Life time prediction for analog circuit components utilizing digital degradation monitors

如今，在安全关键应用中使用集成电路变得越来越普遍。这就要求在整个生命周期内满足这些电路的规格要求。然而，集成电路的特性会受到时间变化的影响。半导体结构中的物理效应是造成这些影响的原因。材料性能随时间变化的变化称为老化或降解。这些与时间相关的机制导致集成电路的特性发生变化，从而可能导致电子系统故障或缩短其使用寿命。尽管在可靠性范围内集成电路的设计中出现了一些创新的解决方案，但电路行为的改变是不可避免的。本研究计划的主要目标是测量模拟集成电路退化条件的新方法。在了解退化条件的基础上，将发展出预测寿命的理论基础。为了实现这两个目标，测量退化和寿命预测，将开发一种基于自适应滤波器系数的确定标称电路性能的形式化方法。除了这些标称性能外，还将评估最小和最大规格边界。此外，还将确定这些性能老化引起的时间依赖性变化。通过观察自适应滤波器的系数随时间的变化，可以确定任何电路的实际状态。对这些系数的连续观察可以用来推断系统在运行过程中发生的故障。为了在此基础上建立对微电子电路退化的估计，必须以一种随后可区分这些影响的方式来测量对电路的寄生影响。因此，在项目中建立了多个传感器接口来确定模具pvt的影响。有了标称电路性能、系统实际状态和pvt影响的知识，就可以开发出确定电路退化的方法。由此导出的退化状态为集成电路寿命预测的增强方法的发展提供了基础。在这个项目中，这些方法将以第一理论方法为例来实现。

Nowadays it becomes more common to use integrated circuits within safety critical applications. This necessitates that specifications of these circuits are satisfied over the whole lifetime. However, the characteristic properties of an integrated circuit are affected by temporal changes. Physical effects in the semiconductor structures are responsible for these influences. Time dependent changes in material properties are called aging or degradation. These time dependent mechanisms cause the change in characteristic properties of an integrated circuit which may cause a malfunction of an electronic system or shorten its lifetime. In spite of several innovative solutions for the design of integrated circuits with a scope on reliability a shift in the behavior of the circuit cannot be foreclosed. The main objectives of this research proposal are new methods for measuring the conditions of degradation of analog integrated circuits. Based on the knowledge of the condition of degradation theoretical rudiments for the prediction of lifetime will be evolved. In order to achieve these two goals, measuring the degradation and lifetime prediction, a formal method based on the coefficients of adaptive filters for the determination of nominal circuit performances will be developed. Besides these nominal performances the minimal and maximal specification boundaries will be evaluated. Moreover, the time dependent change caused by aging of these performances will be determined. By observing the coefficients of an adaptive filter over time the actual state of any circuit can be determined. The continuous observation of these coefficients can be utilized to draw inferences about system failures occurring during operation. In order to build an estimation of the degradation of microelectronic circuits on this basis the parasitic influences on the circuit have to be measured in a way that these effects are subsequently distinguishable. Therefore, multi sensor interfaces are built in the project to determine die PVTinfluences. With the knowledge of the nominal circuit performances, the actual state of the system, and the PVT-influences methods to determine the degradation of a circuit can be developed. The therewith derived degradation state represents the origin for the development of enhanced methods for the lifetime prediction of integrated circuits. In this project these methods will be exemplarily realized involving first theoretical approaches.