Operating lifetime study of ultraviolet (UV) light-emitting diode products

Light-emitting diodes (LEDs) can emit radiation that spans the range from near infrared (IR) to all three bands of ultraviolet (UV) radiation: UV-A, UV-B, and UV-C. This report focuses on LEDs that emit in one of the three UV bands because they have the potential to displace low-pressure mercury vapor (LPMV) lamps in a variety of industrial processes, including ink and adhesive curing, medical procedures, and germicidal disinfection. However, before emerging UV LED technologies can displace LPMV lamps, the efficiency and reliability of these sources must meet the user’s expectations in each application. An earlier report focused on the construction and initial performance of commercial UV LED products in radiometric and current-voltage (I-V) tests [1]. This report focuses on the long-term performance and reliability of the same set of commercial products. The intent of this report is to provide to the lighting industry a benchmark of the state of UV LEDs as of mid-2021 when these products were purchased. Understanding the failure modes and failure rates of UV LEDs is important in improving UV product reliability at the LED, lamp, and luminaire level and is critical to developing products with higher efficiency, lower carbon footprint, and significantly reduced environmental impact than LPMV lamps.
The lifetime testing results for a series of 13 commercial UV LED products, spanning the range from UV-A to UV-C, are provided in this report. These products were chosen from commercially available UV LEDs in conjunction with the LED Systems Reliability Consortium (LSRC). All products were tested as packaged die, soldered to a metal-core printed circuit board (MC-PCB), and mounted on a heat sink. Three stress tests were used during this study to examine the reliability of each product when subjected to current, temperature, and humidity stressors that may be encountered during normal operation. Separate, independent populations of devices were used for each product during the three tests, and no device under test (DUT) experienced more than one stress test. The three stress tests are as follows:
• A room temperature operating life (RTOL) test conducted near the maximum forward current (If) for each product as given by the manufacturer’s specifications. This test is termed RTOL-1, and there were 10 DUTs for each product used in this test matrix.
• A second RTOL test conducted at a lower If value that was 15% to 65% of that used during RTOL-1. This test is termed RTOL-2, and there were 10 DUTs for each product used in this test matrix.
• A temperature-humidity (T-H) storage test conducted in an environmental chamber set to 50 degrees Celsius (°C) and 90% relative humidity. This T-H storage test was performed with the LEDs in an unpowered state. A subset of three products from the test matrix was chosen for the T-H storage tests, and there were three DUTs for each product tested.
The 13 UV LED products were divided into two groups: the RTOL-1 test lasted for 1,000 hours (hrs) for Group-1 and 500 hrs for Group-2. Between the two groups, a total of 57 failures (which is 44% of the total product population) occurred during RTOL-1 that were classified as either abrupt failures (i.e., produced no UV radiation) or parametric failures (i.e., radiant flux maintenance [RFM] was 0.50 or lower). Four products (i.e., UV-3, UV-5, UV-9, and UV-11) did not experience any failures during RTOL-1, but the other nine products had failure rates ranging from 10% to 100%. The most common failure mode was a parametric failure, which accounted for 50 failures during the test. Only seven DUTs failed abruptly, and four out of these seven devices were UV-2 DUTs. These findings demonstrate that there is a wide range of reliability in commercially available UV LED products.
Different DUTs of the 13 UV LED products were divided into two groups for RTOL-2 based on their power rating. Group-A, which had If values less than 55 milliampere (mA), was operated for 2,500 hrs during RTOL-2, whereas Group-B, which had If values greater than 100 mA, was operated for 3,000 hrs during RTOL-2. Even though product operation times were longer in RTOL-2 compared with RTOL-1, the cumulative failure rate was lower, as might be expected for the milder stress conditions (i.e., lower If values). A total of 32 failures occurred during RTOL-2 (which is 25% of the total product population): 11 abrupt failures and 21 parametric failures. This finding corresponds to a cumulative failure rate of 25%. Most of the failures during RTOL-2 were concentrated in fewer products; this observation was in contrast to findings from RTOL-1, during which failures were more widespread. For example, during RTOL-2, 7 abrupt failures occurred in the UV-2 DUTs and a greater than 50% parametric failure rate was measured for the UV-8 and UV-14 DUTs. These three products accounted for 22 out of the 32 failures recorded in RTOL-2.
Based on this work and combined with an extensive literature review, the following failure mechanisms could be identified across these UV LEDs:
• Abrupt failure caused by macroscopic defects in the semiconductor material. This failure mode was most prevalent in UV-2, but appeared sporadically in a couple other products that were tested.
• Contributions to parametric failure caused by the growth of parasitic diode circuits that reduce radiant carrier recombination through modification of Shockley-Read-Hall (SRH) mechanisms. The parasitic diode emerges early in the operational life of the UV LED and can be identified through a characteristic hump in the I-V profile. This mechanism was especially prominent in the aluminum gallium nitride (AlGaN) LEDs with nominal maximum emission wavelength (λmax) at either 275–280 nanometers (nm) or at 310 nm. This mechanism occurred to a smaller extent in indium gallium nitride (InGaN) LEDs with λmax of 365 nm.
• Contributions to parametric failure likely caused by the degradation of silicones (e.g., cracking, discoloration) used in encapsulation of the UV-A LEDs. This failure mechanism only occurred in the InGaN LEDs with λmax of 365 nm because the AlGaN LEDs do not use silicones. This failure mechanism may account for the poor RFM performance of UV-12 and UV-14 during RTOL-1 and UV-14 during RTOL-2.
• Contributions to parametric failure caused by package effects likely arising from organic residues from lid seal processes and processes that use conductive adhesives (e.g., protective diode attach). The existence of this failure mechanism was supported by T-H storage tests, during which the radiant flux declined even though the DUTs were unpowered during test and their I-V curves did not change during the 148 hrs of this test.
In addition to identifying some of the failure mechanisms occurring in commercial UV LED products, one of the most significant findings from this research was that not all of the LEDs of a given product exhibited the same behavior within the test duration. For example, 60% of the UV-8 DUTs during both RTOL-1 and RTOL-2 were parametric failures by the end of each test. The DUTs that exhibited parametric failures also displayed a rise in leakage current flowing through a parasitic parallel diode circuit as shown by the changes in the I-V curve presented in Figure ES-1. In contrast, 40% of the UV-8 DUTs during RTOL-1 and 20% of the DUTs during RTOL-2 exhibited no changes in the I-V curve at test completion. In short, these DUTs showed no evidence of the emergence of a significant parallel diode circuit during operation. During the milder RTOL-2 experiment, there was an addition sample sub-population in-between these two extremes. During RTOL-2, 20% of the UV-8 DUTs exhibited significant leakage current that was less than that of the parametric failures; however, these DUTs exhibited acceptable RFM values that were lower than those of DUTs with minimal leakage current. Radiometric studies of the sub-populations indicated that the RFM values were significantly higher for DUTs that did not exhibit the parallel diode circuit. This finding suggests that a burn-in test of 500-hr duration followed by I-V measurements can be used to quickly identify UV LED products that are prone to significant RFM loss through the formation of a parasitic diode circuit during operation.


Figure ES-1. I-V curves for six DUTs of UV-8 as initially received (solid lines) and after 2,000 hrs of RTOL-2 (dashed lines). The emergence of the parasitic diode in parallel with the UV LED is demonstrated by the hump in the I-V curve between 0.5 volts (V) and 4 V.
The wide variation in behavior in radiometric and I-V tests observed for UV-8 and other UV LEDs during this study was likely the consequence of process variability during semiconductor manufacturing. Reducing this process variability across the wafer will likely produce a more consistent lifetime and efficiency for UV LED products. Such process consistency would reduce or eliminate the formation of parallel diode circuits during UV LED operation, a leading failure mechanism found in the products examined in this study.
Improving the packaging of UV LEDs by developing better encapsulation polymers and reducing or eliminating outgassed organic residues from adhesives and solder fluxes would likely also improve the RFM performance of UV LEDs. Encapsulation polymers are important to improve light extraction efficiency and to provide environmental protection. Silicones are a common encapsulant and can be found in three UV-A products that were tested during this study. Unfortunately, two out of the three products quickly displayed large RFM losses during testing likely because of silicone degradation. The other product (i.e., UV-11) exhibited RFM values greater than 0.95 under all test conditions, demonstrating that some silicones can be used as encapsulants in UV-A LEDs. In general, the materials and processes currently used for fabricating blue and white LEDs may not be appropriate for UV LEDs because of their different emission wavelengths and semiconductor chemistries.
More research is needed by the industry to improve the efficiency and long-term performance of InGaN and AlGaN materials and their LED packages. The findings from this research will be critical to delivering on the promise of high efficiency and a long lifetime for UV LED products The findings from this research will also be helpful in creating new UV technologies for building systems, personal safety, and industrial and medical application that will be energy efficient, low carbon producing, and environmentally friendly and will improve human well-being and health.