A rectangle of dark glass on a cleanroom bench, about the size of a hardback book laid open.
Thin silver lines run across it. Under the lines, the layered sandwich every module has carried for twenty years.
One layer in that stack is missing, and nothing about the panel gives it away.
The film that carries current out of the top cell has always been built around a metal that comes out of the ground as a leftover.
This one does not use it.
The reading barely moved.
Why the old film could not simply be swapped
A tandem cell needs a window that conducts. It has to pass light down to the silicon and carry electrons sideways to the contacts at once.
Indium tin oxide does both well, which is why it is everywhere. Indium is a byproduct of zinc refining rather than mined for its own sake, so supply is tied to somebody else’s decisions.
Tin oxide has been a candidate for years. The problem was never the material. It was how you put it down.
Conventional sputtering fires high energy particles at a target, and they reach the surface still carrying that energy. On silicon it does not matter. On a perovskite layer it wrecks the interface.
Reactive plasma deposition works at much lower energy. Metal vapor meets reactive oxygen in a plasma and arrives gently, which lets a dense film grow on something fragile.
The recipe was never the obstacle. The delivery was.
Three numbers, and the difference between them
The work reports three results and they are not interchangeable. Reading them as one figure is how this gets overstated.
With the tin oxide used only as the recombination layer, a small device reached a certified 33.6 percent. It still had indium in its electrodes.
Extending the same film to the front and rear electrodes removed the indium entirely. That device reached 33.2 percent as a champion result.
So going all the way cost about four tenths of a point, which is worth stating rather than rounding away.
The number that matters commercially is the third. A minimodule of about 32 square inches, a size that means something outside a laboratory, came in at a certified 31.0 percent.
Small area flatters everything. Module area does not.
What the durability testing actually covers
The modules went through thermal cycling, damp heat and outdoor operation, the standard trio for anything claiming to be more than a curiosity.
The outdoor figure has a hard number attached. After 105 days outside, the indium free minimodules held 65 percent of peak efficiency.
That is honest reporting of a hard problem. A third of the output gone in three and a half months is nowhere near a warranty.
It is the expected shape at this stage. Perovskite degradation outdoors is the open question of the field, and no electrode swap was going to close it.
Tin runs at roughly one percent of the price of indium.
Outdoors took a third. Indoors proved the film.
Where the silver story goes the other way
Indium is not the only constrained element in a panel. Silver forms the gridlines on every conventional cell.
The usual telling has silver demand from solar climbing while installations rise. That is not what the current forecasts say.
Photovoltaic silver demand is projected to fall about 19 percent this year, from roughly 187 million ounces to 151 million. Manufacturers have been printing less of it per cell.
Industrial silver fabrication overall is forecast to slip about 2 percent, and photovoltaics are the reason. Solar is not pulling that market up.
What comes back at the far end was covered here, with froth flotation lifting almost all the silver out of crushed cells.
The industrial version of that runs in Georgia, where worn panels arrive by the truckload at a 255,000 square foot plant.
Demand fell by design. The deficit stayed anyway.
What the finding does and does not settle
The silver market is still short, for the sixth year running, at a projected deficit near 67 million ounces. Solar thrift did not fix that, because solar was never all of it.
Which is the lesson for indium too. Designing a scarce metal out of one layer removes one dependency. It does not make a supply chain comfortable.
The deposition method is the transferable part. The equipment already exists in display manufacturing, so the path to a line is short.
What is untested is what is always untested. Nobody has run one of these outside for a decade, and the curve so far points the wrong way.
The cells, the certified figures and the deposition route are set out by an industry title.
The silver demand forecasts and the sixth consecutive deficit are published by the trade body.
The tin oxide did its job, and the scarcity that gets talked about most is already shrinking without any of this.
