When the fire suppressant is the pollutant

The news first. On 5 October, Suffolk County Executive Ed Romaine announced that NextEra Energy would suspend operations at two battery storage facilities on Long Island while officials investigate drinking-water contamination, as Energy-Storage.News reported on 7 October, citing local sources. The investigation goes back to 31 May 2023, when the East Hampton battery energy storage system on Cove Hollow Road caught fire.

The contaminant is perfluoropropionic acid, PFPrA. The Suffolk County Water Authority started testing for this class of chemical in January 2026 with a newly developed method, and found it in four public supply wells at its Bridgehampton Road wellfield. Levels rose between early tests and later ones. The authority says April and May results topped 50 parts per billion, and it took two wells out of service.

"Battery fire poisons wells" is the headline that writes itself. The regulator's own data tell a more specific and, for anyone who builds these systems, more useful story.

What the state actually found

The New York State Department of Environmental Conservation (DEC) published a community update in September 2026. Its preliminary conclusion is worded carefully:

The investigation points to the fire suppression system, not battery storage, as a PFPrA source.

The reasoning rests on three pieces of evidence the update lays out. First, scale: operators told DEC that more than 180,000 battery cells were on site and 672 burned, under 0.4%. Second, soil: after the fire, DEC sampled soil where fire water had soaked in and found no battery-related metals. Third, location: the highest PFPrA readings so far, in both shallow soil and groundwater, sit right next to the battery building, with elevated groundwater concentrations to the south, which is the regional downhill direction of groundwater flow toward the wellfield. DEC also notes that responders confirmed no PFAS-containing firefighting foam was used.

What was used is the interesting part. The building had three fire protection systems: emergency sprinklers, a water injection system, and a "waterless" clean-agent system that discharges a chemical sold as Novec 1230. During the response, DEC says, the sprinklers ran for about 30 hours, local officials estimate more than a million gallons of water were used, and about 4,900 pounds of Novec 1230 were released into the building.

The chemistry: a ketone that falls apart in water

Novec 1230 is a fluorinated ketone, perfluoro-2-methyl-3-pentanone, also listed as FK-5-1-12. Its structure is CF₃CF₂C(O)CF(CF₃)₂: a carbonyl group with a two-carbon perfluorinated arm on one side and a branched three-carbon one on the other. Clean agents like it are popular in rooms full of electronics because they put out fires without leaving water behind.

Ordinary ketones shrug off water. This one is different: the fluorines pull so hard on electrons that the carbonyl carbon becomes an easy target for water, and the fluorinated fragment that breaks off is stable enough to leave. A 2011 study in Environmental Science & Technology by Jackson and colleagues measured it directly. Their summary: the ketone "can undergo hydrolysis to produce PFPrA and CF₃CFHCF₃ (HFC-227ea) in a manner analogous to the Haloform reaction." In plain terms, water attacks the carbonyl and the molecule splits in two. The two-carbon arm leaves as perfluoropropionic acid, CF₃CF₂COOH, which is exactly the compound in the East Hampton wells. The other arm leaves as the gas HFC-227ea.

DEC's update cites a separate 2008 engineering paper (Tuma, an IEEE thermal-management symposium) for the same point: Novec 1230 "can degrade into PFPrA in the presence of water." DEC's preliminary picture is that a large volume of water mixed with the agent's vapor or residue, ran off south of the building, soaked in, and has been moving through the aquifer toward the wells for three years.

Here is the honest wrinkle, and it cuts both ways. Jackson's team measured the hydrolysis rate and found it "too slow to be of importance in the atmosphere," where the ketone mostly breaks down by sunlight within days to weeks. That finding is about trace amounts in air and cloud water. It does not tell you what happens when thousands of pounds of the agent meet a building flooded for 30 hours and then sit in wet soil for years. So the chemistry makes DEC's hypothesis plausible, and a direct match on the product compound makes it specific. It does not prove the quantity on its own. That is what DEC's final report, promised for this fall with validated data, needs to show.

Why this matters beyond one town

Clean-agent gas systems are designed for dry rooms. DEC says so plainly: Novec 1230 and similar systems "are not designed to be used with water," and are typically installed in data centers, labs and archives where water would cause damage. The East Hampton building had a clean agent and sprinklers in the same space. In a fire that ran long enough for the sprinklers to keep going for more than a day, the two met.

That design question is the signal for the storage industry. It is not "lithium batteries poison groundwater." DEC's evidence actually points away from that. It is that a fire protection plan is a chemistry problem as well as a fire problem, and that the water you pour on a battery fire has to go somewhere. DEC says it is working with NYSERDA and the state fire office to find out how common these clean-agent systems are at commercial and industrial sites, and that this type of system was not used at three other battery storage fires in New York.

For the drinking-water side, keep the numbers in proportion. There is no enforceable federal or state standard specific to PFPrA. New York's Department of Health set a preliminary site-specific screening level of 3 parts per billion in July 2026, well below the 50 ppb catch-all limit that applies to unspecified organic contaminants. DEC describes PFPrA as less toxic than the long-chain PFAS, PFOA and PFOS, likely because it is a smaller molecule with a shorter half-life in the body. Being ultra-short, though, it also travels faster through soil and groundwater, because it doesn't stick to soil particles as well. And testing methods for this class have not yet been certified by federal or state governments. So: a real contamination problem, with the regulatory yardstick still being built.

If you are specifying a battery system, two questions are worth asking out loud: what is in the suppression system, and where does the fire water go? I care about this because I help design the AI battery-cycling systems for a veteran-owned (HUBZone) energy-storage integrator, run by fellow veterans. I do not own that company and earn nothing from this link. Full policy here. Nothing in this report involves them or any product they sell.

What I could not confirm

DEC's finding is preliminary. Its soil and groundwater data were still undergoing validation as of the September update, and the final report has not been published. If that report changes the conclusion, this post will be corrected in place.

Water volume differs by source. DEC's update says more than one million gallons. The Suffolk County Executive was quoted by ABC7 in July saying 2.2 million gallons went into the ground. I use DEC's figure in the text and note both here.

I have not seen the suspension order itself or confirmed whether the Novec 1230 has since been removed from the buildings. The suspension is reported by Energy-Storage.News citing local sources. I could not open the Tuma 2008 paper; it is cited here only as DEC cites it, and the hydrolysis chemistry rests on the Jackson 2011 paper, of which I read the abstract and metadata, not the full text.

I make no claim about liability. The water authority has sued; that case is unresolved, and nothing here assigns fault.

The signal

New York's regulator thinks the PFAS in East Hampton's wells came from the fire suppressant, not the burned battery: only 672 of 180,000-plus cells burned, no battery metals turned up in the soil, and the contaminant, PFPrA, is precisely what Novec 1230 splits into when it meets water. About 4,900 pounds of that agent met roughly 30 hours of sprinkler water. The finding is preliminary and the final report is due this fall. The lesson for anyone building storage is already clear: a "waterless" agent plus a lot of water is a different chemical system, and the fire water has to be planned for.