Performance of Anion Exchange Membrane Water Electrolysis with High Ionic Strength Electrolyte

Self-adhesive anion-conducting ionomer preparation

OER and HER electrodes preparation

Membrane electrode assembly (MEA) preparation

Electrolysis testing

The cell configuration and electrolysis testing protocol followed a previous report. ²⁰ The current-voltage curves in this study are not iR corrected. Testing of the same cells was conducted several times and the reproducibility was less than 10 mV difference between cell repetitions, < $\pm$1%. The electrochemical impedance spectrum (EIS) of the cell was obtained galvanostatically at 1 A/cm² using SP-300 (BioLogic) after stabilizing the cells at 1 A cm⁻² for 5 min. The interfacial capacitance (C_int) was calculated with Eq. 1.

$$\begin{eqnarray}&&{C}_{{int}}=\displaystyle \frac{1}{2\pi f{R}_{{int}}}\end{eqnarray} \tag{ 1 }$$

In Eq. 1, C_int is the interfacial capacitance (F/cm²), f is frequency (Hz), R_int is the interfacial resistance (Ω cm²).

The polarization curves were recorded with a DC power supply (N5742A, Keysight) and a voltmeter. Each data point in the polarization curves was recorded by holding at a specific current density for 3 min to achieve steady-state mass transport of reactants and products.

The total cell potential (${U}_{{cell}}$) was obtained from the polarization curve and was analyzed based on reversible cell potential (${E}_{{rev}}$), activation overpotential (${\eta }_{{act}}$), ohmic overpotential (${\eta }_{{ohm}}$), and mass transport overpotential (${\eta }_{{mt}}$), according to Eq. 2.

$$\begin{eqnarray}&&{U}_{{cell}}={E}_{{rev},T}+{\eta }_{{act}}+{\eta }_{{ohm}}+{\eta }_{{mt}}\end{eqnarray} \tag{ 2 }$$

${E}_{{rev}}$ was calculated using the empirical temperature dependent expression, Eq. 3. ²⁴

$$\begin{eqnarray}&&{E}_{{rev},T}=1.229-0.9\times {10}^{-3}(T-298)\end{eqnarray} \tag{ 3 }$$

In Eq. 3, T is the absolute temperature (K). ${\eta }_{{act}}$ was calculated using an extrapolated Tafel plot fitted in low current density region (5 to 50 mA cm⁻²) where there were no significant ohmic and mass transport losses using the Tafel relationship Eq. 4.

$$\begin{eqnarray}&&{\eta }_{{act}}=b/{10}^{3}\times \mathrm{log}\left(\frac{i}{{i}_{0}}\right)\end{eqnarray} \tag{ 4 }$$

In Eq. 4, b is the Tafel slope (mV/dec), $i$ is the current density (A/cm²), and ${i}_{0}$ is the exchange current density (A/cm²). ${\eta }_{{ohm}}$ was calculated from the high frequency resistance (HFR) in the EIS spectrum using Ohm's law, Eq. 5.

$$\begin{eqnarray}&&{\eta }_{{ohm}}={HFR}\times i\end{eqnarray} \tag{ 5 }$$

${\eta }_{{mt}}$ was calculated by subtracting ${E}_{{rev}}+{\eta }_{{act}}+{\eta }_{{ohm}}$ from ${U}_{{cell}}.$

Cation transference number measurement

The cation transference number in the solid polymer electrolyte was evaluated by a modified Bruce-Vincent method. ²³ The drop in current from its initial value to a steady-state value was recorded. Hydroxide ions are formed at the cathode by water reduction and oxidized at the anode by the oxidation of hydroxide to oxygen. Thus, a steady state gradient in hydroxide ions was formed where all other ions are electro-inactive. The anion conductive membranes contained mainly immobilized quaternary ammonium cations tethered to the solid polymer electrolyte, however, there are mobile cations due to salt permeation from the electrolyte. A perturbation in the cell voltage causes gaseous products produced at the electrodes by electrolysis. The cation transference number, ${t}_{{M}^{+}},$ can be calculated by Eq. 6.

$$\begin{eqnarray}&&{t}_{{M}^{+}}=1-{t}_{O{H}^{-}}=1-\displaystyle \frac{{I}_{SS}\left({\rm{\Delta }}{V}_{mod}-{I}_{0}{R}_{int,0}\right)}{{I}_{0}\left({\rm{\Delta }}{V}_{mod}-{I}_{SS}{R}_{int,SS}\right)}\end{eqnarray} \tag{ 6 }$$

In Eq. 6, ${t}_{O{H}^{-}}$ is the anion transference number, ${I}_{0}$ is the initial current (A), ${I}_{SS}$ is the steady-state current (A), ${R}_{int,0}$ is the initial interfacial resistance (${\rm{\Omega }}$), ${R}_{int,SS}$ is the steady-state interfacial resistance (${\rm{\Omega }}$), and ${\rm{\Delta }}{V}_{mod}$ is the modified perturbation voltage, which is equal to the applied voltage subtracted by ${E}_{rev,T}$ (V).

It has been previously shown that an alkaline electrolysis using a low hydroxide concentration anolyte (pH ∼ 12) suffers significant ohmic and mass transport losses at the cathode. ²⁰ Increasing the anolyte cation concentration (independent of the hydroxide concentration) assisted in water delivery to the cathode through an AEM. The cathode overpotential was lowered when the anolyte ionic strength increased, at constant anolyte pH. The improvement in performance was likely due to the improvement in water transport from the anode to the cathode. It was also shown that the salt rejection of the AEM is not perfect and a measurable amount of NaNO₃ from the anolyte was found in the AEM. Only sodium cations were used (0.05 M to 1.0 M) in the previous study. The present study investigates the effect of different anolyte cations and transport through the AEM on cathodic electrolysis performance.

AEM electrolysis with different anolyte cations was investigated. Figure 1 shows the effect of Li⁺, Na⁺, K⁺, and Cs⁺ in the anolyte on the electrolysis overpotential. It was previously found that 0.15 M Na⁺ provided a near optimum electrolysis performance when 0.01 M NaOH was also contained in the anolyte. The cathode overpotential decreased with the addition of any alkali cation, compared to only 0.01 M NaOH, as shown in Fig. 1a. At high current density, K⁺ was the highest performing alkali cation and Li⁺ was the lowest with the order of performance: K⁺ > Na⁺ > Cs⁺ > Li⁺. In Fig. 1a, there is a small change in the activation overpotential at low current density (i.e., < 0.1 A cm⁻²). This shows that the overpotential with K⁺ was also the lowest at low current density (<0.1 A cm⁻²). Huang et al. analyzed the effect of different alkali cations on the HER exchange current density for the hydrogen oxidation reaction (HOR) through rotating disk electrode (RDE) experiments. ²⁵ It was found that the HER/HOR exchange current density was cation-dependent, showing that Li⁺ had the highest exchange current density, followed by Na⁺, K⁺, and Cs⁺. When the interfacial water molecules were replaced by a bulkier cation, the reorganization energy and entropy of the hydrogen-bonded water structure increased resulting in a change of the solvation configuration at the interface. However, in this study, K⁺ had the smallest activation overpotential, indicating that in the dry HER cathode water content is also involved but may be a limiting reagent. It is also noted that the 0.01 NaOH anolyte has the lowest (most favorable) anode overpotential and the overpotential became worse as a nitrate salt was added to the anolyte. This is reasonable because at least some of the anode catalyst is covered with ionomer. A majority of the anions in that ionomer would be nitrate which lowers the hydroxide concentration in the ionomer and shifts the equilibrium anode potential to more positive (less favorable) values.

Zoom In Zoom Out Reset image size

The overpotential with K⁺ was also the lowest at high current density (1 A cm⁻²), compared to other cations tested (Fig. 1a). The overpotential at high current density includes the effect of ohmic and mass transport resistances. In this study, the anode and cathode overpotentials were investigated separately because the anolyte affects both electrolysis performance. ²⁰ Figure 1b shows the anode potential (top four curves) and cathode potential (bottom four curves) vs pRE as a function of current density. The potential difference between the anode and cathode in Fig. 1b corresponds to the full cell potential in Fig. 1a. The anode potential remained nearly constant vs current density for each anolyte because the pH in each electrolyte was the same. An expanded scale graph of Fig. 1b is shown in Fig. s2. The largest single change in the half-cell potentials occurs at the cathode where there was a decrease in cathode overpotential with the addition of anolyte salt, even though the anolyte is in contact with the anode only. Further, there is a significant improvement in cathode overpotential (i.e., lower overpotential) when the ionic strength of the anolyte was increased. This is likely due to improved water transport to the cathode with high cation concentration in the anolyte. ²⁰ Water diffuses to the cathode catalyst, and ions migrate away from the cathode. Unlike the anode ionomer, the cathode ionomer plays a critical role in ion and water transport because the electrode is operated dry (i.e., the ionomer is the only pathway for water and ions between the membrane and catalyst surface). Of the four alkali cations tested, K⁺ shows the lowest cathode overpotential at high current density (i.e., 1 A cm⁻²). It is noted that the alkali ion trend correlates with the mobility of the ions.

Figure 1c shows the Nyquist plots for the different anolytes. The extracted impedance values are summarized in Table I. A single out-of-phase component (near semi-circle interfacial impedance) was observed for the 0.01 M NaOH anolyte. The difference between the low frequency x-intercept and high-frequency x-intercept is called the interfacial impedance here. When the salt was added to the 0.01 NaOH electrolyte, the out-of-phase component is broken into two contributions composed of a cathode (higher frequency loop) and an anode (lower frequency loop) interfacial impedance. ^20,26 The ohmic series resistance, high frequency resistance (HFR), decreased with added anolyte salt due to the improved ionic conductivity. The HFR for the five anolytes were in the following order (smaller is better): Cs⁺ < K⁺ < Na⁺ < Li⁺ < none. The bulkier cations have a smaller water solvation shell and weaker interaction with water which leads to higher ionic mobility and water diffusivity. Notably, the anolyte with K⁺ showed the lowest overall cell voltage.

Table I. Impedance analysis of 0.01 M NaOH and high ionic strength electrolytes with 0.15 M of LiNO₃, NaNO₃, KNO₃, and CsNO₃.

	HFR (${\rm{\Omega }}$ cm2)	Rint (${\rm{\Omega }}$ cm2)	Cint (F/cm2)	${E}_{{overall}}$ @ 1 A cm−2 (V)
0.01 M NaOH	0.263	0.205	0.011	2.11
0.01 M NaOH + 0.15 M LiNO3	0.215	0.165	0.061	1.88
0.01 M NaOH + 0.15 M NaNO3	0.198	0.182	0.055	1.854
0.01 M NaOH + 0.15 M KNO3	0.186	0.136	0.066	1.802
0.01 M NaOH + 0.15 M CsNO3	0.177	0.173	0.033	1.876

The interfacial resistance (R_int) decreased (i.e., smaller is better) in a slightly different sequence than the HFR: K⁺ < Li⁺ < Cs⁺ < Na⁺ < none. R_int consists of the charge transfer and mass transfer resistances. The interfacial capacitance (C_int) is related to the total effective catalyst surface area (i.e., bigger is usually better): none < Cs⁺ < Na⁺ < Li⁺ < K⁺. A greater water flux delivered by the cation via migration increases the effective electroactive surface area of the catalyst. This leads to a decrease in the charge and mass transfer resistance at the three-dimensional cathode because the catalyst is more effectively used. ²⁷ The trend of R_int and C_int shows that the cation and water mass transfer improvement and HER electroactive surface area are improved with addition of a more mobile anolyte cation.

A graphical summary of the activation, ohmic and mass transfer overpotentials of cells with 0.01 M KOH (left side) and 0.01 M KOH + 0.15 M KNO₃ (right side) are shown in Fig. 2. The use of additional K⁺ in the anolyte improves the ohmic, activation, and mass transport overpotential compared to only use of 0.01 M KOH. Other cations also show results consistent with the K⁺ results in Fig. 2 (see Fig. S1). Notably, most of the anolyte improvements are derived from the mass transport enhancement due to the augmented water delivery. ²⁰ These improvements all appear tied to the mobility of the hydrated.

Zoom In Zoom Out Reset image size

The mobile cation concentration in the AEM is small due to salt incorporation from the anolyte. The quaternary ammonium cations within the solid polymer electrolyte are tethered to the cross-linked polymer backbone, however, anion exchange membranes do not perfectly reject other cations. In this case, the Na⁺ concentration in the AEM was found to be 0.147 meq g⁻¹. ^20,28 If there were no mobile alkali cations within the AEM, only tethered quaternary ammonium chain segment mobility would contribute to cation transport within the polymer. The cation transference number was measured in this study by using the Bruce-Vincent method to investigate the trend in transference number with cation type. ²³

The methodology here is analogous to the original Bruce-Vincent method except for using electroactive hydroxide. Figure 3 shows the measurement protocol with 0.01 M KOH. First, linear sweep voltammetry was used to check the applied voltage needed to induce both OER and HER in the cell, Fig. 3a. The current initiated at 1.4 V generates hydroxide at the cathode and consumes hydroxide at the anode. Figure 3b shows chronoamperometry (CA) data (see inset) and Nyquist plots at the beginning and steady state portion of the CA at 1.5 V. The Nyquist plots show an out-of-phase element at 1.5 V, which incorporates the HER and OER. The characteristic time constant (${\tau }_{{char}}$) was calculated by an equation (${\tau }_{{char}}$ ∼ ${L}^{2}$/(2D₊), where L is the thickness of electroactive layer and AEM, and D₊ is the cation diffusivity) to determine the time period needed to approach a steady-state for to determine the cation transference number. Using the K⁺ diffusivity of 1.96 × 10⁻⁹ m²/s in water and 150 $\mu$m AEM thickness yields a time constant of about 10 s. ¹³ In the CA experiments, the current approached a steady state value in several seconds.

Zoom In Zoom Out Reset image size

The transference number for each alkali cation in the anolyte at 0.01 M and 0.1 M concentration was obtained by the Bruce-Vincent method (Table II). ²³ When the anolyte concentration was 0.01 M, the cation transference number was 0.13 to 0.16. When the anolyte concentration was increased to 0.1 M, the cation transference number increased to 0.20 to 0.23. The higher concentration of alkali ion in the anolyte increases the alkali cation concentration in the AEM, leading to a higher cation transference number because the ratio of mobile alkali cations to immobilized cations (i.e., pendant quaternary ammonium) increases as the anolyte concentration increases. It was previously reported that the IEC of Na⁺ in AEMs after soaking it in DI water increased after soaking it in 0.01 M NaOH + 1.0 M NaNO₃ electrolytes. In DI water, there were no sodium ions detected in the AEM. The IEC of sodium ions in the AEM when soaked in 0.01 M NaOH + 1.0 M NaNO₃ was 0.147 meq g⁻¹. There was also a measurable sodium ion concentration in DI water after soaking an AEM tested in an electrolysis with 0.1 M NaOH anolyte for 40 h at 1 A cm⁻². ²⁰ This previous result and the cation transference numbers in Table II show that an AEM does absorb salt from the anolyte (i.e., AEM salt rejection is not perfect), and cation migration and diffusion occur in the AEM during electrolysis. Cs⁺ shows the highest cation transference number, however, the electrolysis voltage (i.e., cathode overpotential) of the cell with Cs⁺ was not the lowest. This result shows that water transport is not only related to the cation mobility, but also affected by the cation hydration number.

The size of the hydrated alkali ions is tabulated in Table III. The small ionic radius for lithium ions results in a larger hydration shell and larger overall dynamic radius, which lowers its ionic mobility compared to the other alkali ions. Potassium ions have the smallest radius resulting in a relatively high ionic mobility, compared to other alkali ions shown. A pictorial representation of the ionic radius and solvent shell is shown in Fig. 4. Although all four alkali ions improve the cathode performance, as shown in Fig. 1b, potassium ions show the best performance, which correlates with the smallest dynamic ionic radius and high mobility.

Zoom In Zoom Out Reset image size

Lower pH (∼12) water electrolysis may contribute to improved AEM durability by suppressing metal corrosion and polymer hydrolysis. Using the optimal electrolyte with additional K⁺ (i.e., 0.01 M NaOH + 0.15 M KNO₃), the long-term electrolysis durability was investigated at 1 A cm⁻² (Fig. 5). Electrolysis with 0.01 M NaOH + 0.15 M KNO₃ shows a very small increase in voltage during 100 h of electrolysis. Once a voltage plateau was developed, the cell voltage was very stable and maintained a value of 1.732 V for 800 h. This corresponds to 46.4 kWh/kg_H2 of energy input and ∼85%_HHV energy efficiency. The voltage slightly increases after 800 h with a marginal degradation slope of ∼40 $\mu$V/hr. Thus, an alkaline water electrolysis with low pH high ionic strength electrolyte and optimal K⁺ salt concentration may be beneficial for the long-term electrolysis durability due to the better chemical stability at lower hydroxide concentration electrolyte, while preserving low electrolysis overpotential from the alkali ion water delivery mechanism from the anode to the cathode.

Zoom In Zoom Out Reset image size

Table III. Ionic radius (r_ion) ^a), hydration shell width ($\unicode{x02206}$ r_hyd) ^a), and dynamic ionic radius (r_dyn) of alkali cation. Dynamic ionic radius is sum of ionic radius and hydration shell width (r_dyn = r_ion + $\unicode{x02206}$ r_hyd).

Alkali cation	Ionic radius, rion (nm)	Hydration shell width, $\unicode{x02206}$ rhyd (nm)	Dynamic ionic radius, rdyn (nm)
Li+	0.069	0.172	0.241
Na+	0.102	0.116	0.218
K+	0.138	0.074	0.212
Cs+	0.170	0.049	0.219

^a)Ionic radius and hydration shell width were found in Ref. 27.

AEM alkaline water electrolysis performance with high ionic strength electrolytes comprising different alkali cation was investigated. The anode and cathode overpotential were examined in a three-electrode electrolysis cell using a pRE. The cathode overpotential was significantly improved when an additional 0.15 M of alkali cation was added into the 0.01 M NaOH anolyte. Cathode overpotential significantly changed while the anode overpotential remained relatively constant. The high ionic strength electrolyte with K⁺ (i.e., 0.01 M NaOH + 0.15 M KNO₃) had the lowest overall electrolysis overpotential among the tested cations. The alkali cation transference number was measured with an AEM alkaline water electrolysis cell to investigate the effect of mobile cation mobility on electrolysis performance. The cation transference number increased for the more mobile cations. Cs⁺ had the highest cation transference number among the ones tested. K⁺ had the smallest dynamic ionic radius resulting in the high mobility and most effective water transport. The optimal high ionic strength electrolyte (i.e., 0.01 M NaOH + 0.15 M KNO₃) had very good alkaline water electrolysis durability likely due to the more moderate pH in the high ionic strength electrolyte.