Dégradation du joint des boîtiers UV

A UV-C disinfection module that met its dose specification on the day of commissioning and fails to meet it two years later has usually not lost its LEDs to an electrical fault. It has lost them optically. Deep-ultraviolet photons carry enough energy to break the carbon-carbon and carbon-hydrogen bonds in organic materials, and every organic material in the optical path — the encapsulant, the adhesive, the sealing gasket, the lens coating — is slowly being consumed by the light it is transmitting. This page explains the degradation mechanisms specific to UV-C LED packages, distinguishes them clearly from the quite different ageing of low-pressure mercury lamp tubes, and sets out the material and design choices that determine whether a module holds its output for two years or ten. It is written for disinfection system integrators, water and air treatment engineers, and buyers writing performance specifications for UV-C sources.

1. Why UV-C degradation is a materials problem, not a device problem

The distinction matters because it changes what can be warrantied. A visible-light LED degrades mainly because the die and the phosphor degrade; the package is largely a passive container. A UV-C LED at 275 nm has no phosphor — the light comes directly from the AlGaN active region — and the die itself is comparatively stable. What degrades is everything the light passes through and everything the light touches on its way out of the package and into the reactor.

The photon energy at 275 nm is approximately 4.51 eV. That is above the bond dissociation energy of C–C (about 3.6 eV), C–H (about 4.3 eV) and C–O (about 3.6 eV) bonds. Absorption of a single deep-UV photon is therefore sufficient to break the backbone of a silicone polymer chain. The photochemical consequence is chain scission, followed by cross-linking and oxidation, which together convert a transparent elastomer into a hazy, increasingly absorbing material. This is not a slow thermal process that can be slowed by cooling; it is photochemical, and it proceeds at a rate set by the accumulated UV dose rather than by temperature alone. Cooling extends electrical life but does not restore a degraded optical path.

Separation notice. Two fundamentally different technologies exist in this product family and must never be combined in a specification, a test plan or a warranty discussion. J-15 is a 3535 UV-C LED with a dual-chip 275 nm emitter. Z-14 is a low-pressure mercury lamp tube operating at 254 nm with a 365 nm variant. An LED and a mercury discharge tube have different failure mechanisms (semiconductor and photochemical ageing versus electrode erosion and mercury depletion), different electrical drive requirements, different warm-up behaviour and different replacement economics. Statements about lifetime, dose, temperature dependence or spectral output that apply to one are not valid for the other.

2. The two degradation mechanisms inside a UV-C package

Two mechanisms act simultaneously and are frequently confused because both produce a fall in measured output. They have different signatures and different remedies.

Encapsulant haze. Where the package uses a silicone dome or a silicone layer over the die, the UV photons progressively break the polymer chain. Chain scission creates low-molecular-weight fragments, which recombine and cross-link; oxidation introduces carbonyl and hydroxyl groups, which absorb in the deep UV. The optical result is a rise in bulk scattering — haze — plus a rise in absorption. Because the degraded layer sits directly above the die, it attenuates the emitted light at the source, so the measured output falls even though the junction is healthy. Visually the dome changes from clear to milky or yellow-tinted. The signature is a fall in output accompanied by a change in the angular distribution, because scattering broadens and softens the beam.

Seal degradation. Where the package uses a hermetic window bonded with an adhesive or sealed with an organic gasket, the bond line is a thin organic film directly in the optical path and also the barrier against the environment. UV exposure embrittles and yellows the adhesive; thermal cycling then cracks it; moisture ingress follows, and the die surface, the metallisation or the bond pads corrode. The signature here is different: the output may hold for a long period and then fall relatively sharply, often with an accompanying electrical change as moisture reaches the die. A bright, clear window with a visibly degraded bond line is the diagnostic picture.

3. Why 275 nm is more demanding than the visible spectrum

The comparison with visible-light LEDs is instructive because it explains why UV-C package design cannot reuse the visible-light bill of materials. At 450 nm the photon energy is 2.76 eV, below the C–C bond energy; at 275 nm it is 4.51 eV, above it. The difference is not one of degree. A blue LED emits photons that a silicone can transmit essentially indefinitely; a UV-C LED emits photons that a silicone absorbs and consumes.

The second difference is the availability of UV-stable transparent materials. In the visible range, designers can choose among silicone, epoxy, polycarbonate, PMMA and glass. In the deep UV below 300 nm, the list collapses to fused silica, a few hard glasses, certain fluoropolymers, and thin films. Every one of those choices involves a trade-off in cost, sealing technology, formability or thermal expansion, and the specification must be written around the constraint rather than around the visible-light habit.

4. Décor and design: the thermal and dose environment

Optical degradation is driven by dose, and dose is driven by irradiance and time. The first design question for a long-life UV-C module is therefore not “which LED” but “how much irradiance does the package see internally”. A LED driven at three times its rated current emits roughly two to three times the photon flux, doubling or tripling the internal dose rate and shortening the optical life correspondingly, with no benefit to the reactor if the reactor’s dose requirement was already met.

Junction temperature acts as a second-order accelerator. The photochemical scission itself is largely temperature-independent, but the oxidation step that follows it is thermally activated, and the mechanical stress from thermal cycling is what turns an embrittled bond line into a cracked one. Keeping the junction temperature low therefore extends life in two ways at once: it reduces the electrical wear-out, and it slows the oxidative part of the optical degradation.

5. Verification: measuring what actually degraded

Because two mechanisms produce the same symptom, the test method must be able to separate them. The following sequence does so without destroying the sample.

Measure output in an integrating sphere or with a calibrated radiometer at a fixed, documented geometry and a fixed drive current and case temperature. The geometry matters: because haze broadens the beam, a measurement made with a narrow detector at normal incidence will show a smaller fall than the true loss of total radiant flux. Always report total flux in a sphere alongside, or instead of, a single-point irradiance reading.

Measure transmission of the optical path by comparing a stressed unit against an unstressed unit of the same construction, using a UV spectrophotometer on the window or dome material where a witness sample is available. Where witness samples were mounted alongside the LEDs in the same reactor, this is the cleanest measurement in the whole programme, because it isolates the material from the device.

Measure electrical parameters at fixed temperature. Vf and reverse leakage that remain stable while output falls confirm an optical rather than an electrical failure, and a late rise in leakage confirms that the seal has been breached.

Measure spectral output. A stable spectrum with reduced amplitude is consistent with haze and absorption. A spectral shift suggests a die or drive change and points elsewhere.

6. Specification guidance for a long-life UV-C path

The following specification clauses prevent the most common early-life failures in UV-C modules, and they are written so that a supplier can be held to them.

State the optical material explicitly: window material, seal technology and whether any organic material lies in the optical path. “Quartz window with a hermetic frit or AuSn seal” and “hard glass with an adhesive bond line” are different products with different lifetimes, and a specification that says only “UV-resistant” will be satisfied by the latter.

State the UV dose for which the transmission guarantee holds, not only the initial output. A specification of “≥ 70 % of initial radiant flux after 5,000 hours at rated current and 25 °C case temperature” is verifiable and meaningful; “high UV stability” is not.

State the Température de Jonction condition at which the lifetime figure applies, and require that the module’s thermal design achieves it. For deep-UV LEDs, the output-versus-junction-temperature coefficient is strongly negative, so an undersized heat sink reduces initial output as well as lifetime.

State the cleaning and handling requirement, including prohibition of organic solvents on the window and prohibition of direct contact with the dome.

State the spectral requirement with a tolerance on peak wavelength and a definition of the UV-C band used for dose calculation, so that the module and the radiometer agree.

Separation requirement. Where a project uses both an LED-based UV-C source and a mercury-tube source, the specification must carry separate sections for each. In this portfolio that means J-15 (3535 UV-C LED, 275 nm) and Z-14 (UV sterilization tube, 254 nm and 365 nm variants) are specified, tested, aged and warrantied under separate clauses, with separate lifetime definitions, because the ageing mechanisms, the temperature behaviour and the restrike characteristics are entirely different.

7. Product family and part selection

Note the wavelength boundary. The 365 nm technology appears twice in this portfolio for different reasons: as the second wavelength of the Z-14 tube, produced by a phosphor in a mercury discharge, and as J-12, a UVA LED. They share a nominal wavelength and nothing else: the photon energy at 365 nm (3.40 eV) is below the C–C bond energy, so silicone is far more durable there than at 275 nm, and the seal constraints that dominate J-15 largely fall away for J-12.

8. Common errors and how to avoid them

9. Verification and reporting

A supplier’s UV-C package claim is verifiable when it names the optical material stack, the test conditions, the dose or duration, and the retention criterion against a defined measurement geometry. A complete qualification package for a UV-C part contains the initial spectral and total-flux data, the optical material declaration including any organic constituents, the ageing protocol with the drive current and case temperature, the retention result at each measurement point, the beam distribution at the start and end of test, and, where available, transmission data from witness coupons of the window and seal materials. Where a project mixes LED and mercury-tube sources, the report must be split so that the two technologies are not averaged or compared directly.

10. Referenced standards

• IEC 62471 — Photobiological safety of lamps and lamp systems
• IEC 62471-6 — Photobiological safety of ultraviolet lamp products
• IEC 60529 — Degrees of protection provided by enclosures (IP Code)
• IES LM-80-21 — Approved method: measuring luminous flux and colour maintenance of LED packages, arrays and modules
• IES LM-84 — Approved method: measuring luminous flux and colour maintenance of LED lamps, light engines and luminaires
• IES TM-21-19 — Projecting long term lumen maintenance of LED light sources
• JESD22-A101 — Steady state temperature humidity bias life test
• ISO 15858 — UV-C devices: safety information, permissible human exposure
• IUVA guidance — UV-C LED measurement and dose reporting practice
• ISO 9001 — Quality management systems, for supplier qualification evidence

11. Contact us and sample requests

QUEENDOM supplies the 3535 UV-C LED (J-15, dual chip, 275 nm) together with the UV sterilization tube line (Z-14, 254 nm and 365 nm), and can provide spectral, radiant-flux and ageing data in the format described above. When requesting samples or design support, state the target dose and exposure time, the distance from the source to the working plane, the medium being treated (air, water or surface) and whether ozone or chlorine will be present in the optical cavity, so that the correct optical material stack and sealing technology can be selected and the lifetime quoted against the environment the module will actually see.

Related products and applications

The UV-C emitter and the UV tube line discussed in this paper are listed below.

• UVC 275 nm LED (J-15)
• UV disinfection tube (Z-14)
• UV-A LED (J-12)
• High-power SMD LED (J-01)
• Application overview: LED components application solutions
• More technical papers: LED knowledge resources