Automated Data Assessment for Semiconductor Process Parameter Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional statistical methods are insufficient for evaluating the impact of experimental splits and cross-dependencies on a large number of electrical parameters in semiconductor manufacturing, leading to inefficiencies and errors in data analysis during device and process development.
Innovation Solution
An automated data assessment method that identifies and ranks parameters affected by experimental splits and cross-splits, using pattern recognition and fuzzy logic to determine significance and reduce data sets, thereby focusing analysis on critical parameters and saving time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional statistical methods are used to evaluate the impact of experimental splits on electrical parameters, then the evaluation can be performed with simple methods, but the analysis is insufficient and prone to errors when dealing with large numbers of parameters
Solution Approach 1:
The patent segments the complex data analysis task into distinct phases: automated data collection from process steps, pattern recognition to identify significant parameters, and focused manual inspection of only those parameters. This segmentation transforms an overwhelming comprehensive analysis into manageable discrete tasks, resolving the contradiction between evaluation accuracy and analysis complexity.
Solution Approach 2:
The patent introduces automated pattern recognition algorithms as an intermediary between raw electrical parameter data and human analysis. This intermediary system pre-processes and filters thousands of parameters, identifying only those with significant split impacts for manual review, thereby enabling accurate evaluation without requiring direct human analysis of all parameters.
2Reliability
If comprehensive testing and evaluation is performed on all parameters, then complete data assessment is achieved, but the time required for data analysis increases significantly
Solution Approach 1:
The patent performs preliminary automated analysis and pattern recognition before manual inspection. By pre-identifying significant parameters through automated algorithms and pre-filtering the data set, the system prepares the analysis in advance, ensuring comprehensive assessment of critical parameters while minimizing the time required for subsequent manual review.
Solution Approach 2:
The patent extracts and isolates only the significant parameters that are truly affected by experimental splits, separating them from the large body of insignificant data. This extraction focuses analytical efforts on the critical subset of parameters, maintaining assessment completeness for relevant variables while dramatically reducing overall analysis time.
3Measurement precision
If manual inspection is performed on all electrical parameters, then detailed analysis is possible, but the cost and time requirements become prohibitive
Solution Approach 1:
The patent extracts only the significant parameters that require detailed manual inspection, separating them from the large set of parameters that can be adequately assessed through automated analysis. This selective extraction maintains high measurement precision for critical parameters while improving overall productivity by limiting manual inspection to a small subset.
Solution Approach 2:
The patent applies different quality levels of analysis to different parameters: automated pattern recognition for the majority of parameters and detailed manual inspection only for significant parameters identified by the automated system. This local differentiation of analysis quality optimizes the balance between measurement precision and productivity by applying intensive resources only where necessary.
Data Source
AI summary
A method of adjusting process variables in a processing flow is disclosed. Processed samples are tested to determine sample parameters of the tested samples. The sample parameters are analyzed analyzing in relation to the process variables applied in the processing steps to determine the impact of the process variables on the sample parameters The process variables are modified in an attempt to change the sample parameters towards predetermined target values. And, the sequence of processing steps is repeated with the modified process variables. The analyzing step includes, for given samples, automated matching between patterns of process variables applied for the samples and corresponding sample data sets of parameters determined from the samples; quantifying the degree of match in terms of score values associated with patterns of process variables; and determining the significance of said score values in terms of significance values based on the deviation of the parameters in the sample data sets from said predetermined target values.