TRANSPARENT MULTIJUNCTION
ORGANIC PHOTOVOLTAICS
By
Margaret Young

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Materials Science and Engineering – Doctor of Philosophy
2017

ABSTRACT
TRANSPARENT MULTIJUNCTION ORGANIC PHOTOVOLTAICS
By
Margaret Young
The widespread adoption of conventional solar cells based on inorganic semiconductors has been
gaining traction in recent years but is still often hindered by high cost and lack of aesthetic appeal.
Transparent organic semiconductor-based solar cells that selectively absorb in the UV and the NIR enable
integration into building windows, automobiles, and consumer electronics in ways that traditional solar
cells cannot. Moreover, integration onto existing infrastructure reduces the racking and installation cost. In
this work, we investigate routes to improve the efficiency of transparent solar cells by utilizing
multijunction architectures. A transfer-matrix optical interference model is developed as a framework to
optimize the full device stack considering the angle-dependent PV performance that is critical for matching
subcell photocurrents in series tandem solar cells. In addition, a new method of fine tuning energy levels of
low-bandgap small molecules with infrared selective absorption was demonstrated using a series of organic
heptamethine salts. By exchanging the counterion from a small, hard anion to a fluorinated weakly
coordinating anion, the frontier energy of the salt is shown to shift without affecting the bandgap, thus
enabling simultaneous optimization of photocurrent generation of photovoltage. We further utilize this
tunability to develop heptamethine molecules with absorption as deep as 1600 nm, the deepest infrared
photoresponse demonstrated to date with organic small molecules ideal for multijunction integration.
Ultimately, transparent solar cells are an exciting new paradigm for solar deployment enabled by organic
and excitonic semiconductors that offer a pathway to integrate solar onto virtually any surface without
impacting the view.

ACKNOWLEDGMENTS

I want to thank my adviser Dr. Lunt for always believing in me both personally and professionally. I
will never forget your enthusiasm for life. I also want to thank my committee members, Dr. Ilsoon Lee, Dr.
Zhang, and Dr. Morelli for their feedback on my comprehensive exam and thesis and guidance throughout
my PhD program.
I want to thank my coworkers in the MOELab. Chris, thank you for saving my butt all the time. You’re
the best teammate in the world. Paddy, my desk buddy, I’ll never forget your love of pop songs and slang.
Pei, thank you for always being calm and optimistic. Lili, your drive and enthusiasm is inspiring. Dianyi,
thank you for sharing your vast chemistry and cleanroom experience with us. Chenchen, aka Cinderella,
your storytelling and sense of humor is unrivaled. Matt, thank you for being the voice of reason. Alex, thank
you for the pizza. Mark, Adam, Audrey and Zach, thank you for your hard work in this group, I hope you
have bright futures ahead of you in research and otherwise. I want to thank my former labmates John, Juan,
Dhanashree, Yimu, Jorge, Yunhua, Tyler, Natalia, Brian, and Nathan. Thank you all for the memories and
support through the years. Though I’ve gotten some questionable nicknames (Peggerson, King Peggy, King
of the Radishes), I’ll miss everyone dearly.
I want to thank our collaborators: Prof. Sophia Lunt and Deanna Broadwater in Biochemistry; Lijun
Chen and Tony Schilmiller at the Mass-spec facility; Carol and Stanley Flegler at the Center of Advanced
Microscopy; Per Askeland at the CMSC facility. I would never have made it to where I am without your
help and guidance. Especially Per with those hallway light switches.
I want to thank my professors in the MTH, ECE and MSE courses I took at MSU. I want to thank the
organizers of all the talks I went to at MSU. I always learned something new, and the talks helped me
develop breadth as a scientist. These included: CORE-CM, Science at the Edge, and the CHEMS
department seminars. Thank you. I also want to thank the session organizers at outside conferences (MRS,
APS, and AVS) that gave me the opportunity to present, as well as the Graduate School for travel funds to

iii

go to professional conferences. Meeting the people who wrote papers I read gave a human context to
research and provided a sense of scientific community.
I’m grateful for having had the opportunity to teach (TA) for three semesters while in grad school.
Thank you to the EGR 100 students in FS12 and Mr. Hinds. Mindstorm LEGO robots and marshmallows
will always remind me of you. Thank you to the students of the XRD classes in SS14 and SS15 and Dr.
Lunt. Those lab reports were something else. Being a TA gave me a glimpse of the hard work that goes into
teaching, and it made me appreciate my past teachers much more.
I want to thank my friends on and off-campus. I have so many fond memories with the CHEMS GSA
and football tailgates / social hours / game nights (Chris, Markus, Amrita, Aseel, Gloria, Harsha, Mollie,
Kevin, Trey, Ajith), singing karaoke at Buddies (Carolyn, Amanda, Ben), playing Heroes of the Storm
(Mario, Gautam), hanging out with HKSA (Jason), studying the Bible with ACIV (Esther, Stephanie,
Cecelia, Jeff), and playing mahjong / eating hotpot (Xinting, Zhongyu, Chenchen, Zeyang, Yuelin). I want
to especially thank my roommate of four years, Xinting, for her continual support, as well as my labmate
and fellow ex-band geek Chris, without either of whom I would’ve never made it through grad school.
Thank you, Mom and Dad, for only being an hour and a half away in Troy and letting me visit you
whenever I wanted to. Thank you, Dad, for driving to pick me up at MSU even when I could drive myself
home. Thank you, Mom, for inspiring me to try my best. Thank you to my twin brother Vinson for keeping
it real from Georgia Tech. Thank you to my cousin Suzi in LA for always believing in me. I love you all.

iv

TABLE OF CONTENTS

LIST OF TABLES .................................................................................................................................... vii
LIST OF FIGURES ................................................................................................................................. viii
KEY TO ABBREVIATIONS.................................................................................................................... xi
KEY TO SYMBOLS.............................................................................................................................. xviii
Chapter 1 – Introduction to Organic Photovoltaics................................................................................. 1
1.1 Organic semiconductors ...................................................................................................................... 2
1.1.1 Semiconductor basics................................................................................................................... 2
1.1.2 Organic molecular bonding.......................................................................................................... 3
1.1.3 Electronic states and transitions ................................................................................................... 5
1.1.4 Energy transport in organic semiconductors .............................................................................. 11
1.2 Organic photovoltaics: working principles ....................................................................................... 12
1.3 OPV background ............................................................................................................................... 14
1.3.1. Early OPVs: 1980-2000. ........................................................................................................... 15
1.3.2. Polymer OPVs. ......................................................................................................................... 15
1.3.3. Small molecule OPVs: thermal evaporation. ............................................................................ 17
1.3.4. Solution processed small molecule OPVs................................................................................. 17
1.3.5. Tandem OPVs ........................................................................................................................... 18
Chapter 2 – Introduction to Transparent Photovoltaics ....................................................................... 22
2.1 TPV motivation................................................................................................................................. 22
2.2 TPV figures of merit ......................................................................................................................... 23
2.3 TPV background and state of the art ................................................................................................. 26
2.3.1 Spatially segmented PVs ............................................................................................................ 26
2.3.2 Semi-transparent single junction thin film PVs ......................................................................... 28
2.3.3 Luminescent solar concentrators and scattering concentrators .................................................. 30
2.3.4 Visibly transparent PVs ............................................................................................................. 31
2.3.5 Transparent luminescent solar concentrators ............................................................................. 34
2.4 TPV transparent electrodes ............................................................................................................... 34
2.4.1 Transparent conductive oxides................................................................................................... 35
2.4.2 Ultra-thin metals ........................................................................................................................ 37
2.4.3 Polymers and conductive nanoparticles ..................................................................................... 39
2.5 TPV efficiency limits ........................................................................................................................ 40
Chapter 3 – Experimental Techniques.................................................................................................... 42
3.1 Thin film optical measurements: UV-Vis and VASE ....................................................................... 42
3.2 Heptamethine counterion exchange .................................................................................................. 44
3.3 Purity assessment via high-resolution mass spectrometry ................................................................ 45
3.4 Solar cell device fabrication.............................................................................................................. 47
3.5 Solar cell testing (JV-EQE)............................................................................................................... 52
3.5.1 Current-voltage (JV) .................................................................................................................. 52
3.5.2 External quantum efficiency (EQE) ........................................................................................... 52
Chapter 4 – Angle Dependence Modeling............................................................................................... 55
v

4.1 Motivation for angle dependent PV performance modeling ............................................................. 55
4.2 Optical interference modeling ........................................................................................................... 57
4.3 Modifications for angular incidence ................................................................................................. 59
4.4 Verification of model (internal consistency and experimental) ........................................................ 62
4.5 Application for real-world solar irradiance data ............................................................................... 71
Chapter 5 – NIR-Absorbing Organic Heptamethine Salts for TPV .................................................... 74
5.1 Polymethine PV background ............................................................................................................ 74
5.2 Counterion exchange effect on open-circuit voltage ........................................................................ 75
5.3 Deep NIR (1600 nm) absorption ....................................................................................................... 78
5.4 TPV demonstrations with organic salts ............................................................................................ 86
Chapter 6 – Tandem TPV Optimization................................................................................................. 89
6.1 Tandem TPV background ................................................................................................................. 89
6.2 Simulations and contour mapping..................................................................................................... 89
6.3 Experimental results.......................................................................................................................... 91
Chapter 7 – Conclusions and Outlook .................................................................................................... 98
7.1 TPV efficiency: increase NIR EQE, multijunction TPV, better TE design ...................................... 98
7.2 TPV applications............................................................................................................................... 99
7.3 TPV scalability................................................................................................................................ 100
7.4 TPV lifetime.................................................................................................................................... 101
7.5 Future applications: organic salts for cancer theranostics............................................................... 103
7.6 Final summary ................................................................................................................................ 106
APPENDIX .............................................................................................................................................. 108
BIBLIOGRAPHY ................................................................................................................................... 111

vi

LIST OF TABLES

Table 2.1. Selected semi-transparent and transparent PV efficiencies. ...................................................... 27
Table 4.1. Layer thicknesses (Å) to maximize Jsc and Jsc* for 0°, 70°, or all angles of incidence .............. 67
Table 5.1. Device parameters and molecular properties for each salt......................................................... 80
Table 6.1. Selected tandem semi-transparent OPV performance metrics ................................................... 90

vii

LIST OF FIGURES

Figure 1.1. Hybridizations of carbon: sp, sp2, and sp3. ................................................................................. 4
Figure 1.2. Illustrations of σ and π bonds. .................................................................................................... 4
Figure 1.3. Molecular structures of archetypal organic semiconductors used in organic solar cells. ........... 5
Figure 1.4. Formation of molecular bonding and antibonding orbitals in sp2 hybridized carbon................. 6
Figure 1.5. Energy landscape for ground state S0 and the first excited state S1 as a function of nuclear
coordinates for a molecule. ........................................................................................................................... 7
Figure 1.6. Singlets and triplets in organic molecules. ................................................................................. 8
Figure 1.7. Absorption spectra for organic and inorganic semiconductors. ................................................. 9
Figure 1.8. Frontier energy levels for a series of acenes. ............................................................................ 10
Figure 1.9. FĂśrster and Dexter energy-transfer mechanisms. ..................................................................... 11
Figure 1.10. Four steps to photocurrent generation in molecular organic solar cells. ................................ 12
Figure 1.11. JV curve for solar cell with PV performance metrics shown on it. ........................................ 13
Figure 1.12. Non-fullerene acceptor with 11.2% PCE................................................................................ 16
Figure 1.13. Polymer triple junction tandem solar cell. .............................................................................. 21
Figure 2.1. Transparent photovoltaic applications and the solar spectrum. ................................................ 22
Figure 2.2. Survey of different TPV technologies as a function of AVT. .................................................. 26
Figure 2.3. Examples of segmented solar cells. .......................................................................................... 28
Figure 2.4. Examples of thin-film, semi-transparent solar cells. ................................................................ 29
Figure 2.5. Scattering solar concentrators. .................................................................................................. 30
Figure 2.6. UV- and NIR-selective transparent solar cells. ........................................................................ 31
Figure 2.7. Low bandgap donors. ............................................................................................................... 31
Figure 2.8. High performance transparent solar cell module. ..................................................................... 33
Figure 2.9. Transparent luminescent solar concentrators............................................................................ 34
Figure 2.10. Thin silver and ITO comparison. ............................................................................................ 36
Figure 2.11. Efficiency limits of TPVs. ...................................................................................................... 40
viii

Figure 3.1. Ellipsometry fitting for thin and thick films on silicon. ........................................................... 43
Figure 3.2. Counterion exchange synthesis. ............................................................................................... 44
Figure 3.3. High resolution mass spectra for Cy1-TPFB and Cy2-TPFB. ................................................. 46
Figure 3.4. Device architecture and layer patterns...................................................................................... 48
Figure 3.5. Evaporator diagrams. ................................................................................................................ 48
Figure 3.6. Custom-built system assembly for two evaporation chambers and a nitrogen glovebox
(Angstrom Engineering). ............................................................................................................................ 49
Figure 3.7. Diagram of spin-coating process. ............................................................................................. 51
Figure 3.8. Photographs of JV testing setup. .............................................................................................. 52
Figure 3.9. EQE for tandem PV. ................................................................................................................. 54
Figure 4.1. The position of the sun and the average yearly flux for PVs mounted in different configurations.
.................................................................................................................................................................... 55
Figure 4.2. Conventional and inverted configurations for TPV.................................................................. 56
Figure 4.3. Schematic showing xyz coordinates for transfer matrix equations. ......................................... 58
Figure 4.4. Substrate corrections for optical interference simulation. ........................................................ 61
Figure 4.5. Consistency check A + R + T = 1............................................................................................. 63
Figure 4.6. Schematic showing how solar flux changes with incident angle.............................................. 64
Figure 4.7. Predicted and experimental JSC with respect to incident angle. ................................................ 65
Figure 4.8. Example Jsc contour plot where ClAlPc and C60 thicknesses are varied to maximize 0° Jsc for
ITO thickness = 1000 Å in the conventional configuration. ....................................................................... 66
Figure 4.9. Simulated normalized Jsc under 1 sun illumination as a function of angle of incidence for two
optimizations. .............................................................................................................................................. 68
Figure 4.10. Simulated absorption, reflection, transmission and IQE of 735 nm light as a function of incident
angle in three optimizations. ....................................................................................................................... 68
Figure 4.11. Simulated electric field profile for conventional and optically inverted ClAlPc-C60 TPVs. .. 70
Figure 4.12. Simulated electric field profile for normal and oblique incidence on ClAlPc-C60 TPVs. ...... 70
Figure 4.13. Effect of improved angular dependence on yearly PV power output. .................................... 72
Figure 5.1. Molecular structure of Cy cation and anions used in Ref. [31]. ............................................... 76
Figure 5.2. Absorption and JV for Cy salts................................................................................................. 77
ix

Figure 5.3. Cy-PF6 and Cy-TPFB alloying. ................................................................................................ 77
Figure 5.4. Cy1 and Cy2 salt structures, absorption, and mass spectra. ..................................................... 79
Figure 5.5. Device architecture, JV and EQE for Cy1 and Cy2. ................................................................ 79
Figure 5.6 Proposed energy level schematic for Cy1, Cy2, and C60. .......................................................... 81
Figure 5.7. Interface gap change in Cy1 salt and band bending in Cy2-TPFB. .......................................... 81
Figure 5.8. VOC and NIR EQE as a function of thickness for Cy1 and Cy2. .............................................. 82
Figure 5.9. Specific detectivity D* spectra for each deep NIR salt (Cy1 and Cy2).................................... 85
Figure 5.10. Contour map of predicted JSC and AVT for TPV with Cy-TPFB. .......................................... 86
Figure 5.11. Experimental optimization of Ag/Alq3 electrode for TPV with Cy-TPFB. ............................ 87
Figure 5.12. TPV demonstration with Cy-TPFB: JV, EQE, and transmission. .......................................... 87
Figure 6.1. Examples of managing electric field profiles in TPV............................................................... 90
Figure 6.2. ClAlPc and SnPc tandem device. ............................................................................................. 91
Figure 6.3. ClAlPc and Cy-TPFB tandem devices. .................................................................................... 93
Figure 6.4. Contour maps for transparent Cy-TPFB/C60 – ClAlPc/C60 tandem devices showing (a) AVT, (b)
CRI, and (c) JSC as a function of the middle MoO3 and the top ITO thicknesses. ...................................... 95
Figure 6.5. Contour maps for transparent Cy-TPFB/C60 – ClAlPc/C60 tandem devices showing (a) AVT, (b)
CRI, and (c) JSC as a function of the ClAlPc and the Cy-TPFB layer thicknesses. .................................... 95
Figure 6.6. Dilute donor devices with ClAlPc, DBP, and TAPC. .............................................................. 96
Figure 6.7. ClAlPc BHJ and dilute C60 tandem devices.............................................................................. 96
Figure 7.1. Examples of OPV lifetime. ..................................................................................................... 101
Figure 7.2. Optical window of absorbance for biological tissue............................................................... 103
Figure 7.3. Types of photodynamic therapy. ............................................................................................ 104
Figure 7.4. Optoelectronic tunability of Cy salts for therapy and for imaging. ........................................ 105
Figure 7.5. A549 cell viability with different concentrations of organic salts dissolved in DMSO. ........ 105
Figure A.1. Frontispiece for Advanced Optical Materials Volume 4, Issue 7 (July 2016). ...................... 110

x

KEY TO ABBREVIATIONS

A

acceptor

Abs.

absorption

AFM

atomic force microscopy

AgNW

silver nanowire

Alq

short for “Alq3”

Alq3

tris(8-hydroxyquinolinato)aluminum

AM1.5G

“air mass 1.5 global”, also known as the solar spectrum on the surface of the Earth

AO

atomic orbital

a-Si

amorphous silicon solar cell

AZO

aluminum-doped zinc oxide

BBAR

broadband anti-reflection

BCP

bathocuproine

BDT

benzodithiophene

BF-DPB

N4,N4'-Bis(9,9-dimethyl-9H-fluoren-2-yl)-N4,N4'-diphenylbiphenyl-4,4'-diamine

BHJ

bulk heterojunction

BIPV

building integrated photovoltaics

BODIPY

boron-dipyrromethene

Bphen

bathophenanthroline

C12

alkyl group with twelve carbons

CIE

International Commission on Illumination

CIGS

cadmium indium gallium selenide

ClAlPc

chloroaluminum phthalocyanine

CoCB

cobalticarborane

Conv

conventional
xi

CuPc

copper phthalocyanine

CY

1-(6-(2,5-dioxopyrrolidin-1-yloxy)-6-oxohexyl)-3,3-dimethyl2-((E)-2-((E)-3-((E)-2(1,3,3-trimethylindolin-2-ylidene)ethylidene)cyclohex-1-enyl)vinyl)-3H-indolium
chloride

Cy

2-[2-[2-chloro-3-[2-(1,3-dihydro-3,3-dimethyl-1ethyl-2H-benz[e]indol-2ylidene)ethylidene]-1-cylohexen-1-yl]-ethenyl]-3,3dimethyl-1-ethyl-1H-benz[e]indolium

Cy1

(1-Butyl-2-(2-3-2-(1-butyl-1H-benzo[cd] indol-2-ylidene)-ethylidene-2-diphenylaminocyclopent-1-enylvinyl)-benzo[cd]indolium

Cy2

(1-Butyl-2- (23-2-(1-butyl-1H-benzo[cd]indol-2-ylidene)-ethylidene-2-phenylcyclopent1-enyl-vinyl)-benzo[cd]indolium

Cy7

2-[2-[2-chloro-3-[2-(1-ethyl-1,3-dihydro-3,3-dimethyl-2H-indol-2-ylidene)ethylidene]-1cyclohexen-1-yl]ethenyl]-1ethyl-3,3-dimethyl-3H-indolium iodide

Cy7-P

3-Butyl-2-(2-[3-[2-(3-butyl-1,1-dimethyl-1,3-dihydro-benzo[e]
indol-2-ylidene)-ethylidene]-2-chloro-cyclohex-1-enyl]-vinyl)-1,1dimethyl-1Hbenzo[e]indolium hexafluorophosphate

D

donor

D-A

donor acceptor pair

DBP

dibenzoperiflanthene

DBR

distributed Bragg reflector

DC

direct current

DCM

dichloromethane

DCV

dicyanovinyl

DCV6T

α,ω-bis(dicyanovinyl-sexithiophene)-Bu(1,2,5,6)

DI

deionized

DMF

N,N-dimethylformamide

DMSO

dimethyl sulfoxide

DOS

density of states

DPP

diketopyrrolopyrrole

DSSC

dye-sensitized solar cell

xii

DTDCTB

2-{[7-(5-N,N-ditolylaminothiophen-2-yl)-2,1,3-benzothiadiazol-4-yl]methylene}
malononitrile

DTTz

2-((2-(5-(4-(diphenylamino)- phenyl)thieno[3,2-b]thiophen-2-yl)thiazol-5-yl)methylene)malononitrile

ΔLUMO

energy difference between LUMO of donor and LUMO of acceptor

Ex

excitation

EPR

electron paramagnetic resonance

ESI

electrospray ionization

ETL

electron transport layer

F

fluorescence

F6TCNNQ

hexafluorotetracyanonaphthoquinodimethane

F4-ZnPc

fluorinated zinc phthalocyanine

f-CNT

free-standing multiwall carbon nanotube

FDA

Food and Drug Administration

FDPP

a DPP unit with a furan moiety used in place of thiophene

FPhB

tetrakis(4-fluorophenyl)borate

FRET

FĂśrster resonance energy transfer

GO

graphene oxide

H2Pc

phthalocyanine

Hb

hemoglobin

Hb-O2

oxyhemoglobin

HOMO

highest occupied molecular orbital

HTL

hole transport layer

IC60BA

indene C60 bis adduct

ICG

indocyanine green

ICL

interconnecting layers for tandem solar cells

IoT

Internet of Things
xiii

IR

infrared

isc

intersystem crossing, also called “ISC”

ITIC
ITO

3,9-bis(2-methylene-(3-(1,1-dicyanomethylene)-indanone))-5,5,11,11-tetrakis(4hexylphenyl)-dithieno[2,3-d:2’,3’-d’]-s-indaceno[1,2-b:5,6-b’]dithiophene
indium tin oxide

IZAO

indium zinc aluminum oxide

LC

liquid chromatography

LCD

liquid crystal display

LED

light emitting diode

Low-E

low emissivity

LSC

luminescent solar concentrator

LUMO

lowest unoccupied molecular orbital

MeOH

methanol

MDMO-PPV

poly[2-methoxy-5-(3′,7′-dimethyloctyloxy)-1,4-phenylenevinylene]

Me-PTC

N, N′-dimethyl-3,4,9,10-perylenedicarboximide

MJ

multijunction

MO

molecular orbital

NIR

near-infrared

NREL

National Renewable Energy Laboratory

OFET

organic field effect transistors

OLED

organic light-emitting diode

OptInv

optically inverted

OPV

organic photovoltaics

P

phosphorescence

P3HT

poly,3-hexylthiophene

pBBTDPP2

poly[3,6bis(4’-dodecyl-[2,2′]bithiophenyl-5-yl)-2,5-bis(2-ethylhexyl)-2,5-dihydropyrrolo
[3,4-]pyrrole-1,4-dione]
xiv

PBDB-T

poly[(2,6-(4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)-benzo[1,2-b:4,5-b’]dithiophene))-alt(5,5-(1’,3’-di-2-thienyl-5’,7’-bis(2-ethylhexyl)benzo[1’,2’-c:4’,5’-c’]dithiophene-4,8dione)]

PBDTT-DPP

poly{2,6′-4,8-di(5-ethylhexylthienyl)benzo[1,2-b;3,4-b]dithiophene-alt-5dibutyloctyl-3,6-bis(5-bromothiophen-2-yl)pyrrolo[3,4-c]pyrrole-1,4-dione

PBDTTT-C-T 2,6-Bis(trimethyltin)-4,8-bis(5-(2-ethylhexyl)thiophen-2-yl)benzo[1,2-b :4,5-b']
dithiophene
PbPc

lead phthalocyanine

PC61BM

(PCBM) poly-C61-butyric methyl ester

PC71BM

poly-C71-butyric methyl ester

PCDTBT

poly[N-9′-heptadecanyl-2,7-carbazole-alt-5,5-(4′,7′-di-2-thienyl-2′,1′,3′benzothiadiazole)]

PCPDTFBT

poly[2,6-(4,4-bis(2-ethylhexyl)-4H-cyclopenta[2,1-b;3,4-b’]dithiophene)-alt-4,7-(5fluoro-2,1,3-benzothiadiazole)]

PDT

photodynamic therapy

PEDOT

poly(3,4-ethylene dioxythiophene)

PET

polyethylene terephthalate

PFTBT

poly[2,7(9,9-didecylfluorene)-alt-5,5-(4′,7′-di-2-thienyl-2′,1′,3′benzothiadiazole)]

PHJ

planar heterojunction

PID

proportional, integral and derivative control

PIDT-PhanQ

polyindaceno-dithiophene-alt-quinoxaline

PMHJ

planar mixed heterojunction

PMMA

polymethyl methacrylate

PSEHTT

poly[(4,4′-bis(2-ethylhexyl)dithieno-[3,2-b:2′,3′-d]silole)-2,6-diyl-alt-(2,5-bis(3-(2ethylhexyl)- thiophen-2-yl)thiazole[5,4-d]thiazole)]

PSS

polystyrene sulfonic acid

PTB

series of polymers formed by combination of TT and BDT units

PTB7

poly[[4,8-bis[(2-ethylhexyl)oxy]benzo[1,2-b:4,5-b']dithiophene-2,6-diyl][3-fluoro-2-[(2ethylhexyl)carbonyl]thieno[3,4-b]thiophenediyl]]

PTCBI

3,4,9,10-perylenetetracarboxylic bisbenzimidazole
xv

PTT

photothermal therapy

PV

photovoltaics

PVD

physical vapor deposition

QCM

quartz crystal monitor

QToF

quadropole and time of flight detector, a type of tandem mass spec

Re

recombination

RF

radio frequency

SAM

self-assembled monolayer

SeDPP

DPP unit with a sulfur atom substituted by selenium

SEM

scanning electron microscopy

SJ

single junction

SnNcCl2

tin(IV) 2,3-naphthalocyanine dichloride

SnPc

tin phthalocyanine

SPR

surface plasmon resonance

SQ

Shockley Queisser, or squaraines dyes

STPV

semitransparent photovoltaic

Sub

substrate

SubNc

subnaphthalocyanine

TAPC

4,4′-cyclohexylidenebis[N,N-bis(4-methylphenyl)benzenamine]

TCO

transparent conductive oxide

TCOA

transparent conductive oxide acting as an anode (hole-collecting)

TCOC

transparent conductive oxide acting as a cathode (electron-collecting)

TEM

transmission electron microscopy

TFM

tetrakis[3,5-bis(trifluoro methyl)phenyl]borate

TiOx

TiO2, titanium dioxide

xvi

TLSC

transparent luminescent solar concentrator

TMM

transfer matrix method

TPV

transparent photovoltaics

TPFB

tetrakis(pentafluorophenyl)borate; could also be short for Cy-TPFB

TRIS

Δ-tris(tetrachloro-1,2-benzene diolato) phosphate(V). also called TRISPHAT or ΔTRISPHAT

TRISPHAT

see “TRIS”

TT

thieno[3,4-b]thiophene

UV

ultraviolet

UV-Vis

ultraviolet-visible spectroscopy

VASE

variable angle spectroscopic ellipsometry

VIS

visible light

W2(hpp)4

tetrakis(hexahydropyrimidinopyrimidine)ditungsten(II)

XRD

x-ray diffraction

ZnPc

zinc phthalocyanine

xvii

KEY TO SYMBOLS

Å

Angstrom, equivalent to 0.1 nm or 10-10 m

A

absorption; device area

aB

exciton Bohr radius

AAL

absorption in active layers

Ac

electron affinity

ACL

absorption in contact layers

ATF

absorption through the thin film stack

AVT

average visible transmission

AVT0

average visible transmission with no solar spectrum weighting

c

speed of light

CRI

color rendering index

d

thickness of film

D*

specific detectivity

E

energy, or electric field

EB

exciton binding energy

|EDA|2

electric field intensity at the donor-acceptor interface

Eg

bandgap

EQE

external quantum efficiency

Eref

reference spectrum for mismatch factor, usually solar spectrum

ES

solar simulation lamp (source) spectrum

f

fraction of s-polarized light

FF

fill factor

G(x)

exciton generation rate as a function of position x in optical interference model

h

Planck constant
xviii

ħ

reduced Planck constant

I

current

I0

dark saturation current

Ic

ionization energy

IG

interface gap, energy difference between HOMO of donor and LUMO of acceptor

Ijk

interface matrix between layers j and k

Iph

photocurrent

IQE

internal quantum efficiency

j

layer index in optical interference model

J

current density

JMP

current density at maximum power point

JSC

short circuit current density

JV

current-voltage

k

Boltzmann constant, or the extinction coefficient

L

lightness (part of CRI calculation)

LC0

charge carrier collection length

LD

exciton diffusion length

Lj

layer matrix for layer j

⃗⃗
𝑀

surface normal vector of solar cell module

MF

mismatch factor (spectral mismatch factor)

m

total number of layers in optical interference model

ml

magnetic quantum number

𝑚𝑟∗

reduced effective mass of exciton

ms

magnetic spin quantum number

n

the principal quantum number; the lone pair electrons on oxygen or nitrogen; the ideality
factor; the refractive index; or the exciton population density
xix

𝑛0

refractive index of air

𝑛𝑠

refractive index of substrate

Ăą

complex index of refraction

P(Îť)

photopic response, sensitivity of human eye to different wavelengths of light

PCE

power conversion efficiency

PGen

generated electrical power

Pin

incident optical power

q

elementary charge

Q

power dissipation rate

QA

normalized yearly power output of Device A, an unoptimized transparent solar cell

R

reflection; responsivity (A/W)

RD

differential resistance of a solar cell in the dark at zero bias

rjk

Fresnel reflection coefficient for interface between layer j and layer k

𝑝

𝑟𝑗𝑘

Fresnel reflection coefficient for p-polarized light for interface jk

𝑠
𝑟𝑗𝑘

Fresnel reflection coefficient for s-polarized light for interface jk

rp

reflectivity of p-polarized light

rs

reflectivity of s-polarized light

Rp

parallel (also called shunt) resistance

RS

sheet resistance; reflection at the substrate-air interface; series resistance

RTF

reflection through the thin film stack

rab

distance between molecules

S

the total spin; the solar spectrum photon flux (units of photons nm-3 s-1); the Stokes shift;
or the transfer matrix

𝑆

position vector for the sun

S0

ground state

S1

first excited singlet state
xx

S2

second excited singlet state

SN-1

current spectral noise density

SR

spectral response of reference photodiode

ST

spectral response of test device; thermal (Johnson-Nyquist) noise

T

transmission or temperature, depending on context

t

thickness

T1

first excited triplet state

T50

time for solar cell performance to decrease to 50% of initial value

T80

time for solar cell performance to decrease to 80% of initial value

tD

donor film thickness

tf

final (calculated) thickness, used in calculating tooling factors

TFf

final (calculated) tooling factor

TFi

initial (guessed) tooling factor

Tg

glass transition temperature

ti

initial (guessed) thickness, used in calculating tooling factors

tjk

Fresnel transmission coefficient for interface between layer j and layer k

𝑝

𝑡𝑗𝑘

Fresnel transmission coefficient for p-polarized light for interface jk

𝑠
𝑡𝑗𝑘

Fresnel transmission coefficient for s-polarized light for interface jk

TS

transmission at the substrate-air interface

TTF

transmission through the thin film stack

u

chromaticity coordinate, one of two (u and v)

v

chromaticity coordinate, one of two (u and v)

V

voltage

VMP

voltage at maximum power point

VOC

open circuit voltage

xxi

W

peak width

x

position vector in optical interference model
Greek symbols:

Îą

azimuthal angle of solar cell module; absorption coefficient

Îąj

absorption coefficient of layer j

β

tilt angle of solar cell module

δB

improvement in power generation of Device B (PV optimized for oblique incidence) over
Device A

Δ

phase component of ellipsometric angle

Îľ

permittivity

Îľr

relative permittivity, also called dielectric constant

Îľ0

permittivity of free space

Ρ

power conversion efficiency (see PCE)

ΡA

absorption efficiency

ΡCC

charge collection efficiency

ΡCT

charge transfer efficiency

ΡED

exciton diffusion efficiency

θ

incident angle of solar cell on solar cell module

0

ambient incident angle

θj

(incident) angle of light in layer j

θk

(refraction) angle of light in layer k

Îş

extinction coefficient, imaginary index of refractive index also notated as k

Îť

wavelength

Îťmax

peak absorption wavelength

vn

intramolecular vibrational mode

π

pi bonding orbital, 1 per double bond, 2 per triple bond
xxii

π*

pi antibonding orbital

ρ

complex reflectance

σ

sigma bonding orbital, 1 per single, double or triple bond

σ*

sigma antibonding orbital

τ

exciton lifetime before recombination

φ

zenith angle of the sun

Ψ

amplitude component of ellipsometric angle

ω

azimuthal angle of sun

xxiii

Chapter 1 – Introduction to Organic Photovoltaics
Solar power is an attractive source of renewable energy because of the tremendous energy potential
of sunlight reaching the earth. Traditional solar cells fabricated with inorganic semiconductors are
important technologies for offsetting worldwide fossil fuel consumption. However, their adoption in
various locations is often hindered by their high cost and lack of aesthetic appeal. To address the limitations
of traditional photovoltaics (PV), various thin-film architectures have been developed based on materials
such as amorphous silicon, CdTe, cadmium-indium-gallium-selenide (CIGS), organic semiconductors, and
halide perovskites. Among these technologies, organic and other excitonic materials have the unique ability
to selectively absorb ultraviolet (UV) and near-infrared (NIR) light without absorbing visible light. Because
70% of the solar photon flux is in the near-infrared (Îť > 700 nm), efficient transparent solar cells can be
made using excitonic materials.
In this chapter, we will introduce the working principles of organic photovoltaics, and summarize
the status of transparent organic photovoltaics in the following chapter. Experimental methods used
throughout the research in this thesis will be covered in Chapter 3; these include device fabrication and
testing methods, as well as thin film analysis. Chapters 4-6 will cover three major projects completed for
transparent PV development. The first of these, Chapter 4, discusses optical interference simulations for
thin film PVs, specifically, how to modify the standard treatment for oblique incident angles, which are
important for transparent PVs installed onto buildings. The next chapter, Chapter 5, will discuss the
development of heptamethine salts for NIR OPVs with tunable energy levels and photoresponse out to 1600
nm. These materials expand the catalogue for transparent PVs and enable broader NIR absorption. Chapter
6 will cover the progress made on developing tandem OPVs and TPVs. Tandem, or multijunction,
architectures were explored to improve the TPV power efficiency. The last chapter, Chapter 7, describes
the outlook of transparent OPVs, as well as the cancer theranostic application of the heptamethine salts
developed in Chapter 5.

1

1.1 Organic semiconductors
1.1.1 Semiconductor basics
Metals, semiconductors and insulators are distinguished by different bonding and band structures.
In metals, the conduction band is partially filled with electrons, which gives them high conductivities. In
semiconductors and insulators, the valence band is filled with electrons while the conduction band is
completely empty. The energy difference between the top of the valence band and the bottom of the
conduction band is called the bandgap (Eg) and is lower in semiconductors (0.5 – 3 eV) than in insulators
(>3 eV). In semiconductors, an electron can be excited to the valence band via optical or thermal excitations.
The excited electron is coulombically bound to the hole left behind in a quasi-neutral particle called the
exciton. The binding energy (EB) of the exciton is:
𝐸𝐵 =

ħ2
2𝑚𝑟∗ 𝑎𝐵 2

(1.1)

where ħ is the reduced Planck constant (ħ = h/2π), 𝑚𝑟∗ is the reduced effective mass of the exciton, and aB
is the exciton Bohr radius:
𝑎𝐵 =

ħ2 𝜀
𝑚𝑟∗ 𝑞2

(1.2)

q is the elementary charge, and Îľ is the permittivity (Îľ = Îľr Îľ0). The relative permittivity Îľr (also called the
dielectric constant) is the polarizability of a material and is related to its density. Because the dielectric
constant is much larger for inorganic semiconductors (Îľr = 11 for Si) than in organic semiconductors (Îľr =
2-4), the exciton radii and binding energies are very different. In inorganic semiconductors, the exciton
Bohr radius is large (140 Å for GaAs) and the binding energy is small (4.2 meV for GaAs);[39] these
excitons are called Wannier-Mott excitons. In contrast, organic semiconductors have small Bohr radii (510 Å) and large binding energies (0.5 – 1 eV); these excitons are called Frenkel excitons.
Semiconductors have electrical properties that can be modified by doping, which, for inorganic
semiconductors, is done by adding an impurity atom with different numbers of valence electrons. If the
impurity atom has more valence electrons than the atom it replaces, then it is an electron donor, and the
semiconductor is said to be n-doped, where n stands for negative. In the opposite case, if the impurity atom
2

has fewer valence electrons, then it is called an electron acceptor, and the semiconductor is said to be
positively p-doped with holes, which are electron vacancies. By combining p-type and n-type materials,
electrical devices can be made with biasing (diodes, solar cells) and switching abilities (transistors). In an
inorganic p-n junction, a p-type material is in contact with an n-type material. The excess holes from the ptype material diffuse into the n-type material, leaving behind a negative space charge at the junction.
Similarly, the excess electrons from the n-type material diffuse into the p-type material, leaving behind a
positive space charge. Eventually, the positive and negative space charge lead to the build-up of an electric
field that stops the diffusion of holes and electrons. Doping and p-n junctions are accomplished differently
in organic semiconductors, which will be described later.
1.1.2 Organic molecular bonding
Organic molecules consist of carbon atoms bonded to hydrogen. Carbon has a total of six electrons
(1s2, 2s2, 2p2), two of which are in the inner shell and four of which participate in reactions and are called
the valence electrons. Electrons have different types of atomic orbitals (AO) that correspond to the shape
of their probability distributions. Carbon can have hybridized orbitals depending on the types of
participating bonds. In methane (CH4), the four valence electrons of carbon are in hybridized sp3 orbitals,
four lobes that are 109.5 degrees apart and form a tetrahedron (Figure 1.1). The two electrons in the C-H
bond have a probability distribution that exists between the two atoms: this is called a molecular orbital
(MO). This type of bond is called a σ-bond or a σ-orbital (Figure 1.2). In ethene C2H4, each carbon is
bonded to three atoms, one carbon and two hydrogens. The carbon atoms are double bonded to each other.
In this case, one valence electron is in the 2p orbital and three valence electrons are sp2-hybridized, meaning
that they exist in three coplanar lobes 120° apart. The carbon double bond consists of one sp2 electron from
each carbon and one 2p electron from each carbon. The two sp2 electrons in the bond have a σ orbital, and
the two 2p AOs are lobes that overlap above and below the molecular plane. The overlapping p orbitals are
called a π bond or π orbital. Therefore, a double bond consists of one σ bond (two sp2 electrons) and one π
bond (two 2pz electrons). In carbon triple bonds (for example, ethyne C2H2), the four valence electrons of
carbon consist of two 2p electrons and 2 sp-hybridized electrons, which are two separate lobes 180° apart
3

Figure 1.1. Hybridizations of carbon: sp, sp2, and sp3.

Figure 1.2. Illustrations of σ and π bonds.(a) In ethene (C2H4), both C atoms have p and sp2 hybrid
orbitals. The atoms are far apart, so atomic orbitals are shown. (b) The overlapping sp2-hybrid orbitals make
a σ bond, and the overlapping p-orbitals make a π bond in the C=C double bond in ethene.
from one another in the molecular plane. The two carbons in this triple bond have one σ bond (two sp
electrons) and two π bonds (two 2py and two 2pz electrons). σ-bonds have shorter distances than π-bonds
and maintain the molecular structure. π-bonds come from overlapping pz-orbitals and are responsible for
the interesting optical transitions in molecules. A molecule is “conjugated” when it has alternating single
and double bonds. The position of the single and double bonds can be switched, leading to a resonant
structure, and the π-orbitals (also called p-orbitals) are delocalized along the alternating single and double
bonds. The delocalized p-orbitals allow for intramolecular charge transfer, and the overlap of p-orbitals
across different molecules allows for intermolecular charge transfer via a hopping mechanism. At low

4

Figure 1.3. Molecular structures of archetypal organic semiconductors used in organic solar cells.
CuPc = copper phthalocyanine, P3HT = poly,3-hexylthiophene, PC61BM = poly-C61-butyric methyl ester.
temperatures, hopping occurs less frequently, and intermolecular charge transfer can occur by coherent
charge transfer processes. Therefore, organic semiconductors can be described as a set of conjugated
organic molecules and polymers with bandgaps between 0.3 – 3 eV. Examples of organic semiconductors
used in organic photovoltaics are shown in Figure 1.3.
1.1.3 Electronic states and transitions
The energy levels of an electron in a finite potential well are discretized. Similarly, an electron
orbiting a positively charged nucleus occupies discrete energy levels. Because an electron has half-integral
spin, it is considered a fermion. A photon, in contrast, has integral (zero) spin, making it a boson. By the
Pauli exclusion principle, no two fermions can occupy the same quantum state. The electrons in an atom
occupy a quantum state characterized by four numbers: the principal number n, the orbital number l, the
magnetic quantum number ml, and the magnetic spin quantum number ms. As two atoms are brought closer
together, their valence electrons cannot occupy the same quantum state, so their atomic orbital levels split
into one bonding orbital (with lower energy) and one antibonding orbital (with higher energy). A bond is

5

therefore energetically favored to form, and the lower-energy bonding orbital is occupied with two electrons.
Optical transitions take place for electrons in σ and π orbitals, as well as non-participating electrons (e.g.,
lone pair electrons on oxygen or nitrogen) called n. Typical transitions that can be excited with UV-visible
light include: σ  σ*, n  σ*, π  π*, n  π*, where * indicates an antibonding orbital. The amount of
energy split between the bonding and antibonding orbitals depends on how much the atomic orbitals overlap.
Because the orbitals from each atom overlap significantly in a sigma bond, the energy minimized by the
bond formation is greater, and the sigma bonding and antibonding orbitals are far apart. In contrast, the π
bond consists of overlapping pi orbitals that only weakly overlap, and the energy difference between the π
and π* orbitals is smaller. For organic molecules, the σ  σ* transition is usually seen in UV light
absorption, while π  π* transitions can be activated by visible light. The smallest amount of energy needed
to an excite an electron is the energy from the highest occupied molecular orbital (HOMO), which is the π
orbital in this case, and the lowest unoccupied molecular orbital (LUMO), the π* orbital (Figure 1.4).

Figure 1.4. Formation of molecular bonding and antibonding orbitals in sp2 hybridized carbon.
The highest occupied molecular orbital (HOMO) is the π bonding orbital, and the lowest unoccupied
orbital (LUMO) is the π* antibonding orbital.

6

Figure 1.5. Energy landscape for ground state S0 and the first excited state S1 as a function of nuclear
coordinates for a molecule. Horizontal lines are vibronic substates, and vertical lines correspond to
electrical transitions.
Electronic transitions in organic molecules need to consider vibrational and electronic states
together because the atoms are smaller and their vibrational frequencies are larger.[39] Optical transitions
result in electrons moving between vibronic states, the combination of vibrational and electronic states. The
potential energy landscape for the ground state S0 and the excited state S1 for different nuclear coordinates
(e.g., different interatomic distances for the two-atom case) is shown in Figure 1.5. Higher excited states
(S2, S3, … Sn) are possible for more electrons being excited to unoccupied molecular orbitals. Intramolecular
vibrational modes are marked as vn or v’n. The Frank-Condon principle states that electron transitions
happen faster than the nucleus can respond, resulting in vertical excitation (“Ex”) and recombination (“Re”)
transitions in Figure 1.5. After excitation from v0 to v’1, the electron eventually relaxes to the bottom of S1
(v’0), accompanied by a shift in the nuclear coordinates of the molecule. When the electron relaxes to the
ground state, the energy can be emitted as either a photon (radiative decay) or a phonon (nonradiative decay).

7

Figure 1.6. Singlets and triplets in organic molecules. (a) Electron spin configurations for ground state
S0, first excited singlet state S1, and first excited triplet state T1. Note that electron spins are opposing for
singlet states and parallel in triplet states. (b) Jablonski diagram showing fluorescence and phosphorescence.
Solid and dashed lines indicate radiative and nonradiative decay respectively.
Other types of electronic transitions can occur as well, and these can be summarized in a Jablonski
diagram (Figure 1.6 (b)), which removes the nuclear coordinate axis. A single electron is called a doublet
because the sum of its spin (S = Σms) is ½, and the multiplicity 2S + 1 is equal to 2. A pair of spin-opposite
electrons has a total spin S = 0 and a multiplicity equal to 1, making it a singlet. A pair of spin-parallel
electrons has a total spin of S = 1 and a multiplicity of 3, giving it the name triplet. The spin configurations
of singlets and triplets in organic semiconductors are shown in Figure 1.6(a). Changes in total spin are not
allowed by the conservation of spin and momentum, so transitions with ΔS = ±1 are said to be spin forbidden.
Despite being spin forbidden, spin flips can occur via spin-orbit coupling. Phosphorescent emission, which
is a transition from T1 to S0, is spin forbidden and therefore occurs on a much longer timescale (Îźs to ms)
than fluorescence, which is a transition from S1 to S0 (ns).
Another consequence of the molecular nature is discrete and strong electronic transitions in photon
absorption. As shown in Figure 1.7(a), absorption for organic semiconductors occurs as transitions to
distinct states (S1, S2 … Sn) from the ground state (S0), where the optical bandgap EG is the energy between
S0 and S1. In porphyrin systems, two absorption bands are known as the Q Band (S0  S1) and the Soret
Band (S0  S2). The overlap (or similarity) of the vibrational modes of each orbital sub-state dictates the

8

Figure 1.7. Absorption spectra for organic and inorganic semiconductors. (a) Jablonski diagram for
organic semiconductors highlighting the gap in density of states between top of S1 and bottom of S2. This
leads to two separate bands of absorption, as seen in the bottom left inset. The middle inset shows the
vibronic substates in the S1 and S2 orbitals as a function of rab, the distance between molecules. (b) Jablonski
diagram for an inorganic semiconductor showing absorption of all photons with energy above the bandgap.
strength of transition, leading to variations in the peak absorption intensity, which are highlighted in Figure
1.7(a) as “weak”, “moderate”, and “strong” transitions. In many acene and porphyrin molecules, these
vibronic features can be resolved in absorption and emission spectra but are often broadened into featureless
bands due to additional vibrational coupling and splitting in the solid state. Nonetheless, the gap between
the S1 and S2 levels (and corresponding vibronic substates) can create a discontinuity in the density of states
(DOS) that can be exploited to allow visible light transmission with selective NIR harvesting. This is in
sharp contrast to the continuum in the density of states for traditional inorganic semiconductors (Figure 1.7
(b)), which leads to a continuously decaying absorption profile. The continuum of states for inorganic
semiconductors arises from the periodicity in crystal structure and therefore the nuclear potential energy
distribution (Bloch’s theorem). In either class of materials, the excited state electron rapidly relaxes to S1

9

Figure 1.8. Frontier energy levels for a series of acenes. From left to right: naphthalene (C10H8),
anthracene (C14H10), tetracene (C18H12) and pentacene (C22H14). Ac is electron affinity and Ic is ionization
energy. Figure adapted from Ref. [3] with permission.
via thermalization after a photon is absorbed into an Sn state.
The properties of organic semiconductors can be finely tuned via changes in the molecular structure,
conjugation, and crystal structure. Bandgap engineering can be achieved by adjusting the conjugation length
or by designing push pull molecules. The progression of the bandgap with conjugation is well demonstrated
in the acene series where the addition of conjugated benzene rings starting from naphthalene to pentacene
results in additional orbital energy level splitting that decreases the bandgap (Figure 1.8).[3, 40] Push pull
molecules also rely on energy level splitting to decrease the bandgap and are composed of alternating
electron donating and electron accepting units (D-A). When the D and A units are close, the highest
occupied molecular orbital (HOMO) levels split, leading to a shallower HOMO level than either unit has.
Similarly, the lowest unoccupied molecular orbital (LUMO) levels split and produce a deeper LUMO level.
The bandgap of the D-A unit is thus narrowed by the HOMO-LUMO level splitting.

10

Figure 1.9. FĂśrster and Dexter energy-transfer mechanisms. (a) Schematic of FĂśrster energy transfer
shown for singlets. D = donor, A = acceptor and * = excited state. (b) Schematic of Dexter energy transfer
shown for triplets.
1.1.4 Energy transport in organic semiconductors
This section will cover the intermolecular transfer of excitons, the second important energy carrier
in organic semiconductors. Excitons can move within a material via two main mechanisms: FĂśrster and
Dexter transfer. In FĂśrster transfer, also called FĂśrster resonance energy transfer (FRET), the exciton on
the donor (D) undergoes long-range dipole-dipole coupling with the acceptor (A) molecule, and the energy
non-radiatively transfers from donor to acceptor (Figure 1.9 (a)). This energy transfer can happen between
similar or dissimilar molecules. The rate of FRET depends on 1) the distance between the donor and
acceptor, 2) the amount of spectral overlap between the emission spectrum of the donor and the absorption
spectrum of the acceptor, and 3) the relative orientation of the emission dipole of the donor and the
absorption dipole of the acceptor. For aligned D-A dipoles, FRET can occur between molecules up to 1100 nm apart depending on the strength of the coupling.[39] On the other hand, Dexter transfer is highly
distance sensitive, occurring typically at <1 nm length scales via direct orbital overlap. In Dexter transfer,
two carriers move in concerted exchange: the electron in the donor LUMO moves to the acceptor LUMO,
and an electron moves from the acceptor HOMO to the donor HOMO (Figure 1.9 (b)). Both singlet and
triplet excitons may diffuse in Dexter transfer.

11

Figure 1.10. Four steps to photocurrent generation in molecular organic solar cells. (1) photon
absorption, (2) exciton diffusion, (3) charge transfer, and (4) charge collection. (b) Blue arrows illustrate
the movement of only the electron for clarity – for each electron collected at the anode there is one collected
at the cathode.
1.2 Organic photovoltaics: working principles
A solar cell consists of a p-n junction sandwiched by two electrodes. This junction has a built-in
voltage, and when light is incident on a solar cell, the excess carriers move from one electrode to the other
due to the built-in electric field. Organic p-n junctions are distinct from inorganic p-n junctions because
free carriers are not typically produced directly by light excitation. The photoexcited state that is separated
by the built-in electric field in inorganic p-n junctions remains coulombically bound in organic materials,
requiring an energetic driving force for dissociation greater than the exciton binding energy. In organic
semiconductors, photocurrent production occurs in four distinct steps with different efficiencies: absorption
(ΡA), exciton diffusion (ΡED), charge transfer (ΡCT), and charge collection (ΡCC) (Figure 1.10). At a specific
wavelength, a photon can be absorbed with efficiency ΡA (absorption efficiency), promoting an electron
from the HOMO to the LUMO. The exciton then diffuses to a heterojunction with efficiency ΡED (exciton
diffusion efficiency). The average distance the exciton diffuses in a material before recombining is called
the exciton diffusion length, which can vary from 10 to 20 nm in organic materials.[41] The exciton then
separates into an electron on the acceptor material and a hole on the donor material with efficiency ΡCT
(charge transfer efficiency). The isolated charge carriers are then swept to the electrodes by the internal
electric field or via diffusion with efficiency ΡCC (charge collection efficiency).

12

Figure 1.11. JV curve for solar cell with PV performance metrics shown on it. The inner shaded box
shows the maximum power producible, and the outer box shows the area of JSC*VOC.
In any diode (organic or inorganic), current flows in one direction. When light is incident on the
diode, the current-voltage (I-V) relationship of the solar cell can be described as an exponential curve shifted
by the photocurrent:
𝑞𝑉

𝐼 = 𝐼0 [exp (𝑛𝑘𝑇) − 1] − 𝐼𝑝ℎ

(1.3)

Where I0 is the dark saturation current, q is the elementary charge, V is voltage, n is the ideality factor, k is
the Boltzmann constant, T is temperature, and Iph is the photocurrent. The current is usually divided by the
area of the device to give J, the current density. This makes it easier to compare devices with different areas
and to calculate the incident power, which at 1 sun intensity is 100 mW/mm2. The power conversion
efficiency (PCE or Ρ) of a solar cell can be calculated from the current-voltage (J-V) curve, which is
collected by measuring the current through an illuminated solar cell while sweeping the voltage (Figure
1.11). The PCE is defined as the ratio of generated electrical power (PGen) to the total incident optical power
(Pin) on the active area of the device:



PGen J SCVOC FF

Pin
Pin

(1.4)

where JSC is the short-circuit current density, VOC is the open-circuit voltage, and FF is the ratio of the
maximum power generated PGen to the product of JSC and VOC.

13

Photocurrent can be broken down into its spectral components by measuring the external quantum
efficiency (EQE), which is the ratio of collected electrons to the number of incident photons. Since every
photon produces up to one electron (in the absence of any carrier multiplication) regardless of the initial
photon energy, integrating EQE over the solar photon spectrum S (units of photons/nm3-s) then yields the
JSC:

J SC  q  ( EQE  S )d 

(1.5)

where q is the elementary charge. Equation 1.5 also is useful as a consistency check on the measurement
of the JSC. In OPVs, the EQE is the product of the four efficiencies described above: absorption (ΡA), exciton
diffusion (ΡED), charge transfer (ΡCT), and charge collection (ΡCC) efficiency. Internal quantum efficiency
(IQE) is the fraction of collected charge carriers over the number of absorbed photons and is related to EQE
by: EQE   A  IQE . For organic semiconductors, the exciton diffusion lengths range from 10-20 nm,
which limits the thickness of the active layers to a similar range, and this tradeoff between absorption and
exciton diffusion efficiency is known as the exciton diffusion bottleneck.
1.3 OPV background
Two approaches to making OPVs are to use small molecules or to use polymers. Small molecule
organic semiconductors can be thermally evaporated or solution processed and have the benefit of ease of
purification and good batch-to-batch consistency. Polymer OPVs are typically solution processed but
require additional consideration of the final molecular weight and regioregularity. Thermal evaporation is
a physical vapor deposition (PVD) technique requiring high vacuum. In this work, we utilized both
approaches depending on the suitability for the active layer utilized. At this point it is not clear which
approach (vacuum or solution) will ultimately be cheaper, but nearly all electronic devices are made by
vacuum deposition.
In the following sections, we provide a brief overview of early OPVs polymer OPV development,
thermally evaporated and solution processed small molecule OPVs, and tandem OPVs.

14

1.3.1. Early OPVs: 1980-2000.
The first organic solar cells consisted of a single organic thin film sandwiched between two metal
electrodes with different work functions, but they had very low efficiencies (0.001 – 0.01%).[42, 43] In
1986, Tang demonstrated the first bilayer OPV (also called planar heterojunction) with an evaporated ptype phthalocyanine derivative and an n-type perylene derivative.[44] It had a breakthrough efficiency of
around 1%. Around this time, conjugated polymers were developed and used in a single layer organic solar
cell.[45] The next breakthrough was the use of C60 in the bilayer solar cell, which enabled efficiencies over
2%.[46, 47] The higher electronegativity of C60 and a high electron mobility made it energetically favorable
for photoexcited electrons on the p-type organic thin film to transfer onto C60, thus arriving at the naming
convention of p-type (electron) donors and n-type (electron) acceptors.
1.3.2. Polymer OPVs.
A soluble derivative of C60, phenyl-C61-butyric acid methyl ester (PCBM), was developed in 1995
and used in a polymer bulk heterojunction device with 2.9% efficiency.[48] The bulk heterojunction
structure, which was based off the dye-sensitized solar cell architecture, consisted of interpenetrating
domains of the donor and acceptor to help overcome the exciton diffusion bottleneck. This was a critical
step in improving the efficiency because the exciton diffusion length of the donor polymer poly-3hexylthiophene (P3HT) was ~10 nm, and a ~25 nm thick layer was needed to absorb most of the incoming
light.[49] As a result, many photo-generated electron-hole pairs never reached a dissociating D-A interface
and instead recombined. P3HT:PCBM bulk heterojunctions were developed in the early 2000s, with many
studies focusing on control of morphology. Thermal annealing was shown to significantly improve
efficiencies. Analytical techniques such as atomic force microscopy (AFM) and x-ray diffraction (XRD)
confirmed that with thermal annealing: the PCBM fullerene molecules would diffuse out of P3HT:PCBM
mixed domains, PCBM would self-aggregate, and P3HT polymers would line up with one another, forming
crystallites.[50, 51] Essentially, with thermal annealing, phase segregation and crystallization would
occur.[49] Notably, other donors and acceptors have been developed since P3HT and PCBM. P3HT has a
bandgap of 1.6 eV, but devices with PCBM are limited to VOCs of ~0.6 V. Since then, low bandgap polymers
15

Figure 1.12. Non-fullerene acceptor with 11.2% PCE. (a) Device structure and molecular structure. (b)
Absorption profiles of PBDB-T, PC71BM, and ITIC. (c) Solution absorption of PBDB-T as a function of
temperature. (d) Energy level schematic for PBDB-T/PC71BM and PBDB-T/ITIC devices. Figure
reproduced with permission from Ref. [12].
with higher photocurrents have been developed (PDCTBT, PTB7, etc.) to absorb in the NIR portion of the
solar spectrum. Efforts to improve PCBM led to the development of PC71BM, which has more absorption
in the visible spectrum and absorption out to 700 nm.[52] However, fullerene-based acceptors overall have
been challenging to replace due to their high electron mobilities. The HOMO/LUMO levels are hard to
modify for fullerenes, and precise control of the energy levels is needed to optimize interface gaps (voltage)
and ΔLUMO gaps (photocurrent). To overcome this, a record non-fullerene device was recently
demonstrated with 11.2% PCE (Figure 1.12).[12] The new acceptor ITIC significantly differed from PCBM
in its molecular structure, its primarily NIR absorption (peak at 710 nm, shoulder ~630 nm), and its higher
HOMO and LUMO levels (3.78/5.51 vs. 3.98/5.96 for PCBM). The raising of the LUMO levels contributed
to a higher interface gap and therefore VOC. Polymer-based OPVs have crossed the 10% PCE threshold in
research efficiencies.

16

1.3.3. Small molecule OPVs: thermal evaporation.
The first bilayer OPV used copper phthalocyanine, an evaporated small molecule, as the donor.
Phthalocyanines, merocyanines, squaraines, diketyopyrrolopyrroles, borondipyrromethene, and isoindigo
dyes are examples of the wide variety of small molecules that can be evaporated or spin-coated as donors
for OPVs. Phthalocyanines consist of four isoindole groups linked together by nitrogen atoms, leading to a
planar aromatic structure. A great deal of functionalization can be done either on the perimeter (e.g. adding
sulfonate or carboxylate groups for water solubility) or in the center coordinating bivalent (+2) metal atom.
The size and shape of the metal atom can lead to either planar (CuPc) or bent (PbPc, SnPc, ClAlPc)
molecules. In the case of chloroaluminum phthalocyanine, the out-of-plane Cl atom changes the packing
from cofacial to slip-stack, leading to a bathochromic shift in absorption which can be exploited in nearinfrared absorbing OPVs.[53] Subphthalocyanines, a smaller derivative of phthalocyanine molecules, have
been used in solar cells with up to 3.0% PCE.[54] Merocyanines, another small molecule, have been used
in solar cells with 3.9% PCE.[55] One of the earliest attempts to make push-pull molecules for OPV donors
was in 2006 and introduced dicyanovinyl (DCV) electron withdrawing units to a 1-D oligothiophene.[56]
Putting electron donating and electron withdrawing groups on the same molecule can promote
intramolecular charge transfer and increase absorption.[53] Another push-pull molecule DTDCTB was coevaporated with C70 to form a planar mixed heterojunction device with 7.9% PCE, a record for single
junction evaporated small molecule OPV at the time.[57] More recently, an 8.0% efficient device was
demonstrated using thermally evaporated small molecules in a ternary architecture (C 70, DTTz, and
DTDCTB).[58]
1.3.4. Solution processed small molecule OPVs
Some small molecules decompose during thermal evaporation; for these materials, solution
processing methods such as spin-coating and doctor blading are used to deposit uniform thin films.
Merocyanines were cast from solution in a metal/dye/metal sandwich structure as early as 1978.[59] In dye
sensitized solar cells (DSSCs), a mesoporous layer of TiO2 is soaked in a solution containing small molecule
photosensitizers; the first was demonstrated in 1991 with an efficiency of 7.1%.[60] Subnapthalocyanine
17

(SubNc) was spincoated to form amorphous thin films; the PCE of the devices were 1.5% after annealing
at 120°C for 40 minutes.[61] Squaraines were spincoated with PCBM or use evaporated C60 for the acceptor
and achieved up to 5.7% PCE after thermal annealing at 90°C.[62] Mainly, 1-D oligothiophene-based push
pull molecules dominate among solution processed small molecule donors. A D-A-A planar mixed
heterojunction (PMHJ) with C70 with 6.4% PCE, which was the record in 2009.[63] More recently, >11%
PCE was obtained using a small molecule ternary-blend OPV.[64]
1.3.5. Tandem OPVs
A key approach to limit thermalization losses in OPV is to stack cells together in a tandem (or
multijunction) OPV. There are predominately two types of tandem solar cells, solar cells connected in series
and solar cells connected in parallel. For series tandem solar cells, the layers in both subcells can be
deposited in a monolithic stack because the intermediate layer does not need to be externally contacted.
The majority of tandem OPVs are monolithically configured in series because of the compatibility with
single junction device fabrication processes. The open circuit voltage of a series tandem solar cell is the
sum of the open circuit voltages of its subcells:
𝑉𝑂𝐶,𝑡𝑎𝑛𝑑𝑒𝑚 = 𝑉𝑂𝐶,𝑏𝑜𝑡𝑡𝑜𝑚 + 𝑉𝑂𝐶,𝑡𝑜𝑝

(1.6)

and the photocurrent of the tandem device is limited by the subcell that produces less current (also called
the limiting or current-limiting subcell):
𝐽𝑆𝐶,𝑡𝑎𝑛𝑑𝑒𝑚 = min(𝐽𝑆𝐶,𝑏𝑜𝑡𝑡𝑜𝑚 , 𝐽𝑆𝐶,𝑡𝑜𝑝 )

(1.7)

Because of this condition, the currents produced in both the top and bottom subcells must be
matched to maximize JSC. This can be accomplished by tuning the absorber layer thicknesses in each subcell
or tuning the thicknesses of optical spacer layers, which are transparent and conductive layers within the
structure. Tuning the absorber layer thickness straightforwardly adjusts the number of photogenerated
carriers within a subcell, and tuning the optical spacer layer thicknesses adjusts the intensity of the electric
field profile across the structure, thereby allowing the maximum light absorption to occur near the donor-

18

acceptor interface. Because of the relatively large number of layers, tuning these thicknesses is a task greatly
aided by simulations.
Between the two subcells in a tandem OPV is the intermediate layers, also called interstitial layers
or the recombination zone. These layers are responsible for recombining photo-generated holes and
electrons from the two subcells. As such, intermediate layers consist of hole transporting layer (HTL), a
thin metal (to reset the work function and serve as recombination centers), and an electron transporting
layer (ETL). Extra care must be taken for solution processed tandem OPVs to ensure that solvents used to
spincast intermediate layers do not dissolve the first subcell. Other optimizations for the intermediate layers
involve tuning the work function of the thin metal and improving the carrier mobilities in the ETL and
HTLs by doping or other treatments.
The Shockley Queisser (SQ) limit for a single junction solar cell can be calculated based on
thermodynamic limits. The ideal bandgap for a single junction solar cell is around 1.1-1.4 eV, leading to a
maximum theoretical efficiency of 33% PCE.[65] The open circuit voltage depends on the bandgap Eg, but
high JSC requires absorption of many photons so there is a natural trade-off. A small bandgap leads to
absorption of many photons, but because the electrons rapidly relax to the conduction band edge (in OPV,
the LUMO level) upon photoexcitation, much of the energy of high-energy photons is lost. A large bandgap
allows for more efficient collection of high-energy photons, but photons with less energy than the bandgap
are not absorbed, leading to a loss in JSC. A considerable amount of energy is lost in the thermalization of
electrons excited by high-energy photons to the LUMO level. Using multiple subcells with complementary
absorbers reduces thermalization loss and has been a traditional way to increase efficiencies for inorganic
(Si, GaAs, etc.) as well as organic solar cells. Empirical limits for organic solar cells include a loss in
voltage from the donor-acceptor heterojunction / energy offset required to dissociate excitons. Based on the
trend of best reported PVs with a range of bandgaps, the highest achievable VOC is approximately 80% of
the SQ limit;[66] this heuristic, when combined with 75% EQE and 0.75 FF (the highest values
demonstrated by OPV), a practical limit of 17% PCE was calculated for single junction OPVs.[66] For
current-matched, two-junction OPVs, the empirical efficiency limit was calculated to be 24%, using
19

absorber bandgaps of 0.9 and 1.5 eV.[66] These efficiency limits show why tandem architectures are
attractive for higher efficiency devices. The efficiency limits of transparent solar cells (both single junction
and multijunction) are treated differently and will be described in more detail in Chapter 2.
The first tandem OPV was developed by Hiramoto et al in 1990 using a gold interstitial layer
between two identical small molecule subcells (Me-PTC/H2PC).[67] The voltage of the tandem device was
almost double of the single junction (0.74 V vs. 0.44 V). From that point, the development of tandem OPVs
has incorporated improvements made to single junction devices (by using a planar mixed heterojunction,
for example)[68] while solving challenges inherent to tandem structures, such as managing optical
interference by tuning optical spacer layers[69] and changing the work function of the metal in the
interstitial layer.[70] As noted previously, solvent orthogonality is critical to managing polymer tandem
solar cells. Although out of the scope of this thesis, polymer tandem OPVs have had an interesting
developmental path. In 2005, Kawano et al demonstrated a tandem device with two polymer subcells, both
consisting of bulk heterojunctions of MDMO-PPV with PC61BM, and an intermediate layer of sputtered
ITO and spincast poly(3,4-ethylene dioxythiophene):polystyrene sulfonic acid (PEDOT:PSS).[71]
Polymer subcells were also combined with evaporated small molecule subcells: the first example developed
by Dennler et al in 2006 used P3HT:PCBM and ZnPc:C60 bulk heterojunction and had a PCE of 2.3%.[72]
Then, in 2007, Gilot et al developed the first spincast intermediate layer for tandem polymer OPV using
ZnO nanoparticles dispersed in acetone and pH-neutral PEDOT:PSS.[73] Triple junctions were also
demonstrated in the study.
Recently, many studies of tandem OPVs have achieved efficiencies over 10%. In 2013, Liu et al
demonstrated a solution processed small molecule tandem device with 10.1% PCE.[74] It had two identical
subcells connected with a polyelectrolyte intermediate layer. Chen et al developed a solution processed
triple junction device in 2014 with 11.5% PCE using absorbers with 1.4, 1.58 and 1.9 eV bandgaps.[9] The
intermediate layers consisted of PEDOT:PSS / ZnO with spincast WO3 layer on top of the middle subcell
to align energy levels for hole transport from PTB to PEDOT (Figure 1.13).[9] The current record for
tandem OPVs is held by the German chemical company Heliatek. They set a record efficiency of 13.2% in
20

Figure 1.13. Polymer triple junction tandem solar cell. (a) Device structure, (b) absorber optical
constants, and (c) energy level schematic. Figure reproduced with permission from Ref. [9].
2016 for a triple-junction tandem with in-house designed and thermally evaporated small molecule
absorbers.[75]
In summary, organic semiconductors form the basis of organic electronic devices ranging from
OPVs, organic light emitting diodes (OLEDs), organic field effect transistors (OFETs), and organic
thermoelectrics. The main feature of organic semiconductors is the formation of Frenkel excitons due to
their low dielectric constants. Their non-crystallinity results in a discontinuity in the density of states, giving
rise to distinct absorption bands, which can be tuned to be outside of the visible spectrum. This feature can
be used to make visibly transparent solar cells, which is the topic of the next chapter.

21

Chapter 2 – Introduction to Transparent Photovoltaics
Organic photovoltaics (OPV) have long offered the promise of low cost, flexible solar cells. While
recent advances have led to efficiencies beyond 10%, the large scale commercial viability of OPVs as a
replacement for traditional technologies remains an open question. In this chapter, we describe the emerging
field of transparent photovoltaics (TPV) that uniquely exploit the excitonic nature of organic
semiconductors. Transparent photovoltaics can create entirely new functionality and markets for solar
deployment while complementing traditional opaque PV technologies. Moreover, TPVs provide a new and
key motivation for reinvigorating research efforts into OPVs. In this chapter, we discuss the development
of semi-transparent and transparent solar cells along with their key figures of merit, characterization
strategies, and theoretical efficiency limits.
2.1 TPV motivation
Visibly transparent excitonic solar cells based on organic and excitonic semiconductors that
selectively absorb in the ultraviolet (UV) and the near infrared (NIR) have recently been developed.[14]
These new kinds of solar cells can enable solar PV integration into buildings, windows, greenhouses,
automobiles, and consumer electronics (Figure 2.1(a)) and can turn essentially any surface into a power

Figure 2.1. Transparent photovoltaic applications and the solar spectrum. (a) Example applications
for transparent photovoltaics, clockwise from upper-left: skyscraper windows, greenhouse windows, ereader screens, and automobile windows. (b) AM1.5G solar spectrum in photon flux plotted with the
photopic response, figure adapted with permission from Ref. [2].
22

generating surface. Integrating PVs into building and car structures produces electricity at the point of
consumption, thereby reducing electricity transmission costs. Also, mounting PVs onto existing surfaces
also reduces the racking cost associated with installing PVs. Despite not being able to absorb photons in
the visible region, TPVs can still generate power using a significant portion of the solar spectrum.
Approximately 10% of the photon flux is in the ultraviolet (Îť < 430 nm) and 70% of the photon flux is in
the infrared (Îť > 680 nm), meaning that 80% of the solar spectrum is invisible to the human eye and can be
used to generate electricity (Figure 2.1(b)).
2.2 TPV figures of merit
The approach to TPVs using wavelength selective harvesting around the visible spectrum allows
for the highest possible combination of efficiency and clear transparency. For TPVs, three independent
objective criteria are important in their performance: power conversion efficiency (Ρ or PCE), average
visible transmission (AVT), and color rendering index (CRI). Optimization of TPVs therefore requires the
simultaneous consideration of all these metrics. In the absence of information on the CRI, the product of
PCE*AVT can be utilized as a figure of merit to assess transparent device progress and overall system
efficiencies (power generation and lighting efficiency).
Solar cell transparency is reported as the average visible transmission (AVT) weighted by the
photopic response (sensitivity to different wavelengths of light, P(Îť)) and, in some cases, also by the solar
photon spectrum (AM1.5G, S(Îť)). The standard definition accepted and reported in the window industry
of AVT is described by:
AVT 

 T ( ) P ( ) S (  ) d 
 P ( ) S ( ) d 

(2.1)

where the wavelength range is selected to cover the range of the photopic response. As a reference point,
typical AVT requirements are 50-80% for architectural glass, 55-90% for automotive glass, and 80-90% for
mobile electronics.[2] For integration onto displays where the solar spectrum weighting is less important,
Equation 1.2 can be simplified by substituting S = 1. We assign the notation “AVT0” for AVT with no solar

23

spectrum weighting. However, because the solar spectrum is roughly flat throughout the visible spectrum,
the difference between AVT0 and AVT is usually small enough to be negligible. The solar spectrum and the
photopic response are both plotted for reference in Figure 2.1 (b).
The color rendering index (CRI), typically used in the lighting industry, quantifies the ability of a
light source to faithfully reproduce the color of an object when compared to an illuminating reference light
source (typically a blackbody radiator). The color temperature of the reference blackbody radiator is chosen
so that its chromaticity is closest to that of the light source being tested. The CRI formalism can also be
used to assess the impact of transmission through a surface to reproduce the color quality of the initial
source color temperature (such as the solar spectrum, AM1.5G). A CRI > 90 is considered to be excellent
for an illumination source.[76] Commercial cool-white fluorescent light typically have a CRI of ≥70,[77,
78] and an incandescent bulb has a CRI of 100 due to being a blackbody radiator. To calculate CRI, eight
test-color samples (CIE 1974) are “illuminated” by a reference spectrum. This illumination is characterized
by lightness (L) and chromaticity (u and v), which make up the three-dimensional uniform color space
CIELUV. The illumination of the eight test-color samples by the light transmitted through the solar cell
(S(Îť)T(Îť)) is then calculated. The difference in L, u, and v between illuminating the eight test color samples
with the sun or with the light transmitted through the solar cell is then corrected for chromatic adaption (the
ability of the eye to discern the same color under different illumination conditions) and averaged into the
CRI by the following expression:

CRI 

1 8
 (100  4.6 (L*i )2  (ui* )2  (vi* )2 )
8 i 1

(2.2)

For reference, the CRI for any neutral density absorption profile (flat absorption through the visible) is 100
regardless of the reference illumination spectrum.

The color coordinates are also useful in many

applications where they can indicate the degree of coloring. For example, a* and b* are the chromaticity
indices of the perceptually uniform color space CIELAB system which are related to the CIELUV
coordinates u and v. Positive values of b* (0 to 100) indicate a yellow coloring, and negative values of b*
indicate blue coloring, while positive a* indicates red coloring, and negative a* indicates green coloring.
24

The photon balance is an important consistency check for any solar cell, but it is particularly
important for TPVs. The photon balance is:

A  R T 1

(2.3)

where A is the absorption, R is the reflection, and T is the transmission of the entire device film stack at
each wavelength. While reflection and transmission can be measured by standard spectrometers, absorption
is more difficult to measure directly for thin film layers. To estimate A, it is important to recognize that
there can be absorption in both the active layer (AAL) and in the contact layers (ACL) so that A = AAL+ ACL.
The active layer absorption can be obtained from the definition of the EQE (IQE = EQE / AAL). Therefore,
the photon balance becomes:

EQE
 ACL  R  T  1
IQE

(2.4)

It has been shown that AAL can be measured directly by measuring the EQE under reverse bias where IQE
can approach 100% via field-driven dissociation and carrier collection.[79] Measurements of R and T are
generally straightforward but require careful attention for reference samples: R measurements need
appropriate highly reflective reference mirrors with known reflectivity, and T measurements on thin-film
devices should avoid the use of reference samples (unlike with solution measurements). We note that the
measurement of ACL is notably difficult because variations in the thin-film architecture will change thinfilm optical interference and the amount of contact absorption. One solution is to model the entire device
using transfer matrix modeling described in Section 4.2 to estimate ACL. Nonetheless, because the IQE can
reach 100% for some bulk heterojunction architectures and ultrathin planar devices,[80] the following
balance should still be satisfied: EQE  R  T  1 . This consistency check confirms that neither EQE nor T
of the entire device has been overestimated. We emphasize that this check should be reported for all semitransparent and transparent devices (e.g. EQE, R, and T should be reported for each device) and applied to
Eq. 2.4 like the photocurrent check in integrating the EQE using Eq. 1.5, which is listed as a necessary
criterion for many journal publications including Nature journals.[81]

25

Figure 2.2. Survey of different TPV technologies as a function of AVT. Si and GaAs efficiencies are
included for reference. The green shaded portion is the combination of AVT and PCE that is enabled by
wavelength selective approaches. Figure adapted from Ref. [1] with permission from the authors.
2.3 TPV background and state of the art
Several approaches to making semi-transparent PVs (STPV) have been explored since the first PVs
were developed, including 1) spatially segmented photovoltaics, which involves spacing opaque PVs; 2)
thin film solar cells, which can make semi-transparent PVs with tint; 3) scattering solar concentrators, which
impart optical haze; and 4) luminescent solar concentrators, which can be adapted for both color and high
transparency. In contrast, selectively absorbing organic materials have been used to make visibly
transparent PVs. In general, we categorize PVs with AVT < 50% as “semi-transparent” (neutral, hazy, or
colored) and AVT > 50% as “transparent” (neutral or tinted), analogous to the language used for transparent
conductors. An overview of the performance (PCE) and transparency (AVT) for organic and inorganic TPV
is provided in Figure 2.2. Selected results are tabulated in Table 2.1 for single-junction organic STPV/TPV.
The research developments for each category are summarized below.
2.3.1 Spatially segmented PVs
Semi-transparent photovoltaics have applications in building-integrated photovoltaics (BIPV)
either as facades or colorful architectural installations (e.g. being commercialized by Solaronix, Onyx Solar,
26

Table 2.1. Selected semi-transparent and transparent PV efficiencies.

Device Structure

VOC
(V)

JSC
(mA
cm-2)
0.45 2.5
0.53 7.4

FF

Ρ
(%)

Ρ (%) * Ref.
AVT(frac.)
(%)
0.62 26
0.16
[4]
1.95 10.4*
0.20
[19]
AVT
(%)

ITO/CuPc/PTCBI/BCP/Ag/ITO
0.55
ITO/TiO2/P3HT:PC71BM/PEDOT:PSS/
0.40
Ag grid
ITO/ZnO/SQ:PC61BM/PEDOT:PSS
0.54 2.0
0.35 0.4 52*
(TLSC) CY in PMMA with edge- mounted Si solar 0.50 1.2
0.66 0.4
86
cells
ITO/W2(hpp)4 doped C60/C60/
0.56 3.8 0.595 1.3
28
ZnPc:C60/F6TCNNQ doped BF-DPB/
f-CNT
ITO/TiO2/P3HT:PC61BM/MoO3/ITO
0.53 6.3
0.57 1.9 21.0*
AgNW/TiOx/ZnO/P3HT:PC61BM/
0.58 8.2
0.49 2.3 22.5*
PEDOT:PSS/GO/PEDOT:PSS
ITO/MoO3/Cy-TPFB/C60/BCP/Ag/
0.69 2.5
0.53 0.9
61
MoO3/ITO
ITO/PEDOT:PSS/Cy7-P/C60/Ag/Alq3
0.40 4.7 0.463 0.9 62.4*
ITO/PEDOT:PSS/PCDTBT:PC71BM/
0.85 9.2 0.481 3.9 21.5*
TiOx/Al/ZnO:Al
ITO/MoO3/ClAlPc/C60/BCP/ITO
0.71 4.2
0.46 1.3
65
BBAR/Quartz/ITO/MoO3/ClAlPc/
0.78 4.7
0.46 1.7
56
C60/BCP/ITO/DBR/BBAR
ITO/MoO3/DTDCTB/DTDCTB:C60/
0.779 5.50 0.494 2.11 60.0
C60/Bphen/Ag/MoO3
ITO/PEDOT:PSS/
0.84 7.98 0.63 4.2 31.3*
PIDT-PhanQ:PC71BM/Surfactant/Ag
ITO/MoO3/DTDCTB/DTDCTB:C70/
0.805 7.68 0.524 3.24 46.4
C70/Bphen/Ag/MoO3
ITO/PEDOT/PTB7:PC71BM/BCP/Ag/
0.733 10.0 0.70 5.6
28
(LiF/MoO3)3
ITO/MoO3/Cy7/C60/Alq3/Ag/Alq3
0.63 6.4
0.54 2.2 71.2*
ITO/ZnO/C60-SAM/
0.76 11.9 0.62 5.6 29*
PBDTTT-C-T:PC71BM/MoO3/Ag
ITO/ZnO/PCPDTFBT:PC71BM/
0.74 11.4 0.58 5.0 51.2*
PEDOT:PSS/Ag
ITO/PEDOT:PSS/
0.77 9.30 0.56 4.0 64.3*
PBDTT-DPP:PC61BM/TiO2/AgNW
(Ubiquitous Energy, Inc certified device)
0.68 2.35 0.59 5.1+ 50.9
*
AVT recalculated using transmission curves provided in references
+
measured at 0.185 suns

27

0.21
0.34

[25]
[5]

0.36

[28]

0.40
0.52

[30]
[11]

0.54

[31]

0.56
0.84

[32]
[33]

0.85
0.95

[14]
[14]

1.27

[34]

1.31

[21]

1.50

[34]

1.57

[35]

1.57
1.62

[36]
[37]

2.56

[38]

2.57

[22]

2.60

[1]

Figure 2.3. Examples of segmented solar cells. (a) Schematic of segmented PV. (b) Patterned electrode
for a-Si solar cell, reproduced from Ref. [13]. (c) Photo of a cafÊ façade with segmented PV installed,
reproduced from Ref. [20].
and Polysolar).[82, 83] Spatially segmented PVs are solar cells that achieve some transparency through the
patterning of open spaces in the module (Figure 2.3) and have been investigated for over three decades.
Any PV technology can, in principle, be spatially separated by mounting a spaced array of opaque modules
onto a glass superstrate (e.g. being commercialized by Solar Constructions).[84] For less obstructive views,
PVs can be segmented at the mm and Îźm scale by patterning material in stripes or grids via etching or laser
ablation – the latter approach was used with Si PVs to create devices with 8-10% PCE and 10% AVT.[20,
85] A similar approach of spatial segmentation has been demonstrated for OPVs to create limited
transparency.[19] Although spatial segmentation is a conceptually simple way to add transparency to highly
efficient solar cells, transparency and power conversion efficiency cannot be simultaneously optimized for
the highest AVTs.
2.3.2 Semi-transparent single junction thin film PVs
Like spatially segmented PVs, most PV technologies can also be made thin enough to transmit a
fraction of visible light to make a STPV (Figure 2.4(a)). This approach generally leads to significant tinting
due to non-uniform absorption profiles but has been widely demonstrated with a variety of PV technologies,
including a-Si, CdTe,[86] CIGS,[87] quantum dots,[88] GaAs,[89], and organics. However, these
approaches to semi-transparent PVs are less effective at the highest AVT values since much of the solar flux
simply passes through the active layer. At lower AVT, this approach often this leads to red, brown, or yellow
coloring with traditional semiconductors and a full spectrum of colors with organic semiconductors, which
28

Figure 2.4. Examples of thin-film, semi-transparent solar cells. (a) Schematic of thin-film, semitransparent PV. (b, c) Demonstrations of semi-transparent organic solar cells adapted from Ref. [11] and
[21], respectively.
could be advantageous in applications for colorful architecture. Nonetheless, semi-transparent solar panels
based on crystalline and amorphous silicon have been produced on a commercial scale, and multiple studies
have been conducted on their real (and simulated) power-generation on different buildings throughout the
world.[90, 91]
Several semitransparent OPVs have been made using this approach. The archetypal small molecule
copper phthalocyanine (CuPc)[44] was used with 3,4,9,10-perylenetetracarboxylic bis-benzimidazole
(PTCBI) as an acceptor and indium tin oxide (ITO) as the transparent electrode for a device with PCE of
0.62% and AVT of 26%.[4] Similar results were obtained with a CuPc/C60 heterojunction and a solutionprocessed silver nanowire (AgNW) mesh.[92] Another phthalocyanine, zinc phthalocyanine (ZnPc), was
used as both the donor and as the high-index anti-reflection coating for a device with PCE 0.6% and AVT
of 24%.[93] The archetypal polymer bulk heterojunction composed of poly-3-hexylthiophene (P3HT) and
[6,6]-phenyl-C61-butyric acid methyl ester (PCBM) has been used in many studies of semitransparent
polymer PV.[19, 94-96] For example, Ameri et al used a silver grid with an underlying layer of PEDOT:PSS
to make P3HT:PC70BM devices with 2% PCE and 10% AVT.[19] Because the peak absorption of P3HT
(~550 nm) lies in the middle of the visible region, the visible transparency is reduced in these systems
(Figure 2.4 (b)).[11]
Considerable efforts were made to develop polymers with smaller bandgaps for higher efficiency
semitransparent and opaque devices.[97] The combination of the thieno[3,4-b]thiophene (TT) unit (weak
acceptor strength; strong quinoidal resonance) and the large planar benzodithiophene donor gave rise to the

29

Figure 2.5. Scattering solar concentrators. (a) Schematic of scattering solar concentrator, modified from
Ref. [5]. Incoming light is scattered and wave-guided to the edge-mounted solar cell. (b) Photo of a
scattering concentrator window module, reproduced from Ref. [15].
PTB series, which had a bandgap of 1.6 eV and improved polymer-based solar cell performance from 4 to
6% PCE.[97, 98] Further improvements in side chain engineering, such as adding fluorine to the TT unit
to deepen the HOMO level, led to the development of PTB7, a polymer that has enabled PCEs over 8% in
both conventional and inverted architectures[99] and has been used in STPVs with 5.6% PCE and 28%
AVT.[35, 100]
2.3.3 Luminescent solar concentrators and scattering concentrators
Luminescent solar concentrators and scattering concentrators are two types of solar concentrators
that utilize a wave-guiding plate to focus light at the edges. In a luminescent solar concentrator
photoluminescent chromophores, i.e. compounds or molecules that absorb and reemit light of a longer
wavelength, are embedded in the wave-guiding plate. For scattering concentrators, light scattering particles
are embedded in the plate (Figure 2.5 (a)). In both cases, a high-efficiency solar cell (e.g. Si or GaAs) is
mounted at the edges to convert the concentrated light to electricity. The simple construction means that
this approach can be very scalable, however, efficiencies have been modest (< 7%). Luminescent solar
concentrators have been made using organic compounds, typically heavily colored since their
inception.[101-103] In laboratory studies, scattering concentrators using TiO2 nanoparticles have been
made with 1.0% PCE and 85% transmittance with a device area of 2.5 cm2,[15, 104, 105] and a range of
scattering particles have been demonstrated.[106] However, scattering concentrators face a fundamental

30

Figure 2.6. UV- and NIR-selective transparent solar cells. (a) Schematic of selectively UV and NIRabsorbing excitonic transparent solar cells from Ref. [1]. (b, c) Demonstrations of transparent organic solar
cells adapted from Ref. [14]and [22], respectively.

Figure 2.7. Low bandgap donors. (a) Plot of solar spectrum overlaid with the normalized film absorption
spectra of different low bandgap donors spanning a large near-infrared spectral range with NIR selective
harvesting. (b) Molecular structures of the different low bandgap donors in (a).
tradeoff between transparency and PCE, and often exhibit larger losses at large device sizes due to
outcoupling.[107]
2.3.4 Visibly transparent PVs
The development of visibly transparent solar cells based on selective harvesting of the invisible
parts of the solar spectrum has emerged within the last 5-10 years. Wavelength selective TPVs offer the
highest potential in combined AVT and PCE and requires three key components: 1) active layers with
selective absorption potential in the UV and NIR, 2) highly conductive electrodes that are highly transparent
throughout the solar spectrum, and 3) an overall device configuration that enables selective absorption in
the active layers rather than in electrode layers (Figure 2.6 (a)). The highest reported efficiencies currently

31

range from 5% (single junction) to 8% (multijunction) (Table 6.1). The development of narrow bandgap
small molecule (ClAlPc,[14] Cy,[32] etc.) and polymer (PBDTT-DPP,[22] PCPDTFBT,[38] etc.) donors
with selective NIR harvesting was an important step to enabling high efficiency transparent solar cells.
Shown in Figure 2.7 are the absorption spectra for free-standing films and molecular structures of key
organic molecules and polymers with selective NIR harvesting.
The first TPV based on selective harvesting in the infrared was demonstrated using
chloroaluminum phthalocyanine (ClAlPc).[14] C60 was used as the acceptor and sputtered ITO was used as
the top and bottom electrodes to produce a PCE of 1.3% with an AVT of 65% (Figure 2.6 (b)) that was
increased to a PCE of 1.7% and an AVT of 56% using a NIR selective transparent mirror. Because ITO was
used as both the anode and cathode, metal oxide layers were utilized to obtain a high work function
anisotropy and high photovoltages (>0.7 V).[14] TPVs have also been made using push-pull molecules
such as DTDCTB, which consists of a thiophene electron-donating group and a benzothiadiazole (BT)
electron-withdrawing group.[34] DTDCTB was thermally evaporated in a PMHJ structure, which has neat
donor and acceptor layers on either side of the BHJ to aid in carrier collection. MoO3 was used as the optical
incoupling layer on top of a thin layer of Ag. The device has a high product of efficiency and AVT, with
60% AVT and 2.1% PCE for the device using C60.
Several solution processable small molecule donors have also been used in TPVs including
heptamethine salts,[31, 36] squaraines (SQs),[25, 108] and boron-dipyrromethene (BODIPY)
derivatives.[109] Heptamethine salts in particular are uniquely tunable by anion exchange and blending that
can precisely modulate the HOMO and LUMO levels by up to 1 eV,[31] which is desirable for aligning
energy levels to balance electron transfer efficiency and interface recombination that can push performance
towards the excitonic limit. This energy level tuning can also help to experimentally determine exciton
binding energies and optimize the donor-acceptor LUMO level offset for maximum JSC and VOC.[110] Due
to their small bandgaps and energy level tunability, heptamethine salts have been used in TPVs with high
AVT (>70%) and PCEs of up to 2.2%.[36] Moreover, opaque tandem devices have been demonstrated using

32

Figure 2.8. High performance transparent solar cell module. (a) JV characteristics for Newport-certified
high performance single junction transparent solar cell from Ubiquitous Energy using proprietary materials.
(b) Transmission, reflection and EQE spectra for device in (a). (c) Photo of large-area, series-integrated
module made using devices in (a).
heptamethines and trimethines with 4.3% efficiency,[111] making these a promising material set for tandem
TPVs in the future.
High efficiency solution processed conjugated polymer TPVs have also been demonstrated based
on

poly(2,60-4,8-bis(5-ethylhexylthienyl)benzo-[1,2-b;3,4-b]dithiophene-alt-5-dibutyloctyl-3,6-bis(5-

bromothiophen-2-yl)pyrrolo[3,4-c]pyrrole-1,4-dione) (PBDTT-DPP) with PC60BM as the acceptor.[22]
PBDTT-DPP is composed of BDT donor units and DPP acceptor units (strong acceptor behavior with weak
quinoidal resonance) with a peak polymer film absorption positioned at ~780 nm[97]. When combined
with a composite AgNW electrode, a PCE of 4.0% and AVT of 64% was achieved, the highest combination
of PCE and AVT for solution-processed single junction solar cells (Figure 2.6 (c)).
Recently, a certified PCE of 5.1% with 50.9% AVT was reported,[1] which is currently the highest
combination of PCE*AVT for single junction devices (Figure 2.8). A highly transparent large area (10 cm
x 10 cm) module was similarly demonstrated by combining these devices in series integration, which
resulted in voltages over 100V under mid-day outdoor illumination.

33

Figure 2.9. Transparent luminescent solar concentrators. (a) Schematic of luminescent solar
concentrators reproduced from Ref. [5]. Incoming light is absorbed and re-emitted by a luminophore
embedded in a wave-guiding plate. (b) Absorption and emission spectra of luminescent dye with Stokes
Shift S and peak width W. (c) Photo of a TLSC, from Ref. [5].
2.3.5 Transparent luminescent solar concentrators
Transparent luminescent solar concentrator (TLSC) are a related technology to TPV, which can
exhibit high AVT, CRI, and defect tolerance (Figure 2.9 (a)). As with other LSCs, energy is guided over the
device optically rather than electrically. TLSCs are composed of luminophores that absorbs either UV or
IR light and reemits deeper NIR light. To minimize reabsorption of the emitted light, the Stokes shift of
the luminophore (the difference between absorption and emission peak, in nm) should ideally be > 100nm
(Figure 2.9(b)).[112] Reabsorption losses are also a primary limiting factor for the scale-up of TLSCs > 1
m2. UV-absorbing inorganic luminophores with large down-conversion shifts (~400 nm) have been used
for TLSCs with PCE > 0.5% and AVT > 85%.[113] However, due to the lower photon flux in the UV region
of the solar spectrum, it is important to develop NIR-selective luminophores with large Stokes shifts for
TLSCs as well. TLSCs using NIR-absorbing organic salts have been made with 0.4% PCE and 86% AVT
(Figure 2.9(c)).[5]

Looking ahead, the key challenges for TLSCs are associated with the design of

luminophores with high luminescence efficiency, high Stokes shifts, and deeper NIR-selective harvesting.
2.4 TPV transparent electrodes
Transparent electrodes are a critical component of TPV devices as they are required in both the
anode and cathode with both high and low work functions. Transparent electrodes have been made using
transparent conductive oxides, ultra-thin metal films, conductive polymers, carbon nanotubes, and metallic

34

nanowires. We summarize these key electrode materials that can be integrated into such devices and discuss
considerations for their full optimization. Future advances in transparent electrodes will greatly aid the
development and scalability of TPVs.
2.4.1 Transparent conductive oxides
Transparent conductive oxides (TCOs) have wide bandgaps, which lead to low absorbance in the
visible spectrum. Indium tin oxide (ITO) is the most commonly used transparent electrode for thin film
solar cells (OPV,[44] a-Si,[114] CdTe,[115] etc.), organic light emitting diodes (OLEDs),[116] LCDs,[117],
and touchscreens. It has a combination of good transparency (~80% in visible wavelengths) and
conductivities (sheet resistances of around 10-20 Ω/□) and accordingly has been used as a top electrode in
many TPV studies.[4, 14, 30, 31] ITO is typically deposited via magnetron plasma sputtering but can also
be deposited using pulsed laser deposition or solution deposition of nanoparticles. A high temperature (400500°C) annealing step is typically used for ITO to achieve its maximum transparency and conductivity. The
deposition of ITO is straightforward when deposited as the first/bottom electrode onto hard substrates like
glass or steel, but it can be less compatible with flexible substrates due to the inherent brittleness of
polycrystalline ITO and the high temperature annealing step. The ion bombardment of sputtering may
damage underlying organic films, leading to lower device yields when ITO is used as a top electrode.[118]
Barrier layers such as organics (e.g. CuPc),[119] metal oxides (MoO3),[30] or thin metals (Ag)[4] can be
deposited prior to sputtering ITO to protect the underlying organic films. Low energy and oblique
depositions of ITO have also been explored to minimize damage to underlying films. For example, ITO has
been directly deposited onto organic layers using low operating powers of 7-25 W and RF sputtering at a
low rate (0.05-0.3 Å/s).[14]
Another key consideration of using ITO (and many other conductors) is the potential for strong
surface plasmon resonances (SPR) in the NIR, which commonly results in significant absorption at
wavelengths in the range of 1100-2000 nm (Figure 2.10(a)). SPR are excited surface waves of conduction
electrons found in any conductor. These oscillations and absorption features are highly sensitive to the
dielectric environment, angle of excitation, conductivity, and density; thus, the SPR are also sensitive to
35

Figure 2.10. Thin silver and ITO comparison. (a) Transmission spectra for AgNW electrodes with
different transparencies and sheet resistances with ITO as a reference. Note the pronounced IR absorption
of the ITO electrode for Îť > 1200 nm due to SPR. Graph reproduced from Ref. [7]. (b) Transmission spectra
of ITO processed under different annealing conditions. Graph reproduced from Ref. [16] (c) SEM image
of an AgNW electrode reproduced from Ref. [7]. (d) Normalized extinction coefficient for Ag nanoplates
of different sizes, reproduced from Ref. [26].
the processing conditions (such as annealing, doping, etc.). This is highlighted in Figure 2.10(b) for ITO,
where annealing in air can shift this SPR deeper in the IR. Ultrathin metals and metallic nanoparticles also
often exhibit SPR in the visible and NIR and therefore also need to be carefully considered, while metal
nanowires often have SPR deeper in the IR. As will be discussed in Section 2.5, the ideal bandgap for a
completely transparent solar cell is around 1.1 eV, which corresponds to a wavelength of 1100 nm. Thus,
SPR could result in considerable amounts of parasitic NIR absorption in the contact layers, which will be
important to consider in future development of TPVs.
Alternative transparent conductive oxides have been considered for ITO due to the limited
availability of In and cost. One example is zinc oxide, which is earth-abundant and can be doped with
aluminum and gallium for increased conductivity. Aluminum-doped zinc oxide (AZO) has been used with
36

TiO2 and Al interlayers for poly[9’-hepta-decanyl-2,7carbazole-alt-5,5-(4’,7’-di-2-thienyl-2’,1’,3’benzothiadiazole) (PCDTBT):PC71BM solar cells to achieve 3.9% PCE with 21.5% AVT.[33] The 8 nm
TiO2 layer was spincoated on top of the mixed PCDTBT:PC71BM layer for electron selectivity and
protection during the sputtering process. The Al interlayer provided partial reflectance as well as chemical
protection between the AZO and TiO2. Schubert demonstrated a highly transparent gallium doped zinc
oxide top electrode using pulsed laser deposition with an average transmission of 82.7% (400-800 nm) and
sheet resistance RS of 83 Ω/□.[120] By placing a metal shield between the target and the sample during
PLD, direct line of sight ballistic particles were reduced that limited damage to the underlying organic films.
Amorphous sputtered oxides may also be suitable for flexible substrates. This was shown, for
example, with InZnAlO (IZAO) where the resistance did not change over 10,000 bending cycles.[121]
Additionally, the IZAO electrode showed high transmission (AVT ~88%) and low sheet resistivities (85
Ω/□).[121] When used as a bottom electrode in P3HT:PCBM flexible devices on polyethylene terephthalate
(PET), IZAO had higher efficiencies (2.5% vs. 1.9%) compared to cells with ITO and PEDOT:PSS.[121]
Combining amorphous sputtered oxides like IZAO with methods to mitigate damage from sputtering could
enable their use as both bottom and top electrode in rigid and flexible TPVs.
2.4.2 Ultra-thin metals
Ultra-thin metals are attractive candidates to replace ITO for their compatibility with roll-to-roll
processes, low-temperature processing, and improved flexibility. They can be deposited using thermal
evaporation, which causes less damage to underlying organic films compared to sputtering. However, ultrathin metals face an inherent tradeoff between transparency and conductivity at low thicknesses and are in
general less transparent than ITO except in the deep NIR spectrum (Îť > 1000 nm) (Figure 2.10 (a)).
Moreover, many metals such as Ag can often degrade over time via oxidation.[122]
The sheet resistance at low thicknesses increases due to the rise of electron scattering at surface
and grain boundaries, which can be predicted for a variety of metals using the Fuchs-Sondheimer and
Mayadas-Shatzkes models.[123] Ag films with thicknesses below 12 nm are not typically continuous or
conductive,[124] but new approaches via seeding have enabled ultrasmooth and ultrathin metals.[125, 126]
37

More recently, a flexible TPV using thin Ag as the electrode and PET as the substrate was demonstrated
with little drop in performance after 1500 bending cycles at 2.7% tensile strain.[127]
The deposition of composite metal/semiconductor layers, originally conceived as a way to move
the electric field intensity closer to the donor acceptor interface |EDA|2 , has become a widely-used method
of making semi-transparent solar cells.[34, 36, 109, 128] In composite electrodes consisting of
metal/semiconductor multilayers, the metal layer is generally used for lateral conductivity while the
semiconductor functions both to select carriers and manage optical field profiles. O’Connor modeled the
electric field intensity at the donor-acceptor interface for a generic capping layer as a function of index of
refraction and thickness.[128] In the planar CuPc/C60 device, an enhancement of 30% was seen in both the
modeled |EDA|2 for Îť = 470 nm (absorption of C60) and experimental JSC for a capping layer consisting of 75
nm of aluminum hydroxyquinoline (Alq3).[128] VĂŠron used this approach in organic capping layer for
enhanced light in/outcoupling in a semitransparent electrode of 12 nm Ag with 20 nm Alq 3 to make a
transparent solar cell with 2.2% PCE and 71.4% AVT.[36] Inorganic semiconductors like MoO3 can also
be used to make transparent electrodes, as in the study by Lin that used an electrode of Ag (14 nm) / MoO3
(40 nm) in a DTDCTB:C70 device that had 3.24% PCE and 46.4% AVT.[34] This approach could also be
used to eliminate ITO from the device entirely since the bottom ITO can also be replaced by a
semiconductor/metal electrode.

This was done using MoO3/Ag in the study by Wang[129] and

MoO3/Ag/MoO3 was used to replace ITO as a bottom anode to make PCDTBT:PC71BM solar cells with
4.4% PCE.[130] In this study, 2 nm of MoO3 served as a seed layer for the 6 nm Ag, which was deposited
onto -5°C substrates in order to prevent Ag interdiffusion.[130] A separate study by Schubert used a similar
approach with 1 nm of Ca, Al, or Au as seed layers for the lateral-conducting Ag layer,[125] where the
highest transparency electrode consisted of 3 nm MoO3 / 1 nm Au / 7 nm Ag / MoO3 5 nm / Alq3 4 nm and
has a peak transmittance of 83% at 580 nm and a resistivity of 19 Ω/□.

38

2.4.3 Polymers and conductive nanoparticles
As an alternative to the vapor deposition of oxides and metals, polymer-based conductors have
been explored as transparent electrodes that can be deposited using spincoating or spraycoating. However,
these polymer-based conductors typically exhibit lower conductivities and transparencies when compared
to other alternative electrodes and may be incompatible on some systems due to solubilities or wettability
of the active layers. PEDOT:PSS, which is typically used as an anode modification layer or can be used as
a transparent electrode directly, is deposited via spincoating[131-133] spraycoating,[25, 134, 135], or vapor
polymerization. For example, a transparent inverted OPV was made with the following structure:
ITO/ZnO/SQ:PC60BM/PEDOT:PSS, where ZnO was spincoated on from a sol-gel solution, the
SQ:PC60BM solution was spincoated and the low and high-conductivity formulations of PEDOT:PSS were
spraycoated.[25] As PEDOT:PSS is an aqueous solution, it is difficult to spincoat onto hydrophobic organic
surfaces. In these cases, dilution in isopropanol or surfactant additives can be used to achieve enhanced
wetting. For example, a layer of AI4083 (lower conductivity formulation of PEDOT for interfaces) diluted
in isopropanol and n-butanol was first spray coated, rendering it suitable for deposition of PH500 (high
conductivity PEDOT).
Conducting nanoparticles can also be suspended in solution and spincoated or spraycoated to make
transparent electrodes. Examples include silver nanowires (AgNW),[136-140] ITO nanoparticles,[22]
graphene[141, 142] and carbon nanotubes.[28, 143, 144] These conducting nanoparticles usually need to
be embedded in a conductive matrix or mechanically contacted to achieve higher conductivity. One
example is the development of a composite AgNW electrode in TPVs, where the gaps between silver
nanowires were filled with a spincoated layer of ITO nanoparticles.[22] Other hybrid electrodes have
combined thin Ag, silicon nanowires, PEDOT:PSS, and Alq3 to improve transmission over a long range of
wavelengths, highlighting the potential benefits in incorporating multiple optimization strategies.[145]

39

2.5 TPV efficiency limits
The theoretical efficiency limits are useful in understanding the overall potential of TPVs and can
be used to establish reasonable performance goals for different applications. Calculation of the efficiency
limits of transparent solar cells starts with the Shockley-Queisser (also known as detailed balance) limit.[65]
Every photon outside of the visible band with energy higher than the bandgap EG is assumed to be absorbed
and converted to an electron with 100% efficiency, setting the upper limit for JSC. The electron then relaxes
to the bandgap and is collected at the potential defined by the maximum voltage from the dark
recombination rate of a black body radiator. For AM1.5G sunlight, the maximum efficiency for an opaque,
single junction cell is 33.1%. We note that variations in this number have been reported due to slight
variations and approximations in the solar spectrum used to perform the calculation.[2] The range of visible
wavelengths for calculating the efficiency limit was defined to be 435-670 nm, where the CRI can be > 95

Figure 2.11. Efficiency limits of TPVs. Figures reproduced from Ref. [2]. (a) Calculated CRI and AVT
values for ideal transmission curves with different UV and IR wavelength cutoffs. (b) Idealized stepfunction absorption curves for a solar cell with bandgap EG. For a AVT of 10%, the absorption would be
10% from 435-670 nm, as shown by the dashed line. (c) Thermodynamic efficiency limits for TPVs with
different AVTs as a function of their bandgap. (d) Thermodynamic efficiency limit for TPVs with different
AVTs and number of junctions.
40

and the AVT can be > 99.5% (Figure 2.11(a)).[2] While it is possible to see wavelengths outside this range
if bright enough (e.g. 370-435nm ad 670-750nm), these wavelengths do not strongly contribute to our
perception of color rendering. (Figure 2.11(b)).[2] The idealized step-function absorption curves envisioned
in Figure 2.11(b) can be achieved with molecular absorption profiles alone or enhanced with the use of 1D
photonic crystals. The thermodynamic efficiency was calculated for varying bandgaps and for varying
amounts of neutral absorption profiles in the visible (Figure 2.11(c)). A single junction transparent solar
cell with 100% transparency has a theoretical efficiency limit of 20.6%.[2] This efficiency limit can be
further increased via multijunction integration (as described in Section 1.3.5), where complementary
bandgaps enable high energy photons to be absorbed in higher voltage subcells. Indeed, one of the biggest
loss mechanisms in the single-junction efficiency limit is the thermal relaxation of hot electrons to the
bandgap. For a transparent two, three, and infinite-junction solar cell with series integration and perfect
photocurrent matching, the thermodynamic efficiency limits are 27.1%, 29.8%, and 37%, respectively
(Figure 2.11(d)).[2]

41

Chapter 3 – Experimental Techniques
In this chapter, various experimental techniques used to characterize thin films, synthesize new
compounds and fabricate organic solar cells will be discussed. Details of the device testing will also be
described.
3.1 Thin film optical measurements: UV-Vis and VASE
Ultraviolet-visible spectroscopy (UV-Vis) is a technique used to characterize the transmission of
either a thin film or a thin film stack. With UV-Vis, a light source consisting of broad spectrum light source
(e.g. an incandescent bulb) is passed through a monochromator. The monochromatic light is split in two
beams, one that goes through the thin film sample and one that is used as a reference for the intensity. A
detector collects the intensity of the two beams of light and subtracts one from the other to get the
transmission of the sample (0-100%). For thin-film transmission measurements, reference samples (e.g.,
clear glass slides) should not be utilized as in the case for solution transmission measurements. In solution
measurements, a reference solution containing just the solvent can be utilized to subtract front/back and
solvent reflections so that the transmission data can be straightforwardly converted to absorbance. The case
for thin films is different and reflections cannot be simply subtracted due to the complex optical interference.
Accordingly, no reference sample should be used for thin film measurements. The reflection of a thin film
sample can also be measured via UV-Vis. For this measurement, a different sample mount is used with
triangular mirrors. The sample beam first hits the mirror, then the sample, then the mirror, and is then finally
collected at the detector. With the combination of experimental measurements for reflection and
transmission, absorption spectra can be estimated Equation 2.3, A + R+ T = 1. However, this assumes no
scattering losses, which means that absorption can be overestimated by this method.
Variable angle spectroscopic ellipsometry (VASE) is another technique used to characterize thin
film optical characteristics and nm-level thicknesses. Ellipsometry measures the change in polarization in
light after it is reflected off a surface. A monochromatic beam of light is polarized and sent through a fiber
optic directed at a thin film sample on a reflective surface. The light reflected off the sample is collected at

42

Figure 3.1. Ellipsometry fitting for thin and thick films on silicon. (a) Experimental Ψ for three different
incident angles, 65°, 70° and 75° for 100 Å thick Cy-TPFB. Thickness was fit in transparent region (λ >
1000 nm). (b) Experimental Ψ and Δ for 70° incidence for 2000 Å thick TiO2. Oscillations are visible for
thicker films.
a detector, and, based on the phase difference and light attenuation of the reflected light, the ellipsometric
angles Ψ (amplitude component) and Δ (phase component) can be calculated:[146]
𝜌=

𝑟𝑝
𝑟𝑠

= tan(𝛹) 𝑒 𝑖𝛥

(3.1)

where ρ is the complex reflectance, rp is the reflectivity of p-polarized light, and rs is the reflectivity of spolarized light. Ψ and Δ are a polar description of the ratio of reflectance for p- and s-polarized light. The
goal of the measurement is to obtain the optical constants n and k (refractive index and extinction coefficient,
respectively) and the thickness d of the thin film sample. With three unknowns (n, k and d) and two
experimental parameters (Ψ and Δ), it is not advisable to fit for all three at once. One approach is to measure
Ψ and Δ for in a wavelength range where the film is transparent (k = 0) and fit for n and d. The wavelength
dependence of n in the transparent region is given by the Cauchy Urbach model. Then, the thickness d is
fixed, and n and k are calculated wavelength by wavelength from the transparent region. A model (Snell’s
laws, the Fresnel coefficients, and thin film interference equations) is then used to fit the optical constants
n and k from Ψ and Δ. The measured thickness can be used to calculate the tooling factor (discussed in next
section on device fabrication) of thermally evaporated materials, and the optical constants n and k can be
used in optical interference models (see Section 4.2). The range of measurable thicknesses for ellipsometry
is around 0.1 to 1000 nm. For thin (~100 Å) films, Ψ and Δ are relatively stable across the transparent region

43

Figure 3.2. Counterion exchange synthesis. (left) Photograph of 1-step counterion exchange synthesis.
The more nonpolar Cy-TPFB (orange solid) precipitates out of a mixture of Cy-I and K-TPFB in 5:1
MeOH:DCM. (right) Dried silica plug after silica gel chromatography to separate unreacted Cy-I (dark band
on top) from Cy-TPFB.
of the film (Figure 3.1 (a)). For thick (> 1000 Å) films, optical interference patterns can be seen in the form
of oscillations in Ψ and Δ (Figure 3.1 (b)).
3.2 Heptamethine counterion exchange
Heptamethine cations typically carry a +1 charge and is coordinated with a -1 anion, which can be
a single ion (i.e., I-) or a polyatomic ion (such as BF4-). One example of the reaction is shown below for
exchanging the counterion from I- to tetrakis(pentaflurophenyl)borate (TPFB):

An example synthesis starts with equimolar amounts of potassium tetrakis(pentaflurophenyl)borate (KTPFB, Boulder Scientific Company) and Cy-I (American Dye Source) which are dissolved in 5:1
methanol:dichloromethane (MeOH:DCM) at 10 mg/ml and stirred at room temperature for 15 minutes in
ambient atmosphere. In this case, a solid orange precipitate forms due to the more nonpolar product CyTPFB being formed in a relatively polar solution of primarily methanol (Figure 3.2). The crude product is
then collected using vacuum filtration and an MeOH wash. The crude product is redissolved in minimal
DCM (~10 mg/ml) and poured through a plug of silica using DCM as eluent to remove impurities and
44

unreacted Cy-I. In silica gel chromatography, solvent is drawn through silica gel by a variety of methods
(capillary motion, gravity or vacuum), and components of a mixture can be separated by polarity. The more
polar reactant Cy-I binds to the silica gel, and the less polar product Cy-TPFB is eluted with the nonpolar
DCM through the silica plug. The first fraction with similar color to Cy-I is collected (in this case, green),
and excess DCM is removed in a rotary evaporator at 55°C for 20 min at atmospheric pressure.
3.3 Purity assessment via high-resolution mass spectrometry
Mass spectrometry is an analytical technique that identifies the mass over charge (m/z) ratio of a
given compound. The goal of mass spectrometry is to obtain the molecular mass of a sample and any
impurities. Mass spectrometry consists of three main components: an ionization source, an analyzer, and a
detector. The ionization source used for our experiments was electrospray ionization (ESI), which means
that a dilute (1-100 nM) solution is led to the end of a nozzle where a capillary voltage (1 kV or higher) is
applied to create ions in solution. The solution is then mixed with a drying gas and sprayed, thereby
generating gaseous ions. The next component is the analyzer, which for our experiments was both a
quadrupole and a time of flight detector (QToF). This type of mass spectrometer containing two analyzers
is called a tandem mass spectrometer and has higher resolution. A quadrupole consists of four rods, two of
which are positive and two of which are negative. A varying voltage is applied onto the quadrupole such
that only ions with certain m/z make it from one end to the other. In a time-of-flight analyzer, the ions are
accelerated in a tube, and their m/z is known by when the ions arrive at a detector by gravity. The final
component, the detector, is usually an electron multiplier.
Different elements have isotopes in varying relative abundances and each element will therefore
generate a specific isotope pattern in mass spectrometry. Examples of isotope patterns for the cation and
anion for Cy1-TPFB and Cy2-TPFB are shown in Figure 3.3.
The verification of cations and anions and ion purity assessment were performed using a Waters
Xevo G2-XS QToF mass spectrometer coupled to a Waters Acquity ultra-high pressure LC system. Cations
were analyzed in positive ion mode electrospray ionization, and anions were analyzed in negative ion mode

45

Figure 3.3. High resolution mass spectra for Cy1-TPFB and Cy2-TPFB. (a) Positive and (b) negative
mode electrospray ionization for Cy1-TPFB. (c) Positive and (d) Negative mode electrospray ionization for
Cy2-TPFB. Predicted isotopic abundance peaks for each compound were generated using the Isotope
Model tool in MassLynx software.

46

ESI. Solutions were prepared in acetonitrile and directly injected for 2 minutes using an eluent of 50:50
water:acetonitrile. Mass spectra were acquired using a dynamic range extension over m/z 50 to 1,500, with
mass resolution (M/ΔM, full width-half maximum) of approximately 20,000. Other parameters include
capillary voltage of 2 kV, desolvation temperature of 350°C, source temperature of 100°C, and desolvation
gas (N2) at 400 L h-1. For ion purity assessment, solutions of the exchange precursors and products were
prepared in acetonitrile with concentrations varying from 10 nM to 500 nM and analyzed by mass
spectrometry as described above.
3.4 Solar cell device fabrication
Organic solar cells consist of two electrodes, a donor, and an acceptor. A typical solar cell device
consists of ITO (transparent bottom electrode), MoO3 (a transparent HTL), a donor, an acceptor (usually
C60), an electron transport layer called bathocuproine (BCP), followed by the cathode, which is silver or
aluminum for opaque cells. For transparent devices, the top transparent electrodes used include ITO and
Ag / Alq3. The general device structure is shown in Figure 3.4 (a).
The active area for an organic cell is outlined by the overlap of the top and bottom electrodes
because the organic layer has poor lateral conductivity compared to Si. The patterns used for the bottom
ITO, organic layers, and top electrode are shown in Figure 3.4 (b). The bottom ITO layer is prepatterned
onto 12 x 12 mm polished glass substrates (Xin Yan Tech). The pattern allows for ITO to be contacted
easily during electrical testing. The organic layers (MoO3 through BCP) are deposited by thermal
evaporation through a stainless-steel shadow mask. The top electrode is deposited through another shadow
mask, creating a pattern of six “fingers”. Each finger is a different device, and this pattern allows for six
5.4 mm2 devices to be made on each substrate. For organic layers that have higher conductivities, a shadow
mask can be placed on top of the substrate during testing to define the device area.
The substrates need to be free of dust and organic residue before the deposition step. To accomplish
this, a substrate cleaning procedure is used as follows. Substrates are sonicated for three minutes each in
soap, DI water, and acetone. Then, the substrates are submerged in boiling isopropanol for three minutes
and dried with nitrogen. Finally, the substrates are treated with oxygen plasma for 90 seconds and kept in
47

Figure 3.4. Device architecture and layer patterns. (a) Typical device structure for an OPV. Cy stands
for cyanine, but could be any organic donor. (b) Different layer patterns used for ITO, organic layers, and
the top electrode. The device area is outlined. This pattern for a 12-mm substrate makes six 5.4 mm2 devices.

Figure 3.5. Evaporator diagrams. (a) Diagram of thermal evaporation chamber. The labelled parts are:
(1) source boat, (2) copper heating agents, (3) source shutters, (4) QCM, (5) substrate shutters, (6)
substrates, and (7) stage. (b) Schematic showing the positioning of the six evaporation sources and three
QCM sensors.
a nitrogen-filled glovebox to prior to the subsequent deposition steps.
Thermal evaporation is a type of physical vapor deposition where source material is sublimed under
vacuum. A simplified diagram for one evaporation source is shown in Figure 3.5(a). The source material
(in powder form) is loaded into tungsten boats with a baffle design that emits sublimed material by line of
sight to the substrate where it condenses. The tungsten boats are heated by copper filaments. The pressure
is usually maintained at around 10-6 torr, which is achieved through a two-stage pump (a rough mechanical

48

pump and a cryopump). Between the active-layer and electrode deposition, layer masks are changed by
unloading the substrates with a transfer fork into a nitrogen glovebox to prevent exposure to air before the
device is completed or packaged. The position of the pump, transfer fork, evaporation chambers and
nitrogen glovebox can be seen in Figure 3.6.
In thermal deposition, the deposition rate (usually 0.5 – 2 Å/s) is monitored using quartz crystal
monitors (QCM) above each source. Quartz crystals resonate at specific frequencies that depend on the

Figure 3.6. Custom-built system assembly for two evaporation chambers and a nitrogen glovebox
(Angstrom Engineering). A. EvoVac evaporation chamber. B. EvoVac cryopump. C. EvoVac helium
compressor. D. Nitrogen glovebox. E. Solvent glovebox. F. AMOD evaporation chamber. G. AMOD
cryopump. H. AMOD helium compressor.

49

total volume of thin film deposited on them. The electrical power delivered to the source boat for resistive
heating is controlled using standard PID control. Once the deposition rate is stable, the substrate shutters
open, allowing for a controllable deposition rate and a controllable thickness. QCMs are calibrated to relate
the thickness arriving on the QCM to the thickness on the substrate by the ratio called the tooling factor
(TF), given below:
𝑡𝑖
𝑇𝐹𝑖

𝑡

= 𝑇𝐹𝑓

𝑓

(3.2)

where ti is the initial measured thickness based on the initial guess of the tooling factor TFi and tf is the
thickness measured on a reference substrate (usually undoped single-crystalline Si) using ellipsometry and
TFf is the calibrated tooling factor. The TF depends on the source material (density, Z-factor), boat position,
and the overall boat-QCM-Substrate geometry. The Z-factor (also called Z-ratio) is the ratio of the acoustic
impedance of quartz to that of the deposited material and is a correction factor in the frequency-change-tothickness function. Accordingly, if a boat is repositioned or a QCM is moved, the TF needs to be remeasured.
Because the QCM is calibrated, the co-deposition of multiple materials in a single layer should be reported
as a vol. % (from the ratio of the deposition rates) and not as mass or molar %.
In our AMOD evaporation chamber (Angstrom Engineering), there are six thermal evaporation
sources and three QCMs: each QCM is responsible for monitoring the deposition rates of two sources. The
positioning of the six evaporation sources is shown in Figure 3.5(b). Different materials can be deposited
at the same times because of this design; this process is referred to as codeposition and is necessary for
creating bulk heterojunctions. The ratio of donor and acceptor can be varied using codepositon, providing
a useful tuning parameter for improved film morphologies, doping, and device performances.
Some materials are unsuitable for thermal evaporation because they decompose at high
temperatures prior to sublimation. For these materials, solution processing methods such as spin-coating
and doctor blading can be used to deposit thin films from solution. In this research, spin-coating is used to
deposit thin films of the heptamethine salts discussed in Chapter 5, which are too bulky for evaporation. In
spin-coating, the substrate is held on a central axis by a vacuum chuck and rotated at around 2000-4000

50

Figure 3.7. Diagram of spin-coating process. (a) Solution is pipetted onto substrate. (b) The substrate is
rotated, causing solvent to evaporate and shear off. (c) Particles left behind on the surface form a thin film.
rpm (Figure 3.7). A solution is pipetted onto the substrate either before or during the rotation; these are
respectively called static or dynamic spincoating. The film is formed by a combination of solvent
evaporation and solvent shearing. Using a high vapor pressure solvent (low boiling point) leads to fast
solvent evaporation, creating thick, rough films. Using a low vapor pressure solvent (high boiling point)
leads to solution that spin off the substrate and result in films that could be too thin. Therefore, the optimum
solvent system can often be a mixture of a high and low vapor pressure solvents; for example, a mixture of
1:3 of dichloromethane (boiling point = 50°C) to chlorobenzene (boiling point = 131°C) was used as the
solvent system for spincoating the heptamethine salts covered in Chapter 5. Spincoating speeds and times
are other parameters that can be controlled for the spincoating process, with faster speeds and longer spin
times creating thinner films. The parameters used for the heptamethine salts were 2000 rpm and 30 seconds
of spin-coating time.
Opaque devices with a top electrode of 80 nm Ag are made for reference or control devices. For
transparent devices, the top electrode is either ITO or a combination thin metal/semiconductor layer (e.g.
Ag/Alq3). Sputtering is used to deposit ITO, which has too high of a sublimation temperature to be deposited
using thermal evaporation. In sputter deposition, plasma ions (Ar+ in our case) bombard a target of the
material to be deposited, and the ejected atoms are ballistically transported under low vacuum (10-2 torr) to
the substrate where they condense. As mentioned in Chapter 2, there are tradeoffs between using ITO or
Ag/Alq3 as the transparent electrode. ITO is more transparent and conductive, but the sputtering process
can damage the underlying organic films and lead to low device yield. Low deposition rates (0.02 Å/s) are

51

Figure 3.8. Photographs of JV testing setup. (a) Calibrating the intensity of the xenon arc lamp for JV
testing. (b) Substrate holder for testing. Gold pins contact different devices on the substrate. (c) Screenshot
of Labview program used to interface with sourcemeter.
typically used to prevent this degradation from excessive high energy plasma ions. Ag/Alq3 is less
transparent but can be deposited at higher rates without damaging the organic films underneath.
3.5 Solar cell testing (JV-EQE)
3.5.1 Current-voltage (JV)
Obtaining the JV curve for a solar cell under 1-sun illumination allows us to calculate its power
conversion efficiency (Figure 3.7). Current density (J) was measured as a function of voltage (V) using a
Labview-controlled sourcemeter. The design of the electrode patterns allowed for independent and rapid
testing of all the devices on a substrate. The illumination source was a xenon arc lamp calibrated for
AM1.5G (100 mW/mm2) intensity using a NREL-calibrated Si reference cell with a KG5 filter to match
the response of the Si cell to our testing cells (for details, see mismatch factor discussion in Section 3.5.2).
AM 1.5G refers to the solar irradiance at the surface of Earth after being absorbed by air with a diameter of
1.5 times the thickness of the atmosphere and is a standard illumination source for PV testing.
3.5.2 External quantum efficiency (EQE)
The external quantum efficiency (EQE) for a solar cell is the ratio of collected electrons over
incident photons for a given wavelength. EQE measurements are performed by illuminating the device area
with monochromatic light from a tungsten halogen lamp chopped at specific frequency (200 Hz was used
in our lab). The small current signal from the device is sent through a picoammeter and amplified by a lock52

in amplifier. The light intensity at the end of the IR-fiber was measured using a Newport calibrated Si diode
for 350-800 nm and a Newport calibrated Ge diode for 800-1600 nm. The EQE can be integrated with the
solar spectrum to give the JSC, making it a good consistency check for the JSC measured by JV testing, which
is easy to overestimate. Errors in device area and illumination intensity can result in overestimations of JSC.
EQE is also used for calculations of the spectral mismatch factor (MF), which accounts for the
difference in the reference detector for setting the solar simulator power and the spectral response of the
tested solar cell considering the non-ideal solar simulation source. The mismatch factor is used to correct
the incident power Pin from 1-sun intensity (PCE = JSC*VOC*FF / Pin*MF). In practice, EQE is measured
before JV to calculate MF and adjust the intensity of the simulation lamp. For example, if an MF of 1.05
were measured, then the intensity of the lamp would need to be decreased by 5% so that the incident power
would be 1-sun (100 mW mm-2). To calculate MF, the EQE of a silicon reference cell with a known response
is measured using the solar simulation lamp, as well as the EQE of the test device. The equation for MF is
as follows:
𝑀𝐹 =

∫ 𝐸𝑟𝑒𝑓 (𝜆)∗𝑆𝑅 (𝜆)𝑑𝜆
∫ 𝐸𝑟𝑒𝑓 (𝜆)∗𝑆𝑇 (𝜆)𝑑𝜆

∗

∫ 𝐸𝑆 (𝜆)∗𝑆𝑇 (𝜆)𝑑𝜆
∫ 𝐸𝑆 (𝜆)∗𝑆𝑅 (𝜆)𝑑𝜆

(3.3)

Where Eref is the solar spectrum, SR is the spectral response of the silicon reference photodiode, ST is the
spectral response of the test device, and ES is the solar simulation lamp (source) spectrum. The silicon
reference cell normally has photoresponse out to 1100 nm. We use a KG filtered silicon reference cell
because the KG5 filter has low transmission in the infrared; the KG5-filtered reference cell has a more
similar photoresponse to OPVs, and therefore, using it makes the MF closer to 1.[147]
In Figure 3.9(a), the set-up for EQE measurement of a multijunction solar cell is shown.[6] As in
the single junction case, chopped monochromatic light is incident on the sample, and a lock-in amplifier
amplifies the PV current responding to the same frequency as the chopping of the monochromatic light. To
measure the EQE of an individual subcell (e.g. subcell 1) of a tandem solar cell, the other subcell (e.g.
subcell 2) must be filled with carriers so that every carrier excited by the chopped monochromatic light in
subcell 1 either is collected at the electrode or recombines with the carrier of the other type at the

53

Figure 3.9. EQE for tandem PV. (a) Set-up for EQE measurement of tandem PV. Optical bias lights pass
through a neutral density filter to control for intensity before illuminating the sample. (b) Normalized
absorption spectra of two active materials in Ref. [6], PFTBT:PCBM in wine and pBBTDPP2:PCBM in
green. Figures reproduced from Ref. [6] with permission.
interconnecting layers (ICL, also called recombination zone or intermediate layers). The optical biasing
lights are un-chopped, so the current they generate in the solar cell are not amplified by the lock-in amplifier.
Neutral density filters and hot mirrors are used to easily switch which subcell is being optically biasing
different subcells.

54

Chapter 4 – Angle Dependence Modeling
Integrating transparent solar cells into building envelopes is a unique opportunity to reduce the
levelized electricity cost of solar power. However, this integration requires a consideration of the angular
dependence of these devices because illumination around the building envelope is rarely at normal
incidence. In this chapter, we explain how we updated standard transfer-matrix equations to accurately
model the quantum efficiency and optical properties under oblique illumination. We use this model to
demonstrate the various angular performance characteristics possible for proof-of-concept optimizations of
transparent planar-heterojunction solar cells and discuss considerations needed to fully account for optical,
electrical, and positional configurations in this optimization.
4.1 Motivation for angle dependent PV performance modeling
The deployment of PVs in BIPV applications warrants the consideration of the angle-dependence

Figure 4.1. The position of the sun and the average yearly flux for PVs mounted in different
configurations. (a) Schematic showing angles related to the position of the sun. S stands for the position
of the sun, and M stands for the surface normal of the module. Adapted from Ref. [148]. (b) Table of
average yearly flux for solar cells mounted in different positions in different U.S. cities. Adapted from Ref.
[148].

55

Figure 4.2. Conventional and inverted configurations for TPV. Schematic showing device layer
structure, illumination orientation, and incident angle h for the (i) conventional, (ii) optically inverted, (iii)
electrically inverted, and (iv) optically and electrically inverted configurations. (TCOA denotes the holecollecting transparent anode, and TCOC denotes the electron-collecting transparent cathode).
of both the solar flux and oblique angle device performance. In general, the photon flux follows a cosine
decay as a surface is tilted away from a normal-incidence beam. However, the orientation of solar
illumination varies throughout the day and year such that the average solar flux only drops modestly. This
is in part because no surface stays at normal incidence for long without solar tracking and east/west facing
surfaces can collect greater flux in the morning/evening. Figure 4.1 shows the average yearly solar flux for
vertically mounted surfaces facing south, southeast, east (identical to west), and north in three northern
hemisphere cities in the U.S. A comparable amount of solar flux is received for horizontal and vertically
oriented surfaces over the year except in north facing vertical surfaces where there is no direct sunlight
(only diffuse solar illumination). When combined around the entire surfaces of a building the overall
vertical flux from east-west-south facings surfaces can be substantially greater than the horizontal or tilted
fluxes. Transparent solar cells can also be oriented in multiple optically and electrically inverted
configurations (Figure 4.2). Therefore, obtaining an accurate model for solar cell performance at oblique
illumination angles and non-conventional device configurations is critical.

56

To further improve the efficiencies of OPV and TPV and reduce thermal losses, a multijunction
approach is commonly employed.[2, 23] The photocurrent matching requirement of series-integrated
subcells in multijunction PVs mandates a precise understanding of the angular performance of each subcell
to ensure photocurrent matching across all angles. Previous efforts to model the angle dependence of
conventional organic photovoltaics[149, 150] based on Ref. [151] lack clear descriptions of the correct
angle-dependent model (e.g., fields and generation terms for both s- and p-polarizations) and do not show
consistency checks necessary to assess the accuracy of these results. Here, we update the transfer matrix
method (TMM), which has been widely used to model solar cell performance at normal illumination,[151,
152] to account for oblique illumination as a function of optical orientation, and we demonstrate complete
self-consistency. We then model and optimize the performance of planar heterojunction (PHJ) TPVs under
normal and oblique illumination for the two optical configurations.
4.2 Optical interference modeling
Complex optical interference effects arise because optimal layer thicknesses for organic and
transparent PVs are typically in the range of 10-100 nm, which is the same order of magnitude as
wavelengths of incident light. The resulting field profiles for thin film structures often resemble sinusoidal
functions, rather than pure exponential decays of the Beer-Lambert equation. PV performance models that
account for this optical interference are used in many thin-film solar cell studies to aid in efficiency
optimization and understand angle dependencies. The transfer matrix model (TMM) is a popular model
used to describe optical interference.[151, 152] TMM accounts for the complex interface
reflections/transmission in a multilayer film stack. For a multilayer film stack with j = 1 to m layers and
each layer having a complex index of refraction Ăąj = nj + i*kj and thickness dj, the electric field propagated
at an interface between layer j and k can be written as:

1

 Ē j 
 Ēk   t jk
    I jk     
 Ēk   rjk
 Ē j 
 t jk

57

rjk 

t jk   Ēk 
 
1   Ēk 

t jk 

(4.1)

Figure 4.3. Schematic showing xyz coordinates for transfer matrix equations. E refers to electric field,
with Ei, Er, and Et referring to incident, reflected and transmitted electric field component respectively.

where the positive sign denotes propagation in the positive direction (+x, for example) and the negative
sign is propagation in the negative direction (Figure 4.3). rjk and tjk are the Fresnel reflection and
transmission coefficients respectively, with rjk = (ñj – ñk)/(ñj + ñk) and tjk = 2ñj/(ñj + ñk). The electric field
propagating within a layer j can be described as:

ei j d j
Lj  
 0

0 
 2
 , j  
 
e 
i j d j


 ùj


(4.2)

The product of all the interface matrices Ijk and layer matrices Lj results in the transfer matrix S:
S
S   11
 S21

S12   m

   I ( n 1) n Ln  I m ( m 1)

S22   n 1


(4.3)

which is the relationship between the electric field coming into and out of the multilayer stack, as in:

 Ē0 
 Ēm 1 

S
 
  
 Ē0 
 Ēm 1 

(4.4)

When these are combined with a 1-D exciton diffusion model, the exciton concentration profile can be
calculated:
L2D

2n  x 
 n  x    G ( x)  0
x 2

(4.5)

where LD is the exciton diffusion length, n is the exciton density, τ is the exciton lifetime, and G is the
exciton generation (or photon absorption) rate, which is proportional to the magnitude of the electric field

58

intensity |E(x)|2. These models typically assume 100% charge transfer and 100% charge collection
efficiencies, so the flux of excitons that diffuse to the D-A interface over the incident photon flux yields the
EQE at each wavelength. The EQE can then be integrated to calculate the short circuit photocurrent density
JSC (Eq. 1.5). The most relevant applications of optical interference modeling for TPVs are 1) simulating
the A, R, and T of a film stack at normal or oblique angle to verify the measured transmittance or internal
quantum efficiency of an entire device film stack [37], 2) extracting the parasitic absorption losses in
transparent contacts, 3) optimizing for photocurrent generation, and 4) designing layers for light
management such as with 1D photonic reflectors (e.g. distributed Bragg reflectors).[14, 153] For optimizing
device structures, electric field profiles are simulated as a function of position within the device while
varying the thicknesses of various layers (electron or hole transport layers[28] or transparent electrode
layers[34, 154, 155]) to optimize for high electric field intensity at the D-A interface to help overcome the
exciton diffusion bottleneck. In practice, there are three or more independent objective criteria to maximize:
efficiency, overall transparency, and color rendering.
4.3 Modifications for angular incidence
While the transfer matrix method has been described in detail in the literature,[156] we focus on
corrections needed for oblique illumination and optical inversion. The electric fields of two adjacent layers

j and 𝑘 = 𝑗 + 1 are related at their interface by a matrix consisting of the complex Fresnel reflection and
transmission coefficients, 𝑟𝑗𝑘 and 𝑡𝑗𝑘 , which differ for s- and p- polarized light as:[157]
𝑠
𝑟𝑗𝑘
=

𝑠
𝑡𝑗𝑘
=

𝑝

𝑟𝑗𝑘 =
𝑝

𝑡𝑗𝑘 =

ñ𝑗 cos(𝜃𝑗 )−ñ𝑘 cos(𝜃𝑘 )
ñ𝑗 cos(𝜃𝑗 )+ñ𝑘 cos(𝜃𝑘 )
2ñ𝑗 cos(𝜃𝑗 )
ñ𝑗 cos(𝜃𝑗 )+ñ𝑘 cos(𝜃𝑘 )
ñ𝑗 cos(𝜃𝑘 )−ñ𝑘 cos(𝜃𝑗 )
ñ𝑗 cos(𝜃𝑘 )+ñ𝑘 cos(𝜃𝑗 )
2ñ𝑗 cos(𝜃𝑗 )
ñ𝑗 cos(𝜃𝑘 )+ñ𝑘 cos(𝜃𝑗 )

59

(4.6)
(4.7)
(4.8)
(4.9)

where ñ𝑗 is the complex index of refraction for layer j (e.g. n + iκ), j is the incident angle in layer j, and k
is the refraction angle in layer k. s- and p- polarization refer to the electric field oscillating parallel (p-) or
perpendicular (s-) to the plane of incidence, which is the plane containing both the normal vector of the
interface and the light propagation vector. The polarization dependence can be neglected in the normal
incidence case (Eqs. 4.6-4.7 become equivalent to Eqs. 4.8-4.9) but not at oblique angles. Substituting
Snell’s law at each interface, cos(j) can be expressed as a function of the ambient incidence angle, 0, and
2

ambient index of refraction, 𝑛0 : 𝑐𝑜𝑠(𝜃𝑗 ) = √1 − (𝑛0 /ñ𝑗 ) sin2 (𝜃0 ). In the notation of Ref. [7], power
dissipation Qj for s- and p-polarized light in the thin film layers are:
2

1

𝑄𝑗𝑠 (𝑥) = 2 𝑐𝜀0 𝛼𝑗 𝑛𝑗 |𝐸𝑗𝑧 |
2

1

𝑝

(4.10)
𝑦 2

𝑄𝑗 (𝑥) = 2 𝑐𝜀0 𝛼𝑗 𝑛𝑗 (|𝐸𝑗𝑥 | + |𝐸𝑗 | )

(4.11)

where c is speed of light, 0 is permittivity of free space, j is the absorption coefficient of layer j, nj is the
𝑥,𝑦,𝑧

real index of refraction for layer j, and 𝐸𝑗

are the time and y-integrated Euclidean electric field

components in layer j (Figure 4.3):[158]
𝐸𝑧,𝑗 =
𝐸𝑦,𝑗 =
𝐸𝑥,𝑗 =

1
√cos 𝜃0
cos 𝜃𝑗
√cos 𝜃0

ñ0 sin 𝜃0
ñ𝑗 √cos 𝜃0

where ñ𝑗 cos 𝜃𝑗 = √ñ𝑗2 − 𝑛02 sin2 𝜃0 , 𝜉𝑗 =

+ 𝑖𝜉𝑗 𝑥
− −𝑖𝜉𝑗 𝑥
𝑒
+ 𝑡𝑗,𝑠
𝑒
(𝑡𝑗,𝑠
)

(4.12)

+ 𝑖𝜉𝑗 𝑥
− −𝑖𝜉𝑗 𝑥
[𝑡𝑗,𝑝
𝑒
+ 𝑡𝑗,𝑝
𝑒
]

(4.13)

+ 𝑖𝜉𝑗 𝑥
− −𝑖𝜉𝑗 𝑥
[−𝑡𝑗,𝑝
𝑒
+ 𝑡𝑗,𝑝
𝑒
]

(4.14)

2𝜋
ñ cos 𝜃𝑗 ,
𝜆 𝑗

−
and 𝑡𝑗,𝑝
is the same as 𝑡𝑗− of Pettersson et al (Ref.

[151]) except that the interface matrix Ijk (that the total transfer matrix comes from) consists of the Fresnel
+
−
+
complex reflection and transmission coefficients for p-polarized light (expressions for 𝑡𝑗,𝑝
, 𝑡𝑗,𝑠
, and 𝑡𝑗,𝑠
are

also defined by the total transfer matrix). The absorption coefficient is 𝛼𝑗 = 4𝜋𝜅𝑗 /𝜆, where κj is the
imaginary index of refraction for layer j. Total power dissipation is then calculated by taking the sum of
the s- and p-polarized power dissipations weighted by each fractional component of each polarization in
60

the incident light. Solar irradiance is unpolarized and therefore can be simulated with a 1:1 ratio of s- and
p-polarized light. To account for the substrate, we derive incoherent corrections from Figure 4.4 for the
conventional optical orientation with illumination through the substrate as:
𝑅 = 𝑅𝑆 +
𝑇=

𝑇𝑆 2 𝑅𝑇𝐹
1−𝑅𝑆 𝑅𝑇𝐹

𝑇𝑆 𝑇𝑇𝐹
1 − 𝑅𝑠 𝑅𝑇𝐹

𝐴 = 𝐴 𝑇𝐹

𝑇𝑆
1−𝑅𝑇𝐹 𝑅𝑆

(4.15)
(4.16)
(4.17)

And for the inverted optical orientation (illumination through the thin film) as:
𝑅 = 𝑅𝑇𝐹 +
𝑇=

2
𝑇𝑇𝐹
𝑅𝑠
1 − 𝑅𝑠 𝑅𝑇𝐹

𝑇𝑆 𝑇𝑇𝐹
1 − 𝑅𝑠 𝑅𝑇𝐹
𝑇

𝑅𝑆
)
𝑇𝐹 𝑅𝑆

𝑇𝐹
𝐴 = 𝐴 𝑇𝐹 (1 + 1−𝑅

(4.18)
(4.19)
(4.20)

where R, T, A are the reflection, transmission and absorption through the device at each wavelength; Rs, Ts
are the reflection and transmission at the substrate-air interface (non-film side); and RTF, TTF, ATF are the
reflection, transmission, and absorption through the thin film layers. Once the reflection, transmission, and

Figure 4.4. Substrate corrections for optical interference simulation. Schematic showing the first three
terms for reflection and transmission through the whole device. The clear block represents the substrate
while the gray block represents the thin film layer stack. RTF and TTF are the reflection from and
transmission through the thin film layers, and RS and TS are the reflection from and transmission through
the substrate.

61

absorption are found for the thin film layer stack (RTF, TTF, and ATF), a correction must be made to find the
reflection, transmission, and absorption for the overall device (R, T, and A), which includes the substrate.
The reflection and transmission associated with the substrate interface on the non-film side will be referred
to as RS and TS and it is assumed the substrate always has negligible absorption. Light that is transmitted
through the device consists of light that has been transmitted through the substrate and the thin film layer,
as well as light that has been reflected multiple times before being transmitted through both substrate and
thin film stack. Likewise, light that is reflected by the device consists of light that is reflected by the incident
surface or has undergone multiple reflections within the substrate before being transmitted through the
incident medium.
Given the exciton diffusion lengths (provided by Ref. [14]and Ref. [152]) and the position
dependent Qj, it then becomes possible to solve the one-dimensional diffusion equation to calculate the
external quantum efficiency (EQE) of the device. The short-circuit current density (Jsc) is subsequently
calculated by integrating the convolution of the EQE and the solar spectrum S as 𝐽𝑠𝑐 = ∫(𝐸𝑄𝐸 ∗ 𝑆) 𝑑𝜆.
4.4 Verification of model (internal consistency and experimental)
To confirm the accuracy of the optical model (i.e., an internal consistency check), we show that A
+ R + T = 1 at all wavelengths, angles, and polarization combinations (see Figure 4.5) where A, R, and T
are all independently determined from the calculated field profile:
Conventional
[𝐴] + [𝑅] + [𝑇] = [

𝑇𝑆
∙
1−𝑅𝑆 (𝑓|𝑟 𝑠 |2 +(1−𝑓)|𝑟 𝑝 |2 )

𝑝
𝑠
∑𝑚
𝑗=1 ∫(𝑓𝑄𝑗 + (1 − 𝑓)𝑄𝑗 )𝑑𝑥𝑗 ]

𝑇 2 (𝑓|𝑟 𝑠 |2 +(1−𝑓)|𝑟 𝑝 |2 )

(𝑓|𝑡 𝑠 |2 +(1−𝑓)|𝑡 𝑝 |2 )𝑇𝑆
𝑠 2
𝑝 2 ]
𝑆 (𝑓|𝑟 | +(1−𝑓)|𝑟 | )

𝑆
+ [𝑅𝑆 + 1−(𝑓|𝑟
𝑠 |2 +(1−𝑓)|𝑟 𝑝 |2 )𝑅 ] + [1−𝑅
𝑆

=1

(4.21)

Optically inverted
[𝐴] + [𝑅] + [𝑇] = [(1 +

𝑅𝑆 (𝑓|𝑡 𝑠 |2 +(1−𝑓)|𝑡 𝑝 |2 )
)∙
1−𝑅𝑆 (𝑓|𝑟 𝑠 |2 +(1−𝑓)|𝑟 𝑝 |2 )

+ [(𝑓|𝑟 𝑠 |2 + (1 − 𝑓)|𝑟 𝑝 |2 ) +

𝑝
𝑠
∑𝑚
𝑗=1 ∫(𝑓𝑄𝑗 + (1 − 𝑓)𝑄𝑗 )𝑑𝑥𝑗 ]

𝑅𝑆 (𝑓|𝑡 𝑠 |2 +(1−𝑓)|𝑡 𝑝 |2 )2
(𝑓|𝑡 𝑠 |2 +(1−𝑓)|𝑡 𝑝 |2 )𝑇𝑆
+
]
[
]
𝑠
2
𝑝
2
1−(𝑓|𝑟 | +(1−𝑓)|𝑟 | )𝑅𝑆
1−𝑅𝑆 (𝑓|𝑟 𝑠 |2 +(1−𝑓)|𝑟 𝑝 |2 )

62

= 1 (4.22)

Figure 4.5. Consistency check A + R + T = 1. Left, center, and right panels show consistency check for
different polarizations of light, and solid and dashed lines show it for normal and oblique incidence
respectively. Device structure: Glass Substrate / 1200 Å ITO / 200 Å MoO3 / 125 Å ClAlPc / 400 Å C60 /
80 Å BCP / 1200 Å ITO).
where m is the total number of layers, Qj is the power dissipated in the thin-film layer j, xj is the position in
layer j, f is the fraction of light that is s-polarized, rs,p are the Fresnel complex reflection coefficients for s𝑠,𝑝

and p-polarized light, ts,p are the Fresnel complex transmission coefficients for s- and p-polarized light, 𝑟0
𝑠,𝑝

and 𝑡0

are the Fresnel coefficients for the substrate-ambient interface, 𝑇𝑆 = (𝑓|𝑡0𝑠 |2 + (1 −

𝑝 2

𝑝 2

𝑓)|𝑡0 | ) 𝑞𝑠 ⁄𝑞0 in the conventional case and 𝑇𝑆 = (𝑓|𝑡0𝑠 |2 + (1 − 𝑓)|𝑡0 | ) 𝑞0 ⁄𝑞𝑠 in the optically inverted
𝑝 2

𝑝

𝑝

𝑝

𝑠
𝑠
𝑠
case, 𝑅𝑆 = (𝑓|𝑟0𝑠 |2 + (1 − 𝑓)|𝑟0 | ), 𝑟 𝑝 = 𝑆21 /𝑆11 , 𝑟 𝑠 = 𝑆21
/𝑆11
𝑡 𝑝 = 1/𝑆11 , 𝑡 𝑠 = 1/𝑆11
(as in Ref.

[122]), 𝑞𝑠 = 𝑛𝑆 cos𝜃𝑆 = √𝑛𝑠2 − 𝑛02 sin2 𝜃0, and 𝑞0 = 𝑛0 cos𝜃0. The Fresnel coefficients for the substrateambient interface for the case of conventional illumination are:
𝑟0𝑠 =
𝑡0𝑠 =
𝑝

𝑟0 =

𝑛0 cos(𝜃0 )−𝑛𝑠 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )
𝑛0 cos(𝜃0 )+𝑛𝑠 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )
2𝑛0 cos(𝜃0 )
𝑛0 cos(𝜃0 )+𝑛𝑠 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )
𝑛0 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )−𝑛𝑠 cos(𝜃0 )
𝑛0 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )+𝑛𝑠 cos(𝜃0 )

63

(4.23)
(4.24)

(4.25)

𝑝

𝑡0 =

2𝑛0 cos(𝜃0 )
𝑛0 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )+𝑛𝑠 cos(𝜃0 )

(4.26)

where 𝑛𝑠 and 𝑛0 are the refractive indices of the substrate and ambient, respectively. The Fresnel
coefficients substrate-ambient interface for the optically inverted case are:
𝑟0𝑠 =

𝑛𝑠 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )−𝑛0 cos(𝜃0 )

𝑡0𝑠 =

2𝑛𝑠 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )

𝑝

𝑟0 =
𝑝

𝑡0 =

𝑛𝑠 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )+𝑛0 cos(𝜃0 )

𝑛𝑠 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )+𝑛0 cos(𝜃0 )
𝑛𝑠 cos(𝜃0 )−𝑛0 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )
𝑛𝑠 cos(𝜃0 )+𝑛0 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )
2𝑛𝑠 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )
𝑛𝑠 cos(𝜃0 )+𝑛0 √1−(𝑛0 /𝑛𝑠 )2 sin2(𝜃0 )

(4.27)

(4.28)

(4.29)

(4.30)

Figure 4.6. Schematic showing how solar flux changes with incident angle. (a) Assuming one photon is
incident on area A0, the photon flux is 1/A0. (b) At incident angle θ, the area of illumination becomes A,
which is greater than A0 by a factor of 1/cos(θ).

64

Figure 4.7. Predicted and experimental JSC with respect to incident angle. (a) Experimental data
(squares) and corresponding simulation predictions (line) shown for normalized Jsc under conventional
illumination as a function of incident angle for the device with structure: Glass Substrate / 1200 Å ITO /
100 Å ZnO / 300 Å C60 / 150 Å ClAlPc / 1000 Å MoO3 / ITO 1300 Å. (b) Simulated Jsc as a function of
incident angle for five different top ITO layer thicknesses (x = 1000, 1200, 1400, 1600, 1800), where the
structure is as follows: Glass Substrate / 1200 Å ITO / 200 Å MoO3 / 125 Å ClAlPc / 400 Å C60 / 80 Å
BCP / (x) Å ITO).
We note that the performance under oblique illumination for all plots in this work is simulated
assuming 1 sun photon flux (100 mW/cm2) at all incident angles for comparability. However, the flux on
deployed PVs monotonically decreases with increasing incident angle (Figure 4.6) and varies widely with
PV tilt angle, orientation, time of day, and location. Internal quantum efficiency (IQE) is also calculated
from A as 𝐸𝑄𝐸 = 𝐴 ∗ 𝐼𝑄𝐸, assuming 100% charge transfer and charge collection efficiency for both IQE
and EQE.[156]
Transparent PHJ solar cells were first fabricated to measure EQE under oblique illumination and
further confirm the accuracy of the model. Thin film layers of ZnO, C60, chloroaluminum phthalocyanine
(ClAlPc), MoO3, and ITO were then deposited using thermal evaporation (for C60, ClAlPc, and MoO3) at a
rate of 1 Å/s or DC magnetron sputtering (for ZnO and ITO) at a rate of 0.05 Å/s. The full experimental
device stack was as follows: Glass Substrate / 1200 Å ITO / 100 Å ZnO / 300 Å C60 / 150 Å ClAlPc / 1000
Å MoO3 / ITO 1300 Å. The EQE spectrum was measured for this 60 mm2 device (20 mm x 3 mm) under

65

Figure 4.8. Example Jsc contour plot where ClAlPc and C60 thicknesses are varied to maximize 0° Jsc
for ITO thickness = 1000 Å in the conventional configuration.
varying degrees of incidence ranging from 0° to 80°. EQE was also predicted from the derived model, using
optical constants measured by variable angle spectroscopic ellipsometry (J. A. Woollam) and exciton
diffusion lengths of 8 nm and 20 nm for ClAlPc and C60, respectively. The EQE spectra were then integrated
with the solar spectrum to calculate Jsc, which is normalized to the 0° incidence Jsc and plotted in Figure
4.7(a) along with the Jsc predicted by the model, verifying good accuracy between the model and
experimental data. Using the transfer matrix model described and verified above, we investigate the
performance of the archetypal TPV bilayer ClAlPc-C60 PHJ solar cells described in Ref. [14]in various
configurations and orientations.

Here, we examine the conventional electrical configuration (Glass

Substrate / ITO 1200 Å / MoO3 200 Å / ClAlPc (x) Å / C60 (y) Å / bathocuproine (BCP) 80 Å / ITO (z) Å).
We find that by varying the thickness of one layer alone, e.g. the top ITO layer, the angular response of Jsc
can change from decreasing monotonically to exhibiting distinct angular maxima or reduced angular
performance roll-off (see Figure 4.7(b)). This highlights the important role that optical interference plays
not only on the normal incidence Jsc, as shown previously in TPVs,[14] but also the angular performance.
To study these effects we optimize the thicknesses of three key layers of the device (the donor, acceptor
and top ITO layers) simultaneously to maximize Jsc, which is proportional to power conversion efficiency
66

(𝑃𝐶𝐸 = 𝐽𝑆𝐶 𝑉𝑂𝐶 𝐹𝐹, where VOC is open circuit voltage and assumed not to vary over the optimization
thickness range), at various angles of incidence: x and y were allowed to vary from 100 to 800 Å, and z was
varied from 400 to 2000 Å (a range of ITO thicknesses for which fill factor FF varies by less than 5%[14]).
Optimum layer thicknesses were then obtained by generating contour plots of JSC versus x and y and finding
maximum Jsc in this three-parameter space (see Figure 4.8). The optimum layer thicknesses for maximum
0° Jsc, maximum 70° Jsc, and maximum Jsc,ave for the conventional and optically inverted configurations are
listed in Table 4.1, where 70° is the typical incident angle of light on a south-facing window at noon in the
summer for East Lansing, MI (43° N. latitude),[159] and Jsc,ave is the average of Jsc across all angles. The
angle of illumination to optimize Jsc will depend on the application: for building-integrated TPVs, the
performance under oblique illumination is more important than the performance at normal illumination
because light that is incident on the building-integrated TPV will consist of more direct sunlight (roughly
70% of the incident spectrum[160]), which is mostly oblique illumination, than diffuse sunlight (the roughly
30% remaining), which impinges upon the device at all angles.

Table 4.1. Layer thicknesses (Å) to maximize Jsc and Jsc* for 0°, 70°, or all angles of incidence
ITO
MoO3 ClAlPc C60
BCP
ITO
0° Conventional JSC
1200 200
125
400
80
1200
Modified JSC
125
225
1400
Optically
JSC
1200 200
150
275
80
400
inverted
Modified JSC
125
200
400
70° Conventional JSC
1200 200
125
475
80
1600
Modified JSC
125
250
1800
Optically
JSC
1200 200
150
250
80
400
inverted
Modified JSC
125
175
400
All angles Conventional JSC
1200 200
125
475
80
1200
Modified JSC
125
250
1600
Optically
JSC
1200 200
150
275
80
400
inverted
Modified JSC
125
175
400
Jsc* was modified from Jsc by multiplying it by charge collection efficiency, assuming a total charge
collection length LC0 of 1000 Å for the donor/acceptor (see Eq. 4.32).

67

Figure 4.9. Simulated normalized Jsc under 1 sun illumination as a function of angle of incidence for
two optimizations. (a) Devices with layer thicknesses optimized for maximum 0° Jsc for conventional case.
(b) Devices with layer thicknesses optimized for maximum 0° Jsc for each case. (c) Devices with layer
thicknesses optimized for 70° Jsc for each case. The normalization factor, 0° Jsc, is shown in the inset. Note
that the illumination flux is held constant at all angles for comparison; the flux will vary with angle
depending on geographical position, time of day, and orientation.

Figure 4.10. Simulated absorption, reflection, transmission and IQE of 735 nm light as a function of
incident angle in three optimizations. (a) Devices with layer thicknesses optimized for maximum 0° JSC
for conventional case. (b) Devices with layer thicknesses optimized for maximum 0° JSC for each case. (c)
Devices with layer thicknesses optimized for 70° JSC for each case.
In Figure 4.9, we plot relative Jsc as a function of incident angle, where the inset legend gives the
magnitude of Jsc at 0° used for normalization. We also plot A, R, and T of a single wavelength of light (735
nm) as functions of incident angle, where 735 nm is the absorption peak of ClAlPc, the largest contribution
to photocurrent among the active layers (Figure 4.10). In both figures, we include two approaches to
thickness optimization: the devices in the left panel of Figure 4.9 have layer thicknesses optimized for 0°
68

Jsc for the conventional configuration; devices in the center panel have thicknesses optimized for 0° Jsc for
their respective configurations; and the devices in the right panel have layer thicknesses optimized for 70°
Jsc for their respective configurations. Optimizing the thicknesses for the optically inverted configuration
improved its normal incidence Jsc by 15% (moving from left to center panel), and optimizing the thicknesses
for 70° Jsc improved the it by 5% for the conventional case and 0.1% for the optically inverted case (moving
from center to right panel). The normalized Jsc in Figure 4.9 decreases sharply at high angles of illumination
for all devices; however, optically inverted devices are shown to retain the normal incidence value over
greater ranges of angles. The absorption curves show a similar trend of dropping off more quickly in the
optimized conventional device. Losses to absorption at this wavelength (735 nm) come from increased
reflection. Therefore, we note again that minimizing losses from reflection over all wavelengths is
paramount for TPVs. The shift in reflection curves between the conventional and optically inverted devices
can be explained by the different Brewster angles for the air-glass interface and the air-ITO interface: 55°
and 62°, respectively. At angles lower than Brewster angle, the p-polarized component of light, which
makes up 50% of solar irradiance, has a transmission that increases with higher incidence angles until it
reaches 100% transmission at the Brewster angle, after which it begins to decrease, leading to a sharp
increase in total reflection (reflection due to both s- and p-polarized components of light).

This

phenomenon also explains the different angles at which Jsc begins to drop off in Figure 4.9 for the
conventional and optically inverted cases. Similar behavior around the Brewster angle has also been
reported for polymer OPVs.[161]
To gain insight as to why devices optimized for optical inversion have better normal incidence
performance and less-varying angular responses, we plotted the electric field profiles for Îť = 735 nm, the
peak absorbance of ClAlPc (Fig. 4.11). In the conventional illumination case, the top ITO layer was made
thicker (1200 Å) to move the shoulder of the standing wave profile (which is largest in the bottom ITO
layer where the photons first enter) closer to the donor-acceptor (D-A) interface. In the optically inverted
orientations, moving the peak of the profile to the D-A interface is not as important as reducing the distance

69

Figure 4.11. Simulated electric field profile for conventional and optically inverted ClAlPc-C60 TPVs.
Simulations were done for Îť = 735 nm, the peak absorbance of ClAlPc. 1 = MoO3, D = ClAlPc, A = C60,
and 2 = BCP. (a) Layer thicknesses optimized for illumination through substrate (b) Layer thicknesses
optimized for illumination through thin film stack.

Figure 4.12. Simulated electric field profile for normal and oblique incidence on ClAlPc-C60 TPVs.
Simulations were done for Îť = 735 nm, the peak absorbance of ClAlPc. 1 = MoO3, D = ClAlPc, A = C60,
and 2 = BCP. (a) Layer thicknesses optimized for illumination through substrate (b) Layer thicknesses
optimized for illumination through thin film stack.
70

from the point of light entry to the D-A interface, so the optimum top ITO layer in these cases was 200 Å.
These optically inverted devices have a less-varying angular response also due to the thin top ITO layer
(Fig. 4.12); the increase in path length at oblique incidence spreads out the standing wave profile in an
overall thicker device more than it does in the thinner devices optimized for optical inversion.
The TMM-based optimization for Jsc can be modified to account for electrical considerations by
including the thickness-dependence of the 1) donor-acceptor charge collection lengths, 2) resistance of the
electrode layer on the FF and 3) changes in the active layer bandgap.[162-164] While 3) is notably complex
and well beyond the scope of the present work, we note that for 2), the FF was shown to be independent of
the ITO cathode for thicknesses beyond 40 nm for small devices[14] and can be modeled for large area
scaling as:[66, 165]
𝐹𝐹(𝑑) = 𝐹𝐹(𝑅𝑆 = 0, 𝑅𝑃 = ∞)[1 −

𝐽𝑆𝐶 (𝑑)𝑅𝑆 (𝑑)
]
𝑉𝑂𝐶

(4.31)

where d is the total active layer thickness, Rp is the parallel (also called shunt) resistance, and the series
resistance, Rs, will be dependent on both the specific layer thickness and area. For 1), we can account for
the influence of active layer thickness on charge collection efficiency under zero bias, as:[166]
𝜂𝐶𝐶 (𝑑, 𝑉 = 0) = 𝐿𝐶0 ⁄𝑑 ∙ [1 − exp(− 𝑑⁄𝐿𝐶0 )]

(4.32)

where LC0 is the charge carrier collection length. For larger active layer thicknesses, the charge collection
efficiency will decrease, which will reduce the optimum layer thickness. We show an example comparing
optimizations with, and without, the charge collection correction in Table 4.1. Finally, we note that
complete optimization would also consider the position of the sun (changing incident angle) and flux
variations throughout the day for specific PV orientations/locations. Nonetheless, the demonstration of this
model provides the starting point for analyzing such performance and optimization.
4.5 Application for real-world solar irradiance data
Proof-of-concept simulations were performed to quantify the impact of improving the angular
performance of transparent solar cells.[148] Hourly sun position and solar irradiance data was collected
from the National Solar Radiation Database and used to calculate the global tilted irradiance by the Hay

71

model.[148] An unoptimized device (Device A) was compared with the device optimized for oblique
incidence in Ref. [167] (Device B) (Figure 4.13 (a)). Both devices were used in several simulated
configurations with different geographical locations, PV tilt angles, and PV facing direction. In Figure 4.13
(b), the yearly power output for a TPV in Lansing, MI is shown, normalized to the power output of a
horizontal (tilt angle β = 0°) TPV. The highest yearly power output was seen for a south-facing (azimuth
angle ι = 180°) TPV with a tilt angle of 30°. However, the vertically oriented TPV incorporated onto a
building window (β = 90°) had the lowest yearly power output across all PV facing directions. The yearly
power output was calculated for Device B and compared to Device A, with the improvement plotted in
Figure 4.13 (c). For a south-facing window (ι = 180°, β = 90°), the angular response of Device B increased
the yearly power output by 15%, and for a north-facing window (ι = 0° or 360°), the improvement was
closer to 17%.[148] These calculations show that the yearly power output of a solar cell can be increased
by up to 15% and up to 40% elsewhere by reducing the angular roll-off of performance with no change to
the normal incidence performance.
In summary, we extended the transfer matrix model to simulate the angular response of thin-film
solar cells and demonstrated self-consistency of A+R+T=1 across all wavelengths, polarizations of light and
angles of incidence. This model shows the range of angle-dependent performances in transparent planar
heterojunction ClAlPc-C60 solar cells under conventional illumination and optical inversion. Proof-of-

Figure 4.13. Effect of improved angular dependence on yearly PV power output. (a) Normalized
responsivity for Device A (unoptimized) and Device B (optimized for improved angular roll-off). (b) Yearly
power output of Device A in Lansing, MI with tilt angle β, where β = 0° is horizontal, and azimuthal angle
ι, where ι = 180° is south-facing. (c) Improvement in yearly power output in (b) by using Device B. All
figures reproduced with permission from Ref. [148].

72

principle optimizations suggest that optically inverted structures can potentially provide enhanced angular
performance and therefore have the potential to increase maximum power output around the building
envelope. This work opens the door for many important advances and designs of thin-film transparent (and
opaque) solar cells for installation in the building environment, facilitates subcell photocurrent matching in
multijunction solar cells, and demonstrates the additional factors that should be considered for complete
optimization in fixed deployment configurations.

73

Chapter 5 – NIR-Absorbing Organic Heptamethine Salts for TPV
Only a few organic photovoltaics and photodetectors have demonstrated photoresponse past 900
nm, an underutilized spectral region for tandem solar cells, transparent solar cells and infrared
photodetectors. In this chapter, heptamethine salts with selective deep near-infrared (NIR) photoresponse
are demonstrated with EQE cutoffs at Îť = 1400 nm or 1600 nm, the deepest photocurrent generated by
organic small molecules. Anion exchange is shown to deepen frontier orbital levels with minimal changes
in absorption properties, leading to decreases in dark current, increases in open-circuit voltage (approaching
excitonic limits), and increases in specific detectivity. Balancing exciton binding energy and charge transfer
efficiency is shown to be key for enhancing the performance of very small bandgap NIR-absorbing devices.
Organic heptamethine salts represent a pathway to inexpensive infrared solar cells and detectors and expand
the catalog of existing donor materials for transparent and multijunction solar cells.
5.1 Polymethine PV background
Polymethines are a class of molecules consisting of an odd number of methine (C-H) groups that
are connected by alternating single and double bonds. Trimethine, pentamethine, and heptamethines are
commonly used in dyes and bio-imaging.[168, 169] The smaller polymethines (trimethine, pentamethine)
typically have absorption in the visible spectrum due to a smaller degree of conjugation, and the larger
polymethines (heptamethine, nonamethine) have smaller bandgaps and absorption in the NIR.[168]
Cationic polymethines can be synthesized by quarternization reactions of the nitrogen in the indole
group. One of the five valence electrons on the nitrogen is shared with another carbon atom leaving it with
a +1 charge. The aromatic rings on either side of the polymethine backbone can also play a role on the
absorption profile of the dye. Functional groups can be added to the molecular structure for increased
solubility in water (sulfonate and carboxylate groups) or in organic solvents (long alkyl chains). The center
position of the heptamethine backbone also plays an important role. The electron withdrawing / donating
group (Cl) on the center carbon atom can greatly affect the absorption profile by tuning the degree of

74

conjugation or bond length alteration. The three center carbon atoms may also be part of a cyclopentene or
cyclohexene moiety, which can stabilize the structure via rigidity, and change the energies of vibronic states.
A review of organic salts primarily featuring polymethines was recently published [170] Here, we
briefly summarize the introduction of these materials in PV applications. Polymethine PVs were first
demonstrated using visibly absorbing trimethine and pentamethine salts (400-600 nm). A trimethine was
used with a ClO4- anion by Meng et al in 2003 for an OPV with 0.25 VOC and 0.1 % PCE using C60 as the
acceptor.[171] Several studies have since explored different hole transport layers, such as PEDOT:PSS
(leading to a much higher VOC),[172] TiOx (leading to smoother films and higher PCEs of 3.7%),[173] and
MoO3 (PCE = 2.9%).[174] Doping was also studied by air/oxygen for trimethine OPVs.[175]
Polymethines with greater degrees of conjugation (heptamethines and nonamethines) were used for
near-infrared OPVs in 2009 when Bouit et al used a heptamethine salt as a bulk heterojunction donor with
PCBM as an acceptor with 0.4% PCE.[176] VĂŠron et al used a heptamethine salt to make higher efficiency
NIR OPVs as well as transparent OPV.[36] However, very little was still known about the tunability of
these compounds and it was often assumed that the anion played very little role. In the sections below, we
challenge this notion and show that the anion can indeed play a significant role in the tunability of these
compounds.
5.2 Counterion exchange effect on open-circuit voltage
Iodide (I-) is a common counterion after the quarternization reaction is completed using ammonium
iodide (NH4I). While the cation is the photoactive component, the counterion affects how the molecule
aggregates in solution and thin films. For example, the solid packing was studied using single crystal
diffraction for a heptamethine with PF6- anion and Δ-TRISPHAT anion.[36] A higher symmetry crystal
structure was obtained for heptamethine coordinated with the Δ-TRISPHAT anion (monoclinic) than for
the PF6- anion (triclinic).[36]
A study on the effect of counterion exchange on heptamethine OPV performance was carried out
in our lab.[31] In the study, the heptamethine Cy and various counterions were investigated as shown in
Figure 5.1. A variety of size counterion was tested, ranging from monoatomic iodide (I-) to polyatomic
75

tetrakis pentafluorophenyl borate (TPFB-). Changing the counterion did not affect the thin film absorption
or the optical bandgap (Figure 5.2(a)), but the frontier energy levels (HOMO and LUMO) of the molecule
were affected. The change in energy levels could be seen from the different VOC acquired by JV testing
(Figure 5.2(b)). The device VOC increases from 0.4 V (Cy-PF6) to 0.7 V (Cy-TPFB), with no change in
optical bandgap. VOC is difficult to change without synthesizing new molecules, making this result
particularly exciting. The magnitude of the VOC change also resulted in a doubling of the device PCE, from
0.9% (Cy-PF6) to 2.0% (Cy-TPFB). As mentioned previously in Chapter 1, the device VOC depends on the
interface gap, which is the difference between the HOMO of the donor and the LUMO of the acceptor.
Because the device VOC increased (donor HOMO shifted deeper) with no change in the optical bandgap
(energy between HOMO and LUMO of the donor stayed the same), both the HOMO and LUMO levels of
the donor shifted deeper by approximately 0.3 eV when going from the PF6 to the TPFB anion.

Figure 5.1. Molecular structure of Cy cation and anions used in Ref. [31]. (a) Cy and (b) anions, (1) I-,
(2) PF6-, (3) SbF6-, (4) Δ-TRIS–, and (5) TPFB–.

76

Figure 5.2. Absorption and JV for Cy salts. (a) Absorption spectra of spincoated Cy films (>12 nm) with
different counterions. Solution absorption measured for methanol. (b) JV curve for OPV devices with Cy
films with different counterions.

Figure 5.3. Cy-PF6 and Cy-TPFB alloying. (a) VOC as a function of molar fraction of Cy-TPFB, showing
a roughly linear relationship. (b) Proposed energy schematic illustrating DOS of pure anions and mixed.
Interface gap IG of Cy/C60 linearly varies with Cy-TPFB fraction.
Blending two heptamethines with different counterions (i.e., fractional alloying) is interestingly
found to create intermediary energy level shifts resulting. in linear modulation of VOC with molar
concentration (Figure 5.3). This allowed for the fine tuning of the VOC in devices, which represents a
powerful tool for finely tuning the energy levels of a molecule without reverting to more complicated
synthetic methods (the one-step counterion exchange synthesis is described in Chapter 3). Ultimately, this
energy level tuning can be applied in a range of applications where energy level requirements are stringent
for full optimization: 1) near infrared / transparent OPVs 2) OLEDs, 3) photocatalysts, and 4) and
diagnostics and therapeutics (see Section 7.5).
77

5.3 Deep NIR (1600 nm) absorption
Organic semiconductors that absorb in the NIR, i.e. at Îť > 800 nm, are promising for applications
in broadband and transparent solar cells.[14, 22] Organic compounds with NIR photovoltaic response have
been demonstrated including cyanines,[36, 163, 177-179] carbon nanotubes,[180] and polymers.[97, 181]
However, EQEs in these studies have only extended to 1100 nm for SnNcCl2 and 1450 nm for carbon
nanotubes. Design strategies for redshifting the IR absorption of organic molecules have included
increasing the conjugation[163, 182] and modifying the ligand structures to affect aggregation, crystal
structure, and intermolecular proximities. However, once molecules are designed and integrated into
optoelectronic devices, their performance typically suffers from arbitrary energy level alignments, resulting
in lower-than-ideal open-circuit voltages, low carrier mobilities and diffusion lengths, and limited
absorbance past 1000 nm. In this work, we synthesize a new series of heptamethine salts[183] with highest
occupied molecular orbital (HOMO) levels that can be tuned by varying the anion electronegativity.[31, 36]
These organic salts are used in photovoltaic and photodetector cells to demonstrate photoresponse at deep
NIR wavelengths and open-circuit voltages nearing their excitonic limit. Using optical modeling and opencircuit voltage tuning[184, 185] we identify limiting factors for performance and strategies for performance
enhancement.
Heptamethine salts Cy1 (1-Butyl-2-(2-[3-[2-(1-butyl-1H-benzo[cd]indol-2-ylidene)-ethylidene]-2diphenylamino-cyclopent-1-enyl]-vinyl)-benzo[cd]indolium, Îťmax = 996 nm) and Cy2 (1-Butyl-2-(2-[3-[2(1-butyl-1H-benzo[cd]indol-2-ylidene)-ethylidene]-2-phenyl-cyclopent-1-enyl]-vinyl)-benzo[cd]indolium,
Îťmax

=

1024

nm)

coordinated

with

the

counterions

tetrafluoroborate

(BF4-)

and

tetrakis(pentafluorophenyl)borate (TPFB-) are shown in Figure 5.4(a). We focus on these molecules for
their absorption ranges that extend to 1400 nm and 1600 nm for cations Cy1 and Cy2 respectively (Figure
5.4(b)). Figure 5.4(c) shows a summary of the m/z synthesis verification for the cation and anion masses.
In previous studies, weakly coordinating anions like TPFB have been shown to modulate the frontier energy
levels of organic cations used as donors in photovoltaic configurations, thereby increasing the open circuit
voltage (VOC) with little or no impact on the bandgap or absorption range.[31, 36, 176]
78

.

Figure 5.4. Cy1 and Cy2 salt structures, absorption, and mass spectra.(a) Chemical structures of the
heptamethine salt cations (Cy1 and Cy2) and anions (BF4- and tetrakis pentafluorophenylborate, TPFB-).
(b) Normalized thin film absorption (1-Transmission) of each organic salt. (c) Summary of high-resolution
mass spectrometry m/z spectra for cations and anions used in this study. The multiple peaks and their
relative heights represent the isotopic abundances of the constituent elements in the molecule.

Figure 5.5. Device architecture, JV and EQE for Cy1 and Cy2. (a) Device architecture used for
photovoltaic and photodetector structures. (b) JV and (c) EQE curves for devices with salts with thicknesses
of 12 nm (inset highlights the NIR EQE).
79

Table 5.1. Device parameters and molecular properties for each salt

FF

Abs. edge
[nm]

NIR EQE
peak
[%]

D* peak
[Jones]

0.13 Âą 0.01

0.34 Âą 0.01

1440

2.1

3.7 x 10

1.9 Âą 0.2

0.33 Âą 0.01

0.49 Âą 0.01

1460

1.1

5.3 x 10

Cy2-BF4

3.4 Âą 0.3

0.15 Âą 0.01

0.42 Âą 0.05

1590

1.4

7.0 x 10

Cy2-TPFB

1.7 Âą 0.2

0.25 Âą 0.01

0.40 Âą 0.04

1500

0.8

1.7 x 10

JSC
[mA cm-2]

VOC
[V]

Cy1-BF4

3.4 Âą 0.3

Cy1-TPFB

Salt

J0
[ÂľA cm-2]

9

48

10

0.014

9

7.0

10

1.1

Solar cell devices with the structure indium tin oxide (ITO)/10 nm MoO3/t nm salt/40 nm C60/7.5
nm bathocuproine (BCP)/80 nm Ag were prepared using the four salts as a function of thickness (Figure
5.5(a)). Donor layers of each organic salt were spin-coated from N,N-dimethylformamide (DMF) under
nitrogen while other layers were thermally deposited in vacuum. The thickness for each salt was controlled
by varying the solution concentration. For comparison purposes, the JV and EQE for devices with similar
salt thicknesses (12Âą1 nm) are plotted in Figure 5.5(b-c) and average performance metrics are shown in
Table 5.1. The fill factors (FF) for these devices, 0.3-0.5, are slightly low due to decreased shunt resistances
from low-bandgap materials and series resistance from a potential interface barrier between the donor and
MoO3. The exchange of BF4 for TPFB nearly doubles the VOC from 0.13 to 0.33 V for cation Cy1 and 0.17
to 0.25 V for cation Cy2. This enhancement in the voltage is due to the shift in energy levels (Figure 5.6)
and increased interface gap (Figure 5.7(a)). However, this exchange reduces the NIR EQE peak by more
than 50% due to the substantial decrease in the donor-acceptor lowest unoccupied molecular orbital level
offset (ΔLUMO). To understand the effect of gradual shifts in interface gap on EQE, alloyed blends of Cy1or Cy2-BF4 with varying molar ratios of Cy1- or Cy2-TPFB were prepared. The VOC and EQE trends as a
function of TPFB molar fraction are plotted in Figure 5.8(a).

80

Figure 5.6 Proposed energy level schematic for Cy1, Cy2, and C60. Work functions and HOMO levels
were extracted from ultraviolet photoelectron spectroscopy. LUMO levels were estimated by adding the
optical bandgap and calculated exciton binding energies to the HOMO.

Figure 5.7. Interface gap change in Cy1 salt and band bending in Cy2-TPFB. (a) Energy schematic
illustrating the deepening of the LUMO level and increase in interface gap (IG) after counterion exchange
from a small anion BF4 to the bulky and weakly-coordinating anion TPFB-. (b) Schematic D-A band
structure as a function of donor thickness (tD).
The thickness trends of the pure salts are plotted in Figure 5.8(b), where VOC either remains flat
(Cy1-BF4 and Cy2-BF4) or decreases (Cy1-TPFB and Cy2-TPFB) with increasing thickness and EQE
monotonically increases. In general, the VOC is found to be independent of cyanine salt thickness in the 515 nm range. In some cases, open-circuit voltages for OPVs increase with thickness as parallel shunting
pathways are eliminated by the formation of more complete films.[186] In the case of Cy1- and Cy2-TPFB,
however, VOC shows a modest decrease of 20% over the thickness range of 4 to 15 nm. Decreases in VOC

81

Figure 5.8. VOC and NIR EQE as a function of thickness for Cy1 and Cy2. (a) VOC (left axis) and EQE
(right axis) as functions of mole fraction TPFB for blends of (black) Cy1-BF4 and Cy1-TPFB and (blue)
Cy2-BF4 and Cy2-TPFB at tD = 6 nm. (b) EQE and VOC as a function of donor thickness. Wavelength of
EQE is shown in inset: 1200 nm for Cy1-BF4 and Cy1-TPFB and 1350 nm for Cy2-BF4 and Cy2-TPFB.

with increasing thickness have been attributed to (1) increased recombination due to presence of disorderinduced gap tail states,[187] (2) increased recombination due to electric field profile broadening[188] and
(3) shifts in the interface gap due to band bending (Figure 5.7(b)).[189] Mechanisms (1) and (2) are unlikely
due to the small thickness range (1 nm) over which the voltage drop occurs; thus, the VOC decrease is most
likely due to incomplete band bending in Cy1- and Cy2-TPFB devices as a function of thickness. In contrast,
devices with Cy1- and Cy2-BF4 have no thickness dependent photovoltage and therefore likely have smaller
depletion widths stemming from either larger carrier densities or smaller dielectric constants.

82

Quantum efficiencies past 1000 nm have been limited in magnitude to < 15% even for many
quantum dot systems.[163, 190] To identify the limiting factors in our NIR EQE, we examine the
component efficiencies: ΡA (absorption), ΡED (exciton diffusion), ΡCT (charge transfer), ΡCD (charge
dissociation) and ΡCC (charge collection). Through exciton diffusion and optical interference modeling,[152,
167] EQE curves were well fit for effective exciton diffusion lengths, which were calculated assuming 100%
charge transfer, charge dissociation, and charge collection efficiencies. From this analysis, we find that the
effective diffusion lengths in these four salts are all around 0.5-1 nm due to the modest EQEs. However,
we also find that absorption profiles already reach 70% at the peak wavelength for Cy2-TPFB films that
are only 25 nm thick, suggesting that these devices are not limited by absorption. With the extracted
diffusion lengths, the optical interference model predicts that the EQE should decrease for all the tested
salts with increasing thickness due to the inability of excitons to diffuse to the dissociating interface. This
predicted trend of decreasing EQE is indeed seen experimentally in other larger gap cyanine salt devices.[31,
188] However, this behavior contrasts the experimental trends that show EQE monotonically increasing
for donor layer thicknesses past 25 nm. This suggests that the intrinsic diffusion length is in fact longer than
0.5-1 nm and the EQE of these salts is instead limited by charge transfer, charge dissociation, or charge
collection efficiency, at least one of which should not be modeled as 100%.
While there is not a clear method to directly distinguish between the component efficiencies (charge
transfer, dissociation, and collection), we can infer insight about charge collection from other measurements.
For example, since the experimental C60 EQE peak (Îť = 430 nm) does not decrease with increasing salt
thickness and is similar in magnitude to other salt based OPVs with larger bandgaps,[31, 36] this implies
that hole collection from excitons originating on C60 (which still have to transport through the donor salt)
is not a limiting factor. Thus, devices are most likely limited by charge transfer or dissociation efficiency
because of the tradeoff between the lowest unoccupied molecular orbitals of the donor and acceptor and the
exciton binding energy.
To estimate the exciton binding energy, we turn to anion mixing experiments and look for the
emergence of sharp cutoffs in the EQE (Figure 5.8(a)). Indeed, while there is a linear variation in the VOC
83

that stems from a monotonic modulation of interface gap recombination, there is a sharp EQE cutoff at a
molar fraction of 10% TPFB, suggesting that there is just enough energy (ΔLUMO) available at that
concentration to efficiently overcome the exciton binding energy. The remaining quantum efficiency
beyond this concentration likely stems from a combination of field- and temperature-driven dissociation.[24]
We can then estimate the energy available for exciton dissociation by subtracting the interface gap
(calculated from the VOC)[66] from the optical bandgap, yielding exciton binding energies of ~0.55 eV for
Cy1 and ~0.40 eV for Cy2, which is close to other reported values in organic molecules (0.2 to 1.4 eV).[191193] These exciton binding energies make up roughly 50% of the optical bandgap (~ 0.8 eV), limiting the
interface gap (and therefore VOC) to modest values at which efficient exciton dissociation can still take place,
despite the ability to achieve higher VOCs. Moving forward, several strategies can be explored to decrease
the exciton binding energy. For example, molecular modifications can be designed to enhance the
delocalization of the electron/hole orbitals to increase the exciton radius,[194] e.g. via central methine
substitution.[183] Another design strategy involves the coupling of smaller solubilizing groups or anions
that allow for denser packing to increase the dielectric constant.[192, 195, 196] Thus, this presents an
interesting design challenge for the future optimization of very small bandgap organic photoactive devices.
To understand the ultimate potential of these material sets in photovoltaic applications, we modeled the
EQE of a device with a 100 nm thick Cy2-TPFB layer having an exciton diffusion length of 100 nm, a 20
nm thick C60 layer, and charge transfer and charge collection efficiencies approaching 100%. Such a device
would have an EQE of about 70-80% with a JSC on the order of 25 mA/cm2, and could be realized using a
bulk heterojunction architecture and optimized energy level tuning. FF values can be increased from 0.3 to
0.65 (achievable for most organic systems) by optimizing the interface energetics through the selection of
different transport layers or the modification of solvent and/or spincoating conditions. Combined with a
slightly improved VOC of 0.55 V, which is around the Shockley-Queisser excitonic limit, ideal devices
would be around 10% efficient with high transparency and would be well suited for multijunction cells with
complementary absorption.

84

Figure 5.9. Specific detectivity D* spectra for each deep NIR salt (Cy1 and Cy2).
These salt-based devices are also shown to be viable for near infrared photodetectors.[197]
Photodetector devices were fabricated with the same photovoltaic structure as those above. Specific
detectivity (D*) curves for each salt are plotted in Figure 5.9, where D* is proportional to the EQE and
inversely proportional to the differential resistance at zero bias.[198] Specific detectivity D* (cm Hz1/2 W1

) was calculated based on JV measurements at short circuit (V = 0). D* is obtained from:
𝐷 ∗ = 𝑅 √𝐴 𝑆𝑁 −1

(5.1)

where R is responsivity in A/W, A is device area in cm2, and SN-1 is current spectral noise density in A Hz1/2

. At room temperature and 0 V, the noise is dominated by thermal (Johnson-Nyquist) noise ST (A Hz-1/2),

which is estimated as:
4𝑘𝑇

𝑆𝑇 = √ 𝑅

𝐷

(5.2)

where k is the Boltzmann constant (J K-1), T is temperature (K), and RD is the differential resistance of a
solar cell in the dark at zero bias. Calculated D*s are comparable to those of the limited reports in other
organic systems[182, 198-200] and reach values of 6.5 x 1010 Jones at Îť = 1140 nm and 1.7 x 1010 Jones at
Îť = 1390 nm for Cy1 and Cy2 salts respectively. This compares well to the very limited reports in organic
systems, such as porphyrin tapes (2.3 x 1010 Jones at Îť = 1350 nm)[182] and inorganic carbon nanotubes
(8.8 x 1011 Jones at Îť = 1350 nm).[198] Exchanging the anion from Cy1-BF4 to Cy1-TPFB increases the

85

detectivity by an order of magnitude from 3.7 x 109 to 5.3 x 1010 Jones, largely due to lower noise currents
for devices with the TPFB anion. Compared to other organic systems, heptamethine salts have readily
tunable properties via counterion or ligand exchange in addition to being easy to synthesize and fabricate.
Moreover, they exhibit both broader and wavelength-specific photoresponsivity, which is promising for a
range of applications in the near-infrared and visibly transparent photodetectors.
5.4 TPV demonstrations with organic salts
Because of the strong NIR absorption of the heptamethine salts we studied (Cy, Cy1, and Cy2), we
decided to explore their use in transparent solar cells.[31] We began by performing optical simulations to
optimize the thicknesses of the Ag / Alq3 top electrode (Figure 5.10). The optimum combination of JSC and
AVT was found for thin layers of Ag (< 80 Å) and thick layers of Alq3 (600 Å). Thicker layers of Alq3 lead
to both higher transparency and performance, pointing to the importance of managing electric fields and
the shortcomings of only considering Beer Lambert’s law. The contour map was experimentally confirmed

Figure 5.10. Contour map of predicted JSC and AVT for TPV with Cy-TPFB. Device structure: 100 Å
MoO3 / 120 Å Cy-TPFB / 200 Å C70 / 75 Å BCP / x Ag / y Alq3. Black squares indicate experimental values.

86

Figure 5.11. Experimental optimization of Ag/Alq3 electrode for TPV with Cy-TPFB. (a) JV, (b) EQE
and (c) Transmission of TPVs with Cy-TPFB and 300 Å C70 with different thicknesses of Ag and Alq3
shown in legend (Å).

Figure 5.12. TPV demonstration with Cy-TPFB: JV, EQE, and transmission. (left) JV and (middle)
EQE of Cy-TPFB in transparent and opaque devices with peak EQEs of 20 % and 25% respectively. (right)
Transmission of the corresponding device stack with one NIR reflector attached (AVT = 50%).
using several points, shown as black boxes in Figure 5.10; the JV, EQE, and transmission curves for these
devices are shown in Figure 5.11. Thicker layers of Ag (50, 80, and 120 Å were tried) yielded higher JSCs,
and thicker layers of Alq3 yielded more transparent devices (compare 80/300 Ag/Alq3 with 80/600 Ag/Alq3),
confirming both major trends shown by the optical interference calculations in Figure 5.10.
Shown in Figure 5.12 are transparent and opaque planar heterojunction devices based on Cy-TPFB
with absorption in the 800-1000 nm range. The optimal opaque device structure was 10 nm MoO3, 10-15
nm donor, 50 nm C70, 7.5 nm bathocuproine (BCP), and 80 nm Ag. An efficiency of 2.7% was achieved
with 25% external quantum efficiency between 800-1000 nm. For transparent devices, we replaced the Ag
layer with a layer comprised of 8 nm Ag and 70 nm Alq3 (optimized by optical interference simulations)
and made the C70 layer thinner (changed from 50 to 30 nm) to improve transmission. The highest

87

performance transparent device reached an efficiency of 2.0%, peak EQE of 20% between 800-1000 nm,
and AVT = 50% with a NIR reflector attached to the glass substrate.
NIR-absorbing organic salts have the potential to enable highly efficient, tunable, and scalable
TPVs. They have strong absorption in the near infrared, and their frontier energy levels can be easily shifted
by exchanging the counterion. Counterion alloying can be done to meet stringent energy level requirements
in TPVs and other applications such as OLEDs and cancer theranostics. Organic salts can also be deposited
using solution processing methods, allowing TPV device fabrication to be scaled up. Increasing the EQE
of these devices is a key challenge this material set faces; likely, a bulk heterojunction approach will need
to be taken (blending Cy-TPFB with PCBM, for example). Using organic salts in a multijunction device
can reduce thermalization losses; this approach will be discussed in the next chapter.

88

Chapter 6 – Tandem TPV Optimization
6.1 Tandem TPV background
In traditional tandem OPVs, complementary absorbing subcells are used to capture low/high energy
photons with better matched low/high bandgaps to reduce thermalization losses. In semi-transparent tandem
OPVs (5-8% PCE, Table 6.1), complementary absorbing subcells are also used, however, more emphasis
is placed on absorption in the NIR. For example, an NIR absorbing (600-850 nm) subcell composed of 2{[7-(5-N,N-ditolylaminothiophen-2-yl)-2,1,3-benzothiadiazol-4-yl]methylenemalononitrile

(DTDCTB)

and C60 was combined with a visible-absorbing subcell composed of tetraphenyldibenzoperiflanthene (DBP)
and C70 for a semi-transparent tandem device with 5.3% PCE and 31% AVT.[17] In another study, two
different NIR-absorbing polymers were used in a tandem TPV: PBDTT-FDPP-C12 (film absorption max
~750 nm) and PBDTT-SeDPP (film absorption max ~800nm).[23] Using PC61BM in both subcells resulted
in a greenish tint, whereas substituting PC71BM for PC61BM in one of the subcells resulted in a neutral
color due to the redshifted absorption of PC71BM.[23] The use of PC61BM in one subcell and PC71BM in
the other has been used in several tandem TPV studies since PC71BM or C70 can be used to obtain devices
with higher CRI and PCE.[27, 29] Indeed, the highest combination of PCE and AVT for semi-transparent
OPV to date uses PSEHTT:IC60BA and PBDTT-DPP:PC71BM, achieving 8.0% PCE with 41.7% AVT.[29]
6.2 Simulations and contour mapping
Predicting JSC using optical interference modeling becomes especially important for tandem OPVs
because photocurrent matching between subcells is essential for simple series-integrated structures and
experimental optimizations are difficult to perform due to large number of variables and thicknesses. The
overall performance is very sensitive to the thicknesses of the interconnect layers and the electrode
thicknesses, which can all affect PCE, AVT and CRI differently. Example modeling results for a two-cell
tandem incorporating bulk-heterojunction subcells of ClAlPc and tin phthalocyanine (SnPc) are shown in

89

Table 6.1. Selected tandem semi-transparent OPV performance metrics

Front Subcell

JSC
VOC (mA FF
(V) cm-2)
DCV6T:C60
1.54 5.2 0.61
DBP:C70
1.70 6.2 0.51
PBDTT-SeDPP: PC71BM 1.47 8.4 0.59
Back Subcell

F4-ZnPc:C60
DTDCTB:C60
PBDTT-FDPP-C12:
PC61BM
PIDT-PhanQ: PC61BM PCPDTFBT:PC71BM
1.70 5.81 0.67
PBDTT-FDPP-C12:
PBDTT-SeDPP: PC61BM 1.46 7.2 0.61
PC61BM
PSEHTT:IC60BA
PBDTT-DPP:PC71BM
1.62 7.62 0.64
*
AVT recalculated using transmission curves provided in references

Ρ
(%)
4.9
5.3
7.3

Ρ (%) *
AVT AVT(frac.) Ref.
(%)
(%)
19.8*
0.97
[10]
31.8*
1.69
[17]
34.0*
2.48
[23]

6.7
6.4

38.6*
47.4*

2.59
3.03

[27]
[23]

8.0

41.7*

3.34

[29]

Figure 6.1. Examples of managing electric field profiles in TPV. (a) Optical field profile management
for a transparent semiconductor-metal-semiconductor electrode, of an STPV adapted from Ref. [154]. (b)
Optical field profile simulation for the (LiF/MoO3)6 electrode of an STPV at different wavelengths,
reproduced from Ref. [155]. (c) Example tandem TPV optimization, where the thickness of the interconnect
layer and the top ITO cathode are varied to optimize PCE, AVT, or CRI.
Figure 6.1 (c) as a function of the interconnect and top transparent cathode thickness. This model is partially
optimized for two of the eight key-parameters and indicates that an efficiency of > 6% with a transparency >
70% and CRI of ~85 for a two-cell junction is achievable. These simulations demonstrate the complex
multidimensional optimization landscape that can emerge in these systems. Meiss also used the transfer
matrix method to calculate the photocurrent of the limiting subcell for different thicknesses of the thin metal
/ semiconductor electrode (Ag / Alq3) and demonstrated a tandem STPV with PCE of 4.9% and AVT of
19.8%.[10]
90

6.3 Experimental results
Given the importance of the tandem structure, we explored a range of subcell configurations for
TPVs. The first architecture we explored used a subcell of ClAlPc/C60 and SnPc/C60, but of which contain
donors that selectively harvest UV/NIR light ideal for TPV and transparent multijunctions ClAlPc has been
demonstrated as an excellent candidate for TPVs,[14] and SnPc had been demonstrated as a good NIR
solar cell and TPVs.[163, 201] SnPc and ClAlPc can both be thermally evaporated, which is advantageous
for consistent processing in this study. Also, SnPc has deeper NIR absorption than ClAlPc by about 150
nm, so the two subcells can have complementary NIR absorption. The tandem device architecture is shown
in Figure 6.2(a), where ClAlPc subcell was deposited first followed by the SnPc subcell. In Figure 6.2(b),
the JV curves for the opaque devices are shown. The standalone “subcells” are devices made with ITO /
MoO3 / ClAlPc or Cy-TPFB / C60 / BCP / Ag. The ClAlPc subcell device has a much higher VOC (0.74 V)
than the SnPc device (0.38 V), which is not surprising consider the smaller bandgap of SnPc. The VOC of
the tandem device (1.03 V) was almost equal to the sum of the VOCs of the two subcells (0.74 + 0.38 = 1.12
V), indicating good balance of charge recombination in the interconnecting layers. However, the JSC of the
tandem device (0.026 mA mm-2) is less than half of either subcell JSC (~0.06 mA mm-2). While the overall
PCE of the tandem device (1.4%) was lower than that of the single junction ClAlPc device (2.4%), this

Figure 6.2. ClAlPc and SnPc tandem device. (a) Device architecture of ClAlPc/C60 – SnPc/C60 tandem
OPV. (b) JV of opaque devices with silver top electrode. JV is shown for tandem devices and ClAlPc/C60
and SnPc/C60 subcell devices. (c) JV of the tandem device with ITO top electrode.

91

architecture still shows promise considering the high voltages achieved. Photocurrent matching is the key
challenge for this architecture, with no improvement seen by changing either SnPc or ClAlPc thicknesses.
In Figure 6.2(c), the top silver electrode of the tandem device was replaced with 1000 Å of transparent
sputtered ITO to make a transparent tandem OPV. These transparent tandem devices had a very low VOC
(0.49 V) compared to the opaque version (1.03 V), possibly due to the sputtering process of ITO causing
partial shorts within the device stack. The PCE was lower by a factor of 4 (1.4% vs. 0.3%), so clearly more
optimization is needed to successfully replace silver with ITO for transparent tandem devices.
We further explored another potential subcell based on the Cy-TPFB salt. Because of the higher
VOC of Cy-TPFB devices (0.7 V) compared to SnPc devices (0.4 V), SnPc was replaced with Cy-TPFB in
the tandem architecture optimization. This tandem structure is shown in Figure 6.3(a). The Cy-TPFB
subcell was deposited first to avoid damaging other layers with the spincoating process. Control subcell
devices with ClAlPc and Cy-TPFB were also made with the structure ITO / MoO3 / ClAlPc or Cy-TPFB /
C60 / BCP/ Ag. As seen in Figure 6.3(b), the ClAlPc and Cy-TPFB control devices have similar values for
VOC: 0.77 and 0.69 V, respectively. The VOC of the tandem device is 1.38 V, which is close to the sum of
the subcell VOCs (0.77 + 0.69 = 1.46 V). This VOC is significantly higher than the ClAlPc-SnPc tandem
device (1.03 V). The JSC of the ClAlPc and Cy-TPFB tandem device (0.033 mA mm-2) is also closer to the
JSC of the subcells (0.056 and 0.054 mA mm-2 for ClAlPc and Cy-TPFB, respectively) than the ClAlPc and
SnPc tandem device. As a result, the PCE of the tandem device (2.4%) was higher than that of the control
subcell devices (2.1% for both ClAlPc and Cy-TPFB).
Optically biased EQE measurements were performed to determine which subcell was limiting the
photocurrent. The experimental method was covered in Chapter 2; only brief highlights will be covered
here. In Figure 6.3(c), the EQE spectra of the single-junction, control devices are shown. ClAlPc and CyTPFB have noticeably different absorption in the NIR, with the absorption peak of ClAlPc being at 735 nm
and the absorption peak of Cy-TPFB being at 880 nm. Because there are regions where the absorption
spectra do not overlap, each subcell can be illuminated separately: the ClAlPc subcell can be individually
illuminated with 735 nm, and the Cy-TPFB subcell can be targeted with 940 nm. The EQE of the tandem
92

Figure 6.3. ClAlPc and Cy-TPFB tandem devices. (a) Device structure of tandem device with CyTPFB/C60 and ClAlPc/C60. (b) JV of a ClAlPc/C60 device, a Cy-TPFB/C60 device, and three tandem devices
with different concentration solutions used for Cy-TPFB. (c) EQE of ClAlPc/C60 and Cy-TPFB/C60 devices.
(d) EQE of tandem device with 11 nm of Cy-TPFB unbiased (blue curve), biased with 735 nm LED light
(red curve), and biased with 940 nm bias (black curve).
device with and without optical bias is shown in Figure 6.3(d). The unbiased EQE curve is the lowest
because at each point, the quantum efficiency of the subcell with lower current is being measured. When
we optically bias one subcell in a tandem device, we fill it with excess carriers and measure the photocurrent
in the other subcell. This is more representative of real-world conditions because in JV testing, all
wavelengths of light are incident on the device at the same time. When the device is biased with 940 nm
light, the photoresponse of the ClAlPc subcell is being measured (black curve), and when the device is
biased with 735 nm light, the photoresponse of the Cy-TPFB subcell is measured (red curve). The

93

photoresponse of the ClAlPc subcell is higher overall, especially in the region of C60 photoresponse (~400
nm); this imbalance causes a reduction of current density of the tandem device as seen in Figure 6.3(b). The
integrated JSC is 2.3 and 3.5 mA cm-2 for the TPFB and ClAlPc subcells in the tandem device, respectively,
while the measured JSC for the tandem device was 3.0 mA cm-2. These values are consistent with each other
since we expect the current of the tandem device to be between the integrated JSCs of the subcell; when one
subcell produces less current, a reverse bias is created by the other subcell to make it produce more current.
Because the Cy-TPFB subcell was determined to be the current-limiting subcell, changes will need to be
made to the device architecture to improve quantum efficiency within that subcell, which can be
accomplished either by changing layer thicknesses or by adopting a bulk heterojunction.
To guide these experiments, simulations were performed to generate contour maps of the
performance metrics JSC, AVT, and CRI. The initial data shown above highlights the complex nature of
optimizing and fully current matching these types of devices since there are eleven variables (each layer
thickness is a variable). For clarity and visualization purposes, we focus on varying the middle MoO3, top
ITO, ClAlPc and Cy-TPFB thicknesses since these have been identified as the most critical. The results of
simulating AVT, CRI and JSC are split into Figure 6.4 (varying MoO3 and ITO) and Figure 6.5 (varying
ClAlPc and Cy-TPFB). Changes in MoO3 and ITO result in significant changes in AVT (0.75 to 0.80) and
CRI (85% to 95%), meaning a total improvement of 10% in optical properties is possible with complete
optimization. In electrical performance, JSC varied from 2.1 to 2.6 mA cm-2, which marks a potential
improvement of 25%. For changes in ClAlPc and Cy-TPFB, with thicker layers, the AVT intuitively
decreases as there is more overall absorption across the spectrum. CRI is less sensitive but also starts to
decrease above a certain ClAlPc thickness because of tail absorption in the visible region. For the JSC
optimization, we find that thicknesses near the exciton diffusion lengths for the two materials is ideal, which
is well known in the OPV field. Based on these contour maps, the AVT and JSC can potentially be improved
by more than 20% just by small changes in thickness variations. Experimental device fabrication is needed
to confirm these trends; however, these contour maps represent a starting point and a guide to optimization.
Full optimization will require simulation across the entire multivariable space, utilizing larger
94

Figure 6.4. Contour maps for transparent Cy-TPFB/C60 – ClAlPc/C60 tandem devices showing (a)
AVT, (b) CRI, and (c) JSC as a function of the middle MoO3 and the top ITO thicknesses. For these
simulations, Cy-TPFB and ClAlPc thicknesses were both held constant at 200 Å.

Figure 6.5. Contour maps for transparent Cy-TPFB/C60 – ClAlPc/C60 tandem devices showing (a)
AVT, (b) CRI, and (c) JSC as a function of the ClAlPc and the Cy-TPFB layer thicknesses. For these
simulations, the middle MoO3 and the top ITO thicknesses were held constant at 100 and 1200 Å,
respectively. Explain why the range was limited for the thickness of Cy-TPFB and that a larger range would
expose optical interference effects (e.g. starting to be seen in the CRI at 400 Å Cy-TPFB).
supercomputers or genetic algorithms for sampling.
Another approach for tandem TPVs is a C60-only subcell capable of harvesting just the UV part of
the solar spectrum with a higher voltage. These C60-only devices generally only work well when the C60
is dilutely doped (5 volume %) with a donor, for example 1,1-bis-(4-bis(4-methyl-phenyl)-amino-phenyl)cyclohexane (TAPC).[202] Several donors can be used to dilutely dope C60, but TAPC has less tail
absorption in the visible light compared to other donors such as ClAlPc and DBP (Figure 6.6). The working
principle of these devices is that the exciton dissociates within the doped C60 layer almost immediately, and
the electron and hole mobility of C60 is high enough to allow for efficient ambipolar charge collection. This
device has high quantum efficiency (50-60%) in the UV (300-400 nm) region (see Figure 6.7(a)), as well

95

Figure 6.6. Dilute donor devices with ClAlPc, DBP, and TAPC. (a) Device structure and molecular
structure for chloroaluminum phthalocyanine (ClAlPc), dibenzoperiflanthene (DBP), and 1,1-bis-(4-bis(4methyl-phenyl)-amino-phenyl)-cyclohexane (TAPC). (b) JV and (c) EQE of devices with 5% volume
doping (donor:acceptor).

Figure 6.7. ClAlPc BHJ and dilute C60 tandem devices. (a) EQE of a single junction, 5% TAPC:C60
device and of a ClAlPc BHJ device. (b) Device structure of an opaque tandem device with TAPC:C60 front
subcell and ClAlPc BHJ back subcells. (c) JV of TAPC:C60 – ClAlPc BHJ tandem device and the TAPC:C60
and ClAlPc BHJ subcells.
as high VOC (~ 1.0 V). The combination of this subcell with ClAlPc/C60 in tandem devices has resulted in
good VOC summations of the subcells: 0.92 V for TAPC:C60, 0.80 V for ClAlPc BHJ, and 1.65 V for the
tandem device (compare with 0.92 + 0.80 = 1.72 V) (Figure 6.7(b) and (c)). Moreover, 1.65 V is the highest
VOC for a tandem device achieved thus far in this study. However, optically biasing the individual subcell
is challenging because every wavelength absorbed by the TAPC:C60 subcell is also absorbed by the
ClAlPc/C60 subcell. This complicates the process of determining the current-limiting subcell and requires
more manual tuning of probe light intensity during EQE measurements and a posteriori simulations to
predict the photocurrent contributions from each subcell. Nevertheless, the high quantum efficiency and
VOC of the dilutely doped C60 subcell make it a promising candidate for further study in tandem TPVs.

96

The multijunction architecture is attractive for making higher voltage devices and reducing
thermalization losses. As discussed in Chapter 2, the theoretical efficiency limit for a transparent solar cell
increases from 20% PCE to 27% PCE going from a single-junction to a two-junction device. We tested a
variety of subcells (ClAlPc/C60, SnPc/C60, Cy-TPFB/C60 and TAPC:C60) for OPVs, finding that good
voltage summations could be obtained with using an ICL of 50 Å BCP / 1-5 Å Ag / 50 Å MoO3. More
simulations and experimental work will be needed to current match the subcells in TPV architectures, i.e.
using a transparent top electrode.

97

Chapter 7 – Conclusions and Outlook
In this chapter, future work in approaching the efficiency limits for efficient transparent organic
solar cells will be discussed. Also, challenges to face for large scale commercial adoption of the technology
will be considered.
7.1 TPV efficiency: increase NIR EQE, multijunction TPV, better TE design
Key challenges for improving the efficiency of transparent solar cells include increasing the
quantum efficiency in the ultraviolet and near infrared, optimizing the architecture of multijunction TPVs,
and designing better performing transparent electrodes for better scalability. The lifetime is arguably just
as important of an important performance metric as PCE because the power generated over the lifetime of
the device is PCE * lifetime; accordingly, it will be discussed in its own section, Section 7.4. The limiting
factor in current TPVs is the exciton diffusion efficiency since most of the exciton diffusion lengths of
donors are small (< 10 nm), which limits the thickness (and therefore absorption) or charge collection
efficiency for the near-infrared absorbing donor layer. To increase the exciton diffusion efficiency, the
morphology of the donor/acceptor blend needs to be improved. The morphology can be affected by
changing deposition conditions, such as changing solution compositions, deposition rates and substrate
temperatures. Different post-deposition treatments such as thermal or solvent annealing can also be
explored. To analyze morphological changes, SEM, TEM, and small-angle XRD can be used. Another area
of development lies in the design of the molecular structure of the absorber. New donors or acceptors can
be synthesized with different absorption spectra, higher carrier mobilities, lower exciton binding energies,
and better processability (solubility in the case of solution processable molecules). Different strategies like
extending the conjugation, using the push-pull approach on a single molecule, and improving π- π stacking
can be used to accomplish these various objectives.
Multijunction solar cells have been demonstrated to be a reliable way to improve efficiency for
many types of solar cell. The optimization of multijunction transparent solar cells will require more
modeling and experimental studies to port tandem organic solar cell development techniques to tandem

98

transparent solar cells. Extensive modeling and optical cavities can be used to create double passes of both
UV and NIR light in the solar cell, increasing the absorption and the quantum efficiency. Experimental
studies will be needed to optimize the interconnecting layers between junctions and to tune absorber
thicknesses in the individual subcells.
Transparent electrodes for transparent solar cells have been thus far dominated by ITO for the
bottom layer. For the top electrode, the requirement of a layer with compatible processing methods with
the underlying layers has made conventional ITO sub-optimal. However, ITO replacements have not been
as transmissive or conductive, so transparent top electrode development remains a key challenge for
transparent solar cells.
Sputtered oxides have potential for high transparencies, and with different dopants, the
conductivities can be improved. Moreover, with different deposition techniques, such as eclipsed PLD,
damage to the underlying film can be prevented. AgNWs are highly conductive but have surface plasmon
resonances that lead to absorption in the near infrared. For solution processed transparent electrodes, a
combination of conductive oxides and silver nanowires will likely be necessary.
7.2 TPV applications
TPVs can be deployed over large amounts of surface area previously inaccessible for solar energy
harvesting. With approximately 4 billion m2 of glass surface area in the United States alone,[203, 204]
TPVs are capable of offsetting a significant fraction of energy consumption in buildings and generating
electricity close to the point of use while also reducing heating loads similar to low-emissivity (low-E)
coatings. Combined with the 8 billion m2 of rooftop surface area suitable for PV installation,[205] TPVs
can serve as an important complement to rooftop photovoltaics.
Smart windows are a particularly attractive starting point for TPVs because of their low power
demands. A 2% PCE TPV is sufficient to keep even a north-facing smart window powered indefinitely. In
addition, the maximum AVT of most smart windows is only around 70% with varying degrees of CRI and
color tint. These thresholds are already within the demonstrated capabilities of the newest TPV architectures.

99

Beyond power generation, TPVs can also serve as viable replacements for low-E window coatings
which reflect NIR light responsible for carrying heat.[206] Low-E coatings are used to offset cooling loads,
or the amount of energy required to cool a given area. By instead harvesting the NIR light for power
generation, these technologies can simultaneously offset cooling load while contributing energy to the
building.
Consumer electronics such as e-readers are great near-term targets for TPVs due to their large
volumes, low power requirements, and common exposure to lighting and sunlight. TPVs could allow
devices to recharge while in use, allowing manufacturers to minimize battery size to keep devices thin.
Internet of Things (IoT) devices such as electronic shelf labels are another near-term possibility. These and
other IoT devices that are becoming integral components in smart building, retail space, and home networks
feature low enough power consumption that they can already be powered by preliminary TPVs.
TPVs are also ideal components for future all-electric and solar-powered vehicles,[207, 208] where
high AVT (55-90%) is key for coating over windows and preserving aesthetic freedom. While a completely
coated vehicle would require PCEs of 10-15% to provide 10-20 miles worth of travel (~3-3.5 mi./kW-hr)
from a typical charge on a sunny day, modern TPVs are applicable on smaller scales such as internal fans
or smart windows. These simple features would allow vehicles to cool themselves while parked in the sun
without draining the battery.
7.3 TPV scalability
Scaling up and integrating TPVs for given commercial applications requires the simultaneous
consideration of PCE, AVT, and CRI. Entry into architectural applications typically require PCE > 3-5% to
appreciably offset energy consumption in a building, and the AVT should generally be > 60% for non-tinted
windows.[209] While some inorganic semitransparent PV technologies are applicable here, they should be
neutrally colored (high CRI) to avoid adverse effects on the circadian rhythms of inhabitants.[210] Lowpower smart windows and electronic displays can be powered with PCEs > 2%, but the latter require AVTs >
80% and CRI > 90 so as not to interfere with the view. Preliminary TPVs can already meet many of these
metrics as discussed above, and higher performance devices are likely to continue emerging.
100

7.4 TPV lifetime
Lifetime is a key consideration to any PV technology and needs to be longer than the application
lifetime or the cost/energy payback time to become commercially viable. Organic and excitonic materials
are well known to be sensitive to oxygen and moisture [211]. Despite these potential sensitivities,
commercialized organic light emitting diodes (OLEDs) made from similar materials exhibit lifetimes on
the order of decades[212] and a number of OPV lifetimes are now reaching into 5-20 years. Organic PV
performance often has a region of fast decay (also known as the burn-in) followed by a region of slower
decay and can overall be fitted to a biexponential equation or a combination of exponential and linear decay

Figure 7.1. Examples of OPV lifetime. (a) OPV performance decay over time with three primary
regimes: burn-in, long-term, and failure. Reproduced from [8]. (b) Polymer (PCDTBT:PCBM) OPV
lifetime in an oxygen and moisture-free environment. Reproduced from Ref. [18]. (c) Performance
parameter decay curves over 4400 hrs. for encapsulated polymer OPV in indoor 1-sun testing conditions.
Reproduced from [24]

101

(Figure 7.1 (a)).[213] When encapsulation fails, there is often a quick decay in performance until the device
no longer functions at all.[8] PV lifetimes are typically reported as T80 (or T50), which is defined as the time
for performance to decrease to 80% (or 50%) of either the initial performance or the performance after the
burn-in. When the length of the experiment is less than T80 or T50, the rate of decay is extrapolated to
calculate T80 and T50. Primary degradation mechanisms often include moisture or oxygen diffusion in the
active organic layer, which can lead to photo-oxidation of organic layers (e.g. via triplet quenching to
generate singlet oxygen), oxidation of electrodes (e.g. Ag), and crystallization of low Tg (glass-transition
temperature) buffer layers.[8, 213-215] Structural changes in the morphology, particularly for
nanostructured bulk heterojunctions can also result from thermal fluctuations and the presence of moisture
or other solvents. Many of the potential stability issues can be overcome with encapsulation between glass
panels or thin-film encapsulation (for flexible applications).[216] For encapsulated PCDTBT:PCBM
devices held under 1-sun illumination in a controlled moisture- and oxygen-free environment, projected T80
lifetimes of up to 20 years were demonstrated (Figure 7.1 (b)).[18] The same polymer, PCDTBT, had a
projected T80 of 6.2 years when tested under ambient, indoor 1-sun conditions (Figure 7.1 (c)).[24]
Changing the active layer morphology from planar to bulk heterojunction has generally been shown to
increase OPV lifetime due to reduction of the probability of generating new trap-states from exciton-exciton
and exciton-polaron interactions, which are reduced by the increased exciton dissociation efficiency and
leads to lower exciton concentrations.[217]
Nonetheless, when considering lifetime and stability it is often equally important to determine the
required lifetime for a given application. If the application lifetime is longer than the PV lifetime, then the
ease of replacing the PV module becomes the key question. Mobile electronics are often replaced at least
once every 10 years, and most smart phones are replaced as frequently as every 2-3 years. For these
applications, a TPV could be pre-coated onto or beneath display screens. In contrast, for architectural
applications, no PV technology will last as long as a building or window. In this case, TPVs could be

102

Figure 7.2. Optical window of absorbance for biological tissue. Hb = Hemoglobin, Hb-O2 =
Oxyhemoglobin. The shaded portion is the absorption range of Cy salts.
applied as laminates to the inside or outside of windows, which would then allow for cost-effective
replacement.
7.5 Future applications: organic salts for cancer theranostics
Inspired by the fine tuning afforded by organic salts, we have begun to investigate the potential to
translate our solar cell materials to cancer diagnostic probes and therapeutics. Organic salts that absorb and
fluoresce in the infrared can be used as bioimaging agents due to the NIR optical window in biological
tissue that, to date, has been underutilized (Figure 7.2). They can also be used as cancer therapy drugs,
killing tumor cells by generating reactive oxygen species (ROS) or by producing local thermal stress.
Fluorescent molecules have been used clinically as a bioimaging agent since the FDA approved
indocyanine green (ICG) for use in humans in 1959.[183, 218] It was shown to preferentially accumulate
in tumor cells because of the mitochondrial membrane potential.[219] More recently, ICG and other
polymethines have been explored as theranostic drugs, i.e. the combination of tumor diagnosis as well as
tumor killing (therapy).[220, 221] Theranostic drugs are attractive as they combine multiple functions into
one drug, but bright imaging and cytototoxicity are hard to optimize for the same molecule.

103

Figure 7.3. Types of photodynamic therapy. (a) Optical processes leading to the formation of peroxide
(O2-) and singlet oxygen (1O2). A = Absorption, F = Fluorescence, P = Phosphorescence, ISC = intersystem
crossing. (b) Illustration of how the excited triplet state on the dye triplet state (T 1) transfers an electron to
ground state oxygen (3O2) to create singlet oxygen (1O2).
There are two main types of cancer therapy, photodynamic therapy (PDT) and photothermal
therapy (PTT) (Figure 7.3). In PDT, the dye absorbs light and then creates a reactive oxygen species (ROS).
ROS examples include superoxide (O2-), peroxide (H2O2), hydroxyl radicals (OH-) and singlet oxygen (1O2),
and they are generated during normal cell metabolism and can be found in larger amounts when the cell is
under stress. ROS causes damage to different membranes as well as to DNA, leading to cell death. In PDT
a photon is absorbed creating an excited singlet state on the dye. The excited state can then dissociate to
oxidize molecular O2 and form superoxide, or can undergoes intersystem crossing to the triplet state and
energy transfer to ground state triplet oxygen to create singlet oxygen. These two types of therapy can be
distinguished by electron paramagnetic resonance (EPR) because triplet states with the spins going the same
way have a specific magnetic signature.[222] Photothermal therapy (PTT) is another type of therapy that is
considered for these molecules where dyes accumulated in the tumor absorbs NIR light, reradiates this
energy as heat (as opposed to light), which then damages to the surrounding cell environment. Because the
formations of different ROS have different redox potentials, being able to tune the energy levels of the Cy
could allow us to turn on or off cytotoxicity at will. Having a facile way to change the phototoxicity of a
cationic dye is a powerful tool to be able to kill tumors (high phototoxicity) and produce bright images (low
phototoxicity) using the same molecule.
104

Figure 7.4. Optoelectronic tunability of Cy salts for therapy and for imaging. The ball-and-stick model
and Connolly surfaces of the anions are shown: hexafluorophosphate (PF6-), tetrakis(4-fluorophenyl)borate
(FPhB-); cobalticarborane (CoCB-); tetrakis (pentafluorophenyl) borate (TPFB-); tetrakis[3,5-bis(trifluoro
methyl)phenyl]borate (TFM-); Δ-tris(tetrachloro-1,2-benzene diolato) phosphate(V) (TRIS-). Figure
courtesy of Prof. Sophia Lunt.

Figure 7.5. A549 cell viability with different concentrations of organic salts dissolved in DMSO. (left)
cell viability for cultures kept in the dark. (right) cell viability for cultures illuminated with NIR flashlight
centered at 850 nm with daily ~10 minute treatments showing unprecedented tunability (figure courtesy of
Prof. Sophia Lunt).

105

Preliminary work now being carried out with our collaborator (Prof. Sophia Lunt) has shown that
anion exchange can dramatically change the cytotoxicity of the Cy salt (Figure 7.4).[31] Unlike Cy-I and
Cy-PF6 which are cytotoxic at even small concentrations of 1-5 ÎźM in the dark, Cy-FPhB is found to be
highly photocytotoxic, causing cell death only when exposed to NIR light (Figure 7.5). These are attractive
for laser treatments of tumors as they can now provide true double selectivity through the use of both the
dye and light. By deepening the HOMO and pairing the Cy salt with TRIS or TPFB, initial data has found
that the salt can become entirely nontoxic even at much higher concentrations of >20 ÎźM in dark and in
light, making it attractive for non-toxic in-vivo cellular imaging. The range in tunable toxicity of the Cy
dye allows it to be used for both bright imaging as well as for photo-activated therapy, making it an exciting
new platform in biomedical engineering. Future work will need to explore this systems in actual mammalian
systems in vivo with a variety of cancer types and conditions.
7.6 Final summary
Transparent solar cells are an exciting new paradigm for solar deployment specifically enabled by
organic and excitonic semiconductors. They offer a pathway to integrate solar onto virtually any surface
without impacting the view or aesthetics and open new applications that have been less accessible to
traditional semiconductors. By selectively harvesting the invisible parts of the solar spectrum it is possible
to achieve the highest possible combination of efficiency and transparency. Optical interference, low
bandgap donors, and transparent electrodes have all been critical to the optimization of TPVs thus far.
Record transparent solar cells have already reached PCEs of >5% with 60% AVT, enabling deployment
over the smart windows and low-power electronic displays of today. It is likely that TPVs will eventually
approach OPV record efficiencies above 10% while also imparting a high degree of visible transparency.
To reach this lofty goal, there is still a need to develop a broader catalog of donors and acceptors that absorb
outside of the visible spectrum while simultaneously exhibiting high exciton diffusion lengths, charge
carrier mobilities, and lifetime. These important design strategies, combined with the development of new
transparent multijunction architectures, will enable higher energy applications for TPVs in the future and

106

complement traditional PVs in the quest to offset a significant fraction of our energy needs with renewable
energy.

107

APPENDIX

108

List of publications from this thesis
1. Young, M., Traverse, C. J., Pandey, R., Barr, M. C., and Lunt, R. R., Angle dependence of
transparent photovoltaics in conventional and optically inverted configurations. Applied Physics
Letters, 2013. 103(13): p. 133304.
2. Ding, Y., Young, M., Zhao, Y., Traverse, C. J., Benard, A., and Lunt, R. R., Influence of
photovoltaic angle-dependence on overall power output for fixed building integrated
configurations. Solar Energy Materials and Solar Cells, 2015. 132: p. 523-527.
3. Young, M., Suddard‐Bangsund, J., Patrick, T. J., Pajares, N., Traverse, C. J., Barr, M. C., Lunt, S.
Y., and Lunt, R. R., Organic Heptamethine Salts for Photovoltaics and Detectors with Near ‐
Infrared Photoresponse up to 1600 nm. Advanced Optical Materials 4, 1028 (Frontispiece) 2016.
4. Suddard-Bangsund, J., Traverse, C. J., Young, M., Patrick, T. J., Zhao, Y., and Lunt, R. R., Organic
Salts as a Route to Energy Level Control in Low Bandgap, High Open-Circuit Voltage Organic
and Transparent Solar Cells that Approach the Excitonic Voltage Limit. Advanced Energy
Materials, 2016. 6(1): p. 1501659.
5. Traverse, C. J., Young, M., Wagner, S., Zhang, P., Askeland, P., Barr, M. C., and Lunt, R. R.,
Efficient zinc sulfide cathode layers for organic photovoltaic applications via n-type doping.
Journal of Applied Physics, 2014. 115(19): p. 194505.
6. Young, M., Traverse, C. J., and Lunt, R. R. (2018). Chapter 1: Transparent Photovoltaics. In World
Scientific Handbook of Organic Electronics, Volume IV: OPVs. Copyright 2017. World Scientific
Publishing Co.
7. Traverse, C. J., Pandey, R., Barr, M. C., and Lunt, R. R., Emergence of Highly Transparent
Photovoltaics. In re-review.
8. Traverse, C. J., Young, M., Patrick, T., Bangsund, J., Bates, M., Chen, P., Redoute, A., Wingate,
B., Lunt, S. Y., Anctil, A., and Lunt, R. R. Anions for Near-Infrared Selective Organic Salt
Photovoltaics. In preparation.
9. Lunt, R. R., Suddard-Bangsund, J., Traverse, C., and Young, M. 2016. Organic salts for high
voltage organic and transparent solar cells. WO2016176008 A1, filed Apr 6, 2016, and issued
Nov 3, 2016.
10. Lunt, R. R., Lunt, S. Y., Broadwater, D., Young, M. and Bates, M. 2017. Tunable Luminescent
Organic Salts For Enhanced Imaging And Photodynamic Therapy. Filed Aug 8, 2017.
Notes:
Large portions of Chapter 2 will be published in 2018 as a book chapter entitled “Chapter 1: Transparent
Photovoltaics” in World Scientific Handbook of Organic Electronics, Volume IV.
The Frontispiece for Organic Heptamethine Salts for Photovoltaics and Detectors with Near-Infrared
Photoresponse is shown on the next page.

109

Figure A.1. Frontispiece for Advanced Optical Materials Volume 4, Issue 7 (July 2016).
Corresponding article: Organic Heptamethine Salts for Photovoltaics and Detectors with Near-Infrared
Photoresponse.

110

BIBLIOGRAPHY

111

BIBLIOGRAPHY

1.

Traverse, C. J., Pandey, R., Barr, M. C., and Lunt, R. R., Emergence of Visibly Transparent
Photovoltaics. In re-review.

2.

Lunt, R. R., Theoretical limits for visibly transparent photovoltaics. Applied Physics Letters, 2012.
101(4): p. 043902.

3.

Lunt, R. R. (2010), The Growth, Characterization, and Application of Highly Ordered Small
Molecule Semiconducting Thin Films, in Chemical Engineering. 2010, Princeton University.

4.

Bailey-Salzman, R. F., Rand, B. P., and Forrest, S. R., Semitransparent organic photovoltaic cells.
Applied Physics Letters, 2006. 88(23): p. 233502.

5.

Zhao, Y., Meek, G. A., Levine, B. G., and Lunt, R. R., Near-Infrared Harvesting Transparent
Luminescent Solar Concentrators. Advanced Optical Materials, 2014. 2(7): p. 606-611.

6.

Gilot, J., Wienk, M. M., and Janssen, R. A., Measuring the external quantum efficiency of two‐
terminal polymer tandem solar cells. Advanced Functional Materials, 2010. 20(22): p. 3904-3911.

7.

Hu, L., Kim, H. S., Lee, J.-Y., Peumans, P., and Cui, Y., Scalable coating and properties of
transparent, flexible, silver nanowire electrodes. ACS nano, 2010. 4(5): p. 2955-2963.

8.

Mateker, W. R. and McGehee, M. D., Progress in Understanding Degradation Mechanisms and
Improving Stability in Organic Photovoltaics. Advanced Materials, 2017. 29(10).

9.

Chen, C. C., Chang, W. H., Yoshimura, K., Ohya, K., You, J., Gao, J., Hong, Z., and Yang, Y., An
efficient triple‐junction polymer solar cell having a power conversion efficiency exceeding 11%.
Advanced materials, 2014. 26(32): p. 5670-5677.

10.

Meiss, J., Menke, T., Leo, K., Uhrich, C., Gnehr, W.-M., Sonntag, S., Pfeiffer, M., and Riede, M.,
Highly efficient semitransparent tandem organic solar cells with complementary absorber
materials. Applied Physics Letters, 2011. 99(4): p. 043301.

11.

Yim, J. H., Joe, S.-y., Pang, C., Lee, K. M., Jeong, H., Park, J.-Y., Ahn, Y. H., de Mello, J. C., and
Lee, S., Fully solution-processed semitransparent organic solar cells with a silver nanowire
cathode and a conducting polymer anode. ACS nano, 2014. 8(3): p. 2857-2863.

12.

Zhao, W., Qian, D., Zhang, S., Li, S., Inganäs, O., Gao, F., and Hou, J., Fullerene‐Free Polymer
Solar Cells with over 11% Efficiency and Excellent Thermal Stability. Advanced Materials, 2016.
28(23): p. 4734-4739.

13.

Takeoka, A., Kouzuma, S., Tanaka, H., Inoue, H., Murata, K., Morizane, M., Nakamura, N.,
Nishiwaki, H., Ohnishi, M., Nakano, S., and Kuwano, Y., Development and application of seethrough a-Si solar cells. Solar Energy Materials and Solar Cells, 1993. 29(3): p. 243-252.

14.

Lunt, R. R. and Bulovic, V., Transparent, near-infrared organic photovoltaic solar cells for
window and energy-scavenging applications. Applied Physics Letters, 2011. 98(11): p. 61.

112

15.

Chen, R.-T., Chau, J. L. H., and Hwang, G.-L., Design and fabrication of diffusive solar cell
window. Renewable energy, 2012. 40(1): p. 24-28.

16.

Veith, M., Bubel, C., and Zimmer, M., A novel precursor system and its application to produce tin
doped indium oxide. Dalton Transactions, 2011. 40(22): p. 6028-6032.

17.

Xiao, X., Lee, K., and Forrest, S. R., Inverted, semitransparent small molecule photovoltaic cells.
Applied Physics Letters, 2015. 107(3): p. 033901.

18.

Mateker, W. R., Sachs-Quintana, I., Burkhard, G. F., Cheacharoen, R., and McGehee, M. D.,
Minimal long-term intrinsic degradation observed in a polymer solar cell illuminated in an oxygenfree environment. Chem. Mater, 2015. 27(2): p. 404-407.

19.

Ameri, T., Dennler, G., Waldauf, C., Azimi, H., Seemann, A., Forberich, K., Hauch, J., Scharber,
M., Hingerl, K., and Brabec, C. J., Fabrication, Optical Modeling, and Color Characterization of
Semitransparent Bulk-Heterojunction Organic Solar Cells in an Inverted Structure. Advanced
Functional Materials, 2010. 20(10): p. 1592-1598.

20.

Fath, P., Nussbaumer, H., and Burkhardt, R., Industrial manufacturing of semitransparent
crystalline silicon POWER solar cells. Solar energy materials and solar cells, 2002. 74(1): p. 127131.

21.

Chueh, C.-C., Chien, S.-C., Yip, H.-L., Salinas, J. F., Li, C.-Z., Chen, K.-S., Chen, F.-C., Chen,
W.-C., and Jen, A. K. Y., Toward High-Performance Semi-Transparent Polymer Solar Cells:
Optimization of Ultra-Thin Light Absorbing Layer and Transparent Cathode Architecture.
Advanced Energy Materials, 2013. 3(4): p. 417-423.

22.

Chen, C.-C., Dou, L., Zhu, R., Chung, C.-H., Song, T.-B., Zheng, Y. B., Hawks, S., Li, G., Weiss,
P. S., and Yang, Y., Visibly transparent polymer solar cells produced by solution processing. Acs
Nano, 2012. 6(8): p. 7185-7190.

23.

Chen, C.-C., Dou, L., Gao, J., Chang, W.-H., Li, G., and Yang, Y., High-performance semitransparent polymer solar cells possessing tandem structures. Energy & Environmental Science,
2013. 6(9): p. 2714.

24.

Peters, C. H., Sachs-Quintana, I. T., Kastrop, J. P., BeauprĂŠ, S., Leclerc, M., and McGehee, M. D.,
High Efficiency Polymer Solar Cells with Long Operating Lifetimes. Advanced Energy Materials,
2011. 1(4): p. 491-494.

25.

Peh, R. J., Lu, Y., Zhao, F., Lee, C.-L. K., and Kwan, W. L., Vacuum-free processed transparent
inverted organic solar cells with spray-coated PEDOT:PSS anode. Solar Energy Materials and
Solar Cells, 2011. 95(12): p. 3579-3584.

26.

Chen, S. and Carroll, D. L., Silver Nanoplates:  Size Control in Two Dimensions and Formation
Mechanisms. The Journal of Physical Chemistry B, 2004. 108(18): p. 5500-5506.

27.

Chang, C.-Y., Zuo, L., Yip, H.-L., Li, C.-Z., Li, Y., Hsu, C.-S., Cheng, Y.-J., Chen, H., and Jen, A.
K. Y., Highly Efficient Polymer Tandem Cells and Semitransparent Cells for Solar Energy.
Advanced Energy Materials, 2014. 4(7): p. 1301645.

113

28.

Kim, Y. H., MĂźller-Meskamp, L., Zakhidov, A. A., Sachse, C., Meiss, J., Bikova, J., Cook, A.,
Zakhidov, A. A., and Leo, K., Semi-transparent small molecule organic solar cells with laminated
free-standing carbon nanotube top electrodes. Solar Energy Materials and Solar Cells, 2012. 96: p.
244-250.

29.

Yusoff, A. R. b. M., Lee, S. J., Shneider, F. K., da Silva, W. J., and Jang, J., High-Performance
Semitransparent Tandem Solar Cell of 8.02% Conversion Efficiency with Solution-Processed
Graphene Mesh and Laminated Ag Nanowire Top Electrodes. Advanced Energy Materials, 2014.
4(12): p. 1301989.

30.

Schmidt, H., FlĂźgge, H., Winkler, T., BĂźlow, T., Riedl, T., and Kowalsky, W., Efficient
semitransparent inverted organic solar cells with indium tin oxide top electrode. Applied Physics
Letters, 2009. 94(24): p. 243302.

31.

Suddard-Bangsund, J., Traverse, C. J., Young, M., Patrick, T. J., Zhao, Y., and Lunt, R. R., Organic
Salts as a Route to Energy Level Control in Low Bandgap, High Open-Circuit Voltage Organic
and Transparent Solar Cells that Approach the Excitonic Voltage Limit. Advanced Energy
Materials, 2016. 6(1): p. 1501659.

32.

Zhang, H., Wicht, G., Gretener, C., Nagel, M., NĂźesch, F., Romanyuk, Y., Tisserant, J.-N., and
Hany, R., Semitransparent organic photovoltaics using a near-infrared absorbing cyanine dye.
Solar Energy Materials and Solar Cells, 2013. 118: p. 157-164.

33.

Bauer, A., Wahl, T., Hanisch, J., and Ahlswede, E., ZnO:Al cathode for highly efficient,
semitransparent 4% organic solar cells utilizing TiOx and aluminum interlayers. Applied Physics
Letters, 2012. 100(7): p. 073307.

34.

Lin, H.-W., Chen, Y.-H., Huang, Z.-Y., Chen, C.-W., Lin, L.-Y., Lin, F., and Wong, K.-T., Highly
efficient bifacial transparent organic solar cells with power conversion efficiency greater than 3%
and transparency of 50%. Organic Electronics, 2012. 13(9): p. 1722-1728.

35.

Betancur, R., Romero-Gomez, P., Martinez-Otero, A., Elias, X., MaymĂł, M., and Martorell, J.,
Transparent polymer solar cells employing a layered light-trapping architecture. Nature Photonics,
2013. 7(12): p. 995-1000.

36.

Véron, A. C., Zhang, H., Linden, A., Nüesch, F., Heier, J., Hany, R., and Geiger, T., Nir-absorbing
heptamethine dyes with tailor-made counterions for application in light to energy conversion.
Organic letters, 2014. 16(4): p. 1044-1047.

37.

Chen, K.-S., Salinas, J.-F., Yip, H.-L., Huo, L., Hou, J., and Jen, A. K. Y., Semi-transparent
polymer solar cells with 6% PCE, 25% average visible transmittance and a color rendering index
close to 100 for power generating window applications. Energy & Environmental Science, 2012.
5(11): p. 9551.

38.

Chang, C.-Y., Zuo, L., Yip, H.-L., Li, Y., Li, C.-Z., Hsu, C.-S., Cheng, Y.-J., Chen, H., and Jen, A.
K. Y., A Versatile Fluoro-Containing Low-Bandgap Polymer for Efficient Semitransparent and
Tandem Polymer Solar Cells. Advanced Functional Materials, 2013. 23(40): p. 5084-5090.

39.

Sun, S.-S. and Dalton, L. R. (2008) Introduction to organic electronic and optoelectronic materials
and devices. CRC Press.

114

40.

Song, J.-H., An, Y.-S., Kim, S.-G., Lee, S.-J., Yoon, J.-H., and Choung, Y.-K., Power output
analysis of transparent thin-film module in building integrated photovoltaic system (BIPV). Energy
and Buildings, 2008. 40(11): p. 2067-2075.

41.

Lunt, R. R., Giebink, N. C., Belak, A. A., Benziger, J. B., and Forrest, S. R., Exciton diffusion
lengths of organic semiconductor thin films measured by spectrally resolved photoluminescence
quenching. Journal of applied physics, 2009. 105(5): p. 053711.

42.

Chamberlain, G., Organic solar cells: A review. Solar cells, 1983. 8(1): p. 47-83.

43.

WĂśhrle, D. and Meissner, D., Organic solar cells. Advanced Materials, 1991. 3(3): p. 129-138.

44.

Tang, C. W., Two‐layer organic photovoltaic cell. Applied Physics Letters, 1986. 48(2): p. 183185.

45.

Karg, S., Riess, W., Dyakonov, V., and Schwoerer, M., Electrical and optical characterization of
poly (phenylene-vinylene) light emitting diodes. Synthetic metals, 1993. 54(1-3): p. 427-433.

46.

Sariciftci, N. S., Smilowitz, L., Heeger, A. J., and Wudl, F., Photoinduced electron transfer from a
conducting polymer to buckminsterfullerene. Science, 1992. 258(5087): p. 1474-1476.

47.

Smilowitz, L., Sariciftci, N., Wu, R., Gettinger, C., Heeger, A., and Wudl, F., Photoexcitation
spectroscopy of conducting-polymer–C 60 composites: Photoinduced electron transfer. Physical
Review B, 1993. 47(20): p. 13835.

48.

Yu, G., Gao, J., Hummelen, J. C., Wudl, F., and Heeger, A. J., Polymer photovoltiac cells:
Enhanced efficiencies via a network of internal donor-acceptor heterojunctions. Science, 1995.
270(5243): p. 1789.

49.

Brabec, C. J., Gowrisanker, S., Halls, J. J., Laird, D., Jia, S., and Williams, S. P., Polymer–fullerene
bulk‐heterojunction solar cells. Advanced Materials, 2010. 22(34): p. 3839-3856.

50.

Erb, T., Zhokhavets, U., Gobsch, G., Raleva, S., StĂźhn, B., Schilinsky, P., Waldauf, C., and Brabec,
C. J., Correlation between structural and optical properties of composite polymer/fullerene films
for organic solar cells. Advanced Functional Materials, 2005. 15(7): p. 1193-1196.

51.

Yang, X., Loos, J., Veenstra, S. C., Verhees, W. J., Wienk, M. M., Kroon, J. M., Michels, M. A.,
and Janssen, R. A., Nanoscale morphology of high-performance polymer solar cells. Nano letters,
2005. 5(4): p. 579-583.

52.

Wienk, M. M., Kroon, J. M., Verhees, W. J., Knol, J., Hummelen, J. C., van Hal, P. A., and Janssen,
R. A., Efficient methano [70] fullerene/MDMO ‐ PPV bulk heterojunction photovoltaic cells.
Angewandte Chemie, 2003. 115(29): p. 3493-3497.

53.

Lin, Y., Li, Y., and Zhan, X., Small molecule semiconductors for high-efficiency organic
photovoltaics. Chem Soc Rev, 2012. 41(11): p. 4245-72.

54.

Gommans, H. H. P., Cheyns, D., Aernouts, T., Girotto, C., Poortmans, J., and Heremans, P.,
Electro-Optical Study of Subphthalocyanine in a Bilayer Organic Solar Cell. Advanced Functional
Materials, 2007. 17(15): p. 2653-2658.

115

55.

Ojala, A., Petersen, A., Fuchs, A., Lovrincic, R., PĂślking, C., Trollmann, J., Hwang, J., Lennartz,
C., Reichelt, H., and HĂśffken, H. W., Merocyanine/C60 planar heterojunction solar cells: effect of
dye orientation on exciton dissociation and solar cell performance. Advanced Functional Materials,
2012. 22(1): p. 86-96.

56.

Schulze, K., Uhrich, C., Schßppel, R., Leo, K., Pfeiffer, M., Brier, E., Reinold, E., and Bäuerle, P.,
Efficient Vacuum-Deposited Organic Solar Cells Based on a New Low-Bandgap Oligothiophene
and Fullerene C60. Advanced Materials, 2006. 18(21): p. 2872-2875.

57.

Zou, Y., Holst, J., Zhang, Y., and Holmes, R. J., 7.9% efficient vapor-deposited organic
photovoltaic cells based on a simple bulk heterojunction. Journal of Materials Chemistry A, 2014.
2(31): p. 12397.

58.

Shim, H.-S., Moon, C.-K., Kim, J., Wang, C.-K., Sim, B., Lin, F., Wong, K.-T., Seo, Y., and Kim,
J.-J., Efficient vacuum-deposited ternary organic solar cells with broad absorption, energy transfer,
and enhanced hole mobility. ACS applied materials & interfaces, 2016. 8(2): p. 1214-1219.

59.

Morel, D., Ghosh, A., Feng, T., Stogryn, E., Purwin, P., Shaw, R., and Fishman, C., High‐efficiency
organic solar cells. Applied Physics Letters, 1978. 32(8): p. 495-497.

60.

O’regan, B. and Grätzel, M., Low cost and highly efficient solar cells based on the sensitization of
colloidal titanium dioxide. Nature, 1991. 335(24): p. 737-740.

61.

Ma, B., Woo, C. H., Miyamoto, Y., and FrĂŠchet, J. M., Solution processing of a small molecule,
subnaphthalocyanine, for efficient organic photovoltaic cells. Chemistry of Materials, 2009. 21(8):
p. 1413-1417.

62.

Wei, G., Lunt, R. R., Sun, K., Wang, S., Thompson, M. E., and Forrest, S. R., Efficient, ordered
bulk heterojunction nanocrystalline solar cells by annealing of ultrathin squaraine thin films. Nano
Lett, 2010. 10(9): p. 3555-9.

63.

Winzenberg, K. N., Kemppinen, P., Fanchini, G., Bown, M., Collis, G. E., Forsyth, C. M., Hegedus,
K., Singh, T. B., and Watkins, S. E., Dibenzo [b, def] chrysene derivatives: solution-processable
small molecules that deliver high power-conversion efficiencies in bulk heterojunction solar cells.
Chemistry of Materials, 2009. 21(24): p. 5701-5703.

64.

Nian, L., Gao, K., Jiang, Y., Rong, Q., Hu, X., Yuan, D., Liu, F., Peng, X., Russell, T. P., and Zhou,
G., Small‐Molecule Solar Cells with Simultaneously Enhanced Short ‐Circuit Current and Fill
Factor to Achieve 11% Efficiency. Advanced Materials, 2017.

65.

Shockley, W. and Queisser, H. J., Detailed Balance Limit of Efficiency of p‐n Junction Solar Cells.
Journal of Applied Physics, 1961. 32(3): p. 510-519.

66.

Lunt, R. R., Osedach, T. P., Brown, P. R., Rowehl, J. A., and Bulovic, V., Practical roadmap and
limits to nanostructured photovoltaics. Adv Mater, 2011. 23(48): p. 5712-27.

67.

Hiramoto, M., Suezaki, M., and Yokoyama, M., Effect of thin gold interstitial-layer on the
photovoltaic properties of tandem organic solar cell. Chemistry Letters, 1990. 19(3): p. 327-330.

116

68.

Xue, J., Uchida, S., Rand, B. P., and Forrest, S. R., Asymmetric tandem organic photovoltaic cells
with hybrid planar-mixed molecular heterojunctions. Applied Physics Letters, 2004. 85(23): p.
5757-5759.

69.

Drechsel, J., Männig, B., Kozlowski, F., Gebeyehu, D., Werner, A., Koch, M., Leo, K., and Pfeiffer,
M., High efficiency organic solar cells based on single or multiple PIN structures. Thin Solid Films,
2004. 451: p. 515-517.

70.

Cheyns, D., Gommans, H., Odijk, M., Poortmans, J., and Heremans, P., Stacked organic solar cells
based on pentacene and C60. Solar energy materials and solar cells, 2007. 91(5): p. 399-404.

71.

Kawano, K., Ito, N., Nishimori, T., and Sakai, J., Open circuit voltage of stacked bulk
heterojunction organic solar cells. Applied Physics Letters, 2006. 88(7): p. 073514.

72.

Dennler, G., Prall, H.-J., Koeppe, R., Egginger, M., Autengruber, R., and Sariciftci, N. S.,
Enhanced spectral coverage in tandem organic solar cells. Applied Physics Letters, 2006. 89(7):
p. 073502.

73.

Gilot, J., Wienk, M. M., and Janssen, R. A., Double and triple junction polymer solar cells
processed from solution. Applied Physics Letters, 2007. 90(14): p. 143512.

74.

Liu, Y., Chen, C. C., Hong, Z., Gao, J., Yang, Y. M., Zhou, H., Dou, L., Li, G., and Yang, Y.,
Solution-processed small-molecule solar cells: breaking the 10% power conversion efficiency. Sci
Rep, 2013. 3: p. 3356.

75.

Heliatek sets new Organic Photovoltaic world record efficiency of 13.2%. 2016 [cited 2017;
Available
from:
http://www.heliatek.com/en/press/press-releases/details/heliatek-sets-neworganic-photovoltaic-world-record-efficiency-of-13-2.

76.

Schubert, E. F., Gessmann, T., and Kim, J. K. (2005) Light emitting diodes. Wiley Online Library.

77.

Thornton, W. A., Luminosity and Color-Rendering Capability of White Light. Journal of the Optical
Society of America, 1971. 61(9): p. 1155-1163.

78.

Color Rendering Index (CRI) Explained.
[cited
http://www.fullspectrumsolutions.com/cri_explained.htm.

79.

Pandey, R. and Holmes, R. J., Characterizing the charge collection efficiency in bulk
heterojunction organic photovoltaic cells. Applied Physics Letters, 2012. 100(8): p. 083303.

80.

Park, S. H., Roy, A., BeauprĂŠ, S., Cho, S., Coates, N., Moon, J. S., Moses, D., Leclerc, M., Lee, K.,
and Heeger, A. J., Bulk heterojunction solar cells with internal quantum efficiency approaching
100%. Nature Photonics, 2009. 3(5): p. 297-302.

81.

A checklist for photovoltaic research. Nat Mater, 2015. 14(11): p. 1073-1073.

82.

Solaronix
Solar
Cells.
2017;
https://www.solaronix.com/documents/solaronix_solar_cells.pdf.

83.

Thin-film Photovoltaic Glazing for BIPV solutions.
http://www.polysolar.co.uk/Technology/thin-film.
117

2017;

Available

Available

from:

from:

[cited 2017; Available from:

84.

Transparent
solar
panels.
2017;
Available
constructions.com/wordpress/transparent-solar-panels/.

85.

Kuhn, R., Boueke, A., Kress, A., Fath, P., Willeke, G. P., and Bucher, E., Characterization of novel
mono-and bifacially active semi-transparent crystalline silicon solar cells. IEEE Transactions on
Electron Devices, 1999. 46(10): p. 2013-2017.

86.

Kranz, L., Buecheler, S., and Tiwari, A. N., Technological status of CdTe photovoltaics. Solar
Energy Materials and Solar Cells, 2013. 119: p. 278-280.

87.

Saifullah, M., Ahn, S., Gwak, J., Ahn, S., Kim, K., Cho, J., Park, J. H., Eo, Y. J., Cho, A., Yoo, J.S., and Yun, J. H., Development of semitransparent CIGS thin-film solar cells modified with a
sulfurized-AgGa layer for building applications. Journal of Materials Chemistry A, 2016. 4(27): p.
10542-10551.

88.

Zhang, X., Eperon, G. E., Liu, J., and Johansson, E. M. J., Semitransparent quantum dot solar cell.
Nano Energy, 2016. 22: p. 70-78.

89.

Han, N., Yang, Z. X., Wang, F., Dong, G., Yip, S., Liang, X., Hung, T. F., Chen, Y., and Ho, J. C.,
High-Performance GaAs Nanowire Solar Cells for Flexible and Transparent Photovoltaics. ACS
Appl. Mater. Interfaces, 2015. 7(36): p. 20454-9.

90.

Eke, R. and Demircan, C., Shading effect on the energy rating of two identical PV systems on a
building façade. Solar Energy, 2015. 122: p. 48-57.

91.

Chatzipanagi, A., Frontini, F., and Virtuani, A., BIPV-temp: A demonstrative Building Integrated
Photovoltaic installation. Applied Energy, 2016. 173: p. 1-12.

92.

Lee, J.-Y., Connor, S. T., Cui, Y., and Peumans, P., Solution-processed metal nanowire mesh
transparent electrodes. Nano letters, 2008. 8(2): p. 689-692.

93.

Koeppe, R., Hoeglinger, D., Troshin, P. A., Lyubovskaya, R. N., Razumov, V. F., and Sariciftci,
N. S., Organic Solar Cells with Semitransparent Metal Back Contacts for Power Window
Applications. ChemSusChem, 2009. 2(4): p. 309-313.

94.

Tao, C., Xie, G., Liu, C., Zhang, X., Dong, W., Meng, F., Kong, X., Shen, L., Ruan, S., and Chen,
W., Semitransparent inverted polymer solar cells with MoO3/Ag/MoO3 as transparent electrode.
Applied Physics Letters, 2009. 95(5): p. 053303.

95.

Shen, L., Xu, Y., Meng, F., Li, F., Ruan, S., and Chen, W., Semitransparent polymer solar cells
using V2O5/Ag/V2O5 as transparent anodes. Organic Electronics, 2011. 12(7): p. 1223-1226.

96.

Han, J. I., Park, J.-K., and Park, S. K., High-Performance Semitransparent Bulk-Heterojunction
Organic Photovoltaics with Ag Interfacial Layer. Japanese Journal of Applied Physics, 2012. 51:
p. 024104.

97.

Dou, L., Liu, Y., Hong, Z., Li, G., and Yang, Y., Low-Bandgap Near-IR Conjugated
Polymers/Molecules for Organic Electronics. Chem Rev, 2015. 115(23): p. 12633-65.

118

from:

http://www.solar-

98.

Liang, Y., Wu, Y., Feng, D., Tsai, S.-T., Son, H.-J., Li, G., and Yu, L., Development of new
semiconducting polymers for high performance solar cells. Journal of the American Chemical
Society, 2008. 131(1): p. 56-57.

99.

He, Z., Zhong, C., Su, S., Xu, M., Wu, H., and Cao, Y., Enhanced power-conversion efficiency in
polymer solar cells using an inverted device structure. Nature Photonics, 2012. 6(9): p. 591-595.

100.

Romero-Gomez, P., Betancur, R., Martinez-Otero, A., Elias, X., Mariano, M., Romero, B.,
Arredondo, B., Vergaz, R., and Martorell, J., Enhanced stability in semi-transparent
PTB7/PC71BM photovoltaic cells. Solar Energy Materials and Solar Cells, 2015. 137: p. 44-49.

101.

Slooff, L. H., Bende, E. E., Burgers, A. R., Budel, T., Pravettoni, M., Kenny, R. P., Dunlop, E. D.,
and BĂźchtemann, A., A luminescent solar concentrator with 7.1% power conversion efficiency.
physica status solidi (RRL) - Rapid Research Letters, 2008. 2(6): p. 257-259.

102.

Currie, M. J., Mapel, J. K., Heidel, T. D., Goffri, S., and Baldo, M. A., High-efficiency organic
solar concentrators for photovoltaics. Science, 2008. 321(5886): p. 226-228.

103.

Mulder, C. L., Theogarajan, L., Currie, M., Mapel, J. K., Baldo, M. A., Vaughn, M., Willard, P.,
Bruce, B. D., Moss, M. W., McLain, C. E., and Morseman, J. P., Luminescent Solar Concentrators
Employing Phycobilisomes. Advanced Materials, 2009. 21(31): p. 3181-3185.

104.

Chau, J. L. H., Chen, R.-T., Hwang, G.-L., Tsai, P.-Y., and Lin, C.-C., Transparent solar cell
window module. Solar Energy Materials and Solar Cells, 2010. 94(3): p. 588-591.

105.

Yeh, C.-H., Chang, F.-Y., Young, H.-T., Hsieh, T.-Y., and Chang, C.-H., Transparent solar
concentrator for flat panel display. Japanese Journal of Applied Physics, 2012. 51(6S): p. 06FL03.

106.

Alghamedi, R., Vasiliev, M., Nur-E-Alam, M., and Alameh, K., Spectrally-selective all-inorganic
scattering luminophores for solar energy-harvesting clear glass windows. Scientific reports, 2014.
4: p. 6632.

107.

Batchelder, J., Zewail, A., and Cole, T., Luminescent solar concentrators. 2: Experimental and
theoretical analysis of their possible efficiencies. Applied Optics, 1981. 20(21): p. 3733-3754.

108.

Zhang, K., Qin, C., Yang, X., Islam, A., Zhang, S., Chen, H., and Han, L., High-Performance,
Transparent, Dye-Sensitized Solar Cells for See-Through Photovoltaic Windows. Advanced
Energy Materials, 2014. 4(11): p. 1301966.

109.

Meiss, J., Holzmueller, F., Gresser, R., Leo, K., and Riede, M., Near-infrared absorbing
semitransparent organic solar cells. Applied Physics Letters, 2011. 99(19): p. 193307.

110.

Young, M., Suddard‐Bangsund, J., Patrick, T. J., Pajares, N., Traverse, C. J., Barr, M. C., Lunt, S.
Y., and Lunt, R. R., Organic Heptamethine Salts for Photovoltaics and Detectors with Near ‐
Infrared Photoresponse up to 1600 nm. Advanced Optical Materials, 2016.

111.

Zhang, H., Niesen, B., Hack, E., Jenatsch, S., Wang, L., VĂŠron, A. C., Makha, M., Schneider, R.,
Arroyo, Y., Hany, R., and NĂźesch, F., Cyanine tandem and triple-junction solar cells. Organic
Electronics, 2016. 30: p. 191-199.

119

112.

Yang, C. and Lunt, R. R., Limits of Visibly Transparent Luminescent Solar Concentrators.
Advanced Optical Materials, 2017.

113.

Zhao, Y. and Lunt, R. R., Transparent Luminescent Solar Concentrators for Large-Area Solar
Windows Enabled by Massive Stokes-Shift Nanocluster Phosphors. Advanced Energy Materials,
2013. 3(9): p. 1143-1148.

114.

Derkacs, D., Lim, S., Matheu, P., Mar, W., and Yu, E., Improved performance of amorphous silicon
solar cells via scattering from surface plasmon polaritons in nearby metallic nanoparticles.
Applied Physics Letters, 2006. 89(9): p. 093103.

115.

Durose, K., Edwards, P., and Halliday, D., Materials aspects of CdTe/CdS solar cells. Journal of
Crystal Growth, 1999. 197(3): p. 733-742.

116.

Geffroy, B., Le Roy, P., and Prat, C., Organic light‐emitting diode (OLED) technology: materials,
devices and display technologies. Polymer International, 2006. 55(6): p. 572-582.

117.

Snell, A. J., Mackenzie, K. D., Spear, W. E., LeComber, P. G., and Hughes, A. J., Application of
amorphous silicon field effect transistors in addressable liquid crystal display panels. Applied
physics, 1981. 24(4): p. 357-362.

118.

Gu, G., Bulović, V., Burrows, P. E., Forrest, S. R., and Thompson, M. E., Transparent organic
light emitting devices. Applied Physics Letters, 1996. 68(19): p. 2606-2608.

119.

Parthasarathy, G., Burrows, P., Khalfin, V., Kozlov, V., and Forrest, S., A metal-free cathode for
organic semiconductor devices. Applied Physics Letters, 1998. 72(17): p. 2138-2140.

120.

Schubert, S., Schmidt, F., von Wenckstern, H., Grundmann, M., Leo, K., and MĂźller-Meskamp, L.,
Eclipse Pulsed Laser Deposition for Damage-Free Preparation of Transparent ZnO Electrodes on
Top of Organic Solar Cells. Advanced Functional Materials, 2015. 25(27): p. 4321-4327.

121.

Cho, D.-Y., Kim, K.-H., Kim, T.-W., Noh, Y.-J., Na, S.-I., Chung, K.-B., and Kim, H.-K.,
Transparent and flexible amorphous InZnAlO films grown by roll-to-roll sputtering for acidic
buffer-free flexible organic solar cells. Organic Electronics, 2015. 24: p. 227-233.

122.

Pettersson, L. A. A. and Snyder, P. G., Preparation and characterization of oxidized silver thin
films. Thin Solid Films, 1995. 270(1): p. 69-72.

123.

O’Connor, B., Haughn, C., An, K.-H., Pipe, K. P., and Shtein, M., Transparent and conductive
electrodes based on unpatterned, thin metal films. Applied Physics Letters, 2008. 93(22): p. 223304.

124.

Sennett, R. and Scott, G., The structure of evaporated metal films and their optical properties.
JOSA, 1950. 40(4): p. 203-211.

125.

Schubert, S., Meiss, J., MĂźller-Meskamp, L., and Leo, K., Improvement of Transparent Metal Top
Electrodes for Organic Solar Cells by Introducing a High Surface Energy Seed Layer. Advanced
Energy Materials, 2013. 3(4): p. 438-443.

126.

Logeeswaran, V., Kobayashi, N. P., Islam, M. S., Wu, W., Chaturvedi, P., Fang, N. X., Wang, S.
Y., and Williams, R. S., Ultrasmooth silver thin films deposited with a germanium nucleation layer.
Nano letters, 2008. 9(1): p. 178-182.
120

127.

da Silva, W. J., Kim, H. P., Rashid bin Mohd Yusoff, A., and Jang, J., Transparent flexible organic
solar cells with 6.87% efficiency manufactured by an all-solution process. Nanoscale, 2013. 5(19):
p. 9324-9.

128.

O’Connor, B., An, K. H., Pipe, K. P., Zhao, Y., and Shtein, M., Enhanced optical field intensity
distribution in organic photovoltaic devices using external coatings. Applied physics letters, 2006.
89(23): p. 233502.

129.

Wang, Z., Zhang, C., Chen, D., Tang, S., Zhang, J., Sun, L., Heng, T., and Hao, Y., ITO-Free
Semitransparent Organic Solar Cells Based on Silver Thin Film Electrodes. International Journal
of Photoenergy, 2014. 2014: p. 1-7.

130.

Sergeant, N. P., Hadipour, A., Niesen, B., Cheyns, D., Heremans, P., Peumans, P., and Rand, B. P.,
Design of transparent anodes for resonant cavity enhanced light harvesting in organic solar cells.
Adv Mater, 2012. 24(6): p. 728-32.

131.

Dong, Q., Zhou, Y., Pei, J., Liu, Z., Li, Y., Yao, S., Zhang, J., and Tian, W., All-spin-coating
vacuum-free processed semi-transparent inverted polymer solar cells with PEDOT:PSS anode and
PAH-D interfacial layer. Organic Electronics, 2010. 11(7): p. 1327-1331.

132.

Kim, H. P., Lee, H. J., Mohd Yusoff, A. R. b., and Jang, J., Semi-transparent organic inverted
photovoltaic cells with solution processed top electrode. Solar Energy Materials and Solar Cells,
2013. 108: p. 38-43.

133.

Nickel, F., Puetz, A., Reinhard, M., Do, H., Kayser, C., Colsmann, A., and Lemmer, U., Cathodes
comprising highly conductive poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) for semitransparent polymer solar cells. Organic Electronics, 2010. 11(4): p. 535-538.

134.

Colsmann, A., Reinhard, M., Kwon, T.-H., Kayser, C., Nickel, F., Czolk, J., Lemmer, U., Clark,
N., Jasieniak, J., Holmes, A. B., and Jones, D., Inverted semi-transparent organic solar cells with
spray coated, surfactant free polymer top-electrodes. Solar Energy Materials and Solar Cells, 2012.
98: p. 118-123.

135.

Kang, J.-W., Kang, Y.-J., Jung, S., You, D. S., Song, M., Kim, C. S., Kim, D.-G., Kim, J.-K., and
Kim, S. H., All-spray-coated semitransparent inverted organic solar cells: From electron selective
to anode layers. Organic Electronics, 2012. 13(12): p. 2940-2944.

136.

Guo, F., Zhu, X., Forberich, K., Krantz, J., Stubhan, T., Salinas, M., Halik, M., Spallek, S., Butz,
B., Spiecker, E., Ameri, T., Li, N., Kubis, P., Guldi, D. M., Matt, G. J., and Brabec, C. J., ITO-Free
and Fully Solution-Processed Semitransparent Organic Solar Cells with High Fill Factors.
Advanced Energy Materials, 2013. 3(8): p. 1062-1067.

137.

Guo, F., Kubis, P., Stubhan, T., Li, N., Baran, D., Przybilla, T., Spiecker, E., Forberich, K., and
Brabec, C. J., Fully solution-processing route toward highly transparent polymer solar cells. ACS
Appl Mater Interfaces, 2014. 6(20): p. 18251-7.

138.

Ariyarit, A., Manabe, K., Fukada, K., Kyung, K.-H., Fujimoto, K., and Shiratori, S.,
Semitransparent polymer-based solar cells via simple wet lamination process with TiO2layer using
automatic spray layer-by-layer method. RSC Adv., 2015. 5(65): p. 52427-52435.

121

139.

Hunger, C., Rao, K. D. M., Gupta, R., Singh, C. R., Kulkarni, G. U., and Thelakkat, M.,
Transparent Metal Network with Low Haze and High Figure of Merit applied to Front and Back
Electrodes in Semitransparent ITO-free Polymer Solar Cells. Energy Technology, 2015. 3(6): p.
638-645.

140.

Krantz, J., Stubhan, T., Richter, M., Spallek, S., Litzov, I., Matt, G. J., Spiecker, E., and Brabec, C.
J., Spray-Coated Silver Nanowires as Top Electrode Layer in Semitransparent P3HT:PCBM-Based
Organic Solar Cell Devices. Advanced Functional Materials, 2013. 23(13): p. 1711-1717.

141.

Liu, Z., Li, J., Sun, Z.-H., Tai, G., Lau, S.-P., and Yan, F., The application of highly doped singlelayer graphene as the top electrodes of semitransparent organic solar cells. ACS nano, 2011. 6(1):
p. 810-818.

142.

Liu, Z., You, P., Liu, S., and Yan, F., Neutral-Color Semitransparent Organic Solar Cells with AllGraphene Electrodes. ACS nano, 2015. 9(12): p. 12026-12034.

143.

Tanaka, S., Zakhidov, A. A., Ovalle-Robles, R., Yoshida, Y., Hiromitsu, I., Fujita, Y., and Yoshino,
K., Semitransparent organic photovoltaic cell with carbon nanotube-sheet anodes and Ga-doped
ZnO cathodes. Synthetic Metals, 2009. 159(21-22): p. 2326-2328.

144.

Hu, X., Chen, L., Zhang, Y., Hu, Q., Yang, J., and Chen, Y., Large-Scale Flexible and Highly
Conductive Carbon Transparent Electrodes via Roll-to-Roll Process and Its High Performance
Lab-Scale Indium Tin Oxide-Free Polymer Solar Cells. Chemistry of Materials, 2014. 26(21): p.
6293-6302.

145.

Ren, X., Li, X., and Choy, W. C. H., Optically enhanced semi-transparent organic solar cells
through hybrid metal/nanoparticle/dielectric nanostructure. Nano Energy, 2015. 17: p. 187-195.

146.

(2012) Guide to Using WVASE Spectroscopic Ellipsometry Data Acquisition and Analysis Software.
J. A. Woollam Co. Inc.

147.

Shrotriya, V., Li, G., Yao, Y., Moriarty, T., Emery, K., and Yang, Y., Accurate Measurement and
Characterization of Organic Solar Cells. Advanced Functional Materials, 2006. 16(15): p. 20162023.

148.

Ding, Y., Young, M., Zhao, Y., Traverse, C. J., Benard, A., and Lunt, R. R., Influence of
photovoltaic angle-dependence on overall power output for fixed building integrated
configurations. Solar Energy Materials and Solar Cells, 2015. 132: p. 523-527.

149.

Meyer, A. and Ade, H., The effect of angle of incidence on the optical field distribution within thin
film organic solar cells. Journal of Applied Physics, 2009. 106(11): p. 113101.

150.

Dennler, G., Forberich, K., Scharber, M. C., Brabec, C. J., TomiĹĄ, I., Hingerl, K., and Fromherz,
T., Angle dependence of external and internal quantum efficiencies in bulk-heterojunction organic
solar cells. Journal of Applied Physics, 2007. 102(5): p. 054516.

151.

Pettersson, L. A. A., Roman, L. S., and Inganäs, O., Modeling photocurrent action spectra of
photovoltaic devices based on organic thin films. Journal of Applied Physics, 1999. 86(1): p. 487496.

122

152.

Peumans, P., Yakimov, A., and Forrest, S. R., Small molecular weight organic thin-film
photodetectors and solar cells. Journal of Applied Physics, 2003. 93(7): p. 3693-3723.

153.

Bronnbauer, C., Hornich, J., Gasparini, N., Guo, F., Hartmeier, B., Luechinger, N. A., Pflaum, C.,
Brabec, C. J., and Forberich, K., Printable Dielectric Mirrors with Easily Adjustable and WellDefined Reflection Maxima for Semitransparent Organic Solar Cells. Advanced Optical Materials,
2015. 3(10): p. 1424-1430.

154.

Tao, C., Xie, G., Meng, F., Ruan, S., and Chen, W., Tailoring Spatial Distribution of the Optical
Field Intensity in Semitransparent Inverted Organic Solar Cells. The Journal of Physical Chemistry
C, 2011. 115(25): p. 12611-12615.

155.

Zhang, D. D., Jiang, X. C., Wang, R., Xie, H. J., Ma, G. F., Ou, Q. D., Chen, Y. L., Li, Y. Q., and
Tang, J. X., Enhanced performance of semitransparent inverted organic photovoltaic devices via
a high reflector structure. ACS Appl Mater Interfaces, 2013. 5(20): p. 10185-90.

156.

Peumans, P., Yakimov, A., and Forrest, S. R., Small molecular weight organic thin-film
photodetectors and solar cells (vol 93, pg 3693, 2003). Journal of Applied Physics, 2004. 95(5): p.
2938-2938.

157.

Ulaby, F. T., Michielssen, E., and Ravaioli, U., Fundamentals of Applied Electromagnetics 6e.
Boston, Massachussetts: Prentice Hall, 2010.

158.

Kim, J., Jung, S., and Jeong, I., Optical Modeling for Polarization-dependent Optical Power
Dissipation of Thin-film Organic Solar Cells at Oblique Incidence. Journal of the Optical Society
of Korea, 2012. 16(1): p. 6-12.

159.

Khavrus, V. and Shelevytsky, I., Introduction to solar motion geometry on the basis of a simple
model. Physics Education, 2010. 45(6): p. 641.

160.

Mlawer, E. J., Brown, P. D., Clough, S. A., Harrison, L. C., Michalsky, J. J., Kiedron, P. W., and
Shippert, T., Comparison of spectral direct and diffuse solar irradiance measurements and
calculations for cloud-free conditions. Geophysical Research Letters, 2000. 27(17): p. 2653-2656.

161.

Cheyns, D., Rand, B. P., Verreet, B., Genoe, J., Poortmans, J., and Heremans, P., The angular
response of ultrathin film organic solar cells. Applied Physics Letters, 2008. 92(24): p. 243310.

162.

Yang, F., Lunt, R. R., and Forrest, S. R., Simultaneous heterojunction organic solar cells with
broad spectral sensitivity. Applied Physics Letters, 2008. 92(5): p. 053310.

163.

Pandey, R., Kerner, R. A., Menke, S. M., Holst, J., Josyula, K. V., and Holmes, R. J., Tin
naphthalocyanine complexes for infrared absorption in organic photovoltaic cells. Organic
Electronics, 2013. 14(3): p. 804-808.

164.

Vasseur, K., Rand, B. P., Cheyns, D., Froyen, L., and Heremans, P., Structural Evolution of
Evaporated Lead Phthalocyanine Thin Films for Near-Infrared Sensitive Solar Cells†. Chemistry
of Materials, 2011. 23(3): p. 886-895.

165.

Rand, B. P., Burk, D. P., and Forrest, S. R., Offset energies at organic semiconductor
heterojunctions and their influence on the open-circuit voltage of thin-film solar cells. Physical
Review B, 2007. 75(11).
123

166.

Xue, J., Rand, B. P., Uchida, S., and Forrest, S. R., Mixed donor-acceptor molecular
heterojunctions for photovoltaic applications. II. Device performance. Journal of Applied Physics,
2005. 98(12): p. 124903.

167.

Young, M., Traverse, C. J., Pandey, R., Barr, M. C., and Lunt, R. R., Angle dependence of
transparent photovoltaics in conventional and optically inverted configurations. Applied Physics
Letters, 2013. 103(13): p. 133304.

168.

Luo, S., Zhang, E., Su, Y., Cheng, T., and Shi, C., A review of NIR dyes in cancer targeting and
imaging. Biomaterials, 2011. 32(29): p. 7127-7138.

169.

Chen, P.-S., Su, J.-L., Cha, S.-T., Tarn, W.-Y., Wang, M.-Y., Hsu, H.-C., Lin, M.-T., Chu, C.-Y.,
Hua, K.-T., Chen, C.-N., Kuo, T.-C., Chang, K.-J., Hsiao, M., Chang, Y.-W., Chen, J.-S., Yang,
P.-C., and Kuo, M.-L., miR-107 promotes tumor progression by targeting the let-7 microRNA in
mice and humans. The Journal of Clinical Investigation, 2011. 121(9): p. 3442-3455.

170.

Bates, M. and Lunt, R. R., Organic Salt Photovoltaics. Sustainable Energy & Fuels, 2017.

171.

Meng, F., Chen, K., Tian, H., Zuppiroli, L., and Nuesch, F., Cyanine dye acting both as donor and
acceptor in heterojunction photovoltaic devices. Applied Physics Letters, 2003. 82(21): p. 37883790.

172.

NĂźesch, F., Tornare, G., Zuppiroli, L., Meng, F., Chen, K., and Tian, H., Interface modification to
optimize charge separation in cyanine heterojunction photovoltaic devices. Solar energy materials
and solar cells, 2005. 87(1): p. 817-824.

173.

Berner, E., Jäger, T., Lanz, T., Nßesch, F., Tisserant, J.-N., Wicht, G., Zhang, H., and Hany, R.,
Influence of crystalline titanium oxide layer smoothness on the performance of inverted organic
bilayer solar cells. Applied Physics Letters, 2013. 102(18): p. 183903.

174.

Malinkiewicz, O., Grancha, T., Molina-Ontoria, A., Soriano, A., Brine, H., and Bolink, H. J.,
Efficient, Cyanine Dye Based Bilayer Solar Cells. Advanced Energy Materials, 2013. 3(4): p. 472477.

175.

Fan, B., Hany, R., Moser, J.-E., and NĂźesch, F., Enhanced cyanine solar cell performance upon
oxygen doping. Organic Electronics, 2008. 9(1): p. 85-94.

176.

Bouit, P.-A., Aronica, C., Guy, L., Martinez, A., Andraud, C., and Maury, O., Diastereoselective
supramolecular ion-pairing between the TRISPHAT anion and pro-chiral heptamethine cyanine
dyes. Organic & Biomolecular Chemistry, 2009. 7(15): p. 3086.

177.

Rand, B. P., Xue, J., Yang, F., and Forrest, S. R., Organic solar cells with sensitivity extending into
the near infrared. Applied Physics Letters, 2005. 87(23): p. 233508.

178.

Hiramoto, M., Kitada, K., Iketaki, K., and Kaji, T., Near infrared light driven organic p-i-n solar
cells incorporating phthalocyanine J-aggregate. Applied Physics Letters, 2011. 98(2): p. 023302.

179.

Shim, H.-S., Kim, H. J., Kim, J. W., Kim, S.-Y., Jeong, W.-I., Kim, T.-M., and Kim, J.-J.,
Enhancement of near-infrared absorption with high fill factor in lead phthalocyanine-based
organic solar cells. Journal of Materials Chemistry, 2012. 22(18): p. 9077.

124

180.

Shea, M. J. and Arnold, M. S., 1% solar cells derived from ultrathin carbon nanotube
photoabsorbing films. Applied Physics Letters, 2013. 102(24): p. 243101.

181.

Xia, Y., Wang, L., Deng, X., Li, D., Zhu, X., and Cao, Y., Photocurrent response wavelength up
to 1.1 Îź m from photovoltaic cells based on narrow-band-gap conjugated polymer and fullerene
derivative. Applied physics letters, 2006. 89(8): p. 081106.

182.

Zimmerman, J. D., Diev, V. V., Hanson, K., Lunt, R. R., Yu, E. K., Thompson, M. E., and Forrest,
S. R., Porphyrin-tape/c(60) organic photodetectors with 6.5% external quantum efficiency in the
near infrared. Adv Mater, 2010. 22(25): p. 2780-3.

183.

Bertolino, C. A., Caputo, G., Barolo, C., Viscardi, G., and Coluccia, S., Novel heptamethine
cyanine dyes with large Stoke's shift for biological applications in the near infrared. J Fluoresc,
2006. 16(2): p. 221-5.

184.

Khlyabich, P. P., Burkhart, B., and Thompson, B. C., Efficient ternary blend bulk heterojunction
solar cells with tunable open-circuit voltage. J Am Chem Soc, 2011. 133(37): p. 14534-7.

185.

Khlyabich, P. P., Rudenko, A. E., Thompson, B. C., and Loo, Y.-L., Structural Origins for Tunable
Open-Circuit Voltage in Ternary-Blend Organic Solar Cells. Advanced Functional Materials, 2015.
25(34): p. 5557-5563.

186.

Osedach, T. P., Iacchetti, A., Lunt, R. R., Andrew, T. L., Brown, P. R., Akselrod, G. M., and
Bulović, V., Near-infrared photodetector consisting of J-aggregating cyanine dye and metal oxide
thin films. Applied Physics Letters, 2012. 101(11): p. 113303.

187.

Blakesley, J. C. and Neher, D., Relationship between energetic disorder and open-circuit voltage
in bulk heterojunction organic solar cells. Physical Review B, 2011. 84(7).

188.

Jenatsch, S., Geiger, T., Heier, J., Kirsch, C., Nuesch, F., Paracchino, A., Rentsch, D., Ruhstaller,
B., A, C. V., and Hany, R., Influence of chemically p-type doped active organic semiconductor on
the film thickness versus performance trend in cyanine/C60 bilayer solar cells. Sci Technol Adv
Mater, 2015. 16(3): p. 035003.

189.

Kahn, A., Koch, N., and Gao, W., Electronic structure and electrical properties of interfaces
between metals and π‐conjugated molecular films. Journal of Polymer Science Part B: Polymer
Physics, 2003. 41(21): p. 2529-2548.

190.

Brown, P. R., Lunt, R. R., Zhao, N., Osedach, T. P., Wanger, D. D., Chang, L. Y., Bawendi, M. G.,
and Bulovic, V., Improved current extraction from ZnO/PbS quantum dot heterojunction
photovoltaics using a MoO3 interfacial layer. Nano Lett, 2011. 11(7): p. 2955-61.

191.

Hill, I., Kahn, A., Soos, Z., and Pascal Jr, R., Charge-separation energy in films of π-conjugated
organic molecules. Chemical Physics Letters, 2000. 327(3): p. 181-188.

192.

Kraner, S., Scholz, R., Koerner, C., and Leo, K., Design Proposals for Organic Materials
Exhibiting a Low Exciton Binding Energy. The Journal of Physical Chemistry C, 2015. 119(40): p.
22820-22825.

193.

Dkhissi, A., Excitons in organic semiconductors. Synthetic Metals, 2011. 161(13-14): p. 14411443.
125

194.

Knupfer, M., Exciton binding energies in organic semiconductors. Applied Physics A, 2003. 77(5):
p. 623-626.

195.

Cho, N., Schlenker, C. W., Knesting, K. M., Koelsch, P., Yip, H.-L., Ginger, D. S., and Jen, A. K.
Y., High-Dielectric Constant Side-Chain Polymers Show Reduced Non-Geminate Recombination
in Heterojunction Solar Cells. Advanced Energy Materials, 2014. 4(10): p. 1301857.

196.

Kraner, S., Koerner, C., Leo, K., Bittrich, E., Eichhorn, K. J., Karpov, Y., Kiriy, A., Stamm, M.,
Hinrichs, K., and Al-Hussein, M., Dielectric function of a poly(benzimidazobenzophenanthroline)
ladder polymer. Physical Review B, 2015. 91(19).

197.

Zhang, H., Jenatsch, S., De Jonghe, J., Nuesch, F., Steim, R., Veron, A. C., and Hany, R.,
Transparent organic photodetector using a near-infrared absorbing cyanine dye. Sci Rep, 2015. 5:
p. 9439.

198.

Arnold, M. S., Zimmerman, J. D., Renshaw, C. K., Xu, X., Lunt, R. R., Austin, C. M., and Forrest,
S. R., Broad spectral response using carbon nanotube/organic semiconductor/C60 photodetectors.
Nano letters, 2009. 9(9): p. 3354-3358.

199.

Xie, Y., Gong, M., Shastry, T. A., Lohrman, J., Hersam, M. C., and Ren, S., Broad-spectralresponse nanocarbon bulk-heterojunction excitonic photodetectors. Adv Mater, 2013. 25(25): p.
3433-7.

200.

Su, Z., Hou, F., Wang, X., Gao, Y., Jin, F., Zhang, G., Li, Y., Zhang, L., Chu, B., and Li, W., Highperformance organic small-molecule panchromatic photodetectors. ACS Appl Mater Interfaces,
2015. 7(4): p. 2529-34.

201.

Bulovic, V. and Lunt, R. R. (2012), Transparent photovoltaic cells. 2012, Google Patents.

202.

Zhang, M., Wang, H., Tian, H., Geng, Y., and Tang, C. W., Bulk heterojunction photovoltaic cells
with low donor concentration. Adv Mater, 2011. 23(42): p. 4960-4.

203.

Gallagher, S. J., Norton, B., and Eames, P. C., Quantum dot solar concentrators: Electrical
conversion efficiencies and comparative concentrating factors of fabricated devices. Solar Energy,
2007. 81(6): p. 813-821.

204.

Guo, F., Kubis, P., Przybilla, T., Spiecker, E., Hollmann, A., Langner, S., Forberich, K., and Brabec,
C. J., Nanowire Interconnects for Printed Large-Area Semitransparent Organic Photovoltaic
Modules. Advanced Energy Materials, 2015. 5(12): p. 1401779.

205.

Gagnon, P., Margolis, R., Melius, J., Phillips, C., and Elmore, R. Rooftop solar photovoltaic
technical potential in the United States: A detailed assessment. 2016; Available from:
www.nrel.gov/docs/fy16osti/65298.pdf.

206.

Hammarberg, E. and Roos, A., Antireflection treatment of low-emitting glazings for energy efficient
windows with high visible transmittance. Thin Solid Films, 2003. 442(1): p. 222-226.

207.

Ubiquitous Energy’s Truly Transparent Solar Technology Demonstrated Globally in BMW Brand
Stores. 2016 28 September 2016]; Available from: http://ubiquitous.energy/bmw_iexhibition/.

126

208.

Baker, D. Elon Musk's solar roof is pretty, but it's not the first. 2016 30 October 2016]; Available
from:
http://www.sfchronicle.com/business/article/Elon-Musk-s-solar-roof-isn-t-the-first10418690.php.

209.

Fisette, P. Windows: Understanding Energy Efficient Performance. 2003 28 September 2016];
Available from: https://bct.eco.umass.edu/publications/by-title/windows-understanding-energyefficient-performance/.

210.

Berson, D. M., Dunn, F. A., and Takao, M., Phototransduction by Retinal Ganglion Cells That Set
the Circadian Clock. Science, 2002. 295(5557): p. 1070-1073.

211.

Traverse, C. J., Young, M., Wagner, S., Zhang, P., Askeland, P., Barr, M. C., and Lunt, R. R.,
Efficient zinc sulfide cathode layers for organic photovoltaic applications via n-type doping.
Journal of Applied Physics, 2014. 115(19): p. 194505.

212.

Min-hyung, L. Thorough inspection brings perfection to LG's OLED TV. 2016.

213.

Jørgensen, M., Norrman, K., and Krebs, F. C., Stability/degradation of polymer solar cells. Solar
Energy Materials and Solar Cells, 2008. 92(7): p. 686-714.

214.

Cheacharoen, R., Mateker, W. R., Zhang, Q., Kan, B., Sarkisian, D., Liu, X., Love, J. A., Wan, X.,
Chen, Y., Nguyen, T.-Q., Bazan, G. C., and McGehee, M. D., Assessing the stability of high
performance solution processed small molecule solar cells. Solar Energy Materials and Solar Cells,
2017. 161: p. 368-376.

215.

Roesch, R., Eberhardt, K.-R., Engmann, S., Gobsch, G., and Hoppe, H., Polymer solar cells with
enhanced lifetime by improved electrode stability and sealing. Solar Energy Materials and Solar
Cells, 2013. 117: p. 59-66.

216.

Lungenschmied, C., Dennler, G., Czeremuzskin, G., Latrèche, M., Neugebauer, H., and Sariciftci,
N. S. Flexible encapsulation for organic solar cells. in SPIE 6197, Photonics for Solar Energy
Systems. 2006.

217.

Tong, X., Wang, N., Slootsky, M., Yu, J., and Forrest, S. R., Intrinsic burn-in efficiency loss of
small-molecule organic photovoltaic cells due to exciton-induced trap formation. Solar Energy
Materials and Solar Cells, 2013. 118: p. 116-123.

218.

Intes, X., Ripoll, J., Chen, Y., Nioka, S., Yodh, A. G., and Chance, B., In vivo continuous‐wave
optical breast imaging enhanced with Indocyanine Green. Medical Physics, 2003. 30(6): p. 10391047.

219.

Castano, A. P., Demidova, T. N., and Hamblin, M. R., Mechanisms in photodynamic therapy: part
one—photosensitizers, photochemistry and cellular localization. Photodiagnosis and
Photodynamic Therapy, 2004. 1(4): p. 279-293.

220.

Zheng, M., Zhao, P., Luo, Z., Gong, P., Zheng, C., Zhang, P., Yue, C., Gao, D., Ma, Y., and Cai,
L., Robust ICG Theranostic Nanoparticles for Folate Targeted Cancer Imaging and Highly
Effective Photothermal Therapy. ACS Applied Materials & Interfaces, 2014. 6(9): p. 6709-6716.

221.

Song, X., Chen, Q., and Liu, Z., Recent advances in the development of organic photothermal nanoagents. Nano Research, 2015. 8(2): p. 340-354.
127

222.

Hadjur, C., Wagnières, G., Ihringer, F., Monnier, P., and van den Bergh, H., Production of the free
radicals O.-2 and. OH by irradiation of the photosensitizer zinc (II) phthalocyanine. Journal of
Photochemistry and Photobiology B: Biology, 1997. 38(2): p. 196-202.

128