A research team centered at the University of California, Los Angeles (UCLA) has demonstrated, across two peer-reviewed papers, that charge-discharge efficiency alone cannot evaluate the lifetime of next-generation metal batteries. The paper in Joule compared four types of non-aqueous metal anodes—Li, Na, Al, and Mg—by resting them for 14 days. The paper in Nature Communications reconstructed the electrolyte for aqueous Zn batteries. Both are lab-scale cell tests, but they capture with concrete numbers the challenge of designing for the time a battery spends idle, not just the time it spends running.
High cycling efficiency does not guarantee rest-period lifetime
The study published in Joule by Jin Koo Kim et al. examined Li, Na, Al, and Mg anodes, all of which exceeded 98.5% Coulombic efficiency during cycling. Coulombic efficiency refers to the fraction of charge put in during charging that can be retrieved during discharge. However, when a 14-day open-circuit rest period was inserted, the normalized recovered capacity of Li, Na, and Al dropped from 100% to approximately 80–92%, corresponding to losses of roughly 8–20%. Mg, by contrast, dropped from 100% to about 99.6%, limiting the loss to roughly 0.4%.
The contrast is most striking for Na. While its cycling Coulombic efficiency was 99.89% under normal cycling, the recoverable fraction after a 14-day rest was only 92.5%. Across these four metal-electrolyte combinations, there was no correlation between high cycling efficiency and stability during rest. However, the electrolyte composition also differs for each metal, so this comparison cannot isolate causation to the metal species alone.
The test subjects were half-cells using four types of metal anodes—not vehicle data from commercial EVs or complete battery packs. While the paper explicitly specifies four chemical systems, it does not state the number of independent cells used for each measurement point. These results cannot be directly extrapolated to metal batteries in general.
Mg forms a resistive layer only during rest
The behavior of Mg is supported by microscopic observation and impedance measurements. In Li, the solid electrolyte interphase (SEI) continued to thicken even during rest; in Na and Al, localized pitting corrosion progressed. In Mg, however, a high-resistance interfacial layer formed during rest, kinetically slowing the corrosion reaction. This layer disappeared upon subsequent stripping, meaning it did not become a fixed film that would continuously impede Mg-ion transport.
The design principle the research team proposes is not to treat interfacial resistance uniformly as a sign of degradation, but rather to leverage it as a reversible protective layer that appears only during rest. What the Joule paper demonstrated is limited to this specific combination with an Mg-complex electrolyte. Whether the same dynamic interface can be engineered for Li, Na, and Al to suppress rest-period losses remains a matter for future verification.
Zn combines dilution with additives
The Nature Communications paper by Haoyang Wu et al. took a different approach for aqueous Zn batteries. Conventional aqueous electrolyte designs often increase salt concentration to reduce water's reactivity. The research team did the opposite: they diluted ZnSO4 to 0.1 M, added 0.5 M Li2SO4 and 5 vol% DMF, and replaced the solvent with heavy water (D2O). The final formulation is called "1Z5L."
Standard testing used 2032-type coin cells. At 25°C, Zn was deposited at 1 mAh/cm² with a current density of 1 mA/cm², and after the second deposition cycle, the cell was rested open-circuit for 24 hours. The electrolyte volume was 30 μL/cm² per electrode area. The measurements in Figure 4 involved 3–5 independent measurements per condition, and the capacity retention of 1Z5L dropped from an initial 100% to 98.4%—a difference of 1.6 points. By comparison, the 1 M ZnSO4 formulation cited in the paper dropped from 100% to 64.4%, a loss reaching 35.6%.
The effect was not limited to 24 hours. While 1Z5L maintained 84.4% of its initial 100% after 7 days, a 2 M ZnSO4 aqueous solution retained only 3.0%. Furthermore, in an anode-free full cell where a 2-hour open-circuit rest at 100% state of charge was inserted every 5 cycles, the capacity retention after 100 cycles at 25°C and 0.5 mA/cm² was 90.4%. The equivalent 2 M control under the same conditions retained only 0.1%. This full-cell test does not specify the number of independent cells per condition, and should be considered separately from the 3–5 measurements reported in Figure 4. Both sets of results are coin-cell comparisons, not measurements of lifetime in larger cells.
Cycling test results should also be considered separately. A different full cell combining Zn with LiMn2O4 maintained 80% of its initial 100% capacity after 3,300 cycles at 4C, with an average Coulombic efficiency of 99.8%. This 3,300-cycle test did not incorporate rest periods. Long-term charge-discharge performance and 24-hour calendar-life testing should not be conflated as the same lifetime metric.
Corrosion mechanism examined through both computation and experiment
The Zn paper examined, through both computation and experiment, the mechanism by which solvated water molecules promote corrosion more readily than free water molecules. Density functional theory simulations assumed a Zn(002) three-layer 4×4 surface with two explicit water layers, among other conditions. The reaction barrier for the Volmer step, in which a proton is transferred from pure water, was calculated at approximately 1.2 eV; for water coordinated to Zn, this dropped by about 0.6 eV. This is a theoretical value derived from the assumed atomic model, not an experimentally measured barrier within an actual cell.
Experimentally, reducing the ZnSO4 concentration from 3 M to 0.1 M raised the pH from 3.4 to 6.5, and the 24-hour retention rate correspondingly rose from 52.2% to 87.2%. What this shows is a correlation among concentration, pH, and retention. Building on this, the research team progressively tested substitution with heavy water and the addition of Li2SO4 and DMF. Combined with electrochemical measurements and microscopic observation, they confirmed that corrosion decreased with the final formulation. Of the 1.5% irreversible Zn loss observed after resting, 0.65% was attributed to electrically isolated metallic Zn and 0.85% to corrosion-derived Zn2+.
Meanwhile, 1Z5L changes multiple factors simultaneously. The individual contributions of dilution, heavy water substitution, and the addition of Li2SO4 and DMF to the final 1.5% loss cannot be separated. The authors believe this approach could be applicable to other aqueous batteries, but at this stage it remains a possibility demonstrated only for Zn.
Larger cells and independent replication are the next hurdles
The Nature Communications paper was submitted on September 17, 2025, accepted on June 18, 2026, and published as an early, pre-edited version on August 6. The Joule paper was also published online as a peer-reviewed Article on May 19. Neither is merely a preprint claim, but neither paper includes tests where an independent research group reproduced the same conditions.
For practical implementation, testing must move from coin cells to larger pouch cells and verify performance under conditions where electrolyte cannot be used abundantly. For 1Z5L, which contains heavy water, material cost and manufacturing scalability, as well as safety and environmental impact assessments, remain to be addressed. The rest periods tested—14 days for Mg, and 24 hours and 7 days for Zn—must be extended to months and years. What these two papers demonstrate is not a paradigm shift in established theory, but rather the necessity of incorporating capacity recovery after rest into lifetime evaluation. Alongside cycle count, the percentage of capacity recoverable from an initial 100% after a rest period will become a key criterion for selecting next-generation metal batteries.