Quick answer
The same UV-C that inactivates microbes also injures people. It burns the eye (photokeratitis — "welder's flash") and reddens skin (erythema), and the permitted human exposure is very low: a few seconds near an unshielded germicidal lamp can exceed an entire 8-hour limit. Worse, the symptoms appear hours later, so a worker can be overexposed without feeling anything at the time.
This makes every real UV-C installation a safety-engineering problem, not only an efficacy one. There are two regimes:
- Conventional UV-C (254 nm) — keep people out or shielded. No safe occupied exposure at disinfection intensities.
- Far-UV-C (222 nm) — a genuinely different, far more tolerant hazard profile that may allow disinfection while people are present — but only filtered, with ozone managed, and still within limits. Not a free pass.
1. What UV-C does to the human body
UV-C is absorbed very superficially, which shapes the injury pattern:
- Eyes — the cornea absorbs UV-C, producing photokeratitis: painful, gritty, light-sensitive inflammation that typically appears hours after exposure and is usually transient (the cornea regenerates). It does not normally reach the lens or retina at 254 nm.
- Skin — erythema (sunburn-like reddening), again often delayed. At the 254 nm mercury line, acute effects dominate; the long-term skin-cancer risk is considered lower than for the longer UV-B wavelengths (>280 nm), though not zero.
The deceptive part is the delay and the absence of warning: there is no heat, no immediate pain, so people don't self-limit. Engineering controls — not human caution — have to do the protecting.
2. The exposure limits — and how low they really are
Human UV exposure is capped by the ICNIRP guidelines (adopted with the WHO) and the closely-aligned ACGIH Threshold Limit Values (TLVs) — the dose a worker may receive over an 8-hour day:
- The harmonised limit is 30 J/m² = 3.0 mJ/cm² of spectrally-weighted effective radiant exposure over 8 hours (weighted by the actinic action spectrum across 180–400 nm; the 3.0 mJ/cm² figure references 270 nm, where the cornea is most sensitive).
- For the 254 nm mercury line specifically, ACGIH gives a permissible 8-hour exposure of about 6.0 mJ/cm² — equivalent to roughly 0.2 µW/cm² for 8 hours, or 60 mW/cm² for just 0.1 second.
| Reference | Limit (8 h) | Basis |
|---|---|---|
| ICNIRP / ACGIH effective | 3.0 mJ/cm² (30 J/m²), spectrally weighted | 270 nm corneal max |
| ACGIH at 254 nm | ~6.0 mJ/cm² | mercury germicidal line |
Put concretely: at an irradiance of a few µW/cm² — the kind of stray light around a running germicidal lamp — the 8-hour limit is reached in seconds to a couple of minutes. (A scientific debate exists on whether the 254 nm limit is over- conservative given how little UV-C reaches living corneal tissue; reputable bodies are reviewing it, but the current limits remain the design basis.)
3. Far-UV-C (222 nm) — a genuinely different hazard profile
Far-UV-C from filtered krypton-chloride (KrCl) excimer lamps behaves differently because 222 nm penetrates tissue even less than 254 nm — it is largely absorbed in the outer, non-living layer of skin and in the tear film and corneal epithelium, so it struggles to reach living, dividing cells.
The numbers reflect that much higher tolerance:
- ICNIRP sets a far-UV exposure limit on the order of 23 mJ/cm² over 8 hours — several times the 254 nm value.
- In studies of filtered KrCl deployment, no eye irritation was observed up to ~50 mJ/cm² head exposure; corneal surface integrity was maintained to ~600 mJ/cm², with meaningful penetration only at far higher doses.
This is why far-UV-C is promising for disinfecting occupied spaces. But three honest caveats keep it from being "safe light":
- Filtering is mandatory. Unfiltered KrCl lamps emit longer-wavelength tails that are hazardous; the safety case depends entirely on optical filtering.
- Ozone. 222 nm generates ozone, a separate respiratory hazard that must be ventilated/managed.
- It is still limited. "More tolerant" is not "unlimited" — exposure limits and good design still apply.
(For the efficacy and concept of far-UV-C, see the dedicated far-UV-C article; this section is the human-safety lens only.)
4. Engineering controls — how UV-C is made safe
Safety comes from design, layered defence-in-depth, not from trusting people:
- Shield / aim away — keep direct UV off occupants. Upper-room systems put the germicidal zone above head height behind louvers, relying on vertical air mixing (see air disinfection) to bring air up to the light, not light down to people.
- Interlocks — door switches that cut the lamps the instant a room is entered.
- Occupancy / vacancy sensing — motion + "no-motion" timers (e.g. wait 1–4 minutes after last movement) before energising; ideally two independent sources (e.g. vacancy plus CO₂) for redundancy.
- Timers & sequencing, warning signage (eyes and skin must be protected; standardised UV warning symbols), and building-management integration so a fault visibly fails safe.
- PPE (UV-blocking eyewear, skin cover) for any maintenance with lamps live.
The principle mirrors the efficacy side: don't trust the nameplate or the operator — engineer the exposure to zero for anyone who shouldn't receive it.
5. Where the rules live (regulatory frame)
- ICNIRP UV exposure guidelines (adopted with the WHO) and ACGIH TLVs — the international basis for the limits above.
- National occupational frameworks translate these into workplace duty — in Germany via the DGUV (statutory accident-insurance) rules on optical radiation; equivalents exist per country. Pair with the LUVEX regulatory wayfinder for the authority pathways.
- Far-UV-C limits specifically are still evolving as the 222 nm evidence base matures — design with margin and watch for updates.
6. Safety is part of the spec, not an afterthought
Every UV-C design should answer three safety questions alongside the efficacy ones:
- Who could be exposed, and at what dose?
- How is that exposure prevented (shield / interlock / occupancy / scheduling)?
- What is the failure mode — does a fault fail safe (lamps off) or open?
Disinfection that protects health by harming the people who run it is not a solution. The complete UV-C engineering loop is: design the dose, validate it per medium — and protect everyone who shares the space.
Cross-references
- The efficacy foundation — Dose · Validation · Air/Surface/Water.
- Far-UV-C (concept & efficacy) — the dedicated far-UV-C article (this one is the safety lens).
- Regulatory wayfinder — DGUV/ICNIRP authority pathways for occupational UV.
Sources
- ACGIH 2021 Threshold Limit Values (TLVs) and Biological Exposure Indices — UV summary — 254 nm 6.0 mJ/cm² 8-h limit.
- ICNIRP — Guidelines on Limits of Exposure to UV Radiation (180–400 nm) and ICNIRP — Protection of Workers against UV Radiation — 30 J/m² effective, action spectrum.
- University of Rochester EH&S — UV Light Guidelines — photokeratitis/erythema, delayed onset, 254 nm hazard.
- AIHA — Occupational Safety & Health Guide for Surface Disinfection using Germicidal UV — occupational controls.
- A Need to Revise Human Exposure Limits for UV-C Radiation (PMC) — the limit-revision debate.
- 222 nm Far-UVC from filtered KrCl does not cause eye irritation (Kousha 2024, Wiley) — far-UV eye tolerance (50/600 mJ/cm²).
- Care222 — far-UVC research & ICNIRP 23 mJ/cm² far-UV limit — far-UV exposure limit.
- GLHN Architects & Engineers — All About Safety: UV-C Hazards & Risks — engineering controls, interlocks, occupancy.