Various States of Water Species in an Anion Exchange Membrane Characterized by Raman Spectroscopy under Controlled Temperature and Humidity

Anion exchange membrane fuel cells (AEMFCs) hold the key to future mass commercialisation of fuel cell technology, even though currently, AEMFCs perform less optimally than proton exchange membrane fuel cells (PEMFCs). Unlike PEMFCs, AEMFCs have demonstrated the capability to operate independently of Pt group metal-based catalysts. Water characterization inside the membrane is one factor that significantly influences the performance of AEMFCs. In this paper, different water species inside an anion exchange membrane (AEM), QPAF-4, developed at the University of Yamanashi, were studied for the first time using micro-Raman spectroscopy. Spectra of pure water, alkaline solutions, and calculations based on density functional theory were used to identify the water species in the AEM. The OH stretching band was deconvoluted into nine unique Gaussian bands. All the hydrogen-bonded OH species increased steadily with increasing humidity, while the CH and non-H-bonded OH remained relatively constant. These results confirm the viability of micro-Raman spectroscopy in studying the various water-related species in AEMs. The availability of this technique is an essential prerequisite in improving the ionic conductivity and effectively solving the persisting durability challenge facing AEMFCs, thus hastening the possibility of mass commercialisation of fuel cells.


Density Functional Theory
a. Methods A representative section of the QPAF-4 structure that included both hydrophobic and hydrophilic moieties was used for the density functional theory DFT calculations.The hydrophilic moiety included two tetraalkylammonium groups, each with an associated hydro ide anion, and 20 water molecules were randomly placed to surround these functional groups and provide a hydrogen-bonded network between them Figure S6 .First, this structure was subjected to a molecular dynamics MD calculation in order to obtain a realistic arrangement of the water molecules.Then, two stages of geometry optimisation were carried out, first with medium quality and second with fine quality, as described below in more detail.The final optimised structure was then subjected to vibrational analysis of both OH -anions, the 20 water molecules, and representative parts of the ionomer itself.The vibrational frequencies of the various O-H contributions were finally used to generate a spectrum.Normalized Intensity a.u.The simple Nosé-Hoover thermostat for 200 fs at 300 K was used 5,6 .The final configuration of the MD calculation was then subjected to geometry optimisation, first with medium settings 2  10 -5 Ha energy convergence, 4  10 -3 Å gradient convergence, displacement convergence 5  10 -3 Å, scf density convergence 1  10 -5 Ha , and then with fine settings 1  10 -5 Ha energy convergence, 2  10 -3 Å gradient convergence, displacement convergence 5  10 -3 Å, scf density convergence 1  10 -6 Ha .Both were all-electron calculations and used the v. 4.4 basis file.The vibrational frequencies were also carried out with fine settings.To generate the spectrum, the individual bands were subjected to Gaussian broadening factors of 1 cm -1 minimal broadening and 200 cm -1 realistic broadening .

b. Results
The calculated Raman spectrum contains some of the features of the e perimental spectrum but overemphasises the high wavenumber region.The main reasons for this are a all of the bands have been given equal intensity in the calculation, whereas it is well known that the intensity should decrease with increasing wavenumber of O-H stretching as the degree of hydrogen bonding decreases; and b in the calculation, there would be a disproportionate number of free OH groups, i.e., those with essentially zero H-bonding, with resulting frequencies in the 3700-3800 cm -1 region.There are several predicted vibrations in the low-wavenumber region that are due to water OH groups that are Hbonded to hydro ide o ygen.As e plained in the main te t, although these bands might be observed e perimentally, they are difficult to distinguish from interference features.

Figure S1. 1 . 0 M
Figure S2.16.0 M NaOH spectrum deconvoluted into seven Gaussian bands Figure S3.Change in the area under each of the five deconvoluted OH sub bands within the OH band of NaOH with changing molar concentration.

Table of Contents 1. Supporting Experimental Results
8.4969.
CalibrationFigureS6.The spectra of de-hydrated QPAF-4 membrane on silicon substrate at different temperatures.