Research On Bonding Order Optimization For 3D SIC Based On A Novel Cost Model

The bonding of 3D SIC cannot be removed after its implementation.If any bonding failure occurs,all the finished bonding chips needs to be abandoned.And the mid-bond testing can test all the issues it produces during the process of bonding in case the problematic dies enter the next round of bonding in order to ensure the yield,save the resources.However the testing time increases during the process of bonding.This thesis focus on the reduction of testing time and test cost overhead.This thesis proposes a novel approach to optimization of bonding order for 3D SIC.The proposed method is flexible and does not require the bonding order of dies to follow bottom-up style,or only dies of one layer can be bound at a time.The proposed algorithm enjoy the freedom of starting bonding from any layer,and multiple layers can be bound at a time.And a group of dies can be bound to form a partial stack,thus can be bound to another partial stack.Three types of different 3D SICs are built using ITC'02benchmark.Experimental results show that,Pyramid type 3D SIC reduces testing time by 40%,and upside down Pyramid 3D SIC and diamond 3D SIC both reduces testing time by 10%.Comparison of testing power dissipation of three types of 3D SIC,advisable test results can be obtained under lower power consumption.For the computation of bonding cost,abandon test produces when bonding fails.However,for the success bonding ratio of 3D SIC,only die failure and TSV failure are considered without taking consideration of the influence of the difference of the adjacent dies on the success bonding ratio.In terms of the situation,this thesis proposes the concept of difference factor,thus completing the cost model and forms a novel cost model.Under the proposed model,the new bonding sequence algorithm is implemented.Experimental results show that the proposed algorithm reduces test cost overhead effectively.