Reliability of GaN on Si FETs and MMICs
MRS Online Proceeding Library 12/2008; 1195. DOI: 10.1557/PROC-1195-B06-04
This paper reviews the reliability results for the gallium nitride on silicon (GaN-on-Si) technologies for commercial and military communications markets. Two technology platforms have been qualified for volume production: one consisting of discrete heterostructure field effect transistors (HFETs) and the other consisting of HFETs integrated with passive components to form monolithic microwave integrated circuits (MMICs). The technology platform qualifications for volume production have been achieved through intrinsic reliability tests on the active and passive device elements as well as extrinsic reliability tests at the product level. This paper presents reliability results on accelerated life test (ALT), high temperature operating life under DC and RF stress (DC/RF-HTOL), electrostatic discharge (ESD), ramped voltage breakdown, electromigration, temperature cycling, robustness under voltage standing wave ratio (VSWR) mismatch conditions, and diode stability. Degradation and breakdown mechanisms are discussed in relation to material properties reliability. The results show that the HFET and MMIC technology platforms display reliable performance for 20 year product lifetime at worst case operating conditions.
Conference Paper: Time-Zero Dielectric Reliability Test by a Ramp Method[Show abstract] [Hide abstract]
ABSTRACT: Using a long-established feature of time dependent dielectric breakdown (TDDB) it is demonstrated that a ramped voltage breakdown histogram of a sample population can be used to accurately forecast the rate of breakdown failures in the field. It is shown that such a histogram can be interpreted as the field dependence of failure at constant time. The tamp-TDDB relationship involves no fitting parameters and only a single material-related parameter. The temperature dependence of this parameter is established for SiO2 Extensive ramp-life test measurements have verified the relationship experimentally. It is argued that the usual models used to relate laboratory life tests to reliability failures are inherently faulty. The faults stem from the temperature dependence and the distributions of failure times, both of which must be assumed in order to extrapolate accelerated life tests to use conditions. On the other hand the actual distribution is measured in a ramp test and the temperature acceleration is not needed. This finding has far-reaching implications for reliability assessment. Dielectric life tests can be replaced by the relatively simple and rapid ramp test with increased confidence in projection. From the analysis it is shown that the effect on reliability of a high field screen can be quantitatively determined in an absolute manner.Reliability Physics Symposium, 1981. 19th Annual; 05/1981
- [Show abstract] [Hide abstract]
ABSTRACT: Al<sub>0.26</sub>Ga<sub>0.74</sub>N-GaN heterojunction field-effect transistors were grown by metal-organic chemical vapor deposition on high-resistivity 100-mm Si (111) substrates. Van der Pauw sheet resistance of the two-dimensional electron gas was 300 Ω/square with a standard deviation of 10 Ω/square. Maximum drain current density of ∼1 A/mm was achieved with a three-terminal breakdown voltage of ∼200 V. The cutoff frequency and maximum frequency of oscillation were 18 and 31 GHz, respectively, for 0.7-μm gate-length devices. When biased at 50 V, a 2.14-GHz continuous wave power density of 12 W/mm was achieved with associated large-signal gain of 15.3 dB and a power-added efficiency of 52.7%. This is the highest power density ever reported from a GaN-based device grown on a silicon substrate, and is competitive with the best results obtained from conventional device designs on any substrate.IEEE Electron Device Letters 08/2004; 25(7-25):459 - 461. DOI:10.1109/LED.2004.831190 · 2.75 Impact Factor
Data provided are for informational purposes only. Although carefully collected, accuracy cannot be guaranteed. The impact factor represents a rough estimation of the journal's impact factor and does not reflect the actual current impact factor. Publisher conditions are provided by RoMEO. Differing provisions from the publisher's actual policy or licence agreement may be applicable.