university of groningen fabrication of metallic
Post on 22-May-2022
1 Views
Preview:
TRANSCRIPT
University of Groningen
Fabrication of metallic nanostructures with ionsRibas Gomes, Diego
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite fromit. Please check the document version below.
Document VersionPublisher's PDF, also known as Version of record
Publication date:2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):Ribas Gomes, D. (2018). Fabrication of metallic nanostructures with ions: Theoretical concepts andapplications. University of Groningen.
CopyrightOther than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of theauthor(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne-amendment.
Take-down policyIf you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediatelyand investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons thenumber of authors shown on this cover page is limited to 10 maximum.
Download date: 22-05-2022
Fabrication of metallic nanostructures with ions
Theoretical concepts and applications
PhD thesis
to obtain the degree of PhD at the University of Groningen on the authority of the
Rector Magnificus Prof. E. Sterken and in accordance with
the decision by the College of Deans.
This thesis will be defended in public on
Tuesday 9 October 2018 at 11.00 hours
by
Diego Ribas Gomes
born on 28 August 1985
in Ponta Grossa, Brazil
Supervisor
Prof. J. T. M. De Hosson
Co-supervisor
Dr. D. I. Vainchtein
Assessment committee
Prof. H. A. De Raedt
Prof. P. Rudolf
Prof. A. A. Turkin
Fabrication of metallic nanostructures with ions Theoretical concepts and applications
Diego Ribas Gomes
PhD thesis University of Groningen Zernike Institute PhD thesis series 2018-23 ISSN: 1570-1530 ISBN: 978-94-034-0925-2 (Printed version) ISBN: 978-94-034-0924-5 (Electronic version) Print: Ipskamp Printing The research presented in this thesis was performed in the Materials Science group of the Zernike Institute for Advanced Materials at the University of Groningen, the Netherlands. This work was funded by the Coordination for the Improvement of Higher Level Personnel (CAPES) and the Zernike Institute for Advanced Materials. Cover design by Diego Ribas Gomes. Engraving by Matthias Greuter (1620-1647) © Trustees of the British Museum.
CONTENTS
Introduction: flashback and look ahead on nano’s 1
1.1 References ......................................................................................................... 3
Experimental methods 5
2.1 Introduction to electron microscopy ............................................................... 5 2.2 Transmission Electron Microscopy ................................................................ 8 2.3 Illumination system ......................................................................................... 8 2.4 Interaction with specimen .............................................................................. 12 2.5 Quantitative X- Ray Microanalysis ............................................................... 24 2.6 Scanning Electron Microscopy ...................................................................... 26 2.7 Focused Ion Beam .......................................................................................... 28 2.8 Free-standing thin films ................................................................................ 32
2.8.1 Solid bulk films .................................................................................... 33 2.8.2 Nanoporous films ................................................................................ 34
2.9 References ...................................................................................................... 35
Electrochemical fabrication of nanoporous materials 41
3.1 Actuators .......................................................................................................... 41 3.2 Electrochemical nanofabrication .................................................................. 43
3.2.1 Anodized aluminum oxide (AAO) templates ..................................... 43 3.2.2 Templated electrodeposition of metals .............................................. 44
3.3 Experimental results and discussion ............................................................ 45 3.3.1 Commercial membranes ..................................................................... 45 3.3.2 Electrochemical deposition of nanowires .......................................... 46 3.3.3 The goal-structure ............................................................................... 49 3.3.4 Home-made AAO templates ................................................................ 51
3.4 Conclusion ...................................................................................................... 55 3.5 References ...................................................................................................... 56
Radiation Damage 59
4.1 Ion-solid interactions ..................................................................................... 59 4.2 Deep damage ................................................................................................... 65 4.3 Defect formation and diffusion ..................................................................... 69 4.4 Reaction rate theory ...................................................................................... 73 4.5 References ...................................................................................................... 76
Dense solid samples 81
5.1 Initial observations on Au cantilevers ............................................................ 81 5.2 Model of ion induced bending ....................................................................... 84
5.2.1 Point defect kinetics ............................................................................. 84 5.2.2 Analogy to thermoelasticity ................................................................ 87 5.2.3 Accumulation of interstitial clusters .................................................. 92 5.2.4 Gallium build-up ................................................................................. 95 5.2.5 Comparison with experiments ............................................................ 95
5.3 Aluminium cantilevers ................................................................................... 98 5.3.1 Kinetics of interstitial clusters ........................................................... 101 5.3.2 Comparison with experiments .......................................................... 108
5.4 Considerations about the model .................................................................. 110 5.4.1 Temperature increase due to irradiation ........................................... 110 5.4.2 Effects of dynamic composition change ............................................ 112
5.5 Conclusions ................................................................................................... 112 5.6 References ..................................................................................................... 113
Nanoporous specimens 116
6.1 Introduction ................................................................................................... 116 6.2 Experimental considerations ....................................................................... 117 6.3 Ion-induced coarsening ............................................................................... 120 6.4 Model of (IB)² for nanoporous gold ............................................................ 124
6.4.1 Kinetics of coarsening ........................................................................ 127 6.4.2 Young’s modulus of nanoporous Au .................................................129
6.5 Comparison with experiments ..................................................................... 132 6.6 Conclusions ................................................................................................... 142 6.7 References ..................................................................................................... 142
A compendium for bending of nanostructures: summary and outlook
147
7.1 Thesis summary ............................................................................................. 147 7.2 Samenvatting .................................................................................................149 7.3 Outlook ......................................................................................................... 150
7.3.1 Layered systems .................................................................................. 151 7.3.2 Size-dependent coarsening of ion-irradiated np-Au ........................ 155
7.4 Opportunities of Ion Beam Induced Bending or (IB) ² ..............................158 7.5 References...................................................................................................... 161
List of publications (thesis subject and related) 163
Acknowledgements 165
Curriculum Vitae 167
Introduction: flashback and look ahead on nano’s
“What is it that we humans depend on? We depend on our words... Our task
is to communicate experience and ideas to others. We must strive uninterruptedly
to extend the scope of our description, but in such a way that our messages do not
thereby lose their objectives or unambiguous character… We are suspended in
language in such a way that we cannot say what is up and what is down. The
word "reality" is also a word, a word which we must learn to use correctly.” [1]
As much as the clichés might hurt the readers' sensibility, in talking about
miniaturization there seems to be little way to escape mentioning Richard
Feynmann's lecture given at Caltech in 1959, "There is plenty of room at the
bottom" [2]. The revolutionary idea that a given material could be arranged at the
atomic level and give rise to novel properties and applications paved the way to the
rise of a field that still sounds fancy and intriguing 60 years later - the field of
nanoscience. Just a few years after that legendary lecture, Moore laid out his
famous law proposing that the number of components in integrated circuits would
not stop doubling every two years [3]. At the time of this writing, Moore's law has
just started to slow down and that is probably fair enough considering that inside
the laptop where this text is being typed, the processor transistors are mere 22
nanometers apart from each other. As the properties of materials at those very
small scales differ substantially from bulk systems, the exploration of all that "room
at the bottom" has required and continues to require theoretical and experimental
2 Chapter 1
studies that have the potential to revolutionize existing scientific fields and create
new ones.
With the reader's excuse for another paragraph of historical background, the
onset of the theme of this Thesis started before the episodes mentioned above. The
displacement of atoms caused by energetic irradiation - although with neutrons
instead of ions - was a problem for Eugene Wigner who was, during World War II,
in charge of a group tasked with the design and production of the first nuclear
reactors (incidentally, the project to which his group belonged was originally called
"Development of Substitute Materials" but was renamed posteriorly, for concealing
purposes, to the well-known name “Manhattan” – probably better so for the
material scientists!). The build-up of energy in the graphite moderators presented a
danger for certain types of reactors and became known for some time as "Wigner's
disease" [4], [5]. In the best spirit of Paracelsus, we try here to show that sola dosis
facit venenum: only the dose makes the poison, that is, that the displacements
induced by energetic impacts can very well be used to achieve beneficial effects, i.e.
accurate and reproducible bending of nanostructures.
The Thesis is structured as follows. Chapter 2 describes the experimental
techniques and procedures that were used for the synthesis and characterization of
the samples described in the chapters regarding ion-induced bending (Chapters 5
and 6 as explained below). Exploratory studies on nanofabrication using
electrochemical methods are described in Chapter 3. Those were inspired in the
pursuit of a structure with enhanced actuation performance by combining different
approaches described in the literature, i.e. maximization of surface area;
intercalating solid layers in between the high-surface actuation layers to improve
strain in the direction normal to the planes by Poisson effect and facilitating the
transport of charge by using aligned nanowires rather than random ligaments. In
that endeavor, the process of template electrodeposition of metals was explored
along with the fabrication of porous aluminium oxide templates. Due to the
experimental nature of this Chapter, the procedures are described in its text instead
of in Chapter 2.
A compilation of the theory on radiation damage in solids is presented in
Chapter 4 which provides the fundamental base for the approaches used in
Chapters 5 and 6. Those deal with the investigation of ion-induced bending of
dense solid and nanoporous cantilevers, respectively.
Introduction 3
Dense solid bulk Au and Al cantilevers were studied and presented different
bending behaviours under 30 keV Ga+ irradiation. While Au cantilevers bend
continuously towards the inciding ion beam with increasing fluence, Al cantilevers
bend initially away from the beam and reverse direction as irradiation continues. In
both cases, experimental results were succesfully reproduced by a reaction-rate
model based on the diffusion of crystallographic point defects and clusters. In the
case of nanoporous Au, the bending sensitivity to irradiation was greatly enhanced
in comparison to the solid bulk counterparts. The process was described in terms of
ion-induced coarsening of the structure top layers, which generates a volume
contraction responsible for the bending moment.
Chapter 7 wraps up the Thesis with an outlook on ion-induced bending
research. Preliminary experiments with bilayered cantilevers of Au and Pt
suggested a behaviour controlled by the thickness of the irradiated layer and the
stiffness of the underlying material. Deposition of a thin layer of Al2O3 onto solid Al
cantilevers markedly accentuated bending in the direction away from the beam and
increased the fluence at which reversion occurs. Furthermore, the densification of
ion-irradiated nanoporous gold cantilevers was observed to occur by different
means depending on the initial ligament size distribution.
1.1 References
[1] R. G. Newton, The Truth of Science: Physical Theories and Reality. Harvard University
Press, 1997.
[2] R. Feynman, ‘APS meeting lecture’, Caltech Engineering and Science, vol. 23, no. 5, pp.
22–36, 1960.
[3] G. E. Moore, ‘Cramming more components onto integrated circuits, Reprinted from
Electronics, volume 38, number 8, April 19, 1965, pp.114 ff.’, IEEE Solid-State Circuits
Society Newsletter, vol. 11, no. 3, pp. 33–35, Sep. 2006.
[4] E. P. Wigner, ‘Theoretical Physics in the Metallurgical Laboratory of Chicago’, Journal
of Applied Physics, vol. 17, no. 11, pp. 857–863, Nov. 1946.
[5] F. Seitz, ‘Radiation effects in solids’, Physics Today, vol. 5, no. 6, p. 6, 1952.
Experimental methods
“We become what we behold. We shape our tools and then our tools shape us.” [1]
2.1 Introduction to electron microscopy
As it goes without saying, microscopy in the field of materials science is
devoted to linking microstructural observations to properties. However, the actual
linkage between the microstructure studied by microscopy on one hand and the
physical property of a material on the other hand is almost elusive. The reason is
that various physical properties are determined by the collective behavior of
moving defects rather than by the behavior of a single stationary defect. For
instance, there exists a vast but also bewildering amount of electron microscopy
analyses concerned with post-mortem observation of ex-situ deformed materials,
which try to link observed patterns of defects to the mechanical property. However,
in spite of the enormous efforts which have been put in both theoretical and
experimental work since the theoretical concept of a crystal dislocation was
introduced in 1934, a clear physical picture that can predict even one simple stress-
strain curve based on these microscopy observations is still lacking. In fact there
does not exist an easy and unique back transformation from microscopy
observations to properties.
There are at least three reasons which hamper a straightforward correlation
between microscopic structural information and materials properties, both from
fundamental and practical viewpoints. First, in materials we are facing non-
6 Chapter 2
equilibrium effects. Defects determining the physical performance are usually not
in thermodynamic equilibrium and their behavior is very much non-linear. Non-
linearity and non-equilibrium are both tough problems in materials physics and in
fact unsolved. Second, metrological considerations of quantitative electron
microscopy of materials pose very critical questions to the statistical significance of
the electron microscopy observations. In particular, in situations where there is
only a small volume fraction of defects present or where there is a very
inhomogeneous distribution, statistical sampling may be a serious problem. Third,
even when statistical sampling is adequate, still the physical property might be
determined, not by its mean value, say grain size, but by the extremes, e.g. in a
nanostructured metallic material with many nano-grains but still having one
singular big grain, macroscopic yielding is determined by the local plasticity in that
largest grain and not by the many nano-sized grains present! Therefore statistical
averaging of grain-size distributions is not the right thing to do when explaining the
onset of plasticity in metallic systems. As a consequence, do not only magnify to the
highest point-point resolution in TEM - which is quite popular these days- but also
de-magnify so as to clarify and not falsify the essentials in the structure-property
relationship.
Electron microscopy is the interaction between a plane-wave (wave vector is
very large because of picometer wavelength) and an object. The wave-like
characteristics of electrons that are essential for electron microscopy were first
postulated in 1924 by Louis de Broglie (in fact in his PhD thesis [2]! , Nobel Prize in
Physics 1929), with a wavelength far less than visible light. In the same period
Busch revealed that an electromagnetic field might act as a focusing lens on
electrons [3]. Subsequently, the first electron microscope was constructed in 1932
by Ernst Ruska. For his research, he was awarded, much too late of course, in 1986
the Nobel Prize in Physics together with Gerd Binnig and Heinrich Rohrer who
invented the scanning tunneling microscope. The electron microscope opened new
horizons to visualize materials structures far below the resolution reached in light
microscopy. The most attractive point is that the wavelength of electrons are much
smaller than atoms and it is at least theoretically possible to see details well below
the atomic level.
However, currently it is impossible to build transmission electron microscopes
with a resolution limited by the electron wavelength, mainly because of
Experimental methods 7
imperfections of the magnetic lenses. In the middle of the 70's the last century
commercial TEMs became available that were capable of resolving individual
columns of atoms in crystalline materials. High voltage electron microscopes, i.e.
with accelerating voltages between 1 MV and 3 MV have the advantage of shorter
wavelength of the electrons, but also radiation damage increases. After that period
HRTEMs operating with intermediate voltages between 200 kV – 400 kV were
designed offering very high resolution close to that achieved previously at 1 MV.
More recently, developments are seen to reconstruct the exit wave (from a
defocus series) and to improve the directly interpretable resolution to the
information limit. During the last two decades, electron microscopy has witnessed
a number of innovations that enhanced existing approaches and introduced
qualitatively new techniques. One of the most important novel technologies is
aberration correction. Correction of spherical aberration Cs has been demonstrated
by Haider et al [4]. In the years that followed, this hexapole corrector was applied
to various materials science problems [5]–[7]. A quadrupole/octopole corrector
was developed and applied by Dellby et al. [8] for scanning transmission electron
microscopy (STEM) mode. Since then, aberration correction has been used to help
solve various material science problems [9]–[11]. Recently, correction of chromatic
aberration has been demonstrated [12] and successfully applied to improve
resolution in energy-filtered TEM (EFTEM) by a factor of five or so. Furthermore,
there has been significant progress in EELS, resulting in improved energy
resolution (monochromator) and a larger field of view for electron-spectroscopic
[13] and energy-filtered imaging [14]. For a review reference is made to [15].
Elastically scattered electrons generate the basis of the image formation in
HRTEM and they are the predominant fraction of the transmitted electrons for
small sample thickness (<15 nm). In contrast, the thicker the sample the more
electrons become inelastically scattered and this must be prevented as much as
possible, because they contribute mostly to the background intensity of the image.
Therefore, the thinner the specimen (≤15 nm) the better the quality of the HRTEM
images. The inelastic scattered electrons can be removed by inserting an energy
filter in the microscope between specimen and recording device. In the following
section on image formation, only the elastic scattered electrons are considered. The
main technique for detailed examination of the defect structure of nanoligaments
8 Chapter 2
in our work concerns transmission electron microscopy and scanning electron
microscopy.
Here we present a concise review for non-experts, without going into every
detail. This part is based on summaries that were also presented in several reviews
[16], [17] and by former PhDs of our MK group, in particular by Bas Groen, Patricia
Carvalho, Wouter Soer, Zhenguo Chen, Sriram Venkatesan, Stefan Mogck, Mikhail
Dutka and Sergey Punzhin [18]–[25].
2.2 Transmission Electron Microscopy
A schematic picture of a TEM is shown in Figure 2.1a. A transmission electron
microscope consists of an illumination system, specimen stage and imaging system,
analogous to a conventional light microscope. Several textbooks and reviews have
been written on the subject of (high resolution) microscopy, sadly with different
notation conventions. This chapter adopts the notation used in [26], [27] Spence
[28] has presented a more in-depth description of imaging in HRTEM theory.
Williams and Carter [29] have written a very inspirational textbook on general
aspects of electron microscopy. In our work a JEOL 2010F TEM (FEG, 200kV) was
used for atomic structure observations and EDS analysis. For a textbook on theory
behind elastic and inelastic scattering processes in TEM reference is also made to
[30].
2.3 Illumination system
Electrons are generated in an electron gun, accelerated towards the anode and
focused at the specimen with condenser lenses. High resolution TEM requires
planar coherent electron waves, since high-resolution micrographs are formed by
phase contrast. For elemental analysis, it is also important to have the possibility to
focus the electron beam within a diameter (FWHM) of the order of 1 nm to
determine chemical compositions in the nanometer range. Several demands must
be fulfilled: high brightness, small source size and little energy spread of the
electrons.
Experimental methods 9
Figure 2.1 – (a) Cross section of a basic TEM. Simplified ray diagram showing
the two basic operations modes of the TEM. (b) Diffraction pattern mode (c)
Imaging mode. Figures adapted from Refs. [29], [31].
(a)
(b) (c)
10 Chapter 2
The brightness of the beam is an important parameter and is defined as
follows:
I
BA
, (2.1)
i. e. the brightness value B is related to the electron current I emitted from the area
A (which also determines the spatial coherency) into the spatial angle Ω. The
conventional way to generate electrons is to use thermionic emission. Any material
that is heated to a high enough temperature will emit electrons when they have
enough energy to overcome the work function. In practice this can only be done
with high melting materials (such as tungsten) or low work function materials like
LaB6. Richardson’s law describes thermionic emission as a function of the work
function W and temperature T:
2 exp
WJ CT
kT
(2.2)
with the current density J at the tip and C the so-called Richardson-Dushman's
constant depending on the material used for the tip. In electron microscopes
tungsten filaments were most commonly used until the introduction of LaB6. LaB6
sources have the advantages of a lower operating temperature that reduces the
energy spread of the electrons and increases the brightness.
Another way of extracting electrons from a material is to apply a high electric
field to the emitter that enables the electrons to tunnel through the barrier.
Sharpening the tip may enhance the electric field since the electric field at the apex
of the tip is inversely proportional to the radius of the apex:
ER
V
k , (2.3)
with k a correction factor for the tip geometry (usually around 2). The advantage
of this cold field emitter gun (cold FEG) is that the emission process can be to done
at room temperature, reducing the energy spread of the electrons. The small size of
the emitting area and the shape of the electric field results in a very small (virtual)
source size in the order of nanometers with a brightness that is three orders of
magnitude higher than for thermionic sources. It is thus possible to focus the beam
to a very small probe for chemical analysis at an atomistic level or to fan the beam
Experimental methods 11
to produce a beam with high spatial coherence over a large area of the specimen.
The disadvantages of this type of emitter are the need for (expensive) UHV
equipment to keep the surface clean, the need for extra magnetic shielding around
the emitter and the limited lifetime. For electric fields E > 107 V/cm the electron
current density J can be described according the modified Fowler-Nordheim
relation:
6 7 3/2
2
2
1.54 10 6.8 10 ( )exp
( )
v y WJ E
EWt y
, (2.4)
where J is the field-emitted current density in A/m2, E is the applied electric field at
the tip, and W is the work function in eV. The functions v and t of the variable
4 1 23.79 10y E depend weakly on the applied electric field and have been
tabulated in literature [32]. In our FEG-TEM it is therefore possible to focus the
beam to a very small probe for chemical analysis at a sub-nanometer range and
produce a beam with high spatial coherence over a large area of the specimen. The
small (virtual) source size reaches values for the brightness in the order of B = 1011
to 1014 Am-2sr-1 with an energy spread of 0.2–0.5 eV compared with ~ 109 Am-2sr-1
and an energy spread of 0.8–1 eV for thermionic emission.
Some of these disadvantages of cold FEG emitters can be circumvented by
heating the emitter to moderate temperatures (1500C) in the case of a thermally
assisted FEG or by coating the tip with ZrO2 which reduces the work function at
elevated temperatures and keeps the emission stable (Schottky emitter). For
thermal assisted FEGs the work function is often reduced by coating the tip with
ZrO2 which keeps the emission stable (Schottky emitter). This increases the energy
spread of the emitted electrons by about a factor of 2 and some reduction of the
temporal coherency compared with cold FEGs. The Schottky emitter is widely used
in commercial FEG-TEMs, because of the stability, lifetime and high intensity.
Table 2.1 gives an overview of the technical data of the different electron sources.
12 Chapter 2
Table 2.1 – Properties of different electron sources [33]
Parameter Tungsten LaB6 Cold FEG Schottky Heated
FEG
Brightness, A/m2sr (0.3-2)109 (3-20)109 1011-1014 1011-1014 1011-1014
Temperature, K 2500-3000 1400-2000 300 1800 1800
Work function, eV 4.6 2.7 4.6 2.8 4.6
Source size, μm 20-50 10-20 <0.01 <0.01 <0.01
Energy spread, eV 3.0 1.5 0.3 0.8 0.5
Transmission electron microscopes operate generally between 80 kV and
1000 kV meaning that the velocity of the electron includes relativistic effects. This
must be taken into account applying the De Broglie relationship to calculate the
wavelength:
1/2
00 0 2
0
2 12
eEh m eE
m c
(2.5)
For 400 kV electrons = 1.64 pm which is much smaller than the resolution
of any electron microscope because the resolution is limited by aberrations of the
objective lens and not by the wavelength of the electrons. The electron beam is now
focused on the specimen with the condenser lenses and aligned using several
alignment coils. The function of the condenser lens system is to provide a parallel
beam of electrons at the specimen surface. In practice this is not possible and the
beam always possesses a certain kind of convergence when imaging at high
resolution, usually in the range of 1 mrad for LaB6 emitters and 0.1 mrad for FEGs.
2.4 Interaction with specimen
After entering the specimen most of the electrons are scattered elastically by
the nuclei of the atoms in the specimen. Some electrons are inelastically scattered
by the electrons in the specimen. Compared to X-ray or neutron diffraction the
interaction of electrons with the specimen is huge and multiple scattering events
are common. For thick specimens at lower resolutions an incoherent particle model
can describe the interaction of the electrons with the specimen. However, with thin
Experimental methods 13
specimens at high resolution this description fails because the wave character of the
electrons is then predominant. The electrons passing the specimen near the nuclei
are somewhat accelerated towards the nuclei causing small, local, reductions in
wavelength, resulting in a small phase change of the electrons. Information about
the specimen structure is therefore transferred to the phase of the electrons.
After the (plane) electron wave strikes the specimen a part of the electrons
experience a phase shift. The essential part of image formation in TEM is the
transformation of the phase shift stored in the exit wave into amplitude
modulation, and therefore into visible contrast. With the assumption that the
specimen is a pure phase object the phase shift can be described with the image
function:
( )( ) i r
image r e (2.6)
For sufficiently thin objects the phase can be considered as weak ϕ(𝑟) ≪ 1 and
( )image r can be approximated to first order: ( ) 1 ( )image r i r . r is the
projected potential distribution in the materials slice, averaged over the electron
beam direction. The technique of HRTEM has a base in the technique of phase
contrast microscopy, introduced by Zernike [34] for optical microscopy in the
physics department of our own university, the University of Groningen. In 1953 he
received the Nobel Prize in Physics for this invention which made an undisputed
immense impact, in particular into bio- and medical sciences. The principle and
very basic problem in microscopy, both in optical and electron microscopy, is to
derive information about the object we are interested in. The object is defined by
amplitude and phase and to retrieve information about ( )object r from the intensity
of the image2
( )image r is a very complex. Even in an ideal but non-existing
(because of aberrations in any optical system in practice) microscope the intensity
of the image is exactly equal to the intensity of the object function but still phase
information in the intensity is lost.
Zernike had a brilliant and at the same time an utmost simple idea, i.e. most
probably a prerequisite of any Nobel Prize. He realized that if the phase of the
diffracted beam can be shifted over 2 it is as if the image function has the form
exp ( ) 1 ( )image r r and in fact the weak phase object acts as an
14 Chapter 2
amplitude object, i.e. the intensity observed, i.e. I= 2
( ) 1 2 ( )image r r , can be
connected to ( )r . High-resolution TEM imaging is based on the same principles.
Phase-contrast imaging derives contrast from the phase differences among the
different beams scattered by the specimen, causing addition and subtraction of
amplitude from the forward-scattered beam. Components of the phase difference
come from both the scattering processes itself and the electron optics of the
microscope. Just because of a fortunate balance between spherical aberration and
defocus we can perform high-resolution TEM with atomic resolution as will be
shown in the following.
A possibility to visualize the phase contrast in TEM was introduced by
Scherzer. Perfect lenses show no amplitude modulation, but imaging introduces an
extra phase shift between the central beam and the beam further away from the
optical axis of the objective lens (deviation from the ideal Gaussian wave front).
Deviations from the ideal Gaussian wave front in lenses is known as spherical
aberration. For particular frequencies the phase contrast of the exit wave will be
nearly optimally transferred into amplitude contrast. Therefore the observed
contrast in TEM micrographs are mainly controlled by the extra phase shift of the
spherical aberration and the defocus. The influence of the extra phase shift on the
image contrast can be taken into account by multiplying the wavefunction at the
back focal plane with the function describing the extra phase shift as a function of
the distance from the optical axis. The lenses can be conceived cylindrical
symmetric, and will be represented as a function of the distance of the reciprocal
lattice point to the optical axis 1/22 2 U u v , where u and v are the angular
variables in reciprocal space. The extra phase factor ( )U depends on the
spherical aberration and defocus [28]:
2 3 4( ) 0.5 sU fU C U (2.7)
with f the defocus value and sC Cs the spherical aberration coefficient. The
function that multiplies the exit wave is the so-called transfer function ( )U :
( )( ) ( ) i UU E U e . (2.8)
Experimental methods 15
For the final image wave after the objective lens, assuming a weak phase
object:
1( ) ( ) ( ) ( ) ( )image object objectr r U r r
, (2.9)
where and 1 are symbols to represent the Fourier transform and inverse
Fourier transform, respectively. In fact ( )object r is the electron diffractogram in
the back focal plane of the objective lens. As said before because of a fortunate
balance between spherical aberration and defocus, i.e. with the transfer function
( )r in Eq. (2.9) going close to unity, information about the object can be obtained
from the image. After the multiplication of Eq. (2.9) with its complex conjugate and
neglecting the terms of second order and higher the image formation results in:
1( ) 1 2 ( ) [ ( )sin( ( ))]imageI r r E U U . (2.10)
Under ideal imaging conditions the wave aberration must be /2 or
sin( ( ))U = 1 for all spatial frequencies U . Therefore sin( ( ))U is called the
contrast transfer function (CTF). Only the real part is considered since the phase
information from the specimen is converted in intensity information by the phase
shift of the objective lens. In the microscope an aperture is inserted in the back
focal plane of the objective lens, only transmitting beams to a certain angle. This
can be represented by an aperture function A U which is unity for 0U U and
zero outside this radius. When ( )U is negative the atoms in the specimen would
appear as dark spots against a bright background and vice-versa. For ( ) 0U no
contrast results. An ideal behaviour of ( )U would be zero at 0U (very long
distances in the specimen) and 0U U (frequencies beyond the aperture size) and
large and negative for 00 U U .
The contrast of a high-resolution image depends strongly on the microscope
settings and parameters. In practice not all the information in ( )U is visible in the
image. This is caused by electrical instabilities in the microscope causing a spread
of focus because of the chromatic aberration of the objective lens, resulting in
damped higher frequencies. Mechanical instabilities and energy loss due to
16 Chapter 2
inelastic scattering of the electrons by the specimen also contribute to the spread in
defocus. The inelastic scattered electrons contributing to the image can be removed
by inserting an energy filter in the microscope between the objective lens and the
image recording media. Another factor that damps the higher frequencies is the
beam convergence. Since the electron beam has to be focussed on a small spot on
the specimen there is some convergence of the beam present. This also affects the
resolution since the specimen is now illuminated from different angles at the same
time.
These effects which affect the resolution can be represented by multiplying
sin( ( ))U by the damping envelopes E and E which represent the damping by
the convergence and spread in defocus respectively:
2 2 2 412
22 2 2 2 2
( ) exp ,
( ) exp .s
E U U
E U U f C U
(2.11)
The resulting contrast transfer function (CTF) for the microscope we used is
plotted in Figure 2.2 (bottom) with the damping envelopes E and E . For higher
frequencies the CTF is now damped and approaching zero. It becomes clear that
defining a resolution for the HRTEM is not obvious.
Experimental methods 17
0 2 4 6 8 10 12
-1.0
-0.5
0.0
0.5
1.0
E
E
sin((U))
Contr
ast,
a.u
.
Scattering vector, U, nm-1
EEsin((U))
0 2 4 6 8 10 12
-1.0
-0.5
0.0
0.5
1.0
Contr
ast,
a.u
.
Scattering vector, U, nm-1
E
E
EEsin((U))
Figure 2.2 – Contrast transfer functions and damping envelopes of the JEOL
4000 EX/II (top) and JEOL 2010F (bottom) at optimum defocus of 47 and 58
nm, respectively. The highly coherent electron source used in the 2010F, a FEG,
is apparent from the many oscillations in the CTF of the 2010F.
Several different resolutions can be defined as stated by O’Keefe [35] :
(1) Fringe or lattice resolution: This is related to the highest spatial frequency
present in the image. In thicker crystals second-order or non-linear
interference may cause this. Since the sign of sin((U)) is not known there
18 Chapter 2
is in general no correspondence between the structure and the image. This
resolution is limited by the beam convergence and the spread in defocus.
(2) Information limit resolution: This is related to the highest spatial
frequency that is transferred linearly to the intensity spectrum. These
frequencies may fall in a passband, with blocked lower frequencies. Usually
this resolution is almost equal to the lattice resolution. A definition used is
that the information limit is defined as the frequency where the overall
value of the damping envelopes corresponds to 2e (e.g. the damping is 5%
of the CTF intensity).
(3) Point resolution. This is the definition of the first zero point in the contrast
transfer function, right hand side of the Scherzer band. For higher spatial
frequencies, the image contrast and thus the structure cannot be
unambiguously interpreted. The optimal transmitted phase contrast is
defined as a Scherzer defocus 1/2
4 3 sC and the highest transferred
frequency is equal to 1 4 3/41.5 sC .
In Figure 2.2 the CTFs of the JEOL 4000EX and JEOL 2010F at optimum
defocus are plotted with the damping envelopes. The corresponding electrical-
optical properties are listed in Table 2.2. For the two microscopes with a different
illumination system, it is clearly visible that with the JEOL 2010F a rapid
oscillation of the CTF occurs. These oscillations arise because the spatial coherence
is higher for the 2010 F (spread of defocus and in particular beam convergence is
smaller) than for the 4000 EX/II. Correspondingly, the information limit of the
JEOL 2010F is better than of the 4000EX/II. For the JEOL 2010F the information
limit is a factor 2 higher compared with the point resolution. For microscopes with
higher operating voltages, a FEG will not significantly increase the information
limit, because the higher the voltage the larger the energy spread of the electrons
and the damping envelope is limited by the spread of defocus, instead of beam
convergence. For lower (≤200 kV) voltage TEMs the situation is clearly more
favorable and a FEG thus significantly improves the information limit. With higher
voltage instruments the higher acceleration voltage increases the brightness of the
source and the damping envelope is limited by the spread of defocus. The spherical
aberration of the objective lens can be lowered but generally at the expense of a
decrease in tilt capabilities of the specimen. It is possible to compensate sC by a set
Experimental methods 19
of hexapole lenses as suggested by Scherzer [36] in 1947 but it was only feasible
recently due to the complex technology involved. Haider demonstrated a sC
corrected 200 kV FEG microscope [37] in which sC was set at 0.05 mm to reach an
optimum between contrast and resolution. The point resolution of this microscope
was now equal to the information limit of 0.14 nm. Successful application of this
technology would put the limiting factors of the microscope at the chromatic
aberration and mechanical vibrations that currently limit the resolution around
0.1 nm.
Table 2.2 – Technical data of JEOL 4000EX and JEOL 2010F microscopes
Parameter JEOL 4000EX/II JEOL 2010F
Emission LaB6
(filament)
FEG
(Schottky emitter)
Operating voltage, 0E , kV 400 200
Spherical aberration coefficient, sC , mm 0.97 1.0
Spread of defocus, , nm 7.8 4.0
Beam convergence angle, , mrad 0.8 0.1
Optimum defocus, f , nm –47 –58
Information limit, nm 0.14 0.11
Point resolution, nm 0.165 0.23
For a correct interpretation of the structure the specimen has to be carefully
aligned along a zone axis, which is done with the help of Kikuchi patterns and an
even distribution of the spot intensities in diffraction. The misalignment of the
crystal is in this way reduced to a fraction of a mrad. Beam tilt has a more severe
influence on the image because the tilted beam enters the objective lens at an angle
causing phase changes in the electron wave front. To correct the beam tilt the
voltage centre alignment is used by which the acceleration voltage is varied to find
the centre of magnification that is then placed on-axis. This alignment might not be
correct due to misalignments in the imaging system after the objective lens. For a
better correction of the beam-tilt the coma-free alignment procedure is necessary.
This is done by varying the beam tilt between two values in both directions and
optimizing the values until images with opposite beam tilts are similar. This
20 Chapter 2
procedure is only practically feasible with a computer-controlled microscope
equipped with a CCD camera and is not used here; instead, the voltage centre
alignment is used which could result in some residual beam tilt.
Besides high-resolution transmission electron microscopy ‘conventional’
transmission electron microscopy has been employed in this thesis work trying to
unravel the character of the defects after using Focused Ion Beam methods. Here,
to interpret electron micrographs it is essential to understand the factors that
determine the intensities of Bragg-diffracted beams. Various approaches can be
followed, i.e. ranging from the kinematical theory to the dynamical theory of
electron diffraction. The former is based on the assumptions that only elastic
scattering takes place (hence no absorption) and that an electron can be scattered
only once, whereas the latter also allows for interaction between the diffracted
beams. This interaction is particularly well-defined when the crystal is tilted in such
a way that, besides the direct beam, only one beam is strongly diffracted (i.e. gs
>>0 for the other reflections). As the transmitted wave with amplitude 0 z
propagates through the crystal, its amplitude is depleted by diffraction and the
amplitude zg of the diffracted beam increases, i.e. a dynamical interaction
between 0 z and zg exists. If we assume that sg is parallel to the electron
beam, the interaction can be described by the following pair of coupled differential
equations, known as the Howie-Whelan equations:
2
0
0
is zd i ie
dz
gg
g
g
(2.12)
and
20
0
0
is z
g
d i ie
dz
g
g
, (2.13)
where sg = |sg| and ξ0 and ξg are the extinction distances for the direct and
diffracted waves, respectively. The extinction distance is a characteristic length
scale determined by the atomic number of the material, the lattice parameters and
the wavelength of the electrons, it typically lies between 10 and 100nm. The above
equation shows that the change in 0 z as a function of depth z is the sum of
Experimental methods 21
forward scattering and scattering from the diffracted beam, taking into account a
phase change of π/2 caused by the scattering. Solving the previous 2 equations for
zg gives:
2 2g g g2 2
gg g
sin 11
i ts
s
, (2.14)
where t is the thickness of the crystal. Accordingly, the intensity of the diffracted
beam becomes:
2
2eff
2 2g eff
sin tsI
s
g g , (2.15)
where seff is an effective value of sg defined by 1/22 2 effs s g g
. In fact, by
replacing seff by sg , the intensity according to the kinematical approximation is
found. Absorption can be included in the dynamical theory by adding appropriate
terms to both ξ0 and ξg. Mathematically speaking, absorption is included simply by
allowing the arguments of the sines and cosines to become complex.
Dislocations, dislocation loops or stacking faults give rise to contrast because
they locally distort the lattice and thereby change the diffraction conditions. If the
distortion is given by a displacement field R, i.e.
0
0
exp 2 g
d i ii s z
dz
g
g
g
g R (2.16)
In order for a dislocation to contribute to contrast formation, the dot product
g·R must be nonzero. A screw dislocation in an isotropic elastic medium has a
displacement field parallel to its Burgers vector, and therefore produces no contrast
when g·b = 0. General dislocations have a displacement field with more
components; their image contrast also depends on g·be and g·(bu), where be is
the edge component of the Burgers vector and u is a unit vector along the
dislocation line. In practice however, only very faint contrast occurs when g·b = 0
but g·be 0 and g·(bu) 0. Therefore, the “invisibility criterion” g·b = 0 is used
commonly to determine Burgers vectors of dislocations in elastically isotropic
22 Chapter 2
solids. The determination of a Burgers vector involves finding two reflections g1
and g2 for which the dislocation is invisible, so that b is parallel to g1g2.
The situation where only one beam zg is strongly diffracted is referred to
as a two-beam condition. This type of diffraction is widely used in conventional
TEM of crystalline materials because the contrast is well defined and the Burgers
vectors of the dislocations can be determined as described above. By using the
objective aperture in the microscope, either of the two beams can be selected to
form an image and accordingly, two imaging modes may be distinguished: bright
field (BF) when the direct beam 0 z is used, and dark field (DF) when the
diffracted beam zg is used. Generally, the diffracted beams do not coincide with
the optical axis of the microscope and consequently the DF image will not be of
maximum quality due to spherical aberration. To overcome this problem, the
incident beam is normally tilted in such a way that the desired diffracted beam
passes along the optical axis.
While two-beam conditions produce high contrast of dislocations, the
resolution at which these defects are resolved is not optimal since the lattice planes
around the dislocations are distorted over a relatively large area. To obtain the
maximum resolution, the crystal should be tilted slightly further by such an amount
that the exact Bragg condition is only fulfilled within a small region near the
dislocation. In this way, a high-resolution image is obtained in which the
dislocation shows up as a bright line. This technique is referred to as weak-beam
imaging. The deviation from the exact Bragg condition for perfect crystal is given
by sg as mentioned above and can be determined accurately by the relative position
of the so-called Kikuchi pattern with respect to the diffraction pattern. The origin of
the Kikuchi pattern lies in the elastic re-scattering of inelastic scattered electrons as
further explained in the following section.
The width of a dislocation image is approximately 0.3ξg, i.e. several tens of
nanometers in conventional bright- and dark-field imaging. This width can be
detrimental to the observations of dislocations that are very closely spaced. Since
the effective extinction distance decreases for increasing deviation away from the
Bragg condition, the width of a dislocation image can be reduced to values in the
order of 1 to 5 nm in weak-beam imaging.
Experimental methods 23
At lower magnifications, the average crystal orientation may vary considerably
within the observed area, so that only a small area of the specimen can be set up in
two-beam condition. This is especially relevant in specimens that have been
deformed prior to preparation. In these cases, it is often convenient to orient the
specimen close to a zone axis, so that many reflections are weakly excited,
independently of small changes in orientation. By using the direct beam for
imaging, the dislocation structure can be imaged with good contrast over a
relatively large area. However, the contrast from individual dislocations is generally
smaller than in two-beam condition and not well defined since it results from many
different reflections.
Besides dislocations also dislocation loops appear in this Thesis, generated by
radiation damage. In the past years the nature of small point defect clusters has
been analyzed successfully by means of the black-white contrast figures produced
under dynamical two-beam conditions. In particular, this technique has been
applied to dislocation loops which are so small that their geometrical shapes are
not resolvable under dynamical two-beam conditions or under other strong beam
conditions [38]–[40].
From the black-white contrast the following properties of small loops can in
principle be determined: (a) the Burgers vector, (b) the normal n of the loop plane,
(c) the type of the loop (differentiating between loops of vacancy and interstitial
type). For the assessment of (d), the diameters and (e), the number densities, the
so-called kinematical two-beam conditions were mainly applied. However, there
are a number of experimental difficulties which sometimes complicate the
application of the black-white technique. Cockayne et al [41], [42] have shown that
imaging with a weakly excited beam (weak-beam technique, WB) has considerable
advantages when high resolution is attempted. For instance, Häussermann has
shown that this method, applied to small defect clusters, facilitates the
differentiation between isolated single loops and clusters of loops lying closely
together (multiple loops). These multiple loops are not resolvable under strong-
beam conditions. For a theoretical study of the black-white contrast figures on
electron micrographs produced by small dislocation loops under dynamical two-
beam conditions reference is made to [43].
24 Chapter 2
2.5 Quantitative X- Ray Microanalysis
Quantitative X-ray microanalysis was performed using a JEOL 2010F
analytical transmission electron microscope. Additionally to the operation in the
TEM-mode (parallel incidence of the electron beam) there exists also the possibility
to operate in X-ray energy-dispersive spectrometry (EDS) and nano-beam-
diffraction (NBD) mode. In the NBD-mode the convergences angle α of the beam
incidence is smaller, whereas in the EDS-mode these angles are larger. In the latter
case, the diameter of the electron probe can be reduced to approximately 0.5 nm
(FWHM) and thus enables very localized chemical analysis. The NBD-mode will be
not considered in the present work.
In EDS the inelastic scattered electrons are essential. In general, the highly
accelerated primary electrons are able to remove one of the tightly bound inner-
shell electrons of the atoms in the irradiated sample. This “hole” in the inner-shell
will be filled by an electron from one of the outer-shells of the atom to lower the
energy state of the configuration. After recombination each element emits its
specific characteristic X-rays or an Auger electron. The electron of the incident
beam also interacts with the Coulomb field of the nuclei.
These Coulomb interactions of the electrons with the nuclei lower their
velocity and produce a continuum of Bremsstrahlung in the spectrum. The results
is that the characteristic X-rays of a detected element in the specimen appear as
Gaussian shaped peaks on top of the background of Bremsstrahlung. This
background of the EDS spectrum must be taken into account in quantitative
analysis.
A Si(Li)-detector is mounted between the objective pole pieces with a ultra-
thin window in front of it. This has the advantage that X-rays from light elements
down to boron can be detected. The EDS unit in an analytical TEM has three main
parts: the Si(Li)-detector, the processing electronics and the multi-channel
analyzer (MCA) display. After the X-rays penetrate the Si(Li)-detector a charge
pulse proportional to the X-ray energy will be generated that is converted into a
voltage. This signal is subsequently amplified through the field-effect transistor.
Finally, a digitized signal is stored as a function of energy in the MCA. After manual
subtraction of the background from the X-ray dispersive spectra the quantification
of the concentration iC (i = A,B) of elements A and B can be related to the
Experimental methods 25
intensities iI in the X-ray spectrum by using the Cliff-Lorimer [31] ratio technique
in the thin-film approximation:
A AAB
B B
C Ik
C I . (2.17)
For the quantification, the Cliff-Lorimer factor kAB is not a unique factor and
can only be compared under identical conditions (same: accelerating voltage,
detector configuration, peak-integration, background-subtraction routine). The
ABk factor can be user-defined or theoretical ABk values are stored in the library of
the quantification software package [44]. With modern quantification software, it is
possible nowadays to obtain an almost fully automated quantification of X-ray
spectra using the MCA system. The intensities IA and IB are measured, their
background is subtracted and they are integrated. For the quantification the Kα-
lines are most suitable, since the L- or M-lines are more difficult due to the
overlapping lines in each family. However, highly energetic Kα-lines (for energies >
20 keV non-linear effects in the Si(Li)-detector) should be excluded for
quantification in heavy elements and L- or M-lines must be taken into account
instead. The possible overlapping peaks in X-ray spectra must be carefully
analyzed. Poor counting statistics, because of the thin foil can be a further source of
error in particular for detection of low concentrations. A longer acquisition time
increases the count rates (better statistics), but this may have the drawback of
higher contamination and sample drift during the recording of the spectra. Sample
drift is extremely disadvantageous in experiments where spatial resolution is
essential.
The correction procedure in bulk microanalysis is often performed with the
ZAF correction; Z for the atomic number, A for absorption of X-rays and F for
fluorescence of X-rays within the specimen. For thin electron-transparent
specimen the correction procedure can be simplified, because the A- and F factors
are very small and only generally the Z-correction is necessary. All acquisitions of
the present work were performed using a double-tilt beryllium specimen holder.
The beryllium holder prevents generation of detectable X-rays from parts of the
specimen holder. A cold finger near the specimen (cooled with liquid nitrogen)
reduces hydrocarbon contamination at the surface of the specimen.
26 Chapter 2
2.6 Scanning Electron Microscopy
The Scanning Electron Microscope (SEM) is a type of electron microscope
that study microstructural and morphological features of the sample surface by
scanning it with a high-energy electron beam in a raster scanning pattern. The
electrons interact with the sample matter producing a variety of signals that
contain information about the sample surface topography, composition and other
properties.
The first SEM image was obtained by Max Knoll, who in 1935 obtained an
image of silicon steel showing electron channeling contrast [45]. The first true
scanning electron microscope, i.e. with a high magnification by scanning a very
small raster with a demagnified and finely focused electron beam, was produced by
von Ardenne a couple of years later [46]. The instrument was further developed by
Sir Charles Oatley and his postgraduate student Gary Stewart and was first
marketed in 1965 by the Cambridge Instrument Company as the “Stereoscan”.
The signals produced by an SEM include secondary and back-scattered
electrons (SE&BSE), characteristic X-rays, cathodoluminescence (light), specimen
current and transmitted electrons. SE detector is the most common one in SEM. As
shown by the schematic Figure 2.3, in an FEI /Philips XL-30 FEG-SEM (Field
Emission Gun) of the Applied Physics – Materials Science group electrons are
generated by the field emission gun using a high electrostatic field. They are
accelerated with energies between 1 keV and 30 keV down through the column
towards the specimen. While the magnetic lenses (condenser and objective lenses)
focus the electron beam to a spot with a diameter of approximately 10 nm, the
scanning coils sweep the focused electron beam over the specimen surface. If the
microscope is operated in the backscattered mode, the result is a lateral resolution
on the order of micrometers. The number of back-scattered electrons produced is
proportional to the atomic number of the element bombarded. The result is that
material with a high(er) atomic number produces a brighter image. To capture this
information a detector is required which can either be metal, which is the least
effective, but is versatile and used in environmental scanning electron microscopy
(ESEM); semi-conductor, which is most common or a scintilator/light
pipe/photomultiplier, which are the most efficient.
Experimental methods 27
Figure 2.3 – Schematic view of a Scanning Electron Microscope
The primary electrons current is approximately 10-8 to 10-7A. The large
penetration depth of the high energy electrons will cause the electrons to be
trapped in the material. When studying conducting materials, the electrons will be
transported away from the point of incidence. If the specimen is a non-conducting
material, the excess electrons will cause charging of the surface. The electrostatic
charge on the surface deflects the incoming electrons, giving rise to distortion of
the image. To reduce surface charging effects, a conducting layer of metal, with
typical thicknesses 5-10 nm, can be sputtered onto the surface. This layer will
transport the excess electrons, reducing the negative charging effects. An adverse
effect of the sputtered layer is that it may diminish the resolving power of the
microscope, since topographical information is no longer gained from the surface
of the material, but from the sputtered layer. Charging of the surface is not the only
factor determining the resolution of a scanning electron microscope. The width of
the electron beam is also an important factor for the lateral resolution. A narrow
electron beam results in a high resolution. The spot size however, is a function of
the accelerating voltage
28 Chapter 2
2 2
2 2 2 6
20
1 1
2p c s
i Ed C C
B E
. (2.18)
The broadening of the spot size is the sum of broadening effects due to several
processes. The first contributor is the beam itself, B is the brightness of the
source, i is the beam current and its divergence angle. The second part is the
contribution due to diffraction of the electrons of wavelength by the size of the
final aperture. The last two parts are the broadening caused by chromatic and
spherical aberrations. Where 0E is the electron energy and E is the energy
spread, sC represents the spherical aberration and cC is the chromatic aberration
coefficient. To achieve the smallest spot size, all contributions should be as small as
possible. Decreasing the accelerating voltage will not only cause the wavelength of
the electrons to increase, but also the chromatic aberration increases as well,
resulting in increasing of the spot size and, as a consequence, a decrease in
resolving power of the microscope.
A field emission gun has a very high brightness B , reducing the contribution
in broadening due to the beam itself. The energy spread E in the electron
energies is also small. Together with the fact that the coefficients sC and cC can be
reduced by optimizing the lenses for low-energy electrons, provides the FEG low
voltage scanning electron microscope with very high resolving power.
2.7 Focused Ion Beam
The effects of high-energy particles striking a material are a major concern in
the nuclear materials industry, where reactor housings and the components used
within have to be designed with consideration of the influence of alpha, beta,
gamma and neutron radiation. Such effects can be extremely detrimental to the
operation of a reactor. Amongst them are, for example, the formation of voids
within reactor walls, the embrittlement of affected surfaces, the formation of
secondary phases or the formation of entirely new materials due to nuclear effects
which can lead to the reduction of functionality of critical components.
High-energy particles have a beneficial use, however. The modern focused ion
beam (FIB) instrument has a wide variety of applications in many fields [47], and is
a versatile tool for sample preparation for scanning and transmission electron
Experimental methods 29
microscopy. The FIB operates through either deposition or sputtering of material to
perform topological alterations to a sample. In both cases, ions of a particular
species, usually Ga+, He+, Ne+ or Ar+ are excited to a high energy, typically 30 keV,
and accelerated towards their target. An incident ion imparts a large amount of
energy to the target surface and it results in a series of effects: sputtering of
material i.e. the release of the sample’s surface atoms, ion implantation and
formation of defects along the ion’s trajectory within the material until it reaches a
stopping point.
The sputtering effect is the most common and traditional use for the FIB,
where material can be precisely and quickly removed. However, a high-energy
particle emitted from a FIB entering a material will also transfer its energy to
adjacent particles through inelastic collisions. These particles will do the same to
their own neighbors in a process of so-called “knock-on” effects. Whilst some
particles displaced in this manner will recover to their initial locations in the
atomic lattice, others will remain displaced as interstitial atoms, leaving behind a
vacancy in a defect type called a vacancy-interstitial or Frenkel pair. In this
manner, a single high-energy particle traveling through a material will create a
large amount of defects, and continuous exposure to a beam of high-energy
particles will create a defect-rich layer in a material that is approximately as deep
as the maximum penetration depth of the accelerated ions.
In this Thesis work we have intensively used a Focused Ion Beam integrated
into a dual beam FEG scanning electron microscope. The typical function of an ion
beam apparatus is either the removal or the addition of material - depending on the
nature of ions and their acceleration energy - through bombarding a target with
accelerated ions. Incoming ions typically have much more energy than the bond
energy of the atoms of their target and, upon impact, break a large amount of
atomic bonds as they move into the material. The release of atoms as they break
free from the source material due to this imparted energy is called sputtering, and
the use of this process to selectively change the shape of an object using the ion
beam is called milling.
The dual beam FIB/scanning electron microscope (SEM) microscope (Lyra,
Tescan, CZ) apparatus was used in all ion-beam related experiments. The
apparatus produces a Ga ion beam with an acceleration voltage of 30 keV. The ion
beam’s spot size and ion current can be varied. During all experiments, the spot
30 Chapter 2
size remained unchanged. The ion current operates on factory presets, with settings
for 1, 10, 40, 150, 200, 1000 and 10000 pA. True ion current is measured internally
using a Faraday cup, and it is consistently observed that the true ion current varies
from that of the setting at which that current is produced. The variation can be up
to a factor of two above or below the preset value. While this variation is small on
the absolute scale at low ion current values, it quickly becomes significant as ion
current increases.
The FIB apparatus is equipped with a computer-controlled moving stage that
allows for 360 degree rotation as well as tilt that ranges from –20 to +70 degrees.
Through a combination of rotation and tilt a sample can be observed from any
angle (for schematics see Figure 2.3). There is a 55 degree angle between the
electron and ion emitters, such that a sample being observed from a perpendicular
inclination with the SEM will be 55 degrees tilted away from the normal when
observed with the ion beam and vice-versa. As such, it is customary for a sample to
be tilted 55 degrees once the working area has been found.
The FIB apparatus is capable of masking its ion beam during emission to
allow for a variety of milling types – random and polish. During random milling a
selected area is continuously bombarded with ions with each ion striking a random
point within the area. During polish milling the ion beam removes material a row
of ion “pixels” at a time, which often ensures a far cleaner edge than random
milling.
Figure 2.4 shows a general schematic of the steps involved in the preparation
and irradiation of bending cantilevers. The details for fabrication of the initial free-
standing thin films is described in the next section. The surroundings of the regions
of interest were ion cut and removed, leaving arrays of ~ 5×2 µm2 cantilevers. A
rectangular area that slightly exceeded the cantilever edges was scanned using a
parallel strategy, i.e. the ion beam scanned the selected areas sequentially in lines,
repeating for the necessary number of times until the desired fluence was achieved.
Experimental methods 31
Figure 2.4 – Schematic diagram of cantilevers fabrication and bending: (a)
free-standing film; (b) ion cutting of surrounding areas; (c) linear irradiation
(over the dashed lines) to remove adjacent areas from the line of sight; (d)
cutting of the cantilevers; (e) bending experiment: area scan up to desired
fluence and subsequent SE imaging; (f) y-axis is used for modeling defect
kinetics in the cantilever as a function of depth (chapters 5 and 6), (x, z) is the
larger-scale coordinates describing cantilever curvature.
The experimental fluence was calculated as the average number of ions
over the scanned area according to the formula
22 1
dwell
I abjt t
abe r o
, (2.19)
where j is the averaged flux (ion rate per area considering single charged ions), t is
the irradiation time considering the number of scanning spots in the irradiated
area, I is the ion current, e- is the elementary charge, a and b are the dimensions of
the scanned rectangle, r is the ion beam radius, o is the scanning beam overlap
ratio and dwellt is the dwell time in each scan spot. The values dwellt = 1 µs, r = 50
nm and o = 0.5 were kept fixed while those of I and ab varied between experiments.
After irradiation to a desired fluence, the cantilevers were imaged with the
SEM. Steps of irradiation and SE imaging proceeded until the cantilevers were
32 Chapter 2
almost vertical. The sequential images were measured with ImageJ [48] and the
vertical dimensions of the images were adjusted by a factor of 1/sin(55º) = 1.2208
due to the 55º perspective of the electron beam to the samples normal (Figure
2.5a). The deflections corresponding to each fluence step were finally measured
using Engauge [49] to extract the cantilevers coordinates. Each measurement
corresponds to the average of the coordinates of at least three simultaneously
irradiated cantilevers (Figure 2.5b).
Figure 2.5 – Diagram of the required correction for the cantilevers
displacement due to the relative positions of the electron and ion beams (a); and
example of measurement of displacement from SE images (b).
2.8 Free-standing thin films
In this Thesis, most of the work was done on free-standing metallic films with
sub-micrometer thickness. As counterintuitive as it may appear, the (nano-)
technology to achieve such small thicknesses has been available since ancient times
and artifacts found in Egypt dating to as back as 2690 BC show signs of having
been covered in gold leaves [50], [51]. The method employed to produce such
leaves, i.e. repeatedly beating a stack of intercalated gold and paper or gut layers;
was discussed by Pliny the Elder in the first century and is still in use today. In fact
the precursor alloy films for the synthesis of nanoporous gold (section 2.8.2), are
commercialized mainly for gilding and are still made via that fabrication route.
Experimental methods 33
Present-time students could probably find some useful advice about handling these
leaves in the colorful description of a 12th century Benedictine monk [50] :
"In laying on the gold take glair, which is beaten out of the white of an egg
without water, and with it lightly cover with a brush the place where the gold is to
be laid. Wet the point of the handle of the brush in your mouth, touch a corner of
the leaf that you have cut, and so lift it up and apply it with the greatest speed. Then
smooth it with a brush. At this moment you should guard against drafts and hold
your breath, because if you breathe, you will lose the leaf and find it again only with
difficulty."
The development of the transmission electron microscope stirred new interest
in thin specimen preparation techniques and the fast development of thin film
technologies since then led to breakthroughs in areas as diverse as energy
harvesting and storage, electronic semiconductor devices, corrosion or abrasion
protective coatings and others, making it a significant subfield of Materials Science.
Several techniques exist for the synthesis of thin films and they can be generally
divided in top-bottom and bottom-up approaches. The top-bottom techniques
involve thinning of bulk material, typically by chemical, electrochemical or
mechanical means. The bottom-up approach means building a thin deposit by e.g.
electrodeposition, sputtering or vacuum evaporation, which was the technique of
choice in the present work.
2.8.1 Dense solid bulk films
Thin films of gold, platinum and aluminum were synthetized via physical
vapor deposition using a Temescal FC-2000 electron-beam evaporator. In such a
machine, an electron beam is used to heat and evaporate a solid material source
inside a high vacuum chamber. The evaporated atoms condense on the chamber’s
walls and the deposited thickness can be monitored by means of a quartz crystal
microbalance.
Free-standing films were produced by evaporating the aforementioned metals
on mechanically polished NaCl substrates that were posteriorly dissolved in
distilled water. The pressure in the deposition chamber was kept at 8×10−7 Torr
and an evaporation rate of 0.1 nm s−1 was used for all samples. After dissolving the
34 Chapter 2
rocksalt substrates, the floating films were collected with TEM grids, rinsed with
distilled water and allowed to dry in air. The grids were finally mounted in a home-
designed holder that enables the grid surface to be positioned normally to the
incident ion beam in the Tescan Lyra FIB-SEM dual system.
2.8.2 Nanoporous films
“Separate thou the earth from the fire, the subtle from the gross sweetly with
great indoustry…” [52]
The process of dealloying the less noble component from a metallic alloy does
not lack its own historical curiosities. Andean pre-colombian metalsmiths already
knew the process of depletion gilding, in which the surface of a gold-alloy artifact
would be treated to give it the appearance of being made of pure gold – much for
the disappointment of Spanish looters [53]. What the Andeans did not know (or so
we assume!) was that the dealloyed morphology consisted of an open, bicontinuous
nanoporous gold structure, as shown by Forty [54].
The precursor materials used for the present studies were commercial AuAg
foils (Noris Blattgold GmbH, Schwabach, Germany) with thickness between 100
and 200 nm. The formation of nanoporous gold by free dealloying of these foils was
investigated in our MK group using transmission electron backscatter diffraction
(t-EBSD). This rather recent technique is based on the collection of Kikuchi
patterns formed by transmitted electrons in a SEM and has improved resolution
compared to standard EBSD [55], [56].
The AuAg foils microstructure consists of a matrix of large whirlpool-shaped
grains containing clusters of elongated smaller grains (Figure 2.6). The initial
<001> and <111> fiber texture was found to be enhanced after etching, while the
residual strain measured by the average misorientation inside grains was reduced.
Additionally, the formation of low-angle grain and twin boundaries at the expense
of high-angle grain boundaries hinted to grain boundaries as nucleation points for
the etching process [57].
Experimental methods 35
Figure 2.6 – The [001] inverse pole figure maps of AuAg foil before (left) and
after (right) dealloying. Black lines represent grain boundaries. The arrows
denote locations with a changed grain boundary. From ref. [57]
Ligament and pore sizes of the nanoporous structure can be tuned by
adjusting the dealloying potential during electrochemical etching, as well as by acid
or thermal post-treatments [58]. The free-standing thin films of nanoporous gold
used for ion-induced bending experiments (Chapter 6) were synthetized by free
etching on 65% nitric acid solution. Foils with composition of Au35Ag65 and
Au25Ag75 were dealloyed for times between 15 minutes and 1 day in order to achieve
various ligament and pore sizes. The dealloyed samples were collected from the the
acid solution with gold TEM grids and thoroughly rinsed with distilled water before
drying in air. Thermal post-treatments for further coarsening were performed at up
to 400ºC under Ar atmosphere. The characteristic size of the nanoporous
structures was measured using the rotationally averaged FFT method [59] using
either SEM micrographs or bright field TEM images.
2.9 References
[1] J. M. Culkin, A schoolman’s guide to Marshall McLuhan. Saturday Review,
Incorporated, 1967.
36 Chapter 2
[2] L. D. Broglie, ‘Recherches sur la théorie des Quanta’, Annales de Physique, vol. 10, no.
3, pp. 22–128, 1925.
[3] H. Busch, ‘Berechnung der Bahn von Kathodenstrahlen im axialsymmetrischen
elektromagnetischen Felde’, Annalen der Physik, vol. 386, no. 25, pp. 974–993, 1926.
[4] M. Haider, S. Uhlemann, E. Schwan, H. Rose, B. Kabius, and K. Urban, ‘Electron
microscopy image enhanced’, Nature, vol. 392, no. 6678, pp. 768–769, Apr. 1998.
[5] B. Kabius, M. Haider, S. Uhlemann, E. Schwan, K. an. R. Urban, and H., ‘First
application of a spherical-aberration corrected transmission electron microscope in
materials science’, Journal of Electron Microscopy, vol. 51, pp. 51–58, 2002.
[6] S. Mantl, Q. T. Zhao, and B. Kabius, ‘Self-Assembled Patterning of Ultrathin Silicides by
Local Oxidation’, MRS Bulletin, vol. 24, no. 8, pp. 31–35, Aug. 1999.
[7] M. Lentzen, B. Jahnen, C. L. Jia, A. Thust, K. Tillmann, and K. Urban, ‘High-resolution
imaging with an aberration-corrected transmission electron microscope’,
Ultramicroscopy, vol. 92, no. 3, pp. 233–242, Aug. 2002.
[8] N. Dellby, L. Krivanek, D. Nellist, E. Batson, and R. Lupini, ‘Progress in aberration-
corrected scanning transmission electron microscopy’, Journal of Electron Microscopy,
vol. 50, no. 3, pp. 177–185, May 2001.
[9] K. van Benthem, A. R. Lupini, M. P. Oxley, S. D. Findlay, L. J. Allen, and S. J.
Pennycook, ‘Three-dimensional ADF imaging of individual atoms by through-focal
series scanning transmission electron microscopy’, Ultramicroscopy, vol. 106, no. 11,
pp. 1062–1068, Oct. 2006.
[10] F. Houdellier, B. Warot-Fonrose, M. J. Hÿ, CEMES-CNRS, and France, ‘New Electron
Energy Loss Magnetic Chiral Dichroïsm (EMCD) configuration using an aberration-
corrected transmission electron microscope’, Microscopy and Microanalysis, vol. 13,
no. S03, pp. 48–49, Sep. 2007.
[11] A. Kirkland, L.-Y. Chang, S. Haigh, and C. Hetherington, ‘Transmission electron
microscopy without aberrations: Applications to materials science’, Current Applied
Physics, vol. 8, no. 3, pp. 425–428, May 2008.
[12] B. Kabius et al., ‘First application of Cc-corrected imaging for high-resolution and
energy-filtered TEM’, J Electron Microsc (Tokyo), vol. 58, no. 3, pp. 147–155, Jun.
2009.
[13] A. Berger, J. Mayer, and H. Kohl, ‘Detection limits in elemental distribution images
produced by energy filtering TEM: case study of grain boundaries in Si3N4’,
Ultramicroscopy, vol. 55, no. 1, pp. 101–112, Jul. 1994.
[14] J. M. Zuo, M. Kim, M. O’Keeffe, and J. C. H. Spence, ‘Direct observation of d-orbital
holes and Cu–Cu bonding in Cu2O’, Nature, vol. 401, no. 6748, pp. 49–52, Sep. 1999.
[15] R. Erni, Aberration-corrected Imaging in Transmission Electron Microscopy: An
Introduction. London: Imperial College Press, 2010.
Experimental methods 37
[16] J. T. M. D. Hosson, ‘In situ Transmission Electron Microscopy on Metals’, in Handbook
of Nanoscopy, Wiley-Blackwell, 2012, pp. 1099–1151.
[17] J. T. M. Hosson and A. Van Veen, ‘Nanoprecipitates and nanocavities in functional
materials’, in Encyclopedia of Nanoscience and Nanotechnology, vol. 7, 349 vols,
American Scientific Publishers, 2004, pp. 297–349.
[18] H. B. Groen, ‘Interface dislocation patterns studied with high-resolution TEM’,
University of Groningen, 1999.
[19] P. M. C. C. A. de Carvalho, ‘Planar defects in ordered systems’, University of Groningen,
2001.
[20] S. Mogck, ‘Transmission electron microscopy studies of interfaces in multi-component
systems’, University of Groningen, 2004.
[21] W. A. Soer, ‘Interactions between dislocations and grain boundaries’, University of
Groningen, 2006.
[22] Z. Chen, ‘Superplasticity of coarse grained aluminum alloys’, University of Groningen,
2010.
[23] S. Venkatesan, ‘Structural domains in thin films of ferroelectrics and multiferroics’,
University of Groningen, 2010.
[24] S. Punzhin, ‘Performance of ordered and disordered nanoporous metals’, University of
Groningen, 2013.
[25] M. Dutka, ‘Processing and structure of nanoparticles : characterization and modeling’,
University of Groningen, 2014.
[26] P. Buseck, J. Cowley, and L. Eyring, High-Resolution Transmission Electron
Microscopy: and Associated Techniques. Oxford University Press, 1989.
[27] J. T. M. D. Hosson, D. Schryvers, and G. V. Tandeloo, ‘Metals and Alloys’, in Handbook
of Microscopy, Wiley-Blackwell, 2008, pp. 5–110.
[28] J. C. H. Spence, Experimental high-resolution electron microscopy. Oxford University
Press, 1988.
[29] D. B. Williams and C. B. Carter, Transmission Electron Microscopy: A Textbook for
Materials Science, 2nd ed. Springer US, 2009.
[30] Z. Wang, Elastic and Inelastic Scattering in Electron Diffraction and Imaging.
Springer US, 1995.
[31] R. F. Egerton, Physical Principles of Electron Microscopy: An Introduction to TEM,
SEM, and AEM, 2nd ed. Springer International Publishing, 2016.
[32] R. E. Burgess, H. Kroemer, and J. M. Houston, ‘Corrected Values of Fowler-Nordheim
Field Emission Functions $v(y)$ and $s(y)$’, Phys. Rev., vol. 90, no. 4, pp. 515–515,
May 1953.
[33] L. Reimer, Image Formation in Low-voltage Scanning Electron Microscopy. SPIE
Press, 1993.
38 Chapter 2
[34] F. Zernike, ‘Das Phasenkontrastverfahren bei der mikroskopischen Beobachtung’,
Zeitschrift für technische Physik, vol. 16, no. 454, 1935.
[35] M. A. O’Keefe, ‘“Resolution” in high-resolution electron microscopy’, Ultramicroscopy,
vol. 47, no. 1, pp. 282–297, Nov. 1992.
[36] O. Scherzer, ‘The Theoretical Resolution Limit of the Electron Microscope’, Journal of
Applied Physics, vol. 20, no. 1, pp. 20–29, Jan. 1949.
[37] M. Haider, H. Rose, S. Uhlemann, E. Schwan, B. Kabius, and K. Urban, ‘A spherical-
aberration-corrected 200kV transmission electron microscope’, Ultramicroscopy, vol.
75, no. 1, pp. 53–60, Oct. 1998.
[38] M. Wilkens and M. Rühle, ‘Black–white contrast figures from small dislocation loops. I.
Analytical first order perturbation solution’, physica status solidi (b), vol. 49, no. 2, pp.
749–760, 1972.
[39] F. Häussermann, M. Rühle, and M. Wilkens, ‘Black-white contrast figures from small
dislocation loops II. Application of the first order solution to small loops in ion-
irradiated tungsten foils’, physica status solidi (b), vol. 50, no. 2, pp. 445–457, 1973.
[40] F. Häussermann, K. H. Katerbau, M. Rühle, and M. Wilkens, ‘Calculations and
observations of the weak-beam contrast of small lattice defects’, Journal of Microscopy,
vol. 98, no. 2, pp. 135–154, 1973.
[41] D. J. H. Cockayne, ‘A Theoretical Analysis of the Weak-beam Method of Electron
Microscopy’, Zeitschrift für Naturforschung A, vol. 27, no. 3, pp. 452–460, 1972.
[42] D. J. H. Cockayne, I. L. F. Ray, and M. J. Whelan, ‘Investigations of dislocation strain
fields using weak beams’, The Philosophical Magazine: A Journal of Theoretical
Experimental and Applied Physics, vol. 20, no. 168, pp. 1265–1270, Dec. 1969.
[43] P. B. Hirsch, Electron Microscopy of Thin Crystals. R. E. Krieger Publishing Company,
1977.
[44] EDAX Phoenix software, TEM Quant Materials. Mahwah (NJ): EDAX Inc., 2000.
[45] M. Knoll, ‘Aufladepotentiel und sekundäremission elektronenbestrahlter körper’,
Zeitschrift für technische Physik, vol. 16, pp. 467–475, 1935.
[46] M. von Ardenne, ‘Das Elektronen-Rastermikroskop’, Z. Physik, vol. 109, no. 9–10, pp.
553–572, Sep. 1938.
[47] S. Reyntjens and R. Puers, ‘A review of focused ion beam applications in microsystem
technology’, J. Micromech. Microeng., vol. 11, no. 4, p. 287, 2001.
[48] C. A. Schneider, W. S. Rasband, and K. W. Eliceiri, ‘NIH Image to ImageJ: 25 years of
image analysis’, Nat Meth, vol. 9, no. 7, pp. 671–675, Jul. 2012.
[49] M. Mitchell, B. Muftakhidinov, T. Winchen, Z. Jędrzejewski-Szmek, and T. G. Badger,
‘engauge-digitizer: Version 9.2 Bug fix release’, Zenodo.
[50] E. G. Thomsen and H. H. Thomsen, ‘Hammer forging of thin sections in antiquity’, J.
Applied Metalworking, vol. 1, no. 1, pp. 50–58, Jul. 1978.
Experimental methods 39
[51] E. D. Nicholson, ‘The ancient craft of gold beating’, Gold Bull, vol. 12, no. 4, pp. 161–
166, Dec. 1979.
[52] I. Newton, ‘Keynes MS. 28’, The Chymistry of Isaac Newton, Jun. 2010.
[53] H. Lechtman, ‘Pre-Columbian Surface Metallurgy’, Scientific American, vol. 250, no. 6,
pp. 56–63, Jun. 1984.
[54] A. J. Forty, ‘Corrosion micromorphology of noble metal alloys and depletion gilding’,
Nature, vol. 282, no. 5739, pp. 597–598, Dec. 1979.
[55] R. r. Keller and R. h. Geiss, ‘Transmission EBSD from 10 nm domains in a scanning
electron microscope’, Journal of Microscopy, vol. 245, no. 3, pp. 245–251, Mar. 2012.
[56] R. van Bremen, D. Ribas Gomes, L. T. H. de Jeer, V. Ocelík, and J. T. M. De Hosson, ‘On
the optimum resolution of transmission-electron backscattered diffraction (t-EBSD)’,
Ultramicroscopy, vol. 160, pp. 256–264, Jan. 2016.
[57] L. T. H. de Jeer, D. R. Gomes, J. E. Nijholt, R. van Bremen, V. Ocelík, and J. T. M. D.
Hosson, ‘Formation of Nanoporous Gold Studied by Transmission Electron Backscatter
Diffraction’, Microscopy and Microanalysis, vol. 21, no. 6, pp. 1387–1397, Dec. 2015.
[58] E. Detsi, M. van de Schootbrugge, S. Punzhin, P. R. Onck, and J. T. M. De Hosson, ‘On
tuning the morphology of nanoporous gold’, Scripta Materialia, vol. 64, no. 4, pp. 319–
322, Feb. 2011.
[59] T. Fujita and M. W. Chen, ‘Characteristic Length Scale of Bicontinuous Nanoporous
Structure by Fast Fourier Transform’, Jpn. J. Appl. Phys., vol. 47, no. 2R, p. 1161, Feb.
2008.
Electrochemical fabrication of nanoporous materials
This chapter describes exploratory efforts aimed at the fabrication of high
surface area actuators by means of electrochemical methods. A structure was
idealized which consisted of stacked layers of 1. aligned nanowires and 2. solid
metal, inspired by reports of improved performance in literature. The aligned
nanowire arrays would be produced by potentiostatic electrodeposition of metals
into the nanopores of an aluminium oxide template that could be selectively
dissolved. As preliminary results indicated that commercial membranes did not
match that purpose, the focus shifted to the production of adequate home-made
templates. Although the idealized structure could not be obtained, this chapter
describes some potentially interesting observations made along the way.
3.1 Actuators
A specific problem in the design of any machine (micromachines included) is
that of moving and controlling its mobile parts. The component with that function
could well be called a "mover" but the technical community surely looks smarter
calling it an "actuator". In fact, if an external stimulus (say potential difference) is
not needed we call the system ‘a sensor’; when an external source is still applied we
call it an ‘actuator’. In several cases, nanoporous materials systems may respond
both as actuator and sensor [1]–[3].
42 Chapter 3
Leaving aside pneumatic or hydraulic actuators and focusing now on
microsized elements, the most common type of actuators are piezoelectric
materials. The piezoelectric effect consists in the generation of charge as result of a
mechanical strain and vice-versa. It has numerous applications from day-to-day
devices such as cigarette lighters and wristwatches to scientific uses where ultrafine
motion is required such as in the alignment of optical assemblies or in scanning
probe microscopy. Since the typical piezoactuator delivers ~0.2% strain at a
relatively high potential of 150V [4], an alternative that delivered comparable or
higher strains at lower potentials would be desirable for the use in miniaturized
devices such as MEMS. One recent development that showed particular aptitude to
accomplish that goal was the use of nanoporous metallic materials as actuators.
Contrary to piezoelectric materials where the charge-strain relation stems from the
occurence of electric dipole moments, dimensional change in metallic nanofoams is
achieved by the redistribution of charge at its surfaces. Since the surface-to-volume
ratio in these materials is extremely high, a change in the interatomic distance at
the surface (due to, for example, adsorption of charged species) is enough to
produce macroscopic displacements.
In our research group, previous research successfully increased the strain of
nanoporous metallic actuators at the time by roughly two orders of magnitude [5].
By means of a clever manipulation of the precursor alloy microstructure via cold-
rolling, a nanoporous material with dual lenght scale was synthesized in which
actuation was amplified by buckling of partially delaminated grains. Another
example of optimization of the actuation strain by manipulating its architecture
instead of relying in intrinsic material properties was demonstrated in ref. [6]. In
that study, an increase of 20% in relation to monolithic np-Au was achieved by
fabricating multilayer solid/nanoporous composites. Since the actuating
nanoporous layers were laterally confined by the solid layers, strain was increased
in the normal direction by the Poisson effect. Yet another strategy in tailoring
nanoporous actuators architecture involves facilitating ion transportation, as high
strains require short charging rates [7]. A semiordered hierarchical nanoporous
structure with µm-sized tubes consisting of nm-sized ligaments was shown to
produce remarkable strains that were attributed to the fast ion transfer kinectics.
Inspired by these concepts, work has been put into the pursuit of a structure
that would combine (1) the high surface to volume ratio of actuating nanoporous
Electrochemical nanofabrication 43
metals, (2) the concept of intercalated solid layers improving the actuation strain
and mechanical robustness and (3) linear (non-tortuous) channels that allow easy
diffusion of charge-transporters. The following sections present some of the
exploratory efforts made into the fabrication of such a structure by means of
electrodeposition of metals in anodized aluminium oxide templates.
3.2 Electrochemical nanofabrication
3.2.1 Anodized aluminum oxide (AAO) templates
Pure aluminium metal is a relatively recent tool in mankind box, having been
first isolated less than two centuries ago by Danish scientist Hans Christian Ørsted.
A “symbol of the future” during the nineteenth century, the metal was showcased in
world fairs at the time and Napoleon III of France would serve his honoured guests
in aluminium plates and utensils, while the less noble guests had to manage with
gold and silver [8]. As the processes of mass-production of aluminium evolved and
its price dropped, it quickly found uses in a widespread range of day-to-day
applications. Besides the metal high abundancy and low density which made it an
important part of modern aerospace, transportation and building industries;
aluminium alloys presents good resistance to corrosion thanks to a naturally
forming passive oxide layer. The ability to electrochemically oxidize aluminium
parts to form thicker layers of protective oxide was called “anodization” and was
applied in industrial scales already in the 1920s to improve the corrosion resistance
of seaplane parts.
Depending on the solubility of the formed oxide in the anodization electrolyte,
the resulting coating may be either of barrier or porous type (Figure 3.1). The
porous-type anodization received considerable attention over the last century due
to the fact that the high porosity provided excellent base for decorative coloration
in consumer products. Additionally, while the thickness of barrier-type oxides is
controlled by the anodization potential, porous-type oxide thickness can be
controlled by the anodization type as it is linearly proportional to the total charge.
It wasn’t until 1995, however, that it became a hot topic in nanotechnology with the
discovery that a tunable self-ordered nanoporous structure could be obtained by
44 Chapter 3
performing two-step anodization under the right electrolytes and parameter ranges
[9], [10]. These ordered nanoporous films were the starting point of the efforts
described below for the production of an actuating metallic nanowire structure. A
detailed review on the various aspects of the AAO production is given in ref. [11].
Figure 3.1 – The two types of anodized aluminium oxide: barrier-type (a)
and porous-type (b). Adapted from [11].
3.2.2 Templated electrodeposition of metals
In general, electrodeposition consists in the reduction of metallic species
present in a given electrolyte on the surface of a conductive substrate by means of
an applied electric current [12]. When using an open-through AAO membrane as
template for electrodeposition, one of its sides is typically covered by a conductive
film that acts as the working electrode while the other side is exposed to the
electrolyte so that electrodeposition takes place inside the pores. Four typical stages
can be defined, as depicted in Figure 3.2 [13], [14]. When a potential is applied
(stage I), the current density increases sharply and decreases with time as the
diffusion layer increases from zero to the thickness of the membrane, according to
the so-called Cotrell equation for diffusion-controlled redox reactions. As the pore
filling (stage II) proceeds, the current density stays relatively constant. When the
wires reach the top surface of the membrane (stage III), the increase in electrode
area leads to an increase in current and finally a continuous film is formed (stage
IV) and its thickness continues to grow at constant current.
Electrochemical nanofabrication 45
Figure 3.2 – Typical stages of metal electrodeposition in AAO templates (a) and
associated behavior of current density in time for the potentiostatic case (b).
Adapted from ref. [13] and [15].
3.3 Experimental results and discussion
3.3.1 Commercial membranes
The first attempts at producing nanowire arrays made use of commercial
Anodisc Whatman AAO filter membranes (Figure 3.3). These are available as 13
mm diameter discs with approximately 65 µm thickness. Two pore sizes were
selected for the preliminary experiments, i.e. 20 and 100 nm. The original
justification was that smaller pores would be advantageous considering the highest
specific surface of the final deposited structures while the bigger pore size was
46 Chapter 3
selected for comparison. These membranes quickly proved unfit, however, due to
the fact that only a small fraction of their thickness is composed by pores with the
nominal sizes. As the membranes are intended for filtration, most of their thickness
is composed by ~200 nm pores making them useful only for testing purposes.
Figure 3.3 – Commercial AAO membranes. SEM micrographs of the top surface
and cross-section of 100-nm membrane (a-b); top-view of 20-nm membrane
(c). The different pore sizes are visible through broken regions in (a) and (c).
The scale bar is 1 µm in all images.
3.3.2 Electrochemical deposition of nanowires
The electrode design for deposition inside the pores of the membranes was
inspired by the protocol provided in reference [16]. A layer of gold was evaporated
Electrochemical nanofabrication 47
onto the the filtering side of the membranes and on one side of small glass slides
(Figure 3.4a-b). The coated membranes were attached to the glass slides and kept
in place using small strips of adhesive tape. A copper tape was connected to the
coated region of the glass slide for electrical contact (Figure 3.4c) and the whole
part except for the membrane was covered with nail polish for insulation
(Figure 3.4d, after successful electrodeposition of copper).
Figure 3.4 – Preparation of working electrodes for templated metal deposition.
AAO membranes and glass slides are coated with a thin layer of gold (a-b);
membranes are assembled onto the slides and secured with conductive copper
tape (c); electrode is covered with nail polish leaving only the middle part of the
membranes exposed (d).
Electrochemical deposition of copper, nickel and magnesium into the
templates was performed using a Metronohm Autolab potentiostat and a three-
electrode setup consisting of the aforementioned working electrodes, a platinum
counter-electrode and a Ag/AgCl reference electrode. The deposition baths and
potential were selected from refs. [17], [18] and are listed in Table 3.1.
48 Chapter 3
Table 3.1 – Electrodeposition parameters
NW metal Deposition bath Deposition conditions (DC)
Cu 0.4 M CuSO4 +
0.2 M H3BO3 in water
0.15 V
Ni 0.1 M NiSO4 +
0.2 M H3BO3 in water
1 V
Mg 0.4M (PhMgCl)2-AlCl3 in THF 40 mA cm-2
Figure 3.5 shows an example of a successfully fabricated solid-nanowire-solid
structure, after dissolution of the AAO template by immersion in 1 M NaOH at
room temperature. The nanowires were composed of pure copper as confirmed by
EDS. These structures were robust and easy to handle and the bottom Au layer was
intentionally damaged to allow observation of the middle layer. Figure 3.5a shows a
wide view of the damaged Au layer with the exposed inner nanowire array. The
spherical particles that can be seen on top of the nanowires are an indication of the
early stages of the overfill of the pores before a continuous film is formed on the
surface of the AAO template [19]. Nanowires that remained attached to the top
layer can be seen in the bottom-right corner of Figure 3.5b.
These preliminary results indicated the feasibility of fabricating mechanically
stable, high-surface structures with a relatively uncomplicated setup. It should be
noted, however, that although the deposition of nanowires was shown to be
achievable, the I-V behavior of these cathodes during deposition was not
satisfactorily reproducible. This issue was attributed mainly to the difficulty in
assuring a good contact between the glass slides and the membranes due to the
fragility of the latter; and the imperfect covering of the membranes pores by the
evaporated gold layer. This probably caused the electrolyte to leak in between the
slide and the membrane leading to deposition in that space. Consequently,
incomplete or inconsistent pore filling (Figure 3.6b, described ahead) and general
deviations from the expected characteristic stages were often observed. The
difficulty to achieve reproducible behavior has been mentioned in the literature and
attributed to the current distribution through the cell [14], the statistical nature of
nucleation at the ealier deposition stages as well as to the blocking of pores by
residual air and formed hydrogen bubbles [15].
Electrochemical nanofabrication 49
Figure 3.5 – SEM micrographs of electrodeposited Cu nanowire array. The top
layer was intentionally damaged to allow for observation of the inner nanowire
layer (a-b). A bundle of broken nanowires lay upon the dense array of
nanowires oriented parallel to the viewing direction (c), with zoomed-in details
(d-e). Various views of the exposed nanowires (f-h).
3.3.3 The goal-structure
Bearing in mind the objective of a multilayered structure consisting of
multiple layers of solid and nanowire-arrays, a rather practical problem appears.
While the length of the nanowire-array layer is defined by the thickness of the AAO
template, control of the solid layers thicknesses is not so straightforward.
Copper electrodeposition was performed in an electrode like the one described
in the previous section, but using two stacked membranes instead of a single one.
The resulting structure after dissolution of the template can be seen in Figure 3.6.
50 Chapter 3
Due to inhomogeneous pore filling of the bottom membrane, filling of the space in
between membranes and of the top membrane pores was incomplete and therefore
all these layers are visible in Figure 3.6b. The solid layer that was formed in the
space between the membranes had approximately 3 µm thickness but a method to
tune it isn’t immediately obvious (Figure 3.6a).
Figure 3.6 – SEM micrographs of multilayered structure after deposition in two
stacked templates. The solid layer in between the membranes formed a solid
layer with thickness of aproximately 3 µm (a). Top view shows inconsistent
filling of the pores (b).
Since the fabrication of our own membranes seemed unavoidable considering
the large pores and thickness from the commercial ones, the strategy to tackle the
Electrochemical nanofabrication 51
problem of controlling the thickness of the solid layers was inspired by an approach
to AAO fabrication named “pulse anodization” [20]. When anodization is
performed at different potential regimes, the chemical composition of the formed
oxide layers can be tailored allowing for selective dissolution. If a supporting
material is infiltrated in some areas of the AAO multilayers prior to dissolution, a
template could be achieved for electrodeposition of the aimed solid/nanowire-
arrays with tunable thicknesses (Figure 3.7). Not only this process would allow for
the control of the solid layers thickness, but the possibility to fabricate thin
nanowire templates would be highly desirable. The reduction in aspect ratio of the
nanowires would likely benefit both mechanical stability and pore filling uniformity
during electrodeposition.
Figure 3.7 – SEM micrograph of multilayered AAO structure after etching of
the most soluble layers. From ref. [20].
3.3.4 Home-made AAO templates
For the anodization studies, high purity (99.999%) aluminium sheets with a
thickness of ~300 µm were used as the starting material. Pieces of ~20x20 mm
52 Chapter 3
were cut and annealed at 500 ºC for 3 h under Ar atmosphere. Anodization was
performed in a double-wall cell for temperature control and the sample was
exposed to the electrolyte through a 1.5 cm² circular opening in the bottom of the
cell. A platinum sheet was used as counter-electrode and the potential bias was
applied using a Keithley 2601 sourcemeter.
The “two-step anodization” process described in the seminal papers by
Masuda [9], [10] takes advantage from the fact that the growth of the porous oxide
leaves a textured pattern of concaves in the underlying aluminium substrate. If the
first grown oxide layer is removed and a second anodization step is performed,
pores nucleate preferentiably on those concaves and the resulting oxide layer
exhibit orderly arranged nanopores (Figure 3.8).
Figure 3.8 – Two-step anodization in oxalic acid. SEM micrographs of oxide
layer formed after first step anodization in 0.3M oxalic acid at 40V for 24h at
room temperature (a); patterned aluminium substrate after dissolution of the
oxide (b); oxide layer after second step anodization under same conditions for
75 min. The scale bars are 2 µm in all images.
The geometric parameters of the AAO such as pore diameter and interpore
distance are determined by the temperature and concentration of the electrolyte,
but mainly by the applied voltage. Among the most common electrolytes, sulfuric
and oxalic were chosen aiming at the smallest sizes as their optimal self-ordering
potential is reported to be 25-27 V and 40 V respectively [21], [22].
While the process of anodization may sound straightforward, it was found that
a great deal of experience is required before high quality templates can be obtained.
Electrochemical nanofabrication 53
Important practical details are often omitted from the reported methods, making
reproducibility difficult. For instance, while uniform AAO films with long-range
ordering were successfully grown in oxalic acid, films grown in sulfuric acid were
highly defective, with extensive cracking and signs of partial dissolution (Figure
3.9). The reasons for this are not clear but they are probably related to the higher
solubility of the aluminum oxide in sulfuric acid and the need to quickly rinse the
electrolyte once anodization is complete.
Figure 3.9 – AAO membrane grown in 0.3 M sulfuric acid at 25 V for 2h at 0ºC.
The damage was visible to the naked eye (a). Although the ordered nanopore
structure could be discerned (b), extensive damage was observed over its
surface (c) with signs of partial dissolution (d) and interpore cracking (e).
The process of pulse anodization involves combining two self-ordering
potential regimes i.e. mild and hard anodization (MA and HA). These regimes
differ in their typical current-time behaviours in that HA processes present a sharp
initial increase in current followed by an exponential decrease while MA processes
are characterized by a steady-state current throughout anodization [23]. Since the
54 Chapter 3
dependence of interpore distance on potential differs for each regime, the
production of oxide layers with continuous pores requires the HA potential and
pulse durations to be carefully tuned. Attempts to use that process in oxalic instead
of sulfuric acid were reported to fail because the HA pulse forms a barrier layer
(Figure 3.1b) that is too thick and prevents current recovery in a reasonably short
time [20]. Yet another attempt was made to produce such layered structures, based
on the principle of producing oxides with different composition to allow selective
dissolution as described in the last section.
The production of through-hole membranes for applications such as filtering
or the present case of nanowire synthesis requires detachment of the oxide layer
from the aluminum substrate and the opening of the barrier layer. This is usually
accomplished by dissolving the aluminium substrate and posteriorly etching the
barrier layer. The process brings complications as the top layer must be protected
against dissolution and it is hard to avoid the increase in pore size due to
infiltration of the etchant. Dealing with the problem of reliably detaching and
opening the AAO membranes, a method was proposed which consists in
performing an additional anodization step in concentrated sulfuric acid [24]. As the
alumina formed under this electrolyte incorporates a larger amount of sulphur, it
can be selectively dissolved leaving a free-standing through-hole membrane and
allowing further use of the already patterned aluminium substrate. In the present
case, that approach was tried as an alternative to pulse anodization with the
advantage that the voltage could be kept constant and therefore no tuning would be
required to guarantee continuous pores. Figure 3.10 shows the structure produced
under 15 V at 0ºC using 2.4 M and 12 M sulfuric acid for 34 and 5 minutes
respectively. The potential is quite far from the 25 V reported in the literature but it
was the one that gave best results for single layer anodization. The method of two-
layered anodization failed nonetheless, with extensive interpore cracks in the upper
layer that probably resulted from the very high concentration of sulfuric acid.
Electrochemical nanofabrication 55
Figure 3.10 – SEM micrographs of AAO layered structure after anodization in
2.4 M and 12 M H2SO4 under 15 V at 0ºC. The scale bars are 1 µm in both
images.
3.4 Conclusion
“(…) Let your workings remain a mystery. Just show people the results.” [25] The fabrication of Cu, Ni and Mg nanowire/bulk layered structures was
performed by electrodeposition in commercial membranes. These structures with
high specific-area were mechanically robust and relatively easy to handle. Their use
into useful structures for e.g. actuation would benefit from optimized templates
and electrode designs to allow for uniform pore filling during electrodeposition.
AAO membranes were successfully grown using oxalic acid and mild
anodization conditions. Although the production of highly ordered pores with
~60nm diameter is relatively easy to accomplish, reproduction of the reported
process in sulfuric acid to achieve smaller pores was not satisfactory. Furthermore,
the use of these membranes as a template for electrodeposition involves additional
steps that can make the whole process rather cumbersome.
56 Chapter 3
The attempts to produce dettached open-through membranes revealed that
frequent examination under the microscope would be required after every couple of
minutes of etching to avoid pore dilation or even the complete dissolution of the
oxide. The fabrication of such membranes likely requires stringent controls of bath
composition and temperature as well as in-situ monitoring of current density
evolution for acceptable reproductibility.
3.5 References
[1] E. Detsi, Z. G. Chen, W. P. Vellinga, P. R. Onck, and J. T. M. De Hosson, “Actuating and
sensing properties of nanoporous gold”, J Nanosci Nanotechnol, vol. 12, nr. 6, pp.
4951–4955, jun. 2012.
[2] E. Detsi and J. T. M. De Hosson, “Nanoporous Metals: New Avenues for Energy
Conversion and Storage”, Nano World Magazine, apr. 2017.
[3] J. Fu, E. Detsi, and J. T. M. De Hosson, “Recent advances in nanoporous materials for
renewable energy resources conversion into fuels”, Surface and Coatings Technology,
vol. 347, pp. 320–336, aug. 2018.
[4] E. Detsi, S. H. Tolbert, S. Punzhin, and J. T. M. D. Hosson, “Metallic muscles and
beyond: nanofoams at work”, J Mater Sci, vol. 51, nr. 1, pp. 615–634, aug. 2015.
[5] E. Detsi, S. Punzhin, J. Rao, P. R. Onck, and J. T. M. De Hosson, “Enhanced Strain in
Functional Nanoporous Gold with a Dual Microscopic Length Scale Structure”, ACS
Nano, vol. 6, nr. 5, pp. 3734–3744, mei 2012.
[6] S.-M. Zhang and H.-J. Jin, “Multilayer-structured gold/nanoporous gold composite for
high performance linear actuation”, Applied Physics Letters, vol. 104, nr. 10, p. 101905,
mrt. 2014.
[7] C. Cheng, L. Lührs, T. Krekeler, M. Ritter, and J. Weissmüller, “Semiordered
Hierarchical Metallic Network for Fast and Large Charge-Induced Strain”, Nano Lett.,
vol. 17, nr. 8, pp. 4774–4780, aug. 2017.
[8] S. Venetski, “‘Silver’ from clay”, Metallurgist, vol. 13, nr. 7, pp. 451–453, jul. 1969.
[9] H. Masuda and K. Fukuda, “Ordered metal nanohole arrays made by a two-step
replication of honeycomb structures of anodic alumina”, Science, vol. 268, nr. 5216, pp.
1466–1468, jun. 1995.
[10] H. Masuda and M. Satoh, “Fabrication of Gold Nanodot Array Using Anodic Porous
Alumina as an Evaporation Mask”, Jpn. J. Appl. Phys., vol. 35, nr. 1B, p. L126, jan.
1996.
Electrochemical nanofabrication 57
[11] W. Lee and S.-J. Park, “Porous Anodic Aluminum Oxide: Anodization and Templated
Synthesis of Functional Nanostructures”, Chem. Rev., vol. 114, nr. 15, pp. 7487–7556,
aug. 2014.
[12] Y. D. Gamburg and G. Zangari, Theory and Practice of Metal Electrodeposition. New
York: Springer-Verlag, 2011.
[13] D. A. Bograchev, V. M. Volgin, and A. D. Davydov, “Simple model of mass transfer in
template synthesis of metal ordered nanowire arrays”, Electrochimica Acta, vol. 96, pp.
1–7, apr. 2013.
[14] P. L. Taberna, S. Mitra, P. Poizot, P. Simon, and J.-M. Tarascon, “High rate capabilities
Fe3O4-based Cu nano-architectured electrodes for lithium-ion battery applications”,
Nature Materials, vol. 5, nr. 7, pp. 567–573, jul. 2006.
[15] K. S. Napolskii e.a., “Tuning the microstructure and functional properties of metal
nanowire arrays via deposition potential”, Electrochimica Acta, vol. 56, nr. 5, pp. 2378–
2384, feb. 2011.
[16] A. W. Maijenburg, E. J. B. Rodijk, M. G. Maas, and J. E. ten Elshof, “Preparation and
Use of Photocatalytically Active Segmented Ag|ZnO and Coaxial
TiO<sub>2</sub>-Ag Nanowires Made by Templated Electrodeposition”,
Journal of Visualized Experiments, nr. 87, mei 2014.
[17] W. J. Stępniowski and M. Salerno, Fabrication of nanowires and nanotubes by anodic
alumina template-assisted electrodeposition. One Central Press: Manchester, UK,
2014.
[18] W. Gu, J. T. Lee, N. Nitta, and G. Yushin, “Electrodeposition of Nanostructured
Magnesium Coatings”, Nanomaterials and Nanotechnology, p. 1, 2014.
[19] M. P. Proenca, C. T. Sousa, J. Ventura, M. Vazquez, and J. P. Araujo, “Distinguishing
nanowire and nanotube formation by the deposition current transients”, Nanoscale Res
Lett, vol. 7, nr. 1, p. 280, mei 2012.
[20] W. Lee, K. Schwirn, M. Steinhart, E. Pippel, R. Scholz, and U. Gösele, “Structural
engineering of nanoporous anodic aluminium oxide by pulse anodization of
aluminium”, Nat Nano, vol. 3, nr. 4, pp. 234–239, apr. 2008.
[21] K. Nielsch, J. Choi, K. Schwirn, R. B. Wehrspohn, and U. Gösele, “Self-ordering
Regimes of Porous Alumina: The 10 Porosity Rule”, Nano Lett., vol. 2, nr. 7, pp. 677–
680, jul. 2002.
[22] R. B. Wehrspohn, Red., Ordered Porous Nanostructures and Applications. Springer
US, 2005.
[23] W. Lee, R. Ji, U. Gösele, and K. Nielsch, “Fast fabrication of long-range ordered porous
alumina membranes by hard anodization”, Nat Mater, vol. 5, nr. 9, pp. 741–747, sep.
2006.
58 Chapter 3
[24] T. Yanagishita and H. Masuda, “High-Throughput Fabrication Process for Highly
Ordered Through-Hole Porous Alumina Membranes Using Two-Layer Anodization”,
Electrochimica Acta, vol. 184, pp. 80–85, dec. 2015.
[25] Laozi and S. Mitchell, Tao Te Ching: A New English Version. HarperCollins, 2000.
Radiation Damage
4.1 Ion-solid interactions
Several macroscopic properties of materials are strongly affected by changes
in composition and the structure of the sub-surface layer of about 1 micrometer
thickness. Among these properties are the electronic characteristics of semi-
conductor devices and the corrosion resistance, the mechanical hardness and
friction and lubrication properties of metals. There are even indications for an
influence on the fatigue lifetime.
Since the advent of semiconductors the process of ion implantation has been
increasingly adopted as an effective way of notifying near-surface properties of
materials. With ion implantation, impurity atoms penetrate into a material at high
velocities. Normally, ion energies between 10 and 200 keV are used. At these
energies heavy ions penetrate to a depth of 10 to 100 nm. After injection of ions, the
surface layer often is not in thermodynamic equilibrium. This is extremely
important, because it allows us to bypass the normal chemical solubility rules to
achieve impurity levels far above those obtainable by conventional treatments.
Clearly the modified surface layer will differ from the bulk in chemical and physical
properties.
Numerous examples of the beneficial effects of ion implantation can be found
in the literature. Semi-conductors can be doped in a highly controllable fashion by
ion implantation, and the benefits of ion implantation for the corrosion resistance
of metals have been indicated by several investigators [1]–[4].
60 Chapter 4
The influence of ion implantation is twofold. The first effect is a change in
chemical composition, the second a change in crystal structure. In most cases these
aspects cannot be separated. The changes in structure may be influenced by the
presence of the implanted foreign atoms. For instance, these foreign atoms may
adhere to defect structures and form stable configurations. The two effects may be
separated to a certain extent by applying self-ion implantation into ultra-pure
material.
If an energetic ion penetrates into a metal, it loses its energy by two
mechanisms. The first mechanism, electronic stopping' is due to the interaction of
the incident ion with the electrons of the crystal. The collisions may be elastic or
inelastic, but no permanent damage remains in metals, because the electronic
structure quickly recovers. The energy transferred to the electrons is largely
dissipated as heat. At an energy of about 100 keV, the contribution of electronic
stopping is about 10%. The second mechanism, nuclear stopping, is due to
collisions with lattice atoms. If the incident ion approaches a lattice atom to within
the Thomas-Fermi screening distance, it experiences a strong repulsive force. Since
the energies used in this work are too low to produce nuclear reactions the nuclear
collisions are always elastic.
If sufficient energy is transferred to a lattice alone, it is displaced from its site
to become a self-interstitial. The vacant lattice site is called a vacancy and the
combination of the interstitial and the vacancy a Frenkel pair. The energy needed
for the formation of a Frenkel pair in copper is about 29 eV. After implantation of a
self-ion of 100 keV, one expects about 1000 Frenkel pairs to be formed. The region
where these Frenkel pairs are produced is called the damage cascade. The change
of structure (damage) due to self-implantation is the result of the loss of energy of
the incident ions by nuclear collisions only.
To make the description a bit more quantitative the following remarks are
made: as said above the distribution of implanted ions with depth in a target is
determined by the energy loss processes which slow down the ion on its way
through the target. Generally it is considered that the ion loses energy by two
independent processes, namely nuclear and electronic collisions. In nuclear
collisions kinetic energy is transferred to the target atoms. These collisions are
elastic and result in relatively large angular deflections of the ion. Electronic
collisions are inelastic and they involve energy transfer to the electrons of the target
Radiation Damage 61
atoms. Contrary to the nuclear collisions, these electronic collisions result in
negligible angular deflection of the ions.
Ever since the discovery of energetic particle emission by radioactive
materials the interest in how these particles were slowed down in solids has driven
the steady development of stopping theories. The present understanding of
stopping mechanisms is mainly due to the work of Bohr [5], [6], Bethe [7], Firsov
[8], Lindhard, Scharff and Schlott [9].
If we assume that nuclear and electronic losses are independent processes the
energy loss per distance can be written as
( ) ( )n e
dEN S E S E
dx , (4.1)
dE is the energy loss, dx is the travelled distance by ion, N is the target density, nS
is the cross-section for nuclear stopping and eS is the electronic stopping power.
The nuclear scattering process, with conservation of energy and momentum,
is depicted in Figure 4.1.
Figure 4.1 – Scattering between particles of mass M1 and M2, impact parameter
p and scattering angles 1 , 2 in the laboratory system.
In the non-relativistic case, the kinetic energy transferred to the target atom is
given by
1 2
21 2
( ) (1 cos )( )
m mT E E
m m
, (4.2)
62 Chapter 4
where E is the energy before scattering, is the deflection angle in the center of
mass system (in which the total momentum of the system is zero). is a function
of the impact parameter p and of the interatomic potential V r . In a head-on
collision (p = 0) equals and from Eq. (4.2) it can be seen that the maximum
amount of energy is now transferred.
In our consideration of the electronic stopping power we restrict ourselves to
the energy range far below 1 MeV, which is of practical interest for ion
implantation. The LSS theory, named after Lindhard, Scharff and Schiott [9], was
the first to focus on this low energy region were the ion velocity is less than the
velocity of the target electrons at the Fermi level (i.e. less than about 25 keV/amu).
Here it is assumed that electrons form a free electron gas and from this it is derived
that the stopping is proportional to the ion velocity (or the square root of the
energy):
elS ion velocity k E . (4.3)
Figure 4.2 gives the nuclear and electronic stopping powers calculated from
the LSS-theory for the case of aluminum and nitrogen implanted in copper [10]. It
also gives the ion range R as obtained by integrating the nuclear and electronic
stopping powers using
0 0
1
0 0
= ( ) (E)
E E
n e
dxR dE N S E S dE
dE
. (4.4)
A major advance in electronic stopping calculations has been achieved by
Brandt and Kitagawa [11]. In their approach the effective charge of the ion is
calculated by assuming all electrons to be stripped which have velocities less than
the relative velocity between the ion and the Fermi velocity of the target electrons.
This revises the Bohr concept which stated that all electrons with a velocity less
than the absolute ion velocity would be stripped. Their approach is incorporated in
the TRIM simulation program of Biersack and coworkers [12].
Radiation Damage 63
Figure 4.2 – a) Stopping powers for Al and N in amorphous copper as
calculated by LSS-theory. b) Projected ranges of Al and N in amorphous copper
as calculated by LSS theory [10].
The LSS-theory as discussed so far, is widely applied for calculating ion range
distributions. On doing this one should bear in mind that this theory assumes the
target to be structureless (i.e. amorphous). In reality however, most of the metal
targets to be implanted are polycrystalline and, moreover, they very often have a
pronounced texture. Particularly the semiconductor targets implanted for device
applications have an extremely high perfection in crystallinity.
This crystallinity has been found to lead to discrepancies in the ion
distribution, especially in the deeper tail, as compared to the distribution calculated
from Eq. (4.4). In 1960 Davies [13] reported long ranges for heavy ions implanted
in polycrystalline aluminum and tungsten. At about the same time it was shown
[14] that the sputtering yield for ions bombarding single crystals depends markedly
on the crystal orientation (Figure 4.3).
64 Chapter 4
<100> <110>
<111> random
Figure 4.3 – XTEM micrographs of copper crystals with <100>, <110>, <111>
and random surface orientation perpendicularly implanted with 170 keV
aluminum ions to a dose of 1016 Al/cm2 at room temperature (scale: width of
sub-figure is 1 micrometer) [15].
As a result channeling contributions were investigated and described
theoretically by the continuum model of Lindhard [16] where the actual periodic
Radiation Damage 65
potential of the atomic axes or planes is replaced by a potential averaged over a
direction parallel to these axes or planes. As the incidence angle to a row becomes
larger than some critical angle, the ion trajectory no longer can remain channeled.
These axial critical angles as calculated by Lindhard appear to correspond well with
experimental values [17].
The relative amount of ions being channeled is very sensitive to the precise
orientation of the ion beam. Nevertheless it may also happen that ions that start in
a random direction, get scattered into open channels. This is especially so for heavy
ions at low energies because of their large critical angles. Furthermore, the relative
amount of channeled ions depends on the ion dose because of the accumulation of
radiation damage and the eventual destruction of the channels by amorphization
If the ion profile ( )N x is assumed to be Gaussian it can be expressed in terms
of the average projected range pR (in cm), the spread
pR (in cm) and the
implanted dose D (in ions/cm2) as
2
2( ) exp
2( )2 .
p
pp
x RDN x
RR
(4.5)
in ions/cm3. The projected range pR is defined as the range projected onto
the incidence direction of the ion beam. For ions which terminate at substitutional
lattice positions the atomic concentration is obtained from ( )N x through dividing
by the atomic density of the target. For interstitially implanted ions the atomic
density is increased with ( )N x . In deriving Eq. (4.5) the erosion of the target by
sputtering and the reflection of incident ions is not taken into account.
4.2 Deep damage
In this work it is anticipated that after ion implantation in metals damage is
created not only in the region of the damage cascades, but in a much deeper region
as well. The range in which the damage cascades are expected coincides
approximately with the range in which the incident ions lose their energy.
Figure 4.4 shows the XTEM micrographs of copper (110) single crystals
implanted with copper ions along the <110> axis with ion energies of 170 and 700
66 Chapter 4
keV. The LSS-projected ranges for these implants are pR = 0.04 µm and
pR = 0.16
µm respectively. The insets show the vacancy profiles as calculated by the
simulation code MARLOWE. The extension of the tail of these profiles is in good
agreement with experimental damage ranges. There is a clear dependence of the
damage range on the ion energy.
Figure 4.4 – XTEM micrographs of copper (110) crystals implanted with 170
keV and 700 keV copper ions at a dose of 1016 Cu/cm2. The insets give the
vacancy depth distribution as calculated by MARLOWE [15].
For the 170 keV implant the damage extends to 0.7 µm and for the 700 keV
implant this is about 1.4 µm. This observation confirms the square root dependence
of the damage range on the ion energy which is expected if this damage range is
Radiation Damage 67
determined by the range of channeled ions. It should be noted that the dependence
of the LSS projected ranges on the ion energy is about linear for these implants.
The maximum ion ranges can again be calculated by using the electronic stopping
constant k = 1.05 eV/nm as obtained from SRIM [12], which gives maxR (170 keV) =
0.8 µm and maxR (700 keV) = 1.6 µm, in reasonable agreement with the damage
ranges in Figure 4.4.
But in actual fact, different damage regions can be observed: a region
coinciding with the range of the incident ions, and a region containing less damage
extending far beyond the first region. Other authors have also reported a damage
range extending beyond the range of the incident ions [18]–[21]. It has been
suggested that the deep damage may be due to mobile defects, i.e. this damage is
due to migration and aggregation of self-interstitial atoms escaping from the
damage cascades.
Deep damage due to the clustering of interstitials escaping from the damage
cascades at pre-existing trapping centers have been studied in [22], [23]. The
results described gave a strong indication that preexisting trapping centers do not
play a ro1e in the formation of the deep damage. The main problem is the
formation of nucleation seeds by clustering of interstitials. Once the seeds are
present they are expected to grow to larger clusters by further capture of
interstitials, because the capture radius is large and there are enough free
interstitials available. Large clusters are expected to collapse to dislocation 1oops
(see below for a theoretical analysis).
The growth of dislocation loops has been studied, solving rate equations, by
many authors (see e.g. [24]). In these calculations, however, the problem of
nucleation has not been treated: a fixed cluster density has been assumed as an
initial condition of the calculation. The results are usually compared with the
results of in-situ high voltage electron microscopy irradiation experiments. If the
cluster seed formations were due to the aggregation of migrating interstitials, there
should be a sufficient probability for the formation of clusters in the region where
the deep damage has been found.
In general deep damage does not depend on the implantation dose-rate, and
therefore we conclude that these cluster seeds are formed by interstitials created by
one incident ion only. A calculation of the diffusion of interstitials during the time
68 Chapter 4
interval between two ions incidents within an appropriate distance from each
other, leads to the same conclusion.
As has been observed in FIM studies [25] and confirmed in simulation studies
[26] a primary damage cascade consists of one or several depleted zones containing
clusters of vacancies and a concentration of (di)vacancies. This network of vacancy
type of damage may collapse partly to form vacancy clusters or dislocation loops
(see below). The interstitial atoms are formed for the major part in a region
separated by some length of nm from the depleted zone [27]. The most elementary
theory to estimate the number of Frenkel pairs is that of Kinchin and Pease [28].
Assuming an amorphous so1id, two body hard-sphere collisions, a sharp
displacement threshold energy Ed and no inelastic losses, they obtain for the
number of Frenkel pairs created by one incident ion with energy E :
2
d
d
EN
E . (4.6)
Robinson and Torrens have modified this theory, using a more realistic
interatomic potential and a crystalline structure. In addition, these authors
replaced the total kinetic energy of the incident ions by E - Ein, where Ein represents
the inelastic energy loss to the electrons.
The Kinchin-Pease formula may then be rewritten as [29]
2
ind
d
E EN
E
, (4.7)
where is called the displacement efficiency, which depends on the scattering law
assumed. With 0.85 , and / 0.28inE E [30] and a displacement threshold
energy of 29 eV [31] the number of Frenkel pairs produced by one incident ion of
100 keV in copper is about 1000.
In the evaluation of the experimental results (see Chapters 5 and 6) it should
be kept in mind that during implantation the atomic collisions in the near surface
region may lead to ejection, or sputtering, of target atoms from the surface. This
process can be quantified by the sputtering yield S , defined as the number of
ejected target atoms per incident ion, which can take values from less than one to
ten.
Radiation Damage 69
Sigmund [32] has developed a rigorous theoretical treatment where
sputtering arises from discrete nuclear energy loss processes. In this so called
linear cascade region the intersection of the collision cascade with the surface
determines the magnitude of the sputtering yield S , which, for normal incidence
on elemental targets, can be approximated by
( )S n
yield
surface atomic binding
S E
E . (4.8)
Sputtering yields calculated with Eq. (4.8) have often shown remarkable
agreement with experimental values, when the binding energy is taken equal to the
sublimation energy of the target material. This agreement even appears to hold
when the underlying assumptions of this linear cascade theory are violated (e.g. for
heavy ions at high energies producing very dense collision cascades or spikes).
It is found that, except for glancing angles, the sputtering yield is about
inversely proportional to the cosine of the angle between the incident ion beam and
the surface normal [33]. Sputtering in particular puts a practical upp
er limit to the concentration which can be obtained during high dose implantation.
This is a serious drawback to the general claim that ion implantation, being a non-
equilibrium process, is able to produce new metastable compounds. Also, the
angular dependence of the sputtering yield makes it difficult to achieve uniformity
in retained dose over a target of complex shape. Furthermore, during high dose
implantations, sputtering may lead to very profound topography on crystalline
surfaces by the formation of facetted pits and regular pyramidal features [34]. The
development of these features depends on the crystallographic orientation of the
surface with respect to the ion beam. In addition, surface irregularities and
contaminations initially present can also give rise to topography changes due to
local differences in sputter yield.
4.3 Defect formation and diffusion
Except at very low temperatures interstitials are highly mobile. A part will
migrate into the collapsed depleted zones and recombine. The number of defects
lost in this way may be estimated from the recovery of radiation damage after stage
I (where interstitials become mobile), as measured by resistivity measurements.
70 Chapter 4
Although the interpretation is difficult it is assumed that 70 to 90% of the
interstitials will be lost by recombination after diffusion. Most remaining
interstitials are assumed to be lost at sinks, such as the surface and dislocations. A
few, however may aggregate to form small clusters with a high binding energy.
These clusters form sinks for interstitials created by subsequent incident ions. We
expect that the deep damage consists of these interstitial clusters.
It has been suggested that clusters of interstitials of considerable size, even
containing up to 13 interstitials, are mobile as single interstitials by regrouping
[35]. If this is true the deep damage cannot be explained without the assumption of
pre-existing trapping centers. This assumption is unlikely, however, in our rather
pure materials. This does not imply that larger clusters do not regroup, but we
assume that in such a process the center of mass of the clusters will not move
appreciably.
In the rate-equations as we will explain later it is assumed that single, di- and
tri-interstitials are highly mobile, but that the center of mass of larger clusters is
only s1owly mobile or immobile. The clusters will grow by capturing mobile
(clusters of) interstitials. Clusters containing 7 to 13 interstitials will convert to
stable Frank sessile loops [35]–[38]. All diameters from 5 to 50 nm change to a
perfect loop, because with larger diameters it is energetically favorable to form a
loop with a perfect Burgers vector (see below), but without a stacking fault. The
energy due to the stacking fault increases with the loop area, whereas the energy
due to the dislocation line increases mainly with the circumference of the loop. An
absolute maximum in the damage may be reached if the loops overlap and a
dislocation entanglement occurs. Such an entanglement has been observed in semi-
conductors [39], [40] but not in pure metals after (self-)implantation [41].
Upon annealing, small clusters may become mobile and aggregate to larger
clusters or loops. At greater depth, where the concentration of damage is low, these
clusters may escape without fusing.
The binding energy of a di-interstitial has been estimated to be about 1 eV in
FCC metals and the binding energy of an interstitial to a di-interstitial is estimated
at 1.5 eV [38]. This value increases further with the number of interstitials. The
formation energy of a single interstitial is 2.5 to 4 eV [37], [42]. These binding
energies are sufficiently high that dissociation never occurs. Already at lower
temperatures than those where dissociation could occur, all defects of interstitial
Radiation Damage 71
type have been removed by annihilation by thermally created vacancies. For copper
a complete recovery of the deep damage was observed at 750 K. At this temperature
the concentration of thermally created vacancies increases sharply.
In the following we will analyze the competition between point defect clusters
(vacancy and interstitial type) and dislocation loops as will be a subject of
evaluation in Chapter 5. If we create a vacancy in the interior of a spherical solid
and ‘spread’ the atom which is taken out over the surface leading to an increase of
the rigid volume (conserved) of R :
2 34
43
a aR R r , (4.9)
or
3 2/3aR r R . (4.10)
The formation energy of a vacancy, i.e. the difference in surface energy of
the solid with and without the cavity follows
24 (1 2 /3 )v
f a aU r r R . (4.11)
Neglecting terms in 2
R and since ar R , the vacancy formation energy
for n vacancies of volume a can be written as
2/3
2 2/34 6sphere nv vf f a a fU U r n n n U . (4.12)
The binding energy is expressed as
1/31binding
nv vfU nU n , (4.13)
i.e. for a divacancy 2 1.6v vf fU U
and 2 0.4v vb fU U
.
Let’s turn now to the conversion a spherical point defect cluster to a
dislocation loop. Suppose the n vacancies are formed on a flat circular disk of
thickness of a single vacancy ( 2 1/3a an r ) and radius r:
2 2/32 2disc
f aU r n . (4.14)
72 Chapter 4
From Eqs. (4.12) and (4.14) it is clear that a sphere has the lowest surface area
and therefore the lowest formation energy. However, when the sphere collapses to
a disc it releases a large amount of surface energy (for metals of the order of 1 J/m2)
but it will lose upon strain energy; in fact the surface energy has to be converted to
strain energy of a dislocation loop.
Roughly speaking the elastic strain energy per unit of length of a dislocation
goes with the product of the shear modulus and the magnitude of the Burgers
vector b squared, i.e.
2 1/3 22 2loop
fU r b n b . (4.15)
Comparing Eq. (4.15) with Eq. (4.14) indicates that when r exceeds a critical
value cr a dislocation loop becomes always favorable:
2
c
br
. (4.16)
Typical values for metals leads to 2 4 nmcr and therefore one does not
expect stable discs at larger cr . Also a dislocation loop is energetically more
favorable than a spherical point defect cluster for large enough values of n :
2 1/6
2/3 1/6 1/3
2
6
disloopf
spheref
U b
U n
. (4.17)
It is quite obvious from the above that we ignore in the analysis the
crystallography. In fact it is an elastic continuum–rigid solid approach. The
collapse of a disc does not mean that the crystallographic planes join in the correct
manner (e.g. the threefold ABC {111) planes stacking in FCC Au leads to the
generation of a stacking fault on {111} condensed vacancy planes) and unfaulting
reactions may occur depending in the diameter of the dislocation loop, residual
stress etc, bound by a partial dislocation 1/3 111 . In fact for unfaulting another
partial dislocation 1/6 112 should be nucleated so as to produce, together with
the bounding partial, a perfect dislocation loop as assumed in Eq. (4.17):
1/3 111 1/6 112 1/2 110 . (4.18)
Radiation Damage 73
These unfaulting reactions in FCC Au are favorable, getting rid of the stacking
fault energy of about 20 mJ/m2, above a critical radius of about 5 nm.
So, the general conclusion which will hold is: upon ion radiation at moderate
energies, say below 1 MeV, point defects clusters are produced. For metals,
spherical clusters will collapse to either faulted or unfaulted dislocations loops
depending on the size and specific stacking fault energy. Spherical voids/clusters
and disks are not stable at RT irradiation according to predictions based on linear
elastic continuum mechanics.
4.4 Reaction rate theory
During irradiation, vacancies and interstitials are formed as Frenkel pairs.
These vacancies and interstitials may diffuse, combine, or be lost at sinks. The
theoretical descriptions of radiation damage by so-called rate theory has been
around since the seventies and were used to calculate swelling and irradiation
creep phenomena in nuclear reactor materials. The first approaches condensed to
[43], [44]
2vv v v v i v
dCK k D C C C
dt , (4.19)
2ii i i i i v
dCK k D C C C
dt , (4.20)
where is the rate constant for bulk recombination of interstitials and vacancies,
vK and iK are their production rates, 2,i vk is the total sink strength and
,i vD
represents diffusion coefficients of interstitials and vacancies .
The reactions describing the main processes occurring during and after
irradiation can be specified in a bit more detail as listed in Table 4.1 [45]
74 Chapter 4
Table 4.1 – Rate Equations
Pirr i+v Production of vacancies and (self)interstitials
Ptherm v Production of thermal vacancies at sinks
ki+lv Kki,lv (k-l)i , k>1;
0, k=l;
(l-k)v, k<1
Recombination of vacancy and interstitial clusters
ki+li Kki,li (k+l)i Clustering of interstitials
kv+lv Kkv,lv (k+l)v Clustering of vacancies
kv Kd,kv (k-1)v+v Vacancy dissociation from vacancy cluster
ki+s Ks,ki s Loss of interstitial clusters at sinks
kv+s Ks,kv s Loss of vacancy clusters at sinks
The symbol v denotes a vacancy, i an interstitial, s an unsaturable sink, and k,l
are natural numbers. The parameter Pirr is the production rate for Frenkel pairs
due to the irradiation. The different K’s denote the rate-constants for
recombination, cluster merging, etc. Dissociation of interstitial clusters is not
expected, because the binding energy is too high.
With these reactions we can describe the change with time in the cluster
concentrations by a set of differential rate equations. For clusters containing n
vacancies this change is given by
, ,
, s,nv ( )
nvnv nv ki nv ki nv kv nv kv nv
k k
eqd nv nv nv nv s
dCD C K C C K C C
dt
K C K C C C
(4.21)
The terms represents subsequently diffusion, recombination, clustering,
dissociation and loss at sinks, respectively.
The interstitial cluster concentration is given by
, , s,ni
ni ni ni kv ni kv ni ki ni ki ni sk k
dCD C K C C K C C K C
dt . (4.22)
The subsequent terms reflect diffusion, recombination, clustering, and loss at
sinks, respectively. Compared to Eq. (4.21) here the dissociation term ,d ni niK C is
Radiation Damage 75
absent, in contrast to ,d nv nvK C , because dissociation of interstitial cluster will not
appear due to the high binding energy.
In both equations k,l and n denote natural numbers. For monovacancies and
mono-interstitials (n=1) the production term Pirr must be added. If only the effect
of a single incident ion is considered, the production can also be introduced via the
initial conditions. The production of thermal vacancies is included as an
equilibrium concentration eqC in the term describing the loss of vacancies at sinks.
If these rate equations could be solved simultaneously for all possible cluster
sizes one could calculate the nucleation and growth of these clusters. However,
because the computation time is limited, the number of rate-equations cannot be
very large. In the present work we have restricted the number of equations to only
those that are absolutely necessary to yield an estimate for the nucleation of stable
interstitial clusters (see Chapter 5).
In order to check the model by a calculation one must know the values of the
parameters involved. However, the values of most parameters are not known
directly from measurements. The diffusion constant D is
0
mE
kTD D e
, (4.23)
with mE being the migration energy, k the Boltzmann constant and T the
temperature. The pre-exponential- constant 0D depends on the migration entropy,
on the jump frequency and on the geometry. For mono-interstitials we used
estimates of Young [38] 0 12.5 2 110i oD a s and mE =0.117 eV, with 0a the nearest
neighbor distance. For the migration energy, interstitials become mobile at a
temperature of about 40 K (end of stage I). In early publications it has been
assumed that di-vacancies become mobile at RT (stage III) and vacancies at about
500 K (stage IV) [46]. In more recent work it has been reported that vacancies
become mobile in stage III ( 0.72 eVmvE ), and that in most FCC there is no
evidence for highly mobile di-vacancies in this stage [47].
Either way, the migration of all vacancy defects (v,2v,3v,..) at RT is many
orders of magnitude slower than the interstitial migration and it can be neglected
on the time-scale of the interstitial cluster formation. We therefore used the value
nvD = 0 (n=1,2,3..). On the basis of Huang scattering experiments it has been
76 Chapter 4
concluded that, di-interstitials are more mobile than mono-interstitials in copper
[48]. Since the vacancy mobility is set equal to zero in the calculation, no vacancy
clustering and neither dissociation will occur. We treat the vacancy clusters as sinks
for interstitials, neglecting saturation.
As will be explained in Chapter 5 in our experiments irradiation is done with a
focused ion beam of Ga+ ions, and therefore it is not a self-implantation
methodology. As a consequence the rate Eqs. (4.21) and (4.22) have to include also
the contributions of Ga, i.e.
, s,v
(1 )
( )
vv v i iGa i i v iGa iGa v
eqd v v v v s
dcD c K c c D c c D c c
dt
K c K c c c
(4.24)
where the last term depends on the relative concentration of sinks, i.e. dislocations,
grain-boundary, cavities, interstitial loops densities etc.
, s,(1 )ii i v iGa i i v i i i i i s
dcD c K c c D c c K c c K c
dt (4.25)
vc and ic are the fractional concentrations of vacancies and interstitials,
respectively.
4.5 References
[1] J. R. Morris, R. A. Collins, and G. Dearnaley, “The influence of ion implantation on the
thermal oxidation of copper”, J. Phys. F: Met. Phys., vol. 8, nr. 6, p. 1333, 1978.
[2] L. G. Svendsen, “A comparison of the corrosion protection of copper by ion
implantation of Al and Cr”, Corrosion Science, vol. 20, nr. 1, pp. 63–68, jan. 1980.
[3] G. Dearnaley and P. D. Goode, “Techniques and equipment for non-semiconductor
applications of ion implantation”, Nuclear Instruments and Methods in Physics
Research, vol. 189, nr. 1, pp. 117–132, okt. 1981.
[4] F. H. Stott, Z. Peide, W. A. Grant, and R. P. M. Procter, “The oxidation of chromium-
and nickel-implanted nickel at high temperatures”, Corrosion Science, vol. 22, nr. 4, pp.
305–320, jan. 1982.
[5] N. Bohr, “II. On the theory of the decrease of velocity of moving electrified particles on
passing through matter”, The London, Edinburgh, and Dublin Philosophical Magazine
and Journal of Science, vol. 25, nr. 145, pp. 10–31, jan. 1913.
Radiation Damage 77
[6] N. Bohr and J. Lindhard, “Kgl. Dan. Vidensk. Selsk.”, Mat.-Fys. Medd., vol. 18, nr. 8,
1948.
[7] H. Bethe, “Bremsformel für elektronen relativistischer geschwindigkeit”, Zeitschrift für
Physik, vol. 76, nr. 5–6, pp. 293–299, 1932.
[8] O. B. Firsov, “A qualitative interpretation of the mean electron excitation energy in
atomic collisions”, Zhur. Eksptl’. i Teoret. Fiz., vol. 36, 1959.
[9] J. Lindhard, M. Scharff, and H. E. Schiott, “K. danske Vidensk”, Selsk., Math.-fys.
Meddr, vol. 33, nr. 1, 1963.
[10] E. Gerritsen, “Surface modification of metals by ion implantation”, University of
Groningen, 1990.
[11] W. Brandt and M. Kitagawa, “Effective stopping-power charges of swift ions in
condensed matter”, Phys. Rev. B, vol. 25, nr. 9, pp. 5631–5637, mei 1982.
[12] J. F. Ziegler, M. D. Ziegler, and J. P. Biersack, “SRIM – The stopping and range of ions
in matter (2010)”, Nuclear Instruments and Methods in Physics Research Section B:
Beam Interactions with Materials and Atoms, vol. 268, nr. 11–12, pp. 1818–1823, jun.
2010.
[13] J. A. Davies, J. D. McIntyre, R. L. Cushing, and M. Lounsbury, “The range of alkali
metal ions of kiloelectron volt energies in aluminum”, Can. J. Chem., vol. 38, nr. 9, pp.
1535–1546, sep. 1960.
[14] P. K. Rol, J. M. Fluit, and J. Kistemaker, “Theoretical aspects of cathode sputtering in
the energy range of 5–25 keV”, Physica, vol. 26, nr. 11, pp. 1009–1011, nov. 1960.
[15] E. Gerritsen and J. T. M. D. Hosson, “Cross Section Transmission Electron Microscopy
(XTEM) on Inert Gas Implanted Metals”, in Fundamental Aspects of Inert Gases in
Solids, Springer, Boston, MA, 1991, pp. 143–152.
[16] J. Lindhard, “Influence of Crystal Lattice on Motion of energetic charged particle”, Kgl.
Danske Vidensk. Selskab. Mat. Fys. Medd, vol. 34, nr. 14, 1965.
[17] D. S. Gemmell, “Channeling and related effects in the motion of charged particles
through crystals”, Rev. Mod. Phys., vol. 46, nr. 1, pp. 129–227, jan. 1974.
[18] J. A. Borders and J. M. Poate, “Lattice-site location of ion-implanted impurities in
copper and other fcc metals”, Phys. Rev. B, vol. 13, nr. 3, pp. 969–979, feb. 1976.
[19] D. K. Sood and G. Dearnaley, “Ion implanted surface alloys in nickel”, Radiation
Effects, vol. 39, nr. 3–4, pp. 157–162, jan. 1978.
[20] I. Nashiyama, P. P. Pronko, and K. L. Merkle, “Annealing studyof self-ion-irradiated
gold by proton channeling”, Radiation Effects, vol. 29, nr. 2, pp. 95–98, jan. 1976.
[21] P. P. Pronko, “Dechanneling measurements of defect depth profiles and effective cross-
channel distribution of misaligned atoms in ion-irradiated gold”, Nuclear Instruments
and Methods, vol. 132, pp. 249–259, 1976.
78 Chapter 4
[22] R. J. T. Lindgreen, D. O. Boerma, and J. T. M. de Hosson, “A study of shallow and deep
damage in Cu after implantation of 100 keV Cu and Ag ions”, Nuclear Instruments and
Methods in Physics Research, vol. 209–210, pp. 963–967, mei 1983.
[23] R. J. T. Lindgreen, D. O. Boerma, and J. T. M. D. Hosson, “A study of shallow and deep
damage in Cu and Al after self-implantation”, Radiation Effects, vol. 71, nr. 3–4, pp.
289–314, jan. 1983.
[24] M. Kiritani, K. Urban, and N. Yoshida, “Formation of submicroscopic vacancy clusters
and radiation-induced refining in metals by electron-irradiation at high temperatures”,
Radiation Effects, vol. 61, nr. 1–2, pp. 117–126, 1982.
[25] D. N. Seidman, M. I. Current, D. Pramanik, and C.-Y. Wei, “Direct observations of the
primary state of radiation damage of ion-irradiated tungsten and platinum”, Nuclear
Instruments and Methods, vol. 182–183, pp. 477–481, apr. 1981.
[26] J. Mory and Y. Quéré, “Dechanneling by stacking faults and dislocations”, Radiation
Effects, vol. 13, nr. 1–2, pp. 57–66, mrt. 1972.
[27] D. Wieluńska, L. Wieluński, and A. Turos, “The effect of dislocations on the planar
dechanneling. I. Analytical model”, physica status solidi (a), vol. 67, nr. 2, pp. 413–419,
1981.
[28] G. H. Kinchin and R. S. Pease, “The Displacement of Atoms in Solids by Radiation”,
Rep. Prog. Phys., vol. 18, nr. 1, p. 1, 1955.
[29] I. M. Torrens and M. T. Robinson, “Computer Simulation of Atomic Displacement
Cascades in Solids”, in Interatomic Potentials and Simulation of Lattice Defects,
Springer, Boston, MA, 1972, pp. 423–436.
[30] M. T. Robinson and I. M. Torrens, “Computer simulation of atomic-displacement
cascades in solids in the binary-collision approximation”, Phys. Rev. B, vol. 9, nr. 12,
pp. 5008–5024, jun. 1974.
[31] P. Lucasson, “Production of Frenkel defects in metals”, 1975.
[32] P. Sigmund, “Sputtering by ion bombardment theoretical concepts”, in Sputtering by
Particle Bombardment I, Springer, Berlin, Heidelberg, 1981, pp. 9–71.
[33] P. C. Zalm, “Quantitative sputtering”, Surface and Interface Analysis, vol. 11, nr. 1–2,
pp. 1–24, 1988.
[34] G. Carter, I. V. Katardjiev, M. J. Nobes, and J. L. Whitton, “Sputtering-induced surface
topography on F.C.C. metals”, Materials Science and Engineering, vol. 90, pp. 21–32,
jun. 1987.
[35] H. R. Schober and R. Zeller, “Structure and dynamics of multiple interstitials in FCC
metals”, Journal of Nuclear Materials, vol. 69–70, pp. 341–349, feb. 1978.
[36] W. Schilling, “Self-interstitial atoms in metals”, Journal of Nuclear Materials, vol. 69–
70, pp. 465–489, feb. 1978.
Radiation Damage 79
[37] P. H. Dederichs, C. Lehmann, H. R. Schober, A. Scholz, and R. Zeller, “Lattice theory of
point defects”, Journal of Nuclear Materials, vol. 69–70, pp. 176–199, feb. 1978.
[38] F. W. Young, “Interstitial mobility and interactions”, Journal of Nuclear Materials, vol.
69–70, pp. 310–330, feb. 1978.
[39] T. Tokuyama, “Use of Ion Implantation in Device Fabrication at Hitachi CRL”, in Ion
Implantation in Semiconductors 1976, Springer, Boston, MA, 1977, pp. 519–533.
[40] J. Fletcher, J. Narayan, D. H. Lowndes, and T. Y. Tan, Defects in Semiconductors.
North-Holland Publishing Co. New York, 1981.
[41] R. A. Johnson, Physics of Radiation Effects in Crystals. North-Holland, 1986.
[42] M. W. Thompson, Defects and radiation damage in metals. London,: Cambridge U.P.,
1969.
[43] A. D. Brailsford and R. Bullough, “The rate theory of swelling due to void growth in
irradiated metals”, Journal of Nuclear Materials, vol. 44, nr. 2, pp. 121–135, aug. 1972.
[44] A. D. Brailsford and R. Bullough, “The theory of sink strengths”, Phil. Trans. R. Soc.
Lond. A, vol. 302, nr. 1465, pp. 87–137, jul. 1981.
[45] R. J. T. Lindgreen, “Radiation damage in fcc metals”, University of Groningen, 1984.
[46] H. Mehrer and A. Seeger, “Interpretation of Self-Diffusion and Vacancy Properties in
Copper”, physica status solidi (b), vol. 35, nr. 1, pp. 313–328.
[47] R. W. Balluffi, “Vacancy defect mobilities and binding energies obtained from annealing
studies”, Journal of Nuclear Materials, vol. 69–70, pp. 240–263, feb. 1978.
[48] P. Ehrhart and U. Schlagheck, “Investigation of Frenkel defects in electron irradiated
copper by Huang scattering of X-rays. I. Results for single interstitials”, J. Phys. F: Met.
Phys., vol. 4, nr. 10, p. 1575, 1974.
Dense solid samples
5.1 Initial observations on Au cantilevers
The investigations discussed in this Chapter start focusing on gold as a typical
example of face-centered cubic material. Irradiation with Ga+ ions caused upwards
bending (towards the beam) in all observations (Figure 5.1). The curvature along
the cantilevers length remains mostly uniform during the initial stages and
assumes a U-shape as irradiation proceeds (Figure 5.2). This effect was later
observed for all the other investigated systems and it can be explained considering
the change in angle between the incident ion beam and the sample normal along
the cantilever in the course of the experiment. As irradiation progresses, the
deflection of the free-end becomes more pronounced than the supported end. The
irradiation angle affects sputtering rate and implantation range resulting in the
observed non-uniform curvature in the late stages. In the following, only the early
stages in which curvature remains uniform along the cantilever length are
considered.
82 Chapter 5
Figure 5.1 – Ion-beam irradiation of the cantilever (a); the implantation profile
of Ga+ ions is shown schematically. Deflection of Au cantilever (d0 = 144 nm)
after irradiation to a fluence of 1.4×1020 m−2 (b).
The observed deformation requires the introduction of some sort of
gradient along the films’ thickness. An intuitive attempt to explain upwards
bending would be to assume a negative volume change in the subsurface region,
e.g. by the accumulation of vacancies. However, a stationary description is not
appropriate as the defects continue to migrate by thermal diffusion after the
cascade relaxation stage and are subject to recombination and/or annihilation.
Added to that, each irradiation step in these experiments consists of an order of 108
collision events (considering r=50 nm, I=40pA, o=0.5 and tdwell=10µs in Eq. (2.19))
making it necessary to model the time-dependent evolution of the generated point
defects. The obtained defect distributions along the film thickness can then be
translated into a volume change distribution. This linear expansion/contraction
can finally be used in the equations of thermoelastic bending of beams to find the
solution to the problem of cantilever bending induced by ion irradiation.
Dense solid samples 83
Figure 5.2 – Deflection and curvature evolution of Au cantilevers with nominal
initial thickness of 200 nm (a, b) and 89 nm (c–f) irradiated with 30 keV Ga+
with parallel (a–d) and serial (e–f) scan strategies. The labels next to the lines
indicate the fluence in units of Ga ions/m2. The curvature 1/R was calculated
from experimental data using Eq. (5.17). The error bars of the experimental
data are not repeated in the calculated curvature representation.
84 Chapter 5
5.2 Model of ion induced bending
5.2.1 Point defect kinetics
The rates of ion implantation and defect generation as well as the sputtering
yield can be simulated for amorphous targets using SRIM [1]. The depth-dependent
PD generation rate K y and Ga implantation rate GaK y are calculated as
SRIMK y j K y , (5.1)
Ga
Ga SRIMK y j K y , (5.2)
where j is the flux of Ga+ ions at the film surface, is the volume per lattice site,
SRIMK y is the distribution of Frenkel pairs over the film depth per incident ion
and Ga
SRIMK y is the probability density to find the incident ion at a depth y after
it spent its kinetic energy in collisions with the target atoms.
According to SRIM simulations, PDs are generated mostly within a range of
about 30 nm during irradiation of Au with 30 keV Ga+ ions (Figure 5.3). The
“monolayer collisions – surface sputtering” option of SRIM was selected with
displacement energy of 44 eV [2].
Figure 5.3 – The production rate of PD and the implantation rate of Ga were
calculated with SRIM for 30 keV Ga ion implanted into amorphous Au target at
normal incidence.
Dense solid samples 85
The impingement of the film with energetic ions results in sputtering of the
film, i.e. movement or receding of the film boundary with the rate
dy
v j Ydt
, (5.3)
where Y is the sputtering yield (number of sputtered atoms per incident ion).
A set of rate equations for the concentrations of vacancies, self-interstitials, Ga
interstitials and Ga substitutional atoms was defined as follows:
α α ρ Δ1vv i i v iGa iGa v v v v v v
dCK y C D C C D C C Z D C D C
dt , (5.4)
α ρ Δ1 iv Ga i i v i i i i i
dCK y C C D C C Z D C D C
dt , (5.5)
α ρ ΔiGaGa Ga iGa iGa v i iGa iGa iGa iGa
dCK y K y C D C C Z D C D C
dt , (5.6)
α ρ GaiGa iGa v i iGa iGa Ga
dCD C C Z D C K y C
dt , (5.7)
where concentrations are defined per lattice site, α ,4 /i vR is the
recombination rate constant, , ~i vR a is the radius of spontaneous recombination (
a is the lattice spacing), , ,i v iGaD are the diffusion coefficients (𝐷𝑖,𝑖𝐺𝑎 ≫ 𝐷𝑣), ρ is
the dislocation density and ,i vZ are the efficiencies of dislocations as a sink for
interstitial atoms and vacancies [3]. For simplicity, the diffusion coefficient of self-
interstitials and interstitial Ga atoms are assumed to be the same.
The dislocations’ capture efficiency is estimated for the edge configuration as
1
,
,
2 lni v
i v
LZ
r
, (5.8)
where ,i vr is the capture radius of a PD by a dislocation and L is the mean spacing
between sinks. Dislocations attract interstitial atoms more strongly than vacancies
therefore i vr r and i vZ Z . PDs can be absorbed also by grain boundaries,
86 Chapter 5
however it is assumed here that the distance between grain boundaries is larger
than the dislocation spacing and the film thickness.
The physical meaning of different terms in the set of equations can be
explained conveniently by using Equation (5.4) as an example:
α i i vD C C : mutual recombination of vacancies and self-interstitial atoms
controlled by thermally-activated diffusion of SIA since in metals 𝐷𝑖,𝑖𝐺𝑎 ≫
𝐷𝑣;
α iGa iGa vD C C : recombination of implanted thermalized Ga with vacancies;
ρv v vZ D C : absorption of vacancies by dislocations;
Δ v vD C : diffusion of vacancies from the implanted layer to external
surfaces.
The rate theory equations do not deal with damage cascade collapse, i.e.
formation of either faulted or unfaulted loops of PDs. Zero boundary conditions are
used at both external surfaces for PDs Equations (5.4)-(5.6):
, , 0v i iGa
y vt
dC
dy
(5.9)
0
, ,0
v i iGa
y d
dC
dy
(5.10)
Concentration profiles of vacancies, substitutional Ga atoms and interstitial
atoms are shown in Figure 5.4 for various irradiation doses. The concentration
profiles move inside the film as the material is removed from the surface due to
sputtering. The concentration of vacancies is relatively high, whereas
concentrations of interstitial atoms are very low due to fast diffusion to external
surfaces. It should be noted that distributions of vacancies and Ga atoms in the
subsurface region of the film do not depend on time after a short transient period.
Vacancies are essentially immobile, since at the room temperature in metallic
systems the timescale of vacancy diffusion across the projectile range is (30 nm)²/
vD ~ 105 s.
Dense solid samples 87
Figure 5.4 – Concentration profiles of PD and implanted Ga atoms. Irradiation
time is indicated near curves. The left boundary of the film corresponds to the
irradiated surface and moves to the right (indicated with arrow) because of
sputtering. One second of irradiation time corresponds to a fluence of 3×1019
m−2.
5.2.2 Analogy to thermoelasticity
The implanted ions and point defect distributions result in an inhomogeneous
volume change. Bending is a consequence of the stress originating from the volume
change localized within the layer where implanted ions and point defects are
distributed. The relative volume change Ω V V associated with PDs is given
by the sum
Δω Δω
Ωω ω
v GaPD v GaC y C y (5.11)
88 Chapter 5
where Δωv is the vacancy relaxation volume and ΔωGa
is the excess volume
associated with a substitutional Ga atom. Here the contribution of interstitial
atoms is neglected because under typical conditions the remaining concentrations
of interstitial atoms are very small due to high diffusion mobility at room
temperature. They are accounted for in a second term regarding absorption of
point defects by dislocations bounding prismatic loops, which may result in
dislocation climb. The rate of local volume change due to dislocations is given by
Ω
ρdisli i i i iGa iGa v v v
dZ D C Z D C Z D C
dt . (5.12)
The relaxation volume of vacancies is negative in all metals. The excess
volume associated with substitutional Ga atom depends on the specific film
material, being negative in Au (Table 5.1). It can be deduced from the dependence
of lattice parameter on Ga concentration in solid solutions. The volume change due
to dislocations is always positive even when both interstitials and vacancies are
mobile, because of the dislocation bias to absorption of interstitial atoms [3].
Dense solid samples 89
Table 5.1 – Material parameters used in model calculations for gold
Parameter Value
Flux of Ga+ ions j , m-2s-1 3x1019
Temperature, T , K 300
Atomic volume, ω , m-3 1.7x10-29
Sputtering yield, Y (SRIM) 15.5
Recombination rate constant, α , m-2 3x1020
Diffusion coefficient of vacancies, vD , m2s-² 2.43x10-6exp(-0.89 eV/kBT)
[4]
Diffusion coefficient of interstitials, iD , m2s-1 5x10-6exp(-0.06 eV/kBT)
[5], [6]
Relaxation volume of vacancies, Δωv -0.37 ω [7]
Relaxation volume of interstitial atoms, Δωi ≈ω
Misfit volume of gallium, ΔωGa -0.076 ω [8]
Dislocation density, ρ , m-2 1011
Dislocation capture efficiency for SIA, iZ 3.3
Dislocation capture efficiency for vacancies, vZ 3
The bending that results from the inhomogeneous volume change
Ω Ω ΩPD disl can be evaluated using an analogy with thermoelasticity. It is well
known that thermoelastic stresses due to the linear thermal expansion αT y
result in bending of beams and multilayered materials [9]. In the equations of
thermoelastic bending of the cantilever beams we replace the linear thermal
expansion αT y with the linear expansion/contraction λ Ω y /3y and find
the solution to the problem of cantilever bending induced by ion irradiation. The
strain in the film is given by
ε ε0
1
2xx
dy
R
, (5.13)
90 Chapter 5
ε λ0
0
1
d
y dyd
, (5.14)
λ3
0
1 12
2
dd
y y dyR d
, (5.15)
where 0d d vt is the cantilever thickness, x is the coordinate in the direction
of the free end, R is the radius of curvature at the neutral plane and ε0 is the axial
strain in the x-direction at the neutral plane. The linear elastic stress in the
cantilever is given by
σ ε λ0
1/
2xx
dE y y
R
, (5.16)
where E is the Young’s modulus. The deflection of the beam towards the ion beam
(Figure 5.2) is related to curvature by
2 /2u x R . (5.17)
It should be noted that the radius of curvature (5.15) depends only on the
inhomogeneous volume change, with no explicit dependence on the elastic
modulus. One may expect that the volume change can generate high stresses
leading to plastic deformation. The bending direction depends on the sign of the
local volume change and the location of the layer with respect to the neutral axis in
which the radiation defects are created.
The local volume change due to vacancies and Ga atoms is negative (Figure
5.5) and results in bending towards the ion beam. Both vacancies and
substitutional Ga atoms contribute to volume change. The contribution of
substitutional Ga atoms, however, is small because of their small misfit volume.
Dense solid samples 91
Figure 5.5 – Depth dependence of volume change due to vacancies and
substitutional Ga atoms after irradiation for 10 s (ion fluence of 3×1020 m−2).
In the model described so far there is no accumulation of radiation damage
except for the subsurface layer. The steady-state profiles of PD and implanted Ga
ions are formed after a short transient period. For this reason the cantilever
curvature depends only on maximum vacancy concentration, the thickness of
damaged layer and the current cantilever thickness. Assuming a triangular profile
of vacancy concentration, it can be easily shown by direct integration of Equation
(5.15) that
Δω
ω3
3 21 1
3v
vmax
d t hC h
R d t
, (5.18)
where vmaxC is the maximum vacancy concentration, h is the thickness of
damaged layer and 0d t d vt is the cantilever thickness. The use of vmaxC
0.05 and 20h nm in Equation (5.18) reproduce well the time dependence of
curvature derived from the numerical solution. However, contrary to experimental
observations, Equation (5.18) does not contain a dependence of curvature on initial
thickness of the cantilever, i.e. the model predicts that upon reaching the same
92 Chapter 5
thickness cantilevers of different initial thicknesses would exhibit the same
curvature. Another factor is clearly lacking in this description.
5.2.3 Accumulation of interstitial clusters
Mobile interstitial atoms can form clusters which can evolve into dislocation
loops. Redistribution of material between top and bottom layers could occur due to
the migration of self-interstitial atoms and the interstitial clusters formed in
cascades.
Molecular Dynamics simulations carried out to date predict that a significant
fraction of interstitial defects created in cascades is produced in ‘sessile’ and
‘glissile’ clusters [3], [10]. The grouping of SIAs into clusters is significant because
these defects are thermally stable and they can migrate away from their parent
cascades. In the undamaged region (without cascades) the mobile clusters can form
sessile clusters due to collisions with each other and absorption of single SIAs.
Eventually these clusters may grow into dislocation loops. Contribution of one
atom in the sessile clusters to the volume increase of the film is about the relaxation
volume of a SIA, i.e. Δω Δω~i. At the same time, there is no accumulation of
excess volume in the irradiated subsurface region because cascades create new
clusters and destroy existing ones due to cascade overlap and the effect of
radiation-induced mixing [11], [12]. In addition, the ion beam sputters the surface
atoms and thus removes the damaged subsurface layers. The conclusion is that ion
beam irradiation not only creates point defect distributions within the ion
penetration range but also causes mass transfer across the whole thickness of the
cantilever.
Change of material microstructure well below the implantation depth (long-
range effect) was observed in various materials [13], [14]. Early transmission
electron microscope (TEM) studies [15]–[17] on Cu and Au foils bombarded with
1–5 kV Ar ions showed that interstitial clusters (in the configuration of Frank
sessile dislocation loops) are formed below the bombarded surface at a depth
remarkably larger than the calculated random range of Ar ions. The difference
between ion-induced damage depth and theoretically simulated range was
observed to be more prominent in fcc metals compared to bcc-Fe [18]. This fact was
attributed to a higher value of the Peierls force opposing dislocation glide in bcc
structures. Both Rutherford backscattering spectrometry (RBS) channeling spectra
Dense solid samples 93
and cross-sectional TEM characterization of single crystalline nickel and Ni-based
binary alloys irradiated with 3 MV Au ions clearly demonstrated that the range of
radiation-induced defect clusters far exceed the theoretically predicted depth in all
materials after high fluence irradiation [14]. The range of defect distribution
beneath the irradiated surface increased dramatically with increasing ion fluence.
The range of visible damage almost doubled when the irradiation fluence increased
from 2×1017 to 5×1019 m−2. The composition of the material was found to have a
great impact on defect distribution, suggesting very different defect migration
properties. Radiation defects in pure Ni stretched much deeper than in NiCo and
NiFe, indicating a higher defect migration rate in nickel for both high and low dose
irradiation.
It should be noted that according to MD simulation, vacancy clusters
(including mobile ones) may also form during early stages of cascade evolution.
However, small vacancy clusters and vacancy voids are thermally unstable [3]. For
this reason, their effect to bending is expected to be small as compared to
interstitial clusters. To estimate the contribution of PD and interstitial clusters to
cantilever bending, the analysis presented above is used to formulate a description
of radiation damage evolution in a thin film.
Assuming that sessile clusters form homogeneously across the undamaged
region, the volume increase during time interval can be estimated as
Ω0
dcl i
jnd dt
d h vt
, (5.19)
where dn is the total number of defects generated by a Ga ion and 𝜀 ≪ 1 is the
fraction of interstitial atoms in mobile clusters which go to formation of sessile
clusters. Assuming that weakly depends on the cantilever thickness, we find the
relative volume increase due to mass transfer from the irradiated zone:
Ω 0
0
lni dcl
n d h
Y d h vt
. (5.20)
The curvature can be calculated as
94 Chapter 5
Ω
Δω
ω
0
0
3
0
0 03
00
1 4
( ) 2
( )2 ln
d vt
cl
h
i d
d vty dy
R d vt
n h d h vt d h
Y d h vtd vt
(5.21)
The lower integration range is limited to the thickness of damaged layer.
There is no accumulation of excess volume in this region because cascades create
new clusters and destroy existing ones due to cascade overlap and the effect of
radiation-induced mixing [11], [12].
Figure 5.6 shows the fluence dependence of the relative volume increase in the
unirradiated region of the cantilever due to formation of sessile clusters. The
volume increase or swelling Ωcl is very high and reaches values of several tens of
percent (compare with Figure 5.5).
Figure 5.6 – Fluence dependence of the relative volume increase (swelling) in
the unirradiated region of the cantilever due to formation of sessile clusters for
cantilevers with three different initial thicknesses.
Dense solid samples 95
5.2.4 Gallium build-up
It is reasonable to assume that the mobile interstitial clusters formed during
cascade evolution are composed from Au and Ga interstitial atoms in the same
proportion as the concentration of Au and Ga in the irradiated zone. This means
that mobile clusters deliver Ga atoms to the zone beyond the implantation range.
Consider the cantilever at a steady state after a transient period of ion beam
irradiation. Assuming a triangular profile of substitutional Ga atoms in the
irradiated zone, from the balance of Ga fluxes:
2
ss d Ga
Ga
jn Cj YjC
, (5.22)
the surface concentration of Ga can be estimated:
1
2
s dGa
jnC Y
. (5.23)
Due to the flux of Ga atoms, the concentration of Ga in the unirradiated zone
increases with time similarly to the volume increase given by Equation (5.20):
0
0
ln2
s
d GaGa
n C d hC
Y d h vt
. (5.24)
The average Ga concentration is a sum of two contributions from irradiated
and unirradiated zones:
0 0
0 0
1 ln2
s
Ga dGa
d h vthC n d hC
d vt Y h d h vt
. (5.25)
5.2.5 Comparison with experiments
The set of equations formulated above was solved numerically by the method
of lines using the RADAU code [19]. The average curvature in the cantilevers’ free-
end half is plotted against fluence in Figure 5.7. Bending rates were inversely
related to thickness and the gradual increase in slope for each curve is hypothesized
to be an effect of the thickness reduction due to sputtering.
96 Chapter 5
Figure 5.7 – Fluence dependence of the curvature 1/R: comparison of
experimental data with model calculations. The initial thickness of cantilevers is
indicated near the corresponding curves. Dash-dotted curves are given by the
sum of contributions to bending from vacancies and substitutional atoms,
Equation (5.18), and sessile clusters, Equation (5.21), with the parameters ε =
1.4×10–2, nd = 440 and h = 20 nm. The dashed curve is calculated only with
contributions of vacancies and substitutional atoms to volume change,
Equation (5.18) (initial thickness 200 nm).
The red dashed curve in Figure 5.7 was calculated only with contributions of
vacancies and substitutional Ga atoms ΩPD; the black dash-dotted curves are given
by the sum of contributions to bending from vacancies and substitutional atoms
(Equation (5.18)) and sessile clusters (Equation (5.21)). It demonstrates that
bending caused by isolated PD is small as compared to bending due to sessile
clusters. The number of defects dn and the thickness of damaged layer h =20 nm
were estimated with SRIM (Figure 5.3); the parameter = 1.4x10-2 was adjusted to
reproduce the behavior of all three experimental curves. It can be concluded from
Figure 5.7 that Equation (5.21) well agrees with the experimental dependence of
curvature on fluence and initial cantilever thickness. Here, it is also relevant to
mention that the cluster hypothesis of bending is supported by a previous
observation that bending was not removed by annealing at 2000C for 5 hours,
Dense solid samples 97
whereas bending due to isolated PD would anneal completely. No recovery of
bending was also observed in ZnO nanowires annealed at temperatures up to
800oC [20].
Transmission electron micrographs of the 89 nm-thick sample with
progressive irradiation fluences are shown in 5.8. The grain structure is clearly
visible as well as growth twins from the films synthesis by evaporation. These
images provide a qualitative impression of the defect structure evolution with the
typical diffraction contrast associated with small PD clusters[21].
Figure 5.8 – Kinematic bright-field images of 89 nm thick Au film before
irradiation (a) and irradiated with 30 keV Ga+ ions to fluences of 6.7x1019 (b);
1.3x1019 (c) and 2.0x1020 m-2 (d).
The concentration of Ga for different fluences as measured by EDS is shown in
Figure 5.9. The values refer to the average concentration along the film thickness.
The measurements dispersion is attributed to the low concentrations of gallium,
close to the precision [22], but the trend agrees with the model prediction.
98 Chapter 5
Figure 5.9 – Concentration of Ga in the gold film with initial thickness of 144
nm as measured by EDS (circles) and predicted by the model.
5.3 Aluminium cantilevers
The choice of aluminum to further test the model based on the diffusion of
clusters was founded on the fact that it shares gold’s face centered cubic structure,
has a similar lattice constant (4.0495 vs. 4.0782 Å) [23], [24] but considerably
different atomic weights and electronic structures, which translates in distinct
diffusion constants and stacking fault energies. A rather unexpected and surprising
observation was made on these cantilevers. Literature on ion-induced bending
reports that the irradiated structures bend most of the time towards the incident
beam. Yoshida et al. [25] reported bending away from the beam when the ion
accelerating voltage was increased so that the implantation region would fall on the
opposite side of the cantilever’s neutral axis. The aluminium cantilevers under
irradiation with 30 keV Ga+ ions exhibited bending away from the beam at low
doses and a reversal of bending direction as irradiation progressed (Figure 5.10 and
Figure 5.11). This is quite an interesting observation from a fundamental viewpoint
but also for the field of fabrication of nano- and microsized objects.
Dense solid samples 99
Figure 5.10 – Ion-induced bending of aluminum cantilevers (d0=200 nm): a
initial position; b deflection after irradiation fluence of 8.0×1020 m−2 (before
direction reversal took place); and c deflection after fluence of 1.8×1021 m−2
Recent experiments confirmed formation of a dislocation network [26] in Au
nanoparticles and dislocation loops in Al thin films [27] at fluences typical for
focused ion beam (FIB) milling. TEM observation of as-fabricated Al nanopillars
revealed the formation of nm-sized dislocation loops [28]. According to [27], the
dislocation loops formed by a high energy Ga+ ion impact in Al at room
temperature are most likely of the interstitial type.
100 Chapter 5
Figure 5.11 – Deflection (left column) and curvature (right column) of cantilever
of initial thickness a, b 200 nm, c, d 144 nm and e, f 89 nm irradiated to various
fluences indicated in units of m−2. The lines at y=0 represent the initial
condition.
Dense solid samples 101
5.3.1 Kinetics of interstitial clusters
The observed change in bending direction means that the approach described
previously for gold cantilevers isn’t suitable anymore since the volume increase in
the undamaged layers cannot be considered uniform. The dynamical evolution of
clusters needs to be accounted for.
As in the previous model description, a cantilever beam of thickness 0d
irradiated with Ga+ ions is considered. The primary defects produced by
displacement cascades are now isolated PDs and mobile interstitial clusters. For
simplicity, it is assumed that clusters are of the same size m. Doing this has the
major advantage of reducing the number of fitting parameters. Mobile clusters
undergo random walks and form sessile clusters when they collide with each other.
The subsequent evolution of sessile clusters is not followed. In this study the most
important property of sessile clusters is their volume as a function of depth. The
terms for general capture of PDs by dislocations in e.g. Equations (5.4)-(5.7) are
substituted by individual equations for the diffusion of clusters.
The PD generation rate, Ga implantation rate per second and per lattice site
and sputtering yield are calculated with SRIM using Equations (5.1)-(5.3). The
time-dependent rate equations for concentrations of vacancies vC , self-interstitial
atoms iC , interstitial Ga atoms
iGaC , substitutional Ga atoms GaC , mobile clusters
mC and sessile clusters sC are as follows:
1vv v i i iGa iGa v v
dCK C C D C D C D C
dt , (5.26)
1 1 2i
v Ga m S
i i v i i
dCK C C Km C C
dt
D C C D C
, (5.27)
1iGaGa Ga iGa iGa v iGa iGa
dCK K C D C C D C
dt , (5.28)
1GaiGa iGa v Ga
dCD C C K C
dt , (5.29)
102 Chapter 5
21 2mv m m m m m m
dC KC KC D C D C
dt m
, (5.30)
2S
m m m S
dCD C KC
dt , (5.31)
where concentrations are defined per lattice site, ε is the fraction of clustered
interstitial atoms, dc
d
V
n
is a cluster destruction parameter, β is the probability
to destroy a cluster by a cascade, Vdc is volume of a displacement cascade, nd is the
number of defects of one type produced in a cascade and Di,v,iGa,m are the diffusion
coefficients (Dv < Dm < Di,iGa). The parameter dc
d
KVK
n
is the inverse lifetime
of a cluster due to ballistic and thermal spike effects [3], [11], [12]. Here dK n is
the impingement rate per second per unit volume. It is assumed that if a cascade of
volume Vdc develops in the neighborhood of a cluster, then it destroys a cluster with
the probability β<1. The rate constant for cluster formation is written in the so-
called Smoluchowski approximation [29]
8
m m md D
(5.32)
1 33
24
m
md
, (5.33)
where dm is the effective cluster diameter.
For mobile clusters, we use mixed boundary conditions
m
m y vt
y vt
dCC
dy
(5.34)
0
0
mm y d
y d
dCC
dy
, (5.35)
assuming that the flux of clusters at the external surfaces is proportional to the
concentration of clusters near surfaces, i.e., the surface is not a perfect sink for
Dense solid samples 103
clusters as distinct from PD. A physical interpretation of this assumption is related
to how fast interstitial clusters are transformed to be absorbed by external surfaces.
The volume change originating from clusters is now defined as
2 imS m Sm C C
(5.36)
and the inhomogeneous volume change Ω Ω Ω PD mS is used to substitute the
linear thermal expansion T y in the equations of thermoelastic bending of
cantilevers in the same way as described in the previous section (Equations (5.13)-
(5.15)).
Material parameters used in these simulations are listed in Table 5.2. The
cluster diffusion coefficient mD and parameters , and were adjusted to
reproduce experimental behavior of the cantilever curvature. In the model the
cluster migration energy is 0.3 eV, i.e., higher than reported by MD simulations for
1D migration [30], [31]. This difference can be understood considering that in our
case mD is the effective coefficient of 3D diffusion which requires cluster
reorientation.
104 Chapter 5
Table 5.2 – Material parameters used in model calculations for aluminium
Parameter Value
Flux of Ga+ ions j , m-2s-1 3x1019
Temperature, T , K 300
Sputtering yield, Y (SRIM) 3.8
Recombination rate constant, , m-2 4.8x1020
Rate constant for cluster formation, m , m-2 7.6x1020
Diffusion coefficient of vacancies, vD , m2s-² 7x10-6exp(-0.62 eV/kBT) [32],
[33]
Diffusion coefficient of interstitials, iD , m2s-1 5x10-6exp(-0.1 eV/kBT)*
Diffusion coefficient of clusters, mD , m2s-1 10-5exp(-0.3 eV/kBT)
Atomic volume, , m-3 1.66x10-29
Relaxation volume of vacancies, Δ v -0.33 ω
Relaxation volume of interstitial atoms, Δ i ≈ω
Misfit volume of gallium, Δ Ga -0.08 ω [35]
Number of atoms in cluster, m 4
Fraction of interstitial atoms produced in
clusters,
3.3
Destruction parameter, 3
Simulations carried out with SRIM (‘monolayer collisions-surface sputtering’
option with displacement energy of 25 eV [36]) predict that PD’s are generated
mostly within a range of about 50 nm during normal incident irradiation of Al film
with 30 keV Ga ions (Figure 5.12). According to recommendations [1] the data for
number of defects produced by an ion were divided by factor of 2, since SRIM
overestimates the number of displaced atoms as compared to more realistic MD
simulations.
* At room temperature simulation results do not depend on the diffusion coefficient of
interstitials at typical migration energies ~ 0.1; see also calculation in [34]
Dense solid samples 105
Figure 5.12 – PD production rate K and Ga implantation rate KGa calculated
with SRIM for 30 keV Ga+ ion implanted into amorphous Al target at normal
incidence. The Ga+ flux is 7×1018 m−2s−1 .
For a cantilever of initial thickness 200 nm, the concentration profiles of
vacancies and substitutional Ga atoms are shown in Figure 5.13. The concentration
profiles move inside the film as the material is removed from the surface due to
sputtering. Concentrations of SIAs and Ga interstitial atoms remain below 10−12
because of fast diffusion to external surfaces. At the simulation temperature, the
timescale of vacancy diffusion is (100 nm)²/Dv ~ 37 s. It should be noted that the
concentration of Ga atoms in the subsurface region of thickness 50 nm do not
depend on time after a short transient period.
Figure 5.13 – Concentration profiles of vacancies (a) and implanted Ga atoms
(b). The initial thickness is 200 nm, the Ga+ flux is 7 × 1018 m−2s−1. The left
boundary of the film is moving to the right (indicated with arrow) because of
sputtering.
106 Chapter 5
Figure 5.14 shows concentration profiles of mobile and sessile clusters. It is
seen that the sessile clusters are destroyed near the film surface exposed to the ion
beam. As the irradiated surface recedes due to sputtering, concentration of sessile
clusters increases in the region beyond the penetration range of ions.
Figure 5.14 – Concentration profiles of mobile (a) and sessile clusters (b). The
initial thickness is 200 nm; the Ga+ flux is 7 × 1018 m−2s−1 .
Accumulation of radiation defects results in a high swelling of material, up to
40% in the maximum (Figure 5.15). Vacancies that escaped recombination with
interstitials are removed from the system because of sputtering. Clustered
interstitial atoms survive in the form of sessile clusters, which evolve into
dislocation loops, giving rise to swelling of the film.
Dense solid samples 107
Figure 5.15 – Swelling Ω as a function of depth. The initial thickness is 200 nm;
the Ga+ flux is 7 × 1018 m−2s−1. The vertical solid lines show the position of the
film left boundary subjected to ion irradiation. The dash-dotted lines show the
center lines of the film (neutral axis).
Mobile interstitial clusters that form during cascade evolution can transport
Ga atoms to the unirradiated region. Assuming that the probability to find Ga
atoms in clusters equals the Ga concentration averaged over the implantation
profile thickness imp
GaC (see Figure 5.13b), the increment of Ga concentration in
this region during small time interval is written as
( ) ( )
1
tr imp mSGa Ga
mS
ddC C
(5.37)
108 Chapter 5
Since the Ga implantation profile practically do not change after a short
transient period, i.e. imp
GaC const , Equation (5.37) is easily integrated. The total
Ga concentration in the film is given by
( ) ( ) ( ) ( ) ln 1imp tr imp imp
Ga Ga Ga Ga Ga mSC C C C C (5.38)
5.3.2 Comparison with experiments
The curvature values in the cantilevers’ free-end half were averaged and
plotted against fluence as shown in Figure 5.16. Up to a fluence of 8×1020 m−2
(114 s) the 200 nm cantilever deflects away from the beam; the curvature is
negative. The reason is that most of the swelling occurs between the irradiated
surface and the neutral axis (Figure 5.15). As the film becomes thinner, the bending
direction changes.
Figure 5.16 – Fluence dependence of the curvature 1/R: comparison of
experimental data (symbols) with model calculations (solid lines). The initial
thickness of cantilevers is indicated near the corresponding graphs. The
negative curvature corresponds to downwards bending. The dot-dashed curve
is calculated only with contributions of vacancies and substitutional Ga atoms
to volume change, Equation (5.11) at ε = 0, when cascades do not produce
mobile clusters.
Dense solid samples 109
The minimum curvature reached by the films with initial thickness 200 and
144 nm is similar, while the film with d0=89 nm showed direction reversal at a
smaller negative curvature. This can be reasoned considering the induced swelling
distribution width relative to film thickness is larger in that case. Figure 5.16 shows
also the fluence dependence of bending with ε = 0, i.e. in a system without
production of clusters. In this case the negative value of curvature is due to the
positive volume misfit of substitutional Ga atoms which dominates vacancy
contribution. It is seen, as in the previous chapter, that distributions of isolated
point defects in the damaged region cannot produce the observed cantilever
curvature.
Figure 5.17 compares EDS measurements of Ga concentration in a 200 nm Al
cantilever with the model calculated depth profile and thickness average. The
model prediction that Ga can be found beyond the implantation peak agrees with
the measured values.
Figure 5.17 – Ga concentration in the Al film. a The depth dependence at fluence
8×1020 m−2. b Fluence dependence of Ga concentration averaged over film
thickness as measured by EDS (symbols) and predicted by the model
imp tr
Ga Ga C C (solid line); the dotted line corresponds to the average Ga
concentration without contribution of Ga transported by clusters. The initial
thickness is 200 nm; the Ga+ flux is 7 × 1018 m−2 s−1.
110 Chapter 5
5.4 Considerations about the model
5.4.1 Temperature increase due to irradiation
The model equations take into account the diffusion-limited reactions
between vacancies, interstitial atoms and dislocations, which are controlled by the
mean temperature of the cantilever. The ion irradiation can be expected to cause a
temperature increase in the cantilever.
d Q
Ga+
0 l
T0
-l0
T
x
T0
Figure 5.18 – Temperature distribution in the irradiated cantilever with heat
exchange at x=-l0.
In order to estimate an upper bound of temperature increase δT we assume
that:
the cantilever tip 0 x l is scanned with the Ga+ beam of average flux j;
all kinetic energy of ions transforms into heat in the film (no energy loss to
sputtering, defect creation etc);
the cantilever side-walls are thermally insulated (no heat loss by radiation
to a surrounding medium);
the heat dissipates through the unirradiated cantilever neck of the length l0
into the holder that is kept at constant temperature T0 (Figure 5.18).
The timescale of heat transfer across the 200 nm cantilever is very small and
estimated as κ2 / ~d 3×10-10 s, where κ = 1.27×10-4 m2s-1 is the thermal diffusivity
Dense solid samples 111
of gold. Therefore, on a larger timescale the production rate of heat per unit time
per unit volume can be considered to be uniform across the thickness of the
irradiated part of the cantilever:
00 at 0
/ at 0Ga
l xQ
jE d x l
(5.39)
The temperature distribution satisfies the steady state one-dimensional
equation
2
20
d Tk Q
dx (5.40)
subject to boundary conditions
0
0 x l
T T
(5.41)
0 x l
dT
dx
(5.42)
where k is the thermal conductivity.
The solution of Equations (5.40) and (5.41) is given by
0 0
0 2
0
at 0
/2 at 0
x l l xQlT x T
x l x l x lk
(5.43)
The maximum temperature increase is at the cantilever tip x l
2
021
2GajE l l
Tdk l
(5.44)
For typical parameters j = 3x1019m-2s-1, GaE = 30 keV,
0l l = 5 μm, d = 100
nm and the thermal conductivity of gold k = 314 Wm-1K-1 the temperature increase
is small δT = 0.17 K. Therefore in the cantilever temperature is assumed to be close
to room temperature.
112 Chapter 5
5.4.2 Effects of dynamic composition change
A more sophisticated Monte Carlo code Trydin [37], based on the sputtering
version of the TRIM program for multicomponent targets, computes range profiles
of implanted ions, composition profiles of the target and sputtering rates for a
dynamically varying target composition. However, while SRIM can be readily
downloaded from the internet, Trydin requires a rather bureaucratic procedure.
The concentration of Ga at the surface reaches values of about 5 at.%. By
running the SRIM simulation with the composite target Au95Ga5 it was found that
the influence of implanted Ga on the PD production rate and the sputtering yield is
in the range of a few percent. For this reason SRIM is considered an adequate tool
for our problem; usage of dynamic Trydin code for calculations of the PD
production rate and the sputtering yield would not change the results essentially.
Moreover the average Ga concentration derived from the model well agree with
experimental measurements (Figures 5.9 and 5.17).
5.5 Conclusions
The study demonstrates the ability to bend small structures with sub-micron
precision in a controlled manner using the focused ion beam. A quantitative
relationship between curvature and fluence has been derived which proved to be
helpful in controlling micro- and nanofabrication of small-sized products. Details
of our findings include:
A series of bending experiments with cantilevers made from Au and Al was
performed and different bending responses were observed. The Au
cantilevers bent always upwards (towards the beam) while the Al
cantilevers bent initially downwards and reversed direction as fluence
increased.
The experimental data are discussed in terms of local volume change due to
mass transfer from the cascade region to the so-called undamaged region.
To this end the model for defect diffusion kinetics in the crystalline thin
films under irradiation with energetic ions has been formulated. An
analogy with thermoelasticity is used to convert the inhomogeneous
volume change along the thickness of a free-standing cantilever into
bending curvature.
Dense solid samples 113
It was found that the bending under ion beam irradiation cannot be
explained just by the generation of isolated PDs and deposition of Ga ions
since their effects are too small. Our experimental findings are consistent
with the mechanism of gliding/diffusion of interstitial clusters, which
eventually form sessile clusters. The amount of material transferred to the
unirradiated zone grows with irradiation fluence and leads to swelling in
regions well beyond the penetration range of Ga ions.
The proposed model predicts that the range of ion-induced microstructural
changes far exceeds the SRIM predicted implantation depth. The
application of FIB to prepare samples causes accumulation of radiation
defects and may lead to contamination of the whole sample thickness with
Ga.
5.6 References
[1] R. E. Stoller, M. B. Toloczko, G. S. Was, A. G. Certain, S. Dwaraknath, and F. A. Garner,
“On the use of SRIM for computing radiation damage exposure”, Nuclear Instruments
and Methods in Physics Research, Section B: Beam Interactions with Materials and
Atoms, vol. 310, pp. 75–80, 2013.
[2] P. Jung, “Average atomic-displacement energies of cubic metals”, Phys. Rev. B, vol. 23,
nr. 2, pp. 664–670, jan. 1981.
[3] G. S. Was, Fundamentals of Radiation Materials Science: Metals and Alloys. Berlin
Heidelberg: Springer-Verlag, 2010.
[4] A. Seeger and H. Mehrer, “Interpretation of Self-Diffusion and Vacancy Properties in
Gold”, phys. stat. sol. (b), vol. 29, nr. 1, pp. 231–243, jan. 1968.
[5] S. M. Foiles, M. I. Baskes, and M. S. Daw, “Embedded-atom-method functions for the
fcc metals Cu, Ag, Au, Ni, Pd, Pt, and their alloys”, Phys. Rev. B, vol. 33, nr. 12, pp.
7983–7991, jun. 1986.
[6] S. M. Foiles, M. I. Baskes, and M. S. Daw, “Erratum: Embedded-atom-method functions
for the fcc metals Cu, Ag, Au, Ni, Pd, Pt, and their alloys”, Phys. Rev. B, vol. 37, nr. 17,
pp. 10378–10378, jun. 1988.
[7] W. Maysenhölder, R. Bauer, and A. Seeger, “Effect of three-body interactions on the
formation entropy of monovacancies in copper, silver and gold”, 1985.
[8] B. Predel, “Au-Ga (Gold-Gallium)”, in Ac-Au – Au-Zr, vol. a, O. Madelung, Red.
Berlin/Heidelberg: Springer-Verlag, 1991, pp. 1–5.
[9] R. B. Hetnarski and M. R. Eslami, Thermal Stresses -- Advanced Theory and
Applications. Springer Netherlands, 2009.
114 Chapter 5
[10] D. J. Bacon, F. Gao, and Y. N. Osetsky, “The primary damage state in fcc, bcc and hcp
metals as seen in molecular dynamics simulations”, Journal of Nuclear Materials, vol.
276, nr. 1, pp. 1–12, jan. 2000.
[11] V. Naundorf, “Diffusion in metals and alloys under irradiation”, Int. J. Mod. Phys. B,
vol. 06, nr. 18, pp. 2925–2986, sep. 1992.
[12] A. A. Turkin, C. Abromeit, and V. Naundorf, “Monte Carlo simulation of cascade-
induced mixing in two-phase systems”, Journal of Nuclear Materials, vol. 233, pp.
979–984, okt. 1996.
[13] D. Kiener, C. Motz, M. Rester, M. Jenko, and G. Dehm, “FIB damage of Cu and possible
consequences for miniaturized mechanical tests”, Materials Science and Engineering:
A, vol. 459, nr. 1, pp. 262–272, jun. 2007.
[14] C. Lu e.a., “Direct Observation of Defect Range and Evolution in Ion-Irradiated Single
Crystalline Ni and Ni Binary Alloys”, Scientific Reports, vol. 6, p. 19994, feb. 2016.
[15] H. Diepers and J. Diehl, “Nachweis und Entstehung von Zwischengitteratom-
Agglomeraten in ionenbestrahlten Kupferfolien”, phys. stat. sol. (b), vol. 16, nr. 2, pp.
K109–K112, jan. 1966.
[16] B. Hertel, J. Diehl, R. Gotthardt, and H. Sultze, “Formation of Interstitial Agglomerates
and Gas Bubbles in Cubic Metals Irradiated with 5 keV Argon Ions”, in Applications of
Ion Beams to Metals, Springer, Boston, MA, 1974, pp. 507–520.
[17] J. T. M. D. Hosson, “On the formation of argon-vacancy clusters in copper irradiated
with 4 to 6 kV argon ions”, physica status solidi (a), vol. 40, nr. 1, pp. 293–301, 1977.
[18] E. Friedland and H. W. Alberts, “Deep radiation damage in metals after ion
implantation”, Nuclear Instruments and Methods in Physics Research Section B: Beam
Interactions with Materials and Atoms, vol. 33, nr. 1, pp. 710–713, jun. 1988.
[19] E. Hairer and G. Wanner, “Solving Ordinary Differential Equations II. Stiff and
Differential-Algebraic Problems”, 1996.
[20] C. Borschel e.a., “Permanent bending and alignment of ZnO nanowires”,
Nanotechnology, vol. 22, nr. 18, p. 185307, 2011.
[21] B. L. Eyre, “Transmission electron microscope studies of point defect clusters in fcc and
bcc metals”, J. Phys. F: Met. Phys., vol. 3, nr. 2, p. 422, 1973.
[22] J. Goldstein e.a., Scanning Electron Microscopy and X-ray Microanalysis: Third
Edition, 3de dr. Springer US, 2003.
[23] W. Witt, “Absolute Präzisionsbestimmung von Gitterkonstanten an Germanium- und
Aluminium-Einkristallen mit Elektroneninterferenzen”, Zeitschrift für Naturforschung
A, vol. 22, nr. 1, pp. 92–95, 1967.
[24] A. Maeland and T. B. Flanagan, “Lattice Spacings of Gold–Palladium Alloys”, Can. J.
Phys., vol. 42, nr. 11, pp. 2364–2366, nov. 1964.
Dense solid samples 115
[25] T. Yoshida, M. Nagao, and S. Kanemaru, “Development of Thin-Film Bending
Technique Induced by Ion-Beam Irradiation”, Applied Physics Express, vol. 2, p.
066501, mei 2009.
[26] F. Hofmann e.a., “3D lattice distortions and defect structures in ion-implanted nano-
crystals”, Scientific Reports, vol. 7, p. 45993, apr. 2017.
[27] H. Idrissi e.a., “Point Defect Clusters and Dislocations in FIB Irradiated Nanocrystalline
Aluminum Films: An Electron Tomography and Aberration-Corrected High-Resolution
ADF-STEM Study”, Microscopy and Microanalysis, vol. 17, nr. 6, pp. 983–990, dec.
2011.
[28] S. Lee, J. Jeong, Y. Kim, S. M. Han, D. Kiener, and S. H. Oh, “FIB-induced dislocations
in Al submicron pillars: Annihilation by thermal annealing and effects on deformation
behavior”, Acta Materialia, vol. 110, pp. 283–294, mei 2016.
[29] M. v. Smoluchowski, “Versuch einer mathematischen Theorie der Koagulationskinetik
kolloider Lösungen”, Zeitschrift für Physikalische Chemie, vol. 92U, nr. 1, pp. 129–168,
1918.
[30] D. J. Bacon and Y. N. Osetsky, “Multiscale modelling of radiation damage in metals:
from defect generation to material properties”, Materials Science and Engineering: A,
vol. 365, nr. 1–2, pp. 46–56, jan. 2004.
[31] D. A. Terentyev, L. Malerba, and M. Hou, “Dimensionality of interstitial cluster motion
in bcc-Fe”, Phys. Rev. B, vol. 75, nr. 10, p. 104108, mrt. 2007.
[32] R. W. Balluffi, “Vacancy defect mobilities and binding energies obtained from annealing
studies”, Journal of Nuclear Materials, vol. 69–70, pp. 240–263, feb. 1978.
[33] J. Burke and T. R. Ramachandran, “Self-diffusion in aluminum at low temperatures”,
Metallurgical and Materials Transactions B, vol. 3, nr. 1, pp. 147–155, 1972.
[34] R. Qiu, H. Lu, B. Ao, L. Huang, T. Tang, and P. Chen, “Energetics of intrinsic point
defects in aluminium via orbital-free density functional theory”, Philosophical
Magazine, vol. 97, nr. 25, pp. 2164–2181, sep. 2017.
[35] B. Predel and D.-W. Stein, “Konstitution und thermodynamische eigenschaften der
Legierungen des systems aluminium-gallium”, Journal of the Less Common Metals,
vol. 17, nr. 4, pp. 377–390, apr. 1969.
[36] ASTM E521-16, “Standard Practice for Investigating the Effects of Neutron Radiation
Damage Using Charged-Particle Irradiation”, ASTM International, West
Conshohocken, PA, 2016.
[37] W. Möller and W. Eckstein, “Tridyn — A TRIM simulation code including dynamic
composition changes”, Nuclear Instruments and Methods in Physics Research Section
B: Beam Interactions with Materials and Atoms, vol. 2, nr. 1, pp. 814–818, mrt. 1984.
Nanoporous specimens
“Thirty spokes share the wheel's hub,
It is the center hole that makes it useful.
Shape clay into a vessel,
It is the space within that makes it useful.
Cut doors and windows for a room,
It is the holes which make it useful.
Therefore profit comes from what is there,
Usefulness from what is not there.” [1]
6.1 Introduction
Nanoporous metals have particularities that make them an especially
interesting case for investigating ion-induced bending. Studies have suggested that
nanoporous structures could be resistant to radiation damage when their
characteristic ligament size and the irradiation flux fall within an optimal window
[2]. The reasoning is that if the ligaments are too small, i.e. smaller than the
displacement cascade produced by each collision event, a cascade would cause the
ligament to break or collapse. If on the other hand the ligaments are too large,
consecutive cascades would result in accumulation of defects as in the bulk case.
The sweet spot would be a compromise between ligament size and irradiation flux
such that the irradiation-induced defects are able to diffuse and be annihilated at
the surfaces before another collision event happens in that vicinity.
Nanoporous specimens 117
Those arguments about diffusion lengths and collision frequency do not
consider, however, other effects that may arise from high specific surface area –
enhanced sputtering for instance. Irradiation of gold nanoparticles and thin films
with 25 keV Ga+ ions have been simulated with Molecular Dynamics and the
sputtering yield was found to be size-dependent with a maximum of roughly three
times that of a bulk surface [3]. This effect can be expected to result in high
amounts of internal redeposition and subsequent reconfiguration or densification
of irradiated nanoporous structures. Another important effect arising from ion
irradiation of nanoporous structures is ion-induced coarsening. This effect has not
received much attention in literature. Radiation-induced coarsening was
mentioned in [4], where the response of nanoporous Au samples to irradiation with
400 keV Ne2+ ions to a fluence of ~1019 m-2 was studied. These irradiation
parameters correspond to a uniform radiation damage of 1 dpa in the entire sample
thickness of 150 nm. In this Chapter, ion-induced coarsening is argued to be the
main factor causing the bending of nanoporous cantilevers.
6.2 Experimental considerations
The experimental procedures to cut the bending cantilevers from the free-
standing film in the present work had to be adapted in order to minimize ion-
induced changes in the nanoporous structure. Initial attempts to use the same
cantilever size as the bulk metal samples (around 5x2 µm²) resulted in a curvature
along the cantilever width, which restrains bending along its length, compromising
its measurement. Besides that, the bending of simultaneously irradiated cantilevers
varied considerably (Figure. 6.1) with some amount of lateral bending and cracking
along grain boundaries.
118 Chapter 6
Figure 6.1 – Preliminary attempts to perform bending experiments in
nanoporous cantilevers. Cracking at grain boundaries is clearly visible (red
arrows) as well as curvature induced along the cantilevers width.
In order to minimize these undesired effects, attempts were made to make the
cantilevers as narrow as possible. The intention was to decrease the probability of
the cantilevers being crossed by a grain boundary and at the same time to reduce
any heterogeneities along the cantilever width that could result in lateral bending
and complicate the analysis. By gradually milling the cantilevers lateral edges, free
standing cantilevers could be achieved. Even though the response to irradiation in
these cantilevers did become more uniform than the micro-sized ones shown in
Figure 6.1 the rate of bending seemed to be unexpectedly low and transmission
electron microscopy images of the cantilevers suggest that they experienced
densification (Figure 6.2).
Nanoporous specimens 119
Figure 6.2 – Scanning electron micrograph of ~150 nm-wide cantilever array
after irradiation (a) and bright-field image of as-cut cantilever showing
densified structure (b).
After some trial-and-error iterations, a succesful ion-cutting procedure for the
nanoporous cantilevers was achieved which involved two steps: first the cutting of
two parallel lines in the film, forming a bridge, and then the release of one end of
this bridge, forming a cantilever. When these parallel lines were cut below a certain
distance (roughly 200 nm) the bridges collapsed by themselves in the middle,
leaving equally sized edges in both sides similar to those seen in specimens
subjected to tensile ductile fracture (Figure 6.3).
Figure 6.3 – Preliminary attempts at producing narrow cantilevers with
collapsed bridges.
120 Chapter 6
The optimal width for the cantilevers to avoid densification and achieve
reproducible bending was found to be of around 300 nm (Figure 6.4).
Figure 6.4 – Bending cantilevers with optimized dimensions.
6.3 Ion-induced coarsening
The nanoporous specimens differed from the bulk counterparts in that the
sensitivity to irradiation was considerably higher, and also the displacement
induced by the cutting step, i.e. the position of the cantilevers in the beginning of
the experiments deviate slightly from the plane of the supporting film. These
observations suggest that the governing mechanism in these samples is different
from the solid case, where bending results from bulk mass transfer due to the
diffusion of crystallographic defects. Ion-induced bending of nanoporous
nanopillars was reported in the past by our group [5] and a beam mechanics
description was provided in terms of a stress profile being generated close to the
surface. That was attributed to an ion-induced distribution of vacancies. In the
present experiments, however, some samples had ligament sizes of the same scale
or smaller than that of the displacement cascades as simulated with SRIM for
normal incident 30 keV Ga+ ions. The implantation range for these parameters is
roughly 25 nm while the characteristic length of ligaments in nanoporous gold
prepared by free dealloying lies usually between 5 to 35 nm [6]. Since the stress
Nanoporous specimens 121
profile cannot be assumed to relate to a continuous distribution of defects, a new
hypothesis is presented based on coarsening of the nanoporous structure.
Nanoporosity formation via dealloying involves, in the example of nanoporous
gold, the concurrent removal of Ag and surface diffusion of Au atoms [7]. The
porous structure is very fine in the early dealloying stages and coarsens as it
progresses [8], [9]. Two mechanisms have been suggested to explain the
phenomenon of coarsening, namely localized plasticity in the ligaments and nodes
leading to network restructuring [10]; and surface mediated solid-state Rayleigh
instabilities leading to ligament pinch-off and bubble formation [11]. Both studies
presented reasonable causes for the observation made with HAADF-STEM
tomography of 1-5 nm voids within ligaments [12]; the first as trapping by the
collapse of neighboring ligaments and the second as the result of Rayleigh
breaking-up of the void phase. The voids and the high number of vacancies
resulting from dealloying are thought to play an important role in the process of
coarsening.
The surface instability argument received experimental support by positron
annihilation spectroscopy studies of thermal coarsening [13]. This technique
consists in measuring the time between the injection of positrons in a solid and the
detection of gamma-rays that result from their annihilation with electrons. Since
the local electron density in defects differs from the rest of the crystal, analysis of
the measured lifetimes and their frequencies allows for the characterization of the
number and size of lattice defects such as vacancies, dislocations and voids present
in a sample [14]. In nanoporous gold coarsened at various annealing temperatures,
two positron-lifetime components were identified and assigned to positron
annihilation in: (1) vacancy defects inside the ligaments and (2) larger vacancy
clusters/voids at the ligament-pore interface. The temperature at which the first
component started to decrease, pointing to vacancies annealing, coincided with the
onset of ligament growth. The non-linear variation of the second component with
temperature was interpreted as a signature of the surface instability previously
argued to drive the ligaments’ coarsening, with large voids at the surfaces being
broken down into smaller voids.
Interestingly, thermal coarsening has been shown to be sensitive to the
atmosphere as coarsening was hindered under vacuum or argon when compared to
diatomic nitrogen and oxygen at temperatures up to 400ºC [15]–[17]. The
122 Chapter 6
calculated activation energy for coarsening under vacuum was very close to that of
lattice diffusion in gold, suggesting that surface diffusion is not an active
mechanism in that case. It was proposed that the chemical adsorption of oxygen
and nitrogen (which is generally inactive with gold) causes, as a result of their high
electronegativity, a decrease in the bonding energy of the top-layer gold atoms and
hence facilitates their diffusion as adatoms. A previous study on clusters of
adatoms and vacancies on gold (111) surfaces showed a similar behaviour, where
the area of those features was observed to decay linearly with time under
atmospheric conditions but to be stable under ultrahigh vacuum conditions [18].
Those observations were found to agree well with a model where the mechanism
activating mass flow was the generation of vacancies and adatoms at step edges
instead of the enhancement of free adatoms diffusion. That interpretation was
reinforced by in-situ E-HRTEM studies where nanoporous gold was observed to
coarsen during catalytic CO oxidation [19], [20] by the diffusion of gold atoms at
surface steps.
In nanoporous gold samples subjected to thermal coarsening above 500°C,
grain-boundary cracks were observed as a secondary effect [21]. This was reported
as an observation that was not the focus of the referenced investigation.
Interestingly, coarsening and grain-boundary cracks were also reported to result
from ion irradiation at very low doses , i.e. a single imaging scan with the focused
ion beam [22], and highly strained ligaments were observed across the FIB-induced
cracks. A hint of ion-induced coarsening was also reported in [9], where a
nanoporous gold pillar with diameter of 4 µm was produced by ion milling and
analyzed with X-ray microdiffraction. In the course of the present experiments a
coarsened structure was observed under the scanned lines made amid the
cantilevers cutting steps (Figure 6.5) as well as in the irradiated cantilevers (Figure
6.4).
Nanoporous specimens 123
Figure 6.5 - Coarsened ligaments in the ion-irradiated cutting tracks during the
preparation of nanoporous cantilevers.
An intuitive picture of ion-induced coarsening can be provided based on the
observations above, as follows. In the case of ion-irradiation, one may assume that
a large amount of ligaments will rupture if their sizes are comparable to that of the
displacement cascades. This collapse of ligaments is a pre-requisite for coarsening
according to MD simulations where the decrease of surface area was shown to lag
behind the decrease in topological genus [11]. Additionally, ion-irradiation is
expected to generate point defects that will diffuse and be absorbed preferentially
at the surface steps, leading to either enhanced surface diffusion or direct surface
smoothing via bulk transport [23]. In MD simulations of irradiation of gold
ligaments with 1.5 keV PKA, coarsening was shown to be dose-rate independent,
suggesting that the effect is due to individual cascade effects [4]. Ion-induced
coarsening can then be related to thermal coarsening; assuming (i) direct breakage
of small ligaments by cascades instead of spontaneous collapse by Rayleigh
instabilities and (ii) surface smoothing due to the radiation-induced generation of
point defects instead of thermal activation.
Based on this reasoning, we propose a model that assumes a coarsening
gradient as the main factor generating the bending moment in irradiated
nanoporous cantilevers.
124 Chapter 6
6.4 Model of (IB)² for nanoporous gold
In Chapter 5 we have shown that the mass transfer from the irradiated
cantilever layer to the deeper layers due to transport of interstitial clusters beyond
the Ga penetration range results in cantilever bending. Notwithstanding, ion-
induced bending of nanoporous cantilevers cannot be explained by this
mechanism. The obvious reason is that interstitial clusters cannot escape the
ligament in which they were produced.
To construct a model of nanoporous sample bending under ion irradiation we
use two major experimental observations: (i) nanoporous structure undergo
radiation-induced coarsening and (ii) nanoporous samples are more sensitive to
ion irradiation than solid samples.
The assumption of the model is that ion-induced coarsening (or sintering) of
the nanoporous structure is accompanied with compaction which results in
contraction of the subsurface ligament structure subjected to irradiation. The
nanoporous cantilever is freestanding in the sense it is without constraints against
mechanical deformations; therefore the inhomogeneous volume change causes the
cantilever to bend towards the ion beam. It seems that the ion-induced coarsening
is similar to the effects of annealing on nanoporous gold structures which leads to
coarsening of pores and ligaments [24]–[28]. The underlying mechanism of
thermal coarsening and ligament growth is related to surface diffusion of gold
atoms [11], [28]. The driving force for coarsening is the reduction of total surface
area of ligaments. Evidently, ion beam irradiation enhances atomic mobility in a
similar way as temperature increase. Several mechanisms can control ion-induced
coarsening/sintering:
migration of radiation-induced interstitial atoms to the ligament surface and
surface diffusion of adatoms as controlled by the ligament surface
curvature;
migration of small point defect clusters inside ligaments and to ligament
surfaces.
As discussed in Chapter 5, high mobility of PD clusters of both vacancy and
interstitial types has been confirmed experimentally (for recent results see [29],
[30]) and by means of MD simulations [31]–[33].
Nanoporous specimens 125
In the following our goal is to estimate the relative volume change of the
cantilever material as a function of ligament mean size, ion fluence and depth from
the sample surface. Consider time evolution of a small volume element ( , )V y t
inside the ion-irradiated cantilever, here y is the distance from the cantilever
surface. The volume of all ligaments AuV in the volume ( , )V y t does not change
during irradiation, while volume of pores pV decreases with time due to coarsening
and shrinkage of empty space.
Ion fluence
Ga+
Contraction Contraction
00 pAu VVV pAu VVV
x
y
Figure 6.6 – Schematic view of coarsening and contraction of cantilever
irradiated with ions.
The relative volume change due to coarsening is given by
0 0
( , )( , )
p
Au p
V y ty t
V V
, (6.1)
where 0( , ) ( , ) 0p p pV y t V y t V is the volume decrease from the initial state.
The volume fractions of gold and the porosity are
126 Chapter 6
0
1( , )
1Au Auf y t f
, (6.2)
0( , )
1
p
p
ff y t
, (6.3)
where 0Auf and 0pf are the initial volume fractions.
It is likely that the coarsening is isotropic in all directions, therefore the
relative change in linear dimensions – the linear contraction – is estimated as
1 3
( , ) 1 ( , ) 1y t y t . (6.4)
As in Chapter 5 we use the analogy with the thermoelasticity. Unlike the solid
cantilever the properties of the irradiated nanoporous cantilever such as the linear
contraction and Young’s modulus E depend on position; hence we use formulas of
Ref. [34] describing the thermal stress in an inhomogeneous beam, in which the
linear expansion and Young’s modulus are functions of position y as shown in
Figure 6.6. In equations of thermoelastic bending of the cantilever beam [34] we
replace the linear thermal expansion ( )T y with the linear contraction ( , )y t and
obtain the curvature radius in our problem
0 1
20 2 1
1 T E T E
E E E
M I P I
R I I I
, (6.5)
where
( )
0
( ) ( , ) , 0, 1, 2d t
nEnI t E y t y dy n , (6.6)
( )
0
( ) ( , ) ( , )d t
TP t E y t y t dy , (6.7)
( )
0
( ) ( , ) ( , )d t
TM t E y t y t ydy . (6.8)
Integration is carried out over the cantilever thickness ( )d t which decreases with
time due to ligament sputtering by incident ions and compaction due to coarsening.
Nanoporous specimens 127
0
0
( ) ( , )d
d t d Y jt y t dy
, (6.9)
where Y is the sputtering yield, j is the ion flux and is the atomic volume
(Chapter 5); the upper integration limit d is selected beyond the ion penetration
range in the undamaged region where 0 .
Note that in Eqs. (6.6)-(6.9) the modulus ( , )E y t is the effective Young’s
modulus of the volume element ( , )V y t , which should not be confused with
Young’s modulus of a separate ligament. In our model we use the linear theory of
elasticity. At high ion fluences, when the cantilever deflections is large, one may
expect plastic deformation of ligaments and even microcracking on the far side of
the cantilever.
6.4.1 Kinetics of coarsening
To find the linear contraction ( , )y t we construct below a simple model with
unknown parameters which we will determine from comparison with experimental
data. The total surface area of ligaments S , the characteristic mean ligament size
L and the characteristic mean spacing between ligament (pore size) pL is a
minimal set of variables describing the structure of the volume element under
consideration (Figure 6.6). Parameters defined above are related to gold, pore and
total volumes
AuV ASL , (6.10)
p pV BSL , (6.11)
0 0 0 0 0pAS L BS L V , (6.12)
where A and B are factors associated with the topology of a particular
nanoporous structure. In the following we assume that A and B do not change
essentially with time.
It is natural to assume that (i) the structure with small ligaments coarsens
faster and (ii) the coarsening rate is proportional to the defect production rate
128 Chapter 6
( ) ( )SRIMK y j K y (Chapter 5). Hence the mean ligament size changes with time
according to
1
( )Au
n
dLK y
dt L
, 1n , (6.13)
where n depends on dominating mass transport mechanism, 0Au is a constant.
The solution to this equation can be represented in the form
3
30 ( )
n
n nAu SRIML L n K y
, (6.14)
where is the ion fluence, Au is the dimensionless constant.
In a similar way we construct the equation for the pore size pL
1
( )pP
mp
dLK y
dt L
, (6.15)
where 0p is a constant, the minus sign corresponds to decrease of mean spacing
between ligament. The solution to Eq. (6.15) is given by
3
30 ( )p
m
m mp p SRIML L m K y
, (6.16)
where p is the dimensionless constant.
In the volume element under consideration the volume of solid gold does not
change with time, i.e. SL const , therefore the fluence dependence of the total
area is given by
13
30 00 0 0 ( )
nn
nAu SRIM
S LS S L L n K y
L
. (6.17)
Using Eqs. (6.10),(6.16) and (6.17) we find the pore volume
00
0
( , )( , )
( , )
p
p p
p
L yLV y V
L L y
(6.18)
and the fluence dependence of the relative volume change
Nanoporous specimens 129
00
0
( , )( , ) 1 1
( , )
p
Au
p
L yLy f
L L y
. (6.19)
Using Eq. (6.12) we can obtain the relation between the initial size of pores
0pL and other parameters of the nanoporous structure
00 0
0
1 Aup
Au
fL gL
f
, (6.20)
where g A B
.
6.4.2 Young’s modulus of nanoporous Au
The curvature radius given by Eq. (6.5) depends on Young’s modulus ( , )E y t .
Note that the curvature radius does not change if we substitute ( , ) nAuE y t E for
( , )E y t , where nAuE is Young’s modulus of the undamaged nanoporous material.
This is an important property of the bending phenomenon: it is the relative
modulus (‘relative stiffness’) ( , ) nAuE y t E that contributes to bending rather than
the absolute value of Young’s modulus. Schematically the relative modulus is the
ratio
( , )
nAu
E y t
E
upper layer
bottom layer
E
E, (6.21)
which controls the amount of bending caused by coarsening. A higher modulus of
the upper layer facilitates bending and vice versa: if the upper layer is more
compliant than the bulk the bending is less pronounced.
The Gibson-Ashby scaling relation [35], [36] is used frequently to describe the
dependence of Young’s modulus on density of porous materials, which in our
problems reads
0
( , )1 ( , )
k
kAu
nAu Au
fE yy
E f
, (6.22)
where ~ 2k for the classical Gibson-Ashby relation.
130 Chapter 6
Being designed for low-density cellular (foam) materials, the Gibson-Ashby
relation does not automatically apply to nanoporous gold. The network topology of
a nanoporous structure differs from the cellular architecture of foams. However, in
some materials, this relation is working quite well. Liu and Jin [37] report that in a
Pt-doped nanoporous gold, in which the network connectivity maintained constant,
Young’s modulus varied with relative density according to power-law relation with
an exponent 2.2 0.1k , which agree well with the classical Gibson-Ashby low (
2k ).
It has also been observed that the elastic modulus of nanoporous gold
decreased by more than one order of magnitude in structure coarsening [37], [38]
during which the relative density changed only little. Mathur and Erlebacher [39]
examined Young’s modulus of freestanding, large-grained and stress-free
nanoporous gold films of 100 nm thick over a controlled range of ligament sizes.
Measurements by a buckling-based method [40] have shown a dramatic increase in
the effective Young’s modulus with decreasing ligament size, especially below 10
nm.
All these results indicate that the mechanical response (including the elastic
modulus) of nanoporous metals depends on not only the relative density but also
the topology of the nanoporous structure. The network connectivity [37], [38], [41]
has been suggested as a structural feature that accounts for the anomalous behavior
of Young’s modulus of nanoporous gold. Indeed, strength and stiffness would
decrease if a good number of ligaments in nanoporous metal are ‘broken’ and the
network structure is less connected compared with a foam materials with the same
density (Figure 6.7).
The hypothesis seems to be promising, but then questions arise as to how the
network connectivity parameter can be defined and experimentally measured in
the nanoporous metals, and how it can be correlated to mechanical properties.
Mameka et al [41] proposed a connectivity parameter, which is a ratio between the
diameter of the unit cell of nanoporous structure and the diameter of the closed
rings in a network of load-bearing paths.
Nanoporous specimens 131
Figure 6.7 – Schematic illustration of nanoporous structure with broken or
dangling ligaments (left), and a mechanically equivalent structure obtained by
removing dangling ligaments (right). The density of material with only load-
bearing ligaments is less than the total density (after Ref. [38])
Figure 6.8 shows the influence of coarsening during thermal annealing on the
elastic modulus of the as-dealloyed nanoporous samples [37]. It was found that
Young’s modulus of high density nanoporous samples changed a little when the
structure size was coarsened from 5 nm up to 300 nm during annealing. On the
contrary the elastic modulus of low density samples decreased dramatically after
coarsening. Note a nonmonotonic behavior of Young’s modulus at volume fraction
of gold less than 0.35.
In literature the coarsening of nanoporous structure was proposed to proceed
by the ‘pinch-off’ of some ligaments due to Rayleigh instability [11], [42], [43] and
simultaneous removal (healing) of dangling ligaments by surface diffusion
[11], [44]. Evidently, low density samples with thin ligaments are more prone to
‘pinch-off’ damage caused by annealing which result in decrease of network
connectivity. Increasing ligament size and relative density would decrease the
aspect ratio of ligaments, which decreases the probability of ‘pinch-off’ and
prevents the network connectivity reduction as was observed in [37].
132 Chapter 6
Ligament size, nm
Young m
odulu
s, E
np, G
Pa
Figure 6.8 – Variations of Young’s modulus during coarsening of nanoporous
samples with various initial relative densities, which may increase during
coarsening particularly for low-density samples (after Ref. [37]).
We believe that the influence of ion-induced coarsening on the properties of
nanoporous metals is similar to the thermal coarsening discussed above. Moreover,
the ion irradiation with its sputtering effects is especially efficient to destroy thin
ligaments thereby reducing the connectivity and Young’s modulus. As the sample
become denser with larger ligaments, the radiation-induced coarsening results in a
Young’s modulus increase. In our model the relative modulus is a function of
ligament and pore sizes and initial volume fraction of gold
00 0
0 0
( , )( , )1
( , )
kk
pAuAu Au
nAu Au p
L yf LE yf f
E f L L y
. (6.23)
To model the nonmonotonic dependence on ligament size we will adjust the power
law exponent k for ‘small’ and ‘large’ ligament sizes.
6.5 Comparison with experiments
Figure 6.9 shows the fluence dependence of curvature for nanoporous samples
of different thicknesses and ligament sizes, which are denoted as
Nanoporous specimens 133
‘Thickness/Ligament’. One set of samples was irradiated with 20 keV Ga+ ions; all
other samples were irradiated with 30 keV Ga+ ions. The samples had their
structure tailored by means of varying dealloying times and one was coarsened
thermally. The precursor AuAg alloy foils have thicknesses ranging between 100
and 200 nanometers and therefore coarsening is restricted to that size. In fact, the
sample that was thermally coarsened to the point where the ligament sizes reached
the same scale of the film thickness presented the same bending behaviour as solid
annealed Au samples.
0 1x1019 2x1019 3x1019
0
1
2
3
4
Thickness/Ligament [nm]
128/21
144/47
196/46
196/46 (EGa
= 20 keV)
103/78
89/89 Bulk sample
Curv
atu
re, m
-1
Fluence, m-2
Figure 6.9 – Fluence dependence of curvature for np-samples with different
thicknesses and characteristic lengths.
Figure 6.9 exemplifies the increased bending sensitivity to ion fluence of
nanoporous samples when compared to bulk samples (labelled as 89/89). The
increased sensitivity of the nanoporous samples is evidenced by the fact that the
thickness of the solid cantilever used for comparison was nearly half of the
nanoporous ones.
The four other samples invite a closer look. The argument of ion-induced
ligament collapse and coarsening would predict higher bending rate per fluence for
finer nanoporous structures. However, the finer of the structures produced in this
134 Chapter 6
study, which was dealloyed for 30 minutes (L~25 nm), presented a lower bending
dependence on fluence than the sample dealloyed for 3 hours (L~45 nm). The
reason for such a behavior is the anomalous dependence of Young’s modulus on
ligament size during coarsening of small-sized nanostructures (see below).
0 10 20 30 40 50 60 70
0
5
10
15
20
Ga+ irradiation
C 20 keV
B 30 keV
PD
pro
ductio
n r
ate
KS
RIM
, n
m-1
Depth, nm
Figure 6.10 – The distribution of Frenkel pairs over the sample depth per
incident Ga+ ion at normal incidence. The ‘monolayer collisions – surface
sputtering’ option of SRIM [45] was selected with displacement energy 44 eV
[46].
Depending on the size distribution of ligaments an incident ion can travel
through a ‘surface’ ligament and hit another ligament in a deeper location. On
average in nanoporous materials the ion penetration range is larger than in solids.
To take this into account in our model, which is a mean-field approximation, the
defect generation rate ( )SRIMK y is calculated for an amorphous target of 50%
density of solid gold (Figure 6.10). In reality, nanoporous gold density depends on
fluence and depth y , however for our qualitative model such a refinement is a
second order correction.
The initial fraction of gold in all calculations is 0.3. The initial ligament size
0L was measured in a sample micrograph. To stay in framework of our model the
initial value of 0pL was evaluated using Eq. (6.20) with g 0.45. Other
Nanoporous specimens 135
parameters of the model were adjusted to reproduce the experimental dependence
of curvature radius on fluence: 410Au , 23.0 10p , 2n , 0.5m .
First we show dependence of nanoporous gold geometrical parameters on
fluence of 30 keV Ga+ ions. Figure 6.11 demonstrates that in the near surface region
the volume fraction of gold increases drastically after irradiation to a fluence of
2x1019 m-2 which is a typical maximum fluence in bending experiments with
nanoporous Au
0 20 40 60 80 1000
20
40
60
80
100
Fluence
2x1018 m-2
2x1019 m-2
Au
fra
ctio
n,
%
Depth, nm
a
0 20 40 60 80 1000
20
40
60
80
100b
Fluence
2x1018 m-2
2x1019 m-2
Po
rosity,
%
Depth, nm
Figure 6.11 – Depth dependence of gold volume fraction (a) and porosity (b) at
two values of ion fluences. The initial ligament and pore sizes 0L 46 nm and
0pL = 42 nm.
Figure 6.12a shows depth dependence of ligament size and pore sizes after
radiation-induced coarsening (initial parameters and fluences corresponds with
Figure 6.11). The fluence dependence of the ligament size at a distance of 5 nm
from the surface is presented in Figure 6.12b. This distance can be considered as a
maximum depth which can be seen by SEM in real samples.
136 Chapter 6
0 20 40 60 80 1000
20
40
60
80
100
2x1019 m-2
Siz
e,
nm
Depth, nm
a2x1018 m-2
Ligaments
Pores
0.0 5.0x1018 1.0x1019 1.5x1019 2.0x1019
0
20
40
60
80
100b
Siz
e,
nm
Fluence, m-2
Ligaments
Pores
Figure 6.12 – (a) Depth dependence of ligament size (red) and pore sizes (blue)
at fluences 2x1018 m-2 (dashed curves) and 2x1019 m-2 (solid curves). (b) Fluence
dependence of the ligament size at a depth of 5 nm. Solid and dashed curves
corresponds to ligament and pore sizes respectively.
0.0 5.0x1018 1.0x1019 1.5x1019
40
60
80
100
120
L+
Lp (
FF
T s
ize
), n
m
Fluence, m-2
Figure 6.13 - Evolution of characteristic FFT size with ion irradiation for two
distinct initial average ligament sizes (symbols) Solid curves represents the sum
of ligament and pore sizes calculated with the model.
Figure 6.13 shows the values of the FFT characteristic sizes for samples with
initial average ligament sizes of ~ 27 and ~ 42 nm. The FFT characteristic size is
defined as the position of the first maximum in the dependence of FFT intensity
Nanoporous specimens 137
versus wave vector. For comparison the model predictions for the sum of ligament
and pore sizes are plotted. Examples of the nanoporous structure at different
fluence values are shown in Figure 6.14. Densification can be clearly seen at the last
irradiation step for the sample with smaller ligament size (Figure 6.14c).
Figure 6.14 – Evolution of nanoporous structure under irradiation for samples
with 0L ~ 27 nm (a-c) and 42 nm (d-f) showing unirradiated structure,
intermediate (~7.5×1018 m-2) and final fluences shown in Figure 6.14 In all SEM
micrographs the field of view is 1x1 µm2.
In our experiment it was observed that the bending rate of nanoporous
structures with ligament sizes of ~ 20 nm was lower than the bending rate of
nanoporous structures with 46 nm ligaments. Such a behavior cannot be explained
with the power law exponent 0k in the Gibson-Ashby relation, Eq. (6.23). We
managed to describe bending of nanoporous structures with small ligaments using
the power law exponent 2k . Figure 6.15 shows depth dependence of Young's
modulus (Eq. (6.23)) at a fluence of 1019 m-2 for two types of scaling relation: for the
initial ligament size 20 nm the exponent 2k in Eq. (6.23), for the initial
ligament size 46 nm the exponent 2k . In case of small ligament sizes we
138 Chapter 6
multiplied the right-hand-side of Eq. (6.23) by a factor of 10 in order to emphasize
that undamaged nanoporous material has a higher Young's modulus than the
irradiated subsurface material (as we mentioned above multiplication of the
modulus by a constant does not change the curvature).
0 10 20 30 40 50 60 70
0
2
4
6
8
10
L0 =46 nm k = 2
L0 =20 nm k = -2
Mo
du
lus (
norm
.)
Depth, nm
Figure 6.15 – Depth dependence of Young modulus at a fluence of 1019 m-2 for
two types of scaling relation (6.23).
0.0 5.0x1018 1.0x1019 1.5x1019 2.0x1019
0
2
4
6
8
10
L0 =20 nm k = -2
L0 =46 nm k = 2
a
Mo
du
lus (
no
rm.)
Fluence, m-2
20 30 40 50 60 70 80 900
2
4
6
8
10
L0 =20 nm k = -2
L0 =46 nm k = 2
Mo
du
lus (
no
rm.)
Ligament size, nm
b
Figure 6.16 – Young modulus as a function of fluence (a) and ligament size (b)
at a depth of 20 nm.
Figure 6.16 shows behavior of Young's modulus at a depth of 20 nm. As in
Figure 6.15, in Eq. (6.23) for the initial ligament size 20 nm the exponent 2k ,
whereas for the initial ligament size 46 nm the exponent 2k Figure 6.16b was
plotted as a parametric dependence (20 nm, )K versus (20 nm, )L , where the
Nanoporous specimens 139
parameter is the fluence that changes from 0 to 3x1019 m-2. Note the similarity of
model calculations (Figure 6.16b) with Figure 6.8 which shows a compilation of
experimental data on the effect of annealing on Young's modulus.
In Figs. 6.17-6.19 we compare the curvature radius as predicted by the model
with the curvature found from processing of the experimental data. Figure 6.17
shows that a good agreement with experimental data can be obtained if we use
different power law exponent k for ‘small’ and ‘large’ ligament sizes.
0.0 5.0x1018 1.0x1019 1.5x1019
0
1
2
3
4
Model
L0 = 21 nm k = -2
L0 = 21 nm k = 2
L0 = 46 nm k = 2
Curv
atu
re, m
-1
Fluence, m-2
Figure 6.17 - Fluence dependence of cantilever curvature: influence of initial
ligament size and scaling relation type. Symbols denote experimental data for
the initial ligament size 21 nm (squire) and 46 nm (circles). The blue curve was
calculated with the classical Gibson-Ashby relation (k = 2) for the initial
ligament size 46 nm. The red solid curve was calculated with k = -2 in Eq. (6.23)
for the initial ligament size 21 nm. The dashed curve corresponds to calculations
for the initial size 21 nm with the classical Gibson-Ashby relation (k = 2). Ga+
ion energy is 30 keV.
Figure 6.18 compares the predicted fluence dependence of the cantilever
curvature with experimental data at initial thicknesses 144 and 200 nm. In both
cases the initial ligament size is the same 0L = 46 nm, and the classical Gibson-
Ashby relation (k = 2) was used in calculations.
140 Chapter 6
0.0 5.0x1018 1.0x1019 1.5x1019
0
1
2
3
4 Thickness/Ligament [nm]
144/47
196/46
Model 144/47
Model 196/46
Cu
rva
ture
, m
-1
Fluence, m-2
Figure 6.18 – Influence of initial thickness on the fluence dependence of
curvature. Symbols denote experimental data for the sample thickness 144 nm
(squires) and 200 nm (circles). The model curves are calculated with k = 2 in
Eq. (6.23). The initial ligament size is 46 nm. Ga+ ion energy is 30 keV.
Figure 6.19 shows the fluence dependence of cantilever curvature at Ga+ ion
energies 20 and 30 keV. The model curves reproduce the experimental trend. As it
could be expected the ion beam with lower energy produces lower bending rates,
because of smaller ion penetration range and lower efficiency to create radiation
defects and the consequent structure coarsening.
Nanoporous specimens 141
0 1x1019 2x1019 3x1019
0.0
0.5
1.0
1.5
2.0
2.5
Curv
atu
re, m
-1
Fluence, m-2
20 keV
30 keV
Thickness 200 nm
Ligaments 46 nm
Figure 6.19 – Influence of Ga+ ion energy on the fluence dependence of
curvature. Symbols denote experimental data for the ion energy 20 keV
(squares) and 30 keV (circles). The model curves (solid lines) are calculated
with k = 2 in Eq. (6.23). The initial ligament size is 46 nm.
As an additional test of the model we calculated the thickness decrease of
irradiated nanoporous samples. The total thickness decrease exhibits weak
dependence on ligament size. At the ligament size 46 nm and the fluence
1.5x1019 m-2 the thickness decrease due to sputtering and coarsening is estimated as
sputteringd Y -4 nm (6.24)
0
( , )d
coarseningd y t dy
-7 nm (6.25)
The sum of these values well agree with measurements. However, it should be
noted that experimental measurement of thickness decrease is difficult, because it
is comparable with the ligament size and surface irregularities.
142 Chapter 6
6.6 Conclusions
The stability of nanoporous gold cantilevers with respect to focused ion beam
irradiation was studied and all samples bent toward the 30 kV Ga+ beam. Their
sensitivity to irradiation was observed to be considerably higher than that of solid
samples. A novel message of the model proposed here is that the bending of
nanoporous gold samples under ion irradiation can be explained by coarsening and
sintering of the ligament structure which results in decrease of porosity and
shrinkage of the sample subsurface layer.
An analogy with thermoelasticity was used to convert the inhomogeneous
volume change due to radiation-induced coarsening into the bending curvature of a
free-standing cantilever. The curvature radius depends essentially on the effective
Young's modulus of the damaged nanoporous structure. It turned out that in order
to explain experimental behavior of the cantilevers under irradiation we had to
assume non-monotonic dependence of effective Young's modulus on mean
ligament size: decreasing for small ligament sizes and increasing, of the Gibson-
Ashby type, for larger ligament sizes. The model calculations reproduce
experimental observations on ion-beam-induced bending of nanoporous Au
samples.
The study suggests that nanoporous materials with open-cell porosity are ill-
suited for nuclear applications as structural materials. If on the one hand the high
specific surface area of nanoporous gold facilitates the removal of radiation defects
from the bulk of ligaments; on the other hand it stimulates the radiation-induced
effects of coarsening and sintering, which result in negative swelling, i.e.
dimensional instability.
6.7 References
[1] L. Tsu, Tao Te Ching, Vintage Books, 1972.
[2] E. M. Bringa et al., “Are Nanoporous Materials Radiation Resistant?,” Nano Lett., vol.
12, no. 7, pp. 3351–3355, Jul. 2012.
[3] T. T. Järvi, J. A. Pakarinen, A. Kuronen, and K. Nordlund, “Enhanced sputtering from
nanoparticles and thin films: Size effects,” EPL (Europhysics Letters), vol. 82, no. 2, p.
26002, Apr. 2008.
Nanoporous specimens 143
[4] E. G. Fu et al., “Surface effects on the radiation response of nanoporous Au foams,”
Appl. Phys. Lett., vol. 101, no. 19, p. 191607, Nov. 2012.
[5] S. Punzhin, E. Detsi, A. Kuzmin, and J. T. M. D. Hosson, “Deformation of nanoporous
nanopillars by ion beam-induced bending,” J Mater Sci, vol. 49, no. 16, pp. 5598–5605,
May 2014.
[6] L. H. Qian and M. W. Chen, “Ultrafine nanoporous gold by low-temperature dealloying
and kinetics of nanopore formation,” Applied Physics Letters, vol. 91, no. 8, p. 083105,
Aug. 2007.
[7] J. Erlebacher, “An Atomistic Description of Dealloying Porosity Evolution, the Critical
Potential, and Rate-Limiting Behavior,” J. Electrochem. Soc., vol. 151, no. 10, pp. C614–
C626, Oct. 2004.
[8] O. Zinchenko, H. A. De Raedt, E. Detsi, P. R. Onck, and J. T. M. De Hosson,
“Nanoporous gold formation by dealloying: A Metropolis Monte Carlo study,”
Computer Physics Communications, vol. 184, no. 6, pp. 1562–1569, Jun. 2013.
[9] S. V. Petegem et al., “On the Microstructure of Nanoporous Gold: An X-ray Diffraction
Study,” Nano Lett., vol. 9, no. 3, pp. 1158–1163, Mar. 2009.
[10] K. Kolluri and M. J. Demkowicz, “Coarsening by network restructuring in model
nanoporous gold,” Acta Materialia, vol. 59, no. 20, pp. 7645–7653, Dec. 2011.
[11] J. Erlebacher, “Mechanism of Coarsening and Bubble Formation in High-Genus
Nanoporous Metals,” Phys. Rev. Lett., vol. 106, no. 22, 2011.
[12] H. Rösner, S. Parida, D. Kramer, C. A. Volkert, and J. Weissmüller, “Reconstructing a
Nanoporous Metal in Three Dimensions: An Electron Tomography Study of Dealloyed
Gold Leaf,” Adv. Eng. Mater., vol. 9, no. 7, pp. 535–541, Jul. 2007.
[13] R. N. Viswanath, V. A. Chirayath, R. Rajaraman, G. Amarendra, and C. S. Sundar,
“Ligament coarsening in nanoporous gold: Insights from positron annihilation study,”
Applied Physics Letters, vol. 102, no. 25, p. 253101, Jun. 2013.
[14] J. T. M. D. Hosson and A. V. Veen, “Nanoprecipitates and nanocavities in functional
materials,” in Encyclopedia of Nanoscience and Nanotechnology, vol. 7, H. S. Nalwa,
Ed. American Scientific Publishers, 2004, pp. 297–349.
[15] Y. Sun, S. A. Burger, and T. J. Balk, “Controlled ligament coarsening in nanoporous
gold by annealing in vacuum versus nitrogen,” Philosophical Magazine, vol. 94, no. 10,
pp. 1001–1011, Apr. 2014.
[16] S. Kuwano-Nakatani et al., “Environment-Sensitive Thermal Coarsening of Nanoporous
Gold,” Mater. Trans., vol. 56, no. 4, pp. 468–472, Apr. 2015.
[17] S. Baier et al., “Influence of gas atmospheres and ceria on the stability of nanoporous
gold studied by environmental electron microscopy and in situ ptychography,” RSC
Adv., vol. 6, no. 86, pp. 83031–83043, Aug. 2016.
144 Chapter 6
[18] D. R. Peale and B. H. Cooper, “Adsorbate-promoted mass flow on the gold (111) surface
observed by scanning tunneling microscopy,” Journal of Vacuum Science & Technology
A, vol. 10, no. 4, pp. 2210–2215, Jul. 1992.
[19] T. Fujita et al., “Atomic origins of the high catalytic activity of nanoporous gold,”
Nature Materials, vol. 11, no. 9, pp. 775–780, Sep. 2012.
[20] T. Fujita et al., “Atomic Observation of Catalysis-Induced Nanopore Coarsening of
Nanoporous Gold,” Nano Lett., vol. 14, no. 3, pp. 1172–1177, Mar. 2014.
[21] N. Badwe, X. Chen, and K. Sieradzki, “Mechanical properties of nanoporous gold in
tension,” Acta Materialia, vol. 129, pp. 251–258, May 2017.
[22] Y. Sun and T. J. Balk, “A multi-step dealloying method to produce nanoporous gold
with no volume change and minimal cracking,” Scripta Materialia, vol. 58, no. 9, pp.
727–730, May 2008.
[23] K. F. McCarty, J. A. Nobel, and N. C. Bartelt, “Vacancies in solids and the stability of
surface morphology,” Nature, vol. 412, no. 6847, pp. 622–625, Aug. 2001.
[24] E. Seker et al., “The effects of post-fabrication annealing on the mechanical properties
of freestanding nanoporous gold structures,” Acta Mater, vol. 55, pp. 4593–4602,
2007.
[25] Y. Sun, K. P. Kucera, S. A. Burger, and T. J. Balk, “Microstructure, stability and
thermomechanical behavior of crack-free thin films of nanoporous gold,” Scripta
Mater, vol. 58, pp. 1018–1021, 2008.
[26] M. H. and Mabuchi, “Microstructural evolution in nanoporous gold by thermal and acid
treatments,” Mater. Lett, vol. 62, pp. 483–486, 2008.
[27] M. Hakamada and M. Mabuchi, “Thermal coarsening of nanoporous gold: Melting or
recrystallization,” J. Mater. Res, vol. 24, pp. 301–304, 2009.
[28] Y. K. Chen-Wiegart et al., “Structural evolution of nanoporous gold during thermal
coarsening,” Acta Mater, vol. 60, pp. 4972–4981, 2012.
[29] J. Li et al., “In situ heavy ion irradiation studies of nanopore shrinkage and enhanced
radiation tolerance of nanoporous Au,” Scientific reports, vol. 7, p. 39484, 2017.
[30] J. Li, C. Fan, Q. Li, H. Wang, and X. Zhang, “In situ studies on irradiation resistance of
nanoporous Au through temperature-jump tests,” Acta Mater., vol. 143, pp. 30–42,
2018.
[31] Y. N. Osetsky, D. J. Bacon, A. Serra, B. N. Sing, and S. I. Golubov, “Stability and
mobility of defect clusters and dislocation loops in metals,” J. Nucl. Mater., vol. 276, pp.
65–77, 2000.
[32] Y. Matsukawa and S. J. Zinkle, “One-dimensional fast migration of vacancy clusters in
metals,” Science, vol. 318, pp. 959–962, 2007.
Nanoporous specimens 145
[33] K. Arakawa, K. Ono, M. Isshiki, K. Mimura, M. Uchikoshi, and H. Mori, “Observation of
the one-dimensional diffusion of nanometer-sized dislocation loops,” Science, vol. 318,
pp. 956–959, 2007.
[34] R. B. Hetnarski, Ed., Encyclopedia of Thermal Stresses. Springer Netherlands, 2014.
[35] L. J. Gibson and M. F. Ashby, “The mechanics of three-dimensional cellular materials,”
Proc. R. Soc. Lond. A, vol. 382, pp. 43–59, 1982.
[36] L. J. Gibson and M. F. Ashby, Cellular solids, 2nd ed. Oxford: Pergamon Press, 1999.
[37] L.-Z. Liu and H.-J. Jin, “Scaling equation for the elastic modulus of nanoporous gold
with ‘fixed’ network connectivity,” Applied Physics Letters, vol. 110, no. 21, p. 211902,
2017.
[38] L.-Z. Liu, X.-L. Ye, and H.-J. Jin, “Interpreting anomalous low-strength and low-
stiffness of nanoporous gold: Quantification of network connectivity,” Acta Mater., vol.
118, pp. 77–87, 2016.
[39] A. Mathur and J. Erlebacher, “Size dependence of effective Young’s modulus of
nanoporous gold,” Appl. Phys Lett., vol. 90, p. 061910, 2007.
[40] C. M. Stafford et al., “A buckling-based metrology for measuring the elastic moduli of
polymeric thin films,” Nature materials, vol. 3, no. 8, pp. 545–550, 2004.
[41] N. Mameka, K. Wang, J. Markmann, E. T. Lilleodden, and J. Weissmuller, “Nanoporous
gold-testing macro-scale samples to probe small-scale mechanical behavior,” Mater.
Res. Lett., vol. 4, p. 27, 2016.
[42] M. Ziehmer, K. Hu, K. Wang, and E. T. Lilleodden, “A principle curvatures analysis of
the isothermal evolution of nanoporous gold: Quantifying the characteristic length-
scales,” Acta Mater., vol. 120, p. 24, 2016.
[43] Y. K. Chen et al., “Morphological and topological analysis of coarsened nanoporous gold
by x-ray nanotomography,” Appl. Phys. LEtt., vol. 96, p. 043122, 2010.
[44] S. Karim et al., “Morphological evolution of Au nanowires controlled by Rayleigh
instability,” Nanotechnology, vol. 17, p. 5954, 2006.
[45] J. F. Ziegler, M. D. Ziegler, and J. P. Biersack, “SRIM – The stopping and range of ions
in matter (2010),” Nuclear Instruments and Methods in Physics Research Section B:
Beam Interactions with Materials and Atoms, vol. 268, no. 11, pp. 1818–1823, 2010.
[46] P. Jung, “Average atomic-displacement energies of cubic metals,” Phys. Rev. B, vol. 23,
pp. 664–670, 1981.
A compendium for bending of nanostructures: summary and outlook
7.1 Thesis summary
In addition to exploratory research in the electrochemical fabrication of
nanoporous materials (see Chapter 3), this Thesis has focused primarily on a novel
understanding of the physical mechanisms responsible for the ion-induced bending
phenomenon in metallic materials. Two distinct mechanisms were found to govern
bending in 1. solid bulk and 2. nanoporous thin cantilevers.
The dense solid bulk case, as explained in Chapters 4 and 5, was shown to
involve the formation and accumulation of a defect structure along the samples
thickness. Since the observed cantilever curvatures were too large to be attributed
to the volume change generated by isolated point defects, we postulated that
bending involves the formation of highly mobile interstitial clusters which diffuse
and combine to form sessile clusters in the regions beyond the ion implantation
zone. This hypothesis was supported by the agreement of a rate-equation model
with the experimental observations of two metals with distinct ion-induced
bending behaviors, namely gold and aluminum.
Gold cantilevers, as most of the materials reported in the literature, bended
towards the incident ion beam in all experiments. The unexpected response of
aluminum cantilevers to irradiation was to bend initially away from the ion beam
148 Chapter 7
and to reverse its direction with increasing fluence. This difference in behavior
stems from the distinct defect kinetics in each material. While in gold the sessile
clusters form a broad distribution over depth that extends all the way to the surface
opposite of the irradiated one; in aluminum this distribution over depth is more
narrow and ‘moves’ along the cantilever thickness as sputtering proceeds. In this
manner, the cantilever bends away from the ion beam when most of the interstitial
clusters distribution falls on the ‘top’ side of its neutral axis. The bending direction
is reversed as the cantilever thickness decreases by sputtering and the clusters
distribution falls mostly on the ‘bottom’ side. The phenomena depend critically on
the sputter rate versus the mobility of the clusters, both of which are material
specific.
The nanoporous case: a similar approach was attempted to explain ion-
induced bending in cantilevers made of nanoporous gold. This revealed a series of
particularities that led to the design of a model description taking into account the
volumetric compaction of the irradiated zone rather than the diffusion of defect
clusters.
To start with, the experimental procedures to fabricate the solid bulk bending
cantilevers was found not to be applicable to the nanoporous thin films for a
number of reasons. Special care had to be taken to avoid some effects that would
have affected the analysis. For one, the cantilevers had to be smaller than their bulk
counterparts to minimize the chance of having a grain boundary from the parent
alloy in the bent area. The ion irradiation produces cracks along the grain
boundaries in the dense solid case, which makes the cantilevers displacement non-
uniform and therefore not suitable for study. If the cantilevers were too small,
however, ion-induced densification in the porous systems occurred and made the
bending behavior deviate towards that of the dense solid case. Secondly, the cutting
steps in the cantilever fabrication procedure had to be adjusted carefully to avoid
excessive displacement of the cantilevers prior to the actual experiment, as the
sensitivity to fluence was much higher than in the dense solid case and the mere
scanning of the ion-beam along its cut edges could already cause considerable
bending. After reaching the optimal preparation method that allowed for
reproducible bending, the experimental results could be satisfactorily described
theoretically by assuming a non-monotonic dependence of the irradiated
structure’s effective Young modulus on mean ligament size.
Summary and outlook 149
7.2 Samenvatting
Dit proefschrift concentreert zich op het verkrijgen van nieuwe inzichten in
de fysische mechanismen van ionen-geïnduceerde buig fenomenen in metallische
materialen. Daarnaast is er ook verkennend onderzoek gedaan naar de
elektrochemische vervaardiging van nano-poreuze materialen (zie Hoofdstuk 3). Er
zijn twee verschillende mechanismen gevonden voor de buiging van: 1. massieve- ,
en 2. nano-poreuze dunne vrijdragende balken.
Zoals beschreven in de Hoofdstukken 4 en 5 blijkt in geval van een massieve
hefboom (een balkje dat aan één kant vastzit), die we cantilever noemen, sprake te
zijn van de vorming en accumulatie van defecten langs de doorsnede. De
waargenomen buiging van de cantilever is te groot om toegeschreven te worden aan
de volumeverandering ten gevolge van geïsoleerde puntdefecten, vacatures en
interstitiëlen. Daarom stellen we voor dat de werking van het buigeffect het gevolg
is van mobiele interstitiële clusters van puntfouten. Deze clusters diffunderen en
combineren om vervolgens stationaire clusters te vormen in de gebieden buiten de
ionen-implantatiezone. Deze hypothese wordt onderbouwd met een
productiesnelheidmodel van defecten en met experimentele waarnemingen aan het
door ionen-geïnduceerde buiggedrag van twee verschillende metalen, te weten
goud en aluminium.
De cantilevers van goud buigen in alle experimenten naar de bundel van
invallende ionen toe. Aluminium laat daarentegen een onverwacht verschijnsel
zien door eerst van de ionenbundel af te buigen, om vervolgens bij toenemende
bestraling om te keren. Dit verschil in gedrag wordt toegeschreven aan de kinetiek
van de defecten in de afzonderlijke materialen. In goud zijn de stationaire clusters
breed verspreid over de diepte, tot aan de zijde tegenovergesteld aan de bestraalde
oppervlakte toe. In aluminium is de distributie in de diepte van de clusters vel
smaller en ‘bewegen’ ze over de diepte van de cantilever als de bestraling langer
voortduurt. De balk buigt van de ionenbundel weg zolang de meeste interstitiële
clusters aan de ‘voorzijde’ van de neutrale as blijven. De buiging gaat vervolgens in
omgekeerde richting wanneer de cantilever door aanhoudend sputteren in dikte
afneemt, waardoor de clusters zich vooral aan de ‘achterzijde’ bevinden. Dit
150 Chapter 7
fenomeen is sterk afhankelijk van de sputtersnelheid en de mobiliteit van de
clusters, en is daardoor materiaalgebonden.
Voor de nano-poreuze objecten is voor een vergelijkbare aanpak gekozen, om
ion-geïnduceerde buiging in cantilevers van nano-poreus goud te verklaren. Dit
heeft geleid tot een reeks van nieuwe waarnemingen waaruit de invloeden van
volumetrische verdichting werden ontdekt in plaats van de diffusie van defecten-
clusters.
Zo bleek de experimentele procedure om massieve cantilevers te maken niet
toepasbaar voor de nano-poreuze dunne lagen om meerdere redenen. Extra
aandacht moest geschonken worden om bepaalde effecten te vermijden, die van
invloed kunnen zijn op de analyse. Het eerste effect is dat de cantilevers kleiner
moeten zijn dan de massieve variant om zo de kans op een korrelgrens van de
oorspronkelijke legering in de buigzone te vermijden. De samentrekking ten
gevolge van ionen-bestraling, zorgt voor scheuren langs de korrelgrenzen waardoor
de cantilevers niet gelijkmatig vervormen en daarom niet geschikt zijn voor
gedetailleerde bestudering. Wanneer de cantilevers echter te klein zijn, vindt een
ionen-geïnduceerde verdichting plaats waardoor het buiggedrag gaat afwijken en
meer op het gedrag van massieve cantilevers lijkt. Ten tweede moet het uitsnijden
van de cantilever aangepast worden om vervorming voorafgaand aan het
daadwerkelijke experiment te voorkomen. Deze dunne cantilevers zijn veel
gevoeliger voor bestraling dan de massieve varianten. Slechts een geringe
blootstelling aan de bewegende ionenbundel langs de snijranden kan al een
aanzienlijke buiging veroorzaken. Na optimalisatie van de preparatiemethode
konden reproduceerbare resultaten worden verkregen, welke op een passende
theoretische manier zijn beschreven. Deze beschrijving is gebaseerd op de aanname
dat er een niet-monotone afhankelijkheid van de effectieve stijfheid van de
bestraalde structuur is op de gemiddelde ligamentgrootte.
7.3 Outlook
A natural continuation of the studies on the ion-induced bending
phenomenon is to investigate the behavior of cantilevers made of more than a
single material. Below some preliminary results on the ion-induced bending
behavior of two layered systems are discussed, which have not been investigated
further but present potentially interesting starting points for future research. The
Summary and outlook 151
methods for fabrication of these systems were the same as described in Chapter 2
for the single component samples.
7.3.1 Layered systems
Au-Pt
The choice of Au and Pt is justified by their lack of oxidation issues (its
possible effects are discussed shortly ahead) and the absence of intermetallic
phases that could introduce complex effects to account for. Figure 7.1 shows the
response to irradiation of films with the same thickness (100 nm) made of pure
metals, i.e. Au and Pt; and of bilayered films composed of the two metals irradiated
from both sides.
The Pt film presented the same bending behavior as the Au film, i.e. toward
the ion beam but at a lower rate. MD studies of irradiation-induced stress in thin Pt
films suggested that the initial stress in the film could change the defect structure
development as much as to result in opposed - tensile or compressive - final stress
states[1]. In the present experiments, the evaporated thin films were not annealed
in order to minimize diffusion and ‘blurring’ of the metal-metal interface. This led
to a higher bending rate of the Au cantilever than observed in annealed samples of
similar thickness in Chapter 5.
152 Chapter 7
Figure 7.1 - Fluence dependence of the curvature 1/R of films made of pure Au
(black); pure Pt (green) and bilayered films with the same total thickness
irradiated on the Au side (red) and Pt side (blue).
The layered systems showed bending rates that fell between those of the pure
metal cases. According to SRIM[2] simulations for irradiation with 30 keV Ga+
ions, the implantation range does not cross the interface for any of the two
irradiated sides (Figure 7.2). The role of the interface in interacting with the
diffusing defects would be an interesting question to investigate. We showed earlier
that the bending response to irradiation is inversely proportional to the film
thickness. Assuming that the interface acts rather as a barrier than as a sink, the
irradiation of the bilayered systems can be seen as the irradiation of a single 50 nm
layer that is restrained by another undamaged layer. With this point of view, the
smaller difference between the bending responses of the Au and Pt layers in
comparison to the pure cases could be understood since the Young’s modulus of Pt
is roughly twice the modulus of Au (168 x 78 GPa [3]) making it a better ‘anchor’
for the bending of the Au layer than vice-versa.
Summary and outlook 153
Figure 7.2 - Ion implantation ranges for both sides of the Au-Pt bilayered
system as calculated by SRIM.
Al-Al2O3
In Chapter 5, the abnormal bending behavior of Al cantilevers under
irradiation was explained in terms of the position of the steady-state distribution of
interstitial clusters relative to the total film thickness. One experimental result
particularly in favor of that hypothesis was that the thinner irradiated samples (89
nm) presented downward bending in a much smaller degree than the thicker
samples before reversing direction and bending towards the beam. That
observation was understood to result from the fact that in the thinner samples the
thickness reduction due to sputtering caused the ‘crossing’ of the distribution to the
bottom side of the cantilever to occur at a smaller fluence than in the samples with
144 and 200 nm thickness.
Since Al is known to form a thin oxide layer on its surface when exposed to air,
the question arises of what would be the influence of that layer in the ion-induced
bending phenomenon. To shine light on that issue, a 10 nm layer of Al2O3 was
evaporated on one side of a 100 nm film of pure Al. The surprising bending
behavior observed after irradiation of those samples on both sides is shown in
Figure 7.3.
154 Chapter 7
Figure 7.3 – Effect of the addition of a 10 nm layer of Al2O3 on both sides of a
100 nm thick Al film on the fluence dependence of curvature 1/R.
While the pure Al cantilevers behaved similarly to the cantilevers with 89 nm
presented in Chapter 5 as expected, those with the added oxide layer showed a
much more pronounced bending in the direction opposed to the beam. More
remarkably, the effect occurred when the cantilever was irradiated on the oxide
side but also when irradiated on the opposite side.
An explanation for the downward bending of Al cantilevers in general and the
increase in magnitude when an extra layer is added on the irradiated (‘top’) surface
is as follows. The 30 keV Ga+ ions can pass through the 10 nm oxide layer and
cause displacements in the underlying metal bulk according to SRIM simulations
(Figure 7.4). While the displacements in the metal will give rise to the formation
and accumulation of clusters as discussed earlier, these are unable to diffuse and
annihilate at the top surface before the oxide layer is sputtered away. This way, the
cantilever bends downward due to the positive volume change induced by the
distribution of clusters being formed in the vicinity of the irradiated surface. When
the oxide layer is removed by the impinging ions, the irradiated surface becomes a
sink causing the defect distribution to move deeper into the cantilever’s thickness
Summary and outlook 155
and the bending direction is reversed. The fact that the bending behavior is similar
when the extra oxide layer is at the non-irradiated side of the cantilever is harder to
interpret. This could mean that annihilation at both surfaces plays a crucial role
from the earliest stages of irradiation – a puzzling question for future investigation.
Figure 7.4 – Trajectories of ~10k 30 keV Ga+ ions and distribution profile of
collision events on a bilayered cantilever of Al2O3 (10 nm) and Al (100 nm).
7.3.2 Size-dependent coarsening of ion-irradiated np-Au
In Chapter 6, the volume contraction of the irradiated surfaces of nanoporous
gold was attributed to ion-induced coarsening. This was hypothesized to result
from the direct fracture of smaller ligaments by the ion beam, followed by surface-
mediated rearranging of the structure. More recent measurements of the irradiated
structures were made using images with higher resolution and an automatic code
AQUAMI [4] which has the advantage of not requiring any subjective interventions
from the user. These revealed a size-dependent behavior where densification
followed from thickening of the ligament diameters only for the sample with finer
ligaments (t0~15 nm); the sample with thicker ligaments densifies rather by the
decrease in ligament length. The normalized probability density function
distributions of the nanoporous samples prior and after irradiation are shown in
Figure 7.5.
156 Chapter 7
Figure 7.5 – Comparison of normalized probability density functions of
nanostructure network parameters of samples with t0~15nm (a-c) and t0~35nm
(d-f) before and after irradiation. The ligament and pore diameters were fitted
to Gaussian distributions while the ligament lengths were fitted to log-normal
distributions.
It has to be emphasized that properties such as plastic yield and ductile
fracture are sensitive to the extremes in the distribution, not to the averaged values
[5]. As can be seen in Figure 7.6, a small volume of these samples can present
variations of ligament diameters in the range of roughly one order of magnitude.
The different responses to ion irradiation between samples is here suggested to be a
consequence of the smaller ligament diameters in the sample with t0~15nm (Figure
7.5a). Making a bridge to the case of thermal coarsening of nanoporous gold,
topology has been shown to be an important driving factor. In MD
simulations, a decrease in the topological genus caused by ligament ‘pinch-off’
(a)
(b)
(c)
(d)
(e)
(f)
Summary and outlook 157
events preceded the increase in ligament diameter [6]. Additionally, studies using a
conserved dynamics phase-field model showed that bicontinuous structures exhibit
an asymptotic behavior of the scaled genus upon coarsening, with samples of
different relative densities reaching the same value of scaled genus [7]. Direct
fracture of fine ligaments by the ion beam followed by surface-mediated
rearrangement provides the most probable explanation for the ion-induced
coarsening of the sample with finer ligaments.
Figure 7.6 – Bright field image of sample with L0~35nm. Red arrows highlight
the dispersion of ligament diameters.
So far ion-induced effects on nanoporous gold have been reported either as
secondary observations or with a focus on the effects on single ligaments or pores
[5], [8]–[11]. The evolution of the nanoporous network in samples with two distinct
initial ligament average sizes is shown here to differ under irradiation with Ga+
ions. Both samples showed an increase in the irradiated surface relative density
associated with an increase in the ligaments width-to-length (t/l) ratio.
Interestingly, the densification rate was higher for the sample with smaller
158 Chapter 7
characteristic sizes (t0~15nm), where the increase in t/l proceeded by simultaneous
increase in diameter and decrease in length. For the sample with to~35 nm, the
change in aspect ratio occurred only by decrease of ligament length with no
observed thickening. A more detailed understanding of the effects of ion irradiation
on nanoporous structures may provide a facile alternative to tune their parameters
over the ion-implantation range with a FIB.
Figure 7.7 – Relationship between relative density and ligament aspect ratio for
samples with t0=15 nm (blue) and t0=35 nm (red).
7.4 Opportunities of Ion Beam Induced Bending or (IB) ²
Throughout the present studies, irradiation was performed by scanning the
whole cantilevers area. This was done in order to obtain uniform curvatures along
the cantilever length for each fluence, allowing for the use of the thermoelasticity
analogy. This does not need to be the case, however, as sharp folding of thin films
achieved by irradiation of lines have been reported before [12], [13]. While this
Thesis focused only on a few materials in an attempt to elucidate the physical
Summary and outlook 159
processes, early reports of the ion-induced phenomenon reported bending of
several classes of materials such as metals, semiconductors and compounds [14]–
[16]. Figure 7.8 shows some examples of 3D structures fabricated by single- and
multi-folding achieved by sequential irradiation strategies. Those publications,
however, concentrated rather in demonstrating the feasibility of producing useful
devices providing rather a qualitative picture than a quantitative one. The ion-
implantation range as calculated with SRIM, for instance, is often taken to result in
a static stress distribution with no regards to defect diffusion and recombination.
Figure 7.8 – Examples of structures fabricated using Ion Beam Induced
Bending: (a) Pt mechanical switch for in-situ testing of nano-electronic circuits
[17]; (b) Ti/Al/Cr microcage for trapping of particles [15]; (c) Si thin film
transistor field emitter array [13]; (d) Al helical optical antennas [16].
Regarding the size limitations for structures fabricated by this method and
restricting the discussion to the bulk metal case, one must consider an interplay
between all parameters that define the depth of the ion-induced volume changes,
(a) (b)
(c) (d)
160 Chapter 7
i.e. film thickness, ion source and energy and defect kinectics. Controlled bending
in both directions (toward and opposed to the ion beam) has been demonstrated
elsewhere by varying beam energy and cantilever thickness and no influence was
observed of the cantilever’s length (in the range of 1 to 2 µm) or width (in the range
from 1 µm up to 20 mm)[14].
The cantilever dimensions need to be considered, however, when they are in a
similar size scale as the film’s grains. Our experiments in strongly textured AuAg
films (Figure 7.9) revealed that the effects of differential sputtering for particular
orientations can be large enough to produce significant variation in thickness in the
cantilevers upon irradiation. This uneven thickness results in non-uniform bending
at higher fluences, compromising the reproducibility of the process. The
orientation effect was not observed in the experiments described in Chapter 5 and
this is attributed to the small grain size of the films produced by evaporation, in the
same scale of the film’s thickness hence 1 to 2 orders of magnitude smaller than the
cantilevers dimensions.
Figure 7.9 – Left: superimposed inverse pole figure and image quality maps
showing typical structure of 1x5 µm² Au15Ag85 cantilevers, with lattice
orientation indicators. Right: SEM image of a set of cantilevers after low dose
irradiation, with visible ion channeling-induced contrast and differential
sputtering.
Yet another aspect that hasn’t yet been studied is the effect of irradiation angle
in bending. The depth of the volume changes that govern bending could be tuned
not only by adjusting the beam energy but also by changing the irradiation angle
relative to the surface. The angular dependence of both penetration range [18] and
Summary and outlook 161
sputtering yield [19] could be explored to e.g. minimize the amount of accumulated
damage in a given structure by decreasing the diffusion path for annihilation of
defects at the surface at the same time as the rate of removal of the top layers is
increased.
To conclude, Ion Beam Induced Bending is not only a promising tool in the
fabrication of micro- and nanosized devices or MEMS components but also a
method of characterization of the physical properties of extremely fine nano-
features that are beyond the capacity of typical physical contact-based techniques
to measure. The open questions exposed earlier in this chapter provide exciting
topics of investigation that will ultimately lead to a deepened understanding of the
details involved in FIB-processing of materials.
7.5 References
[1] W.-L. Chan, ‘Stress evolution in platinum thin films during low-energy ion irradiation’,
Phys. Rev. B, vol. 77, no. 20, 2008.
[2] J. F. Ziegler, M. D. Ziegler, and J. P. Biersack, ‘SRIM – The stopping and range of ions
in matter (2010)’, Nuclear Instruments and Methods in Physics Research Section B:
Beam Interactions with Materials and Atoms, vol. 268, no. 11–12, pp. 1818–1823, Jun.
2010.
[3] G. V. Samsonov, Handbook of the Physicochemical Properties of the Elements.
Springer US, 1968.
[4] J. Stuckner, K. Frei, I. McCue, M. J. Demkowicz, and M. Murayama, ‘AQUAMI: An
open source Python package and GUI for the automatic quantitative analysis of
morphologically complex multiphase materials’, Computational Materials Science, vol.
139, pp. 320–329, Nov. 2017.
[5] N. Badwe, X. Chen, and K. Sieradzki, ‘Mechanical properties of nanoporous gold in
tension’, Acta Materialia, vol. 129, pp. 251–258, May 2017.
[6] J. Erlebacher, ‘Mechanism of Coarsening and Bubble Formation in High-Genus
Nanoporous Metals’, Phys. Rev. Lett., vol. 106, no. 22, 2011.
[7] Y. Kwon, K. Thornton, and P. W. Voorhees, ‘The topology and morphology of
bicontinuous interfaces during coarsening’, EPL (Europhysics Letters), vol. 86, no. 4, p.
46005, May 2009.
[8] Y. Sun and T. J. Balk, ‘A multi-step dealloying method to produce nanoporous gold with
no volume change and minimal cracking’, Scripta Materialia, vol. 58, no. 9, pp. 727–
730, May 2008.
162 Chapter 7
[9] S. V. Petegem et al., ‘On the Microstructure of Nanoporous Gold: An X-ray Diffraction
Study’, Nano Lett., vol. 9, no. 3, pp. 1158–1163, Mar. 2009.
[10] M. Caro et al., ‘Radiation induced effects on mechanical properties of nanoporous gold
foams’, Appl. Phys. Lett., vol. 104, no. 23, p. 233109, 2014.
[11] E. G. Fu et al., ‘Surface effects on the radiation response of nanoporous Au foams’, Appl.
Phys. Lett., vol. 101, no. 19, p. 191607, Nov. 2012.
[12] Z. Liu et al., ‘Spatially oriented plasmonic “nanograter” structures’, Scientific Reports,
vol. 6, p. 28764, 2016.
[13] T. Yoshida, T. Nishi, M. Nagao, T. Shimizu, and S. Kanemaru, ‘Integration of thin film
transistors and vertical thin film field emitter arrays using ion-induced bending’,
Journal of Vacuum Science & Technology B, vol. 29, no. 3, p. 032205, 2011.
[14] T. Yoshida, M. Nagao, and S. Kanemaru, ‘Characteristics of Ion-Induced Bending
Phenomenon’, Japanese Journal of Applied Physics, vol. 49, no. 5, p. 056501, May
2010.
[15] K. Chalapat, N. Chekurov, H. Jiang, J. Li, B. Parviz, and G. S. Paraoanu, ‘Self-Organized
Origami Structures via Ion-Induced Plastic Strain’, Advanced Materials, vol. 25, no. 1,
pp. 91–95, 2013.
[16] Y. Mao et al., ‘Programmable Bidirectional Folding of Metallic Thin Films for 3D Chiral
Optical Antennas’, Advanced Materials, vol. 29, no. 19, p. 1606482, 2017.
[17] S. K. Tripathi, N. Shukla, N. S. Rajput, S. Dhamodaran, and V. N. Kulkarni, ‘Fabrication
of nano-mechanical switch using focused ion beam for complex nano-electronic
circuits’, Micro & Nano Letters, vol. 5, no. 2, pp. 125–130, 2010.
[18] H. Gnaser, Low-Energy Ion Irradiation of Solid Surfaces. Berlin Heidelberg: Springer-
Verlag, 1999.
[19] Q. Wei, K.-D. Li, J. Lian, and L. Wang, ‘Angular dependence of sputtering yield of
amorphous and polycrystalline materials’, J. Phys. D: Appl. Phys., vol. 41, no. 17, p.
172002, 2008.
List of publications (thesis subject
and related)
Journal papers
1. L.T.H. de Jeer, D.R. Gomes, J.E. Nijholt, R. van Bremen, V. Ocelík, J.T.M. De
Hosson, Formation of Nanoporous Gold Studied by Transmission Electron
Backscatter Diffraction, Microscopy and Microanalysis 21 (2015) 1387–1397.
doi:10.1017/S1431927615015329.
2. R. van Bremen, D.R. Gomes, L.T.H. de Jeer, V. Ocelík, J.T.M. De Hosson, On
the optimum resolution of transmission-electron backscattered diffraction (t-
EBSD), Ultramicroscopy 160 (2016) 256–264.
doi:10.1016/j.ultramic.2015.10.025.
3. D.R. Gomes et al., On the mechanism of ion-induced bending of
nanostructures, Applied Surface Science 446 (2018) 151–159.
doi:10.1016/j.apsusc.2018.02.015.
4. D.R. Gomes et al., On the fabrication of micro- and nano-sized objects: the
role of interstitial clusters, Journal of Material Science 53 (2018) 7822–7833.
doi:10.1007/s10853-018-2067-0.
5. D.R. Gomes et al., Ion-induced bending of nanoporous gold thin films: a
modelling approach (2018). Manuscript submitted for publication.
6. D.R. Gomes et al., Size-dependent ion-induced coarsening of nanoporous gold
(2018). Manuscript submitted for publication.
164 List of publications
Conference proceedings
1. D.R. Gomes et al., On the fabrication of functional nanostructures with ions,
7th Latin American Conference on Metastable and Nanostructured Materials
(NANOMAT 2017), Brotas, Brazil.
2. D.R. Gomes et al., The Role of Interstitial Clusters in Ion Beam Induced
Bending, 12th International Conference on Surfaces, Coatings and
Nanostructured Materials (NANOSMAT 2017), Paris, France.
Acknowledgements
First of all I wish to express my gratitude for my promotor Jeff De Hosson for
allowing me to pursue my PhD degree under his aegis, guiding and supporting me
throughout. I will always gratefully remember our weekly meetings which Jeff
coined Tempora Diegonis (whose day and time coincided, by a caprice of fate, with
the official PhD graduation ceremony!) and how delightful it was to discuss the
courses of research, the unexpected results, the ever more frivolous bureaucracy,
the anecdotes involving big names of science and whatnot. It is hard to express the
huge privilege it has been to enjoy your enlightening and stimulating influence!
I am deeply obliged to David Vainchtein, my co-promotor, for the help I
received not only on research-related topics but also on broader life and carreer
matters. More than once during casual conversations your questions and advices
led me to open my eyes about important issues I’ve been neglecting – I value those
guidances dearly!
I am grateful to Anatoliy Turkin, whose knowledge (and patience to discuss
and explain out so many misconceptions I brought to him so confidently!)
definitely brought my understanding of radiation damage in materials to a
heigthened level.
I am also grateful for all the people who were somehow involved in the
achievement of this Thesis. I’m aware that the inelegance of forgetting one name is
hard to avoid. In any case, my special words of appreciation to Paul Bronsveld for
his enthusiasm in teaching the operation of the arc-furnace during my first year
and his continual interest on my research; to Vašek Ocelík for all the trainings,
frequent assistance and fruitful collaboration on the optimum resolution of
transmission-electron backscattered diffraction (t-EBSD) leading to one of my first
co-authored publications; and to Yutao Pei for early discussions on template
electrodeposition. And to all the people who passed by the MK group during these
years as students or visitors – it was great sharing laboratory and lunch-table with
you!
166 Acknowledgements
A special mention goes to the students I had the chance to supervise: Jorrit
Nijholt, Marten Koopmans, Bauke Steensma, Lars Gaastra, Sameer Rodrigues and
Konrad Migacz. My gratitude for your trust and the enjoyable cooperations and my
wishes of success in your lifes!
Last but not least, I am grateful for those who were not directly related to the
work that culminated in this Thesis but whose absence would have made it much
more difficult. My friends Roger de Bortoli with his inspiring easy-going and
sincere attitude and Diego Blaese for exchanging weekend visits and the most
fruitful, insightful ontological discussions. My girlfriend Henrika Lüders for being
there and understanding me so well. Meus amigos Rodrigo de Barros, Daniel
Krzsynski e Frederico Mussi que apesar de estarem separados cada um por umas
tantas centenas de quilômetros, estão sempre perto graças à tecnologia. O humor
doentio de vocês me faz rir alto até nos momentos mais difíceis. Obrigado por
estarem aí, piazada do diabo.
Thanks for all your encouragement!
Curriculum Vitae
Diego Ribas Gomes
28 August 1985 Born in Ponta Grossa, Brazil
Education and Activities
8/2014 – 8/2018 PhD in Applied Physics, Materials Science research group
Zernike Institute for Advanced Materials
University of Groningen
Thesis title: Fabrication of metallic nanostructure with
ions: Theoretical concepts and applications
Supervisor: prof. Jeff Th.M. de Hosson
3/2012 – 3/2014 Master degree in Materials Science and Engineering
Federal University of Santa Catarina
Thesis title: Nanosecond ablation of alumina with an
ytterbium fibre-laser: experimental study, topography
and damage evaluation
Supervisor: Prof. Márcio Celso Fredel
9/2012 – 9/2013 Joint period in the context of the Brazilian-German
Collaborative Research Initiative on Manufacturing
Technology
Institute of Advanced Ceramics/Institute of Laser and
System Technologies
Hamburg University of Technology
Supervisor: dr. Rolf Janssen
2/2003 – 12/2008 Bachelor degree in Materials Engineering
State University of Ponta Grossa
top related