A solar window about 12 inches across converted 14 percent of the sunlight falling on it, and under nothing but the ceiling lights of an ordinary room it reached 22 percent

Point a solar cell at the sun and it works.

Point the same cell at a ceiling light and it usually does almost nothing, because a lamp is dim and its light is the wrong shape.

That is the rule.

And a square of tinted glass in a London laboratory has just broken it, by doing better under office lighting than it does outdoors.

Not equally well.

Better, by eight percentage points, which is the wrong way round for anything that calls itself solar.

The number that runs backwards

The solar window prototype is a panel about 12 inches on each side, thin enough to sit in a frame. Under a standard outdoor sun test it converted 14 percent of the light into electricity, which is respectable for anything you can see through.

Then they carried it inside.

Under bright indoor lighting at 1,000 lux, roughly what a well lit office runs at, the same device reached 22 percent. The team reports that as a record for a see through module of this size measured under indoor conditions, and it is the reason the paper exists at all.

So one piece of glass covers two jobs. It works while the sun is up, and it keeps working after dark, on the light already burning inside the room.

Why a lamp is easier than the sun

Sunlight arrives as an enormous spread of wavelengths, from ultraviolet through the infrared, and a silicon cell is built to swallow as much of it as it can. Indoor light is nothing like that.

A ceiling lamp emits a narrow band.

Because it is narrow, a material whose absorption can be tuned to sit exactly on that band will capture almost all of it, while silicon, tuned for the sun, throws most away. The material here is a perovskite, and adjusting its bandgap is a matter of changing the recipe rather than the machine.

That tunability is the whole trick. It is also why the indoor figure beats the outdoor one, because matching a narrow source is simply an easier problem than harvesting a wide one.

The gold hidden inside the glass

Getting current out of a cell without blocking the view is the harder half. A normal electrode is opaque metal, which would turn a window back into a wall.

So they made the metal thin enough to see through.

The team sandwiched gold between two layers of molybdenum oxide, thin enough to pass light while still carrying current. The light absorbing layer itself is 185 nanometers thick, roughly five hundred times thinner than a human hair, which is part of why any of this stays transparent.

Then there is the durability problem. They added a molecule with a long chemical name to plug the defects in the crystal where electrons get trapped, and the panel held 80 percent of its efficiency through 300 hours of continuous light.

What 30 percent transmission actually looks like

Here is the compromise nobody puts in a headline. The panel lets through 30 percent of the daylight hitting it, and ordinary window glass lets through 80 to 90.

That is not a light tint. That is dark.

You could put it in an atrium or a stairwell or an office wall already fitted with blinds, but not in a bedroom. The honest framing is a heavily tinted pane that also earns its keep, not clear glass with a bonus, which is a distinction worth holding onto after a solar highway taught everyone to read these announcements slowly.

Mojtaba Abdi-Jalebi, who led the work at the university, put the commercial case plainly. “Rooftops are commonly fitted with solar panels but the vast window areas of many modern buildings remain largely untapped as an energy resource,” he said, and a trade report sets out the rest.

The distance between a bench and a tower

A 12 inch square in a clamp is a long way from a curtain wall. Scaling a perovskite film evenly across a pane the height of a room is an unsolved manufacturing problem, and losses grow with area.

Then the weather arrives.

Perovskites dislike moisture and heat, and 300 hours under a steady lamp is a laboratory result rather than a decade on a facade. Rain, thermal cycling and ultraviolet will push the crystal far harder than any bench can, so the encouraging stability number is a starting point and the team says as much in the paper.

What survives all that caution is still worth having. A material that reads a lamp better than the sun is a genuinely odd object, and it turns every lit room into a place where a surface earns something, the way a plastic lawn turned out to behave nothing like the thing it copies.

Whether it reaches a real building depends on the boring parts, on cost per square foot and how it looks after ten winters. But the backwards number is measured, published and repeatable, and that is where ordinary technology usually starts.