Abstract
Advancing nanometer technology scaling enables higher integration on a single chip with minimal feature size. As a consequence, the effects of signal and power integrity issues such as crosstalk noise between interconnects, power supply noise and ground bounce in the supply networks significantly increases. Also, reliability issues are eventually introduced by variations in the manufacturing process. These issues will negatively impact the timing characteristics in an integrated circuit (IC), as they give rise to delay defects. Delay-related parametric failures increase the defect escape rate, yield loss and diminish reliability rate. Hence, design-for-test techniques are employed to have a better controllability and observability on the internal nodes to easily detect and locate the faults. However, they are not always detected by the traditional fault models. In our work, we target these challenges and propose novel physical design-aware path delay test methods to deal with delay faults coming from manufacturing defects or physical design issues. They include the investigation of path delay variations in the presence of crosstalk noise, power supply noise, ground bounce and process variations. Based on this, we develop technology independent test methods for identifying the test patterns that may cause a worst-case delay on a target path. Then, we develop a dedicated test pattern generation method for path delay testing in the presence of crosstalk noise, power supply noise and ground noise. The proposed methods can be used to characterize the path speed and it helps to address the speed binning problem. Also, they can be employed in improving the classical ATPG approach of pattern generation. The application of these contributions can bring tremendous improvements to the IC test quality by ensuring better defect coverage and for an increased manufacturing yield during speed binning of IC chips.