Machine Vision Lighting for Semiconductor Inspections

Incredibly small manufacturing defects can cause semiconductors to operate outside the tolerances required. Conversely, modest improvement in process precision can cause a performance boost.

Today’s sophisticated machine vision systems are used in proactive ways to improve semiconductor manufacturing. It is no longer enough to inspect the final product – now, machine vision can monitor the entire process.

Wafer Bumping Ignites New Interest in Semiconductor Inspection

With new wafer bumping technology, it is more important to optimize the efficiency of semiconductor manufacture. Wafer bumping replaces wire bonding to produce greater numbers of semiconductors more precisely. The inspection process uses high magnification optics to detect changes in the status of individual wafer bumps.

Sophisticated Inspection Systems Tackle Production Challenges

Aceris 3D is one system designed to handle this high-intensity application.

A single Aceris 3D inspection head incorporates three independent light sources, two high-res cameras, and a structured light source. Using laser triangulation, it can capture surface geography information quickly with minimal speed reductions.

This architecture easily overcomes limitations of traditional 3D cameras, which must stop to acquire quality images. It also has the advantage of being functional in darkfield, brightfield, and side mode, making it especially good at detecting bump characteristics.

Effective lighting is a central element.

Mirrors, beamsplitters, and filters are used to separate the laser light from the visible light required in 2D imaging. In this way, both lighting modules can use the same optical axis and view through the ringlight. Reflected light is captured by a CMOS imager as the laser source is projected over the wafer at a 45-degree angle.

Systems like the Aceris 3D show that precision manufacturing will continue to evolve hand-in-hand with machine vision. The more powerful inspection systems become, the easier it will be to capture peak performance from both traditional and new technologies.