Biological low voltage field emission scanning electron microscopy

Biological Low-Voltage Scanning Electron Microscopy Biological Low-Voltage Scanning Electron Microscopy Edited by Hei...

Author: James Pawley | Heide Schatten

75 downloads 930 Views 13MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Biological Low-Voltage Scanning Electron Microscopy

Biological Low-Voltage Scanning Electron Microscopy Edited by

Heide Schatten Department of Veterinary Pathobiology, University of Missouri-Columbia, Columbia, Missouri, USA

James B. Pawley Department of Zoology, University of Wisconsin, Madison, Wisconsin, USA

Heide Schatten Department of Veterinary Pathobiology University of Missouri-Columbia 1600 E. Rollins Street Columbia, MO 65211 USA [email protected]

James B. Pawley Department of Zoology University of Wisconsin 250 N. Mills Street Madison, WI 53706 USA [email protected]

Library of Congress Control Number: 2007931613 ISBN 978-0-387-72970-1

e-ISBN 978-0-387-72972-5

c 2008 Springer Science+Business Media, LLC All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. Printed on acid-free paper. 9 8 7 6 5 4 3 2 1 springer.com

Preface

Ever since its advent in the 1950s, the scanning electron microscope hs offered an image of surface structures that was both uniquely detailed and easily interpretable. Although initially, the resolution of the images it produced did not match that of those made using the transmission electron microscope, this deficit no longer relevant on biological specimens where, for both instruments, image quality of is now limited fundamentally by the fragility of the specimen. High-resolution, low-voltage scanning electron microscopy (LVSEM) is now a powerful tool to study biological structure. Particularly when coupled with novel specimen preparation techniques, it has allowed us to understand in three dimensions objects that previously could only be imaged from serial-sectioned material analyzed with transmission electron microscopy. Because LVSEM provides images with clear topographic contrast from specimens coated with only an extremely thin metal coating, it can provide highresolution images of macromolecular complexes and of structural interactions that are free from the confusion of structural overlap. What is more, it provides this analysis in the context of being able to view the large specimen areas needed to provide context. These capabilities are particularly useful for applications in cellular biology. In addition, specific molecular components on surfaces and internal cell structures can be identified by using colloidal-gold labeling techniques. Particular internal structures can be viewed either by isolating them or by using novel fracturing and sectioning techniques that cause the internal components to occur on the outer surface of the specimen. There, the LVSEM can image them with a degree of topological precision that is often not possible with conventional TEM. Given these tremendous capabilities, it seemed to us both surprising and unfortunate that that LVSEM was not used more often to study biological structure. In a time when interest is extending from the genome to the proteome and when we increasingly want to know not only the existence but also the localization and interactions of specific molecules and molecular complexes, it seemed to us that LVSEM was the ideal modality for answering a myriad of important questions in cellular biology and development. What seemed to be needed was a way to make potential users more aware of LVSEM’s unique and powerful capabilities and also to provide the reader with both meaningful examples from a variety of applications and suitable protocols for preparing specimens. v

vi

Preface

We approached a number of leaders in the field with this idea and received a most enthusiastic response. The topics chosen were selected to be of interest to scientists, technicians, students, teachers, and to all who are interested in expanding their knowledge related to LVSEM. The specific topics covered in this book include highresolution LVSEM applications to cellular biology and detailed specimen preparation techniques for molecular labeling and correlative microscopy, cryoSEM of biological samples, and new developments in LVFESEM instrumentation in x-ray microanalysis at low beam voltage. Biological Low Voltage Scanning Electron Microscopy covers many aspects of specimen preparation and provides specific protocols for practical applications that are commonly not available in research papers. It also gives general as well as detailed insights into the theoretical aspects of LVSEM. The book is intended for a large audience as a reference book on the subject. By providing both theory and practical applications related to imaging biological structures with LVFESEM, we hope that it will fill a gap in the literature. During the editing process of this book, two of our most treasured colleagues, who have advanced the field immensely, passed away. Both the late Dr. Hans Ris and the late Dr. Stanley Erlandsen were passionate about the usefulness of LVSEM to enhance their own research, and as such, they left a wealth of new knowledge, novel techniques, and ideas for new applications for the scientific community. Their contributions are of great value to future scientist, students, technical staff, and many other using LVSEM. The editors are most grateful to all authors who have contributed their superb and unique expertise to this project and shared their insights with the present community interested in microscopy and those who will enter the field in the future. We would like to thank Kathy Lyons, our ever-so-patient editor at Springer. In addition, one of the editors (JP) would also like to thank Bill Feeny, the Zoology Departmental artist, and Kandis Elliot, the Botany Department artist, for their help in preparing the figures. Heide Schatten, Columbia, Missouri USA Jim Pawley, Madison, Wisconsin USA

Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

v

List of Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix 1 The Early Development of the Scanning Electron Microscope Dennis McMullan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 LVSEM for Biology James B. Pawley . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

27

3 The Aberration-Corrected SEM David C. Joy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4 Noise and Its Effects on the Low-Voltage SEM David C. Joy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 5 High-Resolution, Low Voltage, Field-Emission Scanning Electron Microscopy (HRLVFESEM) Applications for Cell Biology and Specimen Preparation Protocols Heide Schatten . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 6 Molecular Labeling for Correlative Microscopy: LM, LVSEM, TEM, EF-TEM and HVEM Ralph Albrecht and Daryl Meyer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 7 Low kV and Video-Rate, Beam-Tilt Stereo for Viewing Live-Time Experiments in the SEM Alan Boyde . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 8 Cryo-SEM of Chemically Fixed Animal Cells Stanley L. Erlandsen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 9 High-Resolution and Low-Voltage SEM of Plant Cells Guy Cox, Peter Vesk, Teresa Dibbayawan, Tobias I. Baskin, and Maret Vesk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 vii

viii

Contents

10 High-Resolution Cryoscanning Electron Microscopy of Biological Samples Paul Walther . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 11 Developments in Instrumentation for Microanalysis in Low-Voltage Scanning Electron Microscopy Dale E. Newbury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305

List of Contributors

Ralph Albrecht Department of Animal Sciences Department of Pediatrics Division of Pharmaceutical Sciences University of Wisconsin-Madison Madison, Wisconsin, USA Tobias I. Baskin Biology Department University of Massachusetts Amherst, Massachusetts, USA Alan Boyde Biophysics OGD, QMUL Dental Institute London, UK Guy Cox Electron Microscope Unit University of Sydney NSW, Australia Teresa Dibbayawan Electron Microscope Unit University of Sydney NSW, Australia Stanley L. Erlandsen (Deceased) Department of Genetics Cell Biology and Development University of Minnesota Medical School Minneapolis, Minnesota, USA

David C. Joy Science and Engineering Research Facility University of Tennessee Knoxville, Tennessee, USA Dennis McMullan 59 Courtfield Gardens London SW5 0NF, UK Daryl Meyer Department of Animal Sciences University of Wisconsin-Madison Madison, Wisconsin, USA Dale E. Newbury Surface and Microanalysis Science Division National Institute of Standards and Technology Gaithersburg, Maryland, USA James B. Pawley Department of Zoology University of Wisconsin Madison, Wisconsin, USA Heide Schatten Department of Veterinary Pathobiology University of Missouri-Columbia Columbia, Missouri, USA Peter Vesk Lecturer in the School of Botany University of Melbourne Melbourne, AU ix

x

Maret Vesk Electron Microscope Unit University of Sydney NSW, Australia

List of Contributors

Paul Walther Central Electron Microscopy Unit University of Ulm Ulm, Germany

Chapter 1

The Early Development of the Scanning Electron Microscope Dennis McMullan

It has been forty years since the scanning electron microscope (SEM) became a significant instrument in the scientific community. In 1965, the Cambridge Instrument Company in the United Kingdom marketed their Stereoscan 1 SEM, which was followed about 6 months later by JEOL of Japan with the JSM-1. Before 1965, there were about thirty years of intermittent SEM development in Germany, the United States, England, and Japan, although Japanese development was apparently not covered in the published literature. Development began in the 1930s in Germany, and then began towards the end of that decade in the United States. During these early years, there were two different approaches: the first, which had some specific relevance to the low voltage scanning transmission microscope (LVSEM); and the later one that was linked to the transmission electron microscope (TEM) and led to the scanning transmission electron microscope (STEM) and the future form of the current SEM. But first, we must share a few words about the early history of scanning and its application in microscopy.

Invention of Scanning In the 1840s, Alexander Bain, a Scottish clockmaker, invented the principle of dissecting an image by scanning, and he was granted a British patent (Bain 1843) for the first fax machine (McMullan 1990). At the transmitter, a stylus mounted on a pendulum contacts the surface of metal type forming the message, thus closing an electrical circuit. At the receiver, a similar stylus, also on a pendulum, records electrochemically on dampened paper. Following each swing of the pendulums, the type and the recording paper are lowered by one line. The means for starting the pendulums swinging simultaneously and synchronising them magnetically are described in the patent.

Scanning Optical Microscopy The first proposal in print for applying scanning to microscopy was made in Dublin by Edward Synge (1928). The proposal was for a scanning optical microscope, and his goal was to overcome the Abbe limit on resolution by H. Schatten, J. B. Pawley (eds.), Biological Low-Voltage Scanning C Springer 2008 Electron Microscopy.

1

2

D. McMullan

what is now called “near-field microscopy”—that is, the production of a very small light probe by collimation through an aperture smaller than the wavelength of the light. Synge was a scientific dilettante who had original ideas in several scientific fields, but did not attempt to put them into practice (McMullan 1990). However, he considered some of the problems that would be encountered with a scanning microscope and he proposed the use of piezo-electric actuators (Synge 1932), which are now used with great success in the scanning tunneling microscope and other probe instruments—including, of course, the near-field optical microscope itself. He envisaged fast scanning of the sample so that a visible image could be displayed on a phosphor screen, and he also pointed out the possibility of contrast expansion to enhance the image from a low contrast sample—probably the first mention of image processing by electronic means (as distinct from photographic).

Charged Particle Beams A proposal for using an electron beam in a scanning instrument was described in German patents by Hugo Stintzing of Giessen University (1929). These patents were concerned with the automatic detection, sizing, and counting of particles using a light beam, or—for those of below-light microscopic size—a beam of electrons. The focusing of electrons was (at that time) unknown to him, as to most others, and he proposed obtaining a small diameter probe using crossed slits. The samples would be either mechanically scanned in the case of a light beam, or one could use electric or magnetic fields to deflect an electron beam. Suitable detectors would be used to detect the transmitted beam that was attenuated by absorption or scattering. The output was to be recorded on a chart recorder so that the linear dimension of a particle could be given by the width of a deflection, and the thickness by the amplitude—the production of a two-dimensional image was not suggested. Stintzing did not apparently attempt the construction of this instrument and there are no drawings accompanying the patent specification. Thomas Mulvey (1962), however, published a block-schematic diagram of Stintzing’s proposal much later.

The Transmission Electron Microscope In the early 1930s, the main center for the development of the electron microscope was the Berlin Technische Hochschule in the laboratory of Professor Matthias where Max Knoll was a research assistant supervising students—including Ernst Ruska, whose subject area was electron optics. Arising from this work, Knoll and Ruska demonstrated the first transmission electron microscope with a magnification of x16. From the very beginning of electron microscopy, the imaging of solid samples was an important goal, particularly as the methods for producing thin samples were not developed until later. The first attempt was by Ruska (1933), with the sample

1 The Early Development of the Scanning Electron Microscope

3

Fig. 1.1 Early TEM image of an oxide replica of etched aluminium (Mahl 1940)

surface normal to the viewing direction and illumination by an electron beam at grazing incidence to the surface. Ruska obtained images of copper and gold surfaces but at a magnification of only x10. A few years later, he made a second attempt (Ruska & Müller 1940) with the same geometry and with only marginally better results. von Borries (1940) was much more successful with his grazing incidence method in the transmission electron microscope (TEM), where the sample surface is at a few degrees both to the viewing direction and to the illuminating beam. A breakthrough in the microscopic imaging of surface topography in the TEM was the 1940s introduction of replicas by H. Mahl (1940), and these set the standard for the next 25 years—although they were tedious to make and could be subject to serious artifacts. An early example is shown in Fig. 1.1.

Electron Beam Scanner Knoll, the co-inventor of the TEM with Ruska, was the first to publish images from solid samples obtained by scanning an electron beam (Knoll 1935). In 1932, very soon after the building of the first TEM at the Berlin Technische Hochschule, he joined the Telefunken Company as the director of research to develop television (TV) camera tubes. There, he designed an electron beam scanner for studying the targets of these tubes. A schematic block diagram is shown in Fig. 1.2 (the sample was mounted at one end of a sealed-off glass tube, and an electron gun was located at the other). The accelerating potential was in the range of 500–4,000 V, and the beam was focused on the surface of the sample and scanned by deflection coils in a raster of 200 lines and 50 frames/s. The current collected by the sample (the difference of the incident and secondary emitted currents) was amplified by a thermionic tube

4

D. McMullan

Fig. 1.2 Schematic diagram of Knoll’s (1935) electron beam scanner; (labels translated by T. Mulvey)

amplifier and intensity-modulated cathode-ray tube that was scanned by deflection coils connected in a series with those on the electron-beam scanner. By changing the ratio of the scan amplitudes, the magnification could be varied. Knoll used mainly unity magnification, but he could increase it to about x10 before the resolution was limited by the diameter of the scanning probe. This apparatus had virtually all the features of an SEM, but Knoll surprisingly (in view of his earlier work on the TEM) did not use additional electron lenses to reduce the size of the probe below 100 μm. The resolution he obtained, however, was entirely adequate for his purpose. The beam current was relatively high—on the order of microamps—and therefore thermionic tubes could be used to amplify the signal current in spite of the fast scan rate. He must have realized that reducing the size of the probe would be counter-productive because there were no suitable high-gain electron detectors in existence at that time. Similar images were produced by others working on the development of TV cameras in the 1930s (e.g., von Ardenne 1985), but Knoll was the only one at the time who looked at samples other than camera tube targets (e.g. silicon iron in Fig. 1.3), and he also elucidated the contrast mechanisms of secondary electron coefficients and topography. The images were true secondary electron images because the electron gun and sample were enclosed in the highly evacuated and baked glass envelope, and there was therefore little or no contamination of the surface. It is only comparatively recently that ultra-high vacuum (UHV) SEMs have been available that can work in this imaging regime. Knoll continued using his electron beam scanner (which he named der Elektronenabtaster) for a number of purposes, including the study of oxide layers on metals (Knoll 1941).


5

Fig. 1.3 Electron-beam scanner image of a silicon iron sheet showing electron channeling contrast; horizontal field width = 50 mm (Knoll 1935)

A few years later, Manfred von Ardenne, also in Berlin, built a very different instrument that was in fact a scanning transmission electron microscope (STEM) that he intended to also use for solid samples. His hope was not realized at the time, but his work established many of the principles used in all future SEMs. This is described further in this chapter, but first we will consider Vladimir Zworykin’s RCA microscope—which actually came after von Ardenne’s but was related much more closely to Knoll’s pioneering work—and also to some aspects of later LVSEMs.

The RCA Scanning Electron Microscope Zworykin, who was director of research at the RCA Laboratories in Camden, N.J., initiated a development program for SEM (Zworykin et al. 1942) in 1938 that continued until about 1942. He was a pioneer of electron-scanned TV camera tubes

6

D. McMullan

dating back to the 1920s, and had also developed the first optical microscope with video output (Zworykin 1934). The development of the SEM was done in parallel with that of a TEM, and by the same staff—in particular, J. Hillier, E.G. Ramberg, A.W.Vance and R.L. Snyder as well as Zworykin himself. Although Zworykin had every microscope paper from Germany translated as soon as it was received (Reisner 1989), he was apparently not influenced by von Ardenne’s work on the STEM/SEM. Instead, he started by repeating Knoll’s beam scanner experiments (in effect) using a monoscope. The monoscope was a pattern-generating, cathode-ray tube that had been invented in Knoll’s department at Telefunken (1935) and further developed by RCA for television use (Burnett 1938). His team then built an SEM based on the monoscope, but with two magnetic lenses to produce a very small focused probe, and a demountable vacuum system so that the sample could be changed (Zworykin et al. 1942). The scan rate was the US TV standard—441 lines and 30 frames/second—and the signal was amplified by a thermionic tube video amplifier. For a signal-to-noise ratio of 10, the signal current had to be 3 × 10−8 A, which could only be reached if the probe diameter was about 1 μm. He then tried to obtain a high current in a smaller probe by using a field emission gun with a single-crystal tungsten point, presumably based on experience with the point projection microscope that had been built in the RCA Laboratories by G.A. Morton and E.G. Ramberg (1939). To reach a sufficiently high vacuum, Zworykin had to return to having the gun and the sample in a glass envelope that had been baked and sealed off. A single magnetic lens was used, and fleeting images were obtained at x8,000 magnification, with scanning at TV rate and a thermionic tube amplifier. Stable images could no doubt have been achieved, but at that time a practical microscope would not have resulted because demountable UHV techniques did not then exist. To overcome the noise problem, Zworykin therefore decided to build an SEM with an efficient electron detector and a slower scan. The detector was the combination of phosphor and photomultiplier that T.E. Everhart and R.F.M. Thornley (1960) used nearly twenty years later in an improved form. To bring the secondary electrons to it, he designed an electrostatic immersion lens that retarded the beam electrons and accelerated the secondaries. Figure 1.4 shows the final electron optical arrangement. Electrostatic lenses were used to produce a demagnified image of the source on the sample that was held at +800 V relative to the grounded gun cathode. The electron beam leaving the gun was accelerated to 10 keV in the intervening electron optics. The secondary electrons returning from the sample were similarly accelerated, and diverged as they passed through the 4th electrostatic lens and hit the phosphor screen with an energy of 9.2 keV. In the first instrument, the scanning was done by electro-mechanically moving the sample relative to the beam using loudspeaker voice-coils and (later) hydraulic actuators—it was only in the final version that magnetic scanning of the beam was employed. The scan time was fixed at 10 min by the facsimile recorder that was used for image recording, and also controlled the microscope scans. There was no provision for a faster scan or the production of a visible image on a TV monitor—this seems strange remembering Zworykin’s TV background, but it may have been because the signal bandwidth was seriously limited by the decay time of


7

Fig. 1.4 The electron optics of the SEM built by Zworykin et al. (1942)

the phosphor, which was a problem important in later work at Cambridge (McMullan 1952). The optimum focus setting was found by maximizing the high frequency components in the video waveform observed on an oscilloscope, a method that was originally proposed by von Ardenne (1938b). Although the intention was to produce contrast by differences in the secondary emission ratio of the surface constituents, and the incident beam energy of 800 eV was chosen with this in mind, contamination of the surface in the rather poor vacuum prevented meaningful compositional contrast from being obtained. Surprisingly, Zworykin did not anticipate this, although he was an experienced vacuum physicist. Only two years earlier, secondary emission measurements and the effects of contamination had been published by Bruining and de Boer (1938). Actually, all of Zworykin’s published micrographs were of etched or abraded samples, and contrast was topographic (Zworykin et al. 1942)—for example, etched brass (see Fig. 1.5). The quality of the recorded images was rather disappointing, and together with the lack of a visible image, must have been a factor in RCA’s decision to discontinue the project. The main reason, however, was undoubtedly the excellent results that were, as mentioned earlier, being obtained with replicas viewed in the

8

D. McMullan

Fig. 1.5 Micrograph of etched brass produced by the SEM of Zworykin et al. (1942)

TEM (Mahl 1940). In any event, all available technical effort had to be directed to the highly successful RCA EMB TEM, which was just then coming into production (Reisner 1989).

Von Ardenne’s Scanning Electron Microscope While RCA was developing its SEM, Manfred von Ardenne (a private consultant who had his own laboratory) was developing the first scanning electron microscope with a submicron probe. In 1936, he was contracted by Siemens & Halske AG to investigate the possibility of using a scanned electron probe to avoid the effects of objective lens chromatic aberration with thick samples in TEM. In the course of this work, he laid the foundations of electron probe microscopy by making and publishing (von Ardenne 1938a,b) a detailed analysis of the design and performance of probe-forming electron optics using magnetic lenses. The analysis covered the limitations on probe diameter due to lens aberrations and the calculation of the current in the probe. He also showed how detectors should be placed for bright-field


9

and dark-field STEM and for imaging a solid sample in a SEM, and considered the effects of beam and amplifier noise on imaging. To fulfill the Siemens contract, von Ardenne built the first scanning transmission electron microscope (STEM) and demonstrated the formation of probes down to 4 nm in diameter. But in the short time available, he was limited to employing existing technology, and because there was no suitable low-noise electronic detector, he used photographic film—consequently there was no immediately visible image. A schematic of the microscope column is shown in Fig. 1.6. A demagnified image of the crossover of the electron gun was focused on the sample with two magnetic lenses, and X-Y deflection coils were mounted just above the second of these. Immediately below the sample, was a drum around which was wrapped the photographic film. The image was recorded by rotating the drum and simultaneously moving it laterally by means of a screw while the currents in the deflection coils were controlled by potentiometers mechanically coupled to the drum mechanism. The intensity of the beam was very low (about 10−13 A) and it was necessary to record the image over a period of about 20 minutes. Because the image was not visible until the film had been developed, focusing could only be accomplished indirectly by using the stationary probe to produce a shadow image of a small area of the sample on a single-crystal Zinc sulfide screen that was observed through an optical microscope and prism system. The recordings were inferior to those from the TEM that was being constructed by Ruska and von Borries at Siemens, and the hoped-for advantages of STEM with thick samples were not realized. Von Ardenne spent a short time trying to use the instrument in the SEM mode on bulk samples, but could only obtain low resolution images because of the detector problem. The sample current was amplified by thermionic tubes and a large probe current was needed. He did not publish any images. In total, von Ardenne worked for less than two years on scanning electron microscopy before concentrating on the development of his universal TEM (von Ardenne 1985). Then, with the start of World War II, he began work on a cyclotron and isotope separators for nuclear energy projects. If he had been able to continue, there is little doubt that he would have built an efficient SEM within a year or two: this is evidenced by a patent (von Ardenne 1937) that included a proposal for double-deflection scanning, two papers (von Ardenne 1938a,b), and a book (von Ardenne 1940). Two of the chapters in the book were on scanning microscopy and were based on the 1938 papers, but included additional material relating to imaging the surfaces of solid samples. Most importantly, he proposed a detector using an electron multiplier with beryllium copper dynodes (see Fig. 1.7) that could be opened to the atmosphere and worked with efficiently under poor vacuum conditions. Measurements of the secondary emitting ratio of beryllium copper and its stability when exposed to the atmosphere were otherwise only first reported in 1942 by I. Matthes (1942) of the AEG Research Institute in Berlin, but von Ardenne was probably aware of this research a year or two before. In his book, von Ardenne also discussed the interaction between the beam electrons and the sample, and suggested that back-scattering would cause a loss of resolution, illustrating this with a diagram that has quite a modern look

10

Fig. 1.6 Cross-section of the column of von Ardenne’s (1938b) STEM

D. McMullan


11

Fig. 1.7 Electron multiplier with beryllium copper dynodes proposed by von Ardenne (1940) as a secondary electron detector for SEM. The drawing shows the first three stages of the multiplier and its position relative to the objective lens and sample

(see Fig. 1.8). He argued that the incident beam electrons produce secondary electrons at or near the surface from an area approximately equal to the beam diameter, and give a high-resolution image (nutzbare Strahlung). The beam electrons penetrate the sample, and a proportion of them is backscattered and reach the surface where they produce further secondaries. These two signals are now generally referred to as SE-I and SE-II, respectively (Drescher et al. 1970, Peters 1982). The backscattered electrons are emitted from an area of diameter comparable to the penetration depth, and the secondaries they produce (schädliche Strahlung) may impair the resolution (he did not, however, consider the case of a sample with small inclusions below the surface). He concluded that good resolution might be obtained either with a very low-energy beam (1 keV), or with one having a high energy (50 keV). In the first case, the backscattered electrons would emerge from an area of the surface a little larger than the incident beam, and the resolution would be unaffected. On the other hand, the secondary electrons produced by the backscatter of a 50-keV beam would affect a much larger area, and would be evenly distributed so that their main effect would be to increase the background (reduce the contrast) rather than affect the resolution. Von Ardenne’s scanning microscope was destroyed in an air raid on Berlin in 1944, and after the war he did not resume his work in electron microscopy but researched in other fields—first in Russia and then in Dresden (in 1955), which was then in East Germany. Additional information about von Ardenne’s scientific work is available in his autobiography (von Ardenne 1972) and by McMullan (1988).

12

D. McMullan

Fig. 1.8 Diagram illustrating von Ardenne’s (1940) discussion of secondary electron imaging of a surface

The Cambridge Scanning Electron Microscopes Apart from a theoretical analysis of resolving power by a French author (Brachet 1946), no other substantial work on SEMs was reported until 1948, although recent research has revealed that some rather primitive experiments were done by A. Léauté (1946) at L’Ecole Polytechnique in Paris during World War II (Hawkes & McMullan 2004). In the 1940s, and for many years after, the feeling among most electron microscopists was that the SEM was not worth further consideration in view of the apparent failure at RCA—if such an experienced team could be that unsuccessful, it seemed very unlikely that anyone else could produce an effective instrument. A notable exception to this general opinion was that of Denis Gabor (1945). It was then that Charles Oatley at the Department of Engineering of the University of Cambridge decided to take another look at the SEM, although he related that “several experts expressed the view that this [the construction of an SEM] would be a complete waste of time” (Oatley et al. 1985). He explained at some length the reasons that brought him to this decision, but the main technological justifications were that “Zworykin and his collaborators had shown that the scanning principle


13

was basically sound and could give useful resolution in the examination of solid surfaces” and “improvements in electronic techniques and components had resulted from work during the war” (Oatley, 1982). He also felt that the RCA detector had a low efficiency and only a small proportion of the secondaries were reaching it, with the result that the images were noisy in spite of the long recording time. Independently of von Ardenne, he proposed to use an electron multiplier with beryllium-copper electrodes (Allen 1947), having been promised one by A.S. Baxter at the Cavendish Laboratory, who was making multipliers of this type (Baxter 1949). The full story of Oatley’s achievement is presented in a recent publication (Breton et al. 2004). I was selected by Oatley to build an SEM as a Ph.D. project—it was a challenging task because electron microscopy was a completely new subject for everyone in the laboratory, although I had had some experience in the radar and television industries, including the development and manufacture of cathode-ray tubes. I first completed a 40 keV electrostatically focused TEM that had been started by another PhD student, K.F. Sander. He abandoned it at an early stage and changed the subject of his research project to electron trajectory plotting (Sander 1951). I converted it to a STEM, and then to an SEM, by the addition of scan coils, an electron multiplier detector, and a long persistence cathode-ray tube monitor (McMullan 1952). It was not apparent how Zworykin’s results might be improved upon. A higherincident beam energy was expected to be beneficial, but it was not clear how image contrast would be formed. As mentioned earlier, Bruining and de Boer (1938) had shown that the secondary emission from a surface is critically dependent on the vacuum conditions, and it was plain that the achievable vacuum would not be good enough for there to be meaningful secondary-electron compositional contrast from a polished sample. Images of surfaces were obtained at grazing incidence and viewing direction (2 deg) in the TEM by Bodo von Borries (1940) and others, and it seemed probable that similar images could be produced in the SEM. I therefore mounted a sample of etched aluminium at a rather larger angle (30 deg; because the backscattered electrons did not have to be focused) and was rewarded by the now commonplace threedimensional appearance that is the hallmark of SEM images and a consequence of their large depth of focus. One of the first images—of etched aluminium—is shown in Fig. 1.9: (a) the direct view image (about 0.9-sec frame period), and (b) a 5-min recording. The beam energy was 16 keV and the resolution about 50 nm, limited by astigmatism in the objective lens and insufficient magnetic shielding. A block diagram of the SEM is shown in Fig. 1.10 (McMullan 1953). There was a relatively fast-scan, long-persistence cathode-ray tube display (405 lines, 1.8 fields/sec interlaced and a 5-min frame scan for photographic recording). Other features included a nonlinear amplifier for gamma control, and beam blanking for DC restoration. Double-deflection scanning coils were added later. The most important differences between this instrument and Zworykin’s were the much higher incident beam energy (15-20 keV), and the contrast produced mainly by scattered electrons from the tilted sample. The mechanism of contrast formation was investigated and shown to be topographic. No attempt was made to collect

14

D. McMullan

Fig. 1.9 One of the early images (etched aluminium surface) produced with SEM1. Angle of incidence of 16 keV electrons 25◦ : (a) visible image, 0.95 frames/s; beam current 1.5 × 10−10 A. (b) 5 min recording; 10−13 A. (McMullan 1952, 1953)

Fig. 1.10 Block schematic of SEM1. (McMullan 1952, 1953)


15

Fig. 1.11 Photograph of SEM1 taken in 1953 when K.C.A Smith took over

low-energy secondaries. In fact, I thought that they would be detrimental because of the inevitable contamination on the surface of the sample. I overlooked the increase in signal that is obtained from the low energy secondaries. I realized that there was another advantage in using a high-energy scanning beam: this was that in principle atomic number contrast was possible using backscattered electrons. An experimental curve of emission ratio (for 20 keV primaries) versus atomic number had recently been published by Palluel (1947), but an attempt at obtaining atomic number contrast failed. Some years later, Oliver Wells (1957) was more successful. The obvious disadvantage of high-beam energies was that the resolution was limited by penetration of the primary electrons. I suggested low-loss electron imaging to minimize this, but was not able to implement it (this was also done many years later by Wells, who published in 1971). One other contrast mechanism that I tried was cathodoluminescence, and I was able to demonstrate that phosphors with too long a decay constant to be used for producing images at a 0.9-sec frame time with a Zworykin-type detector were completely satisfactory when excited at a high-current density by a focused probe (Smith & Oatley 1955). Figure 1.11 is a photograph of the microscope (now named SEM1) taken in 1953 shortly before K.C.A. Smith assumed responsibility for it and turned it into an SEM that could produce images comparable with some of those from modern microscopes.

Further Development of the Microscope Smith introduced many improvements to SEM1, including a stigmator and a tilting sample stage, and he increased the efficiency of the detection system by moving the electron multiplier nearer the sample so that low-energy secondary electrons were

16

D. McMullan

collected, thus increasing the signal current. He showed that metalized insulating samples could be imaged, and he examined a wide variety of samples including germanium point-contact rectifiers and biological specimens. He also built an environmental cell for wet specimens (anticipating the environmental scanning electron microscope, ESEM); this had thin windows to admit the focused beam and allow the scattered electrons to reach the multiplier. Although the results were rather disappointing, it led Oatley to suggest replacing the second window with a shortdecay-time plastic scintillator and photomultiplier, and dispensing with the bulky electron multiplier (Smith 1956). This, in turn, resulted in the development of the Everhart and Thornley (1960) detector. Further SEMs were built in the engineering department at Cambridge: SEM2 (Wells 1957); SEM3 (Smith 1960); SEM4 (Stewart 1962); and SEM5 (Pease & Nixon 1965). All were used on a wide variety of samples and for the development of new techniques. Other important instrumental advances made by Oatley’s group during the remainder of the 1950s through to 1965 included: atomic number contrast (Wells 1957); stereomicroscopy (Wells 1960); voltage contrast (Oatley & Everhart 1957); low-voltage (1–2 keV) SEM (Thornley 1960a); high temperature (1200 ◦ C) imaging of thermionic cathodes in a SEM (Ahmed 1962); high-resolution (10 nm) SEM (Pease & Nixon 1965); etching of surfaces in a SEM by ion bombardment (Stewart 1962); ion etching and microfabrication in SEM (Broers 1965); and microelectronics in SEM (Chang 1966). Most of this work is described in papers by Oatley (1982) and Oatley et al. (1985).

Materials Research In 1955, Smith used SEM1 for the first three applications of a scanning electron microscope in materials research: 1. Smith was visited by F.P. Bowden, head of the Surface Physics Laboratory in the Department of Physical Chemistry at Cambridge University, and J. McAuslen of Imperial Chemical Industries, who brought a sample of silver azide crystals. The thermal decomposition of these was being investigated in their laboratory using a TEM in the reflection mode, but had failed due to the premature ignition of the crystals under the intense illumination that was needed. In the SEM, with the crystals mounted on a small hot-plate, the decomposition could be readily controlled and observed without difficulty (Bowden & McAuslan 1956; McAuslen & Smith 1956; see Fig. 1.12. 2. J.H. Mitchell, controller of research at Ericsson Telephones, Ltd. approached Oatley about their work on the etching of germanium surfaces and the emergence of edge dislocations. They wished to establish whether there were pits or raised areas of sublight-microscopic size. J.W. Allen brought specimens to Cambridge and the micrograph in Fig. 1.13 shows a feature produced by the etchant CP4 (a mixture of acids with a small amount of bromine). (Allen & Smith 1956). 3. The third event occurred when Dr D. Atack—then on sabbatical leave from the Pulp and Paper Research Institute of Canada (PPRIC) where he was director of


17

Fig. 1.12 Partial decomposition of a small needle of silver azide. (Smith 1956)

Fig. 1.13 Crystallographic feature on a germanium surface etched with CP4. (Allen & Smith, 1956)

18

D. McMullan

Fig. 1.14 Application of the SEM to the study of wood fibers: a surface of newsprint. (Smith 1956)

the Applied Physics Division—came and asked if he could try some of his pulp and paper samples in the SEM. In this case, a TEM had also been tried with unsatisfactory results (Page 1958). The experiments were highly successful, as shown in Fig. 1.14 (Smith 1956) and led to PPRIC purchasing a SEM (see next section).

The First Low Voltage Scanning Electron Microscope (LVSEM) Imaging In the present context, the work of R.F.M. Thornley on LVSEM in Oatley’s laboratory (Thornley 1960a) needs to be described in more detail. He investigated some of its possibilities, and by meticulous testing and modification was able to greatly improve the performance of the instrument he had been allocated (SEM2), which had been built and used by Wells (1957). This included eliminating the many sources of interference (mainly 50 Hz) he had identified and making alterations to various components—some quite minor—so that eventually he was able to obtain a probe diameter of 200 nm at 1 keV. This is, of course, ridiculously large by today’s standards, but for then it was a considerable achievement. His modified SEM2 was the first LVSEM (Thornley 1960a) of the modern type (strictly speaking, Zworykin’s (1942) was the first, but the low voltage was irrelevant to the rather poor images it was able to produce).


19

To quote from a paper Thornley presented at the European Regional Conference on Electron Microscopy in Delft in 1960 (Thornley 1960b): “Previous work at Cambridge has used electron accelerating voltages greater than 6 kV, limiting the use of the instrument to the examination of conducting surfaces in order to avoid the formation of charging artefacts. Insulators can be covered with a thin metallic film to overcome this restriction, but deformation while under observation is difficult because the conducting film cracks away from the substrate as shown by Wells (1957). If the beam voltage is reduced until the secondary emission coefficient of the specimen is equal to or greater than unity, the surface potential will be automatically stabilised at that of its surroundings, in this case, at earth potential. Under these conditions, the effective secondary emission coefficient of an insulator is unity, regardless of the angle of incidence of the primary beam and the picture contrast is controlled only by collector modulation. . . . .” “. . . . .[Fig. 1.15], of a fractured ceramic surface, shows that, with a suitable choice of collector position, contrast similar to that expected from oblique viewing at high voltage can be obtained at low voltages, in this case, 1.5 kV. Provided the stability requirements can be met, the ultimate resolution, for voltages above 500 V, is limited by the same factors as in the high voltage case to between 50 and 100 Å as shown by Everhart (1958). It has been found that the reduction in gun brightness at low voltages is largely compensated by the increase in secondary emission, so no change in recording time has been necessary. The micrograph shown was recorded over 2 min, using an instrument originally designed for 25 kV operation, but fitted with a modified gun, permitting operation down to 300 volts. Contrast due to surface films is enhanced at low voltages because differences in secondary emission coefficients are more pronounced and penetration effects are reduced, the range of a 500-V electron in aluminium being roughly 30Å. . . . . .”

Further work on ceramics was reported in a paper with L. Cartz of Morganite Research and Development Ltd. (Thornley & Cartz 1962):

Fig. 1.15 Imaging the surface of a sintering fault in an alumina ceramic with 1.5-keV electrons. (Thornley 1960)

20

D. McMullan “A direct electron-optical method of observing an insulator surface is described and applied to a series of alumina ceramics. The surface of the object requires no previous treatment of any kind, and a resolving power of 2,000 Å has been obtained with a depth of focus of about 50 μm. Different phases and components can be distinguished. Fractured surfaces, a fault region, and polished surfaces of various alumina ceramics are examined.”

Commercial Production of SEMs Following the encouraging results obtained using SEM1 to study wood fibers (Atack & Smith 1956), L.R. Thiesmeyer, director of the Canadian Pulp and Research Institute, ordered a fully engineered microscope (SEM3) from the Cambridge University Engineering Department. This was used for many years in their Montreal Laboratories and was the earliest industrial application of an SEM on a daily basis. Smith developed SEM3 and completed it in 1958: it was the first magnetically focused SEM (Smith 1959). The lower section of the column below the table consisted of a modified Metropolitan Vickers (later AEI)-type EM4 TEM (Page 1954) and contained the electron gun, condenser lens, transmission sample stage, objective lens, and double pole piece projector lens. For scanning operations, the transmission objective and projector were used together in various combinations and powers according to the spot diameter required to provide the first stage of spot demagnification. Immediately above the table, there was a section of the column containing the scanning coils and the objective lens that was of the pin-hole type (Liebmann 1955), with three adjustable apertures. There was a tilting sample stage and the Everhart-Thornley type of secondary electron detector. Thiesmeyer and Atack were among the very few who (at that time) saw the great potential of SEM. Although Oatley’s group had produced and published highquality micrographs from many different samples, there was still considerable resistance to SEM among microscopists. Over several years, Oatley expended much effort in trying to persuade electron microscope manufacturers to market an SEM (Jervis 1971, 1972; Oatley 1982; Breton et al. 2004) but he was only finally successful in 1962 when the Cambridge Instrument Company decided to go ahead with the production of an SEM based on the instruments developed by Oatley’s group. This decision was influenced by A.D.G. Stewart, one of Oatley’s students who agreed to join the company and later played a leading part in the development of the Stereoscan, as the new SEM was named (Stewart & Snelling 1965; Stewart 1985). The prototype Stereoscan (see Fig. 1.16) went to the Dupont Chemical Corporation in the United States in 1964, and in the following year the first two production models were sold to P.R. Thornton at the University of North Wales and to J. Sikorski at Leeds University in the United Kingdom, the third to G. Pfefferkorn at Münster University in Germany, and the fourth to the Central Electricity Laboratories in Leatherhead, United Kingdom. In the words of Professor Sir Charles Oatley (1982), “By this time the Company had launched a publicity campaign and orders began to roll in. An additional batch of twelve microscopes was put in hand; and then a further forty . . . . . . . the scanning microscope had come of age.”


21

Fig. 1.16 The Cambridge Instrument Company’s Stereoscan Mk 1 prototype (Stewart & Snelling 1965)

The first commercial competitor was the Japanese firm, JEOL, who marketed their JSM-1 SEM about six months later and were soon followed by others (McMullan 2004).

Other SEMs up to 1965 and Beyond SEM developments in other laboratories prior to 1965, as evidenced in scientific publications, included the following: • An SEM was built in France by Bernard & Davoine (1957) at the National Institute of Applied Science in Lyon. It had a probe size of the order of 1 μm and was used over a period of years mainly for cathodoluminescence studies. • AEI in the United Kingdom—the major TEM manufacturer at that time— developed an SEM but did not proceed after the first instrument, sold in 1959, turned out to be unsatisfactory (Jervis 1971, 1972). • Wells et al. (1965) built an advanced SEM for semiconductor studies and microfabrication for the Westinghouse Laboratories in Pittsburgh, Pennsylvania, and demonstrated EBIC imaging.

22

D. McMullan

• In the USSR, there was a SEM at Moscow University as early as 1960 (Kushnir et al. 1961). • There were other groups, especially in Japan, who did not publish at the time (Fujita 1986). The year 1965 marked the beginning of the general use of scanning electron microscopy, but there were other developments coming to fruition at the time that had a great importance for low-voltage SEM. These were in ultra-high vacuum technology that enabled vacuum systems that were capable of reaching 10−8 Pa or less to be easily put together using standard components: in particular, the Conflat flange (Wheeler & Carlson 1962), all-metal vacuum valves (Wheeler 1976), and the development of the sputter-ion pump, starting with Hall (1958). Therefore, just about the time that the SEM was becoming generally used with the Stereoscan and the JEOL JSM-1, it was becoming possible to realize Zworykin’s (1942) initial concept of a SEM with a low-voltage scanning beam (800V), a cold field-emission gun cathode, secondary emission contrast from a polished sample, and a fast scan. It seems probable that the success of the Stereoscan-type SEMs had the initial effect that electron microscopists were so occupied in using the new instruments that there was no immediate interest in the possibilities of still more powerful techniques. Therefore, the application of UHV to microscopy was initially confined to the development of the STEM, particularly by Albert Crewe and his coworkers at the Argonne Laboratory (Crewe 1966; Crewe et al. 1968). The first UHV, low-voltage SEM with field-emission gun was described by Welter and Coates (1974), who had earlier collaborated with Crewe.

References Ahmed, H. (1962) “Studies on high current density thermionic cathodes”. PhD Dissertation, University of Cambridge. Allen, J.S. (1947) An improved electron multiplier particle counter. Rev. Sci. Instrum. 19, 739–749. Allen, J.W. and Smith, K.C.A. (1956) Electron microscopy of etched germanium surfaces. J. Electronics 1, 439–443. D. Atack and K. C. A. Smith, (1956) “The scanning electron microscope. A new tool in fiber technology,” Pulp Pap. Mag. Can. (Convention issue) 57, 245–251. Bain, A. (1843) Electric time-pieces and telegraphs. British Patent No 9745, filed 27 May 1843. Baxter, A.S. (1949) “Detection and analysis of low-energy disintegration particles.” Ph.D. Dissertation, University of Cambridge. Bernard, R. and Davoine, F. (1957) The scanning electron microscope (in French). Ann. Univ. Lyon Sci. Sect. B[3] 10, 78–86. Bowden, F.P. and McAuslan, J.H.L. (1956) Slow decomposition of explosive crystals. Nature 178, 408–410. Brachet, C. (1946) Note on the resolution of the scanning electron microscope (in French). Bull. Assoc. Tech. Marit. Aeronaut. 45, 369–378. Breton, B., McMullan, D. and Smith, K.C.A. (eds, 2004). “Sir Charles Oatley and the scanning electron microscope”, Adv. Imaging Electron Phys. 133 (P.W. Hawkes, editor-in-chief), Elsevier Academic Press: San Diego, London. Broers, A.N. (1965) “Selective ion beam etching in the scanning electron microscope”. PhD Dissertation, University of Cambridge.


23

Bruining, H. and de Boer, J.H. (1938) Secondary emission. Physica (Amsterdam) 5, 17–30. Burnett, C.E., (1938) The Monoscope. RCA Review 2, 414–420. Chang, T.H.P. (1967) “Combined micro-miniature processing and microscopy using a scanning electron probe system”. PhD Dissertation, University of Cambridge. Crewe, A.V. (1966) Scanning electron microscopes: is high-resolution possible? Science 154, 729–738. Crewe, A.V., Eggenberger, D.N., Wall, J. and Welter, L.M. (1968) Electron gun using a fieldemission source. Rev. Sci. Instrum. 39, 576–583. Drescher, H., Reimer, L., and Seidel, H. (1970) Back-scattering coefficient and secondary electron yield from 10 - 100 keV electrons in the scanning electron microscope (in German). Z. angew. Phys. 29, 331. Everhart T.E. and Thornley R.F.M. (1960) Wide-band detector for micro-microampere low-energy electron currents. J. Sci. Inst. 37, 246–248 (1960). Fujita, H. (1986) “The History of Electron Microscopes”, 11th International Congress on Electron Microscopy: Kyoto, Japan. pp. 187–193. Gabor, D. (1945) “The Electron Microscope”. Hulton Press: London. Hall, L.D. (1958) Electronic ultra-high vacuum pump. Rev. Sci. Instrum. 29, 367–370. Hawkes, P.W. and McMullan, D. (2004) A forgotten French scanning electron microscope and a forgotten text on electron optics. Proc. Roy. Microsc. Soc. 39, 285–290. Jervis, P. (1971/72) Innovation in electron-optical instruments – two British case histories. Research Policy 1, 174–207. Knoll, M. (1935) Static potential and secondary emission of bodies under electron irradiation (in German). Z. tech. Phys. 16, 467–475. Knoll, M. (1941) Detection of attached oxide layers on iron with the scanning electron microscope (in German). Phys. Z. 42, 120–122. Kushnir,Yu.M., Fetisov, D.V. and Raspletin, K.K. (1961) Scanning electron microscope and X-ray microanalyser. Bull. Acad. Sci. USSR. Phys. Ser. (Engl. Transl) 25, 709–714. Léauté, L.[A.] (1946) Applications of the electron microscope in metallurgy (in French). In “L’Optique Electronique” (L. de Broglie ed.) Editions de la Revue d’Optique Théorique et Expérimental: Paris, 1946, 209–220. Liebmann, G. (1955) The magnetic pinhole electron lens. Proc. Phys. Soc. Ser. B 68, 682–685. Mahl, H. (1940) Supermicroscopic determination of the orientation of single aluminium crystals (in German). Metallwirtschaft 19, 1082–1085. Matthes, I. (1942) Investigation of the secondary electron emission from various alloys (in German).. Z. tech. Phys. 22, 232–236. McAuslan, J.H.L. and Smith, K.C.A. (1956) The direct observation in the scanning electron microscope of reactions, in Electron Microscopy: Proceedings of the Stockholm Conference, Sept. 1956, edited by F. S. Sjostrand and J. Rhodin (Academic, New York 1957), pp. 343–345. McMullan, D.(1952) “Investigations relating to the design of electron microscopes”. Ph.D. Dissertation, Cambridge University. McMullan, D. (1953) An improved scanning electron microscope for opaque specimens. Proc. Inst. Electr. Engrs. 100, Part II, 245–259. McMullan, D. (1988) Von Ardenne and the scanning electron microscope. Proc. Roy. Microsc. Soc. 23, 283–288. McMullan, D. (1990) The prehistory of scanned image microscopy, Part 1: scanned optical microscopes. Proc. Roy. Microsc. Soc. 25, 127–131. McMullan, D. (1995) Scanning electron microscopy 1928 – 1965. Scanning 17, 175–185. McMullan, D. (2004) A history of the scanning electron microscope, 1928 – 1965. Adv. Imaging Electron Phys. 133, 523–545. Morton, G.A. and Ramberg, E.G. (1939) Point projector electron microscope. Phys. Rev. 56, 705. Mulvey, T. (1962) Origins and historical development of the electron microscope. Brit. J. Appl. Phy. 13, 197–207.

24

D. McMullan

Oatley, C.W. (1982) The early history of the scanning electron microscope. J. Appl. Phys. 53, R1-R13. Oatley, C.W. and Everhart, T.E. (1957) The examination of p-n junctions with the scanning electron microscope. J. Electron. 2, 568–570. Oatley, C.W., McMullan, D., and Smith, K.C.A. (1985) The development of the scanning electron microscope. in “The Beginnings of Electron Microscopy” (P.W. Hawkes ed.) Adv. Electronics Electron Phys. Suppl. 16, 443–482. Page, D.H. (1958) Reflexion electron microscopy at high angles. Brit. J. Appl. Phys. 9, 60–67 Page, R.S. (1954) A compact console-type electron microscope. J. Sci. Instrum. 31, 200–205. Palluel, P. (1947) Backscattered components of electron secondary emission from metals (in French) C.R. Acad. Sci. 224, 1492–1494. Pease, R. F. W. and Nixon W.C. (1965) High-resolution scanning electron microscopy. J . Sci. Instrum. 42, 31–35. Peters, K-R. (1982) Generation, collection and properties of an SE-1 enriched signal suitable for high-resolution SEM on bulk specimens, in “Electron Beam Interactions with Solids”, SEM (Inc), Chicago, pp 363–372. Reisner, J.H. (1989) An early history of the electron microscope in the United States. Adv. Electronics Electron Phys. 73, 134–231. Ruska. E. (1933) The electron microscopic imaging of surfaces irradiated with electrons (in German). Z . Phys. 83, 492–497. Ruska E, and Müller, H.O. (1940) Progress on the imaging of electron irradiated surfaces (in German). Z Phys 116, 366–369. Sander, K.F. (1951) “An automatic electron trajectory tracer and contributions to the design of an electrostatic electron microscope”. PhD Dissertation, University of Cambridge. Smith, K. C. A. (1956) “The scanning electron microscope and its fields of application”. Ph.D. Dissertation, University of Cambridge. Smith, K. C. A. (1959) Scanning electron microscopy in pulp and paper research. Pulp Pap. Mag. Can. 60, T366-T371. Smith, K. C. A. (1960) A versatile scanning electron microscope, in The Proceedings of the European Regional Conference in Electron Microscopy, Delft, 29 August–3 September 1960 (Houwink, A.I. and Spit, B.J. eds.; Nederlandse Vereniging voor Elektronenmicroscopie, Delft n.d.) pp. 177–180. Smith, K. C. A. and Oatley, C. W. (1955) The scanning electron microscope and its fields of application. Br. J. Appl. Phys. 6, 391–399. Stewart, A.D.G. (1962) Investigation of the topography of ion bombarded surfaces with a scanning electron microscope, in Electron Microscopy, Fifth International Congress for Electron Microscopy, Philadelphia, Pennsylvania, 29 August – 5 September, 1962 (Breeze, S.S., ed.; Academic Press, New York, 1962) pp. D12-D13. Stewart, A.D.G. (1985) The origins and development of scanning electron microscopy. J. Microsc. 139, 121–127. Stewart, A.D.G. and Snelling, M.A. (1965) A new scanning electron microscope, in Electron Microscopy 1964, Proceedings of the Third European Regional Conference, Prague, 26 August – 3 September 1964 (Titlbach. M. ed.: Publishing House of the Czechoslovak Academy of Sciences: Prague) pp. 55–56. Stintzing, H. (1929) Method and device for automatically assessing, measuring and counting particles of any type, shape and size (in German). German Patents Nos 485155–6. Synge, E.H. (1928) A suggested method for extending microscopic resolution into the ultramicroscopic region. Phil. Mag. 6, 356–362. Synge, E.H.(1932) An application of piezo-electricity to microscopy. Phil. Mag. 13, 297–300. Telefunken A G (1935) Improvements in or relating to cathode-ray tube picture transmitters. British Patent No. 465715 (Convention Date (Germany) Oct. 3 1935). Thornley, R.F.M. (1960) “New applications of the electron microscope”. PhD Dissertaion, Universty of Cambridge. Thornley, R.F.M. (1960) Recent developments in scanning electron microscopy, in The Proceedings of the European Regional Conference in Electron Microscopy, Delft, 29 August–3


25

September 1960 (Houwink, A.I. and Spit, B.J. eds.; Nederlandse Vereniging voor Elektronenmicroscopie, Delft n.d.) pp. 173–176. Thornley R.F.M. and Cartz L. (1962) Direct examination of ceramic surfaces with the scanning electron microscope. J. Am. Ceram. Soc. 45, 425–428. von Ardenne, M.(1937a) Improvements in electron microscopes. British Patent No 511204, convention date (Germany) 18 Feb. von Ardenne, M. (1938a) The scanning electron microscope. Theoretical fundamentals (in German). Z. Phys. 109, 553–572. von Ardenne, M. (1938b) The scanning electron microscope. Practical construction (in German). Z . tech. Phys. 19, 407–416. von Ardenne, M. (1940) “Electron Microscopy” (in German). Springer Verlag: Berlin . von Ardenne, M. (1972) “A Happy Life in Engineering and Research” (in German) Kinder Verlag: Munich and Zurich. von Ardenne, M. (1985) On the history of scanning electron microscopy, the electron microprobe, and early contributions to transmission electron microscopy. in “The Beginnings of Electron Microscopy” (PW Hawkes ed), Adv. Electronics Electron Phys. Suppl. 16, 1–21. von Borries, B. (1940) High-resolution images from the electron microscope used in reflection (in German). Z. Phys. 116, 370–378. Wells, O.C. (1957). “The construction of a scanning electron microscope and its application to the study of fibres”. Ph.D. Dissertation, Cambridge University. Wells O.C. (1960) Correction of errors in stereomicroscopy. Br. J. Appl. Phys. 11, 199–201. Wells, O. C. (1971) Low-loss image for surface scanning electron microscope. Appl. Phys. Lett. 19, 232–235. Wells, O.C., Everhart, T.E., and Matta, R.K.(1965) Automatic positioning of device electrodes using the scanning electron microscope. IEEE. Trans. Electron. Dev. ED-12, 556–563. Welter, L.M. and Coates, V.J. (1974) High-resolution scanning electron microscopy at low accelerating voltages. Proc. 7th Ann. SEM Symposium, IIT Research Institute, Chicago. (O. Johari ed.) pp. 59–66. Wheeler, W.R. (1976) Recent developments in metal-sealed gate valves. J. Vac. Sci. Technol. 13, 503–506. Wheeler, W.R. and Carlson, M.A. (1962) Ultra-high vacuum flanges. Trans. AVS Nat. Vac. Symp. 1961, p. 1309–1318, Pergamon Press: Oxford Zworykin, V.A. (1934) Electric Microscope. 1st Congresso Internazionale di Electroradiobiologia 1, pp 672–686. Zworykin, V.A., Hillier, J., and Snyder, R.L. (1942) A scanning electron microscope. ASTM. Bull. 117, 15–23; (Abstract) Proc. Inst. Radio Engrs. 30, 255. Zworykin, V.A., Morton, G.A., Ramberg, E.G., Hillier, J. and Vance, A.W. (1945) “Electron Optics and the Electron Microscope”, Wiley, New York.

Chapter 2

LVSEM for Biology James B. Pawley

Key words: low voltage, high-resolution, scanning electron microscopy, radiation damage

Introduction1 Two Approaches to Microscopical Imaging Early methods of microscopical imaging involved the use of lenses to focus and magnify the pattern of light transmitted, refracted, or reflected by the specimen. Contrast in the final image depended on the extent to which the features of the specimen absorbed, refracted, or reflected the light. Most early methods of transmission electron microscopy (TEM) also followed this approach in that the pattern of transmitted electrons emerging from the far side of the specimen was focused by appropriate lenses to form the final image. In 1935, however, Irwin Knoll pioneered a new approach where the properties of the specimen were not imaged directly in space but were instead sampled in time by a small beam of electrons that sequentially illuminated one point on the object at a time. The final image was built up from a time-sequence of data, and was displayed by a second electron beam on a cathode ray tube (CRT). The two beams swept in synchrony in a rectangular pattern, or raster, over both the specimen and the CRT. The brightness of the beam in the CRT was made proportional to the intensity of some signal generated by the beam striking the specimen, and the magnification was the ratio of the dimensions of the two rasters (Knoll 1935) (see Fig. 2.1). Although few microscopical sampling methods could be implemented in Knoll’s time, the sampling approach embodied in this second type of microscope has many potential advantages. To produce an image or map, the results of an interaction between the beam and the specimen need only be detectable, rather than focusable (i.e., the electron microscope (EM) specimen no longer has to be thin enough to 1

This chapter relies to a considerable extent on Pawley (1992).

H. Schatten, J. B. Pawley (eds.), Biological Low-Voltage Scanning C Springer 2008 Electron Microscopy.

27

28

J. B. Pawley

Fig. 2.1 Knoll’s diagram of a scanning-type microscopical imaging system (Knoll 1935)

transmit the impinging electrons.) In addition, because the image information is carried as a time-varying electronic signal, a variety of analog and digital signal processing procedures can easily be applied to this signal to emphasize the particular aspects of it that are of interest to the viewer (i.e., contrast can be arbitrarily manipulated electronically). Unfortunately, Knoll’s instrument was really more like a video image sensor than a microscope. It operated only at very low magnification and, as discussed in more detail in Chapter 1 (this volume), the performance of a more advanced design by von Ardenne (von Ardenne 1938) was also limited by the capabilities of the electronics of the period. In Knoll’s instrument, the scanning transmitted electron signal was recorded directly on a mechanically-scanned photographic plate, and the complexity of the mechanism needed to scan the plate in synchrony with the beam effectively limited application of the scanning approach to transmission electron microscopy for some time. Scanned electronic imaging was instead applied to the field of television because there the time-sequential nature of the electronic signal that is present between the imaging tube and the display CRT greatly simplifies its widespread dissemination by broadcasting (McMullan 1990).

The Rise of the Modern Surface-imaging SEM In the 1950’s, the scanning approach to electron microscopy was rejuvenated by the Cambridge group under Oatley (Oatley 1972, 1982; McMullan 1953a, 1953b; Wells 1975). This group developed improved detectors for secondary electrons (SE) and backscattered electrons (BSE) (Everhart & Thornley 1960; Wells & Oatley 1959) and also benefited from wartime improvements in the performance of the

2 LVSEM for Biology

29

electronics used for scanning the beam in a rectangular raster and for the electronic display and recording of the detected signals as images. The SEMs developed and used world-wide since that time owe most of their important features to early developments by this group (Oatley et al. 1965).

Electrons as Probes in Scanning Microscopes Compared to light, x-rays, or other elementary particles, electrons are perhaps the ideal excitation source for scanned probe microscopic imaging. Four of their important features include: • Monoenergetic sources of high specific brightness (quanta/cm2 /ster) are readily available and easy to maintain (Oatley 1975; Hainfeld 1977) • The electron wavelength is very short and available lenses can focus electrons into a Gaussian probe as small as 0.5–2 nm in diameter depending on beam voltage • The charge on the electron makes it possible to use electromagnetic fields to scan the probe over the surface of the specimen rapidly and accurately • Energetic electrons striking a solid specimen surface are capable of exciting a wide variety of detectable signals (Everhart et al. 1959; Clarke 1970). The resulting signals include electrons produced by secondary emission (SE) (Everhart et al. 1959), auger emission (AE) (MacDonald 1971; Gerlach & MacDonald 1976), electron channeling (Coates 1969; LeGressus et al. 1983), and backscattering (BSE) (Ball & McCartney 1981), as well as characteristic and continuum x-rays (Duncumb 1957; Newbury et al. 1988; Statham 1988), electron-hole pairs (Breese 1982), light (Jakubowicz 1987) and heat (measured as sound) (Rosencwaig 1982). Because these signals are only elicited from the area of the specimen immediately under the beam, they need only be collected (rather than focused) to produce an image or map of some aspect of the material producing the interaction. The spatial resolution of the map will depend on the size of the probe/specimen interaction volume. Though all of these interactions and others have been used to produce useful SEM images, it is fair to say that images using either the SE or the BSE signal constitute the vast majority of recorded images. The reasons for this are: • SE and BSE can be both produced and detected with high quantum efficiency • On many specimens, the amount of SE signal collected from each point is roughly proportional to the angle between the viewing direction and the normal surface. Such an image conveys a fairly accurate impression of surface topography to the human brain. (Everhart et al. 1959; Wells & Oatley 1959) • In favorable circumstances, much of the SE or BSE signal can be generated from the volume of the specimen immediately adjacent to the beam impact point. Consequently, given a small beam, these signals have the potential for transmitting structural information with high spatial resolution

30

J. B. Pawley

In the case of the SE signal, specimen topography produces variations in the detected signal that closely mimic those that produce changes in the apparent brightness of a macroscopic surface with the same relative shape and illuminated with diffuse light coming from the direction of the SE collector. As a result, an SE image of a rough microscopic specimen can be easily and accurately interpreted in terms of its topographic shape (Hayes 1980). The total BSE signal is a strong function of the density of the specimen under the beam, and BSE images of flat specimens are therefore primarily two-dimensional maps of material density versus position (Boyde 2003; Ferguson et al. 2003). Because of the predominant importance of these signals, the remainder of this chapter will concentrate on those aspects of SEM design and operation that affect the contrast and resolution of the SE and BSE images (Chapter 3 contains a comprehensive discussion of SEM resolution.).

Limitations Associated with the Use of Electrons as the Probing Radiation Electrons are in many ways an ideal excitation source for scanned probe microscopy, but they also have some disadvantages that are not associated with other focusable quanta such as ions (Levi-Setti et al. 1984; Wang et al. 1989), light photons (Pawley 2006), or x-rays (Cheng & Jan 1987; Atwood & Barton 1989). These disadvantages are generally so well-known and accepted, however, that their existence often passes without comment—they are perhaps worth considering here more explicitly. The rationale for this is that success in using the SEM to image complex, organic surfaces will, in large part, depend on the ability of the researcher to avoid or ameliorate the effects of these disadvantages. It is also true that many of these effects vary strongly with beam voltage (Vo ) and, as a result, this parameter becomes the major determinant of SEM performance (Joy 1984, 1985, 1991b; Joy & Pawley 1993). The Characteristics of Electron Lenses All available electron lenses are converging and hence, in practice, the effect of lens aberrations can only be limited by reducing the lens aperture angle, α. The dominant lens aberrations are chromatic and spherical2 . The former produces a blurred spot of diameter: dc = Cc α

V0 V0

(2.1)

where Cc is the chromatic aberration coefficient and V0 is the voltage spread of the beam. Spherical aberration produces a blurring: 2 See Chapter 3, for a description of methods to correct spherical and chromatic aberrations in electron lenses and of the limits on depth-of-focus that then become paramount.

2 LVSEM for Biology

31

ds =

1 Cs α3 2

(2.2)

where Cs is the spherical aberration coefficient. Both Cc and Cs are lengths on the order of the focal length of the electron lens.3 Although the de Broglie wavelengths (λ) of 1–30 keV electrons are short (39 to 7 pm) compared to those of light (400–700 nm), diffraction remains a serious limitation because α must be kept very small to reduce the effects of aberrations. This is especially true for beam energies in the lower part of this energy range. Blurring due to diffraction is: dd = 0.6α/λ

(2.3)

The Intensity of Electron Sources Because electron lenses must be used at small acceptance angles (α = 10−2 −10−3 radians), it is not always easy to provide enough current in the beam to produce a well-defined image in a reasonable length of time. The current density in the focused spot can never be greater than that it is at the source. Early sources consisted of a heated tungsten hairpin and could produce a maximum useable intensity of only about 10 A/cm2 (Oatley 1975; Ohshita et al. 1978) or, assuming perfect optics, about 10−12 A in a 10 nm probe. This current corresponds to 6×106 electrons/sec, and if we imagine an image made up of 1,000 lines, each with 1,000 picture elements (pixels) that are scanned in one second, this corresponds to an average of only 6 electrons/pixel. The actual number is governed by Poisson statistics and is there√ fore 6 ± 6. Clearly, only a very high-contrast specimen can be imaged at all under these conditions, and even then only defined by two or three gray levels. In fact, the contrast of small surface features imaged with SEs is often only a few percent, and the current density in the spot may be reduced well below that at the cathode by practical and theoretical considerations. Under these conditions, small features can only be detected by producing more signal (i.e., 104 –105 quanta/pixel). This can be done either by scanning much more slowly or by using higher brightness electron sources to provide more beam current, thereby improving the statistical accuracy of the brightness measurement in each pixel (Wells 1975, 1978; Pawley 1990, see also Chapter 4) for a more comprehensive discussion of the effect of contrast and statistics on resolution.

The Effects of the Electron Charge Although the fact that the electron has a charge simplifies the process of focusing the probe and of scanning it in a raster, it complicates SEM observations of nonconductive specimens. As the total yield of BSE, plus SE, per beam electron is usually 3

See Chapter 3, for a discussion of aberration correctors as used for SEM.

32

J. B. Pawley

less than 100% at V0 > 5kV, negative charge accumulates within the scanned area of bulk specimens (i.e., those thicker than the electron range). Fields associated with this charge can defocus or deflect the beam and interfere with the collection of low energy ( SE Conversion Plate

Low voltage BSE detector

Concentric YAG BSE detector for high-kV SE detector Ex B Upper polepiece

Specimen

Objective lens

Lower polepiece

Brightfield/STEM aperture STEM detector (b)

Fig. 2.13 (continued)

2 LVSEM for Biology

55

(c)

(d)

Fig. 2.13 Modern field-emission SEM performance. (a) Shows the Hitachi S-5200. The column of the newer S-5500 is entirely contained in a rectangular box that provides shielding from sound and electromagnetic fields. (b) Both instruments offer a variety of SE and BSE detectors optimized for use at low and high beam voltage. A set of coils (ExB) mounted above the specimen, provides a magnetic field perpendicular to the electron-optical axis that compensates for the displacement produced by the electrostatic SE-collection field of the Everhart-Thornley detector. (c) A highmagnification image of a gold-on-solid-carbon substrate made at 1 kV using a Hitachi 5200. (d) An image of a similar specimen made at a beam voltage of 500v using a Hitachi S-4800 SEM. The S-4800 has a normal SEM specimen chamber mounted entirely below the objective lens. (All the images in this figure were generously provided by Hideo Naito and Vinh Van Ngo on behalf of Hitachi Inc, Pleasanton, CA.)

56

J. B. Pawley

vs 5 × 10−9 pascal), but the significance of this difference is somewhat offset by the fact that the TF gun operates hot, and the increased outgassing associated with this fact makes the vacuum conditions for stable TF emission almost as difficult to meet as those for FE. On theoretical grounds, performance should be quite similar for beam currents around 10−9 A, with TF being more suited to higher currents and FE capable of a smaller probe size when smaller currents are sufficient. No critical comparisons of actual, comparable, high-resolution, TF and FE instruments have yet been published, but one difference that may be important is total tip current. TF operates with a total current from the tip (Ig ) of 100–500 μA, while FE usually operates at Ig = 5–10 μA. The charge-charge interactions within the beam that cause increases in the effective V0 (Boersch 1954; Pfeiffer 1972; Barth et al. 1990) are proportional to J2g , where Jg refers to the current density in the first crossover. If the gun is designed and operated in such a way that Jg is higher in the TF system, this will reduce performance at low-kV where V0 has such a strong effect on beam diameter (Shao & Crewe 1987). Aside from different guns, modern, high-resolution FE-SEMs have important differences in the design of their electron optics, as well as that of their vacuum systems, specimen stages, and digital control/display electronics. The author’s experience is limited to the S-900, which was the first to be introduced and the following discussion will reflect this bias. The S-900 at the Integrated Microscopy Resource was the first commercial, inlens, FE-LVSEM to be installed anywhere in the world. Because it permitted one to obtain ∼5x better resolution on biological specimens than had been possible with any previous machines, our early work revealed a number of problems, both in the operation of the instrument and in the preparation of specimens. Although much of this chapter concentrates on these problems, the reader is reminded that almost all of the instrumental problems have now been corrected on later microscopes. Likewise, the improvements in specimen preparation described here and in the other chapters allow one to reach ever closer to the potential of the technique for revealing the three-dimensional intricacies of biological structure.

The SE Performance of Early FE-SEMs at Low Vo The S-900 at the Integrated Microscopy Resource was the first in-lens, cold FE-SEM to be delivered. Consequently, a number of modifications were needed to optimize its LVSEM performance (Pawley 1990). At first, problems with the vacuum system had the effect that a surface viewed at room temperature was often obscured by contamination after a single scan. Figure 2.14 shows two images of the surface of a specimen made by depositing fibrinogen molecules onto a carbon foil attached to an EM grid. Figure 2.14b was recorded first at 100 kx, 1.5 kV, and then Fig. 2.14a was recorded at 50 kx, and the result was magnified x2. Only the top left corner of the field of view is shown, but one can easily identify the three blobs present in both figures and also see how much clearer the image looks away from the area first scanned at 100 kx.

2 LVSEM for Biology

57

Fig. 2.14 Contamination obscures macromolecular structure. These two images were taken of the same area of a specimen made by depositing fibrinogen molecules onto a thin carbon film on a EM grid. After freeze-drying and coating, the specimen was imaged at 100 kx, 1.5 kV in the Hitachi S-900, before the vacuum modifications noted in Fig. 2.16. Immediately afterwards, a second 100 sec scan was recorded at 50 kx. (a) Shows a corner of the second image. The lighter area in the lower right corner is the area covered with contamination during the first scan. (b) Shows the small area of the first 100 kx scan that covered the same area as the box in (a). Although the darker area of (a) provides a noisier image (because the same 100 s exposure time was used to cover a 4x larger area of the specimen), the contrast of the fibrinogen molecules is higher because they have not been so covered with contamination

In terms of the vacuum environment, it is helpful to remember that materials present at a partial pressure of 10−4 Pa will deposit approximately one monolayer per second onto any adjacent solid surface. Even though microscope columns commonly operate at a much higher vacuum, this is not always true of the airlock vacuum. A rough-pumping oil with a vapor pressure of only 10−7 Pa will still coat all airlock surfaces in an hour or two. To solve this problem, we changed over to oil-free vacuum pumping throughout the system, as is shown in Fig. 2.15. The mechanical rough pumps were replaced with oil-free molecular-drag pumps (Danielson 1987), backed by diaphragm pumps, and an additional Gatan anti-contaminator was added. The sliding O-ring that sealed the side-entry stage rod to the inner side of the differentially-pumped airlock region was replaced with an oil-free seal made of spring-loaded teflon. Other changes involved efforts to reduce the effect of internal mains-frequency stray magnetic field (Pawley 1987a) and modifications to the control circuits to simplify the alignment process. As noted above, the side-mounted, Everhart-Thornley SE detector produces a collection field that displaces the probing beam before it reaches the final lens (Zach & Rose 1986) (see Fig. 2.10). As a result, the beam enters the lens field off axis (Pawley 1990). Because the displacement is proportional to Vo , realignment is required more often and this process is assisted if the controls for aligning the effective axis of the stigmator are readily accessible and if all the lens and stigmator currents can be wobbled. Many of these minor changes are now standard features on present commercial instruments. Once they are implemented, an image similar to that in Fig. 2.13b

340 l /s turbo

LN2

10-4

T

mBar

buffer

control

PM

20 l/s

20 l/s

MDPI

tc

buffer ωL > ωM . Electron-excited x-ray spectra also contain a significant x-ray background under all characteristic peaks. This background is the continuous x-radiation created during deceleration of the beam electron within the Coulombic field of the atoms. The

11 Developments in Instrumentation for Microanalysis in LVSEM

265

energy lost from the beam electron in one of these deceleration events is converted into an x-ray photon whose energy can take on a continuum of values from the lower threshold (arbitrarily 100 eV) up to the incident beam energy E0 . The total energy carried by an incident electron can be converted to a photon in a single event, creating the Duane-Hunt limit. The intensity of the x-ray continuum generated at a particular energy Eν follows the relation: Icm ∼ Z(E0 − Eν )/Eν

(11.3a)

Considering the continuum that creates the background directly under a characteristic peak, then Eν = Ech . Although Ech < Ec for a given element, we can approximate U = E0 /Ec as E0 / Eν . With this approximation, Equation (11.3a) can be rewritten as: Icm ∼ Z(E0 − Eν )/Eν ∼ Z(U − 1)

(11.3b)

The spectral peak-to-background ratio can then be obtained from Equations (11.2) and (11.3b): P/B = Ich /Icm = (U − 1)n /Z(U − 1) = (1/Z)(U − 1)n−1

(11.4)

For a typical value of n = 1.5, the P/B thus increases slowly as the 0.5 power of (U-1). Critical aspects of analytical x-ray spectrometry depend strongly on P/B and thus upon U. For example, our ability to detect the presence of elements at various concentration levels depends strongly upon U. For this discussion, the concentration, C, categories major, minor, and trace will be defined as: Major: C > 0.1 mass fraction Minor: 0.01 ≤ C ≤ 0.1 Trace: C < 0.01 While it is possible to detect major constituents as long as the overvoltage U > 1, providing sufficient time is spent to accumulate characteristic peak counts above the background, U = 1.1 represents a practical minimum for the acceptable overvoltage to achieve results when long-time expenditures are possible (>1,000 seconds). For a more reasonable expenditure of time (e.g., 100 seconds), in conventional analytical practice it is generally desirable to select E0 so that U ≥ 2 for the highest value of Ec in the suite of elements to be studied. For measurement of minor and trace elements, a high overvoltage is necessary to achieve adequate P/B for detectability because of the reduced exponent given by Equation (11.4). Thus, careful attention must be paid to the choice of beam energy and overvoltage when developing the analytical strategy to solve a particular problem. Ec for a given shell, e.g., the K-shell, depends upon atomic number, ranging from EK = 54.75 eV for Li, the lowest atomic number element that produces an x-ray, to EK = 115,600 eV for U. The high values of EK

266

D. E. Newbury Table 11.1 X-ray shell choices for conventional beam energy x-ray microanalysis K-shell Z = 4 Be (EK = 116 eV) to Z = 36 Kr (EK = 14323 eV) L-shell Z = 23 V (EL = 512 eV) to Z = 83 Bi (EL = 13424 eV) M-shell Z = 58 Ce (EM = 883 eV) to Z = 92 U (EM = 3551 eV)

that occur for intermediate and high atomic number elements force the analyst to choose a lower energy shell, such as the L- or M-shell, for those elements to give access to a value of Ec that can be excited with the available beam energy of the SEM (typically a maximum of E0 = 30 keV). Table 11.1 presents the elemental shell assignments chosen for conventional analytical practice to measure the naturally occurring elements with a minimum overvoltage of U = 2 (beam energy selected at the maximum of the conventional operational range 10 ≤ E0 ≤ 30 keV). Note that this selection of shells gives a comfortable range of overlap since several elements can be measured with two shells: e.g., Cu from the K and L shells and Ta from the L and M shells, if the beam energy is 20 keV or higher. Figure 11.1 depicts a Periodic Table shaded to indicate typical shell choices for conventional microanalysis conditions. The sampling volume of electron-excited x-ray spectrometry depends upon the electron range within the target. The electron range is both composition and beam energy dependent (Kanaya & Okayama 1972): R(nm) = 27.6(A/Z0.89 ρ)E0 1.67

(11.5a)

Semiconductor EDS (129 eV resolution at MnKα) E0 ≥ 20 keV

K-shell

K&L

U ≥ 2 (Ec ≤ 10 keV)

L-shell

L&M

M-shell Not detectable

H Li

He

Be

B

C

N

O

F

Ne

Na

Mg

Al

Si

P

S

Cl

Ar

K

Ca

Sc

Ti

V

Cr

Mn

Fe

Co

Ni

Cu

Zn

Ga

Ge

As

Se

Br

Kr

Rb

Sr

Y

Zr

Nb

Mo

Tc

Ru

Rh

Pd

Ag

Cd

In

Sn

Sb

Te

I

Xe

Cs

Ba

La

Hf

Ta

W

Re

Os

Ir

Pt

Au

Hg

Tl

Pb

Bi

Po

At

Rn

Fr

Ra

Ac Ce

Pr

Nd

Pm

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

Th

Pa

U

Np

Pu

Am

Cm

Bk

Cf

Es

Fm

Md

No

Lr

Fig. 11.1 Periodic Table showing typical choices of atomic shells for operation in the beam energy range for “conventional” microanalysis, E0 = 10 keV – 30 keV


267

where A is the atomic weight (g/mole), Z is the atomic number, ρ is the density (g/cm3 ), and E0 is the incident beam energy (keV). Energy loss occurs along the electron range, so that calculation of the volume of x-ray production for a particular atomic species involves a correction to Equation (11.3) to reduce the range to account for the cessation of x-ray production when the beam electron loses energy down to the value of Ec : Rx (nm) = 27.6(A/Z0.89 ρ)(E0 1.67 − Ec 1.67 )

(11.5b)

Table 11.2 contains values of Rx for various atom species with K-edge ionization energies ranging from 7.11 keV (Fe) to 0.69 keV (F) present as dilute constituents (C < 0.01 mass fraction) in a carbon matrix. For the conventional beam energy range, the x-ray production range is always in excess of 1 μm except for the hardest radiation considered, FeK, at the lowest beam energy, 10 keV, where the range decreases to approximately 0.7 μm. While these ranges give a measure of the depth of penetration of the beam into the target, Monte Carlo electron trajectory simulation suggests that the lateral spread of the beam is of a similar value. A crude but useful estimate of the sampling volume is to consider that volume to have the form of a hemisphere centered on the beam impact point whose radius is the range. Of course, within this sampling volume the local density of x-ray production varies greatly with position relative to the beam impact point, but the hemispherical sampling volume gives a conservative estimate of the spatial resolution of analysis. Since we are often interested in the partitioning of an element between two structures, an estimate of the x-ray sampling volume is useful in estimating the likelihood of inadvertent sampling of a second, nearby structure while interrogating the first. Examination of Table 11.2 suggests that, for most elements of interest in biological matrices, electron-excited x-ray spectrometry is restricted to super-micrometer spatial resolution when the beam energy is selected in the conventional range, 10 – 30 keV.

X-Ray Production and Spatial Resolution in Low Beam Energy Microanalysis The strong beam energy dependence of the electron and x-ray ranges, which follow an exponent of 1.67 in Equations (11.3) and (11.4), leads to the key rationale for low voltage microanalysis: The lateral and depth spatial resolution can be greatly Table 11.2 Conventional beam energy microanalysis: total electron range and x-ray production range for various atomic species present as traces (C < 0.01 mass fraction) in a matrix of carbon (ρ = 2 g/cm3 ) E0 (keV)

Range

FeK

CaK

ClK

PK

NaK

FK

30 25 20 15 10

9.85 μm 7.26 μm 5.00 μm 3.09 μm 1.57 μm

8.96 μm 6.37 μm 4.11 μm 2.20 μm 0.68 μm

9.43 μm 6.85 μm 4.59 μm 2.68 μm 1.16 μm

9.66 μm 7.07 μm 4.81 μm 2.90 μm 1.38 μm

9.73 μm 7.14 μm 4.88 μm 2.97 μm 1.06 μm

9.81 μm 7.22 μm 4.96 μm 3.06 μm 1.53 μm

9.83 μm 7.24 μm 4.98 μm 3.08 μm 1.55 μm

268

D. E. Newbury

reduced by selecting the beam energy at or below 5 keV. The upper bound of 5 keV for the low-beam energy range is based upon the observation that this is the lowest beam energy for which a useful characteristic x-ray peak can be detected for all of the naturally occurring elements of the Periodic Table (further discussion below). The electron and x-ray ranges for the same elements in a carbon matrix as measured in the low beam energy range are listed in Table 11.3. The decrease in the various values of the range in the low beam energy regime compared to the conventional beam energy regime is quite striking. For example, the x-ray production range for ClK is 4.8 μm (4800 nm) at 20 keV and only 304 nm at 5 keV. The reduction in the sampling volume (and mass) follows the cube of the linear dimension, so that the analytical volume and mass are reduced by a factor of approximately 4000 by reducing the beam energy from 20 keV to 5 keV. Analytical electron microscopy (AEM) with x-ray spectrometry performed at high beam energy, E0 ≥100 keV, is capable of even better lateral resolution, by a factor of at least ten, in a low atomic number matrix like carbon, but AEM requires that the specimen be prepared as a thin section, 100 nm or thinner (Joy et al. 1986). Low beam energy analysis in the SEM has the great advantage of being able to work directly with the as-received sample, or at most requiring only one prepared surface so that specimen thickness is not an issue. LVSEM-microanalysis has the added advantage that large lateral areas, at least a square centimeter in size or larger, can be readily accessed with typical SEM stage movements. The development of the variable pressure/environmental SEM has further improved the compatibility of biological specimens with the vacuum environment of the SEM

Other Advantages of Low Beam Energy Microanalysis While the great improvement in spatial resolution obtained with low-beam-energy microanalysis compared to the conventional analysis strategy is the prime rationale for its utilization, there are additional positive factors. The x-ray spectrum measured under conventional beam energy conditions is subject to significant modification due to the physics of x-ray generation and propagation. The spectrum of x-rays that emerges from a thick target is substantially different from the ideal spectrum Table 11.3 Low beam energy microanalysis: total electron range and x-ray production range for various atomic species present as traces (C < 0.01 mass fraction) in a matrix of carbon (ρ = 2 g/cm3 ) E0 (keV)

Range

FeL

CaL

ClK

PK

NaK

FK

5 4 3 2 1

494 nm 340 nm 211 nm 107 nm 34 nm

475 nm 321 nm 191 nm 88 nm 15 nm∗

488 nm 334 nm 205 nm 101 nm 28 nm

304 nm 151 nm∗ nd∗∗ nd nd

374 nm 221 nm 91 nm∗ nd nd

456 nm 302 nm 172 nm 69 nm∗ nd

476 nm 322 nm 192 nm 89 nm 16 nm∗

∗ Overvoltage ∗∗ nd,

below U = 2 but above U = 1.25 not detectable due to U < 1.


269

generated within the target, i.e., the spectrum that only incorporates the physics of generation of characteristic x-rays and the electron-induced bremsstrahlung (braking radiation) x-rays. The measured x-ray intensity is subject to effects of electron backscattering, electron energy loss, x-ray absorption and x-ray fluorescence, which are collectively referred to as “matrix effects,” since the exact values of the effects for a given photon energy depend upon the exact composition of the target matrix. The absorption, atomic number and secondary fluorescence correction factors derived from these physical effects form the basis of quantitative electron probe microanalysis, in which the composition of the unknown target is determined relative to standards of known composition (Goldstein et al. 2003). The standards can be as simple as a pure element or a stoichiometric compound, which is especially useful for those elements that either are not solid at room temperature in a vacuum, such as chlorine, or which are unstable under electron beam bombardment, such as sulfur. Operation in the low beam energy microanalysis regime has a significant impact on reducing the magnitude of these matrix correction factors.

Minimized X-ray Absorption The as-generated spectrum (E0 = 20 keV) for a carbon target containing 0.005 mass fraction of each of the elements listed in Table 11.2 is shown in Fig. 11.2 along with the spectrum after propagation through the specimen. X-rays are subject to photoelectric absorption within the target: I/I0 = exp−(μ/ρ)ρs

(11.6)

where I0 is the x-ray intensity generated at a particular location in the target, I is the intensity that emerges at the surface after traveling along a path s, (μ/ρ) is the mass absorption coefficient, and ρ is the density (g/cm3 ). Generally, the mass absorption coefficients increase as photon energies decrease for a given matrix. Sharp increases in the mass absorption coefficient occur for photon energies just above the critical ionization energy for an atomic species, which can be seen as sharp steps in the intensity of the continuum background (bremsstrahlung) x-rays. Table 11.2 indicates that for E0 = 20 keV in a carbon matrix, the x-rays are produced in a distribution that extends to depths exceeding 4 μm, leading to the strong absorption effects seen in Figs. 11.2a and 11.2b, which show simulations of the x-ray spectrum as generated within the specimen and upon exiting the specimen after absorption. Despite the low atomic number of the sample matrix, which is predominantly carbon (97%) in this simulation, the absorption of the low energy photons is so severe that the generated and emitted x-ray intensities do not converge until a photon energy of approximately 5 keV; above this value, the spectrum is not significantly affected by x-ray absorption. In the low beam energy regime, the electron range and x-ray production ranges are much shorter, by a factor of 10 or more, and consequently, the x-ray absorption paths are reduced proportionally. Since the absorption function follows an

270

D. E. Newbury (a)

Intensity

70,000

FeL 0

0

1.0

2.0

3.0

4.0 5.0 6.0 7.0 Photon Energy (keV)

8.0

9.0 10.0

(b)

Intensity

5000

FeL

0 0

1.0

2.0

3.0

4.0 5.0 6.0 7.0 Photon Energy (ke V)

8.0

9.0

1 0.0

Fig. 11.2 Simulation of the x-ray spectrum with an incident beam energy of 20 keV for a target containing 0.005 mass fraction each of F, Na, P, Cl, Ca, and Fe in C (balance = 0.97 mass fraction) as generated within the specimen (line trace) and upon exiting the specimen after absorption (filled): (a) vertical axis scaled to CK peak; (b) expanded vertical scale. The generated and emitted x-ray intensities do not converge until a photon energy of approximately 5 keV

exponential dependence on distance, this reduction in absorption path length has a dramatic effect on the spectrum, as can be seen in Figs. 11.3a and 11.3b, where the generated and emitted spectra are compared at E0 = 5 keV. The generated and emitted spectra converge above a photon energy of approximately 1.8 keV, so that, with respect to absorption, the measured spectrum is unaffected from 1.8 keV to 5 keV, which is a much reduced absorption situation compared to analysis under conventional beam energy conditions in Fig. 11.2. Because x-ray absorption is minimized, secondary x-ray fluorescence, which is initiated by x-ray absorption, becomes negligible in the low voltage analysis regime, so the fluorescence matrix correction factor is essentially unity.


271

(a)

Intensity

35000

0 0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5 .0

3.5

4.0

4.5

5.0

Photon Energy (keV) (b)

Intensity

4000

FeL

0 0

0.5

1.0

1.5

2 .0

2. 5

3.0

Photon Energy (keV)

Fig. 11.3 Simulation of the x-ray spectrum with an incident beam energy of 5 keV for a target containing 0.005 mass fraction each of F, Na, P, Cl, Ca, and Fe in C (balance = 0.97 mass fraction) as generated within the specimen (line trace) and upon exiting the specimen after absorption (filled): (a) vertical axis scaled to CK peak; (b) expanded vertical scale. The generated and emitted x-ray intensities converge at a photon energy of approximately 1.8 keV

Electron Scattering Effects Electron scattering affects x-ray production and spatial resolution through two mechanisms: (1) Elastic scattering causes electrons to deviate from their initial incident trajectory, and a significant fraction undergoes sufficient collisions to escape the specimen as backscattered electrons, reducing the total possible x-ray production by carrying off energy that could have made additional x-ray events; and (2) inelastic scattering processes (in addition to the inner shell ionization that

272

D. E. Newbury

initiates x-ray emission) that collectively reduce the energy of the beam electrons. Both of these effects are strongly dependent on the beam energy and the atomic number (composition) of the target and form the basis of the “atomic number” correction in quantitative electron probe microanalysis. In the low-beam-energy regime, the effects of electron scattering become relatively less significant because of the reduced overvoltage, so that although backscattering still occurs, most of the possible x-ray production is actually created in the specimen. When LVSEM x-ray production from a particular element, e.g., Mg, dispersed in a light atomic number matrix that consists mostly of carbon, is compared to x-ray production from a standard that consists of the same element in pure form, the atomic number correction factor, which includes the ratio of the efficiencies of production in the sample and standard, is close to unity.

Table 11.4 X-ray shell choices for low beam energy x-ray microanalysis (E0 = 5 keV and U > 1.1) K-shell Z = 4 Be (EK = 116 eV) to Z = 21 Sc (EK = 4496 eV) L-shell Z = 22 Ti (EL = 454 eV) to Z = 53 I (EL = 4559 eV) M-shell Z = 54 Xe (EM = 672 eV) to Z = 92 U (EM = 3551 eV)

Semiconductor EDS (129 eV resolution at MnKα) E0 = 5 keV

K-shell

U0 ≥ 1.1(Ec ≤ 1.5 keV)

L-shell M-shell

Not detectable

H

He

L i Be Na

Mg

K

Ca

Sc

Ti

V

Cr

Rb

Sr

Y

Zr

Nb

Mo Tc

Cs

Ba

La

Hf

Ta

W

Fr

R a Ac Ce Th

B

C

N

O

F

Ne

Al

Si

P

S

Cl

Ar

C o Ni

Cu

Zn Ga

Ge

As

Se

Br

Kr

Ru Rh

Pd

Ag

Cd

In

Sn

Sb

Te

I

Xe

Re

Os

Pt

Au

Hg Tl

Pb

Bi

Po

At

Rn

Pr

Nd

P m Sm E u

Pa

U

Mn

Fe

Np

Ir

Pu

Gd Tb

Dy

Ho Er

Am C m Bk

Cf

E s F m M d No

T m Yb L u Lr

Fig. 11.4 Periodic Table showing choice of atomic shells available for microanalysis in the lowbeam-energy range, E0 = 5 keV and U = 1.1


273

General Limitations of Low Beam Energy Microanalysis The choice of a beam energy of 5 keV or less obviously limits the span of photon energies of characteristic x-rays that can be excited. The value 5 keV is taken as the beginning of the low-beam energy range based on the observation that this is the lowest beam energy for which the elements of the entire Periodic Table (except for H and He) can be excited and detected, at least when present in the specimen as a major constituent (C > 0.1 mass fraction). Even this analytical strategy for shell choice for low-beam-energy microanalysis is substantially compromised from that listed above for the conventional beam energy range because U = 1.1 must be accepted as a minimum overvoltage. With E0 = 5 keV, the shell choices for U > 1.1 are as follows in Table 11.4, and are shown graphically in Fig. 11.4: (a)

(b)

(c)

Fig. 11.5 Appearance of K, L, and M peaks excited with low overvoltage from elements at the upper end of the shell range for low beam energy analysis: (a) scandium, U = 1.1 for Sc K-shell; (b) KI, U = 1.1 for I L-shell; U = 1.39 for K K-shell; (c) uranium, U = 1.41 for U M-shell; all at E0 = 5 keV

274

D. E. Newbury

While this strategy can, in principle, reach all naturally occurring elements, it must be realized that elements on the upper end of each range are very poorly excited since U is just above the threshold. Figure 11.5 contains examples of EDS spectra for elements at the upper end of each range with E0 = 5 keV (Sc K-shell, U = 1.11; I L-shell in KI, U = 1.10; and U M-shell, U = 1.41). The peak-to-background is seen to be low for these elements, considering that Sc and U are measured as pure elements. For the compound KI, I is present at 0.76 mass fraction and yet the I Lα peak is very low compared to the KKα, due partially to the lower overvoltage for the iodine (U = 1.39 for KKab while U = 1.097 for ILIIIab ). A second physical factor also impacts the KI spectrum. Many elements that would normally be measured with K- or L-shell x-rays in the conventional-beam-energy range must instead be analyzed with L- and M-shell x-rays in the low-beam-energy regime. For a given atomic shell, the fluorescence yield ω, which is the fraction of ionization events that yield x-rays, is a strong increasing function of photon energy and the shell. In general ωK >> ωL > ωM for photons of similar energy. The lower fluorescence yield of the L-shell compared to the K-shell has a strong effect on lowering the iodine intensity compared to the potassium intensity in the KI spectrum of Fig. 11.5b. The practical consequence of this fluorescence behavior on low voltage analysis is that the intensity that can be measured for many intermediate and high atomic number elements relative to the continuum background will be much lower than that obtained in conventional analysis. The real impact of the x-ray generation physics becomes more

Semiconductor EDS (129 eV resolution at MnKα) E0 = 2.5 keV

K-shell

U ≥ 1.1(Ec ≤ 2.25 keV)

L-shell M-shell Not detectable

H

He

L i Be Na

Mg

K

Ca

Sc

Ti

V

Cr

Rb

Sr

Y

Zr

Nb

Cs

Ba

La

Hf

Ta

Fr

Ra

Ac

B

C

N

O

F

Ne

Al

Si

P

S

C l Ar

Fe

C o Ni

Cu

Zn

Ga

Ge

As

Se

Br

Kr

Mo Tc

Ru

Rh

Pd

Ag

Cd

In

Sn

Sb

Te

I

Xe

W

Re

Os

Ir

Pt

Au

Hg

Tl

Pb

Bi

Po

At

Rn

Ce

Pr

Nd

P m Sm E u

Gd

Tb

Dy

Ho

Er

T m Yb

Lu

Th

Pa

U

A m C m Bk

Cf

E s F m M d No

Lr

Mn

Np

Pu

Fig. 11.6 Periodic Table showing choice of atomic shells available for microanalysis in the low beam energy range, E0 = 2.5 keV and U = 1.1. Note significant loss of elements that can be effectively measured


275

obvious as the beam energy is reduced below 5 keV, whereupon large sections of the Periodic Table become effectively inaccessible. There is simply no shell with adequate fluorescence yield available. As shown schematically in Fig. 11.6, at E0 = 2.5 keV many elements are not detectable at any concentration. If minor and trace level concentrations were considered, even more elements would have to be rated inaccessible.

Current X-ray Spectrometry Capabilities for LVSEM Microanalysis With existing technology, the analyst has two choices for performing analytical x-ray spectrometry in the low beam energy regime: (1) the semiconductor energy dispersive spectrometer (EDS) and (2) the wavelength dispersive spectrometer (WDS).

EDS The simplest, easiest to use, and by far most common form of x-ray spectrometry implemented in the SEM is the energy dispersive x-ray spectrometer, which is based upon a semiconductor detector. Usually the EDS is made of silicon (Si-EDS), although germanium detectors (Ge-EDS) are also available. The popularity of the EDS for all SEM applications, including low-beam-energy analysis, arises principally because the energy dispersive aspect of its operation permits the analyst to record the complete x-ray spectrum at every location sampled. The photon detection mechanism of the Si-EDS is based upon photoelectric absorption within the Si crystal, followed by inelastic scattering of the photoelectron, creating charge carriers proportional in number to the original photon energy. While this detection process is serial in time for single photons, it is effectively parallel in photon energy. That is, the entire energy range of photon production can be continuously monitored with the Si-EDS, from a threshold of about 100 eV to the photon energy equal the incident electron beam energy E0 , which is the upper limit (Duane-Hunt limit) of the x-ray bremsstrahlung (braking radiation) or continuum. Si-EDS does suffer from several drawbacks in its capacity to measure the x-ray spectrum. The principal limitation of Si-EDS is relatively poor energy resolution when compared to the natural peak width. Measured at a photon energy of 5890 eV (MnKα), the optimum resolution is approximately 125–130 eV (depending on detector material and size), where resolution is defined as the full peak width at half the maximum peak intensity (FWHM), which can be compared with a “natural” peak width of approximately 1.5 eV for MnKα. This substantial peak broadening is a consequence of the limited number of charge carriers (electron-hole pairs) generated by inelastic scattering of the

276

D. E. Newbury (a) 1000

Resolution, Peak Width (eV)

Large Area Si-EDS 100 High Resolution Si-EDS LiF

10 PET

TAP

Kα1–Kα2

1

0.1 0

2000

4000

6000

8000

10000

Photon energy eV

(b) 1000

Resolution, Peak Width (eV)

Large Area Si-EDS 100 High Resolution Si-EDS

10 PET

TAP 1

Kα1–Kα2 0.1 0

Fig. 11.7

500

1000 Photon energy (eV)

1500

2000


277

photoelectron that is ejected from a silicon atom following the absorption of an x-ray photon. The number of charge carriers (electrons or holes) liberated is approximately: ne,h = Ep /3.6 eV

(11.7)

where Ep is the photon energy. For MnKα at 5890 eV, this gives an average √ of 1636 charges. One sigma (σ for counting or Poisson statistics is equal to n, where n is the number of counts) of this number is 40.4, or about 2.47%. The FWHM of a Gaussian peak is 2.355σ, which thus predicts a value of E/E of 5.8% based upon the simple counting statistics, while the observed optimum FWHM of the Si-EDS is about 129 eV/5890 eV = 2.2%. The resolution depends on the photon energy, as plotted in Fig. 11.7 for a small cross-sectional area detector (10 mm2 , referred to as high-resolution) and for a large cross sectional area (60 mm2 , “large solid angle”) detector. Some individual values of the FWHM are listed in Table 11.5 for a small area, high-resolution (129 eV at MnKα) Si-EDS. The impact of the Si-EDS detection process upon the ideal spectrum of Fig. 11.2 is shown in Figs. 11.8a (linear intensity scale) and 11.8b (logarithmic intensity scale), where the narrow, high peaks in the generated spectrum are substantially broadened and reduced relative to the x-ray continuum background. Actual performance of Si-EDS in the low photon energy region is illustrated in Fig. 11.9, which shows the PKα and PKβ region of the spectrum of Fluorapatite. The resolution of the Si-EDS is inadequate to separate the PKα and PKβ so that a composite peak is observed. There are several negative consequences of this significant broadening that the characteristic x-ray peak undergoes due to the poor resolution of the Si-EDS measurement process:

“Lost” Elements As noted in the discussion of shell selection strategy, the analyst seeking to work in the low-beam-energy regime must deal with the low relative intensity of characteristic peaks due to the low overvoltage and the restriction to shells with low fluorescence yield. This combination of factors results in an inherently low P/B (as-generated) for many elements, and the peak broadening action of the EDS further lowers the measured P/B, all of which acts to further restrict access

Fig. 11.7 (Continued) Resolution (FWHM) versus photon energy for various types of x-ray spectrometers. (a) 0 – 10 keV; (b) 0 – 2 keV. Key: Solid lines: Si-EDS large area (150 eV at MnKα) and “high” resolution (129 eV at MnKα); dashed lines: WDS for various diffractors, LiF, TAP (thallium acid phthalate), PET (pentaerythritol); open squares, diamonds, circles: WDS with synthetic multilayer diffractors; x: first generation NIST microcalorimeter EDS, with analog processing; +: second generation NIST microcalorimeter EDS, optimized for low photon energy, with analog processing; filled triangles: second generation NIST microcalorimeter EDS, broad range version (at MnKα) and low photon energy version (at Al Kα)with optimized digital processing; pointcentered squares: Kα1 width (FWHM) for various elements

278

D. E. Newbury Table 11.5 Comparison of Si-EDS and WDS characteristics Si-EDS WDS (diffractor)

1. Resolution at MnK α(5890 eV) at CaKα (3691 eV) at PKα (2015 eV) at AlKα(1487 eV) at MgKα (1254 eV) at OK (523 eV) at CK (282 eV) 2. Peak Interference 3. Limit of detection (Mass fraction) 4. Photon energy range

5. Photon energy coverage 6. Time constant

7. Typical beam current 8.∗ Count rate E0 = 5 keV Pure element equivalent Ca5 (PO4 )3 F 9. Best application 10. Elemental Mapping ∗

129 eV

12 eV (LiF)

102 eV 83 eV 76 eV 71 eV 60 eV 63 eV Frequent 0.001 to 0.01

12.4 eV (PET) 2.4 eV (PET) 7.5 eV (TAP) 6.2 eV (TAP) 30 eV (LDE1) 11.7 eV (LDEC) Rare 10−5 to 10−4

100 eV to E0 (E0 ≤ 30 keV)

100 eV – 12 keV with 6 diffractors: LDEB, LDEC, LDE1 TAP, PET, LiF ∼resolution 1 μs

continuous entire range 50 μs (129 eV best resolution;3 kHz) 5 μs (30 kHz count rate;178 eV) 1 nA OK 1.78 × 105 c/s/nA/sr FK 1.26 × 105 c/s/nA/sr PK 8.78 × 104 c/s/nA/sr CaKα 7.65 × 103 c/s/nA/sr Qualitative analysis Major, some minor

100 nA 3.60 × 07 × 102 c/s/nA 9.07 × 102 c/s/nA Trace analysis Major, minor and trace

Note difference in dimensions

to elements. Figures 11.4 and 11.6 show the practical situation for Si-EDS at (a) E0 = 5 keV and (b) E0 = 2.5 keV considering all elements present as major constituents (C > 0.1 mass fraction). While generalizations such as those embodied in Figs. 11.4 and 11.6 can be made, the actual detectability situation must be examined for each element of interest. Consider the case of the biologically important element calcium. When the incident-beam energy is reduced below the K-edge energy of calcium, EK = 4.04 keV , the analyst must choose the Ca L-shell with ELIII = 0.349 keV . The CaLα peak at 0.341 keV is certainly within the measurable energy range of EDS, which can readily measure CK at 0.282 keV. However, the fluorescence yield is so low that CaLα is not detectable with EDS on a practical basis. Figure 11.10 shows the CaL region of a spectrum of the mineral Fluorapatite recorded at E0 = 5 keV for the extreme case of 3,000 seconds at 40% deadtime, resulting in a peak channel intensity of 510,000 counts for OK. The spectrum is also shown after multiple linear least squares peak stripping of the CK and OK. No peak structure for CaLα can be discerned in the continuum background despite the extremely high spectral statistics, which are certainly at the limit of a practical measurement dose. For EDS measurements, the CaK-shell is the only practical choice, which restricts E0 to a minimum of 4.4 keV to achieve a minimum value of U = 1.1.


279

(a)

(b)

Fig. 11.8 Effect of EDS detector broadening. Simulation of the x-ray spectrum with an incident beam energy of 5 keV for a target of 0.005 mass fraction each of F, Na, P, Cl, Ca, and Fe in C (0.97 mass fraction): upon exiting the specimen after absorption and after the detector broadening: (a) linear intensity scale and (b) logarithmic intensity scale

Peak Interference There are numerous practical examples of peak interference that are encountered in Si-EDS spectrometry beyond the intrafamily interference situations that occur for elements such as Na, Mg, Al, Si, and P, for which the Kα-Kβ pair is unresolved, and Cr, Mn, Fe, Co, Ni, Cu, Zn, etc. for which the L-family peaks are unresolved. Examples of significant interelement interferences include VL, CrL with OK, SK with PbM, MnL, FeL with FK, ZnL with NaK, AsL with MgK, BrL with AlK, etc. When conventional beam energy analysis is employed, it is usually possible to identify these interference situations by examining peaks arising from other shells, e.g., if PbM is suspected of interfering with SK, then the presence of lead can be confirmed by exciting PbL with a beam energy above 13 keV (PbLIII excitation

280

D. E. Newbury

Fig. 11.9 P K region of the x-ray spectrum of Fluorapatite, Ca5 (PO4 )3 F, as measured with a SiEDS (resolution 129 eV at MnKα) and with WDS (PET diffractor; resolution approximately 9 eV at a photon energy of 2 keV). Note the complete separation of the PKα and PKβ peaks with WDS, and the observation of the satellite peak on the shoulder of the PKα peak. E0 = 5 keV

edge). This possibility is lost with low beam energy microanalysis because of the maximum beam energy of 5 keV. Instead, a peak deconvolution procedure must be applied assuming that one of the interfering peaks is present, and the spectrum residuals studied for evidence of the other peak. Another practical problem with Si-EDS is the impact that the elements C and O have on the interference situation. This interference is obvious in the case of CaL shown in Fig. 11.10, but the region impacted by the C and O peak spans the low

Fig. 11.10 Fluorapatite, Ca5 (PO4 )3 F, as measured for 3,000 seconds at 40% deadtime (8 nA) and E0 = 5 keV (OK peak channel = 510,000 counts). Dose = 24,000 nA-s. Background under the peaks is shown after multiple linear least-squares peak stripping of CK and OK from references measured on pure carbon and SiO2 Note that no peak structures for CaLα and CaLl can be seen at the appropriate locations marked


281

Fig. 11.11 Interference of CK and OK peaks with L-shell and M-shell elements

photon energy peaks of many other elements. Figure 11.11 shows the position of the L- and M- peaks of these elements relative to CK and OK as measured with a 129-eV Si-EDS. Moreover, although the fluorescence yields of CK and OK are low (ωK = 0.0035 for C and 0.0090 for O) compared to K-shell yields of higher atomic number elements (ωK = 0.45 for Cu), the K-shell yields are much higher than the L- and M- shell fluorescence yields in this energy range (ωL = 0.00047 for K and 0.00067 for Ca; ωM = 0.00033 for Sr and 0.00066 for Ag), so that carbon and oxygen, even when present at low concentrations or as surface layers, tend to dominate this region of the spectrum.

Deteriorated Limits of Detection The mere spreading out of the characteristic photons over a wider energy range due to the Si-EDS measurement process would have no particular consequences upon detection sensitivity except for the fact that the characteristic peak spreads out over the x-ray continuum of all photon energies. The characteristic x-ray peaks are situated upon this continuum background, and any attempt at a quantitative measurement must proceed from a separation of the two spectral components so that an accurate measurement of the characteristic x-ray intensity is obtained. An estimate of the background under the peak must be subtracted from the total intensity. The natural statistical variance in this continuum background forms the eventual limit to the recognition of the characteristic peak, and thus defines the limit of detection. As the resolution becomes poorer, more background radiation is incorporated in any measurement of a characteristic peak, and therefore the variance of this background is greater, giving a poorer limit of detection.

282

D. E. Newbury

For a general estimate of the concentration limit of detection, CDL , from the spectrum of a pure element, the background NB,DL within the energy window that defines the peak integral for the desired trace element is first determined. The intensity ratio of unknown to pure element standard, or k-value, that corresponds to the detection limit is calculated for the condition that the intensity of the unknown has fallen to 3 times the standard deviation of the background counts, Iunk = 3 NB,DL 1/2 : kDL = Iunk /Istd = 3NB,DL 1/2 /NS

(11.8)

where NS = NP,S – NB,S is the pure element standard intensity, corrected for background. The corresponding CDL is then found with the appropriate ZAF matrix correction factors for the trace element(s) of interest calculated as part of the analytical procedure: CDL = kDL (ZAF)i

(11.9)

For low beam energy analysis, the ZAF factors generally converge toward unity, so that the assumption CDL = kDL is generally adequate for estimating detection. Applying this procedure to the problem of detecting elements in the spectrum of Fluorapatite measured at E0 = 5 keV shown in Fig. 11.12, the limit of detection is 0.002 mass fraction for MgK and 0.012 mass fraction for KKα, with a dose of 100s at approximately 12% deadtime (yielding an OK peak channel intensity of 4900 counts with Si-EDS performance of 135 eV resolution at MnKα). There is no peak interference for MgK and KKα in this situation. Determining the limit of detection when peak interference occurs requires deconvolution of the interfering peak. The variance due to the high peak channels increases the background variance after peak

Fig. 11.12 Fluorapatite, Ca5 (PO4 )3 F, as measured for 500 seconds at 12% deadtime and E0 = 5 keV (OK peak channel = 24,490 counts). The limit of detection for MgKα, β is calculated to be 0.002 mass fraction for this favorable case of no peak interference and high dose. For KKα, the limit of detection is 0.012 mass fraction due to lower overvoltage


283

removal, so that the CDL is degraded. The detection limit for potassium is poorer than that for magnesium because of the reduced overvoltage for the potassium K-shell relative to the magnesium K-shell. CDL depends on measurement conditions, particularly overvoltage, and the specimen composition, but generally for Si-EDS, CDL will be in the range 0.001 to 0.01 mass fraction unless very long spectrum measurement times are used (500s and higher). The behavior of the peak-to-continuum background as U is reduced is shown in Fig. 11.13 for a pure silicon target. From the measured P/B values and the peak counting rate of these spectra, the CDL can be calculated. Figure 11.14a shows a plot of CDL for a silicon matrix determined from these experimental measurements as a function of beam energy for the low-beam-energy regime. As the beam energy is decreased, the limit of detection moves from the trace constituent region (a)

(b)

Fig. 11.13 Appearance of a pure element K-shell peak (SiKα, β) as a function of overvoltage: (a) Linear intensity scale. Note that the spectrum for lowest overvoltage, U = 1.09 at E0 = 2 keV, does not appear to show a peak with the scaling used; (b) logarithmic intensity scale which reveals the SiKα, β peak at U = 1.09

284

D. E. Newbury (a) Silicon Overvoltage Study

5 keV 100

1

P/B

Minor

10

Trace

0.1

0.01

CMDL (100s)

P/B

Major

0.001 CMDL (100s)

1

0

5

10 U = E0 / Ec

0.0001 20

15

(b) Silicon Overvoltage Study

5 keV 1

Major

CMDL (mass fraction)

0.1

Minor

CMDL (100s) CMDL (200s) CMDL (1000s)

0.01

Trace

0.001

0.0001

0

5

10 U = E0 / Ec

15

20


285

(C < 0.01 mass fraction) into the minor constituent region (0.01 ≤ C ≤ 0.1). Extending the measurement time can reduce CDL , as shown in Fig. 11.14b, but other factors, such as possible alteration of the specimen due to radiation damage and instability due to instrumental drift may occur with increased electron dose and prevent the use of long dwell times.

WDS Physical Basis and Limitations Wavelength-dispersive x-ray spectrometry (WDS) historically preceded Si-EDS as the x-ray measurement technology for electron beam systems. The WDS is based upon Bragg diffraction of the x-rays, which are produced as a virtual point source at the beam impact on the specimen, which has micrometer source dimensions compared to the centimeter dimensions of the diffractometer. By establishing the proper geometric arrangement for the specimen x-ray source, the diffraction crystal, and the detector, x-ray diffraction will produce a scattering maximum for a crystal with atomic plane spacing, d, and for a particular x-ray wavelength, λ, at the Bragg angle, θB , nλ = 2d sin θB

(11.10)

where n is an integer giving the order of the reflection. For a given wavelength, the intensity generally decreases as n increases. The WDS is much sharper in energy resolution than the Si-EDS because the diffraction process changes efficiency rapidly with a small angular change in crystal orientation relative to the x-ray source (the beam impact on the specimen). The resolution depends upon the crystal chosen and the photon energy, but it is generally about 2 to 15 eV FWHM for the commonly-used diffractors needed to span the photon energy range of interest, as illustrated in Fig. 11.7. The performance of the WDS in resolving the PKα and PKβ peaks is shown in Fig. 11.9, where the resolution is sufficient to easily separate these peaks, as well as to reveal a satellite peak on the shoulder of the PKα peak. The WDS as mounted on an electron beam column consists of a high precision mechanical platform that can move the diffractor and detector along a focusing circle relative to a fixed x-ray source (beam striking the specimen) with tolerances of milliradians in angular position and micrometers in spatial position. In fact, in order to present the specimen at a location that is reproducible within micrometers, a fixed optical microscope with shallow depth of focus is usually used to define

Fig. 11.14 Limits of detection from the conventional-beam-energy range into the low-beamenergy regime: (a) P/B (measured) and CMDL for a silicon matrix for 100 s spectrum measurement at 40% deadtime as a function of overvoltage. (b) CMDL for a silicon matrix for 100 s, 200 s, and 1000 s spectrum measurement at 40% deadtime as a function of overvoltage

286

D. E. Newbury

the position of optimum WDS transmission. The specimen is moved mechanically along the z-axis (i.e., parallel to the optic axis of the SEM) to bring it into focus in the fixed optical microscope, thus consistently establishing a position with a reproducibility of a few micrometers. As an alternative, the SEM focus can be used to determine WDS focus, but because the SEM normally operates with such a large depth of focus compared to an optical microscope, the necessary precision in specimen positioning is not adequate for highly precise WDS focusing, where a defocus of 10 μm can cause a significant loss in spectrometer efficiency, depending on the exact spectrometer optical design being used. The diffraction process acts to limit the instantaneous transmission of x-rays through the spectrometer to a narrow band that is essentially a fraction of the resolution, or typically a few eV at most. Thus, the WDS is a serial spectrometer in photon wavelength (or equivalently, photon energy) in which the vast majority of the spectrum emitted from the specimen is wasted at any instant. To view the peak shape, or an extended portion of the spectrum, the WDS must be mechanically scanned by moving the diffractor and the detector on an ideal focusing circle so that Equation (11.8) is continuously satisfied for a succession of wavelengths. The geometry of the spectrometer constrains the wavelength range that can be diffracted from a crystal with a specific d-spacing. To cover the wavelength range that corresponds to x-rays with energies from 100 eV to 5 keV for low beam energy microanalysis requires at least six different diffractors, where the photon energy range of each diffractor is shown in Fig. 11.7. WDS spectrometers are typically constructed to accommodate two or four diffractors on a turret. Thus, to obtain a spectrum by WDS that covers the complete energy range, each crystal must be scanned in succession and changed, a process that is typically automated. However, the absolute efficiency of a WDS is low because of the small solid angle of collection (compared to the typical Si-EDS placement) and the losses in the diffraction process, which lead to requirements for high beam currents (>100 nA) and long dwell times per spectrometer step. Thus, a scan of the complete photon energy range of a diffractor might require 100 s to 1000 s at beam currents as high as 500 nA. Once the peaks of interest have been identified by scanning, peak intensity measurements can be performed much more rapidly by sequentially moving the spectrometer to on-peak and off-peak (background) positions and if necessary, the beam current can be reduced. As an example of the increased sensitivity of WDS, consider the problem of detecting the CaL peaks that were undetectable with EDS under high dose conditions following multiple linear least squares fitting of the interfering CK and OK peaks. Figure shows the results of a WDS scan with a synthetic layered material diffractor where the CaLα peak is just detectable above background. The improved P/B performance of the WDS is critical to detecting the CaLα peak. The

Fig. 11.15 WDS scans of Fluorapatite, Ca5 (PO4 )3 F, with LDE1 synthetic layered material diffractor; (a) Broad scan, showing the OK peak and CaL region, E0 = 5 keV, iB = 10 nA, 1 s per channel, 2000 channels. (b) Narrow scan showing the CaL region with a dose 100× greater, E0 = 5 keV, iB = 200 nA, 20 s per channel, 500 channels. Dose = 2,000,000 nA-s


287

(a) 140

E0 = 5 keV

120

iB = 10 nA 1 s per channel

100 Counts per second

2000 channels 80 Fluorapatite, Ca5(PO4)3F

60

40 CaLα

CaLl

20

0 100

120

140

160

180

200

Spectrometer Position (mm) (b) Apatite (CaL) 160 CaLα E0 = 5 keV iB = 200 nA

140

Counts per second

20 s per channel 500 channels

120

100

CaLl

80

60 Fluorapatite, Ca5(PO4)3F

40 150

160

170 180 Spectrometer Position (mm)

190

200

288

D. E. Newbury

dose necessary for WDS detection, however, actually exceeds that used for the failed EDS attempt shown in Fig. 11.10 by a factor of approximately 80. This is nevertheless a reasonable comparison because the WDS is capable of exploiting the high dose-rate regime while the EDS cannot. WDS can operate with very high beam current (200 nA vs. 8 nA for EDS) by using the action of the diffraction process to accept only those x-rays in the narrow region of interest and to exclude most of the x-rays which would contribute to detector deadtime and paralyze the EDS. Scanning the full CaLα peak in Fluorapatite along with the adjacent background requires such a high electron dose that, for biological specimens such scanned measurements may be impractical due to radiation damage limits. In actual quantitative analysis measurement practice, the WDS would be addressed under computer control to the peak position and then to nearby background positions to gather the necessary information to measure the peak above background, which can be accomplished with a much lower total dose. Thus, if the specimen can withstand the high beam current, the measurement of CaLα by peak jumping WDS would be a possibility. Such high beam currents are generally not consistent with high-resolution, low-voltage SEM performance, however.

Combined Si-EDS and WDS: The complementarity of WDS and Si-EDS Because of the large time penalty required to record a full WDS spectrum, it is not common practice to perform a full qualitative analysis with WDS at every specimen location analyzed. The Si-EDS, however, is extremely well suited for qualitative analysis, since it inevitably measures the entire excited x-ray spectrum at every location. This is just one aspect that illustrates the complementary nature of the Si-EDS and WDS spectrometries. As shown in Table 11.5, on many points critical to efficient and successful analysis, the strengths and weaknesses of the Si-EDS and the WDS actually complement each other. This complementarity has led to the development of an EPMA equipment configuration that combines Si-EDS and WDS capabilities, often with multiple WDS spectrometers. Such an EDS/WDS system is supported by computer-aided analysis software to optimize the data collection and processing and to combine EDS and WDS measurements. Extensive analytical procedures can be executed under completely automatic control. For example, good analytical strategy for an EPMA with both Si-EDS and WDS uses the Si-EDS spectrum for qualitative analysis at every location being analyzed, at least for major and minor constituents. The WDS can then be addressed to measure the elemental peaks that present special problems such as interference from a nearby major constituent peak, or if an element of special interest is anticipated to be near the Si-EDS limit of detection. Finally, to perform a quantitative analysis, the peak intensities measured with Si-EDS for major constituents can be combined with those measured by WDS, typically assigned to minor and trace elements.


289

Alternatively, for maximum data quality, the WDS can be used to measure peak intensities for all constituents identified during the EDS qualitative analysis phase. Combined EDS-WDS systems have been highly successful for electron beam x-ray microanalysis in the conventional-beam energy range where high-beam current (10 nA – 500 nA) can be readily delivered so that reasonable counting rates can be achieved with WDS. With the large beam currents necessary for reasonable WDS counting rates, it is usually necessary to withdraw the EDS to reduce the solid angle so as to reach acceptable counting rates. For low beam energy microanalysis, the EDS has a great advantage because of the very large solid angle that can be obtained by placing a large (30 mm2 or larger) detector in close proximity (e.g., 1 cm) to the beam impact on the specimen. Such a detector arrangement would yield a solid angle of 0.3 sr, which represents about 5% of the total solid angle above the target. With this collection efficiency, it is possible to obtain useful EDS spectra in reasonable integration times, e.g., 100 – 500 s, with the 100 pA to 1 nA beam current carried by the focused beam of the LVSEM. When large beam diameters can be used at the severe expense of imaging resolution, it is possible under low voltage conditions to obtain beam currents in the 10 nA – 50 nA range with the conventional tungsten thermionic source, which makes WDS practical. The SEM spatial resolution under microanalysis conditions can be improved with a higher brightness source, but such a source must also be capable of delivering high total current as well. Sources based upon lanthanum hexaboride provide an improvement of a factor of five in brightness, but recently the thermal field emission gun has been widely adopted because of its even greater brightness, total current capability and stability.

Recent Instrumentation Developments for LVSEM Microanalysis Three recent and continuing developments in x-ray spectrometry instrumentation should impact positively upon low-beam-energy microanalysis performed in the high-resolution, low beam current SEM. These developments improve the resolution, the solid angle of collection and/or the limiting count rate.

The Silicon Drift Detector: Improving Solid Angle and Counting Speed The new class of semiconductor-based energy-dispersive x-ray spectrometer known as the silicon drift detectors (SDD) employs the same x-ray detection physics as the Si(Li)-EDS, but the structure of the SDD represents a radical departure from the conventional monolithic Si(Li)-EDS design in several aspects, as shown schematically in Fig. 11.16:

290

D. E. Newbury

Silicon Drift Detector (SDD) X-rays 300 μm

SDDs are thin

SDD Backsurface SDDs have a complex back surface electrode structure.

Ring electrodes Resistor bridge

Central anode, 80 μm diameter

Area 5 mm2 to 100 mm2

The anode of an SDD is ~ 0.005 mm2 for a 50 mm2 detector, about 1/10,000 the area of EDS

Fig. 11.16 Schematic diagram of a silicon drift detector (SDD) listing main differences with conventional Si-EDS

Thickness The conventional Si-EDS is constructed from a thick crystal of silicon, typically 3 mm thick, whereas the SDD is based upon a thin silicon crystal wafer, typically 300 – 400 μm in thickness. This change in thickness reduces the collection path for deposited charge by an order of magnitude. Of course, the reduced thickness of the SDD lowers the efficiency for high energy photons compared to the conventional Si(Li)-EDS due to penetration, but for 14 keV photons, which corresponds to the upper energy limit for L-shell peaks of naturally occurring elements (ULα = 13.6 keV), the SDD (300 μm thick wafer) still retains about 50% efficiency, which is adequate for practical spectrometric applications, and thicker wafer detectors (400 μm) to extend the efficiency range are possible.

Applied Potential Distribution The conventional Si-EDS has uniform electrodes on the entrance and exit surfaces, while on the exit surface, the SDD has a complex pattern of nested electrode rings with a resistor bridge across the rings that permits application of a stepped potential distribution. This applied potential creates a lateral as well as a transverse field pattern that produces a collection channel through the wafer thickness, which is tilted toward the central collection anode. Thus, charge deposited because of the capture of a photon anywhere in the cross sectional area of the SDD, which is typically 50 mm2 and can be 100 mm2 , is efficiently transported to the very small central anode.


291

Electrodes The Si(Li)-EDS anode occupies the entire surface of the detector, while the SDD anode is a disk less than 100 μm in diameter, about 1/10000 the area of the anode of a 50 mm2 detector. This greatly reduces the noise contribution that results from the anode-silicon interface. With this noise term effectively eliminated from the detector noise budget, the SDD can be operated at higher temperature than the Si(Li)-EDS.

Operating Temperature The Si(Li)-EDS is usually operated with liquid nitrogen cooling, while the SDD operating temperature is about –20 ◦ C to –40 ◦ C, which can be achieved with Peltier electrical cooling. An airflow or water-cooling system exhausts the heat from the Peltier cooler. The reduced cooling demands and the relative simplicity of the cooling system make it easier to accommodate the SDD on electron beam instruments, and novel x-ray detector configurations become possible.

Resolution at MnKα For an equivalent detector area, the SDD has been demonstrated to actually achieve superior resolution: a resolution of 127 eV for an SDD at an operating temperature of −20 ◦ C compared to 129 eV for Si-EDS at an operating temperature of −190 ◦ C for equivalent area, 10 mm2 detectors. For a 50 mm2 detector, a resolution of 134 eV can be obtained, compared to approximately 142 eV for Si(Li).

Detector Counting Rate For an equivalent detector resolution, the SDD is capable of higher counting rates, by a factor of five to ten, when short time constants are used. A limiting output count rate greater than 500 kHz has been demonstrated. Measured input count rate vs output count rate response at several time constants for a 50 mm2 detector is shown in Fig. 11.17.

Low-energy Photon Detection The entrance electrode of the SDD can be made extremely thin, which increases the detection efficiency for low-energy x-rays. A spectrum for manganese (MnLα = 0.636 keV) is shown in Fig. 11.18a, demonstrating the high MnLα/MnKα ratio that can be achieved. Note that the high MnLα/MnKα ratio is not achieved by losing MnKα x-rays due to penetration through the thin SDD. At the energy of MnKα, 5.89 keV, the efficiency of the SDD is unity. Penetration through the SDD first

292

D. E. Newbury

Fig. 11.17 Output count rate and deadtime measured as a function of beam current on an Mn target at E0 = 20 keV for a silicon drift detector operating with several peaking time constants: 8 μs (resolution 134 eV at MnKα), 1 μs (163 eV), 500 ns (188 eV), and 250 ns (217 eV)

begins at approximately 8 keV. SDD spectra of the transition elements from Mn to Cu are superimposed in Fig. 11.18b.

Detector Area and Detector Arrays Individual SDD detectors can readily be made with a large active area. SDD detectors with an area of 50 mm2 are routinely produced, and detectors as large as 100 mm2 have been demonstrated. Moreover, arrays of detectors that can occupy very large solid angles, approaching π steradians, have been successfully assembled and operated to merge the individual detector spectra into a single output spectrum. For application of the SDD to low-beam-energy microanalysis, properties (4), (5), (7) and (8) (as listed in Table 11.5) are most attractive. The restriction to low-beam currents that results under high spatial resolution, low-beam-energy microanalysis operation means that the high counting speed of the SDD is not likely to be tested. However, the capability to create individual SDD detectors with large areas and SDD detector arrays should be very helpful. Because the SDDs can be cooled much more simply than Si-EDS, it should be possible to accommodate the SDD in close proximity to the specimen, which when combined with the


293

Cu

15000

SDD TC = 4 μs E0 = 20 keV

Intensity (counts)

Ni Co

Cr Mn Fe

Fe Mn

Co Ni Cu

Cr

0 0

1.0

2.0

3.0

4.0 5.0 6.0 7.0 Photon Energy (keV)

8.0

9.0

10.0

Fig. 11.18 (a) Silicon drift detector spectrum of manganese at E0 = 15 keV and a 4 μs peaking time constant showing excellent sensitivity for the MnL peak family. (b) SDD spectra for Cr, Mn, Fe, Co. Ni, and Cu showing L- and K-families

large detector area, will yield a solid angle of x-ray collection that is much larger than the conventional Si-EDS. Since the total dose delivered in low-beam-energy microanalysis is usually small compared to conventional beam conditions, more efficient collection of the x-rays with a larger solid angle detector will improve all aspects of analytical x-ray spectrometry, especially detection sensitivity. For those specimens that can sustain higher beam currents, the SDD makes it possible to capture x-ray spectrum images in which the complete energy-dispersive x-ray

294

D. E. Newbury

spectrum is recorded at each pixel of a scan. Such a procedure captures all possible elemental information about the specimen region within the performance limitations imposed by the electron beam dose and the spectrometer characteristics.

X-ray Optics-Augmented WDS Improvements to WDS: optic-enhanced WDS As noted above, the WDS has the spectral resolution to solve most problems related to peak interference and detection of peaks with low P/B that are encountered in either conventional or low-beam-energy microanalysis. Unfortunately, the inherent inefficiency of the WDS severely limits operation with the low beam currents available with the LVSEM. To increase the efficiency, a new class of WDS spectrometer has been demonstrated that incorporates polycapillary x-ray optics to substantially increase the solid angle of collection. X-ray steering is achieved by means of the high degree of reflection of x-rays that approach a smooth glass surface below the critical angle (typically milliradians) for scattering. A large area to increase the scattering below the critical angle is created through the use of polycapillary optics (Kumakhov 1990, 1998). Each channel of the optic is a single thin-walled glass capillary, many of which are bundled to form close-packed structures and then heated to bond the polycapillaries. The uniform bundle can be further heated and drawn to create a tapered optic that can, for example, collect x-rays over a large solid angle and then gradually bring those x-rays into a long parallel section for transport and presentation to the next optical component. Because the solid angle for reflection is inversely proportional to photon energy, polycapillary optics are most efficient for low energy photons. To increase the effective solid angle of WDS (geometric collection limits), a two-zone polycapillary optic has been incorporated to more efficiently couple the x-ray source at the specimen to the diffractor. As shown schematically in Fig. 11.19, the first zone of the polycapillary is a conical section that couples the x-ray emission collected over a large solid angle to the second section that parallelizes and transports the beam to a flat crystal diffractor, which is followed by a large detector. An example of a spectrum of MnF2 , also containing oxygen, is shown in Fig. 11.20, where the MnLα (637 eV) is well separated from the FK (677 eV) and the OK (525 eV). Various performance measures of the opticaugmented WDS for several elements measured with K-shell x-rays are listed in Table 11.6. The rapid developments in positionally-sensitive detectors (PSD) of high spatial resolution may make possible another interesting variation of the WDS. Charles Fiori et al. (1991) proposed a scheme to make use of an aspect of the focusing properties of the WDS to simultaneously detect a sufficiently wide photon energy range E to permit imaging a characteristic x-ray peak and the adjacent background. Their proposal noted that in a conventional focusing WDS, the geometry requires the x-ray source (beam-excited region of the specimen with micrometerdimensions), the diffractor and the detector to be placed on the focusing circle


295

Fig. 11.19 Schematic diagram of a polycapillary optic coupling the x-ray source to a flat diffractor in a wavelength dispersive x-ray spectrometer. In an accurate representation, the capillaries in the conical section would be tapered to a convergence at the source

(Rowland circle), so that convergence occurs for x-rays of a very narrow energy band at the detector. By moving the source and detector off the focusing circle, a much broader range of photon energies, with each increment dE diffracting from a different location on the diffractor, is brought to the focus at the conjugate point. By intercepting these converging rays before the conjugate point with a planar

Fig. 11.20 Spectrum of MnF2 measured with an optic-enhanced WDS. (Example courtesy Parallax Research, Inc.)

296

D. E. Newbury

Table 11.6 Measured Performance of an Optic-Augmented WDS (Parallax Research, Inc. LEXS) Measured LEXS Performance Element

Energy(eV)

Counts/sec/nA

P/B

Resolution (eV)

Sensitivity(ppm)

Be B C N(BN) O(SiO2 ) Mg Al Si

108 183 277 392 525 1254 1487 1740

350 3500 5750 416 375 600 500 400

40 50 >100 40 80 400 300 300

8 18 18.6 16 17 14 19 24

100 30 14 130 60 18 25 25

detector, a range of photon energy, E, could be simultaneously imaged on a linear detector device, although with reduced efficiency because only a small portion of the diffractor satisfies the Bragg equation for any paw photon energy. Being able to image the full width of an x-ray peak would be of special value for low photon energy peaks (Ech < 2 keV) because these peaks are subject to “chemical effects” from the differing energy levels of electrons involved in bonding. The peak position and shape can be strongly affected by chemical effects, making it necessary in conventional WDS practice to scan and integrate the peak to obtain an accurate measure of the intensity. This procedure necessarily incurs a large time penalty and inefficiency due to the loss of x-rays that are not diffracted.

Microcalorimetry The x-ray microcalorimeter is a radically different approach to x-ray spectrometry that is especially promising for LVSEM/microanalysis Wollman er al., (1997, 2000). The detection physics process of the microcalorimeter consists of measuring the temperature rise when a single x-ray photon is absorbed in a metal target, as illustrated schematically in Fig. 11.21a. Maximizing the response to the capture of a photon and minimizing the time needed to return to the baseline state both depend on minimizing the heat capacity of the detector, which depends on physical parameters such as detector volume and material parameters such as the specific heat. By operating at approximately 100 mK, a suitably low value of the heat capacity can be obtained with a metal absorber such as gold or bismuth. The temperature rise of the absorber can be measured by several different techniques, of which a leading example is a circuit incorporating a transition edge sensor (TES), illustrated schematically in Fig. 11.21b. The TES is a binary metal foil (e.g., Cu-Mo) whose layer thicknesses can be manipulated to adjust the superconducting transition temperature. The measurement circuit consists of a current source, an inductor, the metal (e.g., Bi) x-ray absorber, and the TES, arranged so that heat from the absorber must pass through the TES to reach a low-temperature reservoir. When this circuit is cooled by an external refrigeration circuit, the overall circuit resistance falls, and Joule heating I2 /R increases. When the TES reaches its superconducting


297

(a) X-ray Temperature

Thermometer ΔE

C

G Thermal Conductance

Heat Capacity

τ=

ΔE C

C G

Time

(b) Electrothermal Feedback Transition-Edge Sensor Microcalorimeter I V V

SQUID

X-ray absorber

Electrical Circuit

2

Pjoule V R R

Thermal conductance

TES Thermometer Psink

Stable equilibrium V2 = Psink R

Thermal Circuit

Heat Sink

Fig. 11.21 (a) Principle of operation of the x-ray microcalorimeter, showing the capture and the time history of the resulting temperature pulse in the metal absorber. (b) Electrical and thermal circuit for a transition edge sensor (TES) temperature measuring system for the x-ray microcalorimeter

temperature the circuit establishes electrical and thermal equilibrium. When an x-ray is now absorbed by this quiescent circuit, the extra heat deposited is automatically compensated by a decrease in the internal current flowing in the circuit to balance the Joule heating. The changing current induces a changing magnetic field in the inductor, which is measured by a superconducting quantum interference device (SQUID). The energy of the x-ray photon is thus measured as the time integral of the magnetic field. The microcalorimeter measurement process is inherently energy dispersive. The resolution performance of the microcalorimeter EDS as a function of photon energy is shown in Fig. 11.7, where it is compared with Si-EDS and WDS with various diffractors. The demonstrated resolution of the microcalorimeter detector is in the range 2 to 15 eV, which makes it comparable to WDS over most of the photon energy range used for analysis. For photon energies below approximately 700 eV, the microcalorimeter resolution is actually substantially better than the resolution achieved with WDS using the layered synthetic material (LSM) diffractors. Like the Si-EDS, the microcalorimeter EDS process is subject to paralyzable deadtime with a limiting count rate between 500/s to 1,000/s.

298

D. E. Newbury

Fig. 11.22 Microcalorimeter EDS x-ray spectrum of PbS (galena) showing separation of the SKα from the PbMα and Mβ peaks

The microcalorimeter thus combines the energy dispersive character of the semiconductor EDS with the energy resolution of the EDS. An example of the microcalorimeter spectrum of PbS, where the SK and PbM shell x-rays interfere with conventional Si-EDS, is shown in Fig. 11.22. The performance characteristics of the microcalorimeter, listed in comparison with WDS and Si-EDS in Table 11.5, make it especially attractive for low energy x-ray microanalysis in the LVSEM. The full analytical x-ray range can be continuously monitored and the entire spectrum excited in the LVSEM is continuously available. Figures 11.23a and b show an LVSEM microanalysis example of the extreme interference case of NK, OK and TiL, where despite the high-resolution of the microcalorimeter, interference still occurs between NK and TiL, so that the contributions of each element must be determined from a series of measurements on TiN, Ti metal, and Ti oxide. A strong chemical effect on the shape of the TiLα peak is observed in the TiN spectrum in Fig. 11.23a when compared to Ti metal or the Ti-oxide spectra, Fig. 11.23b. The principal limitations are the small physical size of the detector, about 0.25 mm2 , and need to cool the detector and cryoelectronics near absolute zero. The small size of the microcalorimeter detector can be compensated by the use of a polycapillary x-ray optic to couple the electron-excited x-ray source to the detector. An improvement in the detector solid angle by a factor of 300 has been demonstrated with a polycapillary optic consisting of a divergent section at the x-ray source and a convergent section at the detector. For the first commercial presentation of an x-ray microcalorimeter intended for use on an SEM, the necessary cooling system has been realized as a combination of a liquid helium refrigerator to 4 K and an adiabatic demagnetization refrigerator from 4 K to 50 mK (low temperature reservoir), with the detector operating at 100 mK. While the microcalorimeter-EDS is in its initial phase of practical investigation as an x-ray spectrometry tool for the LVSEM, the future of this technique is


299

(a) 300 μcal EDS

N Kα

Ti

Ti Lα1,2

TiN

Counts

200

Ti Lβ3,4 100

Ti Lβ1

Ti Ll Ti Lη

0 400

440 480 Energy (eV)

520

(b) 300

μcal EDS Ti TiOx

Ti Lα1,2

O Kα

Counts

200

100

Ti Ll Ti Lη

Ti Lβ1

Ti Lβ3,4

0 400

440 480 Energy (eV)

520

Fig. 11.23 X-ray microcalorimeter spectra of (a) titanium metal (scraped under inert gas) and TiN and (b) titanium metal (scraped under inert gas) and titanium oxide. Note the interference of TiLl and N K despite high-resolution (2 eV FWHM at AlKα). Also, note the severe shape change in the TiLa peak in the TiN compared with the Ti metal and the titanium oxide

extremely promising. In particular, the possibility exists that an array (n x m) of microcalorimeter detectors can be made. In this way, the limitations imposed by the small size of the individual microcalorimeter detectors can be overcome by effectively increasing the total detector area to n∗ m, and the total output count rate, less than 1 kHz with a single detector, can be increased by the same factor to the range of 10 kHz to 100 kHz, and perhaps even higher. Such a detector would equal or substantially improve upon the resolution of a conventional WDS while providing the energy dispersive function and a count rate similar to the conventional WDS for a single energy channel.

300

D. E. Newbury

An Application of Low Voltage Microanalysis: Particles Particle analysis provides an example of broad interest in the biological and physical sciences as well as in engineering fields where low voltage microanalysis can significantly benefit by incorporating new spectrum measurement technology. Under conventional-beam energy conditions, the x-ray microanalysis of microscopic particles becomes especially difficult when the particle dimensions approach the electron and/or x-ray ranges, i.e., linear dimensions below 10 micrometers (Goldstein et al. 2003). Electrons can penetrate through such particles, directly affecting the x-ray generation function for the particle but also indirectly affecting the measured x-ray spectrum through the remote generation of x-rays from the substrate and nearby features. Further, x-ray absorption can be profoundly modified by particle geometry effects, especially for low-energy photons. High energy photons generated under conventional beam energy conditions, especially the x-ray continuum, create secondary fluorescence almost entirely outside the particle dimensions, and this contributes to the measured spectrum, which can be an important problem when minor and trace-level constituents are important. Theoretical studies as well as careful experimentation have established the value of low-voltage microanalysis for application to particle studies. As illustrated in Figs. 11.24 and 11.25 from the work of Small (2002), the sharp reduction in the size of the interaction volume with the reduction in beam energy makes it possible to treat particles much more like bulk targets. Under LVSEM conditions, even simple normalization of the analytical total can produce quantitative results with substantially reduced relative error distributions compared to analysis in the conventional beam energy regime. Moreover, not only can the direct excitation at low-beam energy be constrained to occur within the particle dimensions, but electrons that escape after scattering within the particle have such reduced overvoltage that remote excitation is also diminished. Finally, because of lower photon energy, the continuum and characteristic x-ray photons produced at low-beam energy are less likely to escape the particle and cause remote secondary fluorescence. These are real advantages, but the limitations imposed on x-ray measurements by the physics

(a)

(b)

Fig. 11.24 Monte Carlo plots for the interaction of a 20 kV electron beam with a 2 μm K-411 particle. (a) Electron trajectories. (b) Mg Kα x-ray generation. (from Small 2002)


301

Fig. 11.25 Monte Carlo plots for the interaction of a 5 kV electron beam with a 2 μm K-411 particle.( a) Electron trajectories. (b) Mg Kα x-ray generation (from Small 2002)

of low beam energy excitation, as well as the measurement limitations of EDS (resolution) and conventional WDS (efficiency), conspire to restrict practical utility. Considering the instrumentation limitations, the microcalorimeter EDS has demonstrated particular utility for LVSEM particle analysis (Wollman et al. 2000). Figures 11.26 and 11.27 show examples of nanoscale particles on a silicon substrate excited with a beam energy of 5 keV. Although even under LVSEM conditions there is scattering into the silicon substrate, the resolution of the microcalorimeter EDS is sufficient to separate the Al K-family and the W M-family from the Si K-family contributions from the substrate, as shown in Fig. 11.26. In the case of the titanium

Fig. 11.26 Microcalorimeter EDS spectrum of a 300 nm particle of tungsten on a silicon substrate. Note detection of W M family and separation from SiKα. E0 = 5 keV (Wollman et al., 2000)

302

D. E. Newbury

microcalorimeter EDS

Counts

150

OKα

0.3 μm TiO2 particle on Si 1.8 keV beam voltage 0.53 nA beam current 400 s live time 406 s real time

100

30 s–1 input count rate 29 s–1 output count rate 1% dead time

50

CKα Ti Lα,β Ti Ll,η

Ti Lβ3,4

0 200

300

400 Energy (eV)

500

600

Fig. 11.27 Microcalorimeter EDS spectrum of a 300 nm particle of titanium on a silicon substrate. Note detection of TiL family and separation from OK. E0 = 5 keV

particle in Fig. 11.27, the resolution of the microcalorimeter EDS is sufficient to permit detection of the Ti L-family peaks despite their low yield (ωL = 0.0016) from the much more intense OK peak (ωK = 0.0090). These examples illustrate the utility of high performance spectrometry for the solution of real problems.

Summary Low voltage SEM/microanalysis can be a powerful tool for the biological microanalyst by providing improved spatial resolution and reduced matrix effects on the measured x-ray intensities. There are inevitable limits imposed by the physics of x-ray generation with low beam energy upon the elemental coverage and detection sensitivity that can be achieved. Optimal performance of the x-ray spectrometer is critical to achieving the best results. For most situations, the conventional EDS is the most effective choice because of its large solid angle, which helps to compensate for the low intensity of the low-voltage electron beam. When the resolution of conventional EDS is inadequate, WDS can solve the spectrometry problem, but the inherent inefficiency of WDS generally limits its application because of the need for high beam current. Augmenting the WDS with x-ray capillary optics can improve the collection efficiency to permit effective operation with lower beam current. The development of the SDD should permit improved geometric collection efficiency over the conventional EDS and a factor of ten greater count rate for the same resolution. The microcalorimeter EDS is the most promising spectrometry technology for


303

the future, combining as it does WDS resolution with energy-dispersive operation. An array of microcalorimeters with large collection angle and an output count rate of 10 kHz to 100 kHz would be extraordinarily powerful for all aspects of biological microanalysis under LVSEM operating conditions.

References Echlin P (ed) (1984) Analysis of biological and organic surfaces. Wiley, New York Echlin P (ed) (1992) Low temperature microscopy and analysis. Plenum, New York Echlin P (2002) Low voltage energy dispersive quantitative X-ray microanalysis of inorganic light elements in bulk frozen hydrated biological specimens. Microsc & Microanalysis 8:120 Echlin P and Galle P (eds) (1976) Biological microanalysis. Societe Francaise de Microscopie Electronique, Paris Goldstein JI et al (2003) Scanning electron microscopy and x-ray microanalysis, 3rd edition. Kluwer, New York Joy DC et al (eds) (1986) Principles of analytical electron microscopy. Plenum, New York Kanaya K, Okayama S (1972) J Phys D: Apll Phys, Penetration and energy-loss theory of electrons in solid targets, 5:43–58 Kumakhov MA (1990) Nucl Instrum Methods Phys Res B, Channeling of photons and new X-ray optics, 48:283–286 Kumakhov MA (1998) Adv X-ray Anal, Development of X-ray Polycapillary Optics, 41:214 Small JA (2002) J Res NIST, The Analysis of Particles at Low Accelerating Voltages (< 10kV) With Energy Dispersive X-Ray Spectroscopy (EDS), 107:555–566 Wollman DA et al (1997) J M, High-resolution, energy-dispersive microcalorimeter spectrometer for X-ray microanalysis, J. Micros 188:196–223 Wollman DA et al (2000) Microcalorimeter energy-dispersive spectrometry using a low voltage scanning electron microscope. J Microscopy 199:37–44

Index

Aberrations, electron-optical, 30, 59, 110–112, 122 chromatic, see Chromatic aberration interactions, 59, 116 higher order, 120–122 reduction by use of a retarding lens, 46, 53 spherical, see Spherical aberration Aberration-corrected SEM (ACSEM), 53, 107–108, 172 aberration correctors, 53, 115–116, 120–127 commercial systems, 115 cost, 126 chromatic aberration, 110–112 depth of field, 113–115 depth-of-field, limited, 123–125 diffraction, 112–113 optical transfer function (OTF), 121–122 optimizing spot size, 113–115 options in, 116–120 problems associated with, 123 results, 121 spherical aberration, 109–110 Aberration correctors, 53, 115–116 ACSEM, 120–122 alignment of, 115, 124–125 options in, 116–120 Ronchigram adjustment, 124 ACSEM, see Aberration-corrected SEM (ACSEM) Actin filaments colloidal gold labeling of, 192 critical-point drying, 60, 148–149 cryo-preparation of, 41, 216–218, 247, 258, 263 images of, 42, 149, 157, 162–163, 218, 234–235, 237, 239, 258 insect flight muscle, 158–160 plant cells, 233–235, 237, 239

sub-membranous network, 150–152 Toxoplasma gondii, 147–149 Aldehyde fixation, 59, 80–81, 146, 217–219 cytoskeleton, 148–149, 237 insect flight muscle, 158 human platelets, 183, 221–223 maceration, in plant tissue, 233 microorganisms, 222–224 neutralization with glycine, 191 nuclear-pore complex, 80–81, 152 plants, 230–231 post fixation, 236 protocols, 146, 149, 152, 161, 183 stabilization prior to cryofixation, 217–218, 221, 260 Toxoplasma Gondii, 148, 163 yeast, 218–219, 261 See also Glutaraldehyde Alignment, electron-optical, 48, 57, 92 aberration correctors, 115, 124–125 stereo imaging SEM, 92 stigmator, 108 Ampex video recorder, 202 Anti-contaminator, 57–58 Antibody labeling, 38, 66, 171–174, 177, 183, 187 See also Colloidal-metal labeling, and Correlative LM/LVSEM/TEM conjugation, 177–178 Fab fragments, 188–189 fluorescent metal particles, 177 methods, 190–191 plants, 236 Apicomplexan parasites, 148 Arabidopsis thaliana, 239–241 Artifacts, 37, 60, 147, 218, 250 charging, 62–64, 66, 71, 76, 84, 197, 247 critical-point drying (CPD), 147, 157, 254 decoration, 66, 257 extraction, 183

305

306 Artifacts (cont.) freeze drying, 216 freezing, 82, 224, 226, 230, 249–253 incomplete dehydration, 59, 147, 149–150, 157, 161, 178, 181, 184, 216, 218, 226 microtrabeculae, 60, 184, 226 membrane extraction, 183 replicas, 3 specimen preparation, 60, 147 stabilization, see Stabilization Astigmatism, 13, 108, 115, 125 of aberration correctors, 115, 125 higher order, 108, 125 in real-time stereo imaging, 91 Astigmator, alignment, 57, 83, 91, 115 Auger emission (AE), 29, 268 Background Bremsstrahlung, 273, 279, 268–269, 278 control, action of, 135 effect of, 142 in energy-filtering TEM, Color plate, 3 in environmental SEM, 35 from subsurface scattering, 11, 41–44, 66, 70 in x-ray microanalysis, 268–269, 273, 278–279, 281–287, 290, 292, 298 Backscattered electron detectors, 28, 48–51, 54, 222, 224, 236, 257, 262 conductive coating, 49, 51 detector quantum efficiency, 141, 255–260, 262 measurement, 141 performance, 142 shadowing effect, 49 Backscattered electrons (BSE), 11, 13, 28–31, 37–39, 42–43, 46–52, 54–55, 61–62, 64, 66, 69–70, 72, 83–86, 131, 141, 171–172, 182, 222, 236, 255, 260, Color plate, 13 atomic-number contrast, 15–16, 30, 49, 171, 247, 259 coefficient, 62, 141 collection contrast, 11, 49 density contrast, 30, 49, 171, 231, 259 detector, 49, 50–51, 54, 222, 224, 236, 257, 262 detector quantum efficiency (DQE), 139, 141, 255–260, 262 double-layer coating, 221–223, 225, 247, 259, 261–263

Index imaging, 11, 13–14, 37–39, 49, 51–52, 70, 83–86, 221–223, 225, 236, 247, 253, 257, 259–261, Color plate 13 noise, 131, 260 range, 42 reduces charging, 83–86 reduces S/N, 131 reduces spatial resolution, 11 reduces x-ray signal, 275 signal, 15 signal/noise, 131 statistics, 131 uncoated specimens, 51–52, 257 Bacteria, 75, 77, 216, 222–226 Bain, Alexander, 6 Bandwidth, detector, 6, 15, 46, 49, 50, 135 secondary electron, 6, 15, 46, 50 backscattered electron, 49, 135 phosphor limit, 15 Barrel distortion, of electron lenses, 124 Beam-induced, conductivity, 63 Beam-induced surface contamination, 37, 57 Biological specimens, see Specimens, biological Boric-acid/borate buffer, 204, 207 BSE, see Backscattered electrons (BSE) Buffers, 80, 146–150, 183, 189, 239 boric-acid/borate, 204, 207 cacodylate, 261 cold extraction, 155 cytoskeleton-preserving, 147, 149, 151, 183, 236–237, 243 low-salt, 80, 152, 153 PBS, 238, 239 plant cells, 231, 233 PIPES/HEPES, 146, 158, 161, 183 pre-cryo buffers, 216–218, 221–222, 226 Cacodylate buffer, 261 Cambridge SEMs, 12–15 Carbon replica specimens, 3, 7, 67, 87, 161, 216–217, 247, 251, 254–255, 259 Cathode-ray tube, 4, 6, 13, 27–28, 47 Cathodoluminescence, 15, 21, 37, 138 Cellular interactions, 146, 151–152, 161–165 Centriole-centrosome complex, 155 Centrosomes, 153–155 Charged particle beams, 2, 29 Charging, insulating specimen artifacts, 19, 31, 33–34, 61, 63–64, 146, 171, 245, 257, 259, 261 avoidance, 62, 71 beam-induced, conductivity, 63 of BSE detector, 49, 51 causes, 29, 31, 61–64, 66

Index coating, see Coating of column, worse at low-V0 , 46 distortion produced by, 32, 83–84, 198 examples, 63–64, 84 reduced in BSE imaging, 84, 86, 247, 257 reduced at low-V0 , 45, 129, 146, 171, 263 scan-speed effects, 198–199 specimen mounting, 71 theory, 61–64, 66 uncoated specimens, 33, 45, 64–65, 83–84, 171–172, 176, 179, 255, 257, 260, Color plate 3 when metal coat disrupted, 197 voltage contrast, 16 Chemical dehydration, 215 Chemical fixation, 217 cytoskeleton, 148–149, 237 glutaraldehyde, see Glutaraldehyde of human platelets, 183, 221–223 of insect flight muscle, 158 maceration, in plant tissue, 233 of microorganisms, 222–224 neutralization with glycine, 191 of nuclear-pore complex, 80–81, 152 osmium, see Osmuim-tetroxide, (OsO4 ) of plants, 230–231 post fixation, 236 stabilization prior to cryofixation, 217–218, 222 of Toxoplasma Gondii, 148, 163 of yeast, 218–219, 261 See also Fixation Chisels, dental, for cutting in SEM, 203–204, 206 Chloroplasts, 229, 232–233, 235, 237, 238, Color plate 13 Chromatic aberration, 30, 110–112, 118 coefficient, 30, 111, 118 correction, 110–112 energy spread, 46, 53, 56, 59, 110, 117–118 Coating, 44–45, 147, 61–74, 147, 155, 157–158, 173, 191, 216, 240, 249–250, 256, 259–260 anti-reflecting, 50 artifacts, 66, 69, 257 carbon, 39, 57, 65, 172, 174, 172, 221–223, 236, 247, 251, 254, 259 charging, 66, 73 chromium, 65–66, 218, 255 of colloidal particles, 172, 182 conductivity, 69, 199, 259–261 contrast, 64–65, 256–257 cryo-coating, 41, 83, 216, 218, 221, 253, 256–260, 263 cryosputter, 253, 260

307 decoration, 66, 257 double-layer coating, 221–223, 226, 249–250, 256, 259–260 effective thickness, 64 glow-discharge, 68–69 ideal, 259 ion-beam sputtering, 59, 68, 75, 79, 88–89, 147, 162, 191, 226 limitations caused by, 59, 61–74 for LVSEM, 61–74, 155, 173 oxide formation, 66 Penning sputtering, 65 of plants, 231, 236, 241 platinum (Pt), 63–64, 68, 75, 79–80, 149, 162, 216, 240 protocols, 68, 157, 261 reduces beam penetration, 45 of scintillator detectors, 49, 51 structure, 62, 65, 70, 79 tantalum/tungsten (Ta/W), 66, 222 for TEM, 191 theory, 61–73, 259–260 thickness, effective, 64, 260 topographic-Z contrast, 64 tungsten, 66, 68, 247, 258–260, 263 vacuum required, 69 Colloidal-metal labeling, 51–52, 171–193 antibody, see Antibody labeling for correlative imaging, 38, 66, 171–174, 183, 187 double labeling, 51, 177, 191 EF-TEM for elemental analysis, 190–193 epitope density, 173–174, 178, 182 fabricating different shapes, 188 faceted cPd particle, 174–175, 188, 192 generation of, 184–185 gold labels, 39, 50–51, 155, 172, 180, 223, 236, 242 ionic double layer, 184, 189 lectins, 38, 51, 175 with LM of living cells, 179, 181–183, 190 particle size, 185–188 plant specimen, 236 popcorn cPd particles, 174–175, 188, Color plate 1 stabilizing colloids, 174, 176, 189 tannic acid, 183, 185–186 using different metals, 186–188 Composite rich image, 262 Conductive staining, 231, 234 Conductivity, electrical, specimen, 31, 39, 44, 49–50, 51, 61–66, 69, 199, 247, 259 beam-induced, 62–63 conductive staining, 60, 204–206, 234 See also Charging and Coating

308 Conductivity, thermal, ice, 252 Contamination, 7, 34, 57, 69, 71, 82 avoidance by double-layer coating, 222, 254, 259 beam-induced, 37, 57 of column, worse at low-V0 , 46 reduces contrast, 140 reduction using cryo, 38, 52 surface-diffusion, 37–38, 57, 67, 71 worse at low V0 , 15, 46, 71, 73 theory, 37 vacuum modifications, 4, 58 Continuum x-rays, 29, 269, 273, 278–279, 282–285, 300 peak-to-continuum, 287 Contrast, image, 44, 136–139, 172, 236, 247, 251, 256–257, 262 adjustment, 134–135, 190 affected by V0 , 33–5 atomic number contrast, 13, 15–16, 40, 172 backscattered electrons, 49, 59, 172 See also Z-contrast cathodoluminescence, 15, 21, 37, 138 chanelling contrast, 5 and charging, 29, 31, 61–64, 66 chemical contrast, lack of, 37–8 collection contrast, 40 and contamination, 46, 57, 140 control, action of, 135 density contrast, 7, 13, 30, 42, 61, 86, 269 differential interference contrast, 177, 179, 181–182, 190, Color plate 2 double-layer coating, 254, 259 effective coating thickness, 61–62, 64–65, 172, 259, 262 expansion, 2, 28 fluorescence, LM, 38, 193 interaction volume, 59 See also Penetration, electron mechanical disruption of, 197, 205 and resolution, 11, 30, 44, 111, 142–143 secondary electron, 7, 32, 50–51, 64–65, 251 and signal-to-noise, 31, 33–35, 57 of small features, 31, 34, 44, 52, 66, 125, 146, 157, 262 in TEM, 40 theory, 4, 31, 33, 66–69 topographic, 7, 13, 19, 31, 33, 40, 44, 64–65, 142–143, 146, 155 topographic-Z contrast, 64 uncoated specimens, 33, 45, 64–65, 83–84, 171–172, 176, 179, 255, 257, 260, Color plate 3 very-low V0 SEM, 53

Index voltage contrast, 16 See also Charging visibility, 142–143 Z-contrast, 13, 15–16, 40, 172 Correlative LM/LVSEM/TEM, 172, 176–177, 190–193 fluorescent labels w/metal particles, 190–192 quantum dots, 180–182 specimen preparation, 183–184, 190–191 CPD, see Critical-point drying (CPD) Critical-point drying (CPD), 147, 149, 184, 191, 215–219, 233–234 ameloblasts, 209 artifacts, 59–60, 215–219 bacteria, 75, 77, 216, 222–226 bound water, 72 centriole/centrosome, 155 cleans specimen, 69, 74 of cryo-specimens, osteoblasts, 88, 205–206 dissection of, 204–206, 207, 209 GH3 cell, 75 importance of proper dehydration, 59–60, 147–148, 150, 152, 157, 184, 254 liver, 207 lung, 210 mitotic apparatus, 76, 154 mouse embryo, 206 nuclear-pore complex, 79–80, 153, 158, 160–161 plants, 231, 233–234, 236–237, 239, 241 platelets, human, 39, 51, 175, 179–181, 190–192 sea-urchin, embryo, 68 sperm-whale dentine, 78 tissue culture cell, 74, 89 Toxoplama gondii, 161 yeast, 219–221, 261 Cryo-SEM, 82–89, 250–251 artifacts, ice-crystal, 82, 224, 226, 230, 249–253 cryo-immobilization, 215–226 cryo-planing, 224 fracture faces, 254–255 fixed animal cells, 215–216, 224–226 focused-ion beam cutting (cryoFIB), 225–226 freeze-dried surfaces, 86, 88, 253–254 freeze-substitution, 157, 184, 191, 216, 247, 254 frozen-hydrated surfaces, 86, 88, 253 high-pressure freezing, 191, 224, 236, 243, 247, 251–252, 255, 261, 263 ice, thermal conductivity, 252

Index ice-crystal artifacts, 82, 224, 226, 230, 249–253 micro-crystalline ice, 224, 226 pre-fixation, 216 processing of frozen samples, 252 protocols, 260–263 specimen preparation, 251–252 stabilization with aldehydes, 217–218 surface coating, 256–260 thaw-fix, 88–89 thickness, properly frozen, 247, 252 Cryofixation, 215–226, 247, 251–252 Cryosputter, 253, 260 Cytoskeleton artifacts (trabeculae), 60, 184 plants, 231, 235, 237, 239 protocols, 147 results, 74, 149–151, 179–180 stabilizing buffer, 145–149, 217, 236–237, 245 See also Microtubules and Actin Cryotomy of frozen samples, 224 Cyst wall of Giardia, 33 Damage, see Specimen damage De Broglie wavelengths, 31 De-embedding, crown-ether, 155–158 Deflection, beam, 2–4, 9, 197 double-deflection scanning, 9, 13 for live-time stereo, 90–93, 198, 201–202, 206 Dehydration, 59, 147, 149–150, 161, 178, 181, 216, 218 artifact, “microtrabeculae”, 60, 184, 226 bound water, 72, 212, 215 crown-ether de-embedding, 161 incomplete, causes artifacts, 59–60, 147–148, 150, 152, 157, 184, 236, 250, 254 limitations, 60–61, 215, 218, 236, 254 plants, 233, 236 protocols, 147, 149, 161, 177 Delocalization, of SE production, 72 limit on high-resolution, 74–75 Density contrast, 30, 42, 61, 86, 259 Dental enamel, 203 Der Elektronenabtaster, 4 Detector quantum efficiency (DQE), 137–140, 143 digital detectors, 139–140 importance of, 140–141 table, 139, 141 Detectors, electron backscattered electrons (BSE), 28, 48–51, 54, 141, 222, 224, 236, 257

309 collection field, effect, 48–49, 139 DQE, 137–141, 143 Everhart–Thornley SE detector, 20, 47, 55, 57, 133, 139, 198 improvements in, 47 performance, measurement, 132 secondary electrons, 20, 47, 55, 57, 133, 139, 198 tables of, 139, 141 Detectors, x-ray microcalorimeter, 296–301 SiLi (EDS), 279–288 silicon drift detector (SDD), 293–298 wavelength dispersive (WDS), 289–292 x-ray optics augmented WDS, 298–300 Detergent extraction, 51, 81, 231, 234–235, 237 Triton X-100, 148–151, 153, 155, 161 Diamond ultra-microtome knives, 203 Diffraction, 31, 45–46, 59–60, 112–114, 117–118 correction, 120 in light microscopy, 178, 190, Color plate 2 limit, 116 pattern, 50, 124, 258 wavelength-dispersive x-ray detector, 289–292 x-ray, 250 Diffraction-limited condition, 114 Diffusion, surface, contamination, 38 Disc of least confusion (DOLC), 109 Disruption/isolation, plant cells, 235–237 Distortion, 32, 47–48, 82–84, 88 of beam, by collection field, 47–48 of beam, by spherical aberration, 109 of SE collection field, by charging, 32, 83–84, 198 of image, barrel/pincushion, 124 of image, caused by charging, 32, 83–84 of specimen, improper drying, 147, 215–219 of specimen, radiation damage, 82, 88 Double-layer coating, 221–223, 225, 247, 257–259, 261, Color plate 13 advantages, 260 DQE, see Detector quantum efficiency (DQE) tables of, 139, 141 EDS, see Energy-dispersive spectrometer (EDS) Electron, 29–30 Electron beam advantages as imaging probe, 30 current density, 15, 31, 45, 56, 83 induced conductivity, 62–63 limitations, 30–36

310 Electron beam (cont.) penetration, 11, 257, 271, 294–295 range, 32, 73, 270–273 scanner, 3–5 See also Electron source Electron lenses, characteristics of, 30–31, 122 convergence angle, 109, 112–113, 118–120, 123 distortion of, 124 magnetic, 6, 8–9, 20, 46 multipole, 115–116 retarding, 46, 53 Electron microscope development of, 2 scanning approach to, 4, 28–29 See also Scanning electron microscope (SEM), Low-voltage scanning electron microscope, and Transmission electron microscope Electron range, 19, 32, 42–43, 73, 270–273 See also Penetration Electron sources, 31–32, 45, 56, 83, 108, 117–120 Boersch effect, 32, 56 brightness, 32, 45–46 current density, 15, 31, 45, 56, 83, 108 effective source size, 107 electron-electron interactions, 56 energy spread, 46, 53, 56, 59, 110, 117–118 field-emission (FE) sources, 22, 46, 53, 56, 74, 94, 146, 171, 250–251, 256, 260 Langmuir equation, 45 lanthanum hexaboride (LaB6 ), 46, 293 monochromator, 117 Schottky sources, 46, 53, 56, 111, 117–120, 122 source size, 31, 107 thermal-field (TF), sources, 53, 56 tungsten hairpin, thermal, 31, 45, 111, 293 Enamel, dental, 201 Energy-dispersive spectrometer (EDS), 279–281 detector arrays, 296–298 detector counting rate, 295 energy resolution at MnKα , 281–282, 295 limits of detection, 285–289 “Lost” elements, 281–283 low-energy photon detection, 295–296 operating temperature, 295 peak interference, 283–285 spatial resolution, 268, 271–272, 275, 293 Energy-filtering TEM (EF-TEM), 176, 178 results, 193, Color plate 3

Index Environmental SEM, 33, 61, 123, 199–200, 207, 212, 272 early history, 16 Everhart, Thomas E., 6, 16, 131 Everhart-Thornley detector, 20, 47, 55, 57, 133, 139, 198 Faceted cPd particle, 174–175, 188, 192 Fax machine, prototype, 1 Field-emission (FE) sources, 22, 46, 53, 56, 74, 94, 146, 171, 250–251, 256, 260 early FE-SEM, 46, 56, 74 modern microscopes, 55, 59, 110, 157, 235 outlook, 93 results, 55 for video imaging, 211 Fixation, 59, 216–222, 236, 247–248 artifacts, 59–61, 217, 250, 252 buffers, see Buffers correlative LN/LVSEM/TEM, 178–181, 191 glutaraldehyde, see Glutaraldehyde Karnovsky’s, 158 maceration, 233 paraformaldehyde, 146, 158, 231, 233, 238, 239 osmium tetroxide, see Osmium tetroxide (OsO4 ) osmium-thiocarbohydrazide (OTO), 204, 206 perfusion fixation of rat, 209 plant cells, 230, 233, 236 polyethylene-glycol, 189, 240 post fixation, 81, 146, 161, 236 prefixation for cryo, 191, 216–222, 226, 233, 237, 251 protocols, 146–153, 157–158 tannic acid, 80, 146–147, 149, 152–153, 161, 183, 185, 231, 233 thaw-fix, 88–89 Focused-ion beam cutting (cryoFIB), 225–226 Fracture, to reveal internal structure carbon, 65 ceramic, 19–20 cleanliness, 67, 82 coating, 251, 262 disadvantages, 155 dry fracture, 74, 76 fracture faces, 254 freeze-fracture, 67, 74, 82–88, 155, 230–233, 235, 249, 253, 259, Color plate 13 osmotic shock, 237, 242 performed in SEM, 204 of plants, 224, 230–233, 235

Index protocol, 261–263 thaw-fix, 89 uncoated, 45, 83–84, 255, 257, 260, Color plate 3 Freeze drying, 41, 57, 59, 72, 184, 209–212, 215–217, 234, 247, 252, 253–254, 261 low-temperature, 258 in SEM, 41, 57, 209–211, 247, 252–254, 261 Freeze-fracture, 67, 74, 82–88, 155, 231–233, 234, 249, 253, 259, Color plate 13 mimicked by cryo-LVSEM, 84–87, 89, 219–220, 225, 255, 261 replicas, 67, 161, 247, 251, 254–255, 259 Freeze-substitution, 157, 184, 191, 216, 247, 254 Frog oocytes, nuclear-pore complex, 152 Giardia, parasites, 33, 35–36, 74–75, 222 Glutaraldehyde fixative, 59, 146–147, 149, 152, 217–226 correlative LM/LVSEM/TEM, 183, 191 flight muscle, 158 Giardia, 222 membrane, 60 microorganisms, 224–226 neutralization with glycine, 191 nuclear-pore complex, 80–81, 153 plants, 231, 233, 236–239 platelets, 183, 221–223 protocols, 146–147, 149, 152 Toxoplasma, 161 yeast, 218–221 Gold labels, 39, 50–51, 155, 172, 180, 223, 234, 242 GPI-IX complex, 221–222 High-pressure freezing, 191, 224, 236, 243, 247, 251–250, 255, 261, 263 of pancreas tissue, 262–263 High-voltage electron microscopy (HVEM), 60, 158–159, 161–163, 180–182, 190–193 correlative microscopy, 190–193 specimen preparation, 172 Human platelets, 39, 51, 175, 179–181, 190–192, 221–223 fixation of, 221–223 Ice crystal damage, 82, 224, 226, 230, 249–253 Image recording, video, 199 Immunolocalization, 38, 150, 152–154, 176, 216, 218, 221, 234, 236–237

311 correlative, see Correlative LM/LVSEM/TEM fluorescence microscopy, 150, 153–154 gold labeling, 155 human platelets, 221, 223 plants, 234 See also Colloidal metal labeling Insect flight muscle, 158–161 Insects and arthropods in situ experiments, 204–206 live observations, 209–211 charging when metal coat disrupted, 197 osmium-thiocarbohydrazide (OTO), 204, 206 Interaction volume, electron, 19, 40–41, 43, 45, 59, 130, 260, 271 See also Penetration Isolated organelles centrosomal material, 153–161 frog oocytes, nuclear envelope, 152 mitotic spindles, 152–153 Johnson noise, 131 Karnovsky’s, fixative, 158 Knoll, Max, 2, 3–4 diagram scanning microscope, 3, 28 Langmuir equation, 45 Lanthanum hexaboride (LaB6 ) cathodes, 46, 293 Laser confocal fluorescence microscope, 40 Lectin labels, 38, 51, 175 Lens aberrations, 30–31 Light microscopy (LM), 27, 29, 89, 153, 174, 176–177 colloidal gold specimens, 38 confocal, 40, 226, 250 correlative microscopy, 176–177, 183, 193, 226, 240 fluorescence, 37–38, 40, 150, 153–154, 177, 180, 182, 190–193, Color plate 2 quantum dot, 180–182 scanning, 2, 30 wavelength, 31 Low beam energy, see Low-voltage Low-voltage scanning electron microscope (LVSEM), see Low-voltage SEM Low-voltage microanalysis advantages of, 272–273 applications, 300–302 capabilities, 279–281 combined Si-EDS/WDS, 292–293 detector arrays, 296–298

312 Low-voltage microanalysis (cont.) detector counting rate, 295 for discriminating metal-particle labels, 172 electron scattering effects, 275–276 energy-dispersive spectrometry, (EDS), see Energy-dispersive spectrometry instrumentation, 267–268 limitations of, 277–279 limits of detection, 285–289 low-energy photon detection, 295–296 “Lost” elements, 281–283 microcalorimetry, 296–299 minimized x-ray absorption, 273–275 operating temperature, 295 particles, 300–302 peak interference, 283–286 physical basis/limitations, 289–292 rationale for, 268–271 recent instrumentation, 294 resolution at MnKα , 295 spatial resolution, improved, 272–273 silicon-drift detector (SDD), 293–294 wavelength-dispersive detector, 289–292 x-ray optics-augmented WDS, 296–298 x-ray production/spatial resolution, 271–272 Low-voltage SEM (LVSEM), 59–61 barriers to operation at low V0 , 45–46 beam-induced surface contamination, 37 colloidal metal labels, see Colloidal metal labels composition contrast, 13, 15–16, 40, 172, 176 correlative studies, 176–183 electrons as probing radiation, 29–30 first instrument, 19, 45 history, 18–20, 268–269 imaging, 45–46 instrumentation for high-resolution, 53–56 interaction volume, 19, 40–43, 45, 59, 130, 260, 271 limitations, 59–61 performance of early FE-SEMs, 56–59 re-emergence, 52–53 shape/size, 173–175 specimen preparation, see Specimen preparation stereo SEM, real-time, 89–93, 198–209 Thornley R.F.M., 6, 16, 18, 45, 69, 251 vacuum requirements, 33–36, 46, 53, 56–59, 66, 272 Low-voltage SEM of animal cells actin cytoskeleton, 147–148 ameloblasts, 209 bacteria, 75, 77, 216, 222–226

Index centriole/centrosome, 153–161 of cryo-specimens, osteoblasts, 88, 205–206 dissection of, 204–207, 209 GH3 cell, 75 liver, 207 lung, 210 mitochondria, 83, 88, 165, 235 mitotic apparatus, 76, 152–154 mouse embryo, 206 nuclear-pore complex, 79–80, 152–153, 158, 160–161 osteocytes, 149–151 plants, 231, 233–234, 236–237, 239, 241 platelets, human, 39, 51, 175, 179–181, 190–192, 221–223 sea-urchin, embryo, 68 sperm-whale dentine, 78 tissue culture cell, 74, 89 Toxoplama gondii, 148–150, 161–162 yeast, 219–221, 261 Low-voltage SEM of plant cells, 229–230 Arabidopsis thaliana, 239–241 Chara, 234, 237, 238 clathrin-coated vesicles, 234–235 coated vesicles, 235 detergent extraction, 231, 234–235, 237 diffusion of fixatives, 235 disruption and isolation, 235–237 handling, 230 HRSEM, 230–232 osmic maceration, 232–233 osmotic shock, 237, 242 plasmolysis and protoplasts, 238–240 Low-voltage SEM microanalysis, see Low-voltage microanalysis Maceration, plants, 232–233 Markers, see Colloidal metal labeling Measuring secondary electron S/N, 132–137 Microcalorimetry, x-ray detector, 296–299 resolution, 297, 298–299 Microscopy development, 15–16 rise of surface imaging, 28–29 two approaches, 27–28 Microtubules, 60, 76, 148, 150, 153–154, 169, 234–236, 237 cortical, plants, 241–242 pellicle cytoskeleton, 148 stabilizing buffer, 147, 149, 151, 183, 236, 242 Mitochondria, 83, 88, 165, 229, 235 Mitotic spindles, 76, 152–154 Monoscope, 6

Index Mouse embryo, 206 Multipole electron lenses, 115–116

313

Optical transfer function (OTF), 121–122 Optimizing spot size, 113–115 Osmic maceration, 232–233 Osmium-thiocarbohydrazide (OTO) conductive staining, 60, 204, 206, 234 density effects, 231 Osmium tetroxide (OsO4 ) fixative, 60, 146, 232–233 boric acid buffer, 207 for colloidal-gold labeling, 224 freeze-substitution, 217 maceration, plants, 231–233 membrane fixation, 60 nuclear-pore complex, 80–81, 153, 157–158, 161 osmium-thiocarbohydrazide (OTO), 204, 206, 232 perfusion fixation of rat, 209 plants, 231–232, 234, 237 platelets, 179, 221 Proteus mirabilus, 224 protocols, 146–147, 149, 152, 161, 165, 179, 183 yeast, 219–221, 261 Osteocytes, 149

parasitophorous vacuole, (PV), 162–163 Toxoplasma gondii, 146, 147–149, 157, 161–166 Particles, 171 contrast in BSE, 52 counting, early, 2 intramembrane, 67, 83, 87, 257 metal coating, 69–70 SE image of, 120 virus, 41, 217, 247, 263 x-ray microanalysis, 300–302 See also Colloidal metal particles Pellicle microtubule cytoskeleton, 148 Penetration, electron beam, 11, 15, 19, 32, 40, 42, 52, 57, 65, 139, 256–257, 271–273, 277 for BSE imaging, 139, 79 effects spatial resolution, 11, 15, 19, 32, 40, 52, 65, 256–257, 271 of electron beam, 11, 257, 271 electron range, 271 of fixative, 155 of gold labels, 174, 180 reduced by heavy-metal coating, 45, 171 in x-ray microanalysis, 294–295 Pincushion distortion, of electron lenses, 125 Plant cells, 229–243 Arabidopsis thaliana, 239–241 Chara, 234, 237, 238 chloroplasts, 229, 232–233, 235, 237, 238 coated vesicles, 235 clathrin-coated vesicles, 234–235 detergent extraction, 231, 234–235, 237 diffusion of fixatives, 235 disruption and isolation, 235–237 handling, 230 osmic maceration, 232–233 osmotic shock, 237, 242 plasmolysis and protoplasts, 238–243 Plasmolysis and protoplasts, 238–243 Poisson statistics, 31, 131–135, 256, 281 Polaroid materials, 198 Polyethylene-glycol, preservative, 189, 240 Popcorn cPd particles, 174–175, 188 Protoplasts, 230, 238–242 PV, see Parasitophorous vacuole (PV)

Paraformaldehyde, fixative, 146, 158, 231, 233, 238, 239 Parasites, 148–150, 163 on fleas, 209 Giardia, 33, 35–36, 74–75, 222 interactions, 162–166

Quantum dot, 180–182 Quantum efficiency, 137–138 detector, 29, 47, 50, 137–138 See also Detector quantum efficiency (DQE)

Nanoparticles, see Colloidal metal particles Near-field microscopy, 2 Noise, in the SEM signals, 129–143 detector, effect of, 137–139 DQE, importance of, 140–141 DQE, of digital detectors, 139–140 effect on microscope performance, 142–143 measuring noise in SE signal, 132–135 measuring signal-to-noise ratio, 135–137 origin of, 129–131 Poisson statistics, 31, 131–135, 256, 281 statistics of, 6, 50, 114, 117, 120–121, 131, 135–138, 142, 247, 257, 260, 262 NPC, see Nuclear pore complex (NPC) Nuclear envelope, 75, 79–81, 145, 152–153, 157–158 Nuclear pore complex (NPC), 79–81, 152, 153, 157–158, 160–161

314 Radiation damage, of specimen, 16, 32–33, 44, 73, 82–88, 129, 155, 171, 252–253, 259 RCA scanning electron microscope, 5–8 Resin-extracted thick sections, 155–166 Resolution, spatial, of LVSEM, 29, 31, 40, 43–44, 52, 105–115, 120–123 aberrations, 30–31, 107–111 See also Aberrations and Aberration corrector affected by electron diffusion, 40, 254 beam current limit, 108, 142 See also Electron source, brightness, and Visibility in biology, 59, 61, 72–73 coating, 61–69 contamination, 69–72 delocalization, 72, 74–75 detectors, 47, 50 diffraction, 31, 112 early SEMs, 9, 11, 13, 15–16, 19 electron optics, 46 electron sources, 56 interaction volume, 19, 40–41, 43, 45, 59, 130, 260, 271 light microscope, 12 multi-factorial, 72–73 optical transfer function (OTF), 121–122 optimizing performance, 113–115 probe diameter, 6, 8, 18, 41, 50, 59115, 142 of small features, 31, 34, 44, 52, 66, 125, 146, 157, 262 source brightness, 45 spatial frequency, 7, 44, 121–122 stray magnetic fields, 46–47, 57 visibility, see Visibility versus TEM, 67 in x-ray microanalysis, 272–273 Ris, Hans, 59–60, 74, 79–82, 146, 150, 152, 157 micrographs, 74, 79–82, 159–163 modern instruments, 53–56, 120 muscle structure, 162–163 nuclear-pore complex, 79–82, 159–160 results, 55, 121 SEM specimen preparation, 59–60, 150, 152, 157, 184 Ronchigram, for correcting aberrations, 124 Rose Criterion, 142–143 Ruska, Ernst, 2–3, 9 Scanned electronic imaging and television, 28 Scanning approach to microscopy, 28–29 invention of, 1

Index Scanning electron microscope (SEM) history, 1 Cambridge commercial SEMs, 12–15 charged particle beams, 2 commercial production of, 20–21 cutting and dissecting in, 203–205 damp-sample SEM, 199 early history, 21–22 electron beam scanner, 3–5 electron optics of, 7 electrons as probes in, 29 evolution, 45–47 freeze drying in, 41, 57, 209–211, 247, 252–254, 261 live, arthropods, 209–211 low V0 performance, 56–59 micro-imaging surfaces, 197 noise performance, 129–143 prototype, SEM1, 14–16 prototype, SEM2 (LVSEM), 16, 18, 20 prototype, SEM3, commercial, 16, 20 RCA early SEM, 5–8 scanned imaging, invention of, 1 static 2-dimensional viewing, 198 temporal resolution, 211 three-dimensional surfaces, 207–209 video-rate LVSEM, 198, 200–201 video-rate stereo SEM, 201–203 Von Ardenne’s SEM, 8–12 See also Low-voltage SEM Scanning optical microscopy, 1–2 Scanning transmission electron microscope (STEM), 5, 8–12, 22 detector layout, 54 Scanning-type microscope, 28–29 Bain, Alexander, 6 double deflection, 13 Knoll’s diagram of, 28 scanning system, 2–4, 9, 29 Stintzing, H., 2 Scherzer, Otto, 110, 115 and spherical aberration, 110 Schottky emitter, 46, 53, 111, 117–120, 122 energy spread, 117–118 Scintillator-PMT detector, 198 SE and BSE detectors, see Detectors Sea urchin, 67–68, 74–76, 83–87, 153, 155–156 Secondary electron emission (SE), 29–30, 132–134 coefficient, 4, 19, 62, 66, 132 coefficient, effective, 19, 62–63, 76 detectors, 20, 46–49, 55, 57, 133, 137, 139, 198 silicon, 138

Index SEM, see Scanning electron microscope (SEM) Shielding, magnetic, 13, 46–47, 53, 55, 123 Shot noise, 131 See also Statistical noise Signal-to-noise ratio (SNR), 6, 50, 114, 117, 120–121, 131, 135–138, 142, 247, 257, 260, 262 early SEM, 6 limits microscope performance, 135–137, 142–143 Rose criterion, 142–143 visibility, 39, 52, 142–143 Silicon drift detectors (SDD), 293–294 thickness, 294 Source, electron, see Electron sources Spatial frequency, of images, 7, 44, 121–122 Specimens, biological actin cytoskeleton, 147–148 ameloblasts, 209 bacteria, 75, 77, 216, 222–226 centriole/centrosome, 153–161 of cryo-specimens, osteoblasts, 88, 205–206 dissection of, 204–206, 207, 209 GH3 cell, 75 liver, 207 lung, 210 mitochondria, 83, 88, 165, 235 mitotic apparatus, 76, 152–154 mouse embryo, 206 nuclear-pore complex, 79–80, 152–153, 158, 160–161 osteocytes, 149–151 plants, 231, 233–234, 236–237, 239, 241 platelets, human, 39, 51, 175, 179–181, 190–192, 221–223 sea-urchin, embryo, 68 sperm-whale dentine, 78 tissue culture cell, 74, 89 Toxoplasma gondii, 145, 147–150, 161, 165 viruses, 41–42, 173, 258–259, 263 yeast, 219–221, 261 Specimen charging, see Charging Specimen damage, 32, 44, 73, 82–88, 129, 155, 171, 259 advantage of LVSEM, 82, 198 beam current density, 83 caused by fixation, 61 caused by glow-discharge coating, 68–69 contamination, see Contamination critical-point drying (CPD) cryo-techniques, 82, 252–253 dehydration, 142 depth effect, 32

315 example images, 36, 83, 85, 88 fluorescence, LM, 38 freezing artifacts, 82, 224, 226, 249–253 and image contrast, 66, 140 ionizing radiation, 32 mass loss, 259 microtrabeculae, 60, 184 radiation damage, 32–36, 44, 66–67, 73, 82–88, 129, 155, 171, 250, 257, 259, 288 reduced at low V0 , 66–67, 82–88, 257 reduced by double-layer coating, 260–261 and scan speed, 198 SEM compared to TEM, 67 theory, 32, 37, 67 tin x-ray microanalysis, 289, 292 Specimen preparation, animal cells antibody staining, see Immunolocalization and Colloidal metal labeling bound water, 72, 212, 215, 250–251 coating, see Coating CPD, see Critical-point drying (CPD) correlative methods, 183–184, 190–193 de-embedded TEM specimens, 155–166 dehydration, see Dehydration detergent extraction, see Detergent extraction effect of fixation and CPD, 216–217 fixation, see Fixation focused-ion beam cutting, 225 freeze drying, see Freeze drying nuclear-pore complex, 152, 157–159 stages of, 250–251 substrates, see Substrates thaw-fix, following cryo, 88 theory, 250–251 whole-mounts, 60, 74, 150, 152, 158–163, 172, 176, 180, 190–191, 230 Specimen preparation, plant cells, see Plant cells Spherical aberration, 31, 109–110 coefficient, 31, 109–110, 142 Scherzer’s rule, 110, 115 Stabilizing before cryofixation, 217–218, 221–222, 226 Stabilizing buffer, cytoskeleton, 145–149, 217, 236–237, 242 Stabilizing colloids, 174, 176, 189 Statistical noise in SEM signal, 31, 131, 134, 256, 281 Poisson noise, 31, 131–135, 256, 281 STERECON method, 220–222

316 Stereo SEM, 16, 41–42, 67, 200–209 high-resolution, 41–42 real-time, 89–93, 200–209 results, 92–93 X-alignment control of, 202 Stereo-tilt coils, 201 Stigmator, 15, 57 alignment, 57, 83, 91, 115, 125 for real-time stereo imaging, 91 Stintzing, H., 2 Stray field, magnetic, 46–47, 57 mains-frequency, 57 Substrates, specimen carbon films, 89–90 carbon grid, 258, 263 chip holder, 71 copper platelet, for cryo, 261 cryo-specimens, 89–90 metal-coated glass, 80, 157–158 sapphire, 221 silicon chip, 62, 66, 71, 129–130 wetting, 217 Surface coating, see Coating Synge, Edward, 1–2 Tables Conventional E0 microanalysis: electron range and x-ray production range for trace species in a carbon matrix, 271 Definition of Orders of Aberration, 10 DQE values for typical SEM BSE detectors, 141 DQE values for typical SEM SE detectors, 139 Low E0 microanalysis: total electron range and x-ray production range for various trace species in a carbon matrix, 272 Measured Performance of an OpticAugmented WDS (Parallax Research, Inc. LEXS), 296 Periodic Table showing choice of atomic shells available for microanalysis at E0 = 2.5 keV, 278 Synthesis of colloidal palladium, 187 Synthesis of cPt w/nucleating sol procedure, 187 The two basic plant tissue protocols, 231 X-ray shell choices for conventional beam energy x-ray microanalysis, 270 X-ray shell choices for low beam energy x-ray microanalysis (E0 =5 keV), 276

Index Comparison of Si-EDS and WDS characteristics, 282 Test specimens carbon, simulation, 32, 43 coating analysis, 70 gold-on-carbon, 59, 120–121 polished silicon, 129–130, 132, 141 T4 polyhead, 217 Pt on carbon, 55, 59, 65 for x-ray microanalysis, 271, 273, 284 TEM, see Transmission electron microscope Thaliana, 239–241 Thermal (tungsten) electron sources, 37 Thermal-field (TF), electron sources, 53, 56 energy spread, 117–118 Thick specimen, definition, 32 Thick-thin filaments, 60 Thornley, R.F.M., first LVSEM, 6, 16, 18, 45, 69, 251 Everhart–Thornley SE detector, 20, 47, 55, 57, 133, 139, 198 first LVSEM paper, 19, 45 Three-dimensional recording/viewing, 198, 207–209 Topographic imaging, 45–46 coding for shape, 30, 33–35 collection contrast, 40 contrast, see Contrast evolution of, 45–46 first images, 3, 7, 13 at low-V0 , 53 modulates metal coating thickness, 64 of small features, 34, 44, 52, 66, 125, 146, 157, 262 See also Resolution Topographic z contrast, 64 Toxoplasma gondii, 145, 147–150, 161, 165 cytoskeleton, 147–148, 150 internal structures, 161, 165 Transmission electron microscope (TEM), 3–4, 6, 18–20, 27 aberration correction, 121, 124 in biology, compared to SEM, 146, 162–165 cold stages, 83 collecting secondary electrons, 46–47 contrast transfer function measurement, 40 conversion to SEM, 20 correlative studies, 176, 190, 193 cryo-specimen flatness, 89–90 damage, 82, 88, 177 electron crystallography, 82, 88 energy-filtering TEM (EF-TEM), 176, 178, 190, 193, Color plate 3 fixation, plant cells, 233

Index freeze-fracture replicas, 251, 254 grazing incidence images, 13 high-voltage electron microscopy (HVEM), 60, 158–159, 161–163, 180–182, 190–193 imaging speed, 67 limitations, 147, 154 of nuclear-pore complex, 80–82 scanning transmission, 5–6, 9–10 specimen damage, 259 specimen preparation, 155, 157 specimen substrate, flatness, 89–90 surface replica specimens, 3, 7, 67, 87, 161, 216–217, 247, 251, 254–255, 259 TEM/SEM, 65, 182 test specimens, 217 whole-mount specimens, 60, 172, 176, 180, 190 Tungsten hairpin thermal electron source, 31, 45, 111, 293 Ultra-high vacuum (UHV) SEM, 4, 6, 22 sputter-ion pump, 22 Uncoated specimens, 33, 45, 64–65, 83–84, 171–172, 176, 179, 255, 257, 260, Color plate 3 yeast fracture faces, 83–84, 262 sea urchin, 83–86 Vacuum requirements, 33–36, 46, 53, 56–59, 66, 272 coating, 68–69 demountable, 6 environmental SEM, 33, 61, 123, 199–200, 207, 212, 272 for field-emission sources, 7, 53 living arthropods, 209 low-temperature freeze drying, 215, 261, 263 molecular drag pump, 57–58, 69 for Schottky TF sources, 56 for specimen coating, 68, 147 sputter-ion pump, 22 surface analysis, 13, 66, 69 system, 58 “wet-SEM”, 199 UHV SEMs, 4, 22, 37 See also Contamination Van der Waals forces, contamination, 37 Very-low V0 SEM, 53 Vibration, 123 sound shielding, 55 Video-rate image recording, 199

317 Video-rate LVSEM, 198 Viruses, 41, 258 adenovirus, 258, 263 as labels, 173 reovirus, 42 tobacco-mosaic virus, (TMV), 258–259, 263 Visibility, 39, 52, 108, 142–143 Rose criterion, 142–143 signal-to-noise ratio, 6, 50, 114, 117, 120–121, 131, 135–138, 142, 247, 257, 260, 262 Voltage contrast, 16 Von Ardenne, Manfred, 4–8, 13 early STEM, 8–12 resolution limit, 12, 40 Wavelength dispersive spectrometer (WDS) combined Si-EDS and, 292–293 physical basis and limitations, 289–292 resolution, 285, 286 x-ray optics-augmented, 296–298 WDS, see Wavelength dispersive spectrometer (WDS) Weiner filter, SE detection, 49, 55 Whole-mount specimens, 60, 74, 150, 152, 158–163, 172, 176, 180, 190–191, 230 correlative studies, 172, 176 dry-fractured cell, 74 frog oocyte nuclear membranes, 150, 152, 158 mitotic apparatus, 75–76, 152–156 plant protoplasts, 230 platelets, 39, 51, 175, 179–181, 190–192, 221–223 X-ray microanalysis detectors, see Detectors, x-ray of particles, 278, 300–302 peak-to-background, for particles, 278 See also Low-voltage microanalysis X-ray optics-augmented WDS, 296–298 YAG BSE detector, 50–51, 54, 222, 224, 236, 258 Yeast, 219–221, 261 cryo-immobilization of, 218–219 Z-contrast, 49, 59, 172 Zworykin, 5–8 early SEM, 7

Biological low voltage field emission scanning electron microscopy

Biological Low-Voltage Scanning Electron Microscopy

Scanning Auger Electron Microscopy

environmental scanning electron microscopy (VP-ESEM)

Scanning Transmission Electron Microscopy: Imaging and Analysis

Scanning Transmission Electron Microscopy: Imaging and Analysis

Scanning transmission electron microscopy : imaging and analysis

Electron Microscopy

Scanning force microscopy

Scanning Probe Microscopy

Scanning Electrochemical Microscopy

Atomic Force Microscopy Scanning Tunneling Microscopy

HandbookScanning Electron Microscopy

Low Voltage Wiring

High-Resolution Electron Microscopy

Electron Microscopy, 2nd Edition

Introduction to scanning tunneling microscopy

Particle Induced Electron Emission II

Particle induced electron emission I

Particle Induced Electron Emission I

Particle induced electron emission II

Introduction to Scanning Tunneling Microscopy

Scanning Force Microscopy of Polymers

Roadmap of Scanning Probe Microscopy

Low-Voltage Low-Power CMOS Current Conveyors

BiCMOS ULSI: Low Voltage, Low Power

Roadmap of Scanning Probe Microscopy

Electron Microscopy Methods and Protocols

Electron Microscopy Methods and Protocols

Electron Microscopy Methods and Protocols

Low-Voltage CMOS VLSI Circuits

Biological low voltage field emission scanning electron microscopy

Biological Low-Voltage Scanning Electron Microscopy

Scanning Auger Electron Microscopy

environmental scanning electron microscopy (VP-ESEM)

Scanning Transmission Electron Microscopy: Imaging and Analysis

Scanning Transmission Electron Microscopy: Imaging and Analysis

Scanning transmission electron microscopy : imaging and analysis

Electron Microscopy

Scanning force microscopy

Scanning Probe Microscopy

Scanning Electrochemical Microscopy

Atomic Force Microscopy Scanning Tunneling Microscopy

HandbookScanning Electron Microscopy

Low Voltage Wiring

High-Resolution Electron Microscopy

Electron Microscopy, 2nd Edition

Introduction to scanning tunneling microscopy

Particle Induced Electron Emission II

Particle induced electron emission I

Particle Induced Electron Emission I

Particle induced electron emission II

Introduction to Scanning Tunneling Microscopy

Scanning Force Microscopy of Polymers

Roadmap of Scanning Probe Microscopy

Low-Voltage Low-Power CMOS Current Conveyors

BiCMOS ULSI: Low Voltage, Low Power

Roadmap of Scanning Probe Microscopy

Electron Microscopy Methods and Protocols

Electron Microscopy Methods and Protocols

Electron Microscopy Methods and Protocols

Low-Voltage CMOS VLSI Circuits

Recommend Documents