As the name implies, direct liquid fuel cells (DLFCs) operate directly on a liquid fuel instead of on hydrogen as in proton exchange membrane fuel cells. DLFCs have the advantages of higher energy densities and fewer issues with transportation and storage of their fuels compared with compressed hydrogen, and are adapted to mobile applications1. Among DLFCs, direct borohydride hydrogen peroxide fuel cells (DBHPFCs) are fed by sodium borohydride (NaBH4) as the fuel and hydrogen peroxide (H2O2) as the oxidant. As BH4 and H2O2 are only stable in high2 and low3 pH environments, respectively, the operation of DBHPFCs must rely on a very efficient separator to prevent cross-over of the anolyte and catholyte. Writing in Nature Energy4, Vijay Ramani and co-workers from Washington University in Saint Louis report a nanoengineered bipolar membrane interface that effectively regulates the local pH and mass transport of the reactants, enabling high-performance and durable DBHPFCs.

Today’s best practical fuel cell membranes are widely considered to be cation-exchange membranes (CEMs)5. However, under an applied current, a CEM-based DBHPFC cell (Fig. 1a) would allow Na+ migration, thereby displacing the NaOH supporting electrolyte (an essential BH4 stabilizer) from the anode to the cathode, which lowers the cell performance. In the open-circuit condition, H+ would diffuse to the anode and consume both OH and BH4. Although anion-exchange membranes (AEMs) have improved tremendously in the past decade6, they cannot impede the anions (OH, BH4 and SO42-) cross-over (Fig. 1b), and are therefore also not ideal for DBHPFCs. Bipolar membranes (BPMs) combine a CEM and an AEM, and can offer some unique advantages for fuel cell applications, but they suffer from sluggish ion transport, are also often ill-defined and have unstable CEM|AEM junctions7,8.

Fig. 1: Operation of a DBHPFC in a CEM-based and an AEM-based cell.
figure 1

a, In an all-CEM cell, Na+ migration under current and H+ diffusion under open-circuit conditions could take place, owing to the improper separation of the anode/cathode compartments, leading to parasitic formation of H2O and NaBO2 at the anode and of Na2SO4 at the cathode. b, In an all-AEM cell, SO42− migration could proceed under current, and OH, BH4 and SO42− could all diffuse under open-circuit conditions, leading to parasitic formation of Na2SO4 at the anode and H2O and H3BO3 at the cathode. OCP, open-circuit potential; j, current density.

In the study reported in Nature Energy, the researchers built an asymmetric BPM, consisting of a 50–180-μm-thick commercial CEM that is mechanically strong, highly proton conducting and BH4 impermeable, and a very thin (~8 nm) AEM that serves as a proton barrier, which is then placed in the vicinity of an anode electrocatalyst (Fig. 2). This unique configuration creates a sharp pH-gradient-enabled microscale bipolar interface (PMBI), enabling an effective separation of the anolyte and catholyte at the electrocatalytically active sites. This effect, cleverly demonstrated with recessed planar electrode model experiments, leads to unprecedented open-circuit voltage (1.95 V) and current density (330 mA cm−2 at 1.5 V and 630 mA cm−2 at 1.0 V), which are beyond reach in classical fuel cells. Moreover, a short-term stability of 50 hours for the DBHPFC was demonstrated. By tuning the thickness of their CEM support and the loading of Pd/Ni-foam anode electrocatalyst, the researchers further demonstrated that their DBHPFC performances could approach those of state-of-the-art proton exchange membrane fuel cells.

Fig. 2: Representation of the different processes taking place at the PMBI under current flow.
figure 2

a, An autoprotolysis of water takes place at the AEM|CEM interface, where H+ cannot be transported in the AEM and OH cannot be transported in the CEM. b,c, In open-circuit conditions at the AEM|CEM interface, diffusion of anions in the AEM and of cations in the CEM yield the consumption of acid of the catholyte and base from the anolyte into water (b) and formation of boric acid (H3BO3) from BH4 reaction with OH and H+ that diffuse in the AEM and CEM, respectively (c). d, Details of the possible formation of Na2SO4 in the catholyte via the cross-over of Na+ through the thin AEM, either under current or at open-circuit conditions. Figure adapted from ref. 4, Springer Nature Ltd.

While the study of Ramani and team provides an interesting proof-of-concept of using tailored BPM interfaces to boost fuel-cell performances, several issues remain. First, and importantly, BPMs can operate efficiently only when the CEM and AEM moieties enable fast water transport and autoprotolysis7. As shown in Fig. 2a, H2O species must be transported to the CEM|AEM junction region, where they should decompose into H+ and OH to ensure transport of current in the CEM and AEM portions of the BPM, respectively. However, these processes in the PMBI are not well understood and in particular it remains to be seen how to accelerate the autoprotolysis of water to reach high current densities.

Second, Ramani and colleagues assumed perfect BPM permselectivity in their discussion of ion transport (that is, only cations should migrate in the CEM and only anions in the AEM). However, in practical DBHPFCs, both the anolyte and catholyte could diffuse within the ‘electrolyte soaked’ BPM. So, one cannot rule out that operating the PMBI would consume acid from the catholyte and base from the anolyte, to form water in the junction region (Fig. 2b). This is more likely to be the case in the open-circuit condition, where the concentration gradient between the anolyte and the catholyte could lead to ion diffusion. As such, if the energy density of the DBHPFC were to be calculated, the calculation should be based not only on the fuel and oxidant molecules, but also on the whole anolyte and catholyte.

Third, to increase the fuel/oxidant Faradaic efficiencies, any parasitic reactions, such as the BH4 hydrolysis in Fig. 2c, should be avoided. Note too that Na+ can diffuse through the thin AEM portion and the thick CEM portion, and combine with SO42− moieties of the anolyte to form another parasitic product (Na2SO4) (Fig. 2d). Although a thicker AEM could potentially prevent these problems, the overall BPM resistance would be affected. Last but not least, the work of Ramani and team could benefit from further development of the electrocatalysts used at the two electrodes, which have not yet been fully optimized.

Nevertheless, the PMBI of Ramani and colleagues provides a new and fascinating design to engineer fuel-cell membrane electrode assemblies, although one must recognize that such a BPM will always suffer from an unavoidable junction potential7. Such a bipolar membrane also holds some interest in the field of water electrolysis, where the membrane could separate an acidic H2-evolution cathode (a very fast reaction) from an alkaline O2-evolution anode (possible on non-Pt-group metals) in a system combining the benefits of the best electrodes of present acidic/alkaline water electrolysers.