Plasma Immersion Into The Ta Film Modified And Cu / Ta-x / Sio <sub> 2 </ Sub> System Failure Mechanism

To optimize the performance of IC chips and to minimize the RC delay caused by interconnection structures, copper is induced as an interconnection material for its lower resistivity and high resistance of electromigration (EM) compared with aluminum. However, copper is quite mobile in Si based materials and its adhesion with SiO₂ is poor, which could seriously degrade the performance and reliability of copper based interconnections. Therefore, a barrier layer between copper and Si based materials is needed to resist the diffusion of copper into Si based substrate and to improve the adhesion of interface.Tantalum is a good candidate for barrier materials to copper, because it has high melting point, low resistivity, good adhesion with SiO₂ and no any alloy phase with copper.In this paper, Cu(200nm)/Ta(25nm)/SiO₂ structures are prepared by self-ionized-plasma physical vapor deposition (SIP PVD) Ta and magnetron sputtering copper deposition. Different dose of N⁺ and C⁺ are implanted into Ta by Plasma Immersion Ion Implantation (Pâ…¢) to improve the performance of barrier layer. Samples are annealed from 500℃to 800℃. The surface morphology, depth profile, reactions, interface and microstructure are investigated. The mechanism of copper diffusion, oxygen induced diffusion, the improvement of Pâ…¢ and the simulation of EM performance of samples are discussed.Grain boundary diffusion dominates the copper diffusion in polycrystalline Ta based films. Oxygen acts as a promoter for the copper diffusion. After C⁺ and N⁺ Pâ…¢, the fast diffusion path is stuffed, O is difficult to pass through grain boundary because due to the calculation of Krasko, the embedded energy of 0 in Ta is higher than C and N. Copper is also stopped effectively. An amorphous-like structure is observed after C⁺Pâ…¢ in Ta. α -Ta is supposed to form in β -Ta after N⁺ Pâ…¢. These are all reasons why the performance of barrier is improved. Harrison's B model and Whipple's analysis is used to calculate the diffusion coefficient and active energyof copper in Ta, this method modify the solution of Junji, Imahori. The ANSYS simulation of two layer of copper interconnection shows that, the thermal mismatch of different layer is a serious problem to consider. EM in copper is caused by both "electron wind" and current density gradient. The higher the resistivity of barrier the better in performance of current crowding.