Techniques to Address Manufacturing Defects in Deep Sub-micron

Author

Puneet Savanur, Southern Illinois University CarbondaleFollow

Date of Award

5-1-2022

Degree Name

Doctor of Philosophy

Department

Electrical and Computer Engineering

First Advisor

Tragoudas, Spyros

Abstract

Shrinking transistor sizes has resulted in increased manufacturing defects. Therefore, an efficient and accurate method is essential to test and diagnose such defects in an Integrated Circuit (IC). It also helps with improving the IC design cycle for future products. However, various challenges are faced while developing a universal testing method for an IC.Chapter 2 addresses the challenge of process variation, which causes no two ICs to be the same. Process variation affects the transistor's physical characteristics, and it is challenging to characterize the billions of transistors in an IC. A novel approach is presented to derive the oxide thickness and threshold voltage of a P-channel metal-oxide-semiconductor (pMOS) and N-channel metal-oxide-semiconductor (nMOS) transistor. Experimental results show the accuracy of identifying the oxide thickness values and the threshold voltage of an embedded transistor.Chapter 3 focuses on physical defects that occur during manufacturing. The progress made in decreasing transistor sizes has increased manufacturing defects. Many types of defects can occur, and there is a need for a universal method to generate tests for all such defects. Chapter 3 presents a new model called the Alternate Gate Fault (AGF) model to detect multiple types of manufacturing defects that change a gate's functionality in a combination circuit. Various Automatic Test Pattern Generation (ATPG) methods that use the Modified Condition Decision Coverage (MCDC) principles to generate tests to detect AGFs are also presented. Experimental results demonstrate the efficiency of the proposed ATPG methods over existing fault models in detecting AGFs in a combinational circuit.Chapter 4 focuses on diagnosing manufacturing defects using the MCDC test. It studies the diagnostic capabilities of the MCDC test vectors generated using the various ATPG techniques presented in Chapter 3. A statistical effect-cause diagnostic approach is used to evaluate the diagnostic resolution of the MCDC test sets. Experimental results compare various test sets generated using different test pattern generation techniques from Chapter 3.