556 



THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1956 



it can attack the oxide. Two specimens of 0.34 ohm cm p-type silicon 

 doped with boron were cut from adjacent parts of a crystal. Each 

 specimen w^as about 1 cm long, 0.2 cm wide, and 0.15-cm thick. The 

 samples were lapped on No. 400 silicon carbide paper, etched in HF and 

 HNO3 and sealed in helium-flushed evacuated quartz tubes, one con- 

 taining a small grain of P2O5 . The tubes were then heated at 1,200°C. 

 for 24 hours. This treatment introduced an n-type layer, highly doped 

 with phosphorus and about 0.001-cm thick, into the surface regions of 

 the specimen in the tube containing P2O5 . Upon removal from the tube 

 this specimen was lapped on the end to remove the n-skin. Complete 

 removal was determined by testing with a thermal probe. 



Small cubes of lithium (0.038 cm on a side) were placed on the ends of 

 both samples (the lapped end of phosphorus-doped one) and alloyed to 

 the silicon for 30 seconds at 650°C in an atmosphere of dry helium. After 

 this treatment the various junction contours should have looked like 

 those in Fig. 6, in which the bottom crystal is shown with the phosphorus- 

 doped skin (cross hatched). During the alloying process a small amount 

 of spherical diffusion of lithium occurs so that small hemispherical 

 n-regions form with the alloy beads as origins. These are shown in Fig. 6. 



Next the specimens were heated in vacuum for about six hours at 

 400°C. Diffusion of lithium into the body of the crystal should occur 

 during this period. However in the sample not protected by the n-type 

 skin lithium should leak to the oxide sink on the surface so that the 

 n-type region due to the lithium should have the pear-shaped contour 

 shown in the upper part of Fig. 7. If the built-in field at the p-n junction 



Fig. 6 — Initial stage following alloying in the diffusion experiment to demon , S 

 strate the impermeability to lithium of a heavily doped n-type skin on silicon. 



