© 2009 dojun kimufdcimages.uflib.ufl.edu/uf/e0/04/10/42/00001/kim_d.pdf · 2013. 5. 31. ·...
TRANSCRIPT
1
CHEMICAL VAPOR DEPOSITION OF TUNGSTEN-BASED DIFFUSION BARRIER THIN
FILMS FOR COPPER METALLIZATION
By
DOJUN KIM
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
2009
2
© 2009 Dojun Kim
3
To my parents, Hwakyum Kim and Hyosun Kim
4
ACKNOWLEDGMENTS
I would like to thank my research advisor, Dr. Timothy J. Anderson, for his full support
and excellent guidance through my four year‟s study in University of Florida. I would like to
thank the members of supervisory committee, Dr. Lisa McElwee-White, Dr. Valentin Craciun,
and Dr. Fan Ren who provided me with valuable comments in my work. The work on
maintenance of CVD and ALD systems would not have been successful if there was no help and
support of Dennis Vince (Chemical Engineering, UF), Jim Hinnant (Chemical Engineering, UF),
and Rob Holobof (A&N Corporation). The excellent facilities and helpful staffs for material
characterizations at Major Analytical Instrumentation Center (MAIC) were highly helpful for me
to obtain valuable results. I would like to thank Eric Lambers (XPS/AES), Kerry Siebein
(TEM/EDS) for their assistance. I also would like to thank Dr. Ivan Kravchenko (Sputter
deposition system) at Nanofabrication Facilities. I also would like to thank Dr. Khalil Abboud
(Structural chemistry of X-Ray diffraction) at Department of Chemistry. I give my thanks to my
colleagues of research project for their assistance: Oh Hyun Kim, Jooyoung Lee, Dr. Jürgen
Koller, Dr. Lii-Cherng Leu, Dr. Kee Chan Kim, Dr. Hiral M. Ajmera, Dr. Michael June, and
Christopher O‟Donohue. Especially, my fellow doctoral student, Oh Hyun Kim was an excellent
collaborator for my research in many ways. The last but not the least, I would like to give my
sincerest thanks to my parents, Hwakyum Kim and Hyosun Kim, for their unconditional love and
support. I am very grateful to my wife, Sora Park, for her love and care in my life. My lovely
son, Jinho Kim and my lovely daughter, Katherine Nayoun Kim, always made me happy to work
harder and harder during my studies.
5
TABLE OF CONTENTS
page
ACKNOWLEDGMENTS ........................................................................................................... 4
LIST OF TABLES ...................................................................................................................... 8
LIST OF FIGURES .................................................................................................................... 9
ABSTRACT ............................................................................................................................. 14
CHAPTER
1 INTRODUCTION ............................................................................................................. 16
2 LITERATURE REVIEW ................................................................................................... 19
2.1 Diffusion Mechanism in Cu Metallization ................................................................ 19
2.2 Ta/TaN Bilayer Structure as a Diffusion Barrier ...................................................... 21
2.3 Chemical Vapor Deposition of Tungsten-Based Diffusion Barrier ........................... 23
2.3.1 Tungsten Nitride as a Diffusion Barrier ........................................................ 23
2.3.2 Tungsten Carbonitride as a Diffusion Barrier................................................ 25
2.4 Atomic Layer Deposition of Tungsten-Based Diffusion Barrier ............................... 29
2.4.1 Tungsten Nitride as a Diffusion Barrier ........................................................ 29
2.4.2 Tungsten Carbonitride as a Diffusion Barrier................................................ 31
3 EXPERIMENTAL PROCEDURE ..................................................................................... 40
3.1 Precursor Synthesis .................................................................................................. 40
3.2 Film Growth ............................................................................................................ 40
3.3 Film Characterizations ............................................................................................. 40
3.4 Diffusion Barrier Testing ......................................................................................... 41
4 DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(N-pip) AS A SINGLE-SOURCE
PRECURSOR .................................................................................................................... 47
4.1 X-ray Crystallographic Study of Cl4(CH3CN)W(N-pip) ........................................... 47
4.2 Preliminary Precursor Screening .............................................................................. 47
4.3 Film Structure .......................................................................................................... 48
4.4 Chemical Composition ............................................................................................. 49
4.5 Chemical Bonding States ......................................................................................... 50
4.6 Lattice Parameter ..................................................................................................... 51
4.7 Average Grain Size .................................................................................................. 52
4.8 Electrical Resistivity ................................................................................................ 52
4.9 Film Growth Rate .................................................................................................... 53
4.10 Diffusion Barrier Testing ......................................................................................... 53
4.11 Conclusions ............................................................................................................. 54
6
5 DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(N-pip): EFFECT OF NH3 ON FILM
PROPERTIES .................................................................................................................... 67
5.1 Film Structure .......................................................................................................... 67
5.2 Surface Morphology ................................................................................................ 68
5.3 Chemical Composition ............................................................................................. 68
5.4 Chemical Bonding States ......................................................................................... 70
5.5 Film Growth Rate .................................................................................................... 72
5.6 Electrical Resistivity ................................................................................................ 73
5.7 Diffusion Barrier Testing ......................................................................................... 73
5.8 Conclusions ............................................................................................................. 74
6 DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(NNPh2) AS A SINGLE-SOURCE
PRECURSOR .................................................................................................................... 83
6.1 Film Structure .......................................................................................................... 83
6.2 Lattice Parameter and Average Grain Size ............................................................... 83
6.3 Chemical Composition ............................................................................................. 84
6.4 Chemical Bonding States ......................................................................................... 85
6.5 Film Growth Rate .................................................................................................... 87
6.6 Electrical resistivity ................................................................................................. 87
6.7 Diffusion Barrier Testing ......................................................................................... 88
6.8 Conclusions ............................................................................................................. 89
7 DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(NNPh2): EFFECT OF NH3 ON
FILM PROPERTIES .......................................................................................................... 97
7.1 Film Structure .......................................................................................................... 97
7.2 Chemical Composition ............................................................................................. 97
7.3 Chemical Bonding States ......................................................................................... 99
7.4 Surface Morphology .............................................................................................. 100
7.5 Film Growth Rate .................................................................................................. 101
7.6 Electrical Resistivity .............................................................................................. 101
7.7 Diffusion Barrier Testing ....................................................................................... 102
7.8 Conclusions ........................................................................................................... 103
8 DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(NNMe2): EFFECT OF NH3 ON
FILM PROPERTIES ........................................................................................................ 114
8.1 Film Structure ........................................................................................................ 114
8.2 Chemical Composition ........................................................................................... 114
8.3 Chemical Bonding States ....................................................................................... 116
8.4 Surface Morphology .............................................................................................. 118
8.5 Film Growth Rate .................................................................................................. 118
8.6 Electrical Resistivity .............................................................................................. 118
8.7 Conclusions ........................................................................................................... 119
9 REACTOR MODELING USING CFD SOFTWARE ...................................................... 125
7
9.1 Description of the Raman-Assisted CVD reactor.................................................... 125
9.2 Multiphase Flow Simulation of the Raman-Assisted CVD reactor ......................... 126
10 CONCLUSIONS AND FUTURE WORK ........................................................................ 137
10.1 Ru-WNxCy for Diffusion Barrier and Cu Direct-Plate Applications ........................ 137
10.2 WNxCy for Realistic Diffusion Barrier Testing ....................................................... 138
LIST OF REFERENCES ........................................................................................................ 140
BIOGRAPHICAL SKETCH ................................................................................................... 146
8
LIST OF TABLES
Table page
2-1 Precursors used for film growth of WNx by CVD .......................................................... 35
2-2 Precursors used for film growth of WNxCy by CVD ...................................................... 35
2-3 Precursors used for film growth of WNx by ALD........................................................... 36
2-4 Precursors used for film growth of WNxCy by ALD ....................................................... 36
4-1 Crystal data and structure refinement for Cl4(CH3CN)W(N-pip) (1) .............................. 56
4-2 Selected bond distances (Å ) and angles (°) for Cl4(CH3CN)W(N-pip) (1) ...................... 57
4-3 Reported binding energy (BE) values ............................................................................ 58
9-1 Boundary conditions for CVD reactor .......................................................................... 129
9
LIST OF FIGURES
Figure page
1-1 The device delay as a function of device generation. Adopted from M. T. Bohr,
“Interconnect scaling – the real limiter to high performance ULSI”, Proceedings of
IEEE International Electron Devices Meeting (1995) 241-242. ...................................... 18
1-2 SEM cross-sectional images: A) Cu deposition without Cu diffusion barrier; B) Cu
deposition with Cu diffusion barrier. .............................................................................. 18
2-1 Microstructure of Cu diffusion barrier materials: A) single crystal; B)
polycrystalline; C) polycrystalline columnar; D) nano-crystalline; E) amorphous.
Adopted from A. Kaloyeros and E. Eisenbraun, “Ultrathin diffusion barrier/liners for
gigascale copper metallization”, Annu. Rev. Mater. Sci. 30 (2000) 363-385. ................. 37
2-2 Diagram showing the applications of metals and nitrides in modern semiconductor
devices. Adopted from H. Kim, “Atomic layer deposition of metal and nitride thin
films: Current research efforts and applications for semiconductor device
processing”, J. Vac. Sci. Technol. B 21 (2003) 2232-2261. ............................................ 38
2-3 Simplified processing steps in dual-damascene structure for Cu metallization. ............... 39
3-1 The diorganohydrazido(2-) tungsten complexes Cl4(CH3CN)W(NNR2) (1: R2 =
-(CH2)5-; 2: R2 = Ph2; 3: R2 = Me2). ............................................................................... 43
3-2 Schematic diagram of the aerosol-assisted CVD system................................................. 44
3-3 Process flow on film properties. (MAIC, http:\\maic.mse.ufl.edu, October, 2008). ......... 45
3-4 Process flow on diffusion barrier testing. (MAIC, http:\\maic.mse.ufl.edu, October,
2008). ............................................................................................................................ 46
4-1 Thermal ellipsoids diagram of the molecular structure of Cl4(CH3CN)W(N-pip) (1).
Thermal ellipsoids are drawn at 50% probability. H atoms are omitted for clarity. ........ 59
4-2 XRD spectra for films deposited on Si(100) in H2 carrier: A) 300 °C, B) 700 °C, C)
between 300 and 700 °C, and D) standard diffraction plots for β-W2N and β-WC1-x. ..... 60
4-3 Variation in chemical composition of W, N, C, and O content in the films with
deposition temperature. Data are measured by XPS after 10 min Ar+ ion sputter. .......... 61
4-4 Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature. Data are from XPS after 10 min Ar+ ion sputter. ...................... 62
4-5 Change in lattice parameter with deposition temperature for polycrystalline films
deposited from 1 based on β-WNxCy(111) diffraction peaks. ......................................... 63
10
4-6 Change in average grain size with deposition temperature for polycrystalline films
deposited from 1 based on β-WNxCy(111) diffraction peaks. ......................................... 63
4-7 Change in film resistivity with deposition temperature. Data are measured by four-
point probe. ................................................................................................................... 64
4-8 Change in growth rate with deposition temperature. Thickness measured by cross-
sectional SEM. .............................................................................................................. 64
4-9 The performance of diffusion barrier by XRD measurement for Cu/WNxCy/Si stacks
before and after annealing at 500 °C. ............................................................................. 65
4-10 SEM images of Si surface after etch-pit test A) before annealing and B) after
annealing at 500 °C. ...................................................................................................... 65
4-11 The performance of diffusion barrier by AES depth profile for Cu/WNxCy/Si stacks
after annealing at 500 °C. .............................................................................................. 66
5-1 XRD spectra for films deposited on Si(100) with NH3: A) between 300 and 700 °C;
B) standard diffraction patterns for β-W2N and β-WC1-x. ............................................... 76
5-2 Surface morphology of films deposited on Si(100) substrate at various temperature:
A) 300 °C without NH3; B) 600 °C without NH3; C) 300 °C with NH3; D) 600 °C
with NH3. ...................................................................................................................... 77
5-3 Variation in chemical composition of A) W, B) N, C) C, and D) O content in the
films with deposition temperature with and without added NH3. Data are measured
by XPS after 10 min Ar+ ion sputter............................................................................... 78
5-4 Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature in the presence of NH3. Data are from XPS after 10 min Ar+
ion sputter. ..................................................................................................................... 79
5-5 Change in growth rate with deposition temperature for films deposited with and
without added NH3. Thickness was measured by cross-sectional SEM. ......................... 80
5-6 Change in film resistivity with deposition temperature with and without added NH3.
Data are measured by four-point probe. ......................................................................... 80
5-7 The performance of diffusion barrier by XRD measurement for Cu/WNxCy/Si stacks
before and after annealing at 500 °C. ............................................................................. 81
5-8 TEM cross-sectional images of Cu/WNxCy/Si stacks: [A) and B)] before annealing
and [C) and D)] after annealing at 500 °C. ..................................................................... 82
6-1 XRD spectra for films deposited on Si(100) at various temperatures: A) 300 °C, B)
700 °C, C) between 300 and 700 °C, and D) standard powder diffraction pattern for
β-W2N and β-WC1-x. ...................................................................................................... 90
11
6-2 Change in A) lattice parameter and B) average grain size with deposition
temperature for polycrystalline films deposited from 2. The estimates are based on
position and shape of diffraction peaks. ......................................................................... 91
6-3 Variation of W, N, C, and O content in the films deposited from 2. Data are from
XPS measurements after 10 min Ar+ ion sputter. ........................................................... 91
6-4 Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature. Data are from XPS after 10 min Ar+ ion sputter. ...................... 92
6-5 SEM images of films grown on Si(100) substrate: A) cross-sectional view of film
grown at 300 °C; B) cross-sectional view of film grown at 700 °C; C) surface
morphology of film grown at 300 °C; D) surface morphology of film grown at
700 °C. .......................................................................................................................... 93
6-6 Change in growth rate with deposition temperature for films deposited from 2.
Thickness measured by cross-sectional SEM. ................................................................ 94
6-7 Change in film resistivity (four-point probe) with deposition temperature for films
deposited from 2. ........................................................................................................... 94
6-8 Cross-sectional TEM images of Cu/WNxCy/Si stacks: [A) and B)] before annealing
and [C) and D)] after annealing at 500 °C. ..................................................................... 95
6-9 EDS depth profile of Cu/WNxCy/Si stacks annealed at 500 °C. ...................................... 96
6-10 The performance of diffusion barrier by XRD measurement for Cu/WNxCy/Si stacks
before and after annealing at 500 °C. ............................................................................. 96
7-1 XRD spectra for films deposited on Si(100) with NH3: A) 300 °C, B) 700 °C, C)
change in XRD spectra, and D) standard diffraction plots for β-W2N and β-WC1-x. ..... 105
7-2 XPS spectra for films deposited on Si(100) with NH3. Note that Cl peaks are evident
as a function of growth temperature. ............................................................................ 106
7-3 Comparison of W, N, C, and O content in the films deposited in the presence and
absence of NH3. Data are measured by XPS after 10 min Ar+ ion sputter. ................... 107
7-4 Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature. Data are from XPS after 10 min Ar+ ion sputter. .................... 108
7-5 Surface morphology of films grown on Si(100) substrate: A) film grown at 300 °C
without NH3; B) film grown at 700 °C without NH3; C) film grown at 300 °C with
NH3; D) film grown at 700 °C with NH3. ..................................................................... 109
7-6 SEM images of films grown on Si(100) substrate: A) cross-sectional view of film
grown at 300 °C; B) cross-sectional view of film grown at 700 °C. ............................. 110
12
7-7 Change in growth rate with deposition temperature for the films deposited in the
presence and absence of NH3. Thickness was measured by cross-sectional SEM. ....... 110
7-8 Film resistivity as a function of deposition temperature for the films deposited in the
presence and absence of NH3. ...................................................................................... 111
7-9 A) TEM image and B) EDS depth profile of a Cu/WNxCy/Si stack annealed at
500 °C for 30 min. ....................................................................................................... 111
7-10 Change in XRD patterns with annealing temperature for Cu/WNxCy/Si stacks. ............ 112
7-11 Change in sheet resistance with annealing temperature for Cu/WNxCy/Si stacks.
Data are measured by four-point probe. ....................................................................... 112
7-12 Cross-sectional TEM images of Cu/WNxCy/Si stacks: A) as-grown and B) after
annealing at 700 °C. .................................................................................................... 113
7-13 Cross-sectional SEM images of Cu/WNxCy/Si stacks: A) as-grown and B) after
annealing at 700 °C. .................................................................................................... 113
8-1 XRD spectra for films deposited on Si(100) with NH3: A) 300 °C; B) 400 °C; C)
change in XRD spectra; D) standard powder diffraction pattern for β-W2N and β-
WC1-x. ......................................................................................................................... 120
8-2 XPS spectra for films deposited on Si(100) with NH3. No Cl peaks detected. ............. 121
8-3 Variation in the chemical composition of W, N, C, and O contents in the films with
deposition temperature. Data are measured by XPS after 10 min Ar+ ion sputter. ........ 121
8-4 Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature. Data are from XPS after 10 min Ar+ ion sputter. .................... 122
8-5 Surface morphology of films grown on Si(100) substrate: A) film grown at 300 °C
with NH3; B) film grown at 700 °C with NH3. ............................................................. 123
8-6 SEM images of films grown on Si(100) substrate: A) cross-sectional view of film
grown at 300 °C; B) cross-sectional view of film grown at 700 °C. ............................. 123
8-7 Change in growth rate with deposition temperature for the films deposited from 3.
Thickness was measured by cross-sectional SEM. ....................................................... 124
8-8 Change in film resistivity (four-point probe) with deposition temperature for the
films deposited from 3. ................................................................................................ 124
9-1 Schematic photographs of A) CVD reactor system that is interfaced to the Raman
spectrometry ; B) nebulizer system; C) the impinging jet probe reactor. ....................... 130
9-2 Mesh design of CVD reactor using GAMBIT™. ......................................................... 131
13
9-3 Color filled contours of static temperature (K) and contour line of static temperature
(K) in the vicinity of the heater. ................................................................................... 132
9-4 Contours of velocity magnitude (m/s) and velocity vector colored by velocity
magnitude (m/s) in the vicinity of the heater. ............................................................... 133
9-5 Contours of velocity magnitude (m/s) and volume fraction of solvent phase in
multiphase flow model................................................................................................. 134
14
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
CHEMICAL VAPOR DEPOSITION OF TUNGSTEN-BASED DIFFUSION BARRIER THIN
FILMS FOR COPPER METALLIZATION
By
Dojun Kim
December 2009
Chair: Timothy J. Anderson
Major: Chemical Engineering
The ternary material WNxCy was investigated for Cu diffusion barrier application. Thin
films were deposited from tungsten diorganohydrazido(2-) complexes Cl4(CH3CN)W(NNR2) (1:
R2=-(CH2)5-; 2: R2=Ph2; 3: R2=Me2) using metal-organic aerosol-assisted CVD. The films
deposited from these novel precursors were characterized for their composition, bonding state,
structure, resistivity, and barrier quality.
WNxCy films from 1, 2 and 3 were successfully deposited in the absence and the presence
of NH3 in H2 carrier in the temperature range 300 to 700 °C. All WNxCy films contained W, N,
C, and a small amount of O as determined by XPS. The Cl content of the film was below the
XPS detection limit (~ 1 at. %). The chemical composition of films deposited with 1 in H2/NH3
exhibited increased N levels and decreased C levels over the entire temperature range of this
study as compared with to films deposited 1 in H2. As determined by XPS, W is primarily
bonded to N and C for films deposited at 400 C, but at lower deposition temperature the binding
energy of the W-O bond becomes more evident. The films deposited at 400 °C were X-ray
amorphous and Cu/WNxCy/Si stacks annealed under N2 at 500 °C for 30 min maintained the
integrity of both the Cu/WNxCy and WNxCy/Si interfaces.
15
Comparison of films deposited from 2 with H2 only and H2/NH3 shows that the best films,
in terms of composition, resistivity, surface roughness, and microstructure, are deposited using
H2/NH3 carrier. The microstructure of films deposited with NH3 was X-ray amorphous below
450 °C. XPS measurements revealed that W is primarily bonded to N and C for films deposited
between 300 and 700 °C. An Arrhenius plot of growth rate was consistent with surface reaction
limited growth and the activation energy was lower for growth in the presence of NH3. It was
observed that the surface roughness improved with added NH3. Samples annealed at higher
temperature showed evidence of failure only when annealed at 700 °C. These results support the
conclusion that WNxCy thin film deposited from 2 is a viable Cu diffusion barrier material.
As anticipated, the film N content was higher for films deposited from 3 with added NH3
as compared to those deposited from 1 and 2. The films deposited with NH3 in H2 carrier at
400 °C had the highest N content of all films (27 at. %). An amorphous film microstructure was
observed for films deposited below 500 °C. The apparent activation energy for the film growth
in the kinetically controlled growth regime was 0.31 eV. The observation of AFM monograph
indicates that the surface roughness improved with added NH3.
Film growth of WNxCy by metal-organic aerosol-assisted CVD using 1, 2, and 3 highlights
the importance of precursor selection, co-reactant selection (H2 only, H2/NH3, N2 only, and
N2/NH3), and operating parameters (deposition temperature, pressure, and flow rate) on film
properties and barrier performance. Preliminary material characterization and diffusion barrier
testing reveals that films deposited using 2 with NH3 in H2 carrier is most promising for diffusion
barrier applications.
16
CHAPTER 1
INTRODUCTION
Statement of Problems
The continuous challenges in microelectronic integrated circuits are increasing speed and
improving reliability. The RC time delay hinders further increasing of speed in integrated circuits
(Figure 1-1). Device dimensions continue to decrease on integrated circuits, and the industry is
transitioning from Al-based interconnects to Cu-based interconnects is required for multilevel
metallization to minimize the RC time delay. Cu-based interconnects show greater resistance
toward electromigration and 40% lower electrical resistivity (ρCu ~ 1.67 μΩ-cm and ρAl ~ 2.65
μΩ-cm), as compared to Al-based interconnects [1-3]. As a result of the high diffusivity of Cu in
Si and SiO2 (DCu ~ 2 × 10-5
cm2/s at 500 °C), high priority has been placed on developing Cu
diffusion barriers (Figure 1-2) [4]. The presence of Cu in Si and SiO2 results in serious
degradation of device performance associated with contact resistance, barrier height, p-n
junctions, contact layers, and electrical connections [5]. Therefore, an effective Cu diffusion
barrier is required to block Cu transport and intermixing with adjacent dielectric materials for Cu
interconnect technology.
To provide excellent diffusion barrier performance characteristics, deposited films need to
possess certain properties such as good step coverage, low electrical resistivity, low deposition
temperature and amorphous microstructure. Various transition metal nitrides have been
investigated as Cu diffusion barriers including TiN, TiSixNy, TaN, TaSixNy, WNx, WSixNy, and
WBxNy [6-12]. Ta/TaN bilayers deposited by physical vapor deposition (PVD) are the currently
utilized Cu diffusion barriers in semiconductor device technology. However, limitations of PVD
due to the directional nature of deposition cause problems upon scaling down the barrier
thickness. Chemical vapor deposition (CVD) and atomic layer deposition (ALD) of TaNx thin
17
films also have difficulties in depositing conductive TaN due to the preferential formation of
Ta3N5 ( ~ 2 × 108 μΩ-cm) [13, 14]. The binary phase material, tungsten nitride (WNx), is a
promising candidate for replacing the prevailing diffusion barrier of Ta/TaN bilayer structure [15,
16]. WNx film shows good thermal stability with Cu, acceptably low resistivity when deposited
by CVD, and reasonable chemical mechanical planarization (CMP) processing [10]. The ternary
phase material, tungsten carbonitride (WNxCy), is also a promising candidate for diffusion barrier
applications. WNxCy film has low electrical resistivity, good adhesion to Cu, good resistance to
diffusion of Cu, and acceptable film growth on SiO2. The efficacy of WNxCy film as a diffusion
barrier has been demonstrated for films grown by both CVD and ALD [13, 17-19]. It has been
shown that X-ray amorphous ternary phase materials such as TiSixNy, TaSixNy, WSixNy, and
WBxNy have better performance as Cu diffusion barriers than binary phase materials due to
higher recrystallization temperature and thus lack of grain boundaries, which can serve as Cu
diffusion pathways [12, 20-22]. Aerosol-assisted CVD (AACVD) is a useful technique for
growing films of refractory metal nitrides because aerosol-assisted delivery permits use of low
volatility precursors and thermally sensitive precursors that decompose before sublimation can
be used [23]. Recently, we reported the synthesis of the diorganohydrazido(2-) tungsten
complexes Cl4(CH3CN)W(NNR2) (1: R2 = -(CH2)5-; 2: R2 = Ph2; 3: R2 = Me2) and
Cl4(pyridine)W(NNR2) (4: R2 = Ph2) by reacting 1,1-diorganohydrazines with tungsten
hexachloride (WCl6) followed by treatment with acetonitrile (CH3CN) or pyridine (C5H5N) [24].
The diorganohydrazido(2-) tungsten complexes (1-3) were demonstrated to be single-source
precursors for the metal-organic CVD (MOCVD) of WNxCy thin films in the absence of NH3 in
H2 carrier. The effect of NH3 in H2 carrier was demonstrated on the properties of WNxCy thin
films deposited from 1, 2, and 3. The diffusion barrier testing was performed to investigate the
18
Cu diffusion barrier properties and the onset of failure process via formation of more resistive
copper silicide (CuxSi).
Figure 1-1. The device delay as a function of device generation. Adopted from M. T. Bohr,
“Interconnect scaling – the real limiter to high performance ULSI”, Proceedings of
IEEE International Electron Devices Meeting (1995) 241-242.
Figure 1-2. SEM cross-sectional images: A) Cu deposition without Cu diffusion barrier; B) Cu
deposition with Cu diffusion barrier.
B) A)
Cu3Si
x
19
CHAPTER 2
LITERATURE REVIEW
2.1 Diffusion Mechanism in Cu Metallization
The substitution of Cu has been a recent technological innovation for the standard Al-Cu
metal interconnects in order to reduce resistance and RC time delay in microelectronic integrated
circuits [2, 3]. The current Cu technology shows improved current-carrying capability by greater
resistance toward electromigration and no device contamination by Cu migration. The success of
the shift to Cu includes the development of an electroplating process for the Cu interconnects,
dual-damascene CMP, and an effective liner material for a Cu diffusion barrier and adhesion
promoter. It is required to establish a fundamental understanding of the predominant diffusion
mechanisms for atomic mobility and associated diffusion phenomena in order to identify an
effective liner for Cu technology. The placement of chemically different atoms in close
proximity causes atomic migration for the purpose of reducing the overall free energy and
establishing equilibrium. Typical reasons for atomic migration are the presence of concentration
differences, existence of a negative free energy of reaction, application of an electrical field,
availability of thermal energy, generation of a strain gradient, or a combination of some or all of
these factors.
Atomic migration could result in a diffusive flux. The net flow of atoms by diffusion is
described by Fick‟s law.
)(dx
dCDJ (2.1)
where C is the atomic concentration, J is the atomic flux per unit area per second, and x is
distance. The temperature dependence of the diffusion coefficient D takes the form of an
Arrhenius relationship.
20
)exp(0kT
QDD (2.2)
where D0 is a constant, Q is the activation energy for diffusion, k is Boltzmann‟s constant, and T
is the temperature in degrees Kelvin.
There are three typical failure mechanisms in the Cu/liner material system. First, Cu
diffuses along grain boundaries. Second, Cu (or substrate atoms) diffuses through bulk defects in
the liner (vacancies and dislocations). Third, loss of liner integrity results from a metallurgical or
chemical reaction with the Cu and/or substrate. Lattice diffusion rates are proportional to the
absolute melting temperature Tm.
mATD ~ (2.3)
where A is a proportionality constant that depends on a variety of factors, including lattice
structure and type of material. Diffusion along grain boundaries has the highest diffusion rates
(or largest A), which result from a large misfit between adjoining grains. Diffusion by
dislocations shows intermediate diffusion rates. Diffusion due to atom-vacancy exchange has
the lowest diffusion rate (or smallest A). This indicates that Cu barrier materials with higher
melting points could act as better Cu diffusion barriers. Also, the microstructure of Cu barrier
materials plays an important role in the resulting diffusion barrier performance. Film
microstructures in Fig. 2-1 can be categorized as single crystal, polycrystalline, nano-crystalline
(i.e. polycrystalline with grain size below ~ 5 nm), and amorphous. Single crystalline materials
are the ideal microstructure of Cu diffusion barriers. Lattice mismatchs with the underlying
substrate and thermal budget limitations make it difficult to deposit liners in single crystal
microstructure. Hence, amorphous phase Cu diffusion barriers are the most desirable for
diffusion barrier applications. There are three basic requirements for diffusion barrier materials.
First, a viable barrier material must not react with Cu or the underlying substrate under thermal,
21
mechanical, and electrical stress conditions. Second, the density of the diffusion barrier must be
as close to ideal as possible for the purpose of eliminating diffusion through voids, defects, or
loosely packed grain boundaries. Third, the microstructure of diffusion barrier must have no
grain boundary diffusion paths.
2.2 Ta/TaN Bilayer Structure as a Diffusion Barrier
The introduction of Cu interconnect technology results in the need for refractory metal
nitride films in modern semiconductor technology (Figure 2-2). Since Cu rapidly diffuses in Si,
a diffusion barrier should be employed between the metals and dielectrics to prevent Cu transport
and intermixing with adjacent dielectric materials. Even if Cu-interconnects show lower
electrical resistivity and greater resistance toward electromigration than Al-interconnects, the use
of Cu-interconnects requires conducting layers in the metallization structure which enhance the
Cu adhesion to dielectrics. Excellent adhesion to the underlying layer or interconnect material is
required to prevent reliability problem such as electromigration and gross delamination during
the CMP process. Additional requirements for Cu diffusion barrier include amorphous
microstructure, low electrical resistivity, high electromigration resistance, good step coverage,
low deposition temperature (≤ 400 °C), and minimal thickness [25].
Cu interconnect metallization has introduced new concepts in integration schemes: dual-
damascene structures, CMP process, and Cu electroplating. Figure 2-3 depicts the simplified
processing steps for the barrier film process that is used for the Cu dual-damascene structures.
The requirements are grouped into film properties and process compatibility. The film property
requirements includes ultra low thickness (< 100 Å ), low resistivity (< 500 μΩ-cm), low halide
residues (< 2 at. %), good step coverage (> 90 %), and reasonable process rate (30 Å /min). The
process compatibility requirements includes CMP compatible, low deposition temperature (≤
400 °C), good adhesion on the etch stopper, good adhesion on SiO2, and good adhesion on Cu.
22
These requirements must be met by any new barrier material to be used in the Cu dual-
damascene structure [26].
Ta/TaN bilayer structure has been used for diffusion barrier applications in dual-
damascene structures for current Cu interconnect metallization in the semiconductor industry.
Ta shows a high melting point (2669 °C) and good stablility with Cu. TaN shows high thermal,
mechanical, and chemical stability. TaN shows good adhesion to SiO2 and low-κ materials and
Ta shows the lack of Cu-Ta compound and better adhesion than Cu/TaN adhesion. Therefore,
Ta/TaN bilayer structure has been used for Cu interconnect technology. Ta/TaN liner has very
low in-plane electrical resistivity, since α-phase Ta deposited on TaN surface is spontaneously
formed with a resistivity in the range 15 to 60 μΩ-cm.
Although PVD TaN has been successful so far as a Cu diffusion barrier, due to the
downscaling of device dimensions in microelectronic integrated circuits, future Cu interconnects
require Cu diffusion barriers deposited by CVD or ALD. The drawbacks of the present Ta/TaN
bilayer structure are both in process and material. PVD is „a line of sight process‟, indicating the
limitations of PVD due to the directional nature of deposition. This cause problems upon scaling
down the barrier thickness. The application of PVD techniques is limited by concerns over their
ability to provide good conformality in sub-100 nm device technology. Also, the underlying
device layer on the substrate can be damaged due to high energy particles. Many researchers
have attempted to grow TaN by CVD and ALD. TaN has many polymorphs with different film
properties depending on N content: solid solution α-phase Ta, hexagonal Ta2N, hexagonal TaN,
cubic TaN, hexagonal Ta5N6, tetragonal Ta4N5, and orthorhombic Ta3N5 [27]. The growth of the
insulating Ta3N5 phase during growth of TaN by CVD and ALD results in an increase in
electrical resistivity of films [14].
23
2.3 Chemical Vapor Deposition of Tungsten-Based Diffusion Barrier
2.3.1 Tungsten Nitride as a Diffusion Barrier
In previous works, WNx has been deposited using PVD techniques such as reactive
sputtering of a W target under N2 atmosphere. However, this technique results in poor step
coverage, a major disadvantage when applied to device structures with high aspect ratio features
[28-31]. Table 2-1 shows the halide and metal-organic precursors used for film growth of WNx
by CVD. WF6 precursor has been used with NH3 coreactant to deposit WNx thin films for
application as a barrier and glue layer for advanced metallization [32, 33]. XRD data shows a
consistent (111) orientation of W2N cubic structure at 450, 550, 650, and 700 °C. The resistivity
obtained for these films ranged from about 900 to 2800 μΩ cm. The F content was detected
from a maximum 0.9 at. % at 450 °C to less than 0.1 at. % at 625 °C [32]. Addition of H2 gas to
the mixture of WF6 and NH3 facilitates the reaction of binary mixtures by breaking a W-F bond
of WF6 or N-H bond of NH3 causing a decrease in activation energy for the reaction. XRD data
shows only W2N was obtained without any diffraction line indicating WN. The N 1s spectrum
from XPS shows two peaks at 397 and 400 eV. The latter peak is ascribed to a N atom or
molecule present in interstitial sites of W2N. The release of N is due to desorption of N2 gas
when it is heated to a high temperature [33]. Another inorganic precursor, WCl6, has been used
for film growth of WNx with a mixture of NH3, H2, and Ar at temperatures of 500 to 900 °C at
0.1 to 10 Torr. The temperature dependence of the Gibbs free energy shows a preferential
reaction with WCl6 in the temperature range of this study. XPS analysis shows three W 4f7/2
peaks at 31.5, 33.6, and 37.2 eV for films at 500 °C. Although no oxide peaks were observed by
XRD, the surface of the film was contaminated with a small amount of oxide [34]. The halide
precursors such as WF6 and WCl6 required high deposition temperatures (> 450°C) and
incorporated the halogen impurities during film growth. W(CO)6 has been explored for film
24
growth of WNx with low impurities at low temperatures compared to other tungsten-based
precursors because the binding energy of W-CO is low [16, 35, 36]. WNx was deposited using
W(CO)6 and NH3 in the temperature range 250 to 500 °C. The film resistivity varied from 590
to 950 μΩ cm. The growth rate varied from 3 to 1930 Å /min. Below 450 °C, the growth regime
shows an Arrhenius type dependence on the deposition temperature. The film growth was
kinetically controlled with the activation energy of 1.00 eV. Sheet resistance measurement and
XRD analysis showed that the diffusion barrier (15 nm thick) blocked the diffusion of Cu up to
600 °C for 1 h annealing [16]. Both results showed that W2N film prevented diffusion of Cu up
to 600 °C, and started to fail at 620 °C, while no barrier and the CVD-W samples failed at 100 to
150 °C and 525 to550 °C. Barrier failure at 620 °C is thought to be due to the diffusion of Cu via
undesired grain boundaries [35]. H2 was premixed with precursor vapor at the reactor inlet and
NH3 as the N source was introduced directly into the chamber through a separate feedthrough.
Film deposited below 275 °C was amorphous, while those deposited between 275 and 350 °C
were polycrystalline. Resistivity as low as 123 μΩ cm was obtained with corresponding step
coverage better than 90 % in a nominal 0.25 μm trench structure with aspect ratio of 4:1 [36].
(tBuN)2W(NH
tBu)2 has been used for film growth of WNx as the single-source precursor [37, 38].
Polycrystalline WNx thin films were grown by low pressure MOCVD using (tBuN)2W(NH
tBu)2
in Ar or H2 carrier. XRD studies showed that the films have cubic structures with the lattice
parameter of 4.154 to 4.180 Å . XPS showed the binding energies of the W 4f7/2 and N 1s were
33.0 and 397.3 eV, respectively. The secondary ion mass spectrometry (SIMS) compositional
depth profiling indicated C and O levels were low in the films. Possible reaction pathways were
suggested by detecting isobutylene, acetonitrile, hydrogen cyanide, and ammonia using gas
chromatography-mass spectroscopy (GC-MS) and nuclear magnetic resonance (NMR) [37].
25
Annealing to 700 K caused the loss of N content from the bulk deposited WNx layer as N2 [38].
Plasma-enhanced CVD (PECVD) using WF6 has been used for film growth of WNx. N2 was
used as the N source and H2 was used to remove F from halide precursor. F, O, and C present in
the films were below 1% based on XPS. W2N films have good adhesion to PVD Cu, CVD W, Si,
SiO2, and Si3N4, as observed by tape peel tests. Despite higher step coverage for films deposited
at 300 °C, XPS indicated F impurity. Rapid thermal annealing (RTA) is used to treat the
deposited films to reduce the F impurity level [39]. W2N films were deposited at a wafer
temperature of 350 °C on Si, SiO2, and Ta2O5, with and without an electron cyclotron resonance
plasma formed SiO2 (ECR-SiO2) top layer. The resistivity of W2N films is 190 to 240 μΩ cm.
XRD patterns are X-ray amorphous [40]. The resistivity for stoichiometric W2N, W rich W2N (x
> 1.0), and N rich W2N (x < 1.0) is different. The resistivity of W rich W2N is 145 μΩ cm and
that of N rich W2N 3000 to 5000 μΩ cm. The decrease of N levels in W2N due to N2 desorption
is confirmed by AES [41]. The resistivity of as-deposited films is 95 to 100 μΩ cm. In order to
improve the adhesion strength of CVD W films, W2N glue layer is interposed between W and Si.
The number of vacancies at N lattice sites is reduced because N atoms occupy interstitial
positions in the W lattice. The more adhesive contact is due to N interstitials due to the
modification of the structural properties such as porosity and vacancies in the W2N [42].
Diffusion barrier test results from SEM and XRD using Cu/WNx/SiCOH/Si stacks showed that
W2N films were stable up to 500 °C. Above 600 °C, WO3 nanorods were grown from the
sample surface due to the residual O in the films [15].
2.3.2 Tungsten Carbonitride as a Diffusion Barrier
Table 2-2 shows that the metal-organic precursors that were used for film growth of
WNxCy by CVD. [W(μ-NtBu)Cl2(H2N
tBu)]2, [W(N
tBu)Cl2(TMEDA)] (TMEDA = N, N, N′, N′ -
26
tetramethylethylenediamine), [W(NtBu)Cl2(py)2] (py = pyridine), and [W(N
tBu)2Cl(N{SiMe3}2)]
have been used to deposit WNxCy in N2 carrier at the deposition temperature of 550 °C. Those
compounds can be used for film growth as single-source precursors or dual-source in the
presence of NH3. In all cases the Cl levels present in the films were less than 1 at. %. Film
growth using NH3 shows lower O level and no change in C content of the resulting films. XRD
pattern of all the films indicated the formation of β-WNxCy. SEM surface images of films
suggest an island growth mechanism. The films were uniform, adhesive, abrasion resistant,
conformal and hard, being resistant to scratching with a steel scalpel [43]. WH2(iPrCp)2 and
WH2(EtCp)2 have been used for film growth of WNxCy in NH3/H2/N2 carrier. Film growth was
carried out on SiO2 substrates using N2 carrier gas at temperature range 350 to 400 °C. NH3
(99.96 %) and H2 (99.9999 %) were used as reactant. The W 4f7/2 and W 4f5/2 peaks at 31.6 and
33.8 eV are well matched with the WCx phase. The C 1s peak located at 283.2 eV is well
matched with the carbidic form. XRR and XRD analyses show no peaks indicating
crystallization. The addition of NH3 causes the O incorporation to decrease significantly. The
lowest value of resistivity was 565 μΩ cm when no coreactant was used at 350 °C. This is
correlated with the decrease of the C level present in the films. The addition of NH3 causes an
increase of the film resistivity because mobility is reduced by the scattering effect of
incorporated N atoms [44]. The tungsten isopropylimido complex Cl4(CH3CN)W(NiPr) has
been used for film growth of WNxCy as a single-source precursor [45, 46]. The precursor
structure was chosen so that the W-N multiple bond of the precursor would survive while the
ancillary ligands and the isopropyl imido substituent dissociated under CVD. Film
microstructure at a temperature below 500 °C was X-ray amorphous, with the minimum value of
film resistivity (750 μΩ cm) and sheet resistance (47 Ω/□) of this study occurring for CVD at
27
450 °C. Film growth rate varied from 10 to 27 Å /min within a temperature range of 450 to
700 °C. The apparent activation energy for film growth in the kinetically controlled regime was
0.84 eV. C levels increased from 12 to 49 at. % in the temperature range 450 to 700 °C.
Fragmentation of ligands and solvent would leave C containing moieties at the film surface,
indicating C incorporation into the WNx film [45]. WNxCy thin films were deposited using
solutions of Cl4(CH3CN)W(NiPr) in 1,2-dichlorobenzene (1,2-DCB). The results show the
solvent affected deposition of C into the films in comparison with the films deposited with
solutions of Cl4(CH3CN)W(NiPr) in benzonitrile (PhCN). The increased N levels for films from
PhCN solutions suggest that the nitrile (CN) group was a significant C source. The activation
energy for film growth from PhCN solutions weas 0.70 eV, while that from 1,2-DCB solutions
was 1.0 eV. This shift in activation energy upon changing the solvent is evidence for an
alternative C deposition process [46]. The tungsten phenylimido complex Cl4(PhCN)W(NPh)
has been used for film growth of WNxCy as a single-source precursor [47]. Film growth rates
varied from 2 to 21 Å /min in the temperature range 475 to 750 °C. The apparent activation
energy for film growth in the kinetically controlled regime was 1.41 eV. Film microstructure
was X-ray amorphous below 500 °C, with minimum film resistivity (225 μΩ cm) and sheet
resistance (75 Ω/□), observed for CVD at 475 °C. Films deposited from Cl4(CH3CN)W(NiPr)
exhibited higher growth rates and higher N level in the same temperature range. These different
results are due to the higher dissociation energy of the imido N-C bond in Cl4(PhCN)W(NPh).
Films from Cl4(CH3CN)W(NiPr) are superior to these from Cl4(PhCN)W(NPh) for diffusion
barrier applications due to lower amorphous deposition temperature, lower sheet resistance, and
higher N level [45, 47]. The tungsten isorpopylimido complex Cl4(CH3CN)W(NiPr) has been
used for film growth of WNxCy in NH3/H2 carrier. AES results initiated that films deposited
28
with NH3 had higher N levels for low deposition temperature (450 – 550 °C), along with
decreased C and O levels as compared with films deposited without NH3. Film microstructure
was X-ray amorphous for film deposited with NH3, in contrast to polycrystalline for phase
present in the films deposited without NH3 at 500 °C. An increase in N level in the amorphous
films would increase film resistivity because the film resistivity is higher for WNx phase relative
to WCx phase and replacement of C by additional N causes electron scattering. Film growth in
the presence of NH3 was mass transfer controlled across the entire temperature range (450 –
700 °C), while film growth in the absence of NH3 had a kinetic to mass transfer control transition
point near 600 °C [48]. A mixture of the tungsten allylimido complex Cl4(CH3CN)W(NC3H5)
and Cl4(PhCN)W(NC3H5) has been used for film growth of WNxCy in the presence and absence
of NH3 [18, 19]. Cl4(PhCN)W(NC3H5) was not isolated but was produced in situ by the
substitution of the acetonitrile ligand of Cl4(CH3CN)W(NC3H5) with PhCN. The rapid rate of
exchange of nitrile ligands in Cl4(CH3CN)W(NC3H5) ensures that the precursor is completely
converted to Cl4(PhCN)W(NC3H5) before film growth starts. Films deposited from a mixture
show X-ray amorphous phase below 550 °C. Film growth rate varied from 5 to 10 Å /min in the
temperature range 450 to 650 °C, and the apparent activation energy for film growth was 0.15 eV.
The values of activation energy for film growth using Cl4(RCN)W(NR′) [R = CH3, Ph, and R′ =
Ph, iPr, allyl] against the N-C plotted against the bond strengths for the amines R′NH2 is linear.
The linear relationship between activation energy for film growth using Cl4(RCN)W(NR′) [R =
CH3, Ph, and R′ = Ph, iPr, allyl] and the N-C bond dissociation energy for the amines R′NH2
suggests that cleavage of the N-C bond is the rate-determining step in film growth. The strength
of the N-C imido bond has an effect on the amount of N incorporated in the film [18]. Films
deposited at 450 °C with NH3 as a coreactant showed 23 at. % N level, which is higher than film
29
growth without added NH3 (4 at. % in N level). O incorporation remained below 6 at. % in the
temperature range 450 to 750 °C. The films deposited below 500 °C were X-ray amorphous and
the X-ray diffraction patterns suggest that either the mixture of β-W2N and β-WC1-x or the solid
solution β-WNxCy exist in the films. The presence of Cl in the precursor raises the concern of
Cl-free films. XPS spectra show no Cl peaks were observed either the Cl 2s (270 eV) or Cl 2p3/2
(199 eV), confirming that Cl level in the films was lower than the XPS detection limit (~ 1 at. %).
An apparent activation energy for films with added NH3 is 0.34 eV, as compared with the value
of 0.15 eV for films without NH3. The film resistivity for films deposited with NH3 exhibited
higher film resistivity, with the lowest film resistivity of 1700 μΩ cm observed for films
deposited at 550 °C [19].
2.4 Atomic Layer Deposition of Tungsten-Based Diffusion Barrier
2.4.1 Tungsten Nitride as a Diffusion Barrier
Table 2-3 shows that the halide and metal-organic precursors that were used for film
growth of WNx by ALD. The WF6 precursor has been used with NH3 coreactant to deposit WNx
thin films for application as a Cu diffusion barrier layer for advanced metallization [26, 49-52].
WNx on SiO2 was deposited at 350 °C. The growth rate was fairly high, saturating at a level of
0.42 Å /cycle. Even though the F impurity was as low as 2.4 at. % in the film, the value of film
resistivity is 4500 μΩ cm. Introduction of a third precursor between WF6 and NH3 pulses caused
improved reduction of W and reduced the formation of HF in order to reduce the resistivity of
WN and avoid Cu pitting [26]. WNx was deposited on Si and tetraethylothosilicate (TEOS)/Si
substrates in the temperature range 200 to 400 °C, synchronizing the NH3 plasma (NH*, NH
+,
NH2+, NH3
+, and H
+), instead of NH3 gas at the NH3 exposure cycles during ALD. The
conventional ALD shows that a 22nm-thick W layer is deposited and a 3 nm-thick WNx layer
appears on the top of this W layer during the 100 cycles exposing WF6 and NH3. AES depth
30
profiles for films deposited by pulse-plasma-enhanced ALD show a uniformly distributed N
concentration in the WNx films on Si and non-Si surfaces. WF6 either reacts with Si quickly due
to the catalytic reaction of Si, forming a thick W layer instead of WNx, or does not adhere to the
non-Si surfaces. High-resolution transmission electronic microscopy (HRTEM) reveals that
WNx (22 nm thick) in the Cu/WNx/Si stack prevents Cu diffusion during the annealing process at
700 °C for 30 min [49]. The deposition rate was about 3 Å /cycle at 350 °C. There are two
different growth regimes: one is the incubation regime and another is the linear and self-limiting
growth regime. Rutherford backscattering spectroscopy (RBS) revealed that WNx (22 nm thick)
in the Cu/WNx/Si stack prevents Cu diffusion during the annealing process at 600 °C for 30 min
[50]. Alternating exposures of NH3 (A) and WF6 (B) in an AB reaction sequence were used to
deposit the WNx at the substrate temperature between 323 and 523 °C. Transmission Fourier
transform infrared (FTIR) spectroscopy studies indicated that NH3 and WF6 surface reactions
were complete and self-limiting at deposition temperature over 323 °C. AFM images exhibit a
root-mean-square (rms) roughness of ± 0.61 nm. The rms roughness of initial SiO2 on Si(100)
was ± 0.25 nm. The XPS spectra indicate that the surface of WNx exhibited characteristic
signals for W, C, N, F, and O. The XPS depth profiling reveals that the WNx films had a W to N
ratio of ~ 3:1. The films also contained 5 at. % C and 3.6 at. % O. Glancing angle XRD results
indicate the films consisted of W2N crystallites with a diameter 11 nm and a (111) texture [51].
Successive exposure to WF6 and Si2H6 (or NH3) in an ABAB… reaction sequence produced W
(or W2N) deposition at substrate temperature 152 - 423 °C (or 323 - 523 °C). Between the WF6
and Si2H6 reactant exposures, the deposition chamber was purged with N2 for several minutes.
Si2H6 serves only a sacrificial role to remove surface species without incorporation into the film
[52]. (tBuN)2(Me2N)2W precursor has been used with NH3 coreactant for film growth of WNx
31
[53-55]. WNx barrier films were deposited by ALD using (tBuN)2(Me2N)2W and NH3 in the
temperature range 250 to 350 °C. Film microstructure was X-ray amorphous as deposited and
100% step coverage was obtained inside holes with aspect ratio greater than 40:1. RBS showed
that O was not detected and C was less than the detectable limit (< 0.5 at. %). WNx film (1.5 nm
thick) proved to be good barriers to the Cu diffusion for temperature up to 600 °C. Numerous
crystals of Cu3Si were observed due to complete breakdown of the barrier for a sample annealed
at 650 °C. RBS and XPS confirmed the loss of N in the annealed film at temperatures greater
than 725 °C, indicating the WNx was converted to pure polycrystalline W [53]. Films deposited
above 350 °C contained C in addition to W and N and their step coverage is not as good as that
for films deposited within the range 250 to 350 °C. The films deposited at 400 °C were more
conductive, 420 μΩ cm. No films were deposited at deposition temperature below 250 °C. ALD
of Cu on the WNx could not be removed by adhesive tape applied to the Cu [54]. ALD has been
used to seal porous low-κ material with silica (4 nm thick), and to add a WNx diffusion barrier
(1.0 nm thick), a Co adhesion layer (1.0 nm thick), and a Cu seed layer (10 nm thick). Tape pull
tests showed the Cu/Co/WNx/silica/low-κ /Si stack has good adhesion. Samples annealed at
400 °C for 30 min showed no agglomeration of Cu observed by SEM and no diffusion of Cu
detected by RBS [55].
2.4.2 Tungsten Carbonitride as a Diffusion Barrier
Table 2-4 shows the halide and metal-organic precursors that were used for film growth of
WNxCy by ALD. The properties of WNxCy films deposited by ALD using WF6, NH3, and TEB
as a source gases were characterized as a diffusion barrier for Cu metallization [13, 17, 56-63].
ALD WNxCy was deposited in the temperature range 275 to 325 °C by supplying WF6, NH3, and
triethylboron [B(C2H5) or TEB] in cyclic pulses. The growth rate was 0.8 Å /cycle; the WNxCy
was conductive (300 – 400 μΩ cm) and dense (15.4 g/cm3). XPS spectra indicate that C in ALD
32
WNxCy is in the WCx phase, which is more conductive than the WNx phase [13]. The films
deposited at 313 °C show resistivities of about 350 μΩ cm with densities of 15.4 g/cm3. The
chemical composition measured by RBS shows W, C, and N of 48, 32, and 20 at. %, respectively.
TEM analysis shows that the as-grown film was composed of a face-centered-cubic (fcc) phase
with a lattice parameter similar to both β-WN2 and β-WC1-x with an equiaxed microstructure.
Diffusion barrier test results show that ALD-WNxCy films (12 nm thick) deposited between Cu
and Si failed after annealing at 700 °C for 30 min. The superior diffusion barrier performance is
the consequence of both the formation of films with equiaxed microstructure and high density
[17, 60, 61]. Film morphology by AFM reveals island growth and fractal behavior of individual
ALD WNxCy on the methyl-terminated self-assembled monolayers (SAMs) for film deposited at
300 °C. Initially, the film grows by deposition of WNxCy on the substrate defect sites. This
deposition causes increased surface area and as a result, film roughness increases. This situation
continues until the film coalesces where the surface area is reduced and accordingly film
roughness decreases. The film area and roughness become constant when the substrate is
completely covered. This growth of ALD WNxCy is enhanced on N containing surfaces such as
N2 plasma-treated SILK polymer films because of good binding states for TEB [56]. TEM
analysis reveals the island growth of individual ALD WNxCy nanocrystals on the PECVD SiO2
during early stages of film growth. The capacitance-voltage (C-V) measurements after bias-
temperature stressing (BTS) reveal that WNxCy thin film (5.2 nm thick) acts a good diffusion
barrier for Cu migration [57]. WNxCy growth on SiC is similar to that on PECVD SiO2. This is
due to the presence of a C-rich layer from TEB precursor decomposition [58]. WNxCy was
deposited at 300 °C in a process sequence using WF6, TEB, and NH3 as precursors. The bulk
resistivity of WNxCy has low resisitivity about 300 – 400 μΩ cm. XPS results show a ratio of
33
W:N:O:C of 60:20:10:10 at. % throughout a very homogeneous layer. The rms roughness of 0.47
nm was determined for a WNxCy layer with AFM [59]. WNxCy was deposited by introducing
TEB as a reducing agent for W. WNxCy shows excellent film properties: good compatibility
with the Cu metal, strong adhesion on the Cu surface, and no pitting on the Cu surface. The
growth rate is 0.08 nm/cycle and it remains constant in the temperature range 300 to 350 °C
(ALD window). XPS spectra indicate the chemical composition of W:N:C is 55:15:30. Boron
(B) residues were below the XPS detection limit (0.5 at. %). F levels were below 2 at. % for
films deposited from 225 to 400 °C. XRD results show that the crystalline phase is β-WC1-x.
The resistivity is as low as 210 μΩ cm, indicating that C is bound in the WCx phase [62]. TEM
images show that the step coverage of WNxCy barrier film is nearly 100 % in one via in a via
chain. AES analysis indicates that the chemical composition of W, C, and N is 57, 30, and 13
at. %. The resistivity was 600 to 900 μΩ cm [63]. W2(NMe2)6 precursor has been used with
NH3 coreactant for film growth of WNxCy thin films between 150 and 250 °C. NH3 was used as
a N source and Ar was used as the carrier and purge gas. At 180 °C, surface-limited growth was
achieved with W2(NMe2)6 pulse lengths over 2.0 s. Shorter pulse of W precursor results in sub-
saturative growth and lower growth rate. The ALD window was detected at the deposition
temperature between 180 and 210 °C. XPS spectra indicate that W 4f7/2 binding energy was 31.5
eV, which is well matched with the binding energy of the WCx and WNx phases. The binding
energy of C 1s at 282 eV and N 1s at 397.6 eV are consistent with C in carbides and N in nitrides.
The binding energy of O 1s was 530. eV. Films deposited at 180 °C exhibited a resistivity value
of 810 ± 50 μΩ cm. The resistivity of WNx films is sensitive to the W to N ratio. Further
exposure of the same film to ambient atmosphere (an additional 30 days) caused an increase in
film resistivity values over 10000 μΩ cm. XRD results for film deposited at 180 °C indicates X-
34
ray amorphous microstructure. AFM analysis shows that the rms roughness (2 μm by 2 μm area)
was 0.9, 0.8, and 0.7 for films deposited at 150, 180, and 210 °C, respectively [64]. (η5-
C5H5)W(CO)2NO precursor has been used with NH3 coreactant to deposit WNxCy thin films by
PEALD [65, 66].
35
Table 2-1. Precursors used for film growth of WNx by CVD
Technique Precursor Coreactant Reference
CVD WF6 NH3 [32]
CVD WF6 NH3 + H2 + Ar [33]
CVD WCl6 NH3 + H2 + Ar [34]
CVD W(CO)6 NH3 + Ar [16, 35]
CVD W(CO)6 NH3 + H2 [36]
CVD (tBuN)2W(NH
tBu)2 H2 or Ar [37]
CVD (tBuN)2W(NH
tBu)2 [38]
PECVD WF6 NH3 + H2 +N2 [39]
PECVD WF6 NH3 + H2 [40-42]
PECVD W(CO)6 NH3 [15]
Table 2-2. Precursors used for film growth of WNxCy by CVD
Technique Precursor Coreactant Reference
CVD [W(μ-NtBu) (N
tBu)Cl2(H2N
tBu)]2 NH3 + N2 [43]
CVD [W(NtBu)2Cl2(TMEDA)] NH3 + N2 [43]
CVD [W(NtBu)2Cl2(py)2] NH3 + N2 [43]
CVD [W(NtBu)2Cl(N{SiMe3}2)] NH3 + N2 [43]
CVD WH2(iPrCp)2 NH3 + H2 + N2 [44]
CVD WH2(EtCp)2 NH3 + H2 + N2 [44]
CVD Cl4(CH3CN)W(NiPr) H2 [45, 46]
CVD Cl4(PhCN)W(NPh) H2 [47]
CVD Cl4(CH3CN)W(NiPr) NH3 + H2 [48]
CVD Cl4(CH3CN)W(NC3H5) H2 [18]
CVD Cl4(PhCN)W(NC3H5) H2 [18]
CVD Cl4(CH3CN)W(NC3H5) NH3 + H2 [19]
CVD Cl4(PhCN)W(NC3H5) NH3 + H2 [19]
36
Table 2-3. Precursors used for film growth of WNx by ALD
Technique Precursor Coreactant Reference
ALD WF6 NH3 [26, 49-52]
ALD (tBuN)2(Me2N)2W NH3 [53-55]
Table 2-4. Precursors used for film growth of WNxCy by ALD
Technique Precursor Coreactant Reference
ALD WF6 NH3 + (C2H5)3B [13, 17, 56-62]
ALD WF6 NH3 [63]
ALD W2(NMe2)6 NH3 + Ar [64]
PEALD (η5-C5H5)W(CO)2NO NH3 [65]
PEALD (η5-C5H5)W(CO)2NO NH3 [66]
37
Figure 2-1. Microstructure of Cu diffusion barrier materials: A) single crystal; B)
polycrystalline; C) polycrystalline columnar; D) nano-crystalline; E) amorphous.
Adopted from A. Kaloyeros and E. Eisenbraun, “Ultrathin diffusion barrier/liners for
gigascale copper metallization”, Annu. Rev. Mater. Sci. 30 (2000) 363-385.
A)
B)
C)
D)
E)
38
Figure 2-2. Diagram showing the applications of metals and nitrides in modern semiconductor
devices. Adopted from H. Kim, “Atomic layer deposition of metal and nitride thin
films: Current research efforts and applications for semiconductor device processing”,
J. Vac. Sci. Technol. B 21 (2003) 2232-2261.
Cu diffusion barrier/adhesion promoter
Cu seed layer
Tungsten plug for via hole
Diffusion barrier Metal gate electrode
39
Dielectric deposition and patterning
TaNx diffusion barrier by PVD
Ta liner/adhesion promoter by PVD
Cu seed layer by PVD
Bulk Cu fill by ECD
Cu/liner/barrier by CMP
Dielectric
TaNx PVD
Ta PVD
Cu seed/Cu ECD
Dielectric deposition and patterning
TaNx diffusion barrier by PVD
Ta liner/adhesion promoter by PVD
Cu seed layer by PVD
Bulk Cu fill by ECD
Cu/liner/barrier by CMP
Dielectric
TaNx PVD
Ta PVD
Cu seed/Cu ECD
Figure 2-3. Simplified processing steps in dual-damascene structure for Cu metallization.
40
CHAPTER 3
EXPERIMENTAL PROCEDURE
3.1 Precursor Synthesis
The diorganohydrazido(2-) tungsten complexes Cl4(CH3CN)W(NNR2) (1: R2 = -(CH2)5-;
2: R2 = Ph2; 3: R2 = Me2) were prepared as described in the literature [24].
3.2 Film Growth
The each precursor was dissolved in benzonitrile (PhCN) in a concentration of 8.1 mg/mL,
9.6 mg/mL, and 7.4 mg/mL for 1, 2, and 3, filled into a gas-tight syringe, and pumped into a
nebulizer. A quartz plate in the nebulizer vibrates at a frequency of 1.44 MHz generating a mist
of precursor and solvent. Carrier gas flows through the nebulizer assembly and transports the
aerosol through the capillary tube from the syringe into a heated impinging jet. The mixture of
precursor and PhCN flows from the showerhead to reach the substrates on a heated graphite
susceptor. A custom-built vertical quartz cold wall CVD reactor system shown in Figure 3-1
was used to deposit the thin films on p-type boron-doped Si(100) single crystal substrates with
electrical resistivities in the range 1 to 2 Ω-cm. A graphite susceptor was heated by radio-
frequency (rf) induction coils to maintain the substrates at the specific deposition temperature.
The deposition temperature was varied from 300 to 700 °C in steps of 50 °C. The operating
pressure was maintained at 350 Torr using a mechanical roughing pump and pressure control
valve. The H2 (99.999 %, Airgas) carrier gas flow rate was 1000 sccm (sccm denotes cubic
centimeters per minute at STP), the NH3 (99.9999 %, Air Liquide) coreactant flow rate was 30
sccm, and the deposition time for all depositions was 150 min.
3.3 Film Characterizations
Several methods were used to characterize the composition, chemical bonding states,
microstruture, surface morphology, growth rate, and electrical properties of the films. X-ray
41
photoelectron spectroscopy (XPS) was used to identify the chemical composition and the
chemical bonding states of the elements in the film using a Perkin-Elmer PHI 5600 ESCA
system. XPS spectra were obtained by monochromatic Mg Kα ionizing radiation (1254 eV) with
the X-ray source operating at 300 W (15 kV and 20 mA). Prior to XPS measurement, Ar+ ions
were used to sputter as-deposited samples for 10 min to remove residual surface contamination.
X-ray diffraction (XRD) was used to identify the film microstructure with a Philips APD 3720
system, operating with Cu Kα radiation (40 kV and 20 mA). XRD was performed from 30 to 80
2θ° with 0.02 ° step size. Atomic force microscope (AFM) was used to measure the surface
roughness with a Digital Instruments Dimension 3100 system, operating in tapping mode. AFM
was performed with 2 Hz scan and with 512 by 512 resolution. Cross-sectional scanning
electron microscopy (SEM) was used to measure the thickness of the film on a JEOL JSM-
6335F to obtain the growth rate. The sheet resistance of the film was measured by the four-point
probe method using an Alessi Industries four-point probe to obtain film resistivity along with
thickness from cross-sectional SEM images.
3.4 Diffusion Barrier Testing
Cu (100 nm thickness) was deposited by reactive sputtering using a Kurt Lesker CMS-18
Sputter system at room temperature. Samples of WNxCy (15 - 20 nm thickness) deposited by
CVD at 400 °C on the Si(100) single crystal substrates were loaded via a load-lock system into
the process deposition chamber with a base pressure of 3 10-7 Torr. The chamber pressure
during deposition was 5 mTorr. The forward sputtering power for Cu was 200 W, while the
WNxCy/Si stacks were rotated at 20 rpm during deposition. Cu was deposited on top of
WNxCy/Si stacks to evaluate their performance as Cu diffusion barriers. The Cu/WNxCy/Si
stacks were then annealed in the CVD reactor at 500, 600, and 700 °C for 30 min/step.
42
Annealing was performed under N2 (99.999 %, Praxair) to protect the Cu layer from oxidation.
XRD and four-point probe were used to investigate the onset of the failure process via the
formation of Cu3Si. SEM imaging was used to reveal the Cu surface morphology. Cross-
sectional transmission electron microscope (TEM) imaging was used to detect the presence of
Cu3Si in the WNxCy/Si interface. Energy dispersive X-ray spectroscopy (EDS) qualitative
analysis was used to identify the presence of Cu Kα signal in the WNxCy/Si interface. The cross-
sectional image was taken by TEM using JEOL TEM 2010F to allow for high-resolution
imaging of multilayered interfaces. Prior to TEM imaging, focused ion beam (FIB) was used to
prepare samples for cross-sectional TEM using FEI Strata DB 235 to allow for precise cross-
sectioning in specific location. FIB was operated with a finely-focused beam of Ga+ ions.
43
1 2 3
Figure 3-1. The diorganohydrazido(2-) tungsten complexes Cl4(CH3CN)W(NNR2) (1: R2 =
-(CH2)5-; 2: R2 = Ph2; 3: R2 = Me2).
44
To Vacuum Pump
Heated Transfer Tube
Quartz Tube
Graphite SusceptorRF Coils
Impinging Jet
Water Cooled Flange
Carrier Gas
to Nebulizer
Dissolved Precursor
from Syringe Pump
Gate Valve
Precursor Aerosol
Plastic Tubing
Quartz Plate
Figure 3-2. Schematic diagram of the aerosol-assisted CVD system.
45
Figure 3-3. Process flow on film properties. (MAIC, http:\\maic.mse.ufl.edu, October, 2008).
XRD
CVD
H 2 (or NH 3 )
CVD
H 2 (or NH 3 )
Si WN x C y Si WN x C y
FE SEM
XPS
AFM
4PP
AES
Material Characterization
Diffusion Barrier CVD
46
Pre-anneal
Si Si
WNxCy
Si
WNxCy
Si
WNxCy
Cu
Si
WNxCy
Cu
Si
WNxCy
Cu
Si
WNxCy
CuWNxCy CVD Cu Sputter Anneal
XRD
FIB TEM/EDS
500~700 °C400 °C R.T.
Post-anneal
Pre-anneal Post-anneal
Sputter
15 nm 100 nm 30 min
4PP SEM
CVD
Figure 3-4. Process flow on diffusion barrier testing. (MAIC, http:\\maic.mse.ufl.edu, October,
2008).
47
CHAPTER 4
DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(N-pip) AS A SINGLE-SOURCE
PRECURSOR
4.1 X-ray Crystallographic Study of Cl4(CH3CN)W(N-pip)
Single crystals suitable for X-ray diffraction were obtained from compound 1 and
subjected to X-ray crystallographic structure determination (Table 4-1). The solid state structure
of 1 reveals the W metal center in a distorted octahedral geometry (Figure 4-1). Four Cl atoms
occupy the basal positions with the W-Cl bonds averaging 2.34 Å (Table 4-2) which is within the
expected range for W(VI)-Cl bonds [67]. The diorganohydrazido(2-) ligand is strongly bound to
the central metal atom as indicated by a short W(1)-N(1) distance of 1.752(3) Å . The short N(1)-
N(2) bond distance (1.265(4) Å ) within the hydrazido ligand suggests a high degree of
delocalization and multiple bond character throughout the W(1)-N(1)-N(2) unit. This
phenomenon has been reported in the literature for other hydrazido complexes of W with
multiple Cl ligands such as W(η5-C5Me5)Cl3(NNPh2) [68] (W-N 1.769(2) Å , N-N 1.296(3) Å ),
cis-[WCl3(NNH2)(PMe2Ph)2] [69] (W-N 1.752(10) Å , N-N 1.300(17) Å ) and
(CH3CN)Cl4W(NNMe2) [70] (W-N 1.769(5) Å , N-N 1.271(8) Å ). The piperidyl unit can be
found in a typical chair-like conformation with the C(1)-C(5)-N(2)-N(1) unit exhibiting a
trigonal planar arrangement as evidenced by the sum of the bond angles (359.7°) around N(2).
The remaining coordination site is occupied by a neutral acetonitrile solvent molecule. The
W(1)-N(3) bond distance of 2.237(3) Å is significantly shorter than those reported for related
tungsten imido compounds [18], suggesting a decreased trans-influence of the
diorganohydrazido(2-) ligand compared to the imido moiety.
4.2 Preliminary Precursor Screening
Multiple spectroscopic techniques were applied to evaluate the viability of 1 as a precursor
for WNxCy deposition [70]. Kinetic data obtained by 1H NMR spectroscopy confirmed an
48
expected weak bond between the acetonitrile ligand and the metal center. Positive-ion chemical
ionization (CI) mass spectrometry was performed to obtain some insight into the fragmentation
behavior of 1. The absence of molecular ion peaks is in good agreement with the acetonitrile
ligand being labile. Mass envelopes containing the piperidine moiety ([pip]+ and [H2pip]
+) were
observed in high abundance, suggesting that cleavage between the two hydrazido N atoms is
facile under high energy conditions (ionization during MS or pyrolysis during CVD). Thermal
behavior studies of 1 via thermogravimetric analysis (TGA) showed a drop in mass
corresponding to loss of the acetonitrile ligand at approximately 100 °C.
4.3 Film Structure
The XRD spectra in Figures 4-2A and 4-2B show amorphous and polycrystalline
microstructure for films deposited at 300 and 700 °C, respectively. The four characteristic
polycrystalline peaks exhibit locations that are consistent with β-WNxCy. The XRD spectra
contain no evident characteristic peaks observed up to 450 °C, indicating that amorphous films
were deposited from 300 to 450 °C. The polycrystalline peaks appearing in spectra of material
deposited from 500 to 700 °C indicate no preferred crystal orientation. Primary peaks at 37.24
and 42.98 2θ° are consistent with (111) and (200) orientation, while the other peaks at 62.58 and
74.98 2θ° are from (220) and (311), respectively. The XRD spectra in Figure 4-2C show the
evolution of film crystallinity with increasing deposition temperature. As the deposition
temperature increases to 700 °C, the (111) and (200) β-WNxCy peaks sharpen further. All of the
films show three additional sharp peaks at 33.08, 61.76, and 69.14 2θ°, indicating Si(200) Kα,
Si(400) Kβ, and Si(400) Kα radiation, respectively. Additionally, all films displayed one peak at
65.99 2θ°, representing W Lα radiation. This W peak comes from deposition on the target due to
evaporation of the W filament [71]. The ability to deposit amorphous films of WNxCy at low
49
temperature is highly significant for diffusion barrier applications since the formation of
polycrystalline films facilitates diffusion of Cu to the underlying Si via the grain boundaries.
4.4 Chemical Composition
XPS results for chemical composition (Figure 4-3) show that W, N, C, and O were present
in the films. No Cl contamination in the films was observed within detection limit of XPS (~ 1
at. %). The W level is highest between 450 and 500 °C, while the N, C, and O levels are fairly
steady in this range. From 300 to 400 °C, the C level is below 10 at. %, with the lowest level of
6 at. % for depositions at 400 °C. Between 500 and 700 °C, the C level increases gradually from
15 to 67 at. %. The overall trend for C content is consistent with the faster decomposition of
hydrocarbon groups in both the precursor and the solvent as the growth temperature increases,
leading to C incorporation into the film. As the deposition temperature increased from 300 to
400 °C, the N level increased from 10 to 18 at %. However, above 500 °C, N levels start to
decrease, as a consequence of increased C concentration in this range. When the deposition
temperature reaches 700 °C, the N level has declined to 5 at. % due to the steep rise in C levels at
high deposition temperatures. Typically, refractory metal nitride diffusion barriers show a
decreasing tendency of N incorporation with increasing deposition temperature due to N2
desorption [29, 33, 38, 72]. Films deposited at 300 C° show over 20 at. % of O, which decreased
drastically to 14 at. % at 450 °C. As the deposition temperature increased from 450 to 700 °C,
the O level decreased gradually to 5 at. %. From XRD spectra in Figure 4-2C, the
polycrystalline microstructure becomes evident for depositions performed at 500 °C. As the
deposition temperature increases, the average grain size increases as well. As the film starts to
crystallize, the microstructure gets denser, which inhibits post-growth of O interdiffusion into
the lattice of the film [73]. This result comes from the densification of film by polycrystal grain
growth between 500 to 700 °C.
50
4.5 Chemical Bonding States
XPS was used to measure the binding energy (BE) of atoms in the films. The W 4f
photoelectron line is a doublet due to spin orbit splitting into W 4f7/2 and W 4f5/2, while the N 1s,
C 1s, and O 1s photoelectron lines show a single peak. Figure 4-4A indicates the evolution of
XPS spectra in the W 4f BE region as the deposition temperature changes. The major W 4f7/2
and W 4f5/2 peaks at 400 °C are at 31.6 and 33.5 eV, which are close to values for WCx and WNx
(Table 4-3). These two values are higher than those for metallic W and lower than those for
WO3. These BE values are consistent with W in the β-WNxCy chemical bonding state. As the
deposition temperature rises to 700 °C, the BE increases to 31.8 and 33.8 eV, which correspond
to WCx and WNx as well. This slight increase in BE comes from more carbon-laden samples at
the higher temperature. For material deposited at 300 °C, the BE more closely resembles that of
WO3 but shifts toward values for WCx or WNx as the deposition temperature rises to 350 °C.
This indicates a chemical bonding state from WO3 dominant to WNxCy as the deposition
temperature increases from 300 to 700 °C. Figure 4-4B illustrates the change of the XPS spectra
in the region of the N 1s BE as the deposition temperature changes. The N 1s peaks are observed
at 397.3 eV, which is the reported value for WNx (Table 4-3). Hence, the N in the films is all
bound in the WNx polycrystals. N at the grain boundary can be excluded due to a single N 1s
peak without a second peak near 399 eV. The maximum N intensity is seen in the spectra of
films deposited at 400 °C, which is consistent with the high N level in those films. Figure 4-4C
indicates the evolution of the XPS spectra in the region of the C 1s BE as the deposition
temperature changes. As the deposition temperature increases to 700 °C, the C 1s peaks exhibit
higher intensity. At lower deposition temperature, the C 1s peaks occur at 283.1 eV, while at
higher temperature, the C 1s peaks are shifted to higher BE. Between 300 and 600 °C, the BE is
lower than reported for amorphous C, while over 600 °C, the BE is higher than that of WCx.
51
Deconvolution of the broad C 1s peak at 700 °C in Figure 4-4C yields two separate peaks, which
are at 283.1 and at 284.5 eV. The peaks at the lower BE are due to C in the β-WNxCy bonding
states, while the peaks at higher BE are consistent with amorphous C present outside of the β-
WNxCy nanocrystals. For deposition at temperatures higher than 650 °C, a small portion of
amorphous C starts to show up with WCx in the film. As the deposition temperature increases to
700 °C, much more amorphous C exists with WCx in the film. Figure 4-4D indicates the
dependence of the XPS spectra in the region of the O 1s BE as the deposition temperature
changes. The O peaks were observed near 530.5 eV, which is consistent with the presence of
WO3 (Table 4-3). O levels are at the maximum in the low temperature film growth due to film
crystallization and C incorporation. As the deposition temperature increases up to 700 °C, the
film density is increased by film crystallization and the grain boundaries are infiltrated by C
incorporation.
4.6 Lattice Parameter
The lattice parameter was determined by XRD using the 2θ position of the β-WNxCy(111)
diffraction peaks. The β-WNxCy peak position was calibrated to the Si(400) diffraction peak at
69.14 2θ°. The dashed line at 4.126 Å in Figure 4-5 indicates the value of the standard lattice
parameter for β-W2N and the dashed-dot line at 4.236 Å in Figure 4-5 indicates the value of the
standard lattice parameter for β-WC1-x. The lattice parameter in Figure 4-5 shows the increasing
tendency as deposition temperature increases from 500 to 650 °C. The change in lattice
parameter shows a composition change in polycrystals. The main reason that the lattice
parameter increases between 500 and 650 °C is that C is incorporated into the β-WNxCy
polycrystals, not at the grain boundary. As the indicated by the chemical composition in Figure
4-3, the C level continues to increase with deposition temperature over 500 °C. Between 650
and 700 °C, the lattice parameter decreases. At this range, the C level increases and W decreases
52
with almost no compositional change in N and O. The decrease in lattice parameter between 650
and 700 °C indicates the limitation of solubility for C in β-WNxCy polycrystals [45]. These
results are consistent with the chemical bonding states in Figure 4-4C, where the C 1s peak in
this range is shifted to higher BE and has a broad peak that can be deconvoluted into two
separate peaks. The lattice parameter decreases between 650 and 700 °C because C exists at the
grain boundary, not in the β-WNxCy polycrystals.
4.7 Average Grain Size
Average grain size was calculated using Scherrer‟s formula [71]. The most dominant β-
WNxCy(111) diffraction peak of the four characteristic polycrystalline peaks was used to
determine FWHM as the reference peak for Scherrer‟s formula. Average grain size (Figure 4-6)
increases between 500 and 600 °C, varying from 31 to 47 Å . As seen in Figure 4-2C, the films
were X-ray amorphous between 300 and 500 °C, which places a limit of 31 Å on the maximum
grain size. The overall tendency of polycrystal grain size increases with the deposition
temperature, varying from 500 to 700 °C.
4.8 Electrical Resistivity
The variation of film resistivity with deposition temperature is shown in Figure 4-7. The
lowest resistivity is 0.9 mΩ-cm at 550 °C and the values of film resistivity fluctuate with the
interplay of polycrystal grain growth, C content, O content, and film thickness between 500 and
700 °C. As shown in Figure 4-4, an increase in the amorphous C level as the deposition
temperature rises from 650 to 700 °C results in an increase in electron scattering, which causes
the film resistivity to increase. The highest film resistivity is 9.4 mΩ-cm for films deposited at
350 °C. The high N level in those films is consistent with increased film resistivity in the β-
WNxCy polycrystal structures, due to the higher resistivity for β-W2N relative to β-WC1-x.
53
4.9 Film Growth Rate
The growth rate is in the range 2.7 to 29.4 Å /min, as determined by cross-sectional SEM.
For films deposited between 650 and 700 °C, the growth rate increased drastically suggesting a
change in the growth mechanism at these temperatures. This observation was also confirmed by
the formation of WCx instead of WNxCy. Figure 4-8 is consistent with the presence of two
growth regimes. The region with the shallow slope is a mass transfer limited growth regime
between 450 °C and 600 °C. The region with the steep slope is a kinetically controlled growth
regime between 300 °C and 450 °C. The apparent activation energy calculated for the activated
process is 0.28 eV.
4.10 Diffusion Barrier Testing
XRD measurement, detection of etch-pits, and AES depth profile were used to detect Cu
transport through the film. XRD measurement was used to search for formation of Cu3Si, which
occurs after barrier failure on Si substrates. As shown in Figure 4-9, the XRD data show no
Cu3Si peaks in the region of 44 to 46 2θ°. Before annealing, only the Cu(111) peak is observed.
After annealing, there are three Cu peaks observed: Cu(111) at 43.44, Cu(200) at 50.80, and
Cu(220) at 74.42 2θ ° [74]. Cu recrystallization upon annealing resulted in increase in the
intensity of the Cu-related textures due to nucleation of the new grains or growth of preexisting
ones in the Cu/WNxCy stacks [71]. The etch-pit test was also used to search for Cu3Si on the Si
surface, which would be evident if there was Cu transport through the barrier. Figure 4-10
shows results of the etch-pit test on samples before and after annealing. However, there are no
etch-pits observed in either sample, indicating that no Cu was interdiffused or intermixed, and
the barrier did not fail upon annealing under N2 at 500 °C for 30 min. If there were Cu transport
through the barrier, defects caused by the formation of Cu3Si on Si substrate would be shown as
54
inverse pyramidal shaped etch-pits after etch-pit test. AES depth profile in Figure 4-11 shows
only negligible background signals for Cu where there is no Cu transport. The sharp interface
between WNxCy and Si indicate that there is no detectable Cu signal at this interface. Thus,
WNxCy deposited at 400 °C is a viable Cu diffusion barrier material to prevent Cu transport and
intermixing in Si during annealing under N2 at 500 °C for 30 min.
4.11 Conclusions
The tungsten piperidylhydrazido complex Cl4(CH3CN)W(N-pip) (1) was used as a single-
source precursor for film growth of WNxCy to investigate the film properties for diffusion barrier
applications. XRD results suggest that films deposited at temperature below 500 °C are X-ray
amorphous and films deposited at higher temperature are polycrystalline. The XPS of the W 4f
bonding state is consistent with W is present in WNxCy and WO3. For films deposited at the low
end of the temperature range, WO3 predominates and as the deposition temperature increases,
WNxCy becomes the dominant W species. XRD results, however, do not indicate any WO3
peaks. The XPS data on the N 1s bonding state suggest that N is present in WNx, while results
on the C 1s bonding state indicate that C is present in WCx and amorphous C. For depositions at
temperature higher than 650 °C, amorphous C coexists with WCx. The XPS data on the O 1s
bonding state suggest that O is present in WO3 and the O levels are the highest for growth
temperatures below 400 °C. As the deposition temperature varies, the film growth rate changes
from 2.7 to 29.4 Å /min, with the transition from a kinetically controlled growth regime to a mass
transfer controlled growth regime occurring near 550°C. Film resistivity changes with the
interplay of polycrystal grain growth, C content, O content, and film thickness. The WNxCy
films were evaluated to determine their suitability as Cu diffusion barriers. WNxCy deposited
from 1 is a viable Cu diffusion barrier material to prevent diffusion of Cu into Si after annealing
55
under N2 at 500 °C for 30 min. Further diffusion barrier testing and film characterization are
underway.
56
Table 4-1. Crystal data and structure refinement for Cl4(CH3CN)W(N-pip) (1)
Empirical formula C7H13Cl4N3W
Formula weight 464.85
Temperature (K) 173(2)
Wavelength (Å ) 0.71073
Crystal system Monoclinic
Space group P21/n
Unit cell dimensions (Å ) a = 9.7113(15) α = 90°
b = 14.874(2) β = 92.777(3)°
c = 9.7939(15) γ = 90°
Volume (Å3) 1413.0(4)
Z 4
Density (Mg/m3) 2.185
Absorption coefficient (mm-1
) 8.906
F(000) 872
Crystal size (mm3) 0.19 × 0.11 × 0.02
θ range for data collection (°) 2.49 - 27.50
Index ranges -10≤h≤12
-19≤k≤17
-12≤l≤9
Reflections collected 9391
Independent reflections 3245 [Rint = 0.0546]
Completeness to = 24.60° (%) 99.9
Absorption correction Integration
Max and min transmission 0.8420 and 0.2825
Data/restraints/parameters 3245/0/137
Goodness-of-fit on F2 1.155
Final R indices [I > 2σ(I)] R1a = 0.0203, wR2
b = 0.0541
Largest diff. peak and hole (Å-3
) 1.309 and -1.035
a R1 = Σ(||Fo| - |Fc||)/Σ|Fo|.
b wR2 = [Σ[w(Fo
2 – Fc
2)
2]/ Σ[w(Fo
2)
2]]
1/2.
S = [Σ[w(Fo2 – Fc
2)
2](n - p)]
1/2, w = 1/[σ
2(Fo
2) + (mp)
2 + np], p = [max(Fo
2,0) + 2Fc
2]/3.
57
Table 4-2. Selected bond distances (Å ) and angles (°) for Cl4(CH3CN)W(N-pip) (1)
W(1)-N(1) 1.752(3) N(1)-W(1)-Cl(1) 92.03(9)
W(1)-N(3) 2.237(3) N(1)-W(1)-Cl(2) 96.70(8)
W(1)-Cl(1) 2.3609(9) N(1)-W(1)-Cl(3) 95.00(9)
W(1)-Cl(2) 2.3252(8) N(1)-W(1)-Cl(4) 98.17(8)
W(1)-Cl(3) 2.3563(9) N(3)-W(1)-N(1) 178.24(11)
W(1)-Cl(4) 2.3444(8) W(1)-N(1)-N(2) 176.6(3)
N(1)-N(2) 1.265(4) N(1)-N(2)-C(1) 120.7(2)
N(2)-C(1) 1.461(4) N(1)-N(2)-C(5) 120.3(3)
N(2)-C(5) 1.473(4) C(1)-N(2)-C(5) 118.7(3)
58
Table 4-3. Reported binding energy (BE) values
W 4f7/2 W 4f5/2 N 1s C 1s O 1s Ref.
Metallic W 31.2-31.7 33.4 [75-77]
WNx 32.7–33.6 33.3-35.8 396.2-398.2 [76-80]
N at grain boundary 399.2-400.0 [33, 34]
WO3 35.5-36.7 37.6-37.8 528.2-531.6 [33, 76-
78, 81]
WCx 31.6-32.3 33.7-33.9 279.7-283.8 [75-77,
82]
Amorphous C 284.2-285.2 [76, 77,
83, 84]
59
Figure 4-1. Thermal ellipsoids diagram of the molecular structure of Cl4(CH3CN)W(N-pip) (1).
Thermal ellipsoids are drawn at 50% probability. H atoms are omitted for clarity.
60
A) B)
C) D)
Figure 4-2. XRD spectra for films deposited on Si(100) in H2 carrier: A) 300 °C, B) 700 °C, C)
between 300 and 700 °C, and D) standard diffraction plots for β-W2N and β-WC1-x.
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K β
Si(200)
300 °C
(a)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K β
700 °C
β -WNxCy(111)
β -WNxCy(220)
β -WNxCy(311)
Si(200)
β -WNxCy(200)
(b)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
300 °C
Si(400)
Si(400) K β
Si(200)
350 °C
450 °C
500 °C
550 °C
600 °C
β -WNxCy(200)
β -WNxCy(111)
β -WNxCy(220)
400 °C
650 °C
β -WNxCy(311)
700 °C
(c)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
β -W2N(200)
β -W2N(220)
β -W2N(311)
β -WC1-x(200)
β -WC1-x(220)
β -WC1-x(311)
β -WC1-x(111)
β -W2N(111)
(d)
JCPDS 25-1257
JCPDS 20-1316
β -W2N(220)
β -WC1-x(220)
61
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
) C
W
O
N
Figure 4-3. Variation in chemical composition of W, N, C, and O content in the films with
deposition temperature. Data are measured by XPS after 10 min Ar+ ion sputter.
62
A) B)
C) D)
Figure 4-4. Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature. Data are from XPS after 10 min Ar+ ion sputter.
28303234363840
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(a) W 4f
390392394396398400402
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(b) N 1s
276278280282284286288
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(c) C 1s
524526528530532534536
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
(d) O 1s
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
63
Figure 4-5. Change in lattice parameter with deposition temperature for polycrystalline films
deposited from 1 based on β-WNxCy(111) diffraction peaks.
Figure 4-6. Change in average grain size with deposition temperature for polycrystalline films
deposited from 1 based on β-WNxCy(111) diffraction peaks.
0
10
20
30
40
50
60
70
450 500 550 600 650 700 750
Temperature (°C)
Averag
e G
rain
Siz
e (
Å)
4.11
4.13
4.15
4.17
4.19
4.21
4.23
4.25
450 500 550 600 650 700 750
Temperature (°C)
Latt
ice P
aram
ete
r (
Å)
β -WC1-x: 4.236
β -W2N: 4.126
64
0
2
4
6
8
10
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Film
Resis
tivit
y (
mΩ
-cm
)
Figure 4-7. Change in film resistivity with deposition temperature. Data are measured by four-
point probe.
Figure 4-8. Change in growth rate with deposition temperature. Thickness measured by cross-
sectional SEM.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
1000/T (K-1
)
ln G
(Å
/m
in)
700 600 500 400 300
Deposition Temperature (°C)
65
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K β
Si(200)
As-grown
Cu(111)
500 °C
Cu(200)
No Cu3Si peaks
Cu(220)
Figure 4-9. The performance of diffusion barrier by XRD measurement for Cu/WNxCy/Si stacks
before and after annealing at 500 °C.
Figure 4-10. SEM images of Si surface after etch-pit test A) before annealing and B) after
annealing at 500 °C.
B) A)
66
0 5 10 15 20 25 30 35 40
Time (min)
In
ten
sit
y (
a.u
.)
W
Si
C
N
O
Cu
Figure 4-11. The performance of diffusion barrier by AES depth profile for Cu/WNxCy/Si stacks
after annealing at 500 °C.
67
CHAPTER 5
DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(N-pip): EFFECT OF NH3 ON FILM
PROPERTIES
5.1 Film Structure
Figure 5-1A shows the progression of XRD patterns with increasing deposition
temperature for films deposited in NH3/H2 atmosphere. The XRD patterns have been
compressed to include the results from all nine growth runs, and thus the resolution is diminished
in the figure. An analysis of the original data, however, reveals four reflections, which were
calibrated to the Si(400) diffraction peak. The reflections at 62.02 and 75.48 2θ° show low
intensity, as compared with the primary ones at 37.48 and 42.78 2θ°. The peak positions in these
patterns are well matched with a two-phase mixture of β-WN2 and β-WC1-x phases or their solid
solution β-WNxCy with the same crystal structure. Both standards exhibited a face-centered
cubic (fcc) structure with similar lattice parameter (β-WN2: 4.124 Å , β-WC1-x: 4.236 Å ). In
addition to the WNxCy peaks, three sharp single crystal peaks were detected at 33.08, 61.76, and
69.14 2θ°, and associated with Si(200) Kα, Si(400) Kβ, and Si(400) Kα radiation, respectively.
The XRD spectra indicate that amorphous films were deposited from 300 to 450 °C, while
polycrystalline materials were deposited at and above 500 °C. The relative intensities of the four
characteristic reflections are consistent with random grain orientation and as expected their
variation with growth temperature follow the same pattern as the film thickness. Note that the
growth rate data is proportional to film thickness since the growth time is constant for all runs.
Primary peaks at 37.48 and 42.78 2θ° are assigned to the (111) and (200) orientations, while the
other two reflections at 62.02 and 75.48 2θ° are attributed to (220) and (311) orientations,
respectively. As the deposition temperature was increased from 500 to 600 °C, the peak
intensities increased, primarily as a result of increased film thickness and possibly changes in
crystallinity. The intensities then successively decreased for the 650 and 700 °C deposited
68
samples. This result is in contrast to that of the previous work using only H2 as the carrier gas in
which the growth rate continued to increase with temperature (Figure 4-8). The decreased
intensity when NH3 is present is likely a result of precursor depletion from parasitic gas phase
reactions. Transamination with NH3 has been postulated to remove the hydrocarbon group in the
precursor, changing the rate-determining step [85], and thus the growth rate of the films
deposited in the presence of NH3 is less than that for H2 only.
5.2 Surface Morphology
The root-mean-square (rms) roughness of the surface of films deposited at 300 °C in the
absence of NH3 was determined by AFM to be 0.99 nm, with a rise to 17.17 nm for deposition
at 600 °C. From the AFM micrographs in Figures 5-2C and 5-2D, the rms roughness of the film
surface for films deposited in the presence of NH3 was 0.81 nm at 300 °C and 1.28 nm at 600 °C,
indicating addition of NH3 results in films with smoother surfaces. The increase in surface
roughness with the increase in the deposition temperature up to 600 °C is accompanied by
increased crystallinity and grain size (Figure 5-1). The decrease in roughness is consistent with
an amorphous microstructure and more facile migration of absorbed species on the surface at
higher deposition temperatures [86].
5.3 Chemical Composition
Despite the presence of Cl in the precursor, no peaks were observed for either Cl 2p3/2 or
Cl 2s at 199 and 270 eV, respectively, ruling out Cl contamination in the films within the
detection limit of XPS (~ 1 at. %). The absence of Cl signals is consistent with prior
computational results on the related imido complex Cl4(CH3CN)W(NiPr), for which a
mechanistic pathway was found for reaction of the H2 carrier gas with W-Cl bonds to produce
HCl in the gas phase [87]. Figure 5-3A shows the W levels have their highest value between 450
and 500 °C. Between 650 and 700 °C, the measured W levels in the films deposited with NH3
69
are higher than that in films deposited without NH3. This difference is related with the chemical
bonding states of C 1s in Figure 5-4C. Between 650 and 700 °C, C 1s binding energy (BE)
shifted from lower BE to higher BE, indicating that amorphous C is more dominant than WCx.
The increased extent of amorphous C formation dilutes the amount of W deposited. Figure 5-3B
shows that the N levels increased over the entire temperature range after the addition of NH3.
The highest N levels for films deposited with NH3 (24 at. % at 400 °C) is greater than that of
films without NH3 (18 at. % at 400 °C). The NH3 coreactant is used as an additional N source,
which allows deposition of high N level films compared to depositions without NH3. However,
the increase in the flow rate of NH3 shows no significant variation in N levels [88]. As
deposition temperature increases up to 700 °C, the N levels drop gradually. For films deposited
at 700 °C, the N levels in single-source deposition indicate an N concentration without including
NH3 is 5 at. %, while for films deposited with NH3 the N increased to 12 at. %. For refractory
metal nitride diffusion barriers, the N levels generally decrease with increasing deposition
temperature in film deposited without NH3 and with NH3. It has been suggested that the higher
deposition temperature increases the rate of N desorption as N2 gas, evidenced by Figure 5-3B
[40]. The results shown in Figure 5-3C point to lower C levels in films deposited with NH3 than
that in single-source deposition from 500 to 700 °C. The decrease in the C levels is attributed to
increased competition from N when NH3 is present. Addition of NH3 seems to have no
significant effect on the C levels in the films deposited below 500 °C, likely due to the lower
reactivity at lower temperature. Figure 5-3D shows that deposition with NH3 has lower O levels
than deposition without NH3. As the deposition temperature increased from 450 to 700 °C, the
O levels in the presence and absence of NH3 decreased gradually to 5 at. %. The low O
incorporation is consistent with dense WNxCy films at higher deposition temperature.
70
Crystallization occurs to a greater extent at higher temperature reducing diffusion of O from air
into the films [89].
5.4 Chemical Bonding States
The values of the BE relative to the emitted the kinetic energy (KE) determined by XPS
were used to identify the elemental chemical bonding states. Figure 5-4A displays the evolution
of XPS patterns for the W 4f BE as deposition temperature is varied for films deposited with
NH3. For films deposited at 300 °C, the W 4f BE is higher than WCx and WNx phases. The
major W 4f7/2 and W 4f5/2 peaks for films deposited at 300 °C are at 36.8 and 37.7 eV, which are
close to these in the WO3 phase. These values for W 4f7/2 and W 4f5/2 peaks agree well with the
reported values of 35.5 – 36.7 eV and 37.6 – 37.8 eV for the WO3 phase [33, 76, 78, 90]. As the
deposition temperature increased from 300 to 350 °C, the W 4f BE shifted from the higher BE
(WO3) to the lower BE (WCx and WNx). The major W 4f7/2 and W 4f5/2 peaks for films
deposited at 350 °C are at 31.7 and 33.6 eV, which are close to these of the WCx and WNx phases.
These values for W 4f7/2 and W 4f5/2 peaks agree well with the reported values of 32.7 – 33.6 eV
and 33.3 – 35.8 eV in WNx phase. Also, these values for W 4f7/2 and W 4f5/2 peaks agree well
with the reported values of 31.6 – 32.3 eV and 33.7 – 33.9 eV in WCx phase [76, 78, 80, 90].
This indicates that the peak shift of W 4f occurs at lower temperature than for films deposited
without NH3. The increased N levels in the films as shown in Figure 5-3 B is believed to be
responsible for this shift. The major W 4f7/2 and W 4f5/2 peaks for films deposited at 400 °C are
at 31.7 and 33.7 eV, which are close to the reported values for WCx and WNx phases. These two
values are higher than metallic W and lower than WO3 [76, 90]. From 350 to 700 °C, the major
W 4f7/2 and W 4f5/2 peaks correspond to WCx and WNx phases, indicating that the chemical
bonding state in W changes from a dominant WO3 phase to the mixture of β-WN2 and β-WC1-x
phases or β-WNxCy single solid solution as deposition temperature increases to 700 °C.
71
The evolution of XPS patterns for the N 1s BE with deposition temperature for films
deposited with NH3 is summarized in Figure 5-4B. The N 1s peak located at 397.3 eV is close to
the reported value for WNx phase. This value agrees well with the reported values of 396.2 –
398.2 eV in WNx phase [76, 78, 80, 90]. It appears that the N in the film is bound to W in the
WNx phase. The intensity of this N 1s peak is much higher over the entire temperature range, as
compared to those for films deposited without NH3. Films deposited at 400 °C have the highest
intensity of N, indicating the highest N levels in the films as shown in Figure 5-3B. A single N
1s peak indicates that N has the same metal nitride bonding state over the entire temperature
range, irrespective of the other components in the film. From 300 to 700 °C, there is only a
single N 1s peak near 397.3 eV (i.e., no second N 1s peak near 400.0 eV).
XPS patterns for C 1s BE are shown in Figure 5-4C over the range of deposition
temperature for films deposited with NH3. As shown in this figure, the BE of C 1s peak located
at around 283.2 eV and corresponding to WCx phase is evident in films grown up to 650 °C. For
films deposited at 700 °C, the bonding states of C 1s shifted from lower BE to higher BE.
Deconvolution of the broad C 1s peak for films deposited at 700 °C using a Gaussian-Lorentzian
function with background subtraction yields two separate peaks. The BE of the C 1s peak
located at 284.7 eV corresponds to amorphous C present outside of the β-WNxCy nanocrystalline
regions, while the BE of the C 1s peak located at 283.4 eV corresponds to WCx in the β-WNxCy
nanocrystals. The former value for C 1s peak agrees well with the reported values of 284.2 –
285.2 eV for the amorphous C phase, while the latter value for C 1s peak agrees well with the
reported values of 279.7 – 283.8 eV for the WCx phase [75, 76, 82-84, 90]. Amorphous C begins
to appear with the WCx phase in the film deposited at 700 °C in the presence of NH3, while
72
amorphous C appears at 650 °C in the absence of NH3. This indicates the addition of NH3
promotes metal carbide bonding at higher temperature (650 °C).
O levels were also probed by XPS. Figure 5-4D traces the evolution of the O 1s BE with
deposition temperature for films deposited with NH3. The XPS pattern in O 1s BE over the
entire temperature range is similar to that measured in films deposited with no added NH3. The
O 1s peaks were near 530.4 eV, which is close to the reported value for WO3 phase. This value
for the O 1s peak agrees well with the reported values of 528.2 – 531.6 eV for the WO3 phase
[33, 76, 78, 90]. As deposition temperature increases to 700 °C, the peak intensity of O 1s
decreased as a result of film crystallization and C incorporation, blocking uptake along grain
boundary.
5.5 Film Growth Rate
The growth rate in the presence of NH3 was low, in the range 0.6 to 4.2 Å /min as
compared to the range 2.7 to 29.4 Å /min for films deposited in the absence of NH3. Figure 5-5
shows the variation of growth rate with deposition temperature for films deposited with and
without NH3. Both plots reveal a transition from a kinetically controlled growth regime to a
mass transfer controlled one. Films deposited with NH3 had a transition point near 450 °C, while
films deposited without NH3 had a slightly higher transition temperature near 500 °C. These
differences in activation energy, transition temperature, and absolute growth rate are consistent
with a change in deposition mechanism due to the addition of NH3. It is also noted that the
growth rate for films deposited with NH3 was decreased at higher temperature (650 to 700 °C),
likely a result of precursor depletion in the gas phase near the substrate surface. Transamination
with NH3 has been postulated to remove the hydrocarbon group in 1, changing the rate-
determining step, and thus the growth rate of the films deposited with NH3 is different from that
without added NH3 [85].
73
5.6 Electrical Resistivity
The film resistivity was determined from measurement of the sheet resistance (four-point
probe) and film thickness (cross-sectional SEM). The effect of growth temperature on the film
resistivity for films deposited with NH3 is shown in Figure 5-6. Films deposited at 300 °C show
the lowest film resistivity (290 μΩ-cm) and the values of film resistivity increase with the
interplay of grain boundary density, film microstructure, film density, chemical bonding states,
and film thickness over the entire temperature range. At lower deposition temperature, an
increase in N levels with WNx bonding states results in a decrease in film resistivity and at
higher deposition temperature, an increase in film thickness results in an increase in film
resistivity. Films deposited at 600 °C show the highest film resistivity (5450 μΩ-cm). The
resistivity for films deposited with NH3 at lower deposition temperature (350 – 450 °C) was
lower than those deposited without NH3. This is attributed to increase in the film N content since
the resistivity of WNx is lower than WO3. At higher deposition temperature (500 – 700 °C), the
film resistivity when deposited with NH3 was higher due to the decrease in C content since the
resistivity of WNx (4000 μΩ-cm) is higher than of WCx (300 – 400 μΩ-cm) [59]. The formation
of WCx is an important factor in decreasing film resistivity. This observation was confirmed by
the XPS results for the bonding states of C in the films, as shown in Figure 5-4C. Hence, the
proper combination of WNx and WCx is important in formation of ternary-based refractory metal
nitrides for diffusion barrier applications.
5.7 Diffusion Barrier Testing
XRD measurements were used to confirm the formation of Cu3Si that occurs after barrier
failure for Si substrates. As shown in Figure 5-7, the XRD patterns show no reflections
attributable to Cu3Si. Generally, the standard diffraction peaks of Cu3Si appear near 44.00 to
46.00 2θ° with barrier failure. After annealing, there are three peaks clearly observed, which are
74
assigned to Cu(111) at 43.44 2θ°, Cu(200) at 50.80 2θ°, and Cu(220) at 74.42 2θ°, while patterns
on samples before annealing only indicate Cu(111). Cu recrystallization upon annealing resulted
in an increase in the intensity of the Cu-related reflections due to nucleation of the new grains or
growth of preexisting ones in the Cu/WNxCy/Si stacks [71]. It is noted that for metallization
applications, the Cu(111) texture is preferred since it shows a higher resistance to
electromigration [91]. In typical diffusion barrier test results, the Cu XRD peak intensities
decrease as silicide peaks are detected along with the evolution of Cu peaks near 44.00 to 46.00
2θ° [74]. Cross-sectional TEM images were taken to observe the quality of the Cu/WNxCy and
WNxCy/Si interfaces. As shown in Figure 5-8, cross-sectional TEM images reveal that there is
no Cu transport through WNxCy before and after annealing under N2 at 500 °C for 30 min. Both
Cu/WNxCy and WNxCy/Si interfaces are clearly defined without any evidence of intermixing
between the layers after annealing. The XRD patterns and cross-sectional TEM images reveal
no failures of diffusion barrier in Cu/WNxCy/Si stacks. Hence, WNxCy is a promising Cu barrier
material.
5.8 Conclusions
The tungsten piperidylhydrazido complex Cl4(CH3CN)W(N-pip) (1) was used to deposit
WNxCy with NH3 to investigate the effect of this coreactant on the film properties for diffusion
barrier applications. The deposited films show higher N levels with lower C incorporation as
compared to films deposited without NH3. XRD results suggest that films deposited at a lower
deposition temperature (below 500 °C) were amorphous with crystallinity evolving at higher
deposition temperature. The XPS W 4f bonding state indicates that most of the W is present as a
mixture of WNx and WCx phases or a WNxCy single solid solution. XPS results for both the W
and O indicate WO3 is present at low deposition temperature (300 °C) in the amorphous state
(XRD results) and as deposition temperature increases, WNxCy becomes the dominant W phase
75
rather than WO3. XPS spectra of the O 1s bonding state show low O incorporation at higher
temperature, which produces films with higher density. An examination of the XPS N 1s
bonding state indicates that N is present in the WNx phase. XPS spectra show films deposited at
400 °C have the highest N levels. The XPS C 1s bonding state results suggest that C is present
as WCx and amorphous C. The C 1s BE shifted from lower energy (283.1 eV) to higher energy
(284.5 eV) for films deposited at 700 °C, indicating that amorphous C coexists with WCx. XPS
observation of the O 1s bonding state indicates that O is present as WO3. XPS spectra also show
lower O incorporation at higher temperature, which produces films with higher density. The film
growth rate with NH3 addition varied in the range 0.6 to 4.2 Å /min over the entire temperature
range of study. Large film resistivity changes were observed and can result from various reasons
including grain boundary density, film microstructure, chemical bonding states, and film
thickness. Films deposited at 300 °C show the lowest film resistivity (290 μΩ-cm), while the
film resistivity was lower at low deposition temperatures as compared to films deposited without
NH3. WNxCy thin films of 20 nm thickness were tested for barrier performance. The results
show that WNxCy films are viable diffusion barriers to prevent Cu interdiffusion and intermixing
with Si after annealing under N2 at 500 °C for 30 min.
76
A) B)
Figure 5-1. XRD spectra for films deposited on Si(100) with NH3: A) between 300 and 700 °C;
B) standard diffraction patterns for β-W2N and β-WC1-x.
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
β -W2N(200)
β -W2N(220)
β -W2N(311)
β -WC1-x(200)
β -WC1-x(220)
β -WC1-x(311)
β -WC1-x(111)
β -W2N(111)
(b)
JCPDS 25-1257
JCPDS 20-1316
β -W2N(220)
β -WC1-x(220)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
300 °C
Si(400)Si(400) Kβ
Si(200)
350 °C
450 °C
500 °C
550 °C
600 °C
β -WNxCy(200)β -WNxCy(111)
400 °C
650 °C
700 °C
(a)
β -WNxCy(200)β -WNxCy(111)
77
Figure 5-2. Surface morphology of films deposited on Si(100) substrate at various temperature:
A) 300 °C without NH3; B) 600 °C without NH3; C) 300 °C with NH3; D) 600 °C
with NH3.
A) B)
D) C)
78
A) B)
C) D)
Figure 5-3. Variation in chemical composition of A) W, B) N, C) C, and D) O content in the
films with deposition temperature with and without added NH3. Data are measured
by XPS after 10 min Ar+ ion sputter.
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
With NH3
Without NH3
(a) W
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
(b) N With NH3
Without NH3
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
(c) C With NH3
Without NH3
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
(d) O With NH3
Without NH3
79
A) B)
C) D)
Figure 5-4. Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature in the presence of NH3. Data are from XPS after 10 min Ar+
ion sputter.
28303234363840
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(a) W 4f
390392394396398400402
Binding Energy (eV)N
(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(b) N 1s
276278280282284286288
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(c) C 1s
524526528530532534536
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
(d) O 1s
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
80
Figure 5-5. Change in growth rate with deposition temperature for films deposited with and
without added NH3. Thickness was measured by cross-sectional SEM.
0
2000
4000
6000
8000
10000
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Film
Resis
tivit
y (
μΩ
-cm
)
With NH3
Without NH3
Figure 5-6. Change in film resistivity with deposition temperature with and without added NH3.
Data are measured by four-point probe.
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
1000/T (K-1
)
ln G
(Å
/m
in)
With NH3
Without NH3
700 600 500 400 300
Deposition Temperature (°C)
81
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K β
Si(200)
As-grown
Cu(111)
500 °C
Cu(200)
No Cu3Si peaks
Cu(220)
Figure 5-7. The performance of diffusion barrier by XRD measurement for Cu/WNxCy/Si stacks
before and after annealing at 500 °C.
82
Figure 5-8. TEM cross-sectional images of Cu/WNxCy/Si stacks: [A) and B)] before annealing
and [C) and D)] after annealing at 500 °C.
A) B)
C) D)
Si WNxCy Cu Si
WNxCy
Cu
Si
Cu
WNxCy
Si
WNxCy
83
CHAPTER 6
DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(NNPh2) AS A SINGLE-SOURCE
PRECURSOR
6.1 Film Structure
Figure 6-1 displays XRD patterns for films grown in the temperature range 300 to 700 °C.
The XRD spectra indicate amorphous film deposition below 500 °C (Figure 6-1), while the films
grown above 500 °C yield four characteristic peaks positions that are consistent with β-WNxCy
(Figure 6-1D). The primary peak at 37.20 2θ° is consistent with the (111) orientation, while
additional peaks at 43.66, 63.24, and 75.38 2θ° are consistent with (200), (220), and (311)
orientation, respectively. As the deposition temperature increases from 500 to 700 °C, the
reflection associated with the (111) orientation sharpens, likely a result of increasing grain size.
Since the film grown at and above 500 °C produce a polycrystalline microstructure, rapid
diffusion of Cu along grain boundaries the underlying Si renders these films impractical for
barrier applications. This is not an issue since deposition above 400 °C is undesirable as many
low κ materials have weak thermal stability [27, 92]. Thus, this precursor is capable of
depositing amorphous films at deposition temperature ≤ 400 °C.
6.2 Lattice Parameter and Average Grain Size
Analysis of the diffraction patterns for the polycrystalline films was performed to estimate
the lattice parameter and average grain size. The standard lattice parameters for face-centered
cubic (fcc) β-W2N (4.126 Å ) and β-WC1-x (4.236 Å ) are shown in Figure 6-2A along with the
estimated lattice parameter of the deposited films as a function of temperature. The rock salt
structure for β-W2N and β-WC1-x consists of W located on the fcc positions, and N and C located
on the octahedral interstitial sites [93]. The 2θ position of the β-WNxCy(111) diffraction peaks
can vary due to the change in chemical composition in the films. If the dominant peak position
84
is between 37.01 2θ° (β-WC1-x) and 37.77 (β-W2N), according to Bragg‟s law, the lattice
parameter will be between 4.126 and 4.236 Å , indicating that N, C, and vacancies are mixed on
the interstitial sublattice. This is the case for films deposited at and above 500 °C. The lattice
parameter increases from 4.15 to 4.20 Å between 500 and 650 °C, indicating that C composition
increases with growth temperature. The lattice parameter at 700 °C, however, decreases to 4.18
Å , suggesting that the incorporation of C into the interstitial sublattice results in expanding the
lattice in the film structure [18].
The most dominant β-WNxCy(111) diffraction peak was used to estimate the average grain
size using Sherrer‟s formula from the experimental FWHM of this reflection. As shown in
Figure 6-2B, the average grain size for the polycrystalline phase increases with temperature in
the range 500 to 700 °C, varying from 25 to 55 Å (Figure 6-2B). A polycrystalline phase with
grain size below 50 Å is generally categorized as nanocrystalline [2]. Increasing grain size with
deposition temperature is common and is often attributed to higher surface diffusivity of
absorbed species leading to formation of stable nuclei and subsequent growth, then decreased
because an increase in the C levels on the film surface inhibits surface diffusion, nucleation and
growth [18].
6.3 Chemical Composition
The presence of Cl in 2 is of interest due to the need for deposition of Cl-free thin films for
diffusion barrier applications. As a first test, no peaks from Cl were observed within the
detection limit of XPS (~ 1 at. %) for either Cl 2p3/2 or Cl 2s at 199 and 270 eV, respectively.
Figure 6-3 shows the elemental composition in the films deposited from 2 as a function of
temperature. The W level is highest for the films grown at 450 and 500 °C, varying from 54 to
56 at. %, while N, C, and O levels are fairly constant for T ≤ 500 °C. As deposition temperature
increases from 500 to 700 °C, the W level decreases to 28 at. %. As deposition temperature
85
increases from 300 to 550 °C, the N level varies from 10 to 14 at %. The N level in films
deposited above 550 °C, however, shows a gradual decrease, accompanied by an increased C
level. The N level for films deposited at 400 °C has the highest value (14 at. %). The C level is
lowest for films deposited at 400 °C (14 at. %), while at lower deposition temperature (≤ 400 °C),
the C level varies from 30 to 24 at. %. As deposition temperature increases up 700 °C, the C
level increases to 62 at. %. The overall trend for C content is due to the faster decomposition of
hydrocarbon groups in the precursor and the solvent (PhCN) as the deposition temperature
increases, leading to C incorporation into the film. As the deposition temperature increases from
300 to 700 °C, the O level decreases gradually to 5 at. %. The low O incorporation indicates that
WNxCy films have dense microstructure at higher deposition temperature.
6.4 Chemical Bonding States
XPS was used to give information for the bonding states in films and the results are
summarized for each elements in Figure 6-4A-D. The evolution of XPS patterns in the BE of W
4f as the deposition temperature varies is shown in Figure 6-4A. For films deposited at 300 °C,
the major W 4f7/2 and W 4f5/2 peaks are at 31.5 and 33.7 eV, which are close to WNx and WCx
phases. These values for W 4f7/2 and W 4f5/2 peaks agree well with the range in reported values
of 32.7 – 33.6 eV and 33.3 – 35.8 eV in the WNx phase [76, 78-80, 90]. Also, these values for
the W 4f7/2 and W 4f5/2 peaks agree well with the reported ranges of 31.6 – 32.3 eV and 33.7 –
33.9 eV for the WCx phase [75, 76, 82, 90]. As deposition temperature increases from 300 to
700 °C, the major W 4f7/2 and W 4f5/2 peaks are observed at bonding energies of 31.5 – 31.6 and
33.5 – 33.8 eV, which agree well with WNx and WCx phases. The XPS results indicate that W is
primarily bonded to C and N. The evolution of XPS patterns in BE of the N 1s as the deposition
temperature varies is shown in Figure 6-4B. For films deposited at 300 °C, the major N 1s peak
is located at 397.4 eV, which is contained within the reported range of 396.2 – 398.2 eV in WNx
86
phase [76, 78-80, 90]. As deposition temperature increases from 300 to 700 °C, the major N1s
peak shows a slight shift for higher BE at 397.3 – 397.5 eV, which agrees well with WNx phase.
All N in the film is bound in the WNx phase, not in the N at the grain boundaries. The intensity
of N is higher at lower deposition temperature (< 400 °C). A single N 1s peak indicates that N
has the same metal nitride bonding state over the entire temperature range, irrespective of the
other contents in the films. The evolution of XPS patterns in BE of C 1s as the deposition
temperature varies is shown in Figure 6-4C. For films deposited at 300 °C, the major C 1s peak
is at 283.1 eV, which is close to WCx phase. As deposition temperature increases to 600 °C, the
BE of C 1s peak is at 283.1 – 283.2 eV, which agrees well with WCx phase. Deconvolution of
the broad C 1s peak for films deposited at 700 °C using Gaussian-Lorentzian function with
background subtraction yields two separate peaks associated with W-C and C-C. The BE of the
C 1s peak located at 284.4 eV agrees well with C-C bonding outside of the β-WNxCy
nanocrystals, while the BE of C 1s peak located at 283.8 eV agrees well with W-C bonding in
the β-WNxCy. The former value for C 1s peak agrees well with the reported ranges of 284.2 –
285.2 eV in amorphous C phase, while the latter value for the C 1s peak agrees well with the
reported ranges of 279.7 – 283.8 eV in WCx phase [75, 76, 82-84, 90]. The presence of the
bulky phenyl ligands in 2 is of interest since thermal decomposition would lead to more C in the
films. For diffusion barrier applications, low electrical resistivity is significant and the WCx
phase is more conductive than WNx phase [13]. The evolution of XPS patterns in the BE of O 1s
as a function of deposition temperature is shown in Figure 6-4D. For films deposited at lower
deposition temperature (≤ 400 °C), the major O 1s peak is at 530.2 eV, which is within the
reported range of 528.2 – 531.6 eV associated with WO3 phase [33, 76, 78, 81, 90]. As
evidenced by Figure 6-4D, the peak intensity of O 1s decreased with increasing deposition
87
temperature to 700 °C. IN the range of higher deposition temperature, film crystallization (>
500 °C) and C incorporation (W-C and C-C phases) are important factors to block grain
boundary regions in the films. The XPS results indicate that W is primarily bonded to N and C
for films deposited over the entire temperature range (300 – 700 °C).
6.5 Film Growth Rate
The growth rate was determined from the measured film thickness by cross-sectional SEM
images. Figures 6-5A and 6-5B shows images for films grown at the lowest (300 °C) and
highest (700 °C) growth temperature. Figures 6-5C and 6-5D display the SEM images for
surface morphology indicate that films deposited at 700 °C are polycrystalline with a rough
surface. As shown in the Arrhenius plot in Figure 6-6, the growth rate varied from 1.0 to 25.4
Å /min. The growth rate at 700 °C increased drastically suggesting a change in growth
mechanism at this temperature. This observation is consistent with the shift in binding energy in
XPS results, indicating amorphous C coexists W-C bonding states. The Arrhenius plot indicates
the presence of two growth regimes and their slopes consistent with mass transfer limited growth
at higher temperature and surface reaction limited at lower temperature. The apparent activation
energy calculated for the activated process is 0.49 eV in the surface reaction limited growth
regime.
6.6. Electrical resistivity
Film resistivity varied from 0.6 to 7.9 mΩ-cm (Figure 6-7). The lowest resistivity of 0.6
mΩ-cm is obtained for films deposited at the lowest temperature (300 °C). The value for films
deposited at 300 and 350 °C agrees well with the reported value of 0.3 to 0.4 mΩ-cm in WNxCy
phase. WNxCy phase shows a much lower film resistivity than WNx (~ 4.5 mΩ-cm) [13]. There
are several factors that influence the WNxCy resistivity, including the distribution of W-N and
88
W-C bonding states, the presence of W-O bonding state, film microstructure, and vacancies in
sublattice. The interplay of these factors can lead to a complex variables of the film resistivity
with deposition temperature. The low film resistivity at low temperature is likely due to higher
W-C phase. The formation of WCx in WNxCy (0.3 – 0.4 mΩ-cm) leads much more conductive
films as compared with WNx (4.0 mΩ-cm ) [59].
6.7 Diffusion Barrier Testing
Cross-sectional TEM images show that there is no diffusion of Cu into WNxCy and Si
before and after annealing under N2 at 500 °C for 30 min (Figure 6-8). Both Cu/WNxCy and
WNxCy/Si interfaces are clearly observed without any intermixing between the layers after
annealing, which indicates WNxCy thin films block the Cu diffusion. The EDS depth profile
shows that the Cu Kα peak decreases sharply at the Cu/WNxCy interface (Figure 6-9). Although
it is clear that Si Kα and W Lα peaks are present in the films, it is impossible to separate two
overlapping peaks due to the limitation of EDS. The EDS depth profile, however, clearly shows
that no Cu diffusion is observed between Cu/WNxCy and WNxCy/Si interfaces after annealing.
Cross-sectional TEM images and EDS depth profiles reveal no onset of failure in the
Cu/WNxCy/Si stacks. XRD measurements show that there is no formation of Cu3Si that occurs
after failure for Si substrates either before or after annealing under N2 at 500 °C for 30 min
(Figure 6-10). The XRD patterns show no reflections attributable to Cu3Si. Before annealing
there is only one peak clearly observed, which is assigned to Cu(111), while diffraction patterns
after annealing evidence Cu(111), Cu(200), and Cu(220). Cu recrystallization upon annealing
causes an increase in the intensity of Cu texture due to grain growth in the Cu/WNxCy/Si stacks.
The enlargement of Cu grains results in the reduction of the density of Cu grain boundaries,
which contributes to lower film resistivity due to low electron scattering. It is noted that for Cu
interconnect technology, the Cu(111) texture is preferred since it shows greater resistance toward
89
electromigration. In summary, the cross-sectional TEM images, EDS depth profiles, and XRD
measurements reveal no onset of failure of the diffusion barrier in the Cu/WNxCy/Si stacks.
6.8 Conclusions
Cl4(CH3CN)W(NNPh2) (2) was evaluated as a single-source precursor for CVD of WNxCy.
XRD patterns show that films deposited below 500 °C were amorphous, while polycrystalline
films were grown between 500 and 700 °C. The lattice parameter of the polycrystalline films
varied from 4.15 to 4.20 Å , while the average grain size increased from 25 to 55 Å over the
temperature range of growth. Examination of the XPS W 4f bonding state indicates that most of
the W is present as a mixture of WNx and WCx or a WNxCy single solid solution. XPS
measurements revealed that W was predominantly bonded to N and C, with C portion increasing
with growth temperature. This was attributed to in part to decomposition of the solvent. The
amount of W bonded to O, however, was limited. The XPS N 1s bonding state indicates that N
is present in tWNx, while the XPS C 1s bonding state indicates that C is present in WCx and as
amorphous C. A large variation in film resistivity was measured and is due to the interplay of
the combination of W-N and W-C bonding states, the presence of W-O bonding state, film
microstructure, and film thickness. The results show that WNxCy films are viable Cu barrier
materials to prevent diffusion of Cu into Si after annealing under N2 at 500 °C for 30 min.
Therefore, WNxCy is a viable Cu barrier material to prevent diffusion of Cu into Si for Cu
interconnect technology.
90
A) B)
C) D)
Figure 6-1. XRD spectra for films deposited on Si(100) at various temperatures: A) 300 °C, B)
700 °C, C) between 300 and 700 °C, and D) standard powder diffraction pattern for
β-W2N and β-WC1-x.
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
300 °C
(a)
Si(400) K β
Si(200)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
700 °C
(b)
Si(400) K β
β -WNxCy (111)
β -WNxCy (220)
β -WNxCy
Si(200)
β -WNxCy (200)
Si(400)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
300 °C
Si(400)
Si(400) Kβ
Si(200)
350 °C
450 °C
500 °C
550 °C
600 °C
β -WNxCy(200)β -WNxCy(111)
β -WNxCy(220)
400 °C
650 °C
β -WNxCy(311)
700 °C
(c)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
β -W2N(200)
β -W2N(220)
β -W2N(311)
β -WC1-x(200)
β -WC1-x(220)
β -WC1-x(311)
β -WC1-x(111)
β -W2N(111)
(d)
JCPDS 25-1257
JCPDS 20-1316
β -W2N(220)
β -WC1-x(220)
91
A) B)
Figure 6-2. Change in A) lattice parameter and B) average grain size with deposition
temperature for polycrystalline films deposited from 2. The estimates are based on
position and shape of diffraction peaks.
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
C
W
N
O
Figure 6-3. Variation of W, N, C, and O content in the films deposited from 2. Data are from
XPS measurements after 10 min Ar+ ion sputter.
4.11
4.13
4.15
4.17
4.19
4.21
4.23
4.25
450 500 550 600 650 700 750
Temperature (°C)
Latt
ice P
aram
ete
r (
Å)
β -WC1-x: 4.236
β -W2N: 4.126
(a)
0
10
20
30
40
50
60
70
80
450 500 550 600 650 700 750
Temperature (°C)
Averag
e G
rain
Siz
e (
Å)
(b)
92
A) B)
C) D)
Figure 6-4. Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature. Data are from XPS after 10 min Ar+ ion sputter.
28303234363840
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(a) W 4f
390392394396398400402
Binding Energy (eV)N
(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(b) N 1s
276278280282284286288
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(c) C 1s
524526528530532534536
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
(d) O 1s
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
93
Figure 6-5. SEM images of films grown on Si(100) substrate: A) cross-sectional view of film
grown at 300 °C; B) cross-sectional view of film grown at 700 °C; C) surface
morphology of film grown at 300 °C; D) surface morphology of film grown at
700 °C.
A) B)
WNxCy WNxCy Si Si
C) D)
94
Figure 6-6. Change in growth rate with deposition temperature for films deposited from 2.
Thickness measured by cross-sectional SEM.
0
2
4
6
8
10
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Film
Resis
tivit
y (
mΩ
-cm
)
Figure 6-7. Change in film resistivity (four-point probe) with deposition temperature for films
deposited from 2.
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
1000/T (K-1
)
ln G
(Å
/m
in)
700 600 500 400 300
Deposition Temperature (°C)
95
Figure 6-8. Cross-sectional TEM images of Cu/WNxCy/Si stacks: [A) and B)] before annealing
and [C) and D)] after annealing at 500 °C.
WNxCy
A) B)
Cu
C) D)
Si
WNxCy
Si
Cu
Cu WNxCy Si
Si
WNxCy
96
0 50 100 150
Depth (nm)
In
ten
sit
y (
a.u
.) Si Kα
Cu Kα
W Lα
Figure 6-9. EDS depth profile of Cu/WNxCy/Si stacks annealed at 500 °C.
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K β
Si(200)
As-grown
Cu(111)
500 °C
Cu(200)
No Cu3Si peaks
Cu(220)
Figure 6-10. The performance of diffusion barrier by XRD measurement for Cu/WNxCy/Si
stacks before and after annealing at 500 °C.
A)
B)
Cu WNxCy Si
97
CHAPTER 7
DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(NNPh2): EFFECT OF NH3 ON FILM
PROPERTIES
7.1 Film Structure
The XRD spectra in Figure 7-1 are used to identify the crystalline phases and to measure
the lattice parameter and polycrystalline grain size. The peak positions in the X-ray diffraction
patterns are well matched with a polycrystal mixture of β-W2N and β-WC1-x phases or the solid
solution β-WNxCy. The XRD spectra show that amorphous films were deposited from 300 to
450 °C, while the polycrystalline materials deposited at temperature from 500 to 700 °C. These
four observed characteristic peaks with primary peaks at 37.62 and 43.18 2θ° are consistent with
(111) and (200) orientations, respectively. Two additional reflections at 62.74 and 75.72 2θ° are
attributed to the (220) and (311) orientations, respectively. Comparing the observed relative
peak intensities to those of the standard powder diffraction intensities [Figure 7-1 D)] indicate
that no preferred crystal orientation (texture) exists. As the deposition temperature was increased
to 700 °C, the peak intensity increased. Film deposition with NH3, however, resulted in decrease
of intensity in XRD peaks as compared to film deposition without NH3.
7.2 Chemical Composition
The measured photoelectron intensities of XPS are used to identify unknown elements and
measure the atomic concentration. The XPS spectra in Figure 7-2 indicate that W, N, C and O
were present in the films. The AES spectra shown in Figure 7-3 were taken at the same time as
the XPS spectra. In AES, the ejected electrons are not the primary ionized electrons but the
secondary ionized electrons, which are produced by the decay of ionized atoms from exited
states to lower energy states [94]. Despite the presence of Cl in the precursor, no peaks were
observed for either Cl 2p3/2 or Cl 2s at 199 and 270 eV, respectively, ruling out Cl contamination
in the films within the detection limit of XPS (~ 1 at. %). The absence of Cl signals is consistent
98
with prior computational results on the related aryl- and alkylimido complex Cl4(CH3CN)W(NR),
(R = Ph, iPr, C3H5) for which a mechanistic pathway was found for reaction of the H2 carrier gas
with W-Cl bonds to produce volatile HCl in the gas phase [87]. Figure 7-3A shows that at lower
deposition temperatures (≤ 400 °C), the measured W levels in the films in the presence of NH3
are higher than for films deposited without NH3. Between 300 and 600 °C, the W levels are
between 47 and 51 at. %.
The N levels, shown in Figure 7-3B, increased over the temperature range 400 to 700 °C
after addition of NH3. The N levels have their highest value (19 at. %) for films deposited at
450 °C as compared with those of films grown without NH3 (14 at. %). As the deposition
temperature increases to 700 °C, the N levels gradually drop. For films deposited at 700 °C, N
levels without NH3 are 4 at. %, while for films deposited without NH3 the N level is 11 at. %. It
is postulated that the higher deposition temperature increases the rate of N desorption as N2 gas,
as shown in Figure 7-4B. C competes with N for bonding with W, and as shown in Figure 7-3C,
the C levels in films deposited with NH3 than are lower than those from the single-source only
deposition from 300 to 400 °C. The decrease in the C levels is due to increased competition
from N when NH3 is present. For films deposited at 400 °C, C levels without NH3 show 24 at. %,
while films deposited without NH3 shows 15 at. %. Ternary phase metal carbonitride barrier
materials show that the C levels lower the resistivity because W-C phase has lower resistivity
than W-N phase.
The Figure 7-3D shows that deposition with NH3 has lower O levels than deposition
without NH3 between 450 and 700 °C. As the deposition temperature increased, the O levels in
the presence and absence of NH3 decreased, with the O levels reaching at 5 at. % in films
deposited at 700 °C. The low O incorporation is consistent with a dense film microstructure [89].
99
Crystallization and thus denser films at higher temperature results in reduction of O adsorption
and reaction from air exposure post-growth.
7.3 Chemical Bonding States
The binding energies (BE) from XPS measurement are used to identify the chemical
bonding states of the elements in the films. This is accompanied by measuring the kinetic energy
of emitted elements and relating it to the binding energy according to:
bEhvE
where E is kinetic energy of the ionized electron, hv is incident radiation, and Eb is the binding
energy of the electron [94]. This method was used to investigate the chemical bonding states for
four elements in the films (W, N, C, and O). The photoelectron line of W 4f is a doublet due to
two spin-orbit states, 4f7/2 and 4f5/2, while the photoelectron lines of N 1s, C 1s, and O 1s are
singlets. The evolution of XPS patterns for W 4f BE with deposition temperature for films
deposited in the presence of NH3 is summarized in Figure 7-4A. The major W 4f7/2 and W 4f5/2
peaks are at 31.7 and 33.7 eV, which are close to the values for WCx and WNx, respectively,
over the entire deposition temperature range of this study (300 – 700 °C). The values for the W
4f7/2 peaks agree well with the reported ranges of 32.7 to 33.6 eV for WNx and 31.6 to 32.3 eV
for WCx. The values for W 4f5/2 peaks agree well with the reported ranges of 33.3 to 35.8 eV for
WNx and 33.7 to 33.9 eV for WCx [76, 78, 80, 90]. The evolution of XPS patterns shown in
Figure 7-4A indicates W bonding state is dominant in the physical mixture of β-WN2 and β–
WC1-x or β-WNxCy solid solution.
The evolution of XPS patterns for N 1s BE with deposition temperature for films deposited
in the presence of NH3 is summarized in Figure 7-4B. The major N 1s peak is at 397.3, which is
close to the values for WNx. The peak position of this BE remained constant over the deposition
100
temperature range of this study (300 – 700 °C) and the BE associated with the N at the grain
boundary at 400.0 eV is absent. The value for N 1s peak agrees well with the reported range of
396.2 to 398.2 eV for WNx [76, 78, 80, 90]. The XPS pattern shown in Figure 7-4B indicates
that the N in the film is bound to W in the WNx. Only a single N 1s peak is located at near 397.3
eV without a second N 1s peak near 400.0 eV.
The evolution of XPS patterns for the C 1s BE with deposition temperature for films
deposited in the presence of NH3 is summarized in Figure 7-4C. The major C 1s peak observed
for T ≤ 600 °C is at 283.3 eV, which is close to the value for WCx. This value for the C 1s peak
agrees well with the reported range of 279.7 to 283.8 eV for WCx [75, 76, 83, 84, 90]. For films
deposited above 600 °C, the bonding state of C 1s is shifted from lower to higher BE. The
higher BE value of the C 1s peak is located at 284.7 eV, which is close to the value for
amorphous C of 284.2 to 285.2 eV for WCx [75, 76, 83, 84, 90]. This peak shift indicates that
WCx in the WNxCy nanocrystals coexists with amorphous C.
The evolution of XPS patterns for O 1s BE with deposition temperature for films deposited
in the presence of NH3 is summarized in Figure 7-4D. The major O 1s peak is at 540.4 eV,
which is close to the value for WO3. This value remained extra over the entire temperature range
of this study (300 – 700 °C). The value for O 1s peaks agree well with the reported range of
528.2 to 531.6 eV for WO3 [33, 76, 78, 90]. As deposition temperature increases to 700 °C, the
peak intensity of O 1s is decreases due to film crystallization and C incorporation.
7.4 Surface Morphology
The root-mean-square (rms) surface roughness of the film deposited at 300 °C without
NH3 was determined by AFM to be 5.0 nm, while that of films deposited at 700 °C with NH3
was 87.4 nm (Figures 7-5A and 7-5B). The surface roughness shown in Figures 7-5C and 7-5D
shows the value of rms roughness was 1.1 nm at 300 °C and 5.7 nm at 700 °C, indicating the
101
addition of NH3 results in films with smoother surfaces. The increase in surface roughness is
accompanied by increased crystallinity in film microstructure as the deposition temperature
increases to 700 °C. The decrease in roughness is due to an amorphous microstructure and more
facile migration of absorbed species on the surface [86].
7.5. Film Growth Rate
Cross-sectional SEM images as exemplified in Figure 7-6 were used to measure the film
thickness. The growth rate in the presence of NH3 was low, in the range 7.3 to 14.3 Å /min as
compared to the range 1.0 to 25.4 Å /min for films deposited in the absence of NH3. The
Arrhenius plot in the presence of NH3 reveals one growth regime while the plot in the absence of
NH3 reveals a transition from a kinetically controlled growth regime to a mass transfer controlled
one. These differences in growth rate and transition in growth regime are consistent with a
change in deposition mechanism due to the addition of NH3. The difference in growth rate
indicates a shift in deposition mechanism due to the addition of NH3. Transamination with NH3
has been postulated to remove the hydrocarbon group in 1, changing the rate-determining step of
this study. For films deposited at 700 °C using a single source, the growth rate increased
drastically suggesting a change in the growth mechanism with increasing temperatures. This
observation was also confirmed by the formation of WCx, indicating most of C exists with C-C
bonding states with small portion of W-C bonding states.
7.6 Electrical Resistivity
The film resistivity was determined from the measured sheet resistance (four-point probe)
and films thickness (cross-sectional SEM). The effect of growth temperature on the film
resistivity for films deposited with NH3 is shown in Figure 7-8. Films deposited at 400 °C show
the lowest film resistivity (1.9 mΩ-cm) and the values of film resistivity vary with the interplay
of grain size, film microstructure, film density, metal to non-metal ratio, and film thickness. At
102
lower deposition temperature, increase in N levels with W-N bonding states result in decrease in
film resistivity. However, after addition of NH3, a decrease in C levels with W-C bonding states
causes an increase in the film resistivity. Also, over 450 °C, the film microstructure changes
from amorphous to polycrystalline, which causes an increase in film resistivity. W-C bonding is
an important factor in decreasing film resistivity as the both sensitivity of WCx is considerably
lower than WNx. Hence, the proper combination of W-N and W-C bonding states is significant
in formation of ternary-based metal nitrides for diffusion barrier applications.
7.7 Diffusion Barrier Testing
Diffusion barrier testing was performed to evaluate the performance for Cu interconnects
on Si. Cross-sectional TEM images and the EDS depth profile were used to observe Cu/WNxCy
and WNxCy/Si interfaces after annealing. The TEM images shown in Figure 7-9A reveal that
there is no Cu diffusion through WNxCy after annealing under N2 at 500 °C for 30 min. Both
Cu/WNxCy and WNxCy/Si interfaces are clearly observed without any evidence of Cu transport
and intermixing between the layers. The EDS depth profile shown in Figure 7-9B indicates that
the Cu Kα signal decreases sharply at the Cu/WNxCy interface, demonstrating that there is no Cu
diffusion observed either before or after annealing under N2 at 500 °C for 30 min. Even if a
small trace of Cu transported to the Cu/WNxCy interface, the Cu Kα signal was not detected
when the scan moved into the Si substrate. Cross-sectional TEM images and the EDS depth
profiles reveal no onset of failure in Cu/WNxCy/Si stacks.
XRD was also employed to identify the phase of Cu-related textures before and after
annealing Cu/WNxCy/Si stacks (Figure 7-10). Only the Cu peak at 43.46 2θ° for the (111)
orientation was observed for the as-deposited sample. After annealing at 500 °C, three Cu peaks
at 43.46, 50.96, and 74.60 2θ° were observed, for the (111), (200), and (222) orientations,
respectively. As the annealing temperature increases to 600 °C, the gradual increase in the
103
intensity of Cu peaks indicates grain growth of Cu, evidenced by the decrease in value of full
width half maximum (FWHM). However, after annealing at 700 °C, the appearance of new Cu-
related peaks at 37.14 and 57.24 2θ° indicates the formation of Cu3Si. The decreasing intensity
of three Cu peaks and the decreasing thickness of Cu films were due to the diffusion of Cu in Si.
Four-point probe was employed to measure the change in sheet resistance before and after
annealing Cu/WNxCy/Si stacks (Figure 7-11). As the annealing temperature increases to 600 °C,
the decrease in the sheet resistance indicates enlargement of Cu grains. The increase in grain
size reduces the Cu grain boundaries, which contributes to the decrease in resistivity due to the
lower electron scattering. However, after annealing at 700 °C, the rapid increase in the sheet
resistance is consistent with the formation of Cu3Si. The cross-sectional TEM images shown in
Figure 7-12 indicate Cu/WNxCy and WNxCy/Si interfaces were clearly observed without Cu
transport and intermixing in the layers before annealing, whereas the Cu3Si crystallite exists in
WNxCy/Si interface after annealing at 700 °C. The SEM images shown in Figure 7-13B indicate
a deterioration of Cu surface morphology after annealing at 700 °C. The color change in the
surface from reddish yellow to dark grey indicates a decrease of thickness in Cu layer, the
increase in surface roughness of Cu, and the formation of Cu3Si, which are all indications of the
transport and intermixing of Cu in Si.
7.8 Conclusions
The tungsten diphenylhydrazido complex Cl4(CH3CN)W(NNPh2) (2) was used to deposit
WNxCy with NH3 coreactant to investigate the effect of this coreactant and the onset of failure
process on the film properties for diffusion barrier applications. The N levels in the films in the
presence of NH3 were higher than those in the absence of NH3. The result shown in the XRD
patterns suggests that film microstructure was amorphous for films deposited at a lower
deposition temperature (below 450 °C). The XPS W 4f bonding state indicates that most of the
104
W is present in the carbide and nitride mixture or a WNxCy single solid solution. The dominant
W bonding state is WNxCy rather than WO3 from 300 to 700 °C. An investigation of the XPS N
1s bonding state indicates that N is present as the nitride. XPS spectra show the highest N levels
for films deposited at 450 °C. An examination of the XPS C 1s peak indicates that C is present
as the carbide. However, for films deposited over 600 °C, the BE in C 1s shifted from the lower
energy to higher energy, indicating that the W-C phase coexists with a C-C phase. An
observation of XPS O 1s indicates that O is present as WO3 or O in the WNxCy. XPS spectra
show lower O incorporation at higher temperature, which produces films with higher density.
AFM micrographs indicate that addition of NH3 causes deposition of films with smoother
surface as compared to those from single-source deposition. The growth rate with added NH3
varied in the range 7.3 to 14.3 Å /min over the entire deposition temperature of study. A large
variation of film resistivity is due to the interplay of various reasons such as grain size, film
microstructure, film density, metal to non-metal ratio, and film thickness. Films deposited in the
absence of NH3 have lower film resistivity than that of films deposited in the presence of NH3.
Optimal combination of WNx and WCx phase is important in formation of ternary phase
materials for diffusion barrier applications. The diffusion barrier test results support the
conclusion that WNxCy deposited from 2 is a viable Cu diffusion barrier material for Cu
interconnect technology.
105
A) B)
C) D)
Figure 7-1. XRD spectra for films deposited on Si(100) with NH3: A) 300 °C, B) 700 °C, C)
change in XRD spectra, and D) standard diffraction plots for β-W2N and β-WC1-x.
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
300 °C
(a)
Si(400) K β
Si(200)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K β
700 °C
Si(200)
(b)
β -WNxCy(111)
β -WNxCy(220)
β -WNxCy(311)
β -WNxCy(200)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
300 °C
Si(400)
Si(400) KβSi(200)
350 °C
450 °C
500 °C
550 °C
600 °C
β -WNxCy(200)β -WNxCy(111)β -WNxCy(220)
400 °C
650 °C
β -WNxCy(311)
700 °C
(c)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
β -W2N(200)
β -W2N(220)
β -W2N(311)
β -WC1-x(200)
β -WC1-x(220)
β -WC1-x(311)
β -WC1-x(111)
β -W2N(111)
(d)
JCPDS 25-1257
JCPDS 20-1316
β -W2N(220)
β -WC1-x(220)
106
-10001002003004005006007008009001000
Binding Energy (eV)
In
ten
sit
y (
a.u
.)
300 °C
350 °C
450 °C
500 °C
550 °C
600 °C
400 °C
650 °C
700 °C
No Cl peaks
W 4fO 1s N 1s C 1sO KVV
C KVV
N KVVW 4dW 4p
Figure 7-2. XPS spectra for films deposited on Si(100) with NH3. Note that Cl peaks are
evident as a function of growth temperature.
107
A) B)
C) D)
Figure 7-3. Comparison of W, N, C, and O content in the films deposited in the presence and
absence of NH3. Data are measured by XPS after 10 min Ar+ ions sputter.
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
(a) W With NH3
Without NH3
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
(b) N With NH3
Without NH3
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
(c) C With NH3
Without NH3
0
10
20
30
40
50
60
70
80
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
(d) O With NH3
Without NH3
108
A) B)
C) D)
Figure 7-4. Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature. Data are from XPS after 10 min Ar+ ions sputter.
28303234363840
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(a) W 4f
390392394396398400402
Binding Energy (eV)N
(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(b) N 1s
276278280282284286288
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(c) C 1s
524526528530532534536
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
(d) O 1s
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
109
Figure 7-5. Surface morphology of films grown on Si(100) substrate: A) film grown at 300 °C
without NH3; B) film grown at 700 °C without NH3; C) film grown at 300 °C with
NH3; D) film grown at 700 °C with NH3.
A) B)
C) D)
110
Figure 7-6. SEM images of films grown on Si(100) substrate: A) cross-sectional view of film
grown at 300 °C; B) cross-sectional view of film grown at 700 °C.
Figure 7-7. Change in growth rate with deposition temperature for the films deposited in the
presence and absence of NH3. Thickness was measured by cross-sectional SEM.
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
1000/T (K-1
)
ln G
(Å
/m
in)
700 600 500 400 300
Deposition Temperature (°C)
With NH3
Without NH3
A) B)
WNxCy Si Si WNxCy
111
0
2
4
6
8
10
12
14
16
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Film
Resis
tivit
y (
mΩ
-cm
)
With NH3
Without NH3
Figure 7-8. Film resistivity as a function of deposition temperature for the films deposited in the
presence and absence of NH3.
A) B)
0 50 100
Depth (nm)
In
ten
sit
y (
a.u
.) Si Kα
Cu Kα
W Lα
Figure 7-9. A) TEM image and B) EDS depth profile of a Cu/WNxCy/Si stack annealed at
500 °C for 30 min.
Si
WNxCy
Cu
112
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K βSi(200)
As-grown
Cu(111)
500 °C
Cu(200)
700 °C
600 °C
Cu3Si Cu3Si W5Si3 Cu(220)
Figure 7-10. Change in XRD patterns with annealing temperature for Cu/WNxCy/Si stacks.
1.0E+02
1.0E+03
1.0E+04
1.0E+05
1.0E+06
0 100 200 300 400 500 600 700 800
Annealing Temperature (°C)
Sh
eet
Resis
tan
ce (
log
mΩ
/sq
uare)
Figure 7-11. Change in sheet resistance with annealing temperature for Cu/WNxCy/Si stacks.
Data are measured by four-point probe.
113
Figure 7-12. Cross-sectional TEM images of Cu/WNxCy/Si stacks: A) as-grown and B) after
annealing at 700 °C.
Figure 7-13. Cross-sectional SEM images of Cu/WNxCy/Si stacks: A) as-grown and B) after
annealing at 700 °C.
A) B)
WNxCy
Cu
Si Cu
Cu3Si
Si WNxCy
SiO2
A) B)
114
CHAPTER 8
DEPOSITION OF WNxCy FROM Cl4(CH3CN)W(NNMe2): EFFECT OF NH3 ON FILM
PROPERTIES
8.1 Film Structure
The progression of XRD patterns in Figure 8-1C are used to identify the crystalline phase
for films deposited in the presence of NH3 in H2 carrier with increasing deposition temperature.
The XRD pattern of the film deposited below 500 °C indicates that it is X-ray amorphous, while
the films deposited at higher deposition temperature (≥ 500 °C) are polycrystalline. The XRD
patterns shown in Figure 8-1C have been compressed to include the results from all nine growth
runs of this study, and thus the resolution is decreased in the figure. An analysis of original data,
however, reveals four reflections at and over 500 °C, which were calibrated to the Si(400)
diffraction peak. The primary peaks at 37.50 and 42.92 2θ° show relatively high intensity, as
compared with the reflections at 61.58 and 75.40 2θ°. All four peaks are between the standard
diffractions of β-W2N and β-WC1-x. Both standards exhibited a face-centered cubic (fcc)
structure with similar values of lattice parameter (β-W2N: 4.124 Å and β-WC1-x: 4.236 Å). XRD
results suggest the existence of a two-phase mixture (β-W2N or β-WC1-x phases) or the presence
of their solid solution (WNxCy). The crystallinity of films deposited in the absence of NH3
increases with deposition temperature [18]. However, the films deposited in the presence of NH3
show a different trend, suggesting that NH3 coreactant in H2 carrier gas alters the growth
mechanism of CVD [95]. A transamination reaction with NH3 has been postulated to remove the
allyl substitute on the imido group in the precursor, changing the rate-determining step [85].
8.2 Chemical Composition
The measured photoelectron intensities of XPS were used to identify elements present in
the films and measure their atomic concentration. XPS spectra in Figure 8-2 indicate that W, N,
115
C, and O were identified in the films. No Cl contamination in the films was observed within the
detection limit of XPS (~ 1 at. %). Figure 8-3 shows the variation in the chemical composition
of W, N, C, and O contents in the films with deposition temperature. The W level is constant (50
at. %) for runs between 300 and 600 °C. Over 600 °C, the W level drops gradually because
amorphous C starts to coexist with W-C. As the deposition temperature increased from 300 to
450 °C, the N level increased from 24 to 27 at %. N levels above 450 °C, however, start to
decrease, as a consequence of increased C concentration in this range. When the deposition
temperature reaches 700 °C, the N level has declined to 13 at. % due to the steep rise in C levels
at high deposition temperatures. It has been suggested that the higher deposition temperature
increases the rate of N desorption as N2 gas [29, 33, 38, 72]. Typical refractory metal nitride
diffusion barriers show the decreasing tendency of N level with increasing deposition
temperature because higher thermal energy in the lattice structure of the film comes from a
higher temperature. From 300 to 450 °C, the C level is below 10 at. %, with the lowest level of 7
at. % for film growth at 300 °C. As deposition temperature increase up to 550 °C, the C level
increases gradually, while between 600 and 700 °C, the C level increases drastically from 19 to
46 at. %. The overall trend for C content is consistent with the faster decomposition of C-H
groups in both the precursor and the solvent as the growth temperature increases, leading to C
incorporation into the film. This is well matched with the pyrolysis of PhCN around 600 °C.
Films deposited at 300 C° show 30 at. % of O, which decreased drastically to 10 at. % at 500 °C.
As the deposition temperature increased from 500 to 700 °C, the O level decreased gradually to 4
at. %. From XRD spectra in Figure 8-1C, the polycrystalline microstructure becomes evident for
depositions performed at 500 °C. As the film starts to crystallize, the film microstructure gets
116
denser by polycrystal grain growth, which prevent interdiffusion of O into the lattice of the film
after film growth [73].
8.3 Chemical Bonding States
The binding energy (BE) of XPS are used to identify the chemical bonding states of the
elements present from any variation in the determined BE from measurement of kinetic energy
(KE) emitted from the elements:
bEhvE
where E is kinetic energy of the ionized electron, hv is incident radiation, and Eb is the
binding energy of the electron [94]. They were used to investigate the chemical bonding states
for four atoms in the films. The photoelectron line of W 4f is a doublet due to two spin-orbit
states, 4f7/2 and 4f5/2, while the photoelectron lines of N 1s, C 1s, and O 1s are singlets.
Figure 8-4A displays the evolution of XPS patterns in BE of W 4f as deposition
temperature increases. The major W 4f7/2 and W 4f5/2 peaks are at 31.4 and 33.8 eV, which are
close to WNx and WCx in the temperature range of this study (300 - 700 °C). These values for
W 4f7/2 and W 4f5/2 peaks agree well with the reported range of 32.7 to 33.6 eV and 33.3 to 35.8
eV for WNx [76, 78-80, 90] while these values for W 4f7/2 and W 4f5/2 peaks agree well with the
reported range of 31.6 to 32.3 eV and 33.7 to 33.9 eV for WCx [75, 76, 82, 90], respectively.
The major W 4f7/2 and W 4f5/2 peaks correspond to WNx and WCx. These results indicate that a
chemical bonding state in W is the mixture of β-WN2 and β-WC1-x or one single solution of β-
WNxCy.
XPS patterns for N 1s BE are shown in Figure 8-4B over the range of deposition
temperature for films deposited with NH3. This value for the N 1s peak agrees well with the
reported range of 396.2 to 398.2 eV for WNx [76, 78-80, 90]. The major N1s peak is at 397.3 to
117
397.5 eV, which agrees well with WNx in the temperature range of this study (300 – 700 °C).
All N in the films is bound in the WNx. A single N 1s peak shows the metal nitride bonding state,
regardless of the other contents in the films. The results shown in Figure 8-4B suggest that N at
the grain boundary can be ruled out due to a single N 1s peak without a second peak near 399 eV.
Films deposited at 450 °C have the highest intensity of N, indicating the highest N levels in the
films, as shown in Figure 8-3.
The evolution of XPS patterns for the C 1s BE with deposition temperature for films in the
presence of NH3 is summarized in Figure 8-4C. Up to 600 °C, the BE of the C 1s peak located at
283.0 – 283.3 eV agrees well with WCx. Deconvolution of the broad C 1s peak for films
deposited from 650 to 700 °C using Gaussian-Lorentzian function with background subtraction
yields two separate peaks (W-C and amorphous C). The BE of C 1s peak located at 284.4 eV
agrees well with an amorphous C phase present outside of the β-WNxCy nanocrystals, while the
BE of C 1s peak located at 283.7 eV agrees well with a W-C phase in the β-WNxCy. The former
value for C 1s peak agrees well with the reported range of 284.2 to 285.2 eV for amorphous C,
while the latter value for C 1s peak agrees well with the reported range of 279.7 to 283.8 eV for
WCx [75, 76, 82-84, 90].
O levels were also probed by XPS in the temperature range of this study (300 - 700 °C).
The major O 1s peak is at 530.2 – 530.3 eV, which is close to WO3. This value for the O 1s peak
agrees well with the reported range of 528.2 to 531.6 eV in WO3 [33, 76, 78, 81, 90]. As
evidenced by Figure 8-4D, the peak intensity of O 1s decreased with deposition temperature
increases to 700 °C. O levels are lower in films grown at high temperature due to great extent of
film crystallization and C incorporation.
118
8.4. Surface Morphology
The root-mean-square (rms) roughness of the film surface was determined by AFM to be 1.23
nm for films deposited at 300 °C in the presence of NH3, with an increase up to 3.47 nm for
deposition at 700 °C (Figures 8-5A and 8-5B). As the deposition temperature increases up to
700 °C, the increase in surface roughness is accompanied by the increase in crystallinity, while
the decrease in surface roughness is consistent with an amorphous microstructure (Figure 8-1).
The AFM micrographs indicate that films with smoother surface are due to deposition at lower
temperature and addition of NH3.
8.5 Film Growth Rate
The growth rate is in the range 1.6 to 32.0 Å /min, as determined by cross-sectional SEM
(Figure 8-6). For films deposited at 700 °C, the growth rate increased drastically suggesting a
change in the growth mechanism at these temperatures. Figure 8-7 is consistent with the
presence of two growth regimes below 700 °C. The region with the shallow slope is a mass
transfer limited growth regime between 400 °C and 650 °C. The region with the steep slope is a
kinetically controlled growth regime between 300 °C and 400 °C. The apparent activation
energy calculated for the activated process is 0.31 eV.
8.6 Electrical Resistivity
The variation of film resistivity with deposition temperature is shown in Figure 8-8. The
lowest resistivity is 3.7 mΩ-cm at 300 °C and the highest film resistivity is 19.4 mΩ-cm for films
deposited at 700 °C. The values of film resistivity fluctuate with the interplay of polycrystal
grain growth, C content, O content, and film thickness in the temperature range of this study.
The high N level in those films is consistent with increased film resistivity in the β-WNxCy
polycrystal structures, due to the higher resistivity for β-W2N relative to β-WC1-x. As shown in
119
Figure 8-4, an increase in the amorphous C level as the deposition temperature rises from 650 to
700 °C results in an increase in electron scattering, which causes the film resistivity to increase.
8.7 Conclusions
The tungsten dimethylhydrazido complex Cl4(CH3CN)W(NNMe2) (3) was used to deposit
WNxCy with NH3 to investigate the effect of NH3 on the film properties for diffusion barrier
applications. The deposited films show higher N levels with lower C incorporation as compared
to films deposited without NH3. XRD results suggest that films deposited below 500 °C were X-
ray amorphous with crystallinity evolving at higher deposition temperature. The XPS W 4f
bonding state indicates that most of the W is present as a mixture of WNx and WCx phases or a
WNxCy single solid solution. XPS results for the W indicates WNxCy is the dominant W phase
in the temperature range of this study. XPS spectra of the O 1s bonding state show low O
incorporation at higher temperature, which produces films with higher density. An examination
of the XPS N 1s bonding state indicates that N is present in the WNx phase. XPS spectra show
films deposited at 450 °C have the highest N levels. The XPS C 1s bonding state results suggest
that C is present as WCx and amorphous C. The C 1s BE is shifted from lower energy (283.1
eV) to higher energy (284.5 eV) for films deposited at 700 °C, indicating that amorphous C
coexists with WCx. XPS observation of the O 1s bonding state indicates that O is present as
WO3. XPS spectra also show lower O incorporation at higher temperature, which produces films
with higher density. The film growth rate with NH3 addition varied in the range 1.6 to 32 Å /min
in the temperature range of 300 to 700 °C. The values of film resistivity fluctuates with the
interplay of polycrystal grain growth, C content, O content, and film thickness in the temperature
range of this study. The high N level in those films is consistent with increased film resistivity in
the β-WNxCy polycrystal structures, due to the higher resistivity for β-W2N relative to β-WC1-x.
Film resistivity varied in the range 3.7 mΩ-cm (300 °C) to 19.4 mΩ-cm (700 °C).
120
A) B)
C) D)
Figure 8-1. XRD spectra for films deposited on Si(100) with NH3: A) 300 °C; B) 400 °C; C)
change in XRD spectra; D) standard powder diffraction pattern for β-W2N and β-
WC1-x.
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K β
Si(200)
300 °C
(a)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
Si(400)
Si(400) K β
700 °C
β -WNxCy(111)
β -WNxCy(220)
β -WNxCy(311)
Si(200)
β -WNxCy(200)
(b)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
300 °C
Si(400)
Si(400) KβSi(200)
350 °C
450 °C
500 °C
550 °C
600 °C
β -WNxCy(200)β -WNxCy(111) β -WNxCy(220)
400 °C
650 °C
β -WNxCy(311)
700 °C
(c)
30 35 40 45 50 55 60 65 70 75 80
2θ Degrees
In
ten
sit
y (
a.u
.)
β -W2N(200)
β -W2N(220)
β -W2N(311)
β -WC1-x(200)
β -WC1-x(220)
β -WC1-x(311)
β -WC1-x(111)
β -W2N(111)
(d)
JCPDS 25-1257
JCPDS 20-1316
β -W2N(220)
β -WC1-x(220)
121
-10001002003004005006007008009001000
Binding Energy (eV)
In
ten
sit
y (
a.u
.)
W 4p
300 °C
350 °C
450 °C
500 °C
550 °C
600 °C
400 °C
650 °C
700 °C
No Cl peaks
W 4fO 1s N 1s C 1sO KVV
C KVV
N KVV
W 4d
Figure 8-2. XPS spectra for films deposited on Si(100) with NH3. No Cl peaks detected.
0
10
20
30
40
50
60
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Con
cen
trati
on
(A
tom
ic %
)
C
W
O
N
Figure 8-3. Variation in the chemical composition of W, N, C, and O contents in the films with
deposition temperature. Data are measured by XPS after 10 min Ar+ ion sputter.
122
A) B)
C) D)
Figure 8-4. Change of binding energies in A) W 4f, B) N 1s, C) C 1s, and D) O 1s with
deposition temperature. Data are from XPS after 10 min Ar+ ion sputter.
28303234363840
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(a) W 4f
390392394396398400402
Binding Energy (eV)N
(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(b) N 1s
276278280282284286288
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
(c) C 1s
524526528530532534536
Binding Energy (eV)
N(E) (
a.u
.)
450 °C
(d) O 1s
500 °C
550 °C
600 °C
650 °C
400 °C
350 °C
300 °C
700 °C
123
Figure 8-5. Surface morphology of films grown on Si(100) substrate: A) film grown at 300 °C
with NH3; B) film grown at 700 °C with NH3.
Figure 8-6. SEM images of films grown on Si(100) substrate: A) cross-sectional view of film
grown at 300 °C; B) cross-sectional view of film grown at 700 °C.
A) B)
A) B)
WNxCy WNxCy Si Si
124
Figure 8-7. Change in growth rate with deposition temperature for the films deposited from 3.
Thickness was measured by cross-sectional SEM.
0
2
4
6
8
10
12
14
16
18
20
250 300 350 400 450 500 550 600 650 700 750
Temperature (°C)
Film
Resis
tivit
y (
mΩ
-cm
)
Figure 8-8. Change in film resistivity (four-point probe) with deposition temperature for the
films deposited from 3.
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
1000/T (K-1
)
ln G
(Å
/m
in)
700 600 500 400 300
Deposition Temperature (°C)
125
CHAPTER 9
REACTOR MODELING USING CFD SOFTWARE
9.1 Description of the Raman-Assisted CVD reactor
To better understand the decomposition mechanism of the tungsten dimethylhydrazido
complex Cl4(CH3CN)W(NNM2) (3), a set of preliminary experiment was performed in the probe
CVD reactor shown in Fig. 9-1. This unique system is interfaced with an in-situ Raman
spectrometer (Ramanor U-1000, Jobin Yvon), which includes double additive monochromator
and uses the 532.08 nm line of Nd:YAG solid-state laser as the light source. As described in
detail elsewhere [96, 97], this CVD reactor is an up-flow, impinging-jet, cold-wall reactor that
was custom-built to quantitative study of the gas phase decomposition kinetics. The inlet to the
reactor consists of three concentric tubes – center, annulus and sweep flows. Each inlet line is
packed with 3 mm glass beads to provide parallel flow inlet boundary condition.
Complex 3, which was tested as a metal-organic precursor for tungsten-based diffusion
barrier material, is introduced through a center line and N2 carrier gas was input into the outer
tow lines to prevent wall deposition. Based on the previous study [96] that developed and
validated a steady-state, two-dimensional mass transport model using results from a CH4 tracer
experiment, conditions are selected such that recirculation flow was not present in the reactor.
In a preliminary experiment, aerosol-assisted CVD (AACVD) for complex 3 was adopted
because this technique has less strict volatility limitation in selecting precursors. The solid
precursor 3 was dissolved in benzonitrile (PhCN) to the concentration 7.4 mg/mL (0.0174
mol/L) and then pumped into a nebulizer from a syringe. A piezoelectric material in the
nebulizer vibrates at a frequency of 1.44 MHz, which generates a mist of precursor 3 and PhCN,
and the mist is then transported to the reactor with a carrier gas. The N2 (99.999%, Airgas)
carrier gas flow velocity was 0.025 m/s, and the mixture of precursor and solvent was injected at
126
a rate of 0.5 ml/h. Pure N2 gas was also delivered to the annulus and sweep lines after same flow
velocity (0.025 m/s) and sufficient time was allowed to reach steady-state before measurements
were taken. Then the 1W Nd:YAG solid-state laser line (532.08 nm) was used to excite complex
3 and several vibrational Raman excitation lines were observed.
9.2 Multiphase Flow Simulation of the Raman-Assisted CVD reactor
Simulations on flow and thermal patterns in the reactor were performed using FLUENT™
computational fluid dynamics (CFD) packages. Equations of conservation for mass, momentum,
and energy were solved with geometry and boundary conditions specific to this reactor geometry.
The conservation of mass can be written as follows.
)( vt
(9-1)
This equation describes the time rate of change of the fluid density at a fixed point in the
space. The vector v is the mass flux, and its divergence is the net rate of mass efflux per unit
volume.
The conservation of momentum can be written as:
gpvvvt
][][ (9-2)
This equation describes the rate of increase of momentum per unit volume. The term
][ vv is the rate of momentum addition by conservation per unit volume. The two terms
][ p are the rate of momentum addition by molecular transport per unit volume. The
term g is the external force of gravity on the fluid per unit volume. The conservation equation
for momentum is equivalent to Newton‟s second law of motion: the statement of mass x
acceleration = sum of forces.
The conservation of energy can be written as:
127
)():(])[()()()2
1()
2
1( 22 gvvvvppvv
t
(9-3)
This equation describes the rate of increase of kinetic energy per unit volume. Two major
energy terms commonly used in computational fluid mechanics are kinetic energy associated
with observable fluid motions of the molecules, plus the energy of interaction between molecules.
The term )2
1( 2v is the rate of addition of kinetic energy by convection per unit volume.
The term )( pv is the rate of work done by pressure of the surroundings on the fluid. The
term )( vp is the rate of reversible conversion of kinetic energy into internal energy. The
term ])[( v is the rate of work done by viscous forces on the fluid. The term ):( v is
the rate of irreversible conversion from kinetic to internal energy. The term )( gv is the rate of
work by gravity on the fluid.
FLUENT™ uses a Finite Volume (FV) method to convert the governing equations to
algebraic equations. Algebraic equations can be solved numerically in order to solve these
coupled conservation partial differential equations (PDE). The solution domain is subdivided
into a finite number of contiguous control volumes (CV). Then, the conservation equations are
applied to each CV. This CV technique has two parts for integrating the governing equations
about each CV and yielding discrete equations to conserve each quantity on a CV basis.
The mesh was generated using the GAMBIT™ (version 2.2.30) with cylindrical
coordinates in two-dimensional format. An unstructured quadruple grid was employed. Three
types of boundary conditions were assigned during the mesh design step: inlet flow velocity at
reactor inlet, outflow type at reactor outlet, and surface wall temperature at heater surface. The
segregator solver was used, where the governing equations (momentum, continuity, and scalar)
128
are solved sequentially rather than in a simultaneous way. Note that “sequentially” means
“segregated from one another”.
The simulation of the behavior of the flow and thermal pattern in the reactor was
performed using FLUENT™ (version 6.2.16). The operating pressure was fixed at 101325 Pa,
that is an atmospheric pressure growth condition for the CVD reactor, and the gravitational
acceleration was turned on to the minus Y axis direction (- 9.8 m/s). Boundary conditions used
in this simulation are summarized in the following table (Table 9-1). The inlet velocity was
0.025 m/s, while the outlet boundary condition uses a typical outlet pressure of reactor, which is
760 Torr. The temperature at the heater surface is set at 1200 K and there is no heat flux through
the side-walls of internal reactor. Figures 9-3 and 9-4 show the calculated contours of static
temperature and velocity magnitude from the CFD simulation. Figure 9-5 shows the calculated
contours of velocity magnitude and volume fraction of the secondary phase in the multiphase
flow model.
129
Table 9-1. Boundary conditions for CVD reactor
Boundary location Boundary type Specific condition
Inlet flow Velocity inlet 0.025 m/s
Outlet flow Outlet pressure 760 Torr
Heater temperature Wall 1200 K
Reactor wall Wall No heat flux
130
Vent Vent
C eramic cap
Heater
Quartz wall
S ubstrate
S tainless screen
Quartz ball
S weep inlet
Annulus inlet
C enter flow
Annulus flow
S weep flow
L AS E R
C arrier gas
M.O. source
Nebulizer
r
z
Vent Vent
C eramic cap
Heater
Quartz wall
S ubstrate
S tainless screen
Quartz ball
S weep inlet
Annulus inlet
C enter flow
Annulus flow
S weep flow
L AS E R
C arrier gas
M.O. source
Nebulizer
r
z
1,200 KHeater wall
No heat fluxReactor wall
760 TorrOutlet pressure
0.025 m/sVelocity inlet
SpecificationsBoundary Type
1,200 KHeater wall
No heat fluxReactor wall
760 TorrOutlet pressure
0.025 m/sVelocity inlet
SpecificationsBoundary Type
Figure 9-1. Schematic photographs of A) CVD reactor system that is interfaced to the Raman
spectrometry ; B) nebulizer system; C) the impinging jet probe reactor.
A) B) C)
131
Figure 9-2. Mesh design of CVD reactor using GAMBIT™.
132
Figure 9-3. Color filled contours of static temperature (K) and contour line of static temperature
(K) in the vicinity of the heater.
133
Figure 9-4. Contours of velocity magnitude (m/s) and velocity vector colored by velocity
magnitude (m/s) in the vicinity of the heater.
134
A) B)
C) D) Figure 9-5. Contours of velocity magnitude (m/s) and volume fraction of solvent phase in
multiphase flow model.
135
E) F)
G) H) Figure 9-5. Continued
136
I) J)
K) L) Figure 9-5. Continued
137
CHAPTER 10
CONCLUSIONS AND FUTURE WORK
Films grown by CVD using 1, 2, and 3 were used to investigate the film properties and
diffusion barrier quality. The results detailed in chapters 3 – 8 indicate that tungsten-based films
have many positive properties important to diffusion barrier application. First, the work
demonstrates that WNxCy is an effective Cu diffusion barrier material to prevent the transport of
Cu into Si. Films deposited at 400 °C using 1, 2, or 3 are able to prevent Cu interdiffusion after
annealing at 500 °C for 30 min under N2 atmosphere. In particular, samples annealed at higher
temperature using 3 showed evidence of failure only when annealed at 700 °C. Second, each
precursor 1, 2, and 3 yielded film growth at temperatures as low as 300 °C, indicating that facile
precursor decomposition pathway and aerosol-assisted metal-organic CVD can be used at an
acceptable deposition temperature (< 400 °C) for diffusion barrier applications. Third, WNxCy
promotes the PVD growth of Cu with the preferred (111) orientation on a WNxCy/Si stack. It is
noted that for metallization applications, the Cu(111) texture is preferred since it shows a higher
resistance to electromigration. Fourth, addition of C to WNx causes to lower the film resistivity
because WCx phase has a lower resistivity than WNx. It was also found that incorporation of
NH3 in the gas stream results in the deposition of higher resistivity films due to the greater
incorporation extent of N. Finally, WNxCy films show good adhesion to Cu, indicating the
Cu/WNxCy/Si stack is thermally and mechanically stable after annealing at 500 °C for 30 min.
10.1 Ru-WNxCy for Diffusion Barrier and Cu Direct-Plate Applications
Several groups have attempted the growth of bilayer direct plate liner/diffusion barrier
materials for Cu integration without the need for a Cu seed layer. A mixed phase Ru-WNxCy
deposited by aerosol-assisted metal-organic CVD (or metal-organic ALD) is proposed as a novel
direct-plate liner for advanced Cu metallization [66]. From present study, it is know that WNxCy
138
diffusion barriers an effective Cu barriers. Based on the combination of known barrier properties
and electrochemical properties, indicating a higher affinity to the direct-plate process, a tungsten-
based diffusion barrier can be selected as a candidate material to combine with Ru for evaluation
as an extendible direct-plate liner technology. The Ru-WNxCy mixture can be deposited using
the precursors from this study and Ru precursors supported by Dr. McElwee-White in the
Department of Chemistry at the University of Florida. By mixing both precursors in the solvent,
The Ru:W overall composition in the films can be varied simply by changing the relative
concentration of both precursors in PhCN. The current Cu barrier/seed stacks are a trilayer
consisting of PVD Cu on top of Ta (adhesion layer)/TaN (diffusion barrier). Using Ru-WNxCy
mixture phase films, it may be possible to replace the traditional PVD Cu/Ta/TaN stack with a
single layer of physical mixtures. That is the Ru should provide good adhesion since the WNxCy
showed nucleate Cu (111) [66].
10.2 WNxCy for Realistic Diffusion Barrier Testing
The diffusion barrier films tested in the present study were 15 to 20 nm in thickness. As
features in interconnects continue to shrink to align with the International Technology Roadmap
for Semiconductor (ITRS), the thickness of the diffusion barrier is required to be 2.9 nm in 2015
for metallization. Thus, a study of how thin can the barrier layer be grown and still be effective
should be made. The film thickness can be easily reduced by either reducing the reaction time or
the precursor concentration in the solvent by aerosol-assisted metal-organic CVD. The diffusion
barrier test results in the present study were characterized by XRD measurement, AES depth
profiling, Secco-etch test, cross-sectional TEM imaging with the EDS analysis, sheet resistance
measurement, and SEM surface imaging. These techniques require significant Cu transport
across the barrier to be effective. It is suggested that electrical characterization such as triangle
voltage sweep (TVS) techniques should be used to detect trace of Cu transport through the
139
barrier film to detect the time barrier limit. The effective capacitance of the structure is
employed as a measure of the free charge of Cu ions that has diffused through the barrier into the
adjacent dielectric. This result could then be correlated with the other methods to give a sense of
their sensitivity.
LIST OF REFERENCES
[1] E. Kolawa, P. J. Pokela, J. S. Reid, J. S. Chen, M. A. Nicolet, Appl. Surf. Sci. 53 (1991)
373-376.
[2] A. E. Kaloyeros, E. Eisenbraun, Annu. Rev. Mater. Sci. 30 (2000) 363-385.
[3] R. Rosenberg, D. C. Edelstein, C. K. Hu, K. P. Rodbell, Annu. Rev. Mater. Sci. 30
(2000) 229-262.
[4] A. A. Istratov, J. Electrochem. Soc. 149(1) (2000) G21-G30.
[5] M. Wittmer, Appl. Phys. Lett. 36 (1980) 456-458.
[6] D. H. Kim, S. L. Cho, K. B. Kim, J. J. Kim, J. W. Park, Appl. Phys. Lett. 69 (1996) 4182-
4184.
[7] G. Llauro, R. Hillel, F. Sibieude, Chem. Vapor Depos. 4 (1998) 247-252.
[8] M. H. Tsai, S. C. Sun, H. T. Chiu, C. E. Tsai, S. H. Chuang, Appl. Phys. Lett. 67 (1995)
1128-1130.
[9] T. Hara, M. Tanaka, K. Sakiyama, S. Onishi, K. Ishihara, J. Kudo, Jpn. J. Appl. Phys. 36
(1997) L893-L895.
[10] N. Gonohe, Mater. Trans. 43 (2002) 1585-1592.
[11] O. H. Gokce, S. Amin, N. M. Ravindra, D. J. Szostak, R. J. Paff, J. G. Fleming, C. J.
Galewski, J. Shallenberger, R. Eby, Thin Solid Films 353 (1999) 149-156.
[12] D. J. Kim, H. S. Sim, S.-I. Kim, Y. T. Kim, H. Jeon, 20 (2002) 194-197.
[13] W. Koh, D. Kumar, W. M. Li, H. Sprey, I. J. Raaijmakers, Solid State Technol. 48 (2005)
54-58.
[14] R. Fix, R. G. Gordon, D. M. Hoffman, Chem. Mater. 5 (1993) 614-619.
[15] S. Jeon, K. Yong, S. G. Park, S. W. Rhee, J. Vac. Sci. Technol. B 24 (2006) 1428-1431.
[16] B. H. Lee, K. J. Yong, J. Electrochem. Soc. 151 (2004) C594-C597.
[17] S.-H. Kim, S. S. Oh, H.-M. Kim, D.-H. Kang, K.-B. Kim, W.-M. Li, S. Haukka, M.
Tuominen, J. Electrochem. Soc. 151 (2004) C272-C282.
[18] O. J. Bchir, K. M. Green, H. M. Ajmera, E. A. Zapp, T. J. Anderson, B. C. Brooks, L. L.
Reitfort, D. H. Powell, K. A. Abboud, L. McElwee-White, J. Am. Chem. Soc. 127 (2005)
7825-7833.
[19] H. M. Ajmera, A. T. Heitsch, O. J. Bchir, D. P. Norton, L. L. Reitfort, L. McElwee-
White, T. J. Anderson, J. Electrochem. Soc. 155 (2008) H829-H835.
[20] E. Eisenbraun, A. Upham, R. Dash, W. Zeng, J. Hoefnagels, S. Lane, D. Anjum, K.
Dovidenko, A. E. Kaloyeros, B. Arkles, J. Sullivan, J. Vac. Sci. Technol. B 18 (2000)
2011-2015.
[21] Y.-J. Lee, B.-S. Suh, M. S. Kwon, C.-O. Park, J. Appl. Phys. 85 (1999) 1927-1934.
[22] J. G. Fleming, E. Roherty-Osmun, P. M. Smith, J. S. Custer, Y. D. Kim, T. Kacsich, M.
A. Nicolet, C. J. Galewski, Thin Solid Films 320 (1998) 10-14.
[23] L. McElwee-White, Dalton Trans. (2006) 5327-5333.
[24] J. Koller, H. M. Ajmera, K. A. Abboud, T. J. Anderson, L. McElwee-White, Inorg.
Chem. 47 (2008) 4457-4462.
[25] H. Kim, C. Cabral Jr., C. Lavoie, S. M. Rossnagel, J. Vac. Sci. Technol. B 20 (2002)
1321-1326.
[26] K.-E. Elers, V. Saanila, P. J. Soininen, W.-M. Li, J. T. Kostamo, S. Haukka, J. Juhanoja,
W. F. A. Besling, Chem. Vapor Depos. 8 (2002) 149-153.
[27] H. Kim, J. Vac. Sci. Technol. B 21 (2003) 2232-2261.
[28] F. C. T. So, E. Kolawa, X. A. Zhao, E. T. S. Pan, M. A. Nicolet, J. Appl. Phys. 64 (1988)
2787-2789.
[29] H. P. Kattelus, E. Kolawa, K. Affolter, M. A. Nicolet, J. Vac. Sci. Technol. A 3 (1985)
2246-2254.
[30] M. Uekubo, T. Oku, K. Nii, M. Murakami, K. Takahiro, S. Yamaguchi, T. Nakano, T.
Ohta, Thin Solid Films 286 (1996) 170-175.
[31] K. Nakajima, Y. Akasaka, K. Miyano, M. Takahashi, S. Suehiro, K. Suguro, Appl. Surf.
Sci. 117/118 (1997) 312-316.
[32] S. D. Marcus, R. F. Foster, Thin Solid Films 236 (1993) 330-333.
[33] T. Nakajima, K. Watanabe, N. Watanabe, J. Electrochem. Soc. 134 (1987) 3175-3178.
[34] M. Nagai, K. Kishida, Appl. Surf. Sci. 70-1 (1993) 759-762.
[35] B. H. Lee, K. J. Yong, J. Vac. Sci. Technol. B 22 (2004) 2375-2379.
[36] J. E. Kelsey, C. Goldberg, G. Nuesca, G. Peterson, A. E. Kaloyeros, B. Arkles, J. Vac.
Sci. Technol. B 17 (1999) 1101-1104.
[37] H. T. Chiu, S. H. Chuang, J. Mater. Res. 8 (1993) 1353-1360.
[38] E. L. Crane, H.-T. Chiu, R. G. Nuzzo, J. Phys. Chem. B 105 (2001) 3549-3556.
[39] J. P. Lu, W. Y. Hsu, J. D. Luttmer, L. K. Magel, H. L. Tsai, J. Electrochem. Soc. 145
(1998) L21-L23.
[40] J. Lin, A. Tsukune, T. Suzuki, M. Yamada, J. Vac. Sci. Technol. A 16 (1998) 611-614.
[41] J. Lin, A. Tsukune, T. Suzuki, M. Yamada, J. Vac. Sci. Technol. A 17 (1999) 936-938.
[42] Y. T. Kim, C. W. Lee, S. K. Min, Appl. Phys. Lett. 61 (1992) 537-539.
[43] S. E. Potts, C. J. Carmalt, C. S. Blackman, T. Leese, H. O. Davies, Dalton Trans. (2008)
5730-5736.
[44] A. C. Anacleto, N. Blasco, A. Pinchart, Y. Marot, C. Lachaud, Surf. Coat. Technol. 201
(2007) 9120-9124.
[45] O. J. Bchir, S. W. Johnston, A. C. Cuadra, T. J. Anderson, C. G. Ortiz, B. C. Brooks, D.
H. Powell, L. McElwee-White, J. Cryst. Growth 249 (2003) 262-274.
[46] O. J. Bchir, K. M. Green, M. S. Hlad, T. J. Anderson, B. C. Brooks, L. McElwee-White,
J. Cryst. Growth 261 (2004) 280-288.
[47] O. J. Bchir, K. M. Green, M. S. Hlad, T. J. Anderson, B. C. Brooks, C. B. Wilder, D. H.
Powell, L. McElwee-White, J. Organomet. Chem. 684 (2003) 338-350.
[48] O. J. Bchir, K. C. Kim, T. J. Anderson, V. Craciun, B. C. Brooks, L. McElwee-White, J.
Electrochem. Soc. 151 (2004) G697-G703.
[49] H. S. Sim, S.-I. Kim, Y. T. Kim, J. Vac. Sci. Technol. B 21 (2003) 1411-1414.
[50] C. W. Lee, Y. T. Kim, J. Vac. Sci. Technol. B 24 (2006) 1432-1435.
[51] J. W. Klaus, S. J. Ferro, S. M. George, J. Electrochem. Soc. 147 (2000) 1175-1181.
[52] J. W. Klaus, S. J. Ferro, S. M. George, Appl. Surf. Sci. 162 (2000) 479-491.
[53] J. S. Becker, R. G. Gordon, Appl. Phys. Lett. 82 (2003) 2239-2241.
[54] J. S. Becker, S. Suh, S. Wang, R. G. Gordon, Chem. Mater. 15 (2003) 2969-2976.
[55] P. de Rouffignac, Z. Li, R. G. Gordon, Electrochem Solid-State Lett. 7 (2004) G306-
G308.
[56] A. M. Hoyas, J. Schuhmacher, C. M. Whelan, T. F. Landaluce, D. Vanhaeren, K. Maex,
J. P. Celis, J. Appl. Phys. 100 (2006) 114903-6-114903-6.
[57] K. S. Kim, M. S. Lee, S. S. Yim, H. M. Kim, K. B. Kim, H. S. Park, W. Koh, W. M. Li,
M. Stokhof, H. Sprey, Appl. Phys. Lett. 89 (2006) 081913-1-081913-3.
[58] A. M. Hoyas, C. M. Whelan, J. Schuhmacher, K. Maex, J. P. Celis, Microeletron. Eng. 83
(2006) 2068-2071.
[59] H. Volders, Z. Tokei, H. Bender, B. Brijs, R. Caluwaerts, L. Carbonell, T. Conard, C.
Drijbooms, A. Franquet, S. Garaud, I. Hoflijk, A. Moussa, F. Sinapi, Y. Travaly, D.
Vanhaeren, G. Vereecke, C. Zhao, W. M. Li, H. Sprey, A. M. Jonas, Microelectron. Eng.
84 (2007) 2460-2465.
[60] S.-H. Kim, S. S. Oh, K. B. Kim, D.-H. Kang, W.-M. Li, S. Haukka, M. Tuominen, Appl.
Phys. Lett. 82 (2003) 4486-4488.
[61] S.-H. Kim, S. S. Oh, H.-M. Kim, D.-H. Kang, K.-B. Kim, W.-M. Li, S. Haukka, M.
Tuominen, Mat. Res. Symp. Proc. 766 (2003) E10.9.1-E10.9.7.
[62] K. E. Elers, V. Saanila, W. M. Li, P. J. Soininen, J. T. Kostamo, S. Haukka, J. Juhanoja,
W. F. A. Besling, Thin Solid Films 434 (2003) 94-99.
[63] S. Smith, W.-M. Li, K.-E. Elers, K. Pfeifer, Microelectron. Eng. 64 (2002) 247-253.
[64] C. L. Dezelah, O. M. El-Kadri, K. Kukli, K. Arstila, R. J. Baird, J. Lu, L. Niinisto, C. H.
Winter, J. Mater. Chem. 17 (2007) 1109-1116.
[65] W. X. Zeng, X. D. Wang, S. Kumar, D. W. Peters, E. T. Eisenbraun, J. Mater. Res. 22
(2007) 703-709.
[66] D. Greenslit, T. Chakraborty, E. Eisenbraun, J. Vac. Sci. Technol. B 27 (2009) 631-636.
[67] A. G. Orpen, L. Brammer, F. H. Allen, O. Kennard, D. G. Watson, R. Taylor, J. Chem.
Soc. Dalton. (1989) S1-S83.
[68] C. Redshaw, V. C. Gibson, W. Clegg, A. J. Edwards, B. Miles, J. Chem. Soc. Dalton.
(1997) 3343-3347.
[69] J. Chatt, M. E. Fakley, P. B. Hitchcock, R. L. Richards, N. T. Luong-Thi, J. Chem. Soc.
Dalton. (1982) 345-352.
[70] J. Koller, H. M. Ajmera, K. A. Abboud, T. J. Anderson, L. McElwee-White, Inorg.
Chem. (2007) 4457-4462.
[71] B. D. Cullity, S. R. Stock, Elements of X-ray diffraction: Third Edition; Third ed.;
Prentice Hall: Upper Saddle River, NJ, 2001.
[72] J. S. Reid, E. Kolawa, R. P. Ruiz, M. A. Nicolet, Thin Solid Films 236 (1993) 319-324.
[73] P. J. Pokela, C. K. Kwok, E. Kolawa, S. Raud, M. A. Nicolet, Appl. Surf. Sci. 53 (1991)
364-372.
[74] D. J. Kim, Y. B. Jung, M. B. Lee, Y. H. Lee, J. H. Lee, J. H. Lee, Thin Solid Films 372
(2000) 276-283.
[75] R. J. Colton, J. W. Rabalais, Inorg. Chem. 15 (1976) 236-238.
[76] C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder, G. E. Muilenberg, Handbook of
X-ray Photoelectron Spectroscopy; Perkin-Elmer Corporation: Eden Prairie, MN, 1978.
[77] J. F. Moulder, W. F. Stickle, P. E. Sobol, K. D. Bomben, Handbook of X-ray
Photoelectron Spectroscopy; J. Chastain and R. C. King Jr., Eds.; Physical Electronics,
Inc.: Eden Prairie, MN, 2000.
[78] H. L. Zhang, D. Z. Wang, N. K. Huang, Surf. Coat. Technol. 115 (1999) 140-144.
[79] Y. G. Shen, Y. W. Mai, Mater. Sci. Eng. B 76 (2000) 107-115.
[80] J. S. Lee, C. S. Park, J. Y. Kang, D. S. Ma, J. Y. Lee, J. Vac. Sci. Technol. B 8 (1990)
1117-1121.
[81] P. J. C. Chappell, M. H. Kibel, B. G. Baker, J. Catal. 110 (1988) 139-149.
[82] G. Leclercq, M. Kamal, J. M. Giraudon, P. Devassine, L. Feigenbaum, L. Leclercq, A.
Frennet, J. M. Bastin, A. Lofberg, S. Decker, M. Dufour, J. Catal. 158 (1996) 142-169.
[83] V. Crist, Handbook of Monochromatic XPS Spectra: The Elements and Native Oxides;
John Wiley & Sons, Ltd.: New York, 2000.
[84] J. Luthin, C. Linsmeier, J. Nucl. Mater. 290-293 (2001) 121-125.
[85] Y. S. Won, Y. S. Kim, T. J. Anderson, L. McElwee-White, Chem. Mater. 20 (2008)
7246-7251.
[86] K. Takahashi, M. Nakayama, S. Tokoyama, T. Kimura, E. Tokumitsu, H. Funakubo,
Applied Surface Science 216 (2003) 296-301.
[87] Y. S. Won, Y. S. Kim, T. J. Anderson, L. L. Reitfort, I. Ghiviriga, L. McElwee-White, J.
Am. Chem. Soc. 128 (2006) 13781-13788.
[88] S.-L. Cho, K.-B. Kim, S.-H. Min, H.-K. Shin, S.-D. Kim, J. Electrochem. Soc 146 (1999)
3724-3730.
[89] S.-H. Kim, D.-S. Chung, K.-C. Park, K.-B. Kim, S.-H. Min, J. Electrochem. Soc. 146
(1999) 1455-1460.
[90] J. F. Moulder, W. F. Stickle, P. E. Sobol, K. D. Bomben, Handbook of X-ray
Photoelectron Spectroscopy; J. Chastain and R. C. King Jr., Eds.; Physical Electronics,
Inc.: Eden Prairie, MN, 1995.
[91] A. Furuya, M. Tagami, K. Shiba, K. Kikuta, Y. Hayashi, IEEE Trans Electron Devices 49
(2002) 733-738.
[92] K.-E. Elers, V. Saanila, P. J. Soininen, W.-M. Li, J. T. Kostamo, S. Haukka, J. Juhanoja,
Chem. Vapor Depos. 8 (2002) 149-153.
[93] L. E. Toth, Transition Metal Carbides and Nitrides; J. L. Margrave, Ed.; Academic Press:
New York, 1971.
[94] A. C. Westerheim, J. M. Bulger, C. S. Whelan, T. S. Sriram, L. J. Elliott, J. J. Maziarz, J.
Vac. Sci. Technol. B 16 (1998) 2729-2733.
[95] H. M. Ajmera, A. T. Heitsch, O. J. Bchir, D. P. Norton, L. L. Reitfort, L. McElwee-
White, T. J. Anderson, J. Electrochem. Soc. 155 (2008) H829-H835.
[96] J. Y. Hwang, C. Park, M. Huang, T. Anderson, J. Crst. Growth (2005) 521-530.
[97] Y. S. Kim, Y. S. Won, H. Hagelin-Weaver, N. Omenetto, T. Anderson, J. Phys. Chem. A
(2008) 4246-4253.
BIOGRAPHICAL SKETCH
Dojun Kim was born in 1975 in Pusan, Korea. He entered Seoul National University in
March 1994 and received his Bachelor of Science degree in chemical engineering in February
1998. Following undergraduate, he started graduate school in Seoul National University and
received his Master of Science degree in chemical engineering in February 2000. Upon
graduation, he worked for more than five years as a process engineer for SK Engineering &
Construction (SKEC) in Seoul, Korea (2000 – 2005). The main role is process design,
simulation, control, and consultation for petrochemical and refinery processes. Following his
five years industry, he started his doctoral studies in the Department of Chemical Engineering at
the University of Florida in August 2005. He joined the electronic materials processing group
under the guidance of Dr. Timothy J. Anderson in December 2005. His research topic is
chemical vapor deposition and atomic layer deposition of metal nitride thin films for diffusion
barrier application. While he was working in SKEC and UF, he married Sora Park on February
23, 2002 and had a son, Jinho Kim, on October 28, 2003 (Seoul, Korea) and a daughter,
Katherine Nayoun Kim, on January 23, 2007 (Gainesville, FL). Upon graduation, he plans to
work as a senior process engineer in Intel‟s Portland Technology Development Division based in
Hillsboro, OR.