Bi-exponential Statistical Leakage Modeling for IC Yield
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As process node sizes shrink, process variations in integrated circuit manufacturing lead to significant leakage power issues, causing reduced manufacturing yield and increased failures due to excessive leakage currents, which existing technologies struggle to accurately analyze and manage effectively.
Innovation Solution
The implementation of a bi-exponential modeling approach for statistical leakage power analysis, which accounts for sensitivity to process variations and combines cell leakage sensitivities, addressing non-linearity and resource efficiency by naturally grouping similar leakage functions without external criteria, and treating random variations in a compact-size file while maintaining correlations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of moving object
If process node sizes are shrunk to improve integration density, then manufacturing precision deteriorates due to process variations causing leakage power issues
Solution Approach 1:
The patent applies preliminary action by performing statistical leakage analysis and bi-exponential modeling during the design phase to predict and characterize leakage behavior before manufacturing. This allows designers to identify and mitigate leakage issues proactively, preventing yield loss rather than detecting it after fabrication.
Solution Approach 2:
The patent replaces traditional deterministic leakage analysis with a statistical modeling approach using bi-exponential distributions. This substitution enables the system to capture the probabilistic nature of leakage variations at scaled node sizes, providing more accurate predictions of leakage behavior under process variations.
2Measurement precision
If statistical leakage analysis is performed to improve leakage power estimation accuracy, then computational complexity increases
Solution Approach 1:
The patent segments the leakage analysis into distinct components: bi-exponential modeling for capturing leakage distributions, statistical characterization of process variations, and hierarchical analysis methods. This segmentation allows complex leakage behavior to be broken down into manageable analytical components that can be computed efficiently.
Solution Approach 2:
The patent transforms the leakage analysis problem by changing parameters from deterministic values to statistical distributions. The bi-exponential model uses parameters such as scale factors and shape parameters that can be extracted from process data, enabling accurate leakage prediction without requiring exhaustive simulation of every possible process variation.
Data Source
AI summary
A method, system, and computer program product are disclosed for performing statistical leakage power characterization to estimate yield of a circuit in terms of leakage power. According to some approaches, this is performed with consideration of bi-exponential modeling.