Executive take
When a formulation that looked strong in the lab dies at pilot, the autopsy almost never finds broken chemistry. It finds variables the lab program never exercised: residual water at production levels instead of glovebox levels, electrolyte quantities cut by an order of magnitude, formation schedules compressed to line takt, raw-material lots that drift inside specification. Cell production during ramp-up runs scrap rates anywhere from under 5% to as high as 90% [2], and the difference is usually not the recipe — it is whether the transfer program tested the recipe's sensitivities before the pilot line did.
This note is for teams deciding whether lab data on a formulation, a sample, or a supplier claim justifies committing pilot-line time. The short answer: not until the five failure modes below have been deliberately stressed at small scale.
The decision or failure mode
The situation is standard. A candidate formulation — an electrolyte package, an electrode system, a coated cathode — performs well in coin cells or single-layer pouches. A decision follows: book pilot-line slots, order materials in kilogram quantities, commit engineering time.
The trap is structural. Lithium-ion manufacturing is a chain of consecutive, tightly interacting process steps [1], and changing one parameter propagates downstream in ways lab equipment does not reproduce [2]. Lab cells are built with deliberate excesses — electrolyte, lithium inventory, relaxed schedules — that "can readily mask practical performance" [4]. So the lab result is real, and the pilot failure is real, and both are measuring different systems.
Getting this wrong typically costs one to two development cycles: months of pilot access, materials at scale-up quantities, and the engineering time to run a failure investigation that could have been a two-week lab study.
Technical analysis
1. The test ladder lies by omission
A lithium-metal coin cell can carry roughly 30 g of electrolyte per Ah; a 78 Ah automotive pouch cell runs near 1 g/Ah [4]. Published cell-level results for new materials are generated under conditions — electrolyte amount, lithium thickness, cathode loading — that "cannot be easily realized in realistic cells" [5]. Even with identical electrodes and electrolyte, moving from coin to pouch format changes impedance substantially, with knock-on effects on rate capability and cycle life [6].
The practical consequence, and one our network has watched repeatedly in large-format development: additive packages qualified in flooded small cells under- or over-dose in multi-layer cells, because consumption rates and spatial distribution do not transfer (EXP-3). Each rung of the ladder — coin, single-layer pouch, multi-layer build — unmasks failure modes the previous rung physically cannot show. Teams that skip rungs to save calendar time buy the skipped rung's failure modes at pilot prices (EXP-5).
2. Trace water is a formulation variable, not a housekeeping detail
LiPF6 reacts with trace water to release HF and further decomposition products that degrade the cell [7], and it is unstable against even ppm-level water — the hydrolysis is governed by the anion's solvation environment in the electrolyte itself [8]. That much is textbook. What the transfer program has to internalize is that water tolerance belongs to the formulation, not the facility.
The Dahn group's LFP/graphite study makes this concrete: electrode drying conditions set residual water (roughly 500 ppm after 100 °C vacuum drying versus 100 ppm after 120 °C), and the damage that water does depends on the additive package. Additive-free electrolyte suffered severe fade and iron dissolution; cells with 2% VC, 2% FEC, or 1% LiPO2F2 tolerated up to 500 ppm of excess water with minimal impact [9]. A formulation that looked moisture-robust in the lab may simply never have met production-level water — or may owe its robustness entirely to an additive that gets consumed doing that job instead of the job it was selected for.
The obvious fix — dry everything harder — has a wall on the other side: the most aggressively post-dried electrodes in a controlled comparison had the lowest residual water and the worst electrochemical performance, from structural damage to the electrode [10]. The drying operating point is a window, not a floor, and it has to be located for each electrode system (our reading; the underlying data is [10]).
Lot-to-lot variation deserves a note of proportion here. In our network's experience, solvent and salt lots inside specification are usually consistent enough not to break a formulation on their own (EXP-4) — but incoming inspection often does not measure the trace impurities that matter for a given additive chemistry, so the residual risk sits exactly where routine QC is not looking. That is an argument for qualifying against more than one lot, not for panic.
3. Filling, wetting, and formation do not scale linearly
Electrolyte filling gets intrinsically harder as formats grow: wetting distances increase while the void volume available for uptake shrinks [12]. A systematic review of the filling/wetting literature (544 records screened) found the field lacks holistic measurement methods and that studies relevant to series production are underrepresented [13] — meaning published lab wetting data gives you little to design a production filling recipe from. In our network's experience, the filling strategy for large-format cells — vacuum profile, staged injection, rest protocol — is a separate engineering deliverable that usually has to be developed, not copied, at pilot scale (EXP-1).
Formation is the second half of the same problem. Wetting plus formation occupies 3–7 days in industrial practice, with aging adding up to two more weeks — one of the most time- and cost-intensive blocks in the plant [11]. Formation protocols tuned on lab cyclers with unconstrained schedules rarely survive contact with line takt unchanged; the SEI-conditioning window narrows with cell size and schedule pressure (EXP-2).
A documented lab-to-pilot upscaling of aqueous-processed NMC622 electrodes shows how these process interactions surface in practice: drying programs had to be redesigned after fast, high-convection drying cracked coatings, and current-collector delamination emerged as a distinct failure at the pilot stage [14]. None of that is exotic chemistry. All of it kills transfer schedules.
Dry-room exposure, at least, is quantifiable rather than mystical: for a ~91% nickel cathode, short moisture exposure during dry-room downtime at −20 °C dew point produced no measurable cell impact [15]. Exposure budgets can be measured and specified — which is exactly why they belong in the transfer package instead of in tribal knowledge.
4. Materials drift inside specification
Scaling active-material synthesis from grams to kilograms and tons exposes what a Nature Energy review bluntly calls blind spots in yields, impurities, and quality control [3]. The mechanism is mundane: specifications are written around the parameters the lab measured, and the parameters that shift at scale — impurity profiles, particle-size distribution, surface area — propagate through slurry rheology and coating density into electrochemical performance [16].
The implication for buyers: a supplier's sample lot and their production lot can both be "in spec" and still behave differently in your cell — because the parameters that shift at scale are often the ones the specification never covered [3]. Qualification against a single good lot is thin evidence.
5. The transfer package records the recipe, not the sensitivities
This one is organizational, and it is ours (ORI). Transfer documentation usually captures compositions, setpoints, and acceptance criteria — all defined on lab equipment. What it rarely captures is the sensitivity map: which variables the formulation is known to be touchy about, what the failure signature looks like when each one moves, and which lab result predicts which pilot behavior. Without that map, every pilot anomaly starts a from-scratch investigation, and the calendar burns.
What changes in practice
Before committing pilot-line time to a formulation, run the stress tests the lab program skipped:
| Stress test | What it does | Failure mode addressed | |---|---|---| | Worst-case-lot qualification | Qualify against spec-edge and multi-lot materials, not the golden sample lot | §4 materials drift | | Moisture tolerance mapping | Test the formulation across the residual-water range your line will actually produce, with and without the additive package | §2 water sensitivity | | Formation window mapping | Sweep formation parameters inside realistic line-takt constraints; find the window edges before the line does | §3 formation transfer | | Mandatory ladder rungs | Insert single-layer and multi-layer pouch builds between coin cells and format cells; no skipping | §1 test-ladder gaps | | Filling protocol as deliverable | Develop and document vacuum/injection/rest protocol for the target format as its own work package | §3 wetting at scale | | Sensitivity register | Ship a living document with the transfer: known sensitive variables, failure signatures, lab-to-pilot predictors | §5 knowledge transfer |
If a formulation's owner — internal team or supplier — cannot tell you its sensitivity register, that is not a paperwork gap. It means the sensitivities have not been measured, and the pilot line will measure them for you.
Sources and evidence
Claims marked EXP are generalized first-hand experience from our advisory network across consumer, energy-storage, and power-cell formulation development and mass production; they carry no employer, client, or product identifiers. ORI marks our own synthesis. Numbered sources:
- Kwade, A. et al. Current status and challenges for automotive battery production technologies. Nature Energy 3, 290–300 (2018). DOI: 10.1038/s41560-018-0130-3
- Attia, P. M., Moch, E. & Herring, P. K. Challenges and opportunities for high-quality battery production at scale. Nature Communications 16, 611 (2025). DOI: 10.1038/s41467-025-55861-7
- Xiao, J. et al. From laboratory innovations to materials manufacturing for lithium-based batteries. Nature Energy 8, 329–339 (2023). DOI: 10.1038/s41560-023-01221-y
- Frith, J. T., Lacey, M. J. & Ulissi, U. A non-academic perspective on the future of lithium-based batteries. Nature Communications 14, 420 (2023). DOI: 10.1038/s41467-023-35933-2
- Chen, S. et al. Critical parameters for evaluating coin cells and pouch cells of rechargeable Li-metal batteries. Joule 3, 1094–1105 (2019). DOI: 10.1016/j.joule.2019.02.004
- Son, Y. et al. Analysis of differences in electrochemical performance between coin and pouch cells for lithium-ion battery applications. Energy & Environmental Materials 7, e12615 (2024). DOI: 10.1002/eem2.12615
- Stich, M. et al. Hydrolysis of LiPF6 in carbonate-based electrolytes for lithium-ion batteries and in aqueous media. J. Phys. Chem. C 122, 8836–8842 (2018). DOI: 10.1021/acs.jpcc.8b02080
- Sheng, L. et al. Unraveling the hydrolysis mechanism of LiPF6 in electrolyte of lithium-ion batteries. Nano Letters 24, 533–540 (2024). DOI: 10.1021/acs.nanolett.3c01682
- Logan, E. R. et al. Performance and degradation of LiFePO4/graphite cells: the impact of water contamination and an evaluation of common electrolyte additives. J. Electrochem. Soc. 167, 130543 (2020). DOI: 10.1149/1945-7111/abbbbe
- Huttner, F., Haselrieder, W. & Kwade, A. The influence of different post-drying procedures on remaining water content and physical and electrochemical properties of lithium-ion batteries. Energy Technology 8, 1900245 (2020). DOI: 10.1002/ente.201900245
- Wood, D. L., Li, J. & An, S. J. Formation challenges of lithium-ion battery manufacturing. Joule 3, 2884–2888 (2019). DOI: 10.1016/j.joule.2019.11.002
- Hagemeister, J. et al. Numerical models of the electrolyte filling process of lithium-ion batteries to accelerate and improve the process and cell design. Batteries 8, 159 (2022). DOI: 10.3390/batteries8100159
- Kaden, N. et al. A systematic literature analysis on electrolyte filling and wetting in lithium-ion battery production. Batteries 9, 164 (2023). DOI: 10.3390/batteries9030164
- de Meatza, I. et al. From lab to manufacturing line: guidelines for the development and upscaling of aqueous processed NMC622 electrodes. J. Electrochem. Soc. 170, 010527 (2023). DOI: 10.1149/1945-7111/acb10d
- Lechner, M. et al. Identification of critical moisture exposure for nickel-rich cathode active materials in lithium-ion battery production. J. Power Sources 626, 235661 (2025). DOI: 10.1016/j.jpowsour.2024.235661
- Amin, A. et al. Impact of supraparticle sizes and morphology on interparticle spacing, slurry rheology, coating density, and electrochemical performance in Si/C anodes for Li-ion batteries. ACS Applied Energy Materials (2024). DOI: 10.1021/acsaem.4c02578
Scope note
This note generalizes across lithium-ion chemistries; specific systems (chemistry, format, process route) shift the weight of each failure mode. It is based on public literature and de-identified industry experience — no confidential employer, client, or supplier information. Cited findings apply to the systems studied in each source; check transferability to your chemistry before acting on a specific number. Any process or formulation change must be validated through your own safety, quality, regulatory, and change-control systems. This is decision-support material, not a substitute for your own qualification program.