Research On New Token Parameter Of Metallic Interconnection Electromigration Reliability

With the development of modern electronic technology, the miniaturization trend of the integrated circuit is more and more obvious, the characteristic measurement of the device is becoming the sub-micrometer and even deep, the volume proportion of metallic interconnection in circuit and the current density dropped on interconnection are more and more greater, electromigration has become one of the major failure mechanisms in VLSI.Based on a brief description to the damage mechanism for electromigration and traditional token parameter, the time series analysis methods are taken into electromigration noise analysis. The noise series of electromigration is one kind of fractal signals, fractal dimension and self-similarity of the system can be obtained through it. Correlation integral is a method of fractal theory, using it in electromigration analysis, we find that through electromigration process, the primary component of noise signal is changed, the dominant component is changed from random component to certainty component, and electrons movement is changed from equilibrium to non-equilibrium, it shows the generation mechanism of noise has changed. And through the estimate of correlation dimension, we can evaluate the fractal dimension of signal which brings us some useful information in system modeling and simplification. High order statistics reflects the non-gauss of signal, our analysis shows that the non-gauss is more and more obviously though electromigration process. These two methods are valid in characterizing electromigration process and predicting interconnection invalidation, they can be used as news analysis methods in electromigration research.