Pouch Cell Edge Seal Integrity

A pouch cell’s case is not a can. It is a five-layer laminate folded around the electrode stack, wound into a jellyroll or built up flat as a stack, and closed on all four sides by melting two facing plastic surfaces into one continuous joint. That joint, more than the aluminium foil running through the middle of it, is what keeps moisture out and electrolyte in for the life of the cell. When engineers talk about a pouch cell “failing at the seal,” they usually mean one of two different things happening at that joint: it was never fully closed (a process defect), or it opens gradually over years of cycling (a durability problem). Both trace back to the same handful of variables: laminate architecture, seal process control, and the mechanical and chemical loads the joint has to survive in service.

The laminate: three functional layers, two adhesive interfaces

The aluminium-laminated film (ALF) that forms a pouch cell’s case is built from three layers, each doing a different job, bonded together as two separate adhesive interfaces rather than a single stack. Working from the outside in:

Outer layer, biaxially oriented polyamide (nylon, sometimes called ONY or BOPA). This is the layer that takes abuse: scuffing, puncture, and the tension of being drawn into a forming die. It contributes essentially nothing to the moisture barrier.
Core layer, aluminium foil. This is the actual hermetic barrier. It is typically an 8021 or 8079 alloy foil with iron deliberately held in the 0.9–1.5 wt% range. Iron content this low is what lets the foil take the 15–26% tensile elongation it needs to survive deep drawing without splitting.^1,2
Inner layer, cast polypropylene (CPP). This is the sealant. It is the layer that actually melts and fuses to the CPP on the opposing wall to close the seal, and it is also the layer in direct contact with the electrolyte.

Because nylon, aluminium and CPP won’t bond to each other on their own, two adhesive interfaces hold the stack together. A polyester-polyurethane adhesive bonds the nylon to the aluminium at roughly 4–5 g/m², and a maleic-anhydride-grafted polypropylene (PP-g-MAH) bonds the aluminium to the CPP at roughly 2–3 g/m². The PP-g-MAH grade matters because it’s chosen specifically for its affinity to the metal oxide on the foil surface, and it’s this second interface, not the seal weld itself, that tends to be where long-term delamination starts.⁴ One China National Intellectual Property Administration patent covering ALF construction describes essentially this architecture, with the inner layer sometimes built as a more complex multi-resin blend (PP, CPP, PE, PA, PVDC and EVOH options alongside a PVA moisture-resistant coating) rather than plain CPP, aimed specifically at improving resistance to the organic solvents and HF that electrolyte breakdown produces.¹

ALF is sold in a handful of standard thickness bands, and the choice is really a choice about application: roughly 88 µm for slim consumer devices, 113 µm as the workhorse grade for mobile and portable power, and 152–153 µm heavy-duty grades for EV and stationary storage cells that need to survive more aggressive handling and a longer service life.⁴

Getting the foil into a cup shape at all is its own constraint. Cold-forming (deep-drawing) the laminate to create the pocket that houses the electrode stack thins the aluminium at the corners and walls of the draw, and if it thins too far the barrier is compromised before the cell is even filled. One published aluminium-laminated-film patent puts a number on this: cold-stamping depth of 3–8 mm, with no delamination, cracking or pinhole light leak permitted, and for a 40 µm baseline foil, draw depth generally can’t exceed 4 mm before post-forming thickness drops below the 30 µm floor needed to hold the barrier.¹ Those figures come from one specific patent’s process claims rather than an industry-wide standard, so the specific limits still vary by foil supplier, temper and press tooling in practice.

Needs review: The source report supplied for this page also quoted a specific heat-seal process window (0.2–0.5 MPa clamping pressure, 2–5 second dwell), attributed to a monograph titled “Computational and Chemomechanical Analysis of Pouch Cell Edge-Seal Integrity Under Cyclic Swelling.” I couldn’t locate that publication anywhere, under that title or a close variant, in the supplied references or in general search, so I can’t verify it’s a real, citable source. Worth tracking down the original document (or your own supplier process data) before publishing that pressure/dwell figure with a citation attached.

Making the seal: thermal fusion and the tab problem

A pouch seal is formed by heating the two facing CPP layers past their melting point and holding them together under light clamping pressure long enough for the polymer chains on each side to interdiffuse and entangle across the interface. Independent sources are consistent that CPP’s practical melt window for sealing sits around 180–190°C. Below that the chains don’t move far enough to entangle properly, and running well above it for longer than necessary risks scorching the film or driving heat into the active stack behind the seal.^3,4,9 Pressure and dwell time are the two variables manufacturers tune by design-of-experiments against their own film, equipment and cell format, rather than reading off a single published number.⁹

Two different heating technologies dominate industrial sealing equipment, and they trade off against each other in a way that matters for battery assembly specifically. Flexible resistive heating (marketed by ROPEX as RESISTRON) uses a heated metal band and is the conventional choice: it runs without an external water-cooling loop, handles long or contoured seal geometries, and cycles at roughly 80 strokes per minute. Ceramic thick-film heating (ROPEX’s CIRUS) heats and cools far faster, on the order of 6,000 K/s, and can run up to about 120 cycles per minute, which suits high-throughput lines, but it needs active water cooling and is limited to flatter, two-dimensional seal geometries, which matters if cleanroom water infrastructure or seal contour is a constraint.³

Whichever heater is used, the tab seal is the hard part of the job. The current collector tabs (nickel on the anode side, aluminium on the cathode side) are far more thermally conductive than the surrounding CPP, so they pull heat away from the sealing zone during the brief sealing pulse. That local heat sink effect is exactly why tab seals get separate qualification from the field seal: if the polymer immediately around a tab doesn’t reach full melt because the metal next to it is wicking heat away, you get a void or a thin spot right where the seal is under the most mechanical stress from vibration and tab flex in service. It’s also why anode (nickel) and cathode (aluminium) tab seals commonly carry different minimum peel-strength requirements in supplier specifications: nickel’s lower thermal conductivity relative to aluminium means it retains heat at the interface a little better, which tends to produce a stronger bond for a given process window. Exact minimum peel-strength figures vary by cell supplier and film system, so they’re not reproduced here as a fixed number.

Geometry pushes the seal to its limits

Two form-factor trends put extra strain on the seal design: ultra-thin cells for wearables and slim devices, and curved or arc-profile cells for products that wrap around a wrist or a device edge. Custom cell manufacturers building to these geometries commonly quote ultra-thin cell thicknesses from roughly 0.4–0.95 mm held to ±0.05 mm, with reinforced film grades specified to hold a 24-month shelf life despite the reduced foil cross-section available to act as a moisture barrier. Curved and arc-profile designs are typically built from 1.0–10.0 mm thickness stock, held to a looser ±0.1 mm tolerance, across curvature radii from about R10 to R150.⁸

For curved geometries specifically, stacked-electrode construction is the practical requirement rather than wound construction: winding a jellyroll around a tight corner radius concentrates tension exactly where the case is also under the most forming stress, and that combination is a reliable way to get casing cracks and seal delamination at the corner. Stacking sidesteps the winding-tension problem, even though it trades away some of winding’s manufacturing throughput advantage.

Why seals fail I: swelling load and where it concentrates

Lithiation swells the electrode stack, and that swelling pushes outward against the pouch wall and, ultimately, against the seal. In-situ pressure-mapping of a 2000 mAh NCM/graphite pouch cell recorded peak expansion force of roughly 125 kg at full charge, concentrated in the centre of the cell face where the electrode stack is least constrained by the seal perimeter; force at end-of-discharge on the same cell dropped to roughly 57 kg.⁵ The absolute force is highest in the middle of the cell, but the rate at which that force changes through a charge cycle is highest at the corners and edges, which is the more useful number for seal design: a joint doesn’t fail because of the biggest number on the map, it fails because of the fastest-changing one nearby.

How that load reaches the seal depends on how the cell is constrained. Under a rigid, constant-gap constraint (a fixed-volume module cavity, for example), the stack has nowhere to go as it swells, and electrode layers can wrinkle and buckle internally. That buckled material acts like a wedge working against the inner shoulder of the seal, applying a steady peeling/shear load to the polymer joint every cycle rather than a single overload event. That’s a slower, more insidious failure mode than an outright burst, and one that’s easy to miss in short-duration qualification testing. A constant-force (flexible, spring- or foam-backed) constraint lets the stack breathe within a controlled pressure band instead of fighting a hard wall, which is the direction cell and module designers increasingly favour for silicon-containing anodes specifically because those chemistries swell more and more irreversibly than graphite alone.

The seal doesn’t have to be uniformly strong around the whole perimeter, either. It’s common practice to deliberately under-seal one edge, typically the bottom edge, away from the tabs, relative to the rest of the perimeter, so that in a gas-generating fault (overcharge, internal short) that edge opens first and vents gas in a controlled way rather than letting pressure build until the case ruptures somewhere unpredictable. The exact release pressure engineered into that weak point is a design choice tuned to the specific cell’s mechanical margins elsewhere, so it varies cell to cell rather than sitting at one fixed industry figure.

Why seals fail II: the seal is under chemical attack too

Mechanical load is only half the story. The electrolyte inside the cell is chemically hostile to the seal from the inside out. Carbonate-based electrolytes use lithium hexafluorophosphate (LiPF₆) as the conducting salt, and LiPF₆ is notoriously moisture-sensitive: trace water hydrolyses it to release hydrofluoric acid (HF).

\text{LiPF}_6 + \text{H}_2\text{O} \rightarrow \text{POF}_3 + \text{LiF} + 2\text{HF}

That HF doesn’t stay put. It diffuses slowly through the CPP sealant layer and attacks the hydroxyl groups on the surface oxide of the aluminium foil, which is the very surface chemistry the PP-g-MAH adhesive is relying on for its bond. Weaken that oxide surface chemically and the adhesive bond weakens with it, quite independently of anything happening mechanically at the joint. This is a slow-burn failure mode: it doesn’t show up in a fresh cell, it shows up after months or years of cycling as gradually declining peel strength at the aluminium-to-sealant interface, and it’s compounded by whatever moisture ingress the cell has already picked up over its life, since more moisture means more HF generated for the same trace-water starting point.

It’s worth being clear that this is a different failure mechanism from pouch-can corrosion, where the aluminium itself is consumed by direct electrical contact with the anode through a damaged inner layer; that mechanism, and how to design against it, is covered separately in Addressing Cell Pouch Corrosion through Cell Design. HF attack on the adhesive bond line is a seal-integrity problem; galvanic corrosion of the foil is a cell-design problem. Both start with the same electrolyte chemistry and both end with a compromised pouch, but the mechanisms and the design fixes differ.

Reading the early warning signs

Seal and swelling problems both tend to announce themselves through cell thickness before they show up anywhere else, which is why in-situ thickness and force measurement has become a standard diagnostic rather than a research curiosity. Charging a pouch cell too fast at low temperature is a good example: on cells tested at 5°C across a range of charge rates, thickness-versus-SOC curves for 1.0C and 1.5C charging diverge sharply from the lower-rate curves once the cell passes roughly 15–27% state of charge, and post-mortem teardown of the 1.0C and 1.5C cells showed the grey-white surface staining characteristic of metallic lithium plating on the anode, while cells held at 0.8C and below stayed a normal graphite gold-brown.⁶ The thickness inflection is a genuine physical signature of plated lithium changing the electrode’s mechanical stack-up, not just noise. And because plated lithium is itself a puncture and internal-short risk, a diverging thickness curve at a given charge rate and temperature is a legitimate early-warning signal for cell and BMS charge-current limits, separate from anything happening at the seal itself.

Silicon-containing anodes complicate this picture further. An individual silicon particle can expand by roughly 300% on full lithiation (more than an order of magnitude beyond graphite), but because the surrounding carbon matrix and porosity in a real silicon-carbon composite electrode absorb most of that particle-level change, cell-level thickness growth over 50 cycles for typical 3–5 wt% silicon-carbon blends comes out far lower, on the order of 9–11% rather than anything close to 300%.⁷ That gap between particle-level and cell-level swelling is exactly why the seal doesn’t need to survive a 300% volume change directly, but it does need enough margin to absorb steadily increasing irreversible growth over the cell’s life as more of that particle-level expansion fails to fully reverse on discharge.

Where seal design is heading

Two things are pushing pouch-seal design harder than they used to: higher-silicon anodes with their larger irreversible swelling, and EV/ESS formats that need a multi-year, high-cycle-count service life rather than a phone’s two or three years. The response on the containment side is the shift already mentioned above: away from rigid constant-gap constraint and toward constant-force bracing that lets the stack breathe within a bounded pressure range instead of fighting a fixed cavity, which reduces the internal buckling and corner-strain concentration that constant-gap designs are prone to.

On the materials side, the direction of travel is sealant and adhesive chemistry that resists the HF attack described above for longer, and lamination processes (thermal lamination in particular) that are more resistant to organic-solvent and HF ingress than older dry-lamination approaches. In-situ instrumentation (conformal pressure and swelling maps of the kind used to generate the expansion-force data above) is also moving from a lab characterisation tool into a production and formation-line diagnostic, catching plating and localised buckling before a cell ships rather than only in post-failure teardown.

References