Differential Capacity Analysis for Detecting Metallic Plating in High Current Graphite Cells
Differential capacity analysis detects metallic lithium plating on high-current graphite cells by identifying distinct stripping peaks during low-rate discharge.

Staging
Channel four on a Maccor series 4000 cycler records terminal voltage during a 3C continuous charge of a 65 Ah prismatic lithium-ion cell. The negative active material is secondary artificial graphite blended with five weight percent silicon oxide. As lithium ions cross the carbonate electrolyte into the host matrix, the anode’s thermodynamic open-circuit potential drops through distinct phase transformations: dilute stage 1L, stage 4, stage 3, stage 2 marked by LiC12 stoichiometry, and stage 1 at LiC6.
Each transition occurs at a defined equilibrium potential on the Li/Li+ reference scale.
Under open-circuit equilibrium at 25 degrees Celsius, the stage 2 to stage 1 conversion sits at approximately 85 millivolts versus Li/Li+. Full-cell terminal voltage simply reflects cathode potential minus anode potential, less cumulative overpotentials under load. At high current densities, polarization across the composite anode layer becomes severe.
Ohmic drop through the separator pores and porous matrix pulls the local potential downward, while solid-state diffusion resistance inside the crystallites builds a steep internal concentration gradient, compounded by charge-transfer impedance at the solid electrolyte interphase.
Thermodynamic equilibrium potential shifts toward zero volts as graphite saturation approaches complete lithiation.
Once these combined overpotentials exceed 85 millivolts during charging, the graphite surface potential drops below zero volts versus Li/Li+, making metallic lithium deposition thermodynamically favorable. The liquid electrolyte does not distinguish between intercalation charge transfer and direct electrodeposition of neutral metal once that surface potential goes negative. Deposition initiates on the graphite flakes nearest the separator, where current distribution through the coating thickness peaks and electrochemical stress is highest.
- Ohmic Polarization drives immediate voltage drop across electrolyte-filled separator pores and tortuous composite channels.
- Charge Transfer Resistance establishes the desolvation activation overpotential across the inorganic solid electrolyte interphase.
- Solid Diffusion Impedance steepens the concentration gradient between particle surfaces and interior crystallite lattices under sustained high current.
Differential capacity analysis converts raw voltage and charge data into derivative curves of dQ/dV versus terminal voltage, turning flat plateaus into recognizable peaks aligned with specific intercalation phase shifts, even as graphite expansion shifts particle contact. Paired with a nickel-rich cathode like NMC811 in a balanced full cell, discharge de-intercalation produces signature peaks tracking LiC6 back to LiC12 and onward into higher-stage structures. At low currents, these graphite transitions maintain clean, predictable voltage separation across state-of-charge ranges.
High charging rates distort that baseline. Joule heating lowers bulk electrolyte viscosity while elevated current magnifies local overpotentials, so whereas low-rate cycling gives sharp derivative peaks, fast charging flattens and broadens them. When current outpaces the kinetic limits of the graphite microstructure, metallic lithium accumulates right alongside intercalation compounds, establishing the physical conditions for subsequent stripping detection.

Stripping
Upon current reversal, reversibly deposited metallic lithium dissolves back into lithium ions before the graphite lattice begins de-intercalating. Because metal dissolution proceeds with faster exchange current kinetics than de-intercalation from LiC6, an anomalous discharge plateau appears below the normal stage 1 extraction plateau. Differentiating this early discharge segment generates a distinct peak on the dQ/dV curve, whose integrated area measures the precise quantity of reversible metallic lithium recovered during that cycle.
Plated lithium also reacts directly with carbonate solvents to grow parasitic solid electrolyte interphase film, consuming cyclable lithium inventory and mobile solvent molecules. As the deposit strips unevenly, fragments lose electrical contact with the graphite matrix to form isolated domains of dead lithium that drive up cell impedance. The total plated mass divides into reversible metal, passivated reaction products, and electrically isolated debris, leaving differential capacity analysis to capture only the reversible fraction still wired to the conductive network.
Integration of the low-voltage stripping peak on a 65 Ah cell tested at minus five degrees Celsius yields 4.2 ampere-hours of reversible metallic lithium following a 2C fast charge.
Quantifying reversible plating requires numerical integration across the stripping peak baseline. Superimposing a post-2C discharge curve over a C/20 reference discharge ~ where no plating occurs ~ reveals an emergent dQ/dV peak centered between 3.52 volts and 3.58 volts in an NMC811 graphite system. Integrating the derivative curve between baseline inflection points gives total capacity in that interval, from which subtracting the reference intercalation capacity isolates the stripped metallic fraction.

Will Low Temperatures Mask Graphite Phase Transitions?
Below 10 degrees Celsius, solid-state lithium diffusivity in graphite drops by roughly an order of magnitude, while ethylene carbonate electrolyte conductivity plummets and kinetic overpotentials climb. Increased internal resistance drags normal graphite de-intercalation peaks toward lower terminal voltages during discharge. Stripping peaks shift downward as well, closing the gap with the primary de-intercalation envelope.
| Temperature | Charge Rate | Stripping Peak Voltage | Plated Capacity | Stripped Capacity | Coulombic Recovery |
|---|---|---|---|---|---|
| 25 °C | 1.0C | None Detected | 0.00 Ah | 0.00 Ah | Not Applicable |
| 25 °C | 2.5C | 3.58 V | 1.85 Ah | 1.42 Ah | 76.8 % |
| 10 °C | 1.5C | 3.54 V | 2.90 Ah | 1.98 Ah | 68.3 % |
| 0 °C | 1.0C | 3.51 V | 3.80 Ah | 2.24 Ah | 58.9 % |
| -10 °C | 0.5C | 3.47 V | 4.65 Ah | 2.18 Ah | 46.9 % |
Resolving that stripping peak depends heavily on the temperature chosen for the diagnostic discharge. Discharging at C/20 and 25 degrees Celsius following a cold fast charge maximizes peak separation: lower ohmic drop moves graphite de-intercalation back up to higher cell voltages, leaving the stripping peak isolated at the onset of discharge. If discharge is run cold instead, polarization merges the stripping peak into the broad stage 1 envelope, frustrating quantitative evaluation.
- Chemical Passivation Loss converts active metallic lithium into lithium carbonate and alkyl carbonates through parasitic electrolyte consumption.
- Mechanical Isolation pinches off dendritic deposits during stripping contraction, stranding metallic particles in the separator matrix.
- Thermal Destabilization lowers the onset temperature of separator melting when stranded microstructures react exothermically during high-rate cycling.
Isolating the fraction of plated lithium that chemically re-intercalates into under-lithiated graphite during pre-discharge rests remains difficult, as its signature blends into baseline electrochemical dissolution signals.

Foil
Current distribution across the negative copper substrate dictates where electrodeposition actually begins. In wound cylindrical and stacked prismatic designs, the graphite coating extends 1.0 to 2.5 millimeters past the cathode boundary ~ an anode overhang intended to prevent cathode edges from facing bare foil. While this overhang functions as a sink for stray lithium ions, geometric misalignment concentrates local flux.
At high charge currents, electric field lines wrap tightly around the cathode margin, driving up current density along the coating boundary.
This local current crowding pulls the edge potential into plating territory well before the bulk coating nears full saturation, lining the margins of the copper collector with metallic film. Because differential capacity analysis records the integrated response of the entire jelly-roll or stack, an edge deposit covering less than 0.5 percent of total coating area shows up as little more than a slight shoulder on the full-cell dQ/dV trace ~ a feature standard smoothing filters regularly discard as instrumental noise.
Underwriters Laboratories standard 1642 conditions nullify safety certifications whenever uncontained metallic deposition triggers internal micro-shorting across separators.
Substrate thickness also plays a role in heat dissipation and in-plane potential drops. Ten-micron copper foils exhibit higher lateral resistance than twelve-micron stock, sharpening thermal gradients during fast transients where current crowding accelerates dendritic growth. Because cooler zones impose higher charge-transfer resistance, current diverts toward warmer central regions while edge geometries remain pinned to concentrated field lines, escalating thermal runaway risks.
Low-voltage differential capacity anomalies at cell margins are framed alternatively as benign geometrical overhang relaxation or as destructive metallic lithium formation.

Inspection
Receiving inspection on high-power graphite cells relies on non-destructive differential capacity screening during lot qualification, where diagnostic success hinges on voltage precision. Cyclers with 16-bit analog-to-digital converters provide roughly 76 microvolts of resolution across a 5-volt span, whereas 24-bit hardware resolves steps down to 0.3 microvolts. Without filtering, numerical differentiation amplifies bit-level quantization noise into artifact peaks that distort capacity integration.
Data processing employs Savitzky-Golay polynomial filtering to smooth raw V(t) and Q(t) arrays prior to differentiation. A second-order polynomial evaluated over a 15 to 25 millivolt window suppresses high-frequency flutter without shaving down the stripping peak’s true amplitude. Fitting continuous cubic splines before taking analytical derivatives serves as an alternative route.
In either case the margin is slim: oversmoothing erases the metallic stripping signature altogether, while undersmoothing leaves quantization noise in place.

Are Relaxation Voltage Curves Reliable Validation?
Open-circuit relaxation curves provide an independent check on differential capacity findings. Once fast charging cuts off, a cell bearing metallic deposits shows an extended rest plateau: mixed potentials between the graphite matrix and surface metal pin negative-electrode potential near zero volts versus Li/Li+ until the deposit intercalates or passivates completely. Differentiating this relaxation curve with respect to time ~ plotted as dV/dt against log time ~ yields a distinct inflection marking the moment surface metal disappears.
| Cell Format | Nominal Capacity | Converter Resolution | Sampling Interval | Filter Window | Detection Limit |
|---|---|---|---|---|---|
| Cylindrical 21700 | 5.0 Ah | 24-bit | 100 ms or 1 mV | 15 mV polynomial | 15 mAh |
| Cylindrical 4680 | 25.0 Ah | 24-bit | 100 ms or 1 mV | 20 mV polynomial | 80 mAh |
| Prismatic Hard-Case | 65.0 Ah | 16-bit | 500 ms or 2 mV | 25 mV polynomial | 350 mAh |
| Pouch Cell | 40.0 Ah | 24-bit | 200 ms or 1 mV | 15 mV polynomial | 90 mAh |
Qualification sequences map safe operating boundaries across current and temperature matrices. Automated test channels step charging rates upward in 0.25C increments at target temperatures, running subsequent C/20 diagnostic discharges to monitor the differential capacity response. The appearance of a low-voltage stripping peak flags the threshold C-rate where plating begins.
These transition points assemble into multi-dimensional charge envelopes programmed directly into battery management system firmware.
- Baseline Normalization establishes reference C/20 dQ/dV curves across target temperature bands on factory-conditioned cells.
- Stress Cycling runs targeted fast-charge profiles in thermal chambers to force overpotentials past intercalation thresholds.
- Isothermal Rest Control holds core cell temperature at 25 degrees Celsius for two hours to dissipate internal gradients before readout.
- Diagnostic Discharge runs a low-rate discharge to the lower cutoff voltage to quantify stripping peak area.
- Lot Disposition quarantines batches that breach allowable stripping thresholds to protect pack yields.
On legacy cyclers, digital noise often swallows subtle electrodeposition signals whenever voltage sampling intervals exceed five millivolts.

Ledger
Undetected lithium plating rewrites warranty exposure and liability reserves. Fast-charged cells that repeatedly plate develop dendrites capable of puncturing porous polymer separators, where a single micro-short inside a 100 kilowatt-hour industrial storage unit or passenger EV pack can trigger catastrophic exothermic venting. With automotive battery recalls running into several hundred million dollars per event, an undetected plating vulnerability readily bankrupts pack assemblers.
Scrap and rework allowances in contract manufacturing agreements track incoming inspection rigor directly. A procurement team without differential capacity screening routinely signs off on lots with marginal electrode tortuosity or non-uniform binder distribution ~ defects that induce premature plating under standard fast charging. Coulombic efficiency curves mask the problem early on, as dead lithium buildup reads as mild, linear fade over the first one hundred cycles before tipping into severe nonlinear knee degradation.
Buyers protect capital outlays by writing quantitative differential capacity metrics directly into cell specifications, bounding allowable reversible stripping capacity during lot qualification. If a five-cell sample drawn from a forty thousand unit container shows a stripping peak over 0.5 percent of nominal capacity under 1.5C charging at 10 degrees Celsius, the buyer turns the shipment away at the port of entry and bills qualification testing back to the manufacturer.
International Electrotechnical Commission standard 62660-1 section 6.2 acceptance criteria bind vendors to explicit high-rate capacity retention thresholds, forcing suppliers to replace rejected lots whenever incoming differential capacity scans detect sub-surface metallic deposition during qualification.


