日本語
English
简体中文
In our post about detecting smaller features with metrology cameras, we made the designation between mainstream machine vision cameras and industrial metrology cameras. [See slide 7 in Yole Developpement’s Machine Vision Market report presentation]. Here are some more details on when a metrology camera will provide a performance advantage.
Metrology cameras are optimized for those applications when the pixel information is used as data for measurements such as in bright field/dark field illumination or interferometry. These are typically used in process control systems for semiconductor and electronics manufacturing but also others.
The standard approach with mainstream cameras is to take a good sensor, add in a bunch of functions and a cost cutting design. These cameras are often selected based on resolution, frame speed, and price criteria, which works well for positioning and general inspection/detection applications like food and paint quality verification or factory automation. Image sensor innovation is driving what can be done with mainstream cameras, and they can be used more widely than before.
But, if your imaging chain is used to differentiate your system or if the quality of the starting image is critical to the overall accuracy of the system, then read on.
High quality image data
With metrology cameras, special attention is given to ensure the best pixel data. With a full understanding of the image sensor, there are ways to drive the sensor in a specific way for an application to increase performance. By tuning the sensor to certain settings, camera manufacturers can reduce the amount of defects the sensor generates for example. Defects and non-uniformity generation depends not only on the sensor design but also on the conditions it is operated in, like temperature. Camera embedded calibrations can also be done automatically in the field to adjust to system conditions, such as temperature variations, optics imperfections, and clocking. These measures among others provide the dynamic range, uniformity, linearity, etc. for detection of small details and accurate measurements.
No variations between images and cameras
Variations in the image should be smaller than the variations one is trying to measure!
Also, critical to the accuracy of automated measurements is consistent data such that the variations identified are not those from the camera or cameras. Cameras with consistent performance reduce the metrology variability and serves to better determine the process discrepancy. With both consistent images and camera-to-camera consistency (when multiple cameras are used per measurement or multiple tools are used per process line), any changes detected can be determined as process deviations, allowing root cause analysis to take corrective action.
Image-to-image consistency is maintained through a robust camera design as well as embedded functions mentioned above. Camera-to-camera consistency is achieved through detailed sensor mounting and alignment processes as well as rigorous, individualized assembly and test procedure process so that mechanical and electrical adjustments can be made.
Image sensor optimization, precise sensor alignment and specialized verification procedures are all measures that would be taken with metrology cameras, but may not be done with most mainstream cameras. These extra steps are done to provide the performance leaps required in new technology nodes as well as eliminate workarounds and adjustments.
