Field Emission Scanning Electron Microscopy New Perspectives

SPRINGER BRIEFS IN
APPLIED SCIENCES AND TECHNOLOGY
Nicolas Brodusch
Hendrix Demers
Raynald Gauvin
Field Emission
Scanning Electron
Microscopy
New Perspectives
for Materials
Characterization
123
SpringerBriefs in Applied Sciences
and Technology
Series editor
Janusz Kacprzyk, Polish Academy of Sciences, Systems Research Institute,
Warsaw, Poland
SpringerBriefs present concise summaries of cutting-edge research and practical
applications across a wide spectrum of fields. Featuring compact volumes of 50–
125 pages, the series covers a range of content from professional to academic.
Typical publications can be:
• A timely report of state-of-the art methods
• An introduction to or a manual for the application of mathematical or computer
techniques
• A bridge between new research results, as published in journal articles
• A snapshot of a hot or emerging topic
• An in-depth case study
• A presentation of core concepts that students must understand in order to make
independent contributions
SpringerBriefs are characterized by fast, global electronic dissemination,
standard publishing contracts, standardized manuscript preparation and formatting
guidelines, and expedited production schedules.
On the one hand, SpringerBriefs in Applied Sciences and Technology are
devoted to the publication of fundamentals and applications within the different
classical engineering disciplines as well as in interdisciplinary fields that recently
emerged between these areas. On the other hand, as the boundary separating
fundamental research and applied technology is more and more dissolving, this
series is particularly open to trans-disciplinary topics between fundamental science
and engineering.
Indexed by EI-Compendex and Springerlink.
More information about this series at http://www.springer.com/series/8884

Nicolas Brodusch Hendrix Demers
•
Raynald Gauvin
Field Emission Scanning

Electron Microscopy
New Perspectives for Materials
Characterization
123
Nicolas Brodusch Raynald Gauvin
Department of Mining and Materials Department of Mining and Materials
Engineering Engineering
McGill University McGill University
Montreal, QC Montreal, QC
Canada Canada
Hendrix Demers
Department of Mining and Materials
Engineering
McGill University
Montreal, QC
Canada
ISSN 2191-530X ISSN 2191-5318 (electronic)

SpringerBriefs in Applied Sciences and Technology
ISBN 978-981-10-4432-8 ISBN 978-981-10-4433-5 (eBook)
https://doi.org/10.1007/978-981-10-4433-5
Library of Congress Control Number: 2017954879
© The Author(s) 2018

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Printed on acid-free paper
This Springer imprint is published by Springer Nature

The registered company is Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Developments in Field Emission Gun Technologies
and Advanced Detection Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1 Cold-Field Emission Technology . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 CFE-SEM for Low Voltage Microscopy . . . . . . . . . . . . . . . . . 8
2.3 Scanning Transmission Microscopy in the SEM . . . . . . . . . . . . 11
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3 Electron Detection Strategies for High Resolution Imaging:
Deceleration and Energy Filtration . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1 Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Application of Dual In-Lens Electron Detection . . . . . . . . . . . . 22
3.3 Energy filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4 Low Voltage SEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 37
4.1 Strategy of Characterization: Deceleration and Energy
Filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2 High Resolution Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.3 Low Voltage, Specimen Charging, and Material Contrast . . . . . 41
4.4 Ultra-Low Voltage SEM: Uses and Limitations . . . . . . . . . . . . 43
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5 Low Voltage STEM in the SEM . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
6 The f-Ratio Method for X-Ray Microanalysis in the SEM . . . . . . . 55
6.1 The Limits of X-Ray Microanalysis Models . . . . . . . . . . . . . . . 55
6.2 Description of the f-Ratio Method . . . . . . . . . . . . . . . . . . . . . . 55
v
vi Contents
6.2.1 f-Ratio Method for Binary System . . . . . ........... 57

6.2.2 Generalization of the f-Ratio Method
for Multi-elements . . . . . . . . . . . . . . . . . ........... 59
6.3 Examples of Quantitative X-Ray Analysis Using
the f-Ratio Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.3.1 Binary Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
6.3.2 Multi-elements Example . . . . . . . . . . . . . . . . . . . . . . . . 61
6.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
7 X-Ray Imaging with a Silicon Drift Detector Energy
Dispersive Spectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . ....... 67
7.1 X-Ray Emission Rate with Low Accelerating Voltage
and Thin Film . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.2 Comparison of Silicon Drift Detector Geometry . . . . . . . . . . . . 70
7.2.1 Solid Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
7.2.2 Takeoff Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
7.3 X-Ray Map Acquisition at High Spatial Resolution
and High Signal-to-Noise Ratio . . . . . . . . . . . . . . . . . . . . . . . . 74
7.3.1 Low Accelerating Voltage . . . . . . . . . . . . . . . . . . . . . . 75
7.3.2 Low Voltage STEM . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
7.3.3 Phase Map Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7.3.4 Removal of the Effect of Electron Channeling
on X-Ray Emission in Thin Specimens . . . . . . ....... 81
7.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....... 82
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....... 82
8 Electron Diffraction Techniques in the SEM . . . . . . . . . . . . . . . . . . 85
8.1 Electron Channeling Contrast Imaging . . . . . . . . . . . . . . . . . . . 86
8.2 Low Voltage STEM Defects Imaging . . . . . . . . . . . . . . . . . . . 89
8.3 Electron Backscatter Diffraction . . . . . . . . . . . . . . . . . . . . . . . . 91
8.4 Dark-Field Electron Backscatter Diffraction . . . . . . . . . . . . . . . 93
8.5 Transmission Forward Electron Backscatter Diffraction . . . . . . . 96
8.6 Dark-Field Imaging with a Forecaster Detector
in Transmission Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
9 Magnetic Domain Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
9.1 Type-I Contrast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
9.2 Type-II Contrast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
9.3 Type-III Contrast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Contents vii
10 Advanced Specimen Preparation . . . . . . . . . . . .......... . . . . . . 115

10.1 Surface Preparation . . . . . . . . . . . . . . . . . .......... . . . . . . 115
10.2 Surface Cleaning . . . . . . . . . . . . . . . . . . . .......... . . . . . . 119
10.3 Charging Compensation with Ionic Liquid Treatment . . . . . . . . 122
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . .......... . . . . . . 126
11 Conclusion and Perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Acronyms
a-SDD Annular SDD

BC Band contrast
BF Bright field
BSE Backscattered electron
CBED Convergent beam electron diffraction
CCD Charged coupled device
CFE Cold field emission
CNT Carbon nanotube
c-SDD Conventional SDD
CTEM Conventional TEM
DF Dark field
DSTEM Dedicated STEM
EBSD Electron backscatter diffraction
EBSP Electron backscatter diffraction pattern
ECCI Electron channeling contrast imaging
ECP Electron channeling pattern
EDS Energy dispersive spectrometry
EELS Electron energy-loss spectrometry
FE-SEM Field emission SEM
FSD Forecaster detector
g-SDD Optimized geometry SDD
HAADF High-angle annular dark field
HA-BSE High-angle BSE
HRTEM High-resolution TEM
IL Ionic liquid
IPF Inverse pole figure
LA-BSE Large-angle BSE
LLE Low-loss electron
LV-STEM Low-voltage STEM
MAC Mass absorption coefficient
ix
x Acronyms
MD Magnetic domain
MSA Multivariate statistical analysis
MWCNT Multiwalled CNT
NEG Non-evaporative getter
NOES Non-oriented electrical steel
PCA Principal component analysis
PDBSE Photodiode BSE detector
RCC Rotation contour contrast
RHEED Reflection high-energy electron diffraction
ROI Region of interest
SDD Silicon drift detector
SEM Scanning electron microscope
SE Secondary electron
SNR Signal-to-noise ratio
STEM Scanning transmission electron microscope
t-EFSD Transmission electron forward scatter diffraction
TEM Transmission electron microscope
TE Transmitted electron
TOA Takeoff angle
TWIP Twinning-induced plasticity
WD Working distance
Symbols
B Magnetic induction
b Burgers vector
C Contrast
C(i) Concentration of element i in the standard
Cc Chromatic aberration coefficient
Ci Concentration of element i in the sample
Cs Spherical aberration coefficient
d0 Initial probe diameter
dc Contribution to the probe diameter due to chromatic aberration
dd Contribution to the probe diameter due to diffraction aberration
dp Final probe diameter
E Electric field
E0 Accelerating voltage
E2 Beam energy at which neutral state is reached (high-energy side)
Edec Deceleration voltage
EL Landing voltage
F Lorentz force
fi f-ratio of element i
g Reciprocal lattice vector
I(i) Intensity of element i in the standard
Ii Intensity of phase i (imaging) or element i (X-rays)
Imax Maximum intensity level in the image
Imin Minimum intensity level in the image
Ip Electron probe current
KAB Cliff and Lorimer K factor for element pair A and B
Rm Maximum electron range
Rz Total electron range
t Time
w Deviation parameter
Z Atomic number
xi
xii Symbols
a Convergence angle
b Angle between the closest magnetic easy axis
d SE yield
DE Energy loss
eg Extinction distance of reflection g
η BSE yield
h Angle between the incident beam and a crystal lattice plane
hB Bragg angle
hi Angle of beam incidence with respect to the specimen surface
Ki Calibration factors for element i
s Acquisition time
Chapter 1
Introduction
The true scanning electron microscope (SEM) was invented in 1937 by Von
Ardenne (1938a, b, c) in the form of the scanning transmission electron microscope
(STEM). After World War II, a significant progress was made by Cambridge
University researchers to develop the first commercial SEM for bulk imaging
(McMullan 1985; Oatley and Smith 1955). Later, Crewe et al. developed a cold
field emission gun source (Crewe et al. 1968a) that soon led to the true STEM with
high spatial resolution (Crewe and Wall 1970; Crewe et al. 1968b). Since then, the
impact of SEM and STEM on modern science has been immense and nowadays it is
not conceivable to fabricate and develop new materials without the aid of scanning
electron microscopy. Its principle is quite simple, although practically less
straightforward: a focused electron beam is scanned across a specimen surface in a
raster motion. At each pixel of the image, the primary electrons interact with the
atoms of the specimen through elastic and inelastic collisions and release several
types of signals that are collected by specific electron and x-ray detectors. Pixel by
pixel, the final image is generated using the signal selected by the user, all signals
being available simultaneously with various intensities. For a general description of
the SEM technique and its optics, the reader is referred to Goldstein (Goldstein
et al. 2003) or Reimer (1998).
In the last decades, several books were dedicated to the science of SEMs, among
which the Goldstein’s (Goldstein et al. 2003; Goldstein 1975), Wells (1974),
Reimer’s (1993, 1998) books are obviously the most popular. However, scanning
electron microscopy is a highly evolving technique and the last decade have seen
important technological improvements that justify now to address a practical and
pragmatic review of what are the benefits of this new technology applied to real
materials. Recently, Bell and Erdman (2012) edited an interesting book treating of
low voltage electron microscopy in general although not uniquely focused on
field-emission SEMs (FE-SEMs). However, it documents the electron detection
technologies applied in recent FE-SEMs where the collection of the emitted
© The Author(s) 2018 1

N. Brodusch et al., Field Emission Scanning Electron Microscopy, SpringerBriefs
in Applied Sciences and Technology, https://doi.org/10.1007/978-981-10-4433-5_1
2 1 Introduction
electrons is confined inside the objective lens, thus celebrating the in-lens detectors
era either for secondary electrons (SEs) or backscattered electrons (BSEs). In
addition to their high collection efficiency, these detectors now provide dynamic
energy filtering of the collected signals through their particular geometry combined
to the application of the electro-magnetic fields of the objective lens.
Since a few years, recent developments in the field of scanning electron
microscopy have led to extending the range of application of SEM, especially in the
characterization of thin specimens with low voltage scanning transmission electron
microscopy (STEM). More specifically, Kikuchi diffraction techniques by trans-
mission emerged with the parallel work of Keller (Keller and Geiss 2012), Trimby
(2012) and Brodusch (Brodusch et al. 2013a, b) which now allows orientation
mapping and phase identification of nanomaterials with a spatial resolution close to
that obtained with a transmission electron microscope (TEM). Recently, a new type
of electron backscatter diffraction (EBSD) camera with the screen normal to the
electron beam direction was introduced on the market. By collecting the direct
beam and its surroundings, it permits to collect the major part of the forward
diffracted signal and also participates to the reduction in spatial resolution
(Fundenberger et al. 2015). Ultimately, convergent beam electron diffraction
(CBED) and lattice imaging were demonstrated when the specimen was immersed
inside the space between the upper and lower pole piece of a cold-field emission
SEM (CFE-SEM) (Konno et al. 2014; Orai et al. 2014; Sunaoshi et al. 2012, 2016).
In parallel, the x-ray microanalysis technique also benefited from important
developments. Energy dispersive spectrometry (EDS) is actually facing a revolution
with the introduction of annular (Demers et al. 2013; Zaluzec 2009) and win-
dowless (Burgess et al. 2013) silicon drift detectors (SDD). With its larger solid
angle of collection, the annular SDD permits the collection of a few millions counts
per second at best while limiting the effects of such high count rates like sum and
escape peaks. Its specific location on top of the specimen allows, in addition, a
significant reduction of the well-known shadowing effect that limits the applica-
bility of the conventional SDD detectors located on the side of the chamber. Also,
measuring the inelastic energy loss suffered by a low voltage electron beam passing
through a thin specimen by electron energy loss spectroscopy has been demon-
strated in a SEM, however with a still low spectral resolution compared to the
EELS-TEM technique (Khursheed and Luo 2005; Luo and Khursheed 2008;
Sunaoshi et al. 2016; Yamazawa et al. 2016). While still in development, these
techniques will certainly bring a new era in the spectroscopic techniques in the
SEM in a few years.
Along the course of this book, although not exhaustive, we wished to demon-
strate what can be achieved with CFE-SEMs in terms of imaging and analysis
capabilities, applied on real materials and problematics in the fields of materials and
nanoscience. This not only for education purpose, but also to set a baseline of what
is achievable routinely nowadays in a characterization laboratory equipped with
such equipment in terms of imaging and analysis capabilities.
References 3
References
Bell, D. C., & Erdman, N. (2012). Low voltage electron microscopy: Principles and applications.
USA: Wiley.
Brodusch, N., Demers, H., & Gauvin, R. (2013a). Nanometres-resolution Kikuchi patterns from
materials science specimens with transmission electron forward scatter diffraction in the
scanning electron microscope. Journal of Microscopy, 250, 1–14.
Brodusch, N., Demers, H., Trudeau, M., & Gauvin, R. (2013b). Acquisition parameters
optimization of a transmission electron forward scatter diffraction system in a cold-field
emission scanning electron microscope for nanomaterials characterization. Scanning, 35,
375–386.
Burgess, S., James, H., Statham, P., & Xiaobing, L. (2013). Using windowless EDS analysis of
45–1000 eV X-ray lines to extend the boundaries of EDS nanoanalysis in the SEM.
Microscopy and Microanalysis, 19, 1142–1143.
Crewe, A., Eggenberger, D., Wall, J., & Welter, L. (1968a). Electron gun using a field emission
source. Review of Scientific Instruments, 39, 576–583.
Crewe, A., & Wall, J. (1970). A scanning microscope with 5 nm resolution. Journal of Molecular
Biology, 48, 375–393.
Crewe, A., Wall, J., & Welter, L. (1968b). A high-resolution scanning transmission electron
microscope. Journal of Applied Physics, 39, 5861–5868.
Demers, H., Brodusch, N., Joy, D. C., Woo, P., & Gauvin, R. (2013). X-ray quantitative
microanalysis with an annular silicon drift detector. Microscopy and Microanalysis, 19,
364–365.
Fundenberger, J. J., Bouzy, E., Goran, D., Guyon, J., Yuan, H., & Morawiec, A. (2015).
Orientation mapping by transmission-SEM with an on-axis detector. Ultramicroscopy, 161,
17–22.
Goldstein, J. (1975). Practical scanning electron microscopy: Electron and ion microprobe
analysis. Berlin: Springer Science & Business Media.
Goldstein, J., Newbury, D. E., Echlin, P., Joy, D. C., Romig, A. D., Jr., Lyman, C. E., et al. (2003).
Scanning electron microscopy and X-ray microanalysis: a text for biologists, materials
scientists, and geologists, Springer Science & Business Media.
Keller, R., & Geiss, R. (2012). Transmission EBSD from 10 nm domains in a scanning electron
microscope. Journal of Microscopy. 245(3), 245–251.
Khursheed, A., & Luo, T. (2005). Transmission EELS attachment for SEM. In Proceedings of the
12th International Symposium on the Physical and Failure Analysis of Integrated Circuits,
2005. IPFA 2005.
Konno, M., Ogashiwa, T., Sunaoshi, T., Orai, Y., & Sato, M. (2014). Lattice imaging at an
accelerating voltage of 30 kV using an in-lens type cold field-emission scanning electron
microscope. Ultramicroscopy, 145, 28–35.
Luo, T., & Khursheed, A. (2008). Elemental identification using transmitted and backscattered
electrons in an SEM. Physics Procedia, 1, 155–160.
McMullan, D. (1985). Recollections of the early days of SEM in the Cambridge University
Engineering Department, 1948–53. Journal of Microscopy, 139, 129–138.
Oatley, C., & Smith, K. (1955). The scanning electron microscope and its field of application.
British Journal of Applied Physics, 6, 391.
Orai, Y., Sunaoshi, T., Okada, S., Ogashiwa, T., Ito, H., & Konno, M. (2014). Application of low
energy STEM with the in-lens cold FE-SEM. In Journal of Physics: Conference Series (Vol.
522, p. 012020).
Reimer, L. (1993). Image Formation in low-voltage scanning electron microscopy (SPIE tutorial
text Vol. TT12) (Tutorial texts in optical engineering). USA: SPIE Press.
Reimer, L. (1998). Scanning electron microscopy: Physics of image formation and microanalysis
(Springer series in optical sciences), Berlin: Springer.
4 1 Introduction
Sunaoshi, T., Kaji, K., Orai, Y., Schamp, C. T., & Voelkl, E. (2016). STEM/SEM, chemical
analysis, atomic resolution and surface imaging At 30 kV with no aberration correction for
nanomaterials on graphene support. Microscopy and Microanalysis, 22, 604–605.
Sunaoshi, T., Orai, Y., Ito, H., Ogashiwa, T., Agemura, T., & Konno, M. (2012). 30 kV stem
imaging with lattice resolution using a high resolution cold FE-SEM. In Proceedings of the
15th European Microscopy Congress, Manchester Central, United Kingdom, September
16–21.
Trimby, P. W. (2012). Orientation mapping of nanostructured materials using transmission
Kikuchi diffraction in the scanning electron microscope. Ultramicroscopy, 120, 16–24.
Von Ardenne, M. (1938a). Das Elektronen-Rastermikroskop, Praktische Ausführung. Zeitschrift
für technische Physik, 19, 407–416.
Von Ardenne, M. (1938b). Das Elektronen-Rastermikroskop, Theoretische Grundlagen. Z. Physik,
109, 553–572.
von Ardenne, M. (1938c). Die Grenzen fur das Auflosungsvermogen des Elektronenmikroskops.
Zeitschrift fur Physik, 108, 338–352.
Wells, O. C. (1974). Scanning electron microscopy. USA: McGraw-Hill.
Yamazawa, Y., Okada, S., Yasenjiang, Z., Sunaoshi, T., & Kaji, K. (2016). The first results of the
low voltage cold-FE SEM/STEM system equipped with EELS. Microscopy and Microanalysis,
22, 50–51.
Zaluzec, N. J. (2009). Innovative instrumentation for analysis of nanoparticles: The $$ steradian
detector. Microscopy Today, 17, 56–59.
Chapter 2
Developments in Field Emission Gun
Technologies and Advanced Detection
Systems
2.1 Cold-Field Emission Technology
To improve image or analysis quality a large signal-to-noise ratio (SNR) is nec-

essary for imaging and spectroscopic techniques to provide high quality and precise
measurements. However, the probe current increases as the square of the beam
diameter (Reimer 1998). Therefore, combining high spatial resolution with low
detection limits in spectroscopic techniques has always been a heartbreak, choosing
between imaging resolution or high count rates. Among the kind of SEMs gun
technologies, cold-field emission is the one providing the highest gun brightness
while reducing significantly the chromatic aberration. Thus, the CFE gun provides a
large probe current in a small probe size and tends to conciliate, to a certain extent,
the imaging and analysis SEM conditions. However, inside the gun chamber the
cold-field emitter attracts contaminants even if high vacuum is used and needs to be
cleaned with a daily flashing procedure. The intense flash then results in instability
at the surface of the emitter which in turn results in beam instability and “tip noise”
during the formation of a very thin but uniform oxide layer at its surface. This
instable regime usually begins after a few tens of minutes of high stable emission
and lasts at least one or two hours, depending on the systems. This have dramat-
ically limited the practical use of CFE-SEMs in x-ray microanalysis and orientation
mapping via EBSD, although after the unstable regime, the system may be stable
for several hours but with a significant reduction of the operational beam current.
Recently, carbon nano-tips have been demonstrated as an interesting route for
electron extraction from the emitter and higher stability and brightness were
reported (Houdellier et al. 2015). However, no commercial instrument using carbon
nano-tips is currently available. The last generation of Hitachi CFE guns were fitted
with non-evaporative getter (NEG) pumps that permits to obtain a vacuum one
order of magnitude lower than the previous generation of CFE guns (Kasuya et al.
2014). Additionally, an auto-flash system is used to clean the emitter by applying
a short mild flash after a short period of time, typically a few tens of minutes.

6 2 Developments in Field Emission Gun …
This ensures to benefit from the highest brightness at all time and thus improve,
through the brightness term, the beam diameter/probe current relation, as described
by Eq. 2.28 in Reimer’s (Reimer 1998).
Typical probe current curves, measured with a Faraday cup at the surface plane
of the specimen, are plotted in Fig. 2.1 for CFE guns with normal flashing
(SU-8000, set 1 and 2) and for an auto-flash system (SU-8230). To facilitate the
comparison, the three curves were normalized with the probe current at t = 0 s. The
measurements were recorded with E0 = 30 kV with all column parameters being
equal for the three sets of measurements. Each point corresponds to the average of
sixty measurement in one second. The curve for SU-8230 was obtained with mild
flashes automatically controlled every 65 min. For the SU-8000 SEM, two sets of
data were acquired: For set 1, the emission current was reset every 30 min (which
increase the extraction voltage of the gun) while set 2 is the observed current after a
normal flash without resetting of the emission current. Note that the data shown in
set 1 correspond to the normal operation of the typical CFE-SEM. In fact, as
deduced from the curve of set 2, the emission current drops dramatically after
approximately 30 min with a loss of current of 50% in 1 h. However, when the
emission current is reset during this sudden drop, the emission loss is kept below
30% but the unstable regime duration cannot be decreased. It has to be noted that
the normal operation procedure is to flash the emitter a few hours before the SEM is
being used in order to prevent from the unstable regime during the SEM operation.
Following this procedure, the SEM can be operated 5–6 h with an appreciable beam
current stability of approximately 3% during this period (Fig. 2.1). As one can see,
the auto-flash assisted-CFE gun do not show the dramatic loss of emission observed
with the normal flash operation and the working probe current achievable is higher
1.0 1.0
0.9 0.9
Current SU-8230 (I/I0)
Current SU-8000 (I/I0)
0.8 0.8
0.7 0.7
SU-8000 set 1
SU-8000 set 2
0.6 SU-8230
0.6
0.5 0.5
0.4 0.4
0 1 2 3 4 5 6 7 8 9 10 11 12
Time (h)
Fig. 2.1 Comparison of normalized electron probe current measured with a Faraday cup at E0 =
30 kV at the specimen plane for SU-8000 and SU-8230. The same condenser and current
parameters were kept identical in all three measurements. Each point is the average of 60 points of
1 s each. For SU-8000 curves, set 1 was obtained with resetting the emission current (Ie) every
30 min while set 2 was obtained without resetting Ie
2.1 Cold-Field Emission Technology 7
in average. The current stability is high between each mild flash. However, unex-
pected emission jumps at mild flashing time are observed and are, to this day,
unexplained. Despite the fact that the emission loss is of the order of 15% in 12 h, a
higher beam current is achieved with the auto-flash system. This definitely greatly
improves the capacity of this type of CFE-SEM to perform x-ray and EBSD
analysis in a similar manner to the Schottky FE-SEMs on long periods of time
while it was restricted to a few hours with the previous technology of CFE-SEM.
The combination of the reduction of the chromatic aberration due to the low
energy spread of the CFE gun and the improvement of the in-lens detection led to a
constant improvement of the image quality and resolution as demonstrated in
Fig. 2.2. In this figure, SE images are shown from etched Nb(CN) precipitates in
steel at an accelerating voltage (E0) of 5 kV with the Hitachi S-4500 (a), from
carbon nanotubes decorated with copper nanoparticles with the Hitachi S-4700 at
E0 = 3 kV (b) and from a lithium titanate powder at E0 = 2 kV with the Hitachi
Fig. 2.2 Evolution of the (a)

imaging capabilities of the
5 kV
CFE-SEMs through the last
30 years. a Hitachi S-4500 at
E0 = 5 kV, b Hitachi S-4700
at E0 = 3 kV and c Hitachi
SU-8230 at E0 = 2 kV,
high-resolution micrographs
(b)
3 kV
(c)
2 kV
SU-8230. All these microscopes CFE-SEMs but only the latter provides the
auto-flash system. Through this figure, twenty years of CFE microscopy is illus-
trated and one can clearly see the improvement that was achieved recently with the
advent of the auto-flash CFE-SEM.
2.2 CFE-SEM for Low Voltage Microscopy
The main improvement in the design of new FE-SEMs, and particularly

CFE-SEMs, resides in their ability to work routinely at very low voltages, i.e.,
lower than 1 kV, while keeping a small probe size at the nanometer scale. This is
achieved through the use of a deceleration voltage applied at the specimen surface
which decelerates the electron beam before it penetrates the sample. Combined to
the high brightness and low chromatic aberration provided by the CFE technology,
the spatial resolution thus obtained can be maintained close to the nanometer for
voltages down to 30–50 V. The subtle gain of spatial resolution due to the
auto-flash system can be observed in Fig. 2.3 where an alumina sphere was imaged
in deceleration mode with a landing voltage of 0.7 kV without (SU-8000, Fig. 2.3a)
and with the auto-flash (SU-8230, Fig. 2.3b) system. The same in-lens (top) de-
tector was used in both images. Visually, the image obtained with the auto-flash
option looks sharper with more details being observed compared to the image
without it. The spatial resolution was measured via the SMART-J plugin based on
the SMART program written by Joy (2002). The resulting resolution values were
2.5 and 2.0 nm for Fig. 2.3a, b, respectively, which represent an improvement of
20%.
Fig. 2.3 Effect of the auto-flash system combined with vacuum improvement in the gun chamber
through the comparison of alumina spheres at low accelerating voltage, EL = 0.7 kV. a SU-8000
(E0 = 2.5 kV, Edec = 1.8 kV) in normal mode of operation, b SU-8230 (E0 = 4.2 kV, Edec =
3.5 kV) with the auto-flash system. A spatial resolution of 2.5 and 2.0 nm were measured with the
SMART-J plugin for (a) and (b), respectively
2.2 CFE-SEM for Low Voltage Microscopy 9
The reduction of the primary electrons voltage has several advantages: First,
through the reduction of the electron penetration range, a smaller volume con-
tributes to the emitted signals, especially the BSEs and, as a consequence, the SE2
(Joy 1985; Reimer 1998). Therefore, either SEs (SE1 and SE2) or BSEs signals
benefits from a smaller depth of emission and the prime surface of the specimen can
be analyzed down to a few surface layers at ultra-low voltage (El-Gomati and Wells
2001; Frank and Mullerova 2006; Mikmekova et al. 2007, 2013, 2015) making the
FE-SEMs useful for surface analysis by imaging treated surfaces with the versatility
of the SEM. The reduction of the penetration depth is depicted in Fig. 2.4a where
the electron range was simulated based on Monte Carlo modeling using the Casino
2.42 program (Drouin et al. 2007). A thousand trajectories were simulated and
(a)
(b) Au
Si
1M
100k
10k
Low
R Z (nm)
20 kV 1k Voltage
SEM
100 VeryLow
Energy
SEM
10
100m
0.1 1 10 100 1000
2 kV
E 0 (keV)
(c) CM C
105 CG auvin
Fe MC
Fe Gauvin
104 Al MC
Al Gauvin
Ag MC
Ag Gauvin
103
b (nm)
Au MC
0.2 kV
Au Gauvin
102
101
100
0 20 40 60 80 100
0.05 kV
E0 (keV)
Fig. 2.4 a Effect of reducing the accelerating voltage on the electron diffusion volume, b range of
voltages used in high resolution SEMs and c electron beam broadening b as a function of the
accelerating voltage (E0) for thin foils of 80 nm for C, Al, Fe, Ag, Au. The results obtained by
calculations using the equation from Gauvin (Gauvin and Rudinsky 2016) (dashed lines) in
(c) were compared with those obtained using Monte Carlo simulations (full lines)
displayed with accelerating voltages of 20, 2, 0.2, and 0.05 kV in a Fe target and
the BSEs trajectories are displayed as the red lines. As can be noticed, the BSE
range falls approximately from 300 nm at E0 = 20 kV to 10, 1, and 0.2 nm for E0 =
2, 0.2, and 0.05 kV, respectively. A plot of the total electron range Rz as a function
of the accelerating voltage for Si and Au is presented in Fig. 2.4b.
These curves clearly show the proportional log/log relationship of the two
quantities, regardless of the atomic number for which the range is proportional to
the atomic number Z as Z−8/9 according to Kanaya and Okayama (1972). Secondly,
the tilt dependence of the SE emission decreases (Cazaux 2005; Reimer 1993) with
the accelerating voltage as shown in Fig. 2.5 providing a more uniform image
contrast where the edge effect is highly diminished. Combined to these two
important image-forming effects, the reduction of the beam voltage is also of great
advantage regarding the damage inflicted to the specimen by the electron irradia-
tion. The reduction of the beam voltage reduces greatly the temperature rise, which
is proportional to the beam voltage, as well as the radiolysis that is reduced when
the ionization energy is small in the irradiated volume. Especially for organic
specimens, low voltage prevents from breaking the molecular groups, which is the
main damage process observed for these materials (Reimer 1998). Finally, in some
cases, the reduction of the accelerating voltage may be accompanied with a
reduction of the charge effects. This is mostly applicable to several polymers as they
have smaller E2 voltages compared to ceramics and minerals (Joy et al. 1998; Joy
and Joy 1996). However, this must be taken with precautions because, when the
beam electron density is of the same order as that at higher voltage, the energy loss
is concentrated in a dramatically smaller volume (see Fig. 2.4a) and thus can
degrade the fragile charge balance of nonconductive materials.
Fig. 2.5 Secondary electrons 62

emission loss at a tilt angle of
SE emission loss at tilt =75° (%)
75° for Al, Cu and Au when 60

the accelerating voltage (E0)
is lowered from E0 = 10 kV to 58
E0 = 0.5 kV. Data extracted
from Reimer (Reimer 1993)
56
54
52
50
0 10 20 30 40 50 60 70 80 90
Z
2.3 Scanning Transmission Microscopy in the SEM 11
2.3 Scanning Transmission Microscopy in the SEM
Scanning transmission electron microscopy has revealed as an invaluable technique

since the seventies to characterize material science specimens in the TEM (Crewe
and Wall 1970; Crewe et al. 1968; Pennycook 1989; Pennycook et al. 1996).
However, the use of high accelerating voltages in such microscopes is highly
detrimental to the specimen stability, especially for low Z materials such as lithium
or carbon (Egerton 2012; Egerton et al. 2004). In this regard, decreasing the beam
voltage is becoming an increasing ground of research for the TEM/STEM com-
munity (Bendayan and Paransky 2014; Drummy 2014, Kaiser et al. 2011; Sasaki
et al. 2010, 2014). Especially, due to the small chromatic aberration coefficient
provided by CFE guns, high resolution CFE-SEMs and monochromator-fitted
FE-SEMs nowadays provide the stability and the probe dimensions necessary to
apply efficiently low voltage STEM in the SEM. In addition, SEM is an effective
cost reduction compared to TEM/STEMs and their ease of use make them one of
the most used technique for materials characterization around the world. Typically,
a few minutes are generally necessary to achieve the highest image quality in
STEM mode and this permits to provide high throughput for characterization
laboratories. One interesting advantage of performing STEM in a SEM is its
capability to work at low voltages with high stability, which is generally difficult to
achieve with the high voltage-designed electron columns of TEM/STEMs.
Therefore, the gain in contrast due to the increased interactions of low voltage
electrons with the specimen can be optimized for a wide range of materials,
especially low Z materials. This, combined with the limited beam broadening for
voltages of 20–30 kV (Fig. 2.4c), makes the new generation of FE-SEMs ideal
candidates for low voltage STEM and recently, atomic resolution has been reported
with a CFE-SEM (Konno et al. 2014; Orai et al. 2014; Sunaoshi et al. 2012).
References
Bendayan, M., & Paransky, E. (2014). Perspectives on low voltage transmission electron
microscopy as applied to cell biology. Microscopy Research and Technique, 77, 999–1004.
Cazaux, J. (2005). Recent developments and new strategies in scanning electron microscopy.
Journal of Microscopy, 217, 16–35.
Crewe, A., & Wall, J. (1970). A scanning microscope with 5 nm resolution. Journal of Molecular
Biology, 48, 375–393.
Crewe, A., Wall, J., & Welter, L. (1968). A high-resolution scanning transmission electron
Drouin, D., Couture, A. R., Joly, D., Tastet, X., Aimez, V., & Gauvin, R. (2007). CASINO V2.
42—A fast and easy-to-use modeling tool for scanning electron microscopy and microanalysis
users. Scanning, 29, 92–101.
Drummy, L. F. (2014). Electron microscopy of organic–inorganic interfaces: Advantages of low
voltage. Ultramicroscopy, 145, 74–79.
Egerton, R. (2012). Mechanisms of radiation damage in beam-sensitive specimens, for TEM
accelerating voltages between 10 and 300 kV. Microscopy Research and Technique, 75, 1550–
1556.
Egerton, R., Li, P., & Malac, M. (2004). Radiation damage in the TEM and SEM. Micron, 35,
399–409.
El-Gomati, M., & Wells, T. (2001). Very-low-energy electron microscopy of doped semiconduc-
tors. Applied Physics Letters, 79, 2931.
Frank, L., & Mullerova, I. (2006). The scanning low energy electron microscopy (SLEEM) mode
in SEM. Microscopy and Microanalysis, 12, 152–153.
Gauvin, R., & Rudinsky, S. (2016). A universal equation for computing the beam broadening of
incident electrons in thin films. Ultramicroscopy, 167, 21–30.
Houdellier, F., de Knoop, L., Gatel, C., Masseboeuf, A., Mamishin, S., Taniguchi, Y., et al.
(2015). Development of TEM and SEM high brightness electron guns using cold-field
emission from a carbon nanotip. Ultramicroscopy, 151, 107–115.
Joy, D. (2002). SMART—A program to measure SEM resolution and imaging performance.
Joy, D. C. (1985). Resolution in low voltage scanning electron microscopy. Journal of
Microscopy, 140, 283–292.
Joy, D. C., & Joy, C. S. (1996). Low voltage scanning electron microscopy. Micron, 27, 247–263.
Joy, D. C., Joy, C. S., et al. (1998). Study of the dependence of E2 energies on sample chemistry.
Kaiser, U., Biskupek, J., Meyer, J., Leschner, J., Lechner, L., Rose, H., et al. (2011). Transmission
electron microscopy at 20 kV for imaging and spectroscopy. Ultramicroscopy, 111, 1239–
1246.
Kanaya, K., & Okayama, S. (1972). Penetration and energy-loss theory of electrons in solid
targets. Journal of Physics D: Applied Physics, 5, 43.
Kasuya, K., Kawasaki, T., Moriya, N., Arai, M., & Furutsu, T. (2014). Magnetic field
superimposed cold field emission gun under extreme-high vacuum. Journal of Vacuum Science
& Technology B, 32, 031802.
Mikmekova, S., Yamada, K., & Noro, H. (2013). TRIP steel microstructure visualized by slow and
very slow electrons. Microscopy, 62(6), 589–596.
Mikmekova, S., Yamada, K., & Noro, H. (2015). Dual-phase steel structure visualized by
extremely slow electrons. Microscopy, 64(6), 437–443.
Mullerova, I., Matsuda, K., Hrncirik, P., & Frank, L. (2007). Enhancement of SEM to scanning
LEEM. Surface Science, 601, 4768–4773.
Orai, Y., Sunaoshi, T., Okada, S., Ogashiwa, T., Ito, H., & Konno, M. (2014). Application of low
energy STEM with the in-lens cold FE-SEM. Journal of Physics: Conference Series.
Pennycook, S. (1989). Z-contrast STEM for materials science. Ultramicroscopy, 30, 58–69.
Pennycook, S., Jesson, D., McGibbon, A., & Nellist, P. (1996). High angle dark field STEM for
advanced materials. Journal of Electron Microscopy, 45, 36–43.
Reimer, L. (1993). Image formation in low-voltage scanning electron microscopy (SPIE tutorial
text Vol. TT12) (Tutorial texts in optical engineering), USA: SPIE Press.
(Springer series in optical sciences). Berlin: Springer.
Sasaki, T., Sawada, H., Hosokawa, F., Kohno, Y., Tomita, T., Kaneyama, T., et al. (2010).
Performance of low-voltage STEM/TEM with delta corrector and cold field emission gun.
Journal of Electron Microscopy, 59, S7–S13.
Sasaki, T., Sawada, H., Hosokawa, F., Sato, Y., & Suenaga, K. (2014). Aberration-corrected
STEM/TEM imaging at 15kV. Ultramicroscopy, 145, 50–55.
Sunaoshi, T., Orai, Y., Ito, H., Ogashiwa, T., Agemura, T., & Konno, M. (2012). 30 kV stem
imaging with lattice resolution using a high resolution cold FE-SEM. In: Proceedings of the
15th European Microscopy Congress, Manchester Central, United Kingdom, September
16–21.
Chapter 3
Electron Detection Strategies for High
Resolution Imaging: Deceleration
and Energy Filtration
3.1 Principles
From the early beginning, the reflected and transmitted electrons produced after the
interaction of a high energy electron beam with a specimen surface were collected
by placing electron detectors around and below the specimen at specific locations
inside the specimen chamber (Reimer 1998; Wells 1974). Especially, a BSE
detector with a scintillator or a semiconductor was used either on top or on the side
of the specimen, providing variable collection efficiencies depending on surface tilt
(Murata 1976; Wells 1970, 1971, 1979). For SE detection, the most used electron
detector was, and still is, the Everhart-Thornley scintillator detector (Everhart and
Thornley 1960). It is generally placed on the side of the chamber and collects
mainly SEs attracted by the positively biased detector grid and BSEs depending on
the bias used, the latter signal being very weak due to the very small solid angle.
This detector is still used nowadays in every SEM as a standard SE detector.
However, because its collection angle is broad, it collects mostly the SEs produced
either by the primary electrons (SE1) or by the BSEs during their path back to the
specimen surface (SE2) and those generated by the BSEs at the surface of the
chamber walls and objects (SE3) which act as converting plates. Note that with the
Everhart-Thornley detector, the main contribution to the SE signal is that from the
SE3s (Cazaux 2004, 2005). It is now well known that SE1s carry the high reso-
lution information and SE2/3s are generally considered as background noise in SE
images (Joy 1985; Reimer 1998), although they may carry useful information in
some cases. Furthermore, Monte Carlo simulations and experiments has shown that
the SE1s were produced in a small volume around the primary beam impact point
and resulted in an emission cone of small angle around the beam axis (Reimer
1998). Typically, the lateral resolution of the emitted SE1s is of the order of the
probe size and they are emitted at a few nanometers depth (Koshikawa and Shimizu
1974) while the SE2/3s are emitted from a range similar to that of the BSEs (Reimer
1998). Thus, they are responsible of the lateral resolution loss, the depth resolution

14 3 Electron Detection Strategies for High Resolution …
being, however, identical for both SE1 and SE2/3. According to Monte Carlo
simulations (Joy 1985, 1984), the SE2/(SE1 + SE2) signal ratio could be of the
order of 50% or more depending on the material atomic number.
From this quick summary on SE emission, it is obvious that increasing the part
of SE1 s in the total SEs signal collected by the detector will enhance the spatial
resolution of SE images. Prior to the introduction of the FE-SEM, probe sizes of
several tens to hundreds of nanometers were commonly used which were of the
same order as the BSE spatial resolution, at least at medium to low accelerating
voltages. Thus, the gain would have been negligible if the SE2s/SE3s contribution
would have been attenuated and this may explain why more efforts to isolate SE1s
were produced when FE-SEMs with nanometer scale probe size emerged in the
market.
Nowadays, FE-SEMs are fitted with in-lens or through-the-lens electron detec-
tors. Their particular location inside the objective lens permits to reduce the col-
lection angle to a few degrees while increasing dramatically the SE collection
efficiency by using electro-magnetic fields to attract SEs inside the column prior to
detection (Zach and Rose 1986, 1988). More importantly, because these detectors
are no longer inside the specimen chamber, the contribution of SE3s is drastically
reduced. Furthermore, a higher signal-to-noise (SNR) ratio is obtained provided the
sample lies in the magnetic field of the pole piece as can be seen in Fig. 3.1a
(a) upper (b) top
(c) lower (d) PD-BSE
Fig. 3.1 Simultaneous imaging in a single scan with a upper, b top, c lower and d PDBSE
detectors with E0 = 30 kV, WD = 7.8 mm, Ip = 350 pA pixel dwell time = 26 µs. SNR was
calculated with the SMART-J plugin (Joy 2002) and was 264.9, 1.2, 9.2, and 216.8 for (a–
d) respectively
3.1 Principles 15
(in-lens), c (in-chamber). Additionally, the in-lens detectors, due to their specific

location inside the objective lens, can easily be energy filtered via a grid or elec-
trode biasing. This topic will be discussed further later in Sect. 3.3.
The microscopes used in this study were fitted with two in-lens detectors: The
upper detector at the top of the objective lens or snorkel lens, and the top detector
below the last condenser lens in the column. These detectors are scintillators and are
biased to collect only low-energy electrons, i.e., SEs. They thus require a system of
plates around and normal to the optic axis, for both detectors, to convert BSEs into
SEs (Reimer and Volbert 1979) to allow for their detection by the scintillators.
A schematic of the detection system of the SU8230 CFE-SEM is displayed in
Fig. 3.2. For both detectors, SEs are attracted to the scintillator detector by a high
positive bias. A system of plates, called electrodes, is placed at the bottom of the
snorkel lens and is biased as a function of the detection mode chosen by the user in
the controlling software. In the SE mode, the large angle SEs and low energy BSEs
are not detected while the SEs spiraling around the optic axis back in the column
are not affected and are then detected by the upper detector. This ensures the
detector to collect mostly SE1s that were produced very close to the electron beam
axis and to reject the SE2s/SE3s and low energy BSEs which degrades the spatial
resolution. On the contrary in the LA-BSE mode, all SEs and low and medium
angle, relative to the specimen surface, BSEs attracted inside the pole piece par-
ticipate to the detection process, i.e., SE1s + SE2s and BSEs. This explains the
higher BSE component generally observed in the resulting images. The low energy
SEs can be filtered by using an electrostatic field and are thus eliminated from the
collected signal, the other signals being unaffected. As the bias is increased, the
energy of the SEs being removed increases and the total signal fraction of BSEs,
which are unaffected, increases. At a maximum bias of 150 V, the signal collected
by the upper detector is solely composed of BSEs. In both modes, the top detector
only sees high angle, relative to the specimen surface, BSEs as SEs from the sample
surface are attracted and captured by the upper detector. By applying a bias on a
grid located below the top detector converter plates, the BSE signal is filtered by
removing the low energy component of the signal, acting as a high-pass filter.
Therefore, the signal collected with the top detector is mainly composed of BSEs
with a small angle with the beam axis, which are generally considered as the
low-loss electrons. The upper detector collects mainly SEs and BSEs with a low
and medium angle relative to the specimen surface.
However, even with the reduced chromatic aberration coefficient due to the
small beam energy dispersion provided by the CFE emitter, spatial resolution
carried by the probe diameter drops down when low beam voltages are used. This is
essentially due to the chromatic aberration which is inversely proportional to E0
and, to a lesser extent, to the diffraction aberration (Reimer 1993). Below 1 kV
accelerating voltage, a dramatic loss of spatial resolution (probe size) is observed,
typically from around 1 nm at 3–5 kV to more than 5 nm below 500 V (Fig. 3.3).
In addition, the drop of the probe current with beam voltage reduces greatly the
amount of signal reaching the in-lens detectors even with the high magnetic field
provided by the snorkel lens.
Fig. 3.2 Schematic of the detection system installed on the SU8230 CFE-SEM
One way to overcome these issues is to keep the beam at high voltage during its
path down the column of the microscope and reduce electron velocity before
interacting with the specimen surface. This is achieved by two distinct ways. One is
to decelerate electrons inside the objective lens by using an electrostatic lens after
beam acceleration through the beam booster system developed by Zeiss (Jaksch and
Martin 1995). The other solution is to apply a negative potential (Edec) to the
specimen to generate an electrostatic field at the specimen surface that decelerates
the incoming high energy electrons to the desired landing voltage EL = E0 − Edec
3.1 Principles 17
Fig. 3.3 a Beam enlargement due to chromatic (dc) and diffraction (dd) lens aberrations and
b total probe size (dp) as calculated after Reimer (1998) for assumed Cs = 1 mm, Cc = 2 mm,
DE = 0.2 eV, a = 10 mrad for a CFE gun. The total probe size increases dramatically when E0
drops below 1 kV mostly due to chromatic and to a lesser extent diffraction aberration
(Zach 1989). This technology, known as the deceleration mode, is that used in the
SU8230 and SU8000 microscopes used throughout this work. As a result, the beam
is kept at several kV before entering the specimen thus permitting to reduce the
impact of the aberrations on the final probe size when electrons are leaving the
Upper Top
(a) (b)
Normal
(c) (d)
DeceleraƟon
Fig. 3.4 Resolution test Au on C substrate specimen with E0 = 0.5 kV without (a, b) and with
(c, d) deceleration. a, c upper and b, d top detectors without energy filtration. Working distance
was 1.7 mm, Ip = 81 pA, pixel dwell time was 26 µm. Signal-to-noise ratios were calculated with
SMART-J and are reported in Table 3.1
column. The probe size when hitting the specimen surface is thus smaller than that
when using the normal mode of operation (without deceleration). The impact of
deceleration voltage on spatial resolution and signals collected by the top and upper
detectors is illustrated in Fig. 3.4 and will be discussed later in this section.
In combination to a spectacular spatial resolution improvement, the strong
magnetic and electric fields arising from the immersion lens (E B filter) provide
an improved electron collection efficiency from the two in-lens detectors (Asahina
et al. 2011; Jaksch and Vermeulen 2005; Tsurumi et al. 2010). In fact, due to the
difference in potential at the specimen surface and inside the objective lens, the
emitted electrons are accelerated back to the column, spiraling around the optic axis
of the column (Reimer 1998). However, because the deceleration mode requires to
apply a voltage at the specimen, this directly impacts the SE emission and so, acts
directly on the energy barrier necessary for the SEs to be emitted from the bulk to
the chamber vacuum. Varying E0 and Edec independently permits to attain ultra-low
landing voltages as well as controlling the ratio of SEs detected by the upper and
top detectors.
The detection process when deceleration is used is quite more complicated than
in the normal mode. Due to the potential difference between the specimen surface
and the top part of the column, all the electrons are accelerated towards both
3.1 Principles 19
detectors. Thus, the Z-contrast is lost due to the very large SE collection efficiency
but the top detector still provides high-pass filtration via the biased grid. In de-
celeration mode, the upper detector energy filtration is not available, the bottom
electrodes, used for energy filtration, are at ground voltage for the deceleration to
work.
The other side of the coin is the competitiveness for SE collection efficiency
between the in-lens and in chamber detectors. For the BSEs of energy higher than
1 kV, the deviation due to the strong magnetic field under the pole piece is small
enough to be neglected in this discussion. However, it may have its importance at
very low voltage. Because the magnetic field strength decreases when the distance
from the pole piece (working distance, WD) increases, the number of SEs collected
by the lower detector increases with larger WD. However, at the optimal WD for
in-lens detection few SEs can be attracted to the lower detector and most of them
are sucked up inside the column for detection by the upper and top detectors. Only
those emerging at low angle, relative to the specimen surface, may have a chance to
reach the lower detector. Figure 3.1 shows images from a resolution test specimen
consisting in gold nanoparticles on graphite. All four images were recorded
simultaneously from a single scan with E0 = 30 kV, Ip = 350 pA and a pixel dwell
time of 26 µs. The signals from the upper (a), top (b), lower (c) and retractable
in-chamber photo-diode-BSE (PDBSE) (d) detectors. At first glance, the upper and
PDBSE are those providing the clearest images. Signal-to-noise (SNR) ratios were
calculated using the ImageJ (Rasband) SMART-J plugin (Joy 2002) and were
264.9, 1.2, 9.2, and 216.8 for the upper, top, lower, and PDBSE images, respec-
tively. In addition to confirming the above statement, it also raises the differences in
signal collection between the upper and the top detectors in normal mode (i.e., no
deceleration mode). However, in deceleration mode, the acceleration of the emitted
SEs permits to improve greatly the signal collection from the top detector as seen in
Fig. 3.4.
As expected, the PDBSE image provides high SNR due to its large collection
angle and high channeling contrast because it collects a large part of low and
medium angle BSEs (compared to the specimen surface) (Aoyama et al. 2015;
Cazaux et al. 2013). The lower SE image also provides channeling contrast but with
a dramatically smaller SNR and the SE topography observed in the upper image is
absent. This confirms unambiguously that very few SEs reach the lower detector
and that it collects mostly SE2s and SE3s which are responsible for the BSE
contrast displayed by the detector. The fact that channeling contrast is high con-
firms, on the other hand, the low and medium exit angle of the SEs/BSEs collected
by this detector. When comparing the in-lens signals, one can note that the upper
detector has a high collection efficiency with high SNR while the signal from the
top is very weak, as explained above. The topography observed in Fig. 3.1a con-
firms that most of the SEs attracted towards the column are captured by the upper
detection system while only high angle BSEs (relative to the specimen surface)
reach the top detector (Fig. 3.1b). Compared to the PDBSE image (Fig. 3.1d), this
detector displays BSE contrast with a lower channeling component as expected
from the small collection angle (Aoyama et al. 2015). One would observe a BSE
contrast also with the upper detector if the SE filtration (high pass filter) would be
used but with a smaller SNR compared to the PDBSE image as mentioned earlier
(not shown). Note that the faint channeling contrast observed in the upper detector
image (Fig. 3.1a) was carried by the BSE component of the unfiltered signal
(SE + BSE).
To illustrate the substantial gains of running the SEM in deceleration mode, a
resolution test specimen consisting of gold islands of size smaller than 15 nm
covering a carbon substrate was imaged. The resulting images are shown in Fig. 3.4
with the upper (a, c) and top (b, d) detectors in normal (a, b) and deceleration (c, d)
modes. The accelerating voltage, E0, was 0.5 kV without deceleration while the
landing voltage, EL (with deceleration) was 0.5 kV with Edec = 1.5 kV and
E0 = 2 kV in deceleration mode, Edec being the deceleration voltage applied at the
specimen. All these images were recorded without energy filtration and a set of
images where the top detector energy filtration was applied is presented in Fig. 3.5.
The SNR was calculated with SMART-J and is reported in Table 3.1. The gain of
signal obtained when deceleration was used is striking where an improvement of 27
times was observed for the SNR of the top detector images obtained with and
without deceleration while the SNR only increase of 1.4 times for the upper detector
images. As expected from the above discussion, the signal ratio between the top and
the upper detectors increased significantly from 0.044 without to 0.82 with beam
deceleration.
The spatial resolutions measured with SMART-J suggest that a gain of around
two is obtained from the same set of images when deceleration is used, the main
gain being from the top detector (refer to Table 3.2). However, the spatial reso-
lution gains seem to be underestimated with SMART-J as we already noticed in a
Upper
(a) (b) (e)
Normal
(c) (d)
DeceleraƟon
Fig. 3.5 Resolution test Au on C substrate specimen with E0 = 0.5 kV without (a, b) and with (c,
d) deceleration with Edec = 1.5 kV. a, c upper and b, d top detectors. The bias voltage for filtration
was 300 V with and without deceleration for the top detector images and filtration bias was 200 V
for the upper filtered image of (e). Working distance was 1.7 mm, Ip = 81 pA, pixel dwell time
was 26 µm. Signal-to-noise ratios were calculated with SMART-J and are reported in Table 3.1
3.1 Principles 21
Table 3.1 Signal-to-noise ratio (SNR) calculated with the SMART-J plugin (Joy 2002) in normal
and deceleration modes for the upper and top detectors images shown in Figs. 3.4 and 3.5 with and
without energy filtration
Signal-to-noise ratio Standard deviation
Figure 3.4
Normal, upper (no energy filtration) 20.62 1.00
Normal, top (no energy filtration) 0.90 0.12
Deceleration, upper (no energy filtration) 29.77 3.09
Deceleration, top (no energy filtration) 24.27 2.06
Figure 3.5
Normal, upper with top energy filtration 17.33 0.69
Normal, top with energy filtration 0.59 0.03
Deceleration, upper with top energy filtration 27.85 2.97
Deceleration, top with energy filtration 2.57 0.23
Normal, upper with energy filtration 2.97 0.32
SNR was taken as the average of SNRs of three images acquired from three distinct regions of the
specimen to avoid contamination issues for each SEM condition
Table 3.2 Spatial resolution calculated with the SMART-J plugin (Joy 2002) in normal and
deceleration modes for the upper and top detectors images shown in Figs. 3.4 and 3.5 with and
without energy filtration
Spatial resolution Standard deviation
Figure 3.4
Normal, upper (no energy filtration) 3.28 0.18
Normal, top (no energy filtration) 5.46 0.32
Deceleration, upper (no energy filtration) 2.32 0.02
Deceleration, top (no energy filtration) 2.52 0.10
Figure 3.5
Normal, upper with top energy filtration 3.01 0.08
Normal, top with energy filtration 5.75 0.13
Deceleration, upper with top energy filtration 2.34 0.07
Deceleration, top with energy filtration 3.42 0.09
Normal, upper with energy filtration 3.93 0.13
Spatial resolution values were taken as the average of the values of three images acquired from
three distinct regions of the specimen to avoid contamination issues for each SEM condition (same
images as in Table 3.1)
previous study (Brodusch et al. 2013) and visually the gains seem greater than this.
Surprisingly, one can also notice a more pronounced BSE Z-contrast provided by
the upper and top detectors without sample bias while it is weaker with decelera-
tion. It is assumed that this is related to the increase of the SE yield due to the
acceleration from the surface of the specimen to the column that reduces the energy
necessary for the SEs to be emitted into the vacuum when deceleration is used.
Therefore, the increased contribution of low energy SEs that would have been
absorbed or back reflected into the bulk of the specimen overwhelms the Z-contrast
observed without deceleration in Fig. 3.1a, b. Note that the BSE contrast is
recovered when energy filtration of the top detector is used. By comparing the
upper and top detectors in normal mode it is observed that the top might have a
higher Z-contrast than the upper detector. As mentioned earlier, the SEs in normal
mode are attracted to the upper detector which is the first detector “viewed” by the
electrons attracted back to the column. Then, the high angle and high energy BSEs
left un-deviated hit the top detector collector plates prior to SE detection which
provides a BSE type contrast (Reimer and Volbert 1979). In consequence, the top
detector always demonstrate a higher BSE/SE ratio compared to the upper detector
in normal mode, the SNR always being higher with the upper detector.
However, when deceleration is used, the detected signals from the two detectors
are different. First, the top detector SNR increases depending on the sample bias
voltage (see Fig. 3.6), more electrons being accelerated back to the column at high
Edec, and both detectors benefit from an improved SNR when a small WD and
medium to high Edec are used. Secondly, due to the high velocity of the electrons
travelling back to the column, electrons, SE in majority in addition to the BSEs, are
less attracted by the upper detector field and manage to reach the top detector
collector plates and detector. This clearly results, as shown in Fig. 3.4d, in an
increase of the topographical contrast at the cost of the BSE Z-contrast being
reduced and, at the end, both the upper and top detectors images look very similar
in terms of contrast. To bring the BSE contrast out of the top detector image with
deceleration, one must use energy filtration as will be discussed later in this section.
3.2 Application of Dual In-Lens Electron Detection
Due to the specific location of the collector plates used to collect the BSE at the top
detector, the top detector gives an exact projection of the surface similar to the view
one has from an airplane when looking down at the Earth ground. Although it can
be misleading sometimes if another detector is not used simultaneously, this type of
projection is very useful to reveal the interior features of hollow objects parallel to
the beam. This is exemplified in Fig. 3.7a, b where SBA15 mesoporous silica was
imaged in deceleration mode with the upper (a) and top (b) detectors. In this case,
using only the top detector image was misleading because although the spatial
resolution was high, the topography of the surface was difficult to assess clearly.
With the aid of the upper detector image the true surface relief was unambiguously
revealed with the right and left sides of the image being the inclined walls of the
particle with increasing z (normal to the image plane) and the middle parts being at
an intermediate depth between the top (bottom of the image) and the bottom (top of
the image) of the particle. However, as shown in Fig. 3.7c, d, specific location of
the top detector can solve the difficult problem of having no signal recorded with
the upper detector mostly due to shadowing effects. In Fig. 3.7c, a broken section of
3.2 Application of Dual In-Lens Electron Detection 23
Upper Top
(a) (b)
Bias = 0 V
(c) (d)
Bias = 120 V
(e) (f)
Bias = 1000 V
(g) 20
18
16
14
SNR
12
10
8
Upper SNR
6 Top SNR
4
0 200 400 600 800 1000
Specimen bias voltage (V)
Fig. 3.6 Effect of specimen bias (Edec) voltage on a resolution test Au on C substrate specimen
with E0 = 2 kV. Examples of upper and top images (without filtration) with Edec = 0 (a, b), 120
(c, d) and 1000 (e, f) V (g). Plot of the SNR calculated with the SMART-J plugin as a function of
Edec for the upper and top detectors
(a) (b)
(c) (d)
Fig. 3.7 Comparison of images recorded with the upper and top detectors at low accelerating
voltage in deceleration mode and effect of the collection angle of the two detectors. EL = 1 kV
(E0 = 4.5 kV, Edec = 3.5 kV) and EL = 0.5 kV (E0 = 2.0 kV, Edec = 1.5 kV) for (a, b) and (c, d),
respectively. No filtration was used for (a–c) and DE < 75% for (d). a, b SBA15 mesoporous
silica with 10 nm channels, c, d CNT with iron seeds in the core of the tube
a CNT was imaged with the upper detector and the SE image provided a very weak
image of the tube interior. On the contrary, the top detector image, which was
energy filtered to keep only BSEs with a maximum energy loss of 25% of the
incident energy, revealed a high contrast from the interior of the tube which was
found to be filled with an iron seed resulting from the growth process. Because of
the orientation of the tube, most of the SEs emitted with low and medium angles
were absorbed by the tube wall and resulted in shadowing. The high angle filtered
BSEs captured by the top detector, due to their high energy and high emission angle
(relative to the specimen surface), were not absorbed by the tube walls and con-
tributed to the final BSE image significantly, providing combined topographical and
Z-contrast.
In Fig. 3.8, a Li2FeCoSiO4 nano-powder used as lithium-ion battery cathode
material was imaged at E0 = 3 kV with the upper (a), top (b) and PDBSE (c)
detectors with a single scan. The powder consists of the iron cobalt lithium silicate
appearing dark while the bright grey particles were identified as iron/cobalt oxides
in BSE Z-contrast images. The upper detector was energy filtrated with a bias of
30 V to remove low energy SE and increase the ratio of BSE in the image. The top
detector was filtered with a bias of 1.5 kV to remove low energy BSEs. The PDBSE
provided a large solid angle BSE image with no filtration. The upper detector image
3.2 Application of Dual In-Lens Electron Detection 25
(a) (b) (c)
Fig. 3.8 Comparison of the BSE detection from the upper (a) and top (b) in-lens electron
detectors and the PDBSE retractable solid-state BSE detector (c). The sample was a Li2FeCoSiO4
powder for lithium-ion battery cathodes with E0 = 3 kV. Upper detector bias was 30 V and top
filtration bias was 1.5 kV, WD = 7.8 mm
showed little Z-contrast as expected from the mix of SE/BSE signals whereas the
PDBSE provided high BSE Z-contrast but with a significant topographical com-
ponent due to the large collection angles of the PDBSE detector (Aoyama et al.
2015). On the other hand, the top detector, due to its special “mirror” projection,
provided an image with high BSE Z-contrast with no topographical contrast. This is
obvious when looking at the top right iron oxide particle cluster which did not show
pronounced Z-contrast in the upper and PDBSE detectors images while in the top
detector image this feature had a higher contrast. This is of high importance because
the analysis of powders in the SEM are always performed on rough particles
aggregates for which it is difficult to achieve high Z-contrast across the whole
aggregate with varying height amplitude. The use of a top type detector thus allows
to keep maximum Z-contrast along the whole image regardless of the height
variation along the optic axis and to characterize more accurately the nano-phase
distribution.
The combined use of the upper and top detectors is of high importance as it
allows characterizing the surface topography as well as its chemistry via the
Z-contrast provided by BSEs. Furthermore, if this capability is combined to the
ability of the SEM to use a wide range of accelerating voltages, surface and
in-depth information of great value can be obtained and linked together to provide a
description of the specimen true nature. In Fig. 3.9, the same SBA15 specimen as
used in Fig. 3.7a, b was impregnated with gold nanoparticles and imaged at 1 kV
(Fig. 3.8a, b) and 15 kV (Fig. 3.8c, d) in normal mode with the upper (Fig. 3.8a, c)
and top detectors (Fig. 3.8b, d). The upper detector signal was filtered with a bias
voltage of 15 and 7.4 V at low (Fig. 3.8a) and high (Fig. 3.8c) accelerating voltages
respectively, to remove low energy SEs which were more sensitive to charge
effects. The top detector was also filtered to collect only BSEs having retained at
least 15% (Fig. 3.8b) and 20% (Fig. 3.8d) of their primary beam energy. When low
primary beam voltage was used, the surface of the specimen cross-section was
clearly characterized with channel steps producing a high topographic contrast with
the upper detector. At this accelerating voltage, only a few Au nanoparticles were
observed at the surface on the filtered top detector image in (Fig. 3.8b). On the
other hand, when a high beam energy was applied, the real distribution of the gold
(a) (b)
(c) (d)
Fig. 3.9 Example of depth of analysis as a function of beam voltage with E0 = 1 kV (a, b) and
E0 = 15 kV (c, d). The sample was SBA15 mesoporous silica decorated with Au nanoparticles.
The SE signal from the upper detector was filtered in (a) with a bias voltage of 15 V while 7.4 V
was used for the upper detector image in (c). The top detector was biased to collect only BSEs with
energy at least 15% in (b) and 20% in (d) of their primary beam energy. The combined use of low
and high beam energy allowed for determining the location of the Au nanoparticles which were
found filling the channels instead of covering the surface. The working distance was 4 mm
nanoparticles was revealed, either with the filtered upper or the top detectors im-
ages. This clearly shows the in-depth distribution of these nanoparticles, thus filling
the channels inside the mesoporous silica network. Moreover, due to the high
Z-contrast obtained from the top detector images, a 3D distribution of the Au
particles as a function of E0, corresponding to the probed specimen depth, could be
obtained. Note, however, that the upper detector image shows very low topographic
contrast due to the increased primary electron diffusion length at E0 = 15 kV
compared to 1 kV as well as the decrease of the SE yield with increasing the
accelerating voltage.
3.3 Energy filtration
Since the beginning of scanning electron microscopy, energy filtration of the

backscattered electrons has been of great interest. The emission depth of the
low-loss electrons (LLEs) is dramatically smaller than that of the total BSEs signal,
3.3 Energy filtration 27
because they suffered a small amount of energy loss during their diffusion in the
solid. Wells reported a emission depth smaller than 10 nm in aluminum with an
energy loss (DE) of 400 eV with an primary electron voltage of 15 kV (Wells
1974). In his model, Wells explains the surface shallowness of the low-loss image
from the fact that LLEs are mostly BSEs having undergone a single large angle
scattering event before being backscattered through the surface (Wells 1974). He
designed an in-chamber BSE detector permitting to perform energy filtering at high
sample tilt angles (Wells 1971). A scintillator was used to collect the SEs/BSEs and
two spherical grids were placed in front of it and depending on the voltages applied
to the grids, SEs, BSEs, or low-loss BSEs were collected. The origin of low-loss
signal is not completely understood in regard to the in-lens detectors used nowa-
days. In fact, the contrast obtained by Wells was highly dependent on the takeoff
angle of the detector (Wells 1979) and the specimen was generally tilted to 30° and
more, facilitating the emission of BSEs from a reduced interaction volume, thus
biasing in a way the conclusions he did at that time. Notwithstanding, he demon-
strated the usefulness of using the LLEs as high quality information carriers in the
SEM. In particular, he achieved nanometer resolution imaging of various type of
specimens and obtained resolutions comparable or better than SE images.
Moreover, probably due to the tilted geometry of the specimen he used, the to-
pography of the images he obtained was even more pronounced than that from the
SE images. At the same time, Wells demonstrated the use of the objective lens
magnetic field to deflect low energy electrons prior to their detection with the
scintillator detector (Wells et al. 1973; Wells 1979) and a spatial resolution of 3 nm
was obtained experimentally. Later on, he reported high surface sensitivity and high
spatial resolution of a LLE image obtained by placing the specimen at the center of
the objective lens (Wells et al. 1990). The strong magnetic field around the spec-
imen permitted the deflection of the low energy BSEs while the high energy BSEs,
i.e., the LLEs, where collected by a moveable collector located horizontally in the
pole piece, close to the specimen. The detector position was thus used as an energy
window selector.
Because LLEs are a small fraction of the BSE signal, low-loss imaging is
effective wherever BSE imaging is used. For example, modeling of the electron
channeling contrast showed an increase of contrast from 1.6 to 42% at E0 = 20 kV
(Joy et al. 1982). This was experimentally confirmed by Wells (1974) and Berger
et al. demonstrated that the anomalous absorption and transmission effect respon-
sible of the contrast observed in electron channeling patterns was mostly carried by
the LLEs (Berger and Niedrig 2002). Later, Eades and co-workers reported that
only BSEs with DE of a few percent provided useful contrast in electron backscatter
electron diffraction patterns (Bhattacharyya and Eades 2009; Deal et al. 2008). The
electron channeling contrast imaging was also greatly improved by using the LLEs
collected by the in-lens BSE detectors with energy filtration (Jaksch 2008b, 2012b).
Magnetic contrast of type-II may also be improved by using LLEs with DE < 30%
(Newbury et al. 1973, 1976) although experimental images were not reported yet.
However, among the application fields of BSEs, Z-contrast has been the most
explored in the last three decades. Merli and co-workers used Monte Carlo
modeling applied to AlAs/GaAs superlattices and reported a significant increase of

the Z-contrast when DE < 30% was used. Recently, a silicon drift detector was used
to collect the BSE energy spectrum of varying materials by removing the electron
trap of the detector and converted it to an energy filter to acquire energy-filtered
BSE images (Agemura et al. 2011). They thus confirmed experimentally the
improved Z-contrast as a function of the energy window and that BSEs with DE of
a few percent provided the highest contrast. In-lens detectors also provided
improved Z-contrast of low Z materials with energy filtration (Jaksch 2008a).
Furthermore, low primary beam energy combined with energy filtration provided
high contrast originating from materials atomic properties like hybridization and
band gaps (Jaksch 2011, 2012a) or surface chemical differences of the cell con-
stituents (Kim and Jaksch 2009). To summarize, the LLEs signal has been shown to
be highly surface and chemical state sensitive and results in a dramatic increase in
BSE contrast such as Z, channeling, or magnetic contrasts.
In high-resolution FE-SEMs, in-lens energy filtration of the BSE signal is
generally done by placing a grid below the BSE detector or the collector grids, as in
the SU-8230 used in this study. A negative bias is applied to the grid to repeal the
low energy electrons and define the lower limit of the energy window of the BSEs
to be collected by the top detector. The upper detector that collects both SEs and
BSEs is also filtered when the collector electrode on the sides of the snorkel lens
walls are positively biased to attract low energy SEs on its surface. The resulting
signal is high-pass filtered with a filter bias voltage ranging from 0 to 150 V and a
100% BSE signal is obtained when the high-pass filter is set to 150 V.
Secondary and backscattered electron energy distributions of C, Fe, and Pt are
displayed in Fig. 3.10a, b, respectively. The former was calculated by applying
Eq. 4.30 in Reimer textbook (Reimer 1998) with the y-axis expressed in arbitrary
units and the latter was obtained by Monte Carlo modeling of 5 106 electron
trajectories using CASINO 2.4 (Drouin et al. 2007) with E0 = 3 kV. The contrast
between Pt and C was calculated using Eq. 3.1 with IA and IB being the integrated
intensity of SEs or BSEs from the high-pass cut-off energy to the maximum energy
(primary energy) for Pt and C, respectively.
IA IB
C¼ ð3:1Þ
IA þ IB
As observed from Fig. 3.10a, the SEs of energies between 5 and 20 V are those
providing the highest SE contrast for these materials. On the other hand, as already
reported (Agemura et al. 2011), the contrast increases monotonically with the energy
cut-off of the BSE distribution in Fig. 3.10b and the gain of contrast is more effective
for DE < 30%, confirming the findings of Merli et al. (2001). Consequently, the
contrast will be higher if small energy loss is used. To investigate the LLE signal
fraction, the maximum depth of emission and the BSE yield of the BSE signal
filtered fraction was calculated from Monte Carlo simulations of 2 106 electron
trajectories at normal incidence for E0 = 3 kV with MC X-ray (Gauvin and Michaud
2009) and for the same three elements. The results are shown in Fig. 3.10c, d. The
(a) (b)
100 0.10 15000 0.8
0.08 C
0.06 12500 Fe 0.6
80
0.04 Pt
10000 Contrast Pt/C 0.4
60 0.02
Contrast
DESE /DE
Contrast
C
NBSE
0.00 7500 0.2
Fe
40 Pt -0.02
Contrast Pt/C -0.04 5000 0
20 -0.06
-0.08 2500 -0.2
0 -0.10
0 5 10 15 20 25 30 0 -0.4
0 0.5 1 1.5 2 2.5 3
SE energy (eV)
BSE energy (keV)
(c) (d)
80 0.5
70
C 0.45 C
Fe 0.4 Fe
60 Pt Pt
0.35
Max Z (nm)
50 0.3
BSE yield
40 0.25
30 0.2
0.15
20
0.1
10 0.05
0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Energy loss (% of E0) Energy loss (% of E0)
Fig. 3.10 Energy distributions of SEs (a) and BSE (b) for carbon, iron, and platinum. The SE
energy distribution was calculated using Eq. 4.30 in Reimer textbook (Reimer 1998). The BSE
energy distribution was obtained computing 5 106 electron trajectories by Monte Carlo
modeling with CASINO 2.4 (Drouin et al. 2007). The contrast (black curve) was calculated with
Eq. 3.1 using the cumulative intensities as a function of the high-pass filter cut-off value (DE).
c Maximum emission depth and d BSE yield of the filtered BSE fraction as a function of the
percentage of energy loss (DE). c, d were obtained by simulating 2 106 electron trajectories at
normal incidence with E0 = 3 kV with MC X-ray (Gauvin and Michaud 2009) and shows a
dramatic gain in spatial resolution of the filtered BSE signal, especially with DE < 10%
maximum depth of emission and the BSE yield decrease slowly with increasing the
cut-off energy but drop more significantly for DE < 30%. Especially, the emission
depth and radial spread fall from approximately 72 and 136 nm without filtration,
respectively, to 6 and 12 nm with DE = 5% for carbon. This a reduction of the order
of 11–12 times in spatial resolution while the BSE signal dramatically dropped by
94.2% (Fig. 3.10d). This loss of signal slightly decreases when increasing Z but the
gain of spatial resolution is to a much lesser extent dependent on Z. However, one
has to keep in mind that fine structures of the low-loss region of the spectrum were
not reproduced with the Monte Carlo program used here and even higher contrasts
may be expected. In addition, because the SE yield is at its highest at low energies
(see Fig. 3.10a), the loss of collected signal is expected to degrade rapidly the SNR
when cut-off energy is increased. However, the contrast increases at the same time
and if sufficient beam current is used, the expected contrast is attained with
acceptable SNR (Merli et al. 2001).
To monitor the effects of the high-pass filter for both detectors, cut-off energy
series were acquired with both detectors from a sample of CNTs covered with Pt
nanoparticles. The SNR and spatial resolution were calculated for each image with
the SMART-J plugin (Joy 2002) and the results are reported in Tables 3.3 and 3.4.
Figure 3.11 shows images recorded with E0 = 3 kV with the top detector for cut-off
values of 0 kV (a), 1.5 kV (b), and 2.7 kV (c) and with the upper detector with
cut-off values of 0 V (a), 7.5 V (b), and 22.4 V (c). The SNR and spatial resolution
results were plotted in Fig. 3.12 and the complete set of the upper detector filtered
images are given in Fig. 3.13. As the top detector cut-off energy is increased, the
surface topographical contrast decreases until it vanishes at high energy cut-off
(DE = 10%) as seen when comparing pixels from the CNTs and the carbon sub-
strate underneath, as already observed in Fig. 3.8. At the same time, the Z-contrast
is improved and both SNR and spatial resolution do not vary significantly
(Table 3.3; Fig. 3.12c), following no specific trend. This confirms, as underlined
previously, that the signal collected by the top detector without beam deceleration is
mostly composed of BSEs.
On the other hand, as seen from Figs. 3.11d–f and 3.13, the contrast between the
CNTs and the Pt nanoparticles was significantly improved when the cut-off energy
of the upper detector was higher than 3 V and reached the BSE Z-contrast at
high energy cut-off. This was expected because at energy cut-off values higher than
50–100 V, all secondary electrons are repealed from the detector and only the BSEs
Table 3.3 Signal-to-noise ratio and spatial resolution as calculated using the SMART-J plugin
(Joy 2002) for the top detector as a function of the detector bias voltage
Bias voltage (kV) Resolution (nm) SNR
0.0 4.3 5.6
1.5 4.4 8.4
2.7 4.1 6.2
Table 3.4 Signal-to-noise Upper detector Bias SNR Spatial

ratio and spatial resolution as voltage (V) resolution
calculated using the (nm)
SMART-J plugin (Joy 2002)
Upper Top Upper Top
for the upper and top
detectors as a function of the 0.0 225.2 22.2 2.7 4.5
upper detector bias voltage 3.0 258.8 24.3 2.5 4.0
4.5 243.6 23.2 1.9 4.5
7.5 205.9 24.2 3.2 4.8
10.5 154.8 23.9 3.3 4.2
14.9 132.3 23.5 3.2 4.3
22.4 109.9 23.9 2.9 4.5
150.0 64.0 22.8 3.2 4.9
(a) (b) (c)

Top
(d) (e) (f)

Upper
Fig. 3.11 Effect of energy filtration on images recorded with the top and upper detectors. The
high-pass cut-off voltages were 0 kV (a), 1.5 kV (b), and 2.7 kV (c) for the top detector and 0 V
(d), 7.5 V (e), and 22.4 V (f) for the upper detector (full set of images given in Fig. 3.13)
are captured. At the same time, the topographical contrast carried by the SEs fades
out when the energy cut-off is increased from 0 to 150 V as seen Fig. 3.13i, j
although the spatial resolution calculated with SMART-J did not change signifi-
cantly (Table 3.4; Fig. 3.12b). On the other hand, the cut-off energy impacted the
SNR calculated as seen in Table 3.4 and Fig. 3.12a where a drop of 31.3% was
observed from cut-off energies of 0–10.5 V and 71.6% from 0 to 150 V.
Notwithstanding, the SNR at high energy cut-off was still acceptable to provide
high quality images as seen from Fig. 3.13i, j. In addition, one can notice the small
shining point in Fig. 3.11d on top of the big Pt particle near the center of the field of
view. It is obviously a charging effect and it was rapidly removed when the cut-off
energy of the upper detector was slightly increased (compare Fig. 3.13a–c). This
shows clearly that low SEs are mostly responsible for charging effects and based on
our experience of filtration, most charging issues, if not too severe, can be removed
by using cut-off energies of a few volts. This holds also for the voltage contrast that
disappears with a small cut-off filtration voltage and this observation is consistent
with the fact that low energy electrons are more affected by the internal electrostatic
fields, like those resulting from charge and specimen biasing, present in the
emission volume and at the surface of the specimen.
An important question resides in the impact of the high-pass filtering of the
upper detector, which is “seen” first by the electrons flying back to the column, on
the image quality obtained simultaneously with the top detector. The SNR and
spatial resolution of the upper and top detectors images were calculated identically
on the same set of scans with the same SMART-J parameters and the numerical
values are reported in Table 3.4 and Fig. 3.12a, b. It is clear from these results that
the SNR and spatial resolution obtained with the top detector are not significantly
different when the upper detector cut-off energy is varied and thus the energy
filtration of the upper detector as no or negligible effect on the top detector images
in normal mode (i.e., without deceleration).
Fig. 3.12 Signal-to-noise (a) 350 25

ratio (a, c) and spatial 24
resolution (b, c) as a function 300
23
of the bias voltage of the
upper (a, b) and top 22
250
SNR (upper)
(c) detectors. a, 21
Upper
SNR (top)
b Simultaneously captured 200 20
upper and top images, c top Top
detector images only 19
150
18
17
100
16
50 15
0 25 50 75 100 125 150
Bias voltage (V)
(b) 8.0
Upper
7.0
Top
Spatial resolution (nm)
6.0
5.0
4.0
3.0
2.0
1.0
0.0
0 25 50 75 100 125 150
Bias voltage (V)
(c) 5.0 9.0

8.0
4.8
7.0
Spatial resolution (nm)
4.5 6.0
5.0
SNR
4.3
4.0
4.0 3.0
Resolu on (nm)
2.0
3.8 S/N ra o
1.0
3.5 0.0
0 0.5 1 1.5 2 2.5 3
Bias voltage (kV)
(a) (b)
(c) (d)
(e) (f)
(g) (h)
(i) (j)
Fig. 3.13 Upper detector images acquired with increasing the high-pass energy filtration bias.
a 0.0 V, b 3.0 V, c 4.5 V, d 7.5 V, e 10.5 V, f 14.9 V, g 22.4 V, and h 150.0 V. The specimen
was CNTs covered with Pt nanoparticles and E0 = 1 kV. i and j are zoomed areas extracted from
(a) and (h) to show the loss of topographic contrast
References
Agemura, T., Nomaguchi, T., & Joy, D. (2011). Digital BSE imaging on SEMs. Microscopy and
Microanalysis, 17, 914.
Aoyama, T., Nagoshi, M., & Sato, K. (2015). Quantitative analysis of angle-selective
backscattering electron image of iron oxide and steel. Microscopy, 64(5), 319–325.
Asahina, S., Uno, S., Suga, M., Stevens, S. M., Klingstedt, M., Okano, Y., et al. (2011). A new
HRSEM approach to observe fine structures of novel nanostructured materials. Microporous
and Mesoporous Materials, 146, 11–17.
Berger, D., & Niedrig, H. (2002). Energy distribution of electron backscattering from crystals and
relation to electron backscattering patterns and electron channeling patterns. Scanning, 24, 70–
74.
Bhattacharyya, A., & Eades, J. A. (2009). Use of an energy filter to improve the spatial resolution
of electron backscatter diffraction. Scanning, 31, 114–121.
Brodusch, N., Demers, H., & Gauvin, R. (2013). Dark-field imaging of thin specimens with a
forescatter electron detector at low accelerating voltage. Microscopy and microanalysis: The
Official Journal of Microscopy Society of America, Microbeam Analysis Society,
Microscopical Society of Canada, 1–10.
Cazaux, J. (2004). About the role of the various types of secondary electrons (SE1; SE2; SE3) on
the performance of LVSEM. Journal of Microscopy, 214, 341–347.
Cazaux, J., Kuwano, N., & Sato, K. (2013). Backscattered electron imaging at low emerging
angles: A physical approach to contrast in LVSEM. Ultramicroscopy, 135, 43–49.
Deal, A., Hooghan, T., & Eades, A. (2008). Energy-filtered electron backscatter diffraction.
Ultramicroscopy, 108, 116–125.
Drouin, D., Couture, A. R., Joly, D., Tastet, X., Aimez, V., & Gauvin, R. (2007). CASINO V2. 42
—A fast and easy-to-use modeling tool for scanning electron microscopy and microanalysis
users. Scanning, 29, 92–101.
Everhart, T., & Thornley, R. (1960). Wide-band detector for micro-microampere low-energy
electron currents. Journal of Scientific Instruments, 37, 246.
Gauvin, R., & Michaud, P. (2009). MC X-ray, a new monte carlo program for quantitative X-ray
microanalysis of real materials. Microscopy and Microanalysis, 15, 488.
Jaksch, H. (2008a). Low loss BSE imaging with the EsB detection system on the gemini ultra
FE-SEM. In EMC 2008 14th European Microscopy Congress, September 1–5, 2008, Aachen,
Germany.
Jaksch, H. (2008b). Strain related contrast mechanisms in crystalline materials imaged with AsB
detection. In EMC 2008 14th European Microscopy Congress, September 1–5, 2008, Aachen,
Germany.
Jaksch, H. (2011). The contrast mechanisms of LL-BSE electrons in FE-SEM characterization of
polymer, single proteins, and oxidization states of elements. Microscopy and Microanalysis,
17, 902–903.
Jaksch, H. (2012a). Hybridisation & band gap contrast from LL-BSE electrons. Microscopy and
Microanalysis, 18, 704–705.
Jaksch, H. (2012b). What BSE electrons can tell us. From ECCI via RBS to low loss BSE imaging.
Jaksch, H., & Martin, J. (1995). High-resolution, low-voltage SEM for true surface imaging and
analysis. Fresenius’ Journal of Analytical Chemistry, 353, 378–382.
Jaksch, H., & Vermeulen, J. (2005). New developments in GEMINI FESEM technology.
Microscopy Today, 13, 8–10.
Joy, D. (2002). SMART—A program to measure SEM resolution and imaging performance.
References 35
Joy, D. C. (1984). Beam interactions, contrast and resolution in the SEM. Journal of Microscopy,
136, 241–258.
Joy, D. C. (1985). Resolution in low voltage scanning electron microscopy. Journal of
Joy, D. C., Newbury, D. E., & Davidson, D. L. (1982). Electron channeling patterns in the
scanning electron microscope. Journal of Applied Physics, 53, R81–R122.
Kim, K. W., & Jaksch, H. (2009). Compositional contrast of uncoated fungal spores and stained
section-face by low-loss backscattered electron imaging. Micron, 40, 724–729.
Koshikawa, T., & Shimizu, R. (1974). A Monte Carlo calculation of low-energy secondary
electron emission from metals. Journal of Physics D: Applied Physics, 7, 1303.
Merli, P., Migliori, A., Morandi, V., & Rosa, R. (2001). Spatial resolution and energy filtering of
backscattered electron images in scanning electron microscopy. Ultramicroscopy, 88, 139–
150.
Murata, K. (1976). Depth resolution of the low-and high-deflection backscattered electron images
in the scanning electron microscope. Physica Status Solidi (a), 36, 527–532.
Newbury, D., Yakowitz, H., & Myklebust, R. (1973). Monte Carlo calculations of magnetic
contrast from cubic materials in the scanning electron microscope. Applied Physics Letters, 23,
488–490.
Newbury D., Yakowitz H., & Myklebust L. (1976). A study of type II magnetic domain contrast in
the SEM by Monte Carlo electron trajectory simulation. In Use of Monte Carlo Calculations in
Electron Probe Microanalysis and Scanning Electron Microscopy: Proceedings of a Workshop
Held at the National Bureau of Standards, Gaithersburg, Maryland, October 1–3, 1975. US
Department of Commerce, National Bureau of Standards: for sale by the Superintendent of
Documents, U.S. Government Printing Office.
Rasband, W. S. (1997–2015). Image J. Maryland, USA: Bethesda. https://imagej.nih.gov/ij/.
Reimer, L. (1993). Image formation in low-voltage scanning electron microscopy. In L. Reimer
(Ed.), Image formation in low-voltage scanning electron microscopy. USA: SPIE-International
Society for Optical Engineering.
(Springer Series in Optical Sciences). Berlin: Springer.
Reimer, L., & Volbert, B. (1979). Detector system for backscattered electrons by conversion to
secondary electrons. Scanning, 2, 238–248.
Tsurumi, D., Hamada, K., & Kawasaki, Y. (2010). Energy-filtered imaging in a scanning electron
microscope for dopant contrast in InP. Journal of Electron Microscopy, 59, S183–S187.
Wells, O. C. (1970). New contrast mechanism for scanning electron microscope. Applied Physics
Letters, 16, 151–153.
Wells, O. C. (1971). Low-loss image for surface scanning electron microscope. Applied Physics
Letters, 19, 232–235.
Wells, O. C. (1974). Scanning electron microscopy, USA: McGraw-Hill.
Wells, O. C. (1979). Effects of collector take-off angle and energy filtering on the BSE image in
the SEM. Scanning, 2, 199–216.
Wells, O. C., Broers, A., & Bremer, C. (1973). Method for examining solid specimens with
improved resolution in the scanning electron microscope (SEM). Applied Physics Letters, 23,
353–355.
Wells, O. C., LeGoues, F., & Hodgson, R. (1990). In-lens low-loss electron detector for the upper
specimen stage in the SEM. Electron Microscopy 1990, 1, 382.
Zach, J. (1989). Design of a high-resolution low-voltage scanning electron microscope. Optik, 83,
30–40.
Zach, J., & Rose, H. (1986). Efficient detection of secondary electrons in low-voltage scanning
electron microscopy. Scanning, 8, 285–293.
Zach, J., & Rose, H. (1988). High-resolution low-voltage electron-microprobe with large SE
detection efficiency. In Institute of Physics Conference Series, 81–82.
Chapter 4
Low Voltage SEM
For a long time, the microscopist had to compromise between the lens aberrations,
especially chromatic, and the electron diffusion volume to carry high resolution
imaging in the SEM. As seen from Fig. 3.3, the higher the beam accelerating
voltage, the smaller the chromatic and diffraction aberrations. However, due to its
E5/3
0 dependence (Kanaya and Okayama 1972), the diffusion volume of primary
electrons increases dramatically when E0 is increased. On the contrary and par-
ticularly for BSE imaging, the reduction of the diffusion volume, and thus of E0, is
highly beneficial to improve spatial resolution (Fig. 3.4a, b) which is opposite to the
effects of aberrations. In parallel, the BSE yield, which show little variations with
E0 as a function of Z down to E0 = 5 kV, do not seem to follow this trend for
accelerating voltages smaller than 3–5 kV (Reimer 1993; Schmid et al. 1983)
leading to significant variations of the yield for certain elements. However, these
measurements might be suspicious due to the increasing surface effect of con-
tamination and oxidation at very low voltages (El Gomati et al. 2008) that may have
skewed the reported results. A great advantage resides in the fact that the tilt
dependence of the SE yield is negligible at very low voltages (Reimer 1993) and the
SE image is less affected by the “edge” effect observed at higher accelerating
voltages (Cazaux 2005; Reimer 1998) [see Fig. 6.8 from Reimer (1998)]. The
resulting image provides thus a better description of the edges defining the surface
in addition to a higher surface sensitivity.
The BSE emission volume approaches the range of emission depth of SEs for
approximately E0 < 1 kV, and ultimately that of the probe size at very low voltage.
This is of great importance because the emission depth of the collected signals
becomes of secondary importance and the attention is then focused on contrast and
probe size considerations. Therefore, the deceleration mode was a direct response to
the above discussion. As such, the dramatic increase of the beam aberrations when
decreasing E0 was counteracted and a decrease of the interaction volume at low
primary voltage combined to a small probe size was then rendered possible. In the
following sections, the usefulness of low voltage electron microscopy will be

38 4 Low Voltage SEM
developed and some applications described. Ultra-low voltage will be demonstrated

and the limits of reducing the beam voltage discussed.
4.1 Strategy of Characterization: Deceleration and Energy

Filtration
As discussed above, the variation of the accelerating voltage results in changing

image contrasts and emission volume dimensions. As an example, Fig. 4.1 shows
pairs of images obtained with the upper (Fig. 4.1a, c and e) and top (Fig. 4.1b, d
and f) in-lens detectors (see Fig. 3.2 for the detector definition) from a titania/
amorphous carbon black blend used in solar cells technology at various low
accelerating voltages. From these images, one can rapidly notice that the topog-
raphy and hence, the 3D component of the 2D projection, appears clearly with the
upper detector while it is mostly absent from the top detector images, and this,
regardless of E0 or EL. At E0 = 2 kV (Fig. 4.1a, b), the large penetration depth of
primary electrons permitted to describe more clearly the distribution of each phase
with a reduced contribution of the shallow surface composition (coating or con-
tamination). However, due to the increased penetration of the primaries, the visi-
bility and definition of the nanoparticles were reduced due to the electron beam
diffusion effects at edges (Reimer 1998). Thus, in this case, lowering the acceler-
ating voltage was required and images shown in Fig. 4.1c, d were acquired in
deceleration mode with EL = 0.5 kV (E0 = 4 kV). The impact on spatial resolution
is significant with a 1–2 nm resolution measured on the top detector image
(a) (c) (e)
(b) (d) (f)
Fig. 4.1 Comparison and benefits of combining different low primary beam energies to
characterize multiphase nanoparticle clusters. The specimen was a titania (TiO2)/carbon black mix
used as lithium-ion battery cathode electrodes. a, c, e upper detector, b, d, f top detector with
filtration. SEM conditions were as follows: a, b E0 = 2 kV, c, d EL = 500 V with deceleration
(E0 = 4 kV), e, f EL = 1 kV with deceleration (E0 = 2.5 kV). Filtration bias was 7.5 V for (a),
1.2 kV for (b), 0 kV for (d) and 0.75 kV for (f)
4.1 Strategy of Characterization: Deceleration and Energy Filtration 39
(Fig. 4.1d). However, to obtain the best spatial resolution, the top detector signal
was not filtered to improve the signal-to-noise ratio and did not provide strong
Z-contrast between the two phases. With top detector filtration (DE < 0.75 kV) and
a probe current slightly increased to improve the SNR, the top detector provided
very high contrast between the carbon black particles in dark and the titania ap-
pearing brighter (Fig. 4.1f, EL = 1 kV with E0 = 2.5 kV). Even small features were
observed at the carbon particles surface which were assumed to be impurities or
crystalline carbon nanoparticles. Therefore, varying E0 or EL and comparing the
in-lens detectors with and without deceleration mode is of great value to charac-
terize nanomaterials at the nanometer scale and should be used more systematically
to improve our knowledge of these materials.
4.2 High Resolution Imaging
Pushing the limits of visibility has always been the purpose of SEM designers. Of
course, TEM is the technique of choice when approaching the sub-nanoscale.
However, due to the expensive costs relative to owning or using TEM facilities,
largely spread and low cost FE-SEMs increasingly address a positive response to
this issue. Sub-nanometer spatial resolution is now currently achieved with the new
CFE and monochromators-equipped FE-SEMs (Bell and Erdman 2012; Konno
et al. 2014).
In Fig. 4.2 are reported typical images demonstrating nowadays practical spatial
resolution and contrast standards that can be routinely obtained with a state-of-the-art
CFE-SEM. In Fig. 4.2a, the fractured end of a 100 nm diameter multiwall-CNT
(a) (c) (e)
(b) (d) (f)
Fig. 4.2 High-resolution images with low accelerating voltage. CNTs with a sulfur surface
treatment (a) and covered with Pt nanoparticles (b), MgO ash cubes (c, d) and alumina spheres (e,
f). Corresponding filtrated images obtained simultaneously with the top detector for (a–c) are
available in Fig. 4.3a–c. EL = 0.5 kV and E0 = 2 kV for (a, b), EL = 0.2 kV for (c, d),
EL = 0.7 kV and E0 = 4.2 kV for (e, f)
(MWCNTs) was imaged at EL = 0.5 kV in deceleration mode with E0 = 2 kV. Iron

seed nanoparticles were observed embedded in the MWCNT walls as well as
nanoparticles of 1 nm at the surface resulting from the process steps of growing the
MWCNTs. The embedded nanoparticles were assumed to remain from the Fe cat-
alyst used to grow the MWCNTs and the small ones were assumed to be a result of
the sulfur treatment applied to the newly grown MWCNTs (Verde-Gomez et al.
2017). Also, the surface roughness appeared very clearly with an easily discernible
nanometer scale porosity. In Fig. 4.2b, the same conditions were used to study
MWCNTs covered with Pt nanoparticles and the observed surface roughness was of
the order of the nanometer. As a result, the interaction of the Pt nanoparticles and the
MWCNT substrate was revealed in a more realistic fashion than with TEM. Note that
the top detector images corresponding to the same areas as used in Fig. 4.2a–c are
given in Fig. 4.3a–c.
When the primary electron energy is high and the object size is small regarding
the interaction volume, the edge contrast is weak due to their high penetration depth
and thus the overall contrast of the SE1s + SE2s is not adequate to render the true
topography of a small object. In addition, the incident angle dependence of the SE
yield at medium and high beam energies makes the edges of the object appear
dramatically brighter than the rest of the image. The adjustment of brightness and
contrast can then be complicated to obtain a good image. Conversely and as
described by Reimer (1993), the SE yield angular variation is negligible at very low
voltage, this being slightly more pronounced for low Z materials. Therefore,
decreasing the beam voltage results in reducing the interaction volume to the same
range as the probe size and reducing the intensity variation between the edges of the
object and its core. To demonstrate the usefulness of reducing the beam voltage, a
specimen consisting of MgO nano-cubes was prepared by burning a small Mg strip
(a) (c) (e)
(b) (d) (f)
Fig. 4.3 High-resolution images of CNTs with a sulfur surface treatment (a) or covered with Pt
nanoparticles (b), MgO ash cubes (c, d) and alumina spheres (e, f). a–c are the filtered images
obtained with the top detector corresponding to the same areas as shown in Fig. 4.2a–c. The
images in (d–f) were obtained with the upper detector
4.2 High Resolution Imaging 41
under a graphite SEM stub and was observed with a landing voltage of 0.2 kV with
beam deceleration. The cubes have very straight 90° edges and, in the case of the
larger ones, multiple surface steps making them ideal candidates to demonstrate the
aforementioned statement. In Fig. 4.2c, an image of cubes with a side length
smaller than 60 nm, obtained with the upper detector, is shown. As per the low
landing voltage used, the cubes edges are not over-weighted in the image and the
low angular dependence of the SE yield is just sufficient to render a “natural” visual
aspect of the image. Note that the varying intensity between neighboring cubes
were assumed to be a result of the charging of the specimen combined to the
specimen bias applied for deceleration. A second example is given in Fig. 4.2d,
where a large cube (few hundreds of nanometers) was imaged with the same
conditions. In this image, the multiple steps observed at the edges of the cube show
the same contrast as in Fig. 4.2c, permitting, among others, to distinguish growth
defects very clearly, as shown in the left top corner of Fig. 4.3d. However, if one
looks closer to the very small cubes (side length smaller than 20 nm), bright edge
contrast is observed suggesting that a lower landing voltage should be used for the
smaller cubes to reduce further the edge effect of these objects. Furthermore, the
effect of the substrate backscattering has to be taken into account in this case as SEs
and BSEs contribute to the image via the electron acceleration resulting from the
deceleration-induced electric field at the specimen surface. Consequently, the edge
contrast, related to the backscattering from the other materials in close proximity of
the edge, is larger for MgO (Fig. 4.2d) than for C (SEM stub) in Fig. 4.2c. This
typically demonstrates that the accelerating or landing voltage must be adapted to
the size of the objects that one wants to analyze if edge effects are to be avoided.
Finally, Fig. 4.2e, f are high-resolution images of two alumina spheres with a
landing voltage of 0.7 kV with beam deceleration. The accelerating voltage was
reduced for the same reasons as for the previous example. However, in this case, the
high charging potential of alumina had to be taken into account to choose the final EL
and led to a compromise with EL = 0.7 kV. Again, the use of a low beam voltage
permitted to acquire nanometer scale images with the upper detector where the true
surface was observed with clarity. Due to the high spatial resolution combined to a
low edge contrast of these images, crystal growth defects and directions can now be
directly observed on objects which are difficult to characterize in TEM without
slicing the specimen. Especially, one can notice the nanoscopic parallel surface steps
at the center of the image in Fig. 4.2f as well as the granular topography.
Complementary images of these alumina spheres are given in Fig. 4.3e, f.
4.3 Low Voltage, Specimen Charging, and Material

Contrast
It is well known that biological materials have a high charging potential and require
specific procedures to make them suitable to vacuum immersion prior to SEM
characterization. A review of the charging effects can be found in Reimer (1998) or
Cazaux (2004). The main steps required for preparing such samples are: chemical
fixation to preserve the internal specimen structure, dehydration by exchanging
water with a volatile solvent, drying to eliminate the solvent from the specimen and
surface coating to reduce specimen charging under electron beam irradiation
(Echlin 2009; Schatten and Pawley 2008; Bozzola and Russell 1999). However,
because the surface coating is generally of several nanometers, it prevents from
using low accelerating voltages for obtaining Z-contrast from BSE images. This
range of voltages is highly preferred when surface details are of interest and this
leads to a situation where the surface coating must be avoided. Figure 4.4 exem-
plified what can be obtained when the surface coating is omitted. The specimen was
a solution of cultured cancerous cells treated with functionalized Au coated CNTs
(Dotan et al. 2016) prepared following the procedure briefly described above but
with the surface uncoated. The images presented in Fig. 4.4a, b were obtained with
the upper and top detectors, respectively and an accelerating voltage of 1 kV
without deceleration. Energy filtration biases of 150 and 800 V were used with the
upper and top detectors, respectively. Hence, the upper detector image was purely a
BSE image, while the top detector image corresponded to a low-loss electron (LLE)
BSE image.
Fig. 4.4 Low voltage (a)

imaging of cancerous cells
with the upper (a) and top
(b) detectors after
glutaraldehyde and osmium
tetroxide fixation. No coating
was used and the accelerating
voltage of 1 kV with energy
filtration (150 V for upper and
70% for top) was selected to
reduce charging and enhance
the material contrast
(b)
4.3 Low Voltage, Specimen Charging, and Material Contrast 43
Due to the osmium tetroxide (ZOs = 76, ZOsO4 = 21.6) chemical fixation, the
backscattering coefficient was large and the BSEs collected mostly came from the
organic material surrounding the CNTs and overwhelmed the BSE signal from
them. Therefore, the upper detector image in Fig. 4.4a did not show any contrast
between the CNTs and the cells and provided essentially topographical contrast. In
contrast, the LLE BSE image, originating from a smaller depth (Fig. 3.10c), show a
higher contrast between the two materials (Fig. 4.4b). In fact, because the emission
volume of LLEs is in the nanometer range at this accelerating voltage, the BSEs
collected by the filtered top detector were assumed to interact solely with the CNTs
and the Au coating at their surface and very poorly with the cell underneath. Then,
monitoring the interactions between the CNTs and the cultured cells was made
possible with high Z-contrast and the combined use of the upper and top detectors
with low accelerating voltage permitted to obtain highly valuable information about
these interactions. The charging effects were reduced through the use of low
accelerating voltage with E0 E2 1 kV (Joy and Joy 1996). Using the BSE
signal and energy filtration allowed to attain nanometer resolution with high con-
trast between the materials under investigation and permitted to localize the CNTs
packets between the cells and thus the selective targeting of the CNTs regarding the
cancerous cells was demonstrated.
4.4 Ultra-Low Voltage SEM: Uses and Limitations
As underlined previously, working at low accelerating voltages has two main

advantages: first, by reducing the penetration depth of the primary electrons, surface
information can be obtained. Secondly, when sufficiently low E0 is used, the
emission volume of the BSEs become laterally similar to the probe diameter. Thus,
the resolution gap between SE and BSE imaging no longer holds and it allows a
more straightforward interpretation of both signals due to their similar emission
volume dimensions. Moreover, the availability of the deceleration technology is of
great importance to keep the lens aberrations sufficiently small to produce a
nanometer scale probe size and permit high resolution imaging. Without beam
deceleration, the theoretical probe size would theoretically be of the order of 10 nm
at E0 = 250 V and up to several tens of nanometers at E0 < 100 V (see Fig. 3.3).
However, when deceleration is applied, a significant improvement in probe diam-
eter is achieved as already described previously.
However, how far can we go in this direction and what are the possible appli-
cations and limits of using ultra-low landing voltages? In Fig. 4.5, top detector
images obtained by increasing the deceleration voltage to produce ultra-low landing
voltages are shown. The landing voltages were 100, 50, 30, and ultimately 10 V
with a constant E0 at 1600 V. One can notice the excellent spatial resolution
obtained at 100 and 50 V (Fig. 4.5a, b) with 2–3 nm size object being resolved.
However, it tends to degrade at landing voltages smaller than 50 V, more specif-
ically at 10 V (Fig. 4.5c, d). At 30 V, the resolution was still high enough to
100 V 50 V
(a) (b)
30 V 10 V
(c) (d)
Fig. 4.5 Ultra-low landing voltage imaging in deceleration mode of a sample consisting in CNTs
covered with Pt nanoparticles. a E0 = 1600 V, Edec = 1500 V, EL = 100 V; b E0 = 1600 V,
Edec = 1550 V, EL = 50 V; c E0 = 1600 V, Edec = 1570 V, EL = 30 V; d E0 = 1600 V,
Edec = 1590 V, EL = 10 V
observe surface features as with higher landing voltages, the minimal object size
detected being 3–5 nm. At 10 V, the surface looked very blurry and demonstrates a
dramatic loss of resolution with a minimal measured object size of 5–10 nm.
At the same time, it is interesting to notice that the contrast observed at these
voltages is changing when the landing voltage decreases. At EL > 50 V, the typical
topographical contrast is observed with the top detector because all the emitted
electrons were accelerated towards the top detector converting plate. However, this
contrast seems to decrease at EL = 30 V (Fig. 4.5c) and is finally very faint at
EL = 10 V (Fig. 4.5d). Also, at EL values lower than 50 V bright areas and spots at
the surface of the CNTs are observed and this trend is at its maximum at 10 V.
At ultra-low landing voltage, i.e., 10 V < EL < 100 V, the electron inelastic
mean free path is at its minimum at around 1 nm (Seah and Dench 1979), as well as
the elastic mean free path, around 0.1–1 nm (Ding and Shimizu 1996), and the
incident electrons interact closely with the very first surface layers. Therefore, the
contrast is highly dependent on surface chemistry and electrical as well as magnetic
properties of the surface layers. Especially, charging effects and contamination or
oxidation layers (Dapor et al. 2009) generally observed in a SEM affect the final
image as these layers are generally up to several nanometers thick. At these volt-
ages, the contrast seems influenced by semi-conductors electrical properties
4.4 Ultra-Low Voltage SEM: Uses and Limitations 45
(El-Gomati and Wells 2001) but mainly by the native oxide layer thickness (El
Gomati et al. 2009; Mikmekova et al. 2015; Zaggout et al. 2010) which is material
and orientation dependent. Also, considering the surface barrier from the vacuum to
the matter, a 10 eV electron could only suffer one or two low energy loss inter-
actions before being absorbed or backscattered. To this regard, it may provide
comparable information as Auger electron microscopy with a depth penetration at
the atomic scale.
One of the highest concern in using such low landing voltages is the necessity of
applying a specimen bias to the stage to decelerate the incident electrons prior to
their penetration in the specimen. This bias voltage, generally of a few kV, might
highly modify the electrical properties at and near the specimen surface and
complicates the interpretation of the images thus obtained.
References
Hoboken: John Wiley & Sons.
Bozzola, J. J., & Russell, L. D. (1999). Electron microscopy: Principles and techniques for
biologists. Burlington: Jones & Bartlett Learning.
Cazaux, J. (2004) About the mechanisms of charging in EPMA, SEM, and ESEM with Their Time
Evolution. Microscopy and Microanalysis, 10.
Dapor, M., Jepson, M. A., Inkson, B. J., & Rodenburg, C. (2009). The effect of oxide overlayers
on secondary electron dopant mapping. Microscopy and Microanalysis, 15, 237–243.
Ding, Z.-J., & Shimizu, R. (1996). A Monte Carlo modeling of electron interaction with solids
including cascade secondary electron production. Scanning, 18, 92–113.
Dotan, I., Roche, P. J. R., Paliouras, M., Mitmaker, E. J., & Trifiro, M. A. (2016). Engineering
multi-walled carbon nanotube therapeutic bionanofluids to selectively target papillary thyroid
cancer cells. Plos One, 11.
Echlin, P. (2009). Handbook of sample preparation for scanning electron microscopy and X-ray
microanalysis. Berlin: Springer.
El-Gomati, M., & Wells, T. (2001). Very-low-energy electron microscopy of doped semiconduc-
tors. Applied Physics Letters, 79, 2931.
El Gomati, M., Walker, C., Assa’d, A., & ZadraŽil, M. (2008). Theory experiment comparison of
the electron backscattering factor from solids at low electron energy (250–5000 eV). Scanning,
30, 2–15.
El Gomati, M., Zaggout, F., Walker, C., & Zha, X. (2009). The role of oxygen in secondary
electron contrast of doped semiconductors in LVSEM. In SPIE scanning microscopy.
Joy, D. C., & Joy, C. S. (1996). Low voltage scanning electron microscopy. Micron, 27, 247–263.
Kanaya, K., & Okayama, S. (1972). Penetration and energy-loss theory of electrons in solid
targets. Journal of Physics. D. Applied Physics, 5, 43.
Mikmekova, S., Yamada, K., & Noro, H. (2015). Dual-phase steel structure visualized by
extremely slow electrons. Microscopy, dfv059.
Reimer, L. (1993). Image formation in low-voltage scanning electron microscopy (SPIE Tutorial
Text Vol. TT12) (Tutorial Texts in Optical Engineering). SPIE Press.
Schatten, H., & Pawley, J. B. (2008). Biological low-voltage scanning electron microscopy.
Berlin: Springer.
Schmid, R., Gaukler, K., & Seiler, H. (1983). Measurement of elastically reflected electrons
(E 2.5 keV) for imaging of surfaces in a simple ultra high vacuum scanning electron
microscope. Scanning Electron Microscopy, 2, 501–509.
Seah, M., & Dench, W. (1979). Quantitative electron spectroscopy of surfaces. Surface and
Interface Analysis, 1, 2–11.
Verde-Gomez, Y., Macias, E. M., Valenzuela-Muniz, A. M., Alonso-Lemus, I., Yoshida, M. M.,
Zaghib, K., et al. (2017). Structural study of sulfurated multiwall carbon nanohorns.
Submitted: Nanoletters.
Zaggout, F., Walker, C., & El Gomati, M. (2010). The chemisorption of oxygen and its effect on
the secondary electron emission from doped semiconductors. In Journal of physics:
Conference series.
Chapter 5
Low Voltage STEM in the SEM
Since its birth in 1938 (Von Ardenne 1938a, b and c), spatial resolution and image
contrast have been the central points of attention that pushed the development of
high technological STEMs for the nano-characterization of materials. Theoretically
speaking, spatial resolution is driven by the convolution of the probe size function
with the sample response function which is related to the electron scattering
cross-sections (Morandi et al. 2007; Pennycook et al. 2007). The probe size has
been and is still constantly improved by reducing destructive chromatic, spherical,
and diffraction aberrations through the use of upgraded optics (including cold-field
emitters as described in Chap. 2), aberration correctors and beam monochromators
(Bell and Erdman 2012). At the same time, its resolving power was enhanced by
using high accelerating voltages to benefit from the probe size’s dependence on E0
via the reduction of the chromatic and diffraction aberrations. But most importantly,
the use of high E0 permitted to reduce dramatically the beam broadening generated
by the specimen response function. In fact, beam broadening is inversely propor-
tional to E0 for a constant mass-thickness, as calculated by Goldstein et al. (1977)
and Gauvin (2015). Therefore, a high E0 applied to thinner specimens has been the
rule for decades. Even though this has brought STEM to unprecedented levels of
spatial resolutions, the picture is obscured by three main drawbacks related to the
use of high accelerating voltage. First, the electron beam impacting the surface
might generate electron beam damage in beam sensitive materials like lithium or
carbon which is dramatically increased by the use of higher beam voltages (Egerton
2012; Egerton et al. 2004). Secondly, delocalization effects, related to the increased
interaction forces generated by the high energy electrons, limits the spatial reso-
lution of the final STEM image (Pennycook and Nellist 2011). Finally, because
electron diffraction and scattering cross-sections in solids are inversely related to
E0, the scattering in the specimen is low and leads to low intensity contrasts,
although high quality imaging is still possible.
In the field of scanning electron microscopy, bulk imaging has improved, as
described in previous chapters, but its spatial resolution is still dependent on the
electron diffusion volume inside the specimen (Reimer 1998) although using low
48 5 Low Voltage STEM in the SEM
voltages permit to work with similar diffusion volume and probe size (see Chap. 4).
The increased use of solid state BSE detectors with thin conductive layers permits
now to increase dramatically the detector response to low energy backscattered
electrons (Reimer 1993). Also, the combination of an external BSE detector on top
of the specimen with in-lens BSE detectors provides a better detector angular
selection to achieve high contrast (Aoyama et al. 2011, 2015; Cazaux et al. 2013).
As seen previously, all these upgrades in signal detection combined to the avail-
ability of low landing voltages now allows a dramatic reduction of the spatial
resolution. However, the signal is highly surface sensitive and the intensity level of
the signals collected are quite low due to the fast probe current decrease with
decreasing E0. Especially, the main limitation resides in X-ray microanalysis which,
in the case of low voltage scanning electron microscopy, is limited by the available
incident energy versus the excitation threshold energies of the investigated elements
(Newbury and Ritchie 2016). Although new EDS detectors with thin silicon nitride
(Nylese and Rafaelsen 2017) or no windows (Burgess et al. 2013a, b, 2017) have
proved to provide improved efficiency and energy resolution in the low energy
range of the detected spectrum, the use of low and very low energy lines compli-
cates greatly the quantification procedure as most of these X-rays are L, M, or even
N lines in case of high atomic number materials like Pt or Au (Bearden 1967).
However, if a high SEM beam voltage, typically 20–30 kV, is applied to an
electron transparent specimen and the reflected or transmitted beams are used to
produce the image, the spatial resolution is dramatically improved. In addition,
most of the K and L X-ray lines are thus made available facilitating X-ray
microanalysis, even though special procedures need to be used in the case of thin
specimens (Goldstein et al. 1986). BSE and STEM-BF images from an AA2099
Al–Li–Cu alloy are presented and compared in Fig. 5.1. The BSE image in
Fig. 5.1a was recorded in the bulk form of the sample with E0 = 2 kV which
permitted to reduce the diffusion volume of BSEs to around 30–40 nm. The
Z-contrast is high and the distribution of fine T1 (bright) and d’ (dark) precipitates is
clearly established (Brodusch et al. 2017). However, the diffusion of electrons
inside the specimen significantly reduces the spatial resolution as indicated by the
broadening thickness of the bright T1 plates compared to the STEM images
(a) BSE 2kV (b) STEM-BF 30 kV (c) STEM-BF 200kV
Fig. 5.1 Comparison of a BSE image at E0 = 3 kV (a), a STEM-BF image at E0 = 30 kV in the

SEM (b) and a STEM-BF image at E0 = 200 kV in a DSTEM (c). The sample was an AA2099
Al–Li–Cu alloy and the specimen thickness in (b) and (c) was estimated to 80 nm
5 Low Voltage STEM in the SEM 49
obtained with the same SEM at E0 = 30 kV (Fig. 5.1b) and with a dedicated
aberration-corrected STEM (DSTEM) at E0 = 200 kV (Fig. 5.1c). The minimum
measured thickness was about 10–15 nm in Fig. 5.1a and 4 nm in Fig. 5.1b, c,
which demonstrates the usefulness of combining thin specimens and high voltage to
improve spatial resolution, even in the SEM. However, it is striking to observe that
the BF image contrast is higher at E0 = 30 kV compared to that obtained at
E0 = 200 kV. The T1 plates, which are edge-on in the images have a similar
contrast in both images but those parallel to the specimen surface provide much
more contrast at 30 kV. This is mostly due to the increased elastic scattering at low
voltage and to larger Bragg angles improving bright-field contrast compared to the
high voltage (200 kV) case. The T1 thickness measured recently by HRTEM and
HAADF-STEM was ranging between 1 and 2 nm (Guinel et al. 2014). Thus, the
contribution of one layer of T1s compared to the total emission thickness is too
small at 200 kV to produce a sufficient grey level change between the matrix and
the flat T1s. On the contrary, reducing E0 permitted to increase this contribution
through the corresponding increase in scattering cross-sections, considering an
identical specimen thickness.
The advantages of the bulk and transmission methods are exposed in Fig. 5.2.
Obviously, low voltage BSE imaging is invaluable to monitor precipitation through
the size and space distributions as well as grain orientation and preferential
(a) (b)
(c) (d)
Fig. 5.2 Comparison of bulk low voltage BSE imaging at E0 = 2 kV (a, b) and STEM imaging at
E0 = 30 kV in a CFE-SEM: c bright field (BF), d dark field (DF). The sample was an AA2099
Al–Li–Cu alloy and the specimen thickness in (c) and (d) was estimated to 80 nm
precipitation growth on a large scale (Bois-Brochu et al. 2014; Brodusch et al.

2012; Guinel et al. 2014; Ma et al. 2011; Munoz-Morris and Morris 2010) as seen
from Fig. 5.2a. However, LV-STEM provides higher spatial resolution and higher
contrast, especially when bright field (BF) and dark field (DF) signals are used
simultaneously (Fig. 5.2c, d). The grain boundary complex nanostructure displayed
in Fig. 5.2c, d is far well defined in BF/DF-STEM compared to that in the BSE
image of Fig. 5.2b. Also, the interactions and orientation of the precipitates can be
investigated more quantitatively with high quality STEM images. Note that in both
cases, the thickness probed by the electron beam was of the same order (30–40 nm
at 2 kV and 80 nm thin foil at 30 kV) and thus the information they provide is
highly complementary.
The most important advantage of LV-STEM in a FE-SEM resides in the fact that
multiple electron detectors can be used at the same time, depending on the SEM
manufacturer. Therefore, one can take advantage of the specific information pro-
vided by each detector. Note that X-ray detectors and imaging will be described
specifically in Chap. 7. As we have seen earlier, the transmitted signals provide
mass-thickness, Z and diffraction contrasts while the BSE signal is mainly influ-
enced by the material atomic number, its crystallography and to a lesser extent, its
topography. If the BSE signal is filtered, even at high beam energies, surface
information can be deduced from the resulting images. In fact, all the in-lens
detection systems can be used and provide SE/BSE signals with small collection
angles sensitive to the specimen surface. This detection strategy is exemplified in
Fig. 5.3 where a CNT covered with Ni–Pt nanoparticles was imaged at E0 = 30 kV
in one single scan. The STEM-BF image in Fig. 5.3b is dominated by
mass-thickness and diffraction contrast, the latter originating from the bulk of the
CNT walls where lattice deformation are often observed as shown in the encircled
region in Fig. 5.3a. As expected, Z-contrast dominates the STEM-DF (Fig. 5.3c)
and BSE images (Fig. 5.3d), the latter being clearer than the DF image mostly due
to a variation of the optimum focus between the two modes. However, a strong
insight is given by the SE image in Fig. 5.3a which was recorded with the upper
detector with a filtration bias of 7.4 V. Clearly, this single image permitted to locate
more accurately the nanoparticles covering the surface regarding those inside or on
the back of the CNT. In particular, those in the encircled region seem embedded in
the CNT wall and the contrast is weak in the BF, DF or BSE images which did not
give any clue on their true location. The weak contrast might also suggest that these
particles might be inclusions of carbon clusters produced during the growth of the
CNT. The SE image also inform on the surface roughness state of the specimen
which gives, in addition to the volume information provided by the other detectors,
a better general knowledge of the sample investigated. Due to the energy filtration
applied on the upper detector, low energy SEs, which are generated mostly away
from the prime surface were repealed and only those with high energy coming from
the first surface layers were collected to produce the image. This is mostly due to
the fact that the SE energy loss is proportional to their path length back to the
surface (Koshikawa and Shimizu 1974) and that their escape probability falls
exponentially with their emission depth (D.C. 1995).
5 Low Voltage STEM in the SEM 51
(a) SE (b) BF
(c) DF (d) BSE
Fig. 5.3 Multiple detectors low voltage STEM imaging. a SE (upper) with 7.5 V bias, b BF and
c DF STEM and d BSE (PDBSE) signals were collected simultaneously to combine information
from the bulk and from the specimen surface. The specimen was a CNT covered with Ni–Pt
nanoparticles observed at E0 = 30 kV
Pushing further the decrease of E0, one can still obtain interesting results if the
specimen is sufficiently transparent the low voltage beam. As an example, an
accelerating voltage of 10 kV used to observe the same CNT sample used in
Fig. 5.3 and a SE (Fig. 5.4a) as well as a STEM-BF (Fig. 5.4b) images were
recorded. As in the previous example, these images provided high surface sensitive
contrast with the filtered SE signal collected by the upper detector. At the same
time, the BF image did not suffer from any loss of spatial resolution with 1–2 nm
diameter particles being detected, as in the high beam energy case. As a result of
larger incident electron interactions in the specimen, the BF contrast of small
nanoparticles was greatly improved when decreasing E0. Due to the very small
thickness of graphene sheets, a beam voltage as low as 5 kV could be applied to
them combined with STEM-BF detection and the resulting image is displayed in
Fig. 5.4c. In this case, graphene lattice defects, like in CNTs, were observed
contrary to what was observed at high energy in which these features used to appear
with a very weak contrast.
It has to be noted that even lower energies have been used recently to investigate
free-standing films or biological specimens, the beam energy being in that case of
the order of a few tens or hundreds of volts (Frank et al. 2015; Mullerova et al.
2010). The ability of incident electrons of this energy to be transmitted through the
(a) (b) (c)
Fig. 5.4 Very low voltage STEM imaging. a SE upper detector with no energy filtration and
b BF-STEM images of a CNT covered with Ni–Pt nanoparticles with E0 = 10 kV. c BF-STEM
image of a graphene sheet at E0 = 5 kV
material is, in this case, explained by the fact that the electron elastic mean free path
is at a minimum around 100 eV and starts to increase below this limit (Seah and
Dench 1979) reducing thus the interactions with the material.
References
Aoyama, T., Nagoshi, M., Nagano, H., Sato, K., & Tachibana, S. (2011). Selective backscattered
electron imaging of material and channeling contrast in microstructures of scale on low carbon
steel controlled by accelerating voltage and take-off angle. ISIJ International, 51, 1487–1491.
Aoyama, T., Nagoshi, M., & Sato, K. (2015). Quantitative analysis of angle-selective
backscattering electron image of iron oxide and steel. Microscopy, dfv026.
Bearden, J. A. (1967). X-ray wavelengths. Reviews of Modern Physics, 39, 78.
Hoboken: John Wiley & Sons.
Bois-Brochu, A., Blais, C., Goma, F. A. T., Larouche, D., Boselli, J., & Brochu, M. (2014).
Characterization of Al–Li 2099 extrusions and the influence of fiber texture on the anisotropy
of static mechanical properties. Materials Science and Engineering A, 597, 62–69.
Brodusch, N., Trudeau, M., Michaud, P., Rodrigue, L., Boselli, J., & Gauvin, R. (2012).
Contribution of a new generation field-emission scanning electron microscope in the
understanding of a 2099 Al-Li Alloy. Microscopy and Microanalysis, 18, 1393–1409.
Brodusch, N., Voisard, F., & Gauvin, R. (2017) About the contrast of delta’ precipitates in bulk
Al-Cu-Li alloys in reflection mode with a field-emission scanning electron microscope at low
accelerating voltage. J Microsc.
Burgess, S., Li, X., & Holland, J. (2013a). High spatial resolution energy dispersive X-ray
spectrometry in the SEM and the detection of light elements including lithium. Microscopy and
Analysis, 27, S8–S13.
Burgess, S., James, H., Statham, P., & Xiaobing, L. (2013b). Using windowless EDS analysis of
45–1000 eV X-ray lines to extend the boundaries of EDS nanoanalysis in the SEM.
Burgess, S., Sagar, J., Holland, J., Li, X., & Bauer, F. (2017). Ultra-Low kV EDS—A new
approach to improved spatial resolution, surface sensitivity, and light element compositional
imaging and analysis in the SEM. Microscopy Today, 25, 20–28.
Cazaux, J., Kuwano, N., & Sato, K. (2013). Backscattered electron imaging at low emerging
angles: A physical approach to contrast in LVSEM. Ultramicroscopy, 135, 43–49.
D.C., J. (1995). Monte carlo modeling for electron microscopy and microanalysis. Oxford: Oxford
University Press.
References 53
Egerton, R. (2012). Mechanisms of radiation damage in beam-sensitive specimens, for TEM

accelerating voltages between 10 and 300 kV. Microscopy Research and Technique, 75, 1550–
1556.
Egerton, R., Li, P., & Malac, M. (2004). Radiation damage in the TEM and SEM. Micron, 35,
399–409.
Frank, L., Nebesavrova, J., Vancova, M., Patak, A., & Mullerova, I. (2015). Imaging of tissue
sections with very slow electrons. Ultramicroscopy, 148, 146–150.
Gauvin, R. (2015). A universal equation for computing the beam broadening of incidents electrons
in thin films. Microscopy and Microanalysis, Submitted.
Goldstein, J., Costley, J., Lorimer, G., & Reed, S. (1977). Quantitative X-ray analysis in the
electron microscope. Scanning Electron Microscopy, 1, 315–324.
Goldstein, J. I., Joy, D. C., & Romig, A. D. (1986). Principles of analytical microscopy.
New-York, US: Springer.
Guinel, M.-F., Brodusch, N., Sha, G., Shandiz, M., Demers, H., Trudeau, M., et al. (2014).
Microscopy and microanalysis of complex nanosized strengthening precipitates in new
generation commercial Al–Cu–Li alloys. Journal of Microscopy, 255, 128–137.
Koshikawa, T., & Shimizu, R. (1974). A Monte Carlo calculation of low-energy secondary
electron emission from metals. Journal of Physics. D. Applied Physics, 7, 1303.
Ma, Y., Zhou, X., Thompson, G., Hashimoto, T., Thomson, P., & Fowles, M. (2011). Distribution
of intermetallics in an AA 2099-T8 aluminium alloy extrusion. Materials Chemistry and
Physics, 126, 46–53.
Morandi, V., Merli, P., & Quaglino, D. (2007). Scanning electron microscopy of thinned
specimens: From multilayers to biological samples. Applied Physics Letters, 90, 163113.
Mullerova, I., Hovorka, M., Hanzlikova, R., & Frank, L. (2010). Very low energy scanning
electron microscopy of free-standing ultrathin films. Materials Transactions, 51, 265–270.
Munoz-Morris, M., & Morris, D. G. (2010). Severe plastic deformation processing of Al–Cu–Li
alloy for enhancing strength while maintaining ductility. Scripta Materialia, 63, 304–307.
Newbury, D. E., & Ritchie, N. W. M. (2016). Electron-excited X-ray microanalysis at low beam
energy: Almost always an adventure! Microscopy and Microanalysis, 22, 735–753.
Nylese, T., & Rafaelsen, J. (2017). Improvements in SDD efficiency–from X-ray counts to data.
Pennycook, S., Lupini, A., Varela, M., Borisevich, A., Peng, Y., Oxley, M., et al. (2007). Scanning
transmission electron microscopy for nanostructure characterization. In Scanning microscopy
for nanotechnology. Berlin: Springer.
Pennycook, S. J. & Nellist, P. D. (2011) Scanning transmission electron microscopy: Imaging and
analysis. Berlin: Springer.
Reimer, L. (1993). Image formation in low-voltage scanning electron microscopy (SPIE Tutorial
Text Vol. TT12) (Tutorial Texts in Optical Engineering). SPIE Press.
Seah, M., & Dench, W. (1979). Quantitative electron spectroscopy of surfaces. Surface and
Interface Analysis, 1, 2–11.
Von Ardenne, M. (1938a). Das elektronen-rastermikroskop, Praktische Ausführung. Zeitschrift für
Technische Physik, 19, 407–416.
Von Ardenne, M. (1938b). Das Elektronen-Rastermikroskop, Theoretische Grundlagen. Zeitschrift
für Physik, 109, 553–572.
von Ardenne, M. (1938c). Die Grenzen fur das Auflosungsvermogen des Elektronenmikroskops.
Zeitschrift fur Physik, 108, 338–352.
Chapter 6
The f-Ratio Method for X-Ray
Microanalysis in the SEM
6.1 The Limits of X-Ray Microanalysis Models
Quantitative X-ray microanalysis of bulk samples is usually obtained by measuring

the characteristic X-ray intensities of each element in a sample and in a corre-
sponding standard of known composition. The k-ratio of the measured intensities
from the unknown material over the standard is related to the concentration using
the ZAF or u(qz) correction methods. Under optimal conditions, accuracies
approaching 1% are possible. Routinely, analysis with a SEM-EDS system reach
±5% for 95% of the analyses (Newbury and Ritchie 2013). However, all the
experimental conditions must remain identical during the sample and standard
measurements. This is not possible with a cold-field emission scanning electron
microscope (CFE-SEM) where beam current can fluctuate by 5% in its stable
regime. To address this issue, a method was developed using a single spectrum
measurement (Gauvin 2012; Horny 2006; Horny et al. 2010). It is similar in
approach to the Cliff and Lorimer ratio method (Cliff and Lorimer 1975) developed
for the analytical transmission electron microscope (Goldstein et al. 1986). In this
method, corrections are made for X-rays generated from thick specimens using the
ratio of the characteristic X-ray intensities of two elements in the same material.
The proposed f-ratio method utilizes the ratio of the intensity of a characteristic
X-ray normalized by the sum of X-ray intensities of all the elements measured for
the sample. Uncertainties in the physical parameters of X-ray generation are cor-
rected using an experimental calibration factor that must be previously measured.
6.2 Description of the f-Ratio Method
Quantitative X-ray microanalysis with standard, the k-ratio method, is based on the
measurement of the net X-ray intensities of each element i from the specimen of
unknown composition Ci to that of a standard where the composition of element
56 6 The f-Ratio Method for X-Ray Microanalysis in the SEM
i, C(i), is known. The net intensities of the specimen of unknown composition Ii and
that of the standard I(i) are related to the composition through this equation:

Ci Ii ip s ðiÞ
¼ Zi Ai Fi ð6:1Þ
CðiÞ IðiÞ ip s i
where Zi, Ai, and Fi are the atomic number, the absorption, and the fluorescence
correction factors, respectively. Details on these corrections and the use of Eq. (6.1)
to perform quantitative X-ray microanalysis are given in many textbook (Goldstein
et al. 2003; Reed 1993; Scott et al. 1995). In Eq. (6.1), ip is the probe current and s
is the acquisition time and their product is the dose. Normally, the intensities of the
unknown and standard are acquired with the same probe current and acquisition
time and the ratio of the doses is unity. This explains why this term is not present in
the equations describing the ZAF method. If the acquisition doses are different, their
ratio is included to correct the variation of X-ray intensities caused by the variation
in doses and further difference between the intensities is related to a difference in
composition. However, if the probe current and acquisition time are measured, we
can perform quantitative X-ray microanalysis using Eq. (6.1) even if the dose is not
the same between the unknown and the standard. With cold field electron emitters,
the probe current fluctuations, 5% at best, make the accurate determination of the
dose an issue; quantitative X-ray microanalysis using Eq. (6.1) becomes
problematic.
Quantitative X-ray microanalysis was developed in the STEM in the 1970s years
with electron microscopes were the probe current was notoriety unstable. Cliff and
Lorimer developed a quantitative method based on the ratio of net X-ray intensities
of two characteristic lines from the same EDS spectrum (Cliff and Lorimer 1975).
Since the dose is the same for the emission of line A and B, the ratio of their X-ray net
intensities IA and IB will not depend on the value of the probe current and acquisition
time, even if the probe current fluctuates since the fluctuation is the same for the
emission of line A and B. Therefore, the composition CA and CB are related to the
ratio of intensities through this equation (Cliff and Lorimer 1975):
CA IA
¼ KAB ; ð6:2Þ
CB IB
where KAB is the Cliff and Lorimer K factor. It can be computed theoretically or
measured experimentally with specimens of known composition. Due to uncer-
tainties in the fundamental parameters related to X-ray generation and detection,
measured K factors are more accurate than computed ones. Equation (6.2) is valid
for thin specimens where absorption and fluorescence are negligible but formulas to
correct for these effects exist (Anderson et al. 1995; Nockolds et al. 1980). More
details on this TEM quantitative method are given elsewhere (Gauvin 2012;
Williams and Carter 2009). An extension of the Cliff and Lorimer method, called
the f-ratio method, was developed for bulk specimens in CFE-SEMs.
6.2 Description of the f-Ratio Method 57
6.2.1 f-Ratio Method for Binary System
The f-ratio for a binary system with element A and B and their X-ray net intensities
IA and IB is then defined as:
IA
fA ¼ ð6:3Þ
IA þ IB
Similar ratio method was proposed for the quantification of µ-size particles (Hnizdo
and Wallace 2002).
In the case of very low concentration of element B, Eq. (6.2) becomes unstable.
Low concentration of element B means that IB is close to zero and the intensity
measurement uncertainty creates large uncertainty on the concentration calculation.
However, using the f-ratio Eq. (6.3) when IB is close to zero, the ratio goes to 1 and
the intensity measurement uncertainty effect on the concentration calculation is
smaller. The is more stable statistically than the ratio of IA =IB in the case of low
concentration of one of the element. More details about this method are given
elsewhere (Gauvin 2012; Horny 2006; Horny et al. 2010).
To use the f-ratio method to determine the concentration of unknown binary
systems a calibration curve of f-ratio versus concentration is used. Calibration
curves are computed from Monte Carlo simulations like CASINO (Drouin et al.
2007) or MC X-Ray (Gauvin and Michaud 2009) or alternatively using analytical
uðqzÞ models like PAP, XPP (Pouchou and Pichoir 1991), or PROZA (Bastin et al.
1998). Because of the uncertainties in the fundamental parameters for X-ray gen-
eration and detection, a calibration factor obtained from X-ray measurements of a
specimen of known composition and simulated intensities is calculated. These
references (Gauvin 2012; Horny 2006; Horny et al. 2010; Teng et al. 2017) explain
how to calculate the calibration factor. Examples of calibration curves for a bulk
Al–Mg system at 5 kV are shown in Fig. 6.1. The calibration factor is small for this
system, but needed for accurate quantification. It corrects for inaccuracy in the
window efficiency and/or mass absorption coefficients by decreasing the theoretical
Al intensity and increasing the Mg intensity, which is observed in Fig. 6.1a.
Increasing the accelerating voltage for this system adds large curvature in the f-ratio
curve (Fig. 6.1b). This curvature indicates a difference of mass absorption coeffi-
cient between the two X-ray lines and it is more visible at larger accelerating
voltage. This effect is observed when the two X-ray lines have different ionization
energies. From the measured f-ratio, the composition is obtained from these curves
by interpolation as illustrated in Fig. 6.1c.
The f-ratio method was successfully used with various binary systems (Gauvin
et al. 2006; Horny 2006; Horny et al. 2010) as well as for the determination of
diffusion coefficients (Das et al. 2014; Gauvin et al. 2012; Rudinsky et al. 2014).
(a)
(b)
(c)
6.2 Description of the f-Ratio Method 59
JFig. 6.1 Simulated f-ratio calibration curves for Al–Mg binary system. a f-ratio for Al (blue) and
Mg (orange) versus Al concentration at an accelerating voltage of 5 kV. Dashed lines are f-ratio
calibration curves which include an experimental correction factor. b Effect of the accelerating
voltage on the f-ratio calibration curves for Mg. c Example of how the concentration is determined
from a measured f-ratio value at 20 kV. The red line shows how the experimental f-ratio value of
0.971 gives a weight fraction concentration of 0.889 from the calibration curve
6.2.2 Generalization of the f-Ratio Method

for Multi-elements
The f-ratio method was further developed and generalized for multi-elements by
using the pure element calibration factor (Teng et al. 2017). The generalized f-ratio
equations and calibration curves are computed and the concentration for each
element is obtained using a N dimensions interpolation. More details are given
elsewhere (Teng et al. 2017).
6.3 Examples of Quantitative X-Ray Analysis Using

the f-Ratio Method
6.3.1 Binary Examples
6.3.1.1 Quantification of Au–Cu Binary Alloy with a CFE-SEM
Figure 6.2 shows the relative error on the estimation of the Au composition (weight
fraction) according to the real composition of the standards certified by The
National Institute of Standards and Technology (NIST (Heinrich et al. 1971), using
the standardless routine of an EDS system (Fig. 6.2a) and using the f-ratio and a
calibration factor for each electron beam energy. For the standardless routine, the
inaccuracy ranges to ±40% of the nominal composition. These standardless
quantitative results are in accord with previous studies (Newbury et al. 1995).
Recent development in standardless methods have reduce the inaccuracy to the
ranges of ±10% for 95% of the analyses (Statham 2004; Trincavelli et al. 2014).
However, the f-ratio method provides a higher accuracy that is generally ±5% if
the Ka or La X-ray lines are used as seen from Fig. 6.2b–c. The f-ratio method
gives accurate quantitative X-ray microanalysis results comparable to the k-ratio
method in a SEM-EDS system which reaches ±5% accuracy for 95% of the
analyses (Newbury and Ritchie 2013), and this without a stable probe current.
Therefore, it is possible to perform quantitative X-ray microanalysis with cold field
scanning electron microscopes (Horny 2006; Horny et al. 2010).
(a)
(b)
(c)
6.3 Examples of Quantitative X-Ray Analysis Using the f-Ratio Method 61
JFig. 6.2 Relative error in the estimation of the Au weight fraction according the real weight
fraction of the standards (Heinrich et al. 1971) at different accelerating voltages: a Using the
standardless routine of the EDS system, b using the Au Ma–Cu La f-ratio and calibration factor
obtained for each accelerating voltage, c using the Au La–Cu Ma f-ratio and calibration factor
obtained for each accelerating voltage. The green box represents an error of ±5%. Adapted from
(Horny et al. 2010) under permission from Cambridge University Press
6.3.1.2 Quantification of Al–Mg Diffusion Couple
Figure 6.3 shows the measured composition profiles of Al and Mg in a binary Al–
Mg diffusion couple determined using three different quantification methods at an
accelerating voltage of 5 kV with a CFE-SEM. The difference between the k-ratio
and the f-ratio methods is ±2% and the standardless method is ±8% (Teng et al.
2017).
The f-ratio quantitative results in Fig. 6.3 were calculated using pure element
calibration factors (Teng et al. 2017). Table 6.1 compares the calibration factors
(KAl and KMg) obtained from pure element specimens and factors (KMg–Al) obtained
from an alloy of now composition. The factor KMg–Al was also calculated from the
pure element factors. The differences between the calculated and measured values
were less than 2%, which demonstrated the accuracy of the pure element calibration
factor method (Teng et al. 2017). The pure element calibration factors were a lot
smaller than one because the exact detector parameters like efficiency and solid
angle were difficult to measure and nominal values were used in this case.
6.3.2 Multi-elements Example
6.3.2.1 Quantification of Al–Mg–Zn Ternary Alloy with a CFE-SEM
The f-ratio method was generalized to more than two elements. The correction
factors are either acquired experimentally relatively to two elements or from pure
element specimens and they do not change with composition. They are obtained
from measurement of one known phase. The concentration curves versus f-ratio are
obtained by Monte Carlo simulations and the unknown concentrations are calcu-
lated from these curves via the measured f-ratios by combination of
multi-dimensional interpolation. As an example, a quantitative X-ray line profile of
a ternary Al–Mg–Zn diffusion couple sample was obtained with a cold field
emission SEM at 5 kV, as shown in Fig. 6.4. The Al and Mg concentrations were
obtained from the f-ratio calibration curves and the Zn composition was obtained by
difference. The difference between the k-ratio and the f-ratio methods is ±2% for
Al and Mg and ±5% for Zn. Similar results were obtained at 20 kV (not shown)
and these results at different accelerating voltages demonstrate the accuracy of the
generalized f-ratio quantitative X-ray microanalysis method with cold field emis-
sion scanning electron microscope (Teng et al. 2017).
(a)
(b)
Fig. 6.3 Comparison of quantitative methods for a line profile across an Al–Mg diffusion couple
interface with an accelerating voltage of 5 kV: a Al and b Mg concentrations. (blue line) k-ratio
standard quantification calculated using DTSA-II (Ritchie 2011, 2012). (orange dot) f-ratio with
calibration curves computed with MC X-Ray (Gauvin and Michaud 2009). (green dot)
Standardless obtained from the EDS acquisition software. Adapted from (Teng et al. 2017) under
permission from Cambridge University Press
Table 6.1 Calibration factors for an Al–Mg binary system obtained with pure element specimens
and with an alloy of known composition
E0 (kV) KAl KMg KMg–Al
Calculated Measured
5 0.658 0.666 0.988 0.979
20 0.693 0.723 0.958 0.937
6.4 Summary 63
Fig. 6.4 Comparison of quantitative methods for a line profile across an Al–Mg–Zn diffusion
couple interface with an accelerating voltage of 5 kV: (blue) Al, (orange) Mg and (green) Zn.
(line) k-ratio standard quantification calculated using DTSA-II (Ritchie 2011, 2012) and
(dot) f-ratio with calibration curves computed with MC X-Ray (Gauvin and Michaud 2009).
Adapted from (Teng et al. 2017) under permission from Cambridge University Press
6.4 Summary
The f-ratio method (without a stable probe current) provides an accuracy that is
generally better than ±5% if the Ka or La X-ray lines are used. This is comparable
to the k-ratio method in a SEM-EDS system, which reaches ±5% accuracy for 95%
of the analyses (Newbury and Ritchie 2013) and better than standardless methods
for which the accuracy is in the range of ±10% for 95% of the analyses (Statham
2004; Trincavelli et al. 2014) at best. Furthermore, the calculation of accurate
calibration factors based on pure elements are easier to obtained than using known
composition alloy specimens and have shown a difference of less than 2% (Teng
et al. 2017). Therefore, the generalized f-ratio quantitative X-ray microanalysis
method allows high accuracy quantification of multi-element samples with a cold
field emission scanning electron microscope where the probe current is not stable.
Obviously, the method is also accurate in a conventional SEM-EDS instrument.
References
Anderson, I. M., Bentley, J., & Carter, C. B. (1995). The secondary fluorescence correction for X-ray
microanalysis in the analytical electron microscope. Journal of Micrsocopy, 178, 226–239.
Bastin, G. F., Dijkstra, J. M., & Heijligers, H. J. M. (1998). PROZA96: An improved matrix
correction program for electron probe microanalysis, based on a double Gaussian $(rho z)$
approach. X-Ray Spectrometry, 27, 3–10.
Cliff, G., & Lorimer, G. W. (1975). The quantitative analysis of thin specimen. Journal of
Micrsocopy, 103, 203–207.
Das, S. K., Brodusch, N., Gauvin, R., & Jung, I.-H. (2014). Grain boundary diffusion of Al in Mg.
Scripta Materialia, 80, 41–44.
Drouin, D., Couture, A. R., Joly, D., Tastet, X., Aimez, V., & Gauvin, R. (2007). CASINO V2.42—
A fast and easy-to-use modeling tool for scanning electron microscopy and microanalysis users.
Scanning, 29, 92–101.
Gauvin, R. (2012). What remains to be done to allow quantitative X-ray microanalysis performed
with EDS to become a true characterization technique? Microscopy and Microanalysis, 18,
915–940.
Gauvin, R., Brodusch, N., & Michaud, P. (2012). Determination of diffusion coefficients with
quantitative X-ray microanalysis at high—spatial resolution. In Diffusion in materials—DIMAT
2011. Trans Tech Publications.
Gauvin, R., & Michaud, P. (2009). MC X-ray, a new Monte Carlo program for quantitative X-ray
microanalysis of real materials. Microscopy and Microanalysis, 15, 488–489.
Gauvin, R., Robertson, K., Horny, P., Elwazri, A. M., & Yue, S. (2006). Materials characterization
using high-resolution scanning-electron microscopy and X-ray microanalysis. Journal of the
Minerals, Metals and Materials Society, 58, 20–26.
Goldstein, J. I., Joy, D. C., & Romig, A. D. (1986). Principles of analytical microscopy.
New-York, US: Springer.
Goldstein, J. I., Newbury, D. E., Echlin, P., Joy, D. C., Romig, J., A. D., Lyman, C. E., et al.
(2003). Scanning electron microscopy and X-ray microanalysis: A text for biologists, materials
scientists, and geologists. Plenum Press.
Heinrich, K. F. J., Myklebust, R. L., Rasberry, S. D., & Michaelis, R. E. (1971). Standard
reference materials: Preparation and evaluation of SRM’s 481 and 482 gold-silver and
gold-copper alloys for microanalysis. Washington DC: National Bureau of Standards.
Hnizdo, V., & Wallace, W. E. (2002). Monte Carlo analysis of the detection of clay occlusion of
respirable quartz particles using multiple voltage scanning electron microscopy. Scanning, 24,
264–269.
Horny, P. (2006). Development of a quantification method for X-ray microanalysis with an
electron microscope. McGill University.
Horny, P., Lifshin, E., Campbell, H., & Gauvin, R. (2010). Development of a new quantitative
X-ray microanalysis method for electron microscopy. Microscopy and Microanalysis, 16,
821–830.
Newbury, D. E., & Ritchie, N. W. M. (2013). Is scanning electron microscopy/energy dispersive
X-ray spectrometry (SEM/EDS) quantitative? Scanning, 35, 141–168.
Newbury, D. E., Swyt, C. R., & Myklebust, R. L. (1995). “Standardless” quantitative electron
probe microanalysis with energy-dispersive X-ray spectrometry: Is it worth the risk? Analytical
Chemistry, 67, 1866–1871.
Nockolds, C., Nasir, M. J., Cliff, G., & Lorimer, G. W. (1980). X-ray fluorescence correction in
thin foil analysis and direct methods for foil thickness measurement. In T. Murlvey (Ed.),
Electron microscopy and analysis, 1979. The Institute of Physics.
Pouchou, J. -L., & Pichoir, F. (1991). Electron probe quantitation. In: K. F. J. Heinrich &
D. E. Newbury (Eds.). Plenum Press.
Reed, S. J. B. (1993). Electron microprobe analysis. Press Syndicate of the Cambridge University
Press.
Ritchie, N. W. M. (2011). Standards-based quantification in DTSA-II—Part I. Microscopy Today,
19, 30–36.
Ritchie, N. W. M. (2012). Standards-based quantification in DTSA-II—Part II. Microscopy Today,
20, 24–28.
Rudinsky, S., Gauvin, R., & Brochu, M. (2014). The effects of applied current on one-dimensional
interdiffusion between copper and nickel in spark plasma sintering. Journal of Applied Physics,
116, 154901.
References 65
Scott, V. D., Love, G., & Reed, S. J. B. (1995). Quantitative electron-probe microanalysis. Ellis
Horwood.
Statham, P. J. (2004). A check total for validating standardless and normalised EDX analysis at
Low kV. Microchim. Acta.
Teng, C., Demers, H., Brodusch, N., & Gauvin, R. (2017). f-ratio quantitative analysis of MgAl
and MgAlZn alloy. Microscopy and Microanalysis.
Trincavelli, J., Limandri, S., & Bonetto, R. (2014). Standardless quantification methods in electron
probe microanalysis. Spectrochimica Acta, Part B: Atomic Spectroscopy, 101, 76–85.
Williams, D. B., & Carter, C. B. (2009). Transmission electron microscopy: A textbook for
materials science. Springer.
Chapter 7
X-Ray Imaging with a Silicon Drift
Detector Energy Dispersive Spectrometer
The scanning electron microscope (SEM) was primary developed for imaging
applications. With the introduction of the Si(Li) energy dispersive spectrometer
(EDS), simultaneous imaging and X-ray microanalysis became possible. However,
long working distance and high current were needed because the position and small
solid angle of the EDS detector. SEM was initially and is still optimized for
imaging applications, where the high spatial resolution is generally obtained at short
working distance. This problem is still relevant today and unfortunately X-ray
microanalysis is never performed in the best imaging conditions, i.e., not with the
smallest probe size. The silicon drift detector energy dispersive spectrometry
(SDD-EDS) is now superior to the Si–Li technology. The SDD have better energy
resolution, higher output count rate, and larger detection area, i.e., larger crystal or
sensor. These advantages allow the acquisition of qualitative and quantitative
chemical images (often called X-ray maps) in a couple of minutes (Newbury and
Ritchie 2011). The SDD is now considered as an “imaging detector” like the other
electron detectors. Furthermore, using longer acquisition time, the SDD is used
in situations with low emission rate like low accelerating voltage or low voltage
STEM (LV-STEM). This is possible, because the SDD collects more counts per
second (cps) and gets results in a reasonable acquisition time, which is limited by
the specimen or/and probe drift, specimen contamination, and maximum dose for
probe sensitive specimen. New geometries of SDD were developed to further
increase the collection rate, i.e., with larger solid angle. For example, an annular
silicon drift detector (a-SDD) system with four sensors where the detector is
inserted below the objective lens gives a higher solid angle, up to 1.4 sr. In con-
sequence, a lower working distance and probe current can be used and an improved
spatial resolution becomes possible during X-ray microanalysis. The reduced ac-
quisition time also allows for reducing beam drift, carbon contamination and
damage to sensitive materials. Further improvement of spatial resolution is possible
when working at low accelerating voltage as the interaction volume decreases
significantly. However, the emitted X-ray intensity drops significantly with the

68 7 X-Ray Imaging with a Silicon Drift Detector Energy …
accelerating voltage but is counter balanced by the high count rate achieved with
this type of annular detector.
In this chapter, a review of the intensity emission versus the accelerating voltage
and thin film thickness is given followed by a comparison of three different
geometries of SDDs. The effect of the large solid angle is illustrated using X-ray
maps. Examples of high spatial resolution X-ray “micrographs” acquired with
different imaging conditions are given and application examples of X-ray imaging
with the f-ratio method are given at the end of the chapter.
7.1 X-Ray Emission Rate with Low Accelerating Voltage

and Thin Film
The physics of X-ray emission by the interaction of electrons with the atom of a
specimen is well understood and described in SEM textbooks (Goldstein et al.
2003; Reimer 1998). However, it is difficult to practically grasp the effect of, for
example, accelerating voltage on emitted intensity for a specific specimen. For
example, to decrease the interaction volume of X-ray emission, a low accelerating
voltage should be used or the specimen thickness reduced like in the case of thin
films. Practically, Monte Carlo simulation programs like CASINO (Drouin et al.
2007; Hovington et al. 1997), DTSA-II (Ritchie 2005, 2009), or MC X-ray (Gauvin
and Michaud 2009) are useful to quickly monitor the effect of a parameter on the
X-ray emission.
Figure 7.1 shows the variation of the emitted X-ray intensity with the acceler-
ating voltage for a Fergusonite-Ce specimen and with the thickness for an alu-
minum thin film. The intensities were simulated with the MC X-ray Monte Carlo
program (Gauvin and Michaud 2009), which uses similar models than CASINO,
but allows more complex 3D specimen geometry.
Figure 7.1a shows the variation of the emitted X-ray intensity of various
element-line pairs with accelerating voltage from a complex composition specimen.
The specimen is a Fergusonite mineral containing rare earth elements and the
composition was taken from the Mineralogy Database (webmineral.com 2017). The
intensity variation is complex and many factors influence the X-ray emission:
Concentration of the element in the specimen, X-ray line analyzed (Ka, La, or Ma),
and the type of element. The nature of the element is also of importance: Light
element (oxygen), medium element (yttrium and niobium), or heavy element
(lanthanum and cerium). For bulk specimens, the intensity increases with the
accelerating voltage because of the larger interaction volume, i.e., the electron
interacts with more atoms and thus potentially generates more X-rays. This effect is
observed for most element-line pairs in Fig. 7.1a. However, strong X-ray absorp-
tion will have the reverse effect and decrease the intensity when the accelerating
voltage increases as observed with the oxygen Ka X-ray line. As the accelerating
voltage increases, the X-rays are generated deeper in the specimen and the
7.1 X-Ray Emission Rate with Low Accelerating Voltage … 69
(a)
(b)
Fig. 7.1 Variation of the emitted X-ray intensity with: a accelerating voltage for a Fergusonite-Ce
mineral sample; b thickness of a thin film for a pure aluminum sample
probability of an X-ray interacting with an atom of the specimen increases before

they can reach the surface and decreases the total emitted intensity. Generally, the
absorption is strong for low energy X-rays: K lines for light elements, L lines for
medium elements, and M lines for heavy elements.
For the same element and same accelerating voltage, the relative intensity for
different X-ray line depends on the ionization cross sections and the mass ab-
sorption coefficients (MAC). The cross section is larger for outer atomic shells
(M > L > K) (Llovet et al. 2014) and the interaction volume is larger as well. The
ionization energy is larger for inner atomic shells (K > L > M) which means the
interaction volume of electron with enough energy to ionize the atom is smaller and
closer to the surface. For example, the intensity of La is larger than the Ka for Y and
Nb. However, the Ce Ma intensity is larger than La at low voltage, but the reverse is
observed at larger accelerating voltage. Niobium and yttrium have very similar
X-ray properties, but represent different fractions in the specimen (0.31wt% for Nb
and 0.03wt% for Y), which explains why the Nb Ka and La lines have higher
intensity than Y. This example shows how difficult it is to predict the X-ray
intensity for a specific condition and how useful and essential Monte Carlo simu-
lation software are to predict and understand X-ray emission. Lowering the
accelerating voltage results in an improved spatial resolution, i.e., smaller interac-
tion volume, but limits the choice of the X-ray line that can be analyzed and
decreases the emitted X-ray intensity as shown in Fig. 7.1a. Decreasing the spec-
imen size is therefore another way to obtain smaller interaction volume.
Figure 7.1b shows the quick decrease of the X-ray intensity with decreasing the
film thickness. The intensity decreases by a factor of 40 when we compare a bulk
specimen (*400 10−6 photon/e-/sr) to a 100 nm-thick film (*10 10−6 pho-
ton/e-/sr) at 20 kV. This means that using the same acquisition conditions, one needs
to increase the acquisition time by a factor 40 to obtain comparable results with a thin
film and a bulk specimen. In practice, the acquisition time will be limited by the
specimen or/and probe drift, specimen contamination, and maximum dose for probe
sensitive specimen. Without improvement of the collection count rate, it will not be
possible to collect an X-ray image and only line profiles or point analysis will be
possible due to this limitation. The plateau observed at 20 kV for large thicknesses
indicates that the film is equivalent to a bulk specimen, i.e., the thickness is larger
than the maximum emission depth (*3000 nm) for this accelerating voltage. At
30 kV, the maximum emission depth is around 7000 nm which demonstrates the
large interaction volume at high accelerating voltage. Another interesting observa-
tion on this figure is the comparison of the intensities (for the same thickness) at the
two accelerating voltages. For a bulk specimen, the intensity is larger at 30 kV
because of the small absorption of the Al Ka X-ray in the specimen. Increasing the
accelerating voltage increases the interaction volume which results in a greater
emitted intensity. However, for a thin film, this effect is reversed and the intensity is
larger at 20 kV. Large accelerating voltage results in less interaction of the electron
with the specimen, i.e., less elastic scattering and less ionization events. Lowering
the accelerating voltage increases these interactions, which result in more emitted
X-rays. However, the interaction volume increases which results in degrading
spatial resolution.
7.2 Comparison of Silicon Drift Detector Geometry
The position of the conventional SDD (c-SDD) inside the SEM chamber limits its
solid angle as the specimen-detector distance is limited by the objective lens
geometry. The solid angle is increased using larger detection area (larger sensor).
However, the maximum solid angle is still of the order of tenth of steradian
7.2 Comparison of Silicon Drift Detector Geometry 71
(*0.4 sr). Two designs were recently proposed to overcome this limitation: an
annular SDD (a-SDD) and an optimized geometry SDD (g-SDD). However, larger
solid angle leads to larger range of takeoff angles across the detector sensor and the
effect on the quantification was studied.
7.2.1 Solid Angle
Figure 7.2 shows the solid angle variation with the specimen-detector distance for
three SDD geometries. The detector solid angle values were calculated using
equations developed for different geometries (Zaluzec 2009) and detector geometry
data provided by each SDD manufacturer. The transition between dashed and full
line in Fig. 7.2 shows the minimum distance possible with a FE-SEM.
An a-SDD is positioned just below the objective lens (Terborg et al. 2017) and
specimen-detector distance is not limited, which allows large solid angle (1.4 sr),
3.5 time larger than a large area c-SDD as shown in Fig. 7.2. The annular SDD has
four sensors and a central hole to allows the electron beam to pass through. The
hole diameter explains the decrease of the solid angle observed at very small
specimen-detector distances. The maximum solid angle is found at an optimal
distance. Because of the position of the annular detector, an electron trap magnet
cannot be used and Mylar windows are used instead to prevent the backscattered
electrons (BSEs) from interacting with the SDD sensors. A drawback of this a-SDD
is that C and O Ka X-rays are generated by the BSEs interacting with the Mylar
Fig. 7.2 Variation of the solid angle versus specimen-detector distance for three types of silicon
drift detector energy dispersive spectrometry (SDD-EDS) detectors. The minimum achievable
specimen-detector distance in a FE-SEM for each detector is indicated by the transition from the
dashed line to the continuous line which represents experimental positions
windows and the intensity increases with smaller specimen-detector distance

(Demers et al. 2014).
Another approach is to reduce the size of the electron tap magnet and optimize
the SDD sensor geometry, i.e., using a rectangular shape (Burgess et al. 2017).
These two modifications allow to obtain larger solid angle detectors, which we call
optimized geometry SDD (g-SDD). The improvement is shown in Fig. 7.2 and
further geometry optimization may be possible. The drawback of this design is that
the SDD can only be used at low accelerating voltage as the reduced electron trap
cannot stop high energy BSE to reach the detector sensor.
Another limitation of a c-SDD for detection of low energy X-rays is the strong
absorption in the detector window, which isolate the detector sensor from the SEM
chamber and blocks light to reach the detector sensor. One approach is to
change the material and thickness of the window to improve the transmission of the
window for low energy X-rays (Nylese and Rafaelsen 2017). Another approach
is to remove the window (windowless SDD) and use a SDD with rapid sen-
sor cooling/warming using thermoelectric cooling and a very quick signal-
measurement stability electronics of the detectors after cool down (Burgess et al.
2017).
7.2.2 Takeoff Angle
For quantification microanalysis where the absorption of the X-ray is important, the
value of the takeoff angle (TOA) is important. Lower value increases the absorption
in the specimen. Also, current correction models suppose a fix value of takeoff
angle.
Figure 7.3a shows takeoff angle values for the three detector designs presented
in the previous section. Conventional SDD has a fix nominal value of takeoff angle,
i.e., value at the center of the sensor. This value is used as takeoff angle input in
current correction models. For typical SEM, the value is between 30° and 40°. The
optimized geometry SDD can have a similar value, but in some FE-SEM, the
takeoff angle is small (*16°) to minimize the specimen-detector distance. Because
of the geometry of the annular SDD, the nominal takeoff angle changes with the
specimen-detector distance. Furthermore, because of the large area of the sensor,
the incident X-rays are emitted from a range of takeoff angles and not from a single
value as supposed by the correction models. This range is illustrated in Fig. 7.3a by
the colored area for each SDD.
This effect was studied using the PAP correction model (Pouchou and Pichoir
1991) for an Al–Mg alloy (Mg 90.0wt%). Figure 7.3b–d show the Al Ka k-ratio
values versus the accelerating voltage. Al Ka X-ray line is strongly absorbed in Mg
versus in Al by a factor 10, so the k-ratio will be also strongly affected by the
takeoff angle. The k-ratio was calculated for takeoff angles given by the nominal
TOA and the colored area were obtained with TOA minimum and TOA maximum
values. The effect of the range of TOA, called full calculation, was calculated by
7.2 Comparison of Silicon Drift Detector Geometry 73
(a) (b)
(c) (d)
Fig. 7.3 a Variation of the takeoff angle versus specimen-detector distance for three types of
silicon drift energy dispersive spectrometry (SDD-EDS) detectors. For each detector, the curve
starts at the minimum specimen-detector distance (see Fig. 7.2). b–d Variation of the calculated
k-ratio versus accelerating voltage using the nominal takeoff angle (blue line) or including the
variation of takeoff angle (full calculation, orange circle), see text for details. b Conventional SDD,
c optimized geometry SDD and d annular SDD. The blue area indicates the variation of the k-ratio
versus the minimum and maximum takeoff angle for each detector. The difference between the two
k-ratio calculations is shown at the bottom of each figure
summing the intensity obtained for 1000 TOAs linearly distributed between TOA
minimum and TOA maximum for both the alloy and standard, then the k-ratio was
obtained from these summed intensities and shown with orange circle in Fig. 7.3b–
d. The full calculation represents the k-ratio obtained experimentally and the
nominal calculation is the k-ratio calculated by the correction model to obtain the
concentration. Any difference between the two calculations represent an error in the
calculation of the concentration. A difference up to 2% was obtained for Al Ka with
a c-SDD and a-SDD. A larger difference of 6% for a g-SDD was obtained, because
of the lower value of the nominal TOA. However, the difference obtained for
Mg Ka was negligible for all geometries of SDDs, which shows that only strong
absorption cases are affected by the large range of TOAs.
These calculations show that the difference observed due to the range of TOAs is
not negligible in cases where the absorption is strong. Low energy X-rays have
strong absorption in most cases and are affected by this effect. This large takeoff
angle variation affects the absorption and correction models should include this
effect for accurate quantification results.
7.3 X-Ray Map Acquisition at High Spatial Resolution

and High Signal-to-Noise Ratio
Different information can be shown on an X-ray map. The raw intensity is obtained
using a region of interest (ROI) around an X-ray peak and summing all X-ray
counts in that ROI for each pixel. This raw intensity contains counts from the
background and possibly overlapped peaks. The interpretation of the map is diffi-
cult for low intensity elements as the variation could be from background or
overlapped peaks and not from the element of interest itself. The net intensity is
obtained by removing the background and overlapped peaks contributions from the
spectrum at each pixel. This net intensity map gives a qualitative elemental dis-
tribution, but the quality of the map depends on the method used to remove the
contribution of the background and overlapped peaks. From this net intensity, a
quantification analysis (with or without standards) is used to calculate the weight
concentration (or atomic concentration) and can be displayed as a concentration or
quantitative map. Using the f-ratio method described in Chap. 6, a f-ratio map is
calculated from either the raw or the net intensities.
Figure 7.4 illustrates the effect of a large solid angle by comparing an X-ray map
obtained with a c-SDD and a-SDD in the same conditions with a short acquisition
time. The map pixel size used (1024 768 pixels) is similar to the one normally
used for electron imaging. The map was acquired at 5 kV and with the same
working distance of 15 mm for both SDDs. This is the optimum working distance
for the c-SDD, i.e., optimum count rate. For the a-SDD, the optimum working
distance is around 9–10 mm.
The raw intensity of Zn is shown in Fig. 7.4. After 1 cycle, the c-SDD map
contains only 753 total counts and no information about the phase in the specimen
was observed. However, two regions of high Zn concentration were visible after 1
cycle using the a-SDD, the total number of counts was 35,500. The c-SDD needs 21
cycles to observe the same two regions with a total number of counts of 15,000.
After 21 cycles with the a-SDD, a smaller region of high concentration and a region
of medium concentration were observed with a total of 887,000 counts. The map
acquisition time was 3.15 s for 1 cycle and 66.15 s for 21 cycles. Typical map
acquisition with the c-SDD will need a longer acquisition time to obtain comparable
results collected with the s-SDD.
The large output count rate of the annular SDD allows to work with acquisition
times of the same order as imaging acquisition times, which permits the acquisition
of elemental micrograph of the specimen with a comparable acquisition time as for
BSE imaging (Newbury and Ritchie 2011). Furthermore, using longer acquisition
times, the a-SDD is used in situation with low emission rate like low accelerating
voltage or low voltage STEM (LV-STEM).
7.3 X-Ray Map Acquisition at High Spatial Resolution … 75
Fig. 7.4 Zinc X-ray maps from a Galena sample using a conventional SDD (c-SDD) and annular
SDD (a-SDD). 1 cycle was 3.15 s total acquisition time and 21 cycles was 66.15 s total acquisition
time. The map size was 1024 768 pixels, the field of view 655 µm and the acquisition time per
pixel 4 µs per cycle. The map was acquired at 5 kV and with the same working distance of 15 mm
for both SDDs
7.3.1 Low Accelerating Voltage
Usually a high accelerating voltage, like 20 kV, is used to get large amount of
counts with an overvoltage greater than 2 for all elements in the specimen. In
Fig. 7.5, different accelerating voltages were used to observe the effect of accel-
erating voltage on spatial resolution. Figure 7.5 displays the net intensity maps of
Mg, Zr, and Y at accelerating voltages of 20, 10, and 5 kV. The maps at 20 and
10 kV were acquired for only 15 min, and the maps at 5 kV were acquired for
30 min. With the decrease of the accelerating voltage, the improved spatial reso-
lution makes the edges of each phase much sharper. The careful analysis with a
high spatial resolution shows that there were no Y existing in the Zr phase.
Furthermore, small Mg inclusions were observed in the 5 kV map while they
appeared very faint at 10 kV and were not observed at 20 kV. The acquisition of
low accelerating voltage maps with good signal to noise ratio were possible because
of the a-SDD large solid angle and the maps have an improved spatial resolution,
which gives information about the small phases in the specimen.
Fig. 7.5 Net intensity maps of a Fergusonite mineral sample acquired with the annular SDD at a
high magnification with a field of view of 11 µm and accelerating voltage of 20, 10, and 5 kV.
Adapted from (Teng et al. 2017) under permission from Cambridge University Press
Gold nanoparticles on carbon and carbon nanotubes (CNTs) decorated with Pt

nanoparticles are often used to evaluate the spatial resolution of CFE-SEMs.
Figure 7.6a shows an example of high spatial resolution imaging and X-ray
microanalysis at low accelerating voltage (3 kV) acquired on a resolution test
specimen consisting in gold nanoparticles on graphite. A resolution of 7.3, 8.6, and
9.4 nm were measured with SMART-J (Joy 2002; Kim et al. 2007) on the SE
micrograph, the Au X-ray map, and the C X-ray map, respectively. Another
example of high spatial resolution imaging and X-ray microanalysis was obtained
with CNTs decorated with Pt nanoparticles. Figure 7.6b shows an example of high
spatial resolution imaging and X-ray microanalysis of this specimen using a low
accelerating voltage of 2.5 kV and a working distance of 9.4 mm. A resolution of
19 and 24 nm were measured with SMART-J (Joy 2002; Kim et al. 2007) on the
SE micrograph and the Pt X-ray map, respectively. The Pt X-ray map was acquired
with an a-SDD. The map acquisition time was 24 min with a count rate of 81 kcps.
(a)
(b)
Fig. 7.6 a Secondary electron micrograph of gold nanoparticles on carbon was acquired at an
accelerating voltage of 3 kV and a working distance of 9.7 mm. The corresponding X-ray map was
acquired with an a-SDD. The map acquisition time was 35 min with a count rate of 53.3 kcps.
b Secondary electron micrograph of CNTs decorated with Pt nanoparticles was acquired at an
accelerating voltage of 2.5 kV and a working distance of 9.4 mm. The Pt X-ray map was acquired
with the same a-SDD used in (a). The map acquisition time was 24 min with a count rate of 81 kcps
7.3.2 Low Voltage STEM
Figure 7.7a shows another example of high spatial resolution X-ray map obtained
with an annular SDD of CNTs decorated with Pt nanoparticles with a CFE-SEM in
low voltage scanning transmission electron microscope (LV-STEM) mode at
20 kV. The dark-field (DF) micrograph shown in Fig. 7.7a had a spatial resolution
of 6.5 nm, which was calculated with SMART-J (Joy 2002; Kim et al. 2007).
However, the accelerating voltage and working distance were not optimum for this
kind of imaging. They were chosen to allow simultaneously the acquisition of
X-ray maps with the annular SDD. Currently, the system used here is limited to
accelerating voltages below 20 kV due to the Mylar windows’ thicknesses available
and the shortest working distance is around 9–10 mm, which is shorter than the one
used with a conventional SDD (15 mm on this microscope). Figure 7.7a shows a Pt
X-ray map (right side) obtained with the annular SDD. A probe current of 270 pA
with a count rate of 7 kcps and an acquisition time of 7 min were used. A spatial
resolution of 8.9 nm was calculated for this X-ray “micrograph”.
LV-STEM in the SEM provides now routinely images with sub-nanometer
spatial resolution. This is possible because of the reduced spreading of the probe in
the thin specimen but also due to the large amount of signal available emerging
from the bottom side of the foil. Core excitation of the target atoms is a very low
efficiency process compared to elastic scattering and the small solid angle of c-SDD
detectors does not provide sufficient X-ray signals to acquire chemical maps in a
Fig. 7.7 a Dark field micrograph of CNTs decorated with Pt nanoparticles acquired in LV-STEM
mode. The Pt X-ray map was acquired with an a-SDD. An accelerating voltage of 20 kV and a
working distance of 10.5 mm were used. The map acquisition time was 7 min with a count rate of
7 kcps. b High spatial resolution X-ray elemental map of an Al–Li 2099 alloy at 20 kV in
LV-STEM mode. The T1 plates (Al2CuLi) observed in the bright-field micrograph (TE) are clearly
resolved in the Cu Ka map. The map acquisition time was *2 h with a count rate of 30 kcps
reasonable amount of time due to the small emission volume in transmission mode.
However, the a-SDD detector permits to acquire X-ray data more than 200 times
faster and then provide chemical maps of thin specimens with high resolution and
high signal-to-noise ratio in a reasonable timescale. As an example, Fig. 7.7b shows
a STEM-BF image (TE, transmitted electrons) with the corresponding Cu Ka X-ray
maps obtained with the a-SDD in approximately 2 h with a count rate of 30 kcps on
a thin electropolished specimen of a 2099 Al–Li alloy at 20 kV. The T1 plates
(Al2CuLi) of 1–2 nm thickness are clearly resolved and the spatial resolution is
close to 5 nm. The T1 s are clearly visible in the Cu map because their thickness is
broadened due to electron scattering in the specimen, making them larger (5–
10 nm) than the spatial resolution of *5 nm obtained in these conditions.
7.3.3 Phase Map Analysis
A useful method to display the results of X-ray mapping analysis is to display the
spatial information by phases. An overview of different visual representation was
presented recently (Nylese and Anderhalt 2014). Various commercial platforms
exist for mineral identification and classification to present the analyzed phases of a
specimen. Using a SEM, one approach to identify the phases is to use the scatter
diagram of two element and manually select the observed clustering (Moran and
Wuhrer 2006). The clustering (and phases) can also be found automatically using
multivariate statistical analysis (MSA) and principal components analysis (PCA)
(Kotula et al. 2003), other numerical algorithms (Martins et al. 2013; Müllner 2013;
Münch et al. 2015; Lavoie et al. 2016), or with a toolbox combining various
algorithms (de la Peña et al. 2017).
Figure 7.8 shows a manual phase map analysis using the f-ratio method (Teng
et al. 2017) of an aluminum alloy specimen. In this example, the f-ratio maps for
each element were calculated from the net intensity and a histogram distribution of
f-ratio values was then calculated using an open source Python script, pyXRay
Phase Map (available on GitHub as pyxray phase map repository). The phases were
manually defined by the ranges of the f-ratio values of the constituted elements,
then the phase map was generated. The f-ratio is not the exact composition, but it is
proportional to the concentration of each element as described in Chap. 6 (Fig. 6.1).
During the definition of each phase, a histogram of the f-ratio of each element is
generated and the f-ratio range adjusted visually. Figure 7.8a, b show how the
histogram and the f-ratio map is used to define different phases related to the peaks
in the histogram and the corresponding areas in the f-ratio map clearly show the
phases with different compositions of Al and B. Doing this visual comparison for
each element allows to define each phase of the specimen and generate the phase
map shown in Fig. 7.8c. To help the determination of each phase, a “no phase” and
an “overlap phase” maps can also be generated which indicate if a phase is missing
or if two phases have overlapped definitions of element f-ratio ranges. Furthermore,
the map in Fig. 7.8 was acquired with a large number of pixels, which leads to
Fig. 7.8 Phase map analysis of an aluminum alloy specimen acquired at 3 kV for 11 min with an
a-SDD. a f-ratio distribution histogram and map for aluminum. b f-ratio distribution histogram and
map for boron. c Phase map distribution of all phases analyzed, dark areas represent regions where
the analysis was not able to identify a phase. The map field of view was 41 µm and the map pixel
resolution 1024 768 pixels
noisy f-ratio map as statistical variation is observed in adjacent pixels of the same
phase. This noise was reduced by using a median filter of size 4 when calculating
the f-ratio map for each element.
7.3.4 Removal of the Effect of Electron Channeling

on X-Ray Emission in Thin Specimens
Recently, the f-ratio method was described as a good alternative to remove the
effect of electron channeling on the emission of X-rays in thin crystalline specimens
via the f-ratio method (Brodusch and Gauvin 2017) exemplified on a thin specimen
of a Ti-6Al-4 V alloy. Low magnification net intensity Ti Ka X-ray images were
recorded from several areas of an electropolished Ti-6Al-4 V thin specimen and are
shown in Fig. 7.9 (top). The impact of electron channeling and specimen thickness
on the net X-ray emission is striking. As seen from Fig. 7.9a, b, the defect density
renders the interpretation of the qualitative images very difficult. Similarly, the net
intensity gradient observed in Fig. 7.9b clearly shows a dramatic variation between
the top left and the bottom right corners, which follows the direction of the wedge
due to the specimen thinning process. In Fig. 7.9c, the change of intensity due to
electron channeling is even more dramatic. A contrast of 40% was calculated from
two distinct alpha grains with dark and bright contrast (see white squares on
Fig. 7.9c). Each grain with different crystallographic orientation has different
electron mean free path (because of the channeling effect), which leads to different
(a) (b) (c)

Net intensity
f-ratio
Fig. 7.9 Low magnification net intensity X-ray and corresponding f-ratio images of a Ti-6Al-4 V
thin specimen. a–c Net intensity X-ray images for Ti Ka (top) and corresponding f-ratio images
(bottom). For the sake of clarity, only the Ti Ka images are shown. Reproduced under permission
from John Wiley and Sons (Brodusch and Gauvin 2017)
number of ionization of the K shell of titanium. Since the mean free path is the same
in a grain with a specific crystal orientation, the f-ratio method removes the
channeling effect since the electron mean free path is the same for each element and
the f-ratio is not affected by the change of mean free path of different grain ori-
entation. This is shown in Fig. 7.9 where the f-ratio images were calculated from
the net intensity. The effectiveness of the f-ratio method in cancelling the effects of
specimen thickness (Fig. 7.9b) as well as defect density (Fig. 7.9a, b) and chan-
neling (Fig. 7.9c) is thus clearly observed.
7.4 Summary
Two designs were proposed to increase the solid angle: an annular SDD and an
optimized geometry SDD. The unique geometry and position of the a-SDD com-
bines high count rate capabilities with high collection efficiency (1.4 sr solid angle),
which is good for low-count or probe-sensitive specimens because of the less
specimen/probe drift and faster acquisition time. However, larger solid angle leads
to larger range of takeoff angles across the detector sensor(s) and affect the quan-
tification. This large takeoff angle variation affects the absorption and correction
models should include this effect in the future in order to obtain accurate quan-
tification results. Low energy X-rays have strong absorption in most cases and are
even more affected by this effect.
The large solid angle of some SDDs allows an acquisition time of the same order
as imaging acquisition time, which allows the acquisition of elemental and quan-
titative micrographs of the specimen with simultaneous electron imaging.
Furthermore, using longer acquisition time, the a-SDD is used in situations with
low emission rates like low accelerating voltage or low voltage STEM with X-ray
spatial resolution less than 10 nm. Finally, the f-ratio maps provide a useful and
simple way to reduce the effect of specimen thickness and cancel that of electron
channeling on X-ray emission in thin specimens.
References
Brodusch, N., & Gauvin, R. (2017). The qualitative f-ratio method applied to electron
channelling-induced X-ray imaging with an annular silicon drift detector in a scanning
electron microscope in the transmission mode. Journal of Microscopy.
Burgess, S., Sagar, J., Holland, J., Li, X., & Bauer, F. (2017). Ultra-Low kV EDS—A new
approach to improved spatial resolution, surface sensitivity, and light element compositional
imaging and analysis in the SEM. Microscopy Today, 25, 20–28.
de la Peña, F., Ostasevicius, T., Fauske, V. T., Burdet, P., Jokubauskas, P., Nord, M., et al. (2017).
Hyperspy/hyperspy: HyperSpy 1.3.
References 83
Demers, H., Brodusch, N., Gauvin, R., & Woo, P. (2014). Nanoscale materials characterization by
X-ray microanalysis with high spatial resolution. In: COM 2014 Conference Proceedings.
MetSoc.
Drouin, D., Couture, A. R., Joly, D., Tastet, X., Aimez, V., & Gauvin, R. (2007). CASINO V2.42—
A fast and easy-to-use modeling tool for scanning electron microscopy and microanalysis users.
Scanning, 29, 92–101.
Gauvin, R., & Michaud, P. (2009). MC X-ray, a new Monte Carlo program for quantitative X-ray
microanalysis of real materials. Microscopy and Microanalysis, 15, 488–489.
Goldstein, J. I., Newbury, D. E., Echlin, P., Joy, D. C., Romig Jr, A. D., Lyman, C. E., et al.
(2003). Scanning electron microscopy and X-ray microanalysis: A text for biologists, materials
scientists, and geologists. Plenum Press.
Hovington, P., Drouin, D., & Gauvin, R. (1997). CASINO: A new Monte Carlo code in C
language for electron beam interaction—Part I: Description of the program. Scanning, 19,
1–14.
Joy, D. C. (2002). SMART—A program to measure SEM resolution and imaging performance.
Journal of Micrsocopy, 208, 24–34.
Kim, J., Jalhadi, K., Lee, S.-Y., & Joy, D. C. (2007). Fabrication of a Fresnel zone plate through
electron beam lithographic process and its application to measuring of critical dimension
scanning electron microscope performance. Journal of Vacuum Science Technology B:
Microelectronics and Nanometer Structures, 25, 1771–1775.
Kotula, P. G., Keenan, M. R., & Michael, J. R. (2003). Automated analysis of SEM X-ray spectral
images: A powerful new microanalysis tool. Microscopy and Microanalysis, 9, 1–17.
Lavoie, F. B., Braidy, N., & Gosselin, R. (2016). Including noise characteristics in MCR to
improve mapping and component extraction from spectral images. Chemometrics and
Intelligent Laboratory Systems, 153, 40–50.
Llovet, X., Powell, C. J., Salvat, F., & Jablonski, A. (2014). Cross sections for inner-shell
ionization by electron impact. Journal of Physical and Chemical Reference Data, 43.
Martins, D. S., Josa, V. M. G., Castellano, G., & da Costa, J. A. T. B. (2013). Phase classification
by mean shift clustering of multispectral materials images. Microscopy and Microanalysis, 19,
1266–1275.
Moran, K., & Wuhrer, R. (2006). X-ray mapping and interpretation of scatter diagrams.
Mikrochimica Acta, 155, 209–217.
Müllner, D. (2013). Fastcluster: Fast hierarchical, agglomerative clustering routines for R and
Python. Journal of Statistical Software, 53, 1–18.
Münch, B., Martin, L. H. J., & Leemann, A. (2015). Segmentation of elemental EDS maps by
means of multiple clustering combined with phase identification. Journal of Microscopy, 260,
411–426.
Newbury, D. E., & Ritchie, N. W. M. (2011). Can X-ray spectrum imaging replace backscattered
electrons for compositional contrast in the scanning electron microscope? Scanning, 33,
174–192.
Nylese, T., & Anderhalt, R. (2014). Advanced materials characterization with full-spectrum phase
mapping. Microscopy Today, 22, 18–23.
Nylese, T., & Rafaelsen, J. (2017). Improvements in SDD efficiency—From X-ray counts to data.
Pouchou, J. -L., & Pichoir, F (1991). Electron probe quantitation. In K. F. J. Heinrich &
D. E. Newbury (Eds.). Plenum Press.
Reimer, L. (1998). Scanning electron microscopy: Physics of image formation and microanalysis.
Springer.
Ritchie, N. W. M. (2005). A new Monte Carlo application for complex sample geometries. #sia#,
37, 1006–1011.
Ritchie, N. W. M. (2009). Spectrum simulation in DTSA-II. Microscopy and Microanalysis, 15,
454–468.
Teng, C., Demers, H., Brodusch, N., Jordens, A., Waters, K., & Gauvin, R. (2017). X-ray
microanalysis phase map on rare earth minerals with a field emission scanning electron
microscope and an annular silicon drift detector. Microscopy and Microanalysis, 22, 96–97.
Terborg, R., Kaeppel, A., Yu, B., Patzschke, M., Salge, T., & Falke, M. (2017). Advanced
chemical analysis using an annular four-channel silicon drift detector. Microscopy Today, 25,
30–35.
webmineral.com. (2017). Mineral data from webmineral.com.
Zaluzec, N. J. (2009). Detector solid angle formulas for use in X-ray energy dispersive
spectrometry. Microscopy and Microanalysis, 15, 93–98.
Chapter 8
Electron Diffraction Techniques in the SEM
The first observation of an electron diffraction pattern from a bulk specimen by

Coates and co-workers in 1967 (Coates 1967) in the form of an electron channeling
pattern (ECP) is taken nowadays as the birth of electron diffraction techniques in the
SEM. Before this date, electron diffraction was solely a TEM experiment. Although
Kikuchi (1928) recorded the first Kikuchi pattern in 1928 from a thin film of mica,
it was only in 1935 that Kossel et al. (1935) reported the appearance of Kossel lines
out of a large specimen bombarded by high energy electrons. In this experiment, the
x-rays generated by inelastic interactions of the incident electrons with the core
shell electrons of the specimen’s atoms were diffracted and emitted from the
specimen surface as Kossel cones. These cones were projections of the lattice
planes and were collected using a phosphorescent screen. But because the gener-
ation mechanism of those Kossel cones is highly dependent on the relation between
the energy of the incident electrons and the ionization threshold energies of the
target atoms (Engler and Randle 2010), this limited technique was not used further.
Later, high angle Kikuchi bands were observed by Alam et al. (1954) on various
crystals in reflected mode which led to the technique known as reflection
high-energy electron diffraction (RHEED). While this technique spawned the
electron backscatter diffraction (EBSD) technique two decades later (Venables and
Harland 1973), the discovery of Coates permitted to evidence a new form of
electron diffraction from bulk crystals and led to a new type of BSE imaging, the
electron channeling contrast imaging (ECCI), for which the acquisition of electron
channeling patterns allowed to understand, under the sight of theoretical calcula-
tions, some of the image contrast observed.
Forty years from these findings, ECCI and EBSD techniques are still evolving
and provide new insights never achieved before. In the coming sections, we give a
succinct summary of the mechanisms that explain the observed contrasts supported
by several applications using these techniques. For a more complete description of
the theoretical models, the reader is referred to the literatures cited in this section.

86 8 Electron Diffraction Techniques in the SEM
8.1 Electron Channeling Contrast Imaging
An electron diffraction pattern represents the angular distribution of the BSE yield
obtained when the electron beam is scanned over a large single crystal or rocked
around the optic axis of the microscope on a smaller specimen area. The pattern
consists of pseudo-Kikuchi lines appearing in pairs for planes (hkl) and (-h-k-l).
These lines are kinematically interpreted as the projection of the lattice planes on a
virtual plane situated above the objective aperture and normal to the optic axis. In
fact, although the two-beam approximation of the dynamical theory reproduces
quite well the main features of the ECP, only simulations based on many-beam
dynamical diffraction (Winkelmann 2009; Winkelmann et al. 2007) achieved to
approach quite nicely the contrast observed in experimental ECPs, especially at
zone axis (Winkelmann et al. 2003). In this model, the channeling or scattering
probability of the incident electrons inside the diffusion volume is dependent on the
Bloch wave coefficients defining the power of each wave and their direction of
propagation. In the two-beam approximation, only two waves are considered: Bloch
wave of type I has its maximum between atomic rows while Bloch wave of type II
has its maximum in the vicinity of atomic rows. In fact, at least several beams are
necessary to explain the experimental features (Marthinsen and Hoier 1986; Reimer
1998) but the more accurate simulations were obtained when using a large number
of Bloch waves (Winkelmann 2009).
When the primary beam is scanned in a raster along the specimen surface, two
scenarios are possible: if the scan angle is larger than the beam convergence angle,
i.e., the image field of view is larger than the objective aperture diameter, the
resulting image will be a magnified view of the ECP depending on the magnifi-
cation and the size of the grain. However, if the scan angle is of the order or smaller
than the beam convergence angle, i.e., smaller than the angular resolution of the
ECP according to Joy (Holt et al. 1974), the angle of the primary beam with the
specimen surface is constant and the BSE intensity recorded by the detector at each
pixel is that at the center of the virtual ECP for this specific angle along the optic
axis of the electronic column (Holt et al. 1974). Consequently, the pixel intensity is
only driven by the crystal orientation inside the volume of emission of the diffracted
BSEs. Thus, any local crystal orientation change induced by deformation mecha-
nisms and strain fields automatically generates a contrast in the BSE image.
The depth resolution of an ECCI micrograph is related to the extinction distance
of the excited diffracted beams contributing to the image, i.e., the Kikuchi lines
crossing the optic axis of the microscope. To this end, high voltage is desirable
because the Kikuchi band width is proportional to 1/(ng |g|) with ng the ex-
tinction distance of reflection g in turn proportional to E0 and g, the reciprocal
lattice vector in the two-beam approximation (Schulson 1971). Thus, a small
angular change in orientation results in a larger range of contrast at high voltage
than at lower voltage. However, the concomitant increase of the depth of signal
related to the increase in E0 generates a high background due to the diffracted
beams incoherently scattered before exit. The radial resolution, as mentioned
8.1 Electron Channeling Contrast Imaging 87
earlier, is solely dependent on the probe size. But the contrast, which is dependent
on the BSE intensity at the center (optic axis) of the ECP, is related to the con-
vergence angle which defines the ECP angular resolution [see Fig. 10, Chap. 6 in
Holt et al. (1974)], hence the possible angular deviation that produces a change in
the BSE emission.
Since recently, the main interest in using ECCI was to observe poly-crystallinity
and deformation structures of metal and alloys where the contrast was not quan-
titatively interpretable. Nowadays, a great interest of using this technique reside in
the observation of dislocation structures in a semi-quantitative way. Even single
dislocations can be observed (Gutierrez-Urrutia et al. 2013; Zaefferer and Elhami
2014) in crystalline materials if the dislocation density is low, otherwise the overlap
between dislocations throughout the BSE emission volume makes them undis-
cernible. The mechanism at the origin of dislocation contrast has been explained
some time (Hirsch et al. 1965) before the first image was successfully recorded by
Morin and co-workers in 1979 (Morin et al. 1979) in a FE-SEM. In fact, the main
factors for observing dislocations in a SEM are the probe size, the minimum probe
current necessary to make the dislocation line visible related to a visibility criterion
and the convergence semi-angle a. According to Spencer et al. (1972), the
appropriate beam convergence angle is defined by 2a < 1/(gng), the maximum
probe size by d < (ng g b)/4, with b the Burgers vector, and the probe current Ip
by Ip > k/(C − 1)2 with k a constant dependent on the BSE detector and C = (Imax/
Imin) with Imin and Imax the minimum and maximum intensity levels in the image. In
other words, high probe current combined to a small beam convergence angle and
to a small probe size are the keys to image dislocations with high spatial resolution.
According to Spencer calculations at that time, only field-emission guns where
identified to possibly provide sufficient high brightness to satisfy these require-
ments. For this reason, Morin et al. could image dislocations (Morin et al. 1979)
only in 1979 with one of the first filed-emission column attached to a
Stereoscan SEM.
When the angle between a specific set of lattice planes and the primary beam h is
slightly larger [deviation parameter w = wc, see Fig. 4 in Zaefferer and Elhami
(2014)] than the exact Bragg position hB, which is typically a few degrees at
classical SEM voltages, the channeling of primary electrons is at its maximum.
Thus, the BSE yield is at a minimum value. Then, a dislocation line in a specific
direction induces a strain field around the dislocation core that bends the nearest
neighboring planes on each sides of the dislocation line. This plane bending, as seen
by the incident electron, induces a change in h via w that induces in turn an
excitation enhancement of the Bloch waves with their maximum at the atomic
nuclei. In consequence, the BSE yield becomes larger than that from a defect-free
region of the crystal and the resulting dislocation image appear as a bright line or
spot, depending on the direction of the dislocation line, under a dark background
(Zaefferer and Elhami 2014). The dislocations near the surface will show up with a
black to white contrast across the dislocation line’s strain-field whereas dislocations
located deeper in the BSE emission volume will only provide uniform brightness
over the background (Holt et al. 1974). The same mechanism is involved in stalking
fault imaging but in this case, the fault image appears as bright fringes of periodicity
ng. The number of fringes increases with E0 (Spencer et al. 1972) and depends on
the depth of the stalking fault. Note that a more complicated contrast can be
obtained with different values of the deviation parameter w, but in this case the
resulting contrast is more complicated to interpret quantitatively.
In Fig. 8.1 are presented ECCI images of a deformed iron (a), a compressed
austenitic steel (b), a deformed 5083 aluminum alloy (c), and a Ti-6Al-4V alloy
specimens. For all specimens, a final polishing step with a mixture of colloidal
silica and hydrogen peroxide was used prior to obtaining high quality dislocation
images. Figure 8.1a, c display clear single dislocations in a complicated network.
As seen from these images, the possibility of imaging dislocations and precipitates
at the same time on a large scale provided by bulk imaging in the SEM allows to
observe a large number of orientations when analyzing deformation structures and
mechanisms. Many specific defect arrangements can be characterized like the
density of staking faults as seen from Fig. 8.1b or dislocation alignment at grain
boundaries in Fig. 8.1d. This later configuration was clearly evidenced using
high-resolution BF-STEM imaging combined with EDS mapping as reported in
Fig. 7 of the paper by Brodusch and Gauvin (2017). The dislocations first align to
form a misorientation boundary on which the b-phase crystallites nucleate. In
(a) (b)
(c) (d)
Fig. 8.1 Defect structures observed in bulk materials using electron channeling contrast imaging:
a Dislocations in a deformed iron, b stacking faults in a compressed austenitic steel, c dislocations
in a deformed 5083 aluminum alloy and d dislocations aligned along a grain boundary in an
Ti-6Al-4V alloy. E0 = 20 kV for a, c and 15 kV for b, d
8.1 Electron Channeling Contrast Imaging 89
Fig. 8.1d, the b-phase is identified as the bright line, which is enriched in vanadium,
and the dislocations are perfectly aligned with it as expected.
Dark and bright regions originate from the mechanism above mentioned where
dark regions denote a Kikuchi line (h = hB) crossing the optic axis of the SEM
while bright regions over them correspond to a deviation from the Bragg position
(h = hB ± w) or is due to multiple beam excitations at the optic axis. This case is
generally not desired as it requires the general dynamical theory to be used for
interpretation instead of the more straightforward two-beam approximation. The
contrast observed is dependent on the local (pixel) orientation of the crystal com-
pared to the primary beam and the change in contrast is here directly related to the
deformation of the grains under investigation. As predicted by theory, the highest
contrast is obtained in dark grains for which h * hB. However, one can also notice
contrast in regions with higher intensity background and an inversion of dislocation
contrast (dark dislocations on bright background) was also obtained (not shown)
when tilting the same specimen to a few degrees. This may be explained by the fact
that the same contrast mechanism can be applied if the high intensity apex of the
line profile is selected [see Fig. 4 in Zaefferer and Elhami (2014)]. Another pos-
sibility resides in the fact that the orientation and the depth of the dislocation could
complicate the generation of the contrast and permit dislocations to be imaged with
a variety of excitation conditions. In addition to these dislocation images, bend
contours (Kaboli et al. 2014, 2015) are often observed in plastically deformed
specimens. They correspond to the local bending of the planes due to the internal
energy accumulated during the deformation process.
However, it has to be mentioned that, in highly deformed materials like those
presented here, the contrast may appear more complex than what is generally
presented in the literature. In fact, the rotation contour contrast (RCC) (Kaboli et al.
2015) is superposed with that originated from the dislocations inside the same
interaction volume. Therefore, this complicates greatly the visibility of defects since
only those for which h * hB at a specific rotation angle of the contour will be
visible. When a slight deviation of 1° or 2° is introduced by tilting the specimen, the
dislocations previously visible become faint or invisible while those from a pre-
viously brighter part of the contour become visible over a darker background. In
addition, the density of dislocations is also an important parameter to be considered
because the BSE image is a two-dimensional view of the dislocation structure
inside the BSE emission volume. When the density of dislocations is too large,
dislocation overlaps are likely to happen and finally only a uniform bright contrast
is reported on the image.
8.2 Low Voltage STEM Defects Imaging
Similarly to ECCI, the bright-field STEM image is dependent on the excitation of

Bloch waves and their interferences at the exit face of the thin foil when a crystalline
material is observed, while only elastic forward scattering is at the origin of the
contrast for amorphous specimens. Therefore, defects can be imaged using any
image contrast that makes use of diffracted beams. That includes BF and low-angle
DF as well as BSE imaging. As reported, BF defects image contrast is at a maximum
when the BF detector collection angle is of the order of the primary beam conver-
gence angle, defined by the objective aperture diameter (Maher and Joy 1976).
Fig. 8.2 Dislocation STEM-BF

(a)
structures in STEM mode.
a BF and b BSE images of the
same area and same scan and
c high resolution BF image
showing the distribution of
dislocations and stacking
faults in a Ti-6Al-4V alloy.
Partial dislocations are visible
on both ends of the stacking
faults. a Reproduced under
permission from Elsevier
(Chekir et al. 2017)
(b) BSE
(c) STEM-BF
8.2 Low Voltage STEM Defects Imaging 91
The advantage of using low primary beam voltage to observe defects of crys-
tallinity resides in the fact of using the increase of elastic/inelastic scattering
probabilities to increase the contrast (Morandi and Merli 2007). In combination to
this, the use of thin specimens allows to reduce the incident beam diffusion and
beam broadening and to work close to the two-beam approximation condition.
Hence, with nowadays FE-SEMs, STEM images at 30 kV accelerating voltage are
routinely obtained with a sub-nanometer resolution but the level of contrast is much
higher than that obtained with higher beam voltages. Figure 8.2a, c show BF and
BSE images, respectively, obtained with beam convergence and collection
semi-angles of 10 mrad from the same Ti-6Al-4 V titanium alloy as used in
Fig. 8.1d, prepared by jet-electropolishing and imaged at E0 = 30 kV. Individual
dislocation as well as walls of dislocations are observed, especially at the grain
boundaries. A close look at the contrast of Fig. 8.2c permits to identify stalking
faults and their concomitant partial dislocations as forecasted by the theoretical
calculations (Zaefferer and Elhami 2014). As expected, dislocations were also
visible when a solid-state BSE detector was used on top of the thin specimen
(Fig. 8.2b). Their contrast was maximized in dark grains which corresponded to
bright grains in the BF images because of enhanced channeling (see grain at the top
right of the image).
8.3 Electron Backscatter Diffraction
In 1973, Venables et al. reported that the angular distribution of backscattered

electrons could be observed on a fluorescent screen in spot mode with the sample
tilted to 70–80° with respect to the beam normal (Venables and Harland 1973). This
angular distribution appeared as pseudo-Kikuchi bands in a similar fashion as in an
ECP and gave rise to electron backscatter diffraction (EBSD) patterns (EBSPs). The
incident primary electron beam is first incoherently scattered, generating point
sources from which electron are elastically diffracted based on Bragg’s law. The
diffracted electrons that do not suffer incoherent scattering before reaching the exit
surface carry the diffraction information out of the specimen surface where they will
be detected in vacuum by a charged coupled device (CCD) camera. Following
Bragg’s law and the simple kinematical theory, the BSEs are emitted through cones
similar to Kossel cones and the intersection of these cones with the fluorescent
screen gives rise to pairs of lines. Their intensity is proportional to the square of the
crystal structure factor but a more accurate intensity distribution is obtained when
considering dynamical effects, especially at zone axis (Winkelmann et al. 2003,
2007). Because the primary electrons lose a varying amount of their initial energy
when they are inelastically scattered, the energy distribution of the BSEs partici-
pating to the EBSP is fairly large. However, most of the signal observed on the
EBSD camera comes from BSEs that have only lost 2–5% of their energy (Deal
et al. 2008) and this can be observed on the EBSP as the Kikuchi line width
(Zaefferer 2007). This can be explained by the fact that the energy distribution of
BSEs at high tilt present a narrow peak at high energy followed by a soft decrease at
medium and low energies. Those low energy BSEs that are diffracted will only
contribute to the blurring of the Kikuchi line in addition to the diffuse background.
Thus, the removal of the background from the raw EBSP permits to enhance the
contrast of the captured image by increasing virtually the high energy BSEs con-
tribution. This is combined with the excitation energy threshold of a few keV of
most of commercially available EBSD cameras. The origin of the BSE emission
anisotropy is related to electron channeling by the reciprocity theorem (Reimer
1998). The incident coherent electron distribution of the ECP, when scanning over
a large area or rocking the beam around a stationary point, is similar to the emitted
coherent BSE angular distribution in EBSD. As a result, the incoherent BSEs rising
out from the surface in the case of ECP are similar to the primary electrons inco-
herently scattered prior to diffraction in EBSD. In fact, the origin of the signal in
both techniques is related to the channeling through the Bloch wave theory, electron
channeling being described by the channeling-in of the primary electrons while in
EBSD the diffracted electrons are channeled-out from the exit surface (Wells 1999).
Due to the low signal of the useful diffracted electrons reaching the phospho-
rescent screen of the EBSD detector, a background removal is necessary before
scanning the specimen to acquire an EBSD map. Therefore, this require a stable
probe current in time since the timescale for an EBSD mapping sequence is in
hours. Hence, CFE-SEM have for a long time been considered too unstable to be
seriously used for EBSD map recording. However, with the new generation of
CFE-SEM, long term stability is obtainable, i.e., up to twelve hours (Fig. 2.1,
SU-8000). Especially with the auto-flash gun facility, high probe current with a
stability better than 10% can now be achieved on 24 h acquisitions (Fig. 2.1,
SU-8230). The EBSD band contrast and IPF maps shown in Figs. 10.1 and 10.4
were acquired with a CFE-SEM for several hours acquisition times. No loss of
signal was observed in these figures based on the BC signals, which is based on the
mean contrast of the detected Kikuchi bands over the full EBSP image intensity
(Maitland and Sitzman 2006). Of course, these results could have been obtained
with a Schottky emitter instead of a CFE-SEM. However, the main advantage of
CFE-SEMs is their higher brightness that allows high probe current while keeping
the probe size small, i.e., at the nanometer scale. This is of great importance when
high resolution EBSD mapping is required where the spatial resolution is close to
10–20 nm at best depending on the material atomic number and this permit to work
at the highest achievable EBSD spatial resolution. The most interesting application
of CFE-SEMs is with transmission forward scattering diffraction (t-EFSD) which
will be described in Sect. 8.5 of this chapter. However, one important drawback in
these systems is the influence of the magnetic field from the snorkel lens that
deviates the BSEs from their initial trajectories and results in curved Kikuchi bands
intercepting the EBSD screen. A post-processing procedure is then required to
unwarp the EBSPs before re-indexing all pixels of the map (Hovington et al. 2009).
8.4 Dark-Field Electron Backscatter Diffraction 93
8.4 Dark-Field Electron Backscatter Diffraction
Because the reciprocity theorem relates the electron channeling and the EBSD
patterns, an ECCI-like micrograph can be obtained if a selected region in a refer-
ence EBSD pattern is summed and used to produce a new image along the EBSD
map acquisition (Nowell et al. 2014; Wright et al. 2015) as forecasted by Harland
and co-workers long time ago (Harland et al. 1981). However, the sum of all pixels
from large selected regions allowed to acquire orientation images but their meaning
is questionable related to the mix of tens or hundreds of diffracted beams present in
the region. On the contrary, if a few pixels located at an exact and known reflection,
the resulting image is clearly related to the diffraction data contained in this virtual
aperture. If the EBSPs from an EBSD scan are stored during the acquisition, a
post-processing routine can be used to generate as many images as the number of
pixels in the reference EBSP and select particular reflections to create a DF image
(EBSD-DF image) (Brodusch et al. 2015). A schematic describing the different
steps to achieve EBSD-DF imaging is presented in Fig. 8.3. The contrast obtained
is then directly related to a reference excited beam and makes the quantitative
interpretation of the image possible. To obtain the highest image quality, high
Fig. 8.3 Description of the EBSD-DF process to obtain an EBSD-DF image from an EBSD
dataset
angular EBSP resolution is necessary. For this reason, long detector distance and a
high image pixel resolution are necessary.
An example of EBSD-DF analysis is shown in Fig. 8.4 where a reference EBSP
(Fig. 8.4a), the FSD (Fig. 8.4b), the band contrast (Fig. 8.4c) and EBSD-DF
images (Fig. 8.4d–h) are displayed. The specimen was an indented iron specimen
analyzed by EBSD mapping at E0 = 20 kV. The positions of the 2 2 pixels2
virtual apertures for each image are shown in the reference pattern (Fig. 8.4a). As
can be seen, the contrast of the EBSD-DF images is highly dependent on the
reflection virtually selected. This permits to visualize the deformation structures in
more details, that would have been missed if only one or two images would have
been acquired by ECCI.
As explained above, EBSD-DF is based upon the reciprocity theorem, Fig. 8.5a,
b. However, experimental evidences of reciprocity in the SEM between ECP and
EBSD have never been reported. In a recent paper, Kaboli et al. (2015) used the
EBSD-DF technique to reproduce an ECCI micrograph obtained with the same
reflection at the optic axis of the SEM (center of the ECP). Because EBSD is
performed at high surface tilt angles, in this case +70°, it was deduced from a
simple geometric calculation that the position of the pattern center on the EBSD
camera was equivalent to the beam position in the ECP (optic axis) at a surface tilt
angle of −20° as schematically described in Fig. 8.5c, d [the reader is referred to
Kaboli et al. (2015) for more details].
Therefore, because they were acquired with the same diffraction reflection, the
two images should provide the same contrast if the reciprocity theorem holds. The
resulting ECCI and EBSD-DF micrographs are displayed in Fig. 8.5d, e, respec-
tively. Spatial resolution apart, due a reduced pixel resolution of the EBSD map,
one can notice that the two images contrast are strictly identical and that the bend
(a) Ref EBSP (b) FSD (c) BC (d) Point 1 (e) Point 2
(031)
(310)
[001]
(200)
(f) Point 3 (g) Point 4 (h) Point 5
(020)
(1-
10
)
Fig. 8.4 High contrast EBSD-DF images of a micro-hardness indent on compressed iron obtained
using long EBSD detector distance for high angular resolution EBSPs with an accelerating voltage
of 30 kV and a detector distance of 50 mm as a function of the virtual beam position on the high
angular resolution reference EBSP. a Reference EBSP, b FSD image, c band contrast (BC) image,
and d–h EBSD-DF images with specific reflections marked by arrows in a. The EBSPs’ image
resolution was 1344 1024 pixels. Reproduced under permission from Elsevier (Brodusch et al.
2015)
8.4 Dark-Field Electron Backscatter Diffraction 95
Fig. 8.5 Reciprocity experiment showing identical contrast between the ECCI and EBSD-DF
images. a Schematic for obtaining an ECP and ECCI, b schematic of EBSD pattern collection,
c and d schematics of the reciprocity experiment in the SEM with the ECCI image recorded with
surface tilt of −20° (c), and the EBSD-DF image at +70° tilt (d). e and f are the obtained ECCI and
EBSD-DF images, respectively. Reproduced with permission of the International Union of
Crystallography (http://journals.iucr.org/). Reproduced under permission from John Wiley and
Sons (Kaboli et al. 2015)
contour structures were well reproduced in the EBSD-DF image. This constitutes,
to our knowledge, the first experimental evidence of reciprocity in the SEM
between ECP and EBSD and this allows, now, using EBSD-DF, to interpret and
improve the understanding of the highly diverse ECCI contrast features that are
observed in deformed materials.
8.5 Transmission Forward Electron Backscatter

Diffraction
Although EBSD has revolutionized materials characterization by providing local

orientation and texture information, its spatial resolution is limited by the size of the
BSE emission volume which decreases with E0 and Z−1. Therefore, resolution no
better than 10–20 nm in high Z materials was reported (Randle et al. 2000;
Steinmetz and Zaefferer 2010). Attempts to reduce the beam voltage permitted to
reduce by 1/3rd the spatial resolution on a TWIP steel when reducing the accel-
erating voltage from 15 to 7.5 kV (Randle et al. 2000; Steinmetz and Zaefferer
2010). However, despite the fact that even lower voltages down to 1 kV were
reported (Dorri et al. 2016), the loss of signal due to the increased carbon con-
tamination at low voltage and the large amount of signal absorbed in the coating
layer of the phosphor screen limit greatly the efficiency of the indexing algorithms
(Steinmetz and Zaefferer 2010). Energy filtration applied in front of the EBSD
screen was reported (Deal et al. 2008) and is assumed to improve the spatial
resolution as the emission volume of low-loss is significantly smaller than that of
the non-filtered BSE signal (Merli et al. 2001; Wells 1971, 1974). The elimination
of the low energy BSE signal by thinning the specimen down to a thickness smaller
than the maximum depth of emission of BSEs was another route to improve the
spatial resolution of EBSD (Geiss et al. 2009; Sivel et al. 2005; Small and Michael
2001; Small et al. 2002). However, the improvement in spatial resolution was at the
expense of the SNR of the EBSPs as the BSE signal is dramatically reduced at
small specimen thickness (Reimer 1998) and the collection efficiency of the EBSD
camera is small. On the contrary, complementary to the BSE signal when imaging
thin specimens, the transmitted signal strength is high and carries dynamically
diffracted beams. As in the reflection mode, Kikuchi lines can be observed as the
projection of the crystal lattice planes experienced by the electron beam in the
forward direction. Therefore, any phosphorescent screen combined to a CCD
camera located below the specimen may provide an image of the intersection of the
Kikuchi lines with the detector plane, as in standard EBSD mode. Consequently,
following the original idea of Geiss et al. (2010, 2011), a commercial EBSD camera
normally used in conventional mode was successfully applied at this end and
permitted to acquire Kikuchi patterns from nanoparticles and metallic thin foils
(Keller and Geiss 2012). At this early stage, homemade sample stage dedicated to
this emerging technique had to be fabricated in order to place the specimen in the
optimum position on top of the EBSD detector and several different alternatives
were proposed (Babinsky et al. 2014; Brodusch et al. 2013c; Erdman et al. 2014;
Keller and Geiss 2012; Mortazavi et al. 2015; Rice et al. 2014; Trimby 2012;
Trimby et al. 2014). Inside a focused ion beam microscope (FIB), the foil can be
positioned properly with a micromanipulator after being lift-out and characterized
simply fixed on this device (Bauer et al. 2014). However, if the thin foil is fixed
perpendicular to the holder and the latter is tilted to more than 50–60°, a com-
mercial STEM detector located at the bottom of the specimen chamber can be used
8.5 Transmission Forward Electron Backscatter Diffraction 97
to produce a STEM-BF image permitting to combine transmission imaging with

diffraction (Brodusch et al. 2013b) without resorting to a CTEM or a DSTEM.
A dark-field STEM image can also be obtained simultaneously with the forecaster
detector attached to the EBSD detector (see next section). The main advantage of
using a commercial EBSD system is the availability of the Kikuchi pattern images
post-processing routines that allows orientation mapping (Trimby 2012) and phase
identification (Brodusch et al. 2013b; Keller and Geiss 2012; Robert et al. 2013) as
well as EBSD-DF (Wright et al. 2015).
The origin of the t-EFSD pattern is similar to that involved in EBSD (Zaefferer
2007) and Kikuchi diffraction in the TEM (Fultz and Howe 2013; Williams and
Carter 2009). The primary electron beam penetrates the thin specimen from its top
side and suffers inelastic as well as elastic scattering events, the latter being less
probable due to a large elastic mean free path (MFP). The inelastic collisions,
mostly as plasmon interactions, result in the generation of point sources that dif-
fuses incoherently the primary electrons in small scattering angles. These inco-
herently scattered electrons are then dynamically Bragg diffracted and emitted from
the back surface in a channeling-out fashion. Similarly to the case of EBSD, the
diffracted electrons may be incoherently scattered, mainly through inelastic events
(Winkelmann 2009) due to the different scale of the inelastic and elastic mean free
paths at the accelerating voltage considered here, i.e., 20–30 kV. The probability of
these events dramatically increases with the specimen thickness and because the
inelastic mean free path increases with the accelerating voltage the diffraction signal
is mostly emitted from the bottom surface increasing thus the signal collected by the
EBSD camera below the specimen. This also explains why the transmitted dif-
fracted signal mostly originates from the bottom layers of the foil as demonstrated
by Keller and co-workers (Keller and Geiss 2012; Rice et al. 2014), i.e., from a few
tens of nanometers depending on materials and SEM operating conditions. The
additional background, as in the case of EBSD, results from incoherent and diffuse
elastic scattering and has to be removed by using identical procedures as used for
EBSD (Schwarzer et al. 2009). The optimum thickness, or more precisely
mass-thickness (Rice et al. 2014), should be large enough to have a high number of
point sources but thin enough to limit the probability of incoherent inelastic and
elastic events prior to the diffracted electrons emission. This also drives the mini-
mum thickness for which t-EFSD is applicable. In fact, this thickness is essentially
related to the inelastic mean free path at the entrance of the incoming beam and
prior to the emission of the diffracted electrons (Brodusch et al. 2013b), which is
not well known for all materials except metals. Using a high accelerating voltage
might help in reducing this effect through the increase of the inelastic mean free
path. It has to be noted that, although large mass-thicknesses up to several hundreds
of µgcm−2 still provide useful patterns (Rice et al. 2014), the variation of the
specimen thickness impacts greatly the pattern contrast (Rice et al. 2014; Suzuki
2013; Trimby 2012) rendering the pattern indexing uncertain which should be a
direction of improvement if thick specimens have to be used. However, the loss of
spatial resolution due to the dramatic increase in beam broadening with specimen
thickness (Gauvin 2015) renders the use of t-EFSD unproductive with nearly the
same spatial limits as for EBSD. With proper SEM operating conditions and
optimum thicknesses, orientation mapping was reported with a spatial resolution
better than 5 nm, possibly down to 2 nm (Brodusch et al. 2013b; Trimby 2012) and
thicknesses of 5 nm were reported with high Z materials (Brodusch et al. 2013b;
Rice et al. 2014).
Rice and co-workers suggested that less than 10% of the total number of elec-
trons transmitted through the thin foil contribute to the generation of the t-EFSD
pattern (Rice et al. 2014). They also proposed, based on Monte Carlo simulations,
that only the low-loss electrons with an energy-loss smaller than 10% at the exit
surface contribute to the t-EFSD pattern, which is consistent with the value of a few
percent energy-loss determined by Deal et al. (2008) when applying an energy filter
to an EBSD camera. The range of collection angles involved in t-EFSD and the fact
that only the bottom layers participate to the generation of the pattern strongly
suggests that elastic forward scattering is part of the process, mainly prior to the
inelastic events producing the point sources. Hence, the useful signal for t-EFSD is
a forwarded diffracted signal. It is generally accepted, and largely used in the Monte
Carlo simulation community, that electrons scattered out of the surface with an
angle larger than 90° compared to the incident beam direction are considered as
backscattered and those with angles smaller than 90° are considered as forward
scattered. This, by itself, justifies the use of the acronym t-EFSD for transmission
electron forward diffraction by analogy with EBSD as firstly used by Geiss et al.
(2011). By the way, t-EBSD (Keller and Geiss 2012) and TKD (Trimby 2012), for
transmission EBSD and transmission Kikuchi diffraction, are commonly used in the
related literature.
Since Keller and co-workers reported this new technique, many applications
have been reported with an increasing interest and impact on the understanding of
materials science processes. Orientation mapping is currently the t-EFSD applica-
tion that is reported the most in the literature. The main purpose is the character-
ization of grains and nano-grains structures (Babinsky et al. 2015; Proust et al.
2015; Sha et al. 2014; Sun et al. 2013; Suzuki 2013; Trimby et al. 2014; Zielinski
et al. 2015) and deformation in metals and alloys (Birosca et al. 2015; Meisnar et al.
2015) as well as the characterization of oxides films and their interfaces with
substrates (Garner et al. 2014; Hu et al. 2015). An interesting application was
reported recently based on the use of t-EFSD to choose specific sites for the
preparation of atom probe tomography specimens in the form of sharp needles
(Babinsky et al. 2014). This allows to accelerate the process without requiring
transferring the needle from the FIB to the TEM and so on to characterize the
needle to focus the milling on a specific site. Additionally, this allowed combining
orientation mapping to the atom probe tomography (APT) characterization to
improve the interpretation of the final results (Babinsky et al. 2015; Sha et al. 2014).
Moreover, Bauer et al. (2014) reported that t-EFSD could be applied to thin foils
lifted-out from the specimen surface and attached to a nano-manipulator tip directly
inside a FIB equipped with an EBSD system. This is a great opportunity as it allows
a better control of the foil crystallographic state and thickness during ion beam
8.5 Transmission Forward Electron Backscatter Diffraction 99
thinning. Additionally, this technique will definitely be attractive for moisture and
air sensitive materials as well as brittle objects that would suffer minimum handling
and transfer prior to the analysis.
Phase identification, as it is for EBSD, is an important field of applications. The
identification and surface distribution of nano-objects in multiphase specimens like
glass (Keshavarzi et al. 2015), semi-conductors (Shen et al. 2013), asbestos (Bandli
and Gunter 2014) or metallic alloys (Abbasi et al. 2015; Birosca et al. 2015; Hu
et al. 2015) were reported. Gathering phase information at the nanoscale is
important to understand the chemical, electrochemical, electrical, medical, or
mechanical properties of these materials. Especially in the field of energy storage,
like lithium ion batteries (LIB), the potential impact of t-EFSD is great based on the
few studies reported to date (Robert et al. 2013; Sussman et al. 2014). Managing
impurities and phase transformation during the charge/discharge cycles is manda-
tory to improve the fabrication processes and the electrochemical knowledge of the
nano-materials used. Figure 8.6a shows a typical micrograph of a lithium titanate
powder dispersed on a TEM grid recorded with the upper SE detector without
energy filtration at E0 = 30 kV. The lithium titanate mostly had the form of
nano-flakes of a few tens of nanometers with a thickness of around 5–10 nm.
t-EFSD was applied with the same SEM parameters and the STEM-BF images and
transmission Kikuchi patterns corresponding to the lithium titanate (left) and to
TiO2-rutile (right) are displayed in Fig. 8.6b. Due to their distinct crystallographic
structures, rutile impurities could be identified and their different morphology
pointed out.
(a) (b) Li2 TiO3 TiO2 (rutile)
Fig. 8.6 t-EFSD analysis of a lithium titanate powder at E0 = 30 kV in STEM mode. a SE image
(upper detector without energy filtration) and b BF-STEM images (top), t-EFSD patterns (center)
and indexed patterns of the Li2TiO3 (bottom left) and TiO2-rutile (bottom right) phases
In addition, when an EDS is integrated to the commercial EBSD system used for
t-EFSD, a high degree of accuracy can be achieved in phase identification as
reported by Brodusch et al. (2013b), similarly to what is used in conventional
EBSD (Nowell and Wright 2004). More recently, EBSD-DF imaging has been
combined to t-EFSD to investigate the potential capabilities of the technique on an
AlTi3–TiAl3 alloy (Wright et al. 2015). Although the technique was not fully
explored, one of the interesting point raised by this study was the fact that back-
ground corrected t-EFSD patterns permitted to obtain a high-quality DF image of a
deformed area of the foil with no or little effect of the variation of the mass-
thickness on the contrary to the images obtained with the raw patterns or with the
forecaster detector (FSD). Work is currently in progress to explore more deeply the
combination of these two techniques.
8.6 Dark-Field Imaging with a Forecaster Detector

in Transmission Mode
The recent development of t-EFSD as described in the previous section led to

re-investigate the possible use of the FSD diodes attached to the camera screen in
most EBSD systems. Originally, the FSD was developed to provide orientation and
material contrast through the collection of the forward scattered electrons emerging
from highly tilted flat surfaces (Prior et al. 1996). In a transmission experiment like
t-EFSD, the thin specimen is placed on top of the EBSD camera and the transmitted
electrons are elastically scattered in a wide range of angles which increase as E0 is
reduced. Therefore, large angle transmitted electrons can be collected using the
bottom diodes of the FSD in order to generate a dark-field image (DF-FSD)
(Brodusch et al. 2013a). The collection angle of the detector is selected by changing
the detector distance to the specimen and this allows to collect high-angle as well as
low-angle dark-field images depending if the camera is retracted or brought closer
to the sample [see Fig. 1 and S1 in Brodusch et al. (2013a)]. An example of a
DF-FSD image from carbon nanotubes covered with Pt nanoparticles dispersed on a
standard TEM grid is shown in Fig. 8.7, with a BF image obtained in (a) and a
DF-FSD image in (b) from the same area and same scan.
The signal in the BF image was collected through a cone of 10 mrad semi-angle
and in the DF-FSD image was collected between angles of 426 and 518 mrad. The
image in Fig. 8.7b displays a typical high-angle DF image mostly dominated by
Z-contrast. The diffraction contrast observed in the BF image is no longer visible in
the DF-FSD image. Even though the SNR was one order of magnitude smaller than
in BF, the spatial resolution in the DF-FSD image was measured to be slightly
better. The FSD detector can also be used to record orientation contrast DF images
is the same fashion as in the normal EBSD set-up (Trimby et al. 2014), the flexi-
bility in selecting the collection angle being an advantage.
8.6 Dark-Field Imaging with a Forecaster Detector in Transmission Mode 101
Fig. 8.7 DF imaging with

the bottom semiconductor
diodes of a FSD system
attached to an EBSD camera
at E0 = 30 kV. a BF image
and b DF-FSD micrograph of
carbon nanotubes covered
with Pt nanoparticles.
Reproduced under permission
from Cambridge University
Press (Brodusch et al. 2013a)
This is interesting because most of the t-EFSD work reported do far provides
CTEM or STEM/TEM images of the materials under investigation and the FSD
now permits to carry STEM imaging in the SEM during the t-EFSD analysis
without a dedicated STEM detector allowing more flexibility to select relevant areas
of the thin specimen.
References
Abbasi, M., Kim, D. I., Guim, H. U., Hosseini, M., Danesh-Manesh, H., & Abbasi, M. (2015).
Application of transmitted kikuchi diffraction in studying nano-oxide and ultrafine metallic
grains. ACS Nano, 9, 10991–11002.
Alam, M., Blackman, M., & Pashley, D. (1954). High-angle Kikuchi patterns. Proceedings of the
Royal Society of London. Series A. Mathematical and Physical Sciences, 221, 224–242.
Babinsky, K., De Kloe, R., Clemens, H., & Primig, S. (2014). A novel approach for site-specific
atom probe specimen preparation by focused ion beam and transmission electron backscatter
diffraction. Ultramicroscopy, 144, 9–18.
Babinsky, K., Knabl, W., Lorich, A., De Kloe, R., Clemens, H., & Primig, S. (2015). Grain
boundary study of technically pure molybdenum by combining APT and TKD.
Bandli, B. R., & Gunter, M. E. (2014). Scanning electron microscopy and transmitted electron
backscatter diffraction examination of asbestos standard reference materials, amphibole
particles of differing morphology, and particle phase discrimination from talc ores. Microscopy
and Microanalysis, 20, 1805–1816.
Bauer, F., Sitzman, S., Lang, C., Hartfield, C., & Goulden, J. (2014). Advancing materials
characterization in the FIB-SEM with transmission Kikuchi diffraction. Microscopy and
Birosca, S., Ding, R., Ooi, S., Buckingham, R., Coleman, C., & Dicks, K. (2015). Nanostructure
characterisation of flow-formed Cr–Mo–V steel using transmission Kikuchi diffraction
technique. Ultramicroscopy, 153, 1–8.
Brodusch, N., Demers, H., & Gauvin, R. (2013a). Dark-field imaging of thin specimens with a
forescatter electron detector at low accelerating voltage. Microscopy and Microanalysis, 19,
1688–1697.
Brodusch, N., Demers, H., & Gauvin, R. (2013b). Nanometres-resolution Kikuchi patterns from
materials science specimens with transmission electron forward scatter diffraction in the
scanning electron microscope. Journal of Microscopy, 250, 1–14.
Brodusch, N., Demers, H., & Gauvin, R. (2015). Dark-field imaging based on post-processed
electron backscatter diffraction patterns of bulk crystalline materials in a scanning electron
Brodusch, N., Demers, H., Trudeau, M., & Gauvin, R. (2013c). Acquisition parameters
optimization of a transmission electron forward scatter diffraction system in a cold-field
emission scanning electron microscope for nanomaterials characterization. Scanning, 35,
375–386.
Brodusch, N., & Gauvin, R. (2017). The qualitative f-ratio method applied to electron
channelling-induced x-ray imaging with an annular silicon drift detector in a scanning
electron microscope in the transmission mode. Journal of Microscopy, 267, 288–298.
Chekir, N., Gauvin, R., Brodusch, N., Sixsmith, J. J., & Brochu, M. (2017). Effect of travel speed
and post deposition heat treatments in laser wire deposition of thin Ti-6Al-4V deposits, Part I:
Microstructure characterization. Materials Science & Engineering A (Submitted).
Coates, D. (1967). Kikuchi-like reflection patterns obtained with the scanning electron microscope.
Philosophical Magazine, 16, 1179–1184.
Deal, A., Hooghan, T., & Eades, A. (2008). Energy-filtered electron backscatter diffraction.
Dorri, M., Turgeon, S., Brodusch, N., Cloutier, M., Chevallier, P., Gauvin, R., et al. (2016).
Characterization of amorphous oxide nano-thick layers on 316L stainless steel by electron
channeling contrast imaging and electron backscatter diffraction. Microscopy and
Engler, O., & Randle, V. (2010). Introduction to texture analysis: Macrotexture, microtexture, and
orientation mapping. CRC Press.
Erdman, N., Shibata, M., Nylese, T., & Rampton, T. (2014). Nanoscale crystallographic analysis
in FE-SEM using transmission Kikuchi diffraction. Microscopy and Microanalysis, 20,
864–865.
Fultz, B., & Howe, J. (2013). Transmission electron microscopy and diffractometry of materials.
Springer.
Garner, A., Gholinia, A., Frankel, P., Gass, M., MacLaren, I., & Preuss, M. (2014). The
microstructure and microtexture of zirconium oxide films studied by transmission electron
backscatter diffraction and automated crystal orientation mapping with transmission electron
microscopy. Acta Materialia, 80, 159–171.
Gauvin, R. (2015). A universal equation for computing the beam broadening of incidents electrons
in thin films. Microscopy and Microanalysis (Submitted).
References 103
Geiss, R., Keller, R., & Read, D. (2010). Transmission electron diffraction from nanoparticles,
nanowires and thin films in an SEM with conventional EBSD equipment. Microscopy and
Geiss, R., Keller, R., Sitzman, S., & Rice, P. (2011). New method of transmission electron
diffraction to characterize nanomaterials in the SEM. Microscopy and Microanalysis, 17,
386–387.
Geiss, R. H., Read, D. T., Alers, G. B., & Graham, R. L. (2009). EBSD analysis of narrow
damascene copper lines. In Frontiers of characterization and metrology for nanoelectronics:
2009.
Gutierrez-Urrutia, I., Zaefferer, S., & Raabe, D. (2013). Coupling of electron channeling with
EBSD: Toward the quantitative characterization of deformation structures in the SEM. JOM
Journal of the Minerals Metals and Materials Society, 65, 1229–1236.
Harland, C., Akhter, P., & Venables, J. (1981). Accurate microcrystallography at high spatial
resolution using electron back-scattering patterns in a field emission gun scanning electron
microscope. Journal of Physics E: Scientific Instruments, 14, 175.
Hirsch, P. B., Howie, A., Nicholson, R. B., Pashley, D. W., & Whelan, M. J. (1965). Electron
microscopy of thin crystals. London: Butterworths.
Holt, D. B., Muir, M., Grant, P., & Boswarva, I. (1974). Quantitative scanning electron
microscopy. London, New York, San Francisco: Academic Press.
Hovington, P., Pinard, P. T., Lagacé, M., Rodrigue, L., Gauvin, R., & Trudeau, M. L. (2009).
Towards a more comprehensive microstructural analysis of Zr–2.5 Nb pressure tubing using
image analysis and electron backscattered diffraction (EBSD). Journal of Nuclear Materials,
393, 162–174.
Hu, J., Garner, A., Ni, N., Gholinia, A., Nicholls, R. J., Lozano-Perez, S., et al. (2015). Identifying
suboxide grains at the metal–oxide interface of a corroded Zr–1.0% Nb alloy using (S) TEM,
transmission-EBSD and EELS. Micron, 69, 35–42.
Kaboli, S., Demers, H., Brodusch, N., & Gauvin, R. (2014). Electron channeling contrast
observations in deformed Magnesium alloys. Microscopy and Microanalysis, 20, 1452–1453.
Kaboli, S., Demers, H., Brodusch, N., & Gauvin, R. (2015). Rotation contour contrast
reconstruction using electron backscatter diffraction in a scanning electron microscope.
Journal of Applied Crystallography, 48, 776–785.
Keller, R., & Geiss, R. (2012). Transmission EBSD from 10 nm domains in a scanning electron
microscope. Journal of Microscopy, 245, 245–251.
Keshavarzi, A., Bocker, C., & Rüssel, C. (2015). Nano lamellae composed of yttrium aluminum
garnet and yttrium silicate by surface crystallization of glass. Journal of Materials Science, 50,
848–854.
Kikuchi, S. (1928). Diffraction of cathode rays by mica. Proceedings of the Imperial Academy, 4,
271–274.
Kossel, W., Loeck, V., & Voges, H. (1935). Die Richtungsverteilung der in einem Kristall
entstandenen charakteristischen Rontgenstrahlung. Zeitschrift fur Physik A: Hadrons and
Nuclei, 94, 139–144.
Maher, D. M., & Joy, D. C. (1976). The formation and interpretation of defect images from
crystalline materials in a scanning transmission electron microscope. Ultramicroscopy, 1,
239–253.
Maitland, T., & Sitzman, S. (2006). Electron Backscatter Diffraction (EBSD) techique and material
characterization examples. In: Scanning microscopy for nanotechnology. Springer.
Marthinsen, K., & Hoier, R. (1986). Many-beam effects and phase information in electron
channelling patterns. Acta Crystallographica. Section A, Foundations of Crystallography, 42,
484–492.
Meisnar, M., Vilalta-Clemente, A., Gholinia, A., Moody, M., Wilkinson, A. J., Huin, N., et al.
(2015). Using transmission Kikuchi diffraction to study intergranular stress corrosion cracking
in type 316 stainless steels. Micron, 75, 1–10.
Merli, P., Migliori, A., Morandi, V., & Rosa, R. (2001). Spatial resolution and energy filtering of
backscattered electron images in scanning electron microscopy. Ultramicroscopy, 88,
139–150.
Morandi, V., & Merli, P. G. (2007). Contrast and resolution versus specimen thickness in low
energy scanning transmission electron microscopy. Journal of Applied Physics, 101, 114917.
Morin, P., Pitaval, M., Besnard, D., & Fontaine, G. (1979). Electron–channelling imaging in
scanning electron microscopy. Philosophical Magazine A, 40, 511–524.
Mortazavi, N., Esmaily, M., & Halvarsson, M. (2015). The capability of transmission Kikuchi
diffraction technique for characterizing nano-grained oxide scales formed on a FeCrAl stainless
steel. Materials Letters, 147, 42–45.
Nowell, M., & Wright, S. (2004). Phase differentiation via combined EBSD and XEDS. Journal of
Nowell, M. M., Wright, S. I., Rampton, T., & de Kloe, R. (2014). A new microstructural imaging
approach through EBSD pattern region of interest analysis. Microscopy and Microanalysis, 20,
1116–1117.
Prior, D. J., Trimby, P., Weber, U., & Dingley, D. J. (1996). Orientation contrast imaging of
microstructures in rocks using forescatter detectors in the scanning electron microscope.
Mineralogical Magazine, 60, 859–869.
Proust, G., Retraint, D., Chemkhi, M., Roos, A., & Demangel, C. (2015). electron backscatter
diffraction and transmission Kikuchi diffraction analysis of an austenitic stainless steel
subjected to surface mechanical attrition treatment and plasma nitriding. Microscopy and
Microanalysis: the Official Journal of Microscopy Society of America, Microbeam Analysis
Society, Microscopical Society of Canada, 1–8.
Randle, V., Schwartz, A., Kumar, M., & Adams, B. (2000). Electron backscatter diffraction in
materials science. Kluwer Academic.
(Springer Series in Optical Sciences). Springer.
Rice, K., Keller, R., & Stoykovich, M. (2014). Specimen-thickness effects on transmission
Kikuchi patterns in the scanning electron microscope. Journal of Microscopy, 254, 129–136.
Robert, D., Douillard, T., Boulineau, A., Brunetti, G., Nowakowski, P., Venet, D., et al. (2013).
Multiscale phase mapping of LiFePO4-based electrodes by transmission electron microscopy
and electron forward scattering diffraction. ACS Nano, 7, 10887–10894.
Schulson, E. (1971). Interpretation of the widths of SEM electron channelling lines. Physica Status
Solidi (b), 46, 95–101.
Schwarzer, R. A., Field, D. P., Adams, B. L., Kumar, M., & Schwartz, A. J. (2009). Present state
of electron backscatter diffraction and prospective developments. In: A. J. Schwartz,
M. Kumar, B. L. Adams, & D. P. Field (Eds.), Electron backscatter diffraction in materials
science. Springer.
Sha, G., Tugcu, K., Liao, X., Trimby, P., Murashkin, M., Valiev, R., et al. (2014). Strength, grain
refinement and solute nanostructures of an Al–Mg–Si alloy (AA6060) processed by
high-pressure torsion. Acta Materialia, 63, 169–179.
Shen, Y. Q., Lee, E., Chow, S. Y., Khoo, B. S., Kon, C., Gui, D., et al. (2013). Application of
transmission EBSD in aluminium metal layer and GaAs/AlAs epitaxial layers. In: 2013 20th
IEEE International Symposium on the Physical and Failure Analysis of Integrated Circuits
(IPFA).
Sivel, V., Tichelaar, F., Mohdadi, H., Alkemade, P., & Zandbergen, H. (2005). Crystallographic
analysis of thin specimens. Journal of Microscopy, 218, 115–124.
Small, J., & Michael, J. (2001). Phase identification of individual crystalline particles by electron
backscatter diffraction. Journal of Microscopy, 201, 59–69.
Small, J., Michael, J., & Bright, D. (2002). Improving the quality of electron backscatter
diffraction (EBSD) patterns from nanoparticles. Journal of Microscopy, 206, 170–178.
Spencer, J., Humphreys, C., & Hirsch, P. (1972). A dynamical theory for the contrast of perfect
and imperfect crystals in the scanning electron microscope using backscattered electrons.
Philosophical Magazine, 26, 193–213.
References 105
Steinmetz, D., & Zaefferer, S. (2010). Towards ultrahigh resolution EBSD by low accelerating
voltage. Materials Science and Technology, 26, 640–645.
Sun, J., Trimby, P., Yan, F., Liao, X., Tao, N., & Wang, J. (2013). Grain size effect on deformation
twinning propensity in ultrafine-grained hexagonal close-packed titanium. Scripta Materialia,
69, 428–431.
Sussman, M., Brodusch, N., Gauvin, R., & Demopoulos, G. P. (2014). Transmission electron
forward scattered diffraction and low voltage SEM/STEM characterization of binder-free TiO2
electrodes. Microscopy and Microanalysis, 20, 492–493.
Suzuki, S. (2013). Features of transmission EBSD and its application. JOM Journal of the
Minerals Metals and Materials Society, 65, 1254–1263.
Trimby, P. W. (2012). Orientation mapping of nanostructured materials using transmission
Kikuchi diffraction in the scanning electron microscope. Ultramicroscopy, 120, 16–24.
Trimby, P. W., Cao, Y., Chen, Z., Han, S., Hemker, K. J., Lian, J., et al. (2014). Characterizing
deformed ultrafine-grained and nanocrystalline materials using transmission Kikuchi diffrac-
tion in a scanning electron microscope. Acta materialia, 62, 69–80.
Venables, J., & Harland, C. (1973). Electron back-scattering patterns—A new technique for
obtaining crystallographic information in the scanning electron microscope. Philosophical
Magazine, 27, 1193–1200.
Wells, O. C. (1971). Low-loss image for surface scanning electron microscope. Applied Physics
Letters, 19, 232–235.
Wells, O. C. (1974). Scanning electron microscopy. McGraw-Hill.
Wells, O. C. (1999). Comparison of different models for the generation of electron backscattering
patterns in the scanning electron microscope. Scanning, 21, 368–371.
Williams, D. B., & Carter, C. B. (2009). Transmission electron microscopy: A textbook for
materials science. Springer.
Winkelmann, A. (2009). Dynamical simulation of electron backscatter diffraction patterns. In:
Electron backscatter diffraction in materials science. Springer.
Winkelmann, A., Schroter, B., & Richter, W. (2003). Dynamical simulations of zone axis electron
channelling patterns of cubic silicon carbide. Ultramicroscopy, 98, 1–7.
Winkelmann, A., Trager-Cowan, C., Sweeney, F., Day, A. P., & Parbrook, P. (2007). Many-beam
dynamical simulation of electron backscatter diffraction patterns. Ultramicroscopy, 107,
414–421.
Wright, S. I., Nowell, M. M., de Kloe, R., Camus, P., & Rampton, T. (2015). Electron imaging
with an EBSD detector. Ultramicroscopy, 148, 132–145.
Zaefferer, S. (2007). On the formation mechanisms, spatial resolution and intensity of backscatter
Kikuchi patterns. Ultramicroscopy, 107, 254–266.
Zaefferer, S., & Elhami, N.-N. (2014). Theory and application of electron channelling contrast
imaging under controlled diffraction conditions. Acta Materialia, 75, 20–50.
Zielinski, W., Plocinski, T., & Kurzydlowski, K. (2015). Transmission Kikuchi diffraction and
transmission electron forescatter imaging of electropolished and FIB manufactured TEM
specimens. Materials Characterization, 104, 42–48.
Chapter 9
Magnetic Domain Imaging
9.1 Type-I Contrast
Magnetic domains in ferromagnetic materials generate stray fields above the

specimen surface of ferromagnetics with uniaxial anisotropy, i.e., without closed
domains at the specimen surface. These stray fields can be imaged through type-I
magnetic contrast using a SE detector located inside the specimen chamber
(Banbury and Nixon 1967; Joy and Jakubovics 1968). The SEs emitted from the
specimen surface are deflected by the stray fields above each magnetic domain
leading to variations of the SE yield, d (Reimer 1998). This effect is enhanced at
small takeoff angle and collection solid angle of the SE detector. A quantitative
theoretical description of the phenomenon can be found in (Chim 1994; Joy and
Jakubovics 1969). If the specimen is tilted towards the opposite direction of the SE
detector and if the direction of the magnetic induction, B, is oriented to the SE
detector, the magnetic information carried by the SEs collected by the detector can
be increased significantly (Kammlott 1971). Ultimately, two detectors can be used
to record their signal difference and at the end, improve the magnetic domain
contrast (Yuan et al. 1987). Also, d increases inversely with the accelerating voltage
and this implies that type-I contrast is improved at low voltage as was experi-
mentally observed (Kammlott 1971). In addition, Wells has shown that only
low-energy SEs, typically a few eV, were in fact responsible for the type-I magnetic
contrast (Wells 1974), the optimum value being strongly dependent on the field
strength and position of the SE detector (Newbury et al. 1986). However, the
magnetic field induced by the immersion lens might cancel the deviation from the
specimen stray fields.
Mainly, type-I magnetic contrast was applied to imaging magnetic domains of
cobalt mono- (Joy and Jakubovics 1968, 1969; Szmaja 1994, 1996; Szmaja et al.
1994, 1995) and poly-crystals (Cort and Steeds 1972; Yuan et al. 1987) in bulk
form as well as magnetic tapes recordings (Banbury and Nixon 1967) and heads

108 9 Magnetic Domain Imaging
(Banbury and Nixon 1969). In particular, Szmaja (2000) compared type-I magnetic
contrast and the Bitter colloid method and demonstrated their complementarity in
describing the surface magnetic properties of a cobalt monocrystal. This technique
was applied to Nd2Fe14B single crystals and demonstrated a spatial resolution of
1 µm (Lewis et al. 1998). Recently, Ge and co-workers used type-I contrast to
reveal magnetic domain structures in there study of the magnetic shape memory
effect in Ni–Mn–Ga martensite (Ge et al. 2004).
9.2 Type-II Contrast
As mentioned above, type-I magnetic contrast is detected via SE emission and is

thus a surface analysis. It is known that surface magnetic properties may differ from
those at larger depth, i.e., in the bulk of the specimen. To this end, magnetic
imaging can be carried out using type-II magnetic contrast (Philibert and Tixier
1969) which is based on the BSE signal, an in-depth signal. The BSEs inside the
interaction volume probed by high energy electrons are deflected at each elastic
collision by the Lorentz force F resulting from the magnetic induction
B. Depending on the direction of F, the BSEs are deflected towards the specimen
surface with a higher escape probability or inversely in the opposite direction where
their probability of absorption is increased. In the former case, the BSE yield, η, is
reinforced while in the latter, η is decreased. Hence, the resulting image exhibits
dark/bright contrast related to the direction of B inside the magnetic domains.
Because the deflection takes place deep in the material, both beam energy and
specimen tilt angle influence the contrast, the former controlling the incident
electron path length and elastic mean free path and the latter acting on the BSE
emission probability via the effect of B on the reduction or the increase of the BSE
path length at high tilt angles. In fact, the elastic mean free path length increases
with beam energy which allows the electron to suffer a larger deflection between
two successive collisions thus amplifying the effect of B. The specimen tilt reduces
the total elastic path length and therefore increases the mean BSE emission. In
addition, type-II contrast can be improved by using an energy filter as reported by
Wells (1974, 1976) and the useful signal was measured to be coming from BSE
having lost at least 20%. Monte Carlo simulations where carried out by Newbury
and co-workers and permitted a better understanding of the physics involved in the
contrast formation (Fathers et al. 1974; Newbury et al. 1973): (a) Type-II contrast
varies with E1.4
0 , which means energy higher than 30 keV are highly desirable. In
fact, this led to high energy experiments being the rule in most of the studies
involving type-II magnetic contrast (Yamamoto et al. 1975). As seen from the
calculations of Newbury [Fig. 3 in (Newbury et al. 1973)], the contrast improves
from *0.2% at E0 = 30 kV to *3.0% at E0 = 200 kV. However, the gain of
contrast was at the expense of the spatial resolution with a spatial resolution
approximated to Rm/4 with Rm being the maximum electron range which is of
9.2 Type-II Contrast 109
several tens of µm for transition metals at 200 kV; (b) The optimum tilt angle for
maximum contrast lies between 40 and 70° with a classical Everhart-Thornley
detector [see Fig. 3 in (Fathers et al. 1974)]; (c) The BSEs that carry the magnetic
contrast information are those having lost at most 25% of their initial energy, which
represents a large number of collisions before exit compared to the true LLEs.
However, if tilt angles as large as 80° are used, a high contrast was observed in
specimen current mode as well as when using a forward scatter detector (Fathers
et al. 1974). This permitted to maximize the collection of the high energy BSEs
responsible for the magnetic contrast of type-II, which increase with the tilt angle.
Mainly, type-II contrast has been used to investigate Fe–Si alloys (Fathers et al.
1973; Ikuta and Shimizu 1974). Later on, it was applied and combined with type-I
contrast imaging for the characterization of Ni–Mn–Ga martensite to study the
magnetic shape memory effect of these alloys (Ge et al. 2004, 2005).
In the last twenty years, the development of the EBSD technique was rapidly
followed by the introduction of the FSD for imaging the surface of highly tilted
specimens as required to perform EBSD. Especially when using the bottom
detectors attached to an EBSD camera, orientation contrast images provide a highly
efficient technique to image poly-crystallinity (Prior et al. 1996) and crystal defects
(Trager-Cowan et al. 2007). By chance, the geometrical parameters used to perform
EBSD/FSD imaging are in the range of the parameters necessary to observe type-II
magnetic contrast. Hence, by increasing the accelerating voltage to 30 kV with the
specimen tilted to 70° from the optic axis normal and with the FSD at around 15–
20 mm from the beam impact point at the specimen surface, magnetic domains
were observed in Fe and Fe–Si electrical steels through type-II contrast (Ding et al.
2014; Gallaugher et al. 2014).
An example is presented in Fig. 9.1 where a non-oriented electrical steel
(NOES) was imaged with the set-up described above. The specimen was grinded
and polished using conventional metallographic standards. However, as mentioned
above, type-II contrast with E0 = 30 kV is of the order of 0.2% and any surface
roughness might potentially degrade the contrast. For this reason, flat ion beam
milling was used after mechanical polishing to remove any surface topography. The
Ar+ milling system used was a Hitachi IM3000 with a beam accelerating voltage of
6 kV for 30 min followed by 4 kV for 1 h and 2 kV for 2 h. The angles of inci-
dence of the ion beam were 10°, 10° and 8° respectively, with respect to the
specimen surface.
The two semi-conductor bottom diodes of the FSD were used to acquire the
images presented. Figure 9.1a is the image obtained using the left bottom diode
while Fig. 9.1b was acquired with the right bottom diode. As one can notice,
magnetic domains were visible with various shapes with a strong orientation
contrast as well as a horizontal intensity anisotropy. Because of the opposite
location of the two FSD bottom diodes regarding the phosphorous screen, the two
resulting orientation images demonstrated an appreciable difference in contrast.
Figure 9.1c shows the result of adding the intensity of the two images together. As
a result, the horizontal intensity anisotropy was compensated in the final image but
Bottom diode left Bottom diode right Bottom diode left + right
(a) (b) (c)
(c) – Gaussian blur (c)

(d) (e) (f)
Fig. 9.1 Magnetic domain (MD) imaging with type-II contrast of an electrical steel with a
forecaster detector attached to an EBSD system. a and b MD images obtained with the bottom left
and right diodes of the FSD, respectively. c MD image of the same area obtained by summing
signals from (a) and (b). d Image in (c) post-processed as (c)–f[(c)] where f is a Gaussian blur
function generated via ImageJ (Rasband 2015). The post-processing permits to subtract the
orientation contrast and increase substantially the MD contrast. e and f are examples of MD
observed in the sample with (f) showing high resolution MD imaging with a resolution of 250 nm
it still exhibited a very high orientation contrast in addition to the type-II magnetic
contrast. Electron channeling contrast is of the order of 5–15%, depending on
observation conditions, and thus tend to overwhelm the magnetic contrast which is
of a few tenth of a percent at best as described above (Newbury et al. 1973). To
overcome this issue and reduce the amplitude of channeling contrast, flat fielding
was applied to the sum image. For this purpose, a Gaussian blur was applied using
the ImageJ software (Rasband 2015). The radius of the function was selected to
make any detail of the image disappear and then the resulting blurred image was
subtracted to the original sum image of Fig. 9.1c. The resulting image is displayed
in Fig. 9.1d and the orientation contrast observed in Fig. 9.1c was almost com-
pletely removed from the image, thus dramatically improving type-II magnetic
contrast. Figure 9.1e shows an additional magnetic contrast image post-processed
with the procedure described above where a different domain structure was
observed. A higher resolution image is shown in Fig. 9.1f with a very complex
domain pattern. From this image, the spatial resolution was estimated by point to
point measurement to approximately 250 nm (Gallaugher et al. 2014).
In the past, the Kossel technique was used to study the relation between grain
orientation and magnetic domains structures (Cort and Steeds 1972). However, the
Kossel technique did not permit to map the whole image field of view and suffered
from limitations related to x-rays excitation thresholds limiting its application to
transition metals mostly (Cort and Steeds 1972). More recently, EBSD was used in
combination to magnetic force microscopy (Batista et al. 2014) but using two
9.2 Type-II Contrast 111
separate instruments. As an alternative, TEM can provide both information

(Chapman 1984, 1989) but only from small areas and without, most of the time, the
orientation mapping facility, the sample preparation still being quite a challenge.
As already mentioned above, the position and tilt parameters of the specimen
versus the EBSD/FSD system used here were by chance similar for EBSD analysis
and type-II magnetic contrast imaging. As a result, the relation between the grain
orientation and the magnetic domain structure could be determined with the same
specimen in only one instrument. This permitted to achieve greater accuracy and
efficiency in combining the two characterization techniques compared to the pre-
vious “two steps” attempts. The shape of the domains was closely investigated in
relation to b, defined as the angle between the closest magnetic easy axis, <001>
directions in cubic materials, and the specimen surface. It was shown that in the
NOES used in Fig. 9.1, the neighboring grain orientations were having an influence
on the domain structure inside single grains (Gallaugher et al. 2014).
It has to be noted that recently, type-II magnetic contrast was obtained with
specimen tilt positions near 0° with great image quality (Grüner and Shen 2010;
Ihlefeld et al. 2017). In these studies, a solid-state BSE detector positioned between
the pole piece and the specimen was used with short detector-specimen and
working distances. This implies that a large collection efficiency could be achieved,
improving thus the SNR of the images compared to those obtained with the FSD
diodes. By using the same type of configuration, it was possible to obtain similar
images as those shown in Fig. 9.1 but with 0° tilt angle. Applied to a
PbZr0.2Ti0.8O3/PbZr0.7Ti0.3O3 bilayer film imaged with E0 = 20 kV, this configu-
ration, combined to the reduced BSE emission volume due to the high mean atomic
number of the material, permitted to achieve a spatial resolution of the order of
10 nm which is the best resolution reported to date for type-II magnetic contrast
imaging (Ihlefeld et al. 2017).
9.3 Type-III Contrast
Another technique to record images of magnetic domains in the SEM is the

spin-polarized SEM technique also called type-III magnetic contrast (Koike 2013;
Koike and Hayakawa 1984). It is based on the detection of the spin polarization of
SEs and its quantification by using a dedicated spectrometer attached to the SEM
column. It provides surface information on magnetic structures and stray fields due
to the origin of the SE signal used and its spatial resolution is at the nanometer scale
(Kohashi et al. 2009). This technique is very specialized and a complete description
of its physics is beyond the scope of this chapter. For further information, the reader
is invited to read the cited literatures.
References
Banbury, J. R., & Nixon, W. C. (1967). The direct observation of domain structure and magnetic
fields in the scanning electron microscope. Journal of Scientific Instruments, 44, 889.
Banbury, J. R., & Nixon, W. C. (1969). A high-contrast directional detector for the scanning
electron microscope. Journal of Physics E: Scientific Instruments, 2, 1055.
Batista, L., Rabe, U., & Hirsekorn, S. (2014). Determination of the easy axes of small
ferromagnetic precipitates in a bulk material by combined magnetic force microscopy and
electron backscatter diffraction techniques. Ultramicroscopy, 146, 17–26.
Chapman, J. (1984). The investigation of magnetic domain structures in thin foils by electron
microscopy. Journal of Physics D: Applied Physics, 17, 623.
Chapman, J. (1989). High resolution imaging of magnetic structures in the transmission electron
microscope. Materials Science and Engineering B, 3, 355–358.
Chim, W. K. (1994). An analytical model for scanning electron microscope Type I magnetic
contrast with energy filtering. Review of Scientific Instruments, 65, 374–382.
Cort, D., & Steeds, J. (1972). Some experiments using Kossel lines to study the magnetic domain
structure in poly-crystalline cobalt. Physica Status Solidi (a), 10, 215–222.
Ding, Y. Y., Gallaugher, M., Brodusch, N., Gauvin, R., & Chromik, R. R. (2014). Coating induced
residual stress in nonoriented electrical steel laminations. Journal of Materials Research, 29,
1737–1746.
Fathers, D., Jakubovics, J., Joy, D., Newbury, D., & Yakowitz, H. (1974). A new method of
observing magnetic domains by scanning electron microscopy. II. Experimental confirmation
of the theory of image contrast. Physica Status Solidi (a), 22, 609–619.
Fathers, D. J., Jakubovics, J. P., & Joy, D. C. (1973). Magnetic domain contrast from cubic
materials in the scanning electron microscope. Philosophical Magazine, 27, 765–768.
Gallaugher, M., Brodusch, N., Gauvin, R., & Chromik, R. R. (2014). Magnetic domain structure
and crystallographic orientation of electrical steels revealed by a forescatter detector and
electron backscatter diffraction. Ultramicroscopy, 142, 40–49.
Ge, Y., Heczko, O., Soderberg, O., Hannula, S., & Lindroos, V. (2005). Investigation of magnetic
domains in Ni–Mn–Ga alloys with a scanning electron microscope. Smart Materials and
Structures, 14, S211.
Ge, Y., Heczko, O., Soderberg, O., & Lindroos, V. (2004). Various magnetic domain structures in
a Ni–Mn–Ga martensite exhibiting magnetic shape memory effect. Journal of Applied Physics,
96, 2159–2163.
Grüner, D., & Shen, Z. (2010). Direct scanning electron microscopy imaging of ferroelectric
domains after ion milling. Journal of the American Ceramic Society, 93, 48–50.
Ihlefeld, J. F., Michael, J. R., McKenzie, B. B., Scrymgeour, D. A., Maria, J. P., Paisley, E. A.,
et al. (2017). Domain imaging in ferroelectric thin films via channeling-contrast backscattered
electron microscopy. Journal of Materials Science, 52, 1071–1081.
Ikuta, T., & Shimizu, R. (1974). Magnetic domain contrast from ferromagnetic materials in the
scanning electron microscope. Physica Status Solidi (a), 23, 605–613.
Joy, D. C., & Jakubovics, J. P. (1968). Direct observation of magnetic domains by scanning
electron microscopy. Philosophical Magazine, 17, 61–69.
Joy, D. C., & Jakubovics, J. P. (1969). Scanning electron microscope study of the magnetic
domain structure of cobalt single crystals. Journal of Physics D: Applied Physics, 2, 1367.
Kammlott, G. (1971). Observation of ferromagnetic domains with the scanning electron
Kohashi, T., Konoto, M., & Koike, K. (2009). High-resolution spin-polarized scanning electron
microscopy (spin SEM). Journal of electron microscopy, dfp047.
Koike, K. (2013). Spin-polarized scanning electron microscopy. Microscopy, 62, 177–191.
Koike, K., & Hayakawa, K. (1984). Observation of magnetic domains with spin-polarized
secondary electrons. Applied Physics Letters, 45, 585–586.
References 113
Lewis, L., Wang, J. Y., & Canfield, P. (1998). Magnetic domains of single-crystal Nd2Fe14B
imaged by unmodified scanning electron microscopy. Journal of Applied Physics, 83, 6843–
6845.
Newbury, D., Joy, D., Echlin, P., Fiori, C., & Goldstein, J. (1986). Advanced scanning electron
microscopy and x-ray microanalysis. New York: Plenum Press.
Newbury, D., Yakowitz, H., & Myklebust, R. (1973). Monte Carlo calculations of magnetic
contrast from cubic materials in the scanning electron microscope. Applied Physics Letters, 23,
488–490.
Philibert, J., & Tixier, R. (1969). Effets de contraste cristallin en microscopie électronique à
balayage. Micron, 1, 174–186.
Prior, D. J., Trimby, P., Weber, U., & Dingley, D. J. (1996). Orientation contrast imaging of
microstructures in rocks using forescatter detectors in the scanning electron microscope.
Mineralogical Magazine, 60, 859–869.
Rasband, W. S. (2015) Bethesda, Maryland, USA, 1997–2015. Image J (https://imagej.nih.gov/ij/).
(Springer Series in Optical Sciences). Springer.
Szmaja, W. (1994). SEM investigation of the dependence of magnetic domain structure on the
thickness of cobalt monocrystals. Journal of Magnetism and Magnetic Materials, 130, 138–
146.
Szmaja, W. (1996). The thickness dependence of the magnetic domain structure in cobalt
monocrystals studied by SEM. Journal of Magnetism and Magnetic Materials, 153, 215–223.
Szmaja, W. (2000). Studies of the surface domain structure of cobalt monocrystals by the SEM
type-I magnetic contrast and Bitter colloid method. Journal of Magnetism and Magnetic
Materials, 219, 281–293.
Szmaja, W., Polanski, K., & Dolecki, K. (1994). SEM investigation of the temperature dependence
of magnetic domain structure of cobalt monocrystals. Journal of Magnetism and Magnetic
Materials, 130, 147–154.
Szmaja, W., Polanski, K., & Dolecki, K. (1995). The temperature dependence of magnetic domain
structure in cobalt monocrystals studied by SEM. Journal of Magnetism and Magnetic
Materials, 151, 249–258.
Trager-Cowan, C., Sweeney, F., Trimby, P., Day, A., Gholinia, A., Schmidt, N.-H., et al. (2007).
Electron backscatter diffraction and electron channeling contrast imaging of tilt and
dislocations in nitride thin films. Physical Review B, 75, 085301.
Wells, O. C. (1974). Scanning electron microscopy. McGraw-Hill.
Wells, O. C. (1976). Calculation of type II magnetic contrast in the low-loss image in the scanning
electron microscope. In Use of Monte Carlo calculations in electron probe microanalysis and
scanning electron microscopy: Proceedings of a workshop held at the National Bureau of
Standards, Gaithersburg, Maryland, October 1–3, 1975. US Department of Commerce,
National Bureau of Standards: For sale by the Supt. of Docs., US Govt. Print. Off.
Yamamoto, T., Nishizawa, H., & Tsuno, K. (1975). High voltage scanning electron microscopy
for observing magnetic domains. Journal of Physics D: Applied Physics, 8, L113.
Yuan, J., Senkel, R., & Reimer, L. (1987). Recording of magnetic contrast type I by a two-detector
system. Scanning, 9, 249–256.
Chapter 10
Advanced Specimen Preparation
Specimen preparation cannot be circumvented if one wants to obtain high quality

results. However, due to the intrinsic natures of the incident particles and materials,
the interaction of an electron beam with the matter generates unwanted problems
when analyzing a specimen in a SEM. In this chapter, we concentrate on three of
the main issues encountered daily by the electron microscopist: specimen charging,
carbon contamination, and surface damage resulting from mechanical polishing.
10.1 Surface Preparation
Cross-section imaging is an important mode of observation in the SEM as it pro-

vides rapidly the distribution of the features of interest on a large scale, i.e., from
centimeters down to a few nanometers, which is a great advantage over TEM.
A knowledge of the precipitation mechanism of alloying phases and their distri-
bution in metallic alloys is of primary importance for the design of new competitive
materials as is also the texture and crystallographic information. The geological
concerns about texture and phase distribution also benefits from cross-sectioning.
Generally, cross-section preparation of hard materials involves mechanical
grinding and polishing up to 50 nm colloidal silica or alumina particle size (Vander
Voort 1984), this last step being crucial for diffraction and high resolution BSE
analysis (Vander Voort 2011). However, due to mechanical friction (mainly
cold-work) and heating during this process, a mechanical damaged layer of the
order of a few times the grinding or polishing particle size may form, that is, around
100–200 nm for 50 nm colloidal silica medium. A consequence of modifying the
microstructure at the surface of the specimen, the Kikuchi pattern quality of
medium and soft materials degrades rapidly because of surface deformation due to
the increased surface dislocation density. Therefore, complementary techniques are
necessary in addition to polishing to remove the deformed layers either by chemical
etching or electro-chemical polishing (Mills 1985).
116 10 Advanced Specimen Preparation
Recently, the ion etching technique based on ion sputtering has renewed and
commercial equipment have become available. Focused ion beam is now estab-
lished as a technique of choice for producing thin lamellas for TEM and create
three-dimensional volume analysis. In addition, dedicated dual ion beam milling is
being widely used for final thinning of mechanically pre-thinned TEM discs. In the
field of SEM specimen preparation, broad or equivalently, flat ion beam milling has
been developed to final polish the sample surface with various types of ions, argon
(Ar+) being the most used (Hauffe 1991; Hauffe et al. 2003). In such instruments,
ions with energies from a few hundred to several thousands of electron volts are
used and the ion probe diameter is large, up to several millimeters. The angle of
beam incidence with respect to the specimen surface, hi, varies from glancing
incidence (1°–5°) to normal incidence (90°). Flat milling applied to hard materials
with milling times varying from 2 to 30 min with E0 between 2 and 6 kV and with
incidence angles between 5° and 20° generally allows high quality EBSPs and
ECCI contrast to be obtained once transferred in the SEM. Note however that a
higher cross-section quality has been reported when using a masking plate on top of
the specimen with an incident angle of 90° (Erdman et al. 2011; Ogura et al. 2007;
Woo et al. 2011).
Nevertheless, soft materials have very low hardness and are very ductile. For this
reason, mechanical grinding and polishing generates high deformation and in some
cases, polishing media can remain embedded in the material at the surface [see
Fig. 1 in (Brodusch et al. 2013)]. To circumvent this issue, polishing can be skipped
after the grinding steps and replaced by flat Ar+ milling with a particular sequence
of milling steps where various voltages and angles are applied in the fashion of
slope cutting (Hauffe 1991). For instance, this type of procedure was applied to a
rolled Pb–Ag alloy and the resulting surface is shown in Fig. 10.1a as seen with the
in-chamber SE detector (lower). On this image, two Ag/Pb precipitates are observed
in the Pb matrix and two craters surround each one of them. These are artefacts due
to the sputtering induced by the reflected ions around the precipitates combined to
different erosion rates between the precipitate and the matrix (Czanderna et al.
1998). This artefact is common at small angle of incidence and low accelerating
voltage. At a larger scale, no SiC particle was detected which confirmed that
sufficient material was removed to ensure a damage free surface. In Fig. 10.1b, c are
displayed EBSPs recorded in spot mode in the matrix and in the left precipitate at
E0 = 20 kV with the full resolution of the EBSD camera (1344 1024 pixels2).
These patterns show very high contrast and a high number of Kikuchi bands. High
quality EBSPs at E0 = 5 kV were also recorded and can be consulted in Fig. 5 in
(Brodusch et al. 2013). Because the emitted diffracted BSEs at this voltage originate
from a few nanometers in depth, which was confirmed using Monte Carlo simu-
lations, this demonstrated the low level of amorphization resulting from such Ar+
ion treatment. As a consequence, both EBSD orientation mapping as well as
channeling contrast imaging (ECCI) were being successfully applied to this alloy as
shown in Fig. 3 of the cited reference (Brodusch et al. 2013).
To go further, very soft materials, like pure extruded lithium sheets, are not
suitable neither for grinding nor polishing. In this particular case, the extreme
10.1 Surface Preparation 117
(a) (b) (c)

(c)
(b)
1μm
(d) (e) (f)
Fig. 10.1 Surface preparation with Ar+ flat ion milling of a Pb–Ag alloy (a–c) and a pure lithium
sheet (d–f). EBSPs of the Pb matrix (b) and a precipitate (c) arrowed in (a) at E0 = 20 kV. EBSP
(d), band contrast (e) and IPF maps (f) obtained from a pure lithium sheet at E0 = 30 kV prepared
exclusively using ion milling. Reproduced under permission from Oxford University Press
(Brodusch et al. 2013) and John Wiley and Sons (Brodusch et al. 2015b)
softness combined with the very high surface reactivity to air and moisture of pure
lithium make this material one of the most difficult to prepare for SEM analysis.
Especially, no SEM-based diffraction work has ever been reported in the literature
due to the above-mentioned issues. However, because flat ion milling does not
induce any mechanical stress on the sample surface during milling and operates
completely under high vacuum, it appears to be the best candidate to prepare such a
material. In this case, flat Ar+ milling was applied without any other surface
preparation and a specific procedure was developed in order not to overheat the
specimen (Brodusch et al. 2015b). With a sufficiently high electron beam accel-
erating voltage E0 = 30 kV, it was possible to pass through the thick amorphous
layer generated by ion sputtering. This then permitted to generate the diffraction
signal and collect only the diffracted beams that were not absorbed through this
layer, mainly those from low index planes as shown in Fig. 10.1d. It permitted to
acquire for the first-time orientation maps of several mm2 with a high band contrast
value and a mean angular deviation between the simulated and the experimental
patterns of 0.63°. Note that these data were acquired in the low magnetic field mode
of the microscope, where the BSE trajectories were not affected by the magnetic
field. A band contrast (quality map) (Maitland and Sitzman 2006; Pinard et al.
2009) and the corresponding inverse pole figure map are shown in Fig. 10.1e, f,
respectively. The grain size distribution and the specific texture of the lithium sheet
were determined for the first time using the EBSD technique. However, specimen
heating due to the energy transfer of the Ar+ ions to the target atoms must be
considered. To reduce this effect, a milling system fitted with a cold stage should be
used preferentially. Nevertheless, due to the large milled area it was assumed that
the energy dissipation took place on a large portion of the surface and thus the
specimen heating might have been small for the 15 min milling steps used in the
procedure (Brodusch et al. 2015b).
The surface preparation is also of primary importance before performing ECCI.
For the same reason as for EBSD, the incoming primary electrons could be
de-channeled at the entry of the specimen surface preventing from the BSE yield
differences due to the channeling effect to occur and thus no channeling contrast
will be observed. Similar milling procedures can be applied to prepare the specimen
surface and improve the channeling contrast. In this regard, low energy milling
combined to low incidence angle improve greatly the state of the surface as can be
seen on ECCI images presented in Chap. 8. However, one should be careful and
select optimum milling conditions to reduce the surface topography as exemplified
in Fig. 10.1a. This effect is unavoidable for polycrystalline specimens and alloys
with different phases that provide different sputtering ratios (Czanderna et al. 1998;
Dawson and Petrone 1991; Oechsner 1975).
Another application of advanced surface preparation techniques resides in
preparing polymer composites cross-sections. The preparation technique must not
change the final chemical and mechanical states of the sectionned surface.
Generally, freeze fracture is the technique of choice (Hossain 2012; Michler 2008)
for preparing such materials but due to surface topography and to the multiplicity of
debris left at the surface, this hinders greatly the efficiency of the SEM analysis. In a
recent paper, Brodusch et al. (2015a) developed a specific procedure to prepare
cross-sections of hard polymer nanocomposites. First, the specimen was cut,
grinded and polished following standard metallography procedures and then a
specific ion milling sequence was applied, involving accelerating voltages and
incident angles from high to low values. This ensured to produce a flat specimen
surface but because the reinforcing nanoparticles were mostly below the surface,
their resulting contrast appeared diffuse and did not permit an accurate observation
of their size and distribution. For this reason, a commercially available ozone
cleaner (ZoneSEM, Hitachi High-Technologies Canada) was used to etch-out the
cross-section surface selectively to remove the polymer matrix. In fact, as explained
in more details in the next section, the O radicals reacted with the C bonds at the
surface of organic carbonated materials which were then broken by use of an
ultra-violet (UV) lamp. The resulting CO2 species were evacuated by the vacuum
inside the chamber. The resulting surface, when cleaning times of 0.5 to 2 h were
used, showed the nanoparticles sitting on top of the surface with high contrast and
high spatial resolution. The images displayed in Fig. 10.2a, b were recorded with a
landing voltage of 200 V in deceleration mode from a CNT-reinforced epoxy
nanocomposite polymer prepared with the procedure described above. The top
(a) and upper (b) in-lens detectors were used to record these images and provided
voltage contrast with the top detector (Brintlinger et al. 2002; Kovacs et al. 2007;
Wells 1974) and topographical contrast with the upper detector. The origin of
voltage contrast is beyond the scope of this chapter and a more detailed description
of this effect can be found in the cited literatures. Refer to Chaps. 3 and 4 for a
discussion on the contrast mechanisms between the top and upper detectors. By the
10.1 Surface Preparation 119
Fig. 10.2 High-resolution

images of a cross-section of a
(a)
top
nanocomposite polymer of
epoxy reinforced with carbon
nanotubes and carbon fibers
with the top (a) and upper
(b) in-lens electron detectors
at EL = 200 eV using the
deceleration mode. While a
high contrast was achieved in
(a) between the CNTs and the
polymer matrix, the contrast
in (b) was dominated by
topography. The combination
of the two images permitted to
interpret the distribution and
structure of the CNTs in a 3D
fashion leading to a better
(b)
upper
understanding of the
percolation mechanisms.
Reproduced under permission
from John Wiley and Sons
Brodusch et al. (2015a)
way, the bright voltage contrast of CNTs over the polymer matrix provided by the
top detector was useful to observe the planar network of the CNTs at the surface of
the cross section without any artefact from the specimen preparation. However,
small 30–50 nm spherical isthmus for which the surface etching was not complete
(Brodusch et al. 2015a) were observed and were more apparent in the topographical
image of Fig. 10.2b. This type of image can rapidly provide the distribution and
content of CNTs inside the matrix if a simple image analysis segmentation is
applied (Russ 2011).
10.2 Surface Cleaning
Surface contamination occurs when organic species are cracked and fixed at the
specimen surface due to the interactions of the electron beam (ionization) with the
organic surface contaminants, leaving a layer of amorphous carbon at the surface.
The rate of production of the contamination layer is dependent on the beam current
and the time of exposure. This layer can generate charging and the SE yield
decreases resulting in a visible dark rectangle where the area is scanned by the
beam. The two main organic precursors are: (a) the volatile compounds from the
SEM chamber and (b) those adsorbed at the specimen surface. The former is
nowadays somewhat reduced by using dry rotary pumps to avoid oil
back-streaming towards the SEM chamber (Reimer 1998). This greatly improves
the level and quality of the vacuum in the specimen chamber, which can be ulti-
mately enhanced using a liquid nitrogen cooled cold finger. The contamination
from the specimen is now by far the main contributor to the effect observed during
the scan of a specimen surface area. The affinity of the specimen surface for fixing
organic species is material dependent and is high for metals. The protocol of
preparation greatly impacts the affinity of the specimen to contaminate. To avoid
the remaining of oils and heavy solvents at the specimen surface, cleaning with
several different volatile solvents is necessary, followed by drying ideally combined
with heating. However, even when such precautions are applied, some contami-
nants may still remain, especially in pores or cracks. The surface heating due to the
electron bombardment tends to facilitate the migration of those species towards the
beam impact region and participates to the growing contamination process.
To reduce or eradicate the contamination precursors, a surface treatment is
necessary. Historically, this has been achieved by using plasma cleaning, where a
plasma of 75% Ar and 25% O ions was applied to the specimen surface for a short
period of time (Isabell and Fischione 1998). Recently, the technology of ozone
cleaning was introduced and is now commercially available (Soong et al. 2012). Its
basic principle of operation is as follows: a UV radiation transforms O2 to O
radicals, in a three-step fashion, which react with the C and H radicals at the
specimen surface. The complexes consequently formed are cracked using UV
radiation and the resultant H2O, CO2 and N2 species are removed through the
vacuum flow. Typically, a treatment time of 2–30 min and a pressure of 200–
450 torrs of O2 are used to clean efficiently small sized specimens. However, for
porous specimens, long treatment time may be necessary to get rid of most of the
contaminants.
Figure 10.3a shows a SE image (upper detector) of Al2O3 nano-spheres prepared
by sonication of the pristine powder in ethanol for a few minutes prior to dropping
on an aluminum stub. No cleaning treatment was applied before imaging. Although
the original spheres are round and single, the image is dominated by the contam-
ination layer that hides the surface details and lets us think of possible sintering
between the particles. After a few minutes of ozone cleaning with a O2 pressure of
270 torrs, a large amount of the contaminants was removed and clearly, sintering is
not observed whereas a rough contamination layer is still present at the surface of
the spheres as one can see in Fig. 10.3b. This was due to incomplete surface
cleaning which was not surprising given the observed thickness of the contaminant
layer in Fig. 10.3a and the small cleaning time applied. After an additional ozone
treatment of 15 min, the contamination layer was completely removed and the true
surface of the nano-spheres finally showed up. The full capacity of the
high-resolution microscope can then be used to reveal the surface steps and slip
planes which are characteristic of the spheres’ growth process. This demonstrates
10.2 Surface Cleaning 121
Fig. 10.3 Effect of carbon (a)

contamination on the quality
of the SEM images from
Al2O3 nano-spheres.
a Without cleaning, b with
incomplete ozone cleaning
(few minutes), c with efficient
ozone cleaning (additional
15 min)
(b)
(c)
the crucial importance of specimen preparation to avoid misinterpretation when

high resolution imaging is used. Even with a state of the art FE-SEM it will be
impossible to reach nanoscale resolution if surface contaminants are not removed.
Particularly, CFE-SEMs produce the highest brightness and the current density is
very high. Thus, the power locally available to crack the organics and heat the
surface results in higher rates of carbon contamination layer growth. Therefore, a

pre-observation ozone or plasma treatment is mandatory to obtain the best results
from the SEM.
10.3 Charging Compensation with Ionic Liquid

Treatment
Charging of insulating materials under the electron beam is probably one of the most
insidious and disappointing effect that occurs in the SEM. Its origin lies in the ability
of a material to evacuate the negative charges induced by the electron bombardment.
In fact, in a simplified model which is convenient to describe charge effects,
electron-holes pairs are induced by the primary electrons and the charging behavior
of a material relies on the recombination rate of those pairs. If electrons stay trapped
in trapping sites inside the interaction volume, the specimen becomes negatively
charged whereas if excess electrons are emitted from the specimen as secondary and
Auger electrons, the specimen is positively charged due to the number of holes in
excess (Cazaux 2004). Resulting from this mechanism an electrostatic field
builds-up that in turn repulses electrons and holes, finally ending up rapidly in a
steady charging state. The surface electric field also induces a variation of the true
landing voltage (as in deconvolution mode) and a drift of the primary electrons due
to the repulsive/attractive forces between electrons in relation to the electric field.
Several techniques are available to reduce the charging effects. The more
common one consist of coating the specimen surface with elemental carbon by
thermal evaporation or with noble elements like Pt, Au, or Pd by plasma sputtering
(Echlin 2009). The layer thus produced permits to evacuate most of the negative
charges normally trapped without coating. However, the charge removal is not
complete and charging is still present below the coating layer. For this reason, low
voltage must be used to increase the volume fraction of the coating layer inside the
interaction volume, and hence take full advantage of the coating layer electrical
properties. Charging can also be reduced in ceramics by heating the specimen
in situ to several hundreds of degrees during electron irradiation by using a heating
stage inside the SEM (Wang et al. 2009). In this case, the increase in surface
temperature induces an acceleration of de-trapping of electrons and holes in the
irradiated volume. Charge compensation can also be achieved by recombination of
ions with surface electrons via ionization of a gas introduced in the specimen
chamber or spread locally by means of a needle close to the beam impact point
(Carlton et al. 2004). In some cases, searching by trial/errors or measure the value
of E2 (Brochu et al. 2005; Joy et al. 1998; Echlin 2009) at which the total yield
d + η = 1, d and η being the SE and BSE emission yield, may permit to reduce
considerably the charging effects. However, the technique is not completely effi-
cient (Cazaux 2012) and is not applicable when the sample is made of mixed phases
and pores with different sizes and chemistries.
10.3 Charging Compensation with Ionic Liquid Treatment 123
Recently, ionic liquids (ILs) have received an increasing interest regarding

sample preparation in the SEM (Kuwabata et al. 2006), either in biological
(Dwiranti et al. 2012; Takahashi et al. 2013b) or in materials science (Arimoto et al.
2008; Brodusch et al. 2014b; Imashuku et al. 2012; Takahashi et al. 2013a)
applications. Their low vapor pressure (<5 10−9 Torr) and their high ionic
conductivity make these liquid salts perfect candidates for use in high vacuum
applications (Kuwabata et al. 2010, 2013). Their use for sample preparation can be
divided in two groups related to the properties of the ILs. First, the hydrophilic
property of some ILs can be used as an exchange medium with the water inside of
biological (Dwiranti et al. 2012; Hamano et al. 2014; Ishigaki et al. 2011a, b, c;
Takahashi et al. 2013b) or wet mineral (Takahashi et al. 2011, 2012a, b, 2013a)
specimens. Secondly, the liquid form of ILs, even in high vacuum, allows a highly
efficient penetration through the porosity network when the specimen is immersed
in a diluted solution of IL. This hence allows a better electrical surface conductivity
inside the network along the interaction volume (Brodusch et al. 2014b).
Figure 10.4 illustrates an interesting application of ILs in the materials science
field. One big issue when characterizing specimen edges in a cross-section is the
effect of resin contraction after curing. Even when a conductive filler, either gra-
phite, copper, or nickel particles, is added to the mounting mixture, a gap resulting
from the contraction of the resin is often observed and this promotes greatly the
trapping of negative charges in the gap and hence increases charging effects in the
edge region. In this example, a severe beam drift was observed when a
two-dimension EBSD acquisition was conducted, preventing from achieving high
quality maps. This is illustrated in Fig. 10.4a, b where band contrast (BC) and
inverse pole figure (IPF) maps were recorded from a copper particle splattered by
cold-spray at the surface of a polycrystalline titanium substrate. Although the final
map looks good, the top half witnesses of the large beam drift in the area of the
edge and of the gap between the splat and the resin. After pouring the gap interval
with a diluted solution of 1-butyl-3-methylimidazolium tetrafluoroborate
(BMI-BF4, chemical formula C8H15BF4N2) in ethanol, the surface was cleaned
with Ar+ ion milling with E0 = 2 kV and hi = 8° for a few tens of minutes
with ± 60° swing milling to preserve the specimen edge. The resulting BC and IPF
maps, using the same conditions as for that displayed in Fig. 10.4a, b, are shown in
Fig. 10.4c, d. The IL introduced in the gap is visible on the BC image (Fig. 10.4c)
as a dark layer surrounding the splat as it reduced the reflection of the primary
electrons due to the low mean atomic number of the IL. The drifting was annihi-
lated and several hours of data acquisition were then rendered possible. The
resulting map provided high spatial resolution (50 nm features were mapped) and
high degree of precision and no evidence of drifting was observed.
Similarly, the IL treatment can be applied to minerals and rocks and this pro-
vides a rapid and efficient technique to reduce the specimen charging and allow the
characterization of these highly non-conductive materials in a high vacuum SEM
(Brodusch et al. 2014b; Imashuku et al. 2012). In addition to these advantages, the
IL can be quickly removed by solvent dilution and this is another advantage over
coatings which cannot be removed from the specimen surface without mechanical
Before IL After IL treatment

(a) (c)
(b) (d)
(e) (f)
Fig. 10.4 EBSD band contrast (a, c) and IPF (b, d) maps of two individual Cu particles splattered
on a titanium substrate by cold spray without (a, b) and with (c, d) ionic liquid (IL) treatment and
band contrast maps obtained from a zircon/fergusonite-Y mixed region of a mining ore without
(e) and with (f) IL treatment. This demonstrates the high capability of the IL treatment in reducing
drastically the drifting during a long EBSD scan. Reproduced under permission from John Wiley
and Sons (e, f) Brodusch et al. (2014b)
or chemical treatments. An example extracted from (Brodusch et al. 2014b) is

shown in Fig. 10.4e, f where a mining ore containing, among others, zircon, and
fergusonite-Y crystallites was analyzed by EBSD (only the BC maps are shown
here) without and with the IL treatment, respectively. Without the IL treatment, the
strong electric field due to the charging effect resulted in a large drift and did not
allow to record a good EBSD map (Fig. 10.4e). Again, the impact of the IL
treatment was obvious where no charging/drifting was noticed in Fig. 10.4f, thus
permitting to reveal very small crystallites and to describe the intimate aggregation
state of the zircon and fergusonite-Y particles within these clusters.
Similarly, an identical procedure permitted to enhance electrical conductivity of
nanoparticles in powders (Brodusch et al. 2014a; Torimoto et al. 2006). The powder
was simply dispersed in a solution of BMI-BF4 in ethanol (dilution varies from
100 to 10 µL BMI-BF4/ml ethanol depending on specimen), dropped on a SEM
stub, and left for drying in air. Figure 10.5a shows the best high-resolution image
obtained from a Li2FeSiO4 Li-battery cathode material with E0 = 3 kV without
charge compensation. The image contrast reveals signs of charging through
10.3 Charging Compensation with Ionic Liquid Treatment 125
Fig. 10.5 a Best obtainable (a)

high-resolution image of a
Li2FeSiO4 Li-battery cathode
material obtained at
E0 = 3 kV without charge
compensation. Although the
resolution is acceptable, the
SEM parameters were not
optimum to provide the
highest image quality. b Same
specimen after IL treatment,
upper detector with filtered
topographical contrast with
E0 = 2.2 kV and detector
filter bias = 9 V. c Same as
(b) but with the top detector
with maximum energy loss
(b)
DE = 20%
(c)
uncompensated astigmatism and the spatial resolution was not optimum due to a
slight drifting during the 20 s acquisition scan although the voltage condition was
optimum to reduce charging, i.e., close to the E2 value. Note that this accelerating
voltage was not optimized for the contrast.
The same specimen, after the IL treatment, was imaged with the upper and top
in-lens detectors as shown in Fig. 10.5b, c. Contrary to the image in Fig. 10.5a, the
contrast was optimized so that topographical (Fig. 10.5b) and material (Fig. 10.5c)
contrasts were the highest, without worrying about the charging effects that were
drastically reduced. In this case, E0 = 2.2 kV was used and the full capability of
energy filtration provided by the instrument was explored to achieve the highest
contrast and spatial resolution. More details on the image contrast and energy fil-
tration was presented in Chap. 3. From our experience, the ability of IL to help charge
compensation seems related to the surface reactivity of the nanoparticles. In fact, we
noticed that the IL treatment was not efficient for all types of nanoparticle materials.
However, the high number of ILs available should be tested to find the best candidates
for each specific material. It is also interesting to note that, in earlier nanoparticle
production experiments with IL, the IL was found to provide a significant extension in
the storage life and stability in time of the produced materials (Tsuda et al. 2009).
References
Arimoto, S., Kageyama, H., Torimoto, T., & Kuwabata, S. (2008). Development of in situ
scanning electron microscope system for real time observation of metal deposition from ionic
liquid. Electrochemistry Communications, 10, 1901–1904.
Brintlinger, T., Chen, Y.-F., Durkop, T., Cobas, E., Fuhrer, M., Barry, J. D., et al. (2002). Rapid
imaging of nanotubes on insulating substrates. Applied Physics Letters, 81, 2454–2456.
Brochu, M., Demers, H., Gauvin, R., Pugh, M., & Drew, R. (2005). Determination of E2 for
nitride ceramics using FE-SEM and the duane-hunt limit procedure. Microscopy and
Brodusch, N., Boisvert, S., & Gauvin, R. (2013). Flat ion milling: A powerful tool for preparation
of cross-sections of lead-silver alloys. Microscopy, 62, 411–418.
Brodusch, N., Demers, H., & Gauvin, R. (2014a). Ionic liquid used for charge compensation for
high-resolution imaging and analysis in the FE-SEM. Microscopy and Microanalysis, 20, 38–
39.
Brodusch, N., Waters, K., Demers, H., & Gauvin, R. (2014b). Ionic liquid-based observation
technique for nonconductive materials in the scanning electron microscope: Application to the
characterization of a rare earth ore. Microscopy Research and Technique, 77, 225–235.
Brodusch, N., Yourdkhani, M., Hubert, P., & Gauvin, R. (2015a). Efficient cross-section
preparation method for high resolution imaging of hard polymer composites with a scanning
electron microscope. Journal of Microscopy, 260, 117–124.
Brodusch, N., Zaghib, K., & Gauvin, R. (2015b). Electron backscatter diffraction applied to
lithium sheets prepared by broad ion beam milling. Microscopy Research and Technique, 78,
30–39.
Carlton, R. A., Lyman, C. E., & Roberts, J. E. (2004). Charge neutralization in the ESEM for
quantitative X-ray microanalysis. Microscopy and Microanalysis, 10(6):753–763.
Cazaux, J. (2004). About the mechanisms of charging in EPMA, SEM, and ESEM with their time
evolution. Microscopy and Microanalysis, 10(6):670–684.
Cazaux, J. (2012). From the physics of secondary electron emission to image contrasts in scanning
electron microscopy. Journal of Electron Microscopy, 61(5):261–284. doi:10.1093/jmicro/
dfs048.
Czanderna, A. W., Madey, T. E., & Powell, C. J. (1998). Beam effects, surface topography, and
depth profiling in surface analysis. Berlin: Springer.
References 127
Dawson, P., & Petrone, S. (1991). Crystal orientation effects on the surface morphology produced
by ion bombardment of a pure element: Implications for quantitative surface analysis. Surface
and Interface Analysis, 17, 273–281.
Dwiranti, A., Lin, L., Mochizuki, E., Kuwabata, S., Takaoka, A., Uchiyama, S., et al. (2012).
Chromosome observation by scanning electron microscopy using ionic liquid. Microscopy
Research and Technique, 75, 1113–1118.
Echlin, P. (2009). Handbook of sample preparation for scanning electron microscopy and X-ray
microanalysis. Berlin: Springer.
Erdman, N., Ogura, K., & Campbell, R. (2011). Advanced sample preparation techniques using
broad ar ion beam for optimum EBSD acquisition. Microscopy and Microanalysis, 17, 388–389.
Hamano, T., Dwiranti, A., Kaneyoshi, K., Fukuda, S., Kometani, R., Nakao, M., et al. (2014).
Chromosome interior observation by Focused Ion Beam/Scanning Electron Microscopy
(FIB/SEM) using ionic liquid technique. Microscopy and Microanalysis: The Official Journal
of Microscopy Society of America, Microbeam Analysis Society, Microscopical Society of
Canada, 20, 1340–1347.
Hauffe, W. (1991). Production of microstructures by ion beam sputtering. In: Sputtering by
particle bombardment III. Berlin: Springer.
Hauffe, W., Menzel, S., & Göbel, T. (2003). Advantages of Broad Ion Beam (BIB) processing
compared with Focused Ion Beam (FIB) technology for 3D investigation of heterogeneous
solids. Microscopy and Microanalysis, 9, 148–149.
Hossain, M. K. (2012). Scanning electron microscopy. In V. Kazmiruk (Ed.), InTech, Croatia.
Imashuku, S., Kawakami, T., Ze, L., & Kawai, J. (2012). Possibility of scanning electron
microscope observation and energy dispersive X-ray analysis in microscale region of insulating
samples using diluted ionic liquid. Microscopy and Microanalysis, 18, 365–370.
Isabell, T., & Fischione, P. (1998). Applications of plasma cleaning for electron microscopy of
semiconducting materials. In: MRS Proceedings.
Ishigaki, Y., Nakamura, Y., Takehara, T., Kurihara, T., Koga, H., Takegami, T., et al. (2011a).
Comparative study of hydrophilic and hydrophobic ionic liquids for observing cultured human
cells by scanning electron microscopy. Microscopy Research and Technique, 74, 1104–1108.
Ishigaki, Y., Nakamura, Y., Takehara, T., Nemoto, N., Kurihara, T., Koga, H., Nakagawa, H.,
Takegami, T., Tomosugi, N., Miyazawa, S., & et al. (2011b). Ionic liquid enables simple and
rapid sample preparation of human culturing cells for scanning electron microscope analysis.
Microscopy Research and Technique, 74, 415–420.
Ishigaki, Y., Nakamura, Y., Takehara, T., Shimasaki, T., Tatsuno, T., Takano, F., Ueda, Y.,
Motoo, Y., Takegami, T., Nakagawa, H., & others (2011c). Scanning electron microscopy with
an ionic liquid reveals the loss of mitotic protrusions of cells during the epithelial–
mesenchymal transition. Microscopy Research and Technique, 74, 1024–1031.
Joy, D. C., Joy, C. S., et al. (1998). Study of the dependence of E2 energies on sample chemistry.
Kovacs, J. Z., Andresen, K., Pauls, J. R., Garcia, C. P., Schossig, M., Schulte, K., et al. (2007).
Analyzing the quality of carbon nanotube dispersions in polymers using scanning electron
microscopy. Carbon, 45, 1279–1288.
Kuwabata, S., Kongkanand, A., Oyamatsu, D., & Torimoto, T. (2006). Observation of ionic liquid
by scanning electron microscope. Chemistry Letters, 35, 600.
Kuwabata, S., Torimoto, T., Imanishi, A., & Tsuda, T. (2013). Use of Ionic liquid under vacuum
conditions. In: Kadokawa J-i (ed) Ionic Liquids—New Aspects for the Future. InTech, Rijeka,
Croatia, pp 597–615. doi:http://dx.doi.org/10.5772/52597.
Kuwabata, S., Tsuda, T., & Torimoto, T. (2010). Room-temperature ionic liquid. A new medium
for material production and analyses under vacuum conditions. The Journal of Physical
Chemistry Letters, 1, 3177–3188.
Maitland, T., & Sitzman, S. (2006). Scanning microscopy for nanotechnology, Chap. 2: Electron
Backscatter Diffraction (EBSD) techique and material characterization examples. In: Zhou W,
Wang, Zhong Lin (ed) Scanning Microscopy for Nanotechnology. Springer-Verlag, New York,
pp 41–75. doi:10.1007/978-0-387-39620-0.
Michler, G. H. (2008). Electron microscopy of polymers. Berlin: Springer.

Mills, K. (1985). ASM metal handbook Vol. 9: Metallography and microstructures. Russell
Township: ASM International.
Oechsner, H. (1975). Sputtering—a review of some recent experimental and theoretical aspects.
Applied Physics, 8, 185–198.
Ogura, K., Kamidaira, M., Asahina, S., & Erdman, N. (2007). New methods for cross-section
sample preparation using broad argon ion beam. Microscopy and Microanalysis, 13, 1518.
Pinard, P., Hovington, P., Lagacé, M., Lucas, G., Voort, G. V., & Gauvin, R. (2009). Quantitative
evaluation of metallographic preparation quality using EBSD. Microscopy and Microanalysis,
15, 778.
(springer series in optical sciences). Berlin: Springer.
Russ, J. C. (2011). The image processing handbook. Boca Raton: CRC Press.
Soong, C., Woo, P., & Hoyle, D. (2012). Contamination cleaning of TEM/SEM samples with the
ZONE cleaner. Microscopy Today, 20, 44–48.
Takahashi, C., Pattanayak, D. K., Shirai, T., & Fuji, M. (2013a). Application of hydrophilic ionic
liquid treatment to the morphological observations of hydrated porous ceramic green bodies.
Ceramics International, 39, 1065–1073.
Takahashi, C., Shirai, T., & Fuji, M (2013). Application of Ionic liquids on microscopic
observation of hydrated materials. Annual report Advanced Ceramics Research Center Nagoya
Institute of Technology 1:45–52.
Takahashi, C., Shirai, T., & Fuji, M. (2011). Electron microscopic observation of fine morphology
of wet agar gel using a typical hydrophilic ionic liquid; 1-butyl-3-methylimidazolium
tetrafluoroborate. Advances in Technology of Materials and Materials Processing, 13, 393–
475.
Takahashi, C., Shirai, T., & Fuji, M. (2012a). FE-SEM observation, and mechanism of interaction
of wet agar gel in various swelling conditions using hydrophilic ionic liquid. Materials
Chemistry and Physics, 136, 816–822.
Takahashi, C., Shirai, T., & Fuji, M. (2012b). Observation of interactions between hydrophilic
ionic liquid and water on wet agar gels by FE-SEM and its mechanism. Materials Chemistry
and Physics, 133:565–572.
Takahashi, C., Shirai, T., & Fuji, M. (2013b). FE-SEM observation of swelled seaweed using
hydrophilic ionic liquid; 1-butyl-3-methylimidazolium tetrafluoroborate. Microscopy Research
and Technique, 76, 66–71.
Torimoto, T., Okazaki, K.-I., Kiyama, T., Hirahara, K., Tanaka, N., & Kuwabata, S. (2006).
Sputter deposition onto ionic liquids: Simple and clean synthesis of highly dispersed ultrafine
metal nanoparticles. Applied Physics Letters, 89, 243117.
Tsuda, T., Seino, S., & Kuwabata, S. (2009). Gold nanoparticles prepared with a
room-temperature ionic liquid–radiation irradiation method. Chemical Communications,
6792–6794.
Vander Voort, G. F. (1984). Metallography, principles and practice. Russell Township: ASM
International.
Vander Voort, G. F. (2011). Metallographic Specimen preparation for electron backscattered
diffraction. Praktische Metallographie, 48, 527–543.
Wang, L., Ji, Y., Wei, B., Zhang, Y., Fu, J., Xu, X., et al. (2009). Charge compensation by in- situ
heating for insulating ceramics in scanning electron microscope. Ultramicroscopy, 109, 1326–
1332.
Wells, O. C. (1974). Scanning electron microscopy. New York: McGraw-Hill.
Woo, P., Atsushi, M., & Clarke, J. (2011). Broad beam ion milling applications on a diverse range
of materials for scanning electron microscopy. In M. S. of Canada (Ed.), Abstract book,
Microscopical Society of Canada—MSC-SMC 2011, Ottawa.
Chapter 11
Conclusion and Perspectives
Through the course of this book, the routine capabilities of state-of-the-art

FE-SEMs to provide nanometer resolution was the central theme. Especially, the
focus was directed towards the cold-field emission technology that faced a revo-
lution with the introduction of the auto-flash technology. This technology removes
the drawbacks of the previous CFE technology and allows a higher brightness and
large current density at all time, a small energy spread, and a low beam instabilities
due to an improved gun chamber vacuum and stable beam current. All these
parameters have an impact on the image quality. The high current density and
brightness allow reducing significantly the beam current necessary to obtained SE,
BSE, STEM, and x-ray images while keeping the probe current and beam size
reasonably low, which helps reducing the specimen charging at the same time.
The main specimen preparation requirements and characteristics achievable in
routine characterization with such microscopes can be summarized as follows.
Specimen preparation is a prerequisite to reach the highest image and analysis
quality at the nanoscale. Great care must be taken to remove contaminants from the
sample surface and a specialized cleaning apparatus, plasma or ozone based, is
highly recommended for such cleaning procedures. As metal coating is not relevant
in high resolution electron microscopy where the spatial resolution is high enough
to image the metal coating, new routes need to be evaluated. In this regard, ionic
liquids were identified as promising to provide sufficient surface conductivity
without adding additional surface features and thus reduce specimen charging while
keeping the sample surface unchanged. In addition to provide a damage free surface
finish to as-mechanically polished cross-sections of medium-hard materials, ion
milling can be used as a self-sufficient technique to prepare cross-sections of soft
materials where conventional mechanical techniques failed.
The deceleration technique, where a negative voltage is applied to the specimen
surface, is now well established and allows using landing voltages as low as 10 V.
This opens a new route for scanning electron microscopy, but the interpretation of
SEM images at these ultra-low voltages is not clear and needs to be investigated.
Therefore, down to 30–50 V, the spatial resolution is kept close to the nanometer
130 11 Conclusion and Perspectives
and the images obtained at low voltages provide unprecedented sharpness and
realism. Energy filtration of the signals collected by the in-lens SE/BSE detectors
furnishes, as demonstrated, and forecasted by Oliver C. Wells forty years ago,
improved and new contrasts which promise to be a powerful tool to characterize
nanomaterials in the coming years. Experimental evidences now confirm the results
obtained by Monte Carlo modeling long time ago.
Sub-nanometer STEM BF/DF imaging in the SEM is now well established and
provides impressive images of nano-particles and precipitates in alloys at low
primary beam voltages, i.e., from 5 to 30 kV. The contrast is greatly improved
compared to what is commonly achieved with a TEM/STEM at higher voltage
while the spatial resolution continues to improve to the sub-nanometer level.
Especially, due to the increase of scattering cross-sections with low accelerating
voltages, STEM in the SEM provides outstanding contrasts and fills the gap
between conventional bulk imaging and TEM high resolution imaging. Note that
even lower voltages can offer higher contrasts for low Z materials and very thin
specimens.
The f-ratio method is powerful to reduce the effects of beam instabilities when
performing x-ray microanalysis in CFE-SEM. The generalized f-ratio quantitative
x-ray microanalysis method allows high accuracy quantification of multi-element
sample with cold field emission scanning electron microscope where the probe
current is not stable. Obviously, the method is also accurate in a conventional
SEM-EDS instrument.
The large solid angle of the annular SDD allows an acquisition time of the same
order as imaging acquisition time, which allows the acquisition of elemental and
quantitative micrograph of the specimen with simultaneous electron imaging.
Furthermore, using longer acquisition time, the a-SDD is used in situation with low
emission rate like low accelerating voltage or low voltage STEM with x-ray spatial
resolution less than 10 nm. The f-ratio maps are used to cancel the effects of
specimen thickness, defect density, and channeling.
Crystal defects imaging via electron channeling contrast and BF-STEM imaging
provides similar information as obtained in TEM/STEM at high voltage. Single
defects can be easily observed and orientation can be determined in reflection mode
as well as in transmission mode with the same EBSD equipment. The new trans-
mission Kikuchi diffraction technique provides a simple and straightforward
method to obtain crystallographic information and orientation from localized areas
as small as 2–5 nm. Phase identification is available using the specific functions
available in EBSD software. The post-processing of electron backscattered
diffraction patterns offers a brand-new route for characterizing deformation of
materials and interpret the contrasts observed in ECCI. Its application to further
materials science problematics is assumed, in the future, to provide useful results
for the engineer.
The recent renewal of magnetic domain imaging of type-II combined to EBSD
orientation mapping has permitted to develop a new method to characterize the
intimate relationship between the magnetic domain structure and the local grain
orientation as well the strong influence of the surrounding grains on this structure.
11 Conclusion and Perspectives 131
To this day, the last generation FE-SEMs provide better resolutions and contrasts
but still suffers from limitations, although they tend to be the most versatile imaging
instruments. For example, the maximum accelerating voltage available is generally
25–30 kV but this is limiting for high voltage applications such as magnetic domain
or STEM imaging, the latter suffering from a large electron diffusion when the
specimen thickness and atomic number is high. The combination of detection
signals should be improved although some improvements have been made recently.
However, further design work should be engaged to make all optimum distances
converge towards that providing high resolution and high contrasts which is
especially true for the integration of x-rays and Kikuchi diffraction, by reflection or
transmission, in the high-resolution scheme.
The next generation of CFE-SEMs is now being commercially available and
these instruments send a positive response to some of the drawbacks of the SEMs
described in this book. They combine the auto-flash facility with an in-lens spec-
imen chamber and provide the highest spatial resolution reported. Moreover, the
addition of new detectors for STEM mode allows to obtain crystallographic and
chemical information from a thin specimen at very high resolution without suffering
from any loss of spatial resolution due to a change of working distance or probe
current. A diffraction camera located below the specimen permits to record spot or
convergent beam patterns while electron energy-loss spectrometers are now
available for low voltage spectroscopy. These attachments, combined to BF, DF,
BSE, and SE imaging facilities at optimum working distance, will definitely change
the game. Finally, the opportunity to apply these techniques at low accelerating
voltage in transmission mode in a SEM allows reducing considerably the beam
damage by knock-on compared to a high voltage TEM/STEM. However, because
precisely low voltage is used in this kind of microscope, the response of the
analytical system needs to be accurately assessed in comparison to what is already
known from the operation at higher voltage in order to master its application to all
fields of electron microscopy.
Index
A BSE detector, 13, 25, 27, 28, 48, 87, 91, 111,
Aberration correctors, 47 130
Acquisition time, 56, 67, 70, 74–78, 82, 130 BSE imaging, 27, 37, 43, 49, 74, 85, 90
Alumina, 8, 39–41, 115 BSE range, 10
Amorphization, 116 Bulk imaging, 1, 47, 88, 130
Amorphous carbon, 38, 119
Analytical transmission electron microscope, C
55 Calibration curve, 57, 59, 61–63
Angle of collection, 2 Calibration factor, 55, 57, 59, 61–63
Anisotropy, 92, 107, 109 Cancerous cells, 42, 43
Annular silicon drift detector, 67 Carbon contamination, 67, 96, 115, 121, 122
Anomalous absorption, 27 Carbon NanoTube (CNT), 7, 24, 30, 33, 39,
Atomic number, 10, 14, 48, 50, 56, 92, 111, 40, 42–44, 50–52, 76–78, 100, 101,
123, 131 118, 119
Atomic resolution, 11 Cathode electrode, 38
Atom Probe Tomography (APT), 98 Ceramics, 10
Auger electron microscopy, 45 Channeling effect, 82, 118
Auto-flash, 5, 6, 8, 92, 129, 131 Channeling-in, 92
Channeling-out, 97
B Characteristic x-ray, 55
BackScattered Electron (BSE), 2, 26, 28, 48, Charge compensation, 122, 124–126
71, 91 Charging effect, 31, 124
Band Contrast (BC), 92, 94, 117, 123, 124 Chemical fixation, 42, 43
Beam axis, 13, 15 Chromatic aberration, 5, 7, 8, 11, 15, 17, 37, 47
Beam broadening, 9, 11, 47, 91, 97 Cliff and Lorimer, 55, 56
Beam current, 5, 7, 30, 55, 119, 129 Cold-Field Emission (CFE), 5, 55, 129
Beam current stability, 6, 129 Cold-Field Emission Scanning Electron
Beam damage, 47, 131 Microscope (CFE-SEM), 2, 55
Beam diameter, 5, 6 Cold-spray, 123, 124
Beam drift, 67, 123 Collection angle, 13, 14, 19, 24, 25, 50, 90, 98,
Beam instability, 5, 129, 130 100
Bend contour, 89, 95 Collection efficiency, 2, 13, 14, 18, 19, 82, 96,
Bias, 13, 15, 20–22, 24–26, 28, 30, 32, 33, 38, 111
42, 50, 51, 125 Collection rate, 67
Bloch wave, 86, 87, 89, 92 Collector grids, 28
Bragg angle, 49 Condenser lens, 15
Bragg position, 87, 89 Contaminants, 5, 120, 121, 129
Bright-field contrast, 49 Contamination, 21, 37, 38, 44, 67, 70, 119, 120
Brightness, 5, 6, 8, 40, 87, 92, 121, 129 Convergence angle, 86, 87, 90

in Applied Sciences and Technology, https://doi.org/10.1007/978-981-10-4433-5
134 Index
Convergence semi-angle, 87, 91 Elastic forward scattering, 89, 98

Convergent Beam Electron Diffraction Elastic scattering, 49, 70, 78, 91, 97
(CBED), 2 Electrical conductivity, 124
Converting plate, 13, 44 Electrical properties, 44, 45, 122
Core excitation, 78 Electric field, 18, 41, 122, 124
Correction method, 55 Electrode, 15, 19, 28, 38
Correction model, 72, 73, 82 Electro-magnetic fields, 2, 14
Cross-section, 25, 47, 49, 115, 116, 118, 119, Electron acceleration, 41
123, 129, 130 Electron Backscatter Diffraction (EBSD), 2, 85,
Current density, 121, 129 91
Current stability, 7 Electron Backscatter Diffraction Pattern
Cut-off energy, 28–31 (EBSP), 91–94, 96, 116, 117
Electron Channeling Contrast (ECC), 27, 110,
D 130
Damage, 10, 67, 115, 116, 129 Electron Channeling Contrast Imaging (ECCI),
Deceleration mode, 8, 17–22, 24, 37–40, 44, 27, 85, 88
118, 119 Electron Channeling Pattern (ECP), 27, 85
Deceleration voltage, 8, 18, 20, 43 Electron density, 10
Dedicated Scanning Transmission Electron Electron detector, 13, 14, 25, 50, 67, 119
Microscope (DSTEM), 48, 49, 97 Electron Energy Loss Spectroscopy (EELS), 2
Defect density, 81, 82, 130 Electron inelastic mean free path, 44
Defects, 41, 51, 89–91, 109, 130 Electron interaction, 51
Deformation, 50, 86–89, 94, 98, 115, 116, 130 Electron mean free path, 81, 82
Dehydration, 42 Electron range, 9, 10, 108
Delocalization, 47 Electrostatic field, 15, 16, 31, 122
Depth of emission, 9, 28, 29, 96 Emission current, 6
Depth resolution, 13, 86 Emission depth, 26, 27, 29, 37, 50, 70
Detection area, 67, 70 Emission loss, 6, 7
Detection limit, 5 Emission thickness, 49
Detector grid, 13 Emitted electrons, 2, 18, 44
Deviation parameter, 87, 88 Energy dispersion, 15
Diffracted beam, 86, 90, 93, 96, 117 Energy Dispersive Spectrometry (EDS), 2, 67,
Diffraction aberration, 15, 17, 37, 47 71, 73
Diffuse background, 92 Energy distribution, 28, 29, 91, 92
Diffuse elastic scattering, 97 Energy filter, 28, 98, 108
Diffusion, 9, 26, 27, 37, 38, 47, 48, 57, 61–63, Energy filtering, 2, 27
86, 91, 131 Energy filtration, 18–22, 26–28, 31, 33, 42, 43,
Diffusion couple, 61–63 50, 52, 96, 99, 126, 130
Diffusion volume, 9, 37, 47, 48, 86 Energy loss, 2, 10, 24, 27–29, 45, 50, 125
Direct beam, 2 Energy resolution, 48, 67
Dislocation contrast, 87, 89 Energy spread, 7, 129
Dislocation image, 87–89 Energy threshold, 92
Dislocation line, 87 Energy window, 27, 28
Domain pattern, 110 Escape probability, 50, 108
Domain structure, 108, 110, 111, 130 Etching, 115, 116, 119
Dynamical diffraction, 86 Everhart-Thornley scintillator detector, 13
Dynamical theory, 86, 89 Excitation threshold, 48, 110
Extinction distance, 86
E Extraction voltage, 6
EBSD camera, 91, 92, 94, 96–98, 100, 101,
109, 116 F
Edge effect, 10, 41 Field-Emission SEM (FE-SEM), 1
Elastic collisions, 1, 108 Flashing, 5–7
Index 135
Flat ion beam milling, 109, 116 L

Fluorescence, 56 LA-BSE, 15
Focused Electron Beam (FIB), 1 Landing voltage, 8, 16, 18, 20, 41, 43–45, 48,
Forecaster Detector (FSD), 97, 100, 110 118, 122, 129
Forward diffracted signal, 2, 98 Lateral resolution, 13
Forward scattered electrons, 100 Lattice imaging, 2
F-ratio, 55–57, 59, 61–63, 68, 74, 79–82, 130 Lens aberration, 17, 37, 43
Lithium, 7, 11, 24, 25, 38, 47, 99, 116, 117
G Lithium-Ion Battery (LIB), 24, 25, 38, 124, 125
Glutaraldehyde, 42 Lithium titanate, 7, 99
Lorentz force, 108
H Low-loss electrons, 15, 26, 42, 98
HA-BSE, 19, 24 Low-loss imaging, 27
Heating stage, 122 Low voltage, 1, 2, 8–11, 19, 37, 40, 42, 48, 49,
High accelerating voltages, 11, 47, 70, 75, 97 51, 52, 67, 70, 74, 77, 82, 96, 107, 122,
High-pass filter, 15, 28–31 130, 131
High-resolution, 28, 39–41, 88, 119, 120, 124, Low voltage electron microscopy, 1, 37, 48
125, 131 Low voltage scanning transmission electron
microscopy, 2
I
Image contrast, 10, 38, 47, 49, 85, 90, 124, 126 M
Image pixel, 94 Magnetic contrast, 27, 28, 107–111
Immersion lens, 18, 107 Magnetic domains, 107–111, 130, 131
Impact point, 13, 109, 122 Magnetic force microscopy, 110
Incident angle, 40, 116, 118 Magnetic induction, 107, 108
Incoherent scattering, 86, 91, 92, 97 Magnetic properties, 44, 108
Inelastic collisions, 1, 97 Mass absorption coefficient, 57, 69
Inelastic energy loss, 2 Mass-thickness, 47, 50, 97, 100
Inelastic interaction, 85 Mean atomic number, 111, 123
In-lens detection, 7, 19, 50 Mechanical polishing, 109, 115
In-lens detectors, 2, 15, 18, 27, 28, 38, 39, 118, Mild flash, 5–7
126 Minerals, 10, 123
Interactions, 11, 13, 27, 37, 40, 43, 45, 47, 50, Monochromator, 11, 39, 47
52, 67–70, 97, 115, 119, 122, 123 Monte Carlo modeling, 9, 28, 29, 130
Interaction volume, 27, 37, 40, 67–70, 89, 108, Monte Carlo simulation, 9, 13, 14, 28, 57, 61,
122, 123 68, 70, 98, 108, 116
Interpolation, 57, 59, 61 Multivariate Statistical Analysis (MSA), 79
Inverse Pole Figure (IPF), 117, 123
Ionic liquid, 123, 124, 129 N
Ionization cross section, 69 Nanomaterials, 2, 39, 130
Ionization energy, 10, 69 Nanoparticle, 7, 19, 25, 30, 38–40, 50, 76, 78,
Ion milling, 117, 118, 123, 129 118, 126
Ion sputtering, 116, 117 Nonconductive materials, 10
Non-Evaporative Getter (NEG) pump, 5
K Non-Oriented Electrical Steel (NOES), 109
Kikuchi band, 85, 86, 91, 92, 116
Kikuchi diffraction, 2, 131 O
Kinematical theory, 91 Objective lens, 2, 14, 15, 18, 27, 71
Kossel cones, 85, 91 Optics, 1, 47
K-ratio, 55, 59, 61–63, 72, 73 Organic specimens, 10
136 Index
Orientation mapping, 2, 5, 98, 111, 130 Shadowing effect, 2, 22

Osmium tetroxide, 42 Signal-to-Noise Ratio (SNR), 5, 21
Overvoltage, 75 Silicon Drift Detectors (SDD), 2, 67, 68,
Oxidation, 37, 44 71–78, 80, 82, 130
Ozone cleaner, 118 Slope cutting, 116
Ozone cleaning, 120, 121 SMART-J, 8, 14, 18–21, 23, 30, 31, 76, 78
Snorkel lens, 15, 28, 92
P Solar cells, 38
Partial dislocations, 90, 91 Solid angle, 2, 24, 61, 67, 68, 70–72, 74, 75,
Path length, 50, 108 78, 82, 107, 130
Penetration depth, 9, 38, 43 Spatial resolution, 8, 14, 15, 18, 20–22, 27,
Penetration range, 9 29–32, 37–39, 41, 47, 48, 50, 51, 67, 68,
Phase identification, 2, 97, 100, 130 70, 75–79, 82, 87, 92, 94, 96–98, 100,
Plasma cleaning, 120 108, 110, 111, 118, 123, 125, 126,
Pole piece, 2, 14, 19, 27, 111 129–131
Polymer, 10, 118, 119 Specimen bias, 23, 31, 41, 45
Post-processing, 92, 93, 110, 130 Specimen charging, 42, 115, 123, 129
Precipitate, 7, 48, 88, 116, 130 Specimen-detector distance, 70–73
Primary electron, 1, 9, 13, 26, 27, 38, 43, 91, Specimen heating, 117, 118
92, 118, 123 Specimen preparation, 115, 116, 119, 121, 129
Principal Components Analysis (PCA), 79 Specimen response function, 47
Probe current, 5, 6, 39, 56, 63, 67, 87, 92, 129, Specimen thickness, 48, 49, 68, 81, 82, 96, 97,
131 130, 131
Probe size, 5, 8, 14, 17, 37, 43, 47, 87, 92 Spectral resolution, 2
Spin polarization, 111
Q Stalking fault imaging, 88
Quality map, 117, 123 Standardless, 59, 61–63
Steel, 7, 88, 96, 110
R STEM imaging, 49, 51, 52, 88, 101, 130, 131
Radiolysis, 10 Stray field, 107, 111
Raster motion, 1 Surface coating, 42
Reciprocal lattice, 86 Surface sensitivity, 27, 37
Reciprocity theorem, 92–94 Surface treatment, 39, 40, 120
Reference pattern, 94
Reflection High-Energy Electron Diffraction T
(RHEED), 85 TakeOff Angle (TOA), 72, 73
Reflection mode, 96, 130 Temperature, 10, 122
Region Of Interest (ROI), 74 Thin specimen, 2, 48, 49, 56, 78, 79, 81, 82,
Resolution, 1, 2, 5, 8, 9, 19, 27, 37, 41, 43, 75, 91, 96, 97, 100, 101, 130, 131
76, 86, 87, 90, 92, 94, 96, 110, 111, 115, Tip noise, 5
121, 125 Titania, 38, 39
Rotation angle, 89 Top detector, 15, 18–26, 28, 30–32, 38–40,
Rotation Contour Contrast (RCC), 89 42–44, 118, 119, 125
Topographical contrast, 22, 25, 30, 31, 43, 44,
S 118, 125
Scanning Electron Microscope (SEM), 1, 67 Topography, 19, 22, 25, 27, 38, 40, 50, 109,
Scanning Transmission Electron Microscope 118, 119
(STEM), 1 Transmission Electron Microscope (TEM), 2,
Scattering cross-section, 47, 130 11, 39–41, 56, 85, 97–101, 111, 115,
Schottky, 7, 92 116, 130, 131
Scintillator, 13, 15, 27 Transmission Forward Scattering Diffraction
Secondary Electron (SE), 2, 30 (t-EFSD), 92, 97–101
Semiconductor, 13, 101 Transmission Kikuchi diffraction, 98, 130
Index 137
Transmission mode, 79, 130, 131 W

Transmitted beam, 48 Window efficiency, 57
True landing voltage, 122 Windowless silicon drift detectors, 2, 72
Two-beam approximation, 86, 89, 91
X
U X-ray absorption, 68
Ultra-low voltage, 9, 38, 129 X-ray detectors, 1, 50
Upper detector, 15, 18–20, 22, 24–26, 28, 30, X-ray emission, 68, 70, 81, 82
31, 33, 38, 40–43, 50–52, 99, 118, 120, X-ray images, 81, 129
125 X-ray map, 67, 68, 74–79
X-ray microanalysis, 2, 5, 48, 55, 56, 59, 61,
V 63, 67, 76, 130
Vacuum, 5, 8, 18, 21, 41, 45, 91, 117, 118,
120, 123, 129 Z
Virtual beam, 94 Z-contrast, 19, 21, 22, 24–28, 30, 39, 42, 43,
Visibility criterion, 87 48, 50, 100

Field Emission Scanning Electron Microscopy New Perspectives

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Field Emission Scanning Electron Microscopy New Perspectives

Hochgeladen von

Copyright:

Verfügbare Formate

SPRINGER BRIEFS IN

APPLIED SCIENCES AND TECHNOLOGY

More information about this series at http://www.springer.com/series/8884

Field Emission Scanning

ISSN 2191-530X ISSN 2191-5318 (electronic)

© The Author(s) 2018

Printed on acid-free paper

This Springer imprint is published by Springer Nature

6.2.1 f-Ratio Method for Binary System . . . . . ........... 57

10 Advanced Specimen Preparation . . . . . . . . . . . .......... . . . . . . 115

a-SDD Annular SDD

© The Author(s) 2018 1

2.1 Cold-Field Emission Technology

To improve image or analysis quality a large signal-to-noise ratio (SNR) is nec-

© The Author(s) 2018 5

Fig. 2.2 Evolution of the (a)

2.2 CFE-SEM for Low Voltage Microscopy

The main improvement in the design of new FE-SEMs, and particularly

Fig. 2.5 Secondary electrons 62

75° for Al, Cu and Au when 60

2.3 Scanning Transmission Microscopy in the SEM

Scanning transmission electron microscopy has revealed as an invaluable technique

© The Author(s) 2018 13

(a) upper (b) top

(c) lower (d) PD-BSE

(in-lens), c (in-chamber). Additionally, the in-lens detectors, due to their speciﬁc

3.2 Application of Dual In-Lens Electron Detection

(a) (b) (c)

3.3 Energy ﬁltration

Since the beginning of scanning electron microscopy, energy ﬁltration of the

modeling applied to AlAs/GaAs superlattices and reported a signiﬁcant increase of

Table 3.4 Signal-to-noise Upper detector Bias SNR Spatial

(a) (b) (c)

(d) (e) (f)

Fig. 3.12 Signal-to-noise (a) 350 25

(c) 5.0 9.0

© The Author(s) 2018 37

developed and some applications described. Ultra-low voltage will be demonstrated

4.1 Strategy of Characterization: Deceleration and Energy

As discussed above, the variation of the accelerating voltage results in changing

(a) (c) (e)

(b) (d) (f)

4.2 High Resolution Imaging

(a) (c) (e)

(b) (d) (f)

(MWCNTs) was imaged at EL = 0.5 kV in deceleration mode with E0 = 2 kV. Iron

(a) (c) (e)

(b) (d) (f)

4.3 Low Voltage, Specimen Charging, and Material

Fig. 4.4 Low voltage (a)

4.4 Ultra-Low Voltage SEM: Uses and Limitations

As underlined previously, working at low accelerating voltages has two main

(a) BSE 2kV (b) STEM-BF 30 kV (c) STEM-BF 200kV

Fig. 5.1 Comparison of a BSE image at E0 = 3 kV (a), a STEM-BF image at E0 = 30 kV in the

precipitation growth on a large scale (Bois-Brochu et al. 2014; Brodusch et al.

(c) DF (d) BSE

(a) (b) (c)

Egerton, R. (2012). Mechanisms of radiation damage in beam-sensitive specimens, for TEM

6.1 The Limits of X-Ray Microanalysis Models

Quantitative X-ray microanalysis of bulk samples is usually obtained by measuring

6.2 Description of the f-Ratio Method

6.2.1 f-Ratio Method for Binary System

6.2.2 Generalization of the f-Ratio Method

6.3 Examples of Quantitative X-Ray Analysis Using