恒定温度应力加速实验失效机理一致性快速判别方法

针对加速实验中因失效机理发生改变而导致的无效实验问题, 对失效机理一致性和分别在不同应力下器件的初期退化参数分布的关系进行了研究. 研究证明了判断加速实验失效机理一致的条件: 1) 不同应力下失效分布的形状参数服从一致, 即m_i=m, i=1,2,3,···, 2) 尺度参数 η_i服从η_i= AF_i·η. 从而提供一种在实验初期即可根据参数退化分布快速判别不同应力下失效机理是否一致的方法, 避免了因失效机理发生改变而造成的无效加速实验. 最后对加速实验初期理论退化数据和多芯片组件厚膜电阻初期退化数据进行了威布尔分布参数估计, 并对其在不同应力下的失效机理一致性进行了判别.关键词：失效机理一致性恒定温度应力加速实验威布尔分布

Rapid identification of the consistency of failure mechanism for constant temperature stress accelerated testing

For avoiding the invalid acceleration experiments caused by the changes of the failure mechanism, the relationship of the failure mechanism consistency and the parameters of degradation data distribution in the early stage under different accelerated stress levels has been derived. Conditions for judging the failure mechanism consistency are also given as follows: firstly, the shape parameters of failure distribution has a uniform distribution m_i=m, i=1,2,3,···; secondly, the dimension parameter η_i follows the equation η_i=AF_i·η. A method to rapidly discriminate the consistency of failure mechanism under different experimental stresses in the early stage was obtained, and the invalid acceleration experiments caused by the changes of the failure mechanism could be avoided. Finally, theoretical degenerate data in the early stage of the accelerated test and the initial degenerate data of the MCM thick-film resistor were used for estimating, Weibull distribution parameter, and the consistency of failure mechanism degradation was also judged.