Detection And Application Of Noise In Integrated Circuits Interconnects

With the electronic devices developed to deep sub-micrometer, the interconnectioncross-sectional area of VLSI becomes smaller. The current density of interconnection increasessharply, the failure caused by electro migration becomes more and more prominent, which attractsmore attention.Based on a brief description to the conduct electricity mechanism for metal interconnectionand generation mechanism of noise, the method of noise measurement for metal interconnection arereviewed in this paper, the development of balanced bridge measurement system and low-resistanceVLSI interconnection noise detection have been accomplished. Compared with the measurementresult of the traditional system，the new system does an good performance in detecting the smallestsignal, the accuracy of signal detection and the amplification consistency of each frequency band forthe signal, so the new system is more suitable for detecting the noise of low-resistance metalinterconnection.The resistance and noise model for grain/grain boundary of metal interconnection have beenstudied emphatically. For the characteristic of Gaussian barrier existed in the grain boundaries inpolycrystalline materials, the grain boundary/grain resistance model have been established based ontunneling conduction mechanism in grain boundary and diffusion-drift conduction mechanism ingrain. The types and generation mechanism of low frequency noise of grain/grain boundary havebeen detailed studied, The grain boundary/grain noise model have been established based ontunneling conduction mechanism in grain boundary and diffusion-drift conduction mechanism ingrain. According to the theory above and the grain/grain boundary noise model, the impact ofrelevant parameters in each model on changes of noise has been analyzed. In a small bias voltage,the noise of grain boundary does not vary with the bias voltage, whilenoise of grain increases with the increasing current. Based on these results, the test program forseparating the grain/grain boundary noise has been established by changing the bias current. Finallytwo sets of noise detection experiment by changing current have been done to collect the noise ingrain and grain boundary and separate from each other successfully.