Plasma Processing Predictive Control for Faster Parameter Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current semiconductor processing devices require numerous control parameters to achieve high accuracy, making it difficult to find optimal processing conditions efficiently, leading to prolonged development times and increased costs, as existing methods rely on extensive experimental data collection and mass learning data, which is time-consuming and labor-intensive.
Innovation Solution
A search device with a processor, memory, and search program that includes a model learning unit, target setting unit, processing condition search unit, and convergence decision unit, which learns a prediction model from initial data, sets target output parameter values, estimates input parameter values, and updates the model to converge to goal output parameter values, allowing for efficient optimization of processing conditions with a minimal number of data points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If extensive experimental data collection and mass learning data are used to find optimal processing conditions, then the accuracy of the prediction model is improved, but the time and labor required for data preparation increases significantly
Solution Approach 1:
The patent applies preliminary action by pre-processing and selecting only the most relevant parameters from the extensive experimental data before feeding them to the prediction model. This reduces the data preparation time while maintaining model accuracy by focusing on critical parameters that have the most significant impact on processing outcomes.
Solution Approach 2:
The patent extracts and utilizes only the necessary portions of the extensive experimental data required for model training. By selectively extracting relevant data points and parameters rather than using all available data, the system reduces data preparation time while preserving the accuracy needed for optimal processing condition determination.
2Manufacturing precision
If numerous control parameters are set to achieve high accuracy in semiconductor processing, then the processing precision is improved, but the complexity of finding optimal parameter combinations increases
Solution Approach 1:
The patent transforms the complex multi-parameter optimization problem into a more manageable form by changing the approach from direct parameter tuning to using a prediction model that automatically determines optimal parameter combinations. The system learns from historical data and predicts optimal settings, reducing the complexity of manual parameter optimization while maintaining high processing accuracy.
Solution Approach 2:
The patent introduces a prediction model as an intermediary between the numerous control parameters and the processing outcomes. This intermediary system processes the complex relationships between parameters automatically, eliminating the need for manual exploration of parameter combinations and simplifying the optimization process while preserving manufacturing precision.
3Adaptability or versatility
If comprehensive parameter testing is conducted to determine initial optimal conditions, then the completeness of parameter optimization is improved, but the development period is prolonged
Solution Approach 1:
The patent implements feedback mechanisms where the prediction model continuously learns from actual processing results and refines its predictions. This allows the system to achieve comprehensive parameter optimization more quickly by iteratively improving based on feedback, rather than requiring exhaustive initial testing of all parameter combinations.
Solution Approach 2:
The patent applies preliminary action by pre-training the prediction model with available historical data before actual device development begins. This preliminary preparation allows the system to start with informed predictions rather than starting from scratch, significantly reducing the development period while maintaining comprehensive parameter optimization through subsequent iterative refinement.
Data Source
AI summary
A model learning unit learns a prediction model on the basis of learning data, a target setting unit sets a target output parameter value by interpolating between a goal output parameter value and an output parameter value which is the closest to the goal output parameter value in output parameter values in the learning data, a processing condition search unit estimates input parameter values which corresponds to the goal output parameter value and the target output parameter value, a model learning unit updates the prediction model by using a set of the estimated input parameter value and an output parameter value which is a result of processing that a processing device performs as additional learning data.