Alternatives to organic acid surface modification of ZnO for excitonic photovoltaics

ALTERNATIVES TO ORGANIC ACID SURFACE

MODIFICATION OF ZNO FOR EXCITONIC

PHOTOVOLTAICS

by

A thesis submitted to the Faculty and the Board of Trustees of the Colorado School

of Mines in partial fulfillment of the requirements for the degree of Doctor of Philosophy

(Applied Physics).

Golden, Colorado

Date

Signed:

Thomas M. Brenner

Signed:

Prof. Reuben T. Collins

Thesis Advisor

Signed:

Prof. Thomas E. Furtak

Thesis Advisor

Golden, Colorado

Date

Signed:

Thomas E. Furtak

Professor and Head

Department of Physics

ABSTRACT

Surface modification of metal oxides with molecular monolayers is an effective strategy

for tuning interface properties in excitonic devices employing metal oxides as charge

accept-ing and transport layers. The most commonly used attachment chemistries are acid/base

reactions employing organic acids. The use of acid/base chemistries has presented a problem

for one of the most commonly used and promising metal oxides in excitonic devices, zinc

ox-ide (ZnO). ZnO is easily etched by even weak organic acids, leading to non-ox-ideal monolayers

and the accumulation of surface complexes during etching, which is particularly problematic

for ZnO-based dye sensitized solar cells (DSSCs).

Two ways to address this issue have been explored. The first approach is to employ

a triethoxysilane (TES)-based covalent attachment scheme instead of an acid/base

reac-tion for attaching modifier molecules. We demonstrate that dipolar mixed monolayers of

phenyltriethoxysilane-based molecules tune the work function of ZnO and the performance

of bulk heterojunction photovoltaic devices containing modified ZnO layers. This indicates

these modifiers are effective for tuning interfacial electronic structure.

The second approach is to investigate Zn

Mg

O (ZnMgO) alloys in order to produce a

more etch resistant material with similar electronic properties to ZnO. These alloys, when

exposed to the prototypical modifier benzoic acid (BA), demonstrate a steady-state,

macro-scopic etch rate that decreases up to an order of magnitude (at 20% Mg) compared to ZnO.

Infrared spectroscopic characterization of BA-modified ZnMgO indicates a monolayer of BA

attaches to the ZnMgO surface nearly instantaneously and remains throughout etching.

These results suggest that ZnMgO is a promising alternative material that may alleviate

some of the problems with ZnO etching. However, for applications of this material as a

substrate for dye sensitization, the initial etch rate, and not the steady-state rate, is really the

quantity of interest. We investigated the initial etch rate of ZnMgO exposed to N3 dye

(cis-bis(isothiocyanato)bis(2,2’-bipyridyl-4,4’-dicarboxylato)-ruthenium(II)). We find the initial

etch rate of ZnMgO increases with Mg content, in contrast to the steady-state etch rates

observed for BA-treated ZnMgO. We also find that the primary products of etching are

Zn-carboxylate products. From these results we propose a mechanism for the observed etch

resistance.

TABLE OF CONTENTS

ABSTRACT

. . . iii

LIST OF FIGURES . . . .

viii

LIST OF TABLES . . . xii

LIST OF SYMBOLS . . . .

xiii

LIST OF ABBREVIATIONS . . . xvi

ACKNOWLEDGMENTS . . . xviii

CHAPTER 1 METAL OXIDE SEMICONDUCTORS IN EXCITONIC

PHOTOVOLTAICS . . . 1

1.1

Metal Oxide Semiconductors . . . 1

1.2

Organic and Fullerene Semiconductors . . . 4

1.3

General Photovoltaic Device Physics

. . . 10

1.4

Excitonic Solar Cell Device Physics . . . 17

1.5

Metal Oxide/Organic Interfaces . . . 21

1.6

Monolayer Modification of Metal Oxide Surfaces and Interfaces . . . 25

1.7

Acid Dissolution of Metal Oxide Semiconductors . . . 27

1.8

Thesis Organization . . . 29

CHAPTER 2 SAMPLE PREPARATION AND CHARACTERIZATION

TECHNIQUES . . . 31

2.1

Production of Zn

Mg

O Thin Films . . . 31

2.2

Triethoxysilane Modification of ZnO

. . . 32

2.4

Fabrication of Bulk Heterojunction Solar Cells . . . 34

2.5

UV-Vis Absorption Spectroscopy . . . 35

2.6

Infrared Absorption Spectroscopy . . . 36

2.7

Kelvin Probe Surface Potential Measurements . . . 39

2.8

Tapping Mode Atomic Force Microscopy . . . 42

2.9

X-Ray Photoelectron Spectroscopy . . . 45

2.10 Contact Angle Goniometry . . . 48

2.11 Grazing Incidence X-Ray Diffraction

. . . 49

2.12 Photoluminescence Spectroscopy . . . 51

CHAPTER 3 TUNING ZINC OXIDE/ORGANIC ENERGY LEVEL ALIGNMENT

USING MIXED TRIETHOXYSILANE MONOLAYERS . . . 53

3.1

Introduction . . . 54

3.2

Experimental . . . 58

3.3

Results and Discussion . . . 61

3.4

Conclusions . . . 71

3.5

Acknowledgements . . . 71

CHAPTER 4 ETCH-RESISTANT ZN

MG

O ALLOYS: AN ALTERNATIVE TO

ZNO FOR CARBOXYLIC ACID SURFACE MODIFICATION . . . 72

4.1

Introduction . . . 73

4.2

Experimental . . . 76

4.3

Results and Discussion . . . 78

4.4

Conclusions . . . 92

CHAPTER 5 EXPLORING THE MECHANISM OF ZN

MG

O ETCH

RESISTANCE THROUGH DYE SENSITIZATION

. . . 94

5.1

Introduction . . . 96

5.2

Experimental Methodology . . . 99

5.3

Results and Discussion . . . 104

5.4

Conclusions . . . 116

5.5

Acknowledgements . . . 117

CHAPTER 6 CONCLUSIONS . . . 119

6.1

Project Conclusions . . . 119

6.2

Future Project Suggestions . . . 122

REFERENCES CITED . . . 125

APPENDIX A - FURTHER EXPLANATION OF METHODOLOGY . . . 140

A.1 Infrared Active and Inactive Modes: An Example . . . 140

A.2 Fourier Transform Infrared Spectrometer Design . . . 141

A.3 Simplified Schematic of the Kelvin Probe . . . 142

A.4 TM-AFM Feedback Loop Schematic

. . . 144

APPENDIX B - SUPPORTING INFORMATION FOR CHAPTER 3 . . . 145

B.1 Calculation of Surface Proportion of 4CPTES and PTES from Infrared

Spectrum

. . . 145

B.2 Water Contact Angle Measurements . . . 148

B.3 Atomic Force Microscopy Measurements . . . 148

B.4 Dark J-V Curves . . . 150

B.5 Effect of Light Soaking . . . 150

LIST OF FIGURES

Figure 1.1

Hexagonal wurtzite structure of ZnO

. . . 3

Figure 1.2

Atomic force microscopy height images of Zn

Mg

O films produced by

a sol gel process . . . 5

Figure 1.3

UV-Vis spectra of the thin films of Zn

Mg

O studied in this thesis . . . . 6

Figure 1.4

Examples of common electro-active organic polymers and small molecules . 7

Figure 1.5

Illustration of Fermi level equilibration in junctions of electronic

materials . . . 11

Figure 1.6

Example of a current density - voltage curve for a solar cell . . . 14

Figure 1.7

Equivalent circuit model of a solar cell . . . 17

Figure 1.8

The operational steps of excitonic solar cells

. . . 18

Figure 1.9

Examples of excitonic photovoltaic device architectures . . . 19

Figure 1.10

Examples illustrating issues with metal oxide/organic semiconductor

interfaces . . . 22

Figure 1.11

Electronic structure of a metal oxide and an organic in isolation and in

contact . . . 23

Figure 2.1

UV-Vis absorbance spectra of materials used in this thesis . . . 37

Figure 2.2

Diagram of the working principle of the PM-IRRAS technique

. . . 40

Figure 2.3

Basic working principle of Kelvin probe surface potential measurements . 41

Figure 2.4

Behavior of the tip in contact mode atomic force microscopy and

tapping mode AFM . . . 43

Figure 2.5

Phase shift between AFM tip amplitude and driving force during

TM-AFM . . . 44

Figure 2.6

TM-AFM height and phase images showing the relationship between

topography, sample inhomogeneity, and phase

. . . 45

Figure 2.7

X-ray photoelectron spectroscopy experimental setup

. . . 47

Figure 2.8

Contact angle measurement setup and examples . . . 48

Figure 2.9

Experimental setup of grazing incidence X-ray diffraction . . . 50

Figure 2.10

Illustration of photoluminescence spectroscopy experiment setup and

spectrum of N3 dye . . . 52

Figure 3.1

Energy level alignment at an ideal metal oxide/organic interface can be

tuned by introducing a dipolar monolayer at the interface . . . 56

Figure 3.2

PM-IRRAS measurements of ZnO films treated with PTES and

4CPTES in different proportions . . . 63

Figure 3.3

Relative work function of treated ZnO as a function of 4CPTES and

PTES mole fraction . . . 65

Figure 3.4

Representative light J-V measurements of IBHJ photovoltaic devices

containing mixed monolayer modified ZnO . . . 67

Figure 3.5

Plot of open-circuit voltage of devices against relative work function of

treated ZnO films . . . 70

Figure 4.1

Bandgaps of ZnMgO films as a function of Mg content . . . 79

Figure 4.2

Background-subtracted grazing incidence X-ray diffraction spectra of

Zn

Mg

O . . . 80

Figure 4.3

PM-IRRAS infrared absorption spectra of benzoic acid-treated ZnMgO

films showing the carboxyl stretch region . . . 83

Figure 4.4

Peak fits to the dominant ν

(CO

) feature of the infrared spectra of

ZnMgO films soaked in benzoic acid for 1 hr . . . 84

Figure 4.5

Ratio of the integrated intensity of the ν

(CO

) modes observed on

benzoic acid treated ZnMgO . . . 86

Figure 4.6

Variation of the ν

(CO

) modes at 1532 and 1569 cm

in ZnO

samples as a function of exposure time to benzoic acid . . . 88

Figure 4.7

Plot of integrated intensity of 1550 and 1571 cm

_ν

asym

(CO

) modes

Figure 4.8

Relationship between amount of unreacted acetate in ZnMgO films and

amount of attached benzoic acid . . . 90

Figure 5.1

Molar absorptivity of N3 dye in ethanol solution . . . 101

Figure 5.2

Sensitivity factors of the orbitals measured during XPS as a function of

their binding energy . . . 103

Figure 5.3

Bandgaps for ZnMgO samples with 0 - 20% Mg used in Chap. 5 . . . . 104

Figure 5.4

UV-Vis and PL spectra of ZnMgO treated with N3 dye for 1 hr . . . 105

Figure 5.5

Normalized PL spectra of ZnMgO treated with N3 dye for 1 hr . . . 106

Figure 5.6

Surface coverage and PL integrated intensity of ZnMgO treated with

N3 for 1 hr . . . 107

Figure 5.7

PM-IRRAS Spectra of ZnMgO exposed to N3 dye for 3 hr and 24 hr . . 109

Figure 5.8

XPS composition measurements of ZnO and Zn

Mg

O before and

after treatment with 2 mM benzoic acid for 30 min . . . 111

Figure 5.9

XPS composition measurements of Zn and Mg before and after

treatment of Zn

Mg

O by BA for 30 min . . . 112

Figure 5.10

UV-Vis absorbance spectra of ZnMgO films sensitized with N3 after

undergoing a mineral acid pre-etch

. . . 113

Figure 5.11

Surface coverage as a function of total exposure time to N3 dye for

mineral acid pre-etched ZnMgO samples . . . 114

Figure 5.12

N3 surface coverage of a ZnMgO sample exposed to N3 dye for 24 hr

and a sample exposed to N3 for 2 hr after a pre-etch . . . 115

Figure A.1

Examples of infrared inactive and active modes of carbon dioxide. . . . 140

Figure A.2

Experimental setup of an FTIR spectrometer . . . 142

Figure A.3

Schematic of Kelvin probe circuit

. . . 143

Figure A.4

Diagram demonstrating the feedback loop that controls the AFM

microscope . . . 144

Figure B.1

Diagram demonstrating calculation of IR integrated intensities

Figure B.2

AFM phase images of monolayer treated ZnO surfaces.

. . . 149

Figure B.3

J-V curves of monolayer treated ZnO P3HT:PCBM devices in the dark

150

Figure B.4

J-V curves of 4CPTES-treated ZnO devices before and after light

LIST OF TABLES

Table 1.1

Bandgaps of Zn

Mg

O films calculated from UV-Vis spectra

. . . 5

Table 3.1

Characteristics of inverted bulk heterojunction devices with monolayer

modified ZnO electron contacts

. . . 66

Table 4.1

Etch rate of ZnMgO films exposed to 2 mM benzoic acid in hexane

. . . . 81

Table 4.2

Peaks identified in fits of the IR spectra of BA treated ZnMgO . . . 84

Table B.1

Water contact angle measurements of monolayer treated surfaces . . . 148

Table B.2

RMS roughness determined from tapping mode AFM topographic scans

of TES treated ZnO surfaces . . . 148

Table B.3

Shunt resistances for monolayer treated ZnO P3HT:PCBM devices in the

LIST OF SYMBOLS

Relative dielectric constant, also molar absorptivity . . . ε

Boltzmann constant

. . . k

Temperature

. . . T

Fermi energy in semiconductor physics . . . E

Electric potential . . . Φ

Built-in potential . . . V

Electron quasi-Fermi energy . . . E

Hole quasi-Fermi energy . . . E

Conduction state energy in inorganic and organic semiconductors . . . E

Valence state energy in inorganic and organic semiconductors

. . . E

Electron volume concentration in conduction states . . . .

n

Hole volume concentration in valence states . . . p

Density of states of the lowest energy conduction states . . . N

Density of states of the highest energy valence states . . . N

Current density . . . .

J

Fundamental unit of charge . . . e

Hole mobility . . . µ

Electron mobility . . . µ

Gradient operator

. . . ∇

Voltage (equivalent to electric potential) . . . V

Short circuit current . . . J

Open circuit voltage . . . V

Power generated by solar cell . . . P

Solar cell efficiency . . . η

Electrical current or light intensity . . . I

Fill factor . . . F F

Shunt resistance

. . . R

Series resistance . . . R

Fermi energy . . . E

Work function . . . .

φ

Energy difference between the LUMO level of an organic and the conduction band of

a metal oxide . . . E

Hydrogen ion otherwise known as a proton . . . .

H

Transmittance or transmittivity (equivalent) . . . T

Absorptivity, not to be confused with absorbance . . . Abs

Reflectivity . . . R

Wavelength . . . .

λ

Absorbance . . . A

Absorption coefficient

. . . .

α

Path length through sample . . . l

Molar concentration . . . c

Semiconductor bandgap . . . E

Energy or electric field (context-dependent) . . . E

Phase angle between tip amplitude and driving force in tapping mode atomic force

microscopy

. . . θ

Angular frequency of tapping mode atomic force microscopy tip . . . .

ω

Tip oscillation amplitude in tapping mode atomic force microscopy . . . A

Tip quality factor in tapping mode atomic force microscopy . . . Q

Tip spring constant in tapping mode atomic force microscopy . . . k

Inelastic mean free path of photoelectron in X-ray photoelectron spectroscopy . . . λ

Contact angle . . . θ

X-ray angle of incidence in grazing incidence X-ray diffraction . . . δ

Scattering angle in X-ray diffraction

. . . 2θ

Acid dissociation constant . . . pK

Symmetric (sym) or asymmetric (asym) stretch of the carboxylate group

(CO

) . . . .

ν

(CO

)

LIST OF ABBREVIATIONS

Transparent Conducting Oxide . . . .

TCO

Indium Tin Oxide

. . . ITO

Fluorine-doped Tin Oxide . . . .

FTO

Current density - voltage curve . . . J-V

Maximum power point of solar cell . . . MPP

Dye sensitized solar cell . . . DSSC

Highest occupied molecular orbital . . . HOMO

Lowest unoccupied molecular orbital . . . LUMO

Point of zero charge

. . . .

pzc

Isoelectric point . . . IEP

poly-3-hexylthiophene (semiconducting polymer)

. . . P3HT

phenyl-C

-butyric acid methyl ester (fullerene derivative) . . . .

PCBM

poly(3,4-ethylenedioxythiophene):poly(styrenesulphonate) (conductive polymer

blend) . . . PEDOT:PSS

Metal-to-ligand charge transfer . . . .

MLCT

Fourier transform infrared spectroscopy . . . FTIR

Polarization modulated infrared absorption spectroscopy . . . PM-IRRAS

Atomic force microscopy . . . AFM

Tapping mode atomic force microscopy . . . TM-AFM

Octadecanethiol . . . .

ODT

X-ray photoelectron spectroscopy . . . XPS

Kinetic energy

. . . .

KE

Binding energy . . . .

BE

Grazing incidence X-ray diffraction . . . GIXRD

Photoluminescence . . . PL

Bulk heterojunction

. . . BHJ

Titanium dioxide . . . .

TiO

Work function . . . WF

Monolayer . . . ML

Monolayer . . . ML

Inverted bulk heterojunction . . . IBHJ

Shorthand for Zn

Mg

O . . . ZnMgO

Benzoic acid . . . .

BA

Full width half maximum

. . . FWHM

Incident photon-to-current conversion efficiency . . . IPCE

ACKNOWLEDGMENTS

There are quite a few people who’ve helped me through the doctoral process who I need

to thank. I’d like to thank my family for their support and encouragement throughout this

process. I’m also forever grateful to Dayna Jacob for her companionship, encouragement,

and patience during the years I’ve spent doing the work of this thesis.

I’m also extremely grateful to my advisers, Reuben Collins and Thomas Furtak, for the

advice, guidance, ideas, and mentorship they’ve given me. I’d also like to thank Dana Olson

who has been a valuable mentor.

A number of my fellow students have been instrumental in my learning and in the

ac-complishment of this work. I’d like to thank Darick Baker for his guidance during my early

years of graduate school. I’d like to thank Gang Chen for his camaraderie and collaboration

throughout our graduate years together. Without Gang’s contributions, this thesis wouldn’t

be what it is today. I’d also like to thank Thomas Flores, Erich Meinig, Paul Ndione, and

Xerxes Steirer for their efforts on the projects in this thesis. Again, without them, this thesis

wouldn’t be what it is.

I’d also like to thank all the friends I’ve made along this journey for their kind words,

conversation, and companionship.

Finally, I’d like to thank the members of my committee: Reuben, Thomas, Dana, and

Mark Lusk, Brian Gorman, and Stephen Boyes. Their input and ideas during my thesis

proposal, research, and defense have been extremely valuable.

CHAPTER 1

METAL OXIDE SEMICONDUCTORS IN EXCITONIC PHOTOVOLTAICS

The term ’excitonic photovoltaics’ refers to solar cells in which light absorption generates

excitons, bound electron-hole pairs, instead of free carriers. Excitonic photovoltaic devices

have been around for about 25 years, beginning with the invention of the dye sensitized

solar cell in the 1990s, which took advantage of an ultrafast charge transfer from a light

absorbing dye to a transparent metal oxide semicondutor in order to generate photocurrents

and photovoltages. While the efficiency of these devices quickly saturated, new devices

appeared that involved all-soft matter active layers composed of semiconducting polymers

or smaller organic chromophores and fullerenes, referred to as organic photovoltaics. These

devices have quickly risen in efficiency since the discovery of the bulk heterojunction active

layer morphology in the early 2000’s.[1] Companies and research institutions are reporting

efficiencies above 12% and are designing multijunction organic devices that may surpass

single junction devices in efficiency. Dye sensitized solar cells have been making a comeback

recently as well, after more than a decade of stagnation. A number of different innovations

have allowed the world record to grow to 12.3%.[2] The metal oxide materials and molecular

attachment schemes discussed in this thesis have already played a large role in the progress

of excitonic photovoltaic devices, and likely will continue to do so in the future. In this first

chapter, an introduction to the materials and physics of excitonic solar cells, and the role

that metal oxides play in them, will be provided.

1.1

Metal Oxide Semiconductors

Metal oxide semiconductors form a group of electronically useful materials that are often

high-bandgap, which makes them transparent through most of the solar spectrum, and often

’intrinsically’ doped by defects in their structure, making them relatively good conductors

suited to transporting carriers without further processing. The ability to dope intrinsically

or extrinsically to achieve sufficient conductivity for device applications without significant

free carrier absorption in the visible leads to the term transparent conducting oxide (TCO).

Notable metal oxide examples include zinc oxide (ZnO), titanium dioxide (TiO

), nickel oxide

(NiO), molybdenum trioxide (MoO

), tungsten trioxide (WO

), and tin oxide (SnO

). These

materials haven proven to be extremely valuable in excitonic photovoltaics precisely for the

reasons just mentioned. Many of them can also be deposited through cheap, solution-based

processes, which are an added benefit. Extrinsic degenerate doping can be used to create

transparent conducting oxides such as indium tin oxide (ITO), fluorine doped tin oxide

(FTO), and Al or Ga doped ZnO that have sufficient conductivity to accomplish lateral

transport of carriers without significant resistive losses. These materials are often used as

transparent conducting contacts in optoelectronic devices.

ZnO features prominently in this thesis. It has a hexagonal wurtzite crystal structure

shown in Figure 1.1. ZnO has a bandgap of 3.2 eV. In the thin film form used in optoelectronic

devices, ZnO has n-type conductivity that has been attributed to nonstoichiometric defects in

its structure such as interstitial Zn atoms or oxygen vacancies that generate free electrons.[3]

This origin of conductivity has been under some debate recently.[4] The processing conditions

(such as annealing in oxygen or in the presence of hydrogen) can have a substantial influence

on the conductivity of ZnO and are necessary to consider when comparing ZnO prepared in

different ways.[3, 4] It is also relatively easy to form nanostructures of ZnO, compared to

other metal oxides and this is one of its major advantages.[5, 6] ZnO is also a piezoelectric

material.[7] It is particularly susceptible to dissolution by even weak organic acids, one of the

major subjects of this thesis.[8, 9] The variety of properties and applications of ZnO that have

been studied are too broad to be discussed here, but several reviews are available.[3–6, 10–12]

ZnO presents three well studied crystal faces: a non-polar face, a Zn-polar face, and

an O-polar face, labeled in Figure 1.1. A second non-polar face has also been studied, but

not as extensively.[11] Both the Zn- and O- polar faces could be terminated by either a Zn

or O atom. The Zn- and O- polar faces have opposite dipoles at their surfaces while the

Figure 1.1: Hexagonal wurtzite structure of ZnO. Gray atoms represent Zn and yellow atoms

are oxygen. The three surfaces of ZnO - Zn-polar, O-polar, and non-polar are identified. The

two polar faces can both either be Zn or O terminated. In this figure, the Zn-polar face is

Zn terminated, while the O-polar face is also Zn terminated.

non-polar face has no dipole. The surface structures in Figure 1.1 are hypothetical and

are informed by extrapolation from the bulk geometry. In reality, the non-polar surface

fits the bulk geometry, the Zn-terminated surface fits the bulk geometry but with apparent

defects, and very clean oxygen terminated surfaces show surface reconstruction.[11] However,

hyrdoxyl contaminated oxygen terminated surfaces (from water or hydrogen) do not show

reconstruction, and in fact clean surfaces will convert back to the un-reconstructed surface

upon contamination.[11]

The surface of ZnO is very reactive and will react with water and CO

to form adsorbed

monolayers and multilayers.[11] For this reason it is commonly used as a heterogenous

cata-lyst, especially for methanol synthesis.[11] Water displays a rich set of adsorption structures

on ZnO. As mentioned before, the oxygen terminated polar face is fully hydroxylated under

small amounts of contamination in UHV.[11] The non-polar face in Figure 1.1 displays a

par-tially dissociated monolayer of water at the surface in which dissociated and non-dissociated

molecules arrange in an alternating superstructure on the surface.[13] On nanocrystals of

ZnO, IR spectroscopy showed the presence of the above mentioned adsorption structures

along with other structures due to defects.[14] CO

can react with a hydroxylated ZnO

surface to form adsorbed hydrogen carbonate.[15] It is important to note here that single

crystal and solution processed ZnO appear to have different surface chemistries (regardless

of crystal face), as evidenced by their differing response to surface preparation techniques

used in device processing.[16]

Alloying metal oxides can produce materials with tunable properties. The alloy Zn

Mg

O

is featured in Chps. 4 and5 of this thesis. Substitution of Mg for Zn in the wurtzite

struc-ture causes the bandgap to increase linearly with Mg content and the resistivity to increase

exponentially, though the n-type conductivity is maintained.[17, 18] Other properties can be

altered as well. For instance the resistance to acid dissolution (see Chp. 4). Other alloys of

ZnO have been explored as dilute ferromagnetic oxides, where Zn is exchanged for impurity

atoms intended to induce ferromagnetism, such as Zn

Ni

O, Zn

Mn

O, Zn

Co

O, and

Zn

Fe

O.[19]

The films of ZnO and Zn

Mg

O used in this thesis are produced by a sol-gel process

(see Section 2.1). This process produces a very thin film of thickness 25-50 nm, as measured

by profilometry. Atomic force microscopy height images of the surface of these ZnO and

Zn

Mg

O films are shown in Figure 1.2 below. Analysis of our AFM images show that the

films are composed of grains that are approximately 40 nm in size. X-ray diffraction spectra

(given in Chp. 4) show that the crystal structure of these grains is the standard wurtzite

structure for films up to 30% Mg. UV-Vis spectra of these films are shown in Figure 1.3 and

bandgaps calculated from these spectra are given in Table 1.1.

1.2

Organic and Fullerene Semiconductors

Organic semiconductors are synthesized through the techniques of organic chemistry.

They include polymers, small molecules, and dyes. Some examples are shown in Figure 1.4.

There is also a class of semiconductors formed from fullerene derivatives in which fullerenes

Figure 1.2: Atomic force microscopy height images of Zn

Mg

O films produced by a sol gel

process [18] for x = 0, 0.1, 0.2, 0.3. Images for x > 0 taken by Erich Meinig.

Table 1.1: Bandgaps of Zn

Mg

O films calculated from UV-Vis spectra.

Mg Content (mol. %)

Bandgap (eV)

0

3.24

5

3.33

10

3.43

15

3.52

20

3.62

30

3.77

Figure 1.3: UV-Vis absorption spectra of the thin films of Zn

Mg

O studied in this thesis.

It is apparent that the bandgap increases with Mg content.

(C

, C

, C

, etc.)

are modified with an attached functional group that changes the

fullerene’s electronic properties or makes it soluble.[20] One of the most commonly used is

phenyl-C61-butyric acid methyl ester (PCBM) shown in Figure 1.4. For the sake of simplicity,

the term ’organic semiconductor’ includes fullerenes in this thesis.

The physics of carrier transport and generation in these materials is highly unusual

com-pared to inorganic semiconductors and is extremely important to the design of photovoltaic

devices employing organic semiconductor materials. There are several properties that make

these materials unusual: their low dielectric constants, their weak intermolecular

interac-tions, and their very high absorption coefficients.

The very high absorption coefficients of organic materials compensates for their non-ideal

transport properties (discussed below).[25] While many inorganic semiconductors require

micrometer-thick films to achieve complete light absorption, organic materials can generally

achieve ∼95% light absorption above their absorption onset in films < 300 nm thick.[25]

Thus devices can be extremely thin, making it possible to extract carriers even without

Figure 1.4: Examples of common electro-active organic polymers and small molecules. P3HT

= poly-3-hexylthiophene. F8BT = poly(9,9-dioctylfluorene-alt-benzothiadiazole). CuPc =

copper phthalocyanine. PCBM = phenyl-C61-butyric acid methyl ester. P3HT and F8BT

images are from Ref. 21. N3 Dye is from Ref. 22. PCBM is from Ref. 23. CuPc is from

Wikipedia (Ref. 24).

great transport properties. (Dye sensitized solar cells are usually much thicker because of

their unique architecture, discussed in Section 1.4) However, one of the major challenges of

organic semiconductors has been the design of materials with a low energy absorption onset

(low bandgap in inorganic semiconductor parlance).[25] The initial materials developed all

had high energy absorption onsets (≥2 eV) which is much larger than the ideal threshold

energy of ∼1.3eV for terrestrial solar cells. The design of materials with a lower absorption

onset is a subject of ongoing research.

The relative dielectric constant (ε) of a typical organic semiconductor is ε = 4 while that

of an inorganic semiconductor is ε > 10.[26] ε is inversely proportional to the strength of

the electric field within the material. Thus, the coulomb attraction between an electron and

its corresponding oppositely charged hole is reduced by a factor of ε in a material. If the

electron-hole pair is approximated as a hydrogenic system, the average orbital radius of the

electron is increased by a factor of ε. The screening and orbital radius increase defined by

ε scale the electron-hole binding energy by 1/ε

_{. Because of this, the binding energy of a}

photo-generated electron-hole pair in an organic semiconductor will be increased around an

order of magnitude compared to inorganic semiconductors.

The weak interactions between molecules in an organic semiconductor prevent them from

forming strongly bound crystalline structures as are found in inorganic semiconductors. This

prevents the material from forming a band structure and carriers are localized to regions on

the order of one molecule in a small molecule system or several monomers of a polymer chain,

instead of being delocalized as in inorganic semiconductors. The localization of carriers

and the strong electron-hole interaction means that excitations in the material at ambient

conditions generate bound electron-hole pairs known as excitons, with binding energies much

greater than 0.1 eV, and, hence, much greater than the average thermal energy of around

kT = 0.026 eV at 300 K (k – Boltzmann constant, T – temperature). This is in contrast to

inorganic semiconductors which have exciton binding energies of ∼ kT or less, so that the

average thermal energy is enough to split the exciton and generate free carriers. Exciton

formation has important implications for organic solar cells because the excitons generated

by light absorption must first be split before any current can be generated. This is typically

accomplished by introducing a heterointerface, as discussed below, although excitons can

also split in other ways such as at a defect in the organic.

The localization of carriers also prevents band-like transport in organic semiconductors.

Instead, both excitons and free carriers travel through the material via a hopping transport

mechanism.[27, 28] Excitons, which are neutrally charged, can only be transported through

diffusion. The diffusion length, or how far an exciton can diffuse on average before

recombin-ing, has an upper limit of 10-20 nm in most organic semiconductors used in photovoltaics.

This has a significant impact on photovoltaic device design, discussed in Section 1.4.

Free carriers take the form of polarons, or charges coupled to phonon modes.[29] The basic

idea is that the presence of the free charge causes a relaxation of the surrounding atoms.

This relaxation can be significant in organic semiconductors because these materials have

relatively labile bonds that distort easily, as compared to inorganic semiconductors. This

distortion of molecular structures follows the charge as it moves through the material. Free

carrier transport is also dramatically influenced by the degree of order of the molecules in

the film and the overall film morphology. Grain boundaries, stacking and packing geometry,

and crystallinity can all affect transport dramatically.[30–32]

A distinction must be made about electronic structure in organic semiconductors

com-pared to inorganic semiconductors. Because of weak intermolecular interactions, there are

no valence band or conduction band states. Instead, transitions occur between localized

molecular orbitals defined by the structure of the molecule and its interaction with

neigh-boring molecules. The minimum energy electronic transition is therefore between the highest

energy occupied molecular orbital (HOMO) and the lowest energy unoccupied molecular

or-bital (LUMO). These are analogous to, but distinctly different from, the valence band and

conduction band of crystalline inorganic semiconductors. In a well ordered organic material,

interactions between molecules can give rise to multi-molecule excitations that are lower in

energy than the HOMO-LUMO gap of the isolated molecules. This can be observed, for

in-stance, in well ordered P3HT, and can induce significant changes in the absorption spectrum,

in addition to the impact on charge transport mentioned above.[33]

Organic semiconductors often form the active layer of the photovoltaic devices they are

used in, meaning they are often absorbing light and separating charge. Because of the

importance of excitons in these devices, they are often referred to as excitonic devices.

In these devices, it is necessary to split the exciton before the rest of the photovoltaic

process can occur. The mechanism that typically accomplishes this in these devices is an

ultra-fast charge transfer process at the interface between two organic semiconductors or an

organic semiconductor and inorganic semiconductor. The material that gives up an electron

is referred to as a donor and the material that receives it is referred to as an acceptor. This

charge transfer occurs at the interface between two carefully matched materials. A minimum

condition for charge transfer is that the free energy gained in the charge transfer process must

be high enough to overcome the exciton binding energy. There must be some energy offset at

the interface of these materials to allow this to occur, for instance a difference in the LUMO

energy of the two materials. In standard semiconductor physics, this is referred to as a type

II heterojunction.[34] The discovery of ultra-fast electron transfer from a semiconducting

polymer or small molecule (the donor) to a fullerene derivative (the acceptor) is exploited

in nearly all organic photovoltaic devices to date.[35] The charge transfer rate from polymer

to fullerene has been measured to be less than 50 fs, far faster than any recombination

process.[36] Similarly, charge transfer from a dye molecule attached to a metal oxide scaffold

in a dye sensitized solar cell shows a fast charge transfer component occurring in less than

100 fs, again several orders of magnitude faster than any recombination processes, and a

slower picosecond process.[37]

1.3

General Photovoltaic Device Physics

Excitonic solar cells have a number of specialized requirements that differentiate them

from inorganic solar cells. All of these design requirements derive from the property of

organic semiconductors, discussed in Section 1.2, that light absorption generates excitons

instead of free carriers. This requires a specialized architecture for generating free carriers,

but upon free carrier generation the same general theory used for inorganic solar cells can be

applied to excitonic devices. For this reason a general discussion of solar cell device physics

is provided first, followed by discussion of the new ideas required to understand excitonic

devices in Section 1.4.

Solar cells are essentially layered structures composed of conducting and semiconducting

materials. In isolation, each of these materials has a thermal equilibrium distribution of

electrons and holes. This thermal distribution can be described by an absolute chemical

potential energy referred to as the Fermi energy, E

, in semiconductor physics. Further

explanation about how the Fermi energy describes semiconductors in equilibrium can be

found in the introductory text by Neamen.[34] The chemical potential energy provides a

measure of the potential for charge carriers to diffuse. So when the isolated materials are

brought together in a solar cell, charge will diffuse between them according to their differences

in E

until a new equilibrium is established and E

is constant across the whole device.

However, the redistribution of charge sets up electric fields in the device that can be described

by the gradient of the familiar electric potential, Φ. The equilibrium net electric potential

across the device is referred to as the built-in potential, V

. This process is illustrated in

Figure 1.5 below for a standard pn-junction and for a metal-insulator-metal junction, which

is commonly used as a qualitative model of organic photovoltaic devices.[38]

Figure 1.5: Process of Fermi level equilibration in junctions of electronic materials. (a)

Stan-dard pn-junction equilibrium. Before contact (i), the p and n-type materials have differing

Fermi levels E

and E

. After the junction is formed (ii), the Fermi levels equilibrate by

charge transfer and band bending to achieve a constant Fermi level, E

. During this process

a built-in electric field develops, with potential drop V

. (b) In the metal-insulator-metal

model of organic solar cells, the active layer is approximated as insulating. Before contact

(i), the metal contacts have differing Fermi levels. After contact (ii), the Fermi levels of the

metals equilibrate through metal-to-metal charge transfer. This charge transfer generates a

built-in field and potential across the insulating active layer.

The photovoltaic effect is the generation of a current in a device when light is incident

on it, without the need to apply an external voltage, also implying the generation of a

photovoltage. In order for current to be produced, there must be an asymmetry in the

device that drives the electrons and holes generated by the light in opposite directions. In

solar cells this can be accomplished in many ways, for instance the donor/acceptor structure

in excitonic devices, discussed further below, or the existence of a built-in potential V

,

as in a conventional pn-junction solar cell. When light is incident on the device, thermal

equilibrium is disrupted, and a non-thermal distribution of carriers results. The thermal

distribution of carriers described by E

is no longer valid. Excess electrons and holes are

generated in the conduction and valence states, respectively. Luckily, these carriers generally

equilibrate to a thermal-like distribution in their respective states at a rate much faster than

they recombine with the opposite carrier.[39] This makes it possible to define a quasi-Fermi

(quasi-thermal) energy for both electrons and holes separately [26]:

E

= E

+ kT ln(

n

N

)

(1.1)

E

= E

+ kT ln(

p

N

)

(1.2)

E

and E

are the quasi-Fermi energies for electrons and holes respectively. E

and E

are the energies of the conduction states (c), occupied by electrons, and valence states (v),

occupied by holes, of the materials in the device. n and p are the electron and hole

con-centrations, respectively, in the conduction and valence states. N

and N

are the density

of states of the lowest energy conducting states and highest energy valence states. These

equations are intended to be generally descriptive of either organic or inorganic

semiconduc-tors. However, the interpretations of E

, E

, N

, and N

are complicated by the presence

of disorder in a material which spreads the density of states out so that these quantities

are less clearly defined. This disorder has significant implications for excitonic solar cell

performance.[40] Each quasi-Fermi energy is actually composed of the familiar potential

en-ergies already discussed above. Taking E

as an example, E

is a function of position and

follows the variations in the electrostatic potential energy eΦ through what is commonly

re-ferred to as band-bending.[34] The term kT ln(n/N

) captures the chemical potential because

it is a function of electron concentration, n, which varies with position.

The quasi-Fermi levels E

and E

really define the behavior of the device. They shift

when an external load is applied that alters the electrostatic potential. They shift as the

intensity of light changes or if the rate of electron and hole recombination goes up. The

gradients of E

and E

define the forces on the electrons and holes, respectively, and

therefore define the current in the device [26, 34]:

J = nµ

∇E

+ pµ

∇E

(1.3)

or equivalently,

J = −e (nµ

+ pµ

)∇Φ + ekT (µ

∇n − µ

∇p)

(1.4)

J is the current density, e is the fundamental unit of charge, p and n are the hole and

electron volume densities, respectively, k is Boltzmann’s constant, T is temperature, µ

and

µ

are the hole and electron mobilities, respectively, Φ is the electric potential, and ∇ is the

gradient operator. These equations can be derived from the relaxation time approximation

of the Boltzmann transport equation (or more simply the Drude model) and the Einstein

relation, which has been employed here to give the electron and hole diffusion coefficients as

kT µ

.[34, 39] The values of E

and E

at the contacts to the device define the amount

of usable energy available from a pair of carriers, and therefore the voltage (V ) of the device

[26]:

eV = E

(e

contact) − E

(h

contact)

(1.5)

These ideas make it possible to understand the meaning of the performance metrics

reported for solar cells. These parameters are illustrated in Figure 1.6, where an example

current density-voltage (J-V) curve is shown. The current density J is the current per unit

area flowing through the device, defined by Equations 1.3 and 1.4. V is the voltage across

the load on the device. The voltage generated in the solar cell, defined by Equation 1.5, will

match the voltage drop across the load, V , by Kirchoff’s law.

Figure 1.6: Example of a current density (J) - voltage (V) curve for a solar cell. The

important performance metrics are shown.

The shape of the J-V curve is due to the fact that good solar cells are rectifiers, only

allowing current to flow in one direction. This is a consequence of the asymmetric, charge

separating character of the device that allows photovoltaic action, discussed above. When

the load is zero, the device has been shorted. Though V = 0, gradients in the quasi-Fermi

energies still exist that drive the carriers, resulting in a photo-generated current. This is

known as the short circuit current, J

. At the other extreme, when J = 0, the circuit is

effectively open and the voltage under this condition is referred to as the open circuit voltage

(V

). Because the current switches direction at this point and power cannot be extracted

from positive currents, eV

represents the highest energy per electron that can be collected.

V

is highly sensitive to a number of very important processes occurring in the device.

Whereas J is sensitive to gradients in the quasi-Fermi energies, V is sensitive to their actual

values. Combining Equations 1.1, 1.2, and 1.5 and assuming that J = 0, we find

eV

=

E

+ kT ln(

n

N

)

−

E

− kT ln(

p

N

)

(1.6)

The brackets indicate which contact each of the terms should be evaluated at. The first thing

to notice is that V

depends on the difference in the energy of the charge transporting states

of each of the contacts, E

(e

contact)−E

(h

contact). This makes it clear that the electronic

structure of the contacts to a solar cell, and its interaction with the active layer, can be quite

important. This will be the focus of Chp. 3. Throughout the device, the conducting state

energies, E

and E

, define a ’baseline’ for the quasi-Fermi level and their difference should

be as large as possible in order to maximize V

. The carrier concentrations n and p also play

a significant role in determining V

. n and p are influenced by carrier generation processes

(light absorption and exciton splitting in excitonic solar cells) and carrier recombination

processes and thus V

is a sensitive measure of both. The higher n and p, the better, and

eliminating recombination processes that reduce the carrier concentration is an important

strategy for maximizing current and voltage, a somewhat counter-intuitive result, but a very

important one.[41] The carrier concentration terms in Equation 1.6 also suggest that a V

can be generated even if there is no difference in the energy of the contacts, purely through

a diffusion force.[26, 42] Theoretical calculations also suggest that a diffusional force can

counteract a loss of energy due to non-optimal contacts in bilayer organic solar cells.[43]

Along the J-V curve, between V = 0 and V

, there is a maximum power point (MPP)

with P

= J

V

(Figure 1.6), where P is the power delivered by the solar cell.

Note that this is a power density (power per unit area). The MPP can be used to calculate

the efficiency of the device (η) from:

η =

P

I

(1.7)

where I

is the power available from the incident light. Two rectangles are shown in

Figure 1.6, one representing the actual maximum power generated, with area P

, and

one representing the ideal maximum power, with area J

V

. The ratio of these is called the

F F =

P

J

V

(1.8)

The F F is a quantification of how closely a device approaches ideal power generation, and

a quantification of how current decreases as the load (bias) on the device increases. The

F F characterizes the recombination processes that are manifested as the load on the device

is increased. The loss processes that contribute to the F F , and to device performance in

general, are a topic of current research in excitonic solar cells.[44, 45]

In order to understand the discussion of device performance in Chp. 3 and in the scientific

literature in general, it’s important to understand the equivalent circuit model of a solar cell

shown in Figure 1.7. The device is modeled in terms of three elements: a rectifying junction

(or diode, D), a resistance in parallel, the shunt resistance (R

), and a resistance in series

(R

). The diode is modeled by the ideal diode equation, sometimes with modifications for

non-ideality.[34] The photocurrent is modeled as a voltage-independent current source I

in parallel with the diode. The circuit is completed by the load, which has a voltage drop

V across it (implying a bias of V across the solar cell). The major advantage of this model

is that it provides a simple way of incorporating loss processes into a solar cell model. All

the device performance metrics discussed above (Figure 1.6) can be derived from a standard

circuit analysis. The loss processes are incorporated through R

, which represents the Ohmic

losses accrued from resistance within the device, and R

, which represents the carrier

concentration losses due to recombination and any shorts that might exist in the device.

These losses can be approximated from experimental J-V curves and are often reported in

the literature. In Chp. 3, R

was calculated from the slope of the J-V curve at V = 1V and

R

was calculated from the slope at V = 0V. While this equivalent circuit model is simple

and easy to use for achieving a general understanding of experimental data, more complex

modeling is required to accurately describe the J-V curves that are encountered in real solar

cells.

Figure 1.7: Equivalent circuit model of a solar cell. The device is composed of a

voltage-independent current source, I

, due to the incident radiation, a rectifying junction (D),

a shunt resistance (R

) representing shorting and recombination processes, and a series

resistance (R

) representing Ohmic losses in the device. The device is attached to a load

with voltage drop V across it.

1.4

Excitonic Solar Cell Device Physics

The preceding concerns the behavior of free carriers in solar cells. However, generation

of free carriers upon light absorption is not guaranteed. Excitonic solar cells employ

mate-rials in which light absorption generates excitons instead of free carriers. This requires the

architecture of these devices to deviate from that of conventional solar cells. As discussed

in Section 1.2, it is necessary to build the active layer from two different materials: a donor

and acceptor. The materials are chosen in order to allow excitons to efficiently dissociate

into free carriers at the interface of these two materials. There are four fundamental steps

in the operation of an excitonic solar cell (Figure 1.8)[46]:

1. Generation: Excitons are generated by light absorption.

2. Diffusion: Excitons diffuse to an interface.

3. Exciton Dissociation: the exciton is split by charge transfer across the interface (as

discussed in Section 1.2), and the charges are free to move.

4. Charge Collection: charges are driven by concentration gradients and built-in electric

fields to the contacts where they are collected and used to power a load.

Figure 1.8: The operational steps of excitonic solar cells. Figure based on Ref. 46. The

electron and hole reside below the energy of the HOMO-LUMO optical gap in steps 1-3 due

to their exciton binding energy.

In order to maximize exciton dissociation and light absorption simultaneously an

ar-chitecture unique to excitonic photovoltaics has been developed. The donor and acceptor

must be intimately blended at the nanoscale so that excitons can diffuse to an interface

without recombining. The exciton diffusion length of ∼10 nm defines the length scale of

this blending.[46, 47] This design allows the film to be optically thick while still allowing for

efficient exciton dissociation. Furthermore, a percolation pathway must exist in the donor

and acceptor material phases that allows free carriers to leave the active layer efficiently.

Examples of the specialized architecture required are shown in Figure 1.9. In Figure 1.9(a),

the polymer P3HT is combined with PCBM in a ’bulk heterojunction’ in which domains of

P3HT and PCBM form a large interface area for exciton dissociation. Figure 1.9(b) shows

a hybrid nanowire ZnO/P3HT photovoltaic device. The high interface area created by the

P3HT/nanowire interface is intended to efficiently split excitons. However, careful

investiga-tion of these devices suggests that exciton dissociainvestiga-tion at the metal oxide/organic interface

is very inefficient, and that most of the free carriers generated are due to exciton splitting

at defects in the P3HT.[48] Figure 1.9(c) shows a dye sensitized solar cell (DSSC). In these

devices a light-absorbing dye is chemically bound to a high surface area metal oxide such

as a porous matrix or nanowires.[37] An exciton generated in the dye is efficiently split by

charge transfer to the metal oxide. This design eliminates step two, exciton diffusion, in

the operation of an excitonic solar cell. High light absorption is achieved because of the

extremely high surface area of the porous metal oxide matrix.

Figure 1.9: Examples of excitonic photovoltaic device architectures. (a) Inverted bulk

hetero-junction solar cell. (b) Metal oxide/organic nanowire solar cell. (c) Nanowire dye sensitized

solar cell (DSSC). The figures are meant to be illustrative and are not drawn to scale.

After the exciton has been split, free carriers (polarons) are generated. From Section 1.3,

the two processes that drive carriers to the contacts during charge collection are drift and

diffusion. Diffusion plays an important role in excitonic solar cells because exciton

dissocia-tion creates a high concentradissocia-tion of each carrier type in physically separated regions.[26] The

concentration gradient developed at the exciton-splitting interface causes carriers to diffuse

away from it toward the contacts, aiding in charge separation and transport. Thus, in these

devices the energy offset between donor and acceptor helps to drive efficient charge

separa-tion and transport, in addisepara-tion to separating excitons. The role of drift in charge transport

differs between different device designs. In bulk heterojunction and metal oxide/organic

heterojunction solar cells, fields contributing to drift can only really originate in the Fermi

level difference between the contacts. The reason for this is that, while electric fields may

exist locally along the donor-acceptor interface, the complex, interdigitated architecture of

these devices, along with low carrier concentrations in a very thin active layer, rules out the

possibility of a built-in electric field being established by the active layer itself. However,

differences in the Fermi level of the contact materials can establish a built-in potential V

that extends across the active layer, contributing to drift. In DSSCs the situation is

substan-tially different. The liquid electrolyte permeating the active layer will completely cancel any

external electric fields and in this case carrier transport is governed entirely by diffusion.[42]

There are a number of unsolved questions about the physics of excitonic devices, especially

in the newer organic photovoltaic devices. As mentioned above, recombination in these

devices is an area of active research.

The origin of the open circuit voltage is still an

open question as well. A number of studies have established that the energy difference

between the HOMO of the donor and the LUMO of the acceptor in optimized devices has

a 1 V/eV influence on V

.[18, 49, 50] However, the influence of the contacts on V

in these

devices is not well understood. The discussion in Chp. 3 reviews this question extensively.

Another area of recent interest in organic photovoltaics is contact selectivity.[51] In order

to prevent recombination at the contacts and retain carrier density in the active layer, it

is best if each contact selectively passes or transports only one carrier type. This is of

general interest to all of photovoltaics, however the impacts of it on organic solar cells are

just beginning to be studied.[52] These initial results suggest that charge selective contacts

create an enhancement in V

, likely through maintaining higher carrier densities near the

contacts. Metal oxide contacts such as n-type ZnO or p-type NiO are expected to be selective

because the materials are intrinsically doped to transport one carrier type, and large energy

barriers exist for the extraction of the opposite carrier (see Section 1.5). ITO, on the other

hand, is not expected to be selective because of its metallic character.

1.5

Metal Oxide/Organic Interfaces

Metal oxides are employed in organic electronic devices as contacts to the organic active

layer of the device (as in a bulk heterojunction solar cell) or as a component of the active

layer itself (as in hybrid photovoltaic devices or dye sensitized solar cells).[37, 51, 53] The

performance of the interface between these two layers is critical to overall device performance,

and understanding and optimizing these interfaces is an important research direction in

excitonic photovoltaics and electronics in general. There are three major areas of interest:

1. Morphology of the organic layer at the interface. Does the organic phase wet the metal

oxide, and how well does the organic phase order near the interface with the metal

oxide?

2. Energy level alignment: how do the energy levels of each component shift after the

interface is formed and how do the orbitals/bands of each component interact with

each other? Are there energy barriers formed that could inhibit charge transfer?

3. Chemical reactivity: do chemical reactions occur between the organic and the metal

oxide, and how does this affect interface performance?

Examples of these ideas are illustrated in Figure 1.10. Figure 1.10(a) shows how the metal

oxide/organic interface can impact the morphology of the organic, potentially leading to poor

charge transport. This has been observed in the case of ZnO/P3HT interfaces, where P3HT

near the interface shows a blue shift in its absorbance spectrum due to reduced interchain

interactions.[54, 55] In Figure 1.10(b), holes are unable to transfer from the organic to ZnO

due to the large difference in energy between the organic HOMO energy and ZnO’s valence

band energy. On the other hand, electrons can transfer easily. In this case, the electronic

structure makes the ZnO electron-selective.[51] Figure 1.10(c) shows how dye molecules can

be chemically attached to a metal oxide through a carboxylate bond. In this case, a chemical

bond has been purposefully introduced to anchor the dye close to the surface, allowing charge

transfer and photovoltaic action to occur. In other cases, unexpected chemical reactions

between the interface components may be detrimental.

Figure 1.10: Examples illustrating metal oxide/organic semiconductor interface properties

important to organic photovoltaic device performance. (a) The organic morphology may

become disordered at the interface, preventing efficient charge and exciton transport near the

interface. (b) Interface electronic structure can give charge selective contacts. (c) Chemical

bonding of dye to metal oxide makes efficient charge transfer possible.

The electronic structure of the interface is a major topic of Chp. 3 and will be the

focus of the rest of this section. Figure 1.11 shows the electronic structure of an isolated

n-type metal oxide (n-MO), an organic, and two potential interfaces in more detail. Energy

levels in materials are often referenced to what is referred to as the vacuum level or vacuum

energy, shown as the high energy solid black line in Figure 1.11. This is the energy of a free

electron that has just barely escaped the potential of the material, and resides just outside

the material.[56] Theoretically a vacuum level can be defined within a material, but it is

not experimentally accessible. It is a useful reference, however, because it tracks changes in

the electrostatic potential which a free electron will still experience. The energy required

to extract an electron from the Fermi energy at the surface of the material (E

) to

this vacuum level is called the work function (φ in Figure 1.11) and is a useful quantity

for comparing materials because differences in the work function indicate differences in the

chemical potential of the carriers within each material (Section 1.3), and because it can be

measured experimentally. Also important are the differences in energy between conducting

states on either side of the interface. In the case of Figure 1.11, this is the difference between

the LUMO level of the organic and the conduction band of the metal oxide (E

). When

comparing isolated materials (Figure 1.11(a)), this gives an estimate of what the energy

barriers to charge transfer will be.

Figure 1.11: Electronic structure of an n-type metal oxide and an organic in isolation (a),

after an interface is formed under assumption of vacuum level alignment (b), and after an

interface is formed in which there is a dipole-forming charge transfer or chemical interactions

between the two components (c). n-MO means n-type metal oxide, φ is the work function

of the interface, ∆φ is a dipole-induced work function change, CB is conduction band of the

metal oxide, E

is the Fermi energy, E is energy, E

is the energy difference between the

LUMO of the organic and the conduction band of the metal oxide.

Upon actual formation of an interface, non-idealities are expected. It is useful to first

discuss the ideal situation. The ideal energy level alignment, which is often depicted in the

excitonic photovoltaics literature, is shown in Figure 1.11(b), and is referred to as vacuum

level alignment. In this situation, the vacuum level remains constant as in the case of isolated

films, and the work function of the interface is the work function of the substrate. Vacuum

level alignment has two major assumptions: (1) the materials are non-interacting, i.e. there

are no chemical reactions between them and (2) the organic behaves like an insulator, and

charge transfer between the organic and substrate is inhibited.[57] These approximations are

surprisingly good in metal and metal oxide/organic interfaces where the metal oxide work

function is in the HOMO-LUMO gap of the organic.[57–59] Braun et al. suggest charge

transfer between the organic and substrate is inhibited by a charge-blocking contamination

layer due to sample preparation in an ambient environment.[57] However, Greiner et al. have

prepared metal oxide/organic interfaces under ultra high vacuum conditions by evaporating

an organic layer on top of a metal oxide layer oxidized within the chamber.[59] They still

observe vacuum level alignment and propose an alternative explanation in which the

molec-ular nature of the organic prevents charge transfer unless it is thermodynamically stable

for a molecule to be charged.[59] Explaining the vacuum level alignment observed in these

experiments is still a matter of fundamental research.

Violation of (1) and (2) invariably induces an interface dipole that can significantly alter

the energy level alignment at the interface compared to what would be predicted in the

vacuum alignment case. This situation is depicted in Figure 1.11(c). The dipole can be

approximated as two infinite sheets of charge. A simple application of Gauss’ law shows

that this arrangement has an electric field of zero on either side of the dipole and a constant

electric field between the charges, leading to a linear step in electric potential energy across

the dipole (∆φ). Whether the step raises or lowers the energy depends on the direction of

the dipole. This dipolar energy step is manifested as a shift in the vacuum level, as even

free electrons will experience this dipole. Electrons leaving the material must traverse this

dipole and the work function of the film changes by an amount ∆φ which is the potential

energy shift due to the dipole. Consequently E

is shifted by the same amount. One of

the most dramatic observations of charge transfer and subsequent dipole formation at the

interface is referred to as Fermi level pinning. Fermi level pinning occurs when a high density

of states are present in one material at an interface. In this case, charge transfer occurs to

or from these states until the Fermi energy reaches the energy of these states.[60] In metal

oxide (and metal)/organic interfaces, this is observed when the Fermi level of the metal

oxide approaches either the LUMO or HOMO level of the organic, where a high density of

states is available.[57–59] The Fermi level of the interface becomes pinned to the HOMO or

LUMO level and doesn’t change no matter how high or low, respectively, the work function