 

 

 

 

THEOﬁ

(01H 91¢”?

This is to certify that the
dissertation entitled

STRUCTURAL AND ELECTRONIC PROPERTIES OF
RARE EARTH SILICIDES ON Sl(001): NANOWIRES
AND 30 ISLANDS

presented by

Chigusa Ohbuchi

has been accepted towards fulﬁllment
of the requirements for the

Ph.D. degree in Physics and Astronomy

 

 

,Bug AD

Major Prcﬁessor’s Signature

g/IQ/Okr,

Date

MSU is an Afﬁrmative Action/Equal Opportunity Institution

digs-.. Ac-Q-.-._.-.-.-.-.-._.-I-._.-.--

.—-_.-4

.-.—-—.—.-.._. —.-._.-o-n-l-a-o-.--o-.--_-—. _

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DatoDuopGS-p. 1 5

STRUCTURAL AND ELECTRONIC PROPERTIES OF
RARE EARTH SILICIDES ON SI(OOl): NANOWIRES AND 3D ISLANDS

By

Chi gusa Ohbuchi

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Physics and Astronomy

2004

ABSTRACT

STRUCTURALAND ELECTRONIC PROPERTIES OF
RARE EARTH SILICIDES ON SI(OOl): NANOWIRES AND 3D ISLANDS

By

Chi gusa Ohbuchi

Deposition of rare earth (RE) metals on the silicon(001) surface at elevated
temperature results in the formation of RE silicide islands coexisting with a reconstructed
substrate surface. This thesis presents the growth behavior of holmium, samarium, and
dysprosium on Si(OOl) and the electronic properties of the resulting surface structures.

The growth of Ho and Sm on Si(001) at 600°C was studied using scanning tunneling
microscopy (STM) and low energy electron diffraction.

Ho grows in a Stranski-Krastanov mode with highly elongated nanowires (W8) and a
2x4 reconstructed substrate at low metal coverage. As the coverage increases,
three-dimensional (3D) compact silicide islands coexist with the W3. A 2x7
two-dimensional (2D) phase appears in the medium coverage range, and it always
coexists with the 2x4 phase. Scanning tunneling spectroscopy (STS) data show that
HoSiz NWs are more metallic than the 3D silicide islands or the reconstructed substrate.

Srn/Si(001) induces two different surface reconstructions: a lower coverage 2x3 phase
and higher coverage 3x2/c(6x2) phases. In the high coverage regime, large 3D silicide

islands are observed instead of elongated NW 3.

The topographical evolution of 3D Dy silicide structures and the 2D reconstructed
surface in the temperature range of 600 to 750°C was investigated by STM and low
energy electron microscopy (LEEM). The initial surfaces were covered in the 2x7
superstructure plus NWs of uniform width deﬁned by the 2x7 unit cell. Post—growth
annealing at 700°C increases the number of 3D islands and the average island size. At
the same time, the number of NWs decreases and surface reconstruction is swept away.
Therefore, 3D silicide islands are a stable phase whereas NWs are a metastable phase at
700°C.

Ex-situ transport properties of the grown RE silicide ﬁlms were correlated with film
morphology as observed by STM. A 1 nm thick ﬁlm of interconnected DySiz islands
shows surface resistivity at 4.2 K similar to the reported bulk silicide value. A sample
with sparse, disconnected islands shows higher surface resistivity than the 1 nm ﬁlm but

lower than that of the clean Si surface.

ACKNOWLEDGMENTS

First of all, I would like to thank my major advisor Professor Jun Nogami for having
given me an opportunity to work in his group. I decided to do my research at Michigan
State University because Dr. Nogami was there. He generously accepted me as his
student even though I did not have enough background of surface physics. Without his
offer, I could not even come to the US. and my Ph.D. life would never have stated. I
remember the days when he explained to me how STM works in JAPANESE because my
English comprehension level was poor at that time. I was so grateful to him for his
patience and generosity. I also thank him for giving me his time, support, and valuable
advise, and feedback on my Ph.D. work.

I would like to thank my committee members, Professor Norman Birge, Professor Jack
Baldwin, Professor Phillip Duxbury, Professor Thomas Glasmacher, and the previous
committee, Professor Carl Bromberg who is currently on the Sabbatical, for their helpful
comments and suggestions. Especially, I thank Dr. Birge for providing me technical
assistance and knowledgeable advice on transport measurements.

There are several other people whom I would like to thank for helping me with
transport measurements. Dr. Reza Loloee, a SQUID (superconducting quantum
interference devices) specialist in Physics, made pre-deposited Ta and Ti contacts in the
sputtering system. Michael Crosser and Ion Moraru, graduate students in Dr. Birge’s lab,
helped me with preparations for transport measurements in the clean room. They
always gave me a hand whenever I needed them, even though they are busy with their

own research.

iv

 

I would like to express my appreciation to Dr. Wacek Swiech who is a LEEM specialist
in the Frederick Seitz Materials Research Laboratory at the University of Illinois for his
LEEM operation and image acquisition. I really enjoyed doing new experiments with
him although his lab was extremely cold.

I thank former and current coworkers, Dr. Bangzhi Liu, Michael Katkov, Gangfeng Ye,
and Jiaming Zhang for their kind support and friendship. Bangzhi and Micheal taught
me how to do the STM experiments. Besides research, we used to have a cookout on
summer weekend for fun and relaxing. I am really thankful to Gangfeng and Jiaming
for encouraging me throughout the writing of this thesis.

I also thank Professor Evangelyn Alocilja in Biosystems and Agricultural Engineering
for her continuous encouragement, prayer, and support for my successful research
progress and future proposals.

Sean Davis and Beth Czischke are my best American friends and also my best English
teachers. I learned a lot about American cultures and traditions from them. I am
grateful to them for making my life more fulﬁlling and happier.

Finally, I would like to thank my ﬁancé Yuzuru Honda and my family in Japan for their
loving and caring about me all the time. My parents sent me my favorite Japanese food
for my energy source from time to time. Yuzuru has been waiting for the completion of
my Ph.D. since I started the program, which becomes a powerful motivation for me to
work hard and ﬁnish my Ph.D. within ﬁve years. I have been always feeling his
overwhelming love and support although we are far apart. My gratitude to him is
beyond words.

I thank all the other friends in Japan and the US. who cheered me up.

N- . :,

 

 

TABLE OF CONTENTS
LIST OF TABLES ................................................................................ viii
LIST OF FIGURES ................................................................................. ix
ABBREVIATIONS ............. '. .................................................................. xvi
CHAPTER 1 INTRODUCTION ............................................................. 1
1.1 Rare earth (RE) silicides and self-assembled RE disilicide NWs ................. 1
1.1.1 The rare earths ...................................................................... 2
1.1.1.1 RE silicides ..................................................................... 4
1.1.2 Crystal structures of RESiz.x ...................................................... 5
1.1.2.1 Epitaxial growth of RE silicides on Si substrate .......................... 10
1.1.2.1.1 Epitaxial growth of RE silicides on Si(lll) ........................... 10
1.1.2.1.2 Epitaxial growth of RE silicides on Si(001) .......................... 12
1.1.2.2 Schottky barrier height of RE silicides on Si .............................. 16
1.1.2.3 Electrical resistivity of RE silicides ........................................ 17
1.2 Organization of the thesis ............................................................... 22
CHAPT ER 2 EXPERIMENTAL ............................................................ 23
2.1 Experimental and research methods ................................................... 23
2.2 Scanning tunneling microscopy (STM) ................................................ 24
2.2.1 Basic principles of STM ............................................................ 24
2.2.2 STM data acquisition and image processing .................................... 27
2.2.3 Scanning Tunneling Spectroscopy (STS) ....................................... 28
2.3 Low energy electron microscopy (LEEM) ............................................ 30
2.3.1 Basic physics of the low-energy electron interaction with surface: Elastic
scattering and inelastic scattering ................................................. 32
2.3.2 LEEM instrumentation and image acquisition .................................. 33
2.4 Si (001) surfaces .......................................................................... 35

CHAPTER 3 HOLMIUM GROWTH ON SI(001): SURFACE
RECONSTRUCTIONS AND NANOWIRE FORMATION......... 40

3.1 Results and discussion ................................................................... 40
3.1.1 Coverage dependence ............................................................... 40
3.1.2 Nanowires and 3D islands .......................................................... 42
3.1.3 Two-dimensional substrate surface reconstructions ............................ 47
3.1.4 Ho content of the 2x4 substrate reconstruction and the nanowires. . . . . . .....51
3.1.5 Scanning tunneling spectroscopy ................................................. 53

3.2 Conclusions ................................................................................ 55

vi

CHAPTER 4 SAMARIUM INDUCED SURFACE RECONSTRUCTIONSS OF

SI(001) ............................................................................ 56

4.1 Results and discussion .................................................................. 56
4.1.1 Coverage dependence ............................................................ 56
4.1.2 General surface structure evolution with metal coverage .................. 57
4.1.3 The low coverage 2x3 and zigzag chain structures .......................... 61
4.1.4 The medium coverage regime 3x2 and C(6x2) phases ...................... 66

4.2 Conclusions .............................................................................. 70

CHAPTER 5 TOPOGRAPHIC EVOLUTION OF DY SILICIDE ON SI(001):
SHAPE TRANSITIONS FROM NANOWIRES TO 3D ISLANDS

.................................................................................... 71

5.1 Results and discussion .................................................................. 71
5.1.1 STM experiments .................................................................. 71

5.1.2 LEEM experiments ................................................................ 78

5.2 Discussion ................................................................................ 85

5.3 Conclusions .............................................................................. 89
CHAPTER 6 TRANSPORT MEASUREMENTS ........................................ 91
6.1 Sample preparation ...................................................................... 91
6.1.1 Pre-deposited Ta (or Ti) silicide contacts ....................................... 92

6.1.2 In-situ Au contacts ................................................................. 95

6.1.3 Amorphous Ge passivation ....................................................... 97

6.2 Transport measurements: Results and discussion ................................... 98
CHAPTER 7 FUTURE WORK ............................................................. 107
7.1 Future work on growth kinetics ...................................................... 107

7.2 Reliability and reproducibility of electrical contacts for transport measurements
............................................................................................ 108

7.3 Single nanowire measurements ...................................................... 109

7.4 Temperature dependence and magnetic-field dependence of magnetic,
structural, and transport properties of low-dimensional rare earth silicides... 110

APPENDICES ..................................................................................... 117
REFERENCES ..................................................................................... 128

vii

Table 1.1

Table 1.2

Table 1.3

Table 1.4

Table 6.1

Table 6.2

Table 7.1

Table 7.2

Table A1

LIST OF TABLES

Periodic table of the elements ........................................................ 3
RESiz lattice parameters and mismatches with Si substrate ..................... 8
Schottky barrier heights, (DB, of RE metals and silicides on n- and p-type Si

substrates. ............................................................................... 19
Resistivity of RE silicides. ............................................................ 20
Dy and Ho ﬁlm conductivity at 4.2 K vs. metal coverages. All the values

correspond to those in Figure 6.12. There are no values for the conductivity
of bulk Ho silicide in the literature, but the conductivity of the RE silicides lie

in the range 4.5x1o3 to 6.7x105 (sz-cm)‘l at 4.2 K as in Table 1.4. ........... 104
Summary of transport samples measured. Data for the shaded samples are
collected in Figure 6.12 and Table 6.1. ........................................... 105
Bulk properties of RE elements. ................................................... 113
Magnetic properties of RE disilicides. ............................................. 114

Structure of the H-saturated Si(001) surface and H exposure conditions. 1 19

viii

LIST OF FIGURES

CHAPTER 1

Figure 1.1

Spatial models for (a) the hexagonal Ale structure, and (b) tetragonal ThSiz
structure or orthorhombic GdSiz structure. (c) Projection of Ale structure
along the chex-axis. ((1) Projection of ThSizlGdSiz along the b-axis;
Introduction of shear planes in Ale generates the ThSiZ/GdSiz structure. .. 7

Figure 1.2 Structural model for Si(111)1xl-RE: (a) top view and (b) side view. 12
Figure 1.3 Structural model of Si(001)1xl-RE: (a) top view and (b) side view. The
orientation relationship between hexagonal Ale and NW structures are
shown in (c). .......................................................................... 13
CHAPTER 2
Figure 2.1 Wave function I” for an electron with kinetic energy E tunneling a thin
potential barrier of height V0 and width s. The wave function is oscillatory
in the left and right regions of the barrier. Within the barrier, the wave
function decays exponentially. ..................................................... 25
Figure 2.2 Energy diagram for tunneling between a tip and a sample. Tunneling is
possible only when there are empty states on the right. Such empty states
are created when eV is applied to lower the Fermi level on the right.
......................................................................................... 26
Figure 2.3 Schematic diagram of an STM. A metallic tip is mounted on a
piezoelectric tripod actuator which can move the tip in the x—y plane of the
sample surface, and in the z direction normal to the surface, with atomic
resolution. ............................................................................. 28
Figure 2.4 Schematic of an STS experiment. An ST M tip is held above the sample,
and the current is measured while the voltage is ramped. This measures
the conductivity in the surface normal direction. ................................ 30
Figure 2.5 Diffraction contrast in LEEM (a) Bright ﬁeld, (b) tilted bright ﬁeld, (c) dark
ﬁeld. ................................................................................... 31
Figure 2.6 Schematic of LEEM instrument (IBM Type I). ................................. 34
Figure 2.7 Schematic ray diagrams for LEEM. .............................................. 34
Figure 2.8 (a) Diamond structure of silicon, and (b), (c) top view of model of the dimer

reconstruction on Si(001). .......................................................... 35

ix

Figure 2.9

Figure 2.10

Figure 2.11

(a) ﬁlled- and (b) empty-state STM images of Si(001) surface with a side

view of Si dimer model. 2x1 unit cells corresponding to the dimer position
are marked. ........................................................................... 36

Schematic drawing of Si(001) 2X1 reconstructed surface consisting of three
terraces. Si atoms on top, second, third, and fourth layers are shown with
black, dark gray, light gray, and white colors, respectively. The height of
the terrace lowers from left to right. The directions of surface stress are
marked with black arrows. ......................................................... 37

LEED pattern of a Si(001) surface with 2x1 superstructure. (%, 0) and

(g— , O) diffraction spots and (0, %) and (0, %) diffraction spots are

indicated with white and gray circles, respectively. 2x1 unit cell is marked
with a white rectangle as well. ..................................................... 38

Figure 2.12 Dark-ﬁeld LEEM image of Si(001) surface with opposite contrast. Images
(a) and (b) are acquired using one of the diffraction spots with (a) white and
(b) gray circles, respectively. ...................................................... 39
CHAPTER 3
Figure 3.1 Coverage dependence of LEED patterns and topography for Ho growth on
Si(001) at 600°C. .................................................................... 41
Figure 3.2 Two 400x400 nm2 images of HoSiz W8 and 3D islands. (a) NWs formed
by 0.36 ML of Ho. Each NW grows on a single terrace. (b) As the
coverage is increased (0.90 ML), 3D compact silicide islands are found. .. 42
Figure 3.3 Relationship between coverage and the percentage of surface covered with
Ho surface reconstruction, NW 8, and 3D islands. .............................. 43
Figure 3.4 (a) Triangular (left) and rectangular (right) NWs at 0.3 ML (200x200 nmz).
Steps on the substrate flow to accommodate the NWs. Lines A-B and C-D
show the locations of the cross sections shown in (b) and (c). ................ 44
Figure 3.5 (a) Bundled NWs with second layer growth at 0.36 ML (90x90 nmz). (b)
Atomic resolution on bundled NW 3 (12x14 nmz). A single c(2x2) unit cell
is marked with a white square. ..................................................... 45
Figure 3.6 The distributions of (a) the width and (b) the height of single NW 8 and 3D
islands. ................................................................................ 46
Figure 3.7 (a) Empty- and (b) ﬁlled-state images of slightly different areas of the same

surface with 0.55 ML Ho showing both 2x4 and 2x7 structures (30x30 nmz).
The position of gray circle on the NW in (a) corresponds to that in (b). ..... 47

Figure 3.8

Figure 3.9

Figure 3.10

Figure 3.11

(a) Empty-state (1.76 V) and (b) ﬁlled-state (-1.76 V) images of the 2x4
phase in the reconstructed surface shown in Figure 3.7 (10x10 nmz). (a)
and (b) correspond to the square areas numbered by 1 and 3 in Figure 3.7,
respectively. (c) Line drawing showing the empty-state (gray circles) and
ﬁlled-state (dark circles) maxima of the 2x4 structure. Clean 2x1 Si dimer
positions are shown as a reference on the topmost row. ....................... 48

(a) Empty-state (1.76 V) and (b) ﬁlled-state (-l.76 V) images of the 2x7
phase extracted from Figure 3.7. (a) and (b) correspond to the square areas
numbered by 2 and 4, respectively. A 2x7 unit cell consists of three types
of maxima in rows labeled A and B in both images. (c) Line drawing
showing the empty-state (gray circles and oblongs) and ﬁlled-state (dark
circles) maxima of the 2x7 structure. 2x1 Si dimer positions from the
clean surface are shown as a reference on the topmost row. .................. 50

Comparison of experimental coverage with calculated coverage under
different assumptions of Ho content (see text). ................................. 52

(a) The STS I-V curves and normalized spectra for a NW, 3D island, and the
reconstructed substrate. The surface topography is shown in (b) (150x150
nmz). The image (c) is a partial image of (b). Each STS spectrum was
averaged from the STS data acquired in the areas shown by solid or dashed
lines in the STM image shown in (c) (40x40 nmz). (d) Magniﬁcation of
the STS I-V curves at near-zero bias voltage. .................................... 54

CHAPTER 4

Figure 4.1

Figure 4.2

Figure 4.3

Figure 4.4

Figure 4.5

Coverage dependence of LEED patterns and topography for Sat/Si(001) at
600°C. ................................................................................. 56

LEED pattern of Sm/Si(001) at 0.54 ML shows 2x3 (black rectangle) and
c(6x2) (black arrows) periodicities. The LEED pattern was obtained from
the surface shown in Figures 4.8 (c) and 4.9 (a). The electron beam energy
is 63 eV. ............................................................................... 57

(a) Empty-state (1.90 V) and (b) ﬁlled-state (-1.91 V) STM images of the
same area showing 2x3 phase and the zigzag chain structures at 0.067 ML
(15x15 nmz). Cross-sectional proﬁle along the line A is shown on the
bottom of each image. ............................................................... 58

15x15 nm2 image of a surface at 0.59 ML. Both a 3x2 phase and a highly
defected Si 2x1 surface are seen. The bias voltage is —1.77 V. ............ 60

Sm silicide island at 1.46 ML (300x300 nmz). .................................. 61

xi

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

(a) Empty-state (1.57 V) and (b) ﬁlled-state (-l.57 V) STM images in the
same area showing 2x3 phase at 0.067 ML (10x10 nmz). There are
out-of-phase boundaries shown with dashed lines. Two 2x3 unit cells are
marked with white rectangles. (c) Line drawing showing the empty-state
(white circles and ovals) and filled—state (gray circles and ovals) maxima of
the 2x3 structure. .................................................................... 62

(a) Filled-state (-1.82 V) image of zigzag chain structures with buckled Si
dimers below 0.1 ML (10x10 nmz). The 2x3 unit cell is also shown as a
pair of two maxima on the lower right area. (b) Line drawing illustrating
the zigzag chain structures with surrounding buckled Si dimers. This
dimer buckling causes c(4x2) periodicity. A c(4x2) unit cell is marked
with a black rectangle. ............................................................. 65

(a) Large-scale STM image of a surface consisting of two types of step
edges. Straight step edges (SA step) run along the dimer row direction on
terrace A, and terraces B have jigsaw-shaped step edges (83 step). (b) Both
types of terraces consist of 3x2 and c(6x2) phases and Si 2x1 dimer with
zigzag defects shown with white arrows. (c) Close-up image of zigzag
defects on Si dimer rows. The ends of four zigzag defect structures are
marked with black and white arrows. All three images were at different
Sm coverages: (a) 0.93 ML, (b) 0.59 ML, and (c) 0.54 ML. The image
sizes are (a) 300x300 nmz, (b) 50x50 nmz, and (0) 25x25 nmz. Bias
voltages are (a) —2.31 V, (b) -l.77 V, and (c) 1.28 V. ........................ 67

10x10 nm2 images in (a) empty states (1.28 V) and (b) ﬁlled states (-l.77 V)
showing 3x2 and c(6x2) phases at different coverages: (a) 0.54 ML and (b)
0.59 ML. Both 3x2 and c(6x2) unit cells are marked with white rectangles
on each image. (c) Line drawing showing the empty-state (gray circles)
and ﬁlled-state (dark circles) maxima of the 3x2 and c(6x2) structures. Si
dimers are shown on the leftmost row as a reference. .......................... 68

CHAPTER 5

Figure 5.1

STM images of DySiz NWs at 0.58 ML (a) before and (b) after annealing at
600°C for 8 minutes. (a) Dy was deposited at 600°C. Almost all the 2D
surface is covered with 2x7 stripe structure and W3 have comparatively
uniform width distribution. 2x7 phase generally disappears after 1 min
annealing at 600°C. (b) After annealing, 2x4 superstructure coexists with
bare silicon. More bundled NWs are observed and the width of NWs

becomes wider. There is also more second layer growth on top of the NWs.
......................................................................................... 72

xii

Figure 5.2

Figure 5.3

Figure 5.4

Figure 5.5

Figure5.6

Figure 5.7

Figure 5.8

Figure 5.9

Small-scale images of (a) and an) in Figure 5.1 showing (a) 2x7 (-1.80 V)
and (b) 2x4 (+1.04 V) reconstructed surface around NWs. Two unit cells
are marked with white rectangles in each image. ............................... 73

Annealing behavior of Dy grown on Si(001) at 0.58 ML. 3D structures of
DySiz transit from large 3D islands by annealing. 2x7 superstructure
dominates on the 2D surface after rapid deposition of Dy. After 1 min
annealing at 600°C, 2x7 superstructure is replaced by 2x4 superstructure
and bare Si. Less 2x4 structure appears with higher temperature annealing
or longer annealing time. ........................................................... 74

Transition from NWs to large 3D islands of DySiz by annealing. (a) 0.58
ML of Dy was deposited at 600°C, (b) 1St annealing at 600°C for 8 min, (c)
2nd annealing at 600°C for 20 min, ((1) 6h annealing at 700°C for 10 min,
and2 (e) 7‘h annealing at 720°C for 10 min. All image sizes are 400x400
nm . ................................................................................... 77

(a) The LEED pattern of Dy grown on Si(001) at ~600°C (0.78 ML). 1x7
phase and 1/2 order streaks are shown. The electron beam voltage is 60 V.
(b) Dark-ﬁeld LEEM image (2 eV) taken from 2/7 spot shown with the
white arrow in (a). The ﬁeld of view is 4 urn. (c) Bright-ﬁeld LEEM
image (9 eV) in the same area of (b) showing the stripe structure. Contours
of terraces are drawn with white lines. The stripes are oriented at an angle
of 90 degrees on each terrace, which correspond to the NW 3. ............... 78

Sequence of LEEM images showing the growth of DySiz islands (0.78 ML).
The ﬁeld of view is 4 pm. The white arrows show the same island.

Small grain structures start appearing at ~700°C and grow as islands. BF:
Bright ﬁeld, TBF: Tilted bright ﬁeld. ............................................ 80

Annealing duration vs. temperature. Annealing temperature was gradually
increased from RT (00’00”) to 750°C. The black arrow shows the time
range during which the islands are observed as shown in Figures 5.8 and 5.9.
......................................................................................... 81

Time evolution of the length and the width of 3D islands and the

surrounding Si depletion areas in different growth types during annealing.
......................................................................................... 83

Change in the island size by annealing (0.78 ML of Dy). Black, gray, and
white marks indicated large, medium, and small islands, respectively. The
islands numbered from 1 to 13 correspond to the islands with the same
numbers in Figure 5.6 (f). The labels (a)~(f) are marked for the time at
which the LEEM images in Figure 5.6 were taken. ............................ 84

xiii

CHAPTER 6

Figure 6.1

Figure 6.2

Figure 6.3

Figure 6.4

Figure 6.5

Figure 6.6

Figure 6.7

Figure 6.8

Figure 6.9

Figure 6.10

Sample preparation procedure for transport measurements. Si and Ta (or
Ti) depositions are done in the high vacuum chamber (Step 1 and 2).
Sample ﬂashing, deposition of RE metals, Au contacts, and amorphous Ge
are performed in the UHV chamber (Step 4~8). ................................ 92
(a) Atomic Force Microscopy (AFM) image of the boundary between TaSiz
and Si substrate after annealing 100 nm thick Ta on silicon. (b)
Cross-sectional proﬁle traced along a white line in (a) showing a deep
trench. (Width = 23 um, depth = 800 nm) ................................... 94

Figure 6.3 Scanning Electron Microscopy (SEM) images of (a) TaSiz and
(b) TiSiz ﬁlms. For (a), 200 nm of Si + 100 nm of Ta ﬁlms are annealed
after deposition. For (b), 100 nm of Si + 50 nm of Ti ﬁlms are annealed
after deposition. ...................................................................... 94

(a) Sputtering mask for Ta (or Ti) deposition. (b) Shadow mask for Au
deposition. ............................................................................ 95

(a), (b) Optical microscope images of Au contacts covering beneath of TiSiz
contacts. (b) Close-up image of the area marked with a white rectangle in (a).
(0) 3D STM image of the Au contact. The formation of many small Au
islands is observed. ................................................................. 96

Schematic diagram of Au evaporation with a shadow mask. ................. 97

STM images of the Au boundaries (200x200 nmz). The density of Au
increases from (a) to (c). DySiz NWs can be seen underneath of Au. ..... 97

(a) Before Ge deposition: DySiz NWs with interconnected 3D islands.
(180x180 nmz) Coverage is unknown. (b) After 5 ML of Ge deposition
on top of NWs. (200x200 nmz) ................................................. 98

(a) The STM image of 3 ML of interconnected Dy silicide islands. (b)
Experimental setup for transport measurements. Surface resistances were
measured at liquid helium temperature (4.2 K). (c) In order to measure the
change in surface resistance of the sample (a), the sample was scratched in
the horizontal and vertical directions. The numbers labeled on the bottom
0, l, and 2 indicate ‘before scratch’, ‘scratched in the horizontal direction’,
and ‘scratched in the both directions’, respectively. ........................... 99

(a) Change in 4-terminal surface resistance of interconnected Dy silicide
islands due to the sample scratching. A diagram shows the resistances in
the vertical (open circles) and horizontal (solid circles) directions,
respectively. Dashed lines on top are the surface resistance of clean Si
surface. (b) Schematic diagrams of each step of sample scratching and
corresponding current flow. ...................................................... 100

xiv

Figure 6.11 STM images of (a) DySiz NWs at 0.22 ML and (b) interconnected Dy

silicide islands at 3 ML. ........................................................... 102

Figure 6.12 Dy and Ho ﬁlm conductivity at 4.2 K vs. metal coverages. Solid (open)

circles show the surface conductivity of Dy (Ho) on Si(001). .............. 104

CHAPTER 7

Figure 7.1

Figure 7.2

Post growth deposited contacts Step I: Grow sparse network of NWs.
Step 2: Deposit alignment marks: Use AFM to determine the NW/mark
registration. Step 3: Deposit ﬁne Au contacts. .............................. 108

Magnetic structures of heavy RE metals [after Koehler (1972)]. ........... 111

APPENDICES

Figure A1

Figure A2

Figure A3

Schematic of the various possible hydrogenated structures on Si(001)
surface. .............................................................................. 117

(a) Schematic diagram of a Si dimer ‘before’ and ‘after’ H-passivation. A
half-ﬁlled 7t. anti-bonding orbital is completely ﬁlled due to H-desorption.
(b) Empty-state (1.74 V) and (c) ﬁlled-state (-1.82 V) images of
H—passivated Si(001) surface. The left half of the images is the surface
after it was passivated by applying high tip voltage (20x30 nmz).
Cross-sectional proﬁle in the horizontal direction is shown next to each
STM image. ........................................................................ 120

STM image (90x120 nmz) of a H—passivated Si(001) surface. Brighter
areas correspond to an STM-induced H desorption. By operating the tip
position and the voltage, we can write a “MSU” logo! ..................... 121

Figure A4 (a) Before H-passivation: DySiz NWs on Si(001) at 0.80 ML (180x180 nmz).

Figure Bl
Figure Cl

Figure C2

Figure C3

(b) After H-passivation: NWs still survive, although there are small gaps on
the NWs shown by white arrows (50x50 nmz). The change in the electrical

properties of NWs due to the small gaps is unknown. ....................... 122
Schematic diagram of current passing through a thin ﬁlm. .................. 123
Constant-current circuit using a ballast resistor. .............................. 124

(a) Circuit showing a two-terminal resistance measurement. (b) Equivalent
circuit for (a). ...................................................................... 124

(a) Circuit showing a four-terminal resistance measurement. (b)
Equivalent circuit for (a). ......................................................... 125

XV

ABBREVIATIONS

The following abbreviations are used in this thesis.

1D

2D

3D
AFM
BF
DF
LEED
LEEM

RT
SEM
STM
STS

TEM

one-dimensional

two-dimensional
three-dimensional

Atomic Force Microscopy

bright ﬁeld

dark ﬁeld

Low Energy Electron Diffraction
Low Energy Electron Microscopy
monolayer (see p.23 for the deﬁnition)
nanowire

rare earth

room temperature

Scanning Electron Microscopy
Scanning Tunneling Microscopy
Scanning Tunneling Spectroscopy
tilted bright ﬁeld

Transmission Electron Microscopy
ultra high vacuum

xvi

Chapter 1 Introduction

Our research objective is motivated by the potential for application of self-assembled
rare earth (RE) silicide nanowires (NWs) for electronic and photonic devices in the
semiconductor industry and to explore the structural and electrical properties of those
W5 and their growth kinetic and dynamics.

In this chapter, the background and basic ideas of rare earth (RE) silicides and their
growth behavior and electric properties are introduced. The organization of this thesis

will be described in the end of this chapter.

1.1 Rare earth (RE) silicides and self-assembled RE disilicide NWs

RE disilicide NWs were ﬁrstly discovered by Preinesberger et al. in 1998 and they
showed that Dy deposited on Si(001) forms NWs under certain conditions [1].
Unfortunately, this fascinating phenomenon was not noticed for more than 1 year. On
November 1, 1999, the New York Times carried an article reporting that a
Hewlett-Packard research group had produced conductive wires about 10 atoms wide [2].
These NW 3 were actually made of RE silicide [3]. This result was described as part of
an impending revolution in molecular electronics and nanotechnology.

Since then, epitaxial RE silicides grown on silicon have attracted particular interest.
In the study of RE silicides grown on Si(001), it has been reported that so far eight types
of RE silicides can form self-assembled NWs: YSig [4], ScSiz [5], SmSiz [6], GdSiz [5, 7,
8], DySiz [1, 5, 6, 9-12], HoSiz [9, 13, 14], ErSiz [3, 5, 6, 15, 16], and YbSiz [17]. For
the other RE silicide/Si substrate systems, NW formation has been recently found on

Dy/Si(111) [18], Gd/Si(111) [19], and Dy/Si(110) [20]. The formation of W3 is

ascribed to small lattice mismatch in one direction and large lattice mismatch in the
perpendicular direction on Si(001) substrate. The mechanics of nanowire formation will

be explained on page 15.

1.1.1 The rare earths

The rare earths are a group of 15 metallic elements: La (atomic number 57) through Lu
(71) which appear in the extended sixth row of the periodic table as shown in Table 1.1.
Along with Sc (21) and Y (39), they constitute the “RE elements”. RE elements can be
classiﬁed into two groups: the light RE elements (La to Eu) and the heavy RE elements
(Gd to Lu, Sc and Y). The RE metals are typically trivalent, which are distinct by the
progressive ﬁlling of the 4f localized subshell, while the external electrons are shared
between 63 and 5S subshells which determine the conduction band maximum with a total
charge of between two and three electrons under normal conditions [21]. The 5d and 6s
subshells mix in the solid state, and RE metals have a conduction band of hybrid s-d
character; the extent of hybridization varies along the series in a rather irregular way. In
contrast to most of the trivalent RE metals, Sm, Eu, and Yb are divalent. The divalence

of Eu and Yb metals is well explained in terms of very stable conﬁguration of half ﬁlled

(4 f 7) and completely ﬁlled 4f band (4f ‘4 ), respectively [21]. Sm forms divalent or

trivalent compounds depending on the chemical environment, because two valence states

have rather similar energies [22, 23].

EBB. Eon 5E» uoEmemE 8m $52 E8 :3 £033 2808 mm BEEPQQ Be 232: mm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

:8. oZ~E 32:: Enos amen L08 SE; EUR :23 3&3 m2? D: «as Free o<$
=4: .53 Saree ammo ems hnee £3 as. am? Ema. Ems uZS 5% 00mm 3%
35820 53m 2am ESE 85820 mm Em:
mic—10x

25m: E £52. m: 639.: m: 253 255 55o: HE§ mime :mu: mmee DDS. “m3. «m3 is
cm: “<9“ omen 5% 9mg F; wig :5; EM: h: woes ox? 3: «H? Ema :er amen 6%
oxen Fm £er am; amen 5%. BO: m<$ ems. am: am: out? 02? AZ; HNe. >3 amen am:
gem cm: omen 342 00m «0: :NS =03 2% 00: our” 523 .53 >2 F2 em: «08 M2
5.2 5: m2 m: a: 2: 322 «Z:
02.: ma Ow Z» De mm one Em
0mm F

 

 

 

 

 

 

 

 

 

3.5820 05 no 058 Ever—om ﬂ“ 29¢.

1.1.1.1 REsiIicides

The investigation of the intrinsic character of RE silicide systems started to develop in
the early 1980s; interface and thin ﬁlm formation, electronic structure, and transport and
magnetic properties [24-29]. Since then, studies of RE compound layers on
semiconductors have become increasingly popular. Some of their advantages are the
low growth temperatures, the good thermal stability, the low Schottky barrier on n-type
Si substrates [24], and their compatibility to the Si technology. Moreover, these
silicides are characterized by a small lattice mismatch relative to the Si(lll) surface,
ranging from zero to 2.5 %, thereby allowing epitaxial growth with chemically sharp
interfaces and a high degree of crystallinity and structural perfection. Another interest is
that of magnetic properties of RE compounds, associated with a large value of the
magnetic moment of 4f layer of RE elements. These properties are unusual and can be
studied by electrical transport methods, because the magnetic ions interact strongly with
conduction electrons (see Chapter 7).

The RE silicides occur in three major stoichiometries: in metal rich form as RE5313 and
RE5$i4 silicides, as monosilicides RESi, and as disilicides of varying composition, silicon
rich RESiz.x with 09:50.33 (from RE3Si5 to RESiz) [30-34]. From the present viewpoint,
the disilicides are of most interest, since this type of stoichiometry develops upon

annealing of RE overlayers on Si.

1.1.2 Crystal structures of RESih

The crystallographic properties have been described earlier in several papers [30, 35,
36]. As shown in Figure 1.1, three main crystallographic structures are encountered,
depending on the RE metal and on the silicon content: the hexagonal Ale structure
(p6/mmm space group) [Figure 1.1 (a)], the tetragonal ThSiz structure (I4/amd space
group), and its distorted orthorhombic variant GdSiz (Imma space group) [Figure 1.1 (b)].
The structures and lattice constants are listed on Table 1.2. The ﬁrst three RE metals
(La, Ce, and Pr) crystallize in the tetragonal ThSiz or orthorhombic GdSiz structure. The
next RE metals (Nd to Gd) can crystallize all the three structures depending on the RE
elements. The hexagonal AIB2 structure is found mainly in heavy RE compounds like
YbSiz.x and ScSiz.x for a stoichiometry close to R3Si5 [37, 38], and corresponds to the
structure of epitaxial ﬁlms grown on Si(lll). The type of structure adopted depends on
the nature of the RE, on the value of x in RESiz.x and on the temperature.

‘Ale structure hosts the metal atoms in hexagonal planes along the c-axis ([0001]
direction) [Figure 1.1 (a)]. The hexagonal RE planes alternate with Si atoms in a
honeycomb arrangement. In this lattice, Si atoms occupy the interstitial sites between
the hexagonal layers of RE atoms. However, the growth of epitaxially ordered RE3Si5
layers is observed and this bulk-like silicide forms a defected hexagonal Ale structure,
consisting of stacked hexagonal RE planes and graphite-like Si planes with an ordered
arrangement of vacancies at every sixth Si lattice site [39]. Therefore, the actual
compositions of Si atoms vary between 1.67 and 2.00 per RE atom although the
stoichiometry for perfect Ale-type RE silicides would be RESiz.

Tetragonal ThSi2 structure [Figure 1.1 (b)] is closely related to the MB; structure.

Starting from the Ale structure [Figure 1.1 (c)] and introducing shear planes parallel to

(10I0)hex with a shear vector (b.m + cued/2 and a periodicity of ﬂab“, the ThSiz
structure is derived [Figure 1.1 (d)]. As a consequence, the lattice parameters of the two
structures are related:
a = b = (ahex + chcx)/2 and c z 2 J3 am
The orthorhombic GdSiz structure is a deformation of the tetragonal ThSiz structure

with a rt b.

 

Raga}

_I

9%
m

   

 

 

 

 

 

 

 

((1)

Figure 1.1 Spatial models for (a) the hexagonal Ale structure, and (b) tetragonal ThSiz
structure or orthorhombic GdSiz structure. (c) Projection of Ale structure along the
aux-axis. ((1) Projection of ThSizlGdSiz along the b-axis; Introduction of shear planes in
Ale generates the ThSi2/GdSi2 structure.

Table 1.2 RESiz lattice parameters and mismatches with Si substrate. (as, = 3.840 A)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Silicide Structure Lattice parameter and mismatch Ref
a (A) (%) b (A) (%) c (A) (%>
SCSI” hexagonal Ale 3.66 (-4.69) 3.87 (+0.78) [40]
tetragonal ThSi2
orthorhombic GdSiz
YSiz hexagonal Ale 3.842 (+0.05) 4.144 (+7.92) [36]
tetragonal ThSiz 4.04 (+5.21) 13.42 [40]
orthorhombic GdSi2 4.05 (+5.52) 3.954 (+2.97) 13.36 [40]
LaSiz hexagonal Ale
tetragonal ThSiz 4.31 (+12.2) 13.80 [41]
orthorhombic GdSiz 4.271 (+11.2) 4.182 (+8.91) 14.035 [42]
CeSiz hexagonal Ale
tetragonal ThSiz 4.192 (+9.17) 13.86 [42]
orthorhombic GdSiz 4.19 (+9.11) 4.13 (+7.55) 13.92 [42]
PrSi; hexagonal A132
tetragonal ThSiz 4.22 (+9.90) 13.71 [42]
orthorhombic GdSiz 4.20 (+9.38) 4.16 (+8.33) 13.76 [42]
NdSiz hexagonal Ale 4.12 (+7.29) 4.44 (+156) [43]
tetragonal ThSiz 4.111 (+7.06) 13.56 [42]
orthorhombic GdSiz 4.155 (+8.20) 4.125 (+7.42) 13.67 [42]
PmSiz hexagonal Ale
tetragonal ThSiz
orthorhombic GdSiz
Sm3Si5 hexagonal Ale 3.90 (+1.64) 4.21 (+9.64) [44]
SmSiz tetragonal ThSiz 4.08 (+6.25) 13.51 [421
orthorhombic GdSi2 4.105 (+6.90) 4.035 (+5.08) 13.46 [421
EuSiz hexagonal Ale
tetragonal ThSiz 4.29 (+11.7) 13.66 [421
orthorhombic GdSiz

 

 

Table 1.2 (cont’d)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Silicide Structure Lattice parameter and mismatch Ref
a (A) (%) b (A) (%) c (A) <%)
GdSiz hexagonal Ale 3.877 (+0.96) 4.172 (+8.65) [36]
tetragonal ThSiz 4.10 (+6.77) 13.61 [42]
orthorhombic GdSiz 4.09 (+6.51) 13.44 13.44 [42]
TbSiz hexagonal Ale 3.847 (+0.18) 4.146 (+7.97) [36]
tetragonal ThSiz
orthorhombic GdSiz 4.057 (+5.65) 3.97 (+3.39) 13.375 [42]
DySiz hexagonal A1B2 3.831 (-O.23) 4.121 (+7.32) [36]
tetragonal ThSiz 4.03 (+4.95) 13.38 [42]
orthorhombic GdSiz 4.04 (+5.21) 3.95 (+2.86) 13.34 [421
HoSiz hexagonal AlB2 3.816 (-0.63) 4.107 (+6.95) [36]
tetragonal ThSiz 3.964 (+3.23) 13.297 [45]
orthorhombic GdSiz 4.03 (+4.95) I 3.944 (+2.71) 13.30 [42]
ErSiz.x hexagonal Ale 3.79 (-1.30) 4.09 (+6.51) [46]
(“‘05) tetragonal ThSiz 3.96 (+3.13) 13.26 (+245) [46]
orthorhombic GdSi; 5.6 (+458) | 13.26 (+245) 5.6 [461
5.6 (+453) | 13.26 (+245) 11.2 [461
TmSiz hexagonal Ale 3.768 (-l.88) 4.070 (+5.99) [36]
tetragonal ThSiz
orthorhombic GdSiz
YbSiz hexagonal Ale 3.784 (-l.46) 4.098 (+6.71) [36]
tetragonal ThSiz
orthorhombic GdSiz
LuSiz hexagonal Ale 3.747 (-2.42) 4.046 (+5.36) [36]
tetragonal ThSiz
orthorhombic GdSiz

 

 

1.1.2.1 Epitaxial growth of RE silicides on Si substrate

Since RE silicides are extremely reactive with oxygen, which hindered early attempts
at fabrication, both growth technique and epitaxial conditions are very important for
improving epitaxial quality in silicides. Continued in-situ work in UHV systems is
necessary for the production of RE silicide ﬁlms of adequate quality.

RE silicides can be grown on Si substrates by codeposition of RE and Si at elevated
temperatures or by solid-phase epitaxy, i.e. by RE deposition on Si and subsequent
annealing [25, 26, 36, 47-49]. In the latter case, the diffusion processes necessary for
silicide formation give rise to a complex interplay of energetic and kinetic processes,
requiring a detailed control of the process parameters.

A Si(lll) substrate favors the epitaxial growth on it of the RE silicide phase with the
Ale structure, because it contains three equivalent directions (the three <110>
directions) related with a three-fold symmetry [Figure 1.2 (a)]. In the case of a Si(001)
substrate, there are only two <110> directions related with a two-fold symmetry [Figure

1.3 (a)]and thus a tetragonal phase is favored for the formation of thin ﬁlms.

1.1.2.1.1 Epitaxial growth of RE silicides on Si(lll)

Small lattice mismatch to Si(lll) substrates makes epitaxial layer growth possible and
ensures an ideal interface structure with minimized defect states [36]. For the growth of
trivalent RE metals (except for Sm, Eu, and Yb) on Si(lll) at elevated temperatures, a
p(1x1) interface is formed in the submonolayer range and consists of a bulk-like Si
substrate with a single RE T4 site (second-layer substrate atom) with a rotated Si bilayer

on top on the RE [50, 51] as shown in Figure 1.2. The growth of divalent RE metals

10

(Sm, Eu, and Yb) on Si(lll) behaves differently. Low energy electron diffraction
(LEED) and scanning tunneling microscopy (STM) studies show 3x1, 3x2, 5x1, and 7x1

periodicities for Sm [52] and Yb [52-54], and 3x1, 3x3, and 5x5 for Eu [55].

As the coverage increases, J3 x J3R30° structure has been mainly observed for most

of the RE metals [51, 56, 57] although two-dimensional (2D) Eu or Yb layer on Si(lll)
has 2x1 reconstruction [58, 59]. The J3 xJ3R30° structure is assigned to an ordered

superlattice of Si vacancies producing a «[3 xJ3 mesh in the silicide ﬁlm [60]. This
silicide ﬁlm is identiﬁed to be hexagonal RESiz.x with Ale structure where the Si atoms
are arranged in a hexagonal plane normal to the silicide c-axis, which has a good
matching of the lattice parameters (as,= 0.384 nm) of the Si(lll) surface. This allows
interface conﬁgurations with lattice mismatches of less than 2 %, resulting in high quality
epitaxial growth. The orientation relationships of hexagonal RESiz.x on Si(lll) [46,
61-63] are determined to be:
RESi2-x(0001)// Si(lll) and RESi2-x[1 100] // suizi]

as shown in Figure 1.2.

11

00 9,. 00 00 [11201 QREmetar

  

 

o 0 ._ 0 O 0 -
,/ ‘ _ 0 1st SI layer
.0 0‘; .0 .0 .0 hex [110°] g 2nd St layer
0 O L 0 0 0 [101151
9 D '0/ O i _ _
(a) e o o o O 0 0 “2115‘
0.384nm [11115i
Chex
[1100]
[1120]
[111]sr
(b) [I2I]Si
nails:

Figure 1.2 Structural model for Si(lll)1x1-RE: (a) top view and (b) side view.

1.1.2.1.2 Epitaxial growth of RE silicides on Si(001)

Many studies of thin RE silicide film growth on Si(lll) have been reported due to the
easy epitaxial growth in this crystallographic orientation. However, until the discovery
of self-assembled RE disilicide NWs [1, 3, 13], only a few studies were performed on
Si(001) [61, 64-66] which exhibits a square unit cell as shown in Figure 1.3 (a). Most
prior thin ﬁlm growth studies looked at ﬁlm thicknesses between 25 and 200 nm, where
the most interesting features were the crystallographic phases present, and the epitaxial

relationship with respect to the substrate.

0 {.13 RE metal

Si
0 [0001] Chex

9 [1120]
[1i00]

(a)

é—p [11in]

[0001] Cher

[001]Si

"a651,: e'e % e

[110181
[11019

(b)

 

 

 

 

ahex

Figure 1.3 Structural model of Si(001)1x1-RE: (a) top view and (b) side view. The
orientation relationship between hexagonal Ale and NW structures are shown in (c).

As already mentioned before, a Si(001) substrate contains only two <110> directions
and thus the growth mode is different from Si(lll). In the case of ErSiz thin films, the
lattice parameters of hexagonal Ale—type ErSiz (ahex = 0.379 nm and cm = 0.409 nm)

are relatively close to the lattice parameter of Si(001) (as, = 0.384 nm) (see Table 1.2).

13

Epitaxy with the ahex- and chcx-axes parallel to the two mutually perpendicular <110>51

directions of the substrate has two obvious orientation variants, with the ahex- and

chex-axes interchanging directions in the two variants. Although there are several

epitaxial modes as observed in hexagonal ErSiz.x [67], the orientation relationships of

hexagonal RESiz.x (h-RESi2-x) on Si(001) [46, 61-64] are generally determined to be:
h—RESi2-x(1I00) // Si(001) and 114115312.x [0001] // Si[lIO]

or h-RESi2-x(1 100) // Si(001) and h-RESiz.x [1150] // Si[lIO] (rotated by 90°)

The vacancy ordering superstructure of unit cell (ch/3 aJ3 2c) is also found in
epitaxial hexagonal YSi2-,,, TbSi2-x, DySi2-x, and ErSiz.x thin ﬁlms with Ale structure on
Si(001) [61].

The tetragonal ThSi2 and orthorhombic GdSiz structures are grown in the similar way.
Several RESiz.x thin ﬁlms such as DySiz.x [68], EI‘SIz-x [46, 62], and LuSiz.x [69] on
Si(001) form tetragonal or orthorhombic structures and the orientation relationships of
tetragonal / orthorhombic RESiz_x (t-RESiz-x) on Si(001) are determined to be:

t-RESi2-x(010) // Si(001), t-RESiM [100] // Si[110], and t-RESiz.x [001] // Si[ 1 I0]
For the Gd/Si(001) system, the coexistence of epitaxial h-GdSiz.x and o-GdSiz.x , and
polycrystalline o-GdSiz is observed depending on the growth conditions [48, 63]. The
formation of an epitaxial H0812 layer by solid phase reaction is reported, but no structural
detail is given [70].

When the amount of metal deposited onto the Si surface is insufﬁcient to create a
continuous thin ﬁlm, then the RE silicide grows in what can be described as the
Stranski-Krastanov mode. In reality, the RE grows as a 2D metal induced surface

reconstruction involving at most 1 ML of metal, and 3D silicide nanostructures that can

14

be described as either NWs or islands [1, 3-9, 11-14, 17]. Various types of periodicities
due to the surface reconstruction are observed at low metal coverages: 2x3 and 2x4 for
Nd [71], 2x4 and 2x7 for Gd [72], Dy [10, 72], and Ho [14], 2x3, c(6x2) and 2x1 for Sm
[73], 2x3 and c(2x4) for Br [74], 1x5 [17]or 2x3, 2x4, 2x6, and c(6x2) [71, 75] for Yb.
As mentioned previously, in the initial stages of 3D silicide growth, several RE metals
form silicide NWs.

The formation of these NWs arises from the anisotropic lattice mismatch between the
RE silicide and the Si(001) substrate. The RE disilicide NW has a hexagonal Ale
structure. The orientation relationships of hexagonal RE812 on Si(001) are determined
to be:

h-RESi2(1IOO) // Si(001), h-RESi2[0001]//Si[1IO], and h-RESi2[112O] //Si[110]
as shown in Figures 1.3 (a) and 1.3 (b). The atomic resolution on top of NW shows
p(1x1) [12, 17] and c(2x2) [8, 9, 11, 12, 14] which are consistent with a hexagonal Ale
structure.

Comparing the lattice mismatch in a and c directions of the hexagonal structure with
Si(001) 1x1 unit cell, the lattice mismatch in a direction is very small (within it 1.64 %
for Y, Sm, Gd, Dy, Ho, Br, and Yb) as shown in Table 1.2. On the other hand, the lattice
mismatch in c direction is large (over 6.51 %). Because of those anisotropic
lattice-mismatch strains, hexagonal RESiz grows faster in a direction than in c direction.
Therefore, a and c directions of hexagonal structure correspond to a length and width of
NW, respectively, as shown in Figure 1.3 (c).

For ScSiz, the direction of large and small lattice mismatches is opposite, and thus the

growth direction of ScSiz NW is perpendicular to that of other RESiz NWs. Also, YbSiz

15

NWs grow in a wide variety of orientations [17]. YbSiz NWs grow along the [001]
direction which is 45 degrees rotated with respect to the other RESi,, NW-growth
direction. They also merge into the other NWs running in the different orientation.
This behavior might be controlled by larger lattice mismatch (-1.46 %) in a direction, and
thus there is a competition between different orientations of YbSiz NWs on Si(001).
When the lattice mismatch is large in both a and c directions such as NdSiz,
three-dimensional (3D) compact silicide islands form instead of elongated NWs [71].
Those 3D islands are also observed in most of the RESi7/Si(001) system [1, 7, 12, 14, 16,

76] depending on the annealing temperatures and annealing durations.

1.1.2.2 Schottky barrier height of RE silicides on Si

Almost all metals can form silicides, but most of them result in high Schottky barrier
heights on n-type silicon and low barrier heights on p-type silicon. Unlike most other
silicides, RE silicides produce a high Schottky barrier height of 0.7~0.8 eV on p-type Si
and a low barrier height of 0.3~0.4 eV on n-type Si [24, 25, 49, 56, 60, 77-79] as listed in
Table 1.3. This behavior makes them in particular interesting for applications as Ohmic
contacts for n-type Si [25]. Furthermore, the corresponding high barrier height on
p-type Si substrates [24] is an interesting property for infrared detectors or photovoltaic
applications. In case of ErSiz-x/Si diodes, excellent rectifying behavior is observed for
p-type Si, whereas the electrical characteristics of the ﬁlm to n-type Si are relatively
Ohmic at RT and rectifying at low temperatures [49]. Most of the values in Table 1.3
are quite close to one another, which reﬂects the great physical and chemical similarities

of the different RE metals (including Sc and Y).

16

1.1.2.3 Electrical resistivity of RE silicides

Our knowledge of the electrical properties of RE silicides is very limited and
fragmentary. There are only a few reported works on electrical properties of RE silicide
thin ﬁlms. The values of resistivity of RE silicides are summarized in Table 1.4. RE
silicide thin layers exhibit an almost metallic behavior. The resistivity decreases linearly
with decreasing temperature and tends to a limiting residual resistivity, p0, at very low
temperatures. The abrupt decrease of p(T) at very low temperatures is due to the
antiferromagnetic ordering arising from the incomplete 4f shell of the RE3+ ion, which is
operative below the Néel temperature TN. La [80], Gd [81-86], Th [83, 87], Dy [68, 83],
Er [29, 49, 83, 84, 86, 88, 89], Tm [29, 62], and Yb [29] have a signiﬁcant resistivity
drop, but Y [29, 88] and Lu [69] do not. In case of LaSiz.x (OSXSOA), there is a
superconducting transition at 2.5 K [80].

The resistivity p(T) is determined using Matthiessen’s rule which includes the various
independent scattering mechanisms:

ptT) = p0 + and) + pesto)
where ,00 and pm(T) are the residual resistivity and the magnetic contribution to the
resistivity, respectively. p0 arises due to imperfections in the crystalline lattice such as
defects and impurities. pm(T) is the magnetic contribution to the resistivity above the
Néel temperature TN and generated by scattering from randomly oriented spins [90].
The temperature dependent term pe-ph(T) is the contribution of the electron-phonon
scattering mechanism expressed by the Bloch-Gruneisen formula [91]:
5

4
I T D/T x
._ T = T— dx
p" M ) p (90] f (ex-lxl-e’x)

 

17

where OD is the Debye temperature, p’ is the high temperature limit of paph. At very
low temperatures (GD/T > 20), the Bloch-Gruneisen formula can be expressed as pph(T) ~
T5 and at higher temperatures (T ~ 91)), the relation is approximated by the linear

formula: pph(T) ~ T.

18

69:2: :oamEooaonm ”So—E 348:3 Ewen H2329: ”mm: EoEmomoc omecm.come> ”On; Eonﬁomgo 38.405 ME.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

32 sea 86 H “no 94> a
:E >4 .35 me o>
54 >4 who my 5 .3 >4 wad E.
SE Sosa 86 H moo 94>
4mm. >4 86 H 93 my NEE
4a >4 3o H So we Nwe:
3o H moo 94>
4E >4 86 H So my "55
>4 mod
$3 925 $.o 49
8a eeeo 3o H moo 94>
4E Amine >4 8d H 43 me as >4 3o H two me :25
Re .5 22a 3 H 3 we Em
5: Sega So we o0
>4 who
$3 223 Ed E 25>
>4 woo
:ﬁ 92.4 mod >
:2 >4 3 me :2 >4 to me _mom 3
uﬁQEP—ﬂmﬂo—ﬂ 0:52:93 ,uﬂUEP—ﬂmgnﬂ 0:62-88“ 023—? mu—
aom 4.. 32482 93 aae 8:858 255 com me $282 $3 gee =3...er 255 be .32: mm

 

 

.83593 E 099$ use -= so 330:8 can mEQE mm we .40 .mﬁmﬁc SE43 Esosom m4 033.

19

8:35.488 05 3209. 0240358943 ”m 64324.3 oEEoEozto AOV
.Ecomaboa CC .3835: 95 £35.25 EE 6 FM 3 53560.4 065.55 HQ 53362 34553“ noQ

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

.3 8m NEE
2.3 EN 04 o: co Harm
Ba 8+ ".mou
44$ 98 04 o: om £30
043 v4 m 64 R 64 em 8a a. 2 "a 3.3wa
83 cm m4 m4 :3 co 6 am E cm :3 cm 043 v4 m £53
4me mm 2&3
8: ea 53
44a SN 04 c: we £23
:35 SN em 8E an £3
4me we S am 4me mm 25>
as S4 3 4m;
SE E E»
844 84
$3 mam 8 4m;
52 mm 844 e9. 35%
an 4m 2 844 on _mom
8: a: ﬂaw
«om A65 c ES Gave. AEo Giveq Mom 945 o A80 Givq A80 @1464 you A80 SEQ AEo G5 em
54:95:... 2845484...— o==m osoam

 

 

 

 

.8225 mm o £24284 1 29¢

20

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

82 cm Smtom: 82 ow 8254 E? as 8 2 £53
:5. mam 4: £25
043 vv m er
43 8m 94 4mm;
32 8 E ea E 3% 4mg 8 E m: E 3% as com 8 £495.
3: mm em 3 E on E 43 $502:
:3 8:8 E 24 E34 :a 828 E mom E :2 5: mm no 35:4
as E a ﬂea
ocmi «.3544:
3a cam: 5.44%:
82 we E :4 E .9. 52 we E on E em

Pa 828 E 34 E 9: :a 828 E E E 4.4m £55
Nine
53 8m «44 4mm;
5: we 4o 3% 3455
:3 8.3 E om E 8 :55
$3 a? cam ewe—ow
:3 8m 04 c: mo :36
Ga mam "54.2
:3 ma as: m: 43.2

Lem A5544 25 G15 95 6.3164 mom 945$ Eu Give. E8 33164 Lem Eu ~4in ES @136
SEE—444,4 3:85:54..— aem oEoEm

 

 

 

 

 

2.4.258 v4 055.

21

1.2 Organization of the thesis

The remainder of this thesis is organized as follows. In Chapter 2, experimental
background will be provided. The principles of scanning tunneling microscopy (STM)
and low energy electron microscopy (LEEM), and the basic structure of the of Si(001)
surface are presented.

From Chapter 3 to Chapter 6, the growth behavior and electrical properties of RE
silicides on Si(001) are discussed. In Chapter 3 and Chapter 4, the growth behavior of
two different RE silicides, Ho and Sm silicide NWs and/or islands on Si(001) is studied
by STM. The results of scanning tunneling spectroscopy (STS) show that HoSiz W3
and 3D islands are metallic. Chapter 5 contains LEEM studies of growth kinetics of Dy
silicide islands on longer length scales and at elevated temperatures that are not
accessible by STM. The LEEM studies were carried out at the University of Illinois in
collaboration with Dr. W. Swiech of the Frederick Seitz Materials Research Laboratory.
In Chapter 6, transport measurements of Dy thin ﬁlms are presented. The details of
ultra-high vacuum (UHV) sample preparation for transport are also provided.

In Chapter 7, suggestions for future work will be described, continuing the themes of
this thesis in several different areas.

The Appendices include: a summary of studies of hydrogen on Si(001), a deﬁnition of
sheet resistance, the principles of two- and four-terminal measurements, a description of

the van der Pauw’s method, and a list of publications.

22

Chapter 2 Experimental
In this chapter, the experimental methods used, the basic principles and mechanisms of

scanning tunneling microscope (STM) and low energy electron microscope (LEEM), and

the basic structures of Si(001) substrate are introduced.

2.1 Experimental and research methods

All experiments except transport measurements were performed in an ultra high
vacuum (UHV) system with a base pressure of < 2x10'10 Torr. The chamber was
equipped with an STM, a four-grid low energy electron diffraction (LEED) optics, and
facilities for sample heating and metal deposition. The Si(001) samples were
chemically cleaned, and then outgassed by heating to 975°C. After outgassing, they
were ﬂashed brieﬂy at 1175°C to remove surface oxide, and held at 975°C for 10 min,
and then cooled slowly to room temperature (RT). The temperatures of the samples
were measured using an optical pyrometer. Metals were evaporated from a tungsten
wire basket, and the evaporation rate was calibrated by a quartz-crystal thickness monitor.

For holmium (Ho) deposition, typical evaporation rates were 0.05~0.09 monolayers
(ML) per minute [1 ML = was,2 = 1/(0.384 mm)2 = 6.78x10“ atoms/cm2 for Si(001): see
p.35 for details of silicon surface] and total coverages studied were between 0.06~0.9 ML.
For dysprosium (Dy) and samarium (Sm) depositions, typical evaporation rates were
0.1~0.6 MIJmin and 0.03~0.09 Ml/min and total coverages were between 0.l6~3 ML
and 0.05~1.5 ML, respectively. The chamber pressure remained below 1x10”8 Torr
during evaporation, and immediately recovered to below 5x10‘lo Torr after the metal

source was turned off. All growth was done at a substrate temperature of 600°C unless

23

speciﬁcally mentioned. All STM imaging and LEED observation were done at RT.

2.2 Scanning tunneling microscopy (STM)

STM was invented in 1981 [100, 101] by G Binnig and H. Rohrer, IBM in Zurich,
Switzerland, gaining them the 1986 Nobel prize in physics. Their ﬁrst STM image of
Si(111)-7x7 reconstructed surface with its display of individual atoms [102] was the
stimulus that initiated many STM projects all over the world.

STM is a powerful technique for studying the structural and electronic properties of
surfaces with real-space atomic resolution. When a sharp metallic tip is brought close
enough (~10 A) to a conducting surface and a bias voltage (mV~1OV) is applied, the
tunneling current ﬂows between the tip and sample. The lateral resolution is about 1 A
whereas a vertical resolution up to 0.01 A can be achieved. In addition, STM can be
used not only in UHV, but also under other environments such as in air or in solution.

Recently, atoms and molecules have been able to be moved controllably on surfaces,
either by sliding them by means of the tip or by lifting them off the surface by the tip and
depositing them at a different location [103-116]. The atomic-scale manipulation of
surfaces by STM-tip is a promising candidate for future nanoelectronic device

applications.

2.2.1 Basic principles of STM
The operating principle of the STM is based on the quantum mechanical phenomenon
of tunneling. Electrons transmit through a vacuum barrier between two conductors, in

this case the sample and the tip, when they are brought very close together. A simple

24

understanding of this principle follows from consideration to the solution of the
Schrddinger equation for a one-dimensional square barrier of height V0, as shown in
Figure 2.1. For a particle of energy E < Vo incident on this square barrier from the left,
the solutions of the Schrbdinger equation have the form:

leti’“ +Re“’“ (x<0)

w(x) = < AK” + 36““ (0 S x S s) (1)

TeI‘k" (s < x)
L

 

where hk =J2mE, ﬁx: 2m(V0 —E), R and T are reﬂection and transmission

coefﬁcients, respectively, and A and B are constants.

’lﬂf\\nr
UU Lit

0 s x

 

 

 

 

Figure 2.1 Wave function [[1 for an electron with kinetic energy E tunneling a thin

potential banier of height V0 and width s. The wave function is oscillatory in the left
and right regions of the barrier. Within the barrier, the wave function decays
exponentially.

The wave function [(1 decays exponentially into the classically forbidden region
0 S x S s with a decay length of 1/ K. For an electron at the Fermi level of a conductor
with a work function ¢, this decay length would be 1/ K = ‘ihz / 2m¢ (Vo — E is just

the work function). Since most work functions are around 4~5 eV, the decay length is
typically 1/ K ~ 1 A. When a barrier of width sis much thicker than the wave function

decay length of UK, the transmission probability T, or the tunneling current J decays

25

exponentially as:

J = ﬁlle cc e‘2’° = e—Zn/ZmWhZ (2)
m

Thus the tunneling current drops by an order of magnitude for every 1 A. For the
tip-sample distance of ~10 A, the tunneling current is a few nA.

For tunneling between two conductors with a voltage difference V across the gap, only
the states within eV above or below the Fermi level can contribute to tunneling; electrons
in states within eV below the Fermi level on the negative side tunnel into empty states
within eV above the Fermi level on the positive side. In the case of Figure 2.2, the
tunneling current ﬂows from a tip to a sample, since the positive bias voltage is applied to
the sample. When the sample is positively and negatively biased, it is called “empty”

and “ﬁlled” states, respectively.

Tip r».-. Sample
FL,I‘“L, o. . .

Energy levels from
which tunneling can occur

 

Figure 2.2 Energy diagram for tunneling between a tip and a sample. Tunneling is
possible only when there are empty states on the right. Such empty states are created
when eV is applied to lower the Fermi level on the right.

26

2.2.2 STM data acquisition and image processing

STM can be operated in several imaging modes: constant current mode, constant
height mode, barrier height imaging, and scanning tunneling spectroscopy (STS). Most
of the STM images presented in this thesis are acquired in the constant current mode
unless speciﬁcally mentioned.

In the constant current mode, the tip is scanned in the two lateral (x and y) dimensions,
while a feedback circuit constantly adjusts the tip height to keep the current constant.
The position of the tip is accurately controlled by the piezoelectric drivers. The tip
position is recorded as a function of the x-y surface coordinates as shown in Figure 2.3.
All STM bias voltages were applied to the sample with the tip at virtual ground.

According to equation (2), the tunneling current reﬂects the surface topography only
when the work function does not change during scanning. In addition, the tunneling
current changes with the electronic state of the tip and sample. Therefore, essentially,
STM image shows the electronic density of states at the sample surface instead of surface
topography of the sample.

The most common method of displaying STM data is to construct a grayscale image
where the brightness represents the surface electronic density: Brighter areas are higher
density of electrons at the surface. For easy understanding, brighter areas can be
considered topographically higher than darker areas.

In all of the STM images in this thesis, only minimal image processing has been done,
consisting of planar background subtraction, drift correction, or contrast adjustment in
order to extract the desired information. Image processing was done with either “Image

SXM” which is the public domain image analysis software written by Steve Barrett [117],

27

or NOeSYS “Transform” software for the correction of image distortion.

 

 

z
Tunneling X
Image ‘
Vz = Output
Y
x-y
Raster Reference
V0“ Current
--—-h
Piezoelectric 4
Tripod l I
z v

 

 

 

 

 

V

Feedback
\ A Position
x Controller.
.5 Tip '
. ; Vbias
/ Tunnel
Current
Sample

Figure 2.3 Schematic diagram of an STM [118]. A metallic tip is mounted on a
piezoelectric tripod actuator which can move the tip in the x-y plane of the sample surface,
and in the z direction normal to the surface, with atomic resolution.

2.2.3 Scanning Tunneling Spectroscopy (STS)

STS is used for studying local surface density of states with atomic resolution because

the probe apex is very small consisting of a few atoms. The sensitivity of STS to

28

electronic structure can be a tremendous advantage over other spectroscopic methods
such as ultraviolet photoelectron spectroscopy (UPS), X-ray photoelectron spectroscopy
(XPS), Auger electron spectroscopy (ABS), and inverse photoelectron spectroscopy
(IPES). A single tunneling spectrum can show surface states both above and below the
Fermi level simultaneously, at the same spatial location.

Figure 2.4 shows the schematic of an STS experiment. The sample-tip separation is
fixed and the tunneling current I is recorded as a function of applied bias voltage V (I-V
curve). Since I-V characteristics depend exponentially on the sample-tip separation, the

or the normalized differential conductivity dI/dV ,
dan I /V

 

 

logarithmic derivative

which is a dimensionless quantity, is generally used to remove the inﬂuence of sample-tip
separation. The approximate expression for the tunneling current [119] is given by:
V
I cc 1: ps(E)T(E)dE (3)
where p3 is the surface density of states, T is the transmission probability in equation

(1), and V is the applied sample voltage. After the ﬁrst order approximation of E (or V),

 

the normalized differential conductivity can be expressed as:

dI/dV ~ p,(eV)

I IV 1 V (4)
W E p.(E)dE

 

 

Therefore, is proportional to the normalized surface density of states.

For the STS data plot and graphing, WaveMetrics “Igor Pro” software was used.

29

 

 

 

Figure 2.4 Schematic of an STS experiment. An STM tip is held above the sample,
and the current is measured while the voltage is ramped. This measures the
conductivity in the surface normal direction.

2.3 Low energy electron microscopy (LEEM)

LEEM is a relatively new electron microscopy technique for in-situ real-time studies of
surface dynamics, such as epitaxial growth, phase transitions, interface formations, strain
relaxation phenomena, and chemisorption over a wide temperature range. LEEM was
developed by E. Bauer and his student W. Telieps in 1985 [120] after its invention by E.
Bauer in 1962 [121]. RM. Tromp and MC. Reuter designed a new LEEM that has
higher resolution and wider capability of the instrument in 1991 [122].

Electron microscopes use beams of electrons instead of light, and magnets instead of
lenses. However, the working principle of electron microscopes is more or less the
same as in the optical microscope. In LEEM, the sample is illuminated with a
well-collimated beam of low-energy electrons which diffract from the sample. The
diffracted electrons are then used to form an image. In contrast to STM in which the

probe tip is scanned back and forth across the surface to form images, all pixels are

30

imaged simultaneously from the illuminated area on the surface. At low energies of the
order of 1~50 eV, in particular at very low energies (< 10 eV), high brightness and a good
resolution can be obtained. Contrast is mainly produced by diffraction, surface
t0pography, and local changes in electron work function and/or electron density. For the
diffraction contrast, there are three imaging modes in LEEM as shown in Figure 2.5:
bright-ﬁeld (BF) imaging, tilted bright-field (TBF) imaging, and dark-ﬁeld (DF) imaging.
When electron beam is on the optical axis of objective lens, and the specularly reﬂected
(0,0) beam is used to form a BF image. TBF image is also formed by using (0,0) beam,
but with tilted incident angle. In DF imaging, non-specular diffracted beam (any beam
other than (0,0) beam) is used. TBF or DF contrast must be used for the observation of
azimuthally rotated domains, because domains are equivalent at normal incidence and
they do not produce contrast in BF.

For the interface contrast, surface steps are surrounded by stress and strain ﬁelds which
cause local changes of the diffraction conditions. The resulting diffraction contrast may

enhance or reduce the geometric phase contrast.

(b) (c)

 

specimen \
lens I
(hk) (00)
' """k‘e‘ aperture -
o (oo) (bk) 0

 

 

optical axis

Figure 2.5 Diffraction contrast in LEEM (a) Bright ﬁeld, (b) tilted bright ﬁeld, (c) dark
ﬁeld.

31

The spatial resolution is determined mainly by the spherical and chromatic aberrations
of the objective lens and by diffraction at the contrast aperture. Lateral resolution of
LEEM is 5~10 nm on video-rate imaging [123, 124] , but in principle the best resolution
of 3 nm is achievable [123]. Diffraction contrast provides depth resolution of a single

atomic layer.

2.3.1 Basic physics of the low-energy electmn interaction with surface: Elastic
scattering and inelastic scattering

For the low energy electrons in the LEEM geometry, inelastic scattering and elastic
backscattering are important (No forward scattering). At low energies, the elastic
backscattering does not increase monotonically with nuclear charge because the incident
electrons can ‘feel’ the details of potential distribution of neighbor atoms. Incident
electrons can be reﬂected from surfaces with a high reﬂection coefﬁcient within the ﬁrst
few atomic layers when their energy is in a band gap. (When electrons have wave
vectors at the Brillouin zone at which the band gaps occur in the band structure, they

fulﬁll the Laue condition k — k0 = 27:11 and are reﬂected.) Or they can penetrate deeply

into the crystal when the incident energy E matches to E(k) states in the solid.

The electron energy loss mechanism in solids is excitation of valence band electrons.
For low energy electrons, the inelastic processes do not involve inner electron shells
because of the insufﬁcient energy but only outer electron shells. The electron density in
the valence band (or Fermi energy EF) is nearly constant for most materials, and it gives

the ‘universal curve’ of inelastic mean free path A =1/y , where M E) is the attenuation

coefﬁcient for inelastic scattering.

ME) dominates I/(E), the attenuation coefﬁcient for elastic scattering, above the

32

threshold ET for plasmon excitation. ET varies with E1: from 11 to 27 eV when E:
changes from 5 to 15 eV [125]. At E>ET, inelastic scattering processes determine the
penetration depth corresponding to a mean free path in the range of 0.3 to 1 nm. On the
other hand, at E<E1~, elastic backscattering determines the penetration depth. The
inﬂuence of inelastic scattering is negligible at E<ET, because the inelastic mean free
path increases rapidly as the incident energy E decreases. Therefore, energies up to
20~3O eV are advantageous for LEEM from the point of view of electron-surface

interaction.

2.3.2 LEEM instrumentation and image acquisition

The schematic of a LEEM instrument (IBM Type I) is shown in Figure 2.6. The
design of LEEM requires full UHV compatibility. The sample is held at about —20 kV
to decelerate the electrons to a few eV at the sample surface. The magnetic objective
lens is entirely at ground potential. The incident beam comes from a Schottky emission
gun and is focused with condenser lenses, an octupole stigmator, and steering coils.
Condenser lenses relay the cross-over point into the back focal plane of the objective lens
in which a diffraction pattern is formed as in Figure 2.7. Two sets of steering coils
control the shift and tilt of the e-bearn. A beam-separating deﬂection magnet is
necessary because the reﬂected beam has to be separated from the illuminating beam.
The reﬂected electrons pass through the deﬂection magnet again and enter the imaging
column. The image is projected onto a channel plate screen. LEEM images are
recorded using a video camera outside the vacuum chamber and digitized for analysis at

30 frames per second.

33

sample
magnetic objective lens

:3 : octupole stigmator

deﬂection magnet
steering coil (DZ)

steermg COll
steering coil (Dl)\ quadrupole stigrnator
’ ‘ / transfer lens
octupole stigmator %$i£, ‘ :7 a» '/ ,contrast aperture

condensor lens (C24 :,~ ' '0 / intermediate lens
‘V

condensor lens (Cl) V ‘ $
‘ projector lenses

V7
‘7 V‘;
V’ ‘y channelplate
screen

electrostatic lenses
Schottky emission gun: ZrO on W000)

 

Figure 2.6 Schematic of LEEM instrument (IBM Type I).

(a)

sample (1,) objective lens

 
  
 

 

III

.“"; objective lens

 

 

 

 

 

 

back focal plane
(LEED pattern)

 

 

back focal plane image
(LEED pattern)

Figure 2.7 Schematic ray diagrams for LEEM [126, 127].

34

2.4 Si (001) surfaces

0n unreconstructed Si(001), each Si atom on the top layer has two broken covalent
bonds (dangling bonds) with a nearest-neighbor separation of as] = ads/2 = 0.384 run,
where (10 equals the Si lattice constant of 0.543 nm as shown in Figure 2.8 (a). aSi is
deﬁned as a unit length of Si(001) surface and a square 2D lattice in Figure 2.8 (b)
represents a 1x1 unit cell. Since this surface is not an energetically favorable state, the
surface reconstruction occurs to lower the surface energy. Adjacent atoms on the
surface move toward each other at the expense of inducing a large surface stress and
make a new bond forming a “dimer” [Figure 2.8 (c)]. Intradimer atomic distance is
0.24i0.02 nm which has been studied experimentally [128-130] and theoretically

[131-135].

 

 

 

 

   

0.384 nm
(a) (b) Ideal surface (c) Dimer formation

0.543 nm

Figure 2.8 (a) Diamond structure of silicon, and (b), (c) top view of model of the dimer
reconstruction on Si(001).

35

The valence electrons of an isolated Si atom are in 33 and 3p atomic orbitals. These
orbitals are ‘hybridized’ into sp3 orbitals [136] when the atoms arrange in tetrahedral
structures like in a diamond-lattice crystal. As shown in Figure 2.9, in the dimer surface
state, the ﬂstates, ”bonding and Jz'anti-bonding, are formed from even and odd
combinations of the left and right dangling bond state. On the other hand, 0' states,
O'bonding and a'anti-bonding, arise from the bridge states [137]. During STM
imaging, the spacial distribution of these ﬂbonding and ﬂ'anti-bonding states are
probed by electrons tunneling out of and into the sample, giving the ﬁlled- and
empty-state images, respectively. In the ﬁlled states, bright maxima appear in the center
of dimers where electron density is localized as shown in Figure 2.9 (a). Since each Si
atom has one dangling bond which is a half-occupied electron state, two corresponding
maxima appear at both edges of the imer with small rounded shape in empty states as
shown in Figure 2.9 (b). White rectangles marked in Figures 2.9 (a) and 2.9 (b)

correspond to a black rectangle showing the dimer position in Figure 2.10.

   

HIE" anti-bonding orbital

V r I, /A ‘ ".141. f
8 O 8 G O 6 b
(a) Filled states (b) Empty states

Figure 2.9 (a) ﬁlled- and (b) empty-state STM images of Si(001) surface with a side
view of Si dimer model. 2X1 unit cells corresponding to the dimer position are marked.

36

 

 

 

 

 

« »

.3 ,1 e e etc... o. c.- .e or
125W! mrzr.r.r r: r

:6 c : eff e be “~c~-"‘o~"*~e or
c m r m r r r r C _‘c
¢*‘.\”c“"'. cocoa. 0-....e..-’ .c- ”or
r m t m r i r 3‘ r ; r r r
e ~Q~c~ﬁooojco e- "‘0" r.“ c o c}

x .v -. , . C 1st layer
6 = ‘ .2 I‘» f t f f 61 f C 2nd la er
c4 can. two it. c c- e ewe . y

._ » - ”O O O ~ 6 3rd layer
tmcmrIrgr‘r r. r («Mam
e ----- ococococ’c‘o“ I-o c~~e V

Figure 2.10 Schematic drawing of Si(001) 2x1 reconstructed surface consisting of three
terraces. Si atoms on top, second, third, and fourth layers are shown with black, dark
gray, light gray, and white colors, respectively. The height of the terrace lowers from
left to right. The directions of surface stress are marked with black arrows.

Si dimers form rows consisting of p(2x1) structure on Si(001) surface as shown in
Figure 2.10. Since the dimer rows are rotated by 90° on each atomic step on the surface,
the Si(001) surface has two degenerate phases 2x1 and 1x2 and its surface stress ﬁeld is
anisotropic. The surface is under tensile stress along the dimer row direction and under
compressive stress normal to the dimers [138] as marked with black arrows in Figure
2.10. In this thesis, the Si dimer row direction is defined as [IIO] and the Si
dimerization direction as [110] as indicated in Figure 2.10.

The LEED pattern of a Si(001) surface provides l/2-order spots corresponding to the
2x1 periodicities on the surface as shown in Figure 2.11. The diffraction spots with
white (gray) circles in Figure 2.11 are diffracted from the highest and lowest (the middle)
terraces in Figure 2.10, respectively. For real space imaging, one of these diffraction

spots can be used. In LEEM images, the intensity changes sharply at the step with TBF

37

or DF contrast. Figure 2.12 shows the DF images of Si(001) surface with opposite
contrast [139]. When (%, 0) diffraction spots (marked with white circles in Figure 2.11)
are selected for a DP image, 2x1 terraces appear bright, while 1X2 terraces appear dark
[Figure 2.12 (a)]. When (0, %) diffraction spots (marked with gray circles in Figure
2.11) are selected, the diffraction contrast appears opposite [Figure 2.12 (b)].

Strictly speaking, there are several other reconstructions on Si(001) surface due to the
buckled (twisted) dimers at low temperatures: p(2x1) asymmetric phase, p(2x2) phase, or
c(4x2) phase [135, 140-147]. However, at RT, a fast ﬂip-ﬂop motion of the buckled
dimers is thermally activated, which mostly shows symmetric dimer rows in the STM
images. Therefore, it is plausible to assume that the Si(001) surface has a symmetric

p(2x1) structure as far as the experiments are performed at RT.

 

Figure 2.11 LEED pattern of a Si(001) surface with 2x1 superstructure. (%, 0) and (g,

0) diffraction spots and (0, %) and (0, £0 diffraction spots are indicated with white and

gray circles, respectively. 2x1 unit cell is marked with a white rectangle as well.

38

 

Figure 2.12 Dark-ﬁeld LEEM image of Si(001) surface with opposite contrast.
Images (a) and (b) are acquired using one of the diffraction spots with (a) white and (b)
gray circles, respectively [139].

39

Chapter 3 Ho growth on Si(001): surface reconstructions and
nanowire formation

In this chapter, the growth behavior of Ho on Si(001) is described in detail, including
coverage dependence, nanowire (NW) and 3D island formation, 2D substrate surface

reconstructions, and the results of scanning tunneling spectroscopy (STS).

3.1 Results and discussion
3.1.1 Coverage dependence

Figure 3.1 shows the coverage dependence of Ho growth on Si(001) at 600°C. The
horizontal axis is the metal coverage in monolayers (ML), and both LEED pattem and
topography are shown. Black circles are the metal coverages at which experiments were
done. There are roughly two types of topographies, both of which are comprised of a
2D reconstructed substrate, coexisting with 3D silicide islands. The LEED patterns
reﬂect the 2D ordered surface. For low metal coverage, 0 S 0.4 ML, the islands are
highly elongated NWs with typical width 1.l~7.3 nm, height 0.2~1.2 nm, and lengths
100~500 nm [Figure 3.2 (a)]. As the coverage increases, 3D compact silicide islands
are found of varying shape [Figure 3.2 (b)]. NWs and 3D islands coexist at coverages
above 0.4 ML. A 2X4 phase dominates the 2D reconstructed surface structure at most of
the coverages shown in Figure 3.1 although clean Si 2x1 dimer rows are observed in
places below 0.18 ML. In addition, a 2x7 phase appears in the 0.40~0.55 ML coverage

regime, and it coexists with the 2x4 phase.

40

 

 

O “—o—o—o-q—o-l-ooc . .....
4 Nanowires!

0‘ o.5il“1.o

 

 

r‘ I
i ' l : :30 islands: 1 '

. . . l

I

 

)9 [ML]

Figure 3.1 Coverage dependence of LEED patterns and topography for Ho growth on
Si(001) at 600°C.

The percentage of the 2D surface covered with H0 in either the 2x4 or the 2x7 phase is
not monotonic with coverage as shown in Figure 3.3. At lowest coverage, 0.06 ML,
there is only 2x4 structure and bare Si on the 2D surface. As the coverage is increased,
most of the 2D surface is covered by the 2x4 structure, and both'the density and the
average height of NWs increase. When 3D islands appear, the percentage of the 2D
surface covered by 2x4 decreases and bare Si substrate reappears. From mass
conservation, Ho atoms moved from the 2D surface to the 3D islands. This fact implies
3D silicide islands are preferable to either a Ho reconstructed substrate, or the elongated
NWs in terms of accommodation of Ho. The initial stage of NW growth can be
described as Stranski-Krastanov. However, the reappearance of bare Si once 3D islands

are formed shows that the growth mode changes to Volmer-Weber at higher coverage.

41

     

I I ' '\ ' I V135";

Figure 3.2 Two 400x400 nm2 images of HoSiz W3 and 3D islands. (a) NWs formed
by 0.36 ML of Ho. Each NW grows on a single terrace. (b) As the coverage is increased
(0.90 ML), 3D compact silicide islands are found.
3.1.2 NWs and 3D islands

As seen in Figure 3.2 (a), every NW grows on a single atomic terrace, and runs along a
<1I0> type direction. Each NW usually terminates at a step edge or at a perpendicular
NW, but perpendicular NWs are on different terraces and do not cross. At higher
coverages, rectangular islands can nucleate at the intersection of NWs. The NWs in
Figure 3.2 (a) are comparatively long as the step density on this substrate is low. On
substrates with higher step density, steps can ﬂow to accommodate NWs, as shown in
Figure 3.4 (a). In this image, the steps curve to follow the NWs so that each remains on
a single terrace. Nevertheless, there is some limit to the number of steps that can ﬂow

to accommodate a NW, and so NW length is still limited by step density.

42

(%) (nm)

 

 

 

 

100
C] % of bare Si
I % of Ho/Si surface
80 reconstruction
- + - % of NWs and/or 3D
60 ' ' islands on surface
_D_. average height of
40 NWs and/or 30
islands

 

 

 

 

 

 

 

20'

 

 

 

 

 

(O N ('3 (O (D (D IO N O)
0. N. 0' <0. m, V. o' N. 0' coverage (ML)
0 O O O O O

l V \
\ ’\ 7

30 islands

Figure 3.3 Relationship between coverage and the percentage of surface covered with
Ho surface reconstruction, NWs, and 3D islands.

As shown in Figure 3.4 (a), there are two distinct NW morphologies; one is triangular
in cross section, and the other is rectangular. Figures 3.4 (b) and 3.4 (c) are
cross—sectional proﬁles of these NWs. At 02-03 ML, both types of NWs are seen at
equal probability. As the coverage increases up to 0.4 ML, rectangular NWs are more
often observed and at the same time NWs tend to form parallel bundles as shown in
Figure 3.5 (a). While the narrow NWs in the edges of bundles are mostly triangular in

cross section, those in the middle of the bundles are certainly rectangular.

43

 

 

 

 

 

 

 

 

 

[nm [NM]
1.6 1.6
1 2 1 2
0.8 0.8
A a c o
0.4 0.4
O 5 10 15 20[nm] 0 5 10 15 20 [nm]
(b) triangular nanowire (c) rectangular nanowire

Figure 3.4 (a) Triangular (left) and rectangular (right) NWs at 0.3 ML (200x200 nmz).
Steps on the substrate ﬂow to accommodate the NWs. Lines A—B and OD show the
locations of the cross sections shown in (b) and (c).

The structure of the rectangular NWs can be understood in terms of the bulk silicide
HoSim, which has a hexagonal Ale type structure, with the [0001] axis parallel to the
surface, and the [1120] direction along the long axis of the NW as shown in Figure 1.3 in
Chapter 1. Atomic resolution on the top surface of rectangular NWs shows c(2x2)
periodicity [Figure 3.5 (b)]. The same periodicity is seen on the surface of second layer
NW growth. The height of the second layer NW is always 0.33 nm which corresponds
to 1 bulk HoSiz unit cell height [9]. This height includes one metal layer and two Si
layers. The apparent height of the ﬁrst NW silicide layer varies because the electronic

properties of the NW and substrate surface are different. The structure of the triangular

44

cross section NWs is not clear from the STM data, but it is likely that it has the same
epitaxy characteristic as the rectangular NW with the [1120] direction lying along the
long axis. It is possible that these triangular wires are simply a minimal width

rectangular wire with perhaps a different step edge structure.

 

Figure 3.5 (a) Bundled NWs with second layer growth at 0.36 ML (90x90 nmz). (b)
Atomic resolution on bundled NWs (12X14 nmz). A single c(2x2) unit cell is marked
with a white square.

When the widths and the heights of silicide islands are measured, it is clear that the 3D
compact silicide islands have widely varying dimensions whereas those of NWs are
restricted to narrow ranges. Figures 3.6 (a) and 3.6 (b) are the distributions of the
width and the height of single W3 and compact silicide islands, respectively.

The top graphs show the details of the leftmost bar on the bottom graphs, and include
data from both isolated NWs and NWs in bundles. Narrowcr widths shown by ‘A’ come
from the edges of bundled NWs, and except for these data, there is no particular

difference between the widths of bundled and isolated NWs. For isolated NWs,

45

rectangular NWs are wider than triangular ones. The most probable NW width is in the
range in 2~3 nm, which corresponds to 5~9 silicide cells wide. The maximum NW
width of ~5 nm can be related to the 6.8 % lattice mismatch between HoSiz and Si
substrate along the c axis. This was also noted for DySiz NWs which have similar

lattice mismatch and maximum width [9].

 

 

 

 

 

 

 

 

 

 

 

   

20 40
a I o 17~o. 56ML b I 0.17~0.56ML
g 15 ( ) (nanowires) .8 30 ( ) (nanowires)
"E A E
8 1° H 8 2°
N (U
c 5 c
B I I. I.. 3 1°
#1: =11:
O - L o l ..
0 2 3 4 5 8 7. “8 9 0 0.2 0.4 0.6 0.8 1.0 1.2
width lnml height [ml
3 1 -3
g - 0.17~0.56ML (nanowires) g - 0.17~O.56ML (nanowires)
E m 0.40~0.90 ML (30 islands) E rat: 0.40~0.90 ML (30 islands)
0 , ._ . ..-. D I .
(‘0 (0
a '5 l
2 2 i I
S 2
._ 0 40 80 120 160 200 ‘_ 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
2 Width [mm ‘2 height [nm]

Figure 3.6 The distributions of (a) the width and (b) the height of single NWs and 3D
islands.

The most probable NW height is 0.4~O.6 nm, which is 1~2 bulk HoSiz unit cell heights.
Although we have statistics for comparatively few compact 3D islands, it is clear that

both the width and the height of the compact silicide islands are spread over a much

46

wider range than those for the NWs. The fact that the 3D islands' sizes can exceed the
maximum width of the NWs suggests that their growth is not coherent. It is possible
that there are dislocations at the interface that do not propagate up to the surface of the
islands. However, for most islands there is no clear indication of such dislocations from
the STM images, unlike the dislocation networks observed for other epitaxial silicides
grown on Si(lll) [148]. Further work with transmission electron microscopy (TEM)

would be useful in clarifying the structure of the islands.

3.1.3 Two dimensional substrate surface reconstructions

Figures 3.7 (a) and 3.7 (b) are empty- and filled-state images of slightly different areas
of the same surface with 0.55 ML Ho. Coexistence of 2x4 and 2x7 structures on the 2D
surface is observed. A NW runs diagonally across the image, and gray circles on both

Figures 3.7 (a) and 3.7 (b) indicate the same physical position on the surface.

 

Figure 3.7 (a) Empty- and (b) filled—state images of slightly different areas of the same
surface with 0.55 ML Ho showing both 2x4 and 2x7 structures (30x30 nmz). The
position of gray circle on the NW in (a) corresponds to that in (b).

47

Figures 3.8 (a) and 3.8 (b) show the empty-state (1.76 V) and ﬁlled-state (-I.76 V)
images of the 2X4 structure corresponding to the square areas numbered by ‘1’ and ‘3’ in
Figure 3.7, respectively. These images are of the same area of the surface. Both
ﬁgures have three sets of 2X4 unit cells along the Si dimer row direction [llO] outlined

in white. In the empty—state image [Figure 3.8 (a)], three maxima are visible in each

 

ZIZIZIZIZIZIZIZIZIZIZIZIZIZ

. . . . o
.

 

 

 

 

 

 

o 0'0
O 20400.....o
o ....o
0 00'00
O 0"00

0

O

O

o (C)

o oIo oIo oIo o
3 2

a a

Figure 3.8 (a) Empty-state (1.76 V) and (b) ﬁlled-state (-l.76 V) images of the 2x4 phase
in the reconstructed surface shown in Figure 3.7 (10X10 nmz). (a) and (b) correspond to
the square areas numbered by 1 and 3 in Figure 3.7, respectively. (c) Line drawing
showing the empty-state (gray circles) and ﬁlled-state (dark circles) maxima of the 2x4
structure. Clean 2x1 Si dimer positions are shown as a reference on the topmost row.

48

2x4 unit cell. There are two types of maxima that are lined up in one of two possible
conﬁgurations; a darker maximum (small circle) between brighter maxima (big circles),
or a brighter maximum between two darker maxima. Either combination is equally
probable. In the ﬁlled state image [Figure 3.8 (b)], the contrast between the two types
of maxima is reversed, and more pronounced. Bright oval maxima appear in the
positions of the darker empty-state maxima.

By comparison of both bias images with respect to the Si(001) substrate, the
registration of empty and ﬁlled state of maxima of the 2x4 structure can be diagrammed
as in Figure 3.8 (c). All the empty-state maxima are positioned in what would normally
be Si dimer positions, and the ﬁlled-state maxima are located in the same positions as the
darker empty-state maxima. The overall character of the 2x4 structure is the same as
that seen for Dy on Si(001) [10].

Figures 3.9 (a) and 3.9 (b) show enlargements of the 2x7 structure corresponding to the
square areas numbered by ‘2’ and ‘4’ in Figure 3.7, respectively. Two types of rows
(row A and B) are observed and three types of maxima are visible in both biases. In the
empty-state image [Figure 3.9 (a)], row A consists of oblong maxima elongated along the
7X direction. Row B consists of three maxima and the smaller maximum is located
between two larger maxima. In the ﬁlled-state image, row A includes paired maxima,
and row B has three maxima and the larger maximum is located between two smaller
maxima. The registration of 2x7 phase can be obtained by comparison with positions of
2x4 phase from both bias images.

The registration and approximate spatial extent of the empty- and ﬁlled-state maxima

of the 2x7 structure are diagrammed in Figure 3.9 (c). This illustration contains ﬁve

49

rows (row ABABA) which are seen in Figure 3.9 (b), and the clean 2x1 Si dimer row
positions are shown on top as a reference. The empty-state maxima of the bright
features in row A and B are represented by gray oblongs and circles, respectively. The
ﬁlled-state maxima are illustrated using dark circles in three sizes. In row A, the paired
ﬁlled-state of maxima are located on the bridge site, and an oblong empty-state maximum

is positioned off bridge site. In row B, in both states, the middle maximum is located on

 

 

IIIIIIIIIIIIIIII

I—l

I'—'o‘ o'- 0.0 -

    

o ”o “3.0'
o o

     
      

 

° °o° are
0 I ’V o o I": o o o o
o a - 0.0 0:0 0&0
o I IIIIIIII ‘ o o o o
o” . o - T 0.0 0 ¢ OT
(C) shifted row A and B by 1a

Figure 3.9 (a) Empty-state (1.76 V) and (b) ﬁlled-state (-1.76 V) images of the 2x7
phase extracted from Figure 3.7. (a) and (b) correspond to the square areas numbered
by 2 and 4, respectively. A 2x7 unit cell consists of three types of maxima in rows
labeled A and B in both images. (c) Line drawing showing the empty-state (gray circles
and oblongs) and ﬁlled-state (dark circles) maxima of the 2x7 structure. 2x1 Si dimer
positions from the clean surface are shown as a reference on the topmost row.

50

the bridge site, and both edges of the maxima are positioned off bridge site. Row A or B

in the both states often shifts by laSI in the perpendicular direction to the 2x1 Si dimer

row ([110] direction). In this sense, the overall periodicity should be deﬁned as the
coexistence of 2x7 and c(2x14) unit cells. For simplicity, we denote this phase by 2x7.
Since at a given coverage Ho metal can be distributed in NWs and both the 2x4 and the
2x7 reconstructed surfaces, it is difﬁcult to decide the density of H0 in the 2x7 structure.

Therefore we are not proposing an atomic structure for the 2x7 phase.

3.1.4 Ho content of the 2x4 substrate reconstruction and the NWs

From the results in Figure 3.3, one can compare experimental coverage with calculated
coverage under different assumptions for Ho content of both the 2D substrate
reconstruction and the NWs as shown in Figure 3.10. The calculation was done on the
basis of the ﬁrst ﬁve coverages in Figure 3.3. The table on the right in Figure 3.10
shows two different assumptions of Ho content. The ﬁrst assumption is the number of
Ho atoms per 2x4 unit cell on the 2D surface: ‘A’, ‘B’, and ‘C’ indicate one, one and half,
and two Ho atoms per 2x4 unit cell, respectively. The second assumption is the number
of metal layers in a NW: ‘1’ and ‘2’ represent one and two metal layers in the minimum
height NW, respectively. Here we also assumed that each metal layer in the NW has a
density of 1 ML as implied by the bulk crystal structure as shown in Figure 1.3 in
Chapter 1. The coverages shown in the graph on the left hand side where calculated
under the six possible combinations of assumptions enumerated to the right, along with
the measured percentage of the surface covered by bare surface, 2x4, and NWs as shown

in Figure 3.3. The best model should come closest to the diagonal line. Error bars are

51

omitted for clarity. The error in coverage is less than 20 %, and the error in the vertical
position of the symbols is less than 10 %.

As can be seen in the ﬁgure, the best ﬁt is for ‘BI ’ which implies 1.5 Ho atoms per 2x4
unit cell, and one layer of metal atoms in the minimum height NW. There is one layer
of metal atoms in each bulk unit cell of the silicide, and so this is not a surprising result.
Nevertheless, this coverage dependent analysis was necessary to deduce this fact since
the geometric height of the ﬁrst layer NWs as seen by STM was bias dependent. We
cannot exclude the possibility that the minimum height NW may have more than two
layers of Si atoms, but this depends in some sense on the deﬁnition of where the substrate
ends and the NW begins.

The 1.5 Ho atoms per 2x4 unit cell is somewhat more surprising. The simplest
interpretation is that one of the two types of ﬁlled state maxima is associated with a
single Ho atom. As shown in Figure 3.8, two possible conﬁgurations of 2x4 unit cells

alternate on the surface, one with a bright-dim-bright arrangement of maxima, and the

 

 

 

 

 

 

 

 

0.7
3' 0.6-
8, 0.5
g 04
> - ”
8 0 ° . .
'0 Q3 d’ -— Ideal km: A: 1HO/2X4
2 B X 2; 3:1.5 Ho/2x4
.53 0,2 - 0 0 - 31 C: 2Ho/2x4
2 D 82 , . .
8 0.1 — + 01 1. 1 metal layer In nanowwe
O 02 2:2metal layers in nanowire
0

 

 

 

 

 

O 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Experimental coverage (ML)

Figure 3.10 Comparison of experimental coverage with calculated coverage under
different assumptions of Ho content (see text).

52

other with a dim-bright-dim arrangement. There are equal populations of both types of
unit cells, meaning that the average density of either type of maxima is 1.5 per 2x4 unit
cell. This is not to say necessarily that the maxima seen in either bias is a Ho atom. In
the case of the Dy 2x4, there appears to be three metal atoms per unit cell and yet bright
and dim maxima are seen as well. Determination of the atomic structure of the 2x4
surface and the origin of the differences between the Ho and Dy induced structures will

require complementary information from other surface analytic techniques.

3.1.5 Scanning rltrnneling Spectroscopy

Figure 3.11 (a) shows the STS I-V curves and normalized spectra of a NW, 3D island,
and the adjacent areas of reconstructed substrate at 0.56 ML. Since the 2D surface is
reconstructed by Ho, the spectrum of the substrate is different from that of clean Si. The

upper three curves are the actual I-V data, and the lower three curves are the normalized

 

I/dV

differential conductivity STS data were acquired at every pixel of the image

shown in Figure 3.11 (c) which is a partial image of Figure 3.11 (b). Each STS curve is
an average of the data taken in the area enclosed by the solid or dashed lines. The
normalized spectra show that the NW has non-zero conductivity at near-zero bias.
Figure 3.11 (d) is the STS I-V curves at near-zero bias voltage. It is clear that the I-V
curve of the NW is steeper than that of 3D island and the reconstructed substrate.
Although both NW and 3D island are metallic, NW has greater surface conductivity.

Bulk HoSiz is metallic with a conductivity of approximate 4.4)(103 [9 cm]'1 at RT [99].

53

   
      
   
 

 

(a) -— nanowire ’0'5

- - - SD island
----- substrate

 

 

- 0.4

 

(NI)/(Ap/Ip)

 

 

 

_ nano re
— - - 30 island
..... substrate

 

 

 

-o.2 -03 0'0 o.'1 o.'2
V(Vo|ts)

Figure 3.11 (a) The STS I-V curves and normalized spectra for a NW, 3D island, and
the reconstructed substrate. The surface topography is shown in (b) (150x150 nmz).
The image (c) is a partial image of (b). Each STS spectrum was averaged from the STS
data acquired in the areas shown by solid or dashed lines in the STM image shown in (c)
(40x40 nmz). (d) Magniﬁcation of the STS I-V curves at near—zero bias voltage.

54

3.2 Conclusions

We have studied the initial stages of Ho growth on the Si(001) surface. Metal
coverages between 0.06~0.9 ML were deposited onto substrates heated to 600°C, and the
samples were then cooled to RT for study by STM. Under these growth conditions, Ho
forms either highly elongated silicide NWs or compact 3D silicide islands, together with
several possible Ho induced 2D substrate surface reconstructions. STS spectra reveal
that the NWs are more metallic than either the large 3D islands or the reconstructed

substrate.

55

Chapter 4 Samarium induced surface reconstructions of Si(001)

In this chapter, the growth behavior of Sm on Si(001) is described, including coverage
dependence, 2D substrate surface reconstructions, and 3D island formation. Also we
provide the first report that there are two different surface reconstructions: a lower

coverage 2x3 phase and a higher coverage 3x2 phase.

4.1 Results and Discussion
4.1.1 Coverage dependence

Figure 4.1 shows the coverage dependence of Sm growth on Si(001) at 600°C. The
horizontal axis indicates the metal coverage in monolayers (ML), and both LEED
patterns and topography as seen by STM are shown. Gray circles are the metal
coverages at which the experiments were performed. The error in coverage is less than

20 %. Generally, a 2x3 phase dominates the 2D reconstructed surface for the coverages

 

 

 

aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa

 

. .
I D

 

 

. . 0‘
2D reconstructed surface only 3D island
0 0.5 1.0 1.5 [ 1

Figure 4.1 Coverage dependence of LEED patterns and topography for Sm/Si(001) at
600°C.

56

below 1 ML. At very low coverages (below 0.07 ML), only a 2x1 LEED pattern is
observed. A c(6x2) phase coexists with 2x3 phase above 0.54 ML as in the LEED

pattern shown in Figure 4.2.

 

Figure 4.2 LEED pattern of Sm/Si(001) at 0.54 ML shows 2x3 (black rectangle) and
c(6x2) (black arrows) periodicities. The LEED pattern was obtained from the surface
shown in Figures 4.8 (c) and 4.9 (a). The electron beam energy is 63 eV.

4.1.2 General surface structure evolution with metal coverage

An overall description of all of the ordered surface reconstructions will be presented,
followed by a more detailed discussion of each surface structure.

At low coverages, 0 < 0.2 ML, there are two Sm associated surface structures: a 2x3
ordered phase and a more sparse zigzag chain structure. Figure 4.3 shows a pair of
images (15x15 nmz) in (a) empty and (b) ﬁlled states at 0.067 ML. Two 2x3 unit cells
on the different terraces are shown with white and black rectangles on each image. The
rectangles are slightly distorted as a result of drift while acquiring the images. Also

shown are line traces taken from both images along the dashed lines A shown in Figures

57

4.3 (a) and 4.3 (b). A small terrace of clean 2x1 Si dimers is seen in the central part of
the image. Zigzag chains of brighter features are seen on the upper part of the terrace,
and there is a small patch of brighter ordered 2x3 reconstruction on top of this terrace as
well. Surrounding the central terrace are larger ordered areas of 2x3 reconstruction that
are one atomic height step lower than the brighter 2x3 patch in the central lower part of

the image.

 

 

 

 

 

 

 

 

 

0.20 - - 0.24 - -
0.16 -A - 02° ' '
I ' 0.16 -A -
0.12- - _ _
L l l l l l I l 0'12 -l I I l I I I r-
02468101214nm 02468101214nm

Figure 4.3 (a) Empty-state (1.90 V) and (b) ﬁlled-state (-1.91 V) STM images of the
same area showing 2x3 phase and the zigzag chain structures at 0.067 ML (15x15 nm‘).
Cross-sectional proﬁle along the line A is shown on the bottom of each image.

58

The trace along A in both images shows that the large areas of 2x3 are lower than the
central 2x1 terrace at both bias polarities [0.019 (0.039) nm lower in empty (ﬁlled) states].
This makes it reasonable to assume that the 2x3 is geometrically lower than the 2x1.
The height of the zigzag chains is similar to that of the 2X3 area on top of the central
terrace [about 0.116 (0.065) nm in empty (ﬁlled) states], and the upper 2x3 structure is
about 0.12 nm higher than the lower 2x3, which is about the single atomic height step
(0.125 nm) on the Si(001) surface. This supports the view that the 2x3 structure on both
terraces is in fact the same.

One can describe this 2x3 structure as being built up on top of a Si(001) terrace, or
embedded in a terrace that is one level lower. For the purposes of this thesis, we will
describe the 2x3 as being embedded into the level of the central terrace. This means
that the upper 2x3 structure is effectively embedded in the terrace above the one that is
not visible in this image. With this deﬁnition, the ordered reconstruction is denoted as a
2x3 structure embedded in a 2x1 clean Si terrace, implying that the 2x periodicity is
along the same direction as the Si dimerization direction of the surrounding clean Si
areas.

In the medium coverage regime (0.5 < 0 < 1 ML) the surface looks at first glance to be
similar to the lower coverages. Figure 4.4 shows an image of a surface at 0.59 ML,
which is covered in a combination of 2x1 Si and a 2x3 reconstruction. However, closer
examination reveals several differences. Firstly, the uncovered Si areas have a large
concentration of dark missing dimer defects that are arranged in zigzag patterns, rather
than bright protrusions. Secondly, the 3x periodicity of the “2x3” is perpendicular to

the Si dimer rows in the 2X1 terrace at a similar height in the center of this image. This

59

is perpendicular to the 2x3 structure seen at the lower coverages, and so this structure is
denoted as “3x2”. It should be noted that anti—phase defects in this 3X2 structure can
also produce a c(6x2) periodicity that can also be seen in the LEED patterns as shown in

Figure 4.2.

 

Figure 4.4 15X15 nm2 image of a surface at 0.59 ML. Both a 3x2 phase and a highly
defected Si 2x1 surface are seen. The bias voltage is —1.77 V.

In the high coverage regime (0 > 1 ML), 3D silicide islands are seen in coexistence
with a 2x1 reconstructed substrate. An STM image of a 3D island is shown in Figure
4.5. It is known that Sm-Si system forms Sm38i5 and SmSiz compounds [42, 44].
Although no elongated structure is observed for 3D Sm silicide compounds in this work,
it has been reported that SmSiz.x forms NWs on vicinal Si(001) substrate at the same
growth temperature [6]. This structural difference in the formation of silicide is likely

due to the metal deposition rate, sample annealing temperature, and substrate type.

60

 

Figure 4.5 Sm silicide island at 1.46 ML (300x300 nmz).

4.1.3 The low coverage 2x3 and zigzag chain structures

Figures 4.6 (a) and 4.6 (b) show empty-state (1.57 V) and filled-state (-l.57 V) images
of the same area at 0.067 ML, respectively. No 2x3 unit cells are marked with white
rectangles on each image. An out—of—phase boundary is also indicated with a dashed
line. This type of phase shift in 2X3 structures can be frequently observed. Si dimer
rows running diagonally are shown in the lower right region.

In both empty and filled states, two round maxima can be observed with different size
and contrast within each 2x3 unit cell. Here, the bright and dim maxima are represented
by circles and ovals in the 2X3 unit cell, respectively, on each image. By comparing
both bias images of Figures 4.6 (a) and 4.6 (b) with respect to the Si(001) substrate, the
registration of empty (white) and ﬁlled (gray) states of maxima of the 2x3 structure can
be diagramed as in Figure 4.6 (c). The 2X3 maxima are in registry with the centers of Si
dimers on any line that is drawn along the Si dimer row directions and 3x periodicities

are aligned along the [ITO] direction, as is readily apparent from these images. The

61

registry in the perpendicular direction has been determined from other similar images that
are not shown here. This registry of features is different from Ragan et al.’s result of

Sm on vicinal Si(001) where Sm atoms are positioned on the trenches between the Si

dimer rows [6].

    

(c) 00 1.57 V
:31 -1.57 V
H Si dimer
0 1st Si layer
0 2nd Si layer

 

Si [110]

 

Figure 4.6 (a) Empty-state (1.57 V) and (b) ﬁlled-state (-1.57 V) STM images of the
same area showing 2x3 phase at 0.067 NIL (10x10 nmz). There are out-of-phase
boundaries shown with dashed lines. Two 2x3 unit cells are marked with white
rectangles. (c) Line drawing showing the empty-state (white circles and ovals) and
ﬁlled-state (gray circles and ovals) maxima of the 2X3 structure.

On the line drawing, in empty states, the maxima (white circles and ovals) are situated

almost in the center of an underlying Si square lattice consisting of four Si atoms in each

62

corner, but they are slightly off center. The ﬁlled-state maxima (gray circles and ovals)
lie between two Si atoms, which is shifted by half as] (as, = 0.384 nm) from the
underlying Si square lattice. The dim maxima in empty states (white ovals) are
positioned almost between the ﬁlled-states maxima, while the bright maxima in empty
states (white circles) positioned on the dark region of ﬁlled states.

In fact, the number and the appearance of the maxima appearing in STM images are
bias dependent. In empty states, below 2.56 V, two maxima can be seen with different
contrast in the 2x3 unit cell [Figures 4.3 (a) and 4.6 (a)]. Above 2.74 V, only one
maximum is visible (not shown). In ﬁlled states, below —1.85 V, there are two maxima
in the 2x3 unit cell [Figure 4.6 (b)], while only one maximum can be seen above -1.90 V
[Figure 4.3 (b)]. The bias can also affect the relative height of the 2x3 structure and the
2x1 clean Si surface.

Zigzag chain structures are also observed perpendicular to the Si 2x1 dimer rows as
shown in Figures 4.3 (a) and 4.3 (b). The maxima of the zigzag chain structures in
empty and ﬁlled states are located over the trenches between the Si dimer rows. The
features of maxima in 2x3 structure and zigzag chain structures appear the same,
although they look different due to a tip effect in Figure 4.3 (a). Figure 4.7 (a) shows
the more detailed features of zigzag chain structures with surrounding buckled Si dimers.
An area of 2x3 phase can be also seen in the lower right area. In this bias voltage (-1.82

V), the maxima of dimerlike pairs can be observed to have an oblong shape. Comparing
the two ﬁlled-state images {—1.91 V for Figure 4.3 (b), and -1.82 V for Figure 4.7 (a)], the

paired maxima in Figure 4.7 (a) are positioned over the large maximum in Figure 4.3 (b).

The dimerlike pairs are 2aSI wide in [110] direction and shifted by lag, in [110] direction

63

resulting zigzag chain structures. Since these dimerlike pairs have a width of 203,, there
is not enough space to align straight on the trenches with a 2as; periodicity, and thus
zigzag chains are generated. However, they can align in a row with a 3a5I periodicity
forming 2x3 structure as indicated with a rectangle A in Figure 4.7 (a).

The maxima in the zigzag chain structure can move along the dimer rows under
repeated scanning with STM, and so the pattern of zigzag chain structure slightly changes
from image to image. In fact, the movement of zigzag chain causes noise in STM
images as appears in the oval area B in Figure 4.7 (a). However, the arrangement of 2x3
reconstruction in the small patch does not change even after 10 scans.

In addition, Sm zigzag chain structures induce buckling of the Si dimer rows with
c(4x2) arrangement in which the pattern of buckling is out of phase as illustrated in
Figure 4.7 (b). A c(4x2) unit cell is marked with a black rectangle. The Si dimer
buckling extends possibly more than 20 dimers along a row without decay. However,
the buckling does not affect the adjacent row: if there is no Sm atom in a certain row, the
buckling does not occur in this row even though there are Sm atoms forming zigzag chain
structures in the adjacent rows. This implies that the interaction between Srn and Si
atoms in the same row is very strong, whereas the Sm-Si interaction in the perpendicular

to the dimer rows is ignorable.

83?:
1’23:
32?:

O
0

,:.°.:’
$080.0
,:::8
a?
2:00

0008

3:
3:
3.
3°

8 o
23
12?

0.000

 

O O O 0
0‘0 0
O O '
.3.
0’0 0
O O O 0
O
O O O
0 0:0
0
O
0 0‘0
0 0:0
0 O 0

Figure 4.7 (a) Filled-state (-1.82 V) image of zigzag chain structures with buckled Si
dimers below 0.1 ML (10x10 nm‘). The 2x3 unit cell is also shown as a pair of two
maxima on the lower right area. (b) Line drawing illustrating the zigzag chain
structures with surrounding buckled Si dimers. This dimer buckling causes c(4x2)
periodicity. A c(4x2) unit cell is marked with a black rectangle.

The comparison of the experimental coverage with the percentage of the 2D surface
covered with Sm provides a result that there is one Sm atom in 2x3 unit cell. This is
again different from Ragan et al.’s result which suggests two Sm atoms in a 2x3 unit cell
[6]. It is not possible to propose a model for this phase based on the limited amount of
information that we can deduce from the STM data. However, we note that the
appearance, bias dependence, and registration of STM maxima are all very similar to that
observed for a Ba induced 2X3 phase on Si(001) [149]. Ba is divalent with an ionic
radius of 0.135~0.l63 nm and so it might be expected to behave similarly to divalent Sm

(ionic radius 0.113 nm). However, the Ba model proposed in ref [149] requires 2 metal

atoms per unit cell.

65

4.1.4 The medium coverage regime 3x2 and c(6x2) phases

As noted previously in the coverage regime (0.5 < 0 < 1 ML), the surface is covered in
a combination of highly defective Si 2x1, and Sm reconstructed 3x2 and c(6x2). A
large-scale image is shown in Figure 4.8 (a). The unusual step structure is a signature of
large scale Si adatom displacement during the formation of a reconstructed surface.
Similar step structures have been seen in the case of B on Si(001) [150, 151] and Ge on
Si(001) [152, 153]. Terrace A has a straight step edges (SA step) which run along the
dimer row direction, and terrace B has jigsaw-shaped step edges (83 step). Both terraces
are covered with the same surface structures: the mixture of 3x2 and c(6x2) phases, and
Si dimer rows with zigzag defect structures.

In Figure 4.8 (b), the area of the front edge of the jigsaw shape on terrace B shows the
alternate terraces A and B. 3x2 and c(6x2) phases can be seen in the four corners of the
image and Si dimer rows with diagonal rows of missing dimer defects appear in the
middle. Close-up image of zigzag defects is shown in Figure 4.8 (c). The ends of
zigzag defects are indicated with white or black arrows in Figures 4.8 (b) and 4.8 (c).
The angles between Si dimer rows and the zigzag defects (0° S B S 90°) vary from 41 to
76 degrees depending on the surface area. These missing dimer defects relieve strain
due to a low density of Sm atoms substituting into the Si surface. The Sm atoms do not
show up directly in the empty state images, but they make the ﬁlled state images of the
2x1 structure appear disordered.

Figures 4.4 and 4.8 (b) shows the coexistence of Si 2x1 and the 3x2 phase. From this
image, the 3x2 phase is measured to be about 0.079 (0.078) nm below the surrounding Si
2x1 areas in empty (ﬁlled) states. This measured height is slightly different from that of

the lower coverage 2x3 structure, and once again, the reconstruction is rotated by 90°.

66

 

Figure 4.8 (a) Large—scale STM image of a surface consisting of two types of step edges.
Straight step edges (S A step) run along the dimer row direction on terrace A, and terraces
B have jigsaw-shaped step edges (SB step). (b) Both types of terraces consist of 3x2 and
c(6x2) phases and Si 2x1 dimer with zigzag defects shown with white arrows. (c)
Close-up image of zigzag defects on Si dimer rows. The ends of four zigzag defect
structures are marked with black and white arrows. All three images were at different
Sm coverages: (a) 0.93 ML, (b) 0.59 NIL, and (c) 0.54 ML. The image sizes are (a)
300x300 nmz, (b) 50x50 nmz, and (0) 25x25 nmz. Bias voltages are (a) —2.31 V, (b)
—1.77 V, and (c) 1.28 V.

67

 

Figure 4.9 10x10 nrn2 images in (a) empty states (1.28 V) and (b) ﬁlled states (-1.77 V)
showing 3x2 and c(6x2) phases at different coverages: (a) 0.54 ML and (b) 0.59 ML.
Both 3x2 and c(6x2) unit cells are marked with white rectangles on each image. (c)
Line drawing showing the empty-state (gray circles) and ﬁlled-state (dark circles)
maxima of the 3x2 and c(6x2) structures. Si dimers are shown on the leftmost row as a
reference.

Figures 4.9 (a) and 4.9 (b) show the empty- and ﬁlled-state images of 3x2 and c(6x2)
structures, respectively. Both 3x2 and c(6x2) unit cells are marked with white
rectangles on each image. The STM images were acquired from the surfaces at different
coverages. In empty states [Figure 4.9 (a)], two types of round maxima are observed

with slightly different contrast. Bright and dark maxima appear alternatively with a

1.5a5I periodicity in the perpendicular to the dimer rows (along the [110] direction). The

68

bright and dark maxima each form rows along the dimer row direction ([1 I0] direction)
with a 2aSI periodicity. In ﬁlled states [Figure 4.9 (b)], only one type of maximum is
visible. In the similar way as empty states, the ﬁlled-state maxima lie along the [1T0]
direction with a 2a3i periodicity forming rows. These rows appear along [110] direction
with a 3am periodicity.

Figure 4.9 (c) shows the registration of the maxima in the 3x2 and c(6x2) phases with
respect to the Si(001) substrate. The registration was decided by the same surface area
which is not shown here. In empty states, the bright and dark maxima are distinguished
with large and small gray circles, respectively. The ﬁlled-state maxima are marked with
dark circles. All the maxima are located on the bridge site in both biases. For the 3x2
structure in empty states, bright maxima are situated in the center of underlying 1x1 Si
lattice, while dark maxima locating between bright maxima lie on the two underlying Si
atoms. The ﬁlled-state maxima are positioned between the empty-state bright maxima
on the same rows. When the positions of the maxima are shifted by lag, in the dimer
row direction, c(6x2) structure can be obtained. Thus 3x2 and c(6x2) phases have
basically the same structure. The number of Sm atoms per 3x2 unit cell is deﬁnitely
more than 2 according to the experimental coverages (0.54 and 0.59 ML). However,
this coverage can be affected by the possibility that 3D silicide compounds sparsely exist
somewhere on the surface which STM did not ﬁnd. When the coverage increases to
0.93 ML, more than 90 % of the surface is covered with Si 2x1 with zigzag defects and
few 3x2 phase can be observed, although the LEED pattern shows 2x3, c(6x2), and 2x1

phases.

69

4.2 Conclusions

In the initial stage of Sm deposition on Si(001) at 600°C, the surface is reconstructed
with a 2x3 phase. Sm-induced zigzag chain structures coexist with a 2x3 structure.
STM observations show that the appearance of the 2x3 structure is bias-dependent. As
the coverage is increased, a second 3x2 phase appears with a co-existing c(6x2) phase.
The 3x2 phase is different in appearance from the lower coverage 2x3 phase. At higher
coverages, large 3D Sm silicide islands are found. This research was partially supported

by NSF grant DMR-0305472.

70

Chapter 5 Topographic evolution of Dy silicide on Si(001):
Shape transitions from nanowires to 3D islands

In this chapter, we show that DySiz NWs can be formed with a uniform width by
controlling the deposition rate and the deposition duration, which will play an important
role for nanodevice assembly. The shape transition from NWs to 3D islands is studied
by scanning tunneling microscopy (STM). Low energy electron microscopy (LEEM) is
used to study Dy growth on longer length scales that are not accessible by STM.
Dynamic behavior and growth of 3D Dy disilicide islands at elevated temperatures are

reported.

5.1 Results and Discussion
5.1.1 STM experiments

Figure 5.1 shows STM images of DySiz NWs at 0.58 ML (a) before and (b) after
annealing at 600°C for 8 minutes. As shown in Figure 5.1 (3), almost the whole surface
is covered with a 2x7 substrate reconstruction. There are also some NWs with a
comparatively uniform width of 7 times the lattice constant of silicon as, (= 0.384 nm).
Formation of NWs with uniform widths with My may be attributed to coexistence with
the well-ordered 2x7 substrate reconstruction.

A more detailed image of the 2X7 superstructure is shown in Figure 5.2 (a). Two unit
cells are marked with white rectangles. The detailed atomic structure of Dy 2x7
superstructure on Si(001) surface is reported by 31. Liu et al. [72]

Rapid deposition provides only 2x7 superstructure on the 2D surface regardless of the

metal coverage. Six different metal coverages from 0.19 to 0.84 ML were obtained by

71

adjusting the deposition rates and keeping the deposition time fixed for rapid deposition
experiments. Surface morphology at those coverages only consists of 2X7
superstructure on 2D surface and NWs for 3D structure. More NWs are observed at
higher coverages and NWs tend to bundle above ~0.8 ML. On the other hand, 2x7
superstructure dominates on 2D surface at any coverage between 0.19 and 0.84 ML.
After 8 min annealing at 600°C, bundled NWs are more often observed and there are
more second layer growth on top of the NWs as shown in Figure 5.1 (b). In comparison
with ‘before’ annealing, the NW width distribution ‘after’ annealing is broader because of
the presence of bundled NWs. However, the overall percentage of the surface occupied

by NWs does not change significantly ‘before’ (15.1 %) and ‘after’ (18.5 %) annealing.

 

Figure 5.1 STM images of DySiz NWs at 0.58 ML (a) before and (b) after annealing at
600°C for 8 minutes. (a) Dy was deposited at 600°C. Almost all the 2D surface is
covered with 2x7 stripe structure and NWs have comparatively uniform width
distribution. 2x7 phase generally disappears after 1 min annealing at 600°C. (b) After
annealing, 2x4 superstructure coexists with bare silicon. More bundled NWs are
observed and the width of NWs becomes wider. There is also more second layer growth
on top of the NWs.

72

The 2x7 superstructure is swept away and a disordered 2x4 superstructure appears with
bare silicon on the 2D surface as shown in Figure 5.2 (b). Two 2x4 unit cells are
marked with white rectangles. The 2x4 superstructure consists of one or two round
bright maxima in ﬁlled states. The detailed atomic structure of Dy 2X4 phase on

Si(001) is explained by B.Z. Liu et al. [10]

30 X 30 nm2

 

Figure 5.2 Small-scale images of (a) and (b) in Figure 5.1 showing (a) 2x7 (-1.80 V)
and (b) 2x4 (+1.04 V) reconstructed surface around NWs. Two unit cells are marked
with white rectangles in each image.

Generally, more than 1 min annealing at 600°C replaces the whole 2x7 superstructure
by 2x4 superstructure coexisting with bare silicon surface. Immediate disappearance of
the 2x7 phase can be seen by LEED when the sample heating starts. 2x1 and x4 LEED
patterns are observed during annealing at 600°C after the 2x7 phase disappears.

The extreme sensitivity of the 2x7 superstructure to annealing is characteristic of the
overall ease with which the 2X7 can be converted to the 2x4 structure. It can be further

noted that the balance between 2x7 and 2X4 on the substrate is also sensitive to

73

deposition rate. In earlier work where the Dy deposition was slower, 2x7 and 2x4
coexist on the substrate over a range of metal coverages. The present results show that
fast deposition with post deposition quenching results in a pure 2x7 substrate phase
which can be converted rapidly to 2X4 by annealing.

Figure 5.3 shows the evolution of the 0.58 ML sample shown in Figures 5.1 and 5.2,
after repeated annealing cycling. The percentage of the surface covered with 2D
substrate reconstructions (2x7, 2x4, or 2x1) and 3D structure (NWs or islands) is
expressed in the vertical axis. Annealing processes are shown in the horizontal axis.

The sample was annealed seven times in total: 1St annealing (600°C 8min), 2"‘1 (600°C 20

(%l

   

  

5’ Around
the NWs

 

Around
the Islands

  

 

(a) Before (b) 1st (c) 2nd (d) 6th 531 7th
Annealing Annealing Annealing Annealing neallng
600°C 600°C 700°C 720°C
8 min 20 min 10 min 10 min

Figure 5.3 Annealing behavior of Dy grown on Si(001) at 0.58 ML. 3D structures of
DySiz transit from large 3D islands by annealing. 2x7 superstructure dominates on the
2D surface after rapid deposition of Dy. After 1 min annealing at 600°C, 2x7
superstructure is replaced by 2X4 superstructure and bare Si. Less 2x4 structure appears
with higher temperature annealing or longer annealing time.

74

min), 3rd (635°C 12 min), 4” (655°C 10 nrin), 5th (678°C 10 rrrin), 6th (700°C 10 min),
and 7th (720°C 10 min). STM images were taken after each annealing step. The
large-scale STM images of changes in surface morphology (a)~(e) in Figure 5.4
correspond to the bar graphs labeled with the same letters in Figure 5.3. The measured
areas were averaged over more than ﬁve images after each annealing step. The results
of 3rd ~ 5th annealing are omitted for simplicity because the surface morphologies after 3“1
~ 5th annealing are very similar to that after 2“d annealing.

Elongated NWs with uniform width and 2x7 superstructure dominate on the surface
after rapid deposition without post-annealing [Figure 5.3 (a)] as already shown in Figures
5.1 (a) and 5.2 (a). After 1St annealing, the substrate surface reconstruction completely
changes to 2x4 superstructure while the percentage covered by NWs remains almost the
same. The fraction of the substrate covered with 2x4 superstructure [Figure 5.3 (b)] is
66 i 8 % and the remainder is bare Si. STM images indicate that 2x4 superstructure
does not unifonnly distribute on the 2D surface. Furthermore, there seems to be no
clear inﬂuence of either steps or NWs on the distribution of 2x4 superstructure on the
surface.

The metal content on the surface is conserved before and after the 1" annealing step at
600°C. The calculation below indicates that the increased quantity of Dy atoms in the
NW 3 is equal to the decreased quantity of Dy atoms on the 2D surface. Firstly, the
calculation for the metal content in NWs is performed. Dy metal density in the DySiz is
poy = 19.1 atoms/nm3, where the lattice parameters of hexagonal DySiz structure are a =
0.3831 nm and c = 0.4121 nm [36]. If we assume that the minimum height NWs

include one bulk DySiz unit cell high (= one Dy layer plus two silicon layers = 0.33 nm

75

high), 0.93 ML (= 19.1 atoms/nm3 X 0.33 nm = 6.30x1014 atoms/cm‘) of Dy covers the
whole surface as DySiz NWs. Because there is ~20% of second layer growth on top of
NWs after annealing, the percentage difference of metal content in NWs before and after
annealing is: A(NWs) = 18.5 % x 1.2 (after annealing) — 15.1 % (before annealing) =
7.1 %. Therefore, 0.066 ML (= 7.1% of 0.93 ML) of Dy atoms is expected to have
moved from the 2D reconstructed surface to the NWs by annealing.

The second part is the calculation of the metal content on 2D surface. If we assume
that there are ﬁve (three) Dy atoms per 2x7 (2x4) unit cell [72], the difference in metal
content on the 2D surface before and after annealing is: A(2D) = 0.357 ML x 85 %
(before annealing: 2x7) — 0.375 ML x 66 i 8 % (after annealing: 2x4) = 0.055 i 0.030
ML. Therefore, there are less Dy atoms on the 2D surface after annealing. This
coverage range of 0.055 i 0.030 ML for 2D surface also agrees with the change in
coverage of 0.066 ML from the calculation for NW 5.

After the 2“d annealing, 3D compact silicide islands start appearing, although NWs are
still dominant on the surface [Figure 5.4 (c)]. This phenomenon supports the results
reported by 32. Liu et al. [12] The longer side of small 3D islands tends to be oriented
along the direction of NWs. More 2x4 superstructure can be seen around NWs than 3D
islands [Figure 5.3 (c)]. The percentage of the 2D surface covered with 2x4
superstructure is 36 i 13 % (21 i 4 %) around NWs (3D islands). This result implies
that 3D islands are a more favorable state than NW 5 since there are less Dy atoms around
3D islands than NWs. Additional annealings increase the number of 3D islands and the
island size. At the same time, the number of NWs decreases and the surface

reconstruction is gradually swept away. After the 7th annealing at 720°C, various sizes

76

of 3D islands are visible and very few NWs are observed on the surface [Figure 5.4 (c)].
There is no obvious metal induced reconstruction left on the surface, but surface defects
are present on approximately 10 percent of the 2x1 bare Si(001) substrate. It is difficult
to conﬁrm the mass conservation of Dy in this limit since the 3D islands are very sparse
and the size range of the 3D islands is very wide. Also, Transmission Electron
Microscopy (TEM) study reveals that large 3D silicide islands extend underneath the
plate of the surface [154], making it difficult to accurately calculate their volume from

what is seen in the STM data.

 

Figure 5.4 Transition from NWs to large 3D islands of DySiz by annealing. (a) 0.58 ML
of Dy was deposited at 600°C, (b) 1St annealing at 600°C for 8 min, (c) 2“Cl annealing at
600°C for 20 min, (d) 6‘11 annealing at 700°C for 10 min, and (e) 7Lh annealing at 720°C
for 10 min. All image sizes are 400x400 nmz.

77

5.1.2 LEEM experiments

Figure 5.5 shows (a) the LEED pattern, (b) a dark-field LEEM image, and (c) a
bright-field LEEM image of 0.78 ML of Dy grown on Si(001) at ~600°C. Under these
conditions, the surface should be covered in the 2X7 reconstruction plus NWs as in

Figures 5.1 (a) and 5.4 (a). All three images were taken at RT after deposition. The

 

Figure 5.5 (a) The LEED pattern of Dy grown on Si(001) at ~600°C (0.78 ML). 1x7
phase and 1/2 order streaks are shown. The electron beam voltage is 60 V. (b)
. Dark-ﬁeld LEEM image (2 eV) taken from 2/7 spot shown with the white arrow in (a).
The ﬁeld of view is 4 pm. (c) Bright-ﬁeld LEEM image (9 eV) in the same area of (b)
showing the stripe structure. Contours of terraces are drawn with white lines. The
stripes are oriented at an angle of 90 degrees on each terrace, which correspond to the
NWs.

78

electron beam voltage is 60 V in Figure 5.5 (a). The diffraction pattern with 7x
periodicity and 1/2-order streaks are similar to typical LEED pattern Observed from 2x7
reconstructed surface of Gd, Dy, or Ho grown on Si(001) at 600°C [72]. Figure 5.5 (b)
shows the dark-ﬁeld LEEM image (2 eV) taken from the 2/7-order diffraction spot
denoted with the white arrow in Figure 5.5 (a). The ﬁeld of view is 4 um. A clear
bright-dark contrast is observed on alternate terraces. Each terrace shows a
cross-hatched texture, as well as a few small isolated islands. The spacing of the
cross-hatching is about 50 nm, and it is oriented along <1I0> directions but it is not a
direct image of the NW structures. Figure 5.5 (c) is a bright-ﬁeld image (9 eV) in the
same area of Figure 5.5 (b) showing the striped structure. The contours of terraces are
drawn with white lines. There is a rough alteration in stripe direction across the terraces,
which is consistent with NW related features, but and again, the apparent size and
spacing are too large to be the NWs.

Figure 5.6 shows the sequence of LEEM images presenting the growth of Dy disilicide
islands (0.78 NH.) during annealing. Annealing duration time and the temperature are
indicated on the lower left side, and imaging mode (BF: bright ﬁeld, TBF: tilted bright
field) and the incident electron energy are indicated on the lower right side of each image.

As already mentioned in Section 5.1.1, the 2x7 superstructure is replaced by a 2x4
superstructure after 1 min annealing. Also, for LEED pattern in Figure 5.5 (a), the
position of 2/7-order diffraction spot is very close to that of 2x4-diffraction spot. Since
the aperture used is not small enough to distinguish those two spots, 2/7-order and/or the
2x4-diffraction spot are used to form the dark-ﬁeld image during annealing. The shape

of the terraces in the dark-ﬁeld image remains the same up to ~620°C. This clariﬁes

79

51 '30" ' 56'31"
742°C 745°C

 

Figure 5.6 Sequence of LEEM images showing the growth of DySiz islands (0.78 ML).
The ﬁeld of view is 4 pm. The white arrows show the same island. Small grain
structures start appearing at ~700°C and grow as islands. BF: Bright ﬁeld, TBF: Tilted
bright ﬁeld.

80

that silicon atoms does not move on a macroscopic scale when the 2x7 to 2x4 phase
transition occurs. The terrace contrast becomes weaker above ~700°C, although the
terraces still have the original proﬁle.

The annealing temperature was gradually increased from RT to ~750°C for 60 minutes.
The relationship between annealing duration and temperature change above ~600°C is
shown in Figure 5.7. In Figure 5.6, the white arrow points out the same island and its
surrounding area in each image. Different islands labeled with numbers in Figure 5 .6 (f)
are classiﬁed as follows: (#1-5) large islands, (#6-10) medium islands, and (#11-13)
small islands. The incident electron energies of 2~3 eV, 4.5 eV, and 11~12 eV give the
best contrast for surface terrace in the dark-ﬁeld image [Figure 5.5 (b)], step phase in the
bright-ﬁeld image [Figure 5.6 (f)], and stripe structures in the bright-ﬁeld image [Figures
5.5 (c) and 5.6 (b)], respectively. Tilted bright ﬁeld (TBF) image provides the

information of both island growth and surface terrace morphology.

 

 

 

 

750 9 ﬂ
9. o 0
9
:3 700 ,9. ......................................
92 99
3 o
*‘ o
e 650 .............. 3 .............................................................
3 .
.O M
600 5’ . . - - - - .
30'00" 40'00" 50'00" 60'00"

Annealing duration

Figure 5.7 Annealing duration vs. temperature. Annealing temperature was gradually
increased from RT (00’00”) to 750°C. The black arrow shows the time range during
which the islands are observed as shown in Figures 5.8 and 5.9.

81

The nucleation of small structures that eventually grow into islands starts at ~700°C
[Figure 5.6 (a)]. However, the initial stage of the nucleation process cannot be seen
because of the limited resolution, and small islands are hardly distinguishable from the
darker contrast in the stripe structure. Islands become clearly visible 15 seconds after
nucleation. In most cases, the shape of the islands is a rounded rectangle, but a few
square (#4) and elongated (#5) shapes can be also found [Figure 5.6 (1)].

During annealing process, three types of island growth are observed as shown in
Figure 5.8. Time evolution of the length and the width of three different islands [#4, 5,
and 6 in Figure 5.6 (f)] are plotted. The length and the width of islands are marked with
black squares and white triangles, respectively. Note that the vertical axes in Figure 5.8
(and Figure 5.9) show the length and the width of 3D islands including the area of
surrounding trenches which are clearly visible in STM [Figure 5.4 (c)] but are not
resolvable from the actual islands in the LEEM. Therefore, the actual island dimensions
are slightly smaller than those shown in the graphs. However, it is clear enough to
understand the growth behavior of islands since the surrounding Si depletion areas
approximate the shape of islands.

The ﬁrst type of island growth is square growth [Figure 5.8 (a)] in which the width and
the length of the island increase at the same rate. The square island (#4) grows in this
manner. The second type of growth is elongated growth [Figure 5.8 (b)] in which the
length increases while the width remains constant. The elongated island (#5) is this case.
The third type of growth is the most common growth, rectangular growth [Figure 5.8 (c)].
Both width and length increase uniformly but the growth rate of length is faster than that

of width. Here, island #6 was illustrated by an example in Figure 5.8 (c) although the

82

other islands (#1-3 and #7-13) grow in the similar way. Except for the elongated island
(#5), the island width and length grow uniformly in all the directions while maintaining
the same aspect ratio. The aspect ratios of square (#4) and elongated (#5) islands are
1.2:1 and 7.321, respectively, while those of rectangular islands are in the range of 1.8:1 ~

4.6:1.

 

 

 

 

 

 

 

 

 

 

 

 

 

’é‘ (a) Square growth: #4

5 200 I I

% 150~-----------°-'---2.r-2"Z'é ---------- 'ﬁ """"""

E 100» ------ ﬁg.“- --------------------------------

g I

5 sou-13L ------------------------------------------

g) 0 . . .

3 45 50 55 60
Annealing time (min)

A (b) Elongated growth: #5

g 00 ll""I I I i

1:: 400 -------- 'J' ------------------------------------

E 300+ -----------------------------------------------

I

‘2 200 ------------------------------------------------

m 100 ------------------------------------------------

g» 0 Mat: A A - A A -

3 45 50 55 50
Annealing time (min)

A c Rectan ular rowth:#6

E 50( ) 9 9

V ..................................... 1.1 .....

g 200 . I I I

'5 150» ------- #9- ---------------------------------

g 100 ----- .‘i'x """"" a “8A """""" A‘A """

g 50052214:1M9 --------------------------------------

C

a: 0 ‘ ‘ ‘

-‘ 45 50 55 60

Annealing time (min)

Figure 5.8 Time evolution of the length and the width of 3D islands and the
surrounding Si depletion areas in different growth types during annealing.

83

Figure 5.9 shows the change in the size of the islands by annealing. Black, gray, and
white marks indicate large, medium, and small islands, respectively. Numbers from 1 to
13 correspond to the islands labeled with the same numbers in Figure 5.6 (f). The labels
(a)~(f) indicate times at which the LEEM images in Figure 5.6 were captured. The
island growth starts and ceases in the same time regardless of the island size. This
implies that the growth rate of larger islands is faster than that of smaller islands. After
5 minutes of island growth, the island size remains constant up to ~750°C. In general,
the island sizes do not change up to ~760°C and Dy metals start leaving the surface by

annealing at above ~780°C according to the experiments under similar conditions.

 

 

 

 

 

 

 

 

 

 

Large islands

3 __ 10 2A 39
g 4. 5—
E E Medium Islands
E C 532;. =;:;.
a :32 9 .312: 10-
a) V 2 Small islands
5 t0 11012A130
"o 9
c to
to c
a) o
‘U '4: 1
c a)
1.“ a
.‘2 g
8 '0')

0

40 45

 

Annealing time (min)

Figure 5.9 Change in the island size by annealing (0.78 ML of Dy). Black, gray, and
white marks indicated large, medium, and small islands, respectively. The islands
numbered from 1 to 13 correspond to the islands with the same numbers in Figure 5.6 (f).
The labels (a)~(f) are marked for the time at which the LEEM images in Figure 5.6 were
taken.

84

Silicon terraces start breaking up during the island growth [Figure 5.6 (c)]. After the
island growth ceases, terraces start rearranging themselves to form regular Si(001)
surface [Figure 5.6 (e)] with 2x1 periodicity of the LEED pattern. TBF images of
Figures 5.6 (c) and 5.6 (e) have an opposite contrast due to the different incident electron
energy and the different beam tilt in the azimuthal angle. As the traces of streaked
structure shown in Figure 5.6 ((1), small dots are left on the terraces with weak contrast.
This might be a reﬂection of surface defect areas that are generally observed on STM

images as explained in Section 5.1.1.

5.2 Discussion

A DySiz NW has a hexagonal Ale structure. The formation of DySiz NWs arises
from the anisotropic lattice mismatch between DySiz and the Si(001) substrate. The
orientation relationships of hexagonal DySiz on Si(001) are determined to be [5, 9, 12]:

DySi2(lTOO)// Si(001), DySi2[OOOl] //Si[1'1‘0], and DySi2[11§O] //Si[110].

Comparing the lattice mismatch in a and c directions of the hexagonal structure with
Si(001) 1x1 unit cell, the lattice mismatch in the a direction is very small (-0.23 %). On
the other hand, the lattice mismatch in the c direction is large (+7.32 %). Because of
those anisotropic lattice-mismatch strains, hexagonal DySiz is favored to grow in the a
direction than the c direction. Therefore, a and c directions of hexagonal structure
correspond to a length and width of NW, respectively.

When NWs are grown under conditions where the substrate is at least partially covered
in the 2x4 phase, then the widths of NW5 are relatively broadly distributed between about

2as, and 16a3i. The limit that one would expect for coherent growth is about 14am,

85

given the lattice mismatch in the direction of the NW width. However, as shown in this
work, if the substrate is dominated by the 2x7 phase, most of the NW widths are 7as,
which suggests that the presence of the 2x7 surface reconstruction can affect how the
NWs nucleate and grown. The stronger inﬂuence of the 2x7 phase may be due to two
factors. Firstly, even in the case of the 2x4 surface, most NWs are wider than 3 or 4a3i
which suggests that NWs prefer to grow wider than the width of the 2x4 unit cell.
Secondly, it is clear in images of surfaces with a mixture of 2x4 and 2x7 that the 2x7
phase is much more highly ordered. As can be seen in Figure 5.1 (a), the bright rows of
the 2x7 phase are ordered over a similar length scale as adjacent NWs. Annealing at
600°C [Figure 5.1 (b)] converts some of the 2x7 into 2x4, and at the same time, NWs
appear to have moved and reconﬁgured since they are much more likely to be seen in
bundles, and they are much more likely to show second layer growth. Figure 5.1
illustrates that the NWs are a metastable phase, and that the structural order and
uniformity of the NWs are a very sensitive function of growth conditions.

Islands grow in the temperature regime where conversion from NWs to 3D islands is
seen in the STM images shown in Figure 5.4. Speciﬁcally, signiﬁcant conversion of
NW to 3D islands is not seen even after extended annealing at 6005C [Figure 5.4 (c)],
whereas the annealing above 700°C causes the signiﬁcant conversion of surface
morphology [Figure 5 .4 (d)]: transition from NWs to 3D islands.

When the NW width reaches a certain critical width during annealing, where the
maximum width mismatch between DySiz and Si(001) substrate reaches lag, a
dislocation is generated in order to release the misﬁt strain. Cross-sectional TEM

observation shows two types of misﬁt dislocations at the interface between 3D islands

86

and Si substrate [154]. In a Type I dislocation, there are several tilted misfit dislocations
along the interface between 3D islands and Si substrate. In a Type H dislocation, the
whole island is tilted at a small angle with respect to the Si substrate, but there is no
partial misﬁt dislocation. When the Type I dislocation occurs, the DySiz can readily
grow in both directions resulting in the formation of rectangular islands.

For the elongated island [island #5 in Figure 5.6 (0], there is a larger misfit strain
along the width direction, since the width of an elongated island is narrower than that of
rectangular islands. If the misﬁt dislocation along the width direction is suppressed
during annealing for some reason, the misﬁt strain relaxation occurs in the length
direction and the island length increases. The formation of the elongated island might
be related to a Type H dislocation which has a tilted crystal structure. After all, it is
possible that 3D islands can be formed in different growth modes.

DySiz NW 8 grow in the Stranski-Krastanov (SK) mode. As 3D islands start growing,
the growth mode changes from SK mode to Volmer-Weber (VW) mode [Figure 5.3].
When the island nucleation occurs, Dy atoms are desorbed from the 2D surface and
attach along the island edges resulting the linear increase of the island area. The
island-size distribution is broadened. Once Dy atoms attach to the 3D islands, they
preferentially remain on the island regardless of the island size. It seems that the
attachment probability of Dy atoms to 3D islands is equal, so that larger islands can
attract more atoms because of longer circumferences. Therefore, the island growth rate
is size-dependent: Larger and smaller islands grow with faster and slower growth rates,
respectively [Figure 5.9]. The relative size differences of large and small islands do not

change during the island growth.

87

Even though the general behavior of islands in all size ranges is similar, Figure 5.8
clearly shows that islands can exhibit both a variety of morphologies and associated
growth behaviors. The most common island is rectangular, and although there is a
distribution of aspect ratios seen, each individual island is seen to maintain the same
aspect ratio during growth. Square growth is a variation of this behavior. Finally, a
few islands show elongated growth, where the width of the island is ﬁxed as the island
lengthens. This type of 1D growth has been seen in the case of other silicides, including
ErSiz, and it has been related to various theories. However, the diversity in growth
behaviors and island morphologies seen in the LEEM data in Figure 5.6 make it clear that
at least in this system, one cannot make any generalizations about growth in this system.
Furthermore, it is dangerous to make a case for growth behavior base on just one type of
island.

For the RE silicides, it is possible that some of these different behaviors could be
related to the presence of different silicide structures. DySiz can crystallize into
hexagonal, orthorhombic and tetragonal phases, all of which have different lattice
constants. It is tempting on geometrical grounds to associate square, rectangular, and
elongated growths with the tetragonal, orthorhombic, and hexagonal phases, respectively.
In fact, TEM studies of 3D islands grown in this system have shown that islands can be
either hexagonal or orthorhombic / tetragonal, at least at a growth temperature of 600°C.
Correlation of island morphology and silicide crystal structure requires further plan view
TEM work. The LEEM data relates island morphology to growth behavior.

All NWs are gone at 720°C, which is roughly the temperature at which the 3D islands

stop growing in the LEEM measurement. Essentially, growth stops at the point where

88

there is no longer a source of Dy available, from either the substrate reconstruction, or the
NWs.

After island growth stops, there is no further evolution of the islands, even after
extended annealing at 750°C. No ripening or coarsening is seen. This is consistent
with the behavior reported for Er silicide islands on Si(001) annealed at 800°C [76].

One other interesting point is that the Si steps are observed to wander after island
growth stops, whereas they are largely static during the conversion of NW to islands.
Given the strong interaction between NW and single height step conﬁguration seen in the
STM images, it is possible that the substrate steps are pinned in place by a few remaining

NW, until the conversion to 3D silicide islands is complete.

5.3 Conclusions

After rapid Dy deposition, almost all 2D surface is covered with 2x7 superstructure,
and NW 5 have uniform width with 7a3i which is 7 times as wide as the lattice constant of
silicon. Surface reconstruction phase immediately changes from 2x7 to 2x4
superstructure by annealing, while the percentage of NWs remains almost the same.
Therefore, this fact insists that 2x7 superstructure is the metastable state. The metal
content on the surface also conserves before and after annealing at 600°C. Annealing at
700°C increases the number of 3D islands and the average island size. At the same time,
the number of NWs decreases and surface reconstruction is swept away. This fact
insists that 3D islands are the stable state whereas NWs are a metastable state at 700°C.

When the sample is heated, the silicon terraces have the same contours up to ~620°C.

Silicon terraces start breaking up and the terrace contrast weakens at ~700°C as the

89

nucleation of islands starts. The island growth starts and ceases in the same time
regardless of the island size. Larger 3D islands grow at a faster growth rate than smaller

islands.

90

Chapter 6 Transport Measurements

Deposition of a RE metal on the Si(001) surface at an elevated temperature results in
the formation of silicide islands and W3 coexisting with a reconstructed substrate
surface. Although it is important to understand the electrical properties of RE silicide
NWs and islands in air for practical applications, no one so far has done the transport
measurements of ultra high vacuum (UHV)—prepared samples in air.

We describe our approach for making transport measurements on nanometer scale
thickness ﬁlms grown on atomically clean silicon in UHV. A combination of ex-situ Ti
silicide and in-situ Au deposition is used for contacts. Samples are passivated with
amorphous germanium (Ge) before being removed from UHV for transport
measurements at 4.2 K. The transport properties are correlated with ﬁlm morphology as
observed by STM.

In this chapter, we will present the results of ongoing measurements on macrosc0pic

networks of the Dy silicide islands.

6.1 Sample preparation

The procedure needs several steps for the sample preparation as shown in Figure 6.1.
After a sample is chemically cleaned (Step 1), Si and Ta (or Ti) depositions are carried
out in a dc-magnetron triode sputtering system with a base pressure of 3x10’8 Torr (Step
2). The sputtering Ar pressure is 2X10'3 Torr. Ar and Ta (or Ti) purities are 99.999 %
and 99.99 %, respectively. The sample with pre-deposited Ta (or Ti) contacts is then
transferred into the UHV chamber and ﬂashed brieﬂy at 1175°C to remove surface oxide

(Step 4). At the same time, Ta (or Ti) forms silicides which serve as contacts for

91

transport measurements (Step 5). After a RE silicide thin film is made by the same
experimental procedure as Section 2.1 (Step 6), gold (Au) is deposited through a shadow
mask on top of TaSiz (or TiSiz) (Step 7). Since Au partially covers the RE silicide thin
ﬁlm, there is a direct connection between Au contacts and the RE silicide thin ﬁlm. In
the end, the sample is passivated with amorphous Ge at RT to avoid surface oxidation

(Step 8), and then removed from UHV chamber for transport measurements.

Ta or Ti Oxidizaiion
- Silt. - s -
—
Si subsira-le Flashing!

 

.Nm.m_ﬁ-wa___..u

 

l

i 5 b '7 8 Passivafion
i TaSiz or Tifiz Nanowires "
l
g

:drﬁ .1‘1;.‘-:~’Ly,’:‘, 3:: , . .r,

   

Figure 6.1 Sample preparation procedure for transport measurements. Si and Ta (or
Ti) depositions are done in the high vacuum chamber (Step 1 and 2). Sample ﬂashing,
deposition of RE metals, Au contacts, and amorphous Ge are performed in the UHV
chamber (Step 4~8).

6.1.1 Pre-deposited Ta (or Ti) silicide contacts
Refractory metal disilicides have been of special interest due to their temperature
stability and relatively low resistivity [155-165]. TaSiz and TiSiz are the candidates for

pre-growth deposited contacts for transport measurements, since these silicides can

92

survive the high temperature ﬂash necessary to clean the Si(001) surface [166]. The
contact resistances are in the range of 6.8~52 S2 for TaSiz and 5~15 $2 for TiSiz at 4.2 K
[155].

TaSiz is a hexagonal structure (C40) with a c/a ratio of about 1.38 [155]. The sheet
resistance of the TaSiz thin ﬁlm decreases with an increase in ﬁlm thickness and
annealing temperature [156, 157]. On the other hand, TiSiz exists in two phases: a
metastable C49 base-centered orthorhombic phase that is formed at lower temperature
and the stable face-centered orthorhombic C54 phase formed at higher temperature [158].
The C54 phase is more desirable for device applications because the C54 phase has lower
resistivity than the C49 phase. But the resistivity of TiSiz increases due to the failure of
the C49 to C54 phase conversion as the lateral dimension of TiSi2 shrinks to less than 1
urn [159]. However, the type of TiSiz phases will not affect our macroscopic
measurements since our contact dimension and thickness are not in the submicron range.

Si diffuses into Ti layers from 500°C and then TiSi2 is formed at 600°C. TaSiz is also
formed from 600°C [160], but the crystallization of TaSiz occurs mainly between 800°C
and 900°C [157]. The initial attempts at making Ta silicide contacts involved depositing
just Ta onto the substrate, and then annealing in UHV. As shown in Figure 6.2, when
the sample with a pre-deposited Ta is annealed, a deep trench is created on the boundary
between TaSiz and Si surface because Si around Ta is consumed to form TaSi2_ The
same phenomenon can be seen for TiSiz. Since these trenches might interfere with
connections to a RE thin ﬁlm to TaSiz contacts, the deposition of Si with Ta is necessary
beforehand as shown in Figure 6.1. The ratio of metal to deposited Si thickness is

122.23 (1:227) for Ta (Ti) in order for the stoichiometry of the deposited material to be

93

MSiz. When this is done, the trenches around the contacts are largely gone, and also the
rough boundary region around the contacts is narrower. Figures 6.3 (a) and 6.3 (b) show
the surface morphology of TaSiz and TiSiz films, respectively. Although the ﬁlms do not

appear smooth, they are continuous.

 

 

 

Figure 6.2 (a) Atomic Force Microscopy (AFM) image of the boundary between TaSiz
and Si substrate after annealing 100 nm thick Ta on silicon. (b) Cross-sectional proﬁle
traced along a white line in (a) showing a deep trench. (Width = 23 um, depth = 800
nm)

 

Figure 6.3 Scanning Electron Microscopy (SEM) images of (a) TaSiz and (b) TiSiz
ﬁlms. For (a), 200 nm of Si + 100 nm of Ta ﬁlms are annealed after deposition. For
(b), 100 nm of Si + 50 nm of Ti ﬁlms are annealed after deposition.

94

For the deposition of Si and Ta (or Ti) in Step 2 in Figure 6.1, sputtering masks are
used as shown in Figure 6.4 (a). Three samples are ﬁtted on one mask. A rectangle in
Figure 6.4 (a) indicates the position of the sample sitting in the middle of the sputtering
mask. Other two samples can be put on the top and bottom of the middle sample.
After Si and Ta (or Ti) deposition, four contacts are made on each sample as shown in
Figure 6.5 (a). Since the maximum of 24 samples can be loaded on the sputtering
system, the possibility of making many pre-deposited contacts at once offers a great

advantage.

  

ll

sample size = 1.23m

Figure 6.4 (a) Sputtering mask for Ta (or Ti) deposition. (b) Shadow mask for Au
deposition.
6.1.2 In-situ Au contacts

After the RE silicide growth, Au contacts are deposited at RT by covering a sample
with the middle of a shadow mask as shown in Figure 6.4 (b) (Step 7 in Figure 6.1).
About 50 ML (~6 nm) of Au is deposited on top of TaSiz (or TiSiz) contacts [Figure 6.5
(a)]. The Au thin-film contacts are composed of many small islands which are
connected to each other [Figure 6.5 (c)]. Figure 6.5 (b) shows a magnified view of the
edge of the Ti silicide contact overlaid with Au. As mentioned previously, the
codeposition of Si and Ti before annealing eliminates any trenching around the silicide.

At the same time, it is clear that the edges of the Au contact are not sharp. This is

95

because of a combination of the distance between the shadow mask and the sample
(which is about 0.5 mm), the distance between the shadow mask and Au evaporation
source (about 5 mm), the dimension of the shadow mask (2.5 mm), and the finite size of
the Au evaporation source (about 2 mm diameter). Given the geometry of the source
relative to the mask and the sample, one expects that the boundary of the Au deposition
would be spread over about 0.1 mm as shown in Figure 6.6.

Figures 6.7 (a)~(e) show the STM images of the boundaries on the Au contact. The
distance between the areas shown in these images is about 1 um. The density of Au
increases from image (a) to (0), meaning that the edge of the Au boundary is not

microscopically sharp. DySiz NWs can be also seen underneath the Au.

1‘ ”f“ i ‘ )1 __

\u - lilrrm

'l‘iSi:

 

Figure 6.5 (a), (b) Optical microscope images of Au contacts covering beneath TiSiz
contacts. (b) Close-up image of the area marked with a white rectangle in (a). (c) 3D
STM image of the Au contact. The formation of many small Au islands is observed.

96

  

sample
0.1 mm 25 um

Figure 6.6 Schematic diagram of Au evaporation with a shadow mask.

 

Figure 6.7 STM images of the Au boundaries (200x200 nmz). The density of Au
increases from (a) to (c). DySiz NWs can be seen underneath of Au.

6.1.3 Amorphous Ge passivation

Since RE elements and RE silicides are very reactive with oxygen, it is important to
protect the sample surface from oxidation in air. The materials used for surface
passivation should meet the following two conditions. Firstly, it must not affect the
conductance of the sample. Hence, it should be nonconductive at 4.2 K at which

transport measurements are performed. Secondly, it should not be reactive with RE

97

silicide and Si at RT so that surface morphology underneath does not change. Although
we first used hydrogen for the surface passivation for historical reasons (see Appendix A),
Ge seems to work better in terms of more reliable transport measurements.

Figures 6.8 (a) and 6.8 (b) show the STM images of ‘before’ and ‘after’ 5 ML of Ge
deposition on DySiz W5 and interconnected 3D islands, respectively. NWs still

survive underneath the amorphous Ge in Figure 6.8 (b).

 

Figure 6.8 (a) Before Ge deposition: DySiz NWs with interconnected 3D islands.
(180x180 nmz) Coverage is unknown. (b) After 5 ML of Ge deposition on top of
NWs. (200x200 nm2)

6.2 'II-ansport‘ measurements: Results and discussion

All transport measurements are done at liquid helium temperature (4.2 K) as shown in
Figure 6.9 (b). Four-terminal surface resistances are measured by a Keithley digital
multimeter. Note that four-terminal measurements can exclude the contact resistance
and provide only sample resistance (see Appendix C). From the two-terminal

measurements, contact resistances are a few 9 at RT and >1 M9 at 4.2 K.

98

(b)

 

 

 

 

 

   

 

 

 

Liquid He

Figure 6.9 (a) The STM image of 3 ML of interconnected Dy silicide islands. (b)
Experimental setup for transport measurements. Surface resistances were measured at
liquid helium temperature (4.2 K). (c) In order to measure the change in surface
resistance of the sample (a), the sample was scratched in the horizontal and vertical
directions. The numbers labeled on the bottom 0, l, and 2 indicate ‘before scratch’,
‘scratched in the horizontal direction’, and ‘scratched in the both directions’, respectively.

In the ﬁrst part of the experiments, changes in surface resistance due to sample
scratching was investigated. The surface topography of the sample with 3 ML of Dy is
shown in Figure 6.9 (a). The surface is covered with interconnected Dy silicide islands.
The surface resistances were measured in three steps: without scratch (Step 0), with a
scratch in the horizontal direction (Step 1), and both directions (Step 2). The numbers

labeled in Figures 6.9 (c) and 6.10, indicate ‘before scratch (0)’, ‘scratched in the

99

 

mag , -

horizontal direction (1)’, and ‘scratched in the both directions (2)’. Scratches were
drawn with a diamond tipped pencil. AFM images show that the width and depth of the
scratches are ~6i2 um and ~200i100 nm, respectively.

Figure 6.10 (a) shows the change in four-terminal surface resistance of interconnected
Dy silicide islands. The left (right) diagram represents the surface resistance in the
vertical (horizontal) direction. The resistance values are averaged between left and right
(top and bottom) sides of contacts for the vertical (horizontal) direction, and the
resistances in three steps of sample scratching are marked with open (solid) circles. As
a reference, the surface resistance of clean Si surface is indicated with a dashed line on
top of each diagram. Corresponding current ﬂows are schematically diagramed in

Figure 6.10 (b).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

(a) 4-terminal Measurements (b)
3 8 Rvertical Rhorizontal
7 7
6 6 6.46

E a

s 5 E 5

§ 4 g 4

f u

3 3 E 3

m 2 m 2
1 1 1.2x1o-4 6.6x1o-4
o o e L
0 1 2 0 1 2

 

 

 

 

 

Figure 6.10 (a) Change in 4-terminal surface resistance of interconnected Dy silicide
islands due to the sample scratching. A diagram shows the resistances in the vertical
(open circles) and horizontal (solid circles) directions, respectively. Dashed lines on top
are the surface resistance of clean Si surface. (b) Schematic diagrams of each step of
sample scratching and corresponding current ﬂow.

100

Before the sample scratching (Step 0), surface resistances in both vertical and
horizontal directions are ~1.2x102 52. For this particular sample, the distance between
two contacts in the vertical (horizontal) direction is 1.45 mm (1.75 mm). In case of the
surface resistance in the vertical direction (open circles), the resistance increases
dramatically to 7.2 MQ after scratching in the horizontal direction (Step 1), because there
is a large barrier between contacts and the current cannot pass though easily [Rvertical ‘1’ in
Figure 6.10 (b)]. After scratching in the vertical direction (Step 2), the resistance drops,
because the vertical barrier tends to conﬁne the ﬂowing current to the left side [vamcal ‘2’
in Figure 6.10 (b)] thereby decreasing the voltage drop measured across the right hand
side contacts.

In the similar way, the change in the surface resistance in the horizontal direction can
be explained. After scratching in the horizontal direction (Step 1), the resistance slightly
increases, but the resistance value remains the same order of magnitude as that before
scratching. After scratching in the vertical direction (Step 2), the resistance jumps up to
6.4 M52 due to the barrier between the contacts. It should be also mentioned that the
surface resistance values of interconnected Dy silicide islands are smaller than that of
clean Si surface which is drawn by dashed lines in Figure 6.10 even after scratching the
sample in both directions. This is a reasonable result because there is small electrical
contribution to the surface conductivity from Dy metals on Si substrate, although there
are deep boundaries in the middle of the sample. In the conclusion for the ﬁrst part of

our experiments, the destruction of interconnected Dy silicide islands increases the

surface resistance.

101

,3.)

 

450x450 nm2 . ,5 180x180nm2

Figure 6.11 STM images of (a) DySiz NWs at 0.22 ML and (b) interconnected Dy
silicide islands at 3 ML.

The second part of our transport measurements is the comparison of surface
conductivity at different metal coverages. Figure 6.11 shows the STM images of (a)
DySiz NWs (450x450 nmz) and (b) interconnected Dy silicide islands (180x180 nmz).
A few NWs are seen with 2x4 reconstructed surface at 0.22 ML [Figure 6.11 (a)], while
the varying shape of 3D islands are connected and bare Si is not visible on the surface at
3 ML (=1 nm) [Figure 6.11 (b)].

Figure 6.12 shows Dy and Ho ﬁlm conductivity at 4.2 K vs. metal coverages. Solid
(open) circles show the surface conductivity of Dy (Ho) on Si(001). The conductivity
of Dy thick ﬁlm with ~180 ML (60 nm) is added as a reference [68]. The values of
conductivity are indicated in Table 6.1. A 3 ML (zl nm) thick ﬁlm of interconnected
DySiz islands shows surface conductivity at 4.2 K similar to the reported bulk silicide
value. A sample with sparse and disconnected NWs (0.16 ML) shows lower surface

conductivity than the 3 ML ﬁlm but higher than that of the clean Si surface, because the

102

2D surface is reconstructed with Dy metal. The conductivity of a sample with a DySiz
NW network (0.87 ML) is lower than that of the sample at lower coverage. One
possible explanation for lower surface conductivity on the high-coverage sample is the
contact type: instead of pre-deposited Ti and in-situ Au contacts, ex-situ Al contacts were
used (see sample #12 on Table 6.2). For surface conductivities of Ho on Si(001) with
ex-situ A1 contacts, a sample with a HoSiz NW network shows higher conductivity than
that with 2x4 reconstructed surface. Overall, the surface conductivity rises sharply
somewhere in the coverage range of 1~3’ NIL.

Many variations in sample preparation were made to improve the reproducibility:
changing the choice of contact materials, the thickness of the contacts, and sample
cleaning procedures. The methods of sample preparation and the results of transport
measurements of each sample are summarized in Table 6.2. The samples from which at
least some resistance values could be obtained at 4.2 K, regardless of the adequacy of the
values, are gray shaded on the background of the leftmost column. There is no
consistency among the methods of sample preparation and the results of measurements.
Although both two- and four-terminal surface resistances are all measurable at RT,
low-temperature measurements lose the stability due to the background noise.

Unfortunately, the reliability of this experiments were not achieved from three-year
efforts. Besides instability of low-temperature measurements, the limited amount of
metal source and the multiple failures of the in-situ Au evaporator (Section 6.2.2) made

the measurements very difﬁcult. This work is supported by NSF ECS-0303801.

103

 

 

 

 

 

 

 

 

 

 

105 z’cL
.... 104 O 3‘ .[68]
E 3ML“ 60nm
o 103 8%—
C;
E: 102 8%
IE 101_¢0.16ML n
8 (C
E 100 90.46 ML (,(L
5 10-100.18ML n
00.87 ML “
10'2 1 r r ()(l L
0 l 2 3 180
Metal coverage (ML)

Figure 6.12 Dy and Ho ﬁlm conductivity at 4.2 K vs. metal coverages. Solid (open)
circles show the surface conductivity of Dy (Ho) on Si(001).

 

 

 

 

 

 

 

 

At 4.2 K Metal layer Surface topography Conductivity Sample #
coverage (ML) (9 cm)’1 on Table 6.2

Dy 0.16 2x4 1.0x10l # 20

Dy 0.87 NW networks 3.9x10'2 # 12

Dy 3 Interconnected islands 1.9x104 # l6

Dy 180 (60 nm) Thin film 2.9x104 Ref [68]
Ho 0.18 2x4 3.7x10" # 9

Ho 0.46 NW networks 1.3 # 6

 

 

 

 

 

 

Table 6.1 Dy and Ho film conductivity at 4.2 K vs. metal coverages. All the values
correspond to those in Figure 6.12. There are no values for the conductivity of bulk Ho
silicide in the literature, but the conductivity of the RE silicides lie in the range 4.5><lO3 to
6.7x10S (gr-cm)" at 4.2 K as in Table 1.4.

104

03:80.20: 3:3:8é 9:0 -N ”:3 0%

50:88:05 2003: 7:93. wnd 2 8083:: 093i .3 536200: 2E.

A> o: > N 3 0m83> 8:30 5:5 53:28: 00850 ”90: >~Q

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

03805 20:22 Om 32 dev 2Q CR :0 0H E: om + mm E: 83 050:. 3:20:0805 a m _
0380:: 032 32 meg >9 CM 80 0H E: mm + «m E: om: 050p. 082800-08 a E
a so: u .2. 032 02858:: 32 one 5 ES 40 :o: 0002. a 028:5 Ea 02V 7.. m 2
E93 3 n >SQ E: mmd m 000563: >92 .
Eod cm H >~Q VXN + 032 32 $9 29 :E On :8 000910 “020050 AE: 38 _< m .2
0380:: va 32 $9 on :E cm :8 Doomv :0 0200::0 AE: ©va 10. 52 -Rm a . 3
0302:: VXN + 032 38 0 32 mmdv 03 43200 to 0200: 2 w< 55 808:8 w< 5.2-5 A 9
E93 0.3 u >SQ E: 2.0 m va 3 00053038
Eve :0 u :5 4x0 32 2.8 on 5:: 00 a: 0002. a 028:5 :5: 02V 2 5:40 a 0
E93 m.SXm.m n A.8058; 0:: E08 00:50 35:03:09
90 0Q 5553: 32 5.8 2Q :E mm :8 Doomv :0 32055 AE: 83 _< 52.35 3:96 No.8 2 m
0380:: 32 02 Jim :2 8.8 oz EE mm :8 Uooﬂq 00 020050 AE: as :0. 5.2.06 m 0
Sea 8.: u >05 E: mmd m 30533: 32
80d 8.: u :5 8:950: 32 32 84.8 on :E mm :8 0°09. :0 00.0050 AE: SS 2 52-2% m o,
I 032 “00:00:55: 32 3.8 cm E96 8.8 = m
8.53:0: 32 32 00.9 on Eye 58 a 0
l 20:82 0:0 032 00:09: 32 $9 on 3:93 8.8 = m
0380:: :03E030 >15 E 00:00: at .0303 E _< m N .—
2 S. a E: :0 8
0505050002 85589 00825 099 88:00 0&0 30:35

 

 

 

 

 

 

.33 030 H 6:0 2.0 05wE E 8:800:00 0:0 00350 8:05 05 :8 80D 50526:: 00352 8:80:08 .3 50855 N6 0308

105

 

E: 83 3. 0:: E: 0 :- 53: +

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

53:: 083:0: 32 :52 T: E 60:: a a. E: m: + 2 E: 0:: N22 80:20:00-2: E90 :20 E 00
E: 8023 :2 5.0.0... +32 ... : 0 =< 4.5-E +
| SE 0.2 00 + 3:2 000 :0 609. a 2 E: m: + 2 E: :3 22 80808-2: : :0
£2 a 3. 5.0-0.. +
bEcE 32 3.0 00 609. a F E: m: + 2 E: 0:: N22 0008:0002 0: mm
E90: 2 u 0 32 21:0 3. 5.2-E +
a: 8012 u ...: SE 0.40 00 609. :0 F E: m: + 2 E: 20 ~22 02:88-2: .: 00
SE 32 8on 3. 5.2.5 + 608:.
0300:: 3.0 00 + 0:850: 32 32 t :0 a 2 E: 0: + 2 E: 8: N22 00:82:00-2: : :0
50:26.: M Q E: :0 m 0x0 .6 20:02:: 32 we 3. 5.2.5 + 6000'
0.2 20.: u ...: SE 3.0 00 + vxm 32 2.8 :0 a F E: m: + 2 E: 2: 22 09088.8: a :0
32 :2va :< 5.2.3 + A008:
58:: 32 0.2 00 + 32 0.8 :0 :0 F E: on + 2 E: :0 N2:. 00:00:00-2: : 2
32 20$ 3. 5.0-0.: + 6000.
2:32: 32 0.8 00 a 2 E: m: + 2 E: 80 ~22 00:00:00-8: : 2
E90: 8.: u 0 3:2 :2 :< :02: + ,,
02 8.: n 5: :53 600' :0 a. E: 0: + 2 E: 0:: N2:. 02:88-2: : 2
E901 2 n 0 E: : m 20:00:: :23 32 0.0 00 32 oavv 3. 5.2.3 + .
0 0: n ...: + 5:02 028::85E a 02 o :0 6.01:0 F E: m: + 2 E: :5 ~22 02:88-2: : . 2
M N4: :0 099 _
255050003 53050.80: 008-5m 0%: 88:00 80:23;

 

3.308 N6 0309

 

Chapter 7 Future work

In this chapter, suggestions for future work as a continuation of this thesis will be
described: growth kinetics of RE silicide NWs and 3D islands, reproducibility of
electrical contacts for transport measurements, single NW measurements, and
temperature dependence and magnetic-field dependence of magnetic, structural and

transport properties.

7.1 Further work on growth kinetics

In Chapter 5, we reported the LEEM observation of the evolution of Dy silicide islands
during annealing. Our original purpose was to investigate the growth kinetics of RE
silicide NWs such as NW nucleation, size and shape evolution, and Si step motion.
Unfortunately, the width of RE silicide NWs (the width of single NW is typically < 5 nm)
was below the lateral resolution limit of IBM Type I LEEM. The investigation of
growth dynamics of RE silicide NWs, therefore, requires high-resolution LEEM/PEEM,
or other alternative real-time surface analysis methods.

Recently, a few studies of the growth evolution of elongated Er silicide islands and Ti
silicide NWs have been done by photoemission electron microscope (PEEM) or LEEM
[76, 167, 168]. Since their dimensions are about 50 nm wide and several um long,
which is one order of magnitude larger than those of RE silicide NWs, they are resolvable
in LEEM/PEEM.

Moreover, it is important to interpret a correlation of the growth behavior of the width
and length (lateral directions) and the height (vertical direction) of NWs and 3D islands

so that the surface diffusion mechanism, surface stress and strain relaxation can be better

107

understood. Combining these results with TEM data would provide stronger support,
since TEM data offers information about the crystal structure (hexagonal, tetragonal, or
orthorhombic structure) of the 3D islands and about the interface defects between the 3D

islands and the Si substrate [154].

7.2 Reliability and reproducibility of electrical contacts for transport
measurements

Normally, Si substrates with a resistivity of p ~2.5 Q-cm at RT are used for our
transport measurements so that the substrates do not contribute to the sample resistance in
measurements at 4.2 K [169]. When the resistance of a highly-doped Si substrate which
has low resistivity at 4.2 K (p ~10’2 Q-cm at RT) was measured, resistance values were
very stable and did not fluctuate during the measurements. Therefore, one can assume
that the Si substrate and contacts, and the contacts and leads are well contacted.

However, the reproducibility of our measurements on RE silicides on Si substrates is
rather poor as mentioned in Section 6.3. This implies that there is a possibility of an
unstable connection between the RE silicide thin ﬁlm (or NW network) and the contacts.
To improve this situation, deposition of thicker Au layer (more than 10 nm) will be
helpful in Step 7 in Figure 6.1. Additional Au deposition after Ge passivation (Step 8 in
Figure 6.1) will be worth trying as well.

Ex-situ deposited contacts (e.g. evaporation of Snm of Ti and 200 nm of Au in the
clean room) will be an alternate method, although this has the disadvantage that samples
are no longer totally UHV-prepared. Electron-beam (e-beam) lithography is another
alternative method for contact deposition. In this case, very ﬁne electrons can be

defined so that measurements can be done on much smaller scales.

108

 

The use of intrinsic Si wafers allows transport measurements at RT, because the
intrinsic resistivity of Si (05, ~2.4x105 [Q-cm] at RT) is nine orders of magnitude larger
than that of RE silicide thin ﬁlms (plasma-dc ~104 [Q-cm] at RT) [170]. However, STM

on such high resistivity wafers is problematic.

7.3 Single NW measurements

Single NW measurements are even more challenging, because the measurement
dimension is at the nanoscale. Determination of the precise positioning of contacts on a
NW is a key for this measurement. Figure 7.1 shows a schematic diagram of post
growth deposited Au contacts for single NW measurements. In Step 1, a sparse network
of NWs are grown in UHV chamber. Ex-situ alignment marks are fabricated by e-beam
lithography (Step 2). After the registration of NWs and marks is determined by AFM,
appropriately conﬁgured ﬁne Au contacts (~50 nm) will be deposited by e-beam
lithography. When sparse, parallel NWs are formed on a vicinal Si substrate [7],
deposition of four contacts on a single NW would be very possible, because the length of

these NWs can be microns in length.

   

9‘2: ~':.-‘.
=“ "t i. -..
c K 1 ‘ _ ~
~:j {w ..
w res ......

Figure 7.1 Post growth deposited contacts Step I: Grow sparse network of NWs.
Step 2: Deposit alignment marks: Use AFM to determine the NW/mark registration.
Step 3: Deposit ﬁne Au contacts.

109

7.4 Temperature dependence and magnetic-ﬁeld dependence of magnetic,
structural and transport properties of low-dimensional RE silicides

The investigation of magnetic properties of RE disilicide NWs and 3D islands at low
temperature or in high magnetic ﬁeld would be interesting topics, since the magnetism of
one-dimensional (1D) systems is different from that of 2D or 3D systems. In case of Co
atomic chains on a Pt substrate, long-range ferromagnetic order is stabilized by large
anisotropy energy barriers which arise from large localized orbital moments [171, 172].

RE elements and bulk RE silicides have a variety of unusual magnetic properties. For
example, several magnetic structures can be observed in heavy RE metals below RT:
sinusoidal structure [Figure 7.2 (b)], circular cone structure [Figure 7.2 (d)], spiral
antiferromagnetic structure [Figure 7.2 (c)], and ferromagnetic phase in the basal plane
[Figure 7.2 (f)] [42]. Intermediate mixed phases, where the c axis component is
modulated along the c axis (CAM structure) are often seen [Figures 7.2 (a)~(c)]. The
same metal can have several of these structures at different temperatures.

Since the 4f shell, which is responsible for almost all of the magnetism of RE ions, is
localized deep inside the atoms and shielded by the ﬁlled outer 5s and 5p shells, it is not
perturbed by the crystal ﬁeld. Therefore, spin-orbit coupling (LS-coupling) dominantly
contributes to the magnetic moment. Experimental results are fairly consistent with
theoretical calculations as shown in Table 7.1. Although Sc, Y, La, Yb, and Lu ions are
diamagnetic because of the absence of a partially ﬁlled 4f shell (no unpaired electrons),

they generally show paramagnetic behavior.

110

 

m»
a a «9 @-

9 I,
I
’l

I
—-‘~—-—-—‘-*

a 6m

\

.1

0&000

“a.
a @— (9+9)

000000000

000000000
000000000

0

036-6}
6:?)

Br, Tm Er Ho, Er Tb, Dy, Ho Gd, Tb, Dy
(3) (b) (C) (d) (e) (0

Figure 7.2 Magnetic structures of heavy RE metals [after Koehler (1972)] [42].

The magnetic properties of RE disilicides are also summarized in Table 7.2. There is
a large variety of ordering: no magnetic ordering in CeSiz, ferromagnetism in light RE
silicides, PrSiz and NdSiz, antiferromagnetism in heavy RE silicides, GdSiz, TbSiz, DySiz,
HoSiz, and ErSiz. Pierre et al. mentioned that these behaviors might be explained by the
magnetocrystalline anisotropy, which is governed by the three different crystallographic
structures (see Figure 1.1) in the series [173].

Before then, Sekizawa et al. described that the effective radial extent of 4f

wavefunction of RE metal may inﬂuence the sign of the exchange interaction J (J>O for

111

ferromagnetic, and J<O for antiferromagnetic) [174]. The theoretical calculation
indicates that RE silicide compounds are ferromagnetic when the ratios of the radius of 4f
wavefunction to the half of the nearest neighbor distance of the RE ions in the silicide
compounds, r4drN-N is larger than O.23~O.25 [174]. On the other hand, RE silicides are
ferromagnetic when r4f/rN-N < O.23~O.25. This result is consistent with the fact that the
radial extent of 4f wavefunction of light RES are much larger than that of heavy REs.

For transport properties of RE silicides, there is an abrupt decrease of resistivity at very
low temperatures due to the antiferromagnetic ordering arising from the incomplete 4f

shell of the RE ion as described in Section 1.1.2.4.

112

.uuocwaEEﬂ ”.Eom "ozocwaﬁobotﬁa “ma: motor—@2885 “m

.282 053: 282 Ezomaxun: .23 995$ 05 E :2 55 80:: he m+.._ u .- 22 :2: :25 222 .82 _m-A_ n a 20:?
525808 Enwca :39 a 2 m $338222 .ScoEoE “ﬂaw—8 .358 28 can :38 22 A E2 m 3722 ”822 9580
533822 6 A. m can u \\ E 28 2023222an2 263-0220 ococwaﬁaha ”c 5 “Eu 36 ”838388 2.50 “oh
"champagne“ Bmz ”2,—- uouoﬁ M 2284 ”m AFM :3 :88:me Eom Mo 23:3: 0380th ”SR—+32%

 

 

 

 

 

 

 

 

113

 

 

 

 

 

 

 

 

2.: 2.: SEN $2222» a“;
can :- :. Ea a 22 22 2.: 22.2 £2 32:22.2. aEE-
Eon 2.22 22 Ea s E 82 M: :2 2.2 22.: NE: KNEE. .mm

232:: 0:8 22 2: Ea 2 E 222 .2: 8.2 E... 322? so:
a 22 § Ea : 22 2: 2.2 22.2 Em: $2229 +25
2 222 22 Ea 2 82 222 2.2 23 on: 322%. +2.:
2 :2 S 2 Ea e E .22 222 l 2.2 as E2MW $22K: ace
eases. EE 2 22 A: _v | o: :2 o E: $222.2» :5
SEES | .3 2.2 :22 En: £22222» 22:2
2 I 22.2 :2 $222.9 .mEa
Ea : __ 22
. Ea : l .22 2.2 22.2 a: 12222:. +23
.. Ea 2 :22
Ea : I 22.2 2.2 22.2 .5 KNEE. a:
:2 2.2 22.2 Ema KNEE. .28
022.3 M N4. :2 Czoﬁcoaxov Abound 23m coca-Swmag :0—
922323 as a: . .2 0e :2 850:: 2am a: be as 2:- sSia 2:352 285 28:82 22

 

 

 

 

 

 

 

 

 

 

 

 

 

.2: .2: 3:520 mm 2o 22:an £5 E 2%..-

.ouoamanboEsa “n2

”ouocmaﬁobﬂ um

65.255055 “0 ”Rsowabo: F ”ﬁcowaxo: um

Hosanuocwaﬁ 2:853:82 ”a: PM :2 :8an 3:292: 338?: ”up: ":5 «2.3de .I. aﬁgm n coho—ES: Eom ”:1

6:380:88 5:25.: @520 ”sash

”BEEQQES 2283-250 ouocwmﬁﬁaa ”.5

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2:: ”3. S 6.2 S- o N2:2
C d: _ a v
2:: 22.222122.:ném< 2 3:52:2- o MW._2:z
C d: 2 H 2.2
:2: .22: a: 2.: a :1 I u é m 2 a 2 E 2 a 2- 0 E222
m< 2.2
:2: d: 22
2:: m 35 \\ E :2. a 32:2
cogmcwb mam-56.5 0:0:qu Uconuow h
8:: m : N25
2:: .2 : 22.2 : : N22
35 ﬂ: 2 H 2:
2:: Q 2.: a :3 u .3 a 2.2 1...: \\ E 2 H 2- o :o : 22::
E: ME 8: 9 :38 2:52: EEEE oz 3 .2 .: N2:0
2:: 25:22:82 :22 222- : N2:0
M mN 328 mucosa—523:2 :5
3.2: .M NV 9 :30: wESEo 25:92:: 02 «as
8:: M 2.: 2 :38 2:525 E222 oz :2.»
osslsblm
LEM 252:2 0:82:22 O: ath 2.3 Er: Cc .6 0E:2:wo=52bu 820:8

 

 

822.2% mm :0 20:20:22: caucus: N6 032:.

114

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

8:25.: 5598 one «E vacuum 3
2: :2 S m :7855
2: 6:: S Nae.
3K: "2 2 N9: mm- 0 Swan.
FE M cm H zHAH a SP5 owcﬂéonm 2 mm- 0 25:;
.8228: @590 caveman: 958m @—
5: ”2 mm 0 :53.
HE 2325: :82:on 5 ”2 mm $1sz I 2.25:3.
am: 2225: ER: :0 m< mm and NV- I mine.
5:255 @530 0:0:me vacuum mm
8:: ”2 am ”.95
2K: :2 cm 2: m? 0 £95
2:: :2 E o2 m:- o 9:25
~85:wa wists unocwaﬁ 988m 2: .9:
H 2: :2 3.9: : :__-:._2_5
8w: 222:2 22E: :0 n2 2: $6 mm- I :MmmnD
F3 5:: N H mm mm- m mocmnw
conic“: warmth: 0.53mi: 958m :4:
2:: :2 am :25
a: 6:: M 3 9 =38 252:6:5 :25
Usuoaum
hEm 222:2 user—mg): 91:th A5: :9: O: .6 UEQEMEE:EU moEoEm

 

 

 

 

 

 

 

 

95:03 NS 03$.

115

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

38 3:328:83: 032:0: 6::qu 0:855: 02 m nmmnﬂﬂ

2: 2.52: 32:2: 02 m :22»

2:: 2 S. :25:

2: 2 2::: m 28.:

2:: m 2.: 2:: 2.0- m :22
2: 6:: M S. 8 5:8 F520 0:2: :8 oz 2%

0:2: :89 05 co m A0 :- mv Em
22: :5 : :5 2:2: 2 2.: G \\ E 2.:2 m :25
8:25: @590 05:28 BEE :4
2202:“: mucous 0:229: 950:6 2m

2: 2 E: : amam

2:: 2 E 2.8 S- o :22:
2: 2:: ”2 273 :63:

E 2. 0+: 2%:

2: : 2- : 82o:
2: 4:: 2 7d: :25

2:: 2 2.2 M22 2- o :25

2 a E :2

22: 2.2 2 \\ 5 3 o 525

82: 228:: Ba: 5 2 : 22: m+ : 2:5

2: 2 : o 22::

0.533%
05: 2303.5: 05:22 0: sad- EE :0: 0: 6 0E:::mo=smbo 830:6

 

 

 

 

 

 

 

 

116

8.280 N: 03::

APPENDIX A Hydrogen-passivation of Si(001) surface and DySiz NW 8

A Si surface which is terminated by monolayers of hydrogen (H) protects against
surface oxidation when the sample is exposed to air. It has been reported that the
H—passivated Si(001) surface remains atomically pristine after 15 minutes ambient
exposure [183]. In addition, the oxidation is not observed on this surface by X-ray
photoelectron spectroscopy (XPS) after almost two days of ambient exposure. Our
interest in H adsorption was the potential to use it as a passivation against oxidation of
our NW samples in the course of doing the ex-situ transport measurements.

Interaction of atomic H with Si(001) surface was ﬁrst presented by T. Sakurai et al. in
1976 [184]. They found that a H—saturated Si(001) surface produces 2x1 monohydn'de

(M) and 1x1 dihydride (D) phases as shown in Figure A1. Later on, Y]. Chabal et al.

reported the presence of both M and D phases in the 3x1 structure [185].

H

L Y [001]

I .1 )4 x f:’ H I

s s c s e s s s s s _ [110]
(a)1x1 (b) 2x1 (0) 3x1 [“01

Figure Al Schematic of the various possible hydrogenated structures on Si(001)
surface.

On the bulk-terminated surface [Figure 2.8 (a)], each Si atom has two dangling bonds.
The 1x1 D phase is obtained when two H atoms are attached on each Si atom [Figure A1

(a)] adopting a bulklike 1x1 structure. A 2x1 reconstructed Si(001) surface has one

117

dangling bond per Si atom. A 2x1 M phase is obtained when one H atom is attached to
the rest of dangling bond on each Si atom without changing the superstructure [Figure Al
(b)]. The ‘intermediate’ 3x1 phase consists of alternating M and D sites [Figure A1 (c)].
In considering the relationship between the coverage and the hydride phases, the 2x1 M
phase completely covers the surface at the coverage of 1.0 ML, since the ratio of H and
Si atoms on the surface is one to one. For 1x1 D phase, there are double the number of
H atoms on surface. Hence, the whole surface is covered with 1x1 D phase at the
coverage of 2.0 ML. The coverage associated with 3x1 phase is 4/3 (=1.33) ML
because four H atoms exist on every three Si atoms. Table A1 lists the H exposure
conditions needed to prepare different types of surfaces.

The STM work on the H-Si(001) system has studied subjects such as surface structure
[186-195], electronic structure [196-200], adsorption and desorption [106, 201-219],
dangling bond structures [220-223], and oxidation in air [183, 224-228]. In particular,
we can also use the STM tip to selectively desorb H from the surface to produce
nanoscale patterns.

To prepare the H-terminated Si(001) surface, the sample is first cleaned in the same
way as the experimental procedure described in Chapter 2. A 20-min atomic H
exposure with the chamber pressure of 4x10'8 Torr seems to be sufﬁcient to saturate the
whole surface. The sample surface is exposed to atomic H produced by dissociating Hz
with a 1600°C W-ﬁlament placed 1 cm away from the sample. During the H exposure,
the sample is heated at 330°C [Figures A2 and A3]. For the sample with DySiz NW8

[Figure A4], H exposure is done at RT.

118

Table A1 Structure of the H-saturated Si(001) surface and H exposure conditions.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Surface structure H exposure conditions Ref
(substrate temperature)

Monohydride (M) 2x1 Moderate RT dose [187]
Saturation dose at 600~650 K [185, 187, 229]

Domains of 3x1 on the 2x1 phase 400~600 K [211]

(STM)

M+D 3x1 3801-20 K [185]

Dihydride (D) 1x1 Large dose at RT [185]

2x1—)3x1—>1xl (LEED) Additional dose at 400 K [230]

3x1-92x1 (LEED) A mild annealing at 575 K [185]

2x1—91x1—)2xl (LEED) Additional dose at RT, [185]
then annealing at 600 K

 

 

 

Figure A2 (a) shows the schematic diagram of a Si dimer ‘before’ and ‘after’
H-passivation. For the 2x1 M structure, one H atom can attach to each Si dangling bond
to ﬁll the empty states. Therefore, in STM images, the brightness of the maxima
‘before’ and ‘after’ H-passivation appears different. The empty- and ﬁlled-state images
of H-passivated Si(001) surface with lO-min H exposure are shown in Figures A2 (b) and
A2 (c), respectively. Small white spots are contamination coming from the H source.
In both images, the half left side is the depassivated area produced by STM-induced H
desorption. H atoms are desorbed by applying high tip voltage. Here, a sample bias
voltage of +8 V is used. In empty states [Figure A2 (b)], the depassivated surface
appears brighter than the H-passivated surface because it corresponds to bare Si surface.
In ﬁlled states [Figure A2 (c)], the H-passivated surface appears brighter than the
depassivated surface because the electron density is higher when H atom is absorbed on

the dangling bond.

119

   

Depassivated surface Hydrogen-passivated
by high tip voltage Si(001) surface

 

 

IIIIIrI
'IIIIIIIIJ'

 

 

O 10 20 nm

 

lllllllllll
IIIIlIIllLJ

 

 

 

O 10 20 nm

 

Figure A2 (a) Schematic diagram of a Si dimer ‘before’ and ‘after’ H-passivation. A
half-ﬁlled 1t. anti-bonding orbital is completely ﬁlled due to H-desorption. (b)
Empty-state (1.74 V) and (c) filled-state (-l.82 V) images of H-passivated Si(001) surface.
The left half of the images is the surface after it was passivated by applying high tip
voltage (20x30 nmz). Cross-sectional proﬁle in the horizontal direction is shown next to
each STM image.

This H desorption mechanism can be used for nanoscale patterning as shown in Figure

A3. By manipulating the tip position and the voltage, the letters “MSU” have been

120

patterned with a bias voltage of +8 V. The average linewidth is ~10 nm. It is known
that the desorption yield depends on the sample bias voltage [212]. A wider linewidth is

obtained as the sample bias voltage increases.

 

Figure A3 STM image (90x120 nmz) of a H-passivated Si(001) surface. Brighter
areas correspond to an STM-induced H desorption. By operating the tip position and
the voltage, we can write a “MSU” logo!

So far no study of a H-saturated RE silicide film on Si(001) substrate has been done,
although there are a few studies of H adsorption on Er silicide films on Si(lll) [231-237].
LEED studies show that at low H exposures on ErSi1_7(0001), H is absorbed in the same
manner as on Si(lll), conﬁrming that Er silicide is terminated by a buckled Si plane
without Si vacancies [231, 234]. Moreover, H is not only adsorbed on but also
absorbed in the layer. It is suggested that this form of chemisorbed H corresponds to a
saturation of the dangling bonds related to the Si vacancies in the silicide bulk.
H-passivation of both external and internal dangling bonds strongly stabilizes the films
[233].

Here, our attempt of H passivation of DySiz NWs on Si(001) substrate is briefly

reported. Figures A4 (a) and A4 (b) show the ‘before’ H-passivation of DySiz NWs on

121

Si(001) at 0.80 ML and ‘after’ H—passivation of the same surface, respectively.
Although there are small gaps on the NWs shown by white arrows [Figure A4 (b)], the
structures of NWs still survive after H-passivation. The change in the electrical
properties of NWs due to the small gaps is unknown. There are no changes in the
H-induced features on the NW s after an attempt to depassivate at +8V on the tip. This
fact is very different from the depassivated Si(001) surface as shown in Figures A2 (b)
and A2 (c). This implies that the bonding between the H atom and the RE silicide is

stronger than that between H and Si.

 

Figure A4 (a) Before H-passivation: DySiz NWs on Si(001) at 0.80 ML (180x180 nmz).
(b) After H-passivation: NWs still survive, although there are small gaps on the NWs
shown by white arrows (50x50 nmz). The change in the electrical properties of NWs
due to the small gaps is unknown.

122

APPENDIX B Sheet Resistance [238, 239]

The resistance R of a thin film with a length 1, width w, thickness t, and resistivity p
can be written as

Rzpist

L [Q] R 5
wt w

5

Nlb

[Q/square]

where R3 is the sheet resistance of a layer. Here, we assume that the current density is

uniform throughout the ﬁlm.

I Ni
it 1‘ all?

Figure B1 Schematic diagram of current passing through a thin ﬁlm.

 

t

 

 

 

Sheet resistance is a resistance of a film, material, or layer measured by a four-point
probe. The resistance represents the parallel resistance of an inﬁnite number of
inﬁnitely thin parallel sheets. Strictly speaking, the unit for sheet resistance is the ohm
(since W is dimensionless). However, to avoid confusion between R and Rs, sheet
resistance is expressed as ohms per square area which is equal to the average resistivity

of a layer divided by the thickness of the layer.

FYI

p : i : n; [9.cm]

0' net

 

where m, n, e, and r are the electron mass, electron density, electron charge, and

relaxation time, respectively.

123

f

APPENDIX C Two- and four-terminal measurements [240]

To measure a sample resistance R, the easiest way is to pass a constant current through
the sample as shown in Figure C1. A known voltage V0 is connected to a series circuit
consisting of a sample resistor R and a ballast resistor R3. R3 is chosen to be RB >> R so

that the current is I = Vo/(R+RB) ~ VOIRB, which is independent of change in R.

R3

Vo

Q

 

Figure Cl Constant-current circuit using a ballast resistor.

(a) Rs (b) RB

 

 

 

RC1
R

RC2

0— V0

 

 

 

 

 

 

 

Figure C2 (a) Circuit showing a two-terminal resistance measurement. (b) Equivalent
circuit for (a).

Figures C2 (a) and C2 (b) show a circuit for a two—terminal resistance measurement
and its equivalent circuit, respectively. When the sample is connected to the circuit as in

Figure C2 (a), the measured voltage drop V is V: 1(RC1 +R + RC2) as in Figure C2 (b),

124

where RC] and RC2 are the contact resistances. Therefore, we are not measuring the
sample resistance R itself but a sum of the sample resistance R and contact resistances
RC] and R02.

To eliminate the contact resistances from the total resistance, four-terminal connections
are used as shown in Figure C3 (3). In four-terminal measurement, current leads are
separated from voltage leads. Figure C3 (b) shows the equivalent circuit of Figure C3
(3). The current and voltage contacts are labeled as RC1 ~ RC4. Since the contact
resistances RC; and RC; << R3, the current I is still I = Vo/(RB + RC1 + R + RC2) ~ Vo/RB.
A sample voltage is measured between the other two contacts with a voltmeter which has
very large input impedance so that it draws very little current. Since no current will
flow through the contact resistances RC3 and RC4, the voltage drop across the sample can
be measured as V = IR.

Unless the sample resistance is comparable with the internal resistance of the voltmeter,

contact resistances RC3 and RC4 have no effect on the measurements.

(3) RB (b) RB RC1 RC3

 

 

 

 

 

ll
<‘.
0

II
7::

V0

 

 

 

 

 

 

 

 

-o
1

 

 

 

 

 

RC2 RC4

Figure C3 (a) Circuit showing a four-terminal resistance measurement. (b) Equivalent
circuit for (a).

125

APPENDIX D Van der Pauw’s method

The surface resistivity of a sample ,0 can be determined from four-terminal resistances
in the vertical and horizontal directions, Rvertical and Rhoﬁmm. For a ﬂat sample of
anisotropic material of arbitrary shape with four points, van der Pauw equation [241, 242]
is given by

exp “-ﬂd Rverticall/2 +6Xp "7&1 Rhorizonltzlrlz =1 (1)
(pxpy) (pxpy)

where pJr and py are principal values of the resistivity in the plane of the sample, and d is

1/2

the thickness of a sample (or thin ﬁlm). From equation (1), (,0x p y) is obtained.

 

If the sample is rectangular in shape and the four contacts are placed at the comers of

the sample, the second equation is derived as

 

(px /py)1/2 : _—?—ln tanh MRhOl’llonijlz (2)
a” 16(pxpy)

where a and b are the sides of the rectangle [243].

1/2 1/2
) )

, the individual values of the principal resistivity

From (,0pr and (px lpy

can be obtained:

1/2 1/2
) )

p. = (10.x)y
py =(10..py

Mex/py

)1/2 +(Px/Py

1/2
)

In our transport measurements (Section 6.3), we assumed px = py for simplicity.

Therefore the surface resistivity of the sample p is given by equation (1), which is

p E (pxpyf’z.

126

APPENDIX E List of publications

Nogami, J., B.Z. Liu, M.V. Katkov, and C. Ohbuchi, Self-Assembled rare earth
silicide nanowires on Si( 001 ). Phys. Rev. B, 2001. 63: p. 233305.

Ohbuchi, C. and J. Nogami, Holmium growth on Si( 001 ): surface reconstructions
and nanowire formation. Phys. Rev. B, 2002. 66: p. 165323.

Ohbuchi, C. and J. Nogami, Samarium induced surface reconstructions of Si(001)
(published)

Ohbuchi, C., W. Swiech, and J. Nogami, Topographic evolution of Dy silicide on
Si( 001 ).' Shape transitions from nanowires to 30 islands (in preparation)

127

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

BIBLIOGRAPHY

C. Preinesberger, S. Vandre, T. Kalka, and M. Dahne-Prietch, Formation of
dysprosium silicide wires on Si(001), J. Phys. D: Appl. Phys. 31, p.L43 (1998).

http://www. nytimes. com/libra ry/tech/99/II/biztech/articles/OI nano. html.

Y. Chen, D.A.A. Ohlberg, G Medeiros-Ribeiro, Y.A. Chang, and RS. Williams,
Self-assembled growth of epitaxial erbium disilicide nanowires, Appl. Phys. Lett.
76(26), 114004-4006 (2000).

M. Katkov and J. Nogami. Growth of Yttrium on Si(001). in Bulletin of the
American Physical Society, March meeting. 2002.

Y. Chen, D.A.A. Ohlberg, and RS. Williams, Nanowires of four epitaxial
hexagonal silicides grown on Si( 001 ), J Appl. Phys. 91(5), p.3213-3218 (2002).

R. Ragan, Y. Chen, D.A.A. Ohlberg, G Medeiros-Ribeiro, and RS. Williams,
Ordered arrays of rare-earth silicide nanowires on Si(001), J. Crystal Growth 251,
p.657 (2003).

B.Z. Liu and J. Nogami, Growth of parallel rare earth silicide nanowire arrays on
vicinal Si(001), Nanotechnology 14(8), p.873-877 (2003).

D. Lee and S. Kin, Formation of hexagonal Gd disilicide nanowires on Si(100),
Appl. Phys. Lett. 82(16), p.2619-2621 (2003).

J. Nogami, B.Z. Liu, M.V. Katkov, C. Ohbuchi, and N .0. Birge, Self-Assembled
rare earth silicide nanowires on Si(001), Phys. Rev. B 63, p.233305 (2001).

B.Z. Liu and J. Nogami, An STM Study of the Si(001) (2x4)-Dy Surface, Surf. Sci.
488, p.399-405 (2001).

C. Preinesberger, S.K. Becker, S. Vandre, T. Kalka, and M. Dahne, Structure of
DySi2 nanowires on SI(001), J Applied Physics 91(3), p.1695-1697 (2002).

B.Z. Liu and J. Nogami, An STM study of DySi2 nanowires growth on Si(001), J
Applied Physics 93(1), p.593-599 (2003).

J. Nogami, Growth and characterization of atomic and nanometer scale wires on
the silicon surface, Surface Review and Letters 7(5-6), p.555-560 (2000).

C. Ohbuchi and J. Nogami, Holmium growth on Si(001): surface reconstructions
and nanowire formation, Phys. Rev. B 66, p. 165323 (2002).

Y. Chen, D.A.A. Ohlberg, G. Medeiros-Ribeiro, Y.A. Chang, and RS. Williams,
Growth and evolution of epitaxial erbium disilicide nanowires on Si(001), Appl.
Phys. A 75, p.353-361 (2002).

128

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

J .S. Yang, Q. Cai, X.D. Wang, and R. Koch, Morphological evolution of erbium
disilicide nanowires on Si(001), Surf. Interface Anal. 36, p.104-108 (2004).

M. Kuzmin, P. Laukkanen, R.E. Perala, R.-L. Vaara, and 1.]. Vayrynen, Formation
of ytterbium silicide nanowires on Si(001), Appl. Surf. Sci. 222, p.394 (2004).

Z. He, D.J. Smith, and RA. Bennett, unpublished.

J.L. McChesney, A. Kirakosian, R. Bennewitz, J.N. Crain, J.-L. Lin, and F.J.
Himpsel, Gd disilicide nanowires attached to Si(III) steps, Nanotechnology 13,
p.545-547 (2002).

Z. He, M. Stevens, D]. Smith, and RA. Bennett, Dysprosium silicide nanowires
on Si(IIO), Appl. Phys. Lett. 83(25), [15292-5294 (2003).

K.A. Gschneidner and LR. Eying, Handbook on the Physics and Chemistry of
Rare Earths. Electronic structure of rare earth metals, ed. S.H. Liu. Vol. 1. 1978:
North-Holland Publishing Company.

A. Chatterjee and AK. Singh, Pressure-Induced Electronic Collapse and
Structural Changes in Rare-Earth Monochalcogenides, Phys. Rev. B 6(6),
p.2285-2291 (1972).

A. Stenborg, O. Bjorneholm, A. Nilsson, N. Martensson, J.N. Andersen, and C.
Wigren, Valence changes and core-level shifts of Sm adsorbed on Mo(110), Phys.
Rev. B 40(9). P.59l6-5923 (1989).

H. Norde, J .d.S. Pires, F. d'Heurle, F. Pesavento, S. Petersson, and RA. Tove, The
Schottky-barrier height of the contacts between some rare-earth metals (and
silicides) and p-type silicon, Appl. Phys. Lett. 38(11), p.865 (1981).

K.N. Tu, R.D. Thompson, and B.Y. Tsaur, Low Schottky barrier of rare-earth
silicide on n-Si, Appl. Phys. Lett. 38(8), p.626-628 (1981).

RD. Thompson, B.Y. Tsaur, and K.N. Tu, Contact reaction between Si and rare
earth metals, Appl. Phys. Lett. 38(7), p.535-537 (1981).

A. Franciosi, J.H. Weaver, P. Perfetti, A.D. Katnani, and G Margaritondo,

Samarium valence changes and reactive interdiﬁ‘usion at the Si(111)-Sm interface,
Solid State Communications 47(6), p.427 (1983).

G Rossi, J. Nogami, I. Lindau, L. Braicovich, I. Abbati, U.d. Pennino, and S.
Nannarone, First Spectroscopic Investigation of the Yb / Si Interface at Room
Temperature, Journal of Vacuum Science and Technology A 1(2), p.781 (1983).

L.D. Woolf, Transport and magnetic properties of layered rare earth disilicides,
Solid State Communications 47(7), p.519-523 (1983).

129

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

E. Houssay, A. Rouault, 0. Thomas, R. Madar, and J .P. Senateur, Metallurgical
Reinvestigation of Rare Earth Silicides, Appl. Surf. Sci. 38, p.156-161 (1989).

A. Iandelli and A. Palenzona, in: Handbook on the Physics and Cherrristry of Rare
Earths (North-Holland Publishing Company, 1978) vol.2 Alloys and intermetallics,
ch.l3.

G. Rossi, d and f metal interface formation on silicon, Surf. Sci. Rep. 7, p.1
(1987).

FR Netzer, Rare earth overlayers on silicon, J. Phys. Condens. Matter 7,
p.991-1022 (1995).

S. Auffret, J. Pierre, B. Lambert, J.L. Soubeyroux, and J.A. Chromoczek,
Crystallographic and magnetic-structures of Er3Si5, Physica B 163(3), p.271-280
(1990).

V.M. Koleshko, V.F. Belitsky, and A.A. Khodin, Thin-ﬁlms of rare-earth-metal
silicides, Thin Solid Films 141(2), p.277—285 (1986).

J .A. Knapp and ST. Picraux, Epitaxial growth of rare-earth silicides on (111) Si,
Appl. Phys. Lett. 48(7). [1466-468 (1986).

A. Iandelli, A. Palenzona, and GL. Olcese, Valence Fluctuations of Ytterbium in
Silicon-rich compounds, Journal of Less Common Metals 64, p.213-220 (1979).

0. Thomas, C.S. Petersson, and EM. Dheurle, The reaction of scandium thin-ﬁlms
with silicon — diffusion, nucleation, resistivities, Appl. Surf. Sci. 53, p.138 (1991).

R. Baptist, S. Ferrer, G Grenet, and RC. Poon, Surface Crystallography of YSi2-x
Films Epitaxially Grown on Si(III): An X-Ray Photoelectron Drﬁraction Study,
Phys. Rev. Lett 64 p.311-314 (1990).

M.-A.Nicolet and 8.8. Lau, Formation and characterization of transition-metal
silicides, VLSI electronics -Mictostructure Science Vol.6 Eds N.G Einspruch, GB.
Larrabee (1983), and references therein.

C.C. Hsu, Y.X. Wang, J. Hu, J. Ho, and J .J . Qian, The eﬁect of thermal treatment
on the thin-ﬁlm reaction of La, Ce, and Nd on silicon surfaces, J. Vac. Sci.
Technol. A 7(5). p.3016-3022 (1989).

K.A. Gschneidner and L.-R. Eyn'ng, Handbook on the Physics and Chemistry of
Rare Earths; and references therein. vol.1 Metals (1978) ; vol.2 Alloy and
Intermetallics (1979); and references therein: North-Holland Publishing Company.

130

 

 

 

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

[54]

[55]

A. Vantomme, M.F. Wu, S. Hogg, U. Wahl, W. Deweerd, H. Pattun, G Langouche,
S. J in, and H. Bender, Stabilization and phase transformation of hexagonal
rare-earth silicides on Si(III), Nuclear Instruments and Methods B 17, p.261-266
(1999).

ML. Huang, J .H. Yang, Y.A. Chang, R. Ragan, Y. Chen, D.A.A. Ohlberg, and RS.
Williams, Phase stabilities of ternary rare earth metal disilicides, Appl. Phys. A
78(1). [3.1 (2004).

S. Auffret, J. Pierre, B. Lambert-Andron, R. Madar, E. Houssay, D. Schmitt, and
E. Siaud, Magnetic properties versus crystal structure in heavy rare-earth
silicides RSi2-x, Physica B 173, p.265-276 (1991).

N. Frangis, GV. Tendeloo, J .V. Landuyt, G Kaltsas, A. Travlos, and AG
Nassiopoulos, New Erbium Silicide Superstructures: A Study by High Resolution
Electron Microscopy, Phys. Stat. Sol. 158, p.107-116 (1996).

J .E. Baglin, FM. d'Heurle, and CS. Petersson, Appl. Phys. Lett. 36(7), p.594-596
(1980).

I. Gerocs, G Molnar, E. Jaroli, E. Zsoldos, G Peto, J. Gyulai, and E. Bugiel,

Epitaxy of orthorhombic gadolinium disilicide on <100> silicon, Appl. Phys. Lett.

52(25). p.2144-2145 (1987).

J .Y. Duboz, P.A. Badoz, F.A. Davitaya, and J .A. Chroboczek, Electronic
transport-properties of epitaxial erbium silicide silicon heterostructures, Appl.
Phys. Lett. 55(1). P.84-86 (1989).

L. Stauffer, A. Mharchi, C. Pirri, P. Wetzel, D. Bolmont, and G Gewinner,
Electronic structure and interfacial geometry of epitaxial two-dimensional Er
silicide on Si(III), Phys. Rev. B 47(16), P.10555-10562 (1993).

M. Lohmeier, W.J. Huisman, Gt. Horst, P.M. Zagwijn, and E. Vlieg, Atomic
structure and thermal stability of two-dimensional Er silicide on Si(lll), Phys.
Rev. B 54(3), [32004-2009 (1996).

C. Wigren, J.N. Andersen, and R. Nyholm, Sm- and Yb-induced reconstructions of
the Si(III) surface, Phys. Rev. B 48, p.11014-11019 (1993).

R. Hofmann, W.A. Henle, F.P. Netzer, and M. Neuber, Electronic-structure of
epitaxial Yb silicide, Phys. Rev. B 46(7), p.3857-3864 (1992).

C. Wigren, J.N. Andersen, R. N yholm, U.O. Karlsson, J. Nogami, A.A. Baski, and

CF. Quate, Adsorption-site determination of ordered Yb on Si(III) Surfaces, Phys.

Rev. B 47(15). [19663-9668 (1993).

W.A. Henle, M.G Ramsey, F.P. Netzer, and L. Horn, Formation of divalent Eu
silicides at the Eu-Si(111) interface, Surface Science 254(1-3), p.182-190 (1991).

131

 

r." .. -

[56]

[57]

[58]

[59]

[60]

[611

[62]

[63]

[64]

[65]

[66]

[67]

[68]

S. Vandre, T. Kalka, C. Preinesberger, and M. Dahne-Prietch, Epitaxial growth
and electronic structure of lanthanide silicides on n-type Si(III), J. Vac. Sci.
Technol. B 17(4), p.1682-l690 (1999).

T.V. Krachino, M.V. Kuz'min, M.V. Loginov, and M.A. Mittsev, Yb and Sm
Adsorption and Silicide Formation on Si(III) Surface, Phys. Low-Dim. Struct.
9/10, p.95-106 (1999).

M.V. Kuz'min, N .V. Mikhailov, and M.A. Mittsev, Formation and properties of a
binary adsorbed layer in a two-component adsorption system (Sm+Yb)-Si(111),
Phys. Sol. Sta. 45(3), P.579-585 (2003).

T.V. Krachino, M.V. Kuz'min, M.V. Loginov, and M.A. Mittsev, Growth of an
Eu-Si(111) thin ﬁlm structure: The stage of silicide formation, Phys. Sol. Sta.
46(3). [1563-568 (2004).

S. Vandre, T. Kalka, C. Preinesberger, I. Manke, H. Eisele, M. Dahne-Prietsch, R.
Meier, E. Weschke, and G Kaindl, Growth and electronic structure of Dy silicide
on Si(III), Appl. Surf. Sci. 123, p.100-103 (1998).

CH. Luo, GH. Shen, and L]. Chen, Vacancy ordering structures in epitaxial
RESig.Jr thin ﬁlms on (III)Si and (001 )Si, Appl. Surf. Sci. 113/114, p.457 (1997).

A. Travlos, N. Salamouras, and N. Boukos, Growth, structure and electrical
properties of epitaxial thulium silicide thin ﬁlms on silicon, J. Appl. Phys. 81(3),
p.1217-1221 (1997).

J.C. Chen, GH. Shen, and L]. Chen, Interfacial reactions of Gd thin ﬁlms on
( III ) and (001 )Si, Appl. Surf. Sci. 142, p.291-294 (1999).

Y.K. Lee, N. Fujimura, and T. Ito, Epitaxial-growth of Yttrium silicide YSi2-x
on(100) Si, J. Alloys Comp. 193(1-2), [1289-291 (1993).

S. Kinnou, J.Y. Veuillen, and T. Tan, Formation and electronic-properties of
erbium silicide on Si(100), Surf. Sci. 307(Part A), p.258-263 (1994).

G Kaltsas, A. Travlos, N. Salamouras, A.G Nassiopoulos, P. Revva, and A.
Traserse, Erbium silicide ﬁlms on (100) silicon, grown in high vacuum.
Fabrication and properties, Thin Solid Films 275(1-2), p.87-90 (1996).

N. Frangis, J .V. Landuyt, G Kaltsas, A. Travlos, and AG Nassiopoulos, Growth
of erbium-silicide on (100) silicon as characterized by electron microscopy and
diffraction, J. Crystal Growth 172, p.175-182 (1997).

A. Travlos, N. Salamouras, and N. Boukos, Epitaxial dysprosium silicide ﬁlms on

silicon: growth, structure and electrical properties, Thin Solid Films 397,
p.138-142 (2001).

132

 

EMEVJ ... ...
l

[69]

[70]

[71]

[72]

[73]

[74]

[75]

[76]

[77]

[78]

[79]

[80]

[31]

[82]

A. Travlos, P. Aloupogiannis, E. Rokofyllou, and C. Papastaikoudis, Epitaxial
lutetium silicide: Growth, characterization and electrical properties, J Applied
Physics 72(3). [1948-952 (1992).

G Peto, GL. Molnar, Z.E. Horvath, E. Zsoldos, N.Q. Khanh, J. Gyulai, and J.
Kanski, Formation of epitaxial HoSiz layer on Si( 100), Thin Solid Films 318,
p.168-171 (1998).

M.V. Katkov and J. Nogami, Yb and Nd growth on Si(001), Surf. Sci. 524(1-3),
p.129-136 (2003).

B.Z. Liu and J. Nogami, An STM study of the Si(001)(2x7)-Gd, Dy surface, Surf.
Sci. 540(1), P.136-144 (2003).

RJ. Godowski, J. Onsgaard, F. Orskov, and M. Christiansen, structural aspects of
the Sm/Si(100) interface, J. Mater. Sci. Lett. 9, p.989 (1990).

J.S. Yang, Q. Cai, X.D. Wang, and R. Koch, Initial stages of erbium disilicide
formation on Si(001), Surf. Sci. 526, p.291-296 (2003).

M. Kuzmin, R.E. Perala, P. Laukkanen, R.-L. Vaara, M.A. Mittsev, and 1.].
Vayrynen, Initial stages of Yb/Si( I 00) interface growth: 2x3 and 2x6
reconstructions, Appl. Surf. Sci. 214(1-4), p.196-207 (2003).

L. Fitting, M.C. Zeman, W.C. Yang, and R.J. Nemanich, Inﬂuence of strain,
surface diffusion and 0stwald ripening on the evolution of nanostructures for
erbium on Si(001), J Appl. Phys. 93(7), 114180-4184 (2003).

EM. Dheurle, M.O. Aboelfotoh, F. Pesavento, and CS. Petersson, Schottky
barriers of scandium and scandium monosilicide on silicon, Appl. Surf. Sci. 53,
p.237-239 (1991).

EU. Hillebrecht, Interaction of Ce with Si(100), Appl. Phys. Lett. 55(3),
p.277-279 (1989).

M.H. Unewisse and J .W.V. Storey, Electrical and infrared investigation of erbium
silicide, J Appl. Phys. 72(6), 132367-2371 (1992).

A. T ravlos and N. Salamouras, Superconductivity of lanthanum silicide thin ﬁlms,
Vacuum 48(1), p.13-14 (1997).

SM. Hogg and A. Vantomme, Growth and electrical characterization of GeSi1.7
epilayers formed by channeled ion beam systhesis, J Appl. Phys. 91(6),
p.3664-3668 (2002).

A. Travlos, N. Salamouras, and E. Flouda, Epitaxial erbium silicide ﬁlms on (100)
silicon: growth, structure and electrical properties, Appl. Surf. Sci. 120,
p.355-364 (1997).

133

 

[83]

[84]

[85]

[86]

[87]

[88]

[89]

[90]

[91]

[92]

[93]

[94]

[95]

[96]

[97]

J. Pierre, S. Auffret, J.A. Chroboczek, and T.T.A. Nguyen, Magnetotransport in
the rare earth silicides RSiZ-x, J. Phys.: Condens. Matter 6, p.79-92 (1994).

C. Boragno and MR. Bonansinga, Electrical and magnetic properties of ErSi2
and GdSiZ alloy thin ﬁlms, Solid State Communications 92(6), p.515-518 (1994).

A. Travlos, P. Aloupogiannis, E. Pokofyllou, and C. Papastaikoudis, Growth,
characterization and electrical properties of gadolinium silicide thin layers, phil.
Mag. B 67(4), p.485-495 (1993).

G Guizzetti, E. Mazzega, M. Michelini, and F. Nava, Electrical and optical
characterization of GdSiz and ErSiz alloy thin ﬁlms, J Applied Physics 67(7),
p.3393-3399 (1989).

EH. Kaatz, J .V.d. Spiegal, and W.R. Graham, Anomalous magnetotransport in
epitaxial TbSi2.x, J. Vac. Sci. Technol 9(3), p.426-429 (1991).

SM. Hogg, A. Vantomme, M.F. Wu, and G Langouche, Electrical properties of

rare earth silicides produced by channeled ion beam synthesis, Microelectron Eng.
50, p.211-215 (2000).

J. Chroboczek, Transport and magnetic properties of rare earth compounds
epitaxially grown on semiconductors, Acta Physica Polonica A 80(2), p.179-192
(1991).

BR Wohlfarth, Ferromagnetic materials: a handbook on the properties of
magnetically ordered substances ( North-Holland, Amsterdam, 1980).

J .M. Ziman, Electrons and phonons: the theory of transport phenomena in solids,
Oxford, 2001, chapter 9.

A. Franciosi, P. Perfetti, A.D. Katnani, and J .H. Weaver, Samarium chemisorption
on group-IV semiconductors, Phys. Rev. B 29(10), p.561] (1984).

R.G Long, M.C. Bost, and J .E. Mahan, Metallic behavior of lanthanum disilicide,
Appl. Phys. Lett. 53(14), p.1272-1273 (1988).

F. Canepa, S. Ciraﬁci, F. Merlo, and A. Palenzona, Electrical resistivity in the
R5Si3 systems (R=La, Ce, Pr; Nd, Sm), J. Alloys Comp. 203, p.Lll-L13 (1994).

WE. Henry, Bulletin of the American Physical Society 7, p.474 (1962).

X. Zhenjia, Electrical transport in RE silicides, Properties of metal silicides: Eds
K. Maex, M. Rossum (1995) IMEC. Leuven, Belgium and references therein,
(1994).

D. Tsamakis, M. Vlachos, A. Travlos, and N. Salamouras, Electrical properties of
crystalline Er and Dy silicide layers, Thin Solid Films 418, p.211-214 (2002).

134

[98]

[99]

[100]

[101]

[102]

[103]

[104]

[105]

[106]

[107]

[108]

[109]

[110]

F. Canepa, S. Cirafici, F. Merlo, and A. Palenzona, Electrical resistivity
measurements on some R55i3 phases: R=Gd, Tb, Lu and Y, J. Magn. Magn. Mater.
118, p.182-186 (1993).

V.N. Eremenko, V.E. Listovnichii, S.P. Luzan, Y.I. Buyanov, and RS. Martsenyuk,
Phase diagram of the holmium-silicon binary system and physical properties of
holmium silicides up to 1050C, J. Alloys Comp. 219, p.181-184 (1995).

G Binnig, H. Rohrer, C. Gerber, and E. Weibel, Tunneling through a controllable
vacuum gap, Appl. Phys. Lett. 40(2), p.178-180 (1982).

G Binnig, H. Rohrer, C. Gerber, and E. Weibel, Surface Studies by Scanning
Tunneling Microscopy, Phys. Rev. Lett. 49(1), p.57-60 (1982).

G Binnig, H. Rohrer, C. Gerber, and E. Weibel, 7x7 reconstruction on Si( 111 )
resolved in real space, Phys. Rev. Lett. 50, p.120 (1983).

D.M. Eigler and BK. Schweizer, Positioningsingle atoms with a scanning
tunneling microscope, Nature 344, p.524-526 (1990).

L.J. Whitman, J.A. Stroscio, R.A. Dragoset, and R.J. Celotta, Manipulation of
adsorbed atoms and creation of new structures on room-temperature surfaces
with a scanning tunneling microscope, Science 251, p.1206-1210 (1991).

H. Uchida, D. Huang, P. Grey, and M. Aono, Site-Speciﬁc Measurement of
Adatom Binding Energy Differences by Atom Extraction with the STM, Phys. Rev.
Lett. 70(13), 112040-2043 (1993).

TC. Shen, C. Wang, GC. Abeln, J.R. Tucker, J .W. Lyding, P. Avouris, and RE.
Walkup, Atomic scale desorption through electronic and vibrational excitation
mechanisms, Science 268(1590), ( 1995).

P. Avouris, Manipulation of matter at the atomic and molecular levels, Acc. Chem.
Res. 28, p.95-102 (1995).

S. Hosaka, S. Hosoki, T. Hasegawa, H. Koyanagi, T. Shintani, and M. Miyamoto,

Fabrication of nanostructures using scanning probe microscopes, J. Vac. Sci.
Techno] Bl3, p.2813 (1995).

TA. Jung, R.R. Schlittler, J.K. Gimzewski, H. Tang, and C. Joachim, Controlled
Room-Temperature Positioning of Individual Molecules: Molecular Flexure and
Motion, Science 271(5246), p.181-184 (1996).

G Dujardin, A. Mayne, 0. Robert, F. Rose, C. Joachim, and H. Tang, Veneical
Manipulation of Individual Atoms by a Direct STM Tip-Surface Contact on
Ge(III), Phys. Rev. Lett 80(14), 113085-3088 (1998).

135

 

[111]

[112]

[113]

[114]

[115]

[116]

[117]

[118]

[119]

[120]

[121]

[122]

[123]

[124]

[125]

[126]

U.J. Quaade, K. Stokbro, R. Lin, and F. Grey, Single-atom reversible recording at
room temperature, Nanotechnology 12, p.265-272 (2001).

Y. Okawa and M. Aono, Nanoscale control of chain polymerization, Nature
409(683), (2001).

R. Bennewitz, J.N. Crain, A. Kirakosian, J.-L. Lin, J.L. McChesney, D.Y.
Petrovykh, and F.J. Himpsel, Atomic scale memory at a silicon surface,
Nanotechnology 13(499-502), (2002).

SW. Hla and K.H. Rieder, Engineering single molecules with a scanning
tunneling microscope tip, Superlarrices and microstructures 31(1), p.63-72
(2002).

K. Sattler, Nanolithography using the scanning tunneling microscope, J ap. J. Appl.
Phys. Part] 42(7B), P.4825-4829 (2003).

O. Dudko, A.E. Filippov, J. Klafter, and M. Urbakh, Manipulatioons of individual
molecules by scanning probes, Nano Letters 3(6), p.795-798 (2003).

SD. Barrett, http://www. imagesxrn. org. ukl.

J.A. Kubby and J.J. Boland, Scanning tunneling microscopy of semiconductor
surfaces, Surface Science Reports 26, p.61-204 (1996).

J .A. Stroscio and W.J. Kaiser, Scanning Tunneling Microscopy, Academic Press
Inc. (1993) ch.l and ch.4 and references therein.

W. Telieps and E. Bauer, An analytical reﬂection and emission UHV surface
electron microscope, Ultramicroscopy 17, p.57-66 (1985).

E. Bauer, in: Proc. 5th Intern. Congr. on Electron Microscopy (Academic Press,
New York, 1962), p.D-11.

R.M. Tromp and MC. Reuter, Design of a new photo-emission/low-energy
electron microscope for surface studies, Ultrarnicroscopy 36, p.99-106 (1991).

R.M. Tromp, Low-energy electron microscopy, IBM J. Res. Develop. 44(4),
p.503-516 (2000).
13,

Matter

E. Bauer, Photoelectron Microscopy, J. Phys.: Condens.

p.11391-11404 (2001).

E. Bauer, Methods of surface studies depending on inelastic-scattering of
electrons, Vacuum 22(11), p.539-552 (1972).

R.M. Tromp and MC. Reuter, Imaging with a low-energy electron microscope,
Ultramicroscopy 50, p.171-178 ( 1993).

136

 

[127]

[128]

[129]

[130]

[131]

[132]

[133]

[134]

[135]

[136]
[137]

[138]

[139]

[140]

[141]

http://www. ulvac-ph i. co. jp/PDF/taguch i0] .pdf.

M. Aono, Y. Hou, C. Oshima, and Y. Ishizawa, Low-energy ion scattering from the
Si( 001 ) surface, Phys. Rev. B 49(8), p.567 (1982).

W.S. Yang and F. Jona, Atomic structure of Si(001) 2x1, Phys. Rev. B 28(4),
p.2049 (1983).

H. Over, J. Wasserfall, W. Ranke, C. Ambiatello, R. Sawitzki, D. Walf, and W.
Moritz, Surface atomic geometry of Si(001)-(2x1): A low-energy
electron-diffraction structure analysis, Phys. Rev. B 55(7), p.473] (1997).

J .A. Appelbaum, GA. Baraff, and DR. Hamann, Si(001) surface reconstruction:
spectroscopic selection of a structural model, Phys. Rev. Lett. 35, p.729 (1975).

DJ. Chadi, Si(100) surfaces: Atomic and electronic structures, J. Vac. Sci.
Technol 16, p.1290-1296 (1979).

A. Redondo and W.A. Goddard, Electronic correlation and the Si(100) surface:
Buckling versus nonbuckling, J. Vac. Sci. Technol. 21(2), p.344-350 (1982).

Z. Jing and J .L. Whitten, Ab initio studies of H chemisorption on Si(100), Phys.
Rev. B 46(15), [3.9544 (1992).

Y. Jung, Y. Shao, M.S. Gordon, D]. Doren, and M. Head-Gordon, Are both
symmetric and buckled dimers on Si( 100) minima? Density ﬁmctional and
multireference perturbation theory calculations, J. Chem. Phys. 119(20), p.10917
(2003).

S. Gasiorowicz, Quantum Physics - second edition: Wiley, p.334 (1996).

J.A. Appelbaum, GA. Baraff, and DR. Hamann, The Si(100) surface. 111.

Surface reconstruction, Phys. Rev. B 14(2), p.588-601 (1976).

CL. Alerhand, D. Vanderbilt, R.D. Meade, and J .D. Joannopoulos, Spontaneous
formation of stress domains on crystal surfaces, Appl. Phys. Lett. 61(17),
p.1973-1976 (1988).

http://www. research. ibm. com/leem/.

R.M. Tromp, R.J. Hamers, and J .E. Demuth, Si(001) dimer structure observed
with scanning tunneling microscopy, Phys. Rev. Lett. 55, p.1303-1306 (1985).

T. Tabata, T. Aruga, and Y. Murata, Order-disorder transition on Si( 001 )
-C(4X2)T0(2X1), Surface Science 179, p.L63 (1987).

137

[142]

[143]

[144]

[145]

[146]

[147]

[148]

[149]

[150]

[151]

[152]

[153]

[154]

Y. Enta, S. Suzuki, and S. Kono, Angle-resolved-photoemission study of the
electronic structure of the Si(001) c(4x2) surface, Phys. Rev. Lett 65, p.2704
(1990).

RA. Wolkow, Direct observation of an increase in buckled dimer on Si(001) at
low-temperature, Phys. Rev. Lett 68, p.2636 (1992).

A. Ramstad, G Brooks, and R]. Kelly, Theoretical study of the Si(001) surface
reconstruction, Phys. Rev. B 51(20), p.14504-14523 (1995).

G LeLay, A. Criecenti, C. Ottaviani, P. Perfetti, T. Tanikawa, I. Matsuda, and S.
Hasegawa, Evidence of asymmetric dimers down to 40 K at the clean Si(100)
surface, Phys. Rev. B 66, p.153317 (2002).

K. Hata, S. Yoshida, and H. Shigekawa, p(2x2) Phase of Buckled Dimers of
Si(100) Observed on n-Type Substrates below 40 K by Scanning Tunneling
Microscopy, Phys. Rev. Lett 89, p.286104 (2002).

M. Matsumoto, K. Fukutani, and T. Okano, Low-Energy Electron Diffraction
Study of the Phase Transition of Si( 001 ) Surface below 40 K, Phys. Rev. Lett 90,
p.106103 (2003).

R. Stalder, H. Sirringhaus, N. Onda, and H.V. Kane], Observation of misﬁt
dislocations in epitaxial CoSi;/Si( 111 ) layers by scanning tunneling microscopy,
Appl. Phys. Lett. 59(16), 111960-1962 (1991).

K. Ojima, M. Yochimura, and K. Ueda, STM observation of the 2x3 and c(2x6)
structures on Ba/Si(001), Surf. Sci. 491(1-2), p.169 (2001).

DE. Jones, J.P. Pelz, Y. Hong, E. Bauer, and I.S.T. Tsong, Striped Phase and
Temperature Dependent Step Shape Transition on Highly B-Doped Si(001)-(2x1)
Surfaces, Phys. Rev. Lett. 77(2), p.330 (1996).

J .-F. Nielsen, H.-J. Im, and J .P. Pelz, Scanning tunneling microscope studies of
boron-doped Si( 001 ), J. Vac. Sci. Tec. A 17(4), p.1670 (1999).

DE. Jones, J .P. Pelz, Y.H. Xie, P.J. Silverman, and GH. Gilmer, Enhanced Step
Waviness on SiGe( 001 H 2x1 ) Surface under Tensile Strain, Phys. Rev. Lett. 75(8),
p.1570 (1995).

K.M. Chen, D.E. Jesson, SJ. Pennycook, M. Mostoller, and T. Kaplan, Step
Instabilities: A New Kinetic Route to 3D Growth, Phys. Rev. Lett. 75(8), p.1582
(1995).

G Ye, J. Nogami, and M.A. Crimp (submitted).

138

 

[155]

[156]

[157]

[158]

[159]

[160]

[161]

[162]

[163]

[164]

[165]

[166]

[167]

OF Balkashin, A.GM. Jansen, O. Laborde, U. Gottlieb, GL. Sukhodub, P. Wyder,
and I.K. Yanson, High-Frequency Point-Contact Spectroscopy of IiSiz, TaSiz, and
VSi2, J. Low Temp. Phys. 129(314), [1105-116 (2002).

CC. Lin, W.S. Chen, K.Y.J. Hsu, H.K. Liou, and K.N. Tu, Reliability study of
sub-micron titanium silicide contacts, Appl. Surf. Sci. 92, p.660-664 (1996).

N.M. Ravindra, L. Jin, D. Ivanov, V.R. Mehta, L.M. Dieng, G Popov, O.H. Gokce,
J. Grow, and AT Fiory, Electrical and Compositional Properties of TaSiz Films, J.
of Elec. Matls. 31(10), 111074-1079 (2002).

G Medeiros-Ribeiro, D.A.A. Ohlberg, D.R. Bowler, R.E. Tanner, GAD. Briggs,
and RS. Williams, Titanium disilicide nanostructures: two phases and their
surfaces, Surf. Sci. 431, p.116-127 (1999).

KL. Saenger, J. C. Cabral, LA. Clevenger, R.A. Roy, and S. Wind, A kinetic
study of the C49 to C54 IiSiZ conversion using electrical resistivity measurements
on single narrow lines, J. Appl. Physics 78(12), p.7040-7044 (1995).

P. Gallais, J .J . Hantzpergue, and LC. Remy, Sputter deposition of thin tantalum
layers and low temperature interactions between tantalum and SiOz and tantalum
and silicon, Thin Solid Films 165, p.227-236 (1988).

M.T. Huang, T.L. Martin, V. Malhotra, and IE. Mahan, Electronic transport
properties of tantalum disilicide thin ﬁlms, J. Vac. Sci. Technol 33(3), p.836-845
(1985).

H. Oppolzer, F. Neppl, K. Hieber, and V. Huber, Inﬂuence of slight deviations
from TaSiZ stoichiometry on the high-temperature stability of tantalum
silicide/silicon contacts, J. Vac. Sci. Technol B2(4), p.630-635 (1984).

SB. Hemer and KS. Jones, Investigation of mechanisms of vacancy generation in
silicon in the presence of a TaSiz ﬁlm, J. Appl. Phys. 82(2), p.583-588 (1997).

CV Hul'ko, R. Boukherroub, and GP. Lopinski, Chemical and thermal stability
of titanium disilicide contacts on silicon, J Appl. Phys. 90(3), p.1655-1659 (2001).

N. Lundqvist, J. Aberg, S. Nygren, C.-A. Bjormander, and S.-L. Zhang, Effects of
substrate bias and temperature during titanium sputter-deposition on the phase
formation in IiSiz, Microelectron Eng. 60, p.211-220 (2002).

A.W. Dunn, B.N. Cotier, A. Nogaret, P. Moriarty, and RH. Beton, Molecular scale

alignment strategies: An investigation of Ag adsorption on patterned ﬁdlerene
layers, Appl. Phys. Lett. 71(20). P.2937-2939 (1997).

RA. Bennett, B. Ashcroft, Z. He, and R.M. Tromp, Growth dynamics of titanium
silicide nanowires observed with low-energy electron microscopy, J. Vac. Sci. Tec.
B 20(6), p.2500-2504 (2002).

139

 

in!!!

[168]

[169]

[170]

[171]

[172]

[173]

[174]

[175]

[176]

[177]

[178]

[179]

[180]

[181]

[182]

W.-C. Yang, H. Ade, and R.J. Nemanich, Shape stability of TiSiz islands on
Si( 111 ), J. Appl. Physics 95(3), P.1572-1576 (2004).

E. Abramof, A.F.d. Silva, B.E. Semelius, J.P.d. Souza, and H. Boudinov,
Metal-nonmetal transition and resistivity of silicon implanted with bismuth, J.
Mater. Res. 12(3), [1641-645 (1997).

http://www. tydex. ru/mate rial s/material s2/S i. html .

P. Gambardella, A. Dallmeyer, K. Maiti, M.C. Malagoli, W. Eberhardt, K. Kern,
and C. Carbone, Ferromagntism in one-dimensional monatomic metal chanis,
Nature 416(6878), P301 (2002).

P. Gambardella, Magnetism in monatomic metal wires, J. Phys. Condens. Matter
15, p.82533-S2546 (2003).

J. Pierre, E. Siaud, and D. Frachon, Magnetic Properties of Rare Earth Disilicides
RSiZ, J. Less-Common Met. 139, p.321-329 (1988).

K. Sekizawa and K. Yasukochi, Antiferromagnetism of Disilicides of Heavy Rare
Earth Metals, J. Phys. Soc. Jpn. 21, p.274 (1966).

C. Kittel, Introduction to Solid State Physics. 6 ed. 1986.

S. Labroo, X. Zhang, P. Hill, and N. Ali, Electrical and magnetic properties of
antiferromagnetic rare earth disilicides, J. Less-Common Met. 149, p.337 (1989).

H. Yashima, T. Satoh, H. Mori, D. Watanabe, and T. Ohtsuka, Thermal and
magnetic properties and crystal structures of CeGez and CeSiz, Solid State
Communications 41(1), p.1-4 (1982).

J. Pierre, S. Auffret, E. Siaud, R. Madar, E. Houssay, A. Rouault, and J .P. Senateur,
Magnetic properties of rare earth silicide crystals RSiZ-x (R = Pt; Nd, Gd), J.
Magn. Magn. Mater. 89, p.86-96 (1990).

ET Matthias, E. Corenzwit, and W.H. Zachriasen, Phys. Rev. 112, p.89 (1958).

K.S.V.L. Narasimhan and H. Steinﬁnk, Magnetic Investigations on AlBZ Type
Structures, J. Solid State Chem. 10, p.137-141 (1973).

C. Pescher, J. Pierre, A. Ermolieff, and C. Vannuffel, Magnetic properties of
gadolinium silicide thin ﬁlms produced by different fabrication processes, Thin
Solid Films 278, p.140-143 (1996).

J. Pierre, B. Lambert-Andron, and J .L. Soubeyroux, Magnetic-structures of rare
earth silicides NdSi2.x, HoSi2.x, DySi2.x, J. Magn. Magn. Mater. 81, p.39 (1988).

140

 

[183]

[184]

[185]

[186]

[187]

[188]

[189]

[190]

[191]

[192]

[193]

[194]

[195]

MC. Hersam, N.P. Guisinger, and J.W. Lyding, Atomic-level study of the
robustness of the Si(100)-2x1:H surface following exposure ro ambient conditions,
Appl. Phys. Lett. 78(7), p.886-888 (2001).

T. Sakurai and H.D. Hagstrum, Interplay of the monohydride phase and a newly
discovered dihydride phase in chemisorption of H on Si(100)2x1, Phys. Rev. B 14,
p.1593-1596 (1976).

Y]. Chabal and K. Raghavachari, New Ordered Structure for the H-Saturated
Si( 100) Sun‘ace: The (3x1) Phase, Phys. Rev. Lett 54(10), 111055-1058 (1985).

M. Niwa, H. Iwasaki, and S. Hasegawa, Hydrogen terminated Si( 100) surfaces
studied by scanning tunneling microscopy, X-ray photon spectroscopy, and
auger-electron spectroscopy, J. Vac. Sci. Tec. A 8(1), p.266-269 (1990).

J.J. Boland, Structure of the H-saturated Si( 100) surface, Phys. Rev. Lett. 65,
p.3325-3328 (1990).

J .J. Boland, Role of bond-strain in the chemistry of hydrogen on the Si( 100)
surface, Surface Science 261(1-3), p.17-28 (1992).

Y. Okada, H. Shimomura, and M. Kawabe, Atomic image observation of
hydrogen-saturated Si(100) prepared by atomic-hydrogen irradiation, J ap. J. Appl.
Phys. Part2-Letters 31(8A). P-L1121-L1123 (1992).

K. Kitahara and O. Ueda, Observation of atomic-structure by scanning-tunneling
-microscopy of vicinal Si(100) surface annealed in hydrogen gas, Jap. J. Appl.
Phys. Part2-Letters 33(llB). P.L157l-L1573 (1994).

M. Yoshimura and K. Ueda, Evolution of surface structures on Si(001) during
hydrogen desorption, Appl. Surf. Sci. 121, p.179-182 (1997).

T. Komeda and Y. Kumagai, Si( 001 ) surface variation with annealing in ambient
H2, Phys. Rev. B 58(3). 111385-1391 (1998).

K. Arima, K. Endo, T. Kataoka, Y. Oshikane, H. Inoue, and Y. Mori, Scanning
tunneling microscopy study of hydrogen-terminated Si( 001 ) surfaces after wet
cleaning, Surface Science 446(1-2), p.128-136 (2000).

. T. Hashizume, H. Kajiyama, Y. Suwa, S. Heike, S. Matsuura, J. Nara, and T. Ohno,

Adsorption of Si atom on H-terminated Si(001)-2x1 surface, Appl. Surf. Sci.
216(1-4), P.15-18 (2003).

Y. Suwa, S. Matsuura, M. Fujimori, S. Heike, T. Onogi, H. Kajiyama, T. Hitosugi,

K. Kitazawa, T. Uda, and T. Hashizume, Dopant-pair structures segregated on a
hydrogen-terminated Si(100) surface, Phys. Rev. Lett 90(15), p.156101 (2003).

141

 

[196]

[197]

[198]

[199]

[200]

[201]

[202]

[203]

[204]

[205]

[206]

[207]

T. Hirosugi, T. Hashizume, S. Heike, Y. Wada, S. Watanabe, T. Hasegawa, and K.
Kitazawa, Scanning tunneling spectroscopy of dangling-bond wires fabricated on
the Si(100)-2x1 -H surface, Appl. Phys. A 66, p.8695-S699 (1998).

J .A. Schaefer, Electronic and structural-properties of hydrogen on semiconductor
surfaces, Physica B ”004), p.45-68 (1991).

E.J. Buehler and J.J. Boland, Dimer preparation that mimics the transition state
for the adsorption of H-2 on the Si(100)-2x1 surface, Science 290(5491),
p.506-509 (2000).

K. Bobrov, G Comtet, G Dujardin, and L. Hellner, Electronic structure of
partially hydrogenated Si(100)-(2 x 1) surfaces prepared by thermal and
nontherrnal desorption, Phys. Rev. Lett 86(12), p.2633-2636 (2001).

D.X. Chen and J .J . Boland, Chemisorption-induced disruption of surface
electronic structure: Hydrogen adsorption on the Si(100)-2x1 surface, Phys. Rev.
B 65(16). 13.165336 (2002).

J .J . Boland, Evidence of pairing and its role in the recombinative desorption of
hydrogen from the Si(100)-2x1 surface, Phys. Rev. Lett 67(12), p.1539 (1991).

J .J . Boland, Scanning tunneling microscopy study of the adsorption and
recombinative desorption of hydrogen from the Si(100)-2x1 surface, J. Vac. Sci.
Tec. A 10(4), p.2458-2464 (1992).

X.D. Wang, H. Lu, T. Hashizume, H.W. Pickering, and T. Sakurai,
Atomic-hydrogen chemisorption on Si(100)(2x1) studied by FI-STM, Appl. Surf.
Sci. 57(1-4), P.266-274 (1993).

GC. Abeln, T.C. Shen, J .R. Tucker, and J .W. Lyding, Nanoscale STM-patteming
and chemical modiﬁcation of the Si( 100) surface, Microelectron Eng. 27(1-4),
p.23-26 (1995).

P. Avouris, R.E. Walkup, A.R. Rossi, T.C. Shen, GC. Abeln, J .R. Tucker, and J .W.
Lyding, S TM-induced H atom desorption from Si(100): Isotope effects an site
selectivity, Chem. Phys. Lett. 257(1-2), p.148-154 (1996).

J.W. Lyding, T.C. Shen, GC. Abeln, C. Wang, and JR. 'Ducker, Nanoscale
patterning and selective chemistry of silicon surfaces by ultrahigh-vacuum
scanning tunneling microscopy, Nanotechnology 7(2), p.128-133 (1996).

J .W. Lyding, T.C. Shen, GC. Abeln, C. Wang, P.A. Scott, J .R. Tucker, P. Avouris,
and RE. Walkup, Ultrahigh vacuum scanning tunneling microscope-based
nanolithography and selective chamistry on silicon surfaces, Israel J. Chem. 36(1),

p.3-10 (1996).

142

 

[208]

[209]

[210]

[211]

[212]

[213]

[214]

[215]

[216]

[217]

[218]

[219]

D.H. Huang and Y. Yamarnoto, Manipulating atoms one by one with a scanning
tunneling microscope, Surf. Rev. Lett. 3(3), p.1463-1472 (1996).

T. Hitosugi, T. Hashizume, S. Heike, S. Watanabe, Y. Wada, T. Hasegawa, and K.
Kitazawa, Scanning tunneling spectroscopy of dangling-bond wires fabricated on
the Si(100)-2x1 -H surface, Jap. J. Appl. Phys. Part2-Letters 36(3B), p.L361-L364
(1997).

M. Sakurai, C. Thirstrup, T. Nakaya, and M. Aono, Local modiﬁcation of
hydrogen-terminated silicon surfaces by clean and hydrogen-covered STM tips,
Surf. Sci. 386(1-3), P.154-160 (1997).

C. Thirstrup, M. Sakurai, T. Nakayama, and M. Aono, Atomic scale modiﬁcations

of hydrogen-terminated silicon 2x1 and 3x1 surfaces by scanning tunneling
microscope, Surf. Sci. 411, p.203-214 (1998).

C. Syrykh, J.P. Nys, B. Legrand, and D. Stievenard, Nanoscale desorption of
H—passivated Si(100)-2x1 surfaces using an ultrahigh vacuum scanning tunneling
microscope, J Appl. Phys. 85(7), p.3887-3892 (1999).

Q. Shi, D.H. Huang, and GS. Zhu, Vibrational-energy redistribution in
single-atom manipulation by scanning tunneling microscope, Jap. J. Appl. Phys.

Part] 38(6B), p.3856-3659 (1999).

K. Stokbro, U.J. Quaade, R. Lin, C. Thirstrup, and G Grey, Electronic mechanism
of STM-induced diﬁ‘usion of hydrogen on Si(100), Faraday Discussions 117,
p.231-240 (2000).

M. Durr, Z.H. Hu, A. Biedermann, U. Hofer, and TR Heinz, Real-space study of
the pathway for dissociative adsorption of H—2 on Si( 001 ), Phys. Rev. Lett 88(4),
p.046104 (2002).

MC. Hersam, N.P. Guisinger, J. Lee, K.G Cheng, and J.W. Lyding, Variable
temperature study of the passivation of dangling bonds at Si(100)-2x1
reconstructed surfaces with H and D, Appl. Phys. Lett. 80(2), p.201-203 (2002).

Z.H. Hu, Quantitative study of adsorbate-adsorbate interactions of hydrogen on
the Si(100) surface, Phys. Rev B 68(15), p.155418 (2003).

H. Tsurumaki, K. Iwamura, T. Karato, S. Inanaga, and A. Namiki, Adsorption and
desorption of deuterium on partially oxidized Si(100) surfaces, Phys. Rev. B
67(15). 13.155316 (2003).

L. Soukiassian, A.J. Mayne, M. Carbone, and G Dujardin, Atomic-scale
desorption of H atoms from the Si(100)-2x1 : H sudace: Inelastic electron
interactions, Phys. Rev. B 68(3), p.035303 (2003).

143

 

[220]

[221]

[222]

[223]

[224]

[225]

[226]

[227]

[228]

[229]

[230]

[231]

T. Hashizume, S. Heike, M.I. Lutwyche, S. Watanabe, and Y. Wada, Atom
structures on the Si(100) surface, Surface Science 386(1-3), p.161-165 (1997).

T. Hitosugi, T. Hashizume, S. Heike, H. Kajiyama, Y. Wada, S. Watanabe, T.
Hasegawa, and K. Kitazawa, Electronic structures of dangling-bond structures
fabricated on hydrogen-terminated Si(100)-2x1 surfaces, Deffect and Diffusion
Forum/Journal 162, p.43-57 (1998).

T. Hitosugi, S. Heike, T. Onogi, T. Hashizume, S. Watanabe, Z.Q. Li, K. Ohno, Y.
Kawazoe, T. Hasegawa, and K. Kitazawa, Jahn-Teller distortion in dangling-bond

linear chains fabricated on a hydrogen-terminated Si(100)-2 x 1 surface, Phys.
Rev. Lett 82(20), P.4034-4037 (1999).

R.P. Chen and D.S. Lin, Distribution of dangling bond pairs on partially

hydrogen-terminated Si(100) surface observed by scanning tunneling microscopy,
Surf. Sci. 454, p.196-200 (2000).

H. Jungblut, D.J. Muller, M. Aggour, and HI. Lewerenz,
Scanning-tunneling-microscopy observation of atomic structures on silicon (100)
surface in air, Electrochimica Acta 38(10), p.1367-137l (1993).

N. Kramer, M.R. vandenBerg, and C. Schonenberger, Scanning tunneling
microscope-induced oxidation of hydrogen passivated silicon surfaces, Thin Solid

Films 282(1-2), [1637-639 (1996).

H. Kijiyama, S. Heike, Y. Wada, and T. Hashizume, Initial stage oxidation at an
unpaired dangling bond site on a Si(100)-2 X I-H surface, Thin Solid Films
344(Sp. Iss. SI), p.550-553 (1999).

Z.H. Mai, Y.F. Lu, W.D. Song, and WK. Chim, Nano-modiﬁcation on
hydrogen-passivated Si surfaces by a laser-assisted scanning tunneling
microscope operating in air, Appl. Surf. Sci. 154, p.360-364 (2000).

DA. MacLaren, NJ. Curson, P. Atkinson, B. Holst, D.J. Johnson, and W. Allison,
Simple design for the transportation of ex situ prepared hydrogen passivated
silicon, J. Vac. Sci. Tec. A 20(1), p.285-287 (2002).

S. Maruno, H. Iwasaki, K. Horioka, S.T. Li, and S. Nakamura, Electronic
structures of the monohydride (2x1):H and dihydride (1x1 )::2H Si(001) surfaces
studied by angle—resolved electron-energy-loss spectroscopy, Phys. Rev. B 27(7),

p.4110—4116(1983).

D.T. Jiang, GW. Anderson, K. Grifﬁths, T.K. Sham, and RR. Norton, Adsorption
of atomic hydrogen on Si(199)-2x1 at 400 K, Phys. Rev. B 48(7), p.4952-4955
(1993).

J .Y. Veuillen and T.A.N. Tan, Hydrogen adsorption on ErSi1,7(0001), Phys. Rev. B
52(15), p.10796-10799 (1995). '

144

 

[232]

[233]

[234]

[235]

[236]

[237]

[238]
[239]

[240]

[241]

[242]

[243]

S. Saintenoy, P. Wetzel, C. Pirri, J.C. Peruchetti, D. Bolmont, and G Gewinner,
Hydrogen adsorption on erbium silicide surface, Solid State Communications
94(9), [3719-723 (1995).

S. Saintenoy, P. Wetzel, C. Pirri, D. Bolmont, and G Gewinner, Interaction of H
with epitaxial Er silicide layers on Si(III) adsorption versus absorption, Surface
Science 349, p. 145-154 (1996).

P. Wetzel, C. Pirri, and G Gewinner, Buckling reversal of the Si(III) bilayer
termination of 2-dimensional ErSiz upon H dosing, Europhysics Letters 38(5),
p.359-364 (1997).

P. Wetzel, T. Angot, C. Pirri, and G Gewinner, Origin of the
semimetal-to-semiconductor transition observed in two-dimensional Er silicide

upon H exposure: evidence at two chemisorption sites, Surf. Sci. 383, p.340-349
(1997).

P. Louis, T. Angot, D. Bolmont, and G Gewinner, Pairing mechanism in
interaction of atomic hydrogen with epitaxial erbium silicide, Surf. Sci. 422,

p.65-76 (1999).

M.-H. Thilier, C. Pirri, D. Berling, D. Bolmont, G Gewinner, and P. Wetzel,
Structure of clean and H-saturated epitaxial two-dimensional Er silicide on
Si(III) studied by SEXAFS, Surf. Sci. 555(1-3), 1394-100 (2004).

http://www. icknowl edge. com/glossa ry/s. html .
http://www. ece. gatech. edu/resea rch/labs/vc/theory/sheetRes. html .

L. Maisse] and R. Glang, Handbook of Thin Film Technology. 1983:

McGraw-Hill.
J. Homstra and L.J.v.d. Pauw, J. Electron. Control 7, p. 169 (1959).

W.L.V. Price, Extension of van der Pauw's theorem for measuring speciﬁc
resistivity in disks of arbitrary shape to anisotropic media, J. Phys. D: Appl. Phys.
5(6), p.112? (1972).

W.L.V. Price, Electric potential and current distribution in a rectangular sample
of anisotropic material with application to the measurement of the principal
resistivities by an extension of van der Pauw's method, Solid State Electronics 16,
p.753-762 (1973).

145