Home         Authors   Papers   Year of conference   Themes   Organizations        To MES conference

Improving the Efficiency of Automated Monolithic Integrated Circuits Visual Inspection Algorithm
Authors
	Shiryaev B.V.
	Argunov D.P.
Date of publication
	2022
DOI
	10.31114/2078-7707-2022-4-56-62
Abstract
	The article describes an algorithm of automated visual inspection of MIC and the algorithm modifications improving its efficiency. The algorithm is the following sequence of a MIC microphotograph and a photomask processing: search for two or more matching points, calculation of the affine transformation matrix from vector coordinates of the photomask to raster coordinates of the MIC microphotograph, photomask rasterization to the scale of the MIC microphotograph, selection of the analysis area from the MIC microphotograph and the rasterized photomask, transformation of the MIC microphotograph by a classifier based on artificial neural network, calculation of the defect map, and search for anomalies on the defects map. The first modification of the algorithm is to change the order of the defect map calculation from the photomask to the microphotograph. This makes it possible to estimate the change in the MIC reference image. The second modification is to calculate the weighted average pixel distance of the defect map. This makes it possible to estimate the error responses of the classifier based on the artificial neural network. The third modification makes it possible to change the weight factor of the photomask layers, which makes it possible to specify the most important elements of the MIC topology. After the modification of the algorithm, its work was tested on sets of MIC microphotographs of different topologies. The test results were compared to the test results of the original algorithm. The testing showed an increase in the efficiency of the algorithm by changing the order of defect map calculation and changing the weight coefficient of the photomask layers. The repeatability of automated visual inspection with manual inspection after the modification increased to 98%.
Keywords
	visual inspection, defectoscopy, defect map, artificial neural network, MIC.
Library reference
	Shiryaev B.V., Argunov D.P. Improving the Efficiency of Automated Monolithic Integrated Circuits Visual Inspection Algorithm // Problems of Perspective Micro- and Nanoelectronic Systems Development - 2022. Issue 4. P. 56-62. doi:10.31114/2078-7707-2022-4-56-62
URL of paper
	http://www.mes-conference.ru/data/year2022/pdf/D065.pdf