Further insights into prepeak emission in pulsed radiofrequency glow discharge

Abstract

Side-on optical emission measurements, with spatial and temporal resolution, have been carried out on a modified Grimm type pulsed rf glow discharge, using a copper sample and argon as discharge gas. In particular, the early ignition phase of the pulsed discharge, prepeak, and the beginning of the plateau have been electrically characterized and spectroscopically evaluated along the plasma plume for several argon and copper emission lines. Different pulse repetition frequencies have been used while keeping a constant duty cycle of 25%, covering a range from the μs to the ms pulsed GD. In all the cases a spatially extended discharge has been observed during the first 10–20μs of the pulse, when the dc bias voltage has not been fully established yet. During this early phase of the pulse, the species present in the chamber are excited following the amplitude of the voltage symmetrical oscillations, and argon and copper emission can be detected at all the distances evaluated along the plasma axis at delays as short as 4μs. After the first 20μs approximately, the discharge has contracted close to the sample surface, the dc bias voltage has been almost established and the sputtering process is taking place. Emission registered through side-on observation, at positions relatively far from the sample, decreases and then progressively increases to reach a plateau if the voltage pulse lasts long enough. Moreover, the further the detection position is from the sample surface, the later the increase of the copper and argon emission is observed. Therefore transport phenomena involving both argon and copper species are now evidently leading to an estimated average speed of 100m/s.Additionally, end-on optical emission measurements have been carried out to evaluate the prepeak emission from the negative glow region close to the sample surface, and then to compare this emission with that occurring at farther positions (side-on detection experiments). The results have shown that the basis of both prepeaks might be different as the prepeak signal measured using end-on detection peaks around 10μs later than that observed using side-on detection.