The car that runs on air and water

Why a technology that works has not displaced anything

A The proposition is superficially irresistible. A vehicle that carries hydrogen, combines it with oxygen from the air in a fuel cell and emits nothing but water would deliver the range and refuelling speed of a petrol car with none of the consequences. The chemistry was demonstrated in the 1830s, the engineering was proved by spacecraft in the 1960s, and functioning passenger cars have been sold to the public for over a decade. The technology is not speculative, and the arguments against it are not arguments about whether it works. Its difficulties lie almost entirely elsewhere.

B The first is that hydrogen is not a fuel in the sense that coal or oil is a fuel. There are no hydrogen deposits. Every kilogram in use has been manufactured, using energy, from something else, and the something else is at present overwhelmingly natural gas. Producing hydrogen this way releases carbon dioxide, and unless that carbon is captured and stored — which is done at a small number of installations and nowhere at scale — a vehicle burning it in a fuel cell is, considered from end to end, running on fossil fuel with extra steps. This is not a marginal qualification. The overwhelming majority of the hydrogen manufactured today is made this way, for industrial customers who use it to refine oil and to make fertiliser, and the emissions involved are of the same order as those of a substantial national economy.

C The clean alternative is to split water using electricity. This works, requires no exotic materials in principle, and is where most of the industry's optimism resides. It is also where the arithmetic becomes uncomfortable. Roughly a third of the input energy is lost in the electrolysis itself; more is lost compressing the gas to the several hundred atmospheres at which a car can carry a useful quantity, or liquefying it at temperatures near absolute zero; more again in distribution; and a further third or so in the fuel cell that converts it back to electricity on board. The proportion of the original electricity that reaches the wheels is around a quarter. Put the same electricity into a battery and roughly three-quarters arrives. Each individual loss is defensible and several are being reduced; it is their accumulation along a chain of four or five conversions that produces the result, and no single improvement can rescue it.

D That comparison has driven the passenger-car argument almost to a conclusion. If electricity is the scarce input — and in a decarbonising system it is — then a technology that wastes three units for every one it delivers is difficult to defend against one that wastes one for every three, whatever its other merits. The counter-argument rests on the merits: a hydrogen car refuels in minutes rather than hours, carries its energy in a tank that weighs little, and suffers no loss of range in cold weather. These advantages are real, and they are the ones a driver actually notices. They have not proved sufficient, and the reason appears to be that the drawbacks are structural while the advantages are matters of convenience that improve, on the rival technology, every year.

E Infrastructure has compounded the problem in a way that is self-reinforcing. Hydrogen cannot use existing pipelines without modification, since the molecule is small enough to escape through seals that contain natural gas and, over time, to make some steels brittle. A filling station costs a substantial multiple of a conventional one. Motorists will not buy vehicles they cannot refuel, and nobody will build stations for vehicles that have not been bought, so the number of stations in most countries has stayed in the low hundreds and in several has recently fallen. Battery charging faced the same problem and escaped it because a charger can be installed in a domestic garage, and hydrogen has no equivalent of that. A domestic hydrogen supply is not a plausible proposition for reasons of both pressure and safety, so every kilogram must come from a purpose-built station, and the economics of such a station depend on a level of traffic that does not exist.

F Where the case remains strong is precisely where batteries are weakest, and the distinction is one of duty cycle rather than of principle. A long-distance lorry, a ship, a train on an unelectrified line or a machine that must work three shifts without stopping needs energy stored densely by weight and replaced quickly, and for these the battery penalty is severe while the hydrogen penalty is tolerable. Steelmaking is a separate and possibly larger case: hydrogen can replace coal as the agent that strips oxygen from iron ore, and no battery can do anything of the kind. That single application, if pursued, would require more hydrogen than every car ever proposed, which suggests where the limited supply is likely to end up.

G The confusion in public discussion arises from treating hydrogen as a single proposition to be accepted or rejected. It is better understood as a raw material with a small number of applications for which nothing else will serve, and a much larger number for which something else will serve better. The vehicles that have attracted most of the attention belong firmly to the second category, and the industries that attract almost none belong to the first. That the argument has been conducted mainly through the passenger car is an accident of what the public finds interesting, and it has obscured a question that matters considerably more.