研究目的
To efficiently evaluate the reliability of an RF semiconductor device when several different mechanisms contribute simultaneously to its wearout, focusing on GaN HEMT devices.
研究成果
The developed process for quantifying individual degradation mechanisms in GaN HEMTs and determining mean times to failure in RF applications is economical, quick, and allows clearer identification of signature parameters with the mechanisms. It provides a viable method for high-reliability parts qualification with an order of magnitude margin for error in the MTTF's.
研究不足
The technique requires that each signature parameter measures the extent of one mechanism and is not affected significantly by others, and that the mechanisms remain the same under DC and RF stress. The amount of degradation must be small, and RF operation must have the same RF gain compression as in the application.
1：Experimental Design and Method Selection:
The technique involves finding DC parameters that are 'signatures' of each degradation mechanism, performing separate DC-stress lifetests to find the degradation rates for the signature parameters at several temperatures, and obtaining the corresponding Arrhenius curves. An RF-stress lifetest is then performed with only one stress condition, monitoring all signature parameters and RF performance to determine scaling factors between DC and RF degradation rates.
2：Sample Selection and Data Sources:
Data shown is from an advanced technology with MBE GaN HEMT layers grown on (0001)-oriented Si-face SiC substrates. Discrete transistors were used for DC lifetests, and 1-stage 62 GHz MMIC amplifiers containing the same transistor were used for the RF lifetests.
3：List of Experimental Equipment and Materials:
Parts were kept dark in closed fixtures, with flowing dry nitrogen. DC and RF characterizations were conducted with slow (ms) and fast-pulsed (μs) testing systems.
4：Experimental Procedures and Operational Workflow:
DC and RF lifetests were stopped at appropriate intervals for characterizations. DC characterizations were conducted with slow pulsing, and RF characterizations consisted of measuring Pout versus Pin at 62 GHz.
5：Data Analysis Methods:
Changes in signature parameters were plotted versus time, fitted to linear forms for analysis, and Arrhenius plots were constructed to find thermal activation energies.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
