Feature: Machine Vision
Machine vision software can detect defects, and carry out measurement, matching and alignment during manufacture
Automation allows for high speed and high precision manufacture. Machine vision soſtware can detect defects, and carry out measurement, matching and alignment, all the while making the complex semiconductor manufacturing processes highly optimised and efficient.
Machine vision – a crucial part in semiconductor manufacturing By Klaus Schrenker,
Business Development Manager, MVTec
T
here are few other product categories that require as many production steps as the manufacturing of semiconductors. Tis production process is intricate, involving hundreds of different steps, making their implementation and coordination correspondingly complex.
In this sector, there is a great need for accurate and precise
technologies that can be rapidly implemented. Tus, machine vision becomes a key technology here. With it machines and robots can “see”, allowing the automation of the many inspection and alignment processes found in semiconductor manufacturing.
22 July/August 2025
www.electronicsworld.co.uk
Necessary impetus In semiconductor manufacturing there is at least one stage where checks are made for functional or optical defects of the product. Using machine vision to perform this quality check automatically offers many advantages over manual inspection: Machine vision is much faster, the results are reproducible, and the quality of the inspection is not susceptible to human operator error due to fatigue or monotony of the task. AI-based deep learning technologies bring even more to this sector. For example, deep- learning-based anomaly detection enables automated surface inspections that were previously impossible. AI-enabled systems can be trained against a dataset of defects and anomalies, to quickly check if a product suffers from any such faults. Tis dataset is continually updated through machine learning, as soon as any other type defect surfaces – all this without stopping the manufacturing machines rolling. In terms of quality control, it is essential to determine
dimensional stability in addition to checking for defects. Machine vision measures edges along lines or circular segments with sub- pixel accuracy in a matter of milliseconds – in 2D and 3D. Apart from quality inspection, finding and aligning wafers and
ICs is another application where machine vision provides effective support. Tis primarily relies on subpixel-precise shape matching. Te technology finds the objects in real time, even if they have rotated, distorted in perspective or deformed, or have become partially obscured or even outside the captured image.
Quality and precision In generic terms, the semiconductor production processes are divided into front- and back-end. In front-end production, chemical and physical processes are repeatedly applied to a substrate – a silicon wafer – to produce microelectronic circuits, layer by layer. In the back-end stages the individual wafers are then separated, fitted with contacts and packaged, ready for shipping. Machine vision is a key application in the front end, since it
detects defects on the wafer’s surface. Te technology identifies microscopic cracks, scratches, particle contamination and more, even in difficult lighting conditions; see Figure 1. As mentioned earlier, this can involve various technologies, like deep-learning-
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44
