University of Southern California Dissertations and Theses

Organic solar cells: molecular electronic processes and device development

(USC Thesis Other)

Organic solar cells: molecular electronic processes and device development

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ORGANIC SOLAR CELLS:
MOLECULAR ELECTRONIC PROCESSES AND DEVICE DEVELOPMENT

by
Cody Williams Schlenker

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CHEMISTRY)

December 2010
Copyright 2010 Cody Williams Schlenker
ii
Epigraph
Of what a strange nature is knowledge! It clings to the mind, when it has seized on it,
like a lichen on the rock.
-the creation of V. F.
Mary Shelley, Frankenstein (XIII. 103)
iii
Dedication
To Katie for her vision and resilience
To my parents for teaching the value of critical thought
To my brother for his insight and inspiration
iv
Acknowledgements
I am grateful to Professor Mark Thompson for affording me this introduction to organic
photovoltaics. His support and mentorship have been outstanding and I will appreciate them always. He
has demonstrated to me a research philosophy that I hope to continue as I begin the next step in my career
and onward.
Thank you to my committee members Professor Dan Dapkus and Professor Steve Bradforth for
their willingness to be a part of this process. I would also like to thank Professors Surya Prakash, Anna
Krylov, and Chongwu Zhou for their service in my advance to candidacy. Thank you to Professor Curt
Wittig for always offering a refreshing perspective and supplying me with a wealth of information for
further study. Professor Bradforth’s molecular spectroscopy class was superb and I very much appreciate
his investment in teaching, both in the classroom and as a collaborator.
I am very thankful for the friendship and mentorship of my fellow graduate students. Among
them, I cannot put a value on my abiding friendship with Carsten Borek. From Marco Curreli I have
learned that there is always something to learn. I will forever be indebted to Dolores Perez, as a mentor,
guardian, coconspirator, and friend. Much of this work would not have been possible without several
outstanding collaborations. For this I acknowledge Dr. Matthew Franzman, Professor Richard Brutchey,
Koungmin Ryu, Lewis Gomez de Arco, Akshay Kumar, Professor Chongwu Zhou, Professor Barry
Thompson, Dr. Matthew Whited, Dr. Barry Rand, Dr. Libby Mayo, Erin Morrison, C. Kyle Renshaw,
Professor Stephen Forrest, Dr. Vincent Barlier, Stephanie Chin, Philipp Isken, Francisco Navarro, Dr. Sean
Roberts, Professor Steve Bradforth, Dr. Maria Dolores Perez, and Dr. Carsten Borek. Thank you also to
Judy Hom, Michele Dea, Jim Manahan, Heather Connor, Don Wiggins, and Katie McKissick for all of
their hard work. I must also thank everyone in the Thompson group for teaching me so much.
I am very thankful for the support of my family members. They have been an important part of
making me who I am. I would not have come this far without them.
Katie, I am so very proud of you and thankful for your vision and your ability to adapt. Thank
you for being willing to take each day as it comes.
v
Table of Contents
Epigraph ii
Dedication iii
Acknowledgements iv
List of Tables vii
List of Figures viii
Abstract xvi
Chapter 1 Organic photovoltaics for solar energy conversion 1
1.1 The photovoltaic effect 1
1.2 The solar photovoltaic industry 2
1.3 Organic solar cells 3
1.4 Organic materials and mechanisms 5
1.4.1 Challenges for OPV 5
1.4.2 Excitonic materials 6
1.4.3 Photoelectrochemical processes 8
1.4.4 Photon absorption 9
1.4.5 Exciton diffusion 11
1.4.6 Charge transfer 13
1.4.7 Charge separation 15
1.4.8 Charge transport 15
1.4.9 Charge collection 16
1.5 Electrical behavior 18
1.6 Open-circuit voltage 21
1.6.1 Thermodynamic description of open-circuit voltage 22
1.6.2 Orbital energy level offsets 24
1.6.3 Temperature dependence of photovoltage losses 25
1.6.4 General photovoltage description 26
1.6.5 Molecular electron transfer kinetics 28
1.6.6 CT state coupling, molecular orientation, and V
oc
30

1.6.7 Demonstrated kinetic impact on V
oc
33
1.6.8 Architectures for controlled leakage current 36
1.7 Nanostructures and materials 37
1.8 Summary of topics 38
1.9 Chapter 1 endnotes 39
Chapter 2 Instrumentation and analysis 47
2.1 Standard reporting conditions 47
2.2 Instrumentation 49
2.3 Analysis 51
2.4 Chapter 2 endnotes 61
Chapter 3 Ensemble donor systems 62
3.1 Overcoming counterposing limitations 62
3.2 Lamellar bookend structures 63
3.2.1 Potential enhancements 63
vi
3.2.2 Competing processes 64
3.2.3 Molecular perspective on open-circuit voltage losses 65
3.2.4 Materials and methods 67
3.2.5 Results and discussion 68
3.2.6 Concluding remarks 82
3.3 Host-guest sensitization systems 83
3.3.1 Absorption/diffusion mismatch 83
3.3.2 Exciton diffusion length 85
3.3.3 Preparation of triplet states 87
3.3.4 Host-Guest materials 88
3.3.5 Results and discussion 91
3.3.6 Concluding remarks 97
3.4 Chapter 3 endnotes 99
Chapter 4 Molecular aspects of charge collection in organic photovoltaics 102
4.1 Toward multipurpose buffer materials 102
4.2 Materials and methods 104
4.3 Results and Discussion 107
4.4 Concluding remarks 122
4.5 Chapter 4 endnotes 124
Chapter 5 Nanostructures in organic photovoltaics 126
5.1 Solar cells on a small scale 126
5.2 Tin(II) selenide nanocrystal-polymer composite devices 126
5.3 Solid-state hole transport media for dye-sensitized solar cells 131
5.3.1 Toward solid-state photoelectrochemical cells 131
5.3.2 Materials and methods 134
5.3.3 Results and discussion 135
5.3.4 Concluding remarks 140
5.4 Carbon-based transparent electrodes 141
5.4.1 Carbon nanotube network devices 141
5.4.2 Graphene electrode devices 144
5.4.3 Concluding remarks 154
5.5 Chapter 5 endnotes 155
Bibliography 160

vii
List of Tables
Table 1.1. Dark current and photovoltage for various donor materials 34
Table 3.1. Representative performance metrics for bookend and single-donor organic solar
cell devices 79
Table 3.2. Thickness dependence for host-guest Dpt-PtOEP devices 92
Table 3.3. Thickness dependence for host-guest Dpt-PtTPBP devices 97
Table 4.1. Ligand substitution, electrochemical redox potentials, and bulk orbital energy
levels 105
Table 4.2. Measured 100 Å device performance metrics 111
Table 4.3. Measured 200 Å device performance metrics 113
Table 4.4. Model anti-polar diode parameters for 200 Å buffer devices 116
Table 5.1. Performance metrics for graphene OPV cells fabricated on PET 150

viii
List of Figures
Figure 1.1. Number of organic solar PV reports as a percentage of total solar PV reports in
scholarly peer-reviewed journals between the years of 1990 and 2010,
illustrating the redoubled research effort in organic solar cells beginning ca.
2000. 4
Figure 1.2. Donor materials 1-10 and acceptor materials A and B described in this chapter. 5
Figure 1.3. Contemporary organic solar cell devices are based on donor-acceptor
heterojunction device architectures. a) Energy level diagram. b) Planar
heterojunction configuration. c) Bulk heterojunction configuration. 7
Figure 1.4. Photocurrent generation in organic solar cells 1) exciton generation 2) exciton
diffusion 3) charge transfer state generation 4) charge separation 5) charge
transport 6) charge collection. 9
Figure 1.5. Solar photon flux Φ (derived from ASTM G173-03 AM1.5G spectral irradiance)
is plotted as a function of wavelength in the top panel. The absorption
coefficient α for the active region of an archetypal organic donor/acceptor pair,
CuPc/C
60
(solid red trace, Organic D/A), is compared in the intermediate panel
with that of gallium arsenide (dashed black trace, GaAs), a direct band gap
semiconductor and both are contrasted with crystalline silicon (dotted blue trace,
c-Si), an indirect band gap semiconductor. Plotted in the bottom panel, for an
active layer thickness of x = 300 nm, is the cross-sectional number of photons
captured N
p
= Φ – Φ[exp(-αx)] in a single-pass optical path. 10
Figure 1.6. Simulated solar cell electrical behavior in the dark (dotted lines) and under
illumination (solid lines) comparing the effect of the saturation current
parameter J
s
on V
oc
. The sharp inflection points in the semilog plots (upper
panel) are the points where the current switches form positive to negative. The
‘Large Dark Current’ trace in black represents J
s
× 10
6
that of the ‘Small Dark
Current’ trace in red. Also illustrated are P
max
(filled gray rectangle) and the
numerical product J
sc
× V
oc
(unfilled gray rectangle) used to calculate FF =
P
max
/(J
sc
× V
oc
). 17
Figure 1.7. Non-radiative recombination losses occurring under forward bias in a typical
OPV device. Holes injected from the anode Fermi level (E
F,A
) into the HOMO
level (E
i
) of the donor and electrons injected from the cathode Fermi level (E
F,C
)
into the LUMO level (E
a
) of the acceptor are transported to the D/A interface.
Coulombic attraction between holes and electrons yields the D
+
A
-
CT state with
energy E
CT
. Charge recombination reaction D
+
+ A
-
→ D + A occurs with rate
constant k
rec
. 19
Figure 1.8. Single diode equivalent circuit model commonly employed in estimating solar
cell losses. 20
Figure 1.9. Equilibrium energy diagram for a pn junction in an inorganic semiconductor
material with intrinsic Fermi energy E
Fi
, conduction band energy E
c
, valence
band energy E
v
and with potentials φ
Fp
and φ
Fn
given by the respective impurity
concentrations in the p and n regions. The quantity V
bi
represents the total built-
in electrical potential due to band bending. 22

ix
Figure 1.10. Energy level diagram for a typical D/A heterojunction depicting donor material
with ionization energy of E
i
and excited state ionization energy of E
i
*, acceptor
with electron affinity of E
a
and excited state electron affinity of E
a
*, and the
interfacial energy level offset between E
i
and E
a
illustrated as ΔE
DA
, all relative
to vacuum as described in the text. 23
Figure 1.11. Temperature dependence of V
oc
for CuPc and pentacene donor-based devices
compared with the relative temperature independence of V
oc
≈ 0.9 V for the
NPD based device. Reproduced with permission from Ref. 5, Copyright 2007
The American Physical Society. 25
Figure 1.12. Schematic representation of two hypothetical donor materials embodying the
efficiency limiting trade-off between suppressed voltage losses on the left and
robust spectral coverage on the right. Donor 1 exhibits a large ΔE
DA
and is
expected to produce a large open-circuit voltage, whereas Donor 2 exhibits a
small ΔE
DA
, but offers enhanced coverage of the solar spectral irradiance. 27
Figure 1.13. Free energy curves illustrating the relationship between the free energy of
activation ΔG*, reorganization energy λ, the free energy of reaction ΔG°, and
the coupling element H
ij
that determine the electron transfer rates in an OPV
device. 30
Figure 1.14. Molecular orientation and its influence on calculated charge transfer rates
between pentacene and C
60
, indicating that suppressed recombination and
efficient photoinduced charge transfer are not mutually exclusive. For example,
the CT
0
state to the ground state reaction
2
B
2g
P+
⊗
2
T
1u
C60-
→
1
A
g
P
⊗
1
A
g
C60
rate for
parallel orientation 2B (dotted red trace in Parallel plot) is slightly suppressed,
while the forward charge transfer rates for the same orientation (solid and
dashed red traces) are an order of magnitude or more greater than orientation 2A
(blue trace). Reproduced from Yi with permission. Copyright 2009 American
Chemical Society. 31
Figure 1.15. Semi-log scale J-V characteristics in the dark (filled circles) and under
illumination (open circles) for tetracene (black) and rubrene (red) based OPV
devices illustrating the substantially higher V
oc
in the case of rubrene compared
with tetracene, despite their similar ionization potentials. Adapted from Perez
with permission, Copyright 2009, American Chemical Society. 33
Figure 1.16. Semi-log scale J-V characteristics in the dark (filled circles) and under
illumination (open circles) for CuPc (black), PtTPBP (red), and PtTPNP (blue)
based OPV devices suggesting that both ΔE
DA
and kinetic accessibility of the π-
system play a role in determining V
oc
. Adapted from Perez with permission,
Copyright 2009, American Chemical Society. 35
Figure 1.17. Bookend device structures comprising multi-component lamellar donor or
acceptor regions are a promising route to tailored interfaces for limiting voltage
losses and potentially enhancing spectral coverage. 37
Figure 2.1. Hamamatsu Si photodiode, calibrated at the National Renewable Energy
Laboratory (NREL) used as reference photodetector. 48
Figure 2.2. Instrumentation employed in OPV optoelectronic characterization. 49

x
Figure 2.3. Raw current under monochromatic illumination for a typical CuPc/C
60
OPV
(solid trace) and for an 8RA filtered silicon photodiode reference cell (dotted
trace). 50
Figure 2.4. The Script Window is accessible via the native toolbar in OriginPro and allows
the user to develop custom programs such as the simple data transfer script
‘It_import.ogs.’ 51
Figure 2.5. The OriginPro toolbar can be customized with dedicated user-defined buttons to
expedite repetitive procedures using steps a – d. 52
Figure 2.6. Sample customized data treatment script, ‘PV_ Treatment_ CWS.ogs,’
employed in extraction of OPV metrics, such as short circuit current density
(J
sc
), open circuit voltage (V
oc
), compensation voltage (V
0
), maximum power
density P
max
and spectrally integrated photocurrent density J
sc
QE
. 53
Figure 2.7. The OPV data treatment project template contains step-by-step instructions in
the ‘Directions.’ window for data treatment and system calibration, including a
list of symbol definitions. 54
Figure 2.8. The SRS_Kernel comprises calibration data and performs several important
calculations, relevant for external quantum efficiency and spectral mismatch, in
the background during data treatment. 55
Figure 2.9. Spectral mismatch factor (M) calculation incorporated in the ‘MCalc’
worksheet. 55
Figure 2.10. Activating the data in the ‘MonoChromeTestCurrent’ window and selecting
‘Analysis… Calculus… Integrate’ yields the spectrally integrated photocurrent
density. 56
Figure 2.11. Photocurrent density is automatically spectral mismatch corrected. 56
Figure 2.12. Rudimentary equivalent circuit modeling can be applied to estimate OPV losses. 57
Figure 2.13. User-defined functions can be developed and saved as .fdf in the ‘Advanced
Fitting Tool’ by selecting ‘Function…New.’ 58
Figure 2.14. An externally provided .fdf routine can be accessed by selecting
‘Function/Add…’ and opening the .fdf from its location on the local hard disk. 58
Figure 2.15. The characterization layout. 59
Figure 3.1. Schematic photoactive region for proposed bookend architecture in which the
donor layer comprises three complementary materials B
1
, B
2
, and B
3
. 64
Figure 3.2. a) Chemical structures for tetracene (Ttn), 5,6-diphenyl-tetracene (Dpt), rubrene
(Rbn), 5,6-dinaphthyl-tetracene (Dnt), pentacene (Ptn), chloroaluminum
phthalocyanine (ClAlPc), fullerene (C
60
), and bathocuproine (BCP). b) Current
voltage characteristics of single Ttn (black), Dpt (red), or Rbn (blue) 600Å
donor layer OPV devices comprising, ITO/ donor/ C
60
(400Å)/ BCP(100Å)/ Al,
under AM1.5G 1 sun illumination corrected for spectral mismatch corrected
ASTM G173-03 and in the dark (filled symbols). 69

xi
Figure 3.3. a) Neat film excitation (Ex) and emission (Em) spectra illustrating the cascading
exciton energies for the series diphenyltetracene (Dpt), tetracene (Ttn), and
rubrene (Rbn) and the proposed mechanism of Förster-assisted interlayer
excitation transfer between singlet S
1
states of the lamellar donor materials, and
subsequent charge transfer quenching of the rubrene exciton leading to the
charge transfer CT
n
state at the donor/acceptor interface. b) External quantum
efficiency signal for photovoltaic devices comprising ITO/ Donor / C
60
(400 Å)/
BCP (100 Å)/ Al, where Donor is a 650 Å Ttn film coated with and additional
50 Å of either Rbn (filled symbol), Dpt (cross symbol), or Ttn (open symbol).
c) Optical excitation energies derived from a), suggesting excitation transfer
from Ttn to Dpt will be endothermic by ~ 4kT, while excitation transfer from
tetracene to rubrene will be slightly exothermic resulting in the EQE traces in b). 70
Figure 3.4. Current-density as a function of applied field for ITO/ACENE/Cu devices,
where ACENE = Ttn (circle) or DPT (diamond) and positive potential applied to
ITO. 72
Figure 3.5. a) Grazing incidence x-ray diffraction peaks for 600 Å thick tetracene (Ttn)
films grown on pristine ITO (black filled) or ITO coated with 50 Å Dpt (red
open), diphenyltetracene. b) Schematic representation and topographical images
obtained by atomic force microscopy corresponding to samples in a). 73
Figure 3.6. Dark electrical characteristics for Device B, the tetracene based bookend device
(red filled), relative to Device A, the single donor tetracene-based device (black
filled), resulting in the 200 mV increase in V
oc
for the bookend device
characteristics under illumination (red cross) relative to the single donor-based
device (black open). Devices consist of ITO/ Donor (450 Å)/ C
60
(400 Å)/ BCP
(100 Å)/ Al, where Donor is either a single layer of neat Ttn or a lamellar array
comprising Dpt (50Å)/ Ttn (200Å)/ Rbn (200 Å). 73
Figure 3.7. Thin film absorption coefficient (α) for Dpt (blue), Ttn (black), and Rbn (red). 75
Figure 3.8. Electrical characteristics for Dnt-based bookend device structure comprising
ITO/ Donor/ C
60
(400 Å)/ BCP (100 Å)/ Al in the dark (filled) and under
simulated solar illumination (open). Donor is given by Dnt (50 Å)/ Ttn (350 Å)/
Rbn (50 Å). 76
Figure 3.9. Electrical characteristics for ITO/MoO
3
OPV devices comprising ITO/MoO
3

(100 Å) / Donor/ C
60
(400 Å)/ BCP (100 Å)/ Al. Donor represents the bookend
structure Dpt (50 Å)/ Ttn (200 Å)/ Rbn (200 Å) in the dark (red filled symbols)
and under illumination (red cross symbols) or a single Ttn (450 Å) donor layer
in the dark (black filled symbols) and under illumination (black open symbols). 77
Figure 3.10. a) Absorption spectra for thin films of Ptn (dotted red trace), C
60
(dashed black
trace), and ClAlPc (solid blue trace). b) Quantum efficiencies illustrating the
excitation transfer fingerprint of the pentacene layer in the bookend device
comprising a Ptn (300Å)/ClAlPc (300Å) donor layer (solid red trace) compared
with a single 300 Å ClAlPc donor layer (dashed black trace) in ITO/ donor/
C
60
(400Å)/ BCP(100Å)/ Al devices. 78

xii
Figure 3.11. Current voltage characteristics for Device D, a Ptn-based bookend donor
structure comprising [Ptn (600Å)/ ClAlPc (300Å)] (red crosses) under simulated
AM1.5G 1 sun illumination with spectral mismatch correction to ASTM G173-
03, compared with an archetypal [Pentacene (600Å)] single donor layer (black
open symbols). Complete device structure consists of Glass/ ITO/ Donor/ C
60

(400Å)/ BCP (100Å)/ Al (1000Å). Inset illustrates suppressed dark current
(black filled symbols) for Device D compared on a semi-log scale with Device
C, the Ptn single donor (red filled symbols). 81
Figure 3.12. Fraction of excitons N
p
(λ, L
D
)/Φ(λ) generated within a diffusion length (L
D
)
from the D/A interface relative to incident solar photon flux Φ(λ) for
hypothetical donor materials. Assumes arbitrary L
D
, two-pass optical path, and
absorption coefficient identical to CuPc. 84
Figure 3.13. Potential excitation transfer processes occurring in Host-Guest (H-G)
sensitization systems. 88
Figure 3.14. Chemical structures and optical absorption for solid samples of PtOEP (dotted
trace) and PtTPBP (solid trace) in inert matrices of polystyrene and tetra(9,9’-
dimethylfluoren-2-yl)silane, respectively. 89
Figure 3.15. Absorption spectra (upper panel) for equal film thickness of Dpt (dotted trace)
and Dpt-PtOEP [20%] (solid trace). External quantum efficiency (lower panel)
for devices incorporating a donor layer of Dpt (dotted trace) or a Dpt-PtOEP
composite (solid trace). Device structures consist of ITO/donor (300 Å)/ C
60

(400 Å)/ BCP (100 Å)/ Al. 90
Figure 3.16. External quantum efficiency demonstrating prominent PtOEP signal for OPV
device incorporating a guest/acceptor blocking layer (filled symbol trace) in
contrast to Dpt-based single donor device (open symbol trace). Blocking
architecture given by ITO/Dpt-PtOEP [20%] (200 Å)/ Dpt (100 Å)/C
60
(400 Å)/
BCP (100 Å)/Al. 91
Figure 3.17. a) Solid state absorption spectra (upper panel) for C
60
, Dpt and PtOEP.
Thickness dependence of external quantum efficiency (EQE) for Dpt-PtOEP
host-guest devices. b) Current-density voltage characteristics in the dark and
under simulated AM1.5G 1 sun illumination with spectral mismatch correction
to ASTM G173-03 for devices in a). Device structure consists of Glass/ ITO/
Dpt-PtOEP [20%] (x Å)/ C
60
(200Å)/ BCP (100Å)/ Al. 93
Figure 3.18. External quantum efficiency for Dpt-PtTPBP host-guest devices suggesting that
in a 400 Å donor triplet excitons efficiently reach the D/A interface, while
singlets do not. Device structure consists of Glass/ ITO/ Dpt-PtTPBP [20%] (x
Å)/ C
60
(200Å)/ BCP (100Å)/ Al. 95
Figure 3.19. Thickness dependence of DPt-PtTPBP host-guest electrical characteristics in the
dark and under simulated AM1.5G 1 sun illumination with spectral mismatch
correction to ASTM G173-03. Device structure consists of Glass/ ITO/ Dpt-
PtTPBP [20%] (x Å)/ C
60
(200Å)/ BCP (100Å)/ Al. 96

xiii
Figure 4.1. Mechanism for reciprocal carrier collection in double heterojunction OPVs.
Photogenerated electrons (black circles) in the acceptor layer are collected by
recombination with holes (red circles) from the reversibly oxidized buffer layer.
Black arrows represent electron motion, while red arrows represent motion of
holes. 103
Figure 4.2. Preparation method for compounds 2-6. Ruthenium trichloride is refluxed under
mildly reducing conditions to produce a mixed valent “ruthenium blue” solution.
Addition of excess ligand and multiple fractions of KHCO
3
to neutralize
liberated hydronium ion yield the crude precipitate. 107
Figure 4.3. a) The parent tris-acetylacetonate complex 1 and substituted analogues 2-6 used
as buffer materials between the C
60
acceptor and the Ag cathode. b) Bulk
ionization energies (solid black) for complexes 1-6, reflecting the electron
withdrawing nature of the chelating ligands. Solid black lines represent HOMO
energy levels measured relative to vacuum, using ultraviolet photoelectron
spectroscopy. Dashed gray lines represent LUMO energy levels, estimated
according to literature from measured solution electrochemical data in Table 4.1. 108
Figure 4.4. a) Schematic representation of the studied device architecture, consisting of
ITO/ CuPc(400 Å)/ C
60
(400 Å)/ buffer(x Å)/ Ag(1000 Å). b) Current-density
vs. voltage characteristics for devices fabricated with x = 100 buffer layers of
compounds 1-6, showing comparable overall performance for all devices. c)
The measured (open symbols) and simulated (solid lines) J(V) traces for devices
with x = 200 exhibiting the non-ideal inflection behavior of compounds 2-6, in
contrast to compound 1 at the same thickness. 109
Figure 4.5. Fill factor of devices based on 200 Å films of compounds 1-6, showing
prominent dependence on complex ionization energy (E
i
) measured using
ultraviolet photoelectron spectroscopy. 112
Figure 4.6. a) The APD equivalent circuit model used to simulate the solid traces in Figure
4.4a and to estimate non-ideal devices parameters. Voltage applied across the
external leads is distributed between d
1
and d
2
,

as described in the text. b)
Magnitude of the saturation current-density (J
s2
) for Diode 2 used in APD model
simulations of 200 Å buffer layer devices plotted as a function of experimentally
observed FF showing that J
s2
increases by six orders of magnitude with
observed FF. The dotted line is drawn as guide to the eye. c) Experimentally
determined output power density (open symbols) for 200 Å buffer layer devices
compared with simulated data (solid lines) for the same, showing that increased
saturation current for Diode 2 (J
s2
) corresponds to enhanced output power
density. 115
Figure 4.7. a) Schematic representation of energy offset (ΔE
BC
) between Ag Fermi Level
(E
F
) and the ionization energy (E
i,C
) for each buffer complex measured on Ag
coated substrates. b) The estimated diffusion-controlled saturation current for
Diode 2 (J
s2
) fails to correlate with increasing ΔE
BC
. Interfacial offsets for
complexes 2, 3, 5, and 6 measured on Ag coated substrate, offset for complex 4
estimated from bulk measurement. 118
Figure 4.8. a) Ultraviolet photoelectron spectra measured for ruthenium complexes 1-6 on
ITO/C
60
substrates. b) Interfacial energy offsets (ΔE
AB
) between the electron
transport level of C
60
and the hole transport level of each RuL
3
buffer layer. 120

xiv
Figure 4.9. a) Estimated J
s2
for 200 Å OPV devices decrease with increasing interfacial
energy level offset ΔE
AB
between acceptor layer and buffer complexes 2-6. The
dotted line is included as a guide to the eye. b) Space filling molecular models
for buffer complexes 4 and 5, illustrating their respective thienyl- (Left) and
naphthyl- (Right) substitution and the increase in π-system exposure afforded by
the naphthalene moiety of complex 5. Atoms are color coded according to
carbon (black), sulfur (turquoise), oxygen (red), fluorine (green), and hydrogen
(gray). 121
Figure 5.1. SnSe nanocrystals synthesized using di-tert-butyl diselenide at 180 ˚C. High-
resolution TEM image of a single nanocrystal (a), SAED pattern of the
nanocrystals (b), and low-resolution TEM image of the SnSe nanocrystals (c).
(d) Absorbance spectrum of SnSe nanocrystals dissolved in cyclohexane. 127
Figure 5.2. Topographic AFM images (2.5 µm x 2.5 µm) of representative (a)
glass/ITO/MoO3 substrate, (b) 350 Å PPV layer on glass/ITO/MoO3, and (c)
350 Å SnSe:PPV layer on glass/ITO/MoO3 surfaces with calculated RMS
roughness values of 2.8 nm, 5.5 nm, and 6.3 nm, respectively. 128
Figure 5.3. Overlaid spectral mismatch corrected J(V) (filled) and P(V) (open)
characteristics for ITO/MoO
3
/poly/PTCDI/LiF/Al devices, where poly is a
0.25:1.0 wt/wt SnSe:PPV film or a neat polymer film under 1000 W m
-2
AM
1.5G illumination. 129
Figure 5.4. (a) Absorption coefficient of the hybrid SnSe:PPV and neat PPV films as a
function of wavelength indicating that the absorption near 500 nm is comparable
for both films. (b) Quantum efficiency enhancement for
ITO/MoO3/poly/PTCDI/LiF/Al devices when poly = PPV is replaced with the
poly = SnSe:PPV composite layer. 130
Figure 5.5. Output power degradation of ITO/MoO3/poly/C60/BCP/Al devices, where poly
is (a) PPV or (b) SnSe:PPV. 131
Figure 5.6. a) Illustration of the solid state dye sensitized solar cell configuration. A vertical
array of TiO
2
nanowires is grown directly on FTO substrate, sensitized with
N719 dye molecules, infiltrated with vapor deposited NNP:F
4
TCNQ, and coated
with Cu electrodes. b) Energy levels for various sDSC components. All values
presented in eV. c) Photocurrent generation mechanism involving (I) photon
absorption by the dye molecule (D) to generation D*, (II) electron injection
from D* into the conduction band (CB) of TiO
2
, (III) hole collection from D
+
by
the doped NNP layer hopping assisted transport. Also depicted is the
unfavorable charge recombination process (IV) between the photogenerated
electron in TiO
2
and D
+
. 133
Figure 5.7 a) Top view SEM image of as-grown TiO
2
nanowire array on FTO substrate
prior to NNP infiltration. b) Cross-sectional SEM image of the nanowire array
showing individual nanowires growing vertically on FTO substrate with a length
of 200 nm. c) Top view SEM image of TiO
2
nanowire array after HTM
deposition showing the encapsulated nanowires with NNP molecules forming a
percolating network. d) Cross-sectional SEM image of 200 nm long TiO
2

nanowires coated with doped NNP showing infiltration of HTM molecules
within the nanowire array. 136

xv
Figure 5.8. a) Current density versus voltage characteristics for OPVD-fabricated sDSCs
prepared using 200 nm nanowires in the dark (red) and under simulated solar
illumination (blue). The open circle trace in the lower panel represents the
output power density of the device under illumination. b) External quantum
efficiency (EQE, filled blue semicircles), light harvesting efficiency (LHE, open
red circles), and estimated internal quantum efficiency (IQE, filled black circles)
for OVPD-fabricated sDSC on 200 nm TiO
2
nanowire array. c) Transmittance
(open black circles) for 300 nm NNP:F
4
TCNQ hole transport layer and
reflectance spectrum (open red squares) for copper. d) Overlaid photocurrent
(unfilled symbols) and power output (filled symbols) under simulated solar
illumination for OPVD-fabricated sDSC for 200 nm (black) and 400 nm (red)
nanowire samples coated by OVPD with 300 nm and 500 nm thick HTM layers. 138
Figure 5.9. Electrical characteristics for flexible OPV devices based on transparent
electrode materials comprising carbon nanotube networks (square symbol trace)
or ITO (circle symbol trace). 142
Figure 5.10. From upper panel to lower panel, optical transmittance for CNT (solid trace) and
ITO (dotted trace) electrodes used in Figure 5.9, absorbance of the OPV active
layer used in Figure 5.9, external quantum efficiency for OPV devices in Figure
5.9. 143
Figure 5.11. a) Schematic of the CVD graphene transfer process onto transparent substrates.
b) Transmittance/resistance data for the resulting electrodes from a). 146
Figure 5.12. a) Photograph illustrating high flexibility of CVD graphene transferred on a PET
flexible substrate. b,c) AFM images of the surface of CVD graphene and ITO
films on PET, respectively. d,f) Conductance of the CVD graphene and ITO
films on PET substrates under bending conditions, respectively. The devices
used to monitor the conductance had channel width (W) 1 mm and length (L) 1
mm. e) Optical images of CVD graphene (top) and ITO (bottom) films on PET
before and after being bent at the angles specified in panels b and c. Arrows
show the direction of the bending. 147
Figure 5.13. Logarithmic (top) and linear (bottom) current density and power density vs.
voltage characteristics of CVD graphene a) and ITO b) OPV cells on PET under
dark (red traces) and 100 mW/cm
2
AM1.5G spectral illumination (blue traces).
The output power density of the cells is plotted in panels a and b as open circle
traces. The structure of the devices is given by [CVD
graphene/PEDOT/CuPc/C60/BCP/Al] and [ITO/CuPc/C60/BCP/Al] for CVD
graphene and ITO OPVs, respectively. c) Comparison of the modeled (solid
lines) current density and power density curves obtained from the Shockley
equation against the graphene and ITO device values obtained experimentally
(symbols). 149
Figure 5.14. Current density vs voltage characteristics of CVD graphene a) or ITO b)
photovoltaic cells under 100 mW/cm
2
AM1.5G spectral illumination for
different bending angles. Insets show the employed experimental setup. c) Fill
factor dependence of the bending angle for CVD graphene and ITO devices. d)
SEM images showing the surface structure of CVD graphene (top) and ITO
(bottom) photovoltaic cells after being subjected to the bending angles described
in panels a and b. 153

xvi
Abstract
Presently, collective understanding of the involved processes and requisite components that may
lead to maturation of organic photovoltaic devices from bench top concept to disruptive solar energy
conversion technology have not affected efficiencies within the balance of systems threshold. Primarily,
poor broadband spectral photon capture and exciton diffusion, deleterious charge carrier recombination,
and poor charge carrier collection, appear to be major efficiency limiting factors in these potentially cost
competitive solar cells. This dissertation highlights emerging descriptions for such processes and the
continued development of novel materials, device architectures, and process techniques to redress the
resulting losses. Lamellar and composite multi-donor systems with complementary properties are
examined as a means to robust spectral coverage, enhanced exciton diffusion, and simultaneous
suppression of photovoltage losses. Molecular aspects of charge collection in reciprocal carrier
architectures are examined. Device architectures incorporating nanostructured materials and composites
are probed. The overarching conclusions drawn from the results presented in this work underscore the
molecular nature of OPV device operation, contrasted with that of convention covalent crystalline
semiconductor devices.
1
Chapter 1
Organic photovoltaics for solar energy conversion
1.1 The photovoltaic effect
The photovoltaic (PV) effect, known since the mid 19
th
century,
4
generally describes the onset of
or change in an electrical potential occurring upon illumination when two electrodes are separated by a
suitable material. The material separating the two electrodes in a PV cell, which is any device exhibiting
the photovoltaic effect, may be either a solid or liquid.
5
Depending on the nature of this photoactive
material and the nature of the electrical contacts, one of several different mechanisms
5-10
may be
responsible for the separation of opposite polarity charges that constitutes the PV effect induced by the
absorption of electromagnetic radiation. In all cases, the photovoltaic generation of charge arises due to
some form of asymmetry within the device, either static (existing both in the presence and absence of
illumination) or photoinduced (occurring only under illumination). As a result, PV cells exhibit definite
polarity with respect to their electrical terminals, a feature that distinguishes the photovoltaic effect from
photoconductivity.
11
When charge separation occurs in a PV cell as a result of electromagnetic radiation
within the spectral irradiance of the sun, the device is referred to as a solar cell. Utilizing the voltage
supplied by a solar cell or a modular array of solar cells to power common electrical appliances is the
primary impetus behind the study of photovoltaic solar energy conversion. The capture and storage of solar
energy through the selective cleavage and formation of chemical bonds, referred to as artificial
photosynthesis
12
or solar fuel production
13
is a closely related topic with many similar processes as occur in
PV cells. Differences between various solar cell technologies are characterized most broadly by the nature
of the material in the photoactive region. In solid state solar cells, such a photoactive region may consist of
a junction formed in a single material by a process known as doping via the introduction of controlled (both
in concentration and identity) impurities on opposite sides of the junction. Generally, electron accepting
materials are introduced to form a region of concentrated positive (p) charge carriers on one side of the
junction and electron donating materials are introduced to form a region of concentrated negative (n)
2
charge carriers on the opposite side of the junction. This type of junction is referred to as a pn junction.
Alternatively, a heterojunction may be formed between two different materials, each being either doped or
undoped. Further distinctions can be drawn regarding the nature of the bonding between lattice sites, as
well as the dielectric properties of the material in the photoactive region. Here two primary classes of solar
cell device structures may be defined. The first class incorporates conventional non-carbon-based
semiconductor materials, such as crystalline silicon, that possess strong covalent interactions between
lattice sites and exhibit relatively high dielectric constants. The second class incorporates carbon-based
materials, such as dye molecules and conjugated polymers, which may be processed at low temperature or
from solution. Introducing the mechanism of action in the context of solar energy conversion for this
second class of organic photovoltaic devices relative to their inorganic counterparts is the primary focus in
this chapter. The information is presented with the assumption that the reader has a basic understanding of
chemistry and solid state physics, in so far as to recognize the existence of molecular orbitals
14
and energy
band structures.
15

1.2 The solar photovoltaic industry
While silicon is not the ideal solar cell material, it currently dominates the solar PV market.
Crystalline silicon (c-Si) is an inorganic semiconductor in which the valence band maximum and
conduction band minimum are not directly aligned in k-space, making c-Si an indirect band gap material.
The indirect nature of the band gap in c-Si means that a considerable change in momentum is required for
the promotion of an electron from the highest energy state in the valence band to the lowest energy state in
the conduction band. As a result, the absorption coefficient for c-Si is relatively low in the wavelength
region relevant for solar photon capture (λ = 300 – 1300 nm) and cuts off completely near λ = 1100 nm. To
absorb a significant percentage (~90%) of the incident photons, a 100 µm thick c-Si layer is required,
meaning that minority charge carriers must diffuse on the order of 200 µm in order to be efficiently swept
apart at the pn junction and collected at the external contacts. Consequently, efficient c-Si devices require
material of high purity and high crystal quality.
16
Despite these dubious characteristics, crystalline silicon
in its various forms dominates the solar cell market with a roughly 83% market share.
10
This is largely a
3
result of the maturity of other silicon technologies, such as those of the transistor and microelectronics
industries, that have developed infrastructure and processing techniques required for producing silicon solar
cells with record power conversion efficiency (η
p
= 25.0%). Several alternative high performance
inorganic technologies include III-V materials such as crystalline GaAs (η
p
= 26.4%) and InP (η
p
= 22.1%),
thin film chalcogenide systems such as CuInGaSe
2
known as CIGS (η
p
= 19.4%) and CdTe (η
p
= 16.7%),
and multijunction devices such as GaInP/GaAs/Ge (η
p
= 32.0%) and thin film GaAs/CIS (η
p
= 25.8%).
17
In
total, alternative technologies such as these represent roughly 15% of the global PV market share.
1.3 Organic solar cells
While high efficiency PV solar energy conversion is technologically feasible, the high cost of
solar panels utilizing present systems is an impediment to the technology’s wide deployment.
10
Economic
viability is paramount in establishing a PV platform as a carbon neutral route to meeting the ever-
increasing global energy demand.
18-20
With predicted practical power conversion efficiencies of η
p
= 10-
15%
21,22
and the potential for low cost manufacturing, organic photovoltaic (OPV) devices have recently
garnered considerable attention
23-25
as an energy source that could be cost competitive with fossil fuels.
Significant achievements have been seen over the last 20 years in the field of organic electronics, with the
demonstrated success in commercializing organic light emitting diodes
26
and the advent of high
performance organic field effect transistor devices
27-29
and sensors.
30
However, interest in OPVs for solar
energy conversion on the part of the scientific community began in earnest only within the last decade.
This is illustrated in Figure 1.1 by the low percentage (~ 5%) of reports appearing in peer reviewed
scientific journals pertaining to the topic of ‘organic solar photovoltaics’ relative to all such ‘solar
photovoltaics’ reports through the year 1999. From the year 2000 to present the relative percentage of
organic solar photovoltaic reports has increased to greater than 35% of the total. Correspondingly, the
4

OPV community has demonstrated several noteworthy milestones over the past decade. For a planar
heterojunction architecture device, incorporating copper phthalocyanine (CuPc) and fullerene C
60
,
efficiencies of η
p
= 3.6% at 1.5 suns (1 sun = 1000 W/m
2
)
31
and η
p
= 4.2% at high intensity ~ 4 suns
32
have
been achieved. Incorporating these materials in a multilayer mixed heterojunction archetrecture
33
has lead
to reports of improved efficiency with CuPc/C
60
to η
p
~ 5% and tandem cells
34,35
with efficiency of η
p
= 5.2
– 5.7%. Conductivity doping has recently been identified as a promising route to enhanced efficiencies in
both tandem and single junction cells leading to efficiencies of η
p
~ 4%
36,37
and higher. Notable recent
reports for high efficiency polymer-fullerene composite solar cells
17,38
are in the range of η
p
= 7.4% - 7.9%.
With encouraging findings, rudimentary market analysis,
39
device degradation studies,
25,40
and high
throughput process characterization
41,42
have been performed for OPV solar cells and modules as well.

Figure 1.1. Number of organic solar PV reports as a percentage of total solar PV reports in scholarly peer-
reviewed journals between the years of 1990 and 2010, illustrating the redoubled research effort in organic solar
cells beginning ca. 2000.

5

While these results are very promising, OPV solar energy conversion is presently a relatively immature
technology, with device performance lagging behind that of its counterpart technologies and falling short of
the predictions for practical efficiencies. Thus, further development is required to identify promising new
materials and suppress loss mechanisms for future high performance organic solar cells. The following
sections present current conceptions and perspectives on these latter topics.
1.4 Organic materials and mechanisms
1.4.1 Challenges for OPV
There is strong evidence to suggest that the present limit to improved device efficiency lies in
balancing robust spectral coverage, leading to high current density, with augmented energy level offsets,
leading to enhanced open-circuit voltage (V
oc
). There are a considerable number of photons in the near-
infrared (NIR) and infrared (IR) regions of the solar spectrum that are currently uncollected in a typical
OPV. The short circuit current density (J
sc
, V = 0) could be markedly improved by converting these
photons to electrical charges. However, since obtaining a high maximum output power density (P
max
)

Figure 1.2. Donor materials 1-10 and acceptor materials A and B described in this chapter.
6
corresponding to high power conversion efficiency (η
p
) also requires maintaining large V
oc
, care must be
taken when considering low-energy absorbing materials, so that improved photon collection does not come
at the expense of electric potential. Therefore, it is compelling to explicitly scrutinize the origin of
operational voltage losses in OPV devices
This chapter is intended to introduce a molecular perspective on the present state of the art for
OPV devices as it relates to the information in the following chapters. First, characteristics of archetypal
materials used in OPV devices are described from a photochemical standpoint, highlighting their excitonic
character and electrical behavior. Materials relating to charge carrier collection and transparent electrodes
are introduced. Recent demonstrations of suppressed voltage losses and strategies for controlling exciton
dynamics are related to molecular properties. The scientific and technological implications are introduced.
While, this chapter is meant to emphasize the benefit that examining a wide array of materials for use in
OPV devices can lend to a robust understanding of the involved processes, it is in no way intended as a
comprehensive review of the vast number of materials and device architectures that have been explored.
The molecular materials illustrated in Figure 1.2 are used here to describe the properties of OPVs, however,
there are a large number of both molecular and polymeric materials that have been reported for application
to OPV. The reader is referred to several exceptional recent reviews for more detailed information
regarding materials and OPVs.
23,24,43-49

1.4.2 Excitonic materials
In conventional semiconductor (Si, GaAs, CdTe, etc.) photovoltaic devices, the site of photon
absorption intimately shares its valence electrons with its neighbors in a strong covalent network.
Therefore, the wave function of the resulting excited state can be spatially delocalized over many lattice
sites. The electronic excited state resulting from photon absorption by the condensed phase material is
termed an exciton. The energy required to dissociate this exciton into a free hole and electron is on the
order of a few times kT,
50-52
the thermal energy available at room temperature. Optical excitation of
organic molecules, such as those depicted in Figure 1.2, does not directly generate free electron and hole
pairs. This is because strong covalent bonding exists only intramolecularly, while the local intermolecular
7

interactions in the condensed phase organic material are comparatively weak. The excited state in such a
system is spatially localized, generally on a single molecule. These excitons typically have energies 0.3-1.0
eV below that of the free electron and hole. As a result, OPV device operation often exhibits features
reminiscent of molecular excitation- and charge-transfer reactions, retaining relatively well defined
vibronic features associated with the isolated molecule.
Dielectric constants in organic semiconductors are commonly a factor of four or more lower than
their inorganic counterparts. Thus, the inchoate charge separation induced during photon absorption is
relatively unscreened by the surrounding dielectric. The coulombic attraction between opposing partial
charges of the electronically excited molecule results in an energetic stabilization of the exciton compared
to the free electron and radical anion (electron polaron) or cation (hole polaron) species. Both of these
effects work in concert to produce energetically bound and spatially localized Frenkel-type excitons with
binding energies on the order of 0.5 - 1.5 eV in organic materials.
53
Thus, room temperature optical
excitation in conjugated organic materials leads to excitons, while generally leading to free charges in an
inorganic semiconductor device.

Figure 1.3. Contemporary organic solar cell devices are based on donor-acceptor heterojunction device
architectures. a) Energy level diagram. b) Planar heterojunction configuration. c) Bulk heterojunction
configuration

8
A molecule in its electronically excited state can be a potent oxidizing agent, as well as a potent
reducing agent. The efficient photoinduced charge transfer required for converting solar photons to
electrical or chemical energy may be realized at the interface between an excited-state electron donor (D)
and a well chosen electron acceptor (A). This interface, termed the D/A heterojunction, is the contact point
where a charge transfer event between two different materials can take place. The D/A heterojunction in its
various configurations is the defining feature central to contemporary organic solar cell devices, as depicted
in Figure 1.3. The chemical potential energy gradient in organic D/A OPVs controls the photoconversion
process in the device.
54
As a result, an ongoing research effort has evolved toward developing an
understanding of the specific molecular features that dictate the operational photovoltage of OPV devices.
1.4.3 Photoelectrochemical processes
The fundamental photon to electrical energy conversion in simple D/A systems can be
conceptualized as six photochemical or electrochemical processes. The physical phenomenon of current
generation will be characterized as a set of chemical equilibria as depicted in Figure 1.4, based on the
nature of the transient species at each step. 1) The absorption of a photon leads to a localized exciton with
energy E
00
on either the donor (D*) or acceptor (A*). 2) This exciton diffuses to the donor-acceptor
interface via an energy transfer mechanism (i.e. no net transport of mass or charge occurs). 3) Charge
transfer quenching of the exciton at the D/A interface produces a coulombically bound charge transfer (CT)
state in the form of a geminate diradical donor-acceptor complex D
+
A
-
, sometimes referred to as a CT
exciton. 4) Subsequent separation of the D
+
A
-
pair proceeds to produce fully ionized D
+
(hole) and A
-

(electron) polaron species. 5) Transport of opposite polarity carriers in the donor and acceptor layers
proceeds via self-exchange between localized hopping sites. 6) Electrical contacts facilitate charge
collection in the external circuit, by regenerating the neutral ground state molecular species. The
thermodynamic and kinetic factors associated with each of these chemical processes ultimately determine
the power conversion efficiency of the fabricated device. These processes will be explained in more detail
as reversible chemical reactions with associated reactants, products, and changes in free energy and kinetic
equilibria in the following sections.
9

Here the situation is illustrated for an optically excited donor and an acceptor in its ground state,
however, it is important to note that a similar process can take place with an optically excited acceptor and
a ground state donor. Utilizing photoexcitation of both the donor and acceptor materials is important to
achieving the broadest possible coverage of the solar spectrum.
1.4.4 Photon absorption
Optical excitation in conjugated organic systems involve transitions between orbitals with
substantial wave function overlap. As a result, the absorption features for these materials often possess
exceptionally high oscillator strength. Indeed, the optical and electrical properties of many materials used
in OPVs have implications for a diverse set of disciplines due to their intense optical transitions and good

Figure 1.4. Photocurrent generation in organic solar cells 1) exciton generation 2) exciton diffusion 3) charge
transfer state generation 4) charge separation 5) charge transport 6) charge collection.
!
1) h" +D
# $ $
$ % $ $ D
*
2) D
i
*
+D
j
# $ $
$ % $ $ D
i
+D
j
*
3) D
*
+ A
# $ $
$ % $ $ D
+
A
&
4) D
k
+
A
k
&
+D
l
+ A
l
# $ $
$ % $ $ D
k
+ A
k
+D
l
+
+ A
l
&
5) D
l
+
+D
n&l
# $ $
$ % $ $ D
l
+D
n&l
+
6) D
+
+ M
a
0
# $ $
$ % $ $ D + M
a
+
|A
&
+ M
c
0
# $ $
$ % $ $ A + M
c
&
10

carrier conductivities. Among the most widely studied are the metal phthalocyanines, reported for use in
nonlinear optics,
55
as sensitizers for photodynamic therapy,
56,57
organic thin film transistors.
28,29
In a
related, but non-optoelectronic application, copper phthalocyanine is used as the pigment in many blue car
paints, illustrating its high absorptivity and environmental stability.
58
This means that an extremely thin
film can still absorb a substantial fraction of the incident solar photons. As will be discussed, this is
important since photon absorption doesn’t directly produce free carriers, the excitons have to diffuse to the
D/A interface and low mobility charges need to be collected at opposing terminals.
For many π-conjugated organic molecules the oscillator strength of electronic transitions can be
very high compared with conventional inorganic semiconductors. As a result a relatively thin film of

Figure 1.5. Solar photon flux Φ (derived from ASTM G173-03 AM1.5G spectral irradiance) is plotted as a
function of wavelength in the top panel. The absorption coefficient α for the active region of an archetypal
organic donor/acceptor pair, CuPc/C
60
(solid red trace, Organic D/A), is compared in the intermediate panel with
that of gallium arsenide (dashed black trace, GaAs), a direct band gap semiconductor and both are contrasted with
crystalline silicon (dotted blue trace, c-Si), an indirect band gap semiconductor. Plotted in the bottom panel, for
an active layer thickness of x = 300 nm, is the cross-sectional number of photons captured N
p
= Φ – Φ[exp(-αx)]
in a single-pass optical path.
11
molecules can absorb a substantial fraction of the incident solar photon flux (Φ). This is demonstrated in
Figure 1.5 by the absorption coefficients (α) for an archetypical organic D/A pair active region, CuPc/C
60
,
compared with gallium arsenide (GaAs),
59
a direct band gap material, and crystalline silicon (c-Si),
60
an
indirect band gap material, both of which are commonly used in conventional inorganic solar cell devices.
Consequently, calculating a cross-sectional capture number for incident photons as N
p
(λ, x) = Φ(λ) –
Φ(λ)[exp(-αx)], where x is an active layer thickness of 300 nm, it is shown that at 630 nm, near the peak of
the solar spectral irradiance, the organic D/A pair captures greater than 96% of the incident solar photons,
while GaAs and c-Si only capture 73% and 10% respectively. Note that for the organic D/A pair, this
estimate is rather conservative, as it does not consider additional photon capture due to back reflection.
Additionally, promising molecular materials exist with substantially higher α

compared with CuPc/C
60
.
61

Nonetheless, from Figure 1.5 one may conclude that OPV active layer thicknesses on the nanometer length
scale are sufficient for efficient photon capture, in contrast to active layer thicknesses of several microns
required for c-Si. Thus, with a significantly thinner film that absorbs the same fraction of light, far less
material is necessary for OPV devices, making the energy and charge conduction requirements less
stringent.
1.4.5 Exciton diffusion
Due to strong vibronic coupling, excitation energy absorbed in excess of the energy of the lowest
vibronic state (E
00
) is transferred as heat to the surrounding medium as the site relaxes to the lowest energy
exciton. The spectral bands associated with this excitonic state tend to be broadened and bathochromically
shifted, while retaining similar vibronic structure to the molecular species. For simplicity the excitonic
species will be considered here as a molecular excitation, with energy E
00
numerically equal to the thin film
spectral intersection of excitation and emission.
In the general case, any molecule in the device may undergo direct optical excitation, however
efficient charge transfer only occurs where there is direct electronic coupling between the donor and
acceptor. This means that the energy absorbed by a donor molecule not intimately coupled or adjacent to
an acceptor molecule by direct orbital overlap must be transferred through the medium to the D/A
12
heterojunction in order to contribute to the photocurrent. This energy transfer process, called exciton
diffusion, may occur via a dipolar mechanism or an electron exchange mechanism. The length over which
excitation can propagate prior to decay of the exciton population to 1/e (roughly 35%) of its initial value is
called the exciton diffusion length (L
D
). Ideally, the fractional solar photon flux N
p
(λ, L
D
)/Φ(λ) absorbed
within L
D
of the D/A heterojunction will be equal to one. In practice, however, this can be as low as 0.1,
due to short exciton diffusion length in many organic materials. This means that in a film thick enough to
absorb nearly all of the incident photons, 90% of the resulting excitons will not reach the D/A interface to
undergo charge transfer. As a result, researchers have attempted to circumnavigate this problem by
developing bulk-heterojunction architectures consisting of interdigitating nanoscale phase-segregated donor
and acceptor regions. As illustrated in Figure 1.3, such a structure is attractive for ensuring that excitons are
formed within L
D
of the D/A interface. The photocurrent generated by such bulk-heterojunction devices
can be substantial, however, their performance is directly linked to local heterogeneity on the nanometer
length scale. The characterization and control of these features is currently a major limitation to bulk-
heterojunction device performance.
62,63

Both mechanisms of exciton diffusion, coulombic and electron exchange, are current topics of
scientific interest. The former results from resonant interaction between transition dipole moments of the
excited and ground state molecules and can occur on length scales considerably larger than the sum of their
van der Waals radii. This form of energy migration is known as Förster resonant excitation transfer
(FRET) and is the primary mechanism for singlet exciton diffusion in the OPVs. The FRET process can be
visualized as a direct extension of the coulombic interaction between the electric field of a photon and the
π-system of a molecule. The elementary rate for this process may be expressed as k
FRET
= (8.8×10
-28
/
n
4
τ
o
r
6
) jΚ
2
mol, where n is the index of refraction of the medium, τ
0
is the radiative lifetime of the energy
donor, r is the intermolecular separation, j is the spectral overlap integral, and Κ is an orientation factor.
Since the FRET rate goes as 1/τ
0
the rate of energy transfer for triplet excited states (τ
0
> 1 µs) via the
Förster mechanism is generally very small. In practice, predicting exciton dynamics using the above
expression is complicated by conformational distortion, molecular disorder, and the presence of trap states
13
in the structures used in OPV devices.
64-66
Consequently, there is significant interest in developing accurate
methodologies for probing exciton dynamics
65,67
in OPV structures.
44,68-75

The other non-trivial mechanism for exciton diffusion arises from electron exchange between a
chromophore in its electronically excited state and a ground state molecule in close proximity. This
process, known as Dexter excitation transfer (DET), occurs through direct wave function overlap and is,
therefore, limited to length scales on the order of the van der Waals radii of the two molecules. Since τ
0
for
the triplet excited state in conjugate organic molecules can be in the microsecond to millisecond regime,
DET is the primary mechanism for triplet exciton diffusion. Although the formal rate dependence for DET
is proportional to the spectral overlap integral and attenuates as k
DET
∝ exp(-2r), it contains terms that
cannot be easily related to physically measurable quantities. Therefore, it is difficult to adequately address
even the elementary rate of triplet exciton diffusion in the general case and the utility of triplet exciton
diffusion in OPVs is a current area of interest.
76-78
Section 3.3 of this work is devoted to the development
of model systems for investigating and utilizing triplet excitons in OPV structures.
1.4.6 Charge transfer
Following exciton diffusion to the D/A interface, efficient charge transfer quenching of the excited state
must occur in order to yield the desired photoresponse. In principle, the thermodynamic requirements for
this process are straightforward, the energy of an exciton interacting with a charge acceptor must be greater
than the CT state energy. In practice, however, assigning enthalpic quantities associated with the lowest
unoccupied molecular orbital (LUMO) in condensed phase organic materials is often complicated by
sample degredation.
79,80
Additionally, the physics of semiconductor devices has historically been discussed
in terms of band models for covalent crystalline materials, where the free electron approximation is
appropriate, which is not the case for most organic systems. Terms like conduction band, LUMO, electron
affinity, transport level, and optical LUMO are often inappropriately taken to be interchangeable.
Significant attention
2,81-92
has been given to developing a physically relevant energy level description for
organic electronic materials, where it is more appropriate to discuss weakly interacting molecules in
condensed phase low dielectric media.
14
Of particular importance to formation of the CT state is the energy of the excited state. When a
molecule at the D/A interface becomes electronically excited by FRET, DET, or photon absorption, it may
take part in a charge transfer reaction with the adjacent charge acceptor molecule. The thermodynamic
requirements to form the geminate CT state species [D
+
A
-
] at the D/A interface can be understood by
considering the change in enthalpy for the photoinduced charge transfer reaction D* + A → [D
+
A
-
]. The
ionization energy of this excitonic state (E
i
*), sometimes referred to as the ‘optical LUMO,’ is an extremely
useful quantity for assessing whether forward electron transfer will be exothermic. Succinctly, E
i
* is the
oxidation potential of the excited state [(D*)
0/+
], which can be used to determine which acceptors will
engage in efficient charge transfer with D*. A value for E
i
* may be obtained from the combination of
ultraviolet photoelectron spectroscopy (UPS) and UV-vis absorption/emission spectroscopy by numerically
subtracting -E
00
from the ionization energy of the neutral ground state species, E
i
* = E
i
– (-E
00
). If the
resulting value of E
i
* is greater than the electron affinity (E
a
) of the accepter, then the photoinduced
forward electron transfer reaction will be exothermic. A similar calculation of an acceptor exciton’s
electron affinity (E
a
*) may be made to assess whether acceptor exciton dissociation will be exothermic.
Thus, one can also define an ‘optical HOMO’ for OPV acceptors from the electron affinity and E
00
, which
can be used to assess which donors will be appropriate for dissociation of an exciton in a given acceptor.
Although a cogent kinetic description for the charge generation process is still developing, during
exciton dissociation, some generalized charge transfer (CT) state
81
is thought to form, characterized by D
+

and A
-

polaron species. Very little screening of these charged species takes place, since the dielectric
constant, ε
m
≈ 3, of the surrounding medium is relatively low, and the attractive electrostatic potential
energy E
c
= q
2
/4πε
0
ε
m
r between the geminate D
+
A
-
pair is on the order of 25 kJmol
-1
for a typical geminate
pair spacing, r = 20 Å. This means that the coulombic stabilization of the CT state is roughly 10 times
greater than kT at room temperature and, if not efficiently surmounted, can severely limit the output power
of OPV devices by markedly reducing the fill factor (Section 4.).
15
1.4.7 Charge separation
Once formed, the CT state may decay through one of two basic reversible pathways, excluding
photochemical degradation. In the first, regeneration of some neutral species (excited state or ground state)
may occur. This process, referred to as recombination because it involves combining separated charges,
can be a major loss mechanism. The second is the preferred pathway for CT state decay in which a series
of electron self-exchange reactions occur between the radical cation (anion) of the CT complex and the
surrounding neutral ground state donor (acceptor) molecules. A major experimental challenge in this area
is the development of ultrafast two-dimensional optical spectroscopies with interfacial specificity to
reliably characterize the electronic coupling and CT state dynamics for a broad range of materials and
architectures.
93-96

1.4.8 Charge transport
The weak intermolecular electronic coupling that exists in conjugated organic materials acts, not
only to localize their excitons, but their charge carriers as well. Consider two photogenerated charges, each
100Å from the D/A interface. Since at this distance E
c
= q
2
/4πε
0
ε
m
r ≈ kT, these charges may be regarded as
coulombically separated in a typical organic semiconductor. Thus, the energy of the hole may be crudely
approximated as the E
i
of the donor and the energy of the electron as E
a
of the acceptor. Since there is
minimal extended band structure in these materials, transport of charge proceeds via a localized hopping
mechanism characterized by a radical ion that polarizes the surrounding neutral molecules. The rate of this
hopping process may be expressed
97,98
in terms of nonadiabatic Marcus theory for electron transfer to
account for the electronic coupling of the polarized neutral ground state species to the radical ion and the
internal reorganization energy required for this pair to reach the same geometry. The amorphous nature of
the organic materials used in OPVs gives each molecule a unique environment, leading to inhomogeneities
in the local field experienced by each molecule. This inhomogeneity leads to carrier trapping sites in these
amorphous materials, decreasing the charge transport rates below those given by a Marcus-type treatment
16
of intermolecular charge transfer. The use of crystalline materials in OPVs is being explored, which would
eliminate carrier traps of this type and improve OPV performance.
90,97,99,100

1.4.9 Charge collection
Photovoltaic solar energy conversion requires efficient delivery of photogenerated charges to an electrical
load resistance in the external circuit. This is generally accomplished in the case of OPV devices by
sandwiching the absorbing materials between a substrate-supported transparent electrode, such as tin-doped
indium oxide (ITO) or fluorine-doped tin oxide (FTO), and a metal counter electrode, deposited by thermal
evaporation. While ITO is currently the dominant transparent electrode technology for D/A OPV devices,
its ultimate utility is questionable. Several promising alternative transparent electrode technologies are
discussed in Section 5.4. Notably, the performance of flexible graphene-based electrodes and OPV devices
are found to be competitive with that of ITO. The nature of the electrical contacts employed in device
preparation is largely determined by interface states between the electrode and the organic materials and
can strongly impact device performance. Properties of this organic/electrode interface are an area of
substantial interest since the density of states profile in this region of the device can be quite complex due
to interface chemical and polarization effects.
82-84,101

Even when a system meets the thermodynamic requirements for charge collection, there is some
evidence to suggest that molecular electron transfer kinetics can play a dominant role in determining charge
collection efficiency.
102
The impact of electron transfer kinetics on charge carrier collection is detailed in
Chapter 4, with regard to the electron-collecting buffer layer between the acceptor and cathode of lamellar
OPV devices. This layer has the potential to act in multiple capacities, such as in local optical field
enhancement or in defect passivation for large-area devices. However, the properties that govern charge
collection in this layer are still poorly understood on the molecular level. Thus the molecular properties
required for efficient charge collection were examined by replacing the typical bathocuproine buffer layer
with neat films from a series of six substituted tris(β-diketonato)ruthenium(III) analogues in the lamellar
OPV structure, indium-tin-oxide/ copper phthalocyanine/ fullerene/ buffer/ Ag. The six ruthenium
complexes have frontier orbital energies that vary by nearly 1.7 eV, measured using ultraviolet
17

photoelectron spectroscopy (UPS). At 100 Å buffer thicknesses, the ruthenium-based devices all exhibit
power conversion efficiencies of 1 ± 0.3% - 1.4 ± 0.3% and little performance variation, indicative of metal
mediated carrier collection. However, strongly buffer material dependent inflection points are observed in
the current-density voltage dependence (cf. Section 1. of OPVs with 200 Å buffer layers, suggesting the
onset of a reciprocal carrier collection regime. Simple anti-polar diode circuit analysis suggests that
molecular electron transfer rates at the acceptor/buffer interface play a crucial role in determining charge
collection efficiencies. The general implication for the design of organic optoelectronic devices being that
a molecular kinetic approach is a feasible route for enhancing device performance beyond what may be
expected based on orbital energies alone.

Figure 1.6. Simulated solar cell electrical behavior in the dark (dotted lines) and under illumination (solid lines)
comparing the effect of the saturation current parameter J
s
on V
oc
. The sharp inflection points in the semilog plots
(upper panel) are the points where the current switches form positive to negative. The ‘Large Dark Current’ trace
in black represents J
s
× 10
6
that of the ‘Small Dark Current’ trace in red. Also illustrated are P
max
(filled gray
rectangle) and the numerical product J
sc
× V
oc
(unfilled gray rectangle) used to calculate FF = P
max
/(J
sc
× V
oc
).

18
1.5 Electrical behavior
Although the operation principles of OPV devices are inherently dissimilar to a typical p-n
junction device, the electrical characterization procedures to obtain their performance metrics are the same
as illustrated in Figure 1.6. In this section electrical measurement and analysis are discussed in brief. The
basic electrical measurement consists of connecting the device to an external power source, applying a DC
voltage (V), and measuring the electrical current density (J). Generally, it is desirable to perform such a
measurement while sweeping the applied voltage from V < 0 (reverse bias) to V > 0 (forward bias). The
electrochemical implication of this formalism is that under reverse bias, the organic material is reduced at
the anode and oxidized at the cathode, while under forward bias, the organic material is oxidized at the
anode and reduced at the cathode.
In the dark, the device behaves as a diode, exhibiting saturated current density (J
s
), flowing with
limited voltage dependence under reverse bias, while an exponential rise (rectification) in current density is
observed beyond some applied forward bias. For a relatively simple case involving no resistive losses, the
voltage dependence upon rectification and the magnitude of this saturation current density (J
s
) under
reverse bias are intimately related according to J = J
s
exp[(qV/nkT)-1], where q, k, and T are the elementary
charge, Boltzman’s constant, and temperature, respectively. Ideality factors of n ≈ 2 are common for
organic devices and are associated with current voltage relationships dominated by recombination.
15
While
this relationship is an oversimplification, it demonstrates that J
s
is a defining metric, characterizing the
voltage-dependence of the current density under both forward and reverse bias. For the former, Figure 1.7
represents the processes leading to J
s
. The magnitude of J
s
will depend in most cases on the rate (k
inj
) of
carrier injection from the electrodes, charge mobility of the materials, and the rate (k
rec
) of net charge
recombination at the donor/acceptor interface (D
+
+ A
-
→ D
0
+ A
0
). Note that D
0
and A
0
need not be
ground state species. This metric is integral in determining the magnitude of the open-circuit voltage.
19

Upon illumination, the photogeneration processes discussed in section 3 lead to an increase in the
density of D
+
and A
-
in the device. When a reverse bias is applied at the external contacts, conversion of
the photogenerated D
+
and A
-
ions back to the neutral species results in the measured photocurrent density
under illumination depicted in Figure 1.6. Again, for the ideal case, under illumination the voltage
dependence of the current density is given by J = J
s
exp[(qV/nkT) - 1] - J
ph
, where J
ph
is the photogenerated
current density. At zero applied bias J = - J
ph
and the corresponding short circuit current density (J
sc
) is
shown in Figure 1.6. From the above J(V) dependence, the open circuit photovoltage may be expressed as
V
oc
= q
-1
nkT ln(J
sc
/J
s
+1) since J = 0. For a simplified case this is the most basic photovoltage expression.
For n ≈ 2, the magnitude of the photovoltage depends simply on a ratio of rates, where J
sc
represents how
frequently photogenerated charges are collected and J
s
represents how frequently charges recombine.
In practice, resistive effects and physical aberrations can complicate device characterization.
Generally, one may estimate the importance of these non-idealities by modeling device data according to

Figure 1.7. Non-radiative recombination losses occurring under forward bias in a typical OPV device. Holes
injected from the anode Fermi level (E
F,A
) into the HOMO level (E
i
) of the donor and electrons injected from the
cathode Fermi level (E
F,C
) into the LUMO level (E
a
) of the acceptor are transported to the D/A interface.
Coulombic attraction between holes and electrons yields the D
+
A
-
CT state with energy E
CT
. Charge
recombination reaction D
+
+ A
-
→ D + A occurs with rate constant k
rec
.
20

an equivalent circuit as shown in Figure 1.8. Incorporating all series resistive elements as R
s
, and all
parallel resistance as R
p
one obtains the expression:
(1.1)
for the current density as a function of voltage under illumination. Developing a physically relevant
interpretation for the commonly observed phenomenological J-V behavior
32,44,103
represented by Eq. 1.1 is
an active area of OPV device research.
104-110
Methods are presented in Chapter 4 on adapting Eq. 1.1 to
account for commonly observed departures from its predicted exponential current dependence on applied
voltage.
In accord with Eq. 1.1, at voltages 0 < V < V
oc
, a typical photovoltaic device under illumination
supplies power density (P = J × V) to the external circuit (cf. lower panel of Figure 1.6, dashed trace in first
quadrant). The point of maximum output power density is denoted by P
max
in Figure 1.6 and corresponds
to the area of the filled gray rectangle in quadrant four of the lower panel. A useful quantity for comparing
the maximum output power relative to the photocurrent and photovoltage produced by any particular
device is the fill factor, given by FF = P
max
/ J
sc
× V
oc
. The FF may be visualized as in Figure 1.6 as the
fractional area of the unfilled gray rectangle (i.e. peak values for J and V, but available power at these
points = 0) occupied by the filled gray rectangle. Typical FF values for OPVs range from 0.3 to 0.7.
Finally, the power conversion efficiency (η
P
) is calculated as the numerical quotient of P
max
and the total

Figure 1.8. Single diode equivalent circuit model commonly employed in estimating solar cell losses.

21
integrated spectral irradiance (E
Total
), giving η
P
= P
max
/ E
Total
. Instrumentation and analysis techniques
common to the characterization of OPV devices are described in Chapter 2 of this work. Note that precise
standards of measurement and calibration have been developed, including spectral mismatch correction, for
accurately reporting η
P
. The active researcher is admonished to diligently adhere to them, as outlined in the
literature,
111-114
whenever reporting OPV performance metrics.
1.6 Open-circuit voltage
In the preceding sections, general operational and electrical principles of OPV devices were introduced.
This section is devoted to developing a molecular interpretation of the properties that influence the
operational photovoltage in OPV devices. Considering photons near the peak of the solar spectrum, ~ 2.0
eV (~ 600 nm) are typically efficiently absorbed in OPV devices, but common values for qV
oc
are only in
the range of 0.5 - 1.0 eV, device efficiencies may be substantially improved by understanding the origin of
photovoltage losses in these devices.
Furthermore, a substantial photon flux exists below 2.0 eV in the solar spectrum. Without
understanding the origin of the photovoltage losses, developing materials to efficiently harvest these low
energy photons may ultimately lead to an efficiency limiting trade-off between enhanced photocurrent and
enhanced photovoltage. Consequently, it is compelling to develop a molecular description of the operation
open-circuit voltage in OPV devices to understand the losses in an electrochemical context. By simply
eliminating voltage losses, future efficiencies 2-5 times that of present devices may be achievable.
Based on early observations of the photovoltaic effect in laminated organic systems,
115
the first
modern bilayer OPV device
8
was noted to exhibit a photovoltage that was only weakly dependent, in both
magnitude and polarity, on the nature of the electrical contacts. Rather, the reported V
oc
was dictated
largely by the nature of the organic/organic interface. This observation lead to the later proposal
116
that the
V
oc
is determined by the offset in frontier orbital energies for the removal of an electron from the HOMO of
22

the donor and placing it in the LUMO of the acceptor. Consequently, data has been generated which shows
a correlation of the frontier molecular orbital energy levels with device performance, particularly V
oc
.
21,117-
120
Nevertheless, structural variations on the molecular level are known to substantially impact electron
transfer kinetics in chemical
12,121,122
and biological
123,124
systems and the voltage produced by solution-
based photoelectrochemical cells has also been linked to charge transfer rates.
125,126
Thus, the key
efficiency improvements, required for OPVs to become a truly disruptive technology, may lie in the
detailed kinetics of charge generation and recombination, which, at this time, are poorly understood for
multi-component thin film systems.
1.6.1 Thermodynamic description of open-circuit voltage
The most general description for the open-circuit voltage limit in any photovoltaic device has been
addressed by Gregg.
54
According to the free energy of the system, the obtainable photovoltage will depend
on both the electrical and chemical potential energy gradients in the device. However, the equilibrium
charge carrier concentrations and carrier mobilities in most heavily doped inorganic PV devices are

Figure 1.9. Equilibrium energy diagram for a pn junction in an inorganic semiconductor material with intrinsic
Fermi energy E
Fi
, conduction band energy E
c
, valence band energy E
v
and with potentials φ
Fp
and φ
Fn
given by the
respective impurity concentrations in the p and n regions. The quantity V
bi
represents the total built-in electrical
potential due to band bending.

23

extremely high compared with organic molecules. This means that the effect of the chemical potential in
classic silicon-based cells is negligible. Indeed, the V
oc
limit for typical inorganic photodiodes is
understood to be V
oc
≤ V
bi
,
127-129
where V
bi
is the ‘built-in’ electrical potential energy difference across the
pn junction as illustrated in Figure 1.9.
This canonical description for the limits of silicon solar cells led to early efforts to develop a
similar narrative for the maximum phenomenological photovoltage of OPV devices. As a result, estimates
for V
bi
have been made by measuring the open-circuit voltage produced from composites of various
conjugated polymers
117
and derivatized fullerenes.
116
These results demonstrate a clear correlation between
the observed V
oc
and electrochemical reduction potentials of the fulleroid electron accepting materials.
Thus, the magnitude of the open-circuit voltage appeared to be determined by the offset (ΔE
DA
) between the
ionization energy (E
i
) associated with donor HOMO level and the electron affinity (E
a
) associated with the
acceptor LUMO level as depicted in Figure 1.10.

Figure 1.10. Energy level diagram for a typical D/A heterojunction depicting donor material with ionization
energy of E
i
and excited state ionization energy of E
i
*, acceptor with electron affinity of E
a
and excited state
electron affinity of E
a
*, and the interfacial energy level offset between E
i
and E
a
illustrated as ΔE
DA
, all relative to
vacuum as described in the text.
24
1.6.2 Orbital energy level offsets
Previous researcher has utilized the concept of modifying ΔE
DA
to increase the cell voltage. It is
important here to reiterate that employing new materials with high oscillator strength transitions
overlapping the solar spectrum, while yielding high voltage devices, is paramount for improving power
conversion efficiency. This was demonstrated in the case of boron subphthalocyanine chloride (SubPc),
130

when compared as a donor material against a prototypical standard CuPc control in optimized cells having
the planar lamellar structure ITO/Donor/C
60
/BCP/Al. Possessing a delocalized 14 π electron system, the
general class of subphthalocyanines are lower macrocyclic homologues of the parent 18 π electron
phthalocyanines and are composed of three nitrogen bridged isoindole units around a central boron core.
131

As a consequence of its smaller π-system, SubPc exhibits a more negative E
i
value compared with CuPc a
hypsochromic shift of its Q-band absorption. Representative J(V) curves for both the CuPc and SubPc
based devices demonstrate that for the standard CuPc control device, a V
oc
of 0.42 V was obtained under 1
sun illumination, while the SubPc device produced a V
oc
of 0.97 V. Although the CuPc device exhibits a
broad spectral absorption, from the J(V) dependence, the photocurrent produced by both the SubPc (J
SC
=
3.36 mA/cm
2
) and CuPc (J
SC
= 3.07 mA/cm
2
) devices are comparable. This stems from the narrow, but
intense SubPc Q-band absorption at λ = 590 nm that compensates for the low-energy broadband absorption
exhibited by CuPc. Since the FF values for both the SubPc test device and CuPc control device are
virtually identical, the resulting power conversion efficiency (η
p
) more than doubles, increasing from 0.9%
for CuPc to 2.1% for SubPc.
The demonstration that device performance enhancement in excess of 100% can be achieved by
maintaining substantial optical absorption, while increasing the V
oc
, represents a significant proof-of-
concept and emphasizes the importance of understanding the origin of photovoltage losses in OPV devices.
Further comparisons supporting the validity of tailoring the ΔE
DA
value to control V
oc
have been made by
incorporating boron subnaphthalocyanine chloride (SubNc)
132,133
as a donor material. The SubNc devices
exhibit lower photovoltages than the corresponding SubPc cells, when tested under equivalent light
25

intensities. This has been rationalized on the basis of HOMO destabilization upon benzannulation. Thus,
the reduced photovoltage is attributed to lower offset energy at the SubNc/C
60
interface.
1.6.3 Temperature dependence of photovoltage losses
From the expression V
oc
∝ ln(J
sc
/J
s
), the photovoltage will depend logarithmically on light
intensity, until reaching some saturation voltage. Since J
s
accounts for thermal activation,
1,21,104
according
to J
s
= J
so
exp(-ΔE
DA
/γkT), where the frequency factor J
so
and γ are phenomenological terms that must be
assessed experimentally, an inverse linear relationship between V
oc
and temperature is expected within a
given temperature regime. Beyond this the V
oc
again tends to saturate at a particular maximum voltage.
The notable demonstrations of efficiency enhancement for materials combinations with large ΔE
DA
values
produced further impetus to quantitatively explore the relationship between temperature and photovoltage.
One such study examined the V
oc
dependence on temperature for CuPc, pentacene, and N,N’-di-[(1-
naphthyl)-N,N’-diphenyl]-1,1’-biphenyl)-4,4’-diamine (NPD) donor materials paired with various acceptor
materials,
21
Significantly, select materials appear to exhibit temperature independent photovoltage

Figure 1.11. Temperature dependence of V
oc
for CuPc and pentacene donor-based devices compared with the
relative temperature independence of V
oc
≈ 0.9 V for the NPD based device. Reprinted figure with permission
from B. P. Rand, D. P. Burk and S. R. Forrest, Phys. Rev. B, 75, 115327, 2007. Copyright 2007 by the American
Physical Society
26
response, as evidenced by the NPD/C
60
trace in Figure 1.11. While CuPc and pentacene possess planar
geometrical structures, NPD exhibits a slightly distorted geometry due to steric effects. A similar effect
was also observed for devices fabricated with SubPc,
21
which exhibits a cone-like distortion due to
intramolecular ring strain., This suggests that for these two materials combinations, room temperature loss
mechanisms are inherently minimized.
1.6.4 General photovoltage description
While substantial photovoltage enhancement has been demonstrated by increasing ΔE
DA
, such an
approach represents a significant challenge for the prospect of harvesting low energy solar photons to
produce high efficiency devices. Again, this is because both high voltage and high current are required,
assuming facile charge separation, to realize gains in power conversion efficiency. In Figure 1.12 the
thermodynamic balance required for enhancing the open-circuit voltage, while achieving broad absorption
overlap with the solar spectrum is depicted. Correspondingly, there is an efficiency limiting trade-off
between robust photon capture and minimal photovoltage losses.
Recent results have revealed the prospect that voltage losses may also be minimized according to
the kinetic facility of excited state charge transfer quenching relative to charge recombination. For this
reason it is compelling to return to the general photovoltage treatment presented by Gregg, based on the
electrical and chemical potential energy gradients in the device. From the V
oc
description that emerges, one
may expect molecular and architectural features that kinetically or thermodynamically suppress J
s

selectively, will enhance the operational V
oc
.
As mentioned previously, a significant research effort has gone into addressing the relationship
between the built-in electrical potential (V
bi
), and the V
oc
produced by OPV devices. However, direct
electroabsorption measurements of V
bi
across simple devices consisting of a single organic layer, inserted
between two energetically asymmetric contacts, have demonstrated that the relationship at room
temperature between V
bi
and the observed V
oc
is not straightforward,
134
due to various loss mechanisms.
Moreover, the photovoltaic effect has even been observed for organic molecular systems with no intrinsic
asymmetry,
135-137
hence no built-in electrical
27

potential. From these observations it is clear that the V
oc
in OPV devices is not necessarily limited to any
inherent built-in electrical potential. Importantly, the proposed explanation for the observed photovoltage
in such systems relies on the interfacial kinetics
135
of charge generation and recombination.
While simple single layer symmetric device architectures have not presently produced high power
conversion efficiencies, demonstration that kinetic factors likely play a major role in determining the open-
circuit voltage losses in all OPV devices cannot be overemphasized. For the synthetic chemist, the potential
implication is an additional molecular handle in the design of new materials combinations that
preferentially favor augmented J
sc
, while suppressing J
s
. The following sections outline specific molecular
examples relating to the characterization and development of such materials.

Figure 1.12. Schematic representation of two hypothetical donor materials embodying the efficiency limiting
trade-off between suppressed voltage losses on the left and robust spectral coverage on the right. Donor 1
exhibits a large ΔE
DA
and is expected to produce a large open-circuit voltage, whereas Donor 2 exhibits a small
ΔE
DA
, but offers enhanced coverage of the solar spectral irradiance.

28
1.6.5 Molecular electron transfer kinetics
The conventional wisdom gleaned from V
oc
saturation studies and PES data suggests that when
recombination losses are suppressed, qV
oc
is limited thermodynamically by ΔE
DA
. However, at room
temperature or low incident photon flux, the magnitude of the operational photovoltage of many devices
can be markedly reduced through recombination losses. As such, discerning the origin of these losses has
become an important area of study. It has proven fruitful to electrically represent the ensemble averaged
carrier flux as discussed according to Eq. 1.1 for the equivalent circuit model in Figure 1.8. Details on this
type of analysis may be found in the literature and the interested reader is directed to several outstanding
recent reviews and articles.
104,105,108,109,138-140

It has already been mentioned that when n ≈ 2, as it is for many OPV devices, the saturation
current density under forward bias arises from charge recombination.
15
In this regime the satuaration
current may be expressed as
(1.2)
where q is the elementary charge, k
rec
is the rate constant for the recombination reaction D
+
+ A
-
→ D
0
+
A
0
, and [D
+
] and [A
-
] are, respectively, the concentration of ionized donor and ionized acceptor molecules
at the D/A interface as depicted in Figure 1.7. Here, bulk recombination is implicitly neglected. Solving
Eq. 1.1 for V(J = 0) = V
oc
and substituting for J
s
, the photovoltage may be expressed as
(1.3)
(1.4)
(1.5)
29
where the approximations become valid as , , and J
ph
» J
s
. Now, according to Marcus
theory
121
for outer sphere electron transfer, the Arrhenius behavior for bimolecular recombination may be
expressed as
122

(1.6)
where h is Planck’s constant, H
ij
is the electronic coupling matrix element between the initial CT state and
the final neutral state, λ is the free energy for geometric reorganization, and ΔG° is the total free energy
change for the recombination reaction as depicted schematically in Figure 1.13.
This expression constitutes a unified description for the voltage losses in OPV devices, accounting
for the bimolecular interaction at the D/A interface, as well as the facility of charge injection and transport
throughout the rest of the devices. The magnitudes of the [D
+
] and [A
-
] terms in Eq. 1.5 will depend on the
charge injection rate k
inj
at the electrodes (Figure 1.7), and on the mobility of holes (µ
p
) in the donor and
electrons (µ
n
) in the acceptor. Materials and architectures that minimize [D
+
] and [A
-
] will maximize V
oc
.
For charge recombination involving no exciton generation, Eq. 1.6 shows that k
rec
goes as the
square of the electronic coupling of the CT state to the ground state and decreases exponentially with
increasing reorganization energy. This relationship helps to frame the observed variations in V
oc
with
device architecture, process conditions, molecular orientation, film morphology, and materials properties in
terms of a more conceptually transparent description for actual electrochemical processes that occur during
device operation. The implication of Eqs. 1.5 and 1.6 for molecular design are, materials and architectures
that minimize H
ij
and maximize λ will maximize V
oc
. Thus, the kinetics of molecular electron transfer
forms the basis for assessing and minimizing voltage losses in organic photovoltaic devices. Note that, as
depicted in Figure 1.13, non-radiative charge recombination, following photoinduced charge transfer,
occurs in the Marcus inverted region, with free energy of activation ΔG*.
12,141
This is the origin of the long
lived photoinduced CT states that make OPV device operation possible.
30

1.6.6 CT state coupling, molecular orientation, and V
oc

The charge recombination rate in Eq. 1.6 during outer sphere electron transfer is related to the π–π
interaction of the donor and acceptor species through H
ij
. The application of modern computational
chemistry to this interfacial kinetic process will very likely be invaluable for developing the next generation
of low-loss V
oc
materials. Comparing excited-to-CT state coupling with ground-to-CT state coupling is
particularly interesting. Recently, such calculations were carried out for an archetypical D/A pair,
pentacene/C
60
.
142
Importantly, the results demonstrate that, based on geometric orientation, suppression of
charge recombination by decoupling the ground-to-CT and triplet-to-CT state interactions is not mutually
exclusive with respect to efficient photoinduced charge transfer. This is exemplified by the parallel
orientation 2B, depicted in Figure 1.14, that exhibits the most rapid rate for photoinduced charge transfer,
both from pentacene to C
60
, (
1
B
1u
P
⊗
1
A
g
C60

→
2
B
2g
P+
⊗
2
T
1u
C60-
) and from C
60
to pentacene (
1
A
g
P
⊗
1
T
1g
C60
→

Figure 1.13. Free energy curves illustrating the relationship between the free energy of activation ΔG*,
reorganization energy λ, the free energy of reaction ΔG°, and the coupling element H
ij
that determine the electron
transfer rates in an OPV device.

31

2
B
2g
P+
⊗
2
T
1u
C60-
). However, the back recombination reaction for orientation 2B is suppressed, such that the
rate for recombination from the CT
0
state to the ground state (
2
B
2g
P+
⊗
2
T
1u
C60-
→
1
A
g
P
⊗
1
A
g
C60
) is lower than
orientation 2A. Similar comparisons can be drawn between other molecular orientations, suggesting that
interfacial coupling can, in principle, act to suppress ground-to-CT state recombination, while preserving
excited-to-CT forward electron transfer. The conceptual implication being, it is possible to suppress charge
recombination rates and preserve V
oc
, without reducing J
sc
, simply by deploying a D/A pair with the
appropriate geometrical interaction. This represents a promising result for rationally designing materials

Figure 1.14. Molecular orientation and its influence on calculated charge transfer rates between pentacene and
C
60
, indicating that suppressed recombination and efficient photoinduced charge transfer are not mutually
exclusive. For example, the CT
0
state to the ground state reaction
2
B
2g
P+
⊗
2
T
1u
C60-
→
1
A
g
P
⊗
1
A
g
C60
rate for parallel
orientation 2B (dotted red trace in Parallel plot) is slightly suppressed, while the forward charge transfer rates for
the same orientation (solid and dashed red traces) are an order of magnitude or more greater than orientation 2A
(blue trace). Reproduced from Yi with permission. Copyright 2009 American Chemical Society.

32
combinations with minimal recombination losses. Presently, however, decoupling the ground and CT
states while achieving strong excited-to-CT state coupling, based solely on molecular orientation, is a
formidable challenge.
Experimentally, it has been shown for CuPc that molecular orientation can have an impact on the
energy of molecular electronic states.
92
Moreover, interfacial molecular orientation has also been shown to
have an experimentally quantifiable influence on the interfacial charge recombination kinetics for TiOPc,
CuPc, and PbPc.
91
Molecular orientation has also been observed to influence the efficiency of charge
transfer processes at CuPc/F
16
CuPc interfaces.
143,144
Of course, the implication for device applications as it
relates to open-circuit voltage is that the recombination and injection rates, hence the magnitude of J
s
, are
related to interfacial molecular orientations as predicted from Eq. 1.3. This is to say that controlling the
coupling of the ground state to the CT state may be achieved by controlling the orientation of molecules at
the D/A interface, thus minimizing J
s
via orientationally tailored molecular orbital interactions may be truly
feasible.
It should be noted that care must be taken in addressing interfacial electronic structure, since it
may considerably impact observed macroscopic materials behavior
82,84,101
with respect to the isolated
material. As such, precise characterization of interfacial frontier orbital energies has demonstrated
relationships between molecular properties and electronic interactions for several common OPV
heterojunctions.
85,86,145
In particular, combining ultraviolet photoelectron spectroscopies (UPS) and x-ray
photoelectron spectroscopies (XPS) has helped to characterize the CuPc/C
60
heterojunction contrasted with
that of titanyl phthalocyanine (TiOPc)/C
60
. Commonly used in photoconductivity and electrophotographic
studies, TiOPc may exist in several crystallographic polymorphs with disparate electronic properties,
90,146

depending on process conditions
147
and the nature of the surrounding medium. Thus, detailed UPS and
XPS studies have been invaluable in characterizing the performance enhancement
90
observed in OPV
devices fabricated using TiOPc. Similar studies have been carried out
89,148
using aluminum phthalocyanine
chloride (ClAlPc), a common photosensitizer
56,149
and recently observed to yield photovoltage
enhancements in OPV devices.
150

33

1.6.7 Demonstrated kinetic impact on V
oc

Recently, research in our laboratory has suggested a practical route for decreasing coupling
between ground and CT state D/A pairs, demonstrated by the poignant result of substantially increased
photovoltage for D/A pairs that would not have otherwise been expected from the simple ΔE
DA
model. The
first case presented here addresses the relatively high V
oc
for planar heterojunction devices using
rubrene/C
60
(V
oc
= 0.92 V) compared with tetracene/C
60
(V
oc
= 0.55 V).
1
Both donor materials consist of
four fused six-member aromatic hydrocarbon rings and, although rubrene has phenyl substituents appended
in the 5,6,11, and 12 positions, both molecules exhibit nearly identical ionization energies.
2,3
Steric
hindrance forces the rubrene phenyl rings to twist out of the backbone plane, electronically isolating them
from the fused π-system.
Based on the ionization energies for tetracene and rubrene, values of ΔE
DA
~ 2 eV are estimated
for the D/A heterjunctions formed between C
60
and either donor material. From the thermodynamic model,
the V
oc
of both devices should be comparable. However, it is clear from the J(V) characteristics in Figure
1.15 that the rubrene based device produces a markedly higher photovoltage. Since the pendent phenyl
rings of rubrene lie virtually orthogonal to the tetracene core, intimate intermolecular π-electronic

Figure 1.15. Semi-log scale J-V characteristics in the dark (filled circles) and under illumination (open circles)
for tetracene (black) and rubrene (red) based OPV devices illustrating the substantially higher V
oc
in the case of
rubrene compared with tetracene, despite their similar ionization potentials. Adapted from Perez with permission,
Copyright 2009, American Chemical Society.

34

interaction between adjacent molecules is inhibited. The effect of this reduced coupling can be observed in
the broadened and red shifted absorption bands for tetracene in the neat film compared to solution.
Rubrene, on the other hand, exhibits minimal change in its thin film spectral features compared with
solution. The exposure of the π system in tetracene is expected to lead to increased D/A coupling
compared with the sterically hindered π system of rubrene.
Observations made in tetracene-based versus rubrene-based OPVs and related devices prompted
us to experimentally explore the relationship of Eq. 1.4 in more detail. For devices with ideality factor, n,
close to 2, which is common for OPVs, the saturation current has been expressed as J
s
= J
so
exp(-
ΔE
DA
/2nkT).
21
Substituting this definition of J
s
into Eq. 1.4 and simplifying yields Eq. 7. This clearly
illustrates the relationship of the V
oc
to both DE
DA
and the intermolecular coupling at the D/A interface,
represented by J
so
.
(1.7)
Table 1.1. Dark current and photovoltage for various donor materials
a
Donor V
oc
ΔE
DA
J
so

NPD 0.85 1.9 11
SubPc 0.97 2.0 5.5
Tetracene 0.55 1.6 - 2.0 150
b

Rubrene 0.92 1.8 - 2.0 0.43
b

CuPc 0.48 1.7 1.5 × 10
4
PtTPBP 0.69 1.4 12
a
C
60
acceptor heterojunction with V
oc
in V, ΔE
DA
in eV and J
so
in mA/cm
2
.
1-3

b

Represents lower bound estimate.

35

While rubrene and tetracene have very similar values of ΔE
DA
, their J
so
values differ by three
orders of magnitude (cf. Table 1.1). The relatively isolated π-system in rubrene reduces the coupling at the
D/A interface compared with tetracene. By implication, depopulation of the CT state becomes less
kinetically feasible when going from tetracene to rubrene. This suggests that the enhanced photovoltage
produced by rubrene arises in part from kinetically suppressed dark recombination processes resulting from
a smaller H
ij
term in Eq. 1.6. The smaller H
ij
value for rubrene results from poorer D/A coupling at the
rubrene/C
60
interface, relative to tetracene, due to steric interactions between the phenyl groups of rubrene
and C
60
(cf. Figure 1.15). It is an oversimplification to say that J
so
is given by H
ij
, since J
so
is also
dependent on the reorganization energy (λ in Eq. 1.6). However, for similar systems, such as tetracene and
rubrene, the relative magnitudes of J
so
are largely due to differences in the spatial arrangements of donor
and acceptor, which strongly influences the D/A coupling, i.e. H
ij
. It is also likely that a similar effect acts
to minimize the [D
+
] term of Eq. 1.2 by suppressing dark injection of holes from the anode contact. Both
the latter and the former would act to curb the photovoltage losses in rubrene.
A significant effect on the V
oc
was also observed in the case of platinum
tetraphenylbenzoporphyrin (PtTPBP) when compared against CuPc in C
60
-based heterojunction devices.

Figure 1.16. Semi-log scale J-V characteristics in the dark (filled circles) and under illumination (open circles)
for CuPc (black), PtTPBP (red), and PtTPNP (blue) based OPV devices suggesting that both ΔE
DA
and kinetic
accessibility of the π-system play a role in determining V
oc
. Adapted from Perez with permission, Copyright
2009, American Chemical Society.

36
Both CuPc and PtTPBP possess a 38 electron π-system and intense UV-vis absorption profiles. However,
the ΔE
DA
for the PtTPBP/C
60
junction is 1.4 eV, while that of CuPc/C
60
is 1.7 eV. Based solely on these
ΔE
DA
values, one might expect the photovoltage produced by the PtTPBP/C
60
device to be significantly
lower than that of the CuPc/C60 heterojunction. In reality, however, the PtTPBP-based device exhibits a
V
oc
= 0.69 V, 20% higher than the CuPc based device, V
oc
= 0.48 V, as shown in Figure 1.16. Again, the
origin of this disparity appears to lie in the relative kinetic accessibility of the π electrons for the two donor
materials. The PtTPBP possesses four sterically-hindered phenyl substituents that confer a perturbed non-
planar saddle-shape or ‘butterfly’ contour on the molecule and reduce the accessibility of the core π-system
compared with CuPc. In both the PtTPBP and the rubrene cases the aforementioned phenomenological
frequency factor, J
so
are three orders of magnitude lower than their control device counterparts. Thus,
considerable kinetic suppression of the recombination reaction, D
+
+ A
-
→ D + A, results in the
unexpectedly large photovoltage produced by these devices.
In principle, materials with sterically inaccessible π systems and non-planar geometries seem
attractive for their apparent potential to control electronic coupling in OPV devices. However, in practice,
this same quality is likely to deleteriously impact charge transport and exciton diffusion. Thus, further
study is needed to understand the kinetics of recombination at the D/A interface and the dependence of [D
+
]
and [A
-
] on the nature of the electrical contacts and charge mobility of the incorporated materials.
1.6.8 Architectures for controlled leakage current
Incidentally, the kinetic expression in Eq. 1.5 can also be used to describe the remote impact that
device architecture may have on charge recombination at the D/A interface and the operational
photovoltage. Whereas k
rec
is more or less determined by the properties of the donor/acceptor pair, the
values of [D
+
] and [A
-
] may be determined by a myriad of factors. As a result, many studies have focused
on understanding charge injection at the electrical contacts
83,151,152
and on introducing various charge
blocking architectures
153-156
to influence the charge carrier density at the D/A interface. Architectures such
as these may be a promising route to chemical control of photovoltage losses in OPV devices. In Chapter
37

3, lamellar donor structures are considered such as that depicted in Figure 1.17 where a donor material
possessing certain desirable qualities, like high charge mobility, is bookended by complementary materials
to reduce the charge recombination rate and minimize charge leakage into the device. Although balancing
apposing electrochemical processes may be impracticable using single donor architectures, this distributive
bookend approach effectively partitions the performance requirements amongst several materials. It is
expected that such bookend architectures may have the potential to match donor materials with
complementary properties, to suppress voltage losses and simultaneously enhance spectral coverage to
increase photocurrent, with complete fill factor retention. Such structures are broadly compelling from a
synthetic, theoretical, and spectroscopic standpoint, both as a potential tool for interrogating electronic and
photochemical processes, and as a design element with the potential to enhance efficiency in future organic
photovoltaic devices.
1.7 Nanostructures and materials
In addition to identifying and utilizing molecules for OPV applications, a portion of this
dissertation focuses on the study of materials with dimensions on the nanometer length scale. Structures of
nanometer dimensionality often exhibit exotic properties relative to their macroscale counterparts. Such
features are compelling in solar energy conversion
157
for quantum confinement
158,159
and low temperature

Figure 1.17. Bookend device structures comprising multi-component lamellar donor or acceptor regions are a
promising route to tailored interfaces for limiting voltage losses and potentially enhancing spectral coverage.
38
solution processability
160
in colloidal semiconductor nanocrystals,
161
high surface area semiconducting
photoelectrodes,
162
nanoscale phase segregation,
62,63
and highly transparent electrodes for flexible
electronics.
163
Chapter 5 presents the design, development, and characterization several photovoltaic
device architectures that incorporate nanostructured components. Specifically, devices based on thin film
polymer composites of environmentally benign semiconducting SnSe nanocrystals, vapor processed hole
transport layers for solid state dye sensitized solar cells supported on TiO
2
nanowires, and flexible carbon-
based nanometer dimensional transparent electrodes are discussed.
1.8 Summary of topics
An incomplete understanding of the performance limitations between interacting components
hinders maturation of organic solar cells from laboratory concept into disruptive technology. The
following chapters in this dissertation present current progress in identifying and mitigating photovoltage
losses, achieving panchromaticity, controlling exciton migration, developing concepts for future
multipurpose charge transport materials, and utilizing OPV components with nanometer dimensionality. In
Chapter 2, common instrumentation and analysis techniques applied for characterizing OPV devices are
presented. Approaches for utilizing ensemble donor systems to suppress voltage losses, enhance spectral
coverage, and control exciton dynamics are detailed in Chapter 3. Implications of molecular electron
transfer are addressed in Chapter 4 for the future design of buffer materials to be employed between the
acceptor layer and the electron-collecting electrode. Finally, Chapter 5 summarizes recent results in
developing inorganic-organic composites and flexible transparent electrodes.
39
1.9 Chapter 1 endnotes
1 M. D. Perez, C. Borek, S. R. Forrest and M. E. Thompson, J. Am. Chem. Soc., 2009, 131, 9281.
2 N. Sato, K. Seki and H. Inokuchi, J. Chem. Soc., Faraday Trans., 1981, 77, 1621.
3 N. Sato, H. Inokuchi and E. A. Silinsh, Chem. Phys., 1987, 115, 269.
4 E. Becquerel, C.R. Acad. Sci., 1839, 9, 561.
5 A. W. Copeland, O. D. Black and A. B. Garrett, Chem. Rev., 1942, 31, 177.
6 B. Oregan and M. Gratzel, Nature, 1991, 353, 737.
7 M. Razeghi and A. Rogalski, J. Appl. Phys., 1996, 79, 7433.
8 C. W. Tang, Appl. Phys. Lett., 1986, 48, 183.
9 R. Memming, Top. Curr. Chem., 1988, 143, 79.
10 A. Goetzberger, C. Hebling and H. W. Schock, Mater. Sci. Eng., R, 2003, 40, 1.
11 R. L. Petritz, Physical Review, 1956, 104, 1508.
12 M. R. Wasielewski, Chem. Rev., 1992, 92, 435.
13 H. B. Gray, Nature Chemistry, 2009, 1, 7.
14 D. A. McQuarrie and J. D. Simon, Physical Chemistry: A Molecular Approach, University
Science Books, Sausalito, CA, 1997.
15 D. A. Neamen, Semiconductor Physics and Devices, Third Edition edn., McGraw-Hill, New York,
2003.
16 M. A. Green, Semicond. Sci. Technol., 1993, 8, 1.
17 M. A. Green, K. Emery, Y. Hishikawa and W. Warta, Progress in Photovoltaics, 2010, 18, 346.
18 N. S. Lewis and D. G. Nocera, P. Natl. Acad. Sci. USA, 2006, 103, 15729.
19 M. I. Hoffert, K. Caldeira, A. K. Jain, E. F. Haites, L. D. D. Harvey, S. D. Potter, M. E.
Schlesinger, S. H. Schneider, R. G. Watts, T. M. L. Wigley, et al., Nature, 1998, 395, 881.
20 M. I. Hoffert, K. Caldeira, G. Benford, D. R. Criswell, C. Green, H. Herzog, A. K. Jain, H. S.
Kheshgi, K. S. Lackner, J. S. Lewis, et al., Science, 2002, 298, 981.
40
21 B. P. Rand, D. P. Burk and S. R. Forrest, Phys. Rev. B, 2007, 75, 115327.
22 P. Heremans, D. Cheyns and B. P. Rand, Acc. Chem. Res., 2009, 42, 1740.
23 B. C. Thompson and J. M. J. Frechet, Angew. Chem. Int. Ed., 2008, 47, 58.
24 H. Spanggaard and F. C. Krebs, Sol. Energy Mater. Sol. Cells, 2004, 83, 125.
25 F. C. Krebs, S. A. Gevorgyan, B. Gholamkhass, S. Holdcroft, C. Schlenker, M. E. Thompson, B.
C. Thompson, D. Olson, D. S. Ginley, S. E. Shaheen, et al., Sol. Energy Mater. Sol. Cells, 2009,
93, 1968.
26 S. R. Forrest, Nature, 2004, 428, 911.
27 Z. Bao, A. Dodabalapur and A. J. Lovinger, Appl. Phys. Lett., 1996, 69, 4108.
28 Z. Bao, A. J. Lovinger and A. Dodabalapur, Appl. Phys. Lett., 1996, 69, 3066.
29 C. D. Dimitrakopoulos and P. R. L. Malenfant, Adv. Mater., 2002, 14, 99.
30 M. E. Roberts, A. N. Sokolov and Z. N. Bao, J. Mater. Chem., 2009, 19, 3351.
31 P. Peumans and S. R. Forrest, Appl. Phys. Lett., 2001, 79, 126.
32 J. G. Xue, S. Uchida, B. P. Rand and S. R. Forrest, Appl. Phys. Lett., 2004, 84, 3013.
33 J. G. Xue, B. P. Rand, S. Uchida and S. R. Forrest, Adv. Mater., 2005, 17, 66.
34 J. G. Xue, S. Uchida, B. P. Rand and S. R. Forrest, Appl. Phys. Lett., 2004, 85, 5757.
35 D. Cheyns, B. P. Rand and P. Heremans, Appl. Phys. Lett., 2010, 97, 3.
36 D. Wynands, M. Levichkova, M. Riede, M. Pfeiffer, P. Baeuerle, R. Rentenberger, P. Denner and
K. Leo, J. Appl. Phys., 2010, 107, 6.
37 R. Schueppel, R. Timmreck, N. Allinger, T. Mueller, M. Furno, C. Uhrich, K. Leo and M. Riede,
J. Appl. Phys., 2010, 107, 6.
38 Y. Y. Liang, Z. Xu, J. B. Xia, S. T. Tsai, Y. Wu, G. Li, C. Ray and L. P. Yu, Adv. Mater., 2010,
22, E135.
39 C. J. Brabec, Sol. Energy Mater. Sol. Cells, 2004, 83, 273.
40 M. Jorgensen, K. Norrman and F. C. Krebs, Sol. Energy Mater. Sol. Cells, 2008, 92, 686.
41 F. C. Krebs, T. Tromholt and M. Jorgensen, Nanoscale, 2010, 2, 873.
41
42 F. C. Krebs, Sol. Energy Mater. Sol. Cells, 2009, 93, 394.
43 B. Kippelen and J. L. Bredas, Energ. Environ. Sci., 2009, 2, 251.
44 P. Peumans, A. Yakimov and S. R. Forrest, J. Appl. Phys., 2003, 93, 3693.
45 H. Hoppe and N. S. Sariciftci, J. Mater. Res., 2004, 19, 1924.
46 R. Steim, F. R. Kogler and C. J. Brabec, J. Mater. Chem., 2010, 20, 2499.
47 J. L. Bredas, J. E. Norton, J. Cornil and V. Coropceanu, Acc. Chem. Res., 2009, 42, 1691.
48 C. Li, M. Liu, N. G. Pschirer, M. Baumgarten and K. Müllen, Chem. Rev., 2010, ASAP.
49 A. W. Hains, Z. Liang, M. A. Woodhouse and B. A. Gregg, Chem. Rev., 2010, ASAP.
50 H. Mathieu, P. Lefebvre and P. Christol, Phys. Rev. B, 1992, 46, 4092.
51 D. F. Blossey, Phys. Rev. B, 1970, 2, 3976.
52 D. F. Blossey, Phys. Rev. B, 1971, 3, 1382.
53 I. G. Hill, A. Kahn, Z. G. Soos and R. A. Pascal, Chem. Phys. Lett., 2000, 327, 181.
54 B. A. Gregg, J. Phys. Chem. B, 2003, 107, 4688.
55 G. de la Torre, P. Vazquez, F. Agullo-Lopez and T. Torres, J. Mater. Chem., 1998, 8, 1671.
56 B. W. Henderson and T. J. Dougherty, Photochem. Photobiol., 1992, 55, 145.
57 R. Bonnett, Chem. Soc. Rev., 1995, 24, 19.
58 S. Stachura, V. J. Desiderio and J. Allison, Journal of Forensic Sciences, 2007, 52, 595.
59 H. C. Casey, D. D. Sell and K. W. Wecht, J. Appl. Phys., 1975, 46, 250.
60 S. M. Sze, Physics of Semiconductor Devices, Wiley New York, 1981
61 S. Y. Wang, E. I. Mayo, M. D. Perez, L. Griffe, G. D. Wei, P. I. Djurovich, S. R. Forrest and M.
E. Thompson, Appl. Phys. Lett., 2009, 94, 233304
62 R. Giridharagopal and D. S. Ginger, J. Phys. Chem. Lett., 2010, 1, 1160.
63 C. Groves, O. G. Reid and D. S. Ginger, Acc. Chem. Res., 2010, 43, 612.
64 S. Saini and B. Bagchi, PCCP, 2010, 12, 7427.
42
65 E. Hennebicq, G. Pourtois, G. D. Scholes, L. M. Herz, D. M. Russell, C. Silva, S. Setayesh, A. C.
Grimsdale, K. Mullen, J. L. Bredas, et al., J. Am. Chem. Soc., 2005, 127, 4744.
66 B. Jackson and R. Silbey, J. Chem. Phys., 1983, 78, 4193.
67 G. D. Scholes, X. J. Jordanides and G. R. Fleming, J. Phys. Chem. B, 2001, 105, 1640.
68 H. Fidder, J. Knoester and D. A. Wiersma, J. Chem. Phys., 1991, 95, 7880.
69 T. Q. Nguyen, I. B. Martini, J. Liu and B. J. Schwartz, J. Phys. Chem. B, 2000, 104, 237.
70 T. Stubinger and W. Brutting, J. Appl. Phys., 2001, 90, 3632.
71 D. E. Markov, E. Amsterdam, P. W. M. Blom, A. B. Sieval and J. C. Hummelen, J. Phys. Chem.
A, 2005, 109, 5266.
72 S. R. Scully and M. D. McGehee, J. Appl. Phys., 2006, 100, 034907.
73 R. R. Lunt, J. B. Benziger and S. R. Forrest, Adv. Mater., 2010, 22, 1233.
74 R. R. Lunt, N. C. Giebink, A. A. Belak, J. B. Benziger and S. R. Forrest, J. Appl. Phys., 2009, 105,
053711.
75 B. A. Gregg, J. Sprague and M. W. Peterson, J. Phys. Chem. B, 1997, 101, 5362.
76 B. P. Rand, S. Schols, D. Cheyns, H. Gommans, C. Girotto, J. Genoe, P. Heremans and J.
Poortmans, Org. Electron., 2009, 10, 1015.
77 W. A. Luhman and R. J. Holmes, Appl. Phys. Lett., 2009, 94, 153304.
78 S. T. Roberts, C. W. Schlenker, V. S. Barlier, S. W. Chin, R. McAnally, M. E. Thompson and S.
E. Bradforth, in preparation, 2010.
79 K. Tsutsumi, H. Yoshida and N. Sato, Chem. Phys. Lett., 2002, 361, 367.
80 K. Kanai, K. Akaike, K. Koyasu, K. Sakai, T. Nishi, Y. Kamizuru, Y. Ouchi and K. Seki, Appl.
Phys. A, 2009, 95, 309.
81 E. A. Silinsh and V. Čápek, Organic Molecular Crystals, American Institute of Physics, New
York, 1994.
82 H. Ishii, K. Sugiyama, E. Ito and K. Seki, Adv. Mater., 1999, 11, 605.
83 J. Hwang, A. Wan and A. Kahn, Mater. Sci. Eng., R, 2009, 64, 1.
84 A. Kahn, N. Koch and W. Y. Gao, Journal of Polymer Science Part B-Polymer Physics, 2003, 41,
2529.
43
85 D. M. Alloway and N. R. Armstrong, Appl. Phys. A, 2009, 95, 209.
86 M. Brumbach, D. Placencia and N. R. Armstrong, J. Phys. Chem. C, 2008, 112, 3142.
87 S. Duhm, G. Heimel, I. Salzmann, H. Glowatzki, R. L. Johnson, A. Vollmer, J. P. Rabe and N.
Koch, Nat. Meter., 2008, 7, 326.
88 S. Kera, H. Yamane, H. Fukagawa, T. Hanatani, K. K. Okudaira, K. Seki and N. Ueno, J.
Electron. Spectrosc. Relat. Phenom., 2007, 156, 135.
89 S. Kera, H. Yamane, H. Honda, H. Fukagawa, K. K. Okudaira and N. Ueno, Surf. Sci., 2004, 566,
571.
90 D. Placencia, W. N. Wang, R. C. Shallcross, K. W. Nebesny, M. Brumbach and N. R. Armstrong,
Adv. Funct. Mater., 2009, 19, 1913.
91 H. Yamane, H. Honda, H. Fukagawa, M. Ohyama, Y. Hinuma, S. Kera, K. K. Okudaira and N.
Ueno, J. Electron. Spectrosc. Relat. Phenom., 2004, 137, 223.
92 H. Yamane, Y. Yabuuchi, H. Fukagawa, S. Kera, K. K. Okudaira and N. Ueno, J. Appl. Phys.,
2006, 99, 5.
93 J. Bredenbeck, A. Ghosh, H. K. Nienhuys and M. Bonn, Acc. Chem. Res., 2009, 42, 1332.
94 J. Bredenbeck, A. Ghosh, M. Smits and M. Bonn, J. Am. Chem. Soc., 2008, 130, 2152.
95 M. Khalil, N. Demirdoven and A. Tokmakoff, J. Phys. Chem. A, 2003, 107, 5258.
96 T. Brixner, J. Stenger, H. M. Vaswani, M. Cho, R. E. Blankenship and G. R. Fleming, Nature,
2005, 434, 625.
97 V. Coropceanu, J. Cornil, D. A. da Silva, Y. Olivier, R. Silbey and J. L. Bredas, Chem. Rev., 2007,
107, 926.
98 J. Nelson, J. J. Kwiatkowski, J. Kirkpatrick and J. M. Frost, Acc. Chem. Res., 2009, 42, 1768.
99 B. Servet, G. Horowitz, S. Ries, O. Lagorsse, P. Alnot, A. Yassar, F. Deloffre, P. Srivastava, R.
Hajlaoui, P. Lang, et al., Chem. Mater., 1994, 6, 1809.
100 W. L. Ma, C. Y. Yang, X. Gong, K. Lee and A. J. Heeger, Adv. Funct. Mater., 2005, 15, 1617.
101 Y. Hirose, A. Kahn, V. Aristov, P. Soukiassian, V. Bulovic and S. R. Forrest, Phys. Rev. B, 1996,
54, 13748.
102 C. W. Schlenker, C. K. Renshaw, S. R. Forrest and M. E. Thompson, in preparation, 2010.
44
103 S. Yoo, B. Domercq, S. R. Marder, N. R. Armstrong and B. Kippelen, Proc. SPIE-Int. Soc. Opt.
Eng., 2004, 5520, 110.
104 W. J. Potscavage, A. Sharma and B. Kippelen, Acc. Chem. Res., 2009, 42, 1758.
105 A. Moliton and J. M. Nunzi, Polym. Int., 2006, 55, 583.
106 S. Yoo, B. Domercq and B. Kippelen, J. Appl. Phys., 2005, 97, 103706.
107 B. Mazhari, Sol. Energy Mater. Sol. Cells, 2006, 90, 1021.
108 N. C. Giebink, B. E. Lassiter, G. P. Wiederrecht, M. R. Wasielewski and S. R. Forrest, Phys. Rev.
B, in review.
109 N. C. Giebink, B. E. Lassiter, G. P. Wiederrecht, M. R. Wasielewski and S. R. Forrest, Phys. Rev.
B, in review.
110 J. Cuiffi, T. Benanti, W. J. Nam and S. Fonash, Appl. Phys. Lett., 2010, 96, 143307.
111 K. Emery, in Handbook of Photovoltaic Science and Engineering, ed. H. S. Luque A., Wiley &
Sons, Chichester, UK, 2003.
112 K. A. Emery and C. R. Osterwald, Sol. Cells, 1986, 17, 253.
113 R. J. Matson, K. A. Emery and R. E. Bird, Sol. Cells, 1984, 11, 105.
114 V. Shrotriya, G. Li, Y. Yao, T. Moriarty, K. Emery and Y. Yang, Adv. Funct. Mater., 2006, 16,
2016.
115 D. Kearns and M. Calvin, J. Chem. Phys., 1958, 29, 950.
116 C. J. Brabec, A. Cravino, D. Meissner, N. S. Sariciftci, T. Fromherz, M. T. Rispens, L. Sanchez
and J. C. Hummelen, Adv. Funct. Mater., 2001, 11, 374.
117 M. C. Scharber, D. Wuhlbacher, M. Koppe, P. Denk, C. Waldauf, A. J. Heeger and C. L. Brabec,
Adv. Mater., 2006, 18, 789.
118 C. J. Brabec, A. Cravino, D. Meissner, N. S. Sariciftci, M. T. Rispens, L. Sanchez, J. C.
Hummelen and T. Fromherz, Thin Solid Films, 2002, 403, 368.
119 F. B. Kooistra, J. Knol, F. Kastenberg, L. M. Popescu, W. J. H. Verhees, J. M. Kroon and J. C.
Hummelen, Org. Lett., 2007, 9, 551.
120 M. Lenes, G. Wetzelaer, F. B. Kooistra, S. C. Veenstra, J. C. Hummelen and P. W. M. Blom, Adv.
Mater., 2008, 20, 2116.
121 R. A. Marcus, Rev. Mod. Phys., 1993, 65, 599.
45
122 P. F. Barbara, T. J. Meyer and M. A. Ratner, J. Phys. Chem., 1996, 100, 13148.
123 H. B. Gray and J. R. Winkler, Annu. Rev. Biochem, 1996, 65, 537.
124 M. Fedurco, Coord. Chem. Rev., 2000, 209, 263.
125 F. Pichot and B. A. Gregg, J. Phys. Chem. B, 2000, 104, 6.
126 B. A. Gregg, F. Pichot, S. Ferrere and C. L. Fields, J. Phys. Chem. B, 2001, 105, 1422.
127 W. Shockley and H. J. Queisser, J. Appl. Phys., 1961, 32, 510.
128 S. Guha, J. Yang, P. Nath and M. Hack, Appl. Phys. Lett., 1986, 49, 218.
129 M. Hack and M. Shur, J. Appl. Phys., 1985, 58, 997.
130 K. L. Mutolo, E. I. Mayo, B. P. Rand, S. R. Forrest and M. E. Thompson, J. Am. Chem. Soc.,
2006, 128, 8108.
131 G. de la Torre, C. G. Claessens and T. Torres, Chem. Commun., 2007, 2000.
132 B. W. Ma, C. H. Woo, Y. Miyamoto and J. M. J. Frechet, Chem. Mater., 2009, 21, 1413.
133 B. Verreet, S. Schols, D. Cheyns, B. P. Rand, H. Gommans, T. Aernouts, P. Heremans and J.
Genoe, J. Mater. Chem., 2009, 19, 5295.
134 G. G. Malliaras, J. R. Salem, P. J. Brock and J. C. Scott, J. Appl. Phys., 1998, 84, 1583.
135 B. A. Gregg, M. A. Fox and A. J. Bard, J. Phys. Chem., 1990, 94, 1586.
136 H. Kallmann and M. Pope, J. Chem. Phys., 1959, 30, 585.
137 Geacinto.N, M. Pope and H. Kallmann, J. Chem. Phys., 1966, 45, 2639.
138 A. Ortiz-Conde, F. J. G. Sanchez and J. Muci, Solid-State Electronics, 2000, 44, 1861.
139 A. Jain and A. Kapoor, Sol. Energy Mater. Sol. Cells, 2004, 81, 269.
140 T. C. Banwell and A. Jayakumar, Electron. Lett, 2000, 36, 291.
141 G. Grampp, Angew. Chem. Int. Ed., 1993, 32, 691.
142 Y. P. Yi, V. Coropceanu and J. L. Bredas, J. Am. Chem. Soc., 2009, 131, 15777.
143 W. Chen, H. Huang, S. Chen, Y. L. Huang, X. Y. Gao and A. T. S. Wee, Chem. Mater., 2008, 20,
7017.
46
144 W. Chen, S. Chen, Y. L. Huang, H. Huang, D. C. Qi, X. Y. Gao, J. Ma and A. T. S. Wee, J. Appl.
Phys., 2009, 106, 064910.
145 N. R. Armstrong, W. N. Wang, D. M. Alloway, D. Placencia, E. Ratcliff and M. Brumbach,
Macromol. Rapid Commun., 2009, 30, 717.
146 T. J. Klofta, J. Danziger, P. Lee, J. Pankow, K. W. Nebesny and N. R. Armstrong, J. Phys. Chem.,
1987, 91, 5646.
147 T. Enokida, Jpn. J. Appl. Phys., Part 2, 1992, 31, L1135.
148 V. Chauhan, R. Hatton, P. Sullivan, T. Jones, S. W. Cho, L. Piper, A. deMasi and K. Smith, J.
Mater. Chem., 2010, 20, 1173.
149 Y. Luo and D. Kessel, Photochem. Photobiol., 1997, 66, 479.
150 R. F. Bailey-Salzman, B. P. Rand and S. R. Forrest, Appl. Phys. Lett., 2007, 91, 013508.
151 J. C. Scott and G. G. Malliaras, Chem. Phys. Lett., 1999, 299, 115.
152 I. H. Campbell, J. D. Kress, R. L. Martin, D. L. Smith, N. N. Barashkov and J. P. Ferraris, Appl.
Phys. Lett., 1997, 71, 3528.
153 N. Li, B. E. Lassiter, R. R. Lunt, G. Wei and S. R. Forrest, Appl. Phys. Lett., 2009, 94, 023307.
154 Z. T. Liu, M. F. Lo, H. B. Wang, T. W. Ng, V. A. L. Roy, C. S. Lee and S. T. Lee, Appl. Phys.
Lett., 2009, 95, 093307.
155 Y. Kinoshita, T. Hasobe and H. Murata, Appl. Phys. Lett., 2007, 91, 083518.
156 Y. Kinoshita, R. Takenaka and H. Murata, Appl. Phys. Lett., 2008, 92, 243309.
157 P. V. Kamat, J. Phys. Chem. C, 2007, 111, 2834.
158 Y. Wang and N. Herron, J. Phys. Chem., 1991, 95, 525.
159 X. G. Peng, L. Manna, W. D. Yang, J. Wickham, E. Scher, A. Kadavanich and A. P. Alivisatos,
Nature, 2000, 404, 59.
160 C. B. Murray, C. R. Kagan and M. G. Bawendi, Annu. Rev. Mater. Sci., 2000, 30, 545.
161 C. Burda, X. B. Chen, R. Narayanan and M. A. El-Sayed, Chem. Rev., 2005, 105, 1025.
162 M. Gratzel, Journal of Photochemistry and Photobiology a-Chemistry, 2004, 164, 3.
163 L. G. De Arco, Y. Zhang, C. W. Schlenker, K. Ryu, M. E. Thompson and C. W. Zhou, Acs Nano,
2010, 4, 2865.
47
Chapter 2
Instrumentation and analysis
2.1 Standard reporting conditions
Generating meaningful solar conversion efficiency comparisons between photovoltaic devices
requires adherence to standards for solar spectral irradiance and testing conditions. Internationally
accepted standard solar cell testing procedures were developed in the 1980s by the national photovoltaic
calibration laboratories in many countries around the world, including the National Renewable Energy
Laboratory (NREL) in Golden, CO.
1,2
However, researchers at the intersection of various disciplines,
where emerging solar cell technologies are currently being developed, may possess limited awareness
and/or limited funds to proactively institute these standards of measurement in their pallet of routine
devices characterization techniques. The resulting efficiency values, measured under different testing
conditions, are extremely problematic for reliably comparing device data collected and reported among
different laboratories.
Researchers in the OPV community are becoming increasingly aware of this problem.
3
In an
attempt to adhere to standard reporting conditions (SRC), this chapter attends to the topic of standardized
device characterization and measurement procedures, as developed based on personal consultation with
Keith Emery, photovoltaic device characterization team leader at NREL. This material addresses
procedures to account for the mismatch between the spectral irradiance of the research lamp source (E
s
) and
the reference solar spectrum, ASTM G173-03 AM1.5G (E
R
), Air Mass for 37° south-facing tilt, global
illumination, and for the mismatch between the spectral responsivity of the test device (S
T
) and that of the
reference detector (S
R
). This is affected by calculating a spectral mismatch factor (M). The spectral
mismatch factor, given by
48

(2.1)
integrated over all wavelengths (λ), is used in correcting the short circuit current of the test device ( ),
measured under E
s
, to the short circuit current of the test devices ( ) measured under E
R
, by calculating
= /M for every tested device. Additionally, the measurement procedure verifies that the corrected
white light response equals the integrated multiplicative product of the spectral responsivity and reference
irradiance, .

Figure 2.1. Hamamatsu Si photodiode, calibrated at the National Renewable Energy Laboratory (NREL) used as
reference photodetector.

49
2.2 Instrumentation
As depicted in Figure 2.1, an NREL calibrated silicon photodiode (Hamamatsu S1787-04 or
S1787-04; 8RA or KG5 Filter) is employed in tuning white light illumination intensity from an AM 1.5G
filtered 300 W Xenon arc lamp (Newport
®
Oriel Product Line) coupled to the sample via a fused silica ¼
Wave UV Al mirror, to an incident irradiance (P
inc
) of 1000 Wm
-2
(1 sun) and to perform routine spectral
mismatch correction to E
R
. To measure E
s
(λ) an Ocean Optics USB2000+VIS-NIR spectrometer is situated
at the sample mount position, thus accounting for the Xe arc output and all incorporated optics. Reference
photodiodes are selected to achieve near unit spectral mismatch with respect to the test cell and to minimize
filter degradation over time. Additionally, all broadband measurements are performed with a beam size
significantly larger than the area of the device, to ensure accurate current density assessment. Neutral

Figure 2.2. Instrumentation employed in OPV optoelectronic characterization.

50

density filter are incorporated for attenuated broadband illumination experiments. Conversely, all
monochromatic measurements are performed using an aperture that is smaller than the area of the device,
ensuring that the reference detector and the test device experience the same photon flux. The reference
photodiode is housed in micrometer driven test assembly, to ensure the path lengths from the source to the
active planes of both the reference and the test devices are equal. Photodiodes are submitted to NREL for
recalibration on a 6-month cycle, to minimize detector degradation error. The instrumentation relevant to
both simulated solar and monochromatic irradiation is depicted in Figure 2.2. The current dependence on
applied test voltage I(V) is measured using a Keithley 2420 SourceMeter with 100 pA sensitivity,
controlled via LabView interface, in the dark and under simulated solar illumination. To minimize error
arising from inaccuracies in device area, current density is calculated by pixel mapping for digital cathode
images collected by optical microscopy. Filtered and frequency modulated monochromatic light (250 Hz,
10 nm FWHM) from a Cornerstone
TM
260 ¼ M double grating monochromator (Newport® 74125) and Xe
arc source is used in conjunction with an EG&G 7220 DSP lock-in amplifier to perform all spectral
responsivity and spectral mismatch correction measurements. The phase sensitive nature of this technique
requires that frequency modulation is applied using an enclosed optical chopper head coupled directly to
the outlet of the monochromator to minimize scattering error. A near-UV bandpass filter is

Figure 2.3. Raw current under monochromatic illumination for a typical CuPc/C
60
OPV (solid trace) and for an
8RA filtered silicon photodiode reference cell (dotted trace).

51

employed to maximize shortwave signal-to-noise and order sorting cut-on filters (309 nm and 570 nm) are
applied to suppress second-order artifacts. All filters are housed within LabView controlled automated
carousel.
Templated data processing is performed using custom scripting in an OriginPro environment to
quickly obtain radiant photon flux, M, and device performance metrics, such as short circuit current density
J
sc
(V = 0), open circuit voltage V
oc
(J = 0), maximum output power density P
max
, fill factor FF =
P
max
/J
sc
V
oc
, power conversion efficiency η
p
= P
max
/P
inc
, corrected for spectral mismatch. Additional
rudimentary equivalent circuit modeling is routinely applied to estimate η
p
losses due to factors such as
series resistance (R
s
).
2.3 Analysis
The routine testing procedures presented here were developed as a reliable and high-throughput
solution for accurately generating OPV performance data in accord with SRC for a variety of materials
with relatively disparate spectral response features. As such, E
s
is monitored according to the reference
photodiode current density at short circuit to within 0.02 mA/cm
-2
of the calibration value under E
R
. For
this irradiance level, the obtained photocurrent is roughly linear with illumination intensity,
4
whereas the
photovoltage is far less sensitive to illumination intensity, increasing according to V
oc
∝ ln(P
inc
), and
therefore undesirable as a calibration parameter. It should be noted that, following spectral mismatch

Figure 2.4. The Script Window is accessible via the native toolbar in OriginPro and allows the user to develop
custom programs such as the simple data transfer script ‘It_import.ogs.’

52

correction, this procedure is generally applicable for minimizing spurious efficiency data associated with
E
s
,
2
however it does not completely eliminate a slight systematic error in the resulting voltage. While the
error associated with this method is generally negligible, it can in principle, be alleviated by verifying the
linearity of the test device photoresponse and adjusting E
s
with respect to the calculated mismatch factor
between the reference and test device for each unique device architecture. However, at the basic research
level, employing this method is often impracticable due to low materials availability and poor device
longevity and does not ultimately offer improvement in measurement accuracy greater than trial-to-trial
variability.
Following device fabrication, current voltage characteristics are measured for the test device in the
dark and under calibrated broadband illumination. Within a typical voltage range -1.0 V ≤ V ≤ +1.5 V, raw
current values in reverse bias (cf. Section 1.5) are on the order of I = 0.1 µA to below the sensitivity of the
instrumentation (~ 100 pA) for a ∅1 mm electrical contact. Under illumination, reverse
a) b)
c) d)
Figure 2.5. The OriginPro toolbar can be customized with dedicated user-defined buttons to expedite repetitive
procedures using steps a – d.

53

currents are typically on the order of I = 1 - 10 µA. In forward bias in the dark and under 1 sun
illumination, the measured current can be as high as I = 10 mA. As illustrated in Figure 2.3 for a typical
CuPc/C
60
planar heterojunction device, the current measured at short circuit as a function of incident
/*PV_Treatment_CWS.ogs */

get data4_a -e NV; /* # of voltage points */
NL=1; /* # of light intensities */

/* calculate Jsc and copy into parameters worksheet */

for(i=2;i<=NV;i+1){
if (%(Data4,3,i)>0){
j1=%(Data4,3,i-1);
j2=%(Data4,3,i);
dV1=j1/(j1-j2);
parameters_C[1]=abs(Data4_N[i-1]+dV1*(Data4_N[i]-Data4_N[i-1]));
break;
}
};

/* calculate Voc and copy into parameters worksheet */

for(i=2;i<=NV;i+1){
if (%(Data4,6,i)>0){
j1=%(Data4,6,i-1);
j2=%(Data4,6,i);
dV1=j1/(j1-j2);
parameters_D[1]=Data4_C[i-1]+dV1*(Data4_C[i]-Data4_C[i-1]);
break;
}
};

/* calculate VO and copy into parameters worksheet */

for(i=2;i<=NV;i+1){
if (%(Data4,11,i)>0){
j1=%(Data4,11,i-1);
j2=%(Data4,11,i);
dV1=j1/(j1-j2);
parameters_B[1]=Data4_C[i-1]+dV1*(Data4_C[i]-Data4_C[i-1]);
break;
}
};

/*Calculate Max Power*/
%B=%(Data4,9);
limit %B;
parameters_E[1]=limit.ymax;
data4_G=data4_C-parameters_B[1];

/*Copy Integrated Jsc*/
copy _integ_area Data4_I;
get data4_I -e NW;
parameters_H[1]=Data4_I[NW-2];
Figure 2.6. Sample customized data treatment script, ‘PV_Treatment_CWS.ogs,’ employed in extraction of OPV
metrics, such as short circuit current density (J
sc
), open circuit voltage (V
oc
), compensation voltage (V
0
),
maximum power density P
max
and spectrally integrated photocurrent density J
sc
QE
.
54

wavelength is on the order of I(λ) = 2 – 20 nA for monochromatic radiant flux ca. 0.01 – 0.1 µW between λ
= 300 and 750 nm emanating through ∅0.5 mm aperture. To accurately determine η
p
and the external
quantum efficiency from the untreated test device data and the measured I(λ) characteristics for the NREL
calibrated reference detector, a substantial number of repetitive calculations are required.
A customized data treatment tool was developed in OriginPro and broadly applied as a routine
protocol for comparing performance metrics between various device architectures. This approach
substantially reduces the time invested by the user in data analysis. The following is intended as a brief
introduction and tutorial for facilitating such data processing, but assumes the user’s basic understanding of
the native software. The ‘Script Window’ is accessible via the native OriginPro toolbar as illustrated in
Figure 2.4 and allows the customization of the working environment. From the Script Window, programs
can be written and run to expedite repetitive procedures. Once a program script is generated, such as the
data transfer script ‘It_import.ogs,’ shown in the Script Window of Figure 2.4, which simply copies data
from one worksheet to another, it is often convenient to link such a program to its own dedicated button on
the OriginPro toolbar. This allows the user to quickly run the script, without directly accessing the code
through the Script Window. To activate a script and link it to the OriginPro toolbar, the script is saved to
an accessible folder on the local hard disk, such as C:\Program Files\OriginLab\OriginPro75\. As
illustrated in Figure 2.5, the resulting file is then assigned to a user-defined button, which is subsequently
associated with the OriginPro toolbar, acting as a liaison between the user and the .ogs file. Of specific
interest to OPV research, is the calculation of performance metrics, such as spectral mismatch factor (M),

Figure 2.7. The OPV data treatment project template contains step-by-step instructions in the ‘Directions.’
window for data treatment and system calibration, including a list of symbol definitions.

55

short circuit current density (J
sc
), open circuit voltage (V
oc
), compensation voltage (V
0
), maximum power
density P
max
, fill facotor (FF = P
max
/J
sc
V
oc
), spectrally integrated photocurrent density (J
sc
QE
), etc. Sample
encoding to facilitate these calculations using a user-defined script for a broad range of materials and
device architectures is presented in Figure 2.6.
For standardized data treatment procedures, it is often convenient to develop a project template
with elements common to those identified in a given data treatment script. Such a project template and
instructions for its operation are depicted in Figure 2.7 for a typical OPV device data set. In the data
analysis routine, the user saves the treatment template as a new project and the measured wavelength-

Figure 2.8. The SRS_Kernel comprises calibration data and performs several important calculations, relevant for
external quantum efficiency and spectral mismatch, in the background during data treatment.

Figure 2.9. Spectral mismatch factor (M) calculation incorporated in the ‘MCalc’ worksheet.

56

dependant current data is imported via the native ASCII import function as new columns into predefined
worksheets within the template project, ‘Irtestday’ and ‘It’ for the reference and test cells, respectively.
Subsequently, calling the ‘It_import.orgs’ script affects calculation of the test cell external quantum
efficiency (EQE) and all parameters required for spectral mismatch correction from within
the‘SRS_Kernel,’ accessible as depicted in Figure 2.8. Excepting calibration update, there is generally no
need for the user to access the SRS_Kernel, as the EQE calculations are automated. Within the main
project folder, the ‘MCalc’ window allows the user to calculate the spectral mismatch factor, by simply
clicking the ‘Recalculate’ button, as depicted in Figure 2.9. The main project folder also contains a
window labeled ‘MonoChromeTestCurrent,’ which displays the theoretical photocurrent density produced

Figure 2.10. Activating the data in the ‘MonoChromeTestCurrent’ window and selecting ‘Analysis… Calculus…
Integrate’ yields the spectrally integrated photocurrent density.

Figure 2.11. Photocurrent density is automatically spectral mismatch corrected.

57

at short-circuit under E
R
(λ). Activating the data in this window and selecting
‘Analysis…Calculus…Integrate’ yields the spectrally integrated photocurrent density
(2.2)
as illustrated in Figure 2.10.
Following treatment of the measured I(λ) data to obtain test device EQE and mismatch factor,
voltage dependent current data is imported to the OPV data treatment project. As illustrated in Figure 2.11,
the spectral mismatch correction (cf. Figure 2.9) is embedded in the ‘Data4’ worksheet and is automatically
applied upon calculating the photocurrent density by running the ‘Set All Column Values’ tool with Data4
active. As such, calling the ‘PV_Treatment_CWS.ogs’ script presented in Figure 2.6 determines
performance metrics for the test device based on spectral mismatch corrected data and posts the results to
the ‘parameters’ worksheet. In addition to obtaining the standard performance metrics, here the user may
cross-reference the short circuit current density, obtained under spectral mismatch correction from
simulated solar illumination, with the theoretical spectrally integrated photocurrent density. Running the
treatment script also returns the percent difference (JscPD) between these to values, with JscPD < 10%
indicating relative agreement between the broadband and monochromatic methods. Values returned for
JscPD in excess of 20% may reflect poor calibration or the necessity for additional broadband DC light bias
in I(λ) data collection. Note, if the user has neglected to calculate the spectrally integrated photocurrent

Figure 2.12. Rudimentary equivalent circuit modeling can be applied to estimate OPV losses.
58

density, an error message reminding them to do so will appear upon running the treatment script. At this
point values for J
sc
, V
oc
, FF, P
max
, V
0
, and η
p
have been returned to the user.
From within the OPV treatment template, rudimentary modeling according to the equivalent
circuit model illustrated in Figure 2.12, can be performed by defining a fitting routine with OriginPro’s
‘Advanced Fitting Tool.’ This is accessible under ‘Non-linear Curve Fit’ in the ‘Analysis’ menu. Two
primary options exist for applying the requisite user-defined fitting routine. In the first, the user may

Figure 2.13. User-defined functions can be developed and saved as .fdf in the ‘Advanced Fitting Tool’ by
selecting ‘Function…New.’

Figure 2.14. An externally provided .fdf routine can be accessed by selecting ‘Function/Add…’ and opening the
.fdf from its location on the local hard disk.
59

generate a fitting routine through the function editor, as illustrated in Figure 2.13, for generating the
function definition file, ‘diodefit_cws.fdf.’ Alternatively, an externally provided .fdf routine can be loaded
to the user’s local directory, to be called from within the Advanced Fitting Tool, as shown in Figure 2.14.
Once the .fdf routine is available, diode parameter extraction from the dark current data is performed by
activating the data set in the ‘DiodeFit’ window of the OPV treatment project and selecting ‘Functions,’
‘DiodeFit_CWS,’ ‘Actions,’ ‘Fit,’ from the select functions menu or by clicking the ‘Fit’ button in the
Non-linear Curve Fitting window. At this point, it is convenient to verify initial parameter values of Js =
0.0006 (mA/cm
2
), n = 2 (unitless), and Rsa = 0.001 (kΩcm
2
), and that the ‘Same X as Fitting Data,’ option
is activated in the ‘Fit Curve’ tab before clicking the 200 Iter. button in the Fitting Session window to
perform the non-linear curve fit with 200 iterations. The resulting model data and fitting parameters are
copied to the ‘DiodeFit’ window. An analogous fitting procedure can be applied for the photocurrent data.
Finally, user specific information, such as date experiment number, log page number, device
architecture, substrate number, and contact pad location may be entered into the ‘Characterization’ layout
window for later reference as illustrated in Figure 2.15. This window contains all pertinent graphical
information and extracted parameters corresponding to the user’s device data set. The print format for this

Figure 2.15. The characterization layout.
60
layout is condensed to be accommodated within a half 9¼” × 11¼” page laboratory notebook entry. Upon
loading all provided .ogs and .fdf files and associating them with appropriate local executable elements
within the project environment, saving and closing the project establishes these elements globally and
allows them to be called later from any local project for data analysis.
61
2.4 Chapter 2 endnotes
1 K. A. Emery and C. R. Osterwald, Sol. Cells, 1986, 17, 253.
2 R. J. Matson, K. A. Emery and R. E. Bird, Sol. Cells, 1984, 11, 105.
3 V. Shrotriya, G. Li, Y. Yao, T. Moriarty, K. Emery and Y. Yang, Adv. Funct. Mater., 2006, 16,
2016.
4 S. M. Sze, Physics of Semiconductor Devices, Wiley New York, 1981
62
Chapter 3
Ensemble donor systems
3.1 Overcoming counterposing limitations
As outlined in Chapter 1, achieving balance among opposing materials requirements for efficient
organic photovoltaic (OPV) device operation continues to be a significant performance limitation in
employing these devices for efficient solar energy conversion. This chapter demonstrates the utility of
developing ensemble donor systems, where several materials work in concert to meet the challenges
required for efficient device operation that a single material could not meet alone. Two general model
systems are examined highlight materials requirements and device architectures intended to limit
recombination losses, enhance spectral coverage, and control exciton diffusion dynamics are outlined.
Specifically, in the first approach lamellar bookend structures, comprising a collection of complementary
donor materials with cascading exciton energies are developed to demonstrate the relative importance of
interfacial charge injection, recombination, and bulk transport processes in determining the operational
open circuit voltage (V
oc
). The general efficacy of this new distributive bookend approach for suppressing
voltage losses, while enhancing spectral coverage, is demonstrated by more than doubling the power output
of pentacene-based devices, under spectral mismatch corrected AM1.5G simulated 1 sun illumination.
These results represent a significant development for defining materials systems to achieve high
performance OPV devices.
In the second approach, model Host-Guest donor systems consisting of an aryl-substituted
tetracene host and platinum porphyrin derivative are developed for the efficient sensitized preparation of
triplet excited states in thin film materials with relatively low intersystem crossing efficiency. The impetus
for this study lies in the potential viability of the resulting long-lived triplet exciton to reach the
donor/acceptor interface prior to deactivation. The thickness dependence of the external quantum
efficiency (EQE) signal exhibited by these Host-Guest devices, suggest triplet sensitization by direct
63
optical excitation of the Guest material is a viable route to enhanced exciton diffusion lengths. The
implications of these results represent a compelling backdrop for future synthetic, theoretical, and
spectroscopic investigation into the properties of ensemble donor systems for enhanced organic solar cell
efficiencies.
3.2 Lamellar bookend structures
3.2.1 Potential enhancements
One promising avenue for improving OPV device performance is to increase the short circuit
current density (J
sc
, V = 0) by harvesting the low energy photons that are currently neglected in the near-
infrared (NIR) region of the solar spectrum. However, obtaining high maximum output power density
(P
max
), resulting in high η
p
, necessitates maintaining high open circuit-voltage (V
oc
) without degrading
device fill factor (FF = P
max
/J
sc
V
oc
). Consequently, architectural permutations that augment J
sc
, but
diminish V
oc
, and vice versa, are unacceptable. In this chapter, the prospect of distributing materials
requirements amongst several complementary absorbing layers is explored to control interfacial electronic
coupling for suppressed voltage losses, while enhancing spectral coverage for increased short circuit
current. The potential is demonstrated for this bookend technique to relax counterposing design rules that
are otherwise difficult to fulfill using conventional single-donor architectures.
Optical-to-electrical energy conversion in conventional OPVs occurs as a result of photoinduced
electron transfer between a molecular electron-donor (D) and an appropriate electron-accepting molecule
(A) through 6 photochemical or electrochemical processes. 1) Photon absorption leads to a localized
excited state (exciton) with energy E
00
on the donor material. 2) This local excitation diffuses to the donor
acceptor interface via some generalized excitation transfer mechanism. 3) Charge transfer quenching of the
exciton at the D/A interface produces a coulombically bound charge transfer (CT) state in the form of a
D
+
A
-
polaron pair. 4) Polaron pair separation proceeds via a poorly understood mechanism, to produce
ionized D
+
(hole) and A
-
(electron) polaron species. 5) Holes in the donor layer and electrons in the
64

acceptor layer, are transported away from the D/A interface via localized self-exchange. 6) Regeneration
of neutral ground state molecular species occurs via external electrical contacts, anode connected to the
donor and cathode connected to the acceptor. Ultimately, J
sc
, V
oc
, FF, and η
p
are determined by the
thermodynamic and kinetic aspects of each chemical processes.
3.2.2 Competing processes
Unfortunately, properties of a specific donor material that favorably impact one important
performance metric are commonly mutually exclusive, with respect to the others. That is, achieving high
η
p
requires a balance between opposing chemical properties. For example, molecular geometries with an
innate propensity to frustrate intermolecular electronic interactions have been implicated in suppressing
photovoltage losses.
1
However, strong intermolecular electronic communication is highly desirable for
efficient exciton diffusion and charge transport.
2-4
This mutual exclusivity renders dubious the prospect of
finding a single suitable donor material to fulfill all desired requirements. As a result, in this chapter
examines the proposition of developing bookend architectures as depicted in Figure 3.1, where the donor
region of the device is partitioned into a multilayer lamellar stack of complementary materials, B
1
to B
n
,
each material executing only a few specific functions. In this approach, materials that are outstanding in
one capacity as a single layer, but perform poorly in other capacities, are bookended by complementary
materials that are effective, where the first is unsatisfactory. Here variations in device performance derived

Figure 3.1. Schematic photoactive region for proposed bookend architecture in which the donor layer
comprises three complementary materials B
1
, B
2
, and B
3
.

65
from different materials are conceptualized in terms of subtle perturbations around the archetypal
oligoacene motif. Employing a Marcus theory-based interpretation of electrochemical processes occurring
in model devices allows the development of bookend donor structures based on aryl-substituted tetracenes
as candidates for suppressing V
oc
losses. Building on this bookend concept, the ability is demonstrated to
simultaneously suppress voltage losses while enhancing spectral response to increase J
sc
, with complete FF
retention in pentacene based devices.
3.2.3 Molecular perspective on open-circuit voltage losses
Attenuating the process of charge recombination at the D/A interface, D
+
+ A
-
→ D
0
+ A
0
is
particularly important for minimizing photovoltage losses in OPVs under illumination. However, in the
dark, this reaction can often be driven electrically, by applying a positive (forward) external bias to the
anode contact. The injected charge carriers, D
+
from the anode and A
-
from the cathode, may then take part
in the recombination reaction, with rate constant k
rec
, given by Marcus theory for outer sphere electron
transfer. Hence, in the dark, the magnitude of the forward bias current can provide crude information about
the frequency of recombination events occurring in a given device architecture.
To develop a conceptual framework to assess observed device behavior based on molecular
properties, the OPV current-density (J) may be represented as a function of voltage (V), according to an
equivalent circuit model, to obtain
5

(3.1)
where R
s
and R
p
represent all series and parallel resistive elements, J
ph
the photocurrent density under
illumination, n the ideality factor, and J
s
the saturation current-density, respectively. In the recombination
limited forward bias regime (n ≈ 2)
6
the saturation current-density may expressed in terms of k
rec
as
(3.2)
66
where q is the elementary charge, and [D
+
] and [A
-
] are, respectively, the donor and acceptor polaron
concentrations at the D/A interface. By solving Eq. 1 in the recombination regime for V(J = 0) = V
oc
and
substituting for Eq. 3.2, the photovoltage may be expressed as
(3.3)
where the approximations become valid as R
p
→ ∞, R
s
→ 0, and J
ph
» J
s
. In essence, the recombination
reaction appears to be a major photovoltage loss pathway in most OPV devices. Accordingly, as the
external load increasingly apposes the photovoltage, insuring the infrequency of recombination events is
key for arresting V
oc
losses.
Based on this V
oc
description, one route to minimizing voltage losses is to suppress the emergence
of electrically induced, non-photogenerated, polarons at the D/A interface. In principle, this is achievable
by suppressing charge injection efficiencies on either side of the heterojunction, holes or electrons,
respectively, on the anode or cathode sides, or by limiting the mobility of these charges leaking into the
D/A region. However, since the optimal balance between charge collection and parasitic polaron
concentration is impracticable to discern a priori, this strategy often leads to unintendedly low FF and J
sc
,
in typical bilayer D/A devices.
Alternatively, consider the Arrhenius behavior of the bimolecular recombination rate constant,
k
rec
, in Eqs. 2 and 3. This quantity has been described
7
according to Marcus theory for outer sphere
electron transfer
8
and is given by
(3.4)
67
where h is Planck’s constant, H
ij
is the electronic coupling matrix element between the initial CT state and
the final neutral state, λ is the free energy for geometric reorganization, and ΔG° is the total free energy
change for the recombination reaction. Quite common is the observation that V
oc
often trends with the
interfacial energy offset, ΔE
DA
= E
i
- E
a
, between the ionization energy (E
i
) for removing an electron from
the donor highest occupied molecular orbital (HOMO) energy level and the electron affinity (E
a
) associated
with the acceptor lowest unoccupied molecular orbital (LUMO) energy level. This derives from the fact
that in most cases the charge transfer state, resulting from photoinduced electron transfer, is sufficiently
destabilized in energy, relative to the ground state, that non-radiative recombination occurs in the Marcus
inverted regime, due to the quadratic dependence of the free energy of activation ΔG* = λ/4(1+ΔG°/λ)
2

on
the free energy of reaction. Concisely, in this regime ΔG* increases with increasing ΔE
DA
and the rate of
charge recombination decreases. In practice, employing this strategy to limit V
oc
losses tends to diminish
J
sc
, due to optical absorption redundancy in the donor and acceptor layers. Finally, although compelling for
their potential to limit charge recombination without exacerbating optical redundancy, systems for which
the H
ij
term in Eq. 4 is vanishingly small, due to steric effects, tend to suffer from poor absorption
coefficient, exciton diffusion length, and charge carrier mobility. Balancing these counterposing
requirements is the impetus for the bookend donor concept, which the data presented here suggests as a
promising avenue for enhancing future device performance.
3.2.4 Materials and methods
Tetracene (Aldrich; 98%), rubrene (Aldrich), pentacene (Aldrich; 93.5%), aluminum
phthalocyanine chloride (Aldrich; 85%), C
60
(MER; 99+%), bathocuproine (Aldrich; 96%), and
platinum(II) octaethylporphyrin (Frontier Scientific) were obtained from commercial sources and purified
via thermal gradient sublimation (~ 0.2 µTorr). Aluminum (Alfa Aesar; 99.999%), copper (Arcor
Electronics), and molybdenum trioxide (Aldrich; 99.5%) were obtained from commercial sources and used
as received. Glass substrates commercially coated with ITO (Thickness: 1500±100 Å, Sheet Resistance: 20
± 5 Ω/cm
2
, Transmission: 84% at 550 nm) were purchased from Thin Film Devices Inc.
68
Device substrates were solvent cleansed and placed in an ozone atmosphere (UVOCS
T10X10/OES) for 10 minutes immediately before they were loaded into the high vacuum (~2 µTorr)
deposition chamber (Kurt J. Lesker Company). Layer thickness and deposition rates were monitored by
quartz crystal microbalance (Inficon) calibrated using monochromatic (Rudolph Technologies, Inc.; Auto
EL) or spectroscopic (J.A. Woollam Co., Inc.; WVASE32) ellipsometry. Condensed phase optical
measurements were performed on solvent cleansed glass, quartz, or polished silicon substrates using an
Agilent 8453 spectrophotometer, Photon Technology International fluorimeter, and Perkin Elmer Lambda
950 spectrophotometer coupled to a UV/Vis/NIR integrating sphere module. Atomic force microscopy
(AFM) was performed using a Digital Instruments Nanoscope® Dimension 3100 atomic force microscope.
Grazing incidence X-ray diffraction measurements were performed on a Rigaku Ultima IV diffractometer
using Cu Kα radiation source (λ = 1.54 Å)
Current voltage measurements were performed in ambient atmosphere using a Keithly 2420
SourceMeter® in the dark and under corrected 1000 Wm
-2
white light illumination from a 300W Xenon arc
lamp (Newport® Oriel Product Line). Spectral mismatch correction was performed using a silicon
photodiode (Hamamatsu S1787-04; 8RA filtered or S1787-12; KG5 filtered) calibrated at the National
Renewable Energy Laboratory (NREL). Frequency modulated illumination (250 Hz, 10 nm FWHM) from
a Cornerstone
TM
260 ¼ M monochromator (Newport® 74125) was used in conjunction with an EG&G
7220 DSP Lock-In amplifier in all spectral responsivity measurements.
3.2.5 Results and discussion
Voltage losses were compared in archetypal single-layer donor devices based on tetracene (Ttn), a
molecule with a relatively exposed delocalized π-system, with devices fabricated from several aryl-
substituted Ttn derivatives shown in Figure 3.2a. Here an ITO/ Donor (600Å)/ C
60
(400Å)/ BCP(100Å)/ Al
configuration was used, where Donor denotes Ttn, 5,6-diphenyltetracene (Dpt) or rubrene (Rbn) donor
layers. These three donor materials all exhibit ionization energies of ~ -5.4 eV.
9,10
Since the corresponding
∆E
DA
is invariant amongst these single donor devices, one might expect the V
oc
to be identical. However, in
accord with previous observations,
11,12
their JV curves in Figure 3.2b are quite dissimilar. The tetracene
69

based device exhibits a modest J
sc
= 2.58 mA/cm
2
and V
oc
= 0.54 V, with FF = 0.58. The V
oc
of the Rbn
(0.90 V) and Dpt (0.98 V) devices are significantly higher, while their respective J
sc
(2.15 mA/cm
2
and 1.38
mA/cm
2
) and FF (0.47 and 0.41) are significantly lower than the Ttn based device. Note that the dark
current for V ≤ 0.6 V and V ≤ 0.4 V, of both the Dpt and Rbn devices, respectively, is below the detection
limit of the instrumentation. In the Rbn based device, this effect has been attributed to the limited
accessibility of the π-system imposed by the four phenyl rings oriented orthogonal to the tetracene core and
partially shielding it from engaging in recombination events. This essentially amounts to a suppression of
the H
ij
term in Eq. 4. Presumably, a similar effect arises due to the two orthogonal phenyl rings in Dpt.
However, while Ttn forms polycrystalline thin films, both Dpt and Rbn form highly frustrated amorphous
thin films, where exciton diffusion and charge transport are expected to suffer. These results illustrate the
tradeoff between materials that lend themselves to facile exciton diffusion and charge transport (Ttn),
contrasted with materials that tend to minimize photovoltage losses (Dpt and Rbn), suggesting their
possible utility in formulating a viable bookend donor structure.
To assess these, potentially compatible, donor materials for such bookend structures, a rational set
of preliminary design principles was developed, to which each material should conform. The energy
a) b)
Figure 3.2. a) Chemical structures for tetracene (Ttn), 5,6-diphenyl-tetracene (Dpt), rubrene (Rbn), 5,6-
dinaphthyl-tetracene (Dnt), pentacene (Ptn), chloroaluminum phthalocyanine (ClAlPc), fullerene (C
60
), and
bathocuproine (BCP). b) Current voltage characteristics of single Ttn (black), Dpt (red), or Rbn (blue) 600Å
donor layer OPV devices comprising, ITO/ donor/ C
60
(400Å)/ BCP(100Å)/ Al, under AM1.5G 1 sun illumination
corrected for spectral mismatch corrected ASTM G173-03 and in the dark (filled symbols).

70

ordering of lowest optically accessible excitonic states (E
00
) for materials B
1
– B
3
must be B
3
> B
2
> B
1
,
forming an excitation transfer pathway, to funnel the excitation occurring in each material to the D/A
interface. However, to insure monopolar charge transport, the ionization energy of the exciton (E
i
*
) of, for
example, B
2
relative to the electron affinity of material B
1
, must engender endergonic exciton dissociation
at that interface, as is expected for this series of phenyl-substituted tetracene derivatives.
13
Hole transport
a)
b) c)
Figure 3.3. a) Neat film excitation (Ex) and emission (Em) spectra illustrating the cascading exciton energies for
the series diphenyltetracene (Dpt), tetracene (Ttn), and rubrene (Rbn) and the proposed mechanism of Förster-
assisted interlayer excitation transfer between singlet S
1
states of the lamellar donor materials, and subsequent
charge transfer quenching of the rubrene exciton leading to the charge transfer CT
n
state at the donor/acceptor
interface. b) External quantum efficiency signal for photovoltaic devices comprising ITO/ Donor / C
60
(400 Å)/
BCP (100 Å)/ Al, where Donor is a 650 Å Ttn film coated with and additional 50 Å of either Rbn (filled symbol),
Dpt (cross symbol), or Ttn (open symbol). c) Optical excitation energies derived from a), suggesting excitation
transfer from Ttn to Dpt will be endothermic by ~ 4kT, while excitation transfer from tetracene to rubrene will be
slightly exothermic resulting in the EQE traces in b).
71
from B
1
to B
2
etc. should be either thermoneutral or slightly exothermic, to alleviate interfacial charge
trapping. Accordingly, the Dpt, Ttn, Rbn series is ideal in this regard as well.
Although oriented orthogonally to the tetracene core, the phenyl rings in Dpt and Rbn couple
slightly to the aromatic π-system, bathochromically shifting the solution UV-vis absorption features
compared to Ttn by λ ~10 and ~30 nm, respectively. However, due to molecular aggregation, the thin film
excitonic absorption spectrum for tetracene is bathochromically shifted significantly in contrast with its
solution spectrum, while the respective solution and neat film spectra for Dpt and Rbn are more or less
invariant. As a result, the S
1
singlet excitonic states for the Dpt, Ttn, Rbn series may potentially form an
efficient excitation-cascade, as illustrated by the optical excitation and emission spectra in Figure 3.3a. In
this scheme, optical energy absorbed by Dpt is transferred to Ttn via Förster excitation transfer and,
likewise, Ttn excitation to Rbn. Based on this energy ordering, interfacial exciton dissociation and trapping
of holes are not expected at the Dpt/Ttn or Ttn/Rbn interfaces.
9,10,13

To characterize the excitation transfer process the spectral response for 650 Å thick nanocrystaline
tetracene films grown from a single deposition, on three pristine ITO substrates were compared.
Subsequently, each was coated with 50 Å of Ttn, Dpt, or Rbn, using a combinatorial masking array. This
architecture derives its primary acene absorption from the thick Ttn layer and, more importantly, insures
uniform Ttn crystal growth, and hence uniform L
D
, among all three substrates. Thus, disparities in the
external quantum efficiency (EQE) curves for ITO/ donor (700 Å)/ C
60
/ BCP/ Al devices fabricated
according to this structure are expected to primarily reflect the excitation transfer efficiency at the
Ttn/oligoacene interface, rather than reflecting differences in absolute absorbance or exciton diffusion
length. In Figure 3.3b EQE traces are presented for the Ttn/Ttn and Ttn/Rbn devices, in contrast with the
Ttn/Dpt device. For both the Ttn- and Rbn-capped devices, virtually identical EQE traces are observed.
However, the Dpt-capped device exhibits a 32% reduction in EQE. In examination of the exciton energies
for Dpt, Ttn, and Rbn, this is in fact the logical result. As illustrated in Figure 3.3c, excitation transfer from
the Ttn layer to the Rbn layer is slightly exothermic. Therefore, the measured EQE response is virtually
identical for Ttn/Ttn and Ttn/Rbn devices. However, based on the exciton energies, excitation transfer
from Ttn to Dpt is an endothermic process, requiring approximately four times more thermal energy than
72

available at room temperature. As a result, migration of excitons generated in the Ttn layer to the D/A
interface appears to be occluded by the Dpt layer, leading to a loss in EQE signal. These data characterize
the impact of excitation transfer efficiency to be quite significant in determining device performance for the
prospect of future bookend device structures. Hence, the properties of a model tetracene-based bookend
excitation-cascade system were further investigated according to Figure 3.1, where B
1
= Rbn, B
2
= Ttn, and
B
3
= Dpt.
Given the prospect for tetracene-based devices bookended by Dpt at the ITO/organic interface,
implied by the spectra in Figure 3.3a and the disparity in V
oc
between the single donor Dpt and Ttn devices
in Figure 3.2b, it is compelling to characterize the blocking nature against leakage of holes from ITO at the
ITO/Dpt interface compared to that from ITO at the ITO/Ttn interface. Accordingly, ITO/ACENE/Cu
devices were fabricated with nominally hole-selective contacts, where ACENE represents a single neat
layer of either Dpt or Ttn. Given similar work functions for these two electrodes (φ
F
≈ 4.8 eV), hole
injection from the positively biased ITO electrode is favored over electron injection from the Cu electrode
held at ground potential. The resulting field dependence for the Dpt device in Figure 3.4 exhibits a

Figure 3.4. Current-density as a function of applied field for ITO/ACENE/Cu devices, where ACENE = Ttn
(circle) or DPT (diamond) and positive potential applied to ITO.
73

relatively high-field current onset |E| = 68 kV/cm, approximately three times that for the Ttn device with
current onset at |E| = 20 kV/cm. These results suggest that passivating the ITO surface with a thin layer of
Dpt in the bookend structure may be beneficial for suppressing leakage current losses due to parasitic hole-
injection at the ITO/organic interface.
The amorphous nature of neat thin-film Dpt is likely to adversely frustrate in situ Ttn crystal growth on
Dpt during deposition. Understanding such effects is important for accurately assessing device
performance, since charge mobility
3,14
and exciton dynamics
4,15
can both be strongly linked to film
morphology. To interrogate the nature of Ttn film growth, atomic force microscopy (AFM) was employed
to compare the surface topography and grazing incidence x-ray diffraction (GIXD) was employed to
compare the morphology of thermally-grown Ttn films on Dpt-coated ITO, against Ttn films grown on
pristine ITO as shown in Figure 3.5. The AFM topographs for Ttn films, grown on ITO/Dpt and Ttn grown
on ITO, exhibit similar submicron features, with average height on the order of 30 nm, mean roughness
around 7 nm and RMS around 9 nm. Moreover, the GIXD spectra for ITO/Dpt(50Å)/Ttn(600Å) and
ITO/Ttn(600Å) samples both exhibit Ttn (001) peak appearing at ca. 2θ = 7.1° with identical peak widths
(FWHM = 0.62°). However, the intensity in the ITO/Dpt/Ttn spectrum is approximately 20% that of the
ITO/Ttn sample, implying fewer crystalline domains in the ITO/Dpt/Ttn material. Thus, the Ttn film
grown on Dpt appears to contain a large fraction of amorphous material. Although slightly lower hole
a) b)
Figure 3.5. a) Grazing incidence x-ray diffraction peaks for 600 Å thick tetracene (Ttn) films grown on pristine
ITO (black filled) or ITO coated with 50 Å Dpt (red open), diphenyltetracene. b) Schematic representation and
topographical images obtained by atomic force microscopy corresponding to samples in a).

74

mobility
16
and exciton diffusion length (L
D
∝ 1/a
2
, where the hopping distance a is large in the amorphous
phase)
4
for Ttn deposited on Dpt may be expected in contrast to Ttn on ITO, post deposition thermal
processing has been shown to efficiently affect amorphous-to-polycrystalline Ttn conversion.
17
These data
suggest that both Ttn on Dpt and Ttn on ITO, are viable routes to nanocrystaline Ttn thin films.
In light of these preliminary results, model bookend solar cells were fabricated comprising ITO/
Dpt(x)/ Ttn(y)/ Rbn(z)/ C
60
(400 Å)/ BCP(100Å)/ Al. The total donor thickness was given by (x + y + z) =
450 Å, to evaluate how such a structure might compare against the model tetracene-based single (y = 450
Å) donor device. In Figure 3.6 the, JV relationship for a typical bookend (x = 50 Å, y = 200 Å, z = 200Å)
donor device is compared against a typical model single donor device in the dark and under spectral
mismatch corrected AM1.5G 1 sun illumination. For the latter V
oc
= 0.54 V, J
sc
= 2.86 mA/cm
2
, and FF =
0.54 is observed. However, for the bookend structure, a 200 mV enhancement in photovoltage is observed,

Figure 3.6. Dark electrical characteristics for Device B, the tetracene based bookend device (red filled),
relative to Device A, the single donor tetracene-based device (black filled), resulting in the 200 mV increase in
V
oc
for the bookend device characteristics under illumination (red cross) relative to the single donor-based
device (black open). Devices consist of ITO/ Donor (450 Å)/ C
60
(400 Å)/ BCP (100 Å)/ Al, where Donor is
either a single layer of neat Ttn or a lamellar array comprising Dpt (50Å)/ Ttn (200Å)/ Rbn (200 Å).

75

resulting in V
oc
= 0.74 V with comparable J
sc
= 2.55 mA/cm
2
and FF = 0.50. The very small discrepancy in
J
sc
between the bookend and single donor devices may be attributed to two factors. The first being, a
shorter L
D
based on the low intensity GIXD peak for Ttn (001) grown in Dpt as previously discussed, the
second being, low absorption coefficients in the Dpt and Rbn layers relative to Ttn as illustrated in Figure
3.7.
To clarify the relative impact of the Dpt layer compared with the Rbn layer on suppressing voltage
losses in Device B, (x + y) or (y + z) = 450 Å bookend devices were fabricated for various x, y, and z,
ranging from 50 - 250 Å. In both cases V
oc
enhancements of ca. 100 mV greater than the V
oc
for the single
donor Ttn-based model device were observed. From this, the order of magnitude reduction in forward bias
dark current for Device B, compared with Device A, may be attributed to both a leakage-suppressing effect
by the Dpt, as well as a reduction in k
rec
at the Rbn/C
60
interface, contrasted with the Ttn/C
60
interface.
Considering E
i
for both Dpt and Ttn are - 5.4 eV and based on the relatively high hole-injection current for

Figure 3.7. Thin film absorption coefficient (α) for Dpt (blue), Ttn (black), and Rbn (red).
76

Ttn illustrated in Figure 3.4 compared to Dpt, the apparent leakage-suppressing effect afforded by Dpt in
these bookend devices, appears to be primarily kinetic in nature, due to steric inaccessibility about its π-
system. Two methods were employed in testing this hypothesis. In the first approach, the tetracene core
was further isolated through chemical modification, with the expected result being suppression of both
carrier injection and carrier collection. Accordingly, 5,6-dinaphthyltetracene (Dnt), depicted in Figure
3.2a, was obtained from Vincent Barlier. Despite the orthogonal naphthalene moieties that protrude
approximately 3.5 Å perpendicularly out from the plane of the tetracene core, the estimated E
i
= - 5.37 eV
and E
i
* = - 2.92 eV for Dnt, which are virtually invariant from Dpt. Bookend devices were fabricated
comprising ITO/ Dnt(50 Å)/ Ttn(350 Å)/ Rbn(50 Å)/ C
60
(400 Å)/ BCP(100Å)/ Al. Indeed, as illustrated in
Figure 3.8, these Dnt-based bookend devices exhibited inflection points in their JV characteristics,
indicative of poor hole collection and inefficient separation of coulombically bound polaron pairs (cf.

Figure 3.8. Electrical characteristics for Dnt-based bookend device structure comprising ITO/ Donor/ C
60
(400
Å)/ BCP (100 Å)/ Al in the dark (filled) and under simulated solar illumination (open). Donor is given by Dnt
(50 Å)/ Ttn (350 Å)/ Rbn (50 Å).
77

Chapter 4), resulting in extremely poor FF = 0.32, with J
sc
= 2.28 mA/cm
2
and V
oc
= 0.60 V. These data
suggest that π-system accessibility can substantial influence device performance, irrespective of energy
level alignment. Thus, care must be taken in designing materials to accommodate favorable processes,
while suppressing loss pathways.
To further characterize the V
oc
enhancement observed in these bookend devices, the impact of
varying energy level alignment at the anode contact was investigated. For this tetracene-based single donor
and bookend donor devices were fabricated, nominally identical to those measured in Figure 3.6, but
prepared on ITO substrates coated with molybdenum trioxide for its exceptionally large work function.
18,19

In Figure 3.9, the resulting JV characteristics for such devices are compared. For single donor ITO/ MoO
3

(100 Å)/ organic/ Al devices, a V
oc
= 0.53 V was obtained, nearly identical to that for Device A on pristine
ITO. Likewise for the bookend donor ITO/ MoO
3
(100 Å)/ organic/ Al devices, V
oc
= 0.74 V was

Figure 3.9. Electrical characteristics for ITO/MoO
3
OPV devices comprising ITO/MoO
3
(100 Å) / Donor/ C
60

(400 Å)/ BCP (100 Å)/ Al. Donor represents the bookend structure Dpt (50 Å)/ Ttn (200 Å)/ Rbn (200 Å) in the
dark (red filled symbols) and under illumination (red cross symbols) or a single Ttn (450 Å) donor layer in the
dark (black filled symbols) and under illumination (black open symbols).
78

observed, which is also identical to that obtained for the Ttn-bookend donor structure (Device B) with a
bare ITO anode. The implication being, the observed V
oc
enhancement arises primarily from the kinetics of
molecular electron transfer, rather than changes in the built-in electrical potential imposed by the Fermi
level offset between anode and cathode.

To more clearly characterize the photocurrent generated by Device B, the Ttn-baced bookend
structure, single layer thin film transmittance measurements and center-mount optical transflectance
measurements were performed on organic multi-layer stacks, identical to those used in Devices A and B,
but supported on ITO-free glass substrates and uniformly coated with Al (1000Å). The resulting
absorbance data for the single Ttn and bookend donor structures suggest approximately 20% lower
absorption for Device B between λ = 500 and 550 nm. This is in accordance with the lower absorption
coefficient for Dpt and Rbn of α(522 nm) = 0.19 × 10
5
cm
-1
and 0.55 × 10
5
cm
-1
, respectively, relative to
that of Ttn of α = 1.20 × 10
5
cm
-1
in this region as shown in Figure 3.7. The external quantum efficiency
for Device B is only 61% that of Device A in this region, due in part to this lower photon absorption.
However, estimating an internal quantum efficiency at λ = 520 nm, according to IQE(λ) =
EQE(λ)/[1 – f

(λ)], where f

(λ) = ρ(λ) + T(λ) represents the loss function due to spectral reflectivity ρ(λ) and
transmittance T(λ), one obtains IQE ≈ 35% for the single Ttn donor device, but only IQE ≈ 23% for the
a) b)
Figure 3.10. a) Absorption spectra for thin films of Ptn (dotted red trace), C
60
(dashed black trace), and ClAlPc
(solid blue trace). b) Quantum efficiencies illustrating the excitation transfer fingerprint of the pentacene layer in
the bookend device comprising a Ptn (300Å)/ClAlPc (300Å) donor layer (solid red trace) compared with a single
300 Å ClAlPc donor layer (dashed black trace) in ITO/ donor/ C
60
(400Å)/ BCP(100Å)/ Al devices.

79

Ttn-based bookend sample. This suggests the slight photocurrent loss observed for Device B, compared to
the single donor device, is not simply due to diminished absorption. Rather, the low IQE likely stems from,
either poor exciton diffusion through the Ttn layer of the bookend structure, or inefficient excitation
transfer at the Ttn/Rbn interface. Based on the EQE data presented in Figure 3.3b for device composed of
high-quality nanocrystaline tetracene films grown simultaneously on identical pristine ITO substrates,
excitation transfer at the Ttn/Rbn interface is not expected to limit the IQE for Device B. Thus, this
disparity may be assigned as a primarily morphological effect, arising from frustrated in situ Ttn crystal
growth during device fabrication.
The prospect of bookend excitation-cascade structures, that pair complementary materials, to not
only suppress photovoltage losses, but to simultaneously achieve panchromaticity, renders compelling the
ability to demonstrate bookend devices with enhanced spectral coverage, leading to an increase in J
sc
.
While Ttn is an interesting organic optoelectronic material for its solid-state luminescence and high carrier
mobility, its applicability for solar applications is limited, due to its spectral redundancy with common
fullerene acceptors, such as C
60
. To extend the bookend concepts developed herein from Ttn and to
demonstrate their utility for real device efficiency enhancement, bookend structures using pentacene (Ptn)
were examined.
The principle vibronic progression of Ptn in the absorption spectrum in Figure 3.10a exhibits a
favorable bathochromic shift of ~150 nm compared with Ttn, for which the absorption intensity drops
precipitously between 525 and 550 nm (not shown). As a result, photocurrent for Ptn-based solar cells can
be 2-4 times that of their Ttn counterparts. Moreover, Ptn itself, has been shown to possess exceptional
Table 3.1. Representative performance metrics for bookend and single-donor organic solar cell
devices
Device
a
Donor J
sc

(mA/cm
2
)
b

V
oc
(V)
FF P
max

(mW/cm
2
)
b

A Ttn 2.86 0.54 0.54 0.83
B Dpt/Ttn/Rbn 2.55 0.74 0.50 0.95
C Ptn 5.60 0.25 0.40 0.55
D Ptn/ClAlPc 8.25 0.36 0.46 1.39
a

ITO/ Donor/ C
60
(400 Å)/ BCP (100 Å)/ Al, where Donor = Ttn (450 Å), Dpt (50 Å)/Ttn (200
Å)/Rbn (200 Å), Ptn (600 Å)/ClAlPc(300Å) for Devices A, B, C, and D, respectively
b
Under simulated 1 sun illumination, corrected for spectral mismatch to ASTM G173-03 AM1.5G

80
charge transport properties.
20
However, due to the shallow Ptn ionization energy of E
i
= - 4.9 eV, the
photovoltage for Ptn-based devices is generally very low, with V
oc
= 0.25 V. Thus, the aforementioned
bookend design principles for excitation transfer and charge transport alignment were applied to design a
bookended pentacene-based device for increased J
sc
and V
oc
, with complete FF retention. For clarity, a
simple two-layer bookend donor structure is considered here, where the Ptn film is grown directly on ITO
in both the single donor and bookend donor devices, omitting the B
3
layer suggested in Figure 3.1. Thus,
morphological variation in the Ptn layer may be neglected. With regard to V
oc
losses, aluminum
phthalocyanine chloride (ClAlPc), was incorporated at the D/A interface in the bookend structure, based on
its large ionization energy (E
i
= - 5.3 eV) and its axial chloride substitution and partially obscured π-
system. With regard to optical excitation, ClAlPc was selected as a complementary material for Ptn/C
60
,
due to its high optical transparency between 400 - 600 nm and its intense low energy Q-band signature at
λ
max
= 740 nm, below the Ptn optical absorption edge near 700 nm as illustrated in Figure 3.10a.
Consequently, excitation transfer occurring from Ptn to ClAlPc is expected to be exothermic and readily
identifiable in the quantum efficiency trace for a bookend device prepared with B
1
= ClAlPc and B
2
= Ptn.
To verify this, devices were fabricated with donor layers based on Ptn (300 Å)/ClAlPc (300 Å) and
compared the external quantum efficiency in Figure 3.10b with that for devices based on Ptn-free donor
layers. The Ptn signature, attributed here to excitation transfer from the acene to ClAlPc and subsequent
charge transfer quenching of the excitation-transfer induced ClAlPc excited state at the C
60
interface,
appears in the solid red trace for the Ptn/ClAlPc device as two broad (~ 50 nm) features centered near 590
nm and 675 nm with intensities of approximately 10% and 15%, respectively. The peak at 750 nm and the
shoulder near 650 nm correspond to direct optical excitation of ClAlPc, while the broad feature from 400 –
550 nm is primarily due to C
60
absorption. These data demonstrate the potential of pairing multiple
materials in bookend donor structures that absorb in complementary regions, to enhance spectral coverage,
resulting in efficient broadband photon capture. This constitutes a very promising result for the prospect of
increasing J
sc
, while suppressing voltage losses, potentially leading to higher device efficiency.
Given the spectral response data for the Ptn-based bookend structure in Figure 3.10b, suggesting
excitation transfer from Ptn to ClAlPc as a viable channel for photocurrent enhancement, the current-
81

voltage characteristics for Ptn-ClAlPc bookend structure devices were further interrogated. In the main
panel of Figure 3.11 the linear JV characteristics in the dark and under spectral mismatch corrected
AM1.5G simulated 1 sun illumination for single layer Ptn-based device are contrasted with those obtained
for bookend devices comprising a Ptn (600 Å)/ClAlPc (300Å) a donor layer (Devices C and D,
respectively, in Table 3.1). The single donor Ptn-based device (Device C) exhibits J
sc
= 5.60 mA/cm
2
, FF
= 0.40, and V
oc
= 0.25 V. By comparison, the Ptn-based bookend device (Device D) JV characteristics,
presented in red, exhibit a 44% enhancement in photovoltage, increasing to V
oc
= 0.36 V. According to the
inset semi-log scale trace in Figure 3.11, this enhancement results largely from suppressed dark current
losses in the bookend device, for which the forward bias dark current at V = 0.25 V is thirty times lower
than the single donor device. Owing to the low-energy spectral compliment afforded by the ClAlPc layer,

Figure 3.11. Current voltage characteristics for Device D, a Ptn-based bookend donor structure comprising [Ptn
(600Å)/ ClAlPc (300Å)] (red crosses) under simulated AM1.5G 1 sun illumination with spectral mismatch
correction to ASTM G173-03, compared with an archetypal [Pentacene (600Å)] single donor layer (black open
symbols). Complete device structure consists of Glass/ ITO/ Donor/ C
60
(400Å)/ BCP (100Å)/ Al (1000Å). Inset
illustrates suppressed dark current (black filled symbols) for Device D compared on a semi-log scale with Device
C, the Ptn single donor (red filled symbols).
82
the photocurrent obtained from Device D (J
sc
= 8.25 mA/cm
2
) is nearly 50% higher than the single donor
layer device (J
sc
= 5.60 mA/cm
2
). Thus, demonstrating the utility of deploying a multi-donor system
capable of engaging in donor-donor excitation transfer to enhance photocurrent, while limiting parasitic
recombination at the D/A interface to suppress voltage losses. Moreover, following photoinduced electron
transfer from ClAlPc to C
60
, the hole transport level alignment between B
1
and B
2
layers insures slightly
exothermic carrier injection from ClAlPc (E
i
= -5.3 eV) to Ptn (E
i
= -4.9 eV). As a result, no degradation is
observed in FF for the bookend structure. In fact, the FF = 0.46 for Device D is slightly higher than that
for the Device C, with FF = 0.40. In conjunction with the enhancements in J
sc
and V
oc
, this leads to an
overall power conversion efficiency enhancement of > 150% for the bookend structure.
3.2.6 Concluding remarks
In summary, focusing on the structure-function relationship between π-system accessibility and
organic solar cell performance for various electron donor molecules, the potential is demonstrated for
lamellar bookend architectures, which match donor materials with complementary properties, to suppress
voltage losses and simultaneously enhance photocurrent, with complete fill factor retention. Where
balancing apposing electrochemical processes is impracticable in single donor architectures, this
distributive bookend approach is compelling, since it effectively partitions performance requirements
amongst several materials. Thus, the bookend structure has the advantage of relaxing design rules for each
specific component, dividing the rational development of new materials for enhanced device efficiency into
several more tractable problems: (i) excitation transfer efficiency between complementary materials must
act to funnel excitation to the charge transfer donor-acceptor interface; (ii) interfacial energy level
alignment within the donor array must impel photogenerated charge carriers toward the collecting electrode
to retain fill factor; (iii) materials with exposed aromatic π-systems exhibiting outstanding charge mobility
and exciton diffusion lengths should be bookended with interfacial materials featuring slightly shielded π-
electron systems to control interfacial electronic coupling for suppressed parasitic charge recombination;
83
(iv) complementary materials should be selected for limiting optical absorption redundancy between
materials within the donor array as well as with respect to the charge acceptor, to enhance the breadth of
spectral response.
In conjunction with the prospect for future synthetic, theoretical, and spectroscopic developments,
this bookend approach represents a broadly applicable structure, both as a tool for interrogating electronic
processes in condensed phase photochemistry and as a design element for enhanced organic photovoltaic
device efficiency. In this regard, the present demonstration represents a significant contribution, not only
to the field of organic electronics, but to the broader chemical community as well.
3.3 Host-guest sensitization systems
3.3.1 Absorption/diffusion mismatch
In addition to spectrally enhancing the breadth of photon capture, defining ensemble donor
systems to control exciton dynamics is also attractive for OPV solar energy conversion. Although the
absorption coefficient (α) associated with electronic transitions in π-conjugated chromophores relevant to
solar photon capture can be exceptionally high (cf. Section 1.4.4), the length (L
D
), being typically only on
the order of 5 – 20 nm,
21,22
that the resulting excited state can diffuse prior to deactivation is not. As
illustrated in Figure 3.12, an exciton diffusion length of five to ten times greater than this is required to
ensure that 100% of the absorbed solar photons generate excitons within L
D
from the donor/acceptor
interface. Here the solar photon flux Φ(λ) is derived from ASTM G173-03 Global and the cross-sectional
photon capture number is calculated as N
p
(λ, x) = Φ(λ) - Φ(λ)exp[-αx] for a given path length, x. Thus, for
a set of hypothetical donor materials with α profiles identical to copper phthalocyanine (CuPc), but
exhibiting a range of L
D
between 3 nm and 200 nm, a fractional solar photon flux N
p
(λ, L
D
)/Φ(λ), absorbed
84

within L
D
from the D/A heterojunction can be estimated for each material in a standard two-pass optical
path OPV architecture. The result in Figure 3.12 illustrates that for L
D
≈ 50 nm, nearly 90% of the incident
photons at λ = 625 nm, near the solar peak maximum, generate excitons within L
D
from the D/A interface
and are, therefore, viable excitons for photocurrent production. With L
D
≈ 100 nm, nearly 100% of the
incident solar photons in this region generate such viable excitons. Beyond L
D
≈ 200 nm, the absorbed
photon flux saturates, yielding N
p
/Φ = 1.0, meaning all incident photons generate excitons within L
D
from
the D/A interface. Based on this simple estimation, a modest increase in L
D
between a factor of five and a
factor of ten is required to insure complete photon absorption within an exciton diffusion length away from
the D/A interface.
Assuming exciton dissociation and carrier collection with near unit efficiencies, based on Figure
3.12, photocurrent losses of 60 – 80% are expected for materials exhibiting exciton diffusion lengths in the
5 – 20 nm range. This substantial loss in photocurrent is the impetus for the bulk heterojunction

Figure 3.12. Fraction of excitons N
p
(λ, L
D
)/Φ(λ) generated within a diffusion length (L
D
) from the D/A interface
relative to incident solar photon flux Φ(λ) for hypothetical donor materials. Assumes arbitrary L
D
, two-pass
optical path, and absorption coefficient identical to CuPc.
85
architecture presented in Chapter 1. While such structures can, in principle, achieve relatively high
photocurrents, the intricate phase segregation required on the nanometer length scale can render intractable
their use in fabricating, optimizing, and characterizing new materials systems and architectures.
This substantial limit to device performance underscores the potential gains that could be realized
through achieving enhanced exciton diffusion lengths. Since exciton diffusion losses may arise due to
premature relaxation of the excited state to the molecular ground state, controlling the exciton deactivation
mechanism may be a viable route to enhanced exciton diffusion. The radiative rates for depopulating the
singlet excited state for many organic systems are in the range of a few hundred picoseconds to tens of
nanoseconds, while those for triplet excited states are typically 10
3
-10
4

times less rapid. Thus, the lifetimes
(τ) for triplet states are expected to be three to four orders of magnitude longer compared with singlet
excitons. This behavior arises as a consequence of spin conservation for transitions between states of the
same multiplicity, rendering transitions between spin states of different multiplicities formally spin
forbidden.
23

3.3.2 Exciton diffusion length
Generally, exciton diffusion lengths may be expressed in terms of τ as L
D
= , where D is the
of the excitonic state through the medium. As discussed below, this relationship highlights another
distinction between singlet and triplet excitons in that the mechanism by which their diffusion occurs
strongly influences the relative magnitudes of D for each type of exciton. While diffusivities for triplet
excitons
24
are typically one to two orders of magnitude less than those for singlet excitons,
21
triplet
radiative deactivation rates are generally at least three to four orders of magnitude slower.
24,25
The
important implication being that, in principle, carefully prepared triplet excitons may deliver the critical
factor of five to ten needed to redress the diffusion/absorption length mismatch represented in Figure 3.12
Singlet exciton diffusion arises due to resonant interaction between transition dipole moments of
the excited and ground state molecules. This dipolar interaction can occur on length scales considerably
larger than the sum of their van der Waals radii. This form of energy migration is known as Förster
resonant excitation transfer (FRET) and is the primary mechanism for singlet exciton diffusion in the
86
OPVs. The FRET process can be visualized as a direct extension of the coulombic interaction between the
electric field of a photon and the π-system of a molecule. The elementary rate for this process may be
expressed as k
FRET
∝ j Κ
2
/n
4
τ
o
r
6
mol, where n is the index of refraction of the medium, τ
0
is the radiative
lifetime of the energy donor, r is the intermolecular separation, j is the spectral overlap integral, and Κ is an
orientation factor. This dipolar interaction is the primary mechanism for singlet excitation transfer. In
practice, predicting exciton dynamics using the above expression is complicated by molecular disorder and
the presence of trap states in the structures used in OPV devices.
The other non-trivial mechanism for exciton diffusion arises from electron exchange between a
chromophore in its electronically excited state and a ground state molecule in close proximity. This
process, known as Dexter excitation transfer (DET), occurs through direct orbital overlap and is, therefore,
limited to length scales on the order of the van der Waals radii of the two molecules. Since τ
0
for the triplet
excited state in conjugate organic molecules can be in the microsecond to millisecond regime, DET is the
primary mechanism for triplet exciton diffusion. Although the formal rate dependence for DET is
proportional to the spectral overlap integral and attenuates as k
DET
∝ exp(-2r), it contains terms that cannot
be easily related to physically measurable quantities, rendering its use in precisely predicting triplet exciton
diffusion lengths impracticable. Consequently, there is significant interest in developing accurate
methodologies for probing exciton dynamics in OPV structures.
4,21,26-30

Due to the spin-forbidden transition required in populating triplet excited states, optically
preparing triplet excitons in high yield is uncommon for most organic materials. While there is some
president for efficient intersystem crossing (ISC) in hydrocarbon materials, it is considered a relatively
exotic phenomenon. For example, near unit intersystem crossing quantum yields
31
in spheroidal C
60
leads
to triplet
32
exciton diffusion lengths reported
33
as high as 400 Å
22
and this material is widely employed as
an acceptor to achieve high performance. Additionally, relatively high intersystem crossing quantum yields
(Q
ISC
> 60%) in tetracene have been attributed
34
to facile relaxation from S
1
to T
2
and respectable OPV
efficiencies have been achieved by utilizing relatively thick (1000 Å) tetracene films.
17
While these
87
examples are rather anomalous, several potential routes to efficiently preparing and broadly utilizing triplet
excitons in organic solar energy conversion devices have been proposed.
3.3.3 Preparation of triplet states
Singlet exciton fission
35
to produce two triplet excited states between neighboring chromophores
has been proposed as another viable
36
route to efficiently preparing triplet excitons for solar energy
conversion. Identifying molecular properties that affect efficient singlet fission is an active area of
research.
37,38
Indeed, singlet fission in pentacene single crystals, a process that is exothermic by 0.1 eV, has
been shown to occur on a ca. 1 ps timescale. However, this pathway is suppressed in vapor deposited thin
films, where charge transfer and trapping at grain boundaries and impurity sites in such films leads to
extremely poor singlet fission yields.
39
Thus, understanding triplet excited state dynamics in the thin film
is crucial for the prospect of efficiently utilizing triplet excited states in OPV device applications. This
makes it compelling to identify organic materials systems for unambiguously preparing and probing triplet
exciton dynamics by both optoelectronic and spectroscopic techniques.
One such method is to employ a complementary host (H)-guest (G) system in which the radiative
lifetime of the host singlet excited state (
1
H*) is on the order of 0.1 – 10 ns, but for which ISC occurs on a
0.1 – 1 µs timescale.
25
Thus, direct optical excitation of pristine host material leads to singlet excitons,
since the excited state is depopulated at a rate which is much faster than the requisite time for intersystem
crossing to occur. However, triplet exciton generation in such materials can be optically accessed as
illustrated in Figure 3.13, through an intermolecular excitation transfer process (sensitization) occurring via
electron exchange with a guest material that does undergo rapid ISC, such as a transition metal complex
with significant spin-orbit coupling. This may be achieved by introducing a controlled impurity
concentration of guest material into the host matrix to insure direct orbital overlap for efficient electron
exchange. Some recent results employing platinum
40,41
and iridium
42
complexes as guest materials have
suggested that sensitized preparation of
3
H* triplet excitons may be a promising route to enhanced exciton
88

diffusion lengths. However, such systems exhibit either limited solar photon capture or redundant spectral
features that obfuscate analysis of the relative triplet exciton contribution. Therefore, it is the purpose of
the current study to define materials and device architectures for which the singlet and triplet optoelectronic
response relevant to solar photon capture may be spectrally resolved. This study lays crucial groundwork
for spectroscopically probing the transient species, characterizing the sensitization efficiency, and
determining the intermediary kinetics involved in the triplet sensitization process that will be reported
elsewhere.
43

3.3.4 Host-Guest materials
To identify prospective H-G systems, the estimated triplet energies for two oligoacenes employed
in section 3.2, tetracene (Ttn) and 5,6-diphenyltetracene (Dpt), both materials exhibiting
34
T
1
≈ 1.3 eV (λ ~
950 nm), were considered relative to the phosphorescence peak positions for two platinum porphyrin
complexes with relevance to organic optoelectronic devices, platinum(II) octaethylporphyrin (PtOEP)
44,45

and platinum(II) tetraphenyltetrabenzoporphyrin (PtTPBP).
46,47
These two platinum complexes undergo
ISC
48,49
on timescales significantly faster than their radiative rates of fluorescence and exhibit intense

Figure 3.13. Potential excitation transfer processes occurring in Host-Guest (H-G) sensitization systems.
89

emission at ca. λ
PtOEP
= 635 nm and λ
PtTPBP
= 775 nm from their T
1
states. Thus, they are potential
candidates for triplet excitation transfer to Ttn or Dpt. In contrast with tetracene, frustrated Dpt crystal
growth renders samples prepared using this material relatively insensitive to variations in process
conditions, such as deposition rate and substrate passivation. This is important when analyzing results
obtained for systems comprising pristine material in contrast to systems incorporating controlled impurity

Figure 3.14. Chemical structures and optical absorption for solid samples of PtOEP (dotted trace) and PtTPBP
(solid trace) in inert matrices of polystyrene and tetra(9,9ʼ-dimethylfluoren-2-yl)silane, respectively.
90

concentrations, which may themselves experience frustrated aggregation. Additionally, Dpt is attractive as
a model host material since the native triplet yields in Dpt are substantially suppressed, due to the
destabilization of the T
2
state relative to S
1
in Dpt, contrasted with Ttn.
34,50
Preliminary Stern-Volmer
phosphorescence quenching analysis of from these two potential porphyrin sensitizers by various tetracene
derivatives in 2-methyltetrahydrofuran suggests quenching of the porphyrin triplet emission occurs in the
diffusion controlled regime, with quenching constants on the order of 2 × 10
9

– 6 × 10
9
M
-1
s
-1
.
The
absorption features in Figure 3.14 for dilute PtOEP and PtTPBP in inert solid matrices illustrate their
intense Soret and Q band characteristic. This absorption profile makes these porphyrin molecules
compelling model Guest materials to pair with Dpt for their relatively high optical transmittance in the
wavelength region of λ = 450 – 525 nm, where the primary vibronic progression in the Dpt absorption

Figure 3.15. Absorption spectra (upper panel) for equal film thickness of Dpt (dotted trace) and Dpt-PtOEP
[20%] (solid trace). External quantum efficiency (lower panel) for devices incorporating a donor layer of Dpt
(dotted trace) or a Dpt-PtOEP composite (solid trace). Device structures consist of ITO/donor (300 Å)/ C
60
(400
Å)/ BCP (100 Å)/ Al.
91

spectrum occurs. This is crucial in developing a model system for clearly identifying the spectral
contribution of individual components.
3.3.5 Results and discussion
To assess the utility of employing these model sensitization systems for controlling exciton
dynamics, OPV devices in various architectures were fabricated using Dpt as a host matrix and either
PtOEP or PtTPBP as a triplet sensitizer. Overlaid in the upper panel of Figure 3.15, are the absorption
profiles for 300 Å films of neat Dpt (dotted trace) and a composite film (solid trace) of 20 % PtOEP by
weight in Dpt on glass substrates. For the neat Dpt film, relatively few photons are absorbed in the
wavelength range λ = 525 – 575 nm, however, the Dpt-PtOEP sample exhibits a peak in this region
resulting from the PtOEP Q-band absorption. The Dpt vibronic progression is prominent for both samples

Figure 3.16. External quantum efficiency demonstrating prominent PtOEP signal for OPV device incorporating a
guest/acceptor blocking layer (filled symbol trace) in contrast to Dpt-based single donor device (open symbol
trace). Blocking architecture given by ITO/Dpt-PtOEP [20%] (200 Å)/ Dpt (100 Å)/C
60
(400 Å)/ BCP (100 Å)/Al.
92

between λ = 425 – 525 nm, with the intensity loss in the Dpt-PtOEP sample arising from substituting
PtOEP for 20% of the Dpt material in this sample. The external quantum efficiency (EQE) data for OPV
devices incorporating the corresponding donor layers in the structure ITO/ donor (300 Å)/ C
60
(400 Å)/
BCP (100 Å)/ Al are presented in the lower panel of Figure 3.15. Due to the aforementioned PtOEP Q-
band absorption, the H-G device exhibits a prominent feature at ca. λ = 550 with EQE ~ 6% nm, while the
neat Dpt-based device is relatively featureless in this region with EQE > 2.5%. These data indicate that
optical excitation of PtOEP leads to photocurrent generation in the host-guest device. However, in this
architecture photocurrent arising from direct contact between PtOEP and C
60
at the D/A interface is
indistinguishable from the contribution os sensitized Dpt triplet excitons diffusing to the D/A interface.
Additionally, the EQE traces within the wavelength region between λ = 425 – 500 nm are virtually
identical for both devices, exhibiting a broad feature reminiscent of Dpt absorption convoluted with the
substantial fullerene response from the 400 Å thick acceptor layer.
To rule out the unlikely possibility that the PtOEP photocurrent response in these host-guet
devices might arise from direct photoinduced electron transfer between PtOEP and C
60
, a simple blocking
architecture was devised in which a neat blocking layer of Dpt was inserted between the H-G layer and the
acceptor layer. In this architecture, the blocking layer prevents the direct orbital overlap required for
electron transfer from PtOEP to C
60
, thus precluding any EQE response in this region due to porphyrin-C
60

charge transfer. The measure EQE response in Figure 3.16 for a guest/acceptor blocking architecture given
by ITO/ Dpt-PtOEP [20%] (200 Å)/ Dpt (100 Å)/ C
60
(400 Å)/ BCP (100 Å)/ Al are contrasted with the
EQE response from a single donor ITO/Dpt/C
60
OPV device. Encouragingly, these results demonstrate that
Table 3.2. Thickness dependence for host-guest Dpt-PtOEP devices
Thickness (Å)
a
J
sc
(mA/cm
2
)
b
V
oc
(V) FF P
max
(mW/cm
2
)
b

100 2.09 0.68 0.50 0.71
200 1.33 0.64 0.39 0.33
400 0.80 0.61 0.34 0.17
a

ITO/ Dpt-PtOEP [20%] (x)/ C
60
(200 Å)/ BCP (100 Å)/ Al,
b
Under simulated 1 sun illumination, corrected for spectral mismatch to ASTM G173-03 AM1.5G

93

a)
b)
Figure 3.17. a) Solid state absorption spectra (upper panel) for C
60
, Dpt and PtOEP. Thickness dependence of
external quantum efficiency (EQE) for Dpt-PtOEP host-guest devices. b) Current-density voltage characteristics
in the dark and under simulated AM1.5G 1 sun illumination with spectral mismatch correction to ASTM G173-03
for devices in a). Device structure consists of Glass/ ITO/ Dpt-PtOEP [20%] (x Å)/ C
60
(200Å)/ BCP (100Å)/ Al.
94
PtOEP response is retained with the incorporation of a neat Dpt blocking layer between the H-G layer and
the acceptor layer. Note that the shoulder near 625 nm in both EQE traces arises from C
60
absorption.
These data suggest that photocurrent produced by optical excitation of PtOEP occurs via triplet excitation
transfer from PtOEP to Dpt, with direct PtOEP electron transfer to C
60
being insignificant.
As implied in Figure 3.12, if the thickness L of the absorbing layer is given by L < L
D
, than as L →
L
D
, the external quantum efficiency is expected to increase as more photons are absorbed. However, in the
case where L ≥ L
D
, an increase in the absorbing layer thickness can lead to quantum efficiency losses, as
fewer of the absorbed photons generate excitons spatially viable for engaging in charge transfer at the D/A
interface. For this reason analyzing the dependence of the quantum efficiency on absorbing layer thickness
may aid in characterizing changes in exciton diffusion length. The lower panel in Figure 3.17a compares
the external quantum efficiency for PtOEP triplet sensitized OPV devices with increasing H-G layer
thickness between 100 Å – 400 Å. To minimize background fullerene response in these devices, while
ensuring uniform coating, the C
60
layer thickness is held fixed at 200 Å. As the thickness of the H-G layer
increases from 100 Å to 400 Å, a substantial loss in quantum efficiency is observed within the wavelength
range λ = 400 – 500 nm, where the materials primarily responsible for photon absorption are Dpt and C
60

(upper panel). However, as highlighted in the inset, within λ = 525 – 575 nm the EQE signal
corresponding to PtOEP absorption increases as the thickness of the H-G layer increases. These data
implicate direct optical excitation of PtOEP as a means of preparing and probing triplet excitons diffusion
in the Dpt host matrix. Given the loss in EQE signal between λ = 400 – 500 nm, indirection sensitization of
Dpt triplet excitons via Förster excitation transfer from
1
H* to
1
G* does not appear to be an efficient means
of achieving enhanced exciton diffusion length in this system. Thus, despite gains in EQE between λ = 525
nm and λ = 575 nm, the resulting short circuit current densities in Figure 3.17b under simulated 1 sun solar
illumination decrease from J
sc
= 2.09 mA/cm
2
for the 100 Å H-G layer to J
sc
= 1.33 mA/cm
2
and 0.80
mA/cm
2
for 200 Å and 400 Å H-G devices, respectively. The V
oc
attained by these devices is relatively
independent of thickness, varying by only 70 mV. As summarized in Table 3.2, the FF decreases slightly
with increasing H-G thickness, with the highest of FF = 0.50 for 100 Å.
95

The intense Q-band absorption for PtTPBP illustrated in Figure 3.14 within the wavelength range
λ = 600 - 650 nm is substantially red shifted (ca. 100 nm) contrasted with PtOEP due to benzannulation in
PtTPBP. This is an attractive feature in developing model systems for probing triplet sensitization since the
resulting guest signal is expected to be spectrally distinct from the host and the acceptor responses. As
such, data analysis for optoelectronic and spectroscopic signals becomes tractable, despite response from
multiple materials. This is demonstrated for device architectures identical to those in Figure 3.17, but
where substituting PtTPBP for PtOEP leads to the distinct Q-band feature between 600 nm and 650 nm in
the EQE traces in Figure 3.18. Qualitatively, the thickness dependence for PtTPBP-sensitized devices is
very similar compared with PtOEP, with increasing thickness exhibiting increasing photoresponse between
600 nm and 650 nm. Since this is the region of PtTPBP Q-band absorption, this signal is attributable to
photocurrent arising from
3
G* →
3
H* triplet excitation transfer. The increasing photoresponse in this

Figure 3.18. External quantum efficiency for Dpt-PtTPBP host-guest devices suggesting that in a 400 Å donor
triplet excitons efficiently reach the D/A interface, while singlets do not. Device structure consists of Glass/ ITO/
Dpt-PtTPBP [20%] (x Å)/ C
60
(200Å)/ BCP (100Å)/ Al.
96

region with increasing H-G layer thickness suggests facile diffusion of the resulting
3
H* excitons to the
D/A interface over at least 400 Å. By comparison, the photoresponse between λ = 400 nm and λ = 500 nm
peaks at EQE = 20% for the 100 Å H-G device, diminishing to 15% for 200 Å and 10% for 400 Å. This
inverse relationship between H-G layer thicknesses for PtTPBP devices is also qualitatively similar to
PtOEP and suggests inefficient indirect sensitization upon Dpt singlet excitation.
In Figure 3.19, the JV characteristics in the dark and under simulated 1 sun solar illumination for
the 100 Å, 200 Å, and 400 Å Dpt-PtTPBP devices are compared. The H-G device incorporating a 100 Å
composite donor layer exhibits a modest power conversion efficiency of ca. 1%, owing to FF = 0.56, J
sc
=
2.0 = mA/cm
2
, and V
oc
= 0.85 V. As may be expected based on the EQE data, increasing the H-G layer
thickness to 200 Å and 400 Å results in a decreases in J
sc
of ca. 0.65 mA/cm
2
and 0.85 mA/cm
2
,
respectively. The relatively high photovoltage is virtually invariant with H-G thickness. The thickness

Figure 3.19. Thickness dependence of DPt-PtTPBP host-guest electrical characteristics in the dark and under
simulated AM1.5G 1 sun illumination with spectral mismatch correction to ASTM G173-03. Device structure
consists of Glass/ ITO/ Dpt-PtTPBP [20%] (x Å)/ C
60
(200Å)/ BCP (100Å)/ Al.
97

dependent EQE characteristics for host-guest devices incorporating PtOEP and PtTPBP are qualitatively
similar at short circuit; however, the current voltage characteristics are qualitatively distinct. Specifically,
as the H-G layer thickness is increased from 100 Å to 400Å, the PtTPBP characteristics begin to exhibit an
inflection point in the JV data under illumination. This feature may be indicative of poor charge carrier
collection (cf. Chapter 4). Comparing the ionization energies for the two sensitizers
1
PtOEP (E
i
= - 5.3 eV)
and PtTPBP (E
i
= - 4.8 eV) relative to that of the Dpt host (E
i
= - 5.4 eV), trapping of holes on PtTPBP is
expected to be a substantial impediment to carrier collection as the H-G layer thickness increases. As a
result, the FF values summarized in Table 3.3 for the PtTPBP H-G devices decrease significantly with
increasing donor layer thickness. In this respect, Dpt-PtTPBP is not ideal as a model triplet sensitization
system, since poor carrier collection can impact the overall J
sc
and slightly complicate analysis of the
optoelectronic response.
3.3.6 Concluding remarks
In summary, the evaluation of thin film triplet exciton dynamics is expected to be crucial in
potentially utilizing these long lived excited states for enhanced organic solar cell performance. A host-
guest device architecture was employed as a means of identifying model triplet sensitization systems for
OPV devices. Using 5,6-diphenyltetracene (Dpt) as a host material and phosphorescent platinum porphyrin
complexes platinum(II) octaethylporphyrin (PtOEP) and platinum(II) tetraphenyltetrabenzoporphyrin
(PtTPBP) as guest materials, and evaluating the thickness dependence of the corresponding external
quantum efficiency, direct optical excitation of the guest material appears to generate Dpt triplet excitons
with long (400 Å) range viability. While PtTPBP is attractive in this system for its unique spectroscopic
Table 3.3. Thickness dependence for host-guest Dpt-PtTPBP devices
Thickness (Å)
a
J
sc

(mA/cm
2
)
b

V
oc

(V)
FF P
max

(mW/cm
2
)
b

100 2.03 0.85 0.56 0.97
200 1.38 0.85 0.44 0.52
400 1.18 0.84 0.28 0.28
a

ITO/ Dpt-PtTPBP [20%]/ C
60
(200 Å)/ BCP (100 Å)/ Al,
b
Under simulated 1 sun illumination, corrected for spectral mismatch to ASTM G173-03 AM1.5G

98
signature between λ = 600 nm and λ = 650 nm, its shallow ionization energy relative to Dpt appears to
result in charge trapping, limiting the applicability of this system in solar energy conversion devices.
While the more negative ionization energy of PtOEP appears to mitigate this problem, its spectral signature
is slightly obscured by the optoelectronic response of the host and acceptor materials, making PtOEP less
attractive for fundamental mechanistic investigations. The present study lays the groundwork for future
analysis of triplet exciton transient dynamics and for the development of new materials and architectures
for enhanced efficiency in OPV solar energy conversion devices.
99
3.4 Chapter 3 endnotes
1 M. D. Perez, C. Borek, S. R. Forrest and M. E. Thompson, J. Am. Chem. Soc., 2009, 131, 9281.
2 V. Coropceanu, J. Cornil, D. A. da Silva, Y. Olivier, R. Silbey and J. L. Bredas, Chem. Rev., 2007,
107, 926.
3 B. Servet, G. Horowitz, S. Ries, O. Lagorsse, P. Alnot, A. Yassar, F. Deloffre, P. Srivastava, R.
Hajlaoui, P. Lang, et al., Chem. Mater., 1994, 6, 1809.
4 R. R. Lunt, J. B. Benziger and S. R. Forrest, Adv. Mater., 2010, 22, 1233.
5 A. Moliton and J. M. Nunzi, Polym. Int., 2006, 55, 583.
6 D. A. Neamen, Semiconductor Physics and Devices, Third Edition edn., McGraw-Hill, New York,
2003.
7 P. F. Barbara, T. J. Meyer and M. A. Ratner, J. Phys. Chem., 1996, 100, 13148.
8 R. A. Marcus, Rev. Mod. Phys., 1993, 65, 599.
9 N. Sato, H. Inokuchi and E. A. Silinsh, Chem. Phys., 1987, 115, 269.
10 N. Sato, K. Seki and H. Inokuchi, J. Chem. Soc., Faraday Trans., 1981, 77, 1621.
11 A. K. Pandey and J. M. Nunzi, Adv. Mater., 2007, 19, 3613.
12 C. W. Chu, Y. Shao, V. Shrotriya and Y. Yang, Appl. Phys. Lett., 2005, 86, 243506.
13 H. J. Ding and Y. L. Gao, Appl. Phys. A, 2009, 95, 89.
14 I. McCulloch, M. Heeney, C. Bailey, K. Genevicius, I. Macdonald, M. Shkunov, D. Sparrowe, S.
Tierney, R. Wagner, W. M. Zhang, et al., Nat. Meter., 2006, 5, 328.
15 V. K. Thorsmolle, R. D. Averitt, J. Demsar, D. L. Smith, S. Tretiak, R. L. Martin, X. Chi, B. K.
Crone, A. P. Ramirez and A. J. Taylor, Phys. Rev. Lett., 2009, 102, 017401.
16 H. Graaf, W. Michaelis, G. Schnurpfeil, N. Jaeger and D. Schlettwein, Org. Electron., 2004, 5,
237.
17 Y. Shao, S. Sista, C. W. Chu, D. Sievers and Y. Yang, Appl. Phys. Lett., 2007, 90, 103501.
18 R. Tokarz-Sobieraj, K. Hermann, M. Witko, A. Blume, G. Mestl and R. Schlogl, Surf. Sci., 2001,
489, 107.
19 Irfan, H. J. Ding, Y. L. Gao, D. Y. Kim, J. Subbiah and F. So, Appl. Phys. Lett., 2010, 96, 73304.
100
20 S. F. Nelson, Y. Y. Lin, D. J. Gundlach and T. N. Jackson, Appl. Phys. Lett., 1998, 72, 1854.
21 R. R. Lunt, N. C. Giebink, A. A. Belak, J. B. Benziger and S. R. Forrest, J. Appl. Phys., 2009, 105,
053711.
22 P. Peumans, A. Yakimov and S. R. Forrest, J. Appl. Phys., 2003, 93, 3693.
23 G. A. Crosby, K. W. Hipps and W. H. Elfring, J. Am. Chem. Soc., 1974, 96, 629.
24 E. A. Silinsh and V. Čápek, Organic Molecular Crystals, American Institute of Physics, New
York, 1994.
25 N. J. Turro, Modern Molecular Photochemistry, University Science Books, Sausalito, CA, 1991.
26 H. Fidder, J. Knoester and D. A. Wiersma, J. Chem. Phys., 1991, 95, 7880.
27 T. Q. Nguyen, I. B. Martini, J. Liu and B. J. Schwartz, J. Phys. Chem. B, 2000, 104, 237.
28 T. Stubinger and W. Brutting, J. Appl. Phys., 2001, 90, 3632.
29 D. E. Markov, E. Amsterdam, P. W. M. Blom, A. B. Sieval and J. C. Hummelen, J. Phys. Chem.
A, 2005, 109, 5266.
30 S. R. Scully and M. D. McGehee, J. Appl. Phys., 2006, 100, 034907.
31 M. Lee, O. K. Song, J. C. Seo, D. Kim, Y. D. Suh, S. M. Jin and S. K. Kim, Chem. Phys. Lett.,
1992, 196, 325.
32 J. W. Arbogast, A. P. Darmanyan, C. S. Foote, Y. Rubin, F. N. Diederich, M. M. Alvarez, S. J.
Anz and R. L. Whetten, J. Phys. Chem., 1991, 95, 11.
33 L. A. A. Pettersson, L. S. Roman and O. Inganas, J. Appl. Phys., 1999, 86, 487.
34 C. Burgdorff, T. Kircher and H. G. Lohmannsroben, Spectrochimica Acta Part a-Molecular and
Biomolecular Spectroscopy, 1988, 44, 1137.
35 K. C. Kao, Dielectric Phenomena in Solids, Elsevier Academic Press, London, 2004.
36 C. C. Gradinaru, J. T. M. Kennis, E. Papagiannakis, I. H. M. van Stokkum, R. J. Cogdell, G. R.
Fleming, R. A. Niederman and R. van Grondelle, P. Natl. Acad. Sci. USA, 2001, 98, 2364.
37 A. F. Schwerin, J. C. Johnson, M. B. Smith, P. Sreearunothai, D. Popovic, J. Cerny, Z. Havlas, I.
Paci, A. Akdag, M. K. MacLeod, et al., J. Phys. Chem. A, 2010, 114, 1457.
38 I. Paci, J. C. Johnson, X. D. Chen, G. Rana, D. Popovic, D. E. David, A. J. Nozik, M. A. Ratner
and J. Michl, J. Am. Chem. Soc., 2006, 128, 16546.
101
39 V. K. Thorsmolle, R. D. Averitt, J. Demsar, D. L. Smith, S. Tretiak, R. L. Martin, X. Chi, B. K.
Crone, A. P. Ramirez and A. J. Taylor, Phys. Rev. Lett., 2009, 102.
40 P. Heremans, D. Cheyns and B. P. Rand, Acc. Chem. Res., 2009, 42, 1740.
41 B. P. Rand, S. Schols, D. Cheyns, H. Gommans, C. Girotto, J. Genoe, P. Heremans and J.
Poortmans, Org. Electron., 2009, 10, 1015.
42 W. A. Luhman and R. J. Holmes, Appl. Phys. Lett., 2009, 94, 153304.
43 S. T. Roberts, C. W. Schlenker, V. S. Barlier, S. W. Chin, R. McAnally, M. E. Thompson and S.
E. Bradforth, in preparation, 2010.
44 M. A. Baldo, D. F. O'Brien, Y. You, A. Shoustikov, S. Sibley, M. E. Thompson and S. R. Forrest,
Nature, 1998, 395, 151.
45 Y. Shao and Y. Yang, Adv. Mater., 2005, 17, 2841.
46 M. D. Perez, C. Borek, P. I. Djurovich, E. I. Mayo, R. R. Lunt, S. R. Forrest and M. E. Thompson,
Adv. Mater., 2009, 21, 1517.
47 C. Borek, K. Hanson, P. I. Djurovich, M. E. Thompson, K. Aznavour, R. Bau, Y. R. Sun, S. R.
Forrest, J. Brooks, L. Michalski, et al., Angew. Chem. Int. Ed., 2007, 46, 1109.
48 A. K. Bansal, W. Holzer, A. Penzkofer and T. Tsuboi, Chem. Phys., 2006, 330, 118.
49 Y. Sun, C. Borek, K. Hanson, P. I. Djurovich, M. E. Thompson, J. Brooks, J. J. Brown and S. R.
Forrest, Appl. Phys. Lett., 2007, 90, 213503.
50 C. Burgdorff and H. G. Lohmannsroben, J. Lumin., 1994, 59, 201.
102
Chapter 4
Molecular aspects of charge carrier collection in organic photovoltaics
4.1 Toward multipurpose buffer materials
An organic layer such as bathocuproine (BCP)
1
is commonly employed between the acceptor and
cathode of planar heterojunction organic photovoltaic (OPV) devices to preserve the native acceptor
properties against damage incurred during cathode deposition. Several detailed explanations for the
improvement
2
in device performance obtained from incorporating this buffer layer have been suggested,
such as suppression of parasitic exciton quenching
1
at the acceptor/cathode interface, local optical field
intensity enhancement
3
leading to increased photon harvesting
4,5
and the facilitation of a low resistance
cathode contact
6,7
leading to higher charge collection efficiencies.
Although, this BCP layer has the general potential to act as a powerful design feature for light
management and shunt passivation in large area device applications, charge transport in this material and
similar wide energy gap materials depends on a 100 Å compromised
5,8,9
region of indeterminate energy and
BCP:metal composition, produced during cathode deposition. This thickness dependence precludes using
these materials in applications where optimizing device efficiency requires an arbitrarily thick charge
transport layer. Additionally, there is evidence suggesting that a significant pathway for device failure may
be the growth of crystalline grain boundaries in bathocuproine.
1
Progress toward viable thickness tolerance
was realized
10
with the discovery that the metal complex tris(acetylacetonato)ruthenium(III), addressed
here as Ru(acac)
3
, could be incorporated at three times the thickness of bathocuproine into devices with an
indium tin oxide transparent anode, copper phthalocyanine donor, fullerene C
60
acceptor, and silver cathode
(ITO/CuPc/C
60
/buffer/Ag) without significant reduction in fill factor (FF). The illustration in Figure 4.1
depicts carrier collection in the Ru(acac)
3
buffer system as it is thought to proceed, via a Ru
III/IV
redox
couple, shuttling positive charge from the cathode to the acceptor/buffer (A/B) interface. This mechanism
is qualitatively different from transport of photogenerated electrons through the acceptor layer, where the
103

electron moves away from the interface between donor and acceptor by discrete hopping through the lowest
unoccupied molecular orbital (LUMO) of adjacent acceptor molecules. In the R(acac)
3
system, net current
flow is understood to arise from the motion of reciprocal positive charge, injected from the cathode onto
the highest occupied molecular orbital (HOMO) of the buffer material. This reciprocal carrier motion
facilitates electron collection by hole transport. Thus, rather than electron hopping through the LUMO of
adjacent molecules to the cathode, it is the reciprocal transport of positive charge through the HOMO of the
buffer complex that allows net electron extraction in this ruthenium based system.
Although realizing the cost reduction promised by the advent of OPVs requires minimal use of
cost prohibitive precious metal buffer complexes, it is compelling to first understand the charge collection
process, from a molecular standpoint, in this prototypical thickness tolerant ruthenium system. Moreover,
as a ground state neutral species, Ru(acac)
3
is a low spin d
5
open-shell metal complex,
11
and very little is
known about the implications of incorporating neutral paramagnetic materials into organic electronic

Figure 4.1. Mechanism for reciprocal carrier collection in double heterojunction OPVs. Photogenerated
electrons (black circles) in the acceptor layer are collected by recombination with holes (red circles) from the
reversibly oxidized buffer layer. Black arrows represent electron motion, while red arrows represent motion of
holes.

104
devices. In the present study, a series of tris(β-diketonato)ruthenium(III) analogues, bearing various
aromatic and electron withdrawing substituents, are analyzed in OPV devices as buffer materials on the
bases of their molecular structure and frontier orbital energy levels. Devices employing 100 Å thick buffer
layers all exhibit similar performance to that of the parent complex. However, upon increasing the buffer
thickness to 200 Å, a strong dependence on buffer identity is observed. These devices exhibit departures
from ideality marked by inflection points in the current density voltage characteristics. Photoelectron
spectroscopy was employed to interrogate the relevant interfacial energy levels for charge collection.
Finally, a simple physical model, based on a dual anti-polar diode (APD) equivalent circuit, was employed
to correlate this non-ideal behavior with molecular properties. The results suggest that charge collection in
these systems is limited by the rate of bimolecular charge recombination at the acceptor/buffer (A/B)
interface. As described by Marcus theory, this rate will depend on the free energy of the polaron pair (PP)
state relative to the ground state, arising from local orbital energy level alignment; on the electronic
coupling between the PP state (A
-
B
+
) and the ground state surfaces, arising from the extent of molecular
orbital interactions; and on the free energy of reorganization between the PP state and the ground state,
arising from spatial redistribution of the nuclei required for efficient electron transfer. The broader
implication for the general design of organic electronic devices being, one must attend to both
thermodynamic and kinetic parameters when considering the preparation of new high performance
materials.
4.2 Materials and methods
For the preparation of the studied metal complexes the following materials were purchased from
Sigma-Aldrich and used as received: ruthenium(III) chloride hydrate (99+%); 1,3-dephenyl-1,3-
propanedione (98%); 4,4-difluoro-1-phenyl-1,3-butanedione (97%); 4,4,4-trifluoro-1-phenyl-1,3-
butanedione (99%); 4,4,4-trifluoro-1-(2-naphthyl)-1,3-butanedione (99%); 4,4,4-trifluoro-1-(2-thienyl)-1,3-
butanedione (99%). As per the “ruthenium blue solution” method, preparation was carried out under inert
atmosphere (N
2
) and complexes were purified over an activated Alumina (neutral, Alfa Aesar) column with
benzene (EMD; 99.92%) as the eluant and recrystallized from ethanol or a 1:1 by volume mixture of
105

ethanol and benzene. The resulting products were characterized by electron impact ionization mass
spectrometry (HP 5973), UV-Vis spectroscopy (Agilent 8453), and solution cyclic voltammetry (EG&G
Potentiostat/Galvanostat, Model 283). All electrochemical measurements were performed under nitrogen
against ferrocene/ferrocenium (Aldrich; 98%) at 100 mVs
-1
in dry (CaH
2
: Aldrich; 95%) acetonitrile
(EMD; 99.99%) with tetrabutylammonium hexafluorophosphate (Aldrich; 98%) as the supporting
electrolyte. Silver reference, platinum counter, and glassy carbon working electrodes were used to couple
the samples to the potentiostat. The measured oxidation and reduction potentials presented in Table 4.1 are
in agreement with those reported in literature.
12

All ultraviolet photoelectron spectroscopy (UPS) measurements were performed by C. Kyle
Renshaw in the research laboratory of Professor Stephen R. Forrest at the University of Michigan. The
highest occupied molecular orbital (HOMO) energy levels of the RuL
3
buffer layers on ITO/C60 or Ag
were measured by UPS on commercial ITO coated glass. Substrates with ITO sheet resistance of <
15Ωcm
-2
, were solvent cleansed and UV-ozone treated prior to loading into an ultra-high vacuum (UHV)
chamber. Prior to the deposition of a 50 Å thick RuL
3
film: 1) 50 Å thick C60 films were deposited onto
ITO substrates or 2) 300 Å thick Ag films were deposited onto p-type Si. The samples were immediately
transferred to the analysis chamber for UPS measurements - all deposition and transfer was performed
under UHV (< 1 × 10
-9
Torr). The organic films were deposited by organic molecular beam deposition
Table 4.1. Ligand substitution, electrochemical redox potentials, and bulk orbital energy levels
Complex R R’ Ru
IV
/Ru
III
[a]
[V]
Ru
III
/Ru
II
[a]
[V]
HOMO[b]
[eV]
LUMO[c]
[eV]
1 Ru(acac)
3
CH
3
CH
3
+0.60 -1.16 -4.99 -3.40
2 Ru(bhba)
3
C
6
H
5
C
6
H
5
+0.73 -0.84 -5.39 -3.78
3 Ru(bhf
2
)
3
C
6
H
5
CF
2
H +1.05 -0.55 -6.15 -4.13
4 Ru(fhsa)
3
CF
3
C
4
H
3
S +1.20 -0.35 -6.44 -4.36
5 Ru(fhna)
3
CF
3
C
10
H
7
+1.22 -0.33 -6.59 -4.39
6 Ru(fhba)
3
CF
3
C
6
H
5
+1.28 -0.33 -6.66 -4.39
[a] Reversible half-wave potentials (± 0.1 V) vs.

Cp
2
Fe
+
/Cp
2
Fe
[b] Solid-state HOMO energy level (± 0.1 eV) measured by ultraviolet photoemission spectroscopy as in the text
[c] Estimated solid-state LUMO energy level (± 0.5 eV) calculated after reference 14
106
(OMBD) from Knudsen cells, while the Ag films were deposited by thermal evaporation. Thickness was
monitored by ellipsometrically calibrated quartz crystal microbalance. Spectra were collected using a
Thermo VG hemispherical electron energy analyzer with a pass function FWHM of 0.16 eV, calibrated by
fitting the Fermi step of a Au sample, to filter electrons photoemitted from the sample by a 21.22eV He(I)
emission line from a gas discharge lamp. To minimize sample charging, electrical contact to the ITO film
was maintained by a conducting clip attached to a copper puck and connected to an electrical feedthrough.
The sample was biased at -9.00 V to ensure that the low kinetic energy electrons pass through the analyzer.
For photovoltaic device fabrication, glass substrates commercially coated with ITO (Thickness:
1500±100 Å, Sheet Resistance: 20±5 Ωcm
-2
, Transmission: 84% at 550 nm) were provided courtesy of
Thin Film Devices Inc., Lot # 092606. Substrates were solvent cleaned and placed in an ozone atmosphere
(UVOCS T10X10/OES) for 10 minutes immediately before they were loaded into the high vacuum
deposition chamber (~2 µTorr). All ruthenium analogues, CuPc (Aldrich; 97%), C
60
(MER; 99+%), and
Ru(acac)
3
(Aldrich; 97%) were purified via thermal gradient sublimation (~0.2 µTorr). Organic and silver
(Alfa Aesar, 99.9999%) layer thicknesses and deposition rates were monitored by calibrated quartz crystal
microbalance (Inficon) to yield OPV devices with the structure ITO/ CuPc(400Å)/ C
60
(400Å)/ buffer/
Ag(1000Å).
The J(V) measurements were performed using a Keithly 2420 SourceMeter® in the dark and
under corrected 1000 Wm
-2
white light illumination from a 300W Xe arc lamp (Newport® Oriel Product
Line). Spectral mismatch correction was performed using a calibrated (NREL) silicon photodiode
(Hamamatsu S1787-04; 8RA Filtered). Chopped illumination (250 Hz, 10 nm FWHM) from a
Cornerstone
TM
260 ¼ M monochromator (Newport® 74125) was used in conjunction with an EG&G 7220
DSP Lock-In amplifier to make all spectral responsivity measurements. Antipolar diode modeling was
performed using the ProductLog[z] representation of the Lambert-W function in Mathematica
®
5.1.
107

4.3 Results and Discussion
A versatile synthetic approach to various ruthenium β-diketonate derivatives is the “ruthenium
blue” method of Endo et al.
13
depicted in Figure 4.2. Briefly, this one-step preparation entails refluxing
ruthenium trichloride in mildly reducing aqueous ethanol, introducing excess chelating ligand, and
subsequently quenching the liberated H
+
with multiple fractions of bicarbonate. Based on the parent
Ru(acac)
3
, substituted β-diketonate complexes 2-6 in Figure 4.3a were synthesized according to this
preparation method. Ligands were selected on the basis of their ability to maximize the vapor
processability, increase intermolecular electronic communication, and tailor the HOMO energy level of the
resulting complex. The HOMO energies reflecting the latter phenomenon experimentally determined by
UPS are plotted in Figure 4.3b, with LUMO energy levels estimated using electrochemical methods.
14

Depending on the magnitude of the electron withdrawing effects exerted by the β-position substituents
12,15

on the electron density of the chelating oxygen atoms
16
in each ligand, the d frontier orbitals of each metal
complex may be strongly stabilized via increasingly electron deficient chelating atoms. The numbering
scheme chosen in Figure 4.3 for the complexes correlates the compound number with the ordering of bulk
HOMO energies, i.e. a higher number represents increased HOMO stabilization. Thus, data presented in
order of compound number illustrates how the data varies with bulk HOMO energy.

Figure 4.2. Preparation method for compounds 2-6. Ruthenium trichloride is refluxed under mildly reducing
conditions to produce a mixed valent “ruthenium blue” solution. Addition of excess ligand and multiple fractions
of KHCO
3
to neutralize liberated hydronium ion yield the crude precipitate.

108

The device structure discussed here and depicted schematically in Figure 4.4a is a variation on the
archetypal donor/acceptor double heterojunction, consisting of a copper phthalocyanine (CuPc) donor, a
fullerene C
60
accepter, and the RuL
3
buffer material, sandwiched between an indium-tin-oxide (ITO) anode
and Ag cathode (ITO/ CuPc[400 Å]/ C
60
[400 Å]/ RuL
3
[x Å]/Ag[1000 Å]). Figure 4.4b compares the
current density as a function of voltage, J(V), for double-heterojunction OPVs incorporating a 100 Å thick
neat film of ruthenium complexes 1-6 inserted between the C
60
acceptor and the Ag cathode replacing the
typical bathocuproine layer. These data demonstrate that at 100 Å all of these materials exhibit similar
performance with power conversion efficiencies around 1% - 1.4% ±0.3%. Typical performance metrics
for these 100 Å buffer devices are summarized in Table 4.2 according to the ligand substitution of each
buffer complex. The short circuit current densities (J
sc
) for 2-6 of 5.0 – 5.7 mAcm
-2
are all slightly higher
than the 4.9 mAcm
-2
obtained from the original Ru(acac)
3
based device. The substituted β-diketonate
analogues 2-6 show fill factors (FF) of 0.58 ± 0.03, 0.45 ± 0.03, 0.38 ± 0.03, 0.49 ± 0.03, and 0.46 ±0.03,
respectively, that are comparable to the 0.52 ± 0.03 FF for the parent acetylacetonate complex. Similar
open-circuit voltages (V
oc
) between 428 ± 20 and 535 ± 20 mV were measured for all of these devices.
a) b)
Figure 4.3. a) The parent tris-acetylacetonate complex 1 and substituted analogues 2-6 used as buffer materials
between the C
60
acceptor and the Ag cathode. b) Bulk ionization energies (solid black) for complexes 1-6,
reflecting the electron withdrawing nature of the chelating ligands. Solid black lines represent HOMO energy
levels measured relative to vacuum, using ultraviolet photoelectron spectroscopy. Dashed gray lines represent
LUMO energy levels, estimated according to literature from measured solution electrochemical data in Table 4.1.
109

a)
b)
c)

Figure 4.4. a) Schematic representation of the studied device architecture, consisting of ITO/ CuPc(400 Å)/
C
60
(400 Å)/ buffer(x Å)/ Ag(1000 Å). b) Current-density vs. voltage characteristics for devices fabricated with x
= 100 buffer layers of compounds 1-6, showing comparable overall performance for all devices. c) The measured
(open symbols) and simulated (solid lines) J(V) traces for devices with x = 200 exhibiting the non-ideal inflection
behavior of compounds 2-6, in contrast to compound 1 at the same thickness.

110
Overall, the highest FF (0.58) and J
sc
(5.74 mAcm
-2
) are observed for devices incorporating complex 2,
bearing ligands with 1,3-C
6
H
5
, and complex 4 bearing 1-CF
3
,3-C
4
H
3
S groups, respectively. Considering
the marked differences in chemical functionality between the phenyl, trifluoromethyl and thienyl moieties,
it is noteworthy that these values are only slightly higher than the FF (0.52 ±0.03) and J
sc
(4.93 mAcm
-2
)
for the parent Ru(acac)
3
complex possessing methyl substitution. Furthermore, the resulting power
conversion efficiencies of 1.1% ±0.3% for Ru(fhsa)
3
and 1.4% ±0.3% for Ru(bhba)
3
due to their slightly
low FF and V
oc
values, respectively, are comparable to that of 1.4% ±0.3% for Ru(acac)
3
.
The relatively material independent performance observed for all 100 Å buffer based devices is
commensurate with ultraviolet and x-ray photoelectron spectroscopic studies
8,9
showing that thermal
deposition of metal vapor onto organic layers often yields substantial metal diffusion 50 – 100 Å into the
organic film and the formation of mid-gap states facilitating carrier transport, and/or a metal mediated
charge transport layer with properties relatively independent of the organic material.
17
This intercalated
region has been estimated to be ~115 Å for the case of BCP.
5
Thus, for ruthenium derivative thicknesses of
100 Å, it is not surprising that no clear correlation between materials properties and device performance is
observed, due to the likely presence of such an interfacial layer. In essence, when the RuL
3
film is thin, the
100 Å thick intercalated layer, incurred during metal deposition, appears to have a leveling effect on the
charge collection process and the resulting device performance is similar for all observed cases.
In order to test this hypothesis, devices were fabricated for which the properties of the organic
material itself, not the intercalated layer, would be responsible for the electrical behavior of the resulting
device. To compensate for possible metal mediated transport effects, the buffer layer thickness was
increased to 200 Å. At this thickness, it is expected that charge conduction by metal mediated transport
will be of minor importance to the overall J(V) characteristics of the device, since carriers encounter a
substantial length of pristine organic material between the metal intercalated layer and the acceptor. Thus,
the properties of the ruthenium complex, and not those of the intercalated layer, will dominate the electrical
behavior. The experimental J(V) results for these devices, containing 200 Å neat films of compounds 1-6
as buffer layers, are plotted as open symbols in Figure 4.4c. Clearly, several of these devices behave in a
manner that is decidedly unlike their 100 Å counterparts. Specifically, the concavity changes and their
111

associated inflection points at V = -0.26, -0.19, 0.11, 0.18, and 0.38 V for compounds 2-6, respectively,
result in marked FF reductions and substantial power losses for these devices, when compared with the
Ru(acac)
3
device of equal buffer thickness. This non-ideal inflection behavior, that is often observed
experimentally, but seldom discussed in the OPV literature,
18,19
bears a phenomenological resemblance to
the well documented rectifying back contact
20
case of CdS/CdTe.
21,22
However, in the present case, the
interfacial energy level alignment estimated by UPS at the A/B interface suggests that the observed FF
values are determined, not by charge injection at the electrode, but rather by Marcus-like bimolecular
recombination rates at the A/B interface.
Comparing the electrical characteristics of devices fabricated with 200 Å buffer films, the FF values of
0.52 ±0.03, 0.40 ±0.03, 0.20 ±0.03, 0.12 ±0.03, 0.14 ±0.03, and 0.09 ±0.03 for compounds 1-6,
respectively, tend to decrease with HOMO energy stabilization. That is, values 2-6 are 23%, 61%, 77%,
71%, and 83% lower than the Ru(acac)
3
based device FF, respectively. This can be seen in the linear
correlation between FF and bulk ionization energy of the buffer complex plotted in Figure 4.5. The FF for
these 200 Å devices decreases monotonically with a slope of ca. 0.25 eV
-1
(solid line).
The measured short circuit current density for this series exhibits a less regular, but pronounced
dependence on buffer identity. The observed J
sc
for complexes 1-6 are 5.2, 5.7, 5.2, 3.1, 4.3, and 1.6
mAcm
-2
, respectively. Conversely, no clear dependence on buffer identity is observed in the measured V
oc
.
Table 4.2. Measured 100 Å device performance metrics
Complex FF[a] J
sc
[b]
[mAcm
-2
]
V
oc
[c]
[mV]
P
max
[d]
[mWcm
-2
]
1 Ru(acac)
3
0.52 4.9 535 1.4
2 Ru(bhba)
3
0.58 5.2 457 1.4
3 Ru(bhf
2
)
3
0.45 5.6 468 1.2
4 Ru(fhsa)
3
0.38 5.7 489 1.2
5 Ru(fhna)
3
0.49 5.4 428 1.1
6 Ru(fhba)
3
0.46 5.0 428 1.0
[a] Fill factor ± 0.03
[b] Short circuit current density ± 0.1 mAcm
-2

[c] Open circuit voltage ± 20 mV.
[d] Maximum output power density ± 0.2 mWcm
-2

112

Indeed, the values of 480, 530, 480, 420, 460, and 390 mV are comparable. Typical performance metrics
for these 200 Å devices are collected in Table 4.3 and exemplify that FF and power output are strongly
dependant on the identity of the buffer complex used to fabricate the device.
In accord with the poor electron transport previously observed for the parent Ru(acac)
3
complex,
10

direct electron transport via Ru
III
/Ru
II
self exchange does not appear to be efficient for the substituted RuL
3

complexes. As a result, despite possessing estimated electron transport levels in Figure 4.3b that are
suitable for direct transport of photogenerated electrons to the cathode, complexes 2-6 produce non-ideal
device characteristics. As discussed below, the non-linear response, resulting in the inflection behavior
observed in the 200 Å buffer layer devices, does not arise from a linear resistance effect. Instead, the
observed inflection behavior suggests a mechanism by which charge collection is a thermally assisted
process. The origin of this behavior could naively be attributed to the barrier height for hole injection
increasing at the buffer/cathode interface along with the strength of the electron withdrawing groups on the

Figure 4.5. Fill factor of devices based on 200 Å films of compounds 1-6, showing prominent dependence on
complex ionization energy (E
i
) measured using ultraviolet photoelectron spectroscopy.
113

ligands. However, upon probing interfacial energy level alignment using ultraviolet photoelectron
spectroscopy, not only does this interface appear to have little to no effect on device performance, but this
behavior resembles Marcus-inverted charge recombination at the A/B interface.
To assess the relationship between FF and molecular properties for these non-ideal systems, it is
instructive to consider the nature of the J(V) dependence in more detail. Typically, losses in OPV devices
are estimated according to the modified Shockley equation
23
for a single diode equivalent circuit model.
As such, the current density voltage relationship exhibits no inflection points and is given by the
transcendental equation
(4.1)

where, respectively, J and V are the current density and voltage measured at the electrical terminals, R
s
,
R
sh
, J
s
, J
ph
, n, and V
t
are the lumped series resistance multiplied by cross sectional area, the lumped shunt
resistance multiplied by cross sectional area, the reverse-bias saturation current density, the photocurrent
density, the diode ideality factor, and the thermal voltage. Commonly, however, non-idealities in
experimental systems cannot be adequately modeled according to a single diode circuit, due to J(V)
inflection behavior. As previously noted, a more physical description can be obtained in these cases by
Table 4.3. Measured 200 Å device performance metrics
Complex FF[a] J
sc
[b]
[mAcm
-2
]
V
oc
[c]
[mV]
P
max
[d]
[mWcm
-2
]
1 Ru(acac)
3
0.52 5.2 480 1.3
2 Ru(bhba)
3
0.40 5.7 530 1.2
3 Ru(bhf
2
)
3
0.20 5.2 480 0.5
4 Ru(fhsa)
3
0.12 3.1 420 0.2
5 Ru(fhna)
3
0.15 4.3 460 0.3
6 Ru(fhba)
3
0.09 1.6 390 0.1
[a] Fill factor ± 0.03
[b] Short circuit current density ± 0.1 mAcm
-2

[c] Open circuit voltage ± 20 mV.
[d] Maximum output power density ± 0.2 mWcm
-2

114
introducing, in series, a model auxiliary anti-polar diode (APD). Under illumination and an applied
potential the measured current density for the APD model in Figure 4.6a will equal that flowing through
both Diode 1 and Diode 2, giving J = J
1
= J
2
. The measured voltage being the sum of the voltage across
Diode 1, the voltage across Diode 2, and the voltage across all series resistive elements.
As a practical matter, it should be noted here that as presented, the transcendental nature of Eq. 4.1
precludes an analytical solution and as such is of limited value for direct optimization of the model
parameters. However, it has been shown
24-28
that Eq. 4.1 can be explicitly solved in terms of the Lambert-
W function
29
of the form w = W
0
(x), being the solution to the transcendental equation we
w
= x. This method
is applied here for the explicit solution for the APD model in Figure 4.6a. It follows from V = V
1
+ V
2
+
JR
s
, where the numerical subscripts denote Diode 1 and Diode 2, that the measured voltage may be
expressed in terms of the Lambert-W function as
(4.2)
with the observed nonlinear current density rise in quadrant I near the compensation voltage
30
(V
o
)
interpreted as the onset of Diode 2 breakdown.
The experimental J(V) results for devices containing 200 Å neat films of compounds 1-6 as buffer
layers were presented as open symbols in Figure 4.4c, here those results are analyzed based on the above
two-diode discussion. Model parameters for the parent Ru(acac)
3
device were obtained from nonlinear
regression according to the Lambert-W expression for Eq. 4.1. The extracted ideality factor was
subsequently used to perform iterative optimization of the remaining model parameters in Eq. 4.2 for the
observed electrical behavior of devices fabricated using RuL
3
compounds 2-6. The resulting simulated
J(V) characteristics are presented as solid lines in Figure 4.4c and compared against the experimental data.
Model parameter estimates for the Diode 2 recombination-dominated case are collected in Table 4.4.
115

Resistive losses resulting from the largely insulating nature of conjugated organic materials
employed in thin film OPV devices are often assessed by horizontal optimization of Eq. 4.1 for V > 0 while
simultaneously neglecting parallel conductance. However, the J(V) inflection behavior in many
experimental systems, as in the case presented here, limits the physicality of results obtained in this way,
due to the implicit R
s
voltage dependence. Alternatively, in cases where the J(V) characteristics exhibit
inflection points, the anti-polar diode model accounts for non-idealities in quadrants I and IV and yields a
more physical description of the output power losses incurred due to series resistance effects.
a) b)
c)
Figure 4.6. a) The APD equivalent circuit model used to simulate the solid traces in Figure 4.4a and to estimate
non-ideal devices parameters. Voltage applied across the external leads is distributed between d
1
and d
2
,

as
described in the text. b) Magnitude of the saturation current-density (J
s2
) for Diode 2 used in APD model
simulations of 200 Å buffer layer devices plotted as a function of experimentally observed FF showing that J
s2

increases by six orders of magnitude with observed FF. The dotted line is drawn as guide to the eye. c)
Experimentally determined output power density (open symbols) for 200 Å buffer layer devices compared with
simulated data (solid lines) for the same, showing that increased saturation current for Diode 2 (J
s2
) corresponds
to enhanced output power density.
116

Correspondingly, the R
s
values used to model devices based on compounds 2-6 of 10 ±4, 37 ±1, 41 ±2, 30
±3, and 37 ±2 Ωcm
2
, respectively, are comparable in magnitude and their slight variation does not correlate
with the trend in device FF. As such, the impact on device performance resulting from disparities between
the low-range electrical conductivity of compounds 2-6 appears to be relatively minor.
Similarly, in a hypothetical device suffering a large parallel conductance across R
sh1
, significant
FF reduction
20
is expected. However, the model R
sh1
values in Table 4.4 are all also comparable for
complexes 2-6, and are all approximately 2 orders of magnitude larger than the corresponding model R
s

values. This indicates that the output power losses do not result from shunting of Diode 1. From this
analysis, neither series resistance nor parallel conductance offer a satisfactory explanation for the observed
dependence of FF on the nature of the buffer materials used in this study.
As previously mentioned, the measured short circuit current densities for these 200 Å devices
demonstrate a strong dependence on buffer identity, ranging in magnitude from 1.6 mAcm
-2
to 5.7 mAcm
-2
.
However, at an applied bias of V = -1 V the J(V) traces in Figure 4.4c demonstrate a far more narrow
current density distribution, ranging from only 6.3 mAcm
-2
to 7.4 mAcm
-2
. This indicates that,
notwithstanding the poor fill factor, high carrier collection efficiencies for electron and hole pairs
coulombically bound geminate at the donor acceptor interface can be obtained by applying a substantial
external bias to draw out photogenerated charges. Moreover, the model photocurrent density (J
ph
) values
corresponding to compounds 2-6 of 5.80 ±0.05, 6.99 ±0.01, 6.05 ±0.3, 5.95 ±0.06, and 6.24 ±0.03 mAcm
-2
,
respectively, are comparable, indicating that disparities in exciton generation and/or dissociation
Table 4.4. Model anti-polar diode parameters for 200 Å buffer devices
Complex J
s2
[a]
[µAcm
-2
]
J
s1

[µAcm
-2
]
R
s

[Ωcm
2
]
R
sh2

[kΩcm
2
]
R
sh1

[kΩcm
2
]
J
ph

[mAcm
-2
]
2 Ru(bhba)
3
(9.0 ± 5.0) × 10
2
1.7 ± 0.9 10 ± 4 0.1 ± 0.01 1.10 ± 0.2 5.80 ± 0.05
3 Ru(bhf
2
)
3
3 ± 2 1.7 ± 0.9 37 ± 1 1.9 ± 0.3 1.15 ± 0.09 6.99 ± 0.01
4 Ru(fhsa)
3
(5 ± 4) × 10
-2
1.7 ± 0.9 41 ± 2 1.9 ± 0.3 1.35 ± 0.09 6.05 ± 0.03
5 Ru(fhna)
3
(6 ± 3) × 10
-1
1.7 ± 0.9 30 ± 3 1.3 ± 0.3 0.70 ± 0.1 5.95 ± 0.06
6 Ru(fhba)
3
(6 ± 5) × 10
-4
1.7 ± 0.9 37 ± 2 2.9 ± 0.8 1.10 ± 0.3 6.24 ± 0.03
[a] Recombination limited estimates
117
probabilities among these devices may be neglected. Hence, the model R
s
, R
sh
, and J
ph
parameters
discussed thus far for complexes 2-6 exhibit only limited dissimilarities and do not adequately address the
trend in FF reduction and resulting power losses observed in the 200 Å buffer devices. The significant J
sc

reductions observed for the Ru(fhsa)
3
, Ru(fhna)
3
, and Ru(fhba)
3
are certainly not due to low J
ph
,

since this
model photocurrent parameter is comparable for all observed cases.
While the model parameters corresponding to photocurrent generation and resistance are
comparable for the 200 Å buffer layer devices, the model J
s2
associated with Diode 2 is the only parameter
that differs significantly from device to device. In fact, it varies by six orders of magnitude depending on
the buffer complex and exhibits a marked correlation with the observed FF for compounds 2-6 when
plotted as in Figure 4.6b. This finding suggests that a systematic variation in the energy barrier associated
with a thermally assisted interfacial process is the likely origin of the observed trend in the 200 Å buffer
layer device data. Note that Diode 2 saturation current estimated for complex 1 obtained by substitution of
the non-linear regression parameters obtained for the single diode case as constants into Eq. 4.2 yields J
s2

more than eight orders of magnitude larger that for complex 6 with the smallest Diode 2 saturation current.
When the experimentally observed output power densities are overlaid in Figure 4.6c with
simulated data according to the APD model, the power density is observed to correlate with the magnitude
of J
s2
. For devices with small model J
s2
, the maximum output power density (P
max
) is diminished due to
significant FF reduction. Conversely, as the saturation current density of Diode 2 increases a dramatic
increase in P
max
is observed, such that in the limit of J
s2
→ ∞, the J(V) relation tends toward the single
diode behavior where no inflection point is observed. This is precisely what is observed in the case of
complex 1 for which the current density simulated according to the single diode and dual diode models are
virtually indistinguishable. Taking this as the limiting case of infinite Diode 2 saturation current, the APD
model effectively collapses to the single diode model in the case of the Ru(acac)
3
based device. Thus, it is
compelling to identify the molecular properties that might influence the magnitude of the model Diode 2
saturation current, the corresponding positions of the J(V) inflection points, and as a result, the power
output of the 200 Å buffer layer devices.
118

In order to discriminate between the two plausible energy barrier locations (i.e. buffer/cathode vs.
acceptor/buffer) responsible for the observed J(V) inflection behavior, the interfacial energy level
alignment between the buffer layer and the cathode, compared to that between the acceptor layer and the
buffer layer were evaluated by ultraviolet photoelectron spectroscopy. At the buffer/cathode interface, the
injection barrier for holes was estimated based on the ionization energy (E
i,C
) of compounds 1 – 6 measured
on silver, relative to the Fermi energy (E
F
) of the metal. That is, after accounting for shifting of the
interfacial electronic structure,
9
the interfacial energy level offset ΔE
BC
= E
i,C
- E
F
≈ φ
h
,
20,23
was calculated
as depicted in Figure 4.7a. Note that estimating φ
h
for actual devices according to ΔE
BC
is approximate, as
it neglects the interaction of silver vapor with the organic material during device fabrication.
In the event that the current density at constant voltage is limited by hole injection, an exponential
dependence in accordance with J
s2
= J
so2
exp(-φ
h
/kT)
20,23
is expected for the Diode 2 saturation. The
frequency factor J
so2
may be regarded physically as a kinetic term related to the area of the junction, wave
function overlap between buffer and the electrode, charge mobility, the electron transfer reorganization
energy, and the density of chargeable sites at the junction interface.
31
However, this exponential
dependence is not exhibited by the data presented in Figure 4.7b, where the abscissae represent ΔE
BC
≈ φ
h
,
a) b)
Figure 4.7. a) Schematic representation of energy offset (ΔE
BC
) between Ag Fermi Level (E
F
) and the ionization
energy (E
i,C
) for each buffer complex measured on Ag coated substrates. b) The estimated diffusion-controlled
saturation current for Diode 2 (J
s2
) fails to correlate with increasing ΔE
BC
. Interfacial offsets for complexes 2, 3,
5, and 6 measured on Ag coated substrate, offset for complex 4 estimated from bulk measurement.
119
while the ordinates are values of J
s2
, assuming an ideality factor n
2
of unity. In reality, there is virtually no
observed correlation between ΔE
BC
and J
s2
. In part, this is due to shifting of the vacuum level at the
buffer/Ag interface, causing departures in the calculated φ
h
from what may be expected based simply on the
bulk HOMO orbital energy levels measured for the complexes. Moreover, the induction of extrinsic
electronic states in the buffer layer, that is known to occur upon cathode deposition, is likely to have an
additional leveling effect on the barrier for hole injection in the devices studied here. This observation
suggests that the measured device performance is not limited by the injection of holes at the buffer/Ag
interface.
Observing no apparent correlation between device performance and hole injection barrier at the
buffer/cathode interface, it remains to address the recombination-limited scenario. Under cursory
examination, one may regard as trivial the thermodynamically favorable processes of carrier recombination
at the A/B interface. However, according to Marcus theory the rate of this interfacial charge transfer
reaction may actually determine the charge collection efficiency, and hence the power output in the 200 Å
buffer devices. Based on Marcus theory, the rate constant k
rec
for outer sphere bimolecular electron transfer
can be expressed as

(4.3)
where h is Planck’s constant, H
ij
is the electronic coupling matrix element between the PP state and the
final neutral state, λ is the free energy for geometric reorganization, and ΔG° is the total free energy change
for the recombination reaction A
-
+ B
+
→ A + B. For a series of similar compounds, such as the RuL
3

complexes, Eq. 4.3 suggests that k
rec
should decrease with decreasing driving force (-ΔG°), reach a
maximum at -ΔG° = λ, and finally decrease with any further increase in driving force. The latter case
constitutes the Marcus inverted regime, where, although highly exergonic, the reaction occurs slowly due to
the large geometric reorganization energy involved. As a result lnk
rec
for such a system will exhibit a
quadratic dependence on the driving force.
120

To estimate the driving force for the recombination reaction between the acceptor and the buffer
layer, the energy level offset ΔE
AB
= E
A
– E
i,B
was examined, where E
A
is the C
60
electron transport level
measured by inverse photoemission spectroscopy (IPES).
32
The E
i,B
term represents the ionization energy
for each buffer complex on C
60
-coated ITO substrates, where the C
60
/RuL
3
interfacial vacuum level shift is
negligible,
10
determined by UPS. The measured UPS spectra and the resulting energy level diagram
derived from the measured UPS data are depicted in Figure 4.8. Based on this analysis, the thermodynamic
driving force for the recombination reaction of photogenerated electrons in the C
60
layer with reciprocal
holes injected into the buffer material from the cathode is approximately ΔE
AB
.
a)
b)
Figure 4.8. a) Ultraviolet photoelectron spectra measured for ruthenium complexes 1-6 on ITO/C
60
substrates. b)
Interfacial energy offsets (ΔE
AB
) between the electron transport level of C
60
and the hole transport level of each
RuL
3
buffer layer.

121

Relating the measured UPS data to the device electrical characteristics, the recombination limited
(n
2
= 2) saturation current density for Diode 2 may be expressed in terms of Eq. 4.3 as

(4.4)
where q is the elementary charge, and [A
-
] and [B
+
] are, respectively, the concentration of ionized acceptor
and ionized buffer molecules at the A/B interface. Based on Eq. 4.3, when the driving force is sufficiently
large, the recombination rate and the resulting J
s2
value in Eq. 4.4 will decrease with increasing driving
force. Correspondingly, plotting J
s2
as the ordinate on a semi-log scale versus ΔE
AB
in Figure 4.9a reflects
a diminishing recombination rate with increasing driving force. This behavior is indicative of the Marcus
inverted regime, in which the rate for charge recombination slows as the driving force increases.
Since the expression for J
s2
in Eq. 4.4 implicitly depends on the electronic coupling element H
ij
,
departures from the quadratic behavior predicted by Eq. 4.3 with increasing driving force may be
understood on the basis of intermolecular electronic coupling. For example, the space filling molecular
models in Figure 4.9b illustrate the large exposed naphthalene moieties of complex 5 in contrast to the
a) b)
Figure 4.9. a) Estimated J
s2
for 200 Å OPV devices decrease with increasing interfacial energy level offset ΔE
AB

between acceptor layer and buffer complexes 2-6. The dotted line is included as a guide to the eye. b) Space
filling molecular models for buffer complexes 4 and 5, illustrating their respective thienyl- (Left) and naphthyl-
(Right) substitution and the increase in π-system exposure afforded by the naphthalene moiety of complex 5.
Atoms are color coded according to carbon (black), sulfur (turquoise), oxygen (red), fluorine (green), and
hydrogen (gray).

122
smaller thienyl moieties in complex 4. As a result, the extended naphthalene π-system of complex 5 is
more accessible for intermolecular interaction with C
60
at the A/B interface. This leads to an order of
magnitude increase in saturation current associated with Diode 2 for complex 5 relative to complex 4,
despite the similar ΔE
AB
values measured for both materials. This result translates to a smaller loss in P
max

and demonstrates that kinetic accessibility on the molecular level may significantly impact the observed
device performance. These data suggest the chemical mechanism by which charge carrier collection
proceeds for these devices is reminiscent of an A/B interfacial recombination process in the Marcus
inverted regime. Thus, the origin of the inflection behavior in these 200 Å buffer layer devices reflects the
molecular electron transfer character of OPV device operation.
4.4 Concluding remarks
In summary, the behavior of a series of tris(β-diketonato)ruthenium(III) analogues as reciprocal
carrier transport materials in double heterojunction OPVs was examined on the basis of interfacial energy
level offsets. The performance independence observed for 100 Å neat layers of the materials used in this
capacity suggests that charge transport at this thickness proceeds via a metal mediated route. Conversely,
incorporating thicker 200 Å films leads to device characteristics that strongly depend on the nature of the
buffer material. Specifically, all of these thicker devices based on ruthenium complexes with
acceptor/buffer energy level offsets -ΔE
AB
≥ 1.4 eV exhibit non-ideal inflection behavior in their J(V)
characteristics, which are indicative of thermally activated carrier collection. In conjunction with
ultraviolet photoelectron spectroscopy, simulations of the J(V) behavior suggest that Marcus-inverted
charge recombination at the acceptor/buffer interface may be regarded as the primary origin of this
behavior. The present study highlights the continuing importance of considering thermodynamic and
kinetic factors resulting from molecular properties that influence OPV device performance. The analysis
presented here illustrates, for a specific case using a dual anti-polar diode model to describe the carrier
collection behavior exhibited by ruthenium buffer materials, that molecular structure can have a measurable
impact on device performance, regardless of energy level alignment. However, it should be noted that this
is simply one example where molecular kinetic effects influence device performance. Since organic
123
optoelectronic devices 1) are inherently operated under non-equilibrium conditions and 2) are molecular by
nature, the effects of molecular kinetic accessibility are expected to be general and, at times, even dominate
the observed device performance. One extremely important example where this is expected to play a role
is the origin of photovoltage losses in OPV devices, which are determined by the ratio of photocurrent
density to saturation current density.
124
4.5 Chapter 4 endnotes
1 P. Peumans, V. Bulovic and S. R. Forrest, Appl. Phys. Lett., 2000, 76, 2650.
2 P. Peumans and S. R. Forrest, Appl. Phys. Lett., 2001, 79, 126.
3 S. E. Burns, N. Pfeffer, J. Gruner, M. Remmers, T. Javoreck, D. Neher and R. H. Friend, Adv.
Mater., 1997, 9, 395.
4 S. Yoo, W. J. Potscavage, B. Domercq, S. H. Han, T. D. Li, S. C. Jones, R. Szoszkiewicz, D. Levi,
E. Riedo, S. R. Marder, et al., Solid-State Electron., 2007, 51, 1367.
5 J. Huang, J. S. Yu, H. Lin and Y. D. Jiang, J. Appl. Phys., 2009, 105.
6 H. Gommans, B. Verreet, B. P. Rand, R. Muller, J. Poortmans, P. Heremans and J. Genoe, Adv.
Funct. Mater., 2008, 18, 3686.
7 M. Vogel, S. Doka, C. Breyer, M. C. Lux-Steiner and K. Fostiropoulosa, Appl. Phys. Lett., 2006,
89.
8 Y. Hirose, A. Kahn, V. Aristov, P. Soukiassian, V. Bulovic and S. R. Forrest, Phys. Rev. B., 1996,
54, 13748.
9 A. Kahn, N. Koch and W. Y. Gao, J. Polym. Sci., Part B: Polym. Phys., 2003, 41, 2529.
10 B. P. Rand, J. Li, J. G. Xue, R. J. Holmes, M. E. Thompson and S. R. Forrest, Adv. Mater., 2005,
17, 2714.
11 D. R. Eaton, J. Am. Chem. Soc., 1965, 87, 3097.
12 A. Endo, Y. Hoshino, K. Hirakata, Y. Takeuchi, K. Shimizu, Y. Furushima, H. Ikeuchi and G. P.
Sato, Bull. Chem. Soc. Jpn., 1989, 62, 709.
13 A. Endo, M. Kajitani, M. Mukaida, K. Shimizu and G. P. Sato, Inorg. Chim. Acta, 1988, 150, 25.
14 P. I. Djurovich, E. I. Mayo, S. R. Forrest and M. E. Thompson, Org. Electron., 2009, 10, 515.
15 G. S. Patterson and R. H. Holm, Inorg. Chem., 1972, 11, 2285.
16 R. F. Handy and R. L. Lintvedt, Inorg. Chem., 1974, 13, 893.
17 Z. R. Hong, Z. H. Huang and X. T. Zeng, Chem. Phys. Lett., 2006, 425, 62.
18 M. Glatthaar, M. Riede, N. Keegan, K. Sylvester-Hvid, B. Zimmermann, M. Niggemann, A.
Hinsch and A. Gombert, Sol. Energy Mater. Sol. Cells, 2007, 91, 390.
125
19 K. Schulze, C. Uhrich, R. Schuppel, K. Leo, M. Pfeiffer, E. Brier, E. Reinold and P. Bauerle, Adv.
Mater., 2006, 18, 2872.
20 A. L. Fahrenbruch and R. H. Bube, Fundamentals of Solar Cells: Photovoltaic Solar Energy
Conversion, Academic Press, New York, 1983.
21 M. Burgelman, J. Verschraegen, S. Degrave and P. Nollet, 2005.
22 S. H. Demtsu and J. R. Sites, Thin Solid Films, 2006, 510, 320.
23 S. M. Sze, Physics of Semiconductor Devices, 2 edn., Wiley-Interscience, New York, 1981.
24 A. Ortiz-Conde, F. J. G. Sanchez and J. Muci, Solid-State Electron., 2000, 44, 1861.
25 A. Jain, S. Sharma and A. Kapoor, Sol. Energy Mater. Sol. Cells, 2006, 90, 25.
26 A. Jain and A. Kapoor, Sol. Energy Mater. Sol. Cells, 2005, 85, 391.
27 A. Jain and A. Kapoor, Sol. Energy Mater. Sol. Cells, 2004, 81, 269.
28 T. C. Banwell and A. Jayakumar, Electron. Lett., 2000, 36, 291.
29 B. Hayes, Am. Sci., 2005, 93, 104.
30 G. G. Malliaras, J. R. Salem, P. J. Brock and J. C. Scott, J. Appl. Phys., 1998, 84, 1583.
31 J. C. Scott and G. G. Malliaras, Chem. Phys. Lett., 1999, 299, 115.
32 R. Schwedhelm, L. Kipp, A. Dallmeyer and M. Skibowski, Phys. Rev. B., 1998, 58, 13176.

126
Chapter 5
Nanostructures in organic photovoltaics
5.1 Solar cells on a small scale
Materials composed of structures with dimensions on the nanometer length scale often exhibit
exotic properties relative to their macroscale counterparts. Quantum confinement,
1,2
and low temperature
solution processability
3
in colloidal semiconductor nanocrystals,
4
high surface area semiconducting
photoelectrodes,
5
nanoscale phase segregation
6,7
and highly conductive thin films for flexible electronics
8

make such structures compelling candidates for solar energy conversion.
9
This chapter focuses in the
design, development, and characterization of various photovoltaic device architectures that incorporate
nanostructured components. Specifically, devices based on thin film polymer composites of
environmentally benign semiconducting SnSe nanocrystals, vapor processed hole transport layers for solid-
state dye sensitized solar cells supported on TiO
2
nanowires, and flexible carbon-based nanometer
dimensional transparent electrodes are discussed.
5.2 Tin(II) selenide nanocrystal-polymer composite devices
The I-III-VI family of semiconductor nanocrystals, such as CuInS
2
, CuInSe
2
, and CuIn
x
Ga
1-x
Se
2
,
has shown significant promise as an environmentally friendly alternative to cadmium and lead based
photovoltaic materials.
10-14
For example, solar cells derived from nanocrystal inks of CuIn
x
Ga
1-x
S
2
were
recently shown to demonstrate power conversion efficiencies of up to 5.55%.
14
However, given the ever
increasing demand for copper and the increasing costs and limited supplies of indium and gallium, the long
term feasibility of these materials as solar cell components is questionable.
15-17
Thus, it is compelling to
examine devices fabricated with new types of semiconductor nanocrystals that are composed of more earth
abundant elements.
127

Recently, tin(II) selenide (SnSe) has garnered interest as an earth abundant, nontoxic, and
environmentally benign component for photovoltaic devices, such as solar cells.
18,19
Bulk SnSe has an
indirect band gap of 0.90 eV and a direct band gap of 1.30 eV,
20
while thin films of SnSe have
demonstrated band gaps that are higher in energy as a result of quantum confinement.
18
Significant
advances in the synthesis of related IV-VI semiconductor nanocrystals (e.g., SnS and SnTe) have been
made in the past.
21-23
Recently, the first solution-phase synthesis of well-defined and quantum-confined
SnSe nanocrystals and demonstration of their utility in solar cells was reported by Franzman et al.
24

Nanocrystalline SnSe was synthesized and characterized by Dr. Matthew A Franzman, in the
research group of Professor Richard L. Brutchey, using dialkyl diselenide as the selenium source. The
details of this preparation are reported elsewhere.
24
Briefly, di-tert-butyl diselenide (0.38 mmol) was
injected into a solution of anhydrous tin(II) chloride (0.74 mmol) in a mixture of dodecylamine and
dodecanethiol (2.50 mL/0.50 mL, respectively) at 95 ˚C. After heating to 180 ˚C for 4 minutes, the
reaction solution was cooled and the SnSe nanocrystals were precipitated using ethanol to yield a dark
brown solid that was dispersible in common organic solvents, such as toluene. By controlling the amount

Figure 5.1. SnSe nanocrystals synthesized using di-tert-butyl diselenide at 180 ˚C. High-resolution TEM image
of a single nanocrystal (a), SAED pattern of the nanocrystals (b), and low-resolution TEM image of the SnSe
nanocrystals (c). (d) Absorbance spectrum of SnSe nanocrystals dissolved in cyclohexane.
128

of di-tert-butyl diselenide in the reaction, the composition of thenanocrystals (and oxidation state of tin)
can be rationally controlled.
Transmission electron microscopy (TEM) revealed elongated, anisotropic nanocrystals 19.0 ± 5.1
nm wide, but more polydisperse in length (cf. Figure 5.1). The high-resolution TEM image in Figure 5.1a
shows an apparent single crystalline particle with the (111) lattice planes displayed (d = 0.29 nm). Powder
X-ray diffraction (XRD) analysis of the SnSe nanocrystals revealed that they crystallized in the
herzenbergite orthorhombic phase (Pnma), which consists of a series of strongly bound double layers of tin
and selenium in a highly distorted rock salt structure.
25
Lattice parameters of a = 11.55 Å, b = 4.16 Å, and
c = 4.45 Å were calculated from the diffraction pattern and are in good agreement with literature values for
orthorhombic SnSe (JCPDS no. 048-1224). The lattice parameters calculated from selected area electron
diffraction (SAED) patterns of several randomly chosen regions of the SnSe nanocrystals agree with the
lattice parameters calculated from the XRD pattern for orthorhombic SnSe, with no other crystal phases
observed by SAED analysis (Figure 5.1b). The elemental composition of the nanocrystals was confirmed
to be near stoichiometric by energy dispersive x-ray spectroscopy
24
and x-ray photoelectron spectroscopy
was used to corroborate the elemental composition and confirm the oxidation states of tin and selenium in
the nanocrystals.
The SnSe nanocrystals absorb throughout the visible and into the near-IR region of the spectrum
(cf. Figure 5.1d), resulting in the dark brown color of the material. To assess the potential for these SnSe

Figure 5.2. Topographic AFM images (2.5 µm x 2.5 µm) of representative (a) glass/ITO/MoO3 substrate, (b)
350 Å PPV layer on glass/ITO/MoO3, and (c) 350 Å SnSe:PPV layer on glass/ITO/MoO3 surfaces with
calculated RMS roughness values of 2.8 nm, 5.5 nm, and 6.3 nm, respectively.

.

129

nanocrystals to act as electron accepting materials in photovoltaics, hybrid SnSe:polymer solar cells were
compared to neat polymer devices. Poly[2-methoxy-5-(3´,7´-dimethyloctyloxy)-1,4-phenylene vinylene]
(PPV) was selected as a dispersive matrix, based on its ability to act as an absorbing material and potential
electron donor to the SnSe nanocrystals. Hybrid devices consisted of a 100 Å MoO
3
hole transport layer, a
350 Å SnSe:PPV absorbing layer (0.25:1.0 wt/wt), a 200 Å perylene-3,4,9,10-tetracarboxylic diimide
(PTCDI) acceptor/hole-blocking layer, and a LiF (10 Å)/Al (1000 Å) bilayer cathode on glass/ITO
substrates. This architecture was developed to minimize leakage current losses at the collecting electrodes.
Composite films were obtained from SnSe:PPV suspensions prepared by dispersing a mixture of SnSe (5
mg) and PPV (20 mg) in toluene using a standard probe Misonix S3000 sonicator running at 30-40 W for 1
h. Hybrid and neat polymer films were deposited by spin casting from toluene under ambient conditions
onto the target substrate. Surface topography of the SnSe:PPV composite films was analyzed by atomic
force microscopy (AFM) on a Digital Instruments Nanoscope Dimension 3100 atomic force microscope.
The topographical images in Figure 5.2 of ITO/MoO
3
, ITO/MoO
3
/PPV, ITO/MoO
3
/PPV:SnSe illustrate
that the SnSe nanocrystals are well dispersed by PPV in the 350 Å composite layer.
The current density voltage relationship, J(V), measured under spectral mismatch corrected one
sun illumination for SnSe:PPV hybrid films and neat PPV films are overlaid in Figure 5.3, along with the

Figure 5.3. Overlaid spectral mismatch corrected J(V) (filled) and P(V) (open) characteristics for
ITO/MoO
3
/poly/PTCDI/LiF/Al devices, where poly is a 0.25:1.0 wt/wt SnSe:PPV film or a neat polymer film
under 1000 W m
-2
AM 1.5G illumination.

130

output power density voltage dependence, P(V). The J
SC
of the SnSe:PPV hybrid device (0.39 mA cm
-2
) is
nearly twice that of the neat polymer control (0.20 mA cm
-2
), while V
OC
of the former (455 mV) is
comparable to that of the latter (480 mV). The fill factor for the SnSe:PPV hybrid (0.36) and PPV (0.30)
based devices are similar, suggesting no charge trapping on SnSe. Moreover, the power conversion
efficiency (η) is enhanced by >100% with SnSe:PPV compared to neat PPV. This enhancement results
from quantum efficiency doubling near 500 nm, attributed to PPV electron transfer to SnSe, since the
hybrid film absorption coefficient is comparable to that of neat PPV at 500 nm as shown in Figure 5.4.
Substituting C
60
/BCP for PTCDI/LiF can achieve PCE > 0.25% for SnSe:PPV; however, the reliability of
these devices is compromised by rapid degradation, as illustrated in Figure 5.5. Such degradation has been
previously reported for PPV/C
60
.
26,27

Thus, a facile solution-phase synthesis of well-defined SnSe nanocrystals that exhibit quantum
confinement effects has been reported for the first time. The direct optical band gap of E
g
= 1.71 eV offers
robust coverage of the solar spectrum, which makes the SnSe nanocrystals attractive candidates for
incorporation into solar cells. Functional solar cells were fabricated using a SnSe:PPV active layer with

Figure 5.4. (a) Absorption coefficient of the hybrid SnSe:PPV and neat PPV films as a function of wavelength
indicating that the absorption near 500 nm is comparable for both films. (b) Quantum efficiency enhancement for
ITO/MoO3/poly/PTCDI/LiF/Al devices when poly = PPV is replaced with the poly = SnSe:PPV composite layer.
131

PCE of 0.06% under AM 1.5G illumination. Future studies will focus on the optimization of these low-
cost devices to best utilize this earth abundant and nontoxic photovoltaic material.
5.3 Solid-state hole transport media for dye-sensitized solar cells
5.3.1 Toward solid-state photoelectrochemical cells
A promising excitonic solar cell
28
technology incorporates a photoelectrochemical device known
as a dye-sensitized solar cell (DSC). These DSCs have demonstrated relatively high power conversion
efficiencies (~ 10%) since first proposed by Gratzel.
29
In a typical cell, dye molecules are adsorbed on a
high surface area semiconducting metal oxide, such as mesoporous TiO
2
, forming a type-II heterojunction
between the dye molecule and TiO
2
. Following photon absorption by the dye, an electron is injected into
the conduction band of the mesoporous network of TiO
2
nanoparticles. The neutral state of the dye is
subsequently regenerated by a solution phase I
-
/I
3
-
redox couple that shuttles positive charge to a counter
electrode.
Unfortunately, the volatile solution phase electrolyte system, typically consisting of acetonitrile
and iodide/triiodide, deployed for dye regeneration in traditional DSCs represents a stability and packaging
limitation to this otherwise potentially economically viable solar technology. To address the problems of

Figure 5.5. Output power degradation of ITO/MoO3/poly/C60/BCP/Al devices, where poly is (a) PPV or (b)
SnSe:PPV.
132
corrosion and solvent leakage associated with these liquid systems,
30
there is substantial interest in
developing solid-state (sDSC) devices, where a highly conductive, solid-state hole-transport medium
(HTM) replaces the solution electrolyte system.
31
Currently, HTM deposition is carried out via solution-
based techniques, where materials such as poly(3-hexylthiophene) (P3HT)
32-34
or 2,2’,7,7’-tetrakis-(N,N-di-
p-methoxyphenylamine)9,9’-spirobifluorene (spiro-OMeTAD)
35-38
are dissolved in a solvent and drop cast
or spin cast onto the substrate. However, these techniques rely strictly on capillary action
39
for the uptake
of HTM molecules and require concomitant solvent removal from deep within the nanostructured device.
As a result, a considerable mass transport limitation exists for solution-based deposition methods, due to
the required transverse motion of solute molecules with respect to the surrounding solvent medium. In this
regard, solution based deposition techniques are unattractive for achieving a high quality conformal HTM
coating of the photoanode surface to ensure contact of the HTM domains to the counter electrode. This is a
problem for devices fabricated by solution-based HTM deposition, because efficient dye regeneration
requires intimate orbital overlap between the dye and the molecules of the HTM. Consequently, physical
infiltration of molecules employed in the HTM is critical to achieving high performance.
39
Since affecting
intimate TiO
2
infiltration by HTM materials has proven to be quite challenging using solution based
fabrication methods,

it is compelling to examine alternative HTM deposition techniques.
The present study demonstrates the feasibility of employing a conformal physical vapor
condensation technique, known as organic vapor phase deposition (OVPD),
40
to grow TiO
2
-infiltrating
domains of the HTM material 1,4-bis(2-naphthylphenylamino)benzene (NNP) in intimate contact with the
dye layer of sDSC devices prepared on TiO
2
nanowire arrays. This device fabrication process, which is
expected to be scalable and relatively inexpensive, occurs across a sharp thermal gradient by condensation
of the HTM material from deep within the interstitial spaces of the nanostructured TiO
2
electrode. The
primary advantage is that OVPD is completely solvent-free. Thus, immediately upon deposition, the
desired molecular orientation is achieved, with intimate contact between the HTM and the adsorbed dye on
the surface of the TiO
2
. Rather than relying on capillary action for the uptake of HTM molecules and
subsequent molecular reorientation during solvent removal, OVPD occurs locally at the solid-vapor
interface, with only an inert carrier gas as the surrounding medium. In principle this is highly desirable,
133

because it eliminates the mass transport problems associated with solution deposition methods, for which
the dynamics of solvent removal tend to create voids and non-uniformities in the HTM. Thus, in addition
to facilitating extensive infiltration, the OVPD technique is also expected to yield a highly desirable HTM
structure, by reducing the tendency to form high resistance grain boundaries in the HTM layer, which can
also result during solution-based deposition methods.
a) b)
c)
Figure 5.6. a) Illustration of the solid state dye sensitized solar cell configuration. A vertical array of TiO
2

nanowires is grown directly on FTO substrate, sensitized with N719 dye molecules, infiltrated with vapor
deposited NNP:F
4
TCNQ, and coated with Cu electrodes. b) Energy levels for various sDSC components. All
values presented in eV. c) Photocurrent generation mechanism involving (I) photon absorption by the dye
molecule (D) to generation D*, (II) electron injection from D* into the conduction band (CB) of TiO
2
, (III) hole
collection from D
+
by the doped NNP layer hopping assisted transport. Also depicted is the unfavorable charge
recombination process (IV) between the photogenerated electron in TiO
2
and D
+
.
134
To clearly demonstrate the growth of conformal HTM layers by OVPD, a model photoanode
composed of an array of 1-dimensional TiO
2
nanowires was selected. Related nanowire arrays have been
previously obtained on Ti foil
41-43

and on transparent conducting oxides
44,45
and have recently
46
been
employed
47,48
in various solar cell architectures to improve charge collection efficiency
41-43,49-51
and light
trapping.
52
With respect to the work presented here, TiO
2
nanowire arrays represent a model system for
clarity of demonstration, since scanning electron microscopy (SEM) can be employed to directly probe
HTM infiltration in OVPD-fabricated solid-state dye sensitized solar cells.
5.3.2 Materials and methods
The device structure used in this study is illustrated schematically in Figure 5.6a. A vertically
aligned array of TiO
2
nanowires was first grown on fluorine-doped tin oxide (FTO) coated glass substrates
using a hydrothermal method. Nanowire samples were prepared by Akshay Kumar in the research
laboratory of Professor Chongwu Zhou. Detailed synthetic procedures are reported elsewhere.
46
Briefly,
FTO-coated glass substrates were immersed in sequential ultrasonic solvent baths of acetone, isopropanol,
and deionized (DI) water, dried under nitrogen flow, and placed in a Teflon lined autoclave containing the
reaction mixture of 10 ml DI water, 10 ml HCl, and 1 ml titanium isopropoxide. The autoclave was heated
to 120
o
C and the reaction time varied according to the desired nanowire length. The autoclave was cooled
under flowing water, the sample removed and thoroughly rinsed with DI water, dried under nitrogen flow,
and immersed in a 0.1M aqueous solution of TiCl
4
for 30 minutes at 60
o
C. Each nanowire array was
subsequently annealed at 400
o
C for 30 minutes in air. These relatively mild synthesis conditions have
virtually no deleterious impact on FTO conductivity.
Nanowire substrates were then immersed in a 0.3 mM ethanolic solution of N719 dye for 24
hours. The dye loaded nanowire samples were transferred to the OPVD chamber and coated with a highly
conductive HTM layer of NNP doped with tetrafluorotetracyano-p-quinodimethane (F
4
TCNQ). A glass
reactor OVPD system maintained by Francisco Navarro was used for HTM growth, where the organics are
sublimed inside a tube furnace, impelled by an inert carrier gas (e.g., N
2
) and condense on the surface of a
cooled substrate. The OVPD method offers tenability with respect to film quality by controlling ambient
135
pressure, substrate temperature, deposition rate, and inert carrier gas flow rate. The HTM films were
grown under a pressure of 1 Torr, inert gas flows of 40 sccm and organic deposition source temperatures of
240
o
C - 280
o
C and 140
o
C - 150
o
C for NNP and F
4
TCNQ, respectively. Deposition rates of 0.1 - 0.5 Å/s
and 1 - 30 Å/s were used for F
4
TCNQ and NNP, respectively, with 1 - 5 % F
4
TCNQ doping concentrations.
After HTM deposition, the samples were loaded into a metal evaporator and 100 nm Cu contacts were
deposited through a shadow mask at ~ 1 × 10
-6
Torr as a counter electrode. The hole transport material
NNP was synthesized
53
by Chao Wu.
Current density (J) as function of applied voltage (V) characteristics of the solid state dye solar
cell were measured in air at room temperature, in the dark and under spectral mismatch corrected 100
mW/cm
2
white light illumination from an AM-1.5G filtered 300 W Xenon arc lamp (Newport Inc.).
Routine spectral mismatch correction for ASTM G173-03 was performed using a filtered silicon
photodiode, calibrated by the National Renewable Energy Laboratory (NREL) to reduce measurement
errors. Frequency modulated monochromatic light (250 Hz, 10 nm FWHM) and lock-in detection were
used to perform all spectral responsivity and spectral-mismatch correction measurement.
5.3.3 Results and discussion
As shown in Figure 5.6b, the excited state ionization energy ( E
i
* ≈ - 3.8 eV) of the dye relative to
the conduction band edge ( E
c
= - 4.2 eV) of TiO
2
drives photoinduced electron transfer (II) from the dye to
the metal oxide. Moreover, the ionization energy (E
i
≈ - 4.8 eV) of NNP
53
lies between that of the dye and
the work function of polycrystalline Cu, favoring the shuttling of holes from the dye through the doped
NNP layer to the Cu electrode. Doping of the NNP layer with F
4
TCNQ, a strong oxidant, results in the
appearance of vacant electronic levels, through which charge carriers can migrate via a localized hoping
mechanism, as illustrated in Figure 5.6c (III). The DC conductivity of this chemically doped HTM is
relatively high, with a value on the order of σ ~ 1.9 × 10
-3
S/cm, measured in an FTO/NNP:F
4
TCNQ/Cu
planar configuration. The measured σ for this HTM is comparable to that of archetypal hole transport
layers commonly studied for organic light emitting device applications,
54,55
where conductivity
enhancement is believed to arise from a dopant induced narrowing of the depletion region that facilitates
136

tunneling at the electrical contact.
54
The electron affinity of NNP is estimated to be less negative than E
i
*
of N719 by > 2.0 eV, precluding efficient electron transfer from the excited dye molecule to the HTM, due
to the large energy barrier, ΔE ~ 100 kT. As previously mentioned, the energy offset between the TiO
2

conduction band edge and E
i
* of N719, favors electron injection from the photoexcited dye to the TiO
2

nanowire and subsequent collection at the FTO electrode. The thermodynamic open-circuit voltage limit is
expected to be governed by the energy offset between the Fermi level of TiO
2
and the HOMO level of the
NNP layer.

Figure 5.7 a) Top view SEM image of as-grown TiO
2
nanowire array on FTO substrate prior to NNP infiltration.
b) Cross-sectional SEM image of the nanowire array showing individual nanowires growing vertically on FTO
substrate with a length of 200 nm. c) Top view SEM image of TiO
2
nanowire array after HTM deposition
showing the encapsulated nanowires with NNP molecules forming a percolating network. d) Cross-sectional
SEM image of 200 nm long TiO
2
nanowires coated with doped NNP showing infiltration of HTM molecules
within the nanowire array.

137
To minimize recombination losses (cf. IV in Figure 5.6c), the molecules of the hole transport layer
and the adsorbed dye must be situated in close spatial proximity, within the sum of their van der Waals
radii, to insure direct intermolecular orbital overlap. This requires extensive physical infiltration of the
HTM into the interstices of the TiO
2
network. Achieving this infiltration by liquid-based solution
deposition techniques has proven difficult in practice, apparently due to mass transport limitations. That is,
solvent evaporation and premature nucleation of the solute increase the solution’s viscosity and the HTM
molecules begin to agglomerate before a high quality conformal coating is achieved. Thus, incomplete
infiltration of the hole transport layer via liquid-based solution deposition is a significant challenge.
35,39

In contrast, OVPD employs thermal-gradient-assisted mass transport to induce conformal
deposition of the HTM layer via physical vapor condensation from within the nanostructure. Figure 5.7
depicts top view and cross-sectional SEM images before and after HTM deposition obtained for a typical
array of vertically aligned nanowires with individual diameter of ca. 80 nm and length of ca. 200 nm.
Respectively, the top-view and cross-sectional SEM images in Figures 5.7c and 5.7d illustrate the TiO
2

nanowires from Figures 5.7a and 5.7b, following OVPD-deposition of a nominally 300 nm HTM layer.
From the top-view image, the conformally coated TiO
2
array, which, prior to HTM deposition appeared as
sharp monolithic features, appears encapsulated with the doped NNP material that is beginning to form a
percolating network. Importantly, the SEM image in Figure 5.7d illustrates the absence of any HTM
overlayer that may potentially coat only the pinnacle face of the individual nanowires in the titania array.
Overlayer formation would suggest substantial growth in an undesirable orientation coplanar with the
substrate surface and little coaxial coating. Instead, a fairly homogenous distribution of HTM material is
observed throughout the entire height profile of the nanowire array, obscuring the resolvability of
individual nanowires under the HTM coating in the cross-sectional image. Thus, the OVPD method
appears to allow the HTM to penetrate deep inside the nanowire network and, thus, should allow collection
of charge carriers from dye molecules attached along the entire TiO
2
surface.
The potential for these OVPD-grown HTM layers to operate in sDSC devices was assessed under
simulated solar and monochromatic illumination. Figure 5.8a illustrates typical current density (J) versus
voltage (V) characteristics and the corresponding output power density (P = JV) for OVPD sDSC devices
138

with 200 nm nanowires. Red and blue traces correspond to the current density measured in the dark and
under illumination, respectively. The output power density of the cell is shown as the open circle trace, for
which the maximum point on the curve corresponds to the maximum power output density (P
max
). This cell
exhibits rectifying behavior under dark and shows a significant photovoltaic effect under illumination. For
an incident power density (P
inc
) of 100 mW/cm
2
, the cell exhibits a short circuit current (Jsc) of 0.56
mA/cm
2
, open circuit voltage (V
oc
) of 0.355 V, and a fill factor (FF) of 46%, resulting in a power
conversion efficiency (η
P
= P
max
/P
inc
) of 0.1%.
a) b)
c) d)
Figure 5.8. a) Current density versus voltage characteristics for OPVD-fabricated sDSCs prepared using 200 nm
nanowires in the dark (red) and under simulated solar illumination (blue). The open circle trace in the lower panel
represents the output power density of the device under illumination. b) External quantum efficiency (EQE, filled
blue semicircles), light harvesting efficiency (LHE, open red circles), and estimated internal quantum efficiency
(IQE, filled black circles) for OVPD-fabricated sDSC on 200 nm TiO
2
nanowire array. c) Transmittance (open
black circles) for 300 nm NNP:F
4
TCNQ hole transport layer and reflectance spectrum (open red squares) for
copper. d) Overlaid photocurrent (unfilled symbols) and power output (filled symbols) under simulated solar
illumination for OPVD-fabricated sDSC for 200 nm (black) and 400 nm (red) nanowire samples coated by OVPD
with 300 nm and 500 nm thick HTM layers.
139
The external quantum efficiency (EQE) for these sDSC devices is presented as the blue semicircle
trace in Figure 5.8b. The EQE of ca. 5% is relatively low due primarily to poor photon absorption, as
outlined below. Since the length of the nanowires is only ~ 200 nm, the amount of dye present in the cell is
orders of magnitude lower than that in nanoparticle-based devices, where the typical thickness of TiO
2

layer is 5-8 µm. Using dye desorption analysis, surface coverage of 0.586 nmol/cm
2
was calculated, which
is significantly lower than thick nanoparticle film-based devices. Dye-free control devices, fabricated with
TiO
2
nanowires infiltrated with doped NNP exhibited the expected result, with no photocurrent production
at short-circuit and featureless quantum efficiency signal below the detection limit of the instrumentation
employed in this study. These data corroborate the observed photocurrent contribution due to N719.
From SEM image analysis, 5 faces comprise the available area for dye adsorption on an individual
nanowire, the area of the top face being 80 nm × 80 nm and the area of side face being 200 nm × 80 nm.
The total surface area of one nanowire is 70400 nm
2
and assuming a packing density of 25%, a weight
density of 4.25 g/cm
3
for rutile TiO
2
, a specific surface area of 12.9 m
2
/g was calculated. Given an area of
1.46 nm
2
occupied by each dye molecule, one obtains a geometric roughness factor of ~ 5. From this the
relative number of photons absorbed, or light harvesting efficiency (LHE), can be calculated as a function
of wavelength as LHE(λ) = 1 – 10
-Γσ(λ)
, where Γ represents the number of sensitizer molecules per unit area
in mol/cm
2
, and σ is absorption cross section in cm
2
/mol. Given the spectral reflectance of copper
56
shown
in Figure 5.8c is only 25 – 30% between the wavelength range of λ = 400 – 550 nm and the optical
transmittance of the NNP:F
4
TCNQ layer is less than 90% from λ = 400 – 500 nm, the LHE may be
estimated as the maximum possible number of photons absorbed by dye molecules on the TiO
2
surface in a
single optical pass. The resulting quantity is plotted in Figure 5.8b as the red open circle trace for N719 on
a 200 nm nanowire support. The peak LHE is only around 17%, suggesting that the efficiency of this
device is mainly limited by lack of photon absorption in the dye layer, due to relatively low optical density.
The estimated internal quantum efficiency (IQE = EQE/LHE) for these 200 nm nanowire sDSC devices is
presented in Figure 5.8b as the filled black circle trace. With IQE ca. 30%, these devices are comparable to
conjugated polymer based cells, such as TiO
2
/MEHPPV,
57
TiO
2
/P3HT,
33,58-60
and to TiO
2
/spiro-OmeTAD
based solid-state dye-sensitized solar cells.
35,37

140
To further probe the impact that the number of dye molecules available for photon absorption has
in determining the photocurrent for these OVPD-fabricated sDSC, the length of the incorporated nanowires
was increased. As previously shown,
46
the nanowire synthesis can be tuned to afford nanowires of varying
aspect ratio, by controlling the preparation conditions. Nanowires grown in this fashion are highly
crystalline as determined from high resolution transmission electron microscope (HR-TEM) imaging and
selected area electron diffraction (SAED) analysis.
46
With the (001) face growing the most rapidly and the
(110) face growing more slowly, this method typically yields nanowires with a square cross-section and
sidewalls bound by (001) faces. Nanowires of ~ 400 nm length were prepared and incorporated into
OVPD-fabricated sDSC devices, with the resulting electrical characteristic presented as the red traces in
Figure 5.8d and overlaid with reference data for 200 nm nanowire devices. These thicker nanowire devices
were found to yield an increase of 33% in the power conversion efficiency, due largely to an increase in
short circuit current density (J
sc
), from 0.6 mA/cm
2
for the 200 nm TiO
2
devices to 0.9 mA/cm
2
for the 400
nm TiO
2
devices. Note that slightly less than a two-fold enhancement in J
sc
was observed despite the
increase in nanowire length by a factor of two. This is due to the nanowire growth dynamics, by which
increasing the length also gives rise to larger diameter nanowires. This reduces the packing density of the
nanowires on the substrate and, thus, reduces the total internal surface area available for dye adsorption.
Presently, further attempts to increase the nanowire length using extended growth time have been met with
limited success, since the titania eventually coalesces and the opening to the interstitial space becomes
constricted. Growth of longer nanowires with small diameter may significantly enhance the performance
of these devices, which will be the focus of future work.
5.3.4 Concluding remarks
The results presented here for the OVPD-based fabrication of model devices represent a promising
new avenue for achieving intimate HTM infiltration in future solid-state dye sensitized solar cells. This
work outlines several characteristics pertinent to efficient hole transport in sDSCs. 1) The conductivity of
the HTM layer should be extremely high, 2) the HTM must effectively infiltrate the TiO
2
network to collect
all the photogenerated charge carriers, (3) and the ionization energy of the HTM should less negative than
141
that of the dye molecule in order to facilitate hole transport. Each of these characteristics is addressable in
the vapor deposited hole transport system presented here as follows: 1) The conductivity of the NNP layer
can be controlled by doping the film with an electron accepting material such as F
4
TCNQ. The resulting
HTM conductivity was estimated for doped NNP layers on pristine FTO substrates with 100 nm thick Cu
contact pads. Typical two-probe I-V measurements on 5% F
4
TCNQ doped NNP layers yield conductivity
for this doped HTM material on the order of 2 mS/cm. 2) As shown in Figure 5.7, the vapor deposited NNP
layer efficiently infiltrates the nanowire network. This is achieved via deposition rate optimization for the
HTM from 1 Å/s - 30 Å/s Growth rates in excess of 20 Å/s tend to result in poor hole transport layers for
F
4
TCNQ-doped NNP. 3) The estimated hole transport level for NNP is -4.8 eV
27
, making this material, an
attractive candidate when doped with F
4
TCNQ for transporting holes from photooxidized dye molecules on
the TiO
2
surface.
This work demonstrates use of the physical condensation method of OVPD as a solvent-free
method for sDSC fabrication. Development of sDSC devices is essential to circumnavigate the strict
encapsulation requirements for the general outdoor deployment of dye-sensitized solar cells. In this regard,
a vapor deposited solid-state hole transport layer represents a compelling method for sDSC development.
5.4 Carbon-based transparent electrodes
5.4.1 Carbon nanotube network devices
A critical component of organic photovoltaic cells is the transparent electrode, through which light
couples into the device and allows these devices to be easily manufactured on lightweight flexible
substrates. Conventional OPVs typically use transparent indium tin oxide (ITO) or fluorine doped tin oxide
(FTO) as such electrodes.
61,62
However, limited indium reserves, intensive processing requirements, and
the highly brittle nature of metal oxides
15,63
impose serious limitations on the use of these materials for
142

applications where cost, physical conformation, and mechanical flexibility are germane. This section
examines the performance of solar cell devices fabricated with alternative flexible carbon-based transparent
electrodes, such as carbon nanotube (CNT) networks and graphene obtained by chemical vapor deposition.
Numerous applications for carbon nanotubes have been proposed
64-67
ranging from nanometer
dimensional sensors to mechanically robust composites. Recently, macroscale transparent electrodes
comprising networks CNT bundles have been proposed as a potential alternative to ITO in optoelectronic
devices
68-70
Such networks can be obtained via a stamp transfer method, the preparation and
characterization of such films has been reported elsewhere.
71
Briefly, CNT electrodes in the present study
where prepared by Koungmin Ryu in the laboratory of Professor Chongwu Zhou by dispersing P3
nanotubes, obtained by arc discharge from Carbon Solutions Inc., in 1 wt % aqueous sodium dodecyl
sulfate (SDS), using probe sonication and subsequent centrifugation. The resulting suspension was diluted
and filtered over an alumina membrane with 200 nm pore size to form a CNT mat on the order of 40 nm
thick. The CNT mat was subsequently transferred from the filter membrane to the target substrate using
poly(dimethysiloxane) (PDMS) stamp. The resulting electrode material exhibits a substantial tradeoff
between transparency and sheet resistance, since the conductivity of the electrode depends on the amount
of CNT material deposited, with typical sheet resistance on the order of a few hundred ohms per square at
transparencies greater the 80%. Sheet resistance (R
sh
) as low as R
sh
= 100 Ω/sq. are obtainable, however,

Figure 5.9. Electrical characteristics for flexible OPV devices based on transparent electrode materials
comprising carbon nanotube networks (square symbol trace) or ITO (circle symbol trace).
143

the transparency for these samples was on the order of 50 – 60 % in accord with previous observations
71

and suggesting a major challenge in employing such electrodes in OPV devices.
Compared in Figure 5.9 are the JV characteristics in the dark and under simulated 1 sun
illumination for OPV devices fabricated on ITO or CNT coated with poly(3,4-ethylenedioxythiophene)
poly(styrenesulfonate) (PEDOT) and a polyethylene terephthalate substrate. The active layer employed in
these devices consisted of / Copper phthalocyanine (CuPc) [40 nm] / Fullerene (C
60
) [40 nm] /
Bathocuproine (BCP) [10 nm] / Aluminum (Al). The V
oc
for both devices are comparable, with values of
ca. 0.43 V. However, from the JV data, a clear loss in photocurrent is exhibited by the CNT device with J
sc

= 3.30 mA/cm
2
in contrast to J
sc
= 4.35 mA/cm
2
for the ITO-based device. Moreover, the FF calculated for
the CNT-based device (FF = 0.47) is slightly lower than that for the ITO-based device (FF = 0.52). The
FF losses appear to arise from a greater series resistance (R
s
) in the CNT case (R
s
A = 14 Ωcm
2
) relative to
the ITO device (R
s
A = 5 Ωcm
2
), as estimated according to the modeled dark current (gray trace) in the
semi-log scale plot of Figure 5.9. The origin of the low photocurrent in the CNT-based devices appears to
arise from an optical attenuation effect induced by the poor transmittance of the CNT electrode as
illustrated in the top panel of Figure 5.10. The transmittance for this CNT electrode rises steadily

Figure 5.10. From upper panel to lower panel, optical transmittance for CNT (solid trace) and ITO (dotted trace)
electrodes used in Figure 5.9, absorbance of the OPV active layer used in Figure 5.9, external quantum efficiency
for OPV devices in Figure 5.9.
144
throughout the UV and visible portions of the spectrum from 50% at 400 nm. However, within this region
it never reaches the ca. 70% transmittance exhibited by ITO. This effectively reduces the number of
photons absorbed in the CNT device active layer, leading to a lower exciton generation rate in the CNT-
based device. The resulting loss in external quantum efficiency is illustrated in the lower panel of Figure
5.10 for the CNT-based device relative to the ITO-based device. These results highlight the difficulty
associated with incorporating CNT-based planar electrodes as the anode material in OPV devices. That is,
typical CNT samples are composed of a mixture of semiconducting and metallic species, where the
semiconducting tubes introduce transmittance losses and do not aid in transporting charge. Additionally,
the junction resistance between nanotubes also plays a significant role in determining the overall sheet
resistance of the resulting electrode. Thus, a more transparent electrode can be obtained through sample
preparation with a thinner electrode. However, this increases the sheet resistance and will therefore induce
further FF losses for the CNT-based device compared to the ITO-based device.
5.4.2 Graphene electrode devices
In contrast, graphene is a one-atom thick, two-dimensional crystalline arrangement of carbon
atoms with a quasi-linear dispersion relation, and predicted mobility on the order of 10
6
cm
2
/V·s for a
charge carrier concentration n
i
~ 10
12
cm
-2
.
72
A graphene monolayer has a transparency of 97-98%
73
and
the sheet resistance of undoped graphene is of the order of ~6kΩ.
74-77
These qualities suggest graphene
films as suitable candidates for alternative transparent conductive electrodes, where low sheet resistance
and high optical transparency are essential. Conventional methods to obtain graphene thin films, such as
epitaxial growth,
78
micromechanical exfoliation of graphite,
79
and exfoliation of chemically oxidized
graphite
80,81
are either expensive, unscalable, or yield graphene with limited conductivity due to a high
defect density. Recently, graphene films obtained from reduced graphene oxide (GO) have been explored
for applications as transparent electrodes in solar cells.
82-84
However, the devices obtained exhibited rather
moderate performance, leakage current under dark conditions, and moderate power conversion efficiency
of < 0.4 %. The moderate performance of these devices may be attributed to several factors, including: i)
reduction of oxygen functionalities on the graphene oxide flakes does not completely restore the film’s
145
π-conjugation, and ii) the vacuum filtration or spin coating methods used to prepare reduced graphene
oxide films lead to stacked graphene flakes and thus significant flake-to-flake resistance. For instance, Eda
et. al.
82
reported doped reduced graphene oxide films with a sheet resistance of 40 kΩ/sq, a transparency of
64%, and a solar cell conversion efficiency of 0.1%, while Wu et. al.
84
reported reduced graphene oxide
films of 5–10
3
kΩ/sq, >80% transparency, and a conversion efficiency of 0.4%. Efforts to improve
percolation on the graphene electrode include the use of reduced GO combined with carbon nanotube films,
but this approach requires extra processing steps.
85
As a result, continuous, highly flexible, and transparent
graphene films are still highly desirable for photovoltaic applications.
Chemical vapor deposition (CVD)
86-89
has recently been demonstrated as a compelling method of
obtaining high quality graphene films. In particular, films with sheet resistance of 280 Ω/sq (80%
transparent) and 770 Ω/sq (90% transparent) have been reported for graphene synthesized on Ni films,
while sheet resistance of 350 Ω/sq (90% transparent) has been reported for CVD graphene on Cu films,
which represents an advance in the use of graphene as transparent conductor. Graphene obtained by such
methods is potentially scalable for large area device fabrication.
89
As a result, flexible OPV devices were
fabricated using CVD-graphene films synthesized and characterized by Lewis Gomez De Arco in the
laboratory of Professor Chongwu Zhou. Such films exhibit tunable sheet resistance and transparency
between 230 Ω/sq at 72% transparency, and 8.3 kΩ/sq at 91% transparency. The use of CVD graphene is
attractive because alternative methods produce films comprising stacked micron-size flakes, which suffer
from flake-to-flake contact resistance. In contrast, grain boundaries of CVD graphene films have the
advantage of being formed in situ during synthesis, which is expected to minimize contact resistance
between neighboring graphene domains. Solar cells fabricated with CVD graphene exhibited performance
that compares to ITO devices and surpasses that of ITO devices under bending conditions, exhibiting
power conversion efficiencies of 1.18% and being operational under bending conditions up to 138°.
Graphene films were synthesized by chemical vapor deposition on a nickel surface and transferred
to the target substrate as detailed elsewhere.
8,89
Briefly, poly-methylmethacrylate (PMMA) was deposited
atop the as-synthesized graphene on Si/SiO
2
/Ni substrates, etching of the nickel rendered free-standing
146

PMMA with the synthesized graphene adhered to it, which facilitates transferring the graphene film to
target substrates as illustrated schematically in Figure 5.11.
Optical transmission in graphene is primarily dictated by photon absorption. As a consequence, as
CVD graphene films become thinner, transparency is expected to increase. In principle, the sheet
resistance of a graphene film comprised of several graphene layers should decrease for each additional
layer,
90
therefore it is expected that the thicker the film, the larger the number of layers, the smaller the
sheet resistance, but simultaneously, the lower the transparency. Optical transmittance data illustrating this
is shown for several thicknesses of CVD graphene in Figure 5.11. As such, highly transparent CVD
graphene films can be obtained at the expense of higher resistance. Sheet resistance as low as 230 Ω/sq
(with T=72%) and optical transparency as high as 91% (with R
Sh
= 8.3 kΩ/sq) were achieved, and therefore
a compromise between these parameters must be met for specific applications.
To investigate the flexibility of the CVD graphene electrodes and its influence on the performance
of flexible OPV cells, CVD graphene films were transferred onto PET substrates and the electrical
conductivity of graphene and ITO films were compared under bending conditions. Figures 5.12b and 5.12c
show AFM images of CVD graphene (R
Sh
= 500 Ω/sq and T = 75 %) and ITO (R
Sh
= 25 Ω/sq, T = 86 %),
on PET. Aluminum metal contacts were thermally deposited through a shadow mask onto the
aforementioned films. Two-probe electrical measurements were performed on both films by direct contact
of tungsten micro-probes to the aluminum electrodes, soldering the probe tips to the aluminum pads to
a) b)
Figure 5.11. a) Schematic of the CVD graphene transfer process onto transparent substrates. b)
Transmittance/resistance data for the resulting electrodes from a).
147

assure electrical contact during the measurement. Bending angle dependence of the film conductance was
monitored as shown in the inset of Figure 5.12d. The conductance of the graphene/PET film remained
virtually unperturbed by bending (cf. Figure 5.12d) even after several complete bending cycles and
decreased by only 7.9% after 100 bending cycles. In contrast, Figure 5.12f shows three clearly defined
regions that describe the typical behavior of ITO conductance under bending conditions. For bending
angles from 0° to ~130° a steady decrease in the conductance of the ITO film by three orders of magnitude
with increased bending angle was observed. Immediately, upon reaching a critical angle ca. 130°,
conductance in the ITO sample drops by six orders of magnitude. After the critical angle was reached, the
conductance of the film remained poor, even with decreasing bending angle. An open circuit (σ ≤ 10
-12
S)
was obtained after only one bending cycle. The fact that the conductivity of the ITO film did not recover

Figure 5.12. a) Photograph illustrating high flexibility of CVD graphene transferred on a PET flexible substrate.
b,c) AFM images of the surface of CVD graphene and ITO films on PET, respectively. d,f) Conductance of the
CVD graphene and ITO films on PET substrates under bending conditions, respectively. The devices used to
monitor the conductance had channel width (W) 1 mm and length (L) 1 mm. e) Optical images of CVD graphene
(top) and ITO (bottom) films on PET before and after being bent at the angles specified in panels b and c. Arrows
show the direction of the bending.
148
after bending the ITO film back to lower radius of curvature can be associated with the development of
multiple discontinuity scattering sites on the brittle ITO film that were generated by tensile strain under
bending and may further develop under compressive stress while decreasing the bending angle. Optical
microscopy images were collected on the ITO/PET and Graphene/PET films. Figure 5.12e shows optical
micrographs of graphene and ITO films before and after the first bending cycle (0°150°0°). Very
pronounced cracks were observed in the ITO film, while the graphene film remained intact. These results
demonstrate the advantage of CVD graphene in terms of mechanical flexibility over ITO films, suggesting
the potential for robust, flexible, and lightweight transparent CVD graphene electrodes in OPVs.
Although CVD graphene films outperform ITO as transparent conductive electrodes on flexible
PET substrates under bending conditions, it is important to implement this material into working OPVs in
order to evaluate its performance. Thus, OPV cells on PET substrates were fabricated using graphene and
ITO as transparent electrodes, under identical experimental conditions. Graphene electrodes were
fabricated by transferring as-grown CVD graphene films onto pre-cleaned, 100 µm thick, PET substrates.
PET substrates coated with ITO were obtained from Southwall Technologies Inc. Both substrates were
solvent cleaned and passivated by spin coating a thin layer (10 nm) of PEDOT:PSS with R
Sh
= 1 kΩ/sq. Use
of the PEDOT:PSS coating decreased the conductivity of the PEDOT:PSS/CVD Graphene film to 2.1
kΩ/sq, while for the PEDOT/ITO film it remained ~1 kΩ/sq. PEDOT:PSS was expected to help mitigate
the brittle nature of the ITO electrode to enhance its performance under bending conditions. Finally, the
planarizing effect afforded by the PEDOT:PSS treatment is desirable to compensate for possible folding or
wrinkles that may accompany the CVD graphene film transfer process or irregular wetting between the
electrode and the cell active layers, which would yield device shorting or shunt losses.
The substrates were taken into high vacuum conditions, where the organic thin films and the
aluminum cathode were consecutively deposited by thermal evaporation. The multilayered configuration
employed (Fig 1a) is given as: CVD graphene [<5 nm] or ITO / PEDOT:PSS / Copper phthalocyanine
(CuPc) [40 nm] / Fullerene (C
60
) [40 nm] / Bathocuproine (BCP) [10 nm] / Aluminum (Al). Aluminum
cathodes were deposited through a shadow mask with circular openings with 1 mm diameter.
149

Current density vs. voltage or J(V) characteristics were measured in air at room temperature in the
dark and under spectral mismatch corrected 100 mW/cm
2
white light illumination from an AM 1.5G
filtered 300 W Xenon arc lamp (Newport Co.). Routine spectral mismatch correction for ASTM G173-03
was performed using a filtered silicon photodiode calibrated by the National Renewable Energy Laboratory
(NREL) to reduce measurement errors. Frequency modulated monochromatic light (250 Hz, 10 nm
FWHM) and lock-in detection was used to perform all spectral responsivity and spectral mismatch
correction measurements. The J(V) characteristics of a typical photovoltaic cell obtained with CVD
graphene (R
sh
: 3.5 kΩ/sq, T: 89%) were compared against a typical cell obtained with an ITO anode (R
Sh
:

Figure 5.13. Logarithmic (top) and linear (bottom) current density and power density vs. voltage characteristics
of CVD graphene a) and ITO b) OPV cells on PET under dark (red traces) and 100 mW/cm
2
AM1.5G spectral
illumination (blue traces). The output power density of the cells is plotted in panels a and b as open circle traces.
The structure of the devices is given by [CVD graphene/PEDOT/CuPc/C60/BCP/Al] and
[ITO/CuPc/C60/BCP/Al] for CVD graphene and ITO OPVs, respectively. c) Comparison of the modeled (solid
lines) current density and power density curves obtained from the Shockley equation against the graphene and
ITO device values obtained experimentally (symbols).
150

25 Ω/sq, T: 96%), that were fabricated under identical experimental conditions. Figures 5.13a and 5.13b
show semi-log (up) and linear (down) J(V) plots obtained from CVD graphene and ITO OPV cells,
respectively. Red and blue traces correspond to the current density measured in the dark and under
illumination, respectively. The output power density of the cells (P), which is given by P = J·V, is shown
in Figures 5.13a and 5.13b as open circle traces for which the maximum point on the curve corresponds to
the maximum output power density (P
max
) of the device. For an incident power density, P
inc
= 100
mW/cm
2
, the power conversion efficiency (η = P
max
/P
inc
) and other performance parameters are
summarized in Table 5.1. It is clearly observed from the semi-log plots in Figures 5.13a and 5.13b that
both devices have nearly identical open circuit voltage (V
oc
) (for J=0) of 0.48 V under illumination
conditions, which suggests similar recombination behavior in both cells. Furthermore, from Figure 5.13a, it
can be seen that unlike OPVs reported for reduced GO anodes,
23,24
minimal leakage current densities were
observed for the CVD graphene OPV cells.
The J(V) characteristics of the CVD graphene cell under illumination showed a short-circuit
photocurrent density (J
sc
) (for V=0) of 4.73 mA/cm
2
, an open-circuit voltage (V
oc
) of 0.48 V and a
maximum power (P
max
) of 1.18 mW/cm
2
, to yield a fill factor (FF) of 0.52 and overall power conversion
efficiency (η) of 1.18%. The control device, using an ITO anode on PET, gave J
sc
of 4.69 mA/cm
2
, V
oc
of
0.48 V and P
max
of 1.27 mW/cm
2
, for a FF of 0.57 and an efficiency of 1.27%. Analysis of Figures 5.13a
and 5.13b reveals that despite the lower transparency and higher R
Sh
of the CVD graphene electrode, CVD
graphene solar cell exhibits an output power density nearly 93% of that shown by the ITO device.
Additionally, CVD graphene OPV cells were observed to be more sensitive to the anode conductivity, and
Table 5.1. Performance metrics for graphene OPV cells fabricated on PET.
a
Anode J
sc
(mA/cm
2
) V
oc
(V) FF η (%)
CVD graphene 4.73 0.48 0.52 1.18
ITO 4.69 0.48 0.57 1.27
a
The structure of the devices is given by [CVD graphene/PEDOT/CuPc/C60/BCP/Al] and
ITO/PEDOT/CuPc/C60/BCP/Al] for CVD graphene and ITO OPVs, respectively.
151
hence, to its capacity to pull holes from the active layers than to its transparency. The fact that the two cells
gave very similar device performance is encouraging, especially considering that the ITO substrate gave
~100-fold lower R
Sh
and higher transparency than the CVD graphene film, which would favor the
performance of the ITO device. This may be rationalized by considering that, as demonstrated above, the
sheet resistance increases to similar values on both electrodes after being coated with PEDOT:PSS. In this
case, charge injection from the active layers of the OPV cells may be limited by the PEDOT:PSS layer,
thus yielding similar performance on both cells. Devices fabricated on PET/PEDOT:PSS substrates without
graphene or ITO produced open circuit characteristics. Although PEDOT:PSS was used on both, graphene
and ITO OPV cells, the performance of the cells was measured by puncturing the PEDOT:PSS layer to
contact the underlying electrode material, which confirms that CVD graphene and ITO anodes, instead of
PEDOT:PSS are the ultimate electrodes in the hole extraction process of the devices.
To estimate the impact of resistive losses on device performance the J(V) dependence under
illumination was modeled according to a modified form of the Shockley equation, which is commonly
applied to describe the current density (J) vs. voltage (V) characteristics of organic solar cells, given by:
(5.1)
where R
s
, R
p
, J
s
, J
ph
, n, and V
t
are the lumped series resistance, lumped parallel resistance, reverse-bias
saturation current density, photocurrent density, diode ideality factor, and thermal voltage respectively for a
single diode circuit model. As a practical matter, the transcendental nature of Eq. 5.1 was resolved by
expressing it in terms of the Lambert-W function
91
to give

(5.2)
where W
0
represents Lambert’s function of the form W(x)e
W(x)
=x(V).
91-94

152
In Figure 5.13c, the modeled J(V) and output power density obtained according to Eq. 5.2, are
plotted as solid lines for the CVD graphene and ITO cells depicted in Figures 5.13a and 5.13b, respectively.
The modeled data are compared against the experimentally measured values, plotted as open symbols in
Figure 5.13c, demonstrating that these CVD graphene based devices may be described by the generalized
Shockley equation in the same way that their ITO based counterparts are commonly discussed. Modeling
the data in this way allows us to estimate to what extent series resistive losses, parallel conductance, and
recombination processes may impact device performance. The model ideality factors, parallel resistances
and saturation current-densities were all comparable for the ITO and CVD graphene devices under
illumination, having values of n = 2.4 and 2.6, R
p
=1.47 kΩcm
2
and 1.62 kΩcm
2
, and J
s
=2.0 µA/cm
2
and
3.1 µA/cm
2
,
respectively, suggesting that the recombination and leakage processes are similar for both
devices. The model series resistance calculated from Eq. 5.2 for the CVD graphene device is 12.6 Ωcm
2
,
which is less than 5 times that of the ITO device with R
s
= 2.6 Ωcm
2
, while the model photocurrent density
(J
ph
) for the CVD graphene device (4.75 mA/cm
2
) is higher than J
ph
for the ITO device (4.66 mA/cm
2
).
This indicates that the power output of the graphene based device is primarily limited by charge transport
losses rather than optical transmittance losses. This constitutes a very promising result for CVD graphene
transparent electrodes, which perform comparably to ITO, despite carrying a relatively higher sheet
resistance. Given the comparable performance of OPVs with graphene and ITO electrodes, the question
remains if such devices will perform well under strain-stress conditions. Current-voltage characteristics
under bending of CVD graphene and ITO solar cells are shown in Figures 5.14a and 5.14b, respectively.
The observed performance of both devices was slightly degraded upon bending. For instance, solar cells
using CVD graphene electrodes withstood bending angles (curvature radii, surface strain) up to 138° (4.1
mm, 2.4%) while exhibiting reasonable solar cell performance. In sharp contrast, ITO cells only withstood
bending to 36° (15.9 mm, 0.8%) while showing poor performance, and failed completely to become an
open circuit after being bent to 60° (9.5 mm, 1%).
It is important to note that, with increased bending angle, the current density dropped for CVD
graphene and ITO devices, while their open circuit voltage remained virtually unchanged. In some cases
this effect can be associated with decreased illumination of the devices during bending. However, as both
153

cells are subjected to similar bending conditions, the marked difference exhibited in the conversion
efficiency between them cannot be attributed to irregular illumination induced by bending, but may be
related to the presence of micro cracks on the ITO device. To understand this, consider the fill factor vs.
bending angle data for the OPV cells with CVD graphene and ITO electrodes plotted in Figure 5.14c. The
fill factor (FF=P
max
/J
sc
V
oc
) depends strongly on the output power of the cell, and is directly related to the
cell conversion efficiency (η) by
(5.3)

Figure 5.14. Current density vs voltage characteristics of CVD graphene a) or ITO b) photovoltaic cells under
100 mW/cm
2
AM1.5G spectral illumination for different bending angles. Insets show the employed experimental
setup. c) Fill factor dependence of the bending angle for CVD graphene and ITO devices. d) SEM images
showing the surface structure of CVD graphene (top) and ITO (bottom) photovoltaic cells after being subjected to
the bending angles described in panels a and b.
154
Gradual degradation of the initial fill factor, and hence, the conversion efficiency was observed on the
CVD graphene cell as the bending angle increased. In contrast, the fill factor of the ITO device rapidly
decayed to zero when bent at around 60°. Changes in film topology introduced by bending of the devices
were investigated by SEM measurements. Figure 5.14d shows the appearance of micro-cracks throughout
the ITO device, while no signs of micro-cracks or fissures were observed on the graphene device.
Development of micro-cracks generated by mechanical stress in ITO, even at small bending angles, can
substantially increase the film resistance, which has a key impact in reducing the fill factor. This agrees
well with the observed decrease in output current density and power conversion efficiency of the solar cells
without observing appreciable change in the V
oc
. CVD graphene, being of organic nature and more flexible,
surpasses the performance of ITO, which may easily crack under slight bending albeit PEDOT:PSS
passivation. Therefore, the brittle nature of ITO plays a major role in the poor performance of ITO-flexible
organic solar cells, while the CVD graphene thin films exhibited good performance as flexible transparent
electrodes.
5.4.3 Concluding remarks
In summary, this work demonstrates a feasible, scalable and effective method to employ CVD
graphene as highly transparent, continuous and flexible electrodes for OPVs. This approach constitutes a
significant advance towards the production of transparent conductive electrodes in solar cells. CVD
graphene meets the most important criteria of abundance, low cost, conductivity, stability,
electrode/organic film compatibility and flexibility that are necessary to replace ITO in organic
photovoltaics, which may have important implications for future organic optoelectronic devices.
155
5.5 Chapter 5 endnotes
1 Y. Wang and N. Herron, J. Phys. Chem., 1991, 95, 525.
2 X. G. Peng, L. Manna, W. D. Yang, J. Wickham, E. Scher, A. Kadavanich and A. P. Alivisatos,
Nature, 2000, 404, 59.
3 C. B. Murray, C. R. Kagan and M. G. Bawendi, Annu. Rev. Mater. Sci., 2000, 30, 545.
4 C. Burda, X. B. Chen, R. Narayanan and M. A. El-Sayed, Chem. Rev., 2005, 105, 1025.
5 M. Gratzel, J. Photoch. Photobio. A, 2004, 164, 3.
6 C. Groves, O. G. Reid and D. S. Ginger, Acc. Chem. Res., 2010, 43, 612.
7 R. Giridharagopal and D. S. Ginger, J. Phys. Chem. Lett., 2010, 1, 1160.
8 L. G. De Arco, Y. Zhang, C. W. Schlenker, K. Ryu, M. E. Thompson and C. W. Zhou, Acs Nano,
2010, 4, 2865.
9 P. V. Kamat, J. Phys. Chem. C, 2007, 111, 2834.
10 D. C. Pan, X. L. Wang, Z. H. Zhou, W. Chen, C. L. Xu and Y. F. Lu, Chem. Mater., 2009, 21,
2489.
11 J. Tang, S. Hinds, S. O. Kelley and E. H. Sargent, Chem. Mater., 2008, 20, 6906.
12 M. G. Panthani, V. Akhavan, B. Goodfellow, J. P. Schmidtke, L. Dunn, A. Dodabalapur, P. F.
Barbara and B. A. Korgel, J. Am. Chem. Soc., 2008, 130, 16770.
13 R. G. Xie, M. Rutherford and X. G. Peng, J. Am. Chem. Soc., 2009, 131, 5691.
14 Q. Guo, G. M. Ford, H. W. Hillhouse and R. Agrawal, Nano Lett., 2009, 9, 3060.
15 C. Wadia, A. P. Alivisatos and D. M. Kammen, Environmental Science & Technology, 2009, 43,
2072.
16 K. V. Ragnarsdottir, Nature Geoscience, 2008, 1, 720.
17 R. B. Gordon, M. Bertram and T. E. Graedel, P. Natl. Acad. Sci. USA, 2006, 103, 1209.
18 B. Pejova and A. Tanusevski, J. Phys. Chem. C, 2008, 112, 3525.
19 B. Pejova and I. Grozdanov, Thin Solid Films, 2007, 515, 5203.
20 I. Lefebvre, M. A. Szymanski, J. Olivier-Fourcade and J. C. Jumas, Phys. Rev. B, 1998, 58, 1896.
156
21 S. G. Hickey, C. Waurisch, B. Rellinghaus and A. Eychmuller, J. Am. Chem. Soc., 2008, 130,
14978.
22 M. V. Kovalenko, W. Heiss, E. V. Shevchenko, J. S. Lee, H. Schwinghammer, A. P. Alivisatos
and D. V. Talapin, J. Am. Chem. Soc., 2007, 129, 11354.
23 Y. Xu, N. Al-Salim, C. W. Bumby and R. D. Tilley, J. Am. Chem. Soc., 2009, 131, 15990.
24 M. A. Franzman, C. W. Schlenker, M. E. Thompson and R. L. Brutchey, J. Am. Chem. Soc., 2010,
132, 4060.
25 P. B. Littlewood, J. Phys. C: Solid State, 1980, 13, 4855.
26 M. Jorgensen, K. Norrman and F. C. Krebs, Sol. Energy Mater. Sol. Cells, 2008, 92, 686.
27 H. Neugebauer, C. J. Brabec, J. C. Hummelen, R. A. J. Janssen and N. S. Sariciftci, Synth. Met.,
1999, 102, 1002.
28 B. A. Gregg, J. Phys. Chem. B, 2003, 107, 4688.
29 B. Oregan and M. Gratzel, Nature, 1991, 353, 737.
30 H. J. Snaith and L. Schmidt-Mende, Adv. Mater., 2007, 19, 3187.
31 B. Li, L. D. Wang, B. N. Kang, P. Wang and Y. Qiu, Sol. Energy Mater. Sol. Cells, 2006, 90, 549.
32 L. E. Greene, M. Law, B. D. Yuhas and P. D. Yang, J. Phys. Chem. C, 2007, 111, 18451.
33 G. K. Mor, S. Kim, M. Paulose, O. K. Varghese, K. Shankar, J. Basham and C. A. Grimes, Nano
Lett., 2009, 9, 4250.
34 R. Zhu, C. Y. Jiang, B. Liu and S. Ramakrishna, Adv. Mater., 2009, 21, 994.
35 U. Bach, D. Lupo, P. Comte, J. E. Moser, F. Weissortel, J. Salbeck, H. Spreitzer and M. Gratzel,
Nature, 1998, 395, 583.
36 F. Fabregat-Santiago, J. Bisquert, L. Cevey, P. Chen, M. K. Wang, S. M. Zakeeruddin and M.
Gratzel, J. Am. Chem. Soc., 2009, 131, 558.
37 H. J. Snaith, A. J. Moule, C. Klein, K. Meerholz, R. H. Friend and M. Gratzel, Nano Lett., 2007, 7,
3372.
38 J. H. Yum, B. E. Hardin, S. J. Moon, E. Baranoff, F. Nuesch, M. D. McGehee, M. Gratzel and M.
K. Nazeeruddin, Angew. Chem. Int. Ed., 2009, 48, 9277.
39 I. K. Ding, N. Tetreault, J. Brillet, B. E. Hardin, E. H. Smith, S. J. Rosenthal, F. Sauvage, M.
Gratzel and M. D. McGehee, Adv. Funct. Mater., 2009, 19, 2431.
157
40 M. Baldo, M. Deutsch, P. Burrows, H. Gossenberger, M. Gerstenberg, V. Ban and S. Forrest, Adv.
Mater., 1998, 10, 1505.
41 M. Adachi, Y. Murata, I. Okada and S. Yoshikawa, J. Electrochem. Soc., 2003, 150, G488.
42 G. K. Mor, K. Shankar, M. Paulose, O. K. Varghese and C. A. Grimes, Nano Lett., 2006, 6, 215.
43 H. E. Prakasam, K. Shankar, M. Paulose, O. K. Varghese and C. A. Grimes, J. Phys. Chem. C,
2007, 111, 7235.
44 O. K. Varghese, M. Paulose and C. A. Grimes, Nature Nanotechnol., 2009, 4, 592.
45 C. K. Xu, P. H. Shin, L. L. Cao, J. M. Wu and D. Gao, Chem. Mater., 2010, 22, 143.
46 A. Kumar, A. R. Madaria and C. W. Zhou, J. Phys. Chem. C, 2010, 114, 7787.
47 X. J. Feng, K. Shankar, O. K. Varghese, M. Paulose, T. J. Latempa and C. A. Grimes, Nano Lett.,
2008, 8, 3781.
48 B. Liu and E. S. Aydil, J. Am. Chem. Soc., 2009, 131, 3985.
49 A. Kongkanand, K. Tvrdy, K. Takechi, M. Kuno and P. V. Kamat, J. Am. Chem. Soc., 2008, 130,
4007.
50 K. Zhu, N. R. Neale, A. Miedaner and A. J. Frank, Nano Lett., 2007, 7, 69.
51 J. van de Lagemaat, N. G. Park and A. J. Frank, J. Phys. Chem. B, 2000, 104, 2044.
52 E. Garnett and P. D. Yang, Nano Lett., 2010, 10, 1082.
53 C. Wu, P. I. Djurovich and M. E. Thompson, Adv. Funct. Mater., 2009, 19, 3157.
54 S. H. Baek, C. U. Kim, D. C. Choo, T. W. Kim, J. H. Park, J. H. Seo and Y. K. Kim, Mol. Cryst.
Liq. Cryst. Sci., 2009, 498, 258.
55 M. Pfeiffer, K. Leo, X. Zhou, J. S. Huang, M. Hofmann, A. Werner and J. Blochwitz-Nimoth,
Org. Electron., 2003, 4, 89.
56 L. Q. Wang, G. Z. Liang and G. C. Dang, Spectrochim. Acta, Part A, 2005, 61, 1021.
57 B. Li, L. D. Wang, B. N. Kang, P. Wang and Y. Qiu, J. Photoch. Photobio. A., 2005, 172, 135.
58 K. J. Jiang, K. Manseki, Y. H. Yu, N. Masaki, K. Suzuki, Y. L. Song and S. Yanagida, Adv. Funct.
Mater., 2009, 19, 2481.
59 S. S. Williams, M. J. Hampton, V. Gowrishankar, I. K. Ding, J. L. Templeton, E. T. Samulski, J.
M. DeSimone and M. D. McGehee, Chem. Mater., 2008, 20, 5229.
158
60 C. Goh, S. R. Scully and M. D. McGehee, J. Appl. Phys., 2007, 101.
61 P. Peumans, A. Yakimov and S. R. Forrest, J. Appl. Phys., 2003, 93, 3693.
62 A. Andersson, N. Johansson, P. Broms, N. Yu, D. Lupo and W. R. Salaneck, Adv. Mater., 1998,
10, 859.
63 M. Boehme and C. Charton, Surface & Coatings Technology, 2005, 200, 932.
64 R. H. Baughman, A. A. Zakhidov and W. A. de Heer, Science, 2002, 297, 787.
65 P. Serp, M. Corrias and P. Kalck, Appl. Catal., A, 2003, 253, 337.
66 E. Frackowiak and F. Beguin, Carbon, 2001, 39, 937.
67 A. S. Arico, P. Bruce, B. Scrosati, J. M. Tarascon and W. Van Schalkwijk, Nat. Meter., 2005, 4,
366.
68 A. D. Pasquier, H. E. Unalan, A. Kanwal, S. Miller and M. Chhowalla, Applied Physics Letters,
2005, 87, 203511.
69 M. W. Rowell, M. A. Topinka, M. D. McGehee, H. J. Prall, G. Dennler, N. S. Sariciftci, L. B. Hu
and G. Gruner, Applied Physics Letters, 2006, 88, 233506.
70 J. van de Lagemaat, T. M. Barnes, G. Rumbles, S. E. Shaheen, T. J. Coutts, C. Weeks, I. Levitsky,
J. Peltola and P. Glatkowski, Applied Physics Letters, 2006, 88, 233503.
71 D. H. Zhang, K. Ryu, X. L. Liu, E. Polikarpov, J. Ly, M. E. Tompson and C. W. Zhou, Nano Lett.,
2006, 6, 1880.
72 Y. B. Zhang, Y. W. Tan, H. L. Stormer and P. Kim, Nature, 2005, 438, 201.
73 R. R. Nair, P. Blake, A. N. Grigorenko, K. S. Novoselov, T. J. Booth, T. Stauber, N. M. R. Peres
and A. K. Geim, Science, 2008, 320, 1308.
74 A. K. Geim and K. S. Novoselov, Nat. Meter., 2007, 6, 183.
75 P. Blake, P. D. Brimicombe, R. R. Nair, T. J. Booth, D. Jiang, F. Schedin, L. A. Ponomarenko, S.
V. Morozov, H. F. Gleeson, E. W. Hill, et al., Nano Lett., 2008, 8, 1704.
76 Y. W. Tan, Y. Zhang, K. Bolotin, Y. Zhao, S. Adam, E. H. Hwang, S. Das Sarma, H. L. Stormer
and P. Kim, Phys. Rev. Lett., 2007, 99, 4.
77 T. J. Echtermeyer, M. C. Lemme, J. Bolten, M. Baus, M. Ramsteiner and H. Kurz, European
Physical Journal-Special Topics, 2007, 148, 19.
78 I. Forbeaux, J. M. Themlin and J. M. Debever, Phys. Rev. B, 1998, 58, 16396.
159
79 K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V.
Grigorieva and A. A. Firsov, Science, 2004, 306, 666.
80 L. M. Viculis, J. J. Mack and R. B. Kaner, Science, 2003, 299, 1361.
81 S. Gilje, S. Han, M. Wang, K. L. Wang and R. B. Kaner, Nano Lett., 2007, 7, 3394.
82 G. Eda, Y. Y. Lin, S. Miller, C. W. Chen, W. F. Su and M. Chhowalla, Appl. Phys. Lett., 2008, 92.
83 X. Wang, L. J. Zhi and K. Mullen, Nano Lett., 2008, 8, 323.
84 J. B. Wu, H. A. Becerril, Z. N. Bao, Z. F. Liu, Y. S. Chen and P. Peumans, Appl. Phys. Lett., 2008,
92, 263302.
85 V. C. Tung, L. M. Chen, M. J. Allen, J. K. Wassei, K. Nelson, R. B. Kaner and Y. Yang, Nano
Lett., 2009, 9, 1949.
86 Q. K. Yu, J. Lian, S. Siriponglert, H. Li, Y. P. Chen and S. S. Pei, Appl. Phys. Lett., 2008, 93,
113103.
87 A. Reina, X. T. Jia, J. Ho, D. Nezich, H. B. Son, V. Bulovic, M. S. Dresselhaus and J. Kong, Nano
Lett., 2009, 9, 3087.
88 K. S. Kim, Y. Zhao, H. Jang, S. Y. Lee, J. M. Kim, J. H. Ahn, P. Kim, J. Y. Choi and B. H. Hong,
Nature, 2009, 457, 706.
89 L. G. De Arco, Y. Zhang, A. Kumar and C. W. Zhou, Ieee Transactions on Nanotechnology,
2009, 8, 135.
90 X. S. Li, Y. W. Zhu, W. W. Cai, M. Borysiak, B. Y. Han, D. Chen, R. D. Piner, L. Colombo and
R. S. Ruoff, Nano Lett., 2009, 9, 4359.
91 B. Hayes, Am. Sci., 2005, 93, 104.
92 T. C. Banwell and A. Jayakumar, Electron. Lett, 2000, 36, 291.
93 A. Jain and A. Kapoor, Sol. Energy Mater. Sol. Cells, 2004, 81, 269.
94 A. Ortiz-Conde, F. J. G. Sanchez and J. Muci, Solid-State Electron., 2000, 44, 1861.

160
Bibliography
M. Adachi, Y. Murata, I. Okada and S. Yoshikawa, J. Electrochem. Soc., 2003, 150, G488.

D. M. Alloway and N. R. Armstrong, Appl. Phys. A, 2009, 95, 209.

A. Andersson, N. Johansson, P. Broms, N. Yu, D. Lupo and W. R. Salaneck, Adv. Mater., 1998, 10, 859.

J. W. Arbogast, A. P. Darmanyan, C. S. Foote, Y. Rubin, F. N. Diederich, M. M. Alvarez, S. J. Anz and R.
L. Whetten, J. Phys. Chem., 1991, 95, 11.

A. S. Arico, P. Bruce, B. Scrosati, J. M. Tarascon and W. Van Schalkwijk, Nat. Meter., 2005, 4, 366.

N. R. Armstrong, W. N. Wang, D. M. Alloway, D. Placencia, E. Ratcliff and M. Brumbach, Macromol.
Rapid Commun., 2009, 30, 717.

U. Bach, D. Lupo, P. Comte, J. E. Moser, F. Weissortel, J. Salbeck, H. Spreitzer and M. Gratzel, Nature,
1998, 395, 583.

S. H. Baek, C. U. Kim, D. C. Choo, T. W. Kim, J. H. Park, J. H. Seo and Y. K. Kim, Mol. Cryst. Liq. Cryst.
Sci., 2009, 498, 258.

R. F. Bailey-Salzman, B. P. Rand and S. R. Forrest, Appl. Phys. Lett., 2007, 91, 013508.

M. Baldo, M. Deutsch, P. Burrows, H. Gossenberger, M. Gerstenberg, V. Ban and S. Forrest, Adv. Mater.,
1998, 10, 1505.

M. A. Baldo, D. F. O'Brien, Y. You, A. Shoustikov, S. Sibley, M. E. Thompson and S. R. Forrest, Nature,
1998, 395, 151.

A. K. Bansal, W. Holzer, A. Penzkofer and T. Tsuboi, Chem. Phys., 2006, 330, 118.

T. C. Banwell and A. Jayakumar, Electron. Lett, 2000, 36, 291.

Z. Bao, A. Dodabalapur and A. J. Lovinger, Appl. Phys. Lett., 1996, 69, 4108.

Z. Bao, A. J. Lovinger and A. Dodabalapur, Appl. Phys. Lett., 1996, 69, 3066.

P. F. Barbara, T. J. Meyer and M. A. Ratner, J. Phys. Chem., 1996, 100, 13148.

R. H. Baughman, A. A. Zakhidov and W. A. de Heer, Science, 2002, 297, 787.

E. Becquerel, C.R. Acad. Sci., 1839, 9, 561.

P. Blake, P. D. Brimicombe, R. R. Nair, T. J. Booth, D. Jiang, F. Schedin, L. A. Ponomarenko, S. V.
Morozov, H. F. Gleeson, E. W. Hill, et al., Nano Lett., 2008, 8, 1704.

D. F. Blossey, Phys. Rev. B, 1971, 3, 1382.

D. F. Blossey, Phys. Rev. B, 1970, 2, 3976.

161
M. Boehme and C. Charton, Surface & Coatings Technology, 2005, 200, 932.

R. Bonnett, Chem. Soc. Rev., 1995, 24, 19.

C. Borek, K. Hanson, P. I. Djurovich, M. E. Thompson, K. Aznavour, R. Bau, Y. R. Sun, S. R. Forrest, J.
Brooks, L. Michalski, et al., Angew. Chem. Int. Ed., 2007, 46, 1109.

C. J. Brabec, Sol. Energy Mater. Sol. Cells, 2004, 83, 273.

C. J. Brabec, A. Cravino, D. Meissner, N. S. Sariciftci, T. Fromherz, M. T. Rispens, L. Sanchez and J. C.
Hummelen, Adv. Funct. Mater., 2001, 11, 374.

C. J. Brabec, A. Cravino, D. Meissner, N. S. Sariciftci, M. T. Rispens, L. Sanchez, J. C. Hummelen and T.
Fromherz, Thin Solid Films, 2002, 403, 368.

J. L. Bredas, J. E. Norton, J. Cornil and V. Coropceanu, Acc. Chem. Res., 2009, 42, 1691.

J. Bredenbeck, A. Ghosh, H. K. Nienhuys and M. Bonn, Acc. Chem. Res., 2009, 42, 1332.

J. Bredenbeck, A. Ghosh, M. Smits and M. Bonn, J. Am. Chem. Soc., 2008, 130, 2152.

T. Brixner, J. Stenger, H. M. Vaswani, M. Cho, R. E. Blankenship and G. R. Fleming, Nature, 2005, 434,
625.

M. Brumbach, D. Placencia and N. R. Armstrong, J. Phys. Chem. C, 2008, 112, 3142.

C. Burda, X. B. Chen, R. Narayanan and M. A. El-Sayed, Chem. Rev., 2005, 105, 1025.

C. Burgdorff, T. Kircher and H. G. Lohmannsroben, Spectrochim. Acta, Part A, 1988, 44, 1137.

C. Burgdorff and H. G. Lohmannsroben, J. Lumin., 1994, 59, 201.

M. Burgelman, J. Verschraegen, S. Degrave and P. Nollet, 2005.

S. E. Burns, N. Pfeffer, J. Gruner, M. Remmers, T. Javoreck, D. Neher and R. H. Friend, Adv. Mater.,
1997, 9, 395.

I. H. Campbell, J. D. Kress, R. L. Martin, D. L. Smith, N. N. Barashkov and J. P. Ferraris, Appl. Phys. Lett.,
1997, 71, 3528.

H. C. Casey, D. D. Sell and K. W. Wecht, J. Appl. Phys., 1975, 46, 250.

V. Chauhan, R. Hatton, P. Sullivan, T. Jones, S. W. Cho, L. Piper, A. deMasi and K. Smith, J. Mater.
Chem., 2010, 20, 1173.

W. Chen, S. Chen, Y. L. Huang, H. Huang, D. C. Qi, X. Y. Gao, J. Ma and A. T. S. Wee, J. Appl. Phys.,
2009, 106, 064910.

W. Chen, H. Huang, S. Chen, Y. L. Huang, X. Y. Gao and A. T. S. Wee, Chem. Mater., 2008, 20, 7017.

D. Cheyns, B. P. Rand and P. Heremans, Appl. Phys. Lett., 2010, 97, 3.

C. W. Chu, Y. Shao, V. Shrotriya and Y. Yang, Appl. Phys. Lett., 2005, 86, 243506.
162

A. W. Copeland, O. D. Black and A. B. Garrett, Chem. Rev., 1942, 31, 177.

V. Coropceanu, J. Cornil, D. A. da Silva, Y. Olivier, R. Silbey and J. L. Bredas, Chem. Rev., 2007, 107,
926.

G. A. Crosby, K. W. Hipps and W. H. Elfring, J. Am. Chem. Soc., 1974, 96, 629.

J. Cuiffi, T. Benanti, W. J. Nam and S. Fonash, Appl. Phys. Lett., 2010, 96, 143307.

L. G. De Arco, Y. Zhang, A. Kumar and C. W. Zhou, Ieee Transactions on Nanotechnology, 2009, 8, 135.

L. G. De Arco, Y. Zhang, C. W. Schlenker, K. Ryu, M. E. Thompson and C. W. Zhou, Acs Nano, 2010, 4,
2865.

G. de la Torre, C. G. Claessens and T. Torres, Chem. Commun., 2007, 2000.

G. de la Torre, P. Vazquez, F. Agullo-Lopez and T. Torres, J. Mater. Chem., 1998, 8, 1671.

S. H. Demtsu and J. R. Sites, Thin Solid Films, 2006, 510, 320.

C. D. Dimitrakopoulos and P. R. L. Malenfant, Adv. Mater., 2002, 14, 99.

H. J. Ding and Y. L. Gao, Appl. Phys. A, 2009, 95, 89.

I. K. Ding, N. Tetreault, J. Brillet, B. E. Hardin, E. H. Smith, S. J. Rosenthal, F. Sauvage, M. Gratzel and
M. D. McGehee, Adv. Funct. Mater., 2009, 19, 2431.

P. I. Djurovich, E. I. Mayo, S. R. Forrest and M. E. Thompson, Org. Electron., 2009, 10, 515.

D. R. Eaton, J. Am. Chem. Soc., 1965, 87, 3097.

T. J. Echtermeyer, M. C. Lemme, J. Bolten, M. Baus, M. Ramsteiner and H. Kurz, European Physical
Journal-Special Topics, 2007, 148, 19.

G. Eda, Y. Y. Lin, S. Miller, C. W. Chen, W. F. Su and M. Chhowalla, Appl. Phys. Lett., 2008, 92.

K. Emery, in Handbook of Photovoltaic Science and Engineering, ed. H. S. Luque A., Wiley & Sons,
Chichester, UK, 2003.

K. A. Emery and C. R. Osterwald, Sol. Cells, 1986, 17, 253.

A. Endo, Y. Hoshino, K. Hirakata, Y. Takeuchi, K. Shimizu, Y. Furushima, H. Ikeuchi and G. P. Sato,
Bull. Chem. Soc. Jpn., 1989, 62, 709.

A. Endo, M. Kajitani, M. Mukaida, K. Shimizu and G. P. Sato, Inorg. Chim. Acta, 1988, 150, 25.

T. Enokida, Jpn. J. Appl. Phys., Part 2, 1992, 31, L1135.

F. Fabregat-Santiago, J. Bisquert, L. Cevey, P. Chen, M. K. Wang, S. M. Zakeeruddin and M. Gratzel, J.
Am. Chem. Soc., 2009, 131, 558.

163
A. L. Fahrenbruch and R. H. Bube, Fundamentals of Solar Cells: Photovoltaic Solar Energy Conversion,
Academic Press, New York, 1983.

M. Fedurco, Coord. Chem. Rev., 2000, 209, 263.

X. J. Feng, K. Shankar, O. K. Varghese, M. Paulose, T. J. Latempa and C. A. Grimes, Nano Lett., 2008, 8,
3781.

H. Fidder, J. Knoester and D. A. Wiersma, J. Chem. Phys., 1991, 95, 7880.

I. Forbeaux, J. M. Themlin and J. M. Debever, Phys. Rev. B, 1998, 58, 16396.

S. R. Forrest, Nature, 2004, 428, 911.

E. Frackowiak and F. Beguin, Carbon, 2001, 39, 937.

M. A. Franzman, C. W. Schlenker, M. E. Thompson and R. L. Brutchey, J. Am. Chem. Soc., 2010, 132,
4060.

E. Garnett and P. D. Yang, Nano Lett., 2010, 10, 1082.

Geacinto.N, M. Pope and H. Kallmann, J. Chem. Phys., 1966, 45, 2639.

A. K. Geim and K. S. Novoselov, Nat. Meter., 2007, 6, 183.

N. C. Giebink, B. E. Lassiter, G. P. Wiederrecht, M. R. Wasielewski and S. R. Forrest, Phys. Rev. B, in
review.

N. C. Giebink, B. E. Lassiter, G. P. Wiederrecht, M. R. Wasielewski and S. R. Forrest, Phys. Rev. B, in
review.

S. Gilje, S. Han, M. Wang, K. L. Wang and R. B. Kaner, Nano Lett., 2007, 7, 3394.

R. Giridharagopal and D. S. Ginger, J. Phys. Chem. Lett., 2010, 1, 1160.

M. Glatthaar, M. Riede, N. Keegan, K. Sylvester-Hvid, B. Zimmermann, M. Niggemann, A. Hinsch and A.
Gombert, Sol. Energy Mater. Sol. Cells, 2007, 91, 390.

A. Goetzberger, C. Hebling and H. W. Schock, Mater. Sci. Eng., R, 2003, 40, 1.

C. Goh, S. R. Scully and M. D. McGehee, J. Appl. Phys., 2007, 101.

H. Gommans, B. Verreet, B. P. Rand, R. Muller, J. Poortmans, P. Heremans and J. Genoe, Adv. Funct.
Mater., 2008, 18, 3686.

R. B. Gordon, M. Bertram and T. E. Graedel, P. Natl. Acad. Sci. USA, 2006, 103, 1209.

H. Graaf, W. Michaelis, G. Schnurpfeil, N. Jaeger and D. Schlettwein, Org. Electron., 2004, 5, 237.

C. C. Gradinaru, J. T. M. Kennis, E. Papagiannakis, I. H. M. van Stokkum, R. J. Cogdell, G. R. Fleming, R.
A. Niederman and R. van Grondelle, P. Natl. Acad. Sci. USA, 2001, 98, 2364.

G. Grampp, Angew. Chem. Int. Ed., 1993, 32, 691.
164

M. Gratzel, J. Photoch. Photobio. A, 2004, 164, 3.

H. B. Gray, Nature Chemistry, 2009, 1, 7.

H. B. Gray and J. R. Winkler, Annu. Rev. Biochem, 1996, 65, 537.

M. A. Green, Semicond. Sci. Technol., 1993, 8, 1.

M. A. Green, K. Emery, Y. Hishikawa and W. Warta, Progress in Photovoltaics, 2010, 18, 346.

L. E. Greene, M. Law, B. D. Yuhas and P. D. Yang, J. Phys. Chem. C, 2007, 111, 18451.

B. A. Gregg, J. Phys. Chem. B, 2003, 107, 4688.

B. A. Gregg, M. A. Fox and A. J. Bard, J. Phys. Chem., 1990, 94, 1586.

B. A. Gregg, F. Pichot, S. Ferrere and C. L. Fields, J. Phys. Chem. B, 2001, 105, 1422.

B. A. Gregg, J. Sprague and M. W. Peterson, J. Phys. Chem. B, 1997, 101, 5362.

C. Groves, O. G. Reid and D. S. Ginger, Acc. Chem. Res., 2010, 43, 612.

S. Guha, J. Yang, P. Nath and M. Hack, Appl. Phys. Lett., 1986, 49, 218.

Q. Guo, G. M. Ford, H. W. Hillhouse and R. Agrawal, Nano Lett., 2009, 9, 3060.

M. Hack and M. Shur, J. Appl. Phys., 1985, 58, 997.

R. F. Handy and R. L. Lintvedt, Inorg. Chem., 1974, 13, 893.

B. Hayes, Am. Sci., 2005, 93, 104.

B. W. Henderson and T. J. Dougherty, Photochem. Photobiol., 1992, 55, 145.

E. Hennebicq, G. Pourtois, G. D. Scholes, L. M. Herz, D. M. Russell, C. Silva, S. Setayesh, A. C.
Grimsdale, K. Mullen, J. L. Bredas, et al., J. Am. Chem. Soc., 2005, 127, 4744.

P. Heremans, D. Cheyns and B. P. Rand, Acc. Chem. Res., 2009, 42, 1740.

S. G. Hickey, C. Waurisch, B. Rellinghaus and A. Eychmuller, J. Am. Chem. Soc., 2008, 130, 14978.

I. G. Hill, A. Kahn, Z. G. Soos and R. A. Pascal, Chem. Phys. Lett., 2000, 327, 181.

Y. Hirose, A. Kahn, V. Aristov, P. Soukiassian, V. Bulovic and S. R. Forrest, Phys. Rev. B., 1996, 54,
13748.

M. I. Hoffert, K. Caldeira, G. Benford, D. R. Criswell, C. Green, H. Herzog, A. K. Jain, H. S. Kheshgi, K.
S. Lackner, J. S. Lewis, et al., Science, 2002, 298, 981.

M. I. Hoffert, K. Caldeira, A. K. Jain, E. F. Haites, L. D. D. Harvey, S. D. Potter, M. E. Schlesinger, S. H.
Schneider, R. G. Watts, T. M. L. Wigley, et al., Nature, 1998, 395, 881.

165
Z. R. Hong, Z. H. Huang and X. T. Zeng, Chem. Phys. Lett., 2006, 425, 62.

H. Hoppe and N. S. Sariciftci, J. Mater. Res., 2004, 19, 1924.

J. Huang, J. S. Yu, H. Lin and Y. D. Jiang, J. Appl. Phys., 2009, 105.

J. Hwang, A. Wan and A. Kahn, Mater. Sci. Eng., R, 2009, 64, 1.

Irfan, H. J. Ding, Y. L. Gao, D. Y. Kim, J. Subbiah and F. So, Appl. Phys. Lett., 2010, 96, 73304.

H. Ishii, K. Sugiyama, E. Ito and K. Seki, Adv. Mater., 1999, 11, 605.

B. Jackson and R. Silbey, J. Chem. Phys., 1983, 78, 4193.

A. Jain and A. Kapoor, Sol. Energy Mater. Sol. Cells, 2005, 85, 391.

A. Jain and A. Kapoor, Sol. Energy Mater. Sol. Cells, 2004, 81, 269.

A. Jain, S. Sharma and A. Kapoor, Sol. Energy Mater. Sol. Cells, 2006, 90, 25.

K. J. Jiang, K. Manseki, Y. H. Yu, N. Masaki, K. Suzuki, Y. L. Song and S. Yanagida, Adv. Funct. Mater.,
2009, 19, 2481.

M. Jorgensen, K. Norrman and F. C. Krebs, Sol. Energy Mater. Sol. Cells, 2008, 92, 686.

A. Kahn, N. Koch and W. Y. Gao, J. Polym. Sci., Part B: Polym. Phys., 2003, 41, 2529.

H. Kallmann and M. Pope, J. Chem. Phys., 1959, 30, 585.

P. V. Kamat, J. Phys. Chem. C, 2007, 111, 2834.

K. Kanai, K. Akaike, K. Koyasu, K. Sakai, T. Nishi, Y. Kamizuru, Y. Ouchi and K. Seki, Appl. Phys. A,
2009, 95, 309.

K. C. Kao, Dielectric Phenomena in Solids, Elsevier Academic Press, London, 2004.

D. Kearns and M. Calvin, J. Chem. Phys., 1958, 29, 950.

S. Kera, H. Yamane, H. Honda, H. Fukagawa, K. K. Okudaira and N. Ueno, Surf. Sci., 2004, 566, 571.

M. Khalil, N. Demirdoven and A. Tokmakoff, J. Phys. Chem. A, 2003, 107, 5258.

K. S. Kim, Y. Zhao, H. Jang, S. Y. Lee, J. M. Kim, J. H. Ahn, P. Kim, J. Y. Choi and B. H. Hong, Nature,
2009, 457, 706.

Y. Kinoshita, T. Hasobe and H. Murata, Appl. Phys. Lett., 2007, 91, 083518.

Y. Kinoshita, R. Takenaka and H. Murata, Appl. Phys. Lett., 2008, 92, 243309.

B. Kippelen and J. L. Bredas, Energ. Environ. Sci., 2009, 2, 251.

T. J. Klofta, J. Danziger, P. Lee, J. Pankow, K. W. Nebesny and N. R. Armstrong, J. Phys. Chem., 1987,
91, 5646.
166

A. Kongkanand, K. Tvrdy, K. Takechi, M. Kuno and P. V. Kamat, J. Am. Chem. Soc., 2008, 130, 4007.

F. B. Kooistra, J. Knol, F. Kastenberg, L. M. Popescu, W. J. H. Verhees, J. M. Kroon and J. C. Hummelen,
Org. Lett., 2007, 9, 551.

M. V. Kovalenko, W. Heiss, E. V. Shevchenko, J. S. Lee, H. Schwinghammer, A. P. Alivisatos and D. V.
Talapin, J. Am. Chem. Soc., 2007, 129, 11354.

F. C. Krebs, Sol. Energy Mater. Sol. Cells, 2009, 93, 394.

F. C. Krebs, S. A. Gevorgyan, B. Gholamkhass, S. Holdcroft, C. Schlenker, M. E. Thompson, B. C.
Thompson, D. Olson, D. S. Ginley, S. E. Shaheen, et al., Sol. Energy Mater. Sol. Cells, 2009, 93, 1968.

F. C. Krebs, T. Tromholt and M. Jorgensen, Nanoscale, 2010, 2, 873.

A. Kumar, A. R. Madaria and C. W. Zhou, J. Phys. Chem. C, 2010, 114, 7787.

M. Lee, O. K. Song, J. C. Seo, D. Kim, Y. D. Suh, S. M. Jin and S. K. Kim, Chem. Phys. Lett., 1992, 196,
325.

I. Lefebvre, M. A. Szymanski, J. Olivier-Fourcade and J. C. Jumas, Phys. Rev. B, 1998, 58, 1896.

M. Lenes, G. Wetzelaer, F. B. Kooistra, S. C. Veenstra, J. C. Hummelen and P. W. M. Blom, Adv. Mater.,
2008, 20, 2116.

N. S. Lewis and D. G. Nocera, P. Natl. Acad. Sci. USA, 2006, 103, 15729.

B. Li, L. D. Wang, B. N. Kang, P. Wang and Y. Qiu, J. Photoch. Photobio. A., 2005, 172, 135.

B. Li, L. D. Wang, B. N. Kang, P. Wang and Y. Qiu, Sol. Energy Mater. Sol. Cells, 2006, 90, 549.

C. Li, M. Liu, N. G. Pschirer, M. Baumgarten and K. Müllen, Chem. Rev., 2010, ASAP.

N. Li, B. E. Lassiter, R. R. Lunt, G. Wei and S. R. Forrest, Appl. Phys. Lett., 2009, 94, 023307.

X. S. Li, Y. W. Zhu, W. W. Cai, M. Borysiak, B. Y. Han, D. Chen, R. D. Piner, L. Colombo and R. S.
Ruoff, Nano Lett., 2009, 9, 4359.

Y. Y. Liang, Z. Xu, J. B. Xia, S. T. Tsai, Y. Wu, G. Li, C. Ray and L. P. Yu, Adv. Mater., 2010, 22, E135.

P. B. Littlewood, J. Phys. C: Solid State, 1980, 13, 4855.

B. Liu and E. S. Aydil, J. Am. Chem. Soc., 2009, 131, 3985.

Z. T. Liu, M. F. Lo, H. B. Wang, T. W. Ng, V. A. L. Roy, C. S. Lee and S. T. Lee, Appl. Phys. Lett., 2009,
95, 093307.

W. A. Luhman and R. J. Holmes, Appl. Phys. Lett., 2009, 94, 153304.

R. R. Lunt, J. B. Benziger and S. R. Forrest, Adv. Mater., 2010, 22, 1233.

R. R. Lunt, N. C. Giebink, A. A. Belak, J. B. Benziger and S. R. Forrest, J. Appl. Phys., 2009, 105, 053711.
167

Y. Luo and D. Kessel, Photochem. Photobiol., 1997, 66, 479.

B. W. Ma, C. H. Woo, Y. Miyamoto and J. M. J. Frechet, Chem. Mater., 2009, 21, 1413.

W. L. Ma, C. Y. Yang, X. Gong, K. Lee and A. J. Heeger, Adv. Funct. Mater., 2005, 15, 1617.

G. G. Malliaras, J. R. Salem, P. J. Brock and J. C. Scott, J. Appl. Phys., 1998, 84, 1583.

R. A. Marcus, Rev. Mod. Phys., 1993, 65, 599.

D. E. Markov, E. Amsterdam, P. W. M. Blom, A. B. Sieval and J. C. Hummelen, J. Phys. Chem. A, 2005,
109, 5266.

H. Mathieu, P. Lefebvre and P. Christol, Phys. Rev. B, 1992, 46, 4092.

R. J. Matson, K. A. Emery and R. E. Bird, Sol. Cells, 1984, 11, 105.

B. Mazhari, Sol. Energy Mater. Sol. Cells, 2006, 90, 1021.

I. McCulloch, M. Heeney, C. Bailey, K. Genevicius, I. Macdonald, M. Shkunov, D. Sparrowe, S. Tierney,
R. Wagner, W. M. Zhang, et al., Nat. Meter., 2006, 5, 328.

D. A. McQuarrie and J. D. Simon, Physical Chemistry: A Molecular Approach, University Science Books,
Sausalito, CA, 1997.

R. Memming, Top. Curr. Chem., 1988, 143, 79.

A. Moliton and J. M. Nunzi, Polym. Int., 2006, 55, 583.

G. K. Mor, S. Kim, M. Paulose, O. K. Varghese, K. Shankar, J. Basham and C. A. Grimes, Nano Lett.,
2009, 9, 4250.

G. K. Mor, K. Shankar, M. Paulose, O. K. Varghese and C. A. Grimes, Nano Lett., 2006, 6, 215.

C. B. Murray, C. R. Kagan and M. G. Bawendi, Annu. Rev. Mater. Sci., 2000, 30, 545.

K. L. Mutolo, E. I. Mayo, B. P. Rand, S. R. Forrest and M. E. Thompson, J. Am. Chem. Soc., 2006, 128,
8108.

R. R. Nair, P. Blake, A. N. Grigorenko, K. S. Novoselov, T. J. Booth, T. Stauber, N. M. R. Peres and A. K.
Geim, Science, 2008, 320, 1308.

D. A. Neamen, Semiconductor Physics and Devices, Third Edition edn., McGraw-Hill, New York, 2003.

J. Nelson, J. J. Kwiatkowski, J. Kirkpatrick and J. M. Frost, Acc. Chem. Res., 2009, 42, 1768.

S. F. Nelson, Y. Y. Lin, D. J. Gundlach and T. N. Jackson, Appl. Phys. Lett., 1998, 72, 1854.

H. Neugebauer, C. J. Brabec, J. C. Hummelen, R. A. J. Janssen and N. S. Sariciftci, Synth. Met., 1999, 102,
1002.

T. Q. Nguyen, I. B. Martini, J. Liu and B. J. Schwartz, J. Phys. Chem. B, 2000, 104, 237.
168

K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A.
A. Firsov, Science, 2004, 306, 666.

B. Oregan and M. Gratzel, Nature, 1991, 353, 737.

A. Ortiz-Conde, F. J. G. Sanchez and J. Muci, Solid-State Electron., 2000, 44, 1861.

I. Paci, J. C. Johnson, X. D. Chen, G. Rana, D. Popovic, D. E. David, A. J. Nozik, M. A. Ratner and J.
Michl, J. Am. Chem. Soc., 2006, 128, 16546.

D. C. Pan, X. L. Wang, Z. H. Zhou, W. Chen, C. L. Xu and Y. F. Lu, Chem. Mater., 2009, 21, 2489.

A. K. Pandey and J. M. Nunzi, Adv. Mater., 2007, 19, 3613.

M. G. Panthani, V. Akhavan, B. Goodfellow, J. P. Schmidtke, L. Dunn, A. Dodabalapur, P. F. Barbara and
B. A. Korgel, J. Am. Chem. Soc., 2008, 130, 16770.

A. D. Pasquier, H. E. Unalan, A. Kanwal, S. Miller and M. Chhowalla, Applied Physics Letters, 2005, 87,
203511.

G. S. Patterson and R. H. Holm, Inorg. Chem., 1972, 11, 2285.

B. Pejova and I. Grozdanov, Thin Solid Films, 2007, 515, 5203.

B. Pejova and A. Tanusevski, J. Phys. Chem. C, 2008, 112, 3525.

X. G. Peng, L. Manna, W. D. Yang, J. Wickham, E. Scher, A. Kadavanich and A. P. Alivisatos, Nature,
2000, 404, 59.

M. D. Perez, C. Borek, P. I. Djurovich, E. I. Mayo, R. R. Lunt, S. R. Forrest and M. E. Thompson, Adv.
Mater., 2009, 21, 1517.

M. D. Perez, C. Borek, S. R. Forrest and M. E. Thompson, J. Am. Chem. Soc., 2009, 131, 9281.

R. L. Petritz, Phys. Rev., 1956, 104, 1508.

L. A. A. Pettersson, L. S. Roman and O. Inganas, J. Appl. Phys., 1999, 86, 487.

P. Peumans, V. Bulovic and S. R. Forrest, Appl. Phys. Lett., 2000, 76, 2650.

P. Peumans and S. R. Forrest, Appl. Phys. Lett., 2001, 79, 126.

P. Peumans, A. Yakimov and S. R. Forrest, J. Appl. Phys., 2003, 93, 3693.

P. Peumans, A. Yakimov and S. R. Forrest, J. Appl. Phys., 2003, 93, 3693.

M. Pfeiffer, K. Leo, X. Zhou, J. S. Huang, M. Hofmann, A. Werner and J. Blochwitz-Nimoth, Org.
Electron., 2003, 4, 89.

F. Pichot and B. A. Gregg, J. Phys. Chem. B, 2000, 104, 6.

169
D. Placencia, W. N. Wang, R. C. Shallcross, K. W. Nebesny, M. Brumbach and N. R. Armstrong, Adv.
Funct. Mater., 2009, 19, 1913.

W. J. Potscavage, A. Sharma and B. Kippelen, Acc. Chem. Res., 2009, 42, 1758.

H. E. Prakasam, K. Shankar, M. Paulose, O. K. Varghese and C. A. Grimes, J. Phys. Chem. C, 2007, 111,
7235.

K. V. Ragnarsdottir, Nature Geoscience, 2008, 1, 720.

B. P. Rand, D. P. Burk and S. R. Forrest, Phys. Rev. B, 2007, 75, 115327.

B. P. Rand, J. Li, J. G. Xue, R. J. Holmes, M. E. Thompson and S. R. Forrest, Adv. Mater., 2005, 17, 2714.

B. P. Rand, S. Schols, D. Cheyns, H. Gommans, C. Girotto, J. Genoe, P. Heremans and J. Poortmans, Org.
Electron., 2009, 10, 1015.

M. Razeghi and A. Rogalski, J. Appl. Phys., 1996, 79, 7433.

A. Reina, X. T. Jia, J. Ho, D. Nezich, H. B. Son, V. Bulovic, M. S. Dresselhaus and J. Kong, Nano Lett.,
2009, 9, 3087.

M. E. Roberts, A. N. Sokolov and Z. N. Bao, J. Mater. Chem., 2009, 19, 3351.

S. T. Roberts, C. W. Schlenker, V. S. Barlier, S. W. Chin, R. McAnally, M. E. Thompson and S. E.
Bradforth, in preparation, 2010.

M. W. Rowell, M. A. Topinka, M. D. McGehee, H. J. Prall, G. Dennler, N. S. Sariciftci, L. B. Hu and G.
Gruner, Applied Physics Letters, 2006, 88, 233506.

S. Saini and B. Bagchi, PCCP, 2010, 12, 7427.

N. Sato, H. Inokuchi and E. A. Silinsh, Chem. Phys., 1987, 115, 269.

N. Sato, K. Seki and H. Inokuchi, J. Chem. Soc., Faraday Trans., 1981, 77, 1621.

M. C. Scharber, D. Wuhlbacher, M. Koppe, P. Denk, C. Waldauf, A. J. Heeger and C. L. Brabec, Adv.
Mater., 2006, 18, 789.

C. W. Schlenker, C. K. Renshaw, S. R. Forrest and M. E. Thompson, in preparation, 2010.

G. D. Scholes, X. J. Jordanides and G. R. Fleming, J. Phys. Chem. B, 2001, 105, 1640.

R. Schueppel, R. Timmreck, N. Allinger, T. Mueller, M. Furno, C. Uhrich, K. Leo and M. Riede, J. Appl.
Phys., 2010, 107, 6.

K. Schulze, C. Uhrich, R. Schuppel, K. Leo, M. Pfeiffer, E. Brier, E. Reinold and P. Bauerle, Adv. Mater.,
2006, 18, 2872.

R. Schwedhelm, L. Kipp, A. Dallmeyer and M. Skibowski, Phys. Rev. B., 1998, 58, 13176.

A. F. Schwerin, J. C. Johnson, M. B. Smith, P. Sreearunothai, D. Popovic, J. Cerny, Z. Havlas, I. Paci, A.
Akdag, M. K. MacLeod, et al., J. Phys. Chem. A, 2010, 114, 1457.
170

J. C. Scott and G. G. Malliaras, Chem. Phys. Lett., 1999, 299, 115.

S. R. Scully and M. D. McGehee, J. Appl. Phys., 2006, 100, 034907.

P. Serp, M. Corrias and P. Kalck, Appl. Catal., A, 2003, 253, 337.

B. Servet, G. Horowitz, S. Ries, O. Lagorsse, P. Alnot, A. Yassar, F. Deloffre, P. Srivastava, R. Hajlaoui,
P. Lang, et al., Chem. Mater., 1994, 6, 1809.

Y. Shao, S. Sista, C. W. Chu, D. Sievers and Y. Yang, Appl. Phys. Lett., 2007, 90, 103501.

Y. Shao and Y. Yang, Adv. Mater., 2005, 17, 2841.

W. Shockley and H. J. Queisser, J. Appl. Phys., 1961, 32, 510.

V. Shrotriya, G. Li, Y. Yao, T. Moriarty, K. Emery and Y. Yang, Adv. Funct. Mater., 2006, 16, 2016.

E. A. Silinsh and V. Čápek, Organic Molecular Crystals, American Institute of Physics, New York, 1994.

H. J. Snaith, A. J. Moule, C. Klein, K. Meerholz, R. H. Friend and M. Gratzel, Nano Lett., 2007, 7, 3372.

H. J. Snaith and L. Schmidt-Mende, Adv. Mater., 2007, 19, 3187.

H. Spanggaard and F. C. Krebs, Sol. Energy Mater. Sol. Cells, 2004, 83, 125.

S. Stachura, V. J. Desiderio and J. Allison, Journal of Forensic Sciences, 2007, 52, 595.

R. Steim, F. R. Kogler and C. J. Brabec, J. Mater. Chem., 2010, 20, 2499.

T. Stubinger and W. Brutting, J. Appl. Phys., 2001, 90, 3632.

Y. Sun, C. Borek, K. Hanson, P. I. Djurovich, M. E. Thompson, J. Brooks, J. J. Brown and S. R. Forrest,
Appl. Phys. Lett., 2007, 90, 213503.

S. M. Sze, Physics of Semiconductor Devices, Wiley New York, 1981

S. M. Sze, Physics of Semiconductor Devices, 2 edn., Wiley-Interscience, New York, 1981.

Y. W. Tan, Y. Zhang, K. Bolotin, Y. Zhao, S. Adam, E. H. Hwang, S. Das Sarma, H. L. Stormer and P.
Kim, Phys. Rev. Lett., 2007, 99, 4.

C. W. Tang, Appl. Phys. Lett., 1986, 48, 183.

J. Tang, S. Hinds, S. O. Kelley and E. H. Sargent, Chem. Mater., 2008, 20, 6906.

B. C. Thompson and J. M. J. Frechet, Angew. Chem. Int. Ed., 2008, 47, 58.

V. K. Thorsmolle, R. D. Averitt, J. Demsar, D. L. Smith, S. Tretiak, R. L. Martin, X. Chi, B. K. Crone, A.
P. Ramirez and A. J. Taylor, Phys. Rev. Lett., 2009, 102, 017401.

V. K. Thorsmolle, R. D. Averitt, J. Demsar, D. L. Smith, S. Tretiak, R. L. Martin, X. Chi, B. K. Crone, A.
P. Ramirez and A. J. Taylor, Phys. Rev. Lett., 2009, 102.
171

R. Tokarz-Sobieraj, K. Hermann, M. Witko, A. Blume, G. Mestl and R. Schlogl, Surf. Sci., 2001, 489, 107.

K. Tsutsumi, H. Yoshida and N. Sato, Chem. Phys. Lett., 2002, 361, 367.

V. C. Tung, L. M. Chen, M. J. Allen, J. K. Wassei, K. Nelson, R. B. Kaner and Y. Yang, Nano Lett., 2009,
9, 1949.

N. J. Turro, Modern Molecular Photochemistry, University Science Books, Sausalito, CA, 1991.

J. van de Lagemaat, T. M. Barnes, G. Rumbles, S. E. Shaheen, T. J. Coutts, C. Weeks, I. Levitsky, J.
Peltola and P. Glatkowski, Applied Physics Letters, 2006, 88, 233503.

J. van de Lagemaat, N. G. Park and A. J. Frank, J. Phys. Chem. B, 2000, 104, 2044.

O. K. Varghese, M. Paulose and C. A. Grimes, Nature Nanotechnol., 2009, 4, 592.

B. Verreet, S. Schols, D. Cheyns, B. P. Rand, H. Gommans, T. Aernouts, P. Heremans and J. Genoe, J.
Mater. Chem., 2009, 19, 5295.

L. M. Viculis, J. J. Mack and R. B. Kaner, Science, 2003, 299, 1361.

M. Vogel, S. Doka, C. Breyer, M. C. Lux-Steiner and K. Fostiropoulosa, Appl. Phys. Lett., 2006, 89.

C. Wadia, A. P. Alivisatos and D. M. Kammen, Environmental Science & Technology, 2009, 43, 2072.

L. Q. Wang, G. Z. Liang and G. C. Dang, Spectrochim. Acta, Part A, 2005, 61, 1021.

S. Y. Wang, E. I. Mayo, M. D. Perez, L. Griffe, G. D. Wei, P. I. Djurovich, S. R. Forrest and M. E.
Thompson, Appl. Phys. Lett., 2009, 94, 233304

X. Wang, L. J. Zhi and K. Mullen, Nano Lett., 2008, 8, 323.

Y. Wang and N. Herron, J. Phys. Chem., 1991, 95, 525.

M. R. Wasielewski, Chem. Rev., 1992, 92, 435.

S. S. Williams, M. J. Hampton, V. Gowrishankar, I. K. Ding, J. L. Templeton, E. T. Samulski, J. M.
DeSimone and M. D. McGehee, Chem. Mater., 2008, 20, 5229.

C. Wu, P. I. Djurovich and M. E. Thompson, Adv. Funct. Mater., 2009, 19, 3157.

J. B. Wu, H. A. Becerril, Z. N. Bao, Z. F. Liu, Y. S. Chen and P. Peumans, Appl. Phys. Lett., 2008, 92,
263302.

D. Wynands, M. Levichkova, M. Riede, M. Pfeiffer, P. Baeuerle, R. Rentenberger, P. Denner and K. Leo,
J. Appl. Phys., 2010, 107, 6.

R. G. Xie, M. Rutherford and X. G. Peng, J. Am. Chem. Soc., 2009, 131, 5691.

C. K. Xu, P. H. Shin, L. L. Cao, J. M. Wu and D. Gao, Chem. Mater., 2010, 22, 143.

Y. Xu, N. Al-Salim, C. W. Bumby and R. D. Tilley, J. Am. Chem. Soc., 2009, 131, 15990.
172

J. G. Xue, B. P. Rand, S. Uchida and S. R. Forrest, Adv. Mater., 2005, 17, 66.

J. G. Xue, S. Uchida, B. P. Rand and S. R. Forrest, Appl. Phys. Lett., 2004, 84, 3013.

J. G. Xue, S. Uchida, B. P. Rand and S. R. Forrest, Appl. Phys. Lett., 2004, 85, 5757.

H. Yamane, H. Honda, H. Fukagawa, M. Ohyama, Y. Hinuma, S. Kera, K. K. Okudaira and N. Ueno, J.
Electron. Spectrosc. Relat. Phenom., 2004, 137, 223.

H. Yamane, Y. Yabuuchi, H. Fukagawa, S. Kera, K. K. Okudaira and N. Ueno, J. Appl. Phys., 2006, 99, 5.

Y. P. Yi, V. Coropceanu and J. L. Bredas, J. Am. Chem. Soc., 2009, 131, 15777.

S. Yoo, B. Domercq and B. Kippelen, J. Appl. Phys., 2005, 97, 103706.

S. Yoo, B. Domercq, S. R. Marder, N. R. Armstrong and B. Kippelen, Proc. SPIE-Int. Soc. Opt. Eng.,
2004, 5520, 110.

S. Yoo, W. J. Potscavage, B. Domercq, S. H. Han, T. D. Li, S. C. Jones, R. Szoszkiewicz, D. Levi, E.
Riedo, S. R. Marder, et al., Solid-State Electron., 2007, 51, 1367.

Q. K. Yu, J. Lian, S. Siriponglert, H. Li, Y. P. Chen and S. S. Pei, Appl. Phys. Lett., 2008, 93, 113103.

J. H. Yum, B. E. Hardin, S. J. Moon, E. Baranoff, F. Nuesch, M. D. McGehee, M. Gratzel and M. K.
Nazeeruddin, Angew. Chem. Int. Ed., 2009, 48, 9277.

D. H. Zhang, K. Ryu, X. L. Liu, E. Polikarpov, J. Ly, M. E. Tompson and C. W. Zhou, Nano Lett., 2006, 6,
1880.

Y. B. Zhang, Y. W. Tan, H. L. Stormer and P. Kim, Nature, 2005, 438, 201.

K. Zhu, N. R. Neale, A. Miedaner and A. J. Frank, Nano Lett., 2007, 7, 69.

R. Zhu, C. Y. Jiang, B. Liu and S. Ramakrishna, Adv. Mater., 2009, 21, 994.

Abstract (if available)

Abstract Presently, collective understanding of the involved processes and requisite components that may lead to maturation of organic photovoltaic devices from bench top concept to disruptive solar energy conversion technology have not affected efficiencies within the balance of systems threshold. Primarily, poor broadband spectral photon capture and exciton diffusion, deleterious charge carrier recombination, and poor charge carrier collection, appear to be major efficiency limiting factors in these potentially cost competitive solar cells. This dissertation highlights emerging descriptions for such processes and the continued development of novel materials, device architectures, and process techniques to redress the resulting losses. Lamellar and composite multi-donor systems with complementary properties are examined as a means to robust spectral coverage, enhanced exciton diffusion, and simultaneous suppression of photovoltage losses. Molecular aspects of charge collection in reciprocal carrier architectures are examined. Device architectures incorporating nanostructured materials and composites are probed. The overarching conclusions drawn from the results presented in this work underscore the molecular nature of OPV device operation, contrasted with that of convention covalent crystalline semiconductor devices.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Organic photovoltaics: from materials development to device application

PDF

Molecular aspects of photoconversion processes in organic solar cells

PDF

First steps of solar energy conversion; primary charge generation processes in condensed phase coupled chromophores

PDF

Energy management in organic photovoltaics

PDF

Improving the efficiency and stability of organic solar cells through ternary strategies

PDF

Molecular and morphological effects on the operational parameters of organic solar cells

PDF

Singlet fission in disordered acene films: from photophysics to organic photovoltaics

PDF

Molecular and polymeric donors for bulk heterojunction organic solar cells

PDF

Singlet fission in covalently-linked tetracenes

PDF

Excited state dynamics of organic photovoltaics probed using ultrafast spectroscopy

PDF

Properties and applications of dipyrrin-based molecules in organic photovoltaics

PDF

Theoretical, experimental, device fabrication, and degradation studies of materials for optoelectronic devices

PDF

Porphyrin based near infrared‐absorbing materials for organic photovoltaics

PDF

Dopants for organic light-emitting devices

PDF

Efficient ternary blend bulk heterojunction solar cells with tunable open-circuit voltage

PDF

Non-radiative energy transfer for photovoltaic solar energy conversion: lead sulfide quantum dots on silicon nanopillars

PDF

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

PDF

Nanostructure electronic band modeling for solar cell and lighting applications

PDF

Development of a family of semi-random multichromophoric polymers for application in organic solar cells

PDF

Squaraines and their applications to organic photovoltaics

Asset Metadata

Creator Schlenker, Cody Williams (author)

Core Title Organic solar cells: molecular electronic processes and device development

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Chemistry

Publication Date 09/14/2012

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

Tag flexible electronics,OAI-PMH Harvest,organic electronics,organic photovoltaic,solar cells,solar energy

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Thompson, Mark E. (committee chair), Bradforth, Stephen E. (committee member), Dapkus, P. Daniel (committee member)

Creator Email cschlenk@gmail.com,cschlenk@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m3442

Unique identifier UC1312207

Identifier etd-Schlenker-3972 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-391080 (legacy record id),usctheses-m3442 (legacy record id)

Legacy Identifier etd-Schlenker-3972.pdf

Dmrecord 391080

Document Type Dissertation

Rights Schlenker, Cody Williams

Type texts

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

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu