With the development of electronic products in the direction of miniaturisation and high integration, chip appearance inspection faces new challenges. Traditional 2D inspection can no longer meet the needs of identifying small and hidden defects. So how to achieve more accurate, multi-angle comprehensive inspection？

3D AOI provides a more reliable and accurate solution. The next article will be from the AOI principle, 3D AOI and 2D AOI technology comparison, application areas and other perspectives to explain the advantages of 3D AOI in the semiconductor field applications.

AOI (Automatic Optical Inspection) is a means to achieve automatic inspection through the optical system imaging, the core of the technology is to obtain accurate and high-quality optical images and their identification process.

The working logic of AOI to detect defects in chip appearance can be divided into four simple steps.

(1) image acquisition: the use of CMOS or CCD image sensor, combined with multi-angle light source and lens, the chip surface scanning, access to high-resolution high-contrast images.

(2) Data processing: using GPU or FPGA processor, de-noising, correction, segmentation and enhancement algorithms to improve the image quality and optimise the original captured image.

(3) Image analysis: using traditional image processing algorithms and machine learning algorithms to extract visual features such as edges, corners, areas, etc., and compare them with normal templates to judge the appearance of defects.

(4) Defect report: Generate a report on the distribution of identification defects, provide statistics and analysis of the number of various types of defects, and output detection results and defect-related information.

Therefore, the logic of AOI inspection can be imaginatively understood as a simulation of the human eye inspection process, through the "see" and "judge" to achieve automatic optical inspection. In order to support and realise the above four stages, the hardware system of AOI equipment includes four parts: working platform, imaging system, image processing system and electrical system, which is a multidisciplinary automation equipment integrating machinery, automation, optics and software.

Traditional 2D AOI and 3D AOI in the image processing, comparison and analysis and results output of the basic process is the same, the main difference is that the image acquisition method is different, 2D AOI using a surface array or line array camera from a single point of view to obtain the 2D image of the chip; and 3D AOI using multiple viewpoints or structured light and other ways to obtain the chip's 3D stereo image. As 3D AOI can obtain richer three-dimensional information, it can achieve more accurate and comprehensive defect detection.

Specific advantages are reflected in

(1) wider detection coverage, 3D AOI can scan the component from multiple angles, covering all three-dimensional space, while 2D AOI may have "Blind spots".

(2) Detection is more accurate and precise, based on the 3D image of the 3D AOI comparison detection is more accurate.

(3) can detect more types of defects, 3D AOI can identify some defects that 2D AOI can not detect, such as misalignment, deformation, warping and other defects in three-dimensional shape.

(4) Better image effect, 3D AOI generates a more intuitive three-dimensional view, which helps to quickly locate and identify defects.

In order to meet the chip lightweight and miniaturisation design needs, many domestic and foreign manufacturers have introduced 3D AOI inspection equipment. Suzhou Bozhon Semiconductor Co., Ltd. as a global semiconductor equipment R & D manufacturer, which uses the optical principle, for the chip appearance defect detection to create the MicroStar series — high-speed and highprecision automatic AOI inspection machine. Among them, IR9721 can be used for BGA, LGA, QFN, QFP and other packages to provide comprehensive 6side inspection and 3D measurement to ensure that the final chip package appearance quality and yield improvement.

In short, 3D AOI plays an important role in improving product yield, ensuring product quality and safety, which is widely used in consumer electronics, server motherboards, new energy vehicle circuit boards, X-ray machines, MRI and other precision medical equipment manufacturing, aerospace and other fields. In the future, with the empowerment of computer vision, deep learning and other technologies, the detection speed, accuracy and intelligence level of 3D AOI will be significantly improved, and the cost will be further reduced. It will be applied in a wider range of manufacturing industries, and become the key infrastructure equipment to ensure product quality in Industry 4.0 and smart manufacturing.