l I111}: 4. I11 ‘ a u killiyltlrii'lllJ

 

 

 

 

 

   

 

 

4‘
UBRARY

2% Michigan State
Universﬁy

 

 

 

 

This is to certify that the
dissertation entitled

SUBSTRATES FOR SURFACE-ENHANCED RAMAN
SPECTROSCOPY

presented by

MUHAMMAD AJMAL KHAN

has been accepted towards fulﬁllment
of the requirements for the

PhD degree in Electrical Engineering

 

 

“ CW“ ,4 W

Major’ Prcé’essor’s Signature

“/17/03

Date

MSU is an afﬁrmative-action, equal—opportunity employer

 

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

 

DAIEDUE

DAIEDUE

DAIEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K lProj/AccGPres/ClRC/DaleDue undd

- —- ”w.” .

 

 

SUBSTRATES FOR SURFACE-ENHANCED RAMAN SPECTROSCOPY
By

MUHAMMAD AJMAL KHAN

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment Of requirements
for the degree of
DOCTOR OF PHILOSOPHY

Electrical Engineering

2008

 

ABSTRACT

SUBSTRATES FOR SURFACE-ENHANCED RAMAN
SPECTROSCOPY

By

Muhammad Ajmal Khan

Surface enhanced Raman spectroscopy (SERS) utilizes surface plasmon
resonance in metallic nanostructures to yield nearly a millionfold increase in the Raman
signal of an analyte. The unprecedented sensitivity and speciﬁcity of SERS has great
potential for applications in analytical chemistry and biological sensors. SERS is
inherently a nanoscale phenomenon and recent advances in nanotechnology have
generated an immense Opportunity for the use of nanoparticles, nanowires and nanorods
as substrates for SERS. This dissertation explores different aspects of utilizing bulk
synthesized germanium oxide, zinc oxide nanowires, and metallic nanorods as substrates
for SERS. It discusses the synthesis details and growth kinetics for the oxide nanowires
and metallic nanorods. The germanium nanowire growth is carried out in a simple tube
fumace chamber using two different temperature regimes - the traditionally high
temperature synthesis and a novel low temperature synthesis. The high temperature
synthesis (~850 °C) does not yield good control over length and diameter of the
nanowires. The new low temperature synthesis technique overcomes these limitations by
maintaining the source temperature at ~650 °C.

Zinc oxide nanowires are synthesized by thermal evaporation Of zinc powder in
an oxidizing environment. The as-synthesized bulk nanowires act as nanostructured

template that is coated with a thin gold ﬁlm to create a plasmon active surface for SERS

‘-

 

Y
~.

application. In addition metallic nanorods of aluminum, copper, and silver were
synthesized on glass slides in an e-beam equipped physical vapor deposition system using
oblique angle deposition technique. Structural and chemical compositional
characterization of the SERS substrates was done using a scanning electron microscope,
transmission electron microscope and energy dispersive X-ray spectroscopy (EDS). The
SERS performance is evaluated using model analytes of 4-methylbenzenthiol, 1,2-
benzenedithiol and trans-1,2-bis(4-pyridyl)ethylene. The substrates yield strong and
unambiguous Raman spectra from just a monolayer or few femto moles of analyte. The
Raman enhancement factors are computed by comparing the intensity Of Raman signal
from SERS substrate to that from the bulk analyte. The metallic nanorods substrates show
a metal dependent SERS enhancement with silver yielding an order of magnitude

stronger enhancement compared to other metals. The oxide nanowires based substrates
show an average Raman enhancement factor of ~106 with good reproducibility of signal

over the tested area. The dissertation puts in to perspective how noble metal coated

germanium oxide and zinc oxide nanowires can be used as robust SERS platform for

detection of trace levels of chemical species.

 

 

Copyright by
MUHAMMAD AJMAL KHAN
2008

To my parents and family.

1!.

ACKNOWLEDGEMENTS

First of all, I would like to thank Almighty for granting me the honor of
accomplishing this work. I would also like to thank my committee for their patience,
guidance and understanding throughout the course of this research. I would like to
speciﬁcally express my deep gratitude to my advisor, Prof. Timothy P. Hogan without
whom none of the accomplishments documented in this dissertation would have been
possible. He has always been an outstanding mentor and a great friend throughout my
stay at Michigan State University.

On a personal note, I would like to express my most sincere gratitude to my
parents, my parents-in-law, and my family whose love, prayers, and motivation are the
biggest reason behind my success. I would also like to express my warmest appreciation
and gratitude for my wife, whose passion, sacriﬁce and encouragement gave me the
strength and inspiration to complete this work.

And ﬁnally I would like to express my gratitude and thanks to my long time
friend and fellow researcher, Muhammad Farhan, for his support, dedication and
encouragement that kept me focused during ups and downs of my stay in graduate
school.

Financial support from the National Science Foundation is gratefully

aCknowledged.

vi

TABLE OF CONTENTS

 

 

 

 

 

LIST OF TABLES x
LIST OF FIGURES xi
Chapter 1: Introduction 1
1.1 Raman spectroscopy ............................................................................... l
1.2 Limitations of conventional Raman spectroscopy .................................. 2
1.3 Surface enhanced Raman spectroscopy .................................................. 3
1.4 Common SERS substrates ...................................................................... 6
1.5 Major challenge for SERS ...................................................................... 7
1.6 SERS from Nanowires : Motivation for this work ................................. 8
1.7 Organization Of this dissertation ........................................................... 11
Chapter 2: Theory of Raman Scattering 12
2.1 Classical theory of Raman scattering .................................................... 13
2.2 Quantum mechanical theory of Raman scattering ................................ 20
2.3 Surface enhanced Raman scattering ..................................................... 25
2.3.1 Electromagnetic mechanism ............................................................. 25
2.3.2 The chemical enhancement ............................................................... 47

2.4 Ordinary Raman spectra and SERS spectra .......................................... 48
Chapter 3: Experimental Set up 50
3.1 Oxide nanowire synthesis set up ........................................................... 50
3.2 Silver nanorods synthesis set up ........................................................... 53
3.3 Raman spectroscopy set up ................................................................... 55
3.3.1 Laser spot size determination ............................................................ 59
3.3.2 Raman system calibration ................................................................. 60
Chapter 4: Germanium Oxide Nanowires based SERS Substrates ............ 63
4.1 Nanowire synthesis ............................................................................... 64
4.1.1 Reviews of past efforts ...................................................................... 64
4.1.2 Our synthesis technique .................................................................... 65
4.1.3 Characterization of nanowires .......................................................... 67

4.2 Nanowire growth kinetics ..................................................................... 71
4.2.1 Oxidation Of germanium ................................................................... 71
4.2.2 Stability of germanium oxides .......................................................... 73
4.2.3 Growth model ................................................................................... 74
4.2.4 Synthesis temperature proﬁle ........................................................... 83
4.2.5 Synthesis pressure profile ................................................................. 84

4.3 Factors affecting growth of nanowires ................................................. 87
4.3.1 Role of gold catalyst ......................................................................... 87
4.3.2 Effect of growth time on length of nanowires ................................ 100
4.3.3 Effect of oxygen ﬂow rate .............................................................. 101
4.3.4 Effect Of synthesis temperature ....................................................... 103

vii

4.3.5 Effect of background gas ................................................................ 104

 

 

4.3.6 Effect of background gas pressure .................................................. 107
4.4 Raman characterization of GeOz nanowires ....................................... 109
4.4.1 Experimental parameters ................................................................ 1 10
4.4.2 Raman spectra of GeOz nanowires ................................................. 110
4.4.3 Gold coating of GeOz nanowires .................................................... 112
4.5 Evaluation of SERS substrate ............................................................. 120
4.5.1 SERS of rhodamine and nile blue .................................................. 121
4.5.2 SERS of thiols ................................................................................. 130
4.6 Conclusion .......................................................................................... l 59
Chapter 5: Low Temperature Synthesized GeOz Nanowires based
Substrates for SERS 161
5.1 Limitation of high temperature nanowire synthesis ............................ 162
5.2 Low temperature Synthesis of nanowires ........................................... 167
5.3 Nanowire grth kinetics ................................................................... 172
5.3.1 Oxidation Of germanium ................................................................. 172
5.3.2 Growth kinetics ............................................................................... 174
5.3.3 Synthesis temperature proﬁle ......................................................... 180
5.3.4 Pressure proﬁle ............................................................................... 181
5.4 Factors affecting NWs synthesis ......................................................... 182
5.4.1 Effect of gold ﬁlm thickness ........................................................... 183
5.4.2 Effect of growth time on length of nanowires ................................ 195
5.4.3 Effect Of oxygen ﬂow rate .............................................................. 197
5.4.4 Effect of background gas ﬂow rate ................................................. 200
5.4.5 Effect of temperature ...................................................................... 201
5.4.6 Background gas type ....................................................................... 203
5.4.7 Effect Of background gas pressure .................................................. 209
5.5 Raman characterization of GeOz nanowires ....................................... 210
5.5.1 Raman spectra of substrate ............................................................. 210
5.5.2 Gold coating of GeOz nanowires .................................................... 213
5.6 SERS evaluation of substrates ............................................................ 217
5.6.1 Detection of 4-MBT ........................................................................ 218
5.6.2 Detection of 1,2-BDT ..................................................................... 228
5.7 Conclusion .......................................................................................... 232
Chapter 6: Zinc Oxide Nanowires based SERS Substrates 234
6.1 Zinc oxide nanowires synthesis .......................................................... 234
6.2 ZnO nanowires growth model ............................................................ 242
6.3 Raman characterization of ZnO nanowires ......................................... 248
6.3.1 Raman Characterization of NWs .................................................... 249
6.3.2 Gold coating of ZnO nanowires ...................................................... 250
6.4 SERS study of ZnO nanowires substrate ............................................ 252
6.5 Detection of R6G probe molecule ...................................................... 252
6.6 Detection of nile blue .......................................................................... 254
6.7 Detection using thiols ......................................................................... 256

viii

 

 

 

6.7.1 Gold coating thickness study .......................................................... 257
6.7.2 SERS for 4-MBT ............................................................................ 261
6.7.3 SERS for 1,2-BDT .......................................................................... 271

6.8 Conclusion .......................................................................................... 274
Chapter 7: Metallic Nanorods based SERS Substrates 276
7.1 Oblique angle deposition .................................................................... 276
7.2 Metallic nanorods synthesis ................................................................ 277
7.3 Oblique angle deposition kinetics ....................................................... 287
7.3.1 Incident ﬂux angle .......................................................................... 288
7.3.2 Temperature .................................................................................... 288
7.3.3 Rate of deposition ........................................................................... 289

7.4 Raman characterization of substrates .................................................. 289
7.5 Detection of nile blue .......................................................................... 291
7.6 Detection of MBT .............................................................................. 294
7.7 Detection of BDT ................................................................................ 307
7.8 Conclusion .......................................................................................... 309
Chapter 8: Conclusions and Future Work 310
8.1 Conclusion .......................................................................................... 3 10
8.2 Future work ......................................................................................... 315
Appendix A: Operation of EzRaman-L system 317
A. 1 Introduction ......................................................................................... 3 17
A2 Procedure ............................................................................................ 3 17
A.3 Sample mounting ................................................................................ 317
A.4 Acquiring Spectra ............................................................................... 318
A5 Shutdown ............................................................................................ 319
A6 System Calibration .............................................................................. 319
References - -- - 320

 

ix

LIST OF TABLES

Table 3-1. Isopropyl alcohol Raman peaks and wavenumbers. ....................................... 61
Table 4-1. SERS EF for common substrates. ................................................................. 145
Table 4-2. Raman band assignment for 4-MBT ............................................................. 148
Table 4-3. Raman peak assignment for BPE .................................................................. 152
Table 4-4. Raman band assignment for 1,2-BDT. ......................................................... 155
Table 7-1. Depolarization factors for some common geometries. ................................. 302

'1

V

,1

1h

D
p

LIST OF FIGURES

Figure 2-1. Pictorial representation of Rayleigh and Raman scattering[47] .................... 17
Figure 2-2. Three modes of vibrations of C02 molecule[3]. ........................................... 19

Figure 2-3. Schematic illustration of plasmon oscillation for sphere showing electron
distortion in response to electric ﬁeld [72]. ...................................................................... 27

Figure 24. Energy ﬂux (Poynting vector) around a metal nanoparticles under plane
wave excitation for two frequencies[77]- away from resonance and at plasmon
resonance ........................................................................................................................... 33

Figure 2-5. Effect Of radiation damping for different aspect ratio of gold nanorods. Also
shown is the resonant wavelength for each aspect ratio. .................................................. 44

Figure 2-6. Light interaction with metal nanospheres dimmers[73]. Analyte molecule is

shown by small green ﬁlled circle in between two particles. ........................................... 45
Figure 2-7. Calculated Raman enhancement factors at the junction of 60 nm diameter
dimmers for 497 nm excitation[39]. ................................................................................. 46
Figure 3-1. Photograph of nanowire synthesis set up. ..................................................... 52
Figure 3-2. Quartz support plate with typical sample layout (A top view). .................... 52
Figure 3-3. Physical vapor deposition systems for metallic nanorods synthesis. ............ 54

Figure 3-4. Schematic of oblique angle deposition of silver nanorods using e-beam

evaporation. (a) Normal angle deposition. (b) Oblique angle deposition. ........................ 55
Figure 3-5. Raman system set up. .................................................................................... 56
Figure 3-6. Raman system optical ﬁber head and sample stage. ..................................... 5 7
Figure 3-7. Knife edge method for determining laser beam diameter. ............................ 60
Figure 3-8. Reference Raman spectra of Isopropanol, tax: 20 s. ................................... 61
Figure 4-1. Schematic of germanium dioxide nanowire synthesis. ................................. 67
Figure 4-2. SEM image of GeOz nanowires on different samples of Si. ........................ 68
Figure 4-3. EDS spectrum and quantiﬁcation results for Ge02 nanowires. .................... 69

xi

Figure 4-4. TEM image showing gold at tip and EDS spectra for two regions of nanowire
and the diffraction pattern. ................................................................................................ 70

Figure 4-5. Vapor pressure of GeO at different temperatures. ........................................ 76

Figure 4-6. Formation of cloudy reactants after introducing oxygen into the quartz tube.

........................................................................................................................................... 77
Figure 4-7. Spherical particles and GeOz NWs. .............................................................. 78
Figure 4—8. Faceted germanium oxide particles with NWs. ............................................ 78
Figure 4-9. Ge02 particles on An coated Si substrate after typical synthesis run. .......... 82

Figure 4-10. Plot showing GeOz stability as function of temperature and the hydrogen to

water vapors pressure ratio[112] ....................................................................................... 83
Figure 4-11. Tube fumace typical temperature proﬁle during nanowire synthesis. ........ 84
Figure 4-12. Typical pressure proﬁle during NW synthesis. ........................................... 86
Figure 4-13. GeOz NWs on Si(100) partially coated with 7 nm Au ﬁlm. ....................... 88
Figure 4-14. GeOz NW showing particle at tip. .............................................................. 89
Figure 4-15. SEM image and particle size distribution for 3 nm Au ﬁlm. ...................... 90
Figure 4-16. SEM image and particle size distribution for 7 nm Au ﬁlm. ...................... 91
Figure 4-17. SEM image and particle size distribution for 15 nm Au ﬁlm. .................... 92
Figure 4-18. SEM image and particle size distribution for 21 nm Au ﬁlm. .................... 93
Figure 4-19. SEM image and particle size distribution for 28 nm Au ﬁlm. .................... 94
Figure 4-20. Mean particle size distributions after annealing different thickness Au ﬁlms.
Error bars indicate standard deviation from mean. ........................................................... 95
Figure 4-21. Ge02 NW diameter distribution for 3 nm Au ﬁlm. .................................... 96
Figure 4-22. Ge02 NW diameter distribution for 7 nm Au ﬁlm. .................................... 97
Figure 4-23. GeOz NW diameter distribution for 15 nm Au ﬁlm. .................................. 97
Figure 4-24. Mean diameters of GeOz W5 and annealed Au ﬁlm particles. ................ 98
Figure 4-25. A faceted GeOz whisker. ............................................................................. 99

xii

Figure 4-26. Cross-section of substrate showing long NWs extending from substrate. 101

Figure 4-27. Effect of oxygen concentration on NW synthesis. In all cases, the total ﬂow
rate (background + oxidation mix) was 100 sccm. ......................................................... 102

Figure 4-28. GeOz NWs synthesized at different temperatures (a) 865 °C (b) 840 °C (d)
830 “C (d) 800 °C (e) 765 0C. .......................................................................................... 104

Figure 4-29. SEM image of GeOz NWs synthesized using N2 as background gas. ...... 106

Figure 4-30. SEM image of GeOz NWs synthesized using Ar as background gas. ...... 106

Figure 4-31. Ge02 NWs diameters for different background gases. ............................. 107
Figure 4-32. Effect of background gas pressure on NW synthesis. ............................... 108
Figure 4-33. White GeOz coating at the end of quartz tube. ......................................... 109

Figure 4-34. (a) Raman spectra for two phases of GeOz. Reproduced with permission
from ref [114]. (b) Raman spectra of GeOz nanowires on Silicon substrate, tacc=15 s.. 111

Figure 4-35. Raman spectra of 21 nm Au coated Ge02 NW5, tacc=15 5. Also shown are
spectra for uncoated NWs. .............................................................................................. 112

Figure 4-36. Raman spectra of 15 nm Au coated GeOz NWs, tacc=5 5. Also shown is the
Spectra for uncoated NWs. .............................................................................................. 113

Figure 4-37. Raman spectra for bare silicon wafer and after coating with ~21nm gold.
......................................................................................................................................... 114

Figure 4-38. Variation of 520 cm‘1 Raman peak intensity with Au ﬁlm thickness. ...... 115

Figure 4-39. Raman spectra of lOnm e-beam high purity gold coated GeOz nanowires,
iacc=5 5. Also shown is the spectra from an uncoated part of the sample. ...................... 116

Figure 4-40. Successive Raman spectra from same point on sample for 21 nm Au coated
3602 NW sample dried in a desiccator for 36 hr. .......................................................... 117

Figure 4-41. Effect of ~21nm gold coated GeOz nanowires at 146 °C, tacc=5 5. Insert
Shows room temperature Raman spectra. ....................................................................... 119

Figure 4-42. TEM image of 21 nm Au coated GeOz NW. Insert shows another NW with
1ﬁnger Au particles. .......................................................................................................... 120

Figure 443. Chemical structure and ordinary Raman spectra of R6G with 785 nm
excitation ......................................................................................................................... 122

xiii

 

Figure 4-44. Raman spectra of 2uM R6G in DI water for 14mm gold coated Ge02

nanowires. Bulk solution spectrum of same concentration is also shown, tacc=15 3. Bulk
spectrum is x10 times the value to show features ........................................................... 123

Figure 4-45. Raman spectra of 3 uM R6G in ethanol for 21 nm gold coated GeOz
nanowires. Also shown is the ordinary Raman spectra of 3 uM ethanolic R6G bulk
solution, tacc=15 s. .......................................................................................................... 124

Figure 4-46. Raman spectra of 3 uM R6G in ethanol from GeOz nanowire substrate

decorated with 60 nm gold colloidal. Also shown is the baseline corrected spectra, tacc=
15 s. ................................................................................................................................. 125

Figure 4-47. (a) Nile blue molecular structure. SERS spectra of NB with 531 nm
excitation[124]. ............................................................................................................... 127

Figure 4-48. SERS spectra of 21 nm Au ﬁlm coated GeOz NWs before and after
application of 2 uL of 10 uM NB, tacc= 15 3. Baseline corrected spectra of NB is also

shown. ............................................................................................................................. 128
Figure 4-49. SERS spectra from 2 uL of 10 uM ethanolic NB along a line from GeOz
NWs, tacc= 5 s. ................................................................................................................ 128
Figure 4-50. Raman spectrum of spectra of 15 nm Ag coated Ge02 nanowire substrate
before and after applying 2.73 uM Nile blue in ethanol, tax: 5 s. ................................. 129
Figure 4-51. Chemical structures Raman ethylene analytes. ......................................... 131
Figure 4-52. Effect Of Au thickness on SERS spectra. .................................................. 132
Figure 4-53. SERS from 1mM BDT for 20 min soak .................................................... 133
F igure 4-54. Effect of gold coating age on SERS from 1mM MBT .............................. 134

Figure 4-55. SEM image of GeOz NW substrate after 12 hr soak in 1mM MBT. ........ 135
Figure 4-56. Chemical structure and ordinary Raman spectrum of 4-MBT. ................. 136

Figure 4-57. (3) SERS spectra of a SAM of 4-MBT. Also shown is the Raman spectra of
SHbstrate before formation of SAM, tacc=15 s for both spectra. (b) SERS spectra Of a
SAM of 4-MBT from 5 adjacent points 250 um apart, tacc= 5 s. ................................... 138

Figure 4-58. 4-MBT 1068 cm'1 Raman peak intensity distribution for randomly chosen
area on substrate, taco: 5 s. ............................................................................................. 139

xiv

‘t

A‘-

0‘

.9'
‘l

.4

 

Figure 4-59. Possible structures of 4-MBT SAM on Au(111)[130]. Open circles are gold
atoms and solid circles represent sulﬁir head group of 4-MBT. ..................................... 140

Figure 4-60. Ordinary Raman spectra of 4-MBT and polystyrene, tacc= 5 s. ............... 143

Figure 4-61. (a) EF correction for cylindrical excitation volume assmnption. (b) DFOV
estimation using two Raman bands of solid 4-MBT crystals. ........................................ 144

Figure 4-62. SERS spectra of a SAM of 4-MBT on Klarite chip from 5 different points,

ta“: 5 s. .......................................................................................................................... 146
Figure 4-63. 4-MBT 1068 cm'l Raman peak intensity distribution for randomly chosen
area on Klarite chip, tacc= 5 s. ......................................................................................... 147
Figure 4-64. Chemical structure and ordinary Raman spectrum of BPE ....................... 149

Figure 4-65. SERS spectra of 0.83 uM BPE from Au coated GeOz NWs, tacc= 15 s. .. 150

Figure 4-66. SERS spectra of 0.83 uM BPE from different points on substrate, tacc= 5 s.
. ........................................................................................................................................ 151

Figure 4-67. BPE 1190 cm'1 Raman peak intensity distribution for randomly chosen area
on substrate, tam= 5 s. ..................................................................................................... 151

Figure 4-68. Ordinary Raman spectra of 1,2-BDT in liquid state. Also show is the
chemical structure of molecule. ...................................................................................... 153

Figure 4-69. SERS spectra of self-assembled monolayer of 1,2-BDT from Au coated

GeOz nanowires, tax: 15 s. ............................................................................................ 154
Figure 4-70. SERS spectra from 1mM BDT along line on substrate. ........................... 156
Figure 4-71. Area map of 1028 cm" Raman peak of 1,2-BDT, tacc=5 s. ...................... 157
Figure 4-72. Molecular structure of 1,2-BDT and formation Of SAM on Au(111)
face[126]. ........................................................................................................................ 157
Figure 4-73. Optical image of laser illumination Of Si wafer (left) and G602 NWs
substrate (right). .............................................................................................................. 159
Figure 5-1. SEM image of G602 NW substrate ............................................................. 164
Figure 5-2. SEM image of cracked GeOz ﬁlm on Si substrate during nanowire synthesis.
......................................................................................................................................... 164
Figure 5-3. EDS elemental map for top right comer of cavity in Figure 5-2. .............. 165

XV

.53 '

Figure 5-4. SEM image showing non-uniformity of the ﬁlm on the substrate. ............. 166

Figure 5-5. SEM image of substrate showing different morphologies of G602 particles
and W8 .......................................................................................................................... 166

Figure 5-6. Schematic of low temperature synthesis of germanium dioxide nanowire. 168
Figure 5-7. SEM image of Ge02 nanowires on Si at 400-500 °C ................................. 168

Figure 5-8. EDS spectrum and quantiﬁcation results for GeOz nanowires. .................. 169

Figure 5-9. TEM image showing dark catalyst particle at tip. Right image shows a dark

spherical particle probably broken during sample preparation. ...................................... 170
Figure 5-10. Different morphology of low temperature GeOz NWs. ............................ 171
Figure 5-11. Rate of Ge oxidation versus oxygen pressure at 550 °C [98] .................... 175

Figure 5-12. Plot showing GeOz stability as function of temperature and the hydrogen to
water vapors pressure ratio[112] ..................................................................................... 180

Figure 5-13. Tube fumace typical spatial temperature proﬁle during nanowire synthesis.
. ........................................................................................................................................ 181

Figure 5-14. Typical pressure proﬁle during NW synthesis. ......................................... 182

Figure 5—15. GeOz NWs on Si(100) that has patterned Au ﬁlm (NO gold was coated in
the dark “X” pattern). Right image shows close up of the red circled area of left image.

......................................................................................................................................... 1 84
Figure 5-16. GeOz NW showing particle at tip. ........................................................... 184
Figure 5-17. SEM image and particle size distribution for 3 nm Au ﬁlm. .................... 186
Figure 5-18. SEM image and particle size distribution for 7 nm Au ﬁlm. .................... 187
Figure 5-19. SEM image and particle size distribution for 15 nm Au ﬁlm. .................. 188
Figure 5-20. SEM image and particle size distribution for 21 nm Au ﬁlm. .................. 189
Figure 5-21. SEM image and particle size distribution for 28 nm Au ﬁlm. .................. 190
Figure 5-22. Mean particle size distribution for different initial Au ﬁlm thickness at
500 °C. Error bars indicate standard deviation from mean. ............................................ 191
Figure 5-23. Ge02 NW diameter distribution for 7 nm Au ﬁlm. .................................. 192
Figure 5-24. GeOz NW diameter distribution for 14 nm Au ﬁlm. ................................ 193

xvi

 

Figure 5-25. GeOz NW diameter distribution for 21 nm Au ﬁlm. ................................ 193

Figure 5-26. Mean diameters of GeOz W5 and annealed Au ﬁlm particles. .............. 195
Figure 5-27. Effects of growth time on the length of the NWs. .................................... 196
Figure 5-28. Length of NWs versus growth time. ......................................................... 197
Figure 5-29. Effect of oxygen concentration on NW synthesis. .................................... 198
Figure 5-30. Elemental composition of the sample surfaces after growth for different
oxygen concentrations. ................................................................................................... 199
Figure 5-31. Effect of background gas ﬂow rate on NWs synthesis. ............................ 201
Figure 5—32. GeOz NW growth at different temperatures. ............................................ 202

Figure 5-33. SEM images of GeOz NWs synthesized using Ar as background gas ...... 203
Figure 5-34. SEM image of GeOz NWs synthesized using N2 as background gas. ...... 204
Figure 5-35. GeOz NWs diameters for different background gases. ............................. 205

Figure 5-36. Island like growth of NWs for 7 nm Au coated Si substrate with N2 as
background gas. .............................................................................................................. 206

Figure 5-37. Island like growth of NWs for 7 nm Au coated Si substrate with Ar as the
background gas. .............................................................................................................. 207

Figure 5-38. GeOz NWs found at higher temperature region of furnace on 7 nm Au
coated Si substrate ........................................................................................................... 208

Figure 5-39. GeOz NWs found at higher temperature region Of furnace on 15 nm Au

coated Si substrate ........................................................................................................... 208
Figure 5-40. NW synthesis at a background gas pressure of 1.1 Torr. .......................... 209
Figure 541. NW synthesis at a background gas pressure Of 6 Torr shows location
dependent growth. ........................................................................................................... 210
Figure 5-42. Raman spectra of GeOz NWs of different diameters, tacc=15 s. ............... 211

Figure 5-43. Raman spectra for two different morphologies of NWs. Insert shows SEM
image showing circular islands on uniform background. ............................................... 212

Figure 5-44. Raman spectra of bare and 12 nm Au coated GeOz NWs, tacc=15 s. ....... 214

xvii

l1

 

Figure 5-45. Raman spectra of Au coated GeOz NW5 Of average diameter 108 nm,
tacc=15 s. ......................................................................................................................... 215

Figure 5-46. Raman spectra of Au coated Ge02 NW5 of average diameter 120 nm,
tacc=15 s. ......................................................................................................................... 216

Figure 5—47. SEM images of substrate before (left) and after (right) 12 hr soak in 1 mM
methanolic MBT. ............................................................................................................ 217

Figure 548. Chemical structures of 4-MBT and 1,2-BDT used as Raman analytes. 217

Figure 5—49. SERS spectra of SAM of 4-MBT on 108 nm dia NW5, tacc=15 s. ........... 219

Figure 5-50. Successive spectra from substrate points spaced 250 um along a line,
tacc=15 s .......................................................................................................................... 219

Figure 5-51. Effect of Au coating thickness on SERS activity of substrate having 108 nm
mean diameter NW5. ....................................................................................................... 221

Figure 5-52. Effect of Au coating thickness on SERS activity of substrate having 108 nm
mean diameter NWs. Also shown is spectra (along the x-axis) from a ﬂat Si(100) with 55

nm Au coating that was soaked in 1mM MBT along with the NW5. ............................. 221
Figure 5-53. Variation of 1068 cm'1 Raman peak intensity of 4-MBT with Au ﬁlm
thickness .......................................................................................................................... 222
Figure 5-54. SERS spectra of SAM of 4-MBT for different Au ﬁlm thickness on G602
NW5 with 108 nm mean diameter. ............................................................................... 222
Figure 5-55. Variation of Raman intensity with Au ﬁlm thickness for different diameters
of NW5. ........................................................................................................................... 223
Figure 5-56. Variation Of Raman BF with length of GeOz NW5 coated with 40 nm thick
Au ﬁlm. ........................................................................................................................... 224
Figure 5-57. Spatial map of SERS intensity for 1068 cm'1 peak of 4-MBT .................. 225
IFigure 5-58. Ordinary Raman spectra of 4-MBT and polystyrene. ............................... 228
Figure 5-59. SERS spectra Of SAM of 1,2-BDT on 108 nm diameter NW5, tace = 15 s.
.................................................................................................................................... 229
Figure 5-60. SERS spectra of SAM of 1,2-BDT from different points on substrate ..... 229

Figure 5-61. Spatial map of SERS intensity for 1028 cm'1 peak of 1,2-BDT, tacc=5 s. 230

Figure 6-1. Experimental set up for ZnO nanowire synthesis. ...................................... 237

xviii

 

Figure 6—2. Zinc oxide nanowires on Si 100 substrate using Au as catalyst .................. 237
Figure 6—3. Diameter distribution of ZnO NW5 ............................................................. 238
Figure 6-4. ZnO nanowires on source powder ............................................................... 238
Figure 6-5. ZnO NWs found in uncoated area of Si wafer. ........................................... 239
Figure 6-6. EDS spectrum and quantization results for ZnO nanowires. ...................... 240

Figure 6-7. Different morphologies of ZnO nanostructures. (a) Nanocombs. (b)

Nanorods. (c) Nanowhisker. (d) Nanocages. (e) Flowers. (f) Hexagonal rods. ............. 241
Figure 6-8. Probing structures under ZnO NW5. ........................................................... 245
Figure 6-9. NW5 on catalyst coated (left) and uncoated (right) substrates. ................... 245
Figure 6-10. ZnO nanostructures in low temperature region on gold coated silicon ..... 248

Figure 6-11. Raman spectra of Zinc oxide nanowires on alumina and silicon substrates
showing characteristic peak at 437cm‘1, tacc=20 s. ......................................................... 249

Figure 6-12. Raman spectra of 21mm gold coated ZnO nanowires on Si 100 substrate,
tacc=15 5. Insert shows the SEM image of substrate. ...................................................... 251

Figure 6-13. Raman spectra of ZnO combs on Si100, tacc=15 5. Insert shows SEM image
of the substrate. ............................................................................................................... 251

Figure 6-14. Ordinary Raman spectra of 3 uM ethanolic solution of R6G showing only
the ethanol peaks. ............................................................................................................ 253

Figure 6-15. Raman spectra of 3 uM R6G from 21 nm Au coated ZnO nanowires
Synthesized on Si(100) substrate ..................................................................................... 253

Figure 6-16. Raman spectra of ZnO nanowire. (a) Before adding NB. (b) After adding
NB. .................................................................................................................................. 255

Figure 6—17. Raman spectra of 2.73 uM ethanolic NB from ZnO nanowires coated with
60 nm gold colloidal at two locations, tacc=15 s. ............................................................ 256

Figure 6-18. Effect of gold coating thickness of SERS from SAM of 1,2-BDT on ZnO
Nws, tacc=15 s. ............................................................................................................... 258

Figure 6—19. Effect of gold coating thickness of SERS from SAM of BDT on ZnO NWs,
tacc=15 s. ......................................................................................................................... 258

xix

Figure 6-20. Effect of gold coating thickness of SERS from SAM of 4-MBT on ZnO

NW5, tacc=15 s. ............................................................................................................... 260
Figure 6-21. Effect of gold coating thickness of SERS from SAM of MBT on ZnO NW5,
tacc=15 s. ......................................................................................................................... 260
Figure 6-22. ZnO NW5 substrate. (a) Before 4-MBT soak. (b) After 12 hr soak in 1 mM
MBT. ............................................................................................................................... 261
Figure 6-23. SERS spectra of a SAM of 4-MBT from 5 adjacent points 250 um apart,
tacc= 15 s. ........................................................................................................................ 262
Figure 6-24. 4-MBT 1068 cm'1 Raman peak intensity distribution for randomly chosen
area on substrate, tacc= 5 s. ............................................................................................. 263
Figure 6-25. Ordinary Raman spectra of 4-MBT and polystyrene, tax: 5 s. .............. 266

Figure 6-26. SEM image of two typical morphologies of ZnO NW5. (a) Randomly
oriented. (b) Partially aligned. Scale bar is 2 pm in both images. .................................. 267

Figure 6-27. Comparison Of SERS signal from random and partially aligned ZnO NW5
for SAM of 4-MBT, tacc=15 s. ........................................................................................ 268

Figure 6-28. SERS from SAM of 4-MBT for (a) randomly distributed NW5, tacc=15 s.
Cb) Partially aligned NW5, tacc=15 s. .............................................................................. 269

Figure 6-29. Cross-section for ZnO nanowires. (a) Randomly distributed. (b) Partially
aligned. ............................................................................................................................ 270

Figure 6-30. ZnO nanoparticles in low temperature region of reactor and SERS spectra
0f SAM of MBT after gold coating of particles, tacc=5 s. .............................................. 271

Figure 6-31. SERS spectra of self-assembled monolayer of 1,2-BDT ﬁom Au coated
ZnO nanowires, tacc= 15 5. Also shown is spectra before analyte application. .............. 272

Figure 6—32. Successive SERS spectra from 1mM BDT along a line on substrate, tacc=15
S . .. ................................................................................................................................... 273

Figure 6-33. Area map of 1028 cm'1 Raman peak of 1,2-BDT, tacc=5 s for each point.273

Figure 7-1. (a) CAD experimental set up schematic. (b) Optical image of Silver nanorods
slibstrate (0) Optical image of copper nanorods substrate. ............................................. 278

XX

 

Figure 7-2. SEM images Of nanorods synthesized on glass substrate. (a) Aluminum
nanorods. (b) Copper nanorods. (c) Silver nanorods. (d) Silver nanorods showing

shadowing effect. ............................................................................................................ 280
Figure 7-3. Aluminum NR5 diameter and length distribution. ..................................... 281
Figure 7-4. Copper NR5 diameter and length distribution ............................................ 282
Figure 7-5. Silver NR5 diameter and length distribution. ............................................. 283

Figure 7-6. EDS spectra and quantization results for aluminum nanorods substrate. 284

Figure 7-7. EDS spectra and quantization results for copper nanorods substrate .......... 285
Figure 7-8. EDS spectra and quantization results for silver nanorods substrate ............ 286
Figure 7-9. Silver nanorods that grew on sample clips (clips made of steel). ............... 287

Figure 7-10. Raman spectra of silver nanorods, tacc = 5 5. Insert shows SEM image of
silver nanorods. ............................................................................................................... 290

Figure 7-11. Raman spectra of smooth silver ﬁlm and glass slide, tacc =5 5. ................ 290

Figure 7-12. Raman spectra of copper and silver tipped aluminum nanorods, tacc 5 5.. 291

Figure 7-13. SERS spectra from Ag nanorods substrates before and after applying 3 11L

Of 5 uM ethanolic NB, tacc 15 5. Insert shows ordinary Raman spectra Of 170 mM NB in
ethanol with only weak 582 cm'l peak identiﬁable. ....................................................... 292

Figure 7-14. SERS spectra from Ag nanorods substrate from different spots for 5 uM
ethanolic NB, tax: 5 s. ................................................................................................... 293

Figure 7-16. Raman spectra from silver tipped aluminum nanorods before and after
application of 100 uM 4-MBT, tacc= 15 s. ..................................................................... 295

Figure 7-17. MBT SAM spectra before and after soaking aluminum nanorods in 0.1
mM methanolic for 14 hrs, tacc= 5 s. .............................................................................. 295

Figure 7-18. 4-MBT 1071 cm'1 Raman peak intensity distribution from silver tipped
aluminum nanorods for randomly chosen area on substrate, tacc= 5 5 ............................ 296

Figure 7-19. Raman spectra from copper nanorods before and after formation of SAM
0fMBT, tau:= 15 s. ......................................................................................................... 297

Figure 7-20. Raman spectra of from Silver nanorods substrate before and after formation
0f SAM of 4-MBT, tacc= 15 s. ........................................................................................ 298

xxi

has
5‘

 

Figure 7-21. 4-MBT 1068 cm'I Raman peak intensity distribution for a randomly chosen
area on a silver nanorods substrate, tacc= 5 s. ................................................................. 298

Figure 7-22. (a) Optical constants of silver as ﬁrnction Of incident wavelength[76].
Circles correspond to laser wavelength used in this work. (b) Theoretical enhancement
factor for an isolated silver nanorod as ﬁmction of aspect ratio. .................................... 304

Figure 7-23. (3) Optical constants of copper as function of incident wavelength[76].
Circles correspond to laser wavelength used in this work. (b) Theoretical enhancement
factor for an isolated copper nanorod as function of aspect ratio. .................................. 306

Figure 7-24. Raman spectra of aluminum nanorods before and after application of
formation of SAM of 1,2-BDT, tacc= 15 s. ..................................................................... 307

Figure 7-25. Raman spectra of silver nanorods before and after application of 20 uM 1,2-
BDT, tacc= 15 s. .............................................................................................................. 308

Images in this dissertation are presented in color.

xxii

Chapter 1: Introduction

1.1 Raman spectroscopy

The understanding and control of interaction between light and matter has always
been an area of active research due to its potential applications in many ﬁelds of science
and technology. Vibrational spectroscopic techniques have been extensively employed to
study this interaction and have played a major role in extending our understanding of
structure, bonding, and reactivity in all phases of matter. The main techniques to detect
molecular vibrations are based on the processes of infrared (IR) absorption and Raman
scattering. These are widely used to provide information on chemical structure of matter,
identify substances using their characteristic spectral patterns, and to determine
quantitatively or semi-quantitatively the amount of substance in a sample. Samples can
be examined in a whole range of physical states; for example, as solids, liquids or vapors,

in hot or cold states, in bulk, as microscopic particles, or as surface layers.

Raman spectrum is considered as the optical ﬁngerprint of chemicals and
bimolecules as it represents the vibrational frequencies of functional chemical bonds in
molecules[1]. Raman spectroscopy offers a number of advantages due to its ability to
examine samples inside glass containers without any special preparation and in the
Presence of common solvents like water. Being a completely non-detructive and non-
intrusive technique, it permits the detection and identiﬁcation Of unknown species in any
phase. As an optical technique it has high resolution (1 cm") and immunity to the
presence of an ambient gas phase. On the more fundamental side, it provides spectra

Complementary to those obtained by either infrared absorption or electron energy loss

 

 

spectroscopy[2]. Raman scattering depends upon the change in the molecular
polarizability during a molecular vibration, while infrared absorption depends upon the
change in the dipole moment. High resolution, wide spectral range, high sensitivity,
complementary selection rules, and high spatial resolution make Raman spectroscopy an
ideal probe for surface sciences. It also ﬁnds its usage in a wide variety of ﬁelds
including examination and identiﬁcation of inorganics, minerals, art, archaeology,
polymers, emulsions etc. Raman scattering is extensively employed in biological,
pharmaceutical and forensic applications to study physical structures of the live cells, and

for plant control and reaction following[3].

l .2 Limitations of conventional Raman spectroscopy

Despite its numerous advantages, conventional Raman scattering is less widely
used than infrared absorption. The major shortcoming of the technique is its extremely
Small scattering cross section as compared to the Rayleigh or elastic scattering. Typical
Raman scattering cross sections of molecules are in the range of ~10'29 cm2 whereas
typical Rayleigh scattering cross sections are ~ 10‘26 cm2 [4]. The scattering cross section
of a particle is deﬁned by relating the rate of photons striking a molecule to the rate of
Scattering in all directions and can be viewed as an area presented by a molecule for the
Scattering of incident photons[4]. Thus it might appear that the application of such an
intrinsically inefﬁcient process to the detection of trace amount of species would not be
Very productive. In addition to low signal intensity, Raman spectroscopy is also plagued
by fluorescence[5]. Fluorescence is a strong light emission from the sample (including
Sample impurities) or its surrounding. These broadband ﬂuorescence signals can be of

much higher intensity than the Raman signals and may altogether obscure the Raman

‘nw
-

on-

1.4

s.-

 

 

 

 

spectra (typical ﬂuorescence cross sections of molecules is ~ 10'19 cm2)[4]. It is estimated
that at least 40% of the samples used in Raman spectroscopy suffer from some degree of
ﬂuorescence. Techniques such as time resolved spectroscopy, excitation using deep
ultraviolet (UV) and metal ﬂuorescence quenching have been studied to tackle
fluorescence problem but each has its associated problems like sample degradation from
UV irradiation, spectra perturbation by metal etc. The chance of an unknown sample
exhibiting ﬂuorescence is strongly dependent on the wavelength of the laser used for
excitation. Typical laser wavelengths are 780 nm, 633 nm, 532 nm, and 473 nm, although
others are common. Fluorescence from the sample or its surrounding can be signiﬁcantly
reduced by moving the excitation all the way to the near infrared (for example, 785 nm
will yield considerably less ﬂuorescence than 532 nm). But on the other hand the
efﬁciency of Raman scattering is proportional to UV, so there is a decrease in
enhancement as the excitation laser wavelength becomes longer. Hence a balance needs
be maintained between enhancements from shorter wavelength versus accompanying
ﬂuorescence. Near-infrared (IR) laser excitation greatly reduces the number of samples
Prone to ﬂuorescence and allows higher laser powers to be used without
Photodecomposition. This makes Near-IR excitation a natural choice for the study of
ﬂuorescent prone samples but the low sensitivity due to small scattering cross-section

remains a major problem limiting the widespread application of Raman spectroscopy.

1 .3 Surface enhanced Raman spectroscopy
The Observation of Surface-Enhanced Raman Spectroscopy (SERS) by

Fleishchman et al. in 1974 reinvigorated interest in Raman scattering as a powerful

.r

.u.

a.

‘I

analytical tOOl[6]. The Raman signals from pyridine molecules adsorbed on the
electrochemically roughened silver electrode were found to be over a millionfold stronger
than their corresponding surface densities. The initial enthusiasm following the discovery
of SERS lasted for over a decade and many theories were proposed to explain the strong
enhancement Observed from rough surfaces and colloidal aggregates. Today, it is
generally agreed that the SERS arises from two mechanisms — an electromagnetic
enhancement and a chemical charge transfer effect. In electromagnetic enhancement
model, the analyte located in the close proximity to the metal interacts with the incident
laser light through the excitation of surface plasmons in metal[7]. The metal can be in the
form of a thin rough ﬁlm with nanoscale features or in the form of aqueous colloidal
nanoparticles. Silver, gold and copper are the common metals employed due to their
strong SERS activity. In the theory of chemical enhancement, the analyte chemically
bonds to the metal and excitation is through the transfer of electrons from metal to
molecule and back to metal again[8]. There is evidence for both mechanisms contributing
to SERS but it is difﬁcult to isolate one from the other in actual measurement. However,
it is widely accepted that electromagnetic enhancement plays a more dominant role than
the chemical enhancement. The work in SERS reached a plateau about ten years ago but
became reinvigorated again by the reports of Kneipp and coworkers[9-13] and Nie and
coworkers[14-18] of single molecule detection. Both independently and simultaneously
demonstrated that under favorable circumstances it is possible to detect single molecule
based on its SERS spectra. The reported enhancement factors are of the order of 1014
With an effective Raman cross-section of 10'16 cmZ/molecule at near-infrared non-

resonant excitation for molecules in colloidal silver solution[19].

1

\. .
i T.

w.

‘-

I

The discovery of SERS has Offered an unprecedented sensitivity and selectivity to
Raman spectroscopy. Identifying molecules at very low concentrations is critical for
many analytical applications such as forensics, medical diagnostics, drug discovery, and
chemical development. This giant enhancement of Raman signal offers the potential of
developing an ultrasensitive sensing platform based on SERS with molecular (especially
bimolecular) identiﬁcation capabilities. SERS is an attractive tool for biomedical
applications because it has several advantages over competing optical techniques. Some
of these advantages are: (i) minimal sample preparation; (ii) potential for remote
sampling through ﬁber Optics; (iii) fast analysis; (iv) weak Raman scattering of hydroxyl
group. This implies that water and silica do not manifest themselves as serious sources of
background noise. Thus biological samples in aqueous media can be analyzed without
any interference from water or glass containers. A5 a result, a number of biological
applications using SERS have been reported. Kneipp has demonstrated SERS nanoprobes
utilizing gold nanoparticles for probing intrinsic chemical environment in living
cells[20]. Yonzon e. al. reported a SERS based blood-glucose sensor that can be used for
real-time continuous sensing[21]. The sensor utilized a mixed self-assembled monolayer
(SAM) of decanethiol (DT) and mercaptohexanol (MH) on silver ﬁlm over nanospheres
(AgFON). The purpose of SAM was to protect oxidation of silver and to provide for the
binding of analyte of interest to the AgFON. The silver nanospheres based SERS was
able to efﬁciently detect and quantify glucose in physiological concentration range in the
presence of interfering analytes and in bovine plasma. The sensor showed temporal
Stability for at least 10 days in bovine plasma, making it a potential candidate for

implantable sensing. Similarly Vo-Dinh and coworkers have successfully demonstrated

 

 

 

SERS based gene probes for the detection of HIV DNA, p53 cancer gene, and
hyperspectral imaging of Raman dye-labeled silver nanoparticles in single cells[22, 23].
Alak and Vo-Dinh were the ﬁrst to demonstrate trace level detection of
organophosphorus chemical agents using SERS[24]. Since then, SERS has also been
applied for the detection Of chemical agents like cyanide, VX, Sarin etc., and their
hydrolysis products in water[25]. Many researchers have reported successful detection of
trace amount of bacillus anthracis (anthrax) spores using SERS[26-29]. SERS has also
been utilized to detect common bacteria like E. coli that cause intestinal and extra-

intestinal infections[30].

1.4 Common SERS substrates

The initial Observation of SERS was from metal electrodes roughened by
oxidation-reduction cycles. Later SERS was also reported from silver and gold colloidal
nanoparticles. Currently the colloidals are one of the most widely used substrates for
SERS due to the ease of preparation. However, the techniques exploiting colloids or
roughened surfaces prevent the control of the spacing and periodicity of the features. This
leads to the random and strong variations in the SERS signal from these substrates that
limit their widespread use for analytical applications. Numerous efforts have been
devoted to the search for a homogeneous, repeatable, and mass producible SERS
Substrate that can yield enormous enhancement factors observed in random “hot spots”.
Various forms of nanostructures have been explored to enhance SERS effects, such as
rough metallic surfaces by chemical etching[3l], silver films on TiOz[32], colloidal silver

nanoparticles[15], silver nanoparticle arrays fabricated by nanosphere lithography[33],

electrO-deposition Of silver on silver ﬁlms at high potential[34], aligned monolayer of
silver nanowires[35] and aligned silver nanorods by oblique angle deposition[36, 37].
However, many Of these methods are either expensive or time consuming, and it is not
easy to make reproducible substrates of the correct surface morphology to provide
maximum SERS enhancements. Assembling metal nanostructures synthesized by a
bottom-up approach, such as by Langmuir- Blodgett ﬁlms or by casting metal colloids
onto surfaces, can achieve quasi-regular arrays of metallic nanoparticles or nanowires
with small separation. However, most of the bottom-up approaches usually require a
monolayer coating of surfactants for dispersion during synthesis. This added surfactant
can interfere with the Raman signal of the analyte. Other techniques, such as nanosphere
lithography (N SL) uses self assembled and close-packed nanospheres as the lift-off layer
for evaporated Raman active metal (usually Ag or Au)[38]. This technique also fails to
provide the necessary nanometer-scale gaps between metallic nanostructures formed

between nanospheres following lift-off.

l .5 Major challenge for SERS

The discovery of SERS effect promised to expand the applications of Raman for
the detection of very low concentrations of molecules. The potential of SERS has been
demonstrated by single molecule detection. However, until very recently, the common
SERS substrates (typically colloids or roughened surfaces) gave highly variable signals
and severely limited acceptance of SERS as a viable analytical technique. As a result, the
dream of an ultra-sensitive ﬁeld deployable robust SERS sensor is still far ﬁom reality.

All the giant Raman enhancement demonstrations reported have come from few hotspots

g-”

on the substrate[15]. In the study of SERS from silver (Ag) nanocrsytal aggregates, J iang
and coworkers noted that the hotspots are mostly compact, nonfractal aggregates of Ag
nanoparticles and only fewer than 1% of the aggregates give detectable SERS
activity[39]. The hotspots yielding high SERS activity are interspersed with large areas
of little or no enhancement. This leaves the analyst with the tedious task of manually
exploring the surface of the substrate for these hotspots and subsequent detection of
analyte. Mostly these hot spots result from natural and uncontrollable aggregation of
nano-particles at random locations on substrate that make the SERS measurement highly
erratic. The lack of reproducibility has effectively limited SERS to the research
laboratory. A way around this problem is engineering substrates that have uniform
morphology imitating such hotspots. The difﬁculty of fabricating such substrates can be
gauged from the order of dimensions needed for reasonable SERS enhancement. Both the
theoretical and experimental studies suggest that a separation Of less than 5 nm between
aggregated nanoparticles is necessary for strong SERS enhancement[39]. But this scale of
Controlled separation is far beyond the resolution of any scalable top-down fabrication
approach[40]. Today main issue facing the SERS community is fabrication of a robust,
economical and scalable substrate that can provide high sensitivity with good uniformity

and repeatability over large area.

1 .6 SERS from Nanowires : Motivation for this work
SERS has the potential to greatly enhance our current capability of trace chemical
analysis. This can have signiﬁcant impact on early medical diagnosis and treatment. Also

it can be an effective tool in monitoring the food supply chain against viral and bacterial

:1--

contaminations. At the same time, the ability to detect chemical and nerve agents will
help ﬁght the threat of bio-terrorism. However, the full potential of SERS can be only
realized after developing techniques for economical fabrication of a homogenous SERS
substrate. This research has been motivated by the desire to fabricate an economical and.
reproducible SERS substrate based on different types Of nanowires and nanorods. The
terms nanowire and nanorod are often used interchangeably to denote high aspect ratio

(AR) nanostructures. In this dissertation, the term nanowire is used to denote the

length

nanostructures having lengths much larger than diameter (i.e., AR = >>1). The

diamter

nanorods are also high aspect ratio nanostructures but their aspect ratio is less compared
to nanowires (typically AR S 10). The use of these high aspect ratio nanostructures such

as nanowires and nanorods for SERS has several advantages. These include:

1. The high aspect ratio Of these nanostructures yields strong electric field

enhancement.

2. The surface properties of these nanostructures are highly reproducible and
well deﬁned as compared to colloidal systems.

3. The synthesized nanowires/nanorods offer many unique features (sharp
vertices, noncircular cross-sections, inter-wire coupling) that may lead to
larger ﬁeld enhancement factors, offering higher sensitivity under optimal
conditions[4l, 42].

4. High density of nanowires/nanorods means close interaction between
adjacent nanowires and hence strong inter-wire coupling, which enables

SERS to manifest for a broad selection of excitation sources.

9

5. The nanowires/nanorods can readily be used for molecular detection
either in air or in aqueous solution environment.

6. The semiconducting nanowires Of zinc oxide and germanium dioxide are
high refractive index materials, which can provide strong light

conﬁnement for improved SERS effect.

The major contribution of this dissertation to the ﬁeld of SERS is an attempt to
investigate oxide nanowire based substrates for potential use in detection of chemicals
using Raman spectroscopy. This includes both the fabrication and characterization of the
nanowire substrate for application as SERS platforms. It deals with bulk synthesized
semiconducting nanowires of germanium oxide and zinc oxide to yield SERS. The
germanium oxide nanowires were initially synthesized at high temperature (~850 °C).
Subsequently a novel low temperature (~650 oC) synthesis was developed to achieve
better control over the parameter of substrates. The as-synthesized oxide nanowires are
not Raman active since only noble metals yield appreciable SERS. Gold coating of the
nanowires has been employed as SERS active interface to support surface plasmons.
Selection of gold is based on its chemical inertness, compatibility with biological systems
and good optical properties for near IR laser excitation. The chemical functionality can be
easily added to the gold coated substrates by exploiting well-established surface
chemistry protocols. The coating of SERS substrates with ligand molecules having
Speciﬁc terminal groups is extremely effective to boost the selectivity of the SERS
Substrate. This functionalization of SERS substrates is particularly useful in the case of

complex molecules such as proteins or for molecules with low afﬁnity to metal. The

10

 

LL\
5

 

comparison of SERS performance of nanowires with that of commercially available
Klarite chip showed a stronger enhancement from nanowires. The dissertation also
investigates synthesis of aluminum and copper nanorods using well-established oblique
angle deposition technique. The SERS performance of theses quasi-aligned nanorods is

evaluated using model molecules and compared with those of commonly employed silver

nanorods.

l .7 Organization of this dissertation

This dissertation is organized into 7 chapters. The dissertation begins with an
introduction to the theory of Raman scattering and SERS in chapter 2. The experimental
set up for the preparation of SERS substrates and their Raman characterization is
described in chapter 3. Synthesis and Raman studies of high temperature fabricated
germanium oxide nanowires is presented in chapter 4. Results of Raman analysis using
low temperature germanium oxide, zinc oxide, and silver nanorods are presented in

Chapter 5, 6, and 7 respectively. Finally chapter 8 presents a summary of results and

outlines some future tasks.

ll

Chapter 2: Theory of Raman Scattering

When monochromatic radiation scattered from a system such as gas, liquid or
solids is analyzed, most of the light has same frequency (hence wavelength) but some
fraction is found to have different frequency than the incident light. Such scattering of
light with a change of frequency is known as Rarnan scattering, after Indian scientist C.
V. Raman who, with K. S. Krishnan, ﬁrst Observed this phenomenon in l928[43]. The
elastic scattering from scattering centers like molecules that are very small compared to
the wavelength of incident radiation is called Rayleigh scattering. Another type of elastic
scattering from larger objects (size greater than wavelength of incident radiation) like a
dust particle is also possible and is referred as Mie scattering[44]. Often the term
Rayleigh scattering is used to represent both Rayleigh and Mie elastic scattering[45].
Inelastic scattering originating from the Doppler effect with a very small frequency
change (wavenumber change of the order of 0.1 cm!) can also exist. This type of
Scattering, predicted by Brillouin in 1922 and experimentally veriﬁed by Gross in 1930,
is called Brillouin scattering[46]. However, due to very small wavenumber change,
Brillouin scattering is not separable from the incident radiation under the experimental
Conditions used for Raman spectroscopy.

Light scattering arises from the dipole moments induced in molecules by the
inCident ﬁeld, through the polarizability of the electrons. Polarizability is the ability of
the electron cloud to be distorted by the electric ﬁeld of the incident radiation. The more
polarizable a molecule is, the more the electron clouds of its bonds are distorted by the

1nCident radiation. The electric ﬁeld Of the radiation interacts with the electron cloud of

12

the bonds Of molecule in the sample and this induces a temporary dipole moment in the
bond[47]. The molecular bond emits the scattered light (Rayleigh or Raman scattered) as
it relaxes to a low energy unexcited state. Rayleigh scattering is a result of static
polarizability whereas Raman scattering follows from the modulation of polarization by
electronic, vibrational or rotational motions[45]. Rayleigh scattering, being at the
frequency of the incident radiation, does not convey any useful information but Raman

scattering provides information about the vibrational states of the molecule.
2.1 Classical theory of Raman scattering

A simple classical model based on electromagnetic theory of radiation can be
used to gain insight into the light scattering phenomena. Consider a laser beam
illuminating the sample at frequency v0 with corresponding amplitude of electric ﬁeld
given by

E = E0 cos(27rv0t) (2.1)

The electric ﬁeld can induce a dipole moment ,u in the molecule according to

,u = aE (2.2)

where a is the polarizability of the molecule. At a molecule’s equilibrium nuclei

geOmetry, the polarizability has certain value, (10. As the molecule vibrates, the distance

13

between nuclei changes from that of equilibrium. At some distance, Ar, away from the

equilibrium position, the instantaneous polarizability is given by

a = a0 +[a—a] Ar (2-3)

6r

Here the derivative term represents the change in the polarizability of the
molecule with change in position. Suppose the molecule is vibrating or rotating in some

sinusoidal manner with maximum displacement of rmax , then Ar can be expressed in

terms of vibrational frequency vm as
Ar = rmax cos(27rvmt) (2.4)

Inserting equation (2.4) into equation (2.3) gives
8a
a = a0 + rmax (E—JCOSQm/mt) (2.5)
r

Substituting the value of u and E from equation (2.5) and (2.1) respectively into

(2.2), the time dependence Of the dipole moment becomes

14

,u = aOEO [1+ rmax (gjcosaﬂvmtﬂcosaﬂvot)
1 do:
= aOEO cos(27rv0t) + —2— Eormax E cos(27r(v0 + vm )t) (2.6)

+ l 1204,,ax [ﬂjcosQﬂr/O — vm )r)
2 Or

The ﬁnal form of equation (2.6) shows that the scattered light has been modulated
by the oscillating polarizability. The ﬁrst term in the above equation corresponds to the
Rayleigh scattering at the unshifted frequency. The second term with higher frequency,

v0 + vm , is the anti-Stokes Raman scattering and third term with lower frequency,
v0 — vm , is the Stokes Raman scattering. This simple classical treatment also provides a

selection rule for Raman scattering: the molecular polarizability must change as a
ﬁrnction of nuclear motion during a vibration for that vibration mode to be Raman
active[2]. However, it is deﬁcient in that it does not take into account the quantized
nature of vibrations and lacks any information about intensities of scattering. Quantum
theory on the other hand is able to better explain the relationship between molecular
PI‘Operties and Raman scattering. Nonetheless a simple insight into Raman scattering can
be gained using a simple classical model of the wave nature of light.
Figure 2-1 shows a pictorial representation of the Rayleigh and Raman scattering.
The states marked 0, 1, 2 etc., are different vibrational states (vibronic states) of the
gr Ound electronic state. The incident light considered as an oscillating dipole interacts
With the molecule and polarizes the cloud of electrons. This transfers the energy of the
light into the molecule, which is promoted to a higher energy virtual state as, indicated by

uInward arrows in Figure 2-1. This interaction can be considered as the formation of a

15

very short-lived ‘complex’ between the light and the electrons in the molecule[3]. This
high-energy state, often called virtual state, has different electron geometry compared to
the ground state and is characterized by the fact that the nuclei have not moved
appreciably. This virtual state is a real state of the transitory ‘complex’ formed between
light and molecule. This is an unstable state and energy is released almost simultaneously
in the form of scattered radiation as annotated by the bold downward arrow in Figure
2-1. A word about the virtual state: since it has different electron geometry than the static
state and the nuclei have not moved to reach a new equilibrium state to ﬁt distorted
electron arrangement, this state does not correspond to any of the electronic states of the
molecule. The laser deﬁnes the extent of the distortion of electron cloud and the energy
of the virtual state. As a result, this process is not quantized and the energy of the
molecule can assume any of an inﬁnite number of states depending on the frequency of
radiation of the source. The Rayleigh scattering is the most intense form of scattering and
is essentially an elastic process. This happens when the electron cloud relaxes without
any nuclear motion. The Raman scattering (Stokes and anti-Stokes) happens when light
and electrons interact and the nuclei begin to move at the same time. Since nuclei are
much heavier than electrons, there is appreciable change in the energy of the molecule. If
the molecules start with a ground state and ﬁnally relaxes to an excited ground state, then
it has effectively absorbed the energy difference between two states. The scattered
radiation is Of longer wavelength than the excitation wavelength and the process is

termed Stokes Raman scattering.

16

 

 

f——

V'—

  

,Esc = hv

Raleigh
scattering
Virtual 3' --------------
states
Laser
excitation,
E = 111/

 

Vibronic states; 2
of ground if 1

 

electronic statef
3.. 0

 

I
0

Raman scattering
)\ '
____________ ’ \...........~

\

\
Anti-stokes
scattering

Stokes scattering

......_..~..__ . n.- ...-__..

Esc =hv+AE

 

 

 

 

 

 

Figure 2-1. Pictorial representation of Rayleigh and Raman scattering[47].

On the other hand, if the molecule was initially in an excited state and makes a

transition to ground state, the scattered radiation is of shorter wavelength and the process

is called Anti-Stokes Raman scattering. At room temperature, most molecules are likely

to be in their lowest energy electronic state or ground state ‘0’ represented by bold

horizontal line. Thus the intensity of Stokes Raman scattering will be much higher than

the anti-Stokes scattering. The ratio of the Stokes and anti-Stokes Raman scatterings

depends upon the population of molecules in the ground and excited state. This can be

Calculated using Boltzmann’s equation

ﬂg=§£exp[
N0 80

"(Eex " E0)

.T l

(2.7)

17

rJarerrnunfwéieﬂruv .m. Fur-Jive» a.”

   

5‘6

 
    
   

 

I‘P!’QI\OI'€O"I)$I5!QI.Jol'u gm!-
.._7.. .f.. .l I... . - ... p

 

1"3'
5..

J"...

*‘-i~.
.

‘i
(I:

where

Nex is the number of molecules in the excited vibrational energy level,

No is the number of molecules in the ground vibrational energy level,

g is the degeneracy of the energy levels,
Eex —E0 is the difference in energy between the excited and the ground

state,

k is the Boltzmann’s constant (1.3807 X 10 ‘23 JK").

The intensity of Anti-Stokes band is expected to increase with increasing
temperature as a larger fraction Of a target molecules are expected to be in the ﬁrst
vibrationally excited states under such circumstance. It may be worth reiterating the
fundamental differences between Raman scattering and the IR absorption. Although the
Raman scattering and the IR absorption of a molecule are dependent upon the same
vibrational modes, they arise from processes that are inherently different. The selection
rule for IR absorption requires that the dipole moment must change for a molecular
vibration to be IR active. Raman scattering on the other hand, involves a momentary
distortion of the electron distribution of a bond in a molecule, which is then followed by
reemission of the radiation as the bond returns to one of its ground electronic states. The
hOHIO-nuclear molecules such as nitrogen or oxygen provide a good example to illustrate
this difference. These molecules do not possess any dipole moment in equilibrium
Position or under bond stretching vibration. Thus the bonds of homo-nuclear molecules
are not infrared active as no dipole moment can be induced by the incident radiation.

HOWever, these bonds are Raman active since the polarizability of the bond between the

18

 

 

atoms in such a molecule varies periodically. The polarizability of the bond changes in
phase with the stretching vibrations, reaching a maximum at the greatest separation and a

minimum at the closest approach of the two nuclei.

Symmetrical Bending or Asymmetrical
stretch, y] deformation, yz stretch, 73

«v» @W-v

Figure 2-2. Three modes Of vibrations of C02 molecule[3].

Figure 2-2 shows the three vibrational modes of C02. The symmetric stretches of
C02 molecule are Raman active because polarizability clearly changes during the
vibration. However, symmetric stretching is infrared inactive since there is no change in
dipole. On the other hand, the asymmetric stretches in the molecule are Raman inactive
because there is no change in the polarizability of the molecule. Since bond polarizability
are additive, the polarizability change due to compression of one C=O bond is cancelled
by the Stretching of the other C=O bond. The asymmetric stretch is infrared active
because when one of the oxygen atoms moves toward the carbon and other away from it.
This results in a change in dipole moment, as there is a net change in charge distribution.

Similar arguments apply to the bending vibration, which is IR active but Raman inactive.

The intensity of the Stokes Raman scattering is given by [3]

I = Klaza) (2'8)

19

 

 

where K consists of constants such as speed of light, I is the laser power, a) is the
frequency of incident radiation and a is the polarizability of the electrons in the molecule.
This suggests that higher laser power and shorter wavelength will yield higher Raman
intensities. But the problem of sample degradation, photodecomposition and ﬂuorescence
puts an upper limit on practically usable laser power and wavelength. The intensity has
quadratic dependence upon the polarizability of the molecule. The polarizability is a
function Of the shape and size of the molecule. It usually varies with spatial direction and
is independent of the permanent dipole moment (if present) in a molecule. It is an
anisotropic property of a molecule and hence the scattered Raman intensity shows a
dependence upon the direction and polarization of the incident and scattered radiation.
Therefore, polarizability is often expressed as a tensor to take into account the possible
variation in polarization. The polarizability is calculated quantum mechanically by

perturbation theory or by time dependent theory.

2.2 Quantum mechanical theory of Raman scattering

In quantum mechanics, the Raman scattering is explained by the Kramers
Heisenberg Dirac (KHD) equation derived by Krammers and Heisenberg[48], and
Dirac[49]. The scattering process is described as an excitation to a virtual state (lower in
energy than a real electronic transition) with nearly coincident de-excitation to a different
Vibronic state of ground electronics state. The difference in energy between two Vibronic
states manifests as a change in the wavelength of the scattered radiation. The scattering
event occurs in 10'14 seconds or less. The KHD expression describing the molecular

polarizability tensor for the transition from ground state to ﬁnal state is given by[3]

20

 

 

(a...) _Z[<F|rp|1><1|ral0>+(IlrplG><FlralI>] (,9)

OF I 77001 -h(0L —iF, hwlp '1"th -iF,

Here G is the ground Vibronic state, I is the intermediate state (a Vibronic state Of

the excited electronics state) and F is the ﬁnal Vibronic state of ground state, (armI )GF is

the molecular polarizability tensor component for transition G —> F, and p and o are the
incident and scattered polarization directions. The operator r is the electric dipole
Operator, th is the incident photon energy, arc, is the angular frequency of transition
I —» G, h represents h/27r where h is Planck’s constant and F1 is the natural linewidth of

the intermediate state. The summation z is over all Vibronic states of the molecules
1

due to non-speciﬁc nature of scattering. The numerator in equation (2.9) uses ‘bra’ and

‘ket’ (<l and |>) nomenclature for integrals to simplify the expression. In the integral

< I IrUIG >, the |G> is a wave function to represent ground Vibronic state of the ground

electronics state. The operator r0 operating on |G> and subsequent multiplication by

excited state <1] mixes the two states to describe the distorted electron conﬁguration in
the virtual state. This describes in part the excitation process (transition from ground

Vibronic state to an intermediate state). A similar scattering process described by

< F Irpll > leaves the molecule in the ﬁnal state |F>. Since molecule can be in an

excited Vibronic state to start with, a second term in added in equation (2.9) that mixes
the excited and ground in a similar way. It may be emphasized that the mixing of states
(vibronic, excited and ground) employed in the KHD approximation models the virtual

state of the molecule at the instant it had captured light. In the denominator of equation

21

 

-~-¢ ‘0'- ‘

,. 7,-I'V

5“ § .
v'. ‘
b.» l ..
.‘ .9"
A.
—
.
t.'
I
7': f“
M n‘b

   

 

.
7‘. . ..
\_ .-
‘3."“
.. . ...
‘0. D
\J v
.
7
‘l
V n
1 ~
~L
:y
’..
. .
“~_ ‘ -
-“l
‘
.. ’
.~ ‘N,
'-.,
h
\.
R, ,.
Q‘.
i
. .
.~.
\
‘ L
‘1
‘
.
I l
‘4
P
’

(2.9), the if I term relates to the life time of excited state and is small in energy
compared to th and 11600,. This term keeps the expression from blowing up when the

laser energy approaches the energy difference between ground and the excited electronic
state. It is obvious from the denominators of equation (2.9) that the second term plays
smaller role in describing the polarization process and can be neglected.

The KHD expression models all possible transition from a ground state to a ﬁnal
state and it is very difﬁcult to directly compute polarizability from basic electronic
structure or even interpret spectra from KHD relation[50]. The KHD equation can be
simpliﬁed by splitting states into electronic and vibrational components using the Bom-
Openheimer approximation. In this approximation, one can separate the total wave

function (‘1’) into electronic (9), vibrational ((1)) and rotational (r) components as

.1, = 9(r, R).<D(R).r(R) (2.10)

where r and R are the electronic and nuclear co-ordinates respectively. The
separation works because the time scales for electronic transitions (10'13 sec or less) are
much smaller as compared to vibrational transitions (10'9 sec)[3]. Although rotational
effects can be seen in gas phase Raman, these are relatively weak and can be neglected.
The splitting up of the wave ﬁmctions allows for the integrals of KHD expression to be

split up as

<I|r0|G>=<61'(D1 IrJIOGﬂDG>=§01|r0|902£®1|r0|¢62 (2.11)

electronic vibronic

 

22

No appreciable nuclear movement can occur during Raman scattering due to

extremely small time scale of Raman scattering. Thus the electronic part of the wave

function can be written as
M,G(R)=(9,|r,,|90) (2.12)

Since nuclei will change position during vibrational transition, the electronic

dipole matrix elements M 10(R) of equation (2.12) are varying functions of nuclear co-

ordinates R. Thus movement MIG can be expanded about the equilibrium position R0

using Taylor series as

6M
M,G(R)=M,G(R0)+[ 6R“?

 

] R6 + higher order terms (2.13)
R0

8

where R8 is the normal mode co-ordinate operator of vibration mode 3. Similar

expressions are valid for each of the 3N - 6 possible normal co-ordinates. Normal co-
ordinates of a molecule make use of the natural directions of bonds and are those co-
ordinates in which all atoms vibrating go through the center of gravity of the molecule at
the same time. The energy of a molecule can be divided into a number of different parts
or ‘degrees of freedom’. Three of these degrees of freedom describe translation and
another three describe rotation except linear molecules, which have only two possible

rotations. Thus for a molecule of N atoms, the number of vibration degrees of freedom is

23

 

'I’

‘~-__

3N -6 degrees or 3N -5 in the case of linear molecules. Substituting the Taylor series

expansion into Bom-Openheimer modiﬁed KHD expression yields

(A-fm) ‘
(am) =M120(Ro)z <¢RF |¢RI ><¢R1 lCDRF >
OF

+M R M' R x
1 hmGI—th—il“, IG( 0) IG( 0)

 

 

 

Z<¢RF|R€I¢R1><¢RI|¢RG>+<¢RF|¢R1><CDR1iRs ¢RG>
J ha)”: +th —iF, J
(B-tverm)

(2.14)

Here M and M’ correspond to the zeroth order and 1St order terms in the Taylor
series expansion of equation (2.13). The A-term is known as the Franck Condon term or
the Albrechts A-term and describes the summation of the matrix elements coupling all
vibrational states of a molecule. The contribution from A-term of equation (2.14) is zero
due to “closure theorem”, which demonstrates that when all vibrational wave functions
are multiplied together, the ﬁnal answer is zero[3]. It therefore does not contribute

towards Raman scattering but only the Rayleigh scattering. The co-ordinate operator, R5

in the B-term (also known as the Herzberg Teller term or the Albrechts B-term) describes
the effect of movement along the molecular axis during vibration and appears because
correction terms M’ has been multiplied with the vibrational states. An important feature
of this operator is that the integral will only have ﬁnite values when there is one quantum
of energy difference between initial and ﬁnal states. Thus the B-term describes the
Raman scattering and, due to the existence of the operator, no overtones are available

from this term (overtones are observed only if higher order terms from Taylor expansion

24

\,

are utilized). It also follows from the product of integrals in the B-term that only

symmetric vibrations will be Raman active.

2.3 Surface enhanced Raman scattering

Surface enhanced Raman scattering is a phenomenon which is observed when a
molecule is adsorbed onto, or is close to a roughened metal surface[6]. Originally the
enhancement was observed on silver but since then a wide range of metal surfaces has
been used including noble, alkali and transition metals. However, silver, gold and copper
are the commonly used metals since these have absorption maxima in the visible region
and hence give the greatest enhancement in signal[51]. Many models were reported in the
early days of SERS to explain the enhancement and the mechanisms that cause
enhancements in SERS are still under debate. However, it is generally agreed that there
are two main mechanisms that contribute to SERS and each provides varying degree of
enhancement in parallel, i.e., multiplicatively[52-54]. These are the electromagnetic
enhancement[54] and a charge transfer or chemical enhancement[8]. There is evidence
that both these mechanisms affect the SERS enhancement and it is more appropriate to

consider SERS arising from a combination of the two effects.

2.3.1 Electromagnetic mechanism

The electromagnetic enhancement stems from the special optical properties of
noble metals to support surface plasmons. Plasmons are the collective oscillations of free
electrons and, if oscillations are restricted to the surface of the conductor, then these are

known as a surface plasmon. The ﬁrst observation of the existence of surface plasmons

25

dates back to 1902 when Wood reported a sharp decrease in the reﬂectivity of diffraction
grating at certain frequency that excited surface plasmons[55]. The surface plasmons are
associated with an enhanced electric ﬁeld at the surface and interaction of this enhanced
ﬁeld with the molecule constitutes the electromagnetic enhancement mechanism. On ﬂat
metal surfaces, surface plasmon excitation is prohibited by momentum conservation in
the plane of the surface[56] and hence atomically ﬂat metal surfaces do not show
signiﬁcant SERS activity[52, 53]. The momentum conservation rule does not hold for the
rough surfaces and surface plasmons can be excited[56]. Surface plasmons interact with
the incident light as determined by their dipole (and higher multipole) transition
moments[39]. Moskovits was the ﬁrst to suggest that the huge enhancement observed in
SERS was due to excitation of surface plasmons[7]. Creighton et al. compared the
wavelengths corresponding to the Mie extinction maximum for silver (or gold) aqueous
sol particles and Rarnan enhancement for pyridine molecules adsorbed on these
particles[57]. The sharp Raman enhancement at Mie extinction maximum was observed
and it was concluded that surface plasmon excitation is the underlying mechanism for
SERS. The electromagnetic enhancement models were simultaneously enunciated by
Gestren et al.[58—61] and MeCall et al. [62, 63], and expanded upon by Kerker et al.[64-
71]. The common geometries treated in these models include isolated sphere, isolated
spheroid, interacting spheres, nitrating spheroid, rough surfaces, hemispherical bumps,
and gratings[51]. In the ﬁrst such model, Moskovits treated the rough metal surface as
metal spheres placed on top of a ﬂat metal surface. The simplest form of this model is a
single isolated metal sphere, which explains, at least qualitatively, most of the physics

behind other more complicated systems like rough surfaces or colloidals.

26

2.3.1.1 Isolated single metal sphere

For a small isolated sphere of radius R with a complex wavelength dependent
dielectric constant 8(1) = 8'0.) + is"(2.) that is illuminated by a plane wave (wavelength
larger than the size of the particle), the sphere will develop an internal polarization, P(r) ,
oscillating coherently with the incident ﬁeld E0. This is schematically illustrated in

Figure 2-3.

Electric T
ﬁeld

    

e‘ cloud

Figure 2-3. Schematic illustration of plasmon oscillation for sphere

showing electron distortion in response to electric ﬁeld [72].

Polarization P(r) , also known as optical current, is the dipole moment per unit
volume. The internal polarization of the sphere creates an external (outside the sphere)
EM ﬁeld which can be represented by the ﬁeld of a point dipole p located at the center of

the sphere as[39]

27

 

8-6‘ 3
=47r£ s m R E 2.15
p 0 m[£+2£ J O ( )

m

where so is the permittiVity of free space and am is the dielectric constant of media
surrounding the sphere. In the electrostatic limit (radius of particle much smaller than

wavelength) without retardation, P(r) is related to the local internal electric ﬁeld Em (r)

and the incident ﬁeld E0 by[39]

P(r) = 80 (8 — 5m )El-n (r)

8-8 (2-16)
=3808m[8+2: JED
m

 

Here P and Em (r) are independent of R within the isolated sphere. The net

electric ﬁeld at the surface remains constant as an increase in p is balanced by the
increase in radius of sphere. The EM ﬁeld outside the particle is due to the incident ﬁeld
and the polarized sphere. It is maximum at the surface of the sphere along the particle

axis (deﬁned as the incident electric ﬁeld direction) and is given by

 

 

38
Esurf = 8 + 28 E0
’" (2.17)
= E where g = 36
g 0 £+2£m

28

This expression is independent of R as discussed above. The equation (2.17) has a
pole at the dipole plasmon resonance, which is deﬁned as the wavelength for which the
real part of the complex dielectric constant 8' approaches —2. At the plasmon resonance,
the electric ﬁeld at the surface will become very large as real parts of denominator of

equation (2.17) cancel each other (em =1 for air) and magnitude of Esurf depends upon

imaginary part of sphere’s dielectric constant a", which is relatively small at plasmon
wavelength for noble metals. In fact, the best SERS metal, silver, has the smallest a" and
therefore yields the highest enhancement. Gold and copper have the next lowest values
for a" in that order and their SERS enhancement factors reﬂect this. At plasmon

resonance, g >> 1 and as a result Esurf >> E0. The adsorbed molecule on the sphere
surface will experience a very large ﬁeld Esurf and will yield an enhanced Raman
scattering signal. Quantitatively, the Raman intensity is proportional to the square of the

electric ﬁeld i.e., [SERS at E2 or I SERS oc g2. Generally Raman scattered radiation
frequency is not largely different from the excitation radiation and it also lies within the
plasmon enhancement region. This means that the sphere will also enhance the scattered
radiation ﬁeld in exactly the same way ie, I scattered oc Escauered2 oc g'2 where g' is the

enhancement of scattered ﬁeld. The ‘SERS enhancement’, G , is deﬁned as the ratio of
the Raman scattered intensity in the presence of the sphere to its value in the absence of

the sphere as [73]

(2.18)

 

 

29

where aRO is the Raman polarizability of the isolated molecule. Normally g 3 g’ and the

enhancement factor shows a fourth power dependence on the enhanced local ﬁeld. This
fourth power dependence is the key to the inordinate SERS enhancement. For silver,
g: 30 at 400 nm but still provides a Raman enhancement of G 2 8x105, assuming no
change in Raman polarizability of the molecule upon adsorption on the particle[73].
Neglecting the effect of particle size on its dielectric constant a, the expression for g
(constant relating dielectric constants in equation (2.17)) suggests constant SERS
enhancement is independent of metal particle size. However, the dimension of the metal
particle cannot be larger than the wavelength of light, nor can it be smaller than the
average molecule size. The upper bound is due to the excitation of higher order
multipoles (these modes, unlike dipole, are non-radiative and hence inefﬁcient Raman
scatterers) when the excitation light wavelength approaches the metal particle size. The
lower bound arises from increased electron scattering (reduction in its conductivity) from
the particle surface since the particle dimension approaches the electron mean free
path[66]. At a given wavelength, the local ﬁeld intensity near the metal sphere surface is
related to the induced dipole moment p, which also determines the sphere absorption

(Jabs) and Rayleigh cross—sections (USCG). For a metal particle diameter less than the

wavelength of the incident light, the cross-sections are [39]

Win: ——l=-—”—

(2.19)

abs:

       

 

 

         

and

30

2
a _1287Z'5R6£3,|8—8m| __ 87r3 _p_
SC“ 314 |a+23m| 3,1433 E0

 

 

(2.20)

It is obvious from equations (2.19) and (2.20) that a plasmon creates the same

resonant peak (pole at a = —2£m) in both cross-sections, as well as in local ﬁeld intensity

given by equation (2.17). However, the two cross-sections scale differently with the

particle size and for small particles (Tabs at R3 dominates whereas for large particles

6

0m, oc R is dominant. This means that the nanostructures in the range 5-100 nm are

expected to be good SERS substrates with exact size dependent upon wavelength and the
metal used[73]. For example, the highest Raman enhancement for single silver spheres
was calculated to be for particles with 40-50 nm diameter at ~425 nm excitation
wavelength[74, 75]. Another important parameter affecting the quality of SERS is the
width of plasmon resonance, which depends upon the electronic scattering rate and the
metal interband transitions. For example, the wavelength dependent dielectric function of

a Drude metal is given by

2

0)
5(2) = gb(2)+1——2—!i— (2.21)
a) Hwy

where 8b is the contributions of interband transitions to the metal dielectric function, cop

is the metal plasmon resonance and y is the electronic scattering rate which is inversely

31

 

proportional to electron mean free path (or metal’s DC conductivity). The polarizability
of this sphere in vacuum is

a = 8 "1 R3 . (2.22)

8+2

 

Substituting equation (2.21) into (2.22) yields[73]

R3 6 022-02 +iwa
a- ( b P) b (2.23)

_ ((3,, + 3)w2 — (012,) + iwy(£b + 3)

 

a)
The pole of equation (2.23) is located at (UR = \[_p_§ and the width of this
8b+

resonance is given by 7(6‘b +3) [73]. Now it is easy to see why only noble metals give

good SERS enhancement. A larger y value due to poor conductivity of a given metal (for
example, transition metals) results in increased plasmon width and reduced SERS
enhancement. Similarly a large contribution to dielectric constant (8b) from interband
transitions also widens the resonance width and decreases Raman enhancement. This
precisely explains why SERS enhancement of silver exceeds gold, which in turn exceeds
that of copper. For example, the dielectric constants for thin ﬁlms of silver, gold, and
copper for 3.25 eV excitation light were found to

be —3.4720 + 10.1864, —1.6049 + 15.444, and —2.4131+i5.4397 respectively[76]. The

small imaginary part of dielectric constant for silver as compared to gold and copper is

32

e-r

55 ....

~.- --
-

.I...
nr‘
5‘1
..‘,
'u
l
'h
.
" .
I‘
‘- 11
33’ i-
I] . '
‘- H.
‘ ,
'14 t .
" ‘
1...-
'osA‘
' A
1
-5
Q
‘
~
‘
~
\
‘
~
‘
\
\
K
‘
\
\
\
\
\

the main reason for high SERS activity of silver. The dipolar plasmon resonance

frequency 0),; is also affected by the interband contributions to the dielectric constant for
metal. Although the plasmon resonance frequency for noble metals (mpz 9 eV) lies in
UV region of the spectrum, “’R is still located in visible or near visible region for Cu, Ag
and Au due to varying contributions from 8b. The plasmon resonance is pictorially

depicted in Figure 2-4. When the particle is irradiated by a plane wave with frequency
far from resonance, it does not interact signiﬁcantly with the radiation. At plasmon
resonance, the particle strongly interacts with incident light and the quality of resonance
is dependent upon particle size, material and incident wavelength through factors like a

(which includes 8}, ), 7, am etc.

Away from plasmon At plasmon
resonance resonance

 

 

 

 

 

v v v tr 1! v

 

 

l

 

 

 

 

 

 

" " 1i

 

 

Figure 2-4. Energy ﬂux (Poynting vector) around a metal nanoparticles
under plane wave excitation for two frequencies[77]- away from
resonance and at plasmon resonance.

33

13.1.1

 

*‘4:
‘-
I1. ‘
.; '
7 \

 

 

2.3.1.2 A spheroid body in dielectric medium

The isolated spherical particle model is useful to provide a basic understanding of
EM enhancement mechanism underlying SERS but is not sufﬁcient to accurately
describe real substrates. The real particles mostly consist of irregular shaped particles
arranged in different geometries. Many a times the individual particles can be well
approximated as ellipsoids and result in better analysis than using simple spherical
particle. For example a nanowire can be approximated as prolate spheroid with high
aspect ratio. The particle plasmon resonance wavelength is also dependent upon its shape
and it is useful to get a brief review of the theory behind SERS enhancement for such
shapes. The distribution of electric ﬁeld around a particle of arbitrary shape can be
calculated from Mie theory but the solution becomes quite involved even for simple
shapes. The problem can be simpliﬁed if particle dimensions are small compared to the
wavelength of laser and this assumption is made in following analysis.

Consider a spheroid with principal axes of length 21a , 21b , 21c having dielectric
function 62 and placed in medium of dielectric constant 81 . The spheroid is subject to an
external ﬁeld E0 of frequency (00 applied along one of the principal axes. The potential

inside the body by solving Laplace’s equation in ellipsoidal harmonics is given by[78]

x(E0)x _ Y(E0)y _ 2(E0)2

8 —8 8 -£ 8 -£
1+i——1Aa 1+—2——1A,, 1+ 2 '
81 51 51

 

¢2 = _ (2.24)

 

AC

where A ﬂ is the depolarization factor given by

34

_lalblc°° d3
‘ 1/2 ’ (,B=a,b,C)

A
13
0 (s + ,62){(s +102 )(s +le )(s + 13)}

The depolarization factors always satisfy ZAﬂ =1. For sphere, it is A0 = Ab = Ac = 1/ 3

For a prolate spheroid with a :b = 3 :1 , Aa = 0.1087; Ab = 0.4456.

The electric ﬁeld inside ellipsoid can be found be taking gradient Emsﬂ = —V¢2

and is given by

Emwmo) = {1 +[g(w0)—1]Aﬂ}“ Ewmo) (2.25)

The electric ﬁeld inside the particle is uniform and can be smaller or greater than
applied ﬁeld depending upon value dielectric constant. For dielectric such as quartz,
8(a)) is real and >1.0. This means that the ﬁeld inside a dielectric is in phase, parallel to,
and smaller in magnitude than applied ﬁeld. On the other hand, metals have large number
of free carriers that is associated with a complex dielectric function. Semiconductors can
also have extra carriers excited by light and, to some extent, behave as metals. The
complex dielectric constant of metals (and semiconductors) can have real part (81(00)
that is negative and less than unity. Hence, the ﬁeld inside can be larger and parallel or
anti-parallel to the applied ﬁeld. The possibility of supporting larger ﬁeld than the applied
ﬁeld is the main source behind electromagnetic enhancement from metals. This also
explains Why only a selected set of metals (silver, gold and copper) yield strong SERS

35

effect. It is easy to see from equation (2.25) that maximum of internal ﬁeld occurs when

the real part of denominator vanishes

1+(£1(a)sp)—1)Aﬂ =0

1 2.26
81(wsp) 34-5 ( )

The frequency ((1)5p ) is the surface plasmon frequency and corresponds to optical

excitation frequency that causes resonance or excitation of surface plasmons in particle.

This relationship also shows how the surface plasmon resonance frequency can be tuned

by changing geometry of the particle. For a spherical particle with Aa, =1 and we get

standard condition of 81(wsp)=—2. Although the dielectric constant of metals is a

function of particle size, but for larger particles bulk constants can be used for simplicity.
The laser wavelength used is also typically tied with the experimental set up. Then the
knowledge of optical constant of SERS metal at the laser wavelength can be used in
conjunction with equation (2.26) to determine a particle geometry that will yield
resonance for the given excitation. It is important to mention that the condition of
equation (2.26) correspond to the excitation of lowest order particle plasmon and is

represented by a dipolar distribution of surface charge oscillating at wsp. The ﬁeld just

outside the tip of the axis a of spheroid can be calculated by using continuity of normal

component of electric ﬂux (Dout,,B = Dms,,6 :> soutEowﬂ = ainSEl-nsﬂ) and is given by

36

 

8(a))
1+ (6(a)) — 1)Aﬂ

 

Eou,,ﬂ(a)) = E0,ﬂ(a)) (2.27)

The enhancement of electric ﬁeld experienced by a molecule adsorbed just on the

tip of the spheroid is given by

8(a))

(2.28)
1+ (8(a)) —1)Aﬂ

 

f(w)=

The molecule absorbed at the tip of the spheroid experiences an enhanced electric
ﬁeld compared to the incident ﬁeld and this ﬁeld results in polarization of the molecule.
The total ﬁeld experienced by the molecule will be a sum of incident ﬁeld and ﬁeld of the
plasmons generated on the surface of the metal particle. The ﬁeld at the site of the
molecule can be calculated by replacing metal particle with an equivalent particle dipole

Pe located at particle center. The dipole moment of spheroid is given by

Pen, ((00) = Vac, ﬂ(w0)E0,ﬂ(a20) (2.29)

Here V is particle volume and ae’ﬂﬂUOO) is diagonal element of particle

susceptibility tensor. This tensor element can be calculated by relating dipole moment

with the ﬁeld using relation

37

E+47r£=£E
V

V (2.30)
P = —— a —1 E
47! ( )
Equation (2.30) re-written for the case of polarized particle yields
V
Pe,p(€00) = $0100) ‘1)Ems,ﬁ(0)o)
(2.31)

__I_/_ 8(w0)—1
— 47r1+(8(a)0)-1)Aﬂ

 

50p (600)

From equations (2.29) and (2.31), we ﬁnd expression for susceptibility as

a 2i 8(00)-l
a” 47: 1+(8(w0)-1)Aﬂ

 

(2.32)

The molecule adsorbed at the tip of spheroid (at a distance [[3 from center) will

experience a ﬁeld due to an equivalent dipole at the center of the particle

(2.33)

21’ ((00)
Edip,ﬂ((00) = —e—’ﬂ-3—

The adsorbed molecule is polarized by this dipole ﬁeld and develops a dipole

mo . . . .
ment Am01=am01EdipnB(w0). The vrbratronal motion of the nucler modulates the

38

molecular polarizability am, and hence the dipole moment contains a component

oscillating at the Raman frequency (DR
Hawk) = aR,pﬁEdip,ﬂ(wo) (2-34)
Here the details of Raman scattering has been cast into phenomenological Raman
polarizability cm.
The metal particle is also subjected to Raman radiation, which polarizes it and this

particle acts as antenna to amplify Raman signal. Analogous to incident radiation, the

molecular ﬁeld also produces a ﬁeld at the center of metal particle given by

2y (a) )
Emol, 13001;) = —%—R— (2.35)
I3

This ﬁeld acts exactly like incident laser ﬁeld (except at frequency (OR) and

produces a dipole moment in metal particle
PM) = Vae,ﬂ(wR)Emoz,ﬂ(wR) (2.36)

Inserting expressions for 058.5(0)» and Emo,,ﬂ(a),,) into equation (2.36), one gets

39

 

 

2
V ] 8(600)-l £(wR)-1 aRﬂpWRWOﬁWO)

P :
ﬂ(wR) [272133 1+(g(a)0)-1)Aﬂ 1+(£(wR)—1)Aﬂ

(2.37)

The Raman dipole of the isolated molecule excited by incident ﬁeld E0 along

direction [3 is given by
,uﬂ(a)R) = aR,,BﬂEO,,B(wO) (2.38)

Comparing equation (2.37) and (2.38), we see that relative to Raman dipole of an

isolated molecule subject to same external ﬁeld, the particle dipole is enhanced by the

product f pd (coR ) f pd (0)0) of local ﬁeld enhancement factors at the Raman and incident

frequencies.

V 8(0) —1
27:1,,3 1+(e(w)-1)Ap

 

f Mao) = (2.39)

Here the subscript pd shows point dipole approximation.

2'3'1'3 Correction for radiation damping
The analysis of the interaction of metal particle with the laser presented thus far

has . .
been under electrostatrc approximatron; i.e., the ﬁeld was assumed to be constant

40

across the particle dimension d (kOd <<1, with k0 = woe ). This approach doesn’t take

into account radiative losses, which are proportional to the square of particle dipole and
hence of the volume. The dipole is expected to decrease in magnitude in reaction to its
own radiation. This can be modeled as ﬁrst order correction to the electrostatic
approximation. A radiation reaction ﬁeld E, is deﬁned such that the work per unit time
done by this ﬁeld on the particle dipole equals the power radiated by dipole[74]. In other
words, the radiation reaction ﬁeld tends to reduce the dipole in accordance with power

radiated. This leads to the deﬁnition
. l 3
E, =13-(2k )P (2.40)

This ration reaction ﬁeld is 90° out of phase with the dipole and the metal particle
experiences the external ﬁeld as well as its own radiation ﬁeld. The size of dipole is

determined by self consistently from

 

, 1
Pﬂmo) = Vae,ﬂﬂ(a)0)[E0’ﬂ(a)0) +1§(2k3)Pﬂ(a)0)] (2.41)
Rearranging
Va (0) )
P5000) = 1 8’3” O EO,,B(‘00)
1"" 3(21‘0 )Vae,p/3(wo) (2.42)
= Vaeﬁxﬂﬂ ((00 )EO,’B(0)0)
41

a (a) ) . . .
6’5!) 0 . The correction term in denominator of

where aeffng’g ((1)0) = 1 3
l-i-3- (2kg )Vae,ﬂﬂ(w0)

effective susceptibility expression dominates near particle resonance and limits the
effective polarizability Vaeff to the order of a cubic wavelength x103 . Now we can ﬁnd

ﬁeld enhancement just outside the spheroid using this effective susceptibility.

4—”——P/’(w°) (2.43)

Eou:,p(wo) = 8000) V €(wo)-1

Substituting in the value of P5 ((00) from equation (2.42) in equation (2.43), we

get

47! 8(600)

 

 

 

Eout, 5000) = 7.5—(35:1 Vaeff,ﬂﬂ(wO)E0,ﬂ(w0)
_ 4n 8((00) Vaeﬁﬂ ((00) E (a) ) (2.44)
- “— up 0
V “(UM—1 1-i%(2k03)Vae,ﬁp(wo)
Using deﬁnition of are, ﬂﬂ from equation (2.32) in (2.44), we get
E _ 49(600)
ouz,p(0)0) - Vk 3 Eo,p(€00)
1+[8(0’0)-]][Ap -i( 67‘: )] (2.45)

§ f (0)0)Eo,p((00)

42

 

.r‘:.

.I, an

(.._n

.h--

.r‘.
,.

‘5.

Comparing with (2.28), we see that the depolarization factor A5 has been

replaced by an effective Aeff,,8 that accounts for radiation damping. Radiation damping

is proportional to particle volume and inversely proportional to x13; i.e., radiation
damping is more sever for shorter wavelengths. For high eccentricity particle, the large
increase in enhancement is partly offset by the smaller fractional area near the tips. Thus
enhancement averaging over surface is required and EF can be corrected by multiplying
with a reduction factor (a that can be signiﬁcantly smaller than unity. The radiation
damping limits the maximum achievable SERS enhancement due to dipolar radiation
loss. The effect of radiation damping is more severe for large particles. Figure 2—5 shows
the effect of radiation damping on SERS enhancement for different aspect ratio of gold
nanorods modeled as prolate spheroids. The enhancement factors are calculated at the
resonant wavelengths corresponding to the aspect ratio. The enhancement is negligible
for simple sphere and increases as the aspect ratio of the particle increases due to
concentration of the electric ﬁeld at the tip. For a particular aspect ratio, the enhancement

Starts to decrease as the volume of the particle increases due to radiation damping. It can

be seen that the radiation damping begins to effect enhancement for £3- 210—4.
2

43

N.—

A

$9...

1 l I ‘

C~Vh_ tacos.- u-smvnp-AUHVP-Ewpsn- We

 

 

 

 

 

 

 

 

 

10E . ...... . .Mﬁ . new”?
i -—alb=1,)t=489 nm 3
106’ ..,....... alb=2, A=533 nm ‘
N: ------a/b=4, i=643 nm
(D .
89 ; ______ __ alb=6, i=787 nm ;
E 105:” W wif:i..'."..’:ﬂ'..'ff'; ----- -'---'---alb=7, x=843 nm .,
+0- .' ~~~~~ 1
8 4 ............. ‘
3:, 1o .. .................. . -
c h a... ~33,
OJ : ’ .."'a.a.l 1
as) 3 ’ "“«Z’m ‘
U 10 it. ‘9..".’:‘ ~ ".1
c E x}...
m k :9
.c: r
C 2 r .. ~.,
L” 10 g 2
1.
10 ‘ . - . . .
4 -3 -2
10 10 1O
V/)t3

Figure 2-5. Effect of radiation damping for different aspect ratio of gold
nanorods. Also shown is the resonant wavelength for each aspect ratio.

2.3.1.4 Interacting particles
Appreciable SERS enhancement from an isolated or single particle is, hardly
noticeable. The real substrate often consists of interacting colloidals or rough surfaces.
The Electromagnetic ﬁeld between two closely spaced particles is not a simple coherent
sum of the ﬁelds from individual particles. The particles exhibit capacitive coupling that
Causes signiﬁcant changes in the polarization of each particle. Heuristically, the dipole
near ﬁeld of one particle induces quadruple and higher order moments in its neighbors.
As a result of these mutual inductions, the near ﬁelds from all these moments coherently

ad . . . .
d to produce maxrmum external ﬁeld in the junctron and the Raman enhancement

44

factor can approach 10l0 to 10'2 for separation of l nm[39]. Figure 2-6 shows two
closely spaced metal nanospheres illuminated by plane polarized light and the resulting
induced charges on each sphere. The conjugate nature of charges on each sphere gives
rise to enhanced electric ﬁeld and correspondingly enhanced Rarnan scattering for a
molecule localized in the junction between two particles. It is worth reiterating that the
dipole induced in each particle arises not only from the incident ﬁeld but also for the
intense ﬁeld of its partner. The effect of light polarization relative to the axis of the
dimmer is also illustrated. This simple example can be extended to large aggregates of
particles where each particle will have its own characteristic ﬁeld strength and the

accumulative effect could yield giant enhancement making the cluster a “hot” site.

+++++

+++
+ +*

        

 

Figure 2-6. Light interaction with metal nanospheres dimmers[73].
Analyte molecule is shown by small green ﬁlled circle in between two
Particles.

45

Another important feature of aggregates of particles or other interacting
nanostructures is the short range of the enhanced ﬁeld. This not only requires the metallic
particles to be closely located but the analyte molecule must also be present within the
enhanced ﬁeld range to get good SERS enhancement. Figure 2-7 shows the calculated
Raman enhancement factors at the midpoint of 60 nm diameter dimmer (material: Ag, Al,
Si, SiOz) for 497 nm excitation[39]. The Raman enhancement decreases rapidly with

increasing inter-particle distance and ideal spacing is of the order of few nanometers.

This is due to the fact that the ﬁeld of the dipole scales (DC—1? and the fourth power
r

dependence of enhancement in equation (2.18) results in very sharp gain decay (0c —}—2 .
r

 

1010
0 Ag
0 Al
8 _ 6 Si
10 a SiO2
“S
E
8106 ~ .
E
8
(13104 ~ ~
0:
Lu
(n

_l
O
N

 

 

J l l

0 2 4 6 8 1 0
Separation between particles (nm)

 

10

Figure 2-7. Calculated Raman enhancement factors at the junction of 60
11m diameter dimmers for 497 nm excitation[39].

46

The electromagnetic model provides for most of the SERS observations including
need for nanostructured material for SERS activity, better SERS response for noble
metals, greater enhancement for interacting nanoparticles compared to isolated particle,
and the few hot sites (appropriately structured clusters) among an ensemble of
particles[79]. However, the EM model does not account for all observations, such as

molecular resonance, charge-transfer transitions etc.[39, 80]. For example, EM
enhancement is expected to be an isotropic ampliﬁer of Raman scattering of a molecule
adsorbed on a rough surface but the SERS intensities for CO and N2 under similar
conditions differs by more than a factor of 200[51]. This observation could not be

explained by the EM model alone, and introduces a chemical enhancement mechanism in

the SERS picture.

2.3.2 The chemical enhancement

Many studies support a second enhancement mechanism known as chemical
enhancement or charge transfer[8, 52, 53]. This mechanism is less well understood due to
the problem of experimentally decoupling it from the electromagnetic mechanism as both
operate simultaneously for most SERS experiments involving rough surfaces or colloids.
In this mechanism, the analyte is assumed to form a bond with the metal that facilitates
transfer of electrons between metal and analyte[3]. The chemical enhancement can be
explained by a resonance Raman effect in which either molecular energy levels shift and
broaden, or a new resonant intermediate energy level involving the metal surface is

created; In the resonant intermediate state formation hypothesis, the metal absorbs the

11
ght and charge is transferred into the analyte-metal complex. Then the Raman process

47

occurs, excitation is transferred back into the metal and re-radiation of Raman scattering
takes place from the metal surface[3]. The chemical enhancement predicts larger
enhancement factor of the ﬁrst adsorbate layer compared to the subsequent layers.
However, it is generally agreed that the electromagnetic enhancement mechanism (of the
order 104 to 106) plays a much greater role in overall SERS enhancement than the
chemical enhancement (of the order 102)[52]. This difference in the scales of the two
enhancements makes the EM mechanism a dominant factor and most research has

focused on enhancing the SERS effect through the use of interacting nanostructures.

2,4 Ordinary Raman spectra and SERS spectra

Since we will be dealing with both ordinary as well as SERS Raman spectra of
different species in the later part of this dissertation, it is worthwhile to mention a few

important differences between the two. The ordinary Raman spectra SERS spectra

usually differs from normal Raman spectra in that

a. New peaks may appear.

b. Some strong normal mode peaks may become weak or disappear.

Raman intensity versus analyte concentration relationship may become

non-linear in SERS

(1. High energy bands intensity falls more rapidly in SERS due to different

gain for scattered radiation.
6. Spectra tends to be depolarized.

f- Departure from or“ dependence.

48

g. Broad background may appear[51].

For example, for pyridine the SERS spectra is very weak below monolayer
coverage (because the carbon ring is parallel to surface). At monolayer coverage, a rapid
rise in SERS intensity results as the carbon ring is forced normal to surface to allow more
molecules. The light induced dipole in the metal can be divided into two components
(parallel and normal). It is the molecular polarizability caused by perpendicular
component which leads to scattering from a rough surface[3]. The appearance of new
bands in SERS is due to adsorption of molecule on metal that breaks the molecule’s
center of symmetry. The mutual exclusion rule (IR active, Raman inactive and vice
versa) is no longer valid and allows IR bands to appear in the Raman spectra. Some types
of bands are more intense in SERS than normal. The problem with chemical
enhancement is that the nature of the species formed between the analyte and metal is not
Well deﬁned. The selection rules should refer to the surface species instead of isolated
molecules. However, treating the molecules as separate entities simpliﬁes the rules and
gives reasonable results. Also in SERS the strong intensity means contaminants in trace
amOllnt have a greater impact and may alter the spectra. All these factors need be taken
into consideration while attempting to compare the SERS spectra of any species against

its ordinary Raman spectra.

49

Chapter 3: Experimental Set up

This research effort can be broadly categorized into two steps. First step involves
the fabrication and characterization of SERS substrate while second part involves the
evaluation of substrates using the Raman system. The substrates were prepared by either
growing oxide nanowires in tube furnace system or growing metallic nanorods using an
e-bearn evaporation set up. The evaluation of the SERS substrates was done utilizing a
near IR Raman system. Accordingly the experimental systems used in this study can be
classiﬁed into three categories - a nanowires synthesis set up, nanorods synthesis set up

and Raman Spectroscopy set up.

3.1 Oxide nanowire synthesis set up

The nanowires employed in this study are germanium dioxide (GeOz) and zinc

oxide (ZnO). The GeOz nanowires are synthesized utilizing vapor-liquid-solid (VLS)

8TOWth mechanism ﬁrst proposed by Wagner and Ellis in l964[81]. The ZnO nanowires
Synthesis followed a self-catalytic grth mechanism. The synthesis of nanowires is
carried out using a horizontal tube furnace as shown in Figure 3-1. The furnace consists
0f coiled resistive heating element encapsulated by ceramic thermal insulation in a
hollow cylindrical shape. The furnace is 12 in. long and has an outside diameter of 8 in.
A 3.35 in. diameter quartz tube is inserted through central 4 in. diameter bore of the
fumace to facilitate control of the temperature and environment for the growth. The
quartz tube is sealed on both ends using stainless steel end caps with double ‘0’ ring

Seals, . . .
The upstream end of tube 15 connected to two-gas lines that can Independently

50

 

supply a measured amount of gas ﬂow through respective Mass Flow Controllers
(MFC’S). The downstream end is connected to a mechanical rotary pump that can achieve
a vacuum of less than 100 mTorr inside the quartz tube. The pressure is measured by a
convector type vacuum gauge located at the downstream end of the fumace in the
vacuum line. The temperature of the fumace is monitored by a type ‘K’ thermocouple
(not shown) placed between the quartz tube and the inside wall of fumace. The
maximum temperature that can be achieved at the location of the thermocouple is
~1000 °C. The sample and the source material are placed on a ﬂat quartz plate (2.5 in. X 6
in.) at the desired positions. Figure 3-2 shows a typical layout for the germanium source
material and silicon substrate on the quartz support plate. For ZnO nanowires, the
substrates are placed at the top of boat containing zinc powder and the boat is then loaded
on the support plate. After positioning source material and substrates, the support plate is
inserted into the quartz tube and localized within the heating zone of the furnace. The
desired vacuum level, temperature range and gas ﬂow rate for the experiment can be
controlled using fumace controller, MF C’s and the isolation valve in the vacuum pump
line. In a typical synthesis experiment, the quartz tube is evacuated to ~200 mTorr,
backﬁlled with Ar to ~10 Torr and then evacuated again to improve vacuum. This is
followed by a quick leak test of the system by closing the vacuum isolation valve and
Observing the rise in pressure for ﬁxed period of time. After satisfactory set up of vacuum
inside the quartz tube, the gas ﬂow and pressure are established to desired value utilizing
MFG and Vacuum valve. Next temperature is ramped to the target value using
temp eratur e Controller. At this stage, additional gas/gases can be introduced and/or ﬂow

rate
S can be modiﬁed using two MFCs. The growth of the nanowires is canied out for the

51

 

set amount of time and then furnace is shut down. Argon is usually allowed to ﬂow into
the tube during cooling of furnace to room temperature. The quartz tube can be opened
aﬁer the pressure inside reaches atmospheric pressure. This may require bleeding

additional Ar gas or room air into the tube.

 

Fig‘u‘e 3-2. Quartz support plate with typical sample layout (A top view).

52

3,2 Silver nanorods synthesis set up
Silver nanorods are synthesized using Oblique Angle Deposition (OAD) of the
metal onto a glass substrate[36]. The synthesis of nanorods was carried out in an AXXIS
physical vapor deposition system (PVD) from K. J. Lesker as shown in Figure 3-3. This
deposition system is equipped with one 4-pocket e-beam evaporation gun and two
sputtering guns. Only the e-beam evaporation was utilized for the growth of these
metallic nanorods since it allows precise control over the deposition rate. The e-beam can
be used to evaporate up to four different sources in turn without breaking the vacuum.
This is achieved by rotating the hearth of e-beam so that the crucible with the desired
material is illuminated by the e-beam. Glass slides (l in. X 3 in., VWAR Inc.) are used as
typical substrates for the metallic nanorods. These slides are cleaned and loaded onto the
substrate stage of the PVD system. The deposition chamber is evacuated to ~3><106 Torr
and heated radiatively by a pair of quartz lamps located under the substrate platen. The
OAD synthesis involves ﬁrst depositing a thin layer (typically ~50 nm) of metal on the
glass slide using e-beam evaporation. During this deposition, the substrate faces the
e-beam evaporation source and is continuously rotated as schematically illustrated in

Figure 3-4(a). Then the substrate rotation is stopped and the substrate stage is tilted such

that the incident ﬂux anives at almost grazing angle (515°) to the surface of the
Stationary substrate as shown in Figure 3-4(b). The oblique angle deposition of metal is
done to the desired thickness at a rate of ~0.4 A/s as measured by a Quartz Crystal
Monitor (QCM) operating at 6 MHz. The temperature of the substrate is maintained at
8513 °C throughout the deposition. The nanorods start to grown during oblique

d6 03' '
p ltron due to shadowing effects of nanoscale islands. The substrate can be inspected

53

 

visually from the 6 in. glass view port of the chamber during growth. Typically substrate
surfaces have a mirror ﬁnish alter normal deposition that changes to more dull
appearance after deposition of ~200 nm metal at oblique angle. This is an indirect

evidence of the start of formation of nanorods on the substrate.

 

Fig“ l‘e 3-3. Physical vapor deposition systems for metallic nanorods
synthesis.

54

 

. (no rotation)
substrate

% Evaporation

source

(a) 0?)

Continuous
H rotation Substrate
positioned for
| I . OAD deposition
_ Glass

Figure 3-4. Schematic of oblique angle deposition of silver nanorods
using e-beam evaporation. (a) Normal angle deposition. (b) Oblique angle
deposition.

3.3 Raman spectroscopy set up

The Raman system utilized for this research is a portable system (EzRaman-L)
from Enwave Optronics Inc. as shown in Figure 3-5. It employs a Near IR frequency
stabilized diode laser operating at 785 nm with maximum power output of ~400 mW.
The MR excitation frequency was chosen to minimize ﬂuorescence from the samples.
The laser power is delivered to the sample through 100 um excitation ﬁber and scattered
light is collected by 200 um collection ﬁber. The ﬁber optic probe head contains a
r6placeable lens tube, which focuses the laser light onto the sample and also collects
scattered light in backscattering geometry. In addition, a Rayleigh rejection ﬁlter of
Optical density (OD) > 8 in the collection path provides rejection of the laser line from the
Spectra. The high 0D is a unitless measure of the transmittance of an optical element at a

iv
g en WaVelength and is deﬁned as

55

CD = —log(i) (3.1)
10

where I and 10 are the intensity of the transmitted and incident laser respectively.

  
  
 

 

Probe head. 1

a“: ..

Figure 3-5. Raman system set up.

The spectrograph has an f/4 aperture, symmetrical crossed Czemy-turner with a
resolution of better than 6 cm". The spectral range covered is from 200-2400 cm'l. The
system is equipped with a charge-coupled device (CCD) array detector having a pixel
size of 1 4 pm x 200 um and 16-bit digitization. Computer interface with software for

Collecti on and analysis is provided through a USB cable. The system is supplied with an
oPtical fiber probe and a vial holder to analyze liquid samples. The ﬁber head has a
rePlaceable lens tube and that can carry different numerical aperture (NA) lenses. The as
supplied System has been optimized for the study of liquid samples with laser always
focuSed inside the sample vial when using sample holder. The analysis of samples using
SERS mOStly involves solid substrates and this warrants precise focusing of laser on the
substrate Sui-face. Moreover, the ability to scan the substrate surface for homogeneity is

56

 

also needed for uniformity studies. Due to these considerations, a three axes linear
sample stage was installed to allow precise motion of substrate as shown in Figure 3-6.
The laser head of the optical ﬁber is mounted on a separate linear z-stage that allows
quick and coarse focus of laser beam on substrate. The precise focusing of laser on the
sample is accomplished by adjusting the substrate z-stage. The laser light at 785 nm is
delivered to the sample via an excitation ﬁber. A lens at the end of the probe head focuses
the laser beam to a spot and the sample can be brought into sharp focus by adjusting 2—
height of the probe head. In a typical focusing exercise, the sample is placed at a course
focus depending upon the focal length of the lens and then ﬁne focus is achieved by

adjusting the height of the substrate while monitoring the strength of the Raman signal.

1 Orrclﬁber ‘

Cardboard '
enclosure
1 9 head

 

Figure 3-6. Rarnarr system optical ﬁber head and sample stage.

57

The scattered light is collected in the backscattering geometry (Raman scattering

is isotropic and incoherent) by the same lens and carried to the spectrograph via the

collection ﬁber. The ﬁber head also incorporates a Raleigh rejection ﬁlter to remove the
elastically scattered signal from the sample signal. The spectrograph separates different
wavelengths from the Raman scattered light which is then focused onto a linear CCD
array detector. The output of the CCD is the Raman spectrum of the sample. The Raman
system can be interfaced to a PC or laptop with a USB cable and the collected spectra is
displayed on the computer running proprietary software. The Raman scan always starts
with collection of the CCD dark current with the laser off to account for CCD noise.
Then the laser illuminates the sample and Raman scattered signal is integrated by the
CCD for the duration of excitation. The Rarnan spectrum (only Stokes scattering) is then

presented on the computer monitor as an energy shift from the energy of the laser beam.

. . 1 . . .
The x—ax1s of Raman spectra IS the wavenumber (w = I) of Raman scattered rad1atron 1n

' units[3]. The y-axis shows the intensity of the scattered radiation detected by the

cm'
CCD at a particular wavevector. Often a smoothing algorithm (like boxcar) is applied to
remove sharp discontinuities in the spectra. The near IR laser used in the Raman system

is invisible to the eye and caution has to be exercised in operation of system. The whole
laser head and sample stage is enclosed in a box to contain the laser radiation and to
avoid false detection due to interference from ambient light. The system is also supplied
With protective eyewear to minimize risk of eye injury in case of accidental exposure to

the laser beam. The coarse position of the laser beam on the sample can be monitored

u ' - .
8mg a Corninon CCD camera 11ke a USB computer webcam.

58

3.3.1 Laser spot size determination

The laser is focused to a ﬁne spot on the substrate by focusing lens at the end of
optical ﬁber. The size of the laser spot on the substrate is needed to determine the amount
of analyte excited by the laser for Raman enhancement factor calculations. There are
many ways to measure the diameter of laser beam. These include ray tracing,
ﬂuorescence correlation spectroscopy, and multiphoton ionization yield. Though these
methods provide accurate results, they also require sophisticated experimental set up and
analysis. The knife-edge method is a simple and practical method that can be easily
implemented in most situations. In this method, a knife-edge mounted on a translation
stage is used to eclipse the laser beam more and more by slowly translating it across the
focal plane of the laser. The power of the laser beam reaching a detector is recorded as a
function of translated positions of knife-edge. The diameter of the beam can be deﬁned
in terms of a threshold power that is contained in that diameter. A commonly employed
threshold of D36 was used. The knife-edge measurement was realized using a razor blade
attached to a micrometer stage as depicted in Figure 3-7. The measured diameter of the
laser beam using knife-edge method was found to be ~94 pm. This value agrees well with

the manufacturer speciﬁed diameter of less ~ 100 um.

59

Laser excitation

Focusing lens

 

 

. Razor blade
Laser focal pornt
a... ‘ ‘. s
\~ ~\’. g. ‘ 1.5—7
H‘Ai: " f“; 'r‘ecooooo

O: L.“ 1"
’ Edge translation

Power meter -->

Figure 3-7. Knife edge method for determining laser beam diameter.

3.3.2 Raman system calibration

An important factor in getting accurate spectra is the x-axis calibration of the

spectrum to ensure correct assignment of wavenumbers. This is done calibrating the CCD

pixel positions against the Raman peaks of a reference sample. Isopropanol (also known

as i sopropyl alcohol) was used as reference sample for calibration and its Raman spectra

is shown in Figure 3-8. For calibration, the instrument is set to pixel mode and the pixel

numbers corresponding to the eight Raman peaks of isopropanol spectra are noted.

Finally these pixels are entered into the x-axis calibration dialog box of the Raman

s . . . . .
0th” are Which generates the calibratron coefﬁcrents for accurate pixel to wavenumber

mapping-

60

15000r

I,

1 0000

5000 -

 

Raman lntensity/Arbitr. Units

 

1.. i

500 1000 1500
Wavenumber/cm‘1

; o -

Figure 3-8. Reference Raman spectra of Isopropanol, tacc= 20 3.

Table 3-1 shows the pixel and the wavenumbers corresponding to different

Raman peaks of isopropanol that are used to generate x-axis calibrations data.

Table 3-1. Isopropyl alcohol Raman peaks and wavenumbers.

 

 

 

 

 

 

 

 

 

 

Peaks Pixels Wavenumber (cm'])
1 139 374.2
2 179 433.4
3 222 491.2
T 4 460 817.6
5 564 945.6
6 708 1 129
7 892 1339.8
8 988 1453

 

 

 

 

61

Although y-axis calibration can be done, it is much more complicated due to non-
availability of a suitable standard. Other factors such as laser power, frequency, and
environment also affects the scattering intensity and makes it more difﬁcult to achieve
y—calibration. Fortunately for the qualitative work like detection of material, y—calibration

is not essential

62

Chapter 4: Germanium Oxide Nanowires based SERS Substrates

This chapter presents details of synthesis, growth kinetics and SERS results for
germanium dioxide (GeOz) nanowires (NW 3). Germanium dioxide nanomaterials exhibit

blue photoluminescence (PL) material with peak energies around 3.1 eV and 2.2 eV[82].
Germanium oxide nanowires can emit stable and high brightness blue light at 485 nm
(2.56 eV) under excitation at 221 nm (5.61 eV)[83]. Germanium oxide-based glass is

thought to be more refractive than the corresponding silicate glass so that the Geo; NWs

may be used for nano-connections in future opto-electronic communications. It has also

attracted considerable attention as a potential material for optical ﬁbers with less
transmission loss than SiOz[84]. The NWs were synthesized on silicon (Si) substrates

using the Vapor Liquid Solid (VLS) growth mechanism and were investigated for their
potential use as SERS substrates. Characterization of the NWs was done using a
Scanning Electron Microscope (JEOL 6400V), a Field Emission Scanning Electron
Microscope (JEOL 6300F), and a Transmission Electron Microscope (JEOL 2200FS).
The as synthesized substrates were coated with gold/silver to provide a plasmon active
surface. The SERS performance of these substrates was evaluated using commonly
emPIOyed probe molecule analytes including Rhodamine 6G, Nile Blue, 4-

methJ’lbtenzenthiol, 1,2-benzenedithiol and trans-1,2—bis(4-pyridyl)ethylene.

63

ll

4.1 Nanowire synthesis
There have been a number of reports of Ge02 NWs synthesis in past decade using

a variety of techniques. Some of the past efforts and present synthesis technique are

discussed in this section. A grth model to explain the synthesis is also presented.

4. l .1 Reviews of past efforts

Gennanium dioxide NWs were ﬁrst synthesized by Bai et al.[85] using physical
evaporation of a mixture of Ge and Fe (8%) powder at 820 0C under continuous ﬂow of
1 3O sccm of Ar at 200 Torr. The product was collected on inside walls of the quartz tube
in front of a copper cold ﬁnger. Since then there have been a number of reports of GeOz
NW synthesis using a variety of techniques. Zhang et al.[86] used carbon nanotube-

confined reaction in which pure Ge powder, carbon nanotubes and mixture of Si/SiOz

powders were kept at 850 °C under ﬂowing Ar (150-200 sccm) for 3-4 hrs. Germanium
dioxide nanorods of 50-200 mm diameter and several um lengths were formed at location
of carbon nanotubes. Wu et al.[83] reported carbothennal reduction based synthesis of
NW 5 by mixing equal amounts of Ge and carbon powder and heating to 840 °C under 20

sccm ﬂow of N2 for 3.5 hr. The GeOz NWs having 50-120 nm diameter and hundreds of

11m length were formed on the inside walls of alumina boat containing source powder.
Dang 6t al. [87] used thermal oxidation by heating Ge powder to 1000 °C under 120 sccm
gas ﬂow (80% Ar, 20% 02) for 30 min to get NWs of 80 nm diameter and several tens of
microns long at the location of source material. Hu et al.[88] heated a mixture of 2 g Ge
POWder With 4 g Fe(NO3)3 on an alumina wafer to 800 °C for 30 min. Then the

te
mp er ature was raised to 1300 °C and synthesis was carried out for 6 hr under Ar ﬂow

64

 

 

 

(containing 5% H2) of 50 sccm at a pressure of 400 Torr. Nanowires having diameter of
150 nm with lengths of several tens of microns were collected on inside walls of alumina
tube in the 700 °C temperature region. Su et al.[89] thermally evaporated Ge powder at
1050 °C in an oxidizing environment (90% Ar, 10% 02) for 40 min at 300 Torr. The

self catalytic synthesized nanowires (100 nm diameter and 800 nm length) were collected

on Si substrate kept at 950 °C. The tips of NWs were found to have Ge particles of 500
nm diameter (tip showed 3% Oz in addition to Ge) and NW PL spectra showed two blue
emission peaks at 448 nm and 471 nm and a violet emission peak at 411 nm. Higalgo et
al. [90] annealed compacted GeOz powder discs under ﬂowing Ar for 24-48 hrs and got

NWs and needles having diameters of 100-500 nm. Tang et al.[91] utilized laser ablation
of Ge powder compacted as a disc at 820 °C and 700 Torr under Ar ﬂow of 50 sccm.
Nano whiskers having diameter of ~ 2pm and lengths of up to 2 mm were collected on
quartz substrate located near target. Wu et al.[92] employed thermal annealing of Ge

under oxidizing environment (room air at 200 sccm and 350 Torr) at 850 °C to synthesize

NWs with diameters of 40-500 nm.

4.1.2 Our synthesis technique
The GeOz nanowires used in this work were synthesized utilizing the VLS growth

mechanism[81] in a simple horizontal tube furnace set up as shown in Figure 4-1. In the
VLS mechanism, the catalyst forms a eutectic with the nanowire material and acts as a
Preferential growth site for the nanowires. The location and diameter of the nanowire is
mainly controlled by the position and size of the catalyst particle[93]. The synthesis
teel’mique is similar to thermal annealing growth technique[92] previously reported but

65

employs stoichiometric ﬂow of dry gases instead of room air to allow greater control over
reaction variables. Silicon substrates were cleaned by sonication for 15 min each in
acetone, methanol, and DI water and ﬁnally dried by blowing compressed nitrogen. The
substrates were then coated with 3 nm to 28 nm thick gold ﬁlm using an e-beam
evaporator or by sputtering. In the case of e-beam evaporation, the ﬁlm thickness was
monitored using Inﬁcon Quartz Crystal Monitor. The Au coated silicon substrate was
placed on a quartz support plate, several centimeters downstream of the solid germanium
source material. The support plate was then positioned in the center of a quartz tube
placed in a cylindrical tube furnace. The tube was evacuated to less than 200 mTorr by a
mechanical vacuum pump and subsequently backﬁlled with argon at a ﬂow rate of 50-80
sccm. Temperature of the furnace was raised to 855 °C over 30 min and allowed to
stabilize for 10 min. The temperature was measured using ‘K’ type thermocouple
sandwiched between quartz tube and tube ﬁrmace as shown. After reaching growth
temperature, the downstream isolation valve was closed and pressure was allowed to
increase to about 5-10 Torr. The increased background pressure allows conﬁning the
source vapors over the substrates. Then oxygen mixed with Ar (40% 02 and balance Ar)
Was introduced into the reaction chamber at a ﬂow rate of ~30-50 sccm. The nanowire
STOWth was carried out for nearly 5-30 min and then the gas ﬂow was switched from

Oxygen/argon (40/60) to pure argon. The furnace was allowed to cool to room

temperature before opening the tube to atmosphere.

66

Ge source Quartz support plate Au coated substrate

 

 

 

 

 

 

 

 

 

 

Ar —’ :3 \‘I v x ’1 6
i a i “:3 ——s
Ar + 02 —' :1 3 3 to vacuum
9 3 . . pump
Quartz tube ‘
thermocouple 5

Hollow cylindrical furnace
Figure 4-1. Schematic of germanium dioxide nanowire synthesis.

4.1.3 Characterization of nanowires

The substrate surface was covered with white powder-like material after synthesis

that could be seen with the naked eye. The product was characterized using JEOL 6400

scanning electron microscope (SEM). Figure 4-2 shows the SEM image (15 keV with

~15 mm working distance) of the substrate fully covered with dense nanowires. The x-

 

 

 

ray Energy Dispersive Spectroscopy (EDS) analysis shown in Figure 4-3 conﬁrmed the
NWs were composed of germanium and oxygen. The atomic ratio of germanium to

Oxygen was found to be 25:54 which is quite close to the actual ratio of 1:2 suggesting
the product to be GeOz. The nanowires were also characterized by JEOL 2200FS 200 kV

ﬁeld emission transmission electron microscope (TEM). The TEM samples were
prepared by ultrasonicating the nanowire substrate in ethanol and then dispersing a drop

of solution on a holey carbon-coated TEM grid. Figure 4-4 shows the TEM image of a
Single GeOz nanowire with a diameter of ~31 nm. The TEM image shows a dark

Spherical tip at the end of the nanowire. The EDS of the tip shows some gold while that

fr Om the stem of the nanowire does not show gold in the EDS spectrum. This conﬁrms

67

 

 

 

that growth is taking place by the VLS mechanism. The selected area electron diffraction
(SAED) pattern from the nanowire is also shown in Figure 4-4 and conﬁrms that the

nanowires are single crystalline.

 

 

Figure 4-2. SEM image of Geo; nanowires on different samples of Si.

68

 

 

 

 

Si
Au
0 Au
Ge
Au
.1 Au ﬁéu Ge Au
i 4 6 8 10 12 14 keV
Element App Intensity Weight% Weight% Atomic%
Conc. Corrn. Sigma
0 K 0.55 0.9780 21.61 0.87 53.97
Si K 0.25 0.7934 11.97 0.46 17.03
Ge L 0.88 0.7524 44.66 0.94 24.59
Au M 0.37 0.6500 21.76 1.07 4.41
Totals 100.00

 

 

 

Figure 4-3. EDS spectrum and quantiﬁcation results for GeOz nanowires.

69

 

Ge
Ge

Ge

 

 

 

 

 

 

Ge
' ' . 11% [I'll FJthl-ir
0.5 1.5 keV
[TEM] V i 200 Onm
J EM-22OOFS 200kV x80k 100% 4/25/2007 1058 32 AM

Figure 4-4. TEM image showing gold at tip and EDS spectra for two
regions of nanowire and the diffraction pattern.

70

4.2 N anowire growth kinetics

The synthesis of nanowires involves both oxidation as well as annealing, so
before embarking on explaining oxide nanowire growth kinetic, it is useful to take a

cursory look at the oxidization process of metallic germanium and the behavior of its

oxides under high temperature annealing.

4.2.1 Oxidation of germanium

Unlike silicon the formation of germanium oxide is complicated by the interim
reaction product of a monoxide that is thermodynamically unstable as a solid and
sublimes at high temperature[94]. Germanium is stable to atmospheric oxygen at room
temp since a protective thin oxide ﬁlm covers its surface. Adsorption of water vapors on

the surface causes destruction of protective layer and results in the formation of a thick
layer of oxide[95]. Oh et al. have shown that the native oxide on Ge is mostly Ge02 with
small amounts of Ger (1 < x < 2) [96]. Prabhakaran et al.[97] reported formation of a
mixture of GeO and GeOz upon exposure of Ge to air and the amount of GeOz increases
with increasing exposure time.

Germanium is readily oxidized when heated in air or in a stream of oxygen. A
ﬁlm of (360 always forms at the start of oxidation that transforms into GeOz as the
reaction proceeds[98]. Many studies conﬁrmed formation of Ger (1 <x <2) during

initial stages of thermal oxidation of Ge[99-101]. Prabhakaran et al.[97] found that in situ

oxidation of germanium by leaking oxygen into UHV chamber only produces GeO. In

“1611' Study, GeOz was obtained only by exposing Ge to air and they attributed formation

of G302 to moisture effects. Molle et al.[101] reported formation of signiﬁcant amount

71

of sub-oxides at room temperature during germanium oxidation using atomic oxygen.
Increasing temperature to 300 °C, strongly enhanced formation of GeOz. However, GeOz
started transforming to GeO at 400 °C and completely desorption of oxide occurred at

450 —475 °C.

The main reactions during thermal oxidation of germanium in an oxidizing

environment are[98]

Ge(s)+ 1/202 (g) —) GeO(s) (4.1)
GeO(s) -—) GeO(g) (4.2)
Ge0(s)+ 1/202 (g) —-) GeO 2(3) (4.3)

A study of oxidation rate of Ge at 450-700 °C showed that above 550 °C, the
oxidation rate of Ge increases with decreasing oxygen pressure[98]. The relative
contribution of these reactions is a strong function of oxygen partial pressure. For
example, the rate of Ge vaporization (i.e., loss due to oxidation) is ~5.4 mg/min in stream
of air at 4 mm Hg. The vaporization of Ge, however is nearly zero if stream is at a
pressure of 34 mm Hg[98]. Indeed as the oxygen pressure decreases, all of the GeO
formed on the surface manages to sublimate i.e., reactions (4.1) and (4.2) take place, the
rate of reaction (4.3) being very slow. When oxygen partial pressure increases, the

Oxidation proceeds according to reactions (4.1) and (4.3), the reaction (4.3) begins to

Prevail over reaction (4.2). After a continuous ﬁlm of GeOz has formed on the surface of

the Sample, the reaction is determined by the rate of diffusion of oxygen and GeO

through this ﬁlm, and the reaction slows down considerably. Similarly if Ge is heated

72

under conditions where the amount of oxygen is insufﬁcient to convert it to GeOz (for
example in stream of N; or Ar containing oxygen admixture), formation of volatile GeO

results[95].

4.2.2 Stability of germanium oxides

The oxidation/annealing step at high temperature has signiﬁcant impact upon
oxidation process due to formation of volatile GeO that can diffuse away or decompose
during oxidation/annealing process. Annealing of germanium sample having a thin oxide
film at 500 °C under N2 environment conﬁrmed thermal decomposition of GeOz into
GeO which desorbs from the surface[98]. A similar observation was reported by
Prabhakaran et al. during their study of thermal decomposition of oxides for Si and
Ge[102]. They found that on annealing, both the oxides (GeOz and SiOz) undergo
thermal decomposition and the desorbing species is the corresponding monoxide.
However, their decomposition pathways were entirely different. On annealing the Ge

oxide layer, the GeOz species transforms to GeO on the surface and ﬁnally desorbs from
the surface at ~425°C. In contrast, annealing results in the transformation of SiO to SiOz

Up to a temperature, lower than the desorption temperature (~760 °C). Hansen et al. only
detected GeO(g) as desorption product during exposure of Ge (100) to molecular
Oxygen[103], It has been reported that the GeO desorbs from germanium surface on

annealing at 400 °C in vacuum without undergoing any further transformation[97]. In

Contrast to GeO, the GeOz is very stable and does not melt unless temperatures greater

than 1 1 1 6 i 4 °C are reached[104]. On heating bulk GeOz to 1000-1100 °C, vaporization

73

is hardly noticeable and there is no dissociation to Ge and 02 or GeO and 1/202[95]. The

situation changes when GeOz is in contact with Ge (e.g., thin Ge02 ﬁlm on Ge) and a

volatile GeO may be formed at the interface. In summary, it is widely agreed that GeO is
the only desorption product of oxidation of Ge and the GeO desorption occurs in the

temperature range 400 — 500 °C.

4.2.3 Growth model

The main ingredients in the synthesis of GeOz nanowires are the solid phase

reactant of Ge metal and gas phase species of Ar, H2 and oxygen. Obviously the most
important of these are Ge and oxygen as these are to form the ﬁnal product. The Ar is
used mainly as the carrier gas to transport the source material to the substrate(s). The
beginning source material of Ge is expected to have thin layer of native GeOz since it has
been exposed to atmosphere during storage. As the source temperature is ramped under
low vacuum (300-400 mTorr) with protective ﬂow of Ar and Hz, the native oxide is

expected to be reduced to Ge or will escape by transforming into GeO above 400 °C.
During ramping and even at the reaction temperature, the source evaporation is negligible
since Ge has a negligible vapor pressure at the grth temperature (the melting point of
Ge is 937.2i0.5 °C[105, 106] whereas the actual temperature at Ge source inside the
furnace is ~ 800 °C). Hence, any direct evaporation of Ge is not a signiﬁcant contributor
to nanowire growth. This was further conﬁrmed from a control experiment in which
neither any discemable loss of Ge source nor NWs grth was observed when an

eXperi ment was carried out without ﬂowing oxygen into the chamber.

74

The introduction of oxygen in the gas stream Ar initiates the oxidation of

germanium source. It is hypothesized that the insufﬁcient supply of oxygen results in the
formation of GeO on the surface on source material. Even if small amount of GeOz is
formed, it will react with the underlying layer of Ge and transform into GeO. Any thicker
layer of GeOz is ruled out since thermal stability of GeOz (Melting point lll6i4 0C)
will bring the further oxidation of Ge to a stand still. This is contrary to the experimental

observation that the source material completely disappears in less than 10 minutes of

reaction. Therefore, the ﬁrst reaction occurring during the grth of NWs is

Ge(s)+ 02 (g) -—> 2GeO(s) (4.4)

It may be emphasized that for every temperature, there is an optimum oxygen
concentration at which the oxidation rate of germanium does not surpass the sublimation
rate of the monoxide being formed. Otherwise, a thick layer of dioxide will form on the
surface of the germanium and the oxidation process will virtually come to a halt. The
GeO(s) formed during reaction (4.4) is very volatile and has high vapor pressure as
shown in Figure 4-5.

The GeO sublimates at 710 °C[107] and at the reaction temperature, it

immediately transforms to gas phase according to

G60 (5) —> GeO (g) (4.5)

75

 

  
   

 

 

-—- Pressure over GeO
w... Pressure over Ge+GeO2

2 _
:E’ 1 ~
E
E.
o.
a) 0 *
_o

-1 -

'08 0.9 1 1.1 1.2

1o3rr (K)

Figure 4-5. Vapor pressure of GeO at different temperatures.

However, oxidation of GeO in air begins at 550 °C and the presence of oxygen in
the grth chamber can lead to oxidation of GeO vapors as these diffuse away from the

source under the ﬂow of carrier gas according to reaction[107]

GeO(g)+ 1/202 (g) —) 0602 (s) (4.6)

Oxidation of GeO is not expected to be signiﬁcant due to the low concentration of
oxygen in the environment otherwise thermally stable layer of GeOz will deposit on the

substrate(s) and the walls of reaction chamber. The walls of the quartz tube in the high

temperature region do not get any discemable coating and the substrates tend to turn

black (color of GeO) instead of white (color of GeOz) at the beginning of synthesis.

76

u...

r“
...

 

Nevertheless, the actual vapor phase is expected to be a mixture of mostly GeO along

with some GeOz. The GeOz is likely in the form of ﬁne particles due to its thermal
j stability at reaction temperature. The visual observation of a whitish cloud formation that
( rises from the source material after starting the oxygen ﬂow supports the presence of
some GeOz in the vapor phase as is evident in Figure 4-6. In fact, some experiments did
yield some spherical or faceted particles along with nanowires on the substrate as shown
in Figure 4-7 and Figure 4-8. However, precise quantitative analysis of vapor phase
species requires more elaborate experimental set up (like X-ray photoelectron

spectroscopy) than the one available.

 

 

Figure 4-6. Formation of cloudy reactants after introducing oxygen into
the quartz tube.

77

 

 

 

 

 

 

 

Figure 4-8. Faceted germanium oxide particles with NWs.

78

 

In the vapor phase, there is also a thermal decomposition pathway for GeO(g)
coming off the Ge source. It is known that solid GeO disassociates into GeO; and Ge at

high temperatures. The dissociation reaction of GeO is highly temperature dependent. It
may take several hours for decomposition at 600 °C but at 700 °C it takes a few

minutes[98]. The thermal dissociation of GeO proceeds according to reaction

ZGeO (g) <:> Ge + GeOz

The reaction temperature controls the probability of forward or reverse

reaction[98]. In case of deposition on a condenser (or cold ﬁnger), GeO will deposit
without decomposition. On the other hand, it will decompose into Ge and G602 without a

condenser at fairly high temperatures. Since the substrates are also located in the high

temperature zone, we expect the higher rate of decomposition based deposition i.e.,
substrate gets a mix of GeO, GeOz, and Ge. The situation is further complicated by the

presence of a thin gold ﬁlm on the substrate that forms a eutectic with the underlying

silicon. Kim et al. proposed deposition of GeO on substrates that subsequently oxidizes

to form Ge02 during growth of nano-cones by thermal heating of Ge powder at 950 °C
under an oxidizing environment with a N2 background[108]. The exact manner of

adsorption of GeO or G602 into Au cannot be predicted since no information about the

ternary phase diagram for Ge, Au, and oxygen system could be found. However, it is

postulated that the presence of liquid eutectic droplets facilitate incorporation of GeO,
Ge, and GeOz into the drOplet yielding an initial sub-oxide composition

Ger (l < x < 2) of NW[109]. The composition of the nanowire then transforms to more

stable GeO; by incorporation of oxygen from environment as well as through loss of

79

GeO by vaporization. The excess GeO diffuses away from the reaction zone and
condenses in the colder part of the tube. The ends of tube get a yellowish to brownish
back coating that is characteristic color of GeO. Some ﬁne powder like coating is

obtained further downstream part of tube.
Thus the most plausible mechanism for the formation of GeOz NWs can be

summarized as:

(a) Oxidation of Ge to form GeO(s).

(b) Transformation of GeO(s) to GeO(g).
(c) The transport of GeO(g) (and possibly some small percentage of GeOz

formed in gas phase) to substrate.
((1) Preferential adsorption of GeO on the substrate (black appearance of

substrate) at the sites of gold catalyst and possibly some associated
decomposition into Ge and GeOz.

(e) The growth of Ger (1<x<2) NWs via the VLS mechanism by
absorbing more source material from the vapor phase.

(f) The evolution of stable GeO; NWs by incorporation of oxygen and

loss of GeO.

It may be noted that a different mechanism was proposed in past in which the

reaction results in adsorption of GeO; vapors on the Au eutectic on the substrate[l 10].

However, the thermal stability of GeOz does not support the feasibility of having any

80

signiﬁcant quantity of GeO; in vapor phase at temperature below 1100-1200 °C. The
smooth surface of the NWs also suggest a viscous state of the nanosize Ger (1 < x < 2)

exists at the tip of the NW during growth[100]. The existence of very ﬁne GeOz particles
in the gas phase cannot be ruled out but any large concentration of Ge02 is not expected.

If that was the case, one would expect much coarser NW surface (the GeOz particle are

likely to be incorporated as such into NWs due to high thermal stability). This is contrary
to the actual observation (TEM and SEM studies) of smooth NW surface with the
exception of some instances where ~ micron size particles were observed along with

NWs. Additional evidence relates to the physical appearance of substrates during
synthesis. One will expect visual whitening of the substrate upon adsorption of GeOz

instead of ﬁrst blackening followed by appearance of random white dots on the substrate.

The density of these white dots then gradually increases as reaction proceeds and tends to
cover the whole substrate. In one control experiment, some of the ﬁne GeOz powder
collected from the walls of the quartz tube was placed on Au coated Si substrate to see if
the GeOz on Au ﬁlm can form a eutectic and promote NWs growth. However, GeOz

particles remained as such on the substrate and no NW growth was observed as shown in

Figure 4-9. Only the Au ﬁlm broke into small particles as expected. The EDS of the

substrate next to GeOz particles did not show any Ge or GeOz indicating stability of
GeO; for the given the reaction conditions.

The explanation so far has ignored the presence of H2 in the chamber. It is known

to be an oxide reducing agent depending upon the partial pressures of hydrogen and that

of water in the system for a given temperature. As a practical manner, the partial pressure

81

of water is equal to the base pressure of the system[111]. In our case, the base pressure is

~ 200 mTorr and partial pressure of H2 is ~ 250 mTorr (for a total pressure of 5 Torr).

 

Figure 4-9. GeO; particles on Au coated Si substrate after typical
synthesis run.

Figure 4-10 is a plot of thermodynamic stability of GeOz as a function of ratio of

hydrogen pressure to water pressure for different reaction temperatures[112]. For a

P(Hz)
P(HZO)

z 1.25 , the G60; is unstable for all temperatures above 500 K and is expected
to decompose to constituent elements. Since the NWs synthesis is carried out above this
temperature, we do not expect any GeO; on the source material at the start of oxygen

ﬂow. The situation changes after introduction of oxygen into the system and the

oxidation of Ge/GeO become dominant compared to the reduction of GeO/G602 by H2.

82

'5 V
”l: .

71m
not».

AﬁUApr v. e\« at; a vat

l N

1“

ék

The effect of H2 will be discussed again later in this chapter during study of background

gases on NWs synthesis.

10;

  
  

GeOz unstable

0

.3
O
..,

P(H2)/P(HZO)

GeOz stable

_\
O

 

-2 l r n r 4]
10200 400 600 800 1000 1200

Temperature (K)

 

Figure 4-10. Plot showing GeOz stability as function of temperature and
the hydrogen to water vapors pressure ratio[112].

4.2.4 Synthesis temperature proﬁle
The furnace temperature was measured and controlled by using a ‘K’ type
thermocouple placed between quartz tube and furnace. The actual growth temperature
inside the quartz tube at the site of source and substrates is expected to be somewhat
different from that measured by the outside thermocouple due to physical separation and

environment. To get a more accurate value of temperature inside the fumace, a second

83

‘K’ type thermocouple was placed inside the quartz tube. Figure 4-11 shows a typical
temperature proﬁle during NWs synthesis run measured by the two thermocouples. As
expected the actual temperature inside the quartz tube is lower than the outside

temperature by about 65—75 °C under steady state for an outside temperature of 850 °C.

 

 

 

900" —Temp outside
300 - ----- Temp inside
S5
V 700-
(D
‘5
:3 600-
8
E 500-
(D
|._
400
300-
200

1000 2000 3000 40100
Time (sec)

Figure 4-11. Tube furnace typical temperature proﬁle during nanowire synthesis.

4.2.5 Synthesis pressure profile
Pressure dependent studies during NW5 synthesis revealed a typical variation
pattern as shown in Figure 4-12. It is important to mention that the pressure is measured
at the downstream end of the tube and may have minor deviation from the actual value at
the location of source. However, calculation of conductance of vacuum line from tube
center to gauge location for a viscous ﬂow yields a value of 21 .32 US at 5 Torr and this is

much larger than the total gas throughput of ~1 .267 Torr-L/s (corresponding to 100 sccm

84

.1 in...

I‘Vﬂ'

..~rr

 

ﬂow rate). The calculated difference between measured and true values of pressure at the

location of the reaction is only 0.06 torr for a total ﬂow rate of 100 sccm at 5 Torr and

room temperature. The synthesis is initiated at ~5 Torr by introducing Ar + 02 (Ar 60%,

balance 0;) mix in the system at a ﬂow rate of 50 sccm. The variation of the pressure at

the synthesis temperature of ~850 °C can be analyzed to get some insight into the

different reactions taking place during synthesis. Three distinct pressure regimes can be

identiﬁed in Figure 4-12:

1.

Region I : This region is marked by the rapid rise to a peak value. The rise
in pressure can be attributed to the formation of volatile GeO that is in
vapor phase at the synthesis temperature. Also one can notice a brief hold
of pressure near 10 Torr, which is probably an artifact of the pressure
gauge.

Region II : This region indicates a gradual drop in pressure value from the

peak attained in Region I. This is probably due to the conversion of
GeO(g) to Ge02(s) that reduces the contribution of gas phase to pressure.

Also some of the GeO(g) is expected to escape the high temperature
region and condense in the colder part of the chamber. The lowest value of
pressure is reached once nearly all of the GeO vapors are transformed into
solid phase. The source material is found to completely disappear before
reaching the pressure minimum in this region.

Region III : Pressure starts to slowly rise in this region which is attributed

to the continuous ﬂow of background and reaction gases into the chamber.

85

I51.

It is important to re-iterate that a convector P-type gauge is used to measure the
pressure. The readings of convector type of gauge at higher pressure (>1 Torr) are highly
dependent upon the type of gas. This is due to different thermal conductivity of different
gases. Typically the gauge is calibrated for a certain type of gas (mostly nitrogen or air)
and a correction factor needs be applied to convert indicated pressure to true pressure
when using a gas different from calibration gas. For example, a gas calibrated with
nitrogen will practically give accurate reading for nearly every gas below a pressure of 1
Torr. However, same gauge will indicate a pressure of only 8.83 Torr for a true pressure
of 100 Torr when used with Ar gas. Hence the analysis presented above shows an
accurate trend of pressure variations but the absolute values of pressure are expected to
have signiﬁcant variations from indicated values. This will discussed again later in this

chapter while studying effect of different background gases on NW synthesis.

 

Pressure (torr)

 

 

 

4 _
2 _.
O0 200 400 600 800 1000 1200

Time (s)
Figure 4-12. Typical pressure proﬁle during NW synthesis.

86

4.3 Factors affecting growth of nanowires

The NWs synthesis is affected by a number of factors including thickness of
catalyst ﬁlm, temperature, pressure, and background gas type, ﬂow rate, oxygen ﬂow rate
etc. The effect of these parameters on the resulting NWs will be discussed in turn in this

section.

4.3.1 Role of gold catalyst

The catalyst plays an important role as the nucleation site for initiation and
continued growth of the nanowires in VLS mechanism. In typical semiconductor
nanowire grth by VLS mechanism, the metal catalyst forms a eutectic with
semiconductor and this eutectic has a lower melting point than either of its constituents.
For example, a eutectic containing 19.5 atomic % Si and 80.5 atomic % Au is in liquid
phase for temperature 2 360 °C. This liquid phase serves as the preferential site for
adsorption of gas phase species into the liquid phase catalyst or its eutectic form at the
growth temperature. This results in nanowire growth that is highly localized to the
catalyst coated area. Figure 4-13 shows dense and patterned NWs grth on Si(100)
substrate that was partially coated with 7 nm Au ﬁlm. As can be seen, the NWs growth
(white areas) is strongly limited to the Au coated area with no growth on uncoated area
(bottom left corner of image). The catalyst speciﬁc growth is one of the strong

characteristics of VLS mechanism and gives evidence of VLS mechanism taking place.

87

 

 

Figure 4-13. 6602 NWs on Si(100) partially coated with 7 nm Au ﬁlm.

Another characteristic of VLS is that the NWs generally terminate in catalyst
particle that has a spherical shape. However, not all of our NWs were found terminated in
catalyst particle. This could be due to breaking off of the catalyst particle during growth
or cooling, or the catalyst might be incorporated into the nanowire during cooling.
Another factor could be co-existence of other synthesis mechanisms like oxide-assisted
growth (OAG)[88] or vapor-solid (V S)[91]. However, the argument against OAG or VS
is the complete absence of NWs in uncoated area whereas none of these other
mechanisms need a catalyst to yield NWs. It is postulated that the VLS is the dominant
mechanism but other mechanism may exist. The catalyst coating signiﬁcantly enhances
the adsorption of vapor phase species to initiate growth and subsequently different
growth mechanism including VLS, VS and OAG may contribute to NW growth to

varying degrees. Some of the NWs do show a spherical particle at the tip conﬁrming

88

existence of VLS growth. Figure 4-14 shows the SEM image of such a single NW with

a particle at the tip.

 

Figure 4-14. Ge02 NW showing particle at tip.

4.3.1.1 Gold film annealing

The initial catalyst ﬁlm thickness is the main controllable parameter in VLS
mechanism as it has a direct bearing on the diameter of the nanowires. A systematic study
of the evolution of catalyst ﬁlm into particulates was carried out to get an estimate of
initial size distributing of catalyst particles. Different thickness of gold ﬁlms were
deposited on Si(100) wafer and the samples were subjected to typical synthesis
conditions without germanium source material or oxygen ﬂow. Figure 4-15 through
Figure 4-19 show the SEM images and corresponding particle size distributions after
annealing different thickness ﬁlms. It may be mentioned that the particle size
measurement for very thin ﬁlms (3 nm) is not very accurate (an error of ~20%) due to

poor image quality at very high magniﬁcation of the SEM. Nevertheless such coarse

89

measurement is sufﬁcient to give an idea of the size distribution expected for very thin

ﬁlms.

 

14- mean= 13

std_dev= 3

Number of islands

 

 

9 11 13 15 17 19
Diameter of Au islands (nm)

Figure 4-15. SEM image and particle size distribution for 3 nm Au ﬁlm.

90

 

mean= 39

std_dev= 10

Number of islands

 

 

(20 26 32 38 44 5O 56 62 68 74 80
Diameter of Au islands (nm)

Figure 4-16. SEM image and particle size distribution for 7 nm Au ﬁlm.

91

 

 

 

mean= 92
35
std_dev= 33
w 30
E
g 25—
.9
“g 20
8
E 15~
3
Z

10*

 

 

0 40 80 120 160 200 240 280
Diameter of Au islands (nm)

Figure 4-17. SEM image and particle size distribution for 15 nm Au ﬁhn.

92

 

 

mean= 131

50 F std_dev= 57

A
0

Number of islands
N o.)
O O

 

 

0 100 200 300 400 500
Diameter of Au islands (nm)

Figure 4-18. SEM image and particle size distribution for 21 nm Au ﬁlm.

93

 

mean= 191

32 _ std_dev= 109

N
A

.3
0)

Number of islands

00

 

 

O
O 100 200 300 400 500 600 700 800
Diameter of Au islands (nm)

Figure 4-19. SEM image and particle size distribution for 28 nm Au ﬁlm.

94

Figure 4-20 shows the mean particle diameters for different ﬁlm thickness and
relationship is approximately linear. It may be emphasized that the particle size
distribution is in agreement with earlier reports. For example, Sharma et al. found that
approximately 85% of the eutectic droplets were smaller than 50 nm for 5 nm thick e-

beam evaporated Au ﬁlm on Si substrates that were annealed at 940 K for 5-10 min at 95
Torr in 1 slm H2 ambient[l 13]. Though it is not possible to directly relate island sizes in
our case since we differ in ambient, anneal time, and temperature. Still the mean values
of metal island sizes are comparable (we get ~40 nm diameter islands for ﬁlm thickness
of 7 nm). It may also be noted that the size of the Au particles would be expected to
change from these dimensions when at the actual conditions (ﬂow of reaction gases and

formation of reactants) since the surface energies are dependent upon the ambient phase.

300 ~

N
01
O

200 -

150*

 

100* l

 

.50 ~

Mean diameter of Au islands

 

00 57 10 145 20 215 310

Thickness of Au film (nm)

Figure 4-20. Mean particle size distributions after annealing different
thickness Au ﬁlms. Error bars indicate standard deviation from mean.

95

4.3.1.2 Gold film thickness affect on GeO; NWs diameter

Next we move on to study diameter of the nanowires that result from different
thickness gold ﬁlms. NWs were synthesized on Si substrates that had different initial Au
ﬁlm thickness in a single experiment to seclude any variations from one experiment to
another. The diameters of the NWs were determined from SEM studies by selecting at
least three different areas on each substrate. For each substrate, more than 100 NWs were
characterized using SEM. Figure 4-21 through Figure 4-23 show the diameter
distribution for NWs synthesized on Si substrate from 3 - 15 nm thick Au ﬁlms. It may be
mentioned that the sizes of the particles were measured from high-resolution SEM
images using measurement tool box of SEM software and then Matlab histogram

function was used to plot the size distributions.

100-
90—
80
70
60-
50-
40-
30*
20-
10*

00 50 100 150 200 250 300 350 400 450 500

Diameter of GeO2 NWs (nm)

mean= 118
std_dev= 73

Number of NWs

 

 

Figure 4-21. GeOz NW diameter distribution for 3 nm Au ﬁlm.

96

mean= 125

std_dev= 101
1 12 —

Number of NW5
‘1
O

 

 

0 100 200 300 400 500 600 700 800 9001000
Diameter of GeO2 NW5 (nm)

Figure 4-22. GeOz NW diameter distribution for 7 nm Au ﬁlm.

    

mean= 170
std_dev= 121

 

00 80 160 240 320 400 480 560 640 720 800
Diameter of GeO2 NWs (nm)

Figure 4-23. GeO; NW diameter distribution for 15 nm Au ﬁlm.

97

  
   
 

150 + NW5
' "5' Au particles
1?
5
4°: 100 ' ,A
Q) 0‘"...
E ‘0‘...
.9 ...e‘.
O ..o‘..
50f .. 0‘...
AIN'T. . n .

 

 

4 6 8 10 (2 14
Thickness of initial Au ﬁlm (nm)

Figure 4-24. Mean diameters of Ge02 NWs and annealed Au ﬁlm
particles.

Figure 4-24 shows the mean diameter for the gold particles and resulting NWs
for different catalyst ﬁlm thickness. It is immediately evident that the diameters of the
synthesized NWs do not have a one—to-one correspondence with the measured particle
diameters for that ﬁlm thickness. It is also important to point out that the size of the ﬁnal
eutectic particle at the tip of the NW is typically larger than the diameter of the nanowire
as can be seen in Figure 4-4 and Figure 4-14. Hence we start with a small Au particle
that yields a relatively large diameter NW ending in a tip particle of even larger diameter.
An obvious explanation for the increase in the diameter of the NW compared to the
starting Au particle is the increase in the size of eutectic particle following adsorption of
vapor phase species that increases diameter of NW5. This is supported} by the EDS
analysis of tip particles with TEM that shows presence of germanium as well as oxygen

in the eutectic particle. The oxidation of sub-oxides following nucleation can also

98

 

 

 

increase NW diameter. The dynamic conditions at the reaction temperature and ambient
may dictate different morphological evolution of NW. For example, the Au alloy may be
more distributed at the tip of the nanowire during growth and then the Au gets pushed out
during cooling leading to the formation of larger spherical particle at tip. Also the
mechanisms other than the VLS can be occurring simultaneously that contribute to larger
NWs than expected from VLS alone. For example, VS mechanism can result in micron
size whisker[91]. We did observe some faceted whiskers during synthesis as shown in
Figure 4-25. The surface of the whisker appears rough in the image but is probably due
to gold coating to facilitating SEM imaging. Some lateral growth on the NW can also
result in NWs thicker than the catalyst particle. The difference between annealed Au
eutectic particle size and the resulting NWs diameter is large for thinner ﬁlm. This can be
explained from the Gibbs-Thomson effect that predicts higher vapor pressure for smaller
diameter wires. As a result, very small diameter NW (in case of thin Au ﬁlm) may not be

feasible under the synthesis conditions employed in this work.

 

Figure 4-25. A faceted G602 whisker.

99

4.3.2 Effect of growth time on length of nanowires

In a typical VLS growth, the length of W5 is proportional to growth time.
However, this is valid only if precise control over the amount of reactant supply is
exercisable as is the case of chemical vapor deposition (CVD) reaction. However, such
control in not feasible in the experimental set up used in this work. The main reason is the
high volatility of reactants at grth temperature and the associated thermal inertia of the
system. It was observed that nearly all the source material manages to oxidize even if the
furnace is turned off after reaching growth temperature. The thermal inertia of the system
and fast oxidation of source makes it difﬁcult to exercise control over length using
growth time as variable. Although it is difﬁcult to precisely determine the amount of
reactants (like GeO in this case) but it is speculated that the supply of reactant is high at
the beginning of reaction and then decreases gradually as the source material is
consumed. The rate of grth of W5 is also expected to follow a similar trend. The
typical length of NWs was found to be several tens of microns. The random and
intertwined/entangled orientation of NWs on the substrate makes the precise length
measurement difﬁcult during SEM studies. A better insight can be obtained by studying
cross-section of the substrate and looking at the height of the NW cluster layer. Figure
4-26 shows cross-section of the substrate and conﬁrms randomly oriented NWs
extending up to 100 um above the substrate. It is therefore concluded that the overall

length of NW is of the order of tens of microns.

100

 

Figure 4-26. Cross-section of substrate showing long NWs extending
from substrate.

4.3.3 Effect of oxygen ﬂow rate

The NWs synthesis was found to be very sensitive to oxygen concentration as
expected. Figure 4-27 shows NWs synthesized on Si substrate with 7 nm initial catalyst
coating using different oxygen ﬂow rates. As can be seen, the best coverage (white area)
is obtained for an oxygen ﬂow rate of ~ 50 sccm. This agrees with earlier discussion of
Ge oxidation where it was found that the oxidation of Ge to produce GeO or GeOz is a
strong function of concentration. The observed low yield of NWs for high ﬂow rate of
oxygen can be explained by increased conversion of GeO into GeO; that limits the spatial
coverage of NWs on substrate closer to source. On the other hand, low concentration
allows more of GeO to escape from reaction zone, thus reducing yield of NWs. The total

ﬂow rate was also found to be a signiﬁcant parameter impacting the NW growth.

101

Reducing the total ﬂow rate generates high GeO pressure area over the source and
incoming ﬂow of oxygen cannot penetrate this high-pressure area. Instead GeO vapors
from source diffuse toward the upstream end and meet with 02 in the incoming gas ﬂow
to form GeO; at upstream. This was evidenced by white coating on upstream end of
quartz tube for low total ﬂow rate. High ﬂow rate on the other hand pushes the GeO
vapor quickly out of the reaction zone resulting in increased coating at the downstream

end. A total ﬂow rate of ~ 100 sccm was found to yield optimum yield of NWs on the

substrates.

65 sccm

 

50 sccm

 

40 sccm

 

Figure 4-27. Effect of oxygen concentration on NW synthesis. In all
cases, the total ﬂow rate (background + oxidation rrrix) was 100 sccm.

102

4.3.4 Effect of synthesis temperature

Synthesis temperature showed a pronounced impact on NWs yield. Figure 4-28
shows optical images of Si substrates covered with NWs synthesized in the temperature
range 865 —- 765 °C. The yield of the NWs was found to be optimum in the range 830-840
°C. The density of substrate coverage is less at high temperature as can be seen in Figure
4-28(a). This can be explained in terms of high energy of volatile GeO formed at high
temperature that tends to quickly move out of the reaction zone. Also the corresponding
high temperature of the substrates results in less sticking on substrate and greater
desorption. This implies that most of the source material manages to escapeiout of the
reaction zone at high temperatures (2 850 °C) without yielding nanowires at the
substrates. As the temperature is lowered, better yield of W5 is obtained for

temperatures of 850-800 °C. This may be considered optimal temperature range for
creation of GeO reactants and subsequent conversion into Ge02 nanowires at the
substrate. Lowering the temperature further results in reduced yield as seen in Figure
4-28(e). The low temperature effect can be attributed to low evaporation rate of source
decreases. The growth is also found to be localized closer to source indicating GeO

vapors do not have enough supersaturation to yield uniform growth far from source.

103

 

Figure 4-28. G602 NWs synthesized at different temperatures (a) 865 °C
(b) 840 °C (d) 830 °C (d) 800 °C (e) 765 °C.

4.3.5 Effect of background gas

Although most of the nanowires were synthesized using a mixture of Ar (95%)
and hydrogen (5%), some experiments were carried out using pure Ar or N2 as a
background gas. The morphology of the NWs synthesized using N2 and Ar gases is
similar as shown in Figure 4-29 and Figure 4-30 respectively. However, some variations
in the diameters were noticed. Figure 4-31 shows the mean diameter of GeO; NWs

104

obtained for different initial Au ﬁlm thickness by using different background gases. The

smallest diameter is obtained for Ar and hydrogen mix. This is probably due to reducing
affect of H2, which in a way etches away, or reduces lateral growth of oxide NWs. The
diameter versus gold ﬁlm thickness dependence is similar for the case of Ar and Ar
mixed with H2. However, the case of N2 background has relatively smaller increase in

NW diameter as Au thickness is increased from 7 nm catalyst ﬁlm to 15 nm. There is no

clear explanation for this affect. One possibility is that the annealing studies of the Au
ﬁlm was done in Ar environment and annealing in N2 might somewhat change the size
distribution of Au particles. There have been reports of breaking up of catalyst particles

due to electrical charging in N2 background ﬂow. This might explain why the NWs
synthesized using N2 background show less diameter dependence on catalyst ﬁlm
thickness. Another related observation is the different pressure proﬁle for N2 background

compared to Ar background. In case of N2 background, the pressure initially rises to
~180 Torr (~12 Torr for Ar) and then drops down to 110 Torr (~7 Torr for Ar). The ratio
of the pressure change in the two cases (N2:l80/110 = 1.64 and Ar:12/7 = 1.71) is

compatible. The convector pressure transducer has different calibration curves for
different gases at higher pressure (>1 Torr) due to their different thermal conductivities.
Hence it is mainly the different response of the sensor to the two gases that is responsible
for observed disparity of pressure in case of different background gases. The values of
pressure for nitrogen background are considered more authentic since that is the typical
calibration gas for the sensor. The reading of atmospheric pressure as 760 Torr also

supports the gauge is reading nitrogen pressure more accurately.

105

 

Figure 4-29. SEM image of GeO; NWs synthesized using N2 as
background gas.

 

Figure 4-30. SEM image of GeO; NWs synthesized using Ar as
background gas.

106

180

160

 

Diameter of GeO:2 NWs (nm)

 

120- . . . . .
4 6 8. 10 12 14
Thickness of initial Au ﬁlm (nm)

Figure 4-31. GeOz NWs diameters for different background gases.

4.3.6 Effect of background gas pressure

The background gas pressure at the start of the synthesis affects the yield of the
NWs by changing the degree of supersaturation in the gas phase. The main role of the
background gas pressure is to conﬁne the gas phase reactant near the source and
substrate. Figure 4-32 shows two substrates next to location of Ge source with NWs
synthesized at two different background gas pressures. The low gas pressure allows the
reactants to reach relatively far from the source material. On the other hand, high
background gas pressure keeps the reactants localized and some of the GeOz is found
even at the location of the source. This can be seen for the synthesis at 72 Torr where the

dome shaped white GeO; is formed at the location of Ge source, i.e, high pressure kept

the GeO vapors so localized that these transformed to dioxide at the source instead of

107

reaching to the substrate. Another indirect evidence of background pressure at work can
be from the observation of coatings at the ends of quartz tube. Figure 4-33 shows optical
images of quartz tube downstream end before and after reaction. The amount of white
coating of GeO; can be an indirect measure of the conﬁning capacity of the background
gas. In case of too much coating at the downstream end of the tube, one can conclude
insufﬁcient gas pressure for conﬁnement of reactants in the hot zone. Negligible or no
coating means too much pressure exists in chamber to let the reactants reach out to the

substrates.

P = 28 torr

Ge source
on AlN plate

P = 72 torr

 

Figure 4-32. Effect of background gas pressure on NW synthesis.

108

 

Figure 4-33. White GeOz coating at the end of quartz tube.

The preceding discussion has elaborated various aspects of NWs growth and
identiﬁed a set of optimal reaction parameters in terms of temperature, pressure, gas ﬂow
rates, and catalyst thickness to synthesize GeO; NWs in high yield. This sets the stage for

Raman characterization of NWs and their potential used as substrates for SERS.

4.4 Raman characterization of GeO; nanowires

This section provides details about the Rarnan equipment parameters used and the
results of Raman characterization of GeO; NWs substrates. The substrates were
characterized before and after applying metal coating. A signiﬁcant change in Raman
spectra of substrate was noticed after applying a metal coating. The potential sources of

this change were thoroughly investigated.

109

4.4.1 Experimental parameters

The Raman system used in this study was EzRaman-L from Enwave Optronics
Inc. with 6 cm’1 resolution at 785 nm excitation. The sample was mounted on a xyz—stage
and illuminated by the Raman system laser beam via an optical ﬁber link. The Raman
probe head was attached to a separate z-stage that facilitates precise focusing of laser
beam onto the sample surface independent of sample movement. All spectra were
acquired with a focusing lens (N .A. = 0.22) that yields a spot size of less than 100 um
diameter on the substrate. The laser power on the substrate was nearly 30 mW as
measured with Newport optical meter (model 815 in conjunction with 818-SL detector

and 883-SL attenuator). The same laser power was used in all studies except where noted.

4.4.2 Raman spectra of GeO; nanowires

The bare substrate containing GeOz nanowires was ﬁrst scanned by the Raman
system to get the reference signal and to ensure no impurities are present that can cause
interference with the Raman signal. Figure 4-34 shows the Raman spectra of GeOz
nanowires (NW s) synthesized on Si(100). The Raman lines obtained from the nanowires
are similar to the ones reported by Macoulaut et al. for trigonal phase of (360211141 This
Raman spectrum agrees very well with the trigonal phase spectra of GeOz and
complements the EDS and diffraction results to prove that the synthesized nanowires are
made up of crystalline trigonal GeOz. The water solubility of the synthesized nanowires

is also an indirect-indication of their trigonal growth phase and this has been further

continued by the Raman spectrum.

110

 

 

 

 

 

 

 

“0 444 i
701
3‘
C
3
5
e
'E ’ TrigonalGeOz
S.
3‘ _
-- 516 .
g 263330 593' 881 973
*5 860 A
3Rutile GeOz
. lwtr‘ﬂj J 1 1
200 800 1000 1200
0060Raman shift(cm )
(b) 4
3x10
25
Zr

A
Am

Raman lntensity/Arbitr. Units

F’
01

 

 

 

200 450 700 950 1200

Wavenumber/cm'1

Figure 4-34. (a) Raman spectra for two phases of GeOz. Reproduced with
permission from ref [114]. (b) Raman spectra of GeOz nanowires on
Silicon substrate, tacc=15 s.

111

4.4.3 Gold coating of GeO; nanowires

The nanowire substrates were sputter coated with ~ 21mm of gold to allow
excitation of surface plasmons. The choice of gold is based upon its compatibility with
biological systems, chemical inactivity and good SERS enhancement for NIR
excitation[l 15]. The coated substrates were characterized by the Raman system before
application of an analyte to establish background signal. Figure 4-35 shows the observed
Raman spectra of gold-coated nanowires with 15 5 integration time. This spectrum shows

a broad continuum over all wavenumbers that peaks at about 800 cm'1 and almost
completely obscures the original GeOz spectra. Also shown is the Raman spectrum from

an uncoated part of same substrate under identical conditions. From this comparison it is
clear that the gold coating is causing the broad background. Although the intensity of the
NWs spectra is expected to diminish with Au coating, the source of broad background is

not exactly known.

x104

anncoated NWs
—Au coated NWs

.N
romeo

A
T

 

 

.0
or

Raman lntensity/Arbitr. Units
A
01

 

)- /j~
La. . J1»... “JLJL .
200 400 600 800 1000 1200

Wavenumber/cm'1

Figure 4-35. Raman spectra of 21 nm Au coated GeO; NWs, tacc=15 3.
Also shown are spectra for uncoated NWs.

112

Very similar broad continuum was observed from other locations of the coated

area and for different substrates. For example, Figure 4-36 shows the Raman spectra
from 8 GeO; NW substrate before and after coating it with a 15 nm thick Au ﬁlm and the

same broad background appears after coating. The introduction of this continuum in the
Raman spectrum was very puzzling since it usually arises due to ﬂuorescence, but it is
known that ﬂuorescence is quenched in the vicinity of metal. However, what is observed
is something similar to ﬂorescence introduced by the metal or the deposition process.
Some potential sources of the broad background were hypothesized as impurities in the
deposition process, impurity in metal source material, and moisture on the substrates. A

systematic study was done to conﬁrm or eliminate each of these potential sources.

3000 -
------ Uncoated NWs

-—Au coated NWs
2500 ~

2000

I

1 000

I

Raman Intensity/Arbitr. Units
at
O
‘?

500~ 1.

        

 

. war .. .. ~ t 5'. 1.-.,
1200 1400 1600
1

200 0 e 600 80 000
Wavenumber/cm'

Figure 4-36. Raman spectra of 15 nm Au coated GeOz NWs, tacc=5 3.
Also shown is the spectra for uncoated NWs.

113

The ﬁrst step towards isolating the source of the broad background is an
understanding of how the gold ﬁlm affects the Raman spectra of a reference material like
silicon. For this purpose, a bare silicon wafer was sputter coated with ~21 nm thickness
of gold by the same sputter coating process. Figure 4-37 shows the effect of gold coating
on the Raman spectra of a bare Si(100) wafer acquired at 785 nm laser excitation with 5 5
integration time. The bare wafer spectra shows a strong Si peak at 520 cm'1 and a
characteristic broad ﬂuorescence background towards higher wavenumbers (beyond
1200 cm'l). The intensity of the characteristic peak at 520 cm'I is reduced by more than
75% with gold coating. An even more striking effect of gold coating is the almost
complete elimination of the Si ﬂuorescent tail. This is expected, as the metal tends to
quench ﬂuorescence and shows that the sputter gold coating process is not a likely source
for the introduction of broad continuum seen in the Raman spectra of gold-coated

nanowires.

2500 ' --»- Bare Si(100)
—21 nm Au coated Si(1

2000 -

.3
(11
O
O

1000-

500 -

Raman lntensity/Arbitr. Units

 

 

 

500 1 000 1 500 2000

Wavenumber/cm'1

Figure 4-37. Raman spectra for bare silicon wafer and after coating with
~21nm gold.

114

 

A further insight into the impact of gold-coating on substrate can be achieved by
studying the effect of Au thickness on the Raman spectra of Si. Four Si(100) wafers were
coated with gold ﬁlms having thickness from 5 — 20 nm. Raman spectra from these gold
coated wafers was acquired under the same experimental conditions and the intensity of
the main Raman peak was used to compare the effect of gold thickness. Figure 4-38
shows the variation of the 520 cm'1 Raman peak of Si as the Au ﬁlm thickness increases
from 5 nm to 20 nm. As expected the Raman intensity decreases with increasing Au
thickness due to the attenuation affect. This further conﬁrms that a smooth metal coating

attenuates the Raman signal from underlying substrate.

 

3000 . e
I
2500 ~

2000

1000

Raman Intensity/Arbitr. Units
at
O
O

 

500

 

 

 

5 1O 15 20
Au ﬁllm thickness (nm)

Figure 4-38. Variation of 520 cm'1 Raman peak intensity with Au ﬁlm thickness.

115

A supplemental study was done to verify deposition process purity by using a
much cleaner e-beam evaporation with new gold source material. The nanowire
substrates were loaded in the deposition chamber that was evacuated to 5x106 Torr. The
substrate was set to continuous rotation and 10 nm gold was deposited using e-beam
evaporator as measured by a quartz crystal monitor (QCM). The Raman spectra from the
e—beam deposited Au coated substrate is shown in Figure 4-39 and once again the impact
of the gold coating is evident as a strong continuous background that peaks in the 800-
1000 cm'1 region of spectrum. This experiment conﬁrms that the source of background

does not relate to the impurities in the coating process or source of the metal.

     

------ Uncoated NWs
—Au coated NWs

ty
N
0|
0
O

 

200 400 600 800 1000 1200 1400 1600
Wavenumber/cm'1

Figure 4-39. Raman spectra of 10nm e-beam high purity gold coated

GeOz nanowires, tacc=5 5. Also shown is the spectra from an uncoated part
of the sample.

116

The GeOz is known to absorb moisture from environment and the gold coating
might trap it under the metal. Although water is a weak Raman scatterer, the presence of
gold can enhance its signature and contribute to the broad background observed after
gold coating. To eliminate the inﬂuence of adsorbed moisture, the samples coated with
21 nm gold were stored in desiccator for 36 hrs and then analyzed with the Raman
system. Figure 4-40 shows 10 spectra obtained from the same spot on the sample with 5
5 signal integration and 10 s delay between two acquisitions. No signiﬁcant change in
shape or intensity of the broad continuum is found after drying of the samples and

suggests that moisture is not causing it.

 
 
  

 
  

“it ,1
1 at)?

 

 

200 400 600 800 10100 12100 14100 1800
Wavenumber/cm”1

Figure 4-40. Successive Raman spectra from same point on sample for

21 nm Au coated GeOz NW sample dried in a desiccator for 36 hr.

117

As a last step in the investigation of background continuum, the sample was
heated in situ to burn off or drive out any contaminant that might be causing observed
ﬂuorescence and then observe any change in its Raman spectrum. This was done by
placing sample on top of a resistive button heater and monitoring the temperature by a K-

type thermocouple. Figure 4-41 shows the Raman spectra of ~21 nm gold coated GeOz

nanowires heated to 146 °C. For comparison, the room temperature spectra is also shown
as a dotted line in Figure 4-41. No signiﬁcant change due to heating of sample is
observed. It is therefore, concluded that the broad background introduced is caused only
by the presence of gold coating rather than any contaminant or moisture effect. The exact
cause of this broad continuum is not fully understood yet. It is a common observation that
the molecular SERS spectrum is accompanied by a broad background covering most of
the normal 0-3000 cm'l range[73]. Some researchers attribute it to electronic Raman
scattering from metal[l 16, 117] while others consider it lmninescence[118]. In our case,
the broad background appears following the metal coating of nanowire substrate. Since
the background continuum is present even before the application of analyte molecule, the
most probable cause of the continuum is the electron scattering off defects from rough
gold ﬁlm covering the nanowires that allows for electronic Raman scattering over a broad
energy range[39]. Further understanding of the broad continuum is of interest, however
for the purpose of detection measurements it can be removed through a baseline

correction.

118

(O
O
O
O

----- Room temperature
8000 ” — 146 °C

 

 

200 400 600 800 1000 12100 1400 1800
Wavenumber/cm’1

Figure 441. Effect of ~21nm gold coated GeO; nanowires at 146 °C,
tacc=5 5. Insert shows room temperature Raman spectra.

TEM studies of gold—coated NWs were also carried out to investigate the
morphology of the Au coating. The e-beam deposition of Au on the NW covered
substrate was done at room temperature without any subsequent annealing. Figure 442
shows the TEM image of GeO; NWs broken off from Au coated Si substrate and clearly
conﬁrms that the Au does not wet the NW but forms 3D islands. It is known that mid-to-
late transition and noble metals do not generally wet oxide surfaces, but make particles

that only cover a fraction of the surface[119]. Similar 3D islands of thermally evaporated
Au on TiOz at room temperature have been previously observed[120]. Also only the side

of NW facing the incident Au source vapor is being coated. Most of the NWs showed
similar wetting pattern with some variation in density and size distribution (typically 2-10

mm) of Au islands. This type of metal island formation during low temperature deposition

119

1.0

p"

has been explained using an island growth model in which kinetic effects dominate and a
large density of smaller Au islands are nucleated[120, 121]. The shadowing effects can
also supplement anisotropic growth of 3D islands since the nanowires are oriented at

random angle with respect to the incident vapor[36].

 

Figure 4-42. TEM image of 21 nm Au coated GeO; NW. Insert shows
another NW with larger Au particles.

1 4.5 Evaluation of SERS substrate
SERS evaluation of 060; NWs was done using different analytes including
Rhodamine 6G, Nile Blue, 4-methylbenzenthiol, 1,2-benzenedithiol and trans-l,2-bis(4-
pyridyl)ethylene. These analytes are the one commonly employed for evaluating SERS

substrates and can be broadly categorized into two groups — (a) Dye molecules like R6G

. 120
l

are

an“.

1
‘0' I
I in
it“.
.

'1
I n

I“!

and nile blue, and (b) Thiol molecules like MBT and BDT that form self assembled
monolayers on noble metals. SERS results for GeOz NWs substrates using these two

analyte groups are presented in the following section.

4.5.1 SERS of rhodamine and nile blue

The initial evaluation of the SERS substrates was done using standard dye
molecules. Rhodamine 6G (R6G) and Nile Blue (NB). R6G (dye content 95%) and Nile
blue (dye content 80%) were obtained from Sigam-Adrich as solid powders. These
molecules were chosen due to their common usage by SERS researchers in the

characterization of substrates and non-toxic nature.

4.5.1.1 Detection of rhodamine

Rhodamine 6G is a common laser dye with high photostability and molecular
weight of 479.02 g/mole. It exhibits strong ﬂuorescence that tends to mask its spectra in
ordinary Raman spectroscopy[l22]. This can be seen in the ordinary Raman spectra of
R6G as shown in Figure 4-43. Here one can see the Raman peaks of R6G background at
1178 crn'l (C—H in-plane bending), 1304 cm"1 (C—O— C stretching), and 1360 cm'1
and 1513 cm“1 (C——C stretching of aromatic ring) on the top of strong ﬂorescence
background[123]. The ﬂuorescence is expected to be quenched in SERS spectra due to
presence of metal in the vicinity of analyte that provides a fast path for transfer of energy

between substrate and molecule[122].

121

    

5.
4.5—
.39
5 4~
33.5—
.§‘
(0
5
g 3-
c
E
(02.5”
0:
2-

 

 

1000 1100 1200 1300 1400 1500 1600 1700 1800 1900
Wavenumber/cm’1

Figure 4-43. Chemical structure and ordinary Raman spectra of R6G with
785 nm excitation.

For SERS studies, R6G was dissolved in DI water and the stock solution was
diluted to a concentration of 3 micromoles (uM). Care had to be taken to store the
prepared solution in plastic bottles as R6G has tendency to stick to glass. About 200 11L
diluted solution was applied to the substrate using plastic pipette but it was found to form
a drop on the substrate rather than wetting it. So the substrate was soaked in solution for
10-15 min and then allowed to dry at room temperature. Raman spectra were collected
from different parts of the substrate and very few sites yielded discemable spectra of
R66. Figure 444 shows Raman spectra of a 2 11M R6G in DI water from one of the

active sites. It can be seen that the R66 spectra rides on a slowly varying broad

122

02'

vi

‘F‘.

u-
1
.7.“
\.
.11.

14;

 

background that was observed after gold coating of nanowires. The dotted gray line
shows the Raman spectrum of a bulk 2 uM R6G solution, not placed on the SERS
substrate, and obtained under same conditions. The bulk solution spectrum does not yield
any discemable peaks and this conﬁrms the enhancement is due to gold-coated

nanowires. However, the enhancement is from isolated locations only.

co
8
3

- .. Bulk (x10)
8000) -—SERS

 

  

r
l

.ﬁr ‘ ' 1’
"‘ . l l, . -,- '
l w

4 r ‘1 ' ' 3.11.111... .n.,-...r..r
500 1000 1500 2000
Wavenumber/cm'1

Figure 4-44. Raman spectra of 211M R6G in DI water for l4nm gold

coated GeOz nanowires. Bulk solution spectrum of same concentration is

also shown, tacc=15 8. Bulk spectrum is X10 times the value to show

features.

To avoid the problem of substrate wetting, ethanol was used as solvent instead of

DI in the rest of this study. The use of ethanol as a solvent also helps in preventing the

dissolution of GeOz nanowires which would occur if water is used as the solvent. The

123

 

R6G was dissolved in ethanol (95%) and diluted successively to obtain a concentration of
3 uM. The Raman spectra of the prepared concentration in liquid form did not yield any
discemable spectrum. A small drop of the liquid (~250 11L) was applied to the substrate
using plastic pipette and allowed to dry under ambient conditions. Again the Raman
signal was observed from few isolated spots and Figure 445 shows spectra from one

such spot. The bulk solution (insert of Figure 4-45) yields only ethanol spectra but the
R6G peaks are clearly visible in the enhanced spectra from gold-coated GeOz nanowire

substrate.

------- Bulk
--SERS

C”
O
O
O

1‘ 5000-

.5
O
O
O

 

 

3
.
L—‘mAh—A A- L—A‘h‘.‘

500 1000 1 500 2000
Wavenumber/cm'1

Figure 4-45. Raman spectra of 3 uM R6G in ethanol for 21 nm gold

coated Ge02 nanowires. Also shown is the ordinary Raman spectra of 3
11M ethanolic R6G bulk solution, tacc=15 s.

124

4.5.1.2 Detection of R6G using Au nanoparticles decorated nanowires

To study and compare the effect of gold colloid versus gold ﬁlm on nanowires
substrates, theGeOz nanowires were decorated with gold colloidal. A drop of ~300 11L of

60 nm gold colloidal suspension (from Alfa Aesar) in water was added to the substrate
and allowed to dry. The 3 uM ethanolic solution of R6G was also drop coated to the
substrate. The enhanced Raman signal can be detected at isolated spots from the substrate

as shown in Figure 4-46 with a characteristic broad background. However, most areas of

the substrate just showed the spectra of GeOz.

1 8000 -
—— raw
~---—- baseline corrected

1

1 6000

I'

‘0 14000
1:

 

12000 -

10000 -

8000

6000 -

 

Raman lnstensity/Arbitr.u

4000-

”11.381.11.44... 11.1110. .

800 1000 1200 1400 1600 1800
Wavenumber(cm )

 

 

Figure 4-46. Raman spectra of 3 uM R6G in ethanol from Ge02 nanowire
substrate decorated with 60 nm gold colloidal. Also shown is the baseline

corrected spectra, tacc= 15 s.

The broad background was subtracted from the spectra using baseline correction

feature of the Raman system software as shown in Figure 4-46. This feature of the

125

software takes up to 20 user deﬁned points on x-axis that should have zero Raman
intensity and then ﬁts a curve through these points to determine background signal. This
background signal is then subtracted from the Raman spectra to yield background free
signal. The technique does require few iterations and ﬁne-tuning of parameters before

yielding an acceptable background free spectra.

4.5.1.3 Detection of nile blue

Nile blue A is a common stain used in biology and ﬂuorescence studies. It has
molecular weight of 732.85 g/mole. Figure 4-47 shows the reference Raman spectra of
nile blue (NB) recorded with 531 nm laser excitation after removal of ﬂuorescence
background[l24]. The NB molecule has 126 normal modes but most of the Raman
intensity is contained in seven modes at 590, 114lm 1351, 1429, 1492, 1544, and 1640
cm"[124]. The main peaks in the spectrum are the 590 and 1640 cm'1 modes
corresponding to ring breathing and ethylenic-stretching motions respectively. For SERS
studies, a 2.73 11M solution was prepared by dissolving measured amount of NB in 95%
ethanol and successively diluting the solution until desired concentration was achieved.

About 250 1.1L of this ethanolic NB solution was placed on the GeO; nanowire substrate

that had been sputter coated with 21 nm gold ﬁlm and allowed to dry. The Raman spectra
of the NB from the gold coated nanowire substrate is shown in Figure 4-48 with all the
major peaks of NB clearly visible. The insert gives the Raman spectra recorded for the
2.73 11M ethanolic NB from bulk solution under similar experimental conditions and all
that can be observed are the ethanol peaks with NB signal buried below the noise line. On

the other hand, the strong SERS enhancement form gold-coated nanowires makes all

126

main peaks of NB clearly visible while quenching the ﬂuorescence. Figure 4-49 shows
SERS spectra of NB from different arbitrarily selected points on the substrate and main
peaks are clearly visible. However, the enhancement was found to be location dependent

with some locations yielding low or no enhancement.

(3) (H5C2)2N+ o NH2

(b) Ring breathing
Ethylnic
stretch

 

 

Raman Intensity

 

l l l I I I I

400 800 1200 1600

Wavenumber/cm'l

Figure 4-47. (a) Nile blue molecular structure. SERS spectra of NB with
531 nm excitation[124].

127

Raman lnstensity/Arbitr. Units

—- raw spectrum
---- baseline corrected

 

 

 

 

(0
E10000»
D
‘5 8000
it
.1?
‘3
9 6000-
(I)
E
5 4000
E
(U
0:
2000- l
0 .uluhs ooiulia Ju.‘l‘.t 11‘
400 800 1000 1200 1400 1600
Wavenumber/cm‘1

Figure 4-48. SERS spectra of 21 nm Au ﬁlm coated GeOz NWs before

and after application of 2 uL of 10 11M NB, taco: 15 5. Baseline corrected
spectra of NB is also shown.

 

     
  

1000
1500 3 sample point

Wavenumber/cm'1 2000 1

Figure 4-49. SERS spectra from 2 11L of 10 11M ethanolic NB along a line
from GeOz NWs, tacc= 5 s.

128

4.5.1.4 Detection of nile blue using silver coated nanowires

Silver is another coinage metal that shows strong Raman activity due to excitation
of surface plasmons in the visible region of spectrum. Although gold is a better choice for
SERS due to its chemical inertness, silver coating of NWs was also studied since it

exhibits strong SERS due to its capability to support surface plasmons in optical
spectrum. GeOz nanowires synthesized on Si were coated with a 15 nm silver ﬁlm using

e-beam evaporation and continuous substrate rotation. The coated substrates were

characterized by the same Raman system as shown in Figure 4-50.

3110

----- Substrate
4.5 - — NB applied

Raman Intensity/Arbitr. Units
_. N
'01 N or on

 

.0
01—h

 

 

500 1000 1500 2000
Wavenumber/cm'1

Figure 4-50. Raman spectrum of spectra of 15 nm Ag coated GeO;
nanowire substrate before and after applying 2.73 IIM Nile blue in ethanol,

taco: 5 S-

129

Similar to the case of gold coating, the silver coating has also introduced a strong
and broad background that peaks near 800 cm'l. Next a drop of 2.73 11M NB in ethanol is
applied to the substrate and allowed to dry at room temperature. Figure 4-50 gives the
Raman spectra of substrate after drying of NB. As can be seen, the main peak of NB at
592 cm'l is clearly identiﬁable whereas the remaining peaks show up as weak peaks on
the top of broad background.

The initial evaluation of substrates using R6G and NB as probe molecules
provided evidence of strong enhancement, sufﬁcient to easily detect 1.1M concentrations
of the two species. On the other hand, no discemable Raman spectra could be obtained
from bulk solutions at these concentrations. This clearly conﬁrms the SERS from
nanowires; however only at a few isolated locations. Although there is a high density of
the NWs on the substrate only a few “hot spots”, where signiﬁcant SERS could be
measured, were found. It was thus speculated that the cause lies in inadequate adsorption
of the analytes on the NWs substrates. It is known that the SERS effect diminishes within
a few nm away from the active metal surface[39]. To test this hypothesis, the thiol based
probe molecules were investigated which are known to form self-assembled monolayers

on gold.

4.5.2 SERS of thiols

Thiols are chemical compound that contain a ﬁmctional group composed of a
sulfur atom and a hydrogen atom. The aromatic thiols are special class of thiols that
contain an aromatic hydrocarbon (benzene ring) and are commonly employed in SERS

studies due to their ability to form self-assembled monolayers on noble metals. This

130

allows for relatively simple evaluation of the SERS performance of substrates based on
knowledge of the number of molecules contributing to the SERS signal. Two aromatic
thiols 4-methylbenzenthiol (purity 98%) and 1,2-benzenedithiol (purity 96%) and one
non-thiol analyte trans-1,2-bis(4-pyridyl)ethylene (purity 98%) were purchased from
Sigma-Aldrich and used without further puriﬁcation. Figure 4-51 shows the molecular
structure of analyte molecules used in this study. The diluted solutions with various

concentrations were prepared in methanol (99.8% HPLC grade).

 

CH3
SH
SH
SH
4-methylbenzenethiol 1,2-bis(4—pyridyl)ethylene 1,2-benzenedithiol

Figure 4-51. Chemical structures Raman ethylene analytes.

4.5.2.1 SERS dependence on Au thickness
The NWs were coated with a SERS active metal such as gold to provide a surface
plasmon active surface. The thickness of the metal layer is expected to affect the SERS
performance of the substrate. The effect of metal thickness was studied by coating
different parts of the same NWs substrate with different gold thickness. The 4-MBT was
used as target molecule to study the effect of gold thickness since these aromatic thiols

are known to form self-assembled monolayers (SAMs) on gold, silver, and copper“:25 1.

131

The SAM was prepared on the substrate by soaking it in 1 mM methanolic 4-MBT
solution for 14 hrs. The substrate was then rinse in methanol and dried by blowing air.
Figure 4-52 shows SERS spectra from substrate for 4 different thickness of Au ﬁlm. It is
evident that strong enhancement with moderate background continuum is achieved for

~10-20 nm thick Au coatings.

 

4
35x10, . . ﬁ ﬁ . .
6 nm Au
3- —12nm Au -
—23 nm Au
—35 nm Au
2.5— -

 

Raman lntensity/Arbitr. Units
N, _

 

 

J~ L1 .‘1 - st.‘§'wv' "'I 1 " "MW' “‘ro‘ in ‘. )~
an. ,.r 1...! °.r‘- 4...... '- "4...: -.
an.

 

I".
”mmiw.w\k‘~’f \
1 l

400 600 800 10100 1200 1400 1600
Wavenumberlcm'1

 

Figure 4-52. Effect of Au thickness on SERS spectra.

To assist in a generalized conclusion of the effect of Au coating thickness on
SERS enhancement, 8 different analyte was also investigated. In this case, the substrate
was soaked in 1 mM methanolic 1,2-BDT solution for 20 min to form a SAM. The
substrate had been coated with 3 different Au ﬁlm thickness in three different areas of the

same nanowire coated substrate. This ensured that all experimental variables were the

132

same for the synthesis of NWs. The substrate was rinsed with methanol and dried before
acquiring the spectra. Figure 4-53 shows the SERS spectra of a SAM of 1,2-BDT from
three different Au coated NWs. Again good enhancement above a broad background

continuum can be found for Au ﬁlms having thickness of 10-20 nm.

 

6000 . . . . .
—12 nm Au
—23 nm Au 1

5000“ —35 nm Au

Its

i

l
l

.h
C
O
O

-<

!

Raman Intensity/Arbitr. Un
8 e
s o
'\

 

.5
o
3

 

 

400 600 800 1000 12100 1400 .1800
Wavenumber/cm'1

Figure 4-53. SERS from 1mM BDT for 20 min soak.

It is also important to study the affect of the age of the Au coating on SERS
perfonnance of the substrate to assess deterioration over time. Storage of samples in open
environment can degrade their SERS activity due to contaminations on the Au coating.
The nanowire substrate was coated with gold and stored for 10 days in a petri dish at
ambient conditions. Subsequently the SAM of 4-MBT was synthesized as discussed

earlier. Figure 4-54 shows the effect of Au coating aging on SERS spectra for a

133

monolayer of 4—MBT. It can be seen that the 10 days old Au coating exhibit a lower
signal strength as compared to corresponding fresh coating. This can be attributed to
contamination, which reduces the monolayer coverage of analyte. It had been reported
that the aged gold ﬁlms do not produce as strong enhancement as fresh ﬁlms due to

contamination of ﬁlms during storage.

3x10
—12 nm Au
~--------- 12 nm Au 10 days
2'5' —23 nm Au
~----- 23 nm Au 10 days
2* i —35 nm Au

---------- 35 nm Au 10 days

Raman Intensity/Arbitr. Units
- or

.0
or

 

 

 

400 600 800 1000 12100 1400 16100
Wavenumber/cm'1

Figure 4-54. Effect of gold coating age on SERS from 1mM MBT.

Another important aspect in the SERS study of any substrate is the affect of
application of the analyte to the SERS substrate. A commonly employed technique is the
application of a solution containing the analyte to the substrate. The drying of the solvent
creates strong surface tension forces on the nanostructures such as NWs. The resulting

stresses can change the morphology of the substrate as shown in Figure 4-55. Here the

134

 

substrate has been soaked in 1 mM methanolic MBT for 12 hr, then rinsed in methanol

and blown dry. The clustering of NWs due to solvent evaporation is clearly noticeable.

 

 

Figure 4-55. SEM image of GeO; NW substrate after 12 hr soak in 1mM
MBT.
The qualitative SERS study of gold coated GeOz nanowire substrate presented in

this section shows that the thiols can be used to carry out detailed quantitative assessment

of these substrates. This includes the study of reproducibility of the SERS and calculation

ofSERS enhancement factor as discussed in the following section.

4.5.2.2 Detection of 4-metlrlybenzenethiol
The 4-MBT belongs to para-distributed benzene class of chemicals with D211

symmetry. Figure 4-56 shows the chemical structure and ordinary Raman (OR) spectrum

135

of 4-MBT. It can form SAM on gold by forming S-Au bond after loss of hydrogen. The
Man vibrational band ~910 cm"1 is assigned to the S-H bond of the molecule and is
clearly seen in the OR spectra. However, the same band does not show up in SERS
spectra because sulfur atom bonds to the metal surface by giving up the hydrogen
atom[l26]. Thus the study of the Raman peak corresponding to SH vibrations can be

used to determine the type of adsorption of 4-MBT on metal surface.

0 hydrogen
4 . carbon
4 2‘ 10 0 sulfur
CH3
3.5 ~
3 - SCC Ring breathing 0
2.5 * SH

 

1.5

I

5CC

Raman lntensity/Arbitr. Units
N

I

0.5

 

 

 

 

 

400 600 800 1000 1200 1400 1600
Wavenumber/cm'1

Figure 4-56. Chemical structure and ordinary Raman spectrum of 4-MBT.

136

The gold-coated GeO; nanowire substrate was soaked in 100 IIM methanolic 4-

MBT solution for 10 hrs. It has been reported that 1mM methanolic 4-MBT forms SAM
on gold ﬁlm after 12 hr of incubation[127]. The concentration used in this study is an
order of magnitude smaller than the one used for SAM and is expected to yield
sub-monolayer coverage. For conservative estimates of the Raman enhancement factor, a
monolayer coverage assumption is used in our calculations. After the 4—MBT soak, the
substrate was thoroughly rinsed with methanol to remove multilayers and/or crystals that
might have formed[127]. The substrate was dried in the stream of nitrogen before
acquiring the SERS spectra shown in Figure 4-57(a). SERS spectra does not show the S-
H vibrational Raman band at ~916 cm’I conﬁrming the formation of S-Metal bond[128]).
An interesting effect is the suppression of continuous background of raw substrate
following the formation of the 4-MBT monolayer. Figure 4-57 (b) shows Raman spectra
from ﬁve consecutive 250 um spaced points on the substrate and the characteristic peaks
of 4-MBT are easily identiﬁable. It may be mentioned that no Raman signal was detected
from a Si(100) substrate with 21 nm smooth Au ﬁlm that was soaked in 4-MBT
alongside nanowires substrate, conﬁrming the enhancement is indeed from the
nanowires. Similarly the bulk solution of 100 11M methanolic 4-MBT only yielded

spectra for the methanol solvent.

137

A
m
v

 

  
 
 
   

 
 

x10“
w 5?—
g l --------- before
n 4L —after
5 1
.9 ,
1’1 3r
.é‘
2 3
a) 2'
1c” 1
c .
N 1r
E .
(0 ; .__.
m 1 -—...

 

400 600 800 1000 1200 1400 1600 1800

Wavenumber/cm'1

A
U"
v

15000 4

1 0000 ~

5000 -

 

Raman Intensity/Arbitr. Units

4
2000
3 ./ 1500
. 2 1000
Sample pomt 1 50° Wavenumber/cm‘1

Figure 4-57. (3) SERS spectra of a SAM of 4-MBT. Also shown is the
Raman spectra of substrate before formation of SAM, tacc=15 s for both
spectra. (b) SERS spectra of a SAM of 4-MBT from 5 adjacent points 250
um apart, tax: 5 s.

The homogeneity of the substrate has been a challenging issue for the scientiﬁc

community in SERS research. To verify the repeatability of SERS measurement from

138

substrate, SERS spectra were acquired from 100 points covering an area of
1.2 mm x 1.4 mm. The 1068 cm'1 Raman band of these spectra is then plotted in Figure
4-58 to give a spatial map of enhancement from the substrate. It can be seen that the
substrate provides unambiguous and uniform SERS spectra for a monolayer of analyte

over the tested area with a standard deviation of approximately 20% from the mean.

mean=13.2E3
1110‘1 . std dev=2.7E3

 

Raman Intensity/Arbitr. Units

 

       

y-position (x140um) o 0 x— position(x125um)

Figure 4-58. 4-MBT 1068 cm" Raman peak intensity distribution for
randomly chosen area on substrate, tax: 5 s.

The experimental Raman enhancement factor (EF) for our substrate was

calculated using the standard equation[129]

EF=ISERS NREF (47)
[REF NSERS

139

where I SERS is the Rarnan intensity of some speciﬁc band from the analyte
adsorbed on SERS active substrate and [ref is the Raman intensity of the same band
from the bulk analyte. NSERS and N ref are the number of molecules that yield I SERS and
I ref respectively. Thus equation (4.7) gives the Raman enhancement per molecule for a

speciﬁc vibrational band of analyte. The speciﬁc band is usually the one that gives the
strongest Raman scattering and its intensity is directly read from SERS spectra. An
obvious choice therefore in our case is the 1068 cm'l Raman band and we utilize the
average intensity from 100 sample points in equation (4.7). The number of 4-MBT
molecules contributing to SERS can be estimated from the surface density of SAM on
Au. Figure 4-59 shows the possible structures of 4-MBT SAM on Au(100) surface[l30].
The corresponding surface densities of 4-MBT molecules are 3.5x1014 and 4.6x1014

2

molecules /cm for 4x13 and 213x13 respectively. Seo et al. estimated surface

concentration of 4-MBT as 4.5x1014 molecules /cm2 on Au(111)[130].

213xt3

4XV3

 

 

an
><
H
i
U

 

 

 

11
>4
9
>-<
>-<
C4

 

 

   

 

Figure 4-59. Possible structures of 4-MBT SAM on Au(111)[130]. Open
circles are gold atoms and solid circles represent sulfur head group of 4-
MBT.

140

Taking 4.5 x 1014 molecules/cm2 for a monolayer of 4-MBT on gold, we ﬁnd
approximately 5.87 x 10'‘4 moles (3.53 x 1010 molecules) of 4-MBT are exited in the

laser spot (100 pm diameter). The determination of Nref is more challenging and
different methodologies have been reported to estimate it. Ideally the reference I ref and
N ref may be obtained by recording the Raman signal from a ﬂat metal coated part of the

same substrate; however it is very challenging to get ordinary Raman spectra for a
monolayer of analyte. Even such an arrangement does not yield true ordinary spectra due
to the presence of metal in the vicinity of analytem]. The commonly employed techniques
to circumvent this problem are to use solid bulk form or a higher concentration solution
of the analyte to get the approximate values of the reference parameters. Both these
techniques need an estimate of the laser interaction volume. This can be calculated from
laser depth of ﬁeld considerations. We did not detect any ordinary Raman spectra from
~20 mM 4-MBT methanolic bulk solution. Therefore, we used a solid form of 4-MBT to

get values of [ref and NWf. Figure 4-60 shows the ordinary Raman spectrum of 4-

MBT. The number of molecules excited in the solid phase was determined by assuming
the excitation volume to be a cylinder of diameter equal to laser spot size and length

equal to depth of ﬁeld. This assumption underestimates the Nref since the actual

interaction volume has the shape of a two back-to-back truncated cones with waist
diameter equal to the focused laser beam spot size. It is expected to yield a more
conservative estimate of EF. Figure 4-61(a) shows the correction factors for the
calculated EF if the actual interaction volume of the laser beam is taken into account. A

good estimate of depth of ﬁeld is important for an accurate assessment of EF. Many

141

factors inﬂuence the depth of ﬁeld including laser wavelength and power, numerical
aperture and focal length of lens, and type of sample (opaque, transparent or turbid with
respect to laser wavelength). However, only sample type is the main variable in present
case since all other parameters were kept same while measuring SERS signal from
nanowires and bulk sample. The depth of ﬁeld was determined by moving the laser focus
point from under to over-focus on bulk crystal of 4-MBT‘ and noting the 1/2 power
points. The measurement of depth of ﬁeld with 4-MBT bulk crystal ensures a more
accurate estimate of number of molecules contributing to Raman signal from bulk
sample. Figure 4-61(b) shows the intensity variation for two Raman peaks of a 4-MBT
crystal as the laser focus was moved from under focus to over focus on the sample. The
depth ﬁeld of view (DFOV) was estimated to be 902 pm for 790 cm'I band and 842 pm
for 1090 cm‘1 peak. A more conservative estimate of 800 pm for DFOV was used in EF
calculations. Taking the monolayer thickness of 4-MBT as 0.5 nm[127], we get a quick

estimate of Nref /NSERS ratio. The estimated enhancement factor was found to be
2.64 x 106. We also calculated I ref and Nref by using polystyrene as a reference whose

Raman cross—section can be assumed to be equal to that of 4-MBT[127]. Ordinary Raman
spectrum of polystyrene is shown in Figure 4-60. Following a similar line of
calculations, the EF was estimated to be 1.43 x 106. This EF is in good agreement with
that calculated using bulk solid form of 4-MBT and shows that polystyrene can also be
used as reference for 4-MBT enhancement factor calculations. It may be noted that we
have used a ﬂat area approximation while determining NSERS in our EF calculations. The
actual or effective area of the nanowires is somewhat increased (~ 57% for half coated

NW) due to their cylindrical nature. However, the effect is not expected to be signiﬁcant

142

 

due to the line of sight deposition of Au that only coats the part of the nanowires facing

the evaporation source.

Raman lntensity/Arbitr. Units

14000

1 2000

1 0000

8000

6000

4000

2000

I

I

I

I

I

 

400 600 800

....... .. polystyrene
— 4-MBT

 

 

£1

1 ink.
wﬁ‘W‘L‘LI-o- » Wu-

1000 1200 1400 1600 1800
Wavenumber/cm"1

Figure 4-60. Ordinary Raman spectra of 4-MBT and polystyrene,

143

A
m
V
A GI 0)

EF correction multiplier
00

 

 

2 r

1 .

0 2 4 8 8 1'0
DFOV/spot diameter

  
  

A

C'

V
-—L

—790 cm'1 peak
°”"""1090 cm'1 peak

.0 .o .o
A O) CO

Normalized intensity

9
N

 

1

41100 -500 0 500 1000
Offset from focus (micron)

Figure 4-61. (a) EF correction for cylindrical excitation volume
assumption. (b) DFOV estimation using two Raman bands of solid 4-MBT
Crystals.

144

It is pertinent to compare the EF from the nanowire substrate to that of other
commonly employed SERS substrates. Table 4-1 lists EF reported in literature for some
commonly employed SERS substrates. It can be seen that the SERS EF for nanowires is
reasonably placed in the range reported by different researchers. Considering the average

enhancement is of the order of 1 x 106, the EF for nanowires can be categorized above

average.

Table 4-1. SERS EF for common substrates.

 

 

 

 

 

 

 

 

\ Substrate EF Ref. Remarks

Du island ﬁlm 6 x 103 [131] 1064 nm excitation
Au ﬁlm over A1203 2 x 104 [131] 1064 nm excitation
nanoparticles

\joughened Au ﬁlm (peak) 4.3 x 106 [127] 785 nm excitation
Ag ﬁlm over nanospheres 1 x 107 752.3 nm excitation
Ag nanorods (max) 1 x 108 [36] 785 nm excitation
W chip from 3D 1 x 106 Manufacturer Binding molecules.
Tecl’mologies Ltd. 1 x 104 data Most other

molecules.

Mm Real Time Analyzer 1 x 106 Manufacturer

L data

 

 

 

 

 

 

Next we analyze the SERS performance of a commercial SERS chip -— Klarite

from 313 Technologies Ltd. This chip contains lithographically etched pattern of regular

145

pits (inverted pyramids) in silicon wafer and the pattering is optimized for 785 nm
excitation. The chip is coated with nearly 300 nm gold ﬁlm to provide plasmon active
surface. The Klarite chip was soaked in 1 mM methanolic MBT for 14 hr to assemble
3AM of analyte. It was rinsed in methanol and dried before acquiring the spectra. Figure

4-62 shows spectra for SAM of 4-MBT on Klarite chip for randomly chosen points on

sub strate.

O) (D
O O
O O
O O
/ r

4000]

2000 \

 

Raman |ntensrty/Arbitr. Units

 

Wavenumber/cm“1

Sample point

Figure 4-62. SERS spectra of a SAM of 4-MBT on Klarite chip from 5

different points, taco= 5 s.

The spatial variation of 1068 cm" peak of 4-MBT over 6.25 mm2 area is shown in
Figure 4-63 and shows good reproducibility of the signal with a standard deviation of
N 6 %- The average EF for Klarite chip was estimated as ~1.5 x 106 and agrees well with
the Value suggested by manufacturer. The analysis of Klarite chip serves a twofold

146

purpose; (1) it validates the mechanism of EF estimation, and (2) it shows that the

nanowires yield a better EF than the commercially available product.

mean=7727
std dev=454

 

 

 

10000.
.«e
5 8000.
E
g 6000.
.é
2
2 4000.
s
C
E 2000\ 7 1
m
m _ f,
0>\ ,I' \
5 {\\\ \ \ if
- : " l 1":“"~—;-’ 1' ’
4 \\ :1 -1 Lb/ 5
A " « 77 7;“ -4 V——;—/‘ 4
3 \ /,-:R'
2 \ 1 4’7 3
y‘yk 2
y-position (xzsotim) 1 x- position(x250um)

Figure 4-63. 4-MBT 1068 cm'l Raman peak intensity distribution for
randomly chosen area on Klarite chip, tacc= 5 s.

Table 4-2 shows the Raman peak assignment for 4—MBT (wavenumbers in cm'l).
The Raman peak assignment is based on the knowledge of Raman bands for p-distributed

benzme derivatives. The assignment is further reﬁned by using the spectral correlation

 

With the ordinary Raman (OR) spectra of 1 chlcrc 4 “ ,‘L This was based on
the Well—known similarity in the ordinary Raman spectra of mecaptans and the

corresponding chlorides (1 chlcrc 4 ‘L L in this case)[128, 132]. It is evident

 

ﬁ'om Table 4-2 that the SERS Raman peaks of 4-MBT measured here are in good
agreement with those reported earlier.

147

Table 4-2. Raman band assignment for 4-MBT.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

OR p- OR OR SERS 4- SERS 4-MBT Assignment
distributed 4-MBT p-chlorotoluene MBT[136] (this work)
benzene Ref (this work) RefI134]/Ref[l35] (514 nm)
[ 1 32], [133]
”77240-340 309 m -/303 322. 822 w yCH; (10b)
”740-400 368 vw 360/377 VCS,’ Al (7a)
”7380-420 381 m 386/405 386 387.5i0.5 m rcc; A2 (16a)
481-680 636 s 547/636 623 619i0.6 s see; A] (12)
720-840 790 vs 765/797 794 786. 6i1.4 w 6CC; A1 (6a)
1004-1022 1011 w 999/1016 1004.4i1m 6CH;tCC 18a
1 050-1100 1069 w 1078 1067.7i0.48 s Ring; 1
1050-1100 1091 vs 1056/1090 Ring; A1(1)
1100-1128 1112 w 6CH,-18b
1085—1129
1156-1190 1178s 1072/1176 1181 11705110719 6CH;A1 (9a)
1142-1190
1100-1300 1204 s 1194/1208 1209 1199. 5:0. 7 w 6CH;A1 (13)
1260-1310 1305 w -11300 1378 1369.8i1m SCH; B2 (3)
1460- 1 530 1495 vw 1483 1480.8:07 m vCC; 19a
1552-1 605 1572 m -/1574 1579 1579. 7111 sh ICC; 192 (8b)
\
1570-1 628 1606 m 1595/1596 1593 1597i0.6 s vCC; A1 (8a)
\

 

 

 

 

 

 

 

 

szvery weak; w: weak; m: medium; 8: strong; vs: very strong; sh: shoulder.

V: in‘Plane stretching; 8: in-plane bending; y: out-of-plane bending.

148

4.5.2.3 Detection of bis(4-pyridyl)ethylene
The second analyte used was trans-1,2-bis(4-pyridyl)ethylene (BPE). BPE

belongs to N-heterocyclic compounds that are of biological, pharmaceutical and
industrial importance[137]. BPE is centrosymmetric and photostable with well resolved
spectral lines in SERS spectra. BPE has 66 vibrational mode but only 23 modes of Ag
symmetry ( in C2h symmetry) and 10 of Bg symmetry are Raman allowed[138]. The Ag
modes are in-plane while the Bg modes are out—of-plane. Many of these modes are very

weak (like 5 Ag modes of CH stretch near 3000 cm") or below 250 cm" and are difficult

to measure accurately.

ethylenic vCC

vCH+6Ring
Ring vCC ‘

 

A
01
l

 

 

Ring breathing

SCH
vRing+5CH
Ring def.

800 800 1000 1200 1400 1600 1800
Wavenumber/cm'1

Raman Intensity/Arbitr. Units

.0
01
l

 

 

 

 

 

 

Figure 4-64. Chemical structure and ordinary Raman spectrum of BPE.

149

For SERS studies, a 0.83 11M solution of BPE was prepared in methanol and 2 11L
solution was applied to the nanowire substrate. The solution immediately spread on the
substrate to a spot of nearly 5 mm diameter, and was allowed to dry before acquiring the
Raman spectra shown in Figure 4-65. The BPE peaks can be clearly identiﬁed on the top

of broad background that was found after Au coating of the substrate. The BPE spectra
can be easily identiﬁed from different points on substrate with just few seconds of signal

accumulation.

4000 ~

3000 ~

2000 -

 

1000*

Raman Intensity/Arbitr. Units

 

 

O 1 1 1 1 1 1 1 1
200 400 600 800 1 000 1 200 1 400 1 600 1 800
Wavenumber/cm'1

Figure 4-65. SERS spectra of 0.83 11M BPE from Au coated GeOz NWs,
taco: 15 3.

Figure 4-66 shows SERS spectra of BPE from ﬁve different points (~125 11m
SPaCing between points) on the substrate and the main peaks of molecule can be clearly
Seen in all the spectra. The SERS uniformity of our substrate was veriﬁed by plotting
1190 Cm" Raman of BPE for 100 sample points recorded over a random substrate area as
Shown in Figure 4-67. The plot conﬁrms that the substrate provides a relatively uniform

enhancement with a standard deviation of less than 23% from the mean.

150

g 11806m'1

5 3000

E 2000

.é‘

tn

5 1000

‘5‘

E o 7

E o « 5

n: /” 4

1000 ‘ / 3

Wavenumber/cm' 2000 1 Sample point

Figure 4-66. SERS spectra of 0.83 11M BPE from different points on
substrate, tax: 5 s.

mean=1391
std dev=318

§
0

E

S

    
 

Raman Intensity/Arbitr. Units
at
8

500 év’;
1° \\\-‘ r: ,
\ . ‘_ /, . .
' a
5\\ X 6
4

\g/
y-position (x7011m) o 0 2

 

x- position(x63pm)

Figure 4-67. BPE 1190 cm" Raman peak intensity distribution for
randomly chosen area on substrate, tacc= 5 s.

We also calculated the experimental enhancement factor for BPE using the
“90 Cm" Raman band. This band is the strongest in the SERS spectra and also has the
added beneﬁt of being relatively insensitive to molecular orientation on a metal

s|11'face[36]. The molar concentration and applied volume of BPE means that nearly 10‘2

151

molecules are added to the substrate. Assuming uniform spatial adsorption of these
molecules on the substrate, it is estimated that nearly 6.65 x 10'16 moles of BPE are
contributing to SERS signal for the laser spot size of 100 um diameter. The 0.83 11M
solution of BPE could not be used as reference since it did not yield a detectable Raman
Signal. We therefore prepared ~0.75 M solution of BPE in methanol for reference
measurements. The intensity of 1190 cm'1 Raman line from this reference solution was

used as [ref in equation (4.7) to calculate Raman EF. The number of molecules excited
in the bulk solution (Nref) were estimated to be 1.78 x1015 from depth of ﬁeld

considerations. This translates into an average enhancement factor of ~ 1.70 x 106. The
Raman peak assignments for ordinary Raman (OR) and SER spectra of BPE molecules

are given in Table 4-3 (wavenumbers in cm!) and agree well with the reported literature.

Table 4-3. Raman peak assignment for BPE.

 

 

 

 

 

 

 

 

 

 

 

 

OR[1 38] OR (this SERS [138] SERS (this Assignment [137, 138]
work) (532 nm) work

640 662 652 649:2.8 w Ag 17; ring deformation

995 989 1008 1012il.8 s Ag 13; Ring breathing of pyridine

1194 1189 1200 1188117 8 Ag 9; ethylenic C=C stretch

1230 1229 1244 A8 8; CH bending on pyridine ring
\

1348 1348 1338 1336i2.6 w Ag 6; CH wag
\

1418 1409 1421 Ag 5; ring str. + CH bend
\

1491 1492 1493 Ag 4: ring stretch
\

1550 1552 1544 A8 3; ring stretch

1600 1603 1604 1615i3 m Ag 2: pyridine ring C=C stretch

1641 1650 1640 1640i3.1 in As 1: ethylenic stretch

 

 

 

 

 

 

 

 

 

152

 

4.5.2.4 Detection of benzenedithiol
To conﬁrm the enhancement for other molecules, we also investigated 1,2-BDT
as a probe analyte. The 1,2-BDT belongs to Cz‘, symmetry point group if the hydrogen
atoms are ignored and has 36 Raman active with the A1 modes polarized in the
Raman[139]. Figure 4-68 shows the ordinary Raman spectra of 1,2-BDT in liquid state

with signiﬁcant vibrational modes annotated.

6CH 0 hydrogen
3.5~ . 0 carbon
0 sulfur
SH
2.5~ ' 3“

 

SCC
vCS i ’ Ring breathing
1 F i vCC

SSH 1
1 ‘ SCH
0.51 Vcc

1.5

1

Raman lntensity/Arbitr. Units
N

 

 

400 600 800 1000 1200 1400 1600
Wavenumber/cm'1

Figure 4-68. Ordinary Raman spectra of 1,2-BDT in liquid state. Also
Show is the chemical structure of molecule.

The 1,2-BDT molecule is known to form self assembled monolayer (SAM) on

200
11111 thick Au ﬁlm for concentration > 1 ><10'3 M after more than 20 min of

Soak-
111g]: 1 26]. We immersed our SERS-active nanowires substrates in 20 11M solution of
122~BD

in methanol for 20 min. This concentration is less than the one reported for

153

 

monolayer coverage and is expected to yield a more conservative estimate of the

enhancement factor. The sample was then rinsed with methanol and dried under dry

nitrogen stream to ensure removal of excess analyte. Figure 4-69 shows the SERS
spectra from Au coated GeO; NWs, which clearly shows the 1,2-BDT peaks. The broad
continuum is also somewhat suppressed following the formation of the BDT SAM on the
Au coated nanowires. The SERS spectra of 1,2-BDT helps to identify the mechanism of
adsorption molecule on metal surface. Comparing to the ordinary Raman spectra, the
Raman band at ~910 cm'l corresponding to SH bending is absent in SERS spectra and

confirms the molecule absorbs on the surface by forming sulfur to metal bond aﬁer losing

hydro gen[128].

I

14000 — before

--------- after 1,2-BDT

I

12000

I

10000

a)
O
O
O

T

 

6000

 

 

4000 -

Raman lntensity/Arbitr. Units

   

2000 -

 

 

800 400 600 800 1000 1200 1400 1600
Wavenumber/cm'1

Figure 4-69. SERS spectra of self-assembled monolayer of 1,2-BDT from Au
Coat
ed Geog nanowires, taco: 15 s.

154

Table 4-4 shows the Raman band assignment (wavenumbers in cm") for different
vibrational modes of 1,2-BDT. The Raman peak positions in SERS spectra are somewhat
shifted compared to ordinary Raman spectra of 1,2-BDT. This is not unexpected given

that a shift of 10-20 cm'1 has been reported earlier[139].

Table 4-4. Raman band assignment for 1,2-BDT.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[’Ordinary Ordinary SERS [128] SERS [139] SERS (this work) Assignment[l39]
Raman[128] Raman[139] (514.5 nm) (1064 nm) (785 nm)
471 s 474(5) p 466 s 468 (3) 47211.16 vCS; A1 (7a)
658 s 658(5),p 647 m 655:1.81 m 5CC; A1 (6a)
698 vw 700 w 696 w 70312.64 vw rCC; A2 (4)

733 w 734w p 725 w 72812.54 vw 6CC; 32 (6b)
1010 sh 1010 sh 1008 sh 8CC; 32 (12)

1038 p 1034 vs 1035 1028106 s son; A] (18b)
1114 p 1087 vs 1089 107811.67 s Ring; A1 (1)

1160 dp 1161 m 1162w 115512.48m SCH; B2(18a)

1273 p 1263 s 1264 126611.89 s vCC;A1 (14)
SCH; 132 (3)

1452 w p 1433 m 1435 143611.69 m vCC; A1 (19b)

1560 p 1543 w 1544 154711.34 m vCC; A1 (8a)

1572 dp 1563 m 1565 157310.75 m vCC; 132 (8b)

 

Viv - .
' Vel‘)’ weak; w: weak; m: medium; s: strong; vs: very strong; sh: shoulder.

V; '
"1‘13 1 ane stretching; 5: in-plane bending; y: out-of-plane bending.

 

155

 

 

 

Raman Intensity/Arbitr. Units

    

1 500
1 000

I _ t 1 500
samp e pom Wavenumber/cm'1

Figure 4—70. SERS spectra from 1mM BDT along line on substrate.

Figure 4—70 shows successive spectra from the substrate collected from points

Spaced 250 um apart and show good reproducibility of measurement. Figure 4-71 shows

the plot of the 1028 cm1 peak of 1,2-BDT for 100 points scanned in a rectangular area of

1 -25 mm x 1.4 mm. The data show the substrate provides very uniform SERS
enhancement with a standard deviation of approximately 18%.

The SERS EF for 1,2-BDT was again calculated using equation (4.7) and this

time the 1028 cm“l Raman band of 1,2-BDT was utilized. The calculation of I SERS in

CQUatiOn (4.7) requires a knowledge of molecule adsorption of substrate. The 1,2-BDT is
known to self assemble on Au(111) by forming two sulfur-gold bonds as shown in
Fig" re 4-72. Assuming monolayer coverage of substrate with 1,2-BDT and each
"1013c ul e binding to 2 Au atoms on the surface, we get 6.4 x 1014 molecules/cm2 on the

sh),
Strate-

156

 

mean=2426
std dev=439

3500 \
3000 4
2500 4
2000 4
1500 V

1000 ‘1 .— 1. / . ' .

 

Raman Intensity/Arbitr. Units

/
U

 

1. 4
\ l 5‘ \.,_ 5' D 10
5 {/2 6

'/ / 4

y-position(x1400m) o n 2 x-position(x125um)

 

Figure 4-71. Area map of 1028 cm“ Raman peak of 1,2-BDT, tacc=5 s.

hydrogen
0 carbon 0 801d

sulfur

51°

 

 

Figure 4-72. Molecular structure of 1,2-BDT and formation of SAM on
Au(111) face[126].

157

Considering the laser spot size of 100 um diameter, nearly 8.35 x 10"4 moles
(5.03 x 1010 molecules) of 1,2-BDT contribute to the SERS signal. The ordinary Raman
spectra of the source 1,2-BDT was used as reference to calculate Raman EF. The laser
excitation volume was estimated to be ~6.28 nL from depth of ﬁeld considerations. The
average enhancement factor was found to be ~ 1.27 x 105. The low enhancement factor
relative to other analytes is probably due to sub-monolayer coverage of 1,2-BDT on the
substrate. Assuming 1,2-BDT surface coverage depends linearly on bulk solution
concentration, we get 0.02 monolayer of 1,2-BDT from 20 11M solution. The
enhancement factor using this sub-monolayer coverage will be almost 50 times the one
calculated for the complete monolayer assumption.

It is useful to conjecture on the mechanism involving the interaction between laser
and analyte in the present system of Au coating oxide NWs. It is hypothesized that the
high dielectric nature of oxide NWs helps to conﬁne and guide the light to a much larger
area on the substrate than that directly illuminated by the laser beam. This is supported by
the Observation that the laser spot is hardly noticeable in CCD camera when focused on
Si Wafer. However, the illuminated size is much larger (few mm diameter) when NWs

SUbStI‘ate is targeted suggesting light waveguiding and coupling among NW8. This is
depicted in Figure 4-73 where CCD image of laser spot on bare Si wafer and GeO; NWs
SUbStI‘ate is shown. The NWs contain large number of gold particles on top surface,
which are also expected to immobilize analyte molecules. The light coupling between
NW8 e>~Cczites surface plasmons in Au particles that in turn excite Raman modes of
analyte - The analyte Raman radiation is again ampliﬁed by the Au nanoparticles, coupled

back
to t he W5 and ﬁnally travels to the receiver in the reverse direction. This plausibly

158

 

explains the observed laser spot broadening and good reproducibility of SERS from GeO;

NWs substrate.

 

Figure 4-73. Optical image of laser illumination of Si wafer (leﬁ) and
G602 NWs substrate (right).

4.6 Conclusion
This chapter mainly presented the details of synthesis and results of Raman
enhancement from gold-coated G602 nanowires. The EDS, TEM and Raman studies of
nanOWires conﬁrmed the trigonal phase of crystalline GeO; nanowires. The gold coating
introduced a broad background in the Raman spectrum of nanowires. By changing gold
dePOSition process and studying the impact of sample heating and drying in desiccator,
the l'11'1purity and water vapor were eliminated as possible sources for broad background
in the Spectra. The silver coated nanowires also exhibit a background continuum similar
to the one observed for gold-coated nanowires. The exact source of the background
Continuum is not known but is probably due to electronic scattering from metallic
nanostnlctures. The SERS substrates were evaluated using R6G and NB as probe

Thole
cu 1 es. Strong enhancement was found at isolated locations, which was sufﬁcient to

159

 

easily detect uMolar concentrations of these two probe molecules. On the other hand, no
discemable Raman spectra could be obtained from bulk solutions at these concentrations.
This clearly conﬁrms the SERS from nanowires but at few isolated locations. The
isolated enhancement was attributed to the nature of analytes that do not chemically bind
to the substrate. The substrates were also studied for SERS by using 4-MBT, BPE and
1 ,2-BDT as probe molecules. The aromatic thiol based analytes are known to chemically
bind with the metal coating on the substrate and yield better estimate of SERS
enhancement. The substrates yielded an average enhancement factor of the order of 106
with good repeatability and small standard deviation of approximately 20% over the
tested areas. The amount of analyte that yields these strong SERS spectra is on the order
of ~1 0'14 moles and conﬁrms the high sensitivity of these novel substrates. The SERS
performance of the nanowire substrates was also compared with that of commercial
SERS chip, Klarite from 3-D Technologies Ltd. This result demonstrates that the gold

coated nanowires can be a promising candidate for SERS.

160

Chapter 5: Low Temperature Synthesized GeO; Nanowires based

Substrates for SERS

In the last chapter, experimental details and growth kinetics for GeO; NW

synthesized at a temperature (~ 830 — 850 oC) above sublimation temperature of GeO
(710°C) were presented. The SERS evaluation of gold coated NW8 showed strong
enhancement for analyte that bond well to the substrate. However, the synthesis
investigation revealed little control over the parameters of the nanowires like length,
diameter, density etc. Raman enhancement from the substrate is expected to be
signiﬁcantly dependent on these parameters. In an effort to achieve greater control over
these parameter, an alternate synthesis approach was developed that limits the maximum
temperate of the source to ~ 650 0C i.e., below the sublimation temperature for GeO.
Thus this new synthesis is termed low temperature growth. This chapter presents the
synthesis details and Raman studies. Characterization of the substrate was done with
Scanning Electron Microscope (JEOL 6400V), Field Emission Scanning Electron
Microscope (JEOL 6300F) and Transmission Electron Microscope (JEOL 2200F S). The
as synthesized substrates were evaluated for their SERS performance using NIR Raman
system. 4-MBT and 1,2-BDT were utilized as probe molecules for SERS analysis. In this
chapter, ﬁrst the growth environment for nanowires is discussed and this is followed by

results of Raman analysis.

161

5.1 Limitation of high temperature nanowire synthesis
As noted in the last chapter, there have been a number of reports of GeOz NW8

synthesis using thermal evaporation[85], carbon nanotube conﬁned reaction of Ge[86] ,
laser ablation[9l], thermal oxidation[87-89, 92], carbothermal reduction[83] etc. All
these techniques have utilized high source temperatures in the range 820-1300 °C to

generate sufﬁcient vapor pressure of Ge or GeO in an oxidizing or inert environment and
relied on residual or leakage oxygen to form the GeO; NW8 or whiskers. Mostly the

NW8 were collected on the source, or close to it, at almost the same temperature. The
only exception is the synthesis by Bai et al.[85] where the NW8 were collected in the low
temperature zone (exact temperature in the growth zone was not reported by the authors)
in front of a copper cold ﬁnger on the inside walls of quartz tube. Another relatively low
temperature synthesis was carried out by Hu et al.[88] that had the source material at
1300 0C, but substrate was at 700 °C. The main rationale of using the high temperature is
to generate sufﬁcient vapor pressure of reactants to facilitate growth at a reasonable rate.
There are mainly two parameters governing the oxidation of Ge — temperature and the
oxygen concentration. While the temperature is expected to have a monotonic effect, the
oxygen concentration can have a non-trivial dependence on its partial pressure. For most
metals, the oxidation rate above 1 Torr are either independent of pressure or increase with
increasing pressure. However, Ge shows a strong inverse dependence of oxidation rate on
oxygen pressure at and above 550 °C[140]. Since GeO is the main oxidation product of
Ge oxidation, we expect a very high GeO vapor pressure above its sublimation point at
710 0C. Germanium monoxide vapor pressure of 28.48 Torr at 705 0C has been

FCPOrted[98], The high vapor pressure of GeO and its diffusion is responsible for the

162

 

observation that most of the source material is gone within 5-6 minutes of the start of
oxygen ﬂow. The rapid oxidation of Ge source rapidly infuses a lot of GeO onto the
substrate making it difﬁcult to exercise good control over growth. This is evident by
cracking of the top layer of the substrate as shown in Figure 5-1. The corresponding

SEM shows Ge:O atomic ratio of nearly 30:63 without any signature of the underlying Si

substrate. The thick peeled layers are therefore mainly comprised of GeOz and thickness

can be in excess of 10 um. Thinner Ge02 layers are expected to decrease the electron

beam penetration and allow detection of Si in the EDS. This suggests that the substrate is

getting signiﬁcantly thick coating of probably GeO during the initial deposition that later

transforms to GeOz. The NW8 start to grow out of this thick coating as the process

continues and generally become dense enough to shield the underlying thick layer. The
ﬁlm cracking is probably the result of thermal mismatch between top coating and
underlying substrate. The formation of an alloy at the interface cannot be ruled out as we
didn’t see a clean break between the substrate and ﬁhn as shown in Figure 5-2. Here one
can see the underlying Si substrate from region where top ﬁlm patch has been removed.
The corrugated nature of the substrate suggests formation of an alloy at the interface and
strong adhesion between the ﬁlm. and substrate. The elemental mapping of the substrate
(imaged in Figure 5-2) clearly shows the cavity is mainly composed of Si whereas the

surround area is made of germanium and oxygen as illustrated in Figure 5-3.

163

 

Figure 5-1. SEM image of GeOz NW substrate.

 

Figure 5-2. SEM image of cracked G602 ﬁlm on Si substrate during

nanowire synthesis.

164

OKal SiKal
GeKal AuLal

Figure 5-3. EDS elemental map for top right comer of cavity in Figure

-2.

Another issue relates to the uniformity of NWs growth across the surface of the
substrate as shown in Figure 5-4. In many cases agglomerations of particles, especially

closer to the source, are observed alongside the NWs as shown in Figure 5-5. Here one
can see nearly stoichiometric Ge02 (Ge:O ratio 33:63) particles without any signature

from underlying Si substrate.

165

 

 

Figure 5-5. SEM image of substrate showing different morphologies of
GeO; particles and NW8.

166

5.2 Low temperature Synthesis of nanowires

The low temperature synthesis of nanowires also uses the same quartz tube
experimental set up (Figure 5-6) that was used for high temperature synthesis, however
the experimental conditional of temperature, ﬂow rate, pressure and synthesis duration
are varied. This has led to a technique that addresses some of the issues associated with
the high temperature synthesis. The GeOz nanowires synthesis still follows the vapor-
liquid-solid (VLS) grth mechanism[81]. The source material of Ge is placed on a
quartz support plate and silicon substrates coated with 3 - 28 nm thick gold catalyst ﬁlm
are placed few inches from the Ge source material. The support plate is inserted in the
quartz tube such that the source is located in the high temperature zone near the center of
the fumace whereas the substrates are in the low temperature zone towards downstream
end of the fumace. The temperature was measured using ‘K’ type thermocouple placed
inside the quartz tube that can be moved along the length of the tube using magnetic
coupling without braking vacuum. The tube was evacuated to less than 200 mTorr by a
mechanical vacuum pump and subsequently backﬁlled with argon at a ﬂow rate of 50

Sccm. Temperature at the source was raised to 640-650 °C over 20 minutes and the
Vacuum valve at the downstream end was adjusted to obtain a pressure of ~ 2 Torr for the
Constant ﬂow of background gas. The temperature at the substrates was measured at ~
400 —500 °C. The grth of the NWS was initiated by introducing Ar mixed with oxygen
(40% Oz and balance Ar) at the rate of 10 sccm. The growth was done for 1 hr and the
pr essure increased to ~ 3 Torr over this time. The furnace was switched off and allowed

to cool to room temperature before opening the tube to atmosphere. The substrates

167

surface turned white to pink after the experiment. The substrate was characterized using

JEOL 6400 scanning electron microscope (SEM) and TEM.

 

 

 

 

 

 

Quartz support plate Ge source Au coated substrate
r._“_\,_ _ _____ _ -4". ,‘f Pressure gauge
1 r ,. 5

AI ——«> 1:: g '
E ‘3’ 12E —-6
1: ! —1
Ar + 02 " 3 ; h to vacuum
9 i i pump
uartztube .‘ “
Q a thermocouple
Hollow cylindrical fumace

Figure 5-6. Schematic of low temperature synthesis of germanium dioxide
nanowire.

Figure 5-7 shows the SEM image (15 keV with ~15 mm working distance) of the

Substrate fully covered with dense nanowires.

 

Figure 5-7. SEM image of Ge02 nanowires on Si at 400-500 °C.

168

 

 

 

 

 

0 Sr
Au
Au Au Au AuGhe Au Ge
1 2 3 // 9 10 11
keV
Element App Intensity Weight% Weight% Atomic%
Conc. Corm. Sigma
0 K 0.19 1.1035 29.39 1.81 60.99
Si K 0.04 0.6767 10.11 0.90 11.95
Ge L 0.27 0.7751 58.36 2.00 26.70
Au M 0.01 0.5909 2.14 1.78 0.36
Totals 100.00

 

 

Figure 5-8. EDS spectrum and quantiﬁcation results for GeO; nanowires.

Figure 5-8 shows the x-ray Energy Dispersive Spectroscopy (EDS) spectra and
quantization results for the NWS substrate. Though it is difﬁcult to measure the Ge/O

r atio of the wires quantitatively due to the intensity uncertainty of the EDS detection for

I lgrater elements, it is estimated that the chemical composition of the GeOz NW is close to

169

GeOz. The atomic ratio of germanium to oxygen from the quantization results is found to
be 27:61 that is quite close to the actual ratio of 1:2. It may be mentioned that a clear
signature of gold and the underlying Si substrate was observed in EDS, which conﬁrms
thin top layer of NW8. The nanowires were also characterized by JEOL 100 CXII ﬁeld
emission transmission electron microscope (TEM). The TEM samples were prepared by
sliding holey carbon-coated TEM grid across the sample face. Figure 5-9 shows the
TEM image of a two GeOz nanowire with a diameter of ~52 nm - 87 nm. The TEM

image shows a dark spherical tip at the end of the thin nanowire that is probably the
catalyst particle. The other NW doesn’t show catalyst particle at the tip and may have
been broken during sample preparation. The right side image of Figure 5-9 shows some
nanowires with and without catalyst tip. Interestingly it also contains an isolated dark
spherical particle (denoted by red circle) and supports the hypothesis of catalyst particle

being broken during sample handling or preparation.

   

200 nm

Figure 5-9. TEM image showing dark catalyst particle at tip. Right image
shows a dark spherical particle probably broken during sample
preparation.

170

The NWs synthesized at low temperature show good uniformity over large areas
of the substrates but do show some circular clusters of different morphology as shown in
Figure 5-10. Here, one can see isolate circular islands superimposed on more uniform
NWs. The morphology of NW8 in the islands is different from the other NWs in that
these are much longer, curved and narrower. The cross-section of the substrate also
shows the normal length of NWs is few microns but in these circular regions, nanowires

reach tens of microns in length.

 

Figure 5-10. Different morphology of low temperature GeOz NWs.

171

UI
3.“

gen

0111‘

i

011

r‘ .-

m1]

5.3 Nanowire growth kinetics
The synthesis of GeOz NW8 involves oxidation of germanium source to generate

germanium oxide vapors. These vapors then diffuse away from the source and condense
on the gold coated substrate to synthesize NW8. This section presents the details of
oxidation of the germanium source at low temperatures and the subsequent grth of

NW8 from the vapor phase germanium oxide.

5.3.1 Oxidation of germanium

The low temperature synthesis of nanowires is also characterized by thermal
oxidation of germanium source in an oxidizing enviromnent. Germanium is readily
oxidized upon heating in a stream of oxygen/air and a ﬁlm of GeO always forms at the
start of oxidation that transforms into GeOz as the reaction proceeds[98]. Many studies

conﬁrm the formation of Ge0x(1 < x < 2) during the initial stages of thermal oxidation
of Ge[99-101]. The sub-oxides transformed to GeO; with increasing temperature to

300 °C. However, further heating caused the formation of volatile GeO and complete
desorption of oxide occurred at 450 —475 °C.
The main reactions during thermal oxidation of germanium in an oxidizing

environment are: [98]

Ge(s)+ 1/202 (g) —) GeO(s) (5.1)
GeO(s) —> GeO(g) (5.2)
172

 

press

1111:-

The relative contribution of these reactions is a strong function of oxygen partial
pressure. The oxidation rate of Ge increases with decreasing oxygen pressure and the
oxidation rate is independent of crystal orientation of Ge source above 550°C[l40].
Indeed as the oxygen pressure decreases, nearly all of the GeO formed on the surface
manages to sublimate. i.e., reactions (4.1) and (4.2) take place, the rate of reaction (4.3)
being very slow. When oxygen partial pressure increases, the oxidation proceeds
according to reactions (4.1) and (4.3) i.e., the reaction (4.3) begins to prevail over
reaction (4.2) and the oxidation rate slows down considerably. Similarly the oxidation
product is mainly GeO if oxidation is carried out under conditions where the amount of

oxygen is insufﬁcient to converts it to GeOz (for example in a stream of N2 or Ar

containing oxygen admixture)[95]. The volatile GeO can diffuse away from source or
decompose during oxidation/annealing process at high temperatures. In contrast to GeO,

GeOz is very stable and doesn’t melt unless temperatures above 1116i4 oC[104] are
reached. On heating GeOz to 1000-1100 °C, the vaporization rate is very low and there is
no dissociation to Ge and 02 or GeO and 1/202[95]. The only possible decomposition

pathway for GeOz is through reaction with Ge to form GeO at or above ~425°C[102].
The resulting GeO then desorbs from the surface without undergoing any further
transfonnation[97]. In summary, it is widely agreed that GeO is the only desorption
product of oxidation/annealing of Ge and its desorption takes place in the temperature

range 400 — 500 °C.

173

v.)-
S“

4.3.

$011

11':

,9:
f‘.’

Q
.‘J

0 .

rl.‘

5.3.2 Growth kinetics

The synthesis of GeO; nanowires involves a solid phase Ge source and gas phase
species of Ar, H2 and oxygen. The Ar is used mainly as the carrier gas to transport the

source material to the substrates and H2 is included to avoid source and substrate
oxidation during the heating phase. The beginning source material of Ge is likely covered
with a thin layer of native GeO; since it has been exposed to atmosphere during storage.
During heating of the source under low vacuum (2 Torr) with a protective ﬂow of Ar and
Hz, the native oxide is expected to be reduced to Ge or will escape by transforming into

GeO above 400 °C. The integrity of the source (weight loss due to vaporization) is
maintained during ramping and even at the reaction temperature of 650 °C, since the Ge
(melting point ~ 93721.05 °C[105, 106]) has negligible vapor pressure at these
temperatures. Hence, direct evaporation of Ge is not considered to be a contributor to

nanowire growth.
The oxidation of germanium begins with the introduction of oxygen in the gas

stream. It is hypothesized that the insufﬁcient supply of oxygen results in the formation
of GeO on the surface of the source material. Even if a small amount of GeOz is formed,
it will react with the underlying layer of Ge and transform into GeO at 650 °C. We don’t
expect any thicker layer of Ge02 to form on the source for the experimental conditions.
The thermal stability of GeOz (Melting point 1116i4 0C) will prevent the further

OXIdation of Ge but we observe signiﬁcant loss of the source as the reaction proceeds.

Therefore, the ﬁrst reaction occurring during the grth of NWS is

174

Ge(s)+ 02 (g) —+ 2Ge0(s) (5.4)

It may be re-emphasized that for every temperature, there is an optimum oxygen
concentration at which the oxidation rate of germanium doesn’t surpass the sublimation
rate of the monoxide being formed in reaction (5.4). Otherwise, a thick layer of dioxide
will form on the surface of the germanium and the oxidation process will virtually come
to a halt. This is evident from Figure 5-11 where Ge oxidation rate is plotted versus

oxygen pressure at 550 °C. The lower oxidation rate at higher oxygen pressure is due to

the formation of a top GeOz layer that acts as a diffusion barrier for diffusion of Oz and

GeO to/from GeOz/Ge interface.

p—d
N
I

 

00
I

Oxidation rate ><103 mole/cm2 s

 

J 1 l

40 80 120
Oxygen pressure (Torr)

 

Figure 5-11. Rate of Ge oxidation versus oxygen pressure at 550 °C [98].

175

The GeO(s) formed during reaction (5.4) is volatile at high temperatures (vapor

pressure ~ 3.29 Torr at 644 oC[98]), so some part of it goes to gas phase

GeO(s) a GeO(g) (5.5)

The GeO vapors formed in reaction (5.5) can undergo further oxidation (GeO
oxidation starts 550 °C in air) in the presence of oxygen to yield GeOz according to the

reaction[107]

GeO(g)+ 1/202 (g) —> GeOz (s) (5.6)

But oxidation of GeO is not expected to be signiﬁcant due to the low

concentration of oxygen in the environment. If reaction (5.6) proceeds at a signiﬁcant
rate, a thermally stable layer of GeOz will deposit on substrates and the walls of reaction
chamber. Nevertheless, the actual vapor phase is expected to be a mixture of mostly GeO
along with some GeOz. The precise analysis of vapor phase species requires more

elaborate experimental set up (like X-ray photoelectron spectroscopy) than the one

available.

In the vapor phase, there is also a possible thermal decomposition pathway for
GeO. It is known that solid GeO disassociates into GeOz and Ge at high temperature
according to reaction

2060 (g) <=> Ge + G802

176

The reaction temperature controls the relative probability of forward or reverse
reactions[98]. It may take several hours for decomposition at 600 °C but at 700 °C it

takes few minutes[98]. In the case of deposition on a condenser, GeO will deposit
without decomposition. On the other hand, it will decompose into Ge and Ge02 without a

condenser at fairly high temperature. Since the substrates are also located in a relatively
low temperature zone (400-500 °C), we don’t expect a higher rate of decomposition-
based deposition i.e., substrate gets mostly GeO, and some decomposed GeO; and Ge.
Nevertheless the existence of Ge nanocrystals (NCs) in the deposited GeO ﬁlm cannot be
ruled out. In fact, Gorokhov et al. synthesized Ge NCs in Ge02 ﬁlms by depositing from
supersaturated GeO vapors[141]. The situation is further complicated by the presence of
a thin gold ﬁlm on the substrate that forms a eutectic with the underlying silicon. The
exact manner of adsorption of GeO or GeO; into Au cannot be predicted since no
information about phase ternary phase diagram for Ge, Au and oxygen system could be
found. However, it is postulated that the presence of liquid eutectic droplets will facilitate
incorporation of GeO into the droplet yielding an initial sub-oxide composition GeO or

possibly some Ger (l < x < 2) of NW[109]. The composition of nanowire then might

transform to more stable GeOz by incorporation of oxygen from the background gas as

well as through 1088 of GeO by vaporization. Under such a scenario, the excess GeO
diffuses away from the reaction zone and condenses in the colder parts of the growth
chamber. A yellowish to brownish-black coating (characteristic color of GeO) found at

the ends of the quartz tube supports this hypothesis.

177

Thus the most plausible mechanism for the formation of Ge02 NWS can be

summarized as:

(a) Oxidation of Ge to form GeO(s).
(b) Transformation of some part of GeO(s) to GeO(g).

(c) The transport of GeO(g) (and possibly some small percentage of
GeOz) to the substrate under a reduced pressure.

((1) Preferential adsorption of GeO on the substrate at the sites of the gold
catalyst and possibly some associated decomposition into Ge and
GeOz.

(e) The grth of Ge0x(l<x<2) NW8 via the VLS mechanism by

absorbing more source material from the vapor phase at the tip of the

nanowires.
(f) The evolution towards more stable GeOz NW8 by incorporation of

oxygen and loss of GeO.

It may be noted that the low temperature synthesis follows similar kinetics as that
of high temperature growth. The important difference lies in the temperature regime,
which slows down the oxidation rate of Ge and GeO considerably to yield better control
over the amount of gas phase species generated and transported to the substrate. This
slows the rate of material deposition onto the substrate leading to better control over

length and diameter of the NW8. The low substrate temperature also helps to reduce

178

116501

but 11

11:01)

desorption of the incident material yielding better surface converge, i.e, no dots appear
but uniform surface color changes are observed over whole surface of the substrate.

The discussion so far has ignored the presence of hydrogen in the gas stream. It is
known to be an oxide reducing agent depending upon the partial pressures of hydrogen
and that of water in the system for a given temperature. As a practical manner, the partial

pressure of water is equal to the base pressure of the system[111]. In our case, the base

pressure is ~ 200 mTorr and partial pressure of H2 is ~ 100 mTorr (for a total pressure of
2 Torr of a gas mixture containing 5% H2). Figure 5-12 is a plot of thermodynamic

stability of GeOz as a function of ratio of hydrogen pressure to water pressure for

different reaction temperatures[112]. For a £12- z 0.5 , the GeO; is unstable for all
P(H20)

temperatures above 600 K and is expected to decompose to constituent elements. Since
the NW synthesis is carried out above this temperature, we don’t expect any GeO; on the

source material at the start of oxygen ﬂow. The situation changes after introduction of
oxygen into the system and the oxidation of Ge becomes appreciable compared to the

reduction of GeO/GeO; by H2.

179

  
  

 

 

1O ..
o GeOz unstable
6 10 f
N
1'5
Q 1
’31 i
a; 1
0' 10 i GeOz stable
1 0'2 ‘ ‘ ‘ 4 ’
200 400 600 800 1000 1200

Temperature (K)

Figure 5-12. Plot showing GeO; stability as function of temperature and
the hydrogen to water vapors pressure ratio[l 12].

5.3.3 Synthesis temperature proﬁle

The actual temperature inside the furnace was measured with a ‘K’ type
thermocouple placed inside the quartz tube. The thermocouple can be moved along the
length of the furnace by using a magnetic coupling without breaking the vacuum. This
arrangement allowed mapping the temperature proﬁle of the furnace at different spatial
locations. Figure 5-13 shows a typical temperature proﬁle along the length of the furnace
during NW8 synthesis run. To get a more accurate proﬁle, the thermocouple was moved
from its initial downstream position towards center of the furnace and then brought back
' to the initial position. The temperature proﬁles for the inward and outward movement of
the thermocouple showed good agreement. The furnace has good temperature uniformity

for the center zone (center i2 inch) and then the temperature drops rapidly (~ 50 °C /in.)

180

511

l‘is

. \
n.4, ‘

toward open end of the furnace. This temperature gradient allowed the source and

substrate to have different temperatures from the optimal growth of NW8.

 

 

700
A temp inward
0 temp outward
o l
650 ' 8 2
8
O
00 ‘ o
2 600" A
3
e 2
in
Q.
11E) 550* o
1— ° 0
O
5001 ° 0
A
450 1 L 1 1 h J
0 1 2 3 4 5 6

Position furnace Center —-> downstream end (in)

Figure 5-13. Tube furnace typical spatial temperature proﬁle during nanowire
synthesis.

5.3.4 Pressure proﬁle

Observation of pressure during NW8 synthesis revealed a typical variation pattern
as shown in Figure 5-14. It is important to mention that the pressure is measured at the
downstream end of the tube and may have some deviation from the actual value at the
location of the source. The synthesis is initiated at ~2 Torr by introducing a mixture of Ar

and oxygen (Ar 60%, Oz 40%) at a ﬂow rate of 10 sccm. The initial rapid rise of pressure

can be attributed to the increased gas ﬂow as well as the contributions from GeO vapors

181

that are formed as a result of oxidation of Ge source. Then there is a more gradual rise of
pressure to a ﬁnal constant value of ~2.9 Torr, which can be related to the temperature

rise during a long growth period. The temperature varied by almost 5-6 °C during 1 hr of

growth.

.N
(O
T

 

 

0 1000 2000 3000 4000
Time (s)

Figure 5-14. Typical pressure proﬁle during NW synthesis.

5.4 Factors affecting NW8 synthesis
The low temperature synthesis of GeO; NWs is dependent upon a number of

CXperimental parameters that control the yield and characteristics of nanowires. These
include thickness of the catalyst ﬁlm, grth time, gas ﬂow rates, background pressure,

temperature, etc. The detailed study of the affect of these parameters on nanowires is

Presented in this section.

182

1110-51
frougl
11.111;
for .W

1.111111 1

5.1.1.1

.g ~ .
:1 ”Say-,0
4““.

5.4.1 Effect of gold ﬁlm thickness

The control over the diameter of NW8 is an important factor in engineering the
nano-structured substrates for possible applications. This control is mainly exercised
through the initial gold ﬁhn thickness. The thin gold ﬁlm breaks down in small islands
during heating phase and subsequently these islands become nucleation and growth sites
for NW8. Both the annealing of gold ﬁlms and the diameter dependence of NW8 on

initial ﬁlm thickness is studied in this section.

5.4.1.1 Gold film annealing

The catalyst plays an important role as the nucleation site and subsequent growth
of the nanowires in the VLS mechanism. The main emphasis is the preferential
adsorption of gas phase species into the catalyst or its eutectic form that is in liquid phase
at the growth temperature. This results in nanowire grth that is highly localized to the
gold-coated area. Figure 5-15 shows dense and patterned NWs grth on Si(100)
substrate that had Au removed in “X” pattern from the 21 nm thick Au ﬁlm. As can be
seen, the NW8 grth (white areas) is strongly limited to the Au coated area with no
growth on uncoated area (X pattern). The high degree of uniformity of the W8 is also
evident from the low magniﬁcation image. The catalyst speciﬁc growth is one of the
strong characteristics of VLS mechanism and gives evidence of VLS mechanism taking

place.

183

 

Figure 5-15. GeO; NWs on Si(100) that has patterned Au ﬁlm (No gold
was coated in the dark “X” pattern). Right image shows close up of the red
circled area of left image.

Another characteristic of VLS growth is that the NW8 generally terminate in a
catalyst particle that has a spherical shape. Figure 5-16 shows the SEM and TEM image
of NWs. The SEM image shows NW8 growth around two horizontal slits patterned into
the Au ﬁlm on a silicon substrate. The bright particles at the tips of NW can be easily
seen for most NWs that grew to cover the uncoated (dark) area. The TEM image clearly

shows a dark eutectic particle at the tip.

   

7 50 nm
Figure 5-16. GeOz NW showing particle at tip.

184

the n11
113.1111
3111115
5115 .01
the di:
19.11111

here d

03165
4; .‘r
knife]
_ A
53311111

2113'
"“193.

The catalyst ﬁlm thickness plays an important role in determining the diameter of
the nanowires in VLS mechanism. The thermal annealing of the catalyst ﬁlm results in
breaking up of ﬁlm into small islands due to the differences in surface energies of the
catalyst and underlying substrate. These islands or particles then become the nucleation
site for the growth of the nanowire. The size of the metal islands has a direct bearing on
the diameter of the nanowires. A systematic study of evolution of catalyst ﬁlm into
particulates was carried out at the reaction temperature. Different thickness of gold ﬁlms
were deposited on Si(100) wafer and the samples were subjected to typical synthesis
conditions without a germanium source material or oxygen ﬂow. Figure 5-17 through
Figure 5-21 shows the SEM images and corresponding particle size distributions for the
Au particles. The diameters of the particles were determined from high magniﬁcation
images of the sample. For each sample, at least diameters of nanowires from at three
different locations were measured using distance measurement tool of SEM imaging
software. This ensures that the measured diameters are better representatives of general

nanowire sizes. The distribution plots were then generated using histogram function of

Matlab.

185

 

 

12_ mean= 17

std_dev= 3

Number of islands

 

 

(10 12 14 16 18 20 22 24
Diameter of Au islands (nm)

Figure 5-17. SEM image and particle size distribution for 3 nm Au ﬁhn.

186

 

Number of islands

.x _x
N
l

mean= 34

   

std_dev= 12

U1

 

0000)“)

16 24 32 4O 48 56 64
Diameter of Au islands (nm)

Figure 5-18. SEM image and particle size distribution for 7 nm Au ﬁlm.

187

 

   
   

mean= 82

std_dev= 40

_x
0

Number of islands
8

 

0 60 120 180 240 300
Diameter of Au islands (nm)

Figure 5-19. SEM image and particle size distribution for 15 nm Au ﬁlm.

188

 

mean= 99

25 ~ std_dev= 55

Number of islands
_\ _L N
O 01 O

01

 

 

O 80 160 240 320 400
Diameter of Au islands (nm)

Figure 5-20. SEM image and particle size distribution for 21 nm Au ﬁlm.

189

 

mean= 189

16 _ std_dev= 109

Number of islands

 

 

O 80 160 240 320 400 480 560
Diameter of Au islands (nm)

Figure 5-21. SEM image and particle size distribution for 28 nm Au ﬁlm.

190

Figure 5-22 shows the mean particle diameters for different ﬁlm thickness and
relation is approximately linear. It may be emphasized that the particle size distribution is
in line with the earlier reports. For example, Borshch et al. found the metal island size
distribution has spread from 10 - 250 nm for 15-25 nm thick vacuum sputter coated Au
ﬁlm on polished glass after annealed at 740 K for 1 hr[l42]. Though it is not possible to
directly relate island sizes in our case since we differ in underlying substrate (silicon is
known to form eutectic with Au), time and temperature. Still the spread of metal island

sizes is comparable (we get 25 - 360 nm for thickness 15-21 mm).

 

300- ..
8 250-
C
L“
.22
3 200-
<
“6
as 150-
“<15
E 1-
(U
'- 100*
g / L
(0
(D
E

 

 

01
O

 

J l

o 5 110 115 2'0 25 30
Thickness of Au film (nm)

Figure 5-22. Mean particle size distribution for different initial Au ﬁlm
thickness at 500 °C. Error bars indicate standard deviation from mean.

191

5.4.1.2 Effect of gold film thickness on diameter of nanowires

The diameter dependence of NWs was investigated by coating silicon substrates
with different initial thickness of gold ﬁlms. The GeO; NW8 were synthesized from these
substrates under similar experimental conditions. The diameter measurements were done
using SEM images. For each substrate, NW8 from at least three different areas were
characterized to ensure a greater generality of results. Figure 5-23 through Figure 5-25

show the diameter distribution for NWS synthesized from 7 - 21 nm thick Au ﬁlms.

   
  

121
l mean=79
l
1 0% std_dev= 17
1
‘1
w 8
E 1
z 1
3 Bi
as
n l
E
3 1
Z 41
l
l

(40 50 60 7O 80 90 100 110 120
Diameter of GeO2 NWS (nm)

Figure 5-23. GeOz NW diameter distribution for 7 nm Au ﬁlm.

192

N
.A

mean= 108

N
.3

std_dev= 21

Add
moron

Number of NW5
‘9

 

 

 

g0 70 90 110 130 150 170 190
Diameter of GeO2 NWs (nm)

Figure 5-24. GeO; NW diameter distribution for 14 nm Au ﬁlm.

24 * mean= 120

N
A

std_dev= 20

Number of NW3
.8 —8 —L
O) (D N 01 CD

00

 

 

65 80 95 110 125 140 155 170
Diameter of GeO2 NWs (nm)

Figure 5-25. G602 NW diameter distribution for 21 nm Au ﬁlm.

193

The diameters of annealed Au islands and W3 for different gold ﬁlm thickness
are plotted in Figure 5-26. Although the diameter of the NWs is not the same as the
corresponding measured Au island size, the diameters of the synthesized NWs do have a
one-to-one correspondence with the measured particle diameters for that ﬁlm thickness.
An obvious explanation for the increase in the size of NWs compared to corresponding
eutectic particle is that the liquid alloy catalyst can not move up another atomic layer
until the entire layer under that droplet is formed. For a larger diameter liquid alloy
droplet, a larger capture cross-section exists for gathering GeO molecules from the gas
phase, however a larger volume also exists which must be supersaturated before the next
atomic layer is added to the end of the nanowire. The increased volume of the liquid alloy
is reflected in the larger diameter of the corresponding NWs. Another observation that
relates to the large size difference between the initial gold particles and the NWs for very
thin gold ﬁlms. This can be attributed to the Gibbs-Thomson effect that prevents growth
of very small diameter NWs for a given degree of supersaturation. This happens due to
higher vapor pressure for sharp curvatures (small diameter) and the growth of very thin

NWs becomes thermodynamically unfavorable.

194

 

120v

  
   

— NWs

110 *6"- Au particles

1 00
90

80

Diameter (nm)
\

70 ” /"

60 ~ .2"

40 r /"

 

l l

8 10 112 {4 16 18 2o
Thickness of initial Au ﬁlm (nm)

Figure 5-26. Mean diameters of GeOz NWs and annealed Au ﬁlm
particles.

5.4.2 Effect of growth time on length of nanowires

In the VLS growth mechanism, the length of the NWs is controlled by growth
time. The NWs were synthesized for different length of time while keeping the rest of the
experimental conditions the same. The lengths of synthesized NWs were measured using
SEM images. The length of nanowires, however cannot be precisely measured from top
view images of substrates since NWs are located at random orientation on the substrate
and the SEM image just gives the top projection. To avoid this difﬁcultly, cross-sectional
images were taken as shown in Figure 5-27 for different lengths of growth time. It is
obvious that the length of W5 is directly related to the growth time. The same

relationship can be more clearly evident in Figure 5-28. Here one can see that the length

195

is almost linear with the growth time. For thinner metal catalyst ﬁlms, the length of the
NWs is somewhat longer than thicker metal catalyst ﬁlms for same growth time. This is

not unexpected, as thinner will tend to grow faster for same degree of supersaturation of

vapor phase species.

   

10 min 15 min

 

30 min 60 min

Figure 5-27. Effects of growth time on the length of the NWs.

196

5500 -

5000
4500
E 4000
5
«n 3500
E 3000
O
.C
E’ 2500
3 2000
1500
1000
500

l

l

T

F

l

 

—7 nm Au ﬁlm
--b-- 14 nm Au ﬁlm

30 40
Growth time (min)

Figure 5-28. Length of NWs versus growth time.

5.4.3

The synthesis was found to be very sensitive to oxygen concentration as discussed
in the section on kinetics of reaction. Figure 5-29 shows NWs synthesized on a silicon
substrate coated with 7 nm Au catalyst ﬁlm for different concentrations of oxygen. The
oxygen concentration is controlled by varying the ﬂow rate of the oxygen in the chamber
while keeping the total ﬂow rate constant. As can be seen, the best growth is obtained for
an oxygen ﬂow rate of 5-15 scorn (corresponding background gas ﬂow of 55—45 sccm).

Low oxygen ﬂow results in an oxygen deﬁcient synthesis. This decreases the probability
of forming (3602 on the source surface that can inhibit further oxidation of the source. On

the other hand, the low oxygen concentration results in a decreased growth rate.

Increasing the oxygen concentration leads to longer NWs due to increased transformation

Effect of oxygen flow rate

197

 

20 sccm 30 sccm

Figure 5-29. Effect of oxygen concentration on NW synthesis.

of sub-oxide into more stable G602. The growth was found to completely vanish for
ﬂow rates of 30 sccm or more. It is interesting to note the substrate did get a G802

coating for the 30 sccm gas ﬂow rate (Ar with 40% oxygen) but no NWs were formed.
The EDS studies showed Ge:O ratio of 27:66 whereas the case of a 20 sccm ﬂow rate
with NW showed a ratio of 23:61. This implies that the substrate is almost getting the

same amount of reactants for the oxygen ﬂow rates of 20 and 30 sccm with very different

198

NWs results. It is suspected that the higher oxygen is resulting in an increased formation

of (360; in the vapor phase. This GeO; then condenses on the substrate but the absence

of GeO prevents growth of NWs. Perhaps GeO is depositing on the wet Au droplet, then

forming GeO; either while still in the Au, or on the surface of the Au ball, or it is forming

a short section of the nanowire as a GeO nanowire that then converts to GeOz as

suggested in the growth model. The incorporation of some GeO; during growth explains

the roughness seen on the nanowires. The concentration of oxygen affects the balance of

G60 and GeO; on the substrate and inhibits the nanowire growth beyond an optimal

value.

\1 CD (0
O O O
l

O)
O

(a) b
O 0

Atomic percentage (%)

N
O

.3
O

01
AO

 

 

10

«r 31

15 20 25 30
Flow rate (sccm)

Figure 5-30. Elemental composition of the sample surfaces after grth
for different oxygen concentrations.

199

Figure 5-30 shows the results of EDS study of substrates for different flow rates
of oxygen. It is interesting to see how the silicon composition varies for different flow
rates. Silicon being the underlying substrates, its composition gives an indirect estimate
of the density and length of the NWs. As discussed earlier, the low ﬂow rate yields
shorter NWs so we get higher silicon signal for low ﬂow rates. It decreases as the density
and length of NWs increases for increasing ﬂow rate. However, it tends to increase for
higher ﬂow rate (around 20 sccm) where the density tends to diminish. Silicon

concentration shows a decreasing trend at higher ﬂow rate (~30 sccm) that can be

attributed to formation of relatively uniform GeOz coating instead of NWs.

5.4.4 Effect of background gas flow rate

The effect of background ﬂow rate was studied by varying the flow rate while
keeping all other parameters like catalyst thickness, grth time, and temperature the
same. Figure 5-31 shows NWs synthesized on silicon substrate coated with 7 nm Au
catalyst ﬁlm for different background gas flow rates. It is evident that the best growth
occurs for a background ﬂow rate of ~ 60 sccm. The lower and higher rates yield sparse
growth of NWs. Low ﬂow rate, implying lower oxygen concentration, decreases the
formation of the vapor phase reactants and the degree of supersaturation. This decreases
the yield of NWs as is the case for 33 sccm gas ﬂow rate. On the other hand, high ﬂow
rate means fast transport of source material through the reaction zone (to maintain
constant pressure) and this decreases the probability of incorporation into the NWs.
Another effect could be the cooling and/or rapid source vapor drift associated with the

high ﬂow rate that reduces local temperature at the substrate to decrease NW5 yield.

200

 

33 sccm 60 sccm 120 sccm

Figure 5-31. Effect of background gas ﬂow rate on NWs synthesis.

5.4.5 Effect of temperature

Temperature has pronounced impact on NWs yield. Figure 5-32 shows the SEM
images of NWs observed at different substrate temperatures from 600 — 400 °C. The yield
of the NWs was found to be optimum in the range 400-500 °C. All experiments were
carried out with Ar (mixed with 5% hydrogen) as background gas at a ﬂow rate of 50
sccm and 10 sccm of Ar (with 40% oxygen) was used to synthesize NWs. The substrates
only show randomly distributed and interconnected clusters containing small particles
(bright spots) at high temperature (601-591 °C). These clusters tend to diminish in size
leading to more isolate islands as the temperature decreases (577-561 °C). The isolated
growth from 2D to 3D can be seen at ~533 °C where few nanostructures appear to grow.
The substrate at ~523 °C shows a number of isolated and long nanowires. The yield of
nanowires can be seen to increase for the lower temperature regions. The low yield at
higher temperature is probably due to less sticking of the high energy volatile GeO on the

substrate.

201

 

577 °C

~591 °C

601 °C

 

°C

~467

°C

485

515°C

 

~40] °C

 

Figure 5-32. GeO; NW grth at different temperatures.

202

5.4.6 Background gas type

Although most of the nanowires were synthesized using a mixture of Ar (95%)
and hydrogen (5%), some experiments were carried out using pure Ar or N; as a
background gas. The morphology of the NWs synthesized using different gases turned
out to be different. Figure 5-33 shows the some of GeOz NWs obtained for using Ar as
background gas. There have been large variations in the morphology of yield from NWs
to nano-particles. Similar types of morphologies were noted for NWs synthesized with

N2 background as shown in. Figure 5-34.

 

Figure 5-33. SEM images of GeO; NWs synthesized using Ar as
background gas.

203

 

Figure 5-34. SEM image of GeO; NWs synthesized using N; as
background gas.

204

Tl
investigal
hydrogen
or stops
depender
smaller c
substrate
condens:
different

10111 (in

The diameter of NWs obtained using different background gases was also

investigated and is shown in Figure 5-35. The smallest diameter is obtained for Ar and
hydrogen mix. This is probably due to reducing affect of H2, which in a way etches away,
or stops lateral development of oxide NWs. The diameter versus gold ﬁlm thickness
dependence is similar for the case of Ar and N2. The Ar background, however results in

smaller diameter wires than when N2 is used. One possible reason could be greater

substrate cooling with N2 that has a greater thermal conductivity resulting in higher

condensation of vapors leading to larger diameter wires. Another reason could be a
different actual pressure inside the tube than the one measured (synthesis done at ~ 2

Torr) due to different response of pressure sensor to different gases.

12o — ...-»-‘
--b-~Ar+H2 .. ..... .. -
+ N2 “fa-"M.

110— “cw-Ar r"

100*

   
  

.r' "I
,.
,u
.e

V.
20"
' we"!

CD
0

 

Diameter of GeO2 NWs (nm)
(0
C?

70

1 1 1 1

4 6 8 10 12 14 16 18 20
Thickness of initial Au ﬁlm (nm)

Figure 5-35. GeOz NWs diameters for different background gases.

205

Another effect of background gas type is noticeable from the appearance of the
substrate. The substrate was found to have white dots distributed randomly as was the
case when Ar with Hz was used as background gas. However, the density of such island
like formations of NWs was found to be much greater for N2 and Ar background as
shown in Figure 5-36 and Figure 5-37 respectively. The islands are actually formed of
dense, long and curved nanowires. The density and thickness of the NWs in island area

can be gauged from the EDS signature that doesn’t show any Si signature.

 

Figure 5-36. Island like growth of NWs for 7 nm Au coated Si substrate
with N2 as background gas.

206

 

Figure 5-37. Island like growth of NWs for 7 nm Au coated Si substrate with Ar

as the background gas.

Another interesting effect of the background gas lies in the temperature at which
the NWs can be obtained. In case of Ar and H2 environment, the best grth was
obtained in the temperature region 400-500 °C. Although NW growth was observed in
this temperature region for N2 and Ar background, the morphology of the structures
varied from NWs to particles. On the other hand, signiﬁcant NWs growth was found at
higher temperatures in case of Ar background gas. Figure 5-38 and Figure 5-39 show

the GeO; NWs found at the high temperature (~600 °C) region of the furnace for catalyst

207

ﬁlm thickness of 7 nm and 15 nm respectively. It is evident that the NWs are much

longer than the typical low temperature NWs seen in low temperature regions. Another

difference is the morphology where these NWs show strong curvature and sharp turns.

 

Figure 5-38. G602 NWs found at higher temperature region of fumace on
7 nm Au coated Si substrate.

 

Figure 5-39. Ge02 NWs found at higher temperature region of furnace on
15 nm Au coated Si substrate.

208

5.4.7 Effect of background gas pressure

The background gas pressure at the start of the reaction affects the yield of the
NWs by changing the degree of supersaturation in the gas phase. Background gas
pressure mainly helps to conﬁne the gas phase reactant near the source and substrate.
Figure 5-40 shows NWs synthesized on Si substrate having different Au ﬁlm thickness
at background gas pressure of 1.1 Torr. All three substrates show sparse growth
indicating background pressure is insufﬁcient to yield enough supersaturation of vapors
for efficient NWs growth. Figure 5-41 shows the NWs synthesized on Si substrates with
7 nm Au coating at a background pressure of 6 Torr. Here, we see a strong location
dependent growth of NWs depending upon the proximity of the substrate to the source
material. There is some NW growth at the upstream end (closer to source) of the substrate
but diminishes sharply within few m away from it. This can be explained by the effect
of high background gas pressure that keeps the reactants localized and only those parts of
the substrate close to source get enough vapors to initiate growth. As we move away from
source, concentration of vapor phase or supersaturation decreases resulting in

diminishing NW yield.

 

7nmAu

Figure 5-40. NW synthesis at a background gas pressure of 1.1 Torr.

209

   

Upstream end 5 mm 10 mm

Figure 5-41. NW synthesis at a background gas pressure of 6 Torr shows
location dependent growth.

5.5 Raman characterization of GeO; nanowires

The Raman system used in this study was EzRaman-L from Enwave Optronics
Inc. with 6 cm'l resolution at 785 nm excitation. The sample was mounted on a xyz-stage
and illuminated by the Raman system laser beam via optical ﬁber link. The Raman probe
head was attached to a z-stage that facilitates precise focusing of laser beam onto the
sample surface. All spectra were acquired with a focusing lens (N .A. = 0.22) that yields a
spot size of less than 100 um diameter on the substrate. The laser power on the substrate
was nearly 30 mW as measured with Newport optical meter (model 815 in conjunction

with 818-SL detector and 883-SL attenuator).

5.5.1 Raman spectra of substrate

The bare substrate containing GeO; nanowires were ﬁrst scanned by Raman
system to get the reference signal and to ensure no impurities are present that might cause
interference with the Raman signal. Figure 5-42 shows the Raman spectra of GeOz

nanowires of two different diameters synthesized on Si(100) by using different initial

210

6000 ~

-----108 nm
——-120 nm

5000 r

4000 ~

3000 ~

 

2000 ~

Raman Intensity/Arbitr. Units

1 000

 

 

 

560 1600 1600 2000
Wavenumber/cm"1

Figure 5-42. Raman spectra of GeO; NWs of different diameters,

catalyst ﬁlm thickness. Unlike the case of high temperature growth, here only the main
peak at ~444 cm'I is noticeable. Also the strong substrate peaks at ~ 520 cm'1 alongwith a
broad background is clearly visible. This broad background is similar to the one we
observed aﬁer coating high temperature synthesized NWs with Au. We speculate that the
source of this background comes from the Au catalyst for NW growth. The Au coating
breaks down into isolated islands to act as nucleation sites for NW growth in the VLS
mechanism. A portion of these gold particles ends up at the tip of the NWs while some
are broken off during the process. Yet another part may stay on the substrate and it is the

combination of these isolated Au particles that are giving rise to a broad background. The

main difference from the high temperature growth lies in the amount of GeO; deposited

on the substrate. The high temperature grth has a large quantity of GeOz deposited at a

211

high rate and the Au is buried under it. This gives strong GeOz spectra but weak
scattering from the Au and thus a relatively small broad background signal is measured.
Conversely, low temperature growth has little GeO; to give a strong Raman peak but
relatively large Au content and a correspondingly larger broad background signal is
measured. As mentioned before, the low temperature short NWs growth is accompanied
by isolated island of long NW as was shown in Figure 5-10. The Raman spectra from the
uniform short NWs and those in the white island region are shown in Figure 5-43. Here,
we clearly see the main Raman peak of Ge02 is more prominent but background is
substantially reduced for high density of NWs in the white island region. This directly

supports that the greater amount of NWs is responsible for the stronger spectra and the

------- Bare NWs
—White island NWs

   
 

 

       

 

460 600 800 who who 1460 1500

Wavenumber/cm'1

Figure 5-43. Raman spectra for two different morphologies of NWs.
Insert shows SEM image showing circular islands on uniform background.

212

relatively less Au compared to 660; results in a smaller broad background signal.

Nevertheless the Raman spectrum main peak at 444 cm'1 is evidence of trigonal GeOz.

5.5.2 Gold coating of GeO; nanowires

The as prepared nanowire substrates are not suitable for SERS since only noble
metals are known to yield strong enhancement. The substrates were made SERS active by
coating with thin ﬁlm of gold that allows excitation of surface plasmons. The coated
substrates were characterized by Raman system before application of analyte to establish
background signals. Figure 5-44 shows the observed Raman spectra of 12 nm gold-
coated nanowires with 15 s of integration time. The Raman spectrum after Au coating
shows an increase in the intensity of the broad continuum over all wavenumbers peaking
at about 800 cm’l. It is clear from Figure 5-44 that the gold coating is causing an increase
in the intensity of the broad background. Such a broad continuum is commonly observed
in SERS studies but the source of broad background is not exactly known. Some
researchers attribute it to electronic Raman scattering from metal[116, 117] while others
consider it luminescence[118]. It is a common observation that the molecular SERS
spectrum is accompanied by a broad background covering most of the normal 0-3000
cm'l range[73]. In our case, the broad background appears following the metal coating of
nanowire substrate. Since the background continuum is present even before the
application of analyte molecule, the most probable cause of the continuum is the electron
scattering off defects from the rough gold ﬁlm covering the nanowires. This allows for

electronic Raman scattering over a broad energy range[39]. Further understanding of the

213

broad continuum is of interest, however for the purpose of detection measurements it can

be removed through a baseline correction algorithm.

 

8000-
------ Bare NWs
7000+ — 12 nm Au coated
:9.
g 6000
E
s
.é‘
(D
C
33
E
C
(U
E
(U
n:

 

 

400 600 360 1300 12100 14100 1600
Wavenumber/cm'1

Figure 5-44. Raman spectra of bare and 12 nm Au coated GeOz NWs,
tacc=15 3.

Similar broad continuum was observed for different thickness of Au ﬁlm with

more or less identical shape of spectrum. For example, Figure 5-45 shows the Raman
spectra from GeOz NWs having an average diameter of 108 nm after coating of 12 nm,

23 nm, and 35 nm thick Au ﬁlm and similar broad background appears after coating. The
introduction of continuum in the Raman spectrum is not fully understood yet and needs
further investigation. It usually arises due to ﬂuorescence but it is known that

ﬂuorescence is quenched in the vicinity of metal. The observed continuum is similar to

214

ﬁorescence and is introduced by the metal or the deposition process. The investigation of
Raman spectra of ﬂat silicon wafer with Au ﬁhn thickness has revealed a decreasing
trend of Raman intensity with increasing ﬁlm thickness as expected from theory.
However, it is postulated that the broad background appearing in SERS active substrates
is characteristic of the rough metal ﬁlm, which is inherently composed of discontinuous
islands as revealed by TEM study of Au coated NWs. This can be used to explain the
intensity trend seen in Figure 5-45. As the ﬁlm thickness increases, we see an initial
increase in the background intensity but then it starts to decrease for thicker ﬁlms. This
can be attributed to the Oswald ripening of the individual isolated island towards a more
continuous ﬁlm. Similar trend can be seen in case of a different substrate that was

synthesized with 120 nm average diameter NWs as shown in Figure 5-46.

8000

‘ ,_ 12 nm Au
—23 nm Au
7000+ -------- 35 nm Au

 

.60001

5000 l

.5
§

Raman lntensuty/Arbitr Units
8
O
‘?

E

 

 

 

400 600 800 1000 1200 1400 1600
Wavenumber/cm'1

Figure 5-45. Raman spectra of Au coated GeOz NWs of average diameter
108 nm, tacc=15 s.

215

- 12 nm Au
—23 nm Au
700°” —-----35 nm Au

Raman Intensity/Arbitr. Units

 

 

._.L___«-___P

“—400 600 800 1000 1200 1:00 1600
Wavenumber/cm'1

 

 

Figure 5-46. Raman spectra of Au coated GeO; NWs of average diameter
120 nm, tacc=15 s.

It is also useful to look at the effect of application of analyte on the substrate.
This is especially critical with nanostructures since the drying of the analyte containing
solvent exerts strong capillary forces that can deform the nanowire structures. Figure
5-47 shows the SEM image of the substrate before and after application of analyte. The
substrate was soaked in 1mM methanolic 4-MBT solution for 12 hr and allowed to dry
before imaging. It is clearly seen that the nanowires tend to bundle together after
application of analyte. This is due to the capillary forces generated as the liquid dries and
has been observed in the past as well. For example, Chen et al. reported clustering of InP

nanowires in the form of shrubs after application and drying of analyte solution[143].

216

 

 

Figure 5-47. SEM images of substrate before (left) and after (right) 12 hr
soak in 1 mM methanolic MBT.

5.6 SERS evaluation of substrates

SERS performance of the synthesized NWs was evaluated using commonly
employed thiols analytes. SERS analyte molecules 4-methylbenzenthiol (purity 98%) and
1,2-benzenedithiol (purity 96%) were purchased from Sigma-Aldrich and used without
further puriﬁcation. Figure 5-48 shows the molecular structure of analyte molecules used
in this study. The diluted solutions with various concentrations were prepared in

methanol (99.8% HPLC grade) and stored in glass vials.

CH,
b \ \
/ //
SH
SH SH
4-methylbenzenethiol 1 ,2-benzenedithiol

Figure 5-48. Chemical structures of 4—MBT and 1,2-BDT used as Raman
analytes.

217

 

(J!

112

V:
p."

(‘J
(J

l’.
I

a:
I

1

I

SEE

5.6.1 Detection of 4-MBT
The ﬁrst analyte for SERS evaluation of low temperature synthesized GeOz

nanowires was 4-MBT. The aromatic thiols are known to form self-assembled
monolayers (SAMs) on gold, silver, and copper[125]. It has been reported that MBT
forms a self assembled monolayer on gold after 12 hr soak in 1 mM or higher
concentration[127]. The 23 nm gold coated GeO; nanowire substrate was soaked in
1 mM methanolic 4-MBT solution for 12 hrs. After the 4-MBT soak, the substrate was
thoroughly rinsed with methanol to remove multilayer and/or crystals that might have
formed[127]. The substrate was dried in the stream of nitrogen before acquiring the
SERS spectra shown in Figure 5-49. An interesting effect is the suppression of the
continuous background signal following the formation of the 4-MBT monolayer. Figure
5-50 shows Raman spectra from ﬁve consecutive 250 um spaced points on the substrate
and the characteristic peaks of 4-MBT are easily identiﬁable. No discernible Raman
signal was found from a Si(100) substrate with 21 nm smooth Au ﬁlm that was soaked in
4-MBT alongside nanowires substrate, conﬁrming the enhancement is indeed from

nanowire substrate.

218

x10

~~ 23 nm Au coated
—MBT SAM

I"
cn

N

 

Raman Intensity/Arbitr. Units
_. or

.0
or

 

 

 

400 600 800 10100 1200 1&0 1600 1800
Wavenumber/cm'1

Figure 5-49. SERS spectra of SAM of 4-MBT on 108 mm dia NWs,

x10

00
1

N
1

 

 

 

 

f
t

 

Raman lntensity/Arbitr. Units

win

1 000

     

- 1
Sample pomt Wavenumber/cm'1

Figure 5-50. Successive spectra from substrate points spaced 250 um
along a line, tacc=15 s..

219

 

Having shown good SERS activity of 660; NWs, the next step is to determine

the optimal thickness of the Au coating for SERS. Since gold ﬁlm provides the surface
plasmon active surface, its thickness is expected to have a signiﬁcant impact on SERS
from the substrate. To investigate the effect of gold thickness, a systematic study of the
Raman enhancement of 4-MBT analyte as a function of Au ﬁlm thickness was carried
out. The substrates having NWs with mean diameter of 108 nm were coated with
different thickness of Au ﬁlm in the range 12 - 55 nm. These substrates were then soaked
in 1 mM methanolic 4—MBT solution for 12 hrs, rinsed in methanol and blown dry to
form SAM. Figure 5-51 and Figure 5-52 shows the SERS spectra for different thickness
of Au. The variation of the 1068 cm'l Raman peak of 4-MBT with varying gold ﬁlm
thickness for 108 nm diameter NWs is shown in Figure 553. As can be seen, the best
enhancement is found for coating thickness of 30 — 45 nm. The representative SERS
spectra for different ﬁlm thickness are plotted in Figure 5-54 and clearly show the
dependence of SERS on ﬁlm thickness. Similar dependence is observed for different

diameters of NW5 as shown in Figure 5-55. Here, the Au ﬁlm thickness dependence of

1068 cm'1 Raman band of 4-MBT is plotted for different diameters of GeOz NWs.

220

 

x10

4
—0nm
3.5 —12 nm
—23 nm

00

.N
or

.L
0'1

Raman intensity/Arbitr. Units
—b N

.0
on

 

 

 

400 600 800 1000 1200 14100 1600
Wavenumber/cm’1

Figure 5-51. Effect of Au coating thickness on SERS activity of substrate
having 108 nm mean diameter NW3.

4

4x10
- - 35 nm
3.5- —45 nm
—55 nm
3. ~ - Si(100)

N
on

 

u—l
9'

Raman Intensity/Arbitr. Units
_. N

.0
or

 

 

0 4 __..-_._. -

.— L...“

400 600 800 1000 1200 1400 1000
Wavenumber/cm'1

 

 

Figure 5-52. Effect of Au coating thickness on SERS activity of substrate
having 108 nm mean diameter NWs. Also shown is spectra (along the x-
axis) from a ﬂat Si(100) with 55 nm Au coating that was soaked in 1mM
MBT along with the NWs.

221

x10

 

A
l
1
.1

9°
01

l”
or on
X
\
\
xx

’0‘

.3
(’1

Raman lntensity/Arbitr. Units
—It N
' I

.0'”
”0'0”.”
0

.0
.0"
\

 

 

 

0 110 20 310 4‘0 510 60
Au thickness (nm)

Figure 5-53. Variation of 1068 cm'1 Raman peak intensity of 4-MBT with
Au ﬁlm thickness.

x10

.5
4

(a)
1

 

 

 

a.
‘1‘;

<gaman Intensity/Arbitr. Units
N

CO
v

  

1 500
1 000

500

Au thickness (nm) 0 Wavenumber/cm'1

Figure 5-54. SERS spectra of SAM of 4-MBT for different Au ﬁlm
thickness on GeOz NWs with 108 nm mean diameter.

222

 

x10

"OM79 nm
_ —108 nm

A

9"
or

 

0)

.N
or

N
r

.4
01
r

 

1068'1 Raman band intensity (Arbit. Units)

 

L J

010 1 5 20 25 30 35 40 45 50
Au ﬁlm thickness (nm)

A Jm J

Figure 5—55. Variation of Raman intensity with Au ﬁlm thickness for
different diameters of NWs.

» The length of nanowires is another tunable parameter in low temperature
synthesis technique. The length dependence of SERS was studied by synthesizing GeO;

NWs using 14 nm catalyst ﬁlm for different lengths of time. The mean diameter of the
nanowires was found to be ~108 nm whereas average lengths varied from few hundred
nm to 10 microns depending upon grth duration. The nanowires substrates were
coated with ~40 nm thick Au ﬁlm and SAM of 4-MBT was formed by soaking in 1 mM
methanolic solution for 12 hr. The relative Raman enhancement from different lengths of

nanowires was estimated by comparing the Raman intensity of 1068 cm'1 peak. Figure
5-56 shows the variation of the Raman enhancement for different length of GeO;

nanowires. The maximum enhancement was obtained for nanowires that had average
lengths of 700-1000 nm corresponding to 10-15 min of growth time. For longer and

223

 

r)

shorter nanowires, the enhancement sharply decreases. It may be noted that the peak
enhancement lengths are comparable to the wavelength of exciting laser. An accurate

analysis of the enhancement is not feasible due to extremely high complexity of the
nanowire substrates resulting from composite nature (Au and GeOz), non-conformal gold

coating and large number of interacting nanowires.

 

 

 

 

 

6x 10 1 T

E

S

=3 5” i
e

3

Q 4 r 1
8

.9.

.E

2 3 ‘
m

.o

C -
m ﬁv
E 2 -
(U

(I

5 1 _

w

co

2

00 2000 4000 6000 8000 10000

NWs length (nm)

Figure 5-56. Variation of Raman EF with length of GeOz NWs coated
with 40 nm thick Au ﬁlm.

The homogeneity of the substrate has been a major issue in SERS research. The
reproducibility of SERS measurements from different locations on the substrate was
evaluated by acquiring Raman spectra from 100 points covering an area of
2.5 mm x 2.5 mm. Each sample spot was illuminated for 5 s and spectra were collected.
The substrate was then moved in x or y-direction by 250 um and process was repeated to

224

 

cover the target square area. The 1068 cm'1 Raman band of these 100 spectra is shown in
Figure 5-57 to give a spatial map of enhancement from the substrate. It can be seen that
the substrate provides unambiguous and uniform SERS spectra for a monolayer of
analyte over the tested area with a standard deviation of approximately 3 % from the

mean.

mean=2.93E4

4 std dev=824
x 10

 

 

_. Raman Intensity/Arbitr. Units
N

 

1 ‘V \ ”—7;

0;: \" V 7 ’

0 \\ \gﬁ \ . r

g m //// 10
5 \ \ //5/
\_
v/

y-position (x250 11m) 0 0 x- position(x250 um)

Figure 5-57. Spatial map of SERS intensity for 1068 cm’1 peak of 4-MBT.

The experimental Raman enhancement factor (EF) for our substrate was

calculated using the relationship

EF =L9£RA_N’£f_ (5,7)
Iref NSERS

225

where [SERS is the Raman intensity of some speciﬁc band from the analyte adsorbed on

SERS active substrate and 1 ref is the Raman intensity of the same band from the bulk

analyte. NSERS and Nref are the number of molecules that yield [SERS and [ref

respectively. Thus equation (5.7) gives the Raman enhancement per molecule for a
speciﬁc vibrational band of the analyte. The speciﬁc band is usually the one that gives the
strongest Raman scattering and its intensity is directly read from SERS spectra. An
obvious choice therefore in our case is the 1068 cm”1 Raman band and we utilize the
average intensity from 100 sample points in equation (5.7). Taking 4.5 ><10l4
molecules/cm2 for a monolayer of 4-MBT on gold[130], we ﬁnd approximately

5.87 x 10'‘4 moles (3.53 x 10'0 molecules) of 4-MBT are exited in the laser spot (100 um

diameter). The determination of Nref is more challenging and different methodologies

have been reported to estimate it. Ideally the reference 1 ref and Nref may be obtained

by recording Raman signal from a ﬂat metal coated part of the same substrate but it is
hardly possible to get ordinary Raman spectra for a monolayer of analyte. Even such an
arrangement does not yield true ordinary spectra due to presence of metal in the vicinity
of analyte[54]. The commonly employed techniques to circumvent this problem are to
use solid bulk form or higher concentration solution of analyte to get the approximate
values of reference parameters. Both these techniques need an estimate of laser
interaction volume and that can be calculated from depth of ﬁeld considerations. We did
not detect any ordinary Raman spectra from ~20 mM 4-MBT methanolic bulk solution.

Therefore, we use solid form of 4-MBT to get values of [ref and Nref- Figure 5-58

226

lil~

shows the ordinary Raman spectrum of 4-MBT. The number of molecules excited in the
solid phase was determined by assuming the excitation volume to be a cylinder of
diameter equal to laser spot size and the length of interaction equal to depth of ﬁeld. This

assumption underestimates the N ref since the actual interaction volume has the shape of

a two back-to-back truncated cones with waist diameter equal to the focused laser beam
spot size. It is expected to yield a more conservative estimate of EF. The depth of ﬁeld
was determined by moving the laser focus point from under to over-focus and noting the
1/2 power points. Taking the monolayer thickness of 4-MBT as 0.5 nm [127], we get a

quick estimate of Nref / NSERS ratio. The estimated enhancement factor was found to be

5.86 x106. It may be noted that we have used a ﬂat area approximation while
determining NSERS in EF calculations. The actual or effective area of the nanowires is
somewhat increased (~ 57% for half coated NW) due to their cylindrical nature.
However, the effect is not expected to be signiﬁcant due to line of sight deposition of Au
that only coats part of the nanowire facing the evaporation source.

Another approach to calculated Iref and Nref is based on using polystyrene as

reference[127]. This method utilizes the fact that the Raman cross-section of polystyrene
can be assumed to be equal to that of 4-MBT. Ordinary Raman spectrum of polystyrene
is shown in Figure 5-58. Following a similar line of calculations, the EF was estimated
to be 3.2 x 106. This EF is in good agreement with that calculated using bulk solid form
of 4-MBT and shows that polystyrene can also be used as reference for 4-MBT

enhancement factor calculations.

227

 

14000 r
------ polystyrene

-- 4-MBT
12000 _

' 10000»
8000+

6000 -

 

Raman lntensity/Arbitr. Units

 

2000 ~

. l 1
- nip ‘ .
-. _.-.. AIL , ‘. .LV 1 1 1 mM‘W‘L‘tG ' aw»—

400 600 800 1000 _ 1200 1400 1600 1800
Wavenumber/cm'1

4000- i ‘

 

 

Figure 5-58. Ordinary Raman spectra of 4-MBT and polystyrene.

5.6.2 Detection of 1,2-BDT

The generality of nanowires based substrate for SERS was evaluated by probing a
1,2-BDT as a SERS analyte. This molecule is known to form self assembled monolayer
(SAM) on Au for concentration 2 l x 10'3 M after more than 20 min of soaking[126]. The
SERS substrate with 35 nm thick Au ﬁlm was immersed in 1 mM methanolic solution of
1,2-BDT for 20 min. This concentration is the least reported for monolayer coverage and
is expected to yield a more conservative estimate of enhancement. The sample was then
rinsed with methanol to ensure removal of excess analyte. The substrate was allowed to
dry before acquiring Raman spectra. Figure 5-59 is a plot of SERS spectra from Au

coated GeOz NWs, which clearly shows the 1,2-BDT peaks. The broad continuum is also

somewhat suppressed following the formation of BDT SAM on the Au coated nanowires.

228

 

5110;

Figu

Strong SERS spectra from different randomly chosen points on substrate are shown in

Figure 5-60.

x10

~ - before
— after

N (a) #

Raman Intensity/Arbitr. Units

—L
I

 

 

 

400 600 800 10100 1200 1400 1600
Wavenumber/cm'1

Figure 5-59. SERS spectra of SAM of 1,2-BDT on 108 nm diameter
NWs, tacc = 15 s.

4

 

x 10

U)
E 5.
:3
£5 4.
a
$ 3‘
5 ~ 1

2 l
g 1.111
a gx 1'

4\\3\ 1500
2 1000
500
Sample point 1

Wavenumber/cm'1

Figure 5-60. SERS spectra of SAM of 1,2-BDT from different points on
substrate. '

229

 

 

Figure 5-61 shows the plot of the 1028 cml peak of 1,2—BDT for 100 points
scanned in a rectangular area of 2.5 mm x 2.5 mm. The data show the substrate provides

very reproducible SERS enhancement with small standard deviation (~ 8%).

mean=1.5E4
std dev=1.2E3

 

Raman lntensity/Arbitr. Units

 

 

\\)<’//
y-position(x250 pm) 0 ° x-position(x250 um)

Figure 5-61. Spatial map of SERS intensity for 1028 cm'1 peak of 1,2-
BDT, tacc=5 s.

The SERS EF for 1,2-BDT can also be calculated using equation (5.7) and this
time the 1028 cm'l Raman band of 1,2-BDT is selected for estimation. Assuming
monolayer coverage of substrate with 1,2-BDT and each molecule binding to 2 Au atoms
on surface, the aerial density of analyte is estimated at 6.4 x 1014 molecules/cm2 on the
substrate. For a laser spot size of 100 um diameter, nearly 8.35 x 10''4 moles (5.03 x 10l0
molecules) of 1,2-BDT contribute to the SERS signal. The ordinary Raman spectra of the

230

 

 

source 1,2-BDT was used as reference to estimate [ref and N,.ef for use in EF
calculations. Although I ref can be directly measured from ordinary Raman spectra of
bulk analyte, the estimation of Nref required knowledge of laser exaction volume. This

excitation volume was estimated to be ~6.3 nL from depth of ﬁeld considerations. The
average enhancement factor from equation (5.7) was found to be ~ 7.4 x 105. The low
enhancement factor relative to other analytes is probably due to sub-monolayer coverage
of 1,2-BDT on the substrate. Another factor that is peculiar to 1,2-BDT is the storage of
analyte solution. It was observed that the 1,2-BDT glass vial develops some sort of
coating on the walls. It is probably due to sticking of 1,2-BDT on the glass and the
coating did not dissolve back into solution even after ultrasonication. The sticking of 1,2-
BDT in the form of a coating is also likely to decrease the concentration of analyte from
the initially prepared 1 mM concentration. This decrease in the concentration can also
cause sub-monolayer coverage and associated decrease in measured EF.

The above results have shown that the gold coated nanowires provides strong
SERS enhancement to detect monolayer converge of thiol molecules. The SAM of the
same analytes on smooth gold ﬁlm over silicon wafer did not yield any discemable
spectra. It is important to highlight the contributions of nanowires towards SERS. It is
envisaged that the nanowires have a trifold effect towards SERS. First these provide a
rough 3-D surface to host the metal coating that is a prerequisite for exciting surface
plasmons. Secondly the poor wetting of gold on the oxide nanowires yields island like
growth of the metal ﬁlm on individual nanowires with closely spaced metal nano-
particles. The closely spaced gold islands on the nanowire surface provide strong electric

ﬁeld enhancement in the gaps between adjacent islands. Third contribution is derived

231

from the optical properties of nanowires and is a major factor in improving repeatability
of SERS measurement. It is hypothesize that the high dielectric nature of oxide NWs
helps to conﬁne and guide the light to a much larger area on the substrate than directly
illuminated by the laser beam. This is supported by the observation that the laser spot is
hardly noticeable in CCD camera when focused on silicon wafer. However, the spot size
grows to a much larger areas when the NW substrate is illuminated suggesting light
waveguiding and coupling among NWs. Each NW is coated with a large number of gold
particles that are also expected to immobilized analyte molecules. The light coupling
between NWs brings the excited surface plasmons to these metal particles, which in turn
excites Raman modes of analyte. The analyte Raman radiation is again ampliﬁed by the
Au nanoparticles and coupled back to the NWs where they travel to the receiver in

reverse direction.

5.7 Conclusion

In this chapter, a novel low temperature synthesis of germanium oxide nanowires
using thermal evaporation of metallic germanium is presented. This technique offered
good control over diameter and length of nanowires. Nanowire yield was found to be
highly sensitive to temperature, oxygen concentration and background gas pressure. The
nanowire substrates were coated with thin ﬁlm of gold to make these SERS active. The
gold coating introduced a broad background in the Raman spectrum of nanowires. The
performance of substrates was evaluated by using 4-MBT and 1,2-BDT as probe
molecules. Optimal SERS enhancement was obtained for a gold coating thickness of

30 ~ 45 nm. The substrates showed an average enhancement factor of the order of 106

232

 

with good repeatability and small standard deviation over the tested areas. The amount of
analyte that yields these strong SERS spectra is on the order of ~10'14 moles and conﬁrms

the high sensitivity of these novel substrates.

233

 

Chapter 6: Zinc Oxide Nanowires based SERS Substrates

Zinc oxide (ZnO) is a versatile technological material that exhibits both
semiconducting and piezoelectric properties. Its strong piezoelectric and pyroelectric
properties have allowed it to be extensively used in mechanical actuators and
piezoelectric sensors. It is transparent to visible light and is suitable for short wavelength
optoelectronic applications owing to its wide band gap. ZnO grows in a wide variety of
morphologies such as nanowires, nanocombs, nanoshells, nanocages, nanobelts,
nanosprings etc[l44]. The ZnO was mainly synthesized as nanowires for this research.
The synthesis of ZnO nano-structures was done using a simple vapor transport and
annealing mechanism. The substrates were characterization with SEM (JEOL 6400).
Rhodamine 6G (R6G), Nile Blue, 4-methylbenzenthiol, and 1,2-benzenedithiol were
utilized as probe molecules for SERS analysis. In this chapter, details of nanowires

synthesis and the results of Raman analysis are presented.

6.1 Zinc oxide nanowires synthesis

Typical synthesis of ZnO nanowires is done using solid-vapor process in which
source material is evaporated at elevated temperature and the resultant phase condenses
under appropriate conditions of temperature, pressure, gas environment to form desired
nanostructures. Huang et al. were the ﬁrst to synthesize crystalline ZnO NW5 by vapor
transport via VLS mechanism[145, 146]. Nanowires were grown on 5 nm Au-coated Si
by heating a 1:1 mixture of ZnO and graphite powder to 900-925 °C under constant ﬂow

of Ar for 5-30 min. Since then there have been a number of reports of ZnO NWS

234

 

synthesized by variety of techniques including physical vapor deposition[147-l49], laser
ablation[lSO], catalyst-free metal oxide vapor phase epitaxy (MOVPE)[151, 152],
template assisted growth using electrodeposition[153], aqueous solution-based
synthesis[154-156] and chemical vapor deposition (CVD)[157].

Zinc oxide nanowires were synthesized on Si(100) and quartz substrate using
thermal evaporation and transport of Zn source in an oxidizing environment. The
substrates were cleaned by sonication for 15 minutes each in acetone, methanol and D/I
water and ﬁnally dried by blowing compressed nitrogen. Generally thin gold ﬁlm
(3-20 nm) was used as catalyst but some growth was also observed on uncoated part of
substrate. In a typical set up, the substrates were placed on top of a quartz or ceramic
boat containing zinc powder. The distance between source powder and substrates varied
from 1-3 cm depending upon type of boat used. In certain instances, the zinc powder was
mixed with sodium chloride to control evaporation[158]. The source boat alongwith the
substrates was then placed on a quartz support plate that was inserted into the heating
zone of the horizontal tube furnace as shown in Figure 6-1. The tube was evacuated to
less than 300 mTorr and Argon mixed with 5% hydrogen at a ﬂow rate of 50 sccm was
introduced into the tube as a background gas. The reaction gas of Argon-oxygen mixture
(40% oxygen balance argon) at 503ccm was introduced using a separate MFC. The
introduction of gas ﬂow increased the pressure inside the tube to about 550 mTorr. The
synthesis pressure of ~2 Torr was adjusted by partially closing a vacuum valve located at
the downstream end of the tube. After adjusting gas ﬂow rates and pressure, temperature
of the furnace was raised to 650 °C over 10-12 min. The control temperature was

measured by a ‘K’ type thermocouple placed between walls of quartz tube and tube

235

 

furnace. The actual temperature inside the quartz tube at the location of source/substrate
was measured by a second ‘K’ type thermocouple (not shown) placed inside the tube.
During growth the, the pressure of the reaction chamber increased to ~25 Torr. The
grth took about 1-2 hr and subsequently the fumace was allowed to cool to room
temperature. The source powder turned white from original gray color and the substrate
surface appeared light to dark gray after the experiment. The morphology of the source
powder and substrates was examined by SEM and compositional analysis was done by
EDS. Figure 6—2 shows the SEM image of the ZnO NW5 synthesized on Si substrate.
Figure 6-3 shows the diameter distribution of ZnO NWs. The nanowires have a mean
diameter of 62 nm with small standard deviation of 14 rim. Very dense and aligned
nanowires were also found at the location of source as shown in Figure 6-4. However,
these source NWs were not further investigated since present study was mainly aimed at
SERS from NW5 synthesized on ﬂat substrates. The ﬂat substrates have the advantage of
one time focusing of laser beam with predictable and stable metal coating. Nevertheless
such nanowires may be useful in SERS from metallic colloidal by providing 3-D holding

framework for nanoparticles[ 1 43].

236

 

Au coated substrate
Hollow cylindrical furnace

 

 

 

 

 

 

Zn source
inside boat _:
\ f ‘a’
: v' :
Ar H I: l '\ 5W :
i a g :1 _5
Ar + 02 —‘> :1 E I; to vacuum
4 1' '7' (1:2—:‘2; pump
Quartz tube ,1'
thermocouple

Quartz support plate

Figure 6-1. Experimental set up for ZnO nanowire synthesis.

 

Figure 6-2. Zinc oxide nanowires on Si 100 substrate using Au as catalyst

237

 

 

   
   

Number of NWs
co

 

36 44 52 60 68 76
Diameter of ZnO NWs (nm)

Figure 6—3. Diameter distribution of ZnO NWs.

 

mean= 62

std_dev= 14

84 92

100

 

Figure 6—4. ZnO nanowires on source powder.

238

 

 

It may be mentioned that NWs were also found in locations where no catalyst was
present as shown in Figure 6—5. This observation conﬁrms the existence of growth
mechanism other than the typical VLS mechanism. The composition of the nanowires
was conﬁrmed to be of zinc oxide ﬁom EDS analysis as shown in Figure 6-6. The EDS
quantization results yield atomic ratio of Zn:0 is 120.9 and conﬁrm the product is
stoichiometric ZnO within the accuracy of measurement. The high percentage of gold is
due to sample coating that has to be done to make the sample conductive for SEM

studies.

 

 

Figure 6—5. ZnO NWs found in uncoated area of Si wafer.

239

Zn

 

 

 

0 Au
Au Au
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
keV
Element App Intensity Weight% Weight% Atomic%
Conc. Corrn. Sigma

OK 3.17 1.1516 115.12 1.22 45.33
Zn K 9.63 0.7622 69.32 1.83 50.88
Au M 1.86 0.6557 15.56 1.81 3.79
Totals 100.00

 

 

Figure 6-6. EDS spectrum and quantization results for ZnO nanowires.

The NW5 were not the only nanostructures obtained in the synthesis. Some
nanocombs[159], nanorods[160], nanowhiskers[161], nanocages[l62], nanoﬂowers[l60]
and hexagonal faceted nanorods[154] were also found as shown in Figure 6-7. Generally
speaking, the substrate side facing the source (facedown side covering the source boat)
shows greater variation of morphologies. This is probably due to high saturation of
vapors and convoluted ﬂow of source vapors inside the quartz boat. The substrate face
outside the boat have more uniform growth of NWs since gas ﬂow is unrestricted and

vapors diffuse away without any restriction.

240

 

 

Figure 6-7. Different morphologies of ZnO nanostructures. (a)
Nanocombs. (b) Nanorods. (c) Nanowhisker. (d) Nanocages. (e) Flowers.
(1) Hexagonal rods.

The most common morphologies of ZnO synthesized in this research were the

nanowires and hexagonal nanorods. The other morphologies were mostly found on the

241

top of these two structures at isolated locations. Therefore, next section will discuss the

possible mechanism leading to the growth of these nanostructures.

6.2 ZnO nanowires growth model

The NWs synthesis has been reported using either catalyst assisted or catalyst free
environment. However, in this case NWs are observed in both areas of substrate - with
and without catalyst. The catalyst assisted growth occurs via VLS mechanism and this
growth mechanism has been widely used for semiconductor nanowire growth. In this
mechanism, the catalyst forms a eutectic with the NW material and the resulting liquid
droplet serves as a preferential site for absorption of gas phase reactant. Once
supersaturated with the reactant, the liquid droplet acts as nucleation site for
crystallization and growth of nanowire. The growth continues so long as the supply of
reactants is available and temperature is high enough to maintain the catalyst in liquid
phase. The VLS mechanism is therefore characterized by strong localization of growth to
the catalyst coated area and presence of solid catalyst on the tip of the nanowires. The
oxide NW growth through the VLS mechanism is somewhat complicated by the presence
of oxygen. Most oxides have very high melting points and are unlikely to be in liquid
phase at growth temperatures commonly employed. Hence either pure metallic source is
used in an oxidizing environment or some form of reducing agent is utilized when using
oxide source material. For example, mixing of ZnO with carbon or graphite is a

commonly employed technique to synthesize ZnO nanowires[146, 163, 164].

242

 

In this study, pure zinc source was used as the starting material. Zinc is highly
volatile (melting point ~419.52 °C) and at high grth temperature, it can vaporize

according to
Zn(s) —> Zn(g) (6.1)

The zinc vapors then diffuse away from the source and condense on substrate (Au
coated or uncoated) where temperature is less than boiling point (907 °C) of Zn [165].
These liquid droplets of Zn then form ideal nuclei for the self-catalytic growth of ZnO
nanostructures via VLS mechanism[81]. The grth of NWs on the catalyst free area of
substrate also supports the existence of a self-catalytic growth. The presence of oxygen in
the reaction makes the situation more complex and is expected to result in oxidation of

zinc according to reaction
Zn(s) + g 02 —> Zn0(s) (6.2)

Both of the above reactions are expected to be occurring simultaneously. The
ZnO is very stable (melting point 1975 0C) at reaction temperature and is not expected to
yield any signiﬁcant vapor pressure at the synthesis temperature used in this work.
However, the limited supply of oxygen in the reactor (the substrate is covering the top of

the boat containing source material) may result in substantial amount of zinc suboxides

(ZnOx, 0<x <1)[165]. It is known that the suboxides of Zn have low melting points

(~419 °C) and can vaporize and condense on the substrate yielding nucleation sites[166].

243

 

These suboxide droplets can enhance the adsorption and difﬁrsion of Zn oxides at the tips
of the ZnO NWs. Further oxidation of the suboxide droplets leads to the formation of
ZnO NWs. Yet another possible pathway for the growth of NW is from the direct

deposition of islands ZnO on the substrate following oxidation of Zn or

ZnOx (0 < x <1) vapors. It has been previously reported that ZnO islands act as preferred

sites for the growth of NWs[167]. Some evidence of this can be seen by probing substrate
under the top layer of NWs as shown in Figure 6-8. Here, the top layer of NWs is
scratched to reveal existence of relatively large (~ 200-300 nm) crystallite structures of
ZnO. The small diameter NWs appear to be growing between these and lend credibility
to hypothesis that a ZnO seed layer might be forming ﬁrst. Then the small diameter NWs
starts to grow as the supply of Zn is exhausted. The precise mechanism for the growth is
net yet clear but all three mechanisms discussed above are expected to occur
simultaneously with different degrees of contributions towards ﬁnal product. It was also
observed experimentally that the gold—coated substrates gave better yield of NWs
compared to uncoated substrates. The catalyst-coated area typically shows high yield of
the NWs under same conditions of synthesis as can be seen in Figure 6-9. This can be
explained by better sticking of source vapors to catalyst-coated area that results in better

yield.

244

 

Figure 6-9. NWs on catalyst coated (left) and uncoated (right) substrates.

245

The growth process can be summarized as:

(a) Formation of zinc vapors and zinc suboxides upon heating of zinc source
in limited supply of oxygen.

(b) Transport of vapors generated in above step to substrate.

(0) Condensation of vapor on substrate to form seed layer. Better sticking and
condensation for gold coated substrate.

((1) Growth of ZnO nanowires from these seeds (and gold catalyst if present).

It may be re-iterated that some different morphologies were also observed during
synthesis of nanowires. Since bulk yield in this study was of the form of nanowires and,
no attempt is made to explain detailed kinetics of formation for other morphologies. In
fact, detailed kinetics of the synthesis of the different morphologies of ZnO are not
exactly known. These have been attributed to different synthesis conditions like oxygen
concentration, temperature, pressure etc. Hu et al.[161] had reported formation of
nanoneedle-microwhisker formation on silicon without any metal catalyst by thermal
evaporation of ZnS powder at 900-950 °C in continuous ﬂow of 100 sccm Ar containing
2 % oxygen. They attributed the initial hexagonal microwhisker due to high
supersaturation of ZnO vapors during initial phase of reaction. The nanoneedle is formed
due to reduction in supersaturation of ZnO as supply of ZnS is reduced favoring growth
of nanoneedle. Leung et al.[163] reported formation of ZnO tetrapods and bone like
nanorods by controlling Zn vapor release from source. Wang et al.[166] observed that the

formation of ZnO tetrapods was extremely sensitive to oxygen concentration in the

246

reactor. The formation of tetrapods is also attributed to oxidation of Zn vapor during gas
phase transport[l68, 169]. Yao et al.[165] reported formation of different types of
nanostructures at different temperatures confirming sensitivity to temperature. It is
appropriate to conclude that these low yield isolated morphologies of ZnO are highly
sensitive to experimental conditions and are synthesized only at localized spots where
suitable conditions exist. A more uniform and reproducible substrate is prerequisite for
SERS and only nanowires were found to satisfy this requirement.

Before proceeding to the Raman studies of ZnO substrates, it is important to
highlight an observation concerning growth of NWs on gold coated substrate at low
temperature. In the relatively low temperature region.(less than 600 °C), the ZnO
nanorods are strongly localized to location of catalyst. Figure 6-10 shows the ZnO
growth on silicon substrate located in the low temperature region of the ﬁrrnace. It can be
clearly seen that the growth is strongly limited to the gold-coated area of the substrate.
The main difference besides low temperature is also the proximity of the substrate from
the source that will reduce the supply of the reactants to this region. So it is reasonable to
speculate that under low supersaturation (due to greater distance from source), there is
less probability of forming a ZnO seed layer. Hence grth is more governed by the

typical VLS growth mechanism of crystal growth.

247

6.3

  
    

m'

   

Zn: 17.45 %

0 17 64<y Si: 92.14%
$1555.01 °Af Au: 7.86 %
Au: 9.9 %

Figure 6-10. ZnO nanostructures in low temperature region on gold
coated silicon.

Raman characterization of ZnO nanowires

The ZnO NWs were characterized by EzRaman-L from Enwave Optronics Inc.

with 6 cm'1 resolution at 785 nm excitation. The sample was mounted on a xyz-stage and

illuminated by the Raman system laser beam via optical ﬁber link. The Raman probe

head was attached to a separate z-stage that facilitates precise focusing of laser beam onto

the sample surface independent of sample movement. All spectra were acquired with a

focusing lens (NA. = 0.22) that yields a spot size of less than 100 um diameter and a

power of ~30 mW on the substrate.

248

6.3.1 Raman Characterization of NWs

The Raman spectra of ZnO nanowires synthesized on alumina substrate is shown
in Figure 6-11 and shows the characteristic Raman band at 437 cm'l. The position of the
main peak at 437 cm'I agrees well with that reported previously with an underlying rising
background[170]. It may be mentioned that the Raman signal from most of the substrates
was very weak and ZnO peak was hardly noticeable. This is probably due to very thin
layer of ZnO nanowires on the substrate as evidenced by strong Raman peak of

underlying Si substrate.

3500 -

___A|203 substrate
------ Si substrate

1

3000

2500

2000

 

 

1 500

1000

Raman intensity (Arbit. units)

 

500

 

 

300 400 500 600 700 800
Wavenumber/cm”1

Figure 6-11. Raman spectra of Zinc oxide nanowires on alumina and
silicon substrates showing characteristic peak at 437cm'1, tacc=20 s.

249

6.3.2 Gold coating of ZnO nanowires
The ZnO substrates were coated with thin ﬁlm of gold to create a plasmon active

surface. Similar to the observation of slowly varying background in case of gold-coated
GeO; nanowires, a broad continuum was also observed for ZnO nanowires. For example,

Figure 6-12 shows Raman spectrum of ZnO nanowire substrate after coating with 21mm
gold ﬁlm. The broad continuum peaking between 800-1000 cm'1 is appearing after gold
coating of substrate. The silicon 521 cm”1 peak intensity from silicon substrate is reduced
as a result of metal coating. It may be noted that the substrate peak is quite strong and
this is primarily due to very thin top layer of NWs as well as high transparency of ZnO
for IR radiation. The ﬂuorescence peak from substrate for high wavenumber is also
somewhat quenched after gold coating. This is expected since metals are known to help
reduce ﬂuorescence[73]. Figure 6-13 shows the Raman spectra for ZnO nanocombs and
the continuous background is quite prominent. It peaks in the same general region of the
wavenumber and is usually intense compared to the underlying Raman peak of substrate.
It is believed that the metal is enhancing the electric ﬁeld due to surface plasmon
excitation and subsequently light gets trapped in the nanostructures to give rise to such
broad peak at shifted wavelengths. However, this hypothesis needs further investigation
and the efﬁcacy of the ZnO nanowires in detection of R6G and NB probe molecules is

discussed next.

250

14000 - mm Uncoated NWs
—Au coated NWs

    

12000 -

10000 -

8000 ~

6000 '

4000 -

Raman Intensity/Arbitr. Units

2000 *

 

 

,MJ “J.

J- .J a. i 1 41
200 400 600 800 1000 1200 1400 1600
Wavenumber/cm'1

Figure 6-12. Raman spectra of 21mm gold coated ZnO nanowires on Si

100 substrate, tacc=15 5. Insert shows the SEM image of substrate.

4000

3500

units)
(A)
O
O
O

2500

N
O
O
O

1500

 

Raman intensity (Arblt

.L
o
O
O

01
O
O

 

 

400 600 800 1000 1200 1400 1600 1800 2000 2200
Wavenumber/cm’1

Figure 6-13. Raman spectra of ZnO combs on SilOO, tacc=15 5. Insert
shows SEM image of the substrate.

251

6.4 SERS study of ZnO nanowires substrate

The initial SERS studies were done using Rhodamine 6G (R6G) and Nile Blue as
probe molecules. The substrates were coated with gold or decorated with gold particles to
make them SERS active. However, problem of analyte adsorption on substrate lead to the
use of thiol based analytes like 4-methylbenzenthiol, and 1,2-benzenedithiol for detailed

SERS analysis.

6.5 Detection of R6G probe molecule

R6G was the ﬁrst probe molecule investigated using the ZnO nanowires. A 3 uM
solution of R6G in 95% ethanol was prepared to act as SERS probe. This concentration
of R6G is very low for ordinary Raman detection. The ordinary Raman spectra of 3 uM
ethanolic solution of R6G is shown in Figure 6-14 and only ethanol peaks are visible.
Then a drop of 3 uM R6G was applied to the substrate and allowed to dry before
acquiring Raman spectra. Figure 6-15 shows the Raman spectra obtained from 21 nm
gold-coated ZnO nanowires on Si(100) substrate. All the major peaks of NE are clearly
identiﬁable on the top of a broad background that is characteristic of gold-coated
nanowires substrate. However, the uniformity of the substrate to yield detectable spectra

was found to be low and only few spots produces strong spectra.

252

rb
00
o
o
o

 

 

 

500 r i
0‘ 400 600 800 1000 1200 1400 1600 1800
Wavenumber/cm'1

Figure 6-14. Ordinary Raman spectra of 3 uM ethanolic solution of R6G
showing only the ethanol peaks.

Raman intensity (Arbit. units)

 

 

400 600 800 1000 1200 1400 1600 1800 2000 2200
Wavenumber/cm'1

Figure 6-15. Raman spectra of 3 uM R6G from 21 nm Au coated ZnO
nanowires synthesized on Si(100) substrate.

253

6.6 Detection of nile blue

Nile blue A is a common dye used in biology and ﬂuorescence studies. The probe
solution was prepared by dissolving nile blue in ethanol and then diluting the stock
solution to get a concentration of 2.73 uM. This concentration of analyte is so small that
the bulk solution on yield ethanol peaks in the ordinary Raman spectra. First the Raman
spectra from the uncoated ZnO substrate was recorded as shown in Figure 6-16 (a). Then
a drop of probe ethanolic NB was added to this uncoated ZnO NW substrate and allowed
to dry. The Raman spectra after application of 2.73 uM nile blue did not show any
detectable NB peaks as given in Figure 6-16 (b). This conﬁrmed that gold coating is
essential for SERS from these oxide nanowires. Subsequently a drop (about 0.25 ml) of
60 nm gold colloidal was added to the substrate and the Raman spectra were acquired
again. The substrate yielded a clear Raman spectra of nile blue after application of gold
colloidal as shown in Figure 6—17. Here, all major peaks of NB are clearly identiﬁed with
very little background. This conﬁrms that the gold metal in the form of thin ﬁlm or
nanoparticles is necessary for achieving strong Raman enhancement. However, it was
found that certain spots of the substrate do yield a broad background and the probe
molecule spectra rides on the top of such continuum. The issue of hot spots was also

noticed but as a whole enhancement was found from many locations.

254

(a) 900»

800 ~

Raman intensity (Arbit. units)

200
1 00

(b)
900

800
700
600
500
400
300

Raman intensity (Arbit. units)

200
100

700»
600L
500r
400
300

l

 

 

 

 

 

400 600 800 1000 1200 1400 1600“_'
Wavenumber/cm'1

   

 

400 500 800 1 000 1200 1400 1500
Waven umber/cm'1

Figure 6-16. Raman spectra of ZnO nanowire. (a) Before adding NB. (b)
After adding NB.

255

-----Ioc 1
12000 -
— loo 2

1 0000 ~

6000' qumva

4000M

2000 ~

 

Raman lntensity/Arbitr. Units

 

 

400 600 800 1000 1200 1400 1600
Wavenumber/cm'1

Figure 6-17. Raman spectra of 2.73 uM ethanolic NB from ZnO
nanowires coated with 60 nm gold colloidal at two locations, tacc=15 s.

The SERS evaluation of substrates using R6G and NB as probe molecules yielded
strong enhancement to detect uMolar concentrations of analytes. However, the
enhancement was localized to few isolated spots on the substrates. It was postulated that
the problem of hot spots is due to selective adsorption of analyte on the substrate i.e., the
analyte only adsorbs to few isolated locations on substrate that form hot spots. In order to

study this hypothesis, a different class of analytes was investigated.

6.7 Detection using thiols
Thiols are commonly employed analytes in SERS studies due to their tendency to
form SAM on noble metals. The two analytes investigated with ZnO NWs are the 4-

methylbenzenthiol (4-MBT) and 1,2-benzenedithiol (1,2-BDT).

256

6.7.1 Gold coating thickness study

The substrates were coated with different thickness of metal to determine the
optimal thickness for SERS. In order to minimize experimental error due to uncontrolled.
experimental variations, same substrate was coated with different thickness of gold ﬁlm
on different parts. This ensures that any substrate variations from one experiment to
another do not interfere with the systematic gold coating studies. The gold coated
substrates were soaked in appropriate concentration of probe molecule for set amount of
time to yield a SAM of analyte. Figure 6-18 shows the SERS spectra of 1,2-BDT for
different thickness of gold ﬁlm on ZnO NWs. All spectra are acquired under identical
illumination conditions of laser power, exposure time, laser spot diameter etc.

The increasing of SERS with increasing thickness prompted trying even thicker
coating and a second piece of same substrate was coated with thicker gold ﬁlms in
different parts. The probe molecule SAM was formed on this substrate following exactly
the same steps as before. The SERS spectra were again acquired under identical
conditions and are shown in Figure 6-19 for ﬁlm thickness of 28-56 nm. It is
immediately clear that the SERS intensity increases for increasing gold ﬁlm thickness
upto ~40 nm and then starts to decrease for thicker ﬁlms. It is envisaged that the thinner
ﬁlms do not yield closely spaced metal islands on oxide NW5. This results in reduced
coupling between particles and a low SERS enhancement. An optimal is reached for ~40
nm thick ﬁlm where best coupling exists between nanowires and metal particles. A
further increase in ﬁlm thickness results in continuous ﬁlm and the enhancement due to

inter-particle coupling is lost besides.

257

7000 f ' I T l I

 

 

 

 

—6 nm Au
_ ---- 14 nm Au ]

0,6000 -~21nmAu
.t
C
? 5000 -
E
E 4000 ~
.3:
2
9 3000 ~ ,
E.
s .1
g 2000 '- 5“ ‘3" "w”.a-frw’; 1"?! r; i 1* ‘i
a: i, W“ “a. 1M

1OOOLJ’~‘«1,W.»~’M “M’Amﬁf‘i‘ ', “4

0

 

400 600 800 1000 12100 1400 16100
Wavenumber/cm’1

Figure 6-18. Effect of gold coating thickness of SERS from SAM of 1,2-
BDT on ZnO NWs, tacc=15 s.

 

 

x 10
2.5
i —28 nm Au
] ----42 nm Au
,2 27 I 56 nm Au
= . i
s 1 *"
it 1.5
.E‘
(I)
5 1
g 1
c .
m
g .
tr 0.5

      

L

400 600 800 1000 12100 1400 1600

Wavenumber/cm'1

Figure 6—19. Effect of gold coating thickness of SERS ﬁom SAM of BDT
on ZnO NWs, tacc=15 s.

258

In order to ensure that the ﬁlm thickness results obtained above are not limited to
a single analyte, same study was repeated for a different analyte. This time 4-MBT was
used as analyte and its SAM was formed following previously reported procedure[127].
The SERS spectra were acquired for similar conditions for all different gold thickness.
Remarkably similar trend of SERS enhancement versus gold thick was obtained with 4-
MBT probe molecule as shown in Figure 6-20 and Figure 6-21. It is clearly obvious that
the Raman enhancement increases monotonically with increasing gold thickness upto
~40 nm. Then the enhancement start to decrease for thicker ﬁlms. The broad continuum
also shows a similar trend but tends to reach saturation for ﬁlms thicker than 40 nm. It is

known that the thin metal ﬁlms do not yield conformal coating of oxide nanostructures.
Gold coating of GeOz nanowires and TiOz at room temperature has shown a 3D island

growth[120, 171]. It is therefore hypothesized that the very thin ﬁlm yields very small
islands on nanowires and these are far apart to yield any appreciable enhancement. With
thicker ﬁlms, the size of islands increases and so is their mutual surface plasmon
interaction leading to strong Raman enhancement[80, 172]. However, beyond a certain
critical value gold coating starts to take a more continuous form on the nanowires that
decreases the enhancement. Another simultaneous factor could be the dielectric metal
coupling that is reduced for thicker ﬁlms. The experimental results suggest a thickness of

the order of ~40 nm yields the best enhancement.

259

14000 1 . .'- . .

 

 

 

—6 nm Au
---~14 nm Au
”12000? - 21 nm Au i
.‘E
3_10000~
£-
2 8000»
e 1
8 1
g 6000 1
E .
g 1
E 4000'
(U
a:

 

      

 

400 600 800 1000 1200 1400 1600
Wavenumber/cm'1

Figure 6-20. Effect of gold coating thickness of SERS from SAM of 4-
MBT on ZnO NWs, tacc=15 s.

3; 10
—28 nm Au
------42 nm Au
2.5 ----- 56 nm Au

N

 

Raman lntensity/Arbitr. Units
- or

.0
on

 

 

l

400 600 300 1000 12100 14100 1600

Wavenumber/cm“1

Figure 6-21. Effect of gold coating thickness of SERS from SAM of MBT
on ZnO NWs, tacc=15 s.

260

The effect of application of analyte on ZnO nanowire substrate was also studied.
Figure 6—22 shows the SEM image of substrate before and after soaking in 1 mM
methanolic MBT for 12 hr. It is evident that there is no discemable change in the
morphology of the substrate after application of analyte. This shows that the ZnO
nanowires are robust against morphological changes unlike the G602 nanowires that

exhibited signiﬁcant bundling and clustering.

 

Figure 6-22. ZnO NWs substrate. (a) Before 4-MBT soak. (b) After 12 hr
soak in 1 mM MBT.

6.7.2 SERS for 4-MBT

The ZnO nanowires substrate was coated with ~40 nm gold and incubated in 1
mM methanolic 4-MBT solution for 12 hrs. This concentration and soak time is known to
yield a SAM of MBT on gold[127]. After the 4-MBT soak, the substrate was thoroughly
rinsed with methanol to remove multilayers and/or crystals that might have formed[127].
This ensures that the SERS signal recorded is actually from a SAM of MBT and yields a
more accurate estimation of enhancement factor. The substrate was allowed to dry before

acquiring the SERS spectra shown in Figure 6-23. This ﬁgure shows the Raman spectra
261

from ﬁve consecutive 250 um spaced points on the substrate and the characteristic peaks
of 4—MBT are easily identiﬁable. An interesting effect is the suppression of continuous
background of raw substrate following the formation of the 4—MBT monolayer. It may be
mentioned that no discemable Raman signal was detected from a Si(100) substrate
having a 40 nm smooth Au ﬁlm that was soaked in 4—MBT alongside nanowires
substrate, conﬁrming the enhancement was indeed from nanowire substrate. Also the

bulk solution of 1 mM methanolic 4-MBT only yielded spectra for the methanol solvent.

3.5\

I"
01 O)
/ /

 

 

Raman lntensity/Arbitr. Units
0 .11
a A a N
I l g

 

\

 

OI

 

 

   
 

1500

   

1000

Sample point 1

Wavenumber/cm'1

Figure 6-23. SERS spectra of a SAM of 4-MBT from 5 adjacent points

250 um apart, tam: 15 5.

Another issue related to the SERS research is the repeatability of the enhancement
over spatial area of substrate. To verify the reproducibility of SERS measurement from

substrate, SERS spectra were acquired from 100 points covering an area of

262

2.5 mm x 2.5 mm. The 1068 cm'l Raman band of these spectra is then plotted in Figure
6-24 to give a spatial map of enhancement from the substrate. It can be seen that the
substrate provides unambiguous and uniform SERS spectra for a monolayer of analyte

. . . . std. d v' t' n
over the tested area wrth a coefﬁcrent of variation (ﬂ—i x100) of ~ 3.5%.
mean

12000\ '
10000~ - ' ‘
8000

6000... -

 

4000 \ -

2000 V

Raman Intensity/Arbitr. Units

 

 

y-position (x250 pm) x- position(x250 pm)

Figure 6-24. 4-MBT 1068 cm'1 Raman peak intensity distribution for
randomly chosen area on substrate, tam= 5 s.

The experimental Raman enhancement factor (EF) for the ZnO substrate was

calculated using the relation[129]

EF=ISEAM (6.3)
1 REF NSERS

263

where [SERS is the Raman intensity of some speciﬁc band from the analyte

adsorbed on SERS active substrate and [ref is the Raman intensity of the same band
from the bulk analyte. N SERS and Nref are the number of molecules that yield I SERS

and 1 ref respectively. Thus equation (6.3) gives the Raman enhancement per molecule

for a speciﬁc vibrational band of analyte. The speciﬁc band is usually the one that gives
the strongest Raman scattering and its intensity is directly read from SERS spectra. The
highest Raman intensity for MBT is obtained for the 1068 cm'1 Raman band and the
average intensity from 100 sample points of this band is used in equation (6.3). Taking
4.5 x 10I4 molecules/cm2 for a monolayer of 4-MBT on gold[130], we ﬁnd

approximately 5.87 x 10'14 moles (3.53 x 1010 molecules) of 4-MBT are excited in the

laser spot (100 um diameter). The determination of Nref is more challenging and
different methodologies have been reported to estimate it. Ideally the reference I ref and

Nmf may be obtained by recording Raman signal from a ﬂat metal coated part of the

same substrate but it is hardly possible to get ordinary Raman spectra for a monolayer of
analyte. Even such an arrangement does not yield true ordinary spectra due to presence of
metal in the vicinity of analyte[54]. The commonly employed techniques to circumvent
this problem are to use solid bulk form or higher concentration solution of analyte to get
the approximate values of reference parameters. Both these techniques need an estimate
of laser interaction volume and that can be calculated from depth of ﬁeld considerations.
We did not detect any ordinary Raman spectra from ~20 mM 4-MBT methanolic bulk

solution. Therefore, we use solid crystalline form of 4-MBT to get values of [ref and

264

Nref' Figure 6-25 shows the ordinary Raman spectrum of 4-MBT. The number of

molecules excited in the solid phase was determined by assuming the excitation volume

to be a cylinder of diameter equal to laser spot size and length equal to depth of ﬁeld.

This assumption underestimates the Nref since the actual interaction volume has the

shape of a two back-to-back truncated cones with waist diameter equal to the focused
laser beam spot size. It is expected to yield a more conservative estimate of EF. The
depth of ﬁeld was determined by moving the laser focus point from under to over—focus
and noting the 1/2 power points. The DFOV was estimated to be ~842 pm for 1090 cm'1

peak but a more conservative estimate of 800 um was used in EF calculations. Taking the

monolayer thickness of 4-MBT as 0.5 nm[127], we get a quick estimate of Nref /N SER S

ratio. The estimated enhancement factor was found to be 2.19 x 106. It may be noted that
we have used ﬂat area approximation while determining N SERS in EF calculations. The
actual or effective area of the nanowires is somewhat increased (~ 57% for half coated
NW) due to their cylindrical nature. However, the effect is not expected to be signiﬁcant

due to line of sight deposition of Au that only coats part of the nanowire facing the

evaporation source.

265

1 4000

1

~-------- polystyrene
— 4-MBT

I

12000

10000 -

8000

1

6000 -

 

El

4000 ~

Raman lntensity/Arbitr. Units

 

2000

T

 

 

. l:
ONT-"'1"h ‘ “‘ L‘Mﬂ‘t‘am.b~ wt»

400 600 i 800 1000 1200 1400 1600 1800
Wavenumber/cm'1

Figure 6-25. Ordinary Raman spectra of 4-MBT and polystyrene,
ta“: 5 s.

We also calculated [reef and Nref by using polystyrene as reference whose

Raman cross-section can be assumed to be equal to that of 4-MBT[127]. Ordinary Raman
spectrum of polystyrene is shown in. Following a similar line of calculations, the EF was
estimated to be 1.2 x 106. This EF is in good agreement with that calculated using bulk
solid form of 4-MBT and shows that polystyrene can also be used as reference for 4-
MBT enhancement factor calculations.

The SERS enhancement discussed above are for the ZnO NWs synthesized on 7
nm Au coated silicon substrate that was placed facing down on the quartz boat containing
source zinc powder. As discussed previously, these nanowires are randomly oriented as

shown in Figure 6-26(a). The nanowires obtained on the substrate area outside the quartz

266

boat were more aligned as shown in Figure 6-26(b). The exact mechanism of this
alignment is not clear at this stage. It has been attributed to the natural alignment of ZnO
seed nanocrystals on the substrate and subsequent growth of nanowires from these
seeds[l67]. In the present case, the ﬂow pattern of source vapors and degree of
supersaturation appears to be affecting the alignment of nanowires i.e, conﬁned ﬂow with
high supersaturation yields random nanowires. The SERS signal from these partially
aligned NWs was also compared and is shown in Figure 6-27. It is clearly evident that
intensity of SERS signal for randomly oriented W5 is more than twice from partially

aligned NWs.

 

Figure 6-26. SEM image of two typical morphologies of ZnO NWs. (a)
Randomly oriented. (b) Partially aligned. Scale bar is 2 pm in both
images.

267

 

 

4 _
----- random NWs
3.5 , —oriented NWs
(I)
a 3-
D it
a 2-5
SE
b it
.5
C .-
2
E
C
(U
E
(U
0:

 

 

 

400 600 800 1000 1200 1400 1600 1800
Wavenumber/cm'1

Figure 6—27. Comparison of SERS signal from random and partially
aligned ZnO NWs for SAM of 4-MBT, tacc=15 s.

The uniformity of SERS signal for both types of NWs was comparable as shown
in Figure 6-28 where SERS spectra from ﬁve random sample points for each type of
W3 is shown. One obvious explanation for higher SERS from randomly oriented NWs
is their high density as can be seen in SEM image. Also it has been reported that the NWs
crossing yield stronger SERS[173] and randomly orientations has higher number of such
crossings. This can be more easily seen in cross-sectional image of substrate shown in
Figure 6—29. The randomly distributed nanowires have large number of crossings

compared to partially aligned nanowires that exhibit few crossings.

268

A
m
V

A

(.0

_L
/

 

Raman lntensity/Arbitr. Units
N

010
\

 

Wavenumber/cm"1

Raman lntensity/Arbitr. Units
N
/

 

   

  

 

1.
0..
5
3’ ‘\ 2//"T1soo
2 \\ ./”// 1000
\r" 500
Sample point

Wavenumber/cm'1

Figure 6-28. SERS from SAM of 4-MBT for (a) randomly distributed
NWs, tacc=15 s. (b) Partially aligned NWs, tacc=15 s.

269

 

Figure 6-29. Cross-section for ZnO nanowires. (a) Randomly distributed.
(b) Partially aligned.

Another morphology that yielded strong SERS was ZnO nanoparticles that were
formed on gold—coated substrate in the low temperature region (~450 0C). Here the
substrate gets a coating of densely packed nanometric ZnO particles as shown in Figure
6-30. The EDS data analysis shows nearly stoichiometric ZnO (23.37 % Zn, 21.49 % 02)
that is thin enough to pass electron beam resulting in strong signal from underlying
silicon (36.14 % Si, and 19.0% Au). These densely packed ZnO nanoparticles coated
with ~42 nm gold also yielded strong SERS signal for SAM of MBT as shown in Figure
6-30. The strong enhancement from these particles is probably due to there close

proximity that allows for inter-particle couple of plasmonic ﬁeld between particles[39].

270

x10

.5
/

(A)
/

—L
I

Raman Intensity/Arbitr. Units
N
/

 

V

010

 

1 000

500

- 1
Sample pomt Wavenumber/cm'1

Figure 6-30. ZnO nanoparticles in low temperature region of reactor and
SERS spectra of SAM of MBT after gold coating of particles, tacc=5 5.

6.7.3 SERS for 1,2-BDT

To conﬁrm the enhancement for other molecules, detection of 1,2-BDT as a probe
molecule was also investigated using gold-coated ZnO NW substrate. This molecule is
known to form self assembled monolayer (SAM) on Au for concentration 2 l x 10'3 M
after more than 20 min of soaking[126]. We immersed our SERS-active nanowires
substrates in 1 mM solution of 1,2-BDT in methanol for 20 min. This concentration is the

smallest that was reported for monolayer coverage and is expected to yield a more

271

conservative estimate of enhancement. The sample was then rinsed with methanol to
ensure removal of any excess analyte and allowed to dry before acquiring Rarnan spectra.
Figure 6-31 shows the SERS spectra from 42 nm Au coated ZnO NW8, which clearly
shows the 1,2-BDT peaks. The broad continuum is also somewhat suppressed following
the formation of BDT SAM on the Au coated nanowires. Figure 6-32 shows successive
spectra from substrate collected from points spaced 250 um apart and shows good

reproducibility of measurement.

 

 

 

 

 

 

4
2.53‘ 10
------before
—after
.9 2—
'E
3
LE
g 1.5
9?;
U)
15
g 1*
C
(D
E
0: 0.5— 1
EM
OMA’J .: V. m. *2. ..

 

400 600 800 10100 12100 14100 1600 1800
Wavenumber/cm'1

Figure 6-31. SERS spectra of self—assembled monolayer of 1,2-BDT from

Au coated ZnO nanowires, taco= 15 8. Also shown is spectra before analyte
application.

272

x10

 

£2.5\
C
2 2.
.2
E15.
’3 11
c 1‘ .
§°-5~ 1 dllllll 1
6:“ g..\ dim“
,\ //
4 \3 \ rig//"//(1500
é‘\.- /«// 1000
Sample point 1 i 500 Wavenumber/cm"1

Figure 6-32. Successive SERS spectra from 1mM BDT along a line on
substrate, tacc=15 s .

   

 

 

 

 

mean=7855
std dev=340
8400
8200
.3 8000
5
1.; 8000
E 6000
E. ‘ 7800
.3
5 4000
g . . 7600
C 7.1
£2000 . -- , "7"”
g ‘ 7 J g 7400
0>\ i; (5:7,! '- xii-7 :
10 ‘\.- / 720°
8 ‘~ \\\ - ~. ' /\
5 ‘1 / 8 7000
6
6800

y-position (x250 11m)

  

x- position(x250 11m)

Figure 6-33. Area map of 1028 cm"1 Raman peak of 1,2-BDT, tac¢=5 s for

each point.

273

Figure 6-33 shows the plot of the 1028 cm1 peak of 1,2-BDT for 100 points
scanned in a rectangular area of 2.5 mm x 2.5 mm. The data conﬁrms that the substrate
provides very uniform SERS enhancement with small standard deviation (~ 4.3% from
mean). The SERS EF for 1,2-BDT is again calculated using equation (6.3) and this time
the 1028 cm'l Raman band of 1,2-BDT is utilized. Assuming monolayer coverage of
substrate with 1,2-BDT and each molecule binding to 2 Au atoms on surface, we get
6.4 x 1014 molecules/cm2 on the substrate. Considering the laser spot size of 100 pm
diameter, nearly 8.35 x 10'14 moles (5.04 x 1010 molecules) of 1,2-BDT contribute to the
SERS signal. The ordinary Raman spectra of the source 1,2-BDT was used as reference
to calculate Raman EF. The laser excitation volume was estimated to be ~6.3 nL from
depth of ﬁeld considerations. The average enhancement factor from equation (6.3) was
found to be ~ 4.1 x 105. The low enhancement factor relative to 4-MBT is probably due
to sub-monolayer coverage of 1,2-BDT on the substrate. A thin ﬁlm was found to form
on the glass vial holding the analyte suggesting some sort of reaction or deposition of
analyte on walls. The coating did not dissolve back into solvent even after sonication.
This may cause lowering of analyte concentration from that of initially prepared yielding

sub-monolayer analyte coverage and under-estimation of enhancement factor.

6.8 Conclusion
The detailed synthesis of ZnO nanowires using a simple thermal evaporation set

up is presented in this chapter. The NWs synthesis is likely due to a combination of self-
catalytic growth and VLS mechanism. Like the case of GeO; substrate presented earlier,

the gold coating was found to introduce a broad background in the Raman spectrum of

274

nanowires. The ZnO substrates were evaluated using R6G and NB as probe molecules.
Strong enhancement was found at isolated locations, which was sufﬁcient to easily detect
uMolar concentrations of the two probe molecules. On the other hand, no discemable
Raman spectra could be obtained from bulk solutions at these concentrations. This clearly
conﬁrms the SERS from nanowires but at few isolated locations. The isolated
enhancement was attributed to the nature of analytes that don’t chemically bind to the
substrate. This hypothesis was tested by using thiol based probe molecules of 4-MBT and
1,2-BDT. These aromatic thiol based analytes are known to chemically bind with the
metal coating to form SAM on the substrate. The average Raman enhancement factor for
4-MBT and 1,2-BDT were estimated to the order of 106 and 105 respectively. The
repeatability of SERS measurement was tested by recording spectra from 100 sample
points spanning 2.5X2.5 mm2 area of the substrate. A very uniform and repeatable SERS
spectra was found for both types of analytes with small standard deviation of less than
5% over the tested areas. These results demonstrate that the gold-coated ZnO NW

substrates have the potential to provide a very uniform SERS enhancement.

275

Chapter 7: Metallic Nanorods based SERS Substrates

Silver, gold and copper are the common metals for the plasmonic nanostructures
due to their ability to support surface plasmons in the visible region of electromagnetic
spectrum. Exploring the metallic nanostructures made up of these metals for SERS is a
natural extension of this property. The different metallic nanostructures or more
appropriately nanorods were synthesized using a grazing angle deposition technique also
known as “Oblique angle deposition”. The substrates were characterized using SEM and
EDS. SERS studies were conducted using nile blue, 4-methylbenzenthiol, and 1,2-

benzenedithiol as probe molecules.

7.1 Oblique angle deposition

Oblique Angle Deposition (OAD) is a technique to create porous ﬁlms by
utilizing tilt of substrate with respect to incident ﬂux direction. The oblique incidence of
the ﬂux creates atomic shadowing from already deposited islands and this promotes
inclined columnar structures under conditions of limited adatom diffusion. Such features
of deposited ﬁlms were ﬁrst observed in 1959 while studying magnetic anisotropy of iron
ﬁlms and optical activity of ﬂuorite ﬁlms deposited under oblique angle of incident
ﬂux[174-l76]. The technique had been known and utilized previously for enhancing
contrast of weakly reﬂecting surfaces and to emphasize surface roughness by decoration
in electron microscopy[l77]. The technique is quite general and various sources
including e-beam, thermal, laser, and sputter deposition can be used. The substrate can be

ﬁxed or rotated at different speeds during rotation (phased rotation) to yield desired level

276

of isotropy. Substrate can be ﬂat or templated to grow ordered arrays. The technique is
quite robust and practically any material or combination of materials may be synthesized.
Lu et al. have synthesized Si springs and tungsten nanorods on ﬂat and templated
substrates[l78]. Suzuki et al. demonstrated SERS from silver and gold coated arrays of

SiOz nanorods prepared by a variation of CAD technique[l79, 180]. More recently there

have been number of reports of good SERS enhancement by using pure silver nanorods
synthesized by OAD[36, 37, 181]. In this work metallic nanorods of copper, aluminum

and silver were synthesized for use as SERS substrate.

7.2 Metallic nanorods synthesis

The metallic nanorods were synthesized utilizing the oblique angle deposition
technique that yields a porous columnar morphology on substrate [36, 174, 175, 182,
183]. The glass slides were ﬁrst cleaned using ultrasonication in acetone, followed by
methanol and ﬁnally in D/I water for 15 min each. Then the slides were blown dry using
compressed nitrogen. The slides were mounted on the deposition stage of PVD chamber
and ﬁrst a thin layer (~50 rim) of metal (aluminum, copper or silver) was deposited at
normal incidence using standard e-beam evaporation. The base pressure at the start of
deposition was around 3><10’6 Torr and substrate was continuously rotated at ~ 6 rpm
during deposition. The deposited ﬁlm thickness was monitored using quartz crystal
monitor (Inﬁcon XTM/2) operating at 6MHz. The substrate was heated from rear to
85253 °C during deposition using two quartz lamps housed under the substrate holder.

The substrate was then tilted (82 :t 3° with respect to source ﬂux direction) so the

incident ﬂux arrived at a grazing angle and another 400-900 nm of metal was deposited

277

without rotating the substrate to get different length of nanorods. The actual deposited
ﬁlm at grazing angle is expected to be somewhat less than this thickness since quartz
crystal monitor measures the normal incidence ﬂux. In case of aluminum nanorods, ﬁrst
we synthesized the desired length of nanorods and then coated only the tips of nanorods
with ~10 nm of silver by switching the source vapor ﬂux from aluminum to silver. The
substrate angle change was achieved by tilting the substrate assembly with respect to the
ﬁxed incident vapor ﬂux from e—bearn source. Figure 7-l(a) shows the schematic of the
CAD set up utilized to synthesize the nanorods. It is worth mentioning that the substrates
were having a mirror ﬁnish following initial normal deposition. However, the surface
ﬁnished changed to a dull appearance after about 200 nm metal at grazing angle
deposition as can be seen from optical images of silver and copper nanorods on glass

substrates in Figure 7-l(b) and (c) respectively.

(a)

   

: < 15°
Continuous Substrate
rotation positioned for
2; Glass OAD deposition
misubstrate (no rotation)
A
% Evaporation E%
source
Normal deposition Oblique angle deposition

Figure 7-1. (a) CAD experimental set up schematic. (b) Optical image of
Silver nanorods substrate (c) Optical image of copper nanorods substrate.

278

Figure 7-2 shows the SEM (accelerating voltage 15kV, 15mm working distance)
image of the substrates with dense and well-aligned nanorods of different metals. As can
be seen, the incidence ﬂux was arriving from the bottom of the substrate and the
shadowing effect is clearly visible in the form of black V-shape behind the white
particulate towards the bottom-center of the image in Figure 7-2 (0). It is envisaged that a
combination of shadowing effect and high silver mobility even at low temperature is
causing the incident flux to self-coalesce into small nanorods[40]. The aluminum
nanorods show many sharp facets whereas copper and silver rods are mostly cylindrical.
Typical length, diameter, and density of aluminum nanorods was found to be 652i96 rim,
217i66 nm, and 5-6 /um2 respectively from SEM studies. The silver nanorods were
found to be more dense (10-1 1/ umz) with length and diameter averaging 587:1:46 rim and
71i12 nm respectively. The copper nanorods were also dense (12-14/um2) like silver but
exhibit greater bundling perpendicular to the direction of incident ﬂux. The average
length and diameter of copper nanorods was 523:1:58 nm and 64i8 nm respectively. It
may be mentioned that the length of nanorods is the main controllable parameter in this
synthesis technique whereas little or no control exists over density and diameter for the
deposition conditions used in this work. The nanorods in length ranging from 500-700
nm, as shown in Figure 7-2, were found to yield good SERS signal. The diameters and
lengths of the three types of nanorods were measured from SEM analysis. Figure 7-3,
Figure 7-4 and Figure 7-5 show the distribution for diameters and lengths of aluminum,
copper and silver nanorods respectively. The EDS analysis of the aluminum, copper and
silver nanorods substrates is shown in Figure 7-6, Figure 7-7 , and Figure 7-8

respectively. The oxidation of nanorods is more severe for copper and aluminum than

279

silver as is evident from EDS quantization results. The silicon, sodium and calcium form
a part of the underlying substrate glass. The underlying substrate shows up in the EDS
because the electron beam penetrates few micron into the sample whereas the top layer of

silver is only few hundred nm thick.

 

Figure 7-2. SEM images of nanorods synthesized on glass substrate. (a)
Aluminum nanorods. (b) Copper nanorods. (c) Silver nanorods. ((1) Silver
nanorods showing shadowing effect.

280

Aluminum NR diamter distribution
mean= 227

std_dev= 52

14—
12' 6I |
:éoll I II--

190 220 250 280 310 340 370 400
Diameter of Al NRs (nm)

Number of NRs
00 ES

0')
1

A

Aluminum NR length distribution

mean= 670

std_dev= 100

Number of NRs

 

 

200 460 520 580 640 700 760 820 880 940 1000
Length of Al NRs (nm)

Figure 7-3. Aluminum NRs diameter and length distribution.

281

mean= 63

     
  

 

 

16' std_dev= 8
w 12 '
0:
z
“5
0)
Q .
E 8
3
z
4 .
0
40 50 60 70 80 90 100
Diameter of Cu NRs (nm)
mean= 523
16 -
std_dev= 58
w 12 -
0:
z
“5
(D
Q .
E 8
3
z
4 ..
200 450 550 600 650

500
Length of Cu NRs (nm)

Figure 7-4. Copper NRs diameter and length distribution.

282

24-

21 _ mean= 70

std_dev= 12

Number of MRS

 

 

910 50 60 70 80 90 100 110 120
Diameter of Ag NRs (nm)

mean= 587

std_dev= 46

Number of NRs

 

 

O
500 520 540 560 580 600 620 640 660 680
Length of Ag NRs (nm)

Figure 7-5. Silver NRs diameter and length distribution.

283

 

 

 

Al
Si
O Na
Ag
Ca Ca Ca
0 1 2 3 4 5 6
keV
Element App Intensity Weight% Weight% Atomic%
Conc. Corrn. Sigma
0 K 4.38 0.7629 17.18 1.00 27.88
Na K 1.94 1.3700 4.24 0.29 4.79
Al K 19.35 1.0946 52.85 0.9 50.87
Si K 3.10 0.6896 13.45 0.51 12.43
Ca K 0.85 0.9584 2.65 0.29 1.72
Ag L 2.34 0.7273 9.62 0.76 2.31
Totals 100.00

 

 

Figure 7-6. EDS spectra and quantization results for aluminum nanorods
substrate.

284

 

 

 

keV
Element Weight% Atomic%
C K 3.30 8.81
O K 19.87 39.76
Mg K 1.12 1.48
Si K 17.06 19.44
K K 0.58 0.47
Ca K 2.67 2.13
Cu L 55.40 27.91
Totals 100.00

 

 

Figure 7-7. EDS spectra and quantization results for copper nanorods
substrate.

285

 

 

 

 

 

 

 

Si Ag
Ca
0 2 4 6 8 10
keV
Element App Intensity Weight% Weight% Atomic%
Conc. Corrn. Sigma

Na K 0.18 0.8436 2.02 0.66 6.12
Si K 1.65 0.9923 15.37 0.95 38.06
Ca K 0.24 0.9365 2.35 0.65 4.08
Ag L 7.97 0.9202 80.26 1.26 51.74
Totals 100.00

 

 

Figure 7-8. EDS spectra and quantization results for silver nanorods
substrate.

286

 

 

511!

C01

7.3

mm

There does not appear to be any speciﬁc requirement of having glass as the
substrate for nanorods synthesis. The sample holding steel clips were also found to
contain similar nanorods. For example, Figure 7—9 shows the SEM image for one of the

sample clips and the dense and aligned nanorods are covering the entire surface.

 

Figure 7-9. Silver nanorods that grew on sample clips (clips made of
steel).

7.3 Oblique angle deposition kinetics

An excellent review of the different mechanism suggested to explain oblique
angle deposition is given by Abelrnann[183] and will not be repeated here. In summary,
the shadowing effect is advanced as the most plausible argument for the formation of
nanorods. In addition surface diffusion of adatoms also plays an important role. This can
be random or directional due to oblique angle of incident which imparts them a lateral
momentum[183]. For example, silver has high mobility even at room temperature and has
been observed to self-coalesce from smooth ﬁhn into small islands[40]. Other factors

287

such as presence of water or oxides on the surface also affect the process as these change
the diffusion of adatoms[183]. Oxidation usually reduces whereas water increases surface
diffusion of adatoms. So it may be a combination of shadowing and self-coalescing that
is actually driving the growth of nanorods. The nano columns are tilted toward surface
normal which is explained by the lateral momentum of incident ﬂux[184]. Other
operating parameters affecting grth are angle of incident ﬂux, substrate temperature,

and rate of deposition.

7.3.1 Incident flux angle

At increasing oblique angles, the vapor ﬂux encounters enhanced atomic
shadowing. This results in an increased porous structure of isolated grains that are
inclined towards vapor source. This occurs because of atomic shadowing, which creates
areas where incident ﬂux cannot reach due to low adatom mobility for surface diffusion.
Vick et al. reported that angle of greater than 80° is needed for any appreciable
shadowing[185]. Similarly best results for germanium nanorods on Si were obtained for

an angle of 87° with 0.2 — 1.5 A°/sec deposition rate[186].

7.3.2 Temperature

High temperature means greater mobility or surface diffusion of adatoms and this
improves the crystallinity of the deposited material. However, for high temperatures the
column cross-section tends to become elliptical and bundling results. As long as the
increase in temperature is such that the bulk diffusion does not become dominant, the

morphology will be still the columnar.

288

7.3.3 Rate of deposition

Low deposition rate gives the adatoms more time to diffuse on the substrate to
form islands. It also ensures a larger spacing between nuclei in all directions, thus
reducing bundling of growing nanorods. In fact there are two different types of
diffusivities for the adatoms — one over the substrate and another over the growing
ﬁlm[l87]. Better columnar structures are achieved for material having low diffusitivity

over substrate but high over the growing ﬁlm.

7.4 Raman characterization of substrates

The substrates were ﬁrst characterized by acquiring Raman spectra without any
analyte to establish the background reference. Figure 7-10 shows the Raman spectra of
silver nanorods without any probe molecule. The spectra contains broad background that
peaks around 700 cm'l. The exact source of this background is not known but has been
commonly attributed to metal luminescence [188] or electronic Raman scattering from
underlying metal[39, 73]. The spectra were also collected from uncoated part of glass
slide and the shiny smooth silver ﬁlm as shown in Figure 7-11. The smooth and shiny
silver ﬁlm did not yield any discemable spectra whereas broad ﬂuorescence peak can be
seen for glass slide. This ﬂuorescence is strongly diminished for the silver coated part of

the substrate.

289

 

1800-

1600-

1400-

1200' U
WW

1000

800-

600-

Raman Intensity (arbit. units)

200-

 

 

 

400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Wavenumber I cm'1

Figure 7-10. Raman spectra of silver nanorods, tacc = 5 5. Insert shows
SEM image of silver nanorods.

80007
~— Glass
7000 - —Smooth Ag ﬁlm (x10)

0)
O
O
O

01
O
O
0

40001

N 0)
O O
O O
O 0

Raman intensity (arbit. units)

  

1000i

 

0 11.11111‘M1illl . 1 1
500 1 000 1 500 2000
Wavenumber I cm"I

Figure 7-11. Raman spectra of smooth silver ﬁlm and glass slide,

290

 

Figure 7-12 shows the Raman spectra from silver tipped aluminum nanorods and
copper nanorods. Just like case of silver nanorods, both types of substrates are
characterized by broad background. The broad background is most probably due to

electronic Raman scattering from underlying roughness of nanorods.

1200'

—-Al NRs
----Cu NRs

.3
o
o
O

800 -

600 -

 

400 1
l

1111

500 1000 1500 2000
Wavenumber/cm'1

Raman Intensity (Arbitr. Units)

200

 

].
l

   

 

Figure 7-12. Rarnan spectra of copper and silver tipped aluminum
nanorods, tacc 5 s.

7.5 Detection of nile blue

Nile blue was the ﬁrst analyte used for SERS evaluation of metallic nanorods.
The probe solution was prepared by dissolving nile blue in 95% ethanol and then diluting
this stock solution to get a concentration of ~ 5 uM. The samples for SERS measurement
were prepared by applying nearly 3 uL drop of 5 uM solution to the substrate. The
applied volume of analyte immediately spread to nearly 11 mm diameter spot on the

substrate and was allowed to dry before acquisition of spectra. Figure 7-13 shows

291

 

Raman spectra of substrate before and after application of 3 1.1L of 5 uM NB and all
major peaks of NB are clearly distinguishable. The applied analyte volume/concentration
and substrate coverage imply that nearly l.24><lO"5 moles of NB are contributing to
SERS signal. The ordinary Raman spectra from 5 uMolar bulk solution of NB showed
only the peaks for ethanol and did not yield any detectable NB spectra. This clearly
conﬁrms the enhancement from silver nanorods. It may be worth mentioning that no
discemable ordinary Raman spectra could be observed even from highly concentrated

(0.17 M ethanolic) NB solution due to strong ﬂuorescence background as shown in the

insert of Figure 7-13.

 

 

 

 

 

 

 

14000 - background
-------- NB spectrum E60001
12000- 3.
e
A 534000
.53 10000- E
a 5 .
g 3000- g g 2000
s: 11 . - - -
8
2: _ g: 500 1000 1500
.2 6000 :L Raman shift (cm'1)
a r '
g 4000' M “A
2000- W ' "
M- UK...

 

400 600 800 1000 1200 1400 1600
Raman shift/cm'1

Figure 7-13. SERS spectra from Ag nanorods substrates before and after

applying 3 uL of 5 uM ethanolic NB, tacc 15 3. Insert shows ordinary

Raman spectra of 170 mM NB in ethanol with only weak 582 cm'1 peak
identiﬁable.

292

 

Figure 7-14 shows the successive SERS spectra of NB from nine different
randomly selected points on the substrate. The major peaks of NB are clearly identiﬁable
but some variation in signal intensity from different areas of the substrate is noticeable.
Further investigation of substrate showed strong enhancement at most locations but some
variation of Raman signal was observed across the substrate. These can be attributed to
the non-uniform binding of the analyte on the substrate. For a more thorough study,
another type of analyte that are known to strongly bind to the metal like thiols are

investigated in the next section.

5000.
4000 .
e' 3000 s
(0

units)

 

 

N
O
O
O
z

1000. 1 in
O\

lnstensity (

 

500

1 000 6

1500 4 l (50 d)
sam e oint ms ace
Raman shift (cm'1) 2000 2 p p u D

Figure 7-14. SERS spectra from Ag nanorods substrate from different
spots for 5 uM ethanolic NB, taco: 5 s.

293

 

7.6 Detection of MBT

The 4-MBT, being an aromatic thiols, forms self-assembled monolayers (SAMs)
on gold, silver, and copper[125]. The SAM forming analyte allows for easier and more
accurate computation of the number of molecules contributing to SERS, and helps to
obtain a better estimation of the Raman enhancement factor. Also such common probe
molecules allow comparing the performance of the SERS substrates against others
reported in the literature. The SAMs were prepared by soaking the silver tipped
aluminum nanorods in a 100 uM methanolic solution of 4-MBT for 14 hrs. This
concentration of 4-MBT is an order of magnitude smaller than the one reported for
SAM[127, 189, 190] and is expected to yield sub-monolayer coverage. Actual coverage
is expected to be even lower since silver tipped aluminum nanorods have substantial
voids between adjacent rods. After the 4-MBT soak, the substrate was rinsed with
methanol to remove multilayers and/or crystals that might have formed[127]. The
substrate was then dried in the stream of nitrogen before acquiring the SERS spectra.
Figure 7-15 shows Raman spectra from silver tipped aluminum nanorods substrate and
the characteristic peaks of 4-MBT are easily identiﬁable. Also shown is the Raman
spectrum from the substrate alone, which shows a weak background continuum. A
slightly enhanced continuum is also present in the SERS spectrum of 4-MBT monolayer.
It has been almost a universal observation that the molecular SERS spectrum is
accompanied by a broad background covering most of the normal 0-3000 cm'l range[39,
73]. The precise origin of this continuum is still not clear and some researchers attribute
this background to electronic Raman scattering from metal[116, 117] while others

consider it luminescence[118]. However, it can be removed by the baseline correction.

294

 

16000 -

 

—before
m 14000 - --------- after 4-MBT
S 12000—
5 10000-
$
:37; 8000-
C
.‘L’
5 6000-
g 1
g 1m 1
0: ii -" ”N L" " . 1'1
2000' W‘llﬂl

 

 

500 1000 1500
Wavenumber/cm'1

Figure 7-15. Raman spectra from silver tipped aluminum nanorods
before and after application of 100 uM 4-MBT, tam: 15 s.

Figure 7-16 shows the successive spectra from different sample points lying
along a line arbitrarily scanned on the substrate and the characteristic spectrum of 4-MBT

is easily recognizable.

(I)
”E 4000
D
e 3000 .
3
é‘zooo l 1 1H 1
2 1H .
a, V
E 1000 "J“
C
E
a ° - 5
500 4
1000 3
1500 / 2
-1 2000 1
Wavenumber/cm sample point (100nm spaced)

Figure 7-16. MBT SAM spectra before and after soaking aluminum
nanorods in 0.1 mM methanolic for 14 hrs, tacc= 5 s.

295

 

The spatial uniformity and repeatability of SERS measurement from substrate has
been a challenging issue in SERS research. To investigate repeatability of the SERS
signal, we acquired spectra from 100 raster-scanned locations covering an area of
1mm x 1 mm. The 1071 cm‘l Raman peak intensity from each sample point is then
plotted in Figure 7-17 to give a spatial map of enhancement from the substrate. It can be
seen that the substrate provides unambiguous and uniform SERS spectra for a monolayer

of analyte over the tested area with only subtle variations.

mean=2039
std dev=155 2600

2500
2400
$2300
.2200
2100

2000

Raman Intensity/Arbitr. Units

1900

1800

 

 

y—position (xtoorm 0 x- position(x100um)

Figure 7-17. 4-MBT 1071 cm'l Raman peak intensity distribution from
silver tipped aluminum nanorods for randomly chosen area on substrate,

The copper nanorods were also evaluated for SERS using 4-MBT as probe
molecule. The substrate was soaked in 1 mM 4-MBT solution for 15 hrs to form
SAM[190]. After monolayer formation, substrate was rinsed with methanol to remove
any physisorbed molecules and dried with nitrogen. The copper nanorods also yield

296

sufﬁcient enhancement to detect main peaks of 4-MBT as shown in Figure 7-18 but the
signal intensity is low compared to silver tipped aluminum nanorods. This result is not

unexpected given greater SERS activity of silver compared to copper.

 

 

4000 .
a —before
.11 3500" 31 -----arter4-MBT
5 3000» / 1,. 511,.
Q ‘- ~ 11
t r
E 2500* '1’ w \K‘
g ’f ”~1- .
3% 2000» ,I ”'93!
5 "l N
E
g 1000 1" ht‘w‘q
a:
500 M
0 A J

 

400 600 800 1000 1200 1400 1600 1800
Wavenumber/cm"1

Figure 7-18. Raman spectra from copper nanorods before and after
formation of SAM of MBT, tacc= 15 s.

SERS from silver nanorods have been previously reported[36, 191] and hence it is
reasonable to compare the SERS performance of aluminum and copper nanorods to that
of silver. Hence, sliver nanorods were also synthesized using OAD and soaked in
100 uM methanolic MBT to compare their SERS performance with aluminum and
copper nanorods. The silver nanorods also yield strong SERS spectrum of 4-MBT as
shown in Figure 7-19. Quantitatively the intensity of main Raman peak at 1068 cm" is

~6 times stronger for silver nanorods than for aluminum nanorods.

297

 

x10

03

— before
------- after 4-MBT

U!

A

1-1

1 i 1

1i -

- :: ll
5:. 1:, n 3 .1 1
.«1 cw 1"“ \w 161. 11
. I -' ‘.
\' “A.4" ‘W

N

 

Raman lntensity/Arbitr. Units
01)

_I

 

300 400 600 800 1000 1200 1400 1600 1800
Wavenumber/cm'1

Figure 7-19. Raman spectra of from Silver nanorods substrate before and
alter formation of SAM of 4-MBT, tax: 15 s.

mean=1.3E4
x 10“ std dev=4.2E3

   

Figure 7-20. 4-MBT 1068 cm'l Raman peak intensity distribution for a
randomly chosen area on a silver nanorods substrate, tacc= 5 s.

298

 

The spatial reproducibility of SERS signal form silver nanorods is again analyzed
by acquiring spectra from 100 points covering a sample area of 1.4 X125 mmz. The
spatial variation of 1068 cm" peak of 4-MBT from the tested area is shown in Figure
7-20 and conﬁrms good reproducibility of SERS signal.

The experimental Raman enhancement factor (EF) for the nanorods substrate was

estimated using the equation

N
EF ____ [SERS ref (71)
[ref NSERS

where [SERS is the intensity of some speciﬁc Raman band from the analyte
adsorbed on a SERS active substrate and N SERS is the number of molecules contributing

to 15515. Similarly 1 ref is the intensity of the same Raman band from the bulk analyte
and Nref is the number of molecules that yield Iref' Thus equation (7.1) yields the

Raman enhancement per molecule for a speciﬁc vibrational band of analyte. This speciﬁc
band is typically the one that gives the strongest Raman scattering and hence we selected
the 1068 cm'1 Raman band for EF calculation. For a more comprehensive estimate of EF,
we utilized the average intensity from 100 sample points of substrate in equation (7.1).
Taking 4.5 x 1014 molecules/cm2 for a monolayer of 4-MBT on silver (we assume that
the binding of 4-MBT on silver is similar to that of gold[130]), approximately

5.87 x 10'14 moles (or 3.53 x 10'0 molecules) of 4-MBT were excited in the laser spot.

The determination of Nref is more involved and different methodologies have been

299

 

reported to estimate it. One obvious choice to estimate Nref is by recording the Raman

signal from a ﬂat metal coated part of the same substrate but this method seldom yields
any discemable spectra due to extremely weak Raman signal from just a monolayer of
analyte in this conﬁguration. Moreover, the Raman signal obtained in this conﬁguration
is not the true ordinary Raman spectra due to the presence of metal in the vicinity of the
analyte[54]. Altemately one may use solid bulk form or higher concentration solution of
analyte to get the approximate values of the reference parameters. Either of these
techniques needs an estimate of laser interaction volume and that can be calculated from
depth of ﬁeld considerations. We did not detect any ordinary Raman spectra from ~20
mM 4-MBT methanolic bulk solution. Therefore, we used a solid crystal of 4-MBT to get

values of [ref and Nref' The number of molecules excited in the bulk sample was

determined by assuming the excitation volume to have the shape of a cylinder with waist
diameter equal to the focused laser beam diameter and length equal to the depth of ﬁled.
The depth of ﬁeld was determined by changing laser beam focus on sample and recording
the 1/2 power points. The Nsms was computed by taking the monolayer thickness of 4-
MBT as 0.5 nm[127], The average enhancement factor based on mean intensity from 100
sample points was estimated to be ~4.09 x 105 whereas the peak enhancement factor was
found to be ~ 1.03x 106. It may be mentioned that we ignore surface roughness in the EF
calculations that can introduce a minor variation from the calculated value. However, the
increased surface area due to roughness is largely offset by substantial voids between
nanorods. The average EF for the copper and silver nanorods was found to be
~2.23 x 105 and ~2.51 x 106 respectively. The higher EF for silver nanorods is probably

due to greater SERS activity of Ag. The Raman enhancement from copper nanorods is

300

 

not found to be very promising and the obvious reason is the rapid deterioration of the
surface due to oxidation[127].

It is useful to compare the measured Raman EF with that predicted by theory.
Theoretical enhancement factor for nanorods under quasi-static approximation can be

calculated from relationship proposed by Wokaun et al. [192]

EFSEas = If(wo)-f(wR)|2 (7.2)

where f ((1)0) and f ((oR) are, respectively, the enhancement at laser and

scattered Raman wavelengths. According to (7.2), the enhancement consists in two

different factors: f ((00) from particle near-ﬁeld enhancement at incident wavelength and
f (coR) arising from the re-radiation of the molecule Raman near-ﬁeld by the particle.

The best enhancement should be located near average of two wavelengths. This also
explains why the SERS enhancement decreases for higher Raman bands i.e., resonance
width cannot effectively cover both incident and scattered radiation.

For an isolated metallic object of volume V and dielectric function

8(a)) = 81(60)+i82 (co) placed in an external electric ﬁeld of frequency (00, the electric
ﬁeld enhancement f ((1)) is given by[74]
2
If (60)] =

1.10112 (7.3)
{1—[1—£]]Aj +82C}2 +{82Aj +[1-81]C}2

 

301

 

432V

where C = 3
3x1

 

particle eccentricity. For a spheroid with principal axes of length 210, 21,, , Zlc and the

laser ﬁeld applied ﬁeld applied along one of these axis, the depolarization factor is given

by[l92]

A .

 

boo
1:061 ds
0

(H112.){(s+l§)(s+1§x1~+3)}l

The depolarization factors can be computed from above expression knowing only

the geometry of the particle. Table 7-1 lists some of these factors for some common

particle shapes.

Table 7-1. Depolarization factors for some common geometries.

9

j=a, b,c.

and the depolarization factor[l93] 0<Aj <1 characterizes

 

 

 

 

 

 

 

 

 

Shape Depolarization factors

Aa Ab Ac
Sphere 1/3 1/3 1/3
Prolate spheroid with a :b = 2 :1 and b=c 0.1736 0.4132 0.4132
Prolate spheroid with a :b = 3 :1 and b=c 0.1087 0.4456 0.4456
Prolate spheroid with a : b = 3 :1 and b=c 0.0754 0.4623 04623
Oblate spheroid with a : b = l :3 and b=c 0.6354 0.1823 0.1823
Needle with a : b —> co and b=c 0 0.5 0.5

 

 

 

 

302

 

 

1110111
11311011
polarl
11111

like

11111!

Will

2185‘

bul
SP
re:
ne

CZ

Equation (7.3) can be used to predict theoretical SERS enhancement from the
knowledge of dielectric function and depolarization factor of metal particle. The metallic
nanorods synthesized here can be approximated as prolate spheroids with incident
polarization along the long axis of the rods. The dielectric ﬁrnction of metals at the laser
excitation wavelength can be approximated from the published values. For noble metals

like silver, the imaginary part of the dielectric constant is small throughout the visible

range and peak Raman enhancement l f ((0)l2 can be found by maximizing equation (7.3)

. . . 1
With respect to 81(60). The peak enhancement factor is obtained. for 81(wres) zl——

A}

assuming both .92 (co) and I? are much smaller than unity. Figure 7-21(a) is a plot of
xi.

bulk optical constants of silver as reported by Johnson and Christy[76]. We used cubic
spline interpolation to get silver dielectric constants £1 = ~29.7841 at 785 nm from
reported values. This value of 31 will yield resonance in a prolate spheroid of aspect ratio
nearly 6 (A j = 0.0481) with laser excitation along long axis. Figure 7-21(b) shows the
calculated Raman enhancement for 71 nm diameter silver nanorod (treated as prolate
spheroid) as a function of aspect ratio. The SERS enhancement increases with increasing

eccentricity of nanorod and becomes maximum (~ 7.605x105 ) for an aspect ratio of
nearly 6. It may be reiterated that the theoretical model is valid for an isolated rod and
does not take into account dipolar interactions between neighboring structures. The
measured EF is higher than the predicted value due to contributions from neighboring

nanorods.

303

 

 

 

 

 

 

'1001 15 a 2:5 5 3:5
Photon Energy (eV)
b
< >106,
105?

SERS EF, |f((0)|4
8

A
O
(A)

 

 

 

10 . 1 . . .

1 2 3 4 5 6
Nanorod aspect ratio

Figure 7-21. (a) Optical constants of silver as ﬁrnction of incident

wavelength[76]. Circles correspond to laser wavelength used in this work.

0)) Theoretical enhancement factor for an isolated silver nanorod as
function of aspect ratio.

304

Similar theoretical analysis can be repeated for copper nanorods and is shown in
Figure 7-22. The imaginary part of the dielectric function has larger value in case of
copper and represents greater losses (absorption in material) compared to silver. On the
other hand, the real part of dielectric function for copper has a lower magnitude than
silver and this is reﬂected in resonance being excited for lower particle eccentricity (the

maximum of SERS EF is predicted for an aspect ratio of 5 in case of copper as against 6

for silver). The predicted maximum EF is ~ 5.88 ><10s for the copper nanorods. However,
the measured value was smaller than the predicted and difference is probably due to
oxidation of copper nanorods. The rapid oxidation of copper and aluminum was visually
noticeable by the change of substrate color after removing it from deposition chamber.
The EDS data also showed substantial presence of oxygen in case of aluminum and
copper nanorods. The silver nanorods did not show any signiﬁcant oxygen during EDS.
This indicates the oxidation of silver is much slower compared to copper. The non-
conductive top oxide layer on copper and aluminum not only changes the chemical
binding of analyte but will also adversely affect the excitation of surface plasmons. On

the other hand, silver oxide is a conductor and may still support surface plasmons.

305

(a) . . .

 

 

 

 

 

 

207 82
0 :‘N “MM mmmmm M
-20 _ 81
40-
J50-
-80 ..
-1 a 1 n a
001 1.5 2 2.5 3 3.5
Photon Energy (eV)
(b)

 

SERS EF, ”((0)14

 

P

 

 

 

10152521567 8

Nanorod aspect ratio

Figure 7-22. (a) Optical constants of copper as function of incident
wavelength[76]. Circles correspond to laser wavelength used in this work.
(b) Theoretical enhancement factor for an isolated copper nanorod as
function of aspect ratio.

306

7.7

Detection of BDT

The SERS performance of the aluminum and silver nanorods was also evaluated

using 1,2-BDT probe molecule. This molecule also forms self assembled monolayer

(thickness of SAM ~0.49 nm) on silver by forming two sulfur-metal bonds[126]. For

monolayer formation, the silver tipped aluminum nanorods substrate was soaked in 20

uM methanolic 1,2-BDT solution for 40 min. The sample was then rinsed with methanol

and dried under dry nitrogen stream to ensure removal of any exCess analyte. Figure 7-23

shows the Raman spectrum of SAM of 1,2-BDT from silver tipped aluminum nanorods.

It is evident that the substrate exhibits good enhancement to yield strong and

unambiguous spectrum of 1,2-BDT superimposed on slowly varying broad background.

4.5 ~

Raman lntensity/Arbitr. Units

 

.111“-

— before
--------- after 1.2-BDT

 

 

 

 

 

 

 

800 1000 1200 1400 1600
Wavenumber/cm"1

600

Figure 7-23. Raman spectra of aluminum nanorods before and aﬁer
application of formation of SAM of 1,2-BDT, tacc= 15 s.

307

 

 

 

before
---------- after 1.2-BDT

 

 

 

 

Raman Intensity/Arbitr. Units

 

 

 

0 J; L 1 r 1 -
200 400 600 800 1000 1 200 1400 1 600
Wavenumber/cm'1

Figure 7-24. Raman spectra of silver nanorods before and after
application of 20 uM 1,2-BDT, tacc= 15 s.

For comparison we also acquired SERS spectra of 20 uM 1,2-BDT from silver
nanorods substrate that was processed under similar conditions. Figure 7-24 shows that
the silver nanorods yield more intense spectra than aluminum nanorods as was observed
in case of 4-MBT. To quantify the performance of these two substrates for 1,2-BDT, the
SERS EF for 1,2-BDT was again calculated using equation (7.1) by utilizing the
1028 cm'1 Raman band. Assuming monolayer coverage of substrate with liZ-BDT and

each molecule binding to two silver atoms on surface, we get ca. 6.4 x 1014
molecules/cm2 on the substrate. This implies that nearly 8.35 x 10‘M moles (5.03 x 1010

molecules) of 1,2-BDT contribute to the SERS signal for the laser spot size of 100 pm.
The ordinary Raman spectrum of the bulk 1,2-BDT was used as a reference to calculate

the Raman EF. The average intensity of 1028 cm'1 Raman band from 100 sample points

308

was employed to calculate Raman EF. The average enhancement factor from equation
(7.1) was found to be ~ 2.3 x 105 for aluminum nanorods and ~ 8.1 x 106 for silver

nanorods. These values are comparable to those obtained for 4-MBT and conﬁrm strong

enhancement from nanorods substrate.

7.8 Conclusion

Copper and aluminum nanorods were synthesized on glass substrates using OAD
with e-beam evaporation. The tips of aluminum nanorods were coated with silver to make
them SERS active. The EDS of the copper and aluminum nanorods showed presence of
oxygen that was attributed to the oxidation of metal after exposure to atmosphere. The
synthesized substrates were evaluated for SERS by using 4-MBT and 1,2-BDT as probe
molecules. The substrates yielded an average enhancement factor of the order of 105 with
good uniformity over the tested areas. The amount of analyte that yielded these strong
SERS spectra was on the order of femto moles and conﬁrmed the high sensitivity of these
substrates. However, EDS analysis of both copper and aluminum nanorods showed
existence of substantial amount of oxygen that was attributed to the corresponding metal
oxide layer on nanorods. Silver nanorods on glass slides were also prepared and their
SERS performance was compared with that of other two types of nanorods using same
probe molecules. The pure silver nanorods substrates were found to have higher density
of nanorods and larger enhancement factors of the order of ~106. The better SERS from
silver nanorods was attributed to its greater SERS activity and absence of insulating top

layer.

309

Chapter 8: Conclusions and Future Work

8.] Conclusion

This dissertation deals with various types of nanowires and nanorods substrates
for potential use in SERS. The initial SERS studies were done using germanium oxide
nanowires synthesized at high temperature (> 800 °C). The high temperature synthesis
scheme yielded very dense and long nanowires with little control over diameter and
length. The nanowires were found to comprise of crystalline hexagonal phase of GeOz
based on TEM, EDS and Raman studies. The as-synthesized nanowires did not yield any
discemable Raman enhancement. Subsequently nanowires substrates were coated with
thin ﬁlm of gold to provide surface plasmon active surface. The gold was selected
because of its good SERS activity, chemical stability and compatibility with biological
species. The gold coating introduced a broad background in the SERS spectra of
nanowires. The possible sources of background continuum due to contamination from
metal deposition or processing of substrate were thoroughly investigated. The TEM study
of gold-coated nanowires showed formation of nanoscale gold islands. It was concluded
that the source of background is a characteristic of rough metal ﬁlm over the substrate
based on this study and previous reports in literature. The SERS performance of the
substrates was evaluated using rhodamine 6G and nile blue probe molecules. Although
strong SERS signal was obtained for micro molar concentration of these analytes, the
issue of hot spots surfaced. It was hypothesized that the hot spots were due to selective
adsorption of analyte over the substrate rather than the non-homogeneity of substrate.
This hypothesis was tested by investigating thiol based analytes that are known to form

self-assembled monolayers on gold under appropriate conditions. The 4-MBT, BPE and

310

 

1,2-BDT yielded strong and uniform SERS enhancement over tested area of the
substrates. The average enhancement factor for GeOz nanowires based substrates was

estimated to be of the order of 106 without any hot spots. The corresponding thickness of
gold over smooth silicon surface did not yield any SERS enhancement aﬁer application
of analyte alongside nanowire substrates. This showed the important role of nanowires in
enhancing SERS signal. It was hypothesized that the nanowires framework served
threefold purpose; (1) to provide a metal rough surface for plasmon activation, (2) to
conﬁne guide and couple the light between adjacent wires, and (3) to focus the light at
the surface of the metal. The combined affect of these contributions of nanowires resulted
in a strong and reproducible SERS signature from nanowires.

The lack of control over the diameter and length of the nanowires synthesized at
high temperature necessitated ﬁrrther study of the synthesis process. The analysis of high
temperature synthesis revealed very rapid reaction dumping lot of source material on the
substrate in short time. It was hypothesized that the rate of reaction may be lowered by
exploring temperate regime below the sublimation temperature of GeO. This led to the
development of a new low temperature synthesis scheme that provided better control over
the supply and condensation of source material on the substrate. The low temperature
synthesis was carried out at a source temperature of ~650 °C while the substrate was
located in the temperature region of 400-500 °C. This new synthesis employs the lowest
temperature reported for germanium oxide nanowire growth using thermal evaporation.
The technique yields good control over the diameter and length of nanowires. The Raman
characterization of low temperature synthesized nanowires revealed a broad continuum

with moderate strength. This was attributed to the existence of gold nanoparticles from

311

 

the initial gold ﬁlm used to catalyze nanowires. However, the density of these gold
particles was too low for any reasonable SERS enhancement and very weak Raman
signal was detected for uncoated nanowires. The nanowires were coated with top gold
layer to improve their SERS activity. Like the case of high temperature synthesized
nanowires, gold coating resulted in substantial increase in the intensity of broad
background. The SERS evaluation was done using 4-MBT and 1,2-BDT probe
molecules. The substrates yielded strong and repeatable SERS signal for the two analytes
over the tested area. The SERS performance of theses nanowire based substrates is also
compared with that of commercially available SERS chip, Klarite from 3D Technologies
Ltd. Our results indicate strong dependence of SERS on length of nanowires and
thickness of gold coating. The average conservative estimates of enhancement factor for
low temperature synthesized nanowire of nearly 1 pm length and coated with ~40 nm
gold is approximately 6><106 with small standard deviation of nearly 4%. This
enhancement was found to be better than from that of Klarite chip that yielded an
enhancement of ~1.5><106 for same analyte at 785 nm laser excitation. This initial
investigation conﬁrms the potential of oxide nanowires for SERS applications aimed at
developing a robust and sensitive detector platform.

Another type of substrate that was studied in this dissertation was the zinc oxide
nanowires synthesized on silicon substrate. The synthesis was also based on simple
thermal evaporation of zinc powder in an oxidizing environment. The same tube ﬁtrnace
set up used for germanium oxide nanowire synthesis was utilized to grow ZnO nanowires
but with different reaction conditions. Unlike the case of germanium oxide that only grew

at location of gold catalyst, the zinc oxide nanowires were also found at locations without

312

 

 

catalyst. However, the yield of catalyst-coated area was higher compared to uncoated
area for the same experimental conditions. Another difference observed was in the
morphology of the nanowires for the two materials. The germanium oxide nanowires
almost always grew as long cylindrical wires whereas ZnO was found to yield a variety
of morphologies including nanowires, hexagonal nanorods, nano-combs, nano-cages,
nano-particles etc. However, the main morphologies were found to be nanowires and
hexagonal nanorods. The hexagonal nanorods were usually partially aligned normal to
the substrate and less dense compared to the randomly oriented nanowires. The SERS of
both these morphologies of ZnO was also evaluated using 4-MBT and 1,2-BDT probe
molecules. The gold thickness for optimal SERS enhancement was found to be ~40 nm.
The average SERS enhancement of the order of 106 was obtained over the tested area of
the substrate with small variations.

Metallic nanorods of aluminum and copper were also synthesized and evaluated
for use as SERS substrates. The nanorods were fabricated over glass substrate using
oblique angle deposition (OAD) technique. In OAD synthesis, the source material is
evaporated from a point source and the resulting vapor arrives at a glancing angle on the
substrate. The nanostructures are formed due to atomic shadowing of incident vapor on
the substrate. The choice of aluminum and copper was based due to their low cost and
ready availability in most research laboratories. The aluminum nanorods were
synthesized with silver tips to make them SERS active. Silver nanorods have been
previously reported to yield good SERS activity and were also synthesized to compare
their performance against those of aluminum and copper. The synthesized nanorods were

evaluated using NB, 4-MBT and 1,2-BDT probe molecules. The aluminum and copper

313

 

nanorods yielded average SERS enhancement factor of the order of 105 whereas silver
nanorods showed stronger SERS enhancement of ~106. The stronger SERS enhancement
for silver is not unexpected given the higher SERS activity of silver compared to other
metals. However, an added issue with copper and aluminum nanorods is their tendency to
form insulating surface oxide layer. Both aluminum and copper nanorods based
substrates were found to change color after exposure to atmosphere. The EDS analysis
conﬁrmed existence of substantial quantity of oxide on the substrate. The insulating
oxide layer not only changes the adsorption of analyte over the substrate but also affects
the surface plasmon excitation in the top layer. Therefore, these two types were not
considered competitive choices as SERS substrates unless surface stabilization can be
achieved.

The list of major contributions of this dissertation is as follows:

1. Investigation of optimal conditions for high temperature synthesis of
GeO; nanowires.
2. Demonstration of strong and reproducible SERS enhancement from gold

coated G602 nanowires synthesized at high temperature.

3. Development of novel technique for low temperature synthesis of GeO;
nanowires.

4. Identiﬁcation of optimal conditions for low temperature synthesis of GeO;
nanowires.

5. Demonstration of good control over length and diameter of low
temperature synthesis of GeO; nanowires.

314

 

6. Identiﬁcation of optimal parameters for SERS from low temperature
synthesized GeO; nanowires.
7. Demonstration of strong and uniform SERS enhancement from gold

coated GeOz nanowires synthesized at low temperature.

8. Synthesis of two morphologies of ZnO nanowires.

9. Demonstration of strong and uniform SERS enhancement from two
morphologies of ZnO nanowires.

10. Synthesis of aluminum, copper and silver nanorods and investigation of
SERS performance.

11. Comparison of SERS performance of nanowires and nanorods with that of

commercial Klarite SERS chip.

8.2 Future work

This dissertation forms the early work on the potential of using bulk synthesized
oxide nanowires for use as SERS substrates. The results conﬁrm that the nanowires have
the potential to act as economical, scaleable and sensitive platform for detection of
chemical species. The strength of these substrates lies in the good uniformity and
repeatability of the SERS signal over the spatial extent of the substrate without the
serious issue of hot spots that has plagued SERS research in case of random aggregates of
metallic colloidals. However, caution has to be exercised in declaring these as the
substrates of choice before further research. First of all, the most critical issue of analyte
adsorption on the substrate needs be addressed. It is a universal agreement that gold and

silver has high SERS sensitivity but the enhancement decays exponentially with

315

 

increasing distance from metal surface. The distances involved are of the order of few
nanometers and can be reliably established for analytes that chemically bind to metal.
This implies that the enhancement factors are strongly dependent upon type of analyte
and the associated binding to the substrate. In fact, some sort of linker molecules may
have to be used in order to immobilize non-binding target molecules on the substrate. A
second issue relates to the change in the morphology of the substrate upon application of
analyte solution. The resulting capillary forces during drying of the solution tend to pull
nanowires together and leave behind bundled structures. There need to be more work
done to increase robustness of the nanostructures so as to minimize distortion of
nanostructures on the substrate. Thirdly nanowires of additional materials may be
evaluated for use in SERS. Hence future work needs to focus on improving robustness of
the nanowires, exploring additional nanowire materials, and investigating mechanism to
immobilized analyte at close proximity to the substrate. Regarding metallic nanorods,
only silver nanorods have exhibited the potential to yield high enhancement. Future work
on silver nanorods may be focused towards achieving better control over the diameter of

the nanorods to further enhance their SERS performance.

316

 

 

A.1

Appendix A: Operation of EzRaman—L system

Introduction

Following procedure gives a summary of EzRaman-L operation. For detailed

description, please refer to system operation manual located alongside system.

A.2

A.3

Procedure

Power on the computer.

Login using Raman proﬁle. Password is “raman”.

Start the Enwave software by double clicking “Enwave application center” icon
on desktop.

Start CCD camera software by double clicking “Logitech quickcam” icon on

desktop. Start “Quick capture” to view sample image.

. Make appropriate entry in Raman System log book.

Power on the Raman system by using Power ON/OFF switch located on front.

The “SYSEM ON” green LED should lit.

Sample mounting
Set the sample on the sample stage.
Adjust distance between lens and sample to get sample at lens focus (~ 6 mm for

0.22 NA and ~3 mm 0.55 NA lens tube).

. Position CCD camera such that it faces the sample

Place the cover in front of the Raman probe enclosure to ensure laser safety.

. Put on laser safety goggle.

317

 

A.4

. Select X-axis as “cm'

Acquiring Spectra

Set the Raman system parameters. Typical parameters are, “Integration :5,
Average :1 and Boxcar :1”. Use longer integration or higher average to get better
signal in case of weak and noisy spectra.

1”

Select X-axis scale as minimum 250 maximum 2200.

. Select Y-axis scale as minimum 0 maximum 60,000.

. Switch on the laser by turning laser key clockwise from OFF to ON position.

Wait for few seconds for red “LASER ON” LED to light.

. Start desired type of scan by clicking corresponding icon. Typically continuous

scan is used during focusing or spatial scanning of sample surface. Single scan
will do one scan and then stop automatically whereas continuous scan needs be

manually stopped using “Stop scan” button.

1

j a Single A . j Continuous . .. 1! Stop
q" scan ﬁt; scan .i scan

. Adjust ﬁne focus by changing laser focus (use z-translation in quarter turn) and

monitoring signal strength. The maximum signal intensity will be achieved when

sample is in sharp focus. Stop scan if in continuous mode.

. Obtain desired spectra by changing time and average settings if desired. Stop scan

if in continuous scan mode.

318

 

 

A.5

A.6

. Save Raman spectrum in desired ﬁle format (text, excel, jpg) by using save as

option of “FILE” menu.

Shutdown

. Make sure laser scan is OFF. If not, stop laser scan by pressing “Stop scan” icon.

Switch Off the laser by turning laser key clockwise from ON to OFF position.
The red LASER ON LED should extinguish.

Shut down the Raman system by using Power ON/OFF switch located on front

. Close Raman software.

System Calibration
System calibration is required to ensure correct pixel to wavenumber mapping.

Please refer to chapter 5 of Raman system manual.

319

 

References

[1] G. L. Liu and L. P. Lee, "Nanowell surface enhanced Raman
scattering arrays fabricated by soft-lithography for label-free

biomolecular detections in integrated microﬂuidics," Applied Physics
Letters, vol. 87, pp. 047101, 2005.

[2] A. Campion, "Raman Spectroscopy," in Vibrational spectroscopy of
molecules on surfaces, J. T. Yates and T. E. Madey, Eds. New York
and London: Plenum Press, 1987.

[3] E. Smith and G. Dent, Modern Raman Spectroscopy - A practical
approach. West Sussex: John Wiley & Sons Ltd, 2005.

[4] C. L. Stevenson and T. Vo-Dinh, "Signal expressions in Raman
Spectroscopy," in Modern Techniques in Raman Spectroscopy, J. J.
Laserna, Ed. West Sussex: John Wiley and Sons, 1996, pp. 1-39.

[5] D. Porterﬁeld and A. Campion, "Flourescence-free Ramam

spectroscopy," Journal of American Chemical society, vol. 110, pp.
408-410, 1988.

[6] Fleischm.M, P. J. Hendra, and Mcquilla.Aj, "Raman-Spectra of
Pyridine Adsorbed at a Silver Electrode," Chemical Physics Letters,
vol. 26, pp. 163-166, 1974.

[7] M. Moskovits, "Surface roughness and the enhanced intesity of
Raman scattering by molecules adsorbed on metals," The Journal of
Chemical Physics, vol. 69, pp. 4159-4161, 1978.

[8] A. Otto, I. Mrozek, H. Grabhom, and W. Akemann, "Surface-
Enhanced Raman-Scattering," Journal of Physics-Condensed Matter,
vol. 4, pp. 1143-1212, 1992.

[9] K. Kneipp, H. Kneipp, V. B. Kartha, R. Manoharan, G. Deinum, I.
Itzkan, R. R. Dasari, and M. S. Feld, "Detection and identiﬁcation of a

single DNA base molecule using surface-enhanced Raman scattering
(SERS)," Physical Review E, vol. 57, pp. R628l-R6284, 1998.

320

 

 

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

K. Kneipp, H. Kneipp, G. Deinum, I. Itzkan, R. R. Dasari, and M. S.
Feld, "Single-molecule detection of a cyanine dye in silver colloidal
solution using near-infrared surface-enhanced Raman scattering,"
Applied Spectroscopy, vol. 52, pp. 175-178, 1998.

K. Kneipp, H. Kneipp, I. Itzkan, R. R. Dasari, and M. S. F eld,
"Surface-enhanced non-linear Raman scattering at the single-molecule
level," Chemical Physics, vol. 247, pp. 155-162, 1999.

K. Kneipp, H. Kneipp, R. Manoharan, I. Itzkan, R. R. Dasari, and M.
S. F eld, "Near-infrared surface-enhanced Raman scattering can detect
single molecules and observe 'hot' vibrational transitions," Journal of
Raman Spectroscopy, vol. 29, pp. 743-747, 1998.

K. Kneipp, Y. Wang, H. Kneipp, L. T. Perelman, I. Itzkan, R. Dasari,
and M. S. Feld, "Single molecule detection using surface-enhanced

Raman scattering (SERS)," Physical Review Letters, vol. 78, pp.
1667-1670, 1997.

S. M. Nie and S. R. Emory, "Single-molecule detection and
spectroscopy by surface-enhanced Raman scattering," Abstracts of
Papers of the American Chemical Society, vol. 213, pp. l77-Phys,
1997.

S. Nie and S. R. Emory, "Probing Single Molecules and Single
Nanoparticles by Surface-Enhanced Raman Scattering," Science, vol.
275, pp. 1102-1106, 1997.

D. J. Maxwell, S. R. Emory, and S. M. Nie, "Nanostructured thin-ﬁlm

materials with surface-enhanced optical properties," Chemistry of
Materials, vol. 13, pp. 1082-1088, 2001.

W. A. Lyon and S. M. Nie, "Conﬁnement and detection of single
molecules in submicrometer channels," Analytical Chemistry, vol. 69,
pp. 3400-3405, 1997.

J. T. Krug, G. D. Wang, S. R. Emory, and S. M. Nie, "Efﬁcient
Raman enhancement and intermittent light emission observed in

single gold nanocrystals," Journal of the American Chemical Society,
vol. 121, pp. 9208-9214, 1999.

321

 

 

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

K. Kneipp, Y. Wang, H. Kneipp, I. Itzkan, R. R. Dasari, and M. S.
Feld, "Population pumping of excited vibrational states by

spontaneous surface-enhanced Raman scattering," Physical Review
Letters, vol. 76, pp. 2444-2447, 1996.

J. Kneipp, "Nanosensors based on SERS for applications in living
cells," Surface-Enhanced Raman Scattering: Physics and
Applications, vol. 103, pp. 335-349, 2006.

C. R. Yonzon, O. Lyandraes, N. C. Shah, J. A. Dieringer, and R. P. V.
Duyne, "Glucose sensing with Surface-Enhanced Raman
Spectroscopy," Surface-Enhanced Raman Scattering: Physics and

Applications, vol. 103, pp. 367-379, 2006.

T. Vo-Dinh, F. Yan, and M. B. Wabuyele, "Surface-Enhanced Raman
Scattering for Biomedical Diagnostics and Molecular Imaging,"

Surface-Enhanced Raman Scattering: Physics and Applications, vol.
103, pp. 409-426, 2006.

T. Vo-Dinh, N. Isola, J. P. Alarie, D. Landis, G. D. Grifﬁn, and S.
Allison, "Development of a multiarray biosensor for DNA

diagnostics," Instrumentation Science & Technology, vol. 26, pp. 503-
514, 1998.

A. M. Alak and T. Vodinh, "Surface-Enhanced Raman-Spectrometry
of Organophosphorus Chemical-Agents," Analytical Chemistry, vol.
59, pp. 2149-2153, 1987.

S. Farquharson, F. E. Inscore, and S. Christesen, "Detecting Chemical
Agents and their Hydrolysis Products in Water," Surface-Enhanced
Raman Scattering: Physics and Applications, vol. 103, pp. 447-460,
2006.

C. R. Y. Christy L. Haynes, Xiaoyu Zhang, Richard P. Van Duyne,,
"Surface-enhanced Raman sensors: early history and the development
of sensors for quantitative biowarfare agent and glucose detection,"
Journal of Raman Spectroscopy, vol. 36, pp. 471-484, 2005.

X. Y. Zhang, N. C. Shah, and R. P. Van Duyne, "Sensitive and
selective chem/biosensing based on surface-enhanced Raman

spectroscopy (SERS)," Vibrational Spectroscopy, vol. 42, pp. 2-8,
2006.

322

 

 

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

C. R. Yonzon, X. Zhang, J. Zhao, and R. P. Van Duyne, "Surface-
enhanced nanosensors," Spectroscopy, vol. 22, pp. 42, 2007.

S. Farquharson, A. D. Gift, P. Maksymiuk, and F. E. Inscore, "Rapid
dipicolinic acid extraction from bacillus spores detected by surface-

enhanced Raman spectroscopy," Applied Spectroscopy, vol. 58, pp.
351-354, 2004.

A. Sengupta, M. Mujacic, and E. J. Davis, "Detection of bacteria by
surface-enhanced Raman spectroscopy," Analytical and Bioanalytical
Chemistry, vol. 386, pp. 1379-1386, 2006.

K. T. Carton, X. Gi, and M. L. Lewis, "A Surface Enhanced Raman-
Spectroscopy Study of the Corrosion-Inhibiting Properties of

Benzimidazole and Benzotriazole on Copper," Langmuir, vol. 7, pp.
2-4, 1991.

L. M. Sudnik, K. L. Norrod, and K. L. Rowlen, "SERS-active Ag
ﬁlms from photo reduction of Ag+ on TiOz," Applied Spectroscopy,
vol. 50, pp. 422-424, 1996.

T. R. Jensen, M. D. Malinsky, C. L. Haynes, and R. P. Van Duyne,
"Nanosphere lithography: Tunable localized surface plasmon

resonance spectra of silver nanoparticles," Journal of Physical
Chemistry B, vol. 104, pp. 10549-10556, 2000.

G. Sauer, U. Nickel, and S. Schneider, "Preparation of SERS-active
silver ﬁlm electrodes via electrocrystallization of silver," Journal of
Raman Spectroscopy, vol. 31, pp. 359-363, 2000.

A. Tao, F. Kim, C. Hess, J. Goldberger, R. R. He, Y. G. Sun, Y. N.
Xia, and P. D. Yang, "Langmuir-Blodgett silver nanowire monolayers

for molecular sensing using surface-enhanced Raman spectroscopy,"
Nano Letters, vol. 3, pp. 1229-1233, 2003.

S. B. Chaney, S. Shanmukh, R. A. Dluhy, and Y. P. Zhao, "Aligned
silver nanorod arrays produce high sensitivity surface-enhanced

Raman spectroscopy substrates," Applied Physics Letters, vol. 87, pp.
031908, 2005.

323

 

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44]

145]

[46]

[47]

H. Y. Chu, Y. J. Liu, Y. W. Huang, and Y. P. Zhao, "A high sensitive
ﬁber SERS probe based on silver nanorod arrays," Optics Express,
vol. 15, pp. 12230-12239, 2007.

A. J. Haes, C. L. Haynes, A. D. McFarland, G. C. Schatz, R. R. Van
Duyne, and S. L. Zou, "Plasmonic materials for surface-enhanced
sensing and spectroscopy," Mrs Bulletin, vol. 30, pp. 368-375, 2005.

J. Jiang, K. Bosnick, M. Maillard, and L. Brus, "Single molecule
Raman spectroscopy at the junctions of large Ag nanocrystals,"
Journal of Physical Chemistry B, vol. 107, pp. 9964-9972, 2003.

Z. Li, W. M. Tong, W. F. Stickle, D. L. Neiman, R. S. Williams, L. L.
Hunter, A. A. Talin, D. Li, and S. R. J. Brueck, "Plasma-Induced
Formation of Ag Nanodots for Ultra-High-Enhancement Surface-

Enhanced Raman Scattering Substrates," Langmuir, vol. 23, pp. 5135-
5138, 2007.

J. P. Kottmann and O. J. F. Martin, "Plasmon resonant coupling in
metallic nanowires," Optics Express, vol. 8, pp. 655-663, 2001.

J. P. Kottmann, O. J. F. Martin, D. R. Smith, and S. Schultz,
"Dramatic localized electromagnetic enhancement in plasmon

resonant nanowires," Chemical Physics Letters, vol. 341 , pp. 1-6,
2001.

C. V. Raman and K. S. Krishnan, "A new type of secondary
radiation," Nature, vol. 121, pp. 501-502, 1928.

G. Mie, "Articles on the optical characteristics of turbid tubes,

especially colloidal metal solutions," Annalen Der Physik, vol. 25,
pp. 377-445, 1908.

D. A. Long, Raman Spectroscopy. New York: McGraw-Hill Inc,
1977.

E. Gross, "Change of wave-length of light due to elastic heat waves at
scattering in liquids," Nature, vol. 126, pp. 201-202, 1930.

H. Skoog and Neiman, Principles of Instrumental Analysis, 5th ed.
Philadelphia: Harcourt Brace College Pulbishers, 1998.

324

 

[48]

[49]
[50]

[51]

[52]

[53]

[541

[55]

[56]

[57]

[58]

H. A. Kramers and W. Heisenberg, "On the dispersal of radiation by
atoms," Zeitschrift Fur Physik, vol. 31, pp. 681-708, 1925.

P. A. M. Dirac, Proc. R. Soc. London, Ser. A, vol. 114, pp. 710, 1927.

A. C. Albrecht, "Theory of *Raman Intensities," Journal of Chemical
Physics, vol. 34, pp. 1476-1484, 1961.

A. Campion and P. Kambhampati, "Surface-enhanced Raman
scattering," Chemical Society Reviews, vol. 27, pp. 241-250, 1998.

A. Campion, J. E. Ivanecky, C. M. Child, and M. Foster, "On the
Mechanism of Chemical Enhancement in Surface-Enhanced Raman-

Scattering," Journal of the American Chemical Society, vol. 117, pp.
11807-11808, 1995.

P. Kambhampati, C. M. Child, M. C. Foster, and A. Campion, "On the
chemical mechanism of surface enhanced Raman scattering:

Experiment and theory," Journal of Chemical Physics, vol. 108, pp.
5013-5026, 1998.

M. Moskovits, "Surface-enhanced spectroscopy," Reviews of Modern
Physics, vol. 57, pp. 783-826, 1985.

R. W. Wood, "On a Remarkable case of Uneven Distribution of Light

in a Diffraction Grating spectrum," Proc. Phys. Soc. London, vol. 18,
pp. 269-275, 1902.

H. Raether, "Surface-Plasmons on Smooth and Rough Surfaces and
on Gratings," Springer Tracts in Modern Physics, vol. 111, pp. 1-133,
1988.

J. A. Creighton, C. G. Blatchford, and M. G. Albrecht, "Plasma
Resonance Enhancement of Raman-Scattering by Pyridine Adsorbed
on Silver or Gold Sol Particles of Size Comparable to the Excitation

Wavelength," Journal of the Chemical Society-Faraday Transactions
Ii, vol. 75, pp. 790-798, 1979.

J. Gersten and A. Nitzan, "Electromagnetic Theory of Enhanced
Raman-Scattering by Molecules Adsorbed on Rough Surfaces,"
Journal of Chemical Physics, vol. 73, pp. 3023-3037, 1980.

325

 

 

[59]

[60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

[68]

J. Gersten and A. Nitzan, "Spectroscopic Properties of Molecules
Interacting with Small Dielectric Particles," Journal of Chemical
Physics, vol. 75, pp. 1139-1152, 1981.

J. I. Gersten, "The Effect of Surface-Roughness on Surface Enhanced

Raman-Scattering," Journal of Chemical Physics, vol. 72, pp. 5779-
5780, 1980.

J. I. Gersten, "Rayleigh, Mie, and Raman-Scattering by Molecules
Adsorbed on Rough Surfaces," Journal of Chemical Physics, vol. 72,
pp. 5780-5781, 1980.

S. L. Mccall and P. M. Platzman, "Raman-Scattering from

Chernisorbed Molecules at Surfaces," Physical Review B, vol. 22, pp.
1660-1662, 1980.

S. L. Mccall, P. M. Platzman, and P. A. Wolff, "Surface Enhanced
Raman-Scattering," Physics Letters A, vol. 77, pp. 381-3 83, 1980.

M. Kerker, O. Siiman, L. A. Bumm, and D. S. Wang, "Surface
Enhanced Raman-Scattering (Sers) of Citrate Ion Adsorbed on
Colloidal Silver," Applied Optics, vol. 19, pp. 3253-3255, 1980.

M. Kerker, O. Siiman, and D. S. Wang, "Effect of Aggregates on
Extinction and Surface-Enhanced Raman-Scattering Spectra of
Colloidal Silver," Journal of Physical Chemistry, vol. 88, pp. 3168-
3170, 1984.

M. Kerker, D. S. Wang, and H. Chew, "Surface Enhanced Raman-
Scattering (Sers) by Molecules Adsorbed at Spherical-Particles,"
Applied Optics, vol. 19, pp. 4159-4174, 1980.

D. S. Wang, H. Chew, and M. Kerker, "Enhanced Raman-Scattering
at the Surface (Sers) of a Spherical-Particle," Applied Optics, vol. 19,
pp. 2256-2257, 1980.

D. S. Wang and M. Kerker, "Enhanced Raman-Scattering by
Molecules Adsorbed at the Surface of Colloidal Spheroids," Physical
Review B, vol. 24, pp. 1777-1790, 1981.

326

 

[69]

[70]

[71]

[72]

[73]

[74]

[75]

[76]

[77]

[78]

D. S. Wang and M. Kerker, "Absorption and Luminescence of Dye-
Coated Silver and Gold Particles," Physical Review B, vol. 25, pp.
2433-2449, 1982.

D. S. Wang, M. Kerker, and H. W. Chew, "Raman and Fluorescent
Scattering by Molecules Embedded in Dielectric Spheroids," Applied
Optics, vol. 19, pp. 2315-2328, 1980.

H. Wang, C. S. Levin, and N. J. Halas, "Nanosphere arrays with
controlled sub-IO-nm gaps as surface-enhanced Raman spectroscopy
substrates," Journal of the American Chemical Society, vol. 127, pp.
14992-14993, 2005.

K. L. Kelly, B. Coronado, L. L. Zhao, and G. C. Schatz, "The optical
properties of metal nanoparticles: The inﬂuence of size, shape, and

dielectric environment," Journal of Physical Chemistry B, vol. 107,
pp. 668-677, 2003.

M. Moskovits, "Surface-enhanced Raman spectroscopy: a brief
retrospective," Journal of Raman Spectroscopy, vol. 36, pp. 485-496,
2005.

A. Wokaun, J. P. Gordon, and P. F. Liao, "Radiation Damping in
Surface-Enhanced Raman-Scattering," Physical Review Letters, vol.
48, pp. 957-960, 1982.

B. J. Messinger, K. U. Vonraben, R. K. Chang, and P. W. Barber,
"Local-Fields at the Surface of Noble-Metal Microspheres," Physical
Review B, vol. 24, pp. 649-657, 1981.

P. B. Johnson and R. W. Christy, "Optical-Constants of Noble-
Metals," Physical Review B, vol. 6, pp. 4370-4379, 1972.

S. A. Maier, M. L. Brongersma, P. G. Kik, S. Meltzer, A. A. G.
Requicha, and H. A. Atwater, "Plasmonics - A route to nanoscale
optical devices," Advanced Materials, vol. 13, pp. 1501, 2001.

C. J. F. Bottcher, Theory of electric polarization, vol. 1, 2nd ed. New
York: Elsevier scientiﬁc publishing company, 1973.

327

 

 

[79]

[80]

[81]

[82]

[83]

[84]

[85]

[86]

[87]

[88]

M. Moskovits, "Surface-enhanced Raman spectroscopy: a brief
perspective," Surface-Enhanced Raman Scattering: Physics and
Applications, vol. 103, pp. 1-17, 2006.

A. M. Michaels, J. J iang, and L. Brus, "Ag nanocrystal junctions as
the site for surface-enhanced Raman scattering of single Rhodamine
6G molecules," Journal of Physical Chemistry B, vol. 104, pp. 11965-
1 1971, 2000.

R. S. Wagner and W. C. Ellis, "Vapor-Liquid-Solid Mechanism of
Single Crystal Growth," Applied Physics Letters, vol. 4, pp. 89-90,
1964..

M. Zacharias and P. M. Fauchet, "Light emission from Ge and GeOz

nanocrystals," Journal of Non-Crystalline Solids, vol. 227-230, pp.
1058-1062, 1998.

X. C. Wu, W. H. Song, B. Zhao, Y. P. Sun, and J. J. Du, "Preparation
and photoluminescence properties of crystalline GeOz nanowires,"
Chemical Physics Letters, vol. 349, pp. 210-214, 2001.

S. Sakaguchi and S. Todoroki, "Optical properties of GeOz glass and
optical ﬁbers," Applied Optics, vol. 36, pp. 6809-6814, 1997.

Z. G. Bai, D. P. Yu, H. Z. Zhang, Y. Ding, Y. P. Wang, X. Z. Gal, Q.
L. Hang, G. C. Xiong, and S. Q. Feng, "Nano-scale GeO; wires

synthesized by physical evaporation," Chemical Physics Letters, vol.
303, pp. 311-314, 1999.

Y. Zhang, J. Zhu, Q. Zhang, Y. Yan, N. Wang, and X. Zhang,
"Synthesis of GeO; nanorods by carbon nanotubes template,"
Chemical Physics Letters, vol. 317, pp. 504-509, 2000.

H. Y. Dang, J. Wang, and S. S. Fan, "The synthesis of metal oxide
nanowires by directly heating metal samples in appropriate oxygen
atmospheres," Nanotechnology, vol. 14, pp. 73 8-741, 2003.

Q. L. J .-Q. Hu, X.-M. Meng, C.S. Lee, S.T. Lee,, "Synthesis and
Nanostructuring of Patterned Wires of alpha-GeO; by Thermal
Oxidation," Advanced Materials, vol. 14, pp. 1396-1399, 2002.

328

 

[89]

[90]

[91]

[92]

[93]

[94]

[95]

[96]

[97]

[98]

[99]

Y. Su, X. Liang, S. Li, Y. Chen, Q. Zhou, S. Yin, X. Meng, and M.
Kong, "Self-catalytic VLS growth and optical properties of single-

crystalline GeOz nanowire arrays," Materials Letters, vol. 62, pp.
1010-1013, 2008.

P. Hidal go, B. Mendez, and J. Piqueras, "GeO; nanowires and
nanoneedles grown by thermal deposition without a catalyst,"
Nanotechnology, vol. 16, pp. 2521-2524, 2005.

Y. H. Tang, Y. F. Zhang, N. Wang, 1. Bello, C. S. Lee, and S. T. Lee,
"Germanium dioxide whiskers synthesized by laser ablation," Applied
Physics Letters, vol. 74, pp. 3824-3826, 1999.

C.-I. Wu and T. P. Hogan, "Growth of GeO; Nanowires by Thermal
Annealing," presented at Mater. Res. Soc. Symp. Proc., San
Francisco, 2006.

Y. Cui, L. J. Lauhon, M. S. Gudiksen, J. F. Wang, and C. M. Lieber,
"Diameter-controlled synthesis of single-crystal silicon nanowires,"
Applied Physics Letters, vol. 78, pp. 2214-2216, 2001.

E. E. Crisman, Y. M. Ercil, J. J. Loferski, and P. J. Stiles, "The
Oxidation of Germanium Surfaces at Pressures Much Greater Than
One Atmosphere," Journal of the Electrochemical Society, vol. 129,
pp. 1845-1848, 1982.

V. A. Nazarenko, Analytical chemistry of Germanium. N. Y.: John
Wiley & sons, 1974.

J. Oh and J. C. Campbell, "Thermal desorption of Ge native oxides
and the loss of Ge from the surface," Journal of Electronic Materials,
vol. 33, pp. 364-367, 2004.

K. Prabhakaran and T. Ogino, "Oxidation of Ge(100) and Ge(l 11)
surfaces: an UPS and XPS study," Surface Science, vol. 325, pp. 263-
271, 1995.

V. I. Davydov, Germanium. New York, London, Paris: Gordon and
Breach, Science publishers, 1966.

P. J. Wolf, T. M. Christensen, N. G. Coit, and R. W. Swinford, "Thin
ﬁlm properties of germanium oxide synthesized by pulsed laser

329

 

sputtering in vacuum and oxygen environments," Journal of Vacuum
Science & Technology A: Vacuum, Surfaces, and Films, vol. 11, pp.
2725-2732, 1993.

[100] Y. F. Zhang, Y. H. Tang, N. Wang, C. S. Lee, I. Bello, and S. T. Lee,
"Germanium nanowires sheathed with an oxide layer," Physical
Review B, vol. 61, pp. 4518-4521, 2000.

[101] A. Molle, M. N. K. Bhuiyan, G. Tallarida, and M. Fanciulli,
"Formation and stability of germanium oxide induced by atomic
oxygen exposure," Materials Science in Semiconductor Processing
poceedings of Symposium T E-MRS 2006 Spring Meeting on
Germanium based semiconductors from materials to devices, vol. 9,
pp. 673-678, 2006.

[102] K. Prabhakaran, F. Maeda, Y. Watanabe, and T. Ogino, "Thermal
decomposition pathway of Ge and Si oxides: observation of a distinct
difference," Thin Solid Films, vol. 369, pp. 289-292, 2000.

[103] D. A. Hansen and J. B. Hudson, "The adsorption kinetics of molecular
oxygen and the desorption kinetics of GeO on Ge( 100)," Surface
Science, vol. 292, pp. 17-32, 1993.

[104] A. W. Laubengayer and D. S. Morton, "Germanium. XXXIX. The
polymorphism of germanium dioxide," Journal of the American
Chemical Society, vol. 54, pp. 2303-2320, 1932.

[105] F. X. Hassion, C. D. Thurmond, and F. A. Trumbore, "On the Melting
Point of Germanium," Journal of Physical Chemistry, vol. 59, pp.
1076-1078, 1955.

[106] E. S. Greiner and P. Breidt, "Melting Point of Germanium and the
Constitution of Some Ge-Ga Alloys," Transactions of the American

Institute of Mining and Metallurgical Engineers, vol. 203, pp. 187-
188, 1955.

[107] L. M. Dennis and R. E. Hulse, "Germanium XXXV Germanium
monoxide - Germanium monosulﬁde," Journal of the American
Chemical Society, vol. 52, pp. 3553-3556, 1930.

330

 

[108] H. W. Kim, S. H. Shim, and J. W. Lee, "Cone-shaped structures of
GeOz fabricated by a thermal evaporation process," Applied Surface
Science, vol. 253, pp. 7207-7210, 2007.

[109] Z. Jiang, T. Xie, G. Z. Wang, X. Y. Yuan, C. H. Ye, W. P. Cai, G. W.
Meng, G. H. Li, and L. D. Zhang, "GeOz nanotubes and nanorods

synthesized by vapor phase reactions," Materials Letters, vol. 59, pp.
416-419, 2005.

[1 10] C.-I. Wu, "Fabrication and site speciﬁc growth of nanowires," Ph.D.
dissertation, Michigan state university, East Lansing, MI, USA, 2006.

[111] D. P. Norton, J. D. Budai, and M. F. Chisholm, "Hydrogen-assisted
pulsed-laser deposition of (001)CeOzon (001) Ge," Applied Physics
Letters, vol. 76, pp. 1677-1679, 2000.

[112] D. P. Norton, J. D. Budai, and M. F. Chisholm, "Epitaxial Electronic
Oxides on Semiconductors using Pulsed-Laser Deposition," presented
at Mat Res Soc, Fall 1999, 2000.

[113] S. Sharma, T. I. Kamins, and R. S. Williams, "Synthesis of thin
silicon nanowires using gold-catalyzed chemical vapor deposition,"

Applied Physics a-Materials Science & Processing, vol. 80, pp. 1225-
1229, 2005.

[114] M. Micoulaut, L. Corrnier, and G. S. Henderson, "The structure of
amorphous, crystalline and liquid GeOz," Journal of Physics-
Condensed Matter, vol. 18, pp. R753-R784, 2006.

[115] K. Kneipp, H. Kneipp, I. Itzkan, R. R. Dasari, and M. S. Feld,
"Surface-enhanced Raman scattering and biophysics," Journal of
Physics-Condensed Matter, vol. 14, pp. R597-R624, 2002.

[116] E. Burstein, Y. J. Chen, C. Y. Chen, S. Lundquist, and E. Tosatti,
"Giant Raman-Scattering by Adsorbed Molecules on Metal-Surfaces,"
Solid State Communications, vol. 29, pp. 567-570, 1979.

[117] A. Otto, "Raman-Spectra of (Cn)- Adsorbed at a Silver Surface,"
Surface Science, vol. 75, pp. L392-L396, 1978.

331

 

[l 18] R. L. Birke, J. R. Lombardi, and J. I. Gersten, "Observation of a

Continuum in Enhanced Raman-Scattering from a Metal-Solution
Interface," Physical Review Letters, vol. 43, pp. 71-75, 1979.

[119] C. T. Campbell, "Ultrathin metal ﬁlms and particles on oxide
surfaces: Structural, electronic and chemisorptive properties," Surface
Science Reports, vol. 27, pp. 1-111, 1997.

[120] L. Zhang, R. Persaud, and T. E. Madey, "Ultrathin metal ﬁlms on a
metal oxide surface: Growth of Au on TiOz (110)," Physical Review
B, vol. 56, pp. 10549-10557, 1997.

[121] U. Diebold, J. M. Pan, and T. E. Madey, "Growth Mode of Ultrathin
Copper Overlayers on Tioz(110)," Physical Review B, vol. 47, pp.
3868-3876, 1993.

[122] Z. Zhou, F. Xiao, L. Liu, G. Wang, and Z. Xu, "Probing single-
molecule by surface-enhanced resonance Raman scattering with

linearly and circularly polarized laser," Optics Communications, vol.
251, pp. 209-215, 2005.

[123] M. Suzuki, Y. Niidome, and S. Yamada, "Heat-induced
morphological control of gold nanoparticle ﬁlms for surface-enhanced
Raman scattering (SERS) measurements," Colloids and Surfaces A:
Physicochemical and Engineering Aspects A selection of papers
from the 11th International Conference on Organized Molecular
Films (LB! 1), June 26—30, 2005, Sapporo, vol. 284-285, pp. 388-394,
2006.

[124] M. K. Lawless and R. A. Mathies, "Excited-State Structure and
Electronic Dephasing Time of Nile Blue from Absolute Resonance

Raman Intensities," Journal of Chemical Physics, vol. 96, pp. 8037-
8045, 1992.

[125] A. Ulman, "Formation and structure of self-assembled monolayers,"
Chemical Reviews, vol. 96, pp. 1533-1554, 1996.

[126] Y. J. Lee, T. C. Jeon, W. K. Paik, and K. Kim, "Self-assembly of 1,2-
benzenedithiol on gold and silver: Fourier transform infrared

spectroscopy and quartz crystal microbalance study," Langmuir, vol.
12, pp. 5830-5837, 1996.

332

 

[127] G. Sauer, G. Brehm, and S. Schneider, "Preparation of SERS-active
gold ﬁlm electrodes via electrocrystallization: their characterization

and application with NIR excitation," Journal of Raman
Spectroscopy, vol. 35, pp. 568-576, 2004.

[128] S. H. Cho, Y. J. Lee, M. S. Kim, and K. Kim, "Infrared and Raman
spectroscopic study of 1,2-benzenedithiol adsorbed on silver,"
Vibrational Spectroscopy, vol. 10, pp. 261-270, 1996.

[129] B. Nikoobakht, J. Wang, and M. A. El-Sayed, "Surface-enhanced
Raman scattering of molecules adsorbed on gold nanorods: off-

surface plasmon resonance condition," Chemical Physics Letters, vol.
366, pp. 17-23, 2002.

[130] K. Seo and E. Borguet, "Potential-Induced Structural Change in a
Self-Assembled Monolayer of 4-Methylbenzenethiol on Au(111),"J.
Phys. Chem. C, vol. 111, pp. 6335-6342, 2007.

[131] A. Ibrahim, P. B. Oldham, D. L. Stokes, and T. VoDinh,
"Determination of enhancement factors for surface-enhanced FT-
Raman spectroscopy on gold and silver surfaces," Journal of Raman
Spectroscopy, vol. 27, pp. 887-891, 1996.

[132] G. Varsanyi, Vibrational spectra of benzene derivatives. New York:
Academic press, 1969.

[133] F. R. Dollish, W. G. Gateley, and F. F. Bentley, Characteristic Raman
frequencies of organic compounds. New York: Wiley Interscience,
1973.

[134] H. Kojima, T. Suzuki, T. Ichimura, A. Fujii, T. Ebata, and N. Mikami,
"Laser-induced ﬂuorescence of jet-cooled chlorotoluene molecules,"
Journal of Photochemistry and Photobiology A: Chemistry

3rd Japan-Sino Binational Symposium on Photochemistry, vol. 92, pp. 1-5,
1995.

[135] G. Varsanyi, Assignments for vibrational spectra of seven hundred
benzene derivatives, vol. 1. New York: John Wiley & Sons, 1974.

[136] B. H. Jun, J. H. Kim, H. Park, J. S. Kim, K. N. Yu, S. M. Lee, H.
Choi, S. Y. Kwak, Y. K. Kim, D. H. Jeong, M. H. Cho, and Y. S. Lee,

333

 

"Surface-enhanced Raman spectroscopic-encoded beads for multiplex
immunoassay," Journal of Combinatorial Chemistry, vol. 9, pp. 237—
244, 2007.

[137] Z. Ozhamam, M. Yurdakul, and S. Yurdakul, "DF T studies and
vibrational spectra of trans l,2-bis(4-pyridyl)ethy1ene and its

zinc(II)halide complexes," Journal of Molecular Structure- T heochem,
vol. 761, pp. 113-118, 2006.

[138] W. H. Yang, J. Hulteen, G. C. Schatz, and R. P. V. Duyne, "A
surface-enhanced hyper-Raman and surface-enhanced Raman
scattering study of trans-l,2-bis(4-pyridyl)ethylene adsorbed onto
silver ﬁlm over nanosphere electrodes. Vibrational assignments:
Experiment and theory," The Journal of Chemical Physics, vol. 104,
pp. 4313-4323, 1996.

[139] W. P. Grifﬁth and T. Y. Koh, "Vibrational spectra of 1,2-
benzenedithiol, 2-aminothiophenol and 2-aminophenol and their SER
spectra," Spectrochimica Acta Part A: Molecular and Biomolecular
Spectroscopy, vol. 51, pp. 253-267, 1995.

[140] J. T. Law and P. S. Meigs, "Rates of Oxidation of Germanium,"
Journal of the Electrochemical Society, vol. 104, pp. 154-159, 1957.

[141] E. B. Gorokhov, V. A. Volodin, D. V. Marin, D. A. Orekhov, A. G.
Cherkov, A. K. Gutakovskii, V. A. Shvets, A. G. Borisov, and M. D.
Efremov, "Effect of quantum conﬁnement on optical properties of Ge

nanocrystals in Ge02 ﬁlms," Semiconductors, vol. 39, pp. 1168-1175,
2005.

[142] A. A. Borshch, M. S. Brodin, V. I. Volkov, V. P. Lyakhovetskii, and
R. D. Fedorovich, "Giant nonlinear refraction in gold island ﬁlms,"
JETP Letters, vol. 84, pp. 214-216, 2006.

[143] J. N. Chen, T. Martensson, K. A. Dick, K. Deppert, H. O. Xu, L.
Samuelson, and H. X. Xu, "Surface-enhanced Raman scattering of

rhodamine 6G on nanowire arrays decorated with gold nanoparticles,"
Nanotechnology, vol. 19, pp. 275712, 2008.

[144] Z. L. Wang, "Zinc oxide nanostructures: growth, properties and
applications," Journal of Physics-Condensed Matter, vol. 16, pp.
R829-R858, 2004.

334

 

[145] M. H. Huang, S. Mao, H. Feick, H. O. Yan, Y. Y. Wu, H. Kind, E.
Weber, R. Russo, and P. D. Yang, "Room-temperature ultraviolet
nanowire nanolasers," Science, vol. 292, pp. 1897-1899, 2001.

[146] M. H. Huang, Y. Y. Wu, H. Feick, N. Tran, E. Weber, and P. D.
Yang, "Catalytic growth of zinc oxide nanowires by vapor transport,"
Advanced Materials, vol. 13, pp. 113-116, 2001.

[147] J. Y. Li, X. L. Chen, H. Li, M. He, and Z. Y. Qiao, "Fabrication of
zinc oxide nanorods," Journal of Crystal Growth, vol. 233, pp. 5-7,
2001.

[148] Y. C. Kong, D. P. Yu, B. Zhang, W. Fang, and S. Q. Feng,
"Ultraviolet-emitting ZnO nanowires synthesized by a physical vapor

deposition approach," Applied Physics Letters, vol. 78, pp. 407-409,
2001.

[149] Y. J. Ma, Z. Zhang, F. Zhou, L. Lu, A. Z. Jin, and C. Z. Gu, "Hopping
conduction in single ZnO nanowires," Nanotechnology, vol. 16, pp.
746-749, 2005.

[150] H. J. Son, K. A. Jeon, C. E. Kim, J. H. Kim, K. H. Y00, and S. Y. Lee,
"Synthesis of ZnO nanowires by pulsed laser deposition in furnace,"
Applied Surface Science: Photon-Assisted Synthesis and Processing of
Functional Materials - E-MRS-H Symposium, vol. 253, pp. 7848-
7850, 2007.

[151] W. 1. Park, D. H. Kim, S. W. Jung, and G. C. Yi, "Metalorganic
vapor-phase epitaxial growth of vertically well-aligned ZnO
nanorods," Applied Physics Letters, vol. 80, pp. 4232-4234, 2002.

[152] W. 1. Park, Y. H. Jun, S. W. Jung, and G. C. Yi, "Excitonic emissions
observed in ZnO single crystal nanorods," Applied Physics Letters,
vol. 82, pp. 964-966, 2003.

[153] M. J. Zheng, L. D. Zhang, G. H. Li, and W. Z. Shen, "Fabrication and
optical properties of large-scale uniform zinc oxide nanowire arrays

by one-step electrochemical deposition technique," Chemical Physics
Letters, vol. 363, pp. 123-128, 2002.

[154] L. Vayssieres, "Growth of Arrayed Nanorods and Nanowires of ZnO
from Aqueous Solutions," Adv. Mater., vol. 15, pp. 464-466, 2003.

335

[155] C. Xu, Z. Liu, S. Liu, and G. Wang, "Growth of hexagonal ZnO
nanowires and nanowhiskers," Scripta Materialia, vol. 48, pp. 1367-
1371, 2003.

[1.56] H. E. Unalan, P. Hiralal, N. Rupesinghe, S. Dalal, W. I. Milne, and G.
A. J. Amaratunga, "Rapid synthesis of aligned zinc oxide nanowires,"
Nanotechnology, vol. 19, pp. 255608, 2008.

[157] J. J. Wu and S. C. Liu, "Catalyst-free growth and characterization of
ZnO nanorods," Journal of Physical Chemistry B, vol. 106, pp. 9546-
9551, 2002.

[158] J. Yang, W. Z. Wang, Y. Ma, D. Z. Wang, D. Steeves, B. Kimball,
and Z. F. Ren, "High throughput grth of zinc oxide nanowires from
zinc powder with the assistance of sodium chloride," Journal of
Nanoscience and Nanotechnology, vol. 6, pp. 2196-2199, 2006.

[159] H. Y. P. Yang, S. Mao, R. Russo, J. Johnson, R. Saykally, N. Morris,
J. Pham, R. He, H.-J. Choi,, "Controlled Growth of ZnO Nanowires
and Their Optical Properties," Advanced Functional Materials, vol.
12, pp. 323-331, 2002.

[160] J. Zhang, L.-D. Sun, X.-C. Jiang, C.-S. Liao, and C.-H. Yan, "Shape
Evolution of One-Dimensional Single-Crystalline ZnO

Nanostructures in a Microemulsion System," Cryst. Growth Des., vol.
4, pp. 309-313, 2004.

[161] P. A. Hu, Y. Q. Liu, L. Fu, X. B. Wang, and D. B. Zhu, "Creation of
novel ZnO nanostructures: self-assembled nanoribbon/nanoneedle
junction networks and faceted nanoneedles on hexagonal

microcrystals," Applied Physics a-Materials Science & Processing,
vol. 78, pp. 15-19, 2004.

[162] M. Snure and A. Tiwari, "Synthesis, characterization, and green
luminescence in ZnO nanocages," Journal of Nanoscience and
Nanotechnology, vol. 7, pp. 481-485, 2007.

[163] Y. Hang Leung, A. B. Djurisic, J. Gao, M. H. Xie, and W. K. Chan,
"Changing the shape of ZnO nanostructures by controlling Zn vapor

release: from tetrapod to bone-like nanorods," Chemical Physics
Letters, vol. 385, pp. 155-159, 2004.

336

 

[164] M. C. Newton and P. A. Warburton, "ZnO tetrapod nanocrystals,"
Materials Today, vol. 10, pp. 50-54, 2007.

[165] B. D. Yao, Y. F. Chan, and N. Wang, "Formation of ZnO
nanostructures by a simple way of thermal evaporation," Applied
Physics Letters, vol. 81, pp. 757-759, 2002.

[166] F. Z. Wang, Z. Z. Ye, D. W. Ma, L. P. Zhu, and F. Zhuge, "Novel
morphologies of ZnO nanotetrapods," Materials Letters, vol. 59, pp.
560—563, 2005.

[167] L. E. Greene, M. Law, D. H. Tan, M. Montano, J. Goldberger, G.
Somorj ai, and P. D. Yang, "General route to vertical ZnO nanowire

arrays using textured ZnO seeds," Nano Letters, vol. 5, pp. 1231-
1236, 2005.

[168] M. Kitano, T. Hamabe, S. Maeda, and T. Okabe, "Growth of large
tetrapod-like ZnO crystals : I . Experimental considerations on kinetics
of growth," Journal of Crystal Growth, vol. 102, pp. 965-973, 1990.

[169] J. Park, H.-H. Choi, K. Siebein, and R. K. Singh, "Two-step
evaporation process for formation of aligned zinc oxide nanowires,"
Journal of Crystal Growth, vol. 258, pp. 342-348, 2003.

[170] M. Rajalakshmi, A. K. Arora, B. S. Bendre, and S. Mahamuni,
"Optical phonon conﬁnement in zinc oxide nanoparticles," Journal of
Applied Physics, vol. 87, pp. 2445-2448, 2000.

[171] M. A. Khan, T. P. Hogan, and B. Shanker, "Surface-enhanced Raman
scattering from gold-coated germanium oxide nanowires," Journal of
Raman Spectroscopy, vol. 39, pp. 893-900, 2008.

[172] J. Clarkson, C. Campbell, B. N. Rospendowski, and W. E. . Smith,
"Molecular recognition using surface-enhanced Raman scattering
from citrate-reduced silver colloids in non-aqueous solvents," Journal
of Raman Spectroscopy, vol. 22, pp. 771-775, 1991.

[173] S. M. Prokes, O. J. Glembocki, R. W. Rendell, and M. G. Ancona,
"Enhanced plasmon coupling in crossed dielectric/metal nanowire
composite geometries and applications to surface-enhanced Raman
spectroscopy," Applied Physics Letters, vol. 90, pp. 093105-1, 2007.

337

 

[174] T. G. Knorr and R. W. Hoffman, "Dependence of Geometric
Magnetic Anisotropy in Thin Iron Films," Physical Review, vol. 113,
pp. 1039-1046, 1959.

[175] D. O. Smith, "Anisotropy in Permalloy Films," Journal of Applied
Physics, vol. 30, pp. 264S-26SS, 1959.

[176] D. O. Smith, M. S. Cohen, and G. P. Weiss, "Oblique-Incidence
Anisotropy in Evaporated Permalloy Films," Journal of Applied
Physics, vol. 31, pp. 1755-1762, 1960.

[177] H. Konig and G. Helwig, "*Uber Die Struktur Schrag Aufgedampfter
Schichten Und 1hr Einﬂuss Auf Die Entwicklung Submikroskopischer
Oberﬂachenrauhigkeiten," Optik, vol. 6, pp. 111-124, 1950.

[178] J. P. Singh, T. Karabacak, D. X. Ye, D. L. Liu, C. Picu, T. M. Lu, and
G. C. Wang, "Physical properties of nanostructures grown by oblique

angle deposition," Journal of Vacuum Science & Technology B, vol.
23, pp. 2114-2121, 2005.

[179] M. Suzuki, W. Maekita, Y. Wada, K. Nakajima, K. Kimura, T.
. Fukuoka, and Y. Mori, "In-line aligned and bottom-up Ag nanorods

for surface-enhanced Raman spectroscopy," Applied Physics Letters,
vol. 88, pp. 203121, 2006.

[180] M. Suzuki, K. Nakajima, K. Kimura, T. Fukuoka, and Y. Mori, "Au
Nanorod Arrays Tailored for Surface-Enhanced Raman
Spectroscopy," Analytical Sciences, vol. 23, pp. 829-833, 2007.

[181] Y. Liu, J. Fan, Y.-P. Zhao, S. Shanmukh, and R. A. Dluhy, "Angle
dependent surface enhanced Raman scattering obtained from a Ag

nanorod array substrate," Applied Physics Letters, vol. 89, pp.
173134, 2006.

[182] K. Robbie, L. J. Friedrich, S. K. Dew, T. Smy, and M. J. Brett,
"Fabrication of Thin-Films with Highly Porous Microstructures,"

Journal of Vacuum Science & Technology a-Vacuum Surfaces and
Films, vol. 13, pp. 1032-1035, 1995.

[183] L. Abelmann and C. Lodder, "Oblique evaporation and surface
diffusion," Thin Solid Films, vol. 305, pp. 1-21, 1997.

338

[184] K. Hara, M. Kamiya, T. Hashimoto, K. Okamoto, and H. Fujiwara,
"Oblique-Incidence Anisotropy of the Iron Films Evaporated at Low

Substrate Temperatures," Journal of Magnetism and Magnetic
Materials, vol. 73, pp. 161-166, 1988.

[185] D. Vick, T. Smy, and M. J. Brett, "Growth behavior of evaporated
porous thin ﬁlms," Journal of Materials Research, vol. 17, pp. 2904-
2911, 2002.

[186] W. K. Choi, L. Li, H. G. Chew, and F. Zheng, "Synthesis and
structural characterization of germanium nanowires from glancing
angle deposition," Nanotechnology, vol. 18, pp. 385301, 2007.

[187] A. G. Dirks and H. J. Leamy, "Columnar microstructure in vapor-
deposited thin ﬁlms," Thin Solid Films, vol. 47, pp. 219-233, 1977.

[188] C. K. Chen, A. R. B. Decastro, and Y. R. Shen, "Surface-Enchanced
2nd-Harmonic Generation," Physical Review Letters, vol. 46, pp. 145-
148, 1981.

[189] J. Stropp, G. Trachta, G. Brehm, and S. Schneider, "A new version of
AgFON substrates for hi gh-throughput analytical SFRS applications,"
Journal of Raman Spectroscopy, vol. 34, pp. 26-32, 2003.

[190] Y. S. Tan, M. P. Srinivasan, S. O. Pehkonen, and S. Y. M. Chooi,
"Effects of ring substituents on the protective pr0perties of self-

assembled benzenethiols on copper," Corrosion Science, vol. 48, pp.
840-862, 2006.

[191] Y. J. Liu, J. G. Fan, Y. P. Zhao, S. Shanmukh, and R. A. Dluhy,
"Angle dependent surface enhanced Raman scattering obtained from a

Ag nanorod array substrate," Applied Physics Letters, vol. 89, pp.
173134, 2006.

[192] A. Wokaun, Solid state physics, vol. 38. New York: Academic, 1984.

[193] C. J. F. Boettcher, Theory of Electric Polarization, 2nd ed. New York:
Elsevier, 1973.

339

 

”'iliiiiiliILiiilii[i[iiiiliiiiiliiiiilii‘

669