is 2.7 nm along q y and 2.3 nm along q z .This value is close to what was reported for systems with similar IEC. [57]

Conclusions
In this work, we have investigated the membrane formation process occurring during casting of two different proton conducting materials, namely the well-known perfluorosulfonated Nafion and an alternative fluorine-free aromatic sulfonated polysulfone sPSS.Since the membrane formation process in these materials involves the evolution of the system at different length scales, from macro to nano, we have employed a novel multitechnique investigation method allowing simultaneous measurement of X-ray scattering under gracing incident angles, white LR, and impedance spectroscopy.The structural transitions occurring during drying at the nanoscale have been studied using Xray scattering conducted at grazing angles (GISAXS/GIWAXS).The evolution of the film thickness was measured by analyzing the simultaneously acquired optical spectra.Information about the system dynamics has been obtained by measuring the sample resistivity by EIS.
The two studied polymers are different not only in the backbone chemistry and chain architecture but also in the number of sulfonic groups per chain.Nevertheless, they share similarities in the macroscopic membrane formation process, exhibiting the sequence of transitions solution -gel phase -swollen membrane -dry membrane.
The difference in the chemical structure becomes significant when the evolution of the macromolecular chains at the nanoscale is considered.For Nafion, chains aggregate due to hydrophobic interactions of the polytetrafluoroethylene backbone, forming cylindrical core-shell micelles.Upon drying, the micelles get closer, start to interact, and coalesce to form a nanophase separated gel.The sol-gel transition guessed on the basis of the X-ray results can be confirmed by simultaneous EIS measurements.Slowing down of the system dynamics due to increased viscosity is well detected by an abrupt increase of sample resistance.The average separation between polymer layers is a function of the solvent content, determining the position of the correlation peak of the charged gel structure factor.The gel phase is characterized by a not well-defined gel-air interface, while further drying allows for the formation of a swollen membrane characterized by a well-defined polymer-air interface, parallel to the substrate and detectable by the appearance of fringes in the light interference spectra.Coupling of the thickness and X-ray information shows that Nafion exhibits affine drying, especially in the membrane thickness direction, while differences in the evolution of the polymer matrix along the in-plane and the through-plane direction result in anisotropy in the ionomer peak spacing.
Such an affine drying process is, however, not observed for the other system under study.Indeed, sPSS does not exhibit correspondence between thickness and nanoscale changes during drying.This is the result of the formation of an inhomogeneous gel phase evolving from the unstructured aggregation of semi-rigid sPSS chains, with large-scale clustering formation.Contrary to Nafion, where the ionic structure is formed during drying by a natural evolution and coalescence of the micellar structures initially present in solution, the weakly nanophase separated ionic structure of sPSS is found to develop only at the very end of the drying process, forcing the system to experience a late restructuring process.We believe that dedicated systematic studies of the membrane formation process as a function of the IEC, counter ion nature and solvent composition will be necessary to further optimize the structure and properties of these alternative fluorine-free proton conducting membranes.
The multi-technique approach presented here is a powerful tool to investigate membrane formation, hydration/drying cycling, and solvent swelling exposure of ionomers and polyelectrolytes.The experimental configuration allows exploring the processing parameters commonly used to produce cast membranes for fuel cell applications.Casting membranes with thickness below 100 µm allows reducing the overall ion transport resistance if compared to commercial membranes, making our work of interest for the polymer electrochemical community.

Figure 1 .
Figure 1.Nafion in situ drying.a) Multiple panel summary plot of the temporal evolution for the horizontal I(q y ) and vertical I(q z ) GISAXS integrated intensity (panel I and II), the sample resistivity R (panel III), the temperature (panel IV), and the film thickness (panel V). b,c) GISAXS intensity cuts along the horizontal and vertical direction during the drying experiment at selected times with related fitting curves.Note that one point out of four has been plotted for clarity.

Figure 2 .
Figure 2. Summary of the time evolution of the structural parameters extracted from the best fit of the a) horizontal I(q y ) and b) vertical I(q z ) GISAXS intensity cuts during film formation of a 6 µm Nafion film.The colored regions guide the user to visualize the duration of the different stages of drying.Note that one point out of two has been plotted in most of the curves for clarity.Representative corrected GISAXS patterns at selected times of the membrane formation process are also reported in the top line.The intensity scale is from low intensity (blue) to high intensity (yellow) in the range of 200 to 1000 counts.

Figure 3 .
Figure3.a) Comparison between the evolution of the normalized solvent fraction (extracted from the thickness data) and the normalized signal intensity along q y and q z for the ionomer and polymer components (extracted from GISAXS modeling).b) Evolution of the ratio between the normalized GISAXS intensities and the normalized solvent fraction (I norm (q)/SC norm ).

Figure 4 .
Figure 4. a) Variation of the thickness and the resistivity of Nafion film as a function of the RH.b) Variation of the ionomer peak position (q) and the related ionic domain size (d) along q y and q z directions as a function of the RH.c) GISAXS intensity cuts as a function of the RH along horizontal (q y ) direction and d) vertical (q z ) direction.Inset in the d) quadrant shows the presence of a crystalline peak (highlighted by *) in the GIWAXS intensity cut as commented in the main text.

Figure 5 .
Figure 5. sPSS in situ drying.a) Multiple panel summary plot of the temporal evolution for the horizontal I(q y ) and vertical I(q z ) GISAXS integrated intensity (panel I and II), the sample resistivity R (panel III), the temperature (panel IV), and the film thickness (panel V). b,c) GISAXS intensity cuts along the horizontal and vertical direction during the drying experiment at selected times with related fitting curves.Note that one point out of four has been plotted for clarity.

Figure 7 .
Figure 7. a) Variation of the thickness and the resistivity of sPSS film as a function of the RH, b) variation of the ionic domain size along q y and q z directions as a function of the RH.GISAXS intensity cuts as a function of the RH along c) horizontal I(q y ) and d) vertical I(q z ) directions.