Multi-Stage Etching Recipe Control for Hard Mask Stack Variation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The multi-stage etching technology faces challenges in managing numerous variation factors such as material, thickness, gas flow, bias RF power, and E-chuck temperature, making it difficult to achieve consistent etching targets due to varying hard mask layers and stacking relationships, which form a technical development bottleneck.
Innovation Solution
A machine learning model is employed to generate parameter setting recipes for hard mask layers, considering stack information and etching target conditions, enabling accurate control of these factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional monitoring of one or two hard mask layers is used, then the process is simple to operate, but the manufacturing precision of the final etching target deteriorates due to numerous uncontrolled variation factors
Solution Approach 1:
The patent introduces a machine learning model as an intermediary between the multiple hard mask layers and the final etching target. The model receives stack information from multiple layers and outputs optimized process parameters, mediating the complex relationships between numerous variation factors and the etching outcome without requiring direct monitoring of each layer
Solution Approach 2:
The patent changes the approach from monitoring physical layer characteristics to using a machine learning model that processes and transforms stack information into optimized process parameters. This parameter transformation enables precise control of the etching target by leveraging the model's ability to identify optimal parameter combinations from historical data
2Adaptability or versatility
If the same hard mask layers are continuously monitored, then the ease of operation is maintained, but the adaptability to various material changes and stacking relationships deteriorates
Solution Approach 1:
The machine learning model serves as a universal control system that can handle various material changes and stacking relationships. Instead of requiring specific monitoring procedures for each hard mask layer configuration, the model universally processes different stack information inputs and generates appropriate process parameters for diverse etching scenarios
Solution Approach 2:
The patent uses historical process data and stack information as training data to create a virtual model of the etching process. This digital copy or replica of the physical process enables the system to predict optimal parameters for new configurations without requiring physical experimentation or complex operational procedures
3Manufacturing precision
If numerous variation factors are manually set, then the device complexity remains low, but the manufacturing precision of the etching target deteriorates due to difficulty in setting all factors
Solution Approach 1:
The machine learning model enables the system to self-determine optimal process parameters automatically. The model uses its trained knowledge from historical data to autonomously identify and set the numerous variation factors without requiring manual intervention, thereby achieving both high precision and reduced automation burden
Data Source
AI summary
A control method of a multi-stage etching process and a processing device using the same are provided. The control method of the multi-stage etching process includes the following step S. A stack information of a plurality of hard mask layers is set. An etching target condition is set. Through a machine learning model, a parameter setting recipe of the hard mask layers is generated under the etching target condition. The machine learning model is trained based on the stack information of the hard mask layers, a plurality of process parameters and a process result.