Optimisation of Laser Scanner Parameters using Neural Networks

Presentation of neural networks as a tool for optimisation of scanning parameters. Scanning of the devices is the last step in production of semiconductor devices. This is performed at highest level regarding quality and yield. Critical parameters at this stage are the laser sensor height and the laser power. The method using neural networks describes what the proper settings are in order to obtain high yield, quality and avoid over-rejection. The experiment has been performed at a back-end plant of leading European semiconductor company.