When a solar farm comes down, every panel on it has to go somewhere, and the word the industry uses for where they go is recycled. It shows up in procurement documents, sustainability reports and ribbon-cutting releases, usually sitting next to a recovery rate in the high nineties. A Los Angeles research firm spent this year pulling regulatory filings and corporate disclosures to work out what that word covers in practice.<\/p>\n
Sustainable Source Studios published its answer on July 22 and pushed it out to the press on August 19. For roughly nine out of every ten retired American panels, the firm says, the word covers a landfill.<\/p>\n
Two of the recyclers named in the paper have since disputed it. One of them runs the plant in Cedartown, Georgia. And S3 has already corrected one of its own seven headline findings by a factor of fifteen.<\/p>\n
Isaac Nichelson, who founded S3 and wrote the paper, put the central claim in a single line. “Almost none of what the United States calls ‘solar recycling’ is closed-loop recovery,” he wrote, in a passage pv magazine USA quoted on August 20<\/a>.<\/p>\n The National Renewable Energy Laboratory prices recycling a single module at $15 to $45, against $1 to $5 to drop the same module at a landfill. NREL’s own behavioral modeling of the 2020 to 2050 window puts about 80 percent of American modules in landfills, 10 percent through recycling and 1 percent into reuse under baseline conditions.<\/p>\n So S3’s 90 percent sits a little above the federal lab’s baseline projection, not somewhere off the map.<\/p>\n S3 puts the volume the country is walking into at about one million tons of retired modules by 2030 and roughly ten million tons by 2050.<\/p>\n The United States shut its last primary lead smelter in 2013, after the EPA tightened the national air standard for lead. S3 traces what filled the hole. Korea Zinc, through its holding company PedalPoint and the recycler evTerra, runs four shredding plants on American soil. Metal-bearing concentrate from those plants travels roughly 5,500 nautical miles, about 6,330 miles, to Korea for smelting. Korea Zinc has told investors it wants 120,000 tons a year of solar material moving through its Onsan smelter by 2030, and S3 notes that Korea’s own projected solar waste came to about 1,222 tons in 2025.<\/p>\n S3 lists that gap as its third finding. American panels fill it.<\/p>\n Nobody disputes the shipping. What the recyclers dispute is what it means, and Korea Zinc has been moving on that front too. The company bought the Clarksville, Tennessee zinc refinery from Trafigura’s Nyrstar in December and plans to replace it with a $7.4 billion complex opening in phases from 2029, with the US Department of Defense holding 40 percent of the joint venture.<\/p>\n SOLARCYCLE switched on its 255,000-square-foot line in Cedartown on January 29, aiming for a million panels a year by the end of 2026 and a ceiling of 5 gigawatts’ worth annually. We wrote about that plant and the glass factory going up beside it<\/a> in July.<\/p>\n Asked about the report, the company kept it short. “We disagree with the allegations,” it said, while adding that it welcomed attention to competing schools of thought inside the industry.<\/p>\n Then it gave numbers. For most of the panels it has processed, SOLARCYCLE says more than 80 percent of the mass stays in the United States and goes to domestic offtakers, with under 20 percent heading to refining partners abroad because no American refinery runs at that scale yet. The company puts its value recovery at 97 percent, and says the 0.15 percent of silver it misses ends up in a clean slag the smelter sells as a concrete input.<\/p>\n SOLARCYCLE has also decided against R2 certification for now, saying the standard offers thin guidance on material tracking and overlaps with the ISO 9001, 14001 and 45001 certifications it already holds.<\/p>\n Solar Panel Recycling, LLC runs plants in Salisbury, North Carolina, Lawrenceville, Georgia and Breckenridge, Texas. SPR went further.<\/p>\n “We do not shred modules and ship them overseas for smelting,” the company said in its statement, which also said no one from S3 contacted it during the research or set foot in any of the three sites.<\/p>\n SPR says the paper misstates its annual processing volume at 69,000 tons, about a third of what it actually runs. It says its recovered glass, aluminum and copper stay domestic, and that it sends silver offshore for final refining the same way every other operator does, because no US silver refinery works at scale.<\/p>\n Both companies landed on the same point about who sits at the top of S3’s chart.<\/p>\n The release that went out on August 19 said two North American facilities held R2v3 Appendix G, the solar-specific certification, and that one of the two ran at industrial scale.<\/p>\n Business Wire carried a corrected version<\/a> stating that the seventh finding “understated the number of North American facilities holding R2v3 Appendix G,” and that S3 fixed it after querying SERI’s certified facility directory directly.<\/p>\n The corrected count is at least 30.<\/p>\n S3 keeps the argument alive on different ground. Of those 30-plus facilities, the revised finding says 16 are certified only to hand panels to the next vendor with no materials recovery in scope, four hold certificates scoped purely to PV modules, and one operates at industrial scale. The same correction added a line disclosing that S3 has provided sustainability consulting to Comstock Inc. That disclosure was already in the white paper’s footnotes. It was not in the original release.<\/p>\n Comstock Metals holds the top-right box on S3’s quadrant chart, and the paper calls it the only North American operation meeting every criterion for certified, zero-landfill domestic processing. Its plant sits in Silver Springs, Nevada, designed for 100,000 tons a year.<\/p>\n Comstock announced on August 11<\/a> that it had fully integrated and tested the system and run panels through every stage of production. Final integration was due mid-August, with the ramp toward a first production milestone starting after that. Chief executive Corrado De Gasperis put that milestone in September.<\/p>\n SOLARCYCLE and SPR both seized on the same document, arguing the paper reads a design capacity as current throughput. Comstock says S3 handled the research independently and that it supplied no data or assistance.<\/p>\n SERI added Appendix G to the R2 standard in 2024, and under the R2 Code of Practices, facilities handling PV modules are not required to add it to their scope until January 31, 2027<\/a>. S3’s own reading of the standard is that a compliant custody record can stop at the first certified vendor in the chain, or at a smelter’s gate.<\/p>\n Federal rules are not going to settle it first. The EPA announced on October 23, 2023<\/a> that it would move retired panels into the universal waste category, and its own page still describes that rule as one the agency is planning to propose.<\/p>\n Which leaves the market doing the sorting. First Solar has run in-house recycling for its cadmium telluride modules for more than 20 years and reported material recovery above 95 percent in 2025, which is why S3 names it as an exception alongside Comstock, and why its Louisiana plant<\/a> is a different animal from the crystalline-silicon lines everyone else is building.<\/p>\n Comstock’s first production milestone is due in September. S3’s correction is dated August 19. January 31, 2027 is the next date on the calendar that forces anyone to prove anything.<\/p>\n","protected":false},"excerpt":{"rendered":" When a solar farm comes down, every panel on it has to go somewhere, and the word the industry uses … <\/p>\nKorea Zinc sits at the end of the chain, about 6,330 miles from the panels<\/h2>\n
The Georgia plant says four fifths of every panel never leaves the country<\/h2>\n
A second recycler says nobody from S3 walked its floor<\/h2>\n
S3 corrected its seventh finding by a factor of fifteen<\/h2>\n
Comstock started its plant on August 11 and has not shipped commercially yet<\/h2>\n
January 2027 is when a certificate has to start meaning something<\/h2>\n