The highest current efficiency (CE) for all three membranes was in the low LiOH concentration range (0.5M). More specifically, Sx-2301-Wn recorded the highest CE of 89.2% ± 1.1%, while Flemion had a slightly lower efficiency of 88.4% ± 1.0% (Figure 3A). With increasing base concentrations in the catholyte, a trend of decreasing CE was observed with a minimum at 1.7M of 81% ± 2.5% for Sx-2301-Wn and 77.3% ± 2.4% for Flemion. The lower CE is related to the higher OH⁻ back migration over the membrane due to the increased gradient between catholyte and middle compartment. The pH, initially 5.1, gradually decreased over time in the feed compartment and reached at the end of the test 1.7 and 1.83 for Flemion and Sx-2301-Wn respectively. This indicated indicating that proton leakage was dominant over hydroxide back migration, even though the concentration of the anolyte never surpassed 0.5M H₂SO₄. At pH 1.7, protons start to compete slightly with Na⁺ and Li⁺ and decrease the CE.

Sx-2301-Wn recorded the lowest cell voltage, especially in the higher concentration ranges/towards the end of the tests (Figure 3B). This is related to the lower membrane resistance, which stayed constant throughout the test (Table 2). Flemion had a slight increase in resistance (0.068 Ω cm²), while Sx-2301-Wn remained constant. This could be attributed to increase in resistance in the carboxylated layer, which is only present in the Flemion membrane. If this layer is contacted with an acidic solution, the carboxylic functional groups become protonated and lose their charge (Balster et al., 2004). This reduces the IEC capacity of the membrane, increases the water content and leads to more hydroxide back migration. This was tested separately by measuring the membrane resistance after soaking the membranes for 72 h in 0.2M H₂SO₄. Supplementary Table S2 confirms the hypothesis and shows the increased resistance of 13.859 (Ω cm²) for Flemion, while Sx-2301-Wn (0.513 Ω cm²) and CMI-7000 (0.615 Ω cm²) showed a slight decrease in resistance. The lower CE and increased membrane resistance led to a higher energy consumption (E_sec) for Flemion (12.49 ± 0.84 kWh/kg LiOH) compared to Sx-2301-Wn (11.87 ± 0.27 kWh/kg LiOH).

CMI-7000 had a significantly lower CE, both at high and low concentrations and had a higher standard deviation compared to the other CEMs due to a large difference in CE between both tests. The cell voltage between both tests also varied largely and CI data show a decrease in cell resistance (0.114 Ω cm²) before and after the tests. Therefore, the loss in performance is attributed to a loss in functionality of the membrane, and thus should not be used for concentrating LiOH. The lower chemical stability compared to the other CEM is related to the polymer backbone, which is a polystyrene-divinylbenzene (Sty-DVB) polymer and is non-fluorinated. Besides, it also had the highest internal resistance, partially caused by its higher thickness, cell voltage and thus highest E_sec (14.42 ± 1.69 kWh/kg LiOH).

High LiOH concentrations in the catholyte are desired, as they decrease the intensity of the evaporation and crystallization steps. Therefore, Flemion F-9010 and Sx-2301-Wn were selected to be subjected to a more concentrated LiOH solution (3M), while the anolyte volume and concentration was kept constant to minimize the effect of proton leakage. The same membranes were used as in the low concentration tests.

Figure 4A shows that the highest CE recorded was 56.5% ± 0.3% for Sx-2301-Wn at 3.7M LiOH, while Flemion had a slightly lower CE of 53.4% ± 1.4%. Both membranes have however a significantly lower CE compared to the low concentration LiOH tests, and also a slightly higher cell voltage. The water migration for both membranes is also significantly lower for the high concentration tests (Flemion F-9010: −49%, Sx-2301-Wn: −33%, Supplementary Table S2), and is related to the lower Li⁺ and Na⁺ migration. The higher catholyte concentration also decreases the water content in the membrane, but did not improve the hydroxide blocking ability of the membranes.

The membrane resistance data for Flemion (0.513 Ω cm²) and Sx-2301-Wn (0.477 Ω cm²) show a decrease in resistance compared to the pre-testing measurements (Table 3) and to the low concentration tests in Section 3.1.1 (Table 2). Membranes were soaked in 0.5M Li₂SO₄ and had therefore a lower cation content than the membranes after the tests with 3M LiOH. This increased the conductivity of the CEM and decreased its resistance. Flemion did not experience an increase in resistance, compared to the low concentration tests, as the pH in the middle compartment was 13.31 ± 0.03 at the end of the test, indicating that the carboxylic layer did not get protonated. Also for Sx-2301-Wn, the pH in the middle compartment was 13.20 ± 0.04, showing that at high catholyte concentrations and low anolyte concentrations, hydroxide back migration is dominant over proton leakage. The cell voltage for Sx-2301-Wn was however approximately 0.4 V higher on average compared to the low concentration tests, while the CEM resistance was lower. The cause for this could be found in the difference in feed concentration and pH and by increased AEM resistance and were further investigated in the experiments described below. The difference in cell voltage for both Flemion tests could be related to an accumulation of impurities inside of the membrane, to which these are very sensitive to (Momose et al., 1991). Further testing is required to confirm this.

Overall, Sx-2301-Wn had a lower E_sec (16.5 ± 0.1 kWh/kg LiOH) compared to Flemion (18.2 ± 1.4 kWh/kg LiOH), but the E_sec increased drastically compared to the low concentration tests. This indicates that the membranes suffer from the increased concentration gradient between catholyte and middle compartment. In the chlor-alkali industry, a maximum efficiency is observed around 7.5M NaOH in the catholyte. At lower catholyte concentrations, the membrane contains too much water, which enables hydroxide back migration through the tunneling mechanism, and above the maximum, the absorbed electrolyte is not fully balanced by the membrane’s counterions, leading to an increased mobility in anions (Gronowski and Yeager, 1991). Increasing the concentration of LiOH in the catholyte could therefore decrease the water content of the membrane even more and increase the CE, but due to the limited solubility of LiOH (5.34M at 20°C), it is not possible to increase the concentration to similar levels as for NaOH. The sulfate content in the LiOH for Sx-2301-Wn was measured and was only 0.41 wt% and 0.44 wt% for low and high concentrations.

Figure 5A shows that Fumasep had the highest CE (85.7% ± 0.5%) at 0.2M H₂SO₄ compared to Selemion (66.1% ± 2.2%). At 0.6M, Fumasep is still 15% more efficient than Selemion, but seems to be converging towards Selemion. The medium concentration tests were done with a lower anolyte volume (0.5L instead of 1.5L, Supplementary Table S1) and confirmed the converging trend (Figure 5B), with a difference of 13.5% at 0.7M H₂SO₄ and only 3.5% at 1.3M H₂SO₄. The water migration was higher for Fumasep compared to Selemion at both low (Fumasep: 70 ± 6 mL, Selemion 20 ± 6 mL) and medium (Fumasep: 0 mL, Selemion −20 mL) acid concentration. This is related to the higher co migration of water in the hydration shell of SO₄ ^2– over the AEM, but also to the lower proton leakage (and water). However, the water migration with anion migration was significantly lower than the water migration for cation migration (Supplementary Table S3).

The high water migration for Fumasep at low concentrations compared to Selemion can be attributed to the chemistry of the membrane. High ion exchange capacity (IEC) of the membrane tends to increase the water content of the membrane, but also its capacity to migrate anions at low acid concentrations (Jaroszek et al., 2017; Wang et al., 2018). However, at higher acid concentrations, these membranes tend to leak more protons and experience a decrease in CE. The membranes with lower IEC and higher degree of cross-linking or hydrophobicity tend to have better performances at higher acid concentrations (Yu et al., 2022). Based on the CE data from the low and medium concentration tests, one could assume that Fumasep has a higher IEC and water uptake, favoring its performance at low concentrations, while at higher concentrations it gets outperformed by the more hydrophobic Selemion membrane.

Selemion had a slightly lower cell voltage compared to Fumasep at low concentrations (Figure 6A), but were approximately equal at medium concentrations. CI data for both Fumasep and Selemion showed to be very high in neutral environments (5.348 Ω cm² and 27.48 Ω cm²), but much lower in acidic conditions (1.072 Ω cm² and 1.113 Ω cm², Table 4). For this reason, the membrane resistance was measured with the process solutions at the end of the test (Fumasep: 0.64M H₂SO₄, Selemion: 0.61M) and showed that Selemion had a lower resistance than Fumasep, which can explain the slightly lower cell voltage compared to Fumasep. No significant difference in CEM resistance was measured between different AEM tests. Overall, this resulted in a E_sec of 4.99 ± 0.71 kWh/kg H₂SO₄ for Fumasep, compared to 6.79 ± 0.22 kWh/kg H₂SO₄ for Selemion to produce 0.6M H₂SO₄.

Both Aemion + membranes (AF3-HWC9-70-X and AF3-HWK-100-X) had a significantly lower proton blocking ability, as CEs of 46.1% ± 1.5% and 49.5% ± 0.1% were recorded at 0.2M H₂SO₄ (Figure 5A). At elevated concentrations, the efficiency dropped even to 21.5% ± 3.7% and 30.4% ± 0.7% respectively. The pH in the middle compartment was, at the end of the test, 1.45 ± 0.02 and 1.61 ± 0.03 respectively, indicating that the membranes had a higher proton leakage to the middle compartment compared to the Fumasep and Selemion membrane at low concentration. On the contrary, the membrane resistance of both membranes from Ionomr Inc. was drastically lower than Fumasep and Selemion, in both neutral and acidic conditions (Table 4). The lower membrane resistance can be related to the higher density in functional groups, which increase the conductivity of the membrane, but also to the poly-imidazolium functional group, which provides a high chemical stability against hydroxide attack and is used in alkaline water electrolysis (Moreno-González et al., 2023). This lower membrane resistance results in a significant lower cell voltage of 0.7–1.0V compared to Fumasep and Selemion. The lower cell voltage could however not compensate the higher proton leakage, and resulted in a higher energy consumption of 9.58 ± 0.27 kWh/kg H₂SO₄ and 9.53 ± 0.22 kWh/kg H₂SO₄ for AF3-HWC9-70-X and AF3-HWK-100-X respectively. Due to the low current efficiency at 0.6M H₂SO₄ for both membranes, these were not tested further in higher H₂SO₄ anolyte concentrations.

Additionally to the measurements, one could also investigate the lifetime and degradation throughout the harsh conditions of the acids and oxidative species like O₂ and Cl₂, which can sometimes be present as an impurity (Mei et al., 2018). The reproducibility of the measurements of the AEM do not indicate any degradation of the membranes, but require further and longer testing to confirm this.

For the reuse of acids in leaching processes, it is desired to have a high acid concentration, as this increases the leaching efficiency of metals from battery scrap (Jung et al., 2021). Therefore, Fumasep and Selemion were tested with an initial anolyte concentration of 3M H₂SO₄. The CE for both membranes was very low and insufficient for any industrial application (Figure 7). Fumasep even had a negative CE of −1.0%, indicating that more H⁺ migrated towards the middle compartment than SO₄ ^2– towards the anolyte. Selemion did better with a CE of 19.6% and final H₂SO₄ concentration of 3.3M, but is still far from efficiency levels of lower concentrations. The water migration was negative for both Fumasep (−15 ± 2 mL) and Selemion (−5 ± 2 mL) because of the high proton leakage over the AEM. The membrane resistance of Selemion was lower after the tests compared to Fumasep in 0.5M Na₂SO₄, but did not differ much in 3M H₂SO₄ (process solutions) (Supplementary Table S4). The lower cell voltage for Selemion can be found in the difference in resistance of the CEM (Flemion), which was less protonated compared to Fumasep because of the lower proton leakage and migration towards the catholyte (CEM Fumasep: 0.599 Ω cm² vs. CEM Selemion: 0.327 Ω cm²) (Supplementary Figure S2; Supplementary Table S5). Overall, the E_sec of Selemion was 10.93 ± 0.79 kWh/kg H₂SO₄ and is significantly higher than the E_sec at low concentrations. No E_sec was obtained for Fumasep, as the membrane was not capable of concentrating acid at this anolyte concentration.

The addition of LiOH to the middle compartment improved the CE for OH⁻production in the catholyte by reducing the H⁺ competition against Li⁺ and Na⁺ (Figure 8). At the end of the test in the uncontrolled EC with Fumasep, the pH was 1.38, resulting in a higher H⁺ concentration (0.042M). The choice of AEM did not impact the OH⁻efficiency significantly for the pH controlled tests due to the elimination of the effect of proton leakage, and only a slight difference of 2.3% in CE was measured for the uncontrolled test. For Fumasep, at low LiOH concentration (0.5M), no LiOH is dosed and the CE is very close between both tests (pH control: 81.5 ± 1.3; no pH-control: 79.6). At higher concentrations (1.5M), LiOH addition becomes larger and contributes to a higher CE (75.4% ± 1.9%) compared to the uncontrolled EC (71.2%). Above 2M LiOH, the difference for Fumasep 8%. Selemion showed a similar trend.

At low acid concentrations (0.3M H₂SO₄), the CE efficiency of the Fumasep EC with pH control was similar to the uncontrolled EC, as no LiOH was added yet (Figure 9). At higher acid concentrations (0.93M), the pH controlled test clearly outperforms the non-controlled one (63.6% ± 0.7% vs. 57.4%). The difference in CE between both configurations remains 5.1% and 5.0% at 1.3 and 1.6M H₂SO₄ and slightly converges again above 2M H₂SO₄ (3.8%). The difference in migration efficiency can be attributed to the speciation of H₂SO₄ at pH 2 and at pH 1.5 or below. At pH 2, the middle compartment contains a 1:1 ratio of SO₄ ^2–/HSO₄ ⁻, while at pH 1.5 the solution has a 1:4 ratio (Supplementary Figure S3). The electric mobility, the ability of charged ions to move through a medium under an electrical field, of SO₄ ^2– is higher than HSO₄ ⁻ due to its divalent nature (Koter and Kultys, 2008). Based on this principle, one could argue that pH 3 or higher is the optimal pH range because 91% of the anions are HSO₄ ⁻. Increasing the pH too much in the middle compartment could however lead to an increased membrane resistance and higher cell voltage. This was verified by measuring the membrane resistance after soaking the membanes in 0.5M Li₂SO₄ pH 5, 3 and 1 (Supplementary Table S6). The membrane soaked in pH 5 had the highest resistance when measured in both Na₂SO₄ (2.746 ± 0.036 Ω cm²) and H₂SO₄ (1.226 ± 0.022 Ω cm²), while pH 1 had the lowest resistance when measured in Na₂SO₄ (1.878 ± 0.043 Ω cm²). When measured in 1M H₂SO₄, pH 3 (0.882 ± 0.023 Ω cm²) had a slightly lower resistance than pH 1 (0.929 ± 0.008 Ω cm²). Increasing the pH of the feed to alkaline ranges can also decrease the lifetime of the AEM, as some functional groups, especially trimethylammonium-based, are prone to hydroxide attack, irreversibly losing its functionality (Marino and Kreuer, 2015; Ramírez et al., 2018). One can thus conclude that raising the pH in the middle compartment too much would increase the cell voltage, counteract the increased CE and potentially decrease the lifetime of the AEM.

The increase in CE due to the pH control did not result in a significant decrease in E_sec (Fumasep: 9.4 ± 1.1 kWh/kg H₂SO₄, Selemion: 9.9 ± 1.1 kWh/kg H₂SO₄) compared to the uncontrolled EC (Fumasep: 9.87 kWh/kg H₂SO₄, Selemion: 10.15 kWh/kg H₂SO₄). No large differences were detected in AEM or CEM resistance, but a higher conductivity in the middle compartment was detected for the uncontrolled ECs compared to the pH controlled ECs (Supplementary Figure S4). In the former, the pH decreases to pH 1.4 and has a higher proton concentration (0.042M) compared to the pH controlled EC (0.01M), where the Li⁺ content is only slightly higher (7%) due to the addition of LiOH. The large difference in conductivity at the end of the tests suggests that the increased proton content results in a lower cell voltage and thus a lower E_sec, counterbalancing the gains in CE or increased membrane resistance.

The net OH⁻production efficiency was calculated for Fumasep and showed that the addition of LiOH to the middle compartment did not result in an increased OH⁻production (Figure 10). At low concentrations (until 0.6M H₂SO₄, no LiOH was added and both configurations had similar efficiencies. Above 0.6M, LiOH addition increased the CE of LiOH generation in the catholyte, but a large amount of LiOH needs to be recycled, which leads to a lower overall OH⁻production efficiency at higher concentrations. Above 1.6M H₂SO₄, the difference became even more pronounced and results in a slight flattening of the curve of the net OH⁻production. The E_sec was therefore higher (17.3 ± 0.2 kWh/kg LiOH) compared to the uncontrolled (14.98 kWh/kg LiOH) and showed that adding base did not improve the overall E_sec. A better approach could be to balance the concentrations of the catholyte and anolyte so that the CE of proton leakage and hydroxide back migration are balanced and the addition of LiOH can be avoided.