Junction temperature (Tj) is the temperature of the active p-n junction, and in continuous-operation medical devices it is the master variable that governs both output and lifetime. For UVA 365 nm phototherapy panels and UVC 254 nm disinfection modules running around the clock, engineers typically cap Tj at 75-85°C, and conservative 24/7 designs target 70°C or below. Exceeding 100°C can halve UV output within 5,000 hours.
Why Tj Sets the Reliability Boundary
UV LED degradation is Arrhenius-activated. At higher Tj the rate of non-radiative recombination rises, quantum efficiency droops, and package materials age faster. The visible failure modes are silicone encapsulant darkening, defect generation in the active region, peak wavelength drift, and solder fatigue. As a rule of thumb, every 10-15°C reduction in Tj extends the time to a given output threshold — usually the L70 or L90 point — by a large multiple.
The acceleration factor makes this calculable: AF = exp[(Ea/k)(1/T_use − 1/T_stress)], with an assumed activation energy Ea, commonly taken near 0.5 eV unless the supplier states otherwise. Every degree of junction rise you accept compounds through that exponential.
Choosing the Number
Dosing makes the limit quantitative, not cosmetic. A sterilizer or phototherapy panel must deliver a defined mJ/cm² dose; silent output decay becomes under-treatment. Risk management under ISO 14971 therefore requires the design to control Tj and monitor output, while IEC 60601-1 governs thermal safety around the patient.
| Design case | Tj cap | Rationale | Expected outcome |
|---|---|---|---|
| 24/7 sterilizer, conservative | ≤70°C | maximum margin | L70 > 20,000 h |
| Intermittent phototherapy | 75-85°C | duty cycle allows thermal recovery | L70 > 20,000 h |
| Absolute maximum, datasheet | 90°C+ | never a design point | output may halve by 5,000 h at 100°C |
Computing and Verifying Tj
Compute the worst case from the thermal chain: Tj = Tambient(max) + Pd(max) × Rth(total). Here Pd is Vf × If at the highest drive current, and Rth sums junction-to-pad, pad-to-heatsink through the TIM, and heatsink-to-ambient segments. Design for the worst combination: highest ambient, degraded TIM, end-of-life Vf. Re-verify after any TIM or heatsink change; a 20% rise in contact resistance moves Tj by degrees the exponential never forgives.
Verify on prototype hardware with In-Situ Temperature Measurement (ISTMT): thermocouples on the pad plus forward-voltage thermometry, which uses the junction itself as the temperature-sensitive parameter (TSP) with a calibrated K-factor. Add embedded thermistors for field operation and derating software in the controller: when the sensor reports excess temperature, current or duty cycle drops instead of letting the system run out of specification.
Documenting the Limit
Record the Tj limit in the design history file and demonstrate margin, typically 10-20°C below the absolute maximum at the worst-case point. Select UV LEDs whose datasheets state maximum Tj — commonly 75-90°C for UVA dies — junction-to-pad thermal resistance, and L70 curves at multiple Tj points. Request lifetime data at your actual current and Tj combination, extrapolated per TM-21 from LM-80 style testing, and confirm the supplier can show more than 10,000 hours of real test time. Keep the LM-80 test current and case temperature within 10% of your operating point; extrapolation from far-off conditions hides the real degradation rate. A Tj cap without this data chain is a wish, not a reliability boundary.
FAQ
Q: What happens if Tj briefly exceeds the cap?
A single excursion rarely destroys the die; the damage is accelerated aging of the encapsulant and active region. Log the event, verify output against the dose budget, and treat repeated excursions as a design failure.
Q: Is 100°C Tj acceptable for a UVC module?
Not for continuous medical operation. At 100°C, package yellowing and active-region defects can halve output within 5,000 hours, which breaks dose consistency long before the service interval.
Q: Forward-voltage thermometry or a thermistor — which do I need?
They answer different questions. Forward voltage reads the junction itself during design verification; thermistors guard the field product. A validated medical design uses both.
