1.2 The Electromigration Failure

When an electric current passes in a conductor, atoms are driven towards the anode due to the momentum transfer from the electrons to the atoms [1]. Due to the blocking boundary imposed by the barrier layer in Cu dual-damascene lines, there is an accumulation of atoms at the anode side and, at the same time, a depletion at the cathode end. Atom accumulation at the anode leads to the development of a compressive stress, while the depletion of atom at the cathode yields a tensile stress. This stress development can results in two distinct failures. If the compressive stress is sufficiently high and the surrounding dielectrics are weak, metal extrusion may form, causing short circuits [7]. On the other hand, a sufficiently high tensile stress leads to void formation, which can grow and span the line or via, so that the line resistance significantly increases and the interconnect fails [8]. Normally, the critical stress for extrusion formation is larger than that for void formation, and the latter is the dominant failure mechanism.

Electromigration induced failures normally present two distinctive phases. In the first one no electromigration generated voids can be observed in the interconnect and no significant resistance change of the line is detected [9]. This phase lasts until a void is nucleated and is visible in scanning electron microscopy (SEM) pictures. Then, the second phase starts, where the void can evolve in several different ways, until it finally grows to a critical size causing a significant resistance increase or completely severing the interconnect line [10,11]. Thus, the total electromigration lifetime is the sum of the time for a void to nucleate plus the time for the void to develop.

Figure 1.2 shows two typical failures in a copper dual-damascene line [12]. In the first case the void nucleates right under the via, so that the failure tends to occur soon after nucleation. Thus, the time to failure is dominated by the nucleation period. In turn, if the void nucleates away from the via, the void has first to migrate towards the via, where it then grows to cause the failure. Therefore, the lifetime is dominated by the void evolution phase [12].

Figure 1.3 shows the electromigration lifetime, normalized to the expected lifetime for the $1\,\mu$m technology node, as a function of the interconnect cross sectional area [13,14]. As the dimensions of the interconnects decrease, the electromigration lifetime also decreases, because the reduction of via and line dimensions requires a smaller critical void to cause the failure [13]. Moreover, as the line width is reduced beyond $100\,$nm, the growth of copper grains during the line fabrication is also reduced, leading to smaller grain sizes [15,16,17]. Consequently, the interconnect line changes from a bamboo-like to polycrystalline structure, so that grain boundary diffusion provides an additional path for mass transport [18]. Another contribution for shorter electromigration lifetimes comes from the introduction of low-k interlevel dielectrics [2,12].

Figure 1.2: Failures in a damascene line. (a) Failure dominated by the void nucleation phase. (b) Failure dominated by void nucleation migration and growth.

(a)

(b)

Figure 1.3: EM lifetime variation as a function of the interconnect dimensions. This curve is calculated based on equation (1.13).