Yield Prediction Model for Semiconductor Substrates

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current semiconductor manufacturing processes face challenges in predicting electrical characteristics and yield, leading to inefficient use of resources and time-consuming final-stage testing, as well as uncertainty in optimizing process configurations for maximizing functional device production.

Innovation Solution

A method is introduced to predict electrical characteristics by determining the sensitivity of these characteristics to process metrics using analysis of electrical and metrology data from previously processed substrates, allowing for per-substrate and per-layer predictions based on process metrology data, thereby providing yield-related information before final product testing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If final-stage electrical testing is performed on all substrates, then accurate yield information is obtained, but time and resources are wasted on testing substrates that will definitely fail

Engineering Contradiction:

Improveyield information accuracyVSAvoidtesting time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by performing early prediction of electrical characteristics using process metrology data before final electrical testing. A prediction model analyzes process parameters from earlier manufacturing stages to identify substrates likely to fail, allowing early elimination of non-yielding substrates and avoiding waste of testing resources on obviously defective items

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary prediction model that bridges process metrology data and final electrical characteristics. This model acts as a mediator by correlating process parameters with electrical outcomes, enabling indirect assessment of substrate quality before actual electrical testing, thus reducing the need for comprehensive final-stage testing

Inventive Principle:

Principle #24Intermediary (Mediator)

2Productivity

If process configurations are optimized using traditional methods, then manufacturing efficiency is improved, but uncertainty remains in maximizing functional device production

Engineering Contradiction:

Improvemanufacturing efficiencyVSAvoidyield optimization certainty

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The patent implements feedback by establishing a correlation between process metrology data and electrical characteristics through prediction models. This feedback mechanism allows continuous refinement of process configurations based on predicted electrical outcomes, enabling data-driven optimization that reduces uncertainty in maximizing functional device production while improving manufacturing efficiency

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent applies parameter changes by using prediction models to identify optimal process parameter settings that maximize electrical yield. By analyzing relationships between process parameters and electrical characteristics, the system can adjust process configurations to achieve better yields with greater certainty, reducing the trial-and-error nature of traditional optimization

Inventive Principle:

Principle #35Parameter changes

3Reliability

If per-substrate and per-layer predictions are implemented, then early identification of issues is enabled, but computational complexity and data processing requirements increase

Engineering Contradiction:

Improveearly issue identificationVSAvoidprediction system complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies segmentation by implementing prediction at multiple hierarchical levels: per-substrate predictions, per-layer predictions, and per-die predictions. This segmented approach allows early identification of issues at appropriate granularities, managing computational complexity by focusing analysis where most beneficial rather than uniformly across all levels

Inventive Principle:

Principle #1Segmentation

4Ease of manufacture

If traditional final-stage testing is used, then resource allocation is simple, but non-functional devices are produced due to lack of early detection

Engineering Contradiction:

Improveprocess management simplicityVSAvoidfunctional device yield

Core Design Contradiction:

Ease of manufactureVSLoss of substance

Solution Approach 1:

The patent applies preliminary action by predicting electrical characteristics early in the manufacturing process using process metrology data. This early prediction enables identification and elimination of non-yielding substrates before final assembly and packaging, preventing loss of functional devices while maintaining relatively simple process management through integrated prediction models

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS11714357B2Method to predict yield of a device manufacturing process

Publication Date: 2023.08.01 ASML NETHERLANDS BV

AI summary

A method and associated computer program for predicting an electrical characteristic of a substrate subject to a process. The method includes determining a sensitivity of the electrical characteristic to a process characteristic, based on analysis of electrical metrology data including electrical characteristic measurements from previously processed substrates and of process metrology data including measurements of at least one parameter related to the process characteristic measured from the previously processed substrates; obtaining process metrology data related to the substrate describing the at least one parameter; and predicting the electrical characteristic of the substrate based on the sensitivity and the process metrology data.