Noise and reliability in simulated thin metal films

Abstract

a b s t r a c tMany macroscopic aspects of electromigration damage in thin metal films have been investigated by means of Monte Carlo simulations based on simplified physical model. The employed model, can be described as a middle-scale model, in which the physical system is modeled with a high level of abstraction, without a detailed atomic physical model of the system.Among the many effects of the electromigration phenomenon, the simulator has been used to investigate several statistical properties of electromigration failure and the noise behaviour.Notwithstanding this simplicity, it is able to generate results in good agreement with many experimental observations: the lognormal distribution of failures, dependence of the mean time to failure from stress current and film geometry, Black exponent, noise statistics.Furthermore, this simulations confirmed a significant correlation between electromigration noise in the initial phase of stress and time to failure which has been suggested by a few experimentalists. This correlation can be usefully exploited as an early indication of the onset of electromigration damage on a per-sample basis.

Thermal Flow-Sensor Drift Reduction by Thermopile Voltage Cancellation via Power Feedback Control

“…The increase in current density near voids can eventually lead to failure of the thin-film resistor. Electromigration mainly occurs along grain boundaries having lower activation energy E a , which explains the direct correlation of the grain-size distribution with the electromigration noise and reliability performance of the thin-film [14], [15]. Annealing can be applied to increase grain size after recrystallization, which can improve electromigration performance [16], [17].…”

A new model for electromigration grain boundary noise based on free volume

“…Because of its convenient, sensitive and nondestructive capability, noise soon became a hotspot in research. At the present stage, researchers' interests mainly fasten on the comparison between noise and ordinary parameters, [10] or the relationship between noise and failure time, stress, geometry characterisation etc., [11,12] or noise time-frequency analysis. [13−15] However, there are fewer researches on the relativity between noise and the electromigration dominant defect-grain boundary.…”