Diagnostics of N_(2) and O_(2) dissociation in RF plasmas by vacuum ultraviolet emission and absorption spectroscopy

摘要

N_(2) plasmas are a useful source of N atoms and have been applied to the formation of a high-quality, thin oxynitride film which prevents boron from penetrating through a gate oxide film, and also to the growth of III-V semiconductor nitride films by plasma-assisted, molecular-beam epitaxy. O_(2) plasmas have also been widely used in microelectronics fabrication processes such as photoresist ashing. The next probable application of O_(2) plasma is the low-temperature, high-growth-rate formation of gate oxides for thin film transistors that is achievable using highly Kr-diluted O_(2) plasmas. In the above applications, the knowledge of N_(2) and O_(2) dissociation degree is indispensable for process improvement and optimization. In this presentation, we focus on the diagnostics of N_(2) and O_(2) dissociation in N_(2) and rare-gas-diluted O_(2) low-pressure plasmas. As for N_(2) plasmas, a novel, simple procedure by vacuum ultraviolet (VUV) optical emission spectroscopy is applied to evaluation of N_(2) dissociation degree in an N_(2), inductively coupled RF (13.56 MHz) plasma. This procedure is based on actinometry and determines a ratio of N atom number density against N_(2) by comparing N-atom optical emission line intensity at 174.4 nm with the N_(2) second-positive (0,0) band intensity. Depending on the RF power (0.3-1 kW) and N_(2) pressure (0.133-1.33 Pa) chosen for plasma generation, N_(2) dissociation degree is found to range from 0.006 to 0.044. These results are comparable with those reported previously (0.01-0.049) by others, confirming the physically reasonable evaluation of N_(2) dissociation degree by the present procedure. The contribution of stepwise excitation out of N_(2) (A~(3)(SIGMA)_(u))~(+) (N_(2) (A)) metastables to the N_(2) 2nd positive emission, which may cause a large error of N_(2) dissociation evaluation, is found to be negligible from the net excitation rate computed using an assumed cross section of the electron impact excitation from N_(2) (A) to the N_(2) C ~(3)(PI)_(u) state and the N_(2) (A) number density deduced from a simple kinetic model. Using the present procedure, N_(2) dissociation degree is evaluated within a possible error of 27 percent if the electron temperature in N_(2) plasmas is in a range of 3-6 eV. In addition, N_(2) dissociation degree is obtained without the knowledge of N and N_(2) translational temperatures, even when these species are not in thermal equilibrium. These features are the advantages of the N_(2) dissociation evaluation by VUV emission spectroscopy to that by VUV absorption spectroscopy.