Operando Raman Spectroscopy Reveals Degradation Byproducts from Ionomer Oxidation in Anion Exchange Membrane Water Electrolyzers

Abstract

This work showcases the discovery of degradation mechanisms for nonplatinum group metal catalyst (PGM free) based anion exchange membrane water electrolyzers (AEMWE) that utilize hydroxide ion conductive polymer ionomers and membranes in a zero gap configuration. An entirely unique and customized test cell was designed from the ground up for the purposes of obtaining Raman spectra during potentiostatic operation. These results represent some of the first operando Raman spectroscopy explorations into the breakdown products that are generated from high oxidative potential conditions with carbonate electrolytes. We provide a unique design and fabrication method for three-dimensional (3D) printable flow cells that enable spatially resolved Raman spectra collection from the electrode surface into the bulk electrolyte. It is proposed that the generation of breakdown products from the hydroxide-conductive ionomers and membranes originates from a multistep, free radical reaction pathway resulting in chain scission of the poly aryl backbone. This hypothesis is backed by the detection of carboxylic and aromatic functional group Raman signals from small molecules that had dissolved and diffused into the bulk electrolyte.

Figure 1

Renderings of the electrode-parallel cell (E-para cell) design showing the parts that were 3D printed (HER housing, OER housing, and sealing gasket were all 3D printed).

Figure 2

Renderings of the parallel cell design showing the internal assembly.

Figure 3

Renderings of the parallel cell design showing the internal schemes for electrolyte flow path, signal reflection path, and spatial gradient from electrode surface.

Figure 4

Electrochemical data for parallel operando cell testing with 3% potassium carbonate (∼0.2 M) at 400 mL/min. (a) Cyclic voltammetry demonstrates stable cycling over 4 cycles at 50 mV/s. (b) Polarization curve corrected for IR with peak current density over 60 mAh/cm² at 2 V vs Hg/HgO. (c) Short durability testing with various potential holds. A hold at 1.4 V shows virtually no degradation while in OER, whereas increasing voltage led to increasing decay rates.

Figure 5

Electrochemical data for parallel operando cell testing comparing results for 1 and 0.2 M (3% by wt) potassium carbonate electrolytes. The dotted line indicates a 10 mA/cm² reading for the comparison of the oxygen evolution reaction (OER) onset potentials. The potentials for OER onset are 1.53 and 1.28 V for the 0.2 and 1 M electrolytes respectively, or a difference in potential of 250 mV.

Figure 6

Waterfall plots for Raman signature as a function of distance from the electrode using the parallel operando cell with 3% (∼0.2 M) potassium carbonate solution at 400 mL/min. (a) Open circuit voltage (OCV), (b) 1.4 V potentiostatic hold, (c) 1.9 V potentiostatic hold, (d) 2.2 V potentiostatic hold. Peak assignments are listed in Table 1 but also listed here: Si-b indicates the silicon wafer reflector chip, Si-b is due to SiO₂ on the surface of the silicon wafer, 1 is due to carbonate (CO₃ in-phase stretch (i.ph.str)), 2 is an H₂O bend, 3 is possibly due to glycolic acid (O–C–C–O str.), 4 is an OH deformation, 5 is due to C–O, C=C, Aro/ring quad. str., or C=C–H str., and 6 is a C=O stretch. Labeling was performed in order of the appearance of evolved peaks with increasing voltage from lower to higher wavenumbers (right to left).

Figure 7

Raman spectra of different voltage holds taken from x = 30 μm from the electrode using the parallel operando cell with 3% (∼0.2 M) potassium carbonate solution at 400 mL/min. Peak assignments are listed in Table 1 but also listed here: 2 is an H₂O bend, 4 is an OH deformation, 5 is due to C–O, C=C, Aro/ring quad. str., or C=C–H str., and 6 is a C=O stretch. Labeling was performed in order of evolved peaks with increasing voltage. Peak 2 decreases with increasing voltage. Peaks 4, 5, and 6 increase with increasing voltage.

Scheme 1. Proposed Full Reaction Mechanism for Hydroxyl Radical Attack of the PiperION Membrane and the Resulting Intermediates and Products

Can be applied to ionomer (one less phenyl ring on backbone).

Figure 8

SEM images of cross sections of PiperION membranes before and after testing in an electrolyzer. (a, b) show pristine membrane images with a range of 21–28 μm thickness. (c, d) Show used membranes after 260 h of operation with 3% potassium carbonate supporting electrolyte with a range of 6–15 μm thickness. The SEM clearly indicates membrane thinning during typical operating conditions.

Figure 9

Proposed full ionomer/membrane degradation process flow.