REFLECTION 11 



23° or 38° (see Figure 2). The distortion was 

 then by a factor of 2.6 or 1.6 onh^ and the 

 pictures were comparatively easy to inter- 

 pret. One ditHculty encountered was the 

 rapid contamination of the specimen surface 

 by the intense electron beam. Fert overcame 

 this by using ionic bombardment of the 

 surface to remove the contamination con- 

 tinuously or, if desired, to etch the surface. 

 The resolution in the direction of maximum 



o 



magnification was again about 300 A and 

 was affected very little by the value of d^. . 

 The resolution in the direction of minimum 

 magnification was worse by a factor of 1/sin 

 di because of the foreshortening. Hence as 

 02 was increased the resolution in this direc- 

 tion improved markedly. The high quality 

 of the images was at first thought to be 

 rather surprising since it had generally been 

 supposed that the energy losses of the scat- 

 tered electrons would increase as the devia- 

 tion of the electron beam increased, thereby 

 introducing a serious loss in resolution be- 

 cause of the increased chromatic aberration. 

 When the spectrum of energy losses was 

 measured (Fert, Pradal, Saporte, and Simon 

 1958) it was found that this supposition was 

 incorrect, for the proportion of electrons with 

 high energy losses actually decreased as the 

 scattering angle was increased. 



It has thus been shown that for best re- 

 sults the largest practicable value of di + d^ 

 should be used. Page (1958a) has taken good 

 pictures with Qi -{- d^ = 26.5° using a modi- 

 fied commercial instrument and without 

 ionic bombardment of the specimen. Un- 

 fortunately the commercial instruments 

 available at present do not allow very large 

 angles of tilt of the electron gun. 



The choice of angle ^i is important. Since 

 the surface of the specimen is illuminated 

 obliquely any asperity on it will cast a 

 shadow. If di is small this shadow will be 

 long compared with the height of the asper- 

 ity. Vertical features may thus be given very 

 high contrast: suppose for instance that 



Fig. 2. Pearlitic steel specimen after ionic 

 etching: montage of three reflection electron mi- 

 crographs. © 1 = 2°, © 2 = 23°. (Photograph repro- 

 duced by courtesy of Professor C. Fert.) 



01 = 1°; then a step 100 A high would cast 



o 



a shadow 5700 A long and at a magnification 

 mj. of 2000 X would have a length in the 

 image of 0.16 mm if 62 = 7° or 0.50 mm if 



02 = 25°. This shadowing effect is one of the 

 most valuable features of reflection electron 

 microscopy, and in order to obtain good con- 

 trast it is normal to use an angle di of about 

 1° or if the surface is very smooth a smaller 

 angle which may be as low as 0.1°. Also it is 

 necessary to use a small value of 9i to obtain 

 good resolution. Fert, Marty, and Saporte 

 (1955) have shown that di must be less than 



3 or 4° if the resolution is not to suffer. Thej'' 

 attribute this to the fact that if 61 is small 

 the electrons will not penetrate far into the 

 specimen and there will be less chance of an 

 electron scattered there in the direction of 

 the objective suffering a second collision be- 

 fore leaving the specimen (particularly if 62 

 is large). Thus the angles 61 and 62 should be 

 chosen as follows: di should be made small 

 (-^1°) and its exact value chosen to give 

 the desired contrast for the particular speci- 

 men being examined, and 62 should be made 

 as large as is practicable. 



225 



