Machine Learning-Based Defect Classification and Anomaly Detection in Advanced IC Package 3D Interconnects Using Scanning Acoustic Microscopy

Imec; Vlaams-Brabant; On-site; EUR 12,000 - 18,000; Full time

Benefits offered by this job

Imec allowance

Job summary

Imec invites applications for a Master/PhD internship focused on Machine Learning-Based Defect Classification and Anomaly Detection in advanced IC package 3D interconnects using Scanning Acoustic Microscopy. The project aims to automate failure analysis and improve reliability in 3D IC interconnects through data-efficient ML methods.

The role requires strong programming in Python (or MATLAB) and experience with ML frameworks (PyTorch, scikit-learn).

Qualifications

• Strong programming skills in Python (or MATLAB) with ML frameworks such as PyTorch and scikit-learn.
• Knowledge of signal processing, computer vision, or acoustic wave propagation.
• Interest in semiconductor packaging, failure analysis, and hands-on lab metrology (SAM).

Skills

Python

MATLAB

PyTorch

scikit-learn

Education

Master's degree

PhD

Job description

/ Machine Learning-Based Defect Classification and Anomaly Detection in Advanced IC Package 3D Interconnects Using Scanning Acoustic Microscopy

Machine Learning-Based Defect Classification and Anomaly Detection in Advanced IC Package 3D Interconnects Using Scanning Acoustic Microscopy

\"Where Artificial Intelligence Meets Failure Analysis: Advancing Reliability in 3D IC Interconnects.\"

Title: Machine Learning-Based DefectClassification and Anomaly Detection in Advanced IC Package 3D InterconnectsUsing Scanning Acoustic Microscopy

Technical Description:

As transistor scaling approaches its physical limits,advanced packaging has become essential to drive semiconductor performance,system integration, and reliability beyond traditional Moore's Law. In suchpackaging architectures, manufacturing defects such as micro-cracks, badwetting, and underfill voids can jeopardize device reliability. ScanningAcoustic Microscopy (SAM) can non-destructively detect these defects, butcurrent inspection relies heavily on slow, subjective manual review. Automatingthis process with machine learning would greatly speed up industrial failureanalysis, yet two main bottlenecks remain: defects occur too rarely to buildlarge training datasets, and tiny micro-cracks look almost identical to healthyjoints in acoustic scans.

Preliminary work has demonstrated that depth-resolvedC-SAM gating combined with anomaly detection substantially improves defectscreening over conventional supervised models. However, 2D C-SAM images discardvital phase, frequency, and time-of-flight information embedded in the rawacoustic waveform.

Objectives:

The primary goal of this project is to build anautomated, highly accurate failure analysis pipeline for advancedmicro-interconnects by overcoming the severe data scarcity inherent topackaging failure analysis. This project will investigate novel machinelearning frameworks capable of achieving high detection accuracy despiteextreme class imbalance and limited defective labels. The proposed methodology shouldaim to maximize model generalization and accuracy across varied interconnectarchitectures. In summary, some potential key objectives include:

• Expand inspection beyond C-SAM images by processing rawA-scan waveforms, extracting phase shifts, time-of-flight variations, and defectsignatures.
• Benchmark modern data-efficient architectures (e.g.,few-shot prototypical networks, metric learning, etc.) to maximizesensitivity toward rare defect classes.
• Leverage forward acoustic simulation tools to synthesizerealistic, defect-injected waveforms and scans, directly mitigating classimbalance and preventing model overfitting.
• Test model robustness across diverse package types (e.g.,FCBGA with exposed die, over molded, etc.), validating predictions againstdestructive cross-sectional inspections.

References:

[1] M. Gomes Abrantes, \"Defect Detection inAdvanced IC Interconnects,\" Master's Thesis, Faculty of EngineeringSciences, KU Leuven & imec, 2026.

[2] J. Snell, K. Swersky, and R. Zemel. Prototypicalnetworks for few-shot learning. In Advances in Neural Information ProcessingSystems (NeurIPS), 2017.

[3] S. Brand, P. Czurratis, P. Hoffrogge, and M.Petzold. Automated inspection and classification of flip-chip-contacts usingscanning acoustic microscopy. Microelectronics Reliability, 50(9–11):1496–1501,2010.

[4] C. S. Lee, G.-M. Zhang, D. M. Harvey, and A. Qi.Characterization of microcrack propagation through analysis of edge effect inacoustic microimaging of microelectronic packages. NDT & E International,79:19–27, 2016.

[5] M. Wolf, A. Sukumaran Nair, P. Hoffrogge, E.Kühnicke, and P. Czurratis. Improved failure analysis in scanning acousticmicroscopy via advanced signal processing techniques. Microelectronics andReliability, 138:114618, 2022.

Desired Qualifications:

• Strong programming skills in Python (or MATLAB) andmachine learning frameworks (e.g., PyTorch, scikit-learn).
• Knowledge of signal processing, computer vision, oracoustic wave propagation.
• Interest in semiconductor packaging, failure analysis,and hands-on lab metrology (SAM).

Contacts: [e-mail]

Type of internship: Master internship, PhD internship

Duration: 6 months to 1 year

Required educational background: Electrotechnics/Electrical Engineering, Materials Engineering, Physics, Nanoscience & Nanotechnology, Mechanical Engineering, Electromechanical engineering, Other

University promotor: Houman Zahedmanesh (KU Leuven)

Imec allowance will be provided.

Title: Machine Learning-Based DefectClassification and Anomaly Detection in Advanced IC Package 3D InterconnectsUsing Scanning Acoustic Microscopy

Technical Description:

As transistor scaling approaches its physical limits,advanced packaging has become essential to drive semiconductor performance,system integration, and reliability beyond traditional Moore's Law. In suchpackaging architectures, manufacturing defects such as micro-cracks, badwetting, and underfill voids can jeopardize device reliability. ScanningAcoustic Microscopy (SAM) can non-destructively detect these defects, butcurrent inspection relies heavily on slow, subjective manual review. Automatingthis process with machine learning would greatly speed up industrial failureanalysis, yet two main bottlenecks remain: defects occur too rarely to buildlarge training datasets, and tiny micro-cracks look almost identical to healthyjoints in acoustic scans.

Preliminary work has demonstrated that depth-resolvedC-SAM gating combined with anomaly detection substantially improves defectscreening over conventional supervised models. However, 2D C-SAM images discardvital phase, frequency, and time-of-flight information embedded in the rawacoustic waveform.

Cross-sectionalview of a solder bump in a FCBGA (flip-chip ball grid array) package.Extracted from [1]

C-SAM images atvarious depths ranging from solder bump to contact pad. Extracted from [1].

Objectives:

The primary goal of this project is to build anautomated, highly accurate failure analysis pipeline for advancedmicro-interconnects by overcoming the severe data scarcity inherent topackaging failure analysis. This project will investigate novel machinelearning frameworks capable of achieving high detection accuracy despiteextreme class imbalance and limited defective labels. The proposed methodology shouldaim to maximize model generalization and accuracy across varied interconnectarchitectures. In summary, some potential key objectives include:

• Expand inspection beyond C-SAM images by processing rawA-scan waveforms, extracting phase shifts, time-of-flight variations, and defectsignatures.

• Benchmark modern data-efficient architectures (e.g.,few-shot prototypical networks, metric learning, etc.) to maximizesensitivity toward rare defect classes.
• Leverage forward acoustic simulation tools to synthesizerealistic, defect-injected waveforms and scans, directly mitigating classimbalance and preventing model overfitting.
• Test model robustness across diverse package types (e.g.,FCBGA with exposed die, over molded, etc.), validating predictions againstdestructive cross-sectional inspections.

References:

[1] M. Gomes Abrantes, \"Defect Detection inAdvanced IC Interconnects,\" Master's Thesis, Faculty of EngineeringSciences, KU Leuven & imec, 2026.

[2] J. Snell, K. Swersky, and R. Zemel. Prototypicalnetworks for few-shot learning. In Advances in Neural Information ProcessingSystems (NeurIPS), 2017.

[3] S. Brand, P. Czurratis, P. Hoffrogge, and M.Petzold. Automated inspection and classification of flip-chip-contacts usingscanning acoustic microscopy. Microelectronics Reliability, 50(9–11):1496–1501,2010.

[4] C. S. Lee, G.-M. Zhang, D. M. Harvey, and A. Qi.Characterization of microcrack propagation through analysis of edge effect inacoustic microimaging of microelectronic packages. NDT & E International,79:19–27, 2016.

[5] M. Wolf, A. Sukumaran Nair, P. Hoffrogge, E.Kühnicke, and P. Czurratis. Improved failure analysis in scanning acousticmicroscopy via advanced signal processing techniques. Microelectronics andReliability, 138:114618, 2022.

Desired Qualifications:

• Strong programming skills in Python (or MATLAB) andmachine learning frameworks (e.g., PyTorch, scikit-learn).
• Knowledge of signal processing, computer vision, oracoustic wave propagation.
• Interest in semiconductor packaging, failure analysis,and hands-on lab metrology (SAM).

Type of internship: Master internship, PhD internship

Duration: 6 months to 1 year

Required educational background: Electrotechnics/Electrical Engineering, Materials Engineering, Physics, Nanoscience & Nanotechnology, Mechanical Engineering, Electromechanical engineering, Other

University promotor: Houman Zahedmanesh (KU Leuven)

Imec allowance will be provided.

imec's cleanroom