Scatterometry-based critical dimension (CD), also known as optical CD (OCD), significantly matches CD scanning electron microscopy in accuracy and precision, in addition to offering superior full-profile reconstruction. OCD, however, is computationally intensive. We construct an extremely fast screening tool that determines whether a sample should or should not proceed to subsequent manufacturing steps. To this end, we examine the diffraction signals of the grating in order to determine whether a sample is in or out of its specification limits. This allows us to allocate traditional metrology resources only for samples that show unusual behavior. Support vector machines (SVMs) are trained to classify each incoming sample as in-spec or out-of-spec. The constructed classifier is applied to gratings exposed with a focus-exposure matrix for a rectangular silicon-bottom anti-reflective coating-photoresist stack, which include erroneous samples with under–over exposure, necking, and bridging problems. The misclassification rates as well as false and missed alarm rates are analyzed. Results show that our prototype screening system has misclassification errors on the order of 5% to 10%, while the computation time is on the order of one vector dot product.
Scatterometry-based CD, also known as Optical CD (OCD) significantly matches CD-SEM in accuracy and precision, in addition to offering superior full-profile reconstruction. OCD, however, is computationally intensive. In this paper, we construct an extremely fast screening tool that determines whether a sample should or should not proceed to subsequent manufacturing steps. To this end we examine the diffraction signals of the grating in order to determine whether a sample is in or out of its specification limits. This allows us to allocate traditional metrology resources only on samples that show unusual behavior. Support vector machines (SVM) are trained to classify each incoming sample as in-spec or outof- spec. The constructed classifier is applied to gratings exposed with a focus-exposure matrix for a rectangular silicon- BARC-photoresist stack, which include erroneous samples with under-over exposure, necking, and bridging problems. The misclassification rates as well as false and missed alarm rates are analyzed. Results show that our prototype screening system has misclassification errors on the order of 5-10 %, while the computation time is on the order of one vector dot product.
Recently, there has been significant interest in so-called Hybrid or Holistic Metrology, the practice of combining measurements from multiple sources in order to improve the estimation of one or more critical parameters. There also has been significant research in capturing and modeling the hierarchical spatial variability of CDs at the die, wafer, and lot level. However, the information inherent in spatial variability models has not been used towards improving the accuracy/precision of CD estimates. In this paper, we review the current trends in Hybrid Metrology and Spatial Variability Modeling, and provide a simple example based on the work of Zhang et al.1 that illustrates how we can incorporate spatial information for improved measurement estimates.
There are two competing costs that occur in off line semiconductor processing metrology. One is the cost of
operating the metrology tool, and the other is the loss in terms of processing cost and yield due to the time
lapse between the occurrence and the correction of a process fault. Virtual metrology (VM) is an alternative
scheme which takes data produced by the processing tool in real time (e.g. plasma etching data during isolation
trench formation) and predicts an outcome of the wafer (e.g. critical dimension of the trench) utilizing an
empirical model. Although VM prediction quality is not as good as that of conventional metrology, it produces
an immediate, low cost prediction for each wafer going through a process. In real life, we envision that practical
metrology schemes will involve a synergistic blend of VM and actual metrology, the latter being used for the
needed periodic recalibration of the VM empirical model. In this work, we formulate the costs associated with
Type I and Type II errors that result from a blended metrology scheme, and propose a general framework that
can be used to quickly lead to the optimal design of such schemes given the characteristics of the process in
question. We also explore the effects of a faulty process (by means of mean shift) on the cost analysis.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.