CMSIUM-OXYGEN-SILVER PHOTOELECTRIC CELL 367 



and a selective maximum appears as in curves 3 and 4. Then its 

 height rises and its position moves to longer wave-lengths until the 

 maximum occurs at 8,000 A. as in curve 5, and the response at 10,000 A. 

 attains its highest value. From a consideration of prevalent concepts 

 of thin-film effects in thermionic and photoelectric phenomena one 

 should expect the surface in this state, which has apparently the 

 maximum value of the photoelectric threshold, to be covered with a 

 monatomic film of free caesium. As we continue increasing the amount 

 of caesium in the film, the height of the maximum still rises but its 

 position recedes to 7,500 A. and the response at 10,000 A. falls as 

 represented in curves 6 and 7. In the transition from curve 5 to 

 curve 7 the gain in response at 7,500 A. is offset by the loss in response 

 at and beyond 10,000 A. As a result the integral sensitivity to 

 tungsten light is sensibly constant and the optimum states for maxi- 

 mum integral sensitivity lie within this range. With further increase 

 in caesium film thickness the height of the maximum drops, and the 

 positions of both the selective maximum and the photoelectric thres- 

 hold recede towards shorter wave-lengths as in curves 8 and 9. 

 Eventually we should doubtless approach the spectral response curve 

 for bulk caesium ^^ with the selective maximum at 5,400 A. and the 

 photoelectric threshold at 7,100 A. 



The large changes in spectral response following the deposition of 

 minute amounts of caesium upon the cathode surface are further 

 evidence that the photoelectric effect is a phenomenon of the superficial 

 surface only and is conditioned mainly by the thickness of the surface 

 film of free caesium. Although the cathode surface as a whole is 

 prepared by irreversible chemical reactions which are arrested before 

 their completion, the film thickness of the standard surface is deter- 

 mined by the concentration of free caesium in the underlying matrix 

 and is maintained by a diffusion equilibrium. In the preparation of 

 the standard surface this partition is first determined at 220° C. in 

 the heat treatment. Stabilization involves a readjustment of the 

 surface to equilibrium with the adjacent portion of the matrix at 

 room temperature. But the attainment of complete equilibrium at 

 room temperature takes so long that many fluctuations of the ambient 

 temperature intervene and the actual history of such a photoelectric 

 surface involves a long series of diffusion waves accompanied by 

 small fluctuations in the thickness of the surface film which deter- 

 mines its photoelectric behavior. 



We wish to express our appreciation of the assistance of Mr. M. F. 

 Jameson in these experiments. Acknowledgment is also due to Mr. 

 H. W. Hermance for the chemical analyses, and to Miss A. K. Marshall 

 for the microphotographs. 



" Miss Seiler, Astrophys. J., 52, 129 (1920). 



