Embed Size (px)
Citation preview
Photonic Enhancement of Colloidal Quantum Dot Photovoltaics
by
André Jean-Roméo Richard Labelle
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Edward S. Rogers Department of Electrical and Computer Engineering University of Toronto
© Copyright by André Labelle 2015
ii
Photonic Enhancement of
Colloidal Quantum Dot Photovoltaics
André Jean-Roméo Richard Labelle
Doctor of Philosophy
Edward S. Rogers Department of Electrical and Computer Engineering
University of Toronto
2015
Abstract
Colloidal quantum dots, nanocrystal semiconductors that can be cross-linked and assembled into
absorbing thin films, are an attractive material for third-generation photovoltaic applications due
to low-cost fabrication and bandgap tunability.
As a result of their limited charge transport, these solution-processed thin films suffer from a
mismatch in absorption length and charge extraction length. Concepts based on the
interdigitation of n- and p-doped layers, approaches that reduce the distance photogenerated
carriers must travel before extraction, offer promise on overcoming this limitation. In this thesis,
I explore and develop techniques to address the absorption-extraction compromise in CQD
materials by implementing nano- and micro-structuring techniques to enhance light absorption in
the active film.
First, I focus on the development of nanomaterials for light guiding/scattering enhancement in
CQD films. For this, I develop a nanostructured gold reflector that, when suitably designed,
guides light and traps it within the active film. I show that this yields enhanced broadband
absorption with more than 4-fold improvement at the most improved wavelength, which
translated into a 34% improvement in photocurrent in a working solar cell. I also show that
iii
periodic nanostructures employed for absorption enhancement can lead to improvements in solar
cell performance. Limitations in device architecture and film formation, however, prevented
significant performance advances for these nano-scale approaches. Regardless, these early results
pointed me to a new and more impactful strategy.
I focus in on realizing micron-scale structured electrodes to enhance absorption, which I show to
be considerably more useful in view of the need to extract charge carriers with high efficiency. I
discover that conformal film formation atop these structured electrodes is an absolute
prerequisite to enhancing performance. These devices, which I term micro-pyramid CQD cells,
provide a 24% enhancement in current and a consequent 15% improvement in power conversion
efficiency.
Ultimately, this work offers a prescription for photonic enhancements of solar cells within this
emerging class of photovoltaic materials.
iv
Acknowledgments
I would like to begin by thanking my supervisor, Professor Ted Sargent. My experience as a PhD
student has been educational from both an academic and general career perspective. Ted has
taught me the value of persistence and work ethic, and pushed me to the limits of my potential. I
feel this experience will be the most influential in helping me forge a successful career.
I would also like to thank the technicians/research associates in Sargent lab who made all our
research possible, Damir Kopilovic, Remi Wolowiec, Elenita Palmiano, and Dr. Larissa Levina.
All have been incredibly helpful at all times throughout my PhD and I credit them with keeping
the laboratory in order and supporting all my research activities. I would also like to thank the
administrative personnel in the group, Jeannie Ing and Vienna Hehir, whom both worked
extremely hard to organize events and maintain all administrative functions of the group. I’d like
to thank group directors Lukasz Brzozwski and Sjoerd Hoogland who have provided guidance
throughout my degree. I would finally like to include a special thanks to Susanna Thon who was
instrumental in helping get my most significant work off the ground and helped mentor me into a
becoming a productive PhD student.
I would also like to thank all my co-workers who have been actively involved in academic and
non-academic collaboration throughout the years and many of whom I consider to be good
friends: Kyle Kemp, David Zhitomirsky, Alex Ip, Lisa Rollny, Jeff McDowell, Valerio Adinolfi,
Michael Adachi, Xinzheng Lan, Bruce Xu, Graham Carey, Amirreza Kiani, Hamidreza
Movahed, Dan Sellan, Zhijun Ning, Jin Young Kim, Ratan Debnath, Dan Paz-Soldan, Anna Lee,
Illan Kramer, Ghada Koleilat, and Mingjian Yuan. And a special thanks to my arch nemeses,
Gabriel Moreno-Bautista and Brandon Sutherland, whose pranks and antics forced me to sharpen
my wits throughout the years.
I would like to thank my personal friends, who have supported me and helped keep me
grounded/not forget my true self throughout this entire process: Luc Robidas, Jen Robidas, Dan
Blais, Jon Diplock, Denis Sonier, Jones, Syl, and the entire crew from back home. Special thanks
to my academic friends from back home who have been equally supportive, with the added
caveat of living through the same experiences: Dr. Ryan Mailloux, Dr. Julie Gosselin, Dr. Chris
Auger, and Dr. Joe Lemire.
v
I would finally like to thank my girlfriend, Hannah Leveque, who has provided love and support
since the beginning of my PhD, and stuck it out the entire way. And of course my parents,
Hubert and Jo-Anne Labelle, whom have guided me through life and to whom I credit this
marvelous accomplishment.
I would like to acknowledge funding awarded during the course of my PhD, including
scholarships from the University of Toronto (Rogers scholarship, Doctoral completion award),
the government of Ontario (Ontario graduate scholarship, Queen Elizabeth II graduate
scholarship in science and technology), and additional financial support provided by grants
awarded from the King Abdullah University of Science and Technology, Saudi Arabia.
vi
Contributions
I. Thesis related contributions
1. A.J. Labelle, M. Bonifazi, Y. Tian, C. Wong, S. Hoogland, J. McDowell, B.R.
Sutherland, S.O. Kelley, E.H. Sargent, A. Fratalocchi. “Broadband Epsilon-Near-Zero
Reflectors for Ultra-Thin Light Harvesting Systems”, in preparation.
This work forms the content of Chapter 3. I designed and performed most of the experimental
work for this manuscript, including the design and implementation of electroplating
techniques for nanostructured Au electrodes, preparation of CQD films, and characterization
with spectrophotometer and photoluminescence excitation. The design of top-illuminated
CQD solar cells was performed by me and Chris Wong; preparation and characterization
was performed by Chris Wong. The manuscript draft was prepared by myself and
collaborator Professor Andrea Fratalocchi, with input from Professor Ted Sargent. The
epsilon-near-zero concept was conceived and explained by Andrea Fratalocchi, and
simulations were performed by Yi Tian and Marcella Bonifazi.
2. A.J. Labelle, S.M. Thon, J. Kim, X. Lan, D. Zhitomirsky, K.W. Kemp, E.H. Sargent.
“Conformal Fabrication of Colloidal Quantum Dot Solids for Optically Enhanced
Photovoltaics”, (2015) ACS Nano, 9, 5447-5453.
This work forms the content of Chapter 5. All experiments were designed by myself and
Susanna Thon. I performed all FDTD optical and SCAPS optoelectronic simulations for this
project, while David Zhitomirksy performed Sentaurus simulations. I prepared all
microstructured substrates and solar cells, and characterized them all (J-V under simulated
solar illumination, EQE, absorption) and performed all calculations to extract IQE values. I
acquired all SEM images. Fabrication techniques were optimized by myself and Xinzheng
Lan. Contact angle measurements were carried out by myself and Jin Young Kim. I also
wrote the manuscript in its entirety, with edits from Susanna Thon and Ted Sargent.
3. A.J. Labelle, S.M. Thon, S. Masala, M.M. Adachi, H. Dong, M. Farahani, A.H. Ip, A.
Fratalocchi, E.H. Sargent. “Colloidal Quantum Dot Solar Cells Exploiting Hierarchical
Structuring”, (2014) Nano Letters, 15, 1101-1108.
This work forms the content of Chapter 6. In this manuscript I performed all FDTD
simulations with assistance/guidance from Michael Adachi and Maryam Farahani. The
silicon master for pyramid-patterned stamps was created by Silvia Masala. I carried out soft
lithography and optimized the transfer stamping process to create pyramid-patterned TiO2
electrodes. I prepared all solar cells and characterizations, optimized performance, and
vii
acquired all SEM images in the manuscript. I also wrote the manuscript in its entirety with
edits from Susanna Thon and Ted Sargent.
4. M.M. Adachi, A.J. Labelle, S.M. Thon, X. Lan, S. Hoogland, E.H. Sargent. “Broadband
Solar Absorption Enhancement Via Periodic Nanostructuring of Electrodes”, (2013)
Scientific Reports, 3, 2928.
This work forms the content of Chapter 4. For this manuscript I designed and optimized the
fabrication techniques applied for nanostructured substrates. Michael Adachi performed
FDTD simulations and prepared the substrates for the project. I fabricated solar cells which
were characterized by Michael Adachi. Michael wrote the manuscript which I edited for
content and style, with additional input from Ted Sargent.
II. Non-thesis related contributions
1. V. Adinolfi, I.J. Kramer, A.J. Labelle, B.R. Sutherland, S. Hoogland, E.H. Sargent.
“Photojunction Field-Effect Transistor Based on a Colloidal Quantum Dot Absorber
Channel Layer”, (2015) ACS Nano, 9, 356-362.
2. Z. Ning, O. Voznyy, J. Pan, S. Hoogland, V. Adinolfi, J. Xu, M. Li, A. R. Kirmani, J.
Sun, J. Minor, K. W. Kemp, H. Dong, L. Rollny, A.J. Labelle, G. Carey, B. Sutherland,
I. Hill, A. Amassian, H. Liu, J. Tang, O. M. Bakr, E. H. Sargent. “Air-Stable N-type
Colloidal Quantum Dot Solids”, (2014) Nature Materials, 13, 822-828.
3. A.H. Ip, A.J. Labelle, E.H. Sargent. “Efficient, Air-Stable Colloidal Quantum Dot Solar
Cells Encapsulated Using Atomic Layer Deposition of a Nanolaminate Barrier”, (2013)
Applied Physics Letters, 103, 263905.
4. G. Koleilat, I.J. Kramer, C. Wong, S.M. Thon, A.J. Labelle, S. Hoogland, E.H. Sargent.
"Folded-Light-Path Colloidal Quantum Dot Solar Cells", (2013) Scientific Reports, 3,
2166.
5. P. Maraghechi, A.J. Labelle, A.R. Kirmani, X. Lan, M. Adachi, S.M. Thon, S.
Hoogland, A. Lee, Z. Ning, A. Fischer, A. Amassian, E.H. Sargent. “The Donor-Supply
Electrode Enhances Performance in Colloidal Quantum Dot Solar Cells”, (2013) ACS
Nano, 7, 6111-6116.
viii
6. K.W. Kemp, A.J. Labelle, S.M. Thon, A.H. Ip, I.J. Kramer, S. Hoogland, E.H. Sargent.
"Interface Recombination in Depleted Heterojunction Photovoltaics Based on Colloidal
Quantum Dots", (2013) Advanced Energy Materials, 3, 917-922.
7. D. Paz-Soldan, A. Lee, S.M. Thon, M.M. Adachi, H. Dong, P. Maraghechi, M. Yuan,
A.J. Labelle, S. Hoogland, K. Liu, E. Kumacheva, E.H. Sargent. “Jointly Tuned
Plasmonic−Excitonic Photovoltaics Using Nanoshells”, (2013) Nano Letters, 13, 1502-
1508.
8. X. Lan, J. Bai, S. Masala, S.M. Thon, Y. Ren, I.J. Kramer, S. Hoogland, A. Simchi, G.I.
Koleilat, D. Paz-Soldan, Z. Ning, A.J. Labelle, J. Kim, G. Jabbour, E.H. Sargent. “Self-
Assembled, Nanowire Network Electrodes for Depleted Bulk Heterojunction Solar Cells”,
(2013) Advanced Materials, 25, 1769-1773.
9. A.H. Ip, S.M. Thon, S. Hoogland, D. Zhitomirsky, R. Debnath, L. Levina, L.R. Rollny,
G.H. Carey, O. Voznyy, A. Fischer, K.W. Kemp, I.J. Kramer, Z. Ning, A.J. Labelle, K.
Wei Chou, A. Amassian, E.H. Sargent. “Hybrid Passivated Colloidal Quantum Dot
Solids”, (2012) Nature Nanotechnology, 7, 577-582.
10. D. Zhitomirsky, I.J. Kramer, A.J. Labelle, A. Fischer, R. Debnath, J. Pan, O.M. Bakr, E.H.
Sargent. “Colloidal Quantum Dot Photovoltaics: The Effects of Polydispersity”, (2012)
Nano Letters, 12, 1007-1012.
11. G.I. Koleilat, X. Wang, A.J. Labelle, A.H. Ip, G.H. Carey, A. Fischer, L. Levina, L.
Brzozowski, E.H. Sargent. “A Donor-Supply Electrode (DSE) for Colloidal Quantum Dot
Photovoltaics”, (2011) Nano Letters, 11, 5173-5178.
ix
Table of Contents
Acknowledgments ................................................................................................................... iv
Contributions ........................................................................................................................... vi
Table of Contents ..................................................................................................................... ix
List of Tables .......................................................................................................................... xii
List of Figures ........................................................................................................................ xiii
List of Appendices ................................................................................................................. xvi
List of Commonly Used Acronyms ...................................................................................... xvii
Chapter 1 Introduction and Motivation...................................................................................... 1
1.1 Renewable Energy .......................................................................................................... 1
1.2 Photovoltaics ................................................................................................................... 2
1.2.1 History .................................................................................................................. 2
1.2.2 Third-Generation Photovoltaics ............................................................................ 4
1.3 Thesis Objective .............................................................................................................. 6
1.3.1 Absorption-Extraction Compromise ..................................................................... 6
1.3.2 Photonic Enhancement ......................................................................................... 7
1.4 Thesis Outline ................................................................................................................. 8
Chapter 2 Colloidal Quantum Dot Photovoltaics: Key Challenges ......................................... 10
2.1 Solar Spectrum .............................................................................................................. 10
2.2 Operation of Photovoltaics ............................................................................................ 11
2.3 Colloidal Quantum Dots ............................................................................................... 15
2.4 Colloidal Quantum Dot Solar Cells .............................................................................. 17
2.5 Strategies for Overcoming the Absorption-Extraction Compromise ............................ 21
2.5.1 Strategies for Colloidal Quantum Dot Solar Cells .............................................. 21
x
2.5.2 Strategies Employed in Other Material Systems ................................................ 24
2.6 Conclusion .................................................................................................................... 26
Chapter 3 Nanostructured Disordered Reflectors for Thin-Film Photovoltaics ...................... 27
3.1 Introduction ................................................................................................................... 27
3.2 Nanostructured Gold (Au) Reflectors ........................................................................... 27
3.3 Colloidal Quantum Dot Absorption on Nanostructured Au Reflectors ........................ 30
3.3.1 Experimental Results .......................................................................................... 30
3.3.2 Finite-Difference Time-Domain (FDTD) Modelling ......................................... 34
3.3.3 Photoluminescence Excitation ............................................................................ 35
3.4 Solar Cells Employing Nanostructured Au Reflectors ................................................. 38
3.5 Conclusion .................................................................................................................... 40
Chapter 4 Broadband Absorption Enhancement via Periodic Nanostructuring of Electrodes 41
4.1 Introduction ................................................................................................................... 41
4.2 Periodic Nanostructures for Photon Management ........................................................ 41
4.3 Nanostructure Design .................................................................................................... 42
4.4 Nanostructured Substrate Fabrication ........................................................................... 45
4.5 Broadband Absorption Enhancement in a Solar Cell ................................................... 48
4.6 Conclusion .................................................................................................................... 49
Chapter 5 Conformal Fabrication of Colloidal Quantum Dot Solids for Optically Enhanced
Photovoltaics ....................................................................................................................... 51
5.1 Introduction ................................................................................................................... 51
5.2 Colloidal Quantum Dot Film: Transport Properties ..................................................... 51
5.3 Impact of Non-Conformal CQD Films ......................................................................... 56
5.4 Technique for Conformal CQD Film Deposition ......................................................... 57
5.5 Enhanced Solar Cell Performance with Conformal CQD Films .................................. 59
5.6 Conclusion .................................................................................................................... 63
xi
Chapter 6 Colloidal Quantum Dot Solar Cells Exploiting Hierarchical Structuring ............... 65
6.1 Introduction ................................................................................................................... 65
6.2 Effect of Pyramid Sidewall Angle ................................................................................ 65
6.3 Hierarchical Structuring for Optimal Broadband Enhancement ................................... 71
6.4 Fabrication of Pyramid-Patterned Electrodes ............................................................... 73
6.5 Absorption and External Quantum Efficiency Enhancement ....................................... 75
6.6 High-Efficiency Colloidal Quantum Dot Solar Cells ................................................... 77
6.7 Conclusion .................................................................................................................... 79
Chapter 7 Conclusions and Future Work ................................................................................. 81
7.1 Conclusion .................................................................................................................... 81
7.2 Impact of the Work ....................................................................................................... 82
7.3 Future Work .................................................................................................................. 83
7.3.1 Higher Aspect-Ratio Pyramids ........................................................................... 83
7.3.2 Standardized Conformal Deposition Techniques ............................................... 84
7.3.3 Standardized Pyramid-Patterned Electrodes for Different Material Stacks ....... 85
7.3.4 Efficient Top-Illuminated CQD Architectures ................................................... 85
7.3.5 Bottom-Illuminated Nanostructured Disordered Electrodes .............................. 86
7.4 Perspectives of CQD Photovoltaics .............................................................................. 86
7.5 Concluding Remarks ..................................................................................................... 86
References ................................................................................................................................ 88
Appendices ............................................................................................................................... 96
xii
List of Tables
Table 3-1. Summary of CQD solar cell performance as a function of nanostructured Au
roughness (as expressed using correlation length, Lc) .................................................................. 39
xiii
List of Figures
Figure 1-1 Global renewable electricity production by region........................................................ 1
Figure 1-2. World renewable power capacity, including projections to 2020 ................................ 2
Figure 1-3 Global cumulative installed capacity for photovoltaics from 2000-2013. .................... 4
Figure 1-4. Chart of best research-cell efficiencies according to technology. ................................ 5
Figure 1-5. Absorption-extraction compromise. ............................................................................. 7
Figure 2-1. Spectral irradiance of 5777 K blackbody emitter, AM0 solar spectrum and AM1.5
solar spectrum. ............................................................................................................................. 11
Figure 2-2. Current density-voltage (JV) plot for solar cell. .......................................................... 13
Figure 2-3. Solar cell efficiency as a function of bandgap for solar concentrations of 1 and 1000
suns ............................................................................................................................................... 15
Figure 2-4. Synthesis mechanisms of colloidal quantum dots ...................................................... 16
Figure 2-5. Relationship between quantum dot size and bandgap .............................................. 17
Figure 2-6. Structure of the Schottky CQD solar cell ..................................................................... 18
Figure 2-7. Structure of the depleted heterojunction CQD solar cell ............................................ 19
Figure 2-8. Model and charge transport in CQD solar cells .......................................................... 20
Figure 2-9. Concept of the depleted bulk heterojunction CQD solar cell ...................................... 22
Figure 2-10. Concept of the folded-light path CQD solar cell ....................................................... 23
Figure 2-11. Concept of the plasmonic-excitonic CQD solar cell................................................... 24
Figure 2-12. Structured substrate concept as applied to amorphous silicon solar cells. .............. 25
Figure 3-1. SEM images of the nanostructured Au reflectors using 13 mM HAuCl4 solution. ...... 28
Figure 3-2. Cross-sectional SEM of nanostructured Au reflector with active films....................... 29
Figure 3-3. SEM images of the nanostructured Au reflectors prepared with exposure times of 10
minutes ......................................................................................................................................... 30
Figure 3-4. Total absorption of CQD films atop nanostructured Au reflectors ............................. 31
Figure 3-5. Absorption profiles for 100 nm CQD films atop nanostructured Au reflectors. ......... 32
Figure 3-6. Absorption profiles for 300 nm CQD films atop nanostructured Au reflectors .......... 33
Figure 3-7. FDTD simulated absorption profiles for CQD films atop nanostructured Au reflectors
...................................................................................................................................................... 35
xiv
Figure 3-8. Simulated energy field plot to illustrate where light energy is concentrated ............ 35
Figure 3-9. Photoluminescence excitation (PLE) ........................................................................... 38
Figure 4-1. Diagram of the nanostructured colloidal quantum dot solar cell .............................. 43
Figure 4-2. FDTD simulations of the nanostructured colloidal quantum dot solar cell ................ 44
Figure 4-3. Fabrication of nanostructured substrates .................................................................. 46
Figure 4-4. Cross-sectional view of the nanostructured CQD solar cell ........................................ 47
Figure 4-5. Total absorption of nanostructured CQD solar cells ................................................... 48
Figure 4-6. Device performance of nanostructured CQD solar cell............................................... 49
Figure 4-7. Non-conformality of thick CQD films over nanostructured substrate ........................ 50
Figure 5-1. FDTD simulations for periodic micro-structures as compared to planar control ....... 53
Figure 5-2. Simulated electric field distribution in a PbS CQD film (using the Sentaurus
optoelectronic device modeling engine) ....................................................................................... 54
Figure 5-3. SCAPS simulation results ............................................................................................. 55
Figure 5-4. Non-conformal CQD films ........................................................................................... 56
Figure 5-5. Non-conformal CQD films – SEM ................................................................................ 57
Figure 5-6. Contact angle measurements of wetting-engineered surface ................................... 59
Figure 5-7. Conformal CQD films .................................................................................................. 60
Figure 5-8. Quantum efficiency for conformal and non-conformal CQD films ............................. 62
Figure 5-9. Device performance comparing conformal CQD case with planar control ................ 63
Figure 6-1. Diagram illustrating the increased-light-path advantage of pyramid-patterned
electrodes ...................................................................................................................................... 66
Figure 6-2. More detailed diagram of the light path through the pyramid-patterned CQD film . 67
Figure 6-3. Theoretical predictions for pyramid-patterned thin-film solar cell as a function of
sidewall angle ............................................................................................................................... 69
Figure 6-4. Evolution of projected Voc, Jsc and PCE as a function of pyramid sidewall angle ....... 70
Figure 6-5. IQE curve for a standard CQD solar cell ...................................................................... 71
Figure 6-6. Results of 3D FDTD simulations of projected Jph as a function of pyramid pitch........ 72
Figure 6-7. Fabrication of pyramid-patterned titania (TiO2) electrodes ....................................... 74
Figure 6-8. SEM images of pyramid-patterned TiO2 electrodes. .................................................. 75
xv
Figure 6-9. SEM images of full CQD devices ................................................................................. 76
Figure 6-10. Enhanced absorption and external quantum efficiency for pyramid-patterned CQD
solar cell ........................................................................................................................................ 77
Figure 6-11. Enhanced performance of pyramid-patterned CQD thin-film solar cell ................... 79
Figure 7-1. Techniques for preparing higher angle pyramid-patterned electrodes ..................... 84
xvi
List of Appendices
A1 CQD Solar Cell Fabrication .......................................................................................... 96
A2 Optical and Device Modelling ...................................................................................... 96
A3 Electrode Preparation .................................................................................................... 97
A4 Contact Deposition ........................................................................................................ 99
A5 Electrodeposition ......................................................................................................... 100
A6 Contact Angle .............................................................................................................. 101
A7 Scanning Electron Microscopy ................................................................................... 101
A8 Solar Simulator ............................................................................................................ 102
A9 Spectrophotometer ...................................................................................................... 103
A10 Quantum Efficiency .................................................................................................. 103
xvii
List of Commonly Used Acronyms
CQD – colloidal quantum dots
Jsc – short-circuit current density
Voc – open-circuit voltage
FF – fill factor
PCE – power conversion efficiency
PbS – lead-sulfide (specific CQD material)
EQE – external quantum efficiency
IQE – internal quantum efficiency
3-MPA – 3-mercaptopropionic acid
Lc – correlation length
havg – average feature height
PV – photovoltaic
AM1.5 – air mass 1.5
CIGS – cadmium indium gallium selenide
CdTd – cadmium telluride
CZTS – copper zinc tin sulfide-selenide
SEM – scanning electron microscopy
FIB – focused ion beam
FDTD – finite-difference time-domain
ITO – tin-doped indium oxide
TiO2 – titanium dioxide
1
Chapter 1 Introduction and Motivation
1.1 Renewable Energy
Global energy consumption was 5.6 x 1020 J/year, or roughly 17.8 TW on average in 2012, with
fossil-fuel based resources (i.e., oil, coal and natural gas) making up 82% of this figure1.
Approximately 8.16 x 1019 J of this energy is used for electricity generation, or 15%; fossil fuels
now account for 68% of this global electricity generation. Remaining electricity demand is met
using nuclear (11%) and renewable resources (21%). Hydroelectric is the largest renewable
source, meeting 16% of total electricity demand1.
Nuclear energy is considered a clean alternative to fossil-fuel based energy sources; however,
there exists growing concern over the safety of nuclear fission power plants, and fissile elements
require careful and costly disposal. Fossil fuels present significant challenges in their
environmental impact and long-term sustainability. For these reasons, there exists an ever-
growing desire to expand the deployment of renewable energy resources. Electricity generated
from renewables will increase by approximately 2000 TWh by 2020, and will provide more than
25% of total electricity generation (Figure 1-1)2.
Figure 1-1 Global renewable electricity production by region, including projections to 2020. Based on IEA data from Key World Energy Statistics 2014 © OECD/IEA 2014, IEA Publishing; modified by André Labelle. Licence: www.iea.org/t&c/termsandconditions/#d.en.26167.
2
It is anticipated that photovoltaics and wind energy will experience the largest growth in that
period (Figure 1-2)2. This work focuses on solar-based technologies, namely photovoltaic cells,
which offer tremendous potential for growth in electricity generation in light of the abundance of
solar power reaching the earth’s surface.
Figure 1-2. World renewable power capacity, including projections to 2020. Global capacity for wind and PV energy are projected to experience the most growth by 2020 under baseline, enhanced low and enhanced high models. Based on IEA data from Renewable Energy Medium-Term Market Report 2014 © OECD/IEA 2014, IEA Publishing; modified by André Labelle. Licence: www.iea.org/t&c/termsandconditions/#d.en.26167.
1.2 Photovoltaics
Photovoltaics rely on the vast power supply that is the Sun. With an average irradiance of 6
kWh/m2 incident on the Earth every day, enough energy irradiates 0.74% (3.8 x 106 km2) of the
Earth’s surface in one day to satisfy global electricity demand, in principle, for one year.
1.2.1 History
In 1839, Edmond Becquerel noted the production of an electric current when two plates of
platinum or gold were submerged in a solution of variable pH and exposed unevenly to solar
light3. Several decades following this discovery, the first solar cell was developed: in 1883,
Charles Fritts made the first solar cell using selenium (Se) wafers, in which he noted that
conductivity was enhanced in the Se wafer when it was exposed to solar light4. It was not until
3
1954 that the first practically efficient solar cell was developed. In his investigation, Daryl
Chapin (Bell Labs) observed the efficient separation of photoexcited charge carriers in a silicon
wafer that included p- and n-doped regions. This yielded a power conversion efficiency (PCE) of
6%5. This discovery laid the foundation of future solar cell designs.
To date, the most practically successful technologies fall into one of two categories known as
first- or second- generation solar cells. The first generation consists of high quality silicon (Si)
wafers and boast among the highest PCE of all single-junction solar cell technologies. The best
PCE (for single-junction, unconcentrated cells) belonged to the PERL cell based on a single-
crystal Si wafer with additional inverse-pyramid surface structures to reduce reflection and
enhance light penetration into the cell, with a PCE of 25%6 until it was recently surpassed by the
HIT (heterojunction with intrinsic thin layer) cell with a PCE of 26%6. Second-generation cells
include lower-materials-utilization alternatives such as the thin-film technologies CIGS, CdTe,
and amorphous Si. They seek to reduce fabrication costs to offer a competitive edge in cost per
watt electricity produced. As of 2013 the total installed capacity of photovoltaic cells was
estimated at almost 139 GW7 (Figure 1-3)7, and it was dominated by first-generation crystalline
Si technologies (multicrystalline Si followed by monocrystalline/single-crystal Si).
4
Figure 1-3 Global cumulative installed capacity for photovoltaics from 2000-2013. Based on EPIA data from Global Market Outlook for Photovoltaics 2014-2018 © EPIA 2014.
1.2.2 Third-Generation Photovoltaics
The field of third-generation photovoltaics spans a large number of emerging technologies that
seek to drive down overall costs (through fabrication, material utilization, balance-of-system
costs) while achieving high efficiency. If successful, this strategy could lower the cost of solar
electricity production. Technologies under exploration include organic/polymer solar cells,
CZTS, perovskites, colloidal quantum dots, concentrated solar cells, and multi-junction
architectures based on III-V semiconductor materials. We observe a rapid progression in
certified PCE for emerging technologies (orange lines), as shown on the NREL efficiency chart
(Figure 1-4)8.
5
Figure 1-4. Chart of best research-cell efficiencies according to technology. This plot is courtesy of the National Renewable Energy Laboratory, Golden, CO.
One interesting emerging technology for third-generation solar cells is based on colloidal
quantum dots (CQD) – nanocrystal semiconductors which leverage quantum mechanics and can
be cross-linked to form active thin films. CQD materials offer a number of fundamental
advantages in applications. The materials are solution-processed, potentially enabling low-cost
fabrication such as roll-to-roll manufacture. The materials have already been deposited via spray-
coating, suggesting one versatile means of producing cells over large areas and at low cost. The
bandgap of these semiconductor nanoparticles can be tuned by varying the size of the quantum
dots, allowing deliberate control over spectral response. This facilitates the fabrication of tandem
and multi-junction structures to harvest energy efficiently over the entire solar spectrum. These
materials have been highlighted for their efficiency in harnessing infrared radiation, and have
achieved certified PCE of 9.9%8. CQD materials will comprise the test-platform for the work
completed as part of my PhD thesis. All techniques developed for the photonic enhancement of
thin-film solar cells were deployed and tested using optimized CQD materials.
In the general context of photovoltaics, CQD-based solar cells benefit from numerous cost-based
advantages that are typical of third-generation technologies, with fundamental characteristics that
6
make this material class uniquely intriguing. Among standard evaluation metrics for photovoltaic
technologies, power conversion efficiency is only part of the equation – the amount of electricity
that can be produced per unit cost is the ultimate measure of value. In this context, emerging
thin-film technologies, such as CQD solar cells, make use of cheap, earth-abundant materials that
can be easily processed to fabricate functional films with relatively little material. Colloidal
quantum dot-based technologies, more specifically those based on metal chalcogenides, make
use of some of the most abundant, and cheapest-to-manufacture materials9 (using the example of
lead sulfide CQD). The emergence of perovskite solar cells, which equally benefit from cost-
effective material and fabrication solutions and have achieved greater performance benchmarks
(power conversion efficiency of 20.1%9), as well as the reduced cost of Si-based modules, has
reduced the impact CQD solar cells. In the long term, however, the tunability of the CQD band
gap deep into the infrared spectrum will secure its value in the context of photovoltaic
applications. This unique property of CQD materials can ultimately be leveraged in multi-
junction applications, including integration with existing Si-based modules to extend operation
into the infrared and increase overall power conversion efficiency.
1.3 Thesis Objective
In this thesis, I seek to overcome a key issue that limits many solution-processed photovoltaics:
the absorption-extraction compromise described immediately below. Many materials, including
CQDs, exhibit a charge-extraction length (minority carrier diffusion length) that is shorter than
the thickness of material required to absorb all incident above-bandgap light (absorption length).
This key issue will now be discussed in greater detail.
1.3.1 Absorption-Extraction Compromise
The absorption-extraction compromise arises from the limited charge transport properties of the
solution-processed CQD semiconductor films. Taking for example the case of lead sulfide (PbS)
CQD films in a depleted heterojunction architecture, and assuming a doping density10 of 1016 cm-
3 for TiO2 and assuming doping density is negligible for PbS CQD films, we calculate a
depletion width of approximately 300 nm into the absorbing film. With diffusion lengths < 100
7
nm11, this limits the overall film thickness to ~ 400 nm for efficient separation of photogenerated
charge carriers.
The absorption spectrum of a PbS film falls to ~ 104 cm-1 near its band edge of 800 – 1000 nm.
In order to absorb all incoming photons in this infrared spectrum in a single pass, the film would
need to be of a thickness of order ~ 1 μm. With this discrepancy in charge extraction efficiency
and absorption coefficient, the overall short-circuit current density and PCE of CQD solar cells is
limited to well below the available above-bandgap solar photon fluence (Figure 1-5).
Figure 1-5. Absorption-extraction compromise. a) Absorption profile for a PbS CQD film (absorption coefficient spectrum) including line to illustrate the optimal film thickness for charge extraction. b) Diagram to illustrate the ideal film thickness for transport (Ltran) compared to the ideal film thickness for complete absorption (Labs).
1.3.2 Photonic Enhancement
Much of the CQD community is justifiably focused on improving transport lengths in CQD
solids. I sought to take a different approach: I would develop a means to enhance light
absorption, over a broad range of the solar spectrum, to optimize the impact on device
performance, within a given thickness of light-absorbing active material. To this end I built
8
nanostructured metallic reflectors to leverage plasmonic effects to enhance light interaction in
CQD absorbing films; implemented infrared light guiding strategies using nanostructured
substrates for improved absorption in CQD solar cells; and finally, I developed microstructured
electrodes that provide dramatic improvements in device absorption and PCE. I further
demonstrated that conformal film deposition is a critical component to implementing photonic
enhancement techniques, and develop techniques for conformal film deposition atop structured
electrodes.
1.4 Thesis Outline
Chapter 2. I explore the fundamental principles of photovoltaics and colloidal quantum dot
semiconductor materials. I first discuss the solar spectrum as well as the operating principles of
photovoltaic devices, including a theoretical discussion of their fundamental limits. I then delve
into colloidal quantum dots, including synthesis techniques and optical properties. I then discuss
practical devices based on colloidal quantum dot films, more specifically solar cells. Finally, I
discuss past work focused on addressing the absorption-extraction compromise of CQD films for
improving performance of colloidal quantum dot solar cells; as well as photonic enhancement
strategies applied to other thin-film technologies that have inspired some ideas presented in this
thesis.
Chapter 3. In this chapter, I focus on nanostructured gold electrodes for enhanced absorption in
thin-film CQD solar cells. First, I discuss electrochemical techniques adapted for fabrication of
nanostructured gold electrodes, including an overview of conditions used to control
nanostructure feature sizes and the resulting optical properties of the device. Next, I demonstrate
the optical enhancement benefits of these structures in a CQD film, including modelling to
understand the mechanisms behind absorption enhancement in the CQD film. Finally, I include
photoluminescence-based characterizations to assert that optical enhancements, in fact, originate
in the CQD film, and postulate next steps for high efficiency solar cells.
Chapter 4. In this chapter, I discuss nanostructured substrates for depleted heterojunction CQD
solar cells. First, I discuss the broadband enhancement of nanocavities in a CQD film through
optical modelling. Followed by presenting fabrication techniques for nanostructured substrates
and the subsequent device stack of the CQD solar cell. I demonstrate these effects in “real
9
devices” with supporting absorption, external quantum efficiency and current density-voltage
(JV) curves under simulated solar illumination to demonstrate the performance enhancement of
these substrates. Finally, I discuss practical limitation of this approach and propose steps
forward.
Chapter 5. In this chapter, I discuss the importance of conformal film deposition for
performance enhancement in CQD solar cells. I first discuss modelling of charge transport
properties and set a limit on film thickness for efficient extraction. I then present techniques to
engineer surface hydrophilicity for conformal film deposition. I finally compare internal
quantum efficiency for a microstructured solar cell employing non-conformal and conformal
CQD films, demonstrating that there is dramatic improvement in charge extraction efficiency
with conformal films.
Chapter 6. Having discussed the importance of conformal film fabrication when employing
micro-structured electrodes in Chapter 5, I then proceed to optimize these electrodes with the
goal of producing field-leading efficiency CQD solar cells. First, I present optical modelling
results and describe the impact of pyramid angle as well as the importance of hierarchical
structuring for optimal broadband enhancement of thin film CQD solar cells. I then provide
details for the fabrication of high fidelity periodic micro-pyramids. Finally, I characterize CQD
solar cells using micro-pyramid electrodes and demonstrate significant enhancement in
absorption, with corresponding improvements in current generation and PCE without
compromising film quality or device performance.
Chapter 7. Here, I summarize key results from the thesis and discuss the progression to practical
improvements in CQD solar cell efficiency. I include a discussion of the impact the work of my
thesis has had on the field of photovoltaics and beyond. I then discuss future steps to increase the
impact of photonic enhancement strategies deployed in this thesis and demonstrate their
prospects towards the goal of achieving commercially impactful results for solution-processed
CQD solar cells.
10
Chapter 2 Colloidal Quantum Dot Photovoltaics: Key Challenges
In this chapter, I explore the theory and key challenges of colloidal quantum dot materials for
photovoltaic applications. I begin with a discussion of the solar spectrum to understand the
platform of energy production for solar conversion technologies and how they can be optimized.
I then discuss common operation metrics of photovoltaic devices, including a discussion of the
thermodynamic limits of solar power conversion efficiencies. Then, I delve into the topic of
colloidal quantum dot nanocrystals, how they are synthesized and how they have been used for
photovoltaic applications. Finally, in relation to the primary objective of this thesis, overcoming
the absorption-extraction compromise of CQD thin films, I discuss existing strategies that have
attempted to overcome this limitation and lay the foundation/inspiration for the work of this
thesis.
2.1 Solar Spectrum
At the center of our solar system lies the star that sustains life on Earth. This medium-sized star
has a mass of roughly 1.989 x 1033 kg. Its energy is generated through the nuclear fusion of light
elements, predominantly hydrogen, and it generates power at a rate of 3.9 x 1023 kW throughout
its volume. If we extend the solid angle down to Earth, the energy incident on the surface of our
planet is approximately 1367 W/m2, or approximately 1000 W/m2 after traversing the
atmosphere. The spectrum of the Sun (above the atmosphere) follows the curve for blackbody
emission from a 5777 K source, corrected to match the total irradiance of the Sun (1367 W/m2).
We denote this solar spectrum as measured above Earth’s atmosphere as AM0 (Air Mass 0). The
solar spectrum is further characterized as a function of path length through the upper atmosphere,
which can be determined by the zenith angle of irradiation with respect to the Earth’s surface.
The base case, when the irradiance is normal to the Earth’s surface, is known as AM1. The most
commonly used spectrum to characterize solar cells is AM1.5, which corresponds to a relative
path length of 1.5 times through the atmosphere and for a zenith angle of 48°. (Figure 2-1).
11
Figure 2-1. Spectral irradiance of 5777 K blackbody emitter, AM0 solar spectrum and AM1.5 solar spectrum.
Approximately half of the sun’s integrated irradiance (under AM1.5 conditions) is in the visible
spectrum. The remainder is distributed throughout the near infrared and short-wavelength
infrared, and includes significant pockets of power spectral density up to 2500 nm. The total
irradiance is measured to be 930 W/m2, however, 1000 W/m2 is used as standard irradiance for
characterization of solar cell efficiency. The extended infrared component of the solar spectrum
represents an interesting opportunity for solar conversion technologies that can harvest longer
wavelength radiation. In this case, the size-tunable bandgap of CQD materials opens an
interesting avenue to harvest deep into the infrared of the Sun’s spectrum.
2.2 Operation of Photovoltaics
The photovoltaic effect is defined as the generation of voltage or electric current in a material
upon exposure to electromagnetic radiation. It is generally characterized by the excitation of a
valence-band electron to a higher-energy state, known as the conduction band, so that it can
12
move freely through the material, leaving a hole (defined as the absence of an electron) in its
place. The operation of a photovoltaic cell depends on some built-in asymmetry to extract the
excited electron-hole pair and feed it into an external circuit before it can relax to its ground
state12.
This asymmetry is best achieved in a p-n junction, where a Schottky barrier is formed between
unlike materials. This produces an electric field that drives photoexcited electrons in a given
direction within the circuit. The solar cell is a diode that conducts in one direction in the dark,
and the current density of the cell can be represented by the superposition of the short-circuit
current density under illumination and the dark current which flows opposite. The total current
density behavior of the solar cell under illumination can be expressed using equation 2-1:
Equation (2-1): 𝐽(𝑉) = 𝐽𝑠𝑐 − 𝐽𝑑𝑎𝑟𝑘(𝑉) = 𝐽𝑠𝑐 − 𝐽0(𝑒𝑞𝑉
𝑘𝐵𝑇⁄
− 1)
where q is the charge of an electron, V is the voltage, kB is the Boltzmann constant (1.38 x 1023
m2kg/s2K) and T is temperature (in K). Under open-circuit conditions, J is equal to 0, and we can
solve for the open-circuit voltage (Voc) as shown in equation 2-2:
Equation (2-2): 𝑉𝑜𝑐 =𝑘𝐵𝑇
𝑞ln(
𝐽𝑠𝑐
𝐽0+ 1)
These two parameters represent the maximal values of operating current and voltage of a solar
cell under ideal conditions. The overall power conversion efficiency of the cell is equal to the
maximum electrical power generated divided by the optical power incident:
Equation (2-3): 𝜂 =𝑃𝑜𝑢𝑡
𝑃𝑖𝑛=
𝑉𝑚𝐽𝑚
𝑃𝑠
where Vm and Jm are the voltage and current density produced at the maximum power point; Ps is
the incident light intensity. The power conversion efficiency can also be expressed as the product
of the short-circuit current density, the open-circuit voltage, and a unitless fill factor (FF):
Equation (2-4): 𝐹𝐹 =𝑉𝑚𝐽𝑚
𝑉𝑜𝑐𝐽𝑠𝑐
Equation (2-5): 𝜂 =𝑉𝑜𝑐𝐽𝑠𝑐𝐹𝐹
𝑃𝑠
13
The general shape of a current density-voltage plot for a photovoltaic cell is shown in Figure 2-2.
This diagram illustrates the general relationship between open-circuit voltage (Voc), short-circuit
current (Jsc) (or current density), and the voltage (Vm) and current density (Jm) at the maximal
power point.
Figure 2-2. Current density-voltage (JV) plot for solar cell. All relevant parameters related to the evaluation of solar cell efficiency are illustrated for reference.
The external quantum efficiency of a solar cell represents its spectral efficiency in converting
incident photons into extracted electrons. This is generally related to the internal quantum
efficiency (probability that an absorbed photon is converted to an excited and extracted electron,
IQE(λ)) of the solar cell as well as the absorption (probability of absorbing an impinging photon,
A(λ)) spectrum:
Equation (2-6): 𝐸𝑄𝐸(𝜆) =𝐼𝑠𝑐(𝜆)
𝐼𝑖𝑛(𝜆)= 𝐼𝑄𝐸(𝜆) ∗ 𝐴(𝜆)
where Isc(λ) is the measured current under short-circuit conditions while illuminating the sample
with a given wavelength radiation λ, and Iin(λ) is the input current of the radiation, which can be
determined by deconvolving the number of photons from the light power and assuming an
electron is generated for each photon. We can further relate this to the short-circuit current
density by integrating the EQE over all wavelengths (equation 2-7) – this will be an important
14
performance metric in evaluating the effectiveness of photonic enhancement strategies
implemented in this thesis:
Equation (2-7): 𝐽𝑠𝑐 = 𝑞 ∫ 𝜙(𝐸) ∗ 𝐸𝑄𝐸(𝐸)𝑑𝐸∞
𝐸𝑔= 𝑞 ∫ 𝜙(𝜆) ∗ 𝐸𝑄𝐸(𝜆)𝑑𝜆
𝜆𝑔
0
where q is charge of an electron, and φ(E,λ) is the incoming photon flux (solar spectrum in this
case).
Now that I have established the metrics of solar cell operation, I will briefly discuss the
theoretical limits of solar power conversion efficiency. Shockley and Quiesser investigated the
thermodynamic limits of solar to electric power conversion of photovoltaic solid state devices.
Considering the limitations of thermodynamics and including the rates of radiative processes in
semiconductors, Shockley and Queisser calculated an intrinsic limit on the efficiency of solar
cells. For the case of the single-junction Si cell, with its bandgap of 1.1 eV, and approximating
the solar spectrum as that of a 6000 K blackbody source, they determined the upper limit to be
30% PCE for unconcentrated sunlight13. Several decades later, Henry14 extended this calculation
to the AM1.5 terrestrial solar spectrum and to account for different intensities (such as
concentrated sunlight) and additional junctions. The ideal bandgap for a single-junction solar cell
is 1.35 eV and the theoretical limit is 31% at one sun intensity. Concentration to 1000 suns
intensity leads to an upper bound of 37% PCE. In the limit of excellent spectral utilization
achieved through the use of many junctions (36 junctions) illuminated with 1000 suns, the PCE
approaches 72% in principle (Figure 2-3)14.
15
Figure 2-3. Solar cell efficiency as a function of bandgap for solar concentrations of 1 and 1000 suns. Reprinted with permission from Ref. 14. Copyright 1980, AIP Publishing LLC.
2.3 Colloidal Quantum Dots
Colloidal quantum dots benefit from the quantum size effect that allows tunability of the optical
bandgap via nanoparticle diameter. By decreasing the size of individual elements below the
exciton Bohr radius of the material the electrical bandgap of the particle can be effectively
increased. The relationship between the effective bandgap (E’g) and the size of quantum dots can
be expressed as:
16
Equation (2-8): 𝐸′𝑔 = 𝐸𝑔 +ℏ2𝜋2
2𝑟2(1
𝑚𝑒+
1
𝑚ℎ) −
1.8𝑒2
𝜀𝑟
where Eg is the bulk semiconductor bandgap, me is the effective electron mass, mh is the effective
hole mass, and r is the radius of the quantum dot.
Quantum dots are synthesized by mixing precursors in a heated flask with organic compounds
which serve to cap (cover the surface) of nanoparticles to prevent fusing of individual units. The
temperature should be high enough to decompose precursors into monomers, but low enough to
allow for nucleation of nanocrystals. More specifically, reagents are added to a coordinating
solvent at elevated temperatures (up to 300 °C) ultimately forming monomers of the nanocrystal
compound. Once the concentration of monomers breaches the nucleation threshold,
crystallization will occur. As crystallization progresses the concentration of initial monomers
will fall below the nucleation threshold, thus terminating the nucleation phase of the reaction.
The remaining monomers will then react with the nuclei resulting in the growth of nanoparticles
as illustrated in Figure 2-415.
Figure 2-4. Synthesis mechanisms of colloidal quantum dots. a) Diagram illustrating the stages of nucleation and growth for the preparation of quantum dot nanocrystals according to the La Mer model16. b) Image of the synthetic apparatus used to prepare colloidal quantum dot nanocrystals. Reprinted from Ref. 15.
17
For the purpose of this thesis, only PbS quantum dots were employed. The synthesis of PbS
nanocrystals is achieved by heating lead oxide (PbO) in oleic acid (OA) to 150 °C.
Tetramethylsilane (TMS) in octadecene is then injected into PbO containing solution while being
vigorously stirred. The size of the nanoparticles can be tuned by adjusting the temperature steps
of the process, or concentrations of the precursors17. The bandgap of these materials is then
dictated by their size as shown in Figure 2-518, with consequent impact on the absorption onset
(exciton peak) of the material. This is the key characteristic that allows for the material to be
tuned to harvest the entire solar spectrum efficiently.
Figure 2-5. Relationship between quantum dot size and bandgap. a) Using the example of PbS quantum dots, here they illustrate the impact of nanocrystal size on bandgap energy levels, note the reduction in bandgap as a function of increasing nanocrystal diameter. b) This illustrates the shift in absorption onset (exciton peak) as a function of quantum dot size (colours corresponding to the sizes in part (a)). Adapted from Ref. 18 with permission of The Royal Society of Chemistry.
2.4 Colloidal Quantum Dot Solar Cells
The first reported optimal-bandgap CQD solar cells with efficiency > 1% were based on the
Schottky architecture19–21. For this configuration the charge-separating junction is formed at the
interface between the solution-processed CQD film and a metal with a specific work function.
This interface forms a Schottky barrier that creates a depletion region within the light-absorbing
18
CQD film that drives charge separation within the device. To date, record efficiencies of 4.6%
PCE have been reported for lead selenide (PbSe) CQD films in a Schottky junction utilizing
shallow-work function aluminum contacts22. The design and operation of the Schottky solar cell
are illustrated in Figure 2-621.
Figure 2-6. Structure of the Schottky CQD solar cell. a) Physical diagram of the Schottky CQD solar cell, with PbS CQD film, ITO for bottom transparent conducting oxide contact, and aluminum (Al) as the top contact. b) Band energy diagram to illustrate the Schottky junction formed at the PbS-Al interface. Adapted with permission from Ref. 21. Copyright 2008, AIP Publishing LLC.
Despite reported progress, the Schottky solar cell suffers from limitations in its design. With
light incident at the back (ohmic side) of the cell, a significant fraction of photogenerated charge
carriers must travel the thickness of the CQD film to reach the charge-separating junction. Since
charge transport is limited in CQD films, a large fraction of the solar spectrum cannot be
efficiently absorbed and extracted in this configuration.
To address these limitations, the depleted-heterojunction architecture was designed. Inspired by
CQD-sensitized solar cells (based on similar concepts as dye-sensitized solar cells23), the
depleted heterojunction device introduced a large-bandgap, shallow-work function front window
layer to form the junction with the CQD film. As a result, light is incident closest to the charge-
separating interface. The first report of this configuration made use of titanium dioxide (TiO2) as
the n-type window layer that forms a junction with a CQD film, yielding power conversion
efficiencies of 5.1%24. As an added advantage, the large-bandgap TiO2 layer forms a hole-
19
blocking barrier at the junction, ultimately leading to improved charge extraction in the device.
The design of the depleted heterojunction solar cell is shown in Figure 2-724.
Figure 2-7. Structure of the depleted heterojunction CQD solar cell. a) Physical diagram depicting the distinct layers of the depleted heterojunction architecture, including the SnO2:F (fluorine-doped tin oxide – transparent conducting oxide layer), the TiO2 window layer (n-type junction), PbS CQD layer (p-type junction) and Au top contact. b) Energy band diagram to illustrate the conceptual advantages of the depleted heterojunction architecture; the TiO2 layer serves both as charge separating junction and hole blocking layer. Adapted with permission from Ref. 24. Copyright 2010 American Chemical Society.
At the beginning of my thesis the record PCE for depleted heterojunction CQD solar cells was
5.1%. Many materials advances have been reported since then, including atomic ligand
passivation25, doping of quantum dots to create p- and n-type layers26–30, and even strategies to
improve air stability31. A certified PCE record for CQD solar cells was achieved applying a
hybrid passivation strategy, wherein halide passivation is implemented at the solution level with
CdCl2 introduced to the PbS CQD surface, and an organic cross-linker (3-mercaptopropionic
acid or 3-MPA) is used in the solid-state exchange. Halides serve to passivate trenches on CQD
surfaces, reducing surface traps, while the 3-MPA provides maximal packing among quantum
dots in the film. The resulting device, based on a depleted-heterojunction architecture, yielded
record certified efficiencies of 7.0% PCE32. This materials processing strategy was applied for all
of the work presented in this thesis.
A major limitation in the design of CQD solar cells for maximum PCE is the absorption-
extraction compromise, which I defined in the previous chapter. With diffusion lengths on the
20
order of 80 ± 10 nm for hybrid-passivated PbS CQD films11,33, coupled with an estimated 200-
300 nm depletion region when interfaced with TiO2 electrodes (free carrier density 1015-1016 cm-
3), CQD films must be limited to ~ 300-400 nm in thickness for efficient extraction of
photogenerated charge carriers. With absorption lengths on the order of ~ 1 µm for 900 nm
wavelength radiation, a large portion of the infrared remains unutilized. Figure 2-8 further
illustrates limits in charge transport for CQD films wherein the drift-dominated region (known as
the depletion region) actually decreases under maximum power operation in favour for the
diffusion-dominated region (known as the quasi-neutral region). We note that electric field and
carrier concentrations generally drop to 0 at 0.4 μm (400 nm) from the junction.
Figure 2-8. Model and charge transport in CQD solar cells. a) Energy band diagram of a p-type PbS CQD/n-type TiO2 heterojunction at short-circuit condition. b) Energy band diagram of a p-type PbS CQD/n-type TiO2 heterojunction at maximum power-point condition. Electrons in blue, holes in red. c) Simulated electron-hole product as a function of distance from the CQD-TiO2 junction. d) Simulated electric field intensity as a function of distance from the junction. Adapted by permission from Macmillan Publishers Ltd: Nature Communications Ref. 33, copyright 2014.
21
2.5 Strategies for Overcoming the Absorption-Extraction Compromise
2.5.1 Strategies for Colloidal Quantum Dot Solar Cells
Many studies have investigated solutions to the absorption-extraction compromise in solution-
processed thin-film photovoltaics. Solutions include reducing the distance over which carrier
transport is required by structuring the electrode; and seeking to increase overall absorption for
an effective thickness of film. The vast majority of strategies have been of the first variety
wherein a bulk heterojunction was created, either with an interdigitated electrode or by mixing
the materials of the p and n junctions together. This concept of bulk heterojunction is borrowed
from organic solar cells wherein acceptor and donor materials are intermixed to create a large-
area charge separating interface throughout the absorbing volume, increasing the likelihood that
short-lived photogenerated carriers can be extracted34–36. Examples include depleted bulk
heterojunctions using nanoporous electrodes37 (150-200 nm particle size), traditional bulk
heterojunctions in which bismuth sulfide - Bi2S3 (n-type) and PbS (p-type) nanoparticles were
intermixed38, and metal oxide nanowire/nanopillar electrodes combined with planar films39–41.
The basic premise of all these strategies is to reduce the effective distance photogenerated
carriers have to travel to reach the charge separating junction. We see in Figure 2-941 how the
interdigitated nanopillars extend the depletion width into regions of the CQD absorbing film that
are relatively far from the base of the electrode, and simultaneously reduce the distance carriers
must travel to reach the charge separating junction.
22
Figure 2-9. Concept of the depleted bulk heterojunction CQD solar cell. a) Diagram depicting the structure of the depleted bulk heterojunction CQD solar cell with TiO2 nanopillars. b) STEM image of the actual solar cell, fabricated with TiO2 nanopillars, PbS CQD layer, MoO3, Au, and Ag as the top contact. Reprinted with permission from Ref. 41. Copyright 2012, Wiley-VCH.
However, bulk heterojunction strategies, which produce a many-fold increase in junction area,
generally lessen open-circuit voltage since they also increase the recombination rate at the
increased-area interface37,42,43. For this reason they not yet led to net power conversion efficiency
advances for colloidal quantum dot solar cells.
Light recycling (i.e., wherein the interaction length of light is increased) approaches generally
leverage some form of modified geometry to increase the light path through the active film, though
these depend heavily on a conformal back mirror. In a unique report devices were engineered/tilted
to trap light longitudinally through the film, without any intrinsic level of structuring required44
(Figure 2-10).
23
Figure 2-10. Concept of the folded-light path CQD solar cell. a) Diagram depicting the light path in a simple planar device (double pass reflection). b) Demonstration of the scalability of the folded-light path concept, wherein numerous cells can be stacked adjacently to collect over a larger area. c) Light path in the folded-light path CQD solar cell, showing > 3 internal reflections inside the substrate, accounting for > 6 passes through the active film. Adapted from Ref. 44.
In other approaches, plasmonic nanostructures were implemented into CQD films in an effort to
increase near-field absorption45,46. Plasmonic nanoparticles (NP) are characterized by their size,
which is smaller than the electromagnetic radiation with which they are interacting, triggering to
the formation of hot energy spots at the nanoscale. Such energy localization can be then used to
increase the absorption of incoming photons in the CQD film. In one such case, highly symmetric
gold nanoshells were explicitly modelled and integrated into CQD films for an increase in
absorption and consequent improvements in device performance46 (Figure 2-11). However, as in
other strategies enhancements were not impactful enough to yield practical advances in the field
of CQD solar cells.
24
Figure 2-11. Concept of the plasmonic-excitonic CQD solar cell. In this case, gold nanoshells are embedded in the CQD film and are engineered to yield plasmonic scattering at a weakly absorbed wavelength of radiation for CQD materials (λ = 820 nm). This leads to improvements in absorption and consequent improvements in short-circuit current density. Reprinted with permission from Ref. 46. Copyright 2013 American Chemical Society.
2.5.2 Strategies Employed in Other Material Systems
Beyond the domain of solution-processed CQD solar cells, many light-trapping strategies have
been successfully deployed. Battaglia et al. employed periodic and random nanostructured
electrodes (ZnO) for amorphous silicon solar cells47. Power conversion efficiency was
dramatically improved for devices prepared with both pyramids (random) and nanocavities
(periodic), with enhancements of 38% for both cases as compared to planar controls. Cross-
sectional scanning electron microscopy (SEM) images and corresponding absorption (1-R) and
external quantum efficiency spectra are presented in Figure 2-1247.
25
Figure 2-12. Structured substrate concept as applied to amorphous silicon solar cells. a) Cross-sectional SEM image of the periodic nanocavity array. b) Cross-sectional SEM image of the random pyramidal texture. c) Cross-sectional SEM image of the planar control case. d) Absorption and EQE data comparing all three cases, with a demonstration of improved EQE when employing either periodic or random texturing of the electrode as compared to the planar case. Reprinted with permission from Ref. 47. Copyright 2012 American Chemical Society.
26
Deeper work on nanostructuring for enhanced solar cell performance has been detailed in
reviews by Cui48,49, which provide in-depth analyses of photon management techniques for
enhanced light utilization for a range of silicon-based photovoltaic technologies and provide
inspiration for this thesis.
2.6 Conclusion
In this chapter, I discussed the solar spectrum for reference to frame the impact of solar
conversion technologies, followed by a presentation of the underlying operation principles of
photovoltaics and theoretical limitations in conversion efficiency. Finally, the topic of colloidal
quantum dots was presented including the theory and synthesis of these materials, how they can
be implemented for photovoltaic applications, and a literature review of previously reported
strategies for addressing absorption-extraction compromise in these devices, and for other
material systems. Many of these concepts provide the foundation for development of light
recycling (via plasmonics or scattering enhancement) techniques applied to thin-film
technologies, including those presented in this thesis for CQD solar cells.
The ultimate objective of this thesis is to investigate effective strategies for light recycling in
CQD solar cells with an emphasis on overall improvement in power conversion efficiency;
something that has not be thoroughly explored for this class of materials used in this application.
In the following chapters, I will present my work related to addressing the absorption-extraction
compromise via light-recycling techniques (with the implementation of structured substrates and
electrodes). To clarify further, the term light recycling is used to denote any strategy that
effectively increases the interaction length of light beyond the standard geometry of the system.
27
Chapter 3 Nanostructured Disordered Reflectors for Thin-Film
Photovoltaics
3.1 Introduction
I discussed in Chapter 1 the absorption-extraction compromise that ultimately limits power
conversion efficiency in CQD solar cells. Here, I adapt electrochemical techniques to CQD
photovoltaics as a means to increase absorption in the active film. I show that the level of
disorder of the structured reflector is an important consideration in leveraging the full photonic
benefits of this material in a thin film optoelectronic context.
I begin by exploring techniques to prepare disordered nanostructured reflectors using
electrodeposition of gold and demonstrate control over surface features as a function of
deposition conditions. Next, I demonstrate the significant absorption enhancement potential of
these materials, which agree well with FDTD simulated results. Additional characterization
techniques based on the photoluminescence response of the CQD film are employed to isolate
the optical effects in the active film, ultimately demonstrating the practical benefits of employing
the reflectors. Finally, I discuss how these reflectors can be integrated into working CQD solar
cells and discuss next steps to yielding high performance photovoltaic devices using these
materials.
3.2 Nanostructured Gold (Au) Reflectors
To prepare the nanostructured reflectors, I applied a specific manufacturing process known as
electroplating. This process has been successfully applied in biological sensing applications for
the production of multiscale disorder with high-surface-ratio features50–53. I apply this technique
for the first time in an energy harvesting context. Nanostructured gold (Au) films were prepared
using a 3-probe potentiostat in a HAuCl4 (gold-chloride) solution, with the reduction of Au
atoms on the working electrode, which is coated with a thin, planar Au film. Figures 3-1 shows
tilted scanning electron microscopy (SEM) images of samples prepared with exposure times of t
= 3 min (Fig. 3-1a), t = 6 min (Fig. 3-1b) and t = 10 min (Fig. 3-1c) in a 13 mM HAuCl4
solution. SEM images reveal a trend of increased disorder as a function of deposition time.
28
Disorder is quantified by measuring the spatial correlation lengths Lc and average feature heights
havg, which are estimated directly from SEM images. In Figure 3-2 I present a cross-sectional
SEM image of a nanostructured Au sample prepared with 13 mM solution and with an exposure
time of t = 6 min. Lc and havg are estimated based on the average distance between features in the
x-axis (Lc) along the centre line and y-axis (havg) above and below the centre line. For the
samples shown in Fig. 3-1, I measure disordered nanostructured surfaces with Lc = 20 nm, Lc =
40 nm, and Lc = 50 nm, as the deposition time t changes from 3, 6, and 10 min, respectively. The
corresponding average height change is havg = 10 nm, havg = 25 nm, havg = 35 nm, respectively.
Quite interestingly, disordered nanostructures generated through electroplating span multiple
spatial scales, with features varying in the range between 10 and 500 nm for exposure times up to
10 min.
Figure 3-1. Scanning electron microscopy images of the nanostructured Au reflectors using 13 mM HAuCl4 solution. a) For an exposure time of 180 s. b) For an exposure time of 360 s. c) For an exposure time of 600 s.
29
Figure 3-2. Cross-sectional SEM of nanostructured Au reflector with active films. The centre line is used to estimate values of Lc and havg for the Au reflector.
Multi-scale disorder can be further controlled by varying the solution concentration of HAuCl4.
As the concentration is increased, there is progressive growth of sharper, micron-scale Au
structures as a function of concentration and exposure time. Figure 3-3 shows SEM images of
samples prepared with exposure times of t = 10 min in 26 mM (Fig. 3-3a) and in 40 mM HAuCl4
solutions (Fig. 3-3b). When samples are prepared with the higher concentration solution, there is
a nanostructured base similar to that observed for the samples in Figure 3-1 (with 13 mM
solution) with the addition of numerous micron-scale features. With a concentration of 26 mM
(Fig 3-3a) there is formation of numerous micron-scale features (~1-2 μm in length), with a
density of approximately 50 per 10 μm x 10 μm area. For the highest concentration of 40 mM
(Fig 3-3b), these micron-scale features further expand into fishbone-like clusters with sizes of
10-20 µm. In both cases, the multi-size micron-scale disordered features coexist with random
nanostructures varying over spatial scales of 100 nm and less. Increased exposure time leads to
30
increased density and relative size of structured features on the characteristic scale for a given
solution concentration.
Figure 3-3. Scanning electron microscopy images of the nanostructured Au reflectors prepared with exposure times of 10 min. a) For solution concentration of 26 mM HAuCl4. b) For solution concentration of 26 mM HAuCl4.
3.3 Colloidal Quantum Dot Absorption on Nanostructured Au Reflectors
3.3.1 Experimental Results
In the first series of experiments, I characterized the linear absorption properties of the
nanostructured Au substrates when coated with CQD films as a function of solution
concentration and exposure time. Measurements were performed using a spectrophotometer
equipped with an integrating sphere, which measures the total reflected and transmitted signals
from the solid film, allowing for the calculation of total absorption. Figures 3-4 shows the total
absorption of these structures when combined with 50-nm-thick colloidal quantum dot films – in
this case using 1.3 eV PbS quantum dot films. This helps illustrate the relative impact of
different feature length scales, and relative densities on total film absorption. In Fig. 3-4a there is
a trend of increased total absorption as a function of deposition time. As shown in SEM, at 13
mM the features are predominantly on the nano-scale (up to 100’s nm), and as deposition time is
increased, the density and relative size of these features is increased. Fig. 3-4b shows the total
absorption for CQD films on nanostructured Au electrodes prepared with 26 mM concentration
solution. At this concentration, the density of nano-scale structures saturates, and instead micro-
31
scale structures begin to form. Despite this growing density of micro-scale features, increased
deposition time has significantly less impact on overall absorption. This trend suggests that
absorption is most strongly affected by nano-scale features. This trend (increased absorption vs
deposition time) almost completely disappears for the case of 40 mM solution (Fig. 3-4c),
suggesting that the increased size and density of micron scale structures (up to 10-20 µm in this
case) has no impact on absorption in the CQD film. The reason for this trend is that CQD films,
on the order of 50 nm, are much less likely to conformally coat the larger micron-scale structures
on the surface, thus these features contribute nothing to the total absorption of the film. Moving
forward, I focused on deposition conditions that produced predominantly nanostructured features
as these lead to the most interesting behavior in the CQD film.
Figure 3-4. Total absorption of CQD films atop nanostructured Au reflectors. a) Prepared with 13 mM HAuCl4 solution. b) Prepared with 26 mM HAuCl4 solution. c) Prepared with 40 mM HAuCl4 solution.
32
Having identified the most interesting reflectors to work with I prepared nanostructured
substrates using a 13 mM solution of HAuCl4 with exposure times of t = 3 min (Lc = 20 nm), 6
min (Lc = 40 nm) and 10 min (Lc = 50 nm). I then prepared PbS CQD films with bandgap of 1.3
eV and thickness between 100 and 300 nm. Figures 3-5 and 3-6 summarize the experimental
results, which show a dramatic absorption improvement when using disordered nanostructured
reflectors with respect to the planar reflector. With a 100-nm-thick CQD film, absorption was
improved by approximately 70% in the most disordered case at λ = 850 nm, a wavelength that is
poorly absorbed in 1.3 eV PbS films (Fig. 3-5a). The increase in absorption is quite impressive
especially in the infrared region, showing a 40% increase at all wavelengths for the most
disordered configuration. The absorption enhancement peak changes with the disorder,
exhibiting a red-shift for increasing correlation length Lc (Fig. 3-5b). More specifically, for Lc =
20 nm the most dramatic improvement is observed at λ = 600 nm, with an absorption
enhancement of 20%. This peak enhancement is observed at λ = 655 nm and λ = 700 nm for Lc =
40 nm and Lc = 50 nm cases, with 50% and 65% more absorption, respectively. Most notably,
we observe an improvement of 5%, 26%, and 47% at the most weakly absorbed wavelength of λ
= 850 nm for the Lc = 20 nm, Lc = 40 nm, and Lc = 50 nm cases, respectively, with a 9-fold
relative enhancement for the roughest case (Lc = 50 nm) with respect to the structure with
weakest disorder (Lc = 20 nm).
Figure 3-5. Absorption profiles for 100 nm CQD films atop nanostructured Au reflectors. a) Total absorption of the CQD film/nanostructured Au reflector for planar Au, Lc = 20 nm, Lc = 40 nm and Lc = 50 nm cases. b) Relative absorption enhancement as compared to planar case for Lc = 20 nm, 40 nm, and 50 nm nanostructured Au cases.
33
Figures 3-6 depicts the absorption spectra for 300 nm CQD films. In this configuration, the
planar structure shows significant Fabry-Perot resonances, whose effect is amplified by disorder
as seen from Fig. 3-6a. The absorption at λ = 850 nm, in particular, is 17%, 42%, 61% and 75%
for the planar, Lc = 20 nm, Lc = 40 nm, and Lc = 50 nm nanostructured Au cases, respectively.
For the thicker CQD film the absorption is observed to improve most dramatically between 780
and 830 nm for all three nanostructured cases. Comparing absorption improvement at λ = 850
nm, I measure 24%, 43% and 57% for Lc = 20 nm, Lc = 40 nm, and Lc = 50 nm, respectively
(Fig. 3-6b). The enhancement for the 50 nm case is approximately 2-fold greater than for the 20
nm case with the thicker CQD film. This result illustrates the impact of disordered
nanostructured reflectors on the absorption of CQD films. With the thinner film, there is a much
more dramatic impact on broadband absorption improvement for the roughest electrode,
indicative of increased impact due to the near-field plasmonic effects of disordered
nanostructured Au features.
Figure 3-6. Absorption profiles for 300 nm CQD films atop nanostructured Au reflectors. a) Total absorption of the CQD film/nanostructured Au reflector for planar Au, Lc = 20 nm, Lc = 40 nm, and Lc = 50 nm cases. b) Relative absorption enhancement as compared to planar case for Lc = 20 nm, 40 nm, and 50 nm nanostructured Au cases.
34
3.3.2 Finite-Difference Time-Domain (FDTD) Modelling
In all cases, the most disordered surface yields the greatest increase in absorption. In order to
provide further physical insights on this process, Finite-Difference Time-Domain (FDTD)
simulations were carried out by collaborators at the King Abdullah University of Science and
Technology (A. Fratalocchi, et. al.). A simulation space of 50 µm x 3 µm was defined, with the
complete device (y-axis dimensions: 152 nm Au, 50 nm TiO2, and 100-300 nm PbS) comprising
45 µm x 300-500 nm of that space. The spatial dimension of a single grid unit was 2 nm. A plane
wave source was used with TM (transverse magnetic) polarization to investigate the optical
behavior of the structure. FDTD is used to simulate optical phenomena by solving Maxwell’s
equations within individual units of a physical grid. For this analysis, a series of random metallic
surfaces that match our fabricated samples in terms of correlation length Lc and average disorder
height havg were created. Results are summarized in Figure 3-7. In comparing Figs. 3-7 with Figs.
3-5a and 3-6a, I observe excellent agreement with our theoretical model, which reproduces the
experimental absorption increase in all the visible and infrared regions. This is an important
result, which confirms the reproducibility of these results, as they depend on the average features
of the disorder (i.e., the correlation length Lc and average height havg), which can be precisely
controlled during the electroplating process. Figure 3-8 provides details of the energy
localization on the random metallic surface, allowing a direct analysis of the behavior of light in
the nanostructured reflector. In this simulation, a sample with Lc = 50 nm and havg = 35 nm was
considered, subjected to monochromatic light at a wavelength of λ =759 nm. In agreement with
theoretical expectations, the electromagnetic energy distribution (Fig. 3-8) illustrates that light
energy gets accumulated very efficiently in the nanocavities of the structured reflector. Figure 3-
8 shows that the presence of subwavelength nanocavities is also interesting as this provides
regions of plasmonic enhancement. This effect combined with scattering in the active film leads
to increased absorption. The relatively greater impact on 100 nm films vs 300 nm films,
however, suggests that plasmonic effects (as illustrated with the E-field plot in Fig. 3-8) is the
dominant mechanism for increased optical enhancement of the active film.
35
Figure 3-7. FDTD simulated absorption profiles for CQD films atop nanostructured Au reflectors. a) Total absorption of a 100 nm CQD film/nanostructured Au reflector for planar Au, Lc = 20 nm, 30 nm, 40 nm and 50 nm cases. b) Total absorption of a 300 nm CQD film/nanostructured Au reflector for planar Au, Lc = 20 nm, 30 nm, 40 nm, and 50 nm cases.
Figure 3-8. Simulated energy field plot to illustrate where light energy is concentrated. Note that radiative energy is predominantly concentrated in the cavities between nanostructured features.
3.3.3 Photoluminescence Excitation
In order to isolate the effects of absorption enhancement in the quantum dot layer from
potentially parasitic losses from the Au electrode, I performed photoluminescence excitation
(PLE) measurements. For this characterization, the sample is excited with a range of
36
wavelengths (from 400 to 800 nm) while measuring the relative intensity of the
photoluminescence (PL) peak for the CQD film. For 950 nm (1.3 eV) PbS films, the greatest PL
intensity is measured at 1050-1100 nm, and so this wavelength is monitored for PLE
characterization.
In Figure 3-9a, I compare PLE signals for a 50 nm CQD film atop nanostructured Au electrodes
of varying levels of disorder (corresponding to the cases explored in Figure 3-4a using
spectrophotometer measurements). Here, the PLE signal is normalized to the highest count value
over the range of 400 to 800 nm – in essence, this technique serves as a relative measure of CQD
absorption efficiency in the visible compared to infrared. Ideally, when absorption in the CQD
layer has been successfully enhanced there should be a shift of increased relative PLE in the
infrared. For the case under consideration, the relative PLE signal in the infrared is increased as a
function of nanostructured Au disorder. This trend is directly reflective of the absorption trend in
Figure 3-4a, in which the infrared absorption is increased as a function of disorder. For the planar
case, the PLE signal is maximized at 570 nm, with approximately 36% relative response at 800
nm. In the nanostructured case with the least disorder (13 mM – 180s), PLE is maximized at 570
nm as well, however, there is substantially more response at 800 nm, with approximately 55%
relative intensity. For the more heavily disordered cases (13 mM - 360s and 600s), the trend is in
fact reversed in which PLE response is maximized in the infrared (780 nm for 360s, and 760 nm
for 600s), instead with the relative response in the visible dropping to 50-55%. This is to support
the absorption data presented in this work and to assert that increased infrared absorption is in
fact due to increased absorption in the CQD layer.
There is, however, a trend of reduced PLE in the visible as compared to the infrared, as
absorption is enhanced. Ideally there should be comparable PLE for visible and infrared
wavelengths. The loss in the visible is essentially due to absorption in the Au layer. Au has non-
negligible absorption < 600 nm, with a broadening of this effect > 600 nm for disordered
nanostructured electrodes. Since the QD film is only 50 nm, much of the visible and infrared
radiation is not absorbed on the first pass through the film, as a result the relative loss in the
electrode is greatest where Au is most absorptive (i.e., in the visible < 600 nm).
37
In order to understand this effect, and to contextualize the concepts in terms of practically
reasonable CQD film thicknesses, I compared the PLE signal as a function of film thickness. In
Figure 3-9b, I present relative absorption as a function of film thickness for a nanostructured Au
substrate of consistent disorder level (Lc = 50 nm). There is a generally consistent shape in
absorption for all samples, with an increase in overall absorption for the thicker CQD film (18
layers – 300 nm). In Figure 3-9c the PLE of these samples is compared. In all cases, the relative
contribution of the infrared signal is greater than for visible radiation. As anticipated, as the
thickness of the CQD film is increased there is a tendency of relatively increased PLE in the
visible spectrum, indicative that losses in the Au are reduced since more of this radiation is
absorbed on the first pass through the CQD film. For the 300 nm case, in which almost all visible
radiation should be absorbed on the first pass through the CQD film, there is a flat PLE response
across the visible and infrared spectrum. This is indicative that the CQD film’s absorption
efficiency in the infrared has been effectively enhanced to a level comparable to its absorption in
the visible. In essence, the nanostructrured electrode has greatly enhanced the absorption of the
active film in the more weakly absorbed infrared regions > 700 nm.
38
Figure 3-9. Photoluminescence excitation (PLE). a) Comparing PLE response as a function of nanostructured Au disorder for 13 mM substrates and 50 nm CQD fimls. b) Absorption as a function of CQD film thickness for Lc = 50 nm disordered reflector. c) PLE response corresponding to samples from b.
3.4 Solar Cells Employing Nanostructured Au Reflectors
An important consideration is whether the significant absorption enhancement presented in Figs.
3-5 and 3-6 can be translated into practical current improvements for colloidal quantum dot solar
cells that harvest into the infrared spectrum. To this end, a prototype device was prepared with
the goal of translating the absorption enhancement of disordered Au electrodes into photocurrent
39
gains. The test device was prepared in a top-illuminated configuration, with the Au layer (either
100 nm planar Au or nanostructured Au) as the bottom electrode, followed by a titanium dioxide
(TiO2) layer of 50 nm deposited atop the electrode and treated with TiCl4. This treatment serves
to passivate the surface of the n-type TiO2 layer while improving the electron transport properties
for efficient charge extraction. The test cell is coated with a 300-nm-thick film of 1.3 eV PbS
CQD, which was fabricated atop TiO2 using spin-casting techniques, and finally a 200-nm-thick
tin-doped indium oxide (ITO) film serving as the transparent top-contact.
The photovoltaic cells were tested under AM1.5 simulated solar illumination and the results are
summarized in Table 3-1. The most disordered sample with Lc = 50 nm exhibits the highest PCE,
showing a gain factor of 15% above the less nanostructured Au sample, and 25% improvement
over the planar film, with almost all enhancement due to increased short-circuit current density.
The short-circuit current density was explicitly improved by 19% and 34% over the less-
roughened Au sample and planar film, respectively. This improvement in short-circuit current
density exceeding 30% is a quite remarkable achievement in addressing the absorption-
extraction compromise of CQD thin-film photovoltaics.
Table 3-1. Summary of CQD solar cell performance as a function of nanostructured Au roughness (as expressed using correlation length, Lc)
These results demonstrate the potential impact of disordered nanostructured reflectors in a
photovoltaic cell, unfortunately, PCE is ultimately restricted by limitations of the top-illuminated
design. Further discussion is included in the following section to detail how photovoltaic layers
must be improved to integrate nanostructured reflectors for optimal performance in a CQD solar
cell.
40
3.5 Conclusion
In this work, I engineered nanostructured reflectors to harvest a significant portion of energy
from the infrared spectrum, enabling development of CQD devices with significant current
increase. I proposed a new strategy that engineers unconventional broadband absorption from
random metallic materials. In a test-prototype solar cell, the nanostructured material shows a
dramatic enhancement of 34% in generated photocurrent over the planar control.
These nanostructured reflectors open new pathways in energy conversion, and by maximizing
the value of this approach in appropriate architectures, I anticipate this strategy could yield
significant improvements in record-efficiency CQD photovoltaic cells. Here are some points that
could be improved for the successful development of high-efficiency CQD solar cells on
nanostructured reflectors. The development of an inverted architecture, with the n-type window
layer (e.g., TiO2 in this case) fabricated atop the CQD film can be important to optimize the
internal quantum efficiency (IQE) of the device. In order to achieve this result, the CQD layer
needs to be fabricated directly onto the Au substrate, a strategy that requires the development of
surface treatments to ensure the adhesion of the absorbing layer on the Au electrode. While the
ITO top contact exhibits excellent optoelectronic properties directly on glass, I observed a
degradation of quality when depositing on the rougher CQD films. To optimize the performance,
I would need to maintain > 90% transmittance from 400-1200 nm for the ITO top-contact, with <
10 Ω/□ sheet resistance. With these architectural changes, the IQE of the solar cell could be
dramatically improved, which would lead to global improvements in power conversion
efficiency.
41
Chapter 4 Broadband Absorption Enhancement via Periodic
Nanostructuring of Electrodes
4.1 Introduction
In the previous chapter, I discussed the implementation of nanostructured reflectors for
broadband enhancement in colloidal quantum dot solar cells. Despite the excellent potential of
these structures, I encountered restrictions in device architecture, necessitated by the fact that the
device must be fabricated on top of the reflector, that ultimately limit solar cell performance, at
least until further advances in the design of top-illuminated architectures.
In this chapter, I expand on these ideas and focus on transparent substrates so that the bottom-
illuminated configuration known to yield the highest efficiencies for depleted heterojunction
CQD solar cells can be maintained. Since disordered reflectors are no longer employed, I focus
on periodic structures that can provide distinct light management opportunities. To this end, I
pattern glass substrates using polystyrene spheres and etch them to create periodic nanocavities.
The active layers are then deposited in sequence to create the photovoltaic cell.
I begin with a brief overview of optical modelling results to illustrate the ideal periodic
structuring for broadband absorption enhancement in CQD films. I then describe the fabrication
process using polystyrene spheres and buffered-oxide etching, and demonstrate the successful
preparation of nanocavities via SEM imaging. Next, I discuss experimental results, highlighting
broadband absorption enhancement, with corresponding improvements in quantum efficiency
and short-circuit current density. Finally, I discuss the limitations of this design and how further
increases in CQD solar cell performance can be leveraged.
4.2 Periodic Nanostructures for Photon Management
In the field of thin film silicon, a number of studies report on the use of sub-wavelength sized
periodic structures such as 2D arrays of nanodomes/nanostructures for reducing reflectance and
enhancing absorption by coupling to guided modes54,55 and arrays of nanowells for efficient light
collection by optical diffraction56. Other examples of optical absorption enhancement schemes
42
for thin films include, but certainly not limited to: excitation of plasmonic modes in metallic
gratings57–61, localized surface plasmons on metal nanoparticles62, light-trapping using
distributed Bragg reflectors with a grating63,64 or 2D and 3D arrays of photonic crystals65,
formation of whispering-gallery resonant modes in spheres66 and shells67, and enhancement by
leaky-mode resonances in optical antennas68.
Periodic structuring is of particular interest as it can be adapted to well-developed fabrication
methods for CQD solar cells and for the ease of designing structures that leverage light
localization and waveguiding effects. Such periodic structures are typically incorporated into
top-illuminated architectures so that the metal back-reflector can be structured directly before the
active layers are incorporated into the device. For the case of CQD solar cells, however, the best
solar cell efficiencies have been achieved with the bottom-illuminated depleted heterojunction
architecture – the primary restriction arising from the need to heat the n-type window layer
above temperatures suitable for the absorbing CQD film.
This study sought therefore to develop periodic structures into transparent substrates to enhance
broadband absorption while building on the highest efficiency CQD photovoltaic architectures.
4.3 Nanostructure Design
It was required in the design that the collection distance (defined as the maximum distance any
charge carrier would have to travel in the film to be collected at its selective electrode) of 400 nm
should be maintained in order to maximize internal quantum efficiency. In addition, with a
relatively lower index of refraction of n ~ 2.6 for PbS CQD films, the glass should be structured
and conformally coated by a thin tin-doped indium oxide (ITO) bottom electrode to maximize
the refractive index contrast and facilitate multiple internal reflections. Finally, to maximize
absorption while maintaining charge extraction efficiency, the PbS CQD layer should be
conformally coated over the TiO2 window layer.
Three dimensional FDTD simulations were carried out by a collaborator (M. Adachi) to
determine the optimal height and periodicity of features to yield the greatest broadband
absorption enhancement. Simulations were carried out for a hexagonal array of three
dimensional structures (periodicity=1 µm) with periodic boundary conditions in the x and y
directions. A broadband (=400-1200 nm) planewave source polarized along the y-axis was
43
incident from within the glass region of the sample. The absorption in each material was
calculated by integrating the absorption only of matching refractive index of a particular
material. Nanosphere lithography was identified as the best technique for producing periodic
nano-features, and so the shape was approximated to correspond to this patterning technique.
Figure 4-1 illustrates the simulated structure with all materials as labelled in the figure (Au =
gold; MoO3 = molybdenum trioxide; PbS = quantum dot film; TiO2 = titanium dioxide; ITO =
tin-doped indium oxide).
Figure 4-1. Diagram of the nanostructured colloidal quantum dot solar cell. Cross sectional illustration of a 3D hexagonal array of the nanostructured CQD solar cell with an ideally conformal PbS CQD film (with a collection distance = 200 nm) and periodicity of 1 µm 69. Adapted from Ref. 69.
The structure is periodic in the x and y directions but offset from neighboring structures by an
angle of 60, forming a hexagonal array along the x-y plane. Localization of power absorbed per
unit volume, calculated by FDTD at λ = 950 nm, is compared for planar and nanostructured
samples in Figure 4-2a. For the planar device, the power density appears as an interference
pattern between incident and reflected waves. For the nanostructured device, power density is
highest along the edges of the Au contact. The intensity enhancement near the surface of the Au
are due to surface plasmon polaritons, or bound waves that travel along the metal/semiconductor
interface70, in this case the Au/CQD interface.
44
Figure 4-2. Finite-difference time-domain simulations of the nanostructured colloidal quantum dot solar cell. a) The power absorbed per unit volume plot of the equal volume PbS CQD planar
film (upper) and nanostructured PbS CQD film at =950 nm (lower). Only the power absorbed per unit volume in the PbS region is plotted. b) The simulated integrated absorption in the PbS CQD film, comparing the structured case and planar cases with equal film thickness, and equal CQD volume69. Adapted from Ref. 69.
45
In order to appreciate the practical impact of the power density results, absorption in the CQD
layer was simulated for each case (planar and structured) in Figure 4-2b. Three cases are
considered: planar with thickness equal to the 200 nm collection distance; planar with thickness
of 290 nm chosen to equalize CQD volume; and the structured case. The structured PbS film
shows strong broadband absorption enhancement compared to both planar cases, with the
greatest enhancement occurring at long wavelengths ( = 600 to 1200 nm) where the need for
absorption enhancement is the greatest (i.e., where the absorption coefficient is lowest for 1.3 eV
PbS CQD films).
4.4 Nanostructured Substrate Fabrication
Periodic nanostructured substrates were then prepared using nanosphere lithography. The
process flow for substrate patterning is shown in Figure 4-3a. Borofloat glass substrates were
first coated by a closely-packed monolayer of 1 m-diameter polystyrene spheres via spin-
casting in two steps: 10 s at 500 rpm followed by 120 s at 700 rpm. This technique was
employed to optimize surface coverage of nanospheres. Substrates were then subjected to O2
plasma etching to reduce the size of the spheres. Aluminum metal (60 nm in thickness) was then
evaporated (via thermal evaporation) through the spheres and acted as the etch mask for the
reactive ion etch (RIE) etching step. Polystyrene spheres were then lifted-off by ultrasonic
agitation in a dichloromethane bath solution for 1 hour. Reactive ion etching was then performed
using CHF3/O2 gas for 125 s, yielding an etch depth of 400 nm. Finally, the structured substrate
was coated with a 200 nm film of sputtered ITO followed by a 50 nm film of TiO2. A tilted (45°
- for added depth imaging) SEM is included in Figure 4-3b to show the finished substrate. Note
that periodicity in this case is 1 µm as defined by the diameter of nanospheres, with less than 1%
defect density (i.e. misshapen cavities or non-patterned areas) per unit area.
46
Figure 4-3. Fabrication of nanostructured substrates. a) Illustration of the fabrication process for
the periodic nanostructured substrates. The periodicity (sphere size) is 1µm. b) A 45 tilted SEM image of the finished nanostructured substrate69. Adapted from Ref. 69.
Photovoltaic devices were prepared by dip-coating semi-conformal PbS CQD layers onto the
substrates, followed by evaporating MoO3/Au top contacts. Cross-sectional SEM image of the
nanostructured CQD solar cell is shown in Figure 4-4a. A model structure based on the cross-
sectional SEM image is included for clarification in Figure 4-4b. The main difference between
the modelled structure in Figure 4-1 and the actual (Figure 4-4 a and b) is the lack of
conformality of the CQD film. The deposited CQD film is observed to be relatively thin at the
top of the nanostructures, with thickness of approximately 60 nm and relatively thick within the
nanocavities (~ 240 nm). It is postulated that capillary forces led to preferential deposition of
47
CQDs in the cavities of the structured substrate. In addition to the non-conformal CQD film, the
ITO is also slightly thicker at the center of each cavity as compared to the edges, caused by
shadowing effects during sputtered deposition71.
Figure 4-4. Cross-sectional view of the nanostructured CQD solar cell. a) Actual cross-sectional SEM image of the finished device, depicting a non-conformal ITO and CQD film. b) Graphical diagram of the actual nanostructured CQD solar cell illustrating the non-conformal nature of active films in the device69. Adapted from Ref. 69.
48
4.5 Broadband Absorption Enhancement in a Solar Cell
Experimental measurements of total absorption are shown in Figure 4-5 (with and without the
top contact). Both with and without top metal contact, the structured device shows strong
broadband absorption enhancement compared to the planar film, validating original FDTD
simulations.
Figure 4-5. Total absorption of nanostructured CQD solar cells. a) Total absorption for single pass through all layers (without top contact) for planar and nanostructured cases. b) Total absorption of all layers with a reflective top contact69. Adapted from Ref. 69.
The current-voltage (JV) curves under AM 1.5 illumination of nanostructured and planar CQD
solar cells are shown in Figure 4-6a. The structured device exhibits noticeable improvement in
short circuit current density with 20.2 mA/cm2 versus 15.4 mA/cm2 for the planar device,
corresponding to an improvement of 31%. The open-circuit voltage remains unchanged (0.57
V), though the fill factor of the structured device was 51.6% versus 60.2% for the planar device.
The drop in fill factor can be attributed to cracking in the CQD film and less-than optimal
coverage of the top MoO3/Au contact on structured devices since both MoO3 and Au layers were
deposited by a line-of sight deposition technique (thermal evaporation). Overall, the power
conversion efficiency (PCE) of the structured device was 6.0% versus 5.3% for the planar
device. In both the nanostructured and planar devices, the volume of CQDs is less than half of
that used in record performing devices, which typically use film thicknesses in the range of 350 -
400 nm32. External quantum efficiency (EQE) measurements (Figure 4-6b) reflect absorption
49
spectra – the structured device yields greater quantum efficiency in the longer wavelength region
(600 – 1000 nm) as compared to the planar control. The absorption enhancement, and
corresponding improvements in device performance are attributed to enhanced localized
absorption near the metal/semiconductor interface as a result of surface plasmon-polaritons and
strong reduction in broadband reflection.
Figure 4-6. Device performance of nanostructured CQD solar cell. a) J-V measurement under AM1.5 simulated solar illumination for structured and planar samples. b) EQE of both structured and planar samples, which demonstrates improved response in the structured case, reflective of absorption results69. Adapted from Ref. 69.
4.6 Conclusion
In this chapter, it was shown that periodic structured substrates demonstrated broadband
absorption enhancement. This translated into practical improvements in short-circuit current
density and power conversion efficiency in CQD solar cells. Although structured substrates were
prepared by nanosphere lithography as a proof-of-concept in this work, such structuring can also
be prepared by other techniques that can be scaled up to large areas such as nano-imprint
techniques or interference lithography. As such, the structural and optical design discussed in
this work is a viable pathway for better utilizing the full solar spectrum in CQD solar cells.
A major limitation in overall device performance was, however, the non-uniformity of film
thickness for structured devices. The thickness of the CQD film was revealed to be much thicker
within the nanocavities (240 nm) than at the mesas/peaks separating them (60 nm). Increasing
50
CQD film thickness led to even greater non-uniformity, compromising charge extraction
efficiency in the thicker regions of the film. In addition, the Au back-reflector planarized as the
number of layers is increased, reducing the optical benefit. This effect is illustrated in Figure 4-7
wherein a thicker device was fabricated and the film planarized over the nanostructured features.
As a result, overall device thickness was limited to less-than-optimal values (optimal thickness
being approximately 400 nm), ultimately restricting the overall power conversion efficiency of
this strategy.
Figure 4-7. Non-conformality of thick CQD films over nanostructured substrate. Cross-sectional SEM image of a thick CQD film atop a nanostructured electrode, showing the “planarizing” effect that occurs when the film is made too thick.
Future efforts would greatly benefit from realization of a conformal, stress-free semiconductor
film deposition technology for CQD films. In the next chapter, I discuss techniques to engineer
surface hydrophilicity in order to increase the conformal nature of solution-processed CQD
films.
51
Chapter 5 Conformal Fabrication of Colloidal Quantum Dot Solids for
Optically Enhanced Photovoltaics
5.1 Introduction
My goal in this work was to extract the greatest practical benefits from nanostructuring of thin
film absorbers. In the previous chapter, the inability to deposit conformal CQD films severely
limited my pursuit of this goal. While nanocavities that yielded excellent broadband
enhancement in CQD films were deployed successfully, the overall improvements in absorption
and current generation/charge extraction were limited by the lack of conformality of the back
reflector and the additional inability to employ optimal film thickness without further
compromising conformality. These issues are generally less of a problem for vacuum-assisted
deposited materials with greater charge extraction efficiency; this is, however, a much larger
problem for solution-processed materials with inherently limited transport lengths.
In this chapter, I explore conditions for preparing highly conformal solution-processed thin-film
materials for optically enhanced photovoltaic applications. I use a periodic micro-pyramid array
as a test system to investigate techniques for conformal film deposition – which was identified as
the optimal shape for micro-structured devices. I begin with a discussion of device modelling
results of PbS CQD materials to establish the boundaries of film thickness beyond which
transport and performance begin to suffer. I then explore device performance metrics applying
standard film deposition techniques to micro-patterned electrodes. Following compromised
results with non-conformal CQD films I discuss techniques to engineer surface hydrophilicity
and ultimately improve conformality of solution-processed films. I finally demonstrate the value
of conformal films for translating absorption enhancement of optically structured devices into
quantum efficiency, photocurrent generation, and power conversion efficiency.
5.2 Colloidal Quantum Dot Film: Transport Properties
Following studies with nanostructured substrates, I resolved that conformality requirements
become difficult to satisfy when working with structures on the order of film thickness. For this
52
reason, I decided to focus on micro-structured strategies for optical enhancement. To appreciate
the enhancement potential of micro-structured substrates, I first compared the impact on
absorption enhancement of a number of different structures in Figure 5-1. Three dimensional
FDTD simulations were performed to determine the projected improvement for different
periodic shapes with a standard periodicity of 2 µm x 2 µm. The simulations included the full
layer structure of a depleted-heterojunction CQD solar cell24, including 300-nm-thick 1.3 eV PbS
CQD films on patterned/planar titania (TiO2) electrodes. A broadband (λ = 400–1200 nm)
planewave source polarized along the y-axis was incident from within the glass region. The
absorption was isolated for the individual layers and only the absorption of the CQD layer was
presented. Comparing parabolic shapes, cones, and pyramids, I found a distinct improvement in
absorption of longer wavelength photons in all cases relative to the case of a planar sample of
equal film thickness. In the most dramatic case, the structured electrodes led to improvements of
140%, 143%, and 165% at the most weakly absorbed wavelength (~ 840 nm) for parabola, cone
and pyramid structures respectively, as shown in Figure 5-1a. The best-case (internal quantum
efficiency, IQE=100%) short-circuit current density for the three classes of structures relative to
the planar control resulted in improvements of 29%, 33%, and 40% for the parabola, cone, and
pyramid features respectively, as shown in Figure 5-1b. I attribute the high performance of the
pyramid design to its high areal packing density of the structured elements: the pyramids lack
planar plateaus/mesas present in the case of structures based on circular-base periodically-
arrayed repeat units.
53
Figure 5-1. Finite-difference time-domain (FDTD) simulations for periodic micro-structures as compared to planar control. a) Three dimensional(FDTD) simulations comparing the spectral absorption of PbS CQD films atop different periodic shapes compared to a planar control with identical film thickness and b) the integrated short-circuit current density (assuming IQE = 100%) of CQD devices employing each respective periodic shape as its electrode72. Adapted with permission from Ref. 72. Copyright 2015 American Chemical Society.
I explored, in simulation, the importance of ensuring that the absorbing film is conformal over
the structured electrode. I used Sentaurus (3D self-consistent optoelectronic device modelling
software) to compare conformal vs. non-conformal films on a pyramid-shaped electrode for 1.3
eV CQD layers (Figure 5-2). The built-in electric field in the light-absorbing CQD film is, in the
non-conformal case, much less favourable to drift-based charge extraction. It includes regions of
reduced E-field strength where the film thickness exceeds the optimal collection distance (Figure
5-2a). In the conformal case, a uniformly-distributed electric field serves to sweep
photogenerated charges from the device in all locations within the light-absorbing film (Figure 5-
2b).
54
Figure 5-2. Simulated electric field distribution in a PbS CQD film (using the Sentaurus optoelectronic device modeling engine) both in a non-conformal a) and conformal b) context atop a pyramid-shaped electrode72. Adapted with permission from Ref. 72. Copyright 2015 American Chemical Society.
Using additional modeling, I found that the film thickness beyond which charge extraction
efficiency degrades is roughly given by the sum of the depletion width and the diffusion length
for photogenerated charges in the absorbing material. This optimal length for charge extraction is
illustrated in the spatial band diagram of a 1.3 eV CQD film interfaced with TiO2 (Figure 5-3a).
A doping density of ~ 1016 cm-3 is assumed in the TiO2 material73, in which case the depletion
width extends into the CQD film over about ~ 300 nm. The diffusion length has been determined
to be ~ 80 nm for hybrid-passivated CQD solids11 , indicating an extraction length of ~ 380 nm
(Figure 5-3a). More quantitatively, the short-circuit current density (Jsc) is plotted as a function
of film thickness (Figure 5-3b). The predicted Jsc is indeed optimized for a PbS film thickness of
400 nm (W ~ 300 nm and Ld = 80 nm), and then decreases for thicker films despite the increased
volume of absorbing material.
55
Figure 5-3. SCAPS simulation results. a) Band diagram of a depleted heterojunction CQD solar cell
(PbS interfaced with TiO2) to illustrate the band bending at short-circuit conditions and map out
the depletion width and diffusion length in the CQD film. b) Predicted short-circuit current density
(Jsc) as a function of film thickness; Jsc is optimized for a film thickness of 400 nm and then
decreases for increased thickness despite the increased volume of absorptive material. This is
considered the optimal thickness for charge extraction and absorption72. Adapted with
permission from Ref. 72. Copyright 2015 American Chemical Society.
56
5.3 Impact of Non-Conformal CQD Films
A stamp-transfer method was applied to prepare pyramid-patterned TiO2 electrodes (as detailed
in Appendix A3). The purpose was to prepare CQD photovoltaic devices that would leverage
increased optical absorption, and proportionate improvements in Jsc, leading to improved overall
PCE.
In early studies, when preparing the absorbing CQD films, previously-reported film deposition
techniques32 were applied. Spin-coating of oleic acid-capped 1.3 eV PbS CQDs was followed by
a solid-state ligand exchange using 3-mercaptopropionic acid (3-MPA) in a hybrid passivation
approach32. Full-pass absorption measurements reveal a significant improvement for the
pyramid-patterned CQD device compared to the planar control for λ > 700 nm, consistent with
optical modeling predictions. The most dramatic enhancement was observed at ~ 850 nm
wavelength with an approximately 45% increase in absorption (Figure 5-4a). However, in spite
of these substantial absorption enhancements, the pyramid-patterned device exhibited current-
voltage characteristics under AM1.5 simulated illumination that revealed a notable loss in Voc
(26%), FF (12%), Jsc (9%) and consequently PCE (37%) compared to planar controls (Figure 5-
4b).
Figure 5-4. Non-conformal CQD films. a) Full-pass CQD film absorption comparing a pyramid-patterned sample to its planar control. There is drastic improvement in absorption at wavelengths > 700 nm for the pyramid-patterned sample. b) Current-voltage measurements under AM1.5 simulated solar illumination of a PbS device prepared using spin-coated film fabrication techniques72. Adapted with permission from Ref. 72. Copyright 2015 American Chemical Society.
57
Scanning electron microscopy (SEM) revealed non-conformal coverage of the CQD film over
the full extent of the pyramid features (Figure 5-5a). Specifically, the CQD absorber was
excessively (> 1 μm) thick in the valleys between pyramids, and formed a sub-100-nm-thick film
at the apexes. At higher magnification, SEM further illustrates this inconsistency in film
thickness, with a roughly 1 µm collection distance at the thickest point in the film (Figure 5-5b).
The thick regions of the film are expected to reduce charge extraction efficiency and increase
series resistance; furthermore, in the regions beyond the depletion width of the film (> 400 nm),
photogenerated carriers mainly recombine, compromising the open-circuit voltage of the cell.
The thin regions at the pyramid peaks represent lost optical absorption opportunities, and, worse
still, potential shunting paths.
Figure 5-5. Non-conformal CQD films – SEM. a) Scanning electron microscopy (SEM) image of a
pyramid-patterned substrate over-coated with a PbS CQD film deposited using conventional
techniques. Film thickness is inconsistent with very thin regions at the apex, and increasingly thick
regions in the valleys. b) Magnified SEM image of the pyramid valley with non-conformal CQD
film – thickness is measured to exceed 1 µm in this case72. Adapted with permission from Ref. 72.
Copyright 2015 American Chemical Society.
5.4 Technique for Conformal CQD Film Deposition
Therefore, the goal was to develop a procedure to provide conformal deposition of CQD films
atop textured substrates. I took the view that it would be advantageous to reduce the extent to
which the (hydrophobic) solvent used to introduce the CQDs would wet the patterned substrate
surface. A low-concentration primer solution of 3-MPA in methanol was applied to the electrode
58
surface prior to solid-state fabrication of the CQD film, and was re-applied between every
successive layer. The primer treatment is expected to create a monolayer of 3-MPA on the
surface and thereby modify the surface affinity, leading to conformal coating of subsequent CQD
layers. 3-MPA was used for compatibility with the CQD film and for its hydrophilic nature.
To understand the impact of the ligand-primer treatment on conformal film formation, contact
angle measurements were performed on a planar TiO2 substrate prepared using the same material
and treatments as used for pyramid-patterned electrodes. Prior to treatment, the contact angle for
a 4 µL drop of H2O was measured to be 39° (Figure 5-6a). The contact angle for an identical
volume of H2O was measured to be 29° (Figure 5-6b) following ligand-primer treatment of the
substrate. The reduction in contact angle indicates an increased affinity between the surface and
the hydrophilic liquid following ligand-primer treatment. Since the quantum dots (QD) are
dispersed in a hydrophobic solvent, the opposite effect will occur on the QD solution, i.e. the
surface will repel the solution. Wetting properties for patterned surfaces are qualitatively distinct
compared to planar counterparts – capillary forces cause preferential in-filling of porous regions
of the structures74 – leading to a liquid meniscus-effect between adjacent features. With the
ligand-primer treatment applied in this study, the surface becomes more repellant to the QD
solvent and modifies the wetting behaviour, allowing air to create a separation between the
solvent and the surface while the substrate is retracted from the QD solution during film
fabrication. This prevents relaxation of the QD-containing solvent in the pyramid valleys,
ultimately preventing the non-uniform film formation that otherwise occurs as the solvent dries.
This method is, in principle, adaptable and could be used to conformally coat structured features
on a variety of length scales for applications in diverse optoelectronic devices.
59
Figure 5-6. Contact angle measurements of wetting-engineered surface. a) Contact angle
measurement of an untreated TiO2 electrode (using 4 µL deionised H2O). The contact angle was
computed as an average of 39°. b) Contact angle measurement of a ligand-primer treated TiO2
electrode (using 4 µL deionised H2O, again). In this case the contact angle is reduced to 29°,
indicative that this treatment increases the surface affinity for hydrophilic liquids72. Adapted with
permission from Ref. 72. Copyright 2015 American Chemical Society.
5.5 Enhanced Solar Cell Performance with Conformal CQD Films
By applying this ligand-primer technique coupled with dip-coating deposition, I was able to
prepare highly conformal films. SEM images of focused-ion beam (FIB) milled cross-sections
reveal the highly conformal CQD films deposited using this technique (Figure 5-7a). Higher-
magnification SEM images reveal the film to be approximately 400 nm in thickness over a large
range of the pyramid features, close to the optimal film thickness for efficient charge extraction
(Figure 5-7b). For added confirmation a top-view SEM of an area ~ 50 x 50 µm2 of the pyramid-
patterned surface is included, showing CQD films to be conformal over an array of 25 pyramid
features (Figure 5-7c).
60
Figure 5-7. Conformal CQD films. a) Scanning electron microscopy image of a focus ion beam-milled cross-section illustrating the resulting device when applying dip-coating + ligand primer treatment to the pyramid-patterned TiO2 electrode. The film thickness is consistent over the surface of the pyramid with this treatment. b) Magnified SEM of the pyramid valley with conformal CQD film – thickness is measured to be ~ 400 nm over the full range of the image. c) Large-area SEM image illustrating the uniform film coverage across the device72. Adapted with permission from Ref. 72. Copyright 2015 American Chemical Society.
Investigating quantitatively, absorption (isolated CQD film absorption – methods in Appendix
A10) and external quantum efficiency (EQE) are compared in Figure 5-8a. In all cases, film
absorption is approximately equivalent when employing pyramid-patterned electrodes. There is,
however, a much greater translation of absorption benefits into device EQE when employing the
61
surface treatment detailed in section 5.4. These spectra were used to calculate internal quantum
efficiency (IQE) - a measure of charge extraction efficiency - for CQD pyramid-patterned
devices prepared using conventional spin-coating methods, standard dip-coating, and dip-coating
combined with the ligand primer treatment (Figure 5-8b) Compared to the non-conformal spin-
coated case, a global improvement of IQE over all wavelengths is observed when employing
similarly non-conformal passive-drying deposition techniques such as dip-coating. With the
addition of surface functionalization via 3-MPA primer treatment a similar, more quantitatively
significant global improvement is observed for all wavelengths compared to the non-conformal
cases. Improved charge extraction efficiency for photons of λ > 850 nm is further observed
compared to the dip-coated case without the ligand primer treatment. The IQE is most improved
at the exciton peak (~ 960 nm) when using the primer technique, providing enhancements of
26% and 17% relative to the reference spin-coated and dip-coated cases, respectively. This
dramatic improvement in device IQE is attributed to the improved conformality of the CQD film,
which allows for efficient extraction of photogenerated carriers from a larger total volume of the
light-absorbing film.
62
Figure 5-8. Quantum efficiency for conformal and non-conformal CQD films. a) Comparison of the total absorption and EQE for spin-coated, dip-coated and dip-coated + primer samples (conformal case), with the conformal case exhibiting the highest EQE response in light of relatively similar absorption spectra. b) Calculated IQE comparing the same three cases, illustrating again that IQE/charge extraction efficiency is optimized in the conformal case72. Adapted with permission from Ref. 72. Copyright 2015 American Chemical Society.
With the application of the ligand primer treatment I was able to optimize charge extraction and
achieve improved device performance for pyramid-patterned CQD samples. Current-voltage
(JV) measurements under AM1.5 simulated solar illumination reveal a representative PCE of
8.5% using pyramid-patterned TiO2 electrode compared to a planar control PCE of 7.7% (Figure
5-9). With conformal CQD films there is minimal compromise in Voc and FF (no change in Voc,
63
and 11% reduction in FF). The major improvement is in Jsc (increased by 27%) for a global
improvement in power conversion efficiency (10% improvement in this case).
Figure 5-9. Device performance comparing conformal CQD case with planar control. J-V curve as measured under AM1.5 simulated solar illumination showing that device performance is in fact improved when applying conformal CQD films over optically-enhanced micro-patterned electrodes72. Adapted with permission from Ref. 72. Copyright 2015 American Chemical Society.
5.6 Conclusion
The conformal deposition of CQD films onto structured substrates has applications within, and
potentially beyond, photovoltaics. Its benefits could potentially be realized in the formation of
photodetector arrays on curved and structured substrates; in the fabrication of wide-viewing-
angle displays; and in coating active layers on 3D-structured cavities to form lasers such as in
whispering-gallery-mode (WGM) devices. In the photovoltaics context, the technique yielded a
notable performance improvement when CQD films were deposited on pyramid-patterned
electrodes. Conformal films atop the structured substrate produced proportional improvements in
Jsc and power conversion efficiency relative to corresponding planar controls, validating the
method in enabling optically-enhanced CQD photovoltaics.
With proven techniques for conformal film deposition I then turned my attention to optimizing
the structured substrate strategy with the goal of yielding record power conversion efficiencies
64
for PbS CQD photovoltaics. In the next chapter, I explore optimized structures and their impact
on performance when applying optimal device architectures and materials.
65
Chapter 6 Colloidal Quantum Dot Solar Cells Exploiting Hierarchical
Structuring
6.1 Introduction
The previous chapter described techniques to coat active films conformally towards the goal of
optically enhanced CQD film absorption. This chapter moves ahead to optimize the structures
thus-enabled, and ultimately seeks to report a record power conversion efficiency for CQD solar
cells. I identified in the last chapter that microstructures must be employed to maximize
conformality of solution-processed films. I then compared structures and revealed pyramids to be
ideal due to high areal packing density with angled sidewalls/tilted films (for longest path
length).
I then postulated that an ideal enhancement strategy should employ the best available CQD
materials in each sublayer of the active device, and should minimize loss in open circuit voltage
(Voc) and fill factor (FF) associated with increased junction area. I would require that any added
processing steps would be scalable and low in implementation cost.
In this chapter I explore the optimal structures (pyramids) for optical enhancement in thin-film
solar cells and investigate the impact of pyramid sidewall angle on film absorption. I then model
these structures to determine the optimal periodicity, and explain the impact of size vs
wavelength for broadband enhancement. I then develop techniques, leveraging natural etch
planes of silicon, to stamp these structures into n-type window-layer electrodes (TiO2) as part of
a depleted heterojunction CQD solar cell. I finally demonstrate improved absorption, in good
agreement with FDTD optical models, and show that I can translate these dramatic
improvements into quantum efficiency, current generation and power conversion efficiency of
the CQD solar cell applying a hierarchical structuring strategy with conformal films.
6.2 Effect of Pyramid Sidewall Angle
The absorption advantage of the pyramid structure relies on in-film scattering from the back
electrode and results from the increased light path through the active material (Figure 6-1).
66
Figure 6-1. Diagram illustrating the increased-light-path advantage of pyramid-patterned electrodes, with detailed equations of the light-film interaction as a function of pyramid angle given below75. Adapted with permission from Ref. 75. Copyright 2014 American Chemical Society.
In a planar device, with a reflective top contact, light will pass through the active material twice,
with a path length of ~ 2x thickness for normal incidence irradiation. For the pyramid-patterned
case, light is trapped within the structure and will reflect internally as a function of pyramid
angle – with approximately 4 passes through the active material in the case of 54.7°-pyramids,
and up to 12 passes for the 80°-pyramids case. I include a diagram in Figure 6-2 to illustrate the
path length of light through the active film. The spectral absorption of the CQD film, as a
function of thickness, is given by equation 6-1:
Equation (6-1): 𝐴(𝜆) = 1 − 𝑒−𝛼(𝜆)𝑝
where p is the total light path through the active film and α(λ) is the spectral absorption
coefficient for the CQD material. The length of each pass is further increased by a factor of
1/cos(αi), wherein the total light path (p) is given by equation 6-2:
Equation (6-2): 𝑝 = ∑2𝑡
cos(𝛼𝑖)
𝑚𝑖=1
67
where t is the CQD film thickness, m (number of reflections) is largest integer ≤3𝜃−2𝜋
2𝜃−𝜋 and αi
can be described by equation 6-3. Therefore, for θ = 54.7°, m = 2 (4 passes through active
material – 2 passes for every reflection); for θ = 80°, m = 6 (12 passes through active material).
Equation (6-3): 𝑛1𝑠𝑖𝑛𝜑𝑖 = 𝑛2𝑠𝑖𝑛𝛼𝑖
Where n1 is the index of refraction of the substrate, n2 is the index of refraction of the absorbing
material, and φi can be described by equation 6-4:
Equation (6-4): 𝜑𝑖 = (𝑖 − 1)𝜋 − (2𝑖 − 1)𝜃;𝜑1 = 𝜃
where θ is the angle of the pyramid.
Figure 6-2. More detailed diagram of the light path through the pyramid-patterned CQD film, labelling the angles used in equations to calculate the total light path.
To determine the quantitative absorption enhancement in a photovoltaic device I performed 2D
FDTD simulations for the complete layer structure of a depleted heterojunction CQD solar cell,
including 300 nm-thick, 1.3 eV bandgap PbS CQD films on pyramid-patterned TiO2 electrodes.
Note that the simulation space in 2D included mid-line along the z-axis, with periodic boundary
conditions in the x direction. A broadband (λ = 400–1200 nm) planewave source polarized along
the y-axis was incident from within the glass region. The spectral absorption of the quantum dot
68
film is presented as a function of pyramid inclination angle (Figure 6-3a). Absorption is
0significantly increased for pyramid angles that cause additional passes through the CQD film.
The most dramatic enhancement occurs for the 80°-pyramid case, in which near-complete
absorption is expected over the spectral range of interest for this material. FDTD simulations
reveal a relative enhancement of more than 50% at the exciton peak (950 nm wavelength), with a
maximum enhancement of ~ 200% at the weakly-absorbed (excitonic valley) wavelength of 850
nm. For a pyramid angle of 54.7°, there is a relative absorption enhancement of ~ 100% at this
wavelength of 850 nm. Absorption is increased for specific critical angles at which extra
reflections would occur inside the pyramid-shaped active layer. The trend does not, however,
indicate that larger angle leads to increased absorption in all cases. It was noted that for 30°
pyramids, the absorption is in fact slightly higher than for 54.7°. Applying theoretical predictions
according to equations 6-1 and 6-2 presented above, p (interaction length of light with active
material) is found to be approximately 6.0 times the thickness of the film for the 30° case, while
p is 4.8 times the film thickness for 54.7°. Despite this theoretical advantage we chose to work
with 54.7° pyramids for practical reasons: silicon’s natural etch plane of 54.7° directly facilitates
implementation of pyramids exhibiting these sidewall angles.
69
Figure 6-3. Theoretical predictions for pyramid-patterned thin-film solar cell as a function of sidewall angle. a) Results of 2D FDTD simulations of absorption as a function of pyramid sidewall angle. Absorption improves most dramatically at wavelengths > 600 nm with near complete absorption for the 80°-pyramid case. b) Projected spectral current-density plotted as a function of wavelength (assuming 100% IQE)75. Adapted with permission from Ref. 75. Copyright 2014 American Chemical Society.
The spectral photogenerated current-density (Jsph) was plotted as a function of wavelength for
planar, 54.7°-pyramids and 80°-pyramids. Jsph is the current density that would flow if internal
quantum efficiency (IQE) were 100%, i.e., it is the absorbed photon current. The maximum
available current density (assuming 100% absorption of the AM1.5 spectrum) is included for
comparison (Figure 6-3b). Comparing with the planar 300-nm-thick-active-layer CQD device,
the integrated Jph is predicted to improve by approximately 25% for the 54.7°-pyramids case, and
nearly 50% for the 80°-pyramids case.
Next, I investigated the impact of angle of inclination on overall power conversion efficiency,
taking account not only photocurrent enhancement, but also increased interfacial area and its
impact on open-circuit voltage. Note that Voc can be described by equation 6-5 under short-
circuit conditions. With effective area of the junction increasing as a function of pyramid angle
for an equivalent area of illumination, Io will increase with pyramid angle (Io proportional to
junction area). Voc is then modified a quantity ∆Voc according to equation 6-6 as a function of θ
(pyramid angle). Starting from best-case reported values32 for Voc for the planar case (0.60V),
Voc will evolve according to Figure 6-4a:
70
Equation (6-5): 𝑉𝑜𝑐 =𝑘𝑇
𝑞ln(
𝐼𝑠𝑐
𝐼𝑜− 1)
Equation (6-6): ∆𝑉𝑜𝑐 = −𝑘𝑇
𝑞ln(𝐴𝐹) =
𝑘𝑇
𝑞ln(𝑐𝑜𝑠𝜃)
where AF = area factor = area of pyramid side walls/area of base = 4∗[0.5∗(𝑏)∗(0.5∗𝑏/𝑐𝑜𝑠𝜃)]
𝑏2=
1
𝑐𝑜𝑠𝜃.
Further, to add realism and facilitate comparison with experiments, I also applied
experimentally-determined spectra of IQE in estimating the short circuit current (Jsc) (Figure 6-
5). Using optical and electronic parameters for the best previously-reported CQD solids, I
predicted a PCE of ~ 10% for 300 nm-thick PbS CQD films conformally-coated atop 80°-
pyramids, with projected Jsc and Voc of 30 mA/cm2 and 0.56 V, respectively. The PCE for 54.7°-
pyramids was predicted to be ~ 9.2% compared with a predicted PCE of ~ 7.7% PCE for the
planar case (Figure 6-4b).
Figure 6-4. Evolution of projected Voc, Jsc and PCE as a function of pyramid sidewall angle. a) Plot of Voc (red, left axis) and Jsc (blue, right axis) as a function of pyramid sidewall angle. b) Plot of total PCE as a function of pyramid sidewall angle.
71
Figure 6-5. Internal quantum efficiency curve for a standard CQD solar cell. Used for calculation of projected Jsc and PCE in Figure 6-4.
6.3 Hierarchical Structuring for Optimal Broadband Enhancement
I then explored whether the in-plane period of pyramid position impacted the optical structural
enhancements of interest. Three dimensional FDTD simulations were used for this purpose. For
these simulations, periodic boundary conditions were modelled in the x and y directions. Again,
a broadband (λ = 400–1200 nm) planewave source polarized along the y-axis was incident from
within the glass region. Due to the time required to run 3D simulations for periodicity > 1 µm, I
employed only a single polarization of the light source as opposed to rotating the source over a
range of polarization angles to simulate a non-coherent source such as the Sun. Note that results
produced with a similar broadband planewave source polarized along the x-axis (perpendicular
polarization) yield overlapping absorption curves due to the symmetry of the pyramid shape. In
Figure 6-6 photogenerated current-density (Jph) is plotted as a function of pyramid pitch for
54.7°-pyramids. Optimal enhancement in current generation occurs for ~ micron-scale pitch. The
improvement as a function of pitch saturates for period > 2 µm in the case of 300 nm thick CQD
films. Comparing absorbed power density for a pyramid-pitch of 500 nm vs 5 µm (inset of
Figure 6-6) standing wave interference patterns can be observed in the sub-micron case. Such
effects limit the potential for absorption enhancement for longer wavelength radiation (e.g., out
72
to λ = 1000 nm in the CQD absorber case of interest herein). This effect was observed in other
reports investigating the effects of inverted-pyramid pitch on optical reflection for silicon solar
cells76,77. As detailed in one of these reports77, for wavelengths greater than the period of the
pyramids, only a zeroth order wave is present and propagates essentially as a plane wave along
the light incidence axis. For D > 2λ, the system is geometric-optics dominated and transmission
through the surface is subsequently maximized77. In sum, in the case of CQD solar cells, which
absorb up to ~1000 nm, > 2 µm pitch is required to ensure a geometric-optics dominated regime
which leads to the most optimal degree of light scattering into the film. This limitation
necessitates a structuring strategy on the multi-micron length-scale. For added ease of
fabrication, I selected a periodicity of 10 µm for this study.
Figure 6-6. Results of 3D FDTD simulations of projected Jph as a function of pyramid pitch. Note that projected current generation is optimized for pitch of ~ 2 µm and this effect saturates for periodicity > 2 µm. (inset left) Absorbed power density map for 500 nm pitch pyramid (λ = 950 nm). Note standing-wave interference patterns similar to those expected in planar absorption profiles for similar film thicknesses. (inset right) Absorbed power density map for 5 µm pitch pyramid (λ = 950 nm). In this case the absorption is reflective of geometric scattering within the active film for long wavelength radiation (λ = 950 nm)75. Adapted with permission from Ref. 75. Copyright 2014 American Chemical Society.
I term this concept hierarchical structuring, as three different length scales are exploited in these
devices: I first employ quantum dots which are structured on the few-nanometer length scale
(~3-5 nm in diameter, necessary for bandgap tuning), that are assembled into thin films
structured on the 100-nanometer length scale (~300-400 nm thickness, necessary for efficient
73
charge extraction), atop micron length scale pyramids (~2-10 µm, necessary for optimum light
collection), thus applying a hierarchy of size scales for optimal results.
6.4 Fabrication of Pyramid-Patterned Electrodes
I created first a template for pyramids using anisotropic etching of silicon. Wet-etch techniques
were then applied to <100>-oriented wafers, producing inverted pyramid features with
inclination angle of 54.7° (Figure 6-7a) based on a periodic pattern. These have much in
common with the PERL (passivated emitter rear locally-diffused) silicon photovoltaic cell78–80,
which employ inverted pyramidal structures to reduce reflection at the silicon-air interface. In
this work this concept was deployed for increased light recycling in the active film, an
application that would prove unnecessary for silicon as it is already a complete absorber of
above-bandgap photons at typical device thicknesses. The silicon master was then prepared using
photolithography (see Appendix A3 for details). After silanization treatment81 of the patterned
surface of the silicon master, polydimethylsiloxane (PDMS) was moulded to achieve high-
fidelity transfer of pyramid features. A second transfer process was applied to print reciprocal
inverted-pyramid structures into PDMS to form the stamp. After applying a thin viscous film of
TiO2 nanoparticle solution on the substrate, the PDMS stamp (with inverted pyramids) was then
applied and solvents were allowed to dry (Figure 6-7b).
74
Figure 6-7. Fabrication of pyramid-patterned titania (TiO2) electrodes. a) Illustrating the effects of anisotropic etching on <100>-oriented silicon. The <100> plane etches preferentially at a rate 100x faster than the <111> plane, creating inverted pyramid features with angle of 54.7°. b) Fabrication flow for preparing pyramid-patterned electrodes. The PDMS mould is first prepared using the patterned silicon wafer as a master; a reciprocal PDMS stamp is prepared from this mould to obtain inverted pyramid features. This stamp is then used to pattern the TiO2 electrode75. Adapted with permission from Ref. 75. Copyright 2014 American Chemical Society.
The result was pyramid-patterned TiO2 films (Figure 6-8) with pitch 10 µm, pyramid base widths
of 8-9 µm, and pyramid height of approximately 5 µm.
75
Figure 6-8. Scanning electron microscopy images of pyramid-patterned TiO2 electrodes. a) Top-view SEM image of 50 x 50 μm2 area. b) Cross-sectional SEM image of a single TiO2 pyramid75. Adapted with permission from Ref. 75. Copyright 2014 American Chemical Society.
6.5 Absorption and External Quantum Efficiency Enhancement
With conformal PbS CQD films on pyramid-patterned electrodes I was able to achieve consistent
performance improvements over planar controls. Scanning electron microscopy (SEM) images
compare CQD films on both planar (Figure 6-9a) and pyramid-patterned electrodes (Figure 6-
9b). Film thickness was measured to be approximately 350 ± 30 nm for the planar case, and
approximately 400 ± 30 nm in the pyramid case – I consider these thicknesses to be in agreement
taking into account the considerable experimental uncertainty of these measurements.
76
Figure 6-9. Scanning electron microscopy images of full CQD devices. a) Cross-sectional SEM of planar CQD device. The film thickness is ~ 350 ± 30 nm. b) Cross-sectional SEM of a pyramid-patterned CQD device. The film thickness is ~ 400 ± 30 nm75. Adapted with permission from Ref. 75. Copyright 2014 American Chemical Society.
Total absorption (including absorption of FTO and top contact) is compared for both samples
(Figure 6-10a). Results are in excellent agreement with simulated (2D FDTD) spectra of total
absorption (for 54.7° pyramids and 300-nm-thick PbS films); with significantly improved
absorption at wavelengths > 600 nm for the pyramid-patterned sample. In the most dramatic
case, total absorption is improved by ~ 72% for the pyramid-patterned case at weakly absorbed
wavelengths (λ = 830 nm). There are additional peaks in the simulated absorption plots that arise
from thin film interference effects, partially due to the coherent source used in simulation.
However, for the experimentally-built pyramids, it was found that the somewhat curved peaks
and valleys, and their slightly rough sidewalls, produce an effective spectral averaging that
77
diminishes these sharp resonances. A similar effect is observed for planar devices which are built
atop rough fluorine-doped tin oxide (FTO) films.
External quantum efficiency (EQE) of pyramid-patterned and planar control samples was also
compared (Figure 6-10b). The predicted Jsc from measured EQE agrees well, to within 5%, with
measured values of Jsc under AM1.5 illumination. I observe a substantial enhancement of EQE in
the infrared spectral region for the structured sample, with ~ 100% enhancement at λ ~ 850 nm,
indicative that improved device performance is the result of increased light path through the
active layer, as indicated by absorption measurements.
Figure 6-10. Enhanced absorption and external quantum efficiency for pyramid-patterned CQD solar cell. a) Total absorption (including absorption of FTO and top contact materials) of pyramid-patterned device compared to planar control. There is increased absorption at wavelengths > 600 nm for the structured device, and excellent agreement with 2D FDTD simulation data for both samples. b) External quantum efficiency (EQE) of a pyramid-patterned sample vs. a planar control. The EQE response is most improved for the wavelength range between 700 and 1000 nm, with ~ 100% improvement at 850 nm over the planar control75. Adapted with permission from Ref. 75. Copyright 2014 American Chemical Society.
6.6 High-Efficiency Colloidal Quantum Dot Solar Cells
Current density-voltage (J-V) characteristics under AM1.5 simulated solar illumination for the
record pyramid-patterned device were also compared with the best planar control (Figure 6-11a).
A power conversion efficiency of 9.2% was obtained, matching a lab-record for colloidal
quantum dot solar cells. This represents an enhancement of 15% in PCE over the best planar
78
control with an overall improvement of ~ 24% in Jsc. Open-circuit voltage and fill-factor were
comparable (to within experimental uncertainty) for the two devices. Thus, conformal films
preserve the FF in the structured case, and any loss in Voc is minimal, as projected (< 0.02 V) for
54.7° pyramids. Note that minor variances in film thickness, however, do lead to a consistently
lower value of FF for the pyramid-patterned case. As expected from theoretical treatments and
quantum efficiency measurements, the pyramid-patterned device generates additional photo-
current from increased absorption inside the same thickness of active material (perpendicular to
the electron-accepting electrode). As discussed, the limited charge transport of CQD films
dictates that the most efficient thickness for transport is too thin to achieve complete absorption
of incident radiation. The pyramids effectively enhance absorption in the active film without
increasing the electrode-to-electrode distance, and they do so by increasing both the angle of
transmission through the active layer, and the number of optical passes in the device: 4 effective
passes occur in the pyramid devices compared to the 2 that occur in a planar device.
To characterize reproducibility, all device results were plotted over a one month period (with
optimized stamping and film fabrication procedures) in a histogram (Figure 6-11b). For a sample
size of 126 pyramid-patterned devices (i.e., 126 individual devices as defined by the 0.049 cm2
contact area) prepared with similar substrate stamping and film fabrication protocols, an average
power conversion efficiency of 8.2% with a standard deviation of ± 0.6% was reported. For
comparison, the average device performance for a sample size of 124 planar devices (also
defined as 124 individual 0.049 cm2 devices) is 6.5 ± 0.8%.
79
Figure 6-11. Enhanced performance of pyramid-patterned CQD thin-film solar cell. a) J-V curve comparing device performance of a pyramid-patterned sample vs. a planar control. PCE was measured as 9.2% for the pyramid-patterned sample, a 15% improvement over the planar control. b) Histogram of pyramid-patterned CQD device performance over a 1 month period (red) and planar control device performance (blue). Average PCE of 8.2% ± 0.6% was computed for a sample size of 126 pyramid-patterned samples. Comparatively, the lab-wide average (neglecting outliers) for planar CQD devices was 6.5% ± 0.8% for a sample size of 12475. Adapted with permission from Ref. 75. Copyright 2014 American Chemical Society.
6.7 Conclusion
In summary, I made use of a micro-structured substrate for increased absorption in transport-
limited CQD films, and applied conformal absorbing films to leverage the full potential of the
80
structured design. Using micron-scale pyramid features as part of a hierarchical structured
device, I achieved an impressive 9.2% solar PCE as a result of a 24% improvement in Jsc over
the planar control. When statistics were gathered for over 100 devices, I achieved a 26%
improvement in the median pyramid device relative to the median planar control.
As demonstrated in this study, there exists potential for additional absorption enhancement when
using steeper-angle pyramids. The good agreement between experiment and theory herein
suggests that the 80° pyramids could indeed enable devices that reach above 10% solar PCE.
These structures could further be adapted to other thin-film technologies or alternative CQD-
based films for comparable absorption enhancement and subsequent current generation benefits,
with minimal losses in open-circuit voltage and fill-factor.
81
Chapter 7 Conclusions and Future Work
7.1 Conclusion
This work began with a summary of the crucial importance of advancing solar cell technologies
towards the goals of manufacturability, low cost, large area, low materials utilization, and, ever-
importantly, power conversion efficiency.
The objective of my thesis was to advance this goal via photonic enhancement strategies that
focus on increasing the light interaction length within thin film materials used for photovoltaic
applications. The work began with a review of key background material, summarizing advances
in colloidal quantum solids and solar cells, noting the key milestones that led up to the
beginnings of the thesis. I also reviewed advances in structured substrates and plasmonic
materials and their applications in enhancing absorption in thin-film photovoltaics.
In the first key advance, I used electrochemical techniques to create highly disordered metallic
nanostructures and demonstrated control over feature size and density with deposition
conditions. Nanostructured reflectors were combined with PbS CQD films and broadband
absorption enhancement was demonstrated, with increased effect for thinner films. This concept
was finally deployed in a test solar cell, demonstrating improvements in short-circuit current
density (up to 34%) and power conversion efficiency (up to 25%). It was concluded, however,
that ultimate power conversion efficiency of the PbS CQD solar cell was restricted by the need
for top-illuminated architectures necessitated by the implementation of nanostructured reflectors.
Following this work I implemented nanostructured transparent substrates for absorption
enhancement in CQD solar cells. Periodic nanostructures were created in glass using nanosphere
lithography techniques, and all active solar cell layers were deposited on top, conserving bottom-
illuminated architecture of best-efficiency CQD cells. CQD films were dip-coated and
photovoltaic devices were shown to yield improved broadband absorption in the infrared
wavelengths, with corresponding improvements in EQE, short-circuit current density, and power
conversion efficiency. Due to conformal coating issues, however, overall film thickness was
restricted to well below optimal values; it was also noted that film thickness was non-uniform
over the nanostructured surface, ultimately limiting charge extraction efficiency of the device, as
82
reflected in fill factor. Power conversion efficiencies were consequently low in comparison to
field-leading values and I resolved that in order to make impactful advances to the field
conformal CQD films over optically enhanced electrodes would be required.
I then explored the impact of conformal film deposition on the performance of optically
enhanced CQD solar cells more directly. Micro-structured pyramids were identified as promising
structures for broadband absorption enhancement. Techniques to engineer the surface
hydrophilicity of micro-structured electrodes were then developed and successfully deployed to
fabricate conformal CQD films. It was shown that, with conformal films, charge extraction
efficiency (as represented through IQE) of the patterned device can be increased – leading to
proportional improvements in external quantum efficiency for optically enhanced architectures.
These techniques yielded measurable improvements in practical device performance and were
deployed for the culmination of the thesis in the final project.
In my final effort, I sought an optimal strategy for light recycling in CQD solar cells. I had
identified micro-pyramids as good structures for broadband enhancement and the optimal
configuration of pyramid features was further explored in terms of sidewall angle and
periodicity. Absorption enhancement was found to increase as a function of pyramid angle; more
importantly, the greatest broadband enhancement was determined to require micron-scale
pyramids with periodicity > 2 μm (approximately 2x the longest wavelength of interest). This led
to the hierarchical structuring concept in which a hierarchy of size-scales are required for optimal
enhancement, from nanometer-size quantum dots, assembled into 100’s-nm-thick films, atop
micron-scale structures. Micron-scale pyramid-patterned electrodes were prepared taking
advantage of silicon’s natural etch planes and employing wet etch and soft lithography
techniques. These structured substrates were then combined with conformal CQD films to yield
broadband enhancement, which translated into practical enhancement of external quantum
efficiency, photogenerated current (up to 24%), and power conversion efficiency (up to 15%).
7.2 Impact of the Work
My work has implications for the field of CQD solar cells and solution-processed thin film solar
cells in general. The work with nanostructured disordered reflectors opens the door for
engineering materials for complete absorption in extremely thin film devices. The underlying
83
concepts of the nanostructured material enable broadband electromagnetic radiation to be
effectively trapped within an active film. This could be extended to vacuum-deposited organic
thin films that tend to require extremely thin layers due to limited transport properties (and since
they do not employ the bulk heterojunction architecture).
Conformal coating techniques can be extended to other solution-processed materials and for
numerous optoelectronic applications. Many applications can benefit from explicit control over
the charge collection distance and wettability engineering techniques presented in this work can
form the foundation for future advances in that respect. The new materials fabrication method
can be also be applied to alternative textures that provide for increased device enhancement
potential. Finally, the work on micro-pyramidal electrodes is the most significant advance of this
thesis. I discovered the broadband improvement that can be leveraged with micro-scale
structuring and successfully designed and optimized a portable, scalable stamping technique for
photonic enhancement in thin film devices. This technique is the first successful implementation
of a self-contained, light recycling strategy for CQD thin film solar cells that has yielded
impactful improvements in device performance. Furthermore, it can be applied widely and
advantageously to overcoming absorption-extraction compromises across the broad spectrum of
emerging solar technologies, including organic, perovskite, and inorganic photovoltaics.
7.3 Future Work
Further advances in CQD materials, following the bulk of the work of this thesis, led to the
highest published certified efficiency of 8.55% by replacing the TiO2 window layer with zinc
oxide (ZnO), and replacing the cross-linking ligand 3-MPA with TBAI (tetrabutyl ammonium
iodide) and EDT (ethanedithiol). An additional certified PCE record of 9.9% has since been
achieved using a similar device architecture.
7.3.1 Higher Aspect-Ratio Pyramids
As revealed in Chapter 6, broadband enhancement is increased as a function of pyramid sidewall
angle. Therefore, a logical future project would be to extend the work of this thesis to 80°
pyramids. Initial discussions have identified reactive ion etching of silicon as a potential
84
technique to create high-aspect ratio features with a roughly pyramidal/conic shape. Work is
underway to create the features depicted in Figure 7-1 as part of a collaborative project. While
the high angle of these features signal the potential for significant enhancement, the limited areal
packing density (as limited by the RIE technique used to create the patterns) could ultimately
counteract any benefits of this structure (as compared to the 54.7° pyramids developed in
Chapter 6). Future work would need to build on these concepts to create high angle (ideally 80°)
pyramidal features with high areal packing density (> 80%).
Figure 7-1. Techniques for preparing higher angle pyramid-patterned electrodes. Reactive ion etching of individual units using a silicon mold.
7.3.2 Standardized Conformal Deposition Techniques
The conformal coating techniques developed in this work were largely tailored for the
application with micro-structured pyramids. As the aspect ratio increases, or the feature size
decreases, I anticipate these techniques may not hold up. Ideally, I would like to extend the
85
concepts explored in Chapter 5 (on the engineering of surface hydrophilicity) to allow for
conformal deposition over all feature scales and aspect ratios. This would be an interesting future
project coupled with the project proposed in section 7.3.1 on the development of 80° pyramid-
patterned electrodes.
7.3.3 Standardized Pyramid-Patterned Electrodes for Different Material
Stacks
During the course of this work, the optimal device architecture shifted from depleted
heterojunction devices employing TiO2 window layers and MPA cross-linked PbS films to ZnO
window layers with TBAI/EDT treated PbS films. This material stack leads to dramatic
improvements in device stability and consequent improvements in overall power conversion
efficiency. We are actively investigating strategies to incorporate the ZnO/TBAI-treated
PbS/EDT-treated PbS device stack into pyramid-patterned devices. Initial studies suggest that
patterning ZnO at the micro-scale would compromise electron mobility of the device and
ultimately limit device efficiency. A probable solution would be to employ pyramid-patterned
TiO2 electrodes as substrates (non-active component) in the device and build all active layers on
top in the following stack TCO/ZnO/TBAI-treated PbS/EDT-treated PbS/Au.
7.3.4 Efficient Top-Illuminated CQD Architectures
As discussed in Chapter 3, the primary limitation in deploying nanostructured reflectors for high
efficiency CQD solar cells is the lack of efficient top-illuminated architectures. I also discussed
material advances that would be required to progress on the development of top-illuminated
architectures, namely improved TCOs and better adhesion between solution-processed CQD
films and metallic substrates. High efficiency top-illuminated architectures would open the door
for numerous applications with the added versatility of being able to deposit active devices atop
a host of reflective/non-transparent substrates.
86
7.3.5 Bottom-Illuminated Nanostructured Disordered Electrodes
As an idea that can bridge techniques applied in both Chapter 3 and Chapter 6, we could apply
soft-lithography techniques developed in Chapter 6 to create PDMS moulds with the
nanostructured disordered features developed in Chapter 3. This stamp could then be used to
stamp either a transparent conducting oxide layer (solution-phase ITO as an example), TiO2
electrodes, or even the CQD film to imprint a nanostructured surface. Then, the deposited Au (as
top contact) should take the form of the nanostructured disordered reflector used to stamp the
underlying layers. This should allow for leveraging the photonic benefits of nanostructured Au
reflectors, while maintaining optimal bottom illuminated architectures.
7.4 Perspectives of CQD Photovoltaics
Colloidal quantum dot photovoltaics remains an intriguing field among third-generation,
emerging technologies. As iterated numerous times throughout the thesis, these materials, which
are based on relatively complex fundamental units, benefit from fundamental and practical
benefits for photovoltaic applications, including band gap tunability (with a range extending to
0.4 eV into the infrared), and the use of low-cost, abundant materials that can be cheaply
processed into light-absorbing films. While there exists competition in the form of perovskite
materials that leverage many of the practical benefits of CQD technologies, achieving PCE of up
to 20%, the fundamental properties of CQD represents an advantage for multi-junction, high
performance applications. With recent advances in performance and stability, I expect the long
term direction of CQD research will focus on leveraging its band gap tunability into the infrared
for implementation into existing modules (Si-based), or more sophisticated multi-junction
applications, bypassing lattice-matching restrictions of multi-material systems.
7.5 Concluding Remarks
A study of optical enhancement of colloidal quantum dot thin film solar cells was performed on
route to completing my thesis. The most significant contributions have yielded substantial
advances in cell efficiency, and further initiated ideas for future work in the field. I was able to
match theoretical expectations with the most successful project, the micro-patterned pyramid
87
electrodes, and further mapped out the requirements for complete absorption in a CQD thin film
device. With additional efforts on this topic I anticipate that we can maximize the photocurrent
and effectively overcome limitations in charge transport for this solution-processed photovoltaic
material. With increased efficiency, this technology becomes more viable in the context of cost
as a function of power generation, pushing the photovoltaic field forward as a global renewable
resource.
88
References
1. International Energy Agency. Key World Energy Statistics 2014. (2014).
2. International Energy Agency. Renewable Energy Medium-Term Market Report 2014. (2014).
3. Becquerel, A.-E. Mémoire sur les Effets Électriques Produits sous L’influence des Rayons
Solaires. Comptes Rendus Séances Hebd. 9, 561–567 (1839).
4. Fritts, C. E. On a New Form of Selenium Photocell. Am. J Sci. 26, (1883).
5. Chapin, D. M., Fuller, C. S. & Pearson, G. L. A New Silicon p‐n Junction Photocell for
Converting Solar Radiation into Electrical Power. J. Appl. Phys. 25, 676–677 (1954).
6. Green, M. A., Emery, K., Hishikawa, Y., Warta, W. & Dunlop, E. D. Solar Cell Efficiency
Tables (Version 45). Prog. Photovolt. Res. Appl. 23, 1–9 (2015).
7. European Photovoltaic Industry Association. Global Market Outlook For Photovoltaics
2014-2018. (2014).
8. National Renewable Energy Laboratory (NREL). Best Research-Cell Efficiencies. at
<http://www.nrel.gov/ncpv/images/efficiency_chart.jpg>
9. Jean, J., Brown, P. R., Jaffe, R. L., Buonassisi, T. & Bulovic, V. Pathways for Solar
Photovoltaics. Energy Environ. Sci. 8, 1200–1219 (2015).
10. Liu, H. et al. Electron Acceptor Materials Engineering in Colloidal Quantum Dot Solar
Cells. Adv. Mater. 23, 3832–3837 (2011).
11. Zhitomirsky, D., Voznyy, O., Hoogland, S. & Sargent, E. H. Measuring Charge Carrier
Diffusion in Coupled Colloidal Quantum Dot Solids. ACS Nano 7, 5282–5290 (2013).
12. Nelson, J. in The Physics of Solar Cells 1–2 (Imperial College Place).
13. Shockley, W. & Queisser, H. J. Detailed Balance Limit of Efficiency of p‐n Junction Solar
Cells. J. Appl. Phys. 32, 510–519 (1961).
89
14. Henry, C. H. Limiting Efficiencies of Ideal Single and Multiple Energy Gap Terrestrial Solar
Cells. J. Appl. Phys. 51, 4494–4500 (1980).
15. Murray, C. B., Kagan, C. R. & Bawendi, M. G. Synthesis and Characterization of
Monodisperse Nanocrystals and Close-Packed Nanocrystal Assemblies. Annu. Rev. Mater.
Sci. 30, 545–610 (2000).
16. LaMer, V. K. & Dinegar, R. H. Theory, Production and Mechanism of Formation of
Monodispersed Hydrosols. J. Am. Chem. Soc. 72, 4847–4854 (1950).
17. Hines, M. A. & Scholes, G. D. Colloidal PbS Nanocrystals with Size-Tunable Near-Infrared
Emission: Observation of Post-Synthesis Self-Narrowing of the Particle Size Distribution.
Adv. Mater. 15, 1844–1849 (2003).
18. Debnath, R., Bakr, O. & Sargent, E. H. Solution-Processed Colloidal Quantum Dot
Photovoltaics: A Perspective. Energy Environ. Sci. 4, 4870–4881 (2011).
19. Klem, E. J. D., MacNeil, D. D., Cyr, P. W., Levina, L. & Sargent, E. H. Efficient Solution-
Processed Infrared Photovoltaic Cells: Planarized All-Inorganic Bulk Heterojunction
Devices via Inter-Quantum-Dot Bridging During Growth from Solution. Appl. Phys. Lett. 90,
183113 (2007).
20. Clifford, J. P., Johnston, K. W., Levina, L. & Sargent, E. H. Schottky Barriers to Colloidal
Quantum Dot Films. Appl. Phys. Lett. 91, 253117 (2007).
21. Johnston, K. W. et al. Schottky-Quantum Dot Photovoltaics for Efficient Infrared Power
Conversion. Appl. Phys. Lett. 92, 151115 (2008).
22. Ma, W. et al. Photovoltaic Performance of Ultrasmall PbSe Quantum Dots. ACS Nano 5,
8140–8147 (2011).
90
23. O’Regan, B. & Graetzel, M. A Low-Cost, High-Efficiency Solar Cell Based on Dye-
Sensitized Colloidal TiO2 Films. Nature 353, 737–740 (1991).
24. Pattantyus-Abraham, A. G. et al. Depleted-Heterojunction Colloidal Quantum Dot Solar
Cells. ACS Nano 4, 3374–3380 (2010).
25. Tang, J. et al. Colloidal-Quantum-Dot Photovoltaics Using Atomic-Ligand Passivation. Nat.
Mater. 10, 765–771 (2011).
26. Tang, J. et al. Quantum Junction Solar Cells. Nano Lett. 12, 4889–4894 (2012).
27. Voznyy, O. et al. A Charge-Orbital Balance Picture of Doping in Colloidal Quantum Dot
Solids. ACS Nano 6, 8448–8455 (2012).
28. Zhitomirsky, D. et al. N-Type Colloidal-Quantum-Dot Solids for Photovoltaics. Adv. Mater.
24, 6181–6185 (2012).
29. Liu, H. et al. Systematic Optimization of Quantum Junction Colloidal Quantum Dot Solar
Cells. Appl. Phys. Lett. 101, 151112 (2012).
30. Ning, Z. et al. Graded Doping for Enhanced Colloidal Quantum Dot Photovoltaics. Adv.
Mater. 25, 1719–1723 (2013).
31. Ning, Z. et al. Air-Stable N-Type Colloidal Quantum Dot Solids. Nat. Mater. 13, 822–828
(2014).
32. Ip, A. H. et al. Hybrid Passivated Colloidal Quantum Dot Solids. Nat. Nano. 7, 577–582
(2012).
33. Zhitomirsky, D. et al. Engineering Colloidal Quantum Dot Solids Within and Beyond the
Mobility-Invariant Regime. Nat. Commun. 5, 3803 (2014).
34. Beyer, B., Pfeifer, R., Zettler, J. K., Hild, O. R. & Leo, K. Graded Absorption Layers in Bulk
Heterojunction Organic Solar Cells. J. Phys. Chem. C 117, 9537–9542 (2013).
91
35. Scharber, M. C. & Sariciftci, N. S. Efficiency of Bulk-Heterojunction Organic Solar Cells.
Top. Issue Conduct. Polym. 38, 1929–1940 (2013).
36. Huang, Y., Kramer, E. J., Heeger, A. J. & Bazan, G. C. Bulk Heterojunction Solar Cells:
Morphology and Performance Relationships. Chem. Rev. 114, 7006–7043 (2014).
37. Barkhouse, D. A. R. et al. Depleted Bulk Heterojunction Colloidal Quantum Dot
Photovoltaics. Adv. Mater. 23, 3134–3138 (2011).
38. Rath, A. K. et al. Solution-Processed Inorganic Bulk Nano-Heterojunctions and Their
Application to Solar Cells. Nat. Photon. 6, 529–534 (2012).
39. Jean, J. et al. ZnO Nanowire Arrays for Enhanced Photocurrent in PbS Quantum Dot Solar
Cells. Adv. Mater. 25, 2790–2796 (2013).
40. Lan, X. et al. Self-Assembled, Nanowire Network Electrodes for Depleted Bulk
Heterojunction Solar Cells. Adv. Mater. 25, 1769–1773 (2013).
41. Kramer, I. J. et al. Ordered Nanopillar Structured Electrodes for Depleted Bulk
Heterojunction Colloidal Quantum Dot Solar Cells. Adv. Mater. 24, 2315–2319 (2012).
42. Kemp, K. W. et al. Interface Recombination in Depleted Heterojunction Photovoltaics based
on Colloidal Quantum Dots. Adv. Energy Mater. 3, 917–922 (2013).
43. Widmer, J., Tietze, M., Leo, K. & Riede, M. Open-Circuit Voltage and Effective Gap of
Organic Solar Cells. Adv. Funct. Mater. 23, 5814-5821 (2013).
44. Koleilat, G. I. et al. Folded-Light-Path Colloidal Quantum Dot Solar Cells. Sci. Rep. 3, 2166
(2013).
45. Feng Lu, H. et al. Plasmonic Quantum Dot Solar Cells for Enhanced Infrared Response.
Appl. Phys. Lett. 100, 103505 (2012).
92
46. Paz-Soldan, D. et al. Jointly Tuned Plasmonic–Excitonic Photovoltaics Using Nanoshells.
Nano Lett. 13, 1502–1508 (2013).
47. Battaglia, C. et al. Light Trapping in Solar Cells: Can Periodic Beat Random? ACS Nano 6,
2790–2797 (2012).
48. Zhu, J., Yu, Z., Fan, S. & Cui, Y. Nanostructured Photon Management for High Performance
Solar Cells. 3rd IEEE Int. Nanoelectron. Conf. INEC 70, 330–340 (2010).
49. Narasimhan, V. K. & Cui, Y. Nanostructures for Photon Management in Solar Cells.
Nanophotonics 2, 187–210 (2013).
50. Soleymani, L. et al. Hierarchical Nanotextured Microelectrodes Overcome the Molecular
Transport Barrier To Achieve Rapid, Direct Bacterial Detection. ACS Nano 5, 3360–3366
(2011).
51. Vasilyeva, E. et al. Direct Genetic Analysis of Ten Cancer Cells: Tuning Sensor Structure
and Molecular Probe Design for Efficient mRNA Capture. Angew. Chem. Int. Ed. 50, 4137–
4141 (2011).
52. Ivanov, I. et al. Chip-Based Nanostructured Sensors Enable Accurate Identification and
Classification of Circulating Tumor Cells in Prostate Cancer Patient Blood Samples. Anal.
Chem. 85, 398–403 (2013).
53. Lam, B. et al. Solution-Based Circuits Enable Rapid and Multiplexed Pathogen Detection.
Nat. Commun. 4, (2013).
54. Zhu, J., Hsu, C.-M., Yu, Z., Fan, S. & Cui, Y. Nanodome Solar Cells with Efficient Light
Management and Self-Cleaning. Nano Lett. 10, 1979–1984 (2010).
55. Ferry, V. E. et al. Light Trapping in Ultrathin Plasmonic Solar Cells. Opt. Express 18,
A237–A245 (2010).
93
56. Leung, S.-F. et al. Efficient Photon Capturing with Ordered Three-Dimensional Nanowell
Arrays. Nano Lett. 12, 3682–3689 (2012).
57. Pala, R. A., White, J., Barnard, E., Liu, J. & Brongersma, M. L. Design of Plasmonic Thin-
Film Solar Cells with Broadband Absorption Enhancements. Adv. Mater. 21, 3504–3509
(2009).
58. Min, C. et al. Enhancement of Optical Absorption in Thin-Film Organic Solar Cells Through
the Excitation of Plasmonic Modes in Metallic Gratings. Appl. Phys. Lett. 96, 133302
(2010).
59. Sheng, P., Bloch, A. N. & Stepleman, R. S. Wavelength‐Selective Absorption Enhancement
in Thin‐Film Solar Cells. Appl. Phys. Lett. 43, 579–581 (1983).
60. Heine, C. & Morf, R. H. Submicrometer Gratings for Solar Energy Applications. Appl. Opt.
34, 2476–2482 (1995).
61. Munday, J. N. & Atwater, H. A. Large Integrated Absorption Enhancement in Plasmonic
Solar Cells by Combining Metallic Gratings and Antireflection Coatings. Nano Lett. 11,
2195–2201 (2011).
62. Pillai, S., Catchpole, K. R., Trupke, T. & Green, M. A. Surface Plasmon Enhanced Silicon
Solar Cells. J. Appl. Phys. 101, 093105 (2007).
63. Zeng, L. et al. Efficiency Enhancement in Si Solar Cells by Textured Photonic Crystal Back
Reflector. Appl. Phys. Lett. 89, 111111 (2006).
64. Ning-Ning Feng et al. Design of Highly Efficient Light-Trapping Structures for Thin-Film
Crystalline Silicon Solar Cells. Electron Devices IEEE Trans. On 54, 1926–1933 (2007).
94
65. Bermel, P., Luo, C., Zeng, L., Kimerling, L. C. & Joannopoulos, J. D. Improving Thin-Film
Crystalline Silicon Solar Cell Efficiencies with Photonic Crystals. Opt. Express 15, 16986–
17000 (2007).
66. Grandidier, J., Callahan, D. M., Munday, J. N. & Atwater, H. A. Light Absorption
Enhancement in Thin-Film Solar Cells Using Whispering Gallery Modes in Dielectric
Nanospheres. Adv. Mater. 23, 1272–1276 (2011).
67. Yao, Y. et al. Broadband Light Management Using Low-Q Whispering Gallery Modes in
Spherical Nanoshells. Nat. Commun. 3, 664 (2012).
68. Yu, Y., Ferry, V. E., Alivisatos, A. P. & Cao, L. Dielectric Core–Shell Optical Antennas for
Strong Solar Absorption Enhancement. Nano Lett. 12, 3674–3681 (2012).
69. Adachi, M. M. et al. Broadband Solar Absorption Enhancement via Periodic
Nanostructuring of Electrodes. Sci. Rep. 3, 2928 (2013).
70. Atwater, H. A. & Polman, A. Plasmonics for Improved Photovoltaic Devices. Nat. Mater. 9,
205–213 (2010).
71. Blech, I. A. & Vander Plas, H. A. Step Coverage Simulation and Measurement in a DC
Planar Magnetron Sputtering System. J. Appl. Phys. 54, 3489–3496 (1983).
72. Labelle, A. J. et al. Conformal Fabrication of Colloidal Quantum Dot Solids for Optically
Enhanced Photovoltaics. ACS Nano 9, 5447-5453 (2015).
73. Wang, X. et al. Tandem Colloidal Quantum Dot Solar Cells Employing a Graded
Recombination Layer. Nat. Photon. 5, 480–484 (2011).
74. Bico, J., Thiele, U. & Quéré, D. Wetting of Textured Surfaces. Colloids Surf. Physicochem.
Eng. Asp. 206, 41–46 (2002).
95
75. Labelle, A. J. et al. Colloidal Quantum Dot Solar Cells Exploiting Hierarchical Structuring.
Nano Lett. 15, 1101–1108 (2014).
76. Han, Y., Yu, X., Wang, D. & Yang, D. Formation of Various Pyramidal Structures on
Monocrystalline Silicon Surface and their Influence on the Solar Cells. J Nanomater. 2013,
7–7 (2013).
77. Llopis, F. & Tobías, I. Influence of Texture Feature Size on the Optical Performance of
Silicon Solar Cells. Prog. Photovolt. Res. Appl. 13, 27–36 (2005).
78. Campbell, P. & Green, M. A. Light Trapping Properties of Pyramidally Textured Surfaces. J.
Appl. Phys. 62, 243–249 (1987).
79. Zhao, J., Wang, A., Altermatt, P. P., Wenham, S. R. & Green, M. A. 24% Efficient PERL
Silicon Solar Cell: Recent Improvements in High Efficiency Silicon Cell Research. Sol.
Energy Mater. Sol. Cells 41–42, 87–99 (1996).
80. Campbell, P. & Green, M. A. High Performance Light Trapping Textures for
Monocrystalline Silicon Solar Cells. PVSEC 11 Part I 65, 369–375 (2001).
81. Xia, Y. & Whitesides, G. M. Soft Lithography. Angew. Chem. Int. Ed. 37, 550–575 (1998).
96
Appendices
A1 CQD Solar Cell Fabrication
All spin-coated samples (structured substrate and planar controls) were prepared using a layer-
by-layer spin-cast deposition method. PbS quantum dots (synthesized and exchanged following
previously published protocols32) were deposited at a concentration of 50 mg/mL in octane
through a 0.2 µm filter, followed by solid state exchange using 3-mercaptopropionic acid at a
concentration of 1% v./v. in methanol. Finally, the film was rinsed twice with pure methanol. All
samples were spin-cast at 2500 rpm for a total of up to 12 cycles. All dip-coated devices
(structured substrate and planar controls) were prepared using sequential dip-coating. Samples
were manipulated using a medium-sized KSV NIMA multi-vessel dip-coater. The samples with
MPA primer treatment were first dipped into 30 mL beakers containing 15 mL of 0.05% v./v.
3-mercaptopropionic acid (MPA) in methanol for 5 s, then left to dry for 360 s (surface primer
treatment). Samples were then dipped into an adjacent 30 mL beaker containing 15 mL of 7.5
mg/mL PbS quantum dots (same as used for spin-casting) in hexane for 30 s, and left to dry for
240 s. Samples were then dipped into a 50 mL beaker containing 25 mL of 0.2% v./v. MPA in
methanol solution (solid-state exchange) for 3 s followed by 60 s drying. And finally the samples
were rinsed in a 50 mL beaker containing 25 mL of pure methanol for 5 s and dried for 120 s.
This process was repeated for 10-18 cycles for optimal results (number of cycles being
dependent on room humidity). Note that samples without MPA primer treatment were fabricated
using the same sequence, minus the first step with 0.05% MPA.
A2 Optical and Device Modelling
One-dimensional device simulations were performed using SCAPS 3.0.01 software. The device
model was adapted from previously published work33.
Finite-difference time-domain simulations were carried out using Lumerical FDTD Solutions
software (http://www.Lumerical.com) version 8.5 (Chapter 4) and 8.7.4 (Chapter 5, 6). Details
provided for each chapter below:
97
Chapter 4. All simulations were for a hexagonal array of three-dimensional structures
(periodicity=1 µm) with periodic boundary conditions in the x and y directions. A broadband
(=400-1200 nm) planewave source polarized along the y-axis was incident from within the
glass region. The absorption in each material was calculated by integrating the absorption only of
matching refractive index of a particular material.
Chapter 5. All simulations were carried out in 3D for pyramids, parabola, cones or planar
structures with periodic boundary conditions in the x and y directions and a unit size of 2000 x
2000 nm2. A broadband (λ = 400–1200 nm) planewave source polarized along the y-axis was
incident from within the glass region. The absorption was isolated for the individual layers and
only the absorption of the CQD layer is presented in Figure 6-1a, and projected short-circuit
current density was calculated in Figure 6-1b assuming an IQE of 100% and integrating against
the AM1.5 solar spectrum.
Chapter 6. Two dimensional simulations were for pyramid (with periodicity, size, and side-wall
angle as defined in the text) or planar structures, at the mid-line along the z-axis, with periodic
boundary conditions in the x direction. Three dimensional simulations were for pyramid (with
periodicity, size, and side-wall angle as defined in the text) or planar structures with periodic
boundary conditions in the x and y directions. A broadband (λ = 400–1200 nm) planewave
source polarized along the y-axis was incident from within the glass region. The absorption was
isolated for the individual layers and only the absorption of the CQD layer is presented in Figure
7-3, while Figure 7-10 represents the total absorption of all active layers in the device. Note that
photogenerated current density (Jph) presented in Figure 7-6 was computed assuming an internal
quantum efficiency of 100% and integrating the predicted absorption curve against the AM1.5
solar irradiance spectrum.
A3 Electrode Preparation
Silicon master preparation. The silicon (Si) master was prepared from a 4” diameter <100>-
oriented silicon wafer coated with 100 nm of PECVD-deposited Si3N4. A grid pattern (with 10
µm pitch) was transferred to the Si3N4 layer via a positive photolithography process. The wafer
was then heated at 115 °C for 1 minute to evaporate residual surface moisture. An adhesion-
98
promoter, hexamethyldisilazane (HMDS), was spin-cast onto the wafer at 4000 rpm (2500 rpm/s
acceleration) for 40 s to promote adhesion between the photoresist and the wafer. S1811
photoresist was then spin-cast onto the substrate at 4000 rpm (2500 rpm/s acceleration) for 90 s.
The wafer was then baked at 115 °C for 1 minute and exposed to 930W UV radiation for 7.1 s in
hard contact mode. The photoresist contained within the 9 µm x 9 µm squares (between grid-
lines) was developed in MF321 for 28 s – the remaining photoresist covering the 1 µm-wide grid
lines was hard-baked at 115°C for 20 min. The Si3N4 was then removed from the 9 µm x 9 µm
square-regions (between grid-lines) by reactive ion etching (RIE) in CF4/CHF3 (CF4/CHF3: 20
sccm; pressure: 16 mTorr; power: 90 W) for 5 min. The remaining photoresist was removed with
a 10 minute exposure to acetone. The silicon wafer, with Si3N4 hard mask was then dipped in
20% KOH in isopropanol at 70 °C, etching the silicon at a rate of 1 µm/minute. The Si3N4 hard-
mask was then removed with buffered oxide etch (BOE), thus producing a silicon wafer with a
periodic array of inverted pyramid features with pitch of 10 µm, and individual pyramid width of
9 µm. Photograph of completed Si master is shown in Figure A1.
Figure A1. Silicon master – periodic inverted pyramids. a) Photograph of the silicon master used for preparing pyramid-patterned electrodes. b) Optical microscope images of the periodic inverted pyramid features on the silicon master. *make sure no space
Electrode preparation. Polydimethylsilane (PDMS) stamps were prepared using the silicon
wafer as a master. Prior to moulding the PDMS, the Si surface was treated using a silanization
process in which the wafer was treated with an O2 plasma-etch at 50 W for 5 min (Unitronics
PE-25-JW Plasma Cleaning System), then submerged in 30% H2O2 to functionalize the surface
for silanization exposure. The wafer was then loaded into a glovebox antechamber and exposed
to 100 µL of Trichloro-(perfluorooctyl)-silane under low vacuum. PDMS (mixed in a 10:1 ratio
99
of resin:curing agent) was poured over the Si master and heated on a hotplate for 20 min at 125
°C. The PDMS stamp was then used to mould a reciprocal PDMS stamp to be used for preparing
the pyramid-patterned electrodes. The silanization process was repeated for the original PDMS
mould, and 5 mL of PDMS (10:1 ratio) was poured over the original PDMS mould and treated in
an oven for 1 hour at 70 °C. The reciprocal PDMS stamp (with inverted pyramid features) was
then used to structure the TiO2 electrode. Diagram provided in Chapter 7.
TiO2 electrode stamping. 200 µL of viscous TiO2 solution (1:2 Dyesol DS90 TiO2
nanoparticles:ethanol) was spin-cast onto an FTO-coated glass substrate at 1000 rpm for 10 s.
The PDMS reciprocal stamp was then placed onto the substrate, applying no additional pressure.
The TiO2 paste was allowed to dry for a minimum of 8 hours and the stamp was removed,
leaving behind a pyramid-structured TiO2 film. The substrate was then heat-treated at
temperatures of 200 °C, 300 °C and 400 °C for 15 min each to burn off excess organics from the
film. The TiO2 electrode was then treated with titanium tetrachloride solution - the substrates
were submerged in a 120 mM solution of TiCl4 in deionized water and heated in an oven for 30
min at 70 °C. Substrates were then removed from the oven, rinsed with water, dried with
compressed N2 and heated to 400 °C on a hotplate for 1 hour. Diagram provided in Chapter 7.
A4 Contact Deposition
Bottom electrodes (TCO and n-type electrodes). Transparent conductive oxide bottom contacts
were deposited by magnetron sputtering (Angstrom Engineering Åmod deposition system in an
Innovative Technology glovebox) using a 3” diameter Indium-Tin-Oxide (In2O3/SnO2, 90/10
wt%) target. Substrates were heated to 380 C and rotated during deposition. TiO2 layers were
deposited by magnetron sputtering from a 3” diameter TiO2 (99.9% purity) target. The standard
film thickness was 50 nm. All TiO2 electrodes were treated with a 120 mM TiCl4 solution at 70
C for 30 min, followed by an anneal at 400-500 C for 60 min in air ambient. (Chapter 5)
Top electrodes. All photovoltaic cells reported (except in Chapter 4) were prepared with
MoO3/Au/Ag top contacts using standard 0.049 cm2 area pixels. Materials were deposited in an
Angstrom Engineering Amod deposition system inside an Innovative Technology glovebox via
thermal (MoO3 and Ag) and e-beam (Au) evaporation. Total thickness of each material was 30-
100
40 nm MoO3 (deposited at a rate of 1 Å/s) 55 nm Au (deposited at a rate of 1.5 Å/s), and 200 nm
Ag (deposited at a rate of 3 Å/s).
A5 Electrodeposition
Electrodeposition (Chapter 3) was carried out using an Ivium Technologies CompactStat unit,
designed as a portable electrochemical interface and impedance analyser, with power
configuration of ± 30 mA at ± 10 V. The procedure was carried out using a 4 probe/3 electrode
configuration in which the working electrode was connected to the gold-coated (Au) substrate;
the counter electrode was connected to a platinum-coated (Pt) titanium mesh electrode; and the
reference electrode was connected to a BASi Ag/AgCl reference electrode (RE-5B with flexible
wire connector). Solutions of HAuCl4 were prepared in the following concentrations in 0.5 M
HCl stock solution – 13 mM HAuCl4, 26 mM HAuCl4, and 40 mM HAuCl4 using a Gold (III)
chloride solution, 99.99% trace metals, 30 wt. % in dilute HCl solution (Sigma Aldrich). Gold
was electrodeposited using the ElectroAnalysis – Amperometric Detection feature; all samples
were prepared applying a bias of -250 mV with approximately 1.5 – 2 cm separation between Pt
counter electrode and Au working electrode. Electrodeposition was carried out between 180 and
600 s and measured currents varied from approximately - 2 mA to - 9 mA depending on HAuCl4
concentration. Figure A2 illustrates the electrodeposition setup and the basic premise of the
procedure.
101
Figure A2. Diagram of the electrodeposition process. Note that Au atoms in solution reduce onto the working electrode to create a nanostructured surface.
A6 Contact Angle
Contact angle measurements were carried out using a Kruss DSA100 Drop Shape Analysis
System using the static Sessile drop method. Planar substrates were prepared with TiO2 solution
(1:2 Dyesol DS90 TiO2 nanoparticles:ethanol) spin-coated onto an FTO-coated glass substrate at
1500 rpm for 10 s. Planar substrates were then heat-treated and TiCl4-treated according to the
same protocol used for pyramid-patterned substrates. A 4 µL drop of deionised water was
deposited on the substrate while in the DSA100 system. Corresponding software captured
magnified images of the droplet and calculated contact angle from a modelled circular shape of
the image.
A7 Scanning Electron Microscopy
Scanning electron microscopy images were obtained using an FEI Quanta FEG 250
environmental SEM/STEM (scanning transmission electron microscopy). All cross-sectional
SEM images were obtained using 90° holder, and all top-view images using a 45° tilted holder to
highlight surface texture. Images were obtained using 5, 10, and 20keV acceleration voltage
based on the application, and generally magnified to the order of 1-10 micron depending on the
application.
102
A8 Solar Simulator
All photovoltaic measurements were carried out under N2-flow. Current density – voltage curves
were measured using a Keithley 2400 source meter with illumination from a Sciencetech solar
simulator with an irradiance of 100 mW/cm2. The active area of the solar cell was illuminated
through a circular aperture with an area of 0.049 cm2. The power was measured using a Melles-
Griot broadband power meter. The spectral mismatch between measured and actual solar
spectrum was measured using a calibrated Silicon reference solar cell from Newport. The visible
and infrared components of the solar simulator were measured using a USB2000 and NIR512
spectrometers from Ocean Optics, respectively. The spectrum was then compared to the AM1.5
solar spectrum as depicted in Figure A3a. Taking into account the external quantum efficiency
(EQE) of the Si reference cell with respect to the EQE of a PbS CQD test cell (Figure A3b) we
can then calculate the M factor using equation A-1:
Equation (A-1): 𝑀 =∫ 𝐸𝑟(𝜆)𝑆𝑟(𝜆)𝑑𝜆𝜆2𝜆1
∫ 𝐸𝑟(𝜆)𝑆𝑡(𝜆)𝑑𝜆𝜆2𝜆1
∗∫ 𝐸𝑠(𝜆)𝑆𝑡(𝜆)𝑑𝜆𝜆2𝜆1
∫ 𝐸𝑠(𝜆)𝑆𝑟(𝜆)𝑑𝜆𝜆2𝜆1
where Er = reference spectral irradiance (AM1.5G spectrum); Sr = spectra responsivity of
reference silicon cell; Es = source (simulator) spectral irradiance; and St = spectral responsivity
of test cell (PbS cell). M was found to be 0.8816 in this case. We then correct for the mismatch
between the Si reference cell and the broadband power meter (CF) using equation A-2:
Equation (A-2): 𝐶𝐹 = 100 ∗ 𝐴𝑟𝑒𝑓 ∗𝑃𝑟𝑒𝑓
𝑃𝑚
where Aref = area of the Si reference cell; Pref = power measured in suns using Si reference cell;
Pm = power measured using broadband power meter in mW; and 100 [mW/cm2] is added to
convert power measured with Si reference cell into mW (1 sun = 100 mW/cm2). The final
multiplicative factor (Jsc factor) is represented in equation A-3.
Equation (A-3): 𝐽𝑠𝑐 =1
𝑀𝑥𝐶𝐹
103
For this case, the total spectral mismatch was found to be ~4% and was taken into account by
applying a multiplicative factor of 0.96 to measured current density values. The uncertainty of
AM1.5 measurements was estimated to be 7%.
Figure A3. Spectral correction for AM1.5 solar simulator. a) Comparison of the AM1.5 solar spectrum vs. the spectrum of the simulator lamp. These spectra are used in calculating the M factor of the solar simulator. b) EQE spectra of the Si reference cell and the CQD test cell. These spectra are also integrated for the calculation of the M factor.
A9 Spectrophotometer
All absorption measurements were done using a Perkin Elmer Lambda 950 UV-Vis-NIR
spectrophotometer equipped with an integrating sphere. Samples were placed at the centre of the
integrated sphere tilted at an angle of 20 relative to the incident beam, with the objective of
capturing specular reflection from the sample surface inside the integrating sphere. The total
transmission (T) and reflectance (R) was collected by the integrating sphere detector with all
ports closed except the one for the incident beam. Absorption was calculated as 100% - T - R.
The 100% transmission baseline measurement was obtained using an empty sphere.
A10 Quantum Efficiency
External quantum efficiency measurements. All EQE measurements were carried out under N2
flow. External quantum efficiency spectra were measured under monochromatic light (400 W
104
xenon lamp source passing through a monochromator with order-sorting filters) which was
chopped at 220 Hz. A constant 1-sun intensity white-light source simultaneously illuminated the
device during measurements. The monochromatic light power was measured using Newport 818-
UV and Newport 818-IR power meters. The current response was measured using a Stanford
Research Systems lock-in amplifier at short-circuit conditions. The uncertainty of the EQE
measurements, calculated from taking the root-mean-square of the error from all equipment was
3%.
Internal quantum efficiency. Internal quantum efficiency is calculated by dividing the spectral
EQE (measured according to the methods detailed above) by the spectral absorption of the CQD
film. While this calculation is relatively straight forward for a planar sample, isolating the
spectral absorption of the CQD film on a patterned substrate is somewhat more complicated. We
detail the procedures (as applied for IQE results presented in Chapter 5) for a pyramid-patterned
film below. The angles used for calculations are shown in Figure 3-4.
Figure A4. Diagram illustrating the general light path through a pyamid-patterned unit in a structured thin film solar cell.
105
To determine the full-pass absorption of CQD films, single pass absorption data was compiled
for the different active layers of the device and absorption was tracked as light scatters within the
structure, taking into account the remaining transmitted radiation with every interaction.
𝐹(𝜆) = 𝑠𝑝𝑒𝑐𝑡𝑟𝑎𝑙𝑎𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛𝑜𝑓𝐹𝑇𝑂
𝐶(𝜆) = 𝑠𝑖𝑛𝑔𝑙𝑒 − 𝑝𝑎𝑠𝑠𝑎𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛𝑜𝑓𝑃𝑏𝑆𝐶𝑄𝐷𝑓𝑖𝑙𝑚
𝑇(𝜆) = 𝑠𝑝𝑒𝑐𝑡𝑟𝑎𝑙𝑎𝑏𝑠𝑜𝑟𝑝𝑡𝑖𝑜𝑛𝑜𝑓𝑡𝑜𝑝𝑐𝑜𝑛𝑡𝑎𝑐𝑡(𝑀𝑜𝑂3 − 𝐴𝑢 − 𝐴𝑔)
𝐹1(𝜆) = 𝐹(𝜆)
𝐶1(𝜆) = (100 − 𝐹1(𝜆)) ∗ 𝐶(𝜆)
𝑇1(𝜆) = (100 − 𝐹1(𝜆) − 𝐶1(𝜆)) ∗ 𝑇(𝜆)
𝐶2(𝜆) = (100 − 𝐹1(𝜆) − 𝐶1(𝜆) − 𝑇1(𝜆)) ∗ 𝐶(𝜆)
𝐶3(𝜆) = (100 − 𝐹1(𝜆) − 𝐶1(𝜆) − 𝑇1(𝜆) − 𝐶2(𝜆)) ∗ 𝐶(𝜆)𝑃𝐹
𝑇2(𝜆) = (100 − 𝐹1(𝜆) − 𝐶1(𝜆) − 𝑇1(𝜆) − 𝐶2(𝜆) − 𝐶3(𝜆)) ∗ 𝑇(𝜆)
𝐶4(𝜆) = (100 − 𝐹1(𝜆) − 𝐶1(𝜆) − 𝑇1(𝜆) − 𝐶2(𝜆) − 𝐶3(𝜆) − 𝑇2(𝜆)) ∗ 𝐶(𝜆)𝑃𝐹
𝐹2(𝜆) = (100 − 𝐹1(𝜆) − 𝐶1(𝜆) − 𝑇1(𝜆) − 𝐶2(𝜆) − 𝐶3(𝜆) − 𝑇2(𝜆) − 𝐶4(𝜆)) ∗ 𝐹(𝜆)
Where PF = path factor, to account for the fact that the path length of light will be different for
the third and fourth pass through the film.
The respective path lengths will be:
𝑡𝑐𝑜𝑠𝜃⁄ for 1st and 2nd pass, and
𝑡𝑐𝑜𝑠𝛼⁄ for 3rd and 4th pass
106
where α = π - 3θ
In the case of θ = 54.7° then α = 15.9°. The relative path length then becomes:
𝑡/𝑐𝑜𝑠(15.9°)/𝑡/𝑐𝑜𝑠(54.7°) = 𝑐𝑜𝑠(54.7°)/𝑐𝑜𝑠(15.9°) = 0.6
Since absorption is proportional to e-αx, then we conclude that the effective absorption for the 3rd
and 4th pass will be A’ = e-α(0.6x) = e-αx(0.6) = A0.6 (the single pass absorption for 54.7°-incidence to
the power of 0.6), therefore PF = 0.6.
The total absorption of the CQD film in a full-pass scenario is therefore:
𝐶𝑄𝐷(𝜆) = 𝐶1(𝜆) + 𝐶2(𝜆) + 𝐶3(𝜆) + 𝐶4(𝜆)
To verify our calculations we summed all Ci, Fj, and Tk absorption components and compared to
the empirically measured full-pass absorption of the pyramid-patterned device – there is
excellent agreement between the 2 curves as depicted in Figure 3-5.
Figure A5. Comparison of calculated (using the equations presented above) and measured absorption of a CQD solar cell with reflective top contact. The agreement between the two spectra validates the mathematical approach used in deconvolving the absorption of the CQD film from the total absorption of the device.
