Adherence/Diffusion Barrier Layers for Copper Metallization: …/67531/metadc4493/m2/... · I am...
Transcript of Adherence/Diffusion Barrier Layers for Copper Metallization: …/67531/metadc4493/m2/... · I am...
ADHERENCE/DIFFUSION BARRIER LAYERS FOR COPPER METALLIZATION:
AMORPHOUS CARBON:SILICON POLYMERIZED FILMS
Merry Pritchett, M.S.
Dissertation Prepared for the Degree of
DOCTOR OF PHILOSOPHY
UNIVERSITY OF NORTH TEXAS
May 2004
APPROVED: Jeffry A. Kelber, Major Professor Robert Desiderato, Committee Member Teresa Golden, Committee Member Rick Reidy, Committee Member Ruthanne D. Thomas, Chair of the Department of
Chemistry Sandra L. Terrell, Interim Dean of the Robert B.
Toulouse School of Graduate Studies
Pritchett, Merry, Adherence/Diffusion Barrier Layers for Copper Metallization:
Amorphous Carbon:Silicon Polymerized Films, Doctor of Philosophy (Analytical
Chemistry), May 2004, 102 pp., 4 tables, 35 illustrations, reference list, 127 titles.
Semiconductor circuitry feature miniaturization continues in response to Moore’s
Law pushing the limits of aluminum and forcing the transition to Cu due to its lower
resistivity and electromigration. Copper diffuses into silicon dioxide under thermal and
electrical stresses, requiring the use of barriers to inhibit diffusion, adding to the
insulator thickness and delay time, or replacement of SiO2 with new insulator materials
that can inhibit diffusion while enabling Cu wetting. This study proposes modified
amorphous silicon carbon hydrogen (a-Si:C:H) films as possible diffusion barriers and
replacements for SiO2 between metal levels, interlevel dielectric (ILD), or between metal
lines (IMD), based upon the diffusion inhibition of previous a-Si:C:H species expected
lower dielectric constants, acceptable thermal conductivity. Vinyltrimethylsilane (VTMS)
precursor was condensed on a titanium substrate at 90 K and bombarded with electron
beams to induce crosslinking and form polymerized a-Si:C:H films. Modifications of the
films with hydroxyl and nitrogen was accomplished by dosing the condensed VTMS with
water or ammonia before electron bombardment producing a-Si:C:H/OH and a-Si:C:H/N
and a-Si:C:H/OH/N polymerized films in expectation of developing films that would
inhibit copper diffusion and promote Cu adherence, wetting, on the film surface. X-ray
Photoelectron Spectroscopy was used to characterize Cu metallization of these a-
Si:C:H films. XPS revealed substantial Cu wetting of a-Si:C:H/OH and a-Si:C:H/OH/N
films and some wetting of a-Si:C:H/N films, and similar Cu diffusion inhibition to 800 K
by all of the a-:S:C:H films. These findings suggest the possible use of a-Si:C:H films
as ILD and IMD materials, with the possibility of further tailoring a-Si:C:H films to meet
future device requirements.
ii
Copyright 2003
by
Merry Pritchett
iii
ACKNOWLEDGEMENTS
I would like to thank my University of North Texas thesis committee, Dr. Teresa
Golden, Dr. Rick Reidy, Dr. Robert Desiderato, and Dr. Moon Kim from the University of
Texas – Dallas. I appreciate your insights, comments and suggestions.
I especially would like to express my appreciation to my advisor, Dr. Jeff Kelber,
for being such a “good guy” – working with my teaching schedule so that I could finish
my master’s degree and for giving me the opportunity to join his group and do really
interesting doctoral research. I’d also like to thank him for encouraging me to utilize
resources from other groups, enabling me to pursue a pharmaceutical internship, and
providing access to financial support from the Semiconductor Research Corporation
(SRC) -TI/SRC Customized Research, and Robert Welch Research Grants.
Much appreciation also goes to those who provided resources for me to extend
my research: Rick Reidy, for the use of the Materials Synthesis and Processing FTIR
instrument, Brian Gorman, for his FTIR expertise, Bob Wallace and Prakaipetch
Punchaipetch, for access to the Laboratory for the Electronic Materials and Devices.
I am truly grateful to Dr. Ed Dorsey for creating an internship at Alcon
Laboratories, Inc. where there was none, and Brett Scott, Leon Donesky, Ivy Badger
and Phillip Ritter, and Dr. Danny Dunn, for providing a year of invaluable, enjoyable,
learning experiences.
Immeasurable thanks go to my parents for memorable summer trips to national
parks and museums that cultivated in me a lasting appreciation for science and for their
constant encouragement and support during my studies.
iv
This thesis is dedicated to the memory of my grandmother, who constantly
encouraged me to “Hang in there”, to my father, who hung in there for me, and to God,
the creator of atoms, interactions and materials – Jehovah-jireh.
v
TABLE OF CONTENTS
Page ACKNOWLEDGMENTS ......................................................................................... iii LIST OF TABLES ................................................................................................... vii LIST OF ILLUSTRATIONS ..................................................................................... viii Chapter
1. INTRODUCTION .................................................................................... 1 1.1 Interconnect Structure and Resistance-Capacitance Delay ...... 1 1.2 Metallization of Interconnects ................................................... 5 1.3 Diffusion Barriers ...................................................................... 6 1.4 Dielectrics ................................................................................. 9 1.5 Adhesion ................................................................................... 11 1.6 Techniques and Instrumentation .............................................. 15
1.6.1 X-ray Photoelectron Spectroscopy (XPS) ...................... 15 1.6.2 Physical Vapor Deposition (PVD) .................................. 22 1.6.3 Precursor Deposition ...................................................... 25 1.6.4 XPS, PVD and Precursor Doser Apparatus .................... 26
1.7 Chapter References ................................................................. 28
2. HYDROXYLATED AMORPHOUS POLYMERIZED CARBON:SILICON FILMS - ADHERENCE /DIFFUSION BARRIERS FOR COPPER ........ 35
2.1 Motivation for Si:C:H Based Barrier Layer ................................ 35 2.2 Experiment ............................................................................... 38 2.3 Results ...................................................................................... 41
2.3.1. Amorphous-Si:C:H Film Characterization ....................... 41 2.3.2. Copper Deposition by Physical Vapor Deposition (PVD) 44 2.3.3. Annealing Effects ............................................................ 48
2.4 Discussion ................................................................................. 50 2.5 Conclusions .............................................................................. 54 2.6 Chapter References .................................................................. 56
vi
3. NITROGEN MODIFIED AMORPHOUS POLYMERIZED CARBON:SILICON FILMS - ADHERENCE /DIFFUSION BARRIERS FOR COPPER ........ 60
3.1 Motivation for Nitrogen Modification of a-Si:C:H Films ............. 60 3.2 Experiment ............................................................................... 63 3.3 Results ...................................................................................... 66
3.3.1. Nitrogen Modified a-Si:C:H Film Characterization .......... 66 3.3.2. Copper Deposition by Physical Vapor Deposition (PVD) 70 3.3.3. Annealing Effects ............................................................ 77
3.4 Discussion ................................................................................. 79 3.5 Conclusions .............................................................................. 82 3.6 Chapter References .................................................................. 84
4. RESEARCH SUMMARY ........................................................................ 88 4.1 Chapter References .................................................................. 91
REFERENCES........................................................................................................ 92
vii
LIST OF TABLES Table Page 1.1. Aluminum and Copper properties .............................................................. 5 1.2. Comparison of thermal mechanical properties of SOD and SiOC films .... 8 1.3. Dielectric constants of common dielectric materials................................... 10 2.1 Relative surface concentrations for Si, Cl and C on Al at different ............ CTCS coverages at room temperature ...................................................... 39 3.1. Binding Energies of a-Si:C:H Film Components (eV)................................. 67
viii
LIST OF ILLUSTRATIONS Figure Page 1.1. Tungsten via plug.................................................................................... 1 1.2. Comparison of intrinsic gate delay and interconnect (RC) delay as a function of feature size. ................................................................... 2 1.3. Microprocessor clock speed for different metals and dielectrics as a function of feature size. ................................................................... 3 1.4. Hillock and void formation ....................................................................... 4
1.5 Interconnect metallization structure......................................................... 6 1.6. Wetting described by contact angle and temperature ............................. 12 1.7. Horizontal approximation of the 180º Peel Test ...................................... 14 1.8. XPS electron emission process............................................................... 16 1.9. Nitrogen adsorbed on Ni(100) surface. ................................................... 17 1.10. Comparison of angle resolved and normal emission geometry............... 18
1.11. XPS Instrumentation. Angle resolved, θ = 15°, Si 2p spectrum ............. enhances the surface oxide species as compared to Normal Si 2p ....... spectrum, θ = 90°.. .................................................................................. 19 1.12. XPS Instrumentation. .............................................................................. 20 1.13. Physical Vapor Deposition process......................................................... 22 1.14 Three methods of metal film growth. (a) Frank van de Merwe (b) ........ Stranski - Krastanov (c) Volmer Weber................................................... 24 1.15. Uptake curves of Metal/Substrate versus metal deposition time depicting two
different growth modes: a) conformal growth b) 3-D island growth ..... 25 1.16. XPS and CVD and PVD system26 2.1. XPS spectra of unmodified a-Si:C:H film................................................. 39 2.2. XPS spectra of hydroxyl-modified a-Si:C:H film ...................................... 40
ix
2.3. XPS spectra of Cu deposition on unmodified a-Si:C:H film .................... 43 2.4. XPS spectra of Cu deposition on OH-modified a-Si:C:H film ................. 43 2.5. Cu(2p3/2)/Si(2p) uptake curve.................................................................. 44 2.6. Cu(L3VV) Auger peak XPS spectra of hydroxyl modified a-Si:C:H film ... during Cu deposition: ............................................................................ 45 2.7. Hydroxyl modified a-Si:C:H film intensity ratios versus anneal ............... temperature............................................................................................. 47 2.8. Cu Auger parameter indicates both Cu(I) and Cu(0) species are ........... deposited................................................................................................. 50 3.1. Unmodified a-Si:C:H film showing single O(1s), C(1s) and Si(2p) species ................................................................................................... 65 3.2. Comparison of XPS spectra (a) nitrogen modified a-SI:C:H film and (b) nitrogen-hydroxyl modified a-Si:C:H film. .............................................. 66 3.3 XPS Cu(2p) spectra showing absence of Cu(II) shake-up (satellite ) ..... peaks ...................................................................................................... 69 3.4 XPS spectra of copper deposition on nitrogen/hydroxyl modified .......... a-Si:C:H film ............................................................................................ 70 3.5. XPS Cu(2p)/Si(2p) Intensity ratio - Copper uptake curve: a) a-Si:C:H/N film b) a-Si:C:H film ................................................................................ 71 3.6 Cu(2p 3/2)/Si(2p) uptake curve indicating a change in deposition mode .. as copper is deposited on a-Si:C:H/N films............................................. 72 3.7. Cu(2p 3/2)/Si(2p) uptake curve indicating a change in deposition ............ mode as copper is deposited on a-Si:C:H/OH/N films............................. 72 3.8. Copper deposited on a-Si:C:H/N films and a-Si:C:H films show only .... Cu(0) species shifting to higher kinetic energy due to increase in cluster size and final state shielding ................................................................... 69 3.9. Copper deposited on a-Si:C:H/OH/N films shows Cu(I) and Cu(0) ........ species simultaneously, producing the broad peak shape. Copper ...... deposited on a-Si:C:H films show only Cu(0) species, indicated by ....... narrow peak shape that shifts to higher kinetic energy due to increase . in cluster size and final state shielding .................................................... 70
x
3.10. Copper diffusion is signaled by the rapid increase in the Cu(3p)/Cu(2p) Ratio........................................................................................................ 72
3.11. Copper Auger intensity decreases with anneal from 300 K –1000 K, ..... and the peak becomes sharper as it becomes more metallic.................. 73
1
CHAPTER 1
INTRODUCTION
1.1 Interconnect Structure and Resistance – Capacitance Delay
Interconnects are metal circuits that carry electron signals between transistors
and are basic to semiconductor technology. Multilayer structures are used to enable
smaller areas to incorporate more circuitry. Metal lines consisting of vias delineated by
Fig. 1.1 Tungsten via plug [1] (Used with permission, Thompson).
dielectric interlayers, alternate on a tungsten or silicon substrate., ending with a polymer
layer, capping layer, to keep corrosion and extraneous electrons out [2]. As
technological advancements enable semiconductor chips to get smaller, the smaller
feature sizes of chip components require much smaller integrated circuits and metal
2
lines connecting those circuits. The size of metal lines in interconnects significantly
affects the speed of electron flow and is based upon the metal’s resistance, R, cross
sectional area, A, length, L, and resistivity, ρ:
R = ρ L A
Therefore decreasing line length and area increases resistance, due to line width
constraints. The type of dielectric material largely determines the capacitance, C, given
by the dielectric constant, K, area, A, and thickness, t,:
C = K * (A/t) Decreasing feature sizes will decrease the A/t ratio, requiring lower dielectric constant
materials to achieve lower capacitance.
Intrinsic material properties significantly affect the conductivity of electrons as
feature sizes decrease. Two parameters that affect the speed an electrical signal are
Fig. 1.2 Comparison of intrinsic gate delay and interconnect (RC) delay as a function of feature size [3] (Used with permission, Barnes).
3
carried are transistor gate delay (the switching time of an individual transistor) and
resistance-capacitance (R-C) delay; the signal propagation time between transistors.
As shown in Figure 1.2, interconnect delay (R-C delay) increases as feature sizes
decrease, whereas transistor gate delay decreases. R-C delay is dependent on the
metal properties, feature sizes and properties of the dielectric [3]:
RC delay = 2ρε(4L²/P²+L²/T²)
R = metal wire resistance L = line length C = interlevel dielectric capacitance P = metal pitch ρ = metal resistivity T = metal thickness ε = ILD permittivity, or dielectric constant
The continued shrinkage of interconnect feature size has greatly increased the R-C
delay relative to gate delay (Fig. 1.2) and has motivated the switch from aluminum to
Fig. 1.3 Microprocessor clock speed for different metals and dielectrics as a function of feature size [3] (Used with permission, Barnes).
Cu-Low κ Al-Low κ
Cu-SiO2
Al-SiO2
Technology Min. Feature (nm)
250 180 150 30 100 70 50
Clo
ck F
req.
(M
Hz)
3200
2800
2400
2000
1600
1200
800
400
0
4
copper interconnects (lower R) and from SiO2 to lower dielectric constant insulators
(lower C) requiring the adoption of new materials and processes. In the transition from
aluminum to copper, and resultant adoption of new materials, the speed of processors
is related to R-C delay more than transistor gate delay (Fig. 1.3).
Aluminum has been the metal of choice on dielectric silicon dioxide due to an
interfacial passivation reaction that enables aluminum to adhere to SiO2 without
diffusing into the bulk. Increases in resistance and signal processing time due to
feature reduction, necessitate a metal with a lower intrinsic resistance than aluminum.
In addition, aluminum atoms have a tendency to migrate along with the flow of
electrons, electromigration, leaving voids at the starting point and piling up at the
Fig. 1.4 Hillock and void formation [1] (Used with permission, Thompson).
5
end point, creating hillocks (Figure 1.4) [4]. Both of these problems are remedied by
changing to copper. Copper is more conductive, with greater resistance to
electromigration and a higher melting point than aluminum as shown in Table 1.1.
The replacement of aluminum by copper in interconnects poses several
problems. First Cu does not form a passivation layer, therefore presenting a possible
oxidation problem, and it poorly adheres to most of the most commonly used dielectric
material, SiO2. In addition, copper readily diffuses into SiO2 [5] under conditions of
Table 1.1 Aluminum and Copper Properties (Used with permission, NIST).
N.A. Lange, Handbook of Chemistry pp. 100-107 (pp. 854-861) (Handbook Publishers, Sandusky, Ohio, 1956), *Thermometric fixed point; E.A. Brandes, Smithells Metals Reference Book, 6th ed., p. 16-2 (Butterworths, London, 1983)
combined bias and thermal stress (BTS) due to the lack of formation of an interfacial
passivation layer. In addition, copper does not wet SiO2. Therefore, copper requires
Metal At. Wt.
Density (20° C) (g/cm3)
Melt. Point (° C)
Heat capacity
(25° C) J/(kg.K)
Thermalexp. coef.
(20° C) (10-6 K-1)
Electrical Resistivity
(20° C) (10-6 W .cm)
Thermal conductivity
(W/(m.K))
Aluminum 26.98 2.702 660.46* 902.1 23.03 2.62 238
Copper 63.54 8.92 1084.9* 385.2 16.6 1.69 416
Gold 196.97 19.3 1064.4* 127.5 14.2 2.4 311
Silver 107.87 10.5 961.93* 236.3 18.9 1.62 417
6
the use of a barrier material that promotes copper adhesion and conformal growth
(wetting) as well as the inhibition of its diffusion into the dielectric.
1.3 Diffusion Barriers
In the past a number of different species have been explored as barrier layers,
e.g. TiN/Ti [6, 7] and TaN/Ta [8], or modifications of these e.g. W/TiN [9], TiSiN [10].
The problem with Ta barrier layers is that they can be contaminated with small amounts
of carbon and oxygen, inhibiting copper wetting and adhesion [11]. The formation of Ta
oxide at grain boundaries may result in the loss of Cu film continuity resulting in the
Fig. 1.5 Interconnect metallization structure [1] (Used with permission, Thompson).
7
formation of Cu silicide [12]. Therefore, Ta and TaN dewetting is foreseeable on the
proposed low-k dielectrics without stringent control of background contaminants during
copper deposition [13].
Other proposed barrier layers include Co(W,P) thin films [14], amorphous silicon
carbide (a-Si-C:H) films [15], and silicon oxynitride (SiOxNy) films [16]. Organic films
would be less reactive than Ta, but generally exhibit only weak bonding to Cu and
inhibition of Cu diffusion under Bias-Thermal-Stress conditions raises concerns about
film porosity and thermal and electrical stability. Additionally, the metal barrier and
dielectric substrate may have very different coefficients of thermal expansion, leading to
stress-induced delamination or cracking during thermal analysis. Non-metallic barriers
would have a significant advantage with respect to metal barriers, since their
coefficients of expansion would be more closely matched.
Amorphous silicon carbide barrier layers are formed by dissociating
trimethylsilane (TMS) gas by plasma enhanced chemical vapor deposition (PECVD), at
temperatures from 300 to 400 °C [17]. TMS films deposited on high conductivity silicon,
with gate electrodes of Cu, metal-insulator-semiconductor (MIS) capacitors, were
investigated by capacitance-voltage (C-V) and current-voltage (I-V) measurements.
These films resisted copper diffusion at a voltage of 2 MV for almost 105 seconds,
indicating a higher resistance than plasma silicon nitride, and after bias-temperature-
stress analysis, showed no evidence of copper diffusion [17, 18]. Trimethylsilane
(TMS) low-k oxide films, containing both Si-CH-Si and Si-O-Si bond linkages, are low-
density oxide materials with leakage currents and breakdown strengths similar to SiO2.
8
Stress-temperature analysis that show no apparent change in properties, with a-Si:C:H
used as the barrier layer, suggest their use as interlayer dielectrics [17].
SiOC films have been suggested as low-k barrier films due to their stability to
humidity and integration, superior hardness as compared to spin on films and thermal
characteristics similar to SiO2 as shown in Table 1.3 [19]. In addition, they do not form
pinholes or require solvent removal, they fill gaps well, and their thickness is easily
controlled. Such films, deposited by chemical vapor deposition (CVD), are compatible
with current processing equipment. In contrast, spin on dielectrics (SOD) can be
specifically designed, and have well-known structures such as Dow Corning’s FOx®
Flowable Oxides, hydrogen silsesquioxane (HSQ) [20], and Honeywell’s
NANOGLASS® E Extra Low-k Spin-On Dielectric [21]. However, they require solvent
removal, produce many pinholes, must be cured, often poorly control thickness and
SiO2 SiOC Organic SOD
Hardness (GPa) 9.0 2.0 0.3
Elastic modulus (Gpa) 80.0 10.0 2.5
Thermal conductivity (W/mk) 0.90 0.35 0.20
Thermal expansion CTE (ppm) 0.50 12 62
Table 1.2 Comparison of thermal mechanical properties of SOD and SiOC films [19] (Used with permission, Semiconductor Fabtech).
9
inadequately fill gaps [22]. Considering the mechanical and thermal advantages, as
compared to the technological requirements, SiOC films appear to be much better
choices for use as barrier layers than SOD films.
1.4 Dielectrics
Dielectrics, in a passive mode, insulate and isolate conductors from adjacent
conductors and underlying silicon. Ideally, they should act to hermetically seal
conductors and devices from impurities such as water or corrosive ions that lead to
device instabilities. Passivating films should have few pinholes and keep out moisture.
Barriers to silicon should keep out alkaline ions, or scavenge them. Two reliability
measures for insulators are current leakage and breakdown.
Silicon dioxide is the most commonly used dielectric for aluminum metallization.
However, decreasing feature sizes necessitates dielectrics with lower resistivity, to
counteract “cross-talk” between closely spaced metal lines resulting from capacitive
coupling. Changing the dielectric material from silicon dioxide has initiated the
development of inter-layer and intra-layer (gap-fill) dielectrics of various organic
polymers, e.g. polystyrene or polyimide as (Table 1.3) [23]. Polyimides have the
undesirable ability to incorporate copper in the films, leading to Cu diffusion under BTS
conditions.
Other interlevel dielectrics (ILDs), dielectrics separating electrical conductors,
include the TMS low-k oxide film (silicon oxynitride), because of its similarity to silicon
dioxide, and other low-k films such as Dow Chemical’s SiLK™ low-k spin on dielectric
10
films [24], that do not require adhesion promoters. Low-k methylsilsesquioxane (MSQ)-
based α-SiOC:H films, deposited by plasma enhanced chemical vapor deposition
(PECVD) from trimethylsilane gas with tantalum nitride as adhesion promoter and
copper diffusion barrier [25], are also suggested for use as ILDs.
Other possible ILDs are porous silica xerogels, optically transparent
interconnected clusters of SiO2 of a small size, 20 nm, that may also serve as cladding
materials in optical interconnects, and suggested for use as alternatives to metal
interconnects. Copper adheres to α-SiOC:H films with thermal diffusion inhibition to ~
700 K [26]. SIOC films are promoted for use as intermetal dielectrics due to their
Material Dielectric constant (k)
Deposition method
Silicon dioxide (SiO2) 3.9-4.5
Fluorosilicate Glass (FSG) 3.2-4.0
Organosilicate Glass (SiOCH or SiCOH) 2.7-3.3
Fluorinated amorphous carbon (a-C:F) 2.4
PECVD
Parylene-N 2.7
Parylene-F 2.4-2.5 CVD
Polyimides 3.1-3.4
Fluorinated Polyimides 2.5-2.9
Benzocyclobutene polymers (BCB) 2.7
Hydrogen-Silsesquioxane (HSQ) 2.9-3.2
Methyl-Silsesquioxane (MSQ) 2.6-2.8
Poly(arylene ether) 2.8-2.9
Aromatic thermosets (SiLK) 2.6-2.7
Teflon AF 2.1
Porous silica (aerogel or xerogel) 1.1-2.4
Spin-on
Table 1.3 Dielectric constants of common dielectric materials [19]
(Used with permission, Semiconductor Fabtech).
11
higher thermal and mechanical stability as compared to organic films, and the ability to
tailor their density and refractive index by increasing their carbon content. However,
SIOC films take up oxygen radicals upon ashing [27]. Organosilicate glass (OSG) films
[28, 29] atomic layer deposition (ALD) Si:O:H films, mesoporous silica films, as well as
dual damascene structures of SiN/SiO2 have also been suggested.
VLK (very low k) films have been proposed as intra-layer dielectrics (trench level)
with SiO2 as the inter-layer (via level) dielectric, to reduce wiring capacitance and avoid
the lower thermal conductivity that results from full low-k (using low-k for both intra and
inter-layer dielectrics) [30]. In addition, HSQ, siloxane-based hydrogen silsesquioxanes
(HSiO3/2)2 have been proposed as an intermetal dielectric, with ammonia treatment to
form a nitride passivation layer inhibiting copper diffusion into these films [31].
1.5 Adhesion
In metallization, adhesion to the substrate is very important. The strength and
durability of an adhesive bond is significantly dependent on the interactions between the
adhering surfaces, functions of the chemical states, which can be monitored by XPS.
Good adherence means the film-substrate interface is not physically affected during
fabrication or expected use when exposed to reasonably low stress levels. Accepted
mechanisms of adhesive/adherend adhesion are based on mechanical interlocking,
surface energies, wetting adsorption, diffusion theory, and chemical bonding [32].
Wetting involves the establishment of intimate molecular contact described by
the contact angle, θ, of the liquid or molten species on the solid substrate, with smaller
12
contact angles corresponding to greater adherence, or wetting, between the film and the
substrate as a result of the interatomic-interactive forces acting across the film-substrate
interface. Therefore, intermolecular wetting is a prerequisite for developing strong
adhesion [32].
The mechanism of adhesion involves interactions of electrostatic forces or
chemical bonding. Atoms at the interface, held together by electrostatic forces when
ionized with opposite polarities, or polarization that creates dipoles, form attractions that
align to cause weak van der Waals type electrostatic bonding across the interface, that
are not strong enough to withstand dewetting. Larger polarizabilities and vibrational
frequencies increase the adhesion strength at the interface, resulting in partial wetting
Fig. 1.6 Wetting described by contact angle and temperature [34] (Used with permission, Kluwer Academic/Plenum Publishers).
Temperature
Con
tact
ang
le, θ
Dewetting
Partial Wetting
Complete Wetting0
13
[33]. Chemical bond formation at the interface is stronger than electrostatic or van der
Waals forces, resulting in much stronger adhesion across the interface and complete
wetting.
For example, Al wets SiO2 through bond formation from oxygen reduction,
whereas Cu may partially, or initially weakly, wet the surface due to van der Waals
forces. This is attributed to the relatively large difference in linear coefficient of thermal
expansion between Cu and SiO2 (Cu =16.5 x 10–6, SiO2 = 0.5 X 10–6). A strong bonding
force, typically supplied by an adhesion promoter, is needed to keep Cu films from
dewetting on SiO2.
Good adhesion requires strong inter-atomic bonding across the film-substrate
interface, low level of film stress, and absence of reactive environment. The more the
free energy is lowered, the stronger the adhesion. For adhesion agents to work they
must promote interaction between the interface surfaces. The mechanism of adhesion
involves the mechanical interlocking of the deposited metal with the polymer [35].
Titanium is often used as an adhesion promoter because it forms lower free
energy compounds with oxygen, nitrogen and carbon. Coupling agents, such as
organosilanes, are used to react with both the organic and inorganic surfaces by
forming a hydrophobic layer at the interface [36]. For polystyrene, this is done by
dissolving the coupling agent with the polymer molecules, e. g. silane, producing silane
anchoring sites that result in the polymerization of the molecule, with increased amounts
of silane producing increased adhesion [36]. Adhesion promotion is also accomplished
by irradiating individual polymer molecules with ultra-violet radiation [37], or electron
beams [38], or thermal curing [39], to polymerize the molecule by cross-linking.
14
In multilevel interconnect structures that use polymer dielectrics there are two
types of interfaces that have very different interface reactions, metal-on-metal and
polymer-on-metal. XPS studies have shown that reactive metals deposited on
polyimide (PI) break carbonyl bonds, however non-reactive metals such as Cu do not.
For copper, weak adhesion is often improved by plasma [40] or ion beam bombardment
[33], ammonia treatment of the polymer surface [32] or by inserting an adhesion layer
(Cr, Ti, Ta, TiN, TaN), depending on the properties of the metal/polymer interface at the
microscopic level.
Adhesion tests can be qualitative or quantitative. A simple qualitative test often
used to determine a minimum strength of adhesion between surface species is the
“Scotch tape test” [41-45]. In this test, a strip of tape is pressed on the top of the
deposited substance, and pulled back at approximately 180 ° in one smooth movement,
similar to a horizontal 180 º Peel Test shown in Figure 1.7 (P = applied load), and
examined for any film removed [46-48]. The European Metallizers Association
Technical Committee Films suggests the use of Scotch Tape 610 or equivalent with a
peel-force of approximately 3 N/15 mm measured against a steel-plate [49].
Fig. 1.7 Horizontal approximation of the 180 º Peel Test [46] (Used with permission, Elsevier).
P
P
15
Qualitative examination of the test substrate will reveal any metal transfer, with a strong
backlight facilitating visualization of any metal removed. The remaining exposed
surface is then analyzed for the deposited substance to see if any of it remains.
Quantitative measurements of adherence include the actual thickness of the deposited
film that remains after the “Scotch Tape Test” [46], and the peel-force when a calibrated
tape is used.
1.6 Techniques and Instrumentation
1.6.1 X-ray Photoelectron Spectroscopy
X-ray photoelectron spectroscopy (XPS) is a technique often used to analyze the
chemical properties of sample surfaces and thin films. In this technique x-rays are used
as a photon source to irradiate the sample to a depth of about 10 µm causing the
emission of electrons and any excess energy emitted as either fluorescence or, more
likely, as Auger electrons [50]. In XPS, photons with a minimum energy, hν, are
absorbed by atoms causing core electrons to be ejected from the sample. If the photon
energy, hν, is greater than the binding energy of the electron (EB), the excess energy is
the kinetic energy of the emitted photoelectron (EK). Calculation of the kinetic energy of
the emitted photoelectrons is the difference between the kinetic energy of the incident x-
rays, the measured binding energy and the work function, Φs, which is constant for that
specific spectrometer (Fig. 1.8)
EK = hν - EB - Φs
16
The binding energy of the photoelectron can be calculated from the measured
kinetic energy, K.E., work function, energy of the incident x-rays and any charging of the
sample surface, due to a sufficiently insulating sample that develops an imbalance of
charge and develops a potential difference relative to the spectrometer. This can be
calibrated by referencing peak energies relative to a calibrated internal standard e.g. C
1s 285.0 eV, to counteract charging effects [51, 52]:
EB = EK - hν- Φs - Ech
Typically, electrons with binding energies of 100 to 1000 eV are involved in these
transitions, with each chemical species producing electrons with unique energies, and
descriptive peak shapes, enabling XPS to differentiate between atoms with different
oxidation states e.g. Si 2+ and Si4+ by the shift in the peak binding energy called the
chemical shift. The chemical shift is based on the change in electrostatic potential on
the core electron when valence electron charge density is accepted or withdrawn from
Fig. 1.8 XPS electron emission process (Used with permission, Moulder & Froelich).
K (1s)
L1 (2s)
L2,3 (2p)
K (1s)
L1(2s)
L2,3 (2p)
hν
KE
KE
Initial state Final state Final stateK (1s)
L1 (2s)
L2,3 (2p)
K (1s)
L1 (2s)
L2,3 (2p)
K (1s)
L1(2s)
L2,3 (2p)
K (1s)
L1(2s)
L2,3 (2p)
hν
KE
KE
Initial state Final state Final state
17
the atom, enabling XPS for use in determining the chemical composition of surfaces and
the effects of processes on sample surfaces [50, 53]. Since binding energies of core
electrons are characteristic for elements in certain chemical environments, XPS also
enables determination of the electronic structure and band structure of interacting
species [50]. Observed chemical shifts facilitate descriptions of local coordination in a
system and the electronic change in oxidation state on adsorption, which is useful in
distinguishing adsorption sites of molecules adsorbed on surfaces, e.g. N2 (Fig. 1.9)
Fig 1.9 Nitrogen adsorbed on Ni(100) surface [54] (Used with
permission, Nilsson).
420 415 410 405 400 395
Binding energy (eV)
N 2 /Ni (100)
18
adsorbed on a Ni(100) surface. In Figure 1.9 the two N 1s peaks have a chemical shift
of 1.3 eV. The 399.4 eV peak corresponds to ionization of the outer N atom, and the
400.7 eV peak is due to ionization of the inner N atom [54].
XPS spectra can be used quantitatively because the intensity, I, of the peaks is
related to the atomic concentration of the species in the sample, N, the sensitivity of the
species to surface effects, A, as relative ratio [11, 50]:
ABB
BAA
AIN
AIN =
Changing the geometry of an instrument affects the relative intensities of different peaks
in analysis [55]; therefore, the intensity ratios should be calculated between results
obtained on calibrated instruments with the same geometric arrangement. XPS signals
measured at angles that are more surface sensitive (Figs. 1.10, 1.11), can be used to
Fig. 1.10 Comparison of angle resolved and normal emission geometry.
Mean Free PathMean Free Path
Surfacenormal
Escape Depthe - e-θ
Normal Emission, Angle Resolved,
19
approximate the relative thickness of a thin film, <100 nm, and indicate if a species is on
the surface or in the bulk of a sample, using the relationship between intensity and
thickness [38]:
)/()(
λdTiTi II −=
0
Auger spectra produced by XPS are the result of an x-ray induced photoelectron
ejection triggering an Auger electron emission from a valence energy level, and
consequent electron transfer from an outer energy level into the vacated position, core
hole, created by the photoelectron [50]. Auger electrons also give information about the
Fig 1.11 Angle resolved, θ = 15°, Si 2p spectrum enhances the surface oxide species as compared to Normal Si 2p spectrum, θ = 90° [56] (Used with permission, Briggs).
chemical species in the sample because the kinetic energy of the Auger electron is
directly related to the core binding energy of the photoelectron:
θ = 90°
θ = 15°
Binding Energy (eV)114 94
XPS
Inte
nsity
θ = 90°
θ = 15°
Binding Energy (eV)114 94
XPS
Inte
nsity
20
2,311 LLK2,3LKL EEEE −−=
Additional information about the type of species is given by the shape of the Auger
peak. Auger spectra are referenced to the vacuum level or the Fermi level, however the
standard vacuum level is defined as 4.500 eV above the Fermi level which is close to
the uncharacterized vacuum level that varies between instruments, so only the Fermi
level reference is absolute [57].
Figure 1.12 [58] schematically illustrates the basic parts of an x-ray photoelectron
spectroscopy instrument. The emitted electrons are routed through lenses, which
focus electrons upon the input slit, and retard electron kinetic energies, by a pre-set
retarding voltage, ER, collimated and focused on the entry slits of the cylindrical
Fig. 1.12 XPS Instrumentation [58] (Used with permission, Fairley).
x - ray gun
Samplestage
ion gun
slits
Host computer
Hemispherical Analyzer
Detector head
HT electronics
21
hemispherical analyzer. The hemispherical analyzer only allows electrons with
stipulated energies, determined by voltage parameters set for the lens system, to
proceed on to the detector. This energy, the pass energy, EP, is determined by the
operator and set in the parameters of the operational software.
The hemispherical analyzer and transfer lenses are generally operated in the
Fixed Analyzer Transmission (FAT) mode, also known as Constant Analyzer Energy
(CAE) mode. In CAE mode, electrons with kinetic energy, EK, lose energy by
retardation against an electrostatic potential, VR, prior to entry into the analyzer where
they pass with kinetic energy EP so that:
EK = e VR+ EP
Only electrons with EK = e VR+ EP will pass through the analyzer and be detected.
Thus, sweeping the electrostatic potential, VR, linearly with time allows the
measurement of the kinetic energy distribution of the photoelectrons emitted from the
sample:
EP = EK – ER
Whereas for Auger electrons the relationship is a ratio of electron energy and the
retardation ratio, R:
EP = EK /R
Most XPS spectra are acquired using CAE mode.
The resolution of XPS spectra is determined by the full width half maximum
(FWHM) of the detected lines affected significantly by the diameter of the analyzer, pass
energy and spread of energies in the x-ray source. In addition charging can interfere
22
with emitted energies causing broadening of the spectral lines. Non-monochromated x-
ray sources provide relatively large throughput to examine large surface areas [59].
1.6.2 Physical Vapor Deposition
Metallization of films is often accomplished by sputtering the surface of the film
with atoms of the desired metal or alloy. Physical vapor deposition (PVD) (Fig. 1.13)
has been widely used for this purpose by the semiconductor industry because most
films used in VLSI (very large scale interconnects) are alloys and PVD can provide
uniform alloy composition across the surface area of wafers while providing good step
coverage. Metal films commonly deposited by sputtering include Al-Cu, Ti-W, TiN, Pt,
and individual metals for specific purposes.
Fig. 1.13 Physical Vapor Deposition process [60] (Used with permission, University of Virginia).
23
For magnetron sputter deposition, an inert gas, often argon, is exposed to a glow
discharge created by applying dc voltage between the anode and cathode, at pressures
from 10-4 Torr to 1.0 Torr [61, 62]. The glow discharge ionizes the gas atoms and
accelerates them toward the metal target (cathode) with energies between 500 and
2000 eV [62], which then dislodge (sputter) metal atoms from the cathode. Sputtered
metal atoms with energies between 3 to 10 eV [62], leave the target, propelled toward
the sample attached to the anode, in a loop induced by a magnetic field. Metal atoms
with energies less than 5 eV [63] condense to form a thin film on the sample surface.
Those with energies > 5 eV can eject from the sample surface and readsorb on the
target surface. Contaminants in the PVD chamber are also excited by the impinging
argon ions, and atoms or ions with energies greater than 10 eV can damage the sample
surface creating defects, artifacts and induced extraneous reactions as shown in Figure
1.13 [60]. Therefore, the energy used is just enough to accomplish sputtering while
limiting extraneous interactions.
XPS is used to monitor the interaction of the deposited metal and the substrate
by plotting the ratio of the metal species intensity compared to the substrate intensity
against metal deposition time. Strong interactions between a deposited metal species
with free energy, γM, and the substrate, with free energy, γS, result in the metal
preferentially adhering to the substrate and conformal growth (Fig. 1.14(a)) for several
monolayers of deposition, termed Frank van der Merwe (F-M) growth, indicated by
several changes in slope of the metal deposition uptake curve.
γM < γS + γMS
24
Fig. 1.14 Three methods of metal film growth (a) Frank van de Merwe (b) Stranski - Krastanov (c) Volmer Weber
Stranski-Krastanov (S-K) growth begins with strong interactions between the deposited
metal species and the substrate that are overcome by the increase in interface energy
as the layer thickness increases due to strain to try to continue to fit the substrate. This
yields initial conformal growth followed by 3-D island formation indicated by a change in
the slope of the uptake curve (Fig. 1.14(b)). Weak interactions between the deposited
metal (M) and the substrate (S) result in the metal preferentially adhering to itself
resulting only in the formation of three dimensional islands (Fig. 1.14(c)), termed
Volmer-Weber (V-W) growth, indicated by a curve with an unchanging slope as
depicted in Figure 1.15.
(a)SPUTTER DEPOSITION
λMS > λM
Conformal growth
Free energy, λ, favors metal-substrate attraction
(c) λMS < λM
3-D islandingFree energy, λ, favors metal-metal attraction
Conformal growth with 3-D islanding
(b)
25
Fig. 1.15 Uptake curves of Metal/Substrate versus metal deposition time depicting two different growth modes: a) conformal growth b) 3-D island growth.
1.6.3 Precursor Deposition
Precursor deposition was accomplished by a simplified adaptation of Chemical
vapor deposition (CVD). CVD is a process involving the formation of a nonvolatile solid
film on a substrate by the reaction of vapor phase chemicals obtained from gases,
liquids, sublimable solids or combination materials that are stable at room temperature
and volatile with high partial pressure [33]. By definition the process introduces a given
composition of reactant gases and diluent inert gases into a reactant chamber; the gas
species move to the substrate and the reactants are absorbed on the substrate then the
adatoms undergo migration and film-forming chemical reactions after which the
gaseous by-products of the reaction are desorbed and removed from the reaction
M/S
XP
S In
tens
ity R
atio
M/S
XP
S In
tens
ity R
atio
0 2 4 6 8 10
Deposition Time or Coverage
No change in slope of curve
0 2 4 6 8 10
Deposition Time or Coverage
1st growth mode
2nd growth mode
0
2
4
6
8
0
2
4
6
8
(a) (b)
26
chamber [64]. In this simplified system the precursor is introduced into the reactant
chamber, without the need of a solvent. Deposition is carried out at a very low
temperature [65] to immobilize the species on the substrate surface and allow the
incorporation of species such as hydroxide or water [66], and enable UV or plasma
irradiation or ion bombardment, to inculcate the film with the desired properties. To get
good growth rates the reaction temperature should be lower than the melting point of
substrate and produce the desired species on the substrate with easily removable by-
products and low toxicity [67].
1.6.4. XPS, PVD and Precursor Doser Apparatus
The apparatus for physical and chemical vapor deposition combines the ultra-
high vacuum XPS system and DC magnetron PVD chamber depicted in Figure 1.16.
1.16 XPS, PVD and Precursor Doser System.
Residual Gas Analyzer
NH3 /H
2 O Doser
XPSPVD
Tower
SampleIntroduction
e- Beame- Beam
ion gunion gun
TurboPump
PVD Chamber(1 x 10-8 Torr)
Main Chamber(5 x 10-10 Torr)
Turbo Pump(backside)
VTMS Doser
Main Chamber(5 x 10-10 Torr)
27
The main chamber is equipped for CVD with a direct doser for film precursor
introduction, and a backfill doser for water and or ammonia introduction, ion-sputtering
gun for sample cleaning, a dual anode ray source and hemispherical analyzer for XPS
sample analysis. The PVD chamber consists of a water cooled sputter target about 10
cm from the sample and a magnetron power source monitored by a baratron gauge,
isolated from the main chamber by a gate valve.
28
1.7 Chapter References
[1] Thompson, C. V., "Interconnects", http://hackman.mit.edu/6152J/LectureNotes/
6.152J.FT01.Lecture19-3.pdf, 2003.
[2] Dauskardt, R. H., Lane, M., Ma, Q. and Krishna, N., Engineering Fracture
Mechanics 61 (1998) 141-162.
[3] Barnes, C. and Lakins, D., http://nepp.nasa.gov/index_nasa.cfm/934/,
"Technology Readiness Overview - Low-k Interlevel Dielectrics Technology -
Brief description of low-k technology", NASA/NEPP Program, 2003.
[4] Rodbell, K. P., in G. C. Schwartz, K. V. Srikrishnan and A. Bross (Eds.):
Handbook of Semiconductor Interconnection Technology, Marcel Dekker, Inc
1998, p. 473-506.
[5] Sun, Y.-M., Lee, S. Y., Lemonds, A. M., Engbrecht, E. R., Veldman, S., Lozano,
J., White, J. M., Ekerdt, J. G., Emesh, I. and Pfeifer, K., Thin Solid Films 397
(2001) 109-115.
[6] Baumann, J., Werner, T., Ehrlich, A., Rennau, M., Kaufmann, C. and Gessner,
T., Microelectronic Engineering 37/38 (1997) 221-228.
[7] Marcadal, C., Richard, E., Torres, J., Palleau, J., Ulmer, L., Perroud, L., Piaguet,
J. and Rolland, G., Microelectronic Engineering 37/38 (1997) 197-203.
[8] Tsai, M. H., Sun, S. C., Lee, C. P., Chiu, H. T., Tsai, C. E., Chuang, S. H. and
Wu, S. C., Thin Solid Films 270 (1995) 531-536.
[9] Baumann, J., Markert, M., Wemer, T., Ehrlich, A., Rennau, M., Kaufmann, C. and
Gessner, T., Microelectronic Engineering 37/38 (1997) 229-236.
29
[10] Blanquet, E., Dutron, A. M., Ghetta, V., Bernard, C. and Madar, R.,
Microelectronic Engineering 37/38 (1997) 189-195.
[11] Chen, L., Magtoto, N., Ekstrom, B. and Kelber, J., Thin Solid Films 376 (2000)
115-123.
[12] Yin, K.-M., Chang, L., Chen, F.-R., Kai, J.-J., Chiang, C.-C., Chuang, G., Ding,
P., Chin, B., Zhang, H. and Chen, F., Thin Solid Films 388 (2001) 27-33.
[13] Laurila, T., Zeng, K., Kivilahti, J. K., Molarius, J., Riekkinen, T. and Suni, I.,
Microelectronic Engineering 60 (2003) 71–80.
[14] Kohn, A., Eizenberg, M., Shacham-Diamond, Y., Israel, B. and Sverdlov, Y.,
Microelectronic Engineering 55 (2001) 297-303.
[15] Tong, J., Martini, D., Magtoto, N., Pritchett, M. and Kelber, J., Appl. Surf. Sci. 187
(2002) 256-264.
[16] Albertin, K. F., Pereyra, I. and Alayo, M. I., Materials Characterization 5568
(2003) .
[17] Loboda, M. J., Microelectronics Engineering 50 (2000) 15-23.
[18] Vogt, M., Kachel, M., Plötner, M. and Drescher, K., Microelectronic Engineering
37/38 (1997) 181-187.
[19] Gotuaco, M., Lee, P. W., Xia, L. and Yieh, E., Semiconductor Fabtech 15 (2001)
179-183.
[20] Dowcorning, http://www.dowcorning.com/applications/Product_Finder/
PF_details.asp?l1=009&prod=02334909&type=PROD&pg=00000037, " FOX(R)
-12 FLOWABLE OXIDE", Dow Corning, Midland, Michigan, 2003.
30
[21] Chapman, L., http://www.electronicmaterials.com/news_events/download/
111401.pdf, "International SEMATECH Fabricates Electrically Yielding 2-Level
Metal Copper Dual Damascene Via Chains with NANOGLASSâ E Extra Low-k
Spin-On Dielectric", www.electronicmaterials.com., Sunnyvale, California, 2003.
[22] Shim, C. and Jung, D., http://tflab.skku.ac.kr/low-k/low-k%20introduction
%202003.pdf, "Introduction to Inter-Metallic Dielectrics & Cu Integration for
beginners in our lab.", Semiconductor & Nano Application Lab Dept. of Physics,
Sungkyunkwan Universtiy, Chunchun-Dong, Jangan-Gu, Suwon, 440-746, South
Korea, 2003.
[23] Godbey, D. K., Buckley, L. J., Purdy, A. P. and Snow, A. W., Thin Solid Films
308-309 (1997) 470-474.
[24] Lee, G.-W., Lee, J.-H., Kim, Y.-M., Lee, W.-C., Kim, H.-K., You, K.-D. and Park,
C.-K., Journal of the Korean Physical Society 41 (2002) 912-917.
[25] Ahn, J.-H., Lee, K.-T., Oh, B.-J., Lee, Y.-J., Liu, S.-H., Jung, M.-K., Kim, Y.-W.
and Suh, K.-P., Journal of the Korean Physical Society 41 (2002) 422-426.
[26] Ishimaru, T., Shioya, Y., Ikakura, H., Nozawa, M., Ohgawara, S., Ohdaira, T.,
Suzuki, R. and Maeda, K., Journal of The Electrochemical Society 150 (2003)
F83-F89.
[27] Kim, H. J., Shao, Q. and Kim, Y.-H., Surface and Coatings Technology 171
(2003) 39–45.
[28] Vella, J. B., Adhihetty, I. S., Junker, K. and Volinsky, A. A., International Journal
of Fracture 119 (2003) 487-499.
31
[29] Jean, S.-K. J., Wang, Y.-L., Liu, C.-P., Hwang, W.-S., Tseng, W.-T. and Liu, C.-
W., Journal of Applied Physics 94 (2003) 732-737.
[30] Kudo, H., Yoshie, Y., Yamaguchi, S., Watanabe, K., Ikeda, M., Kakamu, K.,
Hosoda, T., Ohhira, K., Santoh, N., Misawa, N., Matsuno, K., Wakasugi, Y.,
Hasegawa, A., Nagase, K. and Suzuki, T., Interconnect Technology Conference,
2000. Proceedings of the IEEE 2000 International , 5-7 June 2000, Burlingame,
CA, 2000, IEEE, (2000) 270 -272.
[31] Liu, P.-T., Chan, T.-C., Yang, Y.-L. and Cheng, Y.-F. S., S.M., IEEE Transactions
on Electron Devices 47 (2000) 1733 -1739.
[32] Kim, S. H., Lee, D. W., Chung, K.-H., Park, J. K., Jaung, J.-Y. and Jeong, S. H.,
Journal of Applied Polymer Science 86 (2002) 812–820.
[33] Murarka, S. P., Verner, I. v. and Gutmann, R. J., Copper-Fundamental
Mechanisms for Microelectronic Applications, John Wiley & Sons, New York,
2000.
[34] Yost, G., Hosking, F. M. and Frear, D. R., The Mechanics of Solder Alloy Wetting
& Spreading, Van Nostrand Reinhold, New York, 1993.
[35] Wu, P. K., Dabral, S., Yang, G.-R., Gittleman, B., Li, C., McDonald, J. F. and Lu,
T.-M., J. Appl. Phys 80 (1996) 5759-5764.
[36] Namkansiorn, A., Ghatak, A., Chaudhury, M. K. and Berry, D. H., J. Adhesion
Sci. Technol. 15 (2001) 1725-1745.
[37] Thundat, T. and Sharp, S. L., Applied Physics Letters 66 (1995) 1563.
[38] Chen, L. and Kelber, J. A., J. Vac. Sci. Technol. A 17 (1999) 1968.
[39] Thundat, T. and Sharp, S. L., Applied Physics Letters 75 (1999) 3998-4000.
32
[40] Shenton, M. J., Lovell-Hoare, M. C. and Stevens, G. C., J. Phys. D: Appl. Phys.
34 (2001) 2754–2760.
[41] Ding, P. J., Lanford, W. A., Hymes, S. and Murarka, S. P., J. Appl. Phys. 75
(1994) 3627 - 3631.
[42] Chang, C.-A., Baglin, J. E. E., Schrott, A. G. and Lin, K. C., Appl. Phys. Lett. 51
(1987) 103.
[43] Liu, G. and Freund, M. S., Macromolecules 30 (1997) 5660-5665.
[44] Chan, C. M., Cao, G. Z., Fong, H., Sarikaya, M., Robinson, T. and Nelson, L., J.
Mater. Res. 15 (2000) 148-154.
[45] Rye, R. R., Knapp, J. A., Chi, K.-M., Hampden-Smith, M. J. and Kodas, T. T., J.
Appl. Phys. 72 (1992) 5941-5947.
[46] Anderson, G. P., Bennett, S. J. and DeVries, K. L., Analysis and Testing of
Adhesive Bonds, Academic Press, New York 1977, p. 15-16.
[47] www.sheldahl.com, "http://www.sheldahl.com/Product/TMQualityAssur.htm",
2003.
[48] Schwartz, G. C., in G. C. Schwartz, K. V. Srikrishnan and A. Bross (Eds.):
Handbook of Semiconductor Interconnection Technology, Marcel Dekker, Inc.,
New York 1998, p. 77-159.
[49] Films, E. M. A. T. C., http://www.eurometallizers.org/download/testmetaltape.pdf,
"TEST PROCEDURE METAL ADHESION TAPE TEST", European Metallizers
Association, 2003.
[50] Briggs, D. and Seah, M. P., in D. Briggs and M. P. Seah (Eds.): Practical Surface
Analysis, Vol. 1, Wiley, New York 1990, p. 657.
33
[51] Nuzzo, R. G.,
http://augustus.scs.uiuc.edu/nuzzogroup/PPT/XPS%20Class%2099.PPT, "X-Ray
Photoelectron Spectroscopy (XPS)", University of Illinois at Urbana-Champaign,
Urbana, Illinois, 2003.
[52] Barr, T. L., J. Vac. Sci. Technol. A 9 (1991) 1793-1805.
[53] McIntyre, N. S. and Chan, T. C., in D. Briggs and M. P. Seah (Eds.): Practical
Surface Analysis, Vol. 1, Wiley, New York 1990, p. 485-529.
[54] Nilsson, A., http://www-ssrl.slac.stanford.edu/nilssongroup/pages/
XPS_description.html, "X-ray Photoelectron Spectroscopy (XPS)", Surface
Science and X-Ray Spectroscopy Group, Sand Hill Road, CA, 2003.
[55] M.P.Seah, P.J.Cumpson and M.T.Brown, http://www.npl.co.uk/
nanoanalysis/x1userguide.pdf, "NPL X-Ray Photoelectron Spectrometer Intensity
Calibration Software, Version 2.0User’s Guide", Crown, Middlesex, UK, 2000.
[56] Briggs, D. and Riviere, J. C., in D. Briggs and M. P. Seah (Eds.): Practical
Surface Analysis, Vol. 1, Wiley, New York 1990, p. 657.
[57] Seah, M. P. and Smith, G. C., in D. Briggs and M. P. Seah (Eds.): Practical
Surface Analysis, Vol. 1, Wiley, New York 1990, p. 534-540.
[58] www.casaxps.com, http://www.casaxps.com/help_manual/
XPSInformation/XPSInstr.htm, "XPS Instrumentation diagram", 2003.
[59] Beebe, J. T. P., http://www.uwo.ca/ssw/archives/1999/nov99/0106.html 2003.
[60] Zou, W., Ph. D. Dissertation, University of Virginia, Charlottesville, Virginia 2001,
p. 303.
34
[61] Wasa, K. and Hayakawa, S., Handbook of Sputter Deposition Technology,
Principles Technology and Applications, Noyes, Park Ridge, NJ, 1992.
[62] Mahan, J. E., Physical Vapor Deposition of Thin Films, John Wiley & Sons, New
York, 2000.
[63] Pellicori, S., http://www.cerac.com/pubs/cmn/cmn6_3.htm, "Topics in Sputter
Deposition", Coating Materials News, Milwaukee, WI, 2003.
[64] Wolf, S. and Tauber, R. N., Silicon Processing for the VLSI Era, Vol.1, Lattice
Press, Sunset Beach, CA, 1986.
[65] Li, J., Nguyen, Q. T., Zhou, L. Z., Wang, T., Long, Y. C. and Ping, Z. H.,
Desalination 147 (2002) 321–326.
[66] Cote, D. R., Nguyen, S. V., Cote, W. J., Pennington, S. L., Stamper, A. K. and
Podlesnik, D. V., IBM J. RES. DEVELOP. 39 (1995) 437-464.
[67] Christensen, T. M., http://www.uccs.edu/~tchriste/courses/PHYS549/
549lectures/index.html, "Physcis of Thin Films", University of Colorado at
Colorado Springs, Colorado Springs, CO, 2003.
35
CHAPTER 2
HYDROXYLATED AMORPHOUS POLYMERIZED CARBON:SILICON FILMS - ADHERENCE
/DIFFUSION BARRIERS FOR COPPER *
2.1 Motivation for Si:C:H Based Barrier Layer
The integration of Cu with SiO2 or lower dielectric constant materials in device
interconnect structures requires a barrier to inhibit Cu diffusion under bias/thermal
stress conditions due to copper’s relatively facile diffusion through SiO2 [1]. To date,
refractory metals such as Ta or Ta/TaN have been satisfactory in this regard, but the
move to porous low-k materials with relatively large coefficients of thermal expansion
has motivated a search for alternative barrier materials. Since many low-k materials
contain carbon, the formation of interfacial metal carbide layers, which may inhibit Cu
wetting [2], is an additional concern.
Recent work [3, 4] has shown that thin amorphous Si:C:H films (a-Si:C:H) exhibit
resistance to thermally stimulated Cu diffusion at temperatures below 800 K. More
recently, studies have shown that such films exhibit potentially suitable resistance to
combined bias/thermal stress (BTS) diffusion of copper [5], especially when partially
nitrided [6]. The use of thin a-Si:C:H films as surface modifications to intralayer low-k
dielectrics to act as Cu diffusion barriers, is therefore of significant interest [7].
*Entire chapter is reprinted from THIN SOLID FILMS, Vol 440, No 1-2, M. Pritchett et al., “Copper metallization of hydroxyl-modified amorphous Si:C:H films,” pp. 100-108, 2003 Elsevier Science B.V., with permission from Elsevier.
36
A major difficulty with the use of a-Si:C:H films in intralayer dielectric applications
is the lack of Cu/film adherence . When a-Si:C:H films are formed on various metal
substrates in the presence of electron or plasma bombardment, excellent adherence of
the film to the metal substrate is observed [3, 4, 8]. Subsequent deposition of Cu
adatoms onto the a-Si:C:H film, however, results in weak chemical interaction, and the
growth of three-dimensional islands (Volmer-Weber growth) [4, 9, 10], during the initial
nucleation process. Such weak Cu/film interaction results in facile thermal
agglomeration, and can result in void formation during via fill or device operation.
Wetting is the result of the attraction between the film and the substrate surface,
and is described thermodynamically as the free energy change of the system (∆γ) due
to the removal of a metal adatom from a metal cluster to the substrate surface:
∆γ = γF + γFS - γs (1)
Where γF, γS and γFS are the surface free energies of the metal film, substrate, and film-
substrate interface, respectively. When the overall change in free energy is negative,
the attraction of the film for the interface is greater than the attraction of the film particles
to each other, and wetting will occur. Since γF >>γs for many metal films on many
dielectric substrates, γFS must be large and negative for wetting to occur, i.e., there must
be a strong attractive chemical interaction between film and substrate.
If there is relatively little stress in the deposited film (e.g., due to lattice mismatch
with the substrate), layer-by-layer, or Frank-van de Merwe (FM) growth will result. More
commonly, lattice mismatch induces significant internal stress, and the increase in
stress with film thickness results in a transition from initial layer-by-layer growth to island
formation (Stranski-Krastanov, or SK growth) [11, 12]. Therefore, considerable insight
37
concerning the effects of substrate chemical structure on film nucleation and growth can
be gained by examining the interactions of the substrate with the first few monolayers of
deposited film [13].
Amorphous-Si:C:H films of controlled thickness and composition can be prepared
by electron bombardment of vinyltrimethylsilane (VTMS) multilayers under UHV
conditions [4]. Such films exhibit a chemical stoichiometry close to that of the VTMS
precursor [4] and display excellent thermal stability on a variety of substrates [3]. In
addition, VTMS films, polymerized by gamma radiation, have shown a decrease in their
gas permeability [14]. Other studies have demonstrated that additional properties of a-
Si:C:H films, such as leakage current and dielectric constant, can be adjusted by
modification of the films with ammonia [6], fluorine [15] and nitrogen [7]. The growth of
a-Si:C:H films can be adjusted by modification with oxygen [16]. Surface science
studies of the wetting of Cu on hydroxylated and dehydroxylated ordered alumina [17,
18] and hydroxylated Ta silicate [19] surfaces have found that hydroxylation promotes
the wetting of Cu. Therefore, modifying VTMS with H2O as a source of hydroxyl groups
might promote copper wetting on the modified a-Si:C:H film.
This work reports the ability of copper to wet, or grow conformally on, the surface
of a hydroxide modified cross-linked polymeric a-Si:C:H film (a-Si:C:H./OH) derived
from VTMS. At 90K physisorption of the VTMS precursor was immediately followed by
H2O physisorption. The addition of OH groups onto the Si:C:H film was achieved by
bombarding the physisorbed VTMS/H2O multilayers with an electron beam at 90K,
followed by annealing at 300K in UHV. XPS measurements indicate that Cu deposited
at 300 K by sputter deposition onto a-Si:C:H/OH grows conformally, with the initial Cu
38
ad-layer forming both Cu(I), from charge transfer between Cu and the Si:C:H film, and
Cu(0) species at the interface. In contrast, wetting and charge transfer were not
observed for Cu deposited onto a-Si:C:H, in agreement with previous results [3].
2.2 Experiment
The experiments were performed in an ultra-high vacuum (UHV) chamber that
has been described previously in detail [19]. Briefly, the system consists of a main
chamber with a base pressure of 6 x 10-10 Torr, and a dual-magnetron sputter
deposition chamber, with a base pressure of 5 x 10-9 Torr. Transferring the sample
between the two chambers was accomplished without atmospheric exposure. The
temperature of the sample was maintained between 90 K and 1200 K by a combination
of liquid nitrogen cooling and resistive heating.
X-ray photoelectron spectra were acquired in the UHV main chamber, using a
non-monochromated MgKα X-ray source at 15 kV and 300 W, with a commercial
hemispherical analyzer operated at 50 eV constant pass energy. Methods of data
acquisition and analysis have been described previously [19]. All spectra reported here
were acquired with the analyzer aligned along the surface normal (normal emission),
unless otherwise stated.
The Ti foil substrate had a purity of 99.97%, and was 0.0076 cm thick with an
area of 1 cm2. The foil was spot welded to Ti leads and a chromel-alumel thermocouple
was spot welded to the back of the sample. Sample cleaning was accomplished
through cycles of argon ion (Ar+) sputtering followed by annealing. The Ti substrate was
heated resistively through the thermocouple, and cooled indirectly by liquid nitrogen.
39
VTMS was procured from a commercial source, purified by freeze-thaw pumping, and
delivered through a stainless steel doser tube regulated by a manual leak valve, ~ 2 cm
from the face of the Ti sample. Gas exposures are reported here in Langmuir, L, (1L =
1 x 10-6 Torr-s), and have not been corrected for ion gauge sensitivity and flux to the
surface. To produce the unmodified film, 30L of VTMS was condensed onto a clean Ti
substrate at 90K, immediately followed by electron beam bombardment to induce
crosslinking [4, 20]. To produce the hydroxide-modified film, 30L of VTMS was
condensed onto a clean Ti substrate at 90K, immediately followed by 30L of H2O, and
electron beam bombardment.
A Kimball Electronics electron gun was used to produce an electron beam of 500
eV energy, with a 3mA emission current, producing an average flux at the sample
surface of 1.5 x 1013 electrons cm-2 s-1 [4]. After electron bombardment, the films were
annealed to 300K in UHV in order to desorb any remaining physisorbed species.
Copper was sequentially deposited by physical vapor deposition (PVD) onto the film’s
surface using an Advanced Energy MDX 500 magnetron source (25 watts), and an
MKS baratron gauge to monitor the Ar+ sputter pressure (2 x 10-5 atm), achieving a rate
of about 0.5 Å/minute, and sputtering Cu atoms with energies of ~ 0.2 eV to ~ 10 eV
[21, 22]. After copper deposition was completed, the sample was annealed in intervals
from 300K to 1000K by resistive heating. Copper and film interactions were
characterized by XPS after each successive deposition and temperature increment.
For both the unmodified and modified films, film compositions were calculated
using XPS core level data, with corresponding intensities, and appropriate elemental
40
sensitivity factors for this analyzer [23]. Relative film composition was determined from
XPS core level intensity data with the general equation:
NA/NB = [IAAB]/ [IBAA] (2)
NA and NB are the atomic concentrations of element A and B respectively, IA and IB are
the respective XPS intensities (proportional to peak areas), and AA and AB are the
appropriate atomic sensitivity factors [24].
The a-Si:C:H films produced were very thin, as demonstrated by the ability to
observe the titanium XPS peak after the 300K anneal, enabling the average thickness,
(d), of the a-Si:C:H films to be calculated as a function of the intensity of the Ti(2p3/2)
core level peak (ITi), and the inelastic mean free path (λ) of the Ti(2p3/2) photoelectron,
21.6 Å , using the following equation:
ITi = I Ti (0)exp(-d/λ) (3)
The thickness of copper deposited on the hydroxide-modified film was estimated from
the Si(2p) intensity using the intensity ratio of copper to silicon. The copper to silicon
intensity ratio, ICu/ISi, intensity ratio of reference element atomic sensitivity factors,
I∞Cu/I∞
Si, and the inelastic mean free paths for copper and silicon, (λCu) and (λSi), are
known, therefore the thickness, (dCu) is derived from the equation [24]:
ICu/ISi = (I∞Cu/I∞
Si)[(1-exp-dCu/λCu)/exp-dCu/λSi] (4)
Absolute copper and Ti intensities from clean samples in the system used for these
experiments have an estimated error in measurement of no more than 10% [4]. The
Scotch Tape Test was performed on the PVD copper film to assess adhesion [25, 26].
41
2.3 Results
2.3.1 Amorphous-Si:C:H Film Characterization
The average thickness of the unmodified a-Si:C:H cross-linked film produced
from the VTMS precursor, by the process described above, was determined (Eq. 3) to
be 15 Å. As shown in Figure 2.1, the Si(2p) XPS spectrum exhibits a peak at 100.8 eV,
-106 -104 -102 -100 -98 -960
1000
-290 -288 -286 -284 -282 -2800
4000 C(1s)
Si(2p)
XPS
Inte
nsity
(cps
)
Binding Energy (eV)
C - C, C - H
Si - C
Fig. 1 Fig. 1 Fig. 2.1 C(1s) and Si(2p) XPS Spectra for Unmodified a - Si:C:H film .
42
attributed to Si-C:H from VTMS [27], and a C(1s) spectra and a peak at 284.5 eV,
attributed to C-C and C-H bond types [27]. The unmodified a-Si:C:H film had negligible
oxygen content. No spectral changes were observed at glancing emission (not shown).
Although the use of extremely thin films (~ 30 Å) presents the risk of pinholes affecting
the results, Cu wetting results for unmodified a-Si:C:H (vide infra) were in agreement
with previously published results for thicker films [4].
Normal emission spectra for the hydroxide modified film, (a-Si:C:H/OH), are
shown in Figure 2.2. This film was determined to have an average thickness of 25 Å
(Eq. 3). The Si(2p) spectrum for the modified film displays a high binding energy
shoulder, while the C(1s) spectrum does not, indicating that the oxygen-containing
-544 -540-536-532-528-5240
1500
-290 -288 -286 -284 -282-2800
4500
-111-108-105-102-99 -96 -930
1000
C(1s)
O(1s)
Si(2p)
XPS
Inte
nsity
(cps
)
C -C, C -H
Si-O x
Si -OH
Si:C:H
Si-O xSi-OH
Fig. 2 Fig. 2
Binding Energy (eV)
Fig. 2.2 C(1s), Si(2p), O(1s) XPS spectra for the a-Si:C:H/OH film.
43
species are bonded primarily to silicon and not to carbon atoms. The O(1s) peak (Fig.
2.2) was fit with two spectral components with FWHM of 2.9 eV and binding energies of
532.3 eV for Si-Ox (x<2) and 533.7 eV for Si-OH [28]. The Si(2p) spectrum (Fig. 2.2),
exhibits a dominant feature at 100.8 eV binding energy for Si-C:H. The other features
are more difficult to assign due to the variety of assignments found in the literature for
Si-OH. The shoulder at higher binding energies is well fit by two peaks, with a
component at 102.3 eV peak assigned to Si-Ox (x<2) as indicated by Hollinger [29] and
Himpsel [30], and a component at 103.8 eV attributed to a hydroxylated silicon oxide
species [31]. A binding energy of 102.3 eV, however is also consistent with a
hydroxylated Si+1 species, Si-OH [32, 33]. Evidence for a hydroxylated Si species is
more clearly indicated by the 533.7 eV peak in the O(1s) spectrum, which is in
agreement with the assignment of 533.5 eV for the O(1s) Si-OH peak [28]. The
absence of significant numbers of C-OH groups is indicated by the absence of a
characteristic peak near 286.0 eV in the C(1s) spectrum [27].
Calculation of the C(1s)/Si(2p)total atomic ratio, using appropriate sensitivity
factors (A), and inelastic mean free path (λ) values, atomic concentrations(N), and peak
intensities(I) [23]: yielded a C(1s)/Si(2p)total ratio of 3.5:1, for a-Si:C:H/OH, which is in
agreement with previous research that calculated the C(1s)/Si(2p)total ratio for a-Si:C:H
as 3.8:1 [4]. This prior work found that chemical and matrix effects, due to differences
in kinetic energies, could be corrected for by dividing atomic concentrations by their
appropriate mean free paths as reported by Ke, et al. [24, 34]:
NC / NSi = [(IC · ASi) / λC ] / [(ISi · AC) / λSi] (5)
44
This treatment was applied to these results using inelastic mean free paths calculated
from the modified form of the Bethe equation [35]: C(1s) = 22.5 Å and Si(2p) = 25.7 Å
and O(1s) = 15.0 Å. The corrected C(1s)/Si(2p)total ratio for both the unmodified a-
Si:C:H film and the modified a-Si:C:H/OH film increased to ~ 4.5:1. Because of the
uncertainty of the actual additional chemical and matrix effects contributed by oxygen in
a-Si:C:H/OH, a comparison calculation was done using only the a-Si:C:H Si(2p)(101)
peak intensity to calculate the C(1s)/Si(2p)total ratio, which resulted in a C(1s)/Si(2p)total
ratio of 4.7:1. Both methods of calculating the C(1s)/Si(2p)total ratio for a-Si:C:H/OH,
correlate well with the five to one C(1s)/Si(2p)total stoichiometry of VTMS. The
O(1s)total/Si(2p)total atomic ratio for the same a-Si:C:H/OH film was calculated to be
0.44:1, and 0.073:1 for the unmodified a-Si:C:H film.
2.3.2 Copper Deposition by Physical Vapor Deposition (PVD)
XPS was used to characterize the interactions of sputter-deposited (often
referred to as physical vapor deposited--PVD) Cu with clean and modified film
substrates. Cu coverage was ascertained from Cu(2p3/2)core level spectra, with Cu(I)
and Cu(0) oxidation states distinguished by the examination of x-ray excited Cu(L3VV)
Auger spectra [17]. Cu(2p3/2) spectra at no time revealed observable satellite peaks,
indicating the absence of significant concentrations of Cu(II) [36, 37]. Copper was
deposited in the sputter deposition chamber at 300K at intervals of several minutes
sputter deposition time, with estimated kinetic energies between ~ 0.2 eV and ~ 10 eV
[21, 22]. XPS data were acquired after each deposition, followed by additional copper
deposition. Energetic particles, such as those that can be deposited during PVD, are
45
known to deform and modify surfaces [38]. In this case, however particle energies do
not seem to be the cause of the chemical modification of the modified a-Si:C:H film that
enabled wetting, since copper PVD on unmodified a-Si:C:H films using the same
parameters, did not produce similar XPS spectra or copper wetting.
The evolution of the Si(2p) spectrum during Cu deposition is shown in Figure 2.3
for the unmodified a-Si:C:H film, and in Figure 2.4 for a-Si:C:H/OH. No changes were
observed for C(1s) spectra (not shown) after deposition on either modified or
in binding energy or intensity with Cu deposition (Fig. 2.3). In contrast, Cu deposition
Fig. 2.3 Fig. 2.4
-105 -103 -101 -99 -97
-105 -103 -101 -99 -97
-105 -103 -101 -99 -97
-105 -103 -101 -99 -97 Binding Energy (eV)
XPS
Inte
nsity
( cp
s)
Cop
per D
epos
ition
tim
e (m
in)
Si(2p)
Fig. 3Fig. 3
- 0.0
- 1.0
- 3.0
- 6.0
Fig. 2.3 Si(2p) XPS spectra during deposition of a-Si:C:H film.
-111-108-105-102-99 -96 -93
-111-108-105-102-99 -96 -93
-111-108-105-102-99 -96 -93
-111-108-105-102-99 -96 -93
- 6.5
- 2.5
- 8.5
- 0.0
Binding Energy (eV)
XPS
Inte
nsity
( cp
s)
Si(2p)
Fig. 4Fig. 4
Cop
per D
epos
ition
tim
e (m
in)
Si:C:H
Si:C:H
Si:C:H
Si:C:H
Fig. 2.4 Si(2p) XPS spectra during
Si:C:H
Si:C:H
Si:C:H
Si:C:H
Si-OH
Si-OH
Si-OH
Si-OH
deposition of a-Si:C:H/OH film.
46
on the hydroxylated substrate results in relative attenuation of the Si(2p) intensity
attributed to Si-C:H sites (Fig. 2.4). Changes in relative Cu(2p3/2)/Si(2p) intensity with
Cu deposition time are plotted in Figure 2.5. The plot of Cu(2p3/2)/Si(2p) intensity vs. Cu
deposition for unmodified VTMS exhibits linear behavior, which is consistent with 3-
dimensional islanding (Volmer-Weber growth) [11]. This contrasts with the behavior
observed on OH-modified substrates-- a linear increase followed by a distinct change in
slope, which is more similar to Stranski-Krastanov growth [11].
The evolution of the x-ray excited Cu(L3VV) spectrum during deposition on
unmodified and on hydroxylated films is shown in Figure 2.6. For the first six minutes of
copper deposition on the hydroxylated film at 300K (Fig. 2.6a), the Cu(L3VV) Auger
0 2 4 6 8 10 12 0 2 4 6 8
10 12 14 16 18 20
(a)
(b)
Cu(0)
Cu(I)
Break Point
Deposition Time (min)
XPS
Inte
nsity
Rat
io C
u(2p
3/
2 )/ (S
i 2p)
Fig. 2.5 Copper uptake curve for copper deposition on a) unmodified a-Si:C:H film b) OH-modified a-SI:C:H film.
47
spectrum remains broad, with features at 914.9 eV and 918.0 eV, consistent with the
formation of both Cu(I) and Cu(0) [39, 40]. (Cu(II) was eliminated as a possible source
of the 918.0 eV peak because there were no characteristic satellite peaks in the
Cu(2p3/2) spectra--not shown). At longer deposition times, the Cu(L3VV) peak
noticeably narrows and shifts to 918.8 eV, indicating the predominance of copper metal
[39, 40].
Fig. 2.6 Cu Auger spectra for copper deposition on a) OH-modified a-Si:C:H film b) unmodified a-Si:C:H film.
Deposition on the unmodified a-Si:C:H film (Fig. 2.6b) results in the exclusive
formation of Cu(0). The feature in Figure 6b, assigned to Cu(0), is observed near 916
eV for short Cu deposition times, and is shifted to higher binding energies upon
additional deposition. Such a shift is characteristic of the growth of Cu(0) clusters [41]
and is due to the enhanced screening of Auger final state holes with increasing cluster
890 900 910 920 930
a-Si:C:H Film
Aug
er In
t ens
ityDeposition Time (min)
Kinetic Energy (eV)900 910 920 930
0
500
1000
1500
2000
2500
3000
3500
a-Si:C:H/OH Film
Aug
er In
tens
ity
Deposition Time (min)
Kinetic Energy (eV)
0.0
7.5
3.5
2.0
1.0
5.5
8.5
Cu(0)Cu(I) Cu(0)
20.5
14.5
8.5
6.53.51.0
890 900 910 920 930
a-Si:C:H Film
Aug
er In
t ens
ityDeposition Time (min)
Kinetic Energy (eV)900 910 920 930
0
500
1000
1500
2000
2500
3000
3500
a-Si:C:H/OH Film
Aug
er In
tens
ity
Deposition Time (min)
Kinetic Energy (eV)
0.0
7.5
3.5
2.0
1.0
5.5
8.5
Cu(0)Cu(I) Cu(0)
20.5
14.5
8.5
6.53.51.0
48
size [42]. There is also a shift to lower binding energy in the Cu(2p3/2) spectrum (not
shown) that has been reported for the formation of Cu clusters [43].
For Cu deposition on the modified film, the narrowing of the Cu(L3VV) Auger
peak shape in increasing Cu coverage, accompanied by the shift to higher kinetic
energy (Fig. 2.6a) marks the evolution of a dominant Cu(0) feature that correlates with
the break point in the Cu(2p3/2)/Si(2p) uptake curve (Fig. 2.5b), and is indicative of
copper wetting, or SK growth, on the modified a-Si:C:H/OH film. This behavior is not
observed for Cu deposition on the unmodified a-Si:C:H film, which produces no change
in the uptake slope, consistent with Volmer-Weber growth (3-D islanding) [9], and Cu(0)
cluster formation [39, 40]. An examination of Figure 6a, however, reveals that Cu
deposited at coverages before the break point, (Fig. 2.5b), includes both Cu(I) and
Cu(0). At higher coverages, only Cu(0) is formed. After the Scotch Tape Test, XPS
characterization reveals the presence of Cu.
2.3.3 Annealing Effects
In order to assess the stability of the Cu/substrate interface for the hydroxylated
film, Cu was deposited on a hydroxylated substrate at 300 K, followed by 15 minute
anneals at elevated temperatures in UHV, with XPS acquisition during cool down.
Results are shown in Figure 2.7 for a copper film with an initial average thickness of 5 Å
prior to annealing. The Cu(2p3/2)/Si(2p) ratio decreases slightly due to island formation
between 500K and 800K, after which there is a much more significant decrease in
relative Cu intensity. Such a decrease can be caused either by island formation
(dewetting) or by diffusion of Cu from the surface into the bulk. A measurement of the
49
relative intensities of the Cu(2p3/2)and Cu(3p) intensities can distinguish between these
two possibilities. Since the inelastic electron mean free paths of the two copper species
in the film are very different, 13.9 Å for Cu(2p3/2) and 38.2 Å for Cu(3p) , the Cu(2p3/2)
peak intensity will be preferentially attenuated relative to the Cu(3p) intensity if copper
diffuses into the bulk. Due to the limitations of XPS sensitivity, such a method is
obviously less sensitive than electrical measurements for the detection of Cu diffusion,
but does serve to delineate the reason for changes in relative Cu XPS intensity. A plot
of the Cu(3p)total/Cu(2p3/2)total intensity ratio against temperature can therefore
distinguish between copper island formation and copper diffusion, because copper
200 400 600 800 1000
0
1
2
3
4
5
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Temperature (K)
(b) X
PS In
tens
ity C
u32p
)/Cu(
2p),
cps
(a)
XPS
Inte
nsity
Cu(
2p)/S
i(2p)
, cps
Fig. 7 Fig. 7
(a) (b)
diffusion
Fig. 2.7 Dewetting is indicated by the decrease in curve (a) after 500 K.
Diffusion is indicated by the increase in curve (b) after 800 K.
Diffusion indicated
50
diffusion will produce a significant decrease in the Cu(2p3/2) peak intensity and a
subsequent increase in the Cu(3p)/Cu(2p3/2) intensity ratio [44]. The data in Figure 2.7
indicate no significant change in the Cu(3p)/Cu(2p3/2) intensity ratio below 800 K.
Above 800 K the ratio increases sharply, indicating that annealing to temperatures of
800 K or higher induces considerable Cu diffusion into the substrate. Similar results
were observed for Cu deposited on unmodified a-Si:C:H [3]. Annealing above 800 K
also produced significant changes in the silicon XPS spectrum of the hydroxide modified
a-Si:C:H film (not shown). The dominant Si(2p) 100.8 eV peak (Si-C:H) at 300K, began
to significantly decrease in intensity after 800K, until only 10% of its intensity remained
at 1000K. Annealing above 800K also produced two new Si(2p) peaks. The 98.1 eV
peak is attributable to Ti silicide formation, and the 99.6 eV peak indicates Si-Si
bonding.
2.4 Discussion
The data presented here demonstrate that hydroxylation of the surface of an a-
Si:C:H film strengthens the Cu/film chemical interactions and results in the deposition of
a conformal first layer of PVD Cu at 300 K. The evolution of a dominant Cu(0) feature
for deposition on the hydroxylated film (Fig. 2.6a) correlates with the break point in the
Cu(2p3/2)/Si(2p) uptake curve, depicted in Figure 2.5b. Such a break point is indicative
of initial conformal growth on the modified a-Si:C:H/OH film [11]. Unlike many cases of
S-K growth [17], however, the first layer of deposited Cu consists of both Cu(I) and
Cu(0) species.
51
Different behavior is observed for Cu deposition on an unmodified a-Si:C:H film,
where no change in the uptake slope is observed (Fig. 2.5), consistent with Volmer-
Weber growth (3-D islanding) [9, 11], and for which only Cu(0) formation is observed
(Fig. 2.6b). The shift of the Cu(L3VV) peak to higher Auger kinetic energy with
increasing Cu deposition time (Fig. 2.6b) is consistent with the growth in size of Cu(0)
clusters, since increasing cluster size results in more effective screening of Auger final
state holes [39, 40]. The conclusion of conformal Cu growth on the OH-modified film
surface is further supported by the fact that the substrate Si(2p) intensity is significantly
attenuated during copper deposition on the OH-modified film, which is expected for
conformal growth (Fig. 4) due to chemical interactions of the Cu adatoms with the
modified substrate. Three dimensional island growth of Cu, on the other hand, will not
significantly attenuate the substrate Si(2p) signal as shown in Fig. 3.
Additional information about the copper-film interface can be obtained from the
Auger parameter, α, defined by the equation:
α = EKCu(L3VV) + EBCu(2p3/2) (7)
since the value of α is independent of the uniform sample charging commonly
encountered for XPS on insulating samples [43]. At very low Cu coverage (< 1 minute
deposition time, ~ 0.25 ML) on a-Si:C:H/OH, a broad x-ray excited Cu(L3VV) feature is
observed (Fig. 2.6a). The two partially resolved peaks have α values of 1848.2 eV and
1850.6 eV (Fig. 2.8). The former value is consistent with Cu(I) formation [39, 45],
although the latter is low for the formation of Cu(0) species, which may be due to cluster
formation [40, 45, 46]. At two minutes deposition time (a Cu coverage of ~ 0.5 ML) the
52
first Auger parameter (corresponding to Cu(I)) does not change significantly, but the
second Auger parameter shifts to 1851.5 eV, which is now in the range generally
reported for Cu(0) [45, 47].
Fig. 2.8 Cu Auger parameter indicates both Cu(I) and Cu(0) species are deposited.
In contrast, Cu deposition on a-Si:C:H results in a similar limiting value for α only
after much longer deposition times (> 8.5 minutes). Comparisons with core-hole
spectra (not shown) indicate that changes in α are due mainly to changes in the Cu(L3
VV) spectra (Fig. 2.6a, 6b), and therefore to changes in final state screening [39, 40,
43]. The fact that an α value characteristic of bulk Cu is observed after only 2.5 minutes
Cu deposition on the hydroxylated surface, but not until after 8.5 minutes on the
unmodified surfaces, indicates a more rapid coalescence of isolated Cu clusters on the
1847.1848.1848.1849.1849.1850.1850.1851.1851.1852.1852.1853.
0 1 2 3 4 5 6 7 8 9
Cu Deposition time (min)
Aug
er P
aram
eter
(α
')
Cu +1
Cu 0
Cu Deposition Time (min)
Aug
er P
aram
eter
(α)
53
hydroxylated surface. This is also consistent with the more rapid attenuation of the
substrate Si(2p) intensity on the hydroxylated surface, and confirms the tendency
toward conformal Cu growth on a-Si:C:H/OH.
The thickness of copper deposited on a-Si:C:H/OH was estimated from the
Si(2p) intensity using the intensity ratio of copper to silicon film substrate (Eq. 4). The
average thickness at the break point in the Cu/Si uptake curve (Fig. 2.5) is 2 Å, which is
very close to the Cu(I) ionic diameter of 1.92 Å [40]. Since the initial deposition of Cu is
marked by the formation of both Cu(I) and Cu(0), it is difficult to give a precise
determination of surface coverage in the first layer, the more so as the film surface is
undoubtedly far from smooth on an atomic scale, but these data indicate that the
coverage of the initial Cu adlayer is slightly less than 1 monolayer. The data in Figures
2.4 - 2.6 therefore indicate that Cu deposition at 300 K on the OH-modified a-Si:C:H film
results in initial conformal growth, with the first layer including both Cu(I) and Cu(0)
species. Upon completion of the first layer, a second, metallic Cu layer begins to form.
In contrast, Cu deposition at 300 K on the unmodified a-Si:C:H film results exclusively in
Cu(0) formation and VW growth (non-wetting) as shown in Figures 5a and 6b. Similar
hydroxyl-induced transitions from non-wetting to wetting behavior for deposited Cu have
been reported for sapphire(0001) surfaces [17], and for Rh deposited on ordered,
ultrathin alumina films [48]. Cu wetting has been reported for deposition at 300 K on
hydroxylated ultrathin Ta silicate films on Si(100) and for air-exposed TaSiN substrates
[49]. Theoretical studies also indicate that Pt nucleation occurs preferentially at surface
OH sites on MgO single crystal surfaces [50]. These results strongly suggest, therefore,
54
that surface hydroxylation may be an effective method for control of the nucleation of
semi-noble metal adatoms on a broad variety of oxide or other dielectric substrates.
Thermally-induced copper diffusion was observed by XPS on OH-modified a-
Si:C:H films only at temperature above 800K in UHV, as shown by the lack of change in
the Cu(3p)/Cu(2p3/2) versus temperature curve (Fig. 2.7). The thermal stability of the
Cu/substrate interface is similar for OH-modified and unmodified a-Si:C:H films [4].
Annealing above 800K caused copper to diffuse into the film and resulted in the
development of larger clusters of Cu(0) [43], at the expense of Cu(I), as indicated by the
peak shift from 918.0 eV to higher kinetic energy, 918.6 eV, corresponding to an Auger
parameter shift in the range for Cu(0), from 1851.3 eV to 1851.9 eV.
In addition, annealing from 500K to 800K resulted in a decrease in Si(100.8 eV)
peak intensity, and increases in Si(102.3 eV) and Si(103.8 eV) peak intensities, which
may be due to uncharacterized reactions within the a-Si:C:H/OH film. Annealing to
1000K resulted in a large decrease in intensity of the Si 100.8 eV, Si 102.3 eV and Si
103.8 eV peaks, and the production of a large peak at 98.1 eV, likely titanium silicide,
and a small 99.6 eV peak, attributed to Si-Si, indicating decomposition of the film.
2.5 Conclusions
This study demonstrates that hydroxyl modification of vinyltrimethylsilane (VTMS)
derived amorphous silicon:carbon:hydrogen (a-Si:C:H) film facilitates the wetting of
sputter deposited, physical vapor deposited (PVD), copper to the hydroxyl modified a-
Si:C:H/OH film’s surface at 300K. This study also shows that modification of a-Si:C:H
55
with hydroxide, does not adversely affect the film’s ability to inhibit copper diffusion to
800K, as characterized by x-ray photoelectron spectroscopy (XPS).
These findings suggest the use of a-Si:C:H/OH films as a combined adhesion
promoter and diffusion barrier for the deposition of the initial copper seed layer in
interconnect circuit development, and indicate the possibility that additional
modifications of these films, to lower their dielectric constant and enhance other
properties, may be possible without altering the adhesion/barrier properties, enabling
their use as intermetal dielectrics that do not require an additional diffusion barrier layer
or adhesion promoter.
56
2.6 Chapter References
[1] McBrayer, J. D., Swanson, R. M. and Sigmon, Y. W., J. Electrochem. Soc. 133
(1986) 1242-6.
[2] Chen, L., Magtoto, N., Ekstrom, B. and Kelber, J., Thin Solid Films 376 (2000)
115-123.
[3] Chen, L. and Kelber, J. A., J. Vac. Sci. Technol. A 17 (1999) 1968.
[4] Tong, J., Martini, D., Magtoto, N., Pritchett, M. and Kelber, J., Appl. Surf. Sci. 187
(2002) 256-264.
[5] Lanckmans, F., Gray, W. D., Brijs, B. and Maex, K., Microelectron. Eng. 55
(2001) 329-335.
[6] Ishimaru, T., Shioya, Y., Ikakura, H., Nozawa, M., Nishimoto, Y., Ohgawara, S.
and Maeda, K., Proceedings of the International Interconnect Technology
Conference, San Francisco, CA, 2001, Institute of Electrical and Electronics
Engineers, Inc., (2001) 36-38.
[7] Lee, S. G., Kin, Y. J., Lee, S. P., Oh, H.-S., Lee, S. J., Kim, M., Kim, I.-G., Kim,
J.-H., Shin, H.-J., Hong, J.-G., Lee, H.-D. and Kang, H.-K., Jpn. J. Appl. Phys. 40
(2001) 2663-2668.
[8] Deplancke, M. P., Powers, J. M., Vandentop, G. J., Salmeron, M. and Somorjai,
G. A., J. Vac. Sci. Technol. A 9 (1991) 450.
[9] Aubin, E. and Lewis, L. J., Phys. Rev. B 47 (1993) 6780-6783.
[10] Miller, T., Rosenwinkel, E. and Chiang, T.-C., Phys. Rev. B 30 (1984) 570-577.
[11] Rhead, G. E., Barthès, M.-G. and Argile, C., Thin Solid Films 82 (1981) 201.
57
[12] Komori, F., Kushida, K., Hattori, K., Arai, S. and Iimori, T., Surf. Sci. 438 (1999)
123.
[13] Varma, S. and Chottiner, G. S., J. Vac. Sci. Technol. 10 (1992) 2857-2862.
[14] Yampol'skii, Y. P. and Volkov, V. V., J. Membr. Sci. 64 (1991) 191-228.
[15] Yoshimaru, M., Koizumi, S. and Shimokawa, K., J. Vac. Sci. Technol. A 15
(1997) 2915-2922.
[16] Lee, M.-S. and Bent, S. F., J. Vac. Sci. Technol. A 16 (1998) 1658-1663.
[17] Niu, C., Shepherd, K., Martini, D., Tong, J., Kelber, J. A., Jennison, D. R. and
Bogicevic, A., Surf. Sci. 465 (2000) 163-176.
[18] Heemeier, M., Frank, M., Libuda, J., Wolter, K., Kuhlenbeck, H., Baumer, M. and
Freund, H. J., Catalysis Letters 68 (2000) 19-24.
[19] Shepherd, K. and Kelber, J., Appl. Surf. Sci. 151 (1999) 287-298.
[20] Toth, A., Bertoti, I., Marletta, G., Ferenczy, G. and Mohai, M., Nucl. Instrum.
Methods Phys. Res., Sect. B 116 (1996) 299-304.
[21] Wasa, K. and Hayakawa, S., in R. F. Bunshah and G. E. McGuire (Eds.):
Handbook of Sputter Deposition Technology, Noyes Publications, Park Ridge,
New Jersey 1992, p. 32.
[22] Zou, W., Ph. D. Dissertation, University of Virginia, Charlottesville, Virginia 2001,
p. 303.
[23] "Vacuum Generators", Hastings, UK,
[24] Seah, M. P., in D. Briggs and M. P. Seah (Eds.): Practical Surface Analysis, Vol.
1, New York 1990, p. 201-255.
58
[25] Chang, C.-A., Baglin, J. E. E., Schrott, A. G. and Lin, K. C., Appl. Phys. Lett. 51
(1987) 103.
[26] Chan, C. M., Cao, G. Z., Fong, H., Sarikaya, M., Robinson, T. and Nelson, L., J.
Mater. Res. 15 (2000) 148-154.
[27] Wagner, C. D., Naumkin, A. V., Kraut-Vass, A., Allison, J. W., Powell, C. J. and
Jr., J. R. R., www.nist.gov, "NIST X-ray Photoelectron Spectroscopy Database",
NIST, 2002.
[28] Miller, M. L. and Linton, R. W., Anal. Chem. 57 (1985) 2314-19.
[29] Hollinger, G., Appl. Surf. Sci. 8 (1981) 318-36.
[30] Himpsel, F. J., McFeely, F. R., Taleb-Ibrahimi, A. and Yarmoff, J. A., Phys. Rev.
B 38 (1988) 6084-6096.
[31] Barr, T. L., J. Phys. Chem. 82 (1978) 1801-1810.
[32] Fujita, K. and Oya, A., J. Mater. Sci. (UK) 31 (1996) 4609-4615.
[33] Iiojima, Y. and Sonoda, N., Clean Technol. 12 (1992) 39.
[34] Ke, R., Haasch, R. T., Finnegan, N., Dottl, L. E., Alkire, R. C. and Farrell, H. H.,
J. Vac. Sci. Technol. A 14 (1995) 80-88.
[35] Powell, C. J., Jablonski, A., Tilinin, I. S., Tanuma, S. and Penn, D. R., J. Electron
Spectrosc. Relat. Phenom. 98-99 (1999) 1-15.
[36] Chusuei, C. C., Brookshier, M. A. and Goodman, D. W., Langmuir 15 (1999)
2806-2808.
[37] Paszti, Z., Peto, G., Horvath, A. E. and Karacs, A., J. Phys. Chem. B 101 (1997)
2109-2115.
[38] Paterson, M. J., Diamond Relat. Mater. 5 (1996) 1407-1413.
59
[39] Klein, J. C., Proctor, A., Hercules, D. M. and Black, J. F., Anal. Chem. 55 (1983)
2055-2059.
[40] Kowalczky, S. P., Pollak, R. A., McFeely, F. R., Ley, L. and Shirley, D. A., Phys.
Rev. B 8 (1973) 2387-91.
[41] Wertheim, G. K. and DiCenzo, S. B., Phys. Rev. B 37 (1988) 844-847.
[42] Jennison, D. R. and Madden, H. H., Phys. Rev. B 21 (1980) 430-435.
[43] Wu, Y., Garfunkel, E. and Madey, T. E., J. Vac. Sci. Technol: A 14 (1996) 1662-
1667.
[44] Chen, J. G., Colaianni, M. L., Weinberg, W. H. and J.T. Yates, J., Surf. Sci. 279
(1992) 223-232.
[45] Moretti, G., J. Electron Spectrosc. Relat. Phenom. 95 (1998) 95-144.
[46] Mason, M. G., Phys. Rev. B 27 (1983) 748-762.
[47] Moulder, J. F., Stickle, W. F., Sobol, P. E. and Bomben, K. D., Handbook of X-
ray Photoelectron Spectroscopy, Physical Electronics, Eden Prairie, MN, 1995.
[48] Libuda, J., Frank, M., Sandell, A., Andersson, S., Bruhwiler, P. A., Baumer, M.,
Martenson, N. and Freund, H.-J., Surf. Sci. 384 (1997) 106.
[49] Zhao, X., Leavy, M., Magtoto, N. P. and Kelber, J. A., Appl. Phys. Lett. 79 (2001)
3479-3481.
[50] Bogicevic, A. and Jennison, D. R., Surf. Sci. Lett. 437 (1999) L741-L747.
60
CHAPTER 3
NITROGEN MODIFIED HYDROXYLATED AMORPHOUS POLYMERIZED
CARBON:SILICON FILMS - ADHERENCE /DIFFUSION BARRIERS FOR COPPER
3.1 Motivation for Nitrogen Modification of a-Si:C:H Films
More densely integrated circuits require a change from aluminum to copper
metallization and a concomitant change to lower dielectric constant (low-k) interconnect
materials. Since copper diffuses into SiO2 and low-k hydrogen silsesquioxane (HSQ)
films [1] under bias/thermal stress (BTS) conditions [2], low-k barrier layers, or
dielectrics, must be developed that will inhibit diffusion into SiO2 [3]. In addition copper
must adhere to, or wet, these barrier layers.
Results detailed in Chapter 2 show that hydroxylated amorphous silicon carbon
hydrogen films (a-Si:C:H/OH) not only resist thermally stimulated copper diffusion at
temperatures up to 800K [4, 5], but also enable copper wetting of these hydroxylated
films [6-8]. Additional modifications of these films with nitrogen may produce films that
possess low dielectric constants [9], while acting as barrier layers to copper diffusion
and promoting copper wetting, suggesting their use as all in one intermetal dielectric
layers that are potentially resistant to combined bias/thermal stress diffusion of copper
[10]. Therefore, the modification of amorphous silicon carbon hydrogen (a-Si:C:H) films
61
with nitrogen, as well as the modification of a-Si:C:H films with hydroxide, is of interest
in determining their use as intermetal low-k dielectrics [9].
The first requirement for nitrogen modified a-Si:C:H films as dielectrics, is that
copper wet, grow conformally, on the surface of these films. The second requirement is
that these nitrogen modified a-Si:C:H films inhibit copper diffusion while adhering to the
substrate surface. It has been previously determined that unmodified and hydroxide
modified a-Si:C:H films, formed on a titanium substrate in the presence of electron or
plasma bombardment, have excellent adhesion of the film to the metal substrate [4, 5,
8, 11]. Subsequent deposition of Cu adatoms onto the surface of hydroxyl modified a-
Si:C:H films results in conformal growth, wetting [8], in contrast with 3-D islanding
observed for unmodified a-Si:C:H films [5, 8, 12]. The question now is how modifying a-
Si:C:H films with nitrogen will affect the adhesion and barrier properties of these films.
Nitrogen is an electronegative element that may affect the attractive forces at the
interface of the film and the substrate, previously described thermodynamically as the
change in free energy change of the system (∆γ). Since wetting occurs when film
particles preferentially attract the copper adatoms, it may be that nitrogen will act to
enhance the ability of copper to lose electrons as compared to the unmodified film. This
may resulting in an increased attraction between the film particles and the copper
adatoms [13] such that the free energy of the film-surface interface, γFS, mitigates the
free energy of attraction between film particles, γF, and in conjunction with the free
energy of the substrate, γS, produces a negative overall change in free energy:
∆γ = (γF + γFS) - γS (1)
62
A strong attraction between the nitrogen modified film and copper adatoms, as was
found for the hydroxylated a-Si:C:H film and copper, would indicate that copper wetting
of the film’s surface is unaltered by the incorporation of nitrogen into the a-Si:C:H film.
Nitrogen modified amorphous Si:C:H films with a chemical stoichiometry close to
the C5H12Si of the precursor Vinyltrimethylsilane (VTMS), with controlled composition
and thickness, were prepared under UHV conditions [8],in anticipation of positively
altering film properties such as leakage current [14], and dielectric constant [9, 15]. This
chapter reports the ability of copper to wet, or grow conformally on, the surface of
nitrogen and nitrogen and hydroxyl modified cross-linked polymeric a-Si:C:H films
derived from VTMS, as compared to unmodified and hydroxyl modified a-Si:C:H films.
At 90K, in UHV, the VTMS precursor was physisorbed on titanium, immediately
followed by electron bombardment and annealing to 300K in UHV, resulting in an
unmodified amorphous Si:C:H film (a-Si:C:H). The nitrogen modified film was formed at
90K, in UHV, by condensing the VTMS precursor on the titanium substrate, immediately
followed by ammonia physisorption, electron bombardment, and annealing to 300K,
resulting in a nitrogen modified amorphous Si:C:H film (a-Si:C:H/N). The nitrogen
modified hydroxylated film was formed at 90K, in UHV, by condensing the VTMS
precursor on a titanium substrate, immediately followed by water physisorption,
ammonia physisorption, and electron bombardment, and then annealing to 300K,
resulting in a nitrogen modified hydroxylated amorphous Si:C:H film (a-Si:C:H/OH/N).
XPS measurements indicate Cu deposited by sputter deposition (PVD) at 300 K onto an
a-Si:C:H/N film grows conformally, as indicated by the break of the Cu/Si uptake curve,
similar to, but not as defined as the wetting seen on hydroxyl modified a-Si:C:H films [8].
63
In contrast, conformal growth of Cu deposited on the nitrogen modified hydroxylated film
(a-Si:C:H/OH/N) is indicated by both a break in the Cu/Si uptake curve, and a broad Cu
Auger spectrum from the formation of Cu(I) and Cu(0) species at the interface. These
results are similar to that seen on the hydroxyl modified a-Si:C:H film [8]. As found in
previous studies the unmodified VTMS precursor, physisorbed on titanium and
bombarded by electrons at 90K, and annealed to 300K did not produce charge transfer
or wetting with PVD copper, [4, 8].
3.2 Experiment
These experiments were conducted at temperatures between 90 K and 1200 K,
in an ultra-high vacuum (UHV) system comprised of a main chamber and sputter
deposition chamber, connected by a sample transfer port, enabling transfer without
exposure to the atmosphere, with temperature controlled by liquid nitrogen cooling and
resistive thermocouple heating. The UHV system included a main chamber, with a
base pressure of 6 x 10-10 Torr, and a dual-magnetron sputter deposition chamber, with
a base pressure of 5 x 10-9 Torr [16].
X-ray photoelectron spectroscopy (XPS) spectra were acquired by an non-
monochromated MgKα (1253.6 eV) X-ray source at 15 kV and 300 W with a commercial
hemispherical analyzer operated at 50 eV constant pass energy, in the UHV main
chamber as previously described [16]. XPS spectra were acquired with the analyzer
aligned along the surface normal (normal emission) and at glancing emission from
surface normal. Glancing emission did not produce consistent results, suggesting
64
surface roughness. Therefore the spectra reported here are with the analyzer at normal
emission.
The substrate sample for this study was a 1cm2 area titanium foil (99.97% purity,
0.0076 cm thickness) that was spot welded to titanium leads with a chromel-alumel
thermocouple spot welded to the back for heating and temperature monitoring. Cycles
of Ar+ gas bombardment at a pressure of 1 X 10-5 Torr, alternated with annealing, were
used to clean the sample. Cooling of the sample was done indirectly through the
titanium leads by nitrogen gas that was cooled by liquid nitrogen. VTMS (97% purity),
from a commercial source, was purified by cycles of freeze-thaw pumping, then
delivered ≈ 2 cm from the face of the Ti sample through a stainless steel doser tube
regulated by a manual leak valve. The concentration of VTMS delivered was measured
as gas exposure in Langmuir, L, (1L = 1 x 10-6 Torr s), uncorrected for ion gauge
sensitivity and flux to the surface.
Unmodified VTMS films were produced by condensing 30L of VTMS onto clean
Ti substrates at 90K, immediately followed by electron beam bombardment, to induce
crosslinking [5, 8, 17]. A Kimball Electronics electron gun was used to produce an
electron beam of 500eV energy (3µA emission current) with an average flux at the
sample surface of 1.5 x 10-13 electrons cm-2 s-1 [5, 8]. Nitrogen-modified films were
produced by condensing 30L of VTMS onto clean Ti substrates at 90K, immediately
followed by 30L of NH3, followed by electron beam bombardment. Nitrogen and
hydroxyl-modified films were produced by condensing 30L of VTMS onto clean Ti
substrates at 90K, immediately followed by 30L of H2O, followed by 30L of NH3 and then
followed by electron beam bombardment. All of the films were annealed to 300K in
65
UHV, after electron bombardment, to desorb any remaining condensed species.
Copper was sequentially sputter deposited (PVD) onto each annealed film’s surface, at
300K, using an Advanced Energy MDX 500 magnetron source at 25 watts, while
monitoring the Ar+ pressure with an MKS baratron gauge achieving a rate of ≈ 0.5
Å/minute. After copper deposition the sample was annealed by resistive heating,
through the thermocouple, stepwise from 300K to 1000K. Each step in temperature
was characterized by XPS to detect any resultant interactions between the PVD copper
and the film.
Film composition for both the unmodified and modified films was calculated using
relative atomic concentrations (NA/NB) for carbon, silicon, nitrogen and oxygen, from
XPS core level data with corresponding intensities (I), and appropriate elemental
sensitivity factors (A) for this analyzer [18], using the general equation:
NA/NB = [IAAB]/ [IBAA] (2)
Very thin a-Si:C:H films were produced, which allowed the XPS titanium peak to
be observed after annealing to 300K. The thickness, (d), of the a-Si:C:H films was
calculated as a function of the primary electron flux on the sample, using (ITi), the
intensity of the Ti(2p3/2) core level peak, (λ), and the inelastic mean free path of the
Ti(2p3/2) photoelectron, which is 21.6 Å [16].
ITi = I Ti (0)exp(-d/λ) (3)
The thickness of copper deposited on the modified films was estimated from the
Si(2p) intensity using the intensity ratio of copper to silicon. The intensity ratio of
reference elements, I∞Cu/I∞
Si, the copper to silicon intensity ratio, ICu/ISi, and the inelastic
66
mean free paths for copper (λCu) and silicon (λSi) are known, therefore the thickness
(dCu) can be derived from the equation:
ICu/ISi = (I∞Cu/I∞
Si)[1-exp-dCu/λCu)/exp-dSi/λSi] (4)
Absolute intensities for copper and titanium, calculated using this experimental
system, should have an error in measurement of no more than 10% [5].
3.3 Results
3.3.1 Nitrogen Modified Amorphous-Si:C:H Film Characterization
The average thickness of the unmodified a-Si:C:H cross-linked film produced
from the VTMS precursor, by the process described above, was determined (Eq. 3) to
be 20 Å. As shown in Figure 3.1, the Si(2p) XPS spectrum exhibits a peak at 100.8 eV
attributed to Si-C:H from VTMS [19], and a C(1s) spectra and a peak at 284.5 eV
attributed to C-C and C-H bond types [19], and an insignificant oxygen peak consistent
with previous results [5, 8]. The nitrogen modified silicon:carbon:hydrogen film, (a-
Si:C:H/N), had an average thickness of 27 Å, and the a-Si:C:H/OH/N films average
thickness was 31 Å. Figure 3.2 shows the normal emission spectra for the nitrogen and
nitrogen-hydroxyl modified films. The Si(2p) spectrum for the nitrogen modified film was
fit with a high binding energy peak at 102.2 eV, which was close to the 102.4 eV peak
attributed to Si-N by [20, 21], because of the limited presence of oxygen, and a 100.8
eV peak for Si-C [19]. The C(1s) spectrum included a 284.5 eV peak for C-C and C-Si,
and 285 eV [22] or 285.5 eV for C-N [21] while the nitrogen spectrum consisted of
peaks at 398.1 eV for N-Si and Si-N-O [23], and 399.2 eV for N-C [24, 25]. The
67
nitrogen-hydroxyl modified film was composed of the Si-C 100.8 eV peak and two high
binding energy peaks, 101.9 eV Si-N and 103.8 eV for Si-OH [26] which is in the
Fig. 3.1 Unmodified a-Si:C:H film showing single O(1s), C(1s) and Si(2p) species.
vicinity of 103.7 eV assigned to Si-O [21]. The C(1s) spectrum exhibits a low binding
energy feature at 283 eV, related to a species of carbon possibly bonded to the titanium
substrate, Ti/SiC [19, 27]. In addition, there is a C-N peak at 285.7 [21] and a peak at
284.5 eV for C-C and Si-C, indicating that the oxygen containing species are not
bonded to carbon atoms. The N(1s) spectrum consists of peaks at 398.1 eV for N-Si
and 399.2 eV for N-C [24, 25] and 400.8 eV for N-O [28]. The O(1s) peak (Fig. 3.2) is
fit with two components with FWHM of 2.9 eV and binding energies of 531.3 eV for NOx
and 532.9 eV for Si-OH [8, 29]. Previous work suggested the Si-OH formed in the
-540 -536 -532 -528 -524
-291 -288 -285 -282 -279
-108 -105 -102 -99 -96
C(1s)
Si(2p)
O(1s)
531.3
100.8
284.5
Binding Energy (eV)
XP
S I
nten
sity
(cps
)-540 -536 -532 -528 -524
-291 -288 -285 -282 -279
-108 -105 -102 -99 -96
C(1s)
Si(2p)
O(1s)
531.3
100.8
284.5
Binding Energy (eV)
XP
S I
nten
sity
(cps
)
68
Fig. 3.2 Comparison of XPS spectra (a) nitrogen modified a-SI:C:H film and (b) nitrogen-hydroxyl modified a-Si:C:H film.
hydroxylated a-Si:C:H films as the site for copper wetting [8]. Since there is no hydroxyl
group apparent in the nitrogen modified a-Si-C:H film environment, copper wetting may
be due to electronegative nitrogen groups C-N or Si-N, and the nitrogen-hydroxyl
modified a-Si:C:H films may enable copper wetting at the hydroxyl and nitrogen sites.
C-OH is eliminated as a possibility in the nitrogen-hydroxyl modified film because there
is no characteristic peak near 286 eV in the C(1s) spectra, and in the nitrogen modified
film there is no O (1s) peak near 533 eV, and only small peaks at 532.15 eV for ambient
oxygen contamination [30]. Table 3.1 summarizes the film component binding energies.
-538 -536 -534 -532 -530 -528 0
2000
-405 -402 -399 -396 -3930
1500
Binding Energy (eV)
XPS
Inte
nsity
(b)
-406 - -402 - -398 - -394 0
1500
-106 -104 -102 -100 -98 -96 0
2500
(a)
Si(2p)
C(1s)
O(1s)
N(1s)
-291 -288 -285 -282 -279 0
11000
-290 -288 -286 -284 -282 -2800
13000
-108 -105 -102 -99 -960
3000
-537 -534 -531 -5280
1000
69
Calculation of the C(1s)/Si(2p)total ratio, using appropriate sensitivity factors,
correcting for differences in mean free paths, yielded a C/Si ratio of 3.7:1, for a-
Si:C:H/N, and a C/Si ratio of 3.4:1, for a-Si:C:H/OH/N. These results are similar to the
C/Si ratio of 3.8:1, previously found for unmodified a-Si:C:H [5]. Prior work has found
that chemical and matrix effects, due to differences in kinetic energies, can be corrected
for by dividing by the appropriate mean free path [5]. This treatment was applied to
these results using calculated inelastic mean free path values of for C(1s) = 22.5 Å and
Si(2p) = 25.7 Å and N(1s) = 20.7 Å. The corrected C/Si ratio for a-Si:C:H/N films
increased to ≈ 4.6:1, and the corrected C/Si ratio for a-Si:C:H/OH/N films increased to ≈
4.3:1. Because of the uncertainty of the actual additional chemical and matrix effects
contributed by nitrogen in the a-Si:C:H/N film, and the effects of nitrogen and oxygen in
Si(2p)C(1s)N(1s)O(1s)a-Si:C:H/N
TiO2
Ti/Si-CSi-OHN-OC-NC-CSi-CSi-N
C-NC-CSi-CSi-N
Si(2p)C(1s)N(1s)O(1s)a-Si:C:H/OH/N
102.4398.1100.8284.5
284.5285.5399.2
283530
103.8532.9400.8531.3
288.5399.2284.5
100.8285.7398.1
Table 1 Modified a-Si:C:H Film Binding Energies (eV)
Si(2p)C(1s)N(1s)O(1s)a-Si:C:H/N
TiO2
Ti/Si-CSi-OHN-OC-NC-CSi-CSi-N
C-NC-CSi-CSi-N
Si(2p)C(1s)N(1s)O(1s)a-Si:C:H/OH/N
102.4398.1100.8284.5
284.5285.5399.2
283530
103.8532.9400.8531.3
288.5399.2284.5
100.8285.7398.1
Table 1 Modified a-Si:C:H Film Binding Energies (eV)
70
the a-Si:C:H/OH/N film, a comparison calculation was done using only the a-Si:C:H
Si(2p)(101) peak intensity to calculate the C/Si ratio for the unmodified film that resulted in
a C/Si ratio of 4.7:1. The comparison calculation using only the a-Si:C:H Si(2p)(101)
peak intensity to calculate the C/Si ratio, resulted in a ratio of ≈ 4.8:1 for the a-Si:C:H/N
film, and a C/Si ratio of ≈ 4.6:1 for the a-Si:C:H/OH/N film. Both methods of calculating
the C/Si ratio for a-Si:C:H/N, correlate with the five to one C/Si stoichiometry of VTMS.
The O(1s)total/Si(2p) total atomic ratio for the same a-Si:C:H/N film was calculated to be
0.48:1 and the N(1s)total/Si(2p) total atomic ratio was calculated to be 0.29:1. The
O(1s)total/Si(2p) total atomic ratio for the a-Si:C:H/OH/N film was calculated to be 0.60:1
and the N(1s)total/Si(2p) total atomic ratio for the same a-Si:C:H/OH/N film was calculated
to be 0.31:1. Whereas for unmodified a-Si:C:H the C(1s)/Si(2p) ratio was 4.7:1, and the
O(1s)total/Si(2p) total atomic ratio was 0.11:1. These results are similar to those found for
the hydroxyl modified film in Chapter 2, which found a C/Si ratio of 4.5:1 that correlates
well with the 5:1 VTMS stoichiometry, and a O(1s)total/Si(2p) total atomic ratio calculated
to be 0.44:1 [8].
3.3.2 Copper Deposition by Physical Vapor Deposition (PVD)
XPS was used to characterize the interactions of sputter-deposited Cu with
unmodified and modified film substrates. Cu(2p3/2) core level spectra were used to
determine Cu coverage, and x-ray excited Cu(L3VV) Auger spectra were used to
distinguish Cu(I) vs. Cu(0) oxidation states [7]. The presence of Cu(II) was monitored
by the presence or absence of typical satellite peaks in the Cu(2p3/2) spectra [31, 32],
and was not observed at any point in this experiment (Fig. 3.3). Copper was deposited
71
in the sputter deposition chamber at 300K at sputtering intervals of several minutes and
XPS data were obtained after each deposition.
Fig. 3.3 XPS spectrum showing no Cu(II) satellite peaks upon copper deposition.
No significant changes were observed for C(1s) spectra or Si(2p) spectra
intensities or binding energies after copper deposition on unmodified films (not shown).
In contrast, Cu deposition on the N-modified film results in relative attenuation of the
N(1s) intensity attributed to C-N sites (Fig. 3.4). In the a-Si:C:H/OH/N film the Si-N
102.2 eV peak was attenuated as well as the 103.8 eV Si-OH peak (Fig. 3.5). Copper
growth on the surface of nitrogen modified a-Si:C:H films, depicted in the plot of
Cu(2p3/2)/Si(2p) intensity vs. Cu deposition time (Fig. 3.6), shows a break in the uptake
curve. This break indicates a change in growth mode from conformal to 3-D island
growth similar to Stranski-Krastanov (S-K) growth, whereas unmodified a-Si:C:H films
exhibit consistent linear behavior corresponding to 3-dimensional islanding (Volmer-
-960 -950 -940 -930
0
10000
Binding Energy, eV
XPS
Inte
nsity
72
Weber type growth) [33]. Nitrogen and hydroxyl modified films produce a much more
definite break in the uptake curve for Cu(2p 3/2)/Si(2p) intensity vs. Cu deposition time
Fig. 3.4 XPS spectra for O, C, Si and N species during copper deposition on a-Si:C:H/N film.
shown in Figure 3.7 consistent with conformal growth, and similar to (S-K) growth [33],
albeit the change is much more apparent for the a-Si:C:H/OH/N films.
-405 -402 -399 -396 -3930
1500
-405 -402 -399 -396 -3930
1500
-405 -402 -399 -396 -3930
1500
-405 -402 -399 -396 -3930
1500
-108 -105 -102 -99 -960
3000
-290 -288 -286 -284 -282 -2800
13000
-537 -534 -531 -5280
1000O(1s)
8.5
0.0
6.5
2.5
Binding Energy (eV)
-108 -105 -102 -99 -960
3000
-108 -105 -102 -99 -960
3000
-537 -534 -531 -5280
1000
-537 -534 -531 -5280
1000
-290 -288 -286 -284 -282 -2800
13000-289 -287 -285 -283 -281
0
13000
-537 -534 -531 -5280
1000
XP
S I
nten
sity
(cps
)
Cop
per D
epos
ition
(min
)
N(1s)
C-N
Si-N
-108 -105 -102 -99 -960
3000 Si(2p)
Si-N
Si-C
-290 -288 -286 -284 -282 -2800
13000 C(1s)
C-N
S i-C, C-C
N-OO-O
-405 -402 -399 -396 -3930
1500
-405 -402 -399 -396 -3930
1500
-405 -402 -399 -396 -3930
1500
-405 -402 -399 -396 -3930
1500
-108 -105 -102 -99 -960
3000
-290 -288 -286 -284 -282 -2800
13000
-537 -534 -531 -5280
1000O(1s)
8.5
0.0
6.5
2.5
Binding Energy (eV)
-108 -105 -102 -99 -960
3000
-108 -105 -102 -99 -960
3000
-537 -534 -531 -5280
1000
-537 -534 -531 -5280
1000
-290 -288 -286 -284 -282 -2800
13000-289 -287 -285 -283 -281
0
13000
-537 -534 -531 -5280
1000
XP
S I
nten
sity
(cps
)
Cop
per D
epos
ition
(min
)
N(1s)
C-N
Si-N
-108 -105 -102 -99 -960
3000 Si(2p)
Si-N
Si-C
-290 -288 -286 -284 -282 -2800
13000 C(1s)
C-N
S i-C, C-C
N-OO-O
73
Figure 3.8 chronicles the evolution of Cu(L3VV) spectra during copper deposition
on both a-Si:C:H and a-Si:C:H/N films. For the first minute of copper deposition, ~ 0.5Å,
the Cu(L3VV) Auger spectra showed peaks at 914.9 eV and 918.0 eV, similar to the
formation of both Cu(I) and Cu(0) [34, 35], indicative of conformal growth [8]. With
Fig. 3.5 XPS spectra for O, C, Si and N during copper deposition on a-Si:C:H/OH/N film.
increased copper deposition, Cu films of ~ 1.0Å or more, the Cu(L3VV) peak narrowed
and shifted to 918.8 eV. This was unlike hydroxyl modified a-Si:C:H films, suggesting
weaker charge transfer for a-Si:C:H/N films as compared to Si:C:H/OH films, enabling
-405 -402 -399 -396 -393 -390
0
1400
XPS
Inte
nsity
-536 -533 -530 -5270
8000
-405 -402 -399 -396 -393 -390
0
1400
-405 -402 -399 -396 -393 -390
0
1400
-405 -402 -399 -396 -393 -390
0
1400
-106 -104 -102 -100 -98 -960
2500
-106 -104 -102 -100 -98 -960
2500-106 -104 -102 -100 -98 -96
0
2500-106 -104 -102 -100 -98 -96
0
2500
-536 -533 -530 -5270
8000
-536 -533 -530 -5270
8000
-291 -288 -285 -282 -2790
11000
-291 -288 -285 -282 -2790
11000
-291 -288 -285 -282 -2790
11000
-291 -288 -285 -282 -2790
11000
Binding Energy (eV)
N(1s)Si(2p)O(1s) C(1s)
-536 -533 -530 -527
0
2000
8.5
0.0
6.5
2.5
Cop
per D
epos
ition
(min
)
C-N Si -NSi -N
Si -C
C-N
Si -C, C-C
Ti-CN-O
O-HN-O
Si -OH
Ti-O2
-405 -402 -399 -396 -393 -390
0
1400
XPS
Inte
nsity
-536 -533 -530 -5270
8000
-405 -402 -399 -396 -393 -390
0
1400
-405 -402 -399 -396 -393 -390
0
1400
-405 -402 -399 -396 -393 -390
0
1400
-106 -104 -102 -100 -98 -960
2500
-106 -104 -102 -100 -98 -960
2500-106 -104 -102 -100 -98 -96
0
2500-106 -104 -102 -100 -98 -96
0
2500
-536 -533 -530 -5270
8000
-536 -533 -530 -5270
8000
-291 -288 -285 -282 -2790
11000
-291 -288 -285 -282 -2790
11000
-291 -288 -285 -282 -2790
11000
-291 -288 -285 -282 -2790
11000
Binding Energy (eV)
N(1s)Si(2p)O(1s) C(1s)
-536 -533 -530 -527
0
2000
8.5
0.0
6.5
2.5
Cop
per D
epos
ition
(min
)
C-N Si -NSi -N
Si -C
C-N
Si -C, C-C
Ti-CN-O
O-HN-O
Si -OH
Ti-O2
74
the predominance of copper(0) from the coalescence of Cu adatoms into Cu clusters
[34, 35]. Deposition on the unmodified a-Si:C:H film (Fig. 3.7) resulted in the exclusive
formation of the typical sharp Cu(0) peak initially observed near 916 eV that shifted to
higher binding energies upon additional deposition. Such a shift was seen before for
Fig. 3.6 Cu(2p 3/2)/Si(2p) uptake curve indicating a change in deposition mode as copper is deposited on a-Si:C:H/N films.
Fig. 3.7 Cu(2p 3/2)/Si(2p) uptake curve indicating a change in deposition mode as copper is deposited on a-Si:C:H/OH/N films.
05
10152025303540
0 5 10 15 20 25Deposition Time (min)
Cu/
SiX
PS
Inte
nsity
Rat
io
Cu(I)
Cu(0)
VTMS-N/OH
VTMSa-Si:C:H
a-Si:C:H/OH/N
Break Point
0
2
4
6
8
10
12
14
16
18
20
0.0 2.0 4.0 6.0 8.0 10.0
Deposition Time (min)
Cu/
SiX
PS In
tens
ity R
atio
Cu(I)
Cu(0)
a-Si:C:H/N
a-SI:C:H
Break Point
75
the unmodified a-Si:C:H film [5], and is characteristic of the growth of Cu(0) clusters
[36], due to enhanced screening of Auger final state holes that takes place with
increased copper cluster size [37].
For Cu deposition on the N-modified film, the narrowing of the Auger peak shape
depicted in Figure 3.8, and the increase in kinetic energy that corresponds with the
evolution of the Cu(0) feature, unlike the a-Si:C:H/OH/N film, does not correlate with the
break point in the corresponding Cu(2p3/2)/Si(2p) uptake curve. This is due to weak
charge transfer at the interface of the deposited copper and the modified a-Si:C:H/N
Fig. 3.8 Copper deposited on a-Si:C:H/N films and a-Si:C:H films show only Cu(0) species shifting to higher kinetic energy due to increase in cluster size and final state shielding.
film, which contrasts with the behavior previously observed for unmodified a-Si:C:H
films [8] that produced no change in the uptake curve, as expected for typical Volmer-
Weber growth (3-D islanding) [12]. The increase of kinetic energy of the Auger peak
890 900 910 920 930
20.5
890 900 910 920 930
1.03.56.5
8.5
14.5
a-Si:C:H Film
Aug
er In
tens
ity
Deposition Time (min)
Kinetic Energy (eV)
a-Si:C:H/N Film
Aug
er In
tens
ity
Deposition Time (min)
Kinetic Energy (eV)
0.0
6.5
3.5
2.5
1.0
5.5
8.5
Cu(0)Cu(I) Cu(0)
890 900 910 920 930
20.5
890 900 910 920 930
1.03.56.5
8.5
14.5
a-Si:C:H Film
Aug
er In
tens
ity
Deposition Time (min)
Kinetic Energy (eV)
a-Si:C:H/N Film
Aug
er In
tens
ity
Deposition Time (min)
Kinetic Energy (eV)
0.0
6.5
3.5
2.5
1.0
5.5
8.5
Cu(0)Cu(I) Cu(0)
76
and its narrowing shape is similar to previous findings [5, 8] and are the result of
shielding effects due to Cu(0) cluster formation [34, 35].
In contrast, Cu deposition on the film modified by both nitrogen and hydroxyl
groups (Fig. 3.9) shows a change in Auger peak shape, and change in kinetic energy,
which corresponds with the break point in the Cu(2p3/2)/Si(2p) uptake curve as shown in
Fig. 3.9 a) Copper deposited on a-Si:C:H/OH/N films show Cu(I) and Cu(0) species simultaneously, producing the broad peak shape.
b) Copper deposited on a-Si:C:H films show only Cu(0) species, indicated by narrow peak shape that shifts to higher kinetic energy due to increase in cluster size and final state shielding.
Figure 3.7. For the first four and a half minutes of copper deposition, ~ 2Å, or one
monolayer of Cu on the nitrogen-hydroxyl modified film at 300K (Fig. 3.8), the Cu(L3VV)
Auger spectra showed peaks at 914.9 eV and 918.0 eV, consistent with the formation
of both Cu(I) and Cu(0) [34, 35]. In this case, the behavior is similar to conformal
890 900 910 920 930 890 900 910 920 930
Cu(I) Cu(0)
Kinetic Energy (eV)
Aug
er In
tens
ity
Deposition Time (min)
a-Si:C:H-N/OH Film a-Si:C:H Film
Aug
er In
tens
ity
Deposition Time (min)
Kinetic Energy (eV)
Cu(0)
20.5
14.5
8.5
6.53.51.0
20.5
14.5
8.5
4.52.51.0
6.5
10.512.5
16.518.5
890 900 910 920 930 890 900 910 920 930
Cu(I) Cu(0)
Kinetic Energy (eV)
Aug
er In
tens
ity
Deposition Time (min)
a-Si:C:H-N/OH Film a-Si:C:H Film
Aug
er In
tens
ity
Deposition Time (min)
Kinetic Energy (eV)
Cu(0)
20.5
14.5
8.5
6.53.51.0
20.5
14.5
8.5
4.52.51.0
6.5
10.512.5
16.518.5
77
growth as previously described for hydroxyl-modified a-Si:C:H films [8], with the typical
change in slope of the uptake curve for growth modes similar to S-K growth [33],
followed by the expected increase in kinetic energy and narrowing Auger peak shape
resulting from Cu(0) cluster formation [34, 35].
3.3.3 Annealing Effects
In order to assess the stability of the Cu/substrate interface for the a-
Si:C:H/OH/N film, Cu was deposited on an a-Si:C:H/OH/N film substrate at 300 K,
followed by 15 minute anneals at increasing temperatures in UHV, with corresponding
incremental XPS acquisition at 300K. Results are shown in Figure 3.10 for a copper
film with an initial average thickness of 10 Å prior to annealing. The Cu/Si ratio is
relatively constant until 800K, after which there is a significant decrease in relative Cu
intensity. As previously stated, this decrease in relative copper intensity can been
attributed to either island formation (dewetting) or diffusion of Cu from the surface into
the bulk [4].
A measurement of the relative intensities of the Cu(2p3/2) and Cu(3p) was used to
distinguish between these two possibilities due to the marked differences in the inelastic
electron mean free paths, λ, of the two copper species, 13.9 Å for Cu(2p3/2) and 38.2 Å
for Cu(3p) [12]. Due to its shorter λ, the Cu(2p3/2) peak intensity is preferentially
attenuated relative to the Cu(3p) intensity if copper diffuses into the bulk.
A plot of the Cu(3p)total/Cu(2p3/2)total intensity ratio against temperature (Fig. 3.10)
shows some change below 800 K attributed to dewetting [4], and a significant decrease
78
in Cu(2p3/2) peak intensity and corresponding increase in the Cu(3p)/Cu(2p3/2) ratio [38]
above 800 K, indicating that annealing to temperatures of 800 K or higher induces Cu
diffusion into the nitrogen-hydroxyl modified a-Si-C:H filmas observed for Cu deposited
Fig. 3.10 Copper diffusion is signaled by the rapid increase in the Cu(3p)/Cu(2p) ratio.
on OH-modified a-Si:C:H films [5, 8], and as previously reported for unmodified a-Si:C:H
films deposited on Cu substrates [4].
Annealing above 800 K caused a large decrease in copper auger intensity (Fig.
3.11) and significant changes in the silicon XPS spectra of the a-Si:C:H/OH/N
0
10
20
30
40
0 200 400 600 800 1000 1200
Temperature, K
XP
S In
tens
it y C
u(2 p
)/Si (2
p), c
p s
XP
S In
tens
ity C
u32p
)/Cu(
2 p),
cps
10
20
30
40
0
10
20
30
40
0
Diffusion indicated
79
Fig. 3.11 Copper Auger intensity decreases with anneal from 300 K – 1000 K, and the peak becomes sharper
as it becomes more metallic.
films, suggesting complete breakdown of the films. The dominant Si(2p) 100.8 eV peak
(Si-C:H) at 300K, significantly decreased in intensity after 800K, with the appearance of
two lower energy peaks. The peak at 98.1 eV was attributed to TiSi, with the 99.6 eV
peak attributed to Si-Si bonding. At 1000 K the Si(2p) 98.1 eV (TiSi) peak became the
dominant peak, followed in intensity by the 100.8 eV (Si-C:H) peak with only 10% of its
300K intensity, and a small 99.6 eV (Si-Si) peak.
3.4 Discussion
The evolution of the Cu(0) Auger feature for deposition on the a-Si:C:H/OH/N
film (Fig. 3.9) correlates with the corresponding points in the Cu(2p3/2)/Si(2p) uptake
890 900 910 920 930
0
5000
10000
15000
20000
1000
300
800
900
300 K
800 K
1000 K
Kinetic Energy (eV)
XP
S A
uger
Inte
nsity
(cps
)
Tem
pera
ture
(K)
890 900 910 920 930
0
5000
10000
15000
20000
1000
300
800
900
300 K
800 K
1000 K
Kinetic Energy (eV)
XP
S A
uger
Inte
nsity
(cps
)
Tem
pera
ture
(K)
80
curve, depicted in Figure 3.7. The curve for the a-Si:C:H/OH/N film shows a break that
is characteristic of initial layer by layer (conformal) growth [33] with 3-D islanding, similar
to the growth seen for the OH modified a-Si:C:H film [8]. Conformal growth occurs on
the modified a-Si:C:H/OH/N film, as is indicated by significant attenuation of Si(2p)
Si:C:H intensity from the substrate during copper deposition on the A-Si:C:H/OH/N film.
This correlates with previous research that showed the conformal growth of Cu on OH-
modified a-Si:C:H films [8], whereas only weak initial conformal growth occurs on the
Si:C:H/N film (Figs. 3.6 and 3.8). Evidence for this weak growth is seen in the copper
Auger spectrum that shows both Cu(I) and Cu(0) species for the first minute of
deposition and only Cu(0) after the first minute. In contrast, the unmodified a-Si:C:H
film showed no change in the uptake slope, consistent with solitary Volmer-Weber
growth (3-D islanding) [12, 33], and a corollary sharp Cu Auger peak depicted in Figure
3.8, indicating only Cu(0) coalescence and cluster formation [34, 35].
The thickness of copper deposited on the nitrogen and nitrogen/hydroxyl
modified films was estimated from the Si(2p) intensity using the intensity ratio of copper
to silicon film substrate (Eq. 4). The average thickness of copper at the break point in
the nitrogen modified film’s Cu/Si uptake curve (Fig. 3.8), and for the nitrogen/hydroxyl
modified film’s Cu/Si uptake curve (Fig. 3.9) is ~ 2 Å, comparable to the Cu(I) ion
diameter of 1.92 Å [33] and similar to results for hydroxyl modified films [8]. Since the
initial deposition of Cu is marked by the formation of both Cu(I) and Cu(0), it is difficult to
give a precise determination of surface coverage in the first layer due to the roughness
of the film’s surface. These data estimate that the coverage of the initial Cu adlayer is ~
0.5 slightly less than 1 monolayer. The copper uptake curve data in Figure 3.7 indicate
81
that Cu deposition at 300 K on the a-Si:C:H/OH/N film results in initial layer-by-layer
growth, with the first layer including Cu(I) species along with some Cu(0) species. Upon
completion of the first layer, a second, more metallic Cu layer begins to form similar to
Cu deposition at 300 K on the N-modified a-Si:C:H film that also showed conformal
growth according to the uptake curve. However, charge transfer was not substantial
enough to accomplish copper Auger transitions to form much Cu(I) evidenced by the
domination of Cu(0) in the Auger spectra as shown in Figure 3.8. In contrast, Cu
deposited on the unmodified a-Si:C:H films shows only V-W growth indicating non-
wetting behavior.
Similar transitions from non-wetting to wetting behavior for deposited Cu have
been reported for sapphire(0001) surfaces [7], and for Rh deposited on ordered ultrathin
alumina films [6] and for hydroxyl-modified a-Si:C:H films [8]. In addition, Cu wetting
has also been reported for deposition at 300 K on hydroxylated ultrathin Ta silicate films
on Si(100) and for air-exposed TaSiN substrates [39], and theoretical studies indicate
that Pt nucleation occurs preferentially at surface OH sites on MgO single crystal
surfaces [40]. These results suggest that surface nitridation does not seem to interfere
with surface hydroxylation, enabling hydroxyl-modified films additionally modified by
nitrogen for use as copper adhesion promoters.
Copper diffusion was inhibited by the a-Si:C:H/OH/N film in a manner similar to
unmodified a-Si:C:H film at temperatures below 800K in UHV [4], as shown by a similar
change in the Cu(3p)/Cu(2p3/2) intensity ratio versus temperature curve. Annealing
above 800K caused copper to diffuse into the film and in the additional formation of
Cu(0) at the expense of Cu(I), indicated by the rapid decline in the Cu(3p)/Cu(2p3/2)
82
intensity ratio versus temperature curve. These results are similar to those found for
OH-modified a-Si:C:H films [8]. However, unlike the OH-modifieda-Si:C:H films,
annealing the a-Si:C:H/OH/N film from 500K to 800K resulted in a decrease in Si(100.8)
peak intensity, and increases in Si(102.3 eV) and Si(103.8 eV) peak intensities. This
may be due to uncharacterized reactions within the a-Si:C:H/OH/N film, or between the
film and the small amounts of ambient oxygen. Annealing to 1000K resulted in a large
decrease in intensity of the Si-C 100.8 eV, Si-N 102.3 eV and Si-OH 103.8 eV peaks,
and the production of a large 98.1 eV peak, most likely Ti3SiC2 and TiC 98.4 eV [41],
and a small peak at 99.6 eV, attributed to Si-Si [41]. These peak changes suggest the
total break down of the modified a-Si:C:H/OH/N film.
3.5 Conclusions
This study demonstrates that nitrogen modification of hydroxyl-modified a-Si:C:H
films enables Cu wetting and diffusion inhibition to 800 K in a manner similar to hydroxyl
modified a-Si:C:H films [8], as characterized by x-ray photoelectron spectroscopy
(XPS). In addition, OH/N a-Si:C:H films inhibit copper diffusion to 800 K similar to
hydroxyl-modified a-Si:C:H films, with some film compositional changes at temperatures
above 500K. These findings suggest the use of a-Si:C:H/OH/N films as possible
interlayer dielectrics for the deposition of the initial copper seed layer in interconnect
circuit. If the a-Si:C:H/OH/N film’s dielectric constant is lowered, as in previous
research, to the point that nitrogen modification decreases the dielectric constant and
leakage current [42], these films may be useful as intermetal dielectrics. Additional
83
adaptations of hydroxyl-modification of a-Si:C:H films, substituting other species in the
place of nitrogen, may result in further tailoring the properties of a-Si:C:H/OH films.
This study also demonstrates that nitrogen modification of a-Si:C:H films enables
Cu wetting as compared to unmodified a-Si:C:H films [5, 8], while approximating the
diffusion inhibition properties of unmodified a-Si:C:H films (not shown) as characterized
by x-ray photoelectron spectroscopy (XPS). These findings suggest the use of nitrogen
modified vinyltrimethylsilane derived a-Si:C:H films as a possible interlayer dielectric for
the deposition of the initial copper seed layer in interconnect circuits. If in accordance
with previous research [42], nitrogen incorporation lowers the film’s dielectric constant,
these a-Si:C:H/N films should have a lower dielectric constant than the more polarizable
a-Si:C:H/OH films, suggesting a-Si:C:H/N films as possible intermetal dielectrics.
Changing the nitrogen modification process, by substituting N2 instead of NH3 as the
source of nitrogen, or incorporating other species may result in additional modifications
of a-Si:C:H films. The resulting in a film that can be tailored in various ways to meet
unforeseen requirements associated with the continued decrease in feature sizes
predicted by Moore’s Law [43].
84
3.6 Chapter References
[1] Liu, P.-T., Chan, T.-C., Yang, Y.-L. and Cheng, Y.-F. S., S.M., IEEE Transactions
on Electron Devices 47 (2000) 1733 -1739.
[2] Vogt, M., Kachel, M., Plötner, M. and Drescher, K., Microelectronic Engineering
37/38 (1997) 181-187.
[3] McBrayer, J. D., Swanson, R. M. and Sigmon, Y. W., J. Electrochem. Soc. 133
(1986) 1242-6.
[4] Chen, L. and Kelber, J. A., J. Vac. Sci. Technol. A 17 (1999) 1968.
[5] Tong, J., Martini, D., Magtoto, N., Pritchett, M. and Kelber, J., Appl. Surf. Sci. 187
(2002) 256-264.
[6] Libuda, J., Frank, M., Sandell, A., Andersson, S., Bruhwiler, P. A., Baumer, M.,
Martenson, N. and Freund, H.-J., Surf. Sci. 384 (1997) 106.
[7] Niu, C., Shepherd, K., Martini, D., Tong, J., Kelber, J. A., Jennison, D. R. and
Bogicevic, A., Surf. Sci. 465 (2000) 163-176.
[8] Pritchett, M., Magtoto, N., Tong, J., Kelber, J. A. and Zhao, X., Thin Solid Films
440 (2003) 100-108.
[9] Lee, S. G., Kin, Y. J., Lee, S. P., Oh, H.-S., Lee, S. J., Kim, M., Kim, I.-G., Kim,
J.-H., Shin, H.-J., Hong, J.-G., Lee, H.-D. and Kang, H.-K., Jpn. J. Appl. Phys. 40
(2001) 2663-2668.
[10] Lanckmans, F., Gray, W. D., Brijs, B. and Maex, K., Microelectron. Eng. 55
(2001) 329-335.
85
[11] Deplancke, M. P., Powers, J. M., Vandentop, G. J., Salmeron, M. and Somorjai,
G. A., J. Vac. Sci. Technol. A 9 (1991) 450.
[12] Briggs, D. and Seah, M. P., in D. Briggs and M. P. Seah (Eds.): Practical Surface
Analysis, Vol. 1, Wiley, New York 1990, p. 657.
[13] Murarka, S. P., Verner, I. v. and Gutmann, R. J., Copper-Fundamental
Mechanisms for Microelectronic Applications, John Wiley & Sons, New York,
2000.
[14] Ishimaru, T., Shioya, Y., Ikakura, H., Nozawa, M., Nishimoto, Y., Ohgawara, S.
and Maeda, K., Proceedings of the International Interconnect Technology
Conference, San Francisco, CA, 2001, Institute of Electrical and Electronics
Engineers, Inc., (2001) 36-38.
[15] Liu, X. W., Lin, J. H., Jong, W. J. and Shih, H. C., Thin Solid Films 409 (2002)
178–184.
[16] Shepherd, K. and Kelber, J., Appl. Surf. Sci. 151 (1999) 287-298.
[17] Toth, A., Bertoti, I., Marletta, G., Ferenczy, G. and Mohai, M., Nucl. Instrum.
Methods Phys. Res., Sect. B 116 (1996) 299-304.
[18] "Vacuum Generators", Hastings, UK,
[19] Wagner, C. D., Naumkin, A. V., Kraut-Vass, A., Allison, J. W., Powell, C. J. and
Jr., J. R. R., www.nist.gov, "NIST X-ray Photoelectron Spectroscopy Database",
NIST, 2002.
[20] Mitu, B., Dinescu, G., Dinescu, M., Ferrari, A., Balucani, M., Lamedica, G.,
Dementjev, A. P. and Maslakov, K. I., Thin Solid Films 383 (2001) 230-234.
86
[21] Machorro, R., Samano, E. C., Soto, G. and Cota, L., Applied Surface Science
127-129 (1998) 564–568.
[22] Yamamoto, K., Koga, Y. and Fujiwara, S., Diamond and Related Materials 10
(2001) 1921�1926.
[23] Ushio, J., J. Appl. Phys. 92 (2002) 2475-2478.
[24] Barber, M., Connor, J. A., Guest, M. F., Hillier, I. H., Schwarz, M. and Stacey, M.,
J. Chem. Soc. Faraday Trans. II 69 (1973) 551.
[25] Smirnova, T. P., Badalian, A. M., Yakovkina, L. V., Kaichev, V. V., Bukhtiyarov,
V. I., Shmakov, A. N., Asanov, I. P., Rachlin, V. I. and Fomina, A. N., Thin Solid
Films 429 (2003) 144–151.
[26] Guimon, C., Gonbeau, D., Guillouzo, G. P., Dugne, O., Guette, A., Naslain, R.
and Lahaye, M., Surf. Interface Anal. 16 (1990) 440.
[27] Waldrop, J. R. and Grant, R. W., Appl. Phys. Lett. 56 (1990) 557-559.
[28] Bhattacharyya, S., Cardinaud, C. and Turban, G., J. Appl. Phys. 83 (1998) 4491-
4500.
[29] Miller, M. L. and Linton, R. W., Anal. Chem. 57 (1985) 2314-19.
[30] Barr, T. L., J. Vac. Sci. Technol. A 9 (1991) 1793-1805.
[31] Chusuei, C. C., Brookshier, M. A. and Goodman, D. W., Langmuir 15 (1999)
2806-2808.
[32] Paszti, Z., Peto, G., Horvath, A. E. and Karacs, A., J. Phys. Chem. B 101 (1997)
2109-2115.
[33] Rhead, G. E., Barthès, M.-G. and Argile, C., Thin Solid Films 82 (1981) 201.
87
[34] Klein, J. C., Proctor, A., Hercules, D. M. and Black, J. F., Anal. Chem. 55 (1983)
2055-2059.
[35] Kowalczky, S. P., Pollak, R. A., McFeely, F. R., Ley, L. and Shirley, D. A., Phys.
Rev. B 8 (1973) 2387-91.
[36] Wertheim, G. K. and DiCenzo, S. B., Phys. Rev. B 37 (1988) 844-847.
[37] Jennison, D. R. and Madden, H. H., Phys. Rev. B 21 (1980) 430-435.
[38] Chen, J. G., Colaianni, M. L., Weinberg, W. H. and J.T. Yates, J., Surf. Sci. 279
(1992) 223-232.
[39] Zhao, X., Leavy, M., Magtoto, N. P. and Kelber, J. A., Appl. Phys. Lett. 79 (2001)
3479-3481.
[40] Bogicevic, A. and Jennison, D. R., Surf. Sci. Lett. 437 (1999) L741-L747.
[41] Riley, D. P., O’Connor, D. J., Dastoor, P., Brack, N. and Pigram, P. J., J. Phys.
D: Appl. Phys. 35 (2002) 1603–1611.
[42] Fang, K. L., Tsui, B. Y., Yang, C. C. and Lee, S. D., Proceedings of the
International Interconnect Technology Conference, San Francisco, CA, 2001,
Institute of Electrical and Electronics Engineers, Inc., (2001) 250-252.
[43] public.itrs.net, public.itrs.net/Files/2002Update/2002Update.pdf, "ITRS Roadmap
for Semiconductors 2002 Update", 2003.
88
CHAPTER 4
RESEARCH SUMMARY
Semiconductor device features continue to decrease in size while increasing in
density in response to Moore’s Law resulting exposing the physical, thermal and
electrical limitations of aluminum and silicon dioxide requiring the development of
replacement metal and dielectric systems.
Copper is the obvious choice to replace aluminum due to its superior
electromigration, conductivity, resistivity and thermal properties. However Cu diffuses
into silicon dioxide under normal operating conditions, requiring the use of barrier
materials that inhibit Cu diffusion into SiO2. Barrier layers add to insulator thickness and
delay time, therefore instead of a separate barrier layer. Therefore, another option is to
replace SiO2 with dielectrics that copper can grow on conformally that also inhibit
copper diffusion.
Various amorphous silicon carbon hydrogen (a-Si:C:H) films have been
suggested as possible diffusion barriers and replacements for SiO2 as interlevel
dielectrics (ILD), or intermetal dielectrics (IMD), due to their lower dielectric constants
[1], acceptable thermal conductivity [2] and the ability of copper to grow conformally on
them [3].
This study condensed Vinyltrimethylsilane (VTMS) precursor on titanium
substrates at 90 K and bombarded the physisorbed VTMS with electron beams to
induce crosslinking and formpolymerized a-Si:C:H films. The film was modified by
89
condensing water, ammonia, or water and ammonia, on the condensed VTMS before
electron bombardment, resulting in hydroxyl modified (a-Si:C:H/OH), nitrogen modified
a-Si:C:H/N, and nitrogen and hydroxyl modified (a-Si:C:H/OH/N) polymerized films. X-
ray Photoelectron Spectroscopy (XPS) was used to analyze the films before and after
copper metallization by physical vapor deposition (PVD) to characterize any interactions
between the film and copper species that indicate the wetting behavior of copper, and
XPS was used after annealing to determine the onset of copper diffusion.
XPS results detailed in Chapter 2 show that a-Si:C:H/OH films are composed
primarily of C-C, Si-C, and Si-OH, bonded species that develop strong interactions
enabling the conformal growth of Cu (I) and Cu (0) species, as indicated by Auger
spectra, for a monolayer of copper deposition, that resists thermally-stimulated copper
diffusion at temperatures up to 800K, and complete film dissociation at ~ 1000K.
Findings for nitrogen modified a-Si:C:H films, outlined in Chapter 3, show XPS
spectra dominated by C-C, Si-C and Si-N, bond types for both a-Si:C:H/OH/N and a-
Si:C:H/N films, and Si-OH only present in a-Si:C:H/OH/N films. Copper grows
conformally on a-Si:C:H/OH/N films for the first monolayer of copper deposition when,
according to Auger spectra, both Cu (I) and Cu (0) species are present, and these films
resisted thermally-stimulated copper diffusion at temperatures up to 800K, similar to a-
Si:C:H/OH films. Before these films dissociate at 1000 K, they interact with titanium
forming Ti3SiC2 and TiC species at temperatures above 800K. a-Si:C:H//N films are
composed of C-C, Si-C and Si-N bonds with little oxygen present. Copper grows
conformally on these films according to Cu/Si intensity ratio data. However, the Cu
Auger spectra does not show the presence of Cu (I) and Cu (0) species, suggesting the
90
interface interactions produce only weak charge transfer that lasts for less than 0.5
monolayers of copper deposition before 3-D islanding dominates.
This study demonstrates thatit is possible to modify a-Si:C:H films so that copper
will grow conformally on them, while preserving their ability to wet titanium and inhibit
copperdiffusion, and indicates that the main component responsible for the conformal
growth of copper is the hydroxyl species, not nitrogen.
91
4.1 Chapter References
[1] Ahn, J.-H., Lee, K.-T., Oh, B.-J., Lee, Y.-J., Liu, S.-H., Jung, M.-K., Kim, Y.-W.
and Suh, K.-P., Journal of the Korean Physical Society 41 (2002) 422-426.
[2] Gotuaco, M., Lee, P. W., Xia, L. and Yieh, E., Semiconductor Fabtech 15 (2001)
179-183.
[3] Tong, J., Martini, D., Magtoto, N., Pritchett, M. and Kelber, J., Appl. Surf. Sci. 187
(2002) 256-264.[1] Ahn, J.-H., Lee, K.-T., Oh, B.-J., Lee, Y.-J., Liu, S.-H., Jung,
M.-K., Kim, Y.-W. and Suh, K.-P., Journal of the Korean Physical Society 41
(2002) 422-426.
[2] Gotuaco, M., Lee, P. W., Xia, L. and Yieh, E., Semiconductor Fabtech 15 (2001)
179-183.
[3] Tong, J., Martini, D., Magtoto, N., Pritchett, M. and Kelber, J., Appl. Surf. Sci. 187
(2002) 256-264.
92
REFERENCES
Ahn, J.-H., Lee, K.-T., Oh, B.-J., Lee, Y.-J., Liu, S.-H., Jung, M.-K., Kim, Y.-W. and Suh,
K.-P., Journal of the Korean Physical Society 41 (2002) 422-426.
Albertin, K. F., Pereyra, I. and Alayo, M. I., Materials Characterization 5568 (2003) .
Anderson, G. P., Bennett, S. J. and DeVries, K. L., Analysis and Testing of Adhesive
Bonds, Academic Press, New York 1977, p. 15-16.
Aubin, E. and Lewis, L. J., Phys. Rev. B 47 (1993) 6780-6783.
Barber, M., Connor, J. A., Guest, M. F., Hillier, I. H., Schwarz, M. and Stacey, M., J.
Chem. Soc. Faraday Trans. II 69 (1973) 551.
Barnes, C. and Lakins, D., http://nepp.nasa.gov/index_nasa.cfm/934/, "Technology
Readiness Overview - Low-k Interlevel Dielectrics Technology - Brief description
of low-k technology",NASA/NEPP Program, 2003.
Barr, T. L., J. Vac. Sci. Technol. A 9 (1991) 1793-1805.
Barr, T. L., J. Phys. Chem. 82 (1978) 1801-1810.
Baumann, J., Markert, M., Wemer, T., Ehrlich, A., Rennau, M., Kaufmann, C. and
Gessner, T., Microelectronic Engineering 37/38 (1997) 229-236.
Baumann, J., Werner, T., Ehrlich, A., Rennau, M., Kaufmann, C. and Gessner, T.,
Microelectronic Engineering 37/38 (1997) 221-228.
Beebe, J. T. P., , http://www.uwo.ca/ssw/archives/1999/nov99/0106.html 2003.
Bhattacharyya, S., Cardinaud, C. and Turban, G., J. Appl. Phys. 83 (1998) 4491-4500.
Blanquet, E., Dutron, A. M., Ghetta, V., Bernard, C. and Madar, R., Microelectronic
Engineering 37/38 (1997) 189-195.
93
Bogicevic, A. and Jennison, D. R., Surf. Sci. Lett. 437 (1999) L741-L747.
Briggs, D. and Seah, M. P., in D. Briggs and M. P. Seah (Eds.): Practical Surface
Analysis, Vol. 1, Wiley, New York 1990, p. 657.
Chan, C. M., Cao, G. Z., Fong, H., Sarikaya, M., Robinson, T. and Nelson, L., J. Mater.
Res. 15 (2000) 148-154.
Chang, C.-A., Baglin, J. E. E., Schrott, A. G. and Lin, K. C., Appl. Phys. Lett. 51 (1987)
103.
Chapman, L., http://www.electronicmaterials.com/news_events/download/111401.pdf,
"International SEMATECH Fabricates Electrically Yielding 2-Level Metal Copper
Dual Damascene Via Chains with NANOGLASSâ E Extra Low-k Spin-On
Dielectric", www.electronicmaterials.com., Sunnyvale, California, 2003.
Chen, J. G., Colaianni, M. L., Weinberg, W. H. and J.T. Yates, J., Surf. Sci. 279 (1992)
223-232.
Chen, L. and Kelber, J. A., J. Vac. Sci. Technol. A 17 (1999) 1968.
Chen, L., Magtoto, N., Ekstrom, B. and Kelber, J., Thin Solid Films 376 (2000) 115-123.
Christensen, T. M.,
http://www.uccs.edu/~tchriste/courses/PHYS549/549lectures/index.html,
"Physcis of Thin Films", University of Colorado at Colorado Springs, Colorado
Springs, CO, 2003.
Chusuei, C. C., Brookshier, M. A. and Goodman, D. W., Langmuir 15 (1999) 2806-
2808.
Cote, D. R., Nguyen, S. V., Cote, W. J., Pennington, S. L., Stamper, A. K. and
Podlesnik, D. V., IBM J. RES. DEVELOP. 39 (1995) 437-464.
94
Dauskardt, R. H., Lane, M., Ma, Q. and Krishna, N., Engineering Fracture Mechanics 61
(1998) 141-162.
Deplancke, M. P., Powers, J. M., Vandentop, G. J., Salmeron, M. and Somorjai, G. A.,
J. Vac. Sci. Technol. A 9 (1991) 450.
Ding, P. J., Lanford, W. A., Hymes, S. and Murarka, S. P., J. Appl. Phys. 75 (1994)
3627 - 3631.
Dowcorning, http://www.dowcorning.com/applications/Product_Finder/PF_
details.asp?l1=009&prod=02334909&type=PROD&pg=00000037, " FOX(R) -12
FLOWABLE OXIDE", Dow Corning, Midland, Michigan, 2003.
Fang, K. L., Tsui, B. Y., Yang, C. C. and Lee, S. D., Proceedings of the International
Interconnect Technology Conference, San Francisco, CA, 2001, Institute of
Electrical and Electronics Engineers, Inc., (2001) 250-252.
Films, E. M. A. T. C., http://www.eurometallizers.org/download/testmetaltape.pdf, "TEST
PROCEDURE METAL ADHESION TAPE TEST", European Metallizers
Association, 2003.
Fujita, K. and Oya, A., J. Mater. Sci. (UK) 31 (1996) 4609-4615.
Godbey, D. K., Buckley, L. J., Purdy, A. P. and Snow, A. W., Thin Solid Films 308-309
(1997) 470-474.
Gotuaco, M., Lee, P. W., Xia, L. and Yieh, E., Semiconductor Fabtech 15 (2001) 179-
183.
Guimon, C., Gonbeau, D., Guillouzo, G. P., Dugne, O., Guette, A., Naslain, R. and
Lahaye, M., Surf. Interface Anal. 16 (1990) 440.
95
Hackman, "Interconnects", http://hackman.mit.edu/6152J/LectureNotes/
6.152J.FT01.Lecture19-3.pdf, 2003.
Heemeier, M., Frank, M., Libuda, J., Wolter, K., Kuhlenbeck, H., Baumer, M. and
Freund, H. J., Catalysis Letters 68 (2000) 19-24.
Himpsel, F. J., McFeely, F. R., Taleb-Ibrahimi, A. and Yarmoff, J. A., Phys. Rev. B 38
(1988) 6084-6096.
Hollinger, G., Appl. Surf. Sci. 8 (1981) 318-36.
Iiojima, Y. and Sonoda, N., Clean Technol. 12 (1992) 39.
Ishimaru, T., Shioya, Y., Ikakura, H., Nozawa, M., Nishimoto, Y., Ohgawara, S. and
Maeda, K., Proceedings of the International Interconnect Technology
Conference, San Francisco, CA, 2001, Institute of Electrical and Electronics
Engineers, Inc., (2001) 36-38.
Ishimaru, T., Shioya, Y., Ikakura, H., Nozawa, M., Ohgawara, S., Ohdaira, T., Suzuki, R.
and Maeda, K., Journal of The Electrochemical Society 150 (2003) F83-F89.
JangJean, S.-K., Wang, Y.-L., Liu, C.-P., Hwang, W.-S., Tseng, W.-T. and Liu, C.-W.,
Journal of Applied Physics 94 (2003) 732-737.
Jennison, D. R. and Madden, H. H., Phys. Rev. B 21 (1980) 430-435.
Ke, R., Haasch, R. T., Finnegan, N., Dottl, L. E., Alkire, R. C. and Farrell, H. H., J. Vac.
Sci. Technol. A 14 (1995) 80-88.
Kim, H. J., Shao, Q. and Kim, Y.-H., Surface and Coatings Technology 171 (2003) 39-
45.
Kim, S. H., Lee, D. W., Chung, K.-H., Park, J. K., Jaung, J.-Y. and Jeong, S. H., Journal
of Applied Polymer Science 86 (2002) 812-820.
96
Klein, J. C., Proctor, A., Hercules, D. M. and Black, J. F., Anal. Chem. 55 (1983) 2055-
2059.
Kohn, A., Eizenberg, M., Shacham-Diamond, Y., Israel, B. and Sverdlov, Y.,
Microelectronic Engineering 55 (2001) 297-303.
Komori, F., Kushida, K., Hattori, K., Arai, S. and Iimori, T., Surf. Sci. 438 (1999) 123.
Kowalczky, S. P., Pollak, R. A., McFeely, F. R., Ley, L. and Shirley, D. A., Phys. Rev. B
8 (1973) 2387-91.
Kudo, H., Yoshie, Y., Yamaguchi, S., Watanabe, K., Ikeda, M., Kakamu, K., Hosoda, T.,
Ohhira, K., Santoh, N., Misawa, N., Matsuno, K., Wakasugi, Y., Hasegawa, A.,
Nagase, K. and Suzuki, T., Interconnect Technology Conference, 2000.
Proceedings of the IEEE 2000 International , 5-7 June 2000, Burlingame, CA,
2000, IEEE, (2000) 270 -272.
Lanckmans, F., Gray, W. D., Brijs, B. and Maex, K., Microelectron. Eng. 55 (2001) 329-
335.
Laurila, T., Zeng, K., Kivilahti, J. K., Molarius, J., Riekkinen, T. and Suni, I.,
Microelectronic Engineering 60 (2003) 71-80.
Lee, G.-W., Lee, J.-H., Kim, Y.-M., Lee, W.-C., Kim, H.-K., You, K.-D. and Park, C.-K.,
Journal of the Korean Physical Society 41 (2002) 912-917.
Lee, M.-S. and Bent, S. F., J. Vac. Sci. Technol. A 16 (1998) 1658-1663.
Lee, S. G., Kin, Y. J., Lee, S. P., Oh, H.-S., Lee, S. J., Kim, M., Kim, I.-G., Kim, J.-H.,
Shin, H.-J., Hong, J.-G., Lee, H.-D. and Kang, H.-K., Jpn. J. Appl. Phys. 40
(2001) 2663-2668.
97
Li, J., Nguyen, Q. T., Zhou, L. Z., Wang, T., Long, Y. C. and Ping, Z. H., Desalination
147 (2002) 321-326.
Libuda, J., Frank, M., Sandell, A., Andersson, S., Bruhwiler, P. A., Baumer, M.,
Martenson, N. and Freund, H.-J., Surf. Sci. 384 (1997) 106.
Liu, G. and Freund, M. S., Macromolecules 30 (1997) 5660-5665.
Liu, P.-T., Chan, T.-C., Yang, Y.-L. and Cheng, Y.-F. S., S.M., IEEE Transactions on
Electron Devices 47 (2000) 1733 -1739.
Liu, X. W., Lin, J. H., Jong, W. J. and Shih, H. C., Thin Solid Films 409 (2002) 178-184.
Loboda, M. J., Microelectronics Engineering 50 (2000) 15-23.
M.P.Seah, P.J.Cumpson and M.T.Brown, http://www.npl.co.uk/nanoanalysis/x1
userguide.pdf, "NPL X-Ray Photoelectron Spectrometer Intensity Calibration
Software, Version 2.0User's Guide", Crown, Middlesex, UK, 2000.
Machorro, R., Samano, E. C., Soto, G. and Cota, L., Applied Surface Science 127-129
(1998) 564-568.
Mahan, J. E., Physical Vapor Deposition of Thin Films, John Wiley & Sons, New York,
2000.
Marcadal, C., Richard, E., Torres, J., Palleau, J., Ulmer, L., Perroud, L., Piaguet, J. and
Rolland, G., Microelectronic Engineering 37/38 (1997) 197-203.
Mason, M. G., Phys. Rev. B 27 (1983) 748-762.
McBrayer, J. D., Swanson, R. M. and Sigmon, Y. W., J. Electrochem. Soc. 133 (1986)
1242-6.
McIntyre, N. S. and Chan, T. C., in D. Briggs and M. P. Seah (Eds.): Practical Surface
Analysis, Vol. 1, Wiley, New York 1990, p. 485-529.
98
Miller, M. L. and Linton, R. W., Anal. Chem. 57 (1985) 2314-19.
Miller, T., Rosenwinkel, E. and Chiang, T.-C., Phys. Rev. B 30 (1984) 570-577.
Mitu, B., Dinescu, G., Dinescu, M., Ferrari, A., Balucani, M., Lamedica, G., Dementjev,
A. P. and Maslakov, K. I., Thin Solid Films 383 (2001) 230-234.
Moretti, G., J. Electron Spectrosc. Relat. Phenom. 95 (1998) 95-144.
Moulder, J. F., Stickle, W. F., Sobol, P. E. and Bomben, K. D., Handbook of X-ray
Photoelectron Spectroscopy, Physical Electronics, Eden Prairie, MN, 1995.
Murarka, S. P., Verner, I. v. and Gutmann, R. J., Copper-Fundamental Mechanisms for
Microelectronic Applications, John Wiley & Sons, New York, 2000.
Namkansiorn, A., Ghatak, A., Chaudhury, M. K. and Berry, D. H., J. Adhesion Sci.
Technol. 15 (2001) 1725-1745.
Nilsson, A., http://www-ssrl.slac.stanford.edu/nilssongroup/pages/XPS_description.html,
"X-ray Photoelectron Spectroscopy (XPS)", Surface Science and X-Ray
Spectroscopy Group, Sand Hill Road, CA, 2003.
Niu, C., Shepherd, K., Martini, D., Tong, J., Kelber, J. A., Jennison, D. R. and Bogicevic,
A., Surf. Sci. 465 (2000) 163-176.
Nuzzo, R. G., http://augustus.scs.uiuc.edu/nuzzogroup/PPT/XPS%20Class%2099.PPT,
"X-Ray Photoelectron Spectroscopy (XPS)", University of Illinois at Urbana-
Cahpaign, Urbana, Illinois, 2003.
Paszti, Z., Peto, G., Horvath, A. E. and Karacs, A., J. Phys. Chem. B 101 (1997) 2109-
2115.
Paterson, M. J., Diamond Relat. Mater. 5 (1996) 1407-1413.
99
Pellicori, S., http://www.cerac.com/pubs/cmn/cmn6_3.htm, "Topics in Sputter
Deposition", Coating Materials News, Milwaukee, WI, 2003.
Powell, C. J., Jablonski, A., Tilinin, I. S., Tanuma, S. and Penn, D. R., J. Electron
Spectrosc. Relat. Phenom. 98-99 (1999) 1-15.
Pritchett, M., Magtoto, N., Tong, J., Kelber, J. A. and Zhao, X., Thin Solid Films 440
(2003) 100-108.
public.itrs.net, public.itrs.net/Files/2002Update/2002Update.pdf, "ITRS Roadmap for
Semiconductors 2002 Update", 2003.
Rhead, G. E., Barthès, M.-G. and Argile, C., Thin Solid Films 82 (1981) 201.
Riley, D. P., O'Connor, D. J., Dastoor, P., Brack, N. and Pigram, P. J., J. Phys. D: Appl.
Phys. 35 (2002) 1603-1611.
Rodbell, K. P., in G. C. Schwartz, K. V. Srikrishnan and A. Bross (Eds.): Handbook of
Semiconductor Interconnection Technology, Marcel Dekker, Inc 1998, p. 473-
506.
Rye, R. R., Knapp, J. A., Chi, K.-M., Hampden-Smith, M. J. and Kodas, T. T., J. Appl.
Phys. 72 (1992) 5941-5947.
Schwartz, G. C., in G. C. Schwartz, K. V. Srikrishnan and A. Bross (Eds.): Handbook of
Semiconductor Interconnection Technology, Marcel Dekker, Inc., New York
1998, p. 77-159.
Seah, M. P. and Smith, G. C., in D. Briggs and M. P. Seah (Eds.): Practical Surface
Analysis, Vol. 1, Wiley, New York 1990, p. 534-540.
Seah, M. P., in D. Briggs and M. P. Seah (Eds.): Practical Surface Analysis, Vol. 1, New
York 1990, p. 201-255.
100
Shenton, M. J., Lovell-Hoare, M. C. and Stevens, G. C., J. Phys. D: Appl. Phys. 34
(2001) 2754-2760.
Shepherd, K. and Kelber, J., Appl. Surf. Sci. 151 (1999) 287-298.
Shim, C. and Jung, D., http://tflab.skku.ac.kr/low-k/low-k%20introduction%202003.pdf,
"Introduction to Inter-Metallic Dielectrics & Cu Integration for beginners in our
lab.", Semiconductor & Nano Application Lab Dept. of Physics, Sungkyunkwan
Universtiy, Chunchun-Dong, Jangan-Gu, Suwon, 440-746, South Korea, 2003.
Smirnova, T. P., Badalian, A. M., Yakovkina, L. V., Kaichev, V. V., Bukhtiyarov, V. I.,
Shmakov, A. N., Asanov, I. P., Rachlin, V. I. and Fomina, A. N., Thin Solid Films
429 (2003) 144-151.
Sun, Y.-M., Lee, S. Y., Lemonds, A. M., Engbrecht, E. R., Veldman, S., Lozano, J.,
White, J. M., Ekerdt, J. G., Emesh, I. and Pfeifer, K., Thin Solid Films 397 (2001)
109-115.
Thundat, T. and Sharp, S. L., Applied Physics Letters 66 (1995) 1563.
Thundat, T. and Sharp, S. L., Applied Physics Letters 75 (1999) 3998-4000.
Tong, J., Martini, D., Magtoto, N., Pritchett, M. and Kelber, J., Appl. Surf. Sci. 187
(2002) 256-264.
Toth, A., Bertoti, I., Marletta, G., Ferenczy, G. and Mohai, M., Nucl. Instrum. Methods
Phys. Res., Sect. B 116 (1996) 299-304.
Tsai, M. H., Sun, S. C., Lee, C. P., Chiu, H. T., Tsai, C. E., Chuang, S. H. and Wu, S.
C., Thin Solid Films 270 (1995) 531-536.
Ushio, J., J. Appl. Phys. 92 (2002) 2475-2478.
Vacuum Generators, Hastings, UK,
101
Varma, S. and Chottiner, G. S., J. Vac. Sci. Technol. 10 (1992) 2857-2862.
Vella, J. B., Adhihetty, I. S., Junker, K. and Volinsky, A. A., International Journal of
Fracture 119 (2003) 487-499.
Vogt, M., Kachel, M., Plötner, M. and Drescher, K., Microelectronic Engineering 37/38
(1997) 181-187.
Wagner, C. D., Naumkin, A. V., Kraut-Vass, A., Allison, J. W., Powell, C. J. and Jr., J. R.
R., www.nist.gov, "NIST X-ray Photoelectron Spectroscopy Database", NIST,
2002.
Waldrop, J. R. and Grant, R. W., Appl. Phys. Lett. 56 (1990) 557-559.
Wasa, K. and Hayakawa, S., Handbook of Sputter Deposition Technology, Principles
Technology and Applications, Noyes, Park Ridge, NJ, 1992.
Wasa, K. and Hayakawa, S., in R. F. Bunshah and G. E. McGuire (Eds.): Handbook of
Sputter Deposition Technology, Noyes Publications, Park Ridge, New Jersey
1992, p. 32.
Wertheim, G. K. and DiCenzo, S. B., Phys. Rev. B 37 (1988) 844-847.
Wolf, S. and Tauber, R. N., Silicon Processing for the VLSI Era, Vol.1, Lattice Press,
Sunset Beach, CA, 1986.
Wu, P. K., Dabral, S., Yang, G.-R., Gittleman, B., Li, C., McDonald, J. F. and Lu, T.-M.,
J. Appl. Phys 80 (1996) 5759-5764.
Wu, Y., Garfunkel, E. and Madey, T. E., J. Vac. Sci. Technol: A 14 (1996) 1662-1667.
www.casaxps.com, http://www.casaxps.com/help_manual/
XPSInformation/XPSInstr.htm, "XPS Instrumentation diagram", 2003.
www.sheldahl.com, "http://www.sheldahl.com/Product/TMQualityAssur.htm", 2003.
102
Yamamoto, K., Koga, Y. and Fujiwara, S., Diamond and Related Materials 10 (2001)
1921 1926.
Yampol'skii, Y. P. and Volkov, V. V., J. Membr. Sci. 64 (1991) 191-228.
Yin, K.-M., Chang, L., Chen, F.-R., Kai, J.-J., Chiang, C.-C., Chuang, G., Ding, P., Chin,
B., Zhang, H. and Chen, F., Thin Solid Films 388 (2001) 27-33.
Yoshimaru, M., Koizumi, S. and Shimokawa, K., J. Vac. Sci. Technol. A 15 (1997) 2915-
2922.
Yost, G., Hosking, F. M. and Frear, D. R., The Mechanics of Solder Alloy Wetting &
Spreading, Van Nostrand Reinhold, New York, 1993.
Zhao, X., Leavy, M., Magtoto, N. P. and Kelber, J. A., Appl. Phys. Lett. 79 (2001) 3479-
3481.
Zou, W., Ph. D. Dissertation, University of Virginia, Charlottesville, Virginia 2001, p. 303.