SELF-ORDERING TITANIA NANOTUBE ARRAYS: ELECTROCHEMICAL

ANODIZATION, FUNCTIONALIZATION, AND

APPLICATION

by

York Reed Smith

A dissertation submitted to the faculty of The University of Utah

in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Department of Metallurgical Engineering

The University of Utah

August 2014

Copyright © York Reed Smith 2014

All Rights Reserved

The U n i v e r s i t y o f Ut ah G r a d u a t e S c h o o l

STATEMENT OF DISSERTATION APPROVAL

The dissertation of York Reed Smith

has been approved by the following supervisory committee members:

Manoranjan Misra Chair May 29, 2014Date Approved

Michael L. Free Member May 29, 2014Date Approved

Swomitra K. Mohanty Member May 29, 2014Date Approved

Raj Rajamani Member May 29, 2014Date Approved

Michael Scarpulla MemberDate Approved

Henry White Member May 29, 2014Date Approved

and by ______________Manoranjan Misra______________ , Chair/Dean of

the Department/College/School of _____________ Metallurgical Engineering_____

and by David B. Kieda, Dean of The Graduate School.

ABSTRACT

Hydrogen is envisioned as a viable fu e l o f the future. Photoelectrochemical (PEC)

hydrogen generation by water splitting reaction is the most promising method to obtain

renewable hydrogen. The U.S. Department of Energy has determined that for PEC

hydrogen to be economically feasible, and competitive with steam reforming hydrogen

production, a solar-to-hydrogen efficiency of 10% maintained for 1 ,0 0 0 hours of

operation is required. Selection of durable photo-electrodes capable of withstanding the

harsh aqueous environment in PEC hydrogen generation is an important factor.

Semiconductor nanostructured metal oxides, such as titanium dioxide, are generally more

stable in such environments, making them suitable candidate materials.

In the present investigation, self-organizing nanotubular titanium dioxide

synthesized by electrochemical anodization and heterostructures thereof were examined

for PEC hydrogen generation. In the first part, new synthesis methods were explored such

as light-assisted anodization, surface treatment prior to anodization to achieve

hierarchical nanotubular titanium dioxide, and binary acid anodization for in situ metal

doping. A mechanism for pore nucleation and nanotube wall separation has also been

proposed. In the second part, titania nanotubes were sensitized with nanocrystalline CdO,

CdS, and Mn2+ or Co2+ doped CdS as visible light absorber layers. The material

properties were examined using different characterization techniques such as scanning

electron microscopy (SEM), x-ray diffraction (XRD), ultra violet-visible (UV-vis)

photospectroscopy, x-ray photon spectroscopy (XPS), and Raman spectroscopy. The PEC

activity of the photoanodes was examined under simulated air mass (AM) 1.5 irradiation.

Electrochemical impedance spectroscopy and Mott-Schotty analysis were also used to

ascertain the PEC results and correlate with material properties.

iv

To Mrs. G, for helping me realize the pleasure of finding things out.

TABLE OF CONTENTS

ABSTRACT.................................................................................................................................iii

LIST OF TABLES................................................................................................................... viii

ACKNOWLEDGEMENTS...................................................................................................... ix

CHAPTERS

1: INTRODUCTION AND BACKGROUD.......................................................................... 1

1.1 Solar energy conversion and storage.....................................................................................21.2 Materials and required properties for solar energy conversion.............................................31.3 Semiconductor photoelectrochemistry..................................................................................5

1.3.1 Thermodynamics of water splitting and redox reactions...............................................81.4 Efficiency benchmarks and scale feasibility.......................................................................11

1.4.1 Heterostructures...........................................................................................................121.5 Objectives and scope...........................................................................................................13

2: LITERATURE SURVEY................................................................................................... 15

2.1 Surface and bulk properties of titania..................................................................................162.1.1 Nanostructured 1D titania............................................................................................18

2.2 Electrochemical anodization...............................................................................................202.2.1 Oxide formation stages during anodization.................................................................252.2.2 Formation of complex nanotube geometries................................................................27

2.3 Photoelectrochemical water splitting..................................................................................292.3.1 Material aspects of PEC hydrogen generation.............................................................29

2.4 Titania nanotube array heterostructures with metal............................................................29chalcogenides ............................................................................................................................29

2.4.1 Doped CdS nanocrystals..............................................................................................31

3: MATERIALS AND METHODS....................................................................................... 33

3.1 Anodic synthesis of titania nanotubular arrays...................................................................343.1.1 Photoelectrochemical anodization...............................................................................343.1.2 Synthesis of hierarchical nanotubular structures on foil..............................................36and wire substrate .................................................................................................................363.1.3 Binary acid anodization...............................................................................................363.1.4 Synthesis of titania nanotube array heterostructures....................................................37

3.2 Postsynthesis treatments......................................................................................................383.2.2 Deposition of titania nanoparticles, CdO nanocrystals,...............................................38

and doped CdS nanocrystlas.................................................................................................383.3 Characterization techniques................................................................................................39

3.3.1 Surface morphology and composition (SEM/EDS and XPS)......................................393.3.2 Crystallinity and optical properties (XRD, UV-vis, Raman spectroscopy).................403.3.3 Photo/Electrochemical characterization.......................................................................40

4: RESULTS AND DISCUSSION........................................................................................ 45

4.1 Synthesis..............................................................................................................................464.1.1 Nanotube growth mechanism......................................................................................464.1.2 Light assisted anodization characterization and formation..........................................53mechanism............................................................................................................................534.1.3 Hierarchical nanotbe characterization and formation..................................................65mechanism............................................................................................................................65

4.2 Functionalization.................................................................................................................684.2.1 Binary acid anodization...............................................................................................684.2.2 Galvanic deposition of CdO.........................................................................................714.2.3 Simultaneous ion layer adsorption-reaction (SILAR).................................................73deposition of CdS and doped-CdS nanocrystals...................................................................73

4.3 Photoelectrochemical application and evaluation...............................................................844.3.1 Light assisted anodized T-NTA...................................................................................844.3.2 Hierarchical T-NTA.....................................................................................................874.3.3 Binary acid anodized T-NTA.......................................................................................904.3.4 T-NTA heterostructures...............................................................................................93

5: SUMMARY........................................................................................................................102

REFERENCES........................................................................................................................108

vii

LIST OF TABLES

Table Page

1.1: Comparison of chemical energy stored in different fuels............................................... 9

3.1: Sample identification and corresponding synthesis description section.....................35

4.1: Summary of nanotube dimensions obtained from SEM analysis for light-assistedanodized samples...............................................................................................................58

4.2: Summary of binding energy of peaks (eV) obtained from XPS analysis...................77

4.3: Summary of lattice constant values obtained from the XRD profiles........................ 81

4.4: Equivalent circuit parameter values obtained through nonlinear fitting of theexperimental impedance data...........................................................................................96

ACKNOW LEDGEM ENTS

I would first like to acknowledge the institutions that have funded my research

endeavors. The U.S. Department of Energy and the Utah Science and Technology

Research Initiative (USTAR) supported much of the work presented here. Financial

support through a Graduate Research Fellowship, University of Utah, for 2013-2014 is

also greatly appreciated.

A sincere thanks is extended to my advisor, Dr. Mano Misra. I have learned a

great deal and have thoroughly enjoyed our discussions, even when we talked science.

Also a great deal of thanks goes to Dr. Swomitra Mohanty. I would also like to thank my

other committee members, Drs. Michael Free, Raj Rajamani, Mike Scarpulla, and Henry

White, for their service and valuable input.

I would also like to thank some colleagues and coauthors who have contributed to

this work: Drs. Biplab Sarma and Krista Carlson, University of Utah, and Ruchi Gakhar,

Augstus Merwin, and Dr. Dev Chidambaram, University of Nevada, Reno. I would also

like to thank members of the Metallurgical Engineering Department faculty, staff, and

fellow colleagues. A special extended thanks goes to Diane Hanes in helping with

editing. It’s the sum of all parts that makes things work.

Most of all, I have to thank my family for all the love and support. Thank you

Mom and Dad for “following the child.”

CHAPTER 1

INTRODUCTION AND BACKGROUD

1.1 Solar energy conversion and storage

Arguably the biggest challenge faced by the scientific and engineering community

for the 2 1 st century is the development and implementation of clean, renewable, and

inexpensive energy conversion systems. Increasing emissions levels from the combustion

of fossil fuels in stationary and mobile energy conversion systems have raised many

environmental and health concerns in recent years. Succinctly, anthropogenic emissions

into the atmosphere have been speculated to play a role in global climate change.1, 2 Solar

energy is an attractive “clean and green” energy option to add to our overall

renewable/sustainable energy portfolio to help offset the use of conventional fuels, reduce

harmful emissions, and meet necessary future energy goals.3 Solar energy, although

intermittent, can be viewed as a pseudo-infinite source of energy. Nearly enough sunlight

irradiates the Earth in one hour (4.3 x 1020 J)4 as all the energy consumed on the planet in

one year (5.5 x 1020 J in 2010) .5 The full utility of such an abundant energy resource is

yet to be realized.

Solar irradiation is currently utilized in two general methods: (i) solar thermal,

which harnesses solar energy for thermal driven systems, generally utilizing infrared-

photons, or (ii) photovoltaics, which generates electrical power by use of photovoltaic

materials, utilizing mainly visible/near-IR photons. Another avenue, rather than

converting solar energy into thermal energy or electrical power, is to store this energy in

chemical bonds through the synthesis of fuels. This process is similar to the principles of

photosynthesis in plants. A simple demonstration of this concept is semiconductor

photocatalytic or photoelectrochemical water splitting to hydrogen (H2) and oxygen (O2).

Over the years there have been numerous studies on semiconductor photocatalytic and

2

photoelectrochemical systems for hydrogen generation (water splitting reaction). The

underlying operating principles have also been extend to many other areas such as

electrochemical photovoltaics,6-8 wastewater remediation,9, 10 antibacterial/self-cleaning

surfaces,11, 12 and fuel synthesis, 13 among others.

Although H2 is envisaged to be a viable fuel of the future, hydrogen is not

available in its free form naturally, and occurs mainly in the from of water (H2O), natural

gas (CH4), or other organic compounds.14 Subsequently, the successful use of hydrogen

as a viable fuel option depends highly on production technology and cost.

1.2 M aterials and required properties for solar energy conversion

Ideal materials for converting solar irradiation are semiconductors. This is

because of their discrete quantum states of electrons. Unlike their metal counterparts that

have a continuum of electron states (bands), semiconductors exhibit an electrical

resistivity, or rather an energy band gap (Eg) that extends from the top of the filled

valance energy band (VB) to the bottom of the vacant conduction energy band (CB)

(Figure 1.1).

The conversion of radiant energy to electrical or chemical energy (i.e., surface

redox reactions) is primarily based upon the intramolecular “pumping” of electrons to

higher energy levels (i.e., molecular excitation) upon the absorption of light in the visible

or ultra violet regions. Intermolecular electron transfer from these excited molecules,

either directly or via an external circuit in an electrochemical cell, provides the basis of

the energy conversion process. Generally, when a semiconductor surface is exposed to

light radiation (hv > E g) electron hole pairs (hVB+ - eCB") are generated. This event is

3

4

Figure 1.1: Schematic diagram of molecular orbitals forming bulk band structures.

represented by equation 1. Heat generation for the reverse reaction is shown by equation

2 (recombination). Efficient utilization of the radiant energy requires separation of the

photogenerated electron-hole pairs before recombination. These events have been found

to occur quite effectively within the electric field formed at the interface between a

semiconductor and an electrolyte solution (i.e., band bending).

photocatalyst h v> E g 1 1 + - \~ * \h VB V e CB j (1)

1.3 Sem iconductor photoelectrochem istry

The photovoltaic effect is a phenomena in which a voltage or electrical current is

generated in a material by being exposed to electromagnetic radiation (i.e., light energy).

This is the basis of semiconductor photocatalysis and photoelectrochemistry. The French

scientist Alexandre-Edmond Becquerel first reported witnessing this effect in 1839.15

Several advances between the 1950s and 1970s improved the understanding of

electrochemical interactions between semiconductor-liquid interfaces, namely by

Gerischer, Memming, and Williams, among others.16-18 These studies resulted in

establishing a fundamental understanding of the semiconductor-electrolyte interface,

including kinetics and energetics of charge transfer across the semiconductor-electrolyte

junction. Despite these early fundamental studies, it was not until the 1970’s works’ by

Fujishima and Honda that the potential application of photoelectrochemical systems for

energy conversion and storage were first demonstrated.19, 20 Their initial studies

demonstrated that the oxidation of water could be carried out at a less negative potential

compared to the standard redox potential when a titania surface is irradiated with light

energy greater than the material’s band gap. These results birthed the realization that

sunlight could be used to split water into hydrogen and oxygen, a process commonly

referred to as photoelectrolysis.21

Figure 1.2A shows a schematic of a liquid-junction photoelectrochemical cell.

The photoelectrochemical cell consists of a photoanode, a counter electrode cathode, and

a reference electrode. In a typical PEC setup, semiconductors utilize light irradiation to

promote reactions on the surface of the electrodes within the system. This example

considers n-type semiconductors, as they have generally shown better stability. The

5

6

Figure 1.2: Schematic representations of (A) a liquid junction, three-electrode photoelectrochemical cell for a p-type semiconductor where D/D+ represents oxidizes species by donating electrons to be scavenged by surface holes (e.g., 2OH- -> ^ O 2 + 2e- + H2O) and A/A- represents reduced species accepting electrons (e.g., 2H2O + 2e- -> 2OH- + H2). (B) A photocatalytic system where the energy bands, VB and CB, act as the anode and cathode, respectively, in a short-circuited photoelectrochemical cell, and (C) a depiction of solid volume and surface photoexcitation recombination events.

principles and applications of p-type materials are generally similar to n-type materials.

Oxidation of the electrolyte (e.g., H2O) occurs at the anode and reduction occurs at the

cathode. For photocatalytic systems (Figure 1.2B), Bard’s concept22 can be applied. This

concept suggests that semiconductor particles can act as a short-circuited PEC cell by

providing both oxidizing and reducing sites for the reactions (valance band (VB) and

conduction band (CB), respectively). When comparing the two systems, the

photocatalytic system is simpler and easy to construct but has its limitations on practical

applications.

Upon photo-excitation of an electron from the VB to the CB, the e~CB and h+VB

pairs can follow several pathways (Figure 1.2C). Ideally, in a PEC cell the photo

generated e"CB’s are driven to the external circuit and subsequently to the cathode with a

small bias potential. The h+VB’s can then migrate to the surface of the semiconductor to

oxidize any absorbed molecules or solvent on the surface. The ability for a photocatalyst

to carry out redox reactions and facilitate effective and efficient charge transfer is

predisposed to two criteria: (i) effective charge separation and (ii) the catalysts’ ability to

absorb reacting species on its surface. At the surface of the semiconductor, electrons can

also be donated to an acceptor species (A/A-). Likewise, the holes can migrate to the

surface to combine with electrons from a donor species (D/D+) (Figure 1.2C, path a and

b). However, the recombination of electrons and holes prevents them from transferring to

the surface to react with absorbed species. Recombination can occur within the volume of

the semiconductor, or on the surface of the semiconductor (Figure 1.2C, path c and d).

The rate at which charge transfer occurs depends upon the position of the bands and the

7

8

redox potential of the absorbed species of interest. Some important factors that should be

met for optimal performance of a photocatalytic include:

• The redox potential of the photogenerated VB holes should be sufficiently

positive for the holes to act as an acceptor;

• The redox potential of the photogenerated CB electron should be sufficiently

negative for the electrons to act as a donor;

• Photocatayst should be economically available and be environmentally inert;

• Photocatalyst should be stable in a wide pH range and in a variety of electrolytes.

1.3.1 Thermodynamics of water splitting and redox reactions

The overall water splitting reaction is endothermic and can be written as:

H2O(l) ^ >/2O2(g) + H2(g) (3)

with the half reactions:

2H+(aq) + 2e- ^ H2(g)

H2O(l) ^ O 2(g) + 4H+(aq)

(4)

(5)

Acid/base balanced:

H2O(l) + 2e" ^ H 2(g) + 2OH"(aq) (6 )

4OH"(aq) ^ O2(g) +2 H2O(l) + 4e- (7)

9

The change in Gibbs free energy for reaction 3 is: AG0 = 56.69 kcal/mol = 237

kJ/mol. The standard cell potential is calculated as

E0 = -AG0/nF (8)

where n = 2, the number of electrons taking part in the reaction, and F is Faraday’s

constant (96,500 C). The calculated theoretical potential required for reaction 3 is 1.23 V.

The potential required for other various fuels is given in Table 1.1.

Narrow band gap semiconductor sulfides (i.e., CdS, CdSe, CdTe) are often

studied for solar hydrogen generation as their valance bands are at relatively negative

potentials and offer good visible light driven photochemical reactions. The benefits of

utilizing combinations of materials or heterostructures are discussed in more detail in the

following subsection (Section 1.4.1).

In these systems, sacrificial electrolytes, such as a mixture of Na2S and Na2SO3,

are used. The Na2S in solution acts as a hole scavenger, and is oxidized into S22- to

Table 1.1: Comparison of chemical energy stored in different fuels.23

Fuel AG0(kcal/mol)

ETheoretical(V)

Energy Density (kWh/kg)

Hydrogen (H2) -56.69 1.23 32.67Methanol (CH3OH) -166.80 1.21 6.13Ammonia (NH3) -80.80 1.17 5.52Hydrazine (N2H4) -143.90 1.56 5.22Formaldehyde (HCHO) -124.70 1.36 4.82Carbon monoxide (CO) -61.60 1.33 2.04Formic acid (HCOOH) -6 8 .2 0 1.48 1.71Methane (CH4) -195.50 1.06 -Propane (C3H8) -503.20 1.08 -

10

prevent corrosion (e.g., CdX(h+) -> Cd2++ X"*, X = Se or Te). The visible light assisted

photoexcitation of a TiO2-CdS photoanode, for example, with subsequent charge

transport mechanism can be summarized by the following reactions:

where equation 9 is the electron transfer from CdS to TiO2, and equation 10 is the hole

transfer from CdS to the redox couple (Red) resulting in oxidized products (Ox). The

electrons transferred to titania via equation 9 are collected and result in the generation of

anodic current. Effective scavenging of holes accumulated in the valance band of CdS is

an essential part to maintain stability of the CdS absorber layer. If hole accumulation

becomes too great, the CdS will begin to oxidize and over time results in diminishing

photoelectrochemical performance. Hole scavenging can also occur by equations 11, 12,

and 14. To ensure hydrogen production at the cathode, Na2SO3 is added to help reduce

disulfides back to sulfides. This has been demonstrated to be beneficial to improve

14 24 25 2photocurrent density and hydrogen production in previous studies. ’ ’ Also, the SO3 '

ions mainly yield thiosulfate ions.26

CdS(e-) + TiO2 ^ CdS + TiO2(e-) (9)

CdS(h+) + Red ^ CdS + Ox ( 10)

2S2- + 2h+ ^ S22- ( 11)

SO32- + H2O + 2h+ ^ SO42 + 2H+ ( 12)

2S22- + SO32- ^ S2O32- + S2- (13)

SO32- + S2- + 2 h+ ^ S2O32- (14)

S2O32- + H+ ^ HSO3" + S

S + 2e" ^ S2-

(15)

(16)

11

1.4 Efficiency benchm arks and scale feasibility

One of the goals of the US Department of Energy’s Hydrogen Program is the

large-scale production of hydrogen utilizing a renewable energy source to split water. In

order to meet DOE goals, a PEC system must be low cost, operate at a solar-to-chemical

conversion efficiency of greater than 10%, and have a long operating lifetime (>1 ,000 h).

For economic feasibility, the PEC system should produce hydrogen at a competitive cost

compared with steam reforming of natural gas ($2-3/kg H2) .27 As a result, DOE has

recently established a production cost target for PEC hydrogen to be $5.7/kg H2 by

2020.28 In order to be cost competitive with steam reforming, however, a solar-to-

hydrogen (STH, ^sth) conversion efficiency of 10-25% for 1,000 h of operation is

required.

When considering approaches to produce hydrogen from water splitting, one

method is photovoltaic powered electrolysis. Using a single crystalline silicone PV

module coupled with a high pressure electrolyzer, a STH efficiency of 9% can be

realized,29 which correlates to a hydrogen production cost of ~$10/kg H2. A major

drawback in using such a ‘brute force’ approach is that the voltage output of traditional

pn-junction photovoltaic cells is limited to 0.5-0.6 V and strongly dependant on the

illumination intensity. Subsequently, the optimal number of cells required to generate

sufficient potential (1.8~1.9 V (+0.6-0.7 V)) required for overpotentials and kinetic

losses) is highly variable due to variations in light intensity. This disadvantage can result

in as much as a 50% energy loss, and challenges the viability of the PV powered

electrolysis approach.30

In a recent study, the net primary energy balance of a solar-driven PEC water-

splitting device was studied by Zhai et a l.31 Their net energy balance was conducted

using a fundamental unit of primary energy (MJ) needed to produce 1 kg of hydrogen.

Uncertainty analysis of their model shows the primary energy required to produce 1 kg of

PEC hydrogen is highly sensitive to STH and longevity. An example of their analysis, for

a STH efficiency of 3%, the longevity must be at least eight years to yield a positive net

energy balance. Jaramillo et al?2 performed a technoeconomic feasibility study on the

centralized production of solar hydrogen by photocatalysis and photoelectrochemical

routes. Taking into account all capital expenditures and operating costs to generate 10

tonne per day delivered at 300 psi H2, a final production cost of $1.6-10.4/kg H2 was

realized. The key finding of this study is that solar hydrogen can be produced at targeted

US DOE goals of $2-4/kg H2 and can be a viable route for renewable hydrogen

production if material performance targets can be met.

1.4.1 Heterostructures

The inherent large band gap of titania limits its utility to only 5-10% of the solar

spectrum (UV light). To improve the visible light absorbance (~50% of solar spectrum)

and reduce the band gap of titania, methods such as doping have been carried out

extensively. Despite such efforts, the material itself has limiting factors. The utility of the

full solar spectrum, to improve efficiency, cannot be realized with titania alone. To

counter this problem, heterostructures can be engineered. These heterostructures combine

12

titania nanotubes with visible light/near-infrared active semiconductor layers. The

advantage of engineering such materials is that the properties of each material can be

synergistically coupled to enhance the overall catalytic performance, while still

maintaining their own respective properties. A potential route to enhance solar-to-

electrical/chemical energy conversion efficiency is through designing and synthesizing

semiconductor nanocrystals (NC) or quantum dots (QD) as an absorber layer in third-

generation cell systems to utilize their size or compositional dependant properties.

Weber and Dignam33 demonstrated that a tandem configuration approach gave

higher STH efficiency than when two cells of the same band gap energy (Eg) were

juxtaposed to one another. Under conditions where E g1 = Eg2 = 1.4 eV, a n STH value of

16% was observed, whereas in the tandem approach where E g1 = 1.8 eV and E g2 = 1.15

eV, a nSTH value of 22% was predicted. Prevot and Sivula30 adopted a more extensive

analysis by Bolton et a/.,34, 35 which accounts for realistic energy losses and uses ASTM

G173-03 Reference Spectra. With the large assumed losses (2 eV), and E g1 = 1.89 eV and

E g2 = 1.34 eV, a nSTH value of 21.6% is predicted. In a recent work by Lewis et a / . ,36 they

accounted for the thermodynamic Shockley-Queisser limit of each absorber layer and

predicted a nSTH value of ~30%. Although the conditions and losses specified in the

models can vary, a STH efficiency that can possibly reach 20% and upwards to 30% in a

tandem cell configuration is a promising outlook.

1.5 O bjectives and scope

The general aim of this research was to develop new synthesis methods to obtain

various forms of titania nanotubular arrays and composite/heterostructure materials with

13

improved photoelectrochemical performance. The following four points can summarize

the objectives and scope of this work.

1. Develop a mechanism for nanotube wall separation and pore nucleation based on

the findings of these studies and those within the literature.

2. Examine morphological changes to titania nanotubes that affect

photoelectrochemical performance

a. The effect of light assisted anodiation on the morphology, composition,

and photoelectrochemical performance.37

b. Develop a new method to synthesize hierarchical titania nanotube arrays

on foil and wire substrate.38

3. Physicochemical changes to titania nanotubular arrays for improved

photoelectrochemical activity

a. Develop a method to synthesize TiO2-WO3 nanotubes by anodization in

binary acid electrolyte.39

4. Heterostructures with titania nanotubular arrays

a. Examine the effect of electrochemically deposited of CdO on titania

nanotubular array photoresponces.40

b. Characterize and examine the effect of doped CdS functionalized titania

nanotubular array photoresponses.41

14

CHAPTER 2

LITERATURE SURVEY

2.1 Surface and bulk properties o f titania

Although TiO2 naturally occurs in several different polymorphic forms, anatase

and rutile are the most commonly studied phases for nanomaterials used in solar

conversion applications.42 The basic structure of both anatase and rutile contains titanium

in six-fold coordination with oxygen.42, 43 Differences between the two phases arise from

the connection of the octahedrally coordinated Ti cations; anatase octahedra share four

edges and are connected in staggered chains parallel to [2 2 1 ], while the octahedra in

rutile form chains along [001] with only two shared edges.43 Polymorph stability is

dependent upon the synthesis method adopted and crystallite size.42-45 In bulk form, rutile

is the most thermodynamically stable phase under ambient conditions. However, as

crystallite sizes drop below 13 nm, anatase becomes the more stable phase.43, 46 This

phase change suggests that many of the ordinary physics and chemistry rules of bulk

materials no longer apply at the nano-level, resulting in their properties to differ

substantially.45

The relationship between the bulk and surface chemistry of a material becomes

apparent when examining the surface charge that develops when it is placed in an

aqueous solution. When immersed in an aqueous solution, the charge that develops on an

oxide surface is mainly dependent upon the electronegativity of the cations in the

material and the solution pH .47 However, the charge developed on the surface has been

found to vary depending on the polymorph, crystallographic planes, crystallite size, as

well as the synthesis and measurement method employed. For example, Mandzy et al.4

reported the isoelectric point (IEP) of different sized nanoparticles of antase and rutile: 5

nm anatase - pH 4, 25 nm anatase - pH 5.5 and 10 nm rutile - pH 3.2. As commercially

16

available materials, the synthesis method was not listed; however, the 5 nm anatase value

differed from the pH ~5.5 reported by Penn et al.43 where 5 nm particles were

synthesized via a sol-gel method. This discrepancy is most likely the result of the

technique used to obtain the IEP value. In Mandzy’s study, the IEP was measured using a

zetameter, while Penn relied on the gelling of suspensions. Difference in crystallite size

could account for the discrepancy between Mandzy’s 10 nm rutile particles exhibiting an

IEP of pH 3.2 and 2 nm diameter particles formed by Bullard et al.,49 which had an IEP

of pH 5.2. Bullard also demonstrated the variation between crystallographic planes with

IEP ranges for the (100), (110) and (001) surfaces of rutile as 3.2 - 3.7, 4.8 - 5.5, and 5.5

- 5.8, respectively.

Differences between polymorphs and crystal surfaces can be explained through

surface chemistry. Bullard theorized that differences between Ti and O coordination

caused the variation in IEP. It was suggested that higher IEP values for the (001) surface

were due to stronger and more numerous Lewis base sites. The increased quantity and

strength of these electron-donating oxygen positions facilitated more adsorption of

solvated hydrogen ions. The (001) surface is also more open than the (100) and (110)

surface, promoting high charge mobility, and hence would be the preferred orientation in

electrochemical experimentation.

The relationship between surface acidity and nanotube structure was examined

extensively by Kitano et al.50 In this study, Fourier transform infrared (FT-IR)

spectroscopy was used to examine TiO2 nanosheets and nanotubes with similar crystal

structures. Nanotubes and nanosheets were found to possess both Bronsted and Lewis

acid sites; however, nanotubes were found to exhibit higher Bronsted acidity than the

17

nanosheets. It was suggested that the higher concentration of bridging OH vs terminal

OH groups found on nanosheets, and the distortion of the Ti octahedra, allowed for

nanotubes to have higher catalytic activity than the nanosheets. Sill, few surface acidity

studies have been performed on titania nanotubes formed via anodization, leaving this

area open for further investigation.

2.1.1 Nanostructured 1D titania

Research focused on the synthesis, characterization, and applications of

nanomaterials has recently become a common ground between many scientist and

engineers. Due to the small size of these materials (x10-9), they often exhibit exciting

electronic, optical, and mechanical properties due to small geometry (e.g., quantum

confinement effects). The late Nobel laureate physicists, Richard P. Feynman, first

envisioned this area of research in his talk “There’s Plenty of Room at the Bottom” at the

American Physical Society meeting at Caltech in 1959;51 however, it was not unit 1974

that this field was coined with the name nano-technology by Norio Taniguchi.52

Nanomaterials are considered materials that have features less than 100 nm in at

least one dimension. New and interesting physical and chemical properties emerge when

the size of a material decreases down to the nanometer domain. As the size decreases, the

surface area and number of bonds on the surface increases. In semiconductor materials

within this size domain, the movement of electrons and holes is governed by the well-

known quantum confinement effect, and the transport properties related to phonons and

photons are largely affected by the size and geometry of the material. As a result,

quantum confinement has an affect on the optical, electrical, and chemical functionalities.

18

Quantum confinement effect in nanomaterials is a result of a dispersed

distribution of allowable energy states. That is, as the size of the particle decreases in

dimension (from bulk to nano-sized), the density of states (DOS) becomes discreet. A

review article by Chen and Mao53 covers numerous synthesis methods to achieve nano

sized titanium dioxide, while an extensive review by Diebold42 covers the bulk properties

and surface science of titanium dioxide. Although there are several methods to

synthetically obtain 1D titania nanostructures (i.e., nanotubes, nanorods), generally three

methods are adopted: i) hydrothermal synthesis,54 (ii) template directed growth using

porous alumina,55 and (iii) electrochemical anodization.56 Out of these three methods,

electrochemical anodization offers several advantages over the former two. For example,

electrochemical synthesis is a scalable process, offers easier reagent handling, and

demonstrates robust control on the nanoscale. In the former two synthesis methods, the

final material is in powder form, which can limit its applications.

Electrochemical synthesis methods have shown promise to synthesize a number

of metal oxide nanostructures,57, 58 in particular electrochemical anodization of so-called

valve metals and their alloys.59 Through the use of various processing techniques, a wide

variety of nanostrucutres can be synthesized with high control. The synthesis of vertically

orientated self-ordered metal oxide nanotube arrays on metal substrates, using

electrochemical anodization technique, has been reviewed,59-61 as well as their application

56, 62, 63in solar energy conversion systems.

19

2.2 E lectrochem ical anodization

Keller et aL64 first reported the formation of porous anodic alumnia using an

electrochemical anodization technique. This synthesis method has been extended to

several other valve metals and their alloys, to form highly self-ordered nanostructures.

Despite previous efforts, it was not until the recent works by Hebert et a / 65’ 66 that a

quantitative relationship of oxide dissolution and nanoporous film formation on Al and Ti

had been established. These works focused on aluminum (Al) and titanium (Ti), but can

potentially be applied to other metals. The basis of nanoporous film formation via

anodization involves a combination of ionic migration in the formed oxide and stress-

driven interface diffusion of metal atoms. A differentiating feature between Al and Ti is

that when Al is anodized, it generally forms a porous oxide layer, whereas anodization of

Ti generally forms a nanotubular oxide layer67 where separate individual tubes are

formed.

Raja et a / 6S suggested that faster generation of cation vacancies by accelerated

dissolution and radial transport of vacancies across the pore walls could be the cause of

tubular formation when anodizing Ti in fluoride containing electrolytes. Valota et a / .69

attributes the separation of nanotubes formed on titania as a result of dissolution of a

fluoride rich layer formed at the cell boundaries of the nanotubes; however, this does not

explain the separation of nanotubes in fluoride free electrolytes.70, 71 More recently, Su et

a / .67 proposed a new model based on localized dielectric breakdown during the

anodization process. Their model suggests that oxygen filled voids can be generated at

the barrier oxide layer of anodic oxides due to localized dielectric breakdown of the

oxide. The number and size of such voids increase with the degree of localized dielectric

20

breakdown and the accumulation of these voids at the cell boundary areas causes the

separation of neighboring pores.67 There is still no generally accepted explanation for

metal oxide nanotubular formation via electrochemical anodization process. A more

detailed discussion and new mechanism on tube wall separation is presented in Section

4.1.1.

Zwilling and co-workers first demonstrated the growth of anodic oxide nanotubes

on titanium using fluoride-containing electrolytes in 1 9 9 972, 73 although they were

initially termed nanoporous and not nanotubular. The work of Macak et al.1A’ 75 developed

critical parameters required to synthesize well-ordered titania nanotubular arrays (T-

NTA) as well as the work of Raja et a l 68 76 for various electrolytes. In these methods, a

Ti substrate typically in the form of a metal foil (~0.2 mm thick) is used as an anode

during the anodization process. Thin films of Ti (350-1000 nm) have also been

successfully used to grow T-NTA. Deposition of Ti on conducting substrates, such as

conductive glass (indium tin oxide, ITO) has been carried out using radio frequency

77 78-80sputtering. Anodization of Ti films deposited on Si substrates via DC sputtering, as

well as e-beam deposited films,81 have also been reported. T-NTA have also been

synthesized using other Ti substrate geometries such as thin wires,82-84 meshes85, 86 and

curved surfaces.87 These studies demonstrate the versatility of electrochemical synthesis

methods on any substrate geometry. Formation of other metal oxide nanotube arrays that

have demonstrated photoelectrochemical applications include hematite nanotubes using

88, 89 90 91iron foil and low carbon steel, TaON nanotubes from Ta2O5 nanotubes, as well as

other Ti-based alloys such as TiN ,92 TiPd,93 TiW ,94 TiRu,95 and TiNbZr.96 Typical

21

expected morphology of nanotubular structures obtained via anodization is shown in

Figure 2.1.

The electrochemical baths typically used in the anodic formation of metal oxide

nanotubes consist of a fluorinated inorganic (e.g., 0.5 M H3PO4 + 0.14 M NaF) or

organic- (e.g., 0.2-0.5 wt% NH4F + 0.2-10 wt% H2O in ethylene glycol or glycerol) based

electrolyte. Important parameters in determining the dimensions of the T-NTA include

anodization potential (1-150 V, D.C.), anodization time (~15 minutes to several hours),

pH, temperature, and fluoride content. The diameter of the nanotubes is essentially

determined by the anodization potential and is a linear relationship where an increase in

22

Figure 2.1: Scanning electron micrographs of the (A) top, (B) side, (C) bottom-side, (D) and bottom view of titania nanotube arrays synthesized by electrochemical anodization outlined by the synthesis method is Section 3.1.4.

potential results in an increase in diameter. Fluoride content and bath temperature are

controlling variables in the wall thickness of the nanotubes. Lower temperatures typically

yield thicker walled nanotubes, while higher fluoride content gives thinner nanotube

walls. The length of the nanotubes is a strong function of electrolyte pH. Low pH

electrolytes result in shorter nanotube lengths regardless of anodization time as a result of

self-etching. Electrolytes with pH ~6 , such as organic-based electrolytes, can for longer

anodization times and yield much longer nanotubes (1 -1 0 0 microns).

A schematic of a typical anodization setup is shown in Figure 2.2. The

experimental setup consists of a two-electrode configuration. The metal to be anodized

serves as the anode, while a Pt flag of larger area than the anode material serves as the

cathode. Potentiostatic anodization is commonly used where the potential is ramped (~5

V/s) from free corrosion potential to the predetermined anodization potential. One study97

examined the galvanostatic anodization on the formation of T-NTA. In this study, the

authors observed oscillation in the voltage with time, which eventually resulted in

unstable oxide films. Other techniques, such as pulse voltage anodization,98 have also

been applied and similar nanotube morphology is obtained compared to potentiostatic

anodization methods.

A common practice employed for electrolyte mixing during the anodization

process is mechanically stirring. The recent works by Sanchez-Tovar et al 100 detail the

effects of hydrodynamic conditions on T-NTA formation. In these studies, a rotating disc

electrode was used and the effects of flouride concentration (diffusion limited conditions)

and Reynolds number on T-NTA morphology were examined. These studies concluded

that highly defined nanotubes can be synthesized using defined flow conditions, in

23

24

Figure 2.2: Typical electrochemical anodization setup for the synthesis of metal oxide nanotube arrays utilizing (a) magnetic stirring and (b) ultrasonication agitation methods. During ultrasonication, to prevent electrolyte heating for prolonged synthesis time, the sonication bath temperature is controlled.

particular the nanotube top morphology, which plays a crutial role in

photoelectrochemical applications.100 Other methods of bath aggitation such as

ultrasonication have also been studied.101 The use of ultrasonication during anodization,

or sonoelectrochemical anodization, results in a more ordered T-NTA morphology. The

kinetics of nanotube formation was increased, as evident by monitoring the anodic

current during anodization. Sonoelectrochemical anodization has also demonstrated

enhanced photoelectrochemical performance over photoanodes prepared using

magnetically stirred anodization baths.101-103

2.2.1 Oxide formation stages during anodization

Metal oxide nanotube formation in the presence of fluoride ions typically occurs

in three stages, as outlined by the shape of the anodization current density-time plot.

Shown in Figure 2.3 is a depiction of the stages of nanotube oxide formation: (i) initial

oxide barrier layer formation, (ii) pore nucleation/formation, and (iii) steady-state

nanotube growth stage.

In stage (i), a large current density is initially measured at the instant of an applied

anodic potential indicating the oxidation of Ti to Ti4+ (T i^ T i4++ 4e-). A rapid decrease in

current density is then observed. This is attributed to the formation of an oxide barrier

layer via hydrolysis reaction:

Ti4+ + 2 H2O ^ TiO2 + 4H+ (17)

During the hydrolysis reaction, H+ ions accumulate, and maintain electroneutrality while

F- ions migrate to the H+ sites. When a critical concentration is reached at local regions,

dissolution of TiO2 occurs by the formation of aqueous hexafluorotitanate.

Ti4+ + 2H+ + 6F- ^ H2TiF6 (18)

The dissolution reaction of Ti cations creates negatively charged cation vacancies in the

oxide which migrate to the metal/oxide interface as a result of the potential gradient

across the oxide.104 The presence of metal-cation vacancies near the metal/oxide interface

25

26

Figure 2.3: Atypical current density vs time plot produced during electrochemical anodization.

facilitate the Ti -> Ti4+ + 4e" reaction as the cations can easily hop to the available

vacancy. This event is marked by an increase in current density (stage ii). During this

stage, the nucleation of pores occurs at the oxide surface. Steady-state growth of the

nanotubular oxide layer is observed when the current density achieves a constant value

over time (region iii). It should be noted that the anodic current density during

anodization is comprised of two components: the first is current due to the dissolution

process at the oxide/electrolyte interface, and the second is current due to the oxidation of

titanium at the metal/oxide interface.62 The pH of the electrolyte plays a large role in the

pore nucleation and growth of the nanotubes as the dissolution rate of the oxide increases

with a decrease in pH .105 Anodization of other metals in a fluorinated electrolyte follow a

similar mechanism. For example, the anodization of Fe to form Fe2O3 nanotube arrays.88-

90

2.2.2 Formation of complex nanotube geometries

Electrochemical anodization allows for growing a variety of tubular- and

hierarchical-type nanostructures by changing a few synthesis variables. The formation of

branching T-NTA is a result from step changes in the voltage during anodization.106-108

To form branching tubes, the voltage is not alternated but maintained at the new voltage

until chemical and field-assisted reactions (i.e., oxidation, ion migration, and dissolution)

have equilibrated and the branched tube has reached the desired length. Chen et al.

reported that the number of tube branches could be controlled by reducing the applied

voltage by a factor of l /V « , where n determined the number of branches.107 Complex

multilayer structures were fabricated as adjustments in voltage caused either an increase

or decrease in b ranching. Mohammadpour et al. observed the formation of large diameter

tubes from the combination of smaller diameter tubes during anodization in low fluorine,

viscous electrolytes at high potentials.106 It was suggested that this structure was the

result of the combination of strong capillary forces banding the tubes together and the

inhibition of electrolyte between the spaces of adjacent nanotubes. Proposed applications

for this hierarchical structure included molecular separation in microfluidic techniques

and photovoltaics owing to the higher surface area and better charge separation at the

“core-leg” interfaces.106

Bamboo T-NTA structures can be formed through electrolyte chemistry

modifications and alternating applied bias. The banded bamboo-like structure can occur

in aqueous or high water content glycerol electrolyte solutions.109, 110 However, glycerol

solutions containing glycerol/H2O ratios higher than 9:1 do not exhibit nodes because of

the lack of periodical current fluctuations at the T-NTA surface. No nodes are observed

27

when ethylene glycol replaces glycerol at similar ratios. This observation was attributed

to the higher diffusion constant of the ethylene glycol. This reduced the fluctuations in

current density during tube formation. More control over the node thickness and

frequency is provided by the application of an alternating bias in an ethylene glycol-

based solution.111 Bamboo structures are often used in applications, such as DSSC, where

maximized dye loading is necessary. Although the stratification between the nanotubes

causes a longer random path for photogenerated charges, improvements in incident

photon to current efficiency (IPCE) are still observed. The bamboo rings also allow for a

higher rate of dye loading due to extra surface area. The extra space introduced between

the tubes due to the rings also allow dye molecules to cover the exterior of the NT

walls.111 Kim et a/. reported that bamboo structures with 70 nm spacing between the

compact nodes provided the largest increase in photoactivity.111

Hierarchical structures of T-NTA such as nanolace-type structures atop the

nanotube arrays can be achieved by alternating voltage technique112 or through surface

treatments.38, 113, 114 Double walled T-NT have also been synthesized using an ionic liquid

fluoride solution in ethylene glycol.115 The common synthesis technique adopted for

growing hierarchical-type T-NTA involves a two-step anodization process. First, T-NTA

are grown on a Ti substrate, and then removed (via ultrasonication for a long duration);

leaving a patterned Ti substrate. The as-formed substrate is then subject to a second

anodization. Smooth nanotubes with a nanoporous/nanolace top layer is the resulting

morphology.

28

2.3 Photoelectrochem ical w ater splitting

2.3.1 Material aspects of PEC hydrogen generation

Materials used for PEC hydrogen generation should perform two fundamental

functions:116

• Optical function required to obtain maximum adsorption of solar energy;

• Catalytic function required for water decomposition.

Other important properties:116

• Low resistance for charge transport

• Effective charge separation

• Low interfacial resistance across the semiconductor/electrolyte interface

• Ability to tune the flat band positions

• Resistance to corrosion in aqueous electrolyte under irradiation

Oxide semiconductors are naturally more stable and have better corrosion resistance

properties than nonoxide semiconductors. Titanium dioxide “fits the bilF for many of the

aforementioned properties, and it has become the focus of many investigations.

2.4 Titania nanotube array heterostructures with metal

chalcogenides

Coupling two or more semiconductor materials together is a common practice to

achieve electrical or electro-optical properties that are otherwise unobtainable by each

material separately. There are a variety of synthesis techniques to obtain such structures

such as CVD, electrochemical, and wet-chemical. In many solar-based applications, large

band gap materials (e.g., TiO2, ZnO, SnO2) are sensitized with smaller band gap

29

semiconductors. These are often nanocrystal (NC) deposits or quantum dots (QD).117, 118

The smaller band gap materials act similar to a dyes in dye-sensitized solar cells (DSSC).

These smaller band gap materials serve the purpose to inject electrons (excited via lower

energy photons) into the conduction band of the larger band gap semiconductor. A

comparison of DSSC and QD solar cells is reported by Hodes.119 The advantage of using

NC or QD as sensitizers lies within the quantum confinement effect, as well as the ability

to tune the particle size120 or composition121 to utilize different wavelengths of light.

117 122 123 117 124Some semiconductor materials with narrow band gap include CdS, ’ ’ PbS, ’

Bi2S3,117, 125 CdSe,126, 127 CdTe,128, 129 and InP.130These have been investigated as

sensitizers for T-NTA as an efficient absorber layer of the visible light spectrum.

Sensitization is commonly carried out with semiconductor sulfides or selenides.

These materials can utilize visible and near-IR photons and can easily be synthesized to

dimensions less than their Bohr radius. Use of nanodimensional metal oxides to enhance

the photoactivity of T-NTA is another method. Although metal oxides generally have

larger band gaps, they are generally more stable. Additionally, integration of different

metal oxides and metal oxide-based semiconductor materials into T-NTA has resulted in

the improvement of the photocatalytic properties of the composite system .131 The

incorporation of such oxides with T-NTA is to alleviate/modify the charge carrier

recombination behavior in the individual oxides. That in turn changes the recombination

properties of the composite. The unidirectional transfer of photogenerated electrons and

holes can be achieved by a suitable matching of the CB and VB bands of the oxide and

the T-NTA. However, the effective conduction of the photogenerated charge carriers and

30

reduced recombination losses depends on various properties exhibited by the oxide

including surface area, defect density, crystallinity, particle size-shape, and composition.

2.4.1 Doped CdS nanocrystals

Utilizing quantum confinement effects, by changing the particle size of the

deposited nanocrystal semiconductors, can have its practical limitations. In some

applications, the quantum confinement effects vanish due to particle agglomeration

either through synthesis techniques or over-extended catalytic use.

One method to alter the intrinsic electronic properties of II-IV semiconductors

without changing the particle size is by introducing optically active metal dopants. This

can be controlled to tune the optical, luminescencet, or magnetic properties.132

Bhargava133 first reported doping ZnS nanoparticles with Mn2+ and examined the

photoluminescence and observed a high quantum yield at room temperature. Both ZnS

and CdS doped with Mn have orange luminescence (580~590 nm) due to Mn d-d (4T1-

6A 1) transition. This transition is both spin and orbital forbidden. These properties arise

from the strong exchange coupling between the localized moments of the paramagnetic

dopant and the band structure of the semiconductor.134 The dopant ions can act as

shallow trap states thereby extending the lifetime of charge carriers. In imaging/display

applications, the trap states act as recombination centers which result in a strong

characteristic luminescence signal. The narrow emission band, broad excitation band,

and better photochemical stability make these luminescent nanocrystals better candidates

than traditional dyes.135

Reports on the technical applications of these types of materials have mainly

31

been in the realm of biological applications such as imaging, targeting, and therapy .136

The extended lifetime of charge carriers, as observed in Mn doped CdS, could be

advantageous to improve the performance of solar energy conversion systems. Lee et

al.137 reported the advantages of using optically active metal dopants in a nanocrystalline

solar cells by sensitizing mesoporous titania with a Mg-doped CdSe/CdSe layer

achieving 1% efficiency in solar power conversion. Recently, Santra and Kamat

demonstrated that a quantum dot sensitized solar cell with Mn-doped-CdS/CdSe

sensitizer layer over mesoporous titania can achieve a power conversion efficiency of

5% .138 Luo et al. constructed a similar solar cell consisting of core-shell CuInS2-Mn

doped CdS sensitizer layer and reported a power conversion efficiency of 5% as well.139

The use of other metal dopants such as Cu140 or Co141 have also been employed to

enhance the performance of nanocrystalline PbS/CdS and CdSe solar cells, respectively.

The long lifetime of charge carriers attributed to the dopant, coupled with an appropriate

redox couple to scavenge holes, leads to the boost in cell performance. Sensitization of

T-NTA with CdS nanocrystals has been carried out through a variety deposition of

methods.24, 123, 142-145 The results presented in these works demonstrate improved

photoelectrochemical performance of T-NTA as well as increased visible light utility

attributable to CdS. In a previous report, the addition of a titania nanoparticle (T-NP)

layer on T-NTA has been shown to improve the CdS loading and photoelectrochemical

performance of a T-NTA/T-NP/CdS system ;85 however, a T-NTA/T-NP (NTNP)

architecture sensitized with Mn or Co doped CdS layer system and the

photoelectrochemical responses has yet to be examined. It is envisaged that with doped-

CdS/NTNP nanoarchitecture, improved photoelectrochemical responses can be realized.

32

CHAPTER 3

MATERIALS AND METHODS

3.1 A nodic synthesis o f titania nanotubular arrays

Sample names have been assigned based on their synthesis procedure and are

outlined in Table 3.1.

3.1.1 Photoelectrochemical anodization

A depiction of the experimental setup for anodization under irradiation is shown

in Figure 3.1. Samples were prepared without irradiation (referred to as T-NTA), with 1

h irradiation (referred to as T-NTA-60), and 0.5 h irradiation for the latter half of

anodization time (referred to as T-NTA-30) under the experimental setup. UV-vis

irradiation of the Ti substrates during anodization was supplied by a medium pressure

mercury immersion-type lamp (Ace glass, 7825-34) with a quartz water jacket. The bath

temperature was maintained at 30 ± 5 oC throughout the experiment. The light intensity

during anodization was held constant and measured to be ~72 mW/cm2 (Newport 70260,

radiant power meter). In summary, Ti foils (1.5 x 2 cm, ESPI Metals, G1 grade) were

degreased under ultrasonication for 15 min in a 50/50 vol% acetone/isopropanol solution.

After drying in air, one side of the Ti foil was masked with Kapton® tape for nanotube

growth to occur on one side of the substrate. Anodization was carried out at 60 V for a

total of 1 h in a fluorinated solution of ethylene glycol (0.25 wt.% NH4F, 10 wt.% DI

H2O) under mechanical stirring with a Pt gauze (52 mesh) cathode. Post anodization, the

samples were washed thoroughly with isopropanol and DI water followed by drying at

110 oC in air for a few hours. Crystallization of the amorphous samples was carried out at

500 oC for 2 h under N2/H2 (5% H2/N2 balance) atmosphere with a heating rate of 1.5

oC/min followed by natural convective cooling.

34

35

Table 3.1: Sample identification and corresponding synthesis description section.

Sample Name__________ Sample Description__________Synthesis Details Section

T-NTA Plain TiO2 nanotubes 3.1.1T-NTA-30 Light assisted anodized 3.1.1T-NTA-60 Light assisted anodized 3.1.1TNA Plain TiO2 nanotubes 3.1.2H-TNA Hierarchical TiO2 nanotubes 3.1.2TNT Plain TiO2 nanotubes 3.1.3TNT-WO3 Binary acid anodized 3.1.3TNTA Plain TiO2 nanotubes 3.1.4TNTA-CdO TiO2 nanotubes with CdO 3.2.2NTNP TiO2 nanotubes/TiO2 nanoparticles 3.2.2NTNP-CdS NTNP with CdS 3.2.2NTNP-CdS-Mn NTNP with Mn doped CdS 3.2.2NTNP-CdS-Co NTNP with Co doped CdS 3.2.2

Figure 3.1: Depiction of the top view of the experimental setup for used for anodization in the presences of UV-vis irradiation. Light is supplied via immersion lamp and the temperature is maintained by water circulation jacket.

3.1.2 Synthesis of hierarchical nanotubular structures on foil

and wire substrate

Titania nanotubes were synthesized on titanium foils (1.5 x 1 cm, ESPI Metals,

99%) and wires (3 cm, O = 0.25 mm, 99.7%, Aldrich) similar to previous reports.82, 83 In

summary, the Ti substrates were degreased under ultrasonication for 15 min. in 50/50

vol% acetone/isopropanol. After drying in air, the titanium substrates were anodized at 60

V for 1 h in a fluorinated solution of ethylene glycol (0.25 wt% NH4F, 10 wt% DI H2O),

followed by a brief cleaning with DI water. The T-NT were further cleaned by dipping

the anodes in a beaker of DI water under ultrasonication for ~1-3 s. After drying in air,

the samples where then calcined at 500 oC for 2 h in air with a heating rate of 1.5 oC/min

followed by natural convective cooling. To synthesize hierarchical T-NT, the foil and

wire substrate, after cleaning and washed with DI water, were etched in a solution of HF

(35%), HNO3 (68-70%), and DI water in the volumes of 1:3:50 ml for various times.

After another washing with DI water immediately after etching, the titanium substrates

were anodized and calcined by the aforementioned procedure. Plain titania nanotubes are

denoted as TNA, while hierarchical nanotubes is denoted as H-TNA. The summary of the

synthesis procedure is shown in Figure 3.2.

3.1.3 Binary acid anodization

Cleaned and polished titanium metal strips were anodized at 20 VDC for 1 h in an

aqueous solution containing 0.5 wt.% NH4F for TNT synthesis, and 0.5 wt.% NH4F with

2.4 wt.% phosphotungstic acid (PTA) for TNT-WO3 synthesis. Preliminary studies

indicate 2.4 wt.% of PTA to be the optimal loading. After anodization the samples were

36

37

(A)

Etch

Anodize Sonicate/Remov

TConventional 2 Step Method

(B)Anodize

Single Step

SecondAnodization

Figure 3.2: Schematic for the growth of hierarchical titania nanotubes. The conventional synthesis methodology (labeled path ‘A’) is through a two-step anodization process where the Ti substrate is anodized and the nanotubes are removed and the substrate is then subject to a second anodization. The method presented here utilized an etching treatment, reducing an anodization step, as outlined by path ‘B \

washed with DI water and then dipped into a beaker of DI water under sonication for a

few seconds to remove surface ions, and then dried in air. The samples were then

calcined at 550 °C for 2 h in air atmosphere with a heating rate of 1.5 °C/min followed by

natural convective cooling.

3.1.4 Synthesis of titania nanotube array heterostructures

Titania nanotubes (TNTA) used for fimctionalization with CdO were prepared

85with similar approach as described previously. In brief, 20 mm x 15 mm size of

titanium metal strip (0.1 mm thick) was extracted, cleaned in DI water followed by

ultrasonic cleaning in acetone and isopropyl alcohol for 15 min. The cleaned samples

were anodized at 40 V for 1 h in a fluorinated solution of ethylene glycol containing 0.5

wt.% NH4F and 10 wt.% H2O. Anodized samples were washed and sonicated for ~ 3 s in

DI water followed by drying in air. Annealing was carried out in air atmosphere at

500°C for 2 h with a heating rate of 1.5°C/min followed by natural cooling.

For titania nanotubes sensitized with doped CdS, the plain nanotubes (NT) were

prepared by sonoelectrochemical anodization of titanium foils (0.2032 mm thick, ESPI

Metals, G1 grade) .146 In summary, Ti coupons (~1.5 x ~1.5 cm) were briefly polished

with emory paper and degreased in acetone/isopropanol by sonication for 5-10 min.

After rinsing with DI water, the Ti coupon was subject to sonoelectrochemically

anodization at 60 V (D.C.) for 1 h in a fluorinated ethylene glycol electrolyte (0.5 wt.%

NH4F + 2 wt.% H2O) using a Pt mesh counter electrode. To apply ultrasonic waves

during anodization, the beaker containing the anodization electrolyte was submerged in

an ultrasonic bath (Branson 5510, 100 W, 42 kHz). Cooling was supplied to the

ultrasonic bath to maintain an electrolyte temperature at 30 oC. Post sonoelectrochemical

anodization, the anodes were rinsed and sonicated in DI water for 2 min. and

subsequently dried under vacuum at 110 oC overnight.

3.2 Post synthesis treatments

3.2.2 Deposition of titania nanoparticles, CdO nanocrystals,

and doped CdS nanocrystlas

CdO was anodically deposited on TNTA at room temperature in an aqueous

solution containing 0.05 M Cd(C2H3O2)2.2H2O and 0.1 M Na2SO4 with constant

magnetic stirring. Cyclic voltammetric studies were performed in the solution to obtain

suitable deposition conditions. Optimized deposition conditions were carried out

38

galvanostatically at a current density of -0.1 mA/cm2, for various times of 200, 300, 500,

and 1000 seconds. After deposition, samples were rinsed with DI water and dried in air.

Samples denoted NT prepared by the procedure outlined in Section 3.1.4 were

then annealing at 350 oC for 2 h in air. Subsequently, nanoparticles of titania were

deposited onto the nanotubes by treatment in 0.2 M TiCl4 at 65-70 oC in a sealed reactor

for 20 min followed by washing with ethanol and water. A second annealing was

performed at 450 oC for 2 h in air147, 148 to deposit titania nanoparticles, forming a

nanotube/nanoparticle structure (NTNP).

Deposition of CdS nanocrystals was carried on NTNP samples out using a

subsequent ionic layer absorption reaction (SILAR) technique similar to previous

reports.138, 147 One SILAR cycle consisted of dipping an annealed T-NTA sample in 0.1

M Cd(NO3)2 in methanol, under stirring for 1 min, followed by washing in methanol, and

subsequent dipping in a 0.1 M Na2S in methanol/water solution (50/50 vol.) for 1 min

under magnetic stirring with subsequent washing in methanol/water. After eight SILAR

cycles, the samples were dried under vacuum at 110 oC overnight. Incorporation of Mn2+

or Co2+ was carried out by adding 0.075 M Mn(CH3CO2)2 or Co(CH3CO2)2 into the

Cd(NO3)2 solution prior to carrying out SILAR deposition.

3.3 Characterization techniques

3.3.1 Surface morphology and composition (SEM/EDS and XPS)

Scanning electron micrographs and electron dispersion spectroscopy mapping

were collected using a Hitachi S-4800 SEM with an Oxford EDS detector. Surface

composition was analyzed by XPS (Phi 5600 model). The spectrometer was calibrated

39

with Ag 3d5/2 line at 368.27±0.05 eV and the excitation source was monochromatic AlKa

(1486.6 eV). The vacuum in the analyzing chamber was maintained less than 1 x 10-9

Torr and the analysis area was 800 ^m2. The survey spectra were acquired at pass energy

of 29.35 eV and narrow scans at 23.95 eV. Charging effects were corrected using

adventitious C 1s line at 284.6 eV as internal reference. All peaks were fitted using SDP

v4.6 Gaussian fitting software from XPS International and underwent five points of

smoothing.

3.3.2 Crystallinity and optical properties (XRD, UV-vis, Raman spectroscopy)

X-ray diffraction analysis was carried out using a Rigaku Miniflex XRD (CuKa =

1.54059 A) from 29 = 2 0 to 60 degrees with a step size of 0 .01 degrees and dwell time of

0.5 degrees/min. The XRD data was analyzed using Rigaku PDXL2 analysis software.

Raman Spectra was recorded using a Thermoscientific DXR Raman microscope operated

with a 532 nm laser. A total of 64 exposures were recorded with 2 seconds per exposure.

The sample was analyzed through a 25 ^m pinhole at optical magnification of 50x.

Absorbance spectra were measured using a Shimadzu UV3600 Spectrophotometer.

3.3.3 Photo/Electrochemical characterization

All PEC experiments were carried out using a setup depicted in Figure 3.3. In

general, a three-electrode system was utilized consisting of the titania sample as the

anode, a Pt mesh as the cathode, and a Ag/AgCl (3 M) served as the reference electrode.

The anodes were irradiated at 100 mW/cm2 AM 1.5 irradiation in 1 M KOH (pH = 14),

0.5 Na2SO4 (pH = 6 ), or 0.25 M Na2S + 0.35 M Na2SO3 (pH = 12.2) electrolyte. All

40

41

Figure 3.3: Experimental setup for photoelectrochemical testing.

potentials are with respect to Ag/AgCl unless otherwise specified. Polarizing from open

circuit potential to a final potential at 10 mV/s. performed as potentiodynamic (J-V)

measurements. Potentiostatic (J-t) measurements were recorded at under chopped light

irradiation. A Gamry Reference 6000 or Princeton Applied Research, Parstat 4000 were

used control the potential and record the current.

Electrochemical impedance spectroscopy (EIS) 149 and Mott-Schottky (M-S)

analysis150 are powerful techniques used to examine interfacial charge transfer events at

an electrode/electrolyte interface. With EIS, it is possible to readily separate the

interfacial capacitance, charge transfer resistance, and charge trap states.151 Based on

well-documented theory of semiconductor/electrolyte interfacial capacitance, the

electronic properties in terms of charge carrier density (NA) and flat-band potential (EFB)

can be obtained from the Mott-Schottky equation:152

42

J _____ 2_

C2 e££nN-Aq>-

0lyA '

-A y = E - Efb

e *(19)

(2 0 )

where, C is the capacitance, e is the elementary electronic charge, e0 is the permittivity in

vacuum, e is the dielectric constant, kB is the Boltzman constant, T is the temperature, and

E is the applied bias. From the slope (m) of 1/C2 vs E, NA can be determined by,

Na = - ^ - (21)eee0m

where the intercept of the line yields EFB. The linear portion of the 1/C2 vs E plots is fit

within the potential domain in which the samples behave as capacitors.

The efficiency for the conversion of light energy to chemical energy under an

applied external potential is determined using the following equation:153

j (E 0 - E ) n(%) = ^ x 100 (2 2 )

Io

where, n is the photoconversion efficiency, j p is the photocurrent density (mA/cm2), I0 is

the incident light irradiance (mW/cm2), E°rev is the standard reversible potential for H2,

which is 1.23 VRHE, and Eapp is the applied potential which is Eapp = Emeas - Eaoc, where

Emeas is the electrode potential (vs Ag/AgCl) of the working electrode and Eaoc is the

electrode potential (vs Ag/ AgCl) of the same working electrode under open circuit

condition under illumination.

Quantitative hydrogen generation experiments were conducted in a two-electrode

electrochemical photocell on H-TNA samples (Figure 3.4). Two separate compartments

for the anode and cathode were separated by a fine porous glass frit. The anodes were

illuminated through a quartz window (2 cm diameter). When wires were used as the

anode, the wires were formed into coils and were juxtaposed to one another by attaching

to a titanium strip with carbon and copper tape. The coil configuration allows for more

area of the wire to be exposed to incident light within reactor window. Moreover, on a

per area basis exposed to incident light, titania nanotubes grown on a wire in a coil

configuration over a straight wire demonstrate enhanced photocatalytic activity.154 A Pt

coil spot-welded to a stainless steel rod served as the cathode. The electrolyte in each

compartment was 1 M KOH with 10 vol% ethylene glycol.155 The cathode was inserted

into a burette where the hydrogen was collected via electrolyte displacement. The volume

of hydrogen was measured by directly reading the variation of the electrolyte level in the

burette for various times. The cell voltage was maintained at 0.5 V by a computer-

controlled potentiostat (Princeton Applied Research, Parstat 4000).

43

44

Figure 3.4: Experimental setup for quantitative hydrogen generation. (1) Solar simulator with AM 1.5 irradiation, (2) TiO2 nanotube anode compartment which is separated from the (3) cathode (Pt wire) by a porous quartz frit, (4) gas burette, (5) gas collection port, and (6 ) close up of the Pt wire showing hydrogen bubbles.

CHAPTER 4

RESULTS AND DISCUSSION

4.1 Synthesis

4.1.1 Nanotube growth mechanism

The formation mechanism outlined in Section 2.2.1 for nanotube synthesis is

generally accepted. However, this generalized mechanism does not readily explain the

driving force for the formation of a tubular morphology over a porous oxide layer.

Despite the efforts outlined in Section 2.2, there is no common agreement between

proposed mechanisms. In this section, we propose a generalized mechanism, based on

critical parameters that are required for pore nucleation and pore wall/tube separation.

A compact barrier layer forms when the oxide is not soluble in the anodization

bath, while a porous film can form when the oxide is moderately soluble. During the

initial stages of anodizing titanium, a soluble barrier layer is formed on the surface. It is

important to distinguish during the anoidzation process whether pore formation occurs

due to chemical dissolution of the oxide or if other factors, such as surface stability, that

play a roll. For example, during anodization of Al, the Al3+ ions in solution are mainly

attributed to field assisted ejection into the electrolyte and not due to chemical

dissolution, as pore initiation is also possible when no oxide is formed at the

oxide/electrolyte interface for Al.156 It is speculated that the initial pore formation is a

result of surface instability of the barrier oxide layer. When an oxide layer is formed, a

stress is developed along the metal/oxide interface. Mechanisms of stress relief include

oxidation-induced strain compensated by a relatively large flow of cations from the

metal, and stress generation due to varying stoichiometry of the oxide layer. Instability of

the formed barrier layer is a result of competing processes, i.e., surface energy acting as a

stabilizing force and increased strain energy due to electrostriction, electrostatic, and

recrystallization stresses trying to destabilize the surface.68 If we consider a perturbed

46

surface (Figure 4.1a), the surface energy density varies along the peaks and wells of the

surface. This leads to higher strain energy in the wells, serving as a driving force for

surface instability. A study by Asaro and Tiller157 demonstrated that below a critical wave

length of perturbation, an oxide surface becomes unstable. This wavelength is determined

by the ratio of the surface energy to strain energy. A recent work by Hebert et al 65

developed a quantitative model, and validated with experimental results, to predict

morphological stability of anodic films on Al and Ti correlated to pore initiation. Their

model was able to predict that for the wavelength of a perturbed surface yielding the

maximum disturbance, the growth rate should be proportional to the oxide thickness. This

relationship quantitatively explains the observed proportionality of pore spacing to

anodization voltage, and self-ordering structures are only satisfied within a narrow range

of oxide formation efficiencies (0.50-0.58 for titania). Therefore, it has been verified that

the initial pore formation mechanism for both Al and Ti is attributed to instability of the

perturbed barrier oxide layer and not due to pitting of the oxide layer from halide species.

Once the surface becomes unstable, and a critical barrier layer thickness is

achieved, the applied electric field drives the anionic species to the wells of the perturbed

surface (Figure 4.1a). This is where the applied field strength is the highest.158 Fluoride

and hydrogen ions contribute to decreasing the surface energy even more so, resulting in

a further increase in surface instability. During the hydrolysis reaction, H+ ions

accumulate maintain electroneutrality, while F- ions migrate to the H+ sites. When a

critical concentration is reach in localized regions, dissolution of TiO2 occurs by reaction

18 (Ti4+ + 2H+ + 6F- -> H2TiF6). The dissolution reaction of Ti cations creates negatively

charged cation vacancies in the oxide, which migrate to the metal/oxide interface as a

47

48

Figure 4.1: Schematic depiction of a (a) smooth surface perturbed by a sinusoidal wave allowing for even distribution of electric field and even pore nucleation. A rough surface (b) results in inhomogeneous distribution of electric and pore nucleation. (c) Depiction illustrating the diffusion profiles inside the nanotubes and through the oxide layer. Dissolution of titania occurs at the tops and wells, while the oxidation occurs primarily at the wells.

result of the potential gradient across the oxide.104 The presence of metal-cation

vacancies near the metal/oxide interface facilitates the ejection of Ti4+ to the available

vacancy (Figure 4.1c). For titania, it is important to distinguish between aqueous and

organic electrolytes. When an aqueous electrolyte is used, pore nucleation can be

attributed to both field-assisted ejection of Ti4+ as well as chemical dissolution of

titania.156 It has been observed that pores and cavities are immediately formed once a

potential is applied to the cell.159 When using an organic electrolyte, the chemical

dissolution is much less. In such cases, the Ti4+ ejection into the electrolyte is an essential

step for pore development.

It is also important to take surface roughness and the crystal planes exposed to the

electrolyte into consideration. The grain structure at the electrode/electrolyte interface

assumes significance in the context or pore nucleation and subsequent tube formation. A

relatively smooth surface enables uniform pore nucleation by even distribution of eclectic

field as well as field-induced and chemical etching of the oxide (Figure 4.1a). When the

surface is rough and the surface grains are comparable in size to the amplitude and

spacing of the pores, the pore formation becomes irregular (Figure 4.1b). On an irregular

surface, the etch rates of various crystal planes will vary across the surface. Studies by

Crawford and Chawla, 105 and Leonardi et a l .160 showed that differences in nanotube

formation rates on different Ti crystal planes arise due to differences in atomic density

and oxidation rates. Recently, we have examined the effect of D.C. magnetron sputtered

films that affect the morphology of titania on Si wafers.161 When the deposited Ti films

are rough with pronounced three-dimensional surface grains, the oxide morphology

transitions from nanotubular to nanoporous. In our other studies80, 81 on anodization of Ti

thin films by various deposition methods, the small surface grains result in a nanoporous

layer on top of a nanotubular oxide layer. Further details on this are discussed in Section

4.1.3. Recent studies by other groups on the effect of surface morphology on Ti thin

films162 as well as surface roughness of metal foils163 on morphology of the formed oxide

layer observed similar results.

49

The separation of the pore walls in titana is probably due to the dehydration of

titanium hydroxide and radial cation vacancies. It has been shown experimentally that the

outer oxide layer exposed to the electrolyte has excess hydroxyl ions compared to the

inner oxide layer not exposed to the electrolyte.164, 165 Three ions, namely F-, O2-, and

OH-, migrate from the electrolyte to the oxide/metal interface (Figure 4.1c). The size and

charge strength on these ions dictate their mobility, as OH- are larger than F- and O2- and

contains less charge than O2- ions. As a result, the OH- ions migrate slower than the F-

and O2- and most likely result in the formation of predominantly [Ti(OH)n](4-n)+ species

on the surface (n = 2-5) .166 This is likely why a fluoride-rich layer is observed between

the metal substrate and oxide layer. 69

Considering a mechanism proposed by Raja et a / . ,68 where cation vacancy

transport in the radial direction leads to cell boundary separation, cation vacancies can

migrate along the electric field due to their negative charge. Along the peaks of the

surface (Figure 4.1a), the field strength is not as high. If the cation vacancies are

transported radially, vacancies would populate within the area of two neighboring pores.

Since the charges are the same, they would repel to maintain an equilibrium distance. To

maintain electroneutrality, oxygen vacancies could be generated. If the dissolution of the

of the oxide is much higher than the generation oxygen vacancies at the metal/barrier

layer interface, the repelling forces of the cation vacancies could cause separation of

neighboring pores. Once the cell boundaries separate, they would become exposed to the

electrolyte while forming a thin titanim hydroxide outer layer.

There is likely a concentration gradient across the film from pure titania (inner

oxide) to hydrated titania (oxide/electrolyte). The degree of hydration is highly dependant

50

on the solution pH and subsequent surface charge. The formation of titanium hydroxide

can be summarized by the following reactions and subsequent dehydration of Ti(OH)x to

TiO2:156, 167

51

Ti ^ Ti4+ + 4e- (23)

H2O ^ H+ + OH- (24)

OH- ^ H+ + O2- (25)

Ti4+ + xOH- ^ Ti(OH)x (26)

Ti4+ + 2O2- ^ TiO2 (27)

Ti(OH)x ^ TiO2 + (x-2)H2O (28)

Reactions 23-27 represent the possible field-assisted oxidation processes, as the

applied electric field controls the rate of ion migration within the metal/oxide interface. In

reaction 28, titanium hydroxide is dehydrated to titanium dioxide. This results in a cell

volume decrease. Neighboring nanotubes are separated by the volume shrinkage of two

neighboring cell walls. It is worth noting that a pH gradient is established within the

nanotube during anodization75 (Figure 4.1c). The pH of the organic electrolyte used in

this study was measured to be ~6 , while a pH of ~2 can be achieved at the base of the

na