"Solid-state batteries will fix everything about EVs." You've heard it for years: no fires, faster charging, longer range. And yet mass production keeps slipping. One of the reasons is almost absurd when you say it out loud: the softest metal in the battery punches through the hardest ceramic in it, and shorts the cell.
Lithium metal cuts like cold butter. Garnet ceramic electrolyte turns a knife edge. Butter is drilling through porcelain. It shouldn't work — and it happens in cell after cell.
For more than two decades, the field split into two camps over why. In April 2026, a Max Planck team published a paper in Nature that settles the argument. This post is about what they found, why it took so long, and what it changes for the battery in your next car.
Solid-state batteries in one minute
The lithium-ion cells in your phone and car use a liquid electrolyte — lithium ions swim through it between the anode and cathode. The problem is that this liquid is a flammable organic solvent. Most EV fire headlines start there.
A solid-state battery replaces the liquid with a solid, usually a ceramic. Ceramic doesn't burn. Better still, it lets you use pure lithium metal as the anode instead of graphite, which packs far more energy into the same volume. Non-flammable and higher capacity: on paper, it's the perfect cell.
Then you cycle it. After repeated charging, tree-like lithium crystals — dendrites — start growing from the anode surface. When a dendrite crosses the ceramic and touches the cathode, the cell shorts and dies.
The 20-year argument: electrons or force?
How does soft lithium get through hard ceramic? Two hypotheses competed.
Hypothesis A — the electrochemical camp: "the ceramic leaks."
Ceramic is made of tiny crystal grains, and the boundaries between grains can leak a trickle of electrons. Wherever electrons leak, lithium ions meet them and plate out as metal inside the ceramic. These isolated deposits link up and form a path through. The analogy: there are hairline channels in a wall, mold grows inside the wall first, then connects and breaks through.
Hypothesis B — the mechanical camp: "lithium pushes the ceramic apart."
Every ceramic has microscopic pre-existing cracks from manufacturing. During charging, lithium is forced into these cracks. Trapped with nowhere to go, it builds internal pressure that pries the crack tip open and splits the ceramic. The analogy: water seeping into a rock crevice, freezing, and splitting the rock.
Both are plausible, which is why the debate lasted two decades. And the two camps prescribe opposite fixes: if A is right, you need to block electron conduction in the ceramic; if B is right, you need a tougher ceramic. Not knowing which, research funding flowed in both directions.
The decisive evidence: nothing ahead of the tip
What Dr. Yuwei Zhang's group at the Max Planck Institute for Sustainable Materials did sounds simple: they looked directly at where a dendrite had passed through, at atomic resolution.
Simple to say, brutal to do. Lithium degrades the instant it meets oxygen or moisture, and an electron microscope's beam damages it too. Sample preparation and imaging had to stay under vacuum and cryogenic temperatures without a single break.
The key result:
No lithium enrichment was detected in the ceramic ahead of the dendrite tip.
If Hypothesis A were right, lithium should already be plated inside the ceramic ahead of the advancing dendrite — the mold should be there before the breakthrough. It wasn't. The region ahead of the tip was clean. What the team did find was stress signatures in the ceramic around the crack and plastic deformation inside the lithium — the fingerprints of metal being forced in under enormous pressure.
Electron backscatter diffraction and phase-field simulations pointed the same way. After twenty years, the mechanical camp won.
The mechanism: a waterjet
The authors offered an analogy I can't improve on:
"The soft lithium metal is able to penetrate the stiff ceramic electrolyte, like a continuous waterjet that penetrates a rock."
Think of an industrial waterjet cutter. Water is about as soft as matter gets. Fired through a fine nozzle at thousands of atmospheres, it slices steel and granite. The water didn't get harder. The pressure got concentrated.
Lithium does the same thing. During charging, lithium ions plate out as metal at the anode. When that happens inside one of the ceramic's micro-cracks, the new metal has no exit. It keeps accumulating. Even a soft material, confined on all sides while its volume grows, develops rapidly rising hydrostatic pressure. That pressure loads the crack tip in tension — and brittle materials like ceramic are weak in tension. Crack. Lithium flows into the new gap, pressure rebuilds, crack again. Repeat until the dendrite reaches the far electrode.
| Hypothesis A (electrochemical) | Hypothesis B (mechanical) — winner | |
|---|---|---|
| Cause | Electrons leaking along grain boundaries | Hydrostatic pressure of lithium in cracks |
| Ahead of the dendrite tip | Lithium should already be plated | Should be clean |
| Observed | No plating ✗ | Clean + stress signatures ✓ |
| Everyday analogy | Mold growing inside a wall | Waterjet cutting rock |
| Prescription | Block electron conduction | Tougher ceramic / manage flaws |
An arXiv model paper says the same thing with equations
Around the same time, an analytical model of exactly this process appeared on arXiv (2603.20113). Its core idea is an energy trade-off. Growing a dendrite costs mechanical energy to crack the ceramic. Not growing it costs electrical energy, because current has to detour around the obstruction. Nature takes the cheaper path, so the dendrite advances the moment cracking costs less than detouring.
One conclusion from the model is practically important: the critical current density for dendrite growth scales inversely with the largest flaw length to the 3/2 power. In plain terms, the single biggest defect in the ceramic decides the fate of the cell. Not the average — the worst case. If you work in semiconductors this sounds familiar: a wafer can be pristine everywhere except the one particle that kills the die. It also explains why nominally identical cells fail so inconsistently, and predicts that the scatter follows a Weibull distribution, just like ceramic tensile strength.
So when does my EV get one?
Settling the mechanism doesn't ship a battery tomorrow. What it does is tell everyone where to spend the money. The Nature team lays out three paths:
- Tougher ceramics (higher fracture toughness) — so pressure builds without the crack opening. In waterjet terms: swap the rock for harder rock.
- Deliberate micro-voids to redirect cracks — engineered empty space inside the ceramic gives dendrites and cracks somewhere to go that isn't straight across to the other electrode. Think of a flood spillway.
- Protective coatings on the lithium electrode — stop dendrites at the point where they start.
And the arXiv model adds a fourth: process control aimed at the largest flaw, not the average. Design the manufacturing line to kill the worst defect, not to improve the mean.
Either way, "stop the electron leakage" just dropped down the priority list. Funding that spent twenty years split between two camps can now concentrate on one. That alone changes the pace.
What stayed with me
What struck me most about this paper wasn't the conclusion — it was the method.
Two hypotheses coexisted for twenty years for one reason: nobody could look directly at a dendrite tip. Lithium dies in air and dies under an electron beam, so everyone reasoned from indirect evidence. The Max Planck team didn't invent a new theory. They perfected the tedious sample-handling technique that keeps vacuum and cryogenic conditions unbroken from start to finish. Then they looked. The answer was sitting there.
I work in the semiconductor industry, and I see this pattern constantly: process and equipment teams arguing for months over a defect's root cause, until someone finally cross-sections the part and puts it under a microscope, and it's over in ten minutes. Far more problems go unsolved for lack of a way to see than for lack of theory.
References
- Yuwei Zhang et al., "Mechanically driven Li dendrite penetration in garnet solid electrolyte," Nature (2026). DOI: 10.1038/s41586-026-10415-9
- Ansgar Lowack, "An Analytical Model of Critical and Subcritical Alkali Metal Dendrite Growth in Ceramic Solid Electrolytes," arXiv:2603.20113 (2026)
This post explains published research for a general audience; see the papers for precise figures and conditions. Part of an ongoing series on solid-state physics in everyday life. The Korean version is on my Naver blog.
