The Battery That Has Not Yet Been Built

ANo component of the electric car has attracted as many announcements as the battery that has not yet been built. For three decades the cell in commercial use has carried its lithium between electrodes through a liquid, an arrangement that works well enough to have sold hundreds of millions of vehicles and badly enough to keep engineers looking for an alternative. The liquid is flammable, it limits how much lithium can be packed into a given volume, and it obliges manufacturers to surround each pack with cooling hardware that stores no energy at all. Replacing it with a solid, whether a ceramic, a glass or a rubbery polymer, is supposed to remove all three drawbacks at once, since a solid cannot burn and can in principle sit directly against a sheet of lithium metal, the most energetic anode available. The phrase solid-state battery, however, covers designs that have almost nothing in common beyond the absence of a liquid, and the difficulty that defeats one of them frequently leaves the others untouched.

BTeodora Mensah, an electrochemist at the Nordkap Institute for Energy Storage, argues that the obstacle now standing in the way is mechanical rather than chemical. Between 2019 and 2024 her group cycled 46 cells built around a single ceramic electrolyte and examined each one after failure under an X-ray tomograph. No cell was run at more than one current density. Cells worked gently, at 0.3 milliamps per square centimetre, passed 900 cycles and gave up only 8 per cent of their capacity; cells worked at 5 milliamps per square centimetre, the rate a driver expects at a motorway charger, were finished in fewer than 40. What the tomographs showed in the failed cells was not a chemical reaction but a crack. Lithium had been forced into pores and grain boundaries already present in the ceramic and had prised them open. A material that conducts ions faster than any liquid, Mensah notes, is of no use if the current that makes it worth having is also the current that splits it.

CThe account yields a prediction, and Mensah has tested it. If failure begins at flaws that are there before the cell is ever charged, two ceramics of identical chemistry ought to behave differently whenever their manufacture leaves different numbers of pores. Pellets pressed to 99 per cent of theoretical density carried 4 milliamps per square centimetre for 300 cycles; pellets of 94 per cent density, cut from the same batch of powder, gave way at less than half that current. The benefit, however, was confined to cells in which the ceramic met the lithium across a surface that had been polished after pressing; where the surface was left as it came from the die, density made no measurable difference. Mensah presents the finding as a constraint on manufacturing rather than a verdict on the technology, and she is explicit that nothing in it tells against the polymer and sulphide designs that other groups pursue.

DHassan Oyelaran, a materials engineer at the Vale do Sol Battery Centre, considers the inference premature, on the ground that a cell which fails in a laboratory fixture may be reporting on the fixture. Coin cells of the sort used in most academic work hold their layers together with a spring, and the pressure it delivers falls as the electrodes breathe during cycling. His group assembled pouch cells under a constant 3 megapascals, the figure a production stack could plausibly maintain, and 82 per cent of them passed 500 cycles at 4 milliamps per square centimetre without the cracking Mensah describes. Testing at pressures of this kind is now ordinary practice in his laboratory, although it was rare when the work began. Oyelaran concedes two points. His cells ran at 45 degrees Celsius, warmer than a parked car becomes in most of the world, and his longest test covered 500 cycles, roughly a tenth of what a warranty demands.

EThe argument has consequences outside the laboratory. Several carmakers announced between 2021 and 2023 that solid packs would reach their showrooms before the decade ended, and the bodies that certify battery safety have begun drafting procedures for cells containing no liquid at all. Investors read laboratory cycle counts as forecasts of production dates, a use those numbers were never designed to bear. Oyelaran observes, awkwardly for his own position, that a pouch cell held at 3 megapascals says nothing about cost, and that the equipment needed to maintain such a pressure may itself weigh more than the energy it preserves. Mensah answers that cost and physics cannot be kept apart in practice, since the polishing step her results require would be charged to every cell produced. Both regard the habit of reporting a single spectacular cell, rather than the spread of results across a batch, as the worst practice in the field.

FA settlement of a kind is taking shape, in which the two accounts are assigned to different operating conditions rather than ranked against each other. Where a cell is charged slowly and held firmly, the ceramic behaves as its conductivity promises, which would explain the long lives several laboratories report; where charging is quick and pressure is allowed to drift, flaws left by the press govern the outcome, and the chemistry of the electrolyte becomes almost beside the point. What remains unsettled is whether the pressures that succeed on a bench can be held for a decade in a vehicle driven over broken roads, and no accelerated test now in use can answer that. What both researchers reject is the assumption behind the announcements, namely that a battery which works once in a fixture is a battery that has been engineered.