788 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1957 



of cathode arcs between active surfaces, and of cathode arcs between 

 inactive surfaces, are markedly different, as has been pointed out eadier. 



The fact that erosion by inactive, or clean-surface, arcs takes the 

 form of a mound on one electrode and a crater in the other means simply 

 that successive arcs tend to occur at the same place on the electrode 

 surfaces. This is because each arc must occur where the electric field 

 between approaching electrodes is highest, and the roughening from one 

 arc will be the site of the highest field before the next discharge occurs. 



With carbon particles on the surface, the situation is different. In 

 the case of active cathode arcs, the electric field between approaching 

 electrodes is highest at a point where a group of carbon particles, perhaps 

 pulled up by electrostatic forces, closes a large part of the electrode gap, 

 and an arc must necessarily strike at such carbon particles. In the ac- 

 tivating process, carbon is always being formed by an arc, but only at 

 its periphery; at the hottest parts of the arc, carbon which was formed 

 earlier, is completely removed. Not only does each arc move during its 

 lifetime, continually searching out new carbon which was formed earlier, 

 but a later arc will not strike at a point from which carbon was just 

 cleaned by an earlier arc. This restless movement from point to point 

 results finally in erosion that spreads over the surface in a way which is 

 likely to be statistically uniform. 



24ib) Silver and Gold. Although the character of the erosion at silver 

 (and gold) surfaces, and also its magnitude, are drastically altered by 

 activation. Section 1.4(b), the "direction" of the erosion is still in most 

 cases that characteristic of anode arcs. The predominant loss of metal 

 on closure is usually from the anode for active silver electrodes at voltages 

 too low for air breakdown, just as it is for inactive silver electrodes at 

 low striking voltages. This is in marked contrast to the behavior of pal- 

 ladium surfaces when they become active ; for active palladium surfaces, 

 loss of metal is always chiefly from the cathode. The behavior of silver 

 leads naturally to the hypothesis that even when the surfaces are active 

 arcs at low striking voltages are anode arcs, as they are when the surfaces 

 are inactive. 



This hypothesis has been subjected to test by F. E. Ha worth by the 

 same method used in the case of palladium surfaces. Small carbon parti- 

 cles were dusted on a polished silver surface, and the surface was ex- 

 amined microscopically after it had been subjected to a single arc under 

 the circuit conditions used in similar tests at palladium surfaces. When 

 the maximum particle diameter was 5 X 10~^ cm, it was found from 

 the microscopic examination that all arcs were of the cathode tj^pe (see, 

 for example, Fig. 7), but when the maximum diameter was 2.5 X 10"'* 



