Efficiency and applications of SAT-based test pattern generation: complex fault models and optimisation problems

Modern technologies have enabled the semiconductor industry to enter a new era of integrated-circuit manufacturing. Modern ICs are not only smaller and signi cantly more high-performing than they used to be only a few years ago; they are also considerably more energy-e cient thanks to the use of new materials with convenient electric properties. However, the use of new materials is also making the fabrication process more di cult to control, and the new chips are more prone to defects. In consequence, the role of fault models that allow to describe complex forms of faulty behaviour is becoming increasingly important in hardware test and diagnosis. Without doubt, automatic test pattern generation (ATPG) is the most important test task. ATPG algorithms need to be not only run-time-e cient and to produce compact test sets, given the large number of faults that need to be targeted in multibillion-transistor ICs; they also need to keep pace with the development of new mechanisms for the description of faulty behaviour. Traditionally, ATPG algorithms used in industrial applications are structural, i.e. their reasoning is based on the circuit’s structure. However, SAT-based algorithms, i.e. methods that map the ATPG problem to the problem of Boolean satis ability (SAT), have recently started to gain relevance because they perform better than structural methods on important classes of faults. ‘is doctoral thesis covers the work on SAT-based test pattern generation performed by the thesis’s author between 2008 and 2012. It presents the SAT-based ATPG tool Tiguan and explains in detail all important aspects that were considered in order to make Tiguan a highly e cient test pattern generator capable of calculating provably optimal solutions for complex ATPG problems. ‘e most important contributions of the work presented in this thesis can be summarised as follows: ▸ ‘e general run-time e ciency of SAT-based ATPG was increased through intelligent mapping of the ATPG problem to SAT, through the optimal utilisa-