798 THE BELL SYSTEM TECHNICAL JOURNAL, JULY 1956 



19 



^ 18 

 O 



z 



tr 1 7 



o 



u. 



\^ 16 



< 1 5 



5 

 z 



m 

 I 



"t. 1 3 

 T 



12 



tr 



U 



1 1- 



10 



4 6 8 10 12 14 16 



GROUP FROM SAME ZONE LEVELED BAR 



18 



20 



Fig. 13 — A pilot production process control chart. 



tively. In later sections the authors have adopted the ratio Ic(Q, —5) 

 /c(0, —20) as a measure of the success of the forming. 



The 2N21 transistor is a hermetic seal version of a point -contact 

 medium-speed switching transistor. During the early stages of the de- 

 velopment of this device, it became evident that although similar ger- 

 manium and point wire are used for l^oth structures, the electrical 

 parameters by which the devices are characterized belong to different 

 universes. However, if the geometery of the 2N21 unit is duplicated in 

 a manipulator transistor, the resulting device parameters do resemble 

 those of the earlier unsealed unit. It is therefore likely that an unknown 

 variable in the 2N21 process is responsible for the different universes 

 mentioned above. The effect of such a variable is shown in Fig. 13, which 

 shows a chart of a continuous process control. Each point represents 

 the average of foin- different units sampled at the particular point in the 

 process denoted in the legend. The micromanipulator data represents 

 measurements taken on wafers which have been processed up to but not 

 including point wire attachment. The curve denoted ''header" repre- 

 sents data taken immediately after the point-wire attachment. This is 

 one additional process step beyond th(^ point at which the manipulator 

 data was found. It is evident from this curve that a severe degradation 



