Affordable Access

Publisher Website

Reliability issues in III–V compound semiconductor devices: optical devices and GaAs-based HBTs

Microelectronics Reliability
Publication Date
DOI: 10.1016/s0026-2714(99)00193-6


Abstract This paper reviews the current status of reliability issues in III–V optical devices, semiconductor lasers and light emitting diodes, and GaAs-based heterojunction bipolar transistors (HBTs). First, material issues in III–V alloy semiconductors and our current understanding of degradation in III–V semiconductor lasers and light emitting diodes are systematically presented. Generation of defects and thermal instability are among these issues for these systems. Defects introduced during crystal growth are classified into two types: interface defects and bulk defects. Defects belonging to the former type are stacking faults, V-shaped dislocations, dislocation clusters, microtwins, inclusions, and misfit dislocations, and the latter group includes precipitates and dislocation loops. Defects in the substrate can also be propagated into the epi-layer. Structural imperfections due to thermal instability are also found. They are quasi-periodic modulated structures due to spinodal decomposition of the crystal either at the liquid–solid interface or growth surface, and atomic ordering which also occurs on the growth surface through migration and reconstruction of the deposited atoms. Three major degradation modes of optical devices, rapid degradation, gradual degradation, and catastrophic failure, are discussed. For rapid degradation, recombination-enhanced dislocation climb and glide are responsible for degradation. Differences in the ease with which these phenomena occur in different heterostructures are presented. Based on the results, dominant parameters involved in the phenomena are discussed. Gradual degradation takes place presumably due to recombination enhanced point defect reaction in GaAlAs/GaAs-based optical devices. This mode is also enhanced by the internal stress due to lattice mismatch. However, we do not observe this mode in InGaAsP/InP-based optical devices. Catastrophic failure is found to be due to catastrophic optical damage at a mirror or at a defect in GaAlAs/GaAs double-heterostructure (DH) lasers, but not in InGaAsP/InP DH lasers. In each degradation mode, the role of defects in the degradation and methods of elimination of degradation are discussed. Secondly, we review the current status of two major reliability issues in GaAs-based HBTs, particularly InGaP/GaAs HBTs: degradation in current gain ( β) and variation of turn-on voltage ( V be). In the case of AlGaAs/GaAs HBTs, the β gradually decreased, then drastically degraded. After degradation, the device exhibits an increase in base current I b, which has an ideality factor n≈2 in the Gummel plot. The activation energy for the degradation was estimated to be 0.6 ± 0.1 eV. On the other hand, in InGaP/GaAs HBTs, much higher reliability than in AlGaAs/GaAs HBTs was achieved although the degradation mode is similar. The estimated E a and time to failure for InGaP/GaAs HBTs are 2.0 ± 0.2 eV and 10 6 h at T j = 200°C, respectively, which are the highest values ever reported. We also review previously proposed degradation mechanisms for GaAs-based HBTs: hydrogen reactivation, microtwin-like defect formation, dark defect formation and carbon precipitation. TEM observation of a degraded InGaP/GaAs HBT indicated that there are at least two possible degradation mechanisms: formation of carbon precipitates in the base region and migration of metallic impurities from the base electrode to the base region. The second issue is concerned with the exponential increase in V be with operating time. The mechanism for the increase in V be has been clarified based on reactivation of passivated carbon acceptors in the base region during operation. If the device suffers from H + isolation, V be rapidly decreases at the initial stage, then exponentially increases. The first stage of V be variation can be explained by the fact that a high density of hydrogen atoms migrating from the isolation region to the intrinsic base region, passivate the carbon atoms at the initial stage. From these results, one can expect that the use of He + as an implant instead of H + can solve this problem.

There are no comments yet on this publication. Be the first to share your thoughts.


Seen <100 times