DEVELOPMENT OF CARBON NANOTUBE-BASED GAS AND VAPOUR SENSORS AND SUPRAMOLECULAR CHEMISTRY OF CARBON NANO-MATERIALS

Lee John Hubble BSc (Forensic Science)(Honours) AMRSC MRACI

Centre for Forensic Science The University of Western Australia

This thesis is presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy/Masters of Forensic Science

2009

© 2009 Lee John Hubble

Dedicated to the memory of my beloved grandmother Linda Watson

“Anybody who has been seriously engaged in scientific work of any kind realizes that over the entrance to the gates of the temple of science are written the words: Ye must have faith. It is a quality which the scientist cannot dispense with.”

Max Planck (1932)

Candidate Declaration I declare that the research presented in this thesis, as part of the degree of Doctor of Philosophy/Masters of Forensic Science, at The University of Western Australia, is my own work. The results of the work have not been submitted for assessment, in full or part, within any other tertiary institute, except where due acknowledgement has been made in the text.

……………… Lee John Hubble

I

DECLARATION FOR THESIS CONTAINING PUBLISHED WORK AND WORK PREPARED FOR PUBLICATION The examination of the thesis is an examination of the work of the student. The work must have been substantially conducted by the student during enrolment in the degree. Where the thesis includes work to which others have contributed, the thesis must include a statement that makes the student’s contribution clear to the examiners. This may be in the form of a description of the precise contribution of the student to the work presented for examination and/or a statement of the percentage of the work that was done by the student. In addition, in the case of co-authored publications included in the thesis, each author must give their signed permission for the work to be included. If signatures from all the authors cannot be obtained, the statement detailing the student’s contribution to the work must be signed by the coordinating supervisor. Please sign one of the statements below.

1. This thesis does not contain work that I have published, nor work under review for publication.

Signature......

2. This thesis contains only sole-authored work, some of which has been published and/or prepared for publication under sole authorship. The bibliographical details of the work and where it appears in the thesis are outlined below.

Signature......

3. This thesis contains published work and/or work prepared for publication, some of which has been co-authored. The bibliographical details of the work and where it appears in the thesis are outlined below.

L. J. Hubble & C. L. Raston, ‘Nanofibers of fullerene C60 through interplay of ball-and-socket supermolecules’, Chem. Eur. J., 2007, 13, 6755-6760. Hubble 90%, Raston 10%, Chapter 5.

L. J. Hubble, T. E. Clark, M. Makha & C. L. Raston, ‘Selective diameter uptake of single-walled carbon nanotubes in water using phosphonated calixarenes and ‘extended

II arm’ sulfonated calixarenes’, J. Mater. Chem., 2008, 18, 5961-5966. Hubble 90%, Raston 5%, Clark 3%, Makha 2%, Chapter 2.

Signature...... (Candidate)

Signature...... (Coordinating Supervisor)

III

ABSTRACT The scientific endeavours described within this thesis attempt to create novel solutions to current scientific, commercial and industrial downfalls, and contribute to the advancement of technologies in these areas. This has been achieved through the application of theoretical and experimental principles, entrenched in the domains of chemistry and physics, which have been harnessed to assist in the transformation from nanoscience to nanotechnology. These solutions range from unique supramolecular systems capable of selective-diameter enrichment of single-walled carbon nanotubes (SWCNTs), to the fabrication of low-cost, potentially remote deployable carbon nanotube-based gas and vapour sensors, and expand right through to the development of water-soluble fluoroionophoric sensors and manipulations of a molecular form of carbon in constructing all-carbon nano-architectures.

For the advancement and successful integration of carbon nanotubes (CNTs) into commercial processes, the advent of scalable separation protocols based on their electronic properties is required. SWCNTs have been successfully solubilised using water-soluble p-phosphonated calix[n]arenes (n = 4, 6, 8) and ‘extended arm’ upper rim functionalised (benzyl, phenyl) p-sulfonated calix[8]arenes. Selective SWCNT diameter solubilisation has been demonstrated and subsequent preferential enrichment of SWCNTs with semiconducting or metallic electronic properties has been achieved. In addition, semiconducting nanotube-enriched supernatants (liquid) have been utilised to fabricate on/off field effect transistors (FET). These water-soluble supramolecular systems can be incorporated into post-growth purification protocols, with direct implications in areas such as carbon nano-electronics and device fabrication.

In the current global environment there is a heightened level of public and governmental disquiet due to the reality of impending terrorist attacks. This is compounded by the inherent ease of manufacture and effectiveness of specific chemical warfare agents

(CWAs) used in small-scale terrorist operations. Phosgene, COCl2, is a CWA which is the frontrunner of modern chemical warfare, and is commercially produced as well as being readily prepared from common chemicals. It is a colourless gas, with toxic effects well below its odour threshold of 0.4 ppm. In combating terrorism associated with phosgene, and other CWAs, new detection technologies are required for the purpose of release monitoring, personnel safety, operational intelligence, and warfare IV tactics. A nanotechnology solution is reported for the detection of phosgene which involves a drop-casting technique in constructing a chemiresistor device based on SWCNTs. The entire process from design and development, through to system verification and analyte testing is described. Phosgene is detected over common organic solvents and other halogenated systems, at 24 ppm limits. Sensing G-series nerve agent simulants for sarin (GB) and soman (GD) are also established using the same device, along with the effectiveness of single interdigitated electrode (IDE) sensor geometry versus multiple-array systems.

The ability to control and manipulate structures at the molecular level is appealing for the development of new materials, through unique nano-scale architectures, forming the basis of crystal engineering. Subsequently, a novel supramolecular system is described where toluene solutions of p-tBu-calix[5]arene and C60 result in a 1:1 complex,

(C60)∩(p-tBu-calix[5]arene), which precipitates as nano-fibres. The principal structural unit is based on a host-guest ball-and-socket nanostructure of the two components, with an extended structure comprised of zigzag/helical arrays of fullerenes, predicted through powder X-ray diffraction data coupled with molecular modelling. Under argon, high temperature all-carbon analogues are produced with interesting properties. In addition, results will be described which indicate the potential of this novel complex as an autophagy inductor in cervical cancer, and as a chemo-sensitiser towards drug-resistant breast cancer cell lines. Additional all-carbon structures are described with the formation of rings of helical SWCNT bundles through post-growth SWCNT modifications, and a variety of fibrous all-carbon structures, most notably novel square-geometry carbon nano-fibres (CNFs), through catalytic-chemical vapour deposition (C-CVD) synthesis strategies.

The current requirement for entirely water-soluble fluorescent sensors is routinely documented in the literature. The autofluorescence properties of p-phenyl-sulfonated calix[8]arene are characterised and this water-soluble cavitand is surveyed as a metal cation sensor candidate. This particular system was found to exhibit a change in fluorescence response when exposed to divalent metal cations, and interactions with 2+ 2+ 2+ 2+ [UO2] , Pb , Co , and Cu ions are discussed in detail. The system is characterised through a variety of analytical techniques to yield sensor calibration data, degradation characteristics, pH sensitivity and suitability as a ‘small molecule’ drug-carrier.

V

ACKNOWLEDGEMENTS Firstly I would like to thank my supervisor, Professor Colin Raston, for believing in this research from its conception and giving me the opportunity to undertake these investigations within his research group. Over the timeframe of my candidature at UWA I have gained a wealth of knowledge from you, and your wisdom in the art of science will only strengthen my research career. I thank you for your constant support and your willingness to let me explore a multitude of different research areas in the quest for knowledge. I could not have asked for a more professional supervisor and we will definitely continue our friendship in the future.

To my co-supervisor, Professor John Watling, thank you immensely for your guidance, encouragement and moral support over the years; it has been an absolute pleasure. You have been a major influence on my research career and I have learnt so much from you, and am still amazed with your wealth of knowledge. Over the years we have worked together, spanning back to my early undergraduate days, we have developed a valuable friendship which I will treasure for years to come.

I truly am grateful to Dr. Kevin Winchester who offered much needed advice on device fabrication and sensor testing; to a chemist this was invaluable. To Dr. Adrian Keating, I cannot thank you enough, I appreciate your patience with me and thank you for your ongoing interest in the research projects.

To all the members of the Raston Research Group, past and present, I would like to thank you. You have made my days in the lab enjoyable and I have gained so much from our fruitful scientific discussions. I would like to extend particular appreciation to: Dr. Nathan Webster (Champ), you immediately made me feel welcome at UWA and we developed a fantastic friendship that will last a lifetime. Dr. Thomas Clark (Pommy Tommy), you constantly made me laugh with your antics and stories, and made my PhD extremely enjoyable.

I would like to thank all of the members, both past and present, of the Forensic & Analytical Chemistry Research Group. I have worked alongside side you all for the

VI majority of my research career, and consider you all friends and value your support. A special mention goes to Allen Thomas, Sven Fjastad, Emma Bartle, Rachel Green, Alex Martin, Anna Bradley, and Isaac Chien. In particular, I would like to thank Chris May, we have been friends long before we started our PhDs together and this friendship has only been compounded by the pressures of completing a PhD. You have been a constant source of support and much needed laughter. I value our in-depth discussions and the ability to bounce ideas off of you. I have no doubt that we will be life-long friends and I wish you all the best for your future endeavours.

I would like to thank the Centre for Forensic Science (CFS) for its support over the years, in particular with the early transition stages of my PhD, and for supporting a timely completion scholarship. In particular, I would like to thank Professor Ian Dadour, you have been more of a friend to me than a Director, and that is the way it should be. I appreciate your constant support throughout my time at UWA, during which we have had some great laughs together. I will always reflect back on our friendship with a smile and look forward to continuing it in the future.

My explicit thanks are extended to all past and present staff members at the Centre for Strategic Nano-Fabrication (CSNF). In particular, Centre Managers, Tara McLaren and Diane Arnott, for facilitating conference travel bookings and accommodation. In addition, I would like to extend my appreciation to all staff members of the Centre for Microscopy, Characterisation and Analysis at the UWA for all their assistance over the years. In particular, Dr. Martin Saunders, your endless hours of TEM advice will always be remembered.

I am grateful to Dr. Gerard Poinern for his guidance in the initial stages of this research, and for the use of the synthesis facilities at Curtin University of Technology. I must also thank Professor Bill van Bronswijk and Dr. Peter Chapman, from the Vibrational Spectroscopy Facility at Curtin University, for their assistance with Raman spectroscopy. I am also grateful to Dr. Thomas Becker for your time and guidance with AFM at Curtin University of Technology.

VII

Many thanks go to the Microelectronic Research Group (MRG) for your guidance and advice on everything electrical and device fabrication related. In particular I would like to thank Dr. Byron Walmsley, Jason Milne, and Ryan Westerhout, you all took time out of your studies to help me with mine, and for that I will always be in debt.

To all the members, past and present, in the School of Biomedical, Biomolecular and Chemical Sciences (BBCS), I thank you. I am most appreciative to the Workshop staff, Nigel Hamilton, Andrew Sawyer, and Damien Bainbridge, you all provided much needed knowledge and guidance for the development of the sensor testing facility. I appreciate that no job was too difficult and you always had time for me. These thanks extend to Sarah Davis, for your constant assistance not only in the Chemistry Store but in the Glassblowing Workshop. I would also like to thank Oscar Del Borrello for generously allowing me to borrow measurement equipment for indefinite amounts of time.

Portions of the research described in this thesis could not have been achieved without collaboration with others. I would like to thank Professor Long-Ping Wen from the University of Science and Technology of China, for the cancer cell testing; Dr. John Stride and Mohammad Choucair from the University of New South Wales for supplying ample amounts of graphene; and Dr. George Favas from The Defense Science and Technology Organisation, for assistance regarding the chemical vapour generator development.

I must thank the University of Western Australia and the Australian Research Council for an Ad hoc scholarship, to provide financial assistance throughout my PhD, and for granting a much needed completion scholarship for the final period of compiling this thesis. I thank all organisations which provided awards and travel grants throughout my candidature, with a special mention to professional affiliations of the Australian and New Zealand Forensic Science Society, and The Royal Society of Chemistry.

VIII

I would like to extend my appreciation to Dr. Frank Lincoln, who unselfishly took the time to painstakingly proofread this thesis in its entirety.

An enormous thank you must be given to my family, Paul, Cindy, Kyle, and Chantelle. Ever since I can remember you have always believed in me and supported me in everyway possible. A special mention has to be given to my Nanna, to whom this thesis is dedicated. You were always so proud of me and I wish you were here to see everything I have achieved.

Last, but by no means least, I would like to thank my loving, and patient, girlfriend Sam. You have provided unconditional support throughout this entire journey, and I can truly say I would not be at this point without you. I can only imagine what it must have been like putting up with my crazy hours, not to mention the ‘thesis beard’. Words cannot begin to describe how much your support means to me.

IX

DETAILS OF PUBLICATIONS, CONFERENCE ABSTRACTS AND PATENTS Publications

L. J. Hubble, R. J. Watling & C. L. Raston, ‘Single-walled carbon nanotube-based chemiresistor for the detection of chemical warfare agents’, targeted at Nat. Nanotechnol., (manuscript in progress)

L. J. Hubble & C. L. Raston, ‘The supramolecular assembly and metal cation interaction of p-phenyl sulfonated calix[8]arene - a water-soluble fluoroionophore’, targeted at JACS, (manuscript in progress)

L. J. Hubble, T. E. Clark, M. Makha & C. L. Raston, ‘Selective diameter uptake of single-walled carbon nanotubes in water using phosphonated calixarenes and ‘extended arm’ sulfonated calixarenes’, J. Mater. Chem., 2008, 18, 5961-5966

L. J. Hubble & C. L. Raston, ‘Nanofibers of fullerene C60 through interplay of ball-and-socket supermolecules’, Chem. Eur. J., 2007, 13, 6755-6760

Conference Abstracts

L. J. Hubble, C. L. Raston & R. J. Watling, 2008, ‘The design, fabrication & testing of carbon nano-material chemiresistor sensors for the detection of low-vapour pressure analytes of national security interest’, Abstracts of Oral Papers, 19th International Symposium on the Forensic Sciences, Melbourne, VIC, Australia, October 6th-9th 2008, p.182

L. J. Hubble, C. L. Raston & R. J. Watling, 2008, ‘Development of carbon nanotube-based sensors for gaseous and vapour-phase analyte detection’, CD-ROM of Abstracts of Poster Papers, 2nd International Conference on Nanoscience and Nanotechnology, Melbourne, VIC, Australia, February 25th-29th 2008, pp. 137-138

X

L. J. Hubble, R. J. Watling & C. L. Raston, 2007, ‘Development of carbon nanotube-based sensors’, Abstracts of Oral Papers, 1st UK-US Conference on Chemical and Biological Sensors and Detectors, Bloomsbury, London, UK, April 22nd-24th 2007, p. O13

L. J. Hubble, C. L. Raston, C., 2006, ‘Nano-fibres based on a fullerene C60 ball-and-socket array’, Abstracts of Oral Papers, 1st International Conference on Nanoscience and Nanotechnology, Brisbane, QLD, Australia, July 2nd-7th 2006, pp. 105-106

L. J. Hubble & C. L. Raston, 2006, ‘Nano-fibres based on a fullerene C60 ball and socket array’, Abstracts of Oral Papers, Australian Research Network for Advanced Materials Annual Conference, Brisbane, QLD, Australia, June 28th-30th 2006, p. 95

L. J. Hubble, C. L. Raston & R. J. Watling, 2006, ‘Development of carbon nanotube-based sensors’, Abstracts of Oral Papers, 18th International Symposium on the Forensic Sciences, Perth, WA, Australia, April 2nd-7th 2006, pp. 85-86

Provisional Patent

L. J. Hubble & C. L. Raston, ‘Methods for selectively separating carbon nanotubes’, 2008, Australian Provisional Patent AU,2008,905,620

XI

TABLE OF CONTENTS

CANDIDATE DECLARATION ...... I

DECLARATION FOR THESIS CONTAINING PUBLISHED WORK AND WORK PREPARED FOR PUBLICATION ...... II

ABSTRACT...... IV

ACKNOWLEDGEMENTS...... VI

DETAILS OF PUBLICATIONS, CONFERENCE ABSTRACTS AND PATENTS ...... X

TABLE OF CONTENTS ...... XII

LIST OF FIGURES ...... XVIII

LIST OF TABLES ...... XXX

LIST OF SCHEMES ...... XXXI

ABBREVIATIONS ...... XXXII

1 General introduction ...... 2 1.1 The chemistry of carbon ...... 2 1.1.1 Diamond...... 4 1.1.2 Graphite...... 4 1.2 The discovery of new forms of carbon ...... 5 1.2.1 Carbon nanotubes...... 7 1.2.1.1 Multi-walled carbon nanotubes...... 7 1.2.1.2 Single-walled carbon nanotubes ...... 8 1.2.2 Graphene ...... 10 1.3 Structure and properties of SWCNTs ...... 11 1.3.1 Electronic properties...... 15 1.3.2 Chemical and electrochemical properties ...... 19 1.3.3 Other properties...... 21 1.3.3.1 Optical and optoelectronic ...... 22 1.3.3.2 Mechanical and electromechanical ...... 22 1.3.3.3 Magnetic and electromagnetic ...... 23

XII

1.3.3.4 Thermal ...... 24 1.4 SWCNT applications...... 24 1.4.1 Carbon nanotube gas and vapour sensors ...... 25 1.4.1.1 Different CNT-sensor configurations ...... 26 1.4.1.2 Mechanisms of action ...... 28 1.4.1.3 Methodologies to increase analyte specificity and sensitivity ...... 31 1.4.1.4 Overview of analyte target areas and the current literature ...... 35 1.4.1.5 Advantages over other gas and vapour sensing technologies ...... 36 1.5 Supramolecular chemistry...... 37 1.5.1 Molecular recognition and host-guest chemistry ...... 38 1.5.2 Self-assembly ...... 40 1.6 Calixarenes...... 41 1.6.1 Synthesis and mechanism of formation ...... 41 1.6.2 Calixarene structural conformations ...... 44 1.7 Calixarene–fullerene solid-state interactions ...... 46 1.7.1 Calix[3,4,6,8]arenes ...... 46 1.7.2 Calix[5]arenes ...... 52 1.8 Functionalisation of calixarenes – water solubility...... 54 1.9 Applications of water-soluble calixarenes ...... 55 1.10 Specific aims and structure of the thesis...... 56

2 Diameter-selective sorting of SWCNTs in aqueous media...... 60 2.1 Introduction...... 60 2.2 Experimental ...... 62 2.2.1 Materials...... 62 2.2.2 Methods...... 63 2.3 Results and discussion...... 65 2.3.1 Raman spectroscopy...... 66 2.3.2 UV-visible spectrophotometry...... 76 2.3.3 Transmission electron microscopy...... 77 2.3.4 Atomic force microscopy...... 79 2.3.5 Host-guest directed SWCNT aligned arrays...... 81 2.3.6 Electrical characterisation through the development of TFTs ...... 81 2.4 Conclusion...... 83

XIII

3 Design, fabrication and verification of a gas and vapour analyte delivery system for sensor testing...... 85 3.1 Introduction...... 85 3.1.1 Gas delivery and dilution methodologies...... 86 3.1.2 Vapour generation methodologies targeting CWAs ...... 86 3.1.2.1 Indirect contact vapour generation methods ...... 87 3.1.2.2 Direct contact vapour generation methods...... 87 3.2 Experimental ...... 90 3.2.1 Materials...... 90 3.2.2 Methods...... 90 3.3 Analyte delivery system - ‘Phase-I’ prototype ...... 91 3.3.1 Chemical vapour generator and delivery system ...... 91 3.3.2 Sensor testing chamber ...... 95 3.3.3 Prototype problems ...... 98 3.4 Analyte delivery system - ‘Phase-II’ prototype ...... 99 3.4.1 Chemical vapour generator and delivery system ...... 99 3.4.2 Syringe pump method ...... 99 3.4.3 Prototype problems ...... 106 3.5 Analyte delivery system - ‘Phase-III’ prototype...... 107 3.5.1 Chemical vapour delivery system...... 107 3.5.2 Vapour dilution system ...... 107 3.5.3 System noise...... 118 3.5.4 Sensor test chamber position verification ...... 120 3.6 Conclusions...... 120

4 SWCNT-based chemiresistor sensor testing for compounds of national security interest ...... 123 4.1 Introduction...... 123 4.2 Experimental ...... 126 4.2.1 Materials...... 126 4.2.2 Methods...... 126 4.3 Results and discussion...... 134 4.3.1 Optimisation...... 134 4.3.1.1 IDE device platform...... 134 4.3.1.2 SWCNT chemiresistor device...... 136

XIV

4.3.1.2.1 Drop-casting SWCNT dispersion ...... 136 4.3.1.2.2 Direct CNT growth on IDE platform...... 138 4.3.1.3 SWCNT chemiresistor sensor optimisation...... 139 4.3.1.3.1 Electrical monitoring parameters ...... 139 4.3.1.3.2 Response dependence on baseline resistance...... 140 4.3.1.3.3 Sensor environment...... 141 4.3.2 Metallic versus semiconducting SWCNTs in gas and vapour sensing.143 4.3.3 CWA analyte exposure trials ...... 144 4.3.3.1 Single interdigitated electrode sensor ...... 144 4.3.3.1.1 DMMP...... 144 4.3.3.1.2 DIMP...... 146 4.3.3.1.3 Phosgene ...... 148 4.3.3.1.4 Phosgene response based on current gradient change...... 150 4.3.3.2 Multiple-array sensors...... 152 4.3.3.2.1 DMMP...... 152 4.3.3.2.2 DIMP...... 153 4.3.3.3 Single versus multiple-array ...... 155 4.3.3.4 Defect-induced versus untreated SWCNTs ...... 156 4.3.3.5 Interference plots...... 156 4.3.3.6 Implications in relation to current exposure limit guidelines...... 159 4.4 Conclusion...... 160

5 Progress towards novel carbon nano-materials and interesting application areas...... 163 5.1 Introduction...... 163 5.2 Experimental ...... 164 5.2.1 Materials...... 164

5.2.1.1 C60 nano-fibres and high temperature annealed analogues...... 164 5.2.1.2 Rings of helical SWCNT bundles...... 165 5.2.1.3 Square-geometry carbon nano-fibres ...... 165 5.2.2 Methods...... 165

5.2.2.1 C60 nano-fibres and high temperature annealed analogues...... 165 5.2.2.2 Rings of helical SWCNT bundles...... 169 5.2.2.3 Square-geometry carbon nano-fibres ...... 169 5.3 Results and discussion...... 170

XV

5.3.1 Nano-fibres of fullerene C60...... 170 5.3.1.1 Supramolecular complex formation and characterisation...... 170 5.3.1.2 Thermogravimetric analysis and the development and characterisation of high-temperature analogues...... 178 5.3.1.3 Chemotherapeutic assessment...... 184 5.3.2 Helical rings of SWCNT bundles ...... 187 5.3.2.1 Transmission electron microscopy ...... 188 5.3.3 Square-geometry carbon nano-fibres...... 191 5.3.3.1 Scanning electron microscopy ...... 191 5.4 Conclusion ...... 193

6 A water-soluble fluoroionophore: p-phenyl sulfonated calix[8]arene...... 196 6.1 Introduction...... 196 6.2 Experimental ...... 198 6.2.1 Materials...... 198 6.2.2 Methods...... 198 6.3 Results and discussion...... 201 6.3.1 pH dependent fluorescence studies...... 201 6.3.2 Supramolecular interaction/conformational studies...... 204 6.3.3 Metal cation interaction studies ...... 212 6.3.3.1 Time effects on the metal cation sensor and metal addition fluorescence kinetics...... 212 6.3.3.2 Metal binding effects on p-phenyl-sulfonato-calix[8]arene...... 215 6.4 Conclusion ...... 226

7 Conclusions and future studies...... 229 7.1 Conclusions...... 229 7.2 Future studies...... 232 7.2.1 SWCNT post-growth processing towards diameter-selectivity...... 232 7.2.2 CWA detection through the development of SWCNT-based chemiresistor sensor technology...... 233 7.2.3 Fabrication of functional all-carbon architectures ...... 236 7.2.4 Development of novel water-soluble fluoroionophores...... 238

8 References...... 241

9 Appendices...... 271

XVI

10 Publications...... 276

XVII

LIST OF FIGURES Figure 1.1 Bonding structures of diamond, graphite, and fullerene allotropes of carbon, with σ-orbitals in grey and π-orbitals in blue; note that a π-orbital exists in out-of-plane in sp2 bonding but has been omitted for clarity...... 3 Figure 1.2 Different allotropes of carbon, a) diamond, b) graphite, c) lonsdaleite, d) fullerene C60, e) fullerene C540, f) fullerene C70, g) amorphous carbon, h) single-walled carbon nanotube (SWCNT)...... 3 Figure 1.3 A side-by-side comparison of the truncated icosahedral structure of a molecular model of fullerene C60 and a soccerball...... 5

Figure 1.4 a) Schematic of the electron density of fullerene C60, and b) the C60 isosurface of ground state electron density, calculated with DFT using the CPMD code. Adapted from [15]...... 6 Figure 1.5 Electron micrographs of the microtubules of graphitic carbon which Iijima discovered, more commonly referred to as MWCNTs. A cross-section of each nanotube is illustrated; a) tube consisting of five graphitic sheets (6.7 nm dia.), b) two-sheet tube (5.5 nm dia.), and a seven-sheet tube (6.5 nm dia.) which has the smallest hollow core (2.2 nm). Adapted with permission from Macmillian Publishers Ltd: Nature[19], © 1993...... 8 Figure 1.6 Electron micrographs of SWCNTs, at a) low magnification and b) high magnification; produced through the original modified carbon-arc evaporation preparation by Bethune et al. Adapted with permission from Macmillian Publishers Ltd: Nature[21], © 1993...... 9 Figure 1.7 a) Molecular model of a section of (9,9) armchair SWCNT, b) electron density map of the first layer of the nanotube, c) electron density map of the second layer of the nanotube. Adapted with permission from [114], © 2001 Elsevier...... 12 Figure 1.8 Conceptually, a SWCNT is formed by rolling up graphene along a specific chiral vector to form a cylinder. Three different types of SWCNTs are presented. Adapted with permission from [121], © 2005 John Wiley & Sons, Inc...... 13 Figure 1.9 Schematic indicating the correlation between chiral indices (n,m) and the electronic properties of the resulting SWCNT at room temperature. CNTs with n = m (armchair nanotubes) and those with n – m = 3q are metallic at room temperature (labelled green). CNTs with n - m = 3q + 1 (labelled pink) and n – m = 3q + 2 (labelled purple) are semiconductors with a bandgap that varies inversely with tube diameter. Reproduced with permission from [122], © 2006 courtesy of Mike Arnold...... 13

XVIII

Figure 1.10 a) Schematic of two SWCNTs with nearly identical diameters but with differing chiral indices, b) schematic of two SWCNTs with identical chiral vectors (Ch) but with right- and left-handed helicity. Reproduced with permission from [122], © 2006 courtesy of Mike Arnold...... 14 Figure 1.11 a) TEM images of SWCNTs a1 (left) and a2 (right) and their fast Fourier transform (FFT) images (inset); b) schematic illustrations of the structures of (10,10) and (11,8) SWCNTs (left and right panels, respectively; c) simulated TEM images based on the models in b) and their FFT images (inset), characteristic bright spots in the FFT image are indicated by red triangles in a) and c) scale bar = 1 nm; d) STM image of the resolved atomic resolution of a SWCNT. Figures a), b), and c) adapted with permission from [124], © 2008 American Chemical Society, and Figure d) adapted with permission from [125], © 1999 American Institute of Physics...... 15 Figure 1.12 The band structure (top) and the BZ (bottom) of graphene. The circumferential quantisation demonstrated by the nanotube leads to the formation of a set of discrete energy sub-bands for each nanotube (red parallel lines). The relation of these lines to the band structure of graphene determines the electronic structure of the nanotube. If the lines pass through the K or K’ points, the nanotube is a metal: if they do not (as in this Figure), the nanotube is a semiconductor. Adapted with permission from Macmillian Publishers Ltd: Nature Nanotechnology[127], © 2007...... 16 Figure 1.13 Electronic properties of two different SWCNTs, a) an armchair (5,5) nanotube exhibits metallic behaviour, b) a zigzag (7,0) nanotube is a small bandgap semiconductor (no charge carriers in the DOS at the Fermi energy). Adapted with permission from [131], © 2002 American Chemical Society...... 17 Figure 1.14 a) A bundle of SWCNTs with different sets of tube diameters, b) a SWCNT rope consisting of metallic armchair (10,10) nanotubes. Adapted with permission from [122], © 2006 courtesy of Mike Arnold, and permission from [133], © 2002 John Wiley & Sons, Inc., respectively...... 18 Figure 1.15 Schematic of a Stone-Wales defect (or 7-5-5-7) on the side-wall of a SWCNT. Adapted with permission from [121], © 2005 John Wiley & Sons, Inc...... 20 Figure 1.16 A schematic depicting different SWCNT functionalisation strategies, a) covalent side-wall functionalisation, b) covalent defect and open-tip functionalisation, c) non-covalent exohedral functionalisation with surfactants, d) non-covalent exohedral functionalisation with polymers, and e) endohedral functionalisation with, for example, [133] C60. Adapted with permission from , © 2002 John Wiley & Sons, Inc...... 21

XIX

Figure 1.17 a) Electrical response of an NT-FET at VG = 0 V to 1% NH3, and b) at [223] VG = +4 V to 200 parts-per-million (ppm) NO2. Reprinted with permission from , © 2000 AAAS...... 26 Figure 1.18 a) Optical micrograph illustrating the chemicapacitor configuration; inset is an AFM image of the random network of SWCNTs, and b) measured relative capacitance change of the SWCNT chemicapacitor in response to repeated 20 s doses of dimethylformamide at varying concentrations noted in the figure. Reprinted with permission from [235], © 2005 AAAS...... 27 Figure 1.19 a) Schematic of a NT-FET device configuration, and b) schematic of a chemiresistor device configuration, with no gate-drain potential. Adapted with permission from [238], © 2006 American Chemical Society, and permission from [151], © 2008 John Wiley & Sons, Inc., respectively...... 28 Figure 1.20 a) Schematic of the work function dependent SB caused by the metal-SWCNT work function mismatch, and b) I-VG transfer characteristic of NT-FET devices composed of semiconducting SWCNTs contacted by metals with different work functions; the devices were operated at VSD = -0.5 V. Adapted with permission from [240], © 2005 American Chemical Society...... 29 Figure 1.21 a) Optical image illustrating three CNT-based devices after micro-spotting with droplets of poly(ethylene imine) (PEI) and/or Nafion® solution, and b) AFM image of Pd nanoparticles electrochemically deposited on the surface of SWCNTs forming the [273] active material for a H2 gas sensor. Adapted with permission from , © 2003 American Chemical Society, and permission from [256], © 2007 American Institute of Physics, respectively...... 33 Figure 1.22 A flow chart linking traditional molecular chemistries and developing supramolecular chemistries. Adapted with permission from [286], © 1989 VCH Publishers...... 38 Figure 1.23 Overview of the molecular mitosis hypothesis experiments. The white- and black-filled circles represent protonated and deuterated residues respectively. Adapted with permission from [341], © 1999 American Chemical Society...... 44 Figure 1.24 Structural conformations of calix[4]arene derivatives...... 45

Figure 1.25 Schematic of the diversity of fullerene C60 ordering into one-, two-, and three-dimensional arrays. Host molecules have been omitted for clarity. Adapted with permission from [358], © 2003 John Wiley & Sons, Inc...... 47 Figure 1.26 Top view projection down the c axis of the extended structure of fullerene

C60 and calixarene layers in the (C60)2(p-benzylcalix[4]arene), with red dotted lines

XX indicating C-H···π interactions. Adapted with permission from [370], © 2006 The Royal Society of Chemistry...... 48 Figure 1.27 (top) Elucidated solid-state structure for

(C60)3(p-benzylcalix[6]arene)(toluene)4.25, and (bottom) the same structure with the calixarene and toluene molecules removed to indicate the extent of the three-dimensional C60 network and the virtual porosity. Fullerenes are coloured purple, toluene molecules are in blue; calixarenes are shown in ball-and-stick projected down the c axis. Adapted with permission from [371], © 2008 American Chemical Society...50

Figure 1.28 (a) Structure of complex (C60)3(p-benzylcalix[6]arene)(toluene)4.25, (b) the same structure with the calixarene and toluene molecules; both images show the pinched double cone conformation of the p-benzylcalix[6]arene. Adapted with permission from [371], © 2008 American Chemical Society...... 51

Figure 1.29 (a) STM image of the C60/calix[8]arene derivative, and (b) the proposed structural model of the ordered array resulting in the observed STM image; scale is given in nm. Adapted with permission from [374], © 2003 John Wiley & Sons, Inc. ....51

Figure 1.30 Encapsulation of an individual fullerene C60, a)

(C60)@(p-benzylcalix[5]arene)2 where C60 is represented in space-filling form with the calixarene molecules in stick form, and b) (C60)@(p-phenylcalix[5]arene)2 where C60 and calixarene molecules are represented as space-filling form. Adapted with permission from [367], © 1999 John Wiley & Sons, Inc., and permission from [381], © 2002 The Royal Society of Chemistry, respectively...... 52

Figure 1.31 a) Crystal structure of the (C60)(calix[5]arene)(toluene) complex illustrating the columnar, slightly helical zigzag fullerene array, and b) the extended crystal structure of the (C60)4(calix[5]arene)5(toluene)2 complex with the projection [358] parallel to the 5-fold, Z-shaped columns of C60. Adapted with permission from , © 2003 John Wiley & Sons, Inc., and permission from [382], © 2002 American Chemical Society, respectively...... 53 Figure 1.32 Water-soluble p-substituted calixarenes ranging from shallow cavity (n = 4, 5, 6, 8) to deeper cavity 'extended arm' sulfonated calix[n]arenes (n = 4, 6, 8) to phosphonic acid calix[n]arene (n = 4, 5, 6, 8)...... 54 Figure 2.1 Schematic of a SWCNT thin-film transistor (TFT) and structure. A 300 nm layer of SiO2 is deposited on a heavily doped Si wafer, which is functionalised by either an amine-terminated (A) or phenyl-terminated (B) silanes. A solution of SWCNTs is spin coated onto the surface, providing a connective pathway between the source (S)

XXI and drain (D) electrodes, with directionality of self-assembled monolayers monitored through AFM. Reprinted with permission from [406], © 2005 AAAS...... 60 Figure 2.2 Sorting SWCNTs using density gradient ultracentrifugation, with optical absorbance confirming selective diameter fractions which correlate to CNTs with either metallic or semiconducting electronic properties. Adapted with permission from [409] Macmillian Publishers Ltd: Nature Nanotechnology, © 2006...... 61 Figure 2.3 Principal calixarene structure...... 63 Figure 2.4 Photographs of a) a solution containing SWCNT/water (left) and a solution containing 2A/SWCNT/water (right), and b) supernatant of 2A/SWCNT/water system...... 66 Figure 2.5 Raman spectrum of the ‘blank’ aluminium foil, inset of the typical RBM frequency range. These ‘baseline’ frequencies were taken into account for the analysis of the as-received SWCNTs and supernatant residues. This also represents the spectrum obtained when analysing calixarene control samples...... 67 Figure 2.6 Raman spectrum of as-received SWCNT, inset of the RBM frequency range...... 67 Figure 2.7 a) Raman spectra obtained for the phosphonated calixarene systems, and b) Raman spectra highlighting the RBM bands...... 70 Figure 2.8 a) Raman spectra obtained for the sulfonated calixarene systems, and b) Raman spectra highlighting the RBM bands...... 71 Figure 2.9 UV-visible spectra of solutions of the p-phosphonated calixarenes (1A - 3A) and p-sulfonated calixarenes (4A - 4C)...... 76 Figure 2.10 UV-visible spectra of the p-phosphonated calixarenes (1A - 3A)/SWCNT supernatants and p-sulfonated (4A - 4C)/SWCNT supernatants...... 77 Figure 2.11 TEM image illustrating SWCNT networks coated with 1C...... 77 Figure 2.12 TEM image illustrating the individual SWCNT nature of the supernatant with the wrapping of 1C around the SWCNT surface...... 78 Figure 2.13 EDX spectrum of 2A/SWCNT supernatant...... 79 Figure 2.14 a) A three-dimensional topographical representation of a calixarene sheathed SWCNT...... 80 Figure 2.15 A height profile obtained from analysis of the calixarene sheathed SWCNT in Figure 2.14. Note the y-axis is in nm, and the x-axis is in µm...... 80 Figure 2.16 I-V curve illustrating the TFT properties of the 2A/SWCNT supernatant. Inset is an I-V curve of a commercial 2N5485 N-channel FET, with y-axis represented in µm. Measurements taken at VGS = 0 V...... 82

XXII

Figure 3.1 A schematic of a bubbler-based chemical vapour generator utilised upstream from a CNT-based SAW sensor testing chamber. Adapted with permission from [232], © 2005 Elsevier...... 88 Figure 3.2 A schematic of a spray atomisation system which is used for the vapour generation of CWA GB (sarin). Adapted with permission from [436], © US Military. ..89 Figure 3.3 Schematic of the ‘Phase-I’ prototype chemical vapour generator and delivery system, including sensor testing chamber and proposed measurement instrumentation...... 92 Figure 3.4 Photograph of the twin glass chemical bubblers with fritted glass caps and ¼ inch glass-to-Kovar® entry and exit ports...... 94 Figure 3.5 Photographs of the fabricated Teflon® sensor testing chamber from a) a side view, and b) a front view. The chamber is supported on a moulded Perspex bracket specifically fabricated to align the gas flow entry port with the upstream system...... 95 Figure 3.6 A side-by-side view of a) the PCB design for the test chamber, and b) the fabricated PCB with components installed; the 25-pin DB-type connector (top) and spring sockets containing TO-8 Headers (bottom)...... 96 Figure 3.7 A side-by-side view of a) the PCB design for the BNC connector break-out board, and b) the fabricated PCB with components installed; the 25-pin DB-type connector (top) and an array of BNC connectors (bottom)...... 97 Figure 3.8 Photograph of an open sensor testing chamber...... 98 Figure 3.9 Schematic of the ‘Phase II’ prototype chemical vapour generator and delivery system, including sensor testing chamber and proposed measurement instrumentation...... 99 Figure 3.10 A layer-by-layer schematic breakdown of the internal components and gas flow path of an Alicat gas MFC unit. Adapted with the permission of [441], © 2003 Alicat Scientific, Inc...... 101 Figure 3.11 Photograph of the quartz nebuliser mounted with the tip into a securing gland, with the gas flow feed wired onto the gas entry port and tubing for the introduction of a liquid analyte secured at the rear...... 102 Figure 3.12 A front-view photograph of the 'Phase-II' prototype chemical vapour generator and dilution system, including the sensor testing chamber. Note: in subsequent modifications the peristaltic pump (centre right) was replaced with a syringe pump...... 104 Figure 3.13 Schematic of the ‘Phase-III’ prototype vapour delivery system, including sensor testing chamber and measurement instrumentation...... 108

XXIII

Figure 3.14 Photograph of the syringe pump injection port sealed with a gas chromatography injection port septum (blue)...... 109 Figure 3.15 Photograph of the syringe pump mounted on a retractable, adjustable z-axis stage with the syringe needle piercing the septum and injecting a headspace sample into the vapour dilution system...... 111 Figure 3.16 Photograph of the detachable glass chemical bubbler secured in place through glass-to-Kovar® tube connectors and supported on a plastic die designed to allow back-pressure release, if necessary...... 111 Figure 3.17 Photograph of the in-house fabricated cyclonic mixer consisting of two gas flow line entry ports and a central ‘submerged’ exit route directed towards the sensor testing chamber...... 112 Figure 3.18 Photograph of the electrical measurement instrumentation utilised in the 'Phase-III' prototype which consists of an eDAQ potentiostat (top) and e-corder (bottom)...... 114 Figure 3.19 Photograph of the 'Phase-III' vapour dilution system, including the sensor testing chamber and notebook computer. Note that the electrical monitoring hardware is not captured in this image...... 115 Figure 3.20 Screen capture images of the FlowVision™ software graphical user interface (left), and the Pump Schedule graphical user interface (right). Note the functional parameters within each system are user-specified...... 117 Figure 3.21 A screen capture image of the EChem™ graphical user interface, with test parameters located to the right of screen, and the main window displaying sensor I profile (µA) with respect to time (s)...... 118 Figure 3.22 A sensor response curve for a SWCNT chemiresistor configuration monitored before and after attempts to lower the effect of electrical noise in the system...... 119 Figure 3.23 Sensor response curve for a Pd nanoparticle/SWCNT hybrid chemiresistor sensor upon exposure to H2 gas (5% to 9%), located at positions A and B in the sensor testing chamber...... 119 Figure 4.1 Typical acid-induced SWCNT surface defect sites, indicating nanotube tip and side-wall functionalisation, along with pentagon-induced structural defects. Adapted with permission from [133], © 2002 John Wiley & Sons, Inc...... 125 Figure 4.2 A schematic of the IDE which represents the basis of the sensor platform, scale bar represents 100 μm...... 128

XXIV

Figure 4.3 A SEM image of the random network of SWCNTs between the chemiresistor source and drain electrodes, scale bar is 1 µm...... 129 Figure 4.4 a) A schematic of the IDE acting as a dielectrophoresis device, and b) an optical image illustrating the alignment of CNTs on an electrode under polarised light. Reprinted with permission from [407], © 2003 AAAS...... 132 Figure 4.5 Optical images a) illustrating the entirety of the IDEs, with 100 µm scale bar, and b) illustrating an increased magnification image of the top right corner of a), with a 50 µm scale bar, of the first generation IDE platform using a glass emulsion photolithography mask coupled with in-house printing...... 134 Figure 4.6 Optical images a) illustrating the entirety of the IDEs, with 100 µm scale bar, and b) illustrating an increased magnification image of the top right corner of a), with a 50 µm scale bar, of the second generation IDE platform coupled with outsourced printing...... 135 Figure 4.7 Optical images a) illustrating the top right corner of the IDEs, with 200 µm scale bar, and b) illustrating an increased magnification image of the top right corner of a), with a 25 µm scale bar, of the third generation IDE platform developed using an outsourced Cr mask...... 135 Figure 4.8 SEM images of SWCNT density between the source and drain electrode for a) 1 x 0.02 µL addition (Device i), b) 2 x 0.02 µL addition (Device ii), c) 3 x 0.02 µL addition (Device iii), and d) 4 x 0.02 µL addition (Device iv). Scale bar is 1 µm...... 137 Figure 4.9 Device resistance with respect to different SWCNT densities on the IDE area, with error bars as standard deviations of the mean...... 137 Figure 4.10 SEM image of a) the carbon material growth over an IDE area, and 2) a higher magnification image detailing portions of wire-like materials. Scale bars are 100 µm and 1 µm respectively...... 138 Figure 4.11 A higher magnification SEM image of the wire-like structures generated on the IDE substrate...... 139 Figure 4.12 Effect of sampling potential density on chemiresistor response to saturated vapour of ethanol, error bars as standard deviations of the mean...... 140 Figure 4.13 Effect of SWCNT density on chemiresistor response to saturated vapour of ethanol, error bars as standard deviations of the mean...... 141 Figure 4.14 Chemiresistor sensor baseline-I in response to change in relative humidity...... 142 Figure 4.15 DMMP exposure curve for the metallic SWCNT-enriched device involving dielectrophoretic separation...... 143

XXV

Figure 4.16 Single IDE sensor response curve to DMMP exposure of increasing concentration...... 145 Figure 4.17 Single IDE sensor calibration curve for DMMP vapour exposure with increasing concentration, with error bars as standard deviations of the mean...... 145 Figure 4.18 Single IDE sensor response curve to DIMP exposure of increasing concentration...... 146 Figure 4.19 Single IDE sensor calibration curve for DIMP vapour exposure with increasing concentration, error bars as standard deviations of the mean...... 147 Figure 4.20 Single IDE sensor response curves for phosgene exposure of increasing dilution factors from a saturated vapour...... 148 Figure 4.21 Single IDE sensor maximum response curve data for phosgene vapour exposure with increasing concentration, error bars as standard deviations of the mean...... 149 Figure 4.22 Single IDE sensor calibration curve for phosgene vapour exposure with increasing concentration, error bars as standard deviations of the mean. Higher concentration response was not possible due to sensor response saturation...... 150 Figure 4.23 Single IDE sensor baseline gradient change for phosgene vapour exposure with increasing concentration...... 151 Figure 4.24 Single IDE sensor calibration curve for baseline-I gradient change upon phosgene vapour exposure with increasing concentration. The data is fitted with a growth curve fitting function to represent the gradient trend...... 151 Figure 4.25 Multiple-array sensor response curve to DMMP exposure of increasing concentration...... 152 Figure 4.26 Multiple-array sensor calibration curve for DMMP vapour exposure with increasing concentration, error bars as standard deviations of the mean...... 153 Figure 4.27 Multiple-array sensor response curve to DIMP exposure of increasing concentration...... 154 Figure 4.28 Multiple-array sensor calibration curve for DIMP vapour exposure with increasing concentration, error bars as standard deviations of the mean. Note: the sensor response for the lowest concentration analysed was considered an outlier and not represented in the plot...... 155 Figure 4.29 Sensor response of the single IDE sensor geometry to common organic solvents and a range of halogenated compounds. Each analyte was delivered at 10% saturated vapour concentration...... 157

XXVI

Figure 4.30 Sensor response of the single IDE sensor geometry to common organic solvents and a range of halogenated compounds. Each analyte was delivered at 10% saturated vapour concentration...... 158 Figure 5.1 Optical micrograph a), and TEM image b) of the 6 complex, the latter revealing the linear striations of fullerene C60 within bundles of nano-fibres...... 170 Figure 5.2 A schematic representation of the formation of the 1:1 supermolecule, 6.171 -2 Figure 5.3 UV-visible spectra for a C60 (9.85 x 10 mmol) stock solution with increasing concentrations of 5, spectra recorded in toluene at 25 °C...... 172 Figure 5.4 IR spectrum of the product of the decomplexation of 6 in dichloromethane, additional absorbance peaks are attributed to residual dichloromethane.[354] ...... 173

Figure 5.5 UV-visible spectra for i) a C60 standard, ii) nano-fibre complex from pure

C60, and iii) nano-fibre complex from fullerene soot filtrate, all spectra recorded in toluene at 25 °C...... 174 Figure 5.6 CP-MAS solid state 13C NMR spectra for the 6 complex, with (blue) and without (red) magic angle spinning...... 174 Figure 5.7 PXRD pattern of the 6 complex...... 176

Figure 5.8 a) Proposed helical array of C60 columns; labels denote a tri-angular set intra-plane fullerenes, b) a line structure representation of a) with A - C fullerenes centroid···centroid dihedral angle illustrated, and c) C60 column rotated 45° about the horizontal x-axis in a)...... 177

Figure 5.9 Proposed calixarene packing arrangement around a C60 column, a) with the same orientation as Figure 5.8a, and b) with the same orientation as Figure 5.8c; based on Accelrys software.[15] ...... 178 Figure 5.10 TGA of the nano-fibres of 6, under both air and argon atmospheres...... 179 Figure 5.11 a) SEM image of as-synthesised bundles of the nano-fibres, and SEM images of bundles of the nano-fibres treated under argon at b) 400 °C, c) 700 °C, and d) 800 °C; all scale bars represent 10 μm...... 180 Figure 5.12 BET surface area correlated to percent weight loss for as-synthesised and high temperature analogues annealed under argon, with associated annealing temperatures specified...... 181 Figure 5.13 I-V curve and resistance curve for the 800 °C argon annealed analogue.182 Figure 5.14 Raman spectra of the high temperature analogues and as-synthesised product...... 183 Figure 5.15 Fluorescence microscopy image of an untreated control sample of HeLa cells transfected with LC3-fused to GFP...... 185

XXVII

Figure 5.16 Fluorescence microscopy image of HeLa cells transfected with LC3-fused to GFP that have been a) treated with the 6 complex at a concentration of 2.5 µg mL-1, -1 and b) treated with nano-C60 at a concentration of 0.5 µg mL . Note that samples were treated for 24 h before being subjected to fluorescence microscopy...... 185 Figure 5.17 Percentage of cell death in drug-resistant MCF-7 breast cancer cells, illustrating a comparative study between different incubation methodologies...... 186 Figure 5.18 TEM image of a ring formed from a helical SWCNT bundle, with no seam observable...... 188 Figure 5.19 TEM image of the helical ring structure indicating the inherent wall structure (centre), and also periodicity of the wall projections...... 189 Figure 5.20 TEM image of a separate ring of helical SWCNT bundles, which slightly diverts from a circular array...... 190 Figure 5.21 TEM image of a) a carbon map, and b) a thickness map, of the asymmetric ring of helical SWCNT bundles...... 190 Figure 5.22 A range of different carbon materials generated through C-CVD processes using a Ni catalyst these structures include a) vermicular or spiral shaped carbon fibres; scale bar 10 µm, b) large carbon fibres; scale bar 1 µm, c) circular carbon nano-fibres; scale 20 nm, and d) hexagonal fibres, in this particular case with an open-tip, where presumably a Ni catalyst particle was once located; scale bar 200 nm...... 192 Figure 5.23 A SEM image of a square-geometry CNF, grown through C-CVD processes; scale bar 200 nm...... 193 Figure 6.1 The structural diagram of 4C...... 201 Figure 6.2 Contour plot of the fluorescence spectra of 4C at pH 6.5, with the fluorescence intensity scale (top right)...... 202

Figure 6.3 a) Dependent series of emission spectra (λex = 335 nm) for 4C, and b) the respective contour plot with fluorescence intensity scale (top right)...... 202

Figure 6.4 a) Dependent series of emission spectra (λex = 423 nm) for 4C, and b) the respective contour plot with fluorescence intensity scale (top right)...... 203 Figure 6.5 Conductivity profile with increasing concentration of 4C...... 204

Figure 6.6 Emission profiles (λex = 335 nm) for solutions of increasing concentration of 4C; concentrations below and above the CMC utilised at detector voltage 800 V. .205 Figure 6.7 Zeta potential measurements for solutions of 4C (1 mM) over a pH range...... 206 Figure 6.8 Particle diameter through DLS measurements for solutions of 4C (1 mM) over a pH range...... 207

XXVIII

Figure 6.9 Tapping mode AFM characterisation of solid-state structures of 4C on a mica substrate at a) pH 2, b) pH 6, and c) pH 13. Representative three-dimensional topographical images correlating to contour plots are also shown (right)...... 208 Figure 6.10 A schematic of the lower rim hydroxyl group, hydrogen-bonding directed intramolecular excimer formation. Note for clarity only selected fragments of the principal calix[8]arene structure have been illustrated...... 210 Figure 6.11 A plot of the emission line fluorescence intensities of 4C with increasing acetonitrile component...... 211

Figure 6.12 Fluorescence intensity at λem = 513 nm (λex = 335 nm) for a freshly prepared solution of 4C (40 µM) a) monitored over a 42 h period, and b) in the first 7.5 h...... 213 Figure 6.13 Long-term fluorescence intensity monitoring of a solution of 4C...... 214

Figure 6.14 Real-time emission line monitoring (λex = 335 nm) of a solution of 4C (40 μM), the arrow indicates the addition of the Co2+ metal cation solution (20 μM). 215 Figure 6.15 Quenching ratio of a solution of 4C (40 μM) with the addition of a range of different metal cations (2 molar equiv.)...... 216 Figure 6.16 (Left) Variation of the fluorescence spectra of 4C (40 µM) upon 2+ 2+ 2+ 2+ + 3+ successive addition of a) [UO2] , b) Pb , c) Co , d) Cu , e) Ag , and f) In at pH 6.5, (Right) quenching of the fluorescence intensity of 4C (40 µM) upon successive 3+ addition of the respective metal cations at pH 6.5; in the In plot, the y-axis is I/Io due to enhancement of the fluorescence intensity...... 219 Figure 6.17 Quenching of the fluorescence intensity of 4C (40 µM) based on successive addition of molar equivalents of the metal cations; in the In3+ plot, the y-axis is I/Io due to enhancement of the fluorescence intensity...... 220 2+ Figure 6.18 Plot of Io/ΔI versus 1/[Co ], with a linear regression function fitted, and regression line extrapolated to indicate the y-intercept...... 221 Figure 6.19 Double-logarithmic plot for the quenching of 4C fluorescence intensity following the addition of Co2+ (2 μM to 50 μM), with a linear regression fitting function...... 222 Figure 6.20 Calibration curves for relative fluorescence intensity change versus metal cation concentration for 4C (40 μM), appropriate fitting functions and associated 2+ 2+ 2+ 2+ + correlation coefficients are described for a) [UO2] , b) Pb , c) Co , d) Cu , e) Ag , and f) In3+; a red datum point indicates an outlier not utilised for calibration curve determination; y-error bars are displayed...... 224

XXIX

LIST OF TABLES Table 1.1 Comparison of the mechanical properties of CNTs, graphite and steel. Adapted with permission from [160], © 1998 World Scientific Publishing Co...... 23 Table 1.2 'One pot' synthesis route yields for a variety of para-substituted calixarenes. Adapted with permission from [301], © 1995 John Wiley & Sons, Inc...... 42 Table 2.1 Notation for calix[n]arene systems...... 64

Table 2.2 Raman spectral G-band frequency shifts and the ID/IG ratios for the residues of the calixarene/SWCNT supernatants...... 68 Table 2.3 Calculated SWCNT diameters present based on RBM frequencies in the Raman spectra...... 73 Table 2.4 Observed RBM bands in the as-received SWCNTs and calixarene system supernatants with topographical and electronic assignment...... 75 Table 4.1 Calculated concentration values for the analytes included in the interference study...... 159

Table 6.1 Calculated Kb and n values of the metal cations with 4C...... 222

XXX

LIST OF SCHEMES Scheme 1.1 General base catalysed synthesis scheme for calixarene formation...... 41 Scheme 1.2 Base catalysed mechanism for the formation of linear oligomers prior to cyclisation. Adapted with permission from [300], © 1989 The Royal Society of Chemistry...... 43 Scheme 1.3 A new five-step synthesis of p-phosphonic acid calix[n]arene, n = 4, 5, 6, 8. Adapted from [393, 394] ...... 55

XXXI

ABBREVIATIONS A-C-CVD Alcohol-catalytic-chemical vapour deposition ac Alternating current

AFM Atomic force microscopy

BET Brunauer, Emmett and Teller

BZ Brillouin zone

C-CVD Catalytic-chemical vapour deposition

CHEF Chelation enhanced fluorescence

CHEQ Chelation enhanced quenching

CMC Critical micelle concentration

CNF Carbon nano-fibre

CNT Carbon nanotube

CTV Cyclotriveratrylene

CVD Chemical vapour deposition

CWA Chemical warfare agent

DCM Dichloromethane

DIMP Diisopropylmethylphosphonate

DLS Dynamic light scattering

DMEM Dulbecco’s Modified Eagle’s Medium

DMF N,N-Dimethylformamide

DMMP Dimethylmethylphosphonate

DNT Dinitrotoluene

DOS Density-of-states

Dox Doxorubicin

DWCNT Double-walled carbon nanotube

EDX Energy dispersive X-ray

ESR Electron spin resonance

XXXII eT Electron transfer

ET Energy transfer fcc Face centered cubic

FCS Fetal calf serum

FET Field-effect transistor

FFT Fast Fourier transform

FT-IR Fourier-transform infrared

FWHM Full-width half maximum

GFP Green fluorescent protein hcp Hexagonal close-packed

HiPco High pressure CO disproportionation

HR-STM High resolution-scanning tunneling microscopy

HR-TEM High resolution-transmission electron microscopy

HTM Heavy and transition metal

ID Internal diameter iPr Isopropyl

IDE Interdigitated electrode

ISFET Ion-selective field-effect transistor

LA-CVD Laser-assisted-chemical vapour deposition

LC3 Protein 1 light chain 3

LOD Limit of detection

LOQ Limit of quantitation

LTEL Long-term exposure limit

MEMS Microelectromechanical system

MFC Mass flow controller

MOS Metal oxide semiconductor

MWCNT Multi-walled carbon nanotube

XXXIII

NMR Nuclear magnetic resonance

NT-FET Nanotube field-effect transistor

OD Outer diameter

PABS Poly(m-aminobenzene sulfonic acid)

PANI Polyaniline

PBS Phosphate buffered saline

PCA Principal component analysis

PE-CVD Plasma-enhanced-chemical vapour deposition

PEI Poly(ethylene imine)

PET Poly(ethylene terephthalate)

PIB Poly(isobutylene)

PMMA Poly(methylmethacrylate) ppb Parts-per-billion ppm Parts-per-million ppt Parts-per-trillion

PPy Polypyrrole

PXRD Powder X-ray diffraction

RBM Radial breathing mode

RCF Relative centrifugal force

RMS Root-mean-squared

RPM Revolutions per minute

SAW Surface acoustic wave

SB Schottky barrier

SE Solid electrolyte

SEM Scanning electron microscopy ss-DNA Single strand-deoxyribonucleic acid

STM Scanning tunneling microscopy

XXXIV

SWCNT Single-walled carbon nanotube tBu tert-Butyl

TEM Transmission electron microscopy

TFT Thin-film transistor

TGA Thermogravimetric analysis tPentyl 1,1-dimethylpropyl tOctyl 1,1,3,3-tetramethylbutyl

UHP Ultra-high purity

UV-Vis-NIR Ultraviolet-visible-near-infrared

VGCF Vapour-grown carbon fibre

VHS van Hove singularities

VOC Volatile organic compound

XRD X-ray diffraction

XXXV

Chapter 1

General Introduction

Chapter 1

1 General introduction

The major concepts in this thesis relate to utilisation of carbon nano-materials and harnessing key manipulations at the molecular level through supramolecular interactions, to allow access to new materials and technology, in an attempt to provide solutions for current ‘real-world’ issues. This thesis explores a myriad of topical research areas including the supramolecular assembly of fullerene C60 into novel materials with interdisciplinary applications, novel post-growth treatments for CNT purification and separation, the fabrication of CNT-based chemiresistor sensors for the detection of phosgene gas with potential military prospects and the development of an entirely water-soluble fluroionophore utilising a higher-order sulfonated calixarene system. The following general introduction will introduce the reader to key developments in the areas necessary to assess the contribution made by the subsequent research described in this thesis.

1.1 The chemistry of carbon

The properties of carbon that arise from the different bonding structures will be discussed in order to understand the full versatility of this element and to provide valuable insight into the generation of new structures. Carbon has an atomic number of six and is the lightest atom in group 14 of the Periodic Table. The other elements in this group, Si, Ge, Sn, and Pb, have sp3 bonding in their cubic solid ground states, with condensed phase carbon capable of multiple bonding types leading to materials with varying degrees of dimensionality.[1] The carbon atom has an electronic configuration of 1s22s2p2,[2, 3] owing to its tetravalent nature, with the outer four electrons being able to fill the sp3 (tetrahedral) or sp2 (trigonal planar) as well as the sp (linear) hybrid orbital.[2, 4] The orbital filled will have a direct effect on the type of all-carbon allotrope produced (Fig. 1.1), and consequently the physical and chemical properties of the allotrope, making carbon one of the most versatile elements that exists on earth.[5]

Until recently it was thought that carbon only existed as the two naturally occurring allotropes, diamond and graphite. While these two allotropes will be discussed in this thesis, the breakthrough discovery of the fullerene allotrope family (Fig. 1.2) will be discussed in greater detail as they have far more relevance to this thesis.

2 Chapter 1

Figure 1.1 Bonding structures of diamond, graphite, and fullerene allotropes of carbon, with σ-orbitals in grey and π-orbitals in blue; note that a π-orbital exists in out-of-plane in sp2 bonding but has been omitted for clarity.

Figure 1.2 Different allotropes of carbon, a) diamond, b) graphite, c) lonsdaleite, d) fullerene C60, e) fullerene C540, f) fullerene C70, g) amorphous carbon, h) single-walled carbon nanotube (SWCNT).

3 Chapter 1

1.1.1 Diamond

The bonding configuration of diamond consists of the four valance electrons of each carbon atom occupying the sp3 hybrid orbital as illustrated in Figure 1.1. This configuration allows the covalent bonding of four additional carbon atoms at tetrahedral angles, creating four equivalent sigma (σ) bonds, propagation of this structure creates a three-dimensional material with nearly isotropic properties.[3, 6] The bonding structure found in diamond dictates the chemical and physical properties of the material. Diamond is the hardest naturally occurring material known, with this physical property derived from the strength of the C-C σ-bonds in the tetragonal interlocking three-dimensional arrangement. Diamond is an electrical insulator, unlike graphite, as only σ-bonds are present with no opportunity for the formation of delocalised π-bonds required for electron conduction. Diamond is a particularly sought after gemstone, due to its high refractive index and colourless appearance, where the electrons only absorb light in the ultraviolet region and not in the visible or infrared.[6]

1.1.2 Graphite

In contrast to diamond, the four valence electrons of the carbon atom occupy the planar sp2 hybrid orbital (Fig. 1.1). This dramatically effects the carbon bonding capacity and allows additional carbon atom attachment through three in-plane σ-bonds, with an out-of-plane π-orbital. This bonding type gives rise to a planar hexagonal network of strong in-plane bonding, but only weak inter-planar bonding, termed van der Waals forces, producing the structure depicted in Figures 1.1 and 1.2b. Due to the out-of-plane π-orbital, graphite exhibits semi-metal behaviour and is a good thermal and electrical conductor.[1] It is interesting to note that in terms of mechanical properties graphite is actually stronger in-plane than diamond. This can be explained by the shorter length σ-bond of the sp2 hybrid orbital (0.14 nm) with an associated bond strength of 347 kJ mol-1, when compared to the sp3 hybrid orbital bond length (0.15 nm) resulting in a reduction in bond strength to 611 kJ mol-1.[7] Even the physical appearance of graphite is vastly different from diamond, where wavelengths in the visible range are absorbed and interact with the delocalised π-electrons to give a black appearance.[3]

4 Chapter 1

1.2 The discovery of new forms of carbon

Ultimately it was the intrigue and quest for further understanding of the formation mechanisms of long-chain carbon molecules in interstellar space which led to the serendipitous discovery of entirely new allotropes of carbon. This occurred in the now famous vaporisation of graphite experiments undertaken in 1985 at Rice University, Houston where Richard Smalley, Harry Kroto, and Robert Curl and their research collaborators went on to produce one of the most influential scientific discoveries in recent times which would significantly shape the direction of nanoscience and nanotechnology research for many years to come.[8-11] This influential paper, published in the prominent journal Nature, reported that vapourising graphite in a helium (He) atmosphere produced remarkably stable carbon clusters. Upon optimising the experimental conditions, a cluster size of 60 carbon atoms predominated, with lower percentages of 70 carbon atom clusters. The rationale behind the discovery was that only a closed-cage carbon structure containing 60 carbon atoms could produce a structure of unique super-stability. A truncated icosahedron structure, with 60 vertices and 32 faces, 12 of which are pentagonal and 20 hexagonal, which is commonly encountered as the soccerball structure, was proposed (Fig. 1.3). [12] This was the beginning of a new-era for carbon science, and the enormous impact of this pioneering research, by Kroto, Smalley and Curl, led to the award of the 1996 Nobel Prize in Chemistry.[10, 11, 13]

Figure 1.3 A side-by-side comparison of the truncated icosahedral structure of a molecular model of fullerene C60 and a soccerball.

5 Chapter 1

Further research to prove the proposed structural model of C60 was limited by the insufficient quantities prepared through the laser vaporisation of graphite. Later, Huffman et al. published a methodology that was able to produce macroscopic [14] quantities of predominately fullerene C60, using much simpler equipment consisting of a bell jar where a carbon arc was utilised to vapourise graphite, in an atmosphere of He, and the soot that condensed on the vessel walls was collected. This soot was dispersed in benzene to produce a red solution and, upon drying, plate-like crystals of [14] C60 (90%) and C70 (10%) were formed. This was a major breakthrough which gave researchers everywhere access to workable quantities of fullerene C60, which led to the acquisition of experimental data to support the proposed closed-cage structure, and essentially opening the floodgates for chemical and physical characterisation and the development of novel applications.

Figure 1.4 a) Schematic of the electron density of fullerene C60, and b) the C60 isosurface of ground state electron density, calculated with DFT using the CPMD code. Adapted from [15]

2 Fullerene C60, has a unique structural arrangement consisting of predominately sp characteristics as seen in graphite, yet it does retain some form of sp3 hybrid orbital filling, as in diamond, due to the inherent surface curvature (Fig. 1.1). This particular bonding type is referred to as sp2 rehybridisation, as there are three σ-bonds between neighbouring carbon atoms (sp2), with an out-of-plane π-orbital which is deformed to allow a non-planar, pseudo-tetragonal bonding arrangement (Fig. 1.1). Due to the presence of the delocalised π-orbitals (Fig. 1.4), fullerene C60 has very rich electronic and optical properties. These fullerene molecules, more specifically C60, have shown interesting characteristics for areas of study such as metal-insulator transitions, chemical functionalisations, superconductivity and molecular architecture.[4]

6 Chapter 1

The ability to build and control molecular architecture consisting of fullerene C60 will be explored in Chapter 5. The synthesis of novel fullerene arrangements, leading to the realisation of creating all-carbon architecture for applications as broad as molecular electronics to medicine, will be described in Chapter 5.

1.2.1 Carbon nanotubes

With the discovery of a new nano-scale form of molecular carbon, the fullerenes stimulated not only significant amounts of research interest but intrigued researchers with the dream of an all-carbon nano-scale architecture. Perhaps one of the most extravagant conceptions was the possibility to extend the C60 cage in one direction, to a [16] point where one was left with a hollow tube with C60 hemispheres capping both ends. This seemingly conceptual pipe-dream was soon to become a reality and with this came one of the largest shifts in scientific research in the modern world and would be the beginning of the “nanotube age”.[17, 18]

1.2.1.1 Multi-walled carbon nanotubes

The discovery of nanometer sized graphitic tubules was through the efforts of Sumio Iijima, an electron microscopist for the NEC laboratories in Japan. Iijima was intrigued by the Krätschmer-Huffman carbon-arc method for fullerene production and undertook detailed transmission electron microscopy (TEM) studies on the condensed soot on the vessel walls in the search for novel carbon architecture.[1] Due to mainly encountering amorphous carbon, with no sign of extended graphitised structures, the focus of these studies was eventually shifted to the deposits which formed on the graphitic cathode after the arc-evaporation process.[1] This was the eureka moment as he discovered that the cathodic soot contained a wide range of novel graphitic structures, most interesting were the long hollow fibres ranging from 4 nm to 30 nm in diameter with lengths up to 1 μm.[19] High-resolution TEM imaging of the fibres revealed that they consisted of seamless concentric cylinders of increasing diameter about a common axis (Fig. 1.5), with each cylinder separated by a distance of 0.34 nm, the inter-plane distance observed in graphite.[19] These fibres are more commonly referred to as multi-walled carbon nanotubes (MWCNTs).

7 Chapter 1

Figure 1.5 Electron micrographs of the microtubules of graphitic carbon which Iijima discovered, more commonly referred to as MWCNTs. A cross-section of each nanotube is illustrated; a) tube consisting of five graphitic sheets (6.7 nm dia.), b) two-sheet tube (5.5 nm dia.), and a seven-sheet tube (6.5 nm dia.) which has the smallest hollow core (2.2 nm). Adapted with permission from Macmillian Publishers Ltd: Nature[19], © 1993.

1.2.1.2 Single-walled carbon nanotubes

The discovery of MWCNTs validated conceptions that the ‘ideal nanotube’, a seamless cylinder of one atomic layer of carbon might be achievable. Only two years after the discovery of MWCNTs both Iijima et al.[20] and Bethune et al.[21] reported the synthesis of the elusive SWCNT. This was achieved by using the same process for MWCNT production, however, the graphite electrodes were modified by transition metal catalysts, specifically iron and cobalt. A web-like soot was produced where the typical fullerene soot was deposited, on the vessel walls, and resulted in the formation of CNTs 8 Chapter 1 with very small diameters (~1 nm) (Fig. 1.6), with structures of high similarity to the theorised ‘ideal nanotube’.

Figure 1.6 Electron micrographs of SWCNTs, at a) low magnification and b) high magnification; produced through the original modified carbon-arc evaporation preparation by Bethune et al. Adapted with permission from Macmillian Publishers Ltd: Nature[21], © 1993.

These pioneering SWCNT synthesis methodologies were plagued by low yields, with high volumes of impurities such as amorphous carbon, carbon nanoparticles and metal catalyst particles. During the following year, it was discovered, by Ebbesen and Ajayan, that the yield of SWCNTs in the soot could be dramatically improved by increasing the pressure of He in the arc-discharge chamber.[22] This was the critical step required for the ability to produce laboratory-scale quantities which would, in turn, dramatically boost world-wide research into these fascinating materials.

With the wide spread nanotube research phenomenon developing, SWCNTs have now been produced using several different methods including arc-discharge,[23-40] laser ablation,[41-54] and chemical vapour deposition (CVD).[55] Over the years the low-cost, large scale synthesis of CNTs utilising CVD methods has made it the preferred process for the commercial production of SWCNTs. Currently, many sub-categories of the CVD synthesis have been established including, but are not limited to, catalytic-CVD (C-CVD),[56-70] plasma-enhanced-CVD (PE-CVD),[71-81] laser-assisted-CVD (LA-CVD),[82-84] alcohol-catalytic-CVD (A-C-CVD),[85-88] and high pressure CO disproportionation (HiPco).[89-92] It should be noted that recently, fluidised-bed C-CVD synthesis regimes have shown significant promise for the large scale production of

9 Chapter 1

SWCNTs,[93] with diameter-selective growth capabilities.[94] The synthesis of carbon-based materials realised through a C-CVD methodology is described in Chapter 5.

1.2.2 Graphene

Graphene can be conceptualised as a monolayer of the parent structure graphite (Fig. 1.2b), consisting of a two-dimensional sheet of carbon atoms organised in a hexagonal lattice. The carbon bonding structure is purely sp2 orbital hybridisation resulting in exceptional in-plane bond strength, due to three σ-bonds, and the out-of-plane π-orbital resulting in a cloud of delocalised electrons. It has not been until recently that the synthesis of graphene has been achievable, some synthesis method include vacuum graphitisation of silicon carbide (SiC) substrates,[95, 96] and the direct epitaxial growth on metal substrates.[97-99] However, the standard procedure to gain access to graphene is through micromechanical cleavage processes.[100]

Graphene has a number of extremely interesting electronic properties. The presence of Dirac charge carriers in graphene has been experimentally observed, which translates to gapless and approximately linear electron dispersion at the vicinity of the Fermi level at two points in the Brillouin zone (BZ), leading to a zero-bandgap semiconductor.[101, 102] These electronic properties are thought to have significant implications in the future of carbon nano-electronics and also graphene-based sensors. Recently, the potential of graphene as a gas sensing matrix was reported, and it was suggested that the increase in the graphene charge carrier concentration induced by adsorbing gas molecules, resulted in low-noise, measureable responses and opens the possibility of single-molecule binding event detection.[103] A recent report by Snow et al. circumvents the problem of obtaining and depositing pristine graphene, by depositing graphene oxide on a patterned substrate and reducing it to near-pristine graphene in situ with hydrazine before exposing it to vapours such as acetone, dinitrotoluene (DNT) and hydrogen cyanide.[104] This methodology increases the ease of sensor fabrication, however, the reduction step is not ideal for commercial application and it is known that this type of reduction will not remove all the oxide-related functional groups from the graphene surface which will have a detrimental effect on the electrical properties of the sensor.[105]

10 Chapter 1

It is clear that, in terms of proven applications, graphene is in its infancy. However, with the advent of uncomplicated, high yielding syntheses such as the process described by Coleman and coworkers[106] with up to 12 %wt production of pristine monolayer graphene is possible, there will be a dramatic increase in scientific papers detailing progress in the gas and vapour sensor area. The dramatic effect CNTs have had on the scientific community and their integration into commercial applications can be clearly seen, graphene is the next highlight carbon nano-material and has the future potential to surpass the impact of CNT technologies. For a more detailed overview of graphene, inherent properties, and potential applications the reader is directed to a number of reviews on the topic.[107-109]

1.3 Structure and properties of SWCNTs

At the most fundamental level the structure of a SWCNT can be visualised as the rolling of a seamless cylinder from a single sheet of graphene. The typical experimentally observed SWCNT diameter is between 0.6 nm and 2.0 nm,[110] while it has been suggested that a SWCNT should be at least 0.4 nm in diameter to afford strain energy, and at most around 3.0 nm in diameter to maintain tubular structure and prevent collapse.[111-113] A SWCNT’s length can vary over several orders of magnitude. As the structure is based on the wrapping of a single sheet of graphene on itself the bonding in SWCNT is essentially sp2 hybridisation. However, due to the circular curvature of the nanotubes surface, quantum confinement will occur and cause a deformed sp2 bonding structure, commonly referred to as σ-π rehybridisation or sp2 rehybridisation (Fig. 1.1).[110] This bonding structure is similar to the orbital rehybridisation observed in the fullerene structures (Section 1.2), however, the large aspect ratio presented by the tube structure provides one-dimensional quantum confinement, with unique electron density characteristics (Fig. 1.7).

11 Chapter 1

Figure 1.7 a) Molecular model of a section of (9,9) armchair SWCNT, b) electron density map of the first layer of the nanotube, c) electron density map of the second layer of the nanotube. Adapted with permission from [114], © 2001 Elsevier.

Most of the importance of CNT materials is directed towards their efficient utilisation in device electronics.[115] This arises as the electronic properties of a SWCNT are dependent on the direction in which the ‘graphene sheet’ was rolled up to form the nanotube.[116-118] Each carbon atom in the hexagonal graphene lattice can be identified with a pair of integers (n,m). The topology or helicity of a SWCNT can be uniquely characterised by a chiral vector, Ch, in terms of a set of two non-negative rolling [110, 119, 120] integers (n,m) which correspond to the graphite vectors a1 and a2 (Fig. 1.8).

1h  mn aaC 2 (1)

The SWCNT cylinder is produced by rolling up the graphene sheet such that the two end-points of the vector Ch are superimposed. The nanotube formed is denoted as (n,m) nanotube and the tube diameter, dt, is given by

2/1 C 22  nmmna d h    (2) t  

[119] where a = 0.249 nm (due to 3aa c-c  0.249 nm [ac-c = 0.142 nm]).

12 Chapter 1

Figure 1.8 Conceptually, a SWCNT is formed by rolling up graphene along a specific chiral vector to form a cylinder. Three different types of SWCNTs are presented. Adapted with permission from [121], © 2005 John Wiley & Sons, Inc.

Figure 1.9 Schematic indicating the correlation between chiral indices (n,m) and the electronic properties of the resulting SWCNT at room temperature. CNTs with n = m (armchair nanotubes) and those with n – m = 3q are metallic at room temperature (labelled green). CNTs with n - m = 3q + 1 (labelled pink) and n – m = 3q + 2 (labelled purple) are semiconductors with a bandgap that varies inversely with tube diameter. Reproduced with permission from [122], © 2006 courtesy of Mike Arnold.

13 Chapter 1

As detailed in Figures 1.8 and 1.9, the classification of the SWCNT as zigzag (m = 0) or armchair (n = m) stems directly from the geometric arrangement of the carbon atoms at the seam of the cylinder. Both zigzag and armchair types of SWCNTs clearly have mirrored symmetry, however, depending on the chiral angle, θ, where

2/1 1  3 m    tan   (3)  2nm 

any chiral angle 0 < θ > 30° represents a chiral SWCNT.[119] There can be many different chiral SWCNTs that can exist with many different chiral indices (Fig. 1.9) and although two SWCNTs may have extremely similar diameters they can have different chiral vectors (Fig. 1.10a) which directly effects there electronic properties (Fig. 1.9). It should be noted that chiral SWCNTs with identical chiral vectors (Ch) can be enantiomers with right- and left-handed helicity.[121]

Figure 1.10 a) Schematic of two SWCNTs with nearly identical diameters but with differing chiral indices, b) schematic of two SWCNTs with identical chiral vectors (Ch) but with right- and left-handed helicity. Reproduced with permission from [122], © 2006 courtesy of Mike Arnold.

The actual atomic structure of a SWCNT can be studied using characterisation techniques such as high-resolution TEM (HR-TEM), atomic force microscopy (AFM), and scanning tunneling microscopy (STM). The pioneering work by Dekker et al. showed the ability to produce STM topographical images of atomic resolution to indicate SWCNTs with different chiralities (Fig. 1.11d). [123] More recently, Iijima et al. were able to assign SWCNT chirality to a magnitude of individual SWCNTs using

14 Chapter 1 aberration-corrected TEM (Figs. 1.11a-c).[124] It can clearly be seen that the advancement of modern characterisation instruments is validating and supporting current SWCNT structural theory.

Figure 1.11 a) TEM images of SWCNTs a1 (left) and a2 (right) and their fast Fourier transform (FFT) images (inset); b) schematic illustrations of the structures of (10,10) and (11,8) SWCNTs (left and right panels, respectively; c) simulated TEM images based on the models in b) and their FFT images (inset), characteristic bright spots in the FFT image are indicated by red triangles in a) and c) scale bar = 1 nm; d) STM image of the resolved atomic resolution of a SWCNT. Figures a), b), and c) adapted with permission from [124], © 2008 American Chemical Society, and Figure d) adapted with permission from [125], © 1999 American Institute of Physics.

1.3.1 Electronic properties

It can now be appreciated that from a typical synthesis there will be a wide range of SWCNTs of different diameter, chiral vectors and chiral indices produced. This will ultimately produce different metallic and semiconducting SWCNTs, where approximately two-thirds are semiconducting and one-third are metallic at room temperature (Fig. 1.9).[126] Of all the interesting physical and chemical properties of SWCNTs, the inherent electronic properties will be one of the main concepts described in this thesis. Ultimately, the electronic properties of the SWCNTs give rise to the

15 Chapter 1 ability to utilise these extraordinary materials as active sensing matrices for gas and organic vapours. The application to sensors will be introduced in (Section 1.4.1), and the design, fabrication and testing of CNT sensors will be the primary focus of Chapters 3 and 4.

Figure 1.12 The band structure (top) and the BZ (bottom) of graphene. The circumferential quantisation demonstrated by the nanotube leads to the formation of a set of discrete energy sub-bands for each nanotube (red parallel lines). The relation of these lines to the band structure of graphene determines the electronic structure of the nanotube. If the lines pass through the K or K’ points, the nanotube is a metal: if they do not (as in this Figure), the nanotube is a semiconductor. Adapted with permission from Macmillian Publishers Ltd: Nature Nanotechnology[127], © 2007.

To understand the band structure, and ultimately the inherent electronic properties, of a SWCNT, it should be realised that this stems directly from the electronic structure of graphene. The band structure of graphene (Fig. 1.12) is rather unusual as it has six specific points (Dirac points) along certain directions of the first BZ where the valence band touches the conduction band at the Fermi energy, producing a zero-bandgap semiconductor.[117, 118] This means that graphene exhibits metallic behaviour in some directions and semiconducting in others. Due to the one-dimensional nature of a SWCNT there will be a large number of allowed electronic states in the axial direction, however the number of allowed states in the circumferential direction will be very limited.[128] Due to this circumferential quantisation of states, the available states amount to slices through the two-dimensional band structure, represented by red lines in Figure 1.12, resulting in either one-dimensional metals or semiconductors.[117] For example, if a tube axis allows a one-dimensional slice through the BZ vertices, with the 16 Chapter 1 momentum states passing through K or K’, the SWCNT will act as a one-dimensional metal, referred to as a metallic SWCNT. On the other hand, if the tube axis does not allow the momentum states to pass through K or K’, as displayed in Figure 1.12, there will be a bandgap between the valence band and the conduction band, resulting in a one-dimensional semiconductor, referred to as a semiconducting SWCNT.[127] This is commonly described by the SWCNT density-of-states (DOS). The DOS for a metallic armchair (5,5) nanotube and a semiconducting zigzag (7,0) is illustrated in Figure 1.13. It can be clearly observed that for the semiconducting nanotube that there are no charge carriers in the DOS at the Fermi energy (0 eV), hence the bandgap between the valence band and the conduction band (Fig. 1.13b). Furthermore, depending on the tube axis, this bandgap will increase or decrease in proportion to the SWCNT diameter. For a more detailed overview of the electronic properties of SWCNTs, the reader is referred to reviews on the subject.[129, 130]

Figure 1.13 Electronic properties of two different SWCNTs, a) an armchair (5,5) nanotube exhibits metallic behaviour, b) a zigzag (7,0) nanotube is a small bandgap semiconductor (no charge carriers in the DOS at the Fermi energy). Adapted with permission from [131], © 2002 American Chemical Society.

17 Chapter 1

In reality there are van der Waals forces between individual SWCNTs, the same forces which lead to the planar structure of graphite, which typically yields bundles, or ropes, or SWCNTs. Generally, the tubes within these bundles will have a wide diameter distribution (Fig. 1.14a). However, using the laser vaporisation synthesis of Smalley et al. leads to SWCNT ropes with individual diameters of 1.38 ± 0.02 nm.[43] It has been inferred from X-ray diffraction (XRD)[43] and supported by electron nano-diffraction studies,[132] that the ropes primarily consisted of metallic armchair (10,10) nanotubes (Fig. 1.14b).

Figure 1.14 a) A bundle of SWCNTs with different sets of tube diameters, b) a SWCNT rope consisting of metallic armchair (10,10) nanotubes. Adapted with permission from [122], © 2006 courtesy of Mike Arnold, and permission from [133], © 2002 John Wiley & Sons, Inc., respectively.

The one-dimensional structure of a SWCNT and the quantum confinement derived from the circumferential curvature gives rise to SWCNTs behaving like quantum wires.[134] One aspect of a quantum wire is that electron conduction occurs through well-separated, discrete electron states.[128] Quantum wires also can exhibit electron transport which is ballistic in nature, with electrons being able to travel along the axis of a SWCNT without experiencing scattering from impurities or phonons. Essentially, the electrons will travel without dissipating any energy, or experiencing any resistance in the conductor.[128] However, experimental conductance of an individual SWCNT (~10 kΩ)[135] is lower than the theoretical quantised value (6.45 kΩ).[110, 130] In reality this is due to scattering caused by lattice defects, impurities, structural distortions, substrate coupling, and electrode contacts.[128] Even though these experimental values do not match theoretical predictions, the unique one-dimensional structure of a SWCNT provides near-ballistic electron transport properties, which has significant implications in areas such as, sensors, ultra-fast devices and nano-electronics.

18 Chapter 1

Currently, one of the main problems hindering CNT research and commercial implementation is the post-growth purification of CNTs,[136] which can be detailed on different levels. Firstly, the extraction of CNT material from the other soot material, such as metal catalyst nanoparticles and amorphous carbon; secondly, individualisation of the SWCNTs, to overcome the van der Waals interactions between nanotubes; and thirdly, purification to a point where SWCNTs of the same topography/helicity can be obtained. Until post-growth treatments addressing these issues are realised, the practical application of CNTs in areas such as device technology and nano-electronics, will be severally restricted.[136] These critical issues and details of novel purification methodologies for chirality-selective SWCNT purification, which is considered the ‘holey grail’ of SWCNT post-growth treatment, are addressed in Chapter 2.

1.3.2 Chemical and electrochemical properties

In terms of chemical reactivity, defect-free, or ideal SWCNTs can be separated into two distinct regions, the side-walls and the tips. Typically, as-synthesised SWCNTs will have a cap on one end, reminiscent of a fullerene hemisphere, consisting of five-membered rings (Fig. 1.3). These five-membered rings are more reactive then their six-membered counterparts and the reactivity of these caps are comparable to that of a fullerene itself. Chemical functionalisation of the SWCNT tip is more energetically favourable due to this pronounced curvature in two-dimensions when compared to the slight one-dimensional curvature at the side-walls.[121] It should be noted that due to this curvature and the inherent σ-π rehybridisation, the side-wall surface of the SWCNT is more reactive than a planar graphene sheet. Directly opposed to this is the concave curvature demonstrated inside the SWCNT. This type of curvature imparts very low chemical reactivity, again due to the σ-π rehybridisation, and applications such as nano-containers for gases,[137] and nano-test tubes for reactions have been realised.[138, 139]

In reality, SWCNTs will contain structural defects such as vacancies, Stone-Wales defects, pentagons, heptagons and dopants.[133, 140-144] Defects will be imparted during growth or occur through applied stress,[145] will effect the chemical reactivity of the nanotube.[121, 133, 146] There are also purification induced defects such as the introduction of acid groups to the side-wall and tip of the nanotube. Haddon et al., through simple

19 Chapter 1 acid-base titrimetric experiments, observed that 1% to 3% of carbon atoms of a SWCNT were in fact located at a defect site.[147] Among the possible structural defects that can occur, the Stone-Wales defect (Fig. 1.15) is considered the lowest energy defect and the most influential on the nanotube’s mechanical, chemical and electronic properties.[148] Increased curvature imparted to the SWCNT through the Stone-Wales defect (Fig. 1.15) increases chemical reactivity due to the imparted increase in side-wall curvature. However, recent theoretical studies have indicated that this is not always the case.[149, 150] Furthermore, there is a common consensus that pyramidalisation and π–orbital misalignment play a major role in dictating SWCNT reactivity towards addition reactions.[148]

Figure 1.15 Schematic of a Stone-Wales defect (or 7-5-5-7) on the side-wall of a SWCNT. Adapted with permission from [121], © 2005 John Wiley & Sons, Inc.

The chemical functionalisation of SWCNTs has generated significant interest in recent times, with many schemes built around interplay with the structural defect sites. The main chemical strategies for functionalising SWCNTs are covalent side-wall chemistry, covalent defect and open-tip chemistry, non-covalent endohedral functionalisation with surfactants, non-covalent endohedral functionalisation with polymers, and endohedral functionalisation (Fig. 1.16). [133] Another important form of functionalisation is the

20 Chapter 1 decoration of the SWCNT surface with metal nanoparticles, having implications in areas such as sensing and catalysis.[151, 152] The important chemical functionalisation strategies demonstrated in this thesis are the non-covalent wrapping of surfactants, which forms the basis of Chapter 2, defect and open-tip chemistry, which is described in Chapter 4, and nanoparticle loading detailed in Chapter 3. For a more detailed overview of SWCNTs chemical and electrochemical properties and associated functionalisation, the reader is directed to a number of reviews of the subject.[121, 133, 146, 153, 154]

Figure 1.16 A schematic depicting different SWCNT functionalisation strategies, a) covalent side-wall functionalisation, b) covalent defect and open-tip functionalisation, c) non-covalent exohedral functionalisation with surfactants, d) non-covalent exohedral functionalisation with polymers, and e) endohedral functionalisation with, for example, C60. Adapted with permission from [133], © 2002 John Wiley & Sons, Inc.

1.3.3 Other properties

Due to their nanometer scale dimensions and unusual bonding structure there has been a major focus by both theorists and experimentalists on predicting and exploiting the other properties of SWCNTs. The following sub-sections will provide a brief overview of the inherent properties of SWCNTs, other than their electronic and chemical properties.

21 Chapter 1

1.3.3.1 Optical and optoelectronic

The optical and optoelectronic properties of SWCNTs are directly related to the quantum confinement from their cylindrical nature and to their well-defined band and sub-band structure (Fig. 1.13). Basically, the sharp peaks observed in the DOS (Fig. 1.13) are optical transitions, when electrons are excited from one energy level to another. These peaks are termed van Hove singularities (VHS). Optical spectra have been obtained for individual SWCNTs using resonance Raman,[155] fluorescence,[156, 157] and ultraviolet-visible-near-infrared (UV-Vis-NIR) spectroscopies.[158] These techniques have been instrumental in the assignment of the distribution of nanotube diameters, chiralities and chiral indices in SWCNT samples, which typically relate back to the uniqueness of the VHS of an individual SWCNT.

1.3.3.2 Mechanical and electromechanical

Due to the sp2 orbital hybridisation character of the SWCNT, the σ-bonding of the intra-plane carbon network (Section 1.3) gives rise to appreciable mechanical properties. The SWCNT can be regarded as the ultimate fibre, due to the exceptional bond strength within the tube axis, imparted by the strong σ-bonds.[110] Both theoretical calculations and experimental observations indicate SWCNTs are among the stiffest materials known to man, with a high Young’s modulus and tensile strength (Table 1.1). These excellent mechanical properties are compounded by the low density of the SWCNTs, providing access to high strength, low weight materials. In relation to mechanical properties, MWCNTs actually have increased Young’s modulus and tensile strength (Table 1.1), which is due to the Young’s modulus taking the highest value of a SWCNT, plus contributions from coaxial inter-tube coupling (van de Waals interaction). Furthermore, the Young’s modulus is significantly reduced in SWCNT bundles as the van de Waals interactions between tubes induces a strong shearing force among the packed tubes.[159]

The SWCNTs elastic strain response to deformation is also remarkable. In comparison to other hard materials, SWCNTs can withstand up to 15% tensile strain before fracture,[160] whereas other hard materials, such as diamond, fail with a strain of ~1% or less due to propagation of dislocations and defects.[110] SWCNTs can handle these high strain conditions due to their sp2 rehybridisation bonding structure, through which the

22 Chapter 1 strain can be released. It should be noted that induced tensile or torsional strain has the ability to change the electronic properties of a SWCNT.[110]

Table 1.1 Comparison of the mechanical properties of CNTs, graphite and steel. Adapted with permission from [160], © 1998 World Scientific Publishing Co.

Young’s Modulus Tensile Strength Density

(Gpa) (Gpa) (g cm3 -1)

MWCNT 1200 ~150 2.6

SWCNT 1054 75 1.3

SWCNT bundle 563 ~150 1.3

Graphite (in-plane) 350 2.5 2.6

Steel 208 0.4 7.8

1.3.3.3 Magnetic and electromagnetic

The magnetic properties of CNTs have been studied, with the main goal to understanding their electromagnetic properties. Magnetic properties such as the average observed g-value and spin susceptibility have been found in MWCNTs.[161, 162] However, similar observations have not been made with respect to SWCNTs or SWCNT bundles, possibly due to sample alignment difficulties and strong electron correlation, which may block the non-conduction electron spin resonance (ESR) signal.[110] Similarly, since mechanical deformation can alter the electronic properties of SWCNTs, the application of aligned or perpendicular magnetic fields can alter the DOS bandgap resulting in tuning of electronic properties.[110] This property could be utilised in the emerging branch of electronics referred to as spintronics, exploiting electron spin rather than charge.[163]

23 Chapter 1

1.3.3.4 Thermal

Other allotropes of carbon, diamond and graphite, exhibit high thermal capacity and thermal conductivity, so it is natural to want to predict and experimentally determine these properties in SWCNTs. At temperatures above 100 K SWCNTs and SWCNT bundles exhibit the specific heat of graphite, ~700 mJ g-1 K-1.[164] However, as the SWCNT temperature is lowered a linear temperature dependence is observed, which indicates thermal transport by one-dimensional photons, showing that thermal conductivity is one-dimensional analogous to electrical conductivity. This is induced through the quantum confinement effects imposed by the curvature of a SWCNT.[165]

1.4 SWCNT applications

Due to the unique structure and inherent properties of SWCNTs there has been a monumental quantity of published research and patents outlining both realised and speculative applications. Such application areas include, but are not limited to, nano-electronics,[136, 166] composites,[167-172] photovoltaics,[173-175] analytical chemistry,[176-180] field emission,[181-184] water desalination,[185, 186] probe microscopy,[187- 189] hydrogen storage,[190, 191] catalyst supports for fuel cells,[192, 193] medicine,[194-197] cellular studies,[198-200] spintronics,[163, 201] superconductivity,[202, 203] biosensors,[204-209] actuators,[210] and physical,[211-214] mechanical[215-217] and chemical sensors.[151, 204, 218, 219] Where possible, review articles and recent application articles have been cited to inform the reader of the magnitude of studies undertaken in each application area. In addition, for more information detailing general applications, the reader is directed to an extensive review.[220]

Progress in the chemical sensor development, in particular with respect to CNT-based gas and vapour sensor research, will be expanded in more detail in Section 1.4.1. Furthermore, the utilisation of CNTs for vapour-phase sensing is one of the central themes of this thesis.

24 Chapter 1

1.4.1 Carbon nanotube gas and vapour sensors

The aforementioned inherent properties of CNTs make them appealing candidates to potentially revolutionise the sensor industry with the development of “next-generation” sensors.[219] In terms of the active matrix for gas and organic vapour sensing, at present, the properties of SWCNTs are optimal to those of MWCNTs and as such the focus of this current section will be on SWCNTs. SWCNTs have high aspect ratios, are of nano-dimensions and are entirely composed of exposed surface atoms giving rise to extremely large surface areas,[221] with a highest recorded value of 1587 m2 g-1.[222] In addition, it has been demonstrated that molecular adsorbates can interact with these surface atoms and significantly alter their electronic properties.[223, 224] Coupled with SWCNTs near-ballistic electron transport characteristics,[225] allows for a high-quality electrical conduit for transmission of these electrical perturbations to external contacts. Furthermore, the graphene-like surface provides a chemically robust system which leads to long-term operational stability.[226] Consequently, a combination of these properties has recently projected these unique nano-materials into the spotlight of chemical sensing.

The pioneering work of both Kong et al. and Collins et al. demonstrated the potential of [223] SWCNTs as gas sensors for nitrogen dioxide (NO2) and ammonia (NH3), and [224] oxygen (O2) respectively. Kong et al. produced a nanotube-field effect transistor (NT-FET) which incorporates a single, individual semiconducting SWCNT as the conduction channel. As the NT-FET was exposed to NO2 and NH3 a charge-transfer process takes place which causes chemical gating of the device which can be measured.

In terms of characteristic gate voltage (VG) when the device was exposed to NO2 there was a VG shift of approximately +4 V, with exposure to NH3 resulting in a shift of [223] approximately -4 V. The system was also tested with a VG = 0 V, while exposed to

NH3 which resulted in a decrease in conductance (G) (Fig. 1.17a), and VG = +4 V, while [223] exposed to NO2 which resulted in an increase in device conductance (Fig. 1.17b).

Collins et al. demonstrated that these electrical property changes extended to O2, however, in this case, thin films of SWCNT bundles were utilised as the active sensing matrix.[224] These scientific findings in 2000 caused major shockwaves throughout the scientific community and at the time of this thesis compilation over 1300 literature reports of CNT-based sensors exist, with over 500 of these published in relation to CNT-based gas and vapour sensors (based on Scifinder® data[227]). 25 Chapter 1

Figure 1.17 a) Electrical response of an NT-FET at VG = 0 V to 1% NH3, and b) at VG = +4 V to [223] 200 parts-per-million (ppm) NO2. Reprinted with permission from , © 2000 AAAS.

1.4.1.1 Different CNT-sensor configurations

Advancement from the original NT-FET device configuration has led to a variety of different CNT-sensor configurations in the pursuit of more sensitive and selective systems with decreased sensor response and recovery times, producing low-power consumption devices with reduced physical dimensions. Not all devices mentioned utilise SWCNTs, some encompass MWCNTs with specific examples specified. For example, Ong et al. produced a gas-responsive MWCNT/silicon dioxide (SiO2) composite layer on a planar inductor-capacitor resonant circuit which could be [228] monitored, remotely, in response to carbon dioxide (CO2), O2 and NH3. Li et al. produced a SWCNT chemiresistor sensor, created through a random network of SWCNTs on interdigitated electrodes (IDEs) and claimed detection limits of 44 parts- [229] per-billion (ppb) for NO2 and 262 ppb for nitrotoluene. A different device architecture was reported by Chopra et al., in which microwave resonant MWCNT and SWCNT sensors were utilised for the highly sensitive and fast-response detection of

NH3. This particular sensor is based on a circular, disk electromagnetic resonant circuit coated with CNTs, the electrical resonant frequency of the sensor is monitored and a dramatic downfield shift (4.375 MHz) was seen upon NH3 exposure, with recovery and response times of approximately 10 min; which is relatively fast compared to chemiresistor and NT-FET configurations.[230] In addition, Penza et al. developed surface acoustic wave (SAWs) MWCNT and SWCNT sensors for the detection of volatile organic compounds (VOCs),[231, 232] which shares similarities to the microwave resonance sensor of Chopra et al. Device configuration can be taken one step further, for example, where the SWCNTs are grown in an alumina template and sealed, following this the surface coating is etched; resulting in gas flow and adsorption only occurring in the inner walls of the SWCNTs.[233]

26 Chapter 1

A configuration that fully harnesses the nano-dimensions and electrical properties of CNTs is a unique miniaturised gas ionisation sensor utilising MWCNTs. This was produced by Ajayan and coworkers and is based on a MWCNT vertically-aligned “forest” between an anode and cathode electrode but with a 150 μm gap from the MWCNT tips to the cathode.[234] Due to the sharp tips of the MWCNTs it was possible to generate very high electric fields at relatively low voltages. It was shown that different gases, such as He, CO2, NH3, have different breakdown voltages and this could be utilised to distinguish between the detection of different gases. Furthermore, this configuration could easily be manufactured for portable deployment as it was low-powered and could be powered by carbon-zinc batteries in series.[234]

Figure 1.18 a) Optical micrograph illustrating the chemicapacitor configuration; inset is an AFM image of the random network of SWCNTs, and b) measured relative capacitance change of the SWCNT chemicapacitor in response to repeated 20 s doses of dimethylformamide at varying concentrations noted in the figure. Reprinted with permission from [235], © 2005 AAAS.

One of the major revolutionary SWCNT sensor breakthroughs in terms of sensor configuration was the development of a SWCNT chemicapacitor by Snow et al.[235] This configuration consists of a network of randomly orientated SWCNTs deposited on interdigitated metal electrodes which were grown on a heavily doped silicon (Si) substrate (Fig. 1.18a). The device operates using a capacitance mechanism and the SWCNT network serves as one plate of the capacitor and the Si substrate forms the other. Under an applied gate bias a large radial electric field emanates from the SWCNT network. When an analyte molecule comes into contact with the SWCNT surface it is polarised by this electric field and an increase in system capacitance results. This particular configuration gives rise to strong capacitance responses to most polar molecules, and even some non-polarisable molecules gave a response. In comparison to the large majority of CNT-based sensors at the time of conception, and even today, this

27 Chapter 1 configuration is highly favourable for its fast, sensitive, and low-noise transduction mode in the detection of organic vapours.[226, 235]

Snow et al. have also developed a dual-configured device for simultaneous conductance and capacitance monitoring.[236] In addition, Star et al. recently developed a SWCNT configuration which could simultaneously monitor the electrical and spectroscopic [237] properties of a SWCNT film after exposure to NH3 and NO2. Both systems are instrumental in the advancement of the mechanistic understanding of SWCNT gas and vapour sensors.

It is noteworthy that as NH3 and NO2 were the first environmentally important gases with which CNT sensors were shown to interact, and there is a significant proportion of the literature focused on these two gases. The development of CNT-sensors for the detection of organic vapours was not established until 2003, and has been a major focus point ever since. Consequently, this current section provides an overview of a variety of different sensor configurations and variations within these broad configurations can be found throughout the literature.

1.4.1.2 Mechanisms of action

A variety of different device configurations have been described in Section 1.4.1.1, however, only the proposed mechanisms for the sensor type(s) relevant to this thesis will be elaborated here, and these are NT-FET and chemiresistors, as shown in Figures 1.19a and 1.19b, respectively.

Figure 1.19 a) Schematic of a NT-FET device configuration, and b) schematic of a chemiresistor device configuration, with no gate-drain potential. Adapted with permission from [238], © 2006 American Chemical Society, and permission from [151], © 2008 John Wiley & Sons, Inc., respectively.

28 Chapter 1

A typically SWCNT device consists of either an individual SWCNT or a network of SWCNTs that are in intimate contact with metal electrodes which ultimately allow the monitoring of the electrical properties of the nanotube(s). It has already been discussed (Section 1.3) that semiconducting SWCNTs exist and this type of nanotube has a fundamental role in the mechanism of action for SWCNT gas and vapour sensors. In terms of a NT-FET it is important to realise that when a metal and a semiconductor are in direct contact a potential barrier, the Schottky barrier (SB), develops at the immediate interface, due to work function mismatch of the materials.[239] The magnitude of the SB is dependant on the particular metal (Fig. 1.20a) and when a constant source-drain voltage (VSD) is applied the height of the SB can be modulated through the use of a VG. + This VG modifies the SB and either increases or decreases the probability of a hole (h ) travelling from the metal into the valance band of the SWCNT.[151] To experimentally demonstrate this phenomenon Avouris and coworkers performed positive to negative

VG sweeps using different metal contacts to produce metal specific transfer [240] characteristics and current-gate voltage (I-VG) curves (Fig. 1.20b).

Figure 1.20 a) Schematic of the work function dependent SB caused by the metal-SWCNT work function mismatch, and b) I-VG transfer characteristic of NT-FET devices composed of semiconducting SWCNTs contacted by metals with different work functions; the devices were [240] operated at VSD = -0.5 V. Adapted with permission from , © 2005 American Chemical Society.

In terms of individual SWCNT electronic characteristics it has been shown that semiconducting CNTs are p-type under ambient conditions.[134, 241] It is important to specify that for the majority of CNT-analyte interactions the dominant physical mechanism behind sensitivity to ambient is the physical adsorption of species on the surface of the nanotube.[226] If a molecular adsorbate with electron donation characteristics interacts with a nanotube there will be electrons donated into the valence + band which results in h recombination and a shift in the I-VG curve towards a more negative voltage, which in turn relates to a decrease in conductance in the system. On

29 Chapter 1 the other hand, if an electron withdrawing species interacts with the surface this will + increase the h concentration in the semiconducting nanotube, which will shift the I-VG in a positive direction resulting in an overall increase in conductance in the nanotube;[226] this type of mechanism is referred to as charge-transfer or conductance mechanism.[226] It should be noted that this principal mechanism is apparent for chemiresistor devices as they can be visualised as being at VG = 0. There is also another interaction mechanism suggested in the literature, the capacitance mechanism, however, this is associated with chemicapacitor device configurations (Section 1.4.1.1) and a comprehensive comparison between the two mechanisms has been published by Snow et al.[226] To re-iterate, the charge-transfer mechanism is the predominant mechanism for the chemiresistor sensors tested in this thesis (Chapter 1).

Surface adsorbates can introduce scattering sites along the tube axis which alter the charge mobility of the nanotube; this mechanism leads to a change in slope in the I-VG curve,[242] and is referred to as transconductance.[241] Furthermore, the actual device contact itself can have an effect on the nanotube-metal interface SB and this effect can be similar to that observed for a charge-transfer mechanism, which to-date still causes ambiguity over CNT sensor response mechanisms.[151]

Another interesting aspect of the CNT-analyte interaction mechanism is that it has been experimentally observed that analyte surface coverage of the nanotube is not correlated with the concentration of the species at ambient temperature, the partial pressure (P), but actually correlates to a fraction, P/P0, of the equilibrium pressure (P0). Therefore, the sensor response to a particular analyte is based on the probability of condensation on the nanotube surface and not on its local abundance. This translates to a large dynamic range analyte affinity sensing material, which responds equally well to both low and high vapour pressure analytes. This unique feature is yet another appealing aspect for the utilisation of CNTs as active sensing materials.[226]

In relation to the individual molecular interaction mechanisms of NH3 and NO2, there has been significant conjecture in the literature over the years, with not only different mechanistic reasoning from experimentalists and theorists but opposing views within these two groups. The majority of the literature based on theory and experimental data

30 Chapter 1 have indicated that a charge-transfer mechanism either from the adsorbate to the SWCNT, or vice versa, depending on whether the adsorbate is electron-withdrawing

(NO2) or electron-donating (NH3). For a more detailed breakdown of current NO2 and [151] NH3 interaction mechanisms the reader is directed to a recent review article. It should be noted that not all gases and organic vapours will have an interaction mechanism, such as charge-transfer (conductance) or polarisation (capacitance), with a SWCNT. In situations where this is the case alternative methodologies have been employed to develop a sensing mechanism which will lead to a favourable interaction and hence produces a detectable sensor response. This leads to a magnitude of different adsorbate-sensor interactions, with possible CNT sensor modifications described in Section 1.4.1.3.

1.4.1.3 Methodologies to increase analyte specificity and sensitivity

There has been a constant quest to improve the CNT-based sensor analyte selectivity and specificity, while expanding the range of possible analytes. These improvements can range from direct covalent modification of the nanotube surface for the utilisation of multiple sensory arrays in analyte mixture profiling; the current section will provide a brief overview of some of these modification strategies.

A simplistic methodology that has been reported by several research groups, to increase sensor sensitivity and performance, is to operate the CNT sensor at elevated temperatures.[243-246] This approach can be utilised to increase CNT sensor response and recovery times, reduce the effects of humidity on sensor performance and induce increased sensitivity. Another methodology, which at first glance seems relatively simplistic, is the loading or doping of CNTs with metal or metal oxide particles predominately nanoparticles, or thin films. This modification strategy is one of the most frequently reported and has been a major focus for increasing CNT sensor capabilities for gases and organic vapours. These dopants are typically d-block transition metals, e.g. cobalt (Co),[247] and noble metals including but not limited to, rhodium (Rh),[238, 248] palladium (Pd) (Fig. 1.21b), [238, 249-260] gold (Au),[238, 261, 262] and platinum (Pt).[238, 263] The majority of these dopant systems have shown increased sensitivity to gases such as NH3 and NO2. As pristine CNTs have very little to no affinity to hydrogen (H2) gas the Pd, Rh, and Pt systems are utilised to act as catalytic

31 Chapter 1 clusters. These clusters breakdown diatomic hydrogen to molecular hydrogen and hence a charge-transfer from analyte to nanoparticle occurs and effects the conductance of the supporting nanotube. Moreover, Laing et al.[264] and Bittencourt et al.[265] have also utilised metal oxide semiconductors (MOS), tin dioxide (SnO2), and tungsten trioxide (WO3) respectively, as CNT dopants in the detection of typically non-responsive gases such as carbon monoxide (CO) and nitrogen oxide (NO). In addition, Sánchez et al. produced a sensor which consists of CNTs embedded in a sol-gel-derived titanium dioxide (TiO2) matrix, which was sensitive to both NH3 and acetone vapour.[266] It should be noted that Star et al. established a comprehensive study on different metal nanoparticle-doped SWCNTs in NT-FETs and obtained sensing affinity data for each type of metal.[238] Although metal and metal oxide doping is typically used to improve and activate sensing capacities of CNTs for gases, some of the cited publications do expand into the realm of the application of these devices for sensing organic vapours.

There are limited reports regarding covalent functionalisation of CNTs for gas and organic vapour sensing purposes. However, Villalpando-Páez et al. created a NH3 sensor which consisted of nitrogen-doped CNTs, which had significantly faster response times when compared to conventional systems.[267] Other covalent functionalisations of the CNT surface have been studied, by Borguet et al.[268] and Snow et al.[104], in which the effect of CNT surface defects (as discussed in Section 1.3.2) on analyte sensitivity are proposed. In relation to covalent attachment of different functional groups, this type of modification features significantly in terms of liquid-medium CNT biosensors,[206, 218] but does not have a significant impact in the gas and vapour sensor domain. In terms of CNT functionalisation, a significant proportion of studies have been focused on non-covalent interactions with CNTs to increase sensor performance, which has had a large impact on both gas and organic sensing, and in recent years has dominated the organic vapour literature.

The concept behind non-covalent functionalisation typically refers to the use of a chemo-selective polymer coating on the CNT surface which can either ‘filter’ unwanted interfering species or will interact favourably with the analyte of interest and this interaction will effect the electrical properties of the CNT in a predicable and reproducible fashion. An interesting example is provided by Johnson et al. in which 32 Chapter 1 single strand-deoxyribonucleic acid (ss-DNA) was drop-cast onto the surface of a SWCNT in a NT-FET configuration and this functionalisation caused sensitivity towards propionic acid, methanol and trimethylamine, while also responding to chemicals of national security interest such as dimethylmethylphosphonate (DMMP) and 2,6-DNT).[269] Furthermore, polymer systems such as poly(m-aminobenzene sulfonic acid) (PABS),[270, 271] poly(methylmethacrylate) (PMMA),[272] PEI (Fig. 1.21a),[273, 274] Nafion® (Fig. 1.21a),[275] PEI/starch,[275] and polyaniline (PANI)[276] are used to improve sensitivity and selectivity towards gases such as NH3, NO2, CO2, and CO.

Figure 1.21 a) Optical image illustrating three CNT-based devices after micro-spotting with droplets of poly(ethylene imine) (PEI) and/or Nafion® solution, and b) AFM image of Pd nanoparticles electrochemically deposited on the surface of SWCNTs forming the active material [273] for a H2 gas sensor. Adapted with permission from , © 2003 American Chemical Society, and permission from [256], © 2007 American Institute of Physics, respectively.

As previously mentioned, in relation to organic vapour sensing, the non-covalent functionalisation of CNTs with polymers has dominated the sensing literature in recent years. A prominent example, by Snow and coworkers, describes how a SWCNT chemiresistor flow cell was developed with an in-line filter in place, containing an acidic, strong-hydrogen-bonding polymer, termed HC; this not only allowed selective [277] detection of DMMP over NH3, but also achieved sub-ppb detection limits. The use of this DMMP chemo-selective polymer was extended further in a pioneering study which applied a monolayer of the polymer onto a SWCNT chemicapacitor. This system provided excellent response and recovery times, no sensor saturation for a large analyte concentration range, and detection limits as low as 0.5 ppb,[235] which puts it into contention for the most superior SWCNT DMMP sensor system to date.

33 Chapter 1

A flexible SWCNT-based chemiresistor was reported by Cattanach et al.[278] This sensor was functionalised with a chemo-selective polymer, poly(isobutylene) (PIB), to act as a barrier to screen out high concentrations of interferant vapours while allowing the detection of organophosphate-based chemical warfare agent (CWA) simulants like DMMP and diisopropylmethylphosphonate (DIMP). This system provided the versatility of retaining sensor response at large bending angles, due to the poly(ethylene terephthalate) (PET) substrate, and provided chemo-selectivity for the nerve agent simulant target compounds; albeit with no attempt to drive down detection limits. Other CNT/polymer systems have been reported such as poly(3-methylthiophene) for the detection of chloromethanes,[279] poly(vinyl acetate) (PVA) on vertically aligned CNTs for sensitivity towards a suite of organic compounds including N,N-dimethylformamide [280] (DMF), carbon tetrachloride (CCl4) and chloroform (CHCl3), and PMMA for the [281] detection of dichloromethane (DCM) and CHCl3. The selectivity that a tailored polymer can offer to a CNT sensor system is rapidly being realised and will continue to be a prominent feature in future research.

A novel CNT-sensor methodology, was developed by Snow et al., was capable of simultaneous conductance (G) and capacitance (C) measurements. Experimental trials on a variety of organic vapours showed that the ratio of ΔG/ΔC could be utilised to incorporate further orthogonal dimensions of data for mixture separation.[236] Another methodology for increased dimensional space for mixture separation is the utilisation of multiple-arrays of sensors. One of these systems includes a multiple-array of CNTs loaded with a variety of different metal nanoparticles, and monitoring of the array analyte-specific patterns could be utilised to detect different gases, with separation using principal component analysis (PCA).[238] In addition, Li and coworkers utilised a 32-element CNT sensor array containing a mixture of different dopants and chemo-selective polymers, which differentiated, using different principal components, compounds such as NO2, chlorine (Cl2), benzene, hydrogen chloride (HCl) and acetone.[261] Multiple-arrays and the use of chemometric processing will continue to be developed and exhibit prominence in future scientific studies for the successful differentiation of vapour-phase mixtures.

Finally, attention is drawn to a recent report which has the potential to revolutionise the incorporation of SWCNTs in analytical chemistry with the quest for instrument 34 Chapter 1 miniaturisation. Strano and coworkers,[179] successfully engineered a microelectromechanical system (MEMS)-based micro gas chromatography unit coupled with a functionalised-SWCNT sensor onboard a single microchip. The SWCNTs were functionalised with polypyrrole (PPy) and by using a He carrier gas and injecting a DMMP headspace sample, a measureable and reversible response of as little as 109 DMMP molecules was achieved.[179] This represents a significant step forward for CNT-based gas and vapour sensors, in terms of their incorporation in current analytical technologies.

1.4.1.4 Overview of analyte target areas and the current literature

In Section 1.4.1 a large number of sensor systems for both gas and vapour analyte detection has been presented, however, there are three prominent target areas where the large majority of CNT-sensor literature is focused, namely in environmental gases, medical gases and organic vapours, as well as organic vapours which are of military or chemical warfare relevance.

The environmental gases were the first ‘class’ of analytes focused towards CNT-based sensor detection. As previously mentioned, a large majority of the literature has focused on NO2 and NH3, which adversely effect both human health and the environment. Other environmental gases include H2, which can become explosive in air, methane (CH4), which is a powerful greenhouse gas, CO, sulfur dioxide (SO2) and hydrogen sulfide (H2S) which can have a significant environmental impact and are directly related to hydrocarbon refining. Studies into O2 sensors although less prominent are, however, envisaged to have implications in the automotive industry.

A CNT-based sensor niche has been developing in relation to analytes of medical relevance. A large application focus is the development of low-cost sensors which can be utilised for point-of-care breath analysis of patients for the monitoring of gases such as CO2 and NO for health and preventative disease monitoring. This has also been extended to ethanol, with accurate low-power sensors for monitoring a person’s blood alcohol level. The monitoring of a range of organic vapours has been undertaken, as the

35 Chapter 1 presence of particular compounds can lead to diagnosing conditions such as, but not limited to, breast cancer, liver disease, and diabetes.

The final ‘class’ of compounds is highly relevant in the current volatile global environment and relates to chemicals which poses a threat to national security. The unique sensing mechanism based on P/P0, is attractive for the detection of low-vapour pressure compounds, which typically are associated with explosives and some CWAs. This is the main focus of the design, development and testing of the SWCNT chemiresistor devices described in this thesis (Chapters 3 and 4). A detailed breakdown of current technologies is given in Chapter 4 and for an extensive overview of current CNT-based sensor technologies and capabilities the reader is directed to the following reviews.[151, 219, 226]

There is clearly a gap in the sensor literature, and in the marketplace, with respect to a low-cost, remote deployable sensor capable of detecting CWAs, and this is the ultimate directional aim of the research reported in this thesis. The exponential increase in publications surrounding CNT-based sensors has occurred because researchers and industry alike envisage the significant advantages that these carbon nano-materials possess over other sensing technologies.

1.4.1.5 Advantages over other gas and vapour sensing technologies

One of the major advantages that makes CNT-based gas and vapour sensors so appealing in contrast to current technologies, such as solid electrolyte (SE) or metal oxide semiconducting (MOS) sensors, is that the sensing mechanism is functional at room temperature. This not only increases the operational safety of CNT-based sensors in explosive environments, but also reduces the power consumption of the devices, ultimately leading to battery powered portable sensor units. There is also reduced thermal drift in CNT-based sensors in comparison to conventional sensors. Economically, CNT-based sensors offer a low-cost alternative for most sensing applications and coupled with reduced device dimensions facilitate the development of portable, wearable, even disposable sensors right through to high-end lab-on-a-chip scenarios for miniaturisation of analytical instrumentation. Fast analyte response times

36 Chapter 1 and the reduced requirement of pre-concentration protocols also significantly adds to the appeal of this technology, when compared to traditional techniques such as GC-MS. In the case of CNT-based sensors, qualitative, or semi-quantitative measurements can be obtained in a fraction of the time of traditional techniques, and at a fraction of the cost. Moreover, the very nature of the CNT allows significant flexibility in terms of tailoring a system, through various functionalisation techniques, to suit a specific application. With the recent advent of flexible substrates with optical transparency these sensors could be envisaged as coating pipes or entire walls to monitor industrial leaks, or public transport areas to augment counter terrorism measures. In addition, future developments could lead to fabrication of functional nodal networks based on ‘smart dust’ principles for chemical mapping and providing high quality military intelligence. It is a combination of all these properties, with major emphasis on reduced dimensions and low power operation, which highlight CNT-based sensors as the next generation gas and vapour sensors that have the potential to revolutionise the current sensor market.

1.5 Supramolecular chemistry

A significant underlying theme of the research presented in this thesis relates to specific aspects of supramolecular chemistry, in particular non-covalent calixarene chemistries. A general background to supramolecular chemistry will now be provided, with attention specifically focused on calixarene-carbon nano-material interactions spanning to calixarene-based metal cation sensor systems.

The field of supramolecular chemistry is focused on the “chemistry of molecular assemblies and of the intermolecular bond”,[282] and is concerned with non-covalent interactions between independent molecular entities which form assemblies of molecules know as supermolecules.[282] These themes are based around the reversible nature of formed supramolecular assemblies and are vastly different from the traditional molecular arrangement which are based around covalent bonds. The versatility of these non-covalent interactions gives rise to the possibility of user-defined molecular architecture. Providing an additional tool in the chemist’s arsenal for the realisation of Richard P. Feynman’s[283, 284] concept of a bottom-up approach and the encroachment into the realms of nanotechnology.

37 Chapter 1

1.5.1 Molecular recognition and host-guest chemistry

The general underlying concepts within supramolecular chemistry and its links to traditional covalent chemistry are illustrated in Figure 1.22. The receptor – substrate component draws links from the late 19th Century philosophies of Nobel laureate Emil Fischer, with the proposed enzyme-substrate “lock and key” interactions.[285] In a way the origins of supramolecular chemistry are derived from studies of intriguing biological systems and the ability of these systems to form a higher order of complexity. The breakthrough from a chemical sciences point-of-view was the pioneering work of Cram, Lehn, and Pedersen, with the development of synthetic selective ‘host-guest’ complexes and the authors were eventually awarded the 1987 Nobel Prize for Chemistry.

CHEMISTRY

MOLECULAR SUPRAMOLECULAR

SELF-ASSEMBLY SELF-ORGANISATION A B

SYNTHESIS polymolecular RECEPTOR RECOGNITION assemblies covalent bonds C INTERACTION MOLECULAR AND D SUPERMOLECULE TRANSFORMATION SUPRAMOLECULAR intermolecular DEVICES bonds

SUBSTRATE TRANSLOCATION

FUNCTIONAL COMPONENTS Figure 1.22 A flow chart linking traditional molecular chemistries and developing supramolecular chemistries. Adapted with permission from [286], © 1989 VCH Publishers.

Molecular recognition is defined by the energy and the information involved in the binding and selection of guest(s) by a given host molecule, it may also involve a specific function.[287] For host-guest interactions to occur the host molecule must have the appropriate binding sites for binding guest molecules, and hence allow selective recognition by the host.[288] Molecular recognition takes simple binding one step further and binding events must meet the requirements of both steric and electronic complementarity between host and guest. Consequently, molecular recognition involves maximising both the thermodynamic and kinetic stability of a supermolecule through well defined intermolecular interactions, hence “binding with a purpose”.[286]

38 Chapter 1

These intermolecular interactions include electrostatics, hydrogen-bonding, π-π stacking, dispersion forces and induction forces, and hydrophobic or solvatophobic effects.[286, 289, 290]

Generally speaking, the host molecule will consist of a larger molecular or aggregate of molecules possessing a central hole or cavity with convergent binding sites, much like a synthetic macrocyclic ligand.[291] The guest can be a wide range of structures ranging from a monoatomic ion to complex organic multifunctional species such as a protein. In order to maximise the host-guest interactions several factors must be taken into consideration, as follows:[286]

 Steric (shape and size) complementarity; between the host and the guest, for example the presence of convex and concave domains in the correct location on each molecule.

 Interactional complementarity; to have the appropriate attractive forces between the host and guest, such as electronic and electrostatic.

 Large contact areas; this leads to maximisation of the possible interaction to encourage and enhance binding efficiency.

 Multiple interaction sites; as non-covalent interactions are relatively weak, the possibility of multiple sites can lead to stronger cooperative binding.

 Strong overall binding; high stability, in principle, does not indicate high selectivity. However, this is typically the case as host-guest free energy of binding differences are likely to be larger with high efficient binding.

 Medium effects; host-guest interactions can be influenced by the solvation medium, thus the host and guest should present geometrically matched solvophillic/solvophobic or solvophillic/solvophillic domains.

In relation to host-guest interactions and molecular recognition there is a large variety of molecular receptors that have previously been investigated. The main receptor type reported in this thesis is the macrocyclic ligand, more specifically calix[n]arenes; this concept will be discussed in Section 1.6.

39 Chapter 1

The concept of host-guest molecular recognition is described in Chapter 2 and describes a novel methodology for tackling the current problem of separation of SWCNTs based on electronic structure. The concept of host-guest recognition will be further elaborated upon in Chapter 6 where a water-soluble fluroionophore system is designed and applied to cation sensing in aqueous media, an area severely lacking in the current literature.

1.5.2 Self-assembly

The pioneering studies by Lehn et al. on complexation of oligobipyridine ligands with Cu+ ions led to interesting ‘double-stranded helicates’, reminiscent of the DNA double helix.[292] This was seen as a ‘self-assembly process’ or ‘operating through self-processes’ and led to the implementation of self-assembly and self-organisation in supramolecular chemistry.[293, 294] This branch of chemistry exploits designed systems which are capable of spontaneously generating/organising into well defined, functional supramolecular architectures.[286] Self-assembly allows access to new molecular architectures that are practically inaccessible using traditional multi-step covalent chemistry.[289]

For self-assembly to occur the host and guest are typically pre-arranged to have the required complementarity where, under certain conditions, they form into supermolecules and propagate into higher-order assemblies. One of the many efficiencies of self-assembly is that during the organisation process if a thermodynamically unfavourable side product is formed the network has the ability to dis-assemble and re-assemble again until a thermodynamically stable product is obtained. This is in direct contrast to traditional covalent chemistry were errors in bond formation can not be self-corrected. This disassembly process has recently been exploited by Raston et al.[295] The authors describe a situation where self-assembly of C-alkylpyrogallo[4]arenes in chloroform hydrogen-bond to form hexameric capsules. These capsules are ‘dis-assembled’ utilising high shear forces while under an atmosphere of hydrogen, spontaneous re-assembly occurs, trapping molecular hydrogen in the internal cavity of the molecule. Under these conditions the capsule is termed a “nano-container”.[295]

40 Chapter 1

Self-assembly of particular systems may result in a range of collective properties, and produce structures with inherent properties superior to the binary precursors.[286] The application of self-assembly processes in building unique molecular architecture is described in Chapter 5.

1.6 Calixarenes

Calix[n]arenes are a family of synthetic macrocyclic receptors consisting of cyclic arrays of n phenol moieties linked by methylene groups.[289] The chemical process discovered for the calixarene class of compounds dates back to the phenol and formaldehyde reaction studies of Adolf von Baeyer in 1872.[296-298] Over the years other chemists modified this reaction. However it was not until the 1970’s that Gutsche and coworkers revived the chemistry of the cyclic oligophenol products in the hope of producing a range of cavity-containing molecules for enzyme mimics.[290] A range of oligomeric structures were synthetically obtained and structural investigation led to the identification of ‘bowl-like’ structures which Gutsche coined “calixarene” due to their resemblance to a Greek vase called a calix crater and arene representing the presence of aryl units.[299] For a more extensive coverage of the history of calixarenes the reader is directed to a prominent book describing the subject in detail.[300]

1.6.1 Synthesis and mechanism of formation

Several methods have been developed for the synthesis of calixarenes, with the most general and useful being the one-step, base catalysed condensations of para-substituted phenols and formaldehyde, yielding a mixture of cyclic oligomers (Scheme 1.1). The use of a para-substituted phenol, more specifically, those substituents being electron donating with respect to the phenolic hydroxyl group, are typically essential for the reaction to successfully proceed.

R R

base + H2CO

n OH OH R = alkyl, aryl n = 4, 5, 6, 7, 8, 9, ... Scheme 1.1 General base catalysed synthesis scheme for calixarene formation.

41 Chapter 1

As mentioned the ‘one pot’ calixarene synthesis has been the most popular route and has allowed access to a wide range of para-substituted calixarenes (Table 1.2). The ‘one pot’ synthesis route and slight modifications to this methodology are the approach taken for the synthesis of the calixarenes utilised for the studies reported in this thesis.

Table 1.2 'One pot' synthesis route yields for a variety of para-substituted calixarenes. Adapted with permission from [301], © 1995 John Wiley & Sons, Inc.

R n = 4 n = 5 n = 6 n = 7 n = 8

Me 22[302]

Et 24[302]

iPr 10[303] 26[303]

tBu 49[304] 10 - 15[305, 306] 83 - 88[307] 6 - 17[308, 309] 62 - 65[310]

tPentyl [a] 6 - 7[311] 30[311] 37 - 41[311]

tOctyl [b] 31[312] 30[313]

Adamantyl 71[314]

Phenyl 9[315] 5[315] 10[316] 7 - 14[316]

Benzyl 60[317] 33[318] 16[318] 16[319] 12[318]

Cumyl 21[320]

tPentyl [a] = 1,1-dimethylpropyl, and tOctyl [b] = 1,1,3,3-tetramethylbutyl

Besides the aforementioned ‘one-pot’ synthetic procedure various calixarene analogues can be formed through indirect methods. The most notable are the step-wise synthesis to form a monomeric chain prior to cyclisation,[321-328] fragment condensation which induces the fusion of two or more monomeric fragments,[329-332] and heteroatom-bridging which plays on the aryl fragment bridging using a heteroatom instead of the more conventional methylene linker.[333-337]

42 Chapter 1

The proposed calixarene formation mechanism is highly dependant on the synthesis route being based on thermal induction or either base or acid catalysed. The base catalysed reaction proceeds through the formation of the phenoxide ion, which allows the nucleophilic attack of formaldehyde to yield a hydroxymethyl phenol (Scheme 1.2). The hydroxymethyl phenol is converted to an o-quinonemethide intermediate that proceeds to react with a phenolate ion to form the respective diarylmethyl compound. This diarylmethyl compound is subject to a similar set of transformations to form linear oligomers (Scheme 1.2). These linear oligomers are cyclised to form calixarenes, however the cyclisation mechanism is quite complicated and not fully understood. Interestingly, at least three dozen of the linear oligomers are required before cyclisation will occur.[338]

Scheme 1.2 Base catalysed mechanism for the formation of linear oligomers prior to cyclisation. Adapted with permission from [300], © 1989 The Royal Society of Chemistry.

The kinetic product of the based catalysed reaction of p-tBu-phenol and formaldehyde is p-tBu-calix[8]arene. The suggested mechanism of formation of this calixarene is through the formation of a cyclic hydrogen-bonded dimer, from the pairing of two linear tetramers.[339, 340] The thermodynamic product by the base catalysed reaction has been shown to be the lower oligomer calixarenes, predominantly the calix[4]arene. The proposed mechanism for the formation of the thermodynamic product, p-tBu-calix[4]arene, is through molecular mitosis of p-tBu-calix[8]arene at high

43 Chapter 1 temperature.[341] It should be noted that other mechanisms such as fragmentation-recombination processes have been suggested. Gutsche et al. experimentally attempted to verify the molecular mitosis hypothesis by refluxing both deuterated and non-deuterated p-tBu-calix[8]arene. These experiments afforded products which indicate both suggested mechanisms were in action in the formation of p-tBu-calix[4]arene (Fig. 1.23). In comparison, the proposed formation mechanism for p-tBu-calix[6]arene involved the cyclisation of a linear hexameric chain, as opposed to a hydrogen-bonded dimer of two linear dimers.[342]

Figure 1.23 Overview of the molecular mitosis hypothesis experiments. The white- and black-filled circles represent protonated and deuterated residues respectively. Adapted with permission from [341], © 1999 American Chemical Society.

The proposed mechanisms for the thermal-induced and acid catalysed reactions have been studied to a lesser extent and are not as well understood. For the purpose of the studies contained in this thesis the base catalysed mechanism is sufficient to cover the formation of all calixarenes utilised, and therefore the other mechanistic routes will not be detailed.

1.6.2 Calixarene structural conformations

Due to the rotational mobility about the σ-bonds of the methylene links between the aromatic moieties, calixarenes typically exhibit a high level of conformational

44 Chapter 1 flexibility. Factors such as increasing calixarene ring size, reducing the para-substituent size, temperature, and varying solvent systems can all lead to increased conformational flexibility, and can effect the rate of interconversion.[343, 344] In the specific case of the unmodified calix[4]arene there are four possible conformations that have been pointed out by Cornforth and coworkers,[345] and named by Gutsche as cone, partial cone, 1,3-alternate, and 1,2-alternate (Fig. 1.24).[300]

OR

OR OR OR RO OR OR OR

cone partial cone

OR OR OR OR

OR OR OR RO

1,3-alternate 1,2-alternate Figure 1.24 Structural conformations of calix[4]arene derivatives.

Out of the four possible conformations for calix[4]arene, this macrocycle will typically adopt a cone conformation in solution and solid-state, due to the stabilisation of the lower rim hydroxyl functionalities through intramolecular hydrogen-bonding.[289] Structural investigations to determine conformations can readily be achieved through Proton Nuclear Magnetic Resonance (1H NMR)[344] and 13C NMR.[346]

Calix[5]arenes usually adopt the cone conformation,[347-349] as seen with the calix[4]arenes in both solution and solid state. p-tBu-calix[5]arene adopts a slightly distorted cone structure in the solid-state, which approximates to five-fold symmetry.[348] It should be noted that the increased ring size of the calix[5]arenes results in weaker hydrogen-bonding interactions between the lower rim hydroxyl groups, as a result, the cone structure is less stable at room temperature than the calix[4]arenes. However, calix[5]arenes can be conformationally blocked by functionalisation at the lower rim.[350] The adopted conformation of the p-tBu-calix[5]arene as an inclusion complex is the chief concept within Chapter 5.

45 Chapter 1

For calix[n]arenes larger than calix[5]arene the increase in ring size has a significant effect on the conformational flexibility of the molecule and there is a departure from the cone conformation preference observed with the lower ring sizes.[344] p-tBu-calix[6]arene and p-tBu-calix[8]arene can adopt 8 and 16 main conformations respectively, however, 1H NMR data indicates that the pinched cone and the pleated loop are the preferred respective stable conformations.[344] The increased conformational flexibility of the larger ring size calixarenes is exploited in Chapters 2 and 6 to provide solutions to high impact chemical problems.

1.7 Calixarene–fullerene solid-state interactions

1.7.1 Calix[3,4,6,8]arenes

For more than a decade the supramolecular chemistries involving fullerenes for not only fabrication of novel inclusion complexes, but also the design of multi-dimensional molecular architecture, has captured the attention of scientists’ world-wide. Over the years, it has been found that the icosahedral fullerene C60 readily forms intercalation complexes that are structurally diverse, producing materials containing small molecules such as simple aromatics like hydroquinone,[351] through to larger molecules like ferrocene,[352] and ‘bowl shaped’ receptor molecules including γ-cyclodextrin,[353] modified calixarenes,[354] cyclotriveratrylene and related molecules,[355] and porphyrins and metal macrocycles.[356, 357] Through the use of these systems the manipulation of fullerene···fullerene centroid distances can assemble the C60 molecules into one- and two-dimensional arrays, as well as continuous three-dimensional extended networks (Fig. 1.25). For the purposes of this thesis only the solid-state interactions of different fullerenes and calixarenes, with accompanied structural elucidation or proposed models, will be detailed; these studies describe only portions of the fullerene C60 assemblies, as illustrated in Figure 1.25. For a more comprehensive overview of fullerene supramolecular assembly the reader is directed to recent reviews.[358, 359] It should also be noted that calixarene systems have been utilised to produce inclusion complexes and [355, 360- provide avenues for molecular architecture for higher order fullerene such as C70, 362] [363-365] C76 - C120, however to cover such areas is outside the scope of this thesis.

46 Chapter 1

Figure 1.25 Schematic of the diversity of fullerene C60 ordering into one-, two-, and three-dimensional arrays. Host molecules have been omitted for clarity. Adapted with permission from [358], © 2003 John Wiley & Sons, Inc.

A tremendous increase in interest in the complexation of the bowl-shaped calixarene molecules with fullerenes resulted from the discovery that a discrete 1:1 complex forms between p-tBu-calix[8]arene and fullerene C60, which essentially allowed purification of [354, 366] this fullerene from fullerite. Since this discovery, calixarene-C60 complexes have been shown to form a variety of solid-state structures depending on the ring size, upper rim para-substituents, choice of solvent system and, interestingly, the presence of other solutes which induce different fullerene packing, yet they are not incorporated into the final structure.[358] The calix[3]arene derivative, p-benzylhexahomooxacalix[3]arene, will only be briefly mentioned as this shallow cavity system leads to full encapsulation of a single fullerene molecule,[367] and is the only structurally elucidated system in the calix[3]arene class.

As previously mentioned, calix[4]arene readily adopts a cone conformation, which would typically entice the electron deficient C60 molecule into the electron rich cavity. However, the hydrophobic cavity size in the calix[4]arene is too small to accommodate the C60 molecule and it is found that the C60 binding is exo relative to the cavity in the solid state.[368-370] A case in point is the complexation phenomenon between fullerene

C60 and p-benzylcalix[4]arene, where, even with the deeper cavity calixarene with

47 Chapter 1 functionally mobile benzyl functionality, the C60 molecules reside outside the cavity (Fig. 1.26). [370] Even exo-cavity binding by the calix[4]arene can generate novel two-dimensional C60 arrays, in this specific case revealing flat C60 sheets sandwiched between cavitand layers. Atwood et al. previously documented a similar two-dimensional C60 flat sheet arrangement through complexation with p-iodocalix[4]arene benzyl ether.[368] Furthermore, by changing the upper rim halogen and lower rim functionality of the calixarene, p-bromocalix[4]arene propyl ether complexes with C60 to produce one-dimensional, well separated linear columns of close-spaced C60. This indicates how subtle structural changes of the cavitand can have a dominating effect on the solid-state packing structure, and allows the construction of different architectures at the molecular level.

Figure 1.26 Top view projection down the c axis of the extended structure of fullerene C60 and calixarene layers in the (C60)2(p-benzylcalix[4]arene), with red dotted lines indicating C-H···π interactions. Adapted with permission from [370], © 2006 The Royal Society of Chemistry.

Calix[6]arene also forms inclusion complexes with fullerenes, with the larger ring size allowing more conformation flexibility than the calix[4]arene. The increase in ring size, essentially increases the hydroxyl group spacing and weakens the intermolecular hydrogen-bonding of the lower rim, allowing for a higher level of complexity of the

48 Chapter 1 solution and solid-state structures available.[371] To the author’s knowledge there are only two structurally elucidated calix[6]arene/C60 complexes in the solid-state. Firstly, calix[6]arene forms a 2:1 complex with C60, (C60)2(calix[6]arene)(toluene), in which the fullerenes form an extended three-dimensional crisscrossed array of linear strands (Fig. 1.25). [355] Interestingly, the molecular arrangement is also mimicked when complexation of C70 is undertaken. The existence of such a complex had been previously suggested by Shinkai et al. utilising UV-visible spectrophotometry characterisation.[372] The second elucidated structure is that of the inclusion phenomenon between p-benzylcalix[6]arene and C60, which forms a fullerene enriched [371] 3:1 complex, (C60)3(p-benzylcalix[6]arene)(toluene)4.25. This system forms a complex three-dimensional network of fullerenes, which actually shroud columns of the calixarenes (Fig. 1.27).

Remarkably, the calixarenes can be selectively dethreaded out of the crystal structure in the presence of methylene chloride to form an all-carbon hexagonal close-packed (hcp) polymorph of fullerene C60. This approach is indeed pioneering research in the fact that results suggest a porous three-dimensional fullerene network can be engineered through supramolecular origins (Fig. 1.27).

In terms of fullerene binding in relation to the calix[6]arene cavity the two aforementioned structures differ. The unmodified calix[6]arene adopts endo-cavity binding of two fullerene molecules in the shallow cavity of the pinched double cone,[355] whereas the p-benzylcalix[6]arene, which adopts the same conformation, has pendant arms of other calixarenes residing in the cavities resulting in exo-cavity fullerene binding.[371] It should also be noted that in both complexes the unsubstituted lower rim of the calixarene is important and the fact that both complexes adopt the pinched double cone conformation, re-iterates the requirement of maximising the lower rim hydroxyl interactions.

49 Chapter 1

Figure 1.27 (top) Elucidated solid-state structure for (C60)3(p-benzylcalix[6]arene)(toluene)4.25, and (bottom) the same structure with the calixarene and toluene molecules removed to indicate the extent of the three-dimensional C60 network and the virtual porosity. Fullerenes are coloured purple, toluene molecules are in blue; calixarenes are shown in ball-and-stick projected down the c axis. Adapted with permission from [371], © 2008 American Chemical Society.

50 Chapter 1

Figure 1.28 (a) Structure of complex (C60)3(p-benzylcalix[6]arene)(toluene)4.25, (b) the same structure with the calixarene and toluene molecules; both images show the pinched double cone conformation of the p-benzylcalix[6]arene. Adapted with permission from [371], © 2008 American Chemical Society.

As previously mentioned, the 1:1 complex between p-tBu-calix[8]arene and C60, which led to an economical purification route, sparked initial calixarene complexation studies. Solid-state structure elucidation has not currently been possible, however, the complex has a predicted structure based on the calixarenes in the double cone conformation, with three calixarenes shrouding a triangular array of fullerenes in a micelle-like structure.[354, 373] Interestingly, Bai et al. utilised a calix[8]arene derivative to form a 1:1 complex with fullerene C60, and illustrate through high-resolution STM (HR-STM) highly ordered monolayers on a Au(111) surface (Fig. 1.29).[374] The lack of solid-state structural literature on the calix[8]arene system indicates the difficulty when dealing with the larger, more flexible calixarene molecules, ultimately limiting the extended packing structures available for analysis with current characterisation instrumentation.

Figure 1.29 (a) STM image of the C60/calix[8]arene derivative, and (b) the proposed structural model of the ordered array resulting in the observed STM image; scale is given in nm. Adapted with permission from [374], © 2003 John Wiley & Sons, Inc.

51 Chapter 1

1.7.2 Calix[5]arenes

The calix[5]arene class of host molecules have a dominant role in the fullerene C60 complexation literature. This is owing to the principal calix[5]arene structure which prefers the cone conformation, having close to ideal complementarity of size and surface curvature relative to C60. In addition, the alignment of the C5 symmetry axis of the calix[5]arene, in this conformation, with the C5 symmetry axis of the C60 is favoured energetically as it optimises host-guest π···π interactions.[367, 375-377] There are numerous solution studies on different calix[5]arene-fullerene C60 systems documenting possible interaction.[372, 378-380] However, the focus of the current section is on solid-state fullerene architecture and as such these studies will not be included.

To the author’s knowledge the first structure-elucidated calix[5]arene-C60 complex, was with the calix[5]arene, upper rim functionalised with three methyl and two iodo groups [376, 377] (1,3 disposition) in the para position. This produced full C60 encapsulation with the C60 molecules completely isolated from one another. This is also the case with p-benzylcalix[5]arene, which forms a 2:1 complex with C60, where an individual C60 molecule is shrouded by two staggered trans-host molecules in the cone conformation with dangling benzyl groups of two calix[5]arenes fully shrouding a single C60 molecule [367] (Fig. 1.30a). The endo-cavity binding of C60 to this particular calix[5]arene was predicted earlier by Raston et al., however the molecular mechanics calculations had higher association of the upper rim benzyl functionalities with the C60 than was observed in the solid-state structure.[375] More recently, the analogous p-phenylcalix[5]arene has also been shown to form a 2:1 complex with C60, revealing [381] individual C60 isolation (Fig. 1.30b).

Figure 1.30 Encapsulation of an individual fullerene C60, a) (C60)@(p-benzylcalix[5]arene)2 where C60 is represented in space-filling form with the calixarene molecules in stick form, and b) (C60)@(p-phenylcalix[5]arene)2 where C60 and calixarene molecules are represented as space-filling form. Adapted with permission from [367], © 1999 John Wiley & Sons, Inc., and permission from [381], © 2002 The Royal Society of Chemistry, respectively. 52 Chapter 1

It was not until the pioneering work of Atwood et al., that the real potential of fullerene architectures and complex macroscopic arrays using unmodified calix[5]arene were realised.[358, 382] There are two different fullerene packing arrangements in the solid-state, interestingly with 1:1 host-guest stoichiometry. The system with lesser structural complexity, (C60)(calix[5]arene)(toluene), is comprised of slightly helical, zigzag, one-dimensional oligomeric arrays of C60 (Fig. 1.31a). Intriguingly, this structural polymorph only forms in the presence of other globular molecules such as

C70, C84 and o-carborane, yet these components are not incorporated into the final [358] structure. The other polymorph forms with a high ratio of C60 and gives rise to a 4:5 complex, (C60)4(calix[5]arene)5(toluene)2 which assembles C60 molecules into unique linear columns of five-fold, Z-shaped strands (Fig. 1.31b),[382] which to date has not been replicated by any other host-guest system involving fullerene C60. This system is unique in having multiple fullerene···calixarene interactions, including dual calixarene shrouding “ball-and-dual-socket”, half-cavitand binding “ball-and-socket”, and no direct calixarene interaction.[382]

Figure 1.31 a) Crystal structure of the (C60)(calix[5]arene)(toluene) complex illustrating the columnar, slightly helical zigzag fullerene array, and b) the extended crystal structure of the (C60)4(calix[5]arene)5(toluene)2 complex with the projection parallel to the 5-fold, Z-shaped columns [358] of C60. Adapted with permission from , © 2003 John Wiley & Sons, Inc., and permission from [382], © 2002 American Chemical Society, respectively.

The ability of calix[5]arene and analogues to form novel arrays of fullerene C60 sparked the investigation into p-tBu-calix[5]arene and its suitability as a host molecule for C60. Not only was inclusion complexation apparent but a novel fullerene architecture was observed, unlike any previously mentioned system. Through additional manipulations

53 Chapter 1 of this complex, an all-carbon architecture was successful obtained and current applications extend as far as a possible chemotherapy synergistic agent; details of this system and concepts are discussed in Chapter 5.

1.8 Functionalisation of calixarenes – water solubility

The functionalisation of calixarenes to render them soluble in aqueous media has been extensively researched over more than two decades, predominately to substantiate concepts based around synthetic enzyme mimics.[300] Water-soluble calixarenes have now become an important class of compounds, and various synthetic routes have been established to impart water solubility to the calixarenes through the addition of water soluble groups to either the lower or upper rim. The first of such functionalisation strategies was undertaken by Ungaro et al. through lower rim functionalisation of p-tBu-calix[4]arene with carboxylic acid groups.[383] This functionalisation resulted in moderate compound solubility in water of between 5 x 10-4 M and 5 x 10-3 M, depending on the cation used.[383] Since then calixarenes have been made water soluble by a variety of upper and lower rim functional groups additions including diboronic acids,[384, 385] amino,[386] sulfonic acids (Fig. 1.32), [315, 317, 387, 388] and phosphonic acids (Fig. 1.32). [389-394] For a more detailed overview of water soluble calixarenes the reader is directed to an excellent general review of water-soluble calixarenes,[395] more specifically sulfonated,[396] and phosphonated calixarene systems.[397]

SO3H

SO3H

P(O)(OH) SO3H 2

n n n n OH OH OH OH Figure 1.32 Water-soluble p-substituted calixarenes ranging from shallow cavity (n = 4, 5, 6, 8) to deeper cavity 'extended arm' sulfonated calix[n]arenes (n = 4, 6, 8) to phosphonic acid calix[n]arene (n = 4, 5, 6, 8).

The chief water-soluble calixarene systems described in this thesis are the first generation p-sulfonated calixarenes, the next generation ‘extended arm’ sulfonated calixarenes and the p-phosphonated calixarenes. It should be noted that until the

54 Chapter 1 pioneering synthetic work of Clark et al. only the p-phosphonic acid calix[4]arene was accessible. These recent works have not only given access to larger ring size p-phosphonated calix[n]arenes, n = 5, 6, 8, but have generated a new five-step synthetic route (Scheme 1.3) to produce the calixarenes in high yields.[393, 394]

Br Br

Br2 Ac2O

n n n OH OH OAc

P(OEt)3

NiCl2

P(O)(OH)2 P(O)(OEt)2 P(O)(OEt)2 BTMS KOH

n n n OH OH OAc Scheme 1.3 A new five-step synthesis of p-phosphonic acid calix[n]arene, n = 4, 5, 6, 8. Adapted from [393, 394]

1.9 Applications of water-soluble calixarenes

The need to develop and investigate new classes of water-soluble calixarenes is paramount for expanding the area of supramolecular chemistry and keeping it at the forefront of science. A constant theme running through this thesis it an attempt to overcome current problems areas in scientific and commercial areas, by providing novel solutions. These classes of water-soluble calixarenes have been adapted and studied in an attempt to fill current gaps in the literature, and the requirements of the scientific community. This is highlighted in Chapter 2, where the use of water-soluble calixarenes not only effectively solubilises individual SWCNTs into aqueous media but also allows selective-diameter uptake. This selective nanotube diameter uptake allows the choice of tuning the electronic properties of the resulting solutions through the calixarene, thereby providing advancement for areas such as carbon nano-electronics and device fabrication. Furthermore, there is a significant gap in the current literature in the design and development of water-soluble fluoroionophore systems. This is an area where water-soluble calixarenes conceptually show promise. A suitable system based

55 Chapter 1 on the unique properties of a recently synthesised ‘extended arm’ sulfonated calixarene is described in Chapter 6.

1.10 Specific aims and structure of the thesis

As indicated in this introductory section, the specific aims contributing to the research described in this thesis are as follows:

 To develop novel purification methodologies for the post-growth treatment of CNT materials, with specific focus on carbon nano-electronics applications;

 To develop an appropriate device platform for the assessment of carbon nano-materials for their gas and vapour sensing characteristics, ultimately to produce a carbon nano-material-based gas and vapour sensor;

 To design and fabricate an appropriate sensor testing facility with precision gas and vapour-phase analyte delivery capabilities;

 To assess a variety of carbon nano-materials, more predominately SWCNTs, utilising the fabricated device platform and testing facility;

 To extend the field of all-carbon architecture materials though the fabrication of novel structures, utilising methodologies ranging from unique growth mechanisms to supramolecular interplay;

 To develop and extend supramolecular principles to the liquid-medium sensor domain to fill current gaps in the literature.

The structure of this thesis follows the primary aims listed above. A brief summary of the proceeding thesis chapters will be detailed as follows.

Chapter 2, “Diameter-selective sorting of SWCNTs in aqueous media”, describes novel purification strategies employed to solubilise SWCNTs into aqueous media. Chapter 2 describes a comprehensive Raman spectroscopy characterisation study which suggests that selective supernatant enrichment of specific nanotube diameters is possible using water-soluble p-sulfonated and p-phosphonated calixarenes. This observation is

56 Chapter 1 additionally supported through the electrical characterisation of fabricated on/off field effect transistors (FET), from semiconducting enriched calixarene/SWCNT supernatants.

Chapter 3, “Design, fabrication and verification of a gas and vapour analyte delivery system for sensor testing”, illustrates the process of developing all aspects of fabricating a gas and vapour testing apparatus from the ground up. This was an essential requirement for the successful characterisation of the developed CNT sensors, and is a prominent source of data described in subsequent chapters. Within this chapter details are given on systems troubleshooting, in relation to electrical noise. In addition, verification of the system will be achieved through the use of a novel SWCNT/nanoparticle hybrid sensor.

Chapter 4, “SWCNT-based chemiresistor sensor testing for compounds of national security interest”, describes the application of the developed CNT sensors for the detection of chemical compounds of national security interest. This chapter elaborates on the conceptual design and fabrication strategies involving the device platform, and CNT-based chemiresistor sensor. The concepts behind harnessing the properties of CNTs to provide low-power sensors to be utilised in the military/counter-terrorism domain are described. Chapter 4 describes the optimisation of both the device, and testing parameters, to obtain quality results upon gas and vapour exposure trials. The development of a novel SWCNT electrochemical sensor for the detection of the notorious war gas phosgene is detailed, along with other compounds of interest such as G-series nerve agent simulants. This also shows direct comparisons between different sensor geometries and packaging in terms of sensitivity and limits of detection, which are put into context with currently available technologies.

Chapter 5, “Progress towards novel carbon nano-materials and interesting application areas”, describes characterisation methodologies for structural elucidation of a novel solid-state complex. This chapter details the synthesis of a novel host-guest complex, termed ‘C60 nano-fibres’, and describes the molecular architecture of fullerene

C60 with p-tBu-calix[5]arene. High temperature studies also produced selective volatilisation of the calixarene backbone to yield all-carbon structures with retention of

57 Chapter 1 the principal structure associated with the parent complex. Chapter 5 also detailed this material as a synergistic chemotherapeutic agent for breast and cervical cancer cell lines. Furthermore, a range of other all-carbon materials are described including square-geometry carbon nano-fibres, and rings of helical SWCNT bundles.

Chapter 6, “A water-soluble fluoroionophore: p-phenyl sulfonated calix[8]arene”, details the discovery of autofluorescence properties of p-phenyl-sulfonatocalix[8]arene. The fluorescence properties of this calixarene are systematically mapped and an intensive metal cation binding study undertaken. From these studies it was observed that this was in fact a unique water-soluble fluoroionophore which does not require pendant chromophore functionalisation, and is one of only a few fluoroionophores to be entirely water-soluble. This metal cation sensor is selective for divalent metal cations, with binding constants and degradation studies detailed.

The final chapter, “Conclusions and future studies”, outlines the principal findings of the studies described within this thesis. It highlights the significance of these findings to the scientific community and will propose the future directions in which subsequent research should be targeted.

58

Chapter 2 Diameter-Selective Sorting of SWCNTs in Aqueous Media

Chapter 2

2 Diameter-selective sorting of SWCNTs in aqueous media

2.1 Introduction

Since the discovery of CNTs by Iijima[19] and more recently SWCNTs,[20, 21] these purely carbon-based materials have attracted significant attention from both the research and commercial sectors. The extraordinary mechanical and unique electrical properties of CNTs have been described in Section 1.3, with major research efforts focused on areas such as high performance electronics,[136] scanning probe microscopy,[398, 399] fuel cells,[400] composites,[401, 402] chemical,[223, 235] biological[206, 403] and physical[211, 404, 405] sensors. In order to harness the full potential of CNTs, the ability to separate them according to diameter and/or chirality is required.[136] As previously mentioned, the diameter and inherent chirality of SWCNTs directly controls their semiconducting or metallic properties.[123]

Figure 2.1 Schematic of a SWCNT thin-film transistor (TFT) and structure. A 300 nm layer of SiO2 is deposited on a heavily doped Si wafer, which is functionalised by either an amine-terminated (A) or phenyl-terminated (B) silanes. A solution of SWCNTs is spin coated onto the surface, providing a connective pathway between the source (S) and drain (D) electrodes, with directionality of self-assembled monolayers monitored through AFM. Reprinted with permission from [406], © 2005 AAAS. 60 Chapter 2

Recent methods have been trialled to separate CNTs including pioneering work by Bao et al. which utilises different absorption properties of metallic and semiconducting SWCNTs towards amine- and phenyl-terminated silanes to produce a one-step methodology for self-sorting and aligned TFT (Fig. 2.1). [406] Other methods include dielectrophoresis, which take advantage of the difference in relative dielectric constants of metallic and semiconducting CNTs with respect to the choice of solvent.[407] Pathways utilising differential density gradient systems have also successfully been explored (Fig. 2.2). [408, 409] Other research groups have targeted non-covalent supramolecular interactions, as in biopolymers,[410] DNA-wrapping,[411] porphyrinic polypeptides[412] and oligo-acene adducts.[413] The problem with some of these methods is that the supramolecular chemistry is carried out in organic solvents with limited stability in aqueous media, and require lengthy experimental procedures. Progress has been made to combat these underlying problems with the use of amphiphilic polymers.[414] Supramolecular systems that can be utilised to solubilise SWCNTs in aqueous media, are potentially more benign, generating less waste, and are thus more attractive.

Figure 2.2 Sorting SWCNTs using density gradient ultracentrifugation, with optical absorbance confirming selective diameter fractions which correlate to CNTs with either metallic or semiconducting electronic properties. Adapted with permission from [409] Macmillian Publishers Ltd: Nature Nanotechnology, © 2006.

Calixarenes have played an important role in supramolecular chemistry in general

(Section 1.7). In Chapter 5, a unique structural organisation of fullerene C60, with surface curvature analogous to SWCNTs, and p-tBu-calix[5]arene, to form nano-fibres from toluene solution.[415] These selectivity and self-assembly processes may be mimicked in water using p-phosphonated calixarene systems as a route to diameter-selective purification of SWCNTs in aqueous media. The most widely studied water-soluble calixarenes are the p-sulfonato-derivates which relates to their

61 Chapter 2 bio-compatibility and simplicity of synthesis.[416-418] Recently, Clark et al reported a general route for the synthesis of p-phosphonic acid calix[n]arenes, (n = 4, 6, 8), which can be engineered into unique supramolecular frameworks at the nanometer scale.[393] The combination of water solubility and self-assembly is of interest, for not only the solubilisation of SWCNTs but also the possibility of creating different SWCNT-calixarene arrays to facilitate the separation of the SWCNTs.

There are limited studies on the specific use of water-soluble calixarenes for the solubilisation of SWCNTs. Lobach et al. investigated various types of amphiphilic macrocycles, such as different upper and lower rim functionalised calix[4]resorcinarene and p-sulfonato-calix[n]arenes (n = 4, 6), modified by lower rim alkyl substituents (butyl and dodecyl),[419] with the ability to form stable suspensions of SWCNTs in aqueous media. Furthermore, Ogoshi et al. focused on p-sulfonato-calix[n]arenes (n = 4, 6, 8), with successful solubilisation of SWCNTs for the higher order oligomers, p-sulfonato-calix[8]arene forming a dimer as part of a supramolecular SWCNT network polymer.[420]

The subsequent sections detail the solubilisation of SWCNTs using the water-soluble p-phosphonated calixarenes and analogous compounds of differing lower rim functionality. The findings include a preferential solubilisation of specific diameter SWCNTs, as supported by Raman spectroscopy measurements. In addition, a detailed investigation into the effects of using ‘extended arm’ p-sulfonated calix[8]arene systems on SWCNT solubilisation, for comparison to the earlier studies on the parent p-sulfonated calixarenes is undertaken.[419, 420] It should be noted that the ‘extended arm’ calixarenes have a higher hydrophobic surface area relative to the parent p-sulfonated calixarene.

2.2 Experimental

2.2.1 Materials

The SWCNTs (Elicarb®, Thomas Swan & Co. Ltd., County Durham, UK) utilised were produced by a C-CVD synthesis method with an as-received purity >90% (4.79 wt% ash, 2.76 wt% iron catalyst, <2 nm tube diameter). Graphene powder (470 nm average 62 Chapter 2 size, 90.1 wt% carbon, 9.1 wt% oxygen) was also obtained from a collaborator (University of New South Wales, Australia). Tetraphenylphosphonium bromide and ytterbium trichloride hexahydrate were obtained from a chemicals supplier (Sigma-Aldrich, MO, USA) and used without further purification. The calixarenes utilised in this study are displayed in Figure 2.3 and Table 2.1. They included p-phosphonic acid calix[n]arenes (n = 4, 6, 8), (1A, 2A, and 3A), p-phosphonic acid calix[4]arene modified by octadecyl and butyl substituents on the lower rim, (1B and 1C), p-phosphonic acid calix[6]arene modified by methyl substituents on the lower rim, (2B), p-sulfonato-calix[8]arene, (4A), p-benzyl-sulfonato-calix[8]arene, (4B), and p-phenyl-sulfonato-calix[8]arene, (4C).

Compounds 1A - 3A were prepared via recently published synthetic pathways.[393, 394] The p-sulfonato-calixarene (4A) was synthesised through a modified preparation,[388] with the ‘extended arm’ p-sulfonato-calixarenes (4B and 4C) synthesised via published procedures.[315, 317]

2.2.2 Methods

A typical experiment consisted of the addition of as-received SWCNTs (1.0 mg) into a sample vial (10 mL) containing MilliQ water (6 mL, >18 MΩ cm-1). To this mixture the appropriate calixarene (2.60 x 10-2 mmol) was added. This mixture was ultrasonicated for 10 min with a sonic lance (4 W RMS power, Sonifier 150, Branson, CT, USA) and then a further 30 min at 5 W RMS power, affording black to light grey solutions (Fig. 2.4a). To remove impurities and large agglomerations the solution was divided into 1.5 mL aliquots and centrifuged (5000 relative centrifugal force (RCF), Mini Spin plus, Eppendorf, Hamburg, Germany) for 30 min in ambient conditions. The top three quarters of the supernatant was collected and combined, affording the solubilised product (Fig. 2.4b).

X

n Y Figure 2.3 Principal calixarene structure.

63 Chapter 2

Table 2.1 Notation for calix[n]arene systems.

X Y n Notation

PO3H2 OH 4 1A

PO3H2 C18H37 4 1B

PO3H2 OBu 4 1C

PO3H2 OH 6 2A

PO3H2 OMe 6 2B

PO3H2 OH 8 3A

SO3H OH 8 4A

Benzyl-SO3H OH 8 4B

Phenyl-SO3H OH 8 4C

Host-guest self-assembled arrays: For the assessment of building higher-order SWCNT self-assembled arrays, tetraphenylphosphonium bromide (2 mg, 4.77 x 10-3 mmol) was added to an aliquot (5 mL) of the calixarene/SWCNT supernatant, with the pH adjusted to 6. Further addition of YbCl3·6H2O (5.3 mg, 1.37 x 10-2 mmol) was introduced to the above mixture.

Characterisation methodologies: Raman spectroscopy was performed using a R-3000CN portable Raman spectrometer (Raman Systems Inc., TX, USA). The spectral range covered was 100 cm-1 to 1800 cm-1, with an excitation wavelength of 785 nm, and laser power <10 mW. Spectra were acquired from multiple areas and rotations; the samples were prepared by drop-coating the resulting supernatants onto aluminum foil and air drying. Raman spectra of calixarene control samples were also prepared and collected using this methodology; a Raman analysis protocol, developed specifically for this instrument, is described in Appendix A.1. UV-visible

64 Chapter 2 spectrophotometry was performed with a Cary 50 Tablet spectrophotometer (Varian, CA, USA). The spectral range covered was 800 nm to 200 nm, with a scan rate of 600 nm min-1; the samples were prepared by dilution (100x) of the resulting supernatants, and were analysed in 10 mm path length quartz cuvettes. TEM was performed using a 2100 TEM (JEOL, Tokyo, Japan) operating at 120 keV; sample preparation involved centrifugation of the supernatant (100x dilution) and re-dispersing the resultant solid in MilliQ water, from which a drop of the solution was placed onto a holey carbon film supported by copper grids and air-dried. AFM was performed using a PicoScan system with a PicoScan 3000 PicoSPM II controller (Agilent Technologies, CA, USA) utilised in tapping mode. Topographic images were acquired using a grade 14 tapping mode cantilever. Samples were drop-cast onto a freshly cleaved mica substrate and excess solution removed using a gentle N2 flow. FET electrical characterisation was performed using a potentiostat (model EA161, eDAQ Pty. Ltd., NSW, Australia) and electronic recorder (e-corder, model EA201, eDAQ), utilised in a linear sweep configuration (EChem™, eDAQ). Samples were prepared by drop-casting the supernatant solution onto pre-fabricated interdigitated metal electrodes. The NT-FET current-voltage (I-V) curve data was subjected to a Savitsky-Golay smoothing function, using a second-order polynomial function over 25-data point intervals (Origin Pro 8.0, OriginLab Corporation, MA, USA).

2.3 Results and discussion

All the calixarenes studied are essentially acting as surfactant molecules, in solubilising the SWCNTs. It is interesting to note that during ultrasonication all the solutions containing the calixarenes formed a black solution (Fig. 2.4a), with exception of 1A and 1B which produced light grey suspensions. In comparison, the SWCNT/water experiment with no calixarenes formed large aggregates of the SWCNT material (Fig. 2.4a). The transformation from colourless to black and/or grey indicates the effective solubilisation of the SWCNT material into the aqueous media. This is further supported with the intense black colouration maintained in the supernatant after centrifugation (Fig. 2.4b). It should be noted that the collected supernatant suspensions of 1A - 4C are highly homogeneous and are stable for over six months, and thus the p-phosphonated calixarenes and ‘extended arm’ p-sulfonated calixarenes are effective in solublising and stabilising SWCNTs in aqueous media.

65 Chapter 2

Figure 2.4 Photographs of a) a solution containing SWCNT/water (left) and a solution containing 2A/SWCNT/water (right), and b) supernatant of 2A/SWCNT/water system.

In addition, an identical experimental set was repeated with graphene powder. Again, a black suspension occurred following ultrasonication, however, over a 24 h time period the original solutions, similar to the SWCNT blank (Fig. 2.4a), were afforded and solubilisation was not established. These results indicate that surface curvature may be a required parameter for the successful solubilisation of carbon nano-material by these particular calixarene systems and is an area which requires further attention.

2.3.1 Raman spectroscopy

The supernatant solutions from the previous experiment were dried and the resulting residues were analysed using Raman spectroscopy to obtain a greater understanding of the material solubilised by the calixarene systems. Raman scattering produces characteristic peaks from certain phonon modes in CNTs. The frequencies of immediate interest revolve around the radial breathing modes (RBMs) and the G- and D-bands.[421] The low frequency RBMs (100 cm-1 to 300 cm-1) are directly dependant on the nanotube diameters within a sample. The high-frequency band at ~1340 cm-1 is commonly referred to as the D-band (Disorder-induced band) and provides information regarding amorphous impurities and CNT wall disorder. The high-frequency band at ~1585 cm-1 is commonly referred to as the G-band (Graphite band) and relates to the graphite E2g symmetry of the interlayer mode. This mode represents the structural intensity of the sp2-hybridised carbon atoms of the CNTs and is sensitive to the nanotube surroundings. It is common to utilise the ratio of the intensity of the D-band to that of the G-band (ID/IG), to assess the quality of a CNT sample following a purification procedure.[422] It should be noted that Raman spectra were obtained for all

66 Chapter 2 calixarene control samples, and were found to be not Raman active at the laser frequency utilised in this study (Fig. 2.5).

Figure 2.5 Raman spectrum of the ‘blank’ aluminium foil, inset of the typical RBM frequency range. These ‘baseline’ frequencies were taken into account for the analysis of the as-received SWCNTs and supernatant residues. This also represents the spectrum obtained when analysing calixarene control samples.

Figure 2.6 Raman spectrum of as-received SWCNT, inset of the RBM frequency range.

67 Chapter 2

The spectral shifts in the calixarene/SWCNT systems are described in relation to the as-received SWCNT sample in Table 2.2. All of the calixarene systems produce an up-field shift in the G-band frequency relative to the as-received SWCNT sample (Fig. 2.6), as shown in Figures 2.7a and 2.8a. For the p-phosphonated calixarene analogues, the largest shift was seen for 2B, with the smallest shift seen for 1B. For the latter, this may be related to the large hydrophobic tail, decreasing its solubility in aqueous media. The hierarchy of frequency shift is 1B < 1C < 1A, 2A, 3A < 2B (Table 2.2), and this gives some insight into the relative strength of the interaction between the respective p-phosphonated calixarene and the SWCNTs. The results show that the interplay between the calixarenes and SWCNTs leads to their solubilisation into the aqueous media.[423]

Table 2.2 Raman spectral G-band frequency shifts and the ID/IG ratios for the residues of the calixarene/SWCNT supernatants.

Sample G-band frequency Frequency shift ID/IG ratio (cm-1) (cm-1)

As-received SWCNTs 1583.68 - 0.2131

1A 1592.48 +8.80 0.1613

1B 1588.08 +4.40 0.1816

1C 1588.96 +5.28 0.0729

2A 1592.48 +8.80 0.1559

2B 1594.24 +10.56 0.1350

3A 1592.48 +8.80 0.2449

4A 1592.48 +8.80 0.1277

4B 1602.15 +18.47 0.2514

4C 1602.15 +18.47 0.1440

68 Chapter 2

Similarly, with the p-sulfonated calixarene systems significant G-band shifts were observed, with the largest up-field shifts attributed to the ‘extended arm’ p-sulfonated calix[8]arene systems (4B and 4C) (Fig. 2.8a). This could be due to more efficient packing of the calixarenes around the SWCNT surface. The conformational flexibility of the larger ring size calixarene, with a larger hydrophobic surface area, is likely to encourage π-π stacking of the aromatic units of the benzyl and phenyl upper rim functionalities with the SWCNT surface.

The results indicate that the peak width of the D-band at ~1340 cm-1 has decreased significantly from 100 cm-1 in the as-received SWCNTs to 20 cm-1 to 40 cm-1 full-width half maximum (FWHM) in all trialled systems (Figs. 2.7a and 2.8a). This peak width narrowing is attributed to the reduction of the amount of amorphous carbon in the sample, and indicates that all of the systems studied show a low binding affinity to amorphous carbon, and when coupled with centrifugal processing results in minimal amorphous carbon present in the aqueous supernatant. These results support the use of these water-soluble calixarenes in post-growth purification protocols.

It is important to measure the purity of the SWCNTs that have been solubilised by the calixarene host molecules. A method routinely utilised is based on ID/IG ratio, where a lower value of ID/IG implies less disordered carbon attributed to the SWCNTs structure and hence an increased level of pristine SWCNTs. The as-received SWCNTs have a calculated ID/IG ratio of 0.2131. The ID/IG ratios are displayed in Table 2.2, with the hierarchy of ID/IG ratios as follows: 4B > 3A > SWCNT as-received > 1B > 1A > 2A > 4C > 2B > 4A > 1C. This shows that the majority of the calixarene systems increase the level of pristine SWCNTs in the supernatant, with the exception of 3A and 4B. The

ID/IG ratios for the 3A and 4B systems, albeit lower in purity, are relatively close to the as-received SWCNT ratio. Interestingly, the 3A and 4B systems reveal large G-band shifts, indicating significant surface interaction with the SWCNTs. These systems may also bind to carbon particle impurities which can result in their disaggregation and an increased D-band signal.[423] These data further support the use of the majority of these calixarene systems in post-growth purification protocols.

69 Chapter 2

Figure 2.7 a) Raman spectra obtained for the phosphonated calixarene systems, and b) Raman spectra highlighting the RBM bands.

70 Chapter 2

Figure 2.8 a) Raman spectra obtained for the sulfonated calixarene systems, and b) Raman spectra highlighting the RBM bands.

71 Chapter 2

The RBM frequencies assigned within a Raman spectrum can provide direct information regarding the distribution of SWCNT diameters within a sample.[155] The relationship between the RBM frequency and the inverse nanotube diameter is well described in the literature with the equation:

A  RBM  B (4) d t

-1 where  RBM is the RBM frequency in cm , dt is the tube diameter in nm, and the parameters of A = 223.5 cm-1 nm and B = 12.5 cm-1 have been experimentally determined.[156] The Raman spectrum, including the RBM region, for the as-received SWCNT is detailed in Figure 2.6. It should be noted that several recurring frequencies in the RBM range were identified and removed during data analysis. These could be contributed from the aluminium foil, possible laser reflections, and/or the sampling environment (Fig. 2.5).

The dominant SWCNT diameters in the sample are 1.61 nm (151.25 cm-1), 1.19 nm (200.49 cm-1), 1.05 nm (225.99 cm-1) and 0.89 nm (262.04 cm-1), which are displayed in Figure 2.6. If a broad RBM band is observed it is possible that it is a superposition of a few Lorentzian components and could represent multiple diameter SWCNTs.[424] This is possibly what is occurring with the 151.25 cm-1 band, which is subsequently masking the larger diameter SWCNTs individual RBM bands. The spectra obtained from the p-phosphonated calixarene systems are illustrated in Figures 2.7a and 2.7b, and for the p-sulfonated calixarene systems spectra are detailed in Figures 2.8a and 2.8b.

A summary of the calculated SWCNT diameters present in the supernatants in each system is given in Table 2.3. Firstly, the data supports the previous findings on solubilisation of the SWCNTs. The data also indicate that there is selective uptake of particular diameter SWCNTs, which is highly dependent on the nature of the calixarene. In the majority of the supernatants the most intense band can be attributed to SWCNTs with a diameter of 0.89 nm (ca. 262 cm-1). Interestingly, this SWCNT diameter is not

72 Chapter 2 the most abundant in the as-received SWCNT sample (Fig. 2.6), yet this diameter seems to be efficiently solubilised, with the uptake in solution of larger diameter tubes limited.

Table 2.3 Calculated SWCNT diameters present based on RBM frequencies in the Raman spectra.

Sample Calculated SWCNT diameter distribution (nm) [a]

As-received SWCNTs 1.61, 1.19, 1.05, 0.89

1A 1.01, 0.89

1B 1.01, 0.89

1C 1.04, 0.89

2A 0.89

2B 1.00, 0.89

3A 1.00, 0.89

4A 1.01, 0.89

4B 0.91, 0.89

4C 1.35, 1.04, 0.89

[a] SWCNT diameter relating to the most intense band is italicised.

As previously mentioned the smaller diameter SWCNTs were more readily solubilised into aqueous media. However, the data indicate that there are larger SWCNT diameters solubilised when using 4C (1.35 nm). In the case of 4C involving ‘extended arm’ sulfonated calix[8]arenes, the large G-band shift indicates strong nanotube surface interactions. This suggests a different type of interplay with the SWCNT surface, in solublising the larger diameter SWCNTs. This is consistent with the increased conformational flexibility of the larger ring size calixarene.

73 Chapter 2

The data in Table 2.3 also indicate that 1C and 2A are unique in regard to SWCNT solubilisation. 1C shows the ability to preferentially bind to 1.04 nm (ca. 228 cm-1) diameter SWCNTs, with the band having the greater RBM intensity. This diameter represents the smallest SWCNT diameter population in the as-received SWCNTs (Fig. 2.6). In contrast, 2A only solubilises SWCNTs with a diameter of 0.89 nm (ca. 264 cm- 1). These two systems not only significantly increase the purity of the SWCNTs (Table 2.2), they can also be utilised to solublise, preferentially, SWCNTs of a specific diameter.

With the SWCNT diameter distribution in both the as-received SWCNT sample and the different calixarene system supernatants, it is now possible to calculate the integers of assignment (Table 2.4). This can be achieved with the application of Equation 1

(Section 1.3), where a = 0.249 nm (due to 3aa c-c  0.249 nm [ac-c = 0.142 nm]) and the integers (n,m) define the structure of a SWCNT in terms of its diameter and chiral angle.[425] Furthermore, following the topographical assignment, the electronic nature of each specified nanotube topology can be determined (Table 2.4). There are three possible electronic assignments: (i) quasi-metallic SWCNT with |n - m| = 3q; (ii) metallic SWCNT with n = m, and (iii) semiconducting SWCNT with |n - m| = 3q ± 1, where q is an integer.[425]

From the chiral assignment and, subsequently, the electronic nature of the SWCNTs in the supernatants (Table 2.4), it is possible to comment on the preferential solubilisation of semiconducting and metallic SWCNTs. The majority of the SWCNTs analysed are semiconducting in nature, which is expected from using a 785 nm excitation wavelength, where excitation is primarily resonant with the v2 → c2 transitions.[426] In relation to the p-phosphonated calixarenes, all systems preferentially solubilised 0.89 nm, (10,2) semiconducting SWCNTs based on RBM intensities. For the 1C system, while still solubilising 0.89 nm SWCNTs, solubilisation of the 1.04 nm, (12,2) semiconducting or (9,6) metallic SWCNTs occurs. This is the only p-phosphonated calixarene to bind to metallic SWCNTs. Most p-phosphonated systems solubilise two to three different diameter SWCNTs, whereas 2A only solubilised the 0.89 nm (10,2) semiconducting SWCNTs. This selective binding to specific chiral SWCNTs has significant implications for post-growth purification and application selective

74 Chapter 2 processing for many current and future applications, in particular carbon nano-electronics.[136]

Table 2.4 Observed RBM bands in the as-received SWCNTs and calixarene system supernatants with topographical and electronic assignment.

Observed RBM band Calculated SWCNT Assignment Semiconducting (S) or -1 (cm ) ± 1 Diameter (nm) (n,m) Metallic (M)

264 0.89 (10,2) S

257 0.91 (11,1) S

(9,4) S

235 1.00 (12,1) S

232 1.01 (11,3) S

228 1.04 (12,2) S

(9,6) M

226 1.05 (10,5) S

200 1.19 (15,0) S

178 1.35 (15,2) S

(14,5) M

151 1.61 (18,4) S

In relation to the electronic nature of the solubilised SWCNTs the results obtained from the p-sulfonated calixarene systems were similar to the p-phosphonated calixarene systems. However, it is interesting to note that 4C preferentially binds two different diameter metallic SWCNTs, along with the typical 0.89 nm SWCNTs. The intensities

75 Chapter 2 of these respective RBM bands are relatively similar which suggests that there is an enrichment of metallic SWCNTs in the supernatant for this system.

2.3.2 UV-visible spectrophotometry

In a recent study, Attal et al. utilised UV-visible absorbance spectrophotometry to determine the concentration of SWCNTs in aqueous dispersions. The absorption peak at 273 nm is a signature of the surface π-plasmon excitation of dispersed SWCNTs.[427] The supernatants obtained for the different calixarene systems, in this study, were analysed using this technique (Fig. 2.10). In the majority of the systems slight increases in absorbance at the 273 nm peak were observed, when compared to solutions only containing the calixarene (Fig. 2.9). However, these results are deemed inconclusive as the π-plasmon surface excitations of the calixarene adsorb strongly in the 273 nm region and it is possible the contribution from the SWCNTs has been masked.

Figure 2.9 UV-visible spectra of solutions of the p-phosphonated calixarenes (1A - 3A) and p-sulfonated calixarenes (4A - 4C).

76 Chapter 2

Figure 2.10 UV-visible spectra of the p-phosphonated calixarenes (1A - 3A)/SWCNT supernatants and p-sulfonated (4A - 4C)/SWCNT supernatants.

2.3.3 Transmission electron microscopy

TEM was utilised to assess the residues of the supernatants to give information on the solubilisation characteristics and also reveal possible packing arrangement in the solid-state that may relate to arrangement in the solubilised phase. TEM was carried out on the supernatants of the calixarene systems, with 1C results detailed in Figures 2.11 and 2.12.

Figure 2.11 TEM image illustrating SWCNT networks coated with 1C. 77 Chapter 2

Figure 2.12 TEM image illustrating the individual SWCNT nature of the supernatant with the wrapping of 1C around the SWCNT surface.

The TEM image in Figure 2.11 illustrates the typical structures observed on the grid, where networks of SWCNTs can be seen scattered with amorphous coatings of the calixarene sheathing the SWCNTs. As expected, the SWCNTs are observed to be individual (Fig. 2.12), indicating the disruption of van der Waals forces between SWCNT bundles, and subsequent self-assembly of the calixarene around a SWCNT surface for the effective solubilisation into aqueous media. These results provide support for the increase in pristine SWCNT purity (Table 2.2). Energy dispersive X-ray analysis (EDX) has also been undertaken on the sample, with no peaks attributed to any remaining iron catalyst particles (Fig. 2.13). The phosphorous (P) peak observed in the EDX spectrum supports the inference that the calixarene is sheathing the SWCNTs, and confirms the presence of the calixarene in the system. It should be noted that the Si peak observed in the EDX spectrum is attributed to column purification on the calixarene starting material, and the Cu peaks are due to the copper TEM grid. The combination of SWCNT individualisation, an increased level of pristine nanotubes and removal of unwanted catalyst contaminants supports the use of these systems in SWCNT post-growth purification protocols.

78 Chapter 2

Figure 2.13 EDX spectrum of 2A/SWCNT supernatant.

2.3.4 Atomic force microscopy

AFM was utilised to study the supernatant residues in an attempt to obtain greater understanding of the calixarene surface-coating, which ultimately leads to the solubilisation properties of the system. A typical three-dimensional topographical representation of an individual SWCNT from a calixarene/SWCNT supernatant residue is illustrated in Figure 2.14. It can clearly be seen that a coating of some description is on the SWCNT and this information correlates well with TEM data (Fig. 2.12). A height profile was obtained perpendicular to the directionality of the tube (Fig. 2.15). As AFM has excellent z-height resolution capabilities, this profile can be utilised to provide calixarene coating thickness parameters. This profile indicates that from the underlying mica substrate to the maximum height of the sheathed SWCNT is 5.0 nm. From analysis of the RBM in the Raman spectrum the tube diameter is known to be either, 1.04 nm or 0.89 nm, giving rise to a height difference of 3.96 nm or 4.11 nm respectively. Assuming the calixarene sheathing is evenly distributed around the SWCNT, the calixarene coating of either 1.98 nm or 2.05 nm is obtained.

The data obtained from these analyses assist in developing a possible calixarene stacking arrangement around the SWCNT surface. The approximate 2 nm layer thickness could be attributed to a bilayer stacking arrangement or possibly a calixarene-calixarene interdigitation leading to the formation of capsules along the nanotube surface. These stacking arrangements are purely speculative and require future studies of these systems through computational chemistry methodologies. In order to render the SWCNTs soluble in aqueous media, a large proportion of the

79 Chapter 2 available phosphonate and sulfonate functional groups must be accessible, and this should be taken into account in terms of conformational steric effects.

Figure 2.14 a) A three-dimensional topographical representation of a calixarene sheathed SWCNT.

Figure 2.15 A height profile obtained from analysis of the calixarene sheathed SWCNT in Figure 2.14. Note the y-axis is in nm, and the x-axis is in µm.

80 Chapter 2

2.3.5 Host-guest directed SWCNT aligned arrays

Recently, Raston and coworkers reported the self-assembly capabilities between p-sulfonatocalix[4]arene and a bisphosphonium cation, 1,4-bis(phenyl- phosphomethyl)benzene.[428] Structural elucidation through single-crystal XRD indicated that a bilayer stacking arrangement was produced, and was comprised of a staggered ‘up-down’ calixarene arrangement with exo-cavity cation binding. An interesting aspect of this structure was that individual bilayers were linked through endo-cavity binding of an aromatic functional group, which is described as scaffolding. This is an interesting concept and an attempt to translate this principle to the SWCNT/calixarene system was made for investigations into the self-assembly of SWCNT arrays.

A suitable scaffolding ‘linker’ molecule, tetraphenylphosphonium bromide, was added to the preformed SWCNT/calixarene supernatants, the resulting mixtures were analysed using TEM. The results from TEM analysis indicated that there was no evidence of the assembly of the SWCNTs into higher-order arrays, as the spatial distribution of the SWCNTs was similar to Figure 2.11. In an attempt to further mimic the experimental conditions expressed by Raston et al.,[428] a lanthanide metal chloride salt was added in three-fold excess to the system. This was shown to be necessary for the formation of self-assembled structures in the aforementioned study. The resulting solutions were analysed using TEM, and this indicated that there was no evidence of any assembly processes forcing SWCNT array formation.

2.3.6 Electrical characterisation through the development of TFTs

Through a detailed Raman spectroscopy study on the resulting SWCNT/calixarene supernatants, it has been shown that selective enrichment of semiconducting and/or metallic SWCNTs is tunable, depending on the calixarene system utilised. The supernatants were utilised to fabricate a SWCNT network FET, in an attempt to characterise the electronic properties of the resulting supernatants and to support the finding of the Raman study.

81 Chapter 2

Figure 2.16 I-V curve illustrating the TFT properties of the 2A/SWCNT supernatant. Inset is an I-V curve of a commercial 2N5485 N-channel FET, with y-axis represented in µm. Measurements taken at VGS = 0 V.

I-V curves for the thin film NT-FET fabricated from the 2A/SWCNT supernatant, and a commercially available FET are illustrated in Figure 2.16. The I-V curve generated for the commercial FET indicates a threshold voltage of +0.5 V. This indicates that current (I) will not flow through the device until this minimum voltage is applied, and forms the basis of FETs. In comparison, the device fabricated from the 2A/SWCNT supernatant shows a similar I-V curve trend. It should be noted that the 2A/SWCNT device signal is quite noisy, when compared to the commercial FET, this is a result of the much lower current-range (I-range) of the system, nA compared to µA, and inherently more noise will be present in the measureable signal. The fabricated NT-FET expresses a threshold voltage lower that that of the commercial FET, at +0.1 V. This is to be expected, as SWCNT bandgap in the DOS is inversely proportional to the nanotube diameter. Hence, due to the SWCNT diameter of 0.89 nm, observed in the 2A/SWCNT supernatant (Table 2.3), a low threshold voltage would be expected. This electronic behaviour can be observed for a wide range of the supernatant systems produced through the aqueous solubilisation methodology and provides further support for nanotube separation based on their electronic properties. Furthermore, this supports the 82 Chapter 2 utilisation of this procedure in post-growth purification protocols aimed at the advancement of carbon nano-electronics applications.

2.4 Conclusion

The use of p-phosphonated calixarene systems to solubilise SWCNTs into aqueous media has been developed. The p-phosphonated calixarene systems have been directly compared to a selection of p-sulfonated calix[8]arene systems, with a strong binding affinity observed between the ‘extended arm’ p-sulfonated calix[8]arenes and the SWCNT surface, through Raman spectroscopy. Analysis of the resulting supernatants has confirmed the presence of calixarene sheathed, individualised SWCNTs in solution and indicates the efficient removal of any metal catalyst particles imparted from the growth process. Interestingly, the calixarene systems did not solubilise graphene sheets, which indicates that extended surface curvature could be a requirement for calixarene binding mechanisms leading to solubilisation. The thickness of the calixarene sheath surrounding the SWCNT has been assessed and found to be ca. 2 nm, information which can be utilised for the development of calixarene stacking mechanics. Furthermore, attempts to promote self-assembled SWCNT arrays utilising host-guest principles were explored, however, these were unsuccessful.

These results show that the potential for the use of both p-phosphonated calixarenes and p-sulfonated calix[8]arenes for not only solubilisation of SWCNTs, but as a purification route with selective diameter uptake, tunable with the specific system utilised. The distribution of SWCNT topologies has been assessed and utilised to determine the electronic nature of the selectively solubilised SWCNTs. In support of the Raman investigation, NT-FET were successfully fabricated and the threshold voltage compared to a commercially available FET device. These NT-FETs allow experimental conformation that the selective enrichment of semiconducting SWCNTs in the resulting supernatants can be utilised in device fabrication. This selective diameter and chirality solubilisation protocol, provides a step forward in the post-growth treatment of SWCNTs to obtain application specific SWCNTs in areas such as carbon nano-electronics and device technology.

83

Chapter 3 Design, Fabrication and Verification of a Gas and Vapour Analyte Delivery System for Sensor Testing

Chapter 3

3 Design, fabrication and verification of a gas and vapour analyte delivery system for sensor testing

3.1 Introduction

A major concept contained within this thesis is the design, development and testing of CNT-based gas and vapour sensors. Not only is there a requirement to design and fabricate the actual sensor devices, but also a methodology to allow for the accurate and precise delivery of a vapour-phase analyte. It is essential that there is instrumentation available to obtain accurate sensor response data. This current chapter describes the fabrication and commissioning of an in-house built gas and vapour analyte delivery system coupled with sensor housing and electrical parameter monitoring components. In addition, three different prototype systems will be detailed from theoretical design characteristics through to any functional issues encountered. This current chapter should be viewed as the behind-the-scenes technical workings which allowed the goal of developing CNT-based chemiresistor sensors to be realised.

From the initial stages of this research it was evident that a versatile, multi-purpose sensor testing facility would need to be established, in order to avoid restrictions on the type of analytes which could be trialled against the developed sensors. As previously mentioned, the research aims, in terms of chemical sensing, were to target compounds which are classified as being of national security interest (Section 1.10). These types of compounds, including DMMP and DIMP, have relatively low vapour pressures and are typically not available in pressurised cylinders, so other sample introduction methods are required. This current section and sub-sections describe commonly encountered gas and vapour delivery systems, which are not always achievable with the same apparatus, and provide an overview of the desired system functionalities for testing of the CNT-chemiresistor sensors.

For clarity, the terminology utilised in this thesis in relation to the analyte phase will be described. The term ‘gas’ refers to ‘typical’ gaseous compounds, such as CO, and are chemicals that are typically commercially available in pressurised gas cylinders. The term ‘vapour’, should be taken to mean an analyte compound which is in the

85 Chapter 3 gaseous-phase, yet has been generated from either a liquid or solid source, such as DMMP.

3.1.1 Gas delivery and dilution methodologies

In relation to the accurate delivery of gases contained in pressurised cylinders, there are many different commercially available systems that can handle this task, one such is Environics® (CT, USA). The basic principles of these systems use several gas mass flow controllers (MFCs) to control the flow and dilution ratio of a gas analyte. Gas feed lines are typically fed from the pressurised cylinders to the entry ports of the gas mixer and dilution station, and based on user-defined input parameters, a corresponding flow rate of a specific concentration analyte is achieved and can be directed to a sensor testing chamber. This type of system is well suited to ‘typical’ gas analytes and has been utilised extensively for gas testing trials on CNT-based sensors.[223, 228, 238, 275, 429] However, if ‘vapours’ from organic liquids, or even solids, are required then this system is highly restrictive.

A versatile system to allow access to vapour analyte samples, in addition to ‘typical’ gases was required. A major focus of this particular project was to assess the fabricated CNT-based chemiresistors for their sensing capacity towards CWAs, such as G-series nerve agents and war gases. Section 3.1.2 will assess methodologies for vapour generation aimed at low vapour pressure CWAs, such as DMMP.

3.1.2 Vapour generation methodologies targeting CWAs

Through the efforts of both industry and academia there are a variety of vapour generation methodologies expressed in the literature, and the majority of these require the direct contact of a carrier gas stream with the liquid analyte of interest. However, there are also methodologies which allow vapour generation without such intimate contact, through indirect methods. These indirect methods will be described, followed by a selection of direct contact methods.

86 Chapter 3

3.1.2.1 Indirect contact vapour generation methods

One well known indirect vapour generation method is the use of permeation tubes. This encompasses the slow penetration or permeation of a liquid analyte through a polymeric membrane. The rate at which the analyte is released or permeates from the membrane is controlled by factors such as analyte vapour pressure, analyte permeability through the membrane, temperature and the partial pressure difference across the membrane.[430] These tubes can be utilised to give controlled and predictable analyte concentrations over a wide flow range and can be obtained as calibrated products. A carrier gas is passed over an analyte of interest permeation tube, at known flow rate and temperature, and a predictable concentration of analyte vapour is produced. This vapour generation methodology is typically utilised for the calibration of analytical instrumentation, as trace levels of the analyte can be achieved.[431] Snow and coworkers, for example, used this vapour generation methodology for 1 ppb DMMP delivery to SWCNT chemiresistor sensors for analytical testing.[277] This methodology was not employed in this study due to the limited analyte concentration range these tubes offer, and the number of different analyte permeation tubes that would be needed to undertake a comprehensive sensor evaluation.

Two additional indirect vapour generation methods include diffusion and effusion. Diffusion works by having an open vessel containing the liquid analyte and have the carrier gas flowing near to this open vessel. The vapour pressure and diffusion characteristics of the analyte will dictate the concentrations achieved.[431] Effusion on the other hand, is where a constant stream of carrier gas is directed across a small orifice, which beneath lies a reservoir of liquid analyte. The sweeping carrier gas creates a concentration differential at the orifice and, based on orifice size and sample temperature, the concentration of analyte in the carrier gas can be determined.[431] Both of these methodologies were deemed unsuitable for the current sensor testing requirements and were not pursued.

3.1.2.2 Direct contact vapour generation methods

The most direct method of vapour generation is to bubble all, or part, of a carrier gas stream into a reservoir containing a liquid analyte, referred to as a bubbler-based chemical vapour generator (Fig. 3.1).[432] The concentration of the analyte vapour

87 Chapter 3 contained in the carrier-gas stream is dependant on the vapour pressure of the liquid, the fraction of the carrier-gas stream diverted to the chemical bubbler, and the temperature of the liquid analyte. This vapour generation method has been utilised extensively to generate carrier gas streams containing CWA simulants, used to assess the sensing capacity of carbon-based vapour sensors.[229, 433, 434] This methodology has even been extended to generate vapours from solid-phase samples by passing the carrier gas stream directly over the headspace of the solid analyte.[435] This vapour generation methodology was utilised, in-part, in the development of a suitable system to allow sensor testing.

Figure 3.1 A schematic of a bubbler-based chemical vapour generator utilised upstream from a CNT-based SAW sensor testing chamber. Adapted with permission from [232], © 2005 Elsevier.

A more recent development in vapour generation has been the advent of syringe-pump driven spray atomisation processes reported by Muse and coworkers.[436] This methodology utilises a syringe, containing the liquid analyte, mounted in a syringe-pump. A flow tube is connected to a spray atomiser (nebuliser) which, when coupled with a flow of carrier gas, will effectively create a fine mist of the liquid analyte in a controlled fashion. This fine mist of analyte quickly evaporates to analyte vapour, under certain favourable conditions, and this vapour can be directly utilised or diluted to pass over the sensors (Fig. 3.2). This vapour generation methodology was applied to a G-series nerve agent sarin and the concentrations generated were statistically no different to those produced by liquid bubblers and automated solid sorbent sampling systems.[436] This methodology can be slightly adapted and instead of

88 Chapter 3 injecting liquid samples the headspace of a liquid analyte can be directly injected into a carrier gas stream.[431] These syringe-pump based principles were explored and applied to the fabrication of the vapour generation system utilised for the studies described in this thesis.

Figure 3.2 A schematic of a spray atomisation system which is used for the vapour generation of CWA GB (sarin). Adapted with permission from [436], © US Military.

Other vapour generation methods include the use of delta tube saturators, large-scale saturators, high-pressure injection, humidification and solid-state sorption-based methodologies. These ‘direct’ vapour generation methods were not utilised in the present study and for system comparison the reader is direct to the literature.[431]

The requirements of the multi-purpose sensor testing facility are as follows, (i) the ability to generate, dilute and precision-deliver vapours from multi-phase analytes, (ii) the capability of generating a large dynamic concentration range, parts-per-trillion (ppt) to ppm, dependant on specific analyte, (iii) allow for the introduction of chemical interferants and multi-component systems, (iv) to allow a range of carrier gas environments to be established, and (v) to allow automated sensor testing. To achieve the above system requirements the combination of MFC-based gas delivery and dilution, coupled with bubbler-based chemical vapour generation, and modified syringe pump injection methodologies were utilised in the development of a suitable facility for sensor testing. Applications of these processes are described in the subsequent sections.

89 Chapter 3

Furthermore, the design and fabrication of components such as the sensor testing chamber and electrical connectivity to monitoring equipment will be established.

3.2 Experimental

3.2.1 Materials

Prototype development: For clarity all materials and associated manufacturer details are listed in the appropriate sections within this chapter.

Pd nanoparticle/SWCNT hybrid sensor: The required materials for the IDE platform fabrication and SWCNT dispersion are described in Section 4.2.1. Palladium metal [437] nanoparticles were fabricated using a H2 gas reduction procedure, and were dispersed in water.

3.2.2 Methods

Prototype development: For clarity the details of all prototype development methods are described in the appropriate sections within this chapter.

Pd nanoparticle/SWCNT hybrid sensor testing: The fabricated sensor was utilised to provide verification that the developed prototype could perform and, in addition, to assess conditions within the sensor testing chamber. An aliquot (0.02 µL) of the dispersed Pd nanoparticles were drop-cast into a pre-prepared SWCNT coated IDE platform (Section 4.2.2). Utilising the chemical delivery system the device was exposed to H2 gas (5% - 9%), incremented by 1% steps, with ultra-high purity-N2 (UHP-N2) as the diluent. Before and after H2 exposure the system was purged with UHP-N2.

90 Chapter 3

3.3 Analyte delivery system - ‘Phase-I’ prototype

3.3.1 Chemical vapour generator and delivery system

The system to be developed required the following capabilities: (i) generating a vapour- phase sample of an analyte of interest, (ii) accurately diluting a vapour-phase sample, and (iii) providing a stable continuous flow of the analyte at a set concentration to the sensor testing chamber. As previously mentioned there are different avenues for the generation of vapour-phase samples and in this particular situation, being strictly limited to gases was unacceptable, and incorporating the capability to produce vapour-phase samples of liquid and solid analytes of interest was desirable.

To develop an analyte delivery system to deliver solely ‘typical’ gases, such as H2, CO,

CO2, to a sensor testing chamber is a relatively uncomplicated task. Both common and specialised gases can easily be commercially obtained, and, in addition, different concentrations of a specific analyte gas can be tailored relative to a ‘balance’ gas. Following this, computerised gas dilution systems can be utilised and, pre-programmed, to accurately dilute and/or mix multiple gases for the precise delivery to a sensor testing chamber. However, the ease of analyte preparation and introduction dramatically changes if a vapour-phase sample from a liquid or solid analyte is required.

The basis of the ‘Phase-I’ design was to construct an automated, command-prompt executable, chemical vapour generator and delivery system. This is superior to the ‘typical’ gas mixing and dilution system mentioned, and incorporates an additional chemical generator component to expand potential capabilities beyond the merely commercially available compressed gas samples.

A common methodology for the generation of a vapour-phase sample from a liquid analyte is to integrate a chemical bubbler into the delivery system. Fundamentally, a

‘dry’ carrier gas, such as N2, is bubbled through the liquid ‘source’ chemical and in the process becomes saturated with the vapour of the chemical source. This would be a valuable up-stream component for a chemical vapour generator, and would be linked to a mixing and delivery system, allowing the testing of vapour aliquots produced from liquid analyte samples in combination with typical gas analytes. At an early stage of

91 Chapter 3 development, a methodology to further extend the system capability to obtain vapour-phase samples from solid analytes was abandoned.

Pressure transducer Adjustable flow meter Laptop

Bubbler

Water bath

Gas dilution system Testing chamber

UHP

N2 Semiconductor parameter analyser Figure 3.3 Schematic of the ‘Phase-I’ prototype chemical vapour generator and delivery system, including sensor testing chamber and proposed measurement instrumentation.

A schematic of the initial design of the chemical vapour generator and delivery system, introducing the concepts of the sensor testing chamber and measurement instrumentation, is illustrated in Figure 3.3. Details of the components of the system will be discussed in this current section and Section 3.3.2. The working principle of the system is that by monitoring the parameters, including carrier gas flow rate, liquid temperature, and chemical bubbler headspace pressure, the concentration of the delivered carrier gas-vapour mixture can be calculated using the following equation,[432]

 P   v  v  QQ CG   (5)   PP vHS 

where Qv is the bubbler chemical vapour output flow rate in standard litres per minute

(slpm), QCG is the bubbler carrier gas flow rate (slpm), Pv is the thermodynamic vapour pressure of the source liquid in the chemical bubbler in pascals (Pa) and is a function of temperature only, and PHS is the pressure in the bubbler head space above the source

92 Chapter 3 liquid (Pa). This assumes that the vapour carrier-gas mixture exiting the chemical bubbler is fully saturated.[432]

To minimise any possible contamination, preventative measures were incorporated at the initial design stage. This included minimising the use of all plastics within the system by maintaining all wettable surfaces as either stainless steel or glass, and using high quality Swagelok® fittings and Teflon® seals throughout. It is known that the copper component of brass can react with nitro-based compounds to form explosive residues. As nitro-substituted aromatics, such as nitrotoluene, and 2,4-DNT, were of interest to this study, stainless steel components were preferred over brass. The use of only stainless steel and glass for wettable surfaces initially posed a challenge, as the fusing of glass to stainless steel tube fittings for mechanical connections is not easily achievable due to the differing thermal expansion coefficients of the materials.[438, 439] ® To rectify this problem specialised glass to ¼ inch Kovar tube fittings were obtained for any glass to metal connections within the system. Kovar® is a nickel-cobalt ferrous alloy that exhibits comparable thermal expansion coefficients to borosilicate glass over a wide temperature range.

The ‘Phase-I’ system consisted of a ¼ inch outer diameter (OD) stainless steel flow line connected to a pressure regulator, which is attached to a cylinder of the compressed carrier gas, in this case UHP-N2. Immediately downstream from the gas regulator, the flow line was split in to two using a ¼ inch stainless steel union tee, into one line continuing parallel and one line directed perpendicular to the original flow direction. The latter flow line, was directed to an adjustable flowmeter, for the accurate control of the carrier flow rate to a chemical bubbler. As previously mentioned, accurate monitoring of the carrier gas flow rate is required, and is fundamental for the determination of the delivered carrier gas-vapour concentration delivered to the sensor testing chamber. The chemical bubbler contains the liquid analyte of interest and is thermostatted by a temperature controlled water bath (Fig. 3.3).

93 Chapter 3

Figure 3.4 Photograph of the twin glass chemical bubblers with fritted glass caps and ¼ inch glass-to-Kovar® entry and exit ports.

Twin glass chemical bubblers were manufactured (Fig. 3.4) with entry and exit flow ports consisting of ¼ inch glass-to-Kovar® tube fittings. Fritted glass discs are located at the interface between the carrier gas and the liquid analyte contained within the bubblers. These discs consist of a porous body of sintered glass which has been fused to the tip of the gas inlet tube. This tip is submerged into the vessel containing the liquid analyte. The purpose of this fritted disc, or sparger, is to reduce the diameter of the carrier gas bubbles that will diffuse through the liquid, leading to an overall increase in surface area contact between the liquid analyte and the carrier gas, and promoting saturation of the carrier gas.[432] To allow maximum carrier gas saturation, the use of the twinned bubbler system was employed as this provided >99.9% carrier gas saturation and led to more accurate concentration calculations.[432]

Directly downstream of the chemical bubblers, a pressure transducer is attached to the system (Fig. 3.3), to monitor the chemical bubbler headspace pressure. This information, combined with carrier gas flow rates and the temperature of the liquid analyte, allows the calculation of the carrier gas-vapour concentration. Subsequently,

94 Chapter 3 the gas flow line is directed to an entry port of a gas mixer and dilution apparatus which is automated through time-based command prompts.

The secondary flow line, running parallel to the initial flow line propagating from the union tee (Fig. 3.3), is also directed into this gas mixer and dilution apparatus. This apparatus can be pre-programmed to control the gas flow rates from each individual flow line, and in turn will allow the saturated vapour, which has a known concentration, to be diluted with a balance gas (UHP-N2). These binary components are thoroughly mixed and delivered at a constant, user-controlled, flow rate to the sensor testing chamber. It should be noted that this system is capable of diluting and mixing analytes from compressed gas cylinders.

3.3.2 Sensor testing chamber

With the design of a chemical vapour generator and delivery system, a suitable chamber was designed to house the developed sensors for gas and vapour exposure trials. The basic requirements of the chamber were, (i) an entry port for the vapour-phase analyte to be introduced to the sensors, (ii) an exhaust port to safely vent the vapours, (iii) the ability to form a gas-tight seal, (iv) to be constructed from an inert material with a high chemical resistance, and (v) to allow electrical connections for direct sensor monitoring.

Figure 3.5 Photographs of the fabricated Teflon® sensor testing chamber from a) a side view, and b) a front view. The chamber is supported on a moulded Perspex bracket specifically fabricated to align the gas flow entry port with the upstream system.

95 Chapter 3

The fabricated sensor testing chamber is illustrated in Figure 3.5. The majority of the chamber was fabricated from Teflon®, as this material is chemically resistant to a wide variety of chemicals, and in the design stage of this particular project maximum chemical compatibility was necessary. The chamber consists of two sections, the cylinder where the sensors are mounted and a threaded cap, which was designed so that the chamber could be horizontally mounted, and the cap unscrewed to allow access to the internal cavity. In addition, the interlocking thread provides a gas-tight seal for the chamber.

The chamber lid has a ¼ inch threaded bore in which a Swagelok® ¼ inch male NPT to ¼ inch tube connector is secured, connecting the stainless steel flow tube from the gas mixer and dilution system to the testing chamber. At the rear of the chamber an identical arrangement is fabricated, referred to as the exhaust flow port. Directly below the exhaust port there is a ½ inch grommet seal to act as a cable feed-through port to facilitate the electrical connection of the sensors and monitoring instrumentation.

It is necessary to be able to directly monitor the sensors whilst they are positioned in the testing chamber. A suitable platform was designed to allow multiple sensors to be mounted and monitored within the testing chamber. This was achieved through the design and fabrication of a suitable printed circuit board (PCB), attached to the appropriate electrical connector (Fig. 3.6a).

Figure 3.6 A side-by-side view of a) the PCB design for the test chamber, and b) the fabricated PCB with components installed; the 25-pin DB-type connector (top) and spring sockets containing TO-8 Headers (bottom).

96 Chapter 3

Once the PCB was fabricated (in-house) the individual components were soldered into position. These consisted of a chassis mounted 25-pin DB-type connector (RS Components, NSW, Australia), with each pin directly connected to an individual spring socket (RS Components). TO-8 Headers (Spectrum Semiconductor Materials, Inc., CA, USA) are mounted in the spring sockets, with each individual header providing eight individual electrical connections. Each header, therefore, can hold four individual sensors (Section 4.2.2), each with one source and one drain electrode. As there are three headers mounted on the PCB, a maximum of 12 individual sensors can be located in the testing chamber at any given time. Each sensor electrode is wire-bonded to an individually insulated pin of the header.

Figure 3.7 A side-by-side view of a) the PCB design for the BNC connector break-out board, and b) the fabricated PCB with components installed; the 25-pin DB-type connector (top) and an array of BNC connectors (bottom).

In order to monitor the individual sensors in the test chamber an appropriate break-out board was designed and fabricated (Fig. 3.7). This is external to the testing chamber, being connected to the internal PCB through a 25-wire shielded cable with the appropriate DB-type connectors. This external break-out board consists of another chassis mounted 25-pin DB-type connector, to which the shielded cable is connected. The break-out board has 24 mini BNC connectors (RS Components), through a combination of the break-out board wiring, the shielded cable and the in-chamber PCB, each BNC connector is connected to an individual pin of a TO-8 Header (Fig. 3.8).

97 Chapter 3

This allows the source and drain electrode, of any individual sensor, to be monitored during an experiment by an external measurement instrument.

Figure 3.8 Photograph of an open sensor testing chamber.

The BNC break-out board was to be utilised as a connection between the sensors and a measurement instrument, a Precision Semiconductor Parameter Analyzer (4156A, Agilent Technologies). This instrumentation is configured to apply a voltage (V) between the source and the drain electrodes of a sensor to monitor the changes in I, upon exposure to vapour-phase analytes of interest. The mechanism by which the fabricated sensors operate is described in Section 1.4.1.2.

3.3.3 Prototype problems

Several problem areas emerged during the fabrication of the ‘Phase I’ prototype. One was that the gas mixer and dilution equipment, apart from being relatively expensive, did not have the ability to be chemically resistant, and operate accurately and precisely, when exposed to the suite of chemicals intended for sensor testing. Another problem encountered was in relation to the thermostatting of the system, in which the temperature of the liquid analyte requires accurate monitoring, for the carrier gas-vapour concentration determination. Furthermore, the pressure transducer used to monitor the bubbler headspace pressure is used, typically, in systems that have extensive knowledge of the pressure range expected. This is not a parameter that could be accurately predicted from the design stage.

98 Chapter 3

3.4 Analyte delivery system - ‘Phase-II’ prototype

3.4.1 Chemical vapour generator and delivery system

It was apparent from the problems encountered with the ‘Phase I’ chemical vapour generator and delivery system, that a re-design of the overall methodology was required. Certain components of the ‘Phase-I’ prototype could be retained, such as the sensor testing chamber and printed circuit boards, but the majority was rejected.

3.4.2 Syringe pump method

A syringe pump injection methodology has been used in previous studies for a variety of applications and involves the direct injection of a liquid analyte into the flow of a carrier gas stream using an analyte filled syringe coupled with a syringe pump injecting at a controlled rate.[431] Initially, an adaptation of a typical syringe pump methodology was undertaken, in which a peristaltic pump delivered a constant flow of liquid to the system. However, this was subsequently replaced by a syringe pump, as the low flow rates of the liquid analyte required were not readily achievable, coupled with system back-pressure problems.

This syringe pump injection method, also known as the ‘vapouriser’ method, is adopted as an ASAE (American Society of Agricultural Engineers) Standard, utilised when monitoring environmental air quality.[440] These general principles were mimicked in the design of the ‘Phase II’ prototype of the chemical vapour generator and dilution system.

Fabricated from Teflon® - 12 sensor Syringe pump unit testing capacity

Nebuliser Exhaust to fumehood Baffled mixing tube MFC / flow meter Vapouriser Test chamber Sampling port

UHP

N2 MFC / flow meter Semiconductor parameter Check valve to stop flow analyser back into heated vessel due to pressure differential Figure 3.9 Schematic of the ‘Phase II’ prototype chemical vapour generator and delivery system, including sensor testing chamber and proposed measurement instrumentation.

99 Chapter 3

The basic components and layout of the vapouriser-based design are illustrated in Figure 3.9. Different variables in the system can be monitored and directly controlled to deliver the desired vapour-phase analyte concentration to the sensor testing chamber, this is governed by the equation:

QC R  EA (6) DAB

-1 where R is the injection rate (mL min ), CEA is the concentration at the vapouriser, Q is the air flow rate (m3 h-1), D is the analyte density (g mL-1), A is a time conversion parameter (60 min h-1) and B is a mass conversion parameter (1000 mg g-1). For the most accurate concentration determinations, using this methodology, it is important to accurately control and measure the injection rate (R) and air flow rate (Q).

A description of the ‘Phase-II’ prototype chemical vapour generator and dilution system is now given. Again, the UHP-N2 carrier gas, is utilised as the preferred carrier gas and diluent gas for the experimental testing of vapour-phase analytes. However, it should be noted that in this system different types of gases can be utilised as the carrier gases, and more importantly different analyte gases such as CO, CO2, and H2 can be introduced for sensor testing purposes.

Downstream from the compressed gas cylinder pressure regulator, the 1/8 inch OD stainless steel tubing is split into two separate lines by a 1/8 inch union tee, each line coupled to a gas MFC, as shown, in the schematic (Fig. 3.9). The stainless steel 1/8 inch OD tubing is coupled to the entry port of a gas MFC (MC-1SLPM-D, Alicat Scientific, AZ, USA), the internal structure of which is illustrated in Figure 3.10. The range of this gas MFC is from 10 sccm to 1000 sccm and is calibrated and certified for the accurate gas flow control of 16 different gases. The gas MFC is operated through the program FlowVision™, in which time-based command scripts can be written to control the gas flow for an entire experiment. This piece of equipment is essential to the successful operation of the system, as the precise control of airflow (Q) is critical in determining the ‘vapouriser’ analyte concentration.

100 Chapter 3

Figure 3.10 A layer-by-layer schematic breakdown of the internal components and gas flow path of an Alicat gas MFC unit. Adapted with the permission of [441], © 2003 Alicat Scientific, Inc.

A length of 1/8 inch stainless steel tubing extends from the exit port of the gas MFC, which is connected directly to the gas flow port of a spray chamber nebuliser by 1/8 inch Tygon® tubing (Fig. 3.11). This linkage supplies a constant, controllable gas flow into the nebuliser. The nebuliser itself is made of quartz and consists of two concentric tubes, one larger outer tube and a much smaller inner tube both meeting at the tip. The gap between the outer and inner tube is filled with flowing gas which exits via the tip. A solution is delivered through the inner tube, and depending on the gas flow rate can be used to naturally aspirate liquids out of the tip, according to the well established Venturi effect. As the liquid sample is positioned at the interface of the two concentric tubes, at the tip of the nebuliser, the shearing forces across the face of the inner tube disrupt droplet formation and atomise the liquid efficiently. This atomised mist of liquid analyte is directed by the carrier gas to the vaporisation chamber of the system. Before describing the workings of the vaporisation chamber, the liquid analyte introduction methodology is discussed.

A method for accurately delivering the liquid analyte, referred to as the injection rate (R), is another parameter that is essential for the determination of the analyte concentration at the vapouriser. In the earlier stages of development a peristaltic pump (VS2x2-MIDI, Alitea, WA, USA) was installed, where the injection rates could be controlled by the speed of the pump coupled with the size of the internal diameter (ID)

101 Chapter 3 of the tubing utilised. This would allow calibration of the liquid injection rates to the nebuliser for an analytical run. Several problems occurred in system testing trials that led to the rejection of this type of pump. The major problem was the injection rates required for the liquid analytes, were in the order of μL h-1, to obtain low ppm concentrations. The flow rates required were significantly below the minimum obtainable flow rate using this peristaltic pump (0.0005 mL min-1 to 17.5 mL min-1). Furthermore, in order to carryout accurate detection limit determination the chemical vapour generator was required to produce sub-ppm concentration analyte vapours.

Figure 3.11 Photograph of the quartz nebuliser mounted with the tip into a securing gland, with the gas flow feed wired onto the gas entry port and tubing for the introduction of a liquid analyte secured at the rear.

To achieve the low injection rates required, other liquid delivery methods were explored. It was concluded that a syringe pump could produce the flow rates required. A syringe pump operates by having a specific volume syringe clamped to a frame. A step motor is used to move a threaded screw to either push-in or retract the syringe plunger. The volume of the syringe is critical, as this will determine the minimum and maximum injection rates. This apparatus can be controlled using a notebook computer

102 Chapter 3 and a program called Pump Scheduler, to initiate time-based commands, in which an entire experimental program can be constructed and executed. The syringe pump (Micro-Liter Syringe Pump Module, Harvard Apparatus, MA, USA) was installed, and an appropriate case to contain the electronic components was fabricated. The syringes (1700 series GASTIGHT® Syringes, Harvard Apparatus) utilised ranged from 0.5 μL to 25 μL and coupled with the syringe pump achieved minimum and maximum injection rates of 0.0014 μL h-1 to 1191 μL h-1. This liquid analyte injection methodology overcame the injection rate limitations inherent in the peristaltic pump, and the required flow rates were achieved.

Following the configuration of the syringe pump coupled to the nebuliser, the carrier gas-atomised analyte mist is directed into the vaporisation unit. This unit consists of a quartz tube (550 mm length x 50 mm OD) which is mounted through the centre of a tube furnace. The nebuliser is secured to the entry port of the quartz tube and a stainless steel check valve is attached at the exit port, immediately downstream from the furnace, through a glass-to-Kovar® seal connected to Swagelok® fittings. The principle of this vapouriser is to direct the analyte mist through the quartz tube which is set to a temperature above the boiling point of the liquid analyte. As the mist flows through the tube furnace the atomised droplets volatilise rapidly, due to their high surface area. The analyte is now in a volatilised form for the subsequent dilution and delivery to the sensor testing chamber. The temperature within the tube furnace is maintained by using a temperature control box connected to a Variac and an in-house fabricated nickel-chromium K-type thermocouple (-200 °C to +1200 °C range).

Downstream from the exits port of the vapouriser, a one-way check valve, is installed. The check valve selected (¼ inch Stainless Steel Check Valve, Parker-Hannifin Corporation, CA, USA) was manufactured from 316 stainless steel and fitted with a Viton seal for increased its chemical resistance. The directional pressure which opens the valve, termed the ‘cracking pressure’, needs to be quite low in this particular system as the level of back-pressure exerted on the liquid analyte injection interface could potentially cause flow problems. The distributor of this particular check valve specified a 1 PSI cracking pressure, which should not cause significant back pressure on the liquid analyte injection system.

103 Chapter 3

Immediately following the check valve, a flow line was connected to a ¼ inch union tee, which was also connected to the second gas flow line or the diluent line (Fig. 3.9). This diluent line comes from the first union tee previously mentioned, with the gas flow rate of this line controlled by a second gas MFC (MC-1SLPM-D, Alicat Scientific). As the name suggests this gas flow line is responsible for the dilution of the carrier gas-analyte vapour mixture generated in the vapouriser. The control of diluent flow coupled with different injection rates allows the user to assess a range of analyte concentrations during a sensor testing experiment. To ensure efficient mixing of the analyte vapour and the diluent, a baffled tube mixer (Series 250 Stratos Tube Mixers, Koflo Corporation, IL USA) was installed immediately before the entry port of the fabricated sensor testing chamber. The tube mixer consists of fixed right and left hand helical mixing elements designed for the mixing of light viscosity components, with the elements and housing fabricated from 316L stainless steel for increased chemical resistance.

Figure 3.12 A front-view photograph of the 'Phase-II' prototype chemical vapour generator and dilution system, including the sensor testing chamber. Note: in subsequent modifications the peristaltic pump (centre right) was replaced with a syringe pump.

104 Chapter 3

The general setup for the ‘Phase-II’ prototype chemical vapour generator and dilution system is illustrated in Figure 3.12, with the peristaltic pump replaced by a syringe pump. With the control of airflow (Q) and injection rate (R) parameters, it was possible to determine the analyte vapour concentration generated in the vapouriser by the re-arrangement of Equation 5. In a typical calculation the final concentration of analyte delivered to the sensor testing chamber was the vital parameter to be determined. Operationally, a desired concentration to be delivered to the sensor testing chamber was chosen for a particular analyte, and from this each different parameter was calculated so that this concentration could be achieved. When all the other parameters were calculated, the data were used to establish the injection rate (R) required, with the typical total gas flow rate to the sensor testing chamber, typically defined at 500 mL min-1.

To obtain a 20 ppm concentration of acetone (density = 0.785 g mL-1) at 500 mL min-1 total flow rate, the acetone concentration in the vapouriser would be set to 100 ppm. The dilution would be set at 5x, with the gas MFC connected to the nebuliser set at 10% full scale (100 mL min-1) [500 mL min-1 divided by the dilution ratio]. The air flow (Q), gas MFC flow rate, is now known and is converted to m3 h-1. In setting the vapouriser concentration at 100 ppm, CEA can be determined, using the conversion from ppm to mg m3 -1. This makes it possible to calculate the injection rate (R) required, at 1.8156 μL h-1, and with the appropriate syringe volume selected an experimental run can be initiated. A spreadsheet was constructed for ease of calculations which allowed swift injection rate calculation to be undertaken for a wide range of vapouriser concentrations and subsequent dilutions. The only physical data required for a particular analyte was the density and molecular weight.

The sensor testing chamber, printed circuit boards and measurement instrument from the ‘Phase-I’ prototype (Section 3.3) were also utilised in the ‘Phase-II’ prototype. In addition, there is a sampling port, following the exit port of the sensor testing chamber. This is a modification so that subsequent analytical instrumentation, such as in-line gas chromatography, can be coupled to the system for ‘real-time’ monitoring of the vapour characteristics, such as concentration profiles and for quality control measures.

105 Chapter 3

3.4.3 Prototype problems

The problem regarding liquid analyte injection rates required for the operation of this system was discussed in Section 3.4.2, and led to the replacement of the peristaltic pump with a low-flow rate, high precision syringe pump. With the ‘Phase-II’ prototype constructed the system in its entirety it needed to be trialled and calibrated, during these trials several problems with the setup became apparent. Firstly, significant back pressure was created within the quartz tube vapouriser, which resulted in flow in the opposite direction. Obviously, this was not ideal for the system, as the liquid analyte was not reaching the tip of the nebuliser and thus was not entering the vapouriser. The source of the back pressure was the check valve. The back pressure was greater than the positive pressure obtainable using the syringe pump, and on removal and testing, the cracking pressure of the valve was found to be substantially higher than the 1 PSI specified from the distributor. With the check valve removed the exit port of the quartz vapouriser tube was connected to the union tee through stainless steel tubing cut to size. With this alteration the system could operate and no further back pressure problems were evident.

System testing re-commenced and following a few more trials, another problem with the system was encountered, involving the nebuliser. It was found that under the testing parameters the liquid analyte was not being atomised into a fine mist. This was quite apparent at gas flow rates of <25% full scale, as large droplets of the liquid analyte would be produced at the tip, and would drip into the quartz tube. This would not occur at higher gas flow rates (>25%), and operating in excess of this flow rate, significantly reduces the concentration ranges achievable with the system, identifying a large design setback. Even with the operation at higher gas flow rates, it was evident that a pulsing spray effect was occurring at the nebuliser tip. This occurred due to the fact that as the liquid analyte flow approached the tip of the nebuliser, the pressure differential at the tip of the nebuliser would dramatically increase the liquid analyte flow rate by forcing the liquid out of the nozzle to equilibrate pressures. This also had a significant effect with the concentration of the analyte having an inherently large error. Effectively, the dynamics of the nebuliser in the system significantly limited the concentration ranges achievable by dramatically reducing the dilution factors possible, and even at higher flow rates the mechanism of atomisation introduced large errors for analyte concentration calculations.

106 Chapter 3

Since working at low gas flow rates is a requirement these inherent nebuliser problems could not be resolved. Compounding this problem was the fact that there were no available nebuliser units that could be used with the gas flow rates, and injection rates, required for this system to operate. These inherent problems led to the re-design of a more appropriate system and the ‘Phase-III’ prototype development.

3.5 Analyte delivery system - ‘Phase-III’ prototype

3.5.1 Chemical vapour delivery system

The limitations experienced for both the ‘Phase-I’ and ‘Phase-II’ prototypes for a chemical vapour generator and delivery system, necessitated a major system re-design. Analysis of the previous systems revealed that the majority of the problems stemmed from trying to create a vapour of an analyte within the system, or in-line. The reasoning behind trialling this in-line system was to have greater control over the dynamic concentration range achievable by the system, and to make it versatile, i.e. to cover a large range of analytes. The design of the ‘Phase-III’ prototype would revolve around the possibility of external vapour generation and focus more on being a vapour dilution system. It should be mentioned that in addition to the following ‘Phase-III’ prototype, static vapour delivery systems were also investigated but not pursued, as dynamic vapour delivery was preferred.

3.5.2 Vapour dilution system

The ‘Phase III’ prototype would attempt to be highly versatile with the potential to analyse ‘typical’ gases, vapours of liquid and solid analytes. In the design of this prototype attempts were made to use equipment from previous prototypes. As mentioned previously, the handling and dilution of typical compressed gas analytes is an uncomplicated task, but the vapours of liquid and solid analytes had a greater research priority.

This prototype did not revolve around an in-line chemical vapour generator and so other options were explored. It was decided that headspace sampling of an analyte vapour, generated externally to the system, was the most suitable option. This would require a

107 Chapter 3 small volume (~1 mL) of the liquid analyte of interest placed in a round bottom flask (1 L) and a septum fitted. The liquid/vapour phases were allowed to reach equilibrium and, using a gas-tight syringe, the headspace of the round bottom flask was sampled. The vapour-phase analyte concentration contained in the syringe could be determined, based on the analyte vapour-pressure and the temperature when sampled. This methodology can also be readily applied to obtain vapour samples from solid analytes, if vapour pressure data is available. This also removes the inherent problems of in-line vapour generation, observed in previous prototypes.

Syringe pump Parameter Regulating needle valve analyser

UHP Test Mass flow controllers chamber Exhaust N2 Cyclonic mixer

Regulating needle valve Detachable bubbler Figure 3.13 Schematic of the ‘Phase-III’ prototype vapour delivery system, including sensor testing chamber and measurement instrumentation.

The schematic of the ‘Phase-III’ prototype is illustrated in Figure 3.13. The design allows the analyte of interest headspace sample to be introduced through the use of a gas-tight syringe mounted in a syringe pump, which eliminates the majority of the previous problems encountered in earlier prototypes. However, sampling the headspace does significantly limit the concentrations of the vapour-phase analyte obtainable. This is due to the fact that the concentration of the analyte in the vapour-phase is directly correlated to the vapour-pressure of the compound of interest. For example, a compound with a high vapour-pressure will inherently have a high vapour-phase concentration, with the opposite for a low vapour-pressure compound, thus reducing the dynamic concentration range possible. The restricted concentration range was alleviated by the introduction of an analyte chemical bubbler on the second gas flow line. If a higher concentration of a liquid analyte is required then this bubbler can be utilised and the first gas flow line becomes the diluent line. In theory, this prototype

108 Chapter 3 allows for a large concentration range to be accessed (ppt to ppm), depending on the specific analyte.

A description of the components of the ‘Phase-III’ prototype (Fig. 3.13) follows. The carrier gas depicted is UHP-N2, as this is the preferred carrier and diluent gas for the experimental testing of vapours originating from organic liquids and solids. However, in this system, different gas types can be utilised as the carrier gases and, even more importantly, different analyte gases such as CO, CO2, and H2 can be used for sensor testing purposes.

Downstream from the UHP-N2 cylinder pressure regulator, the 1/8 inch OD stainless steel tubing is split into two separate lines by a 1/8 inch union tee. The first line described is the gas flow line connected to the syringe pump in the schematic (Fig. 3.13). A stainless steel 1/8 inch OD is coupled to the entry port of a gas MFC (MC-1SLPM-D, Alicat Scientific), a schematic is illustrated in Figure 3.10. The range of this gas MFC is from 10 sccm to 1000 sccm and is calibrated and certified for the accurate gas flow rate control of 16 different gases. The gas MFC is operated through FlowVision™ software, where time-based command scripts can be programmed to control the gas flow for an entire analytical run.

Figure 3.14 Photograph of the syringe pump injection port sealed with a gas chromatography injection port septum (blue). 109 Chapter 3

Attached directly to the gas MFC exit port is a ¼ inch Teflon® tubing that is connected to a hard mounted ¼ inch union tee (Fig. 3.14) , perpendicular to the gas flow direction, where the needle of a gas-tight syringe, mounted in a syringe pump, can pierce the septum and introduce a headspace sample. To provide a gas tight seal at this connection, a gas chromatography septum (3/8 inch blue septum, Alltech, VIC, Australia) is secured by the standard ferrule and nut of the union tee (Fig. 3.14). As the methodology of this prototype differs substantially from earlier ones with a vapour-phase analyte being injected into the system, and subsequently diluted, significantly increased injection rates were required. A different syringe pump (Milliliter Syringe Pump Module, Harvard Apparatus) was obtained and an appropriate case to contain the electronic components was fabricated and installed. The syringe pump was mounted on a retractable stage, which was capable of manual height adjustment, as illustrated in Figure 3.15. This retractable stage imparted the ability to insert or retract a mounted syringe from the system, while the mounted syringe height required adjustment, as different volume syringes had slightly differing heights when mounted in the syringe pump. This method required increased injection rates and increased volume syringes were obtained. The syringes (Gas-tight syringe pre-fitted with luer lock valve, SGE Analytical Science, VIC, Australia) utilised, ranged from 2.5 mL to 50 mL, and coupled with the syringe pump achieved minimum and maximum injection rates of 10.24 μl h-1 to 26.56 mL min-1. Furthermore, the previously utilised syringes (1700 series GASTIGHT®) could also be integrated into this system. Each SGE Analytical Science syringe was equipped with a manual shut-off valve to contain the analyte vapour within the syringe post-headspace sampling, and to only allow the analyte vapour into the dilution system when intended.

From the exit port of the union tee a length of ¼ inch stainless steel tubing is coupled to a ¼ inch stainless steel regulating needle valve (SS-1RS4, Swagelok®), which regulates the flow of gas into the gas mixing chamber. This feature also doubles as a preventative measure to eliminate any potential back-pressure environments that may be established, with the promotion for gas flow dynamics towards to the sensor testing chamber.

110 Chapter 3

Figure 3.15 Photograph of the syringe pump mounted on a retractable, adjustable z-axis stage with the syringe needle piercing the septum and injecting a headspace sample into the vapour dilution system.

Figure 3.16 Photograph of the detachable glass chemical bubbler secured in place through glass-to-Kovar® tube connectors and supported on a plastic die designed to allow back-pressure release, if necessary.

111 Chapter 3

The second gas flow line will now be described. This refers to the line that is equipped with a detachable chemical bubbler (Fig. 3.13). Attached directly to the gas MFC exit port is ¼ inch Teflon® tubing which is directed through an adjustable screw-mount to a ¼ inch tube connector union. Between this ¼ inch tube union, and a second stainless steel regulating needle valve, lies a detachable chemical bubbler (Fig. 3.16), with glass-to-Kovar® tube connectors for chemical resistance, and to reduce possible contamination. As mentioned previously, due to the chemical vapour generation methodology utilised for the syringe pump gas flow line, the analyte concentrations are dependent on the analyte’s vapour pressure. This chemical bubbler allows the production of increased vapour-phase concentrations from liquid analytes, and significantly extends the dynamic concentration range of this prototype. This gas flow line, in combination with the first gas flow line, can be controlled to dilute the analyte vapour expelled by the chemical bubbler.

Figure 3.17 Photograph of the in-house fabricated cyclonic mixer consisting of two gas flow line entry ports and a central ‘submerged’ exit route directed towards the sensor testing chamber.

Both of the two gas flow lines are directed to an in-house fabricated ‘cyclonic mixer’ (Fig. 3.13), consisting of a modified, stoppered 250 mL conical flask (Fig. 3.17). The conical flask is inverted and two glass-to-Kovar® entry ports are joined, to the now top

112 Chapter 3 of the flask, and are symmetrically positioned in relation to one another. A ¼ inch capillary was installed through the centre of the flask, and runs down most of the flask height. This capillary tube is then bent at a 90° angle and the glass-to-Kovar® exit port is connected to a ¼ inch tube connector union, which directs the analyte mixture to the sensor testing chamber via ¼ inch Teflon® tubing. The Teflon® tubing utilised reduces undesirable 90° angle gas flow through the system, to maximise space utilisation, to limit contamination and increase the chemical resistance of the setup, and to provide structural flexibility at glass-to-Kovar® joins to reduce structural stress on these components. The purpose of this ‘cyclonic mixer’ is to ensure the turbulent and rapid mixing of the analyte vapour and diluent gas, prior to its introduction to the sensor testing chamber. The symmetrical entry ports, and inherent curvature of the internal surface of the flask, should promote the turbulent mixing desired in a cyclonic fashion, as the vapour-carrier gas mixture flows to the bottom of the flask and then is directed vertically to the exit port. The cyclonic mixer is stoppered and is easily detached, for ease of cleaning in order to reduce contamination and analyte ‘ghosting’ between analytical runs.

The sensor testing chamber and printed circuit boards from the ‘Phase-I’ and ‘Phase-II’ prototypes were also utilised in the ‘Phase-III’ prototype (Section 3.3.2). Within the sensor testing chamber a data logger (1-Wire®, iButton, Maxim Integrated Products, CA, USA) to monitor and record temperature and humidity was installed. The intended measurement equipment was varied due to instrument availability problems.

The device measurement instrumentation integrated into the ‘Phase-III’ prototype is based on a potentiostat and electronic signal recorder, as displayed in Figure 3.18. The potentiostat (model EA161, eDAQ Pty. Ltd., NSW, Australia) is an entirely software controlled, three-electrode system that can be used to apply potentials of ± 10 V, where device currents are in the nanoampere to milliampere range. In a typical chemiresistor device platform (Chapter 4) the working and reference electrodes of the potentiostat are coupled to a modified BNC connector, linked to an appropriate BNC connector (device source electrode) on the breakout board. The counter electrode is connected in the same fashion to the respective sensor drain electrode BNC outlet on the breakout board. This system is utilised to apply a user-specified potential (V) between the source and drain electrode of the sensor, with the active sensing materials providing the electrical 113 Chapter 3 bridging of the circuit. With this potential applied, the I, or the electron flow, through the sensing material, is monitored throughout an analytical run (Section 1.4.1.2). The electronic recorder (e-corder, model EA201, eDAQ) is coupled to the potentiostat, and is a two-channel recording and analysis system. This instrument records analogue data from the electrode system, and allows real-time monitoring of electrochemical systems, with packaged data analysis software. Basically, this unit records the changes in the electrical properties of the individual sensors when an analyte exposure experiment is conducted.

Figure 3.18 Photograph of the electrical measurement instrumentation utilised in the 'Phase-III' prototype which consists of an eDAQ potentiostat (top) and e-corder (bottom).

The constructed ‘Phase-III’ prototype, illustrated in Figure 3.19, is the final system developed for the subsequent testing of the fabricated sensors described within this thesis. There are many advantages of this ‘Phase-III’ system over the previous designs, e.g. this system has the capability to deliver analyte vapours from gases, organic liquids and solids. The gases can be directly introduced, and, if required, diluted through the gas MFCs. Vapours of liquids can be introduced through the syringe pump and/or the chemical bubbler, and headspace vapours of solid samples can be introduced using the syringe pump. The combination of the syringe pump and the chemical bubbler in the

114 Chapter 3 same system allows a large dynamic concentration range to be achievable for a suite of intended analytes. Furthermore, it is important that a low analyte concentration range is achievable (ppb), as this allows for the accurate determination of sensor detection limits. Additionally, in designing this vapour dilution system, versatility is a key issue, and flexibility for testing a range of sensors that will emerge from different research projects, beyond the current direction, is a necessity.

Figure 3.19 Photograph of the 'Phase-III' vapour dilution system, including the sensor testing chamber and notebook computer. Note that the electrical monitoring hardware is not captured in this image.

Another capability associated with this design is the ability to alter the carrier gas environment. Typically, UHP-N2 is utilised as the carrier gas environment, however, the gas MFCs are calibrated for 16 different gases, including H2, O2, CO, CO2, He, neon (Ne), argon (Ar), krypton (Kr), xenon (Xe), propane, butane, acetylene, ethylene, isobutane, and sulfur hexafluoride. This allows for not only direct sensor testing with these gases as analytes, but also allows these gases to be the principal environment in which the testing is undertaken, and to monitor if this induces changes in sensors characteristics.

115 Chapter 3

It is common in field applications of gas and vapour sensors that when removed from the controlled environment of laboratory testing, common interferants can have an effect on the operation of the sensor and can lead to false-positives.[277, 278] Through the use of either the chemical bubbler or syringe pump, depending on the application, interferant chemical species can be readily introduced. For example, the chemical bubbler could be utilised to obtain a carrier gas-vapour mixture for acetone, while the diluent line could be spiked, using the syringe pump introduction system, with a chemical interferant such as toluene. This methodology is not only limited to the mimicry of field contaminants, but it can also be utilised to assess vapour-phase analyte mixtures, to determine if the fabricated sensors can differentiate between multi-component systems.

As the different pieces of equipment, such as the gas MFCs, the syringe pump and measurement instrumentation are all linked and controlled through PC-compatible software, automated experiments are possible. The gas MFCs are controlled through the program FlowVision™ and software was created on request (Pump Schedule, SDR Clinical Technology, NSW, Australia) for the control of the syringe pump. The FlowVision™ software (Fig. 3.20a) can be utilised to construct time-based commands, hence the precision mapping of both gas MFC flow rates for an entire analytical run is possible. Time-based commands were also required for the control of the syringe pump, which originally was controlled through the program Windows® Hyper Terminal. The program Pump Schedule (Fig. 3.20b) was developed to allow time-based command prompts to control syringe pump injection rates throughout an entire analytical run. The simultaneous combination of these two programs, coupled with the EChem™ electrical monitoring software, allows the pre-programming and automation of an experiment. Such an experiment can have multiple different concentrations analysed over the user determined time period. This allows convenient sensor testing and it also reduces user interference during an analytical run.

116 Chapter 3

Figure 3.20 Screen capture images of the FlowVision™ software graphical user interface (left), and the Pump Schedule graphical user interface (right). Note the functional parameters within each system are user-specified.

Configuration of the EChem™ software was required to probe the desired parameters of the sensors, and allowing these electrical parameters to be monitored over time. An easy methodology for sensor monitoring is to compare sensor I-V curve data acquired from a linear voltage sweep, such as -1 V to +1 V. However, this sampling methodology is not suitable for ‘real-time’ sensor monitoring over a user-specified time, and the correct software configuration was determined, utilising a multi-pulse amperometry configuration. This configuration allows a specific base potential (V) to be applied between the working and counter electrodes, with a user specified width (time between sampling, ms) and the number of steps (a parameter coupled with width to allow total analysis time to be established). The ‘multi-pulse’ components were not utilised and hence the system is now configured to apply a voltage potential for a specified amount of time, and constant sampling through an analytical run. With this configuration selected ‘real-time’ data is displayed, and logged, I with respect to time (s), hence the change in I for an individual sensor can be monitored (Fig. 3.21). At the end of an analytical run this data can be exported and sensor resistance calculated, Ω = V / I, and sensor response with respect to time can be generated for further analysis. The sensor response (%R) is determined by the change in sensor resistance (ΔR) in relation to the sensor baseline resistance (Ro) as a percentage; %R = (ΔR / Ro) x 100. Utilisation of this data processing methodology allows rapid comparability between sensors with differing baseline resistance values.

117 Chapter 3

Figure 3.21 A screen capture image of the EChem™ graphical user interface, with test parameters located to the right of screen, and the main window displaying sensor I profile (µA) with respect to time (s).

3.5.3 System noise

As with all the prototypes some inherent problems arose once the system was commissioned for sensor testing. The operation of the syringe pump and chemical bubbler, for the dilution of the vapour-phase analytes worked without fault. However, a significant electrical noise component was observed upon testing a fabricated sensor within the system (Fig. 3.22). Through the analysis of the response of an individual SWCNT chemiresistor sensor, large peaks and troughs of similar amplitude and frequency were swamping the parameter output. A full systematic breakdown and protocol was undertaken to locate the source(s) of these electrical noise problems. Many of the components, such as the gas MFCs and syringe pump, were found to be ‘floating’ above electrical earth. Subsequently, these instruments were loop grounded to the metal frame of the portable trolley on which the prototype was secured. This grounding step removed the high frequency large electrical noise signals, however, some medium to low frequency noise still existed. Measures such as grounding each individual BNC connector on the breakout board, and connecting the PCB D-type pin to mains earth, further reduced and removed these electrical noise problems (Fig. 3.22). Other areas assessed included, mains power location, electrical shielding of cables and the sensor testing chamber, and different mains power boards while a sensor I profile was monitored to determine subsequent effect.

118 Chapter 3

Figure 3.22 A sensor response curve for a SWCNT chemiresistor configuration monitored before and after attempts to lower the effect of electrical noise in the system.

Figure 3.23 Sensor response curve for a Pd nanoparticle/SWCNT hybrid chemiresistor sensor upon exposure to H2 gas (5% to 9%), located at positions A and B in the sensor testing chamber.

119 Chapter 3

3.5.4 Sensor test chamber position verification

A Pd nanoparticle/SWCNT hybrid chemiresistor was fabricated and utilised to assess the effect of sensor position in the sensor testing chamber. The sensor underwent two separate analytical runs, exposed to 5% to 9% H2 gas with 1% increments. The response curve in Figure 3.23 clearly illustrates that minimal sensor response differences can be attributed to sensor position. This verifies that sensor location does not need to be considered when monitoring multiple sensors. In addition, maximum sensor response values for each 1% increment of H2 gas, indicate a linear correlation. This provided additional support that the vapour dilution and delivery system is operating as intended.

3.6 Conclusions

This current chapter showcases the importance of downstream analyte preparation methodologies in the area of chemical sensor testing. In the absence of controlled delivery of a known concentration analyte to a particular sensor system, the reliability of the data generated will be in question. Chapter 3 provides an overview of current analyte delivery methodologies and describes the processes involved for the design and development of a suitable gas and vapour generation and delivery system for subsequent sensor testing.

A breakdown of three different system prototypes, including operational principles and mechanisms, design constraints, and working components has been established. Chapter 3 highlights the chronological development of three different gas and vapour generator prototypes (Phase-I - III), and indentifies key system failure areas and attempts to provide practical solutions through system re-design and component manipulation.

The successful fabrication of a fully commissioned gas and vapour generation and delivery system has been established, based on the Phase-III prototype. This prototype provides a nexus between common-place chemical bubbler-based generation and syringe pump injection techniques to provide a system capable of satisfying the original requirements notably: (i) the ability to generate, dilute and precision deliver vapours

120 Chapter 3 from multi-phase analytes, (ii) the capability of generating a large dynamic concentration range ppt - ppm, dependant on specific analyte, (iii) allow the introduction of chemical interferants and multi-component systems, (iv) allow a range of carrier gas environments to be established, and (v) allow automated sensor testing. In meeting these criteria the system shows functionality that is suitable to a wide range of possible sensor testing conditions and becomes applicable to other current and future research projects.

In addition, the primary workings into the development of downstream components necessary for sensor testing, such as a sensor testing chamber and electrical connectivity components are described in Section 3.3.2.

121

Chapter 4 SWCNT-Based Chemiresistor Sensor Testing for Compounds of National Security Interest

Chapter 4

4 SWCNT-based chemiresistor sensor testing for compounds of national security interest

4.1 Introduction

In the current global environment there is a heightened level of public and governmental disquiet due to the reality of impending terrorist attacks. This wide-spread sense of terror is compounded by the inherent ease of manufacture and effectiveness of specific CWAs, utilised in small-scale terrorist operations, which could result in mass casualties in both military and civilian populations.[442] As a direct reaction from these threats it is of paramount importance that new detection technology is developed for purposes of CWA release monitoring, personnel safety, operational intelligence, warfare tactics, in an attempt to minimise and contain the imposed threat.

Phosgene, COCl2, is classified as a choking agent which are the forerunners of modern chemical warfare.[443] This CWA was used prolifically throughout World War I, initially by Germany and subsequently by allied forces.[442] It is estimated that phosgene release alone accounted for approximately 80% to 85% of poison gas deaths in WWI,[442, 444-446] with the equivalent of 852 man-years lost due to hospitalisation.[447] This capacity for devastation coupled with current commercial production[448] and, more importantly, its uncomplicated synthesis from common chemicals,[449] makes phosgene a serious weapon in a terrorist’s arsenal.

Phosgene is a colourless gas, with toxic effects well below its odour threshold of 0.4 ppm.[450] This is important as a potential victim may not be aware of exposure, because the full effects typically appear several hours after.[443] This, coupled with an increased awareness that phosgene is a serious candidate for use in a terrorist attack, has sparked recent research and development into different phosgene sensor technologies.

Pioneering research by Frye-Mason et al., utilised a unique optical detection methodology in which a on/off chromophoric report molecule is utilised to react with an electrophilic molecule like phosgene and create a detectable on/off fluorescence switch.[451] Following this research, limited studies have been undertaken in this area,

123 Chapter 4 utilising similar fluorescence switching principles.[452, 453] Another recent development is the utilisation of MEMS-based technology by Porter et al., where an organic polymer-inorganic particle material is coated on a microcantilever.[454] The interaction of phosgene displaces the polymer creating more free volume which can be monitored through the piezoresistive microcantilever. There has also been an interest in both in situ and remote phosgene detection utilising spectrometry techniques. These techniques range from remote measurements using scanning image FT spectroscopy,[455] to the more recent development of surface-enhanced Raman scattering utilising functionalised nanoparticles.[456] Each of these different systems has its own set of advantages and disadvantages, yet the general consensus is that there is a need for a hand-held or remote system, cost effective, low power consumption sensor for the detection of phosgene, and to this end, albeit with limited published material, electrochemical sensors dominate.[457-461]

Since the discovery of CNTs in 1991 by Iijima[19] and two years later SWCNTs,[20, 21] carbon nano-materials have attracted much attention both in academia and commerce. It was not until recently that Kong et al.[223] and Collins et al.[224] demonstrated that these materials had the potential for gas sensor applications with the construction of a

NT-FET for the detection of NO2 and NH3, and O2 respectively. This sparked a phenomenal amount of research into the area of CNT-based sensors for gas and vapour sensing.[151, 226, 462] Over the years there have been many different types of CNT-based gas sensors, including chemiresistors,[229] chemicapacitors,[235] and gas ionisation sensors.[234] Particular attention should be given to the recent work of Strano et al., where a micro gas chromatography column was coupled to a functionalised SWCNT sensor array for the highly sensitive on-chip detection of DMMP. This research initiates the practical advancement of laboratory systems miniaturisation incorporating CNTs. This is not to mention the scope for chemical and physical modifications of CNTs, such as surface coatings,[277, 463] nanoparticle loading,[238, 251] and DNA-decoration[269] which all have been utilised to increase analyte sensitivity and selectivity of the CNT-based sensors. The research presented below utilises acid-induced defect sites on the SWCNTs to increase the side-wall and tip reactivity due to the introduced carboxylic acid functional groups, as displayed in Figure 4.1.

124 Chapter 4

Figure 4.1 Typical acid-induced SWCNT surface defect sites, indicating nanotube tip and side-wall functionalisation, along with pentagon-induced structural defects. Adapted with permission from [133], © 2002 John Wiley & Sons, Inc.

To the authors’ knowledge there is no published literature to-date where the unique properties of nano-materials, more specifically carbon nano-materials, has been utilised in an electrochemical sensor as the active matrix for the detection of phosgene. The research described in this thesis focuses on harnessing the power of nano-materials, in particular SWCNTs, in a chemiresistor to provide a launch pad for a major shift in the detection of phosgene. An investigation into the sensor response to G-series nerve agents, sarin and soman, is accessed through chemical simulants. A comparison between single IDE sensor geometry and multiple-array systems is also undertaken. Furthermore, this chemiresistor sensor configuration allows for the highly sensitive detection of phosgene over the common organic solvent and other halogenated systems to be trialled. This current chapter details the first reported study utilising a carbon nano-material-based sensor matrix for the detection of the CWA phosgene.

125 Chapter 4

4.2 Experimental

4.2.1 Materials

SWCNTs (SWCNT 90s, Cheap Tubes, Inc. VT, USA) were obtained, which were produced by a C-CVD synthesis method with an as-received purity >90% (<1.5 wt% ash, >5 wt% MWCNT, <3 wt% amorphous carbon, <2 nm tube diameter). DMF, phosgene (20%) in toluene, DMMP, acetone, ethanol, benzene, toluene, benzyl alcohol, chloroform, chlorobenzene, 1,2-dichlorobenzene, carbon tetrachloride, benzyl bromide, concentrated nitric acid and concentrated hydrochloric acid were obtained from Sigma-Aldrich and used without further purification. The concentrated nitric acid was purified by sub-boiling distillation using a quartz still. DIMP was obtained from Alfa

Aesar (Lancashire, UK) and used without any further purification. UHP-N2, UHP-H2, and acetylene cylinders were obtained through BOC Gases (NSW, Australia). Metal catalyst particles were introduced using nickel (II) sulfate hexahydrate (NiSO4·6H2O) (Sigma-Aldrich), from which stock solutions were made (1 M). All carbon-material synthesis was undertaken in the laboratories of Dr. Gerard Poinern (Curtin University of Technology, WA, Australia). The chemical vapour delivery system and sensor testing facility was designed and fabricated in-house, as described in Chapter 3. The chemiresistor device platforms were designed and fabricated in-house, as detailed in Section 4.2.2. The temperature and humidity data logger (1-Wire®, iButton, Maxim Integrated Products), monitored conditions within the sensor testing chamber.

4.2.2 Methods

SWCNT purification and dispersion: The as-received SWCNTs were further purified by a modified literature preparation.[464] SWCNTs (100 mg) were refluxed in re-distilled 70% HNO3 (25 mL) at 130 °C for 8 h. This acid treatment introduces side-wall defects which have been experimentally and theoretically shown to be the primary mechanism for charge transfer (i.e., the conductance response) as preferential binding of an adsorbate occurs at these induced defect sites.[104] The suspension was cooled to room temperature and the supernatant rendered neutral using centrifugal decantation (3000 RCF) with MilliQ water (>18 MΩ cm-1). The resulting SWCNTs were freeze-dried and heated at 420 °C in a furnace in dry air. The debundling of the SWCNTs and further separation of encapsulated catalyst particles was carried out by a

126 Chapter 4 modified published literature method.[465] The amide groups of DMF can easily attach to the surface of the SWCNTs, assisting in the disruption of van de Waals forces in the SWCNT bundles and producing stable suspensions.[229, 466] The purified SWCNTs (2.0 mg) were dispersed in DMF (100 mL) and ultrasonicated for two hours with a sonic lance (4 W RMS power, Branson Sonifier 150) producing a dark black solution. The resulting solution was divided into equal aliquots and centrifuged (13900 RCF, JA-25.5 Rotor, Beckman Coulter, CA, USA) for 90 min, at ambient temperature. The top three quarters of the supernatant was collected to afford the final dispersion, which was utilised for the fabrication of the SWCNT chemiresistors. In relation to the untreated SWCNTs, the as-received material was subjected to ultrasonication in DMF and centrifugation under the same conditions as mentioned above.

Device platform fabrication: The device platform design was based on a suitable model in the literature.[229] A 3 inch diameter (300 μm to 350 μm in thickness) Si <100> n-doped (phosphorus [1 Ω cm-1 to 20 Ω cm-1]) wafer was washed with isopropyl alcohol, then acetone and dried under a N2 gas flow. An insulating Si3N4 layer (300 nm) was then deposited using PE-CVD processes. PE-CVD conditions were as follows, 200 °C chamber temperature, 200 W RF plasma, 300 mTorr chamber pressure, with feedstocks of SiH4/NH3/N2 at flow rates of 5/50/100 sccm respectively, for a deposition time of 10 min. The Si3N4 insulating layer provides a shock and corrosion resistant, low stress film at room temperature that is pinhole-free, when compared to [467] ® SiO2. A negative photoresist (AZ 2035) was drop-coated and spun (Headway Research, Inc., TX, USA) at 4000 revolutions per minute (RPM) for 40 s. A thermal exposure (110 °C) for 60 s was undertaken followed by mask alignment (MJB 3, Karl Suss, Germany), utilising a pre-fabricated chromium photolithography mask (Bandwidth Foundry, NSW, Australia) and exposed to ultraviolet radiation for 20 s. Another thermal exposure (105 °C) for 60 s was completed and the wafer was placed in a beaker of MIF Developer (AZ® 300, Clariant Corporation, NJ, USA) for 60 s. The wafer was removed from the developer solution and thoroughly rinsed with distilled water (>18 MΩ cm-1). Light microscopy examination of the wafer revealed that these fabrication conditions were sufficient for the IDE finger resolution required.

Following photolithography, the samples were secured in an in-house built resistive metal evaporator for physical vapour deposition. The metal evaporator chamber was 127 Chapter 4 evacuated (3.4 x 10-7 Torr) and a chromium binding layer (5 nm) was deposited followed by a gold layer (50 nm) to form the metal electrodes, deposition rates were monitored by a quartz crystal microbalance. After deposition, the wafers were soaked in acetone for 30 min for the required lift-off to occur, yielding the devices. The resulting wafer sections were coated in a protective photoresist (AZ® 4562) and spun at 2000 RPM for 40 s and underwent a thermal exposure (100 °C) for 180 s to minimise the effects of Si dust during cleavage on the surface morphology. The wafers were sectioned as required and stored in the cleanroom to minimise possible contamination.

Figure 4.2 A schematic of the IDE which represents the basis of the sensor platform, scale bar represents 100 μm.

Each individual set of IDEs consists of 15 fingers, each 15 μm in width and 550 μm in length, with a finger gap of 10 μm (Fig. 4.2). Each individual electrode is connected to a bonding pad (200 μm x 250 μm) to provide a sufficient area for subsequent wire bonding. The advantages of utilising an IDE include the large electrode area to reduce nanotube contact problems, commonly seen in single CNT devices. A large number of nanotubes are inherently deposited on the IDEs which results in more robust sensing characteristics and increased reproducibility due to an increased statistical population. Furthermore, the large surface area of the IDE also increases analyte adsorption capacity.[229]

128 Chapter 4

Wire-bonding of the device platform to TO-08 Header: A wafer section (6 mm x 5 mm) containing four separate IDEs was secured to a TO-08 Header (Spectrum Semiconductor Materials, Inc., CA, USA) with a conductive silver epoxy paste. Additional conductive silver epoxy paste was applied to each bonding pad of the IDEs and appropriate lengths of gold wire (25 µm dia.) were attached. Once the epoxy had cured, fresh epoxy was applied to the appropriate pin connectors on the TO-08 Header and the free end of the Au wire was secured in the epoxy. This now allows the individual monitoring of each IDE device.

Deposition of SWCNTs onto the IDE: Prior to the deposition of the SWCNTs, each individual set of electrodes was tested to ensure no conductive bridging of the metal electrodes had occurred. The previously synthesised and dispersed SWCNT solution was drop-casted onto the surface of each IDE using aliquots of 0.02 µL from a 0.5 µL glass gas-tight syringe (Hamilton Company, NV, USA) and subsequently air-dried. I-V sweeps were carried out on each device until an ohmic response was achieved to provide evidence that the SWCNTs were the only material bridging the electrode gaps, with further support from scanning electron microscopy (SEM), illustrated in Figure 4.3. The suitability of low nanotube density random networks as electronic material has previously been established.[468] These initial electrical characterisations were carried out using a eDAQ Potentiostat and e-corder (e-DAQ) in a linear sweep mode (-1 to 1 V), at an appropriate I range, and data recorded using the EChem™ software (e-DAQ). Prior to analyte exposure the SWCNT-based chemiresistors were treated in a modified glass oven Kugelrohr (B-585, Buchi, Switzerland) put under vacuum at 150 °C for a minimum of two hours.[466]

Figure 4.3 A SEM image of the random network of SWCNTs between the chemiresistor source and drain electrodes, scale bar is 1 µm. 129 Chapter 4

Direct CNT growth onto the IDE platform: Direct growth of CNTs onto the IDE platform was undertaken using C-CVD processes. The C-CVD apparatus (fabricated in-house) consisted of pressurised cylinders of acetylene and UHP-He, with flow rates controlled by individual rotometers, joined downstream by a Y-piece tube connector. The gas flow is directed into a quartz tube (ca. 50 mm OD, 600 mm length) which is running through the centre of a horizontal tube furnace, where the substrate will be placed. The exit end of the quartz tube was fused to a glass tube to direct gas flow to a chemical bubbler which acted as a trap for reaction by-products. Ni catalyst particles were introduced to the pre-fabricated IDE by drop-casting a NiSO4 solution onto the substrate and oven-drying. This substrate is positioned on a quartz boat inside the quartz tube in the middle of the tube furnace. The system is flushed with UHP-He (10 mL min-1) and equilibrated at the required growth temperature (800 °C). Once the required growth temperature has been reached an acetylene/UHP-He (15/85) mixture is introduced, at a total flow of between 5 mL min-1 to 8 mL min-1 for a period of 30 min. After this, the acetylene feedstock was turned off and the system was purged with UHP-He and allowed to cool to room temperature.

Multiple-array sensor platform fabrication: Multiple-arrays of the IDEs were fabricated. These consisted of four separate IDEs, with the individual unit represented in Figure 4.2. All the source electrodes were connected and were accessible through a nominated header pin. All the drain electrodes were also connected and accessible through separate pins. This configuration gives rise to the single IDE being connected in parallel.

Analyte vapour delivery system: The chemical vapour delivery system has been previously detailed (Chapter 3). The analyte vapour delivery system is designed to allow a large dynamic concentration range, from ppt to high ppm, governed by the specific analyte vapour-pressure.[469] Fundamentally, this apparatus allows two vapour introduction methodologies. Firstly, a pre-made analyte vapour-phase mixture can be directly introduced to the system and with a secondary diluent gas feed creating the desired analyte concentration delivered to the sensor testing chamber. The second vapour introduction method is based on the generation of saturated vapours from a liquid analyte in a standard chemical bubbler or contained in a gas-tight syringe (Fig. 3.13). The introductory method utilised depends on the analyte of interest and the 130 Chapter 4 concentration range of the analyte to be studied in an analytical run, and to generate the results displayed in this current chapter the liquid bubbler route was chosen.

To introduce the analyte vapour using the chemical bubbler method a carrier gas flow is controlled by gas MFC 2 and directed to the chemical bubbler (Fig. 3.13). The bubbler contains the analyte (2 mL to 5 mL) at ambient temperature (25 °C) and the carrier gas is saturated with the analyte. This gas stream can be delivered directly to the sensor testing chamber or can be further diluted by a second gas stream controlled by gas MFC 1 (Fig. 3.13). As previously mentioned, to ensure appropriate mixing between analyte and diluent the sample flow lines pass through the cyclonic mixing chamber and are directed into the sensor testing chamber. This was the main analyte introduction method utilised in this current study. Different total flow rates were utilised throughout this study (500 mL min-1 and 1000 mL min-1). Based on these dilution calculations and the specific analyte vapour pressure, the concentrations delivered to the sensor testing chamber were calculated.

The phosgene utilised in this study was generated from a phosgene saturated toluene solution. The minimum concentration achievable was dependant on the minimum N2 flow required to successfully disrupt this matrix and release phosgene from the toluene. As toluene acts as a stabiliser of phosgene, lower concentrations from headspace sampling were not achievable. Furthermore, dilution of the saturated solution failed in phosgene generation, in the vapour delivery system.

Real-time electrical testing: The I of each device was monitored over an analytical run using a potentiostat and e-corder. The potentiostat was configured in multi-pulse amperometry mode using the EChem™ software and operational parameters optimised to achieve maximum sensor response; experimental justification for the chosen parameters will be described in Section 4.3.1.3. The typical experimental parameters utilised were an applied potential of 100 mV, one second sampling rate with experiment length and I-range dependant on the analytical run and analyte employed. A typical analytical run consisted of an initial purge of the system with UHP-N2, followed by a period of analyte exposure at a set concentration; this purge/exposure pattern is continued for increased analyte concentrations until the analytical run is finished. Other

131 Chapter 4 balance gases, UHP-H2 and UHP-Ar, were also trialled for sensor optimisation. In addition, temperature and humidity conditions of the sensor testing chamber were monitored to see their effects on the fabricated sensors using a 1-Wire® data logger (iButton, Maxim Integrated Products). The data can then be exported and analysed using an appropriate software package (Origin Pro 8.0).

Separation of semiconducting and metallic SWCNT based on alternating current (ac) dielectrophoresis: The IDE platform was utilised as the basis for dielectrophoresis-based separation of semiconducting and metallic SWCNTs. The source and the drain electrode were coupled to an arbitrary waveform generator (10 MHz, 33210A, Agilent Technologies). For ac dielectrophoresis, the function generator was operated at a sine wave frequency of 10 MHz, with a peak-to-peak voltage of 10 V. Under these conditions and taking into account the 10 μm electrode 5 -1 finger gap this system is capable of field strengths of 10 V m , which is ideal for metallic SWCNT deposition.[470] The function generator was switched on and an aliquot (10 μL) of a SWCNT suspension was drop-cast onto the electrode surface. Following a 10 min equilibrium period the liquid droplet was blown off by a stream of nitrogen and the function generator switched off.

Figure 4.4 a) A schematic of the IDE acting as a dielectrophoresis device, and b) an optical image illustrating the alignment of CNTs on an electrode under polarised light. Reprinted with permission from [407], © 2003 AAAS.

Methodology for sensor noise quantitation and limits of detection: From the response data it is possible to determine the detection limit of the sensor response to an analyte concentration. It should be noted that the current experimental setup is the

132 Chapter 4 limiting factor on the lowest detectable concentrations measured. The method utilised to calculate detection limits is based on the report by Li et al. with the same statistical methods applied to the sensor data described in this chapter.[229] The sensor noise

(RMSnoise) is calculated through this method, using the variation in the relative sensor response at baseline, based on the root-mean-square-deviation (rmsd). Following this method 10 successive data points along the baseline are extracted before analyte exposure. The baseline data were plotted and a fifth-order polynomial fitting curve applied. Statistical parameters and the fitting-curve equation were obtained within the data-point range and the variance (V 2 ) between the measured data and the calculated data can be worked out from the following equation:

2  2 i  yyV  (7)

where yi represents the measured datum point and y represents the calculated datum point based on the fitting function. The sum of the residuals is then applied to the following equation:

V RMS   2 (8) noise N

where N is the number of data points utilised to produce the curve-fitting function.

Once the RMSnoise is established, for each sensor response curve, the detection limit can be calculated. According to the IUPAC definition[471] when the response signal is at least three times the standard deviation of the noise, it can be considered an actual signal. Therefore, detection limits can be calculated by the extrapolation of the linear calibration curve to where the signal is equal to three times the RMSnoise, as described by the following equation:

RMSnoise DL(ppm)  3 (9) slope

133 Chapter 4

To extend this further, the limit of quantitation (LOQ) was assessed, for the analytes tested, to where the signal is equal to ten times the RMSnoise.

4.3 Results and discussion

4.3.1 Optimisation

4.3.1.1 IDE device platform

The IDE device fabrication methodology described in Section 4.2.2, represents the optimised condition which resulted in a high quality IDE product. Initial fabrication trials utilised an in-house fabricated glass emulsion photolithography mask, which was fabricated from an in-house printed mask design. It was quickly realised that the mask was not sufficient for the reproducible production of the first generation IDEs, with electrode arms missing and separate electrodes joined (Fig. 4.5), which would result in a device short if utilised. The yield for working first generation IDEs utilising this mask was <10%. Following these results the mask design print for the glass emulsion method was outsourced (Bandwidth Foundry), and the new print was utilised to fabricate another photolithography mask. The second generation IDEs produced using this mask, were of increased quality with very few defects (Fig. 4.6). The yield of working devices was significantly increased to 85%. It can clearly be seen that the electrode arms are more uniform, however, at this magnification slight width variations along the electrode arm axis can be observed.

Figure 4.5 Optical images a) illustrating the entirety of the IDEs, with 100 µm scale bar, and b) illustrating an increased magnification image of the top right corner of a), with a 50 µm scale bar, of the first generation IDE platform using a glass emulsion photolithography mask coupled with in-house printing.

134 Chapter 4

Figure 4.6 Optical images a) illustrating the entirety of the IDEs, with 100 µm scale bar, and b) illustrating an increased magnification image of the top right corner of a), with a 50 µm scale bar, of the second generation IDE platform coupled with outsourced printing.

After several mask processing runs, it was observed that the quality of the second generation IDEs were significantly reduced, with yields of working devices decreasing to <30%. This effect was traced to the quality of the photolithography mask, with mask definition significantly deteriorated, a common occurrence, apparently, in glass emulsion masks. This problem was rectified with the outsourcing of a chromium photolithography mask (Bandwidth Foundry), which produced high quality electrode feature definition, with working devices in excess of 98% yield. The extent to which the Cr photolithography mask increased the third generation IDE feature definition is illustrated in Figure 4.7.

Figure 4.7 Optical images a) illustrating the top right corner of the IDEs, with 200 µm scale bar, and b) illustrating an increased magnification image of the top right corner of a), with a 25 µm scale bar, of the third generation IDE platform developed using an outsourced Cr mask.

The third generation IDE platform offered a reproducible electrode configuration of sufficient quality to form the basis of the SWCNT chemiresistor sensors. This platform

135 Chapter 4 also offers advantages including scalability, due to the platform fabrication utilising common Si wafer processing technologies. For example, the technology transfer to market would not be hindered utilising these fabrication techniques, because the platform can be produced at low cost and in large quantities. For example, the processing of one Si wafer can produce over 600 individual IDEs, which results in each platform costing

4.3.1.2 SWCNT chemiresistor device

4.3.1.2.1 Drop-casting SWCNT dispersion

SWCNT chemiresistor devices were fabricated utilising a drop-casting strategy, which provided a method of controlling SWCNT density and can be incorporated into automated processes. In order to optimise the drop-casting procedure and obtain a grasp on device reproducibility, a different number of 0.02 µL additions of the SWCNT dispersion were drop-casted onto the IDE platform and subsequent characterisation was carried out. SEM was utilised to give a visual representation of the SWCNT density on the devices, as illustrated in Figure 4.8. It is clear that with increased numbers of 0.02 µL additions the SWCNT density on the IDE platform increases, so that drop- casting is an effective method for deposition of SWCNTs between the source and drain electrodes of the sensor substrate.

Additional analyses were undertaken to characterise the effect that SWCNT deposition density has on device resistance. The fabricated devices were subjected to a linear voltage sweep from -1 V to 1 V and the resistance calculated to 0.1 V, with the results displayed in Figure 4.9. A negative correlation between device SWCNT density and device resistance can be observed. This is to be expected as the higher the density of SWCNTs the more pathways for electron flow between the electrodes and thus less device resistance. In addition, from multiple analyses of SWCNT chemiresistor devices fabricated through the drop-casting methodology a moderate level of sensor baseline resistance reproducibility is evident (Fig. 4.9). In relation to ease of manufacture and

136 Chapter 4 sensor reproducibility the drop-casting method was adopted for the sensors utilised in this study.

Figure 4.8 SEM images of SWCNT density between the source and drain electrode for a) 1 x 0.02 µL addition (Device i), b) 2 x 0.02 µL addition (Device ii), c) 3 x 0.02 µL addition (Device iii), and d) 4 x 0.02 µL addition (Device iv). Scale bar is 1 µm.

Figure 4.9 Device resistance with respect to different SWCNT densities on the IDE area, with error bars as standard deviations of the mean. 137 Chapter 4

4.3.1.2.2 Direct CNT growth on IDE platform

An additional SWCNT chemiresistor fabrication methodology was trialled, which involved the direct growth of CNTs onto the surface of the IDE platform using C-CVD processes. The C-CVD growth conditions led to carbon materials deposited on the surface of the Si substrate (Fig. 4.10). It can seen that there is a slight correlation between the IDE patterning on the Si substrate and where the growth material is deposited, with the linear patterning in the upper right to centre, illustrated in Figure 4.10a. However, there is also a large excess of material deposited in the areas void of electrode patterning, which is unsatisfactory for the reproducible chemical sensor.

Figure 4.10 SEM image of a) the carbon material growth over an IDE area, and 2) a higher magnification image detailing portions of wire-like materials. Scale bars are 100 µm and 1 µm respectively.

The growth clusters displayed in Figure 4.10 were explored in more detail, and were found to consist of intertwined tube-like structures (Fig. 4.11a) and dendritic-branched structures (Fig. 4.11b). From the morphology observed on the IDE substrate it was clear from the dimensions of the tube-like structures, that they could not represent SWCNTs. In addition, the presence of large amounts of carbon impurities, more than likely amorphous carbon, was evident, which is highly undesirable for the developed chemiresistor sensors. Consequently, this direct IDE substrate growth strategy was not beneficial for SWCNT chemiresistor fabrication.

138 Chapter 4

Figure 4.11 A higher magnification SEM image of the wire-like structures generated on the IDE substrate.

4.3.1.3 SWCNT chemiresistor sensor optimisation

A relationship between drop-casting additions and device resistance has been established. In order to obtain the optimal performance out of the fabricated chemiresistors factors such as sensor baseline resistance, parameter sampling conditions and the sensor environment, were analysed for their effect on sensor performance.

4.3.1.3.1 Electrical monitoring parameters

The sensor monitoring configuration has been previously described in Section 3.5, but it is important to understand if variables in data sampling parameters would effect the sensor response, or alter the noise associated with the sensor data. In the current sensor monitoring setup the multi-pulse data acquisition function allows adjustment of sampling potential (V) and step width (s). Data were acquired for devices, upon exposure to saturated ethanol vapours, utilising different sampling potentials, as illustrated in Figure 4.12. These data indicate that as sampling potential was increased, from 1 mV to 100 mV, there was also an increase in sensor response to ethanol. It can be seen that a plateau in sensor response is developing towards increased sampling potentials. Furthermore, the reproducibility of sensor response is increasing at increased sampling potential, suggesting increased sensor stability. In view of these data, a sampling potential of 100 mV was considered optimal, in terms of both sensor response and reproducibility. In addition, the effect of sampling time was assessed and the results indicated no appreciable reduction in sensor noise with increased sampling time; as a result, an arbitrary sampling time of one second was utilised.

139 Chapter 4

Figure 4.12 Effect of sampling potential density on chemiresistor response to saturated vapour of ethanol, error bars as standard deviations of the mean.

4.3.1.3.2 Response dependence on baseline resistance

The effect of chemiresistor baseline resistance on sensor response required investigation. The chemiresistor devices, described in Section 4.3.1.2.1, and an additional device fabricated for a diluted SWCNT dispersed in DMF solution, were analysed using optimal parameters. The sensors were exposed to saturated vapours of ethanol and the device response was derived, as displayed in Figure 4.13. A positive linear correlation between device resistance and increased sensor response is observed, with data <2000 Ω relating to Devices i - iv mentioned previously. There is an increased sensor response associated with the high resistance device (17868 Ω), however, there is also higher response variation associated with devices at this baseline resistance, and some sensors produced sensor responses comparable to that of Device i (1685 Ω, 1 x 0.02 µL addition). Due to the increased reproducibility of sensor response of Device i coupled with the significant sensor response to the advantage of this sensor, the sensors utilised for the subsequent study were produced using the same drop-casting conditions to produce comparable baseline resistances.

140 Chapter 4

Figure 4.13 Effect of SWCNT density on chemiresistor response to saturated vapour of ethanol, error bars as standard deviations of the mean.

The observation that devices with increased baseline resistance produced an increased sensor response can be explained because in a system where the resistance is low there is an increased flow of electrons from the source to drain electrode. In turn as surface adsorbate molecules interact with the SWCNT matrix there are electron transfer events occurring; basically there will be an increase or decrease in the SWCNTs resistance, dependent on the analyte molecule. If there are a large number of electrons flowing through the system then the resistance will be low and the charge transfer of electrons between the SWCNT and analyte will only contribute a small percentage towards changing the sensor resistance. On the other hand, if a sensor has low I flow and high resistance, then these adsorbate-SWCNT interactions will contribute an increased percentage to the system and as a result produces an increased sensor response.

4.3.1.3.3 Sensor environment

The environment the SWCNT chemiresistor device experiences while electrical parameters are sampled may also have associated effects on sensor response. A

141 Chapter 4 chemiresistor sensor was monitored over a 42 h period when exposed to ambient atmosphere; with the sensor chamber lid removed. Over this period of time it was observed that the sensor’s baseline-current (baseline-I) mimicked the relative humidity change over the same time period, as illustrated in Figure 4.14. This is an interesting aspect as direct exposure to saturated water vapour was also trialled and no change in sensor baseline was observed. In relation to humidity and its effects on the following exposure trials, it should be realised that the UHP-N2 carrier gas will maintain a low relative humidity in the sensor testing chamber and the effects from this parameter will be minimised.

Figure 4.14 Chemiresistor sensor baseline-I in response to change in relative humidity.

Different carrier gas environments were tested for these chemiresistor devices, including UHP-N2, UHP-He, and UHP-Ar. No significant difference in the chemiresistor data was detected, and subsequently UHP-N2 was adopted for all vapour exposure studies.

142 Chapter 4

4.3.2 Metallic versus semiconducting SWCNTs in gas and vapour sensing

At the fundamental level gas and vapour sensing is intrinsically related to physical property changes experienced by the semiconducting SWCNTs in a particular device.[223, 472] Herein, the fabricated IDE platform was adapted to provide the basis for dielectrophoretic separation of metallic and semiconducting SWCNTs,[407] noting the Krupke et al. estimate that 80 ± 5% of the deposited SWCNTs, involving this dielectrophoretic procedure, are metallic in nature.[407] The device was exposed to DMMP, as a CWA simulant, in a typical analyte exposure experiment, as illustrated in Figure 4.15. The exposure response clearly indicates high levels of inherent noise associated with this device, and thus low electrical stability, with no direct correlation between analyte exposure events and changes in the electrical properties of the device. These results confirm that by its very nature, SWCNT gas and vapour sensing is highly dependent on the semiconducting SWCNTs, and their enrichment can only be beneficial in sensor development. A typical SWCNT synthesis yields approximately two thirds semiconducting nanotubes, and deposition of a nanotube suspension will be statistically dominated by the semiconducting SWCNT characteristics. This can be observed in the vapour-phase analyte exposure trials described in Section 4. 3.3. However, a previous study by Strano et al. suggests that some metallic nanotube pathways are effected by their chemical environment.[473]

Figure 4.15 DMMP exposure curve for the metallic SWCNT-enriched device involving dielectrophoretic separation. 143 Chapter 4

4.3.3 CWA analyte exposure trials

The fabricated SWCNT sensors were exposed to specific CWAs to monitor the effectiveness of these chemiresistors in detecting such compounds. Initially DMMP and DIMP were utilised to simulate the sensor response to the G-series nerve agents, sarin and soman, respectively, followed by phosgene. It should be noted that, unless otherwise stated the sensor response, ΔR / Ro, is expressed as a percentage, where Ro is the sensor baseline resistance.

4.3.3.1 Single interdigitated electrode sensor

4.3.3.1.1 DMMP

For the concentration range tested, exposure to DMMP using the single IDE sensor indicates a measurable response (Fig. 4. 16) which correlates directly to analyte exposure events. The positive response shows an increase in the resistance of the chemiresistor, which is indicative of an electron donating species interacting at the SWCNT surface. This correlates with the strong electron donor properties of DMMP, possibly due to the transfer of the partial negative charge on the terminal phosphonated oxygen.[277]

The sensor response curve establishes a response time of 500 ± 28 s, indicating the time for the sensor response to reach 90% of the maximum value for a set DMMP concentration, albeit with a slight baseline increase following each analyte exposure.

The system was purged with UHP-N2 for 30 min and it is clear that the recovery time is slightly greater than this time period, which is supported by the return to the original baseline value after the final DMMP concentration exposure, as displayed in Figure 4.16. These experiments are carried out at ambient conditions and recovery periods can be decreased through various methods, such as application of a vacuum and/or heat exposure.

144 Chapter 4

Figure 4.16 Single IDE sensor response curve to DMMP exposure of increasing concentration.

Figure 4.17 Single IDE sensor calibration curve for DMMP vapour exposure with increasing concentration, with error bars as standard deviations of the mean.

145 Chapter 4

From multiple DMMP exposure trials a calibration curve for the single IDE sensor was constructed (Fig. 4.17) with the data fitted with a weighted linear function, the associated correlation coefficient suggesting a strong positive relationship between sensor response and DMMP concentration. This data in conjunction with the calculated

RMSnoise of the sensor, are key parameters in determining the limits of detection for the fabricated SWCNT-chemiresistor sensors.

Based on the variance observed from a 5th order polynomial fitting function, the -3 RMSnoise for the single IDE was calculated at ±8.23 x 10 . Based on the standard limit of detection (LOD) definition outlined by IUPAC,[471] the LOD for these chemiresistor sensors towards DMMP is 1278 ppb, and a LOQ (signal-to-noise equal to 10) at 4259 ppb.

4.3.3.1.2 DIMP

Figure 4.18 Single IDE sensor response curve to DIMP exposure of increasing concentration.

The single IDE sensor response to increasing concentrations of DIMP was also investigated (Fig. 4.18). The response profile for DIMP has similarities to the DMMP profile (Fig. 4.17). The response time towards DIMP was 558 ± 15 s which is slightly

146 Chapter 4 larger than observed for DMMP. A recovery time of 1955 ± 757 s was observed with relatively long recovery times common for gas and vapour chemiresistors.[223] Various sensor refresh methodologies have been incorporated throughout the literature to reduce sensor recovery periods.[229, 277, 473, 474]

Figure 4.19 Single IDE sensor calibration curve for DIMP vapour exposure with increasing concentration, error bars as standard deviations of the mean.

A DIMP single IDE sensor calibration curve can be constructed from data obtained through multiple analyte exposure trials (Fig. 4.19). The data was fitted with a weighted linear fitting function and indicates a reasonable positive correlation for sensor response with change in DIMP concentration. The RMSnoise was calculated at ±4.82 x 10-3, and in conjunction with the calibration curve gradient these chemiresistor sensors have a LOD of 270 ppb, and a LOQ of 901 ppb.

147 Chapter 4

4.3.3.1.3 Phosgene

The single IDE sensors were exposed to increasing concentrations of phosgene generated from a phosgene saturated toluene solution (Fig. 4.20), establishing a measurable sensor response. A negative sensor response change is observed, indicating a reduction in the overall resistance of the sensor upon exposure to phosgene. This decrease sensor resistance suggests chemical modification of the surface of the SWCNTs, compounded by the irreversible sensor response under ambient testing conditions, with extensive UHP-N2 chamber purging periods exhibiting no effect. From the response data it was established that as phosgene concentration is increased a maximum response change is observed, as illustrated in Figure 4.21.

Figure 4.20 Single IDE sensor response curves for phosgene exposure of increasing dilution factors from a saturated vapour.

There is a negative linear trend between sensor response and phosgene concentration over the analyte range studied, with a sensor saturation point evident (Figs. 4.20 and 4.21). Exposure to concentrations above 46700 ppm leads to a plateau response indicative of sensor saturation. This indicates that the majority of available nanotube surface defect sites are occupied with adsorbate and/or the maximum chemical

148 Chapter 4 modification of the surface has been reached. Response times for these sensors were concentration dependent (Fig. 4.20) and the sensor baseline resistance can be restored by heating in vacuo.

Figure 4.21 Single IDE sensor maximum response curve data for phosgene vapour exposure with increasing concentration, error bars as standard deviations of the mean.

The concentration range used for the analysis of sensor response against phosgene concentration was 18,450 ppm to 46,700 ppm. This concentration range was limited by the current dynamic vapour delivery system configuration, and the concentrations used were the lowest phosgene values that could be effectively generated. A weighted linear fitting function was used to represent the experimental data (Fig. 4.22) with the data indicating a strong negative linear correlation between the maximum sensor response and increasing phosgene concentration. The RMSnoise of these single IDE sensors was calculated at ±7.01 x 10-3, which gives rise to a sensor detection limit of 23.3 ppm and a LOQ of 77.7 ppm. It should be noted that these experiments were conducted under ambient conditions and sensor refresh methodologies previously alluded to could be adopted to access the original sensor baseline resistance.

149 Chapter 4

Figure 4.22 Single IDE sensor calibration curve for phosgene vapour exposure with increasing concentration, error bars as standard deviations of the mean. Higher concentration response was not possible due to sensor response saturation.

4.3.3.1.4 Phosgene response based on current gradient change

The phosgene response curves (Fig. 4.20) indicate that when exposed to increasing concentrations of phosgene, the gradient of the response change also increases. From an analysis of the most linear sections of the response curve, gradient values were extracted and plotted (Fig. 4.23). These gradient values are represented as the change in I (μA s-1), and due to the negative sensor response caused by the phosgene interaction mechanism, positive gradient values are observed.

The response curve gradient data displays an initial gradual increase as concentration increases, followed by a high level of significant growth and finally a plateau as the sensor response reaches concentrations indicative of sensor saturation (Fig. 4.23). The data was fitted with a Boltzmann sigmoidal fitting curve (Fig. 4.24) and clearly represents the gradient data with a high correlation coefficient. This methodology could

150 Chapter 4 be successfully applied in a ‘real-life’ operational setting, to confirm the data obtained using the previously detailed methodology based on sensor response.

Figure 4.23 Single IDE sensor baseline gradient change for phosgene vapour exposure with increasing concentration.

Figure 4.24 Single IDE sensor calibration curve for baseline-I gradient change upon phosgene vapour exposure with increasing concentration. The data is fitted with a growth curve fitting function to represent the gradient trend. 151 Chapter 4

4.3.3.2 Multiple-array sensors

4.3.3.2.1 DMMP

CWA exposure trials were undertaken on multiple-arrays of chemiresistor sensors to investigate how this effects sensor performance and stability. The multiple-array sensor response profile to DMMP exposure (Fig. 4.25) follows closely to the response profile observed for a single IDE sensor (Fig. 4.16). The directionality and response times are highly comparable, indicating a continued relationship between DMMP concentration and sensor response. Here a sloping sensor recovery baseline is observed which indicates that the recovery time under ambient testing conditions is not sufficient for the total desorption of the analyte.

Figure 4.25 Multiple-array sensor response curve to DMMP exposure of increasing concentration.

The identical DMMP concentration range was tested for both the single IDE sensors and the multiple-array sensors to achieve a direct comparison. From the multiple-array data a DMMP calibration curve was constructed and a weighted linear fitting function applied (Fig. 4.26). The fitted function indicates a moderate linear correlation between sensor response and DMMP concentration. It should be noted that an increased

152 Chapter 4 correlation coefficient is obtained by the single IDE sensor. The RMSnoise was calculated from the background data and was found to be at a level of ±3.33 x 10-3. It is interesting to note that the RMSnoise level of the multiple-array sensor is lower than that encountered from the single IDE sensors, this is an advantage as decreased noise levels directly lower sensor limits of detection. This investigation also revealed that there was an increase in sensor response for the concentration range analysed, when compared to the single IDE sensors. This increase in sensor response leads to an increased calibration curve gradient, which in turn will produce lower detection limits. The calculated LOD for the multiple-array sensors to DMMP is 266 ppb and the LOQ is 887 ppb.

Figure 4.26 Multiple-array sensor calibration curve for DMMP vapour exposure with increasing concentration, error bars as standard deviations of the mean.

4.3.3.2.2 DIMP

The multiple-array sensors were also exposed to DIMP at the identical concentration range as the single IDE sensors, and they perform in a similar fashion to the single IDE

153 Chapter 4 sensors (Fig. 4.27), with response and recovery times highly comparable for the two sensor geometries. One standout difference between the two systems is the increased sensor response for the multiple-array sensor at the same concentrations analysed for the single IDE sensor. This trend was also observed for DMMP exposure and indicates that using the multiple-array sensor geometry increases the performance of the SWCNT chemiresistor.

Figure 4.27 Multiple-array sensor response curve to DIMP exposure of increasing concentration.

Based on the response curves for the multiple-array sensors, the sensor response with increasing DIMP exposure can be plotted (Fig. 4.28). A weighted linear fitting function was applied to the data which reveals a moderate positive linear correlation. To assess the LOD for DIMP using this sensor geometry the RMSnoise was calculated to be -3 ±3.25 x 10 . Based on the RMSnoise and the calibration curve gradient, the DIMP LOD is 152 ppb and the LOQ is 508 ppb for the fabricated multiple-array SWCNT chemiresistor sensors.

154 Chapter 4

Figure 4.28 Multiple-array sensor calibration curve for DIMP vapour exposure with increasing concentration, error bars as standard deviations of the mean. Note: the sensor response for the lowest concentration analysed was considered an outlier and not represented in the plot.

Multiple-array sensor studies were also undertaken for phosgene exposure experiments and it was observed that the multiple-array sensors were unpredictable and produced irreproducible sensor response changes.

4.3.3.3 Single versus multiple-array

The performance of the single IDE geometry platform was directly compared to the multiple-array geometry sensors. The comparison of the calibration curve data for DMMP exposure indicates that there were some subtle differences between the systems (Figs. 4.17 and 4.26). Firstly, an increased sensitivity (slope) is observed for the multiple-array sensor geometry, as opposed to the single IDE platform. Secondly, the baseline response RMSnoise is reduced by approximately half with the multiple-array sensor geometry. A possible explanation for the device behaviour, can be related to the multiple-array sensor geometry consisting of four separate IDEs, which have been connected in parallel. This will in turn draw increasing current levels from the

155 Chapter 4 potentiostat, resulting in increased current readings, and subsequently lower sensor baseline resistance. Previously, it has been demonstrated that the SWCNT chemiresistor sensors exhibit increased sensor response at high baseline resistance, so the opposite effect should be observed, and the single IDE sensors should have increased sensor response. However, this is not the case, and this can possibly be explained by each individual IDE in the multiple array operating at the optimal baseline resistance, but when combined they draw increased current. This inherently increases the measureable sensor current to levels where high signal to noise ratios are experienced, thus leading to increased sensor response and decreased sensor noise when compared to the single IDE SWCNT chemiresistor configuration.

4.3.3.4 Defect-induced versus untreated SWCNTs

The effect of the defect-inducing SWCNT pre-treatment was investigated against a purification procedure which deliberately reduces the possibility for nanotube defects. The defects were induced through an acid reflux, which not only removes metal catalyst impurities but also leads to the formation of carboxylic acid and hydroxyl functional groups at the SWCNT tip and along the side-walls.[475, 476] The untreated SWCNTs were subject to the same procedure as the acid functionalised SWCNTs; however, the acid reflux step was excluded. The sensor consisting of the untreated SWCNTs was exposed to the highest concentration of phosgene under identical test conditions as the acid functionalised counterpart. It was found that the acid functionalised SWCNTs sensor response, -42.44%, was nearly double the untreated SWCNTs, -22.13%. This indicates that the acid functionalisation pre-treatment plays an important role by inducing defect sites on the SWCNT surface and this supports recently published reports in the literature.[104] Thus the defects site are important in providing low-energy adsorption sites on the SWCNT surface, which subsequently can serve as nucleation sites for the additional condensation of the analyte vapour.[104]

4.3.3.5 Interference plots

The fabricated sensors were exposed to a selection of common organic solvents including halogenated compounds which were selected to provide adequate structural similarities to phosgene to assess the possibilities of false-positives due to interference vapours. For the common organic solvents (acetone, methanol, ethanol, benzene, and 156 Chapter 4 toluene) a positive sensor response is observed (Figs. 4.29 and 4.30), with associated concentrations listed in Table 4.1. In relation to possible phosgene interferences it can be clearly seen that only three of the compounds trialled, excluding phosgene, exhibit a negative sensor response change. These compounds are chloroform, dichloromethane, and benzyl chloride and their response polarity is indicative of an electron withdrawing SWCNT interaction and/or covalent interaction. It is interesting to note that other chlorinated compounds, such as chlorobenzene and carbon tetrachloride, induce a small positive sensor response, with 1,2-dichlorobenzene having no effect on the sensor response. This is to be expected due to the unreactive nature of aryl halides with their delocalised π-bond system strengthening the carbon-halogen bond, and the reduction in the polarity of the carbon-halogen bond. It should also be noted that toluene gives a slight positive sensor response. This merits comment as this is the solution matrix from which the phosgene utilised in this study was generated, and has the ability to slightly impede the full sensor response caused by phosgene.

Figure 4.29 Sensor response of the single IDE sensor geometry to common organic solvents and a range of halogenated compounds. Each analyte was delivered at 10% saturated vapour concentration.

157 Chapter 4

Figure 4.30 Sensor response of the single IDE sensor geometry to common organic solvents and a range of halogenated compounds. Each analyte was delivered at 10% saturated vapour concentration.

Benzyl chloride has a dramatic effect on the sensor response, and the sensors illustrate an even greater sensitivity towards this compound than to phosgene (Fig. 4.29). This is possibly due to the high reactivity with the surface functional groups of the SWCNT and/or the direct reaction with the surface of SWCNT in general. As with phosgene the observed sensor response for benzyl chloride is irreversible under ambient test conditions, suggesting that in both cases covalent chemistry is prevalent which correlates with the high reactivity of benzylic halides. Furthermore, the sensor interference demonstrated by benzyl chloride can be utilised to the advantage of these sensors, as this is a lachrymator and is classified as a choking gas, which has been utilised in chemical warfare.

The possible interference associated with the detection of DMMP and DIMP is more clearly represented in Figure 4.30. From the selected interference analytes studied, it is observed that many of the analytes tested produced a positive sensor responses, with the same signal polarity as the G-series nerve agent simulants DMMP and DIMP, which

158 Chapter 4 means that there is an increased potential for false-positives from common solvents, such as acetone, methanol, and ethanol. It is interesting to note that the sensors have an increased sensitivity to the nerve agent chemical simulants, especially DMMP, as the sensor response for this simulant at a relatively low concentration (127 ppm) is comparable to a high interference concentration (30526 ppm) for acetone. This coincides with experimental data from SWCNT sensors in other studies.[235, 278]

Table 4.1 Calculated concentration values for the analytes included in the interference study.

Compound Concentration at Compound Concentration at 10% 10% of saturated of saturated vapour vapour (ppm) (ppm)

Acetone 30526.3 1,2-Dichlorobenzene 179.0

Methanol 16710.5 Dichloromethane 57236.8

Ethanol 7802.6 Carbon tetrachloride 15131.6

Benzene 12473.7 Benzyl bromide 59.2

Toluene 3736.8 DMMP 126.6

Benzyl alcohol 12.4 DIMP 36.5

Chloroform 25921.1 Benzyl chloride 161.8

Chlorobenzene 1578.9 Phosgene 37368.4

*Note: HCl (2.4%) in MilliQ water was excluded, this was tested as a possible impurity in the phosgene (20%) in toluene solution.

4.3.3.6 Implications in relation to current exposure limit guidelines

The LOD of these sensors requires lowering, especially as phosgene has an odour threshold of 0.4 ppm.[450] There are varying reported values in the literature for the [446, 477, 478] LCt50 of phosgene, and these values range from 500 ppm to 800 ppm. The LOD, 23.3 ppm, and LOQ, 77.7 ppm, observed for the fabricated chemiresistor sensors

159 Chapter 4 are well below these levels. However, current guidelines for workplace exposure limits put in place by The Great Britain Health and Safety Executive suggest a long-term exposure limit (LTEL) of 0.02 ppm.[479] Currently, these sensors could be utilised to monitor for deliberate phosgene release events, such as terrorist attacks, to warn potential victims and to produce a chemical mapping the contaminated area. These would not be suitable for personal monitors or industrial leak detectors at the sensor’s current LOD.

In relation to the G-series nerve agents sarin and soman the focus is on a direct terrorist threat. The human exposure guidelines for the airborne concentration sufficient to induce severe effects in 50% of those exposed for 30 min, EC50, for sarin and soman is 107 ppb and 140 ppb respectively.[480] The fabricated chemiresistor sensors LOD for

DMMP (266 ppb), and DIMP (152 ppb) is close to the EC50 values. However, reduction in the sensor’s LOD is required before these sensors could be routinely utilised. It should be noted that in relation to these G-series nerve agents these SWCNT chemiresistor sensors could be utilised for monitoring the large release of these nerve agents in public areas.

The RMSnoise levels calculated for these sensors is relatively high when compared to similar published systems.[229] Despite all possible efforts to lower the noise of these sensors, they were still an order of magnitude higher than comparable systems in the literature. This, in combination with the limitations of the current vapour delivery system in relation to the minimum analyte concentrations available, is a future focus area. Lowering both the sensor RMSnoise and the minimum deliverable concentration, has the potential to further lower the detection limits of these sensors by at least an order of magnitude. A future prospect to lower inherent noise in the system is to investigate other carbon nano-materials, more specifically graphene, as this material has been shown to lead to decreased noise levels in devices.[481]

4.4 Conclusion

A SWCNT chemiresistor sensor based on defect-induced nanotubes, debundled through ultrasonication and centrifugation processes in DMF has been developed. The versatility of the principal IDE platform has been demonstrated through the

160 Chapter 4 dielectrophoretic separation of CNTs, to produce a metallic nanotube-enriched SWCNT chemiresistor device. This device was utilised to illustrate the importance of semiconducting SWCNTs in the mechanism of analyte detection. A comparative study was detailed for two different sensor configurations; single IDE and multiple-array. Phosgene and G-series nerve agent simulants were utilised to assess the sensing characteristic of these sensor geometries. Interestingly, these data suggest an increased sensor response and lower RMSnoise with the multiple-array geometry.

These sensing capabilities represent the first report on a carbon nanotube-based gas and vapour sensor that can detect the CWA phosgene. Sensor response curves and calibration curves have been presented with a limit of detection of 23.3 ppm. In addition, two data manipulations were presented for the generation of a calibration curve and could be utilised simultaneously to nullify any false-positives or baseline aberrations. In relation to the application in ‘real world’ setting, interference studies were undertaken and the sensor detection limits obtained were directly comparable to current chemical exposure guidelines.

These fabricated SWCNT chemiresistor sensors have direct implication in the area of counter-terrorism, where phosgene is the focus of recent attention, due to its uncomplicated synthesis with common-place chemicals, large commercial production and ease of release in a terrorist operation. These results also have implications in industrial monitoring of on-site production plants and incorporation into smart-materials for first hazard responders. These devices would also have ‘battle-field’ applications for the monitoring of CWA release, in particular phosgene, and could be envisaged to be self-powered by the incorporation of a functional solar-powered cell, for remote deployment of the sensors, including bushfire monitoring in Australia.

161

Chapter 5 Progress Towards Novel Carbon Nano-Materials and Interesting Application Areas

Chapter 5

5 Progress towards novel carbon nano-materials and interesting application areas

5.1 Introduction

Carbon-based materials have a wide variety of applications such as catalysis,[482-485] fuel cells,[486-488] separations,[489-492] extractions and sensor technology.[223, 235, 493] Overall, carbon is remarkably versatile and offers the possibility of fabricating a diverse range of novel products entirely composed of the element. The discovery of fullerene C60 and higher fullerenes, and the later development of CNTs, has added significantly to this possibility becoming a reality.

In developing the full potential of carbon it is important to be able to accurately prepare new structural carbon frameworks, and to be able to generate new materials with different spatial arrangement of arrays of carbon atoms. One way of achieving this is through self-assembly processes (Section 1.5.2). Fullerenes, as the only form of molecular carbon, are in a unique position to be utilised for these processes. A recent example of this is the formation of a one-dimensional C60 array, in solution, templated by helical nanotubes.[494]

An area that has attracted significant attention in recent years is the supramolecular chemistry of fullerenes involving cavitands, with particular interest being devoted to calixarenes and related molecules. Atwood et al.[354] and Suzuki et al.[366] initially developed a convenient and efficient purification method for obtaining C60 from ‘crude’ fullerene soot, which is based on the formation of a discrete complex with p-tBu-calix[8]arene. Furthermore, the different effects of lower phenolic oligomers, and substituted lower and upper ring functionalisation, has led to unique complex arrays of fullerenes that display fullerene···fullerene interplay.[359, 381, 382] Examples of such [495] interplay include the 1:1 complexes between C60 and cyclotriveratrylene (CTV), and [358] calix[5]arene, where the C60 molecules are packed into continuous zigzag chains.

C60···C60 close contacts vary from 9.90 Å to 10.20 Å, with dihedral angles in the chains ranging from 118° to 172°, indicating a transitional packing structure between a linear chain and double column array.[359] More recently, these systems have been utilised in

163 Chapter 5 engineering crystal virtual porosity,[371, 428] an important area recently highlighted.[496] For a more detailed overview of fullerene–calixarene interactions the reader is directed to Section 1.7.

The formation and characterisation of a novel fullerene C60-calix[5]arene self-assembled complex will be described. It is observed that ‘crude’ fullerene soot similarly produces nano-fibres, albeit with a small percentage of the next highest fullerene, C70, incorporated into the structure. Processes have been developed to allow the selective volatilisation of the calixarene component, to fabricate unique all-carbon architectures based on the C60 organisation observed in the parent complex. In addition, the properties of these analogues are detailed. Furthermore, a unique application area is explored with promising results in the high-impact area of chemotherapeutics.

In addition, to the described fullerene supramolecular arrangement, other all-carbon architectural systems were synthesised and studied. These include the post-growth formation of SWCNT ropes which self-organise into helical rings, with no visible seam, and the formation of novel, square-geometry carbon nano-fibres (CNFs). Investigations into these fabricated all-carbon materials will be subsequently detailed.

5.2 Experimental

5.2.1 Materials

5.2.1.1 C60 nano-fibres and high temperature annealed analogues

Fullerene C60 (99%, BuckyUSA, TX, USA) and Fullerite (assayed fullerene content = 11.7%, batch # 3023, MER Corporation, AZ, USA) powders were obtained and utilised without further purification. The p-tBu-calix[5]arene (5) utilised in this study was synthesised using a modified literature preparation.[305] High temperature argon annealed analogues were prepared from the parent complex, by processes [497] described in Section 5.2.2. The nano-C60 preparation is detailed in the literature. Human cervical epithelioid carcinoma cells (HeLa) and doxorubicin (Dox)-resistant human breast carcinoma cells (MCF-7) were obtained from American Type Culture Collection. Green fluorescent protein (GFP) (Santa Cruz Biotechnology, CA, USA),

164 Chapter 5 protein 1 light chain 3 (LC-3) (Novus, CO, USA), Lipofectamine 2000 (Invitrogen, CA, USA), Dox (Sigma-Aldrich), and antibiotic G418 (Promega, WI, USA) were obtained. All other cell culture reagents or assays were obtained from Gibco (CA, USA) and used without further purification, unless noted.

5.2.1.2 Rings of helical SWCNT bundles

Rings of helical SWCNT bundles were observed in post-processed samples from a commercial SWCNT source (Cheap Tubes, Inc., VT, USA). Fullerene C60 (99%,

BuckyUSA) powder was utilised without further purification. Nitric acid (HNO3) (70%) (analytical reagent) was obtained from Sigma-Aldrich, and purified using a quartz sub-boiling acid distillation apparatus.

5.2.1.3 Square-geometry carbon nano-fibres

Square-geometry carbon nano-fibres were observed in materials produced using C-CVD processes. Acetylene feedstock and UHP-He were obtained from BOC Gases. Metal catalyst particles were introduced to using nickel (II) sulfate hexahydrate (NiSO4·6H2O) (Sigma-Aldrich) from which stock solutions were made (1 M). All other reagents were of analytical grade and were obtained from Sigma-Aldrich and used without further purification. All carbon-material synthesis was undertaken in the laboratories of Dr. Gerard Poinern (Curtin University of Technology, WA, Australia).

5.2.2 Methods

5.2.2.1 C60 nano-fibres and high temperature annealed analogues

Synthesis of (C60)∩(p-tBu-calix[5]arene) complex (6): On dissolution of C60 (40.7 mg, 5.65 x 10-2 mmol) in toluene (40 mL), 13 molar equivalent of 5 (596.3 mg, 0.7351 mmol) is added. Gravity filtration, followed by a wash with hexane (3 x 2 mL), produces a light brown/golden solid (48.7 mg, 3.18 x 10-2 mmol), which was dried in vacuo (56% yield, elemental analysis (%) calculated: C 90.17, H 5.22; found: C 90.14, H 5.28). The complex formed using fullerene soot was derived using this synthesis methodology, with the exception of the fullerene soot/toluene solution being filtered

165 Chapter 5 prior to use. The dethreading of the complex was achieved by dispersion in dichloromethane.

Synthesis of high temperature argon annealed analogues: The parent complex, 6, was heat-treated in an argon atmosphere, with products collected from isothermal hold experiments at 50 °C intervals, between a 400 °C to 800 °C range. These analogues were characterised without further purification.

Characterisation methodologies: UV-visible spectrophotometry was performed with a Cary 50 Tablet spectrophotometer (Varian). The spectral range covered was 800 nm -1 -2 to 200 nm, with a scan rate of 600 nm min . A C60 stock solution (9.85 x 10 mmol) of

C60 in toluene was prepared. Solutions of 1.1, 2.1, 2.8, 5.3 and 11.2 molar equivalents of 5 were added to aliquots (10 mL) of the C60 stock solution. Toluene solutions of 6 (1 mg mL-1), and the fullerite complex (1 mg mL-1) were individually produced. All solutions were analysed in quartz cuvettes with a 10 mm path length at 25 °C. Fourier-transform infrared (FT-IR) spectroscopy was performed with a Spectrum One FT-IR spectrometer (PerkinElmer, MA, USA). The spectral range covered was 450 cm- 1 to 4000 cm-1, with a 1.00 cm-1 resolution; samples were mixed with potassium bromide (KBr) and pressed into discs.

TEM was performed using a 3000F TEM (JEOL, Tokyo, Japan) operating at 200 keV; sample preparation involved forming a liquid suspension of the complex in hexane, from which a drop of the solution was placed onto a holey carbon film supported by copper grids and air-dried. SEM was performed using a 1555 VP field emission scanning electron microscope (Zeiss) operating between 2 keV to 5 keV using the in-lens detector. Sample preparation involved the dispersion of material directly onto carbon tape secured on a standard aluminium stub.

Cross-polarisation magic angle spinning (CP-MAS) solid-state C13 NMR spectra were measured on a ARX 300MHz NMR spectrometer (Bruker, MA, USA) at a frequency of 75.468 MHz, a spin rate of 4 kHz, contact time of two milliseconds and relaxation delay of two seconds. Spectra were recorded with the magic angle spinning in an on- and

166 Chapter 5 off-state. Powder XRD (PXRD) patterns were acquired using a D500 Diffractometer

(Siemens, Munich, Germany) utilising CuKα radiation (λ = 0.154 nm); sample preparation involved placing the material (~50 mg) on a XRD sample holder, levelled with a glass microscope slide. Sample orientation effects were assessed through diffraction patterns obtained at different sample holder rotation angles.

Thermogravimetric analysis (TGA) was performed using a SDT 2960 simultaneous DSC-TGA (TA Instruments, DE, USA) in either air or argon (100 mL min-1) with a temperature program of 3 °C min-1 for varied temperature ranges. For isothermal hold experiments a temperature program of 20 °C min-1 under argon (100 mL min-1), with a 20 min hold period at the programmed final temperature. Brunauer, Emmett and Teller (BET) surface area measurements were acquired on a Tristar 3000 (Micromeritics, GA,

USA), by the adsorption of N2 at 77 K; the sample was loaded in a quartz tube evacuated to 0.1 millibar and cooled with liquid nitrogen to 77 K. Nitrogen gas was incrementally released into the quartz tube for absorption measurements at different vapour pressures. I-V parameters and electrical resistance measurements were carried out on the samples using a micromanipulator coupled with a precision semiconductor parameter analyser (4156A, Agilent Technologies). Dispersive Raman spectroscopy was performed using a 1B micro-Raman spectrometer (Dilor Labram, France). The spectral range covered was 127.38 cm-1 to 3063.17 cm-1 using a grating of 600 lines mm-1; samples were mounted and analysed on glass microscope slides.

Molecular modelling: The production of possible C60 structural arrangement within the parent complex was achieved through molecular modelling (Accelrys Software, Inc.

CA, USA). Based on existing structural information a variety of C60 packing arrangements were trialled. Structural refinement was achieved through energy minimisation calculations, and a combination of fullerene···fullerene centroid distances and dihedral angle measurements. The structures probed included, but were not limited to, entire fullerene encapsulation, single, double, triple, quadruple columnar arrays, and single and multi-strand helical arrays. In combination with inferred physical constraints from characterisation techniques a suitable model was established.

167 Chapter 5

Chemotherapeutic assessment: HeLa cells were plated in 35 mm glass-bottomed dishes and grown continuously as a monolayer in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal calf serum (FCS) and were incubated in a humidified atmosphere of 95% air and 5% CO2 at 37 °C for 24 h. Cells were transfected with green fluorescent protein with protein 1 light chain 3 (GFP-LC3) (1 µg) using Lipofectamine 2000, according to the manufacturer’s protocol and were incubated for another 24 h for the expression of GFP-LC3 fusion protein. After 24 h these transfected cells were transferred to a new plate and underwent selection in DMEM medium containing G418 (0.6 mg mL-1, Promega, WI, USA), with change of medium every three days. Cell colonies exhibiting strong green fluorescence were selected through fluorescent microscope examination 10 days post-transfection and were expanded. These cell cultures were treated individually with aqueous solutions of 6 -1 -1 (2.5 µg mL ) or nano-C60 (0.5 µg mL ) and were incubated for a further 24 h. After incubation at 37 °C for 24 h, the medium was removed and the cells were washed three times with phosphate buffered saline (PBS). The localisation of LC3 in transfected cells was examined by fluorescence microscopy at λem = 530 nm (λex = 485 nm) (Olympus IX71, Tokyo, Japan).

The Dox-resistant MCF-7 cell line was derived from the drug sensitive MCF-7 cells by stepwise selection with Dox, with an initial selection in Dox (0.01 µM) for two months, followed by selection in Dox (0.1 µM) for two months, and finally in Dox (1 µM) for three months. The established cell line was verified for drug resistance by MTT and apoptotic assays, and was maintained in DMEM medium containing Dox (1 µg ml-1) and incubated in a humidified atmosphere of 95% air and 5% CO2 at 37 °C. These cell cultures were either left untreated (control) or treated individually with aqueous -1 -1 solutions of Dox (10 µg mL ), or straight 6 (0.5 µg mL ), or straight nano-C60 -1 -1 -1 (0.5 µg mL ), or 6 (0.5 µg mL ) in combination with Dox (10 µg mL ), or nano-C60 (0.5 µg mL-1) in combination with Dox (10 µg mL-1), and were incubated for a further 24 h. Nuclei were detected by fluorescence microscopy using an apoptosis assay; Hoechst/PI assay. Cells were stained with Hoechst 33342 (10 mM) and PI (10 mM) for 10 min and then examined under fluorescence microscopy. Cell death was quantified by counting 500 cells and the results were expressed by the ratio of PI positive cells to total cells (Hoechst positive cells). All cellular studies were undertaken by Professor

168 Chapter 5

Long-Ping Wen (University of Science and Technology of China, Anhui Province, China).

5.2.2.2 Rings of helical SWCNT bundles

Synthesis of rings of helical SWCNT bundles: The commercial SWCNTs were purified using a modified literature procedure.[464] SWCNTs (0.0987 g) were refluxed at 130 °C with re-distilled 70% HNO3 (25 mL) for 8 h. The resulting solution was cooled to room temperature and rendered neutral through centrifugal decanting with MilliQ water (>18 MΩ cm-1), followed by freeze-drying to afford a product in 44% yield. This product was heated in humidified air at 420 °C, to further promote SWCNT tip-opening. Following this the resulting product (5 mg) and C60 powder (5 mg) were sealed in an evacuated quartz tube (4 x 10-6 Torr). The quartz tube was annealed at

400 °C for 48 h. Note the addition of C60 and subsequent annealing was to produce ‘peapod’ structures, however, this process did not occur and TEM analysis revealed ring of helical SWCNT bundles were present in the sample.

Characterisation methodology: TEM was performed using a 2100 TEM (JEOL, Tokyo, Japan) operating at 120 keV, with carbon and thickness maps produces through energy-filtering processes; sample preparation involved the dispersion of the vacuum annealed material into distilled ethanol, dried over molecular sieves, from which a drop of the solution was placed onto a holey carbon film supported by copper grids and air dried. Ultrasonication was not utilised as a dispersion process.

5.2.2.3 Square-geometry carbon nano-fibres

Synthesis of vapour-grown carbon fibres (VGCFs): The methodology utilised for VGCFs is consistent with that described in Section 4.2.2.

Characterisation methodology: SEM was performed using a 1555 VP field emission scanning electron microscope (Zeiss, Carl-Zeiss, Inc.) operating between 2 keV to 5 keV using the in-lens detector. Sample preparation involved the dispersion of the as-grown material directly onto carbon tape on a standard aluminium stub.

169 Chapter 5

5.3 Results and discussion

5.3.1 Nano-fibres of fullerene C60

5.3.1.1 Supramolecular complex formation and characterisation

The synthesis of the 6 complex requires a large excess of calixarene, and is formed in

56% yield based on C60. After two hours the typical magenta colour of the C60 solution turns dark brown, indicating complex formation. At this point optical microscopy reveals a suspension of light brown needle-like, fibrous material (Fig. 5.1a). To ensure a uniform product, continuous stirring is essential. The mother liquor can be recycled with the addition of an equivalent of C60 and 5, thereafter affording a quantitative yield of the complex. A variety of different synthetic and crystallisation conditions were investigated, such as different host-guest molar ratios, host-guest reaction periods, rates of solvent evaporation, solventless reactions and the use of ultrasonication. Indeed, grinding an equimolar mixture of C60 and 5 followed by sonication in hexane produced the same fibrous product, albeit in low purity with un-reacted starting material present.

Figure 5.1 Optical micrograph a), and TEM image b) of the 6 complex, the latter revealing the

linear striations of fullerene C60 within bundles of nano-fibres.

TEM analysis revealed ordered linear striations throughout the entire structure of the

C60 nano-fibres, these striations correlate to columnar arrays of fullerenes (Fig. 5.1b). Bundles of these nano-fibres readily form from the crystallisation process. It is possible that there is a packing arrangement of the calixarenes that is providing sufficient C60 column separation, leading to the visual effect observed. Bowl-shaped calix[5]arenes

170 Chapter 5 have been shown to be effective in controlling the assembly of fullerenes (Section 1.7.2), with dimensionality and curvature complementarity for endo-cavity binding of [377] C60. Drawing conclusions from the observed TEM results, and from compositional microanalysis (Section 5.2.1), it is suggested that 5 forms a complex with C60 from toluene, as nano-fibres based on the 1:1 supermolecule 6, as illustrated in Figure 5.2.

Figure 5.2 A schematic representation of the formation of the 1:1 supermolecule, 6.

UV-visible spectrophotometry studies were undertaken to assess the solution interaction between 5 and C60 in toluene (Fig. 5.3). With increasing concentrations of the calixarene, there is a significant spectroscopic change observed at 438 nm, indicating complex formation. This supports studies previously undertaken by Shinkai et al., which suggest that a primary prerequisite for C60 inclusion is the ability for intramolecular hydrogen-bonding of lower rim hydroxyl functionalities, to lock the calixarene in a cone conformation.[498] It should also be noted that sufficient intermolecular contact between the aromatic rings of the calixarene and the C60 molecule is essential for C60 inclusion, resulting in the necessary charge-transfer π···π interaction.

171 Chapter 5

-2 Figure 5.3 UV-visible spectra for a C60 (9.85 x 10 mmol) stock solution with increasing concentrations of 5, spectra recorded in toluene at 25 °C.

Using a mixture of fullerenes (fullerite) in place of C60 also afforded a nano-fibre complex, with some C70 incorporated (Fig. 5.5), whereas reacting pure C70 with 5 did not form a solid complex. The structure of the complex derived from fullerite is presumably the same as the structure for the complex formed using pure C60 (Fig. 5.9), with the upper limit of C70 inclusion governed by the geometrical mismatch associated with incorporating a non-spheroidal fullerene into the continuous array. The 6 complex undergoes rapid decomposition in dichloromethane, with the precipitation of pure C60

(Fig. 5.4). The analogous (C60)∩(p-tBu-calix[8]arene) complex similarly degrades in this chlorinated solvent.[354] It should also be noted that PXRD analysis of the material formed on decomposition indicates a hcp crystal structure (Fig. A.2), which involves a dethreading process and has been recently been reported with the analogous p-benzylcalix[6]arene system.[371]

172 Chapter 5

Figure 5.4 IR spectrum of the product of the decomplexation of 6 in dichloromethane, additional absorbance peaks are attributed to residual dichloromethane.[354]

Comparative optical absorbance studies were conducted for both the complex derived from pure C60 and from the crude fullerene soot filtrate, against a C60 standard solution (Fig. 5.5). The complex derived from the fullerene soot has the characteristic absorption bands of C60, demonstrating that only a small percentage of C70 is integrated within the complex, with an absorbance peak seen at 475 nm which is a typical [499] absorbance peak observed in the C70 UV-visible absorbance spectrum. It should be noted that C70 absorbs strongly in this wavelength range, and as a result a much lower concentration of C70 will produce peak heights comparable to that of a solution containing a much higher concentration of C60.

173 Chapter 5

Figure 5.5 UV-visible spectra for i) a C60 standard, ii) nano-fibre complex from pure C60, and iii) nano-fibre complex from fullerene soot filtrate, all spectra recorded in toluene at 25 °C.

Solid-state NMR spectroscopy of the complex clearly shows freely rotating C60 at ambient temperature, and also rotation of the tertiary butyl groups (Fig. 5.6).[500] This provides supporting evidence of the presence of both C60 and 5 in the solid-state structure.

Figure 5.6 CP-MAS solid state 13C NMR spectra for the 6 complex, with (blue) and without (red) magic angle spinning. 174 Chapter 5

Calix[5]arenes have complementarity of curvature with C60 and form endo-cavity complexes, either as a “ball-and-socket” nanostructure, or with two calixarenes shrouding a single fullerene molecule “ball-and-dual socket”.[358, 362] This, coupled with the ratio of 1:1 for the two components in the present complex, is consistent with the supermolecule unit as a “ball-and-socket” nano-structure (Fig. 5.2). The supermolecules then propagate to form one-dimensional arrays, attributed to columnar arrays of C60, noting that such an array is encountered in the unmodified calix[5]arene system.[358] Bundles of these arrays agglomerate to form the macroscopic fibrous product. In comparison to the unmodified calix[5]arene system, the present system has additional geometrical packing constraints, due to the bulky tertiary butyl groups on the upper rim of the calixarene, but nevertheless fullerene···fullerene interactions are still possible.

PXRD was utilised in an attempt to gain insight into the possible solid-state structure of the 6 complex. The PXRD pattern is dominated by the 2θ reflection at 3.93° (Fig. 5.7), which is attributed to the inherent distance between individual C60 columns, at a distance of 2.25 nm. The fibrous nature of the sample gives rise to preferred diffraction pattern orientation effects, observed in diffraction patterns when acquired at different sample orientations. The next highest intensity peak is at a 2θ value of 6.89° with a d-spacing of 1.28 nm. Overall, the diffraction pattern and orientation effects correlate well to the proposed structure (Fig. 5.8). Through acquisitions at different sample holder orientations, an increase in peak intensity at a 2θ value of 6.89° was observed.

This corresponds to the three planar C60 molecules (A, B, and C) being parallel with the X-ray beam (Fig. 5.8). The additional reflections in the diffraction pattern correlate to other intra-column fullerene planes, with increasing angle reflections from intra-fullerene domains.

175 Chapter 5

Figure 5.7 PXRD pattern of the 6 complex.

Owing to the nature of the sample, single-crystal diffraction data could not be obtained, however, the PXRD results were incorporated into a model developed using Accelrys software.[15] For illustrative clarity, all the 5 molecules shrouding the fullerenes have been omitted in Figures 5.8a-c. The model consistent with the data has a single column of C60 comprised of three C60 molecules in-plane (A, B, and C) with a dihedral angle of 80.7° (Fig. 5.8b) and centroid···centroid distances of 0.99 nm (A and B) and 1.28 nm (A and C). This is the repeating unit and is successfully rotated 90° with four turns corresponding to a complete rotation for a helical array. Attwood et al.[358] report a similar model based on single-crystal diffraction data with the unsubstituted calix[5]arene, which forms a linear zigzag array. However, this model does not fit the data obtained from PXRD in relation to the 6 complex. The arrangement of calixarenes in a single helical array is illustrated in Figure 5.9. This array propagates along its principal axis to form the strands of fullerene nano-fibres seen in Figure 5.1b. These fullerene nano-fibres continue to extend between calixarene backbone layers to form an extended packing structure (Fig. 5.11a).

176 Chapter 5

Figure 5.8 a) Proposed helical array of C60 columns; labels denote a tri-angular set intra-plane fullerenes, b) a line structure representation of a) with A - C fullerenes centroid···centroid dihedral

angle illustrated, and c) C60 column rotated 45° about the horizontal x-axis in a).

177 Chapter 5

Figure 5.9 Proposed calixarene packing arrangement around a C60 column, a) with the same orientation as Figure 5.8a, and b) with the same orientation as Figure 5.8c; based on Accelrys software.[15]

5.3.1.2 Thermogravimetric analysis and the development and characterisation of high-temperature analogues

TGA studies were carried out under air and argon atmospheres (Fig. 5.10). Under an atmosphere of air there is an initial 3% weight decrease from 23 °C to 213 °C, which is attributed to solvent surface desorption. This is interestingly followed by a gradual 5% weight increase from 213 °C to a maximum at 309 °C, which is attributed to partial [501] oxidation of C60 followed by a gradual weight reduction at a rate of 0.6 weight % °C-1, until final product decomposition at 503 °C. Under an argon atmosphere there is an initial 2% weight loss up to 213 °C, followed by a 34% weight decrease from 309 °C to 517 °C, then a slow declining plateau region from 517 °C to

705 °C corresponding to an 8% weight loss. The total weight loss prior to C60 degradation corresponds to loss of the calixarene for the 1:1 complex. Further weight loss is associated with degradation of C60 (Fig. 5.10) with a change in the rate of vaporisation above 800 °C, as reported previously by Malhotra et al.[502] In that particular study there was a thermally activated transformation of the fullerenes, in

178 Chapter 5 which larger covalently bonded fullerene arrays were formed.[502] In the present study only 5% of the original sample weight is retained at temperatures up to 1200 °C under an argon atmosphere.

Figure 5.10 TGA of the nano-fibres of 6, under both air and argon atmospheres.

Decomposition under argon was further investigated through a series of isothermal hold experiments, ranging from 400 °C to 800 °C, in 50 °C intervals. The post-heat treated sample morphology was evaluated using SEM, as highlighted in Figures 5.11a-d. The resulting materials appear to have crystalline and amorphous domains, with PXRD results confirming crystallinity in the samples. At temperatures below 600 °C the principal topographical structure was retained for bundles of the nano-fibres. The samples derived at 400 °C were constantly plagued by charging effects, as was the case for the as-synthesised complex (Figs. 5.11b and 5.11a, respectively). At annealing temperatures inclusive of 650 °C to 700 °C, samples no longer suffered from charging (Fig. 5.11c). This relates to loss of the calixarenes, and also indicates that the sample is now conductive. Above 700 °C structural degradation of the bundles of the fullerene nano-fibres occurs, resulting in increased surface defects, notably with pitted or

179 Chapter 5

‘skeleton’ array structures (Fig. 5.11d). The conductivity is associated with formation of graphitic material (Fig. 5.14). The degradation at high temperature aside, the selective removal of large cavitand molecules at high temperatures is without precedent in relation to fullerene architecture and engineering virtual porosity.

TGA data from the isothermal hold experiments indicate that temperatures of 650 °C and 700 °C, correlate to a 36% and 37% weight loss, respectively. For temperatures of 750 °C and 800 °C there is a 54% and 56% sample weight loss, respectively. Coupling of the SEM and TGA results suggest that heating under argon at increasing temperatures correlates to the 5 backbone being selectively volatilised from the sample leaving bundles of C60 nano-arrays, albeit with some graphitic material on the surface of the fibres. The formation of an all-C60 core structure was also confirmed by IR -1 -1 spectroscopy, with typical C60 absorbance peaks observed at 526 cm , 575 cm , 1181 cm-1, and 1428 cm-1 (Fig. A.3). In addition, PXRD of a 700 °C treated sample revealed a face centered cubic (fcc) crystal packing structure, which is the structure of [359] pristine C60 (Fig. A.4).

Figure 5.11 a) SEM image of as-synthesised bundles of the nano-fibres, and SEM images of bundles of the nano-fibres treated under argon at b) 400 °C, c) 700 °C, and d) 800 °C; all scale bars represent 10 μm.

180 Chapter 5

BET surface area measurements were undertaken for the as-synthesised and isothermal hold products (Fig. 5.12). Interestingly, there is a 45-fold increase in surface area demonstrated for the product obtained at 800 °C, in relation to the as-synthesised product. The surface area correlates with trends derived from the TGA data (Fig. 5.10). This indicates a steep increase in surface area until a plateau is reached between 600 °C to 700 °C, followed by another steep increase. The surface area increase is due to the selective volatilisation of the calixarene backbone and formation of graphitic material, providing increasing sites for gas absorption. This details a sufficient methodology for the selective virtual porosity increase of a solid-state fullerene complex. It is interesting to note that PXRD and IR data indicate a C60 fcc crystal structure, despite the large surface area established by BET.

Figure 5.12 BET surface area correlated to percent weight loss for as-synthesised and high temperature analogues annealed under argon, with associated annealing temperatures specified.

A study into the macroscopic electrical properties of the high temperature argon annealed analogues was undertaken. The objective of this study was to confirm if the high temperature argon annealing of the parent complex resulted in conductive material,

181 Chapter 5 which was inferred from the higher temperature analogue SEM samples not being susceptible to charging during imaging. The results for each sample were compared to that obtained for single crystals of pristine C60, which at room temperature is known to be a highly insulating n-type material.[503] I-V curves and resistance values were obtained, with results detailed for the 800 °C analogue (Fig. 5.13). As expected a non-ohmic conductance response was obtained for a single-crystal of C60 at room temperature, which is to be expected for an insulating material. Similar results were obtained for all the argon annealed analogues until the annealing temperature of ≥650 °C was reached, in which the materials demonstrated an ohmic conductance response. This is clearly observed for the 800 °C analogue, where a linear I-V curve, and a relatively stable resistance value are produced from the voltage sweep (Fig. 5.13). The bulk material resistance values obtained for the 800 °C analogues were ~1.2 kΩ. These results reinforce the assumption that material formed under an argon atmosphere, at annealing temperatures ≥650 °C, form a conductive material, with the conductivity being attributed to the formation of graphitic material, as discussed below.

Figure 5.13 I-V curve and resistance curve for the 800 °C argon annealed analogue.

182 Chapter 5

Further experiments were undertaken to assess the high temperature analogues, to indicate a probable cause of the change in electrical behaviour at specific temperatures. Dispersive Raman spectroscopy was undertaken for the range of high temperature analogues, and the as-synthesised product. The Raman spectrum of the as-synthesised product was similar to that of pure C60. In contrast, the high temperature annealed material gave additional Raman peaks (Fig. 5.14), notably around 1580 cm-1 and 1330 cm-1. These peaks correspond to the G- and D-bands, respectively, and are in [504] agreement with the literature values. The G-band relates to the graphitic E2g symmetry mode and reflects the structural integrity of the sp2-hybridised carbon atoms of the graphitic material. While, the D-band, provides information regarding amorphous impurities and disordered carbon.

Figure 5.14 Raman spectra of the high temperature analogues and as-synthesised product.

It is apparent that graphitic material is being formed during the annealing process. However, it is interesting to note that only the high temperature annealed products (≥650 °C) demonstrated conductive behaviour, indicated by the electrical resistance

183 Chapter 5 measurements. Through the analysis of the ID/IG ratio, it is evident that at ID/IG ratios >1 the material becomes conductive. This could either be due to the formation of more disordered carbon atoms at the higher temperatures, or simply an increased carbon-based coating is being formed, fully encapsulating the fibres. In this context it should be noted that the conducting material results in little dissolution of fullerene C60 in the presence of toluene. It is suggested that at the higher temperatures (≥650 °C) a graphitic material (not graphite itself), such as activated charcoal, is encapsulating the remaining bundles of C60 nano-fibres and forms a conductive sheath around a central core. The remaining presence of a C60 core is inferred from typical IR absorbance peaks attributed to C60 (Fig. A.3). It is noteworthy that these high temperature analogues were anticipated to be trialled for their gas and vapour sensing capacities, however, upon drop-casting onto pre-fabricated IDEs, subsequent electrical characterisation revealed the materials were not suitable, and this investigation was discontinued.

5.3.1.3 Chemotherapeutic assessment

Autophagy is an evolutionary-conserved intracellular degradation process that delivers cytoplasmic constituents to the lysosome, and plays an important role in a variety of physiological and pathological conditions.[505, 506] Autophagy is also considered to be an important cell death mechanism. Depending on the circumstances, autophagy has been shown to have drastically different effects on the response of tumour cells to anti-cancer therapy.[507, 508] Previous studies have turned to harnessing the unique physicochemical properties of certain nano-materials, such as quantum dots, to serve as a new class of [509, 510] autophagy regulators. Recently, nano-C60, the water-suspended nano-sized [497] crystal of fullerene C60, has shown interesting biological activity for both the induction of autophagy, and the sensitisation of cancer cell lines in laboratory [464] studies. With the unique fullerene C60 structural arrangement observed in the novel supramolecular 6 complex, laboratory-based chemotherapeutic trials were undertaken and results compared to nano-C60.

HeLa cells, are a cervical cancer cell line, commonly utilised for the effectiveness of possible chemotherapeutic agents. The HeLa cells were transfected with GFP-LC3 and samples were left untreated (Fig. 5.15), treated with the 6 complex (Fig. 5.16a), or treated with nano-C60 (Fig. 5.16b). It can be clearly seen that in the cell sample treated

184 Chapter 5 with the 6 complex there is LC3 puncta formation, which is the causation of the appearance of clusters of high fluorescence intensity, as illustrated in Figure 5.16b. LC3 puncta formation provides a valuable method for the detection of autophagosome formation within the cell, and is indicative of autophagy-induced cell processes.[511] It can be concluded that treatment with the 6 complex resulted in autophagy activation in the HeLa cell line. This implies that the novel supramolecular complex has the capacity to be utilised as a chemotherapeutic agent for this particular cancer cell line.

Figure 5.15 Fluorescence microscopy image of an untreated control sample of HeLa cells transfected with LC3-fused to GFP.

Figure 5.16 Fluorescence microscopy image of HeLa cells transfected with LC3-fused to GFP that have been a) treated with the 6 complex at a concentration of 2.5 µg mL-1, and b) treated with -1 nano-C60 at a concentration of 0.5 µg mL . Note that samples were treated for 24 h before being subjected to fluorescence microscopy.

The formation of LC3 puncta in the fluorescence microscopy image of the HeLa cells treated with nano-C60 implies that this structure also induces autophagy (Fig. 5.16b). It was observed that nano-C60 has a higher efficiency of autophagy induction in these cells, and requires concentrations in solution five times lower that of the 6 complex to produce similar results. This could be an effect from the different C60 spatial

185 Chapter 5 organisation within the structure, or possibly an artifact of the material’s solubilisation efficiency.

Figure 5.17 Percentage of cell death in drug-resistant MCF-7 breast cancer cells, illustrating a comparative study between different incubation methodologies.

Another avenue was explored for the assessment of the 6 complex as an effective chemotherapeutic agent, and as previously detailed nano-C60 was trialled for a direct comparison. The complex was trialled to see the effects on drug-resistant MCF-7 breast cancer cells. These cells are resistance to a common chemotherapy drug Dox, with minimal to no tumour cell death observed upon treatment with this drug (Fig. 5.17). Separate samples of MCF-7 cells were treated and incubated individually with the 6 complex or nano-C60, and cell death was assessed by Hoechst 33342/PI staining and results expressed as the percentage of PI-stained cells (Fig. 5.17). Analysis of these results indicate that the cell death expressed for the 6 complex and nano-C60 treatments were 5.22 ± 0.87% and 9.56 ± 1.74%, respectively. These data indicate that there are slightly elevated levels of cell death in the MCF-7 cells treated with the 6 complex, when compared to the control (1.74 ± 0.43%), and roughly double the cell death in the

186 Chapter 5 nano-C60 treated cells in comparison to the 6 complex. Consequently, the use of either

C60-based materials results in insufficient cell death for these to be considered effective chemotherapeutic agents. However, another series of cell cultures of MCF-7 cells were -1 treated with low-doses of the C60 nano-materials (0.5 µg mL ), and also the chemotherapeutic agent Dox (10 µg mL-1). Interestingly, a significant increase in cell death was observed in both nano-material systems, with a value of 53.91 ± 10.44% for the 6 complex, and a value of 53.04 ± 2.18% for nano-C60. This indicates that both of the C60 nano-materials sensitise the tumour cells to the drug Dox, a mechanism which led to significant cancer cell death.

These results not only indicate the successful application of the novel 6 complex to induce autophagy in a cervical cancer cell line, it also shows its ability to sensitise drug-resistant breast cancer cells, leading to cell death induction processes. It should be noted that these experiments should be viewed as preliminary trials. Currently, additional cancer cell line testing is being undertaken utilising the parent complex, and the high-temperature argon annealed analogues.

5.3.2 Helical rings of SWCNT bundles

As previously detailed there has been significant interest in CNTs, in particular SWCNTs, due to their extraordinary electrical and mechanical properties (Section 1.3). SWCNTs represent the ideal one-dimensional all-carbon structure which could pave the way for a revolution in the nano-electronics industry, and it is essential that efforts are directed towards creating unique architecture, or structural arrangements of SWCNTs that could reveal new physical properties. This current section details the investigation of helical rings of SWCNT bundles, which have been developed through a novel vacuum annealing methodology.

Over the past decade there has been a number of methods reported for the preparation of CNTs with annular geometries, with Smalley and coworkers reporting “fullerene crop-circles” and individual SWCNT toroids produced from a laser-grown method, albeit in 0.01% to 1% yields.[512] Song et al. have reported the high-yielding formation of rings of SWCNT bundles, around 100 nm in diameter, through a floating C-CVD

187 Chapter 5 process.[513] Post-growth processing methods have been developed to produce SWCNT rings, including but not limited to, solvent evaporation effects,[514] ultrasonication,[515] and light chemical etching.[516] The formation of rings from other CNT types such as MWCNTs, and double-walled CNTs (DWCNTs) have been reported.[517, 518] To the author’s knowledge there are no synthesis methodologies that utilise vacuum annealing processes to form helical rings of SWCNT bundles.

5.3.2.1 Transmission electron microscopy

From the TEM analysis of the chemical etched and vacuum annealed SWCNT samples, circular features were evident (Fig. 5.18) with defined wall spacings observed, albeit in low yield. The observed structures had a high level of symmetry and had an OD of ca. 250 nm, with a typical width ranging from 20 nm to 40 nm. Investigation of the structure illustrated in Figure 5.18, at an increased magnification (Fig. 5.19) revealed defined wall structure with ca. 1 nm spacings. The spacing dimensions correlate with the wall spacing of bundles of SWCNTs observed elsewhere in the sample, and these data support the assumption that rings of SWCNT bundles have been formed.

Figure 5.18 TEM image of a ring formed from a helical SWCNT bundle, with no seam observable. 188 Chapter 5

Figure 5.19 TEM image of the helical ring structure indicating the inherent wall structure (centre), and also periodicity of the wall projections.

An interesting feature observed in the SWCNT bundle ring structures in the literature is that typically the wall projection observed will span the entire circumference of the ring. This is not the case with the structures produced, where the only well defined wall structure is highlighted in Figure 5.19. This defined wall structure is periodically distributed around the circumference of the ring at approximately regular intervals. This wall projection phenomenon has been previously observed in ‘twisted’ SWCNT ropes and is attributed to projection effects in the direction of the incident electron beam, with different twisting rates leading to different spaced, defined wall periodicity.[519] The presence of these effects reveals the helical nature of the rings of SWCNT bundles that are produced, and suggests that the vacuum annealing process may produce favourable conditions for the joining of highly twisted SWCNT bundles.

The formation of all-carbon structures was supported through TEM analysis with the carbon map (Fig. 5.21a) correlating to the structure observed in Figure 5.20. Ring formation within the sample was not always entirely symmetrical in nature as demonstrated in Figure 5.21. From this TEM image, a twisting geometry is evident, with narrow and wide sections, within the ring. This variation in ring height is confirmed by a generated thickness map (Fig. 5.21b).

189 Chapter 5

Figure 5.20 TEM image of a separate ring of helical SWCNT bundles, which slightly diverts from a circular array.

Figure 5.21 TEM image of a) a carbon map, and b) a thickness map, of the asymmetric ring of helical SWCNT bundles.

As the structures described are produced in extremely low yields further analysis, such as structural information from Raman spectroscopy or electronic characterisation was not undertaken. This study details the formation of rings of helical SWCNT bundles using a chemical etching and vacuum annealing process, without the use of ultrasonication or ring closure chemistries.

From the structural information obtained from the aforementioned analyses it is apparent that the chemical etching prior to vacuum annealing would induce tip-opening and through the high energy process of annealing, joining of the tips of the SWCNT

190 Chapter 5 bundles was initiated, resulting in the helical, all-carbon structures. Improvement of the induction of such effects through synthesis condition optimisation, is an on-going area of investigation.

5.3.3 Square-geometry carbon nano-fibres

An array of different SWCNT synthesis methods have been detailed for the growth of CNTs over the years (Section 1.2.1.2). C-CVD processes are among the most highly reported methods of carbon nano-material generation, and can be scaled to a commercial production level. This process was subsequently utilised in an attempt to generate SWCNT material. However, a wide variety of carbon materials were produces ranging from amorphous carbon through to carbon nano-fibres. Consequently, this current section will briefly discuss a novel square-geometry carbon nano-fibre structure that was commonly encountered in these samples, giving rise to the formation of a unique all-carbon architecture.

5.3.3.1 Scanning electron microscopy

A wide variety of carbon-based structures were formed from the C-CVD process and are illustrated in Figure 5.22. Vermicular or spiral structures are displayed in Figure 5.22a, which are commonly encountered from growth processes utilising acetylene feedstock at temperatures below 1000 °C.[520, 521] Other structures commonly encountered are the large circular carbon fibres (Fig. 5.22b) and smaller-scale circular nano-fibres (Fig. 5.22c), which are frequently documented in the literature. On the other hand, a less frequently encountered carbon structure was observed in the growth material, consisting of geometry similar to that of an elongated hexagonal cylinder (Fig. 5.22d). In this particular case a significant hollowed out internal core can be observed, which is possibly where a Ni catalyst particle could have been according to the tip-growth mechanism widely accepted for VGCFs. To the author’s knowledge this hexagonal structure has not been documented in the literature in association with Ni catalysed CVD, utilising an acetylene feedstock.

191 Chapter 5

Figure 5.22 A range of different carbon materials generated through C-CVD processes using a Ni catalyst these structures include a) vermicular or spiral shaped carbon fibres; scale bar 10 µm, b) large carbon fibres; scale bar 1 µm, c) circular carbon nano-fibres; scale 20 nm, and d) hexagonal fibres, in this particular case with an open-tip, where presumably a Ni catalyst particle was once located; scale bar 200 nm.

A separate noteworthy all-carbon structure commonly observed in the C-CVD growth material is the square-geometry CNF (Fig. 5.23). These structures were typically 200 nm to 400 nm in width and several tens of microns in length. From analysis of the tip of the CNF it was clearly observed that an annular layering structure is propagating from the centre of the fibre. However, once a threshold diameter is reached, the layer growth arrangement clearly changes to flat, planar sheets to produce a square-geometry outer layer of the CNF. Furthermore, observations suggest that a hollow-core exists within these structures (Fig. 5.23).

These square-geometry CNF structures have not been reported in the literature. It is possible that these CNFs undergo a similar growth mechanism to traditional CNFs. This occurs by the surface diffusion of carbon through the catalytic particle, which templates the carbon condensation. As more carbon feedstock enters the system the catalyst particle is projected off the substrate and the CNF will prorogate and layers will continue to grow and increase the OD of the CNF. The way that the square-geometry CNFs could deviate from this is by the continued carbon feedstock addition, at maximum diameter growth. The layer thickness is no longer dictated by the catalyst

192 Chapter 5 particle and the formation of lower energy planar carbon layers will be preferred over high energy curvature growth. Additionally, the growth may entirely occur on the catalyst particle and the square-geometry could be an effect of the different crystal faces of the solid particle.[522] An interesting aspect of study would be the physical properties of these structures, to determine whether deviations are observed from standard CNFs.

Figure 5.23 A SEM image of a square-geometry CNF, grown through C-CVD processes; scale bar 200 nm.

5.4 Conclusion

The results demonstrate the formation of C60 nano-arrays which form bundles of nano-fibres based on complexation of C60 and p-tBu-calix[5]arene from toluene in a quantitative yield. A grinding/sonication method for obtaining the complex has also been established, and complex formation is possible from crude fullerene soot. Annealing under an inert atmosphere demonstrates the use of this complex as a precursor in preparing exclusively carbon-based material which can be electrically conductive. The ensuing large surface area has implications for the development of catalyst support frameworks, hydrogen and other gas storage material architecture, and device technology.

The biological activity of the novel complex, (C60)∩(p-tBu-calix[5]arene), has been assessed, in relation to multiple cancer cell lines. In addition, the activity of the parent

193 Chapter 5 complex was directly compared to nano-C60, the water-suspended nano-sized crystal of fullerene C60. It was demonstrated that (C60)∩(p-tBu-calix[5]arene) induces LC3 puncta formation in HeLa cells, a cervical cancer cell line, and supports autophagy in a free radical-dependant fashion. This complex has also been shown to sensitise MCF-7 cells, a Dox-resistance breast cancer cell line, to the chemotherapeutic agent, and reduces cell drug resistance possibly through an autophagy activation mechanism. This presents a highly topical area of investigation and will form the basis of additional biological studies into the effects of solid-state fullerene architecture facilitated by supramolecular processes.

The formation of additional all-carbon nano-architectures are established, demonstrating synthesis conditions and subsequent characterisation, of rings of helical SWCNT bundles and square-geometry CNFs. The formation of ‘toroid-like’ SWCNT bundles from a combined chemical etching and vacuum annealing methodology is illustrated, albeit in low yield. Furthermore, the unprecedented report of CNFs with square-geometry from low temperature C-CVD processes has been described. These examples not only illustrate the shear versatility of carbon but represent structures that could have significant applications in future target areas, such as field emission materials, sensor development and nano-material composites.

194

Chapter 6

A Water-Soluble Fluoroionophore: p-Phenyl Sulfonated Calix[8]arene

Chapter 6

6 A water-soluble fluoroionophore: p-phenyl sulfonated calix[8]arene

6.1 Introduction

The requirement for water-soluble fluoroionophores is being increasingly addressed in the chemosensor literature.[523-528] The majority of fluorescent chemosensors for cationic analytes are composed of a cation recognition unit (ionophore) combined with a fluorogenic unit (fluorophore) described as fluoroionophores.[529] Research has focused on the attachment of singular or multi-fluorophore moieties to an ionophore to provide selective binding to the analyte, and the fluorophore, either through enhancement or quenching signals of the binding event.[527, 528] The use of fluorescent cavitands as metal ion selective receptors has not been studied intensively,[530] despite the inherent advantages of using cavitands which have remarkable versatility, in building up functional molecules.

In recent years there have been significant developments in exploiting the supramolecular chemistry of calixarenes to develop novel chemosensors. Through upper and lower rim functionalisation of specific calixarenes, their versatility in the sensor field has been well documented in applications including, but not limited to, thin-films for mass-based organic vapour analyte detection,[531] active surfaces for ion-sensitive field-effect transistors (ISFET),[532] potentiometric sensors for metal cations,[533] and the capping of core-shell quantum dots for toxic metal ion fluorescence detection,[534] and, more recently, anion sensors.[535-543] Related to this are application of calixarenes in the biological sector for the chromogenic detection of hypertensive drugs,[533] protein sensors,[544] detection of the neurotransmitter acetylcholine,[545] and sodium ion concentration in clinical urine samples.[546]

There is an increasing demand for the development of new chemosensors, not only for the detection of alkali and alkaline-earth metal ions,[526, 547-551] but also for use in optical probes for heavy and transition metal (HTM) ions.[525, 530, 534, 552-558] Standard analytical techniques utilised for metal ion analysis, including atomic absorption spectrometry,[559] inductively coupled plasma-mass spectrometry,[560] radioanalysis,[561] flame photometry,[562] and ion selective electrodes.[563-565] Supramolecular-based optical

196 Chapter 6 chemosensors offer a variety of advantages over these traditional techniques with the ability to perform real-time, remote and continuous analytical measurements, allowing increased sensitivity and selectivity against competing analytes, reduced instrumental drift and offer miniaturisation capabilities, which are reliable and cost effective.[526, 556, 566-568]

The most widely studied calixarene systems are based on the different conformational isomers of calix[4]arene,[527] which have been successfully utilised as the principal building blocks used to construct suitable ionophores, either podand type or macrobicyclic as calyx-crowns.[569] Once a suitable molecular recognition platform is designed either singular or multi-fluorophoric units are attached to provide a suitable interaction signal correlated with the ion binding event. An inherent disadvantage of these systems is their lack of water solubility and multi-step synthetic pathways. As mentioned previously, one of the main analyte focus areas of this particular class of optical sensors is the HTM ions, with systems operating in solely aqueous media being limited, with mixed aqueous/organic solvent systems typically utilised. While HTM ions are relatively easy to chelate and detect in organic solvents, they are rather difficult to recognise directly in aqueous environments due to their strong hydration energy. This and other current limitations need to be addressed in the initial stages of the HTM chemosensor design for both environmental and biological applications.[528, 570]

Calixarene-based metal cation chemosensors that are soluble in aqueous media are desirable, especially for fluorescence probing and imaging in biological systems.[528] In addition to this the excitation wavelength must be high enough to avoid the effects of autofluorescence, and it must be compatible with wavelengths available in LED (>360 nm).[528] Clearly, the chemosensor should have high sensitivity and selectivity to the analyte of interest, extended shelf-life, effective at low concentrations, and recoverable. A fluoroionophore which is readily prepared in reasonable yield, without the need to attach pendant fluorophoric units, and at the same time is water-soluble represents a significant shift in the development of fluorometric chemosensors.

This chapter describes the use of a novel autofluorescence calixarene, p-phenol sulfonated calix[8]arene,[315] as a water-soluble fluoroionophore for the detection of

197 Chapter 6 divalent metal cations analytes in aqueous systems, completely void of the attachment of chromophore moieties to the principal structure. This represents a significant shift in chemosensor design for metal cations in aqueous media. This ‘extended arm’ calixarene features in Chapter 2 of this thesis, and is utilised in studies on using p-sulfonated and p-phosphonated calixarenes to solublise SWCNTs into aqueous media, with tunable diameter-selective separation realised. Indeed, during this study the p-phenyl-sulfonato-calix[8]arene (4C) was identified as a possible candidate for fluoroionophoric studies.

6.2 Experimental

6.2.1 Materials

4C was prepared as described earlier.[315] Similarly, p-benzyl-sulfonato-calix[8]arene (4B) and the parent compound, p-sulfonato-calix[8]arene (4A) were prepared through literature procedures.[317, 388] Lithium chloride, sodium chloride, potassium chloride, rubidium chloride, cesium chloride, magnesium chloride, calcium acetate, strontium nitrate, barium nitrate, aluminium(III) chloride, chromium(III) nitrate, iron(II) chloride, iron(III) chloride, cobalt(II) chloride, nickel(II) chloride, copper(II) acetate, ruthenium(III) chloride, silver nitrate, cadmium(II) nitrate, indium(III) chloride, mercury(I) nitrate, lead(II) nitrate, and uranyl nitrate, obtained from Alfa Aesar or Sigma-Aldrich, were of analytical grade and used without further purification.

6.2.2 Methods pH dependant intramolecular fluorescence studies: A solution of 4C (40 μM) in MilliQ (>18 MΩ cm-1) was analysed over a pH range of 2 - 12. Other concentration solutions were studied, however, 40 μM was found to be an effective concentration, with solutions as high as 1 mM resulting in auto-quenching.

Metal cation interaction studies: MilliQ water (>18 MΩ cm-1) was employed as the solvent. Stock solutions of the analytes (1 mM), were prepared in MilliQ water (>18 MΩ cm-1), titrated against 4C (40 μM). Fluorescence measurements were performed using a Cary Eclipse fluorescence spectrophotometer (Varian). Dual

198 Chapter 6 excitation/emission lines were monitored at λem = 513 nm (λex = 335 nm) and

λem = 488 nm (λex = 423 nm), with excitation and emission slit widths of 5 nm, and a manual PMT voltage of 720 V. A typical experiment involved successive addition of the metal cation analyte stock solutions (0.1 μM, 1 μM, 2 μM, 5 μM, 10 μM, 20 μM, 30 μM, 40 μM, 50 μM, 80 μM, 100 μM), adjusted to pH 6.5. The solutions were left for 45 min to equilibrate before analysis. Control samples were also utilised to monitor the standard calixarene solution over the experimental time period, and to assess the dilution effect of the MilliQ water component in the metal cation additions. All metal cations utilised in the interaction study were analysed at high concentrations (>1 mM) to 2+ assess for any possible fluorescence contribution. The solution of uranyl ion, [UO2] , did show slight fluorescence; however this did not interfere with the current study.

Time-based assessment of the specified fluorescent wavelengths for p-phenyl-sulfonato-calix[8]arene: A solution of 4C (40 μM) was intermittently monitored over 23 days, using the above mention fluorescence spectrometer, to assess the shelf-life of the compound in aqueous media.

Potentiometric pKa determinations of p-phenyl-sulfonato-calix[8]arene: 4C (18.1 mg, 398 μM) was dissolved in aqueous HCl (0.01 M) and titrated against aqueous NaOH (0.01 M). The titrations were carried out at 25 °C and the mixture was subjected to constant stirring.

Critical micelle concentration (CMC) experiments: The entire CMC experiment was carried out in a thermostatted water bath at 25 °C. Aliquots of a stock solution of 4C (40 μM and 1 mM, respectively) were added to a vessel with MilliQ water (>18 MΩ cm-1) while conductivity measurements provided CMC values for this system.

Micelle and vesicle formation assessment: Two different methodologies were employed to assess for the formation of micelles or vesicles in solution. Firstly, once the CMC was established for 4C, fluorescence measurements were obtained at concentrations below this value to check whether the solutions were fluorescent. Secondly, Rhodamine dye can be regarded as a low molecular weight drug model and is

199 Chapter 6 often used to check whether aqueous compartments exist within supramolecular structures.[571] A modified literature procedure was followed,[571] in which 4C was dissolved in a Rhodamine 6G aqueous solution (3 mM), and allowed to stand for one week to reach equilibrium. Following this, the solution was purified through a size exclusion chromatographic column (NAP-25, GE Healthcare Bio-Sciences Corporation, NJ, USA) and fluorescence spectra recorded; potassium iodide (1.6 M) was used as a fluorescence quencher to check for micelle/vesicle formation. In addition, an aliquot of the dye/4C mixture was dialysed over a week and its fluorescence spectra recorded.

Dynamic light scattering and zeta potential measurements: Dynamic light scattering (DLS) experiments were carried out on 4C (40 μM, 1 mM) with pH variation (pH range of 2 - 12) using HCl and NaOH. Sub-samples were also taken and utilised for zeta potential measurements. These measurements were acquired using a Zetasizer Nanoseries DLS system (Malvern Instruments, Worcestershire, UK) with quartz cuvettes and zeta flow cells.

AFM analysis: A variety of different samples were analysed using a PicoPlus system with PicoScan 2000 and a PicoSPM controller (Agilent Technologies) in tapping mode. Topographic images were acquired using both grade 14 and grade 15 tapping mode cantilevers. Samples analysed included solutions of 4C (40 μM and 1 mM) at varying pHs, solutions from metal interaction studies, solutions below the CMC concentration, solutions with acetonitrile addition (30%) to study hydrogen-bonding in the system, and solutions from drug carrier trials with Rhodamine 6G. All samples were fabricated through drop-casting on freshly cleaved mica and dried using a gentle flow of N2.

200 Chapter 6

6.3 Results and discussion

6.3.1 pH dependent fluorescence studies

SO3H

HO3S SO3H

OH OH HO

HO3S OH HO SO3H

OH HO OH

HO3S SO3H

SO3H Figure 6.1 The structural diagram of 4C.

For 4C (Fig. 6.1) proton dissociation of the upper-rim sulfonic acid groups and lower rim phenolic hydroxyls is possible with changes in the pH of the local environment. Dissociation events can provide insight into the nature of the fluorescence of the system, as well as facilitate metal cation interactions.[568] Prior to metal cation interaction studies on 4C, the effect of such dissociation events on the fluorescence spectrum was evaluated in MilliQ water, with no organic solvent component necessary and the sulfonic acid head groups rendering this compound water-soluble.

201 Chapter 6

Figure 6.2 Contour plot of the fluorescence spectra of 4C at pH 6.5, with the fluorescence intensity scale (top right).

An as-prepared solution of 4C (40 μM) was assessed for its fluorogenic properties (Fig. 6.2). Within the contour plot excitation and emission lines were identified and monitored for subsequent experiments. The major emission bands were observed at

λem = 513 nm (λex = 335 nm) and λem = 488 nm (λex = 423 nm ). Of the two emission lines, λem = 513 nm has a higher relative intensity when compared to λem = 488 nm.

Figure 6.3 a) Dependent series of emission spectra (λex = 335 nm) for 4C, and b) the respective contour plot with fluorescence intensity scale (top right). 202 Chapter 6

Figure 6.4 a) Dependent series of emission spectra (λex = 423 nm) for 4C, and b) the respective contour plot with fluorescence intensity scale (top right).

The fluorescence emission intensity of 4C was monitored at the above mentioned excitation wavelengths with spectra recorded at a series of different adjusted pH values.

For the emission spectra for λex = 335 nm (Fig. 6.3) there is a gradual increase in fluorescence intensity at λem = 513 nm, with pH increasing from 2.5 to ~6. There is a

λmax between pH 5.2 and pH 6.5 with the gradual, yet substantial, decline in fluorescence intensity at this emission wavelength at pH > 7 and pH < 5. Furthermore, at higher pH values (>9.1) a secondary emission peak is evident (λmax = 430 nm).

Emission data from excitation at wavelength λex = 335 nm supports the autofluorescence of 4C, as well as providing solid evidence that pH plays a vital role in the fluorescence properties of the compound. The emission spectra of 4C at

λex = 423 nm, is the same pH regime as above and is given in Figure 6.4. These emission spectra follow the same intensity profile as those of λex = 335 nm spectra (Fig.

6.3), in which the highest intensity for λmax (490 nm) is established at pH 6.1. No additional peaks or peak shifts were observed throughout the pH series at this excitation wavelength (λex = 423 nm).

It is clear that the 4C system is sensitive to pH and this has a dramatic effect on the fluorescence emission lines, and their intensities. Effects of pH on the system are fully reversible, supporting a proton dissociation equilibrium mechanism. pH dependant intramolecular interactions are responsible for the fluorescence properties demonstrated by the solvated system, and the conformational flexibility of this higher-order cavitand may account for the observed fluorescence. The fluorescence may arise from dimer formation resulting from overlapping of π-orbitals from phenolic rings in close

203 Chapter 6 conformational proximity.[572] Obviously, the fluorescence properties demonstrated by this water-soluble calixarene could stem from a variety of supramolecular interactions in solution. Different supramolecular arrangements include inter- and intra-molecular dimer formation,[573, 574] micelle-formation,[388, 575] and higher-order supramolecular structures such as capsule, vesicle formation,[571, 576] or extended bilayers.[393] These possible interactions were investigated and gave insight into the possible mechanism behind the fluorescence of this compound.

6.3.2 Supramolecular interaction/conformational studies

The different possible supramolecular arrangements for this system, mentioned above, were further investigated to provide insight into the fluorescent nature of the compound. Conductivity experiments were undertaken to establish the CMC which involved monitoring the conductivity of the solution with increasing concentration of 4C,[577] as illustrated in Figure 6.5.

Figure 6.5 Conductivity profile with increasing concentration of 4C.

204 Chapter 6

The two domains correspond to pre- and post-aggregation with fitted lines indicating a CMC of 2 x 10-6 mol L-1. This information is important in determining if self-assembly, or intermolecular interactions, give rise to the fluorescence phenomenon. Higher-order supramolecular structures involving such amphiphiles are possible and include bilayers, hydrogen-bonded capsules, micelles, and vesicles. Indeed, fluorescence occurs at concentrations well below the CMC, albeit at much lower intensity (Fig. 6.6), with emission profiles at λem = 513 nm (λex = 335 nm). A similar trend is demonstrated for the emission profiles at λex = 423 nm (Fig. A.5). Thus it appears that fluorescence is not established through the intimate interaction of neighbouring phenol ring systems arising from intermolecular interactions, and that intramolecular association events ultimately leading to excimer formation and system fluorescence.[527] To further support these claims a metal cation binding study has been undertaken and will be described in Section 6.3.3.

Figure 6.6 Emission profiles (λex = 335 nm) for solutions of increasing concentration of 4C; concentrations below and above the CMC utilised at detector voltage 800 V.

205 Chapter 6

The zeta potential of a solution of 4C (1 mM) was measured over a pH range of 1.9 - 13 (Fig. 6.7), which gives a system zeta potential maximum at 5 < pH < 7. There is a rapid zeta potential decrease from pH 2 – 5, and a gradual positive zeta potential increase from pH 5 to pH 13. This indicates that there is a relatively significant negative net surface charge present between pH 5 and 7, which coincides with the system’s maximum fluorescence intensity (Figs. 6.3 and 6.4). The increase in overall negative surface charge cannot be explained by the proton dissociation of the sulfonic acid head groups, as this occurs at pH < 1.[578, 579] Nevertheless, the increase in overall negative surface charge coupled with the pH dependent fluorescence studies could arise from the aforementioned higher-order structures involving intermolecular π-π stacking of neighbouring phenyl functionalities. At concentrations below the CMC intramolecular interactions may dominate the fluorescence intensity with intermolecular interactions contributing to the fluorescence above the CMC.

Figure 6.7 Zeta potential measurements for solutions of 4C (1 mM) over a pH range.

206 Chapter 6

Figure 6.8 Particle diameter through DLS measurements for solutions of 4C (1 mM) over a pH range.

DLS was utilised to investigate whether higher-order supramolecular structures were being formed in solution and to assess and quantify the structure size and monitor any changes over the pH range of 1.9 to 13 (Fig. 6.8). The results are based on percent volume statistics of the structures in solution. It was observed that at pH 2, small diameter particles (5 nm) dominated the solution statistics. As pH increases there is also an increase in particle diameter (35 nm to 45 nm). There is an interesting particle diameter maximum seen at approximately pH 7, which also correlates to approximately the same pH where this system exhibits maximum fluorescence intensity at the characteristic emission wavelengths. Immediately following this maximum there is a decrease in particle size (35 nm to 45 nm). This data indicates that the size of particles/structures that exist in solution are dependant on solution pH. In addition, these results represent a slight correlation between particle/structure diameter and fluorescence intensity, indicating the possibility of higher-order supramolecular structures being formed in solution. The observed pH effect on average particle size may be directly related to dissociation events of the lower rim phenolic hydroxyl

207 Chapter 6 groups, leading to different conformational arrangements and different intermolecular stacking dynamics in solution. While DLS clearly shows aggregation of 4C, the nature of the fluorescence as an inter- or intra-molecular process cannot be established utilising this methodology.

Figure 6.9 Tapping mode AFM characterisation of solid-state structures of 4C on a mica substrate at a) pH 2, b) pH 6, and c) pH 13. Representative three-dimensional topographical images correlating to contour plots are also shown (right).

Solutions of 4C at pH values which featured in the above studies were characterised using AFM (Fig. 6.9). It can be clearly seen that at the pH values of 2, 6, and 13 the

208 Chapter 6 aggregation of 4C is different. Over the pH range investigated it can be seen from the topology that the particles are more spherical on the mica substrate, which is interesting given that if micelle or vesicle formation occurs, circular flat thin-film topology also occurs, differing from the gradient topology observed for a particle.[580] AFM indicates that size, aggregation and to a lesser extend surface morphology is effected by varying the pH. At a pH 2 (Fig. 6.9a) 2 nm to 5 nm particles form random chain-like aggregates, whereas at pH 6 (Fig. 6.9b) particles are more mono-dispersed with diameters of 10 nm to 20 nm. This changes again at pH 13 (Fig. 6.9c), with an increase in particle diameter from 20 nm to >100 nm, and large aggregates. This is interesting, as the size of the particles and aggregates in the solid-state show a slight diameter correlation when compared to DLS solution studies. This shows that over the pH range studied, any inference to supramolecular arrangement in solution based on solid-state characterisation, should be made with caution. At higher pH values the zeta potential, or surface charge of the compound, decreases indicating more likelihood of molecular aggregation. This is compounded by the much greater positive zeta potential of solutions at pH 2, and AFM results clearly show an increase in molecular aggregation. Interestingly, these results show no evidence that the solvated compound is organised in a bilayer arrangement, which is prevalent for the sulfonated calix[n]arenes.[428, 581-584] Furthermore, it is unclear whether these results can provide solid evidence to support or reject the formation of micelles or vesicles over the pH range studied.

To explore whether the formation of larger supramolecular structures are present in solution, experiments were conducted to establish if this system could act as a molecular carrier. For this purpose Rhodamine 6G was used to mimic an encapsulated drug, i.e. whether the molecules are retained within the hydrophilic pocket within the structure. This approach is common in studying micelle/vesicle formation.[571, 585] Two separation principles were utilised involving the separation of the ‘encapsulated’ Rhodamine 6G and the free dye. 4C was added to a Rhodamine 6G solution (3 M) and allowed to equilibrate over the period of one week. The first separation method involved the use of a size exclusion column (GE, NAP-25) and appropriate fractions were sampled for further analysis. The second methodology involved dialysing the mixture to allow the free dye to diffuse out of the tubing while retaining the larger-size encapsulated system, for further analysis. For the analysis of the fractions from the size exclusion column the fluorescence intensity was monitored and a known fluorescence quencher, potassium

209 Chapter 6 iodide (1.6 M) added. Thus any remaining fluorescence could be attributed to the encapsulated dye. In the present study addition of KI quenched the fluorescence, where the dye should not be quenched, if encapsulated by the supramolecular structure. Similar results were obtained through the analysis of the dialysed solution and clearly all the Rhodamine exists as the free dye and was dialysed from the system.

Figure 6.10 A schematic of the lower rim hydroxyl group, hydrogen-bonding directed intramolecular excimer formation. Note for clarity only selected fragments of the principal calix[8]arene structure have been illustrated.

Any higher-order supramolecular structure, if present, in solution is not effective in encapsulating the drug mimic. This, coupled with the pH dependent fluorescence emission data suggest that the system’s fluorescence is derived exclusively from intramolecular interactions through conformational changes, and, since calix[8]arenes are conformationally flexible, this allows alignment of phenyl units resulting in dimer/excimer formation through π-π stacking (Fig. 6.10). This can be explained as the sulfonic acid head groups are deprotonated at low pH (pKa1 = 1) and as the pH is increased (pKa2 = 6.00) the proton dissociation of one of the phenolic hydroxyls will be facilitated, leaving a free O-.[579] This moiety is capable of conformational stabilisation by hydrogen-bonding with an adjacent phenolic hydrogen, to produce a double cone conformation. This facilitates at least one occurrence of π-π stacking of the principal

210 Chapter 6 phenyl components, however, due to the rigid nature of the p-phenyl upper rim functionality, it is possible two separate intramolecular interaction events are occurring per molecule (Fig. 6.10). pKa3, observed at pH 8.86, results in the deprotonation of a second phenolic hydroxyl group and clearly this disrupts the double cone conformation, with intramolecular rearrangement, resulting in a dramatic reduction of fluorescence and a shift to a shorter emission wavelength at λex = 423 nm. The fact that fluorescence continues above this pH indicates a secondary transition to form a different intramolecular excimer.

The fluorescence studies of the parent compound, 4A, were also undertaken. This compound demonstrated no fluorescence within the excitation/emission range analysed, this provides evidence that the extended p-phenyl moiety is critical in the intramolecular arrangement of 4C. Similarly, the analogous 4B demonstrated no fluorescence characteristics, possibly due to more conformational flexibility derived from the methylene bridge between the two phenyl sub-units.

Figure 6.11 A plot of the emission line fluorescence intensities of 4C with increasing acetonitrile component.

211 Chapter 6

A study was also undertaken to assess the effect of acetonitrile on the system (Fig. 6.11), which is known to disrupt the hydrogen-bonding in supramolecular systems. This could give valuable information on whether larger hydrogen-bonded structures were contributing to the sample fluorescence. This study produced interesting results which indicated that, instead of a relationship where the fluorescence intensity decreases due to dilution effects of the increased acetonitrile component in the mixture, a parabolic response is seen. This indicates that up until a 30% component of acetonitrile, an increase in fluorescence intensity at λem = 513 nm (λem = 335 nm) occurs with a steady decrease from 30% to 50% observed. The acetonitrile is obviously having a synergistic fluorescence effect, in which methyl groups can reside in the clefts of the folded calixarene, which could facilitate the intramolecular interactions and thus effect the fluorescence of the system.

6.3.3 Metal cation interaction studies

Due to the fluorescent nature of 4C, the effect of metal ions and possible metal-calixarene interactions were investigated for the potential application in metal cation sensing. This avenue was explored as there is a current gap in the literature in relation to possible candidates for calixarene-based fluoroionophores, specifically with no requirement for a fluorogenic moiety attachment. This not only reduces the number of synthetic steps and is more cost effective, but it also allows increased sites for further selective modifications, such as crown ether bridges and specific functional motifs, to target specific analytes of interest. These interaction studies focused on a wide range of metal cations, 23 in total, in a variety of valent states, with differing counter-ions.

6.3.3.1 Time effects on the metal cation sensor and metal addition fluorescence kinetics

A study of the time-dependent effects on stored solutions of 4C was undertaken to check whether the fluorescence intensity degraded over time. This is an important aspect to investigate as the fluorescence dynamics of a freshly prepared solution may change upon reaching equilibrium, and the shelf-life of a fluorescence-based sensor is critical.

212 Chapter 6

Figure 6.12 Fluorescence intensity at λem = 513 nm (λex = 335 nm) for a freshly prepared solution of 4C (40 µM) a) monitored over a 42 h period, and b) in the first 7.5 h.

The fluorescence intensity of the emission line λem = 513 nm (λex = 335 nm), was monitored over a finite period, commencing immediately after a solution of 4C (40 μM) was prepared (Fig. 6.12). It can be clearly seen that, immediately after dissolution, the fluorescence intensity of this emission line decreases and plateaus after 180 min. There is a decrease of approximately 3% of the fluorescence intensity over the initial 180 min period, with the majority of fluorescence intensity reduction occurring within the first hour. There is a slight deviation from a steady fluorescence signal after the 180 min period, however, it is less than 1% and is considered insignificant. From these results, protocols were set in place for metal binding experiments, where a solution was left to stand for at least three hours before being used for metal binding studies.

The fluorescence intensity was monitoring for an extended period of over three weeks (Fig. 6.13). It was observed that there is an initial decline to 90% intensity within the first 60 h, followed by a gradual decrease between 125 h and 400 h, where the intensity drops approximately 5% to about 85% intensity, where it remains for the final 150 h of the trial. These data imply that monitoring the original solution fluorescence intensity

213 Chapter 6 and establishing control samples will require incorporation into the analytical methodology, for the use of this compound as an analytical metal cation sensor. While it can be seen that the solution fluorescence is stable, with repeat analyses within a 50+ h period, overall fluorescence intensity decay is observed. In addition, it is noteworthy that the solution was not constantly in a sample cell being monitored for the entire 550 h, and slight decreases in fluorescence may be the result of sample storage, dynamic equilibrium effects, and/or fluctuations in ambient temperature. For any fluorescence studies on 4C, and for its use as a metal cation sensor, it is advisable to always utilise freshly prepared solutions of 4C, and allow these solutions to stand for three hours before use. At a bare minimum, it is suggested that these solutions can be utilised for 40 h. Furthermore, it is recommended that a control solution be monitored to allow for baseline correction.

Figure 6.13 Long-term fluorescence intensity monitoring of a solution of 4C.

The fluorescence intensity at λem = 513 nm (λex = 335 nm) was monitored as soon as a Co2+ solution (20 μM) was directly added to a solution of 4C (40 μM) (Fig. 6.14). Immediately following the addition of the metal cation solution, significant quenching

214 Chapter 6 in fluorescence intensity was observed, followed by a secondary increase, and then decrease in intensity. This can be attributed to the metal cation disturbing the system dynamics, resulting in an equilibrium process taking place. After 30 min the system reaches equilibrium, and stability of the fluorescence intensity is established. This information was incorporated into the metal binding experimental protocol, so that between metal cation addition and fluorescence monitoring, solutions were allowed to equilibrate for increased reproducibility of the results. This should be incorporated into any analytical protocol for the use of 4C as a fluorescence-based metal cation sensor.

Figure 6.14 Real-time emission line monitoring (λex = 335 nm) of a solution of 4C (40 μM), the arrow indicates the addition of the Co2+ metal cation solution (20 μM).

6.3.3.2 Metal binding effects on p-phenyl-sulfonato-calix[8]arene

The effect of a range of different metal cations were analysed to check whether certain metal ions interact with the cavitand, and alter the observed fluorescence signal. This included the introduction of 23 different metal cations (80 μM) to solutions of 4C (40 μM), allowing equilibrium to be reached, followed by fluorescence monitoring (Fig. 6.15). These data are categorised into three sections based on mono-, di-, and trivalent 215 Chapter 6 state cations. The alkali metals and monovalent metal cations have an insignificant effect on the fluorescence intensity of 4C, and all divalent cations assessed have a quenching effect on the system fluorescence, with the most prominent decrease 2+ 2+ 2+ 2+ 2+ observed for Ni , Cu , Co , [UO2] , and Pb , all of which reduce fluorescence intensity by a minimum of 25% of the fluorescence intensity in the absence of metal cations (Io). The other divalent cations appear to have a quenching effect on the system, although this is limited to no more than 20% of Io. The effects of the trivalent cations studied suggest <10% quenching of the fluorescence intensity, with slightly more quenching demonstrated by the addition of Ru3+. Interestingly, the addition of In3+ cations results in an increase in fluorescence intensity, and suggests a chelation enhancement of fluorescence (CHEF), whereas all other metal-calixarene interactions result in a chelation enhancement of quenching (CHEQ).[558] Furthermore, from the suite of counter anions utilised in this study, it can be stated that the type of counter anion has little to no effect on the CHEF and/or CHEQ observed.

Figure 6.15 Quenching ratio of a solution of 4C (40 μM) with the addition of a range of different metal cations (2 molar equiv.).

216 Chapter 6

Metal cation addition to a solution of 4C (Fig. 6.15) clearly indicates that this system undergoes significant interaction with divalent metal cations. This quenching effect will 2+ be explored in an attempt to facilitate a quantitative sensor for divalent cations [UO2] , Pb2+, Co2+, and Cu2+, with monovalent silver and trivalent indium cations included for direct comparison.

Solutions of the metal cations were titrated in the presence of 4C, with the results used to construct fluorescence profiles (Fig. 6.16), which in turn can be utilised to assess cation binding constants, the selectivity and sensitivity of the sensor, and to extract representative sensor calibration data. The fluorescence excitation and emission wavelengths monitored for these metal cation interaction studies were modeled from the characteristic peaks obtained from the aforementioned pH dependence study (Section 6.3.1).

2+ 2+ 2+ 2+ + Successive addition of the metal cation solutions ([UO2] , Pb , Co , Cu , Ag , and In3+) had differing effects on the fluorescence intensity of 4C (Fig. 6.16). For all divalent cation titrations, quenching of the fluorescence intensity can be observed over the emission profile (λex = 335 nm). At low concentrations (<5 μM) slight fluorescence quenching is observed with the addition of In3+ cations, however, at concentrations >5 μM fluorescence enhancement is observed. The addition of Ag+ cations has a minimal effect on the fluorescence intensity of 4C, even at concentrations >100 μM. It is noteworthy that within the emission wavelengths monitored, at the selected excitation wavelengths utilised, no additional fluorescence peaks were observed.

The titration data were plotted to represent the fluorescence quenching relative to the molar equivalent of the metal cation in the system (Fig. 6.17). This plot demonstrates the extent to which the individual metal cations effect the fluorescence intensity of 4C. These data points have been fitted with a non-linear regression function, to enhance the inherent trends, which clearly supports the premise that divalent cations have significant quenching effects on the fluorescence of 4C. In addition, these data can be utilised to provide insight into cation-calixarene interactions, through the calculation of binding constants (Kb).

217 Chapter 6

218 Chapter 6

Figure 6.16 (Left) Variation of the fluorescence spectra of 4C (40 µM) upon successive addition of 2+ 2+ 2+ 2+ + 3+ a) [UO2] , b) Pb , c) Co , d) Cu , e) Ag , and f) In at pH 6.5, (Right) quenching of the fluorescence intensity of 4C (40 µM) upon successive addition of the respective metal cations at 3+ pH 6.5; in the In plot, the y-axis is I/Io due to enhancement of the fluorescence intensity.

219 Chapter 6

Figure 6.17 Quenching of the fluorescence intensity of 4C (40 µM) based on successive addition of 3+ molar equivalents of the metal cations; in the In plot, the y-axis is I/Io due to enhancement of the fluorescence intensity.

Physical parameters for the cation-calixarene interactions were obtained from the Kb, which was determined according to a method described in the literature.[586, 587] Several different methods for binding constant determination were trialled, including solely dynamic quenching, static quenching and dynamic quenching combined, and deviations such as “sphere-of-action” models.[588] However, the method which best fit the data, was previously described by Chipman et al.,[586] and will now be detailed. This involved plotting the function

Io 1   (10) I X ]M[ where Io represents the fluorescence intensity in the absence of metal cations, and ΔI represents the relative change in fluorescence intensity at any point during the titration (Fig. 6.18). A linear regression fitting function is applied to these data and an extrapolated y-intercept is calculated, the reciprocal of which represents the fluorescence intensity of the system at infinite metal cation concentration, I∞. 220 Chapter 6

2+ Figure 6.18 Plot of Io/ΔI versus 1/[Co ], with a linear regression function fitted, and regression line extrapolated to indicate the y-intercept.

Using the calculated value for I∞, coupled with the titration data, Kb values can be derived from the following equations:

 n  I   X ]M[  log     (11)  II  K     diss 

where Kdiss represents the dissociation constant, and n represents the number of binding sites. The slope of the double logarithmic plot (Fig. 6.19) can provide valuable cation-calixarene interaction information and indicate binding ratios of metal cation to ligand. The binding constant Kb can be obtained by plotting:

 I   log   X ]Mlog[ (12)  II  

221 Chapter 6

x+ In addition, the value of log[M ], where log[(ΔI / (|I - I∞|))] = 0, gives rise to the dissociation constant (Kdiss) (Fig. 6.19). The reciprocal of Kdiss determines the Kb which allows quantitative comparison for binding affinities. The Kb and n values were obtained for the different cations through titrimetric profiling (Table 6.1).

Figure 6.19 Double-logarithmic plot for the quenching of 4C fluorescence intensity following the addition of Co2+ (2 μM to 50 μM), with a linear regression fitting function.

Table 6.1 Calculated Kb and n values of the metal cations with 4C.

Metal Cations

Parameter 2+ 2+ 2+ 2+ + 3+ [UO2] Pb Co Cu Ag In

4 -1 Binding constant, Kb (x 10 M ) 9.328 4.564 7.679 10.666 - 0.56

Binding sites, n 2.13 0.81 0.87 0.96 - 1.02

222 Chapter 6

From the information gained from the aforementioned procedure, metal cation to ligand interaction stoichiometry can be established. The number of binding sites indicated by this methodology suggests a 1:1 complexation (values within 0.8 - 1.0), between the metal ion and calixarene for the divalent cations, with the exception of the uranyl ion and the trivalent indium ion. Interestingly, the results indicate a 2:1 stoichiometry for the binding of the uranyl ion. This implies that a single molecule of 4C has two available binding sites for the uranyl ions. From these data Kb can be obtained, which provides quantitative classification for the binding affinities of the different metal cations to calixarene. The hierarchy of Kb for the metal cations trialled is as follows: 2+ 2+ 2+ 2+ 3+ + Cu > [UO2] > Co > Pb >> In >> Ag . The binding constant data (Table 6.1) clearly illustrates the preferential interaction of divalent cations with 4C, more 2+ 2+ specifically the higher binding affinity to both Cu and [UO2] . It should be noted that + the Kb value of Ag was very low, which was discounted as having any significant interaction with the calixarene. Furthermore, although the trivalent indium ions have an associated binding constant, and displays 1:1 stoichiometry, the binding constant value is considered to be low. In addition, In3+ displays a CHEF effect, whereas the divalent cations studied exhibit a CHEQ effect.

The observed changes in the fluorescence intensity of 4C, through the successive addition of metal cation species, suggests that this compound could be utilised as a metal cation sensor in aqueous media. As indicated by the Kb values, an binding affinity to divalent metal cations is preferred over mono- and trivalent cations. Calibration curves have been constructed to provide a correlation between relative fluorescence intensity change, (Io - I) / Io, and metal cation concentration (Fig. 6.20). The concept of utilising 4C as a fluorescence-based, quantitative metal cation sensor is fueled by the requirement for the cost-effective and rapid analysis of heavy metals, such as copper and cobalt, with particular interest in the detection of aqueous uranyl ions, in application areas, such as nuclear waste site remediation proficiency testing and drinking water assessment.[589, 590]

223 Chapter 6

Figure 6.20 Calibration curves for relative fluorescence intensity change versus metal cation concentration for 4C (40 μM), appropriate fitting functions and associated correlation coefficients 2+ 2+ 2+ 2+ + 3+ are described for a) [UO2] , b) Pb , c) Co , d) Cu , e) Ag , and f) In ; a red datum point indicates an outlier not utilised for calibration curve determination; y-error bars are displayed.

To assess 4C for its effectiveness as a metal cation sensor, calibration curves were generated for [Mx+] between 1 μM to 40 μM against a solution of 4C (40 µM) (Figs. 6.20a-f). As previously indicated, there is a fluorescence intensity plateau observed when the metal cation concentration exceeds the molar equivalent of 4C (Fig. 6.17), this

224 Chapter 6 suggests 1:1 binding stoichiometry between the metal ion and ligand, which correlates to the n values obtained, with the exception of the uranyl ion interaction. Interestingly, Shinkai et al. reported a “super-uranophile” which has 1:1 binding to a p-sulfonatocalix[6]arene derivative.[578] However, in relation to this study a 2:1 coordination can be rationalised by the flexible conformational arrangement of the larger ring size calix[8]arene, which could allow a metal-induced calixarene re-organisation to minimise the energy for dual coordination of metal cations, as suggested for other analogous systems.[579]

The calibration curves (Figs. 6.20a-f) further support the high binding affinity to 2+ 2+ 2+ 2+ divalent cations, [UO2] , Pb , Co , and Cu . It can be seen that for the divalent 2+ 2+ 2+ cations, [UO2] , Pb , and Co , a quadratic fitting function was the most appropriate regression curve to represent the data, and this resulted in high correlation coefficients (>0.99448), as illustrated in Figures 6.20a-c. For the Pb2+ curve, the 2 μM datum point was considered an outlier and was not utilised to form the 2nd order polynomial fit. The remaining divalent cation is Cu2+, and further it was found that an exponential fitting function best described this system (Fig. 6.20d). This could possibly be correlated to the non-specific quenching of Cu2+ ions, induced in other common fluorescent probes.[553] This also produced a high correlation coefficient (0.99641), suggesting a strong positive correlation between relative fluorescence quenching and metal cation concentration.

The low binding affinity to the monovalent alkali ions, is compounded by the inability to appropriately fit the data. It can be clearly seen that even when considering the 30 μM datum point as an outlier, only a slightly positive linear function (R2 = 0.22682) can be fitted to reflect the data (Fig. 6.20e). In addition, when considering the trivalent indium cation it appears that at concentrations >5 μM an increase in fluorescence intensity is observed, albeit at relatively small levels. However, the data can be fitted with a 2nd order polynomial function but this produces an unsatisfactory correlation coefficient (R2 = 0.98494) (Fig. 6.20f), which makes 4C unsuitable as a fluorescent sensor for this particular cation.

225 Chapter 6

Through the series of metal interaction studies with solvated 4C, it is clear that the divalent metal cations have the most significant effect on the fluorescence of the system. This could possibly be due to the metal cation inducing electron transfer (eT) or energy transfer (ET) to the calixarene,[527] which disrupts the intramolecular conformation induced by phenolic proton dissociation events at selected pH values. These metal binding events could disrupt the molecular energy minimum which allows excimer formation through the π-π stacking of the aromatic units, ultimately quenching the fluorescence.[527] This ‘extended arm’ sulfonated calix[8]arene has shown significant promise for the quantitative determination of multiple divalent metal cations in aqueous media, indicating its suitability as a water-soluble fluoroionophore.

6.4 Conclusion

The assessment of p-phenyl-sulfonato-calix[8]arene as a water-soluble fluoroionophore has been presented, and has been successfully utilised as a quantitative fluorescence-based sensor for the detection of divalent metal cations. Characterisation of the system, utilising both solution-based and solid-state techniques, has provided insight into the autofluorescent nature of the molecule, and how this process is ultimately effected by metal cation interactions. Time-based fluorescence intensity studies gave valuable insight into the solution dynamics, and solvated shelf-life, and significantly enhanced analytical protocols.

The formation of higher-order supramolecular structures, through intermolecular associations, were investigated. The formation of micelles was supported through conventional conductivity techniques, however, no direct link between higher-order structures and autofluorescence was observed. In addition, the system was not effective as a ‘drug carrier’, however, this was only trialled on Rhodamine 6G, which may not be suitable for binding to the current system. Experimental evidence indicated that partial fluorescence was linked to intramolecular hydrogen-bonding of the lower rim hydroxyls, which can significantly alter calixarene conformation and provides opportunities for excimer formation, through π-π stacking.

226 Chapter 6

The majority of current fluorescent ionophoric sensor systems are based on the more rigid calix[4]arene and its associated derivatives. This study of the higher-order calix[8]arene with water-soluble functionality provides a major shift in the approach to this sensor type. The significant conformational mobility of this ‘extended arm’ calix[8]arene provides increased dimensionality and innovation, which will fill a current gap in the sensing literature. Documentation of this system adds to the limited reports on completely water-soluble fluoroionophores, whereas many comparable systems utilise water/organic solvent mixtures. Furthermore, this system can be obtained in high yield through an uncomplicated synthesis and does not require attachment of a chromophore, due to its autofluorescence. This study provides the basis for more exploratory studies on derivatives of this compound, in which specific functional motifs can be attached to provide increased sensitivity and selectivity to target analytes.

227

Chapter 7 Conclusions and Future Studies

Chapter 7

7 Conclusions and future studies

7.1 Conclusions

The principal concept entrenched in this thesis is the development and extension of carbon nano-material-based technologies to generate novel solutions for current scientific, industrial and commercial problems, targeted at a global audience. This encompassing ideology was achieved through the application of theoretical and experimental principles, entrenched in the domains of chemistry and physics, which have been harnessed to assist in the transformation of nanoscience to nanotechnology.

With current silicon-based transistor technology rapidly approaching fabrication and material limits, there is necessity to turn to the development of other materials. SWCNTs are the frontrunners for integration into current electronics processing, and major interest has been generated, directing the ‘next generation’ devices to the realms of carbon nano-electronics. One noted current limitation is the ability to obtain sufficient quantities of nanotubes that exhibit the same electronic properties. One of the aims of this research is to develop novel purification methodologies for the post-growth treatment of CNTs, with specific focus on nano-electronic applications. In order to meet this aim a novel supramolecular-driven pathway for the realisation of diameter-selective SWCNT separation was developed and the tunable enrichment of SWCNTs with semiconducting, or metallic, electronic properties has been achieved. SWCNTs have been successfully solubilised using water-soluble p-phosphonated calix[n]arenes (n = 4, 6, 8) and ‘extended arm’ upper rim functionalised (benzyl, phenyl) p-sulfonated calix[8]arenes. This thesis describes a comprehensive, comparative study into the use of p-phosphonated calixarene and p-sulfonated calix[8]arene systems, to produce not only diameter-selective nanotube separation, and increased amounts of pristine nanotubes but also to allow processing in aqueous media. A detailed Raman spectroscopy study between the different water-soluble calixarene derivatives has demonstrated, the strong binding affinity between the SWCNT surface and the ‘extended arm’ p-sulfonated calix[8]arenes. Analysis of ID/IG ratios, demonstrated that in the majority of these systems, an increase in the amounts of pristine SWCNTs in the supernatant, with elemental analysis revealing little or no residual growth catalyst left in the supernatant. It was explicitly shown that a calixarene sheath, if formed around individual SWCNTs, facilitates water-solubilisation processes. 229 Chapter 7

In addition, NT-FETs were successfully fabricated from the resulting supernatants, supporting the use of this protocol and providing a novel methodology for the advancement of SWCNT post-growth purification strategies. The implications of this research are potentially far reaching, beyond nano-electronics, and into many other multi-disciplinary domains, such as nano-toxicology, green chemistry and chemical sensor technology.

In the current global environment there is a heightened level of public and governmental disquiet due to the reality of impending terrorist attack. This is compounded by the inherent ease of manufacture and the effectiveness of specific CWAs used in small-scale terrorist operations. Phosgene, COCl2, is a CWA which is the forerunner of modern chemical warfare, and is commercially produced as well as being readily prepared from common chemicals. It is a colourless gas, with toxic effects well below its odour threshold of 0.4 ppm. In combating the terrorism associated with phosgene, and other CWAs, new detection technologies are required for the purpose of monitoring accidental or purposeful releases, personnel safety, operational intelligence, and warfare tactics. The specific aims of this research include the design, fabrication and testing of carbon nano-materials as active sensing matrices of gas and organic vapour sensors, specifically targeting chemicals of national security interest. In order to meet these aims a novel nanotechnology method is reported for the detection of phosgene, involving a drop-casting technique to construct a chemiresistor device based on SWCNTs. The entire process from design and development, through to system verification and analyte testing is described. Phosgene can now be detected over common organic solvents and other halogenated systems, at 24 ppm limits. Sensing G-series nerve agent simulants for sarin and soman are also established using the same device, with detection limits of 266 ppb and 152 ppb, respectively. A comparative study, illustrating the effectiveness of single IDE sensor geometry against multiple-array systems is also detailed.

This thesis highlights the sheer versatility of carbon-based materials and their importance in current and future technologies. The ability to control and manipulate structures at the molecular level is appealing especially for the development of new materials. This can be achieved through unique nano-scale architectures, forming the basis of areas such as crystal engineering. A primary aim of this research is to extend the field of all-carbon architecture to the fabrication of novel structures, through various 230 Chapter 7 processes. A novel supramolecular system is described where toluene solutions of p-tBu-calix[5]arene and C60 result in a 1:1 complex, (C60)∩(p-tBu-calix[5]arene), which precipitates as nano-fibres. The principal structural unit of this complex is based on a host-guest ball-and-socket nanostructure of the two components, with an extended structure comprised of zigzag/helical arrays of fullerenes, predicted through PXRD data coupled with molecular modelling. Under argon, at temperatures >309 °C, the fibres undergo selective volatilisation of the calixarenes producing C60-core nanostructures encapsulated in a graphitic material sheath, leading to a dramatic increase in surface area. Above 650 °C the material exhibits an ohmic conductance response, due to the encapsulation process. In addition, the results indicate the potential of this novel complex as an autophagy inductor in a cervical cancer cell type, and as a chemo-sensitiser towards a drug-resistant breast cancer cell type.

Other all-carbon architectures are described in this thesis, with the formation of rings of helical SWCNT bundles through post-growth SWCNT modifications, and a variety of fibrous all-carbon structures, most notably novel square-geometry CNFs, using C-CVD synthesis strategies. These materials were characterised through electron microscopies, TEM and SEM, to reveal the underlying nature of the materials. These analyses produced structural information to support suggested growth mechanisms. The development of different structural constraints in SWCNTs could enhance the chemical, physical, and/or mechanical properties of the system. On the other hand, the square-geometry of the CNFs produced could potentially form the basis of novel composite structures. The ability to extend the field of all-carbon architecture has clearly been demonstrated.

The current requirement for entirely water-soluble fluorescent sensors is routinely documented in the literature. A final objective of the research was to address this issue, and to contribute to the area of water-soluble fluoroionophores for the future advancement of this field. In order to achieve this, the autofluorescence properties of p-phenyl-sulfonated calix[8]arene are characterised and this water-soluble cavitand is surveyed as a metal cation sensor candidate. It is suggested that the autofluorescence of the cavitand arises predominately from intramolecular interactions, with possible excimer formation. This particular system was found to exhibit a CHEQ response when 2+ 2+ 2+ 2+ exposed to divalent metal cations, and interactions with [UO2] , Pb , Co , and Cu 231 Chapter 7 ions are discussed in detail. The system is characterised through fluorescence spectroscopy to produce sensor calibration data, solution degradation characteristics, and pH sensitivity. In addition, quantitative binding affinity data has been established through binding constant determination, which also provided insight to cation coordination stoichiometry. A detailed analysis utilising an array of solid-state, and solution-based, characterisation techniques was undertaken to provide insight into the inherent dynamics of the system in order to assess the organisation of higher-order structures such as micelles, vesicles, or capsules. In addition, the ‘extended arm’ calixarene was assessed for suitability as a ‘small molecule’ drug-carrier.

Overall, the scientific endeavours described in this thesis have significantly contributed to the advancement of knowledge in the areas of all-carbon nano-material generation and processing, CNT-based gas and vapour sensors, and contributed a novel water-soluble fluoroionophore system. The innovative studies described in this thesis have made significant contributions to a diverse range of multi-disciplinary areas and, thus far, have resulted in two important publications in international peer-reviewed journals, and an Australian provisional patent.

7.2 Future studies

The research described in this thesis has been classified for the advancement of four key areas; SWCNT post-growth processing towards diameter-selectivity; CWA detection through the development of SWCNT-based chemiresistor sensors; fabrication of functional all-carbon architectures; and the development of novel water-soluble fluoroionophores. Building on the results obtained from the aforementioned investigations, future research directions, including potential application areas, are suggested in the following section.

7.2.1 SWCNT post-growth processing towards diameter-selectivity

The thesis describes the solubilisation of SWCNTs into aqueous media coupled with tunable diameter-selectivity, resulting in enrichment of semiconducting or metallic SWCNTs. This study presented Raman spectroscopic evidence of system dependent SWCNT surface binding interactions, which suggest a different stacking arrangement

232 Chapter 7 may be possible depending on the cavitand structure. Future studies should include the use of computational chemistry simulations to establish the different calixarene packing arrangements. In addition, a FET was fabricated utilising the supernatant residue, and the effectiveness of the post-growth protocol was demonstrated. Future studies should include in-depth electrical characterisation of the resulting residues with, and without, the calixarene sheath.

Future studies should continue to develop the primary investigations into the possibility of directing CNT architecture utilising host-guest principles. The outer sheathing on individual SWCNTs comprises cavitand structures, and if the appropriate conditions and required host are met, it should be possible to form self-assembled arrays of SWCNTs that could have significant applications in areas such as carbon nano-electronic, device technology, and tissue engineering. In addition, this principle can be extended further through the utilisation of the phosphonic and sulfonic acid head groups, as sites for metal nanoparticle nucleation and growth. This templated nanoparticle growth, could lead to applications in sensor technology and possibly metal nanoparticle interconnects for CNTs in ‘next generation’ computing.

Further investigation is necessary to establish whether the surface curvature of the carbon nano-material dictates successful solubilisation. Further studies into nano-materials such as the fullerenes and graphene are required to add to the validity of concepts suggested in this thesis. Furthermore, the resulting supernatant residues should be trialled for their gas sensing capacities.

7.2.2 CWA detection through the development of SWCNT-based chemiresistor sensor technology

A major aim of this thesis is focused on the design, fabrication and testing of SWCNT-based chemiresistor sensors, with particular attention paid to the detection effectiveness of CWAs. This area of study has enormous scope for further investigations, where individually tailored systems can be fabricated depending on the analyte of interest, utilising CNTs as key components.

233 Chapter 7

From a device platform perspective, various electrochemical device geometries and sensing principles have been described in relation to CNT-based gas and vapour sensors. This thesis specifically explores a chemiresistor sensor geometry which exploits conductance mechanisms between the analyte and CNT matrix. Future studies should encompass designing CNT-based devices that incorporate the advantages of multiple sensing mechanisms, such as simultaneously monitoring conductance and capacitance mechanisms, coupled with measurement of the frequency shift with surface acoustic wave sensors and measurement of the analyte breakdown voltage with ionisation sensors. The utilisation of multiple mechanisms within one sensor has been explored previously (Section 1.4.1.1), with the conductance and capacitance mechanism monitored. The application of this concept urgently requires extension to coupling additional sensor geometries to provide a holistic overview of an analyte system and thus providing additional separation strategies.

At a fundamental level much conjecture exists in the literature regarding the mechanism of action on analyte binding. Future studies need to be directed to this area to provide unequivocal evidence to support or reject current hypothesised mechanisms. With greater understanding of the underlying mechanisms, analyte-specific sensors will become more achievable.

The concept of utilising multiple-arrays, consisting of the same CNT-based sensing material, to obtain lower sensor detection limits, was described in this thesis. Future studies which focus on the sensor configuration at the design level, should be directed towards multiple-arrays of sensing elements. Multiple-arrays of different CNT-based materials have the ability to add dimensionality, through PCA of the data to increase the differentiation capabilities when dealing with multi-component systems. Ultimately, the developed sensors are aimed at ‘real-world’ applications, but, at the moment, these sensors have only been assessed under controlled laboratory conditions. To avoid potential interferences, the use of multiple-arrays is a fundamental step in the advancement of CNT-based gas and vapour sensors.

Future studies are required to increase the sensitivity and specificity of the developed SWCNT chemiresistor sensors, as it was indicated that responses from common organic

234 Chapter 7 solvents had the potential to mask the detection, and induce false-positives, for the G-series nerve agent simulants analysed. To an extent, this was avoided in terms of phosgene detection, as added data dimensionality was achieved through the inherent phosgene interaction mechanism. Possible strategies for increasing sensitivity and selectivity of these devices include, but are not limited to, nanoparticle decoration, chemo-selective polymers, and functional group attachment. Of these functionalisation strategies, it is recommended that future studies should be directed towards the developing area of chemo-selective polymers, in association with CNT-based sensors.

This thesis describes the selective enrichment of either metallic or semiconducting SWCNTs in aqueous media and, in addition, describes a fabricated FET. Future studies are required into the assessment of these systems in terms of their gas and vapour sensing capacities. The fundamental concept that semiconducting SWCNTs are the majority contributors to sensor response has been demonstrated in this thesis. With this in mind, the semiconducting nanotube enriched calixarene systems show promise for the fabrication of higher-sensitivity SWCNT chemiresistor sensors. Furthermore, the inherent calixarene sheathing around individual SWCNTs may produce a mechanism to increase sensitivity and selectivity towards particular classes of compounds.

The multi-phase development of a suitable, yet versatile, chemical delivery system for sensor testing has been described. Future studies should endeavour to create additional functionality with this system. For example, additional instrumentation to measure analyte concentration levels at the test chamber exhaust port, or simultaneous monitoring of the CNT matrix through Raman spectroscopy, to obtain additional ‘real-time’ sensor data. In addition, although electronic noise present in the testing system was significantly reduced, future studies should focus on further lowering the noise levels, as this will only benefit future sensor studies and further reduce device detection limits. Noise reduction strategies include, carrying out analyses in a facility shielded from possible noise sources, isolation and filtering of mains power, and acquiring specialised precision electrical characterisation equipment. Furthermore, the more practical monitoring of multiple-array devices is required; the option currently exists, yet is limited by the electrical characterisation system.

235 Chapter 7

The future direction of carbon nano-material gas and vapour sensors is rapidly approaching integration into miniaturised analytical instrumentation. An aforementioned pioneering development in this area, is the fabrication of an on-chip micro-gas chromatography instrument coupled to a functionalised SWCNT sensor. The realisation of this critical concept will open up possibilities for SWCNT-integrated instrument miniaturisation studies, which will have significant commercial implications. Another current growth area involves the utilisation of graphene in place of CNTs. Graphene-based sensors have already shown decreased inherent noise levels, which is highly beneficial in chemical sensing. It is foreseen that once improved graphene synthesis methodologies become established scientific publications in this area will grow exponentially in years to come.

The future of CNT-based sensors requires the evolution from the laboratory to the ‘real world’, especially in the case of CWA detection. Future studies should be focused on the appropriate packaging of the device to allow for remote deployable sensors. A target area for this application could be the incorporation of the CNT-based sensors into nodal networks, or even ‘smart dust’ devices, scattered in many different locations so that signals can be transmitted to a central communications unit where sensor data is utilised for chemical mapping. This has significant implication in ‘battle-field’ forensics, and can be considered a highly effective level of ground intelligence. Furthermore, incorporation of solar panel rechargeable cells could be utilised to power these devices, which have extremely small physical dimensions and low-power requirements.

7.2.3 Fabrication of functional all-carbon architectures

The versatility of carbon is inherent in this thesis and it is apparent that control over the arrangement of carbon entities at the molecular level can produce novel functional materials. The formation and proposed self-assembly of (C60)∩(p-tBu-calix[5]arene) has led, not only to the innovative C60 arrangement but also to the development of high temperature all-carbon analogues through selective volatilisation.

236 Chapter 7

An interesting application described in this thesis is the ability for the parent complex to induce autophagy and chemo-sensitivity in human cervical carcinoma cells and drug-resistant human breast carcinoma cells, respectively. Current studies are being directed toward the assessment of the high temperature analogues. Future studies should focus on the relationship between crystal packing structure, such as the fcc or hcp structure of fullerene C60, in relation to its chemotherapeutic activity. In addition, water-solubilisation of C60 utilising p-phosphonated and p-sulfonated calixarenes should be thoroughly investigated, with any successful system assessed for autophagy induction of tumour cells. From a biological perspective, assessment of cytotoxicity of the C60 complex is required, accompanied by a significant study on a wide variety of different carcinoma cells, in combination with a suite of chemotherapeutic agents, to facilitate chemo-sensitisation studies. Furthermore, attempts should be made to increase the activity of the system coupled with interaction mechanism studies through computation chemistry.

The premise of post-growth structural re-organisation in all-carbon materials, more specifically, nano-materials is extended to SWCNTs. The inherent physical and chemical properties of SWCNTs, imposes difficulties for the development of new structural arrangements. A chemical etching and vacuum annealing strategy was utilised to produce rings of helical SWCNT bundles, albeit in low yield. Future studies should be directed towards understanding the role of vacuum annealing in ring formation, and to develop strategies to increase synthetic yield, with greater ring diameter control. An application that was unsuccessfully explored is the ability to endohedrally functionalise SWCNTs with fullerene C60 and subsequently form seamless rings, for the development of a C60-nanotube high frequency oscillator. Investigations into these systems are highly topical and are noteworthy future development areas. Furthermore, the significant increase in SWCNT surface strain should be investigated for gas and vapour sensing capabilities, as this induced strain could lead to increased response mechanisms.

Synthesis of an array of carbon-based materials was achieved through C-CVD processes, utilising acetylene feedstock and relatively low-temperature growth conditions. Interestingly, novel, square-geometry CNFs were observed in the growth material, albeit in low yield. In order for the inherent properties of these unique CNFs 237 Chapter 7 to be characterised, future studies should be undertaken to increase yield, using Raman spectroscopy and electrical probe measurements. In addition, computation studies should be carried out into the mechanism by which a transition from the growth of cylindrical CNF layers to square-layer growth occurs. If increased yields can be achieved, then stress and strain studies need to be carried out for applications in structural materials and composite reinforcement.

7.2.4 Development of novel water-soluble fluoroionophores

This thesis illustrates the fundamental investigations into an autofluorescent calixarene, p-phenyl-sulfonato-calix[8]arene, with respect to its application as a water-soluble fluoroionophore. It is clear that this particular compound shows promise in the quantitation of low concentration divalent metal cations in completely aqueous systems. Although its utilisation as a fluoroionophore was demonstrated, it lacked specificity towards an individual divalent cation. Consequently, future studies should be directed at structural modification, through the attachment of ion-specific functional groups or pendant arms, to promote directed ion coordination. Indeed, if fluorescence originates from excimer formation through intramolecular π-π stacking of aromatic moieties, it may be possible to design a system in which the analyte induces molecular re-organisation is such as way as to completely switch off the fluorescence properties. This would significantly reduce measureable detection limits and provide an on/off state molecular beacon for binding events. Furthermore, system modifications to target and exploit anion binding need to be established.

It is important to obtain greater understanding of the molecular foundations of not only the autofluorescence properties but the mechanisms for cation interaction resulting in fluorescence quenching. Detailed computational chemistry studies through a variety of energy minimising parameters will be essential for a more in-depth understanding of these mechanisms, more specifically in relation to the deprotonation of lower rim hydroxyl functionalities. In addition, computational methods could be utilised to predict coordination interactions for analogous systems relating to the principal structure of p-phenyl-sulfonato-calix[8]arene.

238 Chapter 7

The autofluorescence properties of this calixarene coupled with the inherent water-solubility, imparted by the sulfonic acid functionalities, paves the way for applications in biological systems. This compound, or future derivatives, should be trialled for applications, such as bio-imaging, through fluorescence tracking and molecular binding studies. These autofluorescence properties could be harnessed to track the biological pathways of these water-soluble systems. This, in conjunction with cytotoxicity studies, would provide valuable information into inter- and intra-cellular transportation, and could provide a means of transporting molecular cargo to cellular entities while providing a tracking mechanism. Furthermore, this water-soluble fluorescent cavitand should be investigated for molecular binding studies, in a biological context, to see the effects on structures such as G-quaduplex DNA and tumor cell membranes, with the possibility of site-specific drug delivery.

Further investigation is warranted into the self-assembly of higher-order supramolecular structures from p-phenyl-sulfonato-calix[8]arene. The analyses described in this thesis did not unequivocally support or reject higher-order assemblies in solution. Consequently, studies should be directed at lower rim functionalisation strategies to induce greater molecular amphiphilicity and possibly lead to higher-order structures, such as bilayers, vesicles, and capsule formation. In this context, the concept of utilising these systems as drug carriers can be explored, with environmental triggers, such as pH, for drug targeting and release.

Cavitand interactions with organic molecules has been discussed in detail in relation to fullerene interactions. The interactions of p-phenyl-sulfonato-calix[8]arene with small organic molecules is an area that requires attention. Initial studies indicate that fluorescence enhancement is observed for acetonitrile. Extensive studies should be undertaken into small organic molecule binding effects on p-phenyl-sulfonato- calix[8]arene fluorescence, and system modification to induce chemo-selectivity. A realistic application area could be a sensor for petroleum deposit exploration in high conductivity media; such as seawater.

239

References

References

8 References

[1] M. S. Dresselhaus, G. Dresselhaus, P. C. Eklund, R. Saito, M. Endo, in Carbon nanotubes : preparation and properties (Ed.: T. W. Ebbesen), CRC Press, Boca Raton, Florida, 1997, pp. 1-34. [2] W. N. Reynolds, Physical properties of graphite, Elsevier Publishing Co. Ltd., Amsterdam, 1968. [3] B. T. Kelly, Physics of graphite, Applied Science, London, 1981.

[4] D. Koruga, S. Hameroff, J. Withers, R. Loutfy, M. Sundareshan, Fullerene C60 : history, physics, nanobiology, nanotechnology, North-Holland, Amsterdam, 1993. [5] M. Paradise, T. Goswami, Mater. Des. 2007, 28, 1477-1489. [6] M. A. Prelas, G. Popovici, L. K. Bigelow, Handbook of industrial diamonds and diamond films, Marcel Dekker, New York, 1998. [7] P. S. Bailey, C. A. Bailey, Organic chemistry : a brief survey of concepts and applications, 6th ed., Prentice Hall, New Jersey, 2000. [8] J. E. Baggot, Perfect symmetry : the accidental discovery of buckminsterfullerene, Oxford University Press, Oxford, 1994. [9] H. Aldersey-Williams, The most beautiful molecule : the discovery of the buckyball, John Wiley & Sons, Inc., New York, 1995. [10] H. Kroto, Rev. Mod. Phys. 1997, 69, 703-722. [11] R. E. Smalley, Rev. Mod. Phys. 1997, 69, 723-730. [12] H. W. Kroto, J. R. Heath, S. C. O'Brien, R. F. Curl, R. E. Smalley, Nature 1985, 318, 162-163. [13] R. F. Curl, Rev. Mod. Phys. 1997, 69, 691-702. [14] W. Krätschmer, L. D. Lamb, K. Fostiropoulos, D. R. Huffman, Nature 1990, 347, 354-358. [15] Accelrys, MS Visualizer, Accelrys Software, Inc., San Diego, 2006. [16] J. H. Hafner, A. Thess, P. Nikolaev, A. G. Rinzler, D. T. Colber, R. E. Smalley, in The chemical physics of fullerenes 10 (and 5) years later : the far-reaching

impact of the discovery of C60 (Ed.: W. Andreoni), Kluwer Academic Publishers, Netherlands, 1996, pp. 19-26. [17] Frost & Sullivan, Carbon nanotubes - road to commercialization, Technical Insights, Inc., 2007. [18] E. Stach, Nat. Nanotechnol. 2008, 3, 586-587.

241 References

[19] S. Iijima, Nature 1991, 354, 56-58. [20] S. Iijima, T. Ichihashi, Nature 1993, 363, 603-605. [21] D. S. Bethune, C. H. Kiang, M. S. de Vries, G. Gorman, R. Savoy, J. Vazquez, R. Beyers, Nature 1993, 363, 605-607. [22] T. W. Ebbesen, P. M. Ajayan, Nature 1992, 358, 220-222. [23] P. M. Ajayan, J. M. Lambert, P. Bernier, L. Barbedette, C. Colliex, J. M. Planeix, Chem. Phys. Lett. 1993, 215, 509-517. [24] X. Lin, X. K. Wang, V. P. Dravid, R. P. H. Chang, J. B. Ketterson, Appl. Phys. Lett. 1994, 64, 181-183. [25] S. Seraphin, D. Zhou, Appl. Phys. Lett. 1994, 64, 2087-2089. [26] S. Seraphin, J. Electrochem. Soc. 1995, 142, 290-297. [27] C.-H. Kiang, W. A. Goddard, R. Beyers, D. S. Bethune, Carbon 1995, 33, 903- 914. [28] Y. Saito, K. Kawabata, M. Okuda, J. Phys. Chem. 1995, 99, 16076-16079. [29] X. Zhao, M. Ohkohchi, M. Wang, S. Iijima, T. Ichihashi, Y. Ando, Carbon 1997, 35, 775-781. [30] C. Journet, W. K. Maser, P. Bernier, A. Loiseau, M. L. de La Chapelle, S. Lefrant, P. Deniard, R. Lee, J. E. Fischer, Nature 1997, 388, 756-758. [31] Z. Shi, Y. Lian, X. Zhou, Z. Gu, Y. Zhang, S. Iijima, L. Zhou, K. T. Yue, S. Zhang, Carbon 1999, 37, 1449-1453. [32] Y. Ando, X. Zhao, K. Hirahara, K. Suenaga, S. Bandow, S. Iijima, Chem. Phys. Lett. 2000, 323, 580-585. [33] S. Farhat, M. L. de La Chapelle, A. Loiseau, C. D. Scott, S. Lefrant, C. Journet, P. Bernier, J. Chem. Phys. 2001, 115, 6752-6759. [34] H. Huang, H. Kajiura, S. Tsutsui, Y. Hirano, M. Miyakoshi, A. Yamada, M. Ata, Chem. Phys. Lett. 2001, 343, 7-14. [35] H. Huang, J. Marie, H. Kajiura, M. Ata, Nano Lett. 2002, 2, 1117-1119. [36] T. Zhao, Y. Liu, Carbon 2004, 42, 2765-2768. [37] Z. Li, P. Liu, B. Zhao, H. Wang, Y. Zhang, Carbon 2008, 46, 1819-1822. [38] V. V. Grebenyukov, E. D. Obraztsova, A. S. Pozharov, N. R. Arutyunyan, A. A. Romeikov, I. A. Kozyrev, Fullerenes, Nanotubes, Carbon Nanostruct. 2008, 16, 330-334. [39] M. Keidar, I. Levchenko, T. Arbel, M. Alexander, A. M. Waas, K. K. Ostrikov, J. Appl. Phys. 2008, 103, 094318. [40] Z. Li, L. Wei, Y. Zhang, Appl. Surf. Sci. 2008, 254, 5247-5251.

242 References

[41] T. Guo, P. Nikolaev, A. Thess, D. T. Colbert, R. E. Smalley, Chem. Phys. Lett. 1995, 243, 49-54. [42] T. Guo, P. Nikolaev, A. G. Rinzler, D. Tomanek, D. T. Colbert, R. E. Smalley, J. Phys. Chem. 1995, 99, 10694-10697. [43] A. Thess, R. Lee, P. Nikolaev, H. Dai, P. Petit, J. Robert, C. Xu, Y. H. Lee, S. G. Kim, A. G. Rinzler, D. T. Colbert, G. E. Scuseria, D. Tomanek, J. E. Fischer, R. E. Smalley, Science 1996, 273, 483-487. [44] M. Yudasaka, T. Komatsu, T. Ichihashi, S. Iijima, Chem. Phys. Lett. 1997, 278, 102-106. [45] M. Yudasaka, T. Komatsu, T. Ichihashi, Y. Achiba, S. Iijima, J. Phys. Chem. B 1998, 102, 4892-4896. [46] W. K. Maser, E. Muñoz, A. M. Benito, M. T. Martínez, G. F. de la Fuente, Y. Maniette, E. Anglaret, J. L. Sauvajol, Chem. Phys. Lett. 1998, 292, 587-593. [47] A. P. Bolshakov, S. A. Uglov, A. V. Saveliev, V. I. Konov, A. A. Gorbunov, W. Pompe, A. Graff, Diamond Relat. Mater. 2002, 11, 927-930. [48] P. C. Eklund, B. K. Pradhan, U. J. Kim, Q. Xiong, J. E. Fischer, A. D. Friedman, B. C. Holloway, K. Jordan, M. W. Smith, Nano Lett. 2002, 2, 561-566. [49] D. Nishide, H. Kataura, S. Suzuki, K. Tsukagoshi, Y. Aoyagi, Y. Achiba, Chem. Phys. Lett. 2003, 372, 45-50. [50] M. Kusaba, Y. Tsunawaki, Thin Solid Films 2006, 506-507, 255-258. [51] M. A. El Khakani, J. H. Yi, B. Aïssa, Diamond Relat. Mater. 2006, 15, 1064- 1069. [52] J. H. Yi, B. Aïssa, M. A. El Khakani, J. Nanosci. Nanotechnol. 2007, 7, 3394- 3399. [53] B. J. Landi, R. P. Raffaelle, J. Nanosci. Nanotechnol. 2007, 7, 883-890. [54] Y.-T. Wu, H.-C. Su, C.-M. Tsai, K.-L. Liu, G.-D. Chen, R.-H. Huang, T.-R. Yew, Appl. Phys. Lett. 2008, 93, 023108. [55] G. Maohui, S. Klaus, Appl. Phys. Lett. 1994, 64, 710-711. [56] W. Z. Li, S. S. Xie, L. X. Qian, B. H. Chang, B. S. Zou, W. Y. Zhou, R. A. Zhao, G. Wang, Science 1996, 274, 1701-1703. [57] M. Terrones, N. Grobert, J. Olivares, J. P. Zhang, H. Terrones, K. Kordatos, W. K. Hsu, J. P. Hare, P. D. Townsend, K. Prassides, A. K. Cheetham, H. W. Kroto, D. R. M. Walton, Nature 1997, 388, 52-55.

243 References

[58] A. Fonseca, K. Hernadi, P. Piedigrosso, J. F. Colomer, K. Mukhopadhyay, R. Doome, S. Lazarescu, L. P. Biro, P. Lambin, P. A. Thiry, D. Bernaerts, J. B. Nagy, Appl. Phys. A 1998, 67, 11-22. [59] I. Willems, Z. Kónya, J. F. Colomer, G. Van Tendeloo, N. Nagaraju, A. Fonseca, J. B. Nagy, Chem. Phys. Lett. 2000, 317, 71-76. [60] J. F. Colomer, C. Stephan, S. Lefrant, G. Van Tendeloo, I. Willems, Z. Kónya, A. Fonseca, C. Laurent, J. B. Nagy, Chem. Phys. Lett. 2000, 317, 83-89. [61] L. An, J. M. Owens, L. E. McNeil, J. Liu, J. Am. Chem. Soc. 2002, 124, 13688- 13689. [62] J. Geng, C. Singh, D. S. Shephard, M. S. P. Shaffer, B. F. G. Johnson, A. H. Windle, Chem. Commun. 2002, 2666-2667. [63] L. X. Zheng, M. J. O'Connell, S. K. Doorn, X. Z. Liao, Y. H. Zhao, E. A. Akhadov, M. A. Hoffbauer, B. J. Roop, Q. X. Jia, R. C. Dye, D. E. Peterson, S. M. Huang, J. Liu, Y. T. Zhu, Nat. Mater. 2004, 3, 673-676. [64] K. Hata, D. N. Futaba, K. Mizuno, T. Namai, M. Yumura, S. Iijima, Science 2004, 306, 1362-1364. [65] G.-H. Jeong, A. Yamazaki, S. Suzuki, H. Yoshimura, Y. Kobayashi, Y. Homma, J. Am. Chem. Soc. 2005, 127, 8238-8239. [66] H. J. Yang, H. J. Song, H. J. Shin, H. C. Choi, Langmuir 2005, 21, 9098-9102. [67] L.-P. Zhou, K. Ohta, K. Kuroda, N. Lei, K. Matsuishi, L. Gao, T. Matsumoto, J. Nakamura, J. Phys. Chem. B 2005, 109, 4439-4447. [68] P. Ramesh, N. Kishi, T. Sugai, H. Shinohara, J. Phys. Chem. B 2006, 110, 130- 135. [69] S. Bhaviripudi, E. Mile, S. A. Steiner, A. T. Zare, M. S. Dresselhaus, A. M. Belcher, J. Kong, J. Am. Chem. Soc. 2007, 129, 1516-1517. [70] Y. Zhang, W. Zhou, Z. Jin, L. Ding, Z. Zhang, X. Liang, Y. Li, Chem. Mater. 2008, 20, 7521-7525. [71] T. Kato, G.-H. Jeong, T. Hirata, R. Hatakeyama, K. Tohji, K. Motomiya, Chem. Phys. Lett. 2003, 381, 422-426. [72] G. F. Zhong, T. Iwasaki, K. Honda, Y. Furukawa, I. Ohdomari, H. Kawarada, Chem. Vap. Deposition 2005, 11, 127-130. [73] M. R. Maschmann, P. B. Amama, A. Goyal, Z. Iqbal, R. Gat, T. S. Fisher, Carbon 2006, 44, 10-18. [74] E. J. Bae, Y.-S. Min, D. Kang, J.-H. Ko, W. Park, Chem. Mater. 2005, 17, 5141- 5145.

244 References

[75] W. H. Wang, Y. R. Peng, C. T. Kuo, Diamond Relat. Mater. 2005, 14, 1906- 1910. [76] T. Iwasaki, G. Zhong, T. Aikawa, T. Yoshida, H. Kawarada, J. Phys. Chem. B 2005, 109, 19556-19559. [77] M. R. Maschmann, A. D. Franklin, P. B. Amama, D. N. Zakharov, E. A. Stach, T. D. Sands, T. S. Fisher, Nanotechnology 2006, 17, 3925-3929. [78] T. Kato, R. Hatakeyama, Chem. Vap. Deposition 2006, 12, 345-352. [79] P. B. Amama, M. R. Maschmann, T. S. Fisher, T. D. Sands, J. Phys. Chem. B 2006, 110, 10636-10644. [80] E. J. Bae, Y.-S. Min, U. Kim, W. Park, Nanotechnology 2007, 18, 015601. [81] C.-H. Weng, C.-S. Yang, H. Lin, C.-H. Tsai, K.-C. Leou, J. Nanosci. Nanotechnol. 2008, 8, 2526-2533. [82] D. Nishide, H. Kataura, S. Suzuki, S. Okubo, Y. Achiba, Chem. Phys. Lett. 2004, 392, 309-313. [83] J. Shi, Y. F. Lu, H. Wang, K. J. Yi, Y. S. Lin, R. Zhang, S. H. Liou, Nanotechnology 2006, 17, 3822-3826. [84] M. H. Rümmeli, F. Schäffel, M. Löffler, C. Kramberger, D. Adebimpe, T. Gemming, P. Ayala, B. Rellinghaus, L. Schultz, B. Büchner, T. Pichler, Phys. Status Solidi A 2008, 205, 1386-1390. [85] S. Maruyama, R. Kojima, Y. Miyauchi, S. Chiashi, M. Kohno, Chem. Phys. Lett. 2002, 360, 229-234. [86] H. E. Unalan, M. Chhowalla, Nanotechnology 2005, 16, 2153-2163. [87] T. Izák, T. Danis, M. Veselý, M. Marton, M. Michalka, Vacuum 2007, 82, 134- 137. [88] H. Sugime, S. Noda, S. Maruyama, Y. Yamaguchi, Carbon 2009, 47, 234-241. [89] P. Nikolaev, M. J. Bronikowski, R. K. Bradley, F. Rohmund, D. T. Colber, K. A. Smith, R. E. Smalley, Chem. Phys. Lett. 1999, 313, 91-97. [90] M. J. Bronikowski, P. A. Willis, D. T. Colbert, K. A. Smith, R. E. Smalley, J. Vac. Sci. Technol. A 2001, 19, 1800-1805. [91] W. E. Alvarez, F. Pompeo, J. E. Herrera, L. Balzano, D. E. Resasco, Chem. Mater. 2002, 14, 1853-1858. [92] N. Pavel, J. Nanosci. Nanotechnol. 2004, 4, 307-316. [93] R. Philippe, A. Morançais, M. Corrias, B. Caussat, Y. Kihn, P. Kalck, D. Plee, P. Gaillard, D. Bernard, P. Serp, Chem. Vap. Deposition 2007, 13, 447-457.

245 References

[94] Q. Liu, W. Ren, Z.-G. Chen, D.-W. Wang, B. Liu, B. Yu, F. Li, H. Cong, H.-M. Cheng, ACS Nano 2008, 2, 1722-1728. [95] C. Berger, Z. Song, X. Li, X. Wu, N. Brown, C. Naud, D. Mayou, T. Li, J. Hass, A. N. Marchenkov, E. H. Conrad, P. N. First, W. A. de Heer, Science 2006, 312, 1191-1196. [96] T. Ohta, F. E. Gabaly, A. Bostwick, J. L. McChesney, K. V. Emtsev, A. K. Schmid, T. Seyller, K. Horn, E. Rotenberg, New J. Phys. 2008, 10, 023034. [97] S. Marchini, S. Gunther, J. Wintterlin, Phys. Rev. B: Condens. Matter 2007, 76, 075429. [98] A. L. V. de Parga, F. Calleja, B. Borca, J. M. C. G. Passeggi, J. J. Hinarejos, F. Guinea, R. Miranda, Phys. Rev. Lett. 2008, 100, 056807. [99] P. W. Sutter, J.-I. Flege, E. A. Sutter, Nat. Mater. 2008, 7, 406-411. [100] K. S. Novoselov, D. Jiang, F. Schedin, T. J. Booth, V. V. Khotkevich, S. V. Morozov, A. K. Geim, Proc. Nat. Acad. Sci. U.S.A. 2005, 102, 10451-10453. [101] K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, M. I. Katsnelson, I. V. Grigorieva, S. V. Dubonos, A. A. Firsov, Nature 2005, 438, 197-200. [102] Y. Zhang, Y.-W. Tan, H. L. Stormer, P. Kim, Nature 2005, 438, 201-204. [103] F. Schedin, A. K. Geim, S. V. Morozov, E. W. Hill, P. Blake, M. I. Katsnelson, K. S. Novoselov, Nat. Mater. 2007, 6, 652-655. [104] J. A. Robinson, E. S. Snow, S. C. Badescu, T. L. Reinecke, F. K. Perkins, Nano Lett. 2006, 6, 1747-1751. [105] S. Stankovich, D. A. Dikin, R. D. Piner, K. A. Kohlhaas, A. Kleinhammes, Y. Jia, Y. Wu, S. T. Nguyen, R. S. Ruoff, Carbon 2007, 45, 1558-1565. [106] Y. Hernandez, V. Nicolosi, M. Lotya, F. M. Blighe, Z. Sun, S. De, I. T. McGovern, B. Holland, M. Byrne, Y. K. Gun'Ko, J. J. Boland, P. Niraj, G. Duesberg, S. Krishnamurthy, R. Goodhue, J. Hutchison, V. Scardaci, A. C. Ferrari, J. N. Coleman, Nat. Nanotechnol. 2008, 3, 563-568. [107] O. V. Kharissova, B. I. Kharisov, Open Inorg. Chem. J. 2008, 2, 39-49. [108] D. Li, R. B. Kaner, Science 2008, 320, 1170-1171. [109] R. M. Westervelt, Science 2008, 320, 324-325. [110] J. Han, in Carbon nanotubes : science and application (Ed.: M. Meyyappan), CRC Press, 2005, pp. 1-24. [111] D. H. Robertson, D. W. Brenner, J. W. Mintmire, Phys. Rev. B: Condens. Matter 1992, 45, 12592-12595. [112] S.-i. Sawada, N. Hamada, Solid State Commun. 1992, 83, 917-919.

246 References

[113] A. A. Lucas, P. H. Lambin, R. E. Smalley, J. Phys. Chem. Solids 1993, 54, 587- 593. [114] Z. Q. Xue, W. M. Liu, S. M. Hou, J. P. Sun, Z. J. Shi, Z. N. Gu, X. Y. Zhao, Z. X. Zhang, J. L. Wu, L. M. Peng, Q. D. Wu, Mater. Sci. Eng., C 2001, 16, 17-21. [115] K. Tanaka, M. Okada, Y. Huang, in The science and technology of carbon nanotubes (Eds.: K. Tanaka, T. Yamabe, K. Fukui), Elsevier Science Ltd., Amsterdam, 1999, pp. 40-50. [116] J. W. Mintmire, B. I. Dunlap, C. T. White, Phys. Rev. Lett. 1992, 68, 631-634. [117] R. Saito, M. Fujita, G. Dresselhaus, M. S. Dresselhaus, Phys. Rev. B: Condens. Matter 1992, 46, 1804-1811. [118] N. Hamada, S.-i. Sawada, A. Oshiyama, Phys. Rev. Lett. 1992, 68, 1579-1581. [119] M. S. Dresselhaus, G. Dresselhaus, R. Saito, Carbon 1995, 33, 883-891. [120] M. S. Dresselhaus, G. Dresselhaus, P. C. Eklund, Science of fullerenes and carbon nanotubes, Academic Press, San Diego, 1996. [121] K. Balasubramanian, M. Burghard, Small 2005, 1, 180-192. [122] M. S. Arnold, PhD thesis, Northwestern Univ. 2006. [123] J. W. G. Wilder, L. C. Venema, A. G. Rinzler, R. E. Smalley, C. Dekker, Nature 1998, 391, 59-62. [124] Y. Sato, K. Yanagi, Y. Miyata, K. Suenaga, H. Kataura, S. Iijima, Nano Lett. 2008, 8, 3151-3154. [125] C. Dekker, Phys. Today 1999, 52, 22-28. [126] M. C. Hersam, Nat. Nanotechnol. 2008, 3, 387-394. [127] P. Avouris, Z. Chen, V. Perebeinos, Nat. Nanotechnol. 2007, 2, 605-615. [128] P. J. F. Harris, Carbon nanotubes and related structures : new materials for the twenty-first century, Cambridge University Press, Cambridge, 1999. [129] J.-C. Charlier, X. Blase, S. Roche, Rev. Mod. Phys. 2007, 79, 677. [130] M. Biercuk, S. Ilani, C. Marcus, P. McEuen, Top. Appl. Phys. 2008, 111, 455- 493. [131] J. C. Charlier, Acc. Chem. Res. 2002, 35, 1063-1069. [132] J. M. Cowley, P. Nikolaev, A. Thess, R. E. Smalley, Chem. Phys. Lett. 1997, 265, 379-384. [133] A. Hirsch, Angew. Chem. Int. Ed. 2002, 41, 1853-1859. [134] S. J. Tans, M. H. Devoret, H. Dai, A. Thess, R. E. Smalley, L. J. Geerligs, C. Dekker, Nature 1997, 386, 474-477.

247 References

[135] H. T. Soh, C. F. Quate, A. F. Morpurgo, C. M. Marcus, J. Kong, H. Dai, Appl. Phys. Lett. 1999, 75, 627-629. [136] J. Appenzeller, Proc. IEEE 2008, 96, 201-211. [137] S. Trasobares, O. Stephan, C. Colliex, G. Hug, W. K. Hsu, H. W. Kroto, D. R. M. Walton, Eur. Phys. J. B 2001, 22, 117-122. [138] D. A. Britz, A. N. Khlobystov, K. Porfyrakis, A. Ardavan, G. A. D. Briggs, Chem. Commun. 2005, 37-39. [139] A. N. Khlobystov, D. A. Britz, G. A. D. Briggs, Acc. Chem. Res. 2005, 38, 901- 909. [140] J. Rossato, R. J. Baierle, A. Fazzio, R. Mota, Nano Lett. 2005, 5, 197-200. [141] A. J. Stone, D. J. Wales, Chem. Phys. Lett. 1986, 128, 501-503. [142] A. J. Lu, B. C. Pan, Phys. Rev. Lett. 2004, 92, 105504. [143] S. L. Mielke, D. Troya, S. Zhang, J.-L. Li, S. Xiao, R. Car, R. S. Ruoff, G. C. Schatz, T. Belytschko, Chem. Phys. Lett. 2004, 390, 413-420. [144] Y. Fan, B. R. Goldsmith, P. G. Collins, Nat. Mater. 2005, 4, 906-911. [145] M. B. Nardelli, B. I. Yakobson, J. Bernholc, Phys. Rev. B: Condens. Matter 1998, 57, R4277-R48280. [146] T. Lin, V. Bajpai, T. Ji, L. Dai, Aust. J. Chem. 2003, 56, 635-651. [147] H. Hu, P. Bhowmik, B. Zhao, M. A. Hamon, M. E. Itkis, R. C. Haddon, Chem. Phys. Lett. 2001, 345, 25-28. [148] T. C. Dinadayalane, J. Leszczynski, in Nanomaterials : design and simulation (Eds.: P. B. Balbuena, J. M. Seminario), Elsevier, Amsterdam, 2007, pp. 167- 199. [149] X. Lu, Z. Chen, P. v. R. Schleyer, J. Am. Chem. Soc. 2005, 127, 20-21. [150] H. F. Bettinger, J. Phys. Chem. B 2005, 109, 6922-6924. [151] D. R. Kauffman, A. Star, Angew. Chem. Int. Ed. 2008, 47, 6550-6570. [152] D. Vairavapandian, P. Vichchulada, M. D. Lay, Anal. Chim. Acta 2008, 626, 119-129. [153] C. A. Dyke, J. M. Tour, J. Phys. Chem. A 2004, 108, 11151-11159. [154] D. Tasis, N. Tagmatarchis, V. Georgakilas, M. Prato, Chem. Eur. J. 2003, 9, 4000-4008. [155] A. M. Rao, E. Richter, S. Bandow, B. Chase, P. C. Eklund, K. A. Williams, S. Fang, K. R. Subbaswamy, M. Menon, A. Thess, R. E. Smalley, G. Dresselhaus, M. S. Dresselhaus, Science 1997, 275, 187-191.

248 References

[156] S. M. Bachilo, M. S. Strano, C. Kittrell, R. H. Hauge, R. E. Smalley, R. B. Weisman, Science 2002, 298, 2361-2366. [157] M. J. O'Connell, S. M. Bachilo, C. B. Huffman, V. C. Moore, M. S. Strano, E. H. Haroz, K. L. Rialon, P. J. Boul, W. H. Noon, C. Kittrell, J. Ma, R. H. Hauge, R. B. Weisman, R. E. Smalley, Science 2002, 297, 593-596. [158] A. Hagen, T. Hertel, Nano Lett. 2003, 3, 383-388. [159] J. P. Salvetat, J. M. Bonard, N. H. Thomson, A. J. Kulik, L. Forró, W. Benoit, L. Zuppiroli, Appl. Phys. A 1999, 69, 255-260. [160] J. P. Lu, J. Han, Int. J. High Speed Electron. Syst. 1998, 9, 101-123. [161] J. P. Lu, Phys. Rev. Lett. 1995, 74, 1123-1126. [162] G. Baumgartner, M. Carrard, L. Zuppiroli, W. Bacsa, W. A. de Heer, L. Forró, Phys. Rev. B: Condens. Matter 1997, 55, 6704-6707. [163] A. Brataas, Nature 2008, 452, 419-420. [164] J. Hone, M. Whitney, C. Piskoti, A. Zettl, Phys. Rev. B: Condens. Matter 1999, 59, R2514-R2516. [165] J. Hone, B. Batlogg, Z. Benes, A. T. Johnson, J. E. Fischer, Science 2000, 289, 1730-1733. [166] S. J. Kang, C. Kocabas, T. Ozel, M. Shim, N. Pimparkar, M. A. Alam, S. V. Rotkin, J. A. Rogers, Nat. Nanotechnol. 2007, 2, 230-236. [167] J. N. Coleman, U. Khan, Y. K. Gun'ko, Adv. Mater. 2006, 18, 689-706. [168] S. Bal, S. Samal, Bull. Mater. Sci. 2007, 30, 379-386. [169] A. M. K. Esawi, M. M. Farag, Mater. Des. 2007, 28, 2394-2401. [170] T. Fujigaya, S. Haraguchi, T. Fukumaru, N. Nakashima, Adv. Mater. 2008, 20, 2151-2155. [171] M. N. Tchoul, W. T. Ford, M. L. P. Ha, I. Chavez-Sumarriva, B. P. Grady, G. Lolli, D. E. Resasco, S. Arepalli, Chem. Mater. 2008, 20, 3120-3126. [172] S. V. Ahir, Y. Y. Huang, E. M. Terentjev, Polymer 2008, 49, 3841-3854. [173] H. E. Unalan, P. Hiralal, D. Kuo, B. Parekh, G. Amaratunga, M. Chhowalla, J. Mater. Chem. 2008, 18, 5909-5912. [174] T. Umeyama, H. Imahori, Energy Environ. Sci. 2008, 1, 120-133. [175] V. Sgobba, D. M. Guldi, J. Mater. Chem. 2008, 18, 153-157. [176] M. Trojanowicz, TrAC, Trends Anal. Chem. 2006, 25, 480-489. [177] M. Valcarcel, S. Cardenas, B. M. Simonet, Anal. Chem. 2007, 79, 4788-4797. [178] I. Dumitrescu, P. R. Unwin, N. R. Wilson, J. V. Macpherson, Anal. Chem. 2008, 80, 3598-3605.

249 References

[179] C. Y. Lee, R. Sharma, A. D. Radadia, R. I. Masel, M. S. Strano, Angew. Chem. Int. Ed. 2008, 47, 5018-5021. [180] L. Agüí, P. Yáñez-Sedeño, J. M. Pingarrón, Anal. Chim. Acta 2008, 622, 11-47. [181] R. Seelaboyina, W. Choi, Recent Pat. Nanotechnol. 2007, 1, 238-244. [182] P. Zhao, X. F. Shang, Y. P. Ma, J. J. Zhou, Z. Q. Gu, Z. H. Li, Y. B. Xu, M. Wang, Eur. Phys. J. Appl. Phys. 2008, 42, 251-254. [183] S. Park, H. C. Kim, M. H. Yum, J. H. Yang, C. Y. Park, K. Chun, B. Eom, Nanotechnology 2008, 19, 445304. [184] C. J. Shearer, J. Yu, K. M. O'Donnell, L. Thomsen, P. C. Dastoor, J. S. Quinton, J. G. Shapter, J. Mater. Chem. 2008, 18, 5753-5760. [185] B. Corry, J. Phys. Chem. B 2008, 112, 1427-1434. [186] F. Fornasiero, H. G. Park, J. K. Holt, M. Stadermann, C. P. Grigoropoulos, A. Noy, O. Bakajin, Proc. Nat. Acad. Sci. U.S.A. 2008, 105, 17250-17255. [187] J. H. Hafner, C. L. Cheung, A. T. Woolley, C. M. Lieber, Prog. Biophys. Mol. Biol. 2001, 77, 73-110. [188] M. Zhao, V. Sharma, H. Wei, R. R. Birge, J. A. Stuart, F. Papadimitrakopoulos, B. D. Huey, Nanotechnology 2008, 19, 235704. [189] F. Z. Fang, Z. W. Xu, S. Dong, Meas. Sci. Technol. 2008, 19, 055501. [190] C. Liu, H.-M. Cheng, J. Phys. D: Appl. Phys. 2005, 38, R231-R252. [191] A. G. Lipson, B. F. Lyakhov, E. I. Saunin, A. Y. Tsivadze, Phys. Rev. B: Condens. Matter 2008, 77, 081405. [192] A. D. Taylor, R. C. Sekol, J. M. Kizuka, S. D'Cunha, C. M. Comisar, J. Catal. 2008, 259, 5-16. [193] J. Shen, Y. Hu, C. Li, C. Qin, M. Ye, Electrochim. Acta 2008, 53, 7276-7280. [194] N. G. Portney, M. Ozkan, Anal. Bioanal.Chem. 2006, 384, 620-630. [195] J. Chen, S. Chen, X. Zhao, L. V. Kuznetsova, S. S. Wong, I. Ojima, J. Am. Chem. Soc. 2008, 130, 16778-16785. [196] Z. Liu, X. Li, S. M. Tabakman, K. Jiang, S. Fan, H. Dai, J. Am. Chem. Soc. 2008, 130, 13540-13541. [197] Z. Liu, K. Chen, C. Davis, S. Sherlock, Q. Cao, X. Chen, H. Dai, Cancer Res. 2008, 68, 6652-6660. [198] J. Hone, L. Kam, Nat Nano 2007, 2, 140-141. [199] E. Ben-Jacob, Y. Hanein, J. Mater. Chem. 2008, 18, 5181-5186. [200] K. Welsher, Z. Liu, D. Daranciang, H. Dai, Nano Lett. 2008, 8, 586-590. [201] C. A. Merchant, N. Markovic, Phys. Rev. Lett. 2008, 100, 156601.

250 References

[202] M. Ferrier, A. De Martino, A. Kasumov, S. Guéron, M. Kociak, R. Egger, H. Bouchiat, Solid State Commun. 2004, 131, 615-623. [203] N. Murata, J. Haruyama, J. Reppert, A. M. Rao, T. Koretsune, S. Saito, M. Matsudaira, Y. Yagi, Phys. Rev. Lett. 2008, 101, 027002. [204] S. E. Moulton, A. I. Minett, G. G. Wallace, Sensor Lett. 2005, 3, 183-193. [205] K. Balasubramanian, M. Burghard, Anal. Bioanal.Chem. 2006, 385, 452-468. [206] B. L. Allen, P. D. Kichambare, A. Star, Adv. Mater. 2007, 19, 1439-1451. [207] M. D. Shirsat, C. O. Too, G. G. Wallace, Electroanal. 2008, 20, 150-156. [208] D. R. Kauffman, A. Star, Chem. Soc. Rev. 2008, 37, 1197-1206. [209] K. M. Shin, J. W. Lee, G. G. Wallace, S. J. Kim, Sens. Actuators, B 2008, 133, 393-397. [210] R. H. Baughman, C. Cui, A. A. Zakhidov, Z. Iqbal, J. N. Barisci, G. M. Spinks, G. G. Wallace, A. Mazzoldi, D. De Rossi, A. G. Rinzler, O. Jaschinski, S. Roth, M. Kertesz, Science 1999, 284, 1340-1344. [211] S. Ghosh, A. K. Sood, N. Kumar, Science 2003, 299, 1042-1044. [212] P. Dharap, Z. Li, S. Nagarajaiah, E. V. Barrera, Nanotechnology 2004, 15, 379- 382. [213] C. Stampfer, T. Helbling, D. Obergfell, B. Schoberle, M. K. Tripp, A. Jungen, S. Roth, V. M. Bright, C. Hierold, Nano Lett. 2006, 6, 233-237. [214] N.-K. Chang, C.-C. Su, S.-H. Chang, Appl. Phys. Lett. 2008, 92, 063501. [215] B. Mahar, C. Laslau, R. Yip, Y. Sun, IEEE Sens. J. 2007, 7, 266-284. [216] C. Hierold, A. Jungen, C. Stampfer, T. Helbling, Sens. Actuators, A 2007, 136, 51-61. [217] A. R. Hall, M. R. Falvo, R. Superfine, S. Washburn, Nano Lett. 2008, 8, 3746- 3749. [218] J. J. Davis, K. S. Coleman, B. R. Azamian, C. B. Bagshaw, M. L. H. Green, Chem. Eur. J. 2003, 9, 3732-3739. [219] N. Sinha, J. Ma, J. T. W. Yeow, J. Nanosci. Nanotechnol. 2006, 6, 573-590. [220] M. Endo, M. S. Strano, P. M. Ajayan, Top. Appl. Phys. 2008, 111, 13-62. [221] A. Peigney, C. Laurent, E. Flahaut, R. R. Bacsa, A. Rousset, Carbon 2001, 39, 507-514. [222] M. Cinke, J. Li, B. Chen, A. Cassell, L. Delzeit, J. Han, M. Meyyappan, Chem. Phys. Lett. 2002, 365, 69-74. [223] J. Kong, N. R. Franklin, C. Zhou, M. G. Chapline, S. Peng, K. Cho, H. Dai, Science 2000, 287, 622-625.

251 References

[224] P. G. Collins, K. Bradley, M. Ishigami, A. Zettl, Science 2000, 287, 1801-1804. [225] A. Javey, J. Gou, M. Paulsson, Q. Wang, D. Mann, M. Lundstrom, H. Dai, Phys. Rev. Lett. 2004, 92, 106804. [226] E. S. Snow, F. K. Perkins, J. A. Robinson, Chem. Soc. Rev. 2006, 35, 790-798. [227] American Chemical Society, Scifinder, American Chemical Society, Columbus, 2009. [228] K. G. Ong, K. Zeng, C. A. Grimes, IEEE Sens. J. 2002, 2, 82-88. [229] J. Li, Y. Lu, Q. Ye, M. Cinke, J. Han, M. Meyyappan, Nano Lett. 2003, 3, 929- 933. [230] S. Chopra, A. Pham, J. Gaillard, A. Parker, A. M. Rao, Appl. Phys. Lett. 2002, 80, 4632-4634. [231] M. Penza, F. Antolini, M. Vittori-Antisari, Sens. Actuators, B 2004, 100, 47-59. [232] M. Penza, F. Antolini, M. Vittori-Antisari, Thin Solid Films 2005, 472, 246-252. [233] H. Y. Jung, S. M. Jung, J. Kim, J. S. Suh, Appl. Phys. Lett. 2007, 90, 153114. [234] A. Modi, N. Koratkar, E. Lass, B. Wei, P. M. Ajayan, Nature 2003, 424, 171- 174. [235] E. S. Snow, F. K. Perkins, E. J. Houser, S. C. Badescu, T. L. Reinecke, Science 2005, 307, 1942-1945. [236] E. S. Snow, F. K. Perkins, Nano Lett. 2005, 5, 2414-2417. [237] D. R. Kauffman, A. Star, J. Phys. Chem. C 2008, 112, 4430-4434. [238] A. Star, V. Joshi, S. Skarupo, D. Thomas, J.-C. P. Gabriel, J. Phys. Chem. B 2006, 110, 21014-21020. [239] A. Zangwill, Physics at surfaces, Cambridge University Press, Cambridge, 1988. [240] Z. Chen, J. Appenzeller, J. Knoch, Y.-m. Lin, P. Avouris, Nano Lett. 2005, 5, 1497-1502. [241] R. Martel, T. Schmidt, H. R. Shea, T. Hertel, P. Avouris, Appl. Phys. Lett. 1998, 73, 2447-2449. [242] A. Star, J.-C. P. Gabriel, K. Bradley, G. Gruner, Nano Lett. 2003, 3, 459-463. [243] C. Cantalini, L. Valentini, L. Lozzi, I. Armentano, J. M. Kenny, S. Santucci, Sens. Actuators, B 2003, 93, 333-337. [244] L. Valentini, I. Armentano, J. M. Kenny, C. Cantalini, L. Lozzi, S. Santucci, Appl. Phys. Lett. 2003, 82, 961-963. [245] Y. Jang, S. Moon, J. Ahn, Y. Lee, B. Ju, Sens. Actuators, B 2004, 99, 118-122. [246] H.-Q. Nguyen, J.-S. Huh, Sens. Actuators, B 2006, 117, 426-430.

252 References

[247] M. Penza, G. Cassano, R. Rossi, A. Rizzo, M. A. Signore, M. Alvisi, N. Lisi, E. Serra, R. Giorgi, Appl. Phys. Lett. 2007, 90, 103101. [248] R. Larciprete, S. Lizzit, L. Petaccia, A. Goldoni, Appl. Phys. Lett. 2006, 88, 243111. [249] J. Kong, M. G. Chapline, H. Dai, Adv. Mater. 2001, 13, 1384-1386. [250] Y. M. Wong, W. P. Kang, J. L. Davidson, A. Wisitsora-at, K. L. Soh, Sens. Actuators, B 2003, 93, 327-332. [251] Y. Lu, J. Li, J. Han, H. T. Ng, C. Binder, C. Partridge, M. Meyyappan, Chem. Phys. Lett. 2004, 391, 344-348. [252] J. Li, Y. Lu, H. T. Ng, J. Han, M. Meyyappan, Chem. Senses 2004, 20 (Suppl. B), 710-711. [253] I. Sayago, E. Terrado, E. Lafuente, M. C. Horrillo, W. K. Maser, A. M. Benito, R. Navarro, E. P. Urriolabeitia, M. T. Martinez, J. Gutierrez, Synth. Met. 2005, 148, 15-19. [254] J. Sippel-Oakley, H.-T. Wang, B. S. Kang, Z. Wu, F. Ren, A. G. Rinzler, S. J. Pearton, Nanotechnology 2005, 16, 2218-2221. [255] I. Sayago, E. Terrado, M. Aleixandre, M. C. Horrillo, M. J. Fernández, J. Lozano, E. Lafuente, W. K. Maser, A. M. Benito, M. T. Martinez, J. Gutiérrez, E. Muñoz, Sens. Actuators, B 2007, 122, 75-80. [256] Y. Sun, H. H. Wang, Appl. Phys. Lett. 2007, 90, 213107. [257] Y. Sun, H. H. Wang, Adv. Mater. 2007, 19, 2818-2823. [258] S. Mubeen, T. Zhang, B. Yoo, M. A. Deshusses, N. V. Myung, J. Phys. Chem. C 2007, 111, 6321-6327. [259] M. Penza, R. Rossi, M. Alvisi, G. Cassano, M. A. Signore, E. Serra, R. Giorgi, Sens. Actuators, B 2008, 135, 289-297. [260] F. J. Ibanez, F. P. Zamborini, J. Am. Chem. Soc. 2008, 130, 622-633. [261] Y. Lu, C. Partridge, M. Meyyappan, J. Li, J. Electroanal. Chem. 2006, 593, 105- 110. [262] M. Penza, G. Cassano, R. Rossi, M. Alvisi, A. Rizzo, M. A. Signore, T. Dikonimos, E. Serra, R. Giorgi, Appl. Phys. Lett. 2007, 90, 173123. [263] M. K. Kumar, S. Ramaprabhu, J. Phys. Chem. B 2006, 110, 11291-11298. [264] Y. X. Liang, Y. J. Chen, T. H. Wang, Appl. Phys. Lett. 2004, 85, 666-668. [265] C. Bittencourt, A. Felten, E. H. Espinosa, R. Ionescu, E. Llobet, X. Correig, J. J. Pireaux, Sens. Actuators, B 2006, 115, 33-41.

253 References

[266] M. Sánchez, R. Guirado, M. Rincón, J. Mater. Sci. - Mater. Electron. 2007, 18, 1131-1136. [267] F. Villalpando-Páez, A. H. Romero, E. Muñoz-Sandoval, L. M. Martínez, H. Terrones, M. Terrones, Chem. Phys. Lett. 2004, 386, 137-143. [268] X. Feng, S. Irle, H. Witek, K. Morokuma, R. Vidic, E. Borguet, J. Am. Chem. Soc. 2005, 127, 10533-10538. [269] C. Staii, A. T. Johnson, M. Chen, A. Gelperin, Nano Lett. 2005, 5, 1774-1778. [270] E. Bekyarova, M. Davis, T. Burch, M. E. Itkis, B. Zhao, S. Sunshine, R. C. Haddon, J. Phys. Chem. B 2004, 108, 19717-19720. [271] T. Zhang, S. Mubeen, E. Bekyarova, B. Y. Yoo, R. C. Haddon, N. V. Myung, M. A. Deshusses, Nanotechnology 2007, 18, 165504. [272] Y. Li, H.-c. Wang, M.-j. Yang, Sens. Actuators, B 2007, 121, 496-500. [273] P. Qi, O. Vermesh, M. Grecu, A. Javey, Q. Wang, H. Dai, Nano Lett. 2003, 3, 347-351. [274] K. Bradley, J.-C. P. Gabriel, A. Star, G. Gruner, Appl. Phys. Lett. 2003, 83, 3821-3823. [275] A. Star, T. R. Han, V. Joshi, J. C. P. Gabriel, G. Grüner, Adv. Mater. 2004, 16, 2049-2052. [276] T. Zhang, M. B. Nix, B.-Y. Yoo, M. A. Deshusses, N. V. Myung, Electroanal. 2006, 18, 1153-1158. [277] J. P. Novak, E. S. Snow, E. J. Houser, D. Park, J. L. Stepnowski, R. A. McGill, Appl. Phys. Lett. 2003, 83, 4026-4028. [278] K. Cattanach, R. D. Kulkarni, M. Kozlov, S. K. Manohar, Nanotechnology 2006, 17, 4123-4128. [279] K. S. V. Santhanam, R. Sangoi, L. Fuller, Sens. Actuators, B 2005, 106, 766- 771. [280] C. Wei, L. Dai, A. Roy, T. B. Tolle, J. Am. Chem. Soc. 2006, 128, 1412-1413. [281] B. Philip, J. K. Abraham, A. Chandrasekhar, V. K. Varadan, Smart Mater. Struct. 2003, 12, 935-939. [282] J. M. Lehn, Pure Appl. Chem. 1978, 50, 871-892. [283] R. P. Feynman, in Presentation to American Physical Society, California Institute of Technology, 1959. [284] R. P. Feynman, Eng. Sci. 1960, 23, 22-36. [285] E. Fischer, Ber. Dtsch. Chem. Ges. 1894, 27, 2985-2993.

254 References

[286] J. M. Lehn, Supramolecular chemistry, concepts and perspectives, VCH, Weinheim, 1995. [287] J. M. Lehn, Struct. Bond. 1973, 16, 1-69. [288] J. W. Steed, D. R. Turner, K. J. Wallace, Core concepts in supramolecular chemistry and nanochemistry, John Wiley & Sons, Ltd., Chichester, 2007. [289] P. D. Beer, P. A. Gale, D. K. Smith, Supramolecular chemistry, Oxford University Press, Oxford, 1999. [290] J. W. Steed, J. L. Atwood, Supramolecular chemistry, John Wiley & Sons, Ltd., Chichester, 2000. [291] D. J. Cram, J. M. Cram, Container molecules and their guests, Royal Society of Chemistry, Cambridge, 1994. [292] J. M. Lehn, A. Rigault, J. Siegel, J. Harrowfield, B. Chevrier, D. Moras, Proc. Nat. Acad. Sci. U.S.A. 1987, 84, 2565-2569. [293] J.-M. Lehn, Angew. Chem. Int. Ed. Engl. 1988, 27, 89-112. [294] J.-M. Lehn, Angew. Chem. Int. Ed. Engl. 1990, 29, 1304-1319. [295] K. S. Iyer, M. Norret, S. J. Dalgarno, J. L. Atwood, C. L. Raston, Angew. Chem. Int. Ed. 2008, 47, 6362-6366. [296] A. Baeyer, Ber. Dtsch. Chem. Ges. 1872, 5, 25-26. [297] A. Baeyer, Ber. Dtsch. Chem. Ges. 1872, 5, 280-282. [298] A. Baeyer, Ber. Dtsch. Chem. Ges. 1872, 5, 1094-1100. [299] C. D. Gutsche, R. Muthukrishnan, J. Org. Chem. 1978, 43, 4905-4906. [300] C. D. Gutsche, Calixarenes, Royal Society of Chemistry, 1989. [301] V. Böhmer, Angew. Chem. Int. Ed. Engl. 1995, 34, 713-745. [302] Z. Asfari, J. Vicens, Makromol. Chem., Rapid Commun. 1989, 10, 181-183. [303] B. Dhawan, S.-I. Chen, C. D. Gutsche, Makromol. Chem. 1987, 188, 921-950. [304] C. D. Gutsche, M. Iqbal, Org. Synth. 1990, 68, 234-237. [305] D. R. Stewart, C. D. Gutsche, Org. Prep. Proced. Int. 1993, 25, 137-139. [306] K. Iwamoto, K. Araki, S. Shinkai, Bull. Chem. Soc. Jpn. 1994, 67, 1499-1502. [307] C. D. Gutsche, B. Dhawan, M. Leonis, D. Stewart, Org. Synth. 1990, 68, 238- 242. [308] Y. Nakamoto, S.-i. Ishida, Makromol. Chem., Rapid Commun. 1982, 3, 705-707. [309] F. Vocanson, R. Lamartine, P. Lanteri, R. Longeray, J. Y. Gauvrit, New J. Chem. 1995, 19, 825-829. [310] J. H. Munch, C. D. Gutsche, Org. Synth. 1990, 68, 243-246.

255 References

[311] S. R. Izatt, R. T. Hawkins, J. J. Christensen, R. M. Izatt, J. Am. Chem. Soc. 1985, 107, 63-66. [312] K. Araki, A. Yanagi, S. Shinkai, Tetrahedron 1993, 49, 6763-6772. [313] V. Bocchi, D. Foina, A. Pochini, R. Ungaro, G. D. Andreetti, Tetrahedron 1982, 38, 373-378. [314] I. E. Lubitov, E. A. Shokova, V. V. Kovalev, Synlett. 1993, 647-648. [315] M. Makha, C. L. Raston, Tetrahedron Lett. 2001, 42, 6215-6217. [316] C. D. Gutsche, P. F. Pagoria, J. Org. Chem. 1985, 50, 5795-5802. [317] M. Makha, C. L. Raston, Chem. Commun. 2001, 2470-2471. [318] B. Souley, Z. Asfari, J. Vicens, Pol. J. Chem. 1993, 67, 763-767. [319] J. L. Atwood, M. J. Hardie, C. L. Raston, C. A. Sandoval, Org. Lett. 1999, 1, 1523-1526. [320] A. Ettahiri, A. Thozet, M. Perrin, Supramol. Chem. 1994, 3, 191-196. [321] B. T. Hayes, R. F. Hunter, Chem. Ind. (London) 1956, 193-194. [322] B. T. Hayes, R. F. Hunter, J. Appl. Chem. 1958, 8, 743-748. [323] H. Kämmerer, G. Happel, F. Caesar, Makromol. Chem. 1972, 162, 179-197. [324] G. Happel, B. Mathiasch, H. Kämmerer, Makromol. Chem. 1975, 176, 3317- 3334. [325] H. Kämmerer, G. Happel, V. Böhmer, D. Rathay, Monatsh. Chem. 1978, 109, 767-773. [326] H. Kämmerer, G. Happel, Makromol. Chem. 1980, 181, 2049-2062. [327] H. Kämmerer, G. Happel, Monatsh. Chem. 1981, 112, 759-768. [328] H. Kämmerer, G. Happel, B. Mathiasch, Makromol. Chem. 1981, 182, 1685- 1694. [329] V. Böhmer, F. Marschollek, L. Zetta, J. Org. Chem. 1987, 52, 3200-3205. [330] V. Böhmer, L. Merkel, U. Kunz, J. Chem. Soc., Chem. Commun. 1987, 12, 896- 897. [331] J. de Mendoza, P. M. Nieto, P. Prados, C. Sánchez, Tetrahedron 1990, 46, 671- 682. [332] I. Dumazet, J.-B. Regnouf-de-Vains, R. Lamartine, Synth. Commun. 1997, 27, 2547-2555. [333] B. König, M. Rödel, P. Bubenitschek, P. G. Jones, Angew. Chem. Int. Ed. Engl. 1995, 34, 661-662. [334] H. Kumagai, M. Hasegawa, S. Miyanari, Y. Sugawa, Y. Sato, T. Hori, S. Ueda, H. Kamiyama, S. Miyano, Tetrahedron Lett. 1997, 38, 3971-3972.

256 References

[335] G. Mislin, E. Graf, M. W. Hosseini, A. De Cian, J. Fischer, Chem. Commun. 1998, 1345-1346. [336] N. Iki, H. Kumagai, N. Morohashi, K. Ejima, M. Hasegawa, S. Miyanari, S. Miyano, Tetrahedron Lett. 1998, 39, 7559-7562. [337] H. Tsue, K. Ishibashi, H. Takahashi, R. Tamura, Org. Lett. 2005, 7, 2165-2168. [338] E. Ziegler, Sci. Pharm. 1951, 209-212. [339] S. Usui, K. Deyama, Y. Fukazawa, T. Sone, Y. Ohba, Chem. Lett. 1991, 20, 1387-1390. [340] F. Vocanson, R. Lamartine, Supramol. Chem. 1996, 7, 19-25. [341] C. D. Gutsche, D. E. Johnston, D. R. Stewart, J. Org. Chem. 1999, 64, 3747- 3750. [342] F. Vocanson, R. Lamartine, R. Perrin, Supramol. Chem. 1994, 4, 153-157. [343] C. D. Gutsche, L. J. Bauer, Tetrahedron Lett. 1981, 22, 4763-4766. [344] C. D. Gutsche, L. J. Bauer, J. Am. Chem. Soc. 1985, 107, 6052-6059. [345] J. W. Cornforth, P. D'Arcy Hart, G. A. Nicholls, R. J. W. Rees, J. A. Stock, Br. J. Pharmacol. Chemother. 1955, 10, 73-86. [346] C. Jaime, J. De Mendoza, P. Prados, P. M. Nieto, C. Sanchez, J. Org. Chem. 1991, 56, 3372-3376. [347] M. Coruzzi, G. D. Andreetti, V. Bocchi, A. Pochini, R. Ungaro, J. Chem. Soc., Perkin Trans. 2 1982, 1133-1138. [348] R. K. Juneja, K. D. Robinson, G. W. Orr, R. H. Dubois, K. A. Belmore, J. L. Atwood, J. A. Ripmeester, C. I. Ratcliffe, J. Inclusion Phenom. Macrocyclic Chem. 1992, 13, 93-96. [349] J. F. Gallagher, G. Ferguson, V. Böhmer, D. Kraft, Acta Crystallogr., Sect. C: Cryst. Struct. Commun. 1994, 50, 73-77. [350] D. R. Stewart, M. Krawiec, R. P. Kashyap, W. H. Watson, C. D. Gutsche, J. Am. Chem. Soc. 1995, 117, 586-601. [351] O. Ermer, Helv. Chim. Acta 1991, 74, 1339-1351. [352] J. D. Crane, P. B. Hitchcock, H. W. Kroto, R. Taylor, D. R. M. Walton, J. Chem. Soc., Chem. Commun. 1992, 1764-1765. [353] T. Andersson, K. Nilsson, M. Sundahl, G. Westman, O. Wennerström, J. Chem. Soc., Chem. Commun. 1992, 604-606. [354] J. L. Atwood, G. A. Koutsantonis, C. L. Raston, Nature 1994, 368, 229-231. [355] J. L. Atwood, L. J. Barbour, C. L. Raston, I. B. N. Sudria, Angew. Chem. Int. Ed. 1998, 37, 981-983.

257 References

[356] A. L. Balch, M. M. Olmstead, Coord. Chem. Rev. 1999, 185-186, 601-617. [357] T. Ishii, N. Aizawa, R. Kanehama, M. Yamashita, K.-i. Sugiura, H. Miyasaka, Coord. Chem. Rev. 2002, 226, 113-124. [358] J. L. Atwood, L. J. Barbour, M. W. Heaven, C. L. Raston, Angew. Chem. Int. Ed. 2003, 42, 3254-3257. [359] M. Makha, A. Purich, C. L. Raston, A. N. Sobolev, Eur. J. Inorg. Chem. 2006, 507-517. [360] J. L. Atwood, L. J. Barbour, M. W. Heaven, C. L. Raston, Chem. Commun. 2003, 2270-2271. [361] M. Makha, C. L. Raston, A. N. Sobolev, P. Turner, Cryst. Growth Des. 2006, 6, 224-228. [362] M. Makha, J. J. McKinnon, A. N. Sobolev, M. A. Spackman, C. L. Raston, Chem. Eur. J. 2007, 13, 3907-3912. [363] T. Haino, J. Seyama, C. Fukunaga, Y. Murata, K. Komatsu, Y. Fukazawa, Bull. Chem. Soc. Jpn. 2005, 78, 768-770. [364] T. Haino, C. Fukunaga, Y. Fukazawa, Org. Lett. 2006, 8, 3545-3548. [365] T. Haino, C. Fukunaga, Y. Fukazawa, J. Nanosci. Nanotechnol. 2007, 7, 1386- 1388. [366] T. Suzuki, K. Nakashima, S. Shinkai, Chem. Lett. 1994, 4, 699-702. [367] J. L. Atwood, L. J. Barbour, P. J. Nichols, C. L. Raston, C. A. Sandoval, Chem. Eur. J. 1999, 5, 990-996. [368] L. J. Barbour, G. W. Orr, J. L. Atwood, Chem. Commun. 1997, 1439-1440. [369] S. Mizyed, P. R. Tremaine, P. E. Georghiou, J. Chem. Soc., Perkin Trans. 2 2001, 3-6. [370] M. Makha, C. L. Raston, A. N. Sobolev, L. J. Barbour, P. Turner, CrystEngComm 2006, 8, 306-308. [371] M. Makha, C. W. Evans, A. N. Sobolev, C. L. Raston, Cryst. Growth Des. 2008, 8, 2929-2932. [372] A. Ikeda, M. Yoshimura, S. Shinkai, Tetrahedron Lett. 1997, 38, 2107-2110. [373] C. L. Raston, J. L. Atwood, P. J. Nichols, I. B. N. Sudria, Chem. Commun. 1996, 2615-2616. [374] G.-B. Pan, J.-M. Liu, H.-M. Zhang, L.-J. Wan, Q.-Y. Zheng, C.-L. Bai, Angew. Chem. Int. Ed. 2003, 42, 2747-2751. [375] N. S. Isaacs, P. J. Nichols, C. L. Raston, C. A. Sandova, D. J. Young, Chem. Commun. 1997, 1839-1840.

258 References

[376] T. Haino, M. Yanase, Y. Fukazawa, Angew. Chem. Int. Ed. Engl. 1997, 36, 259- 260. [377] T. Haino, M. Yanase, Y. Fukazawa, Tetrahedron Lett. 1997, 38, 3739-3742. [378] T. Haino, H. Araki, Y. Fujiwara, Y. Tanimoto, Y. Fukazawa, Chem. Commun. 2002, 2148-2149. [379] T. Haino, M. Yanase, Y. Fukazawa, Tetrahedron Lett. 2005, 46, 1411-1414. [380] T. Haino, M. Yanase, C. Fukunaga, Y. Fukazawa, Tetrahedron 2006, 62, 2025- 2035. [381] M. Makha, M. J. Hardie, C. L. Raston, Chem. Commun. 2002, 14, 1446-1447. [382] J. L. Atwood, L. J. Barbour, C. L. Raston, Cryst. Growth Des. 2002, 2, 3-6. [383] A. Arduini, A. Pochini, S. Raverberi, R. Ungaro, J. Chem. Soc., Chem. Commun. 1984, 981-982. [384] F. Ohseto, H. Yamamoto, H. Matsumoto, S. Shinkai, Tetrahedron Lett. 1995, 36, 6911-6914. [385] P. Linnane, T. D. James, S. Shinkai, J. Chem. Soc., Chem. Commun. 1995, 1997-1998. [386] C. D. Gutsche, K. C. Nam, J. Am. Chem. Soc. 1988, 110, 6153-6162. [387] S. Shinkai, S. Mori, T. Tsubaki, T. Sone, O. Manabe, Tetrahedron Lett. 1984, 25, 5315-5318. [388] S. Shinkai, S. Mori, H. Koreishi, T. Tsubaki, O. Manabe, J. Am. Chem. Soc. 1986, 108, 2409-2416. [389] M. Almi, A. Arduini, A. Casnati, A. Pochini, R. Ungaro, Tetrahedron 1989, 45, 2177-2182. [390] V. I. Kalchenko, L. I. Atamas, V. V. Pirozhenko, Markovskii, Zh. Obshch. Khim. 1992, 62, 2623-2625. [391] V. I. Kalchenko, J. Lipkowski, Y. A. Simonov, M. A. Vysotsky, K. Suwinska, A. A. Dvorkin, V. V. Pirozhenko, I. F. Tsimbal, L. N. Markovsky, Zh. Obshch. Khim. 1995, 65, 1311-1320. [392] P. Shahgaldian, A. W. Coleman, V. I. Kalchenko, Tetrahedron Lett. 2001, 42, 577-579. [393] T. E. Clark, M. Makha, A. N. Sobolev, H. Rohrs, J. L. Atwood, C. L. Raston, Chem. Eur. J. 2008, 14, 3931-3938. [394] T. E. Clark, M. Makha, A. N. Sobolev, D. Su, H. Rohrs, M. L. Gross, J. L. Atwood, C. L. Raston, New J. Chem. 2008, 32, 1478-1483.

259 References

[395] A. Casnati, D. Sciotto, G. Arena, in Calixarenes 2001 (Eds.: Z. Asfari, V. Bohmer, J. Harrowfield, J. Vicens), Kluwer Academic Publishers, Dordrecht; Boston, 2001. [396] M. Makha, PhD thesis, Monash University (Clayton), 2001. [397] T. E. Clark, PhD thesis, University of Western Australia (Perth), 2007. [398] H. Dai, J. H. Hafner, A. G. Rinzler, D. T. Colber, R. E. Smalley, Nature 1996, 382, 147-150. [399] J. H. Hafner, C. L. Cheung, C. M. Lieber, J. Am. Chem. Soc. 1999, 121, 9750- 9751. [400] G. Girishkumar, T. D. Hall, K. Vinodgopal, P. V. Kamat, J. Phys. Chem. B 2006, 110, 107-114. [401] O. Breuer, U. Sundararaj, Polym. Compos. 2004, 25, 630-645. [402] A. A. White, S. M. Best, I. A. Kinloch, Int. J. Appl. Ceram. Technol. 2007, 4, 1- 13. [403] K. Besteman, J.-O. Lee, F. G. M. Wiertz, H. A. Heering, C. Dekker, Nano Lett. 2003, 3, 727-730. [404] J. R. Wood, D. H. Wagner, Appl. Phys. Lett. 2000, 76, 2883-2885. [405] Y. Gao, Y. Bando, Z. Liu, D. Golberg, H. Nakanishi, Appl. Phys. Lett. 2003, 83, 2913-2915. [406] M. C. LeMieux, M. Roberts, S. Barman, Y. W. Jin, J. M. Kim, Z. Bao, Science 2008, 321, 101-104. [407] R. Krupke, F. Hennrich, H. V. Lohneysen, M. M. Kappes, Science 2003, 301, 344-347. [408] Z. Chen, X. Du, M. Du, C. D. Rancken, H. Cheng, A. G. Rinzler, Nano Lett. 2003, 3, 1245-1249. [409] M. S. Arnold, A. A. Green, J. F. Hulvat, S. I. Stupp, M. C. Hersam, Nat. Nanotechnol. 2006, 1, 60-65. [410] H. Yang, S. C. Wang, P. Mercier, D. L. Akins, Chem. Commun. 2006, 1425- 1427. [411] Q.-H. Yang, N. Gale, C. J. Oton, F. Li, A. Vaughan, R. Saito, I. S. Nandhakumar, Z.-Y. Tang, H.-M. Cheng, T. Brown, W. H. Loh, Nanotechnology 2007, 18, 405706. [412] K. Saito, V. Troiani, H. Qiu, N. Solladie, T. Sakata, H. Mori, M. Ohama, S. Fukuzumi, J. Phys. Chem. C 2007, 111, 1194-1199.

260 References

[413] R. M. Tromp, A. Afzali, M. Freitag, D. B. Mitzi, Z. Chen, Nano Lett. 2008, 8, 469-472. [414] S. Park, H.-S. Yang, D. Kim, K. Jo, S. Jon, Chem. Commun. 2008, 2876-2878. [415] L. J. Hubble, T. E. Clark, M. Makha, C. L. Raston, J. Mater. Chem. 2008, 18, 5961-5966. [416] J. L. Atwood, R. J. Bridges, R. K. Juneja, A. K. Singh, US Pat. Appl. 94- 1786105489612, 1996. [417] N. Douteau-Guével, A. W. Coleman, J.-P. Morel, N. Morel-Desrosiers, J. Phys. Org. Chem. 1998, 11, 693-696. [418] A. Specht, F. Ziarelli, P. Bernard, M. Goeldner, L. Peng, Helv. Chim. Acta 2005, 88, 2641-2653. [419] A. S. Lobach, I. S. Ryzhkina, N. G. Spitsina, E. D. Obraztsova, Phys. Status Solidi B 2007, 244, 4030-4034. [420] T. Ogoshi, T.-a. Yamagishi, Y. Nakamoto, Chem. Commun. 2007, 4776-4778. [421] M. A. Pimenta, A. Marucci, S. D. M. Brown, M. J. Matthews, A. M. Rao, P. C. Eklund, R. E. Smalley, G. Dresselhaus, M. S. Dresselhaus, J. Mater. Res. 1998, 13, 2396-2404. [422] L. F. Su, J. N. Wang, F. Yu, Z. M. Sheng, H. Chang, C. Pak, Chem. Phys. Lett. 2006, 420, 421-425. [423] V. A. Sinani, M. K. Gheith, A. A. Yaroslavov, A. A. Rakhnyanskaya, K. Sun, A. A. Mamedov, J. P. Wicksted, N. A. Kotov, J. Am. Chem. Soc. 2005, 127, 3463- 3472. [424] S. Bandow, M. Takizawa, K. Hirahara, M. Yudasaka, S. Iijima, Chem. Phys. Lett. 2001, 337, 48-54. [425] M. S. Dresselhaus, G. Dresselhaus, A. Jorio, A. G. Souza Filho, R. Saito, Carbon 2002, 40, 2043-2061. [426] S. Banerjee, S. S. Wong, Nano Lett. 2004, 4, 1445-1450. [427] S. Attal, R. Thiruvengadathan, O. Regev, Anal. Chem. 2006, 78, 8098-8104. [428] M. Makha, Y. Alias, C. L. Raston, A. N. Sobolev, New J. Chem. 2008, 32, 83- 88. [429] R. Ionescu, E. H. Espinosa, E. Sotter, E. Llobet, X. Vilanova, X. Correig, A. Felten, C. Bittencourt, G. V. Lier, J. C. Charlier, J. J. Pireaux, Sens. Actuators, B 2006, 113, 36-46. [430] A. E. O'Keefe, G. C. Ortman, Anal. Chem. 1966, 38, 760-763.

261 References

[431] D. E. Tevault, K. Y. Ong, M. B. Wasserman, W. Ahearn, U.S. Army Edgewood Chemical and Biological Center, Maryland, 2001, pp. 1-23. [432] B. Mayer, C. C. Collins, M. Walton, J. Vac. Sci. Technol., A 2001, 19, 329-344. [433] A. R. Hopkins, N. S. Lewis, Anal. Chem. 2001, 73, 884-892. [434] S. V. Patel, T. E. Mlsna, B. Fruhberger, E. Klaassen, S. Cemalovic, D. R. Baselt, Sens. Actuators, B 2003, 96, 541-553. [435] T. E. Mlsna, S. Cemalovic, M. Warburton, S. T. Hobson, D. A. Mlsna, S. V. Patel, Sens. Actuators, B 2006, 116, 192-201. [436] W. T. Muse, S. J. Anthony, L. C. Buettner, C. L. Crouse, L. C. Crouse, R. J. Mioduszewski, S. A. Thomson, Edgewood Chemical Biological Center, 2002, pp. 1-12. [437] C. L. Raston group., unpublished results. [438] J. R. Hutchins, R. V. Harrington, in Kirk-Othmer encyclopedia of chemical technology, John Wiley & Sons, Inc., 1966, pp. 533-604. [439] J. R. Lewis, in Handbook of stainless steels (Eds.: B. Peckner, I. M. Bernstein), McGraw-Hill, New York, 1977, pp. 19/11-19/36. [440] American Society of Agricultural and Biological Engineers, in Agricultural cabs--environmental air quality, 1997, pp. 404-407. [441] Alicat Scientific Inc., Precision gas flow controller operating manual, Alicat Scientific, Inc., Tucson, 2003. [442] S. Chauhan, S. Chauhan, R. D'Cruz, S. Faruqi, K. K. Singh, S. Varma, M. Singh, V. Karthik, Environ. Toxicol. Pharmacol. 2008, 26, 113-122. [443] M. L. Givens, in Medical response to terrorism: preparedness and clinical practice, 1st ed. (Eds.: D. C. Keyes, J. L. Burstein, R. B. Schwartz, R. E. Swienton), Lippincott Williams & Wilkins, Philadelphia, 2005, pp. 38-45. [444] Ministry of Defence, Medical manual of defence against chemical agents, H.M.S.O., London, UK, 1987. [445] T. C. Marrs, F. R. Sidell, Phosgene in chemical warfare agents, John Wiley, Chichester, 1996. [446] J. Borak, W. F. Diller, J. Occup. Environ. Med. 2001, 43, 110-119. [447] K. E. Jackson, J. Chem. Educ. 1933, 10, 622-626. [448] L. Cotarca, H. Eckert, Phosgenations - a handbook, John Wiley & Sons, UK, 2003. [449] W. F. Diller, W. Schneider, in Ullmann's encyclopedia of industrial chemistry, Vol. A 19, 5th ed., John Wiley & Sons, Inc., 1991, pp. 411-414.

262 References

[450] AIHA, Odor thresholds for chemicals with established occupational health standard, Akron: American Industrial Hygiene Association, 1989. [451] G. Frye-Mason, M. Leuschen, L. Wald, K. Paul, L. F. Hancock, Vol. 5778, SPIE, 2005, pp. 337-346. [452] H. Zhang, D. M. Rudkevich, Chem. Commun. 2007, 1238-1239. [453] D. R. Walt, S. Bencic-Nagale, PCT Int. Appl. WO 2008048698 2008. [454] T. L. Porter, R. Stewart, T. L. Vail, M. P. East, PCT Int. Appl. WO 2008088867, 2008. [455] R. Harig, P. Rusch, C. Dyer, A. Jones, R. Moseley, B. Truscott, Vol. 5995, SPIE, 2005, p. 599510. [456] C. E. Talley, T. R. Huser, C. W. Hollars, S. M. Lane, J. H. Satcher, B. R. Hart, T. A. Laurence, U.S. Pat. Appl. 2006050268 2006. [457] M. H. Wittich, Chemie-Technik (Heidelberg) 1999, 28, 16-18. [458] G. Matz, T. Hunte, W. Schroeder, Field Anal. Chem. Technol. 2000, 4, 195-203. [459] D. Koch, Chemie-Technik (Heidelberg) 2000, 29, 49. [460] R. Bernd, Chemie-Technik (Heidelberg) 2005, 34, 40-41. [461] J. H. Kwak, Y. T. Lim, G. H. Kim, N. J. Choi, H. D. Park, G. H. Rue, J. S. Huh, J. C. Kim, D. D. Lee, Xiyou Jinshu Cailiao Yu Gongcheng 2006, 35, 124-128. [462] G. Jimenez-Cadena, J. Riu, F. X. Rius, Analyst 2007, 132, 1083-1099. [463] A. Star, T.-R. Han, V. Joshi, J. R. Stetter, Electroanal. 2004, 16, 108-112. [464] M. Zhang, M. Yudasaka, S. Bandow, S. Iijima, Chem. Phys. Lett. 2003, 369, 680-683. [465] C. A. Furtado, U. J. Kim, H. R. Gutierrez, L. Pan, E. C. Dickey, P. C. Eklund, J. Am. Chem. Soc. 2004, 126, 6095-6105. [466] K. Lee, J.-w. Lee, K.-y. Dong, B.-k. Ju, Sens. Actuators, B 2008, 135, 214-218. [467] J. Li, Y. Lu, Q. Ye, L. Delzeit, M. Meyyappan, Electrochem. Solid-State Lett. 2005, 8, H100-H102. [468] E. S. Snow, J. P. Novak, P. M. Campbell, D. Park, Appl. Phys. Lett. 2003, 82, 2145-2147. [469] K. Man-Ho, J. G. Charles, N. C. Robert, Rev. Sci. Instrum. 2005, 76, 113904. [470] R. Krupke, F. Hennrich, Eur. Pat. Appl. 1997380 2004. [471] L. A. Currie, Pure & Appl. Chem. 1995, 67, 1699-1723. [472] J. Kong, H. Dai, J. Phys. Chem. B 2001, 105, 2890-2893. [473] C. Y. Lee, S. Baik, J. Zhang, R. I. Masel, M. S. Strano, J. Phys. Chem. B 2006, 110, 11055-11061.

263 References

[474] L. Valentini, C. Cantalini, I. Armentano, J. M. Kenny, L. Lozzi, S. Santucci, Diamond Relat. Mater. 2004, 13, 1301-1305. [475] J. Lui, A. G. Rinzler, H. Dai, J. H. Hafner, R. K. Bradley, P. J. Boul, A. Lu, T. Iverson, K. Shelimov, C. B. Huffman, F. Rodriguez-Macias, Y.-S. Shon, T. R. Lee, D. T. Colber, R. E. Smalley, Science 1998, 280, 1253-1256. [476] M. A. Hamon, H. Hu, P. Bhowmik, S. Niyogi, B. Zhao, M. E. Itkis, R. C. Haddon, Chem. Phys. Lett. 2001, 347, 8-12. [477] S. A. Cucinell, Arch. Environ. Health 1974, 28, 272-275. [478] T. C. Marrs, R. L. Maynard, F. R. Sidell, Chemical warfare agents : toxicology and treatment, 2nd ed., John Wiley & Sons, Ltd, 2007. [479] Health & Safety Executive, EH40/2005 workplace exposure limits, The Stationery Office, London, 2005. [480] Institute of Medicine (U.S.), Chemical and biological terrorism : research and development to improve civilian medical response, National Academy Press, Washington, DC, 1999. [481] J. T. Robinson, F. K. Perkins, E. S. Snow, Z. Wei, P. E. Sheehan, Nano Lett. 2008, 8, 3137-3140. [482] M. J. Ledoux, R. Vieira, C. Pham-Huu, N. Keller, J. Catal. 2003, 216, 333-342. [483] M. Hara, T. Yoshida, A. Takagaki, T. Takata, J. N. Kondo, S. Hayashi, K. Domen, Angew. Chem., Int. Ed. 2004, 43, 2955-2958. [484] A.-H. Lu, W. Schmidt, N. Matoussevitch, H. Bonnemann, B. Spliethoff, B. Tesche, E. Bill, W. Kiefer, F. Schuth, Angew. Chem., Int. Ed. 2004, 43, 4303- 4306. [485] N. Keller, G. Rebmann, E. Barraud, O. Zahraa, V. Keller, Catal. Today 2005, 101, 323-329. [486] Y. Fangli, R. Hojin, Nanotechnology 2004, 15, S596-S602. [487] F. Su, J. Zeng, Y. Yu, L. Lu, J. Y. Lee, X. S. Zhao, Carbon 2005, 43, 2366- 2373. [488] G. S. Chai, S. B. Yoon, J.-S. Yu, Carbon 2005, 43, 3028-3031. [489] K. Yamamoto, S. Akita, Y. Nakayama, J. Phys. D: Appl. Phys. 1998, 31, L34- L36. [490] M. A. de la Casa-Lillo, B. C. Moore, D. Cazorla-Amoros, A. Linares-Solano, Carbon 2002, 40, 2489-2494. [491] S. M. Saufi, A. F. Ismail, Carbon 2004, 42, 241-259. [492] L. Shi, X. Liu, H. Li, W. Niu, G. Xu, Anal. Chem. 2006, 78, 1345-1348.

264 References

[493] H. Xie, Q. Yang, X. Sun, J. Yang, Y. Huang, Sens. Actuators, B 2006, 113, 887- 891. [494] G. D. Pantos, J.-L. Wietor, J. K. M. Sanders, Angew. Chem. Int. Ed. 2007, 46, 2238-2240. [495] J. L. Atwood, M. J. Barnes, M. G. Gardiner, C. L. Raston, Chem. Commun. 1996, 1449-1450. [496] L. J. Barbour, Chem. Commun. 2006, 1163-1168. [497] J. D. Fortner, D. Y. Lyon, C. M. Sayes, A. M. Boyd, J. C. Falkner, E. M. Hotze, L. B. Alemany, Y. J. Tao, W. Guo, K. D. Ausman, V. L. Colvin, J. B. Hughes, Environ. Sci. Technol. 2005, 39, 4307-4316. [498] A. Ikeda, Y. Suzuki, M. Yoshimura, S. Shinkai, Tetrahedron 1998, 54, 2497- 2508. [499] H. Ajie, M. M. Alvarez, S. J. Anz, R. D. Beck, F. Diederich, K. Fostiropoulos, D. R. Huffman, W. Kraetschmer, Y. Rubin, K. E. Schriver, D. Sensharina, R. L. Whetten, J. Phys. Chem. 1990, 94, 8630-8633. [500] R. M. Williams, J. M. Zwier, J. W. Verhoeven, G. H. Nachtegaal, A. P. M. Kentgens, J. Am. Chem. Soc. 1994, 116, 6965-6966. [501] A. Datta, R. Y. Kelkar, M. Datta, J. Mater. Sci. Lett. 1994, 13, 996-999. [502] R. Malhotra, D. C. Lorents, Y. K. Bae, C. H. Becker, D. S. Tse, L. E. Jusinski, E. D. Wachsman, in Fullerenes: synthesis, properties, and chemistry of large carbon clusters (Eds.: G. S. Hammond, V. J. Kuck), American Chemical Society, Washington, D.C., 1992, pp. 127-139. [503] R. Maruyama, J. Appl. Phys. 2003, 94, 6871-6874. [504] F. Tuinstra, J. L. Koenig, J. Chem. Phys. 1970, 53, 1126-1130. [505] D. J. Klionsky, S. D. Emr, Science 2000, 290, 1717-1721. [506] N. Mizushima, Gene Dev. 2007, 21, 2861-2873. [507] Y. Kondo, T. Kanzawa, R. Sawaya, S. Kondo, Nat. Rev. Cancer 2005, 5, 726- 734. [508] M. M. Hippert, P. S. O'Toole, A. Thorburn, Cancer Res. 2006, 66, 9349-9351. [509] O. Seleverstov, O. Zabirnyk, M. Zscharnack, L. Bulavina, M. Nowicki, J.-M. Heinrich, M. Yezhelyev, F. Emmrich, R. O'Regan, A. Bader, Nano Lett. 2006, 6, 2826-2832. [510] O. Zabirnyk, M. Yezhelyev, O. Seleverstov, Autophagy 2007, 3, 278-281. [511] W. Martinet, G. R. Y. De Meyer, L. Andries, A. G. Herman, M. M. Kockx, Autophagy 2006, 2, 55-57.

265 References

[512] J. Liu, H. Dai, J. H. Hafner, D. T. Colbert, R. E. Smalley, S. J. Trans, C. Dekker, Nature 1997, 385, 780-781. [513] L. Song, L. Ci, C. Jin, P. Tan, L. Sun, W. Ma, L. Liu, D. Liu, Z. Zhang, Y. Xiang, S. Luo, X. Zhao, J. Shen, J. Zhou, W. Zhou, S. Xie, Nanotechnology 2006, 17, 2355-2361. [514] T. Vossmeyer, S.-W. Chung, W. M. Gelbart, J. R. Heath, Adv. Mater. 1998, 10, 351-353. [515] R. Martel, H. R. Shea, P. Avouris, J. Phys. Chem. B 1999, 103, 7551-7556. [516] M. Sano, A. Kamino, J. Okamura, S. Shinkai, Science 2001, 293, 1299-1301. [517] M. Ahlskog, E. Seynaeve, R. J. M. Vullers, C. Van Haesendonck, A. Fonseca, K. Hernadi, J. B.Nagy, Chem. Phys. Lett. 1999, 300, 202-206. [518] J. F. Colomer, L. Henrard, E. Flahaut, G. Van Tendeloo, A. A. Lucas, P. Lambin, Nano Lett. 2003, 3, 685-689. [519] Z. Liu, L.-C. Qin, Carbon 2005, 43, 2146-2151. [520] S. Motojima, M. Kawaguchi, K. Nozaki, H. Iwanaga, Appl. Phys. Lett. 1990, 56, 321-323. [521] S. Motojima, M. Kawaguchi, K. Nozaki, H. Iwanaga, Carbon 1991, 29, 379- 385. [522] M. Audier, M. Coulon, Carbon 1985, 23, 317-323. [523] A. W. Czarnik, Chem. Biol. 1995, 2, 423-428. [524] G. Arena, A. Casnati, A. Contino, G. G. Lombardo, D. Sciotto, R. Ungaro, Chem. Eur. J. 1999, 5, 738-744. [525] X. Guo, X. Qian, L. Jia, J. Am. Chem. Soc. 2004, 126, 2272-2273. [526] V. Souchon, I. Leray, B. Valeur, Chem. Commun. 2006, 4224-4226. [527] J. S. Kim, D. T. Quang, Chem. Rev. 2007, 107, 3780-3799. [528] B. Valeur, I. Leray, Inorg. Chim. Acta 2007, 360, 765-774. [529] L. Fabbrizzi, A. Poggi, Chem. Soc. Rev. 1995, 24, 197-202. [530] Y. J. Jang, B.-S. Moon, M. S. Park, B.-G. Kang, J. Y. Kwon, J. S. J. Hong, Y. J. Yoon, K. D. Lee, J. Yoon, Tetrahedron Lett. 2006, 47, 2707-2710. [531] I. A. Koshets, Z. I. Kazantseva, Y. M. Shirshov, S. A. Cherenok, V. I. Kalchenko, Sens. Actuators, B 2005, 106, 177-181. [532] A. Bouazizi, H. Maaref, N. Jaffrezic-Renault, R. B. Chaabane, M. Gammoudi, G. Guillaud, Sens. Actuators, B 1996, 34, 487-489. [533] S. K. Menon, P. Jose, U. Harikrishnan, U. Pal, Indian J. Chem., Sect A 2008, 47A, 246-250.

266 References

[534] H. Li, Y. Zhang, X. Wang, D. Xiong, Y. Bai, Mater. Lett. 2007, 61, 1474-1477. [535] P. D. Beer, P. A. Gale, Angew. Chem. Int. Ed. 2001, 40, 486-516. [536] C. J. Woods, S. Camiolo, M. E. Light, S. J. Coles, M. B. Hursthouse, M. A. King, P. A. Gale, J. W. Essex, J. Am. Chem. Soc. 2002, 124, 8644-8652. [537] S. Arimori, M. G. Davidson, T. M. Fyles, T. G. Hibbert, T. D. James, G. I. Kociok-Kohn, Chem. Commun. 2004, 1640-1641. [538] C. R. Bondy, P. A. Gale, S. J. Loeb, J. Am. Chem. Soc. 2004, 126, 5030-5031. [539] S. H. Lee, S. H. Kim, S. K. Kim, J. H. Jung, J. S. Kim, J. Org. Chem. 2005, 70, 9288-9295. [540] R. Miao, Q.-Y. Zheng, C.-F. Chen, Z.-T. Huang, Tetrahedron Lett. 2005, 46, 2155-2158. [541] A. L. Koner, J. Schatz, W. M. Nau, U. Pischel, J. Org. Chem. 2007, 72, 3889- 3895. [542] R. Kumar, V. Bhalla, M. Kumar, Tetrahedron 2008, 64, 8095-8101. [543] S. J. Brooks, C. Caltagirone, A. J. Cossins, P. A. Gale, M. Light, Supramol. Chem. 2008, 20, 349-355. [544] A. Coleman, F. Perret, A. Moussa, M. Dupin, Y. Guo, H. Perron, Top. Curr. Chem. 2007, 277, 31-88. [545] N. Korbakov, P. Timmerman, N. Lidich, B. Urbach, A. Sa'ar, S. Yitzchaik, Langmuir 2008, 24, 2580-2587. [546] F. Phillips, K. Kaczor, N. Gandhi, B. D. Pendley, R. K. Danish, M. R. Neuman, B. Tóth, V. Horváth, E. Lindner, Talanta 2007, 74, 255-264. [547] H.-F. Ji, R. Dabestani, G. M. Brown, R. L. Hettich, Photochem. Photobiol. 1999, 69, 513-516. [548] I. Leray, F. O'Reilly, J.-L. Habib Jiwan, J.-P. Soumillion, B. Valeur, Chem. Commun. 1999, 795-796. [549] T. Jin, Chem. Commun. 1999, 2491-2492. [550] I. Leray, J.-P. Lefevre, J.-F. Delouis, J. Delaire, B. Valeur, Chem. Eur. J. 2001, 7, 4590-4598. [551] I. Leray, Z. Asfari, J. Vicens, B. Valeur, J. Chem. Soc., Perkin Trans. 2 2002, 10-15. [552] L. Tarazi, A. George, G. Patonay, L. Strekowski, Talanta 1998, 46, 1413-1424. [553] K. Rurack, M. Kollmannsberger, U. Resch-Genger, J. Daub, J. Am. Chem. Soc. 2000, 122, 968-969. [554] O. A. Kharenko, M. Y. Ogawa, J. Inorg. Biochem. 2004, 98, 1971-1974.

267 References

[555] H.-F. Ji, Y. Yang, X. Xu, G. Brown, Org. Biomol. Chem. 2006, 4, 770-772. [556] D. T. Quang, H. S. Jung, J. H. Yoon, S. Y. Lee, J. S. Kim, Bull. Korean Chem. Soc. 2007, 28, 682-684. [557] X. Qi, S. K. Kim, E. J. Jun, L. Xu, S.-J. Kim, J. Yoon, Bull. Korean Chem. Soc. 2007, 28, 2231-2233. [558] S. P. G. Costa, E. Oliveira, C. Lodeiro, M. M. M. Raposo, Sensors 2007, 7, 2096-2114. [559] H. Vanhoe, C. Vandecasteele, J. Versieck, R. Dams, Anal. Chem. 1989, 61, 1851-1857. [560] D. A. Skoog, F. J. Holler, T. A. Nieman, in Principles of instrumental analysis, 5th ed., Saunders College Publishers, 1998. [561] K. H. Theimer, V. Krivan, Anal. Chem. 1990, 62, 2722-2727. [562] R. Tavichakorntrakool, V. Prasongwattana, P. Sriboonlue, A. Puapairoj, C. Wongkham, T. Wiangsimma, W. Khunkitti, S. Triamjangarun, M. Tanratanauijit, A. Chamsuwan, W. Khunkitti, P.-T. Yenchitsomanus, V. Thongboonkerd, Transl. Res. 2007, 150, 357-366. [563] J. Shim, E.-J. Jang, K.-C. Chung, Anal. Lett. 2007, 40, 3038-3049. [564] M. Arvand, S. A. Asadollahzadeh, Talanta 2008, 75, 1046-1054. [565] M. R. Ganjali, P. Norouzi, F. Faridbod, N. Hajiabdollah, R. Dinarvand, S. Meghdadi, Anal. Lett. 2008, 41, 3-23. [566] R. Metivier, I. Leray, B. Lebeau, B. Valeur, J. Mater. Chem. 2005, 15, 2965- 2973. [567] F. Mancin, E. Rampazzo, P. Tecilla, U. Tonellato, Chem. Eur. J. 2006, 12, 1844-1854. [568] N. M. Buie, V. S. Talanov, R. J. Butcher, G. G. Talanova, Inorg. Chem. 2008, 47, 3549-3558. [569] A. Casnati, F. Sansone, A. Sartori, L. Prodi, M. Montalti, N. Zaccheroni, F. Ugozzoli, R. Ungaro, Eur. J. Org. Chem. 2003, 2003, 1475-1485. [570] S. Y. Park, J. H. Yoon, C. S. Hong, R. Souane, J. S. Kim, S. E. Matthews, J. Vicens, J. Org. Chem. 2008, 73, 8212-8218. [571] N. Micali, V. Villari, G. M. L. Consoli, F. Cunsolo, C. Geraci, Phys. Rev. E: Stat. Phys., Plasmas, Fluids, 2006, 73, 051904. [572] I. Oueslati, A. W. Coleman, B. de Castro, M. N. Berberan-Santos, J. Fluoresc. 2008, 18, 1123-1129.

268 References

[573] Y. Zhang, R. A. Agbaria, N. E. Mukundan, I. M. Warner, J. Inclusion Phenom.Macrocyclic Chem. 1996, 24, 353-365. [574] A. N. Lazar, N. Dupont, A. Navaza, A. W. Coleman, Chem. Commun. 2006, 1076-1078. [575] M. Kellermann, W. Bauer, A. Hirsch, B. Schade, K. Ludwig, C. Böttcher, Angew. Chem. Int. Ed. 2004, 43, 2959-2962. [576] M. Strobel, K. Kita-Tokarczyk, A. Taubert, C. Vebert, P. A. Heiney, M. Chami, W. Meier, Adv. Funct. Mater. 2006, 16, 252-259. [577] K. Nesmerak, I. Nemcova, Anal. Lett. 2006, 39, 1023-1040. [578] S. Shinkai, Pure & Appl. Chem. 1986, 58, 1523-1528. [579] J. L. Atwood, D. L. Clark, R. K. Juneja, G. W. Orr, K. D. Robinson, R. L. Vincent, J. Am. Chem. Soc. 1992, 114, 7558-7559. [580] S. Houmadi, D. Coquiere, L. Legrand, M. C. Faure, M. Goldmann, O. Reinaud, S. Remita, Langmuir 2007, 23, 4849-4855. [581] S. G. Bott, A. W. Coleman, J. L. Atwood, J. Am. Chem. Soc. 1988, 110, 610- 611. [582] J. L. Atwood, A. W. Coleman, H. Zhang, S. G. Bott, J. Inclusion Phenom. Macrocyclic Chem. 1989, 7, 203-211. [583] L. J. Barbour, A. K. Damon, G. W. Orr, J. L. Atwood, Supramol. Chem. 1996, 7, 209-213. [584] J. L. Atwood, T. Ness, P. J. Nichols, C. L. Raston, Cryst. Growth Des. 2002, 2, 171-176. [585] E. N. Savariar, S. V. Aathimanikandan, S. Thayumanavan, J. Am. Chem. Soc. 2006, 128, 16224-16230. [586] D. M. Chipman, V. Grisaro, N. Sharon, J. Biol. Chem. 1967, 242, 4388-4394. [587] A. C. Tedesco, D. M. Oliveira, Z. G. M. Lacava, R. B. Azevedo, E. C. D. Lima, P. C. Morais, J. Magn. Magn. Mater. 2004, 272-276, 2404-2405. [588] K. A. Connors, Binding constants : the measurement of molecular complex stability, Wiley, New York, 1987. [589] P. Metilda, K. Prasad, R. Kala, J. M. Gladis, T. P. Rao, G. R. K. Naidu, Anal. Chim. Acta 2007, 582, 147-153. [590] J. Liu, A. K. Brown, X. Meng, D. M. Cropek, J. D. Istok, D. B. Watson, Y. Lu, Proc. Nat. Acad. Sci. U.S.A. 2007, 104, 2056-2061.

269

Appendices

Appendices

9 Appendices

Appendix A.1 Raman analysis protocol – R-3000CN portable Raman spectrometer.

A protocol for the analysis of drop-cast and air-dried residues containing SWCNT material, utilising the above mentioned portable Raman spectrometer, is detailed as follows:

1. Once the instrument is turned on and RSIScan® software is open and the system is calibrated with the Teflon® standard (verification cap), adjust the laser power setting to <10 mW. Caution: at laser power >10 mW the SWCNTs will disintegrate.

2. The SWCNT residues should be on a flat surface, for these studies a flat disc with a single layer of aluminium foil was utilised. The sample disc is mounted on a laboratory jack, for manual z-axis height control.

3. The Raman probe is mounted in a clamp on a retort stand, such that the tip of the probe is facing the sample disc; this allows manual adjustment of the probe in the x- and y-axis for sample alignment.

4. Remove the verification cap and align the Raman probe and sample disc, such that the probe tip has a clear pathway to the particular sample of interest. Note: that the solid cap can be utilised at this stage to approximate the necessary z-axis height of the probe, as the laser is focused at the very tip of the solid cap. At this stage the probe should typically be in a position to allow the acquisition of a Raman spectrum.

5. Set the RSIScan® software to run in Focus mode; the x-axis units are the pixels from the detector which there are 1000. The axis can be adjusted by striking the Pause button on the keyboard, with typical settings of x-axis 100-1000, an integration time of 1 s. With the lower x-axis limit of 100 to start acquiring data past the laser reflectance. Caution: Raman laser safety glasses should be warn at all time due to the laser being entirely exposed utilising this protocol.

6. The y-axis varies experimentally, once a spectrum becomes apparent this axis can be manually scaled for the appropriate intensity range, at this stage the x- and y-axis will be in an appropriate position.

271 Appendices

7. The z-axis height of the position can now be manually varied to allow maximum peak intensity for a given sample. The laser focal point is 4.6 mm from the probe lens, typically start at a slightly greater distance and slowly optimise the sample height.

8. The probe and sample are in the optimal position and an appropriate Raman spectrum can be obtained.

272 Appendices

Figure A.2 PXRD pattern of the product of 6 in dichloromethane, demonstrating a hcp crystal

structure of C60.

Figure A.3 IR spectrum of the product of the argon annealed analogue (650 °C), demonstrating the

typical C60 absorbance peaks.

273 Appendices

Figure A.4 PXRD pattern of pristine C60 (blue), and the 700 °C argon annealed product (red).

Figure A.5 Emission profiles (λex = 423 nm) for solutions of increasing concentration of 4C; concentrations below and above the CMC utilised at detector voltage 800 V.

274

Publications

Publications

PAPER www.rsc.org/materials | Journal of Materials Chemistry Selective diameter uptake of single-walled carbon nanotubes in water using phosphonated calixarenes and ‘extended arm’ sulfonated calixarenes†

Lee J. Hubble,ab Thomas E. Clark,a Mohamed Makhaa and Colin L. Raston*a

Received 26th August 2008, Accepted 17th September 2008 First published as an Advance Article on the web 4th November 2008 DOI: 10.1039/b814904f

Single-walled carbon nanotubes (SWCNTs) have been successfully solubilized using water-soluble p-phosphonated calix[n]arenes (n ¼ 4, 6, 8) and ‘extended arm’ upper rim functionalized (benzyl, phenyl) p-sulfonated calix[8]arenes. Selective SWCNT diameter solubilization has been demonstrated and subsequent preferential enrichment of SWCNTs with semiconducting or metallic electronic properties has been achieved. These water-soluble supramolecular systems can be incorporated into post-growth purification protocols with direct implications in areas such as nano-electronics and device fabrication.

Introduction underlying problems with the use of amphiphilic polymers.26 Supramolecular systems that can be utilized to solubilize Since the discovery of carbon nanotubes (CNTs) by Iijima1 and 2,3 SWCNTs in aqueous media are potentially more benign, more recently single-walled carbon nanotubes (SWCNTs), generating less waste, and are thus more attractive. these purely carbon-based materials have attracted significant Calixarenes have played an important role in supramolecular attention from both the research and commercial sectors. CNTs chemistry in general. We have shown a unique structural have extraordinary mechanical and unique electrical properties organization of fullerene C60, with surface curvature analogous with major research efforts focused on areas such as high to SWCNTs, in toluene to form nano-fibers.27 This selectivity performance electronics,4 scanning probe microscopy,5,6 fuel 7 8,9 10,11 12,13 14–16 and self-assembly process may be mimicked in water using cells, composites, chemical, biological and physical p-phosphonated calixarene systems as a route to diameter- sensors, and many more. In order to harness the full potential of selective purification of SWCNTs in aqueous media. The most CNTs, the ability to separate them according to diameter and/or 4 widely studied water-soluble calixarenes are the p-sulfonato chirality is required. The diameter and inherent chirality of derivatives, because of their bio-compatibility and simplicity SWCNTs directly controls their semiconducting or metallic 28–30 17 of synthesis. Recently we reported a general route for the properties. synthesis of p-phosphonic acid calix[n]arenes (n ¼ 4, 6, 8), Recent methods have been trialed to separate CNTs including which can be engineered into unique supramolecular frame- pioneering work by Bao et al. which utilises different absorption works at the nanometer scale.31 The combination of water properties of metallic and semiconducting SWNTs towards solubility and self-assembly is of interest for not only the amine- and phenyl-terminated silanes to produce a one-step 18 solubilization of SWCNTs but also the possibility of creating methodology for self-sorting and aligned thin-film transistors. different SWCNT-calixarene arrays to facilitate the separation Other methods include dielectrophoresis which takes advantage of the SWCNTs. of the difference in relative dielectric constants of metallic and 19 There have been limited studies on the specific use of water- semiconducting CNTs in respect to the choice of solvent. soluble calixarenes for the solubilization of SWCNTs. Lobach Pathways utilizing differential density gradient systems have et al. investigated various types of amphiphilic macrocycles, also been explored.20,21 More recent research has targeted non- 22 such as different upper and lower rim functionalized calix[4]- covalent supramolecular interactions, as in biopolymers, DNA ¼ 23 24 resorcinarene and p-sulfonato-calix[n]arenes (n 4, 6), modified wrapping, porphyrinic polypeptides and oligo-acene 32 25 by lower rim alkyl substituents (butyl and dodecyl), with the adducts. The problem with some of these methods is that the ability to form stable suspensions of SWCNTs in aqueous media. supramolecular chemistry is carried out in organic solvents with Furthermore, Ogoshi et al. focused on p-sulfonato-calix[n]arenes limited stability in aqueous media, and requires lengthy experi- (n ¼ 4, 6, 8), with successful solubilization of SWCNTs with mental procedures. Progress has been made to combat these the higher order oligomers, with p-sulfonato-calix[8]arene forming a dimer as part of a supramolecular SWCNT aCentre for Strategic Nano-Fabrication, School of Biomedical, network polymer.33 Biomolecular and Chemical Sciences, The University of Western Herein, we report the solubilization of SWCNTs using the Australia, 35 Stirling Highway, Crawley, WA, 6009, Australia. E-mail: [email protected]; Fax: (+61) 8-6488-8683; Tel: (+61) water-soluble p-phosphonated calixarenes and analogous 8-6488-3045 compounds of differing lower rim functionality. The findings bCentre for Forensic Science, The University of Western Australia, 35 include a preferential solubilization of specific diameter Stirling Highway, Crawley, WA, 6009, Australia SWCNTs, as supported by Raman spectroscopy measurements. † Electronic supplementary information (ESI) available: Raman We also report a detailed investigation into the effects of using spectrum of aluminium foil sample holder, UV-visible spectra for the calixarene/SWNT systems supernatants and standard calixarene ‘extended arm’ p-sulfonated calix[8]arene systems on SWCNT solutions. See DOI: 10.1039/b814904f solubilization, for comparison to the earlier studies on the parent

This journal is ª The Royal Society of Chemistry 2008 J. Mater. Chem., 2008, 18, 5961–5966 | 5961

276 Publications

p-sulfonated calixarenes.32,33 The ‘extended arm’ calixarenes have a higher hydrophobic surface area relative to the parent p-sulfonated calixarene.

Experimental Materials and methods SWCNTs (Elicarb) were purchased from Thomas Swan & Co. (UK) which were produced by a catalytic chemical vapour Fig. 2 Photographs of (A) a solution containing SWCNT/water (left) deposition synthesis method with an as-received purity >90% and a solution containing 2A/SWCNT/water (right); (B) supernatant of (4.79 wt% ash, 2.76 wt% iron catalyst, <2nm tube diameter). 2A/SWCNT/water system. The calixarenes utilized in this study are detailed in Fig. 1 and ¼ Table 1. They included p-phosphonic acid calix[n]arenes (n 4, Raman spectroscopy was performed using a Raman Systems, 6, 8), 1A, 2A, and 3A, p-phosphonic acid calix[4]arene modified R-3000CN portable Raman spectrometer. The spectral range by octadecyl and butyl substituents on the lower rim, 1B covered was 100 to 1800 cm1, with an excitation wavelength of and 1C, p-phosphonic acid calix[6]arene modified by methyl 785 nm, and laser power <10 mW. Spectra were acquired from substituents on the lower rim, 2B, p-sulfonato-calix[8]arene, multiple areas and rotations; the samples were prepared by 4A, p-benzyl-sulfonato-calix[8]arene, 4B, and p-phenyl- drop-coating the resulting supernatants onto aluminium foil and sulfonato-calix[8]arene, 4C. allowing them to air dry. Raman spectra of calixarene control Compounds 1A–3A were prepared through recently published 31 samples were also prepared and collected using this method- synthetic pathways, with 1B, 1C and 2B being synthesized in ology. UV-visible spectrophotometry was performed with our laboratory.34 The p-sulfonato-calixarene (4A) was synthe- 35 a Varian, Cary 50 Tablet spectrophotometer. The spectral range sized through a modified preparation, with the ‘extended covered was 800 to 200 nm, with a scan rate of 600 nm/min; the arm’ p-sulfonato-calixarenes (4B and 4C) synthesized through 36,37 samples were prepared by dilution (100 ) of the resulting published procedures. supernatants, and were analysed in quartz cuvettes with a 10 mm A typical experiment consisted of the addition of as-received path length. Transmission electron microscopy (TEM) was SWCNTs (1.0 mg) into a sample vial (10 mL) containing Milli-Q performed on a JEOL 2100 TEM operating at 120kV; sample U 1 water (6 mL, 18 M cm ). To this mixture the appropriate preparation involved centrifugation of the supernatant (100 2 calixarene (2.60 10 mmol) was added. This mixture was dilution) and re-dispersing the resultant plug in Milli-Q water, ultrasonicated for 10 minutes with a sonic lance (4W RMS from which a drop of the solution was placed onto a holey power, Branson Sonifier 150) and a further 30 minutes (5W RMS carbon film supported by copper grids and air-dried. power), affording black to light grey solutions (Fig. 2A). To remove impurities and large agglomerations the solution was divided into 1.5 mL aliquots and centrifuged (5000RCF, Results and discussion Eppendorf Mini Spin plus) for 30 minutes in ambient conditions. All the calixarenes studied are essentially acting as surfactant The top three quarters of the supernatant was collected and molecules, in solubilizing the SWCNTs. It is interesting to note combined, yielding the solubilized product (Fig. 2B). that during ultrasonication all the solutions containing the calixarenes formed a black solution (Fig. 2A), with exception of 1A and 1B which afforded light grey suspensions. In comparison, the SWCNT/water experiment devoid of calixarenes formed large aggregates of the SWCNT material (Fig. 2A). The trans- formation from colourless to black and/or grey indicates the effective solubilization of the SWCNT material into the aqueous media. This is further supported with the intense coloration Fig. 1 Principle calixarene structure. maintained in the supernatant after centrifugation (Fig. 2B). It should be noted that the collected supernatant suspensions of 1A–4C are highly homogeneous and are stable for over six Table 1 Notation for calix[n]arene systems months, and thus the p-phosphonated calixarenes and ‘extended arm’ p-sulfonated calixarenes are effective in solubilizing and XYn Notation stabilizing SWCNTs in aqueous media.

PO3H2 OH 4 1A PO H C H 4 1B 3 2 18 37 Raman spectroscopy PO3H2 OBu 4 1C PO H OH 6 2A 3 2 To obtain a greater understanding of the material solubilized by PO3H2 OMe 6 2B PO3H2 OH 8 3A the calixarene systems, the supernatant solutions were dried and SO3HOH84A the resulting residues were analyzed using Raman spectroscopy. Benzyl-SO HOH 84B 3 Raman scattering produces characteristic peaks from certain Phenyl-SO3HOH 84C phonon modes in carbon nanotubes. The frequencies of

5962 | J. Mater. Chem., 2008, 18, 5961–5966 This journal is ª The Royal Society of Chemistry 2008

277 Publications

immediate interest revolve around the radial breathing modes (RBMs) and the G- and D-bands. The low frequency RBMs (100–300cm1) are directly dependent on the nanotube diameters within a sample.38 The high-frequency band at 1340cm1 is commonly referred to as the D-band (disorder-induced band) and provides information regarding amorphous impurities and carbon nanotube wall disorder. The high-frequency band at 1585cm1 is commonly referred to as the G-band (graphite band) and relates to the graphite E2g symmetry of the interlayer mode. This mode represents the structural intensity of the sp2-hybridized carbon atoms of the CNTs and is sensitive to the nanotube surroundings. It is common to utilise the calculated

ID/IG ratio to assess the quality of a CNT sample following a purification procedure.39 It should be noted that Raman spectra were obtained for all calixarene control samples, and were found to not be Raman active at the laser frequency utilised in this study (see ESI).† The spectral shifts in the calixarene/SWCNT systems are detailed in relation to the as-received SWCNT sample (Table 2). All of the calixarene systems produce an up-field shift in the G-band frequency relative to the as-received SWCNT sample (Fig. 3), shown in Fig. 4A and 5A. For the p-phosphonated calixarene analogues the largest shift was seen for 2B, with the smallest shift seen for 1B. For the latter, this may be related to the large hydrophobic tail decreasing its solubility in aqueous media. The hierarchy of frequency shift is 1B < 1C <

I I Table 2 Raman spectral G-band frequency shifts and the D/ G ratios Fig. 4 (A) Raman spectra obtained for the phosphonated calixarene for the residues of the calixarene/SWCNT supernatants systems, (B) Raman spectra highlighting the RBM bands.

1 1 Sample G-band frequency/cm Frequency shift/cm ID/IG ratio

As-received 1583.68 — 0.2131 1A, 2A, 3A < 2B (Table 2), and this gives some insight into SWCNTs the relative strength of the interaction between the respective 1A 1592.48 +8.80 0.1613 1B 1588.08 +4.40 0.1816 p-phosphonated calixarene and the SWCNTs. The results show 1C 1588.96 +5.28 0.0729 that the interplay between the calixarenes and SWCNTs leads to 2A 1592.48 +8.80 0.1559 their solubilization into the aqueous media.40 2B 1594.24 +10.56 0.1350 3A 1592.48 +8.80 0.2449 Similarly, with the p-sulfonated calixarene systems we 4A 1592.48 +8.80 0.1277 observed significant G-band shifts with the largest up-field shifts 4B 1602.15 +18.47 0.2514 attributed to the ‘extended arm’ p-sulfonated calix[8]arene 4C 1602.15 +18.47 0.1440 systems (4B and 4C) (Fig. 5A). This could be due to more efficient packing of the calixarenes around the SWCNT surface. The conformational flexibility of the larger ring size calixarene with a larger hydrophobic surface area is likely to encourage p–p stacking of the aromatic units of the benzyl and phenyl upper rim functionalities with the SWCNT surface. The results indicate that the peak width of the D-band at 1590 cm1 has decreased significantly from 100 cm1 in the as-received SWCNTs to 20–40 cm1 (FWHM) in all tested systems (Fig. 4A and 5A). This peak width narrowing is attrib- uted to the reduction of the amount of amorphous carbon in the sample, and indicates that all of the systems studied show a low binding affinity to amorphous carbon, and when coupled with centrifugal processing results in minimal amorphous carbon present in the aqueous supernatant. This supports the use of these water-soluble calixarenes in post-growth purification protocols. Fig. 3 Raman spectrum of as-received SWCNT, inset of the radial It is important to measure the purity of the SWCNTs that have breathing mode frequency range. been solubilized using the calixarene host molecules. A method

This journal is ª The Royal Society of Chemistry 2008 J. Mater. Chem., 2008, 18, 5961–5966 | 5963

278 Publications

1 where nRBM is the RBM frequency in cm , dt is the tube diameter in nm, and the parameters of A ¼ 223.5 cm1 nm and B ¼ 12.5 cm1 have been experimentally determined.42 The Raman spectrum, including the RBM region, for the as-received SWCNT is detailed in Fig. 3. It should be noted that several recurring frequencies in the RBM range were identified and removed from the analysis. These could be contributed from either the aluminium foil, possible laser reflections and/or the sampling environment (see ESI).† The dominant SWCNT diameters in the sample are 1.61 nm (151.25 cm1), 1.19 nm (200.49 cm1), 1.05 nm (225.99 cm1) and 0.90 nm (262.04 cm1) (Fig. 3). If a broad RBM band is observed it is possible that it is a superposition of a few Lorentzian compo- nents and could represent multiple diameter SWCNTs.43 This is possibly what is occurring with the 151.25 cm1 band, which is subsequently masking the larger diameter SWCNTs individual RBM bands. The spectra obtained from the p-phosphonated calixarene systems are detailed in Fig. 4A and B. The p-sulfonated calixarene systems spectra are detailed in Fig. 5A and B. A summary of the calculated SWCNT diameters present in the supernatants in each system is given in Table 3. Firstly, the data support the previous findings on solubilization of the SWCNTs. The data also indicate that there is selective uptake of particular diameter SWCNTs, which is highly dependent on the nature of the calixarene. In the majority of the supernatants the most intense band can be attributed to SWCNT with a diameter of 0.89 nm (ca. 262 cm1). Interestingly, this SWCNT diameter is not the most abundant in the as-received SWCNT sample (Fig. 3), yet this diameter seems to be efficiently solubilized, with Fig. 5 (A) Raman spectra obtained for the sulfonated calixarene the uptake in solution of larger diameter tubes limited. systems, (B) Raman spectra highlighting the RBM bands. As previously mentioned the smaller diameter SWCNTs were more readily solubilized into aqueous media. However, the data indicate that there are larger SWCNT diameters solubilized when using 4B (1.55 and 1.42 nm) and 4C (1.35 nm). In the case of 4B routinely utilized is the ratio from the intensity of the D-band to 39 and 4C involving the ‘extended arm’ sulfonated calix[8]arenes, that of the G-band (I /I ). A lower value of I /I implies less D G D G the large G-band shifts indicate strong nanotube surface inter- disordered carbon attributed to the SWCNTs structure and actions. This suggests a different type of interplay with the hence an increased level of pristine SWCNTs. The as-received SWCNT surface, in solubilizing the larger diameter SWCNTs. SWCNTs have a calculated I /I ratio of 0.2131. The I /I D G D G This is consistent with the increased conformational flexibility of ratios are detailed in Table 2, with the hierarchy of I /I ratios as D G the larger ring size calixarene. follows: 4B > 3A > SWCNT as-received > 1B > 1A > 2A > 4C > Data in Table 3 also indicate that 1C and 2A are unique in 2B > 4A > 1C. This shows that the majority of the calixarene regard to SWCNT solubilization. 1C shows the ability to systems increase the level of pristine SWCNTs in the superna- tant, with the exception of 3A, 4B. The ID/IG ratios for the 3A and 4B systems, albeit lower in purity, are relatively close to the as-received SWCNT ratio. Interestingly, the 3A and 4B systems Table 3 Calculated SWCNT diameters present based on RBM frequencies in the Raman spectra reveal large G-band shifts, indicating significant surface inter- action with the SWCNT surface. These systems may also bind to Calculated SWCNT carbon particle impurities which can result in their disaggrega- Sample diameter distributiona/nm tion and lead to an increased D-band signal.40 These data further As-received SWCNTs 1.61, 1.19, 1.05, 0.89 support the use of the majority of these calixarene systems in 1A 1.01, 0.89 post-growth purification protocols. 1B 1.01, 0.89 The radial breathing mode (RBM) frequencies assigned within 1C 1.04, 0.89 a Raman spectrum can provide direct information regarding the 2A 0.89 41 2B 1.00, 0.89 distribution of SWCNT diameters within a sample. The rela- 3A 1.00, 0.89 tionship between the RBM frequency and the inverse nanotube 4A 1.01, 0.89 diameter is well described in the literature: 4B 0.91, 0.89 4C 1.35, 1.04, 0.89

A a nRBM ¼ þ B SWCNT diameter relating to the most intense band is italicized. dt

5964 | J. Mater. Chem., 2008, 18, 5961–5966 This journal is ª The Royal Society of Chemistry 2008

279 Publications

Table 4 Observed RBM bands in the as-received SWCNTs and calix- In relation to the electronic nature of the solubilized SWCNTs arene system supernatants with topographical and electronic assignment the results obtained from the p-sulfonated calixarene systems were similar to the p-phosphonated calixarene systems. However, Calculated Observed RBM SWCNT Assignment Semiconducting it is interesting to note that 4C preferentially binds two different band/cm1 1 diameter/nm (n,m) (S) or metallic (M) diameter metallic SWCNTs, along with the typical 0.89nm SWCNTs. The intensities of these respective RBM bands are 264 0.89 (10,2) S 257 0.91 (11,1) S relatively similar which suggests that there is an enrichment of (9,4) S metallic SWCNTs in the supernatant for this system. 235 1.00 (12,1) S 232 1.01 (11,3) S 228 1.04 (12,2) S UV-visible spectrophotometry (9,6) M 226 1.05 (10,5) S In a recent study, Attal et al. utilised UV-visible absorbance 200 1.19 (15,0) S spectrophotometry to determine the concentration of SWCNTs 178 1.35 (15,2) S in aqueous dispersions. The absorption peak at 273 nm is (14,5) M 151 1.61 (18,4) S a signature of the surface p-plasmon excitation of dispersed SWCNTs.46 The supernatants obtained for the different calix- arene systems, in this study, were analysed using this technique. preferentially bind to 1.04 nm (ca. 228 cm1) diameter SWCNTs, In the majority of the systems slight increases in absorbance at with the band having the greater RBM intensity. This diameter the 273 nm peak were seen, when compared to solutions only represents the smallest SWCNT diameter population in the containing the calixarene. However, these results are deemed p as-received SWCNTs (Fig. 3). In contrast, 2A only solubilizes inconclusive as the -plasmon surface excitations of the calix- SWCNTs with a diameter 0.89 nm (ca. 264 cm1). These two arene adsorb strongly in the 273 nm region and it is possible the systems not only significantly increase the purity of the SWCNTs contribution from the SWNTs has been masked (see ESI).† (Table 2), they can also be utilised to preferentially solubilize SWCNTs of a specific diameter. Transmission electron microscopy With the SWCNT diameter distribution in both the as-received SWCNT sample and the different calixarene system supernatants, Transmission electron microscopy (TEM) was utilised to assess it is now possible to calculate integers of assignment (Table 4): the residues of the supernatants to give information on the solubilization characteristics and also reveal possible packing 2 2 1/2 dt ¼ a(n + m + nm) / p arrangement in the solid state that may relate to arrangement in the solubilized phase. TEM was carried out on the supernatants where a ¼ 0.249 nm (due to a ¼ O3ac-c ¼ 0.249nm [ac-c ¼ of the calixarene systems, with 1C results detailed in Fig. 6 and 7. 0.142nm]) and the integers (n,m) define the structure of The TEM image in Fig. 6 illustrates the typical structures a SWCNT in terms of its diameter and chiral angle.44 Further- observed on the grid, where networks of SWCNTs can be seen more, following the topographical assignment, the electronic scattered with amorphous coatings of the calixarene sheathing nature of each specified nanotube topology can be determined the SWCNTs. As expected the SWCNTs are observed to be (Table 4). There are three possible electronic assignments: 1) individual (Fig. 7), indicating the disruption of van der Waals quasi-metallic SWCNT with |n m| ¼ 3q; 2) metallic SWCNT forces between SWCNT bundles and subsequent self assembly of with n ¼ m and 3) semiconducting SWCNT with |n m| ¼ 3q the calixarene around a SWCNT surface for the effective solu- 1, where q is an integer.44 bilization into aqueous media. These results provide support for From the chiral assignment and subsequently the electronic the increase in pristine SWCNT purity (Table 2). Energy nature of the SWCNTs in the supernatants (Table 4), it is possible to comment on the preferential solubilization of semi- conducting and metallic SWCNTs. The majority of the SWCNTs analysed are semiconducting in nature, which is expected from using a 785 nm excitation wavelength, where excitation is primarily resonant with the v2 / c2 transitions.45 In relation to the p-phosphonated calixarenes, all systems preferentially solubilized 0.89 nm, (10,2) semiconducting SWCNTs based on RBM intensities. For the 1C system, while still solubilizing 0.89 nm SWCNTs, solubilization of the 1.04 nm, (12,2) semi- conducting or (9,6) metallic SWCNTs occurs. This is the only p-phosphonated calixarene to bind to metallic SWCNTs. Most p-phosphonated systems solubilize two to three different diam- eter SWCNTs, whereas 2A only solubilized 0.89 nm (10,2) semiconducting SWCNTs. This selective binding to specific chiral SWCNTs has significant implications for post-growth purification and application selective processing for many current and future applications, in particular nano-electronics.4 Fig. 6 TEM image illustrating SWCNT networks coated with 1C.

This journal is ª The Royal Society of Chemistry 2008 J. Mater. Chem., 2008, 18, 5961–5966 | 5965

280 Publications

10 J. Kong, N. R. Franklin, C. Zhou, M. G. Chapline, S. Peng, K. Cho and H. Dai, Science, 2000, 287, 622–625. 11 E. S. Snow, F. K. Perkins, E. J. Houser, S. C. Badescu and T. L. Reinecke, Science, 2005, 307, 1942–1945. 12 K. Besteman, J.-O. Lee, F. G. M. Wiertz, H. A. Heering and C. Dekker, Nano Lett., 2003, 3, 727–730. 13 B. L. Allen, P. D. Kichambare and A. Star, Adv. Mater., 2007, 19, 1439–1451. 14 J. R. Wood and D. H. Wagner, Applied Physics Letters, 2000, 76, 2883–2885. 15 S. Ghosh, A. K. Sood and N. Kumar, Science, 2003, 299, 1042–1044. 16 Y. Gao, Y. Bando, Z. Liu, D. Golberg and H. Nakanishi, App. Phys. Fig. 7 TEM image illustrating the individual SWCNT nature of the Lett., 2003, 83, 2913–2915. supernatant with the wrapping of 1C around the SWCNT surface. 17 J. W. G. Wildoer, L. C. Venema, A. G. Rinzler, R. E. Smalley and C. Dekker, Nature, 1998, 391, 59–62. 18 M. C. LeMieux, M. Roberts, S. Barman, Y. W. Jin, J. M. Kim and dispersive X-ray analysis (EDX) has also been carried out on the Z. Bao, Science, 2008, 321, 101–103. sample with no peaks attributed to any remaining iron catalyst 19 R. Krupke, F. Hennrich, H. V. Lohneysen and M. M. Kappes, particles, further supporting the use of these systems for post- Science, 2003, 301, 344–347. 20 Z. Chen, X. Du, M. Du, C. D. Rancken, H. Cheng and A. G. Rinzler, growth purification. Nano Letters, 2003, 3, 1245–1249. 21 M. S. Arnold, A. A. Green, J. F. Hulvat, S. I. Stupp and Conclusion M. C. Hersam, Nat. Nanotechnol., 2006, 1, 60–65. 22 H. Yang, S. C. Wang, P. Mercier and D. L. Akins, Chem. Commun., We have reported the successful use of p-phosphonated calixar- 2006, 1425–1427. 23 Q.-H. Yang, N. Gale, C. J. Oton, F. Li, A. Vaughan, R. Saito, ene systems to solubilize SWCNTs into aqueous media. We have I. S. Nandhakumar, Z.-Y. Tang, H.-M. Cheng, T. Brown and directly compared the p-phosphonated calixarene systems to W. H. Loh, Nanotechnology, 2007, 18, 405706. p-sulfonated calixarene systems and have revealed a strong 24 K. Saito, V. Troiani, H. Qiu, N. Solladie, T. Sakata, H. Mori, binding affinity present between the ‘extended arm’ p-sulfonated M. Ohama and S. Fukuzumi, J. Phys. Chem. C, 2007, 111, 1194–1199. 25 R. M. Tromp, A. Afzali, M. Freitag, D. B. Mitzi and Z. Chen, Nano calix[8]arenes and the SWCNT surface through Raman Lett., 2008, 8, 469–472. spectroscopy. These results show the potential for the use of 26 S. Park, H.-S. Yang, D. Kim, K. Jo and S. Jon, Chem. Commun., p-phosphonated calixarenes not only for solubilization of 2008, DOI: 10.1039/b802057d. SWCNTs but as a purification route with selective diameter 27 L. J. Hubble and C. L. Raston, Chem. Eur. J., 2007, 13, 6755–6760. 28 J. L. Atwood, R. J. Bridges, R. K. Juneja and A. K. Singh, US Pat. uptake, tunable with the specific system utilized. Furthermore, 94-1786105489612, 1996. we have assigned the different SWCNT topologies and have 29 N. Douteau-Gue´vel, A. W. Coleman, J.-P. Morel and N. Morel- determined the electronic nature of the selectively solubilized Desrosiers, J. Phys. Org. Chem., 1998, 11, 693–696. 30 A. Specht, F. Ziarelli, P. Bernard, M. Goeldner and L. Peng, HeIv. SWCNTs. This selective diameter and chirality solubilization Chim. Acta, 2005, 88, 2641–2653. provides a step forward in the post-growth treatment of 31 T. E. Clark, M. Makha, A. N. Sobolev, H. Rohrs, J. L. Atwood and SWCNTs to obtain application specific SWCNTs in areas such C. L. Raston, Chem. Eur. J., 2008, 14, 3931–3938. as nano-electronics and device technology. 32 A. S. Lobach, I. S. Ryzhkina, N. G. Spitsina and E. D. Obraztsova, Phys. Stat. Sol., 2007, 244, 4030–4034. 33 T. Ogoshi, T.-A. Yamagishi and Y. Nakamoto, Chem. Commun., Acknowledgements 2007, 4776–4778. 34 T. E. Clark, M. Makha, A. N. Sobolev, D. Su, H. Rohrs, M. L. Gross, The authors acknowledge the facilities, scientific and technical J. L. Atwood and C. L. Raston, New J. Chem., 2008, 32, 1478–1483. assistance of the Australian Microscopy & Microanalysis 35 S. Shinkai, S. Mori, H. Koreishi, T. Tsubaki and O. Manabe, J. Am. Chem. Soc., 1986, 108, 2409–2416. Research Facility at the Centre for Microscopy, Characterisation 36 M. Makha and C. L. Raston, Tetrahedron Lett., 2001, 42, 6215–6217. & Analysis, The University of Western Australia, a facility 37 M. Makha and C. L. Raston, Chem. Commun., 2001, 2470–2471. funded by The University, State and Commonwealth Govern- 38 M. A. Pimenta, A. Marucci, S. D. M. Brown, M. J. Matthews, ments. Furthermore, we gratefully acknowledge the Australian A. M. Rao, P. C. Eklund, R. E. Smalley, G. Dresselhaus and M. S. Dresselhaus, J. Mater. Res., 1998, 13, 2396–2404. Research Council for financial support of this work. 39 L. F. Su, J. N. Wang, F. Yu, Z. M. Sheng, H. Chang and C. Pak, Chem. Phys. Lett., 2006, 420, 421–425. References 40 V. A. Sinani, M. K. Gheith, A. A. Yaroslavov, A. A. Rakhnyanskaya, K. Sun, A. A. Mamedov, J. P. Wicksted and N. A. Kotov, J. Am. 1 S. Iijima, Nature, 1991, 354, 56–58. Chem. Soc., 2005, 127, 3463–3472. 2 S. Iijima and T. Ichihashi, Nature, 1993, 363, 603–605. 41 A. M. Rao, E. Richter, S. Bandow, B. Chase, P. C. Eklund, 3 D. S. Bethune, C. H. Kiang, M. S. de Vries, G. Gorman, R. Savoy, K. A. Williams, S. Fang, K. R. Subbaswamy, M. Menon, A. Thess, J. Vazques and R. Beyers, Nature, 1993, 363, 605–607. R. E. Smalley, G. Dresselhaus and M. S. Dresselhaus, Science, 4 J. Appenzeller, Proceedings of the IEEE, 2008, 96, 201–211. 1997, 275, 187–191. 5 H. Dai, J. H. Hafner, A. G. Rinzler, D. T. Colbert and R. E. Smalley, 42 S. M. Bachilo, M. S. Strano, C. Kittrell, R. H. Hauge, R. E. Smalley Nature, 1996, 382, 147–150. and R. B. Weisman, Science, 2002, 298, 2361–2366. 6 J. H. Hafner, C. L. Cheung and C. M. Lieber, J. Am. Chem. Soc., 43 S. Bandow, M. Takizawa, K. Hirahara, M. Yudasaka and S. Iijima, 1999, 121, 9750–9751. Chem. Phys. Lett., 2001, 337, 48–54. 7 G. Girishkumar, T. D. Hall, K. Vinodgopal and P. V. Kamat, J. Phys. 44 M. S. Dresselhaus, G. Dresselhaus, A. Jorio, A. G. Souza Filho and Chem. B, 2006, 110, 107–114. R. Saito, Carbon, 2002, 40, 2043–2061. 8 O. Breuer and U. Sundararaj, Polym. Compos., 2004, 25, 630–645. 45 S. Banerjee and S. S. Wong, Nano Lett., 2004, 4, 1445–1450. 9 A. A. White, S. M. Best and I. A. Kinloch, Int. J. Appl. Ceram. 46 S. Attal, R. Thiruvengadathan and O. Regev, Anal. Chem., 2006, 78, Technol., 2007, 4, 1–13. 8098–8104.

5966 | J. Mater. Chem., 2008, 18, 5961–5966 This journal is ª The Royal Society of Chemistry 2008

281 Publications

Additions and corrections

Selective diameter uptake of single-walled carbon nanotubes in water using phosphonated calixarenes and ‘extended arm’ sulfonated calixarenes

Lee J. Hubble, Thomas E. Clark, Mohamed Makha and Colin L. Raston J. Mater. Chem., 2008, 18, 5961– 5966 (DOI: 10.1039/b814904f) Amendment published 11th December 2008.

The authors regret the following errors:

A sentence in the section Raman spectroscopy in the Results and discussion that reads:

-1 “The results indicate that the peak width of the D-band at ~1590 cm has decreased -1 -1 significantly from 100 cm in the as-received SWCNTs to 20–40 cm (FWHM) in all tested systems (Fig. 4A and 5A).” should read:

-1 “The results indicate that the peak width of the D-band at ~1340 cm has decreased -1 -1 significantly from 100 cm in the as-received SWCNTs to 20–40 cm (FWHM) in all tested systems (Fig. 4A and 5A).”

A sentence in the same section that reads:

“As previously mentioned the smaller diameter SWCNTs were more readily solubilized into aqueous media. However, the data indicate that there are larger SWCNT diameters solubilized when using 4B (1.55 and 1.42 nm) and 4C (1.35 nm). In the case of 4B and 4C involving the ‘extended arm’ sulfonated calix[8]arenes, the large G-band shifts indicate strong nanotube surface interactions. This suggests a different type of interplay with the SWCNT surface, in solubilizing the larger diameter SWCNTs. This is consistent with the increased conformational flexibility of the larger ring size calixarene.” should read:

“As previously mentioned the smaller diameter SWCNTs were more readily solubilized into aqueous media. However, the data indicates that there are larger SWCNT diameters solubilized when using 4C (1.35 nm). In the case of 4C involving an ‘extended arm’ sulfonated calix[8]arene, the large G-band shift indicates strong nanotube surface interactions. This suggests a different type of interplay with the SWCNT surface, in solubilizing the larger diameter SWCNTs. This is consistent with the increased conformational flexibility of the larger ring size calixarene.”

282 Publications

The Royal Society of Chemistry apologises for these errors and any consequent inconvenience to authors and readers.

Additions and corrections can be viewed online by accessing the original article to which they apply.

This journal is © The Royal Society of Chemistry 2009

283 Publications

Selective diameter uptake of single-walled carbon nanotubes in water using phosphonated calixarenes and ‘extended arm’ sulfonated calixarenes

Supporting Information

Lee J. Hubble, a,b Thomas E. Clark, a Mohamed Makha, a and Colin L. Raston *a a Centre for Strategic Nano-Fabrication, School of Biomedical, Biomolecular and Chemical Sciences, The University of Western Australia, 35 Stirling Highway, Crawley, WA, 6009, Australia. Fax: (+61) 8-6488-8683; Tel: (+61) 8-6488-1572; Email: [email protected] b Centre for Forensic Science, The University of Western Australia, 35 Stirling Highway, Crawley, WA, 6009, Australia.

284 Publications

Supplementary Figure 1. Raman spectrum of the ‘blank’ aluminium foil, inset of the typical radial breathing mode frequency range. These ‘baseline’ frequencies were taken into account for the analysis of the as-received SWCNTs and supernatant residues. This also represents the spectrum obtained when analysing calixarene control samples.

285 Publications

Supplementary Figure 2a. UV-visible spectra of the p-phosphonated calixarenes (1A- 3A) and p-sulfonated calixarenes (4A-4C).

Supplementary Figure 2b. UV-visible spectra of the p-phosphonated calixarenes (1A- 3A)/ SWNT supernatants and p-sulfonated calixarenes (4A-4C) / SWCNT supernatants.

286 Publications

FULL PAPER

DOI: 10.1002/chem.200700339

Nanofibers of Fullerene C60 through Interplay of Ball-and-Socket SuperACHTUNGTRENNUNGmolecules

Lee J. Hubble[b] and Colin L. Raston*[a]

Abstract: Mixing solutions of p-tBu- (powder X-ray diffraction data coupled tion of the calixarenes to afford C60- calix[5]arene and C60 in toluene results with molecular modeling). Under core nanostructures encapsulated in a \ 8 in a 1:1 complex (C60) (p-tBu-calix[5]- argon at temperatures above 309 C, graphitic material sheath, which exhib- arene), which precipitates as nanofib- the fibers undergo selective volatiliza- its a dramatic increase in surface area. ers. The principle structural unit is Above 6508C the material exhibits an based on a host–guest ball-and-socket ohmic conductance response, due to Keywords: calixarenes · fullerenes · nanostructure of the two components, the encapsulation process. host–guest systems · nanostructures · with an extended structure comprising supramolecular chemistry zigzag/helical arrays of fullerenes

Introduction cavitands, with particular interest being devoted to calixar- enes and related molecules. Atwood et al.[16] and Suzuki Carbon-based materials have a wide variety of applications et al.[17] initially developed a convenient and efficient purifi- [1–4] [5–7] [8–11] such as catalysis, fuel cells, separations, extractions, cation method for obtaining C60 from “crude” fullerene and sensor technology.[12–14] Overall, carbon is remarkably soot, which is based on the formation of a discrete complex versatile and offers the possibility of fabricating a diverse with p-tBu-calix[8]arene. Furthermore, the different effects range of novel products entirely composed of the element. of lower phenolic oligomers and substituted lower and

The discovery of fullerene C60 and higher fullerenes, and the upper ring functionalization has led to unique complex later development of carbon nanotubes, has added signifi- arrays of fullerenes that display fullerene···fullerene inter- cantly to this possibility becoming a reality. play.[18–20] Examples of such interplay include the 1:1 com- [21] In developing the full potential of carbon it is important plexes between C60 and cyclotriveratrylene (CTV), and [22] to be able to accurately prepare new structural carbon calix[5]arene where the C60 molecules are packed into frameworks and to be able to generate new materials with continuous zigzag chains; C60···C60 close contacts vary from different spatial arrangement of arrays of carbon atoms. 9.90 to 10.20 , with dihedral angles in the chains ranging One way of achieving this is through self-assembly process- from 118 to 1728, indicating a transitional packing structure es. Fullerenes, as the only form of molecular carbon, are in between a linear chain and double-column array.[20] a unique position to be used for these processes. A recent Bowl-shaped calix[5]arenes are effective in controlling the example of this is the formation of a one-dimensional C60 assembly of fullerenes, with dimensionality and curvature [15] [23] array in solution templated by helical nanotubes complementarity for endo-cavity binding C60. Herein we

An area that has attracted significant attention in recent show that p-tBu-calix[5]arene forms a complex with C60 years is the supramolecular chemistry of fullerenes involving from toluene, as nanofibers based on the 1:1 supermolecule \ (C60) (p-tBu-calix[5]arene) [Eq. (1)]. [a] Prof. C. L. Raston Optical and transmission electron micrographs of the School of Biomedical, Biomolecular & Chemical Sciences structure of these bundles of nanofibers are presented in University of Western Australia Figure 1 a and 1b, respectively. We also show that the calix- 35 Stirling Highway, Crawley, 6009 (Australia) arenes can be removed at high temperatures under an inert Fax : (+61)8-6488-8683 atmosphere to afford encapsulated nanofibers of pure C E-mail: [email protected] 60 with a large increase in surface area compared to the parent [b] L. J. Hubble Centre for Forensic Science, University of Western Australia complex. Furthermore, we indicate that “crude” fullerene 35 Stirling Highway, Crawley, 6009 (Australia) soot similarly affords nanofibers, albeit with a small percent-

Chem. Eur. J. 2007, 13, 6755 – 6760 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 6755

287 Publications

the complex derived from fullerite is presumably the same

as the structure for the complex formed by using pure C60

(see below), with the upper limit of C70 inclusion governed by the geometrical mismatch associated with incorporating a non-spheroidal fullerene into the continuous array. The \ (C60) (p-tBu-calix[5]arene) complex undergoes rapid de- composition in dichloromethane, with the precipitation of \ pure C60 (Figure 2). The analogous (C60) (p-tBu-calix[8]ar-

\ Figure 2. IR spectra of the product of the decomplexation of (C60) (p- tBu-calix[5]arene) in dichloromethane, additional absorbance peaks are attributed to residual dichloromethane.[16] \ Figure 1. Optical micrograph (a) and TEM image (b) of the (C60) (p- tBu-calix[5]arene) complex, the latter revealing the linear striations of [16] fullerene C60 within bundles of nanofibers. ene) complex similarly degrades in dichloromethane. Solid-state NMR spectroscopy of the complex clearly

shows freely rotating C60 at ambient temperature, and also age of the next highest fullerene, C70, incorporated into the some rotation of the tertiary butyl groups (Figure 3). The structure.

Results and Discussion

\ The synthesis of the (C60) (p-tBu-calix[5]arene) complex re- quires a large excess of calixarene, and is formed in 56% yield based on C60. After two hours the typical magenta color of the C60 solution turns dark brown indicating com- plex formation. At this point optical microscopy reveals a suspension of light brown needlelike, fibrous material (Fig- ure 1a). To ensure a uniform product, continuous stirring is essential. The mother liquor can be recycled with the addi- tion of an equivalent of C60 and p-tBu-calix[5]arene, there- after affording a quantitative yield of the complex. A variety of different synthetic and crystallization conditions were in- vestigated, such as different host–guest molar ratios, host– 13 \ p t guest reaction periods, rates of solvent evaporation, solvent- Figure 3. CP-MAS solid state C NMR spectra for (C60) ( - Bu-calix[5]- arene), with (bottom) and without (top) magic angle spinning. less reactions, and the use of ultrasonication. Indeed, grind- ing an equimolar mixture of C60 and p-tBu-calix[5]arene fol- lowed by sonication in hexane afforded the same fibrous TEM image shows ordered linear striations throughout the product, albeit in low purity with un-reacted starting materi- nanofibers, which correlate to columnar arrays of fullerenes al present. (see Figure 1b).

Using a mixture of fullerenes (fullerite) in place of C60 Calix[5]arenes have complementarity of curvature with also afforded a nanofiber complex, with some C70 incorpo- C60 and form complexes with the fullerene residing in the rated (UV/Vis), whereas reacting pure C70 with the same cavity, either as a ball-and-socket nanostructure, or with two calixarene did not form a solid complex. The structure of calixarenes shrouding a single fullerene molecule.[22,24] This,

6756 www.chemeurj.org 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2007, 13, 6755 – 6760

288 Publications

Nanofibers of Fullerene C60 FULL PAPER coupled with the ratio of 1:1 for the two components in the an increase in peak intensity at a 2q value of 6.898 was ob- present complex, is consistent with the supermolecule unit served. This corresponds to the three planar C60 molecules as a ball-and-socket nanostructure [Eq. (1)]. The supermole- (A, B, & C) being parallel with the X-ray beam (Figure 5). cules then propagate to 1D arrays, attributed to columnar The additional reflections correlate to other intra-column arrays of C60, noting that such an array is encountered in the fullerene planes, with increasing angle reflections from calix[5]arene system.[22] Bundles of these arrays agglomerate intra-fullerene domains. to form the macroscopic fibrous product. In comparison to Owing to the nature of the sample, single-crystal diffrac- the calix[5]arene system, the present system has additional tion data could not be obtained; however, the PXRD results geometrical packing constraints due to the bulky tertiary were incorporated into a model developed by using Accelrys butyl groups on the upper rim of the calixarene, but never- software.[25] For illustrative clarity, all the p-tBu-calix[5]arene theless fullerene···fullerene interactions are still possible. molecules shrouding the fullerenes have been omitted for The powder X-ray diffraction (PXRD) pattern is domi- Figure 5 a and 5b. The model consistent with the data has a q 8 nated by a 2 peak at 3.93 (Figure 4), which is attributed to single column of C60 comprising three C60 molecules in- 8 the inherent distance between individual C60 columns, at a plane (A, B, & C) with a dihedral angle of 80.7 and cen- troid···centroid distances of 0.99 nm (A & B) and 1.28 nm (A & C). This is the repeating unit and is successfully rotat- ed by 908 with four turns corresponding to a complete turn for a helical array. Atwood et al.[22] report a similar model based on single-crystal diffraction data with the unsubstitut- ed calix[5]arene, which forms a linear zigzag array; however, this model does not fit the data obtained from PXRD in re- \ lation to the (C60) (p-tBu-calix[5]arene) complex. The ar- rangement of calixarenes in a single helical array is illustrat- ed in Figure 6. This array propagates along its principle axis to form the strands of fullerene fibers seen in Figure 1b. \ Figure 4. Powder X-ray diffraction pattern of (C60) (p-tBu-calix[5]arene). distance of 2.25 nm. The fibrous nature of the sample gives rise to preferred diffraction pattern orientation effects in the patterns at different sample orientations. The next highest intensity peak is at a 2q value of 6.898 with a d-spacing of 1.28 nm. Overall, the diffraction pattern and orientation ef- fects correlate well to the proposed structure (Figure 5). Through acquisitions at different sample holder orientations

Figure 6. Proposed calixarene packing arrangement around a C60 column, with the same orientation as Figure 5 b, based on Accelrys.[25]

Thermogravimetric analysis (TGA) studies were carried out under air and argon (Figure 7). Under air there is an ini- tial 3% weight decrease from 238C until 2138C, which is at- tributed to solvent surface desorption. This is interestingly

Figure 5. a) Proposed helical array of C60 columns; labels denote a tri- ACHTUNGTRENNUNG 8 \ angular60 column set of rotated fullerenes. 45 about b) C the horizontal Figure 7. TGA of the nanofibers of (C60) (p-tBu-calix[5]arene) under ACHTUNGTRENNUNGaxis in (a). x both air and argon atmospheres.

Chem. Eur. J. 2007, 13, 6755 – 6760 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 6757

289 Publications

C. L. Raston and L. J. Hubble

followed by a gradual 5% weight increase from 2138Ctoa the selective removal of large cavitand molecules at high maximum at 3098C which is attributed to partial oxidation temperatures is without precedent in relation to fullerene ar- [26] of C60 followed by a gradual weight reduction at a rate of chitecture. 0.6 weight%8C1, until final product decomposition at TGA data from the isothermal hold experiments indicate 5038C. Under argon there is an initial 2% weight loss up to that temperatures of 6508C and 7008C, correlate to a 36 and 2138C followed by a 34% weight decrease from 309–5178C, 37% weight loss, respectively. For temperatures of 7508C then a slow declining plateau region from 517–7058C corre- and 8008C there is a 54% and 56% sample weight loss, re- sponding to an 8% weight loss. The total weight loss prior spectively. Coupling of the SEM and TGA results suggest to C60 degradation corresponds to loss of the calixarene for that heating under argon at increasing temperatures corre- a 1:1 complex. Further weight loss is associated with degra- lates to the p-tBu-calix[5]arene backbone being selectively dation of C60 (see below) with a change in rate of vaporiza- volatilized from the sample, leaving bundles of C60 nanoar- tion above 8008C, as reported previously by Malhotra, et al. rays, albeit with some graphitic material on the surface of

Here there is a thermally activated transformation of the the fibers. The formation of an all-C60 core structure was fullerenes, in which larger covalently bonded fullerene also confirmed by infrared spectroscopy, with typical C60 ab- arrays are formed.[27] In the present study only 5% of the sorbance peaks at 526, 575, 1181 and 1428 cm1. Further- original sample weight is retained at temperatures up to more, PXRD on a 7008C treated sample revealed a face- 12008C under an argon atmosphere. centered cubic (fcc) crystal packing structure, which is the [20] Decomposition under argon was further investigated structure of pristine C60. through a series of isothermal hold experiments, ranging Brunauer, Emmett, and Teller (BET) surface area meas- from 400–8008Cin508C intervals. The post-heat treated urements were undertaken for the as-synthesized and iso- sample morphology was evaluated by using scanning elec- thermal hold products (Figure 9). Interestingly, there is a 45- tron microscopy (SEM), highlighted in Figure 8a-d. The re-

Figure 9. BET surface area against percent weight loss for as-synthesized and argon-annealed products, with associated annealing temperatures. Figure 8. a) SEM image of as-synthesized bundles of nanofibers, and SEM images of bundles of nanofibers treated under argon at b) 400 8C, c) 700 8C, and d) 800 8C, all scale bars represent 10 mm. fold increase in the surface area exhibited for the product obtained at 8008C in relation to the as-synthesized product. sulting materials appear to have crystalline and amorphous The surface area correlates with trends derived from TGA domains, with PXRD results confirming crystallinity in the data (Figure 7). This indicates a steep increase in surface samples. At temperatures below 6008C the principle topo- area until a plateau is reached at 600–7008C followed by an- graphical structure was retained for bundles of nanofibers. other steep increase. The surface area increase is due to the The samples derived at 4008C were constantly plagued by selective volatilization of the calixarene backbone and the charging effects, as was the case for the as-synthesized com- formation of graphitic material providing increasing sites for plex (Figure 8b and 8a, respectively). At annealing temper- gas absorption. This is interesting to note as PXRD and IR 8 atures inclusive of 650–700 C, samples no longer suffered data indicate a C60 fcc crystal structure despite the large sur- from charging (Figure 8c). This relates to loss of the calixar- face area established by BET. enes and also indicates that the sample is now conductive. A study into the macroscopic electrical properties of the Above 7008C structural degradation of the bundles of fuller- high temperature argon annealed analogues was carried out. ene fibers occurs with increased surface defects, notably The objective of this study was to confirm if high tempera- with pitted or ’skeleton’ array structures (Figure 8d). The ture argon annealing of the parent complex resulted in con- conductivity is associated with formation of graphitic mate- ductive material, which was inferred from the higher tem- rial (see below). The degradation at high temperature aside, perature SEM samples not being susceptible to charging

6758 www.chemeurj.org 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2007, 13, 6755 – 6760

290 Publications

Nanofibers of Fullerene C60 FULL PAPER

[29] during imaging. The results for each sample were compared ture values. The G band relates to the graphitic E2g sym- 2 to that obtained for single crystals of pristine C60, which at metry mode and reflects the structural integrity of the sp - room temperature is known to be a highly insulating n-type hybridized carbon atoms of the graphitic material. While, material.[28] I–V curves and resistance values were obtained, the D band, or diamond band, indicates disordered sp2-hy- with results shown for the 8008C analogue (Figure 10). As bridized carbon atoms within the sample. It is apparent that graphitic material is being formed in the annealing process. However, it is interesting to note that only the high-temperature annealed samples (6508C) ex- hibited conductive behavior, indicated by electrical resist- ance measurements. With the analysis of the G to D intensi- ty ratios it is evident that a switch in the G to D ratio to less than one occurs when the material becomes conductive. This could either be due to the formation of more disor- dered carbon atoms at the higher temperatures or simply an increased coating is being formed fully encapsulating the Figure 10. I–V and resistance curves for the analogue annealed under fibers. In this context we note that the conducting material argon at 800 8C. results in little dissolution of fullerene C60 in the presence of toluene. We conclude that at the higher temperatures ( 6508C) a graphitic material (not graphite itself), such as ac- expected a non-ohmic conductance response was obtained tivated charcoal, encapsulates the remaining C60 fiber and for a single crystal of C60 at room temperature which is a forms a conductive sheath around a central core. The re- known insulating material. Similar results were obtained for maining presence of a C60 core is inferred from typical IR all the analogues annealed under argon until the annealing absorbance peaks attributed to C60. temperature of 6508C was reached, whereby the materials exhibited an ohmic conductance response. This is clearly seen for the 8008C analogue where a linear I–V curve and a Conclusion relatively stable resistance value are produced through the voltage sweep (Figure 10). The bulk material resistance The results demonstrate the formation of C60 nanoarrays values obtained for the 8008C analogues were about 1200W. that form bundles of nanofibers based on complexation of

These results reinforce the assumption that the material C60 and p-tBu-calix[5]arene from toluene in a quantitative formed under an argon atmosphere at annealing tempera- yield. A grinding/sonication method for obtaining the com- tures of 6508C forms a conductive material, the conductiv- plex has also been established, and complex formation is ity of which is ascribed to the formation of graphitic materi- possible from crude fullerene soot. Annealing under an al, as discussed below. inert atmosphere demonstrates the use of this complex as a Further experiments were carried out to assess the high precursor in preparing exclusively carbon-based material, temperature annealed samples to indicate a probable cause which can be conducting. The ensuing large surface area has of the change in electrical behavior at specific temperatures. implications for the development of catalyst support frame- Dispersive Raman spectroscopy was undertaken on the works, hydrogen and other gas storage material architecture, range of high-temperature analogues and the as-synthesized and device technology. product. The Raman spectrum of the as-synthesized product was similar to that of pure C60. In contrast, the high temper- ature annealed material gave additional Raman lines (Figure 11), notably around 1580 cm1 and 1330 cm1. These Experimental Section peaks correspond to the so-called G (graphite) and D (dia- mond) bands, respectively, and are in agreement with litera- \ p t Synthesis of (C60) ( - Bu-calix[5]arene): On dissolution of C60 (40.7 mg, 5.65 102 mmol, BuckyUSA 99 %) in toluene (40 mL), 13 molar equiva- lents of p-tBu-calix[5]arene (596.3 mg, 0.7351 mmol) were added. Gravity filtration, followed by a wash with hexane (3 2 mL), afforded a light brown/golden solid (48.7 mg, 3.18102 mmol), which was dried under vacuum (56 % yield, elemental analysis (%) calcd: C 90.17, H 5.22; found: C 90.14, H 5.28). Transmission electron microscopy (TEM) was performed on a JEOL 3000F FEGTEM operating at 200 kV; sample preparation involved form- ing a liquid suspension of the complex in hexane and then putting a drop of the dispersion onto a holey carbon film supported by copper grids and air-dried. As-prepared samples and annealed analogues were directly ex- amined on a Zeiss 1555 VP field emission scanning electron microscope; Figure 11. Raman spectra of the high-temperature analogues and as-syn- sample preparation involved the dispersion of material directly onto thesized product. carbon tape on a standard aluminum stub.

Chem. Eur. J. 2007, 13, 6755 – 6760 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 6759

291 Publications

C. L. Raston and L. J. Hubble

Solid-state C13 CP-MAS spectra were measured on a Bruker ARX [5] Y. Fangli, R. Hojin, Nanotechnology 2004, 15, S596-S602. 300 MHz NMR spectrometer at a frequency of 75.468 MHz, a spin rate [6] F. Su, J. Zeng, Y. Yu, L. Lu, J. Y. Lee, X. S. Zhao, Carbon 2005, 43, of 4 kHz, contact time of two milliseconds and relaxation delay of two 2366– 2373. seconds. Powder XRD patterns were obtained on a Siemens Diffractome- [7] G. S. Chai, S. B. Yoon, J.-S. Yu, Carbon 2005, 43, 3028– 3031. l= ter D500 by utilizing CuKa radiation ( 0.154 nm); sample preparation [8] M. A. de la Casa-Lillo, B. C. Moore, D. Cazorla-Amoros, A. Li- involved placing approximately 50 mg of material on a XRD sample nares-Solano, Carbon 2002, 40, 2489 –2494. holder, levelled with a glass microscope slide. [9] S. M. Saufi, A. F. Ismail, Carbon 2004, 42, 241– 259. Thermogravimetric analysis (TGA) was performed on a TA Instruments [10] T. Yamamoto, A. Endo, T. Ohmori, M. Nakaiwa, Carbon 2004, 42, SDT 2960 Simultaneous DSC-TGA in either air or argon (100 mLmin1) 1671– 1676. with a temperature program of 38C min1 for varied temperature ranges. [11] L. Shi, X. Liu, H. Li, W. Niu, G. Xu, Anal. Chem. 2006, 78, 1345– For isothermal hold experiments a temperature program of 208Cmin1 1348. under argon (100 mLmin1) with a 20 minute hold period at the pro- [12] J. Kong, N. R. Franklin, C. Zhou, M. G. Chapline, S. Peng, K. Cho, grammed final temperature. Brunauer, Emmett, and Teller (BET) sur- H. Dailt, Science 2000, 287, 622 –625. [13] E. S. Snow, F. K. Perkins, E. J. Houser, B. S. C. , T. L. Reinecke, Sci- face area measurements were measured by adsorption of N2 at 77 K using a Micromeritics Tristar 3000; the sample was loaded in a quartz ence 2005, 307, 1942– 1945. tube evacuated to 0.1 millibar and cooled with liquid nitrogen to 77 K. [14] H. Xie, Q. Yang, X. Sun, J. Yang, Y. Huang, Sens. Actuators B 2006, Nitrogen gas was incrementally released into the quartz tube for absorp- 113, 887 –891. tion measurements at different vapor pressures. Electrical resistance [15] G. D. Pantos, J.-L. Wietor, K. M. Sanders, Angew. Chem. 2007, 119, measurements were carried out on the samples using a micromanipulator 2288– 2290; Angew. Chem. Int. Ed. 2007, 46, 2238–2240. coupled with a precision semiconductor parameter analyzer (Hewlett [16] J. L. Atwood, G. A. Koutsantonis, C. L. Raston, Nature 1994, 368, Packard, 4156 A). Dispersive Raman spectroscopy was performed with a 229– 231. Dilor Labram 1B Raman spectrometer instrument. The spectral range [17] T. Suzuki, K. Nakashima, S. Shinkai, Chem. Lett. 1994, 4, 699– 702. covered 127.38 to 3063.17 cm1 using a grating of 600 lines per mm. The [18] J. L. Atwood, L. J. Barbour, C. L. Raston, Cryst. Growth Des. 2002, samples were mounted and analyzed on glass microscope slides. 2,3–6. [19] M. Makha, M. J. Hardie, C. L. Raston, Chem. Commun. 2002, 14, 1446– 1447. [20] M. Makha, A. Purich, C. L. Raston, A. N. Sobolev, Eur. J. Inorg. Acknowledgements Chem. 2006, 507 –517. [21] J. L. Atwood, M. J. Barnes, M. G. Gardiner, C. L. Raston, Chem. The authors gratefully acknowledge the Australian Research Council for Commun. 1996, 12, 1449 –1450. support of this work. We would like to acknowledge that electron micro- [22] J. L. Atwood, L. J. Barbour, M. W. Heaven, C. L. Raston, Angew. scopy was carried out using facilities at the Centre for Microscopy and Chem. 2003, 115, 3376– 3379; Angew. Chem. Int. Ed. 2003, 42, 3254 – Microanalysis/Biomedical Image and Analysis Facility, The University of 3257. Western Australia, which are supported by University, State, and Federal [23] T. Haino, M. Yanase, Y. Fukazawa, Tetrahedron Lett. 1997, 38, Government funding. We would also like to acknowledge Dr. Charlie 3739– 3742. Musca from the School of Electrical, Electronic, and Computer Engi- [24] M. Makha, J. J. McKinnon, A. N. Sobolev, M. A. Spackman, C. L. neering for assistance with the electrical measurements carried out on Raston, Chem. Eur. J. 2007, DOI: 10.1002/chem.200601188. the samples. [25] Accelrys, MS Visualizer, Accelrys Software, Inc., San Diego, 2006. [26] A. Datta, R. Y. Kelkar, M. Datta, J. Mater. Sci. Lett. 1994, 13, 996– 999. [1] M. J. Ledoux, R. Vieira, C. Pham-Huu, N. Keller, J. Catal. 2003, 216, [27] R. Malhotra, D. C. Lorents, Y. K. Bae, C. H. Becker, D. S. Tse, L. E. 333– 342. Jusinski, E. D. Wachsman, in Fullerenes: synthesis, properties, and [2] M. Hara, T. Yoshida, A. Takagaki, T. Takata, J. N. Kondo, S. Haya- chemistry of large carbon clusters (Eds.: G. S. Hammond, V. J. shi, K. Domen, Angew. Chem. 2004, 116, 3015–3018; Angew. Chem. Kuck), American Chemical Society, Washington, D. C., 1992, Int. Ed. 2004, 43, 2955– 2958. pp. 127 –139. [3] A.-H. Lu, W. Schmidt, N. Matoussevitch, H. Bonnemann, B. Splieth- [28] R. Maruyama, J. Appl. Phys. 2003, 94, 6871 –6874. off, B. Tesche, E. Bill, W. Kiefer, F. Schuth, Angew. Chem. 2004, 116, [29] F. Tuinstra, J. L. Koenig, J. Chem. Phys. 1970, 53, 1126 –1130. 4403– 4406; Angew. Chem. Int. Ed. 2004, 43, 4303–4306. [4] N. Keller, G. Rebmann, E. Barraud, O. Zahraa, V. Keller, Catal. Received: March 1, 2007 Today 2005, 101, 323 –329. Published online: May 29, 2007

6760 www.chemeurj.org 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2007, 13, 6755 – 6760

292