Characterization of WF6/N2/H2 plasma enhanced chemical vapor deposited WxN films as barriers for Cu metallization

W x N is a promising candidate as a barrier material for Cu metallization. In this work, we report the characterization of WxN films deposited by plasma enhanced chemical vapor deposition using WF6/N2/H2 gas mixtures. The films are analyzed by Rutherford backscattering spectrometry, Auger electron spectroscopy, atomic force microscopy, x-ray diffraction, transmission electron microscopy, differential scanning calorimetry, and sheet resistance combined with thickness measurements. The diffusion barrier properties are studied by using Cu-gate metal–oxide–semiconductor capacitors and subjecting to either bias-temperature stress (BTS) of 2 MV/cm at 250 °C or thermal anneal up to 700 °C, and evaluated by capacitance–voltage measurement. It is found that the as-deposited films with W/N ratios from 2–19 have an “amorphous-like” nature. Study of the initial growth shows that the WxN films form by nucleation and grow through coalescence, and the films exhibit a granular structure. The transformation from the amorphous-like structure to crystalline phase(s) depends on the film stoichiometry. For the W/N∼2 films, there are two transformations starting at 497 and 578 °C (at a heating rate of 10 °C/min), corresponding to the formation of the W2N phase and to the further growth of the phase together with the formation of a small amount of α-W, respectively. The corresponding activation energies are 2.20 and 2.98 eV. For the W/N∼3 films, there is only one transformation (to the W2N+α-W) starting at 601 °C and with an activation energy of 2.94 eV. The resistivity is found to be insensitive to the film stoichiometry when the film has an amorphous-like nature, but shows a larger decrease for W richer films upon annealing at 500 °C. Moreover, for W/N∼2 and 3 films, the resistivities are largely determined by the film thickness, from about 200 μΩ cm for the 120 nm films to 310–350 μΩ cm for the 10 nm films. BTS results reveal that 10 nm of the W/N∼3 barrier shows no sign of degradation at least up to 105 h. By contrast, it is found that the initial barrier degradation occurs by annealing at 600 °C and above. This is probably due to the formation of α-W grains in the WxN barrier at these temperatures.