7-2 PROGRESS IN MICROSCOPY 



shows the image contrast-values according to various rates of ab- 

 sorption of the phase-plate versus the thickness e of the detail ob- 

 served and of its ;? — «' index difference from the medium encom- 

 passing it. 



Thus, in a detail 10 microns thick, having an index difference 

 n — n =0001 from the ambient medium, a 0-20 contrast obtains 

 when the phase-plate is transparent. This is not a low contrast and 

 yet such a value does not provide comfortable observation. The same 

 object, seen with a phase-plate dividing the incident intensity by 5 

 (// = 5), increases contrast to 0-4: a satisfactory level. Utmost con- 

 trast, for this object, is obtained by a phase-plate dividing incident 

 intensity by 50. The foregoing results imply a small fp. Now, this is 

 not always feasible in microscopy and ascertaining the image-contrast 

 variation versus the phase-shift 99 set in by the object is useful. Fig- 

 ure 2.6 shows the results evinced by two phase-plates setting — in 

 the normal 7i\l phase displacement between the diffracted and the 

 direct hght. The first is transparent {N = 1), the second absorbing 

 (A^ = 10). 



With a transparent plate, the contrast peaks when the phase dis- 

 placement set in by the object is 45° and cancels out when 9' = 90°. 

 With an absorbing phase-plate, variations occur much faster: when 

 A'^ = 10, maximal contrast obtains when the phase-displacement set 

 in by the object approximates 18° and cancels out when 7 == 34° 7'. 

 The contrast is reversed at ^< values exceeding 90 {N ^^ 1) and 34° 7' 

 (A'^ = 10). If fp remains small the portions OA and OB of these two 

 curves may be considered as being virtually rectihnear as shown in 

 formulae 2.8 and 2.9. In both these regions the contrast y is proportional 

 to the phase difference 9. When a specimen contains some details 

 exhibiting manifold 7 values, the curves in Fig. 2.6 evidence that all 

 the details cannot be descried at the same time with the same phase- 

 plate. Some are discerned with a high, some with a low contrast. 

 Details conforming to y = 90° are invisible with a transparent 

 phase-plate. 



Richter has shown that a 7r/2-dcphasing-plate is not the most 

 efficient one. The phase-displacement xp originated by the phase-plate, 

 for an object conforming to phase-difference 7, should be: 



y}=^90°-(pl2 (2.10) 



with absorption N: 



I /A/ - 4cos''v. (2.11) 



