Saying a battery has degraded by 12 per cent describes a symptom. It says nothing about which mechanism produced it, whether the rate will hold, or what to change. Professional degradation work starts by separating the symptom into modes.
The three modes
Dubarry and colleagues set out the classification that most of the field now uses, and it is worth learning because it converts a vague observation into a diagnosis.
Loss of lithium inventory. Cyclable lithium is consumed by side reactions and is no longer available to shuttle between electrodes. The electrodes themselves remain intact. Capacity falls, and the balance between the two electrodes shifts, which changes where each one sits in its own state of charge window.
Loss of active material. Electrode material stops participating, whether through particle fracture, contact loss from the binder or current collector, or structural transformation into an inactive phase. This is tracked separately for the anode and the cathode, because the consequences differ.
Resistance increase. Charge transfer, ionic transport or electronic conduction gets worse. Capacity at low rate may be almost unchanged while power capability falls.
Every mechanism below feeds one or more of these modes. That mapping is what makes the classification useful.
Anode-side mechanisms
Continued SEI growth. The solid electrolyte interphase forms during the first charge, when electrolyte reduces at potentials the solvent cannot survive. The film passivates the surface, which is what makes graphite anodes work at all. It is not perfectly passivating, so it keeps growing, slowly, forever; in storage the growth rate typically slows with roughly the square root of time, because the thickening film limits its own reactant supply. Each increment consumes lithium and adds a small resistance. This is the dominant path in cells at rest, it follows Arrhenius temperature dependence, and it accelerates as the anode potential falls, which is to say at high state of charge. It maps to loss of lithium inventory and, more slowly, to resistance increase.
Lithium plating. When lithium ions arrive at the graphite surface faster than they can intercalate, the potential drops below zero volts against lithium and metallic lithium deposits instead. Triggers are low temperature, high charge rate, high state of charge and an anode whose available surface has been reduced by earlier ageing. Some plated lithium re-intercalates on rest, which is why a plated cell can show partial recovery. The rest reacts to form new SEI, becomes electrically isolated dead lithium, or grows into structures that can eventually reach the separator. It maps to loss of lithium inventory, and it is the mechanism that connects degradation to safety.
Particle cracking and mechanical fatigue. Graphite expands roughly 10 per cent on lithiation. Silicon expands by roughly 300 per cent, which is why silicon content stays modest despite the energy benefit. Repeated expansion cracks particles, exposes fresh surface, and triggers new SEI formation on that surface. It maps to loss of active material and loss of lithium inventory together.
Binder and current collector degradation. Binder decomposition and copper corrosion, the latter accelerated by over-discharge, cause contact loss. Copper dissolved during over-discharge can redeposit as metallic bridges, which is one reason over-discharge protection is treated as a safety function rather than a convenience.
Cathode-side mechanisms
Transition metal dissolution. Manganese in particular dissolves into the electrolyte, especially at elevated temperature and in the presence of acid. It then migrates to the anode and deposits in the SEI, where it catalyses further electrolyte reduction. This cross-talk is why cathode degradation shows up as accelerated anode degradation, and it is a common source of confusion in post-mortem work.
Microcracking in polycrystalline particles. Nickel-rich layered oxides made as polycrystalline secondary particles crack along grain boundaries as the lattice breathes. Cracks admit electrolyte, which reacts with fresh cathode surface. Single-crystal cathode morphologies exist largely to address this.
Surface reconstruction. The layered structure at the particle surface transforms towards spinel and then rock-salt phases, which are poor lithium conductors. The effect is a growing resistive skin, so it maps mainly to resistance increase.
Electrolyte oxidation at the cathode. At high potential the electrolyte oxidises, forming a cathode-side interphase and generating gas. It scales strongly with upper cut-off voltage, which is why raising the charge voltage to buy capacity is always a trade against life.
Electrolyte and separator
Fluorinated salts hydrolyse in the presence of trace water to produce hydrofluoric acid, which attacks both electrodes and accelerates metal dissolution. This is why dry room specifications during manufacturing are strict, and why moisture ingress in a damaged pack matters.
Separator pores clog with deposits over time, raising ionic resistance locally. Local clogging concentrates current elsewhere, which produces uneven ageing across the electrode area.
Mechanism summary
| Mechanism | Degradation mode | Accelerated by | Diagnostic signature |
|---|---|---|---|
| SEI growth | Lithium inventory, some resistance | Temperature, high SoC, time | Smooth capacity fade at rest, DVA peak shift |
| Lithium plating | Lithium inventory, safety risk | Low temperature charging, high charge rate, high SoC | Voltage plateau on relaxation, sharp fade onset |
| Particle cracking | Active material, lithium inventory | Deep cycling, high rate, silicon content | ICA peak broadening and height loss |
| Metal dissolution | Both electrodes via cross-talk | High temperature, acid, manganese content | Anode-side metals in post-mortem analysis |
| Surface reconstruction | Resistance | High cut-off voltage, temperature, nickel content | Impedance growth with little capacity loss |
| Separator clogging | Resistance | Time, deposits, temperature | Impedance growth, uneven current distribution |
The knee, and why it matters more than the early rate
Most cells fade close to linearly for a long period, then turn a corner and fade quickly. Attia and colleagues catalogued the proposed causes, and the practical point is that a knee is a threshold crossing rather than a new mechanism appearing from nowhere.
Common candidates. Anode porosity falls as SEI and plated material fill the pore structure, local current density rises, and plating begins in earnest, which fills more pores. Or loss of active material at one electrode proceeds until that electrode, rather than lithium inventory, becomes the limit, at which point the cell's usable window collapses quickly. Or resistance rise reaches the point where the cell hits its cut-off voltage early in every cycle.
For anyone modelling asset life, this is the whole game. Extrapolating a linear early fade rate to a warranty endpoint will overestimate life if a knee is coming, and no amount of extra data from the linear region tells you when. Accelerated testing that never reaches the knee is a common and expensive mistake.
Telling mechanisms apart from real data
Incremental capacity analysis differentiates capacity with respect to voltage during a slow cycle. Peaks correspond to phase transitions in the electrode materials. Peak positions shifting together indicates lithium inventory loss; individual peaks shrinking indicates active material loss at the associated electrode.
Differential voltage analysis does the inverse and often works better on flat-curve chemistries such as LFP, where incremental capacity peaks are hard to resolve.
Electrochemical impedance spectroscopy separates ohmic, interfacial and diffusion contributions to resistance, which localises where the resistance is growing.
Voltage relaxation after charge carries a plating signature. Re-intercalation of plated lithium produces a characteristic plateau during rest, which can be detected without disassembly and is used in some adaptive charging strategies.
Post-mortem analysis remains the reference. Disassembly in a glovebox followed by microscopy, diffraction and elemental analysis is the only way to confirm what the non-invasive methods inferred, and it is how the non-invasive methods were validated in the first place.
Informational and educational content only. Not professional, financial, legal, or engineering advice.