412 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1952 



of a number of obstacles. While the operation of the first devices was 

 well understood in a general way, several items were limiting and puz- 

 zling, for example: 



a — Units intended to be alike varied considerably from each other — 

 the reproducibility was had. 



b — In an uncomfortably large fraction of the exploratory devices, 

 the properties changed suddenly and inexplicably with time and tem- 

 perature, whereas other units exhibited extremely stable characteristics 

 with regard to time — the reliability was poor. 



c — It was difficult to use the theory and then existing undeveloped 

 technology to develop and design devices to a varied range of electrical 

 characteristics needed for different circuit functions. Performance char- 

 acteristics were limited with respect to gain, noise figure, frequency range 

 and power— the designability was poor. 



Before the transistor could be regarded as a practical circuit element, 

 it was necessary to find out the causes of these limitations, to under- 

 stand the theory and develop the technology further in order to produce 

 and control more desirable characteristics. 



Over the past two years measurable progress has been made in reduc- 

 ing, but not eliminating, the three listed limitations. 



These advances have been obtained through an improved understand- 

 ing, improved processes and very importantly through improved ger- 

 manium materials. As a result: 



a — the beginnings of method have evolved in the use of the theory to 

 explain and predict the electrical network characteristics of transistors 

 in terms of physical structure and material properties. 



b — It is now possible to evaluate some of the effects and physical 

 meaning of empirically derived processes and thereby to devise better 

 methods subject to control. Previously, inhomogeneities in the material 

 properties masked the dependence of the transistor electrical properties 

 even on bulk properties (such as resistivity) as well as on processing 

 effects. 



c — ^As a result, on an exploratory development level, it is now possible 

 to make transistors in the laboratory to several sets of prescribed char- 

 acteristics with usable tolerances and satisfactory yields. 



d — Such transistors are greatly improved over the old ones in so far 

 as life and ruggedness are concerned, and some reduction in temperature 

 dependence has been achieved. However, it is not to be inferred that 

 all reliability problems are solved. 



e — It has become possible in the laboratory to explore experimentally 

 some of the consequences of the theory with the result that point con- 



