DIRECT EXPERIMENTAL OBSERVATION OF 3D VORTEX STATES
INMULTILAYER FE/GD USING SCANNING ELECTRON
MICROSCOPYWITH POLARIZATION ANALYSIS
(SEMPA)
by
RICHMORASKI
A DISSERTATION
Presented to the Department of Physics
and the Division of the Graduate Studies of the University or Oregon
in partial fulfillment of the requirements
for the degree of
Doctor of Philosophy
March 2023
DISSERTATION APPROVAL PAGE
Student: Rich Moraski
Title: Direct Experimental Observation of 3D Vortex States in Multilayer Fe/Gd
Using Scanning Electron Microscopy with Polarization Analysis (SEMPA)
This dissertation has been accepted and approved in partial fulfillment of the re-
quirements for the Doctor of Philosophy degree in the Department of Physics by:
Dietrich Belitz Chair
Benjamin McMorran Advisor
Jayson Paulose Core Member
Cathy Wong Institutional Representative
and
Krista Chronister Vice Provost for Graduate Studies
Original approval signatures are on file with the University of Oregon Division of
Graduate Studies.
Degree awarded March 2023.
2
© 2023 Rich Moraski
This work is licensed under a Creative Commons
Attribution-ShareAlike (United States) License.
3
DISSERTATION ABSTRACT
Rich Moraski
Doctor of Philosophy
Department of Physics
March 2023
Title: Direct Experimental Observation of 3D Vortex States in Multilayer Fe/Gd
Using Scanning Electron Microscopy with Polarization Analysis (SEMPA)
The global market for power solely for data center usage is estimated to
be $12.4 billion by 2027[1]. In 2021, data center electricity consumption was
∼400 TWh, representing almost 2% of the global energy demand[2]. Ongoing
efforts in spintronics, which use spin currents instead of traditional charge
currents at a fraction of the power[3], are paving the way for significant savings,
both financially and environmentally. There has been considerable research into
alternatives for memory[4–6] including a magnetic structure known as a skyrmion,
a self-supporting magnetic texture characterized by a non-trivial topology. Recent
advances in creating room-temperature stable skyrmions has reignited interest in
these objects.
Building on previous work in the McMorran group, this research set out to
build a more complete understanding of the 3D structure of metastable magnetic
skyrmions, specifically in Fe/Gd thin films. This was done using traditional trans-
mission electron microscope (TEM) techniques along with a unique scanning elec-
tron microscope with polarization analysis at the University, the SEMPA. Data col-
lected using a TEM in Lorentz mode, providing information integrated through
the bulk of the material, was combined with data from SEMPA, providing surface-
sensitive information about the top of the material. Analysis of the data suggests a
4
topologically complex winding nature for the magnetization of skyrmions in this
material.
Presented herein is a brief introduction of the magnetic structures found in
Fe/Gd multilayer thin films; an analysis using new analytical tools built for this
purpose of the data collected; and a user’s manual for SEMPA, including mainte-
nance and troubleshooting guidance.
5
CURRICULUM VITAE
NAME OF AUTHOR: Rich Moraski
GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED:
University of Oregon, Eugene, OR
George Mason University, Fairfax, VA
University of Virginia, Charlottesville, VA
Longwood University, Farmville, VA
University of Central Florida, Orlando, FL
DEGREES AWARDED:
Doctor of Philosophy, Physics, 2023, University of Oregon
Master of Science, Physics, 2019, University of Oregon
Bachelor of Science, Physics, 2017, George Mason University
Bachelor of Science, Commerce, 1992, University of Virginia
AREAS OF SPECIAL INTEREST:
Condensed Matter Physics
Electron Microscopy
Nanomagnetism
Spintronics
PROFESSIONAL EXPERIENCE:
Graduate Research Assistant, University of Oregon, Eugene, OR, 9/2017 -
Present
President, Principle Consultant, Orbium Inc., Eugene, OR, 6/1997 - Present
Senior Program Director, Capital One, Mclean, VA, 12/2015 - 9/2017
Advisory Solutions Consultant for Major Accounts, ServiceNow, Mclean,
VA, 9/2014 - 12/2015
Senior Director of Cloud Automation, Windward IT Solutions, Herndon,
VA, 2/2012 - 9/2014
Senior Director of Knowledge Engineering, BMC Software, Herndon, VA,
7/2007 - 2/2012
Chief Architect, Director of Knowledge Engineering, RealOps, Herndon,
VA, 10/2004 - 7/2007
Chief Architect, Windward Consulting Group, Herndon, VA, 1/2004 -
10/2004
Senior Associate, Booz Allen Hamilton, Vienna, VA, 5/2003 - 1/2004
6
Senior Manager, Bell Atlantic, Arlington, VA, 7/1992 - 6/1997
PUBLICATIONS
Rich Moraski et al. “Electron Microscopy Spin Analysis of Topological
Magnetic Domains in Amorphous Fe/Gd Thin Films”. In: Microscopy
and Microanalysis 28.S1 (2022), pp. 1688–1689
Rich Moraski and Benjamin McMorran. “3D Morphology of Magnetic
Bubbles in Layered Ferromagnetic Materials”. In: Microscopy and Mi-
croanalysis 27.S1 (2021), pp. 150–152
Cameron W Johnson et al. “Exact design of complex amplitude holograms
for producing arbitrary scalar fields”. In: Optics express 28.12 (2020),
pp. 17334–17346
7
ACKNOWLEDGEMENTS
I thank Professor McMorran for his assistance in the preparation of this
manuscript. I also thank John Unguris who built the SEMPA during his tenure
at the National Institute of Science and Technology; Eric Fullerton and Sergio
Montoya for helping decipher the results; and my lab mates Amy Turner, Will
Parker, and Jacques Reddinger. Steve Weimholt of CAMCOR was invaluable in
getting and keeping SEMPA going. This research was supported in part by a grant
from the National Science Foundation, 2105400, to Dr. Benjamin McMorran at the
University of Oregon.
8
I dedicate my dissertation work to my family and many friends. A special thank
you is due to my parents, who endured my inquisitiveness from the beginning. My
sister and niece, as well as my new family (if 14 years can really be called new) were
also as supportive as could be.
Brandon Smith (way more than) helped ease the financial burden of being a full-
time student again.
Gromit, Ripley, Wallace, Kevin, Barky, Finn, and Charlie provided uncondi-
tional love as only the four-legged can.
None of this, however, would have happened without Carla. I just might love
you more than ice cream. Maybe.
9
TABLE OF CONTENTS
Chapter Page
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
I. BACKGROUND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Magnetic Bubbles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Phenomenological Description of Magnetic Skyrmions . . . . . . . . 16
Stabilization Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Dzyaloshinskii-Moriya interaction (DMI) . . . . . . . . . . . . . . . . . 24
Frustrated exchange interactions . . . . . . . . . . . . . . . . . . . . . . 25
Four-spin exchange interaction . . . . . . . . . . . . . . . . . . . . . . . 25
Long range dipolar interaction . . . . . . . . . . . . . . . . . . . . . . . 25
Engineering Skyrmions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Nanomagnetic Microscopy Techniques . . . . . . . . . . . . . . . . . . . . 27
Dissertation Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
II. SEMPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Theory of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
The Auger-Meitner Effect . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Mott Scattering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
The Microscope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Plasma Gun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Electromagnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Molecular Beam Epitaxy (MBE) . . . . . . . . . . . . . . . . . . . . . . 36
Multi-pin Sample Holder . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Manipulator Arm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
SEMPA Column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
III. METHODOLOGY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
SEMPA Alignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
10
SEMPA Capture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
Data Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Importing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Registration and scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Removing offsets and rescaling . . . . . . . . . . . . . . . . . . . . . . . 52
Calculating angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Finding features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
IV. ZOOLOGY OF TEXTURES IN FE/GD . . . . . . . . . . . . . . . . . . . . 57
Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Skyrmions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Antiskyrmions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Trivial Bubbles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Worms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Feature Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Distribution of Magnetization Angles Across Features . . . . . . . . . 68
Comparison of Skyrmion Sizes from SEMPA and LTEM . . . . . . . . 72
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Shapes Observed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Skyrmions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
Antiskyrmions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Bubbles and Worms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
IV.0.1 Feature Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
V. CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Dissertation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Implications and the Future . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
A. PYNISTVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
11
B. SEMPA MANUAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
B.2 Concept of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
B.3 Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
B.4 Alignments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
B.5 Startup and Shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
B.6 SEM Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
B.7 Auger Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
B.8 Ion Milling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
B.9 Plasma Milling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
B.10 Electromagnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
B.11 SEMPA Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
B.12 Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
B.13 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
B.14 Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
REFERENCES CITED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
12
LIST OF FIGURES
Figure Page
1.1 Schematic of skyrmions and an antiskyrmion . . . . . . . . . . . . . . 16
1.2 Bloch and Néel domain walls . . . . . . . . . . . . . . . . . . . . . . . . 19
1.3 Schematics of Bloch skyrmions . . . . . . . . . . . . . . . . . . . . . . . 20
1.4 Schematics of Néel skyrmions . . . . . . . . . . . . . . . . . . . . . . . 21
1.5 Schematics of antiskyrmions . . . . . . . . . . . . . . . . . . . . . . . . 22
1.6 Schematics of bubble . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.7 LTEM image formation . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.1 Polarized secondary electron energies . . . . . . . . . . . . . . . . . . . 32
2.2 Electron beam interaction volumes . . . . . . . . . . . . . . . . . . . . 33
2.3 Schematic of SEMPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.4 SEMPA optics schematic . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.5 SEMPA cutaway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.6 Mott detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.7 Magnetization of an Fe whisker . . . . . . . . . . . . . . . . . . . . . . 42
3.1 Fe/Gd layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.2 Stripe domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3 Auger analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.4 Raw SEMPA images . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.5 Image registration homography applied . . . . . . . . . . . . . . . . . . 51
3.6 3D vs. 2D magnitudes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.7 Normalized magnetization magnitudes . . . . . . . . . . . . . . . . . . 54
3.8 Circular features identified . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.1 All contours over θ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.2 All contours over ϕ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.3 Circular contours . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.4 Skyrmion 23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.5 Skyrmion 27 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
13
4.6 Skyrmion 26 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.7 Skyrmion 28 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.8 Antiskyrmion 9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.9 Antiskyrmion 12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.10 Bubble 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.11 Bubble 14 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.12 Worm 38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.13 Worm 50 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.14 Feature Group 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.15 Feature Group 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.16 Alphas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.17 Skyrmion sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.18 Overview of LTEM data . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.19 Decreasing α from center to wall . . . . . . . . . . . . . . . . . . . . . . 76
4.20 Skyrmion 47 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.21 Barrel-shaped domain wall . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.22 Toroid withW = 5 lines . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.23 Observed winding numbers . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.24 Hopfion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
4.25 Comparison of antiskyrmion magnetization orientations . . . . . . . 83
4.26 Worm 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.27 Worm 32 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
14
CHAPTER I
BACKGROUND
Magnetic Bubbles
Stable, circular magnetic domains, generally dubbed magnetic bubbles, have
been a topic of interest for over 60 years. Originally observed in orthoferrites, per-
ovskites, and various oxides[7–10], they were soon engineered into thin films by
Andrew Bobeck at Bell Laboratories to create "bubble memory," a promising tech-
nology at the time for non-volatile data storage due to their stability and relative
ease of movement[11]. Tape storage was slow and physically large; magnetic-core
memory, while fast, was limited in capacity and prohibitively expensive[12], and
spinning hard drives were also still quite large and complex (the IBM 2305[13],
for example, introduced in 1970, filled the equivalent of 4 racks to house 20MB).
Bubble memory required no moving parts and comparatively little power to flip
bits[14]. While the investigation of wider use of bubble memory continued[15–19],
hard drive and, more recently, solid state drive performance advanced at a faster
pace.
Concurrently, superconductor research[20–22], often focused on UPt3 and vari-
ants, observed the topology of magnet bubbles called magnetic skyrmions. Orig-
inally, "skyrmions" referred to stable soliton solutions of a classical field theory
model for nucleons proposed by British physicist Tony Skyrme in 1961[23, 24].
The moniker of this robust, particle-like entity has since been extended to objects
elsewhere, including superconductors[25–28] and magnetic materials[29–34].
In the 90’s skyrmions were found in a number of thin films[35–37] and have
since grown in interest for the budding field of spintronics[38–42], where magnetic
15
Figure 1.1. A schematic representation of a Bloch skyrmion (top left), a Néel
skyrmion (top center), and an antiskyrmion (top right). Projections
on the unit sphere are shown on the bottom. Figure reproduced
from "Skyrmions and Antiskyrmions in Quasi-Two-Dimensional
Magnets" by Kovalev, Alexey A. and Sandhoefner, Shane in Front.
Phys., 27 September 2018, used under CC BY.
domains instead of electric charges are moved with a fraction of the current re-
quired for conventional electronics[3, 43, 44].
Phenomenological Description of Magnetic Skyrmions
There is some disagreement in the literature about what truly constitutes a
magnetic skyrmion: some insist on a strictly mathematical definition, requiring
integer values for the differential equation shown in Equation 1.3 and a specific
topology induced by a particular Hamiltonian (described below in Dzyaloshinskii-
Moriya interaction (DMI)), while others argue that it is the texture itself–small, sta-
ble, axisymmetric, whirling, easily moved, particle-like–that matters. This work ad-
heres to the more practical-natured second description while still calculating values
for the equation.
16
The two simplest magnetic skyrmions shapes, as well as a texture called an
antiskyrmion, are shown schematically in Figure 1.1. Both skyrmions and an-
tiskyrmions are more or less circular, with diameters from several 10’s to a few
hundred nanometers. Skyrmions are typically characterized in three dimensions as
cylindrical[32, 45–47].
In both cases, the z component of the magnetization vector flips from the center
of the texture to the outside (up to down or down to up). Skyrmions are most of-
ten labelled as Bloch, where the magnetization changes direction in the plane of the
domain wall, or Néel, where the magnetization changes direction perpendicular to
the plane of the domain wall. Figure 1.2 shows a schematic of these basic configu-
rations.
Skyrmions and antiskyrmions can be described mathematically by parameteriz-
ing a fixed-spin model. In general,
m̂(r) = (cosΦ(r) sinΘ(r), sinΦ(r) sinΘ(r), cosΘ(r)) , (1.1)
where m̂ is the normalized local magnetization vector in the magnetic material.
For skyrmions with cylindrical symmetry, Equation 1.1 can be expressed in spheri-
cal coordinates as
m̂ = (cosΦ(ϕ) sinΘ(ρ), sinΦ(ϕ) sinΘ(ρ), cosΘ(ρ)) , (1.2)
with in and out of plane components Θ(ρ) and Φ(ϕ), where ρ and ϕ are the radial
distance and azimuthal angles.
17
Magnetic skyrmions are characterized by a non-zero integer skyrmion num-
ber[48],
∫
1
Nsk = ∫ m̂ ·∫(∂xm̂× ∂ym̂) dxdy4π
1 0 2π dΘ(ρ) dΦ(ϕ)
= dρ dϕ · · sinΘ(ρ). (1.3)
4π ∞ 0 dρ dϕ
The proof of this topological invariant is nontrivial and beyond the scope of this
work; see [23]. Nsk = 1 is a skyrmion (Figure 1.3, Figure 1.4), Nsk = −1 is an anti-
skyrmion (Figure 1.5), and Nsk = 0 is a bubble with trivial topology (Figure 1.6).
The texture’s rotation sense is generally characterized by the chirality α.
Four specific values indicate Bloch and Néel textures: α = π is a clockwise Bloch
2
skyrmion, α = 3π is an anti-clockwise Bloch skyrmion (as in Figure 1.3); and α = 0
2
is an inward-pointing Néel skyrmion, ±π is an outward-pointing Néel skyrmion
(as in Figure 1.4). This definition of α follows the convention of Chen et al[49],
namely, the angle between the magnetization in the domain wall and the direction
of increasingMz .
18
(a) Bloch domain (b)Néel domain
Figure 1.2. Through magnetic domain walls, the magnetization direction
changes; in the images above, it transitions from up on the left to
down on the right. The two simplest transition mechanisms are in a
plane parallel to the ends or orthogonal to those planes. (a) For Bloch
walls, the magnetization direction changes parallel to the plane of
the domain wall. (b) For Néel walls, the magnetization direction is
perpendicular to the plane of the domain wall.
19
Figure 1.3. Schematics of Bloch skyrmions with Nsk = 1. The top images have
α = −π2 and are clockwise/right-rotating; the bottom have α = π2
and are anti-clockwise/left-rotating. The inset at the bottom right of
the zoomed images shows the direction of the in-plane magnetiza-
tion.
20
Figure 1.4. Schematics of Néel skyrmions with Nsk = 1. The top images have
α = π with field lines pointing radially outward; the bottom have
α = 0 with lines pointing inward.
21
Figure 1.5. Schematics of antiskyrmions with Nsk = −1. The top images have
α = 0, while the bottom have α = π2 , effectively the first images
rotated 90°.
22
Figure 1.6. Schematics of a bubble with Nsk = 0. Field lines simply flow around
the bubble, entering and exiting in the same direction.
23
Stabilization Mechanisms
There are a number of mechanisms responsible for magnetic skyrmion forma-
tion[48] that are discussed here.
Dzyaloshinskii-Moriya interaction (DMI)
Spin-canting in non-centrosymmetric magnetic crystals, called DMI after its
discoverers, was the first mechanism suggested for magnetic skyrmion creation[50].
Moriya predicted that neighboring electrons in these crystals would have as much
as 90% of their spins leaned into the a axis of the crystals with only ∼10% pointed
into in the c axis. In 2007 Bode et al found that DMI was responsible for "direc-
tional non-collinear magnetic structure[s]" at surfaces[51]. The great majority of
magnetic skyrmion research since has focused on DMI-stabilized textures[52].
DMI has been found to stabilize skyrmions in a number of materials, including
MnSi[32, 34, 53–55] and FeGe[56–61], described by the DMI Hamiltonian,
∑
HDMI = D · Si × Sj , (1.4)
⟨i,j⟩
whereD is the material-dependent interaction vector and Si and Sj are nearest
neighbor spins. The net effect is to cause nearest neighbor spins to cant, tilting rela-
tive to each other[62]. It is significant to note that the energy of this effect is gener-
ally on the order of 1 × 10−2 to 1 × 10−5 that of the energy of the Heisenberg Hamil-
tonian, ∑
HH = −J Ŝi · Ŝj , (1.5)
⟨i,j⟩
where J is the positive-valued coupling constant and the hats indicate unit length
spins, which tends to align the spins to minimize energy.
24
Frustrated exchange interactions
Magnetic frustration in a triangular lattice has been studied theoretically using a
three term Heisenberg Hamiltonian,
∑ ∑ ∑
HF = −J1 Ŝi · Ŝj − J2,3 Ŝi · Ŝj −H Si,z, (1.6)
⟨i,j⟩ ⟨⟨i,j⟩⟩ i
where the middle term is for second nearest neighbors, the third term describes
spin interactions with a magnetic field H in the z direction, and again all J values
are positive. This has been shown to stabilize both skyrmions and antiskymions in
a superposition of three wave vectors called the triple-q state[63–65].
Four-spin exchange interaction
A monolayer hexagonal Fe lattice on an Ir substrate has been shown via spin-
polarized scanning tunneling microscopy (SP-STM) to stabilize skyrmions due to
electrons shared across 4 adjacent lattice sites[66–68].
Long range dipolar interaction
Dipolar exchange in magnetic thin-films prefers an in-plane magnetization.
When one film is overlaid with a dissimilar film, strains within the layers as the lat-
tices struggle to align lead to an effect conceptually the same as DMI, called interfa-
cial DMI. When multiple layers are stacked on top of each other, and even more so
when heavy metal layers such as Pt or Ir are interspersed[69–71], the effects of the
spin canting add up and create perpendicular magnetic anistropy (PMA)[72]. PMA
prefers out-of-plane magnetization. The short range dipolar exchange energy com-
petes with the PMA[73], stabilizing skyrmions in defined parameter space[74, 75].
25
The dipolar Hamiltonian takes the form
∑ (gµ )2B
Hdip = [3 (r̂ij · Si) (r̂3 ij · Sj)− Si · Sj ] , (1.7)
i≠ rj ij
where g is the dimensionless magnetic moment, µB is the Bohr magneton, and r is
the separation between spins Si and Sj .
Characterizing heterostructures of this sort is the primary subject of this work.
Recently the emergence of skyrmions and other magnetic domains in Fe/Gd
multilayer thin films has been studied extensively[76–80], as these materials show
great promise for spintronic applications[81–83].
Engineering Skyrmions
Our collaborators at the Center for Memory and Recording Research (CMRR)
at UC San Diego, Eric Fullerton and Sergio Montoya, have fabricated Fe/Gd multi-
layer thin films using sputter deposition on Si[84]. Samples of these films were
studied by two previous members of the McMorran group.
Jordan Chess[85] produced a number of software tools for modeling skyrmion
structures (one of which was used to produce the simulated images above). He also
used Lorentz-TEM (LTEM) and scanning electron microscopy with polarization
analysis (SEMPA) to examine labyrinth domains in the supplied samples. LTEM
and micromagnetic simulations were used to examine skyrmions in the samples
(no skyrmions were found in the samples imaged by SEMPA). Three items partic-
ularly relevant to this work from his research[76]: a preference (95%) for a single
chirality in a sample which should not be significantly influenced by DMI[49, 86];
a bimodal chirality distribution fairly sharply peaked at α = π and 3π ; and micro-
2 2
magnetic simulations predicting Néel closure caps at the top layer of the skyrmion.
26
Alice Greenberg[87] used LTEM and scanning transmission electron mi-
croscopy holograph (STEMH) to examine a Landau domain in a permalloy square
and skyrmions in an Fe/Gd sample. Results of relevance from her work include:
micromagnetic simulations suggesting the skyrmions in Fe/Gd might, in fact,
be hopfions, a 3 dimensional object with a more complex topology than the 2
dimensional skyrmions typically described; and discussion about the shortcomings
of using LTEM to attempt to image these objects.
Nanomagnetic Microscopy Techniques
Skyrmions can be characterized in many ways, including direct imaging. Sev-
eral magnetic microscopy techniques exist.
Much of the skyrmion characterization work to date has used 2 dimensional
imaging technologies to capture magnetization information, such as TEM[78, 88–
90] and X-ray scattering[91–94].
TEM produces images by passing electrons (300 keV is typical) through the bulk
of a sample; see schematic in Figure 1.7. Operated in Lorentz mode, in which the
objective lens is turned off, eliminating any remanent fields at the sample, the elec-
trons in the beam interact with the components of magnetic field perpendicular to
the beam’s path present in the sample via the Lorentz force, FL = −e(v ×B), act-
ing on the electrons’ paths. The resulting deflection is revealed when the TEM is
adjusted to image the electron beam in a plane beyond the sample. One significant
limitation of this approach is that only the components of the magnetic field per-
pendicular to the beam cause electron deflection, only revealing Bloch components
and effectively rendering Néel components invisible. The net image shows the inte-
grated magnetic field. As such, in cases where there is a mirror symmetry between
27
Figure 1.7. Schematic of LTEM contrast formation. Electrons (blue lines)
passing through the sample plane (gray rectangle) are deflected by
in-plane magnetization (circled X and dot) by the Lorentz force,
creating light and dark regions at the bottom. Perpendicular magne-
tization (red up and down arrows) has no effect on the electron beam
path. Figure courtesy of Jordan Chess.
the top and bottom of a sample, any Lorentz force influence on the beam at the top
of the sample will be undone by the influence at the bottom.
Lorentz TEM provides 2 dimensional projection images of the magnetic field,
but fully understanding the nature of the magnetization, particularly at domain
walls, requires a full 3 dimensional analysis. There are a handful of microscopy
technologies used for capturing 3 dimensional magnetization data; several are de-
scribed below.
Magneto-optical Kerr Effect Microscopy (MOKE)[95, 96] exploits the fact that
light reflected from a surface with magnetization can undergo polarization and in-
tensity changes. These effects can be reversed in software to determine the surface
magnetization. Being a light-based technique, spatial resolution of ∼200 nm[97] is
not quite sufficient for the features examined in this work.
Similar to MOKE, X-ray Magnetic Circular Dichroism in Photoemission Elec-
tron Microscopy (XMCD-PEEM)[98, 99] is another light-based approach. In this
case, however, X-rays are used as the source, emitting electrons from the sample
28
being imaged. The resolution is still limited (∼50 nm[100]), again limiting the utility
of this instrument.
Spin Polarized Low Energy Electron Microscopy (SPLEEM)[101–103] is a rare
method that uses very low energy electrons, ∼1 eV, to image magnetic surfaces at
fairly high resolution (∼10 nm). SPLEEM uses a spin-polarized source and relies on
the exchange interaction between the incident polarized electrons S and surface
polarized electronsM with a strength proportional to S ·M [104]. This technique
typically involves imaging the sample from opposite angles and subtracting the
image pair, and as such is generally limited to imaging crystalline structures with
strong crystal orientations.
Based on a technique first described by Koike and Hayakawa in 1984[105],
Scanning Electron Microscopy with Spin Polarization (SEMPA, also spin
SEM)[105–108] modifies a typical SEM for probing with additional components
added to analyze the the polarization of secondary Auger electrons. At the Uni-
versity of Oregon we have a truly unique instrument, which we simply refer to as
SEMPA, of this type. The microscope was built at NIST[109] in the early 2000’s and
donated to the University when the program there was retired.
In this work, skyrmion data from SEMPA and LTEM are analyzed in an effort
to answer the question: what is the 3 dimensional structure of these skyrmions?
The results strongly suggest the reality is more complicated than a uniform cylin-
drical domain wall through the bulk.
Dissertation Outline
This dissertation covers the revival of a unique magnetic characterization in-
strument, the SEMPA, a new software framework for processing SEMPA’s outputs,
and the application of these tools to the analysis of magnetic textures in Fe/Gd
multilayer heterostructures.
29
Chapter I lays out the motivation for this work, along with the history and
theoretical underpinnings of the research. Chapter II gives an overview of the
SEMPA, including a brief description of how it functions and the myriad capa-
bilities it offers. Chapter III covers the methodology used for analysis, including a
walk through of the main software steps. Chapter IV contains the results of using
SEMPA to characterize an Fe/Gd multilayer sample fabricated by our collaborators
along with discussion of those results. Finally, Chapter V provides a summary of
the outcomes with a look to the future of the material as well as the abilities of the
SEMPA.
There are two appendices: Appendix A provides a link to the most recent code
base for the pyNISTView software, while Appendix B reproduces SEMPA: The
Missing Manual, also available as a stand-alone document for users of the SEMPA.
30
CHAPTER II
SEMPA
Theory of Operation
Spin polarized secondary electrons were first observed experimentally with an
SEM in 1984 by Koike and Hayakawa using a cold field emission source and a high
energy (100 keV) Mott detector[110]. The same year, Penn and Apell found that the
number of spin polarized secondaries emitted is greatest at low energies[111]. Fig-
ure 2.1 shows the number and polarity distributions of secondary electrons for typ-
ical ferromagnets from 0 eV to 30 eV.
The SEMPA is a heavily modified JEOL JAMP-7830F Auger microprobe with
a thermally assisted field emission source. This base platform was chosen because
SEMPA, an inherently inefficient technique, requires ultra high vacuum (UHV) and
clean surfaces. Auger analysis (electron energy spectroscopy for characterizing ma-
terial composition) has the same requirements, and the JEOL is capable of a UHV
environment of 1 × 10−8 Pa. Two Mott detectors are attached to a purpose-built
column to perform polarization analysis.
The primary electron beam incident on the sample will penetrate to some depth
(Figure 2.2(a)), dependent on beam energy, sample material, and angle of incidence
(Figure 2.2(b)). The incoming beam has sufficient energy to ionize atoms in the
sample, scattering a number of electrons (called secondary electrons) and x-rays
from within a tear-drop shaped region called the interaction volume. Lower energy
Auger electrons are collected and analyzed.
31
Figure 2.1. Typical energy distributions of the number (N) and polarization
(P) of secondary electrons measured for ferromagnetic materials.
The polarization distribution is enhanced at low energies due to
spin-dependent filtering of low energy secondary electrons[112].
SEMPA uses an extraction lens to capture low energy (∼10 eV) Auger
electrons whose polarization is recorded by a Mott detector. Figure
reproduced with permission from "Scanning electron microscopy
with polarization analysis (SEMPA) and its applications" by John
Unguris in Experimental Methods in the Physical Sciences. Vol. 36.
Elsevier, 2001, pp. 167–XVI.
32
(a)Normal interaction volume. The (b) Effect of tilting the sample relative to the electron
incoming primary electron beam beam on the interaction volume. Figure reproduced
(yellow, top) penetrates the sample with permission from "Spatial Resolution in ACOM
to an energy and material depen- - What Will Come After EBSD" by R. Schwarzer in
dent depth, causing a number of Microscopy Today 16(1):34-37 (2008).
reactions. Figure reproduced from
"Electron–matter interaction volume
and types of signal generated" at
Wikimedia Commons (CC BY-SA).
Figure 2.2. Electron beam interaction volumes
The Auger-Meitner Effect
The Auger-Meitner effect[113] occurs when an incoming electron of sufficient
energy scatters an electron from an inner shell of an atom, creating a hole. This
hole can be filled by an electron from a higher orbital. If the binding energy of the
lower orbit is considerably less than the transition energy of the lowered electron,
the excess energy can cause another higher shell electron to be ejected. These en-
ergies are relatively low, on the order of 50 eV to 3,000 eV[114], leading to a short
mean free path when the process occurs within a material. The electrons that break
free from the sample are thus most likely from the top several layers (on the order
of 1 nm[109]) of the sample. Tilting the sample (Figure 2.2(b)) puts more of the in-
teraction volume near the surface, which is preferable for Auger analysis, where the
goal is to characterize the very top layer of the sample.
33
Mott Scattering
Similar to Rutherford scattering except using electrons instead of alpha parti-
cles, Mott scattering occurs when electrons penetrate the nucleus of a high z atom
(typically gold since it does not oxidize and can be sputtered into a thin film as is
the case with SEMPA) and inelastically scatter off the Coulomb potential of the nu-
cleus. Spin-orbit coupling between the electrons and the nucleus splits the trajecto-
ries of the electrons depending on their spin state. The cross-section for scattering
takes the form[115]
σ(Θ) = I(Θ)[1 + S(Θ)P · n], (2.1)
where Θ is the angular distribution of the scattering profile, S is the Sherman func-
tion, a characteristic of the detector design, P is the beam polarization, and n is
normal to the detector plane.
The Microscope
A schematic model of the SEMPA is shown in Figure 2.3. Probe electrons inci-
dent on a tilted magnetic sample eject secondary electrons from the top nanometer
or so of the surface[116]. These electrons scatter from the sample in all directions.
There are a number of additional features that have been added to the micro-
scope, described both below and in Appendix B, making it a robust and diverse tool
for analyzing magnetic textures.
Plasma Gun
A tectra GenV plasma source sits atop the sample exchange chamber. Config-
ured to use Ar gas (though at NIST it was connected to a manifold allowing other
34
Figure 2.3. Schematic of SEMPA. After milling the Sample surface (tilted green
line, bottom center) clean with the Ion Gun (bottom left), spin–
polarized secondaries (pink shape traveling up and to the right from
the sample) scattered by the incident electron beam (pink shape
traveling down from the Electron Source at the top) are guided into
the SEMPA column (L-shaped part on the right). These electrons are
then Mott scattered by a gold target (top right where the secondary
beam ends for In Plane and bottom right in the same area for Out of
Plane) before being counted by a cloverleaf–shaped array of sensors
(gray blocks above and below Electron Spin Polarization Analyzers).
Depending on the configuration, either In Plane (x− y, top right) or
Out of Plane with redundancy (x− z, bottom left) spins are exam-
ined. Figure reproduced with permission from "Scanning electron
microscopy with polarization analysis (SEMPA) and its applications"
by John Unguris in Experimental Methods in the Physical Sciences.
Vol. 36. Elsevier, 2001, pp. 167–XVI.
35
gases to be used), it can gently mill samples with energies in the range 50 eV to
2,000 eV. This can be particularly useful for thin and/or delicate samples.
Electromagnet
A custom twin coil electromagnet has been added to the rear of the instrument,
next to the parking chamber. This permits subjecting samples to magnetic flux up
to 1 T at normal incidence and a range of tilts. The polarity of the field can be re-
versed by exchanging current supply leads.
Molecular Beam Epitaxy (MBE)
The main analyzing chamber is equipped with MBE deposition equipment for
sputtering Fe, Co, or Ni onto samples. Sputter rates for Fe are 0.19 nmmin−1 to
0.27 nmmin−1. MBE can be used to enhance the magnetic signal of weak samples
by applying a thin film of a ferromagnetic material to the surface.
Multi-pin Sample Holder
The stage has been configured with a 7 pin feed-through connector, providing a
facility for a number of active experiments. A simple example of this is the Faraday
cup sample holder that has an electrically isolated chamber wired to one pin while
the top isolated aperture plate is wired to another. This allows precise measure-
ments of beam width and current at the sample plane.
Manipulator Arm
A Kleindiek MM3A micromanipulator arm is positioned to reach the sample
holder in the main analyzing chamber. Samples can be moved around in situ using
a PlayStation® controller on the main console.
36
A comprehensive manual covering all of the standard aspects of the SEMPA can
be found in Appendix B.
SEMPA Column
A schematic of the SEMPA optics is shown in Figure 2.4 and a cutaway of the
assembly in Figure 2.5.
The tip of the SEMPA column in the main chamber is the extraction lens, placed
about 1 cm from the sample and DC biased to optimize collection of the most spin–
polarized Auger electrons at around 10 eV. The input lens placement also helps to
capture the electrons before any external fields can significantly affect their spin.
Once through the extraction lens, a series of Einzel lenses, apertures, and de-
flectors focus, align, and zoom the beam up and right to the dark block (the quarter
sphere) in the center of the column. A unique feature of this microscope is the in-
clusion of a voltage-adjustable quarter sphere electrostatic mirror in the center of
the SEMPA column that can deflect the electron beam 90° without affecting spin
polarization, thus permitting mapping the x− y and x− z magnetization vector
components and enabling capture of all three vector components of surface mag-
netization. This directs the beam to one of two detectors: it either allows the elec-
trons to pass to the Straight Through (ST) detector (at the top right of Figure 2.4,
labelled "In Plane"), which capturesMx andMy information, or guides electrons via
the electrostatic mirror to the right-angle (RA) detector (at the bottom right of Fig-
ure 2.4, labelled "Out of Plane"), which capturesMx andMz information.
37
Figure 2.4. Schematic of the SEMPA optics. The electron beam is focused to a
minimum waist at aperture A1 by lenses E1, E2, and E3 and centered
by deflector D1. Lenses E4, E5, and E6, along with deflector D2,
handle similar duty for aperture A2, effectively the beam defining
aperture. Lenses E5, E6, and E7 focus the beam at aperture A3,
providing the object for the quarter sphere. When E8, E9, and E10
are at the same 1,500 V potential, the beam passes straight through;
reducing the voltage at E8 and E9 to 340 V defects the beam 90°.
Lenses E11, E12, and E13 (or E15, E16, and E17 for right angle op-
eration) zoom the beam, with deflectors D5 and D6 (D7 and D8 for
right angle) control angle and position at the Au foil target, extreme
right[106]. After scattering off the target, V1 and V2 are biased to
guide the beam into Grid 1 and Grid 2 which accelerate the beam
into the multichannel plates, MCPs. Electrons striking the MCPs
cause a cascade of secondaries, maximizing the signal into the an-
odes. Image courtesy of NIST.
38
Figure 2.5. Cutaway view of the SEMPA. The input optic is bottom left; the
quarter sphere is at the cross; the ST components top right and
the RA components top left. The copper parts are predominantly
the Einzel lenses. The gray rectangles top left and right are the E-T
detectors and the blue boxes are the signal processors for the Mott
detectors. The gold target positioning knobs protrude down from
the top halves of the column; they are both set to 90°, the position
that would be used for cleaning the gold targets. Image courtesy of
NIST.
39
At this point the electrons Mott scatter off the Au foil at the end of the column.
They are subsequently decelerated to a landing energy, 150 eV, determined during
design that maximizes spin-dependent backscattering.
Each of the 4-channel Mott detectors in the SEMPA (schematic shown in Fig-
ure 2.6) resolve up, down, left, and right spin-polarizations, with net polarization
1 N+ −N−
P = , (2.2)
S N+ +N−
where N is the number of electrons hitting one of the channels (for Figure 2.6,
electrons hitting C and D would be N+ and N−, respectively, forMx, while elec-
trons hitting B and A would be N+ and N− forMy) and S is the Sherman function,
a system-specific measure of the dependence of the electron scattering to its spin-
orbit coupling, calculated for the SEMPA to be 0.5. As magnetization is propor-
tional to spin polarization, data collected from the x− y plane is combined with
x− z data in software to reproduce the 3D surface magnetization.
An example of the output forMx andMy observed on an Fe whisker is shown
in Figure 2.7. A single SEMPA scan provides two of the vector components of sur-
face magnetization as well as a normal SEM image. It is highly surface-sensitive,
providing a direct measurement of the spin polarization of the top few layers of
the material imaged down to a spatial resolution of ∼10 nm. One limitation, due to
the low energy Auger electrons that are ultimately analyzed, is that images must be
taken at remanence.
40
Figure 2.6. Schematic of the Mott detector. As oriented, channeltrons C and D
will collect positive and negative x polarizations, while B and A will
collect positive and negative y polarizations. There are 2 of these on
the instrument: one for straight through operation capturing x and
y as shown, the other for right angle operation capturing x and z
(with the axes shown rotated 90° about x). Figure reproduced with
permission from "Scanning electron microscopy with polarization
analysis (SEMPA) and its applications" by John Unguris in Experi-
mental Methods in the Physical Sciences. Vol. 36. Elsevier, 2001, pp.
167–XVI.
41
(a)Mx (b)My
Figure 2.7. Magnetization of an Fe whisker. (a) Magnetization in the x direction:
white is right, black is left. (b) Magnetization in the y direction:
white is up, black is down. Anti-clockwise from the top left, this
shows 3 domains: down, right, up.
42
CHAPTER III
METHODOLOGY
Image capture on the SEMPA involves 4 steps: sample preparation, SEMPA
alignment, SEMPA capture, and data processing.
Sample Preparation
Several Fe0.34/Gd0.40 x 120 layers samples (subscripts representing approximate
thickness in nm) were created by our collaborators at CMRR; see Figure 3.1. Each
sample consists of multiple thin layers deposited on a silicon substrate capped by a
thin, nonmagnetic metal layer to prevent oxidation of the structure. The films sup-
port a large variety of magnetic domain configurations[85, 87]. To view skyrmions,
the films are magnetically prepared using the following recipe.
Sample preparation after receipt begins by nucleating skyrmions. This is done
by first tilting a sample to 45° and subjecting it to a 400mT magnetic flux density,
saturating the sample’s magnetism. This can be done in situ in either the TEM or
the SEMPA, using the objective lens of the TEM in Lorentz mode when possible
or with the SEMPA electromagnet for general samples. The field is then brought
down to 0 and the sample tilted to 90° relative to the applied field. The field is
then increased to 185mT, creating the skyrmionic domains. The field is again
decreased to 0. An interesting and somewhat unusual feature of this sample is that
the skyrmions remain at remanence (zero applied magnetic field).
The skyrmion nucleation process above takes the sample from a state of satu-
ration (one large domain covering the sample, all in the direction of the external
field) to a stripe state as seen in Figure 3.2. Raising the field back up to ∼185mT
43
Figure 3.1. Fe/Gd sample layout. A Ta base layer was deposited on a SiO sub-
strate. 120 layers of Fe alternating with Gd were sputtered on top of
the base layer and capped off with Pt.
(this value is sensitive and varies somewhat from sample to sample) "pinches off"
sections of the stripes, initially into worms, then further into scattered circular do-
mains, then a fairly ordered hexagonal circular domain lattice.
The sample is next loaded into the SEMPA sample exchange chamber and shut-
tled through the vacuum system into the main analyzing chamber. Standard SEM
imaging at 25 kV accelerating voltage is used to find a location of interest on the
sample and initial surface composition using Auger analysis.
The Auger scan is run across a defined range of energies selected for the ele-
ments in the sample. Each element produces a set of characteristic peaks in the en-
ergy spectrum. If the scan result indicates the presence of a contaminant (often O
or C) or a capping layer (Pt or Ta for the Fe/Gd samples examined), Ar ion milling
is performed to clean the sample surface. To minimize damage to the magnetic
44
Figure 3.2. SEMPA image of surface magnetization showing stripe domains
created by saturating the sample and then reducing the external field
to 0. The circled 1 is a round texture identified by the feature finding
algorithm described at the end of this chapter.
45
structures in the Fe/Gd samples, Ar ions are used at a relatively low milling en-
ergy of 500 eV. The scan/mill process is repeated until the sample was sufficiently
cleaned. Figure 3.3 shows before and after milling one of the Pt capped Fe/Gd sam-
ples. Milling through the Pt took ∼5 minutes.
The sample is then tilted to 52° in anticipation of SEMPA analysis and eucentric
height validated. SEMPA imaging is performed when the main chamber pressure
has fallen into the 1 × 10−7 Pa or (preferably) lower range.
SEMPA Alignment
For each sample to be imaged, the next step in the process is aligning the
SEMPA optics, depicted schematically in Figure 2.4; a cutaway view of the entire
instrument is shown in Figure 2.5.
The input optics (labeled E1 on the schematic) are brought up to 1,500 V to ex-
tract and collect secondary electrons while the base voltage for the rest of the elec-
trostatic lenses are increased to 1,700 V (each lens/deflector sits at a value some-
what above or below this base voltage, depending on its purpose). The gold target
used for Mott scattering is first removed from the electron path by turning the po-
sitioning knob on the SEMPA column to 45°, allowing the electron beam in the col-
umn to bypass the target and hit the Everhart-Thornley (E-T) detector at the end
of the column. The high voltage power supply in the SEMPA rack is turned on and
the output increased to 2,500 V. The E-T selector switch responsible for choosing
the detector used for imaging is turned to the Straight Through (ST) setting.
At this point the SEMPA optics are ready for calibration. The SEM image is
zoomed out to 50x and contrast increased until the screen begins to show an im-
age. The first three deflectors (D1–D3 on the schematic) are adjusted until the im-
age brightened in the center while zooming in to 10,000x magnification and de-
46
(a) Auger plot of Pt.
(b) Auger plot of Fe.
Figure 3.3. Auger analysis of a Pt-capped Fe/Gd thin film showing energy
ranges for O, Fe, Gd, and Pt from left to right. The top plots are
counts at given energies (N(E)); the bottom plots are dN(E). (a) Ini-
tially a several nm capping layer was present, evident by the 2 peaks
at energies around 1,950 eV and 2,020 eV. The leftmost plot shows
no signal for O, while the center two show no signal for Fe or Gd,
respectively. (b) After several minutes, the Pt was milled through, ev-
ident in the flat rightmost plot, and both Fe and Gd started to show.
Interestingly, apparently some O was embedded into the sample
during the original deposition.
47
creasing contrast to avoid image saturation. This scans the area from which elec-
trons are extracted. When the SEMPA E-T detector is used to view the image sig-
nal, the beam looks like a spotlight that can be moved around. The lenses are ad-
justed until the brightest area is centered.
With the optics aligned, the ST target positioning knob is turned to 0°, inserting
the gold target in line with the electron beam. The preceding steps are repeated for
the Right Angle (RA) section.
The high voltage power supply for the E-T is switched off to eliminate interfer-
ence with the Mott detector.
The power supplies for the blue boxes are turned on, and the multi-channel
plate and anode voltages turned on and increased to 1,500 V and 1,600 V, respec-
tively.
SEMPA Capture
Data collection with the SEMPA begins with configuring the capture software,
specifying, among other things, resolution and dwell time. Dwell time, the length
of time that the electron beam illuminates a single spot before rastering to the next,
can be adjusted depending on the strength of the sample’s signal: imaging pure Fe
for relatively large (several µm) domains can be accomplished with dwell times
on the order of 4ms, while the much fainter skyrmion signal from Fe/Gd requires
20ms or more. As such, a single capture can take some time (just under 22 minutes
in this case for 256 x 256 images).
The electron beam is blanked and SEMPA zeros measured, during which dark
counts are recorded for the Mott detectors. The imaging parameters are set as de-
scribed above and images captured.
The SEMPA software outputs several NetCDF4 format files for each imaging
run. For the ST detector, those include redundant contrast files (with suffixes _-
48
Figure 3.4. Output of SEMPA image capture. The leftmost images are typical
SEM scans. The four images on the right show magnetization, with
magnetization strength in that axis from none (black) to maximum
(white).
ix.sempa and _iy.sempa and one file each for magnetic contrast in x and Y, namely
_mx.sempa and _my.sempa. Imaging parameters and image data are recorded in
each file.
The above steps were repeated for the RA detector with outputs _ix2.sempa,
_iz.sempa, _mx2.sempa, and _mz.sempa.
A representative output is shown in Figure 3.4.
49
Data Processing
The software originally written to process the output of SEMPA, written in
IDL, did not come with the microscope to the University. The decision was made
to develop new software for reading, displaying, and analyzing the data in a more
modern, open language. The code described herein was written in python using
Jupyter to facilitate easier maintenance and enhancements over time. It was also
designed to automate as much of preparing and analyzing the images as possible,
minimizing user input.
The .sempa file extension output by the SEMPA software is not a standard ex-
tension. The string "CDF" was found when examining the binary files from the
command line. This led to finding the Network Common Data Form (NetCDF),
a format typically used for Geographic Information System (GIS) applications,
but appropriate (if not terribly common) for general array-based scientific data.
This project took advantage of netCDF software developed by UCAR/Unidata
(http://doi.org/10.5065/D6H70CW6).
With that determined, preparing the collected data for analysis includes the 5
steps described below.
Importing
For 3D reconstruction, files from bothMx,My andMx,Mz runs are imported.
Using the netCDF libraries, image scale and magnification parameters are extracted
from the files, along with the image data. The images are stored in 2D arrays.
Registration and scaling
Imperfect alignment and sample drift cause the field of view for consecutively
collected ST and RA image captures to be slightly different. As a result, when col-
50
Figure 3.5. Image registration homography applied. Several points from Inten-
sity 1 and Intensity 2 are identified, then the homography to rotate,
skew, and scale Intensity 1 is applied to all images. The images are
then cropped to remove empty data (shown in the image on the right
in black along the left and top sides).
lecting x, y, and z data, requiring two scans, the images must be properly regis-
tered so that the vector components at each pixel are correctly aligned. An oriented
BRIEF (ORB) keypoint detector[117] in OpenCV is used to attempt to identify key
points in the x intensity images from both the ST and RA detectors (_ix and _ix2).
The most likely candidate points between the two files are selected.
A homography between the two files is calculated, then used to warp the per-
spective of the first intensity file. The perspectives for the x and y magnetization
images (_mx and _my) are also warped. This operation creates some black pixels
around the edges where rotation has occurred, as seen top left in Figure 3.5.
Using a binarized version of the warped image, the extreme outer contours are
found in each quadrant. The most distant points in each quadrant (the 4 new cor-
ners) are found and used to crop all images.
A light Gaussian filter (σ = 2) is applied to each image to smooth out some of
the noise.
51
Finally, any data points in the images greater than 3σ from the mean are limited
to ±3σ from the mean to help reduce shot noise and surface defect artifacts. This is
an effective value determined by evaluating a number of images.
Removing offsets and rescaling
Though it would seem reasonable that the data in the magnetization images
would be in the range -1 to 1 (based on Equation 2.2), the values are typically some-
where between -4 and 6 with a range of around 0.2. The three files do not have the
same center point. The differing offsets need to be removed and the data normal-
ized.
All the magnetization image arrays are first individually centered around 0 to
provide a seed for fitting. A least squares fit is then applied to square root of all
three added in quadrature as magnetization magnitude,
√
|M | = M2 +M2 +M2x y z , (3.1)
is constant in a given material. The values in each array are then normalized with
respect to each other, such that at all pointsM 2 = 1.
Calculating angles
With the image data normalized, the azimuthal angles of the magnetization vec-
tor in the x− y plane are calculated for each pixel as
−1 Myϕ = tan , (3.2)
Mx
and the polar z angles are calculated as
−1 Mxyθ = tan , (3.3)
Mz
52
Figure 3.6. Magnitudes of the 3D magnetization vector (|M(x, y)|, uniform,
left) compared to 2D magnitude of the in-plane components of the
magnetization vector (|Mxy(x, y)|, showing deviations in z, right).
The black and white speckles on the 3D image are rounding errors
introduced by the normalization algorithm: all values are exactly
(or very close to) 1. The black areas in the plot on the left indicate
no magnetization in the x− y plane (only in z), while white areas
indicate magnetization only in the x− y plane; gray regions have a
mix.
whereMxy is the magnitude of the x− y component of magnetization.
The resultant array values are thereafter verified by comparing 3D vs. 2D mag-
nitudes of the magnetization, an example shown in Figure 3.6. Individual magneti-
zation components are shown in Figure 3.7, providing additional verification that
the data is not systematically flawed in one dimension.
53
Figure 3.7. Normalized magnetization magnitudes in each plane
54
Finding features
The next step is selecting features of interest. For the purposes of this work,
skyrmions are the primary feature being sought, with the goal of determining their
chirality and characterizing their shape at the surface of the material. The achieve
this, the feature identification techniques described below are applied.
A binarized version of the θ array is morphologically closed using a 3 pixel wide
circular kernel with OpenCV. All closed contours are then found in the image. Any
features that reach the edges are discarded in case relevant texture data is out of
frame.
The contours are next fit to an ellipse to get width and height. The difference
between width and height is compared to a threshold factor multiplied by the max-
imum of the two, effectively eliminating overly long ovals. The area within the con-
tour is compared to the expected area of a circle with a radius half of the average
of the width and height summed, eliminating, for example, concave contours with
similar widths and heights. An example of features located is shown in Figure 3.8.
For this work, circular features were the target, and the above methods were
executed. The selection process could be modified for other shapes of interest.
55
Figure 3.8. In this less than ideal sample, only two roughly circular features
were identified by the algorithm, both on the right side of the image.
56
CHAPTER IV
ZOOLOGY OF TEXTURES IN FE/GD
The raw SEMPA data used for the analysis presented in this chapter were col-
lected by John Unguris and BenMcMorran at NIST in the last quarter of 2017. The
raw LTEM data used for α and size analysis were captured by Will Parker using an
FEI Titan in 2022. They have not previously been published.
Results
Figure 4.1 shows an overview of the ∼2.3 µm2 region of Fe0.34/Gd0.40 x 120 lay-
ers that was analyzed. The background colors showMz , with blue regions indi-
cating magnetization into the page (θ = π) and red regions out of the page (θ = 0);
white areas have no z component of magnetization (θ = π ). All contours found are
2
identified and indexed numerically. The outline of each feature indicates average
chirality: orange features are clockwise while purple features are anti-clockwise.
Figure 4.2 shows an overview of the same region, this time with background
colors indicatingMxy , the in-plane components of magnetization. The inset top
right shows the direction ofMxy at each point. Darker areas have a larger z com-
ponent to the magnetization.
For the investigation of bubble features, each of the contours from Figure 4.1
was evaluated for shape to exclude those that clearly were not skyrmion candidates.
The remaining candidate group of features is shown in Figure 4.3.
57
Figure 4.1. Overview of features detected in Fe0.34/Gd0.40 x 120 layers . The
image colors represent magnetization in z, while the numbered
feature outlines indicate chirality: orange is clockwise, purple is
anti-clockwise.
58
Figure 4.2. Overview of features detected in Fe0.34/Gd0.40 x 120 layers . The
image colors represent magnetization in the x− y plane. Darker
areas are regions where the majority of the magnetization vector is
in z.
59
Figure 4.3. Overview of the circular features detected in Fe0.34/Gd0.40 x 120 lay-
ers . Excluding feature 41, these were used for the following analysis.
60
Skyrmions
Select skyrmions are shown in Figures 4.4 to 4.7. These met the identification
criteria of having N πsk ≈ 1 with a standard deviation σ < , ensuring they generally6
followed a consistently-oriented path around the feature. 26 features were identi-
fied as skyrmions in this dataset.
The leftmost image for each figure shows the skyrmion and its immediate sur-
rounding. Background colors indicateMz , with blue into the page, red out of the
page, and white indicatingMz = 0. The arrows show the in-plane direction of mag-
netization,Mxy. Each image includes an index that can be cross-referenced to the
overview in Figure 4.3.
The outermost contours indicate the domain wall at the surface of the sample.
For all four images, p = 1 as the magnetization points out of the page in the center.
Orange contours are clockwise domains with π < α < π; purple contours are anti-
2
clockwise with π < α < 3π .
2
Plotted to the right of each image is αmeasured at each point along 3 separate
paths: the outermost (orange or purple) as well as 1 of the radial distance out from
3
the maximum enclosedMz , represented by the black dot, and 2 of the radial dis-3
tance out to the outermost. These paths are represented by the yellow and green
dotted rings, respectively. In all cases, α is calculated by walking clockwise around
the contour (starting on the outer contour at the red dot, proceeding around to yel-
low, green, blue, and back to red). The averages at each radial distance are plotted
with dashed lines. Errors for the contour are within the light blue shaded areas.
The image locations in the overall scan are displayed, along with feature dimen-
sions. Additional statistics are presented, including average α along all 3 paths, av-
erage path angles ϕ, and winding number n.
61
Figure 4.4. Clockwise skyrmion. The values for α plotted on the right show the
values and averages 1 and 23 3 of the radial distance from the point of
maximumMz (measured at the yellow and green dotted locations
in the image on the right), as well as at the contour. The values of α
for the contour are plotted along the x axis starting from the red dot,
proceeding clockwise through yellow, green, and blue.
Figure 4.5. Anti-clockwise skyrmion.
62
Figure 4.6. Clockwise skyrmion.
Figure 4.7. Anti-clockwise skyrmion.
63
Antiskyrmions
The two antiskyrmions that were identified are shown in Figures 4.8 and 4.9.
These met the identification criteria of having Nsk ≈ −1 and a generally circular
shape (enclosed area within ±30% of the area of a circle with r = width×height ).
2
Image color schemes follow the same pattern as above, though the contour
color only represents the numeric average for α.
Trivial Bubbles
The two topologically trivial bubbles that were found are shown in Figures 4.10
and 4.11. These met the identification criteria of having Nsk ≈ 0 and a generally cir-
cular shape (enclosed area within ±30% of the area of a circle with r = width×height ).
2
As for antiskyrmions, image color schemes follow the same pattern as for
skyrmions, with the contour color again only representing the numeric average for
α.
Worms
Two worm domains that were identified are shown in Figures 4.12 and 4.13.
Worm domains are longer in one dimension than the other, generally created when
labyrinth domains (much longer) get pinched off during skyrmion nucleation.
As above, image color schemes follow the same pattern as for skyrmions, with
the contour color again only representing the numeric average for α.
64
Figure 4.8. Antiskyrmion with magnetization lines flowing in at 12 and 4
o’clock, out at 2 and 9 o’clock.
Figure 4.9. Antiskyrmion with magnetization lines flowing in at 11 and 5
o’clock, out at 2 and 8 o’clock.
65
Figure 4.10. Trivial bubble with lines flowing from 1 to 7 o’clock.
Figure 4.11. Trivial bubble with lines flowing from 4 to 9 o’clock and 1 o’clock.
66
Figure 4.12. Worm domain roughly 290 nm long.
Figure 4.13. Worm domain roughly 375 nm long.
67
Feature Groups
Two regions of skyrmion groups are shown in Figures 4.14 and 4.15.
While the color scheme follows the above images, the numbering here is based
on the indices of circular contours only.
Figure 4.14 maps to features in the top left of Figure 4.3 (namely, 12, 13, 19, 20,
21, 25, 26, 27, and 33).
Figure 4.15 maps to features just left of bottom center of Figure 4.3 (namely, 42,
43, 45, 47, 48, 49, 53, 54, 57, and 60).
Distribution of Magnetization Angles Across Features
Figure 4.16 shows the distribution of average α values calculated from both
SEMPA and LTEM data. As defined in Background, α is a measure of deviation of
the magnetization vector along a path (the domain wall in this case) from the direc-
tion of increasingMz .
The blue bars represent α for the skyrmions identified in the sample analyzed
by SEMPA. The averages along with error for this bimodal distribution are shown
at the top, outlined in blue. The orange bars represent α for the skyrmions iden-
tified in a similar sample analyzed on the TEM. The averages along with error for
this bimodal distribution are overlaid, outlined in orange. The green outlines rep-
resent α for the all features identified in the sample analyzed by SEMPA. The aver-
ages along with error for this bimodal distribution are included, outlined in green.
68
Figure 4.14. A cluster of skyrmions adjacent to each other.
69
Figure 4.15. Another cluster of skyrmions adjacent to each other.
70
Figure 4.16. Distribution of α calculated from SEMPA data (blue bars for circu-
lar features, green outline for all detected features) and LTEM data
(orange bars).
71
Figure 4.17. Comparison of circular feature sizes from SEMPA data (blue) and
LTEM data (orange). The uniform spacing in both cases is due to
the square grid of pixels used to calculate these values.
Comparison of Skyrmion Sizes from SEMPA and LTEM
Figure 4.17 shows the distribution of average feature dimensions measured
from both SEMPA and LTEM data.
SEMPA measurements are shown in blue with error bars; TEM measurements
are shown in orange with errors. The resolution of the TEM is roughly 10x that of
the SEMPA, made apparent by the significantly smaller errors for the TEM data.
Figure 4.18 shows an overview of the ∼3.9 µm2 region of Fe0.34/Gd0.40 x 120 lay-
ers that was analyzed with the TEM.
72
Figure 4.18. Overview of the circular features detected in Fe0.34/Gd0.40 x 120
layers imaged with LTEM. Inset color wheel shows the direction of
magnetization in the domain walls. The numbering is suppressed
to avoid obscuring the features.
73
Discussion
Shapes Observed
As shown in Figures 4.1 to 4.3, a wide variety of textures were found in the
sample, including skyrmions, antiskyrmions, trivial bubbles, and worm domains.
A feature (indeed, the most significant feature for this research) of Fe0.34/Gd0.40 x
120 layers is that the magnetic textures created after the second exposure to an ex-
ternal field (400mT, 0mT, 185mT) remain at remanence, stabilized by the behavior
described in the section Long range dipolar interaction. This feature, appealing for
many spintronic applications, is not the case in many materials.
Skyrmions
Of 61 total features detected, 31 were circular, with 20 of those likely candidates
for skyrmions. Several are shown above in Figures 4.4 to 4.7. α between π and π
2
rotate clockwise in an increasingly outward radial direction as α increases; α be-
tween 3π and π rotate anti-clockwise with an increasingly outward radial direction
2
as α decreases.
Figure 4.16 shows that LTEM data is effectively evenly split between clockwise
and anti-clockwise Bloch walls, as expected, with any other magnetization appar-
ently purely perpendicular to the plane of the sample.
SEMPA data, however, shows a different story. At the surface, magnetization is
only purely in the z direction somewhere near the center of the domains. The mag-
netization develops x and y components, creating hybrid Néel/Bloch walls. The
value of α generally decreased/increased for anti-clockwise/clockwise skyrmions
as the radial distance from maximumMz increased as shown for several skyrmions
in Figure 4.19, with a specific example shown in Figure 4.20.
74
This is particularly interesting when combined with the dimensions of
skyrmions measured with LTEM versus SEMPA, shown above in Figure 4.17. The
skyrmion diameters, defined whereMz = 0, measured at the surface with SEMPA
are, on average, 135 nm ± 4.5 nm, whereas the diameter of the projected image
with LTEM are, on average, 165 nm ± 1.0 nm.
A simulation of this shape at several z heights using Jordan Chess’s skyrmion
plotting model is shown in Figure 4.21. The domain wall edges move outward radi-
ally as the plot moves closer to the center of the bulk.
This suggests that the magnetization continues to rotate in plane while moving
out radially and down into the bulk until the domain wall is purely Bloch-like at
α = ±π , creating more of a barrel shape than the cylinder normally suggested. This
2
corroborates Alice Greenberg’s simulations and observations[87].
Continuing to analyze the 3D shape of the magnetization, the angle α (that is,
relative to a line pointing radially to the center of the toroid as would be the case if
magnetization were pointing up in the center of the toroid) at the topmost surface
of a toroid withW from 1 to 10 were measured. An example of this setup including
2 lines withW = 5 is shown in Figure 4.22.
A plot of the extrema of measured α from SEMPA along with a cubic spline
fit of the toroid wrapping angles is shown in Figure 4.23. From this it would ap-
pear that the lines of magnetization wrap around the barrel shape between 2 and 8
times. Depending on how mathematically rigorous one is, some researchers would
argue this is taking a form at least superficially like a hopfion[118, 119], shown
schematically in Figure 4.24.
75
Figure 4.19. α values for several skyrmions at the film surface. The first four
plots are clockwise skyrmions. In each case, the value of α de-
creases toward π2 as the radial distance increases, indicating that the
direction of magnetization is transitioning from more Néel-like to
more Bloch-like. The bottom 2 plots are anti-clockwise skyrmions,
where the value of α increases toward 3π2 , also becoming more
Bloch-like as the radius increases.
76
Figure 4.20. Average α for this skyrmion decreases from 2.68 along the path of
the yellow dots 13 of the distance from maximumMz at the black
dot to the orange domain wall to 2.63 along the path of green dots
at 23 of the distance to the wall to 2.56 along the domain wall.
77
(a)
(b)
(c)
Figure 4.21. Simulated barrel-shaped domain wall: (a) hybrid Néel/Bloch walls
at the surface; (b) more Bloch-like walls partway into the bulk;
(c) and purely Bloch walls at the center of the bulk. Red is out of
the page, blue into the page, with white indicatingMz = 0. White
arrows showMxy .
78
Figure 4.22. Toroid wrapped with 2 lines withW = 5. Red regions on the wrap-
ping contour are on top of the toroid, while blue regions are on the
bottom. α ≈ 2.675 at each point of maximum z.
79
Figure 4.23. Measured α for several winding numbers around a toroid. Orange
X’s indicate observed α values fitted to a simulated toroid with
similar dimensions; the horizontal green lines show the extrema of
α observed in the sample. The vertical purple lines show the range
of observed winding numbers. The blue dots are integer winding
values calculated from the simulated toroid.
80
Figure 4.24. Schematic of a hopfion, an object with closed lines wrapping
around both θ and ϕ before meeting back up where they started.
Figure reproduced with permission from “Beyond skyrmions:
Review and perspectives of alternative magnetic quasiparticles”
by Börge Göbel, Ingrid Mertig, and Oleg A Tretiakov in Physics
Reports 895 (2021), p. 4.
81
Antiskyrmions
Of the 31 circular features found, 2 were antiskyrmions, shown above in Fig-
ures 4.8 and 4.9.
The α plots confirm the expected double rotation of the magnetization vectors
while traversing around the domain wall.
Two unexpected observations were made with these textures: an asymmetry
in the width of radial flow in versus out, and the point of maximumMz is not cen-
tered within the area circumscribed by the domain wall.
Figure 4.25 shows a histogram of α values along the contour paths of each of
the antiskyrmions. Regions where π < α < 3π , shaded blue, indicate an outward
2 2
direction of magnetization relative to the contour center. Regions where 0 < α < π
2
and 3π < α < 2π, shaded red, indicate an inward direction of magnetization.
2
In both cases the majority of radially outward magnetization were found point-
ing just outside of the Bloch sections of the domain walls. Roughly twice as much
of the circumference of the antiskyrmions had magnetization pointing outward as
inward.
In Figure 4.8, the point of maximumMz was found to be ∼60% of the distance
from the centroid to the domain wall; for Figure 4.9, the point of maximumMz was
around ∼50% of the distance fro the centroid to the domain wall. In both cases the
slope of decreasingMz was shallowest toward the center, dropping more rapidly
out toward the closest wall.
Both of these observations were of interest to our fabrication collaborators.
While not yet examined in detail, the working theory is that some asymmetry
within the bulk stabilizes the off-center maximumMz . The same mechanism might
also be responsible for the uneven flow directions, as the largest outflow regions
are opposite the highMz locations.
82
(a)Magnetization orientations for Anti- (b)Magnetization orientations for Anti-
skyrmion 9. skyrmion 12.
Figure 4.25. Comparison of antiskyrmion magnetization orientations.
83
Bubbles and Worms
Of the 31 circular features found, 2 were topologically trivial bubbles, shown
above in Figures 4.10 and 4.11. The remaining 30 features were worms, 2 of which
are shown above in Figures 4.12 and 4.13.
Figure 4.11 differs from the standard theoretical model of bubbles shown in
Figure 1.6 in that the walls to either side of the inward pointing magnetization are
predominantly Néel in character, flowing significantly outward for the majority of
the circumference of the feature. Figure 4.10 is closer to the model but also shows
considerable outward pointing magnetization along the sides adjacent the inward
pointing magnetization.
The worm domains predominantly appear to be precursors to two or more
skyrmions (had a slightly higher external field been applied to the sample: this has
been observed real-time while nucleating skyrmions using LTEM). This is par-
ticularly evident in Figure 4.13, where a second local maximum inMz is visible
to the top right of the texture. Several of the other worm domains visible in Fig-
ure 4.1 have narrower waists than ends, indicating where they would separate into
skyrmions, antiskyrmions, or trivial bubbles in a higher external field.
Interestingly, many of the worm domains showed αmeasurements very sim-
ilar to skyrmions. When imaged using a TEM, worms, like skyrmions, generally
show Bloch walls with no contrast in the center. In Figures 4.26 and 4.27 the do-
main walls are again a hybrid Néel/Bloch structure. As above, the lines of magneti-
zation apparently wrap around the domain wall in a more complex shape than the
TEM would indicate.
Skyrmions are, ultimately, shrunken worm domains. Their sizes, though some-
what variable, place a lower limit on the size of a worm domain.
84
Figure 4.26. A worm with a hybrid Néel/Bloch domain wall configuration.
Figure 4.27. Another hybrid Néel/Bloch worm.
85
Feature Groups
As expected[25, 78], the groups of skyrmions (including the 2 antiskyrmions in
the first figure) shown in Figures 4.14 and 4.15 are fairly evenly spaced.
In the overview shown in Figure 4.1, it is apparent on a larger scale that sev-
eral worm domains near circular features were on the verge of pinching off and
forming more hexagonal arrays: features 15 and 16, as well as 18 and 24 are good
examples.
The chirality of the textures tend to group, suggesting these smaller features
originated from larger stripe or labyrinth domains. In the case of Fe/Gd, there is no
preference for chirality expected. Recent research by another member of the group
has shown via TEM that the inclusion of Pt and/or Ir layers can force a preferred
chirality on the film as a whole.
86
CHAPTER V
CONCLUSION
Dissertation Results
This dissertation has detailed the background and observation of a variety of
magnetic structures found in Fe/Gd multilayer thin films through a combination
of technologies, particularly SEMPA and Lorentz TEM. This analysis required the
development of new software tools to process the data output by the SEMPA, as
well as detect and characterize textures found in the films.
The results of this analysis revealed a zoology of chiral spin textures found in
Fe0.34/Gd0.40 x 120 layers including skyrmions, antiskyrmions, trivial bubbles, and
worms. When coupled with 2D projection images from LTEM, the ability of the
SEMPA to capture spin polarization (and, by extension, magnetization) information
enabled the discovery of a morphology with greater complexity than previously ex-
pected for skyrmions and worms, providing a more complete answer to the ques-
tion: what is the 3 dimensional structure of these objects?
For Skyrmions and Bubbles and Worms, this analysis found that the Bloch walls
observed in Lorentz TEM do not extend to the surface, where instead Néel walls
were observed. The gradual transition from Néel to Bloch indicates a non-trivial
topology in three dimensions, not two; something closer to a twisted vortex state
(perhaps hopfion, depending on definition) for the magnetization, wrapping around
barrel-shaped walls not in circles but helices, much like a Slinky coiled back on it-
self. The description of skyrmions as "cylindrical Bloch wall domains" is demon-
strably overly simplistic: hybrid Néel/Bloch domains that appear to extend some
distance into the bulk are a more accurate description.
87
For Antiskyrmions, this research found two asymmetries, in location of maxi-
mumMz in the texture as well as the portion of the texture pointing in versus out.
Implications and the Future
Much remains to be learned about the characteristics of skyrmions and re-
lated textures. A complicating factor is that several of these characteristics are tun-
able; e.g., while Fe0.34/Gd0.40 x 120 layers stabilizes skyrmions at room temperature,
Fe0.34/Gd0.40 x 80 layers does not. Similarly, Fe/Gd has no preferred chirality, but
Fe/Gd/Pt/Ir does.
A fundamental part of spintronics is creating and moving skyrmions. Both of
these activities are influenced by the shape of the skyrmions being manipulated.
The SEMPA holds the potential to answer these questions and provide insight
about different magnetic structures in materials of varied sizes and textures.
Through this work the instrument was made operational and its capabilities
exercised to analyze Fe/Gd, increasing our knowledge of the nature of skyrmions.
88
APPENDICES
89
APPENDIX A
PYNISTVIEW
The code developed by the author for this work is available on Github.
90
APPENDIX B
SEMPAMANUAL
91
SEMPA: The Missing Manual
Rich Moraski
January 5, 2023
92
Contents
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
2 Concept of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
3 Components of the Instrument . . . . . . . . . . . . . . . . . . . . . . . . 103
3.1 Sample Exchange Chamber . . . . . . . . . . . . . . . . . . . . . . . . 103
3.2 Parking Chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
3.3 Electron Column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.3.1 Field emission gun (FEG) . . . . . . . . . . . . . . . . . . . . . 107
3.3.2 Sputter ion pumps (SIPs) . . . . . . . . . . . . . . . . . . . . . 109
3.3.3 Condenser and objective lenses . . . . . . . . . . . . . . . . . 111
3.3.4 Aperture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.3.5 Specimen Analyzing Chamber . . . . . . . . . . . . . . . . . . 112
3.3.6 Stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3.4 Auger electron spectrometer (AES) . . . . . . . . . . . . . . . . . . . . 115
3.5 Micro Ion Etching Device (MIED) . . . . . . . . . . . . . . . . . . . . 116
3.6 Everhart-Thornley detector (E-T detector) . . . . . . . . . . . . . . . 117
3.7 Electromagnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
3.8 Plasma Gun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
3.9 Console . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.10 Printer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.11 Auxiliary pump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
3.12 Turbo pump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
3.13 High tension tank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.14 Vacuum system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
3.15 SEMPA Column . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
3.16 SEMPA Rack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
3.17 Computers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
3.18 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
3.18.1 Auger Master . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
3.18.2 sempa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
93
4 Alignments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
4.1 Field emission gun (FEG) . . . . . . . . . . . . . . . . . . . . . . . . . . 134
4.2 Aperture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.3 Micro Ion Etching Device (MIED) . . . . . . . . . . . . . . . . . . . . 136
5 Startup and Shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.1 Auger Master . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.2 Venting for maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.3 Full shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6 Standard Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
6.1 Sample exchange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
6.2 Parking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6.3 Main chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6.4 Eucentric height . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6.5 Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
7 Auger Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
8 Ar Ion Milling with the MIED . . . . . . . . . . . . . . . . . . . . . . . . 148
9 Ar Plasma Milling in the Sample Exchange Chamber . . . . . . . . . . 150
10 Electromagnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
11 Magnetic Domain Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
11.1 Prepare sample . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
11.2 Turn on SEMPA electronics . . . . . . . . . . . . . . . . . . . . . . . . 154
11.3 Imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
12 Maintenance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
12.1 Manual mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
12.2 Bakeouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
12.3 TSP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
12.4 Cleaning the gold targets . . . . . . . . . . . . . . . . . . . . . . . . . . 161
13 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
13.1 Shutdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
13.2 Auger Master will not connect on launch . . . . . . . . . . . . . . . . 163
13.3 No SEM image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
13.4 The MIED will not mill . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
13.5 V2 or V3 will not open . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
13.6 Pulsing SEM image . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
13.7 SEM image darkens and brightens but does not zoom or focus . . . 167
13.8 Accelerating voltage stuck at one value . . . . . . . . . . . . . . . . . . 168
94
13.9 FEG emission current limited to ∼1.9 A despite higher command . . 169
13.10No FEG emission current at normal operating condition . . . . . . . 171
13.11Cannot get SEMPA electrons to E-T detector or targets . . . . . . . . 171
Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
95
List of Figures
2.1 Schematic of SEMPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
2.2 Mott detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
3.1 The exchange chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
3.2 The parking chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
3.3 The electron column . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.4 Schematic of the electron column . . . . . . . . . . . . . . . . . . . . . 106
3.5 field emission gun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.6 FEG pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.7 Magnified image of the FEG from the side . . . . . . . . . . . . . . . . 108
3.8 SIPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.9 SIP currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.10 Aperture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
3.11 The main chamber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.12 Main pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.13 A view looking up from inside the main chamber . . . . . . . . . . . 113
3.14 The stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
3.15 The AES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
3.16 Micro Ion Etching Device (MIED) . . . . . . . . . . . . . . . . . . . . 116
3.17 The circuitry for the MIED . . . . . . . . . . . . . . . . . . . . . . . . 116
3.18 Light pipes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
3.19 E-T detector selector switch . . . . . . . . . . . . . . . . . . . . . . . . 117
3.20 The electromagnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
3.21 Plasma gun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
3.22 The console . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
3.23 The printer for SEM images . . . . . . . . . . . . . . . . . . . . . . . . 121
3.24 The auxiliary pump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
3.25 The turbo pump controls at the back of the room . . . . . . . . . . . 123
3.26 The high tension tank . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
3.27 Schematic of the vacuum system . . . . . . . . . . . . . . . . . . . . . 125
3.28 SEMPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
3.29 The first sections of the SEMPA column . . . . . . . . . . . . . . . . . 127
3.30 More SEMPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
3.31 Still more SEMPA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
3.32 Schematic of the ST optics . . . . . . . . . . . . . . . . . . . . . . . . . 128
96
3.33 A cutaway of the complete ST column . . . . . . . . . . . . . . . . . . 129
3.34 The SEMPA rack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
3.35 The knob box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
3.36 Pinout for the knob box . . . . . . . . . . . . . . . . . . . . . . . . . . 131
3.37 The computers for the SEMPA . . . . . . . . . . . . . . . . . . . . . . 132
3.38 Auger Master observation windows . . . . . . . . . . . . . . . . . . . 133
5.1 Main window for Auger Master . . . . . . . . . . . . . . . . . . . . . 138
5.2 Windows for starting and monitoring the FEG . . . . . . . . . . . . . 139
5.3 Windows for adjusting the FEG startup and runtime parameters . . 139
5.4 Vacuum control panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.1 SEM controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
7.1 A sample spectrum analysis showing O and Fe peaks . . . . . . . . . 146
7.2 ROI configuration window . . . . . . . . . . . . . . . . . . . . . . . . 147
9.1 Plasma controls and glow . . . . . . . . . . . . . . . . . . . . . . . . . 151
11.1 The main window for the sempa software . . . . . . . . . . . . . . . . 157
11.2 Sample images being captured in the X-Y plane . . . . . . . . . . . . . 158
12.1 The SEMPA column wrapped and ready for bakeout . . . . . . . . . 160
13.1 MIED fuses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
13.2 New vs. used light pipes . . . . . . . . . . . . . . . . . . . . . . . . . . 167
13.3 Scan fuses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
13.4 High tension tank interior . . . . . . . . . . . . . . . . . . . . . . . . . 169
13.5 Short between SEMPA lenses . . . . . . . . . . . . . . . . . . . . . . . 172
13.6 Nanoammeters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
13.7 Individual nanoammeter . . . . . . . . . . . . . . . . . . . . . . . . . . 174
97
List of Tables
4.1 Starting values for MIED channel settings . . . . . . . . . . . . . . . . 136
5.1 Venting options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.1 MIED etching rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
10.1 Measured fields for given currents . . . . . . . . . . . . . . . . . . . . 153
10.2 Required currents for desired fields . . . . . . . . . . . . . . . . . . . . 153
13.1 Shutdown codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
98
Chapter 1
Introduction
This document is meant to be a guide for anyone using the Scanning Electron
Microscopy with Polarization Analysis (SEMPA), whether for standard scanning
electron microscope (SEM) imaging or more complex surface magnetization imag-
ing.
The SEMPA instrument is based on a modified JEOL JAMP-7830F Field Emis-
sion Auger Microprobe. The base instrument, which includes a standard SEM,
specializes in capturing Auger spectra of materials. The nature of Auger analysis
requires relatively high vacuum levels. Being able to determine surface composition
in ultra high vacuum provides a good starting platform for surface magnetization
analysis.
Some JAMP-7830F specs:
• Accelerating voltage from 500 eV to 25 keV
• Probe current from 10 pA to 100 nA
• Main chamber pressure down to 3 × 10−8 Pa
• Sample sizes up to 12mm wide by 5mm thick
• Stage movement of ±10mm in X and Y, 6mm in Z
The instrument was built at NIST in Germantown, MD, in the early 2000’s. It
was the third iteration of such a device.
In 2017, John Unguris, one of the original builders of the system, retired. The
SEMPA project was retired with him. Instead of putting the disassembled instru-
ment into storage, on January 24, 2018, the SEMPA was donated to the University
of Oregon. It has been located in the Micro Analytical Facility, Room 87, CAM-
COR, since.
99
Chapter 2
Concept of Operation
SEMPA is a heavily modified JEOL JAMP-7830F Auger microprobe. This base
platform was chosen as Auger analysis (electron energy spectroscopy for charac-
terizing material composition) is both necessary for ensuring a clean surface and
dependent on an ultra high vacuum system (1 × 10−8 Pa), both requirements of
SEMPA, an inherently inefficient technique.
A unique feature of this microscope is the inclusion of quarter sphere in the
center of the SEMPA column that permits viewing in X-Y and X-Z, thus allowing a
3 dimensional reconstruction of the surface magnetization. A cartoon model of the
instrument is shown in Figure 2.1.
Probe electrons incident on a tilted magnetic sample eject spin–polarized sec-
ondary electrons from the top nanometer or so of the surface. These electrons scat-
ter from the sample up and right through the column to one of two detectors: the
dark block (the quarter sphere) in the center of the column either allows the elec-
trons to pass to the straight through detector at the top right (labelled "In Plane"),
which captures X-Y information, or guides electrons via a voltage difference to the
right-angle detector at the bottom right (labelled "Out of Plane"), which captures X-
Z information. Each of the 4-channel Mott detectors, shown in Figure 2.2, resolve
up, down, left, and right spin-polarizations, with net polarization
1 N+ −N−
P = , (2.1)
S N+ +N−
where N is the number of electrons hitting one of the detectors and S is the Sher-
man function, a characteristic of the detector design. As magnetization is propor-
tional to spin polarization, data collected from the X-Y plane is combined with X-Z
data in software to reproduce the 3D surface magnetization.
100
Figure 2.1. Schematic of SEMPA. After milling the sample surface clean with
the ion gun, spin–polarized secondaries scattered by the incident
electron beam are guided into the SEMPA column. These electrons
are then Mott scattered by a gold target before being counted by a
cloverleaf–shaped array of sensors. Depending on the configuration,
either in-plane (X-Y) or out of plane with redundancy (X-Z) spins
are examined. Figure reproduced with permission from "Scanning
electron microscopy with polarization analysis (SEMPA) and its ap-
plications" by John Unguris in Experimental Methods in the Physical
Sciences. Vol. 36. Elsevier, 2001, pp. 167–XVI.
101
Figure 2.2. Cartoon of the Mott detector. There are 2 of these on the instru-
ment: one for straight through operation capturing X and Y as
shown, the other for right angle operation capturing X and Z. Figure
reproduced with permission from "Scanning electron microscopy
with polarization analysis (SEMPA) and its applications" by John
Unguris in Experimental Methods in the Physical Sciences. Vol. 36.
Elsevier, 2001, pp. 167–XVI.
102
Chapter 3
Components of the Instrument
There are many parts that make up the SEMPA, several factory parts from
JEOL, several custom designed at National Institute of Standards and Technology
(NIST).
3.1 Sample Exchange Chamber
Also called the 1st Airlock chamber, this is the starting point for observing sam-
ples in the instrument where samples are moved in and out of the vacuum. Pressing
the EXCH VENT button during normal operation will vent the chamber to atmo-
sphere, allowing the door to be opened and a sample holder inserted/removed: turn
the black knob with the MDC logo counterclockwise until it is loose, leaving it in
place until the pressure has been increased to 1 ATM: this ensures the door will not
fly open, which is generally not a real concern.
There are 2 arms in this chamber: one that moves the sample from the entry
point to laterally to the second, which moves the sample back to the parking cham-
ber.
The pressure for this chamber is monitored on the Pfieffer Vacuum SIngleGauge
in the rack farthest left. Normal values are in the 1 × 10−5 Pa range.
Figure 3.1. The exchange chamber
103
3.2 Parking Chamber
The parking chamber, located just left of the electron column, allows up to 4
sample holders to be maintained in vacuum.
There are 3 arms in this chamber: one that moves samples from the 2nd arm
from the sample chamber to either the parking holder or the main chamber; one
for the parking holder; and one that moves samples from the parking holder to the
electromagnet.
The pressure for this chamber is monitored on the GP 307 Vacuum Gauge Con-
troller in the rack farthest left. Normal values are in the 1 × 10−5 Pa range.
Figure 3.2. The parking chamber
104
3.3 Electron Column
This is where the electrons are extracted and accelerated. It is covered in mu-
metal to help shield against stray magnetic fields that might otherwise disrupt
Auger or SEMPA imaging. The gun is at the top, lenses and apertures in the mid-
dle, and the sample stage at the bottom.
The pressure for this chamber is monitored on the Ionivac IM520 in the 2nd
rack from the left. Normal values are in the 1 × 10−7 Pa range.
Figure 3.3. The electron column is the dark cylinder in the center of the instru-
ment
105
This is a schematic view of the column.
From top to bottom, the significant components are
• Field emission gun (FEG)
• Suppressor
• Extractor
• Gun alignment coils
• C1 stigmator coils
• C1 condenser lens
• C2 stigmator coils
• C2 condenser lens
• Probe current detector (PCD)
• Scan coils
• OL stigmator coils
• Gun shift coils
• OL objective lens
• OL alignment coils
• Stage
Figure 3.4. Schematic of
the electron
column
106
3.3.1 Field emission gun (FEG)
The FEG sits at the top of the column. It is a thermally-assisted Schottky field
emitter that provides the electrons that are accelerated down toward the sample.
A small tip is heated to ∼1,800 K while in a 3,000 V extraction field. The emitted
electrons are funneled down to the chamber aperture by the 300 V suppressor field.
The pressure for this chamber is monitored on the FEG Ionization Gauge at the
top of the 2nd rack from the left. Normal values are in the 1 × 10−7 Pa range.
(a) The FEG holder (b) The FEG viewed from above
Figure 3.5. field emission gun
107
Figure 3.6. The pressure in the FEG is shown on the top gauge; units are Pa
Replacement tips are available from Ted Pella. The part number is TFE174C
SE4; as of 2021, the price per tip was $5,550.
Figure 3.7. Magnified image of the FEG from the side
108
3.3.2 Sputter ion pumps (SIPs)
The sputter ion pumps (SIPs) are ion pumps meant to operate in high vacuum
(1 × 10−5 Pa or lower). There are 5 total: one for the specimen analyzing chamber
(SIP 1), one for the lenses (SIP 2), one for the parking chamber (SIP 3, underneath
the main instrument), and a pair for the FEG chamber (SIP 4).
These are consumable items.
Figure 3.8. 3 of the 5 SIPs are visible as grey boxes on the sides of the electron
column
109
Figure 3.9. The currents for SIP 1 and 4 are shown on the fourth and second
gauges, respectively
110
3.3.3 Condenser and objective lenses
These are located in the main column. They are responsible for shaping (con-
denser) and focusing (objective) the electron beam ultimately into a probe at the
sample. There are 2 condenser lenses and one objective lens in the SEMPA.
3.3.4 Aperture
The aperture is a small strip with 4 same-sized relatively large holes in it that
holds a thin film with 4 different-sized smaller holes. The aperture assists with
beam shaping and blocks electrons that stray too far from the desired beam path.
The aperture is controlled from Auger Master > Observation > Probe Con-
dition > OL > Aperture. 5 removes the strip from the beam path (no aperture);
4–1 are successively smaller apertures. A larger aperture permits more current at
the cost of reduced resolution; a smaller aperture allows a sharper image at a re-
duced current. For regular SEM imaging, ∼1 nA provides a very sharp image. This
might be too low a current for the rather inefficient SEMPA imaging, however.
(a) The aperture removed from the instrument (b) Close-up of the aperture holder; the aperture
strip is the dark insert at the top behind the
holes
Figure 3.10. Aperture
111
3.3.5 Specimen Analyzing Chamber
Located at the bottom of the electron column, this is where the sample resides
during imaging.
Figure 3.11. A view into the main chamber. The objective lens is the triangle at
the top; the SEMPA input optics are the copper colored bit right
center; and the Micro Ion Etching Device (MIED) is darker silver
cylinder to the right.
Figure 3.12. The pressure gauge for the main chamber; units are Pa
112
Figure 3.13. A view looking up from inside the main chamber up at the instru-
ment
113
3.3.6 Stage
The stage holds the sample holder and allows for it to be translated, raised, low-
ered, tilted, and rotated. The stage is grounded to the chassis. There are special
sample holders that allow for up to 7 electrical connections via a feed-through on
the left side of the stage housing. There is also a BNC connector on the side that
allows measuring the current exiting through the stage.
The smallest units in Auger Master are µm; e.g., x = 1.234 is 1.234mm.
Figure 3.14. The stage is visible 45° down and to the left of the sample holder
114
3.4 Auger electron spectrometer (AES)
The Auger detector is the original reason for this instrument. It uses low energy
secondary electrons, namely Auger electrons, that are scattered from very near the
surface of the sample (within the top 1 nm or so) to determine the composition of
the surface.
Figure 3.15. The AES is the hemispherical dome to the right, behind the main
column in the lab
115
3.5 Micro Ion Etching Device (MIED)
The MIED, or simply ion gun, uses ionized Ar gas to abrade the surface of the
sample, like a (much) gentler version of sandblasting. This is useful for removing
oxidation, capping layers, and other contaminants.
(a) The MIED removed from the instrument (b) The control panels for the MIED
Figure 3.16. Micro Ion Etching Device (MIED)
The two green fuses just right of center in Figure 3.17 tend to blow.
Figure 3.17. The circuitry for the MIED
116
3.6 Everhart-Thornley detector (E-T detector)
Everhart-Thornley detectors (E-T detectors) are the parts responsible for con-
structing the SEM contrast. As the SEM beam rasters across the sample, secondary
electrons are pulled toward the +10,000 V biased cap around the front of the E-T
detector. There they collide with a layer of phosphorus, where the photoelectric
effect causes photons to be released. These photons travel down a clear glass light
pipe where they hit a photomultiplier that amplifies the signal. The imaging cir-
cuitry descans the signal and builds the image shown on the screen.
These are consumable (though renewable) items: over time the phosphor coat-
ing will get blasted away by the incoming electrons.
(a) Light pipe removed from the instrument (b) Light pipe with the copper shield removed
Figure 3.18. Light pipes
The E-T detector switch, shown in Figure 3.19, selects between the "normal"
detector (for standard imaging) and the 2 detectors attached to the SEMPA column.
Figure 3.19. E-T detector selector switch
117
3.7 Electromagnet
The electromagnet allows for in-vacuum application of magnetic fields. It relies
on a separate current supply to generate a field. Table 10.1 shows the fields mea-
sured between the yoke and one of the coils at specified currents (the field increases
linearly with current); Table 10.2 extrapolates this to specific field strengths.
Figure 3.20. The electromagnet, shown here with 2 wooden sticks providing
space between the yoke and coil to measure the field strength
118
3.8 Plasma Gun
The plasma gun, located atop the sample exchange chamber, can be used to
clean samples with a low energy Ar plasma. It is currently configured to operate
on Ar but has been used in the past with other gases.
There is a pump and water container in the service chaise that must provide
cooling for the gun when in use. The pump must be started and stopped manually.
The water lines to the gun need to be removed during bakeouts.
(a) The plasma gun for gentle milling of the (b) The plasma cage
entire sample
Figure 3.21. Plasma gun
119
3.9 Console
This is the main work area. There is one keyboard and mouse shared across 3
computers. There are 6 monitors; starting top left and proceeding clockwise: cam-
era image, doorbell, bracket, 2 monitors for whisk, and the SEM image. The stage
and optics controls are also here, along with sample chamber light power supply,
Kleindiek arm controls, HT wobbler, E-T detector switch, printer, and external
scan controller.
The 4 main power switches and MIED controls are underneath on the left. The
bigger area underneath on the right holds
Figure 3.22. The console
3.10 Printer
The printer produces small (roughly 4 inches square) printouts of the SEM im-
age.
120
Figure 3.23. The printer for SEM images
121
3.11 Auxiliary pump
The Pfieffer pump in front of the sample exchange chamber is used to supple-
ment the turbo molecular pump (TMP) initially when pumping down. It displays
units in mbar.
Figure 3.24. The auxiliary pump
122
3.12 Turbo pump
The turbo pump is most effective at higher pressures (down to ∼1 × 10−5 Pa)
than the SIPs. The pump itself is under the microscope; the controls for it are be-
hind the instrument.
Figure 3.25. The turbo pump controls at the back of the room
123
3.13 High tension tank
The HT tank provides 30 kV to the FEG. The tank contains, among other things,
a Cockroft-Walton voltage multiplier bathed in SF6 to help minimize the risk of
arcing. There is a gauge on the side of the tank with a marking indicating the cor-
rect pressure for the gas. Care must be taken when working with SF6 as it can be
toxic.
Figure 3.26. The high tension tank that provides up to 30,000 V
124
3.14 Vacuum system
Below is a schematic view from the JEOL manual of the vacuum system show-
ing chamber, valves, and pumps.
Figure 3.27. Schematic of the vacuum system
125
3.15 SEMPA Column
This is the reason we’re here. The SEMPA column juts up at an angle from the
specimen analyzing chamber. It is designed to move in and out slightly (the input
optics need to be about 1 cm from the sample, but they need to be protected from
being hit while moving the sample); 2 aluminum spacers are used to hold it in place.
The biggest thing that makes the SEMPA unique is its ability to measure all 3
components of magnetization, thus allowing for reconstructing the 3D magnetiza-
tion at the surface of the sample. This is accomplished by having 2 detectors: one
straight through (ST) and one right angle (RA). In both cases gold targets at the ends
of the columns are used to Mott scatter the incoming polarized electrons to a 4
quadrant detector that measures either X and Y (ST) or X and Z (RA).
(a) The SEMPA column is the silver part diago- (b) The SEMPA rack on the right
nally up on the right
Figure 3.28. SEMPA
Inside the vacuum chambers are a number of electostatic lenses (the copper
tubes in Figure 3.29 are several) leading to the detectors at the ends. These are held
at different voltages controlled by the knob box (Figure 3.35a).
126
Figure 3.29. The first sections of the SEMPA column (E1–E7, see Figure 3.32
below)
In the center of the column is the switchyard which houses a quarter sphere
configured to either have the same voltage on all parts (for ST) or different voltages
on the inside versus outside (for RA). In the RA case, the electric part of the Lorentz
force redirects the electrons to the RA detector at the back of the instrument.
(a) An exploded view with ST to the right, RA (b) The quarter sphere used to redirect the beam
on top, the input optic on the left, and the to the RA detector (toward the top in this
switchyard on the bottom image)
Figure 3.30. More SEMPA
127
(a) The complete SEMPA assembly removed (b) The SEMPA input optic is the copper piece at
2 o’clock
Figure 3.31. Still more SEMPA
Figure 3.32 shows a schematic of the optics and Mott detectors.
Figure 3.32. Schematic of the ST optics
128
Figure 3.33. A cutaway of the complete ST column
129
3.16 SEMPA Rack
The rack shown in Figure 3.34 houses the electronics that control voltages
throughout the SEMPA. The top three instruments are meters intended to assist
with aligning the beam in the system; alignment by eye using the E-T detector de-
tector for each arm is easier and less error-prone.
Figure 3.34. The SEMPA rack
The knob box, shown in Figure 3.35, adjusts the voltage for the electrostatic
lenses (knobs labeled with E’s) and deflectors (knobs labeled with D’s). The first 4
lenses, namely E1-4, are the most frequently adjusted.
130
(a) The SEMPA knob box (b) The knob box circuitry
Figure 3.35. The knob box
4 cables from the knob box, namely O1-O4, connect to the 4 inputs on the
switchyard, namely I-IV. The pinouts are represented in Figure 3.36.
Figure 3.36. Pinout for the knob box
131
3.17 Computers
There are 3 computers used for SEMPA, shown in Figure 3.37. The computers
are laid out in the same orientation as the monitors.
The keyboard and mouse are shared across the machines; which machine is
controlled is handled by the small black switch usually found next to the keyboard.
The order is whisk > doorbell > bracket > unused.
doorbell (Windows 7 Professional, top center)
This machine is connected to the Internet. It has most of the SEMPA images
and documents in C:\sempa.
The main user account is sempa
bracket (CentOS 5.10, top right)
This machine hosts Auger Master. It is not connected to the Internet.
The main user account is aes.
whisk (Debian 6.0.10, bottom center and right)
This machine hosts the sempa software. It is not connected to the Internet.
The main user account is sempa.
Figure 3.37. The computers for the SEMPA
132
3.18 Software
3.18.1 Auger Master
This is the JEOL software for operating the SEM. Auger Master is launched
from the Auger Master icon on the desktop or in the Applications menu. The main
windows for controlling the SEM are shown in Figure 3.38.
(a) The probe condition win- (b) The sample manipulation (c) The image observation
dow in Auger Master window in Auger Master window in Auger Master
Figure 3.38. Auger Master observation windows
3.18.2 sempa
This is the NIST software for operating the SEMPA column. sempa is launched
from the command line. See Magnetic Domain Imaging for more details.
133
Chapter 4
Alignments
4.1 Field emission gun (FEG)
The thermally assisted FEG sits atop the electron column. Electrons that are
emitted from the tip (first thermally excited, then pulled out by the extraction volt-
age) need to travel in a well-defined path down to the sample. This process is docu-
mented in Technical Material TI-99052AP001 from JEOL, found in a white binder
in the SEMPA file cabinet.
As a first step, the FEG is physically aligned by turning the 4 hex head screws
as necessary to maximize current to the closed PCD. This should only need to be
done with the FEG is changed. Standing in front of the machine, the screw in front
on the left is the X+ adjuster; the front right is the Y+ adjuster (opposites on the
rear).
• Ensure that the manual valve (knob just above the aperture on the right side
of the main column) is fully open.
• Ensure the accelerating voltage is 25 kV.
• Verify the aperture is set in Auger Master to 5 (aperture removed from beam
path).
• Pull cable A3 from the back left corner of the console.
• Set all gun and CL alignment values to 0 in Auger Master > Probe Condi-
tion, as well as CL Coarse and Fine.
• Insert the PCD.
• Set the Auger Master > Probe Condition > MDA to PCD.
• Adjust the upper screws until the probe current is maximized. Opening the
probe current monitor can be helpful to see small trends when adjusting.
• Remove the PCD and observe a feature at ∼1300x.
134
• While changing the CL Coarse from 0 to ∼50 in Auger Master > Probe Con-
dition, adjust the upper screws as necessary to keep the feature from mov-
ing.
• Turn on the OL Wobbler in Auger Master > Probe Condition.
• Adjust the CL shift and tilt values until the image does not move.
• Set CL Coarse and Fine values to 0.
• Reconnect A3.
• While changing the CL Coarse from 0 to ∼30 in Auger Master > Probe Con-
dition, adjust the lower screws as necessary to keep the feature from moving.
• Change CL values from 0 to ∼400, zoom in and out, and adjust focus to verify
the feature does not move.
• Adjust the aperture as below.
This software part of this process (gun and CL tilt and shift) should be repeated
for each desired accelerating voltage.
4.2 Aperture
The aperture alignment might need to be adjusted for each aperture.
• Select aperture.
• Get an image if one doesn’t show immediately. This is as much art as science.
Keep an eye on the probe current to look for values in the 1 × 10−9 range (less
than that is no real current) while adjusting the objective aperture fine tuning
knobs. The current likely will start in the 1 × 10−12 A range.
• Once an image is acquired, zoom in to 3000x or higher if possible (this makes
astigmatism issues easier to see).
• Adjust focus to best possible, then tweak stigmators as needed.
• Turn on the HT Wobbler and adjust amplitude to something useable.
• Adjust the aperture knobs until the focus racking does not translate.
– The transverse adjustment knob affects horizontal motion of the wob-
bler; the axial knob affects vertical motion.
– It is generally possible to eliminate horizontal motion fairly easily.
135
– If the vertical motion persists up until signal cut-off, adjust to the edge
that shows least motion, then turn up the probe current/CL coarse ad-
justment. The image should brighten and the knob should be able to
turn farther before stopping the signal. Continue adjusting until the ver-
tical motion is eliminated.
• Zoom in to 10,000x+ and further tweak fine focus and stigmation. It might be
worthwhile to also clear the CL and OL lenses (in Auger Master > Observa-
tion > Probe Condition).
4.3 Micro Ion Etching Device (MIED)
The MIED is finicky on a good day (see The MIED will not mill). The alignment
procedure is documented in section 27.3 of the JEOL JAMP-7830F manual. The
values shown in Table 4.1 provide a good starting point for aligning the MIED.
Item X Y
Energy/Current 2000 200
Intensity/Focus 0 152
Beam Position -60 -12
Column Alignment -3 -3
Energy/Current 500 200
Intensity/Focus 0 152
Beam Position -62 -21
Column Alignment -3 -3
Energy/Current 200 90
Intensity/Focus 0 152
Beam Position -60 -12
Column Alignment -3 -3
Table 4.1. Starting values for MIED channel settings
136
Chapter 5
Startup and Shutdown
Starting the base instrument, assuming pressures are in reasonable ranges, is a
matter of turning on 4 switches on the rack on the bottom left of the console right
to left, bottom to top.
• JEOL HT
• ACCEL HT
• OPN POWER
• ACCEL VOLTAGE
Shutdowns of the console should happen in the opposite order: left to right, top
to bottom.
5.1 Auger Master
Auger Master can be started by double-clicking the icon on the desktop on
bracket. A splash screen will come up with the text Connecting to console....
Once that completes successfully (see Auger Master will not connect on launch in
case that does not happen), the main window shown in Figure 5.1 will appear.
137
Figure 5.1. Main window for Auger Master
To start the FEG, select Maintenance > FEG Control, shown in Figure 5.2a.
The Start button will begin the auto start process.
The progress of startup, as well as realtime condition monitoring, is achieved by
selecting Observation > FEG Monitor, shown in Figure 5.2b.
The initial conditions used for the gun startup are set by Maintenance > FEG
Start Up Tuning, shown in Figure 5.3a. Note the startup maximum gun pressure
is fixed at 5 × 10−7 Pa. The filament current should be selected to achieve ∼90 µA of
emission current once warmed up, keeping the filament current below the value
that will heat the tip beyond 1,800 K (maximum temperature for the health of the
FEG tip). At the time of this writing, that value is 2.41 A for the installed tip. Ex-
traction voltages are typically ∼3,000 V.
Once the FEG is running, filament current, suppressor voltage, and extraction
voltage can be modified by selecting Maintenance > FEG Control > Fine tun-
ing..., shown in Figure 5.3b. Suppressor voltage is typically left at 300 V.
138
(a) The FEG startup window (b) FEG monitor window
Figure 5.2. Windows for starting and monitoring the FEG
(a) The FEG startup tuning window for initial (b) FEG fine tuning window for runtime condi-
conditions tions
Figure 5.3. Windows for adjusting the FEG startup and runtime parameters
5.2 Venting for maintenance
When performing maintenance, often at least some part of the system will need
to be vented. The FEG should be turned off for this: Auger Master > Maintenance
> Shutdown FEG. This will take several seconds.
Close the manual valve located just above the aperture on the right side of the
main column. Remember to reopen the valve after the maintenance is complete.
Table 5.1 is reproduced from the JAMP-7830F manual, section 6.3.1d. Select
the appropriate vent button on the vacuum control panel (Figure 5.4). Pressing the
button again within 5 seconds will cancel the vent process. Once the venting be-
gins, AV1 will open (and the corresponding LED will illuminate); once the venting is
complete, AV1 will close and AV2 will open, remaining open throughout the mainte-
nance. At this point it is safe to open the evacuated parts of the system.
139
Button Chamber(s) Vent time
VENT Airlock 2 minutes
2EXC
VENT Airlock
CHAM Main chamber 5 minutes
VENT Electron lens chamber 2 minutes
eGUN
VENT Electron lens chamber
ALL Main chamber 7 minutes
Table 5.1. Venting options
Figure 5.4. Vacuum control panel
After venting, the instrument will automatically initiate a bakeout. How long
it should bake depends on what was done; 2 hours is typically the minimum. The
bake duration is set with the two ⇑ buttons in the BAKE OUT TIMER section of the
panel. The rest of the system should be baked as well to ensure all contaminants get
pumped out; see Bakeouts.
Press the button selected above to begin the pump down and bakeout.
5.3 Full shutdown
This procedure is useful when a power outage is anticipated, e.g. in the case of
building maintenance.
• Turn off the 4 switches listed above.
140
– ACCEL VOLTAGE
– OPN POWER
– ACCEL HT
– JEOL HT
• Close the manual valve.
• Turn off the ION GUN POWER switch.
• Turn off the FIL ON switch on the Ionization Gauge for the MIED.
• Turn off the FIL ON switch on the Ionization Gauge FEG.
• Turn off the EMIS switch on the main chamber gauge (Ionivac IM 520).
• Turn off the ON switch on the Parking gauge (GP 307 Vacuum Gauge Con-
troller).
• Toggle off the POWER switch for SIP 4.
• Press the STOP button on the TMP STP CONTROL UNIT against the back wall.
The turbo will take ∼20 minutes to spin down.
• Turn off the MAIN POWER switch
Restarting is the opposite of the above, with the exception of the pressure gau-
ges: if the pressures have fallen sufficiently, the SIPs will be out of range. Allow the
system some time to pump back down, checking the ion pumps periodically.
141
Chapter 6
Standard Imaging
6.1 Sample exchange
The sample exchange chamber must be vented to atmosphere to insert a new
sample. Plan ahead and stage any samples you might wish to remove to minimize
time pumping down (∼90 minutes when the instrument is relatively clean).
• Press the EXCH VENT button on the vacuum control panel (see Figure 5.4).
• Loosen the knob holding the exchange door closed.
• Wait for the chamber to pressurize.
• Remove any sample waiting to come out.
• Place the new sample in the exchange fork. The sample holder’s top slot
should go into the exchange fork in the instrument.
• Close the door and snug the knob.
• Press the EXCH VENT button and wait for the system to pump back down.
– The Pfieffer vacuum pump in front of the exchange chamber can be used
to help expedite the beginning of the pump down. Note that it shows
pressure in mbar; 1mbar = 100 Pa, so the chamber gauge is 2 orders of
magnitude higher than what shows on the pump.
– The Pfieffer is not useful beyond ∼1 × 10−3 Pa in the chamber.
– The chamber is pumped down enough once it reaches 3 × 10−5 Pa.
Once the exchange chamber has pumped down, it is safe to move the new sam-
ple into parking.
142
6.2 Parking
The parking chamber includes an arm that comes in from the back of the ma-
chine with 4 slots. This allows for storing samples in vacuum without needing to
pump down. Note that the rearmost slot in the parking holder requires V2 to be
open (it is ∼1 cm too long).
Access the parking chamber from the sample exchange area by opening V2.
Note that the exchange arm sticking out from the front of the instrument will need
to be fully retracted for V2 to open; the MGL 1 lamp will illuminate if this is not the
case.
Open V2 by pressing the V2 OPEN button in the vacuum control panel.
6.3 Main chamber
The main chamber is where the sample is viewed. It is accessed via V3. Again,
the exchange arm in front must be fully retracted for V3 to open.
The stage is prepared for sample exchange by selecting Auger Master > Ob-
servation > Sample manipulation... > Special Position... > Sample
Change Position > Move.
6.4 Eucentric height
To optimize the imaging contrast, milling, or SEMPA operations, the sam-
ple must be at eucentric height. This is achieved by setting the Auger Master >
Observation > Sample manipulation... > Stage Position > z such that
changes to the sample tilt do not cause the image to shift left or right.
Sample exchange lowers the stage to ∼-1.1; +1.1 is a good starting point for
finding eucentric height.
• Set stage tilt to 0°.
• Find a feature at ∼500x zoom and center it.
• Press the Auger Master > Observation > Sample manipulation... >
Tilt / Rotation > Set Tilt > 10 button to tilt the stage to 10°. Note that
this will increase the signal from the sample: lower the contrast to prevent
E-T detector saturation.
• Go back to 0°, then return to 10°to minimize backlash in the stage.
• Adjust the z height until the feature has returned to the center.
– Movement to the left indicates the stage is too low: increase z.
– Movement to the right indicates the stage is too high: decrease z.
143
• Return to 0°and verify the feature is still centered.
• Return to 10°and verify the feature is still centered.
Note that while tweaks to focus ideally should not cause the image to move, in a
less-than-perfectly calibrated system there will be some minor impact.
6.5 Controls
The stage control panel, shown in Figure 6.1a, is used for moving the stage.
• Translate across X and Y
• Raise and lower
• Rotate
• Tilt
Note that the HOLD ON lamp indicates that the control is disabled; press HOLD
OFF to regain control.
The optics control panel, shown in Figure 6.1b, is used to change the SEM im-
age.
• Fix stigmation/coma in X and Y
• Adjust the condenser lenses to increase or decrease probe current
• Change magnification
• Adjust focus
• Change contrast (difference between darkest and bright signals) and bright-
ness (effectively black level)
• Change scan speed (not typically used)
(a) The stage control panel (b) The optics control panel
Figure 6.1. SEM controls
144
Chapter 7
Auger Analysis
Auger analysis performs energy spectroscopy on so-called Auger secondary
electrons scattered from the Auger effect (though Ben might prefer these be called
Meitner electrons and the Meitner effect). When an incoming electron of suffi-
cient energy (25 keV will do) scatters an electron from an inner shell of an atom,
it creates a hole. This hole can be filled by an electron from a higher orbital. If the
binding energy of the lower orbit is considerably less than the transition energy of
the lowered electron, the excess energy can cause another higher shell electron to
be ejected. These energies are relatively low, leading to a short mean free path. The
electrons that break free from the sample are thus from the top several layers (on
the order of 1 nm of the sample.
As Auger analysis requires a clean surface, typically the procedure is to take an
initial scan to get a sense of what is on top, then go back and forth between ion
milling and scanning.
There are a number of Auger scan types available. In the interest of timeliness
(as the scan must go through a range of energies, dwelling on each step for a spec-
ified time), it is often useful to run a spectrum scan with a limited set of regions of
interest (ROIs).
Figure 7.1 shows an example of an Auger spectrum analysis: here 4 energy
ROIs, namely for O, Fe, Gd, and Pt, are examined. O shows an unexpected peak
(this was immediately after milling, which would have removed any surface oxy-
gen, suggesting there is oxygen in the sample). Fe shows the 3 characteristic peaks
at roughly 650, 700, and 750 eV. Both Gd and Pt show no signal (a Pt capping layer
was just milled off, and the Fe layer covers the lower Gd layer).
145
Figure 7.1. A sample spectrum analysis showing O and Fe peaks
Figure 7.2 shows the ROI configuration for the scan above. The selected ROI
table includes energy values for all relevant elements, allowing for simply entering
the symbol for the element of interest.
146
Figure 7.2. ROI configuration window
147
Chapter 8
Ar Ion Milling with the MIED
The MIED is used to prepare sample surfaces by removing contaminants with a
stream of energetic Ar ions.
There are 3 configurations stored in channels 10, 15, and 11, for 2,000 eV, 500 eV,
and 200 eV, respectively. The sample being cleaned will drive which energy should
be used. Approximate etch rates measured when the instrument was at NIST are in
Table 8.1.
The MIED is controlled by the components on the bottom left of the console;
see Figure 3.16.
• Ensure the ION GUN POWER button is illuminated.
• Ensure the HEAD, MEAS, and FIL ON buttons on the Ionization Gauge are illu-
minated.
• Press the AVC button; the display will show lines.
• Turn on the POWER switch on the GP 216 Pressure/Flow Controller. Allow
2-3 minutes for it to settle.
• Turn on the CONTROL switch on the GP 216 Pressure/Flow Controller.
• If the gauge spikes and shows E-H:
– Turn off the CONTROL switch.
– Turn the MEAS and FIL ON buttons on.
– Turn the Reference knob down.
– Repeat the previous step.
– If the pressure stabilizes around 5 × 10−2 Pa, increase Reference back to
5.0.
• Select the channel with the desired energy.
• Press BEAM ENERGY/EMIS CURRENT to verify correct energy and current val-
ues.
148
• Press BEAM SPOT if appropriate.
• Press TIMER RESET if desired.
• Press TIMER DISPLAY.
• Turn off Auger Master > Observation > Image Observation > SEI De-
tector to minimize damage to the e-t.
• Close the PCD.
• Press ETCHING ON/OFF to begin etching. The timer will begin.
• Press ETCHING ON/OFF to end etching. The timer will stop.
If appropriate, analyze the sample to see if sufficient milling has been done. Re-
peat the last 4 steps until done.
• Change to channel 1. This maintains a minimal current through the tip.
• Turn off the CONTROL switch on the GP 216 Pressure/Flow Controller.
• Allow 5 minutes for the system to cool down.
• Turn off the POWER switch on the GP 216 Pressure/Flow Controller.
• Press the AVC button. The pressure should drop to ∼1 × 10−5 Pa.
Element 50 eV 100 eV 500 eV 1,000 eV 2,000 eV
Co 0.04 0.3 1.3 2.0 2.8
Cr 0.04 0.25 1.1 2.0 2.6
Pt 0.05 0.3 1.4 2.1 3.0
Au 0.07 0.5 2.1 3.5 5.0
Si 0.005 0.07 0.42 0.7 0.95
Table 8.1. MIED etching rates; units are nmmin−1
149
Chapter 9
Ar Plasma Milling in the Sample
Exchange Chamber
The plasma gun sits atop the sample exchange chamber. It can be used to gently
mill samples with a low energy Ar plasma.
• Check the water level in the tank next to the pump in the service chaise.
• Turn on the pump.
– Validate pressure is ∼20 psi.
– Validate flow rate is ∼0.5 Lmin−1.
• Open the leak valve to allow Ar to enter at a pressure of 0.06 Pa to 0.09 Pa.
• Ensure the Magnetron Supply Power knob is turned down.
• Turn on the Magnetron Supply.
• Increase the Power knob slowly until a reading of 40mA to 50mA. Ensure
there is no purple glow showing in the small window in the gun.
• Ensure the Grid Supply Extraction and Ion Energy knobs are turned down.
• Turn on the Grid Supply.
• Increase the Extraction knob to 200 V. Ensure the current does not go much
above 1mA.
• Increase the Ion Energy knob to the desired beam energy.
• If the power supply clicks, adjust the voltages to make it stop.
When done, decrease the knobs to 0 and turn off the power supplies. Allow a
few minutes for cooling and then turn off the water pump.
150
(a) The rack of controls for the plasma gun (b) The purple glow of the functioning plasma
chamber
Figure 9.1. Plasma controls and glow
151
Chapter 10
Electromagnet
The electromagnet located at the rear of the SEMPA allows for applying a mag-
netic field to a sample once in the vacuum. This can be handy in the case of needing
to magnetically prepare or modify a sample (e.g., nucleating skyrmions in an Fe/Gd
multilayer thin film) after ion cleaning.
The sample to be put in the electromagnet must first be loaded onto the parking
arm, then moved all the way rearward until it can be moved onto the electromagnet
arm. There is a small wiper that can assist with pushing the sample holder from the
parking arm onto the electromagnet arm. The sample is then moved between the
poles of the magnet, visible through the glass apertures. The sample can be rotated
relative to the poles; there is an angle scale on the rotating dial to the right to help
gauge the angle of rotation.
Once in place, turn on the power supply and adjust the current to the desired
value. The simplest technique for this is to set all power supply knobs to the their
minimum, then increase the current knob(s) to maximum. With the voltage knob(s)
down, no current will flow. Slowly increase the voltage, keeping an eye on the cur-
rent.
There is an FW Bell GH-600 Hall effect sensor in the magnet chamber; unfor-
tunately, at the time of this writing, it appears at least one of the contacts in the
vacuum is broken. Ideally this sensor would be used to monitor real-time magnetic
flux in the area of the sample.
As an alternative, to get some characterization of the magnet’s field strength,
Table 10.1 below shows the magnetic field strengths measured with a Lakeshore
410 Gaussmeter inserted between the yoke and one of the coils for given current
inputs to the electromagnet. Table 10.2 extrapolates the currents required for given
strengths.
152
A mT
0.00 22.0
0.50 94.2 mT A
1.00 145.6 100 0.57
1.50 216 150 0.94
2.00 296 185 1.19
2.50 365 200 1.30
3.00 434 250 1.67
3.50 505 300 2.04
4.00 576 400 2.77
4.50 646 500 3.50
5.00 718 600 4.24
5.50 775 700 4.97
6.00 837 800 5.71
6.50 898 900 6.44
7.00 963 1,000 7.17
Table 10.1. Measured fields for Table 10.2. Required currents for
given currents desired fields
153
Chapter 11
Magnetic Domain Imaging
11.1 Prepare sample
In order to gather SEMPA images, the sample must have a surface free from
magnetic signal blocking elements. Most often this will be a layer of oxygen from
oxidation or some capping material added when the sample was created. See Ar Ion
Milling with the MIED and Ar Plasma Milling in the Sample Exchange Chamber.
Care should be taken to not overheat the sample, thus modifying or destroying
the magnetic texture to be imaged, or even the sample itself (as can happen with
transmission electron microscope (TEM) samples in particular due to their thin-
ness).
Once the sample is ready:
• Ensure the MIED is off and pressures have returned to normal.
• Tilt the stage to 52°, adjusting contrast/brightness as necessary.
• Ensure that the stage is at eucentric height.
• Carefully lower the SEMPA optics after removing the aluminum spacer.
• Ensure the sample has a feature that can be tracked easily as turning on the
SEMPA optics will move and distort the image. The Save Positions feature
on the Auger Master > Observation > Sample Manipulation window
can be particularly helpful: save the spot to be imaged before turning on the
SEMPA optics to have a reference point. If the feature used for tracking is in a
different location than the region of interest, that spot can also be saved.
11.2 Turn on SEMPA electronics
The viewed area and image settings must be adjusted before imaging. Once the
voltages are turned all the way up, the image will shift roughly 250 µm to the left
and 20 µm down.
154
• Turn on power supplies
– kepco power supply AC (left switch)
– 5V Descan Power
– NIST HV supply (just above the kepco)
– UEP 15 NIMS power supply
– SEMPA 0°/90° DETECTOR SIGNAL MULTIPLEXER
– JEOL SEMPA APERTURE CURRENTMONITOR
• Ramp up the voltages while keeping the feature to be tracked in view.
– Turn on the kepco DC voltage (right switch) and set the knob to 400 V to
600 V.
– Turn on the left-most NIMS power supply (NMQ 102M, labelled "Input
Lens") and set to 500 V.
– While keeping the kepco voltage 100 V to 200 V above the input lens and
adjusting contrast and stage position, increase the input lens to 1,500 V
and the kepco to 1,600 V to 1,800 V.
• Turn on SEMPA E-T detector
– Toggle on the HV ENABLE switch on the HV supply.
– Ensure the Scintillator HV power supply is set to 0 and toggle it on.
– Ensure the E-T detector selector switch is set to SEMPA ST of SEMPA RA
as appropriate.
– Increase the Scintillator HV dial to 250 (2,500 V).
In all likelihood, the SEM image screen will be dark. The SEMPA column volt-
ages might need tweaking to view an image.
• Turn the target knob on the SEMPA column to 45°.
• If a bright area is not visible, increase contrast/brightness; if still not visible
– Zoom out to ∼50x.
– Adjust the D1, D2, and D3 knobs slightly until the illuminated area is visi-
ble.
– Increase contrast/brightness if necessary; neither should have to go
above 200 as reported in Auger Master Image Observation.
• Adjust the deflector knobs as required to maximize brightness in the area to
be imaged, zooming in to the desired magnification.
• Adjust stigmation as required; typical values are around +100 in Y and +20 in
X.
155
• Adjust focus as required.
• Decrease the Scintillator HV dial to 0.
• Toggle off the HV ENABLE switch on the HV supply.
• Turn the target knob on the SEMPA column to 0°(fully counterclockwise).
• Turn on all the supplies in the NHS 6005p power supply.
• Turn on the ST or RA analyzer as required. There are 2 switches each, one for
the multi-channel plates (MCP) and one for the anode.
• Increase the MCP and anode voltages, keeping the anode ∼100 V above the
MCP, to ∼1,500 V.
• Adjust the voltages until the sum monitor at the top of the rack reads 30 µA.
11.3 Imaging
Switch the keyboard and mouse over to whisk. If not already running, launch
the sempa software from the command line. The window shown in Figure 11.1
should open on the bottom center monitor.
• Select the appropriate folder for the data to be collected.
• Set the prefix for the files to be generated.
• Set the first run number.
• Select straight through (X-Y) or right angle (X-Z).
• Close the PCD and select sempa > Options > Measure SEMPA Zeros. If the
resultant images are not fairly evenly noise, repeat until they are.
• Add one or more scan specifications as required. X/Y number of points and
dwell time will be dependent on the sample material and goal. Note increas-
ing either can have a significant effect on capture time.
• Capture images. The images will be rendered in real time on the bottom left
monitor; Figure 11.2 is an example.
– Free Run is often used for lower resolution/shorter dwell times to find
interesting features; the resultant images will not save or increment the
run number.
– Start saves the four resultant images to the folder selected above with
the specified name and run number: e.g.,
FeGd3x001_ix.sempa, FeGd3x001_iy.sempa, FeGd3x001_mx.sempa,
FeGd3x001_my.sempa
156
– The ...ix... and ...iy... images should be the same: intensity im-
ages from the E-T detector, while the ...mx... and ...my... images
are the magnetization images for X and Y, respectively. Selecting the X-Z
would result in ...ix2..., ...iz..., ...mx2..., ...mz... suffixes.
Turn everything off in the reverse order.
Figure 11.1. The main window for the sempa software
157
Figure 11.2. Sample images being captured in the X-Y plane
158
Chapter 12
Maintenance
12.1 Manual mode
Occasionally the instrument must be operated in manual mode. Most often this
is after it has shut down for some reason. A number of switches controlling pumps
and valves must be manipulated in order to return to automatic mode (the normal
operating condition). The instrument must be returned to "basal state" before auto-
matic mode can be resumed.
• Remove the metal cover at the bottom of the vacuum control panel (Fig-
ure 5.4).
• Press the GO TO AUTO button so that it is deactivated (out).
• Ensure all switches are toggled off (down).
• Turn on the TMP RP switch.
• Depending on the pressures, toggle the SIP and valve switches. The correct
order involves getting the high pressure out to the turbo pump without upset-
ting some other part of the system. This can be hit or miss, and might require
restarting. Perseverance pays off.
• Once the 3 SIP switches, V1, V5, V6, and PEG switch have been turned on with-
out illuminating the MAN OPERATION ERROR lamp, press the GO TO AUTO but-
ton. It should illuminate.
If (when) at any point in the above process the MAN OPERATION ERROR lamp illu-
minates, toggle all switches off and start over again. The process can take multiple
iterations as the pressures in the system get sorted out.
159
12.2 Bakeouts
Even in an ultra high vacuum environment, there are still contaminants that
will congregate and collect on the walls of the instrument. Water vapor is a com-
mon problem that will build over time as samples as exchanged or any work re-
quiring venting is done.
Periodic bakeouts help the SEMPA maintain the necessary vacuum by increas-
ing the temperature in the instrument above boiling, ideally to ∼140 °C.
The main column has heaters built in that will handle this, controlled by the
BAKE OUT section on the Vacuum control panel. The rest of the instrument will
need to be wrapped in heater tape (some is generally left in place).
All cables and water lines should be disconnected. The mu-metal plates should
be removed. Finally, the entire system wrapped in aluminum foil to help keep the
heat in.
Ensure any samples are either removed from the main chamber or put into
Bakeout Position (Auger Master > Observation > Sample manipulation...
> Special Position... > Bakeout Position > Move).
Figure 12.1. The SEMPA column wrapped and ready for bakeout
160
12.3 TSP
The titanium sublimation pump (TSP) helps to achieve the ultra high vacuum
environment required for SEMPA. It works by running a current through a Ti fil-
ament. As the filament heats up, Ti atoms boil off and capture contaminants in the
chamber before sticking to the chamber walls. Bakeouts help remove the build up
that happens with TSP use.
The TSP should be operated weekly during normal operation.
• Ensure the TSP CURRENT ADJ knob is turned fully counterclockwise.
• Press the TSP button. It should illuminate.
• Slowly increase the current by turning the TSP CURRENT ADJ knob clockwise
while monitoring the main chamber pressure. The pressure likely will not in-
crease significantly until the current approaches 20 A. A rapid rise in pressure,
as well as a pressure above ∼1 × 10−5 Pa, will cause the system to shut down.
The pressure can safely be raised (slowly!) up to ∼5 × 10−6 Pa without concern.
• After 2 minutes, the TSP cycle will end and the button will begin flashing.
Turn the TSP CURRENT ADJ knob fully counterclockwise and press the TSP
button. It should extinguish.
Note that the TSP filaments are consumable items.
12.4 Cleaning the gold targets
The gold targets need to be cleaned every few weeks.
• Turn the knobs on the SEMPA columns fully clockwise.
• Attach the blue cables to the columns.
• Attach the high/signal sides of the splitter to the gold targets.
• Attach the low side to the 1,000 V supply.
• Close the butterfly and MIED valves.
• Launch the Granville Philips pressure monitor on doorbell.
• Power on the voltage supplies and set the grids to +180 V, filaments to +30 V.
• Power on the current supplies, set a 2 V limit, reduce the current to 0, and
turn on the outputs.
• Power on the high voltage supply and set -1,000 V for the gold targets.
161
• Turn on the electrometer.
• Increase the current to 1 A. Monitor the pressure in the software.
• Increase the current to 2 A.
• Look for 4mA of emission.
• Go to manual mode on the vacuum control panel. Ensure all switches are off.
• Turn on SIPs 1 and 2.
• Open the Ar leak valve slowly ∼1.5 turns, keeping an eye on the pressure soft-
ware. Autoscale will help.
• Look for pressure to increase to 1 × 10−2 Pa. The emission current on the
electrometer should rise.
• Wait 2 1/2 minutes.
• Close the leak valve.
• Wait 3 minutes.
• Turn down the filament currents and power off supplies.
• Turn the knobs on the SEMPA column fully counterclockwise.
• Open V6, V2, and V3.
• The pressures should drop in the software monitor.
• Power off the voltage supplies.
• Disconnect the cables from the cleaning setup and reconnect the originals.
• Open the remaining valves.
• Go to auto mode.
• Open the butterfly and ion gun valves.
• Turn on the ion gauges.
162
Chapter 13
Troubleshooting
13.1 Shutdown
Especially when first working with the SEMPA, it seems as though it will shut
down if you look at it cross-eyed. There are many safeguards built in that will shut
down the FEG in an attempt to prevent damage to the rest of the instrument (not
always successfully, unfortunately).
The shutdown reason can be found in Auger Master by navigating to Mainte-
nance > FEG Control and looking at the FEG Status section.
Table 13.1 is reproduced from the JAMP-7830F manual, section 8.2. The major-
ity of shutdowns are related either to FEG emission current or vacuum pressure.
With the exception of codes 1 and 2, even numbers generally represent emission
current shutdowns, while odd numbers are vacuum pressure spikes.
13.2 Auger Masterwill not connect on launch
The base instrument stores a number of states in flipflops throughout the sys-
tem. These can get out of sync with apparent reality. Recovering from this can be
fairly quick or somewhat time consuming.
After stopping the Auger Master process, try the following steps in the order
listed, attempting to restart and reconnect Auger Master after each.
• Select Applications > AES > Reset AES
• Turn off ACCEL VOLTAGE and OPN POWER, then turn on OPN POWER and ACCEL
VOLTAGE
• Turn off ACCEL VOLTAGE and OPN POWER, then restart bracket. Once it is back
up, turn on OPN POWER and ACCEL VOLTAGE
163
Number State at shutdown Cause
0 Unknown Unknown
1 Emission
The system was shut down while the filament was current
2 being started up by the automatic gun-startup Vacuum
process. pressure
3 Excess time
4 Emission
The system was shut down while the suppressor current
voltage was being set by the automatic gun-
5 startup process. Vacuum
pressure
6 Emission
The system was shut down while the extracting current
7 voltage was being set by the automatic gun- Vacuum
startup process. pressure
8 Excess time
9 The system was shut down during the automaticgun-startup process. Interruptioncommand
10 Emission
The system was shut down during the automatic current
11 extracting voltage setting process. Vacuum
pressure
12 Emission
The system was shut down during normal opera- current
13 tion. Vacuum
pressure
14 Shutdown
command
15 Vacuum
The system was shut down immediately before pressure
16 the automatic gun-startup process was run or im-mediately before the automatic extracting-voltage Emission
setting process was run. current
17 Gun ready
Table 13.1. Shutdown codes
164
13.3 No SEM image
Check the items below.
• FEG on? Typical emission current is ∼90 µA.
• PCD open?
• Contrast and brightness turned up? Typical values are around 150/130, re-
spectively, for normal SEM imaging, a bit higher for SEMPA.
• E-T detector selection switch set correctly?
• Internal/external scan switch set correctly?
• If SEMPA imaging, SEMPA input optics on, HV switch on, scintillator HV on
and set to 250, column properly aligned?
13.4 The MIED will not mill
Check the items below for necessary initial conditions.
• Ensure the ION GUN POWER button is illuminated.
• Ensure the HEAD and MEAS buttons on the Ionization Gauge are illuminated.
• Ensure the AVC button on the Ionization Gauge is illuminated.
• Ensure the FIL ON button on the Ionization Gauge is illuminated.
• Ensure both the POWER and CONTROL switches on the GP 216 Pressure/Flow
Controller are on.
• Ensure the pressure in the ionization chamber is ∼5 × 10−2 Pa.
If all of that checks out but the ETCHING ON/OFF button seems to have no real
effect, check the fuses in the Micro Ion Etching Device rack unit, shown in Fig-
ure 13.1.
165
Figure 13.1. The two large gray fuses center right are prone to failing
13.5 V2 or V3will not open
Ensure that the 2nd transfer arm is fully retracted. There is a safeguard built in
to prevent the valves from opening if the arm is not all the way back.
13.6 Pulsing SEM image
The most frequent culprit for this is sample charging. This should be handled by
grounding the sample to the holder (which is held at instrument ground).
Less frequent is a damaged light pipe. Light pipes are consumable items. The
E-T detector works by electrons hitting a phosphor screen. Over time the electrons
will erode a hole through the phosphor.
166
(a) A light pipe with the phosphor burned in the (b) A new (reconditioned) light pipe
center
Figure 13.2. New vs. used light pipes
Light pipes can be refurbished by Applied Beams LLC in Beaverton. As of 2021
the price was $725 per pipe.
13.7 SEM image darkens and brightens but does
not zoom or focus
If the zoom is left very low (<100 x) for an extended period of time (no set time
for that), the required current for the coils can cause significant Joule heating, even-
tually blowing fuses.
Check the deflection coil fuses on the right side of the console. There are 2 8 A
fuses toward the back.
167
Figure 13.3. The two white fuses center right sometimes blow when zoomed
out (<100 x) for an extended period
13.8 Accelerating voltage stuck at one value
It is possible for some components inside the high tension tank to fail. This has
been known to peg the voltage to 30 kV.
There should be a ground wire between the Cockroft-Walton multiplier and the
floating FEG power supply that is connected between 2 large resistors nearby. This
wire was missing/vaporized at one point.
Several ICs were previously damaged. The schematics were marked with ∆’s to
indicate changed parts.
Be sure to use a high voltage probe and exercise extreme caution when work-
ing on the high tension tank. Also be sure to refill the tank with SF6 to the marked
pressure on the gauge.
168
Figure 13.4. There are several transistors hiding within
13.9 FEG emission current limited to ∼1.9 A despite
higher command
If the emission current from the FEG cannot go above ∼1.9 A (normal operat-
ing condition is ∼2.3 A or a bit higher), it is possibly a transistor buried in the high
tension tank has failed.
Previous testing to determine the root cause:
• Removed the tip power cable and put the dummy load in (perpendicular to
the guide pin axis). Attempted to run the current up and hit the same 1.9 A
limit.
169
• Pulled the power supply out of the tank and hooked things up on the cart.
Checked all the pins for the DACs (ICs 17 and 18) and found good values.
Checked the output of the op-amps (ICs 5, 6, and 10) and found good values.
• Checked the voltage going to the last transistor before the tip and found a
relatively low collector voltage (∼2.8 V; expected at least 5 V; the output of IC
10 was at 15 V).
• Built a reproduction of the power filtering circuit (a couple capacitors and a
load resistor) and checked the characteristics of the incoming voltage. The
circuit behaved the same (unloaded voltage across the circuit of ∼3.5 V that
dropped to 2.5-2.9 when shown a load). Steve found that peculiar, but couldn’t
call it the real problem.
• Replaced the power supply coming into the filter with a lab power supply.
This had the added benefit of letting us see the current drawn. Set the lab
supply to 3.5 A and commanded the filament current to 1 A, supply showed
1 A. Went to 2 A, supply showed 1.9 A. 2.5 A also showed 1.9 A. This is the
behavior we saw that started this line of investigation. This allowed us to nar-
row down the problem to 1 or more of 3 transistors.
• After replacement, a command of 2.5 A showed 2.5 A through the circuit.
• Adjust the read back variable resistor to fix a small offset.
Replacing these parts is best done with Steve Weimholt’s assistance as high volt-
ages are involved.
• Vent the eGun chamber (see Venting for maintenance).
• Set bakeout timer to 10 hours.
• Remove the hex head bolts holding the high tension line in place at the top of
the instrument.
• Remove the tip being careful not to hit the sides.
• Replace the tip, aligning with 3 set screws as best possible while looking
down through the end of the holder.
• Add SF6 into the gun chamber until it visibly overflows (∼2-3 seconds).
• Pump back down. At low enough pressure, a bake out to TMP will kick off
(and run for 10 hours).
• Bake out to SIP for at least 24 more hours.
170
13.10 No FEG emission current at normal operat-
ing condition
If there is no emission current from the FEG, it is most likely that the tip has
failed. This can be verified by checking the tip’s resistance: it should be below 1Ω.
The replacement part, available from Ted Pella, is TFE174C SE4 Tip for JEOL
JAMP-7830F. As of summer 2021 the part was $5,500.
Replacing the part is best done with Steve Weimholt’s assistance as the align-
ment in the holder is critical (see Figure 3.5).
• Vent the eGun chamber (see section 5.2).
• Set bakeout timer to 10 hours.
• Remove the hex head bolts holding the high tension line in place at the top of
the instrument.
• Remove the tip being careful not to hit the sides.
• Replace the tip, aligning with 3 set screws as best possible while looking
down through the end of the holder.
• Pump back down. At low enough pressure, a bake out to TMP will kick off
(and run for 10 hours).
• Bake out to SIP for at least 24 more hours.
• Replacement instructions can be found at C:\SEMPA\Instructions_and_-
Manuals\FESAM\7830F FE tip Replacement.doc.
Adjusting the accelerating voltage was also an important part of this process:
high accelerating voltage sends the beam pretty straight down the column, while
lower voltages cause the electron beam to spread out a bit, illuminating more of the
FEG chamber (and knocking a lot of stuff off). Adjustements started around 10 keV
and worked down to ∼3 keV, as pressures permitted. This took several days.
Another much less obvious possibility is a blown opto-coupler. Several opto-
couplers are used to electronically isolate the high voltage parts of the system from
the circuits. Perhaps coincidentally, one of these seems to have been damaged dur-
ing an arc-related shutdown. Examine page HT I/O PB (FE) 3 in the schematics.
13.11 Cannot get SEMPA electrons to E-T detector
or targets
Stray currents or improper voltages anywhere along the SEMPA column can
deflect the beam.
171
It is possible that a microscopic (though hopefully larger) bit of contamination
can short the electrostatic lenses in the SEMPA column; this happened to the joint
shown in Figure 13.5. In the worst case, this can prevent the electron beam from
making it to the end of the column where the sensors are.
This was remedied by shutting down the instrument and removing the SEMPA
optics, then testing continuity between adjacent lenses. Each lens should be elec-
trically isolated from the next. There was nothing visible, but there was a short
between two of the lenses. The fix was to blow compressed air across the joint and
retest continuity. If that is insufficient, the lenses will need to be separated and the
interface between them cleaned.
Figure 13.5. There was an invisible short in the space between two of the elec-
trostatic lenses
Another possible culprit is an issue with the nanoammeters at the top of the
SEMPA rack. There are 8 identical cards in the chassis, selectable by the dial on the
172
front. The original design for these was overly reliant on an accurate voltage supply
(needed to be within a few percent of ± 15 volt). The unit was modified by Cliff at
Technical Services to be adjustable to handle drift over time.
It is also possible that one or more of the cards can develop a short. They mea-
sure small currents at high voltages (thus the floating design) and can be damaged
by arcs.
Figure 13.6. This unit houses 8 nanoammeters, each on a separate card
173
Figure 13.7. These are designed to measure tiny currents and are susceptible to
damage during arcing events
174
Acronyms
E-T detector Everhart-Thornley detector. 117, 120, 130, 143,
155, 157, 165, 166
FEG field emission gun. 106, 107, 109, 124, 134, 138,
139, 141, 163, 165, 168, 169, 171
MCP multi-channel plates. 156
MIED Micro Ion Etching Device. 112, 116, 120, 136,
141, 148, 154, 161
NIST National Institute of Standards and Technology.
103, 133, 148
PCD probe current detector. 106, 134, 149, 156, 165
RA right angle. 126, 127, 156
ROI region of interest. 145, 146
SEM scanning electron microscope. 99, 111, 117, 120,
133, 144, 155, 165
SEMPA Scanning Electron Microscopy with Polarization
Analysis. 99, 103, 105, 111, 112, 117, 126, 130,
132–134, 143, 152, 154–156, 160–163, 165, 171,
172
SIP sputter ion pump. 109, 123, 141, 159, 162, 170,
171
175
ST straight through. 126, 127, 156
TEM transmission electron microscope. 154
TMP turbo molecular pump. 122, 141, 170, 171
TSP titanium sublimation pump. 161
176
REFERENCES CITED
[1] Data Center Power: Global Strategic Business Report. (Visited on 11/13/2022) (cit.
on p. 4).
[2] George Kamiya. Data Centres and Data Transmission Networks, IEA, Paris. (Vis-
ited on 11/13/2022) (cit. on p. 4).
[3] Spintronics for Low-Power Computing. Design, Automation, Test in Europe Con-
ference, and Exhibition (DATE). 2014 (cit. on pp. 4, 16).
[4] R. Tomasello et al. “A strategy for the design of skyrmion racetrack memories”.
In: Scientific Reports 4.1 (Oct. 2014), p. 6784. issn: 2045-2322. doi: 10.1038/sr
ep06784. url: https://doi.org/10.1038/srep06784 (cit. on p. 4).
[5] Guoqiang Yu et al. “Room-temperature skyrmion shift device for memory appli-
cation”. In: Nano letters 17.1 (2017), pp. 261–268 (cit. on p. 4).
[6] R Tomasello et al. “Performance of synthetic antiferromagnetic racetrack mem-
ory: domain wall versus skyrmion”. In: Journal of Physics D: Applied Physics
50.32 (2017), p. 325302 (cit. on p. 4).
[7] R. C. Sherwood, J. P. Remeika, and H. J. Williams. “Domain Behavior in Some
Transparent Magnetic Oxides”. In: Journal of Applied Physics 30.2 (1959),
pp. 217–225. doi: 10.1063/1.1735136. eprint: https://doi.org/10.1063/1
.1735136. url: https://doi.org/10.1063/1.1735136 (cit. on p. 15).
[8] C. Kooy and U. Enz. “Experimental and theoretical study of the domain con-
figuration in thin layers of BaFe12019”. In: PHILIPS RESEARCH REPORTS.
Vol. 15. 1960, pp. 7–19 (cit. on p. 15).
[9] C. D. Mee. “Recent measurements of the magneto-optical properties of some
garnets”. In: Contemporary Physics 8.4 (1967), pp. 385–400. doi: 10.1080/0010
7516708202161. eprint: https://doi.org/10.1080/00107516708202161. url:
https://doi.org/10.1080/00107516708202161 (cit. on p. 15).
[10] J. M. Nemchik. “Circular Domain Configuration in GdIG”. In: Journal of Ap-
plied Physics 40.3 (1969), pp. 1086–1087. doi: 10.1063/1.1657535. eprint:
https://doi.org/10.1063/1.1657535. url: https://doi.org/10.1063/1.1
657535 (cit. on p. 15).
[11] Andrew H Bobeck and H E D Scovil. “MAGNETIC BUBBLES”. en. In: SCIEN-
TIFIC AMERICAN (1971), p. 15 (cit. on p. 15).
177
[12] L.A. Russell. “High-Speed Magnetic-Core Memory Technology”. In: ed. by L.
Marton. Vol. 21. Advances in Electronics and Electron Physics. Academic
Press, 1966, pp. 249–286. doi: https://doi.org/10.1016/S0065-2539(0
8)61011-2. url: https://www.sciencedirect.com/science/article/pii
/S0065253908610112 (cit. on p. 15).
[13] IBM Archives: IBM 2305 fixed head storage. url: https://www.ibm.com/ibm/hi
story/exhibits/storage/storage_2305.html (cit. on p. 15).
[14] K.F. Baker. “A review of magnetic bubble memories and their applications”.
English. In: Radio and Electronic Engineer 51 (3 Mar. 1981), 105–115(10). issn:
0033-7722. url: https://digital-library.theiet.org/content/journals
/10.1049/ree.1981.0014 (cit. on p. 15).
[15] R. Suzuki. “Recent development in magnetic-bubble memory”. In: Proceedings
of the IEEE 74.11 (1986), pp. 1582–1590. doi: 10.1109/PROC.1986.13670 (cit.
on p. 15).
[16] J.Egil Juliussen. “Bubble memory as small mass storage”. In:Microelectronics
Reliability 16.4 (1977), pp. 427–430. issn: 0026-2714. doi: https://doi.org/1
0.1016/0026-2714(77)90441-3. url: https://www.sciencedirect.com/sci
ence/article/pii/0026271477904413 (cit. on p. 15).
[17] Tina-Marie D’Ercole. A solid state data recorder for space-based applications us-
ing magnetic bubble memory. Tech. rep. NAVAL POSTGRADUATE SCHOOL
MONTEREY CA, 1986 (cit. on p. 15).
[18] WCMavity. “Bubble memory system applications”. In: NASA STI/Recon Tech-
nical Report N 77 (1977), p. 19391 (cit. on p. 15).
[19] H Kohara et al. “True swap gate design for on chip cache organization megabit
bubble memory”. In: IEEE Transactions on Magnetics 17.6 (1981), pp. 3038–
3040 (cit. on p. 15).
[20] Mona Berciu and Sajeev John. “Charged bosons in a doped Mott insulator:
Electronic properties of domain-wall solitons and meron vortices”. In: Phys-
ical Review B 57.16 (1998), p. 9521 (cit. on p. 15).
[21] Ronald Johannes Keizer et al. Evolution of magnetism and its interplay with
superconductivity in heavy-fermion U (Pt, Pd) 3. Universiteit van Amsterdam
[Host], 1999 (cit. on p. 15).
[22] A Knigavko and Baruch Rosenstein. “Magnetic Skyrmion Lattices in Heavy
Fermion Superconductor UPt 3”. In: Physical review letters 82.6 (1999), p. 1261
(cit. on p. 15).
[23] Tony Hilton Royle Skyrme. “A non-linear field theory”. In: Proceedings of the
Royal Society of London. Series A. Mathematical and Physical Sciences 260.1300
(1961), pp. 127–138 (cit. on pp. 15, 18).
178
[24] T.H.R. Skyrme. “A unified field theory of mesons and baryons”. en. In: Nuclear
Physics 31 (Mar. 1962), pp. 556–569. issn: 00295582. doi: 10.1016/0029-5582
(62)90775-7. url: https://linkinghub.elsevier.com/retrieve/pii/0029
558262907757 (visited on 11/10/2021) (cit. on p. 15).
[25] Qi Li, John Toner, and D. Belitz. “Skyrmion versus vortex flux lattices in p-
wave superconductors”. In: Phys. Rev. B 79 (1 Jan. 2009), p. 014517. doi: 10
.1103/PhysRevB.79.014517. url: https://link.aps.org/doi/10.1103/Phy
sRevB.79.014517 (cit. on pp. 15, 86).
[26] Qi Li, John Toner, and D. Belitz. “Elasticity and Melting of Skyrmion Flux
Lattices in p-Wave Superconductors”. In: Phys. Rev. Lett. 98 (18 May 2007),
p. 187002. doi: 10.1103/PhysRevLett.98.187002. url: https://link.aps.o
rg/doi/10.1103/PhysRevLett.98.187002 (cit. on p. 15).
[27] Eslam Khalaf et al. “Charged skyrmions and topological origin of superconduc-
tivity in magic-angle graphene”. In: Science advances 7.19 (2021), eabf5299 (cit.
on p. 15).
[28] Eric Mascot et al. “Topological superconductivity in skyrmion lattices”. In: npj
Quantum Materials 6.1 (2021), pp. 1–6 (cit. on p. 15).
[29] Alexei N Bogdanov and DA Yablonskii. “Thermodynamically stable “vortices”
in magnetically ordered crystals. The mixed state of magnets”. In: Zh. Eksp.
Teor. Fiz 95.1 (1989), p. 178 (cit. on p. 15).
[30] Ulrich K Roessler, AN Bogdanov, and C Pfleiderer. “Spontaneous skyrmion
ground states in magnetic metals”. In: Nature 442.7104 (2006), pp. 797–801
(cit. on p. 15).
[31] Benedikt Binz, Ashvin Vishwanath, and Vivek Aji. “Theory of the helical spin
crystal: a candidate for the partially ordered state of MnSi”. In: Physical review
letters 96.20 (2006), p. 207202 (cit. on p. 15).
[32] S. Mühlbauer et al. “Skyrmion Lattice in a Chiral Magnet”. en. In: Science
323.5916 (Feb. 2009), pp. 915–919. issn: 0036-8075, 1095-9203. doi:
10.1126/science.1166767. url: https://www.science.org/doi/10.1126/s
cience.1166767 (visited on 11/10/2021) (cit. on pp. 15, 17, 24).
[33] XZ Yu et al. “Real-space observation of a two-dimensional skyrmion crystal”.
In: Nature 465.7300 (2010), pp. 901–904 (cit. on p. 15).
[34] Akira Tonomura et al. “Real-space observation of skyrmion lattice in helimag-
net MnSi thin samples”. In: Nano letters 12.3 (2012), pp. 1673–1677 (cit. on
pp. 15, 24).
[35] Gareth Williams. “Stability of skyrmions in thin films of superfluid A 3”. In:
Physical Review B 48.10 (1993), p. 7704 (cit. on p. 15).
[36] VL Pokrovsky and WM Saslow. Information-bearing structures and magnetism of
thin films. Annual report, 1997–1998. Tech. rep. Texas A and M Univ., College
Station, TX (United States). Dept. of Physics, 1998 (cit. on p. 15).
179
[37] V Ilkovič, S Tuleja, and D Blažek. “Magnetic properties of superlattice-
configured thin films with variable interlayer couplings”. In: Journal of
magnetism and magnetic materials 193.1-3 (1999), pp. 177–180 (cit. on p. 15).
[38] SD Bader and SSP Parkin. “Spintronics”. In: Annu. Rev. Condens. Matter Phys.
1.1 (2010), pp. 71–88 (cit. on p. 15).
[39] Igor Žutić, Jaroslav Fabian, and S Das Sarma. “Spintronics: Fundamentals and
applications”. In: Reviews of modern physics 76.2 (2004), p. 323 (cit. on p. 15).
[40] David D Awschalom, Michael E Flatté, and Nitin Samarth. “Spintronics”. In:
Scientific American 286.6 (2002), pp. 66–73 (cit. on p. 15).
[41] Fabio Pulizzi. “Spintronics”. In: Nature materials 11.5 (2012), pp. 367–367 (cit.
on p. 15).
[42] Mamoru Matsuo, Junichi Ieda, and Sadamichi Maekawa. “Mechanical gener-
ation of spin current”. In: Frontiers in Physics 3 (2015). issn: 2296-424X. doi:
10.3389/fphy.2015.00054. url: https://www.frontiersin.org/articles
/10.3389/fphy.2015.00054 (cit. on p. 15).
[43] Niklas Romming et al. “Writing and deleting single magnetic skyrmions”. In:
Science 341.6146 (2013), pp. 636–639 (cit. on p. 16).
[44] X.Z. Yu et al. “Skyrmion flow near room temperature in an ultralow current
density”. In: Nature Communications 3.1 (Aug. 2012), p. 988. issn: 2041-1723.
doi: 10.1038/ncomms1990. url: https://doi.org/10.1038/ncomms1990
(cit. on p. 16).
[45] N S Kiselev et al. “Chiral skyrmions in thin magnetic films: new objects for
magnetic storage technologies?” In: Journal of Physics D: Applied Physics 44.39
(Sept. 2011), p. 392001. doi: 10.1088/0022-3727/44/39/392001. url: https:
//doi.org/10.1088/0022-3727/44/39/392001 (cit. on p. 17).
[46] Xiuzhen Yu et al. “Magnetic stripes and skyrmions with helicity reversals”. In:
Proceedings of the National Academy of Sciences 109.23 (2012), pp. 8856–8860
(cit. on p. 17).
[47] R Tolley, SA Montoya, and EE Fullerton. “Room-temperature observation and
current control of skyrmions in Pt/Co/Os/Pt thin films”. In: Physical Review
Materials 2.4 (2018), p. 044404 (cit. on p. 17).
[48] Naoto Nagaosa and Yoshinori Tokura. “Topological properties and dynamics of
magnetic skyrmions”. en. In: Nature Nanotechnology 8.12 (Dec. 2013), pp. 899–
911. issn: 1748-3387, 1748-3395. doi: 10.1038/nnano.2013.243. url: http
://www.nature.com/articles/nnano.2013.243 (visited on 11/10/2021)
(cit. on pp. 18, 24).
[49] Gong Chen et al. “Tailoring the chirality of magnetic domain walls by interface
engineering”. In: Nature Communications 4.2671 (2013) (cit. on pp. 18, 26).
180
[50] Tôru Moriya. “New Mechanism of Anisotropic Superexchange Interaction”. In:
Phys. Rev. Lett. 4 (5 Mar. 1960), pp. 228–230. doi: 10.1103/PhysRevLett.4.2
28. url: https://link.aps.org/doi/10.1103/PhysRevLett.4.228 (cit. on
p. 24).
[51] M. Bode et al. “Chiral magnetic order at surfaces driven by inversion asym-
metry”. In: Nature 447.7141 (May 2007), pp. 190–193. issn: 1476-4687. doi:
10.1038/nature05802. url: https://doi.org/10.1038/nature05802 (cit. on
p. 24).
[52] M. Heide, G. Bihlmayer, and S. Blügel. “Dzyaloshinskii-Moriya interaction
accounting for the orientation of magnetic domains in ultrathin films:
Fe/W(110)”. In: Phys. Rev. B 78 (14 Oct. 2008), p. 140403. doi: 10.1103/PhysRe
vB.78.140403. url: https://link.aps.org/doi/10.1103/PhysRevB.78.140
403 (cit. on p. 24).
[53] Taro Nakajima et al. “Skyrmion lattice structural transition in MnSi”. In: Sci-
ence Advances 3.6 (2017), e1602562. doi: 10.1126/sciadv.1602562. eprint:
https://www.science.org/doi/pdf/10.1126/sciadv.1602562. url:
https://www.science.org/doi/abs/10.1126/sciadv.1602562 (cit. on p. 24).
[54] Won-Young Choi et al. “Skyrmion Phase in MnSi Thin Films Grown on Sap-
phire by a Conventional Sputtering”. In: Nanoscale Research Letters 16.1 (Jan.
2021), p. 7. issn: 1556-276X. doi: 10.1186/s11671-020-03462-2. url: https:
//doi.org/10.1186/s11671-020-03462-2 (cit. on p. 24).
[55] D. Okuyama et al. “Deformation of the moving magnetic skyrmion lattice in
MnSi under electric current flow”. In: Communications Physics 2.1 (July 2019),
p. 79. issn: 2399-3650. doi: 10.1038/s42005-019-0175-z. url: https://doi
.org/10.1038/s42005-019-0175-z (cit. on p. 24).
[56] B. Lebech, J. Bernhard, and T. Freltoft. “Magnetic structures of cubic FeGe
studied by small-angle neutron scattering”. In: Journal of Physics Condensed
Matter 1.35 (Sept. 1989), pp. 6105–6122. doi: 10.1088/0953-8984/1/35/010
(cit. on p. 24).
[57] X. Z. Yu et al. “Near room-temperature formation of a skyrmion crystal in
thin-films of the helimagnet FeGe”. en. In: Nature Materials 10.2 (Feb. 2011),
pp. 106–109. issn: 1476-1122, 1476-4660. doi: 10.1038/nmat2916. url: htt
p://www.nature.com/articles/nmat2916 (visited on 11/10/2021) (cit. on
p. 24).
[58] Maxime Leroux et al. “Skyrmion Lattice Topological Hall Effect near Room
Temperature”. In: Scientific Reports 8.1 (Oct. 2018), p. 15510. issn: 2045-2322.
doi: 10.1038/s41598-018-33560-2. url: https://doi.org/10.1038/s41598
-018-33560-2 (cit. on p. 24).
181
[59] Alison C. Twitchett-Harrison et al. “Confinement of Skyrmions in Nanoscale
FeGe Device-like Structures”. In: ACS Applied Electronic Materials 4.9 (2022),
pp. 4427–4437. doi: 10.1021/acsaelm.2c00692. eprint: https://doi.org/10
.1021/acsaelm.2c00692. url: https://doi.org/10.1021/acsaelm.2c00692
(cit. on p. 24).
[60] Sujan Budhathoki et al. “Room-temperature skyrmions in strain-engineered
FeGe thin films”. In: Phys. Rev. B 101 (22 June 2020), p. 220405. doi: 10.1103
/PhysRevB.101.220405. url: https://link.aps.org/doi/10.1103/PhysRev
B.101.220405 (cit. on p. 24).
[61] András Kovács et al. “Mapping the magnetization fine structure of a lattice of
Bloch-type skyrmions in an FeGe thin film”. In: Applied Physics Letters 111.19
(2017), p. 192410. doi: 10.1063/1.5004394. eprint: https://doi.org/10.106
3/1.5004394. url: https://doi.org/10.1063/1.5004394 (cit. on p. 24).
[62] Albert Fert and Peter M Levy. “Role of anisotropic exchange interactions in
determining the properties of spin-glasses”. In: Physical Review Letters 44.23
(1980), p. 1538 (cit. on p. 24).
[63] Tsuyoshi Okubo, Sungki Chung, and Hikaru Kawamura. “Multiple-q States and
the Skyrmion Lattice of the Triangular-Lattice Heisenberg Antiferromagnet
under Magnetic Fields”. In: Phys. Rev. Lett. 108 (1 Jan. 2012), p. 017206. doi:
10.1103/PhysRevLett.108.017206. url: https://link.aps.org/doi/10.11
03/PhysRevLett.108.017206 (cit. on p. 25).
[64] S. von Malottki et al. “Enhanced skyrmion stability due to exchange frustra-
tion”. In: Scientific Reports 7.1 (Sept. 2017), p. 12299. issn: 2045-2322. doi: 10
.1038/s41598-017-12525-x. url: https://doi.org/10.1038/s41598-017-1
2525-x (cit. on p. 25).
[65] Thomas T. J. Mutter, Andrey O. Leonov, and Katsuya Inoue. “Skyrmion insta-
bilities and distorted spiral states in a frustrated chiral magnet”. In: Phys. Rev.
B 100 (6 Aug. 2019), p. 060407. doi: 10.1103/PhysRevB.100.060407. url:
https://link.aps.org/doi/10.1103/PhysRevB.100.060407 (cit. on p. 25).
[66] Stefan Heinze et al. “Spontaneous atomic-scale magnetic skyrmion lattice in
two dimensions”. en. In: Nature Physics 7.9 (Sept. 2011), pp. 713–718. issn:
1745-2473, 1745-2481. doi: 10.1038/nphys2045. url: http://www.natur
e.com/articles/nphys2045 (visited on 11/10/2021) (cit. on p. 25).
[67] Souvik Paul et al. “Role of higher-order exchange interactions for skyrmion
stability”. In: Nature Communications 11.1 (Sept. 2020), p. 4756. issn: 2041-
1723. doi: 10.1038/s41467-020-18473-x. url: https://doi.org/10.1038/s
41467-020-18473-x (cit. on p. 25).
[68] Nguyen Duy Khanh et al. “Nanometric square skyrmion lattice in a centrosym-
metric tetragonal magnet”. In: Nature Nanotechnology 15.6 (2020), pp. 444–449
(cit. on p. 25).
182
[69] F. J. A. den Broeder et al. “Perpendicular Magnetic Anisotropy of Co-Au Multi-
layers Induced by Interface Sharpening”. In: Phys. Rev. Lett. 60 (26 June 1988),
pp. 2769–2772. doi: 10.1103/PhysRevLett.60.2769. url: https://link.aps
.org/doi/10.1103/PhysRevLett.60.2769 (cit. on p. 25).
[70] N. Nakajima et al. “Perpendicular Magnetic Anisotropy Caused by Interfacial
Hybridization via Enhanced Orbital Moment in Co/PtMultilayers: Mag-
netic Circular X-Ray Dichroism Study”. In: Phys. Rev. Lett. 81 (23 Dec. 1998),
pp. 5229–5232. doi: 10.1103/PhysRevLett.81.5229. url: https://link.aps
.org/doi/10.1103/PhysRevLett.81.5229 (cit. on p. 25).
[71] Nam-Hui Kim et al. “Interfacial Dzyaloshinskii-Moriya interaction, surface
anisotropy energy, and spin pumping at spin orbit coupled Ir/Co interface”.
In: Applied Physics Letters 108.14 (2016), p. 142406 (cit. on p. 25).
[72] Woo Seung Ham et al. “Dzyaloshinskii–Moriya interaction in noncentrosym-
metric superlattices”. In: npj Computational Materials 7.1 (Aug. 2021), p. 129.
issn: 2057-3960. doi: 10.1038/s41524-021-00592-8. url: https://doi.org
/10.1038/s41524-021-00592-8 (cit. on p. 25).
[73] P. F. Carcia. “Perpendicular magnetic anisotropy in Pd/Co and Pt/Co thin-film
layered structures”. In: Journal of Applied Physics 63.10 (1988), pp. 5066–5073.
doi: 10.1063/1.340404. eprint: https://doi.org/10.1063/1.340404. url:
https://doi.org/10.1063/1.340404 (cit. on p. 25).
[74] MA Slonczewski and AP Malozemoff. “Magnetic domain walls in bubble mate-
rials”. In: Suppl (1979) (cit. on p. 25).
[75] Motohiko Ezawa. “Giant Skyrmions Stabilized by Dipole-Dipole Interactions
in Thin Ferromagnetic Films”. en. In: Physical Review Letters 105.19 (Nov.
2010), p. 197202. issn: 0031-9007, 1079-7114. doi: 10.1103/PhysRevLett
.105.197202. url: https://link.aps.org/doi/10.1103/PhysRevLett.105
.197202 (visited on 10/18/2022) (cit. on p. 25).
[76] Jordan J. Chess et al. “Determination of domain wall chirality using in situ
Lorentz transmission electron microscopy”. In: AIP Advances 7.5 (2017),
p. 056807. doi: 10.1063/1.4977500 (cit. on p. 26).
[77] S. A. Montoya et al. “Resonant properties of dipole skyrmions in amorphous
Fe/Gd multilayers”. In: PHYSICAL REVIEW B 95.224405 (2017) (cit. on p. 26).
[78] J. C. T Lee et al. “Synthesizing skyrmion bound pairs in Fe-Gd thin films”. In:
Applied Physics Letters 109.2 (2016), p. 022402. doi: 10 . 1063 / 1 . 4955462.
eprint: https://doi.org/10.1063/1.4955462. url: https://doi.org/10.10
63/1.4955462 (cit. on pp. 26, 27, 86).
[79] Soong-Geun Je et al. “Direct demonstration of topological stability of magnetic
skyrmions via topology manipulation”. In: ACS nano 14.3 (2020), pp. 3251–
3258 (cit. on p. 26).
183
[80] MH Seaberg et al. “Spontaneous fluctuations in a magnetic fe/gd skyrmion lat-
tice”. In: Physical Review Research 3.3 (2021), p. 033249 (cit. on p. 26).
[81] Yuqing Zhou et al. “A comparative study of the domain wall motion in ferri-
magnets (Fe, Co) 1- x (Gd, Tb) x”. In: Nanoscale 14.37 (2022), pp. 13526–13531
(cit. on p. 26).
[82] Bert Koopmans. “Femto-magnetism meets spintronics-toward integrated
magneto-photonics”. In: IEEE Transactions on Magnetics 55.12 (2019), p. 1
(cit. on p. 26).
[83] Felipe M de Souza and Ram K Gupta. “Magnetic Nanomaterials for Flexible
Spintronics”. In: Emerging Applications of Low Dimensional Magnets. CRC
Press, pp. 303–318 (cit. on p. 26).
[84] SA Montoya et al. “Tailoring magnetic energies to form dipole skyrmions and
skyrmion lattices”. In: Physical Review B 95.2 (2017), p. 024415 (cit. on p. 26).
[85] Jordan J Chess. “Mapping Topological Magnetization and Magnetic
Skyrmions”. PhD thesis. University of Oregon, 2017 (cit. on pp. 26, 43).
[86] Gong Chen and Andreas K. Schmid. “Imaging and Tailoring the Chirality
of Domain Walls in Magnetic Films”. In: Advanced Materials 27.38 (2015),
pp. 5738–5743. doi: https://doi.org/10.1002/adma.201500160. eprint: htt
ps://onlinelibrary.wiley.com/doi/pdf/10.1002/adma.201500160. url:
https://onlinelibrary.wiley.com/doi/abs/10.1002/adma.201500160
(cit. on p. 26).
[87] Alice T Greenberg. “The Application of Interferometric Electron Microscopy
for Nanomagnetic Imaging”. PhD thesis. University of Oregon, 2020 (cit. on
pp. 27, 43, 75).
[88] Xiuzhen Yu et al. “Observation of the magnetic skyrmion lattice in a MnSi
nanowire by Lorentz TEM”. In: Nano letters 13.8 (2013), pp. 3755–3759 (cit.
on p. 27).
[89] Jin Tang et al. “Lorentz transmission electron microscopy for magnetic
skyrmions imaging”. In: Chinese Physics B 28.8 (2019), p. 087503 (cit. on p. 27).
[90] Yoshinori Tokura and Naoya Kanazawa. “Magnetic skyrmion materials”. In:
Chemical Reviews 121.5 (2020), pp. 2857–2897 (cit. on p. 27).
[91] Ryan D. Desautels et al. “Realization of ordered magnetic skyrmions in
thin films at ambient conditions”. In: Phys. Rev. Materials 3 (10 Oct. 2019),
p. 104406. doi: 10.1103/PhysRevMaterials.3.104406. url: https://link.a
ps.org/doi/10.1103/PhysRevMaterials.3.104406 (cit. on p. 27).
[92] Seonghoon Woo et al. “Deterministic creation and deletion of a single mag-
netic skyrmion observed by direct time-resolved X-ray microscopy”. In: Na-
ture Electronics 1.5 (2018), pp. 288–296 (cit. on p. 27).
184
[93] V Ukleev et al. “Element-specific soft x-ray spectroscopy, scattering, and imag-
ing studies of the skyrmion-hosting compound Co 8 Zn 8 Mn 4”. In: Physical
Review B 99.14 (2019), p. 144408 (cit. on p. 27).
[94] MT Birch et al. “Real-space imaging of confined magnetic skyrmion tubes”. In:
Nature communications 11.1 (2020), pp. 1–8 (cit. on p. 27).
[95] Mahdi Mehrnia, Jeremy Trimble, and Jesse Berezovsky. “3D Kerr Mi-
croscopy of Magnetic Vortex States in the Presence and Absence of Uniaxial
Anisotropy”. In: Conference on Lasers and Electro-Optics. Optica Publish-
ing Group, 2020, JW2F.1. doi: 10 . 1364 / CLEO _ AT . 2020 . JW2F . 1. url:
https://opg.optica.org/abstract.cfm?URI=CLEO_AT-2020-JW2F.1 (cit. on
p. 28).
[96] DA Allwood et al. “Magneto-optical Kerr effect analysis of magnetic nanostruc-
tures”. In: Journal of Physics D: Applied Physics 36.18 (2003), p. 2175 (cit. on
p. 28).
[97] Mark R Freeman and Wayne K Hiebert. “Stroboscopic microscopy of mag-
netic dynamics”. In: Spin Dynamics in Confined Magnetic Structures I. Springer,
2002, pp. 93–126 (cit. on p. 28).
[98] S Imada et al. “Magnetic microspectroscopy by a combination of XMCD and
PEEM”. In: Surface Review and Letters 9.02 (2002), pp. 877–881 (cit. on p. 28).
[99] Frederik Wegelin et al. “Stroboscopic XMCD–PEEM imaging of standing and
propagating spinwave modes in permalloy thin-film structures”. In: Surface
Science 601.20 (2007). Proceedings of the Fifth International Conference on
LEEM/PEEM, pp. 4694–4699. issn: 0039-6028. doi: https://doi.org/10.10
16/j.susc.2007.05.060. url: https://www.sciencedirect.com/science/a
rticle/pii/S0039602807006036 (cit. on p. 28).
[100] Robert Streubel et al. “Equilibrium magnetic states in individual hemispheri-
cal permalloy caps”. In: Applied Physics Letters 101.13 (2012), p. 132419 (cit. on
p. 29).
[101] E Bauer. “Spleem”. In:Magnetic Microscopy of Nanostructures. Springer, 2005,
pp. 111–136 (cit. on p. 29).
[102] E Bauer, T Duden, and R Zdyb. “Spin-polarized low energy electron mi-
croscopy of ferromagnetic thin films”. In: Journal of Physics D: Applied Physics
35.19 (2002), p. 2327 (cit. on p. 29).
[103] Qiang Wu, Chenlu Ji, and Michael Scott Altman. “SPLEEM Imaging of Thin
Film Magnetic Interfacial Effects”. In: (2013) (cit. on p. 29).
[104] E Bauer. “Low energy electron microscopy”. In: Reports on Progress in Physics
57.9 (Sept. 1994), p. 895. doi: 10.1088/0034-4885/57/9/002. url: https://dx
.doi.org/10.1088/0034-4885/57/9/002 (cit. on p. 29).
185
[105] Kazuyuki Koike. “Spin-polarized scanning electron microscopy”. In:Mi-
croscopy 62.1 (Jan. 2013), pp. 177–191. issn: 2050-5698. doi: 10.1093/jmicro
/dfs092. eprint: https://academic.oup.com/jmicro/article-pdf/62/1/17
7/5855419/dfs092.pdf. url: https://doi.org/10.1093/jmicro/dfs092
(cit. on p. 29).
[106] M R Scheinfein et al. “Scanning Electron Microscopy with Polarization
Analysis (SEMPA)”. In: Scanning electron microscopy (Apr. 1990), p. 28 (cit. on
pp. 29, 38).
[107] Teruo Kohashi, Makoto Konoto, and Kazuyuki Koike. “High-resolution spin-
polarized scanning electron microscopy (spin SEM)”. In: Journal of Electron
Microscopy 59.1 (Oct. 2009), pp. 43–52. issn: 0022-0744. doi: 10.1093/jmicr
o/dfp047. eprint: https://academic.oup.com/jmicro/article-pdf/59/1
/43/5853696/dfp047.pdf. url: https://doi.org/10.1093/jmicro/dfp047
(cit. on p. 29).
[108] Hideo Matsuyama and Kazuyuki Koike. “A data acquisition and display sys-
tem for spin-polarized scanning electron microscopy (spin SEM)”. In: Review
of scientific instruments 62.4 (1991), pp. 970–981 (cit. on p. 29).
[109] John Unguris. “Scanning electron microscopy with polarization analysis
(SEMPA) and its applications”. en. In: Experimental Methods in the Physical
Sciences. Vol. 36. Elsevier, 2001, pp. 167–XVI. isbn: 978-0-12-475983-1. doi:
10.1016/S1079-4042(01)80040-0. url: https://linkinghub.elsevier.co
m/retrieve/pii/S1079404201800400 (visited on 11/10/2021) (cit. on pp. 29,
33).
[110] Kazuyuki Koike and Kazunobu Hayakawa. “Scanning Electron Microscope
Observation of Magnetic Domains Using Spin-Polarized Secondary Elec-
trons”. en. In: Japanese Journal of Applied Physics 23.Part 2, No. 3 (Mar. 1984),
pp. L187–L188. issn: 0021-4922. doi: 10.1143/JJAP.23.L187. url: https
://iopscience.iop.org/article/10.1143/JJAP.23.L187 (visited on
11/10/2021) (cit. on p. 31).
[111] David R. Penn, S. Peter Apell, and S. M. Girvin. “Theory of spin-polarized
secondary electrons in transition metals”. In: Phys. Rev. Lett. 55 (5 July 1985),
pp. 518–521. doi: 10.1103/PhysRevLett.55.518. url: https://link.aps.or
g/doi/10.1103/PhysRevLett.55.518 (cit. on p. 31).
[112] David R. Penn, S. Peter Apell, and S. M. Girvin. “Spin polarization of sec-
ondary electrons in transition metals: Theory”. In: Physical Review B 32.12
(Dec. 1985). Publisher: American Physical Society, pp. 7753–7768. doi: 10.1
103/PhysRevB.32.7753. url: https://link.aps.org/doi/10.1103/PhysRev
B.32.7753 (visited on 10/23/2022) (cit. on p. 32).
186
[113] Demetrios Matsakis et al. “A renaming proposal: “The Auger–Meitner effect””.
In: Physics Today 72.9 (2019), pp. 10–11. doi: 10.1063/PT.3.4281. eprint:
https://physicstoday.scitation.org/doi/pdf/10.1063/PT.3.4281.
url: https://physicstoday.scitation.org/doi/abs/10.1063/PT.3.4281
(cit. on p. 33).
[114] Kenjiro Oura et al. Surface science: an introduction. Springer Science & Busi-
ness Media, 2013 (cit. on p. 33).
[115] Joachim Kessler. Polarized electrons. Vol. 1. Springer Science & Business Media,
1985 (cit. on p. 34).
[116] D. L. Abraham and H. Hopster. “Magnetic probing depth in spin-polarized
secondary electron spectroscopy”. In: Phys. Rev. Lett. 58 (13 Mar. 1987),
pp. 1352–1354. doi: 10.1103/PhysRevLett.58.1352. url: https://link.aps
.org/doi/10.1103/PhysRevLett.58.1352 (cit. on p. 34).
[117] Ethan Rublee et al. “ORB: An efficient alternative to SIFT or SURF”. In: 2011
International conference on computer vision. Ieee. 2011, pp. 2564–2571 (cit. on
p. 51).
[118] Börge Göbel, Ingrid Mertig, and Oleg A Tretiakov. “Beyond skyrmions: Re-
view and perspectives of alternative magnetic quasiparticles”. In: Physics Re-
ports 895 (2021), pp. 1–28 (cit. on p. 75).
[119] Radha Balakrishnan, Rossen Dandoloff, and Avadh Saxena. “Exact Hopfions
in a 3D Heisenberg Ferromagnet”. In: arXiv preprint arXiv:2202.07195 (2022)
(cit. on p. 75).
187