Capacitively coupled radio-frequency N2 discharges at low pressures

Abstract

Capacitively coupled radio-frequency discharges (ccrf) in nitrogen mixtures are frequently used for the processing, modification and functionalization of different kinds of materials. Although nitrogen plasmas have been studied for many years, and despite their growing interest in applications, there is only partial knowledge about ccrf nitrogen plasmas. This paper uses experiments and modelling to study ccrf discharges in pure nitrogen, at 13.56 MHz frequency, 0.1–1 mbar pressures and 2–30 W coupled powers [1]. Experiments performed on two similar (not twin) setups, existing in the LATMOS and the GREMI laboratories, include electrical and optical emission spectroscopy (OES) measurements. Electrical measurements give the rf-applied and the direct-current-self-bias voltages, the effective power coupled to the plasma and the average electron density. OES diagnostics measure the intensities of radiative transitions with the nitrogen second-positive and first-negative systems, and with the 811.5 nm atomic line of argon (present as an actinometer). In the particular case of non-equilibrium ccrf discharges in nitrogen, a self-consistent modeling strategy must account for the interplay between the transport of particles, in the presence of density gradients and the rf field, and their production/destruction due to kinetic mechanisms involving both electrons and heavy species. Simulations use a hybrid code that couples a two-dimensional timedependent fluid module [2], describing the dynamics of the charged particles (electrons and positive ions N2 + and N4 +), and a zero-dimensional kinetic module, describing the production and destruction of nitrogen (atomic and molecular) neutral species [3]. The coupling between these modules adopts the local mean energy approximation to define space–time-dependent electron parameters for the fluid module and to work out space–time-averaged rates for the kinetic module. The model gives general good predictions for the self-bias voltage and for the intensities of radiative transitions (both average and spatially resolved), underestimating the electron density by a factor of 3–4.