K. Rademacher, Strukturelle und elektrische Eigenschaften von epitaktischen, ...... Die ausgezeichnete wissenschaftliche Betreuung durch PD Dr. Hans von ...
D!ss.ETH
DISS.ETH Nr. 10994
Electrical and silicide
-
of
optical properties
Silicon heterostructures.
A dissertation submitted to the
SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZÜRICH
for the
degree of
DOCTOR OF NATURAL SCIENCES
presented by Claude Bernard Schwarz
Dipl. Phys.
ETH
bornOctoberl9,1966 Citizen of Richterswil and Zürich
accepted
on
the recommendation of
Prof. Dr. P. Wächter, examiner Prof. Dr. I. PD Dr. H.
Eisele, co-examiner
von
Känel, co-examiner
1995
TO YVONNE AND MY PARENTS
CONTENTS
CONTENTS
KURZFASSUNG
1
ABSTRACT
3
1. THIN EPITAXIAL FILMS
5
1.1.
Epitaxy
5
1.2. Strain of thin films
7
1.3. Strain determination with RBS
channeling
8
1.4. Strain relaxation
2. Co
-
10
12
Si SYSTEM
3. NEW EPITAXIAL CoSi 3.1. Introduction
-
PHASE ON
15
Si(lll)
motivation
15
3.2. Growth
16
3.3. Structure
17
3.4. Electronic structure
19
3.5.
21
Stability
3.6. Electrical and 3.7.
4.
-
CoSi2
24
magnetic properties
28
Summary
ON
Si(lll)
AND
4.1. Introduction
-
Si(001)
29
motivation
29
4.2. Growth
29
4.3. Structure
31
4.3.1.
CoSi2/Si(lll)
31
4.3.2.
CoSySiCOOl)
31
4.3.3. Strain and critical thickness
hc of CoSi2
on
Si(l 11)
32
CONTENTS
II
4.3.4. Strain and critical thickness 4.4. Electric transport
4.5.
5.
CoSij
-
hc
of CoSi2
on
Si(OOl)
properties
36 38
40
Summary
Si HETEROSTRUCTURES ON
5.1. Introduction
-
Si(lll)
42
motivation
42
5.2. Growth
43
5.3. Stracture
44
5.3.1. Strain and critical thickness 5.4.
hc
of Si
50 52
Summary
6. APPLICATION OF SILICIDE
-
SILICON HETEROSTRUCTURES TO
TUNABLE INFRARED DETECTOR 6.1. Introduction 6.2.
Schottky
-
-
motivation
54 55
contacts
photoemission
57
6.3.1. Photoexcitation
57
Transport
58
6.3. Internal
6.3.2.
6.3.3. Emission 6.3.4. 6.4.
54
across
the interface barrier
Responsivity quantum efficiency
62
-
Principle of Tunable Internal Photoemission 6.4.1.
Symmetrical
6.4.2.
Asymmetrical
TIPS TIPS
6.5. Fabrication of TIPS
Sensor
62
63 65 68
6.5.1. Growth and characterization of PtSi
6.5.2.
60
Photolithography process
6.6. Electrical characterization of TIPS
68 70
74
6.6.1.1-V measurements
74
6.6.2. C-V measurements
78
6.7. Photoelectric characterization of TIPS
6.7.1.
Experimental
81
6.7.2. Simulation and 6.7.3.
Photoresponse
81
fitting procedure
measurements
81 83
CONTENTS
A) Symmetrical
6.7.4. 6.8.
TIPS
83
B) Asymmetrical TIPS
85
Detectivity
92
Summary
93
REFERENCES
95
APPENDICES A:
III
102
Elasticity
102
B: Strain evaluation
C: Critical thickness
104
(model by Matthews)
D:
Schottky diode analysis
E:
Schottky
diode
105
(Cheung method)
106
analysis II (Werner method)
108
I
PUBLICATIONS-PRESENTATIONS
110
CURRICULUM VITAE
114
DANK
115
KURZFASSUNG
1
KURZFASSUNG In
einführenden
einem
zusammengefasst.
Wachstums
Filme,
pseudomorpher
Gitterfehlanpassung
Spannung
zum
ist mit einer Form
Verständnis des Wachstums mittels Rutherford Das
Prinzip
solcher
wird
anderem
epitaktischen
Substrat von
das
Stabilisierung
vorgestellt.
-
Wachstum
Phase
die
Filme in der
Übersicht über das Co
Kristallstrukturen im zweiten
wichtig
in dünnen Filmen wurden -
Experimenten
Kapitel,
wird im
-
Si
gewonnen.
Phasendiagram und einigen wichtigen
Kapitel
drei die
erstmalige Stabilisierung
die sogenannte FeSi Struktur auf. Zum ersten Mal
Normalbedingungen
metallische Monosilizid
-
Phase mit Molekularstrahl
-
Epitaxie
auf Si(l
gelang
11)
zu
es
Diese Phase hat ganz andere
Eigenschaften
als die volumenstabile
insbesondere in der elektronischen Struktur durch UPS äussert. Während in der
Transportmessungen
Transport lochartig ist,
ist
pseudomorphen CoSi,
die
er
denjeningen
von
allerdings
vier werden
dem
Wegen
Filmwachstum elastische Filmen auf Einer der
Si(lll)
wichtigsten
Versetzungen
zur
an
einige
von
Totalenergie
strukturelle
stabilisieren. In
Gitteranpassungsfehler
quantitativ
der
was
sich
Phase der elektrische Die Stabilität
Phasenübergang
vom
zur
Rechnungen plausibel gemacht
CoSi2 von
-
insbesondere
Filmen auf
-1.2%
mit der Theorie
von
neue
Si(lll)
erfordert
der Schichten. Die Relaxation
gespeicherten
Phase
verglichen.
Parameter ist dabei die kritische
Verminderung
-
Phase,
-
und in elektrischen
elektronartig.
Eigenschaften,
dünnen kohärenten
Verspannung
konnte
pseudomorphen
sehr beschränkt ist, sowie der
der isostrukturellen FeSi Phasen
Spannungsmessungen
vorgestellt.
neuen
Messungen
in der volumenstabilen Phase
volumenstabilen s-Phase werden anhand
Kapitel
von
mir diese
einem ersten Schritt wird das Wachstum sowie der strukturelle Nachweis der CsCl
Im
für das
CoSi mit CsCl Struktur beschrieben. In Volumenform weist CoSi unter
pseudomorphem
und mit
Regel
Spannungsmessungen wird darin vorgestellt.
Nach einer kurzen
dargestellt.
als
elastisch verspannt. Diese
und ist somit äusserst
Spektrometrie (RBS) Channeling
epitaktischen
vom
einer
epitaktische
Da
haben, sind diese Filme
Energie verknüpft
des
Konzept
epitaktischer Systeme. Spannungen
Rückstreu
-
Unter der
d. h.
einige Eigenschaften
werden
nicht vorkommt, kurz
Gleichgewichtsphase eine
Kapitel
von
Resultate
und
das
Si(001)
kohärente
epitaktischen CoSi2
Matthews beschrieben werden.
Schichtdicke,
elastischen
bei der die
Energie beginnt.
Bildung Für
von
Si(l 11)
KURZFASSUNG
2
beträgt
sie ~45Ä und auf
channeling
an
Suiziden und Silizid
Beherrschen
Das
Si(lOO)
75
Ä. Detaillierte Spannungsmessungen mit RBS
Silizium Heterostrukturen werden vorgestellt.
Wachstums
des
dokumentiert ist. Um die Heterostrukturen für
Opotelektronik
verwenden
Wachstum
Das
sowie
zweidimensionaler
verspannter
erfolgreiches Überwachsen
ein
fiir
Voraussetzung
-
~
mit
Si,
Anwendungen
-
zu
unvermeidbaren Defekten
Im
Gegensatz
Wachstumsprobleme gelöst werden,
Ein
Infrarot
Heterostrukturen auf
Si(lll),
Wesentlichen
zwei
aus
dazwischenliegende
an
den
-
Si(OOl) Substraten
Detektor, hergestellt
wird im
Kapitel
entgegengesetzten
-
sechs
aus
Fehlorientierung
Schottky
-
der
was
deshalb
Si(lll)
zu
die
d. h. mit
neuen
wesentlichen durch die
Detektoren
-
Dieser Detektor besteht im
Kontakten
in
Serie.
Photoemissionsprozesse beobachtbar.
hingegen
Spannung
Schottky
CoSi2/Si/CoSi2
Das
Je nach Polarisation des Detektors sind unter Lichteinfall
Diagram.
einer äusseren
und
verarmt und es resultiert deshalb ein
-
Anlegen
konnten auf
PtSi/Si/CoSi2
heutigen Schottky Infrarot Detektoren besteht in
vorgestellten
muss
geeignete Substratwahl,
vorgestellt.
vollständig
zwischen zwei und vier verschiedene interne
durch
zu
Silizium
-
CoSi2/Si(lll) Grenzflächenstufen,
unter anderem durch
undotierte Silizium ist
trapezförmiges Energie
Die
und/oder
Fehlorientierung.
neuartiger
Nachteil der
-
heraus, dass das Hauptproblem
die Relaxation der Suizide ist. Die kleine
Beginn verspannt aufwachsen.
-
fiinf
Silizid
absolut kohärentes Wachstum, auch dünnster, Suizide verhindert. Silizium
kleiner Substrat
Kapitel
in der Mikro
Charakterisierung solcher
strukturelle
die
Überwachstums
Substrate führt
in
war
können, ist eine einwandfreie kristalline Qualität erforderlich.
zu
Heterostrukturen wird ausführlich beschrieben. Es stellte sich des Silizium
welches
Suizide
haben eben den
variieren
zu
Barrieren bestimmt
Der grosse
ihrem fixen Detektionsbereich.
Vorteil, den Detektionsbereich
können. Die
Grenzenergie,
wird, lässt sich bei
den
die im
vorgestellten
Detektoren zwischen 0.4 und 0.8 eV verschieben. Das Verhalten der Detektoren kann durch
geeignete
Wahl der beiden Metallschichten stark beeinflusst werden.
energieaufgelöster, beherrschten Si
-
d. h.
farbiger,
Technologie,
IR
-
Der
Weg
zu
echt
Detektion, in Kombination mit der heutzutage gut
scheint somit
gegeben zu
sein.
3
ABSTRACT
ABSTRACT In the
introductory chapter
one, some
concept of pseudomorphism, which relates introduced. Since in
Substrate, they when
are
epitaxial
films have
summarized. The
stabilization of bulk unstable
epitaxial
to
are
phases,
is
small lattice mismatch relative to the
a
a
form of energy which has to be considered
epitaxial Systems. Important
information about the strain State of thin
strained.
with
dealing
general epitaxial
aspects of epitaxial growth
Obviously
strain is
films have been obtained from strain measurements
using
backscattering
Rutherford
spectrometry (RBS) ion channeling. The experiment is briefly described therein. After
of Co
-
Si
a
summary overview of the Co
in
phases
structure is
Si System with
epitaxial
the
chapter two,
-
presented in chapter three.
by
growth and the
MBE
on
In bulk form the monosilicide
Si(l 11) using
structural identification
a
are
thin
predominantly
hole
phase, however,
is
-
transport
CoSi2 template.
presented.
phases
following chapter four, Si(OOl)
Si(lll)
and
Si(OOl)
Substrates. Since
of thin films
are
imposes
quantitatively described
has
CoSi2 strain
on
a
~
overgrowth
striking
observable in UPS
-
like. This
stabilization and the
pseudomorph
phase
transition
energy calculations.
structural aspects of coherent
epitaxial CoSi2
films
were
CoSi2
grown
on
films
Si(lll)
lattice misfit of-1.2 % in respect to Si, coherent
the films. The relaxation of
on
through
epitaxial
on
and
growth
silicide films is
Si(l 11)
and
~
formation of misfit dislocations. This thickness 75 Ä
on
by RBS channeling have been carried out on silicides the
particularly
is the
the
with the model of Matthews. One of the main parameter therein is the
45 A for silicides grown
Mastering
called FeSi
paragraphs
important point
it is electron
given by total
some
discussed. Thin
critical thickness for strain relaxation is
is
so
Transport in the pseudomorph CoSi is
only metastable. Explanation for the
of CoSi and the isostructural FeSi In the
phase
adopts the
In the first few
The most
measurements.
like whereas in the bulk
structures
metallic, pseudomorph CoSi with
behavior to the isostoichiometric bulk stable 8-CoSi which is measurements and in electric
important crystal
stabilization of pseudomorph CoSi with CsCl
structure. For the first time I have succeeded to stabilize the
CsCl structure
some
growth
and
on
of thin, strained silicides is
with Si, which is documented in
optoelectronic applications,
Si(001).
chapter
Detailed strain measurements
heterostructures. a
prerequisite
for
a
successful
five. To be useful in micro- and/or
the heterostructures have to have
an
impeccable crystalline
ABSTRACT
4
quality. Growth
and structural charactenzation of such silicide
discussed in detail. The main
problem
of silicide
relaxation of the silicide. Since the wafers in at the
CoSij/SiOll)
grow with
a
interface prevent
could be solved at least for
Si(l 11),
A novel type of infrared
essentially metal
is
on
undoped
strongly by i.
an
e.
and hence
a
Si(OOl)
a
Even a
trapezoidal
detecting
sensing
and
to
problems
obtained.
CoSi2/Si/CoSi2 sensor
consists
The Silicon between the two
though Schottky fixed
the
were
six. This
detector, impinging light leads
energy band to two or
diagram.
even
barrier detectors
are
to four
widely
ränge. In contrast the device
ränge. With the
the choice of the metals, the cutoff energy in
externally applied
PtSi/Si/CoSi2
Schottky barriers.
depleted, yielding
variable
promising results
chapter
defects
growth. Thus the Si has
overgrowth. However,
no
are
with Si turns out to be the
coherent silicide
in the final
have the drawback of
in this thesis has
on
presented
photoemission processes.
nowadays they
presented
is
on
Silicon heterostructures
always slightly misoriented,
detector, fabricated with
the bias condition of the
different internal
used
Si(lll),
whereas
of two back to back connected
layers
Depending
on
are
strain in the initial stages of
compressive
heterostructures
overgrowth
practice
absolutely
-
tuning
PtSi/Si/CoSi2
bias between 0.8 and 0.4 eV. The way to
behavior
sensors
depending
could be tuned
truly wavelength resolved,
colored, IR imaging with existing Si Signal processing technology is open.
1.1.
1. Thin I spent
a
epitaxial
lot of time at the old
presented
in the
channeling.
following chapters,
terminology
some
University Zürich, and several
superlattices, Zn^Py
backscattering spectrometry (RBS).
measured strain of thin films with RBS
addition
de Graaff accelerator at the
van
of other materials with Rutherford
will be
5
films
films of Ni, Fe and Co silicides, SiGe
measuring thin
Epitaxy
Since
In
particular
I
results of strain measurements
new
I want to describe this method in
and concepts used in
samples
epitaxial Systems
are
some
detail. In
presented in
a
rather
short way.
1.1.
Epitaxy Two ancient Greek words
arrangement)
are
homoepitaxy,
on
over
on a
case
straining
the lattice mismatch r|
a,
van
real two
of
In the
are
case
of
a
Single
-
in alkali
are
identical, i.e.
film material is different frorn
heteroepitaxy by
occur
-
distinguished, namely
the lattice parameters match
and the Substrate as differ
occurs
in
one
growth.
perfectly
the other
on
and
no
hand, the lattice
small amount, which is indicated
of three distinct modes
islands coalesce to form
a
one
monolayer
In the Volmer
dimensional islands form frorn the
growth
first observed to
heteroepitaxy, where the
homoepitaxy
occurs.
dimensional
FM and VW
was
to^iC (taxis
to extended
where the film and the Substrate
der Merwe (FM) growth proceeds -
epitaxy. Epitaxy refers
and
by
according to
Epitaxial growth Frank
case
Si Substrate, and
parameters of the film
resting upon)
or
Century ago. Two types of epitaxy
a
the Substrate. In the interfacial bond
placed
top of a crystalline Substrate and
which refers to the
epitaxial Silicon
-
the roots of the modern word
crystal film formation halide crystals
(epi
erci
modes. The first few -
Weber
beginning of growth.
continuous film. Stranski
followed by the formation of three
-
-
as
at
dimensional nuclei.
time which
Figure
With
three
increasing film thickness (SK)
grow in
a
is
a
1.1.
corresponds
(VW) growth mode,
Krastanov
monolayers
a
is illustrated in
to -
the
combination of the
layer by layer
fashion
1. Thin
6
Frank
Figure
films
epitaxial
-
der Merwe
van
Three different
1.1:
Weber
-
the Stranski
before three
It is obvious that
one
-
-
-
Weber
layer by layer growth (i.e.
to
(VW)
mode to the
Krastanov
(SK)
dimensional islands
attempts
to grow
strength
dimensional
epitaxial film growth
and
governed by
lowest energy
principles,
on
growth
(VW)
are
the
some
predictions
when the misfit and the thickness of the
homogeneous
[2],
thickness
5)
the
as
hc
depends
on
-
that the behavior of the System is be made
can
[1].
by misfit dislocations
overgrowth
are
entirely
small
enough
accommodated
the
by
a
strain in the film
misfit dislocations
film whose
layer by layer (FM)
about the System
i.e. the misfit is
when the misfit and/or thickness is
a
growth
growth),
dimensional islands. In
-
mismatch at the interface is accommodated
is coherentX)
dimensional
formed.
assumption
2)
4)
of three
der Merwe
between the film and the Substrate. For two
large
and
two
Krastanov
-
films in the FM mode. This however
1)
overgrowth
van
-
mode the films first grow
the misfit r\ and the interaction
3)
Stranski
modes for thin films. Frank
growth
(FM) corresponds Volmer
Volmer
growth
are
is
large enough
both
homogeneous
strain
present
initially
coherent becomes unstable at
and misfit dislocations
are
film increases in thickness
a
critical
introduced
beyond hc
the
homogeneous
strain
decreases
Another films that
possibility,
adopt
a
crystal
which
recently
has attracted much mterest, is the
structure that is well lattice matched to the
differs from the
crystal
phenomenon
is
called
coherently
the Substrate. Similar to bulk stable
on
The film atoms
are
in
structure
that the film
normally
pseudomorphism [2]. Initially
registry with
the Substrate atoms.
would the
adopt
called
a
of thin
Substrate but which in bulk form.
pseudomorphic phase
films, there exists
(Also
growth
grows
critical thickness h
commensurate)
This
for
1.2. Strain of thin films
pseudomorphic films,
where misfit dislocations
elastic energy of the film. It is at a
a
corresponding
certain film thickness than that of the
he
thicker than
may
undergo
stabilization of pseudomorphic films
elasticity
continuum
phases phase)
as
as was
qualitatively
presented by Zunger bulk
(stable
for FeSi and CoSi
the total energy
be
for both
curves
the discussion of stability of the
new
film is
higher
The film is then in Films
epitaxial stability. favorable
more
phase.
The
understood within the framework of
and Wood
phase,
phase.
for
energetically
transition to the can
he
lower the stored
pseudomorphic
bulk stable
critical thickness
a new
phase
a
to nucleate in order to
that the total energy of a
possible
metastable State and there exists
begin
7
called
[3,4].
s-phase
If there and
are
two
competing CsCl
pseudomorphic
have to be calculated. This will be needed for
phases
phase (Chapter 3).
CoSi
1.2. Strain of thin films Linear
elasticity theory,
as
presented
thin film Systems, which is allowed
equilibrium
lattice parameter
usually elastically (a,,
-
af)
I a{.
distorted
Simultaneously
Poisson effect
[5] (Figure 1.2;
for small misfits r\. When
only
öy-grows coherently
parallel a
also
Figure
1.3 and
a
dealing
with
thin film with
Substrate with lattice parameter as, it is
strain s±
perpendicular
see
on a
interface, resulting in
the
to
is used in all theories
Appendix A,
in
(a±
=
-
af)
a
planar
strain
I af is introduced
by
s,,
=
the
B.l).
A a
a
f
-L.
V A
a s
Y
a s
Figure
1.2:
Schematic and
a
film
drawing of cell, which,
parameter a^ and the film cell in
Now
fl||
a
a
Substrate lattice unit cell with lattice parameter as
coherently
grown
perpendicular
equilibrium
and a± stand for the
to the
Substrate, has
a
parallel
lattice
lattice parameter a±. The broken line shows
condition with lattice parameter ar
parallel
and
perpendicular
film, respectively. For completely coherent growth
the
planar
lattice parameter of the strained
strain
|e,,|
is
roughly equal
to the
(n
lattice rnisfit t|
films
epitaxial
1. Thin
8
is defined relative to the Substrate lattice constant as, and
film).
lattice constant a{ of the relaxed
separately by measuring symmetrical only the trigonal/tetragonal1' s,
The value of et
strain et
=
=
be determined
(see
next
—
—
=
a/
depends
the bulk elastic constants
on
6,
(1 +>4)8||
=
to determine
asymmetrical Bragg reflexes,
and
can
possible
With XRD it is
-
a^ and
a±
whereas with RBS and
paragraph
£11
relative to the
e^
Appendix B).
(1-2)
ex
Ctj according to (1.3)
,
with
1
+
Cn
A
Cn
(111)
For
and
(100)
surfaces C
=
\Cn
+
=
2
C44- Cu
+
different thermal
strain and misfit
expansion coefficients,
1.3. Strain determination with RBS RBS is
probably
quantitative analysis
in
a
of the most
one
of
C= 0,
straightforward
years into
a
a
-
2 MeV
However, in the
one.
See also
parameters in general have
function of temperature.
channeling
frequently
the surface
used charactenzation
origin
in
region (10
crystalline
solid surface, the
are
case
techniques
Ä)
can
for
be determined
and evolved in the last few
In most cases, He
or
H
-
used. The method is often referred to of strain measurements
structure in
majority
-10000
particle physics
major materials characterization method.
bombardment alters the
bombards
near
way. RBS has its
in the ränge of 1
destructive"
are a
respectively [6].
compositions. Stoichiometry, thickness, depth profiles, crystalline
and strain of thin solid films
energies
C12 and
A for the definition of elastic constants. Since the lattice
appendix
quality
(1.4)
.
fC
+
an
irreversible
one
can
manner.
of the ions penetrate the solid,
beams with as
a
"non
-
observe that the
When
slowing
an
ion beam
down due to
ionization and excitation of the target atoms and momentum transfer to the target atoms. At the end of their ränge below the
thus modified
i)
Tetragonal
by
surface, the majority of the ions stop. The surface region is
the bombardment process, i.e. defect fonnation and
strain refers to distortions
lattice, whereas Bravais lattice.
on
(111)
implantation (first
order
along {100} axes since the deformation is towards a tetragonal Bravais trigonal since the deformation is rather towards a trigonal
surfaces the strain is
1.3. Strain determination with RBS
the ion
process). During
beam bombardment
-
a
again
from the solid
(scattering angle > 90°,
e.g. in -
[7]
name
and
energy ion
[8]
or
for
a
films grown
are
process).
ion
imposed by
hence tilted by
a
small
schematically shown in Figure
-
beam
scattering
quantitative analysis
the
For
Since the
more
analysis
information
in
[9]
see
and for low
[10].
Substrate with
on a
of the scattered ions will
scattering law,
recent review of new trends in
at surfaces in
lattice due to the strain silicide
some
Rutherford backscattering spectrometry (RBS).
scattering
Epitaxial
only
second order
process in this energy ränge follows the Rutherford
received the
9
minute amount of the ions will backscatter
from the target nuclei at and beneath the surface, and escape
channeling
a
lattice misfit t|, have
the Substrate. Off
angle
Ad from the
-
normal
slightly distorted
a
crystallographic
corresponding
axes
of the
of the Substrate,
axes
as
is
1.3.
PJ
SUbstrate
Poisson effect
film
o
Substrate
Figure 1.3:
Light
o
circles indicate the Substrate and black circles the film. Dotted circles
show the atoms of the silicide in their bulk
finite lattice misfit grown
coherently
effect leads to the
o
film
are
a
n
between film and
onto the Substrate
relaxation in the
equilibrium positions.
Substrate,
(arrows
to
the
the film is strained when
right side).
perpendicular direction.
therefore tilted by
a
small
Due to the
The
The Poisson
crystal
axes
-
of
angle Ad, which is shown
schematically for one direction. With normal
channeling
axes
measurements it is
of the film and Substrate
possible
separately,
to determine the exact
whenever at least
Substrate is different and the film thickness does not exceed
tetragonal/trigonal way
distortion st is connected with the
(see Appendix B) [11,12].
a
one
position
of off
-
constituent of film and
certain thickness. Therefore the
angular distortion Ad in the following
10
films
1. Thin
epitaxial
Aö
\ fe 0
As
=
long
the film, it
as
successfuUy steered
angle A& [7].
See for
primary beam
energy measured
width \j/1/2, which itself sure
steering
that
\
*
e,
sin(2 ÄS)
i
=
(1
ii) eM sin (2 Ss)
+
the incident beam is not too far away from the exact
be
can
sin(2ös)
-
for
example [5]
depends
can
using the
on
neglected,
be
into this an
a
channeling direction of
direction, thereby falsifying the misalignment
analysis
thick
NiSi2
crystal axis, I therefore
steering
of the
film.
and
on
effect
90.00° for
play
(111)
and
gave
angle
an
difference
(100) surfaces, respectively.
for films thicker than
100
~
In the
to
two
the
One has to be careful
the strain
was
strain value of 70.53° and
no
experiments
beam
compressive
strain
and
function of
[13]. Therefore
implanted
corresponding
to
ten
aecumulated. Results due to the short
dose
since the
came
particle bombardment
the
a
into
recent paper about
They
even
observed
a
change
of tensile strain into
possible
dose
the beam
damage,
the strain behavior of a
examined. The tilt dose
are
used
much
alters
strong relation between the implanted dose and
in
was
used, in order
angle decreased only
typical strain
therefore not corrected for beam
integration times
a
the smallest
assess
was
times
were
MoSi2,
Hardtke et al..
the beam influence. In order to a
steering
Ä which is the upper limit of film thickness for strain
doing strain measurements,
TaSi2
reported by
film, which within the
the microstructure of the strained film and hence the strain of the film. In on
Symmetrie
He+.
measurements with 2 MeV
strain measurements
the Channel
on
the incident beam energy. To be
usually measured
equal
function of
as a
Steering depends
Channels with respect to the surface normal of the Substrate beneath the
experimental uncertainty,
(1.5)
to minimize
CoSi2
10 % after
a
as
dose
had
been
since the Statistical
errors
measurements
damage
film
larger.
1.4. Strain relaxation In
a
thin
film, which has
is stored when it is grown
E
=
The
growth
is
a
coherently onto the
\C\ [sie + zjy el) 1
+
interesting question possible. If
sufficiently small, the
lattice mismatch r\ relative to the
+
Substrate, elastic energy E
Substrate.
Cn (Sx* e^ + £yyezz
+ ea
s**) +
\C*a {£%, s£ e£J +
(1.6)
pseudomorphic
film
+
is now, up to which thickness coherent
or
the film thickness and the misfit between the two lattices
coherent State is indeed the
energetically
were
both
most favorable. Theoretical
1.4. Strain relaxation
[1,14,15,16]
calculations
based
dislocations and the strained critical thickness
introducing dislocation its
so
theraiodynamic equilibrium
epitaxial
film
predict
that coherent
hc is reached. Beyond this film thickness,
called
density
equilibrium
on
misfit dislocations,
appendix
C. In the
system is
not
case
always
grid
takes
of misfit
place
until
films reduce their elastic energy
totally eliminated
bulk lattice constant. The critical thickness the model
growth
a
a
by
which relieve part of the misfit strain. The misfit
increases until the strain is
roughly estimated using
between
11
proposed by
hc
Matthews
[3,4].
can
grow with
and the strain relaxation
[14] which
of pseudomorphic films the Situation is
in the State of lowest energy
and the film
more
can
be
is summarized in
complicated,
since the
2. Co
12
2. Co
-
Si System
Si System
-
In this work
properties
epitaxial
films behave
diagram
of Co and Si is
of
epitaxial films
differently
than the
same
of
and CoSi
CoSi2
material in bulk
are
presented. Although
form, the binary bulk phase
reproduced here [17].
1600
1400-
1200-
ü o
1000-
2
800Q.
E CD
600-
&
400.
200
Co
60
40
20
atomic percent Silicon
Figure
Binary bulk phase diagram of Co and Si [17].
2.1:
I fürst cite the metallic
CoSi2,
since among all the
the favorite because of its very low
ability in
to grow
Figure
2.2
This bulk
-
epitaxially
a).
on
Si
[18].
The unit cell is face
phase
Epitaxial growth
has
a
specific
-
It
electrical
crystallizes
refxactory silicides, CoSi2
seems
resistivity (~
RT)
in the cubic
centered cubic with
good temperature stability with
of coherent
CoSi2
onto
Si is
15
p.Qcm
at
and its
lattice structure, shown
CaF2
a
lattice parameter of 5.364 Ä
a
melting temperature
possible
to be
[19].
of 1326 °C.
because of the moderate lattice
mismatch of-1.2 % relative to Si. The other to
as
FeSi
phase
structure),
formula units and has the rocksalt structure
I want to mention is CoSi with the cubic B20
which is shown in a
Figure
2.2
lattice parameter of 4.447
by
considerable
b).
A
primitive
Ä [19]. This
-
structure
(also
referred
cell contains four CoSi
structure
can
be derived from
displacements along [111] directions [20]. Precipitates
2. Co
of
with
epitaxial CoSi(lll)||Si(lll)
-5.6 % relative to
will in the Si System
be referred to
lattice mismatch of
a
as
8-CoSi
according
to the
phase
nomenclature of the Fe
-
[22].
2.2:
The cubic cobalt sihcide FeSi structure and
size
was
structure
chosen a
The third
phase
existent in the bulk
representation
of
a) CoSi2; CaF2 structure, b) bulk e-CoSi;
c) pseudomorph CoSi; to the
equivalent
CsCl structure. In
shown in
Figure
2.2
Since
to half the Si lattice
a
c),
a)
and
of the
simple are
the CsCl
unit cell of
constant,
-
cubic
primitive
(CsCl)CoSi
the cell
phase
of the
(CsCl)CoSi
eight
unit cell
indicated.
unit cells
has
are
monosilicide, is a
not
lattice parameter of
shown in
the strong similarities between the CsCl and the
visible. The lattice mismatch of the
c)
Si unit cell size. However, for the FeSi
equilibrium lattice parameters
phase diagram.
Ä which is close
phases
simplified representation
is shown Bulk
In this
13
Si System
Si, have been reported by D'Anterroches [21]. The monosilicide bulk phase
following
Figure
2.74
and with
CoSi[-110]||Si[ll-2]
-
CaF2
Figure
2.2
structure
c). are
with Si is +0.9%. As will be outlined below,
14
this
2. Co
phase
can
literature this
-
Si System
only
be stabilized
phenomenon is
by
called
coherent
growth
of thin CoSi films onto Silicon. In the
pseudomorphism [2].
3.1. Introduction
3.1. Introduction
pseudomorphic phases, intriguing, since
of
pressure may
phases bcc-Co
are
on
only
is
can
positive. Being
with the Substrates
measurements
[26] along
kept
stabilized
(DOS)
a
EF
small gap at
adopts
a
and
Fe 45
Ru
Detail of
periodic table, showing
form of thin
epitaxially
[27] a
are
known
density
reported [28].
Pt
only for Ru, Rh,
MBE
on
of
called s-FeSi, for which
78
stabilized films.
of
showed that
low
the position of group VIII transition metals.
Monosihcides with CsCl structure in
Mäder
Pd
lr
these
Electrical transport
metal with
phase,
order
epitaxial Fe, xSi
or
46
77
Os
[25].
by
effect
epitaxial growth
Ni
Rh
76
are
28
Co
same
epitaxial interfaces,
behavior has been
27
44
same
thereirorn, when grown by
nonmagnetic
magnetic
properties.
contrast to the external
hcp structure,
in contrast to the stable bulk
26
Figure 3.1:
the
low temperatures
sufficiently
EF
material. The
examples
with ab initio band structure calculations
at the Fermi level
there is evidence for
a
favorable
discovered
Recently
stoichiometric FeSi with the CsCl structure is states
by
structures derived
or
at
of
since the material is strained in the
which in bulk form
with the CsCl structure
structure
form, is particularly
with unusual
both, negative and positive, in
be
The existence of
in the past few years.
synthesize materials
to
epitaxially,
in thin film form.
GaAs(llO) [24],
crystallizing Si(lll)
normally
stable
have been of great interest both
by epitaxy
change the crystal
films
strain
magnitude. Moreover,
pressure, which
possibilities
new
by growing
be achieved
15
Si(l 11)
on
i.e. of crystal structures which do not exist in bulk
it offers
Externally applied
motivation
motivation
experimentally [23,24]
and
theoretically [2,3]
-
phase
-
transformations induced
phase
Structural
can
CoSi
epitaxial
3. New
-
Os and for Fe
16
monosilicides known to
do
so
below Fe and Co, s-CoSi
(Figure 2.2).
structure has
has
in the
might
never
already been
special
exist
as
[29].
periodic
table
that
speculate
observed
as a
only
those of Ru, Os and Rh, and
they
Note that Ru and Rh
(see Figure3.1).
perhaps
well. To the best of my
reported before. Epitaxial
been
are
since the
interesting,
an
epitaxial
knowledge,
right
situated
e-FeSi
as
CoSi
CoSi
a
are
Further bulk stable
FeSi structure (Pearson symbol cP8)
to
us
this structure
conditions
in the
same
These facts led
CsCl structure
FeSi with the CsCl structure is
respectively,
crystallizes
Si(l 11)
on
crystallize with
under very
only
phase
-
epitaxial
The existence of
seem to
CoSi
epitaxial
3. New
phase
phase
CoSi with FeSi structure
[17,30] with the
with the CsCl
the other hand
on
precipitate during CoSi2 formation by solid phase epitaxy [21].
Results of the first Observation of CoSi with CsCl structure
[31] will
be described in the
with
unintentional wafer
following.
3.2. Growth Si(lll)
(n
Substrates
doped,
-
misorientation of less than 0.3°
System oxide 1000
as
was
desorbed at
MBE System
can
~
840 °C while
was
routinely
by depositing
were
1-2
grown
an
introduced into the molecular beam
exposing
an
annealing
to
outgas the wafer, the native
the wafer surface to
weak Si flux. Then
a
preparation
by
MBE
on
CoSi2 templates.
Co and Si
RT. The results
structural
quality
Epitaxial growth diffraction
were
of pure Co at
at RT followed
when
growing
of the silicide
(RHEED).
With
Kikuchi pattern which
disappearance
stoichiometrically
essentially
the
room
temperature (RT)
by an anneal to
same
temperatures between 100
-
200 °C
or
below, the pseudomorphic CsCl phase
well
as
the
by
templates
made
reflection
was
3 -11
kept an
~
-
close to
energy electron
slowly
175
by
improved
as
well A
annealing (CsCl)CoSi
them thicker than
Ä thick
by codeposition.
films thinner than ~90Ä.
observed when
growing was
for
high
by depositing
then grown
were
templates except
film thickness the spots faded
merely visible
of the RHEED pattern
or
a
with the Substrate
for both kinds of
templates
or
forming
350 °C
~
template
monitored in situ
increasing
was
onto the
onto thick was
as
a
The latter have either been
silicide. Monosilicides with film thicknesses between 25 and 200 Ä
codepositing
epitaxy (MBE)
[32].
monolayer
stoichiometrically
Qcm)
grown. Details about buffer
be looked up, e.g. in
Thin CoSi films
Co and Si
1200
-
received from the manufacturer. After
Ä thick Si buffer
grown
were
800
as
the
sudden films at
Ä. As will be shown
then transformed into the bulk stable
8-phase.
17
3.3. Structure
3.3. Structure In
typical
observations, the intensity of the spots and the Kikuchi pattern
RHEED
of
rather low from the
beginning
RHEED
therefore, did
panorama,
growth
the
yield
not
and
(CsCl)CoSi
of
complete
structure
reaction
patterns of
a
the
phase
44
was
as
to the bad
stability
Ä thick (CsCl)CoSi film identified
CoSi(lll)||Si(lll) way
or
and
to
are
cubic
have
CoSi[ll-2]||Si[-l-12].
underlying CoSi2 template,
the
of this
phase.
shown.
In
symmetry and This
means
i.e. is rotated
by
with
The
which leads to
with TEM and x-ray
epitaxial
an
analysis
orientation
that CoSi is oriented in the 180
a
3.2 characteristic RHEED
Figure
Together
analysis
Information.
deposition
Observation of faint RHEED patterns is either due to the RT
incomplete
structure
a
were
of
same
degrees around the Substrate
normal relative to the Si Substrate.
Figure
3.2:
Typical and
RHEED patterns of
b) [11-2]
azimuth.
pseudomorph (CsCl)CoSi
(W1232,
after 44 Ä
on
Si(lll) a) [01-1]
(CsCl)CoSi
with 7Ä
CoSi2
template)
Figure film in
on
3.3
a)
shows
a
top of a 10 Ä thick
Figure
3.3
b).
low resolution
cross
CoSi2 template.
The
It is indexed
according
to
a
-
sectional TEM
image
of
a
70
Ä thick CoSi
correspondmg diffraction pattern
is
displayed
cubic unit cell with the lattice parameter aSl of
Si,
3. New
18
epitaxial
CoSi
with the result that all odd
CsCl structure and resolution
image
the
zone
[1-10]
Figure
3.3
d).
Figure
-
-
phase
on
order reflections of CoSi
are
lattice parameter of the film close to
a
and the
axis. An
corresponding Computer
image
Low
a)
CoSi2 template
Simulation of the CoSi with the
be
clearly
seen.
cross
can
-
sectional
TEM
region. All spots
Si(l 11). b) Selected
in
the leftmost column
area are
on
of
-
type
onentation
corresponding zone
axis.
With x-ray diffraction
-
(CsCl)CoSi. c) High
were
70 Ä
high
-
along
displayed
in
thick
10 Ä thick
a
according is
to a
consistent
resolution
-
layer,
taken
cubic unit
image
along
the
with and
[1-10]
resolution image of the interface region between the
and the Si Substrate, taken
(XRD) (Cu K^)
observed for FeSi with the CsCl structure
indices
taken
layer, is
a
diffraction pattern of the silicide indexed
Computer Simulation of the CoSi
d) High
CoSi2 template
of
a
top of
cell with the lattice parameter of Si. The diffraction pattern B
c) shows
CoSi2 template
image
CoSi film with CsCl structure, grown on
3.3
asJ2. Figure
region
resolution
pseudomorphic
absent. This is consistent with the
of the interface
There the type B interface
3.3:
Si(l 11)
found to be absent, when
was
along the [1-10]
the found
same
zone axis.
reflection pattern
[25,33].
indexing according
as
previously
All reflections with odd order
to a cubic unit cell with the Si
3.3. Structure
lattice parameter. This is consistent with
of
~
±
thickness was
the CsCl structure and
With XRD the lattice parameter and the Poisson number
aSi/2.
be 2.74
assuming
0.02 Ä and
hc
are
v
under
=
a
were
lattice parameter
a
further determined to
0.32, respectively. Since 2a0 is larger then aSi, films below
compressive
biaxial strain
measured to be at maximum 2.80
±
By
e».
XRD and RBS the
trigonal
1.5 %
were
of comparable miscut, the strain is constant for films with thicknesses up to 100
larger.
the critical thickness for strain relaxation to be wafer surface and
consequently misfit
step in the
CoSi2 template [34].
deeper)
strongly
in
strained
misfit of -1.2 %, i.
e.
a
Moreover RBS
strain
relieving misfit
a
template
coherent
dislocations
phase
are
a
template.
growth of pseudomorphic
to 100
appropriate spacing
electron diffraction where
by
topographic
~
a
of the
Debye
-
175
CoSi with
dislocations. Since small wafer
good crystal quality.
for film thicknesses
phase
exceeding
transition could also
Ä. This transition
Scherrer
Ä
lattice parameter of 4.44
STM exhibited
a
pronounced change
contrast due to buried Si surface
step
at the
a
condition for
was
observed by
RHEED pattern. Additional confirmation for the transition to e-CoSi
rapidly vanishing
observed
relieving
200 °C. The
-
growth
monolayer Steps [34],
e-phase
transition to the stable bulk
a
pronounced (i.e.
were more
The best
without strain
triggered by growing the CsCl phase thicker than
obtained from the
Ä, indicating
Si, relaxation of the CoSi2 template
at interfacial
present
20Ä occurred upon annealing the CsCl phase be
wafers
films, which indicates better crystal quality. Since CoSi2 has
miscut is therefore necessary for A first order
on
found
formed at every monolayer
are
channeling dips
lattice parameter smaller than
CoSi is hence
strain 8,
The wafer misorientation leads to
dislocations
increases the mismatch between CoSi and the
pseudomorph
critical
only
in films with the unintentional wafer misorientation smaller than 0.1 °. In addition
stepped
a
0.30 %. A strong correlation between strain and wafer
prevails. Trigonal strain values significantly higher than
misorientation
19
as
rings observed by
was
derived. Film
transmission
morphology
well, manifested in
phase
was
a
as
vanishing
transition.
3.4. Electronic structure The
(hv
=
angle
21.2eV)
indicating
the
integrated
(CsCl)CoSi,
for
phase
shown in
to be metallic. The
of the spectrum with
replaced by peaks
ultraviolet
a
broad metal 2>d
photoelectron
Figure 3.4, exhibits
phase -
transition is
band emission
(UPS)
spectroscopy a
distinct Fermi
accompanied by
peak
at 1.1 eV for
located at 0.3, 1.1 and 2.5 eV for e-CoSi
a
spectrum -
edge,
striking change
(CsCl)CoSi being
(see Figure 3.4).
These three
20
3. New
peaks
were
Repeated
measurements
For
are
CoSi
-
s-CoSi
on
surface effect.
a
in agreement with
to 600 °C for about 5
minutes)
and
The
Hamann
[37,38,39,40]
Si(l 11)
[37]. is
fair,
as
same
is also included
as
crystalline quality
are more
broadened than in the
spectral features
the
film after transformation to
together
between
of the monosilicide
density
for both monosilicide
CoSi2
films
by
[32].
0
-5
have been carried out
-10
a
a) UPS spectrum (hv structure grown onto
bulk stable e-CoSi
transformation to
capped
with 2
=
a
energy
21.2 7
eV) of
a
CoSi2 (for
monolayers
at 250
Calculated DOS of the monosilicide
phase
°C for 2
better temperature
Si before
[37].
after the
annealing
Due to the
[eV]
transition to the
minutes)
stability,
and after
the silicide
to 600 °C for 2
phases by Miglio [41]
was
minutes). b)
and
the
by Miglio [41]
A thick CoSi film with the CsCl
ÄCoSi2 template,
phase (annealing
Mattheiss and Hamann
44
CoSi2
parametrized
0
-5
binding energy [eV] Figure 3.4:
for
quantitatively understand
(Figure3.4b)).
-10
Mattheiss
in the UPS spectra
(DOS) calculations, using
phases
of the
CoSi2 (annealing
theory
phases, their peaks
In order to
of states
and
experiment
at
phase [35,36].
with DOS calculations
with other UPS spectra of
CoSi2 spectrum.
observed features,
binding model,
pronounced peak,
phase.
pronounced peak
UPS measurements of the stable bulk
correspondence
well
therefore characteristic for this
are
with Si evidenced that the
from this
reported
poor
experimentally
capped
Apart
relatively
-
on
the UPS spectrum of the
comparison
tight
phase
present in all e-CoSi films, and
0.3 eV is not
8-phase
epitaxial
CoSi2 by
3.4. Electronic structure
of the main features of his calculations
positions
The relative
measurements, but
a
discrepancy of nearly
already reported
monosilicides. This shift,
up to date. This is
theoretically
is
(CsCl)CoSi
a
embarrassing
more
even
agreement between theory and experiment that
for the Fe silicides
was
present in the
are
position
0.8 eV in the absolute
[25],
21
is existent for the
is not understood
CoSi2 always good
since for
found. However, the calculations show
metal since the Fermi level is located in
a
region
of rather
as
well
high density
of
metal d states. has
CoSi2 antibonding
a
calculations of RuSi and RhSi
stabilizing
other hand
Ev
is located
states filled up to
(Figure
right
assuming rigid
understood
b)
3.4
-
band
This
can
a
CaF2
of
structure
peak
a
due to
non-bonding
conditions, where Rh having than Ru. In fact,
(CsCl)FeSi [27]
E¥
to
lies in
Ad states
peak [42].
one
shows that the
electron
to be
magnetic [27].
(CsCl)FeSi
the
with
a
3.5.
maximum thickness of < 175
on
the
also be
can
than Ru has
the DOS of
(CsCl)CoSi
position
of E? is
displaced states
phase [39,40,42]
to
(see
and it
This latter Statement is somewhat
spin polarized
bands should be
pseudomorphic (CsCl)CoSi phase should (100-200°)
be less and the
found to be stable at least up to 900 Ä in contrast to CoSi
Ä).
Stability The
stability
discussed in the
Si(lll)
with
around the
CoSi2
was
of low
For RhSi
strong peak representing nonbonding metallic 3 d
likely
(FeSi
region
more
stable which indeed is manifested in the low transition temperature of thick films
[24].
a
This shift of EF
comparing
delicate since in fact the band structure calculations with
instability
and
DOS
CoSi2 [37,39,40].
be related to enhanced destabilization of this
indicates that the material is
performed. Compared
EF separating bonding
near
Jvanovskii showed that in RuSi
higher energies
and falls in
the
in the middle of the DOS
with the DOS of
higher energies Figure 3.4).
by
(DOS) just below
of states
of states
sp3 hybrids and metal 3d states. This quasigap was claimed by Tersoff
states of Si
and Hamann to be crucial in
density
density
in the
'quasigap'
a
of the CsCl
phase
following. The CoSi bulk phase
lattice mismatch of
[111]
formation
of CoSi and FeSi
direction i.e. with
precipitates
-
could in
5.6 % when the
CoSi(lll)||Si(lll)
(CsCl)CoSi
principle
epitaxial and
grow
coherently
orientation is rotated
by
is on
30
°
CoSi[-110]||Si[ll-2] [43]. During
with this orientation have been
The lattice mismatch of pseudomorphic
proposed by Miglio [41],
as
reported by
with Silicon
on
D'Anterroches
[21].
the other hand amounts to
+
(relative
0.9 %
CoSi
epitaxial
3. New
22
phase on Si(l 11)
Thus the
aSi/2).
to
-
epitaxial stabilization of the CsCl phase is mainly due
the fact that its lattice mismatch is much smaller than that of the bulk a
look at the Situation of FeSi
because the former has rj
=
phases
which
lattice parameter
the
one
[41].
For
prevented by
contribution for in the
case
(CsCl)FeSi from
of FeSi
a
internal
kinetic
the interface
bonding.
a
phase
in both Systems. But
showing
structurally
pseudomorphic
that
stable up to 500 °C, apart
towards
with
FeSi2
increasing
very similar to that of CoSi
to take into consideration energy contributions due to
Silicon Substrate atoms has to be included. Since
the energy of
the energy of the
is increased
by assuming equal
phase 15
energy than the
State. This interface
Figure
higher than
3.5
by
the CsCl
by twice
energy per bond for the NaCl and
phases (for
in the coordination number of these two
Silicon
phases.
matching
This corrected
open circles. For the FeSi this energy at
phase
which then renders the CsCl
is stable up to the critical thickness
he,
which is
equivalent
phase to the
Ä [44]. Above this thickness the CsCl phase is only metastable
be transformed into the stable
thickness h.
form is
into account elastic energy
is the most stable
stoichiometry
epitaxial
s-phase
registry,
experimentally observed
higher in
only taking
diagram of FeSi looks
for the s-CoSi is shown in
stable. Thus the CsCl
can
0.75 eV per formula unit for CoSi and
s-phase only
the Silicon lattice parameter is
and
has
function of relative difference from the Si
and hence divides the energy difference of these two
curve
phase
of nearly all atoms is observed and for the coherent
condition) by the difference energy
bulk
corresponding
could be established
of the
broken bond. He estimates this energy
phase
as a
s-phase
perfectly match the
phase registry
of three atoms is in
CsCl
(CsCl)CoSi
He argues that the energy of broken bonds at the interface from atoms of
the silicide which do not
for the CsCl
stable than
turn cannot be understood in terms of elastic energy contributions of
(Figure 3.5). Miglio then proposed
one
3.5
phase diagram
since the energy
only,
more
et al. calculated the total energy of the CoSi and
energy. When
gradual change
a
temperature [44]. This in
interface
2 % and the
films with thicknesses below 15Ä remain
exhibiting
is
of the NaCl structure both in bulk and
coherent interface, the
a
+
(black Squares) by about
occurrence
high
its
Figure
=
(CsCl)FeSi
having
coherency, the energy of the s-phase (black circles) is lower than
of the CsCl structure
0.54 eV for FeSi. The
that
surprised
topic, Miglio
shown in
are
is
lattice mismatch r|
4.7 %. To elucidate this
-
FeSi
a
one
But when
phase.
to
s-phase.
bonding
Thus
s-phase. even
In the
case
the thinnest
energy term decreases
of CoSi the CsCl
(CsCl)CoSi
films
roughly with h
"'
phase
are
for
in
a
is
always
metastable
increasing
film
3.5.
i
-5
i
i
i
i
l
i
i
i
i
l
i
i
i
b) CoSi
a) CoSi
ANaCI
Stability
A
A ~
"6
\.
h
ih
i
m
CsCl
11
-7
%w^ -8
-8
uul
-10
»
'
""!
i
i
iiiml
(a-asi)/asi
iniiJ
104
103
102
101
i
h[Ä] b) FeSi
10
(a Figure
3.5:
0
-5
a)
-
aQi)
Total energy
CsCl
/
a
curves
for CoSi and FeSi with NaCl
(filled Squares), and the E-phase (filled circles)
lattice parameter,
Open
5
expressed
phase
as
a
with respect to
(CsCl)CoSi,
contribution is taken into account. per formula unit of e and CsCl
as a
[41].
monolayer limit of the
when its unfavorable interface
b) Qualitative Variation of the
phases
the
function of the
in terms of the relative deviation frorn a^
circles indicate the total energy increase in the
E-CoSi
(filled triangles),
function of the
total energy
layer thickness h.
23
3. New
24
On the
epitaxial side
right shown
schematically
CoSi
as
of
is about 70
hs
can
Si(l 11) the total
3.5
s-phase.
energy of the
Ä for FeSi [45] and is
not accessible to
to the
can
which
reasoning gives
enormous
At
a
misfit
for CoSi. Due to
by
by undergoing
thermal treatment
very thin
(CsCl)CoSi
temperature of 180
verify
higher temperatures than of elastic and
stability
-
(< 10Ä)
films
200 °C
thick films.
thermodynamical
of pseudomorphic
have been annealed
energy
(CsCl)CoSi
immediately
weak 2x1 surface reconstruction
a
270 °C. With UPS the
~
the structure of this
favorable
phase,
possible.
The much
a
phase transition larger
phase
polycrystalline
appeared
after
which
While the electrical
Since the
from coherent
larger
with
s-phase
(CsCl)CoSi
is the
film
thickness
energetically
are
most
to coherent s-CoSi should be
phase
(>20Ä) always
may
probably
transforms
to
magnetic properties
resistivity
Systems [26], CoSi with the CsCl
resistivity,
average
of thick FeSi films showed
structure is
Hall effect and
technique [46]
photolithography The
phase.
s
s-CoSi.
3.6. Electrical and
der Pauw
and
could not be identified to be
lattice mismatch relative to Si of the latter
explain why (CsCl)CoSi
Electrical
a
by
or
CsCl, because of enhanced template and Substrate contributions. Further experiments
needed to
K.
complex interplay
a
evidence for the bad
remained up to temperatures of or
by
hm
difficulties I had to grow and characterize it.
Recently growth.
are
films. Transformations have been observed for temperatures between 100 and
Hence the kinetic barriers must be
the
s-phase
metastable at every film
are
which may be induced
s-phase
experiment
lower its total energy
200 °C while thinner films tended to be stable up to
terms. This
CsCl- and
The critical thickness of strain relaxation
be grown. However, the System
phase transformation growing thick
Figure
on
pseudomorphic (CsCl)CoSi films,
kinetic barriers,
thickness,
phase
function of film thickness. Indicated is the critical thickness
a
where films transform to the dislocation
-
and wet
-
residual
or
by
resistivity
increasing temperature (see Figure 3.6). -
1200
Qcm)
-
4K
at ~
behavior
magnetically
were
measured
terminal method
etching, in
150±30 uQcm. Until the temperature of
doped Substrate (800
to order
magnetoresistance
the four
chemical
likely
a
on
typical
as
by
for Kondo
explained
above.
the Standard Van
structures
defined
by
the temperature ränge between 1.4 and 300
of
60 K the
(CsCl)CoSi films
scattered
around
resistivity increases roughly linearly
with
Above this temperature electric conduction of the low
leads to
a
decrease of the measured
resistivity. However,
3.6. Electrical and
the temperature
dependence
below 100 K is rather small for conventional electron
(linear increase)
contribution
and
electric transport is determined -
or
by
surface
a mean
primarily by
path
free
for Kondo Systems
and electron
[47]
residual
at
magnetic
resistivity
at low
lattice
-
phonon at low
constant0,
crystal
rather than
moments is
responsible
temperatures (i.e. 1.4 K)
electron interactions [48],
-
25
resistivity
one
the structural disorder of the
increase of
no
a
of the order of
scattering. Possibly scattering
for this effect. In contrast to FeSi
[26], typical
real Saturation towards
no
temperatures is observed. Because of
by phonon
magnetic properties
was
observed in
(CsCl)CoSi (see insert of Figure 3.6).
202
o
I Q.
200
0
20
60
40
80
100
T[K] Figure 3.6:
Typical temperature dependence (CsCl)CoSi, measured is
due
1200
Evidence for observed at low
could
by
changes sum
a
magnetic ordering
Figure
materials the Hall
second linear rise with
a
force, and
a
second term a
resistivity pH
smaller a
slope
behavior
corresponding
800
pronounced
-
at
measurement
increases
higher
setup,
on a
=
R0B
+
no
hysteresis
96 Ä thick film at
linearly at
small fields and
fields. It may be written
expected from to the
anomalous Hall effect
a
simple
as
the
consideration of
spontaneous magnetization of the
strongly temperature dependent proportionality p„
determined
(substrate resistivity:
3.7 shows the Hall resistance measured
material. This latter term has
"
stems from the
of the normal linear Hall effect with
the Lorentz
conduction
60 K
~
Qcm).
ferromagnetic
to
Substrate
increasing
to
resistivity of coherent
96Ä thick film (W1232). The drop above
temperatures. Within the accuracy of the
detected.
4.2 K. For
at a
of the electrical
Ri M.
assuming the Drude model for one conduction band, i.e. p0
factor Rx
[48]. (3.7)
=
mvf/ne2X
3. New
26
Such
an
epitaxial
CoSi
phase
-
anomalous Hall effect is not restricted to
appear in any material where there
Figure
3.7
bears
a)
since there
two
are
regions with
normally
is attributed to
in
case
unit, thus indicating (CsCl)CoSi the
phase transition
deviates
temperature
at B
>
T)
2
to be metallic. as can
pseudomorphic phase,
be
which is
investigated by
XPS after
in
seen
(s)
material
Since surface
have been
they
Figure
probably
[50].
scattering
way
effect
was
with
a
of nonmagnetic
variations in
(RT codeposition
same
positive
and
füll agreement with
change
published
consists of three nested hole one
sitting
in
could argue that the
inhomogeneities
center can
anomalous Hall effect.
The carrier
Si02
normal
of Co
naturally
-
Although the
CoxOy.
changes sign after
the
bcc
be
-
only
the
slightest
on
days.
on
films
bare surfaces
It is
was
clear that
quite
Thus surface effects
can
be
caused
by incomplete
grown
exactly
were
in the
The Hall
deviation from
relation)
a
linear
cm"
in the temperature ränge between 4
sample
to 600 °C for 30 minutes. This is in
-
CoSi2
in
zone
also
Figure 2.2).
excluded and
are
the Fermi surface
center
This
by
[37].). Similar excess
kind
crystalline quality
to
Co atoms
of structural
probably responsible
density obtained assuming a one band model, i.e. RH l/ne. Despite the high Substrate resistivity, Substrate conduction is the major effect -
was
virtually
a
repeated
were
Co Clusters may be formed
better better
like which
stoichiometry adjusted.
at the Brillouin
(see
-
In
magnetic surface oxide
impurities
theory [37,49,50,51] (i.e.
cubes
not
the
of 1.810
pockets centered
magnetic
hole per formula
magnetic impurities.
a
air for several
and not of
with
(not
annealing
data and
by
(CsCl)CoSi, CoSi2 films
density
after
due to
effect, the experiments
exposed to
shown in
The Hall effect in s-CoSi
b).
however. Further the oxide
template)
thin
completely
constant carrier i.e. hole
and 50 K2) and did not
FeSi
on
increasing
sample
one
the Hall effect
induced
In order to estimate the effect of magnetic
stoichiometry
in
curve
contrast to the isostructural bulk FeSi
excluded for or
of the
The latter bulk metal has also
mainly
reaction
density1*
3.7
the oxide consists sure.
also
magnetic materials,
electric transport in s-CoSi is electron
yielding identical results,
with Si,
can
positive indicating predominant
equivalent to
Interestingly,
rise to similar anomalies in the Hall
might give
The hole
independent diamagnetic susceptibility in
paramagnetic [28].
which is
capped
-
The
of the Hall resistance with
dependence
CoSi2 [26,49,37].
to be the case in bulk
reported
[48].
moments
to the anomalous Hall effect of
little from normal behavior which is
a
contrast to the
also
s-CoSi,
to
magnetic
but
strong temperature dependence of the Hall effect, which
a
5-1022 cm"3 (determined
3.7 is
localized
ferromagnetic materials,
As in FeSi the Hall constant is
Rv
is the
as
large,
linear
a
field. Indeed I observed
hole conduction
are
relationship
close
a
magnetic
Figure
Si(l 11)
on
for the
of (CsCl)FeSi in contrast
=
at
temperatures above
~
50 K.
3.6. Electrical and
to the
which
(CsCl)CoSi
anomaly
evidenced
was
of (CsCl)CoSi films is much
0.6
more
experiments,
XRD and RBS
by quantitative
27
magnetic properties
the Hall
pronounced at any temperature.
Ä CoSi(CsCI)
96 0.4
1
-
E o
a 3.
X
Q.
-1
-
-2 -4 -2
2
0
6
4
10
8
B[T] Figure
Hall resistance of
3.7:
of
a
(CsCl)CoSi
a
b)
and
effect is
typical
sign of the Hall
Further the
s-phase
much smaller than in
for
magnetic
magnetic
constant
(CsCl)CoSi
that it is due to localized structure of the CsCl
with the
magnetic
field
temperatures below 40 K with the second
hypothesis,
magnetometer
evidence of magnetic
a
i.e. that
measurements
as
Curie
unity,
magnetic
classical
-
a
well
as
proportional
magnetoresistance,
anomaly
CoSi2
is
Brillouin
a
in
2.
which
is
(CsCl)FeSi
films grown with
equal
might
be
bulk effect, thus relied with the band
magnetic
a
by
Madar
ferromagnetic hysteresis
dependence. material.
at
Thus I tend to favor
Unfortunately SQUID
scattering measurements
failed to
give
direct
films.
at 4 K of all three Co silicide
to B
anomaly.
the Hall
behavior is not yet clear. It
Weiss temperature
ordering in (CsCl)CoSi
fields
it is
on
cause
results of magnetic torque measurements
(CsCl)CoSi
Perpendicular magnetoresistance for low
magnetic
or
to e-CoSi the
electron like conduction.
to the surface showed indeed
parallel
experimental setup
be excluded to
can
For
anomaly in the Hall
to bulk e-CoSi. The Hall
Clusters
phase. Preliminaiy
different
phase transition
and is not observed
magnetic
measured at 4K.
10 T. The
to
changes, indicating
equivalently
a
materials. After
conditions. However, the nature of the
growth
A thick (CsCl)CoSi film (W1202)
(W1209),
fields up
magnetic surface effects behaves
96
second Hall measurement with
carried out with
was
In conclusion
a) pseudomoiphic
70 Ä thick e-CoSi film
Since the occurs
product
phases
on
surface reconstruetion in contrast to the
different surface reconstruetions
2)
3.5°
surface, which transforms into the Si-rich surface
formation when
(V2 W2)R45
reconstruetion of the Si
shows
but
0C
ky
when
=
-
-k_
k(EF+0-hv)
sectional view of escape
The inner
energy E
Ef-
=
cone
model for
The hatched
conservation of parallel
The internal quantum
yield YmX
will escape and is determined
volume of the
Since the
spherical
the internal quantum
by
momentum
shell of
modified Fowler equation
the Fermi surface
photoexcitation
with
by
defines the emission cap obtained with
k^.
is defined
as
the
probability
that
a
photoexcited
carrier
the ratio of the number of escaping carriers to the number of
density
of states is uniform in
cap for escape to the volume of the
yield YmV
emission shown in
indicates energy values accessible
area
area
Schottky
sphere represents
sphere indicates the spherical
Av. The dotted
photoexcitation.
photoexcited carriers.
cap
lK
momentum space for T= 0 K. The outer
k(Ef).
the
(6.5)
l spherical escape
of excitation
Cross
cross
0.
spherical Shell
6.3:
they
0.
k(EF-hv);
Figure
Figure 6.3.
into consideration the band structure of the metal and semiconductor and the
assumption that interface,
electrons to
61
This ratio is
reciprocal
spherical
space, the ratio of the
shell of excitation
given in fürst order of hv/E?
and
(6.6). (In CoSi2 and PtSi Ev lies typically between 3
1 int
(hv
-
O)2
—
SEFhv
yields
OZE? by
and 5
the
eV)
(6.6)
6.3.4.
of silicide
Application
6.
62
-
Silicon heterostructures to tunable infrared detector
Responsivity quantum efficiency -
The defined
which
spectral responsivity R(hv),
can
be determined
by
detector measurements is
by the photocurrent 7phot measured per incident light power P(hv) according to
The
responsivity
shows the well known Fowler
type dependence (eq. (6.6))
-
photon
on
energy,
Ä(*v)=c(^)2 where the
prefactor C
is
now
written
C
A(hv)
is the
crosses
as
AQiv) G(hv).
optical absorption in the silicide
hot carriers due to Fowler Plot
oc
multiple reflections.
(i.e. A{hv)
and
G{h\)
with
the energy axis at the barrier
efficiency
number of photoexcited
and
(6.9)
GQiv)
This relation is
is the internal
noraially applied
negligible dispersion),
to obtain a
efficiency gain of in the
so
straight
jRQiv)
oc
hv
(6.10)
O
-
7QE is the ratio of the number of measured photoelectrons to the
photoelectrons and can be derived from the responsivity employing
Principle of Tunable
are
depleted
separated by so
trapezoidal
a
thin
with
some
With Silicon
diagrams
the
bandgap, incident
always the Silicon between the two come into play.
Silicon Substrate does not
Schottky diodes,
The
separating
where the two metal
Si spacer is therefore
potential barrier, representing
small modifications for
energy smaller than the Si
i)
undoped Silicon1* layer.
that in the energy
(6.11)
Internal Photoemission Sensor
The basic device consists of two back to back films
called
line which
YQE=R(hv)^.
6.4.
-
height O.
hv
The quantum
(6.8)
image on
-
such
force corrections. a
the Si
Light,
rally
bandgap, with
is
photon
heterostructure is absorbed in both
metal films is meant. For all
properties
discussed below, the
Principle of Tunable
6.4.
layers, creating photoexcited
metal
enough carriers
their
over
photoemission
processes. The
former both metals spacer
are
Schottky clarity
are
barriers and therefore
devices,
zero
to an
TIPS and
the two metals
asymmetrical
are
the illumination and
on
asymmetrical on
TIPS. In the
both sides of the Silicon
different, forming different
energy band
TIPS energy band
diagram.
diagrams
For
more
for different
TIPS
bias and at
a
diagram of a symmetrical bias condition
general
according
due to the electric field at the interface Numbers 1
strongly dependent
asymmetrical
and
6.4 the energy band
Figure at
which leads to four different
separately.
Symmetrical
In
leading
thin
are
layer. Photoemission of these
one
occurs
symmetrical
(asymmetrical)
symmetrical
I discuss the
between
is
Si
63
condition that the silicides
identical, thus the Schottky barrier heights
identical. In the latter
bias conditions
6.4.1.
photoresponse
distinguish
on
is absorbed in
barrier into the
respective
the bias condition. I
electrons and holes,
light intensity
and that not the whole
Internal Photoemission Sensor
4 indicate the four different
through
also be used to refer to the
diagram
its energy band
are
TIPS
Schottky
sketched.
to eq.
Symmetrie,
possible photoemission
electric and
CoSi2/Si/CoSi2
barrier
is not included in
(6.1)
Schottky barrier ofthat particular process.
is
e.g. for
as
lowering,
Figure
6.4.
processes, but will
As the device and hence
photoelectric properties
will
by Symmetrie
in bias. For
zero
bias the
photocurrents
towards the
emitted above
bottom) Schottky
Cbelow'
as
3 and 4 cancel each other
be rather low. When
for holes, since
giving
rise to
a
towards smaller
barrier have to
Schottky barrier (i.e.
wavelengths).
cross
Thus
the Silicon
silieide. Since the Si thickness is much
Si, the carriers
photocuirent potential
even
are
as
the
sum
carriers emitted
the electric
larger than
condition
applied,
small bias is
increasing hole -
on
energy is directed
in electric field and
of the
6.4
over
b)).
are
One
applied potential
the cutoff wavelengths of these processes
photoereated
against
a
photocurrent (see Figure
describe the effective barrier of processes 2 and 3
difference Fand the
in
well
barriers 1 and 4, drift down the built
then coUected in the other silieide film can
as
photocurrent will
of equal currents. Thus the total
photoereated carriers
1 and 2
are
the effective
shifted
potential
field, before being coUected in the other
the ballistic
mean
free
paths
of hot carriers
thermalized and processes 2 and 3 do not contribute to the overall
at very small bias.
barrier maximum
by
(i.e. only
hot carriers with
the energy which is lost
during
the
an
energy
exceeding
path through Si,
can
the be
64
6.
Application
of silicide
-
Silicon heterostructures to tunable infrared detector
collected. Thus the cutoff is shifted to smaller hole
photocurrent
reverse
direction
wavelengths).
Electron
photocurrent
1 and the
4 will therefore dominate in the IR ränge. Since the holes head in the
compared to
the electrons, both photocurrents have the
sign.
same
mmm*?
Figure
Energy band diagrams of symmetrical TIPS (e.g. CoSySi/CoSij) for
6.4:
bias
and bias V
a)
b),
zero
which because of the device symmetry represents the
general bias condition. Since the intermediate Si is thin and undoped, it assumed to be The four
completely depleted
possible photoemission
is not included in
b),
The
applied
respective magnitude
bias and of the
mainly due
lowering.
photocurrent
processes
photoelectric
it is omitted in this
of each
spectral absorption
increasing
a
can
be
1000
asymmetrical
Schottky
electric
fields
expected. However,
a
Schottky contacts,
in
one
given voltage,
-
reason
which in the
case
Schottky effect
measurements the Silicon Substrate
figure.
on
the
photon
the
as
well
Silicon
as
a
image
to the
significant
has to bear in mind that
a
energy, the
3105
V/cm for Si
[88].
by thermionic
emission
105
of my
TIPS is the
the
across a
-
symmetrical
across
mainly governed by
in
V/cm which
If the silicide
the current will be
symmetrical
force barrier
increase
to electric fields of the order of ~
-
bias of 1 V
the electric dark current in these
TIPS may be described
barriers. For this
labeled 1-4.
barrier results.
in the silicides. The cutoff modulation is therefore
Ä fhick Silicon layer leads
interfaces form ideal in
are
photocurrent depends
is very close to the electric breakdown field of
as
trapezoidal potential
to the ratio of the electron and hole currents
With
TIPS with
importance,
a
which would lower the barriers of processes 1 and 4.
Since for the electrical and is of no
and
is
Silicon as
well
the four different
the lowest barrier at
CoSi2/p-Si Schottky barrier.
6.4.
6.4.2.
Asymmetrical
For TIPS
equally
valid and
energy band
diagram
negative
positive
this
and
the
as
small and for discussed In
applied
is
bias
are
no
clarity is
the effective
a)
dominating
at
zero
are
bias
-
across
also included in
symmetrical
only change compared Schottky barrier of PtSi. Figure
the
to
Thus the
6.5 the situations for
for the
case
Figure
symmetrical devices,
6.5. The
Schottky effect is
figure. The photoemission
processes
barrier of process 2 and 3 is increased with the
processes in the IR. Their
is the
sum
are
1 and 2 have the
contributions the barrier
hand, the holes have Situation
of the
reverse
direction of holes
of these two processes. The cutoff
to
case
is the
for holes
photoexcited holes
they
in
are
in PtSi have first to
barrier. As
the
larger
originating
CoSi2
a)
wavelength
is
the Si
by
electrons,
to
given by
the
CoSi2 hole barrier. yet horizontal,
but
now
to the PtSi electron
of the two hole
frorn
CoSi2
as
is the
for
case
all four processes contribute
Schottky barrier.
as
well
as
-
quite different. When the
Silicon interface and have been emitted
the electric field. On the other hand the
against the electric
already mentioned
for the
On the other
barriers, namely the CoSi2 barrier. But
and PtSi is
have reached the silicide
accelerated cross
not
are
compared
an
the electric field in the Si is small. For both electron
height corresponds
cross
affected
are
IR, since the barriers for the emission processes 3 and 4
height and
same
Process 1 and
respective Schottky barriers
TIPS. Otherwise the Situation is close to in the
wavelengths.
externally
change of the cutoff energy and simultaneously to
the energy bands in Si
b),
bias
photocurrents
into the Si,
to a small
the smallest barrier which in this
zero
symmetrical
potential
for
band condition, i.e. where the energy
is the
as
not indicated in the
photoresponse. Because
photocurrent
At
the
The
bias and hence the cutoff wavelength is shifted to shorter
increase of
to the
longer
potential
by the Schottky effect, leading
process
paragraph above
at any bias. In
sketched. Since the flat
bias condition
zero
more
65
assuming comparable light absorption in both metal films.
4 will be the
the
repeated here.
not
are
longer symmetrical
no
horizontal, is
are
well
as
TIPS most remarks made in the
device has to be made for the electron and hole
symmetrical
bands in Si
Internal Photoemission Sensor
TIPS
asymmetrical
are
Principle of Tunable
field
photoexcited holes
ballistically before they
can cross
the
asymmetrical TIPS, the ballistic mean free path
of hot holes in Silicon is much smaller than the Si thickness which renders this process to be rather
unlikely. Photocurrents of holes
respective sign.
and electrons in both directions add with their
66
6.
Application of silicide Silicon heterostructures to tunable -
infrared detector
^EF
V -o
3
d)
Figure 6.5:
Energy band diagrams for
negative bias a),
bias in used
for
zero
asymmetrical TIPS (e.g. PtSi/Si/CoSi2) sketched
consistently
in this
work).
-
band condition
Schottky effect in a), b) importance
At the forward flat
-
for the
and
band bias
and
photoemission d) is
devices,
c) and for positive
as
sketched here, is
Since the intermediate Si is thin and
completely depleted
results. The four different
according
flat
d) (the definition of negative and positive bias
it is assumed to be
no
b), the
bias
not
undoped,
trapezoidal potential are
processes
labeled
included. The Silicon Substrate,
is omitted for
VTB c),
a
barrier 1-4.
having
simplicity.
which is correlated with the
Schottky
barriers
to
e
VFB
*
90s.
be observed after about
were
determined
in
a
somewhat
process. In
series
possible
the Si etch rate increases a
larger
etching
higher
on
practice
of etch tests.
as
a
to
lay
rapidly
net area
having only
15
to
which
Ä thick
for #1184
can
my Si films than
density
samples also influences the electric fields
etching
the etch rate of
40 it is therefore
wafers. Thus the etch rate is very sensitive to the defect
had to be
3
factor
indicating
the fraction of
to other mechanisms such
through
as
tunneling,
Additionally the
Substrate, is usually modeled
through which
the diode is therefore lowered
V- IR and for
ideality
other mechanism.
the diode series resistance R, which stems from the a
(D.2)
1
-
nkT
the current / flows. The
as a
effect of series of
voltage drop
the introduced resistance. Thus with
kT/e, eq. (D.2) is approximated to I
=
where the Saturation current
Is
exp
'e(V-IRf\ nkT
)
1
-
(D.3)
Is is
/5=^**7*exp(-|f) A is the effective
barrier
height (A
=
area
120 A
of the cm
is the Richardson constant and O is the
diode, A
K
for
a
(D.4)
free electron gas
'
and A
=
Schottky
112 A cm" K
for
n-Si2)-3).). A method to extract the series resistance R of ideal
proposed by Norde4) and
had been modified
by
Schottky diodes (i.e.,
Sato and
YasumuraS) for the
Cheung, and N. W. Cheung, Appl. Phys. Lett. 49, 85 (1986). Sze, Physics of Semiconductor Devices, (John Wiley & Sons, J. H. Werner, Appl. Phys. A 47,291 (1988). H. Norde, J. Appl. Phys. 50,5052 (1979). K. Sato, and Y. Yasumura, J. Appl. Phys. 58, 3655 (1985). S. K.
S. M.
New
York, 1981).
n
=
case
first
1)
was
(1
< n
,
of current
vs.
sophisticated model
voltage
the linear ränge is small, it is
of Werner
(see Appendix E).
axis of a
while from the
second determination of the resistance R is obtained. However, when in
plot
-
a
semi
advantageous
to
-
slope
a
logarithmic
use
the
more
APPENDICES
108
E:
Schottky diode analysis J. H. Werner
ideality ideal
factor
hardly
model of
a
a
Schottky Diode, where, additional
real
and the Substrate series resistance
n
Schottky
proposed
(Werner method0)
II
diode should have
observed
even
when
a
bias
-
a
shunt resistance
independent
structures
mesa
Rs,
are
reverse
Rp
to the
is introduced. An
current, which in nature is
etched. For small forward bias the measured
I-Vis then also influenced by shunt currents.
Figure
E. 1:
Equivalent
circuit of
shunt resistance
a
and
Rf
series resistance
a
used to describe the current
With this model the current
-
Schottky diode
real
-
proposed by Werner,
as
Rs.
Further
an
ideality
with
factor
n
a
is
voltage dependence.
voltage equation (D.3)
has to be transformed to
(E.1)
Here
ID
resistance
Rp.
resistance
Rp
reverse
current
is the diode current and In this model it is
and
Rs
ID through the
diode
can
describes the shunt current
assumed, that the Saturation
and also the
bias characteristics for
Ip
ideality
factor
n are
through
current
independent
Is,
a
possible parallel
the shunt and series
of the bias V. From the
large negative bias, the shunt resistance then be corrected for the shunt current
can
be derived. The
as
*>-'-£ Then for forward bias
FD
=
V
-
IRS
(E.2)
»3 k T/e the thermionic diode current
7D
is
given
by (e(V , ID
11
J. H.
Is exp
Werner, Appl. Phys. A 47, 291 (1988).
-
IRs)
^
-^-
.
=
v
n
kT
(E.3)
E:
When
calculated0,
introducing
Schottky Diode Analysis II (Werner method)
the conductance G
Werner proposes the
A)
series resistance
Rs
(E.5)
-
=
to the
ideality factor
+
Rs
Rsli+rüa Cheung
n can
formula
One
can
compare
selfconsistency of the evaluation.
the
three
m (D.6).
results
in
Schottky
order
to
In his paper, Werner concludes that
the most reliable and accurate values for the
equation
the
voltage, corrected
for
From each
be extracted. With the
the series resistance, the Saturation current and therefrom the
determined.
or
"-M
equivalent
and the
either be measured
(E.4)
£^ is
can
fi._|-a_GJW 1
Equation C) (E.6)
which
following three different equations.
B,
O
dl^dV,
-
109
barrier O
test
the
can
then be
accuracy
and
equation A) (E.4) yields
Schottky barrier O, the ideality
n
and the series
resistance Rs.
The numerical determination of G from the /- Fcurve in order to get the real
slope
of the
curve.
requires
de
voltage steps typically of less than
1 mV
PUBLICATIONS
110
-
PRESENTATIONS
PUBLICATIONS
PRESENTATIONS
-
Publications Structural and electrical investigation N.
Onda, J. Henz, E. Müller, H.
Helv.
Phys.
Acta
von
ofan epitaxial metallic FeSi2-phase on Si(lll)
Känel, C. Schwarz, R. E. Pixley
64,197 (1991).
Surface study ofthin epitaxial CoSi^/SiflOO) layers by scanning tunneling microscopy and
reflection high
energy electron
diffraction
Stalder, C. Schwarz, H. Sirringhaus, and H.
R.
von
Känel
Surf. Sei. 271, 355 (1992).
Surface stnictures ofstrained epitaxial CoSi^SiflOO) layers studied by scanning tunneling microscopy
Stalder, C. Schwarz, H. Sirringhaus and H.
R.
Mater. Res. Soc.
Symp.
Proc.
von
Känel
237,499 (1992).
Surface strueture ofultrathin, epitaxial CoSi2films on Si(100)
observed
by scanning
tunneling microscopy Sirringhaus,
H.
Helv.
Phys.
R.
Acta
Stalder, C. Schwarz and H.
von
Känel
65,113 (1992).
Application of epitaxial CoSi^Si/CoS^
heterostruetures to tuneable
Schottky-barrier
detectors
Schwarz, U. Schärer, P. Sutter, R. Stalder, N. Onda and H.
C. J.
von
Känel
Cryst. Growth 127,659 (1993).
Epitaxial phase H.
von
transitions in the iron/silicon system
Känel, N. Onda,
H.
Sirringhaus,
C. Schwarz
Appl.
Surf. Sei. 70,559
(1993).
E. Müller
-
Gubler, S. Goncalves
-
Conto and
PUBLICATIONS
ofthepseudomorphic
Observation and characterization
to
-
PRESENTATIONS
111
stablephase transformation of
FeUxSi on Si(lll) N. Onda, H.
Sirringhaus,
S. Goncalves
Conto, C. Schwarz, E. Müller
-
-
Gubler and H.
von
Känel Proc.
Mater. Res. Soc.
Symp.
Epitaxy ofcubic
iron silicides
N.
280,581 (1993).
Si(lll)
on
Onda, H. Sirringhaus, S. Goncalves
Appl. Surf.
Phase transitions in
von
Känel
epitaxial silicides
C. Schwarz, S. Goncalves
-
Conto, E. Müller
on
B.
H.
Sirringhaus
(World Scientific, Singapore, 1994),
ofCoSi/Si/CoSi2
and H.
Lengeier et al., Proceedings
the formation of semiconduetor interfaces
14-18 June 1993
Characterization
Gubler,
-
ofSemiconductor Interfaces, by
international Conference
Germany,
Conto, C. Schwarz, S. Zehnder and H.
73,124 (1993).
Sei.
in Formation
-
heterostruetures
on
von
Känel
of the 4th
(ICFSI-4) Jülich, ,
p. 471.
Si(lll) by scanning tunneling
microscopy H.
Sirringhaus,
in Formation
C. Schwarz and H.
on
von
14 -18 June 1993
Lengeier et al., Proceedings
(World Scientific, Singapore, 1994),
Mater. Res. Soc.
-
p. 467.
iron silicides
G.
Malegori, Rev. B
L.
Kroll, C. Schwarz
Symp. Proc. 320,73 (1994).
Känel, M. Mendik, K.
von
U.
Conto
Elastic and vibrational properties H.
of the 4th
(ICFSI-4), Jülich,
Känel, U. Kafader, P. Sutter, N. Onda, H. Sirringhaus, E. Müller,
and S. Goncalves
Phys.
B.
the formation ofSemiconductor interfaces
Epitaxial semicondueting and metallic H.
Känel
ofSemiconductor Interfaces, by
international Conference
Germany,
von
Miglio and F.
50,3570 (1994)
A.
ofpseudomorphic FeSifilms
Mäder, N. Onda, S. Goncalves
Marabelli
-
Conto, C. Schwarz,
PUBLICATIONS
112
-
PRESENTATIONS
Magnetron sputter epitaxy ofSimGeJSi(001) P.
Ion
Lett. 65
(17),
2220
layer superlattices
Onda, S. Goncalves
Appl. Phys. 76,
New
7256
-
Phys.
Conto and H.
on
von
von
Känel and R. E.
(1994).
the CsCl
-
structure
Känel, C. Schwarz, E. Müller, L. Miglio, F. Tavazza and G. Malegori
Rev. Lett. 74,1163
Känel
Si
Conto, H. Sirringhaus, H.
epitaxially stabilized CoSiphase with
von
-
(1994).
channeling studies ofepitaxial Fe and Co silicides
C. Schwarz, N.
H.
-
Sutter, C. Schwarz, E. Müller, V. Zelezny, S. Goncalves
Appl. Phys.
J.
strained
(1995).
Pixley
PUBLICATIONS
-
PRESENTATIONS
113
presentations
Oral
Growth and characterization contributed talk
ofthin epitaxial CoSi2 layers
presented at the
fall
meeting of the
Swiss
on
Si(100)
Physical Society in Chur,
OctoberlO-11,1991
Growth
of^-FeSi^/Si(001) by molecular beam epitaxy
contributed talk
presented at the
7 th Euro
Pseudomorphic growth oftransition contributed talk
-
MBE, Bardonecchia, Italy, March 7-10,1993
metal monosilicides
presented at the spring meeting
on
Si(lll)
of the Swiss
Physical Society in Neuchätel,
of the Swiss
Physical Society in Bern,
March 24-26,1993
Strain measurements
contributed talk March
on
epitaxial silicides
presented at the spring meeting
17-18,1994
Application ofsilicide
-
Silicon heterostructures to
contributed talk presented at the March
a
tunable IR
spring meeting of the
Swiss
-
detector
Physical Society in Bern,
23-24,1995
Poster presentation
Application of epitaxial CoSi^/Si/CoS^ heterostructures
to
tuneable
Schottky
-
barrier
detectors
presented at the VII
-
MBE
Conference,
Schwäbisch Gmünd,
Germany, August 24 28,1992 -
114
CURRICULUM VITAE
CURRICULUM VITAE name
Claude Bernard Schwarz
born
in
marital State
unmarried
Citizen
Richterswil and Zürich, Switzerland
education
1973-1979
primary school in Richterswil
1979 -1981
secondary school
1981 -1985
Gymnasium type C in Zürich
Zürich, Switzerland,
October
19,1966
Matura type C
in Richterswil
(natural sciences)
1985-1990
studies of physics at the ETH in Zürich
1990
Experimental diploma work
at
the Solid State
Physics Laboratory, ETH Zürich,
in the group
of Prof. Dr. P. Wächter
Diploma
in
Physics
Diploma thesis:
"
of the ETH Zürich
Supraleitende Tunnel-
junctions mit epitaktischen CoSi2 Schichten auf Si(l
11)". Guidance by PD
Dr. H.
von
Känel 1990 -1995
experimental
Laboratory,
work at the Solid State
Physics
ETH Zürich, in the group of
Prof. Dr. P. Wächter.
Supervisor PD 1995
Dr. H.
von
Känel
Ph. D. Thesis "Electrical and
optical properties
of silicide Silicon heterostructures." examiner:
Prof. Dr. P. Wächter
co-examiners: Prof. Dr. I. Eisele PD Dr. H.
languages
German, French, English
von
Känel
DANK
115
DANK An dieser Stelle möchte ich Prof. Dr. Peter Wächter herzlichst
Diplomand
und
ich ganz
als Assistent in seine
ausgezeichnete
speziell
wissenschaftliche
verdanken. Er verstand
für mittägliche Waldläufe
zu
von
es
durch PD Dr. Hans
Betreuung
E.
Dr.
Y.
Lee
und
H.
Ein grosser Dank
immer sofort bereit
gut, mich immer wieder für die Physik und sogar
waren
Diplomarbeit
Sirringhaus
geht
ihre STM
an
Spitze
-
zu
Für die
speziell
speziell
-
Stalder, Dr. N. Onda,
liefern.
interessant
sehr
Spezialisten
über meine
Nico, der
H.
-
-
unserer
Französin
R. Deller und P.
Apparaturen.
äusserst
Henning,
die
scannen
um
zu
dünner Filme
einführte,
danken. vor
allem
Sylvie Goncalves
-
an
der
neuen
CoSi
-
Phase,
Conto danken.
Untersuchungen
bei Dr. E.
Wägli.
Dr. Andre Weber und Peter Suter
Sputter
Dr. U.
mich bereits während meiner
Ebenso bedanke ich mich für die zahlreichen TEM und SEM
Gubler,
und
Roli und
Schichten
Charakterisierung
aufwendigen Röntgenuntersuchungen,
möchte ich ganz
war
die STM
in die Geheimnisse der elektrischen
möchte ich ganz
mit Dr. R.
zusammen
aufschlussreiche Bilder meiner Schichten
-
Känel möchte
von
der Universität der Bundeswehr in München danke ich für
der MBE Maschine
an
abwechslungsreich.
Müller
hat. Somit
Übernahme des Korreferats.
Die Arbeit
Kafader,
mich als
motivieren.
Herrn Prof. Dr. I. Eisele die spontane
Forschungsgruppe aufgenommen
er
Forschungsarbeit sowie diese Arbeit überhaupt ermöglicht.
wurde meine Die
später
danken, dass
vergoldeten
meine Strukturen
jeweils
mit ihren
Für die vielen guten Kontakte und Dioden möchte ich auch Ihnen
von
Herzen danken. Peter Steiner hat mit seinen
Verfugungstellung Die besonders
seines
Messprogrammen und
der überaus
Kryostaten einen wesentlichen Beitrag
angenehme Atmosphäre
zu
grosszügigen
dieser Arbeit
im Labor F5 wird mir noch
lange
in
zur
geliefert.
Erinnerung
bleiben. Ein Dank
gebührt
auch Urs
systematischen Messungen
an
Schärer, der im Rahmen seiner Diplomarbeit die
symmetrischen TIPS gemacht hatte.
ersten
DANK
116
Unserem Techniker H.
Gübeli,
J.
-
der die
Apparaturen
immer im Schuss
Pixley
Physik
hielt, gebührt
ebenfalls mein Dank. Ein besonderer Dank
geht
an
Dr.
Ralph
Zürich, der mir die Geheimnisse und Tricks
E.
von
vom
RBS beibrachte. Er
Institut der Universität
-
war
alten Van de Graaff für mich anzuwerfen. Während den zahlreichen über
Gespräche
viele interessante schwerer Moment Ende 1993. Er ist
war
nun
Physik
und
unser
endgültige Stillegung
die
Messtagen
gemeinsames Hobby,
und
bestrebt, seine Messkammem
jederzeit bereit,
Demontage
am
Tandem
-
den
hatten wir
die Musik. Ein
Seiner RBS
Beschleuniger
-
Apparatur
der ETH
zu
installieren, sodass auch in Zukunft wieder RBS channeling Messungen möglich sein werden. Ein grosser Dank für die zur
Herstellung
geht
an
Prof. Dr. H. Melchior, der mir die Infrastruktur seines Instituts
und der teilweisen elektrischen
Verfügung stellte,
sowie seinen
Charakterisierung
von
TIPS
-
Strukturen
Mitarbeitern, welche mich in die entsprechenden Prozesse
einführten.
Diese Arbeit wurde dank der finanziellen Suisse
pour
recherche
la
Zusammenarbeit
micromecanique)
mit
Dr.
microtechnique) ermöglicht.
en
Y.
Unterstützung
Oppliger
vom
CSEM
von
der FSRM
Für
die
(Centre
Suisse
dabei
(Fondation
verknüpfte
d'electronique
et
möchte ich mich ebenfalls bedanken.
Allen hier nicht namentlich erwähnten Personen und
Gruppe
von
Prof. Dr. P. Wächter möchte ich mich für die
speziell
den Mitarbeitern der
angenehme
Zusammenarbeit
danken. Zuletzt und
vor
geht
mein Dank
an
meine
Eltern, die mich auf meinem Werdegang moralisch
allem mein Studium finanziell unterstützten.
Yvonne möchte ich
spärlichen
Präsenz. Auch
mir immer wieder
März 1995
neue
an
dieser Stelle danken für Ihr Verständnis meiner
wenn es
Kraft
zum
Teil
im Labor mal nicht nach meinen Wünschen lief, konnte Sie
geben.