Thermophotovoltaics

Thermophotovoltaics (TPV) takes a detour that pays off: instead of converting heat directly into electricity, it first lets the heat glow and then catches the light with a semiconductor cell.

The arrangement

Two parts suffice. An emitter is brought to a high temperature and in doing so radiates electromagnetic waves — as every hot body does. A photovoltaic cell opposite converts this radiation into electricity.

Unlike with a solar cell, the maximum of the radiation does not lie in the visible but in the near infrared. At 2,000 kelvin it lies at about 1.5 micrometres.

Why the detour makes sense

Thermoelectrics converts heat directly into voltage, but needs a temperature gradient for that and rarely gets beyond eight percent efficiency. The reason is a conflict of aims in the material: one needs something that conducts electricity well and heat badly, and in most substances the two conductivities go hand in hand.

Thermophotovoltaics does not have this conflict. Its limit is not materials science but the physics of radiation — and that lies considerably higher.

The two decisive tricks

Matched band gap. The cell is tailored to the spectrum of the emitter. For the infrared range semiconductors such as gallium arsenide, indium gallium arsenide or indium gallium arsenide antimonide are suitable — not silicon, whose band gap of 1.12 electronvolts is too large for these wavelengths.

Modern arrangements stack two cells with different band gaps on top of one another, so that each uses the part of the spectrum that suits it.

Recovery. Photons whose energy lies below the band gap could achieve nothing and would be lost. Instead of wasting them, a mirror is fitted behind the cell that throws them back to the emitter. There they help with the heating.

This move is the reason for the leaps of recent years. It turns a loss into a cycle: what the cell cannot use is not lost but goes back into the heating.

The state of the art

In 2022 a working group of MIT and the National Renewable Energy Laboratory reported in Nature on a tandem cell with an efficiency of about 40 percent at emitter temperatures between 1,900 and 2,400 degrees Celsius.

That is a remarkable number. It lies above the efficiency of a steam turbine in an ordinary power station — and that in an arrangement without a single moving part, without water, without maintenance.

The application that is being thought of

The obvious use is the storage of electricity as heat.

Surplus wind and solar electricity heats a storage medium — molten silicon or graphite blocks, say — to over 2,000 degrees. If the electricity is needed later, TPV cells give it back out again.

The appeal lies in the costs. A heat store consists of cheap, abundantly available materials and hardly ages, whereas batteries are expensive and lose capacity. One would have a battery without chemistry.

Distinction from neutrinovoltaic technology

The difference is the starting point. Thermophotovoltaics needs a hot source, and this heat has to come from somewhere — from combustion, from concentrated sunlight, from a store. Without a hot emitter no radiation and no electricity.

Neutrinovoltaic technology starts where no such source is at hand. Its claim is directed at fluxes that are present everywhere and constantly, and thus at applications in the dark, indoors, underground.

For photovoltaics in comparison something similar holds: it needs light from outside; thermophotovoltaics generates its light itself.

What is to be learned from it

Thermophotovoltaics shows two things that hold for other approaches as well.

First: the apparent detour can be the shorter way. Converting heat into electricity via radiation is more cumbersome than the direct way — and delivers five times as much.

Second: the progress came not from a new basic idea but from careful materials work on band gap and mirror. Precisely therein lies also the task that neutrinovoltaic technology has set itself with its multilayered graphene-silicon stacks.

Sources

  • A. LaPotin et al.: Thermophotovoltaic efficiency of 40 %, Nature 604, 287–291 (2022).