As was shown in Figures 5(a) and @), PD channels which reach the
plane electrode form the subsequent pulses after the primary pulse in
the current and light intensity waveforms. In order to discuss the difference
between the primary and subsequent PD pulses, a selective measurement
of the primary and subsequent PD current pulses was carried
out utilizing the phase-gate control method. Figure 6 shows the results
at g=50 mm, Va=22 kV,, and k=50 ppm with different gate width at
8=90". Note that the observation area of PMT is restricted around the
needle tip. From the measured results of both the primary and subsequent
PD pulses in Figure 6(a), PD channels reach the plane electrode.
The selective measurement results for only the primary current pulse in
Figure 6(!3), indicates that the PD channel did not reach the plane electrode.
On the other hand, the PD light emissions for only subsequent
current pulses in Figure 6(c) were observed at two spots; the needle tip
and the plane electrode. At this moment, the subsequent pulse of light
intensity waveform corresponding to that of the current waveform, did
not appear. This means that the PD at the needle tip emitted inside the
observation area of PMT was pulseless as shown in Figure 4(b), and
appeared even during the prebreakdown streamer. Therefore, the subsequent
current pulse was formed by the PD not at the needle tip but
around the plane electrode. In other words, the PD is considered to be
the 'return stroke' [9], which arises from the electron avalanche around
the plane electrode and propagates toward the needle tip. From the
above discussions, the BD in the ultra-dilute SF6/N2 gas mixtures can
result from the propagation of the return stroke after disappearance of
the prebreakdown streamer.