We have discussed electromigration (EM)-induced drift in polycrystalline damascene versus reactive ion etched (RIE) Al(Cu) in part I. For polycrystalline Al(Cu), mass transport is well documented to occur through sequential stages : an incubation period (attributed to Cu depletion beyond a critical length) followed by the Al drift stage. In this work, the drift behavior of bamboo RIE and damascene Al(Cu) is analyzed. Using Blech-type test structures, mass transport in RIE lines was shown to proceed both by lattice and interfacial diffusion. The dominating mechanism depends on the Cu distribution in the line, as was evidenced by comparing as-patterned (lattice EM) and RTP-annealed (interface EM) samples. The interfacial EM only occurs at metallic interfaces. In that case, Cu alloying was observed to retard Al interfacial mass transport, giving rise to an incubation time. Although the activation energy for the incubation time was found similar to the one controlling Al lattice drift, for which no incubation time was observed, lattice EM is preferred over interfacial EM because it is insensitive to enhancing geometrical effects upon scaling. When comparing interfacial electromigration in RIE with bamboo damascene Al(Cu), with the incubation time rate controlling for both, the higher EM threshold observed for damascene was shown to be insufficient to compensate for its significantly increased Cu depletion rate, contrary to the case of polycrystalline Al(Cu) interconnects. Two factors were demonstrated to contribute. First, there are more metallic interfaces, intrinsically related to the use of wetting or barrier layers in recessed features. Second, specific to this study, the additional formation of TiAl3 at the trench sidewalls further enhanced the Cu depletion rate, and reduced the rate-controlling incubation time. A separate drift study on RIE via-type test structures indicated that it is very difficult to suppress interfacial mass transport in favor of lattice EM upon TiAl3 formation.