Abstract
In this paper, seven LED lamps are selected to investigate how the junction temperature ($T_j$) of LED varies during the step-stress temperature accelerated aging test. The ambient temperatures are 80 °C and 70 °C for the first and second step of the aging test, respectively, and the total aging time is 2180 h. A non-contact method for monitoring junction temperatures of samples, by the ratio of white light energy to blue light energy (W/B), is applied. Experimental measurements show that the W/B ratio and $T_j$ satisfy good linear relationship during the aging for all samples. Then, the junction temperatures are acquired at eight typical points of aging time which are used to eliminate the impact of $T_j$ variation on the estimation of the decay rate and accelerated lifetime. It is shown that the estimated lifetime is longer than that acquired by the traditional method. The estimation errors of the accelerated lifetimes by the traditional method are in a range from 12.8% to 18.6% under the aging temperature of 80 °C and from 12.5% to 20.3% under the aging temperature of 70 °C for the seven samples. The estimation accuracy of the decay rates and accelerated lifetimes of LED lamps in the step-stress accelerated aging test is improved by the proposed methodology to a certain extent.
Introduction
LEDs are renowned for their longevity, often rated for tens of thousands of hours. To validate these claims efficiently, engineers employ Accelerated Aging Tests (AAT), subjecting LEDs to high ambient temperatures to accelerate degradation. However, a critical flaw exists in traditional AAT models: the assumption of constant internal temperature.
Standard testing protocols (such as IESLM-79) assume that if the ambient temperature is held constant, the LED's internal junction temperature ($T_j$) remains stable. In reality, as an LED ages, its efficiency declines, and more energy is dissipated as heat. This causes $T_j$ to rise over time. Since luminous flux naturally decreases as temperature increases—a phenomenon known as "thermal sag"—traditional models often mistake this temporary thermal effect for permanent aging damage, leading to overly pessimistic lifetime predictions.
Experimental Methodology
To address this, researchers developed a methodology to decouple thermal shifts from actual aging degradation. The study utilized seven white LED samples containing yellow ($Y_3Al_5O_{12}:Ce$) and red ($CaAlSiN_3:Eu$) phosphors.
The aging process followed a two-step stress profile: 80°C for 980 hours, followed by 70°C for 1200 hours. A key innovation was the use of a non-contact $T_j$ monitoring method based on the W/B ratio (the ratio of white light energy to blue light energy). Because the blue chip and phosphors respond differently to heat, this optical ratio serves as a precise internal thermometer without requiring physical probes that might interfere with thermal dissipation.
Decoupling Thermal Effects
Before aging, the researchers measured the temperature coefficient ($\mu$) for each LED, quantifying how much light output drops per degree of $T_j$ increase.
As shown in the figure above, there is a clear linear relationship between temperature and lumen output. By identifying this coefficient, the researchers could "correct" the measured light output during the aging test, effectively normalizing the data as if the junction temperature had remained constant at its initial value.
Key Findings
The study revealed several critical insights:
- Junction Temperature Creep: Even under stable ambient conditions, internal junction temperatures increased by 9.5°C to 11.4°C over the 2180-hour test period.
- Significant Estimation Errors: Traditional methods, which ignore this $T_j$ rise, resulted in lifetime estimation errors ranging from 12.8% to 20.3%.
- Underestimated Longevity: In every case, the corrected lifetime was longer than the traditional estimate. For example, for "LED1" at 80°C, the traditional method predicted 3330 hours, while the corrected method predicted 3954 hours—a nearly 19% increase in expected life.
Conclusion
By implementing a non-contact optical monitoring method to correct for internal thermal shifts, this research demonstrates that LED lifespans are often significantly longer than traditional accelerated tests suggest. Decoupling thermal "noise" from actual degradation is essential for accurate reliability engineering in the solid-state lighting industry. Future work may explore these patterns in high-power COB LEDs and consider the impact of driver electronics degradation.
Summary updated