Predicting Vt mean and variance from parallel Id measurement with model-fitting technique

Abstract

To measure the variation of device Vt requires long test for conventional WAT test structures. This paper presents a framework that can efficiently and effectively obtain the mean and variance of Vt for a large number of DUTs. The proposed framework applies the model-based random forest as its core model-fitting technique to learn a model that can predict the mean and variance of Vt based on only the combined I d measured from parallel connected DUTs. The experimental results based on the SPICE simulation of a UMC 28nm technology demonstrate that the proposed modelfitting framework can achieve a more than 99% R-squared for predicting both of Vt mean and variance. Compared to conventional WAT test structures using binary search, our proposed framework can achieve 42.9X speedup in turn of the required iterations of I d measurement per DUT.

Retraction:Reliability assurance in early-life-failure test through improved nearest neighbor regression

“…This shift undermines the effectiveness of a simple stuck-at based test solution. Data-based classification algorithms have improved continuously with the change of defects, minimize yield loss [11,12,13,14,15,16,17], ML can be used to distinguish between marginal defects and process variation defects based on circuit delay, depend on different delay distribution. The classification results can be able to locate the defects and identify the root cause.…”

“…In [12], critical parameters were focused on pass/fail characteristics of past test data, but did not detail correlation of performances and characteristics measured accuracy. In [14], hidden characteristics of test data that make a correlation between dies and test items were investigated, but did not consider data overfitting. In [16], a wafer test flow was optimized with a graphical model and in [17], defect characteristics by wafer mapping were D e l e t e d investigated, but did not consider the marginal defects.…”

“…As we can see, most of previous works are based on conventional fault models [11,12,13,14,15,16,17,18], lack of research on marginal defects and process variation defects with machine learning classification methods. To bridge this gap and formally show that, in a quest to reduce ELF and increase reliability, the work presented in this paper combines the concepts of regression and KNN algorithm with data preprocessing method.…”