A computational model for lithium cobalt oxide leaching tracked published data reasonably well under lower-acid conditions, but it overpredicted recovery when hydrogen peroxide was present at 2.5 M hydrochloric acid. The result draws a clear line between the conditions the model followed and the high-acid peroxide cases where its predictions ran ahead of the comparison data.
The work is a computational comparison, not a new laboratory experiment. It models lithium cobalt oxide, or LCO, and checks predicted lithium and cobalt conversion or recovery against published LCO leaching data. The comparison covers hydrochloric acid at 0.5 M, 1.5 M and 2.5 M, alongside hydrogen peroxide at 0%, 0.2%, 0.4% and 0.6% by volume.
Following the chemistry
Rather than tracking only the final recovery, the model follows changing features of the particle and its surrounding chemistry. It includes LCO conversion and particle radius, acid and peroxide, formation and dissolution of a Co3O4 film, film thickness, lithium and cobalt recovery, oxygen moles and passivation. Passivation is represented as part of the model's changing surface-film state.
The simulations were run in COMSOL Multiphysics v6.3 using the Global ODEs and DAEs, or ge, module. The model is a lumped, zero-dimensional representation, and its kinetic, mass-transfer and morphological parameters were assumed and iteratively tuned to predict the published data. The parameter table used an LCO sample mass of 2.5 g.
The fit was strongest at lower acid
Across the comparison data, lithium and cobalt recovery increased with acid and hydrogen peroxide concentration. Lithium was easier and faster to recover than cobalt. The analysis reports this as a directional trend, with no pooled effect size.
At 0.5 M hydrochloric acid, the model trajectories were reasonably reproduced across all four peroxide concentrations. Lithium recovery reached around 50% with no peroxide and around 40% at the other peroxide levels, while cobalt recovery stayed below 35%.
At intermediate peroxide levels, the model showed reasonable overall agreement for both metals, and lithium performed well across the peroxide range. The main deviations appeared later in the trajectories, when film-transition behavior was not captured as well at 0.2% and 0.4% peroxide. Film dynamics were better captured at 0.6% peroxide.
Where the model lost the trail
The clearest failure came at 2.5 M acid when peroxide was added. In the 0.6% peroxide simulation, conversion approached 100%, and lithium and cobalt concentrations reached about 510 mol/m3, described as the theoretical maximum. The peroxide cases overpredicted cobalt conversion and reaction rate, while lithium conversion was also overpredicted at higher peroxide levels. Recovery fit well when no peroxide was used.
The authors suggest that this mismatch may reflect too much sensitivity to acid concentration in two reaction terms, labelled R3 and R4. They also identify oxygen bubbles as a possible source of interference. Proposed refinements include fractional acid orders or acid activities and a way to account for oxygen-bubble effects, but those ideas were not tested in the reported comparison.
A comparison, not an independent test
Those caveats matter because the parameter values were assumed and iteratively tuned to the literature data. The exercise therefore shows how this framework and parameter set tracked the cited trajectories, not how accurately it would predict a new dataset. No formal goodness-of-fit measures, confidence intervals or uncertainty analysis were reported. Because the model is lumped and zero-dimensional, it does not describe spatial variation inside or around a particle.
The source experiment's sample size and raw-data availability were not reported. The document is an arXiv preprint, version 1, dated 26 August 2026, and no journal is listed. Support was strongest at 0.5 M and 1.5 M acid, while the 2.5 M acid peroxide condition remained poorly fitted. The framework was not tested here on other cathode chemistries, so its reach beyond LCO remains untested.
Paper data and sources
Original title: Battery Recycling: Mechanistic Modelling of LiCoO$_2$ Leaching with Coupled Diffusion-Reaction Kinetics and Film Passivation
Authors: Uddipta Sarma, Ganesh Madabattula
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text