This model returns the quality inspection results of Chemical Mechanical Polishing (CMP) pad by checking the pores in SEM images of the pad. The model takes images of the CMP pads as input and returns bounding boxes around pore detections in the image. During the quality insepction process of CMP pads creation, manufacturers extract sample from the pad, take pictures, and measure quality indicators including number, diameter, and density of pores for a given image. As this process is cumbersome and difficult to complete manually, this model can be used as a key accelerator to the quality control process.
See the model in action with a Modzy MLOps platform demo or start a trial
93.7% Pore Detection accuracy 85% accuracy from cross-comparison of inspector results
The CMP Pad is a consumable item that grinds the surface of a semiconductor wafer. To polish the wafer surface evenly, it is important that pores such as the formation of air droplets are evenly distributed on the pad. The pad makers check the quality of the pad before supplying semiconductor companies. So, the quality inspector of the CMP pad maker extracts samples from the sample pad, takes pictures with SEM, and measures quality indicators such as the number/diameter/density of pores for this image. In this process, it is very difficult for the inspector to check the sample images one by one manually, so this model gives the quality inspection results automatically using deep learning algorithms.
This model was trained with pore image datasets, and the data annotation was carried out with the correct answer dataset worked by the quality inspector of the pad maker who can check the pore quality
The model was tested on a 10% subset of the full dataset and achieves 98% pore detection accuracy.
See how quickly you can deploy and run models, connect to pipelines, autoscale resources, and integrate into workflows with Modzy MLOps platform
d o n o t fill t h i s . f i e l d d o n o t fill t h i s . f i e l d
