Research On Defect Detection And Precision Counterpoint Method Of Flip Bonding Process

The high speed development of semiconductor packaging technology cannot be separated from advanced technological equipment,especially the Wafer level equipment that is the difficult problem should be solved now and in the future.The machine vision system is the key technology of high performance inversion equipment,which provides detection and positioning function relative to the "eye"of the device.In this thesis,we researches on the wafer level vision system,implements the chip,base board to decide if qualified defect detection algorithm and the high precision positioning algorithm,the concrete research content is as follows:1.Understanding the position and advantages of inversion in semiconductor packaging process,Firstly analyzing the function of core modules of inversion equipment and the function of three visual systems.Researching flip chip bonding process visual inspection and visual research current situation and the demand of localization algorithm,which shows the visual detection and localization process existing problems,and clear the research idea and content of this article.2.Analyzing the imperfection of flip chip problem,including the lack of the ball,even welding,chip defects and dislocation,the grid fitting method is put forward,the effective identification of welding defect ball position,at the same time by the methods of gauss mixture model,solve the problem of solder ball size defect.3.Aiming at the positioning of the convex point of the inverted bonding process,a sub-pixel edge extraction algorithm based on quadratic polynomial interpolation and the least square round fitting method are adopted to extract the center coordinates of all the chip convex points.The center coordinate and Angle information of the rectangle are utilized to determine the location information of the convex point chip.4.According to Blob principle,defect detection is carried out for the defect type of inverted substrate.Based on the positioning requirements of the substrate,a shape template matching algorithm through edge direction was proposed.The positioning information of the substrate can be obtained effectively as well as the matching speed is improved.It lays the foundation for the precision alignment of chip and substrate.5.Setting up the chip/substrate visual detection and positioning experiment platform of flip chip bonding process.To determine the light source,camera,lens and other options which conducting calibration experiments on the system.Based on C++ programming tools,developing the chip/substrate visual detection and positioning system and the validity of system can be verified by chip or substrate.