Defocused Aerial Image Convolution for Resist Profile Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional photolithography process models inaccurately represent the resist profile in the Z direction due to ignored physical effects, such as acid diffusion, which affects subsequent processes like etch, especially at deep submicron semiconductor manufacturing nodes.
Innovation Solution
A process model is developed to account for the effects of Chemically Amplified Resist Post Exposure Bake (CAR/PEB) by modeling defocused aerial images using Gaussian kernels to differentiate diffusion lengths between the Z direction and X/Y directions, ensuring accurate representation of resist profiles at varying depths.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional photolithography process models ignore physical effects like acid diffusion to improve simulation speed, then computational performance is improved, but manufacturing precision deteriorates due to inaccurate resist profile representation
Solution Approach 1:
The patent applies parameter changes by introducing a defocus parameter (delta z) to model the aerial image at different depths within the resist. This allows the model to capture the vertical variation in acid concentration without requiring full 3D diffusion simulation, thus maintaining computational efficiency while improving resist profile accuracy in the Z direction
Solution Approach 2:
The patent uses the aerial image intensity distribution as an intermediary to represent the acid diffusion effect. Instead of directly simulating acid diffusion, the model uses the optical aerial image at defocused distances to indirectly capture the diffusion phenomenon, simplifying the physics while maintaining accuracy
2Manufacturing precision
If photolithography process models include full 3D acid diffusion simulation to improve manufacturing precision, then resist profile accuracy is improved, but device complexity increases due to additional computational requirements
Solution Approach 1:
The patent extracts only the essential Z-direction diffusion effect from the full 3D diffusion problem. By separating the vertical diffusion component from the lateral diffusion and representing it through defocused aerial image convolution, the model reduces complexity while retaining the critical physics needed for accurate resist profile prediction
Solution Approach 2:
The patent creates a simplified copy of the diffusion process by using aerial image convolution at different defocus distances to represent acid diffusion. This copy captures the essential diffusion behavior without requiring the complex partial differential equation solving of full 3D diffusion, thus reducing model complexity
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 enhances the accuracy of photolithography process models without sacrificing computational performance, improving the prediction of resist profiles and subsequent process behaviors, thereby reducing errors in semiconductor manufacturing.
Implementation Method 1
diffuses the aerial image using a Gaussian kernel to account for the diffusion length difference between the Z direction and the XlY directions
Data Source
Figure 1
Figure 2~3
Figure 4~5C
AI summary
An embodiment provides systems and techniques for determining a process model. During operation, the system may receive a first optical model which models a first optical system of a photolithography process. Next, the system may use the first optical model to determine a second optical model that models a second latent image that is formed by the first optical system at a second distance. The system may also use the first optical model to determine a third optical model that models a third latent image that is formed by the first optical system at a third distance. Next, the system may receive process data which is obtained by subjecting a test layout to the photolithography process. The system may then determine a process model using the first optical model, the second optical model, the third optical model, the test layout, and the process data.