INVESTIGATION OF MATERIAL TRANSFER PHENOMENA FROM ELECTRODES DURING PULSED BREAKDOWN OF HIGH VACUUM

V. I. Eshkenazi

(Presented by the scientific seminar of the Research Institute of Nuclear Physics)

In recent works on the investigation of electrical breakdown of high vacuum, considerable attention is paid to those electrode marks that remain on the electrodes after breakdown 1, 2, 3, 4. However, all investigations concern either the prebreakdown stage with currents up to 2 mA, or complete breakdown, and there has been no work in which phenomena on the electrodes during pulsed breakdown of a single gap at one voltage were studied at different stages from prebreakdown to complete breakdown.

The experimental setup consisted of a rectangular-pulse generator with a rise time of 1 nsec, an amplitude up to 50 kV, a duration up to 0.5 μsec, and a coaxial vacuum chamber, one of whose electrodes was connected to cable \(Л_3\) with the pulse, and the other to an identical matching cable \(Л_4\) [5] (Fig. 1).

Fig. 1. Diagram of the experimental setup:
1 — discharger with nitrogen at a pressure of 10 atm, 2 — input of the firing pulse,
3 — voltage divider, 4 — oscilloscope trigger, 5 — observed signal;
6 — vacuum chamber with a cutting discharger and the gap under investigation,
\(Л_1 \div Л_4\) — RK-3 cables

A coaxial capacitive divider with a resolution no worse than \(10^{-10}\) sec for recording the phenomenon was installed at a certain distance in front of the chamber. The source of information about the process in the chamber was the signal reflected from the chamber.

The chamber and divider were checked for the absence of distortion by pulses with a rise time \(t_{\phi} = 5 \cdot 10^{-10}\) sec from a G5-12 generator.

With the electrodes short-circuited, pulses were recorded at the input and output of the chamber, and with the electrodes open—the reflected pulse. The oscillograms were observed with an S1-14 oscilloscope with a resolution of \(10^{-10}\) sec.

In parallel with the cable carrying the pulse in vacuum, a chopping spark gap was installed, in which the delay time of the discharge along the surface of the dielectric in vacuum was regulated. By varying this delay time of the discharge along the dielectric, it became possible not to bring the main gap to complete breakdown and to study the electrode marks obtained in this way.

A vacuum of \(5 \cdot 10^{-6}\) torr was produced in the chamber by a VA-0.5-1 unit with a TVL-500 thermoelectric trap.

The cathode was made of copper and was a plane 44 mm in diameter with a Rogowski profile, while the anode was a molybdenum hemisphere \(\varnothing 22\) mm. The anode—hemisphere—was positioned eccentrically with respect to the cathode—plane, at 11 mm from the axis. This electrode geometry produced a uniform electric field in the gap up to 1 mm. The chamber design made it possible to vary the distance between the electrodes and to set the investigated gap with an accuracy of 0.0005 cm and, by rotating the plane about its axis, to obtain each time, at a new location on the plane, marks from individual breakdowns.

The electrodes were carefully mechanically polished. The depth of the scratches was no more than 0.1 micron. Before installation in the chamber, the electrodes were degreased with carbon tetrachloride and alcohol and dried.

In the present article the results are described of studying a gap of 0.35 mm. Electrical breakdown of such a gap occurs at \(u = 35\) kV

Fig. 2. Protrusions on the molybdenum anode, formed as a result of electron bombardment from the cathode during 12 nsec. Magnification \(2000\times\).

and \(\tau_{\mathrm{fr}} = 1\) nsec in 30 nsec. Using the chopping spark gap, it was possible to obtain on the electrodes marks from breakdowns in an incomplete stage after 12, 18, and 30 nsec.

From photographs obtained from the electrodes on an MIM-8 metallographic microscope at a maximum magnification of 2000 in oblique light, which made it possible to reveal the microrelief more clearly, the transfer of material from one electrode to the other appeared as follows.

After the passage of a pulse with a duration of 12 nsec, protrusions that had formed on the anode surface are visible, about 1 μm high and 2 μm in diameter, about 20 in number over a surface area of 0.2 mm² (Fig. 2).

On the cathode, after 12 nsec, traces of transferred spheres 1 micron in diameter and smaller are visible over an area of 0.3 mm².

After the passage of a pulse with a duration of 18 nsec, protrusions up to 2 μm high are visible on the anode, with a diameter at the base of 3.6 μm and at the apex of 1.6 μm, as well as protrusions 1 μm high and 2 μm in diameter over an area of 0.2 mm².

On the cathode, after 18 nsec, traces of transferred molybdenum are visible; moreover, among the spheres 1 μm in diameter, traces of isolated spots are clearly visible (Fig. 3).

Fig. 3. Particles of molybdenum transferred by an incomplete discharge in 18 nsec to the cathode. Magnification 800×

After a completed breakdown, a spot 0.3 mm in diameter remains on the anode, consisting of small craters 1–5 μm in diameter, about \(10^4\) in number (Fig. 4).

On the cathode, after completion of the breakdown, a spot of Mo remains, occupying 0.3 mm².

On the basis of the results presented, the picture of initiation and development of the discharge from the prebreakdown stage to the arc stage appears as follows. Under the action of the electric field, electrons begin to be emitted from the surface of the cathode, which is covered with a film of adsorbed gas and oil. This prebreakdown current heats the surface of the anode,

on which protrusions grow, and part of the anode material from the heated tips of these protrusions, in the form of particles of diameter \(1\,\mu\), covers the cathode surface. The protrusions on the anode grow locally to a height of up to \(2\,\mu\), in a number of about 10. On the cathode we see traces of molybdenum that evaporated

Fig. 4. Crater on a molybdenum anode from one complete breakdown, consisting of small craters. Magnification \(280\times\).

from these grown protrusions. Further increase of the current in the gap, which is limited by the wave impedance of the cable, leads to evaporation of the anode, to the formation on the anode of a large spot consisting of \(10^4\) small craters, and to transfer of the entire mass of evaporated molybdenum to the cathode.

In the present work, growth of protrusions on the anode from the anode material and growth of microtips on the cathode due to evaporation of protrusions on the anode have been found.

The photograph in Fig. 4 confirms Maitland’s data [1] on electron beams heating the anode.

The absence of material transfer from anode to cathode, and conversely, during the first \(5\,\mu\text{s}\) denies the role of Cranberg particles [6] in the initiation of vacuum breakdown and the hypothesis of Maitland [1].

REFERENCES

1. Maitland A., 1961, J. Appl. Phys. 32, 2399—407.
2. Hawley R., 1963, Nature Lond 199, 978—80.
3. Little R. P. and Whitney W. I. 1963, J. Appl. Phys. 34, 2430—2.
4. Maitland A., Hawley R., Brit. J. Appl. Phys. 1965, 1591—92, 16.
5. G. A. Vorob’ev, G. A. Mesyats. Technique for Forming High-Voltage Nanosecond Pulses. Gosatomizdat, 1963.
6. Granberg L. 1952, J. Appl. Phys. 23, 518—22.