Adaptive Metrology Sampling Rate Control for Virtual Metrology Accuracy

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional metrology sampling methods in semiconductor and TFT-LCD manufacturing are inefficient, leading to high cycle times and production costs due to fixed and experience-based sampling rates, and poor virtual metrology prediction accuracy when production variations occur.

Innovation Solution

A metrology sampling method that automatically adjusts the workpiece sampling rate using historical process data, actual measurement values, and statistical models like RI, GSI, DQIx, and DQIy to determine when to perform actual measurements and update virtual metrology models, thereby optimizing sampling frequency and accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If a fixed sampling rate of 1/N is used based on experience values, then the sampling process is simple to implement, but the sampling rate cannot be effectively adjusted and leads to increased cycle time and production cost

Engineering Contradiction:

Improvesimplicity of sampling implementationVSAvoidcycle time and production cost

Core Design Contradiction:

Ease of manufactureVSProductivity

Solution Approach 1:

The patent transforms the fixed sampling rate into a dynamic adaptive sampling rate that automatically adjusts based on process conditions. The system continuously monitors process data and VM prediction accuracy, then modifies the sampling rate accordingly - increasing it when VM accuracy degrades and decreasing it when VM accuracy is high, thereby optimizing both productivity and ease of operation

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system changes the sampling rate parameter dynamically based on monitored performance metrics. By adjusting the sampling frequency according to VM model accuracy and process stability, the system achieves optimal balance between measurement frequency and production efficiency without manual intervention

Inventive Principle:

Principle #35Parameter changes

2Productivity

If the sampling rate is reduced to lower production cost and cycle time, then productivity improves, but virtual metrology prediction accuracy deteriorates when production variations occur

Engineering Contradiction:

Improvecycle time and production costVSAvoidvirtual metrology prediction accuracy

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The system implements a feedback mechanism where VM prediction accuracy is continuously monitored against actual measurement values. When the deviation exceeds a threshold, the system automatically increases the sampling rate to obtain fresh measurement data for model retuning, ensuring VM accuracy is maintained while minimizing unnecessary measurements

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system performs preliminary checks of process data and VM prediction accuracy before determining whether actual measurement is needed. This preliminary assessment allows the system to maintain low sampling rates during stable conditions while quickly switching to higher sampling rates when variations are detected, preventing accuracy degradation before it occurs

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If a high sampling rate is used to ensure VM model accuracy, then measurement precision improves, but cycle time and production cost increase

Engineering Contradiction:

ImproveVM model prediction accuracyVSAvoidcycle time and production cost

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The system applies partial measurement action by using VM predictions for most workpieces and only performing actual measurements when necessary. By determining that full sampling is excessive and using conditional sampling based on process stability and VM accuracy thresholds, the system achieves adequate measurement precision with significantly reduced measurement frequency

Inventive Principle:

Principle #16Partial or excessive action

4Ease of operation

If the workpiece sampling rate is fixed at 1/N, then the sampling method is easy to operate, but it cannot automatically adjust to changing process conditions

Engineering Contradiction:

Improvesimplicity of sampling rate selectionVSAvoidability to adjust to production variations

Core Design Contradiction:

Ease of operationVSAdaptability or versatility

Solution Approach 1:

The system performs self-service by automatically monitoring its own performance and adjusting the sampling rate without external intervention. The automated system evaluates VM prediction accuracy, process data quality, and production conditions to dynamically modify sampling parameters, eliminating the need for manual rate selection while maintaining ease of operation

Inventive Principle:

Principle #25Self-service

Data Source

PatentUS10269660B2Metrology sampling method with sampling rate decision scheme and computer program product thereof

Publication Date: 2019.04.23 NAT CHENG KUNG UNIV

AI summary

In a metrology sampling method with a sampling rate decision scheme, a mean absolute percentage error (MAPE) and a maximum absolute percentage error (MaxErr) of visual metrology values of all workpieces in a set of determinative samples (DS), and various index values that can detect various status changes of a process tool (such as maintenance operation, parts changing, parameter adjustment, etc.), and/or information abnormalities of the process tool (such as abnormal process data, parameter drift/shift, abnormal metrology data, etc.) appearing in a manufacturing process are applied to develop an automated sampling decision (ASD) scheme for reducing a workpiece sampling rate while VM accuracy is still sustained.