808 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1957 



changed periodically so that a glow discharge could be made to occur at 

 each contact break, or at every 6th, 60th, or 600th break. The glow (air- 

 rent was always 0.04 ampere lasting for a time that could be accurately 

 set by means of a synchronized shunt tube. 



In all of the tests, the energy in each closure arc was 190 ergs. Meas- 

 urements were made at partial pressures of limonene of 0.05 and 1 mm 

 Hg. At the lower pressure it was found that the contacts remained inac- 

 tive indefinitely whenever the time of glow discharge on break was on 

 the average more than 0.25 microsecond for each closure arc, and activa- 

 tion would ultimately take place if the average glow time per closure 

 arc was less than this value. (At the limonene pressure of 1 mm Hg there 

 was a corresponding critical glow time of about 1 microsecond). The 

 obvious interpretation of these tests is that a glow discharge of 0.04 

 ampere lasting for 0.25 microsecond sputters and burns off as much 

 carbon as is made by an arc of 190 ergs under the conditions of the ex- 

 periment. To test this conclusion, one needs to know how much carbon 

 is produced by an arc of 190 ergs, and one needs to know the sputtering 

 rate of carbon in a normal glow in air at atmospheric pressure. 



Measurements of the sputtering of carbon in a normal glow discharge 

 were undertaken by F. E. Haworth, since such data are not available 

 in published literature. Carbon and graphite electrodes were weighed 

 before and after a normal glow discharge of 0.006 ampere lasting for 

 various lengths of time. The loss of the carbon or graphite negative elec- 

 trode in nitrogen was found to amount to about 0.15 atom per ion of the 

 discharge. In air the loss was much greater, four times larger for graphite 

 and 15 times larger for carbon (2.3 carbon atoms/ion for carbon in air). 

 The increase in air was attributed to burning, and the difference between 

 carbon and graphite losses in air was believed to be due to smaller crystal 

 size and looser bonding in the carbon case.* 



If we use the highest loss figure of 2.3 carbon atoms/ion we find that 

 a glow discharge of 0.04 ampere for 0.25 microsecond should remove 

 14 X 10^" carbon atoms. From Table IV, one finds that a 190 erg inactive 

 arc in activating benzene vapor produces 9.5 X 10^° carbon atoms in the 

 absence of air (line 3), and an active arc produces 34 X 10^° carbon 

 atoms (line 2).t The net carbon which is left after each arc in air is, of 

 course, considerably less than it would be in the absence of air (Section 

 6.1), but the order of magnitude agreement between these numerical 



* The sputtering rate of 0.15 atom/ion for carbon in nitrogen in the normal 

 glow is about what is reported by Guiithersoliulzei" for silver in the ahnormal glow 

 Init is greater by a factor of about 400 than that found for silver in the nurmnl 

 glow in experiments by Hawortli."* ()l)viousIy sputtering rates for carbon are 

 exceptionally high. 



t It is believed that these figures are substantially the same for linioncMic and 

 for benzene. 



