Research Of Methods For Bonds Inspection Of Automatic Wire Bonder

Wire bonding today is used throughout the microelectronics industry as a means of interconnecting the chips, substrates and output pins. As ICs continue to become more integrated and minimum feature sizes keep shrinking, the need for higher accuracy in fine pitch interconnect applications has become more important than ever. To inspect these wire bonds, the standards are stricter and the hardness is increasing. So, lower cost, simpler operation, higher accuracy and the ability for real time and on-line inspection will be the future research directions. In this dissertation, the formation and causes of wire bonds' invalidation are analyzed and classified, and two kinds of available methods of bonds inspection are researched and implemented. The one is a statistical method based on the data acquisition board, and the other is an on-line inspection method based on the machine vision. In this dissertation, the formation and causes of wire bonds' invalidation are researched and classified, which are divided into three groups, the wire-loss invalidation, the form invalidation and the performance invalidation. Besides, some theses including the metallurgical cracks in the heel and the bonds strength, the intermetallic formation in gold-aluminum system, and those factors that cause the bonds' invalidation are discussed. At the same time, this paper also discusses some general methods of bonds inspection, such as electrical testing, mechanical testing and visual inspection.The statistical method of bonds inspection based on the data acquisition board is proposed. This method acquires and analyzes the ultrasonic dynamic signals in the process of ultrasonic wire bonding to judge whether the bonds have wire or not. In this dissertation, the theory of bonds inspection is thoroughly discussed, the process of acquiring and analyzing the ultrasonic dynamic signals and the foundation of inspection standard are emphasized. Besides, the hardware construction of this method and its software module design are introduced in detail.The on-line inspection based on machine vision is emphasized here, which completely utilizes the existing hardware of Wire Bonder. These bonds images are acquired and processed, which are used to find these bonds' invalidation, such as incorrect position, wire-loss, overlong pigtail et al. In this dissertation, the theory of bonds inspection is thoroughly discussed, the key techniques of image processing including image enhancement, model matching, image thresholding and segmentation are emphasized. Besides, the hardware and software implementation of the above mentioned method is introduced in detail. Experimental results testify that this method is convenient and performable.