Semiconductor Depth Profiling from Pixel Intensity Using ML

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current semiconductor examination processes require destructive methods to measure depth profiles, which are inefficient and costly, and there is a need for a non-destructive, accurate, and cost-effective method to determine the depth profiles of semiconductor specimens.

Innovation Solution

A system utilizing a machine learning model trained with a simulated training set to determine the depth profile of semiconductor specimens based on pixel intensity profiles, selecting an optimal landing energy to enhance sensitivity to depth variations, reducing the need for physical specimen cutting.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If destructive methods are used to measure depth profiles, then measurement precision is improved, but loss of substance increases and productivity decreases

Engineering Contradiction:

Improvedepth profile measurementVSAvoidspecimen material

Core Design Contradiction:

Measurement precisionVSLoss of substance

Solution Approach 1:

The patent creates a digital twin (virtual model) of the semiconductor specimen that replicates its physical properties and depth characteristics. By training a machine learning model on simulated pixel intensity data from this digital twin, the system can predict depth profiles from real specimen images without physical destruction. This copying approach resolves the contradiction by replacing destructive physical measurement with non-destructive virtual modeling.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The patent replaces mechanical destructive cutting methods with a computational approach using machine learning and simulation. Instead of physically sectioning the specimen to measure depth, the system uses pixel intensity profiles from imaging tools combined with a trained neural network to predict depth profiles. This substitution eliminates material loss while maintaining measurement capability.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If destructive methods are used to measure depth profiles, then measurement precision is improved, but loss of time increases

Engineering Contradiction:

Improvedepth profile measurementVSAvoidmeasurement time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by pre-training the machine learning model on extensive simulated data before actual measurement. The digital twin is created beforehand with known depth profiles, and the model learns the relationship between pixel intensity and depth characteristics in advance. During actual measurement, this pre-trained model rapidly predicts depth profiles from new images, eliminating the time-consuming destructive sectioning and measurement process.

Inventive Principle:

Principle #10Preliminary action

3Manufacturing precision

If traditional examination methods are used, then manufacturing precision monitoring is achieved, but device complexity increases

Engineering Contradiction:

Improvefeature dimension controlVSAvoidexamination system

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent introduces a machine learning model as an intermediary between the imaging tool and the depth profile measurement. This neural network acts as a mediator that translates pixel intensity profiles into depth information, bridging the gap between simple imaging and complex metrology. The intermediary model simplifies the overall system by replacing complex physical measurement apparatus with a computational layer that processes standard imaging data.

Inventive Principle:

Principle #24Intermediary (Mediator)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Enables accurate, non-destructive determination of semiconductor depth profiles, optimizing landing energy for sensitivity, and reducing the time and cost associated with traditional destructive methods.

Implementation Method 1

feed the data Dpixel_intensity to a machine learning model to determine, based on the data Dpixel_intensity, data informative of a depth of the given specimen

Methodology Applied

Scientific EffectMachine learning:

Implementation Method 2

at least part of the training set has been generated based on a model operative to simulate, based on one or more parameters informative of a specimen, data informative of a pixel intensity profile of the specimen

Methodology Applied

Scientific EffectSimulation modeling:

Data Source

PatentUS20260024186A1Methods and systems enabling determination of a depth profile of a semiconductor specimen

Publication Date: 2026.01.22 APPL MATERIALS ISRAEL LTD

AI summary

There are provided systems and methods comprising obtaining data Dpixel_intensity informative of a pixel intensity profile of a given specimen, feeding the data Dpixel_intensity to a machine learning model to determine, based on the data Dpixel_intensity, data informative of a depth of the given specimen, wherein the machine learning model has been trained with a training set, wherein at least part of the training set has been generated based on a model operative to predict, based on one or more parameters informative of a specimen, data informative of a pixel intensity profile of the specimen.