-
i
Nanoscale Interfaces in Colloidal Quantum Dot Solar
Cells: Physical Insights and Materials Engineering
Strategies
by
Kyle Wayne Kemp
A thesis submitted in conformity with the requirements
for the degree of Doctor of Philosophy
Department of Electrical and Computer Engineering
University of Toronto
© Copyrighted by Kyle Wayne Kemp 2014
-
ii
Nanoscale Interfaces in Colloidal Quantum Dot Solar
Cells: Physical Insights and Materials Engineering
Strategies
Kyle Kemp
Doctor of Philosophy
Department of Electrical and Computer Engineering,
University of Toronto
2014
Abstract
With growing global energy demand there will be an increased
need for sources of renewable
energy such as solar cells. To make these photovoltaic
technologies more competitive with
conventional energy sources such as coal and natural gas
requires further reduction in
manufacturing costs that can be realized by solution processing
and roll-to-roll printing.
Colloidal quantum dots are a bandgap tunable, solution
processible, semiconductor material
which may offer a path forward to efficient, inexpensive
photovoltaics. Despite impressive
progress in performance with these materials, there remain
limitations in photocarrier collection
that must be overcome.
This dissertation focuses on the characterization of charge
recombination and transport in
colloidal quantum dot photovoltaics, and the application of this
knowledge to the development of
new and better materials.
-
iii
Core-shell, PbS-CdS, quantum dots were investigated in an
attempt to achieve better surface
passivation and reduce electronic defects which can limit
performance. Optimization of this
material led to improved open circuit voltage, exceeding 0.6 V
for the first time, and record
published performance of 6% efficiency.
Using temperature-dependent and transient photovoltage
measurements we explored the
significance of interface recombination on the operation of
these devices. Careful engineering of
the electrode using atomic layer deposition of ZnO helped lead
to better TiO2 substrate materials
and allowed us to realize a nearly two-fold reduction in
recombination rate and an enhancement
upwards of 50 mV in open circuit voltage.
Carrier extraction efficiency was studied in these devices using
intensity dependent current-
voltage data of an operational solar cell. By developing an
analytical model to describe
recombination loss within the active layer of the device we were
able to accurately determine
transport lengths ranging up to 90 nm.
Transient absorption and photoconductivity techniques were used
to study charge dynamics by
identifying states in these quantum dot materials which
facilitate carrier transport. Thermal
activation energies for transport of 60 meV or lower were
measured for different PbS quantum
dot bandgaps, representing a relatively small barrier for
carrier transport. From these
measurements a dark, quantum confined energy level was
attributed to the electronic bandedge
of these materials which serves to govern their optoelectronic
behavior.
-
iv
Dedication
To Brian and Lori Kemp
Thanks for all the love and support throughout the years.
-
v
Acknowledgements
I would first like to express my sincere gratitude and
admiration for my supervisor, and mentor,
Professor Edward H. Sargent. I greatly appreciate his dedication
to research, his endless drive for
success, and his continued support of my professional career.
The past 4 years in the Sargent
group have been the most impactful of my life. I will take all
the lessons accrued here with me in
my future endeavors
I would like to thank all members of the Sargent group, my
coworkers and friends, both past and
present for both their support in and out of the lab: Andre
Labelle, Alex Ip, Susanna Thon, Lisa
Rollny, David Zhitomirsky, Jeffrey McDowall, Daniel, Armin
Fischer, Illan Kramer, Brandon
Sutherland, Chris Wong, Michael Adachi, Zhijung Ning, Damir
Jamakosmanovic, Leyla
Soleymani, Ratan Debnath, Xihua Wang, Andras Pattantyus-Abraham,
Aaron Barkhouse, Ghada
Koleilat, Vlad Sukhovatkin, Jin Young Kim, Mingjian Yuan, Silvia
Masala, Fengjia Fan, Osman
Ahmed, Andrei Buin, Haopeng Dong, Gabriel Moreno-Bautista, and
Huan Liu. I would
specifically like to thank Larissa Levina for her advice and
synthesis of quantum dots; Elenita
Palmiano, Remigiusz Wolowiec, and Damir Kopilovic for keeping
the lab from falling into
complete disarray ; Sjoerd Hoogland for this advice,
enlightening discussions, and help with
experimental design; Oleksandry Voznyy for his valued insight
and DFT modelling; and
Lukasz Brzozowksi for his great leadership and continued
guidance throughout the years.
Additionally to all my other friends who have made life in
Toronto so enjoyable even in the most
stressful of days, I would like to thank: Julie Ciurria, Davis
Holmes, Jeff Grant, Stephanie
Fisher, Malgosia Ip, Jason Godfrey, Edmund Lee, Elise Andrey,
Andrew Achkar, Yonni
Friedlander, and Karen Yu.
I want to extend my appreciation to all my collaborators over
the years; Professor John Asbury
and his group for our work pump-probe spectroscopy, Professor
Aram Amassian and his group
for their help with TEM and GISAXS; and Dr. Neil Coombs and Ilya
Gourevich for SEM and
TEM assistance.
-
vi
I would like to thank my brothers Devin and Travis and my sister
Ashlan. Even though we didn't
always get along, I cannot imagine growing up without them. I
know that I will always have
their love and support, as they will always have mine.
Finally, I would like to thank my parents Brian and Lori Kemp
for always supporting me in my
endeavors and for always believing that I could succeed in
anything I put my mind to, no matter
how difficult.
-
vii
Statement of Personal Contributions and
Collaborations
Dr. Larissa Levina synthesized the PbS quantum dots used in the
studies reported herein. The
Cd-oleate treatments were performed by either Dr. Levina or
myself. Purification and isolation
of the quantum dots were performed by our research group
technical specialist Elenita Palmiano
or myself. I performed all ligand exchanges along with all FTIR
and photoluminescence
characterization. TEM imaging of the ligand exchanged samples
was carried out in the Centre
for Nanostructure Imaging at the University of Toronto with
assistance from Dr. Neal Coombs
and Ilya Gourevich.
I prepared all GISAXS samples which were then measured by Kang
Wei Chou at the Advanced
Light Source at Lawrence Berkeley National Laboratories. Device
fabrication for independent
certification by Newport was conducted by Dr. Jiang Tang and
myself. The best samples
between the two of us were shipped for certification
measurement. Energy filtered TEM
elemental mapping were conducted by microscopy specialists at
the King Abdullah University of
Science and Technology on TEM samples prepared by me. I
fabricated the photodiodes and
measured the frequency response and EQE data.
Unless otherwise specified I carried out fabrication of all
devices reported herein and completed
all device I-V testing under AM1.5G conditions. Monochromatic
I-V measurements and voltage
biased EQE measurements were carried out by me. I also conducted
all fitting of the
experimental data and derived the final model.
Atomic layer deposition and all substrate post treatments were
performed by me. TiO2 substrates
were fabricated by the entire group on a rotation schedule.
Transient photovoltage measurements
were performed by me based on procedures developed by Dr. Sjoerd
Hoogland.
The TRIR and ultrafast measurements were conducted by members of
the Asbury group at Penn
State University. At the onset of this study I visited the Penn
State to demonstrate thin film
deposition techniques and assist in initial measurements. I was
responsible for interpretation of
all experimental data and developed the main conclusions based
on these measurements. I also
-
viii
measured the temperature dependent photoconductivity and Stokes
Shift for all quantum dot
samples. Density functional calculations were performed by Dr.
Oleksandr Voznyy based on
discussions with myself and Professor John Asbury.
-
ix
Table of Contents Abstract
...........................................................................................................................................
ii
Dedication
......................................................................................................................................
iv
Acknowledgements
.........................................................................................................................
v
Statement of Personal Contributions and Collaborations
.............................................................
vii
List of Tables
.................................................................................................................................
xi
List of Figures
................................................................................................................................
xi
Chapter 1 Introduction and
Motivation...........................................................................................
1
1.1 A Brief History of Photovoltaics
........................................................................................................
1
1.2 Current Outlook for Solar Energy
.......................................................................................................
2
1.3 Solution Processed Photovoltaics
.......................................................................................................
3
1.3 Thesis Objectives
................................................................................................................................
5
1.4 Thesis Outline
.....................................................................................................................................
6
Chapter 2 Photovoltaics and Colloidal Quantum Dots
...................................................................
8
2.1 Photovoltaics Background
..................................................................................................................
8
2.1.1 Solar Spectrum
.............................................................................................................................
8
2.1.2 Theory and Operation of Photovoltaics
.......................................................................................
9
2.2 Colloidal Quantum dots
....................................................................................................................
13
2.2.1 Synthesis and Optical Properties
................................................................................................
13
2.2.2 PbS CQD Optoelectronics
.........................................................................................................
17
2.3 Conclusions
.......................................................................................................................................
23
Chapter 3 Early Studies: Core-Shell Quantum Dots
.....................................................................
24
3.1 Introduction
.......................................................................................................................................
24
3.2 Core-Shell Heterojunction Strategies
................................................................................................
25
3.3 PbS-CdS Core-Shell Synthesis and Characterization
.......................................................................
27
3.4 Pyridine Ligand Exchange
................................................................................................................
30
3.5 Mercaptopropionic Acid Exchange
..................................................................................................
32
3.6 Conclusion
........................................................................................................................................
35
Chapter 4 The Advent of Atomic Ligand Passivation
..................................................................
36
4.1 Introduction
.......................................................................................................................................
36
4.2 Motivation
.........................................................................................................................................
36
4.3 Cation Passivation
.............................................................................................................................
37
-
x
4.4 Anion Passivation
.............................................................................................................................
42
4.5 Conclusion
........................................................................................................................................
47
Chapter 5 Interface Recombination in Depleted Heterojunction CQD
Solar Cells ...................... 48
5.1 Introduction
.......................................................................................................................................
48
5.2 Interface Recombination
...................................................................................................................
48
5.4 Interface
Modification.......................................................................................................................
54
5.6 Conclusions
.......................................................................................................................................
61
Chapter 6 Direct investigations of photocurrent extraction
efficiency inside an operating device
.......................................................................................................................................................
62
6.1 Introduction
.......................................................................................................................................
62
6.2 Device J-V Characteristics
................................................................................................................
62
6.3 Photocurrent Loss Mechanisms
........................................................................................................
63
6.4 Collection Extraction Models
...........................................................................................................
66
6.5 Intensity Dependence of Diffusion Length
.......................................................................................
71
6.6 Conclusions
.......................................................................................................................................
72
Chapter 7 The fate of photocarriers in CQD solids
......................................................................
73
7.1 Introduction
.......................................................................................................................................
73
7.2 Ultrafast Photoinduced Absorption
...................................................................................................
73
7.3 Time Resolved Infrared Spectroscopy
.............................................................................................
76
7.4 Theoretical Origins of Stokes Shift
...................................................................................................
81
7.5 Conclusions
.......................................................................................................................................
84
Chapter 8 Conclusions and Future Work
......................................................................................
85
8.1 Thesis Findings and Contributions
...................................................................................................
85
8.2 Future Work
......................................................................................................................................
86
References
.....................................................................................................................................
88
Appendix
.......................................................................................................................................
95
Appendix A NMR Spectra
......................................................................................................................
95
Appendix B Temperature Dependent JSC
................................................................................................
97
Appendix C SCAPS Modelling Parameters
............................................................................................
98
Appendix D Publications
........................................................................................................................
99
-
xi
List of Tables
Table 4-1 Exciton shift and enhancement for Cd-TDPA-OLA
treatment of PbS Dots (N=11). Exciton
enhancement is determined from the ratio of the peak-to-valley
ratios defined in chapter 2.. ................... 39
Table 5-1 Device performance summary for modified TiO2
electrodes. ...................................................
57
Table 7-1 Comparison of activation energies (Ea) for TRIR and
Photoconductivity Decay. .................... 80
Table A1 Basic Material Properties for SCAPS Model .
...........................................................................
98
Table A2 PbS Trap States for SCAPS Model
............................................................................................
98
Table A3 TiO2 Trap States for SCAPS Model
...........................................................................................
98
Table A4 Interface Trap States for SCAPS Model.
...................................................................................
98
List of Figures
Figure 1-1: Predicted global energy consumption in less
developed (LDC) and more developed (MDC)
countries [10].
...............................................................................................................................................
2
Figure 1-2: Total installed global photovoltaic capacity.
[Adapted and reprinted with permission] [11]. .. 3
Figure 1-3: Timeline for best research solar cell efficiencies
for different architectures. [Reprinted with
permission] [13].
...........................................................................................................................................
4
Figure 1-4: Timeline for emerging PV technologies from Figure
1-3. .........................................................
5
Figure 2-2: I-V characteristics of a solar cell in the dark and
under illumination. Intersection of the shaded
rectangle and the I-V curve indicates the maximum power point.
[Reprinted with Permission][25]. ........ 10
Figure 2-3: Integrated photocurrent (right) as a function of
bandgap using the AM1.5 spectrum. ............ 11
Figure 2-4: Thermodynamic limit for diode saturation current
density (right). Theoretical limit for VOC
with (red) and without (blue) consideration of thermodynamic
losses. ......................................................
12
Figure 2-5: Ideal solar cell efficiencies as a function of
bandgap (Eg) at 300 K for 1 sun and 1000 sun
concentrations. [Reprinted with Permission][25].
......................................................................................
13
Figure 2-6: Schematics showing the nucleation and growth of
colloidal nanocrystals in which various
sizes of crystals can be isolated at different times. (b) Simple
synthetic apparatus showing the preparation
of CQDs. [Reprinted with Permission][27].
...............................................................................................
15
Figure 2-7: Size dependent absorption of PbS quantum dots.
[Reprinted with Permission][31]. .............. 16
Figure 2-9: a) A schematic representation of a PbS photodiode
device. b) Frequency dependence of
photocurrent at zero bias. [Adapted and reprinted with
Permission][40]. ..................................................
18
Figure 2-10 a) Schematic of the Schottky device architecture
consisted of PbS CQDs as active material
and Al as metal contact. The inset shows the electron micrograph
of PbS film after n-butylamine
-
xii
exchange. b) Energy band diagram showing the band bending takes
place at Al/PbS interface. Under
illumination, electrons and holes are swept away by the built-in
electric field in the depletion layer. c)
Current–voltage data of the device influenced by the variation
of the simulated solar illumination source.
The device shows AM1.5 PCE of 1.8%. d) EQE spectra for devices
using PbS ........................................ 20
Figure 2-11: a) The depleted heterojunction architecture showing
the various components of the device
(FTO/porous TiO2/PbS QD/Au) along with the band diagram close to
maximum VOC. EF,n and EF,p are the
electron and hole quasi-Fermi levels; Ec and Ev are the
conduction and valence band edges; Jp, PV and Jn,PV
are the hole and electron photocurrents (and are equal at
steady-state); Jp,fwd is the hole current in the
forward bias direction. The Fermi level is shown as a dashed
line. b) Apertured dark and illuminated
current density-Voltage (J-V) response of the photovoltaic
device c) EQE and absorption spectra of the
photovoltaic device. [Adapted and reprinted with
permission][46].
........................................................... 22
Figure 3-1: Schematic representation of the energy-level
alignment in different core/shell systems realized
in semiconductor NCs to date. [Reprinted with Permission][47].
..............................................................
25
Figure 3-2 TEM imaging of PbSe-CdSe core-shell nanocrystals.
Inset: Individual core-shell quantum dot.
[Reprinted with Permission][49].
................................................................................................................
26
Figure 3-3: Synthesis of core-shell PbS-CdS quantum dots by
cation exchange. The PbS cores are
synthesized and then purified. Formation of the core-shell dots
is accomplished by reaction with the Cd-
oleate complex.
...........................................................................................................................................
28
Figure 3-4: a) Absorption spectra for Cd-Oleate treated PbS
quantum dots. b) Photoluminescence spectra
for Cd-Oleate treated PbS quantum dots.
....................................................................................................
29
Figure 3-5: Ligand exchange of oleic acid with pyridine.
..........................................................................
30
Figure 3-6: a) Absorption measurements of PbS and pyridine
exchanged PbS-CdS. b) Fourier Transform
Infrared (FTIR) spectroscopy of pyridine exchanged PbS-CdS The
feature at 2360 cm-1 is CO2 absorption
in the baseline subtraction .
.........................................................................................................................
31
Figure 3-7: Ligand Exchange of Oleic Acid with Mercaptopropionic
acid (MPA) ................................... 33
Figure 3-8: Absorption spectra for PbS-CdS and MPA exchanged
PbS-CdS. ........................................... 33
Figure 3-9 TEM imaging of PbS-CdS and MPA exchanged PbS-CdS
nanocrystals (Scale Bar=100 nm). 34
Figure 4-1: Absorption spectra of PbS nanocrystals before and
after Cd-TDPA-OLA treatment. ............. 39
Figure 4-2: Initial device data for cation passivated
(CD-TDPA-OLA) PbS quantum dots. ...................... 40
Figure 4-3: Absorption spectra for PbS nanocrystals before and
after TDPA-OLA treatment. ................. 41
Figure 4-4: a) Device performance data for PbS, Cd-TDPA-OLA
treated PbS, and TDPA-OLA treated
PbS. b) Photoluminescence spectra for devices in a).
.................................................................................
42
Figure 4-5: a) Device data for certified device fabricated with
atomic ligand passivation. b) External
Quantum Efficiency (EQE) for certified device in a). [Reprinted
with permission][61]. ........................... 43
Figure 4-6: A) TEM image of atomic ligand passivated PbS
nanocrystals used for EF-TEM elemental
mapping. B) Bromine elemental mapping. C) Lead elemental
mapping. D) Sulfur elemental mapping. E)
Combined elemental mapping. [Reprinted with permission][61].
..............................................................
44
Figure 4-7: GISAXS spectra for a) MPA treated PbS nanocrystals
and b) CTAB treated PbS
nanocrystals. [Reprinted with permission][61].
..........................................................................................
45
Figure 5-1: a) Band diagram for PbS-TiO2 depleted quantum dot
heterojunction solar cells at open circuit
conditions. Interface recombination is depicted as being
assisted by interface trap levels. b) Temperature
dependent VOC measurements of devices for 1.33 eV and 1.51 eV
bandgap PbS nanocrystals. ................ 49
Figure 5-2: Size dependence of conduction and valence band edges
of PbS nanocrystals (Red).
[Reprinted with Permission][69].
................................................................................................................
51
-
xiii
Figure 5-3: Simulation results for VOC as a function of window
layer (TiO2) electron affinity and interface
trap density. Trap densities are evenly spaced on a logarithmic
scale. [Reprinted with permission][78]. . 53
Figure 5-4: a) Device architecture for CQD depleted
heterojunction solar cells with interfacial buffer
layer. b) Device J-V data for modified TiO2 electrodes with
different buffer layer materials. [Reprinted
with permission][78].
..................................................................................................................................
55
Figure 5-5: Performance comparisons of TiO2 and modified
ZnO-TiO2 electrodes: a) J-V curves; b) Static
VOC; c) Static JSC and d) Static efficiency values at the
maximum power point. [Reprinted with
permission][78].
..........................................................................................................................................
56
Figure 5-6: Dark J-V comparison between TiO2 and ZnO Modified
TiO2. [Reprinted with
permission][78].
..........................................................................................................................................
58
Figure 5-7: Summary of recombination analysis for TiO2 and
ZnO-TiO2 substrates. a) VOC as a function
of incident photo flux. b) Recombination rates. c) Injected
carrier densities. d) Recombination lifetimes
(τ). [Reprinted with permission][78].
..........................................................................................................
59
Figure 6-1: Current-voltage characteristics for a CQD solar cell
under simulated AM1.5 and in the dark.
....................................................................................................................................................................
63
Figure 6-2: Intensity dependent photocurrent density for a
hybrid passivated CQD solar cell. Lines of best
fit are included to demonstrate unity power law relationship.
[Reprinted with permission][88]. .............. 64
Figure 6-3: a) Spectrally resolved voltage biased EQE of a CQD
solar cell. b) Fractional loss (compared
to 0V JSC conditions) in spectral EQE with forward bias.
...........................................................................
66
Figure 6-4: Collection efficiency and analytical fit [Equation
(6.14)] for: a) Hybrid-passivated CQD solar
cell and b) Organically cross-linked organic CQD solar cell.
[Reprinted with permission][88]. ............... 67
Figure 6-5: Transport and Recombination processes for a) Gärtner
p-n junction model [Equation (6.7)]; b)
Hecht drift transport model [Equation (6.8)] and c) Proposed
drift-limited p-n junction model [Equation
(6.13)]. [Reprinted with permission][88].
..................................................................................................
68
Figure 6-6: Intensity dependence of diffusion length for hybrid
passivated CQD solar cells [Equation
(6.12)] [Reprinted with permission][88].
....................................................................................................
72
Figure 7-1: Schematic for a semiconductor system characterized
by photoinduced spectroscopy.
Photocarriers are pumped high into the band. They then relax
quickly to lower lying states (1Se) where
they can recombine back to the 1Sh ground state or undergo
photoinduced absorption from a probe
excitation.
....................................................................................................................................................
74
Figure 7-2: Ultrafast transient absorption kinetic traces for
PbS quantum dot materials treated with
different ligands. A pump energy of 1.55 eV and a probe energy
of 0.24 eV were used to excite the
samples. Normalized kinetic traces are included in the primary
panel with the actual transient data in the
inset.
............................................................................................................................................................
75
Figure 7-3: Normalized ultrafast transient absorption kinetic
traces for hybrid passivated PbS films as a
function of pump intensity. A pump energy of 1.55 eV and a probe
energy of 0.33 eV were used to excite
the samples. At 321 μJ cm-2 it is estimated that 1.84 excitons
are generated per quantum dot. ................. 76
Figure 7-4: a) Time Resolved Infrared Spectroscopy (TRIR)
spectra for different quantum dots of
different bandgap (1Sh-1Se). b) Comparison of TRIR peak
positions with calculated 1Se-1Pe energy
splitting and Stokes Shift for each bandgap material. Error bars
for the ΔStokes + 1Se-1Pe energies are
indicated to the right of each data point.
.....................................................................................................
77
Figure 7-5: New model of PbS quantum dot band structure which
includes the presence of a sub-gap dark
state. The TRIR signal is represented as the sum of both the
1Se-1Pe energy splitting and the Stokes Shift.
....................................................................................................................................................................
78
-
xiv
Figure 7-6: a) Arrhenius plots and activation energies for
photoconductivity decay rates as for different
bandgap quantum dots (Inset: Interdigitized electrode test
structure.). b) Arrhenius plots and activation
energies for TRIR decay rates for different bandgap quantum
dots. ..........................................................
79
Figure A-2: :NMR 1H spectra for MPA capped PbS-CdS core-shell
dots in deuterated dimethylsulfoxide.
Spectra was taken at 300.06 MHz on a Varian Mercury-300
Spectrometer ............................................... 96
Figure B-1: Temperature dependence JSC measurements of devices
for 1.33 eV and 1.51 eV bandgap
PbS nanocrystals.
........................................................................................................................................
97
-
xv
List of Abbreviations
AM1.5 Air Mass 1.5 Global
CQD Colloidal Quantum Dot
DFT Density Functional Theory
D* Specific Detectivity
Ea Activation Energy
EC Conduction Band Edge
EV Valence Band Edge
EQE External Quantum Efficiency
FF Fill Factor
FTIR Fourier Transform Infrared Spectoscopy
IQE Internal Quantum Efficiency
I-V Current-Voltage
JSC Short Circuit Density
MPA 3-Mercaptopropionic Acid
NMR Nuclear Magnetic Resonance Spectroscopy
Nt Trap State Density
PCE Percent Conversion Efficiency
TEM Transmission Electron Microscopy
TRIR Time Resolved Infrared Spectroscopy
VOC Open Circuit Voltage
-
1
Chapter 1 Introduction and Motivation
1.1 A Brief History of Photovoltaics
In 1883 an American inventor by the name of Charles Fritts
coated a purified sample of selenium
with a thin layer of gold [1]. The result was the first
demonstration of a working photovoltaic
device. Although the estimated power conversion efficiencies
were less than 1% it served as a
significant milestone in the development of a technology that is
now poised to make significant
contributions to our energy production in the near future.
The progression of photovoltaic technology proceeded relatively
slowly over the next 60 years.
Some notable advancements over this period include the
development of single crystal silicon
ingots by Jan Czochralski in 1918 [2], and the demonstration of
the photovoltaic response in
cadmium selenide in 1932 [3]. Perhaps most importantly of all
was a seminal paper of Albert
Einstein in 1905 on the photoelectric effect [4] which helped
established the quantum nature of
light.
Despite these major advancements to the field it wasn't until
the early 1940's in Bell Labs that the
technology of photovoltaics really started to make significant
gains. It was, after all, a Bell Labs
engineer by the name of Russell Ohl who, in 1941, discovered the
p-n junction [5] in a silicon
crystal. The discovery, some may argue, was the birth of the
modern solar cell. Bell Labs
invested heavily in research in the fields of materials science,
physics, chemistry, and electrical
engineering in order to use basic scientific research to develop
technologies [6]. It was this
scientific foundation which helped lead to discoveries in
semiconductor physics and crystal
growth which would be responsible for further advancement in
photovoltaics.
In 1954, soon after Russell Ohl's first discovery of a p-n
junction Bell researchers Gerald
Pearson, Daryl Chapin and Calvin Fuller used artificially doped
crystalline silicon to develop a
6% efficiency solar cell device [7]. Since then continued
research in semiconductor physics and
materials science has helped further photovoltaic technology to
the present day where
performance has advanced to 28.8% for single junctions and for
37.9% for multijunction devices
-
2
[8]. Along the way commercialization of different photovoltaic
technologies has led to the
founding of numerous companies including Suntech and First Solar
with 2012 revenues for the
photovoltaic industry reaching $25.5 billion [9].
1.2 Current Outlook for Solar Energy
Currently the threat of climate change, uncertainties in fossil
fuel markets, and the continued
growth in energy demand have made renewable energy sources,
specifically solar cells,
incredibly important. By some projections global energy demand
may increase by over 50% by
2040 from present day, much of this increase occurring in less
developed countries (LDC) [10]
(Figure 1-1). This is expected to put more strain on the
planet's already strained resources and
will make our efforts to curtail climate change even more
challenging.
Figure 1-1: Predicted global energy consumption in less
developed (LDC) and more developed (MDC) countries [10].
Fortunately the outlook for solar energy remains quite positive.
In 2012 installed global capacity
of solar energy surpassed 100 GW of capacity for the first time
with 31 GW installed in 2012
alone [11]. This has been part of an overall trend of
exponential growth over the past decade
(Figure 1-2).
-
3
Figure 1-2: Total installed global photovoltaic capacity.
[Adapted and reprinted with permission] [11].
As part of its Sunshot Initiative [12] the United States
Department of Energy has set a goal for
reduction in solar installation costs of 75% by 2020. This
reduction would ensure a price of solar
electricity of approximately $0.05 per kWh. At this level, it
will become cost-competitive with
conventional energy sources, most notably, fossil fuels such as
coal and natural gas. Continued
research and development of inexpensive photovoltaic
technologies will be necessary to achieve
these milestones and establish a more sustainable energy
future.
1.3 Solution Processed Photovoltaics
A look at the state of current research indicates that all of
the emerging technologies (Figure 1-3,
Figure 1-4) are compatible with solution processing fabrication
techniques. This list includes
Dye-Sensitized Solar Cells, CZTSSe, various Organic based
architectures, and Colloidal
Quantum Dot (CQD) photovoltaics.
-
4
Drop casting, spray coating, and ink-jet printing are all
scalable fabrication techniques allowing
for the possibility of roll-to-roll processing on light weight,
flexible substrates, at low
temperature. As economic viability is so crucial for further
growth of solar energy generation,
these low cost manufacturing methods offer advantages that do
not exist for other established
technologies.
Figure 1-3: Timeline for best research solar cell efficiencies
for different architectures. [Reprinted with permission] [13].
-
5
Figure 1-4: Timeline for emerging PV technologies from Figure
1-3.
CQD photovoltaics, in particular, have seen significant growth
in just a short period of time. In
the past 5 years performances have increased from 2.1% [14] to
7% [15]. Not only do CQDs
readily form ink-like dispersions in organic solvents, their
quantum size effects offer bandgap
tuning from the visible throughout the near infrared spectrum,
making them a very attractive
option for photovoltaic development.
1.3 Thesis Objectives
At the outset of this study colloidal quantum dot photovoltaic
performances were approaching
4% efficiencies, up from sub-percent levels just a few years
before. Despite this progress,
performances were still well below the 31% predicted by the
Shockley-Queisser limit [16] for a
single junction solar cell. As the field matures, further
enhancements to performance will rely
more heavily on deeper insights of basic material
properties.
-
6
This dissertation focuses on the development of new materials
for CQD photovoltaics by
leveraging insights obtained through characterization of
recombination and transport
processes.
This work focuses on the following research questions:
1) How can we use surface passivation to improve upon device
performance?
We know from previous reports how changes in surface passivation
through different ligand
treatments can have a significant effect on the electronic and
optical properties of CQD thin
films[17][18]. Despite these accomplishments very little work
has found its way into improving
photovoltaic device performance. We sought to develop new
surface passivation strategies to
improve device performance.
2) What are the primary loss mechanisms in PbS CQD films?
Our initial work on these materials was based on the premise
that photocarrier recombination
occurs through poor surface passivation. While this has allowed
us to optimize device
performance a deeper understanding is still required to pinpoint
the main contributors to loss in
these CQD photovoltaics. How does interface recombination at the
PbS-TiO2 heterojunction
affect device performance? How do these electronic defects in
the CQD absorber influence
carrier transport and the overall band structure of the
material? Can we understand the origins of
recombination on a molecular level?
1.4 Thesis Outline
Chapter 2 provides necessary background for both photovoltaics
and, more specifically, CQD
solar cells. We start by introducing the solar energy spectrum
and introduce fundamental
photovoltaic concepts to explain the origins of
Shockley-Queisser analysis. We next introduce
the concept of the CQD, its origins, and basic properties. A
brief review of the work done on
CQD optoelectronics by this group follows to establish the
existing prior art.
-
7
Chapter 3 explores our initial attempts to use core-shell
(PbS-CdS) quantum dots to passivate
surface defects and reduce carrier recombination. Initial
characterization of the synthesis is
reported. Attempts to complete solution-phase ligand exchanges
with were conducted to make
easily processible CQD materials with high nanocrystal packing
densities.
Chapter 4 presents the development of atomic ligand passivation
for PbS quantum dots. In this
chapter we use a post-synthesis cadmium cation treatment to
passivate exposed sulfur sites on
the PbS nanocrystal surface. Our devices show a significant
enhancement in open circuit voltage.
We combined this technique with a halide anion treatment that
passivates lead surface atoms
achieving an inorganic atomic ligand passivation strategy which
passivates both types of
nanocrystal surface atoms. Using this technique we were able to
obtain record high performance
with certified results of 5.1% and 6% in the lab.
Chapter 5 establishes the importance of interface recombination
on device performance. In this
chapter we use advanced characterization techniques such as
temperature dependent current-
voltage and transient photovoltage measurements to investigate
recombination kinetics in these
materials. Using atomic layer deposition we designed an
interfacial buffer layer which led to
improved open circuit voltage and enhanced photocurrent
collection.
Chapter 6 details our investigation into photocarrier collection
efficiencies in our devices. Using
device J-V data we examine recombination loss mechanisms as a
function of intensity. An
analytical model is developed to fit the experimental data. We
use this model to accurately
predict a diffusion length of 90 nm for our best CQD
materials.
Chapter 7 builds on the insights on recombination and transport
developed in Chapter 6.
Transient photoinduced absorption and photoconductivity
measurements are used to study
photocarrier kinetics. With this information we establish time
scales for carrier relaxation,
trapping, and transport. From Time-Resolved Infrared
Spectroscopy we identify sub-bandgap
states, which serve as an effective electronic bandgap in this
material.
Chapter 8 summarizes the major findings made throughout this
dissertation. We conclude with
recommendations for future work to further advance the field of
PbS CQD photovoltaics.
-
8
Chapter 2 Photovoltaics and Colloidal Quantum Dots
2.1 Photovoltaics Background
2.1.1 Solar Spectrum
Radiation from the sun is found to closely resemble that of a
black body emitter with a
temperature of 5800 K (Figure 2-1). Much of this power is
concentrated in the visible spectrum
with a long tail extending into the infrared spectrum. At the
top of the Earth's atmosphere the
composition of the solar radiation consists of approximately 50%
infrared, 40% visible, and 10%
in the ultraviolet portions of the electromagnetic spectrum [19]
and is designated AM0 [20]. The
total power over all wavelengths is 1366 W m-2
. After entering the Earth's atmosphere much of
the ultraviolet radiation is absorbed by ozone (O3). Additional
losses occur at distinct bands due
to absorption from O2, H2O and CO2 in the atmosphere. Finally
taking into account the zenith
angle of sunlight at mid-latitudes we can derive the solar AM1.5
spectrum [21] with a total
integrated intensity of 1000.4 W m-2
. The AM1.5 spectrum is considered the standard for
characterizing and comparing photovoltaic technologies.
-
9
Figure 2-1: Spectral irradiance for a 5800 K black body emitter,
AM0 [20] and AM1.5 [21] conditions. Absorption lines in
the AM1.5 for O3, O2, H2O and CO2 are labeled.
2.1.2 Theory and Operation of Photovoltaics
In their 1961 seminal paper, Shockley and Queisser were the
first to calculate the limiting
efficiencies of p-n junction solar cells through thermodynamic
analysis [22]. The basis of their
analysis relied on the assumption that photovoltaics rely on the
Principle of Superposition [23].
According to this principle the total current from the solar
cell under illumination was a linear
supposition of the dark diode current of the p-n junction,
IDiode (V) and generated photocurrent,
IL:
(V) (2.1)
The diode current of a p-n is the sum of diffusion currents on
each side of the junction [24]:
(2.2)
where I0 is the dark saturation current, n is the ideality
factor of the device, V is the applied bias,
k is Boltzmann's constant, and T is the temperature. Under
illumination the I-V characteristics
are shifted vertically by the illumination current IL (Figure
2-2). The result is that power is being
-
10
generated in the 4th I-V quadrant. The boundaries of the I-V
curve are marked by the short-
circuit current ISC and the open circuit voltage VOC.
Figure 2-2: I-V characteristics of a solar cell in the dark and
under illumination. Intersection of the shaded rectangle and
the I-V curve indicates the maximum power point. [Reprinted with
Permission][25].
The maximum efficiency of the solar cell is defined by:
(2.3)
where Vm and Im are the voltage and current at the maximum power
point and Pinc is the total
incident power. A more conventional form of the equation is to
relate efficiency to ISC and VOC
-
11
(2.4)
where FF is the device Fill Factor which is used to describe the
shape of the I-V curve.
Using these I-V relationships Shockley-Queisser were able to
theoretically determine the output
power of a solar cell and the limits to photovoltaic efficiency.
In their analysis they assumed that
every photon with hυ>Eg generates a photocarrier which
quickly relaxes to the bandedges of the
semiconductor. The total photocurrent is given simply by the
integral of all absorbed photons
from the incident solar spectrum [25]. Figure 2-3 shows the
integrated photocurrent as a
function of wavelength overlapped with the AM1.5 spectrum. For
convenience we normalize by
the device area and calculate photocurrent density.
Figure 2-3: Integrated photocurrent (right) as a function of
bandgap using the AM1.5 spectrum.
If we assume an ideality factor of n = 1 for an ideal solar cell
in Equation (2.2) and if I0 is
known the efficiency of the solar cell can be determined. By
treating the photodiode as a black
body, in thermal equilibrium with its surroundings, a lower
limit could be placed on the
saturation current density and loss factors in the device. At
equilibrium, and with no other loss
-
12
mechanisms to consider, photons are still absorbed by the diode
from the black body emission
from the surroundings (T = 300 K). For this condition of
equilibrium to be true the number of
absorbed photons from black body radiation must be equal to the
number of photons emitted
from radiative emission loss. It is these radiative losses which
make up the dark saturation
current I0, which can now be calculated much like was done for
IL by integrating over all photons
absorbed from a 300 K black body spectrum (Figure 2-4). Again
for convenience we normalize
by the device area to yield the saturation current density J0.
The addition of radiative losses
decreases the theoretical work that can be done by an absorbed
photon, reducing the quasi-Fermi
splitting, and open circuit voltage VOC. By re-arranging
Equation (2.1) we can look at the
theoretical limit of VOC as a function of bandgap.
(2.5)
Without losses VOC could be equal to Eg/q as indicated by the
blue curve in Figure 2-4. The
additional losses imposed by J0 (black curve) decreases the
quasi-Fermi level splitting and VOC
(red curve in Figure 2-4).
Figure 2-4: Thermodynamic limit for diode saturation current
density (right). Theoretical limit for VOC with (red) and
without (blue) consideration of thermodynamic losses.
Taking all these considerations into account the theoretical
efficiency of a p-n junction solar cell
operating under the AM1.5 spectrum has been determined as a
function of device bandgap
(Figure 2-5).
-
13
Figure 2-5: Ideal solar cell efficiencies as a function of
bandgap (Eg) at 300 K for 1 sun and 1000 sun concentrations.
[Reprinted with Permission][25].
At AM1.5G operating conditions a maximum theoretical efficiency
of 31% is predicted [25].
With higher intensities this efficiency can be even higher due
largely to the intensity dependence
of VOC as predicted from Equation (2.5). Overlapped with these
efficiency predictions are the
bandgaps of many common semiconductor materials. Two notable
examples are silicon and
gallium arsenide whose bandgaps lie near to the regime of
maximum efficiency.
2.2 Colloidal Quantum dots
2.2.1 Synthesis and Optical Properties
A quantum dot is a semiconductor nanocrystal where the
electronic and optical properties
demonstrate quantum size effects. Generally speaking if the
dimensions of a nanocrystal are
smaller than the Exciton-Bohr radius of the bulk semiconductor
material, or electron-hole pairs,
photocarriers will experience quantum confinement. In addition
to its bulk bandgap, the bandgap
of the quantum dots (E*) will have an additional size-dependent
quantum confinement and
Coloumbic interaction energy according to:
-
14
(2.6)
where ħ is Plank's constant, R is the nanocrystal radius, me and
mh are the masses of the electron
and hole respectively, and ε is the dielectric constant of the
material. Here the second term in
Equation 2.6 represents the quantum confinement energy and the
last term represents the
Coloumbic interaction energy.
This equation was originally developed by Louis Brus [26] in
1986, after being the first to
observe quantum dots in colloidal solutions. Since their first
discovery CQDs have become an
important topic of research due to both their interesting
chemistry and their unique physical
properties which can be leveraged for use in optoelectronic
devices.
Much of the synthesis of CQDs are based on a model developed by
LaMer [27][28] (Figure 2-
6). In a coordinating solvent reagents are added at temperatures
up to 300 °C resulting in the
formation monomers of the nanocrystal compounds [29]. If the
concentration of theses
monomers is high enough to breach a nucleation threshold
crystallization will occur and the
initial nuclei of the nanoparticles will form. As these nuclei
form the concentration of the initial
monomers begin to drop significantly. Under ideal conditions
concentrations will fall below the
nucleation threshold. Once the nucleation phase of the reaction
is complete the remaining
monomers react with the existing nuclei resulting in growth of
the nanoparticles. Proper control
of the synthesis conditions can result in a monodisperse
distribution of particles with standard
deviations of less than 5% without subsequent processing
[30].
-
15
Figure 2-6: Schematics showing the nucleation and growth of
colloidal nanocrystals in which various sizes of crystals can
be isolated at different times. (b) Simple synthetic apparatus
showing the preparation of CQDs. [Reprinted with
Permission][27].
Variation of the temperature, precursor concentrations, and
duration of the reaction can be used
to synthesis quantum dots of different sizes and thus different
bandgaps. In Figure 2-7 we see
absorption spectra for different sized PbS quantum dots with
diameters ranging from 3 to 10 nm
[31]. Using PbS nanocrystals gives access to bandgaps extending
from the visible to the infrared
portions of the electromagnetic spectrum.
-
16
Figure 2-7: Size dependent absorption of PbS quantum dots.
[Reprinted with Permission][31].
The quality of the quantum dot sample can be gauged by comparing
absorption at both the
exciton peak and local minimum at slightly shorter wavelengths
(Figure 2-8). This figure of
merit, hence forth called the peak-to-valley ratio, is a
function of both homogeneous and
inhomogeneous broadening [32][33][34] where the inhomogeneous
broadening is influenced by
polydispersity in both quantum dot size and shape. An important
optical characteristic of
quantum dots is the common presence of a Stokes Shift where
emission occurs at longer
wavelengths than absorption [27][35]. The origin of Stokes Shift
in quantum dots is often
attributed to the presence of a dark exciton state [36][37],
although experimental values for
Stokes Shift energies, such as those reported in Table 7-1, are
significantly larger than those
predicted by theory from Coulombic and exchange interactions
[38][39]. This matter will be
covered in more detail in Chapter 7 of this work.
-
17
Figure 2-8: Absorption and ohotoluminescence spectra for PbS
quantum dots with a 950 nm exciton. The quality of the
quantum dot sample can be quantified by comparing absorption at
the exciton peak and the local minimum. The energy
difference between absorption and photoluminesence peaks is the
Stokes Shift.
2.2.2 PbS CQD Optoelectronics
Over the past several years the Sargent group has focused much
of its research efforts on the design and
characterization of PbS quantum dot based optoelectronics. This
started out in the development of PbS
photoconductors and photodiodes. Eventually this research was
expanded into the design and fabrication
of photovoltaics.
a. Photodetectors
The bandgap tunability of PbS quantum dots allowed for the
engineering of solution processed
photodetectors sensitive to specific regions of the
electromagnetic spectrum, particularly the near infrared.
By using solid state treatments with dithiol passivating
ligands, photodiodes were fabricated with
quantum dots with bandgaps of ~1500 nm [40].
ΔStokes
-
18
Overall a specific detectivity (D* ) of 1011 Jones was obtained
with a bandwidth of 20 kHz (Figure 2-9).
Specific detectivity is defined by:
2-7)
where Δf is the bandwidth of the detector, A is the device area,
and NEP is the Noise Equivalent Power.
The NEP is the optical power where the photodetector signal is
equivalent to the noise floor.
Figure 2-9: a) A schematic representation of a PbS photodiode
device. b) Frequency dependence of photocurrent at zero
bias. [Adapted and reprinted with Permission][40].
-
19
b. Schottky CQD Photovoltaics
The first demonstrated CQD photovoltaics were fabricated using
Schottky architectures
[41][42][43]. The advantage of a Schottky architecture is its
relative simplicity. The
semiconductor absorber material is sandwiched between two
asymmetric contacts. On one side
an Ohmic contact collects the majority carrier whereas on the
other side a Schottky junction
creates a built-in field which drives charge separation and
collection of the minority carrier. The
early Schottky architectures provided an excellent proving
ground for optimization and
demonstrating the photovoltaic properties of CQD solids. By
treating the quantum dot films with
ethanedithiol or butylamine it was possible to passivate surface
defects and bring the quantum
dots closer together into a much more conductive film. With
these advances AM1.5 photovoltaic
performance eventually achieved 1.8% [44] (Figure 2-10).
-
20
Figure 2-10 a) Schematic of the Schottky device architecture
consisted of PbS CQDs as active material and Al as metal
contact. The inset shows the electron micrograph of PbS film
after n-butylamine exchange. b) Energy band diagram
showing the band bending takes place at Al/PbS interface. Under
illumination, electrons and holes are swept away by the
built-in electric field in the depletion layer. c)
Current–voltage data of the device influenced by the variation of
the
simulated solar illumination source. The device shows AM1.5 PCE
of 1.8%. d) EQE spectra for devices using PbS
quantum dots having different excitonic peak. [Adapted and
reprinted with Permission][44].
c. Heterojunction PbS CQD Photovoltaics
Although the initial Schottky architectures demonstrated very
promising results, new
architectures were also being explored in an attempt to greatly
enhance performance. One of the
primary limitations of Schottky solar cells is that the built in
field is limited to half the bandgap
[45]. This places a limit both on open circuit voltage, VOC and
the depletion region thickness for
charge collection. Another limiting factor was that for these
p-type materials the transparent
conductive oxide was used to form the Ohmic contact at the front
of the cell. Low work function
-
21
metals would then form both a Schottky and reflecting contact at
the back of the cell. This
proved to be limiting as most of the photocarriers are generated
furthest from the Schottky
junction where the built-in field is strongest and charge
extraction is most efficient. Inspired by
the potential of work conducted on Dye-Sensitized Solar Cells in
recent years work began on
using PbS quantum dots as the absorber in this architecture.
Contrary to the initial objectives the
best devices were obtained without reliance on infiltration of
the n-type TiO2 window layer with
PbS quantum dots. Instead a depleted N-p heterojunction
architecture was designed with PbS
CQD films deposited directly on TiO2 electrodes. The highly
n-doped TiO2 window layers
allowed for depletion regions several hundred nanometers thick
extending throughout the entire
thickness of the devices at zero bias conditions. Unlike the
Schottky architecture illumination
occurred on the side of the minority collecting TiO2-PbS
interface allowing for more efficient
current extraction. In the move towards the depleted
heterojunction design the solid state
treatment was changed from short dithiols such as ethanedithiol
to mercaptopropionic acid
(MPA) a bidendate ligand containing both thiol and carboxylic
acid functional groups. The new
MPA treatment proved to greatly improve film conductivity and
led to significant enhancement
in device performance. Overall AM1.5 performance efficiencies of
5.1% [46] were obtained, a
near three fold improvement to the Schottky devices developed 2
years earlier (Figure 2-11).
-
22
Figure 2-11: a) The depleted heterojunction architecture showing
the various components of the device (FTO/porous
TiO2/PbS QD/Au) along with the band diagram close to maximum
VOC. EF,n and EF,p are the electron and hole quasi-
Fermi levels; Ec and Ev are the conduction and valence band
edges; Jp, PV and Jn,PV are the hole and electron
photocurrents (and are equal at steady-state); Jp,fwd is the
hole current in the forward bias direction. The Fermi level is
shown as a dashed line. b) Apertured dark and illuminated
current density-Voltage (J-V) response of the photovoltaic
device c) EQE and absorption spectra of the photovoltaic device.
[Adapted and reprinted with permission][46].
-
23
2.3 Conclusions
Device performance efficiencies of 5.1% proved to be a major
advancement over the sub-percent
values obtained just several years earlier. These advancements
have mainly come from
improvements to the device architecture and better surface
passivation of the quantum dot
absorber. Despite this progress, performance is still well below
the theoretical maximum of 31%
predicted by the Shockley-Queisser limit. Further improvements
must be made through reduced
recombination by improved defect passivation at the
heterojunction interface and on the quantum
dot surface in the active absorber.
-
24
Chapter 3 Early Studies: Core-Shell Quantum Dots
3.1 Introduction
In the previous chapter we established that CQDs were
particularly interesting for use in
optoelectronics due to their combination of bandgap tunability
and solution processibility. Each
of these features of CQDs is a direct consequence of the
nanoscopic dimensions of these crystals.
This produces a high surface-to-volume ratio, increasing the
potential for recombination losses in
solar cells based on these materials: unpassivated surface
states or dangling bonds may serve as
recombination centers for photogenerated carriers, limiting
carrier lifetimes and restricting the
efficiency of charge transport and the extent of Fermi-Level
splitting.
Mitigating these challenges requires a strategy to engineer
surface passivation, removing surface
defects while maintaining the desirable properties that make
CQDs so appealing.
Core-shell nanostructures may serve as a possible remedy to this
problem. Growing
semiconductor shells on nanocrystal cores builds off the field
of epitaxial growth of
semiconductors [47]. Previous reports on CQDs have shown that
implementing a core-shell
structure can significantly improve photoluminescent quantum
yield by reduction of the non-
radiative recombination. Additionally, a shell may serve to
protect against photo-oxidation
effects by acting as a barrier between photogenerated carriers
and the external environment.
In this chapter, the early studies of the application of the
core-shell quantum dot concept to CQD
solar cells are reported. Initially an analysis of the various
options for core-shell materials, based
on heterojunction type selection is described. We then report
the development of the synthesis of
these nanoparticles. We build devices and discover some of the
important practical challenges
that limit the use of established core-shell synthesis methods
in CQD photovoltaics. These
studies pave the way for the new materials processing avenues
explored in Chapter 4 and
beyond, approaches that overcome the limitations of the
traditional core-shell approach.
-
25
3.2 Core-Shell Heterojunction Strategies
There exist four different classes of core-shell structures,
which are differentiated by the band-
alignment between the core and shell materials (Figure 3-1). In
type-I, both the electron and the
hole are confined to the core. In reverse type-I, both the
electron and hole are confined to the
shell. In type-II heterostructures, the electron and hole are
localized to different regions (one in
the shell, the other in the core).
Figure 3-1: Schematic representation of the energy-level
alignment in different core/shell systems realized in
semiconductor NCs to date. [Reprinted with Permission][47].
After photogeneration electrons and holes will relax to the
lowest lying bandedge. In the case of
both Type-II heterostructures, along with the Reverse Type-I
heterostructure, the total energy of
the electron-hole pair system will be reduced. Only in the case
of the Type-I heterostructure is
the energy of the absorbed photocarriers preserved after initial
relaxation. If shell thicknesses can
be made thin enough charge transport can be still be efficient
and the improved surface
passivation of the nanocrystals from the shell may reduce
recombination losses from surface
defects. Thus we desire a Type-I heterostructure design to
reduce carrier losses from surface
recombination, protects the core from photo-oxidation, and
maintains energy of the energy of the
photogenerated electron-hole pair.
Type-I Core-shell quantum dots may be synthesized by injecting
the precursors for the shell
material after initial synthesis of the core nanocrystals. This
technique has been used to synthesis
core-shell structures such as CdSe-ZnS [48]. One of the
challenges with this technique is the
-
26
potential that a heterogeneous solution may be produced
consisting of two different nanocrystal
materials instead of the desired core-shell structure.
Another possibility is to make use of a cation exchange [49].
After the initial synthesis of the
core nanocrystals the desired cation is added at elevated
temperatures. Under the right conditions
a cation exchange may occur on the surface of the quantum dots
where the original cation of the
core is replaced, resulting in a core-shell structure. This
technique has been utilized to produce
PbSe-CdSe core shell structures (Figure 3-2). Here the elevated
temperatures and heavy excess
of Cd-oleate lead to favourable conditions for replacement of Pb
with Cd atoms.
Figure 3-2 TEM imaging of PbSe-CdSe core-shell nanocrystals.
Inset: Individual core-shell quantum dot. [Reprinted with
Permission][49].
Here we utilize this approach for our PbS quantum dots. Based on
the varying reports of the
electron affinity for CdS [47][50] we predict that PbS-CdS may
form either a type-II or Type-I
heterostructure.
-
27
3.3 PbS-CdS Core-Shell Synthesis and Characterization
The synthesis of PbS nanocrystals was performed using the
standard air-free Schlenk-line
technique based on the variation of a previously reported method
(Figure 3-3) [35]. A stock
solution of lead oleate was prepared by pumping a mixture of 4.0
mmol of PbO (0.45g), 4.8
mmol of oleic acid (1.34g ), and 56.2 mmol of 1-octadecene
(14.2g) at 95 °C under vacuum for
16 h. The sulfur precursor was made by mixing 1.0 mmol
bis(trimethylsilyl)sulfide (0.18g) with
10 mL of 1-octadecene in a nitrogen-filled glovebox. The stock
solution was stirred vigorously
while being heated to 120 °C under argon in a three-neck flask
equipped with a thermocouple.
The sulfur precursor was swiftly injected into the flask. The
solution turned brown immediately
after injection and the reaction was quenched to 36 °C. To
isolate the nanocrystals, 50 mL of
anhydrous acetone was injected into the flask and the dispersion
was centrifuged. After removing
the supernatant, the mixture was re-dispersed in toluene and
then re-precipitated again with the
addition of 20 mL acetone. The final PbS nanocrystals, with an
average diameter of ~3.5 nm
was redispersed in toluene (~200 mg mL-1
).
The Cd-oleate solution was synthesized by adding 2.0g of CdO,
12.0 mL of oleic acid, and 32.0
mL of diphenyl ether in a three neck flask and then heated under
N2 to 255 °C (Figure 3-3). As
the boiling point of the diphenyl ether is 258 °C we require the
use of a water chilled condenser
for the recondensing diphenyl ether vapor. As the reaction
proceeds the cadmium oxide reacts
with the oleic acid forming the Cd-oleate complex and water.
To complete the Cd-oleate exchange occurs by adding 10 mL of PbS
quantum dots in toluene
(15 mg mL-1
) to a three neck flask under N2 (Figure 3-3). The solution was
heated to 90°C
before injection of 22 mL of Cd-oleate solution. Isolation of
the sample was completed by
adding 5 mL of the CQD solution to a mixture of 3 mL chloroform,
3 mL methanol, and 3 mL
acetone and then centrifugation at 1500 rpm for several minutes.
The addition of chloroform is
specifically important for proper removal of diphenyl ether from
the Cd-oleate complex. Finally
-
28
the resulting precipitate was dried under N2 and the final
product was redispersed in 2 mL
toluene.
Figure 3-3: Synthesis of core-shell PbS-CdS quantum dots by
cation exchange. The PbS cores are synthesized and then
purified. Formation of the core-shell dots is accomplished by
reaction with the Cd-oleate complex.
To gauge the success of the exchange we measured absorption on
the samples throughout the
exchange (Figure 3-4a). As expected, implementation of the
core-shell structure leads to a blue
shift in the absorption spectra as the core size is reduced from
cation exchange at the surface.
Over the course of an hour the exciton position shifts 160 nm
from 1250 nm to 1090 nm with
much of the shift occurring over the first 10 minutes. During
this phase of the reaction excess
cadmium atoms have easy access to lead atoms on the surface of
the dot. As the reaction
proceeds it becomes more difficult for cadmium and lead atoms to
diffuse between the core and
the outer environment. As a result, after 30 minutes there is
very little change in the absorption
spectra and the reaction appears to be self-limiting. It is also
clear that as this blue shift proceeds,
the exciton becomes less well-defined. This is evident if we
compare the relative absorption at
the local minimum and maximum near the exciton as indicated in
Figure 3-4a. If the cation
exchange occurs at different rates for different quantum dots
due to local fluctuations in reagent
-
29
concentrations initial polydispersity in size of surface
faceting the polydispersity of the final
product may be made worse by the cation exchange treatment.
Photoluminescence (PL) measurements were conducted on these
samples (Figure 3-4b). A 633
nm HeNe laser was used to excite the sample and photoluminescent
spectrum was recorded
using an InGaAs (Ocean Optics NIR-512) detector. Absorption of
the solutions was carefully
matched at the excitation wavelength by varying the
concentration of the sample to ensure
deviations of no more than 10-20% in optical density. It is
clear that the Cd cation treatment has
a profound effect on the photoluminescent spectra. With initial
treatment the photoluminescent
intensity and therefore PLQY is greatly enhanced as the outer
surface becomes passivated with
Cd atoms and a PbS-CdS core-shell structure is formed. Over time
there is a loss in PL intensity
which coincides with a loss in exciton definition. It is likely
that for these thicker shells the
greater degree of Cd penetration into the PbS core creates
surface defects which may begin to act
as recombination centers that quench PL.
Figure 3-4: a) Absorption spectra for Cd-Oleate treated PbS
quantum dots. b) Photoluminescence spectra for Cd-Oleate
treated PbS quantum dots.
-
30
3.4 Pyridine Ligand Exchange
Development of core-shell quantum dots is only the first step in
fabrication of high efficiency
CQD photovoltaics. A pathway to densely packed, highly
conductive quantum dot films is also
necessary. To accomplish this we set out to introduce a solution
phase ligand exchange on the
core-shell nanocrystals from the long chain oleic acid to short
capping layers such as pyridine
(Figure 3-5). Pyridine is a much shorter capping molecule with a
proven binding affinity with
cadmium and has been used previously for fabrication of CdSe
nanocrystal films [51]. The short
pyridine capping ligand allows for fabrication of densely packed
quantum dot solids. An
additional benefit is that due to the weak binding energy of the
pyridine nitrogen bond to the
quantum dot surface there is a relatively low activation barrier
for pyridine removal. Through
additional annealing strategies it may be possible to completely
remove the organic pyridine and
yield an all inorganic CQD solid which maintains quantum
confinement due to the protective
CdS shell.
Figure 3-5: Ligand exchange of oleic acid with pyridine.
To conduct the pyridine exchange 1 mL of core-shell quantum dots
in toluene (150 mg mL-1
)
was isolated by adding 2 mL of the methanol and centrifugation
at 1500 rpm for several minutes.
The precipitate was dried for several minutes under nitrogen
before redispersal in 10 mL of
pyridine. The solution was injected into a three neck flask
under nitrogen and stirred for 3 hours
at 90 °C. The quantum dots were then isolated by addition of
hexane (4:1 hexane:pyridine) and
centrifugation. Finally the sample was then dried and
redispersed in chloroform to achieve a final
concentration of ~70 mg mL-1
.
-
31
We measured the absorption of the exchanged dots with the
original material (Figure 3-6a). The
effect of the pyridine treatment was a complete loss of the
characteristic exciton. This was
interpreted as a consequence of the corrosive nature of
pyridine, a base, which may etch or even
destroy the quantum dots.
We evaluated the success of the ligand exchange by using Fourier
Transform Infrared (FTIR)
spectroscopy (Figure 3-6b) to look at absorption of vibrational
modes for specific functional
groups found in the ligands featured in Figure 3-5. Thin films
were prepared by spin casting the
pyridine exchanged quantum dots directly on KBr substrates.
Results show that there is still a
significant amount of C-O and C=O vibrational features in the
new material. This is highly
suggestive that significant oleic acid is still present and has
not been completely exchanged by
pyridine.
Figure 3-6: a) Absorption measurements of PbS and pyridine
exchanged PbS-CdS. b) Fourier Transform Infrared
(FTIR) spectroscopy of pyridine exchanged PbS-CdS The feature at
2360 cm-1
is CO2 absorption in the baseline
subtraction .
Pyridine is a neutral-donor (L-type) ligand [52] whereas oleic
acid, when deprotonated, is a
charged, lattice terminating (X-type) ligand [52] . Displacement
of oleic acid with pyridine thus
requires additional charge balancing to maintain overall charge
neutrality. Charge neutrality can
be acheived by removal of surface atoms from the nanocrystal
surface [52] or through charged
species, such as displaced oleic acid, in the colloidal
solution. This need for charge neutrality
places a barrier preventing the desired ligand substitution from
reaching completion.
-
32
We found that both the loss in the exciton and the incomplete
success of the ligand exchange
would not help us reach our goals of a densely packed,
monodispersed, conductive quantum dot
solid. Instead we explored a solution based ligand exchange with
mercaptopropionic acid
(MPA) (Figure 3-7). Mercaptopropionic acid has consistently
proved to be the best ligand for
fabrication of quantum dot photovoltaics. Since 2009 all device
record efficiencies using PbS
quantum dots have incorporated MPA [46]. Additionally the thiol
functional group is highly
reactive with metal atoms like lead and thus would greatly
facilitate an exchange and removal of
oleic acid.
3.5 Mercaptopropionic Acid Exchange
To complete the exchange core-shell quantum dots diluted in
toluene (50 mg mL-1
) were mixed
with equal amounts of MPA for ~2 hours. Afterwards hexane, a
non-solvent, was added to the
solution and a precipitate immediately began to form. After the
supernatant was removed,
tetrahydrofuran was added to redissolve any excess MPA in the
remaining precipitate. The
suspension was centrifuged and the clear supernatant was
removed. The final precipitate was
dried under vacuum and then finally redispersed in
dimethylsulfoxide (DMSO). To check the
colloidal stability the resulting solution was passed through a
0.2 μm filter. All the solution was
able to pass through the filter without clogging indicating a
good colloidal dispersion had been
formed. Due to the long alkyl chains oleic acid PbS quantum dots
have only demonstrated
colloidal stability in non-polar solvents such as octane or
toluene [35][53]. The fact that colloidal
solutions could be prepared in DMSO, a highly polar solvent, was
an initial indication of a
successful exchange to MPA.
-
33
Figure 3-7: Ligand Exchange of Oleic Acid with Mercaptopropionic
acid (MPA)
We examined the absorption spectra before and after the MPA
exchange (Figure 3-8). Unlike
the previous pyridine exchange there is still a clear exciton
feature present after the exchange. A
slight blue shift of the exciton appears to have occurred and is
likely a result of minor etching
from the acidic environment of the MPA.
Figure 3-8: Absorption spectra for PbS-CdS and MPA exchanged
PbS-CdS.
To further test the extent of the exchange solution Nuclear
Magnetic Resonance (NMR)
spectroscopy was used to measure the ratio of oleic acid to MPA
(Appendix A). Oleic acid has a
specific NMR feature due to the carbon double bond [54]
(5.25-5.35 ppm) that is not present in
MPA. Similairly MPA which binds to the nanocrystal surface
through the thiol functional group
shows a NMR signature of the carboxylic acid [54] (12.15 ppm)
functional group. This feature is
absent for bound, and therefore deprotonated, oleic acid. This
allows us to directly compare the
-
34
presence of MPA and oleic acid in the final product. Results
indicate that the exchange is almost
entirely complete with ~1% of total organic species being oleic
acid.
We performed TEM imaging of the MPA exchanged quantum dots and
compared them with
some of the original oleic acid capped nanocrystals (Figure
3-9a). The TEM imaging showed
that MPA capped nanocrystals (Figure 3-9b) are much closer
packed than those functionalized
with oleic acid thus indicating a high degree of success for
this exchange.
Figure 3-9 TEM imaging of PbS-CdS and MPA exchanged PbS-CdS
nanocrystals (Scale Bar=100 nm).
We attempted to make films for photovoltaic devices but found
that the boiling point of the
DMSO was too high, and the evaporation rate too low, to make the
optically dense films
necessary to make photovoltaic devices. To test the electrical
properties of the core-shell
quantum dots we instead used a solid state treatment with MPA
similar to those previously
reported [46]. Device data showed very little photovoltaic
response with JSC values ~0.1 mA cm-
2. It was believed that the thick CdS shell was too thick for
efficient charge carrier extraction.
-
35
3.6 Conclusion
In summary, we demonstrated the synthesis of PbS-CdS core-shell
quantum dots. These
exhibited the enhanced photoluminescence characteristic of
well-passivated quantum dots. We
attempted device fabrication, but found very low performance
below 0.1%.
These early studies emphasized both the desired
passivation-enhancing, but also the undesired
transport-blocking, role of shells. They motivated
investigations of innovative strategies that
would overcome this compromise. We will see in the ensuing
chapter that – fortunately – new
concepts inspired by the core-shell preliminary findings enabled
a breakthrough in no-
compromise CQD passivation.
-
36
Chapter 4 The Advent of Atomic Ligand Passivation
4.1 Introduction
In the previous chapter we found that the thick CdS shell on the
PbS dots was too insulating and
inhibited photocarrier transport and photovoltaic performance.
Yet we also found that inorganic
strategies have much to offer in improving passivation.
This inspired a new concept: the idea that – perhaps – one of
the constituents (the cations or
anions) employed in shell growth might be the prime agent
benefiting passivation. If one of these
played a predominant role in improving trap state passivation,
then perhaps a shell analogue –
yet less than a lattice constant thick – could confer the
benefits of improved passivation with
little or no cost to transport.
In this chapter, we begin with a brief review of early studies
along these lines. We then present a
detailed hypothesis as to how an atomic ligand passivation
strategy could proceed. We develop
and optimize a new process that shows a dramatic improvement in
photoluminescent quantum
yield. We deployed these new nanoparticles and demonstrated a
dramatic improvement in solar
cell current and voltage. We conclude with an articulation of
the path forward for further
improvements in CQD photovoltaic efficiency.
4.2 Motivation
Previous work exploring the air stability of PbS quantum dots
demonstrated the importance of
surface composition on device performance [55]. Quantum dots
were synthesized with a high
lead to sulfur composition (1.8:1) indicative of a
stoichiometric core and highly lead rich surface.
The highly lead rich surface significantly reduced the formation
of PbSO3 and PbSO4. Both these
species are believed to introduce recombination centers which
are likely detrimental to
performance [56]. Having a cation rich surface is thus
beneficial as it passivates exposed surface
states and reduces the probability of defect formation. Based on
our work in the previous chapter
-
37
we attempted to develop a Cd cation treatment to passivate these
exposed sulfur sites. The higher
binding energy of cadmium vs lead, which was responsible for the
success of the cation
exchange process, should also lead to more complete cation
coverage of the surface and better
protection from defect formation.
For this passivation strategy we wanted to form a cadmium
protection layer on the outside
surface of the quantum dot and avoid the cation exchange
process, which in the previous chapter
we demonstrated to be detrimental to performance. By using a
different coordinating ligand for
the Cd precursor we hoped to develop a less reactive Cd reagent
to prevent Pb removal through
cation exchange. For inspiration we looked to synthesis work on
CdS nanocrystals [57]. Here
phosphonic acids are considered a staple coordinating ligand due
to their strong binding energy
to Cd atoms and their ability to yield colloidally stable
quantum dots in organic solvents. The
stronger binding energy of phosphonic acids compared to
carboxylic acids [58] should help
reduce the reactivity of the Cd precursor.
4.3 Cation Passivation
The PbS quantum dots were synthesized according to procedures
discussed earlier. Parameters
were established to yield excitons of ~950 nm [46]. After
initial synthesis the quantum dots are
isolated in a nitrogen glovebox and redispersed in octane.
Approximately 1 mL of PbS quantum
dots in toluene are added to a test tube with ~12 mL of
methanol. Not only is methanol a non-
solvent for these quantum dots, it also has a strong affinity
for oleic acid and thus allows for
removal of any excess capping ligand. The solution is
centrifuged at 1500 rpm for 10-15 minutes
resulting in a dark PbS nanocrystal precipitate and a clear
supernatant. The supernatant is
removed, and the quantum dots are dried under vacuum then
redispersed in toluene. Another
isolation is performed and the dots are dried again for >1
hour to ensure full removal of any
excess solvent. The dry precipitate is finally redispersed in
octane to a final concentration of 50
mg mL-1
.
A new cadmium complex was synthesized by mixing 2 mmol of CdCl2,
with 0.4 mmol of
tetradecylphosphonic acid (TDPA) and 10 mL oleylamine (OLA) in a
three neck flask attached
-
38
to a Schenk line. The solution is heated under N2 to 100°C and
magnetically stirred for 30
minutes. Here CdCl2 acts at the Cd cation source,
tetradecylphosphonic acid acts as the
coordinating ligand, and oleylamine as the coordinating solvent.
In a typical treatment 12 mL of
the pre-synthesized PbS quantum dot solution is added to 24 mL
of toluene in a three neck flask
and heated to 60 °C under N2. Finally ~4 mL of the Cd-TDPA-OLA
complex is added and the
reaction is allowed to proceed for 5 minutes before quenching
with ~40 mL of acetone at room
temperature. The quantum dot material is isolated by
centrifugation and redispersed in toluene
to a concentration of 150 mg mL-1
. This isolation procedure is repeated an additional two
times
before final dispersal in octane at a concentration of 50 mg
mL-1
.
Absorption measurements before and after the Cd-TDPA-OLA
treatment show a slight red shift
in the exciton position (Figure 4-1). This is in contrast to the
blue shifts witnessed for the Cd-
oleate treatments described earlier. This suggests that this new
Cd source is not displacing Pb
atoms on the surface. Rather it appears that the Cd is being
added onto the surface of the
quantum dots, effectively increasing their size, resulting in
the slight red shift of the exciton. As
indicated in T