AN ABSTRACT OF THE DISSERTATION OF
Heather A. S. Platt for the degree of Doctor of Philosophy in Chemistry presented
on March 3, 2010.
Title: Copper and Iron Chalcogenides for Efficient Solar Absorption
Abstract approved:
Douglas A. Keszler
This dissertation outlines two approaches for improving the efficiency and reducing
the cost of photovoltaic energy generation. First, the structures of known binary
copper and iron compounds are used to choose the promising compositions Fe2 XS4
(X = Si, Ge), Cu3 TaQ4 (Q = Se, Te), and BaCuQ’F (Q’ = S, Se, Te). The syntheses of
these materials are described, and their optical and electrical properties are discussed.
The second approach is to utilize solution deposition as a method that can cover large
areas with thin films very quickly. Two simple aqueous precursors are spin coated to
produce Fe oxides. These thin films are reacted with flowing H2 S (g) to convert to the
promising absorber FeS2 pyrite, and then characterized optically and electrically.
c
Copyright
by Heather A. S. Platt
March 3, 2010
All Rights Reserved
Copper and Iron Chalcogenides for Efficient Solar Absorption
by
Heather A. S. Platt
A DISSERTATION
submitted to
Oregon State University
in partial fulfillment of
the requirements for the
degree of
Doctor of Philosophy
Presented March 3, 2010
Commencement June, 2010
Doctor of Philosophy dissertation of Heather A. S. Platt presented on
March 3, 2010.
APPROVED:
Major Professor, representing Chemistry
Chair of the Department of Chemistry
Dean of the Graduate School
I understand that my dissertation will become part of the permanent collection
of Oregon State University libraries. My signature below authorizes release of
my dissertation to any reader upon request.
Heather A. S. Platt, Author
ACKNOWLEDGEMENTS
Few experimentalists can complete advanced degrees without the assistance of others,
and I am not one of those few. I want to begin by thanking Doug for fostering an
environment where creative, independent people can explore new ideas while being
challenged to stay grounded in fundamentally sound science. I also appreciate his
commitment to choosing practical problems that have the potential to positively
impact the lives of everyone. Jeremy Anderson was an invaluable compass for gauging
the feasibility of projects large and small. I did not anticipate how much I would learn
about building equipment during my graduate work, and Jeremy usually managed to
keep a straight face no matter what I asked him. Between the two of them, Stephen
Meyers and Jason Stowers set the curves in every class I took with them, so it was
a bit of a relief to finally work with them. I appreciate Stephen’s grasp of both
chemistry and the English language, and Jason has an amazing knack for breaking
complex problems down into manageable pieces. Pete Hersh and I worked together
applying solid state chemistry to new materials development for solar technology, and
I am grateful to him for many useful and interesting conversations.
Someday, I hope to be a little like Janet Tate. She is an excellent co-author,
and I will also never forget her encouragement to pursue projects for their technical
merit. Robert Kykyneshi and Andriy Zakutayev have taught me a lot about solid
state physics, and I especially want to thank them for their patience with all of my
questions. I have learned many useful things from John Wager about solar cell design
and the challenges of fabricating a working solid state device. I am grateful for these
insights, and I will always think of him when I see energy band diagrams. Mike Lerner
and Philip Watson have also made valuable contributions to my graduate work, both
formally in the classroom and outside it as members of my committee. I thank them
for their time and acumen.
I have enjoyed working and learning with many other past and current members
of the Keszler, Tate, and Wager research groups including Cheol-Hee Park, Michael
Hrushka, Joa-Young Jeong, Paul Newhouse, Kai Jiang, Bahar Ozmen, Chris Knutsen,
Sharon Betterton, Alan Telecky, Wei Wang, Nutch Jieratum, Annette Richard, Josh
Russell, Jack Spies, and Richard Schafer. I thank them for making my time in the
lab much more enjoyable. I also want to thank Andrew Smith and Sean Muir of the
Subramanian group for their help with the occasional odd experiment.
In addition to my work in the lab, I have also appreciated opportunities to hone
my love of teaching. I spent many hours with Richard Nafshun, James Krueger, and
Michael Schuyler during the Chemistry Department’s Teacher Mentoring seminar
talking about this fascinating subject. I particularly want to thank Dr. Nafshun
for mentoring me in the specifics of syllabus writing, lecture prep, and exam writing
during one of his general chemistry courses.
And finally, I could not have completed my graduate work without John Donovan, Ted Hinke, and my husband, Andy Platt. John runs the Microprobe Lab at
CAMCOR, and I benefited greatly from his tireless efforts to ensure that everyone
who brings him samples gets the best data possible. Every time I went to talk with
Ted, the Chemistry Department’s machinist, I came away with a better design than I
went in with. I am grateful for his attention to detail in his work and his willingness
to take a little extra time to make sure he understood exactly what I needed. While I
cannot list all the ways that Andy truly contributed to the completion of my degree,
I do want to specifically thank him for finding the missing units that I lost in a few
pesky calculations and providing invaluable help with all sorts of computer bugs.
CONTRIBUTION OF AUTHORS
Vorranutch Jieratum grew single crystals and collected optical transmission data from
them for Chapter 2. Robert Kykyneshi performed the electronic structure calculations in Chapter 2, and Wei Wang calculated the electronic structures in Chapter
5. Andriy Zakutayev collected luminescence, photoemission, electrical and optical
data for Chapters 2, 3, 5, and 6. Jason Stowers collected SEM images of the films in
Chapters 3 and 4, and Alan Telecky collected X-ray reflectivity data from the films in
Chapter 4. Josh Russell collected optical data from the films in Chapter 4. Annette
Richard synthesized some of the powders in Chapter 5. Janet Tate provided valuable
interpretation for all of the luminescence, photoemission, electrical and optical data,
as well as the electronic structure calculations. Douglas Keszler contributed to all
aspects of this work.
TABLE OF CONTENTS
Page
1. Introduction ....................................................................................
1
1.1
Importance of electrical energy...................................................
2
1.2
Efficient solar energy absorption .................................................
3
1.3
Leading absorber materials........................................................
5
1.4
Prediction of electrical and optical properties from crystal structures .
7
1.5
Precursor chemistries for spin coating and flowing gas reactions ........
8
1.6
Summary and Perspective ......................................................... 10
2. Optical and electrical properties of Fe2 XS4 (X = Si, Ge)........................... 17
2.1
Introduction ........................................................................... 18
2.2
Experimental..........................................................................
2.2.1 Synthesis .....................................................................
2.2.2 Characterization ............................................................
2.2.3 Band Structures ............................................................
20
20
20
21
2.3
Results.................................................................................. 22
2.3.1 Experimental results....................................................... 22
2.3.2 Electronic Structures ...................................................... 24
2.4
Summary and Perspective ......................................................... 27
3. Spin coated iron hydroxo-oxalate and conversion
to FeS2 pyrite .................................................................................. 39
3.1
Introduction ........................................................................... 40
3.2
Experimental..........................................................................
3.2.1 Precursor synthesis ........................................................
3.2.2 Thin film preparation and anneals.....................................
3.2.3 Thin film characterization ...............................................
41
41
42
42
3.3
Results..................................................................................
3.3.1 Precursor chemistry and film crystallization ........................
3.3.2 Morphology and stoichiometry..........................................
3.3.3 Electrical and optical properties........................................
44
44
45
47
3.4
Summary and Perspective ......................................................... 47
TABLE OF CONTENTS (continued)
Page
4. Conversion of Fe hydroxide nitrate thin films to FeS2 pyrite ...................... 58
4.1
Introduction ........................................................................... 59
4.2
Experimental Methods ............................................................. 60
4.3
4.4
4.2.1
Precursor synthesis and characterization ............................. 60
4.2.2
Thin film preparation and analysis .................................... 61
Results and Discussion ............................................................. 62
4.3.1
Bulk precursor chemistry................................................. 62
4.3.2
Film morphology and stoichiometry................................... 65
4.3.3
Film electrical and optical properties ................................. 67
Summary and Perspective ......................................................... 69
5. Electronic structure calculations of Cu3 TaQ4 (Q = Se, Te) and photoemission studies of Cu3 TaSe4 .................................................................... 80
5.1
Introduction ........................................................................... 81
5.2
Experimental.......................................................................... 82
5.3
5.4
5.2.1
Synthesis ..................................................................... 82
5.2.2
Characterization ............................................................ 82
5.2.3
Band Structures ............................................................ 83
Results.................................................................................. 83
5.3.1
Experimental results....................................................... 83
5.3.2
Electronic Structures ...................................................... 86
Summary and Perspective ......................................................... 88
6. Synthesis of BaCuQF (Q = S, Se, Te)................................................... 98
6.1
Introduction ........................................................................... 99
6.2
Experimental.......................................................................... 100
6.3
6.4
6.2.1
Synthesis ..................................................................... 100
6.2.2
Characterization ............................................................ 101
Results.................................................................................. 102
6.3.1
Synthetic challenges and approach..................................... 102
6.3.2
Material quality ............................................................ 105
Summary and Perspective ......................................................... 106
7. Conclusions ..................................................................................... 112
TABLE OF CONTENTS (continued)
Page
REFERENCES .................................................................................... 115
A. Effectively mentoring undergraduate students......................................... 124
LIST OF FIGURES
Figure
Page
1.1
a) A typical p-n solar cell, and b) the corresponding energy band diagram. 14
1.2
Crystal structure of sphalerite CdTe. ................................................ 15
1.3
Crystal structure of tetragonal Cu(In,Ga)Se2 . .................................... 15
2.1
Crystal structure of orthorhombic Fe2 XS4 , X = Si, Ge. ........................ 32
2.2
Diffuse reflectance data collected from Fe2 XS4 powders and modeled as
direct gap. .................................................................................. 32
2.3
Transmission data collected from a Fe2 GeS4 single crystal. .................... 33
2.4
Temperature dependence of Seebeck coefficients of Fe2 XS4 pellets........... 33
2.5
a) Log resistivity vs 1/T1/4 and b) vs 1000/T for Fe2 SiS4 pellet. ............ 34
2.6
a) Log resistivity vs 1/T1/4 and b) vs 1000/T for Fe2 GeS4 pellet. ........... 34
2.7
Molecular orbital diagrams for Fe2 XS4 . ............................................. 35
2.8
Band structure of Fe2 SiS4 calculated with Wien2k............................... 35
2.9
Density of states for Fe2 SiS4 calculated with Wien2k. .......................... 36
2.10 Crystal orbital overlap population curves for Fe2 SiS4 calculated with Caesar. ........................................................................................... 37
2.11 Absorption coefficient vs energy for Fe2 XS4 calculated with Wien2k........ 37
3.1
Coordination of Fe3+ by C2 O2−
4 in the Fe2 (C2 O4 )3 molecule. ................. 54
3.2
FT-IR spectrum collected from a Fe oxalate thin film. ......................... 54
3.3
XRD data collected from Fe oxalate thin films heated under flowing H2 S
(g) at the specified temperatures. .................................................... 55
3.4
RMS roughness vs anneal temperature calculated from AFM scans of Fe
oxalate thin films heated under flowing H2 S (g) at the specified temperatures. ....................................................................................... 55
3.5
SEM cross-sectional image of a Fe oxalate thin film. ............................ 56
LIST OF FIGURES (continued)
Figure
Page
3.6
SEM cross-sectional image of a pyrite thin film. .................................. 56
3.7
Direct and indirect (inset) optical band gap estimates for a pyrite film. ... 57
4.1
FT-IR spectrum collected from as-cured and air-heated IHN thin films. ... 75
4.2
XRD collected from IHN gels heated in air at 150 ◦ C for 60 min and then
reacted with H2 S (g) for 120 minutes at various temperatures. ............... 75
4.3
XRD collected from IHN gels heated in air at 150 ◦ C for 120 min and
then reacted with H2 S (g) for 120 minutes at various temperatures. ........ 76
4.4
SEM cross-sectional image of an as-cured IHN thin film on Si. ............... 76
4.5
SEM a) cross-sectional and b) top down images of a 400◦ C, 30 min airheated IHN thin film on Si. ............................................................ 77
4.6
XRD collected from as-cured IHN thin films reacted with H2 S (g) for ten
minutes at various temperatures. ..................................................... 77
4.7
XRD collected from 400◦C, 30 min air-heated IHN thin films reacted with
H2 S (g) for ten minutes at various temperatures. ............................... 78
4.8
SEM a) cross-sectional and b) top down images of a pyrite film synthesized from as-cured IHN................................................................. 78
4.9
SEM a) cross-sectional and b) top down images of a pyrite film synthesized from a 400◦ C, 30 min air-heated IHN film. ................................. 79
4.10 Wavelength-dependent absorption coefficient and indirect optical band
gap estimate (inset) for a pyrite film................................................. 79
5.1
Crystal structure of cubic Cu3 TaQ4 , Q = S, Se, Te. ............................. 93
5.2
XRD patterns for Cu3 TaSe4−x Tex , x = 1 - 3. ..................................... 93
5.3
Unit cell parameter vs x for Cu3 TaSe4−x Tex . ...................................... 94
5.4
PL spectra collected from intrinsic and W-doped Cu3 TaSe4 ................... 94
5.5
PL intensity or FWHM vs temperature for Cu3 TaSe4 ........................... 95
5.6
Band structure of Cu3 TaSe4 ............................................................ 95
5.7
Density of states for Cu3 TaSe4 . ....................................................... 96
5.8
Calculated absorption coefficient vs energy for Cu3 TaSe4 and Cu3 TaTe4 . . 97
LIST OF FIGURES (continued)
Figure
Page
6.1
Crystal structure of tetragonal BaCuQF, Q = S, Se, Te........................ 109
6.2
XRD data collected from BaCuSeF prepared with and without the 2 h
anneal under flowing Ar. ................................................................ 109
LIST OF TABLES
Table
Page
1.1
Optical and electrical properties of commercial solar absorbers. ............. 16
2.1
Unit cell parameters for Fe2 XS4 ....................................................... 38
2.2
EPMA data for Fe2 XS4 pellets. ....................................................... 38
6.1
Unit cell parameters for BaCuQF powders......................................... 110
6.2
Gibbs free energies of formation used to calculate ∆Grxn (flowing gas) ∆Grxn (solid state) for BaCuSF and BaCuTeF. ................................... 110
6.3
Gibbs free energies of formation for Q oxides at 527 ◦ C. ....................... 110
6.4
EPMA data for BaCuQF pellets. ..................................................... 110
6.5
XPS data for BaCuQF pellets. ........................................................ 111
Andy, you make the way to “happily ever after” so much more interesting
than I ever imagined. To God be the glory!
Copper and Iron Chalcogenides for Efficient Solar Absorption
CHAPTER 1
Introduction
2
1.1
Importance of electrical energy
As fossil fuel reserves are depleted and concern over global climate rises, alternative sources of electricity become ever more attractive. Factor in the increase in
demand from the growing human population, and it is clear that development of
technologies to generate electrical energy from hydroelectric, wind, nuclear, and solar sources is extremely important. Nuclear power remains controversial because of
safety concerns and the final disposal of spent fuel. Hydroelectric and wind energy, of
course, are already important components of the electrical grid, but they are limited
by the availability of moving water sources and steady wind currents. With approximately 125,000 TW of solar radiation striking the earth at any given time,1 electricity
generation from the sun has the greatest potential for growth. This is much more
power than both the 13 TW/yr the entire human population used in 2007 and the
30 TW/yr projected demand in 2050.1
As is often the case when theory is put into practice, actual photovoltaic (PV)
energy generation falls far short of the vast potential. In 2003, PV in the United States
produced only 0.1% of the amount of electricity that was generated by fossil fuels.1
While PV capacity has grown significantly since then, according to a February 2010
retail survey, electricity generated from solar panels costs $0.351/kWh,2 while the
average residential customer in the United States paid $0.116/kWh in 2009.3 This
price difference demonstrates that the cost of PV must still decrease significantly
before it becomes competitive with entrenched wind, hydroelectric, and fossil fuel
generated power.
In order to accomplish this ambitious goal, new material systems and processing techniques must be developed, and this dissertation explores both. Choosing
promising materials requires consideration of the basic components and operation of
3
a PV cell and the influence of crystal structure and composition on a compound’s
optical and electrical properties. Background for both topics is given below, along
with a summary of the state-of-the-art solar cell absorbing materials to provide target
properties. Another promising cost-reduction strategy is to move from single crystal Si-based devices (February 2010 module price $2.37/Wp2 ) to polycrystalline thin
film devices (February 2010 module price $1.76/Wp2). Solution deposition of the
polycrystalline films could cover large surface areas quickly, providing a pathway to
increase manufacturing speed and further reduce the cost of the final PV module. To
explore this approach, this work utilizes spin coating to deposit aqueous precursors.
Details of that technique are also discussed below along with some comments on the
flowing gas technique that is also employed throughout this dissertation.
1.2
Efficient solar energy absorption
A typical solid state solar cell is a heterojunction formed between n- (majority
carrier electrons) and p- (majority carrier holes) type semiconductors whose purpose
is to absorb sunlight and produce electron/hole pairs (e− /h+ ). To extract electrical
current from this device, it is imperative than the e− /h+ be separated, and drift and
diffusion are the two mechanisms that drive this separation.4 Carrier drift occurs
in response to the electrostatic field created by the lattice-bound cations and anions
at the heterojunction that are left when the intrinsic e− from the n-type material
recombine with the h+ in the p-type material (Figure 1.1a). This field accelerates
electrons toward the n-type layer and holes toward the p-type layer. Once e− and
h+ have reached the n- and p-layers, diffusion takes over, and the electrochemical
potential gradient formed between the regions of high carrier concentration near the
heterojunction and the lower carrier concentrations near the contacts drives the e−
4
and h+ out of the solar cell. This process is depicted in Figure 1.1b.
Another device option is the p-i-n junction in which a very thin insulating layer
(referred to hereafter as the absorber) is inserted between the p- and n-type layers.
p-i-n junction cells provide an important advantage over the more common p-n junction because they allow desired properties to be divided between materials. As a
result, the absorber can be chosen to have a specific band gap to absorb light of desired wavelength. This flexibility is crucial because the solar spectrum has significant
intensity only over certain wavelength ranges.4 The p- and n-layers are then selected
for the higher mobility required for efficient carrier extraction and current generation,
along with sufficiently large band gaps to transmit visible light to the absorber. An
additional benefit of the p-i-n structure is that e− /h+ generated in the absorber are
inside the space charge region where they are much more likely to be separated than
carriers in a p-n junction device which could be generated anywhere in the p- or
n-layers.
The most important figure of merit for a solar cell is its power conversion efficiency,
or the amount of electricity that is generated relative to the light irradiating the
device. The efficiency is calculated as follows:
η=
Voc Isc F F
Etot A
(1.1)
where Voc is the open circuit voltage, Isc is the short circuit current, A is the area
of the solar cell, and Etot is the total incident irradiance. FF, the fill factor, is a
normalized parameter that gives an indication of the ideality of the device. Voc
and Isc are both limited by the optical band gap of the absorber. That band gap
5
dictates the minimum energy that photons must have to excite e− and create e− /h+ ,
so it is directly proportional to Voc . At higher wavelengths, however, there are fewer
photons available from the sun, and when small numbers of e− are excited Isc will also
be small. Since Voc and Isc are inversely related to each other, intermediate absorber
band gaps of 1.0 - 1.6 eV are required to maximize sunlight conversion efficiency in
single heterojunction solar cells.4
If multiple heterojunctions are stacked together to create multijunction solar cells,
the theoretical efficiency that can be generated also increases. While the maximum
possible efficiency for a single heterojunction solar cell with an absorber band gap of
1.1 eV operating at sea level is 33%, a double heterojunction with one absorber band
gap of 1.4 eV and the other at 0.7 eV has a theoretical efficiency of 60% under the
same conditions.5 An actual multijunction solar cell consisting of InGaAs/GaAs/Ge
has been measured at 34% efficiency under concentrated light.5
What all of this means is that the chief factors impacting material choice for the
layers in a working PV device are optical band gap, majority carrier concentration
and type, and mobility. A useful absorber can have a band gap in the range of 0.7
to 1.6 eV, and it should also have a low carrier concentration so photo-generated
carriers do not immediately recombine. High mobility and larger band gap p- and
n-type window layers are also important to let light through to the absorber and then
extract the e− and h+ produced by that light.
1.3
Leading absorber materials
State-of-the-art solar cell manufacturing utilizes Si, CdTe, or CuIn1−x Gax Se2
(CIGS) in p-n junction configurations. Because of concomitant development in the
semiconductor industry, Si PV technology is the most advanced, and it can be pro-
6
duced in single crystal, polycrystalline, amorphous, and ribbon forms. In 2006, 97.2%
of all PV cells produced utilized some form of Si as the absorber, and the best module
efficiencies were 22.7%.1 There is a major limitation however, in that single-crystal
Si is expensive to process. The clear interest in decreasing the cost of PV has spurred
the development of the alternative thin film absorbers CIGS and CdTe, and Table
1.1 provides some important optical and electrical properties of all three materials for
comparison.
At ambient pressure, CdTe crystallizes in the cubic sphalerite structure (Figure
1.2) with Cd2+ forming a face-centered cubic array and Te2− occupying
1
2
of the
tetrahedral holes.6 Both ions have four nearest neighbors in tetrahedral arrangements.
When CdTe thin films are deposited in vacuum as part of a solar cell, the material can
also adopt the hexagonal wurtzite structure.6 Cd2+ and Te2− are still tetrahedrally
coordinated, but the second-nearest neighbor spheres are slightly different. Both
CuInSe2 and CuGaSe2 crystallize in the chalcopyrite structure, so it is not surprising
that In3+ and Ga3+ can be alloyed in any combination to form CIGS. The chalcopyrite
structure is closely related to the sphalerite structure, and it is formed by stacking
two sphalerite unit cells together and replacing the Cd2+ with Cu+ , In3+ , or Ga3+
as shown in Figure 1.3. As with Te2− in sphalerite CdTe, Se2− occupies
1
2
of the
tetrahedral holes in the chalcopyrite lattice.
At 11%7 the best reported CdTe and CIGS modules have only half the efficiency
of the best single-crystal Si devices. The thinner layers of CdTe and CIGS that can
be utilized because of their large absorption coefficients (Table 1.1) overcome this,
however, and the thin film-based devices are still considerably cheaper than single
crystal Si cells. Unfortunately, based on known mineral reserves and extraction costs,
CdTe and CIGS do not have good long term potential for the sort of large scale
7
deployment required to meet projected world electrical energy needs.8 There is clearly
need for even more improvement.
1.4
Prediction of electrical and optical properties from crystal structures
If the current absorbers Si, CdTe, and CIGS are not able to able to provide
cost-effective PV electrical energy, new materials are clearly needed. With over 100
elements in the Periodic Table and an infinite number of possible combinations, good
search strategies will be required to focus efforts in productive directions. A common
theme of the three absorbers discussed above is the tetrahedral environment of each
element or ion. e− and h+ can not move from one lattice site to another unless
there is a bond connecting them, and the tetrahedral bonding provides pathways in
multiple directions. When numerous routes are available, photogenerated carriers
in Si, CdTe, and CIGS can move quickly through the respective lattices and reach
electrical contacts. High symmetry is therefore one characteristic to look for when
choosing new absorbers.
While carrier mobility through a structure is an important consideration, perhaps
the most significant property of an absorber is the optical band gap defined by the
filled valence and empty conduction bands. Filled outer shell atomic orbitals will typically contribute to a material’s valence band, while empty orbitals should fall in the
conduction band. While p-block metal ions like In3+ and Ga3+ have either completely
full or empty s-, p-, and d -orbitals, transition metal ions often have partially filled
d -orbitals. High symmetry environments like tetrahedral or six-coordinate octahedral
will split these valence shell d -orbitals into higher and lower energy groups, and these
groups will usually end up in both the valence and conduction bands. As a result,
8
the coordination environment of transition metals in a crystal structure is a good
metric for predicting the optical band gap of the material, and in high coordination
the transition metals should produce the larger optical band gaps characteristic of
semiconductors. If the site coordination number is lower than four, the d -orbitals will
split into more than two groups, which effectively decreases the band gap.
Majority carrier concentration and type, the other important material properties
of efficient solar absorbers, are unfortunately more difficult to predict. When moving
from a well characterized compound to a relatively unknown but compositionally and
structurally similar one, the majority carrier type will often remain the same. Since
the established thin film absorber compounds are both p-type, other Cd- or Cu-based
chalcogenides could reasonably be expected to be p-type also. Carrier concentration,
however, can only be measured after successful material synthesis is achieved.
1.5
Precursor chemistries for spin coating and flowing gas
reactions
In addition to considering new absorber materials, thin film solution deposition via
spin coating of aqueous solutions is described. Since the band gaps of the deposited
metal oxides are too large to be useful solar absorbers, they are further treated with
flowing H2 S (g) to convert to lower gap sulfides.
Spin coating itself is a simple process.9 A thick layer of liquid made up of carefully
chosen chemical species and solvents is dispensed onto the surface of the substrate.
That substrate is then rapidly rotated (1000 rotations/min or faster) to generate
centrifugal force. The liquid flows radially outward, and some precursor is lost as
droplets from the edges of the substrate. Simultaneous solvent loss and film thinning
increase the concentration of the remaining solids, which in turn increases the viscosity
9
of the liquid film. After an equilibrium thickness is achieved, the film is treated with
heat to polymerize it and drive off any remaining solvent.
While spin coating is straight forward, development of appropriate aqueous precursors to deposit dense metal oxide films is not. The metal-ligand clusters must
be stable enough to remain soluble, and yet require little activation energy to polymerize and densify to a solid film. The former goal can be realized by the common
practice of including large organic ligands, but that strategy then requires significant
and disruptive thermal treatment to remove those ligands. Such processing is likely
to produce rough and porous metal oxide films which cannot be integrated into a
working solid state device where additional layers must be added on top. A better
route works with the natural hydrolysis
[M(H2 O)6 ]z+ + H2 O → [M(H2 O)5 (OH)](z−1)+ + H3 O+
and condensation
2 M(OH) → M h
OH
i M → M-O-M + H2 O
OH
that all metals undergo in aqueous solution.10 In conjunction with choosing ligands
that decompose at low temperatures to form volatile products, this strategy has
proven successful for the solution deposition of ZnO thin films.11
All reactions are driven by the availability of reactants and slowed by the presence
of products. In a flowing gas system these species are described by their partial pressures, which are primarily controlled by gas flow rate. High flow rates will increase
the overall reaction rate by supplying a large excess of reactants and carrying away
volatile products. Conversely, then, slower flow rates and therefore slower reactions
can significantly change reaction mechanisms. Another useful aspect of flowing gas
10
reactions is that they enable the decomposition of metal salts to produce small, reactive particles in-situ, which is particularly advantageous for reactions that require
easily oxidized metal species. With the appropriate salt, it may also be possible to
carry out these reactions at low temperatures to avoid sintering the solid particles
before they can react with the flowing gas.
1.6
Summary and Perspective
Two approaches to reducing the cost associated with PV energy generation are
discussed in this dissertation. First, readily available materials that are prevalent in
the earth’s crust are considered, and the structures of these compounds are used to
choose other promising compositions that are largely uncharacterized. Such readily
available sources will help to reduce manufacturing costs. It is not enough to simply
use abundant materials, however. These materials must also be deposited in thin
film form over large areas, so the second approach to reducing cost is to explore the
solution deposition of metal oxides. Both solution deposited thin films and some bulk
powders are treated under flowing gas to investigate their conversion to chalcogenides,
thus demonstrating a route to materials with a wide range of optical band gaps.
Chapters 2, 5, and 6 detail exploration of several Fe and Cu-based chalcogenides
and chalcogenide fluorides. These materials are Fe2 XS4 (X = Si, Ge), Cu3 TaQ4 (Q =
Se, Te), and BaCuQ’F (Q’ = S, Se, Te), and they were chosen based on similarities to
binary FeS2 pyrite and Cu2 Q’. The binaries are p-type semiconductors with optical
band gaps of 0.9 to 1.2 eV,12–14 and this work demonstrates that the additional
elements in the more complex compounds increase their band gaps while retaining
the p-type character of the simpler compounds. Since FeS2 pyrite itself is a promising
solar absorber,8, 12 Chapters 3 and 4 describe the solution deposition of two different
11
aqueous precursors and their subsequent conversion to sulfides under flowing H2 S (g).
One precursor is similar to the salt Fe2 (C2 O4 )3 , while the other is made up of iron
hydroxide nitrate clusters formed through controlled hydrolysis. While the optical and
electrical properties of the final 100 nm thick FeS2 pyrite thin films prepared from
both precursors are very similar, scanning electron microscopy images demonstrate
that the films produced from the iron hydroxide nitrate precursor are much denser.
This makes the iron hydroxide nitrate route the more viable for producing a pyrite
absorber that can be successfully integrated into a working PV device.
12
References
[1] A. Slaoui and R. Collins. Advanced inorganic materials for photovoltaics.
Materials Research Society Bulletin, 32:211–214, March 2007.
[2] www.solarbuzz.com maintains current photovoltaic panel and power cost
information.
[3] http://www.eia.doe.gov/cneaf/electricity/epm/table5 6 b.html.
[4] J. Gray. Handbook of Photovoltaic Science and Engineering, chapter The
Physics of the Solar Cell, pages 61–112. John Wiley & Sons, 2003.
[5] A. Luque and A. Marti. Handbook of Photovoltaic Science and Engineering,
chapter Theoretical Limits of Photovoltaic Conversion, pages 113–151. John
Wiley & Sons, 2003.
[6] B. McCandless and J. Sites. Handbook of Photovoltaic Science and
Engineering, chapter Cadmium Telluride Solar Cells, pages 617–662. John
Wiley & Sons, 2003.
[7] J. Beach and B. McCandless. Materials challenges for CdTe and CuInSe2
photovoltaics. Materials Research Society Bulletin, 32:225–229, 2007.
[8] C. Wadia, A. Alivisatos, and D. Kammen. Materials availability expands the
opportunity for large-scale photovoltaics deployment. Environmental Science &
Technology, 43:2072 – 2077, 2009.
[9] D. Meyerhofer. Characteristics of resist films produced by spinning. Journal of
Applied Physics, 49(7):3993–3997, 1978.
13
[10] J. Burgess. Metal Ions in Solution. Ellis Horwood: Chichester, England, 1978.
[11] S. Meyers, J. Anderson, C. Hung, J. Thompson, J. Wager, and D. Keszler.
Aqueous inorganic inks for low-temperature fabrication of ZnO TFTs. Journal
of the American Chemical Society, 130(51):17603–17609, 2008.
[12] A. Ennaoui, S. Fiechter, C. Pettenkofer, N. Alonso-Vante, K. Buker,
M. Bronold, C. Hopfner, and H. Tributsch. Iron disulfide for solar energy
conversion. Solar Energy Materials and Solar Cells, 29:289 – 370, 1993.
[13] P. Lukashev, W. Lambrecht, T. Kotani, and M. van Schilfgaarde. Electronic
and crystal structure of Cu2−x S: Full-potential electronic structure calculations.
Physical Review B, 76(19):195202, 2007.
[14] G. Sorokin, Y. Papshev, and P. Oush. Photoconductivity of Cu2 S, Cu2 Se, and
Cu2 Te. Soviet Physics Solid State, 7(7):1810 – 1811, 1966.
[15] D. Schroder. Semiconductor Material and Device Characterization.
Wiley-Interscience, 3rd edition, 2006.
[16] W. Shafarman and L. Stolt. Handbook of Photovoltaic Science and Engineering,
chapter Cu(InGa)Se2 Solar Cells, pages 567–616. John Wiley & Sons, 2003.
14
Figure 1.1: a) A typical p-n solar cell, and b) the corresponding energy band
diagram.
15
Cd
Te
Figure 1.2: Crystal structure of sphalerite CdTe.
c
Se
In/Ga
Cu
Figure 1.3: Crystal structure of tetragonal Cu(In,Ga)Se2 .
16
Table 1.1: Optical and electrical properties of commercial solar absorbers.6, 15, 16
Mobilities given are for single crystals.
Absorber Eg (eV)
α (cm−1 )
Majority Carriers
µ (cm2 /V·s)
Si
1.1
103 (hν ≤ 800 nm)
e− /h+
1400 (e− )/470 (h+ )
CIGS
1.0 - 1.2 105 (hν ≤ 880 nm)
h+
15 - 150
4
CdTe
1.5
10 (hν ≤ 730 nm)
h+
50 - 80
CHAPTER 2
Optical and electrical properties of Fe2XS4 (X = Si, Ge)
Heather A. S. Platt, Robert Kykyneshi, Andriy Zakutayev,
Janet Tate, Vorranutch Jieratum, Douglas A. Keszler
To be submitted for publication in Nature Materials.
18
2.1
Introduction
In this contribution, we describe the optical and electrical properties of the ternary
sulfides Fe2 XS4 (X = Ge, Si). We have been prompted to examine these materials
on the basis of structural similarities to FeS2 pyrite and the long-standing interest in
pyrite as a thin-film solar absorber. FeS2 pyrite is a well-known cubic sulfide with
Fe2+ octahedrally coordinated by S atoms in S2−
2 dumbbells. The electronic structure
of pyrite is characterized by a valence band of nonbonding Fe 3d -t2g orbitals and a
conduction band of Fe 3d -eg and S 3p orbitals.1 Reported single crystal electron
and hole mobilities are as high as 300 cm2 /V·s,1, 2 and thin-film hole mobilities of 5
cm2 /V·s have been demonstrated.2 An optical indirect band gap of 0.9 - 0.95 eV and
an absorption coefficient reaching 105 cm−1 for wavelengths shorter than 1 µm make
the material especially attractive for thin-film solar applications.1–3 The abundance
of Fe and S also add economic incentive to the development of FeS2 pyrite.4
The promise of FeS2 pyrite as a solar absorber has yet to be realized, however,
due in part to the apparent difficulty of producing high-quality single-phase material.
Defects in the sulfur sublattice create states inside the band gap.5, 6 These defect
states have been shown to pin the Fermi level, leading to open circuit voltages in
photoelectrochemical cells that are significantly lower than the value associated with
the optical band gap.7, 8
To potentially circumvent the problems with the Fermi-level pinning, we are considering the potential of other more complex Fe chalcogenides. In sulfides, Fe is known
to occupy six- (octahedral) or four-coordinate (tetrahedral) sites. Tetrahedral coordination will produce a small crystal field splitting of the Fe d orbitals and a band gap
that is too small for a useful solar absorber. The chalcopyrite CuFeS2 , for example,
contains Fe in a tetrahedral environment; its band gap is 0.5 eV.9 Therefore, only
19
materials containing octahedrally coordinated Fe2+ will be of interest. In ternary and
other high-order sulfides, these environments will be realized when the matrix contains other metals that form strong covalent bonds to S. This covalency will deplete
electron density around the Fe ion and force formation of the desired six-coordinate
environment. These considerations led us directly to the title materials. The strongly
covalent X-S interactions contribute to the stabilization of Fe2+ in a slightly distorted
octahedral environment while also providing a mechanism for limiting S vacancies.
Fe2 XS4 crystallize in the well known olivine structure (Figure 2.1). The S anions
form an approximately hexagonal close packed arrangement with half of the octahedral sites filled by Fe2+ and
1
8
of the tetrahedral holes occupied by X4+ .10 There are
two distinct, distorted octahedral Fe sites in the structure. One is characterized by a
center of symmetry. These octahedra share edges to form infinite chains along the b
axis. This connectivity differs from that of pyrite, where the Fe-centered octahedra
share only vertices. The other octahedra in the Fe2 XS4 structure, characterized by a
mirror plane orthogonal to the b axis, bridge the edge-sharing chains. The octahedral
chains are further connected along the c axis by XS4−
4 tetrahedra.
Aside from reports on the structural and magnetic properties,10–12 little is know
about the behavior of Fe2 XS4 . They have yet to be examined optically or electrically,
vital characteristics for determining suitability as solar absorbers. In this paper, we
report results from such measurements on polycrystalline samples, correlating findings
to band structure calculations and making comparisons to FeS2 pyrite.
20
2.2
2.2.1
Experimental
Synthesis
Stoichiometric amounts of the metal powders (Alfa Aesar, ≥99.9%) were ground
with sulfur powder (Alfa Aesar, 99.5%) and sealed in evacuated silica tubes. Polycrystalline Fe2 GeS4 was produced via heating at 750 ◦ C for 48 h, and powder Fe2 SiS4 was
synthesized by heating at 1075 ◦ C for 72 h. A tube furnace with three-zone control
(Carbolite) was used to maintain a uniform temperature profile across the sample.
Tubes containing Fe2 SiS4 were opened and the material from the tube ground in a
N2 -filled glovebox. Fe2 GeS4 powders were ground in air. Pellets of 9.5- or 13-mm
diameter were produced by cold pressing at 1.5 - 2 tons, and then sintered at 20,000
psi in a hot isostatic press (AIP6-30H, American Isostatic Presses, Inc) with Ar (g).
Fe2 GeS4 pellets were heated at 900 ◦ C for 3 h, and Fe2 SiS4 pellets were heated at
1075 ◦ C for 4 h. Final pellet densities were approximately 70% of theoretical values.
Single crystals of Fe2 GeS4 were prepared by combining the powders described
above with I2 and then sealing in evacuated silica tubes. The tubes were heated at
980 ◦ C for one day, and then maintained at 900 ◦ C for three days before cooling to
500 ◦ C at 8 - 10 ◦ C/h. Final cooling to room temperature was not controlled.
2.2.2
Characterization
X-ray diffraction (XRD) was performed on powders by using a Rigaku MiniFlex
II diffractometer with Cu Kα radiation. Electron probe microanalysis (EPMA) was
performed on the pellets by using a Cameca SX-100. Elemental Fe, Ge, and Si
were used as standards for the respective elements, natural pyrite was used as the S
standard, and Al2 O3 was used as the O standard. Each Fe2 XS4 pellet was sampled
21
over several hundred microns, and at least 12 points were used to calculate the final
elemental ratios.
Room-temperature resistivity was measured in the van der Pauw geometry on Hall
Measurement System version 2.4.1 from Lake Shore Cryotronics, Inc. Temperaturedependent resistivity data were collected in the four collinear probe configuration on
a system designed and built in-house. Majority-carrier type was determined from
Seebeck measurements by using the same in-house system.
Optical band gaps were determined from diffuse reflectance measurements on the
powders.13 Data were collected by using a Si photodiode mounted on an Oriel 70491
integrating sphere. The sample was placed in a ceramic holder beneath the sphere,
and light from a W lamp was directed through a double monochromator (Oriel 77250)
and a series of mirrors into the top of the sphere. The reference powder was MgO.
Transmission experiments were performed with an UV-visible (HR4000) detector and
a deuterium tungsten halogen light source (DH-2000-BAL) from Ocean Optics.
2.2.3
Band Structures
The electronic structures of Fe2 GeS4 and Fe2 SiS4 were computed with the code
Wien2k14 in the full-potential linearized augmented plane-wave (LAPW) formalism
with the PBE generalized gradient approximation. The crystal-structure parameters
of Fe2 SiS4 and Fe2 GeS4 were taken from published data.10 The muffin-tin radii were
set to 2.45 au for Fe and 1.98 au for Si and S for the calculations on Fe2 SiS4 , and
to 2.44 au for Fe and 2.06 au for Ge and S for the calculations on Fe2 GeS4 . The
potentials and charge densities were expanded on a k-mesh of 3570 points in the
Brillouin zone, including 495 unique k points. Inside the spheres, these quantities
were expanded up to l = 10. The calculations were iterated until the total energies
22
and charge converged to better than 0.1 mRy and 10−3 e, respectively.
Total and partial densities of states and dispersion curves for Fe2 SiS4 were also
calculated with the semi-empirical extended Hückel tight binding (EHTB) method
by using the software package Caesar.15 The sample size included up to 5 nearestneighbor cells.
2.3
2.3.1
Results
Experimental results
Preparation of the compounds Fe2 XS4 by using high-temperature methods must
account for two main issues. One is the high vapor pressure of the XS2 components
at high temperatures. While XS2 vapor should be a favorable reagent for reaction
with FeS to form the final product, the equilibrium
Fe2 XS4 (s) ⇋ 2FeS (s) + XS2 (g)
can lead to segregation of SiS2 or GeS2 from the bulk.10, 11, 16 Because of the temperature dependence of the equilibrium constant, segregation will be exacerbated if
samples are heated across a temperature gradient. To minimize gradients, we have
used a three-zone tube furnace to maintain uniform sample heating. The other important consideration is the oxophilic character of Si. The ready hydrolysis of SiS2
upon exposure to moist air, resulting in the formation of SiO2 , has been described.10
It is also possible that Fe2 SiS4 crystallites could exhibit a similar reactivity. To limit
these potential problems, all Si-containing samples have been handled under dry N2
(g).
XRD data from multiple samples were consistent with the formation of Fe2 SiS4
or Fe2 GeS4 . All lines in the patterns could be indexed on the basis of the previously
23
reported orthorhombic unit cells. Unit cell parameters calculated from the data are
given in Table 2.1, and unit cell volumes are within 0.5% of published values.10 EPMA
data (Table 2.2) from sintered pellets are consistent with expected stoichiometries.
These data also indicate that the Fe2 SiS4 pellets are contaminated with oxygen. We
do not yet know the source of this impurity: initial synthesis, sintering, or exposure
to air between process steps. No oxygen was detected in the Fe2 GeS4 pellets.
The Kubelka-Munk method was used to calculate k/s from reflection measurements as a function of energy. k represents the energy-dependent absorption coefficient, and s is a constant scattering factor.13 These data are plotted as (E · k/s)2 vs
E in Figure 2.2 for both Fe2 XS4 powders. Extrapolation of the steepest portion of the
curves to the abscissa produces a direct optical gap near 1.5 eV for each compound.
The transmission experiments with single crystals of Fe2 GeS4 also show absorption
onset beginning at 1.56 eV (Figure 2.3). Since the crystals were 80 to 100 µm thick,
it was not possible to determine an accurate absorption coefficient. For comparison,
FeS2 pyrite has an indirect gap of magnitude 0.9 - 0.95 eV.1, 17, 18 The direct nature
of the band gaps in Fe2 XS4 provide a distinct advantage for solar applications. In
addition, the larger magnitudes provide a better match to the solar spectrum than
FeS2 pyrite.
The Seebeck coefficients at room temperature are 780 uV/K for Fe2 GeS4 and 340
uV/K for Fe2 SiS4 , and they remain positive at all temperatures (Figure 2.4). Holes
are clearly the majority carriers in each of the Fe2 XS4 compounds, which is consistent
with the presence of defects that deplete electrons from the Fe d -bands.
Room temperature electrical measurements in the van der Paaw configuration
produced resistivity values of 3.0·105 Ω-cm for Fe2 SiS4 and 5.0·103 Ω-cm for Fe2 GeS4
pellets. These data agree well with measurements using four collinear probes. Room-
24
temperature values measured in the latter geometry are 2.8·105 Ω-cm and 3.2·103 Ωcm for Fe2 SiS4 and Fe2 GeS4 , respectively. Four collinear probes were also used to
measure resistivity as a function of temperature, and these data are presented in
Figures 2.5 and 2.6. The resistivity of both Fe2 XS4 compounds is semiconductorlike, increasing with decreasing temperature. Flattening of the graphs in the low
temperature range is associated with the noise limit of the measurement system.
Temperature-dependent resistivity measurements may be used to elucidate the
nature of the conduction mechanism through a semiconductor. Carriers may move
by band conduction or through localized states in a hopping mechanism.19 In the case
of band conduction, a plot of log resistivity vs 1/T should produce a straight line with
the slope corresponding to the activation energy. If hopping conduction dominates,
a plot of log resistivity vs 1/T1/4 should be linear. For Fe2 SiS4 , neither model yields
a straight line (Figure 2.5). For Fe2 GeS4 (Figure 2.6), careful comparison of the data
in the high temperature range reveals no significant difference in linearity between
the log resistivity vs 1/T1/4 and log resistivity vs 1/T plots, which is not unexpected
given the limited temperature range. Considering these electrical results and the
rather shallow slopes associated with the band edges in the optical measurements
(Figure 2.2), it is clear that defects are contributing significantly to the properties of
the powders. We are currently pursuing further experiments with single crystals to
better define the intrinsic characteristics of the materials.
2.3.2
Electronic Structures
Each of the Fe sites in the Fe2 XS4 structure is characterized as a distorted octahedron, with point group symmetries of D2h for Fe(1) and Cs for Fe(2). Fe-S bond
lengths vary from 2.461 to 2.600 Å in Fe2 SiS4 and 2.449 to 2.629 Å in Fe2 GeS4 , and
25
S-Fe-S angular deviations from an octahedral geometry are characterized by averages
of 5.0◦ for Fe2 SiS4 and 4.6◦ for Fe2 GeS4 .10 In these distorted octahedral geometries,
the Fe d orbitals will split into three lower energy and two higher energy states as
shown in Figure 2.7. Because the FeS6 distorted octahedra form edge- and vertexsharing chains and the XS4 distorted tetrahedra are isolated, carrier transport will
occur through bands dominated by Fe and S orbital character. The lower energy Fe
3d orbitals and intermixed S orbitals will define the valence band, while the higher
energy Fe 3d orbitals and overlapping S orbitals will define the conduction band.
As expected for isostructural compounds, the band structures and density of states
(DOS) calculated with Wien2k for both Fe2 XS4 compounds are very similar. Representative results for Fe2 SiS4 are given in Figures 2.8 and 2.9. Band dispersion near the
valence band maximum (VBM) is generally low, except along the lines Γ - Y and Γ R. The former follows the edge-sharing Fe-S octahedra that extend along the b axis
of the structure, and the latter tracks a chain of edge- and vertex-sharing octahedra
along the unit cell diagonal. In contrast, the conduction band minimum (CBM) has
greater curvature than the VBM in all directions. This is consistent with the significant overlap of the higher energy Fe 3d and S 3p orbitals and their contributions to
the conduction band DOS.
The lowest energy transition across the computed band gap of Fe2 SiS4 occurs
along the Γ - Y line. It is direct with magnitude 0.25 eV. In Fe2 GeS4 , there is an
indirect transition of 0.23 eV from Γ to the Γ - Y line, and the direct transition at Γ
is 0.24 eV. Because the difference is so small, we also modeled the experimental data
collected from Fe2 GeS4 powders to show the direct band gap. The LAPW method
is known to underestimate the magnitude of band gaps,20 so it is not surprising
that the experimentally determined values are higher than those predicted from the
26
calculations.
Because of the similarities of the results from the Wien2k calculations for Fe2 SiS4
and Fe2 GeS4 , Caesar calculations were limited to Fe2 SiS4 . The band structure and
DOS are qualitatively the same as those from the Wien2k calculations. Crystal
orbital overlap population (COOP) curves are given in Figure 2.10. In the COOP
curves, positive contributions correspond to bonding interactions, and negative values
correspond to antibonding interactions.18
The bonding in Fe2 SiS4 can be evaluated by considering both the DOS and COOP
curves. As shown in Figure 2.9, Fe 3d and S 3p orbitals contribute to the valence
band. In Figure 2.10, a small Fe-S bonding peak at the lowest energy end of the
valence band, a much larger bonding peak in the mid-valence band, and finally a
small antibonding peak at the VBM are observed. Because of their orientations, the
lower energy Fe 3d - S 3p orbital overlap is characterized by π interactions, while
the higher energy Fe 3d - S 3p orbitals overlap directly in σ interactions. The small
bonding peak at the bottom of the valence band and the large peak in the mid-valence
band correspond to Fe 3d - S 3p π and σ bonding. The small antibonding peak at
the VBM is assigned to Fe 3d - S 3p π ∗ , and a large antibonding contribution to the
conduction band reveals the corresponding Fe 3d - S 3p σ ∗ . The Si-S COOP curve
in Figure 2.10 positions the strong bonding interactions. Similar results have been
reported for Si-O in silicates21 and Fe-Si or Re-Si in silicide carbides.22
Figure 2.11 provides the Wien2k predictions of absorption coefficient vs energy
for Fe2 XS4 . Both compounds have maxima of 3.7 · 105 cm−1 . If the energy scale is
shifted to place the onset of absorption at the experimentally determined band gaps of
1.5 eV, the absorption coefficients reach 105 cm−1 at approximately 1.8 eV (690 nm).
These large magnitudes are consistent with the Fe 3d -rich character of the valence
27
band and the more strongly mixed Fe 3d - S 3p character in the conduction band.
As a result, the electric-dipole moment matrix elements associated with the optical
transition increase, thereby enhancing the magnitudes of the absorption coefficients.
The band structure, DOS, and COOP results presented here provide pictures of
the electronic structures of both Fe2 XS4 compounds that are very similar to that of
FeS2 pyrite.18 Chains of edge- and vertex-sharing Fe-centered octahedra in Fe2 XS4
dominate both VBM and CBM in the form of Fe 3d - and S 3p-orbitals. Similar features are observed in FeS2 . Again like FeS2 , absorption coefficients for both Fe2 GeS4
and Fe2 SiS4 are predicted to be on the order of 105 cm−1 . There is a stark contrast
in the magnetic properties of these compounds, however. While FeS2 is only weakly
paramagnetic with primarily low spin Fe,1 the Fe2 XS4 compounds are high spin with
four unpaired electrons on each Fe.11 Carrier excitation in Fe2 XS4 is thus a spinpairing process, while in FeS2 paired electrons are separated into different orbitals.
2.4
Summary and Perspective
We have demonstrated that the sulfides Fe2 XS4 (X = Ge, Si) are moderate band
gap (Eg ∼ 1.5 eV) p-type semiconductors. Band structure calculations predict direct
carrier transitions across the band gaps of Fe2 XS4 .
We began this study by anticipating that Fe2 XS4 would have properties similar to
FeS2 pyrite. Each of the materials contains Fe2+ octahedrally coordinated by S, which
contributes to p-type conductivity. FeS2 pyrite has an indirect gap of magnitude
0.9 - 0.95 eV. The higher energy and direct band gaps of the Fe2 XS4 compounds
are distinctly advantageous in thin-film photovoltaics, providing higher open-circuit
voltages and more efficient optical-to-electrical energy conversion. We conclude that
the Fe2 XS4 sulfides are promising solar absorbers. As a result, we are pursuing thin
28
film deposition and single crystal growth of Fe2 SiS4 and Fe2 GeS4 to better delineate
their optical and electrical properties and potential utility.
29
References
[1] A. Ennaoui, S. Fiechter, C. Pettenkofer, N. Alonso-Vante, K. Buker,
M. Bronold, C. Hopfner, and H. Tributsch. Iron disulfide for solar energy
conversion. Solar Energy Materials and Solar Cells, 29:289 – 370, 1993.
[2] P. Altermatt, T. Kiesewetter, K. Ellmer, and H. Tributsch. Specifying targets
of future research in photovoltaic devices containing pyrite (FeS2 ) by numerical
modelling. Solar Energy Materials and Solar Cells, 71:181 – 195, 2002.
[3] G. Willeke, R. Dasbach, B. Sailer, and E. Bucher. Thin pyrite (FeS2 ) films
prepared by magnetron sputtering. Thin Solid Films, 213:271 – 276, 1992.
[4] C. Wadia, A. Alivisatos, and D. Kammen. Materials availability expands the
opportunity for large-scale photovoltaics deployment. Environmental Science &
Technology, 43:2072 – 2077, 2009.
[5] M. Birkholz, S. Fiechter, A. Hartmann, and H. Tributsch. Sulfur deficiency in
iron pyrite (FeS2−x ) and its consequences for band-structure models. Physical
Review B, 43:11926 – 11936, 1991.
[6] B. Thomas, T. Cibik, C. Hopfner, K. Kiesner, G. Ehlers, S. Fiechter, and
K. Ellmer. Formation of secondary iron-sulphur phases during the growth of
polycrystalline iron pyrite (FeS2 ) thin films by MOCVD. Journal of Materials
Science: Materials in Electronics, 9:61 – 64, 1998.
[7] K. Buker, N. Alonso-Vante, and H. Tributsch. Photovoltaic output limitation
of n-FeS2 (pyrite) schottky barriers: A temperature-dependent characterization.
Journal of Applied Physics, 72:5721 – 5728, 1992.
30
[8] K. Mishra and K. Osseo-Asare. Fermi level pinning at pyrite (FeS2 )/electrolyte
junctions. Journal of the Electrochemical Society, 139:749 – 752, 1992.
[9] T. Oguchi, K. Sato, and T. Teranishi. Optical reflectivity spectrum of a CuFeS2
single crystal. Journal of the Physical Society of Japan, 48:123 – 128, 1980.
[10] H. Vincent, E. Bertaut, W. Baur, and R. Shannon. Polyhedral deformations in
olivine-type compounds and the crystal structure of Fe2 SiS4 and Fe2 GeS4 . Acta
Crystallographica B, 32:1749 – 1755, 1976.
[11] C. Meyer, Y. Gros, H. Vincent, and E. Bertaut. Proprietes magnetiques et
etude par effet Mossbauer du thiosilicate Fe2 SiS4 . Journal of Physics and
Chemistry of Solids, 37:1153 – 1158, 1976.
[12] A. Junod, K. Wang, G. Triscone, and G. Lamarche. Specific heat, magnetic
properties and critical behaviour of Mn2 SiS4 and Fe2 GeS4 . Journal of
Magnetism and Magnetic Materials, 146:21 – 29, 1995.
[13] J. Lindberg and D. Snyder. Determination of the optical absorption coefficient
of powdered materials whose particle size distribution and refractive indices are
unknown. Applied Optics, 12:573 – 578, 1973.
[14] P. Blaha, K. Schwarz, G. Madsen, D. Kvasnicka, and J. Luitz. WIEN2k An
Augmented Plane Wave + Local Orbitals Program for Calculating Crystal
Properties. Karlheinz Schwarz Techn Universität, Wien, Austria, 2001.
[15] J. Ren, W. Liang, and M. Whangbo. Crystal and Electronic Structure Analysis
using CAESAR. PrimeColor Software, Inc., North Carolina, USA, 1998.
31
[16] W. Viaene. The Fe-Ge-S system: phase equilibria. Neues Jahrbuch fur
Mineralogie, Monatshefte, 1:23 – 35, 1972.
[17] G. Zhao, J. Callaway, and M. Hayashibara. Electronic structures of iron and
cobalt pyrites. Physical Review B, 48:15781 – 15786, 1993.
[18] V. Eyert, K. Hock, S. Fiechter, and H. Tributsch. Electronic stucture of (FeS2 ):
The crucial role of electron-lattice interaction. Physical Review B, 57:6350 –
6359, 1998.
[19] N. Mott and E. Davis. Electronic Processes in Non-Crystalline Materials.
Clarendon Press, 2nd edition, 1979.
[20] F. Tran, P. Blaha, and K. Schwarz. Band gap calculations with Becke-Johson
exchange potential. Journal of Physics: Condensed Matter, 19:196208, 2007.
[21] W. Bleam and R. Hoffmann. Isomorphous substitution in phyllosilicates as an
electronegativity perturbation: Its effect on bonding and charge distribution.
Inorganic Chemistry, 27:3180 – 3186, 1988.
[22] H. Koo and M. Whangbo. Analysis of bonding and d-electron count in the
transition-metal carbides and transition-metal-silicide carbides with discrete
linear M-C-M units (M = Cr, Fe, Re) by electronic structure calculations.
Inorganic Chemistry, 38:2204 – 2210, 1999.
32
c
S
X
Fe
b
a
Figure 2.1: Crystal structure of orthorhombic Fe2 XS4 , X = Si, Ge.
(E · k/s)2 (A.U.)
150
100
50
0
1.5
2
2.5
Energy (eV)
Figure 2.2: Diffuse reflectance data collected from Fe2 XS4 powders and modeled as
direct gap.
33
Transmission (A.U.)
0.12
0.1
0.08
0.06
0.04
0.02
0
1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
Energy (eV)
Figure 2.3: Transmission data collected from a Fe2 GeS4 single crystal.
S (103 µV/K)
6
5
4
3
2
1
0
150
200
250
300
T (K)
Figure 2.4: Temperature dependence of Seebeck coefficients of Fe2 XS4 pellets.
34
1/T1/4 (K-1/4)
0.245
106
0.250
0.255
ρ (Ω-cm)
106
5x105
105
3.4
3.6
3.8
4.0
4.2
1000/T (K-1)
Figure 2.5: a) Log resistivity vs 1/T1/4 and b) vs 1000/T for Fe2 SiS4 pellet.
1/T-1/4 (K-1/4)
ρ (Ω-cm)
0.240
106
0.250
0.260
106
105
105
104
104
103
3.5
4.0
4.5
1000/T (K-1)
Figure 2.6: a) Log resistivity vs 1/T1/4 and b) vs 1000/T for Fe2 GeS4 pellet.
35
Figure 2.7: Molecular orbital diagrams for Fe2 XS4 . Eg is the band gap.
Figure 2.8: Band structure of Fe2 SiS4 calculated with Wien2k.
36
2
Energy (eV)
Figure 2.9: Density of states for Fe2 SiS4 calculated with Wien2k. The Fermi energy
is demarcated by the vertical dashed line at 0 eV. Contributions from p- or d -orbitals
are multiplied by the number of formula units in one unit
c) cell (4) before summing to
obtain the total contributions from Fe or S. Note the significant differences between
the ordinate scales of the partial DOS.
37
Si-S
Fe-S
Figure 2.10: Crystal orbital overlap population curves for Fe2 SiS4 calculated with
Caesar.
α (105 cm-1)
4
3
2
1
0
0
0.5
1
1.5
Energy (eV)
Figure 2.11: Absorption coefficient vs energy for Fe2 XS4 calculated with Wien2k.
38
Fe2 SiS4
Fe2 GeS4
Table 2.1: Unit cell parameters for Fe2 XS4 .
a (Å)
b (Å)
c (Å)
Volume (Å3 )
12.400(6) 7.190(3) 5.814(2)
518.3(4)
12.479(4) 7.221(3) 5.910(2)
532.6(3)
Table 2.2: EPMA data for Fe2 XS4 pellets.
X
O
S
Ideal stoichiometry
Fe
Fe
Fe
Fe2 SiS4
3.8 ± 0.5 1.0 ± 0.1
0.9 ± 0.6
Fe2 GeS4
3.9 ± 0.2 1.00 ± 0.07 not detected
CHAPTER 3
Spin coated iron hydroxo-oxalate and conversion
to FeS2 pyrite
40
3.1
Introduction
With a fundamental indirect optical band gap near 0.9 eV and an absorption
coefficient exceeding 105 cm−1 for wavelengths shorter than 1 µm,1–3 FeS2 pyrite
(hereafter pyrite) has the potential to be an excellent thin film solar absorber. Because
the absorption coefficient is so large, only 37 nm of pyrite is required to absorb 90%
of incident light,1 and therefore very little material is required in a working device.
Further economic incentives also drive the development. When compared with the
leading absorbers crystalline Si and thin film Cu(In,Ga)Se2 and CdTe, pyrite has a
significantly lower raw material extraction cost.4
Recognizing this potential, others have utilized vacuum deposition methods like
sputtering3, 5 and metal organic chemical vapor deposition6–8 to grow pyrite thin films.
While techniques requiring vacuum systems are useful in development, successful
manufacturing must cover large areas quickly, making solution processing an appealing alternative. Thin films of 1 Fe:1 S have been electrodeposited from aqueous solutions of Na2 S2 O3 and FeSO4 ,9 Fe(NH4 )2 (SO4 )2 ,10 or FeCl2 11 and single phase pyrite is
formed with post deposition S anneal. Aqueous FeSO4 and (NH4 )2 Sx were alternately
spray pyrolyzed in air to produce amorphous films which crystallized as pyrite, FeS2
marcasite, and lower order sulfides after a 400 ◦ C anneal in N2 (g).12 When these
films were annealed with H2 S (g) instead, single phase pyrite formed. With S vapor
present during deposition, spray pyrolysis of aqueous FeCl3 and thiourea produced
pyrite, but the film morphology suffered from pinholes and cracks.13 Screen printing
pyrite powder14 provides a simplified route for stoichiometric control, but the films
are too thick (∼100 µm) for use as solar absorbers. It is clearly difficult to produce
dense, single phase pyrite from solution-based methods.
Since naturally occurring single crystals of pyrite are common, the best approach
41
to dense, stoichiometric thin films may be along a route that mimics the growth
process of nature. Pyrite crystals likely form under hydrothermal conditions in the
Earth’s crust,1 where H2 S is a prevalent S-source.15 The decomposition of H2 S (g)
utilizing α-Fe2 O3 as a catalyst produced pyrite at the surface,16 while S vapor reacts
with Fe2 O3 , Fe3 O4 , or amorphous Fe oxide thin films to produce single phase pyrite
without proceeding through lower order Fe sulfides.17, 18 One approach to single phase
pyrite, then, begins with solution deposition of a dense, uniform Fe oxide film and
ends by reacting that film with H2 S (g).
Based on these ideas, we have deposited smooth, amorphous Fe oxalate films from
a simple aqueous precursor. This chemistry takes advantage of rapid condensation,
which we have recently utilized to deposit exceptionally dense, uniform Zr and Hf
oxide sulfate19 and Al oxide phosphate thin films.20 To obtain pyrite, we reacted the
Fe oxalate films under flowing H2 S (g) at elevated temperatures. The composition,
optical, and electrical properties of the resulting sulfurized films were then examined.
3.2
3.2.1
Experimental
Precursor synthesis
FeO(OH)·x H2 O was prepared by rapidly mixing 1.0 mL 2 M Fe(NO3 )3 (aq)
(prepared with solid Fe(NO3 )3 ·9 H2 O - Alfa Aesar, ACS, 98.0 - 101.0%) with 1.0 mL
3 M NH4 OH (aq) (EMD Chemicals, ACS). After centrifugation, the supernatant was
decanted. Additional H2 O was added, the combination was vigorously shaken, and
centrifugation was repeated. These rinse steps were repeated once more to remove
+
NO−
3 and NH4 . The supernatant was decanted after the final rinse, and the final
precipitate was dark reddish-brown solid.
42
Fresh FeO(OH)·x H2 O solid was re-dissolved in 3.5 mL 1 M H2 C2 O4 (aq) (prepared
with solid H2 C2 O4 ·2 H2 O - J. T. Baker). Approximately 15 min of vigorous shaking
was required to complete the dissolution. The final Fe oxalate precursor solution was
3+
transparent yellow and contained approximately 2 Fe3+ :3 C2 O2−
4 with [Fe ] = 0.2
M. All precursor solutions were used within several hours of preparation. Solution
preparation and precipitate rinsing were done with H2 O purified to 18 MΩ resistivity.
3.2.2
Thin film preparation and anneals
Fe oxalate films were prepared by spin coating. Each substrate was coated with
the precursor and then spun at 3000 rpm for 30 s to form a single coat. The film
was heated for 2 min in air on a 300 ◦ C hot plate after each coat. This process was
repeated five to ten times to build film thickness. Substrates included Corning 1737
glass, B-doped Si (0.02 Ω cm), and 200 nm thick SiO2 thermally grown on Si (Tox).
All substrates were cleaned by sonicating in a 5% Contrad 70 bath for 60 min at 45
◦
C and then rinsing with 18 MΩ H2 O.
Sulfurization was carried out by heating Fe oxalate films for 10 min at various
temperatures under flowing H2 S (g). H2 S (g) flow was maintained within the range
36 - 42 sccm through ramp up, dwell, and ramp down.
3.2.3
Thin film characterization
Polycrystalline phases in sulfurized Fe oxalate films were identified from X-ray
diffraction (XRD) data collected with Cu Kα radiation on a Rigaku Rapid diffractometer. Transmission Fourier transform infrared (FT-IR) spectra from a Fe oxalate
film on Tox were recorded on a Nicolet 5PC spectrometer, and the reference spectrum was collected from a bare substrate. The ratio of S to Fe was calculated from
43
electron probe micro analysis (EPMA) data collected from a pyrite film prepared at
300 ◦ C. Nine collinear positions were sampled with a Cameca SX-50 at three different
accelerating voltages (10, 15, and 20 kV). Elemental Si and natural pyrite were used
as standards for Si, Fe, and S, and MgO was used as the O standard. Quantitative
analysis was performed by comparing experimental k -ratios with simulated values in
Stratagem thin film composition analysis software.
Film morphology was evaluated with atomic force microscopy (AFM) and scanning electron microscopy (SEM). AFM scans were performed on 0.25 µm2 regions
of Fe oxalate and sulfurized Fe oxalate films with a Veeco Nanoscope. SEM images
were obtained for Fe oxalate films and pyrite films produced at 300 ◦ C. They were
collected on a ZEISS Ultra-55 microscope.
Room temperature electrical properties of pyrite films on 1737 glass sulfurized at
300 ◦ C were evaluated with Hall and Seebeck measurements. Hall data were collected
on Hall Measurement System 2.4.1 from Lake Shore Cryotronics, Inc, and an in-house
designed and built system was utilized for Seebeck data collection.
Optical direct and indirect band gaps were estimated from transmission through
and reflection from a pyrite film on 1737 glass produced at 300 ◦ C. Light from a
W bulb passed through a double monochromator (Oriel 77250) and then directed to
the film via a series of mirrors and lenses. A PbS detector was positioned to collect
reflected or transmitted light. Bare 1737 glass was used as the reference.
44
3.3
3.3.1
Results
Precursor chemistry and film crystallization
Strategies to take advantage of prompt condensation to produce smooth, dense
metal oxide films are predicated on the interaction of the metal ion with ligands in
the aqueous solution. The pH of the final precursor was 1 - 2, and in this range the
oxalate anion will exist primarily in its monoprotic form: [HC2 O4 ]− .21 As a result, it
is most likely to coordinate to the Fe3+ in a bidentate fashion similar to that shown
in Figure 3.1, the Fe-C2 O4 arrangement in the compound Fe2 (C2 O4 )3 ·x H2 O. While
the water content of this yellow-green amorphous solid can vary, the composition
Fe2 (C2 O4 )3 ·6 H2 O has been examined with IR and Raman spectroscopy.22 The two
Fe3+ ions are proposed to bridge three oxalate ligands, while the coordination of the
metal ions is completed by adjacent H2 O ligands.
The thermal decomposition of these Fe-C2 O4 molecules in air is an important
consideration for the conversion of Fe oxalate to the desired intermediate Fe oxide
film. The decomposition of Fe2 (C2 O4 )3 ·5 H2 O in air proceeds through several steps.23
Dehydration begins at approximately 150 ◦ C, and most of the oxalate decomposition
is completed by 250 ◦ C. Final crystallization as Fe2 O3 occurs near 325 ◦ C. Although
determining the chemical composition of a Fe oxalate film is beyond the scope of
◦
this study, it is unlikely that much C2 O2−
4 is retained after the 300 C curing step.
This is confirmed by the FT-IR spectrum of one Fe oxalate film, given in Figure 3.2.
Characteristic features of the oxalate group are present as broad, low intensity peaks
between 1300 and 1600 cm−1 ,24, 25 and the sharp peak near 2350 cm−1 originates from
Fe-O modes.26, 27 The even smaller peak at 1050 cm−1 is an artifact of a necessary
background subtraction to account for the substrate.
45
The decomposition of Fe2 (C2 O4 )3 ·5 H2 O in air delivers polycrystalline Fe2 O3 as
the final product. Fe oxalate films are amorphous, and only weak oxalate features
are present in FT-IR spectra. Based on these considerations and recalling that the
precursor contains 2 Fe3+ :3 C2 O2−
4 , the following reaction is proposed for the liquid
to solid film transformation:
2 Fe2 (C2 O4 )3 ·x H2 O (aq) + 3 O2 (g) → 2 Fe2 O3 (s) + 2x H2 O (g) + 12 CO2 (g)
In order to incorporate it into a working solar cell, any thin film absorber must
be smooth and dense. This is a particular concern for the potentially disruptive
process of converting a metal oxide to a sulfide because many bonds must be broken
and new ones formed. Thermal energy is necessary to activate these processes, but
that energy can also allow undesired rearrangement and grain growth. To investigate
temperature effects on the Fe oxalate conversion to pyrite, a series of reactions at
different temperatures under flowing H2 S (g) were examined. While several different
iron sulfide phases are observed in the XRD patterns (Figure 3.3), pyrite dominates
at and above 300 ◦ C. This process likely proceeds as follows:
Fe2 O3 (s) + 4 H2 S (g) → 2 FeS2 (s) + 3 H2 O (g) + H2 (g)
3.3.2
Morphology and stoichiometry
Thin film morphology is an important consideration for electrical applications
like solar energy absorption. The SEM cross-sectional image of a Fe oxalate film
(Figure 3.5) shows that it is smooth and dense with good substrate adhesion. The
crystallization coinciding with conversion to the final sulfide, however, causes significant roughening. Figure 3.4 provides root-mean-square (RMS) roughness vs anneal
temperature data collected via AFM. Because the RMS value doubled when the H2 S
46
reaction temperature increased from 300 to 400 ◦ C, films treated at 300 ◦ C were
chosen for all further characterization. The SEM cross-sectional image of one such
film is given in Figure 3.6. Despite the relatively low RMS roughness value, it is
obvious that the film reacted with H2 S (g) at 300 ◦ C is still porous and somewhat
delaminated from the substrate. The porosity in particular would allow short circuits
to form through the pyrite layer, posing significant barriers to successful integration
into a functional device.
The reported S/Fe of pyrite single crystals varies between 1.74 and 2.10, although
many of these values are taken from chemical analyses that required sample dissolution.28 Side products of typical acid-assisted dissolution are volatile species like
H2 S or SO2 , and loss of these gases would produce a smaller S/Fe than was actually present in the original crystal. Despite this uncertainty in the real stoichiometry
range of pyrite, what is clear is that open circuit voltages in photoelectrochemical cells
are significantly lower than that expected on the basis of the optical band gap.29, 30
Both S- and Fe-vacancies have been proposed to create defect states inside the band
gap that pin the Fermi level and lead to the drop in open circuit voltage,28, 31–33 so
stoichiometry control of pyrite thin films is a central concern. EPMA data collected
from a pyrite film prepared as part of this study produced a S/Fe of 1.99 ± 0.02, in
good agreement with the expected value. No O was detected. These data demonstrate the efficiency of the relatively short 10 minute H2 S (g) reaction, in contrast
with traditional sol-gel deposited Fe oxide films that were heated with S powder in
evacuated tubes for 20 h. The S to Fe ratio reported for these films was 1.9.34
47
3.3.3
Electrical and optical properties
Independent of deposition technique, pyrite films are typically p-type,2 and the
films in this study are no exception. A typical Seebeck coefficient for the pyrite films
discussed here is 60 µV/K, where the positive sign indicates that holes are indeed the
majority carrier. Such character will occur when defects deplete electrons from the
Fe 3d -t2g levels in the valence band, and Fe-vacancies are an obvious candidate.
A typical resistivity measured is 3·10−1 Ω-cm. The sign of the Hall coefficient
varied from positive to negative during measurements on individual films, indicating
that the pyrite films produced from Fe oxalate have mobilities below the detection
limit of the instrument. Such behavior has been reported for pyrite thin films synthesized by a variety of techniques,35 and low mobilities are consistent with the porous
morphology observed in Figure 3.6.
The wavelength-dependent absorption coefficient α can be extracted from transmission (T) and reflection (R) data by using the relationship T/(1-R) = e −αd , when
the film thickness d is known. From Figure 3.6, a typical pyrite film produced from
Fe oxalate is 130 nm thick. Optical data collected from a similar film are plotted as
(αE)2 or (αE)1/2 vs E in Figure 3.7. By extrapolating the steepest portion of each
curve to the abscissa, we find an indirect optical gap of 0.90 eV and a direct gap just
below 1.4 eV. These are consistent with reported values1, 17, 36, 37
3.4
Summary and Perspective
We have deposited smooth, amorphous Fe oxalate films from an aqueous precursor
and examined the conversion of these films to FeS2 pyrite via reaction with flowing
H2 S (g). The 130 nm thick pyrite films prepared at 300 ◦ C are p-type with an optical
indirect band gap of 0.9 eV. While these pyrite films are stoichiometrically satisfactory
48
(FeS1.99±0.02 ), their coarse morphology makes them poor candidates for incorporation
into a working thin film solar cell.
49
References
[1] A. Ennaoui, S. Fiechter, C. Pettenkofer, N. Alonso-Vante, K. Buker,
M. Bronold, C. Hopfner, and H. Tributsch. Iron disulfide for solar energy
conversion. Solar Energy Materials and Solar Cells, 29:289 – 370, 1993.
[2] P. Altermatt, T. Kiesewetter, K. Ellmer, and H. Tributsch. Specifying targets
of future research in photovoltaic devices containing pyrite (FeS2 ) by numerical
modelling. Solar Energy Materials and Solar Cells, 71:181 – 195, 2002.
[3] G. Willeke, R. Dasbach, B. Sailer, and E. Bucher. Thin pyrite (FeS2 ) films
prepared by magnetron sputtering. Thin Solid Films, 213:271 – 276, 1992.
[4] C. Wadia, A. Alivisatos, and D. Kammen. Materials availability expands the
opportunity for large-scale photovoltaics deployment. Environmental Science &
Technology, 43:2072 – 2077, 2009.
[5] D. Lichtenberger, K. Ellmer, R. Schieck, and S. Fiechter. Optical, electrical and
structural properties of polycrystalline iron-pyrite(FeS2 ) layers deposited by
reactive DC magnetron sputtering. Applied Surface Science, 70:583–587, 1993.
[6] B. Thomas, K. Ellmer, M. Muller, C. Hopfner, S. Fiechter, and H. Tributsch.
Structural and photoelectrical properties of FeS2 (pyrite) thin films grown by
MOCVD. Journal of Crystal Growth, 170(1):808–812, 1997.
[7] B. Meester, L. Reijnen, A. Goossens, and J. Schoonman. Synthesis of pyrite
(FeS2 ) thin films by low-pressure MOCVD. Chemical Vapor Deposition,
6(3):121 – 128, 2000.
[8] L. Reijnen, B. Meester, A. Goossens, and J. Schoonman. In situ mass
50
spectrometric study of pyrite (FeS2 ) thin film deposition with metallorganic
chemical vapor deposition. Journal of the Electrochemical Society, 147:1803,
2000.
[9] S. Nakamura and A. Yamamoto. Electrodeposition of pyrite (FeS2 ) thin films
for photovoltaic cells. Solar Energy Materials & Solar Cells, 65:79 – 85, 2001.
[10] A. Gomes, J. Ares, I. Ferrer, M. da Silva Pereira, and C. Sanchez. Formation of
n-type pyrite films from electrodeposited iron sulphides: effect of annealing
temperature. Materials Research Bulletin, 38(7):1123–1133, 2003.
[11] Y. Dong, Y. Zheng, H. Duan, Y. Sun, and Y. Chen. Formation of pyrite (FeS2 )
thin nano-films by thermal-sulfurating electrodeposition films at different
temperature. Materials Letters, 59(19-20):2398–2402, 2005.
[12] A. Yamamoto, M. Nakamura, A. Seki, E. Li, A. Hashimoto, and S. Nakamura.
Pyrite (FeS2 ) thin films prepared by spray method using FeSO4 and (NH4 )2 sx .
Solar Energy Materials & Solar Cells, 75:451 – 456, 2003.
[13] G. Smestad, A. da Silva, H. Tributsch, S. Fiechter, M. Kunst, N. Meziani, and
M. Birkholtz. Formation of semiconducting iron pyrite by spray pyrolysis.
Solar Energy Materials, 18:299 – 313, 1989.
[14] V. Antonucci, A. Arico, N. Giordano, P. Antonucci, U. Russo, D. Cocke, and
F. Crea. Photoactive screen-printed pyrite anodes for electrochemical
photovoltaic cells. Solar Cells, 31:119 – 141, 1991.
[15] M. Schoonen and H. Barnes. Mechanisms of pyrite and marcasite formation
from solution: 3. Hydrothermal processes. Geochimica et Cosmochimica Acta,
55:12, 1991.
51
[16] C. Kim, A. Escuadro, and M. Bedzyk. Interaction of H2 S with α-Fe2O3 (0001)
surface. Surface Science, 601(21):4966 – 4970, 2007.
[17] G. Smestad, A. Ennaoui, S. Fiechter, H. Tributsch, W. Hofmann, M. Birkholtz,
and W. Kautek. Photoactive thin film semiconducting iron pyrite prepared by
sulfurization of iron oxides. Solar Energy Materials, 20:149 – 165, 1990.
[18] B. Ouertani, J. Ouerfelli, M. Saadoun, B. Bessaı̈s, H. Ezzaouia, and J Bernède.
Characterization of FeS2 -pyrite thin films synthesized by sulphuration of
amorphous iron oxide films pre-deposited by spray pyrolysis. Materials
Characterization, 54(4-5):431–437, 2005.
[19] J. Anderson, C. Munsee, C. Hung, T. Phung, G. Herman, D. Johnson,
J. Wager, and D. Keszler. Solution-processed HafSOx and ZircSOx inorganic
thin-film dielectrics and nanolaminates. Advanced Functional Materials,
17:2117 – 2124, 2007.
[20] S. Meyers, J. Anderson, D. Hong, C. Hung, J. Wager, and D. Keszler.
Solution-processed aluminum oxide phosphate thin-film dielectrics. Chemistry
of Materials, 19:4023 – 4029, 2007.
[21] D. Panias, M. Taxiarchou, I. Douni, I. Paspaliaris, and A. Kontopolous.
Thermodynamic analysis of the reactions of iron oxides: dissolution in oxalic
acid. Canadian Metallurgical Quarterly, pages 363 – 373, 1996.
[22] H. Edwards and N. Russell. Vibrational spectroscopic study of iron (II) and
iron (III) oxalates. Journal of Molecular Structure, 443(1-3):223–231, 1998.
[23] D. Broadbent, D. Dollimore, and J. Dollimore. The thermal decomposition of
52
oxalates. Part IX. the thermal decomposition of the oxalate complexes of iron.
Journal of the Chemical Society A, 1967:451 – 454, 1967.
[24] A. Plostinaru and L. Patron. Coordination compounds as raw materials for
mixed oxides. IX mixed complex compounds of lanthanides and Fe(III) with
oxalic acid. Revue Roumaine de Chemie, 43:601 – 606, 1998.
[25] V. Berbenni, A. Marini, G. Bruni, and R. Riccardi. Solid state reaction study
in the systems Li2 CO3 -FeC2 O4 · 2 H2 O and Li2 CO3 -Fe2 (C2 O4 )3 · 6 H2 O.
Thermochimica Acta, 346:115 – 132, 2000.
[26] B. Weckler and H. Lutz. Lattice vibration spectra. Part XCV. Infrared
spectroscopic studies on the iron oxide hydroxides goethite (α), akaganéite (β),
lepidocrocite (γ), and feroxyhite (δ). European Journal of Solid State and
Inorganic Chemistry, 35(8-9):531–544, 1998.
[27] R. Nyquist and R. Kagel. Infrared Spectra of Inorganic Compounds. Academic
Press, 1971.
[28] K. Ellmer and C. Hopfner. On the stoichiometry of the semiconductor pyrite
(FeS2 . Philosophical Magazine A, 75:1129 – 1151, 1997.
[29] K. Buker, N. Alonso-Vante, and H. Tributsch. Photovoltaic output limitation
of n-FeS2 (pyrite) schottky barriers: A temperature-dependent characterization.
Journal of Applied Physics, 72:5721 – 5728, 1992.
[30] K. Mishra and K. Osseo-Asare. Fermi level pinning at pyrite (FeS2 )/electrolyte
junctions. Journal of the Electrochemical Society, 139:749 – 752, 1992.
53
[31] M. Birkholz, S. Fiechter, A. Hartmann, and H. Tributsch. Sulfur deficiency in
iron pyrite (FeS2−x ) and its consequences for band-structure models. Physical
Review B, 43:11926 – 11936, 1991.
[32] M. Bronold, C. Pettenkofer, and W. Jaegermann. Surface photovoltage
measurements on pyrite (100) cleavage planes: Evidence for electronic bulk
defects. Journal of Applied Physics, 76:5800 – 5808, 1994.
[33] J. Ares, A. Pascual, I. Ferrer, and C. Sanchez. Lattice intrinsic defects and
electrical resistivity in pyrite thin films. Thin Solid Films, 451:233–236, 2004.
[34] H. Siyu, L. Xinyu, L. QingYu, and C. Jun. Pyrite film synthesized for
lithium-ion batteries. Journal of Alloys and Compounds, 472(1-2):9–12, 2009.
[35] J. Ares, I. Ferrer, and C. Sánchez. Majority carriers in pyrite thin films: an
analysis based on Seebeck and Hall coefficient measurements. Thin Solid Films,
431:511–513, 2003.
[36] G. Zhao, J. Callaway, and M. Hayashibara. Electronic structures of iron and
cobalt pyrites. Physical Review B, 48:15781 – 15786, 1993.
[37] V. Eyert, K. Hock, S. Fiechter, and H. Tributsch. Electronic stucture of (FeS2 ):
The crucial role of electron-lattice interaction. Physical Review B, 57:6350 –
6359, 1998.
54
Figure 3.1: Coordination of Fe3+ by C2 O2−
4 in the Fe2 (C2 O4 )3 molecule.
Normalized
Transmittance (A. U.)
100
80
60
Fe-O
40
C2 O4
20
0
4,000
3,000
2,000
Wavenumbers (cm-1)
1,000
Figure 3.2: FT-IR spectrum collected from a Fe oxalate thin film.
55
Intensity (A.U.)
200
150
100
50
0
10
20
30
40
50
2θ (degrees)
Figure 3.3: XRD data collected from Fe oxalate thin films heated under flowing
H2 S (g) at the specified temperatures. Dashed arrows represent FeS mackinawite,
and solid arrows represent FeS2 marcasite.
Roughness (nm)
12
10
8
6
4
2
0
0
100
200
300
Temperature (0C)
400
Figure 3.4: RMS roughness vs anneal temperature calculated from AFM scans of
Fe oxalate thin films heated under flowing H2 S (g) at the specified temperatures.
56
Fe oxalate
130 nm
Si
Figure 3.5: SEM cross-sectional image of a Fe oxalate thin film.
pyrite
130 nm
Si
Figure 3.6: SEM cross-sectional image of a pyrite thin film.
(αE)2 (cm-1·eV)2
57
1011
5x1010
1.6
Energy (eV)
0
0.4 0.6 0.8
1
1.2 1.4 1.6
Energy (eV)
Figure 3.7: Direct and indirect (inset) optical band gap estimates for a pyrite film.
CHAPTER 4
Conversion of Fe hydroxide nitrate thin films to FeS2 pyrite
Heather A. S. Platt, Alan Telecky, Jason Stowers, Josh Russell,
Janet Tate, Douglas A. Keszler
To be submitted for publication in Thin Solid Films.
59
4.1
Introduction
FeS2 pyrite (hereafter pyrite) has often garnered attention, first in its mineral
form because its metallic luster mimics that of gold, and more recently for its optical
and electrical properties. It has been studied as a catalyst for oxygen reduction in
fuel cells,1 as a thermoelectric,2 and as an electrode in Li-ion batteries,3 but the
majority of the reported development has focused on producing thin films to act
as solar absorbers. With a fundamental indirect optical band gap near 0.9 eV and
an absorption coefficient exceeding 105 cm−1 for wavelengths shorter than 1 µm,4–6
only 37 nm of pyrite is required to absorb 90% of incident light.4 Not only is very
little pyrite required in a working device, but it has a significantly lower raw material
extraction cost than the leading absorbers crystalline Si and thin film Cu(In,Ga)Se2
and CdTe.7
Successful thin film manufacturing must cover large areas quickly, so solution
processing would be an ideal method to deposit pyrite. S-sources such as thiourea,
Na2 S2 O3 , and (NH4 )2 Sx and the Fe salts FeSO4 , FeCl2 , FeCl3 , and Fe(NH4 )2 (SO4 )2
have been employed in aqueous solutions for electrodeposition8–10 or spray pyrolysis.11, 12 None of these methods produced single phase pyrite thin films without a
post-deposition sulfurization or additional S vapor present during the deposition.
Screen printing pyrite powder13 provides a simplified route for stoichiometric control,
but the films are too thick (∼100 µm) for use as solar absorbers. It is clearly difficult
to produce true thin films of dense, single phase pyrite with solution-based methods.
Since naturally occurring single crystals of pyrite are common, the best approach
to dense, stoichiometric thin films may be along a route that mimics the growth
process observed in nature. Pyrite crystals likely form under hydrothermal conditions
in the Earth’s crust,4 where H2 S is a prevalent S-source.14 The decomposition of H2 S
60
(g) utilizing α-Fe2 O3 as a catalyst produced pyrite at the surface,15 while S vapor
reacts with Fe2 O3 , Fe3 O4 , or amorphous Fe oxide thin films to produce single phase
pyrite without proceeding through lower order Fe sulfides.16, 17 One approach to single
phase pyrite, then, begins with solution deposition of a dense, uniform Fe oxide film
and ends by reacting that film with H2 S (g).
Based on controlled hydrolysis, we have developed a simple aqueous precursor
that can be spin coated in air to form dense, amorphous iron hydroxide nitrate (IHN)
films. The deposition chemistry takes advantage of rapid condensation, which we
have recently utilized to spin coat exceptionally uniform and smooth Zr and Hf oxide
sulfate18 and Al oxide phosphate thin films.19 To obtain pyrite, we reacted the IHN
films under flowing H2 S (g) at elevated temperatures. The composition, optical, and
electrical properties of the resulting sulfurized films were then examined.
4.2
4.2.1
Experimental Methods
Precursor synthesis and characterization
Solid Fe(NO3 )3 ·9 H2 O (Alfa Aesar, ACS, 98.0 - 101.0%) and Fe metal powder (Alfa
Aesar, 99+%) were dissolved and aged with stirring in a mixture of 80% methanol
(Mallinckrodt, ACS, anhydrous) and 20% H2 O. The initial NO3 /Fe was ∼2 with
[Fe] ∼ 0.2 M. After two days of aging, the transparent red IHN solution was dried
under flowing Ar (g) for another day to remove the methanol. The resulting gel was
redissolved in H2 O and used immediately. Solution preparation and gel re-hydration
were carried out with H2 O purified to 18 MΩ resistivity.
The decomposition of the intermediate IHN gel was evaluated with thermogravimetric analysis (TGA) in stagnant air. The sample was placed in an alumina boat,
61
and heated at 10 ◦ C/min in a Mettler Toledo TGA850. Freshly prepared gel was also
heated first in air at 150 ◦ C for 60 or 120 min, and then in flowing H2 S (g) at temperatures between 200 and 600 ◦ C for 120 min. X-ray diffraction (XRD) was performed
on the powders with Cu Kα radiation on a Rigaku MiniFlex II diffractometer.
4.2.2
Thin film preparation and analysis
IHN films were prepared by spin coating the precursor solution at 3000 rpm for 30
s to form a single coat. The film was heated for 1 min in air on a 150 ◦ C hot plate after
each coat. The spin coating and heating were repeated to build film thickness; these
films will be labeled “as-cured” in the remainder of this paper. Substrates included
Corning 1737 glass, B-doped Si (ρ = 0.02 Ω cm), and 200 nm thick SiO2 thermally
grown on Si (Tox). All substrates were cleaned by sonicating in a 5% Contrad 70
bath for 60 min at 45 ◦ C and then rinsing with 18 MΩ H2 O.
To evaluate the effects of hydration on the sulfide phase formed, some IHN films
were further heated in air (“air-heated” henceforth) at temperatures between 300 and
500 ◦ C for 30 or 60 min. Sulfurization was carried out by heating as-cured and airheated IHN films for 10 to 60 min at various temperatures under flowing or stagnant
H2 S (g). During the flowing reactions, H2 S (g) flow was maintained within the range
36 - 42 sccm through ramp up, dwell, and ramp down. Polycrystalline phases in
sulfurized IHN films were identified from XRD data collected with Cu Kα radiation
on a Rigaku Rapid diffractometer.
Transmission Fourier transform infrared (FT-IR) spectra were collected from ascured and air-heated IHN films on Tox using a Nicolet 5PC spectrometer, with the
reference spectrum collected from a bare substrate. The ratios of S or O to Fe
were calculated from electron probe micro analysis (EPMA) data collected from as-
62
cured and sulfurized IHN films. At least ten collinear positions were sampled with
a Cameca SX-100 microprobe at three different accelerating voltages (10, 15, and 20
kV). Elemental Si and natural pyrite were used as standards for Si, Fe, and S; MgO
was used as the O standard; and AlN was used as the N standard. Quantitative
analysis was performed by comparing experimental k -ratios with simulated values in
Stratagem thin film composition analysis software.
Film morphology was evaluated with scanning electron microscopy (SEM). Images
were collected from as-cured, air-heated, and sulfurized IHN films on a ZEISS Ultra-55
microscope.
Optical absorption coefficients and band gaps were determined from transmission
and reflectance data collected with UV-visible (HR4000) and IR (NIR256) detectors
from Ocean Optics. The light was generated by an Ocean Optics deuterium tungsten
halogen source (DH-2000-BAL), and the reflection standard was OO STAN-SSH.
Room temperature electrical properties of the sulfurized films were evaluated with
Hall and Seebeck measurements. The Hall measurements were performed on Hall
Measurement System version 2.4.1 from Lake Shore Cryotronics, Inc., and the Seebeck
measurements were performed on either an ULVAC ZEM-3 Seebeck Coefficient and
Electrical Resistivity Measuring Instrument or a system designed and built in-house.
4.3
4.3.1
Results and Discussion
Bulk precursor chemistry
The dissolution of Fe metal in a partially aqueous environment exposed to air
will almost certainly produce Fe3+ , and the red color of the precursor throughout
aging, drying, and re-dissolution is characteristic of Fe3+ hydroxide species. Since Fe
63
is being oxidized, another species in the solution must be reduced. The pH of the
precursor at the end of the aging process and after re-dissolution is 2-3, so H+ ions
3+
are one possibility. The other obvious choice is NO−
3 . Since the Fe /Fe standard
reduction potential is -0.04 V and the standard reduction potential for the NO−
3 /NO
half reaction is 0.96 V, it is certainly possible for NO−
3 to act as the necessary oxidizing
agent in the precursor solution. This process may be simplified as
Fe (s) + Fe(NO3 )3 (s) + 2 H2 O (l) → Fe2 (NO3 )2 (OH)4 (aq) + NO (g).
If H+ , instead of NO−
3 , acts as the oxidizing agent, the dissolution would be
Fe (s) + Fe(NO3 )3 (s) + 3 H2 O (l) → Fe2 (NO3 )3 (OH)3 (aq) +
3
2
H2 (g).
In both cases, the Fe-containing species are clusters containing both the nitrate and
hydroxide ions, and the NO3 /Fe that are quite different from known Fe nitrate salts.
As is the case for most metals, Fe3+ hydrolyzes in aqueous solutions, and any
other ions present will significantly impact that process. NO−
3 was chosen for this
study for several important reasons. First, it decomposes readily in air to form
volatile products, which makes it a good leaving group. Second, IR spectra collected
from the hydroxide precipitates of Fe(NO3 )3 (aq) solutions indicate that the Fe3+
20
is not directly coordinated by NO−
so the primary anion sphere will be H2 O or
3,
OH− . This configuration is ideal for the rapid condensation that forms dense M oxide
films. Finally, electron microscopy-based estimates of the particle sizes of the polymer
clusters in nitrate solutions aged less than a few days old are 2 - 4 nm diameter.20
Such small particles will be very reactive, so little activation energy will be required
to begin the condensation process required to form a solid film.
The conversion of the liquid precursor to a solid film requires dehydration and
decomposition of NO−
3 . TGA of dried IHN gels revealed weight loss beginning about
64
140 ◦ C and ending by 250 ◦ C, both of which are approximately 50 ◦ C higher than
the corresponding regions in the TGA decomposition of Fe(NO3 )3 ·9H2 O.21 While
the 150 ◦ C curing temperature used to deposit IHN films is at the lower end of the
decomposition range, it is still high enough to expect significant nitrate decomposition
to occur. In fact, no N was detected in as-cured IHN films by EPMA. The FT-IR
data collected from an as-cured IHN film (Figure 4.1) provide additional support. The
broad peak centered at 3400 cm−1 originates from O-H stretching modes and the small
peak at 1300 cm−1 comes from O-H vibrational features in the FeO(OH) phases.22 HO-H bending modes produce the small broad peak at 1500−1 , and the slightly jagged
feature between 1100 and 1000 cm−1 is an artifact of the background correction to
account for the substrate. While the as-cured film clearly retains hydroxide and some
water, there is no detectable nitrate. Based on these considerations and assuming
that NO−
3 is the primary oxidizing agent because it is thermodynamically favored
over H+ , the following reaction is proposed for the liquid to solid film transformation:
2 Fe2 (NO3 )2 (OH)4 (aq) → 4 FeO(OH) (s) + 2 N2 O5 (g) + 2 H2 O (g).
FeO(OH) was chosen as the primary Fe-containing species in the IHN films because αFeO(OH) is the final product of hydroxide-driven precipitation from Fe(NO3 )3 (aq).20
Pyrite and FeS2 marcasite are often found together in natural samples, and their
Gibbs free energies of formation are very similar.23 Pyrite is the more thermodynamically stable phase, but it is possible that marcasite could form in small domains.
Lower sulfides such as Fe7 S8 pyrrhotite may also be intermediates in pyrite formation.
We examined a series of bulk IHN gel samples to determine where single phase pyrite
forms. When IHN gel was heated at 150 ◦ C for 60 min in air and then reacted with
flowing H2 S (g), single phase pyrite is observed only at 500 ◦ C (Figure 4.2). Below
this temperature, both pyrite and marcasite form. Similar experiments with freshly
65
dried IHN gel also showed both pyrite and marcasite at lower temperatures. When
the IHN gel was heated at 150 ◦ C for 120 min before reacting with H2 S (g), single
phase pyrite formed at 400 and 500 ◦ C as shown in Figure 4.3. The longer heating
time in air will drive more H2 O out of the IHN gel, and this may increase the material’s density considerably. More open, disordered solids like the dried IHN gel may
provide more opportunity for rearrangement and the formation of multiple phases.
What is somewhat unexpected is that Figure 4.2 shows pyrite and pyrrhotite are both
present at 600 ◦ C, but no pyrrhotite is observed at lower temperatures. Pyrite does
not decompose until 743 ◦ C,4 so the flowing gas must be pulling sulfur back out of
the films in a process similar to
7 FeS2 (s) + 6 H2 (g) → Fe7 S8 (s) + 6 H2 S (g).
Based on the reactions of the IHN gels with H2 S (g), we chose to focus on the
temperature range of 300 - 600 ◦ C for the films. We also considered the difference
between films of the more hydrated IHN precursor and those heated in air which
crystallize as Fe2 O3 . The target reactions for the amorphous IHN and Fe2 O3 film
conversions to pyrite are
2 FeO(OH) (s) + 4 H2 S (g) → 2 FeS2 (cr) + 4 H2 O (g) + 2 H2 (g)
and
Fe2 O3 (cr) + 4 H2 S (g) → 2 FeS2 (cr) + 3 H2 O (g) + H2 (g).
4.3.2
Film morphology and stoichiometry
When sub-10 nm metal hydroxide clusters form in aqueous solution, little activation energy is required to initiate rapid condensation. Figure 4.4 shows that this
66
strategy effectively produced dense IHN films, analogous to our recent success with
ink-jet printed ZnO.24 When the initially amorphous IHN films are further heated
in air, they begin to crystallize as Fe2 O3 at 300 ◦ C. The FT-IR data in Figure 4.1
show that by 400 ◦ C the O-H features at 3400 and 1300 cm−1 and the H-O-H bending modes at 1500−1 are gone, indicating that the air-dried IHN films are largely
dehydrated. Further, translational lattice modes of octahedrally coordinated Fe3+ in
α-Fe2 O3 produce features at 560 cm−1 , and Fe-O stretches in the FeO(OH) phases
give rise to features between 400 and 500 cm−1 .22 These features show up as two
sharp peaks in the IR spectrum of the 400 ◦ C air-heated IHN film (Figure 4.1), so
after this treatment the films are probably still a combination of crystalline Fe2 O3
and amorphous FeO(OH). The top-down SEM image of such a film in Figure 4.5b
shows that grain growth has occurred, and the cross-sectional image in Figure 4.5a
demonstrates that these grains are still densely packed.
Figures 4.6 and 4.7 reveal that reacting both as-cured and air-heated IHN films
with H2 S produces FeS2 at lower temperatures. The bulk reactions (Figures 4.2 and
4.3) supply evidence that at 300 ◦ C the sulfurized films are probably a combination of
the pyrite and marcasite phases, and at 400 ◦ C single phase pyrite can be expected.
These observations are consistent with reports of S vapor reacting with Fe oxide
thin films to produce single phase pyrite without proceeding through lower order
Fe sulfides,16, 17 and studies of H2 S reactions with Fe films that found final pyrite
formation did proceed through a Fe1−x S phase.25 As with the bulk reactions, the
lower order sulfide pyrrhotite forms at higher temperatures. Since pyrite is the only
phase observed at 400 ◦ C, all further characterization was carried out on films reacted
with H2 S at that temperature.
SEM images of pyrite films produced by heating as-cured IHN at 400 ◦ C in H2 S
67
show that slightly nonuniform crystallization occurs during the reaction. The topdown image in Figure 4.8b reveals that ridges of slightly larger crystallites form around
flatter regions, although the cross-sectional image in 4.8a demonstrates that those
grains are still tightly packed together. Figure 4.9 provides analogous images for
an air-heated and sulfurized IHN film, and it is striking that this film crystallized
more uniformly. The water and hydroxide remaining in the as-cured IHN film clearly
impact the crystallization process, possibly because there are regions where these
species are concentrated. EPMA data also show slight stoichiometry differences. For
pyrite films synthesized from as-cured IHN, S/Fe = 2.11 ± 0.09 and O/Fe = 0.19 ±
0.08, while no oxygen was detected in and S/Fe = 1.98 ± 0.01 for a sulfurized 400
◦
C, 30 min air-heated IHN film.
4.3.3
Film electrical and optical properties
To evaluate the temperature-dependence of the H2 S reactions, both XRD and Hall
measurements were performed on as-cured and air-heated IHN films sulfurized over
the range 150 to 600 ◦ C. When sulfurization was carried out near 200 ◦ C, the films
were amorphous and their Hall coefficients were all positive. Measured mobilities
were in the range 1 - 5 cm2 /V s, and carrier concentrations were on the order of 1021
cm−3 . Morphologically, these films looked very similar to as-cured IHN films: dense
without any obvious grains. After reactions at 300 ◦ C, the temperature where pyrite
begins to crystallize, the signs of the film Hall coefficients became inconsistent, which
is indicative of mobilities too small to reliably measure. These sign changes persist
for all films sulfurized at higher temperatures, including the single phase pyrite films
at 400 ◦ C. The beginning film character, as-cured or air-heated IHN, while clearly
influencing the pyrite morphology did not have an effect on the Hall measurements. As
68
the SEM images in Figures 4.8 and 4.9 show, single phase pyrite films are made up of
small, tightly packed grains. Grain boundaries can certainly disrupt carrier movement
and therefore significantly decrease mobility, which would produce inconsistencies
in the sign of the Hall coefficient. Variation of the H2 S flow rate and increased
reaction times to deliberately discourage or encourage grain growth did not produce
reliable Hall measurements, and in some cases caused the films to delaminate from
the substrates. A typical resistivity measured from the single phase pyrite films is
0.12 Ω-cm, and the usual Seebeck coefficients are positive with magnitudes near 50
µV/K.
The pyrite films reported here are clearly p-type, which seems to be a universal
feature of synthetic films independent of deposition technique.5 Such character will
occur when defects deplete electrons from the Fe 3d -t2g levels in the valence band, and
Fe-vacancies are an obvious candidate. Both S- and Fe-vacancies have been proposed
to create defect states inside the band gap26–29 that pin the Fermi level and produce
open circuit voltages in photoelectrochemical cells that are significantly lower than
expected on the basis of the optical band gap.30, 31 Variation in the sign of the Hall
coefficient, while the Seebeck coefficient is consistently positive, has also been reported
for many pyrite thin films.32 The electrical properties of the pyrite films described
here are clearly similar to many others deposited by a variety of techniques.
The optical data collected from these pyrite films provide one final method to evaluate their quality. Figure 4.10 provides the wavelength-dependent optical absorption
coefficient, which reaches 105 cm−1 by 1.5 eV (∼830 nm). The inset shows the indirect transition of these pyrite films, and this value is approximately 0.95 eV. As
with the electrical characteristics, the magnitudes of the indirect gap and absorption
coefficient are consistent with literature values.4
69
4.4
Summary and Perspective
In this contribution, we report the solution deposition of iron hydroxide nitrate
thin films from a simple aqueous precursor. As-cured, these morphologically dense
films contain significant amounts of hydroxide and water, and when heated in air
they crystallize as densely-packed Fe2 O3 . Both as-cured and air-heated IHN films are
converted to FeS2 pyrite with heating under H2 S at 400 ◦ C for 10 min. The pyrite
films synthesized from as-cured IHN have S/Fe = 2.11 ± 0.09 and contain a small
amount of oxygen, while the pyrite films produced from air-heated IHN are oxygenfree with S/Fe = 1.98 ± 0.01. Morphologically, these pyrite films are all dense, but the
films synthesized from air-heated IHN exhibit more uniform grains. While we were
not able to confirm mobilities or carrier concentrations, the promising morphology
of these films hints that efforts should continue for the production of 100 nm thick
pyrite films for studies to evaluate their defect chemistry and effects on Fermi-level
pinning.
70
References
[1] D. Susac, L. Zhu, M. Teo, A. Sode, K. Wong, P. Wong, R. Parsons,
D. Bizzotto, K. Mitchell, and S. Campbell. Characterization of FeS2 -based thin
films as model catalysts for the oxygen reduction reaction. Journal of Physical
Chemistry C, 111:18715 – 18723, 2007.
[2] K. Kato, Y. Okamoto, J. Morimoto, and T. Miyakawa. The thermoelectric
properties of FeS2 . Journal of Materials Science Letters, 16(11):914–916, 1997.
[3] V. Livshits, A. Blum, E. Strauss, G. Ardel, D. Golodnitsky, and E. Peled.
Development of a bipolar li/composite polymer electrolyte/pyrite battery for
electric vehicles. Journal of Power Sources, 97-98:782 – 785, 2001.
[4] A. Ennaoui, S. Fiechter, C. Pettenkofer, N. Alonso-Vante, K. Buker,
M. Bronold, C. Hopfner, and H. Tributsch. Iron disulfide for solar energy
conversion. Solar Energy Materials and Solar Cells, 29:289 – 370, 1993.
[5] P. Altermatt, T. Kiesewetter, K. Ellmer, and H. Tributsch. Specifying targets
of future research in photovoltaic devices containing pyrite (FeS2 ) by numerical
modelling. Solar Energy Materials and Solar Cells, 71:181 – 195, 2002.
[6] G. Willeke, R. Dasbach, B. Sailer, and E. Bucher. Thin pyrite (FeS2 ) films
prepared by magnetron sputtering. Thin Solid Films, 213:271 – 276, 1992.
[7] C. Wadia, A. Alivisatos, and D. Kammen. Materials availability expands the
opportunity for large-scale photovoltaics deployment. Environmental Science &
Technology, 43:2072 – 2077, 2009.
71
[8] S. Nakamura and A. Yamamoto. Electrodeposition of pyrite (FeS2 ) thin films
for photovoltaic cells. Solar Energy Materials & Solar Cells, 65:79 – 85, 2001.
[9] A. Gomes, J. Ares, I. Ferrer, M. da Silva Pereira, and C. Sanchez. Formation of
n-type pyrite films from electrodeposited iron sulphides: effect of annealing
temperature. Materials Research Bulletin, 38(7):1123–1133, 2003.
[10] Y. Dong, Y. Zheng, H. Duan, Y. Sun, and Y. Chen. Formation of pyrite (FeS2 )
thin nano-films by thermal-sulfurating electrodeposition films at different
temperature. Materials Letters, 59(19-20):2398–2402, 2005.
[11] A. Yamamoto, M. Nakamura, A. Seki, E. Li, A. Hashimoto, and S. Nakamura.
Pyrite (FeS2 ) thin films prepared by spray method using FeSO4 and (NH4 )2 sx .
Solar Energy Materials & Solar Cells, 75:451 – 456, 2003.
[12] G. Smestad, A. da Silva, H. Tributsch, S. Fiechter, M. Kunst, N. Meziani, and
M. Birkholtz. Formation of semiconducting iron pyrite by spray pyrolysis.
Solar Energy Materials, 18:299 – 313, 1989.
[13] V. Antonucci, A. Arico, N. Giordano, P. Antonucci, U. Russo, D. Cocke, and
F. Crea. Photoactive screen-printed pyrite anodes for electrochemical
photovoltaic cells. Solar Cells, 31:119 – 141, 1991.
[14] M. Schoonen and H. Barnes. Mechanisms of pyrite and marcasite formation
from solution: 3. Hydrothermal processes. Geochimica et Cosmochimica Acta,
55:12, 1991.
[15] C. Kim, A. Escuadro, and M. Bedzyk. Interaction of H2 S with α-Fe2O3 (0001)
surface. Surface Science, 601(21):4966 – 4970, 2007.
72
[16] G. Smestad, A. Ennaoui, S. Fiechter, H. Tributsch, W. Hofmann, M. Birkholtz,
and W. Kautek. Photoactive thin film semiconducting iron pyrite prepared by
sulfurization of iron oxides. Solar Energy Materials, 20:149 – 165, 1990.
[17] B. Ouertani, J. Ouerfelli, M. Saadoun, B. Bessaı̈s, H. Ezzaouia, and J Bernède.
Characterization of FeS2 -pyrite thin films synthesized by sulphuration of
amorphous iron oxide films pre-deposited by spray pyrolysis. Materials
Characterization, 54(4-5):431–437, 2005.
[18] J. Anderson, C. Munsee, C. Hung, T. Phung, G. Herman, D. Johnson,
J. Wager, and D. Keszler. Solution-processed HafSOx and ZircSOx inorganic
thin-film dielectrics and nanolaminates. Advanced Functional Materials,
17:2117 – 2124, 2007.
[19] S. Meyers, J. Anderson, D. Hong, C. Hung, J. Wager, and D. Keszler.
Solution-processed aluminum oxide phosphate thin-film dielectrics. Chemistry
of Materials, 19:4023 – 4029, 2007.
[20] C. Flynn. Hydrolysis of inorganic Iron(III) salts. Chemical Reviews, 84:31 – 41,
1984.
[21] A. Gadalla and H. Yu. Thermal decomposition of Fe(III) nitrate and its
aerosol. Journal of Materials Research, 5:1233 – 1236, 1990.
[22] B. Weckler and H. Lutz. Lattice vibration spectra. Part XCV. Infrared
spectroscopic studies on the iron oxide hydroxides goethite (α), akaganéite (β),
lepidocrocite (γ), and feroxyhite (δ). European Journal of Solid State and
Inorganic Chemistry, 35(8-9):531–544, 1998.
73
[23] F. Gronvold and E. Westrum Jr. Heat capacities of iron disulfides
Thermodynamics of marcasite from 5 to 700 K, pyrite from 300 to 780 K, and
the transformation of marcasite to pyrite. The Journal of Chemical
Thermodynamics, 8(11):1039–1048, 1976.
[24] S. Meyers, J. Anderson, C. Hung, J. Thompson, J. Wager, and D. Keszler.
Aqueous inorganic inks for low-temperature fabrication of ZnO TFTs. Journal
of the American Chemical Society, 130(51):17603–17609, 2008.
[25] G. Pimenta and W. Kautek. Pyrite film formation by h2 s reactive annealing of
iron. Thin Solid Films, 238:213 – 217, 1994.
[26] M. Birkholz, S. Fiechter, A. Hartmann, and H. Tributsch. Sulfur deficiency in
iron pyrite (FeS2−x ) and its consequences for band-structure models. Physical
Review B, 43:11926 – 11936, 1991.
[27] M. Bronold, C. Pettenkofer, and W. Jaegermann. Surface photovoltage
measurements on pyrite (100) cleavage planes: Evidence for electronic bulk
defects. Journal of Applied Physics, 76:5800 – 5808, 1994.
[28] K. Ellmer and C. Hopfner. On the stoichiometry of the semiconductor pyrite
(FeS2 . Philosophical Magazine A, 75:1129 – 1151, 1997.
[29] J. Ares, A. Pascual, I. Ferrer, and C. Sanchez. Lattice intrinsic defects and
electrical resistivity in pyrite thin films. Thin Solid Films, 451:233–236, 2004.
[30] K. Buker, N. Alonso-Vante, and H. Tributsch. Photovoltaic output limitation
of n-FeS2 (pyrite) schottky barriers: A temperature-dependent characterization.
Journal of Applied Physics, 72:5721 – 5728, 1992.
74
[31] K. Mishra and K. Osseo-Asare. Fermi level pinning at pyrite (FeS2 )/electrolyte
junctions. Journal of the Electrochemical Society, 139:749 – 752, 1992.
[32] J. Ares, I. Ferrer, and C. Sánchez. Majority carriers in pyrite thin films: an
analysis based on Seebeck and Hall coefficient measurements. Thin Solid Films,
431:511–513, 2003.
75
Normalized
Transmittance (A. U.)
100
80
60
Fe-O
40
20
O-H
0
4,000
O-H
3,000
2,000
1,000
Wavenumbers (cm-1)
Figure 4.1: FT-IR spectrum collected from as-cured and air-heated IHN thin films.
1,400
Intensity (A. U.)
1,200
1,000
800
600
400
200
0
10
20
30
2θ ( 0)
40
50
60
Figure 4.2: XRD collected from IHN gels heated in air at 150 ◦ C for 60 min and
then reacted with H2 S (g) for 120 minutes at various temperatures. Dashed arrows
are Fe7 S8 pyrrhotite.
76
Intensity (A. U.)
1,000
800
600
400
200
0
10
20
30
2θ ( 0)
40
50
60
Figure 4.3: XRD collected from IHN gels heated in air at 150 ◦ C for 120 min and
then reacted with H2 S (g) for 120 minutes at various temperatures.
IHN
100 nm
Si
Figure 4.4: SEM cross-sectional image of an as-cured IHN thin film on Si.
77
b)
a)
Fe2O3
100 nm
Si
200 nm
Figure 4.5: SEM a) cross-sectional and b) top down images of a 400◦ C, 30 min
air-heated IHN thin film on Si.
Intensity (A. U.)
400
300
200
100
0
10
20
30
2θ ( 0)
40
50
Figure 4.6: XRD collected from as-cured IHN thin films reacted with H2 S (g) for
ten minutes at various temperatures. Dashed arrows are Fe7 S8 pyrrhotite.
78
Intensity (A. U.)
400
300
200
100
0
10
20
30
2θ ( 0)
40
50
Figure 4.7: XRD collected from 400◦ C, 30 min air-heated IHN thin films reacted
with H2 S (g) for ten minutes at various temperatures. Dashed arrows are Fe7 S8
pyrrhotite.
a)
pyrite
b)
100 nm
SiO2
Si
200 nm
Figure 4.8: SEM a) cross-sectional and b) top down images of a pyrite film synthesized from as-cured IHN.
79
b)
a)
pyrite
120 nm
Si
200 nm
Figure 4.9: SEM a) cross-sectional and b) top down images of a pyrite film synthesized from a 400◦C, 30 min air-heated IHN film.
4x105
α (cm-1)
3x105
2x105
2.5
Energy (eV)
105
0
0
0.5
1
1.5
2
2.5
Energy (eV)
Figure 4.10: Wavelength-dependent absorption coefficient and indirect optical band
gap estimate (inset) for a pyrite film. This 120 nm thick film was produced by
sulfurizing a 400◦ C, 30 min air-heated IHN film.
CHAPTER 5
Electronic structure calculations of Cu3 TaQ4 (Q = Se, Te)
and photoemission studies of Cu3 TaSe4
81
5.1
Introduction
The ability to control the band gap of a material by selectively varying the stoichiometry across a solid solution has broad applicability. Such control can be used
to engineer a material having an ideal match to the solar spectrum for efficient light
absorption. Photoemission, the process underlying light emitting diode technology,
can also be tuned by controlling stoichiometry and deliberately introducing dopant
species. Solar cells and light emitting diodes require both n- and p-type materials,
and in this contribution we describe the optical and luminescent properties of p-type
Cu3 TaSe4 . We also discuss the synthesis of the solid solution Cu3 TaSe4−x Tex (x = 0
- 4), where the optical band gap should be inversely proportional to x.
The cubic structure of Cu3 TaQ4 is shown in Figure 5.1.1, 2 Both Cu+ and Ta5+ are
coordinated by four Q2− in tetrahedral arrangements. The Ta-centered tetrahedra
are found at the eight corners of the cell, and the Cu-centered tetrahedra are located
in the middle of the twelve cell edges. Each TaQ4 tetrahedron shares edges with six
CuQ4 , and the CuQ4 tetrahedra share vertices with each other in all directions to
form a three-dimensional framework.
We have previously reported optical band gaps for Cu3 TaS4 and Cu3 TaSe4 of 2.70
and 2.35 eV, respectively,3 so it is clear that the structural changes resulting from
the addition of Ta to the Cu2 S- and Cu2 Se-like systems also significantly influences
the optical properties. In addition, we found that Cu3 TaS4 emits in the visible when
excited by UV light.3 To further delineate the properties of the Cu3 TaQ4 system, we
now report the photoluminescent behavior of Cu3 TaSe4 as a function of temperature.
We have also prepared Cu3 TaSe4 :W and measured photoluminescence as a function
of W concentration. Finally, we discuss the preparation of Cu3 TaSe4−x Tex (x = 0 4) based on the prediction that this solid solution should have smaller optical band
82
gaps and may therefore provide useful solar absorbers.
5.2
5.2.1
Experimental
Synthesis
Stoichiometric amounts of Cu powder (Cerac, 99.5%), Ta powder (Cerac, 99.9%),
and Q powder (Alfa Aesar, 99.99%) were ground and sealed in evacuated silica tubes.
Cu3 TaSe4 and W-doped Cu3 TaSe4 (W powder, Cerac, 99.5%) were synthesized by
heating at 750-800 ◦ C for 24 - 48 h. The Cu3 TaSe4−x Tex samples were prepared in
a N2 -filled glovebox, and then heated at 450 ◦ C for 1 week. Cu3 TaQ4 samples were
also prepared with 5% SeO2 or TeO2 to evaluate the impact of oxygen contamination. Following reaction, all tubes heated at 450 ◦ C were opened and ground in the
aforementioned glovebox.
5.2.2
Characterization
X-ray diffraction (XRD) was performed on the powders by using a Rigaku MiniFlex II or a Bruker-AXS D8 Discover diffractometer with Cu Kα radiation.
Temperature-dependent photoluminescence (PL) data were collected from Cu3 TaSe4
powders. These experiments were performed by using a 337 nm pulsed N2 laser, a
CCD detector, and a MMR optical cryostat. The samples were cooled to 80 K by
using a N2 closed-cycle refrigerator built into the cryostat, and then raised to 300 K
at 25 K intervals by heating. The samples were mounted on the copper cold finger
using double-sided carbon tape. This tape exhibited no luminescence under 337 nm
light excitation. In addition, room-temperature PL data were collected from doped
Cu3 Ta1−x Wx Se4 (x = 0.01, 0.03, 0.05) powders.
83
5.2.3
Band Structures
The electronic structures of Cu3 TaSe4 and Cu3 TaTe4 were computed with the code
Wien2k4 in the full-potential linearized augmented plane-wave (LAPW) formalism
with the PBE generalized gradient approximation. The unit cell parameters were
taken from published values.1, 2 The muffin-tin radii were set to 2.10 au for Se and
2.37 au for Cu and Ta for the calculation of Cu3 TaSe4 , and to 2.22 au for Te and
2.50 au for Cu and Ta for the calculation of Cu3 TaTe4 . The potentials and charge
densities were expanded on a k-mesh of 10,000 points in the Brillouin zone, including
286 unique k-points. Inside the spheres, the expansion of these quantities was limited
to l = 10. In the interstitial region, potentials and charge densities were expanded
using 2,708 and 3,074 plane waves for Cu3 TaSe4 and Cu3 TaTe4 respectively. The
calculations were iterated until the charge converged to better than 10−3 e.
For calculating the optical properties, the upper band-index was chosen to be 110
to ensure that the upper band energy is well above the Fermi level. The Lorentz
broadening of the inter-band spectrum was set to 0.03 eV for both compounds.
5.3
5.3.1
Results
Experimental results
Cu3 TaSe4 powders have previously been prepared at 750 ◦ C,5 and Cu3 TaTe4 crystals were grown at 450 ◦ C.2 As a result, the solid solution Cu3 TaSe4−x Tex was prepared
at the lower temperature. Cu3 TaSe4 powders for PL measurements were synthesized
between 750 and 800 ◦ C to increase crystallite size. XRD data from the intrinsic and
doped samples were consistent with the formation of Cu3 TaQ4 . With the exception of
a few very low intensity impurity peaks, all lines in the patterns could be indexed on
84
the basis of the previously reported cubic unit cells.1, 2 The impurity peaks increased
in intensity when Cu3 TaSe4 and Cu3 TaTe4 were synthesized with 5% SeO2 or TeO2 ,
so powders for the selenide-telluride solid solution were handled under inert atmosphere as much as possible to minimize oxide contamination. While XRD patterns
from both Cu3 TaSe4 and Cu3 TaTe4 showed sharp, well-defined peaks, the diffraction
peaks from the intermediate compositions were often split (Figure 5.2). The unit
cell parameters calculated from average peak positions are very similar to those reported previously,6 but when the split peaks of the intermediate compounds were
refined individually, statistically different values were obtained (Figure 5.3). These
unit cell parameters may represent distinct Se- or Te-rich domains within the same
powder. What is certainly clear is that these synthesis conditions have not produced
satisfactory single-phase powders. As a result, no further characterization of the
Cu3 TaSe4−x Tex solid solution was performed.
While PL emission from Cu3 TaTe4 was outside the detectable wavelength range
of the equipment, Cu3 TaSe4 powders show a strong PL peak at 2.01 eV (Figure
5.4). Similar PL in Cu3 TaS4 was attributed to acceptor-like Cu vacancies because of
the direct correlation between PL intensity and x in Cu3−x TaS4 .5 This assumption
produced an ionization energy for the Cu vacancies of 0.44 eV.7 Assuming a similar
mechanism in Cu3 TaSe4 and using the established band gap of 2.35 eV,3 we estimate
the ionization energy of the Cu vacancies as 0.34 eV. Because this simple mechanism
does not account for possible nonradiative or IR emission from the conduction band
to some other state in the band gap, this value may be much higher than the true
ionization energy of the Cu vacancies in Cu3 TaSe4 .
Doping Cu3 TaSe4 with W causes the PL peak at 2.01 eV to decrease in intensity
and shift to 1.85 eV (Figure 5.4). This shift to lower energy demonstrates that
85
W-doping generates donor states in the gap, which is also true of Cu3 TaS4 .5 The
solubility limit of W in the Cu3 TaSe4 matrix must occur somewhere between 1 and
3% W, since the PL peak does not shift any lower than 1.85 eV despite increasing
W concentration. Low intensity diffraction peaks from a WSe2 phase are visible in
the XRD patterns of Cu3 Ta0.97 W0.03 Se4 and Cu3 Ta0.95 W0.05 Se4 , providing additional
evidence for limited solubility.
The configurational coordinate (CC) model was developed to explain the effects
of lattice vibrations on the optical properties of a localized luminescent center.8, 9
This center can be a deliberately introduced dopant species, or an intrinsic defect
in the material. As temperature increases, the additional thermal energy activates
nonradiative decay pathways for photoexcited carriers, and this process decreases the
intensity of the measured luminescence. Luminescent intensity, I, and temperature,
T, are related as follows9
I∝
1
1 + B · exp
h
−EQ
kT
i
(5.1)
where B is the ratio of the radiative lifetime and the pre-exponential factor of the
nonradiative decay function, and k is Boltzmann’s constant. The term of interest is
EQ , which represents the activation energy of the nonradiative decay. From the I vs
T data plotted in Figure 5.5, we estimate that EQ for Cu3 TaSe4 is 140 meV. Equation
5.1 is also plotted in Figure 5.5 using this value, and a quick visual inspection shows
good agreement with the experimental data.
Phonons in a luminescent material are also influenced by temperature, and the
CC model addresses this in the following relationship:9
86
F W HM ∝ S
s
ℏω
.
coth
2kT
(5.2)
In this equation, the luminescent peak full width at half maximum (FWHM) is related
to T by an empirical fitting constant S and ℏω, the phonon energy of the luminescent
center. FWHM is plotted vs T in Figure 5.5, and these data are fit reasonably well
by the solid curve representing Equation 5.2 when ℏω = 44 meV.
5.3.2
Electronic Structures
The p-type Cu2 Q (Q = S, Se, Te) family10 crystallize in multiple structures that
range from stoichiometric Cu2 Q to Cu-deficient compositions. Monoclinic Cu2 S low
chalcocite is the stable phase at room temperature,11 while cubic Cu1.95 S digenite
and hexagonal Cu2 S high chalcocite stabilize above 89 and 105 ◦ C respectively.12, 13
Cu1.8 Se digenite is stable at room temperature, and above 198 ◦ C Cu2 Se berzelianite
forms.14 These cubic selenides are isostructural with Cu1.78 Se.15 In the telluride
system, hexagonal Cu2 Te16 and trigonal Cu1.81 Te weissite17 have been characterized.
Surprisingly, the majority of these binary Cu+ chalcogenides contain both Cu-Cu and
Cu-Q bonds. The only exception is low chalcocite in which each Cu is coordinated
by three S. The presence of both metal-metal and metal-chalcogenide bonds in most
of these compounds makes it difficult to predict or calculate electronic structures.18
The Cu2 Q compounds are all black, but since the Cu stoichiometry is highly
variable a range of optical band gaps have been reported. The band gap of Cu2 S is
most likely between 1.1 - 1.2 eV,19 while the most often cited studies show that Cu2 Se
begins to absorb at 1.2 eV and Cu2 Te at 1.0 eV.10 As discussed above, the addition
87
of Ta to form the Cu3 TaQ4 compounds produces significant structural changes that
increase the experimental band gaps. The electronic structures are also affected.
The Cu and Ta sites in the Cu3 TaQ4 structure are tetrahedrally coordinated by
Q. In tetrahedral geometries, the Cu and Ta d -orbitals will split into two lower energy
e and three higher energy t2 states. The ten d electrons in Cu+ completely fill the
e and t2 orbitals, while the corresponding Ta5+ orbitals are empty. Because the
CuQ4 tetrahedra form vertex-sharing chains with each other and edge-sharing chains
with the TaQ4 tetrahedra, carrier transport will occur through bands containing
contributions from all three species. The filled Cu 3d orbitals and overlapping Q
orbitals will define the valence band, while the empty Ta d orbitals and intermixed
Cu and Q orbitals will define the conduction band.
As expected for isostructural compounds, the band structures and density of states
(DOS) calculated for both Cu3 TaSe4 and Cu3 TaTe4 are very similar, and these are
shown for Cu3 TaSe4 in Figures 5.6 and 5.7. The major differences to note are that the
band gap of the telluride is smaller than the selenide, and the curvature of both the
valence band maximum (VBM) and conduction band minimum (CBM) are greater
for the telluride. The VBM and CBM exhibit similar dispersion in all directions,
which is consistent with the isotropic nature of the Cu3 TaQ4 crystal structure.
The lowest energy transitions across the computed band gaps of Cu3 TaSe4 and
Cu3 TaTe4 are indirect from R to X, and the magnitudes are 1.60 and 1.14 eV respectively. The experimental band gap of Cu3 TaSe4 is larger than the calculated value;3
band gap data have not yet been reported for Cu3 TaTe4 . The LAPW method is
known to underestimate the magnitude of band gaps,20 so it is not surprising that
the experimentally determined value for Cu3 TaSe4 is higher than the calculation. It
is also likely that the experimental band gap of Cu3 TaTe4 will be larger than 1.14 eV.
88
As predicted, Figure 5.7 shows that Cu d and Q p states are the main contributors
to the Cu3 TaQ4 VBM. While Cu, Ta, and Q all overlap to form the CBM, Ta d orbitals
are the most significant contributor. Because the Ta d orbitals are nearly coincident
with the total Ta DOS, only the total Ta DOS is shown. These results are similar
to previously reported extended Huckle tight binding calculations of the electronic
structure of Cu3 TaTe4 , which found that the valence band is mainly Cu 3d and Te
5p states, and the conduction band is primarily Ta 5d.2
Figure 5.8 provides the Wien2k predictions of absorption coefficient vs energy for
Cu3 TaQ4 . Both compounds have maxima on the order of 105 cm−1 . If the energy
scale is shifted to place the onset of absorption at the experimentally determined band
gap of Cu3 TaSe4 , the absorption coefficient reaches 105 cm−1 at approximately 2.5 eV
(500 nm). Assuming a similar shift of 0.75 eV is required to fit the calculated band
gap of Cu3 TaTe4 to the as yet unknown experimental gap, the absorption coefficient
of the telluride will reach 105 cm−1 at approximately 1.9 eV (650 nm).
5.4
Summary and Perspective
We have measured the photoluminescence of Cu3 TaSe4 as a function of temperature and calculated the activation energy of the nonradiative decay and the phonon
energy of the intrinsic luminescent center. These energies are 140 meV and 44 meV,
respectively. We have also demonstrated a photoluminescent shift from the intrinsic
emission at 2.01 eV to 1.85 eV by doping Cu3 TaSe4 with W.
Electronic structure calculations reveal indirect band gaps in both Cu3 TaSe4 and
Cu3 TaTe4 . The predicted absorption coefficients for both Cu3 TaQ4 compounds are on
the order of 105 cm−1 , which would make them excellent solar absorbers. The optical
band gap of Cu3 TaSe4 is too large for a useful solar absorber, but Cu3 TaTe4 may be a
89
good candidate. While we were not able to perform optical band gap measurements
because of incomplete development of the synthesis methodology, on the basis of our
calculations we conclude that the solid solution Cu3 TaSe4−x Tex (x = 0 - 4) should
allow tunability of the optical band gap to energies below 2.35 eV.
90
References
[1] A. Van Arkel and C. Crevecoeur. Quelques sulfures et séléniures complexes.
Journal of the Less Common Metals, 5(2):177–180, 1963.
[2] J. Li, H. Guo, D. Proserpio, and A. Sironi. Exploring tellurides: Synthesis and
characterization of new binary, ternary, and quaternary compounds. Journal of
Solid State Chemistry, 117(2):247–255, 1995.
[3] P. Newhouse, P. Hersh, A. Zakutayev, A. Richard, H. Platt, D. Keszler, and
J. Tate. Thin film preparation and characterization of wide band gap Cu3 TaQ4
(Q= S or Se) p-type semiconductors. Thin Solid Films, 517(7):2473–2476, 2009.
[4] P. Blaha, K. Schwarz, G. Madsen, D. Kvasnicka, and J. Luitz. WIEN2k An
Augmented Plane Wave+Local Orbitals Program for Calculating Crystal
Properties. Karlheinz Schwarz Techn Universität, Wien,Austria, 2001.
[5] P. Hersh. Wide Band Gap Semiconductors and Insulators: Synthesis,
Processing, and Characterization. PhD thesis, Oregon State University, 2007.
[6] E. Riedel, K. Erekul, and S. Yueksel. X-ray and vibrational studies of sulvanite
mixed crystals Cu3 Nb(Sx Se1−x )4 , Cu3 Nb(Sex Te1−x )4 , Cu3 Ta(Sx Se1−x )4 , and
Cu3 Ta(Sex Te1−x )4 . Zeitschrift fur anorganische und allgemeine Chemie,
465:131 – 140, 1980.
[7] P. Newhouse. Growth and Characterization of Wide-Gap Semiconducting Oxide
and Chalcogenide Thin Films by Pulsed Laser Deposition. PhD thesis, Oregon
State University, 2008.
[8] S. Shionoya and W. Yen, editors. Phosphor Handbook. CRC Press, 1999.
91
[9] O. Amir, E. Lifshitz, C. Uzan-Saguy, and R. Kalish. Characterization of
indium doped SnS2 . European journal of solid state and inorganic chemistry,
31(7):631–648, 1994.
[10] G. Sorokin, Y. Papshev, and P. Oush. Photoconductivity of Cu2 S, Cu2 Se, and
Cu2 Te. Soviet Physics Solid State, 7(7):1810 – 1811, 1966.
[11] H. Evans. The crystal structures of low chalcocite and djurleite. Zeitschrift fuer
Kristallographie, 150:299–320, 1979.
[12] R. Cava, F. Reidinger, and B. Wuensch. Mobile ion distribution and
anharmonic thermal motion in fast ion conducting Cu2 S. Solid State Ionics,
5:501–504, 1981.
[13] M. Kazinets. On the structure of certain phases in the copper-sulfur system.
Kristallografiya, 14:704–706, 1969.
[14] K. Yamamoto and S. Kashida. X-ray study of the average structure of Cu2 Se
and of Cu1.8 S in the room temperature and high temperature phases. Journal
of Solid State Chemistry, 93:202–211, 1991.
[15] S. Danilkin, A. Skomorokhov, A. Hoser, H. Fuess, V. Rajevac, and
N. Bickulova. Crystal structure and lattice dynamics of superionic copper
selenide Cu2−d Se. Journal of Alloys and Compounds, 361:57–61, 2003.
[16] K. Anderko and K. Schubert. Untersuchungen im sytem kupfer-tellur.
Zeitschrift fuer Metallkunde, 45:371–378, 1954.
[17] R. Baranova, A. Avilov, and Z. Pinsker. Determination of the crystal structure
92
of the hexagonal phase betaIII in the copper-tellurium system by electron
diffraction. Kristallografiya, 18:1169–1176, 1973.
[18] S. Kashida, W. Shimosaka, M. Mori, and D. Yoshimura. Valence band
photoemission study of the copper chalcogenide compounds, Cu2 S, Cu2 Se and
Cu2 Te. Journal of Physics and Chemistry of Solids, 64(12):2357–2363, 2003.
[19] P. Lukashev, W. Lambrecht, T. Kotani, and M. van Schilfgaarde. Electronic
and crystal structure of Cu2−x S: Full-potential electronic structure calculations.
Physical Review B, 76(19):195202, 2007.
[20] F. Tran, P. Blaha, and K. Schwarz. Band gap calculations with becke-johson
exchange potential. Journal of Physics: Condensed Matter, 19:196208, 2007.
93
Q
Cu
Ta
Figure 5.1: Crystal structure of cubic Cu3 TaQ4 , Q = S, Se, Te.
Intensity (A. U.)
300
250
200
150
100
50
0
10
20
30
40
2θ ( 0)
50
60
Figure 5.2: XRD patterns for Cu3 TaSe4−x Tex , x = 1 - 3.
94
5.95
5.9
a (Å)
5.85
5.8
5.75
5.7
5.65
0 0.5 1 1.5 2 2.5 3 3.5 4
x
Figure 5.3: Unit cell parameter vs x for Cu3 TaSe4−x Tex . The circles represent the
average unit cell parameter for each composition, and the squares were calculated
from obviously split XRD peaks.
Intensity (A. U.)
105
104
103
1.6
1.8
2
2.2
2.4
Energy (eV)
Figure 5.4: PL spectra collected from intrinsic and W-doped Cu3 TaSe4 .
95
0.21
0.8
0.2
0.6
0.19
0.4
0.18
FWHM (eV)
Intensity (A. U.)
1
0.2
0.17
0
100
150
200
250
300
T (K)
Figure 5.5: PL intensity (circles) or FWHM (squares) vs temperature for Cu3 TaSe4 .
The solid lines are theoretical fits based on Equations 5.1 and 5.2.
Figure 5.6: Band structure of Cu3 TaSe4 .
96
60
0
10
8
Energy [eV]
Figure 5.7: Density of states for Cu3 TaSe4 . The Fermi energy is demarcated by
the vertical dashed line at 0 eV. Note the significant differences between the ordinate
scales of the partial DOS.
97
α (105 cm-1)
10
5
0
0
2
4
Energy (eV)
Figure 5.8: Calculated absorption coefficient vs energy for Cu3 TaSe4 and Cu3 TaTe4 .
CHAPTER 6
Synthesis of BaCuQF (Q = S, Se, Te)
Andriy Zakutayev, Heather A. S. Platt, Douglas A. Keszler, Janet Tate
Portions of this chapter have been accepted for publication in the
Journal of Applied Physics.
99
6.1
Introduction
In a photovoltaic cell, appropriate electrical contacts are required to efficiently
extract electron/hole pairs created by the absorption of light. Such contacts should
have sufficiently large band gaps to allow the majority of the incident light to reach
the absorbing layer. One contact must be n-type with appropriate conduction band
alignment, and the other should be a p-type layer with a good match to the absorber
valence band. CdS, ITO, and ZnO:Al are all examples of useful n-type materials, but
p-type materials are more scarce.
The chalcogenides Cu2 Q (Q = S, Se, Te) comprise one family of p-type materials.1
Even though their formulas are simple, all three Cu chalcogenides are compositionally
complicated. Both the sulfide and selenide range from stoichiometric Cu2 Q to Cupoor Cu1.8 Q, and the telluride also adopts multiple phases.2 These complexities have
led to reports of a range of band gaps for Cu2 S, but general consensus places the
value between 1.1 and 1.2 eV.3 Fewer studies have been reported for the selenide and
telluride, but the most often cited gaps are 1.2 eV for Cu2 Se and 1.0 eV for Cu2 Te.1
These band gaps are small enough that all three Cu chalcogenides absorb visible light,
which largely limits their use to tunnel junctions.
To retain p-type character and obtain larger optical band gaps, we have investigated the isostructural chalcogenide fluorides BaCuQF (Q = S, Se, Te). The tetragonal crystal structure is shown in Figure 6.1. Layers of Ba2 F2 alternate with Cu2 Q2
sheets along the c-axis, leading to quantum confinement of each layer. As a result, the
band gaps of the chalcogenide fluorides are dramatically larger than the binary Cu
chalcogenides, with values of 3.1 eV for BaCuSF,4 2.9 eV for BaCuSeF,4 and 2.3 eV
for BaCuTeF.5 Band structure calculations and photoemission measurements show
that the valence band maxima of all three BaCuQF compounds are made up of Cu 3d -
100
and Q np-orbitals,5, 6 while the conduction band minima contain contributions from
all four elements.5 Thus the inverse relationship between band gap and the atomic
size of Q is the result of increasing band dispersion with greater Cu-Q overlap.
While BaCuSF and BaCuSeF can be readily prepared, the synthesis of BaCuTeF
has proven more difficult. The ideal solid state reactant BaTe is very difficult to
synthesize without oxide impurities, which compromises the stoichiometry of the final
product. Flowing gas methods, on the other hand, open the possibility of using
different Ba sources, and BaCuSF has already been successfully prepared with BaCO3
as a reactant.6 Decomposition of BaCO3 likely produces small, highly reactive Ba
particles that form BaS in-situ, and analogous processes should produce BaSe and
BaTe. To test this idea, we synthesized all three BaCuQF compounds under flowing
gas, and evaluated the chemical composition of the powders. We also discuss Gibbs
free energies to explain the results.
6.2
6.2.1
Experimental
Synthesis
BaCuSF powder was prepared by grinding BaCO3 (Cerac, 99.9%), BaF2 (Cerac,
99.9%), and Cu2 S (Strem Chemicals, 99.5%) in stoichiometric amounts followed by
heating in a 130 sccm flow of H2 S (g) (Matheson) at 550 ◦ C for 2 h. BaCO3 and
BaF2 were also combined with Cu2 Se (Cerac, 99.5%) in stoichiometric proportions
and reacted under a mixture of 15% H2 Se (g) - 85% Ar (Air Liquide, 99.8%) at 550
◦
C for 2 h to form BaCuSeF. The flow rate was maintained at 170 sccm. The gas
was switched over to 100% Ar, and the powder was heated for another 2 h at the
same temperature. BaCuTeF was prepared by combining BaCO3 , BaF2 , and Cu2 Te
101
(Cerac, 99.5%) in stoichiometric proportions and grinding with 5 - 20 weight% excess
Te (Alfa Aesar, 99.999%). The mixture was heated at 550 ◦ C under a mixture of 5%
H2 - 95% Ar for 2 h.
BaCuQF powders were cold pressed at 2 tons in a 9.5 cm die to form pellets for
compositional analysis. BaCuSF pellets were heated in a hot isostatic press (HIP,
American Isostatic Presses) for 3 h under 20,000 psi of Ar at 750 ◦ C to increase their
densities. BaCuSeF analogs were treated in the HIP at 650 - 700 ◦ C for 3 h with
15,000 psi of Ar, and BaCuTeF pellets were processed for 2 h at 780 ◦ C with 15,000
psi of Ar. Final densities were approximately 70% of theoretical values.
6.2.2
Characterization
X-ray diffraction (XRD) was performed on the powders by using a Rigaku MiniFlex II diffractometer with Cu Kα radiation.
Electron probe microanalysis (EPMA) was performed on the pellets by using a
Cameca SX-100 electron microprobe. Standards were elemental Cu, Se, and Te;
natural pyrite for S; BaF2 for Ba and F; and MgO for O. Each BaCuQF pellet was
sampled over several hundred microns, and at least 14 points were used to calculate
the final elemental ratios.
The photoelectron spectroscopic experiments were performed on the pellets by
using the Darmstadt Integrated System for solar energy research (DAISY-SOL). This
system is equipped with distribution and transfer chambers (10−8 mbar), multiple
deposition chambers, and an Escalab 250 photoelectron spectroscopy system (10−10
mbar base pressure). X-ray photoemission spectroscopy (XPS) experiments were
performed using the Kα line of an Al x-ray source with a monochromator (1486.6
eV, 0.4 eV energy resolution). XPS binding energies were measured with respect to
102
the Fermi level of the system, which was calibrated with the 3d 5/2 line of a clean
Ag sample (386.27 eV). XPS core level spectra were recorded over an energy range
at least 10 eV larger than the full width at half maximum of each peak, and the
background intensities were subtracted using a Shirley function. Binding energies of
the core levels were found by taking the second derivative of the core level spectra.
Stoichiometry of the BaCuQF surfaces were determined from the integrated area
under the XPS core level peaks using sensitivity factors supplied by Escalab.
6.3
6.3.1
Results
Synthetic challenges and approach
In addition to allowing the use of various Ba salts as reagents, flowing gas syntheses
of the BaCuQF compounds provide several other useful features. Many chalcogenide
powders that are exposed to air will oxidize, changing the stoichiometry of the final
product. The presence of H2 (g) during the reaction should form H2 O (g) and strip
the oxygen away. In addition, the large excess Q available after decomposition of
H2 Q (g) used for the syntheses of BaCuSF and BaCuSeF will favor the products and
minimize vacancies. H2 Te decomposes above 0 ◦ C, so it was not possible to take
advantage of similar chemistry for the BaCuTeF synthesis.7
Because they are still relatively new, the thermodynamics of the BaCuQF compounds have not been investigated. Some insight may still be gained by comparing
the flowing gas processes to typical solid state reactions, however. Gibbs free energies (∆Grxn ) were calculated for BaCuSF and BaCuTeF as follows, but we could not
complete BaCuSeF calculations because thermodynamic data are unavailable. The
formation of BaCuSF under flowing gas may be simplified to the following reaction
103
BaCO3 (s) + BaF2 (s) + Cu2 S (s) + H2 S (g) −→
2 BaCuSF (s) + CO2 (g) + H2 O (g),
while the solid state reaction is
BaS (s) + BaF2 (s) + Cu2 S (s) −→ 2 BaCuSF (s).
The expression ∆Grxn (flowing gas) - ∆Grxn (solid state) provides a way to compare
the flowing gas and solid state reactions that does not require ∆Gf values for all
of the individual reactants and products. When these reactions are combined and
simplified, only the following chemical species remain:
BaCO3 (s) + H2 S (g) −→ BaS (s) + CO2 (g) + H2 O (g).
At 527 ◦ C, the temperature closest to that used for the flowing gas reactions for
which data are available,8 ∆Grxn (flowing gas) - ∆Grxn (solid state) = 14 kJ/mol (see
Table 6.2 for the relevant ∆Gf values). This small value indicates that neither the
flowing gas nor the solid state reaction are strongly thermodynamically favored. The
formation of BaCuTeF under flowing gas may be simplified to the following reaction
BaCO3 (s) + BaF2 (s) + Cu2 Te (s) + Te (s) + H2 (g) −→
2 BaCuTeF (s) + CO2 (g) + H2 O (g),
and the solid state reaction is analogous to that shown for BaCuSF shown above with
Cu2 Te replacing Cu2 S. The simplified expression comparing the BaCuTeF flowing gas
and solid state processes is
BaCO3 (s) + Te (s) + H2 (g) −→ BaTe (s) + CO2 (g) + H2 O (g).
104
At 527 ◦ C, ∆Grxn (flowing gas) - ∆Grxn (solid state) = 112 kJ/mol, which means that
near the reaction temperature, the solid state reaction is thermodynamically favored.
This is surprising because the CO2 (g) and H2 O (g) products of the flowing gas
process have large negative ∆Gf values, but the unavailability of H2 Te must be more
deleterious. If oxide-free single-phase BaTe is available, the solid state reaction is the
better choice to synthesize BaCuTeF.
The ∆Gf for the Q oxide compounds (Table 6.3) provide an approach to gauge
the potential for forming oxide-free BaCuQF powders via flowing gas synthesis. All
of the S and Se oxides form only in the gas phase at 527 ◦ C, which provides another
mechanism to remove oxide contamination from the powders. By contrast, the most
thermodynamically favorable Te oxide phase is TeO2 (s). This is an obvious issue
since the solid oxide will remain in the bulk powder throughout the reaction.
Thermodynamically the BaCuTeF flowing gas process is less favorable than the
solid state and may retain oxygen as TeO2 (s), so successful BaCuTeF flowing gas
synthesis will depend on kinetic factors. Exclusion of atmospheric O2 (g) from the
reaction zone will be critical, and gas flow rate may also play a role by supplying
reactant H2 (g) and removing product gases. These considerations will also influence
the analogous BaCuSF and BaCuSeF reactions, but since the possible S or Se oxide
by-products are gases it is much more likely that the quaternary products will be
oxide-free.
While oxide contamination is not a critical consideration for the flowing gas synthesis of BaCuSeF, BaSe3 was an unlikely secondary phase that formed with only
flowing H2 Se (g) (Figure 6.2). In the Ba-Se phase diagram, BaSe3 exists as a line
compound at 75 atomic% Se from 0 - 760 ◦ C,9 so at this stage in the synthesis the
BaCuSeF powder must be significantly Se-rich. Also, as shown in Figure 6.2, ad-
105
ditional heating under flowing Ar leads to elimination of the peaks associated with
BaSe3 , probably by carrying Se (g) away from the bulk. Simply shortening the H2 Se
(g) reaction time left lower intensity but still obvious BaSe3 peaks.
6.3.2
Material quality
XRD data collected from the powders were consistent with the formation of the
respective BaCuQF compounds, and no impurity peaks were observed. Refined unit
cell parameters are summarized in Table 6.1. All cell volumes are within 0.6% of
published values.4, 10
Analysis of EPMA data collected from BaCuSF and BaCuSeF pellets (Table 6.4)
reveals that the elemental ratios are all within one standard deviation of the expected
values, and the statistically insignificant O/Ba ratios demonstrate that the flowing
gas method was mostly efficacious in excluding oxygen. The BaCuTeF synthesis
was less successful. While the XRD patterns do not exhibit other crystalline phases,
Table 6.4 reveals that there is a significant excess of F in the pellets. There is also a
statistically significant amount of oxygen.
Because the pellets had to be shipped to Darmstadt, Germany for the XPS experiments, several weeks elapsed between collecting the EPMA and XPS data. This
time lag may help to explain the significantly higher O/Ba ratios calculated for all
three BaCuQF compounds from the XPS data shown in Table 6.5. The biggest factor, however, is that XPS is used to monitor electrons ejected from the first 50 nm
of a sample while EPMA records x-rays from the bulk.11 In addition to the significant oxidation, Table 6.5 shows that the surfaces of the pellets of all three BaCuQF
compounds are significantly Cu- and F-poor. Te/Ba is within one standard deviation
of the expected value, but both S/Ba and Se/Ba are significantly lower. These data
106
clearly demonstrate that all three BaCuQF compounds are sensitive to reaction with
atmospheric O2 and/or H2 O, which clarifies previous bulk stability experiments with
BaCuSF.4
6.4
Summary and Perspective
We have demonstrated that the chalcogenide fluorides BaCuQF (Q = S, Se, Te)
can all be synthesized under flowing gas by using BaCO3 as a Ba source. Thermodynamically, both this method and the traditional solid state route are favorable for
forming oxide-free BaCuSF and BaCuSeF, but the solid state route is more favorable
for BaCuTeF. A different and less thermodynamically stable Ba salt may provide a
better starting point for the BaCuTeF flowing gas synthesis.
107
References
[1] G. Sorokin, Y. Papshev, and P. Oush. Photoconductivity of Cu2 S, Cu2 Se, and
Cu2 Te. Soviet Physics Solid State, 7(7):1810 – 1811, 1966.
[2] S. Kashida, W. Shimosaka, M. Mori, and D. Yoshimura. Valence band
photoemission study of the copper chalcogenide compounds, Cu2 S, Cu2 Se and
Cu2 Te. Journal of Physics and Chemistry of Solids, 64(12):2357–2363, 2003.
[3] P. Lukashev, W. Lambrecht, T. Kotani, and M. van Schilfgaarde. Electronic
and crystal structure of Cu2−x S: Full-potential electronic structure calculations.
Physical Review B, 76(19):195202, 2007.
[4] C. Park, D. Keszler, H. Yanagi, and J. Tate. Gap modulation in
MCu[Q1−x Q’x ]F (M= Ba, Sr; Q, Q’= S, Se, Te) and related materials. Thin
Solid Films, 445(2):288–293, 2003.
[5] C. Park, R. Kykyneshi, A. Yokochi, J. Tate, and D. Keszler. Structure and
physical properties of BaCuTeF. Journal of Solid State Chemistry,
180(5):1672–1677, 2007.
[6] H. Yanagi, J. Tate, S. Park, C. Park, D. Keszler, M. Hirano, and H. Hosono.
Valence band structure of BaCuSF and BaCuSeF. Journal of Applied Physics,
100:083705, 2006.
[7] N. Greenwood and A. Earnshaw. Chemistry of the Elements.
Butterworth-Heinemann, 2nd edition, 1997.
[8] I. Barin, F. Sauert, E. Schultze-Rhonhof, and W. Shu Sheng. Thermochemical
Data of Pure Substances. VCH, 1993.
108
[9] H. Okamoto. The Ba-Se (Barium-Selenium) system. Journal of Phase
Equilibria, 12(4):467–469, 1991.
[10] W. Zhu, Y. Huang, F. Wu, C. Dong, H. Chen, and Z. Zhao. Synthesis and
crystal structure of barium copper fluochalcogenides:[BaCuFQ(Q= Se, Se)].
Materials Research Bulletin, 29(5):505–508, 1994.
[11] D. Schroder. Semiconductor Material and Device Characterization.
Wiley-Interscience, 3rd edition, 2006.
109
c
F
Cu
Q
Ba
Intensity (A. U.)
Figure 6.1: Crystal structure of tetragonal BaCuQF, Q = S, Se, Te.
450
400
350
300
250
200
150
100
50
0
10
20
30
2θ ( 0)
40
50
60
Figure 6.2: XRD data collected from BaCuSeF prepared with and without the 2 h
anneal under flowing Ar.
110
BaCuSF
BaCuSeF
BaCuTeF
Table 6.1: Unit cell parameters for BaCuQF powders.
a (Å)
c (Å)
Volume (Å3 )
4.117(4) 9.0019(6)
152.6(1)
4.249(5) 9.1376(7)
164.9(2)
4.435(2) 9.3811(3)
184.53(8)
Table 6.2: Gibbs free energies of formation used to calculate ∆Grxn (flowing gas) ∆Grxn (solid state) for BaCuSF and BaCuTeF at 527 ◦ C.8
△Gf (kJ/mol)
BaCO3 (s)
-1007.0
BaS (s)
-440.2
BaTe (s)
-295.5
CO2 (g)
-395.5
H2 O (g)
-203.6
-45.7
H2 S (g)
Table 6.3: Gibbs free energies of formation for Q oxides at 527 ◦ C.8
△Gf (kJ/mol)
SO (g)
-58.4
SO2 (g)
-298.4
SO3 (g)
-321.9
-123.9
S2 O (g)
SeO (g)
-2.0
SeO2 (g)
-114.7
8.4
TeO (g)
TeO2 (s)
-179.7
TeO2 (g)
-70.7
Te2 O2 (g)
-120.4
Table 6.4: EPMA data for BaCuQF pellets.
Q
F
O
Cu
Ideal stoichiometry
Ba
Ba
Ba
Ba
BaCuSF
0.99 ± 0.06 0.93 ± 0.07 1.08 ± 0.07 0.1 ± 0.1
BaCuSeF
1.1 ± 0.1
1.2 ± 0.2
0.9 ± 0.2 0.1 ± 0.1
BaCuTeF
0.9 ± 0.1
0.9 ± 0.1
1.5 ± 0.3 0.3 ± 0.2
111
Table 6.5: XPS data for BaCuQF pellets.
Ideal stoichiometry
BaCuSF
BaCuSeF
BaCuTeF
Cu
Ba
Q
Ba
F
Ba
0.56 ± 0.09 0.7 ± 0.1 0.60 ± 0.09
0.65 ± 0.09 0.50 ± 0.09 0.50 ± 0.09
0.6 ± 0.1
1.1 ± 0.1
0.5 ± 0.1
O
Ba
1.1 ± 0.1
1.2 ± 0.1
1.4 ± 0.2
CHAPTER 7
Conclusions
113
This dissertation has explored two approaches to reducing the cost associated with
PV energy generation. The promising compositions Fe2 XS4 (X = Ge, Si), Cu3 TaQ4
(Q = Se, Te), and BaCuQ’F (Q’ = S, Se, Te) were chosen based on their structural
similarities to FeS2 pyrite and Cu2 Q’, readily available materials whose components
are prevalent in the earth’s crust.
The sulfides Fe2 XS4 are shown to be direct band gap (Eg ∼ 1.5 eV) p-type semiconductors. Electronic structure calculations also predict large absorption coefficients
of 105 cm−1 . Since the band gaps in the target range of 0.7 to 1.6 eV and the predicted absorption coefficients are so large, these compounds are deemed promising
solar absorbers.
Electronic structure calculations reveal indirect band gaps in both Cu3 TaSe4 and
Cu3 TaTe4 , and the predicted absorption coefficients for both compounds are on the
order of 105 cm−1 . While the optical band gap of Cu3 TaSe4 at 2.35 eV is too large
for a useful solar absorber, Cu3 TaTe4 may be a good candidate. Synthetic difficulties
prevented experimental optical band gap measurements, but the calculations predict
that the solid solution Cu3 TaSe4−x Tex (x = 0 - 4) should allow tunability of the
optical band gap to energies below 2.35 eV.
The wide band gap p-type chalcogenide fluorides BaCuQ’F have all been synthesized under flowing gas by using BaCO3 as a Ba source. While this method formed
oxide-free BaCuSF and BaCuSeF, the BaCuTeF powders contained some oxide. If
single phase BaTe is available, a traditional solid state route will be more successful
for forming single phase oxide-free BaCuTeF.
In addition to using abundant materials, PV manufacturing costs can also be
decrease by choosing methods to rapidly deposit thin films over large areas. Solution
deposition is an excellent candidate, so the conversion of spin coated Fe oxalate and
114
hydroxide nitrate thin films to sulfides was discussed in this dissertation.
The Fe oxalate films reported here are smooth and amorphous, but they roughen
considerably when heated under flowing H2 S (g) to convert to FeS2 pyrite. The final
130 nm thick pyrite films are p-type with an optical indirect band gap of 0.9 eV with
S/Fe = 1.99 ± 0.02, but their coarse morphology makes them poor candidates for
incorporation into a working PV cell.
Iron hydroxide nitrate films are morphologically dense and contain significant
amounts of hydroxide and water. They can be converted to FeS2 pyrite both as-cured
(S/Fe = 2.11 ± 0.09) and after being further dehydrated by heating in air (S/Fe =
1.98 ± 0.01). The optical properties of the final 100 nm pyrite films are similar to
those synthesized from the Fe oxalate precursor. Morphologically, the pyrite films
synthesized from iron hydroxide nitrate are all dense, but the completely dehydrated
precursor films produce the most uniform pyrite grains. As a result, FeS2 pyrite films
produced from this precursor have the potential to be successfully integrated into a
PV device.
115
REFERENCES
www.solarbuzz.com maintains current photovoltaic panel and power cost
information.
http://www.eia.doe.gov/cneaf/electricity/epm/table5 6 b.html.
Altermatt, P., Kiesewetter, T., Ellmer, K., and Tributsch, H. (2002). Specifying
targets of future research in photovoltaic devices containing pyrite (FeS2 ) by
numerical modelling. Solar Energy Materials and Solar Cells, 71:181 – 195.
Amir, O., Lifshitz, E., Uzan-Saguy, C., and Kalish, R. (1994). Characterization of
indium doped SnS2 . European journal of solid state and inorganic chemistry,
31(7):631–648.
Anderko, K. and Schubert, K. (1954). Untersuchungen im sytem kupfer-tellur.
Zeitschrift fuer Metallkunde, 45:371–378.
Anderson, J., Munsee, C., Hung, C., Phung, T., Herman, G., Johnson, D., Wager,
J., and Keszler, D. (2007). Solution-processed HafSOx and ZircSOx inorganic
thin-film dielectrics and nanolaminates. Advanced Functional Materials, 17:2117 –
2124.
Antonucci, V., Arico, A., Giordano, N., Antonucci, P., Russo, U., Cocke, D., and
Crea, F. (1991). Photoactive screen-printed pyrite anodes for electrochemical
photovoltaic cells. Solar Cells, 31:119 – 141.
Ares, J., Ferrer, I., and Sánchez, C. (2003). Majority carriers in pyrite thin films:
an analysis based on Seebeck and Hall coefficient measurements. Thin Solid Films,
431:511–513.
Ares, J., Pascual, A., Ferrer, I., and Sanchez, C. (2004). Lattice intrinsic defects
and electrical resistivity in pyrite thin films. Thin Solid Films, 451:233–236.
Baranova, R., Avilov, A., and Pinsker, Z. (1973). Determination of the crystal
structure of the hexagonal phase betaIII in the copper-tellurium system by
electron diffraction. Kristallografiya, 18:1169–1176.
Barin, I., Sauert, F., Schultze-Rhonhof, E., and Sheng, W. S. (1993).
Thermochemical Data of Pure Substances. VCH.
Beach, J. and McCandless, B. (2007). Materials challenges for CdTe and CuInSe2
photovoltaics. Materials Research Society Bulletin, 32:225–229.
116
Berbenni, V., Marini, A., Bruni, G., and Riccardi, R. (2000). Solid state reaction
study in the systems Li2 CO3 -FeC2 O4 · 2 H2 O and Li2 CO3 -Fe2 (C2 O4 )3 · 6 H2 O.
Thermochimica Acta, 346:115 – 132.
Birkholz, M., Fiechter, S., Hartmann, A., and Tributsch, H. (1991). Sulfur
deficiency in iron pyrite (FeS2−x ) and its consequences for band-structure models.
Physical Review B, 43:11926 – 11936.
Blaha, P., Schwarz, K., Madsen, G., Kvasnicka, D., and Luitz, J. (2001). WIEN2k
An Augmented Plane Wave + Local Orbitals Program for Calculating Crystal
Properties. Karlheinz Schwarz Techn Universität, Wien, Austria.
Bleam, W. and Hoffmann, R. (1988). Isomorphous substitution in phyllosilicates as
an electronegativity perturbation: Its effect on bonding and charge distribution.
Inorganic Chemistry, 27:3180 – 3186.
Broadbent, D., Dollimore, D., and Dollimore, J. (1967). The thermal
decomposition of oxalates. Part IX. the thermal decomposition of the oxalate
complexes of iron. Journal of the Chemical Society A, 1967:451 – 454.
Bronold, M., Pettenkofer, C., and Jaegermann, W. (1994). Surface photovoltage
measurements on pyrite (100) cleavage planes: Evidence for electronic bulk defects.
Journal of Applied Physics, 76:5800 – 5808.
Buker, K., Alonso-Vante, N., and Tributsch, H. (1992). Photovoltaic output
limitation of n-FeS2 (pyrite) schottky barriers: A temperature-dependent
characterization. Journal of Applied Physics, 72:5721 – 5728.
Burgess, J. (1978). Metal Ions in Solution. Ellis Horwood: Chichester, England.
Cava, R., Reidinger, F., and Wuensch, B. (1981). Mobile ion distribution and
anharmonic thermal motion in fast ion conducting Cu2 S. Solid State Ionics,
5:501–504.
Danilkin, S., Skomorokhov, A., Hoser, A., Fuess, H., Rajevac, V., and Bickulova,
N. (2003). Crystal structure and lattice dynamics of superionic copper selenide
Cu2−d Se. Journal of Alloys and Compounds, 361:57–61.
Dong, Y., Zheng, Y., Duan, H., Sun, Y., and Chen, Y. (2005). Formation of pyrite
(FeS2 ) thin nano-films by thermal-sulfurating electrodeposition films at different
temperature. Materials Letters, 59(19-20):2398–2402.
Edwards, H. and Russell, N. (1998). Vibrational spectroscopic study of iron (II)
and iron (III) oxalates. Journal of Molecular Structure, 443(1-3):223–231.
117
Ellmer, K. and Hopfner, C. (1997). On the stoichiometry of the semiconductor
pyrite (FeS2 . Philosophical Magazine A, 75:1129 – 1151.
Ennaoui, A., Fiechter, S., Pettenkofer, C., Alonso-Vante, N., Buker, K., Bronold,
M., Hopfner, C., and Tributsch, H. (1993). Iron disulfide for solar energy
conversion. Solar Energy Materials and Solar Cells, 29:289 – 370.
Evans, H. (1979). The crystal structures of low chalcocite and djurleite. Zeitschrift
fuer Kristallographie, 150:299–320.
Eyert, V., Hock, K., Fiechter, S., and Tributsch, H. (1998). Electronic stucture of
(FeS2 ): The crucial role of electron-lattice interaction. Physical Review B, 57:6350
– 6359.
Flynn, C. (1984). Hydrolysis of inorganic Iron(III) salts. Chemical Reviews, 84:31
– 41.
Gadalla, A. and Yu, H. (1990). Thermal decomposition of Fe(III) nitrate and its
aerosol. Journal of Materials Research, 5:1233 – 1236.
Gomes, A., Ares, J., Ferrer, I., da Silva Pereira, M., and Sanchez, C. (2003).
Formation of n-type pyrite films from electrodeposited iron sulphides: effect of
annealing temperature. Materials Research Bulletin, 38(7):1123–1133.
Gray, J. (2003). Handbook of Photovoltaic Science and Engineering, chapter The
Physics of the Solar Cell, pages 61–112. John Wiley & Sons.
Greenwood, N. and Earnshaw, A. (1997). Chemistry of the Elements.
Butterworth-Heinemann, 2nd edition.
Gronvold, F. and Westrum Jr, E. (1976). Heat capacities of iron disulfides
Thermodynamics of marcasite from 5 to 700 K, pyrite from 300 to 780 K, and the
transformation of marcasite to pyrite. The Journal of Chemical Thermodynamics,
8(11):1039–1048.
Hersh, P. (2007). Wide Band Gap Semiconductors and Insulators: Synthesis,
Processing, and Characterization. PhD thesis, Oregon State University.
Junod, A., Wang, K., Triscone, G., and Lamarche, G. (1995). Specific heat,
magnetic properties and critical behaviour of Mn2 SiS4 and Fe2 GeS4 . Journal of
Magnetism and Magnetic Materials, 146:21 – 29.
Kashida, S., Shimosaka, W., Mori, M., and Yoshimura, D. (2003). Valence band
photoemission study of the copper chalcogenide compounds, Cu2 S, Cu2 Se and
Cu2 Te. Journal of Physics and Chemistry of Solids, 64(12):2357–2363.
118
Kato, K., Okamoto, Y., Morimoto, J., and Miyakawa, T. (1997). The
thermoelectric properties of FeS2 . Journal of Materials Science Letters,
16(11):914–916.
Kazinets, M. (1969). On the structure of certain phases in the copper-sulfur
system. Kristallografiya, 14:704–706.
Kim, C., Escuadro, A., and Bedzyk, M. (2007). Interaction of H2 S with
α-Fe2O3 (0001) surface. Surface Science, 601(21):4966 – 4970.
Koo, H. and Whangbo, M. (1999). Analysis of bonding and d-electron count in the
transition-metal carbides and transition-metal-silicide carbides with discrete linear
M-C-M units (M = Cr, Fe, Re) by electronic structure calculations. Inorganic
Chemistry, 38:2204 – 2210.
Li, J., Guo, H., Proserpio, D., and Sironi, A. (1995). Exploring tellurides:
Synthesis and characterization of new binary, ternary, and quaternary compounds.
Journal of Solid State Chemistry, 117(2):247–255.
Lichtenberger, D., Ellmer, K., Schieck, R., and Fiechter, S. (1993). Optical,
electrical and structural properties of polycrystalline iron-pyrite(FeS2 ) layers
deposited by reactive DC magnetron sputtering. Applied Surface Science,
70:583–587.
Lindberg, J. and Snyder, D. (1973). Determination of the optical absorption
coefficient of powdered materials whose particle size distribution and refractive
indices are unknown. Applied Optics, 12:573 – 578.
Livshits, V., Blum, A., Strauss, E., Ardel, G., Golodnitsky, D., and Peled, E.
(2001). Development of a bipolar li/composite polymer electrolyte/pyrite battery
for electric vehicles. Journal of Power Sources, 97-98:782 – 785.
Lukashev, P., Lambrecht, W., Kotani, T., and van Schilfgaarde, M. (2007).
Electronic and crystal structure of Cu2−x S: Full-potential electronic structure
calculations. Physical Review B, 76(19):195202.
Luque, A. and Marti, A. (2003). Handbook of Photovoltaic Science and
Engineering, chapter Theoretical Limits of Photovoltaic Conversion, pages
113–151. John Wiley & Sons.
McCandless, B. and Sites, J. (2003). Handbook of Photovoltaic Science and
Engineering, chapter Cadmium Telluride Solar Cells, pages 617–662. John Wiley &
Sons.
119
Meester, B., Reijnen, L., Goossens, A., and Schoonman, J. (2000). Synthesis of
pyrite (FeS2 ) thin films by low-pressure MOCVD. Chemical Vapor Deposition,
6(3):121 – 128.
Meyer, C., Gros, Y., Vincent, H., and Bertaut, E. (1976). Proprietes magnetiques
et etude par effet Mossbauer du thiosilicate Fe2 SiS4 . Journal of Physics and
Chemistry of Solids, 37:1153 – 1158.
Meyerhofer, D. (1978). Characteristics of resist films produced by spinning.
Journal of Applied Physics, 49(7):3993–3997.
Meyers, S., Anderson, J., Hong, D., Hung, C., Wager, J., and Keszler, D. (2007).
Solution-processed aluminum oxide phosphate thin-film dielectrics. Chemistry of
Materials, 19:4023 – 4029.
Meyers, S., Anderson, J., Hung, C., Thompson, J., Wager, J., and Keszler, D.
(2008). Aqueous inorganic inks for low-temperature fabrication of ZnO TFTs.
Journal of the American Chemical Society, 130(51):17603–17609.
Mishra, K. and Osseo-Asare, K. (1992). Fermi level pinning at pyrite
(FeS2 )/electrolyte junctions. Journal of the Electrochemical Society, 139:749 – 752.
Mott, N. and Davis, E. (1979). Electronic Processes in Non-Crystalline Materials.
Clarendon Press, 2nd edition.
Nakamura, S. and Yamamoto, A. (2001). Electrodeposition of pyrite (FeS2 ) thin
films for photovoltaic cells. Solar Energy Materials & Solar Cells, 65:79 – 85.
Newhouse, P. (2008). Growth and Characterization of Wide-Gap Semiconducting
Oxide and Chalcogenide Thin Films by Pulsed Laser Deposition. PhD thesis,
Oregon State University.
Newhouse, P., Hersh, P., Zakutayev, A., Richard, A., Platt, H., Keszler, D., and
Tate, J. (2009). Thin film preparation and characterization of wide band gap
Cu3 TaQ4 (Q= S or Se) p-type semiconductors. Thin Solid Films,
517(7):2473–2476.
Nyquist, R. and Kagel, R. (1971). Infrared Spectra of Inorganic Compounds.
Academic Press.
Oguchi, T., Sato, K., and Teranishi, T. (1980). Optical reflectivity spectrum of a
CuFeS2 single crystal. Journal of the Physical Society of Japan, 48:123 – 128.
Okamoto, H. (1991). The Ba-Se (Barium-Selenium) system. Journal of Phase
Equilibria, 12(4):467–469.
120
Ouertani, B., Ouerfelli, J., Saadoun, M., Bessaı̈s, B., Ezzaouia, H., and Bernède, J.
(2005). Characterization of FeS2 -pyrite thin films synthesized by sulphuration of
amorphous iron oxide films pre-deposited by spray pyrolysis. Materials
Characterization, 54(4-5):431–437.
Panias, D., Taxiarchou, M., Douni, I., Paspaliaris, I., and Kontopolous, A. (1996).
Thermodynamic analysis of the reactions of iron oxides: dissolution in oxalic acid.
Canadian Metallurgical Quarterly, pages 363 – 373.
Park, C., Keszler, D., Yanagi, H., and Tate, J. (2003). Gap modulation in
MCu[Q1−x Q’x ]F (M= Ba, Sr; Q, Q’= S, Se, Te) and related materials. Thin Solid
Films, 445(2):288–293.
Park, C., Kykyneshi, R., Yokochi, A., Tate, J., and Keszler, D. (2007). Structure
and physical properties of BaCuTeF. Journal of Solid State Chemistry,
180(5):1672–1677.
Pimenta, G. and Kautek, W. (1994). Pyrite film formation by h2 s reactive
annealing of iron. Thin Solid Films, 238:213 – 217.
Plostinaru, A. and Patron, L. (1998). Coordination compounds as raw materials
for mixed oxides. IX mixed complex compounds of lanthanides and Fe(III) with
oxalic acid. Revue Roumaine de Chemie, 43:601 – 606.
Reijnen, L., Meester, B., Goossens, A., and Schoonman, J. (2000). In situ mass
spectrometric study of pyrite (FeS2 ) thin film deposition with metallorganic
chemical vapor deposition. Journal of the Electrochemical Society, 147:1803.
Ren, J., Liang, W., and Whangbo, M. (1998). Crystal and Electronic Structure
Analysis using CAESAR. PrimeColor Software, Inc., North Carolina, USA.
Riedel, E., Erekul, K., and Yueksel, S. (1980). X-ray and vibrational studies of
sulvanite mixed crystals Cu3 Nb(Sx Se1−x )4 , Cu3 Nb(Sex Te1−x )4 , Cu3 Ta(Sx Se1−x )4 ,
and Cu3 Ta(Sex Te1−x )4 . Zeitschrift fur anorganische und allgemeine Chemie,
465:131 – 140.
Schoonen, M. and Barnes, H. (1991). Mechanisms of pyrite and marcasite
formation from solution: 3. Hydrothermal processes. Geochimica et Cosmochimica
Acta, 55:12.
Schroder, D. (2006). Semiconductor Material and Device Characterization.
Wiley-Interscience, 3rd edition.
Shafarman, W. and Stolt, L. (2003). Handbook of Photovoltaic Science and
Engineering, chapter Cu(InGa)Se2 Solar Cells, pages 567–616. John Wiley & Sons.
121
Shionoya, S. and Yen, W., editors (1999). Phosphor Handbook. CRC Press.
Siyu, H., Xinyu, L., QingYu, L., and Jun, C. (2009). Pyrite film synthesized for
lithium-ion batteries. Journal of Alloys and Compounds, 472(1-2):9–12.
Slaoui, A. and Collins, R. (2007). Advanced inorganic materials for photovoltaics.
Materials Research Society Bulletin, 32:211–214.
Smestad, G., da Silva, A., Tributsch, H., Fiechter, S., Kunst, M., Meziani, N., and
Birkholtz, M. (1989). Formation of semiconducting iron pyrite by spray pyrolysis.
Solar Energy Materials, 18:299 – 313.
Smestad, G., Ennaoui, A., Fiechter, S., Tributsch, H., Hofmann, W., Birkholtz, M.,
and Kautek, W. (1990). Photoactive thin film semiconducting iron pyrite prepared
by sulfurization of iron oxides. Solar Energy Materials, 20:149 – 165.
Sorokin, G., Papshev, Y., and Oush, P. (1966). Photoconductivity of Cu2 S, Cu2 Se,
and Cu2 Te. Soviet Physics Solid State, 7(7):1810 – 1811.
Susac, D., Zhu, L., Teo, M., Sode, A., Wong, K., Wong, P., Parsons, R., Bizzotto,
D., Mitchell, K., and Campbell, S. (2007). Characterization of FeS2 -based thin
films as model catalysts for the oxygen reduction reaction. Journal of Physical
Chemistry C, 111:18715 – 18723.
Thomas, B., Cibik, T., Hopfner, C., Kiesner, K., Ehlers, G., Fiechter, S., and
Ellmer, K. (1998). Formation of secondary iron-sulphur phases during the growth
of polycrystalline iron pyrite (FeS2 ) thin films by MOCVD. Journal of Materials
Science: Materials in Electronics, 9:61 – 64.
Thomas, B., Ellmer, K., Muller, M., Hopfner, C., Fiechter, S., and Tributsch, H.
(1997). Structural and photoelectrical properties of FeS2 (pyrite) thin films grown
by MOCVD. Journal of Crystal Growth, 170(1):808–812.
Tran, F., Blaha, P., and Schwarz, K. (2007). Band gap calculations with
Becke-Johson exchange potential. Journal of Physics: Condensed Matter,
19:196208.
Van Arkel, A. and Crevecoeur, C. (1963). Quelques sulfures et séléniures
complexes. Journal of the Less Common Metals, 5(2):177–180.
Viaene, W. (1972). The Fe-Ge-S system: phase equilibria. Neues Jahrbuch fur
Mineralogie, Monatshefte, 1:23 – 35.
Vincent, H., Bertaut, E., Baur, W., and Shannon, R. (1976). Polyhedral
deformations in olivine-type compounds and the crystal structure of Fe2 SiS4 and
Fe2 GeS4 . Acta Crystallographica B, 32:1749 – 1755.
122
Wadia, C., Alivisatos, A., and Kammen, D. (2009). Materials availability expands
the opportunity for large-scale photovoltaics deployment. Environmental Science &
Technology, 43:2072 – 2077.
Weckler, B. and Lutz, H. (1998). Lattice vibration spectra. Part XCV. Infrared
spectroscopic studies on the iron oxide hydroxides goethite (α), akaganéite (β),
lepidocrocite (γ), and feroxyhite (δ). European Journal of Solid State and
Inorganic Chemistry, 35(8-9):531–544.
Willeke, G., Dasbach, R., Sailer, B., and Bucher, E. (1992). Thin pyrite (FeS2 )
films prepared by magnetron sputtering. Thin Solid Films, 213:271 – 276.
Yamamoto, A., Nakamura, M., Seki, A., Li, E., Hashimoto, A., and Nakamura, S.
(2003). Pyrite (FeS2 ) thin films prepared by spray method using FeSO4 and
(NH4 )2 sx . Solar Energy Materials & Solar Cells, 75:451 – 456.
Yamamoto, K. and Kashida, S. (1991). X-ray study of the average structure of
Cu2 Se and of Cu1.8 S in the room temperature and high temperature phases.
Journal of Solid State Chemistry, 93:202–211.
Yanagi, H., Tate, J., Park, S., Park, C., Keszler, D., Hirano, M., and Hosono, H.
(2006). Valence band structure of BaCuSF and BaCuSeF. Journal of Applied
Physics, 100:083705.
Zhao, G., Callaway, J., and Hayashibara, M. (1993). Electronic structures of iron
and cobalt pyrites. Physical Review B, 48:15781 – 15786.
Zhu, W., Huang, Y., Wu, F., Dong, C., Chen, H., and Zhao, Z. (1994). Synthesis
and crystal structure of barium copper fluochalcogenides:[BaCuFQ(Q= Se, Se)].
Materials Research Bulletin, 29(5):505–508.
123
APPENDICES
APPENDIX A
Effectively mentoring undergraduate students
Importance of Undergraduate Research
Chemistry weaves through the fabric of everything in the physical world. It is a
fascinating model that explains many complex systems, and provides framework to
continue the expansion of human knowledge. In order to awaken excitement for this
vast subject in students, it is vital to present it in a way that is relevant. A premed student must see her future patients being served by her understanding of basic
chemical principles. The chemistry major who is curious about the way the world
works should leave the laboratory with his perspective tuned to see chemical activity
all around him. It is also important to provide a safe environment where students can
learn from their mistakes. A principle arrived at through trial, error, and eventual
success will stick in the mind much longer than an easy solution. Finally, motivation
is the key to success in chemistry, as in all other disciplines. Any student who truly
wants to learn deserves the best that her mentor has to offer.
The most important teaching opportunity an experienced chemist has is in the
laboratory. This is an exciting arena because the students can see chemistry happening and, hopefully, begin to ask their own questions. An original research project also
provides opportunities for undergraduates to learn important skills. Once a chemical
question has been clearly stated, experiments need to be designed and carried out
125
to find the answer. These vital and central skills are difficult to master, so careful
guidance is required of the mentor. No project is complete without communicating
the results to others orally or in writing. Both media require concise well-labelled
figures, and technical writing should also be clear and carefully worded. The best
presentations and papers also weave the technical details together as a story.
The world is a wonderfully complex place, and there are many questions that
haven’t been answered yet. With good mentoring, undergraduate students can contribute to scientific understanding and get a taste of how exciting it is to do something
that no one else has done before. Once this has happened, the student is truly on his
way to becoming a chemist.
Maximizing Limited Time
Time is a huge constraint when working with undergraduate students. They may
have a ten week long summer of working forty hours per week, or they may sign up for
a research credit on top of an already full schedule of classes. Both scenarios provide
opportunites for learning and tackling original. albeit small, problems.
Filling a forty hour work week during a summer requires setting the student up
with two projects, and the mentor should also have a third idea. Sometimes a fast
student will finish in less time than projected, or an integral piece of equipment will
go down unexpectedly. The mentor should expect to spend about a week training the
student in the techniques she will use in the lab. It is also useful to give the student
relevant literature in the first week, so he can begin from a basic understanding of
the problem and why it is relevant. Appropriate formatting for figures should be
discussed as soon as the student has enough data to start putting them together,
and the final report should be started several weeks before the end of the summer to
126
leave time for revision. It is impossible to write a good paper in one draft, and that
should be emphasized as soon as students begin to communicate their work to others
in writing. Most importantly, the mentor needs to encourage the student to actively
participate in interpreting results and adjusting project plans.
Undergraduates working on an original project while simultaneously taking courses
are more difficult to mentor effectively. One project should be sufficient in this case,
but again the mentor needs to prepare a second project. Expectations need to be
clear up front, along with consequences if the student doesn’t follow through on the
commitment. Training on equipment and lab technique will require several weeks in
smaller amounts of time when the student can fit it around classes. The final report
should focus on experiments and results in the style of a lab report instead of a more
formal paper. An oral presenation of some sort is still beneficial for the student, but
it should be treated as optional because it might not fit with her other final projects
and examinations.
Acknowledgements
The ideas contained in this appendix were worked out over several years in conjunction with my own experiments. I want to thank Joshua Albus, Chris HolmesParker, Matthew Young, Felix Jeremias, and Etienne Brodu for learning with me in
the development of this teaching philosophy. These students also impressed me with
their energy, creativity, and questions. I wish all of them the best of luck in the
future.
Thanks are also due Douglas Keszler. Taking risks in mentoring the students
named above was much easier knowing that Doug’s seemingly endless ideas and experience were available to draw on when mine were exhausted.