It’s the most common question we get. The honest answer: not as designed. This brief lays out what an RO system would cost, what it would look like, and the six specific reasons it cannot — by itself — protect Central New York’s drinking water from PFAS.
The November 2025 NYSDEC permit notice and the Brown & Caldwell Conceptual Design Engineering Report both confirm reverse osmosis is part of the Oak Orchard expansion. Two RO trains are specified: one in the Municipal Treatment Train’s effluent reuse facility, and one in the Industrial Treatment Train’s water reclamation facility.
But neither RO train is on the path to the Oneida River. Both sit on the water reuse loop — they clean water that gets sent back to Micron through Outfalls 006 and 007 for cooling and ultrapure water makeup. The river-bound discharge — Outfall 01B — passes through equalization, biological treatment (membrane bioreactor), and UV disinfection. There is no RO polishing step on the discharge to the Oneida River.
Key Distinction: RO permeate is sometimes blended with the river discharge, but the discharge is not required to pass through RO. The RO is for water reuse, not pollution control.
It’s worth pausing on a second timing detail. The CDE Report explicitly states that the ITT’s water reclamation facility with RO comes online with FAB2 — projected for the second quarter of 2030. FAB1 starts up before that. So for roughly two years, FAB1’s full PFAS load will be discharged to the Oneida River through MBR and UV alone, with no RO of any kind in service.
“RO will concentrate the TDS in an RO brine (waste stream) and RO permeate (clean water product) will be disinfected and returned to the Micron plant or will be blended with other water from the plant prior to eventually being discharged at the outfall.”
The RO is described as a TDS management and water reuse system. It is never described as a PFAS removal system, because that is not what it is for.
Suppose Onondaga County did the right thing and required RO polishing on Outfall 01B itself, sized for the Phase 1 ITT design flow of 16.5 million gallons per day (FAB1 + FAB2 combined). Here’s what that engineering looks like.
The visible RO skids are the smallest part of the picture. Roughly 15 to 20 process tanks sit upstream, around, and downstream of the membranes. Equalization tankage upstream alone is two to four tanks at 1–2 million gallons each (4–8 MG total), sized for 4–12 hours of residence at peak flow to absorb fab process swings. Downstream you have the cartridge filter feed tank (50,000–100,000 gallons), the RO permeate break tank (100,000–500,000 gallons), two to four CIP (clean-in-place) tanks at 10,000–25,000 gallons each for citric acid, caustic, surfactants, and biocides, half a dozen antiscalant and chemical day tanks at 500–1,000 gallons, a brine surge tank ahead of the evaporator at 500,000–1,000,000 gallons, and another 200,000–500,000 gallons of evaporator and crystallizer feed tankage. The evaporator and crystallizer themselves are pressure vessels, not tanks — typically two to four falling-film evaporator effects and two to four forced-circulation crystallizers, each 15 to 30 feet in diameter.
Industrial brackish-water RO at 16.5 MGD, with the pretreatment and zero-liquid-discharge back end required to handle semiconductor wastewater, falls in a well-documented cost range. The defining number is not the membranes — it’s the brine-handling train.
* Substantially higher if PFAS content triggers hazardous waste classification.
Even if you spent the $150–350 million to put RO on Outfall 01B itself — which the current design does not require — RO would not solve the PFAS problem. Here’s why.
The RO units are on the water reuse loops back to Micron (Outfalls 006 and 007). Outfall 01B — the river discharge — passes through MBR and UV only. Some RO permeate gets blended into the discharge, but the discharge itself is not required to pass through RO. There is no RO polishing step on the river-bound stream.
The Conceptual Design Engineering Report is explicit: the ITT’s water reclamation facility with RO comes online with FAB2 in Q2 2030. FAB1 starts up before that. So from FAB1 startup until FAB2 ramps, FAB1’s PFAS load goes through MBR and UV and out to the Oneida River. MBR is biological treatment. It does not remove PFAS. That admission — that biological treatment is specified despite being scientifically ineffective for PFAS — is the central technical fact of this campaign.
PFAS molecules are built around carbon–fluorine bonds — the strongest single bonds in organic chemistry. UV at the doses used for pathogen disinfection is several orders of magnitude too weak to break them. UV-based PFAS destruction exists in the research literature (UV/sulfite, UV/persulfate), but not at MBR effluent disinfection doses.
Long-chain PFAS like PFOS and PFOA are rejected by RO at greater than 95%. Short-chain compounds like PFBA and PFBS drop to 80–90%. Ultra-short-chain compounds — including TFA and PFPrA, increasingly on regulators’ radar — can pass at only 30–60% rejection. The bigger problem: Cornell’s 2021 ES&T study identified 133 PFAS compounds in semiconductor wastewater, many of them unidentified “dark PFAS” with unknown molecular weights. You cannot engineer a guaranteed rejection rate against compounds you cannot characterize.
The roughly 4 MGD of PFAS-concentrated brine goes through evaporator, crystallizer, and centrifuge. The output is a salt cake that gets trucked to a landfill. Landfill leachate is one of the dominant documented pathways for PFAS to re-enter surface water and groundwater. In New York, most landfill leachate is sent to municipal wastewater treatment plants, which cannot treat PFAS, and is then discharged to the same watersheds. The PFAS captured by RO ends up back in surface water — on a five- to twenty-five-year lag.
Clean-in-place cycles every one to three months, scheduled membrane replacements, integrity failures, control upsets, and wet-weather flow surges all create bypass events when RO is offline or impaired. Without an enforceable PFAS limit at Outfall 01B, there is no monitoring requirement to detect those events and no violation trigger when they happen. The treatment infrastructure can fail silently — and legally.
The honest framing is this: the RO already in the Oak Orchard design is a TDS-management and water-reuse system, not a PFAS-removal system. It will lower some PFAS in some of the discharge some of the time — while concentrating the rest into a brine that ends up in a landfill that leaks PFAS back into the watershed.
RO without destruction on the back end is a delay strategy, not a solution.
This is exactly why Demand #4 of the ForeverChemicals NY petition specifies destruction technology, not filtration — and specifies that destruction be applied to the concentrated brine, where the PFAS actually accumulates and where flow volumes are small enough to make the energy economics of carbon–fluorine bond destruction practical. Supercritical water oxidation, electrochemical oxidation, plasma destruction, and hydrothermal alkaline treatment are all commercially available technologies that break PFAS down rather than relocating it.
If a county official or Micron representative responds to your concerns by saying “but they’re putting in RO” — that is the response. RO is necessary for water reuse. It is not sufficient to keep PFAS out of the Oneida River, and it was never designed to be.
Brown & Caldwell, OCDWEP IWWTP/WRF Conceptual Design Engineering Report (November 2025).
NYSDEC Notice of Complete Application, Oak Orchard Wastewater Treatment Plant (November 12, 2025).
Carollo Engineers, RFAI response (November 7, 2025).
Onondaga County CLCPA Climate Analysis Report (November 2025).
Jacob et al., “Per- and Polyfluoroalkyl Substances in Semiconductor Industry Wastewaters,” Environmental Science & Technology (Cornell, 2021).
EPA Work Breakdown Structure-Based Cost Model for RO/Nanofiltration Drinking Water Treatment.
SAMCO Technologies, industrial RO/NF system cost analysis. RO recovery and rejection ranges drawn from peer-reviewed literature on PFAS membrane treatment performance.
All flow numbers, RO process configuration, and brine handling design are from the public Oak Orchard permit record. Equipment counts, tankage, electrical loads, and capital and operating cost ranges are engineering estimates derived from published cost models for industrial brackish-water RO with zero-liquid-discharge back-end at the cited flow scale. These estimates are illustrative of what proper PFAS-targeted treatment would require — not predictions of what Onondaga County is currently designing.