460 BELL SYSTEM TECHNICAL JOURNAL 



original magnification is 27-fold. Several lines were scratched on the 

 surface. The filament was mounted in a vacuum tube and activated. 

 Picture "a" shows the electron emission from the various granules of 

 the oxide. At this stage the cathode had the characteristic appearance 

 of ordinary oxide coated filaments. The filament was then flashed 

 at a comparatively high temperature for successive intervals of time. 

 After each interval another electron picture was taken. As a result 

 of this treatment the oxide evaporated so that the filament had a 

 metallic appearance. However, there is good reason for believing that 

 metallic barium had been alloyed with the nickel, and the emission 

 was much greater than that from clean nickel. The pictures show 

 that when the oxide has disappeared, different areas emit electrons 

 with greatly different intensities. The shapes and sizes of these areas 

 are strikingly similar to those obtained in ordinary optical pictures. 

 In fact Knecht states that the pattern is that of the nickel crystals. 



Numerous other pictures similar to Fig. 22 can be found in the book 

 by Briiche and Scherzer referred to above. Some of these indicate 

 that there is a fine structure non-uniformity inside of a single crystal 

 as well as the non-uniformity between crystals. 



The emission from thoriated tungsten has also been investigated by 

 electron optics and while the pictures are not as striking as those for 

 barium on nickel, they prove rather conclusively that the emission 

 varies from one crystal to the next or from one region of a crystal to 

 the next. Finally the emission from surfaces to which no impurity 

 has purposely been added have been investigated. These too show 

 patchy emissions. 



It has generally been felt and frequently stated that the emission 

 from clean surfaces, particularly clean tungsten, varied with applied 

 field according to the Schottky equation. While this is true at moder- 

 ate and high fields, I have never seen data which showed agreement at 

 low fields, but have seen data which showed disagreement. The 

 deviation is less pronounced than it is for thoriated tungsten but in my 

 opinion it is none the less real. This I have interpreted to mean that 

 even in a polycrystalline surface of clean tungsten, individual crystals 

 have work functions which vary by one-tenth or a few tenths of a volt. 

 This interpretation receives strong support from the work of Farns- 

 worth and Rose." They showed that a (111) surface of a single 

 crystal of copper had a contact potential with respect to a (100) surface 

 equal to 0.463 volt. The direction was such that the work function 

 of the (111) surface was 0.463 volt less than that of the (100) surface. 

 Even after heating the crystals to about 900° C. for as much as 1000 

 hours, the contact potential difference was still 0.378 volt and had 



