MATERIAL CHARACTERIZATION BY ENERGY FILTERED …This thesis has also not been submitted for any...
Transcript of MATERIAL CHARACTERIZATION BY ENERGY FILTERED …This thesis has also not been submitted for any...
MATERIAL CHARACTERIZATION BY ENERGY
FILTERED SECONDARY ELECTRON SIGNALS
INSIDE THE SCANNING ELECTRON MICROSCOPE
Avinash Srinivasan
(B.Tech, VIT University, India)
A THESIS SUBMITTED
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
DEPARTMENT OF ELECTRICAL AND COMPUTER ENGINEERING
NATIONAL UNIVERSITY OF SINGAPORE
2014
DECLARATION
I hereby declare that this thesis is my original work and it has been
written by me in its entirety. I have duly acknowledged all the sources
of information which have been used in the thesis.
This thesis has also not been submitted for any degree in any
university previously.
_________________________
Avinash Srinivasan
20 August 2014
i
Acknowledgements
Upon completion of this thesis, I would like to express my deepest gratitude to
those people who have helped make this work possible.
First and foremost I would like to thank my PhD Supervisor, Associate Professor
Anjam Khursheed who has been a constant source of guidance and support. A
world expert in his field, Prof. Khursheed is a fine example of how great knowledge
leads to great humility, listening to ideas of his students and helping them to
develop those ideas with his deep understanding of the subject. I also thank him for
reading through this thesis carefully and providing his valuable comments.
Next I would like to thank the office staff at CICFAR lab, Mrs Ho Chiow Mooi,
Mr Koo Chee Keong and Ms Linn Linn, who are the backbone of the lab’s
operation, for all their administrative help and infrastructure support.
I would also like to mention Dr Hung Quang Hoang and Mr Nelliyan Karuppiah
for their help in the initial stages of my work. I also thank Mr Han Weiding for his
useful assistance in some of the experiments and Mr Suvra Sarkar for useful
discussions and inputs.
On the personal front, this acknowledgement would be incomplete without
mentioning the unwavering support and constant encouragement of my parents,
Mrs Uma Srinivasan and Mr R Srinivasan, who have backed me in all stages of my
life. I also thank my wife Indrani for all her encouragement and understanding
during my work.
ii
Table of contents
Acknowledgements .................................................................................................. i
Summary ................................................................................................................. v
List of Tables ......................................................................................................... vi
List of Figures ....................................................................................................... vii
Chapter 1 – Introduction to the thesis ..................................................................... 1
References ....................................................................................................... 7
Chapter 2 – Introduction to the SEM .................................................................... 11
2.1 Output signals inside the SEM ........................................................... 15
2.2 Objective lens improvements ............................................................. 21
2.3 Secondary electron energy analyzers SEM attachments .................... 24
2.4 Retarding Field Analyzers .................................................................. 25
2.5 Signal–to–Noise considerations ......................................................... 27
2.6 Deflection/multi–channel analyzers ................................................... 33
2.7 Full range deflection/multi–channel analyzer designs ....................... 35
2.8 Objectives of the thesis ....................................................................... 39
References ..................................................................................................... 41
Chapter 3 – Voltage and dopant concentration measurements of semiconductors
using a band–pass toroidal energy analyzer inside a SEM ................................... 47
3.1 Introduction ........................................................................................ 47
3.2 The problem of specimen fringe fields and local surface microfields 51
3.3 Experimental Results .......................................................................... 55
3.3.1 The experimental setup ............................................................... 55
iii
3.3.2 Experimental analyzer SE signals on a doped silicon specimen in
presence of specimen fringe fields ............................................................ 57
3.3.3 Experimental analyzer SE signals in presence of surface
microfields ................................................................................................ 61
3.3.4 Experimental analyzer SE signals along a semiconductor sample
with a potential gradient............................................................................ 64
3.3.5 Experimental SE analyzer dopant contrast signals from abrupt
semiconductor heterojunctions ................................................................. 67
3.4 Conclusions ........................................................................................ 72
References ..................................................................................................... 74
Chapter 4 – New contrast mechanisms and material characterization by energy
filtered secondary electron signals inside the SEM .............................................. 77
4.1 Introduction ........................................................................................ 77
4.2 Probing and analyzing buried interfaces of multifunctional oxides using
a secondary electron energy analyzer. ........................................................... 78
4.2.1 Introduction ................................................................................. 78
4.2.2 Materials and methods ................................................................ 79
4.2.3 Results and Discussion ................................................................ 82
4.3 SE signal contrast in presence of magnetic fields above the specimen
90
4.4 SE analyzer signal contrast due to surface oxidation of a thin film metal
layer specimen ............................................................................................... 94
4.5 Conclusions ........................................................................................ 98
iv
References ..................................................................................................... 99
Chapter 5 – New secondary electron energy analyzer designs for the SEM ...... 102
5.1 Introduction ...................................................................................... 102
5.2 A wide–range parallel energy analyzer design ................................. 104
5.2.1 Need for a high transmittance wide–range parallel energy analyzer
104
5.2.2 A first–order focusing wide–range PRMA design .................... 109
5.3 Experimental prototype of a RMA attachment inside a SEM specimen
chamber ....................................................................................................... 117
5.4 Conclusion ........................................................................................ 124
References ................................................................................................... 125
Chapter 6 – Conclusions and Suggestions for future work................................. 127
6.1 Conclusions ...................................................................................... 127
6.2 Suggestions for future work ............................................................. 128
Appendix A: Publications resulting from this project ........................................ 131
v
Summary
The scanning electron microscope (SEM) today is capable of providing high
resolution nanometer-level topographical images of a specimen. However, there is
potential to transform it into a nano-scale material science analytical tool. The aim
of this dissertation is to devise and develop methods to improve the analytical
capabilities of the SEM by the use of secondary electron (SE) energy analyzer
attachments. This work presents high signal to noise voltage measurements in the
presence of surface fields and dopant concentration measurements on
semiconductors using the second–order focusing toroidal analyzer. An analytical
applications such as a new application of detecting trapped charges at buried
interfaces of multifunctional oxides is demonstrated, and the results point towards
the development a new SEM analytical technique. Also a prototype of the Radial
Mirror Analyzer is developed and tested inside the SEM. A new design of a multi-
channel SE analyzer, one that can capture the SE energy spectrum, in parallel is
also presented.
vi
List of Tables
Table 3.1 – Dopant concentration measurement results for the Si/ZnO p–n
heterojunction for different p–doped samples. ..................................................... 71
Table 5.1 – Simulated energy resolution for first–order focusing PRMA designs.
............................................................................................................................. 115
vii
List of Figures
Fig. 2.1 – Schematic drawing of a conventional SEM ......................................... 12
Fig. 2.2 – Beam/Specimen interaction: (a) Interaction volume and emitted signals
(b) Energy spectrum of electrons that leave the specimen. .................................. 14
Fig. 2.3 – Monte Carlo simulation of primary electrons striking a silicon sample at
energies 1, 5, 10 and 15 keV. ................................................................................ 15
Fig. 2.4 – Conventional detector layout inside the SEM. ..................................... 16
Fig. 2.5 – Secondary electron signal generation. .................................................. 17
Fig. 2.6 – The SE Chung–Everhart energy distribution........................................ 18
Fig. 2.7 – SE energy distribution with work function variations. ......................... 19
Fig. 2.8 – Variation of SE energy distribution with changes in (a) Specimen
potential and (b) Dopant type and concentration across a p–n junction. .............. 20
Fig. 2.9 – Different types of SEM objective lenses: (a) Conventional lens (b)
Magnetic In–lens (c) Single pole lens below the specimen (d) Single pole lens
above the specimen (e) Retarding field lens and (f) Mixed–field immersion lens
[2.10, 2.11]. ........................................................................................................... 22
Fig. 2.10 – Arrangement of an energy spectrometer for conventional objective lens
type SEMs. ............................................................................................................ 23
Fig. 2.11 – General layout of a closed loop retarding field spectrometer with
hemispherical grid. ................................................................................................ 26
Fig. 2.12 – Output S–curve signals of the retarding field analyzer. ..................... 26
Fig. 2.13 – Comparison of the signal–to–noise characteristics of retarding field
analyzers with multi–channel energy analyzers as a function of cut–off energy in
the SE Chung–Everhart spectrum [2.27]. ............................................................. 31
Fig. 2.14 – Schematic of magnetic immersion lens SE analyzer layout of Kazemian
et al. [2.28, 2.29] used for quantitative dopant mapping [2.1]. ........................... 32
Fig. 2.15 – The 63° CDA Hannah voltage contrast spectrometer [2.8] (a)
Spectrometer layout (b) Experimentally acquired SE spectra for different specimen
voltages. ................................................................................................................ 34
Fig. 2.16 – Schematic layout of the electrostatic toroidal deflection analyzer
reported by Rau and Robinson [2.35]. .................................................................. 36
viii
Fig. 2.17 – Layout of a second–order focusing toroidal analyzer prototype
attachment [2.11]. ................................................................................................. 38
Fig. 3.1 – Original layout of the second–order focusing toroidal analyzer prototype
attachment: (a) Experimental layout (b) Original specimen holder layout [3.3]. . 47
Fig. 3.2 – Experimental SE spectrum reported by Hung et al. [3.3] (a) full range
and (b) selected range in which curve 2 (dotted line) is obtained by shifting curve
1 by 12 mV in order to demonstrate the noise limit. ............................................ 48
Fig. 3.3 – Experimental secondary electron output signals at different specimen
biasing voltages reported by Hoang et al. [3.3]. ................................................... 49
Fig. 3.4 – Experimental secondary electron signals showing improved signal–to–
noise when specimen/inner cap is biased at −10 and −10.1 V shown around the
peak value: (a) Deflection voltage range from 7 to 8 V and (b) deflection voltage
range from 7.1 to 7.16 V [3.3]. ............................................................................. 50
Fig. 3.5 – Direct ray tracing of a 0.5 eV electron (polar launch angle 45 degrees)
with fringe fields above the specimen; the specimen is biased more negative with
reference to the inner cap. The dotted path shows the electron trajectory without
fringe fields (Specimen = – 10 V). ....................................................................... 52
Fig. 3.6 – Simulated potential distribution and electron trajectories of 0.5 eV and
0.4 eV electrons (polar launch angle 45) in the presence of Type I surface
microfields. The dotted line shows the simulated trajectory of the electron without
surface fields. ........................................................................................................ 54
Fig. 3.7 – Simulated potential distribution and electron trajectory of a 0.5 eV
electron (polar launch angle 45) in the presence of Type II surface microfields.
The dotted line shows the simulated trajectory of the electron without surface fields.
............................................................................................................................... 54
Fig. 3.8 – The second–order toroidal energy analyzer SEM attachment: (a)
Experimental layout in the SEM chamber (b) Schematic of the modified specimen
holder indicating the bias voltages applied to the various components of the holder.
............................................................................................................................... 56
Fig. 3.9 – Experimental SE signals obtained from an n–type semiconductor sample:
(a) Specimen biasing from – 10 to –13 volts with VC1 = –10 V, VC2 = 0 V. Inset
shows the biasing condition of the sample holder (b) Experimental SE signals at
different specimen biasing voltages where VS = VC1 (c) A plot of PMT signal
expectation value () for specimen potential change (ΔVs) from 0 to 3 V in presence
of specimen fringe fields. ...................................................................................... 59
Fig. 3.10 – Experimental SE signals obtained from a copper wire in presence of
surface fields: (a) Specimen arrangement to generate microfields above the point
of probing (b) SE signals obtained from the specimen for different biasing of the
ix
copper wire (c) A plot of PMT signal expectation value () for copper wire potential
change (ΔVs) from 0 to 4 V in presence of surface fields above the point of probing.
............................................................................................................................... 63
Fig. 3.11 – Experimental SE signals obtained by setting up a potential gradient
along a semiconductor sample: (a) Specimen arrangement using a button cell (b)
SE signals obtained from the specimen along the x direction (c) A plot of
expectation value µ of the SE signal against the change in potential ΔVS along
distance x. .............................................................................................................. 66
Fig. 3.12 – SE analyzer signal contrast from a n–ZnO / p–Si heterojunction: (a)
Schematic representation of the fabricated thin–film ZnO on Silicon substrate (b)
Experimental SE signals obtained from the p–side and the n–side of each
heterojunction sample for samples A, B, C and D (c) Plot of difference in
expectation value Δ of the SE analyzer signal obtained from the p and n sides of
the Si/ZnO heterojunction against the log of doping concentration of the p–type
silicon substrate. The solid bars at each point on the graph represent standard
deviation of Δ taken over 10 SE signals. ............................................................ 70
Fig. 4.1 – Layout of the specimen holder arrangement of the second–order focusing
toroidal energy analyzer showing the mounting of the LAO/STO specimen. ...... 79
Fig. 4.2 – Schematic representation of the specimen: (a) Representation of the
2DEG formed at the interface of crystalline LAO and crystalline STO substrate (b)
Conducting and insulating interface regions side by side on the same sample (the
interface was made insulating by proton irradiation). .......................................... 81
Fig. 4.3 – Experimental SE analyzer signals obtained from an uncoated STO
substrate (shown in dotted line) and from the LAO/STO heterointerface with
conducting interface (shown in solid line). A primary beam acceleration voltage of
3 kV was used. ...................................................................................................... 83
Fig. 4.4 – Experimental SE analyzer signals obtained from the LAO/STO
heterointerface with conducting interface (shown in solid line) and the insulating
LAO/STO heterointerface (shown in dotted line). A primary beam acceleration
voltage of 3 kV was used. ..................................................................................... 84
Fig. 4.5 – Experimental secondary electron signals obtained from LAO/STO
hetero–interface at various primary beam energies. The signals are obtained at
primary beam electron energies of 2 keV, 3 keV, 4 keV and 5 keV (shown in green,
brown, pink and blue respectively). ...................................................................... 86
Fig. 4.6 – Monte Carlo simulation of the electron trajectories: (a) Primary
beam/specimen interaction indicating the interaction volume of the electrons (b)
Energy contour of the percentage energy loss of primary beam electrons along the
depth of the specimen (c) A graphical plot of percentage energy loss of primary
x
beam electrons against the depth from the surface of the specimen. The red dotted
line indicates the LAO/STO interface at a depth of 8nm from the surface. ......... 89
Fig. 4.7 – Schematic representation of the specimen holder with a current carrying
solenoid placed under the specimen to produce magnetic field. (Cross–section view
of the specimen holder is shown here, while the solenoid is shown completely). 91
Fig. 4.8 – Experimental SE analyzer signals obtained from a metal specimen in
presence of magnetic field (B) created by current carrying solenoid under the
specimen: (a) SE analyzer signals obtained with B field along positive z direction
(b) SE analyzer signals obtained with B field along negative z direction (c) A plot
of SE analyzer signal expectation value (µ) against current flowing in the solenoid
creating the magnetic field (negative value of current indicates a current giving rise
to a B field along negative z–axis) ........................................................................ 93
Fig. 4.9 – Mounting of the Al coated silicon specimen inside the specimen holder
of the analyzer. ...................................................................................................... 95
Fig. 4.10 – (a) Experimental SE analyzer signals obtained from a thin film Al layer
under varying time of exposure to air (b) A plot of the SE analyzer signal
expectation value against the exposure time in air of the Al thin film. ................ 97
Fig. 5.1 – Simulated trajectory paths through the Radial Mirror Analyzer (RMA)
design by Hoang et al. [5.1], 13 rays are plot over a polar angular spread () of
6 in uniform angular steps, shown here from the specimen to the detector plane at
the central energy Ep. .......................................................................................... 102
Fig. 5.2 – Simulated trajectory paths through a planar ideal hyperbolic field
analyzer design at the energies 100, 200, 500, 1000, 3000 and 5000 eV. For each
energy, eleven trajectories are plot evenly between −1.1° and 1.1° around a 25°
polar entrance angle. ........................................................................................... 110
Fig. 5.3 – Simulated ray paths for energies 250, 750, 1250, 1750 and 2500 eV in
the ideal HFA electrode layout (R0 = 2 cm and V0 = –1500 volts) transformed into
axi–symmetric cylindrical coordinates. Nine trajectories plot with an angular
spread over 1.1. .............................................................................................. 112
Fig. 5.4 – Simulated trajectory paths through a first–order focusing PRMA design.
Electrons are plot for energies 50, 100, 200, 500, 1000, 1500, 2000 and 2500 eV
with 9 electron trajectories over a polar angular spread of ± 1.1˚ around a central
angle of 24.8˚. V1 = – 45V; V2 = – 120 V; V3 = – 285 V; V4 = –775V ; V5 = –
1150V; V6 = – 1675V; V7 = – 2020V. ................................................................. 113
Fig. 5.5 – Simulated spot size charcteristics of the first–order focsuing PRMA
design: (a) Direct ray paths at the focal plane for 2% energies around the central
energies of 500 eV and 1500 eV (b) Trace–width as a function of input polar
angular spread ranging from –20 mrad (–1.145) to +20 mrad (+1.145). ........ 116
xi
Fig. 5.6 – Schematic diagram showing the integration of the RMA attachment
inside the SEM chamber with other components. Such a mounting of the analyzer
facilitates operation of the SEM in the normal imaging mode. .......................... 119
Fig. 5.7 – Experimental layout of the RMA inside the SEM. ............................. 120
Fig. 5.8 – A prototype of the RMA attachment: (a) Cross–section 3D CAD model
(b) A photo of the attachment integrated inside the SEM.The azimuthal deflection
angle is 100. ....................................................................................................... 122
Fig. 5.9 – Experimental SE analyzer signals obtained using the first experimental
prototype of the RMA. ........................................................................................ 123
Fig. 6.1 – Proposed modification of the second order focusing toroidal analyzer
where the specimen is independent of the main analyzer body, allowing free
movement of the specimen. ................................................................................ 128
Fig. 6.2 – Proposed post analyzer deflector arrangement for the RMA attachment;
7 rays are plot over a polar angular spread () of 6 in uniform angular steps,
shown here from the specimen to the scintillator of the PMT through the post
analyzer deflector, at the central energy Ep. The magnitude of VPD was
experimentally calculated to be 0.436EP. ........................................................... 130
xii
List of Symbols
SYMBOL
DESCRIPTION
χ Analyzer voltage Resolution Constant
ρ Mass density
μ Signal expectation value / mean
Δμ Change in signal mean
Δθ Polar angular spread
ΔVS Change in specimen voltage,
ΔVR Change in Retarding grid voltage
Z Atomic number
W Working distance
VSC Scintillator Voltage
VS Specimen voltage
VR Retarding grid voltage
VPD Post analyzer deflector plate voltage
(magnitude)
VDEF Analyzer deflection electrode
VC2 Outer conical cap voltage
VC1 Inner conical cap voltage
V1, V2, V3, VD RMA electrode potentials
V1, V2, V3, V4, V5, V6, V7 First–Order PRMA electrode potentials
V0 Potential on a curved hyperbolic shaped
electrode
V(x,y) Potential field distribution
V Primary beam voltage
SEM Scanning Electron Microscope
SE1, SE2, SE3 Types of Secondary Electrons
SE Secondary electron
R Interaction volume
PO2 Partial pressure of Oxygen
PMT Photo multiplier tube
xiii
NS0 Total number of emitted secondaries
ND n-type doping level
I Current
EP Pass energy of an analyzer
EDS Energy Dispersive X-ray Spectroscopy
BSE Backscattered electrons
B Magnetic field
AE Auger electron
A Atomic weight
Chapter 1
1
Chapter 1 – Introduction to the thesis
The aim of this thesis is to devise and develop methods to improve the analytical
capabilities of the scanning electron microscope (SEM) by the use of electron
energy analyzer attachments. The SEM today is capable of providing high
resolution nanometer–level topographical images of a specimen. However, there is
potential to transform it into a nano-scale material science analytical tool.
Currently, the main analytical tool used inside the SEM is the well-known Energy
Dispersive X-ray Spectroscopy method (EDS or EDX). But EDS is limited by its
spatial resolution (typically about 1 μm) and also not suitable for low energy
primary beam applications [1.1].
There is a need for an energy spectral analyzer inside the specimen chambers of
SEMs, because generally, the detector system of conventional SEMs cannot
differentiate between scattered electrons of different energies that leave the
specimen when it is irradiated by the SEM primary beam. The type of information
that can be obtained by analyzing the energy spectrum of the scattered electrons
will be illustrated later in this chapter. The additional information obtained by the
use of electron energy analyzers in the SEM, designed to fit as add-on attachments,
can be collected concurrently with the normal topographical signal, and therefore
be mapped on to high resolution images of the specimen. Historically, SEM
electron energy spectrometers were first developed for the purpose of quantifying
specimen surface voltage changes. Electron beam testers were developed to make
contactless quantitative voltage measurements on integrated circuits (IC).
However, with the evolution of IC technology, multiple upper layers of metal lines
Chapter 1
2
were added to the IC, like power busses, high density routing signals, ground plane
and bond pads, preventing the probing of the circuit’s active regions by focused
electron beam testing methods.
At the turn of the 21st century, there was renewed interest in exploring possibilities
of using electron energy analyzers for applications other than voltage contrast,
mainly for material characterization purposes. The higher energy scattered
electrons, known as backscattered electrons (BSE), are commonly used for
qualitative material contrast mapping inside the SEM. However, by capturing their
energy spectrum, some degree of quantitative material characterization can be
performed since the shape of the BSE spectrum is dependent on the atomic number
of the specimen. This possibility was reported by Luo and Khursheed [1.2], who
correlated experimental BSE spectra with Monte Carlo simulations and applied the
technique for single elemental material analysis. Other researchers have reported
the possibilities of using the BSE energy spectrum for microtomography of layered
microstructures [1.3], measurement of surface potential and charge build-up on
insulator surfaces [1.4], and thickness measurements of ultrathin films on bulk
substrates [1.5].
It is in principle also possible to obtain signature Auger electron (AE) peaks in the
SEM scattered electron spectrum for elemental or compositional analysis.
Normally, Auger Electron Spectroscopy (AES) can only be performed under Ultra-
High Vacuum (UHV) conditions, however, Cubric [1.6] and El-Gomati [1.7]
demonstrated that it is possible to perform AES in vacuum High-Vacuum (HV)
level conditions, comparable to the HV inside SEM specimen chambers. In order
Chapter 1
3
to achieve this, a low voltage ion flood gun for cleaning of the specimen surface
must first be used, followed by relatively fast acquisition of the energy spectrum by
an energy analyzer (tens of milliseconds), before significant buildup of hydro-
carbon layers on the specimen surface.
Lower energy scattered electrons, known as secondary electron (SE) signals,
normally used to obtain topographical image of the specimen inside a conventional
SEM, also carry useful analytical information about the specimen. These can be
used to measure variations in parameters such as specimen surface potential,
semiconductor doping concentration and work function. Quantitative mapping of
doping concentration in semiconductors inside the SEM has received particular
attention recently. Recent studies have demonstrated that by monitoring changes in
the SE signal, it is possible to obtain quantitative dopant mapping in
semiconductors [1.8-1.17]. The most common method for doing this is to extract
contrast directly from an SE image [1.8-1.11], but this is not an accurate way to do
it, since the conventional SE detector captures a bulk signal formed from electrons
that leave the specimen over a wide range of different energies and angles, making
the output signal dependent on a number of other dynamic factors besides dopant
concentration levels in the specimen [1.1, 1.18-1.20]. More reliable quantitative
information about dopant concentration changes can be obtained via the use of an
electron energy analyzer, which relies on detecting shifts in the scattered SE energy
spectrum, in a manner similar to that used in Electron Beam Testers [1.21]. This
approach has been reported in a recent work by Kazemian et al [1.16, 1.17] and by
some other researchers [1.12-1.14]. However, they use relatively poor electron
Chapter 1
4
energy analyzer designs, from a signal-to-noise point of view, limiting the accuracy
to which shifts in the SE energy spectrum can be monitored. The minimum shift in
the SE energy spectrum which can be detected by their methods is typically in the
“a-tenths of an eV” range. To determine dopant concentration levels in a
semiconductor specimen, better accuracy is required; shifts typically less than a few
meV need to be detected. The accuracy of measurement is fundamentally
determined by the signal–to–noise characteristics of the SE electron energy
analyzer used to obtain the SE spectrum. Better energy analyzer designs have
already been presented within the context of quantitative voltage contrast, and an
obvious starting point is to use some of these ideas for the application of dopant
concentration mapping. Apart from dopant concentration mapping, the technique
of using SE energy analyzers inside the SEM can be applied to other types of
specimens, in a search to find more useful contrast mechanisms.
Recently, high signal-to-noise experimental results have been obtained from a
second–order focusing toroidal electron energy analyzer by Khursheed et al.[1.22];
they were able to measure shifts in a signal related to the SE energy spectrum in the
sub eV range [1.23]. However these results were obtained under the idealized
condition of a field free region above the specimen. In practice, the SEs that leave
the specimen are affected significantly by a variety of different surface fields and
fringe fields. Fringe fields are generated from potential differences between the
sample and surrounding electrodes, while surface fields occur due to localized
potential variations caused by primary beam induced effects such as specimen
charging and contamination. More work is required in order to establish whether
Chapter 1
5
the superior signal-to-noise characteristics, reported for the second–order focusing
toroidal analyzer, can be obtained under more realistic specimen conditions like
where fringe fields and surface fields are present.
After their work on the second–order toroidal analyzer, Hoang and Khursheed
presented the Radial Mirror Analyzer (RMA) attachment design for the SEM
[1.24]. Simulation results predict that the RMA design has several advantages over
the second–order focusing toroidal energy analyzer, both in terms of its integration
to the SEM’s primary beam optics, and its own dispersive/focusing properties.
However, a practical RMA prototype needs to be made and tested in order to verify
the simulations predictions. Another promising area for research is to find a feasible
parallel energy analyzer attachment design, where the SE spectrum is captured by
a number of detection channels, all operating simultaneously, instead of the
conventional approach of using a single detection channel operating sequentially.
This promises to greatly speed up data-acquisition times, and may therefore provide
a practical way of mapping secondary electron energy spectral information and
overlaying it on top of the SEM’s topographical image.
The main objective of this thesis work is to further develop the use of secondary
electron energy analyzer attachments for the SEM. The first aim is to investigate
how the second–order focusing toroidal energy analyzer attachment functions in
the presence of surface fields, and whether it can be used to make high signal-to-
noise voltage and dopant concentration measurements on semiconductor
specimens. Experimental results were obtained to demonstrate that although the SE
analyzer signals are greatly changed under the influence of fringe fields and surface
Chapter 1
6
fields, they can nevertheless provide high signal-to-noise voltage and dopant
concentration measurements on semiconductor specimens. The second aim of the
thesis is to use the technique of secondary electron energy filtering in the SEM to
find new useful contrast mechanisms. A new application of detecting trapped
charge at the interface of multi-layer thin insulator films was found, and the results
point towards the development of a new SEM analytical technique. The third aim
of the thesis is to experimentally test the RMA design and compare its performance
to simulation predictions. A prototype of the RMA was made and tested inside a
SEM, and the experimental results verified the design principle of the analyzer. The
fourth objective of the thesis is to develop a viable multi-channel secondary
electron energy analyzer design, one that can capture the SE energy spectrum in
parallel. The work carried out in this section of the thesis led to the development of
new Parallel Radial Mirror Analyzer design [1.25].
Chapter 1
7
References
1.1. Goldstein, J., et al., Scanning electron microscopy and X-ray microanalysis.
2003: Springer.
1.2. Luo, T. and A. Khursheed, Elemental identification using transmitted and
backscattered electrons in an SEM. Physics Procedia, 2008. 1(1): p. 155-
160.
1.3. Niedrig, H. and E. Rau, Information depth and spatial resolution in BSE
microtomography in SEM. Nuclear Instruments and Methods in Physics
Research Section B: Beam Interactions with Materials and Atoms, 1998.
142(4): p. 523-534.
1.4. Jbara, O., et al., Surface potential measurements of electron-irradiated
insulators using backscattered and secondary electron spectra from an
electrostatic toroidal spectrometer adapted for scanning electron
microscope applications. Review of Scientific Instruments, 2001. 72(3): p.
1788-1795.
1.5. Schlichting, F., D. Berger, and H. Niedrig, Thickness determination of
ultra‐thin films using backscattered electron spectra of a new toroidal
electrostatic spectrometer. Scanning, 1999. 21(3): p. 197-203.
1.6. Cubric, D., et al., Parallel acquisition electrostatic electron energy analyzers
for high throughput nano-analysis. Nuclear Instruments and Methods in
Physics Research Section A: Accelerators, Spectrometers, Detectors and
Associated Equipment, 2011. 645(1): p. 227-233.
1.7. El-Gomati, M., C. Walker, and X. Zha, Towards quantitative scanning
electron microscopy: Applications to nano-scale analysis. Nuclear
Chapter 1
8
Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment, 2011. 645(1): p. 68-
73.
1.8. Tsurumi, D., K. Hamada, and Y. Kawasaki, Energy-filtered imaging in a
scanning electron microscope for dopant contrast in InP. Journal of Electron
Microscopy, 2010. 59(S1): p. S183-S187.
1.9. Chung, S., et al., Secondary electron dopant contrast imaging of compound
semiconductor junctions. Journal of Applied Physics, 2011. 110(1): p.
014902.
1.10. Tsurumi, D., K. Hamada, and Y. Kawasaki, Energy-Filtered Secondary-
Electron Imaging for Nanoscale Dopant Mapping by Applying a Reverse
Bias Voltage. Japanese Journal of Applied Physics, 2012. 51(10R): p.
106503.
1.11. Tsurumi, D., K. Hamada, and Y. Kawasaki, Highly Reproducible
Secondary Electron Imaging under Electron Irradiation Using High-Pass
Energy Filtering in Low-Voltage Scanning Electron Microscopy.
Microscopy and Microanalysis, 2012. 18(02): p. 385-389.
1.12. Heath, J.T., C.-S. Jiang, and M.M. Al-Jassim, Measurement of
semiconductor surface potential using the scanning electron microscope.
Journal of Applied Physics, 2012. 111(4): p. 046103.
1.13. Rau, E. and A. Tagachenkov, Image contrast of impurity regions of
semiconductor crystals in scanning electron microscopy. Bulletin of the
Russian Academy of Sciences: Physics, 2013. 77(8): p. 943-947.
Chapter 1
9
1.14. Chee, A.K., et al., A quantitative model for doping contrast in the scanning
electron microscope using calculated potential distributions and Monte
Carlo simulations. Journal of Applied Physics, 2011. 109(1): p. 013109.
1.15. El‐Gomati, M., et al., Why is it possible to detect doped regions of
semiconductors in low voltage SEM: a review and update. Surface and
interface analysis, 2005. 37(11): p. 901-911.
1.16. Kazemian, P., et al., High resolution quantitative two-dimensional dopant
mapping using energy-filtered secondary electron imaging. Journal of
Applied Physics, 2006. 100(5): p. 054901.
1.17. Kazemian, P., et al., Quantitative secondary electron energy filtering in a
scanning electron microscope and its applications. Ultramicroscopy, 2007.
107(2–3): p. 140-150.
1.18. Oatley, C.W., The scanning electron microscope. Sci. Prog., Oxf, 1966. 54:
p. 483-495.
1.19. Reimer, L., Scanning electron microscopy: physics of image formation and
microanalysis. Measurement Science and Technology, 2000. 11(12): p.
1826.
1.20. Wells, O.C., Scanning electron microscopy. 1974: McGraw-Hill.
1.21. Thong, J.T., Electron beam testing technology. 1993: Plenum Publishing
Corporation.
1.22. Khursheed, A. and H.Q. Hoang, A second–order focusing electrostatic
toroidal electron spectrometer with 2π radian collection. Ultramicroscopy,
2008. 109(1): p. 104-110.
Chapter 1
10
1.23. Hoang, H., M. Osterberg, and A. Khursheed, A high signal-to-noise ratio
toroidal electron spectrometer for the SEM. Ultramicroscopy, 2011. 111(8):
p. 1093-1100.
1.24. Hoang, H.Q. and A. Khursheed, A radial mirror analyzer for scanning
electron/ion microscopes. Nuclear Instruments and Methods in Physics
Research Section A: Accelerators, Spectrometers, Detectors and Associated
Equipment, 2011. 635(1): p. 64-68.
1.25. Khursheed, A., H.Q. Hoang, and A. Srinivasan, A wide-range Parallel
Radial Mirror Analyzer for scanning electron/ion microscopes. Journal of
Electron Spectroscopy and Related Phenomena, 2012. 184(11): p. 525-532.
Chapter 2
11
Chapter 2 – Introduction to the SEM
A good understanding of the working of the SEM is essential to understand and
appreciate how electron energy analyzers inside the SEM work and how the SEM
can be used to obtain useful analytical information about the sample in addition to
high resolution topographical images, extracted by the use of energy analyzers.
Fig. 2.1 [2.1] shows a schematic drawing of the layout of a conventional SEM. A
typical SEM column consists of an electron gun, two condenser lenses, an aperture,
an objective lens, an electron detection system, and a set of deflectors, all operating
in a vacuum.
The electron gun, which can be of thermionic tungsten or LaB6 gun or field
emission type, acts as a source of electrons which are accelerated to energies in the
range of 1 to 30 keV inside the column and then focused into an electron probe of
diameter 1 to 10 nm carrying current in the range of 1 to 100 pA. For obtaining
higher current, larger probe diameters may be used, this however would result in a
trade–off on the spatial resolution of the SEM.
This electron probe, also called the primary beam spot, is scanned in a raster–like
pattern across the specimen by the deflection system in front of the final lens and
is operated in synchronization with the detector electronics to produce
topographical images of the specimen. Depending on the type of scattered electron
detected from specimen, the SEM can produce different types of high resolution
images like SE images or BSE images.
Chapter 2
12
Fig. 2.1 – Schematic drawing of a conventional SEM
The primary beam/specimen interaction produces various types of signals which
are a useful source of both topographical and analytical information about the
specimen. A schematic representation of this beam/specimen interaction is shown
in Fig. 2.2a. The primary beam penetrates into the specimen, and in the process
scatters electron of various energies through elastic and inelastic collisions.
Electrons which are scattered inelastically near the surface of the specimen are
Electron Gun
Chapter 2
13
known as secondary electrons (SEs), while electrons that are generated through
multiple elastic collisions from deeper levels are known as backscattered electrons
(BSEs). Secondary electrons and backscattered electrons make up the most
common signals for imaging using the SEM. Auger electrons (AEs) are electrons
having characteristic energies that are ejected from atoms by absorbing the energy
released by the relaxation of another electron into a lower energy inner–shell
vacancy. The energy distribution of the SEs, BSEs and AEs is shown in Fig. 2.2b.
SEs are low energy electrons, usually defined to be below 50 eV. Even in this range,
most secondary electrons lie between 0.5 eV and 5 eV. BSEs on the other hand, by
definition, are electrons with energies from 50 eV up to the primary beam energy.
It is important to understand how the primary beam is scattered inside the specimen.
This can be done by using Monte Carlo simulation programs developed for this
purpose.
a
Chapter 2
14
Fig. 2.2 – Beam/Specimen interaction: (a) Interaction volume and emitted signals
(b) Energy spectrum of electrons that leave the specimen.
Fig. 2.3 shows the results of Monte Carlo simulations carried out using one such
program, Casino [2.2]. For a thick specimen (thickness > 1 µm), it can be seen that
the depth of interaction volume, R (refer Fig. 2.2a) increases with increasing
primary beam energy. The depth of interaction volume also depends on the nature
of the specimen such as mass, density and atomic number.
An approximate formula for estimating R has been given by Kanaya and Okayama
[2.3]
𝑅 =0.0276𝐴𝑉1.67
𝑍0.889𝜌(µm)
where V is the primary beam voltage, A is the atomic weight, is the mass density
and Z is the atomic number.
b
Chapter 2
15
Fig. 2.3 – Monte Carlo simulation of primary electrons striking a silicon sample at
energies 1, 5, 10 and 15 keV.
2.1 Output signals inside the SEM
Fig. 2.4 schematically depicts the detector systems inside the SEM to detect
secondary electrons and backscattered electrons. As already stated, these signals
are the two most common electron signals inside the SEM.
The most common type of secondary electron detector was proposed by Everhart
and Thornley [2.4] and is often simply referred to as the E–T detector. The E–T
detector is electrically isolated from the rest of the SEM and consists of an outer
wire mesh that is typically biased up at 200 to 300 V, attracting the low–energy
1 µ
m
1 keV
10 keV
5 keV
15 keV
Chapter 2
16
secondaries inside the mesh, where they are accelerated by a positive voltage on a
scintillator, typically biased up at 10 – 12 keV.
Fig. 2.4 – Conventional detector layout inside the SEM.
The accelerated electrons produce photons when they hit a scintillator, which are
then amplified by a photomultiplier tube to convert the light into amplified
electrical signals.
Backscattered electron detectors on the other hand, are placed under the final lens
lower pole piece and are typically in the form of a disc with a hole in the center
through which the primary beam can pass. These detectors may either be of a
micro–channel plate type or a silicon p–n junction photodiode type. Recent electron
energy analyzers [2.5, 2.6] have also been designed to occupy a similar position as
BSE detectors, since it does not obstruct other detectors (EDS), and this provides
the opportunity for concurrent SE imaging while acquiring SE spectral information.
Chapter 2
17
Fig. 2.5 [1] shows the different ways in which the secondary electrons are generated
inside the SEM. While SE1 are the SEs generated by the primary beam within a few
nanometers from the surface of the specimen, SE2 are SEs generated within the
same region by backscattered electrons. Secondaries generated by the scattering of
backscattered electrons on external surfaces other than the specimen (such as
chamber walls, the lower lens pole–piece) are called SE3. As SE1 emanate from the
top "few–nanometers" layer of the specimen, they are capable of providing high
resolution information about the specimen. SE2 and SE3 on the other hand emanate
from indirect interactions from backscattered electrons and therefore, in general,
degrade the final image resolution.
Fig. 2.5 – Secondary electron signal generation.
Chapter 2
18
In general, the Chung–Everhart distribution [2.7] is a good approximation to the
energy distribution of secondary electrons. The distribution is given by
𝑑𝑁
𝑑𝐸=
6𝑊2𝑁𝑠𝑜𝐸
(𝐸 + 𝑊)4
where E is the kinetic energy of the SEs leaving the specimen, W is the specimen
work function, and NS0 is the total number of secondaries that are emitted. A plot
of this distribution for gold (W = 5.1 eV) is shown below in Fig. 2.6.
As seen in Fig. 2.6, the distribution of SE increases steeply at lower energies and
then falls down gradually for higher energies, indicating that most secondaries have
energies below 5 eV. This graph is usually referred to as the normal SE energy
distribution.
Fig. 2.6 – The SE Chung–Everhart energy distribution.
Gold
Chapter 2
19
Clearly, the SE spectrum will vary for various types of specimens (as their work
functions differ) and for different specimen conditions (like biasing of the
specimen). To illustrate this, Fig. 2.7 below shows the variation of the SE spectrum
for specimens with work functions 4, 4.5 and 5 eV.
Fig. 2.7 – SE energy distribution with work function variations.
Other analytical information that SE signals can provide and which are commonly
extracted from the SE energy distribution inside a SEM are specimen surface
potential and dopant concentration. Fig. 2.8a illustrates the variation of the SE
energy spectrum with specimen voltage. The kinetic energy of a secondary electron
is increased or decreased depending on the sample biasing, and this causes the SE
spectrum to shift. For negative changes in the specimen potential, the SE spectrum
shifts to the right.
The SE spectrum also varies with dopant type and dopant levels, as depicted in Fig.
2.8b above for a p-n junction.
Work Function (eV)
4
4.5
5
Chapter 2
20
Fig. 2.8 – Variation of SE energy distribution with changes in (a) Specimen
potential and (b) Dopant type and concentration across a p–n junction.
Conventional SEMs are not able to extract the information shown in Fig. 2.8, as
they are not designed to capture the energy spectrum of the scattered electrons that
leave the specimen. The bulk output signal, used to form a conventional SEM
image, is formed from SEs that are emitted over a wide range of energies and
angles. In order to capture the SE energy spectrum inside a SEM, some form of
energy analyzer needs to be incorporated into its specimen chamber.
The accuracy with which specimen voltage or dopant concentration changes can be
quantified depends primarily on the ability of the SE energy analyzer to detect small
shifts (fractions of an eV) in the SE energy distribution. For monitoring shifts in
the SE spectrum the usual practice was to track its peak position [2.8], however a
more general and holistic method was suggested by Khursheed [2.9] who
monitored the change in the expectation value of the SE energy curve to quantify
its shifts. The expectation value not only tracks the peak position of a signal but
a b
Chapter 2
21
also responds to changes in its shape and therefore is a more useful parameter for
asymmetric distributions like the SE spectrum. This has been discussed in detail in
chapter 3 of this thesis.
Other common factors that affect the accuracy of the voltage measurement are
surface fields on the specimen to be discussed in more detail in the next chapter. A
variety of different analyzers for quantitative voltage contrast have been developed
and their relative weaknesses and strengths will be discussed in the subsequent
sections.
2.2 Objective lens improvements
One important factor determining the design of a SE energy analyzer design
attachment is how it fits together with the SEM objective lens. A variety of
objective lenses have been developed for the SEM, as shown in Fig. 2.9. In the
conventional SEM, the specimen is placed in the free–field region below the final
pole–piece of the objective lens as illustrated in Fig. 2.9a [2.10]. The working
distance in this case is defined by the distance between the final pole piece of the
SEM objective lens and the specimen, normally ranging from 5 mm to 30 mm.
Electron energy analyzers designed to integrate with such an objective lens
arrangement need to be directly placed in between the lower pole–piece of the
objective lens and the specimen, in the same general region as the standard SE or
BSE detectors, as illustrated in Fig. 2.10.
The main drawback of this arrangement is that placing the analyzer below the SEM
objective lens results in a large working distance (W). A larger working distance
increases the on–axis aberrations of the primary beam spot on the specimen, which
Chapter 2
22
in turn significantly degrades the image resolution. Also, another challenge lies in
acquiring signals from the analyzer while concurrently allowing operation of the
SEM in its imaging mode (using the conventional E–T detector). Analyzer designs
and prototypes reported in chapter 5 allow for this possibility.
Fig. 2.9 – Different types of SEM objective lenses: (a) Conventional lens (b)
Magnetic In–lens (c) Single pole lens below the specimen (d) Single pole lens
above the specimen (e) Retarding field lens and (f) Mixed–field immersion lens
[2.10, 2.11].
c d
a b
e f
Chapter 2
23
Fig. 2.10 – Arrangement of an energy spectrometer for conventional objective lens
type SEMs.
The objective lens designs shown in Figs. 2.9b–f were developed to either to
improve the SEM spatial resolution (smaller primary beam spot) and/or lower the
landing energy of the primary beam on the specimen. They include a magnetic in–
lens design, where the specimen is placed in the lens gap region, as shown in Fig.
2.9b; semi–in lenses where the magnetic field extends beyond a single lens pole–
piece (Figs. 2.9c–d), a retarding field lens, where the primary beam is slowed down
just before it strikes the specimen (Fig. 2.9e); and a mixed field immersion lens,
where the specimen is immersed in both a retarding electric field and a strong
magnetic field (Fig. 2.9f). A more detailed review of these types of objective lens
improvements can be found in the work presented by Khursheed [2.10]. In the
present work, only secondary electron energy analyzers for conventional lenses will
be considered (shown in Fig. 2.9a), since they can be made as attachments that can
be readily placed in the specimen chamber. For the other types of objective lenses,
Chapter 2
24
secondary electrons must first travel back through the objective lens bore before
their energies can be analyzed. General purpose energy analyzer attachments for
SEMs that use such objective lenses are difficult to design, since access to the
region either above or inside the objective lens is not usually provided.
2.3 Secondary electron energy analyzers SEM attachments
Even within the category of conventional objective lenses, a wide variety of
different analyzer designs have been developed. As mentioned earlier, traditionally,
analyzers were developed to fit between the final pole piece of the objective lens
and the sample. This has the undesirable effect of increasing the working distance,
increasing on–axis aberrations of the primary beam spot on the specimen, and
therefore resulting in a poorer image resolution. Ideally, the spectrometer needs to
fit into the vacuum chamber of the specimen chamber, without increasing the
working distance. In addition, it should also be designed to have high transmittance,
high energy resolution and/or high signal–to–noise characteristics. Transmittance
in this context is the collection efficiency of the spectrometer, defined as the
fraction of the number of electrons at a particular energy which reach the detector
to the number of electrons that leave the sample. Depending on the application, the
energy resolution of the analyzer may also be important; that is the ability of the
analyzer to distinguish between two different peaks in the energy spectrum. In
general, a high transmittance and a high energy resolution are preferred for most
applications. However, for SE energy analyzers, energy resolution is not as
important as the ability to detect shifts in the spectrum or changes in its shape,
which requires minimizing the effects of noise on the spectrum (maximizing
signal–to–noise). More will be said about this important point in the next section
Chapter 2
25
and in a later chapter. In the following section, various types of spectrometers for
different applications in the SEM are discussed.
2.4 Retarding Field Analyzers
As already mentioned, SE energy analyzers in the SEM have generally been used
for quantifying voltage contrast. One of the first types of analyzers used for
quantifying voltage contrast was the retarding field analyzer, as illustrated in the
schematic diagram shown in Fig. 2.11.
Retarding field analyzers work on the principle that the scattered SE electrons are
made to travel in a retarding electric field, defined by the difference between a
retarding grid voltage, VR, and the specimen voltage, VS. Only the secondaries with
an energy greater than e(VR– VS) will surmount the retarding grid and reach the
detector. The signal at any particular value of VR is the sum of the contributions of
all the electrons beyond the potential barrier and is represented by the shaded region
in the SE spectrum shown in Fig. 2.11. This type of analyzer therefore works like
a high pass filter, and the SE energy spectrum is collected in its integrated form.
When the retarding field spectrometer is used for voltage contrast applications, it is
operated in a closed feedback mode [2.12]. In such a mode of operation, the strength
of the potential barrier is varied using a feedback loop to maintain a constant current
output. In this case, ΔVR is the change in VR needed to maintain a constant current
output equals the change in specimen voltage. The output signal of the retarding
field analyzer is an S–curve, as illustrated in Fig. 2.12, the integrated form of the
SE spectrum. The figure also shows the shift in the S–curve when the voltage of
the specimen changes from VS1 to VS2.
Chapter 2
26
Fig. 2.11 – General layout of a closed loop retarding field spectrometer with
hemispherical grid.
Fig. 2.12 – Output S–curve signals of the retarding field analyzer.
Objective lens
PE
Extraction Grid
Retarding Grid VR E2
E1 Specimen VS
VS
VS1 VS2
VR
Signal offset
Chapter 2
27
Fentem and Gopinath [2.13] proposed an early voltage contrast retarding field
analyzer using hemispherical grids for voltage contrast applications. Subsequent
designs of retarding field analyzers using planar grids were reported by Plows
[2.14], Flemming and Ward [2.12], and Gopinath and Sanger [2.15].
Several variations of the retarding field analyzer design as an SEM attachment have
been made in the context of electron beam testing of Integrated Circuits [2.16-2.18].
Retarding Field analyzer designs were also developed for magnetic immersion
objective lenses, such as those shown in Figs. 2.9c–e, where the secondaries travel
back through the objective lens bore. These “through–the–lens” arrangements were
reported by Menzel and Buchanan [2.19], Garth [2.20], Frosien and Plies [2.21]
and Dinnis [2.22], decreasing the working distance significantly to about 2–5 mm.
It is important to note that these types of spectrometers were mainly designed to
monitor the shifts in the SE spectrum, primarily for the application of probing tracks
in Integrated Circuits.
2.5 Signal–to–Noise considerations
There are several processes that contribute to the noise of any voltage measurement
carried out inside the SEM. These are variations in the primary beam current, noise
generated when electrons scatter inside the specimen and electron analyzer, and
noise of the detection system. A detailed analysis of this can be found in a book by
Reimer [2.23] and in a book chapter by Dubbeldam [2.24].
The generation of electrons in the electron gun is a random process following a
Poisson distribution that arises from statistical fluctuations in the number of emitted
Chapter 2
28
electrons, and is generally referred to as shot noise. The signal-to-noise ratio of the
primary beam electrons (𝑆
𝑁)
𝑃𝐸 is given by [2.23]
(𝑆
𝑁)
𝑃𝐸= √(
𝐼𝑃𝐸
2𝑒𝑓)
where IPE is the primary beam current, e is the electron charge and f is the
bandwidth of the detector system.
A primary beam electron can either be absorbed or be elastically scattered inside
the specimen and therefore backscattered electron (BSE) generation statistically
has a binomial distribution. The overall effect of the Poisson distribution of
electrons in the primary beam and the binomial distribution of the BSEs is a Poisson
distribution and the signal-to-noise of this process (𝑆
𝑁)
𝐵𝑆𝐸 is given as [2.23]
(𝑆
𝑁)
𝐵𝑆𝐸= √(
𝐼𝑃𝐸
2𝑒𝑓)
where is the backscattering coefficient, a parameter dependent on nature of the
specimen.
In contrast to the statistics that govern the generation of primary beam electrons
and BSEs, the statistics of Secondary electrons (SE) that emanate from the
specimen is neither a Binomial distribution nor a Poisson distribution because a
single primary electron (or a BSE generated by a PE) can generate zero, one or
many SEs. The signal-to-noise on account of the SE generation process is given as
[2.23]
Chapter 2
29
(𝑆
𝑁)
𝑆𝐸= √(
𝐼𝑃𝐸
2𝑒𝑓(1 + 𝑏))
where the variable b is a noise factor which takes into account the fluctuations of
primary electrons and their probability of generating SEs inside the specimen, b is
dependent on the SE yield factor .
The detector noise is relatively small compared to the shot noise of the primary
beam and the SE emission noise, under standard operating conditions [2.24, 2.25].
In the context of retarding field analyzers used for voltage contrast applications, the
shift in the energy spectrum is dominated by shot noise. Gopinath reported that the
minimum resolvable specimen voltage change was mainly limited by shot noise of
the primary beam [2.26]. He derived a formula to characterize voltage resolution of
retarding field analyzers, where the voltage resolution VS is given by
𝑉𝑆 = 𝐾2 (𝑓
𝐼𝐷)
where ID is the detector current, ∆f is the bandwidth of measurement system and
K is a spectrometer constant. This formula established an experimentally verified
figure of merit for retarding field analyzers, and was widely used for quantitative
voltage contrast applications.
The most important demerit of the retarding field analyzer is its relatively low
output signal–to–noise ratio. As the analyzer works in a high pass filter mode, only
a small number of electrons that reach the detector are sensitive to variations of the
filter grid potential. Therefore a small change in the threshold potential causes the
signal to vary on large noise background, reducing the sensitivity of the voltage
Chapter 2
30
change measurement. The contribution of the BSEs and SE3s generated from the
grids add to the output noise.
Detailed signal–to–noise analysis was first carried out by Dubbledam and Kruit
[2.27], comparing the predicted voltage resolution, ΔVS, analyzers that capture the
integrated form of the Chung–Everhart SE energy spectrum (retarding field
analyzer) and energy analyzers that acquire the SE energy spectrum directly (non–
integrated form). The results are summarized by Fig. 2.13, where the Voltage
Resolution Constant χ, links the voltage resolution to the number of electrons that
leave the specimen, N, by χ = (ΔVS)2N. The cut–off energy denotes the value of the
energy barrier in the retarding field spectrometer, e(VR– VS). The voltage resolution
constant χ is therefore proportional to the number of electrons required to obtain a
given voltage resolution, and its value for retarding field spectrometers is compared
to multi–channel spectrometers, which in this context, denotes spectrometers that
capture the SE spectrum directly, without integrating it. Fig. 2.13 indicates that for
small potential barriers (for points close to the top of the S–curve), the number of
electrons that retarding field analyzers need to attain a given voltage resolution is
one to two orders of magnitude higher than that required by analyzers that are able
to capture the SE energy spectrum directly. As the potential barrier strength grows,
where only higher energy secondaries make up the output signal, the difference in
the signal–to–noise characteristics of both analyzer types becomes smaller. These
results show that from a signal–to–noise point of view, there is an optimum point
mode of operation for retarding field analyzers, typically where the internal
potential barrier lies between 1 to 2 volts. They also indicate that operating close to
Chapter 2
31
the top of the S–curve (low potential barrier) should be avoided. These results
predicted analytically by Dubbledam and Kruit [2.27], were later confirmed
numerically by Khursheed using Monte–Carlo simulations [2.9].
Fig. 2.13 – Comparison of the signal–to–noise characteristics of retarding field
analyzers with multi–channel energy analyzers as a function of cut–off energy in
the SE Chung–Everhart spectrum [2.27].
Following on from the development of quantitative voltage contrast for probing of
IC circuits, SE energy analyzers have been proposed for quantification of
semiconductor dopant mapping of p–n junctions. Kazemian et al. recently reported
on using energy filtered SE signals inside the SEM, where shifts in SE spectra are
measured and used to quantify dopant concentration across a p–n junction [2.28,
2.29]. Fig. 2.14 shows the general layout of the “through–the–lens” detector used
for these experiments. Kazemian et al. use a field emission magnetic immersion
Chapter 2
32
objective lens SEM where an electrostatic deflector and off–axis detector are
inserted into the upper pole–piece of the lens.
The specimen is placed below the upper pole piece at the peak position of the axial
magnetic field (or just below it) as shown in Fig. 2.14. An extraction field is applied
to extract the secondaries. Due to the presence of a magnetic field, the secondaries
spiral up, past the lens–bore, and depending on the extraction field strength,
experience some degree of collimation. These electrons are then deflected on to a
detector placed on one side by electrostatic deflector plates. Although the deflection
action is a relatively broad one in terms of the secondary electron energy range that
is detected, there is however, a filtering effect on the higher energy electrons. The
deflector voltage is swept over a certain range and shifts in a signal related to the
SE spectra are obtained.
Fig. 2.14 – Schematic of magnetic immersion lens SE analyzer layout of Kazemian
et al. [2.28, 2.29] used for quantitative dopant mapping [2.1].
Chapter 2
33
The shape of the S–curves as reported by Kazemian et al. are irregular and deviate
from the shape of regular S–curves. This is because, in addition to it being a
retarding field analyzer, the setup also filters higher energy scattered electrons
which travel up through the lens bore. Monitoring of the shifts in the SE spectrum
was therefore limited to the top linear region of the spectra. For the reasons
indicated by Fig. 2.14, this led to poor signal–to–noise ratios. These reasons were
however, not mentioned by Kazemian et al. [2.28, 2.29]. The precision to which
they measured the p–n junction voltage was 0.72 ± 0.15 V, a signal–to–noise ratio
of 4.8 [2.29]. A later attempt measured the surface potential to be 0.81 ± 0.1 V
[2.28].
2.6 Deflection/multi–channel analyzers
Deflection analyzers, in contrast to retarding field analyzers, are band–pass
analyzers and at any given time, they only detect a very small part of the SE energy
range (in the meV range), a narrow pass range. They function by ramping an
electrode voltage inside the analyzer, which has the effect of changing the pass
range, so that the whole SE energy spectrum can be detected as a time varying
signal. In this case, the output signal is not an integrated form of the SE energy
spectrum, but as long as the pass range has a relatively small energy width, the
signal will directly represent the shape of the SE spectrum. An example of a
deflection band–pass analyzer developed for quantitative voltage contrast
measurements on IC circuits was given by Hannah [2.8], and is shown in Fig. 2.15a.
After secondary electrons are extracted up from the specimen, they are deflected
by the lower and upper deflection plates, which travel through an opening, defined
Chapter 2
34
by collimator plates, and are detected by the SEM scintillator. The collimator plates
can be adjusted to pass a small secondary electron energy range, and the deflection
plates are ramped in time. An example of the experimental output signals obtained
by Hannah are shown in Fig. 2.15b, illustrating how they shift as a function of
specimen voltage.
Fig. 2.15 – The 63° CDA Hannah voltage contrast spectrometer [2.8] (a)
Spectrometer layout (b) Experimentally acquired SE spectra for different specimen
voltages.
b
a
Chapter 2
35
Since bandpass analyzers capture the SE energy spectrum directly, they will, for
reasons already explained (Fig. 2.13), have better signal–to–noise ratio
characteristics than retarding field analyzers. Also, bandpass analyzers can track
changes in the shape in the SE energy spectrum instead of only detecting changes
in the overall number of collected electrons, as with the case of retarding field
analyzers. Furthermore, band–pass analyzers have a distinct advantage over
retarding field secondary electron analyzers in situations where the point to be
measured takes high negative potential values; their signals continue to shift right,
unlike the retarding analyzer whose signals remain approximately constant beyond
a value (top of the S–curve).
SE energy analyzer design for voltage contrast applications led to several multi–
channel SE analyzers, where the entire SE spectrum is acquired in parallel by an
array of energy channels, each capturing a different portion of the energy spectrum.
As long as the energy–width of each channel is small, like band–pass analyzers,
they capture the SE energy spectrum directly, not in its integrated form as in the
case of retarding field analyzers. They do not need any analyzer voltage electrode
to be ramped in time, and are therefore have much faster data–acquisition times
than both retarding field analyzers and band–pass analyzers. Multi–channel SE
energy analyzers were reported by Dubbledam and Kruit [2.30], Khursheed and
Dinnis [2.9], Khursheed [2.31], Khursheed and Karuppiah [2.32] and Kienle and
Plies [2.33] and discussed in detail in a book by Khursheed [2.34].
2.7 Full range deflection/multi–channel analyzer designs
The energy analyzers described so far, were designed primarily to capture the
secondary electron energy range. However, there have been some bandpass/multi–
Chapter 2
36
channel energy analyzer proposals designed to capture the complete energy
spectrum of scattered electrons in the SEM. One such spectrometer is an
electrostatic toroidal deflection analyzer attachment reported by Rau and Robinson
[2.35] (depicted in Fig. 2.16) and was used to capture the BSE spectrum from the
specimen under test. This analyzer attachment is designed to be placed in between
the objective lens and the specimen, resulting in a large working distance.
Therefore, it is not suitable for capturing energy spectral information on the nano–
scale range. However, Rau et al. demonstrated that this analyzer can be used both
as an analytical and an imaging tool inside the analyzer; as an imaging tool, the
analyzer was able to provide tomography information of subsurface structures and
as an analytical tool it was used to produce BSE spectra from different materials.
Fig. 2.16 – Schematic layout of the electrostatic toroidal deflection analyzer
reported by Rau and Robinson [2.35].
Recently, two new full range deflection analyzer design attachments were reported
by Hoang and Khursheed, namely the second–order focusing toroidal electron
energy analyzer [2.36] and later, the radial mirror analyzer (RMA) [2.6]. In addition
Chapter 2
37
to capturing both the SE and BSE energy spectra, these electric band–pass analyzer
attachments are designed to have high energy resolution optics, comparable to or
better than energy analyzers that are normally used for Auger Electron
Spectroscopy [2.37, 2.38]. Both analyzers are rotationally symmetric about the
primary beam axis and function concurrently with the SEM’s normal imaging mode
of operation. They are designed to minimize the working distance (distance from
specimen to lower pole–piece of the objective lens) and maximize transport
efficiency of scattered electrons to the analyzer detector. In the category of general
purpose add–on energy analyzer attachments for the SEM, they are the most
versatile designs reported so far, and they form an important starting point for the
work to be carried out here.
Fig. 2.17 shows the layout of the second–order toroidal spectrometer prototype
reported by Khursheed and Hoang [2.39]. This analyzer is predicted to have an
energy resolution that is comparable to the well–known Cylindrical Mirror analyzer
for the same acceptance angle [2.37]. The prototype was designed as an attachment
that fits on to the specimen stage. Although capable of full 2π collection, the
spectrometer was manufactured to collect 90˚ in the azimuthal direction, in order
to enable simultaneous viewing of the specimen by the conventional SE E–T
detector. The electrons which pass through the spectrometer are detected by a small
photo–multiplier tube (PMT) detector located beneath the specimen.
Experimental results obtained from the second–order focusing toroidal analyzer
have been shown to have excellent signal–to–noise characteristics, thereby
demonstrating that the analyzer inherently offers very high signal–to–noise ratios,
Chapter 2
38
typically allowing spectral shift as small as 12 mV or lower to be monitored.
Furthermore, when the specimen is biased at higher negative voltages, the signal–
to–noise of the output improves, allowing spectral changes as low as 4 mV to be
recorded. However, these results were obtained under idealized specimen
conditions; more work is required to investigate whether these high signal–to–noise
characteristics can be made for real applications such as quantitative dopant
concentration mapping.
Fig. 2.17 – Layout of a second–order focusing toroidal analyzer prototype
attachment [2.11].
Experimental results obtained from the second–order focusing toroidal analyzer
have been shown to have excellent signal–to–noise characteristics, thereby
demonstrating that the analyzer inherently offers very high signal–to–noise ratios,
typically allowing spectral shift as small as 12 mV or lower to be monitored.
Furthermore, when the specimen is biased at higher negative voltages, the signal–
Chapter 2
39
to–noise of the output improves, allowing spectral changes as low as 4 mV to be
recorded. However, these results were obtained under idealized specimen
conditions; more work is required to investigate whether these high signal–to–noise
characteristics can be made for real applications such as quantitative dopant
concentration mapping.
The RMA reported by Hoang et al. [2.6], is predicted to have a relative energy
resolution of better than 0.025% for a polar angular spread of ±6°. This is around
an order of magnitude better than the Cylindrical Mirror analyzer for the same
acceptance angle [2.37] and comparable to the Hemispherical Deflection Analyzer
(HDA) [2.38]. The layout of the RMA, designed to fit as attachment inside scanning
electron/ion microscopes is presented later in this thesis (Refer Fig. 5.1).
The RMA simulation predictions, although promising, need to be tested
experimentally. Both the second–order focusing toroidal analyzer and the RMA
provide good starting points for further developing energy analyzer attachments for
the SEM.
2.8 Objectives of the thesis
The work carried out in this thesis aimed to explore new possibilities of using the
electron energy analyzer as an analytical tool inside the SEM. Firstly, the behavior
of the second–order focusing toroidal analyzer prototype of Hoang et al. for more
realistic conditions, like presence of surface fields and fringe fields above the
specimen was investigated. Experimental results were obtained which demonstrate
that the analyzer can be used to obtain high signal–to–noise signals even in the
presence of such fields, and how it can be applied for voltage and dopant
concentration measurements of semiconductors. Secondly, the toroidal energy
Chapter 2
40
analyzer was used on a variety of different specimens, looking for new contrast
mechanisms. This work led to the discovery of a new quantitative measurement
method for probing the buried charge present at interfaces of multifunctional
oxides. Thirdly, a proof–of–concept prototype of the RMA [2.6] was built and
experimentally tested to obtain results that establish the working principle of the
analyzer. Fourthly, numerical simulation techniques were used to design a new
promising full range parallel radial mirror analyzer (PRMA) design, one that is
capable of directly quantitatively mapping SE energy spectral information on to the
SEM’s conventional image.
Chapter 2
41
References
2.1. A. Khursheed, Scanning Electron Microscope Optics and Spectrometers:
World Scientific, 2011.
2.2. D. Drouin, A. R. Couture, D. Joly, X. Tastet, V. Aimez, and R. Gauvin,
"CASINO V2. 42—A Fast and Easy‐to‐use Modeling Tool for Scanning
Electron Microscopy and Microanalysis Users," Scanning, vol. 29, pp. 92-
101, 2007.
2.3. K. Kanaya and S. Okayama, "Penetration and energy-loss theory of
electrons in solid targets," Journal of Physics D: Applied Physics, vol. 5, p.
43, 1972.
2.4. T. Everhart and R. Thornley, "Wide-band detector for micro-microampere
low-energy electron currents," Journal of Scientific Instruments, vol. 37, p.
246, 1960.
2.5. A. Khursheed, H. Q. Hoang, and A. Srinivasan, "A wide-range Parallel
Radial Mirror Analyzer for scanning electron/ion microscopes," Journal of
Electron Spectroscopy and Related Phenomena, vol. 184, pp. 525-532,
2012.
2.6. H. Q. Hoang and A. Khursheed, "A radial mirror analyzer for scanning
electron/ion microscopes," Nuclear Instruments and Methods in Physics
Research Section A: Accelerators, Spectrometers, Detectors and Associated
Equipment, vol. 635, pp. 64-68, 2011.
Chapter 2
42
2.7. M. S. Chung and T. E. Everhart, "Simple calculation of energy distribution
of low‐energy secondary electrons emitted from metals under electron
bombardment," Journal of Applied Physics, vol. 45, pp. 707-709, 1974.
2.8. J. M. Hannah, "SEM applications to Integrated Circuit Testing," PhD,
University of Edinburgh, Scotland, Scotland, 1974.
2.9. A. Khursheed and A. Dinnis, "A time-of-flight voltage contrast detector for
measurements on VLSI circuits," Measurement Science and Technology,
vol. 1, p. 581, 1990.
2.10. A. Khursheed, "Recent developments in scanning electron microscope
design," in Advances in Imaging and Electron Physics. vol. Volume 115,
W. H. Peter, Ed., ed: Elsevier, 2001, pp. 197-285.
2.11. H. Hoang, "Energy spectrometers for the SEM," PhD Thesis, Electrical and
Computer Engineering, National Univeristy of Singapore, 2010.
2.12. J. Fleming and E. Ward, "A technique for accurate measurement and display
of applied potential distributions using the SEM," Scanning Electron
Microscopy, pp. 465-470, 1970.
2.13. P. Fentem and A. Gopinath, "Voltage contrast linearization with a
hemispherical retarding analyser," Journal of Physics E: Scientific
Instruments, vol. 7, p. 930, 1974.
2.14. G. Plows, "Stroboscopic scanning electron microscopy and the observation
of microcircuit surface voltages," University of Cambridge, 1969.
Chapter 2
43
2.15. A. Gopinath and C. Sanger, "A technique for the linearization of voltage
contrast in the scanning electron microscope," Journal of Physics E:
Scientific Instruments, vol. 4, p. 334, 1971.
2.16. Y. Goto, A. Ito, Y. Furukawa, and T. Inagaki, "Hemispherical retarding type
energy analyzer for IC testing by electron beam," Journal of Vacuum
Science and Technology, vol. 19, pp. 1030-1032, 1981.
2.17. L. Balk, H. Feuerbaum, E. Kubalek, and E. Menzel, "Quantitative voltage
contrast at high frequencies in the SEM," Scanning Electron Microscopy.
IIT Research Institute, Chicago. 1976, 615-624, 1976.
2.18. H. Feuerbaum, "VLSI testing using the electron probe," Scanning Electron
Microscopy, vol. 1, pp. 285-296, 1979.
2.19. E. Menzel and R. Buchanan, "In-the-lens secondary electron analyser for
IC internal voltage measurements with electron beams," Electronics letters,
vol. 20, pp. 408-409, 1984.
2.20. S. Garth, J. Sackett, and D. Spicer, "An in-the-lens spectrometer for high
performance E-beam testing," Microelectronic Engineering, vol. 7, pp. 155-
161, 1987.
2.21. J. Frosien and E. Plies, "High performance electron optical column for
testing ICs with submicrometer design rules," Microelectronic Engineering,
vol. 7, pp. 163-172, 1987.
2.22. A. Dinnis, "Detectors for quantitative voltage contrast on submicron
devices," Microelectronic Engineering, vol. 7, pp. 139-146, 1987.
Chapter 2
44
2.23. R. Ludwing, "Scanning electron microscopy: physics of image formation
and microanalysis," ISBN 978-3-642-08372-31998.
2.24. L. Dubbeldam, "Electron spectrometers and voltage measurements," in
Electron Beam Testing Technology, ed: Springer, 1993, pp. 211-239.
2.25. K. Sim, J. Thong, and J. Phang, "Effect of shot noise and secondary
emission noise in scanning electron microscope images," Scanning, vol. 26,
pp. 36-40, 2004.
2.26. A. Gopinath, "Estimate of minimum measurable voltage in the SEM,"
Journal of Physics E: Scientific Instruments, vol. 10, p. 911, 1977.
2.27. L. Dubbeldam and P. Kruit, "Signal-to-noise ratio improvement in electron-
beam testing by using a dispersive analyzer," Scanning Microscopy, vol. 1,
pp. 1647-1650, 1987.
2.28. P. Kazemian, S. A. M. Mentink, C. Rodenburg, and C. J. Humphreys,
"Quantitative secondary electron energy filtering in a scanning electron
microscope and its applications," Ultramicroscopy, vol. 107, pp. 140-150,
2// 2007.
2.29. P. Kazemian, S. Mentink, C. Rodenburg, and C. Humphreys, "High
resolution quantitative two-dimensional dopant mapping using energy-
filtered secondary electron imaging," Journal of Applied Physics, vol. 100,
p. 054901, 2006.
2.30. L. Dubbeldam and P. Kruit, "First experimental results of an e-beam tester
with dispersive secondary electron energy analyzer," Microelectronic
Engineering, vol. 7, pp. 231-234, 1987.
Chapter 2
45
2.31. A. Khursheed, "Multi-channel vs. conventional retarding field
spectrometers for voltage contrast," Microelectronic Engineering, vol. 16,
pp. 43-50, 1992.
2.32. A. Khursheed and N. Karuppiah, "An add-on secondary electron energy
spectrometer for scanning electron microscopes," Review of Scientific
Instruments, vol. 72, pp. 1708-1714, 2001.
2.33. M. Kienle and E. Plies, "An off-axis multi-channel analyzer for secondary
electrons," Nuclear Instruments and Methods in Physics Research Section
A: Accelerators, Spectrometers, Detectors and Associated Equipment, vol.
519, pp. 325-330, 2004.
2.34. A. Khursheed, "Secondary Electron Spectrometers," in Scanning Electron
Microscope Optics and Spectrometers, ed, 2011, pp. 223-261.
2.35. E. Rau and V. Robinson, "An annular toroidal backscattered electron energy
analyser for use in scanning electron microscopy," Scanning, vol. 18, pp.
556-561, 1996.
2.36. A. Khursheed and H. Q. Hoang, "A second-order focusing electrostatic
toroidal electron spectrometer with 2π radian collection," Ultramicroscopy,
vol. 109, pp. 104-110, 12// 2008.
2.37. H. Z. Sar‐El, "Criterion for Comparing Analyzers," Review of Scientific
Instruments, vol. 41, pp. 561-564, 1970.
2.38. E. Benis and T. Zouros, "The hemispherical deflector analyser revisited: II.
Electron-optical properties," Journal of Electron Spectroscopy and Related
Phenomena, vol. 163, pp. 28-39, 2008.
Chapter 2
46
2.39. H. Hoang, M. Osterberg, and A. Khursheed, "A high signal-to-noise ratio
toroidal electron spectrometer for the SEM," Ultramicroscopy, vol. 111, pp.
1093-1100, 2011.
Chapter 3
47
Chapter 3 – Voltage and dopant concentration measurements of
semiconductors using a band–pass toroidal energy analyzer inside
a SEM
3.1 Introduction
In the previous chapter, a brief introduction to the second–order focusing toroidal
energy analyzer attachment for the SEM reported by Khursheed and Hoang was
presented [3.1]. This toroidal analyzer design is predicted to have an energy
resolution comparable to the well–known Cylindrical Mirror analyzer for the same
acceptance angle [3.2]. Fig. 3.1a below shows the original layout of the second–
order focusing toroidal energy analyzer prototype attachment.
Fig. 3.1 – Original layout of the second–order focusing toroidal analyzer prototype
attachment: (a) Experimental layout (b) Original specimen holder layout [3.3].
a b
Hemispherical caps
Specimen
Specimen Holder
Chapter 3
48
Fig. 3.2a shows the experimentally acquired SE spectrum from the toroidal analyzer
prototype reported by Hoang et al., while Fig. 3.2b shows a selected part of the
same spectrum [3.3]. Both the specimen and the inner cap were kept at ground
potential. It can be clearly seen that the noise is very small and the analyzer
inherently offers very high signal–to–noise ratio, allowing spectral shifts as small
as 12 mV or lower to be monitored.
Fig. 3.2 – Experimental SE spectrum reported by Hung et al. [3.3] (a) full range
and (b) selected range in which curve 2 (dotted line) is obtained by shifting curve
1 by 12 mV in order to demonstrate the noise limit.
Hoang et al. further demonstrated that when the specimen and the inner cap are
together biased at higher negative voltages with reference to the outer cap (at
ground potential), the SE signal no longer represents the SE energy spectrum but
becomes more symmetrical and has a more distinct sharp peak. In addition the
signal becomes higher and shifts to the right as the specimen bias is increased. The
b
a
Chapter 3
49
SE energy spectrum signal is transformed into a much more convenient form for
open–loop specimen voltage measurements than the original SE spectrum. The
effect of biasing the specimen and the inner cap at higher negative voltages with
reference to the outer cap (at ground potential) is shown in Fig. 3.3.
Fig. 3.3 – Experimental secondary electron output signals at different specimen
biasing voltages reported by Hoang et al. [3.3].
The improvement of the signal–to–noise ratio was also reported at negative
specimen/inner cap voltages. As seen in Fig. 3.4, when the bias voltage on the
specimen/inner cap changes from –10 to –10.1 V, there is a distinct shift in the peak
value. In this case, the spectrometer deflector voltage is restricted to a small range
(1 volt) around the peak signal value. The presence of shot noise on these signals
is relatively small, less than 4 mV, a few times better than the case of 0 V bias.
Chapter 3
50
a
b
These results demonstrate that the analyzer inherently has a high signal–to–noise
capability in detecting specimen voltage changes, well into the sub–mV region.
Fig. 3.4 – Experimental secondary electron signals showing improved signal–to–
noise when specimen/inner cap is biased at −10 and −10.1 V shown around the
peak value: (a) Deflection voltage range from 7 to 8 V and (b) deflection voltage
range from 7.1 to 7.16 V [3.3].
The signal–to–noise results reported by Hoang et al. and shown in Fig. 3.4 are
promising; they were obtained from large metal samples in a field free region where
the voltage of the specimen and the first analyzer electrode above it were fixed to
be at the same voltage. However, in practice, for a wider range of specimens, there
will be other effects that act to change the SE analyzer signals shown in Fig. 3.4.
The first effect arises from the fact that the applied voltage to the specimen may
need to be different to the one applied to the inner cap, and this will create fringe
Chapter 3
51
electric fields between the specimen and inner cap. This effect will be referred to
here as specimen fringe fields. Also, there are always local variations of specimen
surface potential, which in turn, create microfields above the specimen surface.
Local surface fields can arise due to a variety of reasons, such as variations in
semiconductor doping levels, a difference in voltage between neighboring
microstructures (IC tracks), beam induced contamination and specimen charging.
The experimental work presented in this chapter first starts by investigating how
the second–order toroidal energy analyzer signals change in the presence of
specimen fringe fields and local surface microfields.
3.2 The problem of specimen fringe fields and local surface microfields
A simple illustration of how specimen fringe fields can change the trajectory path
of an electron emitted from the specimen is shown by the simulation results shown
in Fig. 3.5. These simulations were carried out by use of the Lorentz 2EM software
[3.4]. In this simple example, 0.5 eV trajectories leave the specimen with a polar
emission angle of 45 degrees, and the specimen fringe fields are created by a
negative one voltage difference between the specimen and inner cap (– 11 V and –
10 V respectively). The form of the specimen fringe fields is indicated by the
simulated equipotential lines in Fig. 3.5, and their collective effect is to significantly
deflect the electron away from the straight–line path it would have taken if no fringe
fields were present. This effect changes the analyzer’s entrance optics. Electrons
no longer appear to come from a single point located on the specimen; their
apparent starting position is now energy dependent, and the output signal
characteristics are therefore considerably changed. Just how the analyzer output
Chapter 3
52
signals change in the presence of these kinds of accelerating fringe fields is
experimentally investigated in the next section.
Fig. 3.5 – Direct ray tracing of a 0.5 eV electron (polar launch angle 45 degrees)
with fringe fields above the specimen; the specimen is biased more negative with
reference to the inner cap. The dotted path shows the electron trajectory without
fringe fields (Specimen = – 10 V).
How surface microfields change trajectory paths of electrons that leave the
specimen can also be simply illustrated by simulation. Studies on effects of surface
microfields on voltage contrast have been undertaken by researchers over many
years [3.5 – 3.7]. Within this context, surface fields have been classified into Type
I and Type II depending on the nature of the potential variation that occurs around
the point being probed inside the SEM. Type I local field effects are created when
Specimen = – 11 V
Inner cap = – 10 V
Outer cap = – 0 V Electron trajectory – Initial
energy 0.5 eV, emission angle
45
Electron trajectory
(dotted) in absence of
fringe fields.
Specimen
fringe fields
5 mm
Chapter 3
53
the point being probed is more positive with respect to the neighborhood specimen
region, while Type II local surface fields effects are created when the point being
probed is more negative with respect to its surroundings. Fig. 3.6 shows simulated
electron trajectories through the potential distribution above a sample created by
Type I local fields. As a function of height, the potential above the central track
drops below 0 V before rising again. This drop creates a potential barrier for the
emitted electrons, preventing all electrons having initial energies below the barrier
to reach the analyzer entrance. Slightly higher energy electrons which marginally
surpass the potential barrier, escape and are deflected from their original path. In
Fig. 3.6, the simulated electron ray path trapped by the potential barrier has an
emission energy of 0.4 eV, while the ray which manages to escape has an emission
energy of 0.5 eV.
Fig. 3.7 shows a simple simulation example of Type II surface microfields caused
by neighboring IC tracks. It simulates how a 0.5 eV low energy secondary electron
emitted from the central track is deflected by a positive one volt change in the
neighboring track (located at a distance 50 µm away). From an analyzer optics point
of view, this surface field effect changes the apparent position of the source point,
making it energy dependent, and it therefore changes the shape of the analyzer
output signal.
Chapter 3
54
Fig. 3.6 – Simulated potential distribution and electron trajectories of 0.5 eV and
0.4 eV electrons (polar launch angle 45) in the presence of Type I surface
microfields. The dotted line shows the simulated trajectory of the electron without
surface fields.
Fig. 3.7 – Simulated potential distribution and electron trajectory of a 0.5 eV
electron (polar launch angle 45) in the presence of Type II surface microfields.
The dotted line shows the simulated trajectory of the electron without surface fields.
Electron trajectory in
absence of surface fields Trajectory of a
0.5 eV electron
Trajectory of a
0.4 eV electron
0 V –1 V –1 V
Type I surface field
equipotentials
50 µm
100 µm
0 V 0 V +1 V
Electron trajectory in absence
of surface fields
Type II surface field
equipotentials
Deflected 0.5 eV electron
trajectory
50 µm
100 µm
Chapter 3
55
Historically, surface fields caused large changes to occur in the output signals from
the secondary electron energy analyzers used for Electron Beam Testing of ICs.,
and in practice, greatly limited the accuracy to which quantitative voltage contrast
measurements could be made [3.8, 3.9]. In the present context of performing
material analysis using the second–order focusing toroidal analyzer in the SEM,
large changes in the output signal created by local surface fields and fringe fields
will also occur, and just how much it compromises on the performance of the
analyzer needs to be investigated in the next section.
3.3 Experimental Results
In the previous section, preliminary results obtained from the initial prototype of
the second–order focusing toroidal energy analyzer were presented for large bulk
metal specimens. In this section, experimental results are presented for
semiconductor samples in the presence of fringe/surface fields.
3.3.1 The experimental setup
Fig. 3.8a shows the layout of the present second–order focusing toroidal analyzer
SEM attachment. This is the same setup reported by Hoang and Khursheed [3.3]
except that the specimen holder is now modified from the hemispherical
arrangement reported earlier, to a concentric conical structure in which the
specimen and inner/outer caps can be independently biased. The redesign of the
sample holder also allows for bigger sample sizes to be examined. A schematic of
the modified specimen holder is shown in Fig. 3.8b. It must, however, be clarified
here that the redesign of the specimen holder into a concentric conical structure
does not change the fundamental electron optics of the analyzer. The specimen
Chapter 3
56
holder consists of the specimen surrounded by two concentric conical caps. The
voltage bias to the specimen, the inner conical cap and the outer conical cap are
denoted as VS, VC1, and VC2 respectively.
Fig. 3.8 – The second–order toroidal energy analyzer SEM attachment: (a)
Experimental layout in the SEM chamber (b) Schematic of the modified specimen
holder indicating the bias voltages applied to the various components of the holder.
b
VC2
VC1
Vs
52.5
mm
13 mm
a
b
Chapter 3
57
For all the results discussed here, VC2 was set to 0 V. The scattered electron energy
distribution is obtained by ramping the analyzer deflection electrode (VDEF) and
monitoring the photomultiplier (PMT) output. The sample and conical caps are
biased to negative voltages in order to further increase the SE signal sensitivity to
specimen voltage changes [3.3].
All experiments in this chapter were carried out inside a Philips ESEM XL30 FEG
SEM where a 4 kV, 75 pA primary beam was focussed on to a spot on the specimen.
A personal computer (PC) was used to control VDEF and the PMT output was
obtained through a National Instruments Data Acquisition (NI–DAQ) board [3.10].
The deflection voltage was ramped in steps of 200 mV and each point had a primary
beam dwell time of approximately 150 ms. The scintillator voltage of the PMT,
VSC, was biased to 5 kV. The SE signal was collected from regions which were
exposed to the primary beam for the first time in order to minimize the effects of
contamination.
3.3.2 Experimental analyzer SE signals on a doped silicon specimen in
presence of specimen fringe fields
Experimental analyzer SE signals from an n–type doped Silicon wafer (ND ~ 1018
cm–3) were obtained, where the specimen voltage (VS) was varied from – 10 V to –
13 V and inner cap voltage (VC1) was maintained at – 10 V; outer cap voltage (VC2)
was held at 0 V. Fig. 3.9a shows that as the sample becomes more negative with
respect to the inner cap voltage, the SE signal decreases in amplitude as well as
shifts towards the right. This happens because, as the specimen becomes more
negative, low energy secondaries are deflected upwards by accelerating fringe
Chapter 3
58
fields, causing some electrons to be pulled away from the analyzer entrance slit, as
illustrated in Fig. 3.5, and some (smaller angle) electrons, to be pulled into the
analyzer. However, the new electrons pulled into the analyzer do not compensate
for the electrons that are pulled away from it. This is because these electrons go
into it at steeper entrance angles than the previous ones (absence of fringe fields),
and they as a result, strike the analyzer deflector plates and do not contribute to the
SE signal output. The shift of the output signal to the right is caused by the fact that
the kinetic energies of all SEs increase with negative specimen voltage biasing.
These analyzer signals are very different to the case where VS = VC1 (field free
region above the specimen), as shown in Fig. 3.9b, where the SE signal shifts to the
right but grows significantly in height, similar to the results reported previously by
Hoang and Khursheed [3.3].
a
Chapter 3
59
Fig. 3.9 – Experimental SE signals obtained from an n–type semiconductor sample:
(a) Specimen biasing from – 10 to –13 volts with VC1 = –10 V, VC2 = 0 V. Inset
shows the biasing condition of the sample holder (b) Experimental SE signals at
different specimen biasing voltages where VS = VC1 (c) A plot of PMT signal
expectation value () for specimen potential change (ΔVs) from 0 to 3 V in presence
of specimen fringe fields.
VS = – 10 V
VC1 = VS
VC2 = 0 V
b
c
Chapter 3
60
For the case of the specimen being more positive than the first cap, VS = – 9 V, it is
important to note that apart from the expected shift to the left, the signal also falls
in height. This can be explained simply in terms of the creation of a retarding
electric field, where low energy electrons (<1 eV) are now returned back to the
specimen, since the specimen is now more positive than VC1 (by 1 volt).
To better quantify the specimen voltage change in presence of fringe fields above
the specimen (Fig. 3.9a), the output signal mean µ is plot against the change in
specimen voltage, ΔVS, given by the expectation value E(V):
𝜇 = 𝐸(𝑉) = ∑ 𝑃𝑗𝑉𝑗
𝑁
𝑗=1
where the index j runs from 1 to the number of points in the output signal, N; V
refers to the deflection voltage; Pj refers to the probability of each point in the
output signal, obtained from the output height normalized to the area under the
output curve. The expectation function is a convenient way of monitoring small
changes in the peak value, since the peak position shifts significantly as the
specimen voltage changes. Also the expectation value changes with variations in
the shape of the SE signal. Therefore two signals with the same peak amplitude
occurring at the same value of deflection voltage but with variations in their shapes
will yield different values of expectation value.
The relationship between and ΔVS found from the experimental results of Fig.
3.9a is plot in the graph shown in Fig. 3.9c. It shows that they vary in an
approximately linear way, demonstrating that it is possible to quantify surface
Chapter 3
61
specimen voltage changes in non–metallic specimens with the toroidal analyzer
attachment even in the presence of fringe fields above the sample, where the
specimen voltage and inner cap voltage differ. Despite there being a considerable
loss in parts of the SE analyzer signal, the overall shift due to specimen voltage
change can be reliably obtained by calculating the SE analyzer signal expectation
value. After suitable calibration, the SE analyzer signal can therefore be used to
quantify specimen voltage changes.
3.3.3 Experimental analyzer SE signals in presence of surface microfields
In the previous section, SE signals and their variations in the presence of fringe
fields above the sample were studied. However in most cases, the voltage variations
that occur on the sample are typically localized to the micrometer/nanometer scale,
in the form of surface microfields, as discussed earlier in this chapter. A thin piece
of copper enameled winding wire of diameter 200 microns was mounted on top of
a metal base, as shown in Fig. 3.10a. The outer enamel paint surrounding the wire
makes it possible to bias the wire and the metal base to a different voltage. The
enamel covering is carefully scraped off from the top of the wire using a surgical
blade in order to prevent the wire from charging when illuminated by the primary
electron beam. Experimental SE signals were obtained from the center of the wire
for various values of wire voltage, denoted as VS, while VC1 and VC2 are maintained
at –10 V and 0 V respectively.
Chapter 3
62
PM
T O
utp
ut
(a.u
)
Deflection Voltage (volts)
VS = – 10 V
– 11 V
– 12 V
– 13 V
– 14 V
VC1 = – 10 V
VC2 = 0 V
b
PE
VC1 = – 10 V
VC2 = 0 V
Enamelled
Copper wire
VS = VC1 – ΔVS
a
Metal
Specimen (VC1)
PE
Exposed copper
wire (φ 200 µm)
Enamel
insulation
Chapter 3
63
Fig. 3.10 – Experimental SE signals obtained from a copper wire in presence of
surface fields: (a) Specimen arrangement to generate microfields above the point
of probing (b) SE signals obtained from the specimen for different biasing of the
copper wire (c) A plot of PMT signal expectation value () for copper wire potential
change (ΔVs) from 0 to 4 V in presence of surface fields above the point of probing.
The experimental results shown in Fig. 3.10b indicate that as the wire becomes
more negative with respect to the metal base/inner cap voltage, changing from –10
to –14 V (while VC1 and the metal base remain at –10 V), the SE signal shifts to the
right due to an increase in the kinetic energy of the scattered electrons that pass
through the analyzer, however, the amplitude of the SE signal goes through a
maximum value; it first increases, after which (< –12 V), it decreases. The situation
seems to be a combination of the field–free region case, where the amplitude rises
as the specimen voltage becomes more negative, and the specimen fringe field case,
where the SE analyzer signal amplitude goes progressively down. These results
seem to suggest for VS > –12 V, the effect of Type II surface microfields is
c
ΔVs (Volts)
Ex
pec
tati
on
va
lue (
μ)
of
ou
tpu
t si
gn
al
(vo
lts)
Chapter 3
64
compensated by the acceleration fringe fields created between the specimen and
inner cap, however, for VS < –12 V, the specimen fringe fields dominate. Despite
the variations in analyzer signal amplitude, there is a consistent shift in the peak
position as the specimen voltage is biased more negatively. This is demonstrated
by Fig. 3.10, which plots the analyzer signal expectation value with change in
specimen potential, indicating that it increases monotonically. These experimental
results therefore show that the second–order focusing toroidal energy analyzer can
be used to quantify specimen voltage changes even in the presence of surface fields.
3.3.4 Experimental analyzer SE signals along a semiconductor sample with
a potential gradient
Figs. 3.9 and 3.10 show that the second–order focusing toroidal analyzer can
reliably track changes in specimen voltages at a single point on the specimen. The
following experiment sets out to detect voltage changes across a semiconductor
surface. An n–type doped Silicon sample (ND ~ 1018 cm–3) was mounted on top of
a button cell and the two ends of the specimen were biased in order to create a linear
potential gradient along its surface as shown in Fig. 3.11a. The two electrodes of
the button cell are connected to the two far ends of the silicon wafer using metal
wires. The distance between the two ends of the specimen is 6 mm, across which a
potential drop of 1.5 V is applied. The entire setup is biased at – 10 V, as a result
of which the two ends of the specimen are effectively at – 8.5 V and – 10 V
respectively. Experimental SE analyzer signals were obtained from the specimen
at consecutive points on a straight line along the x direction starting from the centre
of the specimen, and are shown in Fig. 3.11b. The distance between each point of
Chapter 3
65
measurement on the specimen was approximately 120 microns. Assuming that the
specimen voltage changes linearly across its surface, there is a voltage change of
30 mV between each measurement point.
Fig. 3.11b shows that as the potential along the surface of the sample becomes more
negative, there is a distinct shift in the SE analyzer signal; the amplitude increases
and shifts to the right. Again, the specimen voltage change ΔVS can be tracked by
calculating and plotting the output signal expectation value , which in this case
changes with distance x, as shown in Fig. 3.11c. The plot of against the change
in potential ΔVS (proportional to the distance moved along the x direction) is, as
expected, almost linear; hence a small change in the voltage along the surface of
the sample can be measured (via the electron beam) by calculating the expectation
of the SE analyzer signal.
a
Chapter 3
66
Fig. 3.11 – Experimental SE signals obtained by setting up a potential gradient
along a semiconductor sample: (a) Specimen arrangement using a button cell (b)
SE signals obtained from the specimen along the x direction (c) A plot of
expectation value µ of the SE signal against the change in potential ΔVS along
distance x.
b
c
Chapter 3
67
The close fit to the theoretical straight line in Fig. 3.11c confirms that the method
of monitoring the expectation value of the SE analyzer signal in order to measure
small voltage changes on the semiconductor specimen seems to be a good one. An
estimate of the voltage measurement accuracy can be obtained by calculating the
standard deviation between the dotted and the solid line in Fig. 3.11c, translating it
into an uncertainty in voltage measurement by using the /ΔVS gradient. For the
experimental conditions and results shown in Fig. 3.11c, the voltage accuracy is
calculated to be 5.1 mV, corresponding to an average signal–to–noise ratio of 7.3.
These experimental results therefore demonstrate that surface voltage variations on
a semiconductor specimen can be measured to millivolt accuracy by the second–
order focusing toroidal analyzer attachment.
3.3.5 Experimental SE analyzer dopant contrast signals from abrupt
semiconductor heterojunctions
Apart from voltage measurement on semiconductors, an important application of
energy analyzers for the SEM is to obtain quantitative dopant concentration
measurements. The following experiment sets out to do this using Zinc Oxide
(ZnO) on p–doped Si substrate. ZnO is a direct wide band gap (3.37 eV)
semiconductor [3.11] and has a wide variety of applications in molecular and nano–
scale electronics, especially for opto–electronics [3.12, 3.13]. For the purpose of
this study, sol–gel synthesized n–ZnO/Si (n–p) thin film heterojunctions were
fabricated using standard processes [3.14, 3.15]. Four differently doped p–type
silicon samples were taken, on which ZnO thin films were grown to form abrupt p–
n junctions. They were cleaved and then examined in the SEM, as shown in Fig.
Chapter 3
68
3.12a. In this way abrupt p–n junctions with different p–doped concentrations were
fabricated under controlled conditions, providing convenient test samples for
quantitative dopant concentration measurements. At thermal equilibrium, the built–
in potential of each p–n junction depends on the doping levels on the p and the n
side. However in this case, the n–ZnO doping level (carrier concentration 2.2 1018
cm–3) was approximately constant for all the samples, as all the ZnO films were
grown using an identical process and under near–identical conditions. Therefore
variations of the built–in potential between the p–n junction samples are dependent
only on the doping level difference between the p–type silicon substrates. The p–
doping levels for the samples were: A – 5.18 1014 cm–3, B – 1.53 1015 cm–3, C
– 2.08 1016 cm–3 and D – 3.83 1018 cm–3, for which the theoretical built–in
potential values were calculated using first principles to be 0.52 V, 0.56 V, 0.62 V
and 0.75 V respectively. Experimental analyzer SE signals were collected from the
p–side and the n–side of each heterojunction sample (for VS = VC1 = – 10 V, VC2 =
0 V), from points far away from the depletion region, and are shown in Fig. 3.12b.
The expectation value was calculated for each signal. The difference Δ between
the two values of (from the p and the n side) is plot as a function of the p–side
dopant concentration (NA) in Fig. 3.12c. The experimental standard deviation (σ) of
Δ was obtained by repeating these measurements; each signal shown in Fig. 3.12b
was acquired 10 times.
Chapter 3
69
a
b
Chapter 3
70
Fig. 3.12 – SE analyzer signal contrast from a n–ZnO / p–Si heterojunction: (a)
Schematic representation of the fabricated thin–film ZnO on Silicon substrate (b)
Experimental SE signals obtained from the p–side and the n–side of each
heterojunction sample for samples A, B, C and D (c) Plot of difference in
expectation value Δ of the SE analyzer signal obtained from the p and n sides of
the Si/ZnO heterojunction against the log of doping concentration of the p–type
silicon substrate. The solid bars at each point on the graph represent standard
deviation of Δ taken over 10 SE signals.
The shift in the SE energy spectrum which arises from p and n doped sides of p–n
junctions has been previously explained in terms of various mechanisms, which
include a band–bending potential at the surface [3.16] and patch fields occurring
due to the built–in potential across the junction [3.17]. Kazemian et al. [3.18, 3.19]
proposed an empirical model wherein the shift in the SE analyzer signal comes
from a linear combination of these two mechanisms.
c
Chapter 3
71
Therefore parameter Δ is expected to be closely related to the p–n junction built–
in potential. In Fig. 3.12b, the theoretical value of the built–in potential is plot
together with Δ, indicating that Δ maintains a small offset below the built–in
potential. Taking Kazemian et al.’s interpretation, this small difference may be
caused by surface band bending. In the present context, the results shown in Fig.
3.12c demonstrate how quantitative dopant concentration measurements of a p–n
junction sample can be made with the second–order focusing toroidal analyzer
attachment in the SEM. Using interpolation on the curve shown in Fig. 3.12c any
subsequent measurement of Δ on a similar p–n junction ZnO/Si thin film sample
will provide an accurate estimate of NA in its substrate. The accuracy to which this
can be achieved is determined primarily by the uncertainty in Δ caused by noise,
which in this case, is indicated by the length of the error bars in Fig. 3.12c. These
results are given explicitly in Table 3.1. They indicate that the experimental signal–
to–noise varies roughly from 31 to 63, for a shift in expectation value (Δ) ranging
from 0.43 to 0.65 volts, where the noise varies between 9 to 18 mV.
Table 3.1 – Dopant concentration measurement results for the Si/ZnO p–n
heterojunction for different p–doped samples.
Chapter 3
72
These results therefore demonstrate that the high signal–to–noise characteristics
predicted for the second–order focusing toroidal analyzer attachment for dopant
concentration measurements can be obtained in the presence of surface fields, on
more realistic samples than previously examined (bulk metal ones). The millivolt
accuracy of the present results are over one order of magnitude better than previous
p–n junction dopant concentration measurements obtained by a retarding field
analyzer [3.18, 3.19]. These results confirm what was already known in the subject
of quantitative voltage contrast that band–pass analyzers have inherently better
signal–to–noise characteristics than retarding field analyzers, they have a lower
background noise level and their signals are more sensitive to specimen voltage
changes [3.20, 3.21].
3.4 Conclusions
Voltage and dopant concentration measurements on Silicon samples by a second–
order focusing toroidal electron energy analyzer operating in a SEM as an add–on
attachment have been presented. These results were obtained in the presence of
electric fields above the sample, originating from the surface voltage on the sample
differing from the first analyzer electrode by several volts. They demonstrate that
it is possible to obtain high signal–to–noise measurements from the second–order
focusing toroidal analyzer even in the presence of such fields above the specimen.
The accuracy of voltage measurements on the surface of silicon test samples
typically lay between 9 to 18 mV, corresponding to signal–to–noise ratios of 31 to
63. These experimental results therefore establish that the analyzer can be applied
to measure semiconductor surface voltage variations and can be used for 2D dopant
Chapter 3
73
profiling of p–n junction based devices like solar cells by monitoring the shift in
the SE signal obtained from point–to–point on the cross section of such devices.
Chapter 3
74
References
3.1. Khursheed, A. and H.Q. Hoang, A second–order focusing electrostatic
toroidal electron spectrometer with 2π radian collection. Ultramicroscopy, 2008.
109(1): p. 104–110.
3.2. Sar‐El, H.Z., Criterion for Comparing Analyzers. Review of Scientific
Instruments, 1970. 41(4): p. 561–564.
3.3. Hoang, H., M. Osterberg, and A. Khursheed, A high signal–to–noise ratio
toroidal electron spectrometer for the SEM. Ultramicroscopy, 2011. 111(8): p.
1093–1100.
3.4. LORENTZ – EM. 2011, Integrated Engineering Software Inc., Canada.
3.5. Nakamae, K., H. Fujioka, and K. Ura, Local field effects on voltage contrast
in the scanning electron microscope. Journal of Physics D: Applied Physics, 1981.
14(11): p. 1939.
3.6. Wager, W.E. and E.D. Wolf, The effects of local electric fields and
specimen geometry on voltage contrast in the scanning electron microscope.
Journal of Vacuum Science & Technology B, 1986. 4(1): p. 209–212.
3.7. Clauberg, R., Microfields in stroboscopic voltage measurements via
electron emission. II. Effects on electron dynamics. Journal of Applied Physics,
1987. 62(10): p. 4017–4023.
3.8. Khursheed, A. and A. Dinnis, A time–of–flight voltage contrast detector for
measurements on VLSI circuits. Measurement Science and Technology, 1990.
1(7): p. 581.
Chapter 3
75
3.9. Khursheed, A. and A. Dinnis, A theoretical comparison of the data‐
acquisition time characteristics of the time of flight voltage contrast detector with
retarding field detectors. Journal of Vacuum Science & Technology B, 1990. 8(6):
p. 1841–1847.
3.10. National Instruments LabVIEWTM 2010. National Instruments.
3.11. Klingshirn, C., The Luminescence of ZnO under High One‐and Two‐
Quantum Excitation. Physica status solidi (b), 1975. 71(2): p. 547–556.
3.12. Gudiksen, M.S., et al., Growth of nanowire superlattice structures for
nanoscale photonics and electronics. Nature, 2002. 415(6872): p. 617–620.
3.13. Huang, M.H., et al., Room–temperature ultraviolet nanowire nanolasers.
Science, 2001. 292(5523): p. 1897–1899.
3.14. Paul, G. and S. Sen, Sol–gel preparation, characterization and studies on
electrical and thermoelectrical properties of gallium doped zinc oxide films.
Materials letters, 2002. 57(3): p. 742–746.
3.15. Sarkar, S., et al., Rectifying properties of sol–gel synthesized Al: ZnO/Si
(N–n) thin film heterojunctions. Physica E: Low–dimensional Systems and
Nanostructures, 2012. 46: p. 1–5.
3.16. Perovic, D., et al., Field–emission SEM imaging of compositional and
doping layer semiconductor superlattices. Ultramicroscopy, 1995. 58(1): p. 104–
113.
3.17. Sealy, C.P., M.R. Castell, and P.R. Wilshaw, Mechanism for secondary
electron dopant contrast in the SEM. Journal of Electron Microscopy, 2000. 49(2):
p. 311–321.
Chapter 3
76
3.18. Kazemian, P., et al., High resolution quantitative two–dimensional dopant
mapping using energy–filtered secondary electron imaging. Journal of Applied
Physics, 2006. 100(5): p. 054901.
3.19. Kazemian, P., et al., Quantitative secondary electron energy filtering in a
scanning electron microscope and its applications. Ultramicroscopy, 2007. 107(2–
3): p. 140–150.
3.20. Khursheed, A., Scanning Electron Microscope Optics and Spectrometers.
2011: World Scientific.
3.21. Khursheed, A., Multi–channel vs. conventional retarding field
spectrometers for voltage contrast. Microelectronic Engineering, 1992. 16(1): p.
43–50.
Chapter 4
77
Chapter 4 – New contrast mechanisms and material
characterization by energy filtered secondary electron signals
inside the SEM
4.1 Introduction
In the previous chapter, experimental results were presented to demonstrate that the
second–order focusing toroidal energy analyzer [4.1] can provide high signal–to–
noise and dopant concentration measurements on semiconductor specimens, even
though the SE signals are greatly changed by fringe fields and surface fields. A
natural continuation to this work is to look for new contrast mechanisms inside the
SEM using the second–order focusing toroidal analyzer and apply these contrast
mechanisms to develop novel material characterization techniques. Analytical
information about the specimen can be obtained using electron analyzers such as
the second–order focusing toroidal analyzer by monitoring changes in the SE
analyzer signal obtained from the specimen and this can be done together with
capturing nanometer–level topographical images of the specimen, with the SEM’s
conventional SE detector. The secondary electron (SE) spectrum inside a Scanning
Electron Microscope (SEM) has been used for a number of applications like dopant
mapping [4.2, 4.3], monitoring specimen charging [4.4] and Electron Beam Testing
(EBT) of Integrated Circuits [4.5]. In this chapter, SE analyzer signals are presented
that point towards a new application of detecting trapped charge at the interface of
multi–functional oxide layers. This chapter also presents results that illustrate how
SE signals obtained using the second–order focusing toroidal analyzer vary in the
Chapter 4
78
presence of magnetic fields above the specimen surface, and how SE analyzer
signals also change due to spontaneous oxidation of a metal surface to metal oxide.
4.2 Probing and analyzing buried interfaces of multifunctional oxides
using a secondary electron energy analyzer.
4.2.1 Introduction
Recently, the discovery of the formation of a two–dimensional electron gas (2DEG)
at the interface of complex oxides such as LaAlO3/SrTiO3 (LAO/STO) reported by
Ohtomo and Hwang [4.6], has formed the basis of many new device concepts [4.7].
The experimental results presented here highlight the possibility of using SE energy
analyzers inside the SEM as a high resolution contactless way to detect and analyze
hidden interfaces between multifunctional oxides such as the presence of 2DEG at
the interface of LAO and STO.
In the following experiments, a second–order focusing toroidal analyzer SEM
attachment, previously designed by Hoang and Khursheed [4.8] for voltage and
dopant concentration measurements is used to detect the 2DEG at the interface of
a LAO/STO sample with high SE analyzer signal contrast. Conditions for obtaining
significant SE analyzer signal contrast in the SEM between conducting and
insulating LAO/STO interface samples were also investigated and the primary
beam energy was found to be an important factor affecting SE signal contrast.
Monte Carlo simulations of the beam/sample interaction were performed to explain
why the SE contrast mechanism is sensitive to primary beam energy and how it can
be optimized.
Chapter 4
79
4.2.2 Materials and methods
Fig. 4.1 shows the mounting of a LAO/STO sample inside the specimen holder
arrangement of the second–order focusing toroidal analyzer SEM attachment (refer
Chapter 3, Fig. 3.8). In the present arrangement, the specimen is mounted on top of
a metal base and placed inside a concentric conical arrangement of the first and
second analyzer electrodes (holder). The specimen and the first analyzer electrode,
denoted by VS and VC1, were biased to the same voltage ( –10 V) and the second
analyzer electrode (V2) was set to 0 V. It may be recalled that the analyzer is
designed to capture an angular spread of 8 with respect to the central entrance
angle of 45 in the polar direction. The input angular spread in the azimuthal
direction is 100.
Fig. 4.1 – Layout of the specimen holder arrangement of the second–order focusing
toroidal energy analyzer showing the mounting of the LAO/STO specimen.
The sample and the first analyzer electrode are biased to negative voltages in order
to increase the SE analyzer signal sensitivity to specimen voltage changes [4.8]. In
VC2 = 0 V
VC1 = – 10 V
LAO / STO
Specimen
PE
VS = – 10 V
Chapter 4
80
the previous chapter it was shown that the second–order focusing toroidal energy
analyzer is able to detect specimen voltage changes with high signal–to–noise
characteristics. The scattered electron energy distribution can be obtained by
ramping the analyzer deflection electrode (VDEF) and monitoring the
photomultiplier (PMT) output, where the deflection electrode voltage is directly
related to the SE energy.
All experiments were carried out inside a Philips ESEM XL30 FEG SEM whose
primary beam was focussed on to a spot on the surface of the specimen in order to
generate secondary electrons. A personal computer (PC) was used to control VDEF
and monitor the PMT output through a National Instruments Data Acquisition (NI–
DAQ) board [4.9]. The deflection voltage was ramped in steps of 200 mV and each
point had a primary beam dwell time of approximately 150 ms. The scintillator
voltage of the PMT, VSC, was biased to 5 kV. The SE analyzer signal was collected
from regions which were exposed to the primary beam for the first time in order to
minimize the effects of contamination. At primary beam energy of 4 keV, the beam
current was measured to be 75 pA.
Fig. 4.2a is a schematic representation of the LAO/STO heterointerface specimen.
The specimen was fabricated by depositing a thin film of crystalline LAO,
equivalent to a thickness of 18 unit cells (U.C) of LAO, on crystalline STO
substrate.
Chapter 4
81
Fig. 4.2 – Schematic representation of the specimen: (a) Representation of the
2DEG formed at the interface of crystalline LAO and crystalline STO substrate (b)
Conducting and insulating interface regions side by side on the same sample (the
interface was made insulating by proton irradiation).
The deposition of LAO was done by using a process called pulsed laser deposition,
which involves ablating a LAO target onto TiO2 terminated STO (100) substrates.
Samples were prepared in a range of oxygen pressures (PO2) of 1 – 5 × 10–3 Torr
Chapter 4
82
at 750 °C. A nanometer thick 2DEG is formed at the buried LAO/STO
heterointerface. One half of the surface area of this sample was exposed to high
energy proton irradiation, while keeping the other half shielded from the proton
beam. A collimated ion beam was used to raster–scan under normal incidence to
expose a specific region to high energy protons. The regions are electrically
isolated, which was confirmed by the absence of conductivity between the regions
when subjected to direct conductivity measurements using Al wire bonding to the
interface. The proton irradiation causes the embedded interface to become
insulating while the other half, not exposed to the high energy proton irradiation, is
unaffected. The conductivity of the interface is controlled by adjusting the fluence
of the proton beam. However for the irradiated samples investigated in this work,
the interface was made completely insulating. The fabrication process of this
specimen has been described in detail by Mathew et al. [4.10].
4.2.3 Results and Discussion
SE analyzer signals were obtained separately from the conducting LAO/STO
heterointerface specimen and an uncoated STO substrate, and are shown in Fig. 4.3.
As seen in the plot, the SE analyzer signal obtained from the STO substrate contains
a single peak, as expected. However the nature of the SE analyzer signal changes
completely when taken from the surface directly above the conducting LAO/STO
heterointerface. In addition to the regular SE peak, which is similar in amplitude
and position to the only SE analyzer signal peak obtained from the uncoated STO
substrate, a second high amplitude SE peak also occurs. This second peak occurs
at a higher deflection voltage (SE energy) in comparison to the first SE peak and
Chapter 4
83
has amplitude, almost twice as high as that of the first SE peak, indicating the
presence of a high SE yield region.
Fig. 4.3 – Experimental SE analyzer signals obtained from an uncoated STO
substrate (shown in dotted line) and from the LAO/STO heterointerface with
conducting interface (shown in solid line). A primary beam acceleration voltage of
3 kV was used.
SE analyzer signals were then obtained from surfaces directly above both the
conducting and insulating LAO/STO parts of the heterointerface, and are shown in
Fig. 4.4.
Chapter 4
84
Fig. 4.4 – Experimental SE analyzer signals obtained from the LAO/STO
heterointerface with conducting interface (shown in solid line) and the insulating
LAO/STO heterointerface (shown in dotted line). A primary beam acceleration
voltage of 3 kV was used.
The process of transforming the conducting LAO/STO heterointerface into an
insulating one has been reported in detail by Mathew et al. [4.10]. Fig. 4.4 shows
that the SE analyzer signals obtained from the surface exposed to the proton
irradiation (insulating LAO/STO interface) are significantly different from the SE
analyzer signals obtained from the surface unexposed to the proton irradiation
(conducting interface) in both amplitude and position of the second peak. While the
first peak in the SE analyzer signal for both cases occurs at the same position,
relatively unchanged, the amplitude of the second peak in the SE analyzer signal
obtained from the surface above the insulating interface goes down by almost a
Chapter 4
85
factor of 2.5 in comparison to its amplitude for the conducting LAO/STO interface,
and its position shifts by around 10 volts. These results primarily show that when
the LAO/STO interface changes from conducting to insulating, significant contrast
occurs in the SE energy analyzer signals. Unlike the normal SE signal, which has a
single peak, there is an additional high amplitude SE analyzer signal peak obtained
when the interface is conducting. This indicates that there is more trapped charge
at the interface. These trapped charges are emitted easily due to interactions with
the primary beam and back scattered electrons (BSEs) and come out at energies
higher than the usual SE electrons, which shows up as a high amplitude second
peak in the SE signal. This is in sharp contrast to the SE signal obtained from the
specimen when the interface is insulating, the second SE peak is much reduced in
height and now shifts to the right, indicating that the electrons are more difficult to
remove after the interface becomes insulating, since higher deflection voltages
capture higher SE energies. Secondary electrons can be generated either by direct
interaction of the primary beam with the specimen volume or the interaction of
BSEs with the specimen. In this case, the specimen is a complex arrangement of a
thin LAO layer on bulk STO substrate and therefore to understand the exact
contribution to SEs from each material and the effect of BSEs on the SE signal is a
matter of future investigation. However the excellent contrast of SE signals
obtained from the LAO/STO sample when the interface is conducting and
insulating confirms the existence of the 2DEG at the LAO/STO interface, as
reported earlier [4.6, 4.11–4.13] and as verified by Mathew et al. by the use of
contact measurements [4.10]. The high contrast between the second peak of the SE
Chapter 4
86
analyzer signal obtained from the conducting and insulting parts of the LAO/STO
interface can only be attributed to the change in the electrical properties of the
interface, because the physical structure of the specimen (film thickness, material
density) remains unaltered by the high energy proton irradiation. These results
suggest that the SEM together with an energy analyzer can provide a contactless
way of monitoring charge present at the interface of thin film oxides.
Fig. 4.5 – Experimental secondary electron signals obtained from LAO/STO
hetero–interface at various primary beam energies. The signals are obtained at
primary beam electron energies of 2 keV, 3 keV, 4 keV and 5 keV (shown in green,
brown, pink and blue respectively).
The primary beam energy was found to be an important parameter in maximizing
contrast in the SE analyzer signals on the LAO/STO heterointerface sample. Fig.
4.5 shows SE analyzer signals obtained from the surface area of the specimen above
Chapter 4
87
the conducting LAO/STO interface at various primary beam energies. The SE
analyzer signal has a high dependency on the primary beam energy; the distinct
high amplitude second peak occurs only at a particular value of primary beam
energy, 3 keV in this case. At the primary beam energies of 2 and 5 keV, the second
peak is not present. At 4 keV, the second peak does occur in the SE analyzer signal
but the amplitude is reduced by around 75% of the 3 keV signal and the peak shifts
to the right.
Monte Carlo simulation of primary beam electron trajectories were carried out
using a software package called Casino [4.14] to help understand the dependence
of SE analyzer signal with primary beam energy. The simulation results, shown in
Fig. 4.6a, indicate that the interaction volume of the primary beams increases as the
primary beam energy increases, as expected.
The percentage energy loss of primary beam electrons is plot against depth from
the specimen surface and is shown in Fig. 4.6b. Using the percentage energy loss
of the primary beam electrons as a measure of the degree of interaction of the
primary beam with the sample, only the 10–50% 3 keV energy loss curve crosses
the interface (depth 8 nm form the surface), the 2 keV curve lies below it, while the
5 keV curve lies well above it. These simulations indicate that a primary beam of 3
keV interacts the most with the interface, confirming the results obtained by
experiment (shown in Fig. 4.5).
Chapter 4
88
a
2keV
4keV
3keV
5keV
b
Dep
th a
lon
g s
pec
imen
(n
m)
LAO/STO Interface
LAO/STO Interface
% Energy
Loss of PE
Chapter 4
89
Fig. 4.6 – Monte Carlo simulation of the electron trajectories: (a) Primary
beam/specimen interaction indicating the interaction volume of the electrons (b)
Energy contour of the percentage energy loss of primary beam electrons along the
depth of the specimen (c) A graphical plot of percentage energy loss of primary
beam electrons against the depth from the surface of the specimen. The red dotted
line indicates the LAO/STO interface at a depth of 8nm from the surface.
In the case of the 2 keV beam, the majority of the primary beam interaction occurs
within the top LAO layer, for higher energies of the primary beam (4 keV and 5
keV), most of the interaction of the beam with the sample occurs within the bulk
substrate. In general, the optimal value of primary beam energy is dependent on the
thickness of the deposited thin film (LAO in this case) and is proportional to the
film thickness. The results shown in Fig. 4.6 demonstrate that Monte Carlo
simulation of the primary beam/sample interaction is able to provide a good guide
on which beam energy should be used to maximize interaction with the interface.
c
Chapter 4
90
4.3 SE signal contrast in presence of magnetic fields above the specimen
In the previous section, experimental SE analyzer signals from multifunctional
oxide interfaces were obtained, and it was demonstrated how they could be used to
determine the conductivity of interfaces between thin–film oxides. In this section,
experimental SE analyzer signals in presence of magnetic fields above the specimen
are studied. A current carrying air–core solenoid coil of radius 13 mm, made up of
20 turns of enameled copper wire (thickness 0.2 mm) was placed under a stainless
steel disc (thickness 1 mm and radius 13 mm) as shown in Fig. 4.7. A 200 nm layer
of gold was thermally evaporated on top of the specimen to enhance the SE yield
and to avoid surface oxidation effects. The specimen and the first analyzer
electrode, denoted by VS and VC1, were biased to the same voltage ( – 10 V) and the
second analyzer electrode (VC2) was set to 0 V. The experiments were carried out
inside a Philips ESEM XL30 FEG SEM where a 5 kV, 100 pA primary beam was
focussed on to a spot on the surface of the specimen in order to generate secondary
electrons. As before, a personal computer (PC) was used to control VDEF and the
PMT output was obtained through a National Instruments Data Acquisition (NI–
DAQ) board [4.9]. The deflection voltage was ramped in steps of 250 mV and each
point had a primary beam dwell time of approximately 150 ms. The scintillator
voltage of the PMT, VSC, was biased to 5 kV.
Chapter 4
91
Fig. 4.7 – Schematic representation of the specimen holder with a current carrying
solenoid placed under the specimen to produce magnetic field. (Cross–section view
of the specimen holder is shown here, while the solenoid is shown completely).
Let the current flowing through the solenoid be positive if the magnetic field created
by the current points in the positive z–axis direction (see Fig. 4.7) at the center of
the solenoid, and conversely, let it be negative if this magnetic field points in the
negative z direction. Fig. 4.8a presents experimental SE analyzer signals in
presence of a B field along the positive z axis at the center of the solenoid. As the
current (I) through the coil increases, the amplitude of the SE analyzer signal
increases and the SE analyzer signal peak shifts to the left indicating a net decrease
in the kinetic energy of SEs passing through the final slit of the analyzer. This is in
sharp contrast to the experimental SE analyzer signals shown in Fig. 4.8b where
PE VC2 = 0 V
VC1 = – 10 V
VS = – 10 V
Current carrying
enameled copper
wire coil (20 turns)
Silver coated
(500 nm) Metal
specimen
z
13 mm
Chapter 4
92
the direction of the B field is now reversed by reversing the direction of the current
flowing through the coil.
B
B
No Current
25 mA
50 mA
75 mA
VC2 = 0 V
VC1 = – 10 V
VS = – 10 V
Deflection Voltage (VD)
PM
T O
utp
ut
(a.u
)
b
Deflection Voltage (VD)
PM
T O
utp
ut
(a.u
)
Current through
solenoid 75 mA
50 mA
25 mA
No current
B
B
VC2 = 0 V
VC1 = – 10 V
VS = – 10 V
a
Chapter 4
93
Fig. 4.8 – Experimental SE analyzer signals obtained from a metal specimen in
presence of magnetic field (B) created by current carrying solenoid under the
specimen: (a) SE analyzer signals obtained with B field along positive z direction
(b) SE analyzer signals obtained with B field along negative z direction (c) A plot
of SE analyzer signal expectation value (µ) against current flowing in the solenoid
creating the magnetic field (negative value of current indicates a current giving rise
to a B field along negative z–axis)
In this case, an increase in the field strength causes a decrease in the SE analyzer
signal while the signal peak shifts to the right. To further quantify these variations,
the output signal mean µ (otherwise referred to as expectation value) of the SE
analyzer signal is plot against the current (I) flowing through the solenoid at the
time of measurement. To reiterate, a negative current value simply means that the
direction of the B field setup by the current I at the centre of the solenoid points in
along the negative z–axis. This plot is shown in Fig. 4.8c.
Although the plot is not completely linear, it shows a definite trend in expectation
value of SE analyzer signal with change in the magnitude and direction of the B
field created by the current carrying solenoid under the specimen. The SE analyzer
Current (I) through the solenoid (mA)
Exp
ecta
tio
n v
alu
e (µ
) of
ou
tpu
t si
gn
al
(vo
lts)
c
Chapter 4
94
signal contrast observed here illustrates that SE analyzer signals obtained from the
second–order focusing toroidal analyzer are sensitive to magnetic field variations
above the specimen and therefore the technique can, in principle, be applied to the
characterization of magnetic samples, like mapping magnetic domain variations on
the surface of magnetic specimens. Further investigation is required to understand
how this magnetic contrast effect changes when the magnetic field is localised into
micro–magnetic domains on the specimen surface. However, these preliminary
experimental results point towards the basis of a new possible application for the
toroidal second–order focusing analyzer. Also, direct ray tracing simulation needs
to be carried out to better understand how specimen magnetic fields change the
trajectory paths of SEs, especially low energy SEs.
4.4 SE analyzer signal contrast due to surface oxidation of a thin film metal
layer specimen
Spontaneous surface oxidation of thin film metal layers is an important effect in the
field of microelectronics. As feature sizes of devices continue to shrink, even
minimal surface oxidation of thin films has a profound effect on device
characteristics especially for a metal like Aluminum, which has a tendency of
readily oxidizing in air to form a passivation layer. This effect has been studied
extensively in the past [4.15–4.19] and continues to be an area of investigation for
present day researchers [4.20–4.23]. In this section, experimental results are
presented which demonstrate that the second–order focusing toroidal analyzer is
capable of detecting changes in the nature of thin film aluminum layers due to
surface oxidation in air.
Chapter 4
95
A 500 nm thick Al layer was thermally evaporated on an undoped silicon substrate
(thickness 550 microns) which was mounted on a metal base and placed inside the
specimen holder arrangement of the toroidal analyzer as shown in Fig. 4. 9. The
experimental conditions and the data acquisition process were kept the same as in
the earlier experiment reported in section 4.3 of this chapter. The spontaneous
oxidation of the aluminum thin film was allowed to take place at room temperature
(22 C) and at ambient atmospheric partial pressures of oxygen.
Fig. 4.9 – Mounting of the Al coated silicon specimen inside the specimen holder
of the analyzer.
The time of exposure (to air) in our context is defined as the time from which the
SEM chamber was vented and the specimen was left in air to the time the specimen
was placed back inside the SEM chamber and the pumping of the chamber was
initiated. The beam conditions and the working distance were kept unchanged for
the entire duration of the experiment.
VS = VC1 = –10 V
VC2 = 0 V
Silicon substrate coated
with Al (500 nm)
PE
Chapter 4
96
Fig. 4.10a shows the experimentally obtained SE analyzer signals for different
times of exposure and they indicate that substantial contrast between signals
obtained between successive exposure times. The SE analyzer signal peak initially
tends to shift to the left and decreases in amplitude indicating a decrease in work
function and a decrease in the SE yield of the specimen. However longer exposure
to air causes the SE analyzer signal peak to shift to the right while the amplitude
increases initially but continues to decrease thereon with exposure time.
Fig. 10b shows a plot of the SE analyzer signal expectation value (µ) against the
exposure time of the specimen in air and as observed in Fig. 4.10a, the expectation
value initially decreases and then increases consistently with subsequent exposure
in air.
Researchers generally agree that oxygen is first chemisorbed on the surface of Al
and then gradually transformed into oxide, but the specifics of the process, as well
as the adsorption sites of the chemisorbed oxygen atoms, are still being debated
upon [4.15–4.19]. In the present context, it is sufficient to note that SE analyzer
signals obtained using the second–order focusing toroidal energy analyzer, show
significant contrast when acquired from Al thin films exposed to air for different
lengths of time, and that these variations can be quantified by simply monitoring
parameters like the signal expectation value. These preliminary results point
towards another possible application of the toroidal second–order focusing analyzer
in the SEM, that of quantifying the degree of oxidation that occurs on metal
surfaces.
Chapter 4
97
Fig. 4.10 – (a) Experimental SE analyzer signals obtained from a thin film Al layer
under varying time of exposure to air (b) A plot of the SE analyzer signal
expectation value against the exposure time in air of the Al thin film.
Deflection Voltage (VD)
PM
T O
utp
ut
(a.u
)
Exposure time = 0 min
10 mins
15 mins
30 mins
45 mins
a
Exp
ecta
tion
valu
e (µ
) of
ou
tpu
t si
gn
al
(volt
s)
Exposure time (mins)
b
Chapter 4
98
4.5 Conclusions
This chapter has presented new possible applications for energy analyzer
attachments inside the SEM. The first new application is the probing of interfaces
between insulating materials, where high contrast SE analyzer signals between
conducting and insulating LAO/STO heterointerface samples can be obtained. The
method may provide an attractive alternative to contact methods [4.10]. In
principle, it can be used to obtain high resolution spatial mapping of the interface
over a large area. This may be useful in situations where interface properties have
been patterned by lithography techniques. Future work is required to investigate the
role and contribution of back–scattered electrons (BSE) to the characteristic SE
analyzer signals obtained from the sample. Other new possible applications of
energy analyzers in the SEM are the probing of magnetic fields above the specimen
and the quantification of metal surface oxidation.
Chapter 4
99
References
4.1. Khursheed, A. and H.Q. Hoang, A second–order focusing electrostatic
toroidal electron spectrometer with 2π radian collection. Ultramicroscopy,
2008. 109(1): p. 104–110.
4.2. Kazemian, P., et al., High resolution quantitative two-dimensional dopant
mapping using energy-filtered secondary electron imaging. Journal of
Applied Physics, 2006. 100(5): p. 054901.
4.3. Gostev, A., et al., Updating of the toroidal electron spectrometer intended
for a scanning electron microscope and its new applications in diagnostics
of micro-and nanoelectronic structures. Technical Physics, 2013. 58(3): p.
447–454.
4.4. Mizuhara, Y., et al., Quantitative measurement of surface potential and
amount of charging on insulator surface under electron beam irradiation.
Journal of Applied Physics, 2002. 92(10): p. 6128–6133.
4.5. Thong, J.T., Electron beam testing technology. 1993: Plenum Publishing
Corporation.
4.6. Ohtomo, A. and H. Hwang, A high-mobility electron gas at the
LaAlO3/SrTiO3 heterointerface. Nature, 2004. 427(6973): p. 423–426.
4.7. Bogorin, D.F., et al., LaAlO3/SrTiO3-based device concepts. arXiv preprint
arXiv:1011.5290, 2010.
4.8. Hoang, H., M. Osterberg, and A. Khursheed, A high signal-to-noise ratio
toroidal electron spectrometer for the SEM. Ultramicroscopy, 2011. 111(8):
p. 1093–1100.
Chapter 4
100
4.9. National Instruments LabVIEWTM 2010. National Instruments.
4.10. Mathew, S., et al., Tuning the Interface Conductivity of LaAlO3/SrTiO3
Using Ion Beams: Implications for Patterning. ACS nano, 2013. 7(12): p.
10572–10581.
4.11. Kalabukhov, A., et al., Effect of oxygen vacancies in the SrTiO 3 substrate
on the electrical properties of the LaAlO 3∕ SrTiO 3 interface. Physical
Review B, 2007. 75(12): p. 121404.
4.12. Schoofs, F., et al., Carrier density modulation by structural distortions at
modified LaAlO3/SrTiO3 interfaces. Journal of Physics: Condensed
Matter, 2013. 25(17): p. 175005.
4.13. Siemons, W., et al., Origin of charge density at LaAlO 3 on SrTiO 3
heterointerfaces: Possibility of intrinsic doping. Physical review letters,
2007. 98(19): p. 196802.
4.14. Drouin, D., et al., CASINO V2. 42—A Fast and Easy‐to‐use Modeling Tool
for Scanning Electron Microscopy and Microanalysis Users. Scanning,
2007. 29(3): p. 92–101.
4.15. Batra, I.P. and L. Kleinman, Chemisorption of oxygen on aluminum
surfaces. Journal of electron spectroscopy and related phenomena, 1984.
33(3): p. 175–241.
4.16. Lauderback, L. and S. Larson, An AES and SIMS study of the effect of
temperature on the interaction of oxygen with Al (100). Surface Science,
1990. 233(3): p. 276–282.
Chapter 4
101
4.17. Michel, R., et al., Initial interaction of oxygen with aluminium single crystal
faces: a LEED, AES and work function study. Surface Science, 1980. 95(1):
p. 309–320.
4.18. McConville, C., et al., Synchrotron radiation core level photoemission
investigation of the initial stages of oxidation of Al (111). Surface Science,
1987. 188(1): p. 1–14.
4.19. Agarwala, V.K. and T. Fort Jr, Nature of the stable oxide layer formed on
an aluminum surface by work function measurements. Surface Science,
1976. 54(1): p. 60–70.
4.20. Li, J.–T., et al., The initial oxidation of poly-crystalline aluminum studied
with x-ray photoelectron spectroscopy. Journal of Physics D: Applied
Physics, 2014. 47(10): p. 105301.
4.21. Benka, O. and M. Steinbatz, Oxidation of aluminum studied by secondary
electron emission. Surface science, 2003. 525(1): p. 207–214.
4.22. Lanthony, C., et al., On the early stage of aluminum oxidation: An
extraction mechanism via oxygen cooperation. The Journal of chemical
physics, 2012. 137(9): p. 094707.
4.23. Kreiter, O., et al., Ion-induced oxidation of aluminum during reactive
magnetron sputtering. Journal of Applied Physics, 2013. 113(14): p.
143303.
Chapter 5
102
Chapter 5 – New secondary electron energy analyzer designs for
the SEM
5.1 Introduction
Recently, a Radial Mirror Analyzer (RMA), designed to be fitted as an attachment
inside the specimen chambers of scanning electron/ion microscopes has been
reported [5.1]. The analyzer is rotationally symmetric about the primary beam axis,
and functions by using an electric field to mirror scattered electrons/ions emitted
from the specimen in a radial direction, transporting them onto a flat ring shaped
detector plane as shown in the analyzer schematic in Fig. 5.1.
Fig. 5.1 – Simulated trajectory paths through the Radial Mirror Analyzer (RMA)
design by Hoang et al. [5.1], 13 rays are plot over a polar angular spread () of
6 in uniform angular steps, shown here from the specimen to the detector plane at
the central energy Ep.
Chapter 5
103
The analyzer is predicted to have a relatively high energy resolution– transmittance
performance arising from its second–order focusing properties; a relative energy
resolution of better than 0.025% for an opening entrance polar angular spread of
6, this is around an order of magnitude better than the well–known Cylindrical
Mirror Analyzer (CMA) [5.2]. The analyzer is also expected to have a relatively
large bandwidth in its parallel energy mode of operation, over 12% ( 6%) of the
central pass energy, a factor of around four times better than the Hemispherical
Deflection Analyzer (HDA) [5.3].
This chapter follows on from the theoretical design of the RMA, using the design
to make and test a first RMA prototype. The prototype analyzer operates as an add–
on attachment that is placed in the same position as a typical backscattered electron
detector in the SEM specimen chamber, thereby allowing for concurrent usage of
the SEM imaging mode.
Since there are significant advantages to widening the energy range of a SE
analyzer, and detecting different electron energies in parallel, another theme of the
research work reported in this chapter is directed towards designing a wide–range
parallel energy analyzer, one in which the energy of detection varies by an order of
magnitude. Starting with the RMA design, simulation methods based upon direct
ray tracing successively led to the design of new analyzer, a wide–range first–order
focusing Parallel Radial Mirror Analyzer (PRMA).
Chapter 5
104
5.2 A wide–range parallel energy analyzer design
5.2.1 Need for a high transmittance wide–range parallel energy analyzer
The potential advantages of using wide–range parallel energy detection of charged
particles to speed–up spectrometer data–acquisition times are already well known.
One class of parallel analyzer designs is based upon the Hyperbolic Field Analyzer
(HFA), first reported by Curtis and Hsieh [5.4], and later rediscovered by Jacka et
al. working in the context of Auger Electron Spectroscopy (AES) [5.5–5.7]. The
Jacka et al. HFA design detects electrons in parallel over an energy range typically
from 75 to 2600 eV [5.7]. Recently, the HFA has been developed into an AES
attachment for Scanning Electron Microscopes (SEMs) by Cubric et al. [5.8], the
Shimadzu PAG4 analyzer, to be used in combination with a low voltage Argon ion
source. The entire Auger electron spectrum from the HFA can be captured in 1–2
seconds after the specimen surface has been cleaned by the ion gun, before any
appreciable hydrocarbon contamination on the specimen surface is allowed to
build–up. In this way, the HFA opens up the possibility of performing AES at high
vacuum pressures (10–7 to 10–6 Torr), obviating the usual requirement of AES
having to be carried out in Ultra High Vacuum (UHV) conditions (10–10 to 10–9
Torr). There are of course, similar possible data–acquisition time speed–up
advantages for SE energy analysis inside the SEM, and this forms the main
motivation for developing a wide–range energy analyzer design here.
Results from earlier chapters in this thesis using the second–order focusing toroidal
analyzer, a sequential bandpass energy analyzer, demonstrated that it can obtain
high signal–to–noise SE analyzer signals with a dwell time of approximately 150
Chapter 5
105
ms per sample (refer chapter 3 and 4), which means a SE spectrum with 100 points,
can be acquired in around 15 seconds. There are many instances where the speed–
up of SE analyzer signal data–acquisition time is required. It is necessary for
applications such as monitoring of hydrocarbon contamination build–up on the
specimen surface inside the SEM or characterization of insulators that tend to
charge up under the primary beam, or when energy filtered SE analyzer signals are
used to form images. Unlike Auger spectroscopy, which requires analyzers to have
high transmittance and high energy resolution at the same time, SE spectroscopy
applications require primarily high transmittance, which translates into good
signal–to–noise characteristics of the analyzer’s output signal. SE energy analyzer
applications typically require monitoring of changes in parameters like analyzer
signal peak position and/or the analyzer signal shape, which do not depend on the
energy resolution of the analyzer. Therefore, it is the transmittance of a wide–range
parallel energy analyzer that is more important when designing it for SE energy
analysis applications.
At present, analyzers based upon the HFA design have relatively small
transmittance; in the Jacka et al. proposal, the analyzer is reported to have an energy
resolution of 0.8% at 100 eV and 0.16% at 2500 eV, and the entrance angular spread
is limited to around 2.2 ( 1.1) in both polar and angular directions, translating
into less than 0.05% of all electrons emitted from the specimen reaching the
detector [5.7]. Although this has been improved by Cubric et al. to around 2.4 in
the polar angular direction and 5 in the azimuthal angular direction, the overall
transmittance is still relatively small, typically 0.1%, around 100 times smaller than
Chapter 5
106
that of the CMA (for a comparable energy resolution). This relatively low
transmittance comes from two limitations. Firstly, current HFA designs are
characterized mainly by first–order focusing properties on the detector plane,
limiting the angular spread in the plane of deflection (polar angular direction) to
relatively small values, typically less than 1.2. Secondly, since HFA designs
have been up to now, planar in geometry, electrons are detected over a narrow out–
of–plane angular range (azimuthal angular direction), typically less than 2.5.
Recognizing the need to improve the HFA transmittance, some authors have
proposed rotationally symmetric wide–range parallel energy analyzer designs.
However, within the context of scanning electron/ion microscopes, none of the
parallel analyzer designs proposed so far are rotationally symmetric with respect to
the scattered electrons/ions that are emitted from the specimen (primary beam axis).
They are, like the planar HFA, located to one side of the specimen, and limit the
range of out–of–plane electrons that can be detected. Cizmar et al. proposed a
rotationally symmetric version of the HFA, where the analyzer has a similar cross–
sectional field distribution through its axis of rotation to the planar HFA field
distribution. However, the axis of symmetry of the analyzer is in the horizontal
direction (perpendicular to the primary beam axis) and the range of energies on the
detection plane only varies by a factor of two [5.9]. Read proposed a wide–range
parallel analyzer design based upon the CMA layout, with its rotational axis of
symmetry lying in the horizontal direction, and refers to it as the Parallel
Cylindrical Mirror Analyzer (PCMA) [5.10]. The PCMA is similar to the CMA,
except that a linear potential variation is applied to the outer cylinder in the axial
Chapter 5
107
direction. The most practical class of PCMA designs was later developed by Read
et al. into an “axis to axis” PCMA configuration, where both the source and
detection plane lie on the analyzer’s axis of rotational symmetry [5.11]. They
assume that the source is located on the PCMA rotational axis, and that it emits
electrons/ions uniformly around it, different to the scanning electron/ion
microscope case, where secondary electrons/ions are emitted above and about the
primary beam axis. The axis to axis PCMA reported by Read et al. has an energy
range that varies from 300 to 1500 eV (change of detection energy by a factor of
5), and its predicted transmittance is over a magnitude better than that for the planar
HFA Jacka et al. design [5.10]. The polar angular spread is, like that of the planar
HFA design, relatively small, 2.1 ( 1.05). Cubric et al. have recently adapted
the PCMA Read et al. design to make it more suitable as an attachment for scanning
electron/ion microscope [5.8]. They relocate the entrance point to lie on the
analyzer front plate, instead of on the inner cylinder plate, so that the distance
between the analyzer entrance and the primary beam axis can be increased,
providing more room to fit it into place. They also redesign the analyzer so that
scattered electrons/ions land over a rectangular region on the detection plane,
instead of producing a line focus along the rotational axis of symmetry, in order to
reduce the intensity of focus on the detector. In the Cubric et al. design, the energy
range is widened so the detected energy range varies by a factor of 25. The
azimuthal angular spread (out–of–plane angular spread) that can be accepted into
the analyzer is 30.
Chapter 5
108
The following work is based upon designing analyzers that are fully rotationally
symmetric with respect to the primary beam axis, and do not place restrictions on
the scattered electron/ion azimuthal angular range, that is, they have a range of 360
in the out–of–plane direction. The following section presents the design of a first–
order focusing wide–range parallel Radial Mirror Analyzer (RMA), which will be
referred to here as a first–order Parallel Radial Mirror Analyzer (PRMA). Although
the analyzer has first–order focusing properties at the detector plane, for SE signal
acquisition, the exit slit may be widened and the polar entrance angle can be
increased to increase the transmittance of the analyzer. This becomes possible as
signal–to–noise characteristics and not energy resolution of the analyzer is the main
consideration as far as SE analyzer signal acquisition is concerned.
The design approach taken in the following work is essentially a simulation based
one, where all field distributions and electron trajectory ray paths were simulated
using the Lorentz–2EM program [5.12]. This program utilizes the boundary
element method to solve for electrostatic field distributions, in which an adaptive
segment technique automatically optimizes the number of charge segments used on
conductor surfaces, refining them according to the local field strength, and therefore
achieving higher accuracy for a given program run time. A 5th order Runge–Kutta
method variable step method is used for direct ray tracing of charged particle
trajectories, where the trajectory step is adjusted according to the local truncation
error. The accuracy of all simulations were continually checked by repeating all
results with smaller boundary segments and trajectory step sizes, ensuring that
important ray tracing parameters, such as the final focal spot–size of the electron
Chapter 5
109
beam at the spectrometer exit did not change significantly (by less than 1 %). Apart
from estimating the energy resolution visually by observing simulated electron ray
paths around the focal plane, it was calculated numerically from trace–width and
energy dispersion characteristics along the detection plane as a function of electron
energy.
5.2.2 A first–order focusing wide–range PRMA design
Fig. 5.2 shows simulated trajectories through the conventional ideal HFA design,
in which the potential field distribution V(x,y) is described by the following simple
analytical equation
𝑉 (𝑥, 𝑦) =2𝑉0
𝑅02 𝑥𝑦
where V0 is the potential on a curved hyperbolic shaped electrode whose tip (at x =
y) is located at a distance R0 from the analyzer bottom left–hand corner (x= 0, y=0).
The bottom and left–hand side boundaries of the analyzer are fixed to be at zero
volts, and an ideal grid is assumed at the bottom boundary where electrons/ions
enter the analyzer. Note that the distance R0 here, corresponds to 2 times the
variable b used by Jacka et al. [5.6]. Eleven trajectories from the specimen (located
outside the analyzer) are evenly plot over a polar angular spread of – 1.1 to 1.1
around a central entrance polar angle of 25 with respect to the x axis for the
electron energies of 100, 200, 500, 1000, 2000, 3000, and 5000 eV. Equipotential
lines of 500 – 4000 volts in 500 volt steps are also indicated. The simulated energy
resolution, corresponding to half the trace–width on the detector plane, starts out
around 0.65% at 100 eV and drops to 0.2% at 5000 eV. If the polar angular spread
Chapter 5
110
ranges from – 3 to 3, the simulated energy resolution varies from 1.68% at 100
eV, reaches a maximum of 2.25% at 500 eV and drops to 1.33% at 5000 eV.
Fig. 5.2 – Simulated trajectory paths through a planar ideal hyperbolic field
analyzer design at the energies 100, 200, 500, 1000, 3000 and 5000 eV. For each
energy, eleven trajectories are plot evenly between −1.1° and 1.1° around a 25°
polar entrance angle.
The HFA layout shown in Fig. 5.2 is the one reported to have a single second–order
focusing point at 100 eV, and first–order focusing properties for all other energies
[5.6], where the energy (E) to voltage ratio (V0) for R0 = 14.14 mm is 1.19 (V0 = –
84.03 volts). However, the simulation work performed here finds the 2nd order focus
to be located at 50 eV for these conditions, and not at 100 eV. In order to obtain 2nd
order focusing at 100 eV, the present simulations indicate that V0 needs to be –
166.67 volts at R0 = 14.14 mm, which corresponds to (E/V0) = 0.6 and not 1.19 as
reported previously. If a (E/V0) ratio of 0.6 is used, then the simulated energy
resolution for the higher energy range is predicted to be 0.38% at 2000 eV and
0.27% at 5000 eV, instead of 0.29% and 0.2% respectively at a (E/V0) ratio of 1.19.
In the present context, this difference is not important, the main point here is that
Primary
Beam 5 cm
100 200 500 1000 2000 3000 5000 eV
Chapter 5
111
there is only one energy for which second–order focusing occurs, while for all other
energies, the analyzer is characterized by first–order focusing properties. To be
consistent with what has been reported previously, a (E/V0) ratio of 1.19 is taken to
be representative of previous HFA designs.
As a first step towards transforming the HFA into a wide–range parallel first–order
RMA, let us use the electrode layout of the HFA depicted in Fig. 5.3, and map it
into a rotational symmetric geometry, where the axis of rotational symmetry lies on
the primary beam axis.
A hyperbolic shaped mirror conductor boundary at –1500 volts for R0 = 2 cm (in
the ideal HFA described by equation 1) becomes a fixed boundary condition in the
new axi–symmetric coordinate system, the zero boundaries maintain their relative
positions, and the resulting field distribution is then solved numerically by the
Lorentz–2EM [5.12] software with the origin of the coordinate system placed at the
specimen. Fig. 5.3 shows simulated ray paths for energies 250, 750, 1250, 1750 and
2500 eV in the transformed rotationally symmetric geometry. They indicate that
simulated electron trajectory paths no longer focus on to a horizontal detector plane.
In general, for rotationally symmetric structures, a different electrode layout to the
ideal HFA one is therefore required.
Chapter 5
112
Fig. 5.3 – Simulated ray paths for energies 250, 750, 1250, 1750 and 2500 eV in
the ideal HFA electrode layout (R0 = 2 cm and V0 = –1500 volts) transformed into
axi–symmetric cylindrical coordinates. Nine trajectories plot with an angular
spread over 1.1.
A rotationally symmetric analyzer, modeled on the geometrical layout of the
present RMA (shown in Fig. 5.1), but using deflection potentials from the ideal
HFA formula (equation 5.1) as starting values was specified. Simulated trajectory
paths indicated that more than three conical deflectors were required in order to
provide better control of the focal point positions on the final detection plane, and
that the main deflector could be successfully represented by a single straight
electrode (in the cross–sectional plane). Continual refinement of the deflection
electrode potentials and geometry was carried out in order to focus all energies on
to the bottom detection plane, and this led to the wide range parallel rotational
symmetric analyzer design shown in Fig. 5.4.
Rotational axis
Specimen
Focal plane
V0
0 V
0 V
Radius 2500 1750 1250 750 250 eV
Chapter 5
113
Fig. 5.4 – Simulated trajectory paths through a first–order focusing PRMA design.
Electrons are plot for energies 50, 100, 200, 500, 1000, 1500, 2000 and 2500 eV
with 9 electron trajectories over a polar angular spread of ± 1.1˚ around a central
angle of 24.8˚. V1 = – 45V; V2 = – 120 V; V3 = – 285 V; V4 = –775V ; V5 = –
1150V; V6 = – 1675V; V7 = – 2020V.
A set of deflection electrodes inside the analyzer, at voltages V1 to V6, lie along a
conical surface above the entrance grid, and can be used to control the focal point
position for different energies. The top curved mirror electrode in the ideal HFA
design is substituted with a single straight electrode V7 (straight in the cross–
sectional plane). Electrons are plot for energies 50, 100, 200, 500, 1000, 1500, 2000
and 2500 eV with 9 electron trajectories over a polar angular spread of ± 1.1˚ around
a central angle of 24.8˚, and the equipotentials in uniform steps of 135 volts range
from 0 to –2020 volts. The electrode voltages (in volts) are: V1 = –45, V2 = –120,
V3 = –285, V4 = –775, V5 = –1150, V6 = –1675, and V7 = –2020.
The voltages V1 to V7 were systematically adjusted to focus electrons on to the
horizontal bottom analyzer plane, starting with the lower energy range. Focusing
Chapter 5
114
onto a horizontal detection plane is predicted for a variety of different electrode
geometries, and for each layout, its electrode voltages need to be systematically
adjusted (starting with the lower energies). As a general guide, for the analyzer
layout shown in Fig. 5.4, the voltages V1 and V2 largely control focal point positions
in the lower energy range (0 – 200 eV), V3 to V5 mostly control the focal point
positions in the middle energy range (200 to 1500 eV), while V5 and V6 mainly
control the focal point positions of higher energy electrons (1500 to 2500 eV). The
voltage V3 has an effect on focal point positions in the lower energy range, but this
can be easily compensated by changing V2. The voltage V7 has the overall effect of
shifting the focal point position horizontally. Although the voltage V1 affects the
focusing of the entire energy range, its effect greatly diminishes as the electron
energy increases, and since only small changes in V1 are required to control the
focal point positions of the lower energy electrons, the effect of these changes on
the higher energy electrons is negligible.
Fig. 5.5a shows simulated rays paths around the focal plane in the rotationally
symmetric wide–range analyzer design for electron energies 2% below and above
the central energies of 500 and 1500 eV (angular spread of 1.1). They provide a
visual estimation of the relative energy resolution (half trace–width) to be in the
0.25 to 0.3% range. A more rigorous approach, based upon calculating the trace–
width and energy dispersion characteristics along the detection plane is presented
in Table 5.1.
Chapter 5
115
Table 5.1 – Simulated energy resolution for first–order focusing PRMA designs.
The simulated energy resolution typically varies from 0.46 to 0.25% across the 50
to 2500 eV energy range, comparable to those predicted for the ideal planar HFA
design, where there is a maximum simulated energy resolution of 0.68% in the
lower energy range (at 200 eV), falling to 0.265% at 2500 eV. Fig. 5b shows the
simulated trace–width as a function of angular spread, indicating that the analyzer
design is predicted to have first–order focusing properties (second–order spherical
aberration) across its entire detected energy range.
Simulated % relative energy resolution
1.1 angular spread
Energy (eV) First–order focusing PRMA
Previous ideal HFA
design
50 0.463 0.114
100 0.408 0.588
200 0.462 0.680
500 0.274 0.514
1000 0.241 0.387
1500 0.250 0.325
2000 0.238 0.291
2500 0.255 0.265
Chapter 5
116
Fig. 5.5 – Simulated spot size charcteristics of the first–order focusing PRMA
design: (a) Direct ray paths at the focal plane for 2% energies around the central
energies of 500 eV and 1500 eV (b) Trace–width as a function of input polar
angular spread ranging from –20 mrad (–1.145) to +20 mrad (+1.145).
a
b
Chapter 5
117
5.3 Experimental prototype of a RMA attachment inside a SEM specimen
chamber
Fig. 5.1 shows the layout of the RMA design and simulated ray paths for scanning
electron/ion microscopes, similar to the one recently reported by Hoang and
Khursheed [5.1]. Scattered electrons/ions leave the specimen located below the
analyzer on its rotational plane of symmetry (primary beam axis), they enter the
analyzer through a grid and are mirrored down by negatively biased electrodes, exit
the analyzer through another grid, and are brought to focus beneath it on a
horizontal detector. The analyzer has an outer zero volt plate, conical in shape on
the top so that it fits under the lower pole–piece of a scanning electron/ion
microscope objective lens. This arrangement minimizes the working distance (the
distance between the objective lens lower pole–piece and the specimen), allowing
the scanning electron/ion microscope to operate in a high spatial resolution imaging
mode. The first series of deflector plates within the analyzer are three conical
electrodes at potentials, V1, V2 and V3, and they are followed by a main top deflector
plate, biased to VD. The required take–off angle from the specimen, with respect to
the horizontal direction, is . The analyzer focal properties, position and quality of
focus on the horizontal detector plane, as well as the energy bandwidth, are all
determined and controlled by the deflector plate voltages and take–off angle. For
the analyzer layout shown in Fig. 1, these design parameters are as follows: =
33.4, V1 = – 0.172EP, V2 = – 0.470EP, and V3 = – 0.570EP, and VD = – 0.540EP,
where EP is the analyzer central (pass) energy (in electron volts), and 13 simulated
rays in uniform angular steps are plot within an angular spread () of 6 at
Chapter 5
118
energy Ep. Although the analyzer design shown in Fig. 5.1 is similar to the one
reported by Hoang and Khursheed, it does however, incorporate a small
improvement which is important to highlight in the present context. Instead of the
main top deflector (at voltage VD) having a concave curved shape (in the cross–
sectional plane), it consists of two straight segments. This not only makes it easier
to manufacture, but it also seems to improve the simulated energy resolution at the
detector plane by approximately a factor of two. The fact that the main curved
deflector plate of the RMA design can be simply substituted by two straight
segments (straight in the cross–sectional plane), suggests that there is no advantage
to be gained by using curved electrodes, and straight deflector electrodes in the
radial cross–section are likely to be adequate also for the present analyzer design.
Based on this theoretical design, the first prototype of the RMA was fabricated to
fit as an attachment inside a Philips ESEM XL30 FEG SEM. The main design
philosophy was to allow the SEM to operate in the normal imaging mode while
concurrently acquiring signals using the analyzer. Fig. 5.6 illustrates how the RMA
prototype fits inside the SEM; the analyzer is mounted on a push-pull linear
manipulator fitted onto one of the ports of the SEM chamber. The push-pull linear
manipulator facilitates the movement of the analyzer to bring it closer to the final
pole piece of the SEM and occupy the typical position of a BSE detector during
operation and to withdraw the analyzer when not in use. In this way, the analyzer
can be used to acquire the energy filtered scattered electron signals while the
conventional SE detector can continue to operate normally, acquiring high
resolution topographical images of the specimen.
Chapter 5
119
Fig. 5.6 – Schematic diagram showing the integration of the RMA attachment
inside the SEM chamber with other components. Such a mounting of the analyzer
facilitates operation of the SEM in the normal imaging mode.
The experimental layout of the RMA, to fit as an attachment inside the SEM, is
depicted in Fig. 5.7. A potential divider arrangement is used to bias the deflector
electrodes at their respective voltages for a given pass energy. The voltage to the
30
cm
Electron Column final
pole piece Conventional SE
detector
PE
Philips ESEM XL30
Vacuum Chamber
Linear manipulator
Radial mirror analyzer
attachment prototype
Specimen
Specimen stage
Chapter 5
120
potential divider is ramped to capture the scattered electron spectrum. The
spectrometer is designed to capture an angular spread of ± 6 with respect to the
central entrance angle of 33.4 in the polar direction, which is achieved by varying
the height of the specimen and position of the analyzer. The effective input angular
spread in the azimuthal direction is 100.
Fig. 5.7 – Experimental layout of the RMA inside the SEM.
SEM
column
Specimen
VD
V1
V2
V3
θ = 33.4
0 V
3.4 cm Rotational
symmetry
Primary
Analyzer
0 V
0 V
shielded
casing
PMT
+
Scintillator
+5 kV
0 V
0 V
80% open
Metal grids
Exit slit (500 µm)
7 cm
Potential divider
Voltage
SEM STAGE
Electron
Trajectory
Chapter 5
121
The outer cover is grounded in order to prevent electric field leakage into the SEM
specimen chamber. Two grounded electrostatic grids are also used to cover the
entrance and the exit of the spectrometer in order to avoid distortion of the electric
field near these regions. The RMA design focuses the transmitted electrons in such
a way that electron trajectories travel radially out upon exit, and do not naturally
converge to a point like they do in the toroidal analyzer. Ideally, an electrostatic
deflector (some kind of post–analyzer toroidal deflector) is required to redirect all
out–of–plane electrons towards the primary beam axis and focus them on to a single
electron detector placed below the specimen stage. However, for this preliminary
prototype experiment, it was not done due to space and time constraints. Instead, a
PMT/scintillator fixed to the bottom of a 0 volt shielded box was placed below the
RMA exit slit aperture, as shown in Fig. 5.7. The shielding prevents electrons from
being affected by other electrostatic fields while they travel to the PMT. The
transport efficiency from the RMA exit slit to the PMT is expected to be low,
however, for these preliminary proof–of–concept experiments, this arrangement
was found to be adequate for capturing the SE energy spectrum. A 3D drawing of
one half of spectrometer attachment prototype and a photo of the fabricated
attachment integrated with the SEM is given in Fig. 5.8.
Chapter 5
122
Fig. 5.8 – A prototype of the RMA attachment: (a) Cross–section 3D CAD model
(b) A photo of the attachment integrated inside the SEM.The azimuthal deflection
angle is 100.
3.4 cm
Metallic Grids
Exit Slit
PMT with
Scintillator
Analyzer deflector
electrodes
a
b
Chapter 5
123
Experimental SE spectra, obtained from the initial prototype of the RMA are
presented in Fig. 5.9. The experiments were conducted inside a Philips ESEM
XL30 FEG SEM with a primary beam energy of 10 keV (beam current 150 pA)
and the deflection voltage was ramped in steps of 200 mV. A silicon wafer coated
with 300 nm of gold was used as the specimen.
The experimental SE analyzer signals obtained here are similar to the expected
Chung–Everhart SE distribution (refer chapter 2) [5.13], and they shift to the right
as the specimen is biased to – 2 V, indicating an increase in initial kinetic energy
(also expected).
Fig. 5.9 – Experimental SE analyzer signals obtained using the first experimental
prototype of the RMA.
Analysis of the SE analyzer signals shown in Fig. 5.9 reveals that the SE signal
levels are much lower than expected, typically a factor of 100 lower than the signal
strength obtained by the second–order focusing toroidal analyzer used for the
PM
T o
utp
ut
(a.u
)
Deflection Voltage (Volts)
Specimen Voltage = 0 V
Specimen Voltage = – 2 V
VS = VC1
Chapter 5
124
earlier experiments as described in chapters 3 and 4 of this thesis. A factor of only
one half is expected due to the 80% transparency of the entrance and exit grids
together with the smaller entrance angular spread (± 6º instead of ± 8º). These
experimental results points towards low transport efficiency from the RMA slit to
the PMT, which is at present not designed to capture the out–of–plane electrons. A
new detection strategy is required, in order to redirect and focus the out–of–plane
electrons. This proposal is discussed in more detail in the next chapter, nevertheless,
preliminary experimental have been carried out which validate the working
principle of the RMA analyzer.
5.4 Conclusion
This chapter has presented a new wide–energy parallel analyzer attachment design
for the SEM, the first–order PRMA, which can be used to speed–up the data–
acquisition time of SE analyzer signals. This chapter has also presented
experimental results from a prototype RMA spectrometer operating as an
attachment inside the SEM chamber, occupying the same position as a typical BSE
detector. The preliminary SE spectral results validate the general principle of the
analyzer, however, further work is required to devise and implement an efficient
technique to collect the transmitted electrons from the RMA exit slit to the detector.
Chapter 5
125
References
5.1. Hoang, H.Q. and A. Khursheed, A radial mirror analyzer for scanning
electron/ion microscopes. Nuclear Instruments and Methods in Physics
Research Section A: Accelerators, Spectrometers, Detectors and Associated
Equipment, 2011. 635(1): p. 64–68.
5.2. Sar‐El, H.Z., Criterion for Comparing Analyzers. Review of Scientific
Instruments, 1970. 41(4): p. 561–564.
5.3. Benis, E. and T. Zouros, The hemispherical deflector analyser revisited: II.
Electron–optical properties. Journal of Electron Spectroscopy and Related
Phenomena, 2008. 163(1): p. 28–39.
5.4. Curtis, C. and K. Hsieh, Compact wide‐aperture hyperbolic analyzers.
Review of scientific instruments, 1988. 59(11): p. 2424–2428.
5.5. Jacka, M., et al., A fast, parallel acquisition, electron energy analyzer: The
hyperbolic field analyzer. Review of scientific instruments, 1999. 70(5): p.
2282–2287.
5.6. Jacka, M., A. Kale, and N. Traitler, Hyperbolic field electron energy
analyzer with second–order focusing. Review of scientific instruments,
2003. 74(10): p. 4298–4300.
5.7. Jacka, M., Scanning Auger microscopy: recent progress in data analysis and
instrumentation. Journal of Electron Spectroscopy and Related Phenomena,
2001. 114–116(0): p. 277–282.
5.8. Cubric, D., et al., Parallel acquisition electrostatic electron energy analyzers
for high throughput nano–analysis. Nuclear Instruments and Methods in
Chapter 5
126
Physics Research Section A: Accelerators, Spectrometers, Detectors and
Associated Equipment, 2011. 645(1): p. 227–233.
5.9. Cizmar, P., et al., New multichannel electron energy analyzer with
cylindrically symmetrical electrostatic field. Review of scientific
instruments, 2007. 78(5): p. 053714.
5.10. Read, F.H., The parallel cylindrical mirror electron energy analyzer.
Review of Scientific Instruments, 2002. 73(3): p. 1129–1139.
5.11. Read, F.H., et al., The parallel cylindrical mirror analyzer: axis–to–axis
configuration. Nuclear Instruments and Methods in Physics Research
Section A: Accelerators, Spectrometers, Detectors and Associated
Equipment, 2004. 519(1–2): p. 338–344.
5.12. LORENTZ – EM. 2011, Integrated Engineering Software Inc, Canada.
5.13. Chung, M.S. and T.E. Everhart, Simple calculation of energy distribution
of low‐energy secondary electrons emitted from metals under electron
bombardment. Journal of Applied Physics, 1974. 45(2): p. 707–709.
5.14. Khursheed, A., H.Q. Hoang, and A. Srinivasan, A wide–range Parallel
Radial Mirror Analyzer for scanning electron/ion microscopes. Journal of
Electron Spectroscopy and Related Phenomena, 2012. 184(11): p. 525–532.
Chapter 6
127
Chapter 6 – Conclusions and Suggestions for future work
6.1 Conclusions
The main objective of this thesis was to improve upon the current analytical
capabilities of the SEM by the use of secondary electron energy analyzer
attachments, transforming it into a more powerful tool for nanometer scale material
analysis.
Experimental results were presented to demonstrate that it is possible to obtain high
signal-to-noise voltage and dopant concentration measurements on semiconductor
specimens using the second–order focusing toroidal energy analyzer attachment
even in the presence of fringe fields and surface fields above the specimen. A
variety of new applications for SE energy analyzers in the SEM were also reported.
High contrast SE analyzer signals were obtained from: multi-functional oxide
interfaces; from specimen in changing magnetic fields; and from oxidizing thin film
metal layers. A new wide-range parallel SEM energy analyzer attachment design,
the first–order parallel Radial Mirror Analyzer (PRMA), was reported. The first-
order PRMA can be used to acquire SE analyzer signals simultaneously across the
complete SE energy range, speeding up data acquisition time by more than an order
of magnitude. Finally, an experimental prototype of the Radial Mirror Analyzer
SEM attachment was made, and preliminary experimental results were obtained to
verify its working principle. The RMA attachment can be placed in the SEM
chamber in the same position occupied by BSE detector attachments, and has the
Chapter 6
128
advantage of having high performance optics while at the same time allowing for
short SEM working distances (< 10 mm).
6.2 Suggestions for future work
The work presented in this thesis forms a good basis for future work in the areas
elaborated below.
The current prototype of the second order focusing toroidal analyzer has a limited
circular field of view of about 1 mm in diameter. Also movement of the specimen,
once they are loaded inside the specimen holder, is not possible. Therefore the
analyzer prototype may be modified in such a way that the analyzer becomes part
of the SEM chamber which can be brought into operation using a linear manipulator
when required as shown in Fig. 6.1.
Fig. 6.1 – Proposed modification of the second order focusing toroidal analyzer
where the specimen is independent of the main analyzer body, allowing free
movement of the specimen.
Chapter 6
129
The results obtained from p-Si/n-ZnO heterojunction samples are a good starting
point for an in-depth study of other p-n junction specimen, like solar cells and these
results can be compared with those obtained using established methods for p-n
junction characterization.
The experimental results obtained from buried interfaces of multi-functional oxides
hold great promise of being the basis of a new contactless technique of
characterizing the conductivity of thin film oxide interfaces. In order to gain a better
understanding of its potential, the technique needs to be applied to multifunctional
oxide devices made up of different materials. The SE analyzer signal contrast
obtained from a specimen in presence of magnetic fields (created by a current
carrying solenoid placed under the specimen) naturally leads to the question of how
this technique might be applied to the study of local micro-magnetic domains on
the specimen.
Another important direction of future work that needs to be carried out is to design
and develop a post analyzer deflector for the Radial Mirror Analyzer (RMA), to
redirect and focus the transmitted electrons exiting at different azimuthal angles to
a single point or line that can be conveniently detected by a single PMT or
channeltron detector. One strategy, shown in Fig. 6.1 is to use a toroidal electric
sector analyzer placed below the RMA exit aperture slit, deflecting all transmitted
electrons back towards the primary beam axis, and focusing them on to single
detector placed below the specimen stage. The simulation results shown in Fig. 6.1
were obtained through the use of Lorentz 2EM. Finally, future work should also be
directed towards making a prototype of the first–order focusing PRMA design and
Chapter 6
130
use it to speed up data acquisition time, and thereby create new possible
applications, such as acquiring nanoscale quantitative SE analyzer images.
Fig. 6.2 – Proposed post analyzer deflector arrangement for the RMA attachment;
7 rays are plot over a polar angular spread () of 6 in uniform angular steps,
shown here from the specimen to the scintillator of the PMT through the post
analyzer deflector, at the central energy Ep. The magnitude of VPD was
experimentally calculated to be 0.436EP.
Radial Mirror
Analyzer attachment
SEM stage
Analyzer exit
+ VPD – VPD
Scintillator voltage
VSC = + 5000 V
SEM Column
Post analyzer deflector
arrangement
θ = 33.4
Appendix A
131
Appendix A: Publications resulting from this project
Journal
1. Khursheed, A., H.Q. Hoang, and A. Srinivasan, A wide-range Parallel Radial
Mirror Analyzer for scanning electron/ion microscopes. Journal of Electron
Spectroscopy and Related Phenomena, 2012. 184(11): p. 525-532.
2. Srinivasan, A., and A. Khursheed., Probing and Analyzing Buried Interfaces
of Multifunctional Oxides Using a Secondary Electron Energy Analyzer.
Microscopy and microanalysis: the official journal of Microscopy Society of
America, Microbeam Analysis Society, Microscopical Society of Canada
(2014): 1-5.
3. Srinivasan, A., and A. Khursheed. "Voltage and dopant concentration
measurements of semiconductors using a band-pass toroidal energy analyzer
inside a SEM" - Submitted to Elsevier-Ultramicroscopy.
Conference Proceedings
1. Srinivasan and A. Khursheed , “Detection of surface voltage changes using a
second–order Focusing toroidal energy analyzer SEM attachment”, Proceeding of
the 13th International Seminar on Recent Trends in charged particle optics and
Surface Physics Instrumentation, Skalsky Dvur near Brno Czech Republic, pp73,
2012.