1172 IEEE ELECTRON DEVICE LETTERS, VOL. 34, NO. 9, SEPTEMBER 2013 Thermal Transient Effect and Improved Junction Temperature Measurement Method in High-Voltage Light-Emitting Diodes Huaiyu Ye, Xianping Chen, Henk van Zeijl, Alexander W. J. Gielen, and Guoqi Zhang Abstract—The diode forward voltage method with pulsed cur- rents was widely used for monitoring junction temperature (T j ) of light-emitting diodes (LEDs). However, a thermal transient effect (TTE) was observed by the pulsed currents and consequent errors were introduced. Thermoelectric physics was conducted to explain this phenomena and a group of experiments was used to reveal the TTE during T j measurement for high-voltage (HV) LEDs. In addition, an improved pulse-free direct junction temperature measurement method was conducted for HV LEDs to reduce the errors and to achieve in situ T j measurement with dc currents, simpler setups, and a less step sequence. Index Terms— Junction temperature measurement, light-emitting diodes (LEDs), thermal transient effect (TTE), thermoelectric (TE) physics. I. I NTRODUCTION L IGHT-emitting diodes (LEDs) composed of InGaN and yellow phosphors have undergone a very rapid develop- ment over the last decades with the high efficiency, tunable chromaticity, excellent reliability, and environmental benefit [1], [2]. Although LEDs boast very high energy conversion efficiency, they are suffering many problems because of the high junction temperature (T j ) [3]. Among the T j measure- ment methods, the conventional diode forward voltage method or pulsed junction temperature measurement (PJTM) is the most attractive one [4], [5]. A few researchers, however, paid attention to the measurement errors resulting from the pulsed currents. Cain et al. [6] reported the errors introduced by the pulsed currents used for T j measurement of an RF power transistor. Recently, beside the traditional materials, III-nitride Manuscript received June 12, 2013; revised July 14, 2013; accepted July 18, 2013. Date of publication August 6, 2013; date of current version August 21, 2013. This work was supported by the “Consumerizing Solid State Lighting,” the European Commission and AgentschapNL in the ENIAC Joint Undertak- ing under Project E63.9.10397 in the framework of the Research Program of the Materials innovation institute (M2i). The review of this letter was arranged by Editor C. Jagadish. H. Ye is with the Materials Innovation Institute, Delft 2628 CD, The Netherlands, the Delft Institute of Microsystems and Nanoelectronics, Delft University of Technology, Delft 2628 CD, The Netherlands, and also with the Netherlands Organization for Applied Scientific Research, Eindhoven 5612 AP, The Netherlands (e-mail: h.ye@tudelft.nl). X. Chen is with the Faculty of Mechanical and Electrical Engineering, Guilin University of Electronic Technology, Guilin 541004, China. H. van Zeijl and G. Zhang are with Delft Institute of Microsystems and Nanoelectronics, Delft University of Technology, Delft 2628 CD, The Netherlands. A. W. J. Gielen is with the Netherlands Organization for Applied Scientific Research, Eindhoven 5612 AP, The Netherlands. Color versions of one or more of the figures in this letter are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/LED.2013.2274473 Current Rise (c) Si substrate 16mmX4mmX500um LED Chip 1mmX1mmX200um Temperature sensors Wire bonding Wire bonding Silver paste (b) R I LED R L I LED R L LED R L (a) Fig. 1. (a) Model of HV LEDs. (b) HV LED chip was attached on the silicon substrate with temperature sensors and wire bonding. (c) Experimental forward voltage versus sensor temperature for different dc currents. alloys have shown promising thermoelectric (TE) figures of merit [7]–[10]. Blue LEDs that are made of InGaN may perform similarly. Therefore, the TE physics can be used to explain the thermal transient effect (TTE) because of pulsed currents on LEDs [11]; the Peltier effect occurred locally con- fined to the junction and the Joule heating occurred volumet- rically. We observed that a suddenly decreasing temperature with a step-up current and rising temperature with a step- down current. By coincidence, the measured lowest/highest temperature appeared ∼100 μs after the current step, which was also calculated in [12] and [13]. A roughly estimation of the TTE in 500 μs according to the equation [13] is as follows: T = t β + 1  ( M - 1) I 0 ST min,dc l wdc m - ( M 2 - 1) I 2 0 ρ m l 2 w 2 c m  - ( M 2 - 1) I 2 0 ρ d 2 W 2 c β  (1) with β = √ kct /lc m . The maximum height of the transient current pulse used in (1) is, 11.7 times higher than the steady-state current, whereas it is 150 times in our experiment. Then, a very low temperature may achieve in the response time < 100 μs according to [14]. Nevertheless, the TTE introduced lots of 0741-3106 © 2013 IEEE