265 nm — the UV-C LED germicidal optimum, and why the damage doesn't reverse

The big practical advantage of UV-C LEDs over mercury lamps is that you get to choose the emission wavelength — a mercury lamp is locked to 254 nm, an LED can be ordered anywhere across the UV-C band. That freedom raises an obvious question: which wavelength actually kills best, and does the answer hold up outside the lab? A new study in Scientific Reports (Nature Portfolio, 3 April 2026) answers it for two of the bacteria that water utilities care about most.

What was tested

The team compared five UV-C LED wavelengths — 255, 260, 265, 270 and 280 nm — against Escherichia coli (Gram-negative) and Enterococcus faecium (Gram-positive), the classic faecal-contamination indicators. Crucially, each species was tested both as a lab culture-collection strain and as a wild isolate recovered from surface water, to check whether environmental bacteria — toughened by real-world sun and stress exposure — resist UV any better. Fluences ran up to 14 mJ/cm². Alongside the standard survival counts, the authors added a single-cell microscopy metric: a skewness analysis of DAPI-stained nucleoids that quantifies how the DNA inside each cell is structurally reorganised by the light — reportedly the first UV-LED study to measure that across wavelengths.

265 nm comes out on top

265 nm produced the fastest inactivation. For the environmental E. coli strain it reached a rate constant of 1.624 ± 0.056 cm²/mJ — the highest of any wavelength or strain in the study — which lines up neatly with the known DNA absorption maximum in the 260–270 nm window. E. faecium, the tougher Gram-positive, was flatter: it inactivated similarly anywhere from 260 to 270 nm and held out a little longer at the lowest doses. But the headline is how little light it took: every wavelength achieved up to 6-log inactivation below 7 mJ/cm², and a 4-log reduction needed just 2–5 mJ/cm². The wild isolates were no harder to kill than the lab strains — environmental adaptation bought the bacteria essentially no extra UV tolerance.

The damage doesn't reverse

Some bacteria can stitch UV damage back together — photoreactivation in light, dark repair without it — which is why under-dosing is dangerous. Here, after 14 mJ/cm², neither species showed substantial recovery under light or dark incubation. Cyclobutane pyrimidine dimers (the signature UV-C DNA lesion) formed at similar levels across all five wavelengths. E. faecium repaired its dimers somewhat better than E. coli, yet neither organism regained the ability to grow — a sign the lethal damage went beyond CPDs alone, into membrane and likely oxidative injury. In short: at an adequate dose, the kill held.

The caveats that matter for system design

Two distinctions keep this honest. First, those strikingly low doses were delivered in clean phosphate buffer. Real drinking water carries turbidity and dissolved organic matter that absorb and scatter UV, so delivered dose at the cell is far lower than dose at the lamp — the study itself notes that operational UV fluences typically run 40–186 mJ/cm². A 2024 review of UV-LED water disinfection makes the same point bluntly: lab-scale success does not automatically translate to field performance, because the water matrix governs how far the light reaches. Read 2–5 mJ/cm² as a wavelength-ranking result, not a plant design number.

Second, the biologically optimal wavelength is not automatically the optimal system. 265 nm wins per delivered photon, but UV-LED wall-plug efficiency, optical output and cost all vary with emission wavelength, and the deepest UV-C wavelengths are not the most efficient to produce — which is why a 2023 study even found 280 nm LEDs out-disinfecting 254 nm mercury lamps in real wastewater on a system basis. Picking a wavelength is an engineering trade-off, not just a microbiology one. And the contrast between E. coli's sharp 265 nm peak and E. faecium's broad plateau is the deeper lesson: there is no single magic wavelength — there is an action spectrum, and the right choice depends on the target organism.