Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
EFFECT OF FLUORINE AND HYDROGEN RADICAL SPECIES ON MODIFIED
OXIDIZED Ni(Pt)Si
Sneha Sen Gaddam
Thesis Prepared for the Degree of
MASTER OF SCIENCE
UNIVERSITY OF NORTH TEXAS
May 2010
APPROVED: Jeffry Kelber, Major Professor Steve Cooke, Committee Member Bill Acree, Chair of the Chemistry Department Michael Monticino, Dean of the Robert B. Toulouse School of Graduate Studies
Gaddam, Sneha Sen. Effect of fluorine and hydrogen radical species on
modified oxidized Ni(Pt)Si. Master of Science (Chemistry), May 2010, 30 pp., 14
illustrations, 22 references,.
NiSi is an attractive material in the production of CMOS devices. The
problem with the utilization of NiSi, is that there is no proper method of cleaning the
oxide on the surface. Sputtering is the most common method used for the cleaning,
but it has its own complications. Dry cleaning methods include the reactions with
radicals and these processes are not well understood and are the focus of the project.
Dissociated NF3 and NH3 were used as an alternative and XPS is the technique to
analyze the reactions of atomic fluorine and nitrogen with the oxide on the surface. A
thermal cracker was used to dissociate the NF3 and NH3 into NFx+F and NHx+H.
There was a formation of a NiF2 layer on top of the oxide and there was no
evidence of nitrogen on the surface indicating that the fluorine and hydrogen are the
reacting species. XPS spectra, however, indicate that the substrate SiO2 layer is not
removed by the dissociated NF3 and NiF2 growth process. The NiF2 over layer can be
reduced to metallic Ni by reacting with dissociated NH3 at room temperature. The
atomic hydrogen from dissociated ammonia reduces the NiF2 but it was determined
that the atomic hydrogen from the ammonia does not react with SiO2.
Copyright 2010
by
Sneha Sen Gaddam
ii
ACKNOWLEDGEMENTS
First and foremost, I would like to express my gratitude to my advisor Dr.
Jeffry Kelber for his encouraging guidance and sincere support during the course of
my graduate study. I greatly appreciate his patience in teaching me about the
important aspects of research. His sincerity, acumen, fortitude and diligence are
qualities which I hope to mirror in my professional life. Special thanks to Ms. Sudha
Manandhar whose contributions are very significant to my thesis. She has been a great
mentor and I have learnt a lot under her guidance and acknowledge her patience in
teaching me the aspects and operation of XPS. I also thank my past and present group
members, Mr. Cameron Bjelkevig, Mr. Justin Wilks, Mr.Brian Copp, Ms.Sundari
Pokharel and Ms. Mi Zhou for their constant motivation and discussions. I would like
to extend my gratitude to my committee member Dr. Steve Cooke for his help in
finishing up my thesis. I would like to express my admiration towards his earnestness
and spirited approach in professional life. A very special thanks to Ms. Susan
Brockington for helping me finish out all the necessary requirements for my
graduation. I also thank my family members for their never ending love and
encouragement and my friends for being my support system and being my family
away from home.
iii
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS ..........................................................................................iii
LIST OF FIGURES ......................................................................................................vi
CHAPTERS
1. INTRODUCTION ................................................................................1
1.1 NiSi and Ni(Pt)Si: Source/Drain Contacts for Ultra –Small
Transistors ..................................................................................1
1.2 Cleaning of NiSi and Pt-NiSi Surfaces .....................................3
1.3 Previous Work ..........................................................................4
2. EXPERIMENTAL METHODS ............................................................5
2.1 Characterization by X-ray Photoelectron Spectroscopy (XPS) .5
2.2 Description of the UHV Chamber ............................................9
2.3 Description of the Thermal Cracker .......................................10
2.4 Sample Preparation .................................................................11
3. EXPERIMENTAL RESULTS ............................................................13
3.1 Formation of a Clean Mono Silicide Surface .........................13
3.2 Oxidation of the Monosilicide Surface ...................................16
3.3 Reaction of Dissociated NF3 on SiO2/ Ni(Pt)Si ......................17
3.4 Reaction of Dissociated NH3 with NiFx Overlayer ................23
iv
v
4. SUMMARY AND CONCLUSIONS .................................................26
4.1 Summary and Conclusions .....................................................26
4.2 Future Work ...........................................................................28
REFERENCES ............................................................................................................29
LIST OF FIGURES
Page
1.1 TEM image of As-doped 18nm Ni with two steps Ni silicidation ....................1
2.1 Pictorial representation of theory of XPS ..........................................................6
2.2 System for in situ O2 , NH3 and NF3 dosing (molecular and dissociated
species) and XPS studies of the surface modification on Ni(Pt)Si .................10
2.3. Mechanism and schematic representation of a thermal cracker ......................11
3.1(a) XPS spectra of Ni(Pt)Si (a): As received spectra for Ni(2p), Si(2p) and
corresponding surface ......................................................................................14
3.1(b) XPS spectrum for carbon and oxygen. As received spectra and the spectra for
the clean sample ..............................................................................................15
3.2 Graph of the intensity of oxygen vs exposure in langmuir .............................16
3.3 XPS spectra of Ni(2p), Pt(4f), Si(2p) clean surface (top trace) and the XPS
spectra of Ni(2p), Pt(4f), Si(2p) after the oxidation with molecular oxygen at
5x10-6 torr , 8x106 L ........................................................................................18
3.4 Evolution of the F(2p) XPS spectrum upon increasing exposure to dissociated
NF3 : 600L (bottom trace), after 1200L (middle trace) and after 1800L (top
trace) ................................................................................................................19
vi
3.5 XPS of Ni spectra after oxidation with molecular oxygen of 79200L (a), after
atomic NF3 dose of 600L (b), after atomic NF3 dose for 1200L (c), after
atomic NF3 dose for 1800L (d) ........................................................................20
3.6 Graph of the total Si(2p) and O(1s) intensity vs exposure to fluorine in
langmuir oxygen(■) and silicon(▲) ...............................................................21
3.7 Graph of the total intensity vs exposure to fluorine in langmuir. nickel (■) and
platinum (▲) ...................................................................................................22
3.8 XPS spectra of F(1s), Ni(2p), O(1s) and Si(2p) after the reaction with NF3
with dissociated ammonia ...............................................................................25
4.1 Illustration of the reaction of fluorine and NH3 on oxidized Ni(Pt)Si. ...........27
vii
CHAPTER 1
INTRODUCTION
This thesis is a compilation of the work done in the Surface Science Laboratory
at the University of North Texas under the guidance of Dr. Jeffry Kelber. This thesis
focuses on the interactions of free radical species with oxidized NiSi and Pt-doped
NiSi (Ni (Pt) Si) surfaces under controlled ultra-high vacuum (UHV) conditions. The
first chapter explains the scientific background for the research; the second chapter
contains an explanation of the experimental methods involved. The third chapter
contains results for the reaction of molecular vs. dissociated NF3 and NH3 with
oxidized Ni(Pt)Si. The fourth chapter contains a discussion of the results and possible
future avenues of research.
1.1. NiSi and Ni(Pt)Si: Source/Drain Contacts for Ultra–Small Transistors
NiSi has become an attractive material for source, drain and gate contacts in
advanced complementary metal-oxide semiconductor (CMOS) devices (Fig 1.1).
Fig 1.1: TEM image of As-doped 18nm Ni with two steps Ni silicidation 1.
1
The reasons for this are that this silicide material has the lowest achievable
specific resistivity among silicides of interest (e.g., CoSi2, TiSi2) 2 and also consumes
less silicon than disilicides during formation, hence reduces the risk of junction
spiking due to substrate Si depletion. NiSi formation is a one step annealing process
and thus proves to be a simple silicidation process3. Additionally, there is no adverse
line-width dependence of sheet resistance, which is a big problem in TiS2 4. This is
due to the fact that the temperature for the Ni mono silicide formation is lower than
for TiSi2 and is controlled by the rapid diffusion of Ni, rather than Si 5. This diffusion
-controlled formation leads to smoother interfaces and surfaces, and helps in reducing
the voiding observed in Co silicided narrow poly-Si lines 6. NiSi is also known to be
compatible with Ge 6, thus making it a suitable contact material for Si-Ge transistors.
The mechanical stress of NiSi on silicon is also very small.
The major drawback of NiSi is its poor thermal stability, with formation of
disilicides and reduced conductivity occurring for temperatures above 750oC 4. The
introduction of the Pt metal into the NiSi (typically, ~ 5 wt %) improves its thermal
stability and reduces the nucleation of other silicide phases 4. In the course of device
fabrication, NiSi and Pt-NiSi can undergo oxidation due to various processes, like
plasma etching and oxygen exposure. Previous studies in our group8 have shown that
exposure of Pt-NiSi (or NiSi) to molecular oxygen or atmosphere yields a formation
of a thin (6-15 Å) SiO2 layer followed by passivation. In contrast, exposure to atomic
oxygen yields the kinetically-driven formation of a SiO2/metal silicate overlayer
without passivation. The presence of Pt retards but does not halt this process8.
2
1.2. Cleaning of NiSi and Pt-NiSi Surfaces
The removal of oxide or silicate over layers prior to metallization of the Pt-
NiSi surface is important to ensure good interfacial nucleation and adhesion, as well
as low contact resistance.
Sputtering, thermal treatment and reactive cleaning are widely used techniques
to clean the surfaces of bulk crystalline samples. While a combination of Ar ion
sputtering and heating at elevated temperatures is proven to be an effective method in
the preparation of the surfaces, it has been observed that with multi-component
systems, a complication is introduced. Sputtering and heating cause the surface
composition to change with respect to the bulk composition. There is considerable
roughening of the surface and annealing in vacuum (not always practical in industrial
situations) is necessary to restore the surface structure and desired chemical
composition 9.
Wet cleaning used to clean ultra-large scale integrated circuits (USLI) has
faced difficulties: introduction of cleaning agents and de-ionized water in to fine gaps
and holes. Dry cleaning (e.g., plasma or vapor chemical cleaning) is also expected to
be applicable from the view point of chemical integration and other chemical vapor
deposition techniques. Dry cleaning methods like the HF vapor process have been
studied in detail for the removal of Si native oxides 10. Other dry etching methods like
NF3/H2O exhibit a low oxide etching rate of substrate and excessive etching of Si 10.
Partially dissociated NF3 and heated NH3 gases achieved a high etching rate, as it was
found that the etchant was directly generated by the thermally excited species 10. Such
dry cleaning methods involve the reaction of radical species with surfaces. Such
processes are not well understood, and are a focus of research in this project.
3
1.3. Previous Work
Previous work in Dr. Kelber’s group8, 11 has characterized the evolution of the
NiSi and Ni(Pt)Si surface structure on exposure to molecular oxygen(O2) and to a
mixed flux of atomic and molecular oxygen (O+O2). It was observed that the
exposure to O2 resulted in a growth of a thin (6 Å-15 Å, depending on O2 pressure)
SiO2 layer that grows slowly and passivates the surface toward further oxidation.
Under high vacuum conditions at 300 K (PO2 ~ 10-6 torr), the limiting oxide average
thickness is 6-7 Å 8. In contrast the exposures to dissociated oxygen resulted in greatly
increased oxidation rate. The high rate of oxidation, combined with the ability of
atomic oxygen to diffuse through SiO2 12, resulting in the kinetically-driven oxidation
of the metal (Ni, Pt) The presence of Pt in the surface region retards, but does not stop
the oxidation process, for which passivation is not observed 8. Since plasma treatment
(and therefore the presence of atomic O) is common in processing of these materials,
the formation of thick metal silicate layers on NiSi or Ni(Pt)Si must be considered in
designing appropriate surface cleaning methods.
X-ray electron spectroscopy was also used to study the effect of atomic oxygen
and dissociated ammonia on ruthenium and RuO2 respectively. It was found that the
exposure to atomic oxygen results in the formation of a disordered multi layer
ruthenium oxide. The oxide formation proceeded until an average thickness of 68Å.
There was no evidence of higher oxides. The oxide was then exposed to dissociated
ammonia which resulted in the removal of lattice oxide and no nitrogen adsorption
was observed 11.
4
CHAPTER 2
EXPERIMENTAL METHODS
2.1. Characterization by X-ray Photoelectron Spectroscopy (XPS)
2.1.1. Background
XPS (X-Ray photoelectron spectrometry) is a surface analytical technique used
to characterize the chemical composition and electronic structure of the surface region
of materials (typically, the outermost ~ 50 Å). The method involves the detection of
photoelectrons from (usually) the surface region of a material due to the absorption of
photons in the UV or x-ray region of the electromagnetic spectrum. XPS has its
origins in the investigations of the photoelectric effect discovered by Hertz. Innes in
1907 conducted experiments involving a Rontgen tube with a platinum cathode. In
1914, Rutherford stated the first equation of XPS which was later modified13. After
the World War II, Steinhardt and Serfass revived the idea of utilizing XPS as an
analytical tool 13. In 1954, the first X-ray electron spectrum of cleaved sodium
chloride was obtained. During the late 60’s and 70’s, the Siegbahn group (Uppsala)
went on to develop the field of electron spectroscopy for chemical analysis of
molecules and surfaces. Commercial instruments started to appear in 1969-70. Today,
XPS and related methods are commonly used to address a broad variety of issues in
academic and industrial research involving the electronic and chemical structure of
the surface regions of materials. This derives from the inherent surface sensitivity of
the technique under typical laboratory conditions, where the inelastic mean free paths
of photoelectrons are ~ 10 Å – 100 Å.
5
2.1.2. Theory
The primary process of XPS is the ionization of an atom, typically from a core
level, and is illustrated schematically in Fig. 2.1.
If the energy of the excitation photon is known, the electron binding energy
(BE) of each of the emitted electrons can be determined by using the following
equation
------------------------------------ (1)
Where Ebinding is the binding energy of the electron, Ephoton is the energy of the
X-ray photons being used, Ekinetic is the kinetic energy of the electron as measured by
the instrument and φ is the work function of the spectrometer (not the
material).
KE = hν ‐ BE
Fig 2.1: Pictorial representation of theory of XPS.
6
http://en.wikipedia.org/wiki/Electron_binding_energyhttp://en.wikipedia.org/wiki/Work_function
Note that addition of the work function term (eqn. 1) due to the photoelectron
entering the spectrometer detector is not shown here.
Since binding energies for core levels of different elements are typically well-
separated in energy, Ebinding serves as a “fingerprint” for elemental identification.
Typical values for Ebinding in XPS are between 20 eV and 1100 eV. Additionally,
Ebinding is sensitive to the oxidation state of the atom and other factors in the chemical
environment. Such “chemical shifts” (ΔEbinding) are typically ~ 0.1 eV – 10 eV,
depending on the element involved and other factors. Detailed discussions and
tabulations of chemical shifts for elements in specific compounds, and their
relationship to electronic structure, can be found in standard text books. The ability to
determine chemical shifts with high precision using modern electron energy analyzers
allows XPS to serve as a powerful tool for characterizing both surface elemental
composition and chemical bonding/electronic structure.
The surface sensitivity of XPS derives from the limited inelastic mean free
paths (IMFP) for photoelectrons in XPS. The IMFP is the average distance an electron
can travel in a solid before undergoing an inelastic collision and therefore losing any
chemical information (eqn. 1). IMFP values very somewhat with the density of the
material through which the photoelectron passes, but are in general a function of the
photoelectron kinetic energy. An approximate universal (element-independent)
relationship is given by:
7
)]()()ln([ 22
ED
ECEE
E
p +−=
γβλ
MNE vp
ρ8.28= = Free‐Electron Plasmon Energy (eV)
Nv = Number of valence electrons per atom or moleculeρ = Materials density (g cm‐3)M = Atomic or Molecular Weight
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐ (2)
XPS spectra for the experiments discussed here were acquired with non-
monochromatic Mg Kα radiation (hv=1253 eV), and with the analyzer in constant
pass energy mode (22 eV). The electrons escape the surface at about 1000V, which is
very high for the detector to capture and analyze them. Pass energy is the retardation
voltage applied in an electron to slow it down and be captured by the detector. The
high retard ratios Ekin / Epass for XPS lines at high energies principally reduce the
intensities in XPS (and AES) spectroscopy. The pass energy should be as high as
possible, taking in to account the needed minimum energy resolution. Analysis of
XPS spectra was performed according to established methods 14, with commercially
available software (ESCATOOLS) and shirley background subtraction 13. The
analyzer energy scale was calibrated against sputter-cleaned polycrystalline Cu foil.
Sample peak intensities were taken as proportional to the peak area (after background
subtraction) and were determined by fitting to Gaussian–Lorentzian components. The
relative Ni/Si atomic ratio (XNi/XSi) was estimated using the intensities of the
Ni(2p3/2) and Si(2p) intensity at 100 eV according to:
XNi/XSi = FNiSi (INi/SNi)/ (ISi/SSi ) ------------------------------------- (3)
8
where I and S represent, respectively, the corresponding intensities and atomic
sensitivity factors appropriate to this analyzer. The matrix correction factor, FNiSi is
given by
FNiSi ═ (λNidNi)/(λSidSi ) ----------------------------------------- (4)
Where λ represents the electron inelastic mean free path and d represents the atomic
number density.
2.2. Description of the UHV Chamber
All the experiments were carried out in a system consisting of an ultra-high
vacuum (UHV) analysis chamber for surface characterization, and a processing
chamber permitting the exposure of samples to reactive environments (for, e.g.
chemical vapor deposition) without compromising the UHV integrity or
instrumentation in the analytical chamber. This system is shown schematically in Fig.
2.2. The analytical chamber (turbo molecularly pumped, base pressure 1 x 10-10 torr)
consists of a 100 mm mean radius hemispherical analyzer (Vacuum Generators) and
non- monochromatic X-ray source (Physical Electronics)for acquisition of XPS
spectra, LEED (low energy electron diffraction) as well as an ion gun (Physical
Electronics) for sample bombardment/cleaning. The processing chamber (turbo
molecularly pumped, base pressure 5 x 10-8 torr) consists of a precursor dosing
apparatus, and a commercially available (Oxford Applied Research) thermal catalytic
cracker for the generation of free radical species compatible with UHV operating
conditions. Sample temperature in the main chamber is controllable between 150 and
1200 K by a combination of liquid nitrogen cooling and resistive heating. Sample
transport between chambers occurs via magnetically-coupled feed-thru under UHV
9
conditions, with chamber separation accomplished by a differentially pumped Teflon
seal when the sample is in the main chamber. Sample temperature is monitored with a
type-k thermocouple attached to the sample holder. The pressure in each chamber is
monitored by nude ion gauges.
LEED
Turbo 1P = 10‐8 Torr
Turbo 2P ~ 5 x 10‐10 Torr
Thermal Cracker
Hemispherical Analyzer (XPS)
View port
Fig 2.2: System for in situ O2 , NH3 and NF3 dosing (molecular and dissociated
species) and XPS studies of the surface modification on Ni(Pt)Si.
2.3. Description of the Thermal Cracker
The TC 50 (Fig. 2.3) is a thermal gas cracker which is designed for use in
UHV. It has the capability to dissociate molecular gases like hydrogen, ammonia and
generating a stream of highly reactive atomic species. The cracker, from Oxford
Applied research employs an iridium tube which is heated up to 1273o K by electron
beam induced heating. The cracker tube is directly connected to the gas lines, thus
10
reducing the gas load and providing a large surface area and a long path length. The
head is directly water cooled to keep the temperature at ambient and to minimize out
gassing even at maximum power. The gases are sent through the feed and the heated
tube and are then converted to atoms from molecules by catalytic dissociation at the Ir
surface and subsequent desorption as radical species. Dissociation efficiencies vary
with species and pressure, but typically were in the range of 40%-60% for conditions
typical of these experiments.
Gas in
Mixed atomic and molecular flux
H.
Water
Capillary Tube
Filament
H2
Fig 2.3: Mechanism and schematic representation of a thermal cracker.
2.4. Sample Preparation
The sample, 1 cm x 1 cm, was scored from a Ni(Pt)Si wafer (approximately
200Å thick, 5% at. wt. Pt deposited on a Si(100) film) provided by
STMicroelectronics-Crolles. Resistivity measurements indicated that the Si substrate
was As-doped with a density of 5x 1015 cm3. The sample as received was mounted on
11
a Ta foil which was in turn spot-welded to Cu leads for resistive heating/liquid
nitrogen cooling. The sample was introduced into UHV through the intro chamber
and then transferred in to the main chamber.
The sample was then cleaned in the main chamber to remove any oxide or
carbon over layer present due to the exposure to atmosphere. Sample cleaning and
preparation were accomplished by a combination of Ar+ bombardment (1 keV) and
annealing to about 600 K. XPS was taken to make sure that there is minimum amount
of carbon and SiO2 on the surface. Suitable ion bombardment and annealing in UHV
results in an oxide and carbon-free sample, with a Pt-enriched surface region 8.
12
CHAPTER 3
EXPERIMENT RESULTS
3.1. Formation of a Clean Monosilicide Surface
The Ni(Pt)Si sample was first mounted in to the introduction chamber and then
transferred in to the main chamber. The sample was cleaned in cycles of Ar+
sputtering and annealing up to 800K for about 15 minutes. Ni (2p), Si (2p) and Pt (4f)
XPS spectra are displayed in Fig. 3.1(a) for the as-received surface and the samples
after Ar+ sputtering and annealing (clean sample). Corresponding C (1s) and O(1s)
spectra are displayed in Fig. 3.1(b). The analysis of the XPS spectra acquired for the
as-received sample indicates the presence of an oxide thickness of 15(2) Å. The Ni
(2p) spectrum after the sputter/anneal cycle shifts to a lower binding energy
corresponding to the Ni in Pt-NiSi binding energy.8. The Si(2p) spectrum for the as-
received sample displays a feature at 103eV and another near 100eV. XPS studies of
Pt-free NiSi report the Si (2p) binding energy at 98.85 and 98.75 for Ni2Si and NiSi
respectively15 and binding energies ~100.2 for Pt silicides.16. We therefore attribute
the Si (2p) peak at around 100eV to Si due to both Pt and Ni silicides.8. After the
sputter anneal, there is a reduction on the peak intensities of carbon and oxygen along
with decrease in the Si (2p) peak at 103eV. This accounts for the removal of the oxide
as well as the carbon on the surface to form a clean surface on the sample. The
persistence of the carbon and oxygen indicates that the Ni(Pt)Si sample is reactive in
the UHV conditions.
13
880 872 864 856 848-1000
0
1000
2000
3000
4000
108 104 100 96
0
200
400
600
800
80 76 72 68
0
100
200
300
400
500
Clean
As recieved
880 872 864 856 848
-200
0
200
400
600
112 108 104 100 96-50
0
50
100
150
200
80 76 72 68-50
050
100150200250300
Ni(Pt)Si (854.0) Pt and NiSi (103.1)
SiO2 (103.1)
PtSi( 73.2)
Binding energy(eV)
Inte
nsity
Figure 3.1 XPS spectra of Ni(Pt)Si (a): As received spectra for Ni(2p), Si(2p), Pt(4f)
(bottom trace) and corresponding spectra for the clean surface (after sputter and
anneal) (top trace).
14
-292 -288 -284 -280-40-20
020406080
-537 -534 -531 -528-50
050
100150200
-292 -288 -284 -280-500
50100150200250300
-537 -534 -531 -528-2000
200400600800
10001200
As received
Clean
C(1s) O(1s)
Fig 3.1(b): XPS spectrum for carbon and oxygen. As received spectra (bottom trace)
and the spectra for the clean sample (top trace).
After annealing, XPS indicated a surface stoichiometry of 1:1 for Si relative to
total (Ni+ Pt). The Ni(2p) peak maximum for the clean surface (Fig. 2.1a, top left)
was 854.0 eV, consistent with a Ni/Si ratio of 1:1 4.
15
3.2. Oxidation of the Monosilicide Surface
The clean surface was exposed to molecular oxygen at 5x10-6 torr for varying
lengths of time to form a layer of SiO2 on the surface. Increase in O(1s) intensity
(normalized by total Ni(2p) intensity) is shown in Fig. 3.2 The O(1s) intensity
increases monotonically up to a total oxygen exposure of 1x10-6 Langmuir (L) (1 L =
10-6 Torr-sec). The absence of further surface oxidation beyond this point indicates
that a continuous SiO2 film inhibits further O2 chemisorptions, dissociation and
surface oxidation.
0 2 4 6 8
600
900
1200
1500
1800
2100
2400
Inte
nsity
Exposure x105(L)
Molecular Oxygen Uptake
O(1s)
Fig 3.2: Graph of the intensity of oxygen vs exposure in langmuir; P (O2) = 5x10-6
torr T = 298K.
At this point the average SiO2 thickness, calculated from relative Si and O
intensities is about 6-8 Å.
16
During the oxidation process, the Ni(2p), , and Pt(4f) peaks display a gradual
decrease in the intensity with a negligible change in the binding energies with
increasing exposure to O2 consistent with passivation, no further decrease in Ni(2p) or
Pt(4f) intensity is observed for O2 exposure > 8x106L. The decrease in core level
intensities is due to SiO2 attenuation of photoelectron signals from the substrate, while
the absence of any binding energy shifts indicates that the amount of Si transported
into the oxide overlayer during the oxidation process is too small to significantly alter
the electronic state of Ni or Pt.
XPS of Pt(4f), Ni(2p) and Si(2p) intensities before oxidation (clean) and after
exposure to 8x106 L O2 at room temperature are shown in Fig. 3.3. A feature
corresponding to SiO2 is clearly visible after oxidation, consistent with an O(1s) peak
(not shown) characteristic of SiO2. In summary, figure 3.2 indicates an increase in the
O(1s) intensity with complete passivation at 8x106 L. The SiO2 is formed as a
continuous layer on the top of the sample. As mentioned before, Manandhar et al, has
observed that the oxidation on Ni (Pt)Si with atomic oxygen is much more aggressive
when compared to molecular oxygen.
3.3. Reaction of Dissociated NF3 on SiO2/ Ni (Pt)Si:
The sample was first exposed to molecular NF3 at 1x10-6 torr for 15 min. It was
observed that the molecular NF3 did not have any effect on the sample. The oxidized
Ni(Pt)Si was then exposed to dissociated NF3 (F + NFx; x = 1,2) at 1x10-6 torr for an
exposure of 1800 L. The dissociation efficiency of NF3 was assumed to be around
48%. The exposure is calculated in langmuir. A langmuir is determined by
multiplying the pressure of the gas with the time of exposure. One langmuir is equal
17
to the gas pressure of 1x10-6 torr for 1 second. The dissociation energy was
previously measured for ammonia. It was also noticed from the nitrogen spectrum that
there was no uptake of nitrogen and thus indicates that the fluorine is the major
species that reacts with the sample.
880 872 864 856 848
0
1000
2000
3000
4000
84 80 76 72
0
100
200
300
400
500
112 108 104 100 96
0
200
400
600
800
880 872 864 856 848-500
0500
10001500200025003000
84 80 76 72-50
050
100150200250300350400450
112 108 104 100 96-100
0100200300400500600700
Binding Energy
Inte
ns
ity(
CP
S)
Clean
After Oxidation
Ni(Pt)Si (854.0)PtSi (73.2) Ni(Pt)Si (100.1)
SiO2 (103.2)
Ni(2p)
Ni(2p)
Pt(4f)
Pt(4f)
Si(2p)
Si(2p)
Fig 3.3 XPS spectra of Ni(2p), Pt(4f), Si(2p) clean surface (top trace) and the XPS
spectra of Ni(2p), Pt(4f), Si(2p) after the oxidation with molecular oxygen at 5x10-6
torr , 8x105 L.
18
The corresponding evolution of the XPS spectrum indicates that the fluorine
first reacts with the silicon forming a fluoride species (fig 3.4). The peak at 687eV on
the fluorine spectrum indicates the formation of fluorine bound to silicon 17along with
fluorine bound to Si and oxygen at 687.3 18. Further reaction of the sample proceeds to
drive the nickel in the bulk of the sample to the surface to form NiFx species on the
top of all the other layers. The XPS spectrum indicates the formation of the peak at
685 eV which corresponds to NiF2 formation19.
696 694 692 690 688 686 684 682 680
0400800
12001600696 694 692 690 688 686 684 682 680
0400800
12001600696 694 692 690 688 686 684 682 680
0500
1000150020002500
Increase in the Intensity of this side of the peak at about 685eV. NiF2 formation
After Atomic NF3 dose for 600L
After Atomic NF3 dose for 1200L
After Atomic NF3 dose for 1800L
Binding Energy
Inte
nsity
(CP
S)
Fig 3.4: Evolution of the F(2p) XPS spectrum upon increasing exposure to dissociated
NF3 : 600L (bottom trace), after 1200L (middle trace) and after 1800L (top trace).
19
890 885 880 875 870 865 860 855 850 845-700
0700
140021002800
After oxidation with Mol Oxygen for 79200L890 885 880 875 870 865 860 855 850 845
-4000
400800
12001600 After atomic NF3 dose of 600L
890 885 880 875 870 865 860 855 850 845-400
0400800
1200
After 1800 L Atomic NF3 dose
After 1200 L Atomic NF3 dose890 885 880 875 870 865 860 855 850 845
-4000
400800
1200
Ni rich Si(853.1, 852.8)
NiF2(864.3)
NiF2(858.3)
PtNiSi(854.0)
Ni 2p3/2(857.6)
Binding Energy
Inte
nsity
(CP
S)
Fig 3.5: XPS of Ni spectra after oxidation with molecular oxygen of 79200L (a), after
atomic NF3 dose of 600L (b), after atomic NF3 dose for 1200L (c), after atomic NF3
dose for 1800L (d).
It is also evident from the corresponding XPS spectra for Ni that there is a
reaction of Ni with the fluorine atoms. The peak at 858.2(1) eV (Fig. 3.5) proves that
there is a formation of NiF3 species 20. It is observed from the spectra that there is a
20
shift in the binding energy which makes it evident that NiSi is first transformed in to a
nickel rich silicide before the reaction with the fluorine species occurs.
Variation in Si(2p) and O(1s) intensities as a function of exposure to NFx +F
are shown in Fig. 3.5. Corresponding changes in Ni(2p) and Pt(4f) intensities are
shown in Fig. 3.6.
-200 0 200 400 600 800 1000 1200 1400 1600 1800 2000500
1000
1500
2000
2500
3000
3500
4000
4500
5000
5500
Oxygen Silicon
Inte
nsity
exposure to Fluorine (L)
Fig 3.6: Graph of the total Si(2p) and O(1s) intensity vs exposure to fluorine in
langmuir oxygen (■) and silicon (▲).
21
The intensity of the silicon and oxygen decreases with increasing exposure to F
(fig 2.5). This might be due to the fact that the fluorine reacts with silicon oxide first,
consuming all of the silicon. In contrast, (fig 3.6) the Ni(2p) intensity decreases up to
a total F exposure of 1200 L. Subsequent F exposure results in increase in Ni(2p)
intensity (Fig. 2.6) corresponding to formation of NiF2 . During this process, there is
only minor change, if any, in the Pt(4f).
-200 0 200 400 600 800 1000 1200 1400 1600 1800 2000
0
200
400
600
800
1000
1200
1400
1600
1800
Ni Pt
Inte
nsity
Exposure to Fluorine (L)
NiF2 over layer formationF attack on SiO2
Ni(2p)
Pt(4f)
Fig 3.7 Graph of the total intensity vs exposure to fluorine in langmuir. nickel (■) and
platinum (▲)
22
The data in Figures 3.6 and 3.7 indicate that the fluorine reacts first with the
silicon dioxide and then eventually reacts with the nickel. It is been observed that the
Ni intensity increases at 1200L. This might be due to the fact the fluorine draws the
Ni up in to the surface after the consumption of silicon, which leads to a decrease in
the intensity of all the other elements. The Ni is drawn up to the surface through the
Cabrera-Mott mechanism. The hypothesis is that a strong field is set up on the oxide
film due to a contact potential difference between the metal and the oxygen atoms
which enables the metal ions to diffuse through the oxide layer without much help
from the change in the temperature 21.
3.4. Reaction of dissociated NH3 with NiFx overlayer
The sample was first treated with molecular NH3 for any reduction of the
fluoride. It was observed that there was no change in the spectra after the dose with
molecular ammonia. The sample was treated with dissociated ammonia at 5x10-6 torr
after the sample was reacted with NF3. The dose of the gases is measured in langmuir
(L): 1 L corresponds to an exposure of 1x10-6 torr for 1 second. Dissociation
efficiency of ammonia is measured to be 48%. The efficiency was previously
measured using a mass spectrometer. Fig 2.7 indicates that hydrogen from dissociated
ammonia preferentially reduces the Ni fluoride species (no N(1s) signal is observed in
the XPS) and forms nickel in its metallic form. After the initial decrease in the NiFx
species, the hydrogen then reduces the SiFx and there was no change in the total
silicon intensity silicon intensity. It was also observed that there was no uptake of
nitrogen in the spectra and this indicates that hydrogen is the main active ingredient in
the reduction of fluoride.
23
As shown in Fig 3.8 it is observed that the hydrogen mainly reacts with the
nickel and fluorine while the other elements remain largely unaffected. This process
might be due to the fact that the nickel fluoride species is the overlayer covering all
the other elements. The hydrogen is only able to react with the NiFx species. The
dissociated ammonia first has to reduce the NiFx over layer to reach the other
elements. The presence of a (now reduced) Ni overlayer will prevent further atomic H
interaction with SiO2. In any case, atomic H reduction of SiO2 does not become
spontaneous for T < ~ 1020 K due to free energy considerations in forming the SiO
intermediate 22. The silicon oxide requires higher temperatures to be reduced which
might contribute to the selectivity of the hydrogen reaction.
24
-693 -690 -687 -684 -681-500
0500
10001500200025003000350040004500
-890 -880 -870 -860 -850-500
0500
100015002000250030003500
-538 -536 -534 -532 -530 -528 -526-100
-500
50100150200250300350400
-108 -106 -104 -102 -100 -98 -96-50
0
50
100
150
200
After N F3 dose for 10 m in After N H 3 dose for 40 m in(tot) After N H 3 dose for 70 m in(tot) After N H 3 dose for 130 m in(to t) After N H 3 dose for 190 m in(to t)
F(1s)Ni(2p)
O(1s) Si(2p)
Intensity(CP
S)
Binding Energy(eV)
NiF2 decreases
Ni0, Not NiSi
Atomic H preferentially reduces NiF2
No increase in total or individual Si intensity
SiFxcomponent: ~ contstant after initial decrease
Fig 3.8: XPS spectra of F(1s), Ni(2p), O(1s) and Si(2p) after the reaction with NF3
with dissociated ammonia.
25
CHAPTER 4
SUMMARY AND CONCLUSIONS
4.1. Summary and Conclusions
Ni(Pt)Si is a material that is been recently used in the advanced
complementary metal oxide devices. It is a better material because of its reduced risk
of junction spiking, less consumption of Si, better compatibility with Ge, smoother
interfaces and less voiding. Newer methods of cleaning the surfaces are required for
better adhesion and nucleation of metal overlayers on the sample. XPS was used to
study the interactions of molecular and dissociated NF3 and NH3 with SiO2/Ni(Pt)Si.
Dissociation of NF3 and NH3 in the thermal cracker source results in F +NFx and H +
NHx (x = 1,2) respectively. No evidence of reaction was observed upon exposure of
the SiO2/Ni(Pt)Si surface to molecular NF3 or NH3. Further, no evidence of N surface
incorporation was observed in the presence of dissociated gases, indicating that
atomic F and H are the reactive species. Exposure of SiO2/Ni(Pt)Si to F results in
initial SiFx species, with eventual formation of a Ni difluoride overlayer on the
surface. It was also observed that Pt does not play any role in the process. The NiF2
layer is formed on the top of all the other metals as the intensity of all the other
elements was reduced. XPS spectra, however, indicate that the substrate SiO2 layer is
not removed by the dissociated NF3 and NiF2 growth process. The NiF2 over layer
can be reduced to metallic Ni by reacting with dissociated NH3 at room temperature.
The atomic hydrogen from dissociated ammonia reduces the NiF2 but it was
determined that the atomic hydrogen from the ammonia does not react with SiO2.
26
This reaction might be due to the fact that the NiF2 species is the top most
layer and is the first one to react with hydrogen which is summarized in figure 4.1.
The failure of atomic H to reduce SiO2 at room temperature is explained on
thermodynamic grounds. The chemical reaction for the removal of SiO2 is assumed to
be as follows:
SiO2(s) + 2H*(g) SiO(g) + H2O.
The Gibbs’ free energy at room temperature (298 K) is 94KJ, which means that it is
an endergonic or a non-spontaneous reaction. The estimation at which the reaction
becomes spontaneous is 1023 K or 750oC, and this is in good agreement with
experiment 22.
Fig 4.1: Illustration of the reaction of fluorine and NH3 on oxidized Ni(Pt)Si.
27
4.2. Future Work
Other methods to clean the oxide on the Ni(Pt)Si can be explored. We need to
know whether the oxide or the silicate on the surface is relevant. We also need to
explore the possibility of metallization of Ni after the reduction of NiF2. The surface
composition and the surface thermal compatibility need to be explored.
28
REFERENCES
(1) A. Lauwers JVST 2001, B.
(2) Chi, D. Z.; Lee, R. T. P.; Wong, A. S. W. Thin Solid Films 2007, 515, 8102-8108.
(3) Hsu, P. F.; Tsai, M. H.; Perng, B. C.; Shieh, J. H.; Tao, H. J.; Liang, M. S. Physical and Failure Analysis of Integrated Circuits, 2004. IPFA 2004. Proceedings of the 11th International Symposium on the 2004, 199-200.
(4) Chi, D Z, Mangelinck, D, Zuruzi, A S, Wong, A S W, Lahiri, S K Journal of Electronic Materials 2001, 30.
(5) Lavoie, C.; d’Heurle, F. M.; Detavernier, C.; Cabral, C. Microelectronic Engineering 2003, 70, 144-157.
(6) Lavoie, C.; Detavernier, C.; Cabral, J., C.; d’Heurle, F. M.; Kellock, A. J.; Jordan-Sweet, J.; Harper, J. M. E. Microelectronic Engineering 2006, 83, 2042-2054.
(7) Wang, R. N.; Feng, J. Y.; Huang, Y. Appl. Surf. Sci. 2003, 207, 139-143.
(8) Manandhar, S.; Copp, B.; Kelber, J. A. Appl. Surf. Sci. 2008, 254, 7486-7493.
(9) Castro, G. R.; Küppers, J.; Ivashchenko, Y. Applications of Surface Science 1983, 16, 453-462.
(10) Hiroki Ogawa, Tomoharu Arai, Michihiki yanagisawa, Takanori Ichiki, Yasuhiro Horiike JAP 2002, 41, 5349-5358.
(11) Manandhar, S.; Gaddam, S. S.; Kelber, J. Surf. Sci. 2009, 603, 2469-2473.
(12) Marek A. Szymanski, Alexander L. Shluger, and A. Marshall Stoneham Phys. Rev. Lett. 2001, 63, 224207.
(13) D. Briggs and M. P Seah, Ed.; In Practical surface analysis; Wiley: UK, 1983; Vol. 1, pp 657.
(14) Zhao, X.; Magtoto, N. P.; Leavy, M.; Kelber, J. A. Thin Solid Films 2002, 415, 308-314.
(15) P.J Gruthaner, F.J. Gruthaner, A.Madhukar JVST 1983, 20, 680.
(16) Čechal, J.; Šikola, T. Surf. Sci. 2006, 600, 4717-4722.
29
(17) Lewerenz, H. J.; Jungblut, H.; Rauscher, S. Electrochim. Acta 2000, 45, 4615-4627.
(18) Hanamoto, K.; Yoshimoto, H.; Hosono, T.; Hirai, A.; Kido, Y.; Nakayama, Y.; Kaigawa, R. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 1998, 140, 124-128.
(19) Tasaka, A.; Osada, T.; Kawagoe, T.; Kobayashi, M.; Takamuku, A.; Ozasa, K.; Yachi, T.; Ichitani, T.; Morikawa, K. J. Fluorine Chem. 1998, 87, 163-171.
(20) Hüfner, S. Solid State Commun. 1984, 49, 1177-1179.
(21) N. Cabrera, N. F. M. rep. prog. Phys. 1948, 12, 163.
(22) Aßmuth, A.; Stimpel-Lindner, T.; Senftleben, O.; Bayerstadler, A.; Sulima, T.; Baumgärtner, H.; Eisele, I. Appl. Surf. Sci. 2007, 253, 8389-8393.
30