342 BELL SYSTEM TECHNICAL JOURNAL 



This value of 4.45 micro-moles of oxygen per 100 taps corresponds 

 to 1.32 molecules of O2 or 2.64 atoms per electron equivalent of the 

 charge transported in the glow discharge, assuming complete discharge 

 of the condenser. This value entails a high efficiency in the combina- 

 tion of positive ions with the surface, and also predominantly poly- 

 atomic ions at least as complex as O3+. 



Microchemical analyses were performed on several cells to determine 

 the cathode composition and the final distribution of the caesium. 

 With an accuracy of about 3 per cent all the initial caesium was 

 recovered, 68 per cent being recovered from the silver cathode and 

 the rest from the inner surface of the glass bulb. No caesium was 

 found in the pellet residue. In the cathode surface was also found an 

 average of 0.13 milligram of undecomposed silver oxide. But the 

 actual amount may be greater for we have later evidence that at 

 least 7 per cent of the caesium on the cathode occurs as the free metal 

 which would reduce silver oxide when the cathode is extracted with 

 water and dilute acid prior to the analysis. 



The optimum conditions, as shown later, are obtained with 4.1 

 milligrams of caesium chromate in the pellet and a "ratio" of 19 

 condenser discharges per milligram of caesium chromate. These are 

 equivalent to 83 micrograms of caesium and 3.2 micrograms of oxygen 

 (using the factor from Table 1) per square centimeter of the cathode 

 and a ratio of 3.1 atoms of caesium per atom of oxygen. With 68 

 per cent retention of the caesium in the cathode, we obtain 56 micro- 

 grams of caesium per square centimeter and an atomic ratio of 2.1. 

 If we neglect the free caesium and residual silver oxide in the cathode 

 surface and assume that no oxygen has been lost from the cathode, 

 which implies that the caesium on the bulb has been oxidized by 

 reaction with water or other constituents of the glass — the atomic 

 ratio of 2.1 suggests that CS2O is the main caesium constituent of the 

 cathode surface. But however probable it may seem, it is not possible 

 to demonstrate this by ordinary analytical means in the presence of 

 free caesium and residual silver oxide. 



These initial and retained amounts of caesium are equivalent 

 respectively to 910 and 620 atomic layers on the plane cathode surface. 

 If we assume the surface to be increased by a factor of four in rough- 

 ening, we find the caesium retained in the cathode surface to be equiva- 

 lent to roughly 155 atomic layers of free caesium. This caesium will, 

 of course, occur mainly as caesium oxide and be mixed with the finely 

 divided silver from the reduced silver oxide. It would comprise 

 about 50 layers of CS2O molecules. 



L. R. Roller ^ has given quantitative data on surfaces which were 



* Loc. cit. 



I 



