When most researchers first see lithium dendrites in solid-state batteries, they imagine a straightforward picture: a lithium "needle" grows from one electrode, pushing through the electrolyte until it touches the other side and causes a short circuit.
But a 2021 Nature Materials study from Oxford University, the Faraday Institution, and the Paul Scherrer Institute painted a very different picture. Using operando synchrotron X-ray computed tomography (XCT), they watched Li/LiPSCl/Li symmetric cells cycle in real time—and discovered something counterintuitive.
The crack runs through the electrolyte first. Lithium fills it from behind.
When the crack reaches the other electrode, the battery doesn't short circuit immediately. It keeps working until lithium gradually fills the entire channel. This finding rewrites our understanding of how dendrites form—and more importantly, how to prevent them.
The Intuition Was Wrong
The traditional view assumes lithium drives the crack: the lithium tip pushes forward, forcing the electrolyte to fracture ahead of it. The crack and the lithium front move together.
The operando XCT data shows the opposite:
- A dry crack propagates through the solid electrolyte
- Lithium metal fills the crack from the rear, not from the tip
- The crack can span the full electrolyte thickness while lithium is still near the plating electrode
- Short circuit only occurs when lithium finally fills the entire channel
This two-stage mechanism explains why some batteries continue cycling even after cracks have formed—because a crack alone isn't enough for electronic short circuit.
Why This Matters for Battery Developers
If you've been trying to "prevent dendrites" by focusing only on interfacial coatings or ionic conductivity, you've been targeting only half the problem.
This study splits dendrite prevention into two independent engineering goals:
Goal 1: Stop the Crack from Propagating
Before lithium can fill a channel, the channel must exist. Preventing crack propagation means:
- Increase fracture toughness: Add ductile phases, fiber reinforcement, or transformation-toughening
- Reduce porosity gradients: Pre-existing voids guide crack paths
- Design crack-arresting layers: Interlayers that deflect or stop cracks
- Optimize stack pressure: Too low allows gaps; too high concentrates stress
Goal 2: Stop Lithium from Filling the Crack
Even if a crack forms, lithium still needs to deposit preferentially into that crack. Preventing lithium ingress means:
- Improve interfacial contact: Eliminate voids that concentrate current
- Distribute current uniformly: Avoid edge effects and hot spots
- Reduce local overpotential: Better wetting reduces plating driving force
- Monitor with reference electrode: Detect localized plating before it spreads
Standard two-electrode tests can't separate these two stages. You need either operando imaging or three-electrode cells to tell which mechanism dominates.
How the Experiment Was Done
Materials and Cell Configuration
The team used LiPSCl argyrodite as the solid electrolyte, sandwiched between two lithium metal foils. This sulfide electrolyte offers high ionic conductivity (~1 mS/cm) and forms only a thin (~nm-scale) interphase with lithium—so any large-scale cracks can be attributed to mechanical stress from plating, not runaway interfacial reactions.
Cells were held at ~7 MPa stack pressure and cycled at 1.5 mA/cm² with 0.3 mAh/cm² per half-cycle. These are aggressive conditions designed to accelerate crack formation—not representative of commercial cell lifetimes, but ideal for observing the failure mechanism.
Operando Synchrotron XCT
The cells were cycled inside the synchrotron beamline at the Diamond Light Source (UK) and TOMCAT beamline at the Swiss Light Source. Key imaging parameters:
- X-ray energy: 35 keV (monochromatic)
- Detector: LuAG:Ce scintillator + 4× optical magnification
- Voxel size: ~0.65 μm
- Field of view: ~2.6 × 2.6 mm
The term "operando" needs clarification: the current was paused at specific capacity intervals to acquire a full 3D tomogram (3570 projections per scan), then resumed. This gives a series of time-lapse 3D snapshots, not millisecond-level video—but it's sufficient to track crack propagation and lithium ingress as a function of plated capacity.
Phase-contrast imaging enhances edges and low-density features, making it possible to distinguish empty cracks (nearly zero attenuation) from lithium-filled cracks (higher attenuation) and bulk electrolyte.
What They Saw: A Timeline of Failure
Stage 1: Spallation at the Interface (0.2 mAh/cm²)
The first observable change wasn't in the bulk electrolyte—it was at the lithium/electrolyte interface.
After the first plating cycle, circular or elliptical pits appeared on the electrolyte surface near the plating electrode. These are called spallation craters: small cone-shaped fractures where a chunk of ceramic "pops off" the surface.
Spallation occurs more frequently:
- At electrode edges, where electric field and local current density are higher
- Near interfacial voids, where contact is poor
- In regions with higher pre-existing porosity
Beneath each spallation crater, a vertical crack extends into the electrolyte, already several tens of microns deep. These were not pre-existing defects—XCT scans of pristine cells showed no such features.
The team also observed spallation in Li/Li₃N/Li and Na/β″-alumina/Na cells, suggesting this is a general brittle-failure mode for ceramic electrolytes under localized stress, not specific to sulfides.
Stage 2: Crack Propagates Through the Electrolyte (0.4–0.8 mAh/cm²)
As plating continued, the vertical cracks grew deeper—but the lithium didn't keep pace.
By 0.6 mAh/cm², some cracks had reached halfway through the electrolyte. By 0.8 mAh/cm², several cracks had fully spanned the electrolyte thickness and reached the stripping electrode.
But here's the key observation: the voltage didn't drop to zero yet.
In a traditional two-electrode test, you'd assume a through-thickness crack means immediate short circuit. Not here. The crack had connected both sides mechanically, but not electrically—because it was still mostly empty.
Stage 3: Lithium Fills the Crack from Behind (1.0 mAh/cm²)
The team used grayscale intensity in phase-contrast XCT as a proxy for lithium content. Empty cracks have near-zero attenuation (dark in reconstructed images); as lithium fills them, attenuation increases (brighter grayscale).
Plotting grayscale along the length of a crack showed:
- Near the plating electrode (top): High grayscale → lithium present
- Near the stripping electrode (bottom): Low grayscale → still mostly empty
Lithium was filling the crack from the rear, starting near where it was plated, gradually widening the channel. The crack tip had already run ahead.
3D rendering made this even clearer: gray volumes (empty cracks) extended through the full electrolyte thickness, while green volumes (lithium) clustered near the plating side and partway down the crack.
Stage 4: Short Circuit (Complete Lithium Filling)
Only when lithium had filled enough of the crack to create a continuous electronic pathway did the cell voltage collapse. This happened well after the crack itself had become through-thickness.
The delay between "crack connects both electrodes" and "short circuit" can be several tenths of mAh/cm². For a cell cycling at 1.5 mA/cm², that's additional minutes or even hours of apparent functionality.
This is why some solid-state cells show gradual capacity fade before catastrophic failure—the crack network exists, but lithium hasn't filled it yet.
Confirming It's Really Lithium: Spatially Resolved XRD
XCT grayscale alone can't prove the bright regions are lithium—they could be densified ceramic fragments, residual electrolyte, or interphase layers.
To confirm, the team performed spatially resolved X-ray diffraction at the Diamond Light Source I12 beamline. After cycling a cell to short circuit, they scanned a 1.4 × 1.4 mm grid across the lithium electrode surface with 100 × 100 μm spot size, collecting diffraction patterns at each point.
The Li (110) peak intensity was much stronger at the electrode edges and correlated spatially with the spallation sites seen in XCT. Not every spallation led to a through-thickness dendrite, but every long-distance dendrite originated from a spallation.
This dual confirmation (XCT morphology + XRD phase identity) is the gold standard for operando studies: seeing a feature isn't enough—you also need chemical proof.
Why Cracks Prefer Electrode Edges
Finite element modeling showed two contributors to edge concentration:
- Higher electric field at electrode edges (classic edge effect in electrochemistry)
- Interfacial voids from imperfect contact, especially near edges where pressure is harder to control
Both lead to localized current density spikes—up to 2–3× higher than the nominal value. These hot spots drive faster lithium plating, higher mechanical stress, and earlier spallation.
Pre-existing porosity also matters: regions with ~4% void fraction (detectable by XCT) showed more spallation than regions with ~1.5%. But voids below the XCT resolution limit (~0.65 μm) weren't captured, so the true porosity influence is likely stronger.
Practical takeaway: Controlling interfacial contact and avoiding edge concentration is as important as choosing the "right" electrolyte composition.
Can XCT Really Separate Cracks from Lithium?
This is a common question when reading operando imaging papers.
What XCT Can Do
- Distinguish empty cracks from lithium-filled cracks at micron resolution
- Track crack propagation and lithium ingress as separate, sequential events
- Correlate crack morphology with electrode position, voids, and current history
What XCT Can't Do Alone
- Chemically identify every sub-micron feature—lithium, interphase, and ceramic fragments can have overlapping grayscale
- Resolve nanoscale dendrites below voxel size
- Prove lithium identity without independent confirmation (XRD, chemical mapping, etc.)
The strength of this study is combining time-resolved XCT (to see crack-first, lithium-later sequence) with spatially resolved XRD (to confirm lithium location). Neither alone would be conclusive.
What This Means for Solid-State Battery Development
Rethink Your Failure Analysis
If your cell shorts after 50 cycles, don't assume "lithium grew a dendrite." Ask:
- When did the crack form? (May have been cycles earlier)
- What caused the crack? (Spallation, pre-existing void, inhomogeneous pressure)
- Why did lithium preferentially fill that crack? (Edge effect, poor contact, local current spike)
Use Three-Electrode Cells for Early Warning
A two-electrode cell gives you one number: total voltage. You can't tell if plating is uniform or if one side is concentrating stress.
A three-electrode cell lets you monitor anode and cathode potentials separately. When the anode potential shows localized overpotential spikes, you're seeing the electrochemical signature of current concentration—before the crack becomes through-thickness.
Related product: Our three-electrode test mold is designed for this type of diagnostic testing. PEEK core, 500 MPa pressure capability, split-cell design for glovebox assembly.
Optimize Stack Pressure—But Not Just "Higher Is Better"
This study used ~7 MPa, which is moderate for lab cells. Higher pressure can:
- Reduce interfacial voids → lower current concentration
- Suppress crack opening → slower propagation
But excessive pressure can:
- Crush ceramic particles → create new crack initiation sites
- Deform lithium into cracks → accelerate filling
There's an optimal window. Operando imaging under controlled pressure is the only way to find it for your specific material system.
Related resource: See our solid-state battery assembly troubleshooting series for practical guidance on pressure management.
Key Takeaways
- Cracks propagate first; lithium fills later → Dendrite formation is a two-stage process, not one
- Through-thickness cracks don't immediately short the cell → You have a window to detect and intervene
- Spallation at the interface is the early warning → Monitor surface changes, not just voltage
- Edge effects and voids dominate failure location → Interfacial contact quality matters more than bulk properties
- Prevention requires dual strategies: Stop crack propagation AND stop lithium ingress
This isn't just academic curiosity—it's actionable guidance for cell design, failure diagnosis, and quality control.
Reference
Ning, Z., Jolly, D. S., Li, G. et al. Visualizing plating-induced cracking in lithium-anode solid-electrolyte cells. Nature Materials 20, 1121–1129 (2021). DOI: 10.1038/s41563-021-00967-8
Institutions: University of Oxford, The Faraday Institution (UK), Paul Scherrer Institut (Switzerland)
Related Resources
- Three-Electrode Solid-State Battery Testing — Separate anode and cathode contributions to detect localized plating
- Solid-State Battery Failure Analysis Series — Real cases where three-electrode testing revealed hidden failure modes
- Solid-State Battery Assembly Troubleshooting (Part 2) — Preventing poor contact and edge effects during assembly
Testing Equipment for Advanced Diagnostics
Studying crack propagation and lithium plating requires precise control and real-time monitoring.
Three-Electrode Test Mold for Solid-State Batteries
Monitor anode and cathode potentials independently to detect current concentration before catastrophic failure:
- PEEK core: Electrically insulating, stable to 200°C
- 500 MPa standard pressure (1000 MPa custom available)
- Split-cell design: Easy assembly in glovebox
- Reference electrode integration: Track localized overpotential in real time
Single-Point Pressing Mold
Achieve uniform stack pressure to minimize interfacial voids and edge effects:
- Torque-calibrated lead screw: Reproducible pressure control
- φ10 mm working cavity (customizable 5–20 mm)
- Compact footprint: Fits standard gloveboxes
📧 Technical inquiry: business@labcorematerials.com 🌐 Full catalog: labcorematerials.com/products
We ship worldwide (DDP to major US ports) with quotes returned within 48 hours.
LabCore Materials — Precision tools for solid-state battery research.