A retired solar panel gives up its silicon at 99.998 percent purity after being scanned with green laser light, which never touches the glue that everyone assumes is the target

A panel is lying face up on a steel bench, the size of a door.

A head moves above it, and a thin green line crawls across the glass.

Behind the line the surface changes. What was mirror blue goes dull, the color of unpolished pewter, in a strip as wide as a finger.

The glass itself is untouched. Nothing is smoking, nothing is cracking, the frame is still square.

Something has come apart inside the sandwich without anything opening it.

The light goes through the glue and ignores it

This is the part almost every account gets backwards.

The obvious way to take a panel apart would be to attack the encapsulant, the plastic that bonds the layers, because that is what holds everything together.

The laser does not attack it at all.

Green and near infrared laser light passes straight through the front glass and the encapsulant without depositing much energy in either, because neither absorbs strongly at those wavelengths.

It gets absorbed where it reaches silicon.

The top surface of the cell carries a thin anti reflective coating, and that is what takes the hit. It lifts, and the bond between the cell and the plastic above it goes with it. Stripping that coating also removes a separate acid etch from the line.

So the delamination happens from underneath, at a surface nobody can see, driven by light that went through the two layers above it as if they were not there.

What a shredder does to the same object

Consider what is inside the sandwich.

Glass, an aluminum frame, a thin polymer, silicon wafers, copper ribbon, and a little silver laid down as the contact grid.

Silver is the reason anyone bothers. Roughly a quarter of all silver mined in a recent year went into solar cells.

Conventional recycling takes the frame and junction box off, then crushes what is left.

What comes out is a mixed powder in which glass, silicon and silver are physically inseparable, and the silicon is worth about what sand is worth.

The alternative route heats the whole laminate until the polymer breaks down, which works, and costs roughly 25 kilowatt hours for each module.

The numbers the lab actually published

The recovered silicon comes back at 99.998 percent purity, up from about 98 in the cell, at a yield of 97 percent.

The silver comes back at 99.7 percent purity, also at 97 percent yield.

The energy figure carries the argument. The laser draws under 0.2 kilowatt hours for a one sided module and under 0.5 for a two sided one, which is why the round number quoted everywhere is under one.

It covers the electricity the laser consumes and nothing else.

It was done on full size commercial modules, roughly five feet by three and eight by four, rather than on coupons.

Two papers describe this, and not with the same laser. The peer reviewed one uses near infrared pulses lasting about a trillionth of a second. The conference paper, which is where the purity figures come from, runs green and near infrared pulses a thousand times longer, and finds green needs roughly half the energy density to let go.

The detail that turns this into a different story

You will read that three laser pulses release the panel.

They do not. The head raster scans the entire surface of the module, line by line, in the millions of pulses it takes to cover several square feet, which the open version sets out.

The single pulse language belongs to a calibration step, where the lowest energy at which one pulse does anything is measured. That is a threshold, not the process.

Three more limits belong with it. The method works from the glass side only, and releasing the back of a one sided module is explicitly outside the scope. No throughput figure is published, so how long a module takes is unknown. And the authors call it a proof of principle at laboratory scale, not a pilot, as the journal paper says.

Compare that with what already runs, whether a recycling line handling worn modules by the truckload or the slow grind of carpet recycling, and the gap is not chemistry. It is volume.

Where the famous waste numbers actually come from

Two figures travel with every story like this one.

Eighty six million tons of panel waste by 2050, and more than fifteen billion dollars of material inside it.

Neither belongs to this lab. Both come from international energy agency reports published a decade ago, forecasting the global fleet rather than measuring anything.

The lab’s own figures are national tonnages for one small country.

What makes the work interesting is narrower and better. A panel is a stack of materials laminated on the assumption that nobody would want them apart, and this is the first demonstration of a tool that reaches the one interface that matters without disturbing the rest.

Whether that survives contact with a real line is the open question, and nothing published so far answers it.