Semiconductor Process Simulation for Intermediate Profile Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current semiconductor process simulation technologies are limited in generating profiles for intermediate process steps, making it difficult to trace faults and optimize the manufacturing process effectively due to cost and technological constraints.
Innovation Solution
A simulation method using a recurrent neural network (RNN) with process emulation cells, which trains and predicts profiles for each process step based on a final target profile, prior knowledge, and process condition information, allowing for accurate estimation of intermediate step profiles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional process simulation is used to obtain a final profile, then the final profile can be obtained, but intermediate step profiles cannot be obtained
Solution Approach 1:
The patent segments the manufacturing process into multiple discrete process steps, with each step represented by a separate process emulation cell in the neural network. This segmentation allows the system to estimate profiles for each intermediate step individually rather than only providing the final profile, thereby recovering the lost intermediate information while maintaining accuracy.
Solution Approach 2:
The patent uses prior knowledge information about time series causal relationships between process steps to constrain and guide the neural network training beforehand. This preliminary action of incorporating domain knowledge ensures that the estimated intermediate profiles are physically meaningful and consistent with manufacturing principles, improving estimation accuracy.
2Manufacturing precision
If multiple profiles for multiple process steps are checked to tune a specific process, then process tuning can be performed, but cost and technology limitations prevent this
Solution Approach 1:
The patent creates a virtual copy of the manufacturing process through process emulation cells that simulate each process step. Instead of requiring multiple physical experiments or complex measurement systems to obtain profiles for tuning, the system generates accurate profile estimates through neural network simulation, significantly reducing the complexity and cost of process tuning while maintaining high manufacturing precision.
3Productivity
If intermediate step profiles are not obtained, then simulation is simpler, but fault tracing and problematic step identification become difficult
Solution Approach 1:
By segmenting the process into discrete steps with individual emulation cells, the system enables targeted fault tracing. When a defect is detected in the final product, the segmented structure allows rapid identification of which specific process step caused the issue by comparing intermediate profiles, thereby improving defect inspection efficiency without requiring excessive system complexity.
Data Source
AI summary
Provided is a simulation method performed by a process simulator, implemented with a recurrent neural network (RNN) including a plurality of process emulation cells, which are arranged in time series and configured to train and predict, based on a final target profile, a profile of each process step included in a semiconductor manufacturing process. The simulation method includes: receiving, at a first process emulation cell, a previous output profile provided at a previous process step, a target profile and process condition information of a current process step; and generating, at the first process emulation cell, a current output profile corresponding to the current process step, based on the target profile, the process condition information, and prior knowledge information, the prior knowledge information defining a time series causal relationship between the previous process step and the current process step.