University of Southern California Dissertations and Theses

Non-linear surface spectroscopy of photoswitches annd photovoltaics

(USC Thesis Other)

Non-linear surface spectroscopy of photoswitches annd photovoltaics

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content NON-LINEAR SURFACE SPECTROSCOPY OF PHOTOSWITCHES AND
PHOTOVOLTAICS
by
David Taylor Valley
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulllment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMISTRY)
May 2013
Copyright 2013 David Taylor Valley
Acknowledgments
I would like to graciously acknowledge a number of people who supported me for my time
at USC. First of all I would like to thank my advisor Alex Benderskii, for always being
willing to discuss new ideas and data and his insightful comments. I'm particularly for
the opportunity he gave me to join his group and the support he's provided since then.
I would also thank the members of the Benderskii group, Sergey Malky, Fadel Shal-
hout, Purnim Dhar, Ian Craig, Sean Roberts and, of course, Misha Vinaykin. Sergey
and Fadel were especially helpful with the setup, maintenance and general running of our
laser. I guess I'd also like to thank the laser itself for avoiding the common derogatory
personications it's kind invites. I would also like to thank my summer REU student
Matthew Onstott who was great to have around during the dark days of organic syn-
thesis. The synthesis was also performed with great assistance from Janet Olsen for
help with many laboratory techniques, Barry Thompson and Travis Williams for help in
devising a synthetic protocol and Dana Mustafa for help with working up NMR data.
I'd really like to thank Steve Bradforth serving on both my qualifying and thesis
committee, for our many useful conversations throughout the years both at school and
on the train. His input to my research and his excellent sense of humor made grad
school much more enjoyable. I would like to thank the members of the Bradforth group,
Saptaparna Das, Anirban Roy, Robert Seidel, Konstantin Kudinov, Joseph Masterson,
Diana Suern, Tom Zhang, Chris Rivera and Elsa Couderc. Saptaparna and Sean were
especially helpful with their assistance for the two photon absorption measurements.
ii
Barry Thompson and the members of the Thompson group, Petr Khlyabich, Beate
Burkhart, and Alia Latif, were wonderful collaborators with the SHG work and were
always prompt with supplying very high quality samples for study. I would like to thank
Aiichiro Nakano for being gracious enough to take the time from his schedule to join my
committee. Melancha Gupta was a lot of fun to ride the train with and discuss science.
I would like to thank Steve Cronin and his group, Jesse Theiss, Wayne Hung, Rajay
Kumar, Memo Aykol, Zu-Wei Liu, Chia-Chi Chang, Prathamesh Pavaskar, Wenbo Hou,
Rohan Dhall, Fernando Souto, Adam Bushmaker, Shun-Wen Chang, and I-Kai Hsu for
years of scientic help and friendship.
Of course none of grad school would run without the assistance of the Chemistry
department's wonderful administrative sta. Michele Dea, Katie McKissick, Heather
Connor, and Valerie Childress all managed to smooth over the worst of the red tape the
school could throw in my direction.
I would like to thank my parents, John and Andr ee Valley, my brother, sister in
law, and niece, Matt Valley, Claire Lundberg and Sophia Valley, my In-laws Steve and
Andrey Ferry for their support and especially my aunt and uncle Marcy and George
Valley for their generosity and home in Los Angeles.
Lastly, I would of course like to thank my wonderful wife Vivian for her love, support,
scientic input, L
A
T
E
X help, and especially for proofreading the thesis.
David Valley
December 2012
Los Angeles, CA
iii
Table of Contents
Acknowledgments ii
List of Figures vii
List of Tables xii
Chapter 1: Nonlinear Spectroscopy at the Interface 1
1.1 Second Harmonic Generation Spectroscopy . . . . . . . . . . . . . . . . . 2
1.1.1 SHG of Organic Electronic Materials . . . . . . . . . . . . . . . . . 4
1.2 Sum Frequency Generation Spectroscopy . . . . . . . . . . . . . . . . . . . 4
1.2.1 SFG of SAMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 Theory of Even Ordered Nonlinear Spectroscopy . . . . . . . . . . . . . . 7
1.3.1 Construction of
(2)
. . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.2 Sum Frequency Generation . . . . . . . . . . . . . . . . . . . . . . 11
1.4 Outline of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Chapter 2: Experimental Details of Nonlinear Surface Spectroscopy 14
2.1 Sum Frequency Generation . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.1.1 Setup Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.1.2 SFG with in situ illumination . . . . . . . . . . . . . . . . . . . . . 17
2.1.3 SFG tting procedure . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2 Second Harmonic Generation . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2.1 SHG System Overview . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.2 SHG design considerations . . . . . . . . . . . . . . . . . . . . . . . 23
2.2.3 TOPAS 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.2.4 Spectral Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.5 Alignment procedure and tricks . . . . . . . . . . . . . . . . . . . . 32
2.2.6 SHG data collection . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.2.7 Fitting Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.2.8 Normalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Chapter 3: XFROG 39
3.1 What Information is in SFG Spectra? . . . . . . . . . . . . . . . . . . . . 39
3.2 The Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.3 XFROG experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
iv
3.3.1 Experimental Details . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.4 XFROG, Programmatically . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.4.1 Why is Deconvolution Dicult? . . . . . . . . . . . . . . . . . . . 48
3.5 Using XFROG to Determine S(t) . . . . . . . . . . . . . . . . . . . . . . . 50
3.6 Time Gradient XFROG . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Chapter 4: SFG of an Azobenzene Functionalized SAM 53
4.1 Photochemistry of Azobenzene . . . . . . . . . . . . . . . . . . . . . . . . 53
4.2 Self-Assembled Monolayers . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2.1 Azobenzene functionalized SAMs . . . . . . . . . . . . . . . . . . . 56
4.2.2 Azobenzene SFG literature . . . . . . . . . . . . . . . . . . . . . . 56
4.3 Synthetic Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.4 Synthetic Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.4.1 Synthesis of 4-(4-(Hydroxymethyl)phenyl)diazenyl benzonitrile . . 61
4.4.2 Protection, esterization and deprotection . . . . . . . . . . . . . . 62
4.4.3 Formation of SAM . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.5 Characterization of Products . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.5.1 NMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.5.2 FTIR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.5.3 UV-Vis Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.6 SFG on Azobenzene Functionalized SAMs . . . . . . . . . . . . . . . . . . 65
4.6.1 Isomerization behavior . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.7 Orientation of Azobenzene Functionalized SAMs . . . . . . . . . . . . . . 71
4.7.1 Orientational Preference upon Isomerization . . . . . . . . . . . . . 76
4.7.2 Azobenzene Isomerization Under Other Polarizations . . . . . . . 81
4.7.3 Cone Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.8 Discussion and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Chapter 5: Second Harmonic Generation Spectroscopy of Organic Pho-
tovoltaic Materials 90
5.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5.1.1 Device architectures . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.1.2 OPV Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.2 Synthesis and Sample Preparation . . . . . . . . . . . . . . . . . . . . . . 94
5.2.1 UV-VIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
5.2.2 Sample Roughness and Thickness Measurements . . . . . . . . . . 96
5.2.3 Complex Refractive Index . . . . . . . . . . . . . . . . . . . . . . . 96
5.2.4 Two Photon Absorption of P3HT . . . . . . . . . . . . . . . . . . . 98
5.3 SHG of the P3HT / C
60
Interface . . . . . . . . . . . . . . . . . . . . . . . 100
5.3.1 A Model for Extracting Surface SHG . . . . . . . . . . . . . . . . . 101
5.3.2 Which Surface are we Measuring? . . . . . . . . . . . . . . . . . . 105
5.3.3 Fitting Spectra with Thickness Dependent Model . . . . . . . . . . 108
5.3.4 Interpretations of P3HT orientation from SHG spectra . . . . . . . 112
5.4 SHG of the Red-Absorbing Co-Polymer/ Air Interface . . . . . . . . . . . 115
5.5 Conclusion and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
v
Bibliography 122
Appendix A:SFG Signal Fitting Programs 139
A.1 File Input: WinSpec) Mathematica . . . . . . . . . . . . . . . . . . . . 139
A.2 SFG CN Signal Fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
A.3 Example Fitting Script . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
A.4 Fitting for response function. . . . . . . . . . . . . . . . . . . . . . . . . . 141
Appendix B: SHG Signal Fitting Programs 144
B.1 Spike Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
B.2 Gaussian Fitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
B.3 Fitting Script . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Appendix C: XFROG Mathematica Code 147
C.1 XFROG Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
C.2 XFROG Queues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
C.3 input Files and Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
C.4 Graphing Programs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Appendix D:SFG Orientational Analysis 161
D.1 Initialization and Indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
D.2 Fresnel Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
D.3
(2)
elements for nitrile modes . . . . . . . . . . . . . . . . . . . . . . . . 161
D.4 Amplitude Ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
D.5 Isomerization Models Up and Down . . . . . . . . . . . . . . . . . . . . . 162
D.6 Construction of Circle Model . . . . . . . . . . . . . . . . . . . . . . . . . 162
Appendix E: SHG Modeling 164
E.1 Load Index Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
E.2 Load Index Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
E.3 Dene Fresnel Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
E.4 Dene Normalization Factors . . . . . . . . . . . . . . . . . . . . . . . . . 165
E.5 Fitting Thickness Dependent SHG Spectra . . . . . . . . . . . . . . . . . 165
vi
List of Figures
2.1 Schematic of the vibrational SFG setup. Generated 800 nm light split
into two parts. 40% compressed, stretched with etalon. 60% split using
OPA and dierence taken in DFG to generate ngerprint IR. Both beams
are focused on sample stage (Detail of stage shown in Figure 2.2 and SFG
signal is polarization selected and collected on CCD. . . . . . . . . . . . . 17
2.2 Detail of SFG stage. Purge box full of dry, CO
2
depleted air. . . . . . . . 18
2.3 SFG Setup with UV illumination. Xenon lamp white light source ltered
with a 340 nm band-pass lter and focused onto sample stage with spher-
ical lens. UV light modulated with shutter. . . . . . . . . . . . . . . . . . 19
2.4 Overview of the SHG optical setup. Picosecond 400 nm generated from
40% of 800 nm selected, split and stretched using a pair of grating com-
pressors (Detail, gure 2.6) then summed in BBO crystal. 400 nm light
used to pump OPA generating narrowband tunable light 470 nm - 2000
nm. Light ltered and focused on sample (Detail, Fig. 2.5. Signal col-
lected, polarization selected and collected on CCD. . . . . . . . . . . . . 22
2.5 Detail of the sample stage for SHG setup. . . . . . . . . . . . . . . . . . 23
2.6 Detail of the compressors used for 400 nm generation. Compressor 2
generally left in place between SFG and SHG setups. Compressor 1 used
for SFG and SHG and requires tuning. Tuning each compressor both
changes chirp and relative delay of the 800 nm laser pulse. After the
100 cm lens both pulses are summed though a BBO crustal to produce
transform limited picosecond 400nm pulse. . . . . . . . . . . . . . . . . . . 27
2.7 TOPAS idler output. Power curve details mean and standard deviation
from 5 dierent readings. . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.8 Relative error from gure 2.7. Normal spectra range for SHG is 950 - 1600
nm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.9 Spectra width from SHG o of Y cut quartz with idler tuning. . . . . . . 31
2.10 Control and measurement dependancies of LabVIEW program. . . . . . 33
2.11 Left: A set of SHG spectra taken at a 1 nm spacing. Right: Normalized
SHG spectra t to a single Gaussian function. . . . . . . . . . . . . . . . . 35
3.1 A sketch of the XFROG algorithm. The two pieces of input data are the
captured spectra, I
XFROG
(;!), and the electric eld of the visible laser,
E
vis
(t). The deconvolution is represented with
~
f(
;t) in the center. The
output result from this is the explicit electric eld of the IR pulse, E
IR
(t). 43
vii
3.2 Three dierent measured XFROG images. Each image is two separate
time time dependent SFG scans stitched together and interpolated into
a 512x512 grid of interpolated frequency ! versus delay . For the left
spectra the IR beam was stretched with both a 6.3 mm CaF
2
plate and
an IR Waveplate. The center image the beam only passed through the IR
Waveplate. In the right image a 1 mm piece of Ge was used to stretch the
IR pulse. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.3 Solved intensity and argument ofE
IR
from the measured XFROG spectra
shown in Fig. 3.2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.4 A representation of
~
f(
;t) with three dierent cuto margins overlaid. . 51
3.5 The results of the intensity (red) and argument (blue) of E
IR
resulting
from the three cutos performed on
~
f(
;t) in gure 3.4. . . . . . . . . . 52
4.1 A cartoon of the isomerization excited states azobenzene interacts with
upon isomerization. Shown is the excitation to S
2
(

) for trans azoben-
zene and excitation to the S
1
(n

) for cis. . . . . . . . . . . . . . . . . . . 54
4.2 Some previously used azobenzene surface functionalizing molecules from
the literature. I: From Jaschke et al. 1995. II: From Yu et al. 1996 III:
From Dietrich et al. 2008 IV: The system synthesized for these studies. . 59
4.3 Synthetic scheme for the azobenzene functionalized alkane thiol: a: Oxida-
tion of 4-aminobenzonitrile with 2 equivalents of Oxone. b: Addition of 1
equivalent of 4-aminobenzoalcohol. c: Protection of 15-mercaptopentadecanoic
acid with trityl chloride. d: Esterization between protected thiol and
AZO. e: Deprotection with tri
uoroacetic acid. . . . . . . . . . . . . . . . 61
4.4
1
H NMR of compound IV . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.5 FTIR of compound IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.6 Trans! cis Isomerization of AZO molecule in DCM solvent under 340nm
UV light. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.7 Detail from Fig. 4.6. Trans! cis Isomerization of AZO molecule in DCM
solvent under 340nm UV light. . . . . . . . . . . . . . . . . . . . . . . . . 66
4.8 Thermal cis! trans isomerization of AZO molecule in DCM solvent over
24 hours. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.9 Normalized PPP polarized SFG spectra from a set of dierent percentage
formulated SAMs. The spectral peak at 2220 cm
1
is from the nitrile
group on the azobenzene functionalized. The broad background is the
strong non-resonant SFG signal from gold. . . . . . . . . . . . . . . . . . . 67
4.10 AmplitudeB
CN
of the resonant CN stretch signal in PPP polarized VSFG
spectra (Normalized to linear relation), as a function of the solution frac-
tion of azobenzene-functionalized precursor molecules used to form SAM.
Dierent symbols correspond to dierent synthetic batches and experi-
mental preparations of the SAMs. . . . . . . . . . . . . . . . . . . . . . . . 68
4.11 Normalized SFG spectra of PPP (top) and SSP (bottom) polarization of
samples between 0% and 100% azobenzene terminated C-H stretch modes. 69
viii
4.12 Left: PPP SFG spectra for the nitrile stretch band of a 66% azobenzene
lm. Displayed are the experimental data for the lm under 340 nm
illumination (blue) and without (red). The spectra are t with the non-
resonant portion shown in yellow and the resonant portion shown on the
bottom. Right: Cartoon of mixed SAM on gold surface, formed from a
mixture of azobenzene functionalized thiol and dodecane thiol. . . . . . . 70
4.13 Change in amplitude of nitrile
(2)
Res
for 100%, 66% and 50% azobenzene
SAMs taken at 10 second intervals upon isomerization with 340 nm light
and thermal relaxation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.14 Change in amplitude of nitrile
(2)
Res
for 66% azobenzene SAMs taken at
10 second intervals upon isomerization with 340 nm light and thermal
relaxation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.15 A: Change in amplitude of the nitrile
(2)
Res
under 340 nm illumination
for 100%, 66% and 50% azobenzene samples versus time. B: Change in
amplitude from the nitrile
(2)
Res
for thermal relaxation in the dark following
10 min exposure to 340 nm light, for 100%, 66% and 50% azobenzene
samples versus time. Data sets were averaged over multiple isomerization
cycles. Solid lines show exponential ts described in the text. . . . . . . . 73
4.16 Left: PPP SFG spectra for the nitrile stretch band of a 66% azobenzene
lm. Displayed are the experimental data for the lm under 340 nm
illumination (blue) and without (red). The spectra are t with the non-
resonant portion shown in yellow and the resonant portion shown on the
bottom. Right: Cartoon of mixed SAM on gold surface, formed from a
mixture of azobenzene functionalized thiol and dodecane thiol. . . . . . . 77
4.17 Three dierent scenarios for isomerization used: (1) where isomerization
yields equal amounts of \Up" and "Down" states. (2) where only the
\Up" states are free to isomerize (3) where only the \Down" molecules
can isomerize. The three cones (\Both": Orange,\Up": Blue, \Down":
Red) map out the expected
(2)
Res
change upon UV exposure for the three
models as a function of isomerization fraction at saturation. The cones
represent the range of tilt angles = 45

60

of the trans isomer derived
above (4.16). The black horizontal bar represented the experimentally
observed value of
(2)
Res
upon isomerization, 0.91. . . . . . . . . . . . . . 80
4.18 Simulation expected PPP polarization signal change upon isomerization
for Up (Red), Both (Orange) and Down (Blue) models done at 45 (solid)
and 60(dashed) degrees of nitrile o surface normal. (the results shown
here are redundant with gure 4.17, however this is included for compar-
ison with gure 4.20, and gure 4.22. . . . . . . . . . . . . . . . . . . . . 82
4.19 Simulation expected PPP polarization signal change for unity isomeriza-
tion with regard to trans angle of nitrile from surface normal. Evaluated
for models Up (Red), Both (Orange) and Down (Blue). . . . . . . . . . . 82
4.20 Simulation expected SSP polarization signal change upon isomerization
for Up (Red), Both (Orange) and Down (Blue) models done at 45 (solid)
and 60(dashed) degrees of nitrile o surface normal. . . . . . . . . . . . . 83
ix
4.21 Simulation expected SSP polarization signal change for unity isomeriza-
tion with regard to trans angle of nitrile from surface normal. Evaluated
for models Up (Red), Both (Orange) and Down (Blue). . . . . . . . . . . 83
4.22 Simulation expected SPS polarization signal change upon isomerization
for Up (Red), Both (Orange) and Down (Blue) models done at 45 (solid)
and 60(dashed) degrees of nitrile o surface normal. . . . . . . . . . . . . 84
4.23 Simulation expected SPS polarization signal change for unity isomeriza-
tion with regard to trans angle of nitrile from surface normal. Evaluated
for models Up (Red), Both (Orange) and Down (Blue). . . . . . . . . . . 84
4.24 Visualization of the angle distribution of before and after isomerization
for \Cone" model. For each starting delta distribution represented by an
arrow nal distribution is the projection of the cone below onto . The
grey plane represents the surface plane by which is calculated from. . . 87
4.25 \Cone" model evaluated at 45

(solid) and 60

(dashed) for PPP(orange),
SSP (Red) and SPS(Blue) polarizations with regard to isomerization ratio. 88
4.26 \Cone" model evaluated with regard to starting angle for PPP(orange),
SSP (Red) and SPS(Blue) polarizations. . . . . . . . . . . . . . . . . . . . 88
5.1 The three polymers synthesized for that were studied with the SHG spec-
troscopy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
5.2 Absorption spectra for 4 layers of P3HT spun on glass and with 15 nm
C
60
evaporated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.3 Absorption spectra for 4 dierent layers of P3HTT-10%TPD, and P3HTT-
10%DPP spun on glass. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
5.4 AFM taken of 600 RPM spun P3HT lm. Black Scale bar corresponds
to 1 m. Color bar goes from -16nm to 16nm vertical displacement. The
RMS roughness for this lm is 3.9nm. . . . . . . . . . . . . . . . . . . . . 98
5.5 The real (n) and imaginary (k) components of the index of refraction
for polymers P3HT, P3HT-10%TPD, P3HT-10%DPP and lm of C
60
as
calculated by the Kramers { Kronig relation, Eq 5.2. . . . . . . . . . . . . 99
5.6 One and two photon absorption of poly(3-octylthiophene) in tetrahydro-
furan. This gure is adapted from Figures 1 and 4 from Pfeer, et al.,
1993 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.7 Raw SHG spectra from neat P3HT lms and P3HT lm with 15 nm C
60
evaporated on top for PPP, PSS and S45 polarizations. . . . . . . . . . . 102
5.8 The calculated ratio SHG signal from top, P3HT/air interface to bottom,
P3HT/glass interface from Eq. 5.13. . . . . . . . . . . . . . . . . . . . . . 106
5.9 The angle expected of SHG signal for the P3HT/air interface (blue),
P3HT/glass interface (red), and the P3HT/C
60
interface (brown). . . . . 107
5.10 Thickness dependent PSS SHG signal at 515nm for neat P3HT and P3HT/C
60
lms (points) with t curves (lines) from equation 5.7. Each thickness has
4 redundant data points for better t. The Y intercept of the t on the
graph corresponds to the extrapolated homodyne surface dependent signal.108
5.11 PPP surface dependent homodyne SHG spectra of t surface,j
(2)
Surface
j
2
,
and bulk,j
(2)
Bulk
j
2
, P3HT/Air interface and P3HT/C
60
. . . . . . . . . . . 109
x
5.12 Fit PSS homodyne SHG spectra of t surface, j
(2)
Surface
j
2
, and bulk,
j
(2)
Bulk
j
2
, for neat P3HT and P3HT/C
60
lms. . . . . . . . . . . . . . . . . 110
5.13
zyy
element of
(2)
extracted from PSS polarization SHG spectra. Linear
UV-Vis absorption spectrum is also shown. . . . . . . . . . . . . . . . . . 111
5.14 Elements of
(2)
extracted from PPP polarization SHG spectra. ValuesA,
B,C, andD used are from table 5.1. Linear UV-Vis absorption spectrum
is also shown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.15 A cartoon of the band diagram of the P3HT/C
60
interface. Left side
shows the band diagram for P3HT and right for C
60
. The three arrows
represent the bulk absorption (green), interfacial charge transfer state
(red) and interfacial HOMO - LUMO transition (blue). . . . . . . . . . . . 114
5.16 Raw SHG spectra from neat P3HT, P3HTT-10%DPP, and P3HTT-10%TPD
lms for PPP, and PSS polarizations. . . . . . . . . . . . . . . . . . . . . 116
5.17 Fit PPP polarization homodyne SHG spectra of t surface,j
(2)
Surface
j
2
,
and bulk,j
(2)
Bulk
j
2
, P3HTT-10%DPP, and P3HTT-10%TPD lms. . . . . 117
5.18 Fit PSS polarization homodyne SHG spectra of t surface,j
(2)
Surface
j
2
,
and bulk,j
(2)
Bulk
j
2
, P3HTT-10%DPP, and P3HTT-10%TPD lms. . . . . 118
5.19 Elements of
(2)
extracted from PPP polarization SHG spectra for the
P3HTT-10%DPP/air, and P3HTT-10%TPD/air interfaces.. Values A,
B, C, and D used are from table 5.1. . . . . . . . . . . . . . . . . . . . . . 119
5.20
zyy
element of
(2)
extracted from PSS polarization SHG spectra for the
P3HTT-10%DPP/air, and P3HTT-10%TPD/air interfaces. . . . . . . . . 120
A.1 Output of script showing t
res
. . . . . . . . . . . . . . . . . . . . . . . 142
B.1 Fitter Output if PrintQM is set to"True". . . . . . . . . . . . . . . . . . 145
C.1 Amplitude and argument of deconvolved IR pulse with t for second order
and third order chirp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
xi
List of Tables
5.1 Factorization components for tting of I
PPP
(2!) . . . . . . . . . . . . . . 105
xii
Chapter 1
Nonlinear Spectroscopy at the
Interface
Molecules at the surface have always been a subject of fascination for scientists. The
molecules reside in a completely dierent environment from their bulk counterparts, and
while there are comparatively few molecules on the surface, the signicance of their
properties far out weights their number. Catalytic reactivity, surface tension, vapor
pressure, corrosion, and electrochemistry are all surface dominated processes. To the
spectroscopist surfaces are especially intriguing and elusive. Using linear spectroscopic
methods it is extremely dicult to isolate the tiny surface signal from the massive bulk
response.
A number of spectroscopic methods have been developed to examine surface
molecules each with individual advantages. Attenuated total re
ection Fourier trans-
form infrared (ATR-FTIR) spectroscopy[1, 2], near eld scanning optical microscopy
(NSOM)[3], surface enhanced Raman spectroscopy (SERS)[4] or tip enhanced Raman
spectroscopy(TERS)[5, 6] and photoacoustic spectroscopy[7, 8] each oer a degree of
surface specicity in their measurements. There are also a number of non-optical
spectroscopic methods such as inelastic scanning tunneling spectroscopy[9, 10], neutron
scattering[11, 12], and X{ray scattering[13, 14] that can also be used to look at surfaces,
although these measure dierent physical quantities than truly optical spectroscopies.
One common issue with most of these methods is that they require extreme environ-
ments to measure surfaces: the surfaces may need to be under high vacuum, at low
1
temperature, or adjacent to an apparatus. These requirements both restrict the class of
amenable interfaces to study and often perturb the system of study.
To address this problem, a class of even ordered nonlinear spectroscopies were devel-
oped that are forbidden in bulk centrosymmetric media under the electric dipole approx-
imation. At the interface, however, the symmetry is broken allowing the process to yield
surface specic spectroscopic signal. This can be accomplished either with photon of the
same energy, second harmonic generation (SHG), or by summing two dierent photon
energies, sum frequency generation (SFG). The main advantage of these even order non-
linear spectroscopic methods is their ability to obtain surface-selective information with
minimal perturbation of the local environment. Furthermore, these methods are widely
adaptable to a variety of systems.
There is relatively little restriction on the environment of the surface under investi-
gation, meaning these methods really shine in studying hot, wet, or buried interfaces.
These spectroscopies also provide a number of other advantages such as the ability to
perform orientational analysis of an ensemble of chromophores, and the capability for
picosecond or femtosecond time-resolved, surface specic dynamical measurements.
1.1 Second Harmonic Generation Spectroscopy
Although the second harmonic generation phenomenon is most widely used for light
conversion with the introduction of highly ecient non-centrosymmetric crystals such
barium borate (BBO) and monopotassium phosphate (KDP), this chapter is primarily
concerned with the spectroscopic SHG applications.
The original reported SHG spectra[15] was famously not reported as the peak at
2! but declared an error by the Physical Review editors and removed from the graph.
However, owing to rigorous theoretical work by a number of groups in the subsequent
years[16, 17, 18, 19] this was proven not to be an erroneous result.
2
Typically SHG spectroscopy is used in one of three resonance conditions: either the
primary input beam at frequency ! is resonant with the sample, the doubled signal of
frequency 2! is resonant, or both are. Many of the surface SHG studies, instead of
relying on a spectral approach by tuning the laser wavelength, stay xed at a single
wavelength and monitor modulation of SHG intensity through changes in surface chro-
mophore density, surface morphology, or the surface layer shifting on and o resonance.
SHG has been used as a tool to look at surfaces such as the water/air[20] and
water/alkane[21]. It is sensitive as a tool to study the molecular orientation at the inter-
face and can be used in situ to look at chemical reactions[22], acid base equilibrium,
[23], electrochemistry[24], phase transitions[25] and ionic concentration[20, 26, 27]. In
most of these studies SHG was not used to study the spectral character of the surface
molecules, but instead assigned the change of SHG intensity to structural changes at the
interface.
As SHG requires relatively simple optical set-up, the process has also been used
for high resolution confocal microscopy. For many biological applications SHG pro-
vides high contrast for imaging collagen and other protein structure in tissue[28, 29].
SHG microscopy has also been used to facilitate the screening process with protein
crystallization[30, 31].
A large subeld of SHG is devoted to electric eld induced second harmonic gener-
ation (EFISH). This method takes advantage of the bulk SHG enhancement of signal
which is proportional to the applied electric eld times the third order polarization[32].
EFISH has been used as an analytical tool in combination with static or RF elds to
determine the second order polarization of materials[33, 34]. This method has also been
adapted as an ultrafast transient spectroscopy by examining the decay from an optically
pumped substrates[35, 36]
3
1.1.1 SHG of Organic Electronic Materials
Conductive polymers and other electronically active organic lms have attracted atten-
tion recently for integration into eld eect transistors, light emitting diodes, photo-
voltaics, and other optoelectronics. For many of these applications electronic properties
and morphology of the surfaces are critical for to device performance. Bulk
(2)
allowed
organic crystalline materials have also been proposed for use in light conversion applica-
tions.
Although SHG activity is formally forbidden in the electric dipole approximation
in most of these centrosymmetric materials, the highly delocalized electronic states in
many of these organic systems allows for SHG to be weakly bulk allowed through higher
order processes such as electronic quadruple or magnetic dipole transitions. Attempts to
gather surface specic information from these materials have often been frustrated by the
complicated morphology of these systems. For example, MEH-PPV was shown to have
a electronic quadruple allowed bulk response dependent on the crystallite concentration
within neat lms. The crystalline fraction depended on lm preparation factors such
as spin speed and annealing conditions[37]. The bulk generation signal from copper
phthalocyanine lms have also been attributed to electric quadruple transitions[38] while
bulk generation from C
60
lms has been attributed to magnetic dipole transitions[39].
1.2 Sum Frequency Generation Spectroscopy
The technique of sum frequency generation was developed to provide a surface spe-
cic spectroscopic method capable of resolving the ngerprint IR molecular vibrational
region. The technique was originally developed in the mid eighties [40, 41, 42, 43], and
has several advantages: it is both inherently surface selective, as SHG, but also it is
a zero background spectroscopic technique as the upconversion of the signal into the
visible spectrum carries a unique wavevector from either of the re
ected input beams.
In the SFG experiment one of the incumbent laser beams is resonant with the substrate
4
while a second, often o-resonant, beam is used to upconvert the signal. The SFG
signal carries information not only about the spectroscopic selectivity of the interfacial
molecules probed but also about the chromophore concentration, molecular orientation,
and vibrational dephaseing time.
As the eld developed, a wide variety of systems have been explored. A number
of biological interfaces have been targets of investigation by SFG spectroscopy. The
importance of interfaces in biological processes as well as SFG's ability to probe buried
interfaces in situ make it a powerful tool for retrieving information. The formation and
structures of systems such as the lipid bilayer interface[44], phopholipids [45], interfacial
proteins [46, 47]and chiral protein aggregates[48, 49, 50] have all been studied.
Both the physisorption and chemisorption of molecules on metal surfaces have been a
subject of study with SFG. The properties of molecules on metal surfaces are critical for
a number of applications and areas in the elds of catalysis, electrochemistry, and surface
passivation. Many dierent metal surfaces have been studied including zinc[51], copper,
platinum, gold[52], and steel[53]. VSFG has provided valuable information for deter-
mining the molecular intermediates from catalytic surfaces[54, 55, 56, 57]. In addition
to metals, molecularly capped inorganic systems such as the hydrogen[58] terminated
silicon surface and the methyl terminated[59] silicon surface have been studied.
The organization of molecules at the liquid interface has also been a target of a great
number of studies. A series of papers have looked at the air/water interface focusing on
dierent spectroscopic bands such as the free OH stretch[60, 61, 62, 63], and the HOH
bend[64]. Other water interfaces have been studied such as water/quartz[65], water/ionic
crystal[66], water/organic[67, 68] and ice[69], as well as the eects of ionic concentration
on these surfaces[70, 71]. The ionic liquid/air or ionic liquid/vacuum interface have also
been studied as they represent a model system for with comparison to high vacuum
surface specic techniques[72, 73, 74].
Although SFG is normally performed in a time agnostic manner, without specically
treating the temporal information inherent in SFG spectroscopy, a number of studies have
5
targeted this information specically. Information about the vibrational free induction
decays of surface chromophores can be studied both through line shape analysis[75] and
through explicit time dependent approaches such as direct measurement at the decays
and coherent vibrational quantum beating[76]. The dierential temporal decays between
the resonant vibrational response of the surface chromophores and the non-resonant
electronic response of the surfaces have been used for suppression of the nonlinear signal
using a delay between the two pulses used and a temporally asymmetrical visible pulse[77]
SFG spectroscopic methods necessitate the use of highly temporally compressed
laser pulses as driving non-linear spectroscopic process requires massive electric eld
intensities. Because of this, as spectroscopic setups have proceeded through picosec-
ond into femtosecond laser pulse width, it was natural to use this time resolution to
study transiently perturbed systems. Pump SFG-probe setups, where a non{coherent
perturbation has been applied to the system, have studied thermal shock in self assem-
bled monolayers[78, 79]. An IR-pump, SFG probe setups were built to look at the
water surface[80, 81]. Although this setup explicitly looks at vibrational couplings at
the surface, the IR pump and SFG probe interact with chromophores through dierent
sets of selection rules. Some groups have also built true coherent heterodyne-2D SFG
setups[82, 83].
1.2.1 SFG of SAMs
One area of intense study with SFG is self-assembled monolayers (SAMs). SAMs are a
class of surface coating that take advantage of the strong and easily formed gold-thiolate
bond allowing well-ordered forests of organic molecules to be easily synthesized on metal
surfaces. SAMs, as a surface modication, are an obvious target for study using SFG.
A number of fundamental studies have looked at aspects purely alkane SAMs including
their spectroscopic signals [51, 84], ordering[85, 86, 87], defects[88], and response to
thermal shock[78, 79]. These studies have been performed both on
at gold surfaces and
on nanoparticles[89, 90]. The well ordered alkane systems that SAMs produce also serve
6
as a good platform base for chemical functionalization of a surface. SFG studies have
been performed on surface attachments such as peptides[91] and carbohydrates[92].
1.3 Theory of Even Ordered Nonlinear Spectroscopy
The general theory of surface specic, even ordered, nonlinear molecular spectroscopy
can be derived using perturbation theory. These derivation can also be found in various
forms in a number of papers[93, 94, 95, 96, 97].
At low electric elds the polarizability of a medium, that is the tendency of atoms
and the electron cloud to deform, is regarded as linear. That is the polarization is
linear with the magnitude of electric eld,
~
P =
~
E. At high electric eld intensities
the polarizability of materials can no longer be regarded as a linear. Instead it can be
represented with a Taylor expansion introducing higher order terms.
~
P =
(1)
~
E +
(2)
:
~
E
~
E +
(3)
:
~
E
~
E
~
E +::: (1.1)
The expansion is now written in terms of the electric eld,
~
E, and susceptibility
terms, . During spectroscopic measurement materials interact with all the terms at
once, the spectroscopic setup can be manipulated to be selectively sensitive to certain
terms and in fact specic elements of a specic term.
First let us consider a case with two incumbent beams of frequencies !
1
and !
2
where we are looking for the generated !
3
= !
1
+!
2
. We can write the second order
polarization in terms of the two incoming beams, E(!
1
), E(!
2
), and the second order
nonlinear susceptibility
(2)
.
~
P
(2)
=
(2)
:
~
E(!
1
)
~
E(!
2
) (1.2)
7
Lets look at this equation under the inversion operator ^ as required in centrosym-
metric media. This will
ip the sign of the electric elds, ^
~
E(!
1
) =
~
E(!
1
) and
^
~
E(!
2
) =
~
E(!
2
), and will also
ip the sign of the polarization ^
~
P
(2)
=
~
P
(2)
. However
if we look at the case of a centrosymmetric medium ^
(2)
=
(2)
(the material properties
are invariant under inversion). This now gives us the expression,
P
(2)
=
(2)
(E(!
1
))(E(!
2
)) (1.3)
so that

(2)
E(!
1
)E(!
2
) =(
(2)
(E(!
1
))(E(!
2
)))
=(
(2)
E(!
1
)E(!
2
)) (1.4)

(2)
=
(2)
(1.5)
This can only be resolved when
(2)
= 0. Thus for centrosymmetric media, within
the dipole approximation, there will not be any second order spectroscopic signal. This
proof can trivially be extended to any even ordered spectroscopy.
However when there is an interface, centrosymmetry is inherently broken and thus
we can have a nonzero
(2)
. Within this formulation we can now consider some dierent
conditions for which to extract
(2)
signal. The rst condition will be for second harmonic
generation from one beam, thus E(!
1
) = E(!
2
). The second condition is for sum
frequency generation whereE(!
1
)6=E(!
2
) and both beams come from unique directions.
1.3.1 Construction of
(2)
Now we consider the meaning of term
(2)
, the macroscopic susceptibility of the surface,
that is composed from the ensemble of individual molecular hyperpolarizibilities . To
8
construct
(2)
we consider the individual molecular response the multiplication over two
dierent linear transition terms summed over the all states. For the case of VSFG with
a resonant IR beam we get the term,

R
lmn
=
1
2~
X

hgj
lm
jvihj
n
jgi
!

!
IR
i

(1.6)
withhgj
lm
jvi as the Raman polarizability andhj
n
jgi as the IR transition dipole. For
SHG the molecular hyperpolarizibility is as follows,

R
lmn
=
1
4~
X

hgj
l
jvihj
mn
jgi
!

2!i

(1.7)
withhgj
l
jvi as the one photon absorption cross section andhj
mn
jgi as the two photon
cross section.
To go from the individual chromophore to the ensemble we sum over all molecules,
index i, and all components of , lmn. In this case we use Euler rotation matrices, S,
represent the projection of each individual chromophore onto the lab frame.
(2)
IJK
is still
a 27 element tensor like
lmn
but now represented in lab coordinates, I, J and K.

(2)
IJK
=
X
lmn;i
S(IJK;lmn)
lmn;i
(1.8)
Taking orentational averaging over the ensemble we get a
(2)
IJK
in terms of ensemble
number of chromophores, N, spectroscopic parameters, and an average orientational
factor.

(2)
IJK
=NhS(IJK;lmn)
lmn;i
i
orientation
(1.9)
9
At this point we can also see that from this molecular construction we have also
reproven what was shown in equation 1.5. If we posit the system where
lmn;i
is com-
pletely centrosymmetric (8abc;xyz;
abc
=
xyz
) then through the orientational average

(2)
IJK
= 0.
In the frame of reference of the plane with z as the orthogonal axis and x and y
in the plane of the interface, we'll start by imposing circular symmetric conditions on
our molecule in the x and y planes. This is a reasonable approximation for the systems
we are looking at, none of which should exhibit any strong long range orientational
preference. Out of the 27 elements in the
(2)
tensor we can reduce to 4 independent
non-zero elements,
xxz
=
yyz
,
xzx
=
yzy
,
zxx
=
zyy
, and
zzz
.
For SHG we can consider two dierent individually selected axes of polarization for
the incoming (!) and outgoing (2!) beams. These can be paramertized as P polarization
with the electric eld component of the light parallel to the surface and S polarization
with the component perpendicular. Only PSS, PPP, (2!,!,!) and S45 (corresponding
to a S polarized 2!, and 45

polarized !) contribute non-zero signal.
I
S45
=jA
xxz
j
2
(I
!
)
2
(1.10)
I
PSS
=jB
zxx
j
2
(I
!
)
2
(1.11)
I
PPP
=jC
zxx
+D
xzx
+E
xxz
+F
zzz
j
2
(I
!
)
2
(1.12)
A {F are trigonometric factors determined by the angles of incidence and the angles
of incidence and re
ection of the beams. Each of these polarization combinations is
sensitive to dierent components of
(2)
IJK
. From taking polarization sensitive measure-
ments we can then obtain information about these components of
(2)
IJK
. We can then
coordinate this information to the chemical factors which build up our
(2)
IJK
(eqs. 1.6
- 1.9), such as the chromophore density, the orientational parameters, and the elements
of
lmn
.
10
1.3.2 Sum Frequency Generation
In sum frequency generation spectroscopy we have additional dimensions of control rela-
tive to SHG. The two pulses used are of dierent photon energies, which can be adjusted
separately, and have dierent resonance conditions with the material. The polarization
of the beams can be selected separately, giving 3 unique axes of polarizations to select.
The alignment of the two beams is dierent and must be accounted for. As the two beams
are not inherently temporarily concurrent, the temporal delay between the visible and
IR pulses must be considered.
When thinking about SFG generation from a material, we need to create a slightly
dierent visual model than we were using for SHG. We can consider SFG as two separate
interactions on the material: rst a resonant IR pulse vibrationally excites the molecules
on the surface, then at a time later, the visible pulse up converts the vibrational
coherence with a stimulated anti-Stokes Raman process. From this we can immediately
see why SFG requires fullling both the Raman and IR selection rules.
To express this formally, we represent the rst interaction of the IR pulse with the
sample as creating a rst order polarization P
(1)
.
P
(1)
(t) =
Z
1
1
E
IR
(tt
0
)S(t
0
)dt
0
(1.13)
We have two components here: the IR electric eld, E
IR
, and the surface response
functionS(t
0
), which contains all the chemical information. The response function S(t
0
)
contains both non-resonant electronic interaction terms, and resonant vibrational terms
which have a dephaseing decay lifetime. We need to take an integral overE
IR
to account
for the fact that dierent excitations of the surface can occur during dierent times within
this interaction, and a system excited on-resonance has memory described by S(t
0
).
Next, the upconversion with the visible pulse projects a second-order polarization
onto the sample
11
P
(2)
(t;) =E
vis
(t)P
(1)
(t) (1.14)
This is the interaction between the visible electric eld, E
vis
and induced rst order
polarization P
(1)
(t). We now have an interaction that depends on both time and the
delay between pulses. However, SFG is a frequency domain technique. The signal we
detect is the intensity of Fourier transform with respect to time t of the second order
polarization.
I
SFG
(!
SFG
;)/

Z
1
1
P
(2)
(t;)e
i!
SFG
t
dt

2
(1.15)
Although this is a dierent method of deriving the spectroscopic signal from the one
discussed of SHG, it is still equivalent. Fundamentally, we are interested in the chemical
information in of the SFG signal we measure and we need to extract this information
from the data as in SHG. However the formalism derived here conveys the additional
temporal components of SFG.
1.4 Outline of the Thesis
This thesis contains three dierent stories of surface specic non-linear spectroscopies.
The work shown here both explores the limit of these methods and use them to under-
stand molecular surfaces. The rst part, Chapter 3, is an implementation of the cross-
correlated frequency-resolved optical grating (XFROG) algorithm. This method was
developed to deconvolute the explicit time dependent eld of an IR laser pulse from a
time delayed SFG mapping. Unlike traditional XFROG algorithms [98], our approach
relies on a non-iterative deconstruction procedure. We present a proof of concept set of
experiments that conrm the validity of both the theory and the implemented XFROG
algorithm.
12
The second part of this thesis in Chapter 4 is an SFG study of the isomerization of
azobenzene terminated SAMs. A novel synthesis for an azobenzene terminated alkane
thiol is presented as well as chemical characterization of the product. A number of binary
SAM devices were built of dierent dilutions of the azobenzene thiol mixed with an alkane
thiol to examine at the role packing and intermolecular sterics have on the isomerization.
VSFG was used to quantitatively study the vibrational modes, surface concentration of
molecules, and molecular orientation in these samples. In situ SFG measurements were
taken during isomerization of these lms. It was observed that isomerization proceeded
only for dilute lms implying a steric inhibition of isomerization in neat lms. Modeling
was performed on the change of SFG signal observable upon isomerization, which implied
that for the dilute SAMs there was a strong conformational bias in isomerization away
from the surface.
The last part of this thesis in Chapter 5 describes the construction of a new picosec-
ond scanning SHG spectroscopy setup and it's application of this setup to study the
electronic excitations at the buried lamellar donor/acceptor interface in solar cells. A
new experimental setup was devised using a computer-controlled tuned of optical para-
metric amplier to allow for automated, rapidly scanning SHG spectra to be recorded.
This apparatus was used to look at the dierences between the P3HT/air and P3HT/C
60
interfaces. Thickness dependent modeling was used to extrapolate the surface specic
signal from the data taken as there is a bulk SHG contribution along with the surface
specic signal. This technique was also used to look at some new red absorbing polymers,
P3HTT-10%DPP and P3HTT-10%TPD.
13
Chapter 2
Experimental Details of
Nonlinear Surface Spectroscopy
In this thesis there are three general classes of experiments that were performed. Vibra-
tional sum frequency generation (SFG) experiments require the generation and summing
of two distinct laser pulses, one femtosecond compressed and vibrationally resonant with
the molecule and the other non-resonant stretched pulse in the visible portion of the spec-
trum. The second harmonic generation (SHG) studies require generation of a tunable
pulse narrowband pulse in the near infrared. Each of these spectrscopic setups requires
light conversion both in the temporal and spectral sense. Also, although all these exper-
iments are time agnostic in terms of data analysis, they require careful alignment of the
temporal relation between laser pulses.
Like many ultrafast nonlinear spectroscopic setups the basis of our laser system is a
femtosecond 800 nm Ti:sapphire laser. This laser is the workhorse of our setup, as the
system relies upon a steady stream of 800 nm pulses coming out of the machine day in
and day out. The huge pulse intensity of this laser allows us to bend and stretch the
light to multiple wavelengths tailored to the chemical structures of the systems we are
interested in probing.
2.1 Sum Frequency Generation
Broadband vibrational sum frequency generation spectroscopy was used for the exper-
iments in chapter 4. This surface specic spectroscopic technique involves the upcon-
version of a broadband femtosecond pulse in the IR range with a spectrally narrow
14
picosecond visible pulse. This spectroscopic experimental setup was developed prior to
the azobenzene SAM experiment discussed in chapter 4.
2.1.1 Setup Layout
Our broadband vibrational SFG (BB-SFG) spectroscopy setup is based on a femtosecond
Ti:Sapphire laser system (Spectra Physics Spitre) retrotted with a Coherent Legend
regen cavity, which is pumped with a Nd:YLF laser (Evolution-30, Spectra Physics)
and seeded with a Ti:Sapphire oscillator (Kapteyn-Murnane Labs) centered at800 nm
(full width at half maximum, FWHM 50 nm). Sixty percent of the uncompressed
fundamental output of the amplier (4 mJ per pulse at 1 kHz repetition rate) is sent
through a compressor producing60 fs pulses (1.8 mJ,796 nm, FWHM27 nm)
and used to pump an optical parametric amplier (TOPAS-C, Light Conversion). The
signal and idler pulses ( = 1.1 - 2.6 m) produced from the TOPAS are mixed in a
dierence frequency generator (NDFG, Light Conversion) to yield tunable infrared (IR)
pulses (500 - 4000 cm
1
). We are able to obtain 10 mW of IR centered at 2900 cm
1
(C-H stretch region) with FWHM350 cm
1
. The remaining 40% of the uncompressed
fundamental pulse was directed into a second compressor to produce 60 fs visible pulses,
which were then sent into a high-power air-spaced etalon (TecOptics; FWHM = 17
cm
1
, free spectral range480 cm
1
, and nesse65) at 11

incidence angle from the
surface normal to produce a picosecond narrow-bandwidth pulse. The IR and visible
pulses were focused onto the sample surface by a 25 cm focal length CaF
2
lens and
45 cm BK7 lens, respectively. The incidence angles of the IR and visible beams are
66

and 63

from surface normal, respectively. The laser power at the sample was
typically 8 - 9 J per pulse for IR and up to 20 - 22 J per pulse for the visible at 1
kHz repetition rate. The time delay between the two pulses was varied by a joystick-
controlled translation stage (Newport VX-25, 0.1m (0.67 fs) accuracy). The SFG signal
was recollimated, spatially and frequency ltered, focused onto the entrance slit of a 300
15
mm monochromator (Acton Spectra-Pro 300i), and detected using a liquid nitrogen-
cooled CCD (Princeton Instruments, Spec-10:100B, 1340X100 pixels). We used PPP
(SFG-visible-IR), SSP, and SPS polarizations for the experiments. Polarization of the
visible beam is controlled by using a zero-order quartz half-wave plate (800 nm, CVI
Melles Griot) while the IR beam polarization was controlled by using a zero-order MgF
2
half-wave plate (150 - 6500 nm, 5 mm thick, Alphalas), and the SFG polarization was
controlled by using a zero-order quartz half-wave plate (670 nm, CVI Melles Griot). To
eliminate polarization contamination, we used a wire-grid polarizer (Specac) for the IR
beam and a polarizing beamsplitter cube (Newport) for the SFG beam.
The spectra of the narrow-band visible and broad band IR pulses were recorded
using the same signal collection optics and the same monochromator by replacing the
sample surface with a gold substrate (BioGold Microarray Slides, Thermo Scientic).
The spectra of the narrow-bandwidth visible pulse were recorded using the same grating
and CCD as for SFG detection. The spectra of the IR pulses were measured using an IR
grating blazed at 5m and a liquid nitrogen-cooled mercury cadmium telluride detector
(IR Associates). Temporal proles of the compressed fundamental 800 nm pulses were
measured using a homebuilt single-shot autocorrelator to have FWHM60 fs. The
compressed fundamental pulses were then used to characterize the time width and chirp
of the BB IR pulses using a SFG cross-correlation on a nonresonant substrate (gold).
Temporal proles of the narrow-band picosecond visible pulses produced by the etalon
were measured by scanning the femtosecond IR pulse (characterized by the spectrum
and cross-correlation as described above) across the visible pulse and recording the SFG
cross-correlation signal from a nonresonant gold substrate. For all experiments presented
here, the time delay between the visible and IR pulses was set to maximize non-resonant
SFG signal.
16
comp
IR OPA
DFG
800 nm,
1.8W compressed
1.6W uncompressed
0
50cm
Vis Delay Stage
Evolution
KML Oscillator
Monochromator
CCD
Stage
λ/2 Plate
Polarization Selector
λ/2 Plate
800nm Filter
Ge
λ/2
Etalon
Figure 2.1. Schematic of the vibrational SFG setup. Generated 800 nm
light split into two parts. 40% compressed, stretched with etalon. 60%
split using OPA and dierence taken in DFG to generate ngerprint
IR. Both beams are focused on sample stage (Detail of stage shown
in Figure 2.2 and SFG signal is polarization selected and collected on
CCD.
2.1.2 SFG with in situ illumination
The primary modication of the optical table for the azobenzene isomerization experi-
ments (chapter 4) was the introduction of a Xenon lamp to allow for in situ exposure
of samples to visible and UV light while collecting SFG response, as shown in Fig. 2.3.
The samples could be illuminated by Xenon lamp concurrently with the SFG data col-
lection. The light was collimated, and spectrally selected using a 340 nm band pass lter
17
50cm BK7
10cm CaF2
15cm BK7
Purge Box
Figure 2.2. Detail of SFG stage. Purge box full of dry, CO
2
depleted
air.
(ThorLabs). It was focused onto the sample using a UV coated parabolic mirror and
optimized such that the maximum intensity overlaid the laser spot.
2.1.3 SFG tting procedure
Because of the large amount of data collected in the isomerization experiments (often
>1000 scans) shown in chapter 4 it was necessary to develop an ecient method to collect,
normalize, t and output the SFG spectra taken. Mathematica code developed for this
purpose is provided in appendix A along with details the tting of raw data exported
from the CCD controller program (WinSpec). The rst program, TableReadFiles (D.1)
takes input from WinSpec in the form of a large set of spectra in sequentially numbered
text les and read them into a single Mathematica le.
Collected SFG spectra of surface nitrile contain two types of signal that need to
be accounted for. There is both a broad nonresonant signal and a narrow resonant
signal. These two signals are produced with a relative phase that must be accounted
for. The signal observed should be in the form ofj
(2)
res
+e
i

(2)
nonres
j
2
The program
Fitting2Gaus1Lor (A.2) was written to t the CN spectra. Through some trial and
error it was found that using two dierent Gaussian functions usually does a good job
of modeling the shape of the non-resonant signal determined by the spectral shape of
the IR pulses produced by our laser, and the resonant signal can be t to a single
18
Xe Lamp
340nm BP
Sample
Shutter
Figure 2.3. SFG Setup with UV illumination. Xenon lamp white light
source ltered with a 340 nm band-pass lter and focused onto sample
stage with spherical lens. UV light modulated with shutter.
Lorentzian function. Put together this means the t will now have the form in equation
2.1. This introduces a total of 10 variables that must be t. Three for each Gaussian
function (amplitude (A
1
, A
2
), width (
1
,
2
) and position (x
1
, x
2
)). Three for the
resonant Lorentzian contains the function (amplitude (A
res
), width (
), position(!) and
the relative phase between the resonant and non-resonant parts ()).
I
SFG
(!) =jA
1
1
p
2
2
1
e
(!x
1
)
2
2
2
1
+A
2
1
p
2
2
2
e
(!x
2
)
2
2
2
2
+
A
res
e
i
(!!
0
)i
j
2
(2.1)
19
Because of the large number of variable inputs for the t it's important to assume
a reasonable set of starting conditions. A good strategy is to rst t spectra from a
reference blank substrate to get the 6 non-resonant factors. The amplitude of the non-
resonant signal changes with the addition of the SAM but the relative amplitudes as
well as widths and positions usually remain unchanged and can be corrected by scaling
the gaussian amplitude. From here it is relatively straightforward to t in the resonant
part as an addition onto the non-resonant signal. With these starting values the t will
usually converge, even when tting thousands of (similar) spectra. The program also can
print out the results of the dierent components of the t, which are useful in evaluation
of t quality.
For example, Appendix A.2 shows a script where the program Fitting2Gaus1Lor is
used to t 2000 consecutive points under isomerization and relaxation for a azobenzene
functionalized SAM. This script consists of three parts. First the variables are initialized
and starting t paramaters are chosen. For these isomerization experiments only one set
of initial tting parameters was necessary to achieve good ts on the entire set of spectra.
In the second there is a loop which ts the data and saves the dierent t parameters
into tables sorted by spectral time. Finally these results can be plotted although for the
isomerization experiments this also requires an additional data set of the lamp on/o
illumination times.
A slightly modied sitting scheme is provided in appendix A.4 there is a slightly
dierent tting scheme used for the data. This is used to create response functions like
those seen in gure 4.15. This code bins dierent sets of raw data by time after an
isomerization lamp was turned on and then averages the data before tting.
2.2 Second Harmonic Generation
The second harmonic generation spectroscopy experiments shown in chapter 5 required
a new setup on the optical table to be developed. The experiment was designed to be a
20
simple but powerful instrument that would allow for the interfaces of organic photovoltaic
materials to be probed precisely, rapidly and in a nondestructive fashion.
2.2.1 SHG System Overview
Our scanning picosecond SHG spectroscopy setup is based on a femtosecond Ti:Sapphire
laser system (Spectra Physics Spitre) retrotted with a Coherent Legend regen cavity,
which is pumped with a Nd:YLF laser (Evolution-30, Spectra Physics) and seeded with
a Ti:Sapphire oscillator (Kapteyn-Murnane Labs) centered at800 nm (full width at
half maximum, FWHM 50 nm).
Forty percent of the fundamental pulse was split through a 50:50 beamsplitter and
directed into a pair of grating compressors (Newport) to stretch 6 ps chirped pulses
(positive and negative, respectively). Each stretched pulse was P polarized with a zero-
order quartz half-wave plate (800 nm, CVI Melles Griot). The two stretched 800 nm
pulses were focused with a 100 cm lens and temporally correlated through a 10 x 10
x 1 mm BBO frequency doubling crystal to generate >340 J pulses of 4 ps, 7 cm
1
resolution centered at 400 nm with conversion eciency of up to 35%. The process of
generating narrowband 400 nm light is described in in detail in section 2.2.2. The two
divergent 800 beams were spatially ltered and the 400 nm beam was re-collimated with
a 100cm BK7 lens. The 400 nm beam was used to pump a picosecond 400 nm OPA
(TOPAS-400, Light Conversion Ltd.) which provided continuous spectral wavelength
output P polarized signal 470 nm - 770 nm, and S polarized idler 830 - 2600 nm. This
design and generation of TOPAS-400 is detailed in section 2.2.3.
The idler beam was ltered from the collinear signal beam using a 750 nm long pass
lter (Newport). The idler beam was focused onto the sample surface by a 50 cm BK7
lens at an incidence angle of 63

from surface normal, shown in Fig. 2.5. Directly before
the sample a second 850 nm long pass lter (Newport) was used to clean up any SHG
generated from previous optics and directly after re
ection a 750 or 950 nm short pass
lter (Newport) was used to block the re
ected idler. For all spectra taken, organic lm
21
comp
comp
BBO
800 nm,
1.6W uncompressed
0
50cm
L1
I1
L2
Evolution
KML Oscillator
TOPAS 3
750nm LP
λ/2 plate
Monochromator
CCD
2ω
Signal + Idler
Idler
Stage
Figure 2.4. Overview of the SHG optical setup. Picosecond 400 nm
generated from 40% of 800 nm selected, split and stretched using a pair
of grating compressors (Detail, gure 2.6) then summed in BBO crys-
tal. 400 nm light used to pump OPA generating narrowband tunable
light 470 nm - 2000 nm. Light ltered and focused on sample (Detail,
Fig. 2.5. Signal collected, polarization selected and collected on CCD.
samples were kept under a purged N
2
environment. The SHG signal was recollimated
and focused onto the entrance slit of a 300 mm monochromator (Acton Spectra-Pro
300i), and detected using a liquid nitrogen-cooled CCD (Princeton Instruments, Spec-10
2KB/LN, 2048 x 512 pixels).
The SHG spectra were taken at all 3 active polarization combinations PPP,
PSS(convention 2!, !, ! polarization, respectively), and S45 (2!, !) . Polarization
of the idler was rotated to either S or P with a zero-order MgF
2
half-wave plate (150 -
22
50cm BK7
15cm BK7
Purge Box
750 or 850nm SP
850nm LP
ω 2ω
Figure 2.5. Detail of the sample stage for SHG setup.
6500 nm, 5 mm thick, Alphalas) and selected using a polarizing cube. For the frequency
doubled signal polarization was selected with a polarizing beamsplitter cube (Newport)
and rotated to P polarization to go into the monochromator with a zero-order quartz
half-wave plate (800 nm, CVI Melles Griot).
2.2.2 SHG design considerations
In designing this setup there were a few generalized requirements that needed to be
fullled. We needed a setup that would create a surface specic signal that had: high
spectral resolution, detectable signal in a reasonable time period, automated wavelength
tunability, and compatibility with the existing SFG spectroscopic setup on the optical
table.
To solve this problem there are a few options that were deemed possible to fulll
these requirements. The rst possibility was to reroute the output from the 800 nm
femtosecond OPA ( Light Conversion, TOPAS-C) which would provide tunable 900
to 1500 nm signal output and 1700 to 2000 nm idler beams. The main advantage of this
option is that it would have very high power with a maximum of500J pulses coming
from the OPA. The compressed nature of these pulses would have a very high maximum
intensity. The main disadvantage with this setup is that the compressed nature of the
pulses would limit the minimum spectral resolution to>300 cm
1
and probably wouldn't
23
be acceptable for SHG spectroscopy of OPV materials. It also is unclear if tuning the
femtosecond OPA is easily automatable through computer control. This solution would
have been simple to implement on the existing laser table as a simple reroute of the OPA
beam using a kinetic mount mirror before the DFG box would have been trivial. It is not
clear if the rapid tuning of the OPA would have aected the relatively precise settings
necessary for steady output from the DFG box. Although this solution for a tunable
light source for SHG was rejected for the class of experiments shown in chapter 5 it
would be useful if adapted for an EFISH[33, 34] or other femtosecond resolved SHG[99]
type setup where the temporal resolution of the SHG pulse is more important than the
spectral resolution.
The second proposed, and eventually adopted, solution was to use a 400 nm picosec-
ond OPA (TOPAS 3) as the tunable light source. This provides tunable signal from 470
- 770 nm, up to 60 mw power, and idler from 850 - 2000 nm, up to 10mw average power
(at 1kHz repetition rate). These pulses should be close to Fourier transform limited
and for 1-4ps duration, giving a spectral resolution of10 cm
1
. This is more than
sucient spectral resolution for the types of samples investigated, but the experiment
demonstrated broad capability here. The TOPAS 3 OPA is also equipped with a suite
of LabVIEW VIs that were accessible for easily scripting the device to allow for rapid
and repeatable wavelength tuning.
To implement this specialized setup a few supercial changes are necessary to switch
the to SHG setup from the established SFG setup. The most signicant change is in the
rst compressor (Fig. 2.6) where the 800 nm visible SFG line needed to be recongured
to switch between the two spectroscopic modes. The crossover requires splitting the
800 nm beam with a kinetically mounted 50/50 beam splitter and creating two equally
powered oppositely chirped stretched pulses. As the visible line of the SFG setup only
passes through one of the two compressors, retuning is only required for one line and the
second line can remain in the appropriate setting for SHG. Stretching the 800 nm beam
also changes the delay of the pulse, and with the second compressor left in position the
24
appropriate setting for pumping the OPA can be determined by simply looking for the
already calibrated delay between the two split beams. Retuning to a compressed 800 nm
line for SFG can easily be done by optimizing plasma generation by focusing the beam
into air with a 5 or 10 cm lens. The switchover may also change beam steering which
can be checked by redirecting the beam through pre-aligned irises on the beam path.
On the sample side there are a few adjustments that needed to be made to switch
between SFG and SHG. Most of the optics on the down stream side of the OPA are set
up such that there is minimal perturbation of the SHG visible line although SFG requires
adding some lters to the beam path (Fig. 2.5) which can be trivially added as well as
removing the 800 nm lters used to block the visible beam from the CCD. The SHG
setup also uses a dierent monochromator and CCD than the SFG setup (Princeton
Instruments, Acton SP2500 for SHG vs. Action SpectraPro 300i for SFG). The beam
routing between the two detection systems is controlled by a kinetic
ip mirror. The
switchover between the two setups can be performed in well under an hour.
A third method for this experiment would be to set up a white light continuum plus
narrow band visible electronic sum frequency generation (ESFG) setup as implemented
by the Tahara Group[100, 101]. On our optical table this could be implemented by using
the OPA to produce a tunable narrow band visible or near IR source and summing that
with a temporally compressed white light continuum from 800 nm focused in a Al
2
O
3
or CaF
2
crystal[102, 103]. This method allows for broadband collection of SFG signal,
and with careful use of reference substrates like GaAs[101] allows for non-scanning,
single collection of surface specic electronic spectra dispersed on a wide grating CCD.
This method is more complicated to implement compared to narrow band SHG as it
requires not only setting up the OPA but also creating a temporally correlated, stable,
overlapped and well focused white light continuum. Creating the white light continuum
would require routing a compressed 800 nm line. This could be taken either from before
the femtosecond OPA or some of the stretched 800 nm light could be taken before the
TOPAS 3 and recompressed. The advantage is that this does not disturb the relatively
25
fragile BB-IR generation line and this second method allows for that to be applied later
in a possible pump-VSFG-probe and pump-ESFG-probe congurations. This would also
allow for 8 spectral SFG polarization combinations to be probed. Although ESFG is a
promising method for future experiments, we chose to adopt the simpler scanning-SHG
setup described earlier. This decision was due to both the complexities of building the
ESFG experiment and the additional complications in analyzing and decomposing ESFG
spectra. Overall, although an extremely promising direction for future experiments, for
the scope of the experiments shown in chapter 5, both the time complication from setting
up the aparatus as well as some of the additional barriers to decomposing VSFG spectra
versus scanning-SHG makes the SHG a more prudent stepping stone.
Cross chirp 400nm generation
One of the technical barriers to building this apparatus is in the stretching of our60
femtosecond broadband laser into a non-chirped narrow-band 400 nm pulse. To generate
the narrowband picosecond 400 nm pulses we used the method developed by Raoult and
coworkers[104] by summing oppositely chirped stretched femtosecond 800 nm pulses in
a nonlinear BBO crystal. To achieve a narrow, nearly transform limited pulse, the two
pulsed were tuned to give equal and opposite chirp with regard to time through the
frequency summing crystal such that the sum frequency is narrowband.
We implemented this by splitting uncompressed fundamental output at 800 nm of
our Ti-Sapphire laser and compressing the two beams with a matched pair of grating
compressors(Newport), Fig. 2.6. The exact chirp of each the beams is not characterized
but optimization for the conguration where the two beam have equal and opposite chirp
can be done by monitoring the 400 nm power output after doubling and minimizing the
spectral width of the doubled beam. This procedure does not yield a global optimum,
but for each setting of compressor 2 (top) there is a single optimum for compressor 1 and
delay setting for a transform limited picosecond pulse. The compressor 2 setting can be
changed to adjust the pulse duration and bandwidth of the resultant optimized 400 nm
26
λ/2 plate
λ/2 plate
Delay
50:50 BS
100cm Lens
1
2
Figure 2.6. Detail of the compressors used for 400 nm generation.
Compressor 2 generally left in place between SFG and SHG setups.
Compressor 1 used for SFG and SHG and requires tuning. Tuning
each compressor both changes chirp and relative delay of the 800 nm
laser pulse. After the 100 cm lens both pulses are summed though a
BBO crustal to produce transform limited picosecond 400nm pulse.
output. In the experiments in chapter 5, compressor 2 was optimized initially and not
adjusted further.
2.2.3 TOPAS 3
The TOPAS 3 is a 400nm pumped computer controlled picosecond optical parametric
amplier. The device converts picosecond pulses of 400 nm light into signal and idler
beams tunable from 470 nm - 800 nm and 800 nm - 2000 nm, respectively. The OPA
27
provides tunable narrow band light with easy interface with computer control for fast
and repeatable wavelength tuning.
TOPAS 3 internal operation
The TOPAS employs ve laser passes through a single BBO crystal for light conversion
and amplication. The incoming laser is split three ways with beam splitters, with 10%
of the power going to seed the rst three passes, 10% amplifying the fourth pass and
80% of the incoming power used for the fth pass. The rst pass through the crystal
generates a broad-band super
uorescent seed which is amplied by the following four
passes. After the rst pass the broad band light is re
ected o a diraction grating and
passed twice more through the BBO crystal amplifying the seed. The beam is still very
weak at this point but wavelength selected and it is amplied by the fourth pass. With
the fourth pass it is important that the power of the pump is reduced to avoid generating
additional super
uorescence which will damage the BBO crystal as well as reduce the
eventual output quality. Finally the fth pass, which overlaps specially and temporarily
with the eective seed generated by the rst four passes, amplies the pulse to the nal
power.
The internal wavelength tuning is performed by rotating the nonlinear crystal chang-
ing the phase matching condition. This displaces the beam position and is compensated
by a quartz plate controlled by a stepper motor. The spectrum is also cleaned up by
using a motor controlled grating to select a single wavelength in the pre-amplication
stage. These provide three independently controlled axes by which to optimize the laser
although optimized values can be pre-tabulated by wavelength for regular use. While it
may seem like alignment with the ve passed through the BBO will be complicated, in
normal operation it is only necessary to align the rst three.
28
Characterization of TOPAS 3 output
To evaluate the wavelength dependent power output from the OPA a LabVIEW script
was written to automatically cycle through wavelengths and read o the power from
a high sensitivity thermal power meter (Ophir 3A-SH) connected to an energy meter
(Ophir, Nova II) and read into the computer. The power spectra of the OPA idler
output were normally taken at the sample stage after the beam had been ltered and
rotated with a half wave plate, polarization selected with a cube, and ltered with both a
750 nm and an 850 nm long-pass lter. The power dierences are small but still necessary
to be accounted for when normalizing spectra in dierent polarization selections.
Figure 2.7 shows the idler output for the rotated P polarized OPA idler output. The
center value represented is the average of ve power curves taken with the error bar listed
as the standard deviation between the power values of these measurements. These ve
scans were taken over a period of approximately 4 hours. The sharp discontinuities that
occur in the spectra at1075 nm and 1275 nm are from the change in OPA eciency
as it switches over grating orders in its wavelength selection. The repeatability of the
output from the OPA is extremely important when taking SHG spectra because the
normalization is not performed in situ and thus any comparison between a sample and
reference requires a independent tuning of the laser. We found that the repeatability of
the laser power upon subsequent tunings and over time is excellent for the applications
desired. Figure 2.8 shows is the relative error for the variation of laser power between
measurements. For the primary range in which SHG data was taken, 950-1600 nm
the relative error was under 4%. To translate this into the error expected between two
individual wavelength SHG readings this represents less than 6% relative error introduced
by OPA tuning
uctuations.
Characterizing the spectral bandwidth of the of the idler OPA output is more dicult
than the power. Silicon CCD detectors lose signal for wavelengths longer than900 nm
so the frequency dependent spectral width of the idler was observed by looking of the
width of the doubled signal. Figure 2.9 shows the t spectral width of the idler generated
29
800 1000 1200 1400 1600 1800 2000
0.000
0.002
0.004
0.006
0.008
0.010
0.012
Idler Wavelength HnmL
Power HWL
Figure 2.7. TOPAS idler output. Power curve details mean and stan-
dard deviation from 5 dierent readings.
800 1000 1200 1400 1600 1800 2000
0.00
0.05
0.10
0.15
Wavelength HnmL
Reletive Error
Figure 2.8. Relative error from gure 2.7. Normal spectra range for
SHG is 950 - 1600 nm
SHG. The spectral width of the SHG goes from about 7 cm
1
at 975 nm and broadens
smoothly to about 12 cm
1
at 1800 nm.
30
æ æ
æ
æ æ
æ
ææ
æ æ
æ
æ
æ
æ
æ
æ
æ
ææ
æ
æ
æ
æ
æ
ææ
æ
æ æ
ææ
æ
æ
æ
æ
æ
æ
æ
æ
ææ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ æ ææ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ æ
æ
æ
æ
æ
æ
æ
æ
æ æ æ æ
æ
æ
æ
æ
æ
æ
æ
æ æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
ææ
æ
æ
æ æ
æ
æ æ
æ
æ
æ
æ
æ
æ
æ
æ æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
ææ
æ
æ
æ
ææ
æ
æ
æ
æ æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
æ
1000 1200 1400 1600 1800
0
2
4
6
8
10
12
14
Idler Wavelength HnmL
SHG Width Hcm
-1
L
Figure 2.9. Spectra width from SHG o of Y cut quartz with idler
tuning.
2.2.4 Spectral Filtering
Spectral ltering is a critical part of the scanning SHG setup as it is required for both the
signal and idler from the OPA as well as the re
ected idler and SHG after the sample.
Spectral ltering serves three important functions. Filtering blocks the visible output of
the OPA to keep from damaging the sample and CCD, it blocks the idler from hitting
the CCD, and it isolates the SHG generated from the sample from the spurious SHG
signal generated from other optics. Filtering before the sample is performed with a 750
nm long-pass lter directly after the OPA and a 850 nm long-pass lter placed directly
before the sample. Both of these lters block the vast majority of the signal output
while allowing the majority (>90%) of the idler to transmit for all OPA wavelengths. It
is necessary to use two lters to more eciently eliminate the transmitted signal OPA
output. The lter immediately after the OPA also serves to protect the half-wave plate
from damage from the signal as well as reduce light scattering in the room which may
potentially reach the detector and contaminate the SHG signal. The lter immediately
31
before the sample removes the SHG generated by the idler from any of the surfaces before
the sample. With the 850 nm long pass lter this will lter out all signal wavelengths and
the frequency doubled signal out to idler wavelengths of 1700 nm. It is put immediately
before the sample to insure there is only clean idler at the sample, Fig. ??.
After the sample there is a either a 750 nm or 950 nm colored glass short-pass lter
used to block the re
ected idler from the sample. If the idler is transmitted, the second
order grating re
ection in the monochromator will fall on top of the SHG signal on the
CCD. Fortunately, for idler wavelengths longer than 1000 nm the quantum eciency of
the CCD is very low and small leakage does not impact the experiment. The choice of
short pass lters depends on the scanning range as each blocks some of the generated
SHG. The 750 nm lter is applicable for scanning 475 - 725 nm (2!). The 950 nm lter
is useable for scanning 500 - 900 nm, although the blue edge is partially ltered by the
short pass. In general these short pass lters have a much more gradual cuto than the
long pass lters. The short pass lter also blocks the idler so that extra SHG is not
generated by the optics after the sample.
2.2.5 Alignment procedure and tricks
There are a few important details to be aware of during alignment. A useful trick in
alignment of the SHG is to remove a one of the two long pass lters. This leaks a
very small amount of signal which is collinear with the idler and generated SHG signal.
Although this leaked amount is still too intense for the CCD, a variable neutral density
lter wheel can be included to align the laser into the monochromator. Once aligned by
eye, (with both LP lters in place) the signal OPA output is barely detectable on the
CCD. Setting the OPA idler output to 1200 nm takes advantage of a degeneracy where
the signal output of 600 nm is equal to the doubled signal from 1200 nm . This is usually
the best place to begin searching for SHG signal as the aligned signal and SHG signal
can be caught on the CCD at the same time. The position of the two signals are close
32
OPA Mono CCD
LabVIEW
Computer Control
ω 2ω
Figure 2.10. Control and measurement dependancies of LabVIEW pro-
gram.
but not perfectly collinear and with low signal samples nding the SHG signal without
a good reference can be extremely time consuming and/or impossible.
2.2.6 SHG data collection
To create an automated interface to rapidly and repeatably take SHG scans a LabVIEW
program was written to control the various pieces. To take automated SHG spectra the
computer needs to interface with three components on the optics table (Figure 2.10). The
computer needs to tell the OPA to tune the proper idler wavelength (!), the monochro-
mator needs to look for the SHG wavelength (2 !) and the CCD needs to capture the
signal. This program was adapted from an older LabVIEW script written for taking the
XFROG data, described in Chapter 3.
The rst step in the software is that the OPA must be set to the correct idler wave-
length. The sub-VIs to control the OPA were already provided by Light Conversion
although they needed some editing to put in a wrapper with a simple wavelength input.
When given an input wavelength the OPA can acquire a wavelength and stabilize usually
within a few seconds and signicantly faster for a small change. Nevertheless a small
time delay of 2 seconds was programmed in after the wavelength tuning step to ensure
stability. The monochromator was adjusted concurrently with the OPA to center it's
33
spectral range on the expected SHG signal wavelength. Although we might expect the
SHG wavelength to be twice the wavelength input to the idler, the OPA, while precise, is
slightly inaccurate with regard to the wavelength produced. The actual SHG wavelength
overshot the expected value by a small amount. To account for this, an empirically deter-
mined quadratic t was added to estimate actual SHG wavelength from the input idler
wavelength.
Once the laser and monochromator are set up, data may be collected. The SHG signal
as well as background spectra are collected onto the CCD for a preset accumulation time.
The signal on the CCD is binned into 32 vertical pixels and 1 pixel in the wavelength
direction, which allows for the total SHG signal to fall within one collected strip. The full
captured CCD image is displayed, but the saved data are only the strip with the signal
on it minus a selected blank strip. Subtracting a blank strip both sets the baseline to zero
and mitigates some non-strip dependent signals such as leaky room lights. Regardless,
every eort was taken to minimize stray environmental sources of light. After each scan
the full raw data of the strip are saved as well as the total integrated intensity of the
strip. Each thus acquired spectrum is tted to a Gaussian, and the wavelength and area
from a primitive gaussian t are also saved into a table. This included table is useful for
quick inspection of the spectra taken without the extensive peak tting routine described
below.
There are two dierent settings used for measuring the data scans. The rst option
takes a set number of wavelength points all at set intervals for a specied accumula-
tion time. This is useful for taking quick calibration scans on high signal substrates like
quartz. The program also supports input of a table of arbitrary wavelength and accumu-
lation time combinations. For taking low signal data it is useful to tune the accumulation
time and gridding to accumulate the high signal spectral regions faster and the low signal
areas may be measured with longer accumulation times with a more sparse grid. For
example, SHG spectral scans of neat P3HT require typically 50 spectral points at more
than 60 seconds per point 5. With very low signal samples single wavelength points
34
Figure 2.11. Left: A set of SHG spectra taken at a 1 nm spacing.
Right: Normalized SHG spectra t to a single Gaussian function.
can take over 5 minutes in bad spots and taking an evenly time spaced and wavelength
spaced grid calibrated to the lowest signal wavelength will make a sample take much
longer with minimal signal to noise improvement.
2.2.7 Fitting Algorithm
As in the SFG case described earlier, code must be used to t the raw SHG spectra.
There area a few requirements on the tting code for SHG. First the algorithm must be
able to accurately t a weak signal. Second the code must be able to nd the correct
peak position very reliably to minimize hand tting of errant spectra. Figure 2.11 shows
an example of a set of raw spectra and a t on the peak. The tting of a raw spectrum
requires two steps. The rst step is a spike removal algorithm (Appendix B.1) which
removes the large spectral spikes due to cosmic x-rays that often are present in long
acquisition CCD spectra. The second part (Appendix B.2) ts the SHG peak to a
gaussian line shape. There is also a Mathematica script included (Appendix B.3) for
running these two algorithms on a large number of SHG spectra.
35
The spike removal program (SpikeKill) looks for outlier pixels in the spectra and
removes them. The program is given an envelope by which a specied change in recorded
intensity from the adjacent pixel is considered valid. Points where the absolute value of
the derivative is under the envelope are considered valid and a new spectra is created
only from the non-spike points. This makes the spike-removed spectrum smaller in array
size than the input spectrum but because the data contains a wavelength point for every
intensity it is not necessary to have regular spaced data points. If the envelope is set too
small then this procedure will preferentially remove the SHG peak from the spectra. This
program is equipped to handle both a positive and negative spikes that are produced
from either the signal strip of the CCD or from the reference CCD strip.
The spectrum with the spikes removed is then t with a gaussian, which occurs in
a few steps. First an initial guess of the starting parameters for a gaussian of form
amp
1
1
p
2
2
1
e
(xx
1
)
2
2
2
1
needs to be made. The width is t to a constant estimate of

1
=0.13. The amplitude, amp
1
and peak center, x
1
are estimated by sorting the points
in order of intensity and guessing the peak center as the average of the top 5 most intense
points. The estimate for amp
1
, as it represents the area under the curve, can be set to
the mean height of the points selected times the width (sigma
1
) times 2. The center
value, x
1
, estimate is the mean of the wavelength values for the selected points. From
here, Mathematica uses it's NonlinearModelFit routine to t our estimated gaussian to
the data. There is an included option to graph every raw spectra measured with the
programs, shown in Figure B.1, by setting \Printqm" to \True". This will graph the
raw data overlaid with the estimated gaussian in blue and the t gaussian in red. For
normal tting this should be set to false as this option signicantly slows down spectral
tting. Lastly there is a script that will run these two algorithms on a full number of
recorded SHG spectra. It is set up to produce a table of the t gaussian areas versus t
center gaussian wavelengths.
36
2.2.8 Normalization
The normalization of the SHG signal between measured wavelengths is very important
to implement in a robust and repeatable fashion to get accurate wavelength dependent
SHG spectra. Clearly using a substrate that gave a constant wavelength independent
SHG signal as a reference signal and normalizing to that would be the most ideal solu-
tion. This would account for normalization in wavelength
uctuation of the idler power,
! dependent pulse duration, 2! absorption of the SHG signal, relative spectrometer
eciency and chromatic aberration in the idler focusing optics.
The normalization problem is reduced in complexity as the idler power curve is the
only one of the normalization concerns that is responsible for high frequency spectral

uctuation. The CCD/monochromator eciency curve is nearly
at over a limited spec-
tral range. The primary source of wavelength dependent absorption of the generated 2!
signal is from the short-pass lter after the sample to block the ! idler beam. The two
lters used were a 750 nm and 950 nm short-pass lter, and spectra were taken only
when 2! was further than 50 nm from the spectral cuto. The generalized form for SHG
o a surface is that SHG signal scales inverse linearly with laser pulse duration [93].
However our signal is normalized to arbitrary units so all the non-wavelength dependent
parts of the equation can be neglected in our normalization leading to I(2!)/ P (!)
2
.
A power normalization correction can be determined by taking a power curve (Figure
2.7) and squaring it. For the higher energy side of the idler spectral range the power
curve exhibits sharp gures and normalization is necessary to resolve the structure over
even a small spectral range. The total normalization formula is described in Eq. 2.2.
I
SHG;Norm
(2!)/
I
SHG;Raw
(2!)
P
idler
(!)
2
T
SP
(2!)R
CCD
(2!)
(2.2)
37
Where P
idler
(!) is the power of the incumbent ! beam, T
sp
(2!) is the transmission of
the short-pass lter at the doubled frequency, andR
CCD
(2!) is the frequency dependent
response of the CCD.
It is also worth considering the repeatability of this normalization procedure and the
error it propagates onto the measured SHG spectrum. Out of the four normalization
factors listed earlier, the CCD eciency and lter absorption result in negligible error.
The pulse spectral bandwidth is collected simultaneously with the SHG signal and there
is some error associated with the line shape deviation from gaussian line shape and tting
uncertainty of the width. However the power normalization is the primary source of error
in this procedure. As the power curves are not taken in-situ from the SHG spectra, it is
necessary to consider the error introduced both from time
uctuations of the laser power
and the repeatability of the OPA to tune to the same power-wavelength spot. The
laser power is quite reproducible between 475 nm and 800 nm. With the signal squared
this propagates into6% error introduced by the normalization procedure. Thus this
normalization procedure as a whole does not appear to introduce signicant error.
38
Chapter 3
XFROG
Controlling the shape of femtosecond pulses in the mid IR is very challenging. The gener-
ation of pulses involves multiple non-linear amplication processes; specialized materials
are used for IR windows and lenses, many of which are quite dispersive and can introduce
signicant high order chirp to a previously clean pulse. For this reason it is important
to know explicit electric eld of the incident pulse, E
IR
, before performing any optical
spectroscopy with the IR pulses. Many of the standard methods used to characterize
the electric eld of visible pulses such as autocorrelation only characterize the intensity
envelope of the pulse. Often methods such as frequency-resolved optical gating (FROG)
involve time-consuming iterative deconvolution procedure to extract the amplitude and
phase of the E-eld[98, 105, 106, 107].
To address this Vladimir Chernyak and coworkers developed an algorithm that per-
forms non-iterative deconvolution of the amplitude and phase from the XFROG Cross-
Correlated Frequency-Resolved Optical Gating) measurement. Here sum frequency gen-
eration sums an unknown IR pulse with a well-characterized visible pulse, instead of
with itself as in autocorrelation interferometer. We present a noniteratice deconvolution
procedure for extracting the complex eld of E
IR
(t) from XFROG measurement.
3.1 What Information is in SFG Spectra?
To begin, let's look at the reverse process, rst showing what is the information contained
in the SFG signal, and then how we can use this signal to back out information about the
input laser pulses used to generate it. We will start with a simple example of signal from a
39
single SFG spectrum, and show how to interpret the signal in terms of he time-dependent
electric elds of the two pulses, and discuss the limitations of a single measurement.
We can look at the SFG process as a sequence of two events. The rst is an interac-
tion of the IR light with the sample surface, which has the non-resonant electronic and
resonant vibrational response. We represent this in equation 3.1 with the electric eld
of the IR represented by E
IR
(tt
0
) and the response function S(t
0
).
P
(1)
(t) =
Z
1
1
E
IR
(tt
0
)S(t
0
)dt
0
(3.1)
In the case of a purely non-resonant response we can approximate the fast electronic
response as instantaneous and replace S(t
0
) = (t
0
), the integral in equation 3.1 then
falls out leaving P
(1)
(t) =E
IR
(t).
The second interaction creates the second order polarization (equation 3.2), P
(2)
,
from the upconversion of the signal by the visible light. These two interactions are not
simultaneous and represents the delay between the two.
P
(2)
(t;) =E
vis
(t)P
(1)
(t) (3.2)
In the spectrally and time resolved SFG measurement (STiR-SFG ??), the SFG
signal recorded is then the intensity of the radiation emitted by P
(2)
(equation 3.3) as
a function of SFG frequency ask the time delay between the visible and IR pulses. This
requires a Fourier transform of the electric polarization to represent it in the frequency
domain.
I
SFG
(!
SFG
;)/

Z
1
1
P
(2)
(t;)e
i!
SFG
dt

2
(3.3)
40
Expanding P
(2)
we get equation 3.4 and applying the assumption of purely non-
resonant interaction leads to equation 3.5. Equation 3.6 changes the notation by intro-
ducing Fourier transform F(f)
R
1
1
fe
i!t
dt.
=

Z
1
1
Z
1
1
E
IR
(tt
0
)S(t
0
)dt
0

E
vis
(t)e
i!
SFG
dt

2
(3.4)
=

Z
1
1
E
IR
(t)E
vis
(t)e
i!
SFG
dt

2
(3.5)
=jF (E
IR
(t)E
vis
(t))j
2
(3.6)
From here we can recall the convolution theorem shown in equation 3.7 and rewrite
equation 3.6 in as SFG intensity proportional to the intensity of the convolution of the
Fourier transforms of the IR and visible elds (eq. 3.8).
F(FG) =F(F )F(G) (3.7)
I
SFG
(!
SFG
;)/jF (E
IR
(t))F (E
vis
(t))j
2
(3.8)
In this algorithm we start with premise that we know E
vis
satisfactorily (from a
separate measurement) and have recorded a STiR-SFG spectra (I
SFG
(!;)); we wish
to solve for E
IR
. The diculty that we are immediately confronted with is that we
spectroscopically measure the electric eld intensity, not the eld explicitly. This means
that the phase information withinE
vis
andE
IR
are, to an extent lost in our measurement.
With this information we can easily solve for the intensity of the electric eld in the
IR,jE
IR
j
2
but we want the complex eld, E
IR
(t). The phase information, however is
contained in the two dimensional data set I
SFG
(!;), and the full complex eld of E
IR
can be extracted as shown below.
41
3.2 The Algorithm
A sketch of the non-iterative XFROG algorithm is shown in Fig. 3.1. The starting
requirements for the procedure are a XFROG intensity map I
XFROG
(!;)) dened in
frequency! and time, and the complex representation of the visible beam eldE
vis
(t).
Both inputs need to be put into the units of t and
(the Fourier transformed axes
from ! and ). The deconvolution is performed in the Liouville space by dividing the
transformed input by the visible eld. This is then transformed into the density matrix.
From here, diagonalizing the density matrix and nding the eigenvector corresponding
to the largest eigenvalue yields the output complex E
IR
(t).
The rst step in the formal denition of the XFROG algorithm is to revisit the
denition that derived in equation 3.8 to dene the signalI
XFROG
(!
SFG
;). We replace
the time dependent elds we have been using, E
vis
(t) = E
g
(t), and the unknown pulse
we seek to characterize E(t) with density matrices that represent both the real and
imaginary parts of the eld. These are dened using two time axes with the modulated
complex eld dening one dimension, t
0
, and the complex conjugate dening the other,
t
00
.
(t
0
;t
00
)E(t
0
)[E(t
00
)]

g
(t
0
;t
00
)E
g
(t
0
)[E
g
(t
00
)]

(3.9)
Here and below the subscript \g" refers to the gate visible pulse. Now we can redene
the SFG intensity formula in terms of the two density matrices. Note here that instead
of dening I
XFROG
explicitly with a squared function, we are dening it with a complex
conjugate pair.
42
I
XFROG
(τ,ω)
J(τ,t)
ρ
vis
(t’,t”)
g(τ,t) = ρ
vis
(-τ,-τ-t)
g(Ω,t)
~
J(Ω,t)
~
f(Ω,t)=
J(Ω,t)
~
g(Ω,t)
~
~
f(τ,t)
ρ(t’,t”)= f(t’,t’-t”)
E
vis
(t)
F
-1
[ω→t]
F
-1
[Ω→τ]
F[τ→Ω] F[τ→Ω]
E
IR
(t)
Eigenvector
Algorithm Overview
Figure 3.1. A sketch of the XFROG algorithm. The two pieces of input
data are the captured spectra, I
XFROG
(;!), and the electric eld of
the visible laser,E
vis
(t). The deconvolution is represented with
~
f(
;t)
in the center. The output result from this is the explicit electric eld
of the IR pulse, E
IR
(t).
I
XFROG
(;!) =
Z
1
1
Z
1
1
(t
0
;t
00
)
g
(t
0
;t
00
)e
i!(t
0
t
00
)
dt
0
dt
00
(3.10)
Note that this equation is linear in
g
(t
0
;t
00
). At this point we can see that we can
solve for(t
0
;t
00
) if
g
(t
0
;t
00
) andI
XFROG
(;!) are known. We rst want to apply an
inverse Fourier transform I
XFROG
(;!) with regard to ! to change the the dependence
to and t.
43
J(;t) =F
1
!!t
[I
XFROG
(;!)] (3.11)
=
Z
1
1
Z
1
1
(t
0
;t
00
)
1
2
Z
1
1

g
(t
0
;t
00
)e
i!(t
0
t
00
)
dt
0
dt
00
e
i!(t
0
t
00
)
e
i!t
d!
The integral on the right side is equivalent to the delta function (t
0
t
00
t), which
gives us a simplied expression for J(;t).
J(;t) =
Z
1
1
(t
0
;t
0
t)
g
(t
0
;t
0
t)dt (3.12)
This can be redened as a one dimensional convolution with regards to , the inter-
pulse delay.
J(;t) =f(;t)g(;t) (3.13)
where
f(;t) =(t;t) (3.14)
g(;t) =
g
(;t)
This can be converted to a standard deconvolution by using the convolution theorem
and fourier transform both sides with regards to , and deviding to yield the form.
f(;t) =F
1
!!
"
~
J(
;t)
~ g(
;t)
#
(3.15)
with
44
~
J(
;t) =F
!!
[J(;t)] =
~
f(
;t)~ g(
;t) (3.16)
~ g(
;t) =F
!!
[F (;t)] (3.17)
~
f(
;t) =F
!!
[g(;t)] (3.18)
Finally we can transform f(;t) to the density matrix we are interested in, (t
0
;t
00
),
using equation 3.14. From examining our denition of the density matrix in equation 3.9,
we could in principle take any row or column from the matrix and it would be a proper
complex representation of E
IR
with some relative phase oset. However, we have a redun-
dancy of information in this case with a whole dimension of information more than we
need. To account for this we will use a method from the Forward-Error-Correction[108]
(FEC) eld. The theory dictates that by diagonalizing the density matrix, the eigenvec-
tor corresponding to the largest eigenvalue of the system is a representation of E
IR
(t)
that contains the minimum noise.
3.3 XFROG experimental
For the algorithm discussed above we wanted some simple, proof-of-concept spectra to
demonstrate and prove this method's ability to extract the full complex eld of the IR
pulses in our system. To do this we took three sets of 100 dierent time delayed PPP
spectra centered at 2800 cm
1
from a clean piece of gold as a non-resonant substrate.
In the rst spectra the \normal" IR pulse optimally compressed coming from our dif-
ference frequency generation box was only retarded by the waveplate used to select P
polarization. For the second data set we used a CaF
2
plate to introduce negative rst
and third order dispersion to the pulse. Lastly we added a Ge plate which introduced
positive rst and negative third order dispersion to generate XFROG spectra. Figure
3.2 shows these three spectra. In each case two dierent monochromator settings were
45
used to ensure that the full spectral width of the broadband IR pulse is captured. The
chirp introduced by dispersive material is clearly seen as the tilt of the XFROG image
in Figure 3.2.
3.3.1 Experimental Details
The experimental generation of SFG signal for the XFROG measurements were similar to
those described in chapter 2. Instead of a stretched 800 nm laser pulse the XFROG mea-
surements used a compressed visible (gate) pulse. Temporal proles of the compressed
fundamental 800 nm pulses were measured using a homebuilt single-shot autocorrela-
tor to have FWHM 60 fs. The compressed fundamental pulses were then used to
characterize the time width and chirp of the BB IR pulses using SFG cross-correlation
XFROG on a nonresonant substrate (gold). A 1 mm thick Ge window was placed in
the IR beam path to create a positive chirp in the IR pulse, while a 6.3 mm thick CaF
2
window was used to create a negative chirp in the IR pulse. An acquisition time of 0.5
sec was used per delay step (0.5 m or 3.33 fs). CCD images were processed using a
LabView program.
3.4 XFROG, Programmatically
Code was written in Mathematica to run the XFROG algorithm on real sets of measured
spectra. This code is included in Appendix C, and has three parts. In appendix C.1 we
present a list of the programs used for interpolation, aligning correct units within the
spectra, taking forward and inverse Fourier transforms, generating the density matrices
forE
vis
, and preparing data sets for graphing. Appendix C.2 contains the programs that
take the data and initial values as input and runs the XFROG algorithm. "Deconvolu-
tionRun" was setup to take real experimental data input while DeconvolutionRunFake-
Data was created to debug the code by deconvoluting sets of mock data generated from
a known E
IR
and E
vis
.
46
Figure 3.2. Three dierent measured XFROG images. Each image
is two separate time time dependent SFG scans stitched together and
interpolated into a 512x512 grid of interpolated frequency ! versus
delay . For the left spectra the IR beam was stretched with both a
6.3 mm CaF
2
plate and an IR Waveplate. The center image the beam
only passed through the IR Waveplate. In the right image a 1 mm
piece of Ge was used to stretch the IR pulse.
47
Choosing the simulation time and grid width are very important. Appendix C.3
shows an example of some initial values used in these programs. Values such as the
number of points to be run in the interpolation grid, the width of the simulation in fs,
as well as any osets that need to be applied to the experimental data are set here. It
is also important to maintain proper (Fourier-consistent) units in t, (fs), !, and

(cm
1
) throughout the program. As the program necessitates a squarely gridded data
set it is only necessary to specify one of the 4. Likewise as the raw data is measured in!
(interpolated from wavelength) and it's necessary to chose a large enough such that
the simulation includes all the measured experimental data. However if the chosen is
too large it will make the spectral width in !, which is inversely proportional to , too
small. To fulll this requirement, we choose a such that it includes all the signal taken
in both the and! dimensions, and a grid size such that all Nyquist-Shannon theorem
sampling concerns are fullled. Increasing the grid size will also slow the program down
quadratically.
Appendix C.4 shows some graphing programs that visualize the dierent steps of the
XFROG algorithm as well as the output E
IR
. E
IR
is graphed along with the quadratic
and third order chirp t to the data.
The experimental XFROG data sets shown in Figure 3.2 were evaluated using the
programs in Appendix C. Figure 3.3 plots the solved E
IR
power spectrum and phase
(argument) for the IR beam attenuated through Ge, the IR Waveplate alone, and CaF
2
.
The quadratic chirp extracted via this method obeys the expected trend. We obtain the
values of Ge tting to +3:66 10
5rad
fs
2
, the waveplate alone tting to3:12 10
5rad
fs
2
and the CaF
2
tting to7:73 10
5rad
fs
2
.
3.4.1 Why is Deconvolution Dicult?
The mathematical operation of convolution reduces the noise in a signal of nite Fourier
bandwidth by cutting the high and low-frequency noise. In real space, the operation
of convolution (FG) takes a running average raster of function F across the function
48
Figure 3.3. Solved intensity and argument of E
IR
from the measured
XFROG spectra shown in Fig. 3.2.
G. In Fourier space, taking FG = iF[F(F )F(G)], if G contains high frequency
noise andF , our convolution kernel, is of bandwidth smaller than the noise, then during
the multiplication the noise will be reduced as the high frequency components of G
will be multiplied by a small number. In reverse, the deconvolution of these functions
FG =iF
h
F(G)
F(F )
i
in a noisy system amplies the high frequency noise components.
Because of this noise amplication problem we need to apply a noise lter to the signal
before we perform the deconvolution. Programmatically, we do this by setting all high
frequency values outside of an envelope in Fourier space to zero. Figure 3.4 shows the
values of
~
f(
;t) overlaid with three possible envelope cutos. In the picture, all values
outside of a circle were nullied. The three examples shown in the gure represent a
49
situation where the envelope was too wide (dashed) and amplied noise was included,
a situation where the cuto was too tight and signal was chopped (dot-dashed), and
an intermediate situation (solid) where all the signal was included but amplied noise
on the order of the signal was cuto. The results from using these three cutos are
shown in Fig. 3.5. The red shows the intensity of the solved electric eld and the blue
shows the phase. There is not a signicant dierence between the middle and largest
cuto windows in this example although as the cuto window gets larger the amplied
noise would begin to dominate the signal. In the case where the cuto was too tight we
can see some sync function (the Foruier transform of our square well cuto) character
introduced into the nal result.
3.5 Using XFROG to Determine S(t)
Recall that in the beginning of the derivation (section 3.1) we assumed a purely non-
resonant signal with the induced eld from the IR eld as S(t
0
) = (t
0
). However as
spectroscopic chemists we are interested in the non-instantaneous vibrational excitations
that can be induced on surface bound molecules. In this case for a surface exhibiting a
number of induced free induction decays,
S(t) =(t)e
i
+H(t)
X

i
exp(i!
i
(t)
T
(t)) (3.19)
with H(t) as a Heavyside step function and e
i
setting the relative phase between the
instantaneous non-resonant part and non-instantaneously resonant parts. From here we
can revisit the formula forP
(1)
(t) given in equation 3.1 as motivation. We want perform
a similar procedure as before but to extract the resonant part ofS(t), where the chemical
information is. We perform a similar deconvolution useing the P
(1)
induction density
matrix from a non-resonant substrate which denitely is equal to the IR density matrix
above,
nonres
(t
0
;t
00
) and the P
(1)
from our chemically interesting substrate,
(1)
(t
0
;t
00
)
50
Figure 3.4. A representation of
~
f(
;t) with three dierent cuto mar-
gins overlaid.
resulting in the density matrix of only the resonant parts of the S(t) induction as shown
in Equation 3.20. The resonant signal S
res
(t) should be extractable using the same
Forward-Error-Correction method as before.

res
(t
0
;t
00
) =iF
!
0
!t
0iF
!
00
!t
00

F
t
0
!!
0F
t
00
!!
00(
(1)
(t
0
;t
00
))
F
t
0
!!
0F
t
00
!!
00(
nonres
(t
0
;t
00
))
!
(3.20)
51
-100 -50 0 50 100
-0.5
0.0
0.5
1.0
-0.75
-0.5
-0.25
0.
0.25
0.5
0.75
1.
fs
AU
p Rads
Figure 3.5. The results of the intensity (red) and argument (blue) of
E
IR
resulting from the three cutos performed on
~
f(
;t) in gure 3.4.
3.6 Time Gradient XFROG
The main obstacle to the implementation of XFROG is the time consuming nature of
a running a scanning setup. Taking in
uence from the GRAPES (Gradient-Assisted
Photon Echo Electronic Spectroscopy) method[109] developed by Greg Engel, it should
be possible to take a full XFROG spectrum, single shot, by dispersing onto a CCD
array detector. The current scanning setup scales linearly with time for the number
of scans. This is tolerable but slow for the experiments shown where very high SFG
signal gold was used as a sample. This, however, leaves the vast majority of the CCD
detractor unused. Instead of focusing both the visible and IR pulses into dots on the
substrate, they could be focused into lines using cylindrical lenses. A delay could be set
by introducing an asymmetrical path length between the two sides of the beam. Once
the signal was generated, the column of light could be directed into the monochromator
slit and dispersed on the full CCD. This would result in a single shot measurement of
XFROG and greatly increase the feasibility for the method as an analytical technique.
52
Chapter 4
SFG of an Azobenzene
Functionalized SAM
In this chapter we present a novel synthesis using simple protection chemistry to syn-
thesize an azobenzene functionalized alkane thiol molecule. This molecule can be used
to form SAMs that are either 100% functionalized with azobenzene, or contain a mix-
ture of azobenzene-functionalized thiols and simple alkane thiol \spacers". We then use
VSFG to characterize spectral properties of the monolayers to quantitatively determine
the relative concentrations of azobenzene functionalized thiol to alkane thiol spacer, and
study the isomerization behavior. Modeling of SFG signal was also performed to draw
conclusions about the orientation and intermolecular interactions of the surface bound
azobenzene upon isomerization.
4.1 Photochemistry of Azobenzene
Photoexcitation of trans-azobenzene primarily proceeds through one of two excitation
pathways. There can be excitation to a weak S
1
(n

) band at450 nm or there can
be excitation to a strong S
2
(

) band at340 nm. The standard model of isomeriza-
tion, shown in Fig. 4.1, is upon excitation to an excited state the molecule relaxes via
intersystem crossing to the S
0
state. At the intersection it will proceed to either the
trans-state where it started or the isomerized cis-species. Although the absolute yield of
isomerization can be quite small, the ensemble will transfer with preferential excitation
toward either the cis or trans. As the trans state is of lower energy than the cis state,
the cis! trans isomerization is thermally accessible at room temperature.
53
ΦNNC
Free Energy
60°
-60°
λ=340nm
λ=450nm
S0
Sn
NR
Δ
Figure 4.1. A cartoon of the isomerization excited states azobenzene
interacts with upon isomerization. Shown is the excitation to S
2
(

)
for trans azobenzene and excitation to the S
1
(n

) for cis.
The dynamics of isomerization were rst studied through a series of femtosecond
resolved transient absorption papers by Igor Lednev et al. [110, 111, 112]. With excita-
tion to the trans S
2
state the two time constants were t using a kinetic model. There
is a fast relaxation on the order of 1 ps, down to the S
1
state and a slower relaxation for
the non-radiative component on the order of 15 ps to the ground state. Two pathways
have been shown for azobenzene isomerization to have an inversion
ip over the CNN
bond and a rotation around the CNNC axis. It has been shown that excitation of trans-
azobenzene to the S
1
state proceeds through only the rotation pathway while excitation
to the S
2
state proceeds through both the rotation and inversion pathways.
The ecient non-radiative decay pathway of photoisomerization makes azobenzene a
poor emitter however femtosecond resolved
uorescence studies of azobenzene[113, 114]
54
have been performed to explicitly study the vibrational relaxation of the isomerization
process.
Vibrational spectroscopic studies have also been done. Both femtosecond modulated
resonance Raman [115] and femtosecond stimulated Raman[116, 117] studies have been
performed on azobenzene. The advantage of a vibrational approach is that it gives a
direct handle on molecular orientation through the change in Raman modes, rather than
interpolation of the orientation through kinetic models of
uorescence or absorption.
The isomerization yields have also been investigated in a variety of solvents giv-
ing reported quantum yields of 0.009 - 0.16 for S
2
(

)
trans!cis
, 0.21 - 0.33 for
S
1
(n

)
trans!cis
, 0.30 - 0.50 for S
2
(

)
cis!trans
, and 0.40 - 0.69 for S
2
(

)
cis!trans
[118].
Even though the quantum isomerization yield of the S
1
(n

)
trans!cis
is low it is good tar-
get for isomerization studies of the bulk ensemble as the primary excitation wavelength
at 340 nm is isolated from both the S
2
(

)
cis!trans
transition and the S
1
(n

)
cis!trans
transition.
4.2 Self-Assembled Monolayers
Thiol-alkane molecules are vital to the synthesis of well ordered, high density chem-
ically functionalized surfaces on
at gold surfaces. A thiol-alkane molecule forms a
strong thiolate-gold (100 kJ/mol) bond, allowing for strong and permanent xture of a
molecule to a gold surface with the Van-Der Waals interactions between the chains allow-
ing for vertical alignment at high densities. This alkane \forrest" allows for attachment of
dierent moieties for a variety of chemical surface functionalization applications. Besides
other switchable molecules[119, 120], SAMs have been used to functionalize surfaces with
biomolecules[121], electrochemically active molecules[122, 123], and nanoparticles[124].
Much of the appeal of thiolate-gold SAMs for building monolayers on a surface is the
wide variety of attached functional molecules, the mild conditions required for formation,
55
the high quality of lms, and the potential for economic scalability over other methods
such as Langmuir-Blodgett monolayer assembly or epitaxial growth.
4.2.1 Azobenzene functionalized SAMs
The trans - cis isomerization property of azobenzene makes it a popular target for func-
tionalization of surfaces. The light induced change of azobenzene monolayers have been
applied to a variety of devices including optical information storage[125, 126], controllable
wettable surfaces[127, 128], liquid crystal alignment[129, 130], and molecular electronic
switches[131, 132]. These applications all rely on either the electrical, mechanical, or
optical property change upon isomerization between the cis and trans isomers of the
azobenzene. The ultimate performance of such devices is dependent on a number of fac-
tors that aect isomerization such as stability, transition times and doses, and monolayer
formation[133]. To improve these photoactive surfaces and devices, it is rst important
to characterize the molecular organization and understand critical factors aecting iso-
merization such as packing, steric eects, and transition kinetics.
4.2.2 Azobenzene SFG literature
There already exists a limited literature of studying azobenzene functionalized surfaces
with SFG. The rst paper published was by Oh-e and coworkers[134] and studied the
orientation of a lipophilic azobenzene molecule on the water - air interface. They quanti-
tatively t the ensemble orientation using polarization dependent orientational analysis
for both the trans and cis orientations. A pair of papers by Ohe (no relation to Oh-e)
and coworkers[135, 136] examined the ordering and packing of an azobenzene molecule
containing poly(vinyl alcohol) polymer on the water-air interface. They studied the ori-
entation of the cis and trans phases with regard to surface packing. Orientation analysis
was performed in these papers by simulating the orientational dependent SFG using
DFT methods and tting to the measured results. The results show that for the trans
56
azobenzene there is little orientation change with increased surface packing, however the
cis isomer undergoes a 90

change with increased packing.
Only one paper to date studies a gold surface functionalized with an azobenzene
molecule. Wagner and coworkers[137] measured the SFG change of gold surfaces coated
with an azobenzene functionalized adamantine thiol tripod. They use SFG to observe
a change in signal intensity upon isomerization, and record kinetic values of the isomer-
ization cross-sections of their molecular system.
4.3 Synthetic Background
Most of the azobenzene thiols synthesized in the literature follow a similar general outline.
The azobenzene molecule is linked to an alkyl chain using a convenient chemical linker
with a terminal thiol. Yu and coworkers[138], use an amide linker synthesis to create
an azobenzene thiol (compound I, Figure 4.2). They start by synthesizing the acid
functionalized azobenzene using the classic Wallach synthesis[139]. The amide linker is
formed in two steps. First the carboxylic acid group is converted to an acyl chloride with
thionyl dichloride, and second there is a nucleophilic attack by a terminally functionalized
amine-thiol (H
2
N(CH
2
)
n
SH) to form the amide.
Tamada and coworkers'[140] method of creating a azobenzene thiol (compound II,
Figure 4.2) involves inclusion of an ether linker between the azobenzene and alkyl chain.
First they create an azobenzene ether using a Williamson ether synthesis reaction of
a phenyl-azo-phenol with a di-terminally bromonated alkane (1,12-dibromododecane)
yielding the azobenzene with an ether linked terminally bromonated alkane. The next
two steps convert the brominated alkane into a thiol terminated alkane. First the
azobenzene-alkane compound is re
uxed with thiourea to create a isothiouronium salt,
then upon workup in base, and wash in dry, deoxygenated ethanol yields the nal thiol
molecule. This synthesis has also been modied using the extremely pungent hexam-
ethyldisilathiane, which allows the formation of the thiol to proceed at 0

C[141].
57
Dietrich and coworkers[142] used an ester linker to azobenzene modify silicon (111)
surfaces (compoundII, Figure 4.2). This synthesis is dierent from the other two because
the SAM formed is on a silicon substrate and because the linking esterization reaction
is done not in solution phase but heterogeneously onto a carboxylic acid functionalized
SAM. The reaction is started by capping a Piranha cleaned Si-H (111) surface with a
methyl 10-undecenoate through a 20 hour 160

C reaction. This forms an alkyl SAM
with ester functionalization. The surface functionalized methyl ester is then cleaved and
converted to a carboxylic acid by heating in 2M HCl. The azobenzene is formed using
chemistry developed to allow formations of azobenzene molecules under mild conditions
with nearly arbitrary functional groups on each phenyl ring[143]. In this work they syn-
thesize an azobenzene with a nitrile on any of the 4 phenyl carbon locations of one phenyl
ring and a methanol on the opposing phenyl. The ester linkage is then formed between
the carboxylic acid functionalized surface and the alcohol functionalized azobenzene to
form the azobenzene SAM. The esterization proceeds under mild conditions with a tri-
azoles coupling reagent. The reason for performing the esterization directly onto the
surface instead of in solution and then binding the molecule to the silicon is that the
harsh conditions necessary to form the carbon-silicon bonds that link the SAM would
disrupt the relatively delicate ester.
One of the primary concerns with the creation of a new synthetic procedure for an
azobenzene thiol was to develop a synthesis that, 1: minimized use of highly toxic or
pungent compounds, 2: would proceed at room temperature without extreme oxygen or
water sensitivity, and 3: was synthetically accessible to a novice to multi-step organic
syntheses. The synthesis developed was an adaptation of the procedure developed by
Dietrich and co-authors to azobenzene functionalize a silicon substrate. Their use of
mild esterization reagents[144] developed for biological reactions to join the azobenzene
to a carboxylic acid functionalized alkane allowed for the purchase of a long alkyl chain
thiol instead of dealing with messy thiol chemistry.
58
N
N
O
S
N
N
NH
S
O
O
N
N
O
Si
CN
O
Au
I II III
Si
N
N
O
S
CN
O
Au
IV
Figure 4.2. Some previously used azobenzene surface functionalizing
molecules from the literature. I: From Jaschke et al. 1995. II: From Yu
et al. 1996 III: From Dietrich et al. 2008 IV: The system synthesized
for these studies.
We wanted to functionalize a gold surface over a silicon one for a number of reasons
instead of directly using the Dietrich synthesis. Spectroscopically, gold has the advantage
over silicon in that it provides a strong isotropic non-resonant SFG signal that hetero-
dynes with and amplies the resonant SFG signal from the surfactant molecule. Gold
doesn't have any azmuthilly angular dependence in the non-resonant signal, which would
require constant vigilance to retrieve quantitative spectral information from a function-
alized surface. The creation of an azobenzene functionalized thiol molecule also allowed
for the solution phase characterization of the molecule before formation of a SAM, as
opposed to the direct esterization onto a surface where the yield of the surface reaction
59
is dicult to rigorously characterize. Finally, the much greater bulk of literature on
azobenzene functionalization of gold surfaces versus that of silicon surfaces allows for
the chemical understanding gained from these studies to carry a greater signicance to
the literature.
One major dierence that had to be addressed in creating an azobenzene function-
alized thiol instead of an azobenzene ctionalized silicon surface is that the esterization
procedure required prior protection of the thiol molecule before it could proceed. The
alcohol on the azobenzene and the thiol on the thiol-carboxylic acid are both reac-
tive under esterization with the thiol-carboxylic acid. Without protection a major side
product would be polymerization of the thiol-carboxylic acid. To combat this we used
triphenylmethylchloride to protect the thiol before esterization, which could be removed
with tri
uroacetic acid afterward.
4.4 Synthetic Procedure
The synthesis of the azobenzene functionalized thiol is accomplished in four steps
(Scheme 1). First we synthesize the alcohol and nitrile functionalized azobenzene (I)
from the precursors 4-aminobenzonitrile and 4-aminobenzoalcohol[142, 143]. Second the
thiol group on the linker molecule is protected with tritylphenylchloride(II)[92]. Third
the protected linker and azobenzene are esterized(III)[142]. Finally a deprotection step
yields the product(IV).
The synthetic construction of the molecule is such that the nitrile group on the head
of the azobenzene acts as a spectroscopically convenient vibrational SFG chromophore.
Because of the high anity of thiol groups for gold interface, SAMs of binary composition
can be formed by mixing the azobenzene functionalized thiol with dodecane thiol in
ethanol at various ratios with the total concentration of both thiols 1 mM. A clean gold
substrate is then submersed in the thiol solution for 24 hours forming the SAM[145].
60
Figure 4.3. Synthetic scheme for the azobenzene functionalized alkane
thiol: a: Oxidation of 4-aminobenzonitrile with 2 equivalents of Oxone.
b: Addition of 1 equivalent of 4-aminobenzoalcohol. c: Protection of
15-mercaptopentadecanoic acid with trityl chloride. d: Esterization
between protected thiol and AZO. e: Deprotection with tri
uoroacetic
acid.
4.4.1 Synthesis of 4-(4-(Hydroxymethyl)phenyl)diazenyl benzonitrile
4-Aminobenzonitrile was dissolved in 10mL dichloromethane and stirred with 2 equiv-
alents of Oxone in 45mL water for 3 hours at RT under N2. The solution was
extracted with dichloromethane (DCM) twice and the organic phase was washed with 1M
hydrochloric acid, saturated bicarbonate solution, water, and brine. DCM was then evap-
orated under vacuum resulting in 76% yield 4-nitrosobenzonitrile. 4-nitrosobenzonitrile
and 4-aminobenzoalcohol were then dissolved at 1.2:1 molar ratio in 50mL acetic acid
and stirred under N
2
for 3 days. The solid product was ltered, washed with water,
dried and recrystalized in ethyl acetate (EtOAc) resulting in 36% yield product.
61
4.4.2 Protection, esterization and deprotection
15-Mercaptopentadecanoic acid and triphenylmethyl chloride were dissolved in 8mL N,N-
dimethylformamide (DMF) and stirred under N
2
for 6 hours. 50 mL of water was added
and the solution was extracted with ethyl acetate, washed with water and brine, and
the solvent evaporated under vacuum, resulting in product I with 67% yield. A solution
of 50 mL 3:2 acetonitrile:DMF was made with 1molar eq, compound 1, compound 2,
1H-benzotriazolium 1-(bis(dimethylamino)methylene)-5 chloro-,hexa
uorophosphate (1-
),3-oxide (HCTU), and 4-dimethylaminopyridine (DMAP). N,N-Diisopropylethylamine
(DIEA) was added and the solution was stired under N
2
for 24 hours. The solid was l-
tered, washed with acetonitrile, redissolved in DCM and dried under vacuum. The depro-
tection was performed by adding 1 mL tri
uroacetic acid to 0.2 g compound III at 0

C. 100l triisopropylsilane was added dropwise. The solution was stirred for 10 minutes
and dried under vacuum. The resultant product IV was puried using EtOAc/hexanes
column chromatography.
4.4.3 Formation of SAM
Solutions were prepared of mixtures of compound IV and 1-dodecanethiol up to 1 mM
concentration in absolute ethanol. Gold (111) substrates were cleaned with hot, fresh
Piranha solution (4:1 H
2
SO
4
:H
2
O
2
) for 15 minutes and sonicated in ethanol for 2 minutes.
The gold substrate was placed in solution for 24 hours for SAM formation. Afterward
the substrate was rinsed with ethanol and dried with N
2
.
4.5 Characterization of Products
Characterization of the synthesized products is also important to determine both that
the synthesis procedure is correct and also to determine the spectral properties of the
chemicals before SFG spectroscopy. One of the biggest restrictions of the established set
of characterization tools for organic synthesis is that they work best on solution phase
62
Figure 4.4.
1
H NMR of compound IV
systems. Then we are not certain if the solution phase results directly correspond to the
properties of the monolayer, but it is often a good starting point for an in-depth study
of these surfaces.
4.5.1 NMR
The peaks in the NMR of compound IV shown in gure 4.4 are as follow:
1
H NMR
(DMSO-d6, 400 MHz, ) 8.12 -8.07 (d, 2H), 8.02 - 7.97 (d, 2H), 7.92 - 7.75 (d, 2H),
7.45 - 7.38 (d, 2H), 4.62- 4.40 (d, 2H), 1.30-1.00 (m, 28H). The set of 4 doublet peaks
between 8.12 - 7.38 which integrated total to 8 correspond to the hydrogens attached
to the azobenzenes. The chemical shift of these are similar to that of benzene with some
modication depending on their proximity to the nitrile, azo, or methyl moiety attached
to the azobenzene. They are all doublets because each hydrogen only has one adjacent
hydrogen. The doublet at 4.62- 4.40 corresponds to the CH
2
proximal to the ester.
The broad multiple peak of 1.30-1.00 corresponds to the alkyl chain. The integration
to 28 matches the 14 CH
2
sites in the chain. We would also expect a singlet peak of
integral 1 at 5.50 corresponding to the thiol SH, although it's very possible this peak
was lost in broadening.
63
Figure 4.5. FTIR of compound IV
4.5.2 FTIR
FTIR was also taken on compound IV spotted on a CaF
2
plate, shown in gure 4.5,
with the following peaks: (cm
1
) 3080, 2920, 2860, 2230, 1740, 1690. The peaks from
3100 - 2900 cm
1
correspond to the CH stretch modes. The sharp peak at 2230 cm
1
is the nirtile marker on the azobenzene. The peaks at 1740 and 1690 cm
1
are from the
ester linker.
4.5.3 UV-Vis Spectroscopy
The electronic spectra of our azobenzene molecule can be measured with UV-Vis spec-
troscopy. The visible spectra of trans-azobenzene is dominated by two features, strong
S
2
(

) S
0
peaking at 340 nm and a weak shoulder S
1
(n

) S
0
on the red edge
of the S
2
. Cis-azobenzene has the S
2
(

) S
0
peak shifted to at 280 nm, a weak
shoulder S
2
(n

) S
0
and a more prominent S
1
(n

) S
0
peak peaking at 450 nm.
Figure 4.6 shows the change in spectra for dierent integrated exposure times to a 340
nm lamp. The spectra at T=0 is primarily trans-azobenzene, and by 120 minutes it
64
Figure 4.6. Trans! cis Isomerization of AZO molecule in DCM solvent
under 340nm UV light.
should be primarily in the cis form. Figure 4.7 is a detail on the rise of the cis S
1
S
0
peak. In Figure 4.8 shows the thermal relaxiation over 24 hours in linear absorption
spectrum of product I in DCM after being isomerized by a 340 nm lamp. All of these
spectra were taken with the molecule dissolved in DCM.
4.6 SFG on Azobenzene Functionalized SAMs
We prepare lms of various fractions of azobenzene surface coverages and observe their
spectroscopic properties and isomerization kinetics with VSFG. Figure 4.9 shows the
trend of the VSFG spectra for the PPP polarization in the nitrile stretch region (2220
cm
1
) from samples formed with 100% alkane thiol SAM, (zero solution volume percent
azobenzene functionalzed SAM) to neat (100%) azobenzene functionalzed SAMs. The
65
Figure 4.7. Detail from Fig. 4.6. Trans! cis Isomerization of AZO
molecule in DCM solvent under 340nm UV light.
Figure 4.8. Thermal cis ! trans isomerization of AZO molecule in
DCM solvent over 24 hours.
nitrile stretch at 2224 cm
1
appears as a negative peak against the nonresonant back-
ground signal of the gold substrate. The peak's amplitude decreases with the decreasing
azobenzene percentage.
66
Figure 4.9. Normalized PPP polarized SFG spectra from a set of dif-
ferent percentage formulated SAMs. The spectral peak at 2220 cm
1
is from the nitrile group on the azobenzene functionalized. The broad
background is the strong non-resonant SFG signal from gold.
VSFG provides the ability to quantitatively determine the concentration of a vibra-
tional chromophore on a surface. The spectra are t using the standard VSFG formalism
[75], where the surface 2nd order nonlinear susceptibility
(2)
is represented as a sum
of a nonresonant term
(2)
NR
= A
NR
e
i
describing the background signal from the gold
surface and a resonant Lorentzian term
(2)
Res
=
A
Res
(!!
CN
)+i
representing the nitrile vibra-
tional resonance for the total form of I
SFG
(!)/j
(2)
NR
+
(2)
Res
j
2
. The shape of the non
resonant contribution was determined from the SFG signal o a reference clean Au sub-
strate. Because the
(2)
Res
signal is proportional to surface coverage, we can determine
the relative coverage of dierent samples by tting the amplitude A
res
of the resonant
term. In 4.10 we show correlation between the amplitude of the resonant signal
(2)
Res
and the solution concentration of azobenzene terminated thiol used to form the SAM.
If the azobenzene terminated thiol and alkane thiol had equal binding constants and
non-cooperative absorption, then the amplitude of the resonant
(2)
Res
signal of the nitrile
67
Figure 4.10. AmplitudeB
CN
of the resonant CN stretch signal in PPP
polarized VSFG spectra (Normalized to linear relation), as a function of
the solution fraction of azobenzene-functionalized precursor molecules
used to form SAM. Dierent symbols correspond to dierent synthetic
batches and experimental preparations of the SAMs.
marker would be directly proportional the the azobenzene mole fraction in precursor
solution.
We also looked at the VSFG signal from the 2900 cm
1
region corresponding to the
C-H stretch vibrations of the saturated alkane chains. The primary sources of signal
in this region are the CH
3
vibrational modes of the terminal carbon of the alkane thiol
and the CH
2
vibrational modes of the alkane chain of both the alkane and azobenzene-
functionalized thiols. In the PPP spectra shown in Fig. 4.11, the primary peaks observed
for the neat alkane thiol SAMs are the CH
3
symmetric stretch( 2875 cm
1
), CH
3
asym-
metric stretch(2968 cm
1
) and CH
3
Fermi resonance (2930 cm
1
) vibrational modes. In
the SSP spectra, only the CH
3
symmetric (2875 cm
1
) and CH
3
Fermi resonance (2933
cm
1
) peaks are observed. This is consistent with previously reported literature SFG
spectra for PPP and SSP polarization of well ordered, vertically orientated alkane thiol
SAMs[51, 84, 146]. Underlying these peaks there are broader spectral features from the
CH
2
modes in the alkyl chains[76, 92, 147, 148] which become increasingly dominant as
68
Figure 4.11. Normalized SFG spectra of PPP (top) and SSP (bottom)
polarization of samples between 0% and 100% azobenzene terminated
C-H stretch modes.
the alkane thiol fraction decreases and azobenzene thiol fraction increases. The quali-
tative trend observed with these peaks conrms the expected decrease of CH
3
signal of
the alkane thiol as the solution percentage of azobenzene terminated thiol increases.
4.6.1 Isomerization behavior
VSFG can be also used to monitor orientational changes of the nitrile marker upon
azobenzene trans) cis or cis) trans isomerization. The magnitude and sign of the
69
Figure 4.12. Left: PPP SFG spectra for the nitrile stretch band of
a 66% azobenzene lm. Displayed are the experimental data for the
lm under 340 nm illumination (blue) and without (red). The spectra
are t with the non-resonant portion shown in yellow and the resonant
portion shown on the bottom. Right: Cartoon of mixed SAM on gold
surface, formed from a mixture of azobenzene functionalized thiol and
dodecane thiol.
change in
(2)
Res
upon isomerization of azobenzene is not trivial as it depends on the
distribution of all three orientational Euler angles of the reactant trans-azobenzene, as
detailed below. However, we were able to track in-situ the
(2)
resonant and non-
resonant magnitudes as well as their relative phase upon exposure to UV light (340 nm,
4.5 mW/cm
2
) to both quantify the magnitude of change upon isomerization and extract
kinetic parameters for the monolayers. Figure 4.12 shows VSFG spectra of the nitrile
stretch band of a 66% azobenzene lm without UV light exposure and with exposure
to 340 nm light. Also shown is the gaussian shaped non-resonant signalj
(2)
NR
j
2
(gold
line) and the extracted resonant contribution intensityj
(2)
Res
j
2
amplied by a factor of 10
(red and blue lines on the bottom). There was little observed change of the nonresonant
backgroundj
(2)
NR
j
2
correlated with UV illumination.
70
Figure 4.13. Change in amplitude of nitrile
(2)
Res
for 100%, 66% and
50% azobenzene SAMs taken at 10 second intervals upon isomerization
with 340 nm light and thermal relaxation.
Figure 4.13 shows the resonant SFG signal for the nitrile resonance for 3 consecu-
tive isomerization switching cycles alternating between 10 minute exposures to 340 nm
light and 20 minute thermal relaxation in the dark for 100%, 66% and 50% azobenzene
samples. For the 66% and 50% samples there is a clear modulation of the
(2)
Res
signal
with exposure to 340 nm light. Figure 4.14 shows 10 complete on/o cycles on a 66%
azobenzene SAM. This gives a good picture of the repeatability of isomerization in this
system. Although the modulation of
(2)
Res
is larger amplitude for the rst cycle, the
following 9 cycles show a very repeatable signal modulation upon isomerization.
4.7 Orientation of Azobenzene Functionalized SAMs
VSFG provides the ability to quantitatively determine the concentration of a vibrational
chromophore on a surface. The spectra are t using the standard VSFG formalism [75],
71
Figure 4.14. Change in amplitude of nitrile
(2)
Res
for 66% azobenzene
SAMs taken at 10 second intervals upon isomerization with 340 nm
light and thermal relaxation.
where the surface 2
nd
order nonlinear susceptibility
(2)
is represented as a sum of a
nonresonant term
(2)
NR
describing the background signal from the gold surface and a
resonant Lorentzian term
(2)
Res
representing the nitrile vibrational resonance

(2)
NR
=A
NR
e
i
(4.1)

(2)
Res
=
B
CN
(!!
CN
) +i
(4.2)
for the total SFG signal intensity of I
SFG
(!)/j
(2)
NR
+
(2)
Res
j
2
. The shape of the non reso-
nant contribution was determined from the SFG signal o a reference clean Au substrate.
Because the
(2)
Res
signal is proportional to surface coverage of the nitrile chromophore,
we can determine the surface fraction of azobenzene of dierent samples by tting the
amplitudeB
CN
of the resonant term. 4.10 shows the correlation between the amplitude
72
Figure 4.15. A: Change in amplitude of the nitrile
(2)
Res
under 340 nm
illumination for 100%, 66% and 50% azobenzene samples versus time.
B: Change in amplitude from the nitrile
(2)
Res
for thermal relaxation
in the dark following 10 min exposure to 340 nm light, for 100%, 66%
and 50% azobenzene samples versus time. Data sets were averaged
over multiple isomerization cycles. Solid lines show exponential ts
described in the text.
B
CN
of the resonant signal
(2)
Res
and the solution concentration of azobenzene termi-
nated thiol used to form the SAM. If the azobenzene terminated thiol and alkane thiol
had equal binding constants and non-cooperative absorption, then the amplitude of the
73
resonant
(2)
Res
signal of the nitrile marker would be directly proportional the the azoben-
zene mole fraction in precursor solution. While there may be a deviation from linear
behavior at low azobenzene fractions (below 25%), possibly indicating the cooperativity
in alkane thiol adsorption to exclude azobenzene functionalized thiols, the correlation is
linear within the signal-to-noise at azobenzene fraction of 50% and above, used in the
measurements described below. We therefore assume that the azobenzene mole frac-
tion in solutions used for SAM deposition represents the average surface coverage of the
azobenzene thiols in the formed SAMs on the surface.
The ratio of nitrile signals taken at dierent SFG polarization combinations, PPP
and SSP, can be used to determine the ensemble orientation. 4.16 shows both the
experimentally determined
(2)
PPP
=
(2)
SSP
amplitude ratio for dierent azobenzene samples
and a theoretical calculations described below. The experimental values in the left side
of the panel were collected for PPP and SSP polarizations at six dierent dilutions of
the azobenzene on the surface. The error bars shown account for both the error in the
tting of the
(2)
res
amplitudes and for multiple measurements on the same azobenzene
concentrations. Below 40% azobenzene concentration the relative error in the ratio is
large primarily due to the low level of the
(2)
SSP
signal. There appears to be a signicant
dierence in the
(2)
PPP
=
(2)
SSP
ratio measured for 100% and those of less than 100%,
although, within error, the mixed SAMs do not exhibit signicantly dierent response.
The
(2)
PPP
=
(2)
SSP
ratio for 100% is 21.5 1.2 and for <100% it is 18 1.0.
From these ratios we can obtain quantatively the orientation of the nitrile moiety
relative to the surface normal using a procedure developed by Zhuang, et al [94]. This
can be determined by comparing the ratios of the eective nonlinear susceptibilities:
74

(2)
SSP
=
yyz
L
yy
(!) L
yy
(!
1
) L
zz
(!
2
) sin
2
(4.3)

(2)
PPP
=
xxz
L
xx
(!) L
xx
(!
1
) L
zz
(!
2
) cos cos
1
sin
2
(4.4)

xzx
L
xx
(!) L
zz
(!
1
) L
xx
(!
2
) cos sin
1
cos
2
+
zxx
L
zz
(!) L
xx
(!
1
) L
xx
(!
2
) sin cos
1
cos
2
+
zzz
L
zz
(!) L
zz
(!
1
) L
zz
(!
2
) sin sin
1
sin
2
There is only one non-zero
(2)
tensor element contributing to the SSP, and four
elements for PPP, given in equations 4.5 - 4.7. The eective nonlinear susceptibilities
are composed of three pieces. The trigonometric terms depend on the angles of incidence
of the three beams: for outgoing SFG,
1
for visible and,
2
for IR. The terms are
the non-zero elements of the
(2)
tensor for c
1v
point symmetry:

xxz
=
yyz
=
1
2
N
s
[cos(1 +r)cos
3
(1r)] (4.5)

xzx
=
zxx
=
1
2
N
s
[coscos
3
(1r)] (4.6)

zzz
=N
s
[rcos +cos
3
(1r)] (4.7)
Herer is the ratio of molecular hyperpolarizability tensor elements; for chromophores
of cylindrical symmetry around the c axis, r =
aaa
ccc
=
aac
ccc
. The value r = 0:26 was
used for the nitrile moiety, in agreement with literature[94]. The tilt angle of the moiety
from surface normal is represented by , assuming a-function distribution, and N
s
, a
scaling constant which cancels out upon taking a ratio.
The \L" terms are the Fresnel factors in equations 4.8 - 4.10, which depend on
frequency. These are derived using a three layer model with the index of the two stacked
layers, n
1
and n
2
, as well as the index of the monolayer represented by n
0
[94].
75
L
xx
(!) =
2 n
1
(!) cos
n
1
(!) cos
+n
2
(!) cos
(4.8)
L
yy
(!) =
2 n
1
(!) cos
n
1
(!) cos +n
2
(!) cos
(4.9)
L
zz
(!) =
2 n
2
(!) cos
n
1
(!) cos
+n
2
(!) cos

n
1
(!)
n
0
(!)

2
(4.10)
The refractive index values used were n
air
=1 for all wavelength values and n
Au
=
0.16 + 3.55i at 675 nm, n
Au
= 0.18 + 5.11i at 800 nm, and n
Au
= 2.66 + 31.55i at 4500
nm from the Palik Handbook of Optical Constants of Solids[149]. The eective index of
the interfacial layer n
0
was taken to be 1.458, the index of bulk dodecane thiol.
On the left hand side of Figure 4.16, the calculated
(2)
PPP
=
(2)
SSP
ratio is plotted for tilt
angles values between 0

and 90

from surface normal. The ratio is calculated for three
values of n
0
(1.358, 1.458 and 1.558) to show how dierences in the index of the monolayer
aect the estimated t angle of the nirtile. As a result of this procedure, we can bracket
a range of orientational angles of the nitrile. For 100% azobenzene = 37

53

and for
<100% from = 45

60

.
4.7.1 Orientational Preference upon Isomerization
The magnitude of change in
(2)
Res
signal upon isomerization can be modeled to derive
more orientational information about the surface bound azobenzene moieties. One of the
diculties in modeling isomerization is that the geometry of the cis azobenzene isomer
orientation inherently has more degrees of freedom than the trans isomer. Even small
distribution of orientational angles for the trans azobenzene isomer therefore results
in a bi-or tri-modal distribution of orientational angles for the cis conguration after
isomerization. Here we address two questions: First, we seek to quantify the isomerized
fraction at saturation. Second, we want to determine if there is a preferential orientation
of the resulting cis isomer relative to the surface.
76
Figure 4.16. Left: PPP SFG spectra for the nitrile stretch band of
a 66% azobenzene lm. Displayed are the experimental data for the
lm under 340 nm illumination (blue) and without (red). The spectra
are t with the non-resonant portion shown in yellow and the resonant
portion shown on the bottom. Right: Cartoon of mixed SAM on gold
surface, formed from a mixture of azobenzene functionalized thiol and
dodecane thiol.
The same theory used to determine tilt angle of the nitrile moiety on the trans azoben-
zene precursor can be used to construct a model to determine the expected chance of
VSFG signal level upon isomerization from trans to various possible cis orientations.
The model assumed initial delta function distribution in of the nitrile chromophore of
the trans isomer. This, however, leaves uncertain the torsional orientation of the trans
azobenzene around the long axis of the molecule, i.e. the orientation of the -N=N- bond
relative to the surface (4.17). We assume that upon isomerization, the cis azobenene
77
molecules assume two possible conformations: \Up" and \Down". In the \Up" confor-
mation, the nitrile group has rotated by 114

away from the plane of the surface and
in the \Down" conformation, the isomerization occurs toward the surface. We consider
three dierent scenarios: (1) assumes equal ratio of \Up" and \Down" conformations
after isomerization (labeled \Both" in 4.17); (2) assumes \Up" isomerization for 50% of
the molecules while the remaining 50% are frozen in the trans form, e.g. as if \Down"
isomerization is sterically hindered; (3) assumes only \Down" isomerization for 50% of
the molecules with the remaining 50% frozen. While admittedly simplistic, this model
should capture the essential bias in orientational changes upon isomerization. An alter-
native scenario of free torsional rotation is presented in the Supporting Information, and
yields similar results for the calculated change in
(2)
Res
as scenario (1).
Equations 4.11-4.13 derive the expected relative change in
(2)
Res
signal for PPP polar-
ization (i.e. the ratio of
(2)
Res
amplitude after isomerization to that of the all-trans mono-
layer before isomerization) as a function of the initial trans tilt angle and the fraction
of isomerized molecules at saturation . The rst equation for
(2)
Both
=
(2)
Trans
describes
scenario (1) that starts with an orientation and after isomerization goes to equal con-
tributions of both+114

and114

. The second equation for
(2)
Up
=
(2)
Trans
only allows
\Up" isomerization for half of the molecules, while locking the orientation for the down-
ward isomerizing azobenzenes. The last equation for
(2)
Down
=
(2)
Trans
does the reverse,
allowing only \Down" isomerization for half of the molecules. The
(2)
PPP
() dependence
is evaluated using Equation 5.5. For these calculations the n
0
value was 1.458, although
nearly all of the n
0
dependence cancels when solving for change in signal.
78

(2)
Both

(2)
Trans
(;) =
j(1)
(2)
PPP
() +

2

(2)
PPP
( + 114

) +

2

(2)
PPP
( 114

)j
j
(2)
PPP
()j
(4.11)

(2)
Up

(2)
Trans
(;) =
j(1)
(2)
PPP
() +

2

(2)
PPP
( + 114

) +

2

(2)
PPP
()j
j
(2)
PPP
()j
(4.12)

(2)
Down

(2)
Trans
(;) =
j(1)
(2)
PPP
() +

2

(2)
PPP
() +

2

(2)
PPP
( 114

)j
j
(2)
PPP
()j
(4.13)
Figure 4.17 shows the solutions for Equations 4.11-4.13 for angles within the 45-60

range derived above (4.16). The observed experimental value is 0.91
(2)
Res
modulation
upon isomerization, indicated by the horizontal black bar. Each cone represents the
calculated values with regard to starting orientation and fraction of isomerized molecules
for dierent scenarios: (1) \Both" (Orange), (2) \Up" (Blue), and (3) \Down" (Red).
Although in these calculations the error introduced by the assumed n
0
value is small,
the uncertainty in the value shown in 4.16 remains. Our analysis allows for several
possible mechanisms of isomerization. If there is little preference between the \Up" and
\Down" cis states, then the fraction of isomerized molecules at saturation must be very
low (<10%). This would be possible, for example, if phase segregation occurs between
the azobenzene and alkane thiol spacers, and isomerization occurs only at the edges of
azobenzene domains. Another possibility suggested by 4.17 is that a large fraction of
molecules undergo isomerization but the \Up" conformation is strongly preferred over
the \Down" conformation, e.g due to steric constraint for the \Down" conguration.
This would result in a strong orientational preference of the cis-\Up" conformation
over the cis-\Down" conformation after UV exposure. Since the model assumes 50%
of the molecules undergoing `Up" mechanism isomerization, this is consistent with the
measured quantum yield of 5% (per molecule per photon) in the monolayer on the
surface being approximately half of that in solution ( 11%). Other studies have also seen
a high isomerization fraction for similar binary azobenzene SAMs[150, 151].
79
Figure 4.17. Three dierent scenarios for isomerization used: (1)
where isomerization yields equal amounts of \Up" and "Down" states.
(2) where only the \Up" states are free to isomerize (3) where only
the \Down" molecules can isomerize. The three cones (\Both":
Orange,\Up": Blue, \Down": Red) map out the expected
(2)
Res
change
upon UV exposure for the three models as a function of isomerization
fraction at saturation. The cones represent the range of tilt angles
= 45

60

of the trans isomer derived above (4.16). The black
horizontal bar represented the experimentally observed value of
(2)
Res
upon isomerization, 0.91.
80
4.7.2 Azobenzene Isomerization Under Other Polarizations
After running the modeling shown in Appendix D we went back to look at what pre-
dictions can be made for the observed change of SFG signal upon isomerization with
dierent SFG polarization combinations and starting angles not present in our system.
The results shown above can be extended to look at the expected change in VSFG
signal under isomerization for other polarization combinations besides PPP. Experimen-
tally PPP was used to observe isomerization not only because it had the best signal to
noise (by about a factor of 20) but also because these simulated predictions showed that
we would expect the larger modulation of VSFG signal under PPP for = 4560

than
SSP or SPS polarizations. All of the following gures were evaluated for a n
0
value of
1.458 and raman depolarization,r, of 0.26. Figure 4.18 shows the expected signal change
under isomerization PPP at = 45

and = 60

. This is the same as Figure 4.16 but is
included for reference. Figure 4.19 shows the expected isomerization change upon unity
isomerization with respect to changing the trans angle .
For these calculations, the negative values correspond to results where the argument
of the complex
ISO
has
ipped relative to
Trans
. This wouldn't be a visible change
if the spectroscopic method was only measuring the resonant contribution by itself,
j
(2)
j
2
. The signal measured is interfered with the relative phase of the non-resonant
SFG contribution from the gold surface. A change in relative phase would change the
negative peaks to positive and thus noting the change is signicant.
Figure 4.20 and Fig. 4.21 show the same relative gures for isomerization under
SSP polarization. For angles between 45 and 60

the relative VSFG signal change upon
isomerization is about half of that under PPP. Figure 4.22 and Fig. 4.23 show the change
under SPS polarization. For the angles of our system this polarization is yet even less
sensitive than PPP and SSP polarization. It is worth noting that a system with a highly
vertically oriented nitrile chromophore it should be very highly sensitive to isomerization.
These calculations show that for our system's trans orientation value of = 4560

,
PPP is the best polarization combination to choose, however this is not true for all
81
Figure 4.18. Simulation expected PPP polarization signal change upon
isomerization for Up (Red), Both (Orange) and Down (Blue) models
done at 45 (solid) and 60(dashed) degrees of nitrile o surface normal.
(the results shown here are redundant with gure 4.17, however this is
included for comparison with gure 4.20, and gure 4.22.
Figure 4.19. Simulation expected PPP polarization signal change for
unity isomerization with regard to trans angle of nitrile from surface
normal. Evaluated for models Up (Red), Both (Orange) and Down
(Blue).
82
Figure 4.20. Simulation expected SSP polarization signal change upon
isomerization for Up (Red), Both (Orange) and Down (Blue) models
done at 45 (solid) and 60(dashed) degrees of nitrile o surface normal.
Figure 4.21. Simulation expected SSP polarization signal change for
unity isomerization with regard to trans angle of nitrile from surface
normal. Evaluated for models Up (Red), Both (Orange) and Down
(Blue).
83
Figure 4.22. Simulation expected SPS polarization signal change upon
isomerization for Up (Red), Both (Orange) and Down (Blue) models
done at 45 (solid) and 60(dashed) degrees of nitrile o surface normal.
Figure 4.23. Simulation expected SPS polarization signal change for
unity isomerization with regard to trans angle of nitrile from surface
normal. Evaluated for models Up (Red), Both (Orange) and Down
(Blue).
84
starting distributions. At high angles (near surface parallel) all three polarization PPP
should also be the best combination because not only do all three models result in a large
relative change in signal but also because there is a large change in the \Both" model,
which isn't predicted for the SPS and SSP polarization selections. For surface normal
systems where = 0 PPP would not be the best choice as at total isomerization there
is still only a 50% relative change in signal. In the SSP polarization the signal should
nearly vanish from the trans! cis isomerization of a surface normal chromophore, and
for SPS there should be no trans signal but non-zero cis signal under all three models.
Of course, truly at= 0 all three models are equivalent because there can be no relative
bias towards or away from the substrate under the assumptions made.
4.7.3 Cone Model
We also developed another model for a cis-isomerization conrmation from a starting
delta function in the trans conguration. The model \Cone" takes a starting delta
function and converts to an isomerized angle distribution of a cone with an angle of 114

from the starting angle . Figure 4.24 visualizes this with the three arrows showing a
starting trans orientation and the corresponding color circle showing the cis distribution.
A line integral must then be performed over this cone with regards to conical angle
coordinate
p
in the projection of . The equation for this is shown in Equation 4.14.
In Equation 4.15 angle
p
is integrated over with the same isomerization ratio as used
in the other models. The code in Appendix D.6 implements this model.
(
p
;) =
Sin[
2
15
]Cos[
p
180
]Cos[
2
15
]Sin[
p
180
]Sin[]
q
jCos[
2
15
]Cos[
p
180
]Sin[]Sin[
2
15
]Sin[
p
180
]j
2
+jCos[
2
15
]Sin[
p
180
]Sin[]Sin[
2
15
]Cos[
p
180
]j
2
(4.14)
85

Cone
(;) =j
(1)
ppp
() +
R
2
0

PPP
((
p
;))d
p

ppp
()
j (4.15)
Figure 4.25 shows the results for this model for PPP, SSP and SPS polarizations at
45 and 60

with regards to isomerization ratio. Overall this model gives very similar
results to the \Both" model. Figure 4.26 shows the full angle resolved results from this
at unity isomerization. By the results of this simulation if one had a system to isomerize
with a surface normal orientation close to = 0 than SPS or PPP would be the best
polarization choice for the best relative signal modulation but for a sample with surface
parallel orientation then only PPP would show the desired information.
4.8 Discussion and Conclusions
We have synthesized and characterized a novel azobenzene functionalized SAM. The use
of the same surface attachment chemistry as in regular thiol SAMs on gold surfaces
allows us to prepare mixed monolayers where the azobenzene moieties are interspersed
with shorter alkanethiol spacers. This approach enables investigation of mechanistic
features of isomerization such as steric hindrance eects due to packing. Using VSFG
spectroscopy of the nitrile marker attached to the azobenzene moiety, we were able to
observe the spectral modes of SAMs of a binary composition of the azobenzene function-
alized and alkylated thiols and quantitatively determine the relative surface coverage in
mixed SAMs. Using theory previously established in the VSFG literature and taking
polarization dependent measurements of the nitrile signal we were able to quantitatively
determine molecular orientation relative to the surface normal for dierent dilutions of
the binary SAM. VSFG is also sensitive to the orientational changes of the azobenzene
moiety upon isomerization. Over 10 cycles of repeatable, reversible isomerization was
observed for dilute azobenzene functionalized lms. VSFG also allows measurements
of the photon dose required for photoisomerization from trans to cis conformation and
86
Figure 4.24. Visualization of the angle distribution of before and after
isomerization for \Cone" model. For each starting delta distribution
represented by an arrow nal distribution is the projection of the cone
below onto . The grey plane represents the surface plane by which
is calculated from.
thermal relaxation rates back to trans. We observe a pronounced steric hindrance of pho-
toisomerization: no conformational change occurs upon UV illumination in neat (100%)
azobenzene lms, presumably due to tight packing of the headgroups limiting the con-
formational mobility. However, the photoisomerization and thermal recovery readily
proceed and are reversible over multiple cycles when the azobenzene fraction is lowered
to 66% or 50% (i.e., for as little as 34% of the shorter alkanethiol spacer). Perhaps this
can be rationalized by the large amount of excess energy deposited into the azobenzene
moiety upon 340 nm UV excitation into the S
2
state (likely, over an eV above the curve
87
Figure 4.25. \Cone" model evaluated at 45

(solid) and 60

(dashed)
for PPP(orange), SSP (Red) and SPS(Blue) polarizations with regard
to isomerization ratio.
Figure 4.26. \Cone" model evaluated with regard to starting angle
for PPP(orange), SSP (Red) and SPS(Blue) polarizations.
crossing that leads to either cis or trans ground state conformation[110, 113]. Thus,
only a very tight packing would be able to prevent photoisomerization from occurring.
88
Lastly, we developed some models of what the observed chance in VSFG signal would
be for possible molecular conrmations for the isomerized SAMs. Fitting the data to
the observed VSFG signal modulation and an estimation of what the absolute surface
isomerization yield should be we have evidence that there may be a strong dependence in
isomerization prominently occurring away from surface normal. We hope that this work
will help increase the understanding of the isomerization mechanism by highlighting the
importance of steric eects, packing and crystallinity on the switching in dense lms
of actively isomeric molecules with either azobenzene functionality or other molecular
switches that involve conformational change.
89
Chapter 5
Second Harmonic Generation
Spectroscopy of Organic
Photovoltaic Materials
In this chapter we present Second Harmonic Generation (SHG) spectroscopic results
from conjugated polymer interfaces with a new near-IR scanning SHG setup. The new
setup has the ability to look at the second harmonic generation spectra in the range of
475-850 nm. The rst experiment with the setup is a to study the polymer interfaces
of P3HT/air and P3HT/C
60
. From the raw SHG data, modeling using the wavelength
dependent complex refractive index for each material is used to extract both the surface
and bulk components of the SHG data, and tensor elements of the second order surface
hyperpolarizibility,
(2)
. Lastly we repeat this analysis to look at the surface specic SHG
spectra of two new red absorbing polymers, P3HTT-10%DPP and P3HTT-10%TPD, and
compare their interface to that of P3HT.
5.1 Background
Photovoltaic devices based on conjugated conducting polymers have recently attracted
attention as an alternative solar energy platform. These polymers oer a number of
prospective advantages that could allow for them to be adapted as a viable photovoltaic
material. These include the potentially low cost of material synthesis, earth-abundant
constituent materials, ease of lm deposition from either sublimation or solution casting
90
methods, the prospect of chemical tunability of band gap and other material proper-
ties, and the ability to use
exible or oddly shaped substrates. At the same time the
record eciencies and many of the device parameters (open circuit voltage, short circuit
current, ll factor) are far behind state of the art inorganic designs. The relatively low
exciton mobility and rapid exciton recombination in organic and small molecule solar
devices compared to traditional silicon and thin lm inorganic devices[152] require e-
cient separation of bound exciton. One of the goals of this project is to help gain a robust
understanding of the morphology and energetics at the interface so that the mechanism
and ineciencies of charge transfer can be better understood.
Another issue that organic devices need to improve compared to inorganic materials is
in spectral solar overlap. The optimum material band gap for a single junction Shockley
{ Queisser limited cell is approximately 1.1 eV[153]. For materials such as P3HT the
HOMO-LUMO absorption band is nearly twice that. To this degree, there is interest
in new polymers which can absorb throughout the visible to pick up much more of the
lower energy solar photons[154, 155]. In particular, more understanding is needed of
the molecular orientation, morphology and electronic states on the OPV donor-acceptor
interface.
5.1.1 Device architectures
There are two main device architectures used in making OPV cells: lamellar hetero-
junction cells and bulk heterojunction (BHJ) cells. Many solar cell materials suer from
mismatched optical and electrical lengths scales: if the materials are too think then insuf-
cient light is absorbed, if it is too thick then the carries cannot be collected. Unlike
many inorganic cells most OPV materials have very short exciton diusion lengths, on
order of 10{20 nm[156]. Even though many of these materials have extremely high
extinction coecients[157], building an optically thick absorption layer still results in
the loss of many of captured photons.
91
Bulk heterojunction designs rely on the phase segregation between the donor and
acceptor layers of a cell to form an interlocking network of \ngers". This allows for the
optical thickeness of the cell to be large while at the same time the proximity of a D/A
interface from any given point in the cell reduces losses from exciton deactivation and
charge recombination.
For lamellar cells, although they are limited by this tradeo of exciton diusion length
vs light absorption, the liberation from the stringent morphological requirement to form
interlocking donor/acceptor phases allows for a much greater chemical diversity in cell
formation. The lamellar device geometry also allows for a much greater control of the
donor interface as the two layers are formed sequentially instead of forming from phase
separation.
Interfacial orientation is important to charge extraction and cell eciency in both of
these designs. Although the results shown in this chapter are from sample architectures
created to be in parallel to P3HT/C
60
lamellar cells we anticipate that the results shown
are generically applicable to both geometeries.
5.1.2 OPV Interfaces
There already exists a volume of literature investigating the surfaces and interfaces in
OPV systems, with many fundamental questions about the molecular interfaces. What
are the donor { acceptor molecular orientations? What the time scales are of charge
transfer and how are these aected by morphology and energetics? What exactly are the
interfacial intermediate charge transfer states? How can all of these factors be optimized
for device performance?
Probably the most studied interface is the donor/air interface. Since most surface
specic methods of measurement require a physically accessible interface it is an easy
target for study even though it isn't actually an interface present in any devices. X-ray
re
ectance studies were able to show the morphological changes in P3HT/PCBM BHG
cells upon annealing[158]. This method in conjunction with AFM was able to determine
92
the degree of phase segregation between the P3HT and PCBM and it's impact on surface
roughness. Kelvin probe force microscope experiments were able to image the hole defect
concentrations on the TIPS-pentacene surface in FET type devices[159]. There are a
number of studies which combine studies of surface roughness and morphology with
device properties thorough careful interpretation of the electrical measurements of solar
cells[160, 161]. One study used photoelectron spectroscopy of a pentacene monolayer on
bismuth that showed a hydrogen like orbital character to the interfacial charge transfer
states[162].
Another approach for studying surface specic signal is to study donor { acceptor
blends where the percentage of interfacial molecules is high enough that bulk techniques
will measure the surface specic signal. The main dierence between blended lms and
BHJ lms is the latter optimizes phase segregation to promote charge extrication in
a device format while blends haven't necessarily done that. Femtosecond pump-probe
measurements were taken of MEH-PPV / PCBM blends to study the interaction of the
carbonyl stretch in the PCBM to electronic pumping of the MEH-PPV[163]. These
geometeries have also been used to examine the donor to acceptor charge transfer rate
by bulk microwave conductivity measurements. [164]. Two dimensional
1
H
13
C NMR
studies were able to show the relative molecular position of P3HT andC
60
at the interface
before and after annealing[165].
The interfaces in these systems have also been simulated extensively. The character
of the P3HT { C
60
charge transfer state has been the subject of ab initio studies[166].
DFT calculations were able to determine the energy levels and localization of a CT state
between P3HT and C
60
. Groups have also simulated charge transfer at a much more
systematic level[167, 168]. High level simulations were able to show the role of interfacial
morphology on charge transfer rated on the acene/C
60
or acene/ZnO interfaces.
93
5.2 Synthesis and Sample Preparation
Specic details of the synthesis for the P3HT, P3HTT-10%TPD, and P3HTT-10%DPP,
gure 5.1, lms studied here are detailed in a series of papers by Beate Burkhart, Petr
P. Khlyabich, and Barry C. Thompson [169, 170, 171]. The polymers were synthesized
by using a platinum-tin catalyzed reaction to produce polymers of 17,500-24,500 dalton.
For the donor{acceptor copolymers, P3HTT-10%TPD, and P3HTT-10%DPP, synthesis
incorporates the acceptor block at 10% stoichiometric ratio in a semi-random manner.
Polymer layers were spun-cast in o-dichlorobenzene solution at spin speeds of 600, 1000,
1400, and 1800 RPM for 60 sec onto glass substrates. The fullerene layers were evapo-
rated on a matched set of P3HT samples at a rate of 2

A per second with a thermal
evaporator (Kurt J. Lesker Co.).
5.2.1 UV-VIS
Absorption spectra were measured for all the lms was measured from 200 - 1100 nm
using an UV-Vis spectrometer. The absorption spectra are primarily used to determine
the relative thicknesses of the spun layers, which is the main parameter of the modeling of
these lm's SHG activity and is the input for determining the complex index of refraction
for these materials. Figure 5.2 shows the absorption spectra for the neat P3HT and
P3HT/C
60
lms. The P3HT absorption peak is centered at 580 nm with a vibronic
shoulder visible to the blue. This peak is attributed to a

transition. For the
P3HT/C
60
lms we observe the same P3HT absorption spectra as in the neat P3HT lm
plus a peak due to the C
60
absorption. There is a sharp C
60
peak at 340 nm and broad and
weak absorption throughout the visible. Figure 5.3 shows the absorption spectra for the
P3HTT-10%DPP and P3HTT-10%TPD lms. These are the two polymers engineered
to absorb more in the red than P3HT. P3HTT-10%TPD has an absorption peak around
the same wavelength as P3HT at 560 nm, but the peak is both broader and has much less
pronounced vibronic structure. P3HTT-10%DPP has signicantly more red absorption
94
S
S
S
C
6
H
13
C
6
H
13
C
6
H
13
S
S
C
6
H
13
S
N
N
S
O
O
m
n
o
P3HTT-DPP-10%
P3HT
P3HTT-TPD-10%
S
S
C
6
H
13
S
N O
O
n
o
m: 80%
n: 10%
o: 10%
m: 80%
n: 10%
o: 10%
m
Figure 5.1. The three polymers synthesized for that were studied with
the SHG spectroscopy.
than P3HT, with the P3HT localized peak blue shifted a little to 500 nm, and large DPP
localized peaks at 690 nm and 770 nm.
95
Figure 5.2. Absorption spectra for 4 layers of P3HT spun on glass and
with 15 nm C
60
evaporated.
5.2.2 Sample Roughness and Thickness Measurements
An atomic force microscope image of the P3HT 600 rpm spun sample was measured to
determine the polymer roughness and thickness. Figure 5.4 shows a topographical image
of the P3HT surface. From this the RMS roughness of the lm was determined to be
3.9 nm. A scratch was also made in the lm with a razor blade and the AFM took a
prole scan image over the cut. From looking at the dierence between trench bottom
and P3HT surface over a set of 150 lines the thickness of the 600 RPM P3HT sample
was determined to be 38.7 1.5 nm.
5.2.3 Complex Refractive Index
The complex refractive index needed for SHG data analysis was determined for each
material by using the Kramers { Kronig relation between the imaginary component (k)
96
Figure 5.3. Absorption spectra for 4 dierent layers of P3HTT-
10%TPD, and P3HTT-10%DPP spun on glass.
of the material index and the real part (n). The imaginary component of the index was
solved from the material absorption (abs) at wavelength () and material thickness ()
using the form:
k() =
Abs()

4
(5.1)
The imaginary prat of the index, k, should be unitless with and canceling. The
real component of the index can now be computed using the Kramers { Kronig relation:
n(!) =n
1
+
2

Z
1
0
k(!
0
)
!
0
!
d!
0
(5.2)
This integral should be performed with regard to frequency. One concern is that
at !
0
= ! there is a singularity. This can be resolved using a Cauchy principal value
97
Figure 5.4. AFM taken of 600 RPM spun P3HT lm. Black Scale
bar corresponds to 1 m. Color bar goes from -16nm to 16nm vertical
displacement. The RMS roughness for this lm is 3.9nm.
method of the intergration. The unknown in this process is the asymptotic value of the
index,n
1
. This was calculated using variable angle spectroscopic ellipsometery (Sentech
SE850) on the P3HT, P3HT-10%TPD, and P3HT-10%DPP lms between 300 and 850
nm for angles between 50

and 70

from surface normal. This method was performed
on from absorption spectra and the results are shown in gure 5.5. The computational
code used from this procedure is shown in appendix E.1. This data are used later in the
tting models for the SHG spectroscopy.
5.2.4 Two Photon Absorption of P3HT
For SHG studies, it is important to know the TPA spectra because the selection rules for
a SHG-active transition state are that it must be both one and two photon absorption
active. Two photon absorption for poly(3-octylthiophene) in solution was taken by
Pfeer, et al., 1993 [172] (Fig. 5.6). They assigned two peaks in this spectrum: a strong
peak at 3.6 eV (about 0.6 eV to the blue of the 1PA absorption peak) attributed to
98
Figure 5.5. The real (n) and imaginary (k) components of the index of
refraction for polymers P3HT, P3HT-10%TPD, P3HT-10%DPP and
lm of C
60
as calculated by the Kramers { Kronig relation, Eq 5.2.
the
2
A
g

1
A
g
transition, and a weak peak which parallels the one photon absorption
which is attributed to
1
B
u

1
A
g
transition and centrosymmetic breaking disorder in the
thiophene chain backbone. The lower energy peak is also visible in -sexithienyl, which
has reduced symmetry in the chain direction[173]. Theoretical work has also conrmed
these peak assignments[174].
99
Figure 5.6. One and two photon absorption of poly(3-octylthiophene)
in tetrahydrofuran. This gure is adapted from Figures 1 and 4 from
Pfeer, et al., 1993 [172].
The literature does not cover the TPA spectra from thin spin-cast lms of P3HT
or the two P3HT co-polymers we studied. Although the solution phase poly and oligio-
thiophene TPA spectra provide important characterizations of the manner and symmetry
of the TPA peaks we should expect, they do not tell us the exact position of the peaks
for spun-cast P3HT lms or about the dierence expected in adding a co-polymer. We
made serious attempts to gather these spectra with Sean Roberts and Saptaparna Das
at the Stephen Bradforth group but were not able to get clean data.
5.3 SHG of the P3HT / C
60
Interface
Using scanning SHG electronic spectroscopy as described in Chapter 2, we recorded
spectra for spun-cast organic photovoltaic lms. Figure 5.7 shows the normalized SHG
100
intensity for 3 polarization combinations, PPP, PSS (2!,!,!), and S45 (2!,!), for neat
P3HT and P3HT/C
60
lms. From previous theory[97] we would expect that the PPP,
PSS and S45 signals may exhibit surface activity (shown in Chapter 1), however samples
exhibit very little activity for S45 selected polarization and these were not investigated in
depth. Both the P3HT/air and P3HT/C
60
samples show relatively large activity under
both PPP and PSS polarizations. PSS exhibits the largest SHG response from these
samples. The raw SHG spectra are blue-shifted relative to the one photon absorption
spectra. We note, however, that the raw SHG spectra may contain both surface and bulk
signal contributions. Film thickness analysis presented below allows us the disentangle
these contributions. The PSS spectra also exhibits a small etalon pattern on top of the
signal that is attributed to some interference with signal from the bottom, P3HT/glass
interface. The PPP polarization signal is less intense and shows a blue shifted response
relative to the PSS signal. These spectra in the absence of other data are dicult to
properly interpret. While the lms roughly show SHG response that is similar to that
of the direct product of the 1PA and 2PA spectra as expected[96], there is no trivial
way to directly extract meaningful chemical properties from the data. Spectra was also
taken on neat 15 nm C
60
lms and no signal was observable under any polarization in
this spectral range.
5.3.1 A Model for Extracting Surface SHG
To interpret the raw frequency dependent SHG spectra, it is necessary to account for
both the heterodyned bulk signal and for the Fresnel factors to extract the frequency
dependent surface SHG signal. To decompose the spectra we measure from these SHG
experiments we start with the equation for the SHG spectra we would expect to see from
the surface[93]. This shows that the signal is dependent on the power of the incumbent
laser, the index of the material and the hyperpolarizability of the sample,
(2)
eff
.
I
SHG
(2!)/
!
2
n(2!)n(!)
2
j
(2)
eff
j
2
I(!)
2
(5.3)
101
Figure 5.7. Raw SHG spectra from neat P3HT lms and P3HT lm
with 15 nm C
60
evaporated on top for PPP, PSS and S45 polarizations.
The specic non-zero molecular frame elements of
(2)
eff
tensor for dierent polariza-
tions can be constructed given in terms of the Fresnel factors, L
xx
, L
yy
, and L
zz
, and
incident angle of the laser beam, . For PPP polarization there is only one non-zero
element,
zyy
but for PPP polarization there are 4,
xxz
,
xzx
,
zxx
, and
zzz
.

(2)
PSS
=
zyy
L
zz
(2!)L
yy
(!)L
yy
(!)sin() (5.4)

(2)
PPP
=
xxz
L
xx
(2!)L
xx
(!)L
zz
(!)cos()
2
sin() (5.5)

xzx
L
xx
(2!)L
zz
(!)L
xx
(!)cos()
2
sin()
+
zxx
L
zz
(2!)L
xx
(!)L
xx
(!)cos()
2
sin()
+
zzz
L
zz
(2!)L
zz
(!)L
zz
(!)sin()
3
102
This formulation for the spectra represented with an interaction of a surface second-
order polarizability doesn't yet account for the expected SHG spectra. We also want
to account for the bulk generation from the samples. For this we are expanding our
hyperpolarizability with a bulk term,
(2)
=
(2)
Bulk
+
(2)
Surface
in this model. We make the
approximation that the bulk SHG generation from the samples is linear with thickness,
, and carries a relative phase to the surface SHG signal
(2)
Bulk
=e
i

(2)
Bulk
(!).
I
SHG
(2!)/
!
2
n(2!)n(!)
2
j
(2)
Surface
+e
i

Bulk
(!)j
2
I(!)
2
(5.6)
Since the PSS polarization combination only has one non-zero
(2)
eff
element we can
use this formulation to t spectra and solve for
zyy
.
I
PSS
(2!)/
!
2
n(2!)n(!)
2
j
zyy
L
zz
(2!)L
yy
(!)L
yy
(!) +e
i

PSS Bulk
(!)j
2
I(!)
2
(5.7)
The term
PSS Bulk
(!) also contains dependence on the frequency dependent re
ec-
tion and refraction in the bulk SHG generation. For PPP spectra the expressions are a
little bit more complex. Since
(2)
PPP
has 4 non-zero
(2)
eff
elements we cannot just solve
for them directly. To get around this we developed a factorization procedure so that
even if we can't extract
xxz
,
xzx
,
zxx
, and
zzz
explicitly we can solve for a linear
combination of the 4 of them.
I
PPP
(2!)/
!
2
n(2!)n(!)
2
jL
zz
(2!) (A
xxz
+B
xzx
C
zxx
+D
zzz
)+
+e
i

Bulk
(!)j
2
I(!)
2
(5.8)
The composition of constants A, B, C, and D start with the intuition that as in
our experiments! is non-resonant with the sample, and thus the index of each material
103
and thereby the Fresnel factors far from resonance can be set as constants. Looking
back at equations 5.4 and 5.5, the only wavelength dependent parts are now L
xx
(2!)
andL
zz
(2!) withL
xx
(!),L
yy
(!),L
zz
(!), and the geometrical factors, constants. The
observation was made that over the spectral wavelengths investigated, although oset,
both L
xx
(2!) and L
zz
(2!) followed similar trajectories. To account for this the factor
D
Lxx(2!)
Lzz (2!)
E
was introduced using the approximate spectrum,
L
xx
(2!)L
zz
(2!)

L
xx
(2!)
L
zz
(2!)

(5.9)
with

L
xx
(2!)
L
zz
(2!)

R
!
2
!
1
Lxx(2!)
Lzz (2!)
d!
j!
2
!
1
j
(5.10)
Now we can introduce the 4 factoring constants used for PPP,
A

L
xx
(2!)
L
zz
(2!)

L
xx
(!)L
zz
(!)cos()
2
sin() (5.11)
B

L
xx
(2!)
L
zz
(2!)

L
xx
(!)L
zz
(!)cos()
2
sin()
CL
xx
(!)L
xx
(!)cos()
2
sin()
DL
zz
(!)L
zz
(!)sin()
3
One note about this procedure: although for PSS spectra, the tting to
zyy
should
be material independent, for PPP depending on the material parameters of the system
the factoring constants will change. The factoring constants used for in these procedures
are shown in Table 5.1.
104
Interface A (
xxz
) B (
xzx
) C (
zxx
) D (
zzz
)
P3HT/Air 29.51 29.51 24.30 67.91
P3HT/C
60
-0.32 -0.32 0.36 0.047
P3HTT-10%DPP/Air 11.26 11.26 8.23 29.20
P3HTT-10%TPD/Air 11.61 11.61 8.52 29.96
Table 5.1. Factorization components for tting of I
PPP
(2!)
5.3.2 Which Surface are we Measuring?
As a quick interlude we need to treat the fact that in our SHG measurements there are
a few dierent interfaces which could generate second harmonic signal. The primary
interfaces that show 2! resonance in our spectral range are the P3HT/air interface,
P3HT/glass interface, and the P3HT/C
60
interface. There are two concerns here: both
the relative intensity of SHG signal from the dierent interfaces and the direction we
would expect the signal to be found. To simulate the relative intensities of the expected
signal from the interfaces we can use the formula for intensity of SHG in equations 5.3-
5.5 from the top (P3HT/air ) to bottom (P3HT/glass) interface. The ratio of expected
signal for PSS and PPP polarizations is calculated by taking the ratio shown,

(2)
PSS
(Top)

(2)
PSS
(Bottom)
=
!
2
n
P3HT
(2!)n
P3HT
(!)
2
j
(2)
PSSAir=P 3HT
j
2
I(!)
2
!
2
n
3
glass
j
(2)
PSSP 3HT=Glass
j
2
I(!)
2
T
s
(!)
2
T
p
(2!)e
l
(5.12)

(2)
PPP
(Top)

(2)
PPP
(Bottom)
=
!
2
n
P3HT
(2!)n
P3HT
(!)
2
j
(2)
PPPAir=P 3HT
j
2
I(!)
2
!
2
n
3
glass
j
(2)
PPPP 3HT=Glass
j
2
I(!)
2
T
p
(!)
2
T
p
(2!)e
l
(5.13)
The signal from the top surface is simply from equation 5.3, while for the bottom
surface the intensity of the incoming ! light is attenuated by the Fresnel transmission
factor (T
s
(!)) and the outgoing 2! light is attenuated by the absorption in the lm
(e
l
) and transmission back out through the top interface (T
p
(2!)). The two
(2)
factors for the top and bottom must also be treated with the proper Fresnel factors for
the interface as well as the incoming angle dictated by Snell's law. The results from
105
Figure 5.8. The calculated ratio SHG signal from top, P3HT/air inter-
face to bottom, P3HT/glass interface from Eq. 5.13.
these equations are graphed in Figure 5.8. For both PPP and PSS polarizations the top
interface should contribute the bulk of the SHG signal although for PSS there might be
<10% contribution from the bottom interface, which could be responsable for some of
the interference seen in the PSS spectra in Figure 5.7. In calculating PPP spectra the
approximation was made that all four tensor elements are comparable and have the
same wavelength dependence.
The direction of light refracted across an interface is dictated by Snell's law

2
=ArcSin

n
1
()Sin(
1
)
n
2
()

(5.14)
The change in direction is dictated by the refractive index of the two layers. Normally
with Snell's law of re
ection o of a buried interface the resultant output angle is restored
to the same as the input. In the case of SHG from a buried interface when the frequency
doubles the light nds itself in a medium with a dierent index of refraction and upon
performing Snell's law twice (n
1
!n
2
, then n
2
!n
1
) doesn't result in the same signal
output direction. For a material where n
1
(!) n
1
(2!) the output angle will be close
106
450 500 550 600 650 700 750
40
50
60
70
80
90
Wavelength HnmL
Signal Angle HqL
Figure 5.9. The angle expected of SHG signal for the P3HT/air inter-
face (blue), P3HT/glass interface (red), and the P3HT/C
60
interface
(brown).
to the input angle. However if there is a large change in the refractive index then the
angle of the SHG signal can be very dierent. Figure 5.9 shows the predicted direction of
the signal for the P3HT/air, P3HT/glass, and P3HT/C
60
interfaces for an input angle
of 61

o surface normal. We see that the P3HT/air and P3HT/C
60
expected signal
direction is close to that of the re
ected ! signal. For the buried P3HT/glass interface
the expected angle of the signal changes strongly as it crosses the band gap and for 615
- 645 nm we would expect that the signal is totally internally re
ected and channeled
into a waveguide mode. There is an expected overlap of the input and output signal at
555 nm. Similar behavior has been seen with the generation of four wave mixing in thin
metal lms[175].
107
Figure 5.10. Thickness dependent PSS SHG signal at 515nm for neat
P3HT and P3HT/C
60
lms (points) with t curves (lines) from equa-
tion 5.7. Each thickness has 4 redundant data points for better t.
The Y intercept of the t on the graph corresponds to the extrapo-
lated homodyne surface dependent signal.
5.3.3 Fitting Spectra with Thickness Dependent Model
Figure 5.10 shows SHG signal intensity for the P3HT and P3HT/C
60
samples versus
the P3HT layer thickness as well as the ts generated using from equation 5.7. For
both ts the relative phase used for the bulk spectra was zero. The tting allows us to
separate surface vs. bulk contributions to SHG signals. From the ts the extrapolation
of the intercept is indicative of the homodyne surface-only spectra (j
(2)
Surface
j
2
) and the
slope of the dependence is due to the bulk signal. Using this tting procedure across
the spectra taken we can extract both the surface and bulk SHG spectra as well as the
(often signicant) error of these ts.
108
Figure 5.11. PPP surface dependent homodyne SHG spectra of
t surface, j
(2)
Surface
j
2
, and bulk, j
(2)
Bulk
j
2
, P3HT/Air interface and
P3HT/C
60
.
Repeating the procedure shown in Figure 5.10 where the thickness dependent signal
can be t to equation 5.7, for each wavelength in the raw SHG spectrum extracts a
prediction of both the surface and bulk dependent spectral contributions to the overall
observed signal. Figure 5.11 shows the extractedj
(2)
Surface
j
2
andj
(2)
Bulk
j
2
spectra as well
as the relative error in tting for PPP polarizations of the P3HT/air interface and the
P3HT/C
60
interface. The extracted surface signal for both the P3HT/air and P3HT/C
60
interfaces are very similar with a peak at 540 nm. There is a slight dierence between
the observed bulk spectra for the two interfaces although it is not clear whether it is
statistically signicant. For P3HT/C
60
samples there is very weak and broad spectra
and no signicant signal observed for the neat P3HT samples. The reason for the very
small bulk signal observed in PPP is probably because the back P3HT/Glass interface
re
ects very little P polarized light. For both the surface and bulk PPP spectra the
extracted relative errors are small.
109
Figure 5.12. Fit PSS homodyne SHG spectra of t surface,j
(2)
Surface
j
2
,
and bulk,j
(2)
Bulk
j
2
, for neat P3HT and P3HT/C
60
lms.
In Figure 5.12 we perform this procedure for the thickness dependent PSS spectra,
separating them into surface and bulk components. The surface component,j
(2)
Surface
j
2
,
for the P3HT/air has a broad signal that peaks at 575 nm. The P3HT/C
60
interface
signal shows a similar red edge but is suppressed in the bluer side. There is a much
greater observed bulk signal under PSS polarization than PPP. For both the neat P3HT
and the P3HT/C
60
lms the bulk signals are similar in shape and to the blue of the
surface signal. The bulk signal appears to peak near 540 nm. In general the relative
error in the PSS tting is higher than that for PPP. This probably comes from some
mixing of signal from both the top and bottom interface combined with the signal.
Figure 5.14 shows the spectrum of surface nonlinear susceptibility tensor element

(2)
ZYY
extracted from tting thickness dependent SHG with equation 5.7. The dier-
ence between this gure and the spectra shown in Figure 5.12 is that the end result of
this tting is
zyy
which is a molecularly relevant parameter of the surface, while the
j
(2)
Surface
j
2
t above is representative of the eective spectra we observe that includes
110
Figure 5.13.
zyy
element of
(2)
extracted from PSS polarization SHG
spectra. Linear UV-Vis absorption spectrum is also shown.
frequency-dependent Fresnel factors with the bulk signal removed. The t
zyy
is shown
for PSS compared with the linear bulk absorption response of these lms. The spectra
of the P3HT/air interface shows good agreement with the shape of the 1PA absorption
of the lms on the red edge were showing a vibronic progression, but seems to drop o
on the blue side. However the
zyy
of the P3HT/C
60
interface shows a very dierent
response with suppression of the spectrum from that of P3HT/air signal, and a peak
value shifted from 575 nm to 540 nm.
For the PPP polarization the SHG is sensitive to the
xxz
,
xzx
,
zxx
, and
zzz
elements of
(2)
. The rst thing to note about the decomposition of spectra is that
the factorization of the P3HT/air interface and the P3HT/C
60
have dierent factors
for the four component
(2)
, shown in table 5.1, and are not directly comparable. The
deconstructed P3HT/air interface spectra is relatively similar in shape to the
zyy
spectra
determined from PSS polarization spectra. What appears to be a match of the vibronic
progression in the 1PA spectra is not present. Yet, as in the PSS spectra, the P3HT/C
60
interface spectrum peaks to the blue of the P3HT/Air spectra with a peak at 540 nm.
111
Figure 5.14. Elements of
(2)
extracted from PPP polarization SHG
spectra. ValuesA,B,C, andD used are from table 5.1. Linear UV-Vis
absorption spectrum is also shown.
5.3.4 Interpretations of P3HT orientation from SHG spectra
Revisiting the molecular intuition gained from evaluating these spectra, we recall (Chap.
1.3.1) that the
(2)
tensor can be constructed as the ensemble average of the direct
product of the molecular one photon absorption and two photon cross section projected
onto the interface-xed lab frame. As we know the direction of the transition dipole
for one photon excitation, we can start to evaluate which orientational ensembles the
dierent elements of
(2)
should be sensitive to. For instance, the
zyy
element should be
sensitive to orthogonal to the plane 1PA (z) and in-plane TPA (yy). From analysis of this
sort there are two types of information that can be extracted from the data. First the
actual surface specic transition energies for the polymers, and second, the dierences in
transition energries between orientational sub-ensembles. There are two inputs necessary
for this analysis. We need to know the orientation of the P3HT chromophores at the
surface and we also need to know the surface specic one and two photon absorption
tensors. The orientational information can potentially be obtained from vibrational SFG
experiments currently in progress in our group. We can make assumptions about both
112
these but fundamentally the spectra we measure include all three of these uncertain
factors.
One possibility for the large red shift observed between the extracted bulk and sur-
face SHG is the characterization of the two TPA active transitions observed in solution
polythiophene. In the solution spectra there is a weak TPA band that overlaps the 1PA
spectra where centrosymmetry of the P3HT backbone is locally broken, and a strong
band shifted to the blue which is two photon allowed for linear polythiophene chains
(Figure 5.6). One possibility is that in the bulk SHG there are fewer defect states,
and we primarily observe the integral overlap between the 1PA spectra and the bluer
TPA transition (
2
A
g

1
A
g
). At the interface we would expect more defect states that
break the linearity of the conjugated thiophene chain. The SHG spectra at the surface
then would be indicative of the overlap between the 1PA and the redder TPA transition
(
1
B
u

1
A
g
) that mirrors the 1PA. The red shift from bulk to surface measurements
could thus result from dierences in the defect states.
For both the PPP and PSS polarizations we see a signicant blue shift between
the spectra on the P3HT/C
60
interface to the specter at the P3HT/air interface. The
interfacial charge-transfer states reported in the literature[176, 177, 178] are characterized
with lower energies than the bulk absorption with a character of P3HT HOMO to C
60
LUMO. However the spectra we see can still be characterized as P3HT HOMO to P3HT
LUMO, by including band bending by proximity to the C
60
, which lowers both the
HOMO and LUMO levels relative to vacuum, with a more signicant eect on the
HOMO level. Figure 5.15 shows a cartoon of the band we assign our spectra to. It
shows the bulk absorption (green) and red-shifted charge transfer band (red) and the
blue shifted band from interface band-bending (blue). For both polarizations the shift
is from a peak absorption of about 580 nm for P3HT/air to a peak of 525 nm for
P3HT/C
60
. Band bending only accounts for about 10% of the shift for the band gap
dierences between P3HT and C
60
. A number of studies [165, 178, 179] have shown that
for charge transfer the P3HT/C
60
require the fullerene to be proximally orientated to
113
S
S
6
H
13
C
6
H
13
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
S
S
C
6
H
13
C
6
H
13
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
S
S
C
6
H
13
C
6
H
13
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
S
S
C
6
H
13
C
6
H
13
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
S
S
C
6
H
13
C
6
H
13
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
S
S
C
6
H
13
C
6
H
13
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
C
6
S
S
S
C
6
H
13
C
6
H
13
C
6
H
13
S
C
6
H
13
S
C
6
H
13
S
S
S
S
C
6
H
13
C
6
H
13
C
6
H
13
S
C
6
H
13
S
C
6
H
13
Figure 5.15. A cartoon of the band diagram of the P3HT/C
60
interface.
Left side shows the band diagram for P3HT and right for C
60
. The
three arrows represent the bulk absorption (green), interfacial charge
transfer state (red) and interfacial HOMO - LUMO transition (blue).
the conjugated systems of the P3HT. The transition dipole moment of P3HT has been
shown both experimentally [180, 181] and theoretically [178, 182] to be parallel to the
the polymer chain axis. Out of the non-zero
(2)
elements only
xxz
, and
xzx
, with the
rst \x" sensitive to the surface parallel 1PA dipole transition, will contain information
about the P3HT chromophores oriented horizontally to surface.
114
5.4 SHG of the Red-Absorbing Co-Polymer/ Air Interface
The surfaces of lms of randomly co-polymerized donor-acceptor polymers were also
studied using this scanning SHG spectroscopy. Spectra were taken from lms of P3HTT-
10%DPP and P3HTT-10%TPD. Figure 5.16 shows the raw SHG spectra for PPP and
SSP polarization for neat lms of P3HT, P3HTT-10%DPP, and P3HTT-10%TPD and
their comparable absorption spectra. For all three lms the SHG spectra peaks to the
blue of the 1PA spectra as in the P3HT spectra. The PPP polarization spectra is also
bluer in all cases than the PSS spectra. Both the P3HT and P3HTT-10%TPD have
non-zero SHG generation for nearly all of the overlap with the 1PA, but the P3HTT-
10%DPP goes to zero detectable signal signicantly to the blue of the lowest energy
absorption band edge. For the two donor-acceptor polymers the PPP polarization SHG
spectra have much higher signal than the PSS polarization spectra, which is the reverse
trend as observed with neat P3HT.
Figure 5.17 shows the decomposition of the raw PSS spectra into surface and bulk
contributions for P3HTT-10%DPP and P3HTT-10%TPD from equation 5.6. The bulk
and surface spectra for P3HTT-10%DPP show a similar story of oset with the bulk
spectra shifted to the blue of the surface. For P3HTT-10%TPD both the bulk and
surface components have very similar spectra with a peak at 530 nm.
Figure 5.18 shows the results from the decomposition procedure under PSS polar-
ization selection. For P3HTT-10%DPP the t bulk signal,j
(2)
Bulk
j
2
, shows two bands a
strong band with a peak at 590nm a broad shoulder on the red side. There are also two
bands in the surface spectrum, one at the blue edge of the spectra and a small band
peaking at 650 nm. Because of low signal level the decomposition tting fails to converge
for wavelengths greater than 700 nm, for P3HTT-10%DPP the PSS polarized signal is
much smaller than PPP polarization. For P3HTT-10%TPD the surface and bulk spectra
have very similar shapes with a broad shape and peak at 630 nm. Interestingly, the
115
Figure 5.16. Raw SHG spectra from neat P3HT, P3HTT-10%DPP,
and P3HTT-10%TPD lms for PPP, and PSS polarizations.
116
Figure 5.17. Fit PPP polarization homodyne SHG spectra of t sur-
face,j
(2)
Surface
j
2
, and bulk,j
(2)
Bulk
j
2
, P3HTT-10%DPP, and P3HTT-
10%TPD lms.
P3HTT-10%TPD for both PPP and PSS polarization was the only lm observed where
the surface and bulk had similar spectra.
The factorization of the PPP polarized spectra into components of
(2)
is shown in
gure 5.19 using the factors from Table 5.1. For P3HTT-10%DPP there is a single broad
peak at 690 nm. It's not clear if the spectra would extend out to the red edge of the
1PA as spectra near 800 nm was dicult to measure due to stray laser light from the
fundamental output (800 nm) of the Ti-Sapphire on the laser table. Similar to P3HT,
for the spectra from P3HTT-10%TPD we have very nice agreement with the red edge
of the 1PA spectra although there is drop o of the SHG in the blue side after 520 nm.
The spectra peaks at 530 nm.
For the extracted
zyy
, shown in Figure 5.20, for the P3HTT-10%DPP we have a
clear peak on the blue side of the spectra rising to 500 nm. The rise seen peaking to 650
nm in this analysis, if it's statistically signicant, is very weak. For P3HTT-10%TPD
117
Figure 5.18. Fit PSS polarization homodyne SHG spectra of t sur-
face,j
(2)
Surface
j
2
, and bulk,j
(2)
Bulk
j
2
, P3HTT-10%DPP, and P3HTT-
10%TPD lms.
The spectra peaks at 650 nm re
ecting the red edge of the 1PA absorption spectra. Yet
again, for PSS spectra the extremely low signal made statistically signicant spectra of

zyy
extremely dicult to extract.
One question that is not fully revealed in this analysis is the role of the dierent
chromophores in the co-polymer under this analysis. The most clear bi-chromophoric
system is the P3HTT-10%DPP which the peak at 500 nm attributable to the blue shifted
P3HT in the chain and the peaks at 700 nm from the DPP. The PPP spectra is dominated
by the
zzz
element and seems to show activity of the DPP. If there is also signal from
the DPP it's not statistically signicant. For the P3HTT-10%TPD/air interface we see
a clear dierence between the PPP and PSS spectra. Yet again the largest element in
PPP is from
zzz
. For P3HTT-10%TPD, like P3HT the
zyy
is redder than the factored
PPP spectra. Evaluation with regard to the two chromophores in P3HTT-10%TPD is
more dicult with the near degeneracy of the P3HT and TPD states. One possibility for
118
Figure 5.19. Elements of
(2)
extracted from PPP polarization SHG
spectra for the P3HTT-10%DPP/air, and P3HTT-10%TPD/air inter-
faces.. Values A, B, C, and D used are from table 5.1.
the fact that we see the overlap of the bulk and surface spectra from P3HTT-10%TPD
is that the introduction of the co-polymer breaks the symmetry in the P3HT segments
of the polymer and makes 1PA transition SHG allowed.
5.5 Conclusion and Analysis
Overall this is a phenomenological study of the electronic properties of polymer interfaces
and methods to extract them from SHG signal. The use of a non-directly resonant
scanning SHG spectrometer with high spectral resolution and the ability to cover most
the visible range allows us to construct models that disentangle the bulk second harmonic
119
Figure 5.20.
zyy
element of
(2)
extracted from PSS polarization
SHG spectra for the P3HTT-10%DPP/air, and P3HTT-10%TPD/air
interfaces.
response from that of the interface. A model was developed to extract surface specic
SHG from a mixture of surface signal and thickness dependent bulk signal. With this
study we are able to quantify the surface band bending dierence from the P3HT/air to
P3HT/C
60
interface, as well as explain the dierence in P3HT/C
60
interface from PPP to
PSS polarization. For the donor-acceptor copolymers we are able to gather spectra from
these new types of polymer and red-shift in the surface specic SHG spectra tracking the
red-shifted linear absorption. This method can be applied to other systems where the
interfacial electrical properties are of interest and adapted to evaluate material properties
important to the eciency of organic solar energy production.
120
One of the challenges of this project is the construction of a theoretical framework to
analyze these results. These measurements were performed in concert with vibrational
SFG studies on these same interfaces as well as large scale charge transfer simulations,
although as of yet a clear picture of the systems has not not been developed. The
recorded surface spectra do not readily reveal the interfacial charge transfer state pre-
sented in literature. The two main dierences comparing the interfacial charge transfer
to the donor-acceptor interfacial state that we see are that we are not observing excited
state spectra and that we have unique spectroscopic selection rules. To truly evaluate
the morphological information in these polarization-dependent SHG spectra in depth
they would need to be compared to calculated SHG spectra of dierent polymer surface
orientations.
121
Bibliography
[1] Peter R Rich and Jacques Breton. Attenuated total re
ection fourier transform
infrared studies of redox changes in bovine cytochrome c oxidase: resolution of the
redox fourier transform infrared dierence spectrum of heme a(3). Biochemistry,
41(3):967{973, January 2002. 1
[2] J A Mielczarski, A Denca, and J W Strojek. Application of attenuated total re
ec-
tion infrared spectroscopy to the characterization of coal. Applied Spectroscopy,
page 9981004, 1986. 1
[3] E. Betzig, P. L. Finn, and J. S. Weiner. Combined shear force and near-eld
scanning optical microscopy. Applied Physics Letters, 60(20):2484{2486, May 1992.
1
[4] H Menzel, M D Mowery, M Cai, and C E Evans. Surface{conned nanoparticles as
substrates for photopolymerizable self{assembled monolayers. Advanced Materials,
11:131{134, 1999. 1
[5] B. Pettinger, B. Ren, G. Picardi, R. Schuster, and G. Ertl. Nanoscale probing of
adsorbed species by tip-enhanced raman spectroscopy. Physical Review Letters, 92
(9):096101, March 2004. doi: 10.1103/PhysRevLett.92.096101. 1
[6] B. Pettinger, B. Ren, G. Picardi, R. Schuster, and G. Ertl. Tip-enhanced raman
spectroscopy (ters) of malachite green isothiocyanate at au(111): bleaching behav-
ior under the in
uence of high electromagnetic elds. Journal of Raman Spec-
troscopy, 36(6-7):541{550, June 2005. doi: 10.1002/jrs.1332. 1
[7] F Trager, T J Chuang, and H Coufal. Surface photoacoustic spectroscopy-a new
technique for the study of surface vibrations. Spectroscopy and Related Phenomena,
30:1924, 1983. 1
[8] F Trager, H Coufal, and T J Chuang. Infrared-laser photoacoustic spectroscopy
for surface studies: Sf
6
interaction with silver lms. Physical Review Letters, 49:
17201723, 1982. 1
[9] A Bayman, P K Hansma, W C Kaska, and L H Dubois. Inelastic electron tunnel-
ing spectroscopic study of acetylene chemisorbed on alumina supported palladium
particles. Applications of Surface Science, 14(2):194{208, January 1983. 1
122
[10] W Wang, T Lee, I Kretzschmar, and M A Reed. Inelastic electron tunneling
spectroscopy of an alkanedithiol self-assembled monolayer. Nano letters, 4:643646,
2004. 1
[11] W. Hauser and H. Feshbach. The inelastic scattering of neutrons. Physical Review,
87(2):366{373, 1952. doi: 10.1103/PhysRev.87.366. 1
[12] S. K. Sinha, E. B. Sirota, S. Garo, and H. B. Stanley. X-ray and neutron-
scattering from rough surfaces. Physical Review B, 38(4):2297{2311, August 1988.
doi: 10.1103/PhysRevB.38.2297. 1
[13] G. M. Luo, S. Malkova, J. Yoon, D. G. Schultz, B. H. Lin, M. Meron, I. Benjamin,
P. Vanysek, and M. L. Schlossman. Ion distributions near a liquid-liquid interface.
Science, 311(5758):216{218, January 2006. doi: 10.1126/science.1120392. 1
[14] D. K. Schwartz, M. L. Schlossman, E. H. Kawamoto, G. J. Kellogg, P. S. Pershan,
and B. M. Ocko. Thermal diuse x-ray-scattering studies of the water-vapor inter-
face. Physical Review A, 41(10):5687{5690, May 1990. doi: 10.1103/PhysRevA.
41.5687. 1
[15] P. A. Franken, A. E. Hill, C. W. Peters, and G. Weinreich. Generation of optical
harmonics. Phys. Rev. Lett., 7:118{119, Aug 1961. 2
[16] S. S. JHA. Theory of optical harmonic generation at a metal surface. Physical
Review, 140(6A):2020{&, 1965. 2
[17] P. A. Franken and J. F. Ward. Optical harmonics and nonlinear phenomena.
Reviews of Modern Physics, 35(1):23{39, 1963. 2
[18] D. A. Kleinman. Theory of second harmonic generation of light. Physical Review,
128(4):1761{&, 1962. doi: 10.1103/PhysRev.128.1761. 2
[19] J. Rudnick and E. A. Stern. Second-harmonic radiation from metal surfaces. Phys-
ical Review B, 4(12):4274{4290, 1971. doi: 10.1103/PhysRevB.4.4274. 2
[20] K. B. Eisenthal. Liquid interfaces probed by second-harmonic and sum-frequency
spectroscopy. Chemical Reviews, 96(4):1343{1360, June 1996. 3
[21] J. C. Conboy, J. L. Daschbach, and G. L. Richmond. Studies of alkane water
interfaces by total internal-re
ection 2nd-harmonic generation. Journal of Physical
Chemistry, 98(39):9688{9692, September 1994. 3
[22] K Bhattacharyya and E V Sitzmann. Study of chemical reactions by surface second
harmonic generation: p{nitrophenol at the air{water interface. The Journal of
Chemical Physics, 87:1442, 1987. 3
[23] S Ong, X Zhao, and K B Eisenthal. Polarization of water molecules at a charged
interface: second harmonic studies of the silica/water interface. Chemical Physics
Letters, 191:327335, 1992. 3
123
[24] D J Donaldson, J A Guest, and M C Goh. Evidence for adsorbed so2 at the
aqueous-air interface. The Journal of Physical . . . , 99:93139315, 1995. 3
[25] S Subrahmanyan and K B Eisenthal. Studies of surface diusion by second har-
monic
uctuation spectroscopy. Journal of Physical Chemistry, 94:3370{3373,
1990. 3
[26] P. B. Petersen and R. J. Saykally. Evidence for an enhanced hydronium concen-
tration at the liquid water surface. Journal of Physical Chemistry B, 109(16):
7976{7980, April 2005. doi: 10.1021/jp044479j. 3
[27] K. B. Eisenthal. Photochemistry and photophysics of liquid interfaces by second
harmonic spectroscopy. The Journal of Physical Chemistry, 100(31):12997{13006,
1996. 3
[28] P. J. Campagnola and L. M. Loew. Second-harmonic imaging microscopy for visu-
alizing biomolecular arrays in cells, tissues and organisms. Nature Biotechnology,
21(11):1356{1360, November 2003. 3
[29] E. Brown, T. McKee, E. diTomaso, A. Pluen, B. Seed, Y. Boucher, and R. K.
Jain. Dynamic imaging of collagen and its modulation in tumors in vivo using
second-harmonic generation. Nature Medicine, 9(6):796{800, June 2003. 3
[30] Duangporn Wanapun, Umesh S. Kestur, David J. Kissick, Garth J. Simpson, and
Lynne S. Taylor. Selective detection and quantitation of organic molecule crystal-
lization by second harmonic generation microscopy. Analytical Chemistry, 82(13):
5425{5432, July 2010. doi: 10.1021/ac100564f. 3
[31] Ronald D. Wampler, David J. Kissick, Christopher J. Dehen, Ellen J. Gualtieri,
Jessica L. Grey, Hai-Feng Wang, David H. Thompson, Ji-Xin Cheng, and Garth J.
Simpson. Selective detection of protein crystals by second harmonic microscopy.
Journal of the American Chemical Society, 130(43):14076{+, October 2008. doi:
10.1021/ja805983b. 3
[32] P. Guyot-Sionnest and Y. R. Shen. Bulk contribution in surface second-harmonic
generation. Phys. Rev. B, 38:7985{7989, Oct 1988. 3
[33] L. T. Cheng, W. Tam, W., S. H. Stevenson, G. R. Merdith, G. Rikken, and S. R.
Marder. Experimental investigations of organic molecular nonlinear optical polar-
izabilities .1. methods and results on benzene and stilbene derivatives. Journal of
Physical Chemistry, 95(26):10631{10643, December 1991. 3, 24
[34] L. T. Cheng, W. Tam, S. R. Marder, A. E. Steigman, G. Rikken, and C. W. Span-
gler. Experimental investigations of organic molecular nonlinear optical polariz-
abilities .2. a study of conjugation dependences. Journal of Physical Chemistry,
95(26):10643{10652, December 1991. 3, 24
124
[35] J. G. Mihaychuk, N. Shamir, and H. M. van Driel. Multiphoton photoemission
and electric-eld-induced optical second-harmonic generation as probes of charge
transfer across the Si/SiO
2
interfaces. Physical Review B, 59(3):2164{2173, January
1999. doi: 10.1103/PhysRevB.59.2164. 3
[36] Ming Lei, J. Price, and M. C. Downer. Hot carrier injection from nanometer-thick
silicon-on-insulator lms measured by optical second-harmonic generation. Applied
Physics Letters, 96(24):241105, June 2010. doi: 10.1063/1.3455317. 3
[37] Ian M Craig, Christopher J Tassone, Sarah H Tolbert, and Benjamin J Schwartz.
Second-harmonic generation in conjugated polymer lms: A sensitive probe of
how bulk polymer crystallinity changes with spin speed. The Journal of chemical
physics, 133(4):044901, 2010. 4
[38] T. Yamada, H. Hoshi, K. Ishikawa, H. Takezoe, and A. Fukuda. Origin of 2nd-
harmonic generation in vacuum-evaporated copper phthalocyanine lm. Japanese
Journal of Applied Physics Part 2-letters, 34(3A):L299{L302, March 1995. 4
[39] B. Koopmans, A. M. Janner, H. T. Jonkman, G. A. Sawatzky, and F. Vander-
woude. Strong bulk magnetic dipole induced second-harmonic generation from
c-60. Physical Review Letters, 71(21):3569{3572, November 1993. 4
[40] J. H. Hunt, P. Guyot-Sionnest, and Y. R. Shen. Observation of c-h stretch
vibrations of monolayers of molecules optical sum-frequency generation. Chem-
ical Physics Letters, 133(3):189{192, January 1987. doi: 10.1016/0009-2614(87)
87049-5. 4
[41] X. D. Zhu, H. J. Suhr, and Y. R. Shen. Surface vibrational spectroscopy by
infrared-visible sum-frequency generation. Journal of the Optical Society of Amer-
ica B-optical Physics, 3(8):P252{&, August 1986. 4
[42] X. D. Zhu, H., H. Suhr, and Y. R. Shen. Surface vibrational spectroscopy by
infrared-visible sum frequency generation. Physical Review B, 35(6):3047{3050,
February 1987. doi: 10.1103/PhysRevB.35.3047. 4
[43] P. Guyot-Sionnest, J. H. Hunt, and Y. R. Shen. Sum-frequency vibrational spec-
troscopy of a langmuir lm - study of molecular-orientation of a two-dimensional
system. Physical Review Letters, 59(14):1597{1600, October 1987. 4
[44] Jin Liu and John C. Conboy. Phase transition of a single lipid bilayer measured
by sum-frequency vibrational spectroscopy. Journal of the American Chemical
Society, 126(29):8894{8895, 2004. doi: 10.1021/ja031570c. 5
[45] Mark R Watry, Teresa L Tarbuck, and Geraldine L Richmond. Vibrational sum-
frequency studies of a series of phospholipid monolayers and the associated water
structure at the vapor/water interface. The journal of physical chemistry. B, 107
(2):512{518, January 2003. 5
125
[46] Gibum Kim, Marc Gurau, Joonyeong Kim, and Paul S. Cremer. Investigations of
lysozyme adsorption at the air/water and quartz/water interfaces by vibrational
sum frequency spectroscopy. Langmuir, 18(7):2807{2811, 2002. doi: 10.1021/
la0113365. 5
[47] Jie Wang, Sarah M. Buck, and Zhan Chen. Sum frequency generation vibrational
spectroscopy studies on protein adsorption. J. Phys. Chem. B, 106(44):11666{
11672, October 2002. ISSN 1520-6106. doi: 10.1021/jp021363j. 5
[48] Li Fu, Jian Liu, and Elsa C. Y. Yan. Chiral sum frequency generation spectroscopy
for characterizing protein secondary structures at interfaces. Journal of the Amer-
ican Chemical Society, 133(21):8094{8097, June 2011. 5
[49] Li Fu, Zhuguang Wang, and Elsa C. Y. Yan. Chiral vibrational structures of pro-
teins at interfaces probed by sum frequency generation spectroscopy. International
Journal of Molecular Sciences, 12(12):9404{9425, December 2011. 5
[50] Dequan Xiao, Li Fu, Jian Liu, Victor S. Batista, and Elsa C. Y. Yan. Amphiphilic
adsorption of human islet amyloid polypeptide aggregates to lipid/aqueous inter-
faces. Journal of Molecular Biology, 421(4-5):537{547, August 2012. doi:
10.1016/j.jmb.2011.12.035. 5
[51] Hongping Zhang and Steven Baldelli. Alkanethiol monolayers at reduced and oxi-
dized zinc surfaces with corrosion proctection: A sum frequency generation and
electrochemistry investigation. Journal of Physical Chemistry B, 110(47):24062{
24069, November 2006. doi: 10.1021/jp065248a. 5, 6, 68
[52] S. Rivera-Rubero and S. Baldelli. Surface characterization of 1-butyl-3-
methylimidazollum br-, i-, pf6-, bf4-, (cf3so2)(2)n-,scn-, ch3so3-, ch3so4-, and
(cn)(2)n- ionic liquids by sum frequency generation. Journal of Physical Chemistry
B, 110(10):4756{4765, March 2006. doi: 10.1021/jp0563989. 5
[53] C. R. Romero, H. P. Zhang, and S. Baldelli. Preparation and sfg studies of alka-
nethiol monolayers on mild steel surfaces. Abstracts of Papers of the American
Chemical Society, 230:1552015530, August 2005. 5
[54] P. S. Cremer, X. C. Su, Y. R. Shen, and G. A. Somorjai. Ethylene hydrogenation
on pt(111) monitored in situ at high pressures using sum frequency generation.
Journal of the American Chemical Society, 118(12):2942{2949, March 1996. 5
[55] X. C. Su, P. S. Cremer, Y. R. Shen, and G. A. Somorjai. High-pressure co oxida-
tion on Pt(111) monitored with infrared-visible sum frequency generation (SFG).
Journal of the American Chemical Society, 119(17):3994{4000, April 1997. doi:
10.1021/ja9638723. 5
[56] G. A. Somorjai and G. Rupprechter. Molecular studies of catalytic reactions on
crystal surfaces at high pressures and high temperatures by infrared-visible sum
frequency generation (sfg) surface vibrational spectroscopy. Journal of Physical
Chemistry B, 103(10):1623{1638, March 1999. doi: 10.1021/jp983721h. 5
126
[57] K. R. McCrea and G. A. Somorjai. Sfg-surface vibrational spectroscopy studies
of structure sensitivity and insensitivity in catalytic reactions: cyclohexene dehy-
drogenation and ethylene hydrogenation on pt(111) and pt(100) crystal surfaces.
Journal of Molecular Catalysis A-chemical, 163(1-2):43{53, December 2000. doi:
10.1016/S1381-1169(00)00398-8. 5
[58] S. Ye, T. Saito, S. Nihonyanagi, K. Uosaki, P. B. Miranda, D. Kim, and Y. R. Shen.
Stability of the si-h bond on the hydrogen-terminated si(111) surface studied by
sum frequency generation. Surface Science, 476(1-2):121{128, March 2001. doi:
10.1016/S0039-6028(01)00690-2. 5
[59] Fadel Y Shalhout, Sergey Malyk, and Alexander V Benderskii. Relative phase
change of nearby resonances in temporally delayed sum frequency spectra. The
Journal of Physical Chemistry Letters, pages 3493{3497, November 2012. 5
[60] P. B. Miranda and Y. R. Shen. Liquid interfaces: A study by sum-frequency
vibrational spectroscopy. Journal of Physical Chemistry B, 103(17):3292{3307,
April 1999. doi: 10.1021/jp9843757. 5
[61] Maria Sovago, R. Kramer Campen, Huib J. Bakker, and Mischa Bonn. Hydrogen
bonding strength of interfacial water determined with surface sum-frequency gen-
eration. Chemical Physics Letters, 470(1-3):7{12, February 2009. doi: 10.1016/j.
cplett.2009.01.009. 5
[62] Igor V. Stiopkin, Champika Weeraman, Fadel Shalhout, and Alexander V. Ben-
derskii. Dynamics of dangling od stretch at the air/water interface by heterodyne-
detected sfg spectroscopy. Abstracts of Papers of the American Chemical Society,
238:234503, August 2009. 5
[63] P. B. Miranda, Y. R. Shen, Igor V. Stiopkin, Champika Weeraman, Piotr A. Pieni-
azek, Fadel Y. Shalhout, James L. Skinner, and Alexander V. Benderskii. Hydrogen
bonding at the water surface revealed by isotopic dilution spectroscopy. Nature,
474(7350):192{195, April 2011. doi: 10.1021/jp9843757. 5
[64] Mikhail Vinaykin and Alexander V Benderskii. Vibrational sum-frequency spec-
trum of the water bend at the air/water interface. The Journal of Physical Chem-
istry Letters, 3(22):3348{3352, November 2012. 5
[65] G. L. Richmond. Molecular bonding and interactions at aqueous surfaces as probed
by vibrational sum frequency spectroscopy. Chemical Reviews, 102(8):2693{2724,
August 2002. doi: 10.1021/cr0006876. 5
[66] C. Schnitzer, S. Baldelli, and M. J. Shultz. Sum frequency generation of water on
nacl, nano3, khso4, hcl, hno3, and h2so4 aqueous solutions. Journal of Physical
Chemistry B, 104(3):585{590, January 2000. doi: 10.1021/jp992223l. 5
[67] D. E. Gragson and G. L. Richmond. Investigations of the structure and hydro-
gen bonding of water molecules at liquid surfaces by vibrational sum frequency
127
spectroscopy. J. Phys. Chem. B, 102(20):3847{3861, April 1998. ISSN 1520-6106.
5
[68] D. E. Gragson and G. L. Richmond. Potential dependent alignment and hydrogen
bonding of water molecules at charged air/water and ccl4/water interfaces. J. Am.
Chem. Soc., 120(2):366{375, January 1998. ISSN 0002-7863. 5
[69] P. B. Miranda, L. Xu, Y. R. Shen, and M. Salmeron. Icelike water monolayer
adsorbed on mica at room temperature. Physical Review Letters, 81(26):5876{
5879, December 1998. doi: 10.1103/PhysRevLett.81.5876. 5
[70] M. J. Shultz, C. Schnitzer, D. Simonelli, and S. Baldelli. Sum frequency genera-
tion spectroscopy of the aqueous interface: ionic and soluble molecular solutions.
International Reviews In Physical Chemistry, 19(1):123{153, January 2000. doi:
10.1080/014423500229882. 5
[71] M. J. Shultz, S. Baldelli, C. Schnitzer, and D. Simonelli. Aqueous solution/air
interfaces probed with sum frequency generation spectroscopy. Journal of Physical
Chemistry B, 106(21):5313{5324, May 2002. doi: 10.1021/jp014466v. 5
[72] S. Rivera-Rubero and S. Baldelli. Surface spectroscopy of room-temperature ionic
liquids on a platinum electrode: A sum frequency generation study. Journal
of Physical Chemistry B, 108(39):15133{15140, September 2004. doi: 10.1021/
jp048260g. 5
[73] C. Romero and S. Baldelli. Sum frequency generation study of the room-
temperature ionic liquids/quartz interface. Journal of Physical Chemistry B, 110
(12):6213{6223, March 2006. doi: 10.1021/jp0565150. 5
[74] Steven Baldelli. Surface structure at the ionic liquid-electried metal interface.
Accounts of Chemical Research, 41(3):421{431, March 2008. 5
[75] Igor V. Stiopkin, Himali D. Jayathilake, Champika Weeraman, and Alexander V.
Benderskii. Temporal eects on spectroscopic line shapes, resolution, and sensi-
tivity of the broad-band sum frequency generation. Journal of Chemical Physics,
132(23):234503, June 2010. doi: 10.1063/1.3432776. 6, 67, 71
[76] A. N. Bordenyuk, H. Jayathilake, and A. V. Benderskii. Coherent vibrational
quantum beats as a probe of langmuir-blodgett monolayers. Journal of Physical
Chemistry B, 109(33):15941{15949, August 2005. 6, 68
[77] Alexei Lagutchev, Aaron Lozano, Prabuddha Mukherjee, Selezion A Hambir, and
Dana D Dlott. Compact broadband vibrational sum-frequency generation spec-
trometer with nonresonant suppression. Spectrochimica Acta Part A: Molecular
and Biomolecular Spectroscopy, 75(4):1289{1296, April 2010. 6
[78] Alexei S Lagutchev, James E Patterson, Wentao Huang, and Dana D Dlott. Ultra-
fast dynamics of self-assembled monolayers under shock compression: Eects of
128
molecular and substrate structure. The journal of physical chemistry. B, 109(11):
5033{5044, March 2005. 6
[79] Jerey A. Carter, Zhaohui Wang, Hiroki Fujiwara, and Dana D. Dlott. Ultra-
fast excitation of molecular adsorbates on
ash-heated gold surfaces. Journal of
Physical Chemistry A, 113(44):12105{12114, November 2009. 6
[80] John A McGuire and Y Ron Shen. Ultrafast vibrational dynamics at water inter-
faces. Science, 313(5795):1945{1948, September 2006. 6
[81] Zhen Zhang, Lukasz Piatkowski, Huib J. Bakker, and Mischa Bonn. Ultrafast
vibrational energy transfer at the water/air interface revealed by two-dimensional
surface vibrational spectroscopy. Nat Chem, 3(11):888{893, November 2011. ISSN
1755-4330. 6
[82] Wei Xiong, Jennifer E. Laaser, Randy D. Mehlenbacher, and Martin T. Zanni.
Adding a dimension to the infrared spectra of interfaces using heterodyne detected
2d sum-frequency generation (hd 2d sfg) spectroscopy. Proceedings of the National
Academy of Sciences of the United States of America, 108(52):20902{20907,
December 2011. doi: 10.1073/pnas.1115055108. 6
[83] Prashant Chandra Singh, Satoshi Nihonyanagi, Shoichi Yamaguchi, and Tahei
Tahara. Ultrafast vibrational dynamics of water at a charged interface revealed by
two-dimensional heterodyne-detected vibrational sum frequency generation. The
Journal of chemical physics, 137(9):094706, September 2012. 6
[84] M. Himmelhaus, F. Eisert, M. Buck, and M. Grunze. Self-assembly of n-alkanethiol
monolayers. a study by ir-visible sum frequency spectroscopy (sfg). Journal of
Physical Chemistry B, 104(3):576{584, January 2000. doi: 10.1021/jp992073e. 6,
68
[85] Francesca Cecchet, Dan Lis, Julien Guthmuller, Benoit Champagne, Yves Cau-
dano, Christophe Silien, Alaa Addin Mani, Paul A. Thiry, and Andre Peremans.
Orientational analysis of dodecanethiol and p-nitrothiophenol sams on metals with
polarisation-dependent sfg spectroscopy. ChemPhysChem, 11(3):607{615, FEB 22
2010. ISSN 1439-4235. 6
[86] G. Fonder, F. Cecchet, A. Peremans, P. A. Thiry, J. Delhalle, and Z. Mekhalif. Con-
formational order of n-dodecanethiol and n-dodecaneselenol monolayers on poly-
crystalline copper investigated by pm-irras and sfg spectroscopy. Surface Science,
603(15):2276{2282, August 2009. 6
[87] Manish Dubey, Tobias Weidner, Lara J. Gamble, and David G. Castner. Structure
and order of phosphonic acid-based self-assembled monolayers on si(100). Lang-
muir, 26(18):14747{14754, September 2010. 6
[88] J. E. Baio, T. Weidner, J. Brison, D. J. Graham, Lara J. Gamble, and David G.
Castner. Amine terminated sams: Investigating why oxygen is present in these
129
lms. Journal of Electron Spectroscopy and Related Phenomena, 172(1-3):2{8,
May 2009. 6
[89] C. Humbert, B. Busson, J.-P. Abid, C. Six, H.H. Girault, and A. Tadjeddine. Self-
assembled organic monolayers on gold nanoparticles: A study by sum-frequency
generation combined with uv{vis spectroscopy. Electrochimica Acta, 50(15):3101{
3110, May 2005. ISSN 0013-4686. doi: 10.1016/j.electacta.2004.10.084. 6
[90] Champika Weeraman, Achani K. Yatawara, Andrey N. Bordenyuk, and Alexan-
der V. Benderskii. Eect of nanoscale geometry on molecular conformation: Vibra-
tional sum-frequency generation of alkanethiols on gold nanoparticles. Journal
of the American Chemical Society, 128(44):14244{14245, November 2006. doi:
10.1021/ja065756y. 6
[91] Tobias Weidner, Julia S. Apte, Lara J. Gamble, and David G. Castner. Probing
the orientation and conformation of alpha-helix and beta-strand model peptides on
self-assembled monolayers using sum frequency generation and nexafs spectroscopy.
Langmuir, 26(5):3433{3440, March 2010. doi: 10.1021/la903267x. 7
[92] Achani K. Yatawara, Gopinath Tiruchinapally, Andrey N. Bordenyuk, Peter R.
Andreana, and Alexander V. Benderskii. Carbohydrate surface attachment char-
acterized by sum frequency generation spectroscopy. Langmuir, 25(4):1901{1904,
February 2009. doi: 10.1021/la803113h. 7, 60, 68
[93] YR Shen. Surface-properties probed by 2nd-harmonic and sum-frequency genera-
tion. Nature, 337(6207):519{525, FEB 9 1989. ISSN 0028-0836. 7, 37, 101
[94] X. Zhuang, P. B. Miranda, D. Kim, and Y. R. Shen. Mapping molecular orientation
and conformation at interfaces by surface nonlinear optics. Physical Review B, 59
(19):12632{12640, May 1999. doi: 10.1103/PhysRevB.59.12632. 7, 74, 75
[95] G J Simpson. New tools for surface second-harmonic generation. Applied Spec-
troscopy, 55(1):16{16, 2001. 7
[96] AJ Moat and Simpson GJ. A unied treatment of selection rules and symmetry
relations for sum-frequency and second harmonic spectroscopies. J. Phys. Chem.
B, 108:354862, 2004. 7, 101
[97] Poul B Petersen and Richard J Saykally. On the nature of ions at the liquid water
surface. Annual Review of Physical Chemistry, 57(1):333{364, May 2006. 7, 101
[98] R. Trebino. Frequency-Resolved Optical Gating: The Measurement of Ultrashort
Laser Pulses. Kluwer Academic Publishers, 2002. 12, 39
[99] M. Simon, F. Trager, A. Assion, B. Lang, S. Voll, and G. Gerber. Fem-
tosecond time-resolved second-harmonic generation at the surface of alkali metal
clusters. Chemical Physics Letters, 296(5-6):579{584, November 1998. doi:
10.1016/S0009-2614(98)01073-2. 24
130
[100] Shoichi Yamaguchi and Tahei Tahara. Determining electronic spectra at interfaces
by electronic sum frequency generation: One- and two-photon double resonant
oxazine 750 at the air/water interface. Journal of Chemical Physics, 125(19):
194711, November 2006. doi: 10.1063/1.2375093. 25
[101] Shoichi Yamaguchi and Tahei Tahara. Heterodyne-detected electronic sum fre-
quency generation: "up" versus "down" alignment of interfacial molecules. Journal
of Chemical Physics, 129(10):101102, September 2008. doi: 10.1063/1.2981179. 25
[102] A. Brodeur and S. L. Chin. Ultrafast white-light continuum generation and self-
focusing in transparent condensed media. Journal of the Optical Society of America
B-optical Physics, 16(4):637{650, April 1999. 25
[103] R. L. Fork, C. V. Shank, C. Hirlimann, R. Yen, and W. J. Tomlinson. Femtosecond
white-light continuum pulses. Optics Letters, 8(1):1{3, 1983. 25
[104] F. Raoult, A. C. L. Boscheron, D. Husson, C. Sauteret, A. Modena, V. Malka,
F. Dorchies, and A. Migus. Ecient generation of narrow-bandwidth picosecond
pulses by frequency doubling of femtosecond chirped pulses. Optics Letters, 23
(14):1117{1119, jul 1998. doi: 10.1364/OL.23.001117. 26
[105] C. Iaconis and I. A. Walmsley. Spectral phase interferometry for direct electric-
eld reconstruction of ultrashort optical pulses. Optics Letters, 23(10):792{794,
May 1998. doi: 10.1364/OL.23.000792. 39
[106] G. Stibenz and G. Steinmeyer. Interferometric frequency-resolved optical gating.
Optics Express, 13(7):2617{2626, April 2005. doi: 10.1364/OPEX.13.002617. 39
[107] J. C. Diels, E. Vanstryland, and G. Benedict. Generation and measurement of 200
femtosecond optical pulses. Optics Communications, 25(1):93{96, 1978. 39
[108] D.G.C Maccay. Information Theory, Interference, and Learning Algorithms. Cam-
bridge University Press, 2003. 45
[109] Elad Harel, Andrew F. Fidler, and Gregory S. Engel. Single-shot gradient-assisted
photon echo electronic spectroscopy. The Journal of Physical Chemistry A, 115
(16):3787{3796, 2011. doi: 10.1021/jp107022f. 52
[110] IK Lednev, TQ Ye, RE Hester, and JN Moore. Femtosecond time-resolved uv-
visible absorption spectroscopy of trans-azobenzene in solution. Journal of Physical
Chemistry, 100(32):13338{13341, AUG 8 1996. ISSN 0022-3654. 54, 88
[111] IK Lednev, TQ Ye, LC Abbott, RE Hester, and JN Moore. Photoisomerization
of a capped azobenzene in solution probed by ultrafast time-resolved electronic
absorption spectroscopy. Journal of Physical Chemistry A, 102(46):9161{9166,
NOV 12 1998. ISSN 1089-5639. 54
131
[112] IK Lednev, TQ Ye, P Matousek, M Towrie, P Foggi, FVR Neuwahl, S Umapa-
thy, RE Hester, and JN Moore. Femtosecond time-resolved uv-visible absorption
spectroscopy of trans-azobenzene: dependence on excitation wavelength. Chemical
Physics Letters, 290(1-3):68{74, JUN 26 1998. ISSN 0009-2614. 54
[113] T Fujino, SY Arzhantsev, and T Tahara. Femtosecond time-resolved
uorescence
study of photoisomerization of trans-azobenzene. Journal of Physical Chemistry
A, 105(35):8123{8129, SEP 6 2001. ISSN 1089-5639. 54, 88
[114] CW Chang, YC Lu, TT Wang, and EWG Diau. Photoisomerization dynamics of
azobenzene in solution with s-1 excitation: A femtosecond
uorescence anisotropy
study. Journal of the American Chemical Society, 126(32):10109{10118, AUG 18
2004. ISSN 0002-7863. 54
[115] P. D. Chowdary and S. Umapathy. Ultrafast dynamics and photochemistry of
pi-pi * excited trans-azobenzene - a comprehensive analysis of resonance raman
intensities. Journal of Raman Spectroscopy, 39(11):1538{1555, November 2008. 55
[116] Christina M. Stuart, Renee R. Frontiera, and Richard A. Mathies. Excited-state
structure and dynamics of cis- and trans-azobenzene from resonance raman inten-
sity analysis. Journal of Physical Chemistry A, 111(48):12072{12080, December
2007. doi: 10.1021/jp0751460. 55
[117] David P. Homan and Richard A. Mathies. Photoexcited structural dynamics
of an azobenzene analog 4-nitro-4 '-dimethylamino-azobenzene from femtosecond
stimulated raman. Physical Chemistry Chemical Physics, 14(18):6298{6306, 2012.
doi: 10.1039/c2cp23468h. 55
[118] Giovanni Granucci and Maurizio Persico. Excited state dynamics with the direct
trajectory surface hopping method: azobenzene and its derivatives as a case study.
Theoretical Chemistry Accounts, 117(5-6):1131{1143, May 2007. ISSN 1432-881X.
55
[119] Wesley R. Browne and Ben L. Feringa. Light switching of molecules on surfaces.
Annual Review of Physical Chemistry, 60(6):407{428, 2009. ISSN 0066-426X. 55
[120] B. L. Feringa, R. A. van Delden, N. Koumura, and E. M. Geertsema. Chiroptical
molecular switches. Chemical Reviews, 100(5):1789{1816, May 2000. 55
[121] B. T. Houseman and M. Mrksich. The role of ligand density in the enzymatic glyco-
sylation of carbohydrates presented on self-assembled monolayers of alkanethiolates
on gold. Angewandte Chemie-international Edition, 38(6):782{785, 1999. 55
[122] D. J. Campbell, B. R. Herr, J. C. Hulteen, R. P. VanDuyne, and C. A. Mirkin. Ion-
gated electron transfer in self-assembled monolayer lms. Journal of the American
Chemical Society, 118(42):10211{10219, October 1996. 55
132
[123] D. G. Walter, D. J. Campbell, and C. A. Mirkin. Photon-gated electron transfer in
two-component self-assembled monolayers. Journal of Physical Chemistry B, 103
(3):402{405, January 1999. doi: 10.1021/jp983460b. 55
[124] Roger H. Terrill, Timothy A. Postlethwaite, Chun-hsien Chen, Chi-Duen Poon,
Andreas Terzis, Aidi Chen, James E. Hutchison, Michael R. Clark, and George and
Wignall. Monolayers in three dimensions: Nmr, saxs, thermal, and electron hop-
ping studies of alkanethiol stabilized gold clusters. Journal of the American Chem-
ical Society, 117(50):12537{12548, 1995. doi: 10.1021/ja00155a017. 55
[125] Z F Liu, K Hashimoto, and A Fujishima. Photoelectrochemical information-storage
using an azobenzene derivative. Nature, 347(6294):658{660, OCT 18 1990. ISSN
0028-0836. 56
[126] T Ikeda and O Tsutumi. Optical switching and image storage by means of azoben-
zene liquid-crystal lms. Science, 268(5219):1873{1875, JUN 30 1995. ISSN 0036-
8075. 56
[127] K Ichimura, SK Oh, and M Nakagawa. Light-driven motion of liquids on a pho-
toresponsive surface. Science, 288(5471):1624{1626, JUN 2 2000. ISSN 0036-8075.
56
[128] N Delorme, JF Bardeau, A Bulou, and F Poncin-Epaillard. Azobenzene-containing
monolayer with photoswitchable wettability. Langmuir, 21(26):12278{12282, DEC
20 2005. ISSN 0743-7463. 56
[129] K Ichimura, Suzuki Y, T Seki, A Hosoki, and K Aoki. Reversible change in align-
ment mode of nematic liquid-crystals regulated photochemically by command sur-
faces modied with an azobenzene monolayer. Langmuir, 4(5):1214{1216, SEP-
OCT 1988. ISSN 0743-7463. 56
[130] T. Ikeda, O. Tsutsumi, and T. Sasaki. Liquid crystal photonics: Optical switching
and image storage by means of nematic liquid crystals and ferroelectric liquid
crystals. Synthetic Metals, 81(2-3):289{296, August 1996. 56
[131] C Zhang, MH Du, HP Cheng, XG Zhang, AE Roitberg, and JL Krause. Coher-
ent electron transport through an azobenzene molecule: A light-driven molecular
switch. Physical Review Letters, 92(15):158301, APR 16 2004. ISSN 0031-9007.
doi: 10.1103/PhysRevLett.92.158301. 56
[132] Jerey M. Mativetsky, Giuseppina Pace, Mark Elbing, Maria A. Rampi, Marcel
Mayor, and Paolo Samori. Azobenzenes as light-controlled molecular electronic
switches in nanoscale metal-molecule-metal junctions. Journal of the American
Chemical Society, 130(29):9192+, JUL 23 2008. ISSN 0002-7863. doi: 10.1021/
ja8018093. 56
[133] Nathalie Katsonis, Monika Lubomska, Michael M. Pollard, Ben L. Feringa, and
Petra Rudolf. Synthetic light-activated molecular switches and motors on surfaces.
133
Progress In Surface Science, 82(7-8):407{434, 2007. doi: 10.1016/j.progsurf.2007.
03.011. 56
[134] M. Oh-e, Y. Tabe, and H. Yokoyama. Anisotropy of alkyl chains of azobenzene
molecules at the air-water interface observed by sum-frequency vibrational spec-
troscopy. Physical Review E, 68:061602, 2003. 56
[135] C. Ohe, M. Arai, H. Kamijo, M. Adachi, H. Miyazawa, K. Itoh, and T. Seki.
Interaction between a liquid crystalline molecule (4 `-pentyl-4-cyanobiphenyl)
and a poly(vinyl alcohol) derivative containing an azobenzene group at the air-
water interface: Sum frequency generation spectroscopic study. Journal of Phys-
ical Chemistry C, 112(16):6359{6365, APR 24 2008. ISSN 1932-7447. doi:
10.1021/jp076223u. 56
[136] C. Ohe, H. Kamijo, M. Arai, M. Adachi, H. Miyazawa, K. Itoh, and T. Seki.
Sum frequency generation spectroscopic study on photoinduced isomerization of
poly(vinyl alcohol) containing azobenzene side chain at the air-water interface.
Journal of Physical Chemistry C, 112(1):172{181, JAN 10 2008. ISSN 1932-7447.
doi: 10.1021/jp073602e. 56
[137] Steen Wagner, Felix Leyssner, Christian Koerdel, Sebastian Zarwell, Roland
Schmidt, Martin Weinelt, Karola Rueck-Braun, Martin Wolf, and Petra Tegeder.
Reversible photoisomerization of an azobenzene-functionalized self-assembled
monolayer probed by sum-frequency generation vibrational spectroscopy. Phys-
ical Chemistry Chemical Physics, 11(29):6242{6248, 2009. ISSN 1463-9076. doi:
10.1039/b823330f. 57
[138] H. Z. Yu, Y. Q. Wang, J. Z. Cheng, J. W. Zhao, S. M. Cai, H. Inokuchi,
A. Fujishima, and Z. F. Liu. Electrochemical behavior of azobenzene self-
assembled monolayers on gold. Langmuir, 12(11):2843{2848, May 1996. doi:
10.1021/la950632c. 57
[139] O. Wallach and L. Kiepenheuer. Ueber die umwandlung von azoxybenzol in oxya-
zobenzol. Berichte der deutschen chemischen Gesellschaft, 14(2):2617{2619, 1881.
ISSN 1099-0682. doi: 10.1002/cber.188101402222. 57
[140] K Tamada, J Nagasawa, F Nakanishi, K Abe, T Ishida, M Hara, and W Knoll.
Structure and growth of hexyl azobenzene thiol sams on au(111). Langmuir, 14
(12):3264{3271, JUN 9 1998. ISSN 0743-7463. 57
[141] J. Zhang, J. K. Whitesell, and M. A. Fox. Photoreactivity of self-assembled mono-
layers of azobenzene or stilbene derivatives capped on colloidal gold clusters. Chem-
istry of Materials, 13(7):2323{2331, July 2001. doi: 10.1021/cm000752s. 57
[142] P. Dietrich, F. Michalik, R. Schmidt, C. Gahl, G. Mao, M. Breusing, M. B. Raschke,
B. Priewisch, T. Elsaesser, R. Mendelsohn, M. Weinelt, and K. Rueck-Braun. An
anchoring strategy for photoswitchable biosensor technology: azobenzene-modied
134
sams on si(111). Applied Physics A-Materials Science & Processing, 93(2):285{292,
NOV 2008. ISSN 0947-8396. 58, 60
[143] B. Priewisch and K. Rueck-Braun. Novel photoswitchable amino acids. Journal of
Peptide Science, 10(Suppl. S):230, 2004. ISSN 1075-2617. 58, 60
[144] R. T. Pon, S. Y. Yu, and Y. S. Sanghvi. Rapid esterication of nucleosides to
solid-phase supports for oligonucleotide synthesis using uronium and phosphonium
coupling reagents. Bioconjugate Chemistry, 10(6):1051{1057, November 1999. doi:
10.1021/bc990063a. 58
[145] J. C. Love, L. A. Estro, J. K. Kriebel, R. G. Nuzzo, and G. M. Whitesides. Self-
assembled monolayers of thiolates on metals as a form of nanotechnology. Chemical
Reviews, 105(4):1103{1169, April 2005. doi: 10.1021/cr0300789. 60
[146] N. Nishi, D. Hobara, M. Yamamoto, and T. Kakiuchi. Chain-length-dependent
change in the structure of self-assembled monolayers of n-alkanethiols on au(111)
probed by broad-bandwidth sum frequency generation spectroscopy. Journal of
Chemical Physics, 118(4):1904{1911, January 2003. doi: 10.1063/1.1531098. 68
[147] Himali D. Jayathilake, Jeery A. Driscoll, Andrey N. Bordenyuk, Libo Wu, Sandro
R. P. da Rocha, Claudio N. Verani, and Alexander V. Benderskii. Molecular
order in langmuir-blodgett monolayers of metal-ligand surfactants probed by sum
frequency generation. Langmuir, 25(12):6880{6886, June 2009. doi: 10.1021/
la900168p. 68
[148] M Buck and M Himmelhaus. Vibrational spectroscopy of interfaces by infrared-
visible sum frequency generation. Journal of Vacuum Science & Technology A-
Vacuum Surfaces and Films, 19(6):2717{2736, NOV-DEC 2001. ISSN 0734-2101.
68
[149] E. Palik and G. Ghosh. Handbook of Optical Constants of Solids, Volume 3. Aca-
demic Press, New York, 1997. 76
[150] Yue Bing Zheng, John L Payton, Choong-Heui Chung, Rong Liu, Sarawut Che-
unkar, Bala Krishna Pathem, Yang Yang, Lasse Jensen, and Paul S Weiss. Surface-
enhanced raman spectroscopy to probe reversibly photoswitchable azobenzene in
controlled nanoscale environments. Nano letters, 11(8):3447{3452, August 2011.
79
[151] Ajeet S. Kumar, Tao Ye, Tomohide Takami, Byung-Chan Yu, Austen K. Flatt,
James M. Tour, and Paul S. Weiss. Reversible photo-switching of single azobenzene
molecules in controlled nanoscale environments. Nano Letters, 8(6):1644{1648,
JUN 2008. ISSN 1530-6984. doi: 10.1021/nl080323+. 79
[152] J M Nunzi. Organic photovoltaic materials and devices. Comptes Rendus Physique,
3:523542, 2002. 91
135
[153] W Shockley and H J Queisser. Detailed balance limit of eciency of p{n junction
solar cells. Journal of Applied Physics, 32:510{519, 1961. 91
[154] Petr P Khlyabich, Beate Burkhart, and Barry C Thompson. Compositional depen-
dence of the open-circuit voltage in ternary blend bulk heterojunction solar cells
based on two donor polymers. Journal of the American Chemical Society, 134(22):
9074{9077, June 2012. 91
[155] Letian Dou, Jing Gao, Eric Richard, Jingbi You, Chun-Chao Chen, Kitty C
Cha, Youjun He, Gang Li, and Yang Yang. Systematic investigation of
benzodithiophene- and diketopyrrolopyrrole-based low-bandgap polymers designed
for single junction and tandem polymer solar cells. Journal of the American Chem-
ical Society, 134(24):10071{10079, June 2012. 91
[156] Christoph Winder and Niyazi Serdar Sariciftci. Low bandgap polymers for photon
harvesting in bulk heterojunction solar cells. Journal of Materials Chemistry, 14
(7):1077, 2004. 91
[157] D P Gruber, G Meinhardt, and W Papousek. Spatial distribution of light absorp-
tion in organic photovoltaic devices. Solar energy, 79:697704, 2005. 91
[158] Matthias A Ruderer, Shuai Guo, Robert Meier, Hsin-Yin Chiang, Volker
K orstgens, Johannes Wiedersich, Jan Perlich, Stephan V Roth, and Peter M uller-
Buschbaum. Solvent-induced morphology in polymer-based systems for organic
photovoltaics. Advanced Functional Materials, 21(17):3382{3391, July 2011. 92
[159] M J Jaquith, J E Anthony, and J A Marohn. Long-lived charge traps in func-
tionalized pentacene and anthradithiophene studied by time-resolved electric force
microscopy. Journal of Materials Chemistry, 19:61166123, 2009. 93
[160] N R Armstrong, P A Veneman, and E Ratcli. Oxide contacts in organic photo-
voltaics: Characterization and control of near-surface composition in indium-tin
oxide (ito) electrodes. Accounts of chemical Researcb, 42:1748{1757, 2009. 93
[161] Diogenes Placencia, Weining Wang, R Clayton Shallcross, Kenneth W Nebesny,
Michael Brumbach, and Neal R Armstrong. Organic photovoltaic cells based on
solvent-annealed, textured titanyl phthalocyanine/c 60heterojunctions. Advanced
Functional Materials, 19(12):1913{1921, June 2009. 93
[162] X Y Zhu, Q Yang, and M Muntwiler. Charge-transfer excitons at organic semicon-
ductor surfaces and interfaces. Accounts of Chemical Research, 42(11):1779{1787,
November 2009. 93
[163] L W Barbour, R D Pensack, M Hegadorn, S Arzhantsev, and J B Asbury. Excita-
tion transport and charge separation in an organic photovoltaic material: Watching
excitations diuse to interfaces. The Journal of Physical Chemistry C, 112(10):
3926{3934, March 2008. 93
136
[164] David C Coey, Bryon W Larson, Alexander W Hains, James B Whitaker, Nikos
Kopidakis, Olga V Boltalina, Steven H Strauss, and Garry Rumbles. An optimal
driving force for converting excitons into free carriers in excitonic solar cells. The
Journal of Physical Chemistry C, 116(16):8916{8923, April 2012. 93
[165] Cuiying Yang, Jerry G Hu, and Alan J Heeger. Molecular structure and dynamics
at the interfaces within bulk heterojunction materials for solar cells. Journal of
the American Chemical Society, 128(36):12007{12013, September 2006. 93, 113
[166] Yosuke Kanai and Jerey C Grossman. Insights on interfacial charge transfer
across p3ht/fullerene photovoltaic heterojunction from ab initio calculations. Nano
letters, 7(7):1967{1972, July 2007. 93
[167] W Mou, S Ohmura, and F Shimojo. Molecular control of photoexcited charge
transfer and recombination at a quaterthiophene/zinc oxide interface. Applied
Physics Letters, 136:184705, 2012. 93
[168] W Mou, S Ohmura, S Hattori, and K Nomura. Enhanced charge transfer by phenyl
groups at a rubrene/C
60
interfaces. The Journal of Chemical Physics, 100:203306,
2012. 93
[169] B Burkhart, P P Khlyabich, and B C Thompson. In
uence of the acceptor composi-
tion on physical properties and solar cell performance in semi-random two-acceptor
copolymers. ACS Macro Letters, 1:660666, 2012. 94
[170] B Burkhart, P P Khlyabich, and B C Thompson. Solar cells based on semi-
random p3ht analogues containing dithienopyrrole: in
uence of incorporating a
strong donor. Journal of Photonics for Energy, 2:021002, 2012. 94
[171] P P Khlyabich, B Burkhart, and C F Ng. Ecient solar cells from semi-
random P3HT analogues incorporating diketopyrrolopyrrole. Macromolecules, 44:
50795084, 2011. 94
[172] Nicola Pfeer, Paul Raimond, Fabrice Charra, and Jean-Michel Nunzi. Determina-
tion of the two-photon absorption spectrum of a soluble polythiophene. Chemical
Physics Letters, 201(1-4):357{363, January 1993. 98, 100
[173] N Periasamy, R Danieli, G Ruani, R Zamboni, and C Taliani. Location of the
low-energy 1ag state in a polythiophene oligomer by two-photon absorption spec-
troscopy: alpha -sexithienyl. Physical Review Letters, 68(7):919{922, February
1992. 99
[174] D Beljonne and J L Br edas. Theoretical investigation of two-photon absorption
and third-harmonic generation in a long thiophene oligomer: -heptathiophene.
Journal of the Optical Society of America B, 11(8):1380{1387, 1994. 99
[175] S Palomba, H Harutyunyan, J Renger, R Quidant, N F van Hulst, and L Novotny.
Nonlinear plasmonics at planar metal surfaces. Philosophical Transactions of the
137
Royal Society A: Mathematical, Physical and Engineering Sciences, 369(1950):
3497{3509, August 2011. 107
[176] Harri Aarnio, Parisa Sehati, Slawomir Braun, Mathias Nyman, Michel P de Jong,
Mats Fahlman, and Ronald

Osterbacka. Spontaneous charge transfer and dipole
formation at the interface between p3ht and pcbm. Advanced Energy Materials, 1
(5):792{797, June 2011. 113
[177] Tao Liu and Alessandro Troisi. Absolute rate of charge separation and recombi-
nation in a molecular model of the p3ht/pcbm interface. The Journal of Physical
Chemistry C, 115(5):2406{2415, February 2011. 113
[178] C F N Marchiori and M Koehler. Dipole assisted exciton dissociation at conjugated
polymer/fullerene photovoltaic interfaces: A molecular study using density func-
tional theory calculations. Synthetic Metals, 160(7-8):643{650, April 2010. 113,
114
[179] Tao Liu, David L Cheung, and Alessandro Troisi. Structural variability and dynam-
ics of the p3ht/pcbm interface and its eects on the electronic structure and the
charge-transfer rates in solar cells. Physical Chemistry Chemical Physics, 13(48):
21461, 2011. 113
[180] Frank C Spano. Excitons in conjugated oligomer aggregates, lms, and crystals.
Annual Review of Physical Chemistry, 57(1):217{243, May 2006. 114
[181] G Weiser and S M oller. Directional dispersion of the optical resonance of - *
transitions of -sexithiophene single crystals. Physical Review B, 65(4):045203,
January 2002. 114
[182] William Barford, David G Lidzey, Dmitry V Makhov, and Anthony J H Meijer.
Exciton localization in disordered poly(3-hexylthiophene). The Journal of Chem-
ical Physics, 133(4):044504, 2010. 114
138
Appendix A
SFG Signal Fitting Programs
A.1 File Input: WinSpec) Mathematica
1 TableReadFiles[OutName , InputFileName , NPlots ] := f
2 TempData =
3 Transpose[
4 ReadList[
5 InputFileName <> ToString [1] <> ". txt", fReal , Real , Realg]];
6 Data = Table[0 , fi , 1, NPlotsg];
7 DataN = Table[0 , fi , 1, NPlotsg];
8 TempList = Table[0 , fi , 1, NPlotsg];
9 Do[
10 Temp =
11 Transpose[
12 ReadList[
13 InputFileName <> ToString[ i ] <> ". txt", fReal , Real , Realg]];
14 Temp[[2]] = (Temp[[1]]) (RedLaserCenter);
15 Data[[ i ]] = T[fTemp[[2]] , Temp[[3]]g];
16 DataN[[ i ]] = T[fTemp[[2]] , Rescale [Temp[[3]]]g];
17 , fi , 1, NPlotsg
18 ];
19 Data >>> "Data" <> OutName;
20 DataN >>> "DataN" <> OutName;
21 g
A.2 SFG CN Signal Fitting
1 Fitting2Gaus1Lor[RunInput ] := f
2 Gaussian1 =
3 amp1 1/Sqrt[2 n[Pi] sigma1^2] E^((x x1)^2/( 2sigma1^2));
4 Gaussian2 =
5 amp2 1/Sqrt[2 n[Pi] sigma2^2] E^((x x2)^2/(2 sigma2^2));
6
7 FitForm =
8 Abs[Gaussian1 + Gaussian2 + (
9 ampRE^(Iphase))/((x omegafreak) I Gammaq)]^2;
10
11 Infile = ToExpression["Data" <> RunInput];
12 DataExempt =
13 If[# < NRExemptNumbers[[1]] jj # > NRExemptNumbers[[2]] , 1, 0] &;
14 IncludedElements =
15 T[ Position [Map[DataExempt, T[ Infile ][[1]]] , 1]][[1]];
139
16 Fityyy =
17 NonlinearModelFit[ Infile ,
18 FitForm, ffamp1, NRInputs1[[1]]g , fsigma1 , NRInputs1[[2]]g , fx1,
19 NRInputs1[[3]]g , famp2, NRInputs2[[1]]g , fsigma2 ,
20 NRInputs2[[2]]g , fx2, NRInputs2[[3]]g , fampR,
21 RInputs [[1]]g , fphase , RInputs [[2]]g , fomegafreak ,
22 RInputs [[3]]g , fGammaq, RInputs[[4]]gg , x,
23 MaxIterations > 1000];
24 Unfit = ListPlot [ Infile , PlotRange > All , Frame> True,
25 FrameLabel > f"Wavenumber", "Intensity"g,
26 LabelStyle > Directive [Large , Bold] , AxesOrigin > f1900, 0g,
27 PlotStyle > Thick ];
28 Nowfit =
29 Plot[Evaluate [(Gaussian1 + Gaussian2)^2 /. Fityyy[[1 , 2]]] , fx,
30 1900, 2600g, PlotStyle > fThick , Blueg];
31 NowfitRes =
32 Plot[Evaluate[FitForm /. Fityyy[[1 , 2]]] , fx, 1900, 2600g,
33 PlotStyle > fThick , Redg];
34 NowfitRes1 =
35 Plot[Evaluate[
36 Abs[((ampRE^(Iphase))/((x omegafreak) I Gammaq))]^2 /.
37 Fityyy[[1 , 2]]] , fx, 1900, 2600g, PlotStyle > fGreeng,
38 PlotRange > All ];
39 (Print[Show[Unfit ,Nowfit ,NowfitRes ,NowfitRes1 ]];)
40
41 Fitf [x ] = FitForm /. Fityyy[[1 , 2]];
42 (Print[Fityyy["ParameterTable "]];)
43 g
A.3 Example Fitting Script
1 NPlots = 2000;
2 FTime = Table[0 , fi , 1, NPlotsg];
3 FTimeResAmp = Table[0 , fi , 1, NPlotsg];
4 FTimeNResAmp = Table[0 , fi , 1, NPlotsg];
5 FTimeResPhase = Table[0 , fi , 1, NPlotsg];
6 FTimeRatio = Table[0 , fi , 1, NPlotsg];
7
8 NRInputs1 = f609, 80, 1998g;
9 NRInputs2 = f21888, 197, 2197g;
10 RInputs = f67, 120, 2226, 8g;
11 NRExemptNumbers = f2280, 2150g;
12 RExemptNumbers = f2280, 2280, 2150g;
13
14 Timepoint = 0;
15
16 Do[
17 Timepoint += 1/6.;
18 DataInput = DataIso66fp [[ Index ]];
19
20 (FTime[[ Index]]=Fitting2GausLorVerbtest2["Input "];)
21
140
22 FTime[[ Index ]] = Fitting2GausLortest2 ["Input"];
23 FTimeResAmp[[ Index ]] = fTimepoint ,
24 Fityyy["BestFitParameters"][[7 , 2]]g;
25 FTimeResPhase[[ Index ]] = fTimepoint ,
26 Mod[Fityyy["BestFitParameters"][[8 , 2]] , 2 n[Pi ]]g;
27 FTimeNResAmp[[ Index ]] = fTimepoint ,
28 Fityyy["BestFitParameters"][[1 , 2]] +
29 Fityyy["BestFitParameters"][[4 , 2]]g;
30 FTimeRatio[[ Index ]] = fTimepoint ,
31 Fityyy["BestFitParameters"][[7 ,
32 2]]/(Fityyy["BestFitParameters"][[1 , 2]] +
33 Fityyy["BestFitParameters"][[4 , 2]])g;
34 PrintTemporary[Index ];
35 (Print[Fityyy["BestFitParameters "]])
36
37 , fIndex , 1, NPlotsg
38 ]
39
40 (FP)
41
42 DarkNums =
43 Join[Table[i , fi , 1, 95g], Table[i , fi , 160, 280g],
44 Table[i , fi , 345, 465g], Table[i , fi , 529, 649g],
45 Table[i , fi , 713, 834g], Table[i , fi , 898, 1019g],
46 Table[i , fi , 1082, 1202g], Table[i , fi , 1266, 1386g],
47 Table[i , fi , 1450, 1576g], Table[i , fi , 1639, 1758g],
48 Table[i , fi , 1822, 2000g]];
49
50 UVNums = Join[Table[i , fi , 96, 159g], Table[i , fi , 281, 344g],
51 Table[i , fi , 466, 528g], Table[i , fi , 650, 712g],
52 Table[i , fi , 835, 897g], Table[i , fi , 1020, 1081g],
53 Table[i , fi , 1203, 1265g], Table[i , fi , 1387, 1449g],
54 Table[i , fi , 1577, 1638g], Table[i , fi , 1759, 1821g]];
55
56 RRPDark =
57 ListPlot [FTimeResAmp[[DarkNums]] , PlotRange > All ,
58 PlotStyle > Black , Frame> True,
59 FrameLabel > f"Time (min)", "Intensity"g,
60 LabelStyle > Directive [Medium, Bold] ,
61 PlotLabel >
62 Style ["n[Chi]2 Res intensity of 66% Azo SAM", FontSize > 18] ,
63 Axes > f0, 100g];
64 RRPUV = ListPlot [FTimeResAmp[[UVNums]] , PlotRange > All ,
65 PlotStyle > Purple ];
66 Show[RRPDark, RRPUV]
A.4 Fitting for response function.
1 UVStarts = f96, 281, 466, 650, 835, 1020, 1203, 1387, 1577, 1795g;
2 UVOn = Table[0 , fi , 1, 10g];
3 Do[
4 UVStart = UVStarts [[ Index ]];
141
0 50 100 150 200 250 300
60
65
70
Time HminL
Intensity
c2 Res intensity of 66% Azo SAM
Figure A.1. Output of script showing t
res
.
5 UVOnLength = 60;
6 Points = Table[i , fi , UVStart 6, UVStart + UVOnLengthg];
7 UVOn[[ Index ]] = DataIso66fp [[ Points ]];
8 , fIndex , 1, 10g
9 ]
10 DataIsoMeanedUV = Mean[UVOn];
11
12 NPlots = 67;
13 FTime = Table[0 , fi , 1, NPlotsg];
14 FTimeResAmpMUV = Table[0 , fi , 1, NPlotsg];
15 FTimeNResAmpMUV = Table[0 , fi , 1, NPlotsg];
16 FTimeRatioMUV = Table[0 , fi , 1, NPlotsg];
17
18 NRInputs1 = f609, 80, 1998g;
19 NRInputs2 = f21888, 197, 2197g;
20 RInputs = f67, 120, 2226, 8g;
21 NRExemptNumbers = f2280, 2150g;
22 RExemptNumbers = f2280, 2280, 2150g;
23
24 Timepoint = 1;
25
26 Do[
27 Timepoint += 1/6.;
28 DataInput = DataIsoMeanedUV[[ Index ]];
29
30 (FTime[[ Index]]=Fitting2GausLorVerbtest2["Input "];)
31
32 FTime[[ Index ]] = Fitting2GausLortest2 ["Input"];
33 FTimeResAmpMUV[[ Index ]] = fTimepoint ,
142
34 Fityyy["BestFitParameters"][[7 , 2]]g;
35 FTimeNResAmpMUV[[ Index ]] = fTimepoint ,
36 Fityyy["BestFitParameters"][[1 , 2]] +
37 Fityyy["BestFitParameters"][[4 , 2]]g;
38 FTimeRatioMUV[[ Index ]] = fTimepoint ,
39 Fityyy["BestFitParameters"][[7 ,
40 2]]/(Fityyy["BestFitParameters"][[1 , 2]] +
41 Fityyy["BestFitParameters"][[4 , 2]])g;
42 PrintTemporary[Index ];
43
44 , fIndex , 1, NPlotsg
45 ]
143
Appendix B
SHG Signal Fitting Programs
B.1 Spike Removal
1
2 T = Transpose;
3
4 SpikeKill [Input , DS ] :=
5 Module[fInternalinput = Input , DeltaSpike = DSg,
6 SizeRule = If [Abs[#] > DeltaSpike , 1, 0] &;
7 SpikeElements =
8 Flatten [ Position [Map[SizeRule , Differences [T[Input ][[2]]] , 1] , 1]];
9 Internalinput [[
10 Complement[Table[i , fi , 1, Dimensions[ Internalinput ][[1]]g] ,
11 SpikeElements ]]]
12 ];
B.2 Gaussian Fitting
1 Gaussian = amp1 1/Sqrt[2 n[Pi] sigma1^2] E^((x x1)^2/( 2sigma1^2));
2
3 GaussianSHGFit[Input , Printqm ] := Module[fInternalInput = Inputg,
4 NumberToAverage = 5;
5
6 WidthParameters = famp1, .13 , 0, 1g;
7 PeakHeightParameters = fsigma1 ,
8 Max[T[ InternalInput ][[2]]] WidthParameters[[2]]2 , 0,
9 10Max[T[ InternalInput ][[2]]] WidthParameters[[2]]2g;
10 PeakPosParameters = fx1,
11 Mean[T[ InternalInput ][[1]][[
12 Ordering[T[ InternalInput ][[2]] , NumberToAverage]]]] ,
13 Min[T[ InternalInput ][[1]]] , Max[T[ InternalInput ][[1]]]g;
14
15 WidthConstraints = fg
16 Gaussian =
17 amp1 1/Sqrt[2 n[Pi] sigma1^2] E^((x x1)^2/( 2sigma1^2));
18
19 FitGausPeak =
20 NonlinearModelFit[ InternalInput ,
21 Gaussian , fWidthParameters[[f1 , 2g]] ,
22 PeakHeightParameters[[f1 , 2g]] , PeakPosParameters[[f1 , 2g]]g , x,
23 MaxIterations > 10000];
24
144
498 500 502 504 506 508
50
100
150
Figure B.1. Fitter Output if PrintQM is set to"True".
25 If [Printqm == True,
26 Print[Show[
27 ListPlot [ InternalInput , PlotRange > All , PlotStyle > Black] ,
28 Plot[PeakHeightParameters [[2]] 1/Sqrt[
29 2 n[Pi] WidthParameters [[2]]^2]
30 E^((x PeakPosParameters[[2]])^2/(
31 2WidthParameters [[2]]^2))
32 , fx, PeakPosParameters [[3]] , PeakPosParameters [[4]]g ,
33 PlotStyle > Blue , PlotRange > All ] ,
34 Plot[Gaussian /. FitGausPeak["BestFitParameters"] , fx,
35 PeakPosParameters [[3]] , PeakPosParameters [[4]]g ,
36 PlotRange > All , PlotStyle > fRed, Thickg]]];
37 ];
38 FitGausPeak["BestFitParameters"]
39 ];
B.3 Fitting Script
1 FitterA[Input , Width , Spikes , ForT ] :=
2 Module[fg,
3
4 FitAmps = Table[0 , fIndex , 1, Dimensions[Input ][[1]]g];
5 FitWidths = Table[0 , fIndex , 1, Dimensions[Input ][[1]]g];
6
7 Do[
8 InputFile = SpikeKill [Input [[ Index ]] , Spikes ];
9 TempOut = GaussianSHGFit[ InputFile , ForT];
10 FitAmps[[ Index ]] = TempOut[[f3 , 1g, 2]];
11 FitWidths [[ Index ]] = TempOut[[f3 , 2g, 2]];
12 PrintTemporary[Index ];
145
13 ,
14 fIndex , 1, Dimensions[Input ][[1]]g
15 ];
16 FitAmps
17 ];
146
Appendix C
XFROG Mathematica Code
C.1 XFROG Programs
1 (Run these first)
2 T = Transpose;
3 << PlotLegends `
4
5 (Finds you the approporate time from an energy)
6 OmegaAxisFindFlip[ input ] :=
7 RadpFStoWN[(2 n[Pi]
8 If [input <
9 NumberPoints/2, (input 1), (NumberPoints input +
10 1)])/(NumberPointsTimeResolution )];
11
12 (Finds you the approporately united wavenumber from a t space)
13 OmegaAxisFind[ input ] :=
14 RadpFStoWN[(2 n[Pi](input 1))/(NumberPointsTimeResolution )];
15
16 ReverseOmegaAxisFind[
17 Input ] := fRound[
18 1/(2 n[Pi]) WNtoRadpFS[Input]NumberPointsTimeResolution + 1] ,
19 NumberPoints
20 Round[1/(2 n[Pi]) WNtoRadpFS[Input]NumberPointsTimeResolution +
21 1]g;
22
23 (Converts Wavenumber to Radiant per fs)
24 WNtoRadpFS[Wavenumber ] := Wavenumber2.997810^102 n[Pi]10^(15);
25 (The reverse operation)
26
27 RadpFStoWN[RadpFS ] := RadpFS/(2.997810^102 n[Pi]10^(15))
28 (Converts FWHM to )
29 FWHMtoTau[FWHMfs ] := FWHMfs/Sqrt[2Log[2]];
30 (Converts microns delay to femtoseconds , it is mutliped by a factor n
31 of 2 is because of the retroreflector)
32
33 UMtoFs[uMInput ] := uMInput(10/3)2;
34 (Generates a sample pulse .)
35
36 WaveShape[Amp , OmegaIR , TShift , TauIR ] =
37 AmpExp[IOmegaIR(t TShift)]Exp[(t TShift)^2/TauIR^2];
38 WaveShapeChirp[Amp , OmegaIR , TShift , TauIR , Twist ] =
39 AmpExp[I(OmegaIR + (t TShift)Twist)(t TShift)]
40 Exp[(t TShift)^2/TauIR^2];
41 (Generates a dummy matrix to fit data onto)
147
42
43 DensityPlotPrep[Input ] :=
44 Table[Input [[ i , j , 2]] , fi , 1, Dimensions[Input ][[1]]g , fj , 1,
45 Dimensions[Input ][[2]]g]
46 (Adds gaussian noise should you want that)
47
48 NoiseMatrix[ Size1 , Size2 , NoiseAmp ] :=
49 Table[NoiseAmpRandom[NormalDistribution []] , fi , 1, Size1g, fj , 1,
50 Size2g];
51 (Takes forurer by rows)
52
53 FRows[ Input ] :=
54 Module[fInputINT = Inputg,
55 FInput = Table[0 , fi , 1, NumberPointsg];
56 Do[
57 FInput [[ Index ]] = Fourier [InputINT[[ Index ]]];
58 , fIndex , 1, NumberPointsg
59 ];
60 FInput
61 ]
62 (Takes inverse forurer by rows)
63
64 IFRows[ Input ] :=
65 Module[fInputINT = Inputg,
66 FInput = Table[0 , fi , 1, NumberPointsg];
67 Do[
68 FInput [[ Index ]] = InverseFourier [InputINT[[ Index ]]];
69 , fIndex , 1, NumberPointsg
70 ];
71 FInput
72 ]
73 (Fourier on transposition)
74
75 Fourieromegatot[ Input ] :=
76 Module[fInputINT = Inputg,
77 FInput = Table[0 , fi , 1, NumberPointsg];
78 Do[
79 FInput [[ Index ]] = InverseFourier [T[InputINT ][[ Index ]]];
80 , fIndex , 1, NumberPointsg
81 ];
82 T[FInput]
83 ]
84 (Inverse Fourier on transposition)
85
86 Fourierttoomega[ Input ] :=
87 Module[fInputINT = Inputg,
88 FInput = Table[0 , fi , 1, NumberPointsg];
89 Do[
90 FInput [[ Index ]] = Fourier [T[InputINT ][[ Index ]]];
91 , fIndex , 1, NumberPointsg
92 ];
93 T[FInput]
94 ]
95 (Fourier by rows)
148
96
97 FourierOMEGAtoTau[ Input ] :=
98 Module[fInputINT = Inputg,
99 FInput = Table[0 , fi , 1, NumberPointsg];
100 Do[
101 FInput [[ Index ]] = InverseFourier [InputINT[[ Index ]]];
102 , fIndex , 1, NumberPointsg
103 ];
104 FInput
105 ]
106 (Inverse Fourier by rows)
107
108 FourierTautoOMEGA[ Input ] :=
109 Module[fInputINT = Inputg,
110 FInput = Table[0 , fi , 1, NumberPointsg];
111 Do[
112 FInput [[ Index ]] = Fourier [InputINT[[ Index ]]];
113 , fIndex , 1, NumberPointsg
114 ];
115 FInput
116 ]
117 (Sets up the matrix sizes to use)
118 PrepThangs :=
119 Module[fg, Clear[TimeWidth];
120 StartTime =TimeWidth;
121 EndTime = TimeWidth;
122 TimeResolution = (2TimeWidth)/NumberPoints;
123 Colorblock =
124 Table[ColorData["Rainbow", i/NumberPoints] , fi , 1, NumberPointsg];
125 ]
126 (Picks up experimental data and puts on units)
127
128 GetData[ RootDirectoryInt , FileUsedInt , TauSizeInt ] := Module[fg,
129 SetDirectory [ RootDirectoryInt ];
130 ExpRaw = T[DensityPlotPrep[ReadList[ FileUsedInt ]]];
131 WavelengthAxis = 10000000/T[ReadList[ FileUsedInt ][[1]]][[1]];
132 ExpRaw +=
133 NoiseMatrix[Dimensions[ExpRaw][[1]] , Dimensions[ExpRaw][[2]] ,
134 Max[ExpRaw]NoiseLevel ];
135 MinTauExp =TauSizeInt;
136 MaxTauExp = TauSizeInt;
137 ]
138 GetDataReal[ RootDirectoryInt , FileUsedInt , WavelengthAxisint ,
139 TauSizeInt ] := Module[fg,
140 SetDirectory [ RootDirectoryInt ];
141 ExpRaw = T[Import[FileUsedInt , "Table" ]];
142 WavelengthAxis = T[Import[WavelengthAxisint , "Table" ]][[1]];
143 (ExpRaw+=NoiseMatrix[Dimensions[ExpRaw][[1]] , Dimensions[ExpRaw][[
144 2]] ,Max[ExpRaw]NoiseLevel ];)
145 MinTauExp =TauSizeInt;
146 MaxTauExp = TauSizeInt;
147 ]
148 GetDataRealStitch[ RootDirectoryInt , FileUsedInt , WavelengthAxisint ,
149 TauSizeInt ] := Module[fg,
149
150 SetDirectory [ RootDirectoryInt ];
151 ExpRaw = Import[FileUsedInt , "Table"][[1 , 2]];
152 WavelengthAxis = Import[WavelengthAxisint , "list"][[1 , 2]];
153 (ExpRaw+=NoiseMatrix[Dimensions[ExpRaw][[1]] , Dimensions[ExpRaw][[
154 2]] ,Max[ExpRaw]NoiseLevel ];)
155 MinTauExp =TauSizeInt + TauSmudge;
156 MaxTauExp = TauSizeInt + TauSmudge;
157 ]
158 (Sets up more of the matrix sizings)
159 ExpDataPrep := Module[fg,
160 MaxOmegaExp =
161 10000000/WavelengthAxis [[ Dimensions[WavelengthAxis ][[1]]]];
162 MinOmegaExp = 10000000/WavelengthAxis [[1]];
163 MeanOmega = ((MaxOmegaExp + MinOmegaExp)/2) + OmegaSmudge;
164 MaxOmega = MeanOmega + BandWidth;
165 MinOmega = MeanOmega BandWidth;
166 FrequencyCompression = MaxOmegaExp/(2BandWidth);
167 MaxLamda = WavelengthAxis [[ Dimensions[WavelengthAxis ][[1]]]];
168 MinLamda = WavelengthAxis [[1]];
169 WavenumberGrid =
170 Table[Wavenumbers, fWavenumbers, MinOmega,
171 MaxOmega, (MaxOmega MinOmega)/(NumberPoints 1)g];
172 BackSolvedTime =
173 Solve[(2BandWidth) == OmegaAxisFind[NumberPoints/2] , TimeWidth];
174 TimeWidth = (TimeWidth /. BackSolvedTime [[1]]);
175 StartTime =TimeWidth;
176 EndTime = TimeWidth;
177 TimeResolution = (2TimeWidth)/NumberPoints;
178 MinTau =TimeWidth;
179 MaxTau = TimeWidth;
180 Taus = Table[
181 i , fi , MinTau, MaxTau, (MaxTau MinTau)/(NumberPoints 1)g];
182 OmegaAxis = Table[OmegaAxisFind[ i ] , fi , NumberPointsg];
183 ExpermentalData =
184 Table[DetectorNoise , fi , 1, NumberPointsg, fj , 1,
185 Dimensions[OmegaAxis ][[1]]g];
186 ]
187 (Interpolates raw data so we can impose our new set of points onto n
188 it . Points outside of experimental range are set to constant)
189
190 FROGInterpolation[ExpRawint ] := Module[fg,
191 FExp = ListInterpolation [
192 T[ExpRawint] , ffMinLamda, MaxLamdag, fMinTauExp, MaxTauExpgg];
193 PointsRequested =
194 Table[fOmega, Taug, fOmega, MinOmega,
195 MaxOmega, (MaxOmega MinOmega)/(NumberPoints 1)g, fTau,
196 MinTau, MaxTau, (MaxTau MinTau)/(NumberPoints 1)g];
197 AllOmegas = T[T[PointsRequested ][[1]]][[1]];
198 AllTaus = T[PointsRequested [[1]]][[2]];
199 IfInterpOmega = If[# > MinOmegaExp && # < MaxOmegaExp, 1, 0] &;
200 ElementsOmega = T[ Position [Map[IfInterpOmega , AllOmegas] , 1]][[1]];
201 IfInterpTau = If[# > MinTauExp && # < MaxTauExp, 1, 0] &;
202 ElementsTau = T[ Position [Map[IfInterpTau , AllTaus] , 1]][[1]];
203
150
204 T[Table[
205 If [MemberQ[ElementsTau , Taus] &&
206 MemberQ[ElementsOmega, Wavenumbers] ,
207 FExp[10000000/PointsRequested [[Wavenumbers, Taus, 1]] ,
208 PointsRequested [[Wavenumbers, Taus, 2]]] ,
209 DetectorNoise
210 ]
211 , fTaus, 1, NumberPointsg, fWavenumbers, 1, NumberPointsg]]
212 ]
213 (First transformation , gets to J(n[Tau] ,t))
214
215 FrogFourier[ExpDataInterpint ] := Module[fg,
216 ExpDataInterpint2 = RotateRight[ExpDataInterpint , NumberPoints/2];
217 FIfrog = Fourieromegatot[ExpDataInterpint2 ];
218 FIfrog = RotateRight[FIfrog , NumberPoints/2];
219 FourierTautoOMEGA[ FIfrog ]
220 ]
221 (Makes a unchirped visible pulse for the deconvolution)
222
223 VisGenerate[TauVisint , VisAmpTermint ] := Module[fg,
224 VisAmpTerm = 1;
225 OmegaVis = 0;
226 WaveVisRoh[Tinputs ] :=
227 VisAmpTermExp[IOmegaVisTinputs [[1]]]
228 Exp[Tinputs [[1]]^2/TauVis^2]
229 Conjugate[
230 Exp[IOmegaVisTinputs [[2]]]Exp[Tinputs [[2]]^2/TauVis^2]];
231 TimePoints =
232 Table[fOmega, Omega tg, ft , StartTime ,
233 EndTime, (EndTime StartTime)/(NumberPoints 1)g, fOmega,
234 MinTau, MaxTau, (MaxTau MinTau)/(NumberPoints 1)g];
235 Gvis = Table[
236 WaveVisRoh[TimePoints [[ i , j ]]] , fi , 1, NumberPointsg, fj , 1,
237 NumberPointsg];
238 FourierTautoOMEGA[Gvis]
239 ]
240 (Deconvolves the Vis and IR, The cutoff is really the important
241 varable . This keeps from dividing the IR by supersmall numbers on
242 what should be a convergent function)
243
244 Deconvolve[Froginp , Visinp , Cutoff ] := Module[fg,
245 HowBad = 0;
246 CutoffNorm = Max[Abs[Visinp ]] Cutoff;
247 DeCon = Table[
248 If [Abs[Visinp [[ i , j ]]] > CutoffNorm ,
249 Froginp [[ i , j ]]/ Visinp [[ i , j ]] ,
250 HowBad++;
251 0]
252 , fi , 1, NumberPointsg, fj , 1, NumberPointsg];
253
254 DeConShift = DeCon;
255 Shift = TimeWidth;
256 Do[DeConShift [[ i ]] = DeCon[[ i ]]Exp[ IShiftTaus] , fi , 1,
257 NumberPointsg];
151
258 FDeCon = FourierOMEGAtoTau[DeCon];
259 T[RotateRight[T[FDeCon] , NumberPoints/2]]
260 ]
261 (Makes our density matrix , has to expand the size of the matrix for
262 this)
263 DensityMatGen[FDeConInp ] := Module[fg,
264 PadSize = NumberPoints/2;
265 FDeConPad = ArrayPad[FDeConInp, ffPadSize , PadSizeg, f0, 0gg, 0];
266 RohDeConTimes =
267 Table[FDeConPad[[( i j), i ]] , fj , 1 2 PadSize ,
268 NumberPoints 2 PadSizeg, fi , 1, NumberPointsg];
269 N[RohDeConTimes]
270 ]
271 (does some chopping and organizing to organize and reunit data for
272 graphing)
273 GraphPrep[NumberToGraphInt ] := Module[fg,
274 Colorblock =
275 Table[ColorData["Rainbow", i/NumberToGraph] , fi , 1, NumberToGraphg];
276 TimePoints1D =
277 Table[t , ft , 2StartTime ,
278 2EndTime, (2EndTime 2StartTime)/(NumberPoints 1)g];
279 TimePoints1D =
280 Table[t , ft , StartTime ,
281 EndTime, (EndTime StartTime)/(NumberPoints 1)g];
282 a = Table[0 , fi , 1, NumberToGraphg];
283 b = Table[0 , fi , 1, NumberToGraphg];
284 c = Table[0 , fi , 1, NumberToGraphg];
285 d = Table[0 , fi , 1, NumberToGraphg];
286 e = Table[0 , fi , 1, NumberToGraphg];
287 Do[
288 a[[ i ]] =
289 ListPlot [T[fTimePoints1D , (Abs[Eigens [[2 , i ]]])^2g] ,
290 Joined > True, PlotRange > All , PlotStyle > Colorblock [[ i ]]];
291 b[[ i ]] =
292 ListPlot [T[fTimePoints1D , (Re[Eigens [[2 , i ]]])g] , Joined > True,
293 PlotRange > All , PlotStyle > Colorblock [[ i ]]];
294 c [[ i ]] =
295 ListPlot [T[fTimePoints1D , (Im[Eigens [[2 , i ]]])g] , Joined > True,
296 PlotRange > All , PlotStyle > Colorblock [[ i ]]];
297 d[[ i ]] =
298 ListPlot [(Abs[ Fourier [Eigens [[2 , i ]]]])^2 , Joined > True,
299 PlotRange > All , PlotStyle > Colorblock [[ i ]]];
300 e [[ i ]] =
301 ListPlot [(Abs[ Fourier [Eigens [[2 , i ]]]])^2 + .06i ,
302 Joined > True, PlotRange > All , PlotStyle > Colorblock [[ i ]]];
303 , fi , 1, NumberToGraphIntg
304 ];
305 Carrier = WNtoRadpFS[0];
306 CarrierW = Exp[I (CarrierTaus)];
307 PTGW = 20;
308 Ifgraph = If[# >GraphRangefs && # < GraphRangefs , 1, 0] &;
309 PointsToGraph = T[ Position [Map[Ifgraph , TimePoints1D] , 1]][[1]];
310 FinTime =
311 ListPlot [
152
312 T[fTimePoints1D , (Abs[Eigens [[2 , 1]]])^2g][[ PointsToGraph]] ,
313 Joined > True, PlotRange > All , PlotStyle > fRed, Thickg];
314 PinTime =
315 ListPlot [
316 T[fTimePoints1D , (Arg[Eigens [[2 , 1]]CarrierW])/20g][[
317 PointsToGraph]] , Joined > True, PlotRange > All ,
318 PlotStyle > fBlue , Thickg];
319 ]
C.2 XFROG Queues
1 (The queue that runs the program)
2 DeconvolutionRun := Module[fg,
3 PrepThangs;
4 (GetData[RootDirectory ,FileUsed ,TauSize ];)
5
6 GetDataRealStitch[RootDirectory , FileUsed , WavelengthAxisfile ,
7 TauSize ];
8 ExpDataPrep;
9 (Interpolation)
10 ExpDataInterp = FROGInterpolation[ExpRaw];
11 (Fourier of Spectra)
12 FFIfrog = FrogFourier[ExpDataInterp ];
13 (Vis Treatment)
14 FGvis = VisGenerate[TauVis, 1];
15 (Combine, Reverse FFT)
16
17 FDeCon = Deconvolve[FFIfrog , FGvis, DeconvolveCutoff ];
18 (Turn Back into Density)
19
20 RohDeConTimes = DensityMatGen[FDeCon];
21 (Take EigenValues , Plot EigenVectors)
22
23 Eigens = Eigensystem[RohDeConTimes];
24 GraphPrep[NumberToGraph];
25 ];
26 (The procedural queue for data from the fake data program)
27
28 DeconvolutionRunFakeData := Module[fg,
29 PrepThangs;
30 (GetData[RootDirectory ,FileUsed ,TauSize ];)
31
32 GetDataRealStitch[RootDirectory , FileUsed , WavelengthAxisfile ,
33 TauSize ];
34 ExpDataPrep;
35 (Interpolation)
36 ExpDataInterp = FROGInterpolation[ExpRaw];
37 ExpDataInterp = Abs[FakeData];
38 (Fourier of Spectra)
39 FFIfrog = FrogFourier[ExpDataInterp ];
40 (Vis Treatment)
41 FGvis = VisGenerate[TauVis, 1];
153
42 (Combine, Reverse FFT)
43
44 FDeCon = Deconvolve[FFIfrog , FGvis, DeconvolveCutoff ];
45 (Turn Back into Density)
46
47 RohDeConTimes = DensityMatGen[FDeCon];
48 (Take EigenValues , Plot EigenVectors)
49
50 Eigens = Eigensystem[Reverse[RohDeConTimes]];
51 GraphPrep[NumberToGraph];
52 ];
C.3 input Files and Variables
1 (Start)
2 NumberPoints = 512; (Square matrix side of simulation)
3
4 BandWidth = 2000;(bandwidth of simulation in wavenumber)
5
6 NoiseLevel = 0;()
7
8 VisNoiseLevel =
9 0;(can add noise to vis data)
10 (sets the cutoff of deconvolution function)
11 DeconvolveCutoff = .04;
12 TauVis = 51;(width of fourier limited vis pulsenu)
13 DetectorNoise = 0;(value to apply to field around exp data)
14
15 (RootDirectory="";
16 FileUsed="TestFrog2. txt";)
17
18 (Directory in which to find experimental data)
19 RootDirectory = "";
20
21
22 WavelengthAxisfile = "20110217 xfrog COMBINED. txt";
23 FileUsed = "20110217 xfrog with ir wp COMBINED. txt";
24 FileUsed = "FDnoChirpLarge600fs. txt";
25 WavelengthAxisfile = "FDnoChirpLargeXAxis. txt";
26 TauSize = 300;
27 TauSmudgeFake = 0;
28 OmegaSmudgeFake = 0;
29
30 (WavelengthAxisfile="20110217 xfrog COMBINED. txt";)
31 (FileUsed="20110217 xfrog ge plate COMBINED. txt";)
32 (FileUsed="20110217 xfrog with ir wp COMBINED. txt";)
33
34 FileUsed = "20110217 xfrog 6mm caf2 and ir wp COMBINED. txt";
35
36 TauSize = 832.5;(The width of the simulation in fs)
37
38 (Applies an offset to set "zero delay" to approximatly tau=0)
154
39 NumberToGraph = 1;(How many of the eigenvalues to print)
40 GraphRangefs = 200;(How many fs to graph)
41 TauSmudge = 36;
42 OmegaSmudge = 0;
43
44 (Just a big queue that runs all the programs)
45 DeconvolutionRun
46 (Queue for fake data)
47 (DeconvolutionRunFakeData)
C.4 Graphing Programs
1 (Big Figure Plot List)
2
3 (This makes a list of the plots that visualize the workflow that
4 goes on in the algorithm)
5
6 (It 's really hard to get a good mathematica color scheme to
7 visualize these but this is probably about the best for something
8 that works both color and B/W without being too ugly)
9
10 ColorTheme = ColorData[f"CMYKColors", "Reverse"g];
11
12 PlotStyleOptions = fPlotRange > All , InterpolationOrder > 0,
13 ColorFunction > ColorTheme,
14 LabelStyle > fDirective [Medium, Bold] , FontFamily > "Helvetica"gg;
15
16 Rot[ InputPoints ] := T[RotateRight[T[InputPoints] , NumberPoints/2]];
17
18 (How many fs to graph)
19 GraphSize = 140;
20
21 PointstoGraph =
22 Table[fi , jg, fi , GraphSize , NumberPoints GraphSizeg, fj ,
23 GraphSize , NumberPoints GraphSizeg];
24 SubsetGraph[ InputFile ] =
25 Table[ InputFile [[ i , j ]] , fi , GraphSize ,
26 NumberPoints GraphSizeg, fj , GraphSize ,
27 NumberPoints GraphSizeg];
28
29 TimePointsUsed =
30 Table[t , ft , StartTime ,
31 EndTime, (EndTime StartTime)/(NumberPoints 1)g];
32 OmegaPointsUsed =
33 Table[w, fw, BandWidth,
34 BandWidth, (2BandWidth)/(NumberPoints 1)g];
35 TauSubsetMin = TimePointsUsed [[ GraphSize ]];
36 TauSubsetMax = TimePointsUsed [[ NumberPoints GraphSize ]];
37 OmegaSubsetMin = OmegaPointsUsed[[ GraphSize ]];;
38 OmegaSubsetMax = OmegaPointsUsed[[ NumberPoints GraphSize ]];;
39
40 ListDensityPlot [ExpRaw,
155
41 DataRange > ffMinLamda, MaxLamdag, fMinTauExp,
42 MaxTauExpgg, PlotStyleOptions ,
43 PlotLabel > Style ["XFROG Input Spectra", FontSize > 22]]
44
45 ListDensityPlot [SubsetGraph[ExpDataInterp] ,
46 DataRange > ffTauSubsetMin, TauSubsetMaxg, fOmegaSubsetMax,
47 OmegaSubsetMingg, PlotStyleOptions ,
48 PlotLabel > Style ["XFROG(n[Tau] ,n[Omega])", FontSize > 22]]
49
50 ListDensityPlot [SubsetGraph[Abs[ FIfrog ]] ,
51 DataRange > ffTauSubsetMin, TauSubsetMaxg, fTauSubsetMin,
52 TauSubsetMaxgg, PlotStyleOptions ,
53 PlotLabel > Style ["J(n[Tau] ,t)", FontSize > 22]]
54
55 ListDensityPlot [
56 SubsetGraph[T[RotateRight[T[Abs[FFIfrog ]] , NumberPoints/2]]] ,
57 DataRange > ffTauSubsetMin, TauSubsetMaxg, fOmegaSubsetMin,
58 OmegaSubsetMaxgg,
59 FrameLabel > f"n[CapitalOmega]", "t"g, PlotStyleOptions ,
60 PlotLabel >
61 Style ["n!n(nOverscriptBox[n(Jn), n(~n)]n)(n[CapitalOmega] ,t)",
62 FontSize > 22]]
63
64 ListDensityPlot [SubsetGraph[Abs[Gvis]] ,
65 DataRange > ffTauSubsetMin, TauSubsetMaxg, fTauSubsetMin,
66 TauSubsetMaxgg, FrameLabel > f"n[Tau]", "t"g, PlotStyleOptions ,
67 PlotLabel > Style ["g(n[Tau] ,t)", FontSize > 22]]
68
69 ListDensityPlot [
70 SubsetGraph[T[RotateRight[T[Abs[FGvis]] , NumberPoints/2]]] ,
71 DataRange > ffOmegaSubsetMin, OmegaSubsetMaxg, fTauSubsetMin,
72 TauSubsetMaxgg,
73 FrameLabel > f"n[CapitalOmega]", "t"g, PlotStyleOptions ,
74 PlotLabel >
75 Style ["n!n(nOverscriptBox[n(gn), n(~n)]n)(n[CapitalOmega] ,t)",
76 FontSize > 22]]
77
78 ListDensityPlot [
79 SubsetGraph[T[RotateRight[T[Abs[DeCon]] , NumberPoints/2]]] ,
80 DataRange > ffOmegaSubsetMin, OmegaSubsetMaxg, fTauSubsetMin,
81 TauSubsetMaxgg,
82 FrameLabel > f"n[CapitalOmega]", "t"g, PlotStyleOptions ,
83 PlotLabel >
84 Style ["n!n(nOverscriptBox[n( fn), n(~n)]n)(n[CapitalOmega] ,t)",
85 FontSize > 22]]
86
87 ListDensityPlot [SubsetGraph[Abs[FDeCon]] ,
88 DataRange > ffTauSubsetMin, TauSubsetMaxg, fTauSubsetMin,
89 TauSubsetMaxgg, FrameLabel > f"n[Tau]", "t"g, PlotStyleOptions ,
90 PlotLabel > Style ["f (n[Tau] ,t)", FontSize > 22]]
91
92 ListDensityPlot [SubsetGraph[Abs[RohDeConTimes]] ,
93 DataRange > ffTauSubsetMin, TauSubsetMaxg, fTauSubsetMin,
94 TauSubsetMaxgg, FrameLabel > f"t '", "t"g, PlotStyleOptions ,
156
95 PlotLabel > Style ["n[Rho](t ' ,t)", FontSize > 22]]
96
97 Renorm = Max[(Abs[Eigens [[2 , 1]]])^2];
98 ArgRenorm[ arg ] := (arg/(n[Pi ]));
99 Eigs = T[fTimePoints1D , ((Abs[Eigens [[2 , 1]]])^2)/Renormg][[
100 PointsToGraph ]];
101 Args = T[fTimePoints1D , ArgRenorm[(Arg[Eigens [[2 , 1]]])]g][[
102 PointsToGraph ]];
103
104 ListPlot[fEigs , Argsg, Joined > True, PlotRange > All ,
105 PlotStyle > ffThick , Blueg, fThick , Redgg, Frame> True,
106 FrameLabel > ff"AU", "n[Pi] Rads"g, f"fs", Nonegg,
107 LabelStyle > fDirective [Medium, Bold] , FontFamily > "Helvetica"g,
108 FrameTicks > ffAutomatic ,
109 Join[ Flatten [Table[i , fi , 1, 1, .25g]]]g , fAutomatic ,
110 Automaticgg
111 ]
112
113 (Fitting Chirp)
114
115 (This fits the phase of the chirp to a third order polynomial. The n
116 first order chirp is mainly determined by your choice of tau=0. The n
117 second (C2) and third (C3) order chirps are actually interesting .)
118
119 ArgFit = NonlinearModelFit[Args ,
120 C0 + C1 x + C2 x^2 + C3 x^3, fC0, C1, C2, C3g, x,
121 Weights > T[Eigs ][[2]]];
122 Normal[ArgFit]
123 Show[ ListPlot [Args] , Plot[ArgFit[x] , fx, 100, 100g]]
124 Show[ ListPlot[fEigs , Argsg, Joined > True, PlotRange > All ,
125 PlotStyle > ffThick , Blueg, fThick , Redgg, Frame> True,
126 FrameLabel > ff"AU", "n[Pi] Rads"g, f"fs", Nonegg,
127 LabelStyle > fDirective [Medium, Bold] , FontFamily > "Helvetica"g,
128 FrameTicks > ffAutomatic ,
129 Join[ Flatten [Table[i , fi , 1, 1, .25g]]]g , fAutomatic ,
130 Automaticgg], Plot[ArgFit[x] , fx, 80, 80g]]
157
158
159
Figure C.1. Amplitude and argument of deconvolved IR pulse with t
for second order and third order chirp
160
Appendix D
SFG Orientational Analysis
D.1 Initialization and Indices
1 Needs["PlotLegends `"]
2 Needs["ErrorBarPlots `"]
3 T = Transpose;
4 SetDirectory [""];
5 Indices = ReadList["AU Palik. txt", fReal , Real , Realg];
6 nau[wl ] :=
7 Piecewise[ff0.162 + 3.5530 I , wl == 675g, f0.181 + 5.1178 I , wl == 800g, f2.66 + 31.55 I , wl == 4500gg];
8 (index of refraction Au at different wavelengths
9 Values from LuxPop)
10 nair = 1.; (index of refraction of air)
11
12 beta[wl ] :=
13 Piecewise[ff63 Degree , wl == 675g, f63 Degree , wl == 800g, f66 Degree , wl == 4500gg]
14 (angle of incidence of electric fields in degrees)
D.2 Fresnel Factors
1 gamma[wl ] :=
2 ArcSin[( nair Sin[beta[wl]])/nau[wl ]]; (angle of refraction)
3
4 Lxx[wl ] := (2. nair Cos[gamma[wl]])/(
5 nairCos[gamma[wl ]] + nau[wl] Cos[beta[wl ]]);
6 Lyy[wl ] := (2. nair Cos[beta[wl]])/(
7 nairCos[beta[wl ]] + nau[wl] Cos[gamma[wl ]]);
8 Lzz[wl , nL ] := (2. nau[wl] Cos[beta[wl]])/(
9 nairCos[gamma[wl ]] + nau[wl] Cos[beta[wl]])( nair/nL)^2;
D.3
(2)
elements for nitrile modes
1 Chixxz[Th , r ] := 0.5 (Cos[Th] (1 + r) (Cos[Th])^3 (1 r ));
2 Chiyyz[Th , r ] := Chixxz[Th, r ];
3 Chixzx[Th , r ] := (Cos[Th] (Cos[Th])^3 ) (1 r )/2.;
4 Chiyzy[Th , r ] := Chixzx[Th, r ];
5 Chizxx[Th , r ] := Chixzx[Th, r ];
6 Chizyy[Th , r ] := Chixzx[Th, r ];
7 Chizzz[Th , r ] := r Cos[Th] + (Cos[Th])^3 (1 r );
8
9 Xssp[Th , r , nL ] :=
161
10 Lyy[675] Lyy[800] Lzz[4500, nL] Sin[beta[4500]] Chiyyz[Th, r ];
11
12 Xsps[Th , r , nL ] :=
13 Lyy[675] Lzz[800, nL] Lyy[4500] Sin[beta [800]] Chiyzy[Th, r ];
14
15 Xppp[Th , r , nL ] :=
16 Lxx[675] Lxx[800] Lzz[4500, nL]
17 Cos[beta [675]] Cos[beta [800]] Sin[beta[4500]] Chixxz[Th, r]
18 (Lxx[675] Lzz[800, nL] Lxx[4500]
19 Cos[beta [675]] Sin[beta [800]] Cos[beta[4500]] Chixzx[Th, r ])
20 + Lzz[675, nL] Lxx[800] Lxx[4500]
21 Sin[beta [675]] Cos[beta [800]] Cos[beta[4500]] Chizxx[Th, r]
22 + Lzz[675, nL] Lzz[800, nL] Lzz[4500, nL]
23 Sin[beta [675]] Sin[beta [800]] Sin[beta[4500]] Chizzz[Th, r ];
D.4 Amplitude Ratios
1 Spppssp[Th , r , nL ] := Abs[Xppp[Th, r , nL]]/Abs[Xssp[Th, r , nL]];
2 Sspsppp[Th , r , nL ] := Abs[Xsps[Th, r , nL]]/Abs[Xppp[Th, r , nL]];
3 Spppsps[Th , r , nL ] := Abs[Xppp[Th, r , nL]]/Abs[Xsps[Th, r , nL]];
4 AsqSpppssp[Th , r , nL ] := Abs[Xppp[Th, r , nL]]^2/Abs[Xssp[Th, r , nL]]^2;
D.5 Isomerization Models Up and Down
1 XpppIsoPercentageUp[TransAngle , Betaa , Np ,
2 IsoRatio ] := (1 IsoRatio) (Xppp[(n[Pi]TransAngle)/180, Betaa ,
3 Np]) + (IsoRatio) Xppp[(n[Pi](TransAngle 114))/180, Betaa , Np]
4
5 XpppIsoPercentageDown[TransAngle , Betaa , Np ,
6 IsoRatio ] := (1 IsoRatio) (Xppp[(n[Pi]TransAngle)/180, Betaa ,
7 Np]) + (IsoRatio) Xppp[(n[Pi](TransAngle + 114))/180, Betaa , Np]
D.6 Construction of Circle Model
1 Are = 1;
2 Phi = (n[Pi]114)/180
3 PhipD = ((n[Pi]Phip)/180);
4 Coords = fAre Cos[Th]Sin[Phi] , Are Sin[Th]Sin[Phi] , Are Cos[Phi]g;
5
6 RotMat = ff1, 0, 0g, f0, Cos[PhipD] , Sin[PhipD]g , f0, Sin[PhipD] ,
7 Cos[PhipD]gg;
8
9 CoordsRot = Coords.RotMat;
10
11 AnglePath = ArcTan[CoordsRot [[2]]/ CoordsRot [[3]]]
12 AnglePath = VectorAngle[fCoordsRot [[2]] , CoordsRot[[3]]g , f0, 1g]
13
14 XpppIsoPercentageCone[TransAngle , Betaa , Np , IsoRatio ] :=
15 (1 IsoRatio) (Subscript [n[Chi] , ppp][(n[ Pi]TransAngle)/180, Betaa , Np])
162
16 + (IsoRatio) NIntegrate[ Subscript [n[Chi] , ppp][ AnglePath /.
17 Phip > (TransAngle), Betaa , Np] , fTh, 0, 2 n[Pi]g]/(2 n[Pi])
163
Appendix E
SHG Modeling
E.1 Load Index Data
1
2 P3HTThickness = 38;
3 KP3HT = (T[RawP3HTUVVis][[6]]/ P3HTThickness
4 T[RawP3HTUVVis][[5]] )/(4 n[Pi ]);
5 OmeUVVis600 = T[f10000000/T[RawP3HTUVVis][[5]] , KP3HTg];
6 OmeUVVis600I = Interpolation [OmeUVVis600]
7
8
9 NinfP3HT = 1.65852 `;
10 dnP3HT[Ome ] :=
11 NinfP3HT +
12 2/n[Pi]NIntegrate[ (OmeUVVis600I[Omep])/(Omep Ome), fOmep, 10000,
13 35000g, Method> "PrincipalValue", Exclusions > (Omep == Ome)]
14
15 P3HTN = Table[fOme, dnP3HT[Ome]g , fOme, 10000, 35000, 10g];
16 P3HTSnm = T[f10000000./T[P3HTN][[1]] , T[P3HTN][[2]]g];
17 ListPlot [P3HTSnm]
18
19 Export["P3HTKnm. txt", P3HTKnm, "Table"]
20 Export["P3HTNnm. txt", P3HTSnm, "Table"]
E.2 Load Index Data
1 SetDirectory [RootDir ];
2 P3HTNK = ReadList["P3HTNK2. txt", fReal , Real , Realg];
3 PCBMNK = ReadList["PCBMNK Cleaned. txt", fReal , Real , Realg];
4
5 P3HTcomp =
6 Join[T[fT[P3HTNK][[1]] , T[P3HTNK][[2]] + IT[P3HTNK][[3]]g] ,
7 Table[fWL, P3HTNK[[501]][[2]] + IP3HTNK[[501]][[3]]g , fWL, 802,
8 2000, 5g]];
9 PCBMcomp =
10 Join[T[fT[PCBMNK][[1]] , T[PCBMNK][[2]] + IT[PCBMNK][[3]]g] ,
11 Table[fWL, PCBMNK[[501]][[2]] + IPCBMNK[[501]][[3]]g , fWL, 802,
12 2000, 5g]];
13 nP3HT = Interpolation [P3HTcomp];
14 nC60 = Interpolation [PCBMcomp];
15 nAIR = 1.;
16 AngInc = 2 n[Pi](63/
164
17 180 ) (angle of incidence of electric fields in degrees);
18 GammaNeat[Lambda ] := ArcSin[(nAIR Sin[AngInc])/nP3HT[Lambda]];
19 GammaC60[Lambda ] := ArcSin[(nC60[Lambda] Sin[AngInc])/nP3HT[Lambda]];
E.3 Dene Fresnel Factors
1 LxxN[Lambda ] := (2. nAIR Cos[GammaNeat[Lambda]])/(
2 nAIR Cos[GammaNeat[Lambda]] + nP3HT[Lambda] Cos[AngInc]);
3 LyyN[Lambda ] := (2. nAIR Cos[AngInc])/(
4 nAIR Cos[AngInc] + nP3HT[Lambda] Cos[GammaNeat[Lambda]]);
5 LzzN[Lambda ] := (2. nP3HT[Lambda] Cos[AngInc])/(
6 nAIR Cos[GammaNeat[Lambda]] + nP3HT[Lambda] Cos[AngInc]);
7
8 LxxC60[Lambda ] := (2. nC60[Lambda] Cos[
9 GammaNeat[Lambda]])/(nC60[Lambda] Cos[GammaNeat[Lambda]] +
10 nP3HT[Lambda] Cos[AngInc]);
11 LyyC60[Lambda ] := (2. nC60[Lambda] Cos[
12 AngInc])/(nC60[Lambda] Cos[AngInc] +
13 nP3HT[Lambda] Cos[GammaNeat[Lambda]]);
14 LzzC60[Lambda ] := (2. nP3HT[Lambda] Cos[
15 AngInc])/(nC60[Lambda] Cos[GammaNeat[Lambda]] +
16 nP3HT[Lambda] Cos[AngInc]);
E.4 Dene Normalization Factors
1 xPSSN[Lambda , Th , r ] :=
2 LzzN[Lambda/2] LyyN[Lambda] LyyN[Lambda] Sin[AngInc] xZYY[Th, r ];
3 xPSSC60[Lambda , Th , r ] :=
4 LzzC60[Lambda/2] LyyC60[Lambda] LyyC60[Lambda] Sin[AngInc] xZYY[Th,
5 r ];
6
7 AConst =Cos[AngInc] Cos[AngInc] Sin[AngInc] LxxN[1200] LzzN[1200]
8 BConst Cos[AngInc] Sin[AngInc] Cos[AngInc] LzzN[1200] LxxN[1200]
9 CConst = Sin[AngInc] Cos[AngInc] Cos[AngInc] LxxN[1200] LxxN[1200]
10 DConst = Sin[AngInc] Sin[AngInc] Sin[AngInc] LzzN[1200] LzzN[1200]
11
12 xxTozz = Mean[Table[LxxN[Lambda]/LzzN[Lambda] , fLambda, 500, 700g]]
13
14 xPPPNSimp[Lambda ] :=
15 LzzN[Lambda]xxTozzAConst+
16 LzzN[Lambda]xxTozzBConst +
17 LzzN[Lambda]CConst +
18 LzzN[Lambda]DConst
E.5 Fitting Thickness Dependent SHG Spectra
1 Intercepts = Table[0 , fWavein, 1, NItorg];
2 Slopes = Table[0 , fWavein, 1, NItorg];
3 InterceptsErrors = Table[0 , fWavein, 1, NItorg];
165
4 SlopesErrors = Table[0 , fWavein, 1, NItorg];
5 InterceptsE = Table[0 , fWavein, 1, NItorg];
6 SlopesE = Table[0 , fWavein, 1, NItorg];
7
8 InterceptsSQ = Table[0 , fWavein, 1, NItorg];
9 SlopesSQ = Table[0 , fWavein, 1, NItorg];
10 InterceptsSQE = Table[0 , fWavein, 1, NItorg];
11 SlopesSQE = Table[0 , fWavein, 1, NItorg];
12 InterceptsSQErrors = Table[0 , fWavein, 1, NItorg];
13 SlopesSQErrors = Table[0 , fWavein, 1, NItorg];
14
15 RelPhase = 0;
16
17 PhaseModel = Abs[( Surface + E^(IRelPhase)Bulkx)]^2;
18
19 Do[
20 PhaseModel =
21 Abs[(1/WLs[[ Index]])^2/
22 nP3HT[WLs[[
23 Index ]]] Abs[( SurfacexPSSN[WLs[[ Index ]] , 1, 1, AngInc] +
24 E^(IRelPhase)Bulkx)]^2];
25 TempOut =
26 NonlinearModelFit[WaveTrend[Index] , PhaseModel , fSurface , Bulkg,
27 x];
28 Wavein = WLs[[ Index ]];
29 Slopes [[ Index ]] = fWavein, TempOut["BestFitParameters"][[2 , 2]]g;
30 Intercepts [[ Index ]] = fWavein,
31 TempOut["BestFitParameters"][[1 , 2]]g;
32
33 InterceptsErrors [[
34 Index ]] = ffWavein, TempOut["BestFitParameters"][[1 , 2]]g ,
35 ErrorBar[TempOut["ParameterErrors" ][[1]]]g;
36 SlopesErrors [[
37 Index ]] = ffWavein, TempOut["BestFitParameters"][[2 , 2]]g ,
38 ErrorBar[TempOut["ParameterErrors" ][[2]]]g;
39
40 InterceptsE [[ Index ]] = TempOut["ParameterErrors" ][[1]];
41 SlopesE [[ Index ]] = TempOut["ParameterErrors" ][[2]];
42
43 SlopesSQ [[ Index ]] = fWavein,
44 TempOut["BestFitParameters"][[2 , 2]]^2g;
45 InterceptsSQ [[ Index ]] = fWavein,
46 TempOut["BestFitParameters"][[1 , 2]]^2g;
47
48 InterceptsSQE [[ Index ]] =
49 Abs[2TempOut["ParameterErrors"][[1]]
50 TempOut["BestFitParameters"][[1 , 2]]];
51 SlopesSQE [[ Index ]] =
52 Abs[2TempOut["ParameterErrors"][[2]]
53 TempOut["BestFitParameters"][[2 , 2]]];
54
55 InterceptsSQErrors [[ Index ]] = ffWavein, InterceptsSQ [[ Index , 2]]g ,
56 ErrorBar[InterceptsSQE [[ Index ]]]g;
57 SlopesSQErrors [[ Index ]] = ffWavein, SlopesSQ [[ Index , 2]]g ,
166
58 ErrorBar[SlopesSQE [[ Index ]]]g;
59
60 , fIndex , 1, NItorg
61 ]
62
63 ListPlot[fIntercepts , Slopesg, Joined > True]
64 BulkSurf =
65 ErrorListPlot[fInterceptsErrors , SlopesErrorsg, Joined > True]
66
67 ListPlot[fInterceptsSQ , SlopesSQg, Joined > True]
68 BulkSurf =
69 ErrorListPlot[fInterceptsSQErrors , SlopesSQErrorsg, Joined > True]
70
71 SetDirectory [RootDir <> "Exports"];
72 Export["IntP3HTFre. txt", Intercepts , "Table"]
73 Export["IntEP3HTFre. txt", InterceptsE , "Table"]
74 Export["SlopeP3HTFre. txt", Slopes , "Table"]
75 Export["SlopeEP3HTFre. txt", SlopesE , "Table"]
76
77 Export["IntSQP3HTFre. txt", InterceptsSQ , "Table"]
78 Export["IntSQEP3HTFre. txt", InterceptsSQE , "Table"]
79 Export["SlopeSQP3HTFre. txt", SlopesSQ , "Table"]
80 Export["SlopeSQEP3HTFre. txt", SlopesSQE , "Table"]
167

Abstract (if available)

Abstract This thesis contains three different projects of surface specific non-linear spectroscopies. The work shown here both explores the limit of these methods and use them to under- stand molecular surfaces. The first part, is an implementation of the cross- correlated frequency-resolved optical grating (XFROG) algorithm. This method was developed to deconvolute the explicit time dependent field of an IR laser pulse from a time delayed SFG mapping. Unlike traditional XFROG algorithms, our approach relies on a non-iterative deconstruction procedure. We present a proof of concept set of experiments that confirm the validity of both the theory and the implemented XFROG algorithm. ❧ The second part of this thesis is an SFG study of the isomerization of azobenzene terminated SAMs. A novel synthesis for an azobenzene terminated alkane thiol is presented as well as chemical characterization of the product. A number of binary SAM devices were built of different dilutions of the azobenzene thiol mixed with an alkane thiol to examine at the role packing and intermolecular sterics have on the isomerization. VSFG was used to quantitatively study the vibrational modes, surface concentration of molecules, and molecular orientation in these samples. In situ SFG measurements were taken during isomerization of these films. It was observed that isomerization proceeded only for dilute films implying a steric inhibition of isomerization in neat films. Modeling was performed on the change of SFG signal observable upon isomerization, which implied that for the dilute SAMs there was a strong conformational bias in isomerization away from the surface. ❧ The last part of this thesis describes the construction of a new picosec- ond scanning SHG spectroscopy setup and it's application of this setup to study the electronic excitations at the buried lamellar donor/acceptor interface in solar cells. A new experimental setup was devised using a computer-controlled tuned of optical para- metric amplifier to allow for automated, rapidly scanning SHG spectra to be recorded. This apparatus was used to look at the differences between the P3HT/air and P3HT/C60 interfaces. Thickness dependent modeling was used to extrapolate the surface specific signal from the data taken as there is a bulk SHG contribution along with the surface specific signal. This technique was also used to look at some new red absorbing polymers, P3HTT-10%DPP and P3HTT-10%TPD.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Understanding organic and inorganic semiconductor surfaces by coherent nonlinear spectroscopy

PDF

Vibrational sum frequency spectroscopy of molecules on metal, semiconductor, and aqueous surfaces

PDF

Nanocrystal surface engineering as a route towards improved photovoltaics

PDF

Interfacial vibrational and electronic excited states of photovoltaic materials probed using ultrafast nonlinear spectroscopy

PDF

Molecular orientation from sum frequency generation spectroscopy: case study of water and methyl vibrations

PDF

Structure-function studies of random and semi-random poly(3-hexyl thiophene) copolymers

PDF

Monitoring local electric fields, charge densities at electrode surfaces using in-situ graphene/surface enhanced Raman spectroscopy (GERS/SERS) based Stark-shifts

PDF

Vibrational spectroscopy and molecular orientation at interfaces: correlating the structure of a material's surface to its properties

PDF

Asymmetric response and the reorientation of interfacial water with respect to applied electric fields

PDF

Vibrational sum frequency generation studies of electrode interfaces

PDF

Direct arylation polymerization (DARP) for the synthesis of conjugated polymers

PDF

Excited state dynamics of organic photovoltaics probed using ultrafast spectroscopy

PDF

Expanding the library of surface ligands for semiconductor nanocrystal synthesis and photovoltaic applications

PDF

Diamond surface chemistry for NV quantum sensing

PDF

Water partitioning between the bulk and an electrode surface

PDF

Quantitative computer-aided studies of artificial and enantioselective enzymes

PDF

Vapor phase deposition of polymers in the presence of low vapor pressure liquids

PDF

High-frequency and high-field magnetic resonance spectroscopy of diamond

PDF

The influence of CdSe surface ligands on hybrid inorganic/organic solar cell performance

PDF

Developmnt of high-frequency electron paramagnetic resonance (EPR) spectrometer and investigation of paramagnetic defects and impurities in diamonds by multi-frequency EPR spectroscopy

Asset Metadata

Creator Valley, David Taylor (author)

Core Title Non-linear surface spectroscopy of photoswitches annd photovoltaics

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Publication Date 01/28/2013

Defense Date 12/17/2012

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag azobenzene,OAI-PMH Harvest,photovoltaics,second harmonic generation,sum frequency generation,surfaces

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bradforth, Stephen E. (committee chair), Benderskii, Alexander V. (committee member), Nakano, Aiichiro (committee member)

Creator Email DValley@gmail.com,dvalley@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-130477

Unique identifier UC11290581

Identifier usctheses-c3-130477 (legacy record id)

Legacy Identifier etd-ValleyDavi-1411.pdf

Dmrecord 130477

Document Type Dissertation

Rights Valley, David Taylor

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA