Ultraviolet-C disinfection is a dose problem, not a brightness problem. A reactor that delivers 20 mJ/cm² at 275 nm will not clear a wastewater stream that needs 60 mJ/cm², no matter how many emitters are added to the wall. This white paper is written for engineers who must size a UVC LED chamber — in water treatment, air handling, surface disinfection or appliance integration — and who need to connect emitter specifications to a deliverable disinfection target. We explain the D90 concept, build the dose-versus-survival relationship from first principles, and show how QUEENDOM’s J-15 3535 UVC dual-chip 275 nm emitter fits into that workflow. We also draw a strict line between UVC LEDs and the 254 nm low-pressure mercury tubes that remain in legacy equipment.
1. Why dose, not power, defines a UVC reactor
The germicidal effect of ultraviolet light depends on the product of irradiance and exposure time, integrated over the target surface. That product is fluence, measured in joules per square centimetre (J/cm²) or, in the units used by almost all disinfection literature, millijoules per square centimetre (mJ/cm²). A 10 mW/cm² field applied for one second delivers 10 mJ/cm². The same 10 mJ/cm² can be delivered by 1 mW/cm² for ten seconds or 100 mW/cm² for 0.1 second, and the biological outcome is, to first order, identical.
This is why power ratings on a datasheet are an intermediate quantity, never a deliverable. What the customer actually buys is a reduction ratio — typically 3-log (99.9 %), 4-log (99.99 %) or 5-log (99.999 %) inactivation of a named organism. The engineering task is to translate that reduction into a required dose, and then to design an optical and hydraulic or aerodynamic path that delivers that dose to every part of the flow.
| Target organism | Typical dose for 1-log (mJ/cm²) | 4-log dose (mJ/cm²) | Wavelength basis |
|---|---|---|---|
| Escherichia coli | 5–7 | 20–28 | 265–275 nm |
| Pseudomonas aeruginosa | 6–8 | 24–32 | 265–275 nm |
| Staphylococcus aureus | 6–7 | 24–28 | 265–275 nm |
| Legionella pneumophila | 5–8 | 20–32 | 265–275 nm |
| Bacillus subtilis spores | 22–30 | 90–120 | 265–275 nm |
| Candida albicans | 8–12 | 32–48 | 265–275 nm |
| MS2 bacteriophage | 18–22 | 72–88 | 265–275 nm |
The spread in this table is the single most important design fact in the discipline. A spore-forming organism such as Bacillus subtilis needs roughly four times the dose of E. coli for the same log reduction. A specification that quotes a reduction figure without naming the organism is not a specification.
2. The D90 concept and the log-inactivation model
D90 is the dose required to reduce a population to 10 % of its initial value — one decimal reduction, or one log. It is defined by the log-linear (Chick-Watson) model:
“ N / N0 = exp(-k · D) → log10(N / N0) = -D / D90 “
where N0 is the initial count, N the surviving count, D the delivered dose in mJ/cm², and D90 the decimal reduction dose. Rearranging, the dose needed for an n-log reduction is simply D = n · D90. A 4-log target with D90 = 6 mJ/cm² requires 24 mJ/cm². A 4-log target with D90 = 25 mJ/cm² requires 100 mJ/cm². The D90 value is the organism’s property; the dose is the designer’s decision.
Real survival curves are not perfectly straight. Above roughly 3-log, a shoulder or a tailing region frequently appears because organisms clump, hide in crevices, or are shielded by particles. Good practice is to apply a design margin of at least 1.5× to 2× over the log-linear prediction, and to validate the delivered dose with biodosimetry rather than trusting the optical calculation alone.
3. Key parameters and how they map onto the J-15 emitter
The J-15 is a 3535 ceramic package carrying two UVC dice in a single footprint, peaking at 275 nm. The dual-chip construction matters for two reasons. First, it doubles the radiant flux available per placed component, which reduces board area and component count in a dense reactor wall. Second, it lets the designer drive the pair in series at a lower current per die for the same total output, which lowers junction temperature and slows the flux depreciation that dominates UVC LED lifetime.
| Parameter | J-15 3535 UVC dual-chip | Notes |
|---|---|---|
| Peak wavelength | 275 nm (typ.), bin 270–280 nm | Matches the 260–280 nm germicidal peak |
| FWHM | ≤ 12 nm | Narrow band; no broadband UV spill |
| Radiant flux | 40–60 mW @ 350 mA, 25 °C case | Representative; bin dependent |
| Forward voltage | 5.5–7.0 V @ 350 mA (two dice in series) | Series wiring, single driver channel |
| Max forward current | 500 mA continuous | De-rate above 60 °C case |
| Viewing angle | 120° (typ.) | Lambertian; needs reflector for collimation |
| Thermal resistance, junction-to-case | ~12 K/W | Ceramic body; MCPCB mandatory |
| Package | 3.5 × 3.5 mm ceramic | AlN substrate, quartz or hard-glass window |
| Estimated L70 | > 10,000 h at 350 mA, 55 °C case | Projection basis only |
| Wall-plug efficiency | 2–5 % (UV-C band) | Typical of AlGaN 275 nm devices |
The window material deserves a note. Standard silicone encapsulation degrades rapidly under UV-C, so the J-15 uses a quartz or hard-glass window instead. Any designer who re-flows the part under a profile intended for a visible-white emitter risks contaminating the optical path, and the resulting drop in transmission will look like an emitter failure when it is actually an assembly fault.
3.1 Survival ratio versus dose
Expressed as survival rather than log reduction, the same data show why the last decade of reduction is so expensive. Moving from 3-log to 4-log requires a 33 % increase in dose; moving from 4-log to 5-log requires another 25 %. At the tail of the curve, the marginal cost per additional decade rises sharply.
3.2 Wavelength sensitivity of the germicidal response
Germicidal effectiveness is not flat across the UV-C band. It rises steeply from 240 nm, peaks near 265 nm, and falls back through 280 nm. A 275 nm emitter retains roughly 85–90 % of the effectiveness of an ideal 265 nm source on typical organisms, while avoiding the far-UV absorption problems encountered by shorter-wavelength approaches. This is the engineering reason why 275 nm has become the main commercial UVC LED wavelength.
| Lunghezza d'Onda | Relative germicidal effectiveness (typ.) | Material notes |
|---|---|---|
| 255 nm | ~0.72 | Low Al content, higher output power available |
| 265 nm | ~1.00 | Theoretical peak; harder to manufacture at high flux |
| 275 nm | ~0.85–0.90 | J-15 operating point — best flux/perf balance |
| 285 nm | ~0.60 | Falling off the germicidal peak |
| 254 nm | ~0.90 | Low-pressure mercury tube line, not an LED |
4. Sizing a reactor: from target log reduction to emitter count
The design sequence below is the one we recommend to integrators, because it fails early and cheaply when the target is unreachable.
- Fix the organism and the log target. Write the specification as “4-log inactivation of E. coli in 20 °C potable water”, not as “99.99 % sterile”.
- Convert to required dose. Multiply the log target by the organism’s D90, then apply a design margin. For 4-log E. coli with D90 = 6 mJ/cm², that is 24 mJ/cm² predicted and 36–48 mJ/cm² specified.
- Compute the required irradiance. Divide the specified dose by the minimum residence time of the slowest fluid element, not the mean residence time. In a poorly baffled chamber the slowest path may be three times longer than the mean, and the fastest path may exceed the mean by 80 %, which is the path that survives.
- Budget optical losses. Window transmission, reflector efficiency and water absorption all reduce delivered irradiance. Typical end-to-end efficiency for a quartz-window reactor ranges from 55 % to 75 %.
- Count emitters. Divide required irradiance by the per-emitter contribution at the reactor surface, then add thermal de-rating.
- Validate by biodosimetry. Dose calculations are necessary but not sufficient; the delivered dose must be measured.
4.1 Emitter count versus dose target
The curve below shows how emitter count scales with required dose for a 25 mm water gap, using J-15 emitters on a tubular reactor wall. The relationship is linear in dose and inversely proportional to per-emitter useful flux, which is why double-chip parts reduce component count so effectively.
5. Product comparison: 275 nm UVC LED against 254 nm mercury tube
The single most common error in UVC procurement documents is a table that lists “254 nm LED” as a line item. There is no such device in volume production. The 254 nm line is the resonance line of low-pressure mercury vapour; it belongs to discharge tubes, not to semiconductor emitters. In the QUEENDOM portfolio the split is explicit: J-15 3535 UVC dual-chip 275 nm is the LED, while Z-14 UV sterilization tubes at 254/365 nm are mercury tubes. The table below keeps the two technologies separate, because they differ in warm-up, dimming, form factor, mercury content and regulatory treatment.
| Attribute | J-15 3535 UVC LED (275 nm) | Z-14 UV tube (254 nm) |
|---|---|---|
| Emitting mechanism | AlGaN semiconductor junction | Low-pressure mercury discharge |
| Peak wavelength | 275 nm | 254 nm (plus 185 nm in some types) |
| Warm-up to full output | < 1 s (instant) | 30 s to 3 min |
| Dimming / modulation | Continuous via current; PWM capable | Limited; not suited to fast modulation |
| Mercury content | None | 2–10 mg per tube (RoHS-exempt category) |
| Package | 3.5 × 3.5 mm ceramic SMD | Linear or U-shaped glass tube |
| Typical wall-plug efficiency | 2–5 % | 25–35 % |
| Typical lifetime | 10,000–15,000 h to L70 | 8,000–12,000 h |
| Switching cycles | Effectively unlimited | Degrades electrode with each start |
| Form-factor flexibility | Point source, arrayable | Fixed tube lengths |
| Best fit | Compact appliances, point-of-use, portable, cold-start | Large municipal water, HVAC duct banks |
5.1 Energy cost per delivered dose
Wall-plug efficiency is where the mercury tube wins on paper and where the LED wins in practice once duty cycle is considered. A tube running continuously in an intermittently used appliance wastes energy between cycles; an LED that switches instantly can be energised only during flow. The break-even point usually falls near a 30–40 % duty cycle for the same delivered dose.
| Scenario | J-15 UVC LED (275 nm) | Z-14 UV tube (254 nm) |
|---|---|---|
| Continuous operation, 10 mJ/cm² target | ~14 W electrical | ~3 W electrical |
| 25 % duty cycle, same target | ~3.5 W average | ~3 W average (limited dimming) |
| Cold start to full dose | Immediate | 30–180 s delay |
| Maintenance interval | Emitter replacement, no mercury handling | Tube replacement, mercury disposal |
| Housing size for 10 mJ/cm² | 40 × 20 × 15 mm typical | 150 mm tube plus ballast |
6. Common mistakes and how to avoid them
| Mistake | Consequence | Correction |
|---|---|---|
| Writing “254 nm LED” in a specification | No conforming supplier; project stalls | Specify J-15 at 275 nm for LEDs; Z-14 for tube-based 254 nm |
| Quoting a log reduction without naming the organism | Dose requirement undefined | Always pair reduction with organism and medium |
| Using mean residence time | Fast-path organisms survive | Size on the fastest fluid element, or add baffling |
| Ignoring UV-C absorption in water | Delivered dose 20–60 % below prediction | Measure transmittance at 275 nm before sizing |
| Grading emitter by wall-plug power only | Over-sized thermal solution, mis-set UV dose | Specify on radiant flux in the UV-C band |
| Bonding with standard silicone | Rapid window degradation and flux loss | Use quartz/hard-glass window and UV-stable adhesive |
| No degradation margin | Dose falls below target within a year | Budget 20–30 % flux loss over service life |
| Relying on calculation alone | Undetected short-circuiting of flow | Validate with biodosimetry at commissioning |
7. Verification and test methods
Four tests close the loop on a UVC design.
- Radiant flux and spectrum — measure each emitter or a sampled batch on a calibrated spectroradiometer. Confirm peak wavelength within the 270–280 nm bin and FWHM ≤ 12 nm for J-15.
- Irradiance mapping — scan the disinfection plane with a calibrated UV-C radiometer at a minimum of nine points, reporting maximum-to-minimum ratio and the position of the minimum. A well-designed chamber typically achieves ≤ 1.3:1 with appropriate reflectors.
- Biodosimetry — inject the target organism at the design concentration, operate at the lowest specified flow rate and temperature, and enumerate survivors by standard membrane filtration or spread-plate methods. This measures delivered dose including all hydraulic short-circuiting.
- Lifetime and maintenance — run an LM-80-style flux maintenance test at rated current and report L70 projections using TM-21. Re-verify irradiance at the halfway and end of the maintenance interval.
8. Conclusion and selection guidance
For the great majority of point-of-use, appliance and compact air-treatment designs, J-15 3535 UVC dual-chip 275 nm is the correct starting point: two dice per footprint reduce component count, the 275 nm peak sits close to the germicidal optimum, and the ceramic package with a quartz window tolerates the thermal and optical stress of continuous UV-C operation. Design the reactor on dose, not on power; size on the fastest fluid element; leave a 1.5× to 2× margin over the log-linear prediction; and confirm the result with biodosimetry. Where very large volumes of water must be treated continuously and mercury-based equipment is acceptable in the jurisdiction and application, legacy Z-14 254 nm tubes remain a valid alternative — but they are a different technology and must never be described as LEDs.
9. Referenced standards
- IEC 62471 — Photobiological safety of lamps and lamp systems
- IEC 60335-2-109 — Household and similar electrical appliances: particular requirements for UV radiation water treatment appliances
- ISO 15858 — UV-C devices: safety information — permissible human exposure
- IES LM-80 — Approved method: measuring luminous flux maintenance of LED light sources
- IES TM-21 — Projecting long-term lumen maintenance of LED light sources
- DVGW W294 — UV disinfection systems for drinking water supply (German technical standard, cited for dose-validation methodology)
- US EPA UV Disinfection Guidance Manual — biodosimetry and reactor validation methodology
10. Contact us
QUEENDOM supplies the J-15 3535 UVC dual-chip 275 nm emitter from stock, together with reference reactor layouts, irradiance mapping templates and dose-calculation worksheets. We can provide spectral characterization data and binning statements for design verification. For UV-C systems that require camera or sensor feedback, our J-12 3535 UVA (365/375/385 nm) and J-10 3 mm UVA (405 nm) parts are available as excitation and calibration sources. Contact our component engineering group for samples and application support.
Related products and applications
The UVC emitter and the reference UV-A parts used for comparison are available in the following families.
- UVC 275 nm LED (J-15)
- UV-A LED (J-12)
- UV-A LED (J-10)
- Application overview: LED components application solutions
- More technical papers: LED knowledge resources