EVAPORATION OF ATOMS 33 5 



also opened at Xo. The frame carrying the filaments was next inserted, 

 lower end (of smaller diameter then the inside of the main tube) first. 

 It was sealed at A^i and X2 after which the glass rod having fulfilled its 

 purpose was removed and its supporting tubes sealed off at Di and D2. 

 The filaments remained as previously adjusted. Fig. 6d shows the com- 

 pleted tube. The tube, with charcoal tube and ionization gauge attached at 

 F, was then evacuated, rebaked and tested with the gauge for leaks and 

 slow evolution of gas. The charcoal tube was separately outgassed at a 

 higher temperature. Just before sealing ofiF, a thin walled glass capsule 

 of caesium was broken in a side tube and about one cc of caesium distilled 

 into the large appendix H. The usual gas evolution in this process was 

 avoided by using capsules filled with caesium which had already been well . 

 freed from gas. 



The filaments A and B were 2.85 mm apart as determined by examina- 

 tion with a microscope through the clear glass annular folds. From this 

 spacing and the known diameters of A and B, the fraction (/) of the atoms 

 leaving A which are intercepted by B was calculated, f = 0.0071. 



Aging of tungsten filaments 



In previous work on the electron emission from clean tungsten it has 

 been found necessary to age the filaments to obtain reproducible results. 

 With Cs on W an aging process is perhaps even more important. All 

 filaments used were aged at 2400° K. for at least 10 hours followed by an 

 hour at 2600° and finally several short flashes at 2900°. It was further 

 shown that the rate of evaporation of Cs ions from the filament was a most 

 sensitive guide in this process. The cold filament was allowed to coat with 

 Cs, the tube being at room temperature. The tube was then immersed in 

 liquid air to freeze out all Cs vapor. Next the filament was heated to a 

 temperature (about 850° K.) where the rate of ion evaporation was slow 

 enough to be followed on a sensitive galvanometer. If the filament were not 

 sufficiently aged, the plot of evaporation rate against time or against the 

 amount of Cs, 6, left on the surface at any time, showed a number of 

 maxima or peaks as in Fig. 7. These peaks were also observed by Kingdon 

 in earlier unpublished work. Such irregular evaporation is probably caused 

 by non-uniformities of the tungsten surface and would make it difficult to 

 obtain rates and heats of evaporation cliaracteristic of the whole surface. 

 After sufficient aging, the rates of evaporation increase to maximum 

 values and the peaks merge into a smooth curve as also shown in Fig. 7. 

 Further aging produces no change. It was found that a fine grained filament 

 was most easily brought into this final condition and such filaments were 

 used in this work. The average grain size in these filaments was about one- 

 fifth the diameter of the wire. 



