From hydrogen gas to electrons: what actually happens inside a PEM fuel cell

A fuel cell doesn't burn anything. There's no flame, no pistons, no turbine — just a quiet electrochemical handoff in which hydrogen gas gives up its electrons on one side of a membrane and gets them back, changed, on the other. That handoff is the entire product.

The split: where the gas becomes particles

Hydrogen gas is the simplest molecule there is: two protons, two electrons, traveling as a pair (H₂). Inside a proton-exchange-membrane (PEM) fuel cell, that gas flows through channels in a plate, spreads across a porous carbon layer, and lands on the anode — a surface dusted with platinum catalyst. The platinum's job is persuasion: it loosens the bond until the molecule splits, and each hydrogen atom gives up its single electron. One neutral gas molecule becomes four separate particles — two bare protons, two free electrons. Chemists write it H₂ → 2H⁺ + 2e⁻, and that arrow is the moment fuel stops being fuel and starts being electricity-in-waiting.

The sorting: the membrane is a bouncer

Between the two electrodes sits the proton-exchange membrane — a thin polymer sheet, kept moist, that conducts protons and blocks electrons. Protons pass through the wet membrane, hopping between water molecules and acid sites like stones across a stream. Electrons are refused at the door: their only path to the other side is the long way around — out of the cell, through your wires, through whatever you've plugged in, and back. That forced detour is the electric current. A fuel cell doesn't so much make electricity as force electrons to commute, charging them tolls as work along the way.

The reunion: water is the exhaust

On the cathode side, oxygen from plain air waits on its own catalyst layer. Protons arrive through the membrane, electrons arrive through the circuit, and oxygen collects both: O₂ + 4H⁺ + 4e⁻ → 2H₂O. Water and some heat — that's the exhaust. This cathode reaction is famously the slow, stubborn step; it's where most of the platinum lives and most of the losses happen, which is why so much fuel-cell R&D is really cathode R&D.

The numbers that describe it

• On paper the chemistry offers about 1.23 volts per cell; under real load you get roughly 0.6–0.8 volts — the difference is the price of making it happen fast.
• Useful voltage comes from stacking: dozens or hundreds of cells in series, connected by bipolar plates.
• Electrical efficiency typically runs around 40–60%, with water management — moist but never flooded — as the constant balancing act.

Where the AI agents come in

This process is a fast, invisible, continuous chemical negotiation whose health you can only infer from signals — per-cell voltages, humidity, temperatures, pressure drops. That's an agent's natural habitat: watching voltage spread for the early signature of a drying membrane or a flooded channel, adjusting humidification ahead of load changes instead of after, forecasting degradation so maintenance happens on schedule rather than on failure, and logging everything so performance claims rest on data. The electrochemistry is 19th-century physics — the fuel cell predates the gasoline car. Giving every stack a tireless minder is the new part. (More on that in five jobs for agents on the hydrogen line.)

The honest caveats

None of this makes hydrogen automatically “the answer.” The gas has to come from somewhere — that supply chain has its own economics and emissions story (see green hydrogen, explained) — and every claim about a specific product should be checked against the maker's current documentation. What the transition itself offers is elegant and true: gas in, electrons out, water as exhaust.