Browse Prior Art Database

Effective Model Parameter Override for Circuit Yield Prediction

IP.com Disclosure Number: IPCOM000246576D
Publication Date: 2016-Jun-20
Document File: 4 page(s) / 133K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a method for the efficient extraction of model override methodology to pair real-life manufacturing device data with the most effective override model.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 4

Effective Model Parameter Override for Circuit Yield Prediction

Device models are created for use within circuit analysis/simulation to verify the functionality and reliability. Device models are partly standardized within the industry,

which lends to a set of parameters defining device functioning and operation. Early device technology development uses a large number of measurements, if available, to extract device model parameters. Measured device current/charge data is fit by parameter extraction via Optimization and Genetic Algorithms. For modeling statistical variation of devices, the process needs to model all realistic device behavior, most likely via individual instrumentation data or a dedicated Technology Computer Aided Design (TCAD) model. Current parameter extraction solves the overall model parameters (BSIM, BSIMCMG etc.) in total.

Model parameters may need to be systematically overridden by a few key parameters and updated to reflect the real-life measurement device data.

The novel contribution is a method for the efficient extraction of model override methodology to pair real-life manufacturing device data with the most effective override model. This includes:


• Generating TCAD device function data to reflect atomistic variabilities and randomness of device structures

• Generating model override results
• Matching/finding the best model override set based on a queried TCAD or real-life device dataset

The novel method embodies the collection of TCAD based variation data for device metrics, building a black box model for key parameters to be chosen for the model overrides, and fitting the right key parameters that resemble the dedicated device data

with minimum error. The method uses a software-based model driven to create device variational models in terms of device model overrides. It naturally explores model parameter correlations and dependencies.

The components and process for implementing the effective model parameter override for circuit yield prediction follow:

1. Collect TCAD based variation data for device metrics. Device measurement is critical for studying and evaluating performance. Real trial and error (i.e., test chip + hardware measurements) is costly. For varia...