AVS 45th International Symposium
    Electronic Materials and Processing Division Thursday Sessions
       Session EM-ThP

Paper EM-ThP3
TEM Study of TiAl@sub3@ Formation

Thursday, November 5, 1998, 5:30 pm, Room Hall A

Session: Electronic Materials and Processing Poster Session
Presenter: C.C. Pace, University of North Carolina, Chapel Hill and MCNC
Authors: C.C. Pace, University of North Carolina, Chapel Hill and MCNC
M.K. Lamvik, Microelectronics Center of North Carolina
M.A. Ray, Microelectronics Center of North Carolina
A. McTeer, Micron Technology
Correspondent: Click to Email
Metallizations consisting of Al-Cu alloys and Ti layers are widely used in microelectronic device fabrication, because of their low contact resistance and resistance to electromigration and hillock formation. As device dimensions are reduced, reaction mechanisms between layers must be understood at the atomic level to develop device structures that minimize failures due to resistive heating and stress induced voiding. Structures consisting of 350nm Al-0.5% Cu on 20nm Ti sputter deposited onto SiO@sub2@/Si substrates have been studied using a Philips EM 430 transmission electron microscope (TEM) equipped with a heating stage. TEM specimens of as-deposited samples were prepared and annealed in situ in the TEM. The temperature was varied up to 450 °C and times up to several hours. Within this temperature range the reaction rate for the formation of TiAl@sub3@ in the in-situ experiments agree with the published literature, although the range in published values is quite broad. Cross-sectional TEM micrographs reveal a non-uniform TiAl@sub3@/AlCu interface. The roughness revealed by TEM contradicts interpretations of Rutherford backscattering spectra (RBS). RBS suggest a layer-by-layer growth mechanism, which would produce a TiAl@sub3@ layer of uniform thickness. To show the equivalence of an in situ anneal and a furnace anneal, TEM samples were prepared from a wafer annealed in a furnace at 400 °C for 50 minutes. Cross-sectional micrographs revealed similar interface roughness. For both anneals, less than 10nm of Ti was consumed. An unreacted Ti layer remained even after extended annealing. In-depth profiling with Auger electron spectroscopy (AES) was utilized to confirm the presence of unreacted Ti. A model of the reaction mechanism that results in a non-uniform TiAl@sub3@ layer in contact with unreacted Ti will be proposed. It is critical that this mechanism be understood and controlled as device dimensions are scaled to ever smaller sizes.