Learning Model for Uniform Film Thickness in Sublimation Drying
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to determine optimal parameters for uniform solid film thickness during sublimation drying on substrates due to the complexity and time-consuming nature of simulations, leading to potential pattern collapse from non-uniform film thickness.
Innovation Solution
A learning device and method that utilizes a hardware processor to acquire film thickness data from substrate processing, generate a learning model, and infer film-thickness characteristics to optimize parameters for uniform film thickness, enabling real-time recipe determination for substrate processing devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If simulation is used to determine optimal parameters for uniform solid film thickness, then manufacturing precision is improved, but loss of time increases significantly
Solution Approach 1:
The patent creates a virtual copy of the substrate processing system through a learning model that replicates the behavior of the actual system. This virtual model is trained using simulation data and then used to predict film thickness outcomes without requiring time-consuming real-time simulations, thus resolving the contradiction between precision and time loss
Solution Approach 2:
The learning model is trained in advance using simulation data to establish predictive relationships between processing parameters and film thickness outcomes. This preliminary training allows the model to provide rapid predictions during actual substrate processing without requiring time-consuming simulations at runtime, addressing the time precision contradiction
2Manufacturing precision
If the number of training data pieces is increased for better model accuracy, then manufacturing precision is improved, but loss of time increases due to data preparation requirements
Solution Approach 1:
The patent uses a virtual copy approach where simulation data is used to train the learning model instead of requiring extensive experimental data collection. This virtual data generation through simulation allows for sufficient training data without the time-consuming process of conducting numerous physical experiments, thus resolving the contradiction between accuracy and data preparation time
3Reliability
If sublimation drying is used to prevent pattern collapse, then reliability is improved, but manufacturing precision deteriorates due to non-uniform solidified film thickness
Solution Approach 1:
The learning model provides predictive feedback about film thickness outcomes based on processing parameters before actual substrate processing occurs. This allows operators to adjust parameters to achieve both the rapid drying needed for pattern stability and the uniform film thickness required for manufacturing precision, resolving the contradiction between reliability and precision
Solution Approach 2:
The patent systematically varies and optimizes processing parameters (such as heating temperature, gas flow rate, and substrate rotation speed) to achieve the dual goals of rapid water removal for pattern stability and uniform solidified film formation. The learning model identifies optimal parameter combinations that balance these competing requirements
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
Facilitates the generation of a learning model that infers film-thickness characteristics, allowing for efficient and timely optimization of substrate processing conditions to prevent pattern collapse and ensure uniform film thickness.
Implementation Method 1
there is a technique so-called sublimation drying in which the liquid applied to the surface on which a pattern is formed on the substrate is converted into a solid and dried by a phase change from the solid to a gas
Implementation Method 2
causes a learning model to learn first training data including a film-thickness characteristic, which indicates a characteristic of the film thickness
Data Source
AI summary
A learning device includes a hardware processor, wherein the hardware processor acquires a film thickness of a solid or a liquid formed on the substrate by driving a substrate processing device under a first condition, and causes a learning model to learn first training data including the film-thickness characteristic acquired by driving of the substrate processing device under the first condition.