The hydration of perfluorinated sulfonic-acid ionomers is the most important phenomenon that determines their transport and electrical properties. To bridge the gap between the macroscopic electrical properties and the microscopic water-uptake mechanism, we investigated the hydration process of a Nafion membrane using ambient-pressure x-ray photoelectron spectroscopy (APXPS) from vacuum up to ∼90% relative humidity at room temperature. The O 1s and S 1s spectra provided a quantitative analysis of the water content (λ) and the transformation of the sulfonic acid group (–SO3H) to its deprotonated type (–SO3−) during the water-uptake process. Taking advantage of a specially designed two-electrode cell, the conductivity of the membrane was determined by electrochemical impedance spectroscopy before APXPS measurements with the same conditions, thereby establishing the connection between the electrical properties and the microscopic mechanism. By means of ab initio molecular dynamics simulations based on density functional theory, the core-level binding energies of O- and S-containing species in the Nafion + H2O system were obtained.
Perfluorinated sulfonic-acid (PFSA) ionomers are the most widely used solid electrolytes in proton-exchange membrane (PEM) fuel cells because of their robust chemical-mechanical stability and excellent ionic conductivity.1–3 In general, PFSA ionomers consist of a hydrophobic polytetrafluoroethylene (PTFE) backbone and a side chain containing a hydrophilic ionic group (–SO3H). The transfer of protons is realized in the water chains formed by the hydration of the ionic group (SO3H + H2O → SO3− + H3O+) in the ionomers under working conditions.3 The water-uptake of PFSA ionomers is a fundamental behavior that has received extensive attention because it determines the transport and electrical properties of the membrane.3–8
The hydration of PFSA ionomers can be quantitatively described by the concept of water content (λ), defined as the ratio of water to sulfonic acid groups.3 In the past several decades, most experimental studies on the water sorption on PFSA ionomers were based on gravimetric measurements.3,8 The measurement of λ was performed under various conditions of water vapor activity (aw) and relative humidity (RH) at a constant temperature, providing the sorption isotherm. Recently, in situ spectroscopic techniques such as infrared spectroscopy, vibrational sum-frequency generation (SFG) spectroscopy, and surface x-ray diffraction (XRD) have been used to investigate the process of water sorption on PEMs,5–7,9,10 providing information on the evolution of the water molecule structure and hydrogen bonds. To obtain quantitative information on the elemental composition and chemical state of the PEM during the hydration process, ambient pressure x-ray photoelectron spectroscopy (APXPS) has been employed to study the interaction between water molecules and the polymer surface.8 Herein, to achieve a fundamental understanding of the water sorption of PFSA ionomers, the hydration of the most widely used PEM material, Nafion membrane, was studied using a combination of the Tender x-ray APXPS system and density functional theory (DFT)-based ab initio molecular dynamics (AIMD) simulation.
The experiments were performed using a laboratory-based APXPS system.11 To obtain the ionic conductivity of the Nafion membrane during APXPS measurements, a two-electrode in situ cell was designed. Details of the sample preparation are described in the supplementary material. We have eliminated the beam-damage effect by comparing the spectra with different x-ray irradiation times (Fig. S1). Figure 1(a) showed a photograph of the two-electrode cell in the APXPS analysis chamber. Two Au electrodes on quartz substrates were fabricated by screen printing, and a schematic diagram of the front and side views of the cell is shown in Fig. 1(b). A Nafion membrane with a thickness of tens of micrometers was prepared on the two Au electrodes. Figures 1(c) and 1(d) show the O 1s and S 1s spectra of the Nafion membrane from vacuum up to ∼90% RH at room temperature (RT). The relative humidities corresponding to different pressures of water vapor in this experiment are shown in Table S1.
Under vacuum conditions, three peaks in the O 1s spectra with binding energies (BEs) of 534.9, 533.3, and 532.2 eV [Fig. 1(c)] are attributed to the perfluoropolyether side chains (CF2–O–CF), hydroxyl oxygen atoms bonded to sulfur atoms (S–OH) in protonated sulfonic acid groups (–SO3H) and the oxygen atoms double bond with the sulfur atom (S=O) in –SO3H, respectively.12,13 A single peak with a BE of 2477.9 eV was observed in the S 1s spectra [Fig. 1(d)], which can be attributed to –SO3H. Upon exposing the Nafion membrane to 1 Torr water vapor (4.2% RH), the O 1s peaks with BEs of 534.4 and 533.2 eV [Fig. 1(c)] are assigned to gas phase H2O and adsorbed H2O,14–16 respectively. The appearance of an additional peak at a BE of 531.7 eV indicated the formation of deprotonated sulfonic acid groups (S=O in –SO3−).12 A new component in the S 1s spectra [Fig. 1(d)] with a BE of 2477.4 eV also confirmed the hydration of –SO3H (SO3H + H2O → SO3− + H3O+).
As shown in Figs. 1(c) and 1(d), the intensities of the peaks from both gas and liquid H2O and the degree of hydration of –SO3H increased monotonically with a gradual increase in RH. At 30% RH, –SO3H was almost completely hydrated to form –SO3−. Upon further increasing the RH, both the O 1s and S 1s intensities of the Nafion membrane [Figs. 1(c) and 1(d)] were significantly attenuated because of the scattering of photoelectrons by water.14,17 The scattering of the S 1s photoelectrons by water molecules was more pronounced owing to their lower kinetic energy. The inelastic mean free paths of S 1s and O 1s photoelectrons in the water layer are ∼9 and ∼13 nm, respectively.18 Therefore, the signal due to –SO3− in the S 1s spectra was barely visible at RH above 75% [Fig. 1(d)].
From the ratio of the intensities due to sulfonate acid groups (–SO3H + –SO3−) and water [Fig. 1(c)], we obtained the λ of the Nafion membrane under different aw (aw = RH/100%) [Fig. 2(a)]. Using the two-electrode cell, electrochemical impedance spectroscopy (EIS) measurements were performed before APXPS experiments with the same aw (Fig. S2). The in-plane ionic conductivity (σ) of the Nafion membrane was determined using the following equation:19
where L is the distance between the two electrodes, W is the width of the electrodes (L = 1.0 mm, W = 5.0 mm), and t is the thickness of the Nafion membrane. The cross-section view of a scanning electron microscope image showed that t is 36.2 µm (Fig. S3). Based on the APXPS and EIS results, the water content and ionic conductivity of the Nafion membrane are shown as functions of aw in Fig. 2(a).
The driving force of water sorption on the Nafion membrane is the solvation energy of the ionic groups and the chemical potential. In general, there are three regimes during the hydration process: a bound water regime (λ = 1–2), a bound water + free water regime (λ = 2–6), and a bulk water like regime (λ > 6).3 As shown in Fig. 2(a), when the water activity was less than 0.15, the hydration process was in the bound water regime. At such low RH, the limited water molecules are bound to –SO3H to form hydronium ions [SO3·(H3O)+], resulting in negligible ionic conductivity of the Nafion membrane. According to the water sorption model, water domains were formed in the backbone matrix of the Nafion membrane, and the percolation threshold was reached when the λ was greater than 2.3 Our results showed a significant increase in the ionic conductivity after reaching the percolation threshold [Fig. 2(a)], with values of the ionic conductivity of the Nafion membrane that were consistent with those of previous studies.3,20
Based on a large number of experimental data, the sorption isotherm of Nafion can be described by the following equation:3
where aw is the water activity and bi are the polynomial coefficients. The relationship between λ and aw in this work can be fitted by the equation that was proposed by Springer et al.,21 λ = 36.0aw3 − 39.85aw2 + 17.81aw + 0.043 [red line in Fig. 2(a)], indicating that APXPS can provide reliable quantitative analysis of the Nafion membrane hydration processes. Moreover, our APXPS results show an inhomogeneous hydration process of the Nafion membrane [Fig. 2(b)]. As shown in Fig. 2(a), the λ is larger than 2 at RH at 21.0%. However, the –SO3H still exists in the Nafion membrane at this RH [Fig. 1(d)], illustrating an inhomogeneous distribution of H2O on the –SO3H with RH < 20% [Fig. 2(b)].22 However, there are few in situ XPS studies on the hydration process of Nafion membranes, and references on XPS spectrum analysis are scarce. To better understand and support the APXPS results, DFT-based AIMD simulations were employed to study the atomic structures and BEs of the Nafion membrane during the hydration process.
The atomic structures and BEs of the Nafion membrane were investigated using DFT-based AIMD simulations. The trifluoromethyl sulfonic acid solution (TSA-sol) model will be discussed first. This system comprised 25 TSA (CF3SO3H) and 50 H2O molecules [Fig. 3(a)], which is close to the typical conditions where λ = 2. After the AIMD simulation and optimization, the water molecules were located at the innermost part of the cluster, and the TSA was laid on the interface with hydrophobic CF3 groups facing outside and hydrophilic –SO3H groups facing inside. The above-mentioned simulated structure/morphology of the TSA-sol model was consistent with the generally accepted surface morphology of Nafion while equilibrating with water vapor, where hydrophilic ionic groups remain beneath the surface to minimize interfacial energy.23,24 The structure analysis showed that ∼80% of the –SO3H groups were deprotonated, with their protons transferred to water molecules. Two types of coordination numbers (CN) were used to distinguish the various statuses of the O atoms. CN_H is the CN of H on the O atom, and CN_S is the CN of S on the O atom. Four O statuses in various coordination environments are clearly distinguished in the 2D map (CN_H, CN_S) shown in Fig. 3(b). Two kinds of sulfonic O atoms exist in the system and are labeled SOH and SO in Fig. 3(b), where SOH is the sulfonic O atom with a bonded H atom, corresponding to S–OH in –SO3H, while SO represents the sulfonic O atom without an H atom, which exists both in –SO3H and –SO3−. In addition, two water-related O atoms exist in the system, where H2O labels the O atoms in neutral water molecules and H3O+ labels the O atoms in protonated water molecules. H2O with a larger CN_S may be close to the bound water in hydrated Nafion, while H2O with a smaller CN_S approaches free water. It is difficult to determine the core level BE difference between these two kinds of water species by simulation; identical XPS experimental results are shown in Fig. 1(c). We use the label H3O+ for clarity; however, these charged water molecules were mainly in the form of Zundel cations (H5O2+) and the CN_H of 2.5 for those O atoms.25
In Fig. 3(b), significant core-level shifts (CLSs) can be found among the different statuses of the O atoms. The SOH region is colored white, meaning that the CLSs is ∼0 eV (similar to that of the O atom in an isolated water molecule, as noted below), while the SO region is colored deep blue, indicating that its CLS is ∼−1.7 eV. The CLSs of H2O and H3O+ are ∼−0.2 and +1.2 eV, as shown in the regions colored light blue and red, respectively. The results demonstrate that the CLS of a given atom is strongly affected by the local bonding environment provided by its nearest-neighbor atoms.
The O 1s CLS of the ether O atom on the fluoride vinyl ether (FVE) backbone (CF2–O–CF) was calculated in another model system, shown in the inserted part of Fig. 3(c). The O 1s CLS is plotted as a function of distance from the center of the cluster. At the interface, the CLS reaches 1.1 eV. The main contribution to the CLS is electronic polarization, which stabilizes the ionized molecules in their environment and leads to a lower BE. Thus, the CLS of the O atoms in the inner part of the cluster is lower than that of the O atoms at the interface. The ether O atom in the real hydrated system is closer to that in the inner part of this model cluster. Therefore, the CLS of the ether O atom is taken to be +0.9 eV, which can be compared with that of the above-mentioned oxygen species in the TSA-sol model. It should be emphasized that all the species in O 1s refer to an isolated water molecule.
To allow a direct comparison between the results of the simulation and the experiment, the profiles of the oxygen species in O 1s have been calculated by convolution of the density profiles, as shown in Fig. 3(d). The broadening contributed by our APXPS system (from the x-ray source and analyzer) was measured to be 0.47 eV in FWHM.11 Though protonated and neutral water molecules can be distinguished in calculations, the corresponding O 1s BEs of H3O+ species have not been reported in water/aqueous electrolyte- or polymer-involved APXPS studies so far.8,14,26,27 The experimental relative BEs of S–OH and bound/free water are close to the calculated results. The SO species include mainly S=O in –SO3− with a small number of S=O in –SO3H, which supports the conclusion that the majority of –SO3H has been dissociated when λ equals 2. Therefore, the difference in CLS values between H2O and SO was ∼1.5 eV in the calculation, which is in good agreement with our experiments [Fig. 1(c)]. The CLS of the S atoms in sulfonic groups is shown in Fig. S4. A significant difference between the S atom CLSs in –SO3H and –SO3− can be observed, with the calculated CLS of –SO3H being ∼0.7 eV higher than that of –SO3−, while the experimental difference is ∼0.5 eV.
In this work, our APXPS results provide a quantitative analysis of the surface species of a Nafion membrane during the hydration process, in agreement with previous macroscopic gravimetric measurements. Furthermore, DFT-based AIMD simulations have provided strong support for XPS peak assignments for the different species, indicating that theoretical simulations have gradually become a powerful tool for spectral analysis. Taking advantage of a specially designed in situ cell, we were able to test the performance of the sample during in situ/operando characterization, which offered an atomic-level understanding of the structure–activity relationship in the material under study.
See the supplementary material for the preparation of the in situ cell, the measurements of APXPS and EIS, the thickness of the Nafion membrane, the methods of DFT-based AIMD simulation, and the simulation of the CLS of S atoms.
The authors appreciate the support from the National Natural Science Foundation of China (Grant Nos. 21991152, 21802096, 22072093, 21902179, and 21991150) and the Shanghai-XFEL Beamline Project (SBP) (Grant No. 31011505505885920161A2101001).
AUTHOR DECLARATIONS
Conflict of Interest
The authors have no conflicts to disclose.
Author Contributions
C.L. and J.L. contributed equally to this work.
Chiyan Liu: Data curation (equal); Investigation (equal); Visualization (equal); Writing – original draft (equal). Jian Liu: Data curation (equal); Investigation (equal); Validation (equal); Visualization (equal); Writing – original draft (equal). Yong Han: Data curation (equal); Funding acquisition (equal); Investigation (equal); Validation (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Zhangrui Wang: Data curation (supporting); Investigation (supporting). Hui Zhang: Data curation (supporting); Funding acquisition (supporting); Investigation (supporting); Writing – original draft (supporting). Xiaoming Xie: Conceptualization (supporting); Funding acquisition (supporting); Investigation (supporting); Methodology (supporting). Bo Yang: Data curation (supporting); Investigation (supporting); Writing – original draft (equal); Writing – review & editing (equal). Zhi Liu: Conceptualization (equal); Data curation (equal); Funding acquisition (lead); Investigation (lead); Resources (lead); Supervision (lead); Writing – original draft (equal); Writing – review & editing (lead).
DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding authors upon reasonable request.