Ferroelectric Capacitor Model for Local Statistical Variability

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional models for simulating ferroelectric capacitors in integrated circuits fail to accurately account for local variability in polarization characteristics, leading to inaccuracies in circuit simulations and potential errors in integrated circuit operation, necessitating costly and time-consuming fabrication and testing to validate design.

Innovation Solution

A method is developed to model and simulate ferroelectric capacitors by deriving probability density functions for polarization characteristics based on actual device measurements, incorporating local variability through a multi-domain model that accounts for random domain fluctuations, allowing for accurate calibration and simulation of circuit behavior.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional models for simulating ferroelectric capacitors are used, then the simulation process is simple and fast, but the accuracy of polarization characteristics is insufficient

Engineering Contradiction:
Improveaccuracy of polarization characteristicsVSAvoidcomplexity of simulation model
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The ferroelectric capacitor is divided into multiple independent domains, each with its own polarization characteristics. This segmentation allows the model to capture local variability in polarization behavior while maintaining computational efficiency through modular structure.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Each domain in the multi-domain model is assigned local polarization characteristics that vary according to position and material properties. This enables accurate representation of local variability in coercive voltage and remanent polarization across different regions of the ferroelectric capacitor.

Inventive Principle:
Principle #3Local quality

2Reliability

If conventional models are used, then the simulation is computationally efficient, but local variability in polarization characteristics cannot be captured

Engineering Contradiction:
Improverobustness of circuit operationVSAvoidtime for fabrication and testing
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The multi-domain model with local variability is implemented in the simulation stage before fabrication and testing. This preliminary action allows design validation and optimization to be performed computationally, reducing the need for iterative physical prototyping and testing.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The physical ferroelectric capacitor is replicated in the simulation through a detailed multi-domain model that copies its local variability characteristics. This virtual copy allows comprehensive testing and validation without requiring physical devices, saving time and resources.

Inventive Principle:
Principle #26Copying

3Measurement precision

If a multi-domain model with local variability is implemented, then accuracy of polarization characteristics improves, but computational complexity increases

Engineering Contradiction:
Improveaccuracy of local variabilityVSAvoidcomplexity of multi-domain model
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The model uses parameterized representations of domain characteristics, such as distributed coercive voltages and remanent polarizations, that can be efficiently computed. By changing the mathematical parameters and using analytical solutions where possible, the computational complexity is managed while maintaining high accuracy.

Inventive Principle:
Principle #35Parameter changes

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enables precise modeling and simulation of ferroelectric capacitors, improving the robustness and accuracy of integrated circuit operation by accurately capturing local variability, reducing the need for extensive testing and optimizing circuit design efficiency.

Implementation Method 1

Hysteresis in the charge-vs.-voltage (Q-V) characteristic, based on the polarization state of the ferroelectric material, enables the non-volatile storage of information in the FRAM cell

Methodology Applied
Scientific EffectHysteresis: Hysteresis

Implementation Method 2

If the voltage V applied across the capacitor plates exceeds a 'coercive' voltage Vα, the capacitor is polarized into the '+1' state

Methodology Applied
Scientific EffectPolarization: Polarisation

Data Source

PatentUS8380476B2Modeling of ferroelectric capacitors to include local statistical variations of ferroelectric properties
Publication Date: 2013.02.19 TEXAS INSTRUMENTS INC
  • US8380476B2 patent drawing
  • US8380476B2 patent drawing
  • US8380476B2 patent drawing

AI summary

Simulation of an electronic circuit including a model of a ferroelectric capacitor. The model of the ferroelectric capacitor includes a multi-domain ferroelectric capacitor, in which each of the domains is associated with a positive and a negative coercive voltage. A probability distribution function of positive and negative coercive voltages is defined, from which a weighting function of the distribution of domains having those coercive voltages is defined. To create a model of a small ferroelectric capacitor, a Poisson probability distribution is assigned to each of an array of gridcells defining the probability distribution function of positive and negative coercive voltages, and a number of domains assigned to each gridcell is randomly selected according to that Poisson distribution and an expected number of domains in the modeled capacitor for that gridcell, based on the area of the modeled capacitor. The electrical behavior of the ferroelectric capacitor is evaluated by evaluating the superposed polarization of each of the randomly selected domains.