Jun 3, 2009 - MI150, Dolan-Jenner Inc, USA), a mechanical chopper and a dark box ...... Becaria y amiga, pequeña colega y joven promesa de la Ciencia de.
SEMICONDUCTING THIN FILMS PREPARED BY ELECTROCHEMICAL MODULATION FOR TECHNOLOGICAL DEVICES
Felipe Caballero‐Briones Ph.D. Thesis
FACULTAT DE QUÍMICA DEPARTAMENT DE QUÍMICA FÍSICA
SEMICONDUCTING THIN FILMS PREPARED BY ELECTROCHEMICAL MODULATION FOR TECHNOLOGICAL DEVICES
Felipe Caballero Briones TESI DOCTORAL
UNIVERSITAT DE BARCELONA FACULTAT DE QUÍMICA DEPARTAMENT DE QUÍMICA FÍSICA Programa de Doctorat:
TECNOLOGIA DE MATERIALS Bienni 2005-2007
SEMICONDUCTING THIN FILMS PREPARED BY ELECTROCHEMICAL MODULATION FOR TECHNOLOGICAL DEVICES Memòria que presenta
FELIPE CABALLERO BRIONES per optar al Títol de Doctor per la Universitat de Barcelona Director:
FAUSTO SANZ CARRASCO Catedràtic del Departament de Química-Física de la Universitat de Barcelona
Barcelona, desembre de 2009
“Esforça’t en el teu quefer Com si de cada detall que pensis De cada paraula que diguis De cada peça que posis De cada cop de martell que donis En depengués la salvació de la humanitat Perquè en depèn creu-ho” Joan Maragall i Gorina
"Y si alguno se imagina que sabe algo, aun no sabe nada como debe saberlo" I Corintios 8:2
"Y si tuviese profecía, y entendiese todos los misterios y toda la ciencia, y si tuviese toda la fe, de tal manera que trasladase los montes, y no tengo amor, nada soy" I Corintios 13:2
“…a Su conocimiento no se escapa siquiera el peso de un átomo. No hay en los cielos ni en la tierra nada, que sea más pequeño o mayor que un átomo, que no esté consignado en el Libro evidente…” Sura XXXIV, Saba. Al-Qurán.
A mis hijos: María Isabel y Alejandro Ustedes son lo más importante de mi vida A Wendy la madre de mis hijos por estar con ellos
A mis queridos hermanos: Mariana, Julio y Alejandra Sin ustedes tampoco habría acabado
A mi padre: Arturo Mi mejor amigo Mi mayor ejemplo
A la memoria de mi madre: Alejandra Guadalupe No te dejo de extrañar Los ratos que no pienso en ti son porque me has distraído
A la memoria de mis Abuelos: Gilberto y Carlos Mario
A mis amigos y mentores que nos dejaron este año Álvaro Zapata Navarro y Andrés Martel Arbelo Por su legado
Dios: Sigo por aquí Te Deum
Esta tesis contó con el apoyo del Instituto Politécnico Nacional de México a través del Consejo Técnico para Prestaciones y Apoyo a Becarios (COTEPABE) entre Octubre de 2005 y Enero de 2010.
Las imágenes de las páginas 135, 206 y 221 de esta Tesis están reproducidas de la página web (© 2005 Metalprices.com LLC): http://www.metalprices.com/introduction/symbols_large_with_artists_comments.htm
CONTENTS PREFACE
1
Thin films for technological applications
1
Motivation
5
General objectives
6
Structure
6
CHAPTER 1. THEORETICAL BACKGROUND
9
1. Introduction
10
1.1. Nature and properties of conductors and electrolyte solutions
11
1.1.1.
Conductivity of electrolyte solutions
12
1.1.2.
Chemical transformations
13
1.1.3.
Energy levels associated with ions in solution
14
1.2. Semiconductors
16
1.2.1.
Band structure
16
1.2.2.
Energy levels in a semiconductor
21
1.2.3.
Defects in semiconductor crystals
23
1.2.4.
Energy levels at the surface
26
1.2.5.
Carrier transport at semiconductors
28
1.2.6.
Band gap determination
29
1.2.7.
Other optical transitions and the disorder parameter E0
32
1.3. Interfaces
34
1.3.1.
The metal-vacuum and semiconductor-vacuum interfaces
34
1.3.2.
Contact and electrode potentials
36
1.3.3.
Electrochemical cells and electrode processes
39
1.3.4.
The solid-electrolyte interface
44
1.3.5.
Potential and charge distribution through the EDL
46
1.3.5.1. Metal electrodes
46
1.3.5.2. Adsorption
47
1.3.5.3. Semiconductor electrodes
48
The band diagram of the semiconductor|electrolyte interface
56
1.3.6.
i
1.4. Charge transfer through the solid-electrolyte interface: models and kinetics
58
1.4.1.
Metal electrode
59
1.4.2.
Semiconductor electrode
64
1.5. Photoeffects in semiconductor|electrolyte interfaces
69
1.5.1.
Light absorption and carrier generation
69
1.5.2.
Photocurrent, quantum yield and recombination/relaxation processes
72
1.5.3.
Photopotential
75
1.6. Electrochemically prepared thin films
77
1.6.1.
Basic steps in phase formation onto electrodes
77
1.6.2.
Crystal phase formation (metal deposition)
79
1.6.2.1. Initial stage: incubation
79
1.6.2.2. Nucleation
83
1.6.2.3. Nucleus growth.
82
Reacting metal electrodes
84
1.6.3.1. Anodization of metals
84
1.6.3.2. Film formation
86
1.6.3.
1.6.4.
The electronic structure of semiconducting oxide thin films on metal electrodes
89
1.7. Solid-solid interfaces
91
1.7.1.
Schottky junctions
91
1.7.2.
Ohmic contacts
94
1.8. Solar energy conversion: mechanisms and materials 1.8.1.
The Sun and the band gap of semiconductor materials
1.9. Solar cells
96 96
98
1.9.1.
State of the art in photovoltaic research
98
1.9.2.
Cu2O based solar cells
101
1.9.3.
CdTe/CdS solar cells
103
1.9.4.
Transparent conducting oxides for photovoltaics: ZnO
104
ii
CHAPTER 2. EXPERIMENTAL SECTION
105
2.1. Experimental techniques
106
2.1.1. The electrochemical techniques
106
2.1.1.1. Voltammetry
106
2.1.1.2. Electrochemical Impedance Spectroscopy (EIS)
107
2.1.1.3. Photocurrent spectroscopy (PCS)
110
2.1.2. Reflectance and Transmittance
111
2.1.3. Structural and chemical techniques
112
2.1.3.1 X-ray diffraction (XRD)
112
2.1.3.2. Raman spectroscopy
113
2.1.3.3. X-ray photoelectron spectroscopy (XPS)
115
2.1.4. The Scanning Probe techniques
116
2.1.4.1. Atomic Force Microscopy (AFM)
117
2.1.4.2. Current Sensing Atomic Force Microscopy (CAFM)
118
2.1.4.4. Scanning Tunneling Microscopy and Spectroscopy
119
2.1.4.5. Electrochemical Scanning Tunneling
121
Microscopy (ECSTM) 2.1.4.6. Electrochemical Tunneling Spectroscopy (ECTS)
2.2. Instrumentation and experimental details
122
122
2.2.1. Electrochemical instrumentation
122
2.2.2. Photocurrent measurements
123
2.2.3. Electrochemical cells
124
2.2.4. Optical transmittance and reflectance
126
2.2.5. Structural and chemical measurements
126
2.2.5.1. XRD
126
2.2.5.2. Raman spectroscopy
127
2.2.5.3. XPS
127
2.2.5.4. SEM and EDS
128
2.2.6. SPM measurements
129
2.2.6.1. AFM: topography mode
129
2.2.6.2. CAFM
129
2.2.6.4. ECSTM and ECTS
131
2.3. Methods developed for film preparation
132
2.3.1. Cu2O
132
2.3.2. Cu2-xTe
133
2.3.3. ZnO
134
iii
CHAPTER 3. COPPER OXIDE FILMS
135
3. Introduction
136
Specific objectives
136
3.1. The Cu-O and Cu-H2O systems
137
3.1.1. General features
137
3.1.2. Cu oxidation in electrochemical conditions
138
3.2. Cu2O
139
3.2.1. Importance, technological applications and growth of Cu2O films
139
3.2.2. Crystalline structure
141
3.2.3. Electronic structure
143
3.2.4. Electrochemistry of Copper in Alkaline media
144
3.2.4.1. Cyclic voltammetry
144
3.2.4.2. Effect of the preferential orientation of the Cu
149
subtrate surface in the peaks observed at potentials below the onset of Cu oxidation 3.2.4.3. Is there Cu+ dissolution during Cu anodization?
150
3.3. A methodology for Cu2O film growth
152
3.4. General characterization of the prepared films
154
3.4.1. Morphology of the Cu2O layers
155
3.4.2. Film structure by X-ray diffraction
156
3.4.3. Raman spectroscopy: Cu2O fingerprint
158
3.4.4. Chemical composition by X-ray photoelectron spectroscopy
160
3.4.5. Optical characterization
164
3.4.6. Electrical properties by Conductive-AFM
167
3.5. Evidence and analysis of parallel growth mechanisms in Cu2O
170
films prepared by Cu anodization 3.5.1. Vary the dissolution time: effects on film properties
170
3.5.2. Morphology evolution
170
3.5.3. Structure and optical properties
172
3.5.4. Growth mechanism
176
iv
3.6. Electronic properties of the Cu2O films
181
3.6.1. Electrochemical Impedance Spectroscopy
182
3.6.2. Photocurrent spectroscopy
185
3.6.3. Study of the electronic properties of the Cu2O/Cu by ECSTM
188
3.6.4. ECTS characterization of the Cu|Cu2O|electrolyte interface
191
3.7. Study of the alkaline doping in Cu2O films
195
3.7.1. Chemical and structural characterization
196
3.7.2. Optical and electronic behavior
199
Chapter Conclusions
203
CHAPTER 4. APPLICATIONS OF ADVANCED ELECTROCHEMICAL METHODS TO THE PREPARATION OF SEMICONDUCTING FILMS 4.1. Copper telluride films by Electrocrystallization 4.1.1. Introduction
205 206 206
Specific objectives
207
4.1.2. The Cu-Te system
207
4.1.3. Copper electrochemistry in a Te (VI) electrolyte
209
4.1.4. Experimental design for preparation of Cu-Te films
211
4.1.5. Tellurium deposition
211
4.1.6. Cu2-xTe growth and film structure
215
4.1.7 Electronic and photoelectrochemical properties
218
Section Conclusions
220
4.2. Electrodeposited zinc oxide films
221 221
4.2.1. Introduction Specific objectives
222
4.2.2. The Zn-H2O system.
222
4.2.3. ZnO properties
223
4.2.4. Experimental design
225
4.2.5. Electrochemistry
226
4.2.6. Zinc oxide deposition
228
4.2.7. Pulsed electrodeposition of ZnO films with symmetric pulses
229
4.2.7.1. Current transients
229
4.2.7.2. Film morphology
231
4.2.7.3. Film structure
233
4.2.7.4. Optical characterization
234
v
4.2.8. Pulsed electrodeposition of ZnO films with symmetric pulses
236
4.2.8.1. Current transients
236
4.2.8.2. Film morphology
236
4.2.8.3. Film structure
238
4.2.9. Miscellaneous experiments at different duty cycles
240
4.2.10. Electrical characteristics
241
4.2.11. Photocurrent spectroscopy
241
4.2.12. Electrochemical Impedance Spectroscopy
242
Section Conclusions
243
CHAPTER 5. GENERAL COMMENTS AND CONCLUSIONS
245
APPENDIX A. OUTCOMING WORK
249
APPENDIX B. SYMBOLS AND ACRONYMS
251
APPENDIX C. SELECTED PUBLICATIONS
259
APÉNDICE D. RESUMEN EN CASTELLANO
263
Prefacio
264
Motivación
268
Objetivos Generales
269
Estructura de la Memoria
269
Óxido de Cobre
272
1. Electroquímica del cobre en medio alcalino
272
2. Método para el crecimiento de película de Cu2O
274
3. Caracterización general de las películas preparadas
275
4. Evidencia y análisis de mecanismos de crecimiento paralelos en películas de Cu2O preparadas por anodización de Cu
278
5. Estudio de las propiedades electrónicas de las películas de Cu2O
281
6. Estudio de la impurificación de películas de Cu2O con iones alcalinos
283
Aplicación de métodos electroquímicos avanzados a la preparación de películas semiconductoras
286
1. Cu2-xTe
286
2. ZnO
289
Conclusiones
292
AGRADECIMIENTOS
293 vi
P R E F A C E | 1
PREFACE Thin Films for Technological Devices Since the early discovery of fire and the use of stone and bones to manufacture the first tools, the human
civilization
has
been
increasing its level of sophistication supported by new materials: there was indeed a Stone Age, a Bronze Age and an Iron Age. Then it arrive the
development
of
plastics,
semiconductors, glasses, ceramics and biological inspired materials. Each kind had its own time of “glory”, and then became integrated into the mainstream of applications. The definition of “material” itself as “substance associated to a technological application” has evolved and became wider during this past century. The amazing range of different capabilities developed by the same substance when it is prepared with particle sized ranging from the bulk to the nanoscopic scale together with the advent of designed materials and recently with the possibility of manipulate atoms and molecules and to create hybrid and biological-mimetic devices, allow to said that a “material” is “a natural or synthetic substance with applications that depend on its level of structuring”. Bioengineered and bottom-up nanostructured materials are actually in the top of the wave.
2 | P R E F A C E
Semiconductors are now, at the first decade of the 21st century, “old” materials. The semiconducting property was discovered at the dawn of the 20th century, but it was not until the construction of the first transistor in the 50ties that began a revolution that gave rise to the Electronics Age, that on its side pushed up another revolution that lead to the Information Age. The first family of semiconductors to be described comprised selenium, germanium, gallium and silicon. They are elements of the IV and VI families and are denoted as “intrinsic” semiconductors. With the time, binary compound semiconductors such as the II-VI, III-V and IV-VI families were studied and more recently, selective screening, combinatorial synthesis and band gap engineering allow preparing “multinary” semiconductors with tailored properties which detonated its range of applications. However, the civilization has now reached a stage where the effects of the materials on the health and on the environment are a major matter of concern. On the other side, the intensive use of natural resources during the last century caused the increase of the prizes and the lack of availability of prime materials. These two circumstances are now giving place to a return to “older” materials, taking advantage on the improvement of the preparation techniques and the possibility to prepare materials at reduced structuring scale and thickness. Within the increasing demand for devices that use less amount of material and occupy less space, thin film-based technologies are actually the solution for these challenges. Copper and zinc have been metals of capital importance on the technological development of mankind. In the present moment, they have acquired a wide relevance because their abundance in the Earth crust, their comparatively low prices in the world stocks and their relatively friendly environmental behavior. Specifically in the field of optoelectronics and solar cells, many of their compounds are used both as direct constituents of devices or they are devised to substitute more environmental harmful compounds based for example, on cadmium or germanium. The chalcogenide group (the VI family of the Periodic Table, excluding oxygen) has a quite interesting and exploited family of binary compounds with varied applications as thin films ranging from glassy semiconductors for the non-linear optics (namely GaSe, GaTe) to wide band-gap materials for optoelectronics and solar cells (such as CdTe,
P R E F A C E | 3
ZnTe, HgS, PbS, Cu2-xS, Cu2-xTe, to mention only some) including of course a variety of functional oxides. Among this an extremely ample field of research, the present PhD thesis will focus on the electrochemical preparation and study of thin films of Cu2O, Cu2-xTe and ZnO, because their applications in the optoelectronic and solar cell industries. Cu2O is a ptype semiconductor with an interesting absorption coefficient and band gap used in Cu2O/ZnO solar cells and as a base for novel transparent conducting oxides such as CuSr-O and Cu-Al-O. Copper tellurides are attractive compounds used in thermopower devices and for ohmic back contacts in high efficient CdTe/CdS based solar cells. Zinc oxide is an n-type semiconductor, with applications ranging from sensors to transparent conductive oxides that are used as front electrodes in solar cells. Although widely studied, the synthesis and control of properties of these compounds is not always straightforward, particularly when dealing on thin films that currently are the more technological suitable form of devices. Problems associated with stoichiometry and chemical stability between many other, are commonly encountered. A compromise between properties and stoichiometry, and moreover, structure and morphology has to be achieved to obtain materials with optical and electronic properties with an interest to the industry. Although very suitable methods for stoichiometry and morphology control such as reactive sputtering, electron/ion beam deposition, chemical/physical vapor deposition, and free/molecular evaporation, between other are readily available at the research scale, the industry trends are currently focused on scalable, non-expensive, roll over methods. With more than a hundred years of experience, the electrochemical methods for film preparation had proved to fulfill these requirements, because the physics and chemistry underlying the electrochemical deposition are well understood. In the very foundations of the electrochemical techniques and particularly on the potentiostatic ones, it lies the principle of the control of the energy that it is applied to the system that –in principleallows the precise control of the chemical processes that would happen at the substrate where the layer is forming, an advantage that is above a lot of techniques that rely on indirect measures of the actual energy supply. At the present time, this advantage is being used to develop new electrochemical-based techniques to deposit increasingly
4 | P R E F A C E
complex materials. Electrocrystallization, potential or current modulated deposition, light induced deposition between other techniques are methods that allow both stoichiometry and structure/morphology control. On the other side, the increasing interdisciplinarity of the Materials Science that actually takes advantage on the Surface Science, Chemistry of Materials and Applied Physics tools, has allowed that traditionally techniques can be used to in situ monitor electrochemical processes that give deeper insights on elderly “understood” systems. Very interesting findings are currently being achieved on systems that have been studied by decades by the use of newly developed nano-scale tools, such as effects of self doping, impurities and structural inhomogeneities between other in the electronic and optical properties of functional materials. Also many of the underlying mechanisms of crystal growth, passivity against corrosion and electrical performance between other, had been recently unveiled by the use of these very local, atomic scale techniques. The present PhD Thesis arises from very uneven origins. At the present moment I do a recall and realize that fourteen years ago I had my introduction in the scientific career, when I begin as laboratory technician in the New Materials Group in the Applied Physics Department at CINVESTAV-IPN Mérida while concluding my B.Sc. studies. In March 1998 I obtained my B.Sc. in Chemistry with a thesis devoted to study the oxidation mechanisms of CdTe by reactive sputtering. At the time I continued to be technician at CINVESTAV-Mérida, learning about classical Material Science techniques. It was in that period when I also begin to work on tin oxides prepared by reactive sputtering and with CdS films prepared by Chemical Bath Deposition as the Group had the scope of preparing CdTe/CdS solar cells. By the middle of 1999 I was entrained in the M.Sc. studies and in 2000 I had the opportunity to be one of the founders of the CICATA-IPN Altamira. I obtained my M.Sc. in March 2003 with a Thesis that deal on structural and chemical studies on calcium incorporation to alkaline cooked corn, a quite important subject for Mexicans. During that time I get my first experiences on synchrotron techniques as well as I continued my work on II-VI semiconductors and tin oxides in collaboration with colleagues from Mexican and Cuban institutions, collaborations that are still alive. In a ten-year trajectory in Mexican institutions (CINVESTAV-IPN Mérida and CICATA-IPN Altamira) I get experience on the preparation/properties study of the materials associated to the CdTe/CdS and Cd-InSe based solar cells such as CdTe, CdSe, CdS, In2O3, and SnO2 and in parallel on
P R E F A C E | 5
promising transparent conducting oxides for the CdTe/CdS solar cell such as Cd-Te-O and Cd-Te-In-O. Those studies comprised preparation of thin films by close-spaced vapor transport, reactive sputtering and chemical bath deposition and characterization techniques such as optical absorption, in-situ mass spectroscopy, X-ray diffraction, Xray photoelectron spectroscopy, X-ray absorption and on-line and ex-situ Raman spectroscopy between other.
MOTIVATION The interest to get experience on electrochemistry-based preparation techniques, scanning probe techniques and more important, the scientific need to work on surface chemistry related projects that would lead to an increased understanding of the processes that govern the electronic properties of solar-energy materials, were the main motivations to come to Dr. Fausto Sanz’s Biolectrochemistry and Nanotechnologies Group to do my PhD at the University of Barcelona. I have the incredible luck to arrive at the right moment to pick up the inheritance and direct advising of the exciting developments made by Dr. Ismael Díez-Pérez and Dr. Pau Gorostiza on new in situ, electrochemical/nano-probe-based techniques to characterize the local electronic properties of semiconductor|electrolyte interfaces, specifically electrochemical tunneling spectroscopy that has proven to provide a direct measure on the density of states in a semiconductor-electrolyte system, equivalent to high-costly ultra high vacuum measurements. It is on those pioneer and fundamental works on metal|semiconductor|electrolyte interfaces made in this Group on the Fe-O and on the Sn-O systems as well as the new experimental developments on electrochemical tunneling spectroscopy that many of this Ph.D. thesis’ experimental and theoretical background rely.
6 | P R E F A C E
GENERAL OBJECTIVES •
Development of ad-hoc electrochemical-based routines for semiconductor film preparation onto different substrates
•
Study of growth conditions that allow tailoring of the stoichiometry, structure, morphology and other physical properties of copper oxide, copper telluride and zinc oxide thin films for its application to technological devices.
•
Generate
new
knowledge
about
film
growth
mechanisms,
semiconductor|electrolyte interface electronic processes, film properties and impurity roles for the studied materials
STRUCTURE This Memory is structured as follows:
Chapter 1 is intended to be a general introduction beginning with the nature and properties of the electrolyte solutions including the description of the energy levels of ions in solution, basic to understand their interaction with solid electrodes. The solidstate semiconductor basic concepts such the band structure, the energy levels and how the carriers move into semiconductors, emphasizing the role of defects in the optical transitions and electronic properties are revised. Dedicated sections deal with the properties and behavior of different interfaces i.e. metal and semiconductor interfaces with vacuum, electrolytes and other solids. Specifically the energetic of those interfaces and the charge transfer through them are discussed. The electrochemical concepts are outlined using the Gerischer’s band model to make an according discussion with that of semiconductors, particularly when photoeffects at interfaces are analyzed. The mechanisms that govern the preparation of thin films by electrochemical techniques – electrodeposition and metal anodization- are depicted and the structure of such films is discussed. Finally because the importance of the materials prepared in this work for optoelectronic devices, particularly solar cells, a brief review of solar energy conversion and different solar cells is outlined.
P R E F A C E | 7
Chapter 2 reviews the experimental techniques used in this work to characterize electrochemical systems and thin films. A variety of electrochemical and solid state methods is outlined. The principle of operation and latter the specific details of each method of characterization are discussed. The last section of this Chapter contains the experimental details for the preparation of thin films of Cu2O, Cu2-xTe and ZnO. Chapter 3 is dedicated to Cu2O thin films. The first part presents the general properties of the Cu-O and Cu-H2O systems, emphasizing the electrochemical oxidation of copper and the properties that give to semiconducting Cu2O its technological interest. Experimental results concerning the general electrochemistry of Cu in alkaline conditions, the identification of the main processes that occur during a potential excursion and the question of the possible dissolution stage of Cu are outlined. The second part comprehends the methodology developed for the preparation of submicron Cu2O films by Cu anodization and the characterization of its general properties: morphology, structure, chemical composition, and optical and electrical behavior. Thereafter a detailed investigation on the role of applying a potential where dissolution occurs on the properties of the oxide that concludes with evidence of parallel growth mechanisms of Cu2O films is presented. The next section comprises the study of the electronic properties at the Cu|Cu2O|electrolyte interface. A group of techniques such as electrochemical tunneling
impedance
microscopy
and
spectroscopy,
photocurrent
electrochemical
tunneling
spectroscopy, spectroscopy
scanning provide
complementary data to construct an electronic diagram of the interface. The final section of this chapter is a study of Cu2O doping with alkaline metals that provides new data of the electronic structure of Cu2O and outlines a strategy to modify the oxide properties such as the band gap and carrier density. At the end of the Chapter, the corresponding conclusions are given.
Chapter 4 comprises two sections: the first proposes a potentiodynamic method to prepare semiconducting copper telluride films by reaction with a Cu electrode in a Te (IV)-containing electrolyte in acidic conditions; the second, an electrodeposition method to prepare ZnO films onto ITO substrates. The properties of each system, their importance in technological devices, particularly in solar cells are revised. The preparation method for each material is described as well as the different achievements in the modulation of their properties. For Cu2-xTe films the main goal is to prepare the
8 | P R E F A C E
telluride films with tailored crystal structures. The electronic and photoelectronic properties are also studied. For ZnO films, different conditions of pulsed electrodeposition are considered. The main goal is to tailor the morphology and structures of the material with the scope of prepare nanostructured electrodes. Additionally some other physical properties such as their photocurrent activity, electrical performance and optical behavior are studied. Specific conclusions are given for each material at the end of the respective sections.
Chapter 5 presents the general comments and the general conclusions of this Ph. D Thesis.
References are presented in each Chapter using the end page format recommended in the American Chemical Society Journals
Appendix A outlines the works intended to be concluded in the near future.
Appendix B presents a summary of the symbols and acronyms used in this Memory based on the IUPAC directives.
Appendix C presents the publications related to this Thesis as well as the main presentations in international conferences. Publications of the author related with oxides and semiconductor materials are also mentioned.
Appendix D is the Thesis résumé in Spanish in accomplishment to the Universitat de Barcelona guidelines.
T h e o r e t i c a l B a c k g r o u n d | 9
CHAPTER 1 THEORETICAL BACKGROUND
Ilustration: Woodcut from Hortus Sanitatis, Strasbourg c. 1497
10 | C h a p t e r 1
1. Introduction Electrochemistry has as main scope the study of the interconversion between electric and chemical energy, and in a broader definition, the study of any electrical phenomenon that implies matter. Materials Science is interested in the relationship between structure and properties of the materials. Materials are said to be “matter with a specific use” and are usually grouped by the intended use or application. Technology of Materials implies the understanding and setup of analysis techniques, physical studies and materials development. Thin films are materials forming layers ranging from fractions of a nanometre to several micrometres in thickness. Thin films can be obtained either by deposition – apply a thin film on a surface-, or by growth –as are the oxide layers formed on top of a metal. Materials prepared as thin films are now the core of most of the technological applications in daily life and the production of lighter, denser and cheaper devices comprise the design of the preparation techniques as well as the fundamental studies on film’s properties. The present PhD memory is devoted to the preparation of thin films of semiconducting materials with modulated properties, materials that can be applied for photovoltaic devices –between others, prepared and studied by electrochemical techniques. The development of the necessary techniques and procedures to prepare and analyze the films are the main task of this Thesis. The film’s properties are studied by in-situ and ex-situ techniques and are correlated with the preparation conditions and on their final structure. Conversely, this correlation allows the comprehension of the mechanisms that govern film growth/deposition as well as the a priori modulation of the properties of interest. The application of these films to design a prototype or lab-scale devices is becoming a new research line in our laboratory. This chapter presents the concepts needed to manage the obtained results. The Thesis has an interdisciplinary scope and is intended to be read by both electrochemists and material scientists, thus, a brief review to some of the basic electrochemical and solid state concepts will be offered in the next paragraphs. Comprehensive reviews are easily available to deep into the subjects of interest to the reader. Particularly useful and exhaustive are the textbooks consulted to prepare this chapter 1,2,3,4,5 . 1
V.S. Bagotsky, Fundamentals of Electrochemistry, 2nd Ed. (2006) Wiley Interscience, NJ, USA S. Roy Morrison, Electrochemistry at semiconductor and oxidized metal electrodes, 2nd printing (1984) Plenum Press, New York USA, London UK 3 Rüdiger Memming, Semiconductor Electrochemistry, Wiley-VCH (2001) Weinheim, Germany 2
T h e o r e t i c a l B a c k g r o u n d | 11
1.1. Nature and properties of conductors and electrolyte solutions1-5 It is convenient to classify the materials following it’s the way they conduct the electricity. A useful classification is that of conductors, semiconductors and insulators (dielectrics). The essential attribute of a conductor is to have free electric charges and in the presence of an electric field these charges move through generating an electrical current. The free charge concentration and nature is a characteristic for each conductor. For electronic conduction the charge carriers are electrons or holes and for ionic conduction, the carriers are ions (charged atoms). Electronic conductors are all the metals, some oxides, carbon materials (graphene, graphite, and carbon nanotubes), some other inorganic and some organic compounds. Ionic conductors are solid substances (called solid electrolytes) and dissolutions (in polar solvents) containing ions, and then called electrolyte solutions. Substances such as salts, bases and acids dissociate into ions when dissolved. The dissociation could be total or partial being the fraction of original molecules that have been dissociated the dissociation degree αd * . Substances with low αd are called weak electrolytes and those with values of αd close to 1 are called strong electrolytes. For substances with αd < 1 there are present several types of species in the solution i.e. the ions in which the substance dissociates (for example, H3PO4 in water dissociates into H+, H2PO4-, HPO4-2 and PO4-3) and non dissociated molecules. The concentrations of these substances are interrelated except the concentration of the original compound. Binary electrolyte solutions contain just one solute in addition to the solvent. Multicomponent solutions contain several original solutes and the corresponding number of ions. Sometimes in multicomponent solutions the behavior of just one of them is of interest, the rest of them could be only added to raise conductivity or to equilibrate pH for example. In the bulk of an electrolyte solution, the electroneutrality condition is always accomplished but this condition is disturbed at the interfaces formed by the conductor with other conductors (such as an electrode) or an insulator (such as a vessel), where excess charge of a particular sign exists in the form of monolayers or thin space-charge layers (SCLs).
4
X.G. Zhang, Electrochemistry of silicon and its oxide, DOI 10.1007/b100331 J.M. Costa, Fundamentos de Electródica, Cinética Electroquímica y sus Aplicaciones (1981) 1a Ed. Editorial Alhambra SA, Madrid, Spain. * To avoid confusion with the optical absorption coefficient α, the degree of dissociation will be abbreviated αd instead of α as recommended by IUPAC. See Appendix for a complete list of symbols. 5
12 | C h a p t e r 1
1.1.1. Conductivity of electrolyte solutions The conductivity in any conductor is intimately related to other physical properties of a conductor, such as thermal conductivity (in the case of metals) and viscosity (in the case of an electrolyte solution). The strength of the electric current I is measured in amperes and it depends on the conductor, on the electrostatic field strength E in the conductor and on the conductor’s cross section perpendicular to the direction of the current flow, A. To made the current independent of the conductor’s dimensions, the current density j is employed instead which is the fraction of current associated with the unit of area of the conductor’s cross section j =I/A (A/cm2). In an electrolyte solution the charge carriers are the ions and both the positive (cations) and negative (anion) participate in the conduction process. In contrast with the semiconductors where there are also two types of charge carriers (holes and electrons) but only one type usually dominates because of doping, both types of carriers are always present in the electrolyte solution in similar concentration. The conductivity for a binary electrolyte solution is then given by the same equation that rules the electronic conductivity in solids:
(1.1)
In which σ is the conductivity expressed in (Ω.cm)-1, z1 and z2 are the charges of the ions, µ+ and µ- the mobility of the positively and negatively charged ions respectively (in cm2V-1s-1); c is the concentration of the ions (in mol.cm-3) and F is the Faraday constant (96500 A.s.mol-1). In a multicomponent electrolyte solution the conductivity is calculated with the sum over of all species. The mobility of an ion is related to the velocity ν and the electric field E in which the ion is moving. Into an electric field the ions migrate in the direction of the field, movement that is superimposed over the thermal, non directed movement of the ions in the solution. It is noteworthy that in the presence of an electric field the positive and negative ions will move in opposite directions. The corresponding electric force, |zi|eE, accelerates the ion until the friction drag –aproximated from Stokes’ law as 6πηνr being η the viscosity of the solution and r the radius of the hydrated ion- equilibrates the electric force. From Stokes’ law and from the definition of the mobility, it can be written:
T h e o r e t i c a l B a c k g r o u n d | 13 | | E
(1.2)
In the steady state, the ion local concentration is time invariant under the electric field application, i.e. no ions accumulate or vanish. However, the local concentration varies in those places where ions are consumed (sinks) or produced (sources) by chemical reactions. 1.1.2. Chemical transformations Ions in solution have other interactions than electrostatic attraction-repulsion. Those are chemical interactions of ions with the solvent and with each other. These relations lead to changes in the chemical structure surrounding the ion, and their understanding is crucial to describe both the ions in terms of their electronic energy levels and the exchange of electrons between ions in solution and the corresponding electrode. The overall interaction between an ion and other ions and the solvent molecules is called “solvation”. Solvation requires that the lattice ions are released from the crystal lattice. The energy for this process comes from that liberated when ions of the lattice associate with solvent molecules. This energy is called free energy of solvation and it is of the same order of magnitude that the lattice energy of the solute. The interactions between ions and solvent and other ions can be separated into “inner coordination sphere” referred to those taking place in with nearest neighbors and “outer coordination sphere” made up usually with solvent molecules. Inner sphere complex formation is the strong interaction where the ion and its neighbors form a group that can be considered almost a compound (such as a hydrate or a complex). Outer sphere complex formations are possible but are much weaker interactions almost always between ions of opposite charge, where a shell of solvent molecules separates the ion from the counterion. The solvent structure around the ion depends heavily on the charge of the ion. Accordingly, it changes during or after an electron transfer. Such a “rearrangement” or “reorganization” of the solvent molecules and the corresponding energy change play an important role in the electron transfer. Figure 1.1 depicts a scheme of a water solvated Na+ ion showing only the first hydration shell. Note that the solvent molecules are oriented in a form that depends on the ion charge and on the water dipoles.
14 | C h a p t e r 1
Figure 1.1. First hydration shell around a solvated Na+ ion. It is convenient to classify the chemical transformations that precede or follow electron transfer between an ion and an electrode into two types. First is the above mentioned solvent reorganization in the outer shell due to solvent polarization. Second is the stronger inner sphere chemical transformation, which does not always occur, but which can take many forms. Some classes of these transformations are a change in the ligand shell of an ion (thus changing the complex), dimerization (as when H+ reduces to form H2), precipitation or adsorption of the atom (as Cu2+ ions reduce to form Cu metal on the electrode) and so on. 1.1.3. Energy levels associated with ions in solution The electronic energy levels on an ion (or a molecule) in solution reflect the tendency of that species to give up or to accept an electron when the species approaches to the electrode. The energy levels expected for the ions can be related quantitatively to their chemical behavior. In a metal ion with two possible oxidation states, M2+/M3+, for the ion in the reduced (oxidized) state, M2+ (M3+), to transfer an electron to pass to the oxidized (reduced) state, an amount of energy is required, measured as the redox potential E0redox. Tables with E0redox for many redox couples are easily available. In this tables E0redox.describes the tendency of a species in solution to give or to extract electrons from a solid; the potential value is referred to the hydrogen couple H2/H+ whose E0redox=0 or Eref=-4.9 with respect to the absolute energy scale (where Evacuum=0). Nevertheless, as mentioned before, the ion in solution has many interactions with the solvent, with other ions and with the electrode. The energy associated to the ion in the reduced or oxidized state depends on all this interactions as well as on the thermal
T h e o r e t i c a l B a c k g r o u n d | 15
fluctuations. Thus, the energy levels of ions in solution are better described by “fluctuating energy levels” that are a probability distribution, illustrated in Figure 1.2.
Figure 1.2. Fluctuating energy levels of a dissolved ion. W(E) is the Gaussian-shaped probability that the state Ered or Eox has fluctuated to the energy E. Shaded area accounts for occupied levels. The reorganization energy λr relates Eox and Ered levels. The energy scale is referred to the vacuum energy scale where Evacuum=0. (Adapted from refs 2,4). In the Figure Eox is indicated as the most probable energy level for an oxidizing agent, the Gaussian represents the probability to find the level at energy distinct from Eox, due to thermal fluctuations of the solvent dipoles; the same applies for Ered. Both levels are related with the redox energy level by a quantity called the reorientation energy, λr & , as depicted in Figure 1.2, which is determined by the relaxation process involved in the regrouping reorientation of the solvation shell after electron transfer between the oxidized and reduced states. Detailed, quantitative analysis of energy level fluctuations can be found in the cited references for this section, particularly in refs.2,4. The importance of this fluctuating energy arises when the electrolyte is exposed to a solid phase, particularly semiconductor surfaces. The probability of electron transfer between the electrolyte and the solid phase will depend on the energy level at the instant of transfer, although the effect is noticeable only if the fluctuations are “slow” i.e. dependent on molecular or ion &
the reorganization energy is abbreviated λr in this thesis to avoid confusion with λ, wavelength of light
16 | C h a p t e r 1
motion rather than electronic motion. In Figure 1.2 is also observed that the Ered level is placed at an energy below Eox. This representation arises from the Franck-Condon splitting of energy levels: after the electron transfer occurs, the ion gets different charge that causes a subsequent reorientation of the solvent. For example, if an oxidized form of an ion M3+ reduces to the M2+ form according with: M3+ + e- → M2+
(1.3)
the electron will be captured at an energy level Eox as illustrated in Figure 1.2. Following the electron capture, the solvent dipoles will reorganize around the central ion because the extra negative charge, thereafter, the electronic energy level moves to an electronic energy Ered level lower than the level Eox, the unoccupied level now. The consequences of the fluctuating potential of the ions in solution and the Franck-Condon splitting of levels will be analyzed in the section corresponding to the solid-liquid interface. 1.2. Semiconductors2-4 In this section are explained the basic concepts of semiconductors from a materials science point of view, in the necessary extent to the further study of the phenomena that take place at the solid|electrolyte interfaces as well as to discuss the properties of the prepared thin films, but without exhausting the subject as it is not the scope of this work. The electronic structure of a semiconductor as well as its energetics and nature of defects is presented. 1.2.1. Band structure Before the energy bands of semiconductors can be described, the following basic quantities must be introduced. A free electron in the space can be described by classic or by quantum mechanical methods. Combining both methods, the wavelength λ of the electron wave is related to the momentum p by: (1.4) in which h is the Planck constant, m the electron mass and υ the electron velocity.
T h e o r e t i c a l B a c k g r o u n d | 17
The electron wave can also be described by the wave vector defined by the relation: (1.5) Combining equations (1.4) and (1.5) one obtains: (1.6) The kinetic energy of a free electron is then given by: (1.7) The parabolic relation between energy and the wave vector k is illustrated in Figure 1.3.
Figure 1.3. Parabolic dependence of the energy of a free electron E versus wave vector k In a metal the electrons are not completely free. A quantum mechanical treatment of the problem leads to the consequence that not all energies are allowed. The corresponding wave vectors are now given by (1.8) In which L is the length of a metal cube and n is any non-zero integer. Inserting equation (1.8) in (1.7) (1.9)
18 | C h a p t e r 1
The relation between energy and wave vector is still parabolic but the energy of an electron can only have certain values. Nevertheless, taken the dependence of the allowed values of k with the reciprocal values of L, it can be said that in a metal there is a continuum in the available states the electron occupies. By contrast, a semiconductor presents defined energy states (or energy bands) that electrons can reach. The band model of solids stem directly from the picture of atomic energy levels. As described by the Pauli Exclusion Principle, the energy levels of identical overlapping electronic orbitals cannot be equal. The discrete energy levels of isolated atoms overlap and broad forming bands of energy levels. The upper unfilled band associated with the first excited state is called the Conduction Band (CB) and the lower band almost filled with the valence electrons is called the Valence Band (VB). The energy levels in a semiconductor are characterized by the conduction band edge Ec, the valence band edge Ev and by the Fermi level EF. The Fermi level describes the equilibrium distribution of carriers in the bands in an analog way as the redox potential for the ions in solution as seen in the previous section. Between the highest occupated level in the VB to the lowest unoccupied level in the CB exists a gap of forbidden energies i.e. the so called bandgap or gap, Eg, which amplitude is characteristic of each material, depending on available electron energies. The electrons from the valence band may pass the Eg if enough energy in the form of heat or light is supplied. The conductivity of a semiconductor then increases with the temperature, opposite to metals that increase their resistance. In a perfect crystal there are no allowed energy levels for electrons in the gap, analogously to the atomic energy levels, where there are forbidden energies between the discrete levels. In real semiconductor crystals, the presence of crystallographic defects, impurities (either intrinsic or intentionally added), vacancies, and the dangling nature of the bonds in the surface, cause that states may appear into the band gap. These levels are classified either as donor states or acceptor states. Donors are usually located at energy levels slightly below EC give up excess of electrons to the conduction band, then creating electron conductivity (n-type semiconductors). Acceptors, usually located at energy levels slightly above EV capture valence band electrons from atoms of the material, producing hole conductivity (p-type semiconductors). It is important to know that the nature –donor or acceptor- of an impurity/defect level does not depend on its position relative to the band edges but to its capacity to get neutral when trapping a hole (acceptor) or an electron (donor). Figure 1.4 sketches the band structure for a
T h e o r e t i c a l B a c k g r o u n d | 19
semiconductor. It is important to bear in mind that bands are not flat but present curvatures that appear upon solution of the Schrödinger equation; for contrast, figure 1.5 shows the calculated band structure E(k) for Cu2O.
Figure 1.4. Scheme of a semiconductor energy band diagram. Band edges (EC and EV), the Fermi level (EF ), an acceptor and a donor levels within the band gap (Eg) are shown In Figure 1.5 the k values are indicated in Greek capitals indicating the directions of high symmetry in unit cell of the reciprocal lattice of a semiconductor (the so-called Brillouin zone). The band structure then depends on the three dimensional structure of the solid. The separation between the bands along the Γ line corresponds to the band gap. For Cu2O, as the maximum of VB in the Γ point coincides with the minimum of the CB along Γ, band gap is said to be direct.
20 | C h a p t e r 1
Eg
Figure 1.5. a) Conduction bands of Cu2O, b) Valence bands of Cu2O. Encircled is the minimum separation between VB and CB along the Γ line 6 .
6
R. Clasen, P. Grosse, A. Krost, F. Lévy, S.F. Marenkin, W. Richter, N. Ringelstein, R. Schmechel, G. Weiser, H. Werheit, M. Yao, W. Zdanowicz in Non-Tetrahedrally Bonded Elements and Binary Compounds I, Landolt-Börnstein-Group III Condensed Matter, Numerical Data and Functional Relationships in Science and Technology, Subvolume C Edited by O. Madelung (1998) ISBN 3-54064583-7, 1616-9549 (Online) 10.1007/b71138 Ed. Springer-Verlag.
T h e o r e t i c a l B a c k g r o u n d | 21
1.2.2. Energy levels in a semiconductor For a semiconductor at T > 0 K a dynamic process exists: electrons are constantly thermally excited from the VB to the CB, corresponding to electron-hole generation. At the same time, some electrons are losing energy and falling back to the VB corresponding to electron-hole recombination. At equilibrium, there is a constant density of electrons in the excited states, the rate of generation equaling the rate of recombination. At thermal equilibrium, the distribution of electrons among the allowed energy states in the semiconductor crystal is described by the Fermi-Dirac distribution function, denoted by fD(E), which has the form: 1/ 1
/
(1.10)
where EF is the Fermi energy, k the Boltzmann constant and T temperature. A fD(E) = ½, means that the probability of a state being occupied by an electron for the energy level at the Fermi level is ½. For not too highly doped n-type material, the Fermi level is well below the CB such that (EC-EF) >> kT and the Fermi function reduces to the simpler Maxwell-Boltzmann distribution function: /
(1.11)
The total density of electrons n, for not too heavily doped n-type material can then be found by the product of the density of allowed states in the conduction band g(E) and the probability that these states are filled and then integrating over the conduction band: /
(1.12)
Similarly, for moderately doped p-type material, the density of holes p, in the valence band is given by /
(1.13)
where NC and NV are the effective densities of energy states at the bottom of the CB and at the top of VB respectively. For large dopant concentrations Nd → NC or Na→ NV (around 1019 cm-3 for Si), these equations are no longer valid as the Fermi-Dirac distribution cannot be approximated by the Maxwell-Boltzmann distribution. At very high doping levels (Nd ≥ NC or Na ≥ NV) the semiconductor is said to be degenerated
22 | C h a p t e r 1
because the Fermi level is within the conduction or the valence band. As a result, allowed states exist very near the Fermi level, just as in a metal. Consequently, the properties of heavily doped semiconductors became similar to those of metals. From equations (1.12) and (1.13) the intrinsic carrier density ni, can be calculated: /
(1.14)
Equation 1.14 means that at thermal equilibrium the product of the electron and hole densities for a given semiconductor is a constant, and it can be readily inferred that when there is only a type of dopant, either donor or acceptor, for n-type material n ~ Nd and for a p-type material, p ~ Na. The Fermi energy describes the occupation of energy levels at equilibrium, but there are conditions where equilibrium is not attained. One particular case that will be treated later in this Thesis is when the semiconductor is illuminated by photons of energy greater than the Eg. Then electrons are excited from the VB of the semiconductor to the CB, leading to an excess of both electrons and holes. Non-equilibrium can also be caused by carrier injection from a p-n junction near the surface or carrier injection by certain oxidizing/reducing agents in solution at the surface. In any of these cases there can be far more minority carriers than equilibrium conditions allow (ruled by equation 1.14). The exact density of these minority carriers depends on the kinetics of the system, i.e. how fast the system is deexcited: the “lifetime” of minority carriers in the solid. To deal with this non equilibrium condition, the concept of quasi-Fermi energy has been introduced. For example, if the sample is illuminated, excess photoproduced holes and electrons have quasi-Fermi levels pEF and nEF that describe their respective densities in the CB and VB respectively. For majority carriers (electrons in n-semiconductors and holes in p-type), the quasi Fermi energy is almost the same as the equilibrium Fermi energy because their density is not appreciably increased upon illumination. However, minority carriers increase substantially, in cases by many orders of magnitude so, for example in a p-type semiconductor where EF lies closer to the VB, the nEF would be closer to CB. This is an important thing to take into account when dealing with reactions of holes/electrons with species in solution for example, thus it is interesting to have the quasi Fermi energy as a parameter that describes this density.
T h e o r e t i c a l B a c k g r o u n d | 23
1.2.3. Defects in semiconductor crystals In epigraph 1.2.1 was mentioned that the majority carriers in a real semiconductor arise from defects, traps and other structural-electronic features. Some of these defects are vacancies, interstitials, dislocations and grain boundaries. These defects are important not only in electrode reactions but also in the optical-electronic behavior of the semiconductor. For example, anodic oxidation occurs by migration of vacant lattice site or interstitial atoms, as well as self doping by diffusion. Deposition or corrosion usually occurs in dislocations. Grain boundaries and dislocations are chemical active at an electrode surface. Any of these defects can act as donor or acceptor sites modifying the optoelectronic behavior. It is important to mention that each material has its own characteristic type of natural defects that also depend on the methods of preparation, thermal history and so on. Induced defects are also important and the control of them is crucial to modulate materials properties. Figure 1.6 summarizes some of the different types of defects in a GaAs lattice, where In or B foreign dopants are added. A vacancy is of course, a missing atom in the crystal. Cation vacancies are acceptor defects and anion vacancies are donor defects. Consider a strong ionic oxide where the valence electrons are in orbitals close to the oxygen ions. Thus the VB is the band occupied by electrons on OL2- (lattice oxygen) orbitals. In order to remove a neutral metal atom from the solid to form a cation vacancy leaving the region of the imperfection neutral, electrons must be taken from the neighboring oxygen ions. However, the oxygen ions are still there with only slightly perturbed orbital and electrons can be accepted by these orbitals. Correspondingly, in order to remove a neutral oxygen atom from the solid and form a neutral anion vacancy, two electrons must be left behind with no valence orbital to accommodate them, thus they it is possible that the defect donate this electrons. An interstitial atom is one inserted in an available position between the atoms of a normal crystal. Usually are located into tetrahedral or octahedral sites and can be an atom of the same compound or a foreign species (arsenic or boron in the Fig.1.6 respectively). The interstitial atom can act as a donor or in rare cases as an acceptor.
24 | C h a p t e r 1
Figure 1.6. The different types of defects in a binary semiconductor (GaAs). VGa is a cation vacancy; VAs an anion vacancy; Asi an interstitial atom; GaAs and AsGa are antisite defects; Ins and Bs are foreign substitutional dopants, Bi is an interstitial dopant and the Frenkel pair is an interstitial-vacancy pair of defects closely situated. Dislocations and grain boundaries are bulk defects that are of great importance on the solid|liquid interfaces as well as on the electronic properties of the solids. An edge dislocation is a defect where an extra half-plane of atoms is introduced mid way through the crystal, distorting nearby planes of atoms. When enough force is applied from one side of the crystal structure, this extra plane passes through planes of atoms breaking and joining bonds with them until it reaches the grain boundary. A screw dislocation is much harder to visualize. Imagine cutting a crystal along a plane and slipping one half across the other by a lattice vector, the halves will fit back together without leaving a defect. If the cut only goes part way through the crystal, and then slipped, the boundary of the cut is a screw dislocation. Figure 1.7 schematizes both types of dislocations. Dislocations initiate for example by lattice mismatch between the substrate and a growing crystal –an anodic oxide for example- or by mechanical stress.
T h e o r e t i c a l B a c k g r o u n d | 25
Figure 1.7. Edge (top) and screw (bottom) dislocations generated from ideal lattices. The grain boundaries (GB) are regions where growing grains encounter. Considered as two dimensional defects, the GB is not flat but is arbitrarily bent. The grain boundary contains steps and other local "grain boundary defects". They also contain foreign atoms (impurities) or even precipitates from segregated phases. The grain boundary is not crystalline but consists of a thin amorphous layer between the grains. They can act as an accumulation of dislocations. Figure 1.8a) shows a graphical analogy of the grain boundaries in a crystal where segregation is observed, with a wall composed of brick and mortar. Fig. 1.8b) shows a scanning electron microscope (SEM) image of a real polycrystalline Ag-sample where the grain boundaries as well as screw dislocations can be readily observed.
Figure 1.8. a) Graphic analogy of grain boundaries; b) SEM image of an Ag layer deposited onto ceramic substrate © , at the image center a spiral growth caused by a screw dislocation is observed ©
Copyright by Fernando Estel under GNU Free Documentation License.
26 | C h a p t e r 1
Other interesting defects are traps and recombination centers. The term “trap” refers to energy levels deep into the band gap that come from defects or impurities. The name comes because traps capture electrons or holes. Traps can be occupied or not, naturally or intentionally. Trapped carriers can be reemitted either thermally or by photon excitation and they return to CB or VB (electrons or holes). Reemission can be very slow, in the time scale of days. Trapped carriers build up a space charge layer in their vicinity, thus repelling other carriers to approach: this is a very often cause of reduction of efficiency in electronic devices. Recombination centers are sites –either impurities or charged defects- that tend to capture carriers readily from either band. A majority carrier can be captured, then a minority one. In the site they are annihilated –they recombinebefore any of them is released. When a hole-electron pair is generated by photoexcitation, the generated minority carriers will annihilate at recombination centers at a decay rate proportional to their “lifetime” τ. For a n-type material:
(1.15)
And for a p-type material:
(1.16)
where the subscript e indicates “equilibrium”. If n < ne or p < pe the formulas imply generation of carriers. In electrochemical conditions, it could be possible that minority carriers at the surface are consumed by the electrolyte and be of a density less than the equilibrium value associated with bulk density; in this situation, the formulas account for the rate of minority carrier generation. 1.2.4. Energy levels at the surface Surface state energy levels or just “surface states” (SS) arise from different causes. Intrinsic defects in a clean surface can be present. Adsorbed species as well as solution non-adsorbed ions close enough to change electrons with the surface, provide surface states. In solid-solid interfaces all the energy levels are essentially SS. They will be discussed later in more detail, and they will be termed “interface states”, the SS term restricted only for solid-liquid interfaces.
T h e o r e t i c a l B a c k g r o u n d | 27
Energy levels at the surface can be either acceptor o donor, thus influencing the processes of electron exchange, recombination of minority carriers, the bonding of absorbed species, and corrosion between other. Attending their origin, SS are of different types, there are for example: •
Shockley states: they come from dangling orbitals in covalent solids, because the surface atoms do not have neighbors in one side. In the surface these dangling orbitals may overlap causing a bonding and an antibonding state. There is also the analogous case of dangling bonds in transition metal oxides with partially occupied d orbitals. The d orbitals of the surface cations can be available for covalent or ligand field bonding to surface ions.
•
Tamm states: in a more ionic compound semiconductor or insulator, a surface lattice anion (oxygen for example, OL2-) is surrounded by less than its normal complement of cations, thereafter its energy is higher than that of the bulk because the reduced attraction from the neighboring cations. Because they are occupied when the surface is neutral, these are donor surface states. Correspondingly, the unoccupied energy levels associated with the surface cations can be acceptor states. The resulting orbitals broad at the surface forming bands located close to the valence band and the conduction band respectively.
•
Lewis site. It is a site capable to accept/donate an electron pair if it is acid/basic respectively. Although are not considered normally as a SS (because they interchange paired instead of unpaired electrons), Lewis sites at the solid-liquid interface are of great importance because they could absorb hydroxyl ions or protons from the solution thus changing the local pH. They also affect the surface charge thereafter the double layer at the interface and could be passivating sites of other defects.
•
Adsorbed electroactive species could tend to donate or accept electrons and their states could form band mostly because surface heterogeneity, such as the presence of steps, kinks and terraces.
•
Ions in solution within tunneling distance from the surface could exchange electrons with the bands of the solid. These SS are the main interest of electrochemistry. The energy levels of these ions in solution were discussed in the section 1.1.3.
28 | C h a p t e r 1
1.2.5. Carrier transport at semiconductors Conduction in a semiconductor can occur both by hole/electron transport as well as by charged defects such as vacancies, interstitials or impurities. When an electric field of strength E is applied across a crystal, electrons and holes are forced to move in the material with a corresponding current density given by: (1.17) where σ is the conductivity, the reciprocal value of the resistivity ρ. For semiconductors with both electrons and holes as carriers, the conductivity is determined by:
(1.18)
Where e is the elementary charge and μn and μp are the mobilities of free electrons and free holes respectively. Note that equation 1.18 is the same as equation 1.1 that rules conductivity of ions in solution. In doped semiconductors, one of the terms into the parenthesis dominates and accordingly it is possible to increase the conductivity in several orders of magnitude by increasing doping. The mobility is a material constant and usually takes values a range between 1 and 1x103 cm2V-1s-1, high compared with the mobilities of ions and molecules in solution (that are typically around 10-4-10-3 cm2V-1s-1). Factors affecting the mobility are carrier scattering due to the presence of acoustic phonons and ionized impurities, and the temperature. The conductance G and the resistance R are given by:
⁄
(1.19)
where A is the cross-sectional area of the sample and L the length in the direction of current flow. Interstitial ions and charged vacancies can move through the crystal under the influence of an electric field (field-assisted diffusion) or a concentration gradient (diffusion). An interstitial can jump to successive equivalent sites. Vacancies move through the crystal by a mechanism involving an atom jumping into a neighboring vacancy site, then the next atom jumping into the vacancy site and so on. The result of this process is the motion of the vacancy through the crystal in the same way as the holes move through
T h e o r e t i c a l B a c k g r o u n d | 29
the crystal. Vacancy migration is of major importance in the formation of anodic oxides as will be studied in section 1.1.6. Another important parameter is the carrier diffusion coefficient, Dn or Dp, given by the Einstein relation:
;
(1.20)
In the case of defect transport such as vacancies or interstitials the diffusion coefficient also obeys equation 1.20, but the corresponding current density is given by: ⁄
⁄
(1.21)
Where Nd is the density of defects, μd their mobilities, Dd the defect diffusion coefficient and dϕ/dx is the electric field. The first term en equation 1.21 represents the field-assisted diffusion and the second the diffusion caused by the concentration gradient dNd/dx. 1.2.6. Band gap determination Although the explanation on the band structure of semiconductors presented in section 1.2.1. is basically correct, it is a simplification that does not take into account the tridimensional character of the solids. As mentioned above, the real band structure of the solids is a solution of the Schrödinger equation assuming a potential energy profile being periodic with the periodicity of the lattice. Thereafter, the band structure of a solid E(k) is a function of the three dimensional wave vector within the Brillouin zone. In the case shown in Figure 1.5, the maximum of VB and the minimum of the CB coincide along the vector k, then the transition –if allowed- is called direct; otherwise, when the CB minimum and VB maximum are found at k≠0, the transition is called indirect. The process for direct transitions consists in the absorption of a photon of the precise energy to make an electron overcome the energy gap. In the case of indirect transitions, a phonon is needed to supply the missing momentum, and then the transition consists in a successive series of collisions photon-electron-phonon. Figure 1.9 shows a scheme of both types of transitions.
30 | C h a p t e r 1
Figure 1.9. Optical transitions in semiconductors with direct and indirect band gap The band structure of a semiconductor can be probed in several ways, but the simplest method is to measure the absorption spectrum. The absorption coefficient α, is defined as: (1.22) where d it the thickness of the sample, and I and I0 the transmitted and the incident light intensities, respectively. The fundamental absorption refers to band-to-band excitation which can be recognized by a steep rise in absorption when the photon energy of the incident light goes through this range. Since, however, optical transitions must follow certain selection rules, the determination of the energy gap from absorption measurements in not a straightforward procedure. Since the momentum of photons, h/λ, is small compared with the crystal momentum h/a (a is the lattice constant), the momentum of electrons should be conserved during absorption of photons. The absorption coefficient α(hν) (h is the Planck constant and ν the photon frequency) for a given photon energy (hν = Eph the photon energy) is proportional to the probability, P, for transition from the initial to the final state and to the density of electrons in the initial state as well as to the density of empty final states. On this basis, a relation between absorption coefficient α and the photon energy Eph can be derived.
T h e o r e t i c a l B a c k g r o u n d | 31
For a direct band-to-band transition, for which momentum remains constant, it has been obtained for a parabolic energy structure (near the absorption edge): ~
/
(1.23)
in which Eg is the band gap. Accordingly, a plot of (α Eph)2 vs Eph should yield a straight line and Eg can be determined from the intercept, as shown in Figure 1.10. The Figure inset shows the absorption spectra in the region near the band gap.
Figure 1.10 Direct band-gap calculation from absorption measurement of a Cu2O film. Below Eg the absorption tail extends up to 1.5 eV. Inset: absorption spectrum near Eg. For indirect transitions, the relation between α and Eph is given by: ~
(1.24)
Correspondingly, the band gap is calculated from a plot of (α Eph)1/2 vs Eph.
32 | C h a p t e r 1
1.2.7. Other optical transitions and the disorder parameter E03,7,8,9,10 Other electronic transitions are possible upon light excitation. Figure 1.11 depicts some of them. Additionally to the VB to CB transition (1 in Fig 1.11) there can be transitions from VB to an unoccupied level in the CB of higher energy (1a). In the presence of donor or impurity levels, transitions from these to the conduction band are feasible (2 in Fig 1.11). However, as the impurity concentration is very small, the absorption coefficient will be smaller than that for band-to-band transitions. In some semiconductors, upon light excitation a neutral quasi-particle state, an exciton, can be formed by an electron and a hole as a result of their Coulomb attraction (3 in Fig 1.11). Excitons can move through the crystal and can be split into an independent electron and a hole by thermal excitation. Usually it is only possible to observe excitons at low temperatures or at nanoparticulated semiconductors, but there are some systems where excitons are stable at room temperature. Additionally, intraband transitions may occur in heavily doped semiconductors (4 in Fig. 1.11). Then the importance of the analysis of the optical absorption spectra to determine the electronic structure of the semiconductors can be readily inferred.
Figure 1.11. Optical transitions in a semiconductor. Details in the text. 7
S. John, C. Soukoulis, M.H. Cohen and E. N. Economou, Phys. Rev. Lett. 57 (1986) 1777 A. Iribarren, R. Castro-Rodríguez, V. Sosa and J.L. Peña, Phys. Rev. B 58 (1998-II) 1907 9 A. Iribarren, R. Castro-Rodríguez, V. Sosa and J.L. Peña, Phys. Rev. B 60 (1999) 4758 10 A. Iribarren, R. Castro-Rodríguez, F. Caballero-Briones and J.L. Peña, Appl. Phys.Lett. 74 (1999) 2957 8
T h e o r e t i c a l B a c k g r o u n d | 33
In Figure 1.10 inset, an energy-band tail appears in the low-energy region of absorption spectra. This tail has been observed universally in disordered systems, but it is also observed in single crystalline ones and it is known to be caused by potential fluctuations in the material. It is characterized by the tail parameter (E0), also known as Urbach energy, and from the absorption spectra E0 value can be found by the expression: (1.25) The analysis of E0 related to preparation conditions, structural measurements and defect distribution gives additional information about the electronic structure of the materials. As stated in sections 1.2.3. and 1.2.4. in semiconductors exist a number of defect types that create space charge regions where potentials develop modifying the carrier transport and distribution. In the vicinity of impurity atoms the carriers move in a screened Coulomb potential that depends on the ionization state of the impurity and the impurity spatial distribution. The E0 calculated for this contribution can be written as E0,N(N) for donor impurities and E0,P(P) for acceptor impurities. On the other hand, the ionic displacements due to partially ionic bound cause a static disorder field that the carriers feel as a random potential. Then a term due to carrierphonon interactions dependent on the temperature has to be introduced, E0,W(T). In every material, even in single crystals, lattice distortions due to imperfections are always present, such as dislocations, strain, random substitution of atoms and impurities either substitutional or interstitial. The resulting displacements of the atoms from their ideal sites produce additional potential fluctuations. The corresponding term is E0,X(X). The term E0,X(X) contains the contribution of bulk defects E0,def that arises as defects are associated with localized energy levels that act as traps, causing the depletion of the energy band in their surroundings. There is also the strain constant term E0,Y that describes the effects of the lattice strain and elastic properties of the material on E0. The contribution to E0 from the structural disorder then increases from a single crystalline material to a polycrystalline and to a glassy-amorphous one, and it is independent of the temperature. Finally, there is the grain boundaries contribution E0,GB. As mentioned in previous sections, GBs are bidimensional amorphous regions between the grains that contain defects, impurities, traps, etc. that cause large potential fluctuations. It is readily deduced that a reduction in the grain size lead to increased surface/volume ratio, thus
34 | C h a p t e r 1
increasing the effects of GBs on the disorder. The electric field induced by the charge distribution at the grain surface decays from the surface to the center, thereafter, its value increases with reduced grain size, increasing the contribution to E0. This analysis leads to a general expression for the Urbach energy: , , ,
, ,
,
,
, ,
,
(1.26)
The reader is kindly referred to the cited references to see the detailed formulation to calculate each term. The specific weight of each term to E0 can be determined experimentally with optical, structural or electronic techniques. 1.3. Interfaces1-5 In this section the characteristics of the solid-vaccum, solid-electrolyte and solid-solid interfaces are reviewed. Basic electrochemistry concepts are exposed. The potential distributions as well as the charge transfer through the studied interfaces are pointed out. The problem of oxide growth in the metal-electrolyte interface and the different models developed to deal with it are outlined. The electrochemical response of semiconductor electrodes is also examined. Finally, the solid-solid interfaces are commented to understand the ex-situ electrical characteristics of the prepared samples. Photoelectric and photoelectrochemical effects are also assessed. 1.3.1. The metal-vaccum and semiconductor-vacuum interfaces When clean solid –either metal or semiconductor- is exposed to a vacuum, there is not exchange of free charges thus no electrochemical equilibrium is established. To transfer charges from one phase to the other a net work has to be exerted. The total work employed to extract an electron from a metal into the vacuum is called the electron work function or simply work function ϕ, given in energy units. The work function is always positive, since otherwise the electrons would leave the metal spontaneously. The vacuum level Evac is taken as a reference level where the electrons have zero energy. Figure 1.12(a-c) shows the energy diagrams for the a) metal and b) n-type and c) p-type semiconductors vacuum interfaces. The Fermi level position, EF is determined by the work function. In the metal-vacuum case, the ionization energy Ei is equal in absolute value to the work function ϕ and to the
T h e o r e t i c a l B a c k g r o u n d | 35
electron affinity Eea # . There is another term χ, called the electrostatic term or surface term, that depends on the presence of impurities, adsorbates and so on, but that also depends on the crystallographic face exposed. The electrochemical potential of the electrons is given by μ. Then, the position of the Fermi level can be described with the electrostatic term and the chemical potential: (1.27) In the case of semiconductors the Fermi level position depends on doping as seen in previous sections. The work function for a metal is directly measurable, in the case of semiconductors is calculable from ionization measures, but the relative contributions of μ and χ are not accessible experimentally, and can only be estimated by theoretical approaches.
A
B
C
Figure 1.12. Schematic diagrams for different solid-vaccum interfaces A) Metal; B) ntype semiconductor; C) p-type semiconductor.
#
Is important to remember that the ionization energy Ei is that needed to extract one electron from the atom (from the valence band in a semiconductor) and the electronic affinity Eea is the energy gained when an electron is brought from the vacuum into the atom (in the conduction band in a semiconductor).
36 | C h a p t e r 1
When surface states –intrinsic or extrinsic- are present, the rearrangement of the surface atoms causes the buildup of a space charge layer (SCL) beneath the surface. In the case of n-type semiconductors, the SCL is positive, and then the distance between the Fermi level and the CB is increased at the surface, leading to a corresponding band bending. For p-type semiconductors the situation can be described similarly. Figure 1.13 represents the band bending caused by SS in both types of semiconductor.
Figure 1.13. Scheme of the band bending caused by a SS in n-type (left) and p-type (right) semiconductor-vaccum interfaces 1.3.2. Contact and electrode potentials When two conductors are brought in contact, the accumulated charges in the conductor surfaces will undergo redistribution and a certain potential difference will be setup across the junction (interface) which depends on the conductors. This potential difference is called the Galvani potential of this junction. Because another interface is formed when a measuring device such as a voltmeter or potentiometer is connected, Galvani potentials cannot be measured. Galvani potentials are produced by the difference in chemical forces exerted on the electrons within the surface layers by each of the two conductors. The unidirectional resultant force causes the transition of electrons from one conductor to the other. As result, if the two conductors are uncharged initially, one of them will charge up negatively and the other will charge positively. The excess charges of opposite sign accumulate near the interface and form an electrical double layer (EDL). The field that arises within this layer stops further transition of electrons. Finally, an equilibrium state is established in which the electrical force in the EDL completely balances the effect of chemical forces. Is important to note that this equilibrium is a dynamic state, even the
T h e o r e t i c a l B a c k g r o u n d | 37
net charge flow is zero partial charges cross the interface in both directions, being the value of one of these currents (charges flowing by time unit) called the exchange current (or current density when referred to the interface area). The magnitude of the current density indicates if the equilibrium at the interface is reached fast (thermodynamically reversible) or slow (kinetic controlled). An electronic conductor (metal or semiconductor) in contact with an ionic conductor (an electrolyte solution for example) is called an electrode. When an electrode is introduced into an electrolyte solution, an electrochemical equilibrium between the metal and the ion is established. A Galvani potential is established equal to the difference in the work functions of the electrode and the electrolyte, and as stated before, it cannot be measured. Nevertheless, one is interested to know the “electrode potential” in a determinate electrolyte solution. The International Union of Pure and Applied Chemistry (IUPAC) defines the “Absolute electrode potential” as “the electrode potential of a metal measured with respect to a universal reference system (not including any additional metal/solution interface)” 11 . Therefore, it is needed a suitable “universal reference system”. The IUPAC defines the standard hydrogen electrode (SHE) as the reference electrode for which, under standard conditions, E0(H+/H2) = 0 at every temperature 12 . Nevertheless, under practical conditions is not necessary to give absolute potentials as the hydrogen electrode is not easy to handle experimentally. Instead the electrode potentials are usually quoted to a suitable reference electrode, that following the IUPAC is “an electrode that maintains a virtually invariant potential under the conditions prevailing in an electrochemical measurement, and that serves to permit the observation, measurement, or control of the potential of the indicator (or test) or working electrode”11. Many reference electrodes are known. Some of them are the Ag/AgCl (SSC), the Hg/HgCl (SCE), Cu/CuSO4, etc., and their standard potentials referred to SHE are usually found in tables. Nonetheless, there is a case where the measure of the absolute electrode potential is of capital importance, and that is the comparison of the electrochemical and electronic scales of energy, where the vacuum level Evac=0. In Figure 1.14 is reproduced the IUPAC recommended comparison between the electrochemical and physical scales. The 11
IUPAC. Compendium of Chemical Terminology, 2nd ed. ("Gold Book"). Compiled by A. D. McNaught and A. Wilkinson. Blackwell Scientific Publications, Oxford (1997). XML on-line corrected version: http://goldbook.iupac.org (2006-) created by M. Nic, J. Jirat, B. Kosata; updates compiled by A. Jenkins. ISBN 0-9678550-9-8. doi:10.1351/goldbook. 12 S. Trassati, Pure & Appl. Chem 58 (1986) 955.
38 | C h a p t e r 1
conversion of electrochemical potential to electronic energy follows the formula stated in the top of the figure. In the figure the potentials of several other couples and reference electrodes are included. For the case of the SSC electrode, it is E0=0.222 V vs SHE 13 .
Figure 1.14. Conversion of relative electrode potentials into electronic energy for aqueous systems (reproduced from ref.12)
13
R. G. Bates and J. B. Macaskill, Pure & Appl. Chem. 50 (1978) 1701.
T h e o r e t i c a l B a c k g r o u n d | 39
1.3.3. Electrochemical cells and electrode processes With these considerations, it is now possible to describe the basic unit of an electrochemical experiment: the electrochemical cell (otherwise known as galvanic cell or electrolysis cell). Figure 1.15 shows a scheme of a usual electrochemical cell.
Figure 1.15. The basic electrochemical experiment. Arrows indicate material transport and electric field direction (E). Figure and explanation partially adapted from ref. 14 The electrochemical experiment depicted in Figure 1.15 consists in a homogeneous electrolyte solution (0.1 M NaOH in water), two electrodes of different materials (Cu and Pt), a suitable reference electrode (RE) to which the potential of the Cu electrode is quoted, a current meter and a voltage or current source (usually within the same instrument called potentiostat/galvanostat). Before entering the detailed discussion on the energetics and charge transfer at the solid-electrolyte interface, the experimental phenomena that can be observed in such an experiment will be pointed out: In first place, the case where no voltage is applied will be analyzed. As described above, in the absence of a spontaneous chemical reaction at the electrode, a dynamic equilibrium is attained. At this point it is possible to measure a voltage that implies that the potential of one of the electrodes is more positive than that of the other. The
Sverre Grimnes and Ørjan G. Martinsen in Bioimpedance and bioelectricity basics 2nd Ed. (2008) Academic Press, UK
14
40 | C h a p t e r 1
measured voltage is called the open circuit potential (OCP). A very small flux of electrons flows between the two electrodes, being its absolute value the exchange current j0. In the equilibrium the net current measured is zero. However, there are situations where equilibrium is not attained. One case is when a spontaneous chemical reaction can occur at one electrode, thus the measured potential without applying a voltage will vary until the reaction is finished. Another situation of perturbation the equilibrium potential is when more than a reaction can occur simultaneously in an electrode. For example, in an iron electrode in an HCl solution that contains FeCl2 dissolved, while an H2 atmosphere is mantained: Fe+2 + 2e- ↔ Fe
(1.28)
2H+ + 2e- ↔ H2
(1.29)
Each of these semireactions has its own exchange current density and its own equilibrium potential. The condition of overall balance at this electrode is determined by the equation:
0
(1.30)
where the subscripts (1) and (2) refer to equations (1.28) and (1.29) respectively. In this case reaction of eq.(1.28) occurs in one direction and (1.29) in the other, so Fe dissolves anodically and H2 evolves at the cathode. The new equilibrium potential will be intermediate between the individual equilibrium potentials of each reaction and it is called a mixed potential or, if it can be reproduced, a steady-state or rest potential. This particular case would be revised when the deposition of Te over Cu electrodes is studied in Chapter 4. In the case depicted in Figure 1.15, an external voltage (Uelectrode) is imposed so the current flows in the clockwise direction. The change in the potential with respect to the OCP is called electrode polarization. In the circuit section composed by electronic conductors (the wires from the electrodes to the instruments) electrons are the charge carriers and in the electrolyte solution, ions are: anions migrate to the positively polarized electrode while cations do to the negatively one. The composition of the electrolyte bulk remains constant, as the same number of positive/negative ions is leaving/entering a given volume of the electrolyte.
T h e o r e t i c a l B a c k g r o u n d | 41
In the experiment represented in Figure 1.15 with the direction of the imposed potential the Na+ ions migrate to the Pt electrode and the OH- ions to the Cu electrode. A positive current can be measured depending on the value of applied potential Uelectrode. But before continuing the analysis of the processes that occur during the experiment, is important to define the redox potential. A redox system consists of a couple of molecules or ions in a solution that can be oxidized and reduced by pure electron transfer. The corresponding reaction is: Red ↔ Ox + e-
(1.31)
in which Red is the reduced and Ox the oxidized species respectively (for example the Red: Fe+2/Ox: Fe+3). It is important to note that if the redox species remain in solution and that the electron transfer occurs in the electrode, then for studying redox systems an inert electrode is usually employed. In aqueous solutions, the tendency of one species to gain or to lose electrons is given by the redox potential. The redox potential characterizes the reaction (1.31), but as it is necessary to assign a sign, the potential is given for the reduction reaction then the redox potential is usually named reduction potential. The potential of the redox couple Uredox, is calculated by the Nernst equation: (1.32) Where U0redox is the standard redox potential measured against the SHE at 25ºC and usually found in tables, R the gas constant, T the temperature, F the Faraday constant, n the number of electrons involved in the reaction and cox and cred the concentration of the oxidated and reduced species respectively. It is important to state that the equation (1.32) is valid only for diluted dissolutions (below 1 M) and for more concentrated electrolytes the concentrations of the oxidized and reduced species must be replaced by their activities (although for this work, only diluted solutions are employed, thus the concept of activity will not be employed). The Uredox is related with the energy EF,redox of a redox couple of ions in solution described in the section 1.1.3. and depicted in Figure 1.2, and as it was discussed in section 1.3.2. the values of Uredox and EF,redox are related by: EF,redox = -eUredox-4.44
(1.33)
42 | C h a p t e r 1
In aqueous solutions, the redox potential describes the tendency of the redox system to either gain or lose electrons when it is subject to change by introduction of another redox species. A solution with a more positive reduction potential than the new species will have a tendency to gain electrons from the new species (i.e. to oxidize the new species) and a solution with a more negative reduction potential will have a tendency to lose electrons to the new species (i.e. to reduce the new species). Coming back to the example depicted in Figure 1.15, the copper electrode is in contact with the solution of NaOH. Without examining the specific mechanisms (they will be detailed in Chapter 3), the electrode will react to form an anodic oxide (given that in the proposed experiment the imposed potential causes OH- migration to the Cu electrode). The influence of the pH on an electrode reaction is illustrated by the means of U vs pH phase diagrams. Diagrams of this kind where suggested in 1963 by Marcel Pourbaix and became known as Pourbaix diagrams. In Figure 1.16 a Pourbaix diagram for the copper electrode at pH between pH 0 to 14 in the horizontal scale, and on the vertical axis is the electrode potential referred to the SHE electrode.
Figure 1.16. Pourbaix diagram for the copper electrode in aqueous solution The solid lines represent the transitions between two phases. The areas bounded by the solid lines correspond to regions of thermodynamic stability of the named substances. Dashed lines (from negative to positive potential) correspond to the evolution of
T h e o r e t i c a l B a c k g r o u n d | 43
hydrogen and oxygen respectively. From the Pourbaix diagram depicted in Figure 1.16 can be stated that Cu is inert at negative potentials except at high pH where it is dissolved in Cu(OH)2-form. Solid CuO is stable between pH 7 to 13 at U> 0 V, and solid Cu2O in a range between pH 6 and 11 and 013 into Cu(OH)32+. The Pourbaix diagrams are useful as guides to delimit the experimental conditions. However, the presence of other ions in solution, the temperature and the possible formation of phases out of thermodynamic equilibrium affect the real conditions of the experiment. To continue the description of the experiment, it is the moment to examine the Pt electrode. As a current is flowing at a first glace one might think that Na+ is being reduced at the Pt electrode. Nevertheless, when redox potential tables are consulted it is seen that the redox potential of water is more positive than that of the Na/Na+: Pt2+ + 2e- ↔ Pt
E0 = +1.118 V vs SHE
(1.34)
Na+ + e- ↔ Na
E0 = -2.714 V vs SHE
(1.35)
2H2O + 2e- ↔ H2 + 2OH- E0 = -0.8277 V vs SHE
(1.36)
Thereafter, before Na+ could be reduced, water decomposition (equation 1.36) would take place. To complete the experiment explanation, arrows indicating “diffusion” are shown in Figure 1.15. These correspond to species that can diffuse to the electrodes and be ionized or dehydrated there. This movement of a species j is caused by its concentration gradient in the vicinity of the electrodes, following the first Fick law: ,
(1.37)
where considering the ion mobility uj,the diffusion coefficient Dj (in cm2/s) is given by: (1.38) for diluted solutions. Values of Dj for most ions and neutrals are in the order of 0.62.0x10-5 cm2/s and decrease markedly with concentration.
44 | C h a p t e r 1
1.3.4. The solid-electrolyte interface1-5,14,15 The interface between the electrode and an electrolyte is the heart of electrochemistry. The electrode is the source or the sink of electrons and it is the place where the charge transfer takes place, where gradients in electrical and chemical potentials constitute the driving forces for the electrochemical reactions. Figure 1.17 shows a representation of the metal|electrolyte interface, perpendicular to the electrode surface. As the electrode is immersed in the liquid, the electrode charges distribute in the surface similarly as they do in the metal-vaccum interface, forming electrical double layers (EDLs). Depending on the surface charge ions of the opposite charge approach to the electrode surface, loosing hydration water and becoming absorbed together with water molecules, forming the so-named IHP or Inner Helmholtz Plane. Other ions approach the electrode and are held in place by electrostatic forces in a second row of ions that form the OHP, Outer Helmholtz Plane, with a maximum length around 0.3 nm, the size of a solvated ion. Finally there is an extended region of excess space charge associated with mobile ions, called the Gouy-Chapman region (or “diffuse layer”). This region contains the ions that compensate those adsorbed at the electrode, and it is only studied in very diluted electrolytes ( UFB (for p-type), resulting in injection or majority carriers into the semiconductor from the surface, being these carriers those that form the space charge. An inversion layer occurs when Uelectrode >> UFB (for n-type) or Uelectrode >p0 and using light intensities which are such that Δnp0 the recombination rate takes the form: (1.86)
70 | C h a p t e r 1
The lifetime is defined as τ= Δn/Rnp, so that (1.87) As inferred from equations (1.85-1.87) the recombination rate and the lifetime of a bans-band recombination process depends on carrier density. Recombination will be studied below when referred to solid-electrolyte interfaces. Photogenerated carriers, especially those formed in the semiconductor bulk are able to diffuse towards the space charge layer or to the surface, but the distance they are able to travel is limited because they recombine as explained before if created within a distance >dsc+Ldiff (Ldiff is the diffusion length). However, in the space charge layer the electric field separates the electrons and holes.
Figure 1.32. Excitation of electron-hole pairs by light and separation of charge carriers by the field across the space charge layer in n-type and p-type semiconductors The attainable photocurrent under a given illumination intensity depends on the relative length of light penetration in the semiconductor, the diffusion length of minority carriers and the width of space charge layer. The depth of penetration is on the order of α-1, where α is the light absorption coefficient defined as follows: (1.88) where I is the light intensity,I0 the light intensity entering the electrode, ω the frequency of light and x the distance from the surface to the bulk. Note that equation (1.88) is the same as equiation (1.22). The number of absorbed photons per unit time and unit volume is αI(x) and the rate of generation for one photon to one electron-hole pair is described by:
T h e o r e t i c a l B a c k g r o u n d | 71
,
(1.89)
The penetration depth of light may vary in a wide range since α is a function of light frequency as illustrated with the absorption coefficient of Cu2O in Figure 1.33. A more detailed description of this spectra is given in Chapter 3, but for the present purpose is enough to mention that the absorption at energies higher than the band gap (2.10 eV) are due to the generation of holes and electrons and also of excitons. A longer penetration at higher energies is expected.
Figure 1.33. a) High-energy absorption spectrum (I0/I vs. photon energy) from a thin film of Cu2O at 77 K and 295 K6. Figure 1.34 shows two extreme cases of light absorption in an n-type semiconductor, where α-1 Ldiff+LSC being LSC the width x of the space charge layer. The colors of light rays indicate higher energy (blue) and lower enrgy (green). In the region xxd+Ldiff) will recombine before reaching the surface. In the case of strong light absorption at higher energy so that α-1 > xd) by: ∆
/
(1.98)
where n0 is the bulk minority carrier density and Δn the increase carrier density due to light excitation. As large increases in minority carriers are easily achieved even at low light intensities, large photovoltages can be achieved. Since n0 is constant for a given material and Δn is proportional to light intensity equation (1.98) can be rewritten as: /
1
(1.99)
where b is a positive constant. The same expression can be derived for high absorption. It can be seen that Voc is always negative because when an n-type semiconductor is illuminated the band bending decreases. The increase of photovoltage is linear at low intensities of illumination and logarithmic at high intensities. The maximum attainable photovoltage is the size of band bending which is ideally related with the barrier height Vs=Eredox-ECB-μ. Thus the maximum Vph is about Eg. In a semiconductor|electrolyte interface for photoenergy conversion, the objective is to position Eredox close to EV for n-type materials for photoanodes or close to EC for p-type materials for photocathodes to achieve the largest band bending in the dark and thus the larger photovoltages. The maximum Voc is equal to the amount of band bending or the barrier height, which can be given as: /
(1.100)
The theoretical maximum value is not practically achievable because processes such as majority carrier current and surface recombination, thus photovoltages are not usually indicative of barrier height.
T h e o r e t i c a l B a c k g r o u n d | 77
1.6. Electrochemically prepared thin films1,5,17,18 In the introduction of this chapter thin films have been described as “materials forming layers ranging from fractions of a nanometre to several micrometres in thickness”. It has been also mentioned that films can be either deposited onto a substrate from an external source or formed by reaction of the substrate with an external agent. In this section the mechanisms involved and techniques employed to prepare thin films by electrochemical methods will be described. The models employed to describe the phenomena of electrodeposition of metals and anodic oxide formation, are the usually accepted “Simple Classical Model (SCM)” for electrodeposition and the Point Defect Model (PDM) for anodic oxides, although discussions on the application of these models are still on run in the literature. The results will be interpreted on the light of these models. 1.6.1. Basic steps in phase formation onto electrodes In applied electrochemistry, reactions that lead to formation of a new phase are rather common. Some examples include the formation of bubbles in the electrode surface, the cathodic deposition of metals and the anodic passivation or metals. In this section only the formation of new phases in the form of thin films will be considered. The first step in film formation is an electrochemical step, which produces the primary product not yet differentiated to form a new phase. In gas evolution reactions, this is the step that produces gas molecules dispersed in the electrolyte (bubbles). In cathodic metal deposition, this is the step where metal atoms are formed by discharge of the ions; these atoms are in an adsorbed state (called adatoms) on the electrode substrate and have not yet initiated a new metal phase. These steps may take place at a potential range below the equilibrium potential of the main reaction. Thus, because of their energy of chemical interaction with the substrate, metal adatoms can be produced cathodically even at potentials less negative than the equilibrium potential of a given metal-electrolyte system. This process is called the underpotential deposition of metals. Subsequent steps are the formation of nuclei of the new phase and the growth of these nuclei. These steps have two special features: 17
Gary Hodes, Chemical Solution of Deposition Semiconductor Films; Marcel Dekker:,New York, 2002. J. Mostany, B.R. Scharifker, K. Saavedra, C. Borras. Russian Journal of Electrochemistry 44 (2008) 652.
18
78 | C h a p t e r 1
1. The nuclei and the elements of the new phase generated from them (bubbles, crystallites) are macroscopic entities; their number on the surface is limited, that means that they do not emerge in the entire surface but only at a limited number of surface sites. The primary adatoms should diffuse through the surface to where a nucleus appears or grows. 2. The process as a whole is transient; nucleation is initially predominant, and nucleus growth is predominant at large time. Growth of the nuclei usually continues until they have reached a certain mean size. After some time a quasi steady state is attained, when the number of nuclei that cease to grow in unit time has become equal to the number of nuclei newly formed in unit time. Any of the steps listed can be rate determining: adatom formation, adatom diffusion, nucleation or nucleus growth. Hence a large variety of kinetic behavior is expected. Figure 1.37 gives a picture of the process of formation of a new phase onto a solid substrate where the hydrated ion is added to the lattice site by different paths. The “chosen” path affects the energetic and kinetics of the reaction, because the degree of desolvation depends on the position where the ion is located as will seen below.
Figure 1.37. Schematic representation of the electrodeposition process where a hydrated ion in the dissolution is discharged and finally arrives to a lattice site.
T h e o r e t i c a l B a c k g r o u n d | 79
1.6.2. Crystal phase formation (metal deposition) Electrochemical nucleation occurs at the boundary of an electronic conductor (the electrode) and an electrolyte (an ionic conductor). At this interface single ions discharge (or solvate) onto (or from) a nucleus, eventually developing a crystal or dissolving the deposit. Volmer and Weber and later Becker and Döring developed the so-called “Simple Classical Model” (SCM) of the kinetics of nucleation processes assuming that the crystalline aggregates of the new phase had the same composition, structure and thermodynamic properties as the bulk material. Although there are some limitations because the SCM does not take into account that clusters less than 100 atoms do not have the same properties as the bulk phase, it will be used to describe the electrochemical nucleation and growth, although other approaches have been suggested 19 . 1.6.2.1. Initial stage: incubation In the field of chemical bath deposition, the initial stage that precedes nucleation is often called “incubation” and it refers to heterogeneous ion adsorption and embryo formation. In heterogeneous nucleation, subcritical embryos or even individual ions can adsorb onto the electrode. The energy required to form an interface between the embryo and the solid substrate will usually be less than that required for homogeneous nucleation (that takes place in the solution and that would not be treated here) where no such interface exists. Therefore, homogeneous nucleation is energetically preferred over homogeneous nucleation and can occur at less concentration than that required for homogeneous nucleation. Subcritical embryos can redissolve but in a solid surface they are stabilized because the reduced contact between the embryo and the solution. In metal deposition the embryos are the adsorbed ions yet completely discharged (adatoms) or not (adions). The critical radius Rc is the size where the embryo has a 50:50 chance of either redissolve or grow into a stable nucleus; this size is determined by the balance shown schematically in Figure 1.38 between the surface energy required to form the embryo: 4
(1.101)
19
See for example E. Matthijs, S. Langerock, E. Michailovna and L. Heerman, J. Electroanal. Chem. 570 (2004) 123.
80 | C h a p t e r 1
where σ is the surface energy per unit area, and the energy released when a spherical particle if formed: 4
/3
(1.102)
where ρ is the density of the solid and L is the heat of solution. The typical size of Rc is about 100 molecules (between 1-2 nm in diameter); solvent molecules can adsorb on the embryos and change their surface energy; the critical radius will therefore depend not only on the material of the nucleating phase but also on the solution phase. In figure 1.38 the positive slope indicates dissolution and the negative indicates that embryo grows to become a supercritical nucleus.
Figure 1.38. Energetics of nucleation. The critical radius Rc, depends on the balance between surface and volume energies of the growing particle 1.6.2.2. Nucleation Once located at the electrode surface, the metallic atoms could stay adsorbed or get incorporated to the crystalline structure. The latest is only possible in active positions, where interactions with another species already present in the structure can be established. As mentioned before, the balance between the surface and volume contributions results in the value of RC for an embryo and of course, the corresponding free energy change ΔG* required to form the critical sized nucleus: ΔG
M F
K
(1.103)
T h e o r e t i c a l B a c k g r o u n d | 81
In equation (1.103) γ is the interfacial free energy at the embryo|solution interface, η the electrochemical overpotential, Φ(θ) is a function of the contact angle θ relating the volume of a spherical cap comprising the critical nucleus to the whole sphere with the same radius of curvature, ρ the density of the solid. Equation (1.103) states that the work of formation of critical nuclei depends on the thermodynamic supersaturation, corresponding to the zFη term in the electrochemical case with strong dependence on the surface energy too. According to the SCM, the steady state nucleation rate is expressed as: ∆
(1.104a)
where A0 contains the kinetics frequency factors. Let’s now consider that for deposition a number of nuclei N contribute. Consider also that the term γ can be separated into three contributions γNS, γMN and γMS, correspondingly the interfacial free energies (or surface tensions) between the nuclei-solution, nuclei-electrode and electrode-solution respectively. Then it is possible from equations (1.103) and (1.104a) to write separate nucleation rates for the case of low overpotentials and deposition over a foreign surface: A
N
A exp
MS
NS NS
MN
(1.104b)
And for high overpotentials or for the nucleation of the second and subsequent layers where there are not nuclei-electrode interfaces and the surface tension between the nuclei-electrolyte and electrode (layer)-electrolyte are practically the same: A
A exp
N
M NS FRT
(1.104c)
The total number of nuclei that contribute to the growth of the deposit is a function of the dependence of the nucleation process with time. If N0 represents the potential number of active centers and N the real number of nuclei at a given moment, the potential active centers that become nuclei at a given time is proportional to the “undeveloped” number of centers (N0-N) at the nucleation rate A and the time dt considered, thus: (1.105a)
82 | C h a p t e r 1
that gives upon integration 1
exp
(1.105b)
And that also can be written in its limit forms: (1.105c) (1.105d) For At>>1 and At105 cm-1). They are both used as the photon absorber layer in thin film solar cells bearing their names. CdS is the material of choice to form both CdTe and CIS heterojunction. CdS is a n-type semiconductor because its native Cd-vacancies and produces low or any photocurrent, thus it is called a window layer, as it only allows the photons to reach the absorber. Both in CdTe and CIS solar cells the photons entered from the n-type side of the junction. Current is carried from the front side by a transparent conducting oxide (TCO). Current is carried from the p-type back side of the junction by a metal film. Cells have been succesfully prepared by growth from the metal backcontact to the front or from the transparent window to the layer and the choice depends on the research group. CdTe usually crystallized with the zincblende structure. It has a direct band gap of 1.5 eV and can be prepared either as p-type (Cd vacancies) or n-type (Te vacancies). The scheme in Figure 1.54 shows the CdTe/CdS solar cell as prepared in the National Renewable Energy Laboratory (USA). This particular cell is prepared in a glass substrate from where the light enters; in the opposite side a TCO (Cd2SnO4) and a barrier layer (Zn2SnO4) are applied and then the CdS and CdTe films are deposited. The record efficiency in CdTe solar cells is around 17% and it has been prepared with a CdTe deposition temperature of 600ºC and a postdeposition annealing at 400ºC in the presence of CdCl2 vapor. A Cu-containing back contact forms a low resistance junction to then apply another TCO a metal or a conducting polymer. Despite the many technological challenges that a high efficiency solar cell has, one that interests to the scope of this Thesis is the back contact. It is well known that the addition of Cu to back-contacts is commonly used to improve the performance of CdTe/CdS solar cells. This may be due to the reaction with a Te-rich CdTe layer and the formation of a CuxTe/CdTe back-contact. CuxTe has more than 10 phases and an optical bandgap of 0.5–1.08 eV with different x values. Phase transitions of CuxTe can occur with differing deposition temperatures or deposition techniques, thus optimizing the phase composition is the major drawback for this material.
40
X. Wu, J. Zhou, A. Duda, J. C. Keane, T.A. Gessert, Y. Yan and R. Noufi Prog. Photovolt: Res. Appl. 14 (2006) 471
104 | C h a p t e r 1
Figure 1.54. Scheme of the CdTe/CdS solar cells prepared in the National Renewable Energy laboratories, with a record efficiency of 16% 1.9.4. Transparent conducting oxides for photovoltaics: ZnO Doped ZnO is an important oxide used in PV window and display technology applications. It has been reported to have a thin-film resistivity as low as 2.4x10–4 Ohm cm. Although the resistivity of ZnO TCO thin films is not yet as good as the ITO standard, it does offer the significant benefits of low cost relative to In-based systems and high chemical and thermal stability. In the undoped state, zinc oxide is highly resistive because, unlike In-based systems, ZnO native point defects are not efficient donors. However, reasonable impurity doping efficiencies can be achieved through substitutional doping with Al, In or Ga. Most work to date has focused on Al -doped ZnO, but this dopant requires a high degree of control over the oxygen potential in the sputter gas because of the high reactivity of Al with oxygen. Gallium, however, is less reactive and has a higher equilibrium oxidation potential, which makes it a better choice for ZnO doping applications. ZnO is of particular interest for organic photovoltaic applications because it can be readily grown into nanowires and rods, which makes it ideal for infiltration of the polymeric absorption layer. For solar cell applications, ZnO is usually prepared by electrodeposition and by sputtering. Current challenges on ZnO production are film nanostructuring, efficient deposition onto p-type absorbing layers by electrochemical methods and improvement of their electrical properties.
E x p e r i m e n t a l S e c t i o n | 105
CHAPTER 2 EXPERIMENTAL SECTION
Emblem LXV, the 'Doctor of Fools', from Johann Theodor de Bry Proscenium vitæ humanæ sive Emblematum Secularium, Frankfurt, 1627.
106 | C h a p t e r 2
2.1. Experimental techniques In this chapter the operating principle description and the theoretical basis of the experimental techniques employed in this work are given. The formalism that rules each technique is reduced to the necessary for the sake of brevity. The reader is kindly referred to the cited literature. The section 1 will explain the fundamentals of the experimental techniques employed in this work. The section 2 summarizes the employed instrumentation and the experimental details for the film preparation and characterization. The section 3 details the purpose and particularities of each set of experiments of preparation of the different materials. 2.1.1. The electrochemical techniques 2.1.1.1. Voltammetry 1 Current potential curves are usually measured with metal and semiconductor electrodes by scanning the electrode potential over a certain potential range. The potential scan leads typically to a current peak and at higher potentials the current levels off into the diffusion limited current. The peak occurs because during the first time interval sufficient redox molecules are available. The same type of peak occurs when scanning back in the reverse direction. This kind of behavior is expected for diffusion-controlled as well as for kinetically controlled reactions at metal electrodes, and also at semiconductor electrodes as long as majority carriers are involved in the charge transfer process. As an anodic or a cathodic process would occur in close dependence with the applied overpotential as the electrode potential for reduction or oxidation is reached, the information of the process kinetics and dynamics could be extracted. It has been mentioned that the charge transference at a semiconductor-liquid interface can only take place through the valence or the conduction band. Whether then a corresponding current is possible depends on various factors such as the position of the energy bands and the occupation of the energy states in the bands by electrons. Analyzing the I-V behavior of an electrode illuminated or not, its n- or p-type character can be inferred. Other useful information from I-V curves is the estimation of the thickness of a layer by application the Faraday laws for charge transference.
1
Rüdiger Memming, Experimental Techniques in Semiconductor Electrochemistry, Wiley-VCH (2001) Weinheim, Germany.
E x p e r i m e n t a l S e c t i o n | 107
2.1.1.2. Electrochemical Impedance Spectroscopy (EIS)1,2,3,4,5 As mentioned above, in the description of metal|electrolyte interfaces the simple model of a plate condenser works well for stationary conditions. In the simplest case of an elemental electrochemical system, the solid|liquid interface can be described by a series resistance Rs, a charge transfer resistance Rct and a capacity in parallel with it that stands for the capacity due to the Helmholtz layer CH for a metal electrode or for semiconductor electrodes a space charge capacity (Csc). In real electrodes, the behavior is better described if the capacitor is replaced by a constant phase element (CPE or Q) which characteristics will be described in detail below. This is the so-called R(RC) circuit or Randles circuit, depicted in figure 2.1.
Figure 2.1. Randles circuit. The total current J that circulates upon the circuit is the sum of the current flowing through the resistor and the capacitor element. Electrochemical impedance is usually measured by applying a small AC potential (U(t)) in a range of frequencies (f) typically in the 0.1 Hz to 1000 kHz range to an electrochemical cell and measuring the current through the cell. The AC potential must be little, around kT/e (> n0, light intensities where Δp < p0 and Δn >>>n0, can be employed and thus under illumination the observed Flat Band potential is expected to shift towards cathodic potentials. Figure 2.4 depicts the processes occurring during the photoexcitation with and without applied potential. Additionally to the Flat Band potential under illumination, other information such as rate constants of redox processes, electronic structure insights, and carrier lifetimes, can be extracted from photocurrent measurements. In a photoelectrochemical experiment, light in the visible and near ultraviolet part of the electromagnetic spectrum is employed. The incident radiation is characterized by the energy of its photons, hv (where h is the Planck constant) and by its intensity I0.
Figure 2.4. Scheme of the carrier generation upon illumination and charge separation when bands are bent by application of an external potential.
E x p e r i m e n t a l S e c t i o n | 111
2.1.2. Reflectance and Transmittance The band gap energy is one of the fundamental parameters of a material, particularly for semiconductor compounds. It is optically determined at photon energies above the gap from changes in the optical constants n and k, the real and imaginary parts of the refraction index. The experimental procedures involve in general reflectance (R) or transmittance (T) measurements. Figure 2.5 illustrates the reflection and transmission phenomena. From the multiple reflections in the inner face of the film, interference patterns are formed from where the information of the thickness d can be extracted.
Figure 2.5. Reflection and transmission phenomena In the case of a rough film part of the reflected light is lost by scattering as schematized in Figure 2.6a. For this case an integrating sphere is employed. The light has a low incidence angle (8º) on the film and the interior of the sphere has a high reflecting coating that directs all the light reflected from the film to the detector. Figure 2.6.b shows the scheme of an integrating sphere.
Figure 2.6. a) Diffuse reflectance from a rough surface; b) scheme of an integrating sphere
112 | C h a p t e r 2
2.1.3. Structural and chemical techniques 2.1.3.1 X-ray diffraction (XRD) 6 7 The periodic arrangement in the atoms or molecules in a crystalline material can be modeled as a diffraction grating for waves with a wavelength on the order of the separation of atoms that correspond to the magnitude order of the x-ray wavelength (0.5-2.5 Å). An x-ray diffractogram contains information about the spacing and spatial position of the atoms inside a material. In the Bragg approach, the diffraction is considered as the crystals are arranged in layers of atomic planes and acting as semitransparent mirrors. Some of the incident x-rays will be reflected with a reflection angle equal to the incidence angle while the rest of the x-ray will be transmitted and reflected by the following planes. The diffracted beam will be detected only when the reflections coming from parallel planes interfere constructively as summarized by the Bragg law:
2 d sin θ = n λ
(2.1)
where d is the distance between two planes, θ the incidence angle of the beam and n a of the wavelenght λ, and represented in the Fig 2.7.
Figure 2.7. Scheme of the x-ray diffraction phenomenon From a diffraction pattern the position of the diffraction maxima, the peak intensities and the intensity distribution as a function of diffraction angle can be used to identify and quantify the composition of the sample, as well as to calculate the material’s crystallite size and distribution, crystallinity, stress and strain. A diffraction pattern is 6
Anthony West “Basic Solid State Chemistry”, John Wiley and Sons (1988), Essex UK. Ron Jenkins, X-ray Techniques: Overview, in Encyclopedia of Analytical Chemistry R.A. Meyers (Ed.), pp. 13269–13288, John Wiley & Sons Ltd, Chichester, 2000
7
E x p e r i m e n t a l S e c t i o n | 113
characteristic of the atomic arrangement within a given phase and to this extent it acts as a fingerprint of that particular phase. Thus by use of the ICDD PDF database a series of potential matches can be obtained. In the case of diffraction from thin films it must be taken into account that the amount of material available to diffraction does not correspond to an infinite solid as considered in bulk or powder diffraction. The positions of the diffraction maxima could be shifted either by film strain, by misalignment of the sample with respect to the incident beam, or by lack of planarity of the substrate. Also it must be taken into account that because the small angles of the diffracted beam the planes that reflect are mainly those perpendiculars to the substrate surface. Despite this inconvenient, Grazing Incidence Xray diffraction (GIXRD) provides a convenient way to study films in the case where more sophisticated techniques such as surface diffraction or x-ray reflectometry are not available, although the quantification of the data serves mainly as reference. 2.1.3.2. Raman spectroscopy The Raman effect comes from the inelastic relaxation of absorbed light in a material, due to the energy difference that arise with the absorption or generation of a phonon, plasmon or other vibrational particle. The amount of the polarizability change will determine the intensity, whereas the Raman shift in frequency is equal to the involved vibrational level. In the Figure 2.8 is schematized the phenomenon: the Rayleigh scattering corresponds to elastic relaxation of the supplied energy supplied while the Stokes and antiStokes scattering correspond to the Raman effect. As the vibrational levels are characteristic of each material, identification of unknown compounds is possible with Raman spectroscopy. Moreover, because the vibrational spectrum is sensitive to the local chemical environment, Raman spectroscopy is widely used to characterize the properties of materials. Crystal lattice dimensions change when dopants are added, or materials are subjected to mechanical or thermal stresses. When using a microscope to focus the excitation laser over the sample and collecting the Raman generated photons in the backscattering geometry, spatial resolutions of about 1 micron can be achieved, this setup is called microRaman spectroscopy. As the Raman signal is intrinsically low there exist several strategies to enhance it. Among these is the Surface Enhanced Raman effect. Although the nature of the SERS effect is still matter of discussion, the fact is that the intensity of the Raman signal is enhanced several orders of magnitude. The effect was first discovered by
114 | C h a p t e r 2
electrochemists, and roughened metal electrodes are still widely used for SERS. The roughening is made by cycling an electrode once or several times between potentials at which the metal is oxidized and reduced. The major disadvantages are that roughened electrodes could present several crystalline faces or adsorb Cl- commonly used to the roughening process, however the sensitivity increase allow monolayer detection.
Figure 2.8. Raman Effect scheme For example, Cu oxidation in alkaline media has been studied by SERS and it was found an important dependence of the observed bands in dependence of the electrolyte employed, as observed in Figure 2.9 where it is observed the 625 cm-1 band characteristic of Cu2O and bands at 525 cm-1 that appear with NaOH and a band at 290 cm-1 observed with the use of HClO4 and H2SO4.
Figure 2.9. SER spectra of a Cu electrode in the different media marked in the figure at Uelectrode= -0.2 V vs SCE 8 8
Ho Yeung H. Chan, Christos G. Takoudis and Michael J. Weaver, J. Phys. Chem B (1999), 103, 357
E x p e r i m e n t a l S e c t i o n | 115
2.1.3.3. X-ray photoelectron spectroscopy (XPS) 9 X-ray photoelectron spectroscopy (XPS) is a surface analytical technique, which is based upon the photoelectric effect. Each atom in the surface has core electron with the characteristic binding energy that is conceptually, not strictly, equal to the ionization energy of that electron. When an X-ray beam directs to the sample surface, the energy of the X-ray photon is adsorbed completely by the core electron of an atom. If the photon energy, hν, is large enough, the core electron will then escape from the atom and emit out of the surface. The emitted electron with the kinetic energy of Ek is referred to as the photoelectron. The binding energy of the core electron is give by the Einstein relationship: (2.2) where hν is the X-ray photon energy (for monochromatic Al Kα, hν = 1486.6eV); Ek is the kinetic energy of photoelectron, which can be measured by the energy analyzer; and is the work function induced by the analyzer, about 4~5eV. Since the work function, , can be compensated artificially, it is eliminated from equation (2.2), giving the binding energy as follows: (2.3) For insulating samples, once the photoelectrons are emitted out of the sample surface, a positive charge zone will establish quickly in the sample surface. As a result, the sample surface acquires a positive potential (varying typically from a few to tenths of volts) and the kinetic energies of core electrons are reduced by the same amount, C. (2.4) It can be seen that the surface charging results in the shift of the XPS peaks to higher binding energy. In this case, the binding energy has to be calibrated with an internal reference peak. The C 1s peak from the adventitious carbon-based contaminant, with the binding energy of 284.8eV, is commonly used as the reference for calibration. In order to neutralize the surface charge during data acquisition, a low-energy electron flood gun 9
Surface analysis method in materials science, edited by D. J. O'Connor, B. A. Sexton, R. St. C. Smart, (1992) Springer-Verlag, Heidelberg,
116 | C h a p t e r 2
is used to deliver the electrons to the sample surface. The electron flood gun can be tuned to provide the right current to push the XPS peaks back to the real position. The core electron of an element has a unique binding energy, which seems like a "fingerprint". Thus almost all elements except for hydrogen and helium can be identified via measuring the binding energy of its core electron. Furthermore, the binding energy of core electron is very sensitive to the chemical environment of element. The same atom is bonded to the different chemical species, leading to the change in the binding energy of its core electron. The variation of binding energy results in the shift of the corresponding XPS peak, ranging from 0.1 eV to 10 eV. This effect is termed as "chemical shift", which can be applied to studying the chemical status of element in the surface. Therefore, XPS is also known as electron spectroscopy for chemical analysis (ESCA). Since the number of photoelectron of an element is dependent upon the atomic concentration of that element in the sample, XPS is used to not only identify the elements but also quantify the chemical composition. After the value of peak intensity (the peak area after background removal) is obtained, the atomic concentration of an element, Ci, can be expressed as:
∑
(2.5)
Where Ii is the peak intensity for element i, and Si is the sensitivity factor for the peak i. 2.1.4. The Scanning Probe techniques Scanning Probe Microscopy (SPM) techniques provide three dimensional real space images and allow spatially localized measurements of structure and properties. A scanning probe microscope is similar to a profilometer in that the image is obtained by scanning a tip over the surface of the sample. The motion of the tip is controlled at the angstrom level and the tip can be in contact with the sample or in intermittent contact or in “non contact” at all. The motion in all three directions is controlled by piezoelectric elements that cause the scanning of the surface and within the feedback signal the control of z position. The scanner is common to a variety of techniques that vary in the detection form employed to assess the tip movements, giving access to different property mapping such as magnetism, friction, and conductivity between other. All SP
E x p e r i m e n t a l S e c t i o n | 117
microscopes consist of the tip, the scanner, the detector, the electronic control system and the vibration isolation system. In the Figure 2.10 is shown the scheme of a typical SPM system.
Figure 2.10. Scheme of a basic SPM system 2.1.4.1. Atomic Force Microscopy (AFM) The SPM variants are based on the detection of forces between the tip and the sample that depend on the separation between the tip and the sample. During contact, the dominant interactions are repulsive Van der Waals forces. However, when the tip is lifted above the substrate, longer range forces such as electrostatic or magnetic forces may dominate the interaction between the sample and the tip. This section deal with the interactions measured with the tip in close proximity of the sample, i.e. atomic forces. The underlying principle of AFM is that the interactions between the ends of a probe tip mounted at the end of a cantilever results in a response in the cantilever, notably a deflection. In principle the deflection can be measured and keeping the deflection constant by varying the vertical position of the tip produces a constant force image, that when done sufficiently close to the surface where the van der Waals forces dominate, the image represents the topographic structure of the surface. Visual information as well as quantitative data such as grain size and roughness can be obtained after the appropriate image treatments.
118 | C h a p t e r 2
2.1.4.2. Current Sensing Atomic Force Microscopy (CAFM) 10 This is a working mode derived from the contact AFM. It is based on the measure of the current intensity that passes between a conductive probe (usually a metal-covered AFM tip) when a potential difference is applied to the sample. The probe is grounded and works as current sensor. In this mode it is possible to capture simultaneously the topographic information coming from the tip deflection as well as the intensity map. With both channels is possible to relate the topographic features with their electrical response. Current values between 2 pA and 1 µA are possible to be measured. It is also possible to ramp the applied potential while maintaining the tip on a fixed point to obtain local current-voltage curves that allow quantification of the electrical properties of the sample. Figure 2.11A) shows the electric diagram of the setup for a p-type semiconductor under positive and negative bias. Two overimposed topographic-electric images illustrate the differences in conductivity of the same region of the sample depending on the applied biases. Figure 2.11B) illustrates a local current-voltage measurement in a p-type semiconductor (Cu2O) that shows the direction of the electron flux under forward and reverse bias, considering the semiconductor-tip as a Schottky junction.
A
10
Application Modules: Dimension and MultiMode Manual, Veeco Instruments Inc (2003)
E x p e r i m e n t a l S e c t i o n | 119
B Figure 2.11. General scheme of the CAFM technique for a p-type semiconductor; A) over imposed topographic and current images with the schemes that indicate the flux of the majority carriers at positive bias and minority carriers at negative bias; B) typical IV-curve indicating the forward (positive bias) and reverse (negative bias) directions of the Tip|Cu2O Schottky diode 2.1.4.3. Scanning Tunneling Microscopy and Spectroscopy 11,12 Scanning Tunneling Microscopy (STM) and Spectroscopy (STS) rely on electron tunneling, a phenomenon that is based on quantum mechanics, schematized in Figure 2.12. In a metal or a semiconductor, electrons exist within an energy range as mentioned before. At the interface between that material and an insulator there is an energy barrier or tunnel barrier (Fig 2.12 top). If a second metal or semiconductor is placed near the first, and a voltage is imposed between the two, then the shape of the energy barrier is changes and there is a driving force for electrons to move across the barrier (Fig. 2.12 bottom). In classical mechanics the electron cannot travel across the barrier, but go over it only if its energy is raised to a value larger than Eb,1. In quantum mechanics a finite number of electrons are allowed to traverse the barrier if the thickness z is small enough.
11
Dawn Bonnell (Editor); Scanning Probe Microscopy and Spectroscopy: theory, techniques and applications; 2nd Edition, John Wiley & Sons Ltd, Chichester, 2001 12 Ismael Díez-Pérez, Probing the passivity of Iron by electrochemical scanning tunneling microscopy and spectroscopy, PhD Thesis, Universitat de Barcelona, October 2006
120 | C h a p t e r 2
Figure 2.12. Electron tunneling scheme between two materials, Top: no applied bias, Bottom: tunneling upon applying a bias between the materials The barrier width depends on the potential in the barrier and in the electron energy and the probability that an electron will cross the barrier is the tunneling current (I) and it decays exponentially within the barrier width z as:
I ∝ e − 2 κz
(2.6)
where κ:
κ=
2m(V − E) h
(2.7)
where m is the electron mass, V the potential in the barrier, E the electron energy and ћ the reduced Planck constant. Only electrons between the Fermi levels of the two materials can tunnel, due to the constraint that electrons must exist in a filled state at the energy in the negative material and an unfilled state must exist at the energy in the positive material. In STM imaging a sharp tip often made of Pt or W is brought in close proximity of a planar sample surface. When the two surfaces are sufficiently close that the wave functions overlap, the resulting current I is: 1
I = Cρ t ρs e
zk 2
(2.8)
E x p e r i m e n t a l S e c t i o n | 121
where C is a constant, ρt is the tip electronic structure, ρs the sample electronic structure and z the sample-tip separation. It can be observed the exponential dependence of the tunnel current on the sample-tip distance, and it can be concluded too that this sensitivity allow the tunnel current to be used to control the separation with high vertical resolution. The contrast in any STM image is a convolution of the electronic structure of the sample and the electronic structure of the tip. It is only when the latter is minimal that the variation of contrast in the image can be attributed to electronic properties of the sample. Fortunately, this condition can be easily met and equation (2.8) is often a reasonable approximation. 2.1.4.4. Electrochemical Scanning Tunneling Microscopy (ECSTM) As mentioned in the electrochemistry section in Chapter 1, a typical three electrode system intended to study the processes that occur in the working electrode, consists of the sample that is the working electrode itself, a reference electrode against which the potentials are compared and a counter electrode to complete the circuit through which an external potential is applied. In the case of ECSTM, a fourth electrode is immersed in the electrolyte, i.e. the STM tip that acts also as a second working electrode, which potential is compared against the same reference. As in every electrode, through the immersed tip will flow a faradaic current which magnitude depends on the exposed tip area, the electrolyte composition and the tip potential Et. When the tip approaches to the substrate is affected by its presence and a very close separations, the tunnel current It begins to flow and the total tip current is given by I = If + It. However, the faradaic current could be as large that would interfere with It thus it is necessary to minimize If. Usually this is accomplished by insulating the tip leaving exposed as little surface – ideally an atom- as possible and the If can be minimized to negligible levels (typically half a magnitude order of It or even less to ECTS studies). Another issue in ECSTM is the design of the electrochemical cell that must fulfill the requirements of low volume to avoid vibration problems, good electrical contact between the cell and the working electrode, be small to be introduced into the STM head environment, be hermetic, and have space to hold the counter and reference electrodes. Finally another important requirement for ECSTM is the independent control of the tip and sample potentials. Within the modern instruments, this requirement can be easily accomplished by the use of bipotentiostats.
122 | C h a p t e r 2
2.1.4.5. Electrochemical Tunneling Spectroscopy (ECTS) Tunneling spectroscopy performed in a STM provides information about the electronic structure of the sample by probing the sample density of states as a function of energy. There are also two methods to perform STS. The first referred to as point spectroscopy, involves moving the tip to a feature of interest, disengaging the feedback mechanism, modulating the tip bias and recording the resulting variation of the current. As the bias is ramped, either positively or negatively, the current varies in response to the changing electron density in the energy window. Ramping in both directions probes both the occupied and unoccupied energy states, with the magnitude of the current at a specific voltage directly related to the density of states (DOS) of the sample at that energy. The second method to obtain spectroscopic information involves simultaneously collecting images at various biases. This can be accomplished by modulating the bias at a high frequency and recording the current at several discrete values of applied bias or using the same sample and hold method of point spectroscopy at every image point. A comparison of the periodic contrast in such images with atomic charge superposition and the total energy calculations indicates the spatial position of the energy states with respect to the lattice.
2.2. Instrumentation and experimental details The specific instrumentation and experimental setups employed in this Thesis for film preparation and properties characterization are described in this section. 2.2.1. Electrochemical instrumentation The electrochemical preparation of the layers and the electrochemical impedance spectroscopy
measurements
were
performed
using
an
Autolab
PGSTAT12
potenciostat/galvanostat (Ecochemie, NL) shown in Figure 2.13, equipped with a Frequency Analyzer Module (FRA2 module) and an ECD module for low current measure (minimum current: 100 pA). The maximum current load of the PGSTAT12 is 250 mA and the maximum potential 12 V, and it is possible to operate in 2, 3 and 4 electrode configuration. The FRA2 module allows EIS measurements in a frequency range from 10 μHz to 1 MHz and voltage amplitudes from 0.2 mV to 350 mV.
E x p e r i m e n t a l S e c t i o n | 123
Figure 2.13 Photograph of the Autolab PGSTAT12 potentiostat/galvanostat and the electrochemical cell usually employed for Cu2O growth. Cyclic voltammetry (sweep rates from 0.5 to 100 mVs-1), linear voltammetry, programs consisting in series of potential/current steps and sweeps and pulses were done for preparation and characterization of the films. For pulse cycles the software only can manage up to 32000 cycles, thus for longer experiments, several manual repetitions have to be done. There is the possibility to introduce/extract analogical signals from external devices. This is still to be implemented for improved photocurrent measurements and for external pulse driving. The electrochemical impedance spectroscopy measurements (EIS) were performed with the FRA software of Autolab. The excitation amplitude was set at 25 mV ~ kT and different excitation frequencies between 0.1Hz to 10 kHz were typically employed. Impedance vs frequency spectra were acquired at fixed sample potential. The data were represented in Bode and Nyquist plots. These were adjusted by the convenient equivalent circuit on the FRA software to obtain the circuit characteristics, and to determine the frequency of maximum phase shift, to use this frequency to made the impedance vs potential measurement. The Z vs U data were plotted in the MottSchottky form to extract the information of Flat Band potential, semiconductor type, carrier density and width of the SCL. 2.2.2. Photocurrent measurements The PCS measurements were done using a 150 W high intensity illuminator (Fiber Lite MI150, Dolan-Jenner Inc, USA), a mechanical chopper and a dark box with a mechanical shutter with the electrochemical cell placed inside or with the cell wrapped in aluminum paper and the light directed to the bottom of the cell. The light reaches the
124 | C h a p t e r 2
sample through a mechanical collimating hole and the photocurrent was registered both in the constant wave mode or chopped with a 10 Hz or a 20 Hz frequency at a fixed sample potential or scanning the sample potential from anodic to cathodic potentials to measure the photocurrent amplitude or the photocurrent evolution with the potential.
Figure 2.14. Setup for constant wave photocurrent measurement 2.2.3. Electrochemical cells A home-made glass cell with the working electrode placed at the bottom was employed in the three electrode configuration with the Cu plate as working electrode, a high surface Pt/Ir wire as counter electrode and a true Ag/AgCl reference electrode at the end of a Luggin capillary filled with the same electrolyte. A lower volume cell with the same configuration was employed to prepare the Cu2-xTe films, to reduce the amounts of Te solution needed for experiments. The reference electrode was a 2 mm dia., miniature silver/silver chloride true reference electrode from World Precision Instruments (WPI Inc, USA). This electrode has a small chloride leakage rate around 5.7x10-8 mL/hr that for most of the experiences is quite satisfactory. If lower leakage rates are needed an additional membrane filled with the working electrolyte is added to the electrode tip. Figure 2.15 present the conventional setup for thin film deposition. For the case of ZnO temperatures up to 65ºC are required, thus a glass cell with an external jacket was designed as well as a substrate holder shown in Figure 2.16b). The mains differences are the contact system designed for the ITO substrate that is not conductive in the
E x p e r i m e n t a l S e c t i o n | 125
backside and the reference electrode, that in this case is a Pt/Ir wire, because the potential fluctuation of the Ag/AgCl electrodes with the temperature.
Figure 2.15. Left: armed setup and water thermostat system, Right: detail of the ITO mounting, conductive phase is upwards. An additional cell was designed for photocurrent spectroscopy measurements and insitu transmittance (figure 2.16). The cell fits in a modified 1x1 cm2 cuvette holder. The holder is connected by optical fibers with an intermittent LED light source powered by a DC-signal generator instead of the mechanical chopper (on-off frequencies as high as 2 kHz) and to a mini spectrophotometer that has a CCD detector that measures light intensities in wavelengths from 350 to 1050 nm.
Figure 2.16. Cell mounted on the adapted cuvette holder connected by optical fiber with the mini spectrophotometer from Ocean Optics.
126 | C h a p t e r 2
2.2.4. Optical transmittance and reflectance Measurements were performed in a double-beam Shimadzu UV 2101 PC spectrophotometer in the range from 200-900 nm with a step of 0.2 nm. Transmittance measurements of ZnO/ITO films were done by illuminating the film/substrate system from the substrate size to avoid light scattering due to film roughness. An ITO substrate was placed in the reference side to directly obtain the transmittance spectra of the ZnO film. The Cu2O/Cu films were measured in the same spectral range by diffuse reflectance, using an integrating sphere with a freshly polished Cu substrate as reference. The baseline for transmittance was taken in air while for diffuse reflectance BaSO4 powder was used. Adequate black masks were employed to direct the light spot only on the film area. 2.2.5. Structural and chemical measurements 2.2.5.1. X-ray diffraction The films were measured by grazing incidence x-ray diffraction using unfiltered Cu Kα radiation, in a Siemens D500 diffractometer. Different incidence angles (0.4º, 0.8º and 1.2º) were used to look for the best intensity/noise ratio. Selected samples were measured in the Bragg-Brentano geometry (θ-2θ coupling) with in a Panalytical Xpert Pro diffractometer equipped with an automatic sample loading arm. Step sizes and step times were adjusted to obtain the best signal/noise ratio. The phases were identified within the PDF database version 1997 13 . Figure 2.17 (left) shows the goniometer of the D500 equipment with the detector that pivots around the 2θ circle. Figure 2.17 (right) shows the Panalytical diffractometer.
Figure 2.17 Left. Goniometer of the D500 diffractometer; Right. Panalytical Xpert Pro diffractometer for Bragg Brentano measurements 13 Powder Diffraction File Database, International Center for Crystallographic Data 1997
E x p e r i m e n t a l S e c t i o n | 127
2.2.5.2. Raman spectroscopy Films were measured by microRaman spectroscopy in a T64000 (shown in Figure 2.18) from Jobin Yvon equipped with a triple monochromator and a CCD detector using the Ar laser green line at 520 nm focused onto the sample with different microscope objectives to reach distinct effective power densities. The power density was adjusted to increase the signal or decreased to avoid structural changes or damage by heating or oxidation in the film. For Raman measurements of copper oxide, the Cu2O layers were grown onto a Cu substrate previously roughened by a 15 cycles of oxidation-reduction in a KCl 0.1M electrolyte, cycled from -1000 to 400 mV at 100 mVs-1 to intend to enhance the Raman signal by SERS effect, as preliminary experiences do not give a valuable signal. A Cu2O reference spectra was compared with the sample. Cu2-xTe films were directly measured, as Te usually gives a strong Raman signal.
Figure 2.18. Raman spectrometer Jobin Yvon T64000 2.2.5.3. X-ray photoelectron spectroscopy Selected Cu2O samples were measured by XPS. The XPS analyses were performed in a Perkin-Elmer PHI 560 ESCA-SAM system, equipped with a double-pass cylindrical mirror analyzer, with a base pressure of 1.3x10-7 Pa. Argon ion sputtering was performed with 4 keV energy ions and 0.36 μAcm-2 current beam, yielding an approximate 3 nm/min sputtering rate. All XPS spectra were obtained after 5 min of sputtering with Ar+ ions. The utilized low current density in the ion beam and short cleaning time reduce possible modifications in the stoichiometry of the Cu2O surface. For the XPS analyses, samples were excited with 1486.6 eV energy AlKα X-rays. XPS
128 | C h a p t e r 2
spectra were obtained under two different conditions: (i) a survey spectrum mode of 0– 1200 eV, and (ii) a multiplex repetitive scan mode. No signal smoothing was attempted and a scanning step of 1 eV/step and 0.2 eV/step with an interval of 50 ms was utilized for survey and multiplex modes, respectively. The spectrometer was calibrated using the Cu 2p3/2 (932.4 eV) and Cu 3p3/2 (74.9 eV) lines. Binding energy calibration was based on C 1s at 284.6 eV.
Figure 2.19. X-ray photoelectron spectroscopy (XPS) ESCA-SAM 560 Perkin-Elmer 2.2.5.4. Scanning electron microscopy and Energy dispersive x-ray spectroscopy The morphology of films was measured in a JEOL JSM840 scanning electron microscope equipped with an x-ray microanalysis probe INCA Energy 250 from Oxford Instruments. The base pressure of the microscopy chamber was below 1x10-5 Torr. Images were acquired with an accelerator voltage of 15 kV and with a filament current of 2x10-10 A. Energy Dispersive spectra (EDS) were acquired by limiting the zone to measure at the maximum microscope magnification. The accumulation time for each spectrum was 100 s. The elements were identified by their X-ray emission peaks in the instrument software. Figure 2.20 shows a photograph of the instrument.
Figure 2.20. Scanning Electron Microscope Jeol JSM840 with EDS probe
E x p e r i m e n t a l S e c t i o n | 129
2.2.6. SPM measurements 2.2.6.1. Atomic force microscopy: topography mode The morphology of the samples was studied by Atomic Force Microscopy (AFM) using a Multimode microscope head controlled with a Nanoscope IIIa electronics (Digital Instruments, Veeco Metrology Group, St. Barbara Ca, USA). All the images were performed in air, in the tapping mode with Si cantilevers of 35 N/m spring constants (Nanosensors, Wetzlar-Blankenfeld, Germany).
Figure 2.21.a) Multimode AFM microscope with magnifier; b) microscope head 2.2.6.2. Current sensing atomic force microscopy Current sensing atomic force images and IV curves were obtained in a Dimension 3000 Microscope from Veeco, controlled with a Nanoscope IV electronic system. The microscope is mounted inside a Faraday cage and allows samples of 20x20 cm2. To perform the measurements, a preamplifier module Extended TUNA (Veeco) capable to record signals about 1 pA was installed. The employed tips were conductive Si tips coated with a Cr/Co film, with final tip radius about 30 nm. The images were obtained at constant deflection, at a tip scanning rate of 0.50 Hz at different magnifications. The sample bias was varied to obtain electrical images simultaneously to topographic images at the different band bending; the limits of the applied bias were when visible sample damages or abrupt increase in the current was observed. The IV curves were registered after locating a fairly conductive zone in the electrical signal image and then applying the most negative bias to scan from the negative to the positive voltages and
130 | C h a p t e r 2
back. The scanning rate was 0.1 Hz. At least 10 IV curves were acquired in the same sample to make an average. CAFM measurements were usually made at room temperature with N2 flux to reduce the ambient humidity but also measurements were done without humidity control to assess the effect of surface water on the electrical properties of the films.
Figure 2.22. Dimension 3000 microscope showing details inside the faraday cage.
A
B Figure 2.23 a) Tip holder for the CAFM; b) detail of the AFM scanner and current amplifier
E x p e r i m e n t a l S e c t i o n | 131
2.2.6.4. Electrochemical Scanning Tunneling Microscopy and Spectroscopy The ECSTM and ECSTS studies were performed at room temperature by using a Molecular Imaging microscope head (Phoenix, AZ, USA) controlled by Nanoscope IIIa electronics (Digital Instruments, Veeco Metrology Group, St. Barbara Ca, USA). The STM electrochemical cell was made of Teflon and exposes a 0.30 cm2 area to the solution through an O-ring. A Pt/Ir wire concentric to the sample was used as counter electrode and the mini Ag/AgCl electrode described above was used as reference. The ECSTM and ECTS studies were performed at room temperature by using a Molecular Imaging microscope head (Agilent, Phoenix AZ, USA) controlled by the Nanoscope IIIa electronics. The STM electrochemical cell was made of Teflon and exposes a 0.30 cm2 area to the solution delimitated by an O-ring. A Pt/Ir wire concentric to the sample was used as auxiliary electrode and a SSC electrode was used as true reference. ECSTM tips were prepared with the procedure described elsewhere39. The faradaic tip current at potentiostatic conditions and far from the surface was typically better than 0.1 nA for imaging studies, and better than 0.01nA for spectroscopic recordings. The STM images were recorded in the constant current mode at typical current setpoints ranging from 0.50 nA to 2.0 nA. A CV was acquired before each set of experiments to verify the state of the oxide and the range of sample potentials. Images were acquired at different sample potentials and using the entire electrochemical range of the tip while recording imaging conditions. Before image acquisition, the sample current was allowed to stabilize for 5 min. The AFM and ECSTM images were analyzed with the WSxM freeware v.4.0 Develop 8.21 14 . The ECTS setup and methodology is extensively described elsewhere 15 16 17 18 . Briefly, the measurements were performed with a freshly prepared STM tip engaged at a current setpoint of 1.5 nA. Feedback was momentarily disconnected and a potential ramp was applied to the tip. The tunnel current was recorded and digitized in real time with a digital oscilloscope (Tektronix). Blank curves with the tip placed far from the surface were also recorded to subtract the contribution of the non-tunneling "leakage" current. Spectra were acquired at different sample potentials within the range of electrochemical stability of the oxide. 14
Horcas, I.; Fernandez, R.; Gomez-Rodriguez, J. M.; Colchero, J.; Gomez-Herrero, J.; Baro, A.M. Rev. Sci. Instrum. 2007, 78, 013705. 15 Diez-Perez, I.; Gorostiza, P.; Sanz, F. J. Electrochem. Soc. 2003, 150, B348. 16 Güell, A.G.; Díez-Pérez, I.; Gorostiza, P.; Sanz, F. Anal. Chem. 2004, 76, 5218. 17 Díez-Pérez,I.; Güell, A.G.; Sanz, F.; Gorostiza, P. Anal. Chem. 2006, 78, 7325. 18 Díez-Pérez, I.; Sanz, F.; Gorostiza,P. Electrochem. Comm. 2006, 8, 1595.
132 | C h a p t e r 2
Figure 2.24a) Microscope support, head and chamber; b) microscope head; c) cell installed in the microscope, ready to measure.
2.3. Methods developed for film preparation In the next sections the methodologies developed during this thesis for Cu2O, Cu2-xTe and ZnO film preparation will be described. 2.3.1. Cu2O Cu2O layers were growth onto polycrystalline Cu disks (99.99%, Goodfellow) of 10 mm dia x 1 mm thick, sequentially polished with Al2O3 emery paper of 9 µm, 3 µm and 1 µm and finally with Al2O3 paste up to 0.3 µm. After polishing, disks were ultrasonically rinsed with MQ water and afterwards Ar-blown. Electrode surfaces were mirror-like and present a typical RMS roughness of 5 nm (measured by AFM). The electrolyte employed was 0.10 M NaOH prepared from pro analysis purity grade chemicals from Riedel de Häen and MQ water. The electrolyte was always freshly prepared and its pH ranged from 12.8 to 13.0. Before each experiment the electrolyte was bubbled with high purity Ar to remove dissolved O2/CO2. The Cu2O films were growth in stepwise routine: a) after introduce the electrode onto the cell, the native oxide was reduced at -1800 mV during 5 min. Then the sample potential was graded in 100 mV steps up to -1000 mV by 300 s to reach a metallic Cu surface; b) once stabilized, the potential was then stepped successively to the OHadsorption potential, to the Cu+-dissolution potential and to -400 mV that corresponds to the maximum of the oxidation peak and then left during 5 hr; c) afterwards the potential was stepped to -300 mV and then left during 1.5 hr. The electrode is removed, rinsed with MQ water, Arblown and stored to be measured ex-situ or left at -300 mV before being measured
E x p e r i m e n t a l S e c t i o n | 133
electrochemically. Cu2O layers have reddish or slightly green appearance, smooth surfaces and are remarkably stable upon aging, as observed by its photo electrochemical properties. Figure 2.25 depicts the freshly polished copper substrate and a copper oxide sample prepared with the described procedure.
Figure 2.25. Copper substrate (left) and Cu2O film anodically prepared right
2.3.2. Cu2-xTe We used as substrates 1 cm dia. polycrystalline Cu disks (99.99 %) polished up to 0.3 µm with Al2O3 paste. A solution 25 mM of TeO2 in 0.2 M HCl as supporting electrolyte was employed as Te (IV) source. Experiments were performed in an electrochemical glass cell in a standard three electrode configuration with an Ag/AgCl (SSC from herein) true reference electrode and a platinum auxiliary electrode. To highlight the effect of the applied potential in the structural properties of the films, three different experimental steps were designed. A) The TeO2-HCl solution was left in contact with the copper electrode without intentional potential application until equilibrium is achieved and a deposited layer was observed. B) A second experiment started with the A experiment equilibrium situation, followed by a cathodic potential sweep up to -550 mV vs SSC. C) The third experiment was based on the preceding steps plus an anodic sweep up to the onset of Cu oxidation. Figure 2.26 shows the typical samples obtained within these three experiments.
Figure 2.26. Samples prepared following the experiment A, B and C from left to right respectively.
134 | C h a p t e r 2
2.3.3. ZnO Indium tin oxide (ITO, In2Sn2O7-x) coated glass slides (coating thickness 60-100 nm) with a sheet resistance of 15-25 Ω-square (Aldrich) were used as substrates. Slides were carefully cut (in the non-conductive side) with a diamond pencil in 13x13 mm2 squares, rinsed with ethanol and Milli Q water, then N2 blown. Ultrasonic rinsing was not employed to avoid creation of glass particles that would scratch the ITO coating and create defects on the films. The electrolyte solution was 0.1 M Zn(NO3)2 prepared from high purity reactants (Aldrich) and Milli Q water (R > 18 MOhms). The pH of this electrolyte is around 5 and it was not adjusted. The bath temperature was varied and it was found in preliminary experiments that an optimal temperature is 65ºC. Air was continuously bubbled at a rate of about 60 bubbles per second through a Pasteur pipette. Different type of ZnO films were growth. Figure 2.27, from left to right, shows a ZnO film prepared by sweeping the potential from 0 mV to -1300 mV vs SSC at 1 mV-1, at room temperature; next sample was prepared at room temperature without air bubbling, the analysis showed it consisted mainly of Zn; latest sample was prepared at 65ºC with air bubbling and by pulsing the potential between OCP and Uelectrode= OCP-1300 mV at a pulse rate of 100 Hz. Pulse frequency as well as pulse duty cycle (symmetry factor of the pulse) were varied in order to modify film morphologies.
Figure 2.27. Typical ZnO/ITO films, details in the text.
C o p p e r O x i d e F i l m s | 135
CHAPTER 2. COPPER OXIDE FILMS
136 | C h a p t e r 3
3. Introduction This chapter describes the studies that conduce to the obtaining of semiconducting ptype Cu2O films on Cu electrodes by anodization in alkaline media. The chapter initiates with the description of the properties of the Cu-O system, including the expected behavior of copper in electrochemical conditions, and then emphasizes the interest in cuprous oxide (Cu2O) as promising material for technological applications. The usual preparation techniques of Cu2O are mentioned and the focus will be on the electrochemical methods: electrodeposition and anodization. Then the copper electrochemical behavior in alkaline media is studied by cyclic voltammetry to get information about the processes that occur in the Cu|electrolyte interface just to develop a suitable method to grow semiconducting sub micron Cu2O films. The film growth process is thoroughly described. The studied properties of Cu2O films so prepared are: structure, morphology and opto-electronic properties are studied by in-situ and ex-situ techniques,
such
as
electrochemical
impedance
spectroscopy,
photocurrent
spectroscopy, electrochemical tunneling microscopy/spectroscopy, optical reflectance, current sensing atomic force microscopy, and x-ray diffraction to mention the most relevant. To deep into the fundamentals of the Cu|Cu2O|electrolyte interface, an electronic diagram is constructed with the data from concurrent techniques, including the direct observation of the valence band edge by electrochemical tunneling spectroscopy. A fundamental study of the variation of the structure and optoelectronic properties of the Cu2O layers is also conducted by varying the time where the electrode is exposed to dissolution potential during the film growth. Another fundamental study on the variation of film properties is made by changing the cation employed in the alkaline electrolyte solution. These results are compared with a previously proposed model on electronic-structural disorder.
Specific objectives of this study •
Develop an electrochemical based method to prepare Cu2O layers with
semiconducting properties •
Deep into the mechanisms that govern the Cu2O film growth onto Cu substrates
•
Construct an electronic diagram of the interface Cu|Cu2O|Electrolyte
•
Study the relationship between electronic/structural disorder with surface and
bulk modification of the Cu2O as model system
C o p p e r O x i d e F i l m s | 137
3.1. The Cu-O and Cu-H2O systems. 3.1.1. General features Two chemical oxides of copper, Cu2O (red or yellow) and CuO (black) are known and are commonly found in nature in the mineral forms cuprite and tenorite respectively (figure 3.1) a)
b)
c)
d)
Figure 3.1. Photographs of high purity powders a) Cu2O and b) CuO; and oxide ores c) cuprite (Cu2O) and d) tenorite (CuO). Both oxides are formed if Cu metal is heated in air or O2 gas and also are found as corrosion products on air-exposed copper, CuO forming slowly on the top of the Cu2O first formed layer 1,2,3 . Although is not the scope of this study, it is worthy to mention that corrosion of copper and its alloys is still a subject of intense research 4,5 . Figure 3.2 depicts a calculated Cu-O phase diagram, where it can be observed the coexistence domains of Cu-Cu2O and Cu2O-CuO phases in all the ranges except at the stoichiometric compositions. Mixed valence oxides such as Cu4O3 have been reported but only few mentions for its preparation as single-phased high purity material 6,7 . 1
N.N. Greenwood and A. Earnshaw, Copper, Silver and Gold in Chemistry of the Elements, 1st Ed, Reprint (1986) Pergamon Press, Oxford, New York, Toronto, Sidney, Paris, Frankfurt 2 J.-W. Lim, J. Iijima, Y. Zhu, J. H. Yoo, G.-S. Choi, K. Mimura, M. Isshiki, Thin Solid Films 516 (2008) 4040–4046 3 S. Nakayama, A. Kimura, M. Shibata, S. Kuwabata, T. Osakai J. Electrochem. Soc., 148 (2001) B467B472 4 Search in Scifinder Database with subject “Copper corrosion” leads 994 references since 1998 (489 in since 2004, 59 in 2009) 5 F. Wiame, V. Maurice, P. Marcus, Surface Science 601 (2007) 1193–1204 and references therein. 6 P. E. D. Morgan, D. E. Partin, B. L. Chamberland, M. O’Keeffe J. Sol. State Chem. 121 (1996) 33–37 7 J.F. Pierson, A. Thobor-Keck, A. Billard Appl. Surf. Sci. 210 (2003) 359–367
138 | C h a p t e r 3
Figure 3.2. Calculated Cu–O phase diagram (reproduced from 8 )
3.1.2. Cu oxidation in electrochemical conditions Copper oxidation in aqueous media has been extensively studied, mainly from the corrosion point of view. The Pourbaix diagram shown in figure 3.3 9 predicts the existence of Cu, Cu2O and CuO as solid species. In this diagram was considered that the stable species in solution were Cu+, CuOH(aq), Cu(OH)2-, Cu2+, CuOH+, Cu(OH)2(aq), Cu(OH)3-, Cu(OH)42-, Cu2(OH)22+ and Cu3(OH)42+. The analysis of the diagram indicates that Cu would behave as a noble metal at all pHs in the absence of oxygen or oxidizing species. At low pH values Cu would dissolve as Cu+ and further oxidized to Cu2+. At mild alkaline conditions, Cu would oxidize to Cu2O for an appreciable potential range, then to CuO at further positive potentials. At highly alkaline media drastic Cu corrosion is expected, even having a pH range between 11 and 12.5 that, under positive potentials CuO appears as stable species. Cu(OH)2 has been found to be thermodynamically unstable in the cited report. The review of the thermodynamic data gives a first prospective into the range of stability of the formed films although the data refer to bulk like properties. 8 9
B. Hallstedt, L. J. Gauckler, Calphad 27 (2003) 177 B. Beverskog and I. Puigdomenech, J. Electrochem. Soc. 144 (1997) 3476
C o p p e r O x i d e F i l m s | 139
Figure 3.3. Pourbaix diagram for copper electrode at 25°C in aqueous solution
3.2. Cu2O 3.2.1. Importance, technological applications and growth of Cu2O films Copper (I) oxide is a p-type semiconductor material with a direct band gap of 2 eV that has prompted a renewed interest in applications such as resistive random access memories (ReRAM’s) 10 , anodes in lithium ion batteries 11 , Cu2O covered-carbon nanotube catalysts 12 , electrochromic devices for solar light modulators 13 , 3D-ordered macroporous materials for photoelectrochemical cells 14 and there is also interest on developing Cu2O-based transparent conductive oxides (TCO’s) 15 . As model system, Cu2O is possibly still the best suited bulk and recently thin film semiconductor to observe excitonic Bose-Einstein Condensation 16,17,18 . Very recently, low cost nanocrystalline Cu2O-based solar-cells had been described 19,20 . 10
R. Dong, D.S. Lee, W.F. Xiang, S.J. Oh, D.J. Seong, S.H. Heo, H.J. Choi, M.J. Kwon, S.N. Seo, M.B. Pyun, M. Hasan, H. Hwang, Appl. Phys. Lett. 90 (2007) 042107 11 L.J. Fu, J. Gao, T. Zhang, Q. Cao, L.C. Yang, Y.P. Wu, R.J. Holze, Power Sources 171 (2007) 904 12 Y. Yu, L.-L. Ma, W.-Y. Huang, J.-L. Li, P.-K.Wong, J.-C.J. Yu, Sol. State Chem. 178 (2005)1488 13 M. Ristova, R. Neskovska, V. Mirčeski, Sol. Energy Mater. Sol. Cells 91 (2007) 1361 14 X. Li, F. Tao, Y. Jiang, Z.J. Xu, Colloid Interface Sci. 308 (2007) 460 15 M. Nolan, S.D. Elliott, Phys. Chem. Chem. Phys. 8 (2006) 5350 16 J. Brandt, D. Fröhlich, C. Sandfort, M. Bayer, H. Stolz Phys. Status Solidi C 6 (2009) 556–559 17 C. Klingshirn, M. Jörger, T. Fleck, A. Jolk, Solid State Communications 134 (2005) 155–158 18 Y. Sun, K. Rivkin, J. Chen, J.B. Ketterson, P. Markworth, R.P. Chang, Phys. Rev. B 66 (2002) 245-315 19 Y.-H. Lee, I.-C. Leub, M.-T. Wu, J.-H. Yen, K.-Z. Fung, J. Alloy Comps.427 (2007) 213
140 | C h a p t e r 3
The fabrication of Cu2O by physical techniques is usually done by direct thermal oxidation of copper10 and by radiofrequency 21 or direct current reactive sputtering 22 , although other physical techniques had also been employed 23 . Crystalline, p-type films are usually obtained with the sputtering methods. The control of crystallinity, morphology, transmittance between other properties is done by substrate temperature, plasma power and reactive gas pressure. The reactive sputtering process has been also employed to obtain Co-doped 24 or Mn-doped 25 ferromagnetic Cu2O films as well as Aldoped Cu2O 26 for transparent conductive oxides. Chemical and electrochemical methods have been also employed to prepare Cu2O films. Some of the reported techniques include “cooking” of Cu plates in saturated CuSO4 27 , chemical bath deposition 28 , sol-gel 29 , galvanostatic 30,31 and potentiostatic 32,33,34 electrodeposition and anodization. Interestingly, chemical deposited Cu2O films usually show n-type behavior, explained in the literature because a copper excess in the films 35 . Preparation of Sr-doped 36 and Y 37 -doped Cu2O films has been reported by chemicalbased techniques. Concerning to electrodeposition of Cu2O, it is usually based on the electrochemical reduction onto conductive substrates of alkaline lactate or citrate Cu (II) complexes31 or in acetate baths in acid solution 38 . Uniform films have been reported both potentiostatically and galvanostatically and the deposition occur by a two-electron 20
B. D. Yuhas, P. Yang, J. Am. Chem. Soc., 2009, 131 (10), 3756-3761 S. Ishizuka, S. Kato, Y. Okamoto, T. Sakurai, K. Akimoto, N. Fujiwara, H. Kobayashi, Appl. Surf. Sci. 216 (2003) 94 22 A. Sivasankar Reddy, S. Uthanna, P. Sreedhara Reddy Appl. Surf. Sci. 253 (2007) 5287–5292 23 See for example references in M.R. Gennero de Chialvo, J.O. Zerbino, S.L. Marchiano, A.J. Arvia, J. Appl. Electrochem. 16 (1986) 517 24 J. Antony, Y. Qiang, M. Faheem, D. Meyer, D. McCready, M.H. Engelhard, H. Mark Appl. Phys. Lett. 90 (2007) 1-3. 25 L. Pan, H. Zhu, C. Fan, W. Wang, Y. Zhang, J. Q. Xiao J. Appl. Phys. 97 (2005) 10d318/1-3. 26 M. Zhang, G. Dong, W. Lan, P. Dong, H. Yan, Patent CN101158049 A (2008) 27 Q. Pan, M. Wang, J. Electrochem. Soc. 155 (2008) A452-A457 28 R. Neskovska, M. Ristova, J. Velevska, M. Ristov, Thin Solid Films 515 (2007) 4717–4721 29 Sekhar C. Ray, Solar Energy Mater. Solar Cells (2001) 307-312 30 A.E. Rakhshani, J. Varghese, Solar Energy Materials 15 (1987) 237-248 31 T. Mahalingam, J.S.P. Chitra, S. Rajendran, M. Jayachandran, M. J. Chockalingam, J. Cryst. Growth 216 (2000) 304-310 32 A.E. Rakhshani, J. Varghese, Thin Solid Films 157 (1988) 87-96 33 T.D. Golden, M.G. Shumsky, Y. Zhou, R.A. VanderWerf, R.A. Van Leeuwen, J:A. Switzer Chem. Mater. 1996, 8, 2499-2504 34 Y.L. Liu, Y.C. Liu, R. Mu, H. Yang, C.L. Shao, J.Y. Zhang, Y.M.Lu, D.Z. Shen X.W. Fan Semicond. Sci. Technol. 20 (2005) 44–49 35 C.A.N. Fernando, S.K. Wetthasinghe, Solar Ener. Mater. Solar Cells 63 (2000) 299-308 36 B. Roy, J.D. Perkins, T. Kaydanova, D.L. Young, M. Taylor, A. Miedaner, C. Curtis, H.-J. Kleebe, D.W. Readey, D.S. Ginley, Thin Solid Films 516 (2008) 4093–4101 37 N. Tsuboi, K. Tosaka, S. Kobayashi, K. Kato, F. Kaneko, Jpn. J. Appl. Phys. 47 (2008) 588–591 38 W.Shang, X. Shi, X. Zhang, C. Ma, C. Wang, Appl. Phys. A 87 (2007) 129-135 21
C o p p e r O x i d e F i l m s | 141
mechanism. The observed growth rates agree to those predicted from Faraday's law of electrolysis. There are a number of studies concerning the growth and properties of anodic Cu2O on Cu substrates from the fundamental point of view, mainly motivated by the need to characterize passive layers and its behavior in front of different electrolyte solutions. Studies performed in deaerated solutions showed that the passive film presents a duplex structure in which Cu2O, CuO or CuO/Cu(OH)2 are present. The structure of the Cu2O grown layer depends on the electrode potential, the solution pH, and the presence of different ions and dissolved gases in the electrolyte. Detailed studies on oxide formation from its initial to its later stages in the whole electrochemical potential range in alkaline solutions have produced an important knowledge on the structure of the layers as well as on the electronic properties of the interface. Contrarily, the electrochemical production of technologically suitable Cu2O layers by Cu anodization has been scarcely considered because the anodic Cu (I) oxide layers onto polycrystalline Cu have been reported to consist only of a few atomic layers 39 . Nowadays, there are interesting advantages in preparing Cu2O films by Cu anodization, such as the natural electronic back contact and the development of a field-induced doping that may improve the electronic behavior of the Cu2O layers. Thereafter, are of the main interests of this work is to produce property-tunable Cu2O films by an anodization-based method that can be used to produce film thicknesses in the order of hundreds of nanometers, that is, in the range of commercial devices. In the following sections the characteristics of Cu2O films are explained to give place to the developed method of film growth and thereafter to explain the experimental achievements of this work.
3.2.2. Crystalline structure Cuprous oxide crystallizes in the cuprite form as depicted in Figure 3.4.A). The crystal structure consists in two interpenetrating three-dimensional Cu2O networks (Figure 3.4.B) 40,41 . The two networks are held together by Cu-Cu internetwork interactions. It has been mentioned that disrupting this interaction by cation doping could be a mechanism to modify the Cu2O band gap15,41. In the Figure 3.4.A) each copper ion 39
J. Kunze, V. Maurice, L.H. Klein, H.-H. Strehblow, P. Marcus, J. Phys. Chem. B 105 (2001) 4263. V. E. Henrich and P. A. Cox, The surface science of metal oxides, Cambridge University Press (1994), Cambridge, UK 41 M. Nolan and S.D. Eliott, Chem. Mater. 20 (2008) 5522 40
142 | C h a p t e r 3
(solid) has two oxygen ions (light shading) as its nearest neighbors while each oxygen ion is surrounded by a tetrahedron of Cu atoms. The oxygen atoms form a body centered cubic (bcc) lattice, where the Cu atoms partially occupy the interstitial positions in alternating pattern. The lattice constant of the primitive cell is a = 4.27 Å and contains 2 O and 4 Cu atoms (Fig. 3.4.B). Two of its surfaces have been studied in single crystal samples: the non polar (111) and the polar (100). Figure 3.4.C) depicts the powder diffraction pattern of Cu2O 42 where the reflections are marked with the corresponding crystalline planes.
Figure 3.4.A) unit cell of Cu2O; B) crystal lattice of Cu2O; C) powder diffraction pattern Native Cu2O films onto polycrystalline Cu exposed to ambient air have been found to grow epitaxially with the (111) perpendicular axis parallel with to (111) substrate axis, but with two different azimuthally orientations: the (1-10) plane of Cu2O aligns parallel (aligned cuprite) or antiparallel (reversed cuprite) with the corresponding one in the Cu substrate 43 . Authors in ref.43 compare the native oxide response with anodically grown Cu2O; they found that the reversed cuprite is the most stable in the anodized case while in native oxide both forms are equally present. They attribute the difference to different formation energy of the phases, unbalanced upon potential application. Another important difference they observed was the epitaxial strain and relaxation of the lattice, 42 43
International Centre for Diffraction Data Y.S.Chu, I.K. Robinson, A.A. Gewirth J. Chem Phys 110 (1999) 5952
C o p p e r O x i d e F i l m s | 143
the native oxide begins with an expanded lattice and approaching the bulk values of the lattice spacing when the film thickness increases and, on the other hand the anodic oxide lattice is significantly compressed at the beginning tending to relax towards the bulk value. This behavior was attributed by the authors to the change of stoichiometry in the anodic oxide due to incorporation of lattice vacancies during the growth. The conclusion is that the differences arise from the different diffusion mechanism of the oxidizing species during the growth. This issue will be discussed later when comparing the crystallinity of Cu2O film prepared with potential application and without.
3.2.3. Electronic structure Cu2O is one of the oldest known and employed semiconducting materials. As a p-type semiconductor its conduction arises from the presence of holes in the Valence Band (VB)15,41,44,45,46,47,48,49 . Most oxides show poor mobility of VB holes because the O 2p states of the upper VB are localized. On the other hand, the states at the top of the VB in Cu2O are derived from fully occupied Cu 3d10 states sit above a filled band of O 2p states. These states are mixed in some extent and also a small contribution of Cu 4s states has been noticed. The lowest energy conduction band states are derived from a mixture of Cu 4s and Cu 3d, with Cu 4p states at higher energy. The on-site Cu 3d-4s hybridization is said to introduce significant 3d character into the conduction band so that even though Cu2O is nominally a 3d10 oxide the interband transitions have significant 3d-3d character. Theoretical studies had shown that the p-type behavior can be explained in terms of the presence of stable Cu vacancies (VCu+) with low energy of formation, together with the low probability of hole annihilation. The VCu+ are shown to mediate the Cu transport through Cu2O jumping between split vacancies and normal vacancies, with a barrier energy of 0.3 eV that is low enough to allow VCu+ diffusion at room temperature 50 . Several experimental and theoretical studies on its electronic structure have demonstrated the existence of electronic states above the top of the valence band (VB) that give rise to a so-called electronic sub-band (SB) 51 ,52 . According 44
A.M. Fernandez, J.A. Turner, Sol. Energy Mater. Sol. Cells 79 (2003) 391 A.E.J. Rakhshani Appl. Phys. 69 (1991) 2365 46 A.E. Rakhshani, Y. Makdisi, X. Mathew, Thin Solid Films 288 (1996) 69 47 W.Y. Yang, S.W. Rhee, Appl. Phys. Lett. 91 (2007) 232907 48 Z. Rosenstock, I. Feldman, Y. Gil, I.J. Riess, Electroceramics 14 (2005) 205 49 I. Díez-Pérez, F. Sanz, P. Gorostiza, Curr. Op. Sol. Stat. Mater. Sci. 10 (2006) 144 50 A.F. Wright and J.S. Nelson, Appl. Phys. Lett. 92 (2002) 5849 51 H.-H. Strehblow, V. Maurice, P. Marcus, Electrochemica Acta 46 (2001) 3775 52 V. Maurice, H.-H. Strehblow, P. Marcus, J. Electrochem. Soc. 146 (1999) 524 45
144 | C h a p t e r 3
to Rakhshani45, the SB is in fact composed by two different states: an acceptor state due to Cu vacancies (VCu) located 0.45-055 eV above the VB (theoretical studies show the maximum of this acceptor stated are located at the point of the first Brillouin zone15). An intraband state is a donor trap state reported to be composed by oxygen interstitials (Oi)30. The upper limit of the SB is at 0.84 eV above the VB top. Oxygen vacancy (VO) levels with a formation energy higher than that of VCu should act as “hole killers” and are reported to appear at 0.23 eV and 0.44 eV below the CB minimum (760 mV above the donor level top).
3.2.4. Electrochemistry of Copper in Alkaline media Following the Pourbaix diagram for the Cu-H2O system, it is in alkaline media where copper oxides are expected to be more stable. Thereafter, this study was performed in alkaline electrolyte solutions. Sodium hydroxide was chosen as the first electrolyte to be used in the copper electrochemical characterization and Cu2O film growth due the same ionic size of both ions sodium and copper 3.2.4.1. Cyclic voltammetry The electrochemistry of copper and the copper oxide development in alkaline media has been extensively studied23,39,43,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67 . To illustrate the different processes taking place at the electrode|electrolyte interface, Figure 3.5. shows a potentiodynamic curve of the Cu electrode in 0.1 M NaOH electrolyte solution, initiated from the cathodic to the anodic domain and then swept back. The main anodic peaks are associated first to the formation of Cu2O (UCu2O~-400 mV vs Silver/Silver Chloride electrode, henceforth SSC), secondly to the formation of copper (II) oxide (CuO) (UCuO~+50 mV vs SSC) and finally to the passivation of the electrode by a 53
J. Ambrose, R.G. Barradas, D. Shoesmith, J. Electroanal. Chem. 47 (1973) 47 J.M.M. Droog, C.A. Alderliesten, P.T. Alderliesten, G.A. Bootsma J. Electroanal. Chem. 111 (1980) 61 55 M.E. Martins, A.J. Arvia, J. Electroanal. Chem.165 (1984) 135 56 L.D. Burke, T.J. Ryan, J. Electrochem. Soc. 137 (1990) 1358. 57 Gary Hodes, Chemical Solution of Deposition Semiconductor Films Marcel Dekker Inc. USA 2002 58 S. Härtinger, B. Pettinger, K.J. Doblhofer, Electroanal. Chem. 397 (1995) 335 59 H.Y.H.Chan, C.G.Takoudis, M.J.J. Weaver, Phys. Chem B 103 (1999) 357 60 O. Matsuoka, S.S. Ono, H. Nozoye, S. Yamamoto, Surf. Sci 545 (2003) 8 61 V.Maurice, H.-H. Strehblow, P. Marcus, Surf. Sci.458 (2000) 185 62 M. Kang, A.A. Gewirth J. Phys. Chem B 106 (2002) 12211 63 W. Kautek, M. Geuß, M. Sahre, P. Zhao, S. Mirwald, Surf. Interf. Anal. 25 (1997) 548 64 J.Kunze, V.Maurice, L.H.Klein, H.-H. Strehblow, P.Marcus, J. Electroanal. Chem. 554-555 (2003) 113 65 J.-B. He, D.-Y. Lu, G.-P. Jin, Appl. Surf. Sci. 253 (2006) 689 66 V.D. Jovic, B.M. Jovic, J. Serb. Chem. Soc. 67 (2002) 531 67 K. Nakaoka, J. Ueyama, K. Ogura, J. Electrochem. Soc. 151 (2004) C661 54
C o p p e r O x i d e F i l m s | 145
duplex layer composed of Cu2O/CuO-Cu(OH)2 (U>500 mV vs SSC). In addition to these processes, at potentials negative with respect to Cu2O formation, the presence of low current redox peaks has been reported and its importance with respect to the structural evolution of Cu oxides has been also discussed in the literature53-59. In Figure 3.5. inset is enhanced this potential range. From the early works from Droog54, Arvia23,55 and that of Burke and Ryan56, the processes occurring at potentials previous to the onset of Cu2O formation have been explained considering that adsorbed oxygen species in a freshly prepared polycrystalline Cu electrode are formed by a mechanism that involves sequential transitions from lattice Cu atoms (Cul) to Cu ad atoms (Cu*) and to Cu ad ions (Cus+) that eventually lead to Cu+-hydrous species, Cu*(OH)ad 23. The Cul can also produce these hydrous species Cu(OH)ad namely (Cu2O2H)-aq can also be produced
23
. Copper (I) soluble species,
23
. Cu2O formation would arise from
dehydration of the Cu*(OH)ad /CulOHad and in a minor extent by heterogeneous nucleation of Cu2O onto the previously hydroxylated surface57. More recently, using in situ Electrochemical Scanning Tunneling Microscopy (ECSTM) Matsouka et al.60 revealed that a commensurate structure appears in a Cu (111) single crystal at Us=-950 mV, disappears at Us=-600 mV and appears again when the potential is swept back to Us=-950 mV. They suggested that the formed structure corresponds to the surface-bonded Cu (I)-hydrous oxide that, in the cathodic sweep results on the regeneration of active Cu atoms accordingly to the models of Gennero et al.23 and Burke-Ryan56. They also assigned a reversible wave observed at Us=-500/-530 mV to the organization-dissolution process of an ordered oxy-hydroxide ad-layer61. Maurice et al.51 who investigated the initial stages of Cu (111) oxidation in 0.1 M NaOH observed at about Us=-750 mV a quasi-reversible process assigned to the adsorption-desorption of a 0.19 monolayer (ML) of OH- together with a simultaneous Cu surface reconstruction39,51. The aforementioned works agree in assigning an important role to this hydroxide sub-monolayer with respect to the further oxide growth as OH-ad-layer lattice parameters precisely match with those of the oxygen planes in Cu2O (111) 39,51,62. The Cu2O growth starts from the reconstructed OH-populated surface39 as mentioned before, by a diffusion-dehydration mechanism23 or/and by heterogeneous nucleation from soluble Cu+ species57, as the adsorbed OH causes formation of supercritical surface nuclei and depending on the growth potential employed (Us=-450 mV and Us=400 mV) the thickness, the growth rate, the crystallinity and the nucleation sites will be affected39. At Us=-450 mV nucleation preferentially takes place in the steps and the
146 | C h a p t e r 3
oxide layer grows as 2D islands, while at Us=-400 mV there are not preferential sites for nucleation and 3D oxide islands are present from the initial stages of the process. Regarding the Cu2O growth mechanisms, Kang and Gewirth62 suggested that at pH 13, the Cu2O growth consumes two hydroxyl ions according to the equation 1: 2Cu0 + 2OH- Æ Cu2O + H2O + 2e-
(3.1)
By using in situ quartz microbalance measurements, Kautek et al.63 observed the net uptake of 2 mol of oxygen atoms to form Cu2O as in equation (1), suggesting that the Cu2O differences observed by Kunze et al.39 at different growth potentials they employed could be associated to the different rates of the two consecutive processes of hydroxyl uptake observed by Kang et al.62. An alternative way that can explain the Cu2O growth is from the dissolved Cu+ species that lead to a continuous nucleationgrowth mechanism similar to the very well studied chemical-bath deposition process57. Figure 2.5 shows a CV obtained from a Cu electrode in contact with a NaOH 0.15 M electrolyte solution, at a sweep potential rate of 10mV/s. The CV was obtained by first sweeping from negative to positive potentials. In the anodic sweep, the first electrodic peak appearing in the range of potentials between Us=-550 mV and Us=-250 mV has been associated with the Cu(0) to Cu(I) oxidation and the subsequent formation of Cu2O62,63,66. The two peaks observed in the potential range lying from Us=-250 mV to Us=+500 mV correspond to the further oxidation of Cu(I) to Cu(II), followed by the formation of the passive layer. Some authors report this final film to be a duplex layer of Cu2O and (CuO-Cu(OH)2)62,63,66. The cathodic peaks observed from Us=-500 mV to Us=-800 mV are respectively attributed to the Cu(II) to Cu(I) reduction and Cu(I) to Cu(0)62,63,66. At negative potentials prior to the first anodic peak, we used the label OHads in the CV to indicate a series of processes involving OH- adsorption39,51,61. As the aim of the present work is the formation and properties of Cu2O films, only the potential range Us=-1200 to Us=-200 mV, from hydrogen evolution reaction (HER) to Cu2O formation will be analyzed.
C o p p e r O x i d e F i l m s | 147
Figure 3.5. Cyclic voltammogram of a Cu electrode in 0.1 M NaOH. Potential sweep rate of 10 mVs-1. Inset, detail of the potential range negative with respect the onset of Cu0 to Cu(I) oxidation. The inset in Figure 3.5 shows a CV made in the potential range labeled OHads that corresponds to the processes attributed to hydroxyl adsorption. In the CV from Us=1250 to Us=-450 mV three anodic peaks at Us=-800 mV, Us=-650 mV and Us=-575 mV and the corresponding cathodic peaks at Us=-605 mV, Us=-950 mV and Us=-1025 mV, respectively labeled PA1, PA2, PA3, PC3, PC2 and PC1 can be observed. The anodic and cathodic peaks of the PA1-PC1 and PA2-PC2 pairs are separated by about 300 mV, while the peaks of the redox couple PA3-PC3 are observed at the same potential, a characteristic associated with an adsorption process. Reprising the aforementioned results of Matsuoka60 and Maurice61 and the model proposed by Burke56, the assignation of the redox pairs PA1-PC1 and PA2-PC2 respectively to an ad-ion formation and to the subsequent hydrous oxide development seems to be straightforward. The reversible PA3-PC3 redox pair can be ascribed to the proper adsorption-desorption process of hydroxyl ions51,61. The calculated charge transferred in this potential region, yields 63µC.cm-2 in fine agreement with the ca. 0.2 ML adsorption of hydroxyl ions as pointed out in the literature39,51,62.
148 | C h a p t e r 3
In Figure 3.6, the potential scan is reduced to show only the anodic peak A associated with Cu (0) to Cu (I) oxidation and Cu2O formation. The maximum of the anodic peak is centered on Us=-375 mV and its asymmetric shape suggests a non elementary electron transfer reaction. The corresponding cathodic peak C related with Cu(I) to Cu(0) reduction is centered at Us=-650 mV. The peak potential shift has been reported to be characteristic of a constant high-field growth mechanism by ion transport through the oxide39,64. By quartz microbalance measurements63 it has been determined that the reduction of the oxide is only complete when the potential is swept back to around Us=1100 mV. Finally at more positive potentials with respect to the peak A a current plateau (about Us=-260 mV) followed by the onset potential of the peak related with Cu(II) formation at about Us=-200 mV are also observed.
A
200
100
OHads 0
j (μA)
-100 200 VC fit
-200 150 -300
-400
-500
A1
A2
-400
-300
100 50
OHads
0 -600
C -500
-200
-600 -1200
-1000
-800
-600
-400
-200
Us vs SSC Figure 3.6. Cyclic voltammogram in the potential range below the Cu2O peak in NaOH 0.1 M. Potential sweep rate of 10 mV/s. Inset: detail of the A peak with Gaussian fit. The asymmetry observed in the A peak indeed reflects the consumption of two hydroxyl ions in a step mechanism where the rate limiting step would be an OH- diffusiondehydration process, accordingly with the suggestion of Kang62 and Kautek63. With this assumption peak A may be deconvoluted into Gaussian separated peaks as shown in the inset of the Figure 3.6, after baseline subtraction: between Us=-600 mV and Us=-500 mV the OH-adsorption processes were fitted by the peak labeled OHads. The main peak is fitted to two Gaussian peaks: a) a sharp peak labeled A1, that could be associated to
C o p p e r O x i d e F i l m s | 149
the reaction of the previously adsorbed surface hydroxyl to give rise to a CuOH-like phase and b) a wider peak labeled A2, that would correspond to the diffusion of hydroxyl ions to the electrode surface, and the eventual dehydration of the Cu hydroxide to lead the Cu2O phase62. It is noteworthy that the peaks overlap, suggesting that both processes take place at the simultaneously in this potential range. 3.2.4.2. Effect of the preferential orientation of the Cu subtrate surface in the peaks observed at potentials below the onset of Cu oxidation. In order to assess the effect of preferential orientation on the “prepeaks” two copper substrates were tested. The first one was an as-buyed polycrystalline Cu, the second was prepared by annealing a Cu substrate in Ar/H2 atmosphere at 200ºC for 1h, thereafter the oxide formed was removed by immersing the substrate in a 1 % H2SO4 in EtOH solution during 5 min and then polishing the substrate with the usual procedure. In the Figure 3.7 are shown the x-ray diffractograms of the annealed and the regular substrates. It can be readily appreciated that the annealing remarkably changes the substrate orientation, directing it preferentially from the (220) to the (200) direction. 35000 (200)
Cu substrate Cu H2-Ar annealed
Intensity (counts)
30000
25000 (200)
20000
(220)
15000 (111)
10000
5000
50.0
50.5
51.0
0 40
50
60
70
80
2 theta (deg)
Figure 3.7. X-ray diffractograms of the Cu substrate and after H2-Ar annealing In Figure 3.8 are presented the cyclic voltammograms both in the pre-peak and in the Cu2O region. It can be readily observed that the redox waves PA1-PA3/PC1-PC3 are present at the same potentials in both substrates. This experiment confirms that the redox processes previous to the onset of Cu2O formation do not depend on the crystalline orientation of the substrate.
150 | C h a p t e r 3 45
Non-oriented Oriented
-2
j (μA.cm )
30
PA3
PA2
15
PA1
0
-15
PC1
-30
PC2 PC3
-45
-1200
-1100
-1000
-900
-800
-700
-600
-500
Us vs SSC (mV) Figure 3.8. Cyclic voltammograms obtained at different potential ranges at potentials negative with respect to the Cu2O formation, in an untreated substrate (no preferential orientation) and in a (200)-oriented substrate. 3.2.4.3. Is there Cu+ dissolution during Cu anodization? In order to study the possible dissolution of Cu+ or Cu2+ at the present pH (12.85) series of experiments using a complexing agent were performed. NH4NO3 was added; in alkaline medium it would release NH3 and complex the dissolved Cu+ or Cu2+ by the following equilibriums: NH4NO3 + OH- ' NH3 + H2O
(3.2a)
Cu+/Cu2+(aq) + nNH3 ' [Cu(NH3)x(H2O)y]+/2+
(3.2b)
(where 1
