A typical smartphone cell is 4 mm thick. A smartwatch cell lands around 2–3 mm. AR glasses live below 1 mm. Each halving exposes a new set of physics that you were previously allowed to ignore.
The anatomy of a sub-millimetre cell
A lithium-polymer pouch at 0.45 mm thickness is roughly four layers of electrode coating and three separators, wrapped in an aluminium-laminate pouch that is itself about 100 µm on each face. The active material is only a fraction of the total volume — inert structural layers eat a disproportionate share compared to a thicker cell.
That’s why doubling capacity in an AR cell is rarely possible just by “making it twice as long”. Beyond about 300 mAh you almost always have to grow thickness as well, because the additional electrode layers need additional separator and tab mass, and your ratio of active-to-inert material starts going the wrong way.
Swelling becomes a first-class citizen
Every Li-ion cell swells on charge, typically 3–8% volumetric expansion. On a 4 mm cell that’s a few hundred microns of z-axis movement — the enclosure can easily absorb it. On a 0.6 mm cell it’s 30–50 µm, which matters when the cell is glued against a micro-OLED display or a PCB with sub-millimetre standoffs.
Two practical consequences:
- Leave 0.15–0.25 mm of compressible gap (foam, silicone pad) between the cell and any rigid surface.
- Qualify the cell through 85 °C / 85% RH aging and measure swelling after 200 and 500 cycles. AR customers routinely reject cells that swell past 9% at end of test.
Thermal: you can’t radiate through skin
An AR temple touches the wearer’s skin continuously. That disqualifies the normal approach of radiating heat out through the enclosure. The battery ends up sharing a heat path with the SoC, the display driver, and sometimes a camera ISP — any of which can spike to 3–5 W momentarily.
Three design knobs:
- Graphite or copper foil underside to spread heat along the temple rather than letting it pool near the cell.
- Charge-rate throttling in the BMS when skin-side temperature exceeds ~38 °C.
- Cell-internal NTC, not just a PCB thermistor — the cell surface lags the PCB by 2–4 °C on fast charge.
EMI: the cell is a dipole
A thin pouch cell has relatively large surface area per unit mass, and its tabs form a small loop antenna. Two radios sit millimetres away — Bluetooth and (increasingly) UWB or Wi-Fi. We’ve seen three recurring issues:
- Ground-return current in the battery tab modulating BT audio quality.
- Charging harmonics leaking into the 2.4 GHz band during wired charging.
- Pogo-pin contact noise disrupting touch-capacitive temples.
Mitigations are standard: tab orientation perpendicular to antenna polarisation, ferrite bead on the charge path, and ground the pouch aluminium foil to the system star-ground.
Hermetic seal and the HEVT test
AR glasses are worn in rain, steam showers, and humid gym lockers. The cell itself isn’t sealed, but it sits inside an IPx4–IPx7 enclosure. Customers frequently add a high-temperature humidity test (HEVT) on top of IEC 62133: 85 °C / 85% RH for 168 hours with the cell charged to 100% SOC. If your pouch laminate is not spec’d for this, you’ll see moisture ingress through the seal, electrolyte leakage, and dramatic capacity fade.
One pragmatic rule
Spec the thickest cell you can physically fit, not the thinnest that meets runtime. Every 0.1 mm of additional thickness typically buys you 12–18% capacity, much better cycle life, and noticeably more mechanical tolerance. The moment you start chasing the absolute physical minimum, you are trading warranty returns for a spec-sheet win.