QFP (Plastic Quad Flat Package)chip is a surface mount packaging form of integrated circuit chips, also knownas plastic square flat packaging technology. It is commonly used to packageintegrated circuit chips for analog and digital circuits, with advantages suchas miniaturization, high pin density, good electrical performance, and heatdissipation performance. Very suitable for use in high-frequency circuits andwidely used in the electronics industry, it is one of the most importantsemiconductor chips. Therefore, it is crucial to develop equipment for detectingthe appearance of QFP chips.

At present, there are thefollowing difficulties in the appearance inspection of QFP chips:

(1) The chip pin size andspacing between them are very small, and the pin size requirements are strict,which requires high accuracy of the detection equipment;

(2) The number of chip pins andthe number of detection items for a single chip, as well as the large number ofchip detection batches, pose a great challenge to the detection efficiency ofdetection equipment;

(3) The chip detection projectnot only requires ordinary two-dimensional detection, but also more difficultthree-dimensional detection, which puts higher requirements on the detectionequipment.

        Physicalimage of cube chip                               Physical image of rectangular chip

               Figure1: Physical image of QFN packaged chip

(1) Traditionaldetection methods

Traditional chip appearanceinspection uses manual visual inspection or manual measurement, and afterdefects are found, defective products are manually removed. With the rapiddevelopment of chip design and manufacturing technology, the requirements forchip appearance inspection are becoming increasingly high, and thedisadvantages of traditional inspection methods are becoming more and moreobvious:

(1) Slow detection speed:Traditional detection methods usually use manual visual inspection, which islimited by the reaction and operation speed of the human eye, making itdifficult to meet the needs of large-scale and rapid production;

(2) High cost: Traditionaldetection methods require a large amount of manual operation, with high laborcosts, and require a significant investment in purchasing and maintainingdetection equipment;

(3) Difficulty in achievingautomation: Traditional detection methods are difficult to achieve automation,which is not conducive to large-scale and efficient production line integrationand optimization;

(4) Unstable efficiency: Eachoperator has a different level of understanding of appearance inspectionrequirements, and human visual standards are non quantitative and non constantscales, which can cause fluctuations in inspection standards and directly leadto unstable product quality control;

(5) Low detection accuracy: Themanual visual inspection method is easily affected by factors such as visualfatigue and human error, resulting in low detection accuracy and easy omissionor false detection.

 (2)Machine vision based detection methods

Machine vision based inspectionmethods are an important technical means for the appearance inspection of QFPchips. Machine vision, as the name suggests, combines the automatic controlfunction of a machine with visual sensors. The sensors perceive externalthings, obtain judgment results through certain algorithms, and convert theminto control signals to control the executing mechanism to complete theexpected actions. As an efficient detection method, it has irreplaceableadvantages compared to traditional detection methods, such as non-contact, highaccuracy, high speed, and high reliability. By using high-resolution camerasand image sensors, high-definition images of QFP chips can be obtained, andimage processing algorithms can be used to analyze the images to detect qualityissues such as appearance defects, size errors, and surface contamination ofQFP chips.

Its system generally consistsof seven parts, namely the tested object, light source, optical imaging system,image capture system, image acquisition and digitization module, intelligentimage processing and decision-making module, and control execution module, asshown in Figure 2.

Firstly, select the appropriatelighting method based on the working conditions to illuminate the measuredobject. Then, use an image capture system (such as a camera) to capture theimage signal of the measured object, which is converted into a digital signalthrough a digital module (such as an analog-to-digital converter). Thededicated image processing system uses various specific algorithms to extracttarget features, such as area, quantity, position, length, etc, Finally, basedon the preset judgment criteria, make judgments and output results to controlthe driving execution mechanism for corresponding processing.

In practical applications,machine vision based inspection methods can be integrated with automatedproduction lines to achieve automated and efficient appearance inspection ofQFP chips. Meanwhile, by continuously optimizing image processing algorithmsand improving image resolution, detection accuracy and reliability can befurther improved, reducing missed and false detections.

                         Figure 2 Structure of IndustrialMachine Vision System

At present, many domestic andforeign manufacturers have launched AOI appearance defect detection equipment,and Suzhou Bozhong Semiconductor, as one of the domestic companies targetingall machines. It provides comprehensive 6-side detection and 2D/3D measurementfor various packaged chips such as BGA, LGA, QFN, QFP, etc., to ensure thefinal chip packaging appearance quality and yield improvement.

The algorithms and software of chip appearance defect detectionequipment are constantly being optimized and upgraded to meet the detection needsof various new types of defects. Equipment is constantly evolving in terms ofprecision and accuracy, automation and intelligence, multifunctionality andintegration, adaptability and stability. In the future, with the continuousinnovation and improvement of technology, chip appearance defect detectionequipment will become more efficient, accurate, and reliable, providing strongsupport for the sustainable development of the semiconductor industry.