Chip Surface Defect Detection Technology Based On Digital Holography

The manufacturing process of chips is complex and prone to defects,which may greatly reduce or even damage the operating efficiency of semiconductor components and integrated circuits.Therefore,it is necessary to detect chip defects during the manufacturing process.It is difficult for traditional manual detection methods to meet the high speed and high precision detection requirements of semiconductor chip industry.Machine vision detection method is based on the digital image processing theory,which is unable to complete the detection of three-dimensional topography due to twodimensional limitations,and it is difficult to identify the defects of chip surface on the three-dimensional level.Digital holography uses the principle of optical interference to collect the wrapping phase image that records the amplitude and phase information of the object light wave in the form of interference fringe,and then performs phase unwrapping under certain conditions through the corresponding algorithm,reproduces the original object light wave,and obtains the two-dimensional intensity information and threedimensional contour information of the object to be measured.This method can effectively obtain the three-dimensional topography information of the surface of the object to be measured.Compared with machine vision detection method,chip surface defect detection has the advantages of high precision,high real-time performance and wider detection range,which is an ideal solution.In thesis,according to the surface flatness requirements and defect detection requirements of qualified products,the corresponding digital holographic phase-shift interference optical path system was designed for simulation verification,and the relevant optical components were purchased to complete the construction.The flatness parameters of the sample chip were calculated and verified,and then the defects of unqualified products were specifically distinguished through the corresponding evaluation parameters.Thesis mainly completed the following work:1)A four-step digital holographic image acquisition optical path system based on phase shift interference is designed.The optical path is based on the principle of phase shift interference,and the Mahch-Zender optical path structure is used to effectively reduce the impact of interference image acquisition noise,and the actual construction work is completed to meet the basic requirements of experimental image acquisition.Based on this system,the image acquisition of chip surface after phaseshifting interference is completed,and the corresponding wrapping phase is calculated based on the four-step phase-shifting formula.2)Theoretical analysis and calculation deduction were made on the basic mathematical principles of phase unwrapping and systematic summary was completed.Principles of various representative unwrapping algorithms were briefly introduced,and/or unwrapping operation was carried out on the experimental parcel phase diagram to obtain actual results.The possibility of various algorithms for real-time and efficient detection was analyzed combined with the efficiency and accuracy of the algorithm;3)A template matching image registration method based on mutual information is proposed,which not only takes advantage of the simple characteristics of template matching,but also uses mutual information to improve the matching accuracy,reduce the amount of calculation and improve the detection speed.Compared with the traditional template matching method,the multi-mode registration for spatial phase-shifted digital holography improves the evaluation parameter value by 15%.Compared with the maximum mutual information method,the image registration reduces the operation time by 90%.4)Combined with chip surface topography measurement technology and chip defect detection requirements,an effective,feasible and high-precision chip surface defect compliance detection scheme was designed,and the effectiveness of the scheme was verified through specific experiments,which met the requirements of response time less than 10 s,and the error and detection accuracy reached the order of magnitude.