Dependence of Electromigration-Induced Failure Time on Length and Width of Aluminum Thin-Film Conductors

Abstract

Experiments have been undertaken to determine the length and width dependence of electromigration-induced failure time in aluminum thin-film conductors. A statistical model is presented and compared to the experimental data. The significance of the experimental data and the statistical model are discussed in terms of randomly distributed structural defects which produce flux divergences during electromigration.

Electromigration-induced failure of single layered NiFe Permalloy thin films for a giant magnetoresistive read head

“…These results indicate that all the samples tested in this experiment have various microstructures and different defect densities depending on their positions on the substrate. 12 The MTTF and standard deviation for samples with 160 and 180°C of ambient temperatures had 3.47 ͑2.12͒, and 3.74 ͑1.74͒ h, respectively.…”

The Influence of Temperature on Microelectronic Device Failure Mechanisms. Phase 2

“…They showed that time to failure versus current density varied complexly, and that a simple power-law dependence (as shown by Black) was inadequate to describe the experimental conditions over more than a small range of current densities. The results of I Attardo [1970], Black (1968], and Blair [1970] were tangential to the results from the Venables and Lye model at a temperature of 210°C and current densities ranging from 1 x 104 A/cm 2 to 2 x 106 A/cm 2 (Figure 2.10). Venables and Lye [1972] showed that when the temperature dependence of times to failure due to electromigration was represented as an Arrhenius plot, although the curves appeared as accurate straight lines, the slopes yielded only apparent activation energy, which varied with test conditions, indicating that the time to failure was 3 a complex function of baseline temperature and could not be represented by an Arrhenius plot to give an activation energy (Figures 2.11 and 2.12).…”

“…Conversely, the accumulation 3 of mass occurs wherever the electron flow is in the direction of decreasing temperature [Venables and Lye 1972] [Venables and Lye 1972]. Electromigration in thin-metallized films is confined mainly to grain boundaries [Blech 1967, Agarwala 1970, Rosenberg 1968. Thus, the steady-state temperature dependence of electromigration is of the same magnitude as that of…”

“…The effect is particularly severe in films with a small grain size, because a large number of nodes are available to act as nuclei for void formation. On the other extreme, if the grain size is comparable to stripe width, the probability that a single grain will cover the entire stripe increases, which introduces an additional 3 source of flux divergence acting as a barrier to atoms migrating from the negative side, and preventing the replacement of atoms transported away from connecting boundaiies on the other side of the grain [Attardo 1970, Blair 1970. The flow of current through stripes creates voids at grain-boundary nodes that are suitably oriented relative to the current flow direction and longitudinal temperature gradient.…”

A unified multiple stress reliability model for microelectronic devices — Application to 1.55 μm DFB laser diode module for space validation

“…Reliability models (Electromigration [5,7], Ohmic contact degradation [25,26], Coffin-Manson model [14], Eyring model [6,23,24], Humidity model [8], TDDB [15,16], Hot Carrier Injection [17,18], Hydrogen poisoning [9,[19][20][21], Thermo-mechanical stress [22], NBTI [23], etc…) are generally expressed by a function of stress parameter multiplying the exponential activation energy factor [10]. These expressions may not look much like a multiple-stress model but in fact it is as we have shown in paper presented at the 2013 IEEE Aerospace Conference [4]:…”