3 86 PHENOMENA, ATOMS, AND MOLECULES 



The migration of the Cs may be looked upon as a surface dififusion, the 

 number of atoms per unit time crossing each unit of length of a line on 

 the surface being Dda/dt. Treating the problem as one of diffusion, assum- 

 ing as usual that D is independent of o, it is then possible to calculate D 

 from the foregoing experiments. 



Measurements of the migration were made at To = 654°, 702°, 746°, 

 and 812° K. with times from 200 to 4000 seconds. At each temperature the 

 values of D obtained were approximately independent of ^2 (slight increase 

 with (2) . The values of log D plotted against i/T gave a straight line so that 



logio D = ~ 0.70 - 3060/T. ( I ) 



At 812° this corresponds to D = 3.4 X io'° cm^ sec."^. 



A few experiments were made in which the Cs concentration on the end 

 sections in the standard condition was altered to 1.74 X 10^^. At 812° a 

 value of Z> of 1.4 X 10"^ was found. Thus D varies considerably with so 

 that the experiments have given as yet only a kind of average value of D 

 over the range from amax to o. 



Determinations of Z) at higher concentrations have been made by another 

 method. In presence of a given pressure of Cs, the Cs film on a filament at 

 about 1000° K. surrounded by a negatively charged cylinder can exist in 

 two phases, separated by a definite boundary. From the more concentrated 

 phase (a phase) Cs evaporates as atoms while from the dilute /5 phase 

 Cs evaporates as ions ; both of these rates being equal to the rate of arrival 

 from the vapor phase. At a given filament temperature there is only one 

 pressure po at which these two phases can coexist. If the pressure is changed 

 to pi the phase boundary moves with a definite velocity v along the length 

 of the filament, causing either one or the other phase to disappear. 



At the phase boundary there is a sharp concentration gradient. The 

 resulting surface migration is balanced by differences between the rates 

 of evaporation into and arrival from the vapor. The rates of evaporation 

 of atoms and ions, Va and Vp, both depend on T and a, but the rate of 

 condensation u is fixed by the Cs vapor pressure. The total evaporation rate 



V = Va +Vp rises from o at o = o to a maximum at about = 7X10^-, 

 falls to a minimum at 43 X 10^^ and then rises indefinitely. The curve 



V — Li as a function of a thus intersects the a axis at three points and forms 

 two closed areas Ai and Ao, enclosed by the curve and the o axis. A recent 

 article,^ gives data from which such curves can be prepared. Mathematical 

 analysis shows that the condition for a stationary phase boundary is that 

 Ai = A2. 



^ 1. Langmuir, Jour. Amer. Chem. Soc, 54, 1252 (1932). 



