Influence of tantalum/tantalum nitride barriers and caps on the high-temperature properties of copper metallization for wide-band gap applications

Electromigration behavior of Cu metallization interfacing with Ta versus TaN at high temperatures

“…[1][2][3] As Cu metallization is well-studied as the dominant interconnect in advanced integrated circuits (ICs), it is a good choice to start with by examining its applicability for high-temperature applications. We have, therefore, investigated the thermal stability of various metal stacks with Cu films having Ta and TaN layers as diffusion barrier and/or as surface capping layers at temperatures from 400 to 700 C. 4 Two distinct systems have been studied: asymmetric with Ta on one side and TaN on the other, and symmetric with either Ta or TaN on both sides. For the asymmetric systems, we have found that Ta is released from the Ta layer and moves through the Cu film to react with the TaN layer at the opposite interface via the Cu grain boundaries (GBs) after annealing at 500 C or above.…”

“…For the asymmetric systems, we have found that Ta is released from the Ta layer and moves through the Cu film to react with the TaN layer at the opposite interface via the Cu grain boundaries (GBs) after annealing at 500 C or above. 4,5 Without a capping layer, Ta is also released from the Ta barrier layer and diffuses up to the Cu surface, most likely driven by oxidation of Ta and surface passivation of Cu by Ta. 4,5 For the symmetric systems, no Ta diffusion could be inferred from sheet resistance measurement.…”

“…4,5 Without a capping layer, Ta is also released from the Ta barrier layer and diffuses up to the Cu surface, most likely driven by oxidation of Ta and surface passivation of Cu by Ta. 4,5 For the symmetric systems, no Ta diffusion could be inferred from sheet resistance measurement. 4 Since Ta cannot be released from the stable TaN, the Cu/TaN interface would stay intact after annealing.…”

“…4,5 For the symmetric systems, no Ta diffusion could be inferred from sheet resistance measurement. 4 Since Ta cannot be released from the stable TaN, the Cu/TaN interface would stay intact after annealing. This was confirmed with samples of Cu on TaN without any capping layer.…”

“…This was confirmed with samples of Cu on TaN without any capping layer. 4,5 For Ta on both sides of the Cu film, lack of a driving force could explain the absence of Ta diffusion. Instead, protrusion of Ta into Cu was found to occur at 600 C and above by means of cross-sectional transmission electron microscopy (TEM).…”