Anode-free lithium batteries take “delete the anode” to its extreme: instead of shipping graphite or a lithium foil, the cell starts with bare copper on the negative side. All the lithium comes from the cathode on first charge and plates itself onto that copper. Removing the graphite host lifts both gravimetric and volumetric energy density, since graphite typically accounts for over a quarter of a cell’s stack mass while contributing none of its energy.
These cells aren’t missing an anode as a component: they’re assembled, and go through formation, with a separator, a casing and a bare copper current collector sitting exactly where an anode would sit. What’s absent at assembly is anode active material, no graphite bonded to that copper, no lithium foil laminated onto it. The copper does nothing electrochemically until the first charge plates metal onto it. “Anode-free” describes that missing active-material layer during assembly, not a missing electrode.
That absence is also the whole problem. A normal lithium-metal cell ships with a thick excess foil that quietly absorbs years of inefficiency: lithium lost to SEI formation, dead metal, dendrites, all of it just eating into a reservoir sized to have some spare material. An anode-free cell has no such reservoir. Every irreversible loss comes straight out of the cathode’s one-time lithium budget, so SEI reformation, mossy growth and dendritic plating on bare copper translate almost directly into cycle life.
Why not just plate in a reserve from the start, the way conventional lithium-metal cells do? Because that gives back most of what going anode-free was meant to win. A shipped-in excess of lithium is inactive mass and volume from day one, only a fraction of which ever becomes usable capacity, which undercuts the same energy-density gain that motivated deleting graphite in the first place. Bulk lithium foil also has its own manufacturing headaches: soft, reactive, awkward to laminate uniformly in a way graphite slurry isn’t. So the field keeps the design genuinely zero-excess and tries to engineer around the loss mechanisms directly, which is exactly the gap this paper targets. Instead of a shipped-in reserve, it grows a small, self-generated one out of the cathode’s own first-charge capacity.
A newly accepted manuscript from WMG at the University of Warwick and the Faraday Institution’s degradation project, due in Faraday Discussions, takes a specific run at that problem. What’s notable isn’t a new material. The authors get meaningful cycle-life improvement from a cell built entirely from ordinary parts: a cobalt-free cathode, unmodified copper foil, standard dual-salt electrolyte. The lever is entirely in how the cell is charged.
S. Bolloju, G. Bree, J. Zhao, B. I. J. Johnston, C. Clenahan, D. Proprentner and L. F. J. Piper, “Anode-Free LiNiO₂ Battery: A High Energy Density, Low-Cost Cell Design,” Faraday Discussions, Royal Society of Chemistry, Accepted Manuscript, published online 14 July 2026.
This post has been built based on the support of and sponsorship from: illumion, Custom Power, Quarto Technical Services, Eatron Technologies and About:Energy
A newly accepted manuscript from WMG at the University of Warwick and the Faraday Institution’s degradation project, due in Faraday Discussions, takes a specific run at that problem. What’s notable isn’t a new material. The authors get meaningful cycle-life improvement from a cell built entirely from ordinary parts: a cobalt-free cathode, unmodified copper foil, standard dual-salt electrolyte. The lever is entirely in how the cell is charged.
Why LiNiO2
Most anode-free research leans on cobalt-containing NMC cathodes, which makes sense given how much baseline data has been built on them. This paper however uses LiNiO₂ (LNO): cobalt-free, theoretical capacity around 275 mAh/g, average voltage near 3.8 V vs. Li/Li⁺. What matters most for an anode-free design is that LNO delivers over 220 mAh/g without needing to be pushed past 4.3 V, a meaningfully safer ceiling than full lithium extraction would require.
That matters more here than in a conventional cell, because in an anode-free design the cathode supplies the anode’s entire working lithium inventory on first charge, not just that cycle’s capacity. Designing a cathode that can initially hand over a large fraction of its lithium without cracking or triggering electrolyte breakdown is challenging. Dropping cobalt reduces cost and supply-chain exposure.
The trick: don’t charge formation and cycling the same way
The core idea is to stop treating formation and normal cycling as the same voltage window repeated indefinitely. The authors split them:
- Formation (first two cycles), 3.0 to 4.6 V. A much wider window than the cell will ever see again, deliberately over-delivering lithium so more plates onto the bare copper than that cycle strictly needs. Call it a deposit, not a discharge: roughly 42 mAh/g of extra capacity accessed this way versus stopping formation at 4.1 V.
- Everyday cycling, 3.0 to 4.1 V. Once the reserve is down, the ceiling drops, to stop repeatedly driving the LNO particles through a phase transition (H2 to H3, in the layered-oxide literature) associated with microcracking and surface breakdown at higher voltages.
Seeding a lithium buffer and protecting the cathode’s crystal structure are two different jobs. Decoupling them lets each phase be tuned for what it’s actually trying to achieve.
Watching it happen: cryo-FIB-SEM
Rather than infer the reservoir from capacity numbers alone, the group physically sectioned the retrieved lithium electrodes using cryo-plasma-FIB-SEM (milled and imaged cold, to avoid melting or reacting the lithium). After the first 4.6 V charge, the plated layer measured around 28 µm thick, with porosity around 44% at the electrode centre, pointing to a loose, fibrillar structure. After one full cycle back down through the 4.1 V window, that layer thinned to roughly 10 µm, about 0.56 mAh/cm² left standing as reserve, having stripped 2.70 mAh/cm² out in that first discharge.
There’s a second payoff. Plating lithium onto bare copper costs energy to nucleate, showing up here as a roughly 7 mV dip in anode potential before growth settles into a steady plateau. With a residual layer already present, later plating happens onto lithium rather than copper, and that penalty disappears. The reserve also keeps electrolyte off the bare copper, which the authors argue limits galvanic corrosion, a second-order effect but a real one over dozens of cycles.
Does it actually cycle better?
At coin-cell scale, the two-step protocol modestly outperformed the alternatives: roughly 62% retention over 40 cycles against 55 to 56% for cells held to a single window throughout. Coulombic efficiency followed the same order: 99.2% for the two-step protocol against 98.1 to 98.4% for the single-window cells.
A 2024 study from Manthiram’s group reached a considerably higher 76% retention over 100 cycles from an anode-free LNO cell, but only by combining electrochemical pre-lithiation (a 40% lithium excess up front) with a localized high-concentration electrolyte, both adding cost and process complexity. This paper attempting to make a simpler cell: no pre-lithiation, no exotic electrolyte, no current-collector treatment. Instead engineering this using a smarter charge protocol. This is a more interesting approach that could also be a more realistic route to volume production.
From coin cells to something closer to a pack
Coin cells are a poor stand-in for real plating behaviour, mostly because mechanical pressure across a coin cell’s active area is notoriously uneven: the group’s own cross-sections show 44% porosity at the centre against 25% at the edge of the same nominal cell.
To get past that, they built single-layer pouch cells (35 cm² and around 93 mAh) and cycled them under a genuinely applied 1 MPa of stack pressure, same protocol throughout. Result: 68% retention after 90 cycles for the pressurized pouch cell against 37% for an unpressurized coin cell over the same span. Roughly double the number of cycles, with tighter coulombic efficiency too (99.45 ± 0.44% pouch versus 98.81 ± 1.17% coin). Impedance spectroscopy backs up why: both SEI-related and charge-transfer resistances grow faster in coin cells, consistent with pressure keeping lithium deposition denser and more columnar, with less dead, isolated lithium accumulating over time.
There’s also a distinctive decay shape: capacity falls slowly for roughly the first 50 to 60 cycles, then hits a knee and drops sharply. The authors’ read is that the engineered lithium reserve absorbs per-cycle losses during the slow phase; once it’s used up, failure accelerates fast.
What this means for energy density
Removing the graphite host and running at a higher average discharge voltage (3.85 V for the anode-free cell against 3.71 V for an equivalent graphite full cell, similar specific capacity) adds up to roughly 319 Wh/kg and 1120 Wh/L at the coin-cell stack level, against about 240 Wh/kg and 589 Wh/L for the graphite benchmark: a 33 to 36% gravimetric jump and nearly double the volumetric density.
Worth flagging because it’s easy to read past: these are stack-level figures (cathode, anode current collector, separator, electrolyte only). Tabs and pouch or can housing are excluded, standard practice at this stage but a reminder that these numbers sit above what a packaged commercial cell would deliver.
It also rewrites what “formation” and “ageing” mean
On a conventional line, formation and ageing are steps applied to an anode that already physically exists: electrode manufacturing (coating, calendering, slitting) happens on precision equipment well before assembly.
- Formation: the first controlled charge after electrolyte fill, conditions a surface that’s already there, growing SEI on graphite that already had known, uniform porosity.
- Ageing: the rest period that follows, mostly lets that SEI settle and gives marginal cells time to reveal abnormal voltage decay before grading sorts cells by capacity and impedance.
None of that separation survives an anode-free cell. Formation isn’t conditioning a pre-built anode, it’s building one for the first time inside an already-sealed cell. Every variable that shapes the resulting microstructure (local current density, cell temperature, and mechanical pressure, which this paper shows swinging porosity from 25% to 44% within one coin cell) is now baked into the electrode during formation rather than tuned beforehand on a coating line. Tolerances that used to matter mainly for SEI uniformity now decide whether the anode plates dense and columnar or porous and dendritic, which the pouch-versus-coin result shows is close to the single biggest lever on cycle life.
Ageing changes character too. The rest-and-grade step normally screens for micro-shorts against decades of graphite baseline data. Here, that same window is also the first chance to learn whether the freshly-plated reserve is stable or already reacting away, with no equivalent bank of production history to say what “normal” self-discharge should look like for an anode that’s a few hours old. Grading has a sharper version of the same problem: the thing actually worth grading on, reserve size and morphology, is exactly what this paper needed cryo-FIB-SEM to see, a lab technique, not something usable at line throughput. What stands in for it at production scale (coulombic-efficiency measurement, impedance signatures, ultrasonic detection of plated lithium) is still an open question.
That’s a step beyond what this paper sets out to argue, but it follows from what it shows: if reserve size set during formation determines when a cell hits its knee-point, and the slow-decline-then-knee shape here suggests it does, formation has effectively pre-determined the whole degradation curve before the cell has cycled once.
Thoughts
The mechanism isn’t nailed down: It’s unclear whether the retention improvement comes mainly from a bigger physical reserve, a more stable SEI or cathode-electrolyte interphase forming under the higher formation voltage, or both; the paper states plainly that untangling the two is still open. If it’s mostly SEI chemistry, a much smaller, cheaper buffer might get most of the benefit.
All the direct physical evidence is on the lithium/copper side: The imaging is genuinely strong, but it’s all anode-side. That a lower cycling ceiling protects the cathode’s structure is argued from prior literature on that phase transition, not shown in this paper’s own cycled cathodes. A post-mortem cathode cross-section comparing the two protocols would have closed that loop.
Two cells per condition?: A minimum of two cells per condition is reported across four voltage-window variants plus the coin-versus-pouch comparison, with best performers plotted. That’s not enough to know whether the several-point gaps between protocols are a real voltage-window effect or cell-to-cell noise, especially given this study’s own evidence that local porosity within one coin cell can swing from 25% to 44%. But the fix isn’t simply building five cells per condition instead of two; that mostly re-confirms points already likely fine. More useful at this stage would be widening the parameter space actually explored (more formation ceilings, cycling windows, pressure levels) and using outlier-detection tooling, now routine in battery data analysis, to flag anomalous cells or cycles, then spending the retest budget on just those flagged points. Same number of cell-builds, a richer map of the design space.
90 cycles is very low: The authors are upfront that this is a stepping stone: lithiophilic copper coatings and tailored electrolytes are named as the obvious next additions. Anyone picturing a drop-in satellite or EV cell from the headline energy-density numbers should keep that gap in view.
Everything runs at low rates: Formation at C/20, cycling at C/10 charge and C/3 discharge, is reasonable for isolating a protocol effect cleanly but well below real pack rates. Lithium plating tends to get messier, more mossy, more dendritic, as rate goes up, and it’s open whether the reserve strategy holds at speed.
Final Summary
This paper argues a decent chunk of anode-free cycle life can be bought with a charging-protocol change rather than a materials one. A protocol change is close to free to apply to a design already in production; a new coating or exotic electrolyte is a supply-chain and qualification project.
That’s a different and interesting economics story.
References
- S. Bolloju, G. Bree, J. Zhao, B. I. J. Johnston, C. Clenahan, D. Proprentner and L. F. J. Piper, “Anode-Free LiNiO₂ Battery: A High Energy Density, Low-Cost Cell Design,” Faraday Discussions, Royal Society of Chemistry, Accepted Manuscript, published online 14 July 2026.
- P. Vanaphuti, L. Su and A. Manthiram, “Effect of Electrochemical Pre-Lithiation on Layered Oxide Cathodes for Anode-Free Lithium-metal Batteries,” Small Methods, 2024, 8, 2301159. https://onlinelibrary.wiley.com/doi/abs/10.1002/smtd.202301159
Battery Materials and Cells Group
BMAC is a unique, all-in-one battery research group, offering a complete development pipeline from materials research and cell prototyping to scale-up and advanced metrology through our state-of-art facilities within Energy Innovation Centre (EIC) and Wellesbournecampus. Our comprehensive approach accelerates innovation from lower to higher technology readiness levels (TRLs), all under one roof. Our expertise spans the entire battery lifecycle, from materials research and prototyping to scaled-up production and in-depth analysis.
We leverage cutting-edge electrochemical techniques and materials science knowledge to develop innovative solutions for energy storage. Our research goes beyond lithium-ion, exploring promising next-generation chemistries to meet the demands of a sustainable future.