Comparison of high vacuum and ultra-high-vacuum tantalum diffusion barrier performance against copper penetration

Abstract

We demonstrate that depositing Ta diffusion barriers under ultra-high vacuum conditions without in situ oxygen dosing allows for variations both in microstructure and in the concentration of chemical impurities that severely degrade barrier performance. The effects of deposition pressure, in situ oxygen dosing at interfaces, hydrogen and oxygen contamination, and microstructure on diffusion barrier performance to Cu diffusion for electron-beam deposited Ta are presented. 20 nm of Ta diffusion barrier followed by a 150 nm Cu conductor were deposited under ultra-high vacuum (UHV, deposition pressure of 1×10 -9 to 5 ×10-8 Torr) and high vacuum (HV, deposition pressure of 1×10-7 to 5×10-6 Torr) conditions onto 〈100〉 Si. In situ resistance furnace measurements, Auger compositional depth profiling, secondary ion mass spectrometry, and forward recoil detection along with scanning and transmission electron microscopy were used to determine the electrical, chemical, and structural changes that occurred in thin-film Ta diffusion barriers upon annealing. Undosed HV deposited Ta barriers failed from 560 to 630°C, while undosed UHV barriers failed from 310 to 630°C. For UHV Ta barriers, in situ oxygen dosing during deposition at the Cu/Ta interface increased the failure temperatures by 30-250°C and decreased the range of failure temperatures to 570-630°C. Undosed UHV Ta barriers have no systematic relationship between failure temperature and deposition pressure, although correlations between breakdown temperature, oxygen and hydrogen concentrations, and microstructural variations were measured.