Solid-Electrolyte Interphase (SEI)

A skin that has to grow before the battery can work

Charge a lithium-ion cell for the very first time and something quietly goes wrong on purpose. As lithium ions flood into the graphite anode, the anode's potential drops to about 0.1 V versus a lithium-metal reference — deep inside the region where the liquid electrolyte is no longer chemically safe to touch. The carbonate solvents start grabbing electrons from the electrode and falling apart. Their fragments — lithium salts, lithium-bonded oligomers, gas — precipitate onto the graphite as a film a few dozen nanometers thick. That film is the solid-electrolyte interphase.

Here is the elegant part. The SEI is built from reduced electrolyte, so it is an electronic insulator: once it covers the electrode, electrons can no longer leak out to reduce more solvent, and the destructive reaction shuts itself off. But the film is also a decent ionic conductor: Li⁺ ions can hop through it. So the cell ends up with a barrier that blocks the reaction that would kill it while still passing the ions that make it work. The anode has, in effect, grown its own protective skin.

The catch — and the reason every battery engineer has a love-hate relationship with the SEI — is that the lithium locked into that skin never comes back. It came out of the cathode, it is now bonded into LiF and Li₂CO₃ on the anode surface, and it can no longer carry charge. The protective film is paid for in permanently lost capacity.

The reduction chemistry: where the film comes from

Commercial electrolytes are LiPF₆ dissolved in a mix of cyclic and linear carbonates — ethylene carbonate (EC) plus dimethyl, ethyl-methyl, or diethyl carbonate. EC is the key SEI former. Below about 0.8 V vs Li/Li⁺ it accepts electrons and ring-opens. The single-electron pathway gives a radical anion that couples to form lithium ethylene dicarbonate (LEDC), the dominant organic component of the outer SEI, plus ethylene gas:

2 EC + 2 Li⁺ + 2 e⁻  →  (CH₂OCO₂Li)₂  +  C₂H₄↑
                          lithium ethylene dicarbonate (LEDC)

A competing two-electron pathway drives all the way to the simplest inorganic carbonate, building the dense inner layer that sits directly on the graphite:

EC + 2 Li⁺ + 2 e⁻  →  Li₂CO₃  +  C₂H₄↑

Trace water and the LiPF₆ salt supply the fluorine. PF₆⁻ hydrolyzes and decomposes to release HF and PF₅, and the fluoride ends up as lithium fluoride (LiF) — the mechanically stiff, highly insulating component that the best SEIs are rich in:

LiPF₆        ⇌  LiF  +  PF₅
PF₅ + H₂O    →  POF₃  +  2 HF
2 HF + Li₂CO₃ →  2 LiF  +  H₂O + CO₂↑

The net result is a layered structure — known as the mosaic model (Peled) — with a dense inorganic film (LiF, Li₂CO₃, Li₂O) hugging the electrode and a porous organic film (LEDC, semicarbonates, polymers) facing the electrolyte. The reductions happen at characteristic potentials, so they show up as a distinct plateau on the very first charge curve that is absent on every cycle afterward.

Why the electrolyte gets reduced: the window argument

Whether the SEI forms is a question of energy levels, not of any specific molecule. A battery electrolyte has an electrochemical stability window bounded by the energy of its lowest unoccupied molecular orbital (LUMO) and highest occupied molecular orbital (HOMO). If the anode's electrochemical potential (its Fermi level) sits above the electrolyte's LUMO, electrons spill from electrode into electrolyte and the solvent is reduced. That is exactly the case for lithiated graphite:

Couple / level	Potential vs Li/Li⁺	Consequence
Lithiated graphite (LiC₆)	~0.10 V	Below the electrolyte LUMO → reduces solvent
EC reduction onset	~0.8 V	SEI begins to deposit on first charge
Carbonate electrolyte LUMO (effective)	~1.0 V	Lower edge of the stability window
Li plating threshold	0.0 V	Below this, Li metal deposits — runaway SEI growth
Typical LiCoO₂ cathode	~3.9 V	Near the oxidative (HOMO) edge → CEI on cathode

The driving force is large: the graphite sits roughly 0.9 V below the electrolyte's reduction onset. With Faraday's constant F = 96 485 C/mol, that 0.9 V overshoot corresponds to a per-electron driving energy of ΔG = −nFE ≈ −(1)(96 485)(0.9) ≈ −87 kJ/mol — comfortably enough to make solvent reduction spontaneous. The SEI is the kinetic firewall that stops thermodynamics from consuming the whole electrolyte: once the film is thick enough to block electron tunneling (a few nm) and electron transport, the reaction starves itself.

The arithmetic of irreversible capacity loss

The lithium spent on the SEI is bookkept as first-cycle Coulombic efficiency — the ratio of charge you get back out on the first discharge to the charge you put in on the first charge:

CE₁ = Q_discharge / Q_charge

Graphite, good electrolyte:   CE₁ ≈ 90–95%   →  5–10% Li lost to SEI
Silicon anode (uncoated):     CE₁ ≈ 70–85%   →  15–30% lost on cycle 1
Lithium metal (bare):         CE₁ ≈ 90–99%/cycle, but lost EVERY cycle

That first-cycle hit is a one-time tax on graphite, but the steady-state per-cycle efficiency is what sets calendar and cycle life. Suppose a graphite cell averages 99.95% Coulombic efficiency once formed — 0.05% of the lithium inventory is lost to ongoing SEI repair each cycle. After 1 000 cycles the surviving capacity is:

Q_remaining / Q_initial = (0.9995)^1000 = 0.607  →  ~61% capacity left

Drop the efficiency to 99.9% and the same 1 000 cycles leave only (0.999)^1000 ≈ 37%. Every extra digit of Coulombic efficiency — every nanometer of SEI that doesn't have to be rebuilt — is the difference between a battery that lasts 3 years and one that lasts 10. This is why a stable, self-limiting SEI is worth more to cell life than almost any other single property.

When the skin keeps tearing: silicon and lithium-metal anodes

On graphite the SEI is a one-time investment because graphite barely moves — it expands about 10% on full lithiation, so the film stays intact for thousands of cycles. The two anodes that would massively raise energy density both break this deal:

• Silicon stores ~10× more lithium per gram than graphite (theoretical 3 579 mAh/g for Li₁₅Si₄ vs 372 mAh/g for LiC₆) but swells roughly 300% on lithiation. The brittle SEI cracks open every cycle, exposes bare silicon to electrolyte, and re-forms — a fresh lithium tax on every single cycle. Uncoated silicon's Coulombic efficiency starts low and the capacity bleeds away within tens of cycles.
• Lithium metal has the highest possible anode capacity (3 860 mAh/g) and the lowest potential, but it has no host lattice — lithium deposits directly onto the surface. A non-uniform SEI funnels current to local hot spots, growing needle-like dendrites that can pierce the separator and short the cell, or break off into electrically isolated "dead lithium" wrapped in SEI. Taming the lithium-metal SEI is the central obstacle to commercial lithium-metal and "anode-free" cells.

The fixes are mostly SEI engineering: silicon is buffered with carbon coatings and FEC-rich electrolytes that build a stretchy, LiF-rich film; lithium-metal work pushes toward fluorinated and concentrated electrolytes, artificial SEI coatings, and solid electrolytes that physically block dendrites.

SEI vs the cathode-side film (CEI)

The same instability happens at the other electrode, but in reverse — oxidation instead of reduction — producing a thinner, less-studied cathode-electrolyte interphase (CEI). Comparing them clarifies what makes the anode SEI so dominant:

	SEI (anode)	CEI (cathode)
Driving reaction	Electrolyte reduction (gains e⁻)	Electrolyte oxidation (loses e⁻)
Forms when potential is	Below ~1 V vs Li/Li⁺	Above ~4.3 V vs Li/Li⁺
Typical thickness	10–50 nm	1–10 nm
Main components	LiF, Li₂CO₃, Li₂O, LEDC	LiF, polycarbonates, metal fluorides
Lithium consumed	5–10% on first cycle (large)	Small
Dominant failure role	Capacity fade, impedance rise, dendrites	Transition-metal dissolution, gas
Controlled by	EC/FEC/VC additives, formation protocol	High-voltage additives, coatings

Both are passivation films born from the electrolyte being outside its stability window. The anode SEI is thicker and far more lithium-hungry simply because the reductive driving force at ~0.1 V is much larger than the oxidative one at the cathode — and because graphite's surface area is enormous.

Where the SEI decides whether a battery lives or dies

• The factory "formation" step. Before a cell is sold, manufacturers charge it slowly and carefully — sometimes for many hours, sometimes at elevated temperature — purely to grow a clean, uniform SEI. A bad formation protocol leaves a patchy film that keeps consuming lithium for the cell's whole life. Formation is one of the most expensive and time-consuming steps in battery manufacturing precisely because the SEI is built only once.
• Electrolyte additives. A few percent of vinylene carbonate (VC) or fluoroethylene carbonate (FEC) reduces before the bulk solvent and seeds a tougher film. FEC builds a LiF-rich SEI that can extend silicon-anode cycle life roughly tenfold; VC builds a flexible polymeric film prized in graphite cells. These additives are sacrificial by design — they are consumed building the SEI.
• Cold-weather fast charging. At low temperature Li⁺ moves sluggishly through both the SEI and the graphite, so a given charging current drives the anode surface to a more negative potential. Push it below 0 V and lithium plates as metal, instantly reacting with electrolyte to grow runaway SEI and seed dendrites. This is why electric vehicles throttle charge rate in the cold — it is direct SEI protection.
• Calendar aging. Even a battery sitting unused at full charge slowly thickens its SEI, raising internal resistance and bleeding capacity. SEI growth roughly follows a √t law because the film itself is the rate-limiting transport barrier for its own further growth — a self-slowing process that nonetheless never fully stops.

Common misconceptions and pitfalls

• "The SEI is bad, we should prevent it." Without an SEI the electrolyte would reduce on the graphite forever and the cell would self-discharge to death. The goal is not to prevent the SEI but to make it thin, uniform, self-limiting, and built once. A perfect cell has a good SEI, not no SEI.
• "The SEI is a single uniform layer." It is a heterogeneous mosaic — a dense inorganic inner film (LiF, Li₂CO₃, Li₂O) under a porous organic outer film — and its composition varies across the surface and changes with cycling, temperature, and electrolyte. Models that treat it as one homogeneous slab miss the cracking and re-formation that drive real degradation.
• "Lost capacity from the SEI can be recovered." The lithium bonded into SEI salts is gone. You can pre-lithiate the anode (add sacrificial lithium up front) to compensate for the loss, but you cannot extract the lithium already locked in the film.
• "It only matters on the first cycle." The big visible hit is on cycle one, but ongoing SEI repair on every subsequent cycle is what sets long-term fade. A 0.1% per-cycle SEI tax sounds trivial and is fatal over 1 000 cycles.
• "Imaging the SEI is easy." The film is electron-beam fragile and decomposes in a standard TEM. It took cryo-electron microscopy — freezing the SEI to liquid-nitrogen temperatures — to image its true nanostructure intact, which is why detailed SEI structure was contested for decades.
• "Higher voltage charging is always better for energy." Charging the anode harder or faster pushes the surface toward Li plating and accelerates SEI growth and dendrite risk. There is a kinetic ceiling set by SEI and graphite transport, not just by the thermodynamic cell voltage.