In the semiconductor chip manufacturing process, the quality inspection of bumps is crucial. WeLinkirt provides an efficient and precise inspection solution for semiconductor manufacturers with advanced AI vision technology.
User scenario: A leading semiconductor chip manufacturer needs to conduct quality inspection on the bumps on the chips during the bump soldering process in the chip packaging production line. Bumps are the key components for connecting the chip to the external circuit, and their quality directly affects the performance and reliability of the chip. The inspection objects are the missing and bridging defects of the bumps, which can cause abnormal electrical connections of the chip and affect the product yield.
Pain points: Traditional inspection methods mainly rely on manual visual inspection and rule-based machine vision inspection. Manual visual inspection has great limitations. It is difficult to unify the standards of different inspectors, which is prone to missed detections and false alarms. Moreover, the inspection efficiency is low, and the labor cost is high. Rule-based machine vision inspection has difficulty in accurately identifying complex bump shapes and tiny defects, with obvious quantification difficulties. At the same time, when the product is changed, the inspection program needs to be rewritten, resulting in a long changeover time and affecting production efficiency. In addition, with the continuous improvement of product quality and compliance requirements in the semiconductor industry, traditional inspection methods are difficult to meet the strict quality control standards.
Technical principle
WeLinkirt uses advanced AI algorithms and self-developed 3D camera technology to solve the bump inspection problem. The AI algorithm is based on a deep learning vision foundation model, which can accurately recognize the features of the bumps. Through a large number of positive sample learning, the model can quickly adapt to different types of bump shapes and defect features. The self-developed 3D camera can obtain the three-dimensional topography information of the bumps and realize three-dimensional topography reconstruction. This three-dimensional information can more comprehensively reflect the real situation of the bumps, and details such as hidden solder joints, coplanarity, and micron-level topography can be clearly presented. Compared with traditional 2D imaging, 3D imaging can effectively avoid misjudgments caused by factors such as illumination and angle, greatly improving the accuracy of the inspection.
- The deep learning model quantifies and analyzes the features of the bumps, such as size, color, and surface grading, and transforms the "appearance" standard of the human eye into a consistent and quantifiable machine standard, thereby achieving more accurate defect recognition.
- The three-dimensional topography reconstruction technology of the 3D camera can capture the three-dimensional information of the bumps, and can judge the missing and bridging defects of the bumps from multiple angles, improving the reliability of the inspection.
- The APDT positive sample/few sample learning method enables the model to quickly learn and adapt to new product types with a small number of good samples (1 - 20 pieces), reducing the workload of sample collection and annotation.
- The semantic false alarm filtering function can filter false alarms according to the semantic information of the bumps, improving the efficiency and accuracy of the inspection.
WeLinkirt solution and product introduction
WeLinkirt provides the DaoAI AI AOI software system and the DaoAI 2D / 3D AI AOI equipment to solve the bump inspection problem. The DaoAI AI AOI software system has powerful feature recognition capabilities and can achieve 0-code automatic programming within 5 minutes for a good product. Through the APDT positive sample/few sample learning method, the system can quickly learn and adapt to new product types with only 1 - 20 good samples. The semantic false alarm filtering function can effectively reduce false alarms and improve inspection efficiency. The DaoAI 2D / 3D AI AOI equipment uses self-developed 3D cameras and three-dimensional topography reconstruction technology, which can detect details such as hidden solder joints, coplanarity, and micron-level topography. The equipment can conduct a comprehensive and accurate inspection of the bumps to ensure product quality. In the implementation process, WeLinkirt can deploy according to the specific needs of customers through SDK / API / Docker, etc., and support 100% local privatization to ensure that the data does not leave the factory.
WeLinkirt's AI vision technology provides an efficient and accurate solution for semiconductor bump inspection, effectively solving the pain points of traditional inspection methods.
Quantitative results: By using WeLinkirt's solution, the detection rate of the semiconductor manufacturer's bump inspection has reached 98.5%, and the missed detection rate has been reduced to <1.5%, greatly improving the product yield. The false alarm rate has been reduced by -65%, reducing unnecessary re-inspection work and improving inspection efficiency. The product changeover time has been shortened from several hours to 5min, significantly improving production efficiency and reducing production costs.