560 



THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1956 



lattice as in Fig. 10. In an interstitial position lithium can approach an 

 oppositely charged boron, but the interaction will be, at the most, 

 coulombic so that an ion pair will form (see later sections). A covalent 

 bond is unable to appear not only because there are no electrons avail- 

 able for it, but also because the lithium ion cannot move to a position 

 where it can satisfy the tetrahedral symmetry inherent in sp^ hybridiza- 



tion. Calculations (of the sort appearing in the later sections of this 

 paper) show that at high temperatures, at the ion densities involved, 

 ion pairs of the kind depicted in Fig. 10 are completely dissociated. 



Suppose, however, that as temperature is raised vacancies dissolve 

 in the silicon lattice, and that one such vacancy occupies a position near 



Fig. 10 — Schematic diagram of a silicon lattice showing a lithium ion in an 

 interstitial position near a substitutional boron ion, as it occurs in an ion pair. 



a boron ion, as in Fig. 11, a slight modification of Fig. 10 in which the 

 dots represent electrons (dangling bonds). Unpaired electrons such as 

 these might capture an electron from the valence band of silicon so that 

 the vacancy acquires a negative charge and behaves like an acceptor. 

 It is reasonable to suppose that the positive lithium ion will move into 

 this negative vacancy, in the tetrahedral position, and form a covalent 

 bond as in Fig. 11. The lithium-boron complex so formed retains a nega- 

 tive charge and is thus a complex ion. If the specimen were extrinsic at 

 these high temperatures, there would still appear to be as many net 

 acceptors as before the addition of lithium.* 



If the LiB~ compound is stable enough (a question to which we shall 



* It is possible that rapid cooling may quench some of these LiB acceptors into 

 the crystal at room temperature. If this is so it should be possible to investigate 

 the associated energy level by Hall measurements in the interval of time before ' 

 the complexes anneal out. Similar phenomena might be observed in germanium. 



