X - R A Y M I C R O S C O P Y A N D M I C R O R A D I O G R A P H Y PROCEEDINGS OF A SYMPOSIUM HELD AT THE CAVENDISH LABORATORY, CAMBRIDGE, 1956 Edited by V. E. COSSLETT Cavendish Laboratory, University of Cambridge, England ARNE ENGSTROM Department of Medical Physics, Karolinska Institutet, Stockholm, Sweden H. H. PATTEE, JR. Department of Physics, Stanford University, California 1957 ACADEMIC PRESS INC PUBLISHERS. NEW YORK
S7P- COPYRIGHT, 1957 BY ACADEMIC PRESS, INC. Ill FIFTH AVENUE NEW YORK 3, N.Y. ALL RIGHTS RESERVED NO PART OF THIS BOOK MAY BE REPRODUCED IN ANY FORM,BY PHOTOSTAT, MICROFILM, OR ANY OTHER MEANS, WITHOUT WRITTEN PERMISSION FROM THE PUBLISHERS LIBRARY OF CONGRESS CATALOG CARD NUMBER: 57-11935 s PRINTED IN THE UNITED STATES OF AMERICA
I A Scanning Microscope for X-Ray Emission Pictures P. DUNCUMB AND V. E. COSSLETT Cavendish Laboratory, Cambridge, England The purpose of the microscope is to form a picture of a surface by its X-ray emission and to analyze the elements in a selected volume of about one cubic micron in the surface by the characteristic lines emitted. This paper deals with the microscope, and some results of analysis are described on p. 617. It was convenient to develop the microscope first as a scanning projection microscope, somewhat similar to that proposed by Pattee (1), in order to have direct comparison with the normal projection microscope of Cosslett and Nixon (2). Figure 1 shows schematically the arrange- FIG. 1. Collection of X-rays in the scanning projection method. ment of the final lens with the electron beam scanned over the target by coils before the lens. The X-rays transmitted through each point in the specimen are detected by a counter after restriction by an aperture, and the amplified signal from the counter modulates the brightness of 374
X-RAY EMISSION SCANNING MICROSCOPE 375 a cathode-ray tube scanned in synchronism with the microscope beam. The screen, therefore, displays an absorption picture analogous to that from the static projection microscope. Clearly, it is best to put the specimen in contact with the target so that the counter aperture can be made as large as possible without spoiling the resolution. The intensity of each picture element is proportional to the number of quanta transmitted during the scanning time of the picture, and in order that the statistical fluctuation in this number shall not be visible, it must be greater than a certain minimum. With a screen of average brightness, the eye can just detect a change in brightness of about 15%, so that the statistical fluctuation must be less than this for a noise-free picture, and about 650 quanta per picture element are required. In order to collect sufficient quanta with a spot of 1 p and a copper target at 20 kv, it is necessary to use a counter aperture which subtends an angle of 2 at the target, and this gives a depth of focus for the specimen of 30 p. In practice, however, it is possible to put up with a picture showing some quantum noise on the screen and photograph it with an exposure time of several minutes to reduce the effect. An RCA electron microscope has been modified for projection scanning; the column was rebuilt with condenser and projector lenses only and reduced in length to about 80 cm. The counter first used was a scintillation counter using a sodium iodide crystal, and the picture was produced on a 6-in. cathode-ray tube with a long persistence screen at a scanning speed of about one picture in 2 sec. Figure 2 shows 200-mesh/in. copper grid and 800-mesh/in. silver grid photographed from the display screen. The thick bars of the finer grid are 6 p across, and the resolution is limited in this picture by the size of the spot on the display screen. The magnification was increased by reducing the area of scan of the microscope beam, and the same specimen then gave a resolution of about 2 p, limited by the diffusion of the electrons in the target. The accelerating voltage was 30 kv. The maximum area that could be scanned was about 0.4 mm square and the resolution is maintained out to the edges. Comparison with the static microscope showed that the scanning technique was satisfactory, and it appeared at first that the scanning projection microscope might have certain advantages over the static microscope. These are, first, that it can give a brighter picture than the fluorescent screen of the static microscope, and second, that it permits easy control of the contrast of the picture by electronic means. But it seemed, after some experiments, that neither of these features was sufficiently promising to pursue, and attention was therefore transferred to the emission method,
376 P. DUNCUMB AND V. E. COSSLETT In this mode of operation, the electrons strike the specimen directly and the X-rays emitted from each point are collected by the counter. If the specimen is thick, this must be done in the backward direction, but with a thin specimen, the X-rays may be collected along the axis as before, after passing through the specimen. There is now no restriction on the solid angle of X-rays collected, which can be as large as is practicable. A new final lens was designed to cope with both methods of collection and the emission microscope in its present form is shown in Fig. 3. FIG. 2. 200-mesh/in. copper grid and 800 mesh/in. silver grid imaged by scanning projection method. 30 kv. Magnification 200 x. The electron beam is focused by two magnetic lenses on to the specimen O and is scanned over it by the deflection coils. These coils are similar to those used by McMullan and Smith in their scanning electron microscope (3) but give the beam a double deflection, balanced so that the beam always goes through the lens aperture. The spot size is 1 p and the area of scan about y 2 mm square at largest. Part of the emitted X-rays are collected through the gap in the pole piece by a scintillation counter S, and the amplified signal modulates the brightness of the display tube, which is scanned in synchronism with the microscope beam. When the picture is formed, the scan can be stopped and the microscope beam accurately located on any feature of the specimen by positioning
X-RAY EMISSION SCANNING MICROSCOPE 377 the spot on the afterglow of the picture on the display tube. Through another window, part of the emitted X-rays pass into a crystal spectrometer for analysis of the emission spectrum from the selected feature. Alternatively, the crystal can be removed and the X-rays pass straight into a proportional counter P, which gives for each quantum a pulse of height approximately proportional to its energy. Using a single-channel pulse analyzer, one can pass only pulses corresponding to a given characteristic line and, with the beam scanning, use these to form the picture on the display tube, so that the distribution of the element emitting that line is shown up. ffl Electron Gun Display ' > Amplifier Deflection Coils Time Bases Scint. Counter "Magnetic Lenses /Crystal Spectrometer may be inserted Prop. Counter Amplifier Specimen 0 Rate Meter Pulse Analyzer FIG. 3. The X-ray emission scanning microscope. The second lens is shown in Fig. 4. The electrons are focused on to the specimen which is level with the lower pole face, and X-rays pass through a window into a crystal phosphor just outside the pole-piece gap. The light produced passes down a perspex rod leading to the photocathode of the multiplier outside the lens. Alternatively, the X-rays can be collected along the axis of the beam in the case of a thin specimen or for the projection picture of a thin specimen. On the right-hand side, X-rays pass out through a tube filled with hydrogen into the spectrometer for analysis. The specimen is mounted on a tube gripped by two O rings and can be moved laterally by the screws below the lens, when the tube tilts about the upper O ring. The specimen tube can also be moved up or down and rotated; it is insulated from earth in order to measure the incident electron current. The focal length of the lens is about 4 mm; this is slightly shortened when a thick ferromagnetic speci-
378 P. DUNCUMB AND V. E. COSSLETT men is used, but the spot is not distorted. The top face of the lens is recessed so that the second set of deflection coils can be as close to the aperture as possible, to keep the electron beam within the uniform part of the field at large deflections. V Photo- % Multiplier g Specimen Movements FIG. 4. Final lens in the scanning emission microscope. The resolution is again limited by the penetration of the electrons, and Fig. 5 indicates the resolution obtainable at 10 kv to be about 1 p. It shows 1500-mesh/in. silver grid and the X-rays are collected after transmission through it along the axis. The width of the bars is about 3 p, and the dark line down the center of one set is presumably due to their triangular cross section, the center absorbing more strongly than the edges. As an example of the pulse-sorting technique with a proportional counter and pulse analyzer, Fig. 6 shows 200-mesh/in. copper grid and 800 mesh/in. silver taken at 20 kv. In A, the pulse analyzer was set to accept the AgL characteristic at 4 A, and the silver grid appears brighter than the copper. In B the pulse analyzer was passing CuK pulses at 1.5 A when the copper appears brighter than the silver. This demonstrates the possibility of showing up the distribution of one element only, but the energy resolution of the proportional counter is not good enough to be able to do this when the elements differ by less than about 4 in atomic number. However, some recent experiments indicate that there would be sufficient intensity reflected from a curved crystal in the spectrometer to form a noise-free picture, and this would easily be capable of separating adjacent elements.
X-RAY EMISSION SCANNING MICROSCOPE 379 FIG. 5. 1500-mesh/in. silver grid by the scanning emission method. 10 kv. Magnification 630 X. FIG. 6. 200-mesh/in. copper grid and 800 mesh/in. silver. 20 kv. Magnification 300 X. A. Formed by AgL characteristic emission. B. Formed by CuK characteristic emission. Another way of displaying this same result is to regard Fig. 6 A and B as two components of a color picture and photograph them through different filters in register on the same piece of color film. Thus, if A is photographed with a green filter and B with a red, the silver comes out as green and the copper red. This was demonstrated. In principle one
' 380 P. DUNCUMB AND V. E. COSSLETT can attach a given color to a given X-ray characteristic wavelength, which may be a useful technique for presenting in one picture the relative positions of different elements. The main points about the scanning emission microscope are therefore as follows: 1. It will produce a picture of a surface with contrast depending on the local variation of X-ray emission with a resolution of 1 p and a maximum field of view of about y 2 mm square. 2. It then enables one to position the electron beam very precisely on a selected point on the surface and to analyze the emitted X-rays with a crystal spectrometer. 3. Using a proportional counter or possibly a curved crystal spectrometer, it can show up the distribution of a particular element over the surface. Some applications of these techniques are given on p. 617. References 1. H. H. Pattee, /. Opt. Soc. Amer. 43, 61 (1953). 2. V. E. Cosslett and W. C. Nixon, /. Appl. Phys. 24, 616 (1953); W. C. Nixon, Proc. Roy. Soc. A232, 475 (1955). 3. D. McMullan, Proc. Inst. Elec. Engrs. 100, Pt. 1, 245 (1953); K. C. A. Smith and C. W. Oatley, Brit. J. Appl. Phys. 6, 391 (1955).