(A sibling comment is wildly off the mark, and rather than reply and have my information buried far down the page, I'd rather it stay visible:)

The lead/caesium iodides are from the perovskite photovaic material (eg CsPbI₃, FA₀.₅MA₀.₅PbI₃, CH₃NH₃PbX₃ etc) that's being recycled* in the first place, and is the desired output - new photovaic material. That's the entire point, in fact - that at end of life, perovskite-based PV panels are otherwise toxic waste. This process recovers nearly 100% of these materials and allows their reuse in a new PV panel. If you're talking about perovskite PV, you almost always implicitly mean lead perovskites, and the highest efficiency we've found so far are inorganic perovskites like CsPbI₃.

In particular, perovskite PV are far FAR easier to manufacture - traditional silicon panels involve cleanrooms, high vacuum, and 1000+ C steps, while perovskite panels are very tolerant of defects and quite simple to manufacture - they can even be printed with inkjet or even screen printing technology.

This process is effecient enough that even on the 5th round of recycling (100 years of reuse, assuming a 20 year panel lifespan) they're something like 88% as efficient as a new panel, and the perovskite crystals are still at 99.998644% purity.

The solvent is indeed green - just water, sodium acetate, sodium iodide, and hypophosphorous acid.

The abstract is also clear that this is total panel recycling: "We further extend the scope of recycling to charge-transport layers, substrates, cover glasses and metal electrodes." Later in the paper they elaborate that after a brief low heat (150 C for 3 minutes) treatment to delaminate the EVA encapsulation, the panels are "then layer-by-layer recycled to reclaim cover glass, spiro-OMeTAD, perovskite crystal powders and SnO2-coated ITO substrates."

* As a control, they created fresh perovskite from the same technique as the recycling but from pure, purchased reagents, to assess the efficiency difference between them. That's the only place that fresh PbI and CsI appear.

They say they use green solvents, but in the list of materials at the bottom I see Lead Iodide and Cesium Iodide, which doesn't strike me as too green of a thing to use.

Also, the abstract doesn't really make it clear whether the recycling is for complete panels or the "filling" of panels (sorry for the layman's terms). And whether this applies to any Perovskite-utilizing panel or just certain kinds.

But I am generally heartened at seeing thought regarding industrial processes which consider the end of productive life as well. I wish more products were designed to eventually be taken apart and reduced using standard processes, or even per-product processes - rather than the assumption being that more and more stuff is chucked into landfills.