Abstract: A method includes obtaining, by a processing device, data indicative of locations of defects of a substrate. The method further includes generating an image indicating the locations of the defects. The method further includes providing the image indicating the locations of the defects to a trained machine learning model. The method further includes obtaining, as output from the trained machine learning model, a classification of the locations of the defects. The method further includes performing a corrective action in view of the output from the trained machine learning model.

Abstract: A factory control server stores module configuration data for modules. The modules include processes for producing a final product and have corresponding module requirements. The factory control server analyzes in real-time actual product output data that is generated by a final product tester after a factory produces at least one final product to determine whether the actual product output data meets an expected product output. The factory control server analyzes actual module data in real-time to determine a new module requirement to cause new actual product output data for a subsequent final product to meet the expected product output in response to a determination that the actual product output data does not meet the expected product output. The factory control server notifies a module controller in real-time of the new module requirement. The module controller changes parameters in real-time to manufacture the subsequent final product.

Abstract: A factory control server stores module configuration data for modules. The modules include processes for producing a final product and have corresponding module requirements. The factory control server analyzes in real-time actual product output data that is generated by a final product tester after a factory produces at least one final product to determine whether the actual product output data meets an expected product output. The factory control server analyzes actual module data in real-time to determine a new module requirement to cause new actual product output data for a subsequent final product to meet the expected product output in response to a determination that the actual product output data does not meet the expected product output. The factory control server notifies a module controller in real-time of the new module requirement. The module controller changes parameters in real-time to manufacture the subsequent final product.

Abstract: Interlevel gaps between closely spaced circuit elements, such as closely spaced metal interconnect lines, are filed using a biased electron cyclotron resonance (ECR) deposition process. The gaps between circuit elements may be separated by distances of less than 0.6 microns and the gaps can have aspect rations in excess of 2.0. To fill such gaps between the circuit elements on a semiconductor wafer, the wafer is mounted in an ECR reaction chamber. A continuing flow of oxygen (O.sub.2) and silane (SiH.sub.4) gas is introduced into the ECR system's plasma and reaction chambers, respectively, while applying a microwave excitation so as to generate a plasma. High deposition rates and low film stress are achieved by controlling the flow of oxygen and silane so as to maintain an oxygen to silane gas flow ratio of less than 1.5.